Fundamentals of Modern Physics
Uploaded by
Israel TuxpanFundamentals of Modern Physics
Uploaded by
Israel TuxpanMuyjjjjJHpwi*
FUNDAMENTALS OF
MODERN
PHYSICS
FUNDAMENTALS OF
ROBERT MARTIN EISBERG
Associate Professor of Physics
University of California
Santa Barbara
John Wiley
&
Sons, Inc., New York
London
Sydney
MODERN PHYSICS
SoS'Q
13 14 15 16 17 18 19 20
Copyright
1961 by John Wiley & Sons,
All rights reserved.
Inc.
This book or any part
thereof must not be reproduced in any form
without the written permission of the publisher.
Library of Congress Catalog Card
Number: 61-6770
Printed in the United States of America
ISBN
471 23463
Dedicated to
JACOB LOUIS EISBERG
Prefao
This book has evolved from mimeographed notes which I have been
using at the University of Minnesota as a text for a one-year advanced
undergraduate course in modern physics. Its selection of material re-
some
definite ideas that I have concerning such a course.
convinced that a modern course in modern physics should have
two purposes: In addition to teaching many of the standard topics of
flects
I
the
am
field,
the course also should take full advantage of
tunity to introduce the student to
quantum mechanics
its
unique oppor-
an early stage
and in a way which integrates the theory into its historical development and its applications.
The introduction must be
elementary, but it must not be cursory because such treatment would
only amplify the problems that so often arise in a first contact with
quantum mechanics. This program takes time, but it pays great dividends by laying the foundation for a really mature discussion of the
atom and the nucleus. A student can surely learn much more about
these subjects in one year if half of it is invested in developing their
at
in his education,
theoretical bases than
tially
if all
of
it is
spent in discussing
them
at
an essen-
empirical level.
The
in this
inclusion of
book makes
more than
it
the usual
amount of quantum mechanics
necessary that there be certain omissions.
Because
an undergraduate course in solid state physics, I have chosen most of the omissions from
material that would be
covered in such a course.
most
institutions
now
offer
Although there are items in the table of contents of this book which
usually are found only in textbooks of a strictly postgraduate level,
this
should not be taken as a false indication of its prerequisites.
I have
tried to start near the beginning of each topic, and initially
to employ
PREFACE
only arguments that can be appreciated by a student well trained in
elementary physics, and in mathematics through intermediate calculus.
Almost
all the material of the book has been presented on two or
three
separate occasions to a very heterogeneous group of physics, chemistry,
mathematics, and engineering students, ranging from juniors to post-
graduates. Their performance has shown that students at all levels,
and
from all specialties, are quite capable of following the material.
However, in its final form the book contains about twenty percent
more material than can be covered in 90 lecture-hours. This, plus the
fact that there
is a gradual build-up of level in proceeding
through the
book, should make it adaptable to a variety of needs. A course
for
undergraduate students, who have had no prior contact with modern
physics, might thoroughly cover the material through
Chapter 8 and
leave for outside reading certain topics in the remaining
chapters.
course for postgraduate students, who have already studied
modem
physics at an elementary level, might begin with a very rapid
review of
the material through Chapter 5, leaving the details for
outside reading,
and then thoroughly cover the remaining chapters. There is also
enough
material for a one-semester course in introductory
quantum mechanics.
would
like to
particularly Drs.
thank
W.
my
colleagues at the University of Minnesota
B. Cheston, C. E. Porter, and D. R. Yennie,
for
reading parts of the mimeographed notes and
making
many
useful sug-
would also like to thank my wife Lila for the constant
aid
and encouragement which made the completion of this
book possible.
gestions.
R. M. ElSBERG
February, 1961
Contents
Introduction
1.
What
is
modern physics?
2. Historical sketch, 2.
1.
The theory of
1.
relativity
The Galilean transformation and
classical
mechanics, 3.
Galilean transformation and electromagnetic theory, 7.
Michelson-Morley experiment,
Simultaneity, 16.
9.
2.
3.
4. Einstein's postulates,
The
The
15.
Kinematical effects of relativity, 19.
The Lorentz transformation, 24. 8. Transformation of ve-
5.
7.
locity,
27.
9. Relativistic
momentum and
theory, 37.
6.
mechanics, 30.
energy, 36.
11.
Bibliography, 39.
10. Transformation of
Experimental verification of the
Exercises, 39.
Thermal radiation and the origin of quantum
theory
1.
41
Introduction, 41.
charges, 42.
46.
4.
2. Electromagnetic radiation by accelerated
Emission and absorption of radiation by surfaces,
3.
Black body radiation, 47.
Rayleigh- Jeans theory, 57.
tribution, 57.
theory, 63.
8.
10.
liography, 68.
7.
5.
Wien's law, 49.
The Boltzmann
Comparison with experiment, 62.
Some comments on
Exercises, 68.
6.
The
probability dis9.
Planck's
Planck's postulate, 66.
Bib-
CONTENTS
3
1.
Cathode
Electrons
rays, 70.
and quanta
The e/m
2.
ratio of electrons, 72.
charge and mass of electrons, 73.
5.
The
photoelectric effect, 76.
electric effect, 78.
79.
Quantum
7.
The Compton
8.
effect,
magnetic radiation, 85.
A
1.
3.
70
6. Classical
81.
9.
The dual nature
Bibliography, 86.
son's model, 92.
106.
5.
9.
8.
The
Exercises, 86.
2.
particles, 89.
ThomRuth-
6.
7. Predictions of Rutherford's
Experimental verification and determination of Z,
size of the nucleus,
Bibliography, 709.
Alpha
87
4. Predictions of
Comparison with experiment, 98.
model of the atom, 99.
model, 100.
effect,
of electro-
The discovery of the atomic nucleus
scattering of alpha particles, 91.
erford's
The
theory of the photo-
theory of the photoelectric
Thomson's model of the atom, 87.
The
3.
Bucherer's experiment, 75.
4.
707.
10.
problem, 108.
Exercises, 709.
Bohr's theory of atomic structure
10
1. Atomic spectra, 770.
2. Bohr's postulates, 113.
3. Bohr's
theory of the one-electron atom, 775.
4. Correction for finite
nuclear mass, 722.
5. Atomic energy states, 124.
6. The
Wilson-Sommerfeld quantization
relativistic theory,
9.
A critique of the old
rules,
725.
7.
Sommerfeld's
The correspondence principle, 135.
quantum theory, 756. Bibliography, 137.
131.
8.
Exercises, 138.
6
1.
De
5.
746.
The
3.
2.
4. Interpretation of the
uncertainty principle, 153.
139
Some
properties of the pilot
Experimental confirmation of de Broglie's postu-
uncertainty principle, 759.
1.
and waves
Broglie's postulate, 759.
waves, 747.
late,
Particles
Bohr quantization
6.
rule,
Some consequences
Bibliography, 765.
757.
of the
Exercises, 765.
Schroedinger's theory of quantum mechanics
Introduction, 164.
terpretation of the
2.
wave
The Schroedinger
function, 770.
4.
equation, 765.
3. In-
The time independent
164
CONTENTS
xi
Schroedinger equation, 776.
5. Energy quantization in the
Schroedinger theory, 178. 6. Mathematical properties
of wave
functions and eigenfunctions, 184.
7. The classical theory of
transverse waves in a stretched string, 192.
and
differential operators,
mechanics, 206.
1.
9.
The
Bibliography, 210.
8.
Expectation values
classical limit of
quantum
Exercises, 210.
Solutions of Schroedinger's equation
The
201.
212
free particle, 212.
2. Step potentials, 221.
3. Barrier
Square well potentials, 239. 5. The infinite
square well, 251. 6. The simple harmonic oscillator,
256. Bibliography, 266. Exercises, 266.
potentials, 231.
4.
Perturbation theory
1.
Introduction, 268.
3.
An
268
2. Time independent perturbations, 268.
example, 274. 4. The treatment of degeneracies,
278.
5. Time dependent perturbation theory, 285.
Bibliography, 291.
Exercises, 291.
10
1.
One-electron atoms
Quantum mechanics
cles,
293.
293
for several dimensions
The one-electron atom, 295.
2.
and several
parti-
Separation and
solution of the equation of relative motion,
298.
4. Quantum
numbers, eigenvalues, and degeneracies, 302. 5.
Eigenfunctions
3.
and probability
densities, 304.
6. Angular momentum operators,
Eigenvalue equations, 318. 8. Angular momentum
of
one-electron-atom eigenfunctions, 323.
Bibliography, 325.
Exercises, 325.
313.
7.
Magnetic moments,
spin,
and
relativistic
effects
1
.
397
Orbital magnetic moments, 327.
netic field, 330.
spin, 334.
4.
The
momentum, 346.
atoms, 353.
3.
2. Effects of
an external mag-
The Stern-Gerlach experiment and
spin-orbit interaction, 338.
6.
Relativistic
Bibliography, 358.
corrections
5.
electron
Total angular
for one-electron
Exercises, 358.
CONTENTS.
12
1.
2.
Quantum mechanical description of identical particles, 360.
Symmetric and antisymmetric eigenfunctions, 363. 3. The ex-
clusion principle, 365.
eigenfunctions, 369.
gas, 381.
4.
Additional properties of antisymmetric
The helium atom, 373.
5.
Bibliography, 390.
13
1.
360
Identical particles
Multi-electron
Introduction, 391.
Hartree theory, 396.
states of atoms, 414.
structure, 449.
10.
The Fermi
atoms
391
The Thomas-Fermi theory, 393. 3. The
The periodic table, 407. 5. Excited
6. Alkali atoms, 418.
1. Atoms with sev-
2.
4.
eral optically active electrons, 425.
coupling, 441.
6.
Exercises, 390.
The Zeeman
12. Transition rates
Lifetimes and line widths, 468.
8.
LS
and
coupling, 428.
443.
effect,
11.
9.
selection rules, 453.
Bibliography, 472.
JJ
Hyperfine
13.
Exercises,
473.
14
1.
X-rays
The discovery
spectra, 476.
spectra, 489.
electric effect
475
of X-rays, 475.
3.
X-ray
5.
The
Measurement of X-ray
4. X-ray continuum
of X-rays, 493.
6. The photo-
scattering
483.
and pair production, 505.
and attenuation coefficients, 509.
particles, 57 J. Bibliography, 577.
15
1.
2.
line spectra,
7.
Total cross sections
8. Positrons
and other
anti-
Exercises, 577.
Collision theory
Introduction, 575.
2.
518
Laboratory-center of mass transforma-
The Born approximation, 524. 4. Some applications of the Born approximation, 530.
5. Partial wave analysis,
534. 6. Some applications of partial wave analysis, 546. 7. Abtions,
579.
3.
sorption, 554.
16
1.
Bibliography, 558.
The nucleus
Introduction, 559.
clear sizes
Exercises, 558.
and the
2.
The composition
optical model, 565.
559
of nuclei, 567.
4.
3.
Nu-
Nuclear masses and
CONTENTS
abundances, 580.
cal
5.
The
mass formula, 589.
drop model and the semi-empiriMagic numbers, 592. 7. The Fermi
liquid
6.
gas model, 594.
8. The shell model, 599.
9. The collective
Alpha decay and fission, 609. 11. Beta decay,
Gamma decay and internal conversion, 640. 13. Prop-
model, 604.
620.
12.
10.
erties of excited states,
655.
Nuclear forces, 674.
16.
14. Nuclear reactions, 662.
15.
Mesons, 694.
Bibliography, 703.
Exercises, 704.
Index
707
Introduction
I.
What
Is
Modern
Physics?
Webster's dictionary gives as a primary definition of the word modem:
"pertaining to the present time, or time not long past." However, when a
physicist speaks of modern physics he is using the word in a somewhat
different way. Modern physics is used to designate certain specific fields
of physics. All these fields have two things in common: first, their
development has taken place since, roughly speaking, the year 1900; and,
second, the theories used to explain the phenomena with which these
fields are concerned are startlingly different from the theories
that were
in existence before 1900.
The term
that contrasts with
modern
physics
is
classical physics.
Classical physics comprises fields in
which the subjects studied are:
Newton's theory of mechanics and the varied phenomena which can be
explained in terms of that theory, Maxwell's theory of electromagnetic
phenomena and its applications, thermodynamics, and the kinetic theory
of gases. In modern physics the subjects studied are: the theory of
relativity and associated phenomena, the quantum theories and quantum
phenomena, and, in particular, the application of the relativity and
quantum theories to the atom and the nucleus.
Many physicists are working today in the fields of classical physics. On
the other hand, the theory of relativity, which dates from 1905, is certainly
not modern in the sense of Webster's definition. We see that the words
modern and classical, as used in physics, do not have quite their normal
temporal significance.
In this book we shall be concerned with modern physics. However,
as a point of departure, we shall first give a very brief description of the
status of classical physics
toward the end of the nineteenth century.
I
INTRODUCTION
Historical Sketch
2.
The astronomical observations of Tycho Brahe and their interpretation
by Kepler, plus the mechanical experiments of Galileo (all of which
happened within a few decades of the year 1600), were welded into an
by Newton (1687). By the end of
was very highly developed. It provided
mechanical phenomena known at that
elegant yet simple theory of mechanics
the nineteenth century this theory
a successful interpretation of all
time and served as a basis for the kinetic theory of gases which, in turn,
removed many mysteries from the theory of thermodynamics.
A great variety of facts concerning electric and magnetic fields and their
interaction were discovered during the nineteenth century.
phenomena were brought
All these
together within one framework by the beautiful
theory of Maxwell (1864). This theory provided an explanation for the
wave propagation of light which was in agreement with what was known
at that time of both geometrical and physical optics.
all the experimental data known at that time could be
Newton's theory of mechanics or the electromagnetic
theory of Maxwell. Physicists were beginning to feel quite self-satisfied,
and it appears that the majority were of the opinion that the work of their
successors would be 'merely to "make measurements to the next decimal
In fact, almost
fitted into either
place."
The turn of
saw the shattering of this tranquil
of quite revolutionary experimental and theoretical
developments, such as the theory of relativity, which demands that we
reject deeply seated intuitive ideas concerning space and time, and the
situation
by a
the present century
series
quantum theories, which make similar demands on our intuitive ideas
about the continuity of nature.
We shall begin our study of modern physics with the theory of relativity,
not because of chronological considerations since the relativity and early
quantum theories were developed concurrently, but because this order of
presentation will
make
the discussion
much more
convenient.
fairly
brief treatment of relativity will be followed
early
we
quantum
theories.
shall study in detail
atom and
the nucleus.
This will
by a similar treatment of the
lead us to quantum mechanics, which
and then apply to a thorough discussion of the
CHAPTER
The Theory
of Relativity
I.
The Galilean Transformation and
Classical Mechanics
In classical physics, the state of any mechanical system at some
time
can be specified by constructing a set of coordinate axes and giving the
coordinates and momenta of the various parts of the system
at that
t
time.
laws
If we know the forces acting between the various
parts, Newton's
make it possible to calculate the state of the system at any future time
t in terms of
a calculation
state at
its
we
t
It is often desirable that during or after such
specify the state of the system in terms of a new set of
.
coordinate axes, which is moving relative to the first set. The
two-fold
question then arises: How do we transform our description of the
system
from the old to the new coordinates, and what happens to the equations
which govern the behavior of the system when we make the transformation ?
The same question often arises when treating electromagnetic systems
with Maxwell's equations.
This is the question with which the theory
of relativity concerns itself.
Let us see what answer classical mechanics would give to this question.
Consider the simplest possible mechanical system one particle of mass
m acted on by a force F.f Using rectangular coordinates to describe the
system, we have the situation shown in figure (1-1).
t Boldface type will be used in this
The same symbol
in
normal type
book for symbols
will
that represent a vector quantity.
then represent the magnitude of the vector
quantity.
THE THEORY OF RELATIVITY
The
set
of four numbers
coordinates
x, y, z.
that
m
m
Fx Fy
,
and
Fz
we know
the particle has
Newton's laws,
F,
d 2x
F
7? = *
d2y
7?
dt
where
us that at time
(x, y, z, t) tells
Assuming
[Ch.
= F
,,
(1-1)
are the x, y,
and
components of the vector F,
motion of the
give us a set of differential equations which govern the
y
axis
y
Figure l-l.
system.
set of rectangular coordinates.
These equations are valid only
if
the frame of reference defined
by the coordinates x, y, z is an inertial frame, i.e., a frame of reference in
which a body not under the influence of forces, and initially at rest, will
remain at rest. If we know the three coordinates and three components
of momenta of the particle at an initial time t we can evaluate the two
arbitrary constants which appear in the general solution of each of the
differential equations (1-1) and can make specific predictions of the location
(and momentum) of the particle at any time t.
Now consider the problem of describing the system from the point of
view of a new frame of reference x', y', and z', which is moving to the right
with constant velocity u with respect to the old frame of reference, and
which maintains the relative orientation between the two frames shown in
figure (1-2). These two frames of reference are said to be in uniform
translation with respect to each other. There are now two sets of four
numbers (x, y, z, t) and (x', y', z', t') that can equally well be used to
specify the location of the particle at any instant of time t and t', which
are the times measured in the two frames of reference and defined such
that t t' =
when the two frames of reference coincide. What is the
,
Sec.
GALILEAN TRANSFORMATIONS
I]
AND MECHANICS
relationship between the sets of
numbers (x,y,z,t) and (x', y', z', /') ?
Newton, almost all physicists before 1900, and most probably the reader,
would certainly not hesitate to assert that
=X
y' = y
z' = z
=t
X
ut
(1-2)
t'
These equations are known as the Galilean transformation equations.
constitute the answer which classical physics gives to the first part
of the question that we have posed for the case of uniform translational
motion between the two frames of reference.
They
axis
axis
y
x
z' axis
z axis
Two
Figure 1-2.
supposed to be
To answer
dV/A"
by
axis
frames of reference
in
uniform translation.
The x and
x' axes are
collinear.
the second part of the question, let us evaluate the
quantity
differentiating the first
We have
dx'jdt
of equations (1-2) with respect to
= dxjdt
t.
and
Since
It is
t'
= d*x/dt*
dVfdt*
= d*xldt*
obvious that
2
d*y'ldt'
It is also true that
F in
<Px'ldt2
t,
d*yldt*
and
Fx = Fx Fy = Fy
.
the direction of the x or x' axis
is
F,
dV/dt"
=F
d*z/dt*
z because the component of
the same as seen from either frame
THE THEORY OF RELATIVITY
of reference
and
similarly for its other
two components.
for the unprimed quantities in equations (1-1),
motion in the primed system:
= Fx
we
[Ch.
Substituting
find the equations of
dt'
d 2 y'
dt'
Note that equations (1-3) have exactly the same mathematical form as
equations (1-1). Thus the answer to the second part of the question is the
very interesting one that Newton's laws, which govern the behavior of
the system, do not change when we make a Galilean transformation. We
have demonstrated
this for the case in
reference with respect to the other
is
which the velocity of one frame of
in the direction of the x or x' axis.
We leave it as an exercise for the reader to
show that these results are true
regardless of the direction of the relative velocity vector u. The x,
y, z
frame was an inertial frame because d2x/dt 2
d 2y\dt 2
d 2z/dt 2
if
From
we see that x', y', z' is also an inertial frame
because d 2x'\dt* = d^'jdt* = dV/dt" =
if F = 0.
The physical content of the result that we have just derived is quite easy
0.
equations (1-3)
to understand and, in fact, is something with which the reader is certainly
already familiar. Since Newton's laws are identical in any two inertial
frames, and since the behavior of a mechanical system is governed by
Newton's laws, it follows that the behavior of all mechanical systems will
be identical in all inertial frames even though they are in uniform translation with respect to each other.
Consider a laboratory which
is
located
With the car initially at rest with respect to the ground
be an inertial frame), let us perform a series of mechanical
in a railroad car.
(assumed to
experiments such as measuring the period of a swinging pendulum or
investigating the subject of collision mechanics by means of a serious game
of billiards. If we were to perform exactly the same experiments while the
car was traveling at constant speed over a very smooth and very straight
stretch of track, the fact that equations (1-1) and (1-3) are identical in
form requires that the experiments yield exactly the same results as we
obtain when the car is at rest with respect to the ground. This
in agreement with our everyday experience.
is
certainly
Thus we see that Newtonian mechanics predicts that all inertial frames
in uniform translation with respect to each other are equivalent in so far
GALILEAN TRANSFORMATIONS
Sec. 2]
AND ELECTROMAGNETICS
as mechanical phenomena are concerned.
corollary of this is: No
mechanical phenomena can be used to differentiate between inertial frames,
so it is impossible by mechanical experiments to prove that one of the
frames is in a state of absolute rest while the others move with respect to
it.
If the
windows were covered in the railroad car moving smoothly with
we would not be able to detect its motion.
respect to the ground,
2.
The
Galilean Transformation and Electromagnetic Theory
Let us next inquire into the behavior of electromagnetic phenomena
Galilean transformation. Electromagnetic phenomena
are discussed in classical physics in terms of a set of differential
equations
which govern their behavior (called Maxwell's equations), just as mechanics
is discussed in terms of the set of differential
equations which govern
mechanical phenomena (called Newton's laws). We shall not actually
when we perform a
carry through the transformation of Maxwell's equations because
the
calculation is quite complicated owing to the nature of these equations.
However, when a Galilean transformation is performed, it is found that
Maxwell's equations do change their mathematical formin sharp contrast to the behavior of Newton's equations. The point
is a very important
one in the development of the theory of relativity, and we must discuss
its
physical significance.
As mentioned in the Introduction, Maxwell's equations predict the
existence of electromagnetic disturbances which propagate through
space
in the characteristic
manner of wave motion, with a propagation velocity
on the frequency of the wave motion (at least in a
that does not depend
medium with constant index of refraction such
many different names have been given to
as the vacuum). Although
these disturbances (radio
~ 10
~ 10
waves for frequency
lO^sec, infrared radiation for frequency
light waves for frequency
10 15 /sec, X-rays for frequency
they are
physicists,
all
basically the
same phenomena.
13
19
/sec,
/sec)^
The nineteenth century
who were very mechanistic in their outlook,
felt quite sure that
the propagation of waves predicted by Maxwell's equations
requires the
existence of a propagation medium. Just as water waves
propagate
through water, so, according to their argument, electromagnetic
waves
must propagate through a medium. This medium was given the name
ether. The ether was obliged to have somewhat strange
properties in order
not to disagree with certain known facts. For instance, the ether must
be
massless since it was observed that electromagnetic waves such
as light
can travel through a vacuum. However, it must have elastic properties
in order to be able to sustain the vibrations which are
inherent in the idea
THE THEORY OF RELATIVITY
of wave motion. Despite these
[Ch.
the concept of the ether
was
which was to assume
that electromagnetic disturbances could be propagated without the aid of
a propagation medium. Existence of the ether being assumed, the
disturbances would certainly be expected to propagate with a fixed velocity
with respect to the ether, just as sound waves propagate with a fixed
considerably
more
difficulties,
attractive than the alternative,
velocity with respect to the air.
It was assumed that the electromagnetic equations, in the form
presented by Maxwell, were valid for the frame of reference which was at
rest
with respect to the ether
these equations leads to
electromagnetic
waves.
the so-called ether frame.
solution of
an evaluation of the propagation velocity of
The result was 3.00 x 10" cm/sec = c, in
agreement within experimental accuracy with the value for the velocity
of light measured by Fizeau (1849). However, in a new frame of reference
in uniform translational motion with respect to the ether, Maxwell's
equations would have a mathematical form different from their mathematical form in the ether frame since, as mentioned above, a Galilean
transformation changes the form of the equations. When the propagation
velocity of electromagnetic waves as seen from the new frame of reference
was evaluated by using Maxwell's equations in the form which they assume
in this frame, it was found to have a value different from c. The reason is
that the propagation velocity predicted by the equations depends on the
form of the equations, as certainly must be the case.
Thus the application of a Galilean transformation to Maxwell's equations leads to the prediction that a measurement of the velocity of fight
made in the new frame of reference would yield a result different from the
velocity c measured in the ether frame. It is not necessary to describe
in detail the predicted relation between the velocity of light in the ether
frame, the velocity of light in the new frame, and the velocity of the new
frame with respect to the ether frame, since it is exactly what would be
obtained from a simple vector addition calculation of the type done near
the beginning of any elementary physics text. That is,
* light wrt.
where
Flight wrt
new frame
" light wrt. ether
* new frame wrt. ether
\^~
The somewhat comether
plicated calculation of the velocity of light as measured in the new frame
of reference, performed by transforming Maxwell's equations and then
solving them in the new frame, is in complete agreement with the very
simple physical idea that light propagates in such a way that its velocity
with respect to the ether frame is always c, and that its velocity with respect
to any other frame can be found by doing a simple vector addition of
velocities. Note that the intuitive arguments made to justify the validity
c,
wrt.
with respect
to.
Sec. 3]
THE MICHELSON-MORLEY EXPERIMENT
of the vector addition of velocities are basically the same arguments which
behind the Galilean transformation.
In recapitulation, the theory of physics near the end of the nineteenth
lie
century was based
upon three fundamental assumptions: (a) the validity
of Newton's laws, (b) the validity of Maxwell's equations, and (c) the
validity of the Galilean transformation. Almost everything which
could
be derived from these assumptions was in good agreement with experiment
in every case in which the appropriate experiment had been performed.
With respect to the questions we have been discussing, these assumptions
predicted that
all frames of reference in uniform translation with
respect
to each other were equivalent as far as mechanical phenomena
were
concerned,
they were inertial frames, but in regard to electromagnetic
equivalent; there was only one frame of reference, the ether frame, in which the velocity of light was equal to c.
if
phenomena they were not
3.
The Michelson-Morley Experiment
In 1887 Michelson and Morley carried out an experiment which proved
to be of extreme importance. This experiment was designed to
demonstrate
the existence of the special frame of reference called the ether
frame, and
to determine the motion of the earth with respect to that frame.
The earth travels in a roughly circular orbit about the sun with a period
(by definition) of one year, and with an orbital velocity of the order of
6
3 x 10 cm/sec. It would seem unrealistic to make the a priori assumption
that the ether frame travels with the earth and, as we shall see later,
strong
arguments against this assumption were known at that time.
more
reasonable working hypothesis would be to assume that the ether frame
was at rest with respect to the center of mass of our solar system, or to
assume that it was at rest with respect to the center of mass of the universe.
In the first case the velocity of the earth with respect to the ether frame
would have a magnitude of the order of 106 cm/sec, and furthermore the
orientation of the velocity vector would change during the course of
a
year. (The velocity of a point on the surface of the earth, owing to its
daily rotation, is negligible compared to the orbital velocity.) In the
second case the velocity of the earth with respect to the ether frame would
be even greater.
The
basic idea of the Michelson-Morley experiment
was to measure the
from a frame of reference fixed with respect to
the earth, in two perpendicular directions. A moment's consideration of
the classical theory, as summarized by equation (1-4), will show that
the
apparent velocity of light with respect to the earth would be different for
velocity of light, as seen
THE THEORY OF RELATIVITY
10
light traveling in the different directions.
[Ch.
At first glance it might be expected
would be observed in such
that the order of magnitude of the effect which
a measurement would be
3
10 6
10
|velocity of earth wrt. ether|
r+tj
|velocity
of light wrt. ether|
10
1(T
where we have assumed that the ether frame is fixed with respect to the
solar system. But, in fact, there exists a theorem due to Fresnel and
Lorentz which says that it is impossible to devise an optical experiment
Figure 1-3.
An
interferometer.
where the velocity of the apparatus with respect to the ether can produce
a first order effect. The biggest possible effect is a second order effect.
This means that an apparatus designed to observe the difference in the
velocity of light traveling in two perpendicular directions must be able to
-4
We shall prove this
theorem for a specific case in the course of our discussion of the MichelsonMorley apparatus.
The numerical estimate which we have just made indicates that an
extremely sensitive apparatus would be required to observe effects due to
the motion of the earth with respect to the ether, and an even more
sensitive apparatus if quantitative measurements were desired. In order
to do the experiment, Michelson invented an interesting device called an
detect differences of the order of (10
interferometer,
shown schematically
10~ 8
in figure (1-3).
THE MICHELSON-MORLEY EXPERIMENT
Sec. 3]
||
Let us briefly consider the operation of this device.
(For more details
the reader should consult any text in physical
optics.)
Assume
that the apparatus
is
Light emitted by
Consider the central ray shown in the
source
at first
at rest with respect to the ether.
is collimated by lens C.
This ray is incident at 45 upon the half-silvered mirror
M. Such
a mirror has the property of reflecting half the light incident
upon
figure.
it
and
transmitting the remainder. Thus the incident ray is split
into ray 1 and
ray 2, which travel to mirrors
x and
2 and are then reflected back on
themselves. Half of each of the two rays returning to
is directed into
the telescope T. The intensity, due to the central ray,
of the light seen by
An
Figure 1-4.
the observer
interferometer moving through the ether.
be a function of the phase relationship of the two
Since the two rays are in phase when originally split
by M,
their relative
lengths
lx
I2
will
recombining rays.
and
phase upon recombining
will
depend only on the
as follows
rays in phase
2*
2.
where X
2.
rays out of phase
the wavelength of the light. Such a device can be used in many
ways; the standard meter is now defined in terms of the wavelength of a certain spectral line by using an interferometer.
Next, let us consider an interferometer moving, as indicated in
figure
(1^), with velocity v with respect to the ether, where v is parallel to the
is
different
direction of ray 2.
beam of
Assume
light is split
by
lx
l2
reflection
/.
In the present case the incident
from a mirror which
is
moving with
THE THEORY OF RELATIVITY
12
[Ch.
Assuming that light travels with a constant velocity
easy to show by a Huygens wavelet construction
that the reflected ray of light will be thrown forward at exactly the right
angle such that, after being reflected from
l5 it will be back at the axis
defined by S and C just in time to meet M. This is not a crucial point
because it would be possible to assume that
is a "rough" mirror which
reflects light diffusely. Figure (1-4) shows the situation as seen by an
respect to the ether.
in the ether frame,
it is
M
M
M' indicates the position of M at the instant
it;
M'2 indicates the position of 2
observer in the ether frame;
the ray reflected from
at the instant
ray
To
it
reflects
meets
2.
evaluate the phase relation of the two combining rays,
we
calculate
the time required for the two rays to travel their respective paths.
seen from the ether frame, ray
ct in
vt.
a time
From
t.
During
the figure
this
we
As
an oblique path of length
time the apparatus moves to the right a distance
travels along
see that
cV =
vh 2
2
I
so
c
2
Since v \c
<
1,
we can
use the binomial theorem and obtain
The
total time for ray
1 is
-"(> +
It
to calculate the time required by ray 2 from the point of view
of an observer fixed with respect to the apparatus. According to the last
of equations (1-2), no difficulties will arise from evaluating time intervals
It is easiest
in different frames of reference.
Equation (1-4) predicts
observer, the velocity of ray 2
is
v while
moving to the left. In each part of the journey
The time for the round trip is then
v while
distance
/.
that, for this
moving to the
=
C
+
V
Using the binomial theorem,
c
this
c\
becomes
-?('*3
2
c /
right,
it
and
a
travels
THE MICHELSON-MORLEY EXPERIMENT
Sec. 3]
13
The two combining rays are out of phase by an amount dt/r, where
T = Xjc = the period of the vibration for light waves of wavelength X.
This phase difference
is
dt_
(t 2
-t )c _2llv*c _lr?
1
c2c 2 X
as
Xc 2
In the actual experiment a measurement was made with the apparatus
shown in the figure, and then a second measurement was made after
the apparatus was rotated through 90.
moment's consideration will
show that the phase difference for the apparatus in the second position is
v2
~~_ ~x7
St'
The effect of rotating the apparatus is consequently expected to be a change
in the phase difference by an amount
_
AA =
dt
St'
[The experiment was done in this
that /x be precisely equal to
/2 ,
as
21 v
=T1
X
(1-6)y
<r
way in order to avoid the requirement
we have assumed in our calculation. It
show that equation (1-6) is true if lx is only approximately equal
even though (1-5) would then be incorrect.]
easy to
is
to
/2
In their experiment, Michelson and Morley used light of
wavelength
6 x 10- 5 cm and a length /
10* cm. If we use our estimate of
2
8
If)" , we find that equation (1-6) gives a value
v*lc
for A of about
0.5. This means that, if the combining rays were just in
phase in
the
first
measurement, they would be just out of phase in the second measurement.
An observer looking through the telescope while the apparatus was being
rotated
would
see brightness replaced
by darkness, or
vice versa.
To
the
extreme surprise of Michelson and Morley, no such effect was
observed.
To make sure that the null effect was not due to a fortuitous combination
of velocity vectors such that their laboratory happened to be stationary
with respect to the ether frame during the time their measurements
were
being made, Michelson and Morley made observations during
various
times of the day and various seasons of the year. An effect
was never
observed. Despite the predictions of the classical theory, the
Michelson-
Morley experiment showed that the velocity of light is the same
when
measured along two perpendicular axes in a reference frame which,
presumably, is moving relative to the ether frame in different directions
at
different times of the year.
THE THEORY OF RELATIVITY
14
[Ch.
These remarkable results captured the attention of the entire community
of physicists. The experiment was repeated with increased accuracy in a
number of laboratories, and it was confirmed that the observed effect was
zero to within the experimental accuracy of about 0.3 percent of the
theoretically predicted value. Many people tried to devise explanations for
the negative result.
Three separate ideas were put forth in an attempt to explain the results
of the Michelson-Morley experiment and yet retain as much as possible of
the physical theories then in existence. These were: the "ether drag"
the "Lorentz contraction" hypothesis, and the so-called
"emission theories."
hypothesis,
The
all
ether drag hypothesis
bodies of
finite
assumed that the ether frame was attached to
mass. The hypothesis was attractive because it would
Michelson-Morley experiment and yet
did not involve modification of the theories of mechanics or electro-
certainly lead to a null result in the
it
magnetism. But there were two reasons why it could not be accepted. In
1853 Fizeau had measured the velocity of light in a column of rapidly
flowing water. If the water were to drag the ether frame with it, the
observed velocity would just be the sum of the velocity of light in stationary
water and the velocity, of the water. Fizeau actually did observe a change in
the velocity of light, but the change
was only a
fraction of the velocity of
the .water and, furthermore, was fully accounted for by the electromagnetic
theory of Maxwell, without introducing the ether drag hypothesis, as due
to the motion of charges in water molecules polarized by the electro-
magnetic field. The second argument against this hypothesis comes from
a phenomenon known in astronomy as stellar aberration. The apparent
positions of stars are observed to
angular diameter.
This
of the earth about the sun; in
causes a vertical
at
move
in circular orbits of very small
a purely kinematical
effect due to the motion
same effect as that which
shower of rain to appear to a moving observer as falling
is
an angle. From
this
would not be present
fact, it is the
analogy
if
it is apparent that the astronomical effect
a light ray were to travel with constant velocity
with respect to the ether frame and
frame were fixed with respect
if that
to the earth.
The Lorentz contraction assumed
to the ether,
all
that,
owing to
their
material bodies contract by a factor of
motion
(1
relative
v 2 jc 2 ) 1A in
the direction of motion. This, of course, is a serious modification of the
concept of a rigid body that is basic to classical mechanics. The reader
would find it worth while to demonstrate for himself that, if an interferometer moving through the ether were distorted in this way, a null
result
would be obtained
in the Michelson-Morley experiment.
for an interferometer in which
lx
2,
However,
a calculation similar to the one
Sec. 4]
EINSTEIN'S
POSTULATES
15
above, but including the Lorentz contraction, predicts a
change in phase
difference
by an amount
when
the velocity of the interferometer with respect to the ether
changes
v'. Such a change in velocity could be
provided by the motion of
the earth in its orbit. The appropriate experiment has
been
from
v to
Kennedy, who observed no
effect.
done by
Thus the Lorentz contraction hypoth-
esis is ruled out.
In an emission theory Maxwell's equations are modified in such
a way
wave remains associated with the velocity of its
source. It is easy to see that this too would lead to a null
result in the
that the velocity of a light
Michelson-Morley experiment, but it must be rejected because it
conflicts
with certain astronomical evidence concerning binary stars. Binary
stars
are pairs of stars which are rotating about their common
center of gravity.
Consider such a pair at a time when one is moving toward the earth
and the
an emission theory were valid, the velocity
of light with respect to the earth from one star would be larger
than that
of light from the other star. This would cause them to appear
to have
very unusual orbits about their center of gravity. However,
the observed
motion of binary stars is completely accounted for by Newtonian
other
is
moving away. Then,
if
mechanics.
None of these
attempts to "patch
A more fundamental change was
4.
up"
the classical theories
was
tenable.
needed.
Einstein's Postulates
All the evidence we have presented seems to be consistent
only with the
conclusion that there is no special frame of reference, the ether
frame, with
the unique property that the velocity of light in the
frame
is
equal' to
c.
Just as for inertial frames
and mechanical phenomena, all frames in
uniform translation with respect to each other seem to be
equivalent in
that the propagation velocity of light, measured in any
direction, is equal
to the same value, c. Einstein (1905) was willing not only
to accept this
evidence but also to generalize on it. In fact, he stated as
a postulate:
The laws of electromagnetic phenomena (including the
fact that the
propagation velocity of light is equal to the constant value c), as
well as the
laws of mechanics, are the same in all inertial frames
of reference, despite
the fact that these frames may be in uniform translation
with respect to
each other. Consequently, all inertial frames are completely
equivalent.
THE THEORY OF RELATIVITY
16
[Ch.
This postulate required that Einstein change either Maxwell's equations
or the Galilean transformation, since the two together imply the contrary
of the postulate. Guided by the failure of the emission theories, he chose
not to modify Maxwell's equations.
This he indicated by stating as a
second postulate:
The
velocity
of light
is
independent of the motion of its source.
Einstein was then forced to modify the Galilean transformations. This
was a very bold move. The intuitive belief in the validity of the Galilean
transformation was so strong that his contemporaries had never seriously
questioned
it.
And
yet, as
we
shall see, the very different transformation
which Einstein proposed in lieu of the Galilean one is based on realistic
physical considerations whereas the Galilean transformation is grossly
unrealistic. Another indication of the boldness of this move is that the
arguments of section 1 show that any modification of the Galilean transformation will require some compensating modification of Newton's
equations in order that the first postulate will continue to be satisfied for
mechanics.
first
5.
We
shall see later
we must develop
the
what
interesting results this leads to, but
new transformation
equations.
Simultaneity
Consider the fourth of the Galilean transformation equations
which is
,
(1-2),
is the same time scale for any two inertial
frames of reference in uniform translation with respect to each other, and
This equation says that there
this is equivalent to saying that there exists
such frames.
Is this
true?
investigate the question of time
Let us for the
a universal time scale for
In order to find out
we must
all
realistically
measurement.
moment concern
ourselves with the problem of defining
a time scale in a single frame of reference. Now the basic process involved
in any time measurement is a measurement of simultaneity. For instance,
common
statement of the type "The train arrived at 7 o'clock,"
an abbreviated notation for what really occurred "The train
arrived and the little hand of a nearby clock pointed to 7 simultaneously."
No further discussion of the question of the simultaneity of events which
occur at essentially the same physical location is necessary. Not so
obvious is the problem of determining the simultaneity of events which
occur at different locations, but this, in fact, is the key problem involved
in the setting up of a time scale for a frame of reference. In order to have
the very
is
actually
SIMULTANEITY
Sec. 5]
17
a time scale valid for a whole frame of reference we must have a number
of clocks distributed throughout the frame so that there will everywhere
be a nearby clock which can be used to measure time in its vicinity. These
clocks must be synchronized
of the clocks
"The
that
is,
we must be
hand of clock
little
two
hand of clock B
able to say of any
and the
little
pointed to 7 simultaneously."
number of methods for
They all involve
the determination of simultaneity suggest
themselves.
two
means of transwould be no problem.
the transmission of signals between
physically separated locations. If we
had
at our disposal a
mitting information with infinite velocity, there
y
s Light
-*J
signals >.
-*-
-L>.
^-Midpoint
Illustrating Einstein's definition of simultaneity.
Figure 1-5.
is the point where the Galilean transformation goes wrong by tacitly
assuming the existence of such a method of synchronization. In fact, no
such method exists. Since we have agreed to be realistic in developing a
This
use a real process to transmit the information.
Electromagnetic waves, such as light signals, seem to be most appropriate
time scale,
for
we must
two reasons
first,
propagation velocity e is the highest known,
second, and more fundamental, since we have
their
a clearly desirable feature
postulated that the propagation velocity of light in any frame of reference
will
be
c, it
our time
Two
seems expedient to use the velocity of
scale.
Thus we
instants of time
light itself in setting
up
are lead to Einstein's definition of simultaneity:
x
and
2,
observed at two points x x and x z
in
particular frame, are simultaneous if light signals simultaneously emitted
from the geometrically measured midpoint between xx and x2 arrive at
x x at
This
is
and
at x % at
t2 .
illustrated in figure (1-5).
An equivalent definition is that
and t2 are simultaneous if light signals
x
emitted at tx from x 1 and at t 2 from 2 arrive at the midpoint simultaneously.
These definitions intimately mix the times tlt t2 and the spatial coordinates
xx x2 In Einstein's theory simultaneity does not have an absolute meaning,
,
?x
independent of the spatial coordinates, as
it
does in the classical theory.
THE THEORY OF RELATIVITY
18
[Ch.
consequence of these definitions is that two events which are simulwhen viewed from one frame of reference are in general not
simultaneous when viewed from a second frame which is moving relative
to the first. This can be seen by considering a simple example. Figure
taneous
from the point of view
(1-6) illustrates the following sequence of events
O who is at rest relative to the ground. This observer has so
charges of dynamite at Q and C2 that the distances OC[ and
of an observer
placed two
OC2 are equal. He can cause them to explode simultaneously by sending
simultaneous light signals to Cx and C2 which actuate detonators. Assume
he does this so that, in his frame of reference, the explosions are simultaneous with the instant that he
n3
is
O'
abreast of O', an observer stationed
Co
c
Figure 1-6.
Two
successive views of a train moving with uniform velocity.
arrows indicate the
on a
c.
The
small
light signals.
which is moving by with uniform velocity v. The explosions
marks C[ and % on the side of the train. After the experiment O'
train
leave
can measure distances 0'C[ and 0'C%. He will find they are equal.f The
explosions also produce flashes of light. Both O and O' record the times
at which they receive the flashes. Observer O will, of course, receive the
flashes simultaneously, confirming that in his frame of reference the
explosions occurred simultaneously. However, O' will receive the flash
which originated at C% before he receives the flash from C[ because the
moves during the finite time required for the light to reach him.
Since the explosions occurred at points equidistant from O' but the light
train
were not received simultaneously, he concludes that the explosions
were not simultaneous.
signals
Such disagreements concerning simultaneity lead to interesting
results.
From the point of view of O, C^C2 = C\C^. However, from the point of
view o0\ c; passed C2 before C[ passed Q. Therefore O' must conclude
that
t
QC <
We
C^C^.
shall see
he must find 0'C{
The
simultaneity problem will also bring the two
n t he ne xt paragraph
that, according to O',
0'C'% due to the homogeneity of space.
0'C[
OCl. However
Sec. 6]
KINEMATICAL EFFECTS OF RELATIVITY
19
observers into disagreement concerning the rates of
clocks in their reAs we shall see, the nature of their disagreements about the measurement of distances and time intervals
is such as to
allow both O and O' to find a value of c for the velocity
of the light pulse
which came from x or C2
spective frames of reference.
6.
Kinematical Effects of Relativity
Einstein's two postulates, plus his definition, uniquely
determine the
equations which must be used to transform the coordinates
and time from
B'-r
O'
Figure 1-7.
Two
measuring rods
in
uniform translation.
one frame of reference to another.
There are several ways of actually
carrying out the derivation required to find these equations.
We shall
use here a procedure which is particularly instructive
because of
its
emphasis on physical concepts.
Comparison of Lengths Oriented Perpendicular to the Direction of
Motion
(I)
O and O' so stationed in reference frames that
O with velocity v. As shown in figure (1-7), O
at the center of parallel measuring rods AB and
Consider two observers
O'
is
moving
relative to
and O' are each located
A'B' which, by superposition, are
rest with respect to each other.
known
to be of identical length when at
pose the question: What are the
relative lengths of the two measuring rods, from the point
of view of either
observer, when they are moving with respect to each other?
As
indicated in the figure,
rods coincide
when
we assume
passing. Let
near the ends of AB,
defined by AB, and
We
that the centers of the measuring
O have two helpers Oa and Ob
stationed
who send light signals when A' and B' cross the line
who also mark the location of the points at which
THE THEORY OF RELATIVITY
20
A' and B' cross the
with the length AB.
line.
It is
Afterwards
[Ch.
can compare the length A'B'
apparent from the symmetry of the arrangement
from Oa and Ob simultaneously. Furthermore, so will O'. Therefore O' must agree that Oa and B made their
measurements simultaneously and, consequently, must accept the results
that
of the measurement. Let us tentatively assume that O and O' conclude
that
will receive the signals
TW < AB.
Next repeat the experiment, but now requiring that the length
comparisons be carried out in the primed frame of reference. Since
th e two
Einstein's first postulate requires complete symmetry between
<
A'B'.
apparent that O and O' must then conclude that AB
consistent
possible
only
The
statement.
previous
This contradicts the
conclusion is that AB = A'B'. Thus we find that the lengths of identical
measuring rods, as seen by any observer, are the same independent of
frames,
it is
whether they are at
rest relative to the observer or
moving
in a direction
perpendicular to the direction of their orientation.
(II)
Comparison of Time
Relative Motion
An observer
compare
O',
Intervals
Measured by Clocks Which Are
in
moving with velocity v relative to observer O, wishes to
measured by his clock with a measurement of the
a time interval
Mirror
Mirror
K~
^<g)
Figure 1-8.
sets of clocks in uniform translation.
made with clocks belonging to O.
have established that, when at rest with respect
same time
they
Two
interval
We
assume that
to each other,
all
is
the clocks involved run at the same rate and are synchronized. It
reading
with
the
compared
be
apparent that the reading of an O' clock can
of an O clock that happens to be momentarily coincident with the former
without any complication, even though they are moving relative to each
clocks in the two
other. Thus measurements of a time interval made with
in figure (1-8).
illustrated
frames can be compared by the procedure
KINEMATICAL EFFECTS OF RELATIVITY
Sec. 6]
21
Observer O' sends a light signal to a mirror which reflects
it back to him.
Both O and 0' record the time of transmission of the signal with
clocks
and C", respectively. The time of reception of the signal is recorded
by
the two observers with clocks C and C". The diagram
on the left shows
2
the sequence of events
from the point of view of O', and the diagram on the
from the point of view of O.
The elapsed time between the two events measured by O' is 2At', where
At' = l'/c, and where /' is the distance to the mirror
as measured in his
frame. The elapsed time measured by O is 2At. From the diagram
it is
right
apparent that
c2
where
/ is
Af2
v 2 At 2
the distance to the mirror as measured
by O. Solving for
At,
we have
At2
_
=
c*-v 2
shows that
At
We
Vl
/'.
=L
c
v /c
At
But argument
v jc
Therefore
1
Vl
v /c
= _^1_
Vl - v 2jc
(1 _7)
see that a time interval
between two events occurring at the same
some frame of reference is longer by a factor l/Vl v 2/c 2
when viewed from a frame moving relative to the first frame and, consequently, in which the two events are spatially separated. The time
place in
interval
measured
place
called a proper time.
(Ill)
is
in the
frame in which the events occurred in the same
The
effect
Comparison of Lengths Oriented
Motion
involved
Parallel to
is
called time dilation.
the Direction of
Consider the experiment we have just described and imagine a measuring
rod placed in the O frame with one end at clock C and the other end at
x
clock
C2
Designate by
which it is at rest.
from the O' frame.
to
L the length of the rod in the O frame, with respect
We want to evaluate L', the length of the rod as seen
In this frame the rod is moving in a direction parallel to its own length.
Since the velocity of O' with respect to O is v, the velocity of O (and also
of the rod) with respect to O' must be precisely
v; otherwise there would
THE THEORY OF RELATIVITY
22
[Ch.
be an inherent asymmetry between the two frames which is not allowed by
Einstein's first postulate. Let t[ be the time at which observer O' sees the
front end of the rod pass, and 4 be the time of passage of the rear end.
Then the elapsed time 2At' = t' - t[ is related to the length of the rod as
2
measured in the O' frame, and to
velocity
its
measured
in that frame,
by
the equation!
L'
2Af
(1-8)
We must also establish an equation connecting the corresponding quantities
as seen
from the
v, travels
O frame. In this frame, C, which is moving with velocity
L in a time 2At. Thus
the distance
L =
From
the last two equations
(1-9)
2At
we obtain
L'
LAt'
At
Argument
II
shows that
At' I At
= Vl -
c/c
Therefore
L'
= LVl -
v*lc
(1-10)
We see that a measuring rod is shorter by a factor Vl
which
v 2 /c 2
compared
observed from a
length. The length of
when
its length in a frame
frame in which it is moving parallel to its own
the rod measured in the frame in which it is at rest is called the proper
The effect is called the Lorentz contraction because equation
length.
like the equation proposed by Lorentz (cf. section 3). In
looks
(1-10)
to
in
it is
at rest,
it is
a great distinction between these equations. In Lorentz's
proposals represented the velocity of a material body with respect to the
fact, there is
ether frame.
In the present equation v represents the relative velocity
between the two arbitrary inertial frames O and O
It is worth while to point out that the technique used by the observer to
measure the length of a moving rod in the experiment just described is
different from that used by the observer in the experiment of argument I.
The latter observer simultaneously marked the locations of the ends of the
rod and then measured the distance between the two marks. However,
'.
very easy to prove that the two techniques are equivalent in that, if
both can be applied to the same situation, they will give the same answer.
it is
must, of course, be
t Velocities measured in any given frame
distances measured in the frame and times measured in the frame.
denned
in terms of
Sec. 6]
(IV)
KINEMATICAL EFFECTS OF RELATIVITY
Synchronization of Clocks
Which Are
in
23
Relative Motion
In section 5 we saw that observers in motion
with respect to each other
disagree concerning the simultaneity of two
events. This will lead to a
disagreement between the two observers concerning
the synchronization
of clocks. Our task here is to determine the
quantitative nature of this
disagreement.
Consider again the experiment discussed in arguments
II and III with
between the two events which were
timed by the clocks in both frames. We called the time
interval measured in
the O frame 2&t, and the time interval measured
in the O' frame 2Af ',
and we already know one relation between these two
quantities. From
equation (1-7) we have
special reference to the time lapse
Vi -
Since this equation expresses the estimate which
would be made in the O
frame of the time interval which was measured in the
O' frame, and since
the measurement was made in the O' frame with
a single clock, no problem
of synchronization arose in its derivation. However,
if we wish to write
an equation giving the estimate which would be made in
the O' frame of the
time interval measured in the O frame, we must
account for the fact that
observer O' would contend that the two clocks in
the
frame were
unsynchronized. But, after taking this into account, there
must be the
same dilation of a time interval as it is transformed from
O to O' as there
was when
it
was transformed from O' to O. Thus we put
2At
0A ,,
VI
tr/c
where 8
is a quantity to be determined which
corrects for synchronization.
This quantity can be determined by requiring that
equations (1-8),
(1-9), and (1-10) be satisfied. Substitute (1-8) and (1-9) into
(1-11). We
then have
From
L'
Ljv
Vl -
3
2
/c
(1-10) this becomes
LVl -
ytjc 2
L/v
(.-4)-i +
V\
CI
+
i
THE THEORY OF RELATIVITY
24
So the value of 8
is
[Ch.
given by
Lv^_ _Lv
V c
The minus
sign
ahead of the
means that
first
clock
(1-12)
to observer O' the second clock
C2
appears
Cv
We see again that two clocks fixed in a certain frame with a separation
L, and synchronized in that frame, appear out of synchronization when
viewed from a frame moving relative to the clocks.
It is interesting
to note
from synchronization is proportional to the separation
of the two clocks. Even more interesting is that the sign of the effect
changes as the sign of v changes, since v enters linearly. Thus two observers
that the departure
in different frames investigating the
synchronization of each other's
Not only would each observer
contend that the other observer's clocks were unsynchronized, but the
two observers would even disagree on the sign of the effect. This should
be compared with the effects described by equations (1-7) and (1-10),
in which v enters as a squared quantity. Although in these cases there is
clocks
would be
in total disagreement.
also a disagreement
between the two observers, there
Each observer believes
measuring rod contracts and time scale dilates.
in their disagreement.
7.
is
a certain symmetry
that the other observer's
The Lorentz Transformation
We
are
now
only a short step from our goal of deriving the coordinate
transformation of the theory of relativity.
Consider two observers
and O' who view the same event while moving with
velocity v with respect
to each other. Relative to their respective coordinate systems, the location
and time of the event are specified by the set of numbers (x, y, z, t) or
(x', y', z', t').
The desired coordinate transformation consists of the
equations which allow us to calculate the primed set of numbers in terms
of the unprimed set, or vice versa.
Assume that the orientation of the two coordinate systems, and of the
vector v specifying the velocity of O' with respect to O, is as shown in
figure (1-9). We define the zeros of the t and the t' scales such that clocks
at the origins of the two coordinate systems read / = t' =
when the
origins are coincident.
The separation of the two origins at some later time is vt, as seen by O
by O'). However, the distance x' measured in the O'
(or vt' as seen
frame appears contracted by the factor
Thus
would say that the
Vl
tP/c 2
distance, parallel to v,
when viewed by O.
from
his yz plane to the
Sec. 7]
THE LORENTZ TRANSFORMATION
position of the event was vt + x'Vl - v 2 \c 2
x coordinate of the event. Consequently
x
Solving for
x',
we
vt
x'Vl
25
This
is,
by
definition, the
v 2/c 2
find
Vl - v 2lc2
x
vt
^=^^=^=^
Now O believes that the time
at the point x', y',
z',
t' read on the clock fixed in the O' frame,
must be corrected by an amount 6' = +x'v/c 2
in order to account for the lack of synchronization
axis
between that clock and
y' axis
(x,y,z,t)
(x',y',z',t')
vt (as seen by
O)
->~x axis
Figure 1-9.
Illustrating the
*-*' axis
argument leading to the Lorentz transformation.
the clock fixed at the origin of the O' frame. (It would
be instructive for
the reader to convince himself that in the present
case the quantity d'
positive.)
Furthermore, time intervals measured by the clock at the
O' origin appear to O to be dilated by the factor 1/Vl
v*/c 2
Thus,
according to O, the time t of the occurrence of the event is
is
+ x'v[c 2
Vl - v 2jc
t'
Substituting for x
from the equation above and solving for
xv/c
Vl - v jc
of section 6
tells
we
find
Argument
t',
us that y'
y and
z'
z,
since these
THE THEORY OF RELATIVITY
26
distances
are measured in directions perpendicular
together our results,
to
v.
[Ch.
Gathering
we have
vt
Vi - v /c
X
y'
z'
=
=
(1-13)
z
t
xv/c
Vl -
v jc
These equations constitute the so-called Lorentz transformation which,
according to Einstein, must be used for the transformation of the coordinates and time of an event from the inertial frame O into the inertial
frame O'.f
An algebraic solution of equations (1-13) for x, y, z, and t produces
another form of the Lorentz transformation which is useful in transforming from the primed frame to the unprimed frame. These are
+ vt'
Vi- v /c
x'
y'
(1-13')
z'
+ x'v/c
Vl - v /c
f
Equations (1-13') are identical with (1-13) except that v has been replaced
v. The minus sign is merely a consequence of the fact that, for (1-13'),
still
v
means the velocity of O' with respect to O. If we were to redefine
by
v to
mean
the velocity of
with respect to O', the resulting equations
would be completely identical with (1-13). We could have guessed the
form of (1-13') from the form of (1-13) without actually doing the algebra,
since the observed symmetry between the two is required by Einstein's
first postulate. The postulate demands that all inertial reference frames in
uniform translation be equivalent. Consequently the transformation
which we have derived from this postulate must be the same in going from
O to O' as it is in going from O' to O.
It is' immediately apparent that the Lorentz transformation reduces to
the Galilean transformation
when
vie
<
This makes good physical
t Equations of exactly this form, but with v representing a velocity relative to the
ether,
some
had been proposed in connection with a
work of Einstein.
years before the
classical theory of electrons
by Lorentz
TRANSFORMATION OF VELOCITY
Sec. 8]
27
The fundamental difference between the two transformations is
we realize that the signals, which are needed to synchro-
sense.
that in the former
nize the clocks of a reference frame, propagate
with the finite velocity c,
while in the latter we naively assume the velocity
of these signals to be
infinite. However, if we are required to
make a coordinate transformation
in a situation in which the relative velocity
v is very small compared to the
velocity of the synchronizing signals, it should
not make any difference
whether we take the latter velocity to be c or oo.
Consequently in this
limit the two transformations should merge.
On
the other hand, significant discrepancies arise
between the preof the Galilean transformation and the rigorously
correct
dictions
Lorentz transformation when v
had not been observed
is
in classical
comparable to c. These discrepancies
mechanics because experiments had not
been performed at such high
velocities. It is interesting to note that for
equations (1-13) are meaningless, in that real coordinates
and
times are transformed into imaginary ones. Thus
c appears to play the
role of a limiting velocity for all physical
phenomena.
v/c>
We
shall attain
a better understanding of this as we continue our
discussion.
To close this section, let us check the internal consistency of the Lorentz
transformation with the postulates from which it was
derived. Consider
again figure (1-9), and imagine that at the instant =
=
t
a spark gap located at the
As viewed by
common
origin of the
we
t'
discharge
two frames of reference.
observer O, the resulting light pulse will travel out
in
directions with the velocity
c.
Thus, at any instant
form a spherical wave front of radius r
t,
The wave
ct.
all
the light pulse will
front will
obey the
equation
**
z2
or
x*
/2
a2
c2 ? 2
(1
_ 14)
From the view point of O', according to Einstein's postulates, the
light
pulse must also travel in all directions with the
velocity c. Thus O' must
also see a spherical wave front
*'
y*
+ z' - cY =
2
(1_15)
despite the fact that he
it
as
is moving with velocity v relative
to O. We leave
an exercise for the reader to apply equations (1-13')
to (1-14) and
obtain (1-15), thus giving the desired demonstration
of consistency.
8.
Transformation of Velocity
Consider the particle shown in figure (1-10), moving
with velocity
as seen in a frame of reference O.
would like to evaluate the velocity
We
THE THEORY OF RELATIVITY
28
[Ch.
of the particle, as seen in the frame of reference O' which
moving relative to O with velocity v.
Measured in the O frame, the velocity vector of the
is itself
particle has
com-
ponents
Vx =
The same
Vy =
dxjdt,
dzjdt
measured in the O' frame, has components
velocity vector, as
V'x
V =
dy/dt,
V'y
dx'jdt',
dy'jdt',
V'
dz'/dt'
/
Particle
Figure 1-10.
moving
observed from two frames of reference
particle
in
uniform
translation.
Now from equations
we know that the
(1-13)
and the unprimed coordinates and times
relation between the
primed
is
(x
vt)
V1-/3 2
=
=
y'
z'
y
z
1
vx\
(^
where we have introduced the convenient notation ^ = v/c. Take the
differential of these equations, remembering that v is a constant. This gives
dx'
(dx
Vl-
dy'
dz'
i
dt'
=
=
v dt)
dy
dz
1
=
Vi
j
A/ dt
2V
i
dx\
1
TRANSFORMATION OF VELOCITY
Sec. 8]
29
Then
1
/j
(ax
V __= Vl -
j.\
v at)
dx
v
K.-
dr
ft
J=(dt - fl\
Vi-^
1
-M i_^ i-v
dy
n=
Vl
dt
<fy/rff
/?\
V!
flt
K,Vl
rfr/
ft
"
(1-16)
and, similarly,
dz
V',=
dz/dt
Vl-/S 2 ^
Vl-/32 V
__
vji -
1--.V.
c* dtl
These equations tell us how to transform the observed velocity from one
frame of reference to another frame of reference.
0.9
<
0.9 c
*-
Figure l-l
Illustrating an
example of the addition of
velocities.
First we note that, as Vjc and v\c approach zero,
equations (1-16)
approach those which would be derived from the Galilean transformation.
Another very interesting property of these equations is that it is impossible
to choose V and v such that V, the magnitude of the
velocity which
is
seen in the
new
in figure (1-1
of positive
To
x,
1).
frame,
As
is
greater than
c.
seen by O, particle
Consider the example illustrated
has velocity 0.9c in the direction
and
particle 2 has velocity 0.9c in the negative x direction.
evaluate the velocity of particle 1 with respect to particle 2
we transform
from the
velocity v
O frame
to the O' frame moving in the negative x direction
with
-0.9c, using the first of equations (1-16). We obtain
V{
0.9c
(-0.9c)
1.80c
<
THE THEORY OF RELATIVITY
30
not
It is
[Ch.
a rigorous proof of this property by considering
form of equations (1-16). The velocity transformation
difficult to give
the mathematical
equations demonstrate another aspect of the fact that in the theory of
a limiting velocity for all physical phenomena.
relativity c acts as
9.
Relativistic
Mechanics
It has been emphasized that Einstein's modification of the transformation
equations would, in turn, necessitate a compensating modification in the
laws of mechanics, so that these laws continue to satisfy the requirement
of not changing form under the influence of a coordinate transformation^
In this section we develop the laws of the new mechanics, which is called
relativistic
mechanics.
obviously desirable to carry over into relativistic mechanics as
of classical mechanics as possible. We shall see that it is possible to
It is
much
preserve the classical definition of
where p
and
is
the
momentum,
momentum,
= mu
(1-17)
and u the velocity of a particle,
law of conservation of momentum,
the mass,
also to preserve the classical
2*
Ip, _ initial
(1-18)
final
we are willing to modify the classical concept of mass. It will
be necessary to allow the mass of a particle to be a function of the magnitude of its velocity, i.e.,
providing
m=
m(u)
(1-19)'
The form of this function is to be determined. However, we know a priori
that we must have m(u) -> m as u -- 0, where m is a constant equal to
,
the classically measured mass of the particle. This is true since, as u -> 0,
the Lorentz transformation approaches the Galilean transformation and
no modification
in
mechanics
is
necessary.
In order to evaluate the function m(u),
As seen from the xyz frame
and 2 are moving in directions
ment.
(?!
we
consider the following experi-
indicated in figure (1-12), observers
parallel to the x axis with equal and
These observers have identical balls Br and B2 each
as measured when they are at rest. While passing, each observer
opposite velocities.
of mass
t Just the opposite is true for the laws governing electromagnetic
well's equations), since Einstein's
phenomena (Max-
second postulate prohibits modification of these laws.
RELATIVISTIC
Sec. 9]
MECHANICS
throws his ball toward the other with a velocity which, from
his own point
of view, is directed perpendicular to the x axis and is of
magnitude V.
In the xyz frame, Bx and B will be seen to approach with equal
2
velocity
along parallel lines making angles 6 =
with
x
the
axis,
and
rebound
X
2<
.
at angles 6
and
Assuming conservation of momentum and that the
collision is elastic, it is easy to show that 8
= 2/ and that the magnitude
lf
of the velocity of the balls is the same after the collision
as before. The
actual value of
or 2/ depends on the impact parameter d, which we
lf
assume to be such that 6 =
as shown in the figure.
X
Xf
lf
2/
(Out of paper)
Figure 1-12.
collision
between two
balls of identical rest
mass.
Now look at the process from the point of view of Ox as shown in figure
Observer Ox throws Bx along a line parallel to his y axis with
,
(1-13).
velocity
V.
It
returns along the
same
with velocity of the same
maintain a constant x velocity
relative velocity of
2 with respect to O x
The component of velocity of B2 along his y axis is observed by O to
x
change sign during the collision but to maintain a constant magnitude.
magnitude but opposite sign.
which is just equal to v, the
To
He
sees
line
B%
we realize that the y component of the velocity
Then we transform this to the Ox frame with the
evaluate this magnitude
of B2 as seen by
,
2 , is V.
aid of the second of equations (1-16) and obtain vVl
component of the velocity of B2 as seen by O x
The y momenta of both Bx and B2 as measured in the
v 2jc 2 for the
Ox
frame, simply
change sign during the collision. Consequently the total y momentum of
the system consisting of the two balls changes sign. But the law
of
conservation of momentum says that the total y momentum of the system
before the collision must equal the total y momentum of the system after
the collision. This can only be true if the total
y momentum, as seen by
THE THEORY OF RELATIVITY
32
x , is
zero both before and after the collision.
Evaluating the
[Ch.
momenta
according to the definition (1-17), and equating the total y momentum to
zero in order to satisfy the conservation of momentum (1-18), we obtain
an equation which
is
constant equal to
obviously self-contradictory
.
However,
if
we
allow
if
we
insist that
be a
to be a function of the
velocity as in equation (1-19), the equation reads
Vm(V)= VVl -
v*lc
m(V V\\ -
v 2 jc 2 )
v2)
or
m{V)
= Vl -
iP/c 2
m(V V\l -
where the argument of the function
Figure 1-13.
Acollison between two
balls
on the
v^/c
2
)
v2)
right side of the equations
of identical rest mass, as seen by observer
t.
simply the magnitude of the velocity of B2 as seen by O v This equation
perfectly satisfactory and, in fact, enables us to evaluate the function m
immediately if we take V-*-0. We then have
is
is
m(0)
= m = Vl -
m(v)
=w
2
t>
/c
m(v)
so
/Vl
v 2 jc 2
(1-20)
[Evaluation of the functional form (1-20) from the equation for m(V),
by taking V-*-0, constitutes a short-cut but not an approximation. It
is easy to show that (1-20) satisfies the equation for m(V) exactly.] Thus
a consistent theory of relativistic mechanics demands that the mass of
a particle moving with velocity v be larger than its mass when at rest by a
factor
v /c The mass m(v) is called the relativistic mass of the
mQ is called its rest mass.
quite high velocity v = 0.1c, the relativistic mass is only 0.5
1/Vl
and
particle,
For the
percent greater than the rest mass.
However, with increasing
v the rest
Sec. 9]
MECHANICS
RELATIVISTIC
33
mass rapidly becomes larger since m(v) ->ooasu^e. It is apparent
that the velocity of a particle cannot exceed c if (1-20) is valid.
Continuing our discussion of relativistic mechanics, let us consider a
particle of rest
magnitude
mass
but accelerated by a force of
initially at rest,
F which is applied in
the particle in the distance x
f
the x direction.
Jo
In order to evaluate this
We
assume that
total
work done on
tf
T = )'f dx = P'f dt =
law.
The
is
Jo
dt
we must know
Fv
dt
Jo
the relativistic
form of Newton's
it is
rF
dP
_
=
= d (mv)J
(1-21)
dt
dt
Note that, if m = constant, this is just F = m dvjdt = ma. Actually,
Newton's statement of the law was that of equation (1-21) and
not
F - ma. Inserting (1-21) into the equation for T, we have
T=
Jo
Integrating
by
parts,
we
dt
dt
vdp
Jo
obtain
T=
H? - ['p
dv
Jo
Substituting for
v dv as d(v*)j2,
p from
equations (1-17) and (1-20), and also expressing
we have
T =
Vf
mv
_
2 z
Lvi
v /c J
mo |V
2 Jo
The second term
is
a standard form.
T= m
v /c
U/l
d(v
2
)
Vl -
Integrating
it,
v /c
we have
+ Vl -
v /c
v /c
This reduces to
T= m
c''
-Vl
v /c
Evaluating at the limits and then dropping the subscript /to simplify
our
notation,
we have
T=
Vl
^2
c"
2
v /c 2
m nc'
(1-22)
THE THEORY OF RELATIVITY
34
[Ch.
Now the theorem of conservation of energy implies that the total work
done on the particle should equal its kinetic energy. Thus we would like
T
to call
To check
the kinetic energy of the particle.
this, let v/c
->
0.
Then
T = m
>+-\
V*
= -- m v*
m oCc -
2 v
*
T-^ z
,
x
1
This agrees with the classical expression for kinetic energy and confirms
our identification of T.
Continuing with our interpretation of equation (1-22), we observe that
a function of v which can be written as the difference between a term
depending on v and a constant term, as follows
T is
T(v)
where E(v)
E(v) for v
must
=w
= 0,
i.e.,
also be energies
when
when
its
its
velocity
v jc 2
E(0)
=m
/V
E(v)
= mc
2
,
(0)
and
also
Since
is
where E(0) is the value of
an energy, E(v) and E(0)
E(v) being some energy associated with the particle
and (0) some energy associated with the particle
zero. To identify the meaning of these energies,
is v,
velocity
is
rewrite the equation as
E(v)
The conclusion
is
inescapable.
T(v)
E(0)
We must interpret
of the particle moving with velocity
E(v) as the total energy
the sum of the kinetic
energy T(v) of the particle and an intrinsic energy (0) associated with the
particle when it is at rest. The energy E(v) is called the total relativistic
energy,
and E(0)
is
called the rest
v,
since
it is
mass energy.
We
have established Einstein's well-known relation between mass and
energy: The rest mass energy of a particle is equal to c 2 times its rest mass.
That is,
(0)
and the
That is,
=m
c2
(1-23)
total relativistic energy is equal to c 2 times its relativistic mass.
E(v)
Equation (1-22)
kinetic energy,
tells
and
= mc
(1-24)
us that the relation between total relativistic energy,
rest
mass energy
mc 2
is
= T+ m
c2
(1-22')
RELATIVISTIC
Sec. 9]
MECHANICS
35
For situations in which the particle also has a potential energy V,
possible to extend the argument to show that
mc 2
= T + V+ m
c2
This emphasizes the fact that the relativistic mass
(1-22")
is
a direct measure of the
total energy content of the particle.
content will increase
It is
its relativistic
it is
Anything that increases
mass.
its
energy
sometimes convenient to have a relation similar to (1-22') which
momentum p. It can be obtained by evaluating
explicitly involves the
the quantity
m 2c4 where
/5
vjc.
This
mfc*
is
m v - m y = my
= my
^
1
/r
1
.1
1
2
^1
mW = cY
/3
Thus
=w
2 4
= c/+ m
2
2 4
(1-25)
The relativistic properties of mass and energy, and the relation between
the two, were derived from the Lorentz transformation and the additional
assumptions we made concerning the new form of the laws of mechanics.
These additional assumptions seemed reasonable since they were obviously
designed to preserve as
much
of classical mechanics as possible. However,
can be found only by comparing the predictions
of relativistic mechanics with experiment. We reserve this
comparison for
section 1 1, but perhaps it is appropriate to point out here that
the existence
of a rest mass energy m c 2 is not in conflict with classical physics.
Since the
experiments of classical physics are all such that the total'rest mass involved
their ultimate justification
constant, the appropriate rest
mass energies could be added to both sides
energy balance equations without destroying their validity.
The relativistic theory of energy is, however, of more than academic
interest. One important reason is that there are
processes in nature in
is
of
all
which the total rest mass of an isolated system does not remain constant.
For such processes experimental evidence shows that the change in rest
mass energy is exactly compensated by a change in kinetic, or potential,
energy in a
This
is
way
that conserves the total relativistic energy of the system.
some very practical processes for the conversion of rest
the basis of
mass energy into other forms of energy, as in a nuclear reactor. It
also
shows that in our relativistic theory we must replace the separate classical
laws of conservation of mass and conservation of energy
by a single
comprehensive law of conservation of total relativistic energy:
36
THE THEORY OF RELATIVITY
As observed from a given frame of reference,^
of an isolated system remains constant.
10.
Transformation of
Momentum
[Ch.
the total relativistic energy
and Energy
In certain problems, particularly those involving collision mechanics,
convenient to transform the momentum and total relativistic energy of
it is
a particle from one frame of reference to another. In this section
the appropriate transformation.
Consider a particle of
rest
mass
positive x axis of reference frame O.
of
moving with
In that frame
it
we derive
V along the
have components
velocity
will
momentum
V,
and
total relativistic
Pv
o,
Pz
energy
Wgc 2
E=
Vl - V 2/c 2
Next consider an observer O' who is also moving along the positive x axis
of O, but with velocity v. According to this observer, the particle would
have components of momentum
p'
and
m v
'
p'y
Vi -
F'V
p'z
0,
total relativistic energy
'
where
To
is
we
m Qc 2
Vl - F'V
the velocity of the particle with respect to O'.
establish the relation
frames,
between
momentum and
energy in the two
evaluate the quantities
V'/Vl
Vic*
and
c 2 /Vl
V'
jc
The same restriction applies, of course, to the energy conservation law of classical
For example, the total energy content of a particle will increase if we transform
classically from a system in which the particle is at rest into a system in which it is
moving and consequently has kinetic energy. See section 10.
t
physics.
Sec. II]
EXPERIMENTAL VERIFICATION OF THE THEORY
37
in terms of V. This is
(1-16),
done by using the velocity transformation equation
which in the present case reads
V~
V'=
1
Vvjc 2
The
calculation will not be reproduced here.
forward but lengthy. The result is
'
vE/c
Vi - v
Px
Pz
= Py
= Pz
E'
- E~
Py
It is perfectly straight-
Vl
(1-26)
PxV
2
v /c
Although in the derivation we assumed the velocity of the particle to be
parallel to the x axis, it can be shown that equations (1-26) are valid for
any orientation of its velocity vector.
II.
Experimental Verification of the Theory
The theory of
was designed to agree with the experimental
same in frames of
reference which are in uniform translation with respect to each other.
However, in addition to achieving this, the theory predicts a number of
new phenomena, such as length contraction, time dilation, relativistic
increase in mass, and a relation between mass and energy. This is typical
of a scientific theory. Also typical is the fact that the initial acceptance of
the theory was only tentative even though it appeared to be based upon
relativity
fact that the velocity of light is observed to be the
correct logic.
The theory did not achieve full
new phenomena had been put
concerning the
status until
its
predictions
to the test of experiment.
To
close our discussion of the theory of relativity we shall briefly mention
some of the experiments which confirm its predictions.
The first of these was performed in 1909 by Bucherer. It consisted of a
measurement of the masses of high velocity electrons, using a fairly direct
technique that will be discussed in Chapter
3. Bucherer's results are shown
some more extensive results obtained in
recent years are shown by dots, and the predictions of equation (1-20)
are shown by the solid line. Note that these results prove not only that
by the
crosses in figure (1-14),
equation (1-20) has the correct functional form, but also that the velocity
THE THEORY OF RELATIVITY
38
c,
which
essentially enters the theory
of
3.00
is
equal to the velocity of
10 10 cm/sec.
length contraction
relativity as the limiting velocity
for the transmission of information, actually
light,
[Ch.
number of experimental verifications of
and time dilation are now known. One of the clearest
of these involves measuring the lifetime of unstable particles called mesons,
mesons of various velocities. It is observed that the lifetime of a
for
moving meson
rapidly
dilated in excellent agreement with equation
is
(1-7).
These experiments
chapter
we
"
.,
1.8
1.7
1.6
1.5
be mentioned in Chapter 16. In that
an overwhelming amount of evidence proving
will
shall also present
with c
3.00
x 10
cm/sec
-g-1.4
1.3
1.2
1.1
1.0
0.3
0.4
0.5
0.6
v/c
Figure 1-14.
I.
An experimental
verification of the
0.7
0.8
0.9
dependence of mass on
velocity.
From
Kaplan, Nuclear Physics, 1955, Addison Wesley, Reading, Mass.
the validity of equations (1-23) and (1-24), Einstein's relations between
mass and energy. The predictions of the theory of relativity have been
confirmed on every point, and there is now universal agreement on its
validity.
subject that might logically be discussed at this point
is
the so-called
general theory of relativity, which considers frames of reference that are
accelerating with respect to each other. It is found that there are very
relations between acceleration and gravitational effects.
However, the subject is such that any pretense at a serious treatment
must be given in terms of tensor calculus, and therefore we shall pass it
over. This will not be a handicap because the general theory has had very
little interaction with the other branches of modern physics, and also
because it is possible, by using various subterfuges, to employ the special
theory that we have developed in certain problems involving accelerated
motion. For instance, in Chapter 1 1 we shall make such a calculation for
a case of uniform circular motion.
interesting
Ch.
EXERCISES
I]
39
BIBLIOGRAPHY
Bergmann,
wood
P. G., Introduction to the
Theory of Relativity, Prentice-Hall Enele'
Cliffs, N.J.,
1942.
Leighton, R. B., Principles of Modern Physics,
McGraw-Hill
Book
Co.,
New
York, 1959.
Lorentz, H. A., A. Einstein, H. Minkowski, and
H. Weyl, The Principle of
Relativity, Dover Publications, New York, 1951.
EXERCISES
Write equations analogous to (1-2) for the case in
which the relative
is not parallel to any of the axes,
and use these to show that
equations (1-3) still obtain.
1.
velocity vector u
2.
Derive equation (1-6').
3.
The
years.
distance to the farthest star in our galaxy
Explain
why
it is
this star within his lifetime,
4.
Apply equations
5.
In frame O, particle
velocity
v.
is
possible, in principle, for a
and estimate the required
(1-13') to (1-14)
1
is
at rest
and obtain
and
of the order of 10 5 light
being to travel to
human
velocity.
(1-15).
moving to the right with
moving to the right
the two particles appear in O'
particle 2 is
Now consider a frame O' which,
relative to O, is
with velocity u. Find the value of u such that
to be approaching each other with equal and
opposite velocities.
6.
Verify that the form (1-20) for m{v) satisfies exactly
the equation for miV).
7.
Carry through the calculation leading to (1-26).
In the laboratory (LAB) frame, particle 1 is at rest
with total relativistic
energy Ex and particle 2 is moving to the right with
total relativistic energy
8.
and momentum p 2
Show
masses of the system
is
that the frame in which the center of the
relativistic
at rest is moving to the right with velocity
c-
+E
LAB frame, and show that the total momentum of the system is
zero in this center of mass (CM) frame. Now let
the two particles have the same
rest mass m , and let the total kinetic energy
of the system in the LAB frame be
TLAB Evaluate Tcu the total kinetic energy of the system
relative to the
in the
and show that
rCM
CM
frame
V2m n c* V Tj
in the extreme relativistic limit where T
m ^. What might be the practical
1AB
consequences of this in the study of high energy nuclear
>
reactions ?
EXERCISES
40
9.
[Ch.
Consider a collision in the LAB frame between the moving particle 2 and
Show that
1, both of which have the same rest mass m
the stationary particle
the angle between the velocity vectors of the two particles after the collision is
always 90 in classical mechanics, and show how this result is modified by
relativistic
mechanics.
Hint: Transform to the
then transform back to the
An
LAB
CM
frame, treat the collision,
frame.
-4.8 x 10~ 10
x 10~ 28 gm), initially
held at rest at one plate of a plane parallel condenser, is released and allowed
to fall through vacuum to the other plate under the influence of the electric field
in the condenser. The separation between the plates is 102 cm, and the voltage
10.
electron (charge
across the plates
is
esu,
mass
10 7 volts (10 7 /300 stat-volts).
9.1
Calculate the time required,
in the frame of reference of the condenser, for the electron to travel between
its plates.
Discuss the justification of using the theory developed in this chapter
Hint: The magnitude of
and the magnitude of the component of the electric field
to the direction of motion, are both unchanged by a Lorentz trans-
for the present system, which involves accelerations.
the electric charge,
parallel
formation.
CHAPTER
Thermal Radiation
and the Origin of
Quantum Theory
I.
Introduction
In the course of a successful attempt at resolving certain discrepancies
between the observed energy spectrum of thermal radiation and the
predictions of the classical theory, Planck was led to the idea that a system
executing simple harmonic oscillations only can have energies which are
integral multiples of a certain finite
amount of energy
(1901).
closely
was later applied by Einstein in explaining the photoelectric
effect (1905), and by Bohr in a theory which predicted with great
accuracy
many of the complex features of atomic spectra (1913). The work of these
related idea
three physicists, plus subsequent developments
and Heisenberg
(ca.
1925), constitutes
by de Broglie, Schroedinger,
what is known as the quantum
This theory and the theory of relativity together comprise the two
of modern physics. In this chapter we begin our
study of quantum theory with a discussion of thermal radiation.
The nature of the subject is such that any discussion of thermal radiation
must be based upon the results of advanced treatments of several different
fields of classical physics: thermodynamics, statistical mechanics,
and
theory.
most
significant features
electromagnetic theory. In
and
justify
them with
many
cases
we
shall simply quote these results
arguments since their proofs are most
appropriately left to the texts which treat these subjects. However, we
shall reproduce in detail that step in which Planck introduced the
quantization of energy, and we shall be able to see exactly how its introduction
resolved a serious conflict between experiment and the classical theory.
qualitative
41
THERMAL RADIATION AND QUANTUM THEORY
42
2.
[Ch. 2
Electromagnetic Radiation by Accelerated Charges
The surface of every body at a temperature greater than absolute zero
emits energy, in the form of thermal radiation, due to the motion of electric
charges near the surface. This radiation consists of electromagnetic
waves (often called infrared waves) of exactly the same nature as visible
but of longer wavelength. As we shall be concerned on several
occasions in this book with the emission of electromagnetic radiation by a
light,
moving charge, a description of the classical theory of the process is
appropriate at this point. The description will be qualitative, but some of
the quantitative results of the theory will be presented.
In elementary electrostatics
an electric
Coulomb's law
charge q
is
shown that surrounding every
it is
field specified
by the vector E. These are
stationary
related
^(-:)
by
<->
where r is the vector from the charge to the point at which E is evaluated,
and where (r//-) is a unit vector in that direction.! It is often convenient to
represent the electric field in terms of lines offorce which are constructed
to be everywhere parallel to the local direction of E and of density (number
of
lines crossing
local value of E.
cm2
area normal to the direction of E) equal to the
two dimensional representation of the lines of force
surrounding a stationary charge is shown in figure (2-1). This
time independent) electric field contains stored energy. In
static
energy stored per unit volume,
the equation
E by
(i.e.,
p, the
is
related to the local value of
fact,
=E
/Stt
(2-2)
the field is in a medium of unity permeability such as the vacuum. This
can be verified with little difficulty by considering the case of a parallel
plate condenser in which E is everywhere constant, and equating the
energy required to charge the condenser to the energy stored in the electric
if
field.
The energy stored in the field of a stationary charge is static and is not
away in the form of electromagnetic radiation; if this were not
so, the conservation of energy would be clearly violated. It is also true
that the energy stored in the field of a charge moving with uniform velocity
is not radiated away but moves along with the charge, even though there
radiated
t With a few exceptions that are explicitly indicated, in this book we shall write all
electromagnetic equations in the cgs-Gaussian system of units. In this system charges
are expressed in esu in all equations.
RADIATION BY ACCELERATED CHARGES
Sec. 2]
43
are associated with the charge a non-static electric
field E and also,
according to Ampere's law, a non-static magnetic field H. It is
easy to
see that this must be the case by mentally transforming
to a frame of
reference in which the charge
is stationary, realizing that it cannot radiate
and then applying the relativistic requirement that the
behavior of the charge including whether or not it radiates
cannot
depend upon the frame of reference from which it is viewed. The total
energy stored in the field of a uniformly moving charge is larger
than if it
were stationary because, when there is also a magnetic field H, electromagnetic theory shows that the energy density is
in that frame,
(2-2')
o7T
for unity permeability.
by the
The
excess energy
forces initially producing the
Figure 2-1.
The
lines of force
For a charge undergoing
is supplied from the work done
motion of the charge.
surrounding a stationary charge.
acceleration,
magnetic
fields
instant
suffered a constant acceleration
the non-static electric
and
cannot adjust themselves in such a way that none of the
stored energy is radiated, as they can in the case
of a charge moving with
uniform velocity. We can see this qualitatively by considering
the behavior
of the electric field. In figure (2-2) we describe this
field by drawing some
of the lines of force surrounding a charge which was
at rest at the initial
/,
a to the
right during the interval
t to t', and then continued moving
with constant final velocity. The figure
shows the lines of force at some later instant t", as viewed
from the frame
of reference moving with the final velocity, and
with the restriction that
all times are measured in that frame.
At small distances the lines of force
are directed radially outward from the present
position of the charge.
At
large distances they are directed radially
outward from the initial
position of the charge. The reason is simply that
information concerning
the position of the charge cannot be transmitted
to distant locations with
infinite velocity, but only with the velocity
c. As a consequence, there are
THERMAL RADIATION AND QUANTUM THEORY
44
[Ch. 2
breaks in the lines of force contained between a sphere centered on the
position and of radius c(t"
t), which is the minimum distance
initial
at
which the
field
can "know" the acceleration
started,
centered on the present position and of radius c{t"
t'),
and a sphere
which is the
minimum
distance at which the field can "know" that the acceleration
As t" increases, the region containing the breaks expands outward
with velocity c. The electric field in this region has components which are
stopped.
both longitudinal and transverse to the direction of expansion, but by
constructing diagrams for several values of t" it is easy to see that the
longitudinal component dies out very rapidly, and can soon be ignored,
Figure 2-2.
The
lines
offeree surrounding an accelerated charge. Only
some of the
lines
are shown.
whereas the transverse component dies out slowly. In fact, electromagnetic
theory shows, by calculations based upon the same idea as our qualitative
discussion, that at large distances
(large t") the transverse electric field
bx
from the region of the acceleration
obeys the equation
_ qa sin 6
=
i
(2-3)
c r
= c(t" t) is the magnitude of the vector
from the region at which the acceleration a took place to the point at
which the transverse field is evaluated, and where is the angle between
r and a. The dependence of E on 6 and r can be seen in our diagrams,
x
and it is clear from our discussion that E must be proportional to q and a.
L
The electromagnetic theory also shows that because of the acceleration
for unity permeability, where r
r
RADIATION BY ACCELERATED CHARGES
Sec. 2]
there will be a transverse magnetic field
that at large distances
45
HL moving along with E
and
from the region of acceleration
TI
Hi.
^sin 8
= qa
i
2
(2-3')
c r
for unity permeability, where r
c(t"
t).
These two transverse fields
propagating outward with velocity c form the electromagnetic radiation
emitted by the accelerated charge. For this, or any other, electromagnetic
radiation propagating in a
medium of unity
EX =
HL
and the energy density of the radiation
Figure 2-3.
Illustrating
permeability,
= {El +
(2-2') is
E\)
the energy transport
in
(2-4)
a parallel stream of radiation.
However, equations (2-3) and (2-3') are valid only if the dimensions
of
the region in which the acceleration took place are small
compared to c
times the period of its duration, as we have assumed.
The
intensity / of the emitted radiation
carries per second across a
cm2
is defined as the energy which it
area normal to the direction of pro-
pagation.
in
Consider figure (2-3), which shows the radiation propagating
from the region of acceleration
forms an essentially parallel stream. The radiation is moving with
some
that
it
direction at such a large distance
velocity c cm/sec.
In one second it moves a distance c cm. Consequently
the energy contained in the parallelepiped with a 1 cm 2
base normal to
the direction of propagation, and with length /
c cm, will be carried
across the base in 1 sec. The energy contained in the
parallelepiped is
equal to the energy density p times the volume c cm3 Thus
the intensity
all
pc
(2-5)
THERMAL RADIATION AND QUANTUM THEORY
46
Using equations (2-3) and
(2-4),
I(6)
we have
E!c
[Ch. 2
= tEl^l
(2 _ 6)
where we indicate explicitly that the intensity is a function of the direction
of emission. Note that 1(6) obeys the familiar inverse square law. The
total
energy radiated in
all
directions per second, R, can be evaluated
integrating 1(6) over the area of a sphere of arbitrary radius
R=
\l(0)
= \(6)2^ sin
dA
r.
That
by
is,
d6
which
yields
*=
lV
(2-7)
for the rate of radiation of energy.
Only some of the qualitative ideas of
be used in the next few chapters; the equations will be
used toward the end of the book.
this section will
3.
Emission and Absorption of Radiation by Surfaces
Let us return now to the phenomenon of thermal radiation. As we
have said, every body at a temperature greater than absolute zero emits
In the classical theory, this can be pictured as the result of
the accelerations of electric charges near the surface due to thermal
this radiation.
agitation.
Now,
in a single acceleration lasting for a certain period
of
most of the emitted radiation has a frequency approximately equal
to the reciprocal of the period, and a wavelength equal to c times the
period. But, in the many different acceleration processes which give rise
to the thermal radiation, a whole spectrum of wavelengths is emitted.
On the basis of this picture it would be expected that the rate of emission
of energy, integrated over the entire wavelength spectrum, would increase
as the temperature of the surface increases because of the enhanced
thermal agitation, and it would also be expected that the rate of emission of
energy would be proportional to the area of the surface. This is found to
be the case; an empirical equation due to Stefan (1879) states
time,
IT
aeT*
(2-8)
The quantity IT is the total energy emitted in all frequencies per second
cm2 from a surface at absolute temperature T; e is a constant called
the emissivity which ranges from
to 1 depending on the nature of the
per
Sec. 4]
BLACK BODY RADIATION
emitting
surface;
and
47
Stefan-Boltzmann constant.
0.567 x 10" 4 erg-cm^-deg-^-sec- 1 is the
The energy emitted as thermal radiation is
supplied by the thermal agitation energy.
We shall also be concerned with the absorption of thermal radiation by
a surface. In this process, energy is removed from incident
thermal radiation,
through
its
agitation energy.
action
There
on electric charges, and ends up as thermal
an interesting relation between the efficiency
is
of a surface as an emitter of thermal radiation, as measured
by e, and its
an absorber. The latter property is measured by a constant
called the absorptivity a that is denned as the ratio
of the total thermal
energy absorbed by the surface to the total thermal energy
incident
efficiency as
upon
it.
The
(1895),
relation between e
is
and
a,
which
is
a theorem due to Kirchhoff
simply
e=
(2_9)
This was proved in a thermodynamic argument,
independent of any
detailed assumptions about the emission and
absorption processes,
by
considering the thermal equilibrium of surfaces of different
nature which
are exchanging thermal energy by emission and
absorption. It has also
been verified experimentally.
4.
Black Body Radiation
Consider a body with a surface that has the property of
absorbing all
upon it, i.e., a = 1. Such a body is called a black
body. A number of objects found in nature are quite
good approximations
to a black body. One example would be any object
coated with a diffuse
layer of black pigment; another example will be
given shortly. Since the
absorptivity of a black body is equal to unity, it follows
from Kirchhoff
the radiation incident
's
law that
absorber,
its
is
emissivity
also the
to equation (2-8),
is also.
most
we
black body, which
efficient radiator.
is
the
most
efficient
Applying the condition e
see that the radiant
power emitted per cm2 is the
the same temperature. This strongly
same for all black bodies which are at
suggests that the other properties of the thermal
radiation emitted by
black bodies, such as the spectral distribution of the
radiation, also depend
only on the temperature of the black body but not
on its detailed nature.
It
can be proved by thermodynamic arguments, involving
two black bodies
in equilibrium, that this is indeed the case. It is
evident that the universal
properties of the thermal radiation emitted by black
bodies makes them of
particular theoretical interest.
The
spectral distribution of black
body
radiation
is
specified
by the
THERMAL RADIATION AND QUANTUM THEORY
48
quantity IT {X), which
is
so denned that IT (X) dX
[Ch. 2
equal to the energy
is
emitted per second, in radiation of wavelength in the interval X to X
from
cm2
of a surface at temperature T.
Pringsheim are associated with the
quantity (1899).
dX,
The names Lummer and
earliest accurate
measurements of
this
The measurements were made with an instrument
prism spectrometers used in measuring optical
had to be used in order to have the
transparent to the long wavelength thermal radiation.
essentially similar to the
spectra, except that special materials
lenses, prisms, etc.,
10
1750"K
r=1500K
r=1250K
T=1000K
12
X
Figure 2-4.
The
(in units of
spectral distribution of black
4
-4
10
cm)
body radiation
at several temperatures.
The observed dependence of IT (X) on X and Tis indicated in figure (2-4).
The arrow on the abscissa indicates the wavelength at which the eye has its
maximum
response (green
light).
IT (X) increases with increasing T. The
integral of IT (X) over all X is, of course, just equal to the quantity IT
previously defined. This integral, which is equal to the area under the
curves, does increase with the fourth power of T, in agreement with
We
see that, for
equation (2-8).
any fixed
X,
Figure (2-4) also shows that the spectrum
shifts
toward
Sec. 5]
WIEN'S
LAW
49
shorter wavelengths as
increases.
quantitative inspection of the
figure will demonstrate the validity of
the equation
^max c 1/r
where Xmax
is
the X at which IT {X) has
its
(2-10)
maximum value for a particular
T.
All these results are in agreement with the
everyday experience that
bodies emit more heat as their temperature
increases, and that with
increasing temperature their "color" shifts from
dull red to blue white
(i.e., increasingly more radiant energy
is emitted in the region of short
wavelengths).
Now consider an object containing a cavity which is connected to the
outside by a small hole. Radiation incident upon the
hole from the outside
enters the cavity and is reflected back and forth by the
walls of the cavity,
eventually being absorbed on these walls. If the area of
the hole is very
small compared with the area of the inner surface of the cavity,
a negligible
amount of the incident radiation will be reflected back to the hole. Then all
the radiation incident upon the hole is absorbed. For the hole,
a = 1 and
must have the properties of the surface of a black body.
Next assume that the walls of the cavity are uniformly heated to
a
therefore the hole
The walls will emit thermal radiation which will fill the
small fraction of this radiation incident from the inside upon
temperature T.
cavity.
The
the hole will pass through the hole.
Thus the hole will act as an emitter of
thermal radiation. Since the hole must have the properties of the
surface
of a black body, the radiation emitted by the hole must have
a black body
spectrum. But, as the hole is merely sampling the thermal radiation present
inside the cavity,
it is
clear that the radiation in the cavity
a black body spectrum.
In fact,
it
will
must also have
have a black body spectrum
characteristic of the temperature T of the walls since this is
the only
temperature defined for the system. The spectrum emitted by the hole in
the cavity is specified in terms of an energy flux I (X), but it is
convenient
T
to specify the spectrum of the radiation inside the cavity in
terms of an
energy density p T (X) which is defined such that
p T (X) dX is the energy contained in 1 cm3 of the cavity in the wavelength interval X to X
dX. It is
apparent from the discussion above that
Pt (X) is proportional to IT (X),
with a proportionality constant that does not depend on X or T. This statement, as well as all the other statements of the last two paragraphs, can,
of course, be proved rigorously.
5.
Wien's Law
Many of the properties of black body radiation can be understood from
the classical theory of thermodynamics. In 1884 Boltzmann produced a
THERMAL RADIATION AND QUANTUM THEORY
50
theoretical derivation of the equation giving IT
per
the total energy emitted
cm2
per second, for a black body radiator (equation 2-8 for the case
1).
In this derivation he considered a cavity with reflecting walls in
the form of a cylinder with a movable piston and
radiation at temperature T. Thermal radiation, in
electromagnetic radiation, can be
to
[Ch. 2
its
energy density.f Taking
this
shown
filled
common
with thermal
with
all
other
to exert a pressure proportional
system through a cycle of expansion and
,x-*<>x
X\
/
/
\\
/
/
N*
xrFigure 2-5.
I449K;
An
experimental verification of Wien's
0,T =
Modern
Introduction to
compression,
From
I259K.
K.
F.
Physics, 5th ed.,
called
in
Richtmyer,
E.
law.
#, T
I646K;
H. Kennard, and T.
McGraw-Hill Book Co.,
New
x,T =
Lauritsen,
York, 1955.
thermodynamics a Carnot cycle, Boltzmann
work done by the pressure of the radiation,
obtained a relation between the
and its temperature. This relation leads to the desired result since the
pressure can be expressed in terms of the energy density, which can in turn
be expressed in terms of IT
.
In the process of expansion or compression of such a cavity filled with
radiation, the wavelength of any spectral component of the radiation will
be changed as the result of a Doppler
piston.
From
shift
upon reflection from the moving
fact, Wien (1893) was able
a detailed consideration of this
to derive a general functional
body radiation known
form
for the spectral distribution of black
as Wien's law:
MA)=^P
where /(A T)
is
(2-11)
some function of the product of the wavelength and the
is not specified by Wien's derivation.
This
temperature, whose form
equation
is
in excellent
seen from figure (2-5).
t In this respect
it
agreement with the experimental data, as can be
This figure plots A 5 times p T (X) as a function of
behaves just
like a gas.
Sec. 6]
THE RAYLEIGH-JEANS THEORY
51
XT, using 1259, 1449, and 1646 data of
Lummer and Pringsheim for
IT (X) [which is proportional to
p T (X)]. We see that all sets of data fall
on the same smooth curve, confirming the prediction
of equation (2-11)
that X h p T (X) is equal to a universal function
of the variable XT.
It is
apparent that Wien's law agrees with the empirical
fact expressed by
J
equation (2-10).
6.
The Rayleigh-Jeans Theory
Wien's thermodynamical derivation performed the useful
role of allowall black body spectra measured at
different temperatures to be discussed theoretically in terms of the single function/pr).
But the derivation
did not evaluate the form of this function. It is
typical of a thermodynamical argument to show that certain relations must
obtain between the
variables which describe some physical system,
but not to provide a
complete theory of the behavior of the system. The
reason is that such
arguments are based on general principles that apply to all
physical systems,
but do not involve the details of the composition of
the particular system
in question. A theory which would be able
to evaluate the function
f{XT) must take into account some of the detailed properties
of a black
body.
ing
At
the turn of the present century several attempts
were made to formusuch a theory. In one of them the behavior of accelerating
electric
charges in the walls of a black body cavity was
considered. These
charges were assumed to be the source of the black
body radiation. It
was also assumed that each charge undergoes simple
harmonic oscillation,
of some particular frequency, about its equilibrium
late
position. According
to electromagnetic theory, each charge then
radiates at a frequency
precisely equal to its particular oscillation frequency.
In thermal equilibrium the energy of a particular frequency, or wavelength,
component of
the black body radiation must be proportional
to the average energy of the
corresponding charged oscillator because the radiation and
the oscillator
are constantly exchanging energy. From
detailed considerations of the
absorption and emission properties of a charged oscillator,
the proportion-
constant was evaluated. It contains the square of
the frequency.
Next, certain results of the theory of statistical
mechanics (to be described
later) were used to evaluate the average
energy of the charged oscillators
in terms of the temperature of the walls in which
they are located. From
ality
these results the spectral distribution of the black
body radiation,
the function f(XT), is immediately obtained.
therefore
pursue
this theory further.
We
Instead,
we
shall turn to the theory
and
not
of Rayleigh
shall
THERMAL RADIATION AND QUANTUM THEORY
52
[Ch. 2
and Jeans which
because
leads to the same results, and which has a sounder basis
does not require such a detailed picture of the origin of the black
it
body radiation. In fact, it has been shown that the form off(XT) obtained
by Rayleigh and Jeans is a necessary consequence of the theories of
classical physics.
Consider a cavity with metallic walls at temperature T. The walls emit
thermal electromagnetic radiation. In thermal equilibrium this radiation
has a black body spectrum characteristic of the temperature T. Furthermore, as we shall show, in the steady state attained at equilibrium the
electromagnetic radiation inside the cavity must exist in the form of
standing waves with nodes at the metallic surfaces.
Rayleigh and Jeans
number of standing waves with nodes at the surfaces of the
cavity in the wavelength interval A to 1 + dX. They then calculated the
average energy of these waves. The energy depends on the temperature T
and is evaluated from the theory of statistical mechanics. The number of
calculated the
standing waves in the wavelength interval times the average energy of the
waves divided by the volume of the cavity is equal to the average energy
per cm3 in the wavelength interval X to X + dX, which is just p T (X). From
this f(XT) is evaluated by using equation (2-11).
Let us assume that the metallic walled cavity filled with electromagnetic
radiation is in the form of a perfect cube of edge length a, as shown in
figure (2-6). Then the radiation reflecting back and forth between the
walls can be analyzed into three components along the three mutually
perpendicular directions defined by the edges of the cavity. As the opposing walls are exactly parallel to each other, the three components of the
radiation do not mix, and we may treat them separately. Consider the
x component and the metallic wall at x = 0. All the radiation of this
component which is incident upon the wall is reflected by it, and the
incident and reflected waves combine to form a standing wave. Now,
since electromagnetic radiation is a transverse vibration with the electric
field
vector
perpendicular to the propagation direction, and since the
propagation direction for
question,
an
its
is
this
component
parallel to the wall.
electric field parallel to its surface,
is
perpendicular to the wall in
But a metallic wall cannot support
as currents can always flow in such
as to neutralize the electric field. Therefore E for this component
must always vanish at the wall that is, the standing wave associated with
the x component of the radiation must have a node at x = 0. The standing
wave must also have a node at x = a because there can be no parallel
way
corresponding wall. Furthermore, similar conditions
apply to the other two components the standing wave associated with the
and y
a, and the standing wave
y component must have nodes at y
associated with the z component must have nodes at z
and z a.
electric field in the
THE RAYLEIGH-JEANS THEORY
Sec. 6]
53
These conditions put a limitation on the
possible wavelengths of the
electromagnetic radiation contained in the
cavity.
To determine this limitation, consider radiation of
wavelength X and
frequency v
c\X, propagating in the direction
defined by the three
The radiation must be a standing
wave since all three of its components are standing
waves. We have
indicated the locations of some of the fixed
nodes of this standing wave
by a set of planes perpendicular to the propagation
direction a, y. The
angles
a,
/?,
y, as
Figure 2-6.
shown
in figure (2-7).
cubical cavity containing electromagnetic
radiation.
distance between these nodal planes of the
radiation is just 1/2, where X
wavelength.
have also indicated the locations at the three axes
of the nodes of the three components. The distances
between these nodes
are
is its
We
X^ __
2 cos a
2 cos p
2 cos y
(2-12)
THERMAL RADIATION AND QUANTUM THEORY
54
[Ch. 2
Let us write expressions for the magnitudes at the three axes of the
of the three components. They are
electric
fields
E(x,
i)
E(y,
t)
E(z,
t)
= A sin (Inx/X^ sin {lirvi)
= B sin (2irylXy) sin (2nvt)
= C sin (2ttz/A ) sin (2m>t)
(2-13)
for the x component represents a wave with a maximum
amplitude A, with a spatial variation sin (2Trx/Xx), and which is oscillating
The expression
z
,
T
C\J
CM
Mil
,<
feft
Hji
^X
c)
*-\ xl2-+ ^A*/2-*
Figure 2-7.
The nodal
planes of a standing
wave propagating
in a certain direction in
a cubical cavity.
the
0, 1, 2, 3,
with frequency v. As sin (2irxlXx) vanishes for 2x\X x
separfixed
nodes
has
because
it
X
wavelength
wave
of
standing
wave is a
x
ated by the distance Ax
XJ2. The expressions for the y and
components
represent standing waves of maximum amplitudes B and C and wavelengths X y and X but all three component standing waves oscillate with
the frequency v of the radiation.
Note that these expressions automatically
requirement that the x component have a node at x = 0, the
0, and the z component have a node at
y component have a node at y
z
0. To make them also satisfy the requirement that the x component
satisfy the
THE RAYLEIGH-JEANS THEORY
Sec. 6]
55
have a node at x = a, the y component have
a node at y
component have a node at z = a, set
= nx
= nv
=
2*/4
1y\K
2/ l z
.
ycosa =
and the
=a
for y = a
for z = a
2,
W!C ,
a,
(2-14)
C os
,8
f"
^,
0, 1, 2, 3
Squaring both sides of these equations and adding,
we obtain
2
(2a/A) (cos 2
But the angles
a,
/S,
cos 2 p
cos 2 y)
nx
y have the property
cos 2 a
cos 2
,5
cos 2
Thus
2a/A
= Vn 2 +
where n., , n z take on all possible integral values.
This equation describes
the limitation on the possible wavelengths
of the electromagnetic radiation
contained in the cavity.
It is convenient to continue
the discussion in terms of the allowed
frequencies instead of the allowed wavelengths.
They are
= =
A
la
^ nl +
"**
"*
Now we shall count the number of allowed frequencies in a given
interval
by constructing a uniform cubic
(2
"15
frequency
one octant of a
rectangular coordinate system in such a way
that the three coordinates of
each point of the lattice are equal to a possible
set of the three integers
nx
lattice in
n y n z See figure (2-8). By construction,
each lattice point corresponds to an allowed frequency. Furthermore,
N(v) dv, the number of
allowed frequencies between v and v + dv, is
equal to N(r) dr, the number
of points contained between concentric
shells of radii r and r + dr
where
,
'
From
equation (2-15) this
= Vn 2 +
n2
is
for
where nx = 0, 1, 2, 3,
= 0, 1, 2, 3,
Using equations (2-12) these conditions become
.
2a
(2-16)
THERMAL RADIATION AND QUANTUM THEORY
56
Since N(r) dr
is
equal to the volume enclosed by the shells times the
density of lattice points, and since,
N(r) dr
is
Figure 2-8.
by construction, the density
is
unity,
simply
N(r) dr
frequencies
[Ch. 2
4jrr
dr
The rectangular coordinate system used in counting the number
Only a few of the points are shown.
of allowed
in a cubical cavity.
Setting this equal to N(v) dv,
and evaluating
r2
dr from equation (2-16),
we have
v)dv=^y dv
N{
This completes the calculation except that
by a factor of 2 because,
we must
multiply these results
we have
enumerated, there are actually two independent waves corresponding to
the two possible states of polarization of electromagnetic radiation.! Thus
for each of the allowed frequencies
t The fact that electromagnetic radiation in the visible range (light) can have either
of two independent circular polarizations is discussed in almost any elementary physics
text.
THE BOLTZMANN PROBABILITY DISTRIBUTION
Sec. 7]
the
number of allowed
57
frequencies in the frequency interval v to
v
N(v) dv
dv
is
(2 -17)
can be shown that N(v) dv is independent
of the assumed shape of the
and depends only on its volume, a3
It
cavity
The Boltzmann Probability Distribution
7.
The next step in the Rayleigh- Jeans calculation
the average energy contained in each
standing
is
the evaluation of
wave of frequency v
the particular energy of some wave
can have
According to classical physics,
any value from zero to infinity; the actual
value is proportional to the
square of its average amplitude (cf.
equation 2-4). But, if we have a
system containing a large number of
physical entities of the same kind
which are
thermal equilibrium with each other at a
temperature T such
as the system of standing waves in
equilibrium in
the black body c'avity
the classical theory of statistical
mechanics demands that the energies of
these entities be distributed according
to a definite probability distribution
whose form
is
specified
by
T.
As
the average energy
determined by the
depends on T
It is not difficult to see the
essential ideas underlying these results
of
statistical mechanics, if a sufficiently
simple system is considered. Imagine
a system consisting of entities of the
same kind which contain energy An
example would be a set of identical coil
springs which are vibrating
Assume that the system is isolated so that
its total energy
probability distribution,
it
must have a
is
definite value that
content
is
and assume also that the entities can exchange
energy with each
other through some mechanism so that
the system is in thermal equilibrium
Finally, assume that there are only
four of these
constant,
entities in the
that the total energy of
system
any entity is restricted to the values s =
Ae
2A 3A 4Ae, 5A
and that the total energy of the system
(which
now must be some integral multiple of As) has
the value 3Ae The last
assumptions are made strictly for the purpose
of simplifying the calculation
we are about to perform. After the calculation
we shall imagine letting
As approach zero so that e may have any
value, but at the same time
,
keeping the
energy of the system constant. We
shall also imagine
entities in the system become large.
Since the entities can exchange energy
with each other, all possible
divisions of the total energy 3As
between the four entities can occur The
only possible divisions are indicated
in the diagrams of figure (2-9)
Consider the diagram labeled i
1.
This corresponds to divisions of
lettmg the
total
number of
THERMAL RADIATION AND QUANTUM THEORY
58
total energy in
which
for three entities
But
can be the one in the energy
and e = 3Ae for the
any one of the four
there are actually four such divisions, since
state e
This
3Ae.
is
[Ch. 2
fourth.
entities
indicated in the
column marked "number of duplicate divisions." The same is true for the
divisions of the type = 3. The reader should verify for himself that there
/'
are twelve duplicate divisions of the type
any rearrangement of
entities
In evaluating this number,
2.
different energy states is to
co
CO
CO
CO
co
o c
o
<
to
to
40/20
24/20
P'(S)
Figure 2-9.
between
Illustrating
to
be
a>
^a;
ro
8.2
%'S.
4/20
12/20
12
4/20
12/20
a simple
4/20
calculation
0/20
leading to the
Boltzmann probability
distribution.
counted as a new division since the
entities
could presumably be
dis-
tinguished experimentally when in different energy states even though they
same kind, but any rearrangement of entities in the same energy
not counted because entities in the same energy state could not be
are of the
state is
distinguished.!
We now make the final assumption:
All possible divisions
of the total energy occur with the same probability. Then the probability
that the divisions of a given type will occur is proportional to the number
of duplicate divisions of that type, and the relative probability Pt is just
equal to that number divided by the total number of duplicate divisions.
The
are evaluated in the
'.
The
total
= 4x3x2x1.
state
let
us calculate
number of rearrangements (permutations) of the four entities is
But rearrangements of the two entities within the same energy
do not count, so this number must be divided by 2
t Hint:
column marked "Pt ." Next
THE BOLTZMANN PROBABILITY DISTRIBUTION
Sec. 7]
P'(e), the probability
0.
and the
For
of finding an entity in the energy state
divisions of the type
relative probability
59
s.
Consider
there are three entities in this state,
that these divisions occur is 4/20; for /
2
/
two entities, and Pt is 12/20; for i = 3 there is one entity, and
4/20. Thus P'(0), the probability of finding an entity in this
state,
there are
P
3
is
4/20
lated
12/20
m the same way
= 40/20.
4/20
2A
Figure 2-10.
is
The
values of P'(a) calcufor the other values of s are given in the row marked
1
3A
4A
comparison of the results of a simple calculation and the Boltzmann
probability distribution.
They are also plotted as points in figure (2-10).
the exponential function
"P'(e)."
is
P(e)
where
and
are constants.
adjusted to give the best
fit
Ae- t/
The
solid line
(2-18)
(In the figure the constants have been
to the points representing the results of our
calculation.)
Now we must imagine letting Ae go to zeroand at the same time
changing the condition on the total energy to read that it is equal to
Ae, with
increasing as rapidly as Ae decreases, so that the total energy
remains constant. The result of this step is that the calculated function
becomes defined for values of e which are closer and closer together.
In the limit, the energy e becomes a continuous variable as classical
physics
P'(e)
demands
must, and the probability distribution P'(e) becomes a conFinally, when the number of entities in the system is
allowed to become large, this function is found to be identical with the
P(e) of equation (2-18).
it
tinuous function.
The quantity P(e)
of
its
specifies the probability
energy states between e and s
of finding an entity in one
de for a system containing a large
THERMAL RADIATION AND QUANTUM THEORY
60
[Ch. 2
number of entities of the same kind in thermal equilibrium with each
other. The probability is simply P(e) de. It also specifies the average
energy of one of the
which
entities,
is
eP(e) de
*
Jo
(2-19)
P(e) de
I
The integrand
in the
numerator
the energy e weighted by the probability
is
The denominator
that an entity will have this energy .f
is
the probability
of finding an entity with any energy; it is just the total number of
in the system. Evaluating P(e) from equation (2-18),
P(e)
we
Ae- Zl
Ae-* e
entities
l/e
find
Aee'^de
-
ee~
Jo
de
Jo
Ae~ m de
e-" de
Jo
Jo
Now
X
'de
da.
da,
("e-de
da.
Jo
Jo
e
_a
f
Jq_
de
Jo
Jo
Thus
= -
In
da,
Evaluating the integral,
e~" de
Jo
we have
d
da
-1
da.
da.
a.
_2
1
a.
Therefore
(2-20)
By
treating a system containing
two
different kinds of entities in equili-
can be proved that for a given temperature the constant e
must be independent of the nature of the entities of which the system is
comprised. Consider then the system containing a set of identical coil
brium,
it
t We assume that the number of energy states in the interval e to e + de is independent
of e. The assumption is valid for a system of simple harmonic oscillators to which we
shall apply the equation.
THE BOLTZMANN PROBABILITY DISTRIBUTION
Sec. 7]
springs which can exchange energy
by some process or
are in equilibrium at temperature T.
Assume
other,
61
and which
that the lengths of the
springs can vibrate about their quiescent lengths, but that they cannot
otherwise move.
Then one "coordinate" per
spring
(its
length)
is sufficient
to describe the system at any instant,
and the system is said to have one
degree of freedom per entity. Now it is shown in almost any elementary
physics text that, for a system of gas molecules moving in thermal
equilibrium at temperature T, the average kinetic energy per degree of
-1
Boltztnanris
freedom is k T/2, where k
1.38 x 10~ 16 erg-deg
is
the law of equipartition of energy, and it is true for any
classical system. Thus, for our classical system of springs, the average
This
constant.
is
kinetic energy of vibration of
one of the springs
is
k T/2.
But, for a vibrat-
ing spring, or any entity in which the single "coordinate" executes simple
harmonic
energy.
oscillations, the average total
The reader may
easily
prove
energy
this for
one of the springs of our system, the average
e
With equations (2-20) and
(2-21),
twice the average kinetic
total
energy
e is
= kT
we
(2-21)
find
= kT
eQ
This determines the value of
is
a pendulum. Therefore, for
(2-22)
and equation (2-18) becomes
P(e)
Ae- elkT
(2-23)
is the famous Boltzmann probability distribution.
These equations should apply directly to a system of standing waves in
equilibrium in a black body cavity because each of the entities of this
system also has a single "coordinate" (its amplitude) which executes
This
simple harmonic oscillations.f In the Rayleigh- Jeans calculation for this
system
we
are interested only in equation (2-21).
It
could have been
written directly from the law of equipartition of energy, but later
we
shall
need (2-23) and the calculation preceding (2-20).
t Equations (2-22) and (2-23) also apply to a system of moving gas molecules
however, equation (2-21) is replaced by e = f&Tas the average total energy equals the
average kinetic energy, but there are 3 degrees of freedom. There is no inconsistency
here since equation (2-20) is also changed. It is replaced by s = ie because a factor
proportional to e^ must be included in the integrands of equation (2-19) to account
for the fact that the number of energy states in the interval e to s + fife is proportional to
e^ (cf equation 12-37). The product of this factor and P(e) gives the Maxwellian energy
.
distribution of gas molecules,
Ae ^e~ slkT
,
THERMAL RADIATION AND QUANTUM THEORY
62
8.
[Ch. 2
Comparison with Experiment
The energy per cm3 in the frequency interval v to r + dv of the black
body spectrum of a cavity at temperature T is just the product of the
average energy per standing wave (equation 2-21) and the number of
standing waves in the frequency interval (equation 2-17) divided by the
volume of the
cavity.
That
is,
p T {v) dv
Snv 2 kT
dv
(2-24)
cr
To make
a comparison with our earlier discussion,
the energy density in terms of the wavelength
X.
we must
express
This can be done by
using the relation
v
c/X
Thus
dv=-j- dX
(2-25)
Consider a certain interval of the spectrum, which can be defined in terms
of frequency as v to v + dv, or equally well in terms of wavelength as
X to X + dX. For this interval the following relation obtains
PT (X) dX
= -p T (v) dv
(2-26)
because both sides are equal to the energy content per cm3 in that interval
of the spectrum. [The minus sign compensates for the fact that, according
to equation (2-25), dv and dX are of the opposite sign.] Combining
equations (2-25) and (2-26),
we have
PT(X)dX
Using
this to
We
is
p T (v)j-2 dX
transform equation (2-24) to the variable
Pt {X) dX
This
&nk
X
(2-27)
X,
it
find
XT dX
(2-28)
the Rayleigh- Jeans spectral distribution of black
note that
we
body
radiation.
has the form required by Wien's law (equation 2-11)
it
can be written p T (X) dX =f(XT)/X5 where f(XT) = ZnkXT.
However, as is seen in figure (2-11), the Rayleigh- Jeans spectrum is in
since
PLANCK'S THEORY
Sec. 9]
63
complete disagreement with the experimental data;
form
it
predicts the
wrong
for the function /(AT).
In the limit of long wavelengths, the Rayleigh- Jeans spectrum approaches
the experimental
p T(X) -*
and
is
But, as
results.
A->0,
theory predicts that
their
whereas experiment shows that p T {X) always remains finite
equal to zero at zero wavelength. The unrealistic behavior of the
oo,
Rayleigh- Jeans distribution at short wavelengths
"ultraviolet catastrophe." This terminology
is
is
known in physics as the
suggestive of the importance
Rayleigh-Jeans
spectrum
T=1646K
X(in units of 10
Figure 2-1
1.
A comparison
-4
cm)
of the Rayleigh-Jeans spectrum and experiment.
of the failure of the Rayleigh-Jeans theory. Let us evaluate the total energy
cm3 for the Rayleigh-Jeans spectrum:
per
0. We have here an even more impressive
This goes to infinity unless T
complete
failure
of the Rayleigh-Jeans spectrum. Yet
indication of the
earlier, a necessary consequence
we
have
mentioned
spectrum
is,
as
this
of the theories of classical physics.
9.
Planck's
Theory
The discrepancy between experiment and theory was resolved in 1901
by Planck but only at the expense of introducing a postulate which was
not only new, but also drastically at variance with certain concepts of
THERMAL RADIATION AND QUANTUM THEORY
64
Planck's postulate
classical physics.
Any physical
oscillations
may
[Ch. 2
be stated as follows
whose single "coordinate" executes simple harmonic
a sinusoidal function of time)^ can possess only total
entity
(i.e.,
is
energies s which satisfy the relation
e
where v
An
nhv,
0, 1, 2, 3,
the frequency of the oscillation
is
and h
a universal constant.
is
way of
energy level diagram provides a convenient
behavior of an entity governed by
contrasting this behavior with
(2-29)
and
this postulate,
illustrating the
it is
also useful in
what would be predicted by
classical
= bhv
Classical
Figure 2-12.
= 4hv
= 3kv
= 2hv
= hv
=
Planck
Energy
level
diagrams for a
classical
simple harmonic oscillator and a
simple harmonic oscillator obeying Planck's postulate.
physics.
See figure (2-12).
we
In such diagrams
indicate each of the
possible energy states of the entity with a horizontal line.
The
distance
from the line to the zero energy line is proportional to the total energy to
which it corresponds. Since the entity may have any energy from zero to
infinity according to classical physics, the classical energy level diagram
consists of a continuum of lines extending from zero up. However, the
entity executing simple harmonic oscillations can have only one of the
discrete total energies s
This
postulate.
is
nhv, n
0, 1, 2, 3,
obeys Planck's
if it
indicated by the discrete set of lines in
its
diagram. The energy of the entity obeying Planck's postulate
energy level
is
said to be
and the
quantum number.
Let us recalculate p T (X) under the assumption that Planck's postulate
is obeyed by the electromagnetic standing waves whose amplitudes are,
according to equation (2-13), executing simple harmonic oscillation with
quantized, the allowed energy states are called
integer n
is
quantum
states,
called the
t The word "coordinate" is used in its general sense to mean
describes the instantaneous condition of the entity. Examples are
spring, the angular position of a
any quantity which
:
the length of a coil
pendulum bob, the amplitude of a wave. All
examples happen also to be sinusoidal functions of time.
these
PLANCK'S THEORY
Sec. 9]
frequency
65
No modification will be required in the calculation of N(v) dv
v.
no point
was it necessary to specify the energy
However, the evaluation of s involved integrals in
which the energy of the standing waves, e, was the variable of integration.
This was proper in the original calculation since classically the energy can
assume any value. But, under Planck's postulate, e can only have the
discrete values nhv where n = 0, 1, 2, 3, 4,
Thus we must recalculate
Then
e, replacing all integrals over e by summations over this variable.
since at
in that calculation
of the standing waves.
00
-Q
__
00
00
n=
__ n =
where we have assumed, as did Planck, that the Boltzmann probability
distribution
employed
In a manner completely analogous to that
applies.
still
in the calculation preceding equation (2-20),
we can show
this
yields
1
= -
oo
da.
2 e~"'
n=o
<*7iV
In
hv
Now
tznhv
Ip
p 2<xhv
_ 3aftv
_i_
_i_
or
00
-v
Je
n=
where x
e~
xhv
=1 + X+X +X +
Z
...
But
(1
x)' 1
an " v
(1
x%
x3
Therefore
GC
2=
e~
_0* v )~ 1
71
and
- ln(l
-e-"**)" 1
den
_(1
e -^)(_l)(l
1 '"
- xh
hv
hve""'
1
-"v
^v
*y%-{-hv)e- ahv
hv
e^l^T
(2-30)
i
Evaluating p T (v) dv as before, using equation (2-17), we have
Pt(v) dv
~\
= SJMdv =
hv
THERMAL RADIATION AND QUANTUM THEORY
66
[Ch. 2_
Transforming to the variable X with equation (2-25), we obtain
Pj{X) dX
Znhc
J5
This
is
the black
body
JiclkXT
dl
(2-32)
spectral distribution derived
by Planck.
It
Wien's law with/(Ar) = %-nhc\(e ullc, T 1). This function involves
the constant h whose numerical value is to be determined by fitting the
-
satisfies
\
Figure 2-13.
(in units of
10 ~* cm)
comparison of Planck's spectrum and experiment.
T = I646K.
is theoretical.
The dots are
experimental and the curve
theoretical spectrum to the experimental data.
figure (2-13).
The value providing the
h
6.63
best
X 10"27
fit
Such a
to the data
fit
is
shown
in
is
erg-sec
This constant is known as Planck's constant. The figure shows that Planck's
theoretical spectrum is in extremely good agreement with the experimental
spectrum for T 1646 K. Since it satisfies Wien's law, we know that it
will agree equally well with experiment at any other temperature.
10.
Some Comments on
Planck's Postulate
worth while to give a qualitative explanation of how Planck's
which in terms of the mathematics only has the effect of replacing
integrals by sums, is able to produce such a remarkable change in the
It is
postulate,
results of the calculation.
essentially
The
failure of the Rayleigh- Jeans theory is
due to the rapid increase in N(y) dv as
becomes
large.
Since
SOME COMMENTS ON PLANCK'S POSTULATE
Sec. 10]
67
the classical law of equipartition of energy requires that e be independent
of v, the spectrum itself will also increase rapidly with increasing v (or
decreasing
Planck's postulate has the effect of putting a cut-off into
does not obey the classical law. This keeps the spectrum
bounded and, in fact, brings the spectrum to zero as v -* ao (or as k -* 0).
This behavior of s can be seen directly from equation (2-30). The decrease
s so that
A).
it
comes about in the following way. Under Planck's postuthe energy of the standing waves is either e
0, or s
hv, or e
2hv,
etc. If we now consider the Boltzmann probability distribution
(equation
in e as v ->- oo
late,
we see that for v large enough that hv
kT the probability that a
standing wave will have an energy other than e
becomes negligible.
Consequently, when we evaluate s for standing waves whose frequency
>
2-23),
we find e
0. On the other hand, for the very low frequency
standing waves in which hv
kT, the quantum states are so close together
that they behave as if they were continuously distributed, as in the
classical
theory. Thus in the low frequency limit Planck's value of e
approaches the
v is large,
<
classical value
of i
kT, as
from equation (2-30).
was not so far reaching as it is
in the form we have given. Planck's initial work was done
with the theory,
described in section 6, in which the energy in a particular
frequency
component of the black body radiation is related to the energy of a
In
its
is
also apparent
original form, Planck's postulate
(charged) particle in the wall of the black body cavity oscillating
sinusoidsame frequency, and he postulated only that the energy of the
ally at the
oscillating particle is quantized.
After the development of the Rayleighit is equivalent to quantize directly
the oscillating electromagnetic waves, and the postulate
was broadened
to include any entity whose single "coordinate" oscillates
Jeans theory
it
became apparent that
sinusoidally.
Planck's
first
step consisted in finding,
by
and
an empirical
equation which agreed with the experimental spectra. Then he
set about
finding what modifications of the classical theory were
required to yield
this equation. In considering the derivation of the
Boltzmann probability
distribution P(e), which we described in terms of a simple
example in
section 7, he was led to try holding constant the quantity
that we called
trial
error,
Ae in our example, instead of taking it to zero as required by classical
theory. Although P(e) then has values only for discrete e,
these values are
not changed. However s is changed, as we have seen, and its new
value
depends on Ae. Planck soon found that the desired spectral distribution
is
obtained
Ae
proportional to the oscillation frequency v. Then the
possible energies e
n Ae become e
nhv, as stated in his postulate.
The success of Planck's postulate in leading to a theoretical black body
if
is
spectrum which agrees with experiment requires that we tentatively accept
at least until such time as it may be proved to lead to conclusions
its validity,
THERMAL RADIATION AND QUANTUM THEORY
68
[Ch. 2
which disagree with experiment. It may have occurred to the reader that
there are physical systems whose behavior seems to be obviously in disagreement with the postulate. For instance, an ordinary pendulum executes
simple harmonic oscillations, and yet this system certainly appears to be
capable of possessing a continuous range of energies. Before we accept
this argument, it would be prudent to make some simple numerical
calculations concerning such a system. Consider a pendulum consisting
of a 1-gm weight suspended from a 10-cm string. The oscillation frequency
of this pendulum is about 1 .6/sec. The energy of the pendulum depends on
the amplitude of the oscillations.
Assume the amplitude to be such that
extreme position makes an angle of 0. 1 radian with the
vertical; then the energy E is approximately 50 ergs. If the energy of the
the string in
its
pendulum is quantized, any decrease in the energy of the pendulum, due
for instance to frictional effects, will take place in discontinuous jumps of
AE =
magnitude
not observed. But, if we consider that
10" 26 ergs, whereas
50 ergs,
it is apparent that even the most sensitive experimental equipment would
be totally incapable of resolving the discontinuous nature of the energy
AE =
hv
6.63
hv.
10
This
-27
is
1.6
E~
decrease.
No
evidence, either positive or negative, concerning the validity of
Planck's postulate
The same
we
is
is
is
to be found
consider systems in which v
of the order of
example
is,
radiation.
from experiments involving a pendulum.
Only when
true of all other macroscopic mechanical systems.
E are
we
is
so large and/or
is
so small that hv
One
in a position to test Planck's postulate.
of course, the high frequency standing waves in black body
Other examples will be considered in the following chapters.
BIBLIOGRAPHY
Mayer,
New
J. E.,
and M. G. Mayer,
Statistical Mechanics,
John Wiley and Sons,
York, 1940.
Peck, E. R., Electricity and Magnetism, McGraw-Hill
Book
Co.,
New
York,
1953.
Richtmyer, F. K., E. H. Kennard, and T. Lauritsen, Introduction to Modern
Physics,
McGraw-Hill Book Co.,
New
York, 1955.
EXERCISES
1.
Verify (2-2) by using the procedure suggested immediately below that
equation.
2.
When
the sun
is
directly overhead, the thermal energy incident
upon
the
Ch.
EXERCISES
2]
69
about 1.4 x 10 6 ergs-crrf^-secr 1 The diameter of the sun is
about
1.6 x 10 u cm, and the distance from the sun to the earth is
about 1.3 x 10 13 cm.
Assuming that the sun radiates like a black body, use equation (2-8) to
estimate
earth
its
is
surface temperature.
3.
Repeat the calculation of section 7 for a system with
energy 6 Ac.
Compare
six entities
and
total
the results with equation (2-18), as in figure
(2-10).
4. Prove the fact, used in obtaining
equation (2-21), that the total energy
of a simple harmonic oscillator is twice its average
kinetic energy. Do this for
the case of a pendulum.
5. Integrate equation (2-32) over all
the results agrees with equation (2-8).
6.
To
A,
and show that the
T dependence
of
a certain approximation, an atom of a crystal executes
simple harmonic
about its equilibrium position because it possesses thermal
oscillations
agitation
energy and experiences an approximately linear
restoring force. Use the bulk
compressibility of some crystal to estimate the strength
of the restoring force
and, from this, the oscillation frequency. Use
equation (2-21) to estimate the
oscillation energy at room temperature.
From these results determine the
approximate value of the quantum number that describes
the motion of this
microscopic simple harmonic oscillator, and compare
it with the approximate
value of the quantum number that describes
the motion of the macroscopic
simple harmonic oscillator discussed in section
10.
CHAPTER
Electrons
and Quanta
Cathode Rays
I.
In this chapter
we
the photoelectric effect
shall primarily
be concerned with two phenomena,
and the Compton
effect, the interpretation of which
played a crucial role in the development of the quantum theory. These
phenomena
involve the interaction of electromagnetic radiation with
To pump
Glass tube
Cathode
A
matter
we
specifically
cathode ray tube.
with the electrons contained in matter. Consequently
shall first briefly describe the discovery of electrons
and the quantitative
measurement of their charge and mass.
The conduction of
electricity
through rarefied gases was a popular
subject of investigation in the second half of the nineteenth century.
70
Sec.
I]
CATHODE RAYS
71
typical experimental arrangement is shown in figure (3-1).
high voltage
applied between the negative electrode, or cathode, and the positive
electrode, or anode, causes current to flow through the
tube which is
filled with some gas (such as air) at a low pressure.
The current produces
a striking pattern of layers of glowing gas which alternate with dark
layers.
The exact nature of the pattern depends on the pressure of the gas as well
as on its composition and the magnitude of the voltage.
A complicated
set of phenomena, involving ionization of the gas, is
responsible for the
conduction of current and for the light emitted by the gas. We shall not
discuss these processes here.
pressures, in the region of 0.01
menon
develops.
The gas
When
the tube
is
pumped
to very low
mm of Hg or less, a quite different pheno-
glow (for one reason, because there
ammeter indicates that a current
is still flowing in the tube.
Another indication that something is still
happening in the tube is the presence of a luminous spot S
on the inner
surface of the glass wall at the end of the tube. The
size of the spot and
its position on the end wall of the tube
are observed to depend on the size
and location of the holes in diaphragms D and D' in such
a way as to
imply that the luminous spot is caused by rays of
particles which leave
from the cathode and travel in straight lines, finally impinging
upon the
is
very
little
gas
left).
ceases to
Nevertheless, the
wall.
Additional supporting evidence is obtained from the
observation
that a solid object placed to the right of D' will
cast a sharp shadow on the
luminous spot at the end of the tube. The position of the
luminous spot
can be influenced by the presence of a strong electric
or magnetic field in
the region to the right of D', showing that the
particles traveling from the
cathode to the wall of the tube are electrically charged.
The sense of the
deflection of the spot is such as to indicate that the
charge of the particles
is negative.f This was confirmed
by an experiment in which an insulated
metallic chamber was placed inside the tube
at S; the chamber was
observed to collect a negative charge. These particles,
which travel in
ray-like paths from the cathode, were called
cathode ray particles, or
sometimes simply cathode
rays. The apparatus shown in figure
(3-1)
called a cathode ray tube. It is now known that
the cathode ray particles
are emitted from the cathode as a result of its
bombardment by positive
is
ions of the gas contained in the tube.
t The reader may wonder why the negatively charged particles travel
in straight lines
in the region to the right of D\ instead of being
attracted to the positively charged anode
The reason is that most of the voltage drop across the
tube occurs near the cathode
Thus the charged particles are accelerated in the strong
electric field near the cathode
and then they continue moving in straight lines. The
electric field near the anode is too*
weak to produce an appreciable deflection.
ELECTRONS AND QUANTA
72
2.
[Ch. 3
The e/m Ratio of Electrons
In 1897 Thomson made accurate measurements or e/m, the ratio of
the charge of the cathode ray particles to their mass. This provided the
key to their identity. Thomson's apparatus is shown in figure (3-2).
Cathode ray particles of (negative) charge e and mass m travel from
the cathode C toward the anode A, which is pierced with a small hole and
H.v.
Glass tube-
An
Figure 3-2.
apparatus used to measure the e/m ratio for electrons.
first of two diaphragms, D and D', that define the particles
narrow beam. The particles then pass between two parallel metal
plates of length / and separation d. Finally they strike the end of the tube,
producing a luminous spot at Sv The application of a voltage V across
the two plates creates a force F e Vj'd which acts on the particles in a
also acts as the
to a
direction perpendicular to their velocity
verse acceleration a
t
l/v
acceleration lasts for the time
during which the particles are passing between the plates.
emerging,
The
This force produces a trans-
v.
The
(e/m)(V/d).
transverse
their
deflection
St S2
deflection
is
of the luminous spot
is
amplified
almost exactly 2L/1, where L
the center of the plates to the end of the tube. Thus
for small deflections,
is
SiS 2
VL
by a factor which,
the distance from
is
(3-1)
S^
known, equation
(3-1) allows the evaluation of e/m.
To
measured and the velocity v of the
Thomson made
Upon
\(e/m){V/d){f/vf.
If the deflection
is
%at 2
particles is
a separate measurement employing an
due to a voltage V across the two plates and also a
magnetic force resulting from the interaction of the moving charged
particles with a magnetic field
applied over the region between the plates
determine
v,
electrostatic force
THE CHARGE AND MASS OF ELECTRONS
Sec. 3]
and
73
With the
and magnetic
in a direction perpendicular to the plane of the figure (3-2).
correct orientation of the magnetic field, the electrostatic
forces will be opposed;
in fact, for a certain value of the magnetic field
strength the forces can be
made
to cancel out exactly.
For
this value of
the following relation obtains
eVjd
and the luminous spot
is
Hev/c
(3-2)
no deflection. The left side of this equation
and the right side the magnetic force. From
suffers
the electrostatic force,
equations (3-1) and (3-2), ejm can be evaluated in terms of the quantities
measured in the experiment. The currently accepted value is
ejm
=
=
5.27
10 17 esu/gm
1.76
10 8 coulombs/gm
The results of the experiment were observed to be independent of both the
composition of the cathode and the nature of the residual gas in the tube.
This showed that the value of ejm stated above is characteristic of the
cathode ray particles themselves and that all cathode ray particles have
same value of ejm.
The ratio of the charge to the mass of cathode ray
the
particles is very
larger than the value of that ratio for ionized atoms.
is
measured in experiments involving the
ionized atoms; an electrolytic
cell is
electrolysis
The
much
latter quantity
of solutions containing
operated for a certain period of time
and the total charge carried by the ionized atoms, as well as the total mass
of the atoms deposited at the electrode, is measured. The largest atomic
value of ejm is that for an ionized hydrogen atom, but ejm for a cathode
ray particle
is
1836 times greater than ejm for an ionized hydrogen atom.
Either the charge of a cathode ray particle
is
very
much
larger than the
charge of an ionized atom, or the mass of a cathode ray particle
much smaller than an atomic mass.
Thomson made the assumption that
the second alternative
was
In fact, he assumed that the charge of the cathode ray particles
equal in magnitude to the charge of a singly ionized atom.
charge
is
called the electronic charge.
modern terminology,
3.
The cathode ray
is
is
very
correct.
exactly
This unit of
particles,
in
are called electrons.
The Charge and Mass
of Electrons
The magnitude of the electronic charge was determined in a series of
experiments performed by Townsend and Thomson (ca. 1900) and by
Millikan (1909).
These experiments were based on the observation that
ELECTRONS AND QUANTA
74
[Ch. 3
very small drops of liquid often pick up electric charges. If a voltage V
is applied between two horizontal parallel plates located above and below
the region containing the charged droplets, a droplet can be suspended
against the force of gravity
when
q-=Mg
(3-3)
where d
is
the separation of the parallel plates, q
is
the charge
on the
980 cm-sec-2 In order to determine M,
g
the voltage is removed, and the droplet falls under the influence of gravity
until it reaches terminal velocity because the frictional drag F becomes
equal to the gravitational force Mg. According to Stokes' law, Fv =
6nrjav, where a is the radius of the droplet, v is its terminal velocity, and r\
droplet,
is
mass, and
is its
the coefficient of viscosity of the
Thus
is falling.
medium (air) through which
we have the relation
the droplet
at terminal velocity
Mg =
(3-4)
6irr]av
The radius of the droplet can be expressed in terms of its mass and density
= ^-n&p. Putting this into equation (3-4), we can
from the equation
solve for
in terms of the terminal velocity v. The latter quantity is
measured by observing with the aid of a microscope the distance through
which the droplet falls in a given time. Knowing M, we find q from equation
(3-3).
The results of a large number of measurements on the charge of different
droplets could
all
be expressed as
q
where n
That is,
is
it
4.80
lO
-10
esu
(3-5)
an integer, and where we quote currently accepted values.
was observed that the charge of a droplet was always some
integral multiple of 4.80
meaning that there
exists
10 -10 esu.
It is
reasonable to interpret this as
a basic unit of charge of magnitude
= 4.80
= 1.60
10" 10 esu
10- 19 coulombs
no droplet was ever found to have a charge less than this amount.
his colleagues assumed that the electronic charge was
equal to this basic unit of charge. Then, knowing the value of ejm for
an electron, they were able to evaluate the mass of an electron. The value
they obtained was not far from the currently accepted value,
since
Thomson and
m=
9.11
x 10-28 gm
BUCHERER'S EXPERIMENT
Sec. 4]
75
is now an overwhelming body of evidence proving the correctness
of the assumption that the electronic charge is equal to 4.80 X 10 -10 esu.
There
For
instance,
possible to evaluate the electron
it is
mass from the equation
(e/m) H +
e/m
1836
where (e/w) H+ represents the charge to mass ratio for an ionized hydrogen
atom and m n+ represents its mass. The latter quantity is equal to the
chemical atomic weight of hydrogen divided by Avogadro's number.f
Thus
L2JL_
6.02
L=
10 23 1836
9 ai
llxl0
x 1U
*
-28
This provides an independent check on the value of the electron mass
determined from e/m and the value of e and thereby confirms the correctness of the value of e given above.
In recapitulation, the electron is a particle having a negative charge
the magnitude of which is equal to the magnitude of the charge of a
singly ionized atom, and a mass which is smaller than the mass of an
ionized hydrogen
4.
atom by a
factor of 1836.
Bucherer's Experiment
This is an appropriate place to describe the work of Bucherer
(1909)
which provided the first experimental verification of the relativistic depend-
m=
ence of mass on velocity:
mjVl - ^/c 2 (equation 1-20). This
experiment consisted of a series of measurements of the e/m ratio of high
velocity electrons. The reader will recall that the theory of relativity does
not alter the laws of electromagnetic phenomena; the charge of an
electron,
just as
e, is
it is
a constant independent of its velocity in the theory of relativity
in the classical theory.
Consequently the
relativistic
theory
predicts
cV
W,
/e
cV
where (e/w),,/c_ is the value of e/m for electrons of velocities small compared to c. The quantity (e//n) e/( ^ is what was measured in Thomson's
experiments, since the velocity of the electrons in the cathode ray tubes
which he used is very small compared to c. In order to obtain electrons
t
We are ignoring the 0.05 percent difference between the mass of an ionized hydrogen
atom and
that of a neutral hydrogen atom.
76
ELECTRONS AND QUANTA
[Ch. 3
with velocities comparable to c, Bucherer used the electrons which
are
spontaneously emitted from the naturally radioactive elements (cf.
section 2, Chapter
These electrons are emitted with a continuous
4).
spectrum of velocities which extend, in some cases, to velocities as high as
Bucherer's apparatus was essentially similar to that used by
Electrons of a certain velocity were first selected from the
continuous spectrum by a combination of electric and magnetic fields.
These electrons were then deflected in a magnetic field, and the e\m ratio
0.99c.
Thomson.
was determined from the magnitude of the deflection. The results of the
experiment, which are shown in figure (1-14), are in excellent agreement
with equation (3-6) over the entire range of velocities investigated.
5.
The Photoelectric
Effect
In a cathode ray tube, electrons are emitted from a metallic electrode
bombardment of the electrode by positive
ions of the gas contained in the tube. Another process in which electrons
are emitted from the surface of a metal was discovered by Hertz in 1887.
(the cathode) as a result of the
later
form of
To pump
shown
his apparatus is
in figure (3-3).
-s-
The
glass tube
Ultraviolet light
&
III
Figure 3-3.
photoelectric
cell.
contains a polished metal electrode called a photocathode, and a second
electrode in the form of a perforated metal plate. The two electrodes are
maintained at a potential difference of a few volts (normally with the
second electrode positive with respect to the photocathode). When
ultraviolet light (frequency v
~ 10
16
sec" 1) passes through the perforated
second electrode and is incident upon the inner surface of the photocathode,
a current is observed to flow through the tube. This phenomenon is called
the photoelectric effect. The effect persists even when the tube is evacuated
to a very low pressure, implying that gaseous ions are not the carriers of
77
THE PHOTOELECTRIC EFFECT
Sec. 5]
Experiments in which a magnetic
the current.
field
was applied
to the
region between the photocathode and the second electrode indicated that
the current consisted of the flow of negatively charged particles. Thomson's
discovery of the existence of electrons suggested that the negatively charged
particles of the photoelectric effect could also be electrons which were
liberated
from the photocathode as a result of
This hypothesis was confirmed
violet light.
measured the ejm ratio of the photoelectric
was the same as that for electrons.
-2
v'max
Potential of the
Figure 3-4.
second electrode
The voltage dependence
in
bombardment by
1900 by Lenard,
particles
ultra-
who
and showed that
it
+4
+2
its
wrt.
photocathode
(volts)
of the current liberated in a photoelectric cell.
cleared up the question of the identity of the
but they also demonstrated some properties of the
which, as we shall see, were very difficult to understand
The experiments of Lenard
photoelectric particles,
photoelectric effect
in terms of the theories of classical physics.
Lenard measured the current
reaching the second electrode as a function of the potential between that
electrode and the photocathode, keeping all other parameters fixed.
His data are represented in figure (3-4). An interesting feature of these
data
is
that there
the potential
is still
some current reaching the second
of that electrode
is
electrode
when
negative with respect to the photoof course, tend to repel the negatively
A potential of this sign will,
charged photoelectrons. This must mean that the photoelectrons are
ejected from the photocathode with non-negligible kinetic energy. The
shape of the curve of current versus potential difference indicates that not
cathode.
photoelectrons have the same kinetic energy; but the well-defined
Fmax shows that there is a well-defined maximum
endpoint at V
all
kinetic energy
charge.
It
max = eFmax
was
where e
is
the magnitude of the electronic
soon suggested that the photoelectrons of
maximum
the surface of the photocathode, whereas
energy were emitted from
lower energy photoelectrons originated inside the surface and lost kinetic
energy in the process of reaching the surface. This suggestion implies that
Emax is a good measure of the energy given to an electron in the photoLenard made a series of measurements, of the type
electric process.
ELECTRONS AND QUANTA
78
[Ch. 3
indicated in figure (3-4), as a function of the intensity of the incident
ultraviolet light. He found that the photoelectric current for positive
values of V (for which photoelectrons of all kinetic energies will reach the
second electrode) was directly proportional to the light intensity. However
Vmax was observed to be independent of the light
intensity. Thus the energy E
max acquired by the photoelectrons does not
depend on the intensity of the incident light.
the cut-off potential,
6.
Classical
Theory of the Photoelectric
Effect
Let us consider what the classical theory would predict about the photoeffect.
Thomson's experiments showed that metals contain
electrons, although it was not really known whether the electrons were
electric
free to
move throughout
the metal or were bound to atoms. In either
mechanism for the ejection of these electrons by the absorption of
incident light would involve an interaction between the electric field of the
electromagnetic light waves and the electric charge of the electrons. The
oscillating electric field acting on the charged electrons would set them
into vibration with an amplitude that would be proportional to the amplitude of the electric field. Now, it is easy to show that the average kinetic
energy of any vibrating body is proportional to the square of the amplitude
of its vibration. Furthermore, the electromagnetic theory shows that the
case, the
amplitude of the oscillating
electric field is proportional to the
root of the light intensity
equation 2-6). Thus
(av. kin. energy) cc
(cf.
(amp. of vib.) 2 oc (amp. of
square
we have
elect, fid.) 2 oc
(Vlight
int.)
That is, the average kinetic energy of the vibrating electrons would be
proportional to the intensity of the incident light. This makes it difficult
to understand why the energy acquired by the photoelectrons was
observed experimentally to be independent of the light intensity.
An even more serious problem, which arises in any attempt to explain the
photoelectric effect in terms of the classical wave theory of light, is that
according to such theories a very long time is required for an electron to
absorb enough energy. The problem can be illustrated by a simple calculation.
This calculation requires an estimate of the radius of an atom. The
estimate can be obtained from a knowledge of the macroscopic density of
an element, p, its atomic weight A, and Avogadro's number N. Consider
a typical element,
gm.- 1
Now
of an atom
silver;
~ 10 gm-cm- 3 A ~ 100, also JV~ 6 x 1023
atom is AjN~ 100/6 x 1023 gm. The volume
where r is its radius. Assuming that the atoms
p
the mass of an
is
V=
fn-r
3
,
QUANTUM THEORY OF THE PHOTOELECTRIC
Sec. 7]
in the solid are closely packed, the density of an
mately equal to the density of the solid. So
mass/atom
'
or
1n
10
___"
gm-cm
atom
100/6
c=l
EFFECT
will
79
be approxi-
X 10-23 gm'1
sT
volume/atom
IQr8 cm. This result can be confirmed in several
from which we obtain r
analysis of the measured diffusion rate of a
an
from
ways. For instance,
terms
of the kinetic theory of gases, it is also
in
atoms,
gas composed of
8
10~
cm.
r
found that
us calculate the time required for photoelectrons to absorb
eFmax which is observed experimentally to be of the
the energy .Em ax
lO" 19 joules = 10~ 12 ergs. Let
order of 1.6 x 10~ 19 coulombs x 1 volt
Now
let
us assume that a source which emits energy in the form of ultraviolet
1
10 7 erg-sec- 1 is located 1
1 joule-seclight, at the rate of 1 watt
meter from the photoelectric tube. If the light is emitted with spherical
symmetry, the energy flux per cm2 at the tube is equal to the energy
cm2 to the area of a sphere of radius
X 1/4tt(102)2 cm2 ~ 102 erg-cm^-sec"1
emitted per second times the ratio of
1
that is, 10 erg-sec1 meter;
Let us assume that the electrons emitted in the photoelectric effect are
bound to atoms and that they are somehow able to absorb all the energy
incident upon the atom to which they are bound. The radius of an atom is
-8
of the order of 10 cm. Thus its cross sectional area is of the order of
10
-16
cm2 and
,
the rate at which energy
is
incident
cm =
upon
this area is
10" 14 erg-sec -1
If we
approximately 10 erg-cnr^-sec"" x
their
rate
bound
to
atoms,
were to assume that the electrons were not
can
estimate
Finally
we
be
less.
of absorption of energy would presumably
-12
that the time required for a photoelectron to absorb the observed 10
ergs is about 10 2 sec. This calculation, which is based on the assumption
that the light energy is uniformly
characteristic of the wave theory
2
10" 16
from the source
was turned on
and the emission of the first photoelectrons of about a minute. No
such time delay was observed by Lenard. In fact, experiments performed
in 1928 by Lawrence and Beams, using a light source many orders of
magnitude weaker than we assumed in the above calculation, set an
upper limit on the time delay of about 10~9 sec!
distributed over spherical
wave
fronts spreading out
predicts a delay between the time at
7.
Quantum Theory
light source
of the Photoelectric Effect
In 1905 Einstein enunciated a
effect
which the
quantum theory of
the photoelectric
which was closely related to the quantum theory of black body
80
ELECTRONS AND QUANTA
[Ch. 3
He reasoned that Planck's requirement that the energy content
of the electromagnetic waves of frequency v in a radiant source (e.g.,
the ultraviolet light source) can only be 0, or hv, or 2hv,
or nhv,
radiation.
implies that in the process of going
from energy
nhv to energy state
(n \)hv the source would emit a burst of electromagnetic energy of
energy content hv.
state
assumed that such a burst of emitted energy is initially localized
a small volume of space; and furthermore that it remains localized as it
moves away from the source with velocity c, instead of spreading out in the
Einstein
in
manner
characteristic
of moving waves. He assumed that the energy
quantum of energy, is related to itsfrequency v
content s of such a bundle, or
by the equation
He
also
assumed
hy
that in the photoelectric process one
completely absorbed by an electron
in the
of the quanta
is
photocathode.
This would impart to the electron an energy equal to hv. If this energy is
greater than the quantity &E, the electron can escape from the photocathode, where Aisis the energy which the electron must expend in reaching
the surface of the photocathode plus the energy
W required to overcome
we know must be present at the surface since
do not normally escape. The kinetic energy of the electron after
the attractive forces which
electrons
escaping from the photocathode will be equal to
electron originating at the surface,
will
have
its
maximum
is
will
E=
hv
A".
For an
be just equal to W, and
value
^max
where
AE
hv
(3-8)
a constant which depends on the composition of the photo-
cathode.
According to
this theory, the rate
of emission of photoelectrons will be
proportional to the flux of quanta incident upon the photocathode, which
in turn will be proportional to the intensity of the incident electromagnetic
radiation.
This
is
in agreement with Lenard's observation that the photo-
proportional to the intensity. As can be seen from
equation (3-8), the theory also agrees with the observation that is m a X
does not depend on the intensity. Furthermore, the theory removes the
electric current is
difficulty
concerning the time required for the photoelectrons to receive
enough energy. This difficulty arose in the wave theory because the energy
of the source was spread uniformly over the entire wave front. In the
quantum theory
the energy
concentrated in the quanta. Within a very
turned on, a quantum will strike the photowill be absorbed by one of the electrons, and the ejection
short time after the source
cathode where it
of that electron will
result.
is
is
THE COMPTON EFFECT
Sec. 8]
Einstein's theory, as
diction that the
81
summarized by equation
(3-8),
makes the
maximum kinetic energy of the photoelectrons
pre-
should be a
linear function of the frequency of the incident electromagnetic radiation.
This prediction was tested in 1916 by Millikan, who measured is m ax for
-1
14
The nature of his
radiation in the frequency range 6 to 12 x 10 sec
.
data
is
is
indicated in figure (3-5).
The experiment shows
a linear function of the frequency
v.
that .Emax actually
According to equation
(3-8), the
fci
Figure 3-5.
The frequency dependence
of the
maximum
kinetic energy of photo-
electrons.
slope of the straight line fitting the data should be equal to Planck's
constant
h,
and
its
intercept
on the
v axis should
be equal to Wjh. Thus
the photoelectric experiments, with the aid of Einstein's theory, could be
used to determine the numerical value of h. The value of h which was
obtained agreed to within better than 0.5 percent with the value of that
constant which had been determined by the completely independent method
of fitting Planck's theory to the experimental black body spectrum. This
finding represented a great triumph for the
8.
The Compton
In 1923
quantum
theories.
Effect
Compton
discovered that,
when a beam of X-rays of
well-
by sending the
radiation through a metallic foil, the scattered radiation contains a comThis
ponent of well-defined wavelength ^ which is longer than A
phenomenon is called the Compton effect. As we shall see, it provides
extremely convincing evidence for the existence of quanta. Although a
description of the production of X-rays and the measurement of their
wavelength will not be presented until Chapter 14, it is desirable to discuss
defined wavelength X
is
scattered through
an angle
the
Compton
effect at this time.
ELECTRONS AND QUANTA
82
[Ch. 3
X-rays consist of electromagnetic radiation of wavelength which is
by a factor of the order of 104 This is
proved by experiments which show that X-rays can be diffracted and
shorter than that of visible light
polarized just like light.
In Compton's experiment the wavelength A
of the incident radiation was 0.708 x 10~ 8 cm. The wavelength A was
x
observed to depend on the angle of scattering 6, but not on the material
comprising the foil. A typical set of data is shown in figure (3-6) for
45
:90
TT
\> ^1
Wavelength spectra
From A. H. Compton, Phys.
Figure 3-6.
foil.
135
of quanta scattered at various angles
Rev., 22,
from a carbon
409 (1923).
and 135. These curves are wavelength spectra of the
The abscissas are a measure of the wavelength, and
the ordinates are a measure of the intensity per unit wavelength interval.
At all angles the scattered radiation is observed to contain a component of
wavelength equal to the wavelength X of the incident radiation. The
reason for the presence of this component will be discussed in Chapter 14.
Here we shall be concerned only with the component in the scattered radiation of wavelength X v The wavelength of this component is observed
always to be greater than the wavelength of the incident radiation, and to
6
0, 45, 90,
scattered radiation.
of scattering. At 6 = 90, X x X =
x 10~8 cm. In terms of the frequency of the scattered radiation,
c\l x we see that v x is always less than the frequency of the incident
increase with increasing angle
0.024
v1
radiation v
and decreases with increasing angle of
This suggested to
Compton
scattering.
that, since v t is proportional to the
energy
of the quanta, which according to Einstein are associated with electromagnetic radiation of this frequency, the dependence of E1 on 6 is qualita-
angular dependence of the energy of a particle which
has been scattered from another particle. Pursuing this, he developed a
simple but very successful theory.
tively similar to the
Consider a quantum in the incident
For radiation of frequency
v,
beam of electromagnetic
quantum is
radiation.
the energy of the
E=
hv
(3-9)
Taking the idea of a quantum as a localized bundle of energy quite literally,
THE COMPTON EFFECT
Sec. 8]
we
shall consider
it
83
E and momentum p.
to be a particle of energy
Such a
particle must, however, have certain quite specialized properties. Consider
equations (1-20) and (1-24), which give the total relativistic energy of a
particle in terms of its rest
mass
E=w
Since the velocity of a
hv
is finite, it is
Thus
and
its
velocity v
-r
c 2 /Vl
quantum equals
apparent that the
/c
and since its energy content
mass of a quantum must be
c,
rest
quantum can be considered
to be a particle of zero rest
which is entirely kinetic. The
momentum of a quantum can be evaluated from the general relation
(1-25) between the total relativistic energy E, momentum p, and rest mass
zero.
the
mass, and of total
This
relativistic
energy
is
y+
For a quantum the second term on the
Ejc
(WoC 2)2
right
hv/c
is
zero,
h\l
and we have
(3-10)
where X is the wavelength of the electromagnetic radiation associated with
the quantum. It is quite interesting to note that Maxwell's classical wave
theory of electromagnetic radiation leads to an equation p = Ejc, where
p represents the momentum content per unit volume of the radiation and
E represents the energy content per unit volume.
Now the frequency v of the scattered radiation was observed to be
independent of the material in the
does not involve entire atoms.
due
He
to collisions
foil.
This implies that the scattering
Compton assumed
that the scattering
between the quanta and individual electrons in the
was
foil.
assumed that the electrons participating in this scattering process
and initially stationary. Some a priori justification of this
assumption can be found from considering the fact that the energy of an
X-ray quantum is several orders of magnitude greater than the energy of
an ultraviolet quantum, and from our discussion of the photoelectric
effect it is apparent that the energy of an ultraviolet quantum is comparable to the energy with which an electron is bound in a metal. Consider,
then, a collision between a quantum and a free stationary electron,
as in figure (3-7). In the diagram on the left, a quantum of total relativistic
energy E and momentum p is incident on a stationary electron of rest
mass energy m c 2 In the diagram on the right, the quantum is scattered
at an angle 6 and moves off with total relativistic energy Ex and momentum
with kinetic energy T and
p lt while the electron recoils at an angle
of momentum and total
conservation
applied
the
momentum p. Compton
also
are free
</>
AND QUANTA
ELECTRONS
84
relativistic
Momentum
energy to this collision problem.
Ch. 3
conservation
requires
= Pi cos + p cos
Po
(3-1 1)
<f>
and
sin 6
px
= p sin
(3-12)
<f>
Squaring these equations, we obtain
(p
=p
cos 8) 2
cos 2
<f>
and
p\ sin
2
2
p sin
<
o^o
Before scattering
After scattering
Figure 3-7.
collision
between a quantum and a free stationary electron.
Adding, we find
p
Conservation of total
p\
2poPl cos 6
p*
relativistic
energy requires
E +m
= E + T+ m
(3-13)
c2
(3-14)
Thus
- Ex = T
According to equation (3-10),
this is
c(Po~Pi)
=T
(3-15)
According to equations (1-22') and (1-25),
(T
+m
c 2) 2
= cy +
(w
c 2) 2
So
T2 + 27m
c2
c 2/? 2
or
T2/^ + 2Tm = p2
Evaluating/? 2 from equation (3-13), and
Oo
Pi)
+ 2
c(Po
Pi)
(3-16)
Tfrom
Pt
p\
(3-15),
we have
2PoPi cos
Sec. 9]
DUAL NATURE OF ELECTROMAGNETIC RADIATION
85
which reduces to
<fi(Po
~ Pi) = PoPitt -
or
(1 - -) =
(1 moc
pj
\pi
cos e )
cos 0)
Multiplying through by h and applying equation (3-10),
Ax
Ao
A c (l
we
obtain
(3-17)
cos 0)
where
la
=
m
0.02426
x 10
_8
cm
is
the so-called
equation.
Compton wavelength.
It predicts that
Equation (3-17)
is
the increase in wavelength (A t
scattered electromagnetic radiation depends only
on
the
Compton
A)
of the
the scattering angle
and the universal constant h/m c and is independent of the wavelength of
The equation was verified quantitatively by the
experiments of Compton and others. The recoil electrons were observed
experimentally by Bothe and by Wilson (1923). Bothe and Geiger (1925)
showed that, in the scattering of a single quantum by an electron, the
scattered quantum and the recoil electron appear simultaneously. The
energy of the recoil electrons were measured by Bless (1927) and found to
be in agreement with the predictions of the theory. These experiments, and
the incident radiation.
their interpretation in terms of the theory, provide
even stronger evidence
for the existence of quanta than does the photoelectric effect.
9.
The Dual Nature
Einstein's
of Electromagnetic Radiation
quantum theory of the photoelectric effect was very successful
phenomenon which simply could not be understood on the
in explaining a
basis of the theories of classical physics. Nevertheless, many physicists
were very reluctant to accept it because it was completely contrary to the
well-established wave nature of electromagnetic radiation. After the
discovery and interpretation of the Compton effect, the existence of quanta
could no longer be questioned. As a result physics was in what seemed at
the time to be a very uncomfortable situation. On the one hand, electromagnetic radiation manifests certain properties, such as diffraction, which
could only be explained in terms of wave motion. On the other hand,
some of its properties, such as those evident in the photoelectric and
Compton effects, could only be explained in terms of quanta which, being
localized,
have essentially the characteristics of particles. Electromagnetic
ELECTRONS AND QUANTA
86
[Ch. 3
radiation appeared to possess a split personality, sometimes behaving as
were composed of waves and sometimes behaving as
posed of particles.
if it
The experimental
if it
were com-
facts concerning the properties of electromagnetic
radiation, as well as the interpretations of these properties in terms of the
existence of both wave aspects and particle aspects, remain essentially
unchanged today. However, as a consequence of the broader point of
view provided by the development of the theory of quantum mechanics,
the attitude of physicists concerning this situation is now very different
from
their initial attitude.
The
of electromagnetic radiation
is
duality evident in the wave-particle aspects
is
no longer considered unusual because
now known to be a general characteristic of all physical entities.
more,
this duality is
it
Further-
no longer considered to represent a problem,
since
possible to reconcile the existence of both aspects with the aid of the
it is
theory of quantum mechanics.
This theory will be discussed in detail
later.
BIBLIOGRAPHY
Born, M., Atomic Physics, Blackie and Son, London, 1957.
Richtmyer, F. K., E. H. Kennard, and T. Lauritsen, Introduction
Physics,
McGraw-Hill Book Co.,
New York,
to
Modern
1955.
EXERCISES
1.
Electrons are emitted at thermal velocities from a heated cathode, are
accelerated through a voltage drop V,
and then move
through a magnetic field of strength
motion. Derive an expression relating
based upon this expression, with which
H perpendicular
V,
it
in a circle of radius
to their direction of
H, and R to e/m. Design equipment,
would be possible to make a 1 percent
measurement of e/m.
2. Prove the statement, made in section 6, that the average kinetic energy of
a vibrating system is proportional to the square of the amplitude of its vibration.
Do this for the case of a pendulum.
3.
Derive an expression for the kinetic energy of the recoiling electron in the
Compton
4.
the
scattering process.
Derive the
CM
Compton equation
for the case 8
180 by transforming to
frame, treating the collision, and then transforming back to the
LAB
frame.
5. List the experimental evidence, mentioned in section 9, for the properties
of electromagnetic radiation that can be explained only by wave motion.
CHAPTER
The Discovery
of the Atomic
Nucleus
I.
Thomson's Model of the Atom
That electrons are emitted from the metallic cathode of a cathode ray
or photoelectric tube implies that the atoms of which the cathode is
comprised contain electrons.
atoms contain
electrons.
If so,
it is
This assumption
the simple picture of a positively ionized
or
more
electrons have been removed.
reasonable to assume that
is
attractive because
atom
It
as an
it
all
leads to
atom from which one
agrees with the experimental
observation that the charge of a singly ionized
atom
is
equal in magnitude
to the charge of a single electron, or that the charge of a doubly ionized
atom
is
equal to the magnitude of the charge of two electrons,
etc.
Additional evidence for the existence of electrons in atoms was soon
obtained from the experiments of Barkla and others (1909) concerning the
scattering of X-rays by atoms. These experiments will be discussed in
Chapter 14, but it is appropriate to mention at this point that the experiments provided an estimate of Z, the number of electrons in an atom.
It was found that Z is roughly equal to A/2, where A is the chemical atomic
weight of the atom in question. Another set of experiments which provided
a measure of the number of electrons in an atom will be described in this
chapter.
Since atoms are normally neutral, they must also contain positive charge
equal in magnitude to the negative charge carried by their normal comple-
ment of electrons. Thus a neutral atom has a negative charge of magnitude
87
THE DISCOVERY OF THE ATOMIC NUCLEUS
88
[Ch. 4
Ze, where e
is the electronic charge, and also a positive charge of the same
magnitude. That the mass of an electron is very small compared to the
mass of even the lightest atom implies that most of the mass of the atom
must be associated with the
positive charge.
All these considerations naturally led to the question of the distribution
of the positive and negative charges within the atom. Thomson proposed
a tentative description, or model, of the composition of an atom, according
to which the negatively charged electrons were located within a continuous
Electron
Sphere of
positive charge
Figure 4-1.
Thomson's model of the atom.
The positive charge distribution was
assumed to be spherical in shape with a radius of the order of 10 8 cm,
which was known to be the order of magnitude of the radius of an atom.
Owing to their mutual repulsion, the electrons would be uniformly distridistribution of positive charge.
buted through the sphere of positive charge. This is illustrated in figure
(4-1). In an atom in its lowest possible energy state, the electrons would be
fixed at their equilibrium positions.
In excited atoms
material at high temperature), the electrons
equilibrium positions.
would
(e.g.,
atoms
in a
vibrate about their
Since the electromagnetic theory predicts that an
accelerated charged body, such as a vibrating electron, emits electro-
magnetic radiation, it was possible to understand qualitatively the emission
of such radiation by excited atoms on the basis of Thomson's model.
However, a calculation of the spectrum of radiation which would be
emitted showed that the model did not appear to be able to lead to quan-
agreement with the experimentally observed spectra.
titative
Really conclusive proof of the inadequacy of Thomson's model was
obtained in 1911 by Rutherford from the analysis of certain experiments
involving the scattering of alpha particles by atoms. Rutherford's analysis
showed
charge
that, instead of being spread
is
throughout the atom, the positive
concentrated in a very small region, or nucleus, at the center of the
atom. Rutherford's work comprised one of the most important developments in the subject of atomic physics, as well as the foundation of the
subject of nuclear physics.
Sec. 2]
2.
Alpha
ALPHA PARTICLES
89
Particles
Alpha particles are doubly ionized helium atoms which are spontaneously
emitted at very high velocities from several of the heaviest elements, such
U and Ra. This phenomenon will be discussed in detail later in this
book; here we shall present only a few pertinent points which are required
as background for a description of the experiments which led Rutherford
as
to the discovery of the nucleus.
Photographic
tr
\
To
Collimator
Figure 4-2.
An
pump
"Radioactive
element
apparatus used to study the types of radiation emitted by radioactive
elements.
In 1896-98 Becquerel and
Mme.
Curie discovered that several of the
heaviest elements spontaneously emit penetrating radiation capable of
Soon the radiation emitted by these
was being investigated in many laboratories.
With apparatus of the type indicated in figure (4-2), it was shown that the
blackening a photographic plate.
so-called radioactive elements
radiation consists of three separate components.
Radiation
is
emitted
from a radioactive element at the bottom of a channel in a lead block which
collimates the radiation into a well-defined beam. With no magnetic field,
a photographic plate exposed to the radiation shows, upon development,
a darkened spot at a point in line with the collimating channel.
magnetic
field is
in a direction perpendicular to the plane of the figure, three
spots are seen
When
applied to the region between the collimator and the plate,
on the
plate after
it is
developed.
The two
darkened
spots which are
deviated from the zero field position must be due to charged particles
emitted by the radioactive element, and which are deflected, by the application of the magnetic field, in a direction depending on the sign of the
The undeviated spot is presumably due to uncharged radiation
upon which the magnetic field would have no influence. The particles of
which the positively charged component of the radiation are comprised
charge.
THE DISCOVERY OF THE ATOMIC NUCLEUS
90
The
are called alpha particles.
[Ch. 4
negatively charged particles are called
beta particles, and the uncharged radiation
is
given the
name gamma
radiation.
In the experiments just described the chamber containing the collimator
and
plate is evacuated. If the experiment is repeated with the chamber at
atmospheric pressure, the spot due to the alpha particles does not appear.
Apparently a few centimeters of air is sufficient to stop the alpha particles.
Radioactive
Very
substance
thin-
walled tube
* To pump
Figure 4-3.
An
apparatus used to prove that alpha particles are doubly ionized helium
atoms.
Similar experiments establish that a foil of
some dense substance several
from reaching the plate, but
millimeters thick will stop the beta particles
that the intensity of the spot due to the
decreased only
when
collimator and the plate.
We know now
energy electrons, whereas the
radiation of the
gamma
radiation
is
significantly
several centimeters of lead are placed between the
same nature
that the beta particles are high
gamma radiation
consists of electromagnetic
as light, but of very
much
higher frequency.
The nature of the alpha particles was established in a series of measurements performed by Rutherford. Using apparatus basically similar to
that used by Thomson, he measured the charge to mass ratio of the alpha
particles as well as their velocity. These measurements showed that the
alpha particles were emitted with a well-defined velocity in the range
1.4 x 109 to 2.2 x 10 9 cm/sec, the exact value depending upon the radioactive element from which they were emitted. Note that these velocities
The charge to mass ratio
of the alpha particles was found to be one-half the value of that ratio for
are roughly one-twentieth the velocity of light.
singly ionized hydrogen atoms. Thus, if their charge was equal in
magnitude to a single electronic charge, the alpha particles would have a
mass twice the mass of hydrogen atoms. Alternatively, if the alpha
particles were assumed to have a charge twice the magnitude of an electronic
charge, their mass would be four times the mass of hydrogen atoms (i.e.,
equal to the mass of helium atoms). Rutherford was inclined toward the
THE SCATTERING OF ALPHA PARTICLES
Sec. 3]
latter interpretation since
it
would mean
91
that alpha particles were simply
doubly ionized helium atoms.
By a simple experiment, using the apparatus shown in figure (4-3),
Rutherford proved that this is indeed the case. A very thin-walled glass
tube containing a radioactive substance was surrounded by a glass
enclosure which was initially evacuated.
Some of
the alpha particles
emitted by the radioactive substance were able to penetrate the thin-
walled tube and enter the outer enclosure.
After a time, a sensitive test
of the contents of the outer enclosure showed that
amount of ordinary helium
gas.
it contained a detectable
This finding confirmed the argument that
alpha particles are doubly ionized helium atoms
electrons
which can pick up two
and become neutral helium atoms during or after their passage
through the thin-walled tubing.
3.
The Scattering
of Alpha Particles
Rutherford and his collaborators performed
many
experiments in the
course of a program designed to determine the properties of alpha particles
and of their interaction with matter. By far the most interesting of these
were the experiments concerning the scattering of alpha particles upon
-Microscope
Diaphragm
Alpha particle
source
Figure 4-4.
Thin
An
alpha particles
is
foil
alpha particle scattering experiment.
The region traversed by the
evacuated.
foils of various substances. A typical experimental
arrangement is shown in figure (4-4). The radioactive source emits alpha
particles which are collimated into a narrow parallel beam by the two
diaphragms. The beam is incident upon a foil of some substance, usually
a metal. The foil is so thin that the alpha particles pass completely through
with only a small decrease in their velocity. However, in the process of
passing through thin
traversing the
foil,
each alpha particle experiences
many
small deflections
THE DISCOVERY OF THE ATOMIC NUCLEUS
92
[Ch. 4
due to the Coulomb force acting between its charge and the positive and
negative charges of the atoms of the foil. Since the deflection of an alpha
particle in passing through a single atom depends on the details of its
trajectory through the atom, it is apparent that the net deflection in passing
through the entire foil is different for different alpha particles in the beam.
As a result, the beam emerging from the far side of the foil is divergent.
A quantitative measure of its divergent character is obtained by measuring
the
number of alpha
particles scattered into
each angular range
to
accomplished with the aid of an alpha particle detector
+
consisting of a layer of the crystalline compound ZnS and a microscope.
ZnS has the useful property of producing a small flash of light when struck
d. This
<D
by an alpha
is
particle.
If observed
by a microscope, the
flash
incidence of a single alpha particle can be distinguished.!
ment an observer counts
the
number of
light flashes
due to the
In the experi-
produced per unit
time as a function of the angular position of the detector.
4.
Predictions of Thomson's Model
The deflection of an alpha particle in following a given trajectory through
an atom can be calculated from Coulomb's law and the laws of mechanics,
if the distribution
distribution
of charges in the atom
is
given.
Thus, for the charge
assumed in Thomson's model of the atom,
it is
possible to
calculate the expected results of the experiment just described.
It
is
Alpha particle
trajectory
Figure 4-5.
*\
An
alpha particle passing through a Thomson's model atom.
apparent from the discussion above that the calculation will not be able
to predict the behavior of a single alpha particle, but only the average
behavior of a large number of alpha particles. The average behavior,
however, is precisely what is measured in the experiment.
Consider the traversal of an alpha particle through a single atom of the
type discussed in Thomson's model;
figure (4-5).
mum
We
this is indicated schematically in
shall estimate the order
of magnitude of
possible deflection angle of the alpha particle.
< max ,
the maxi-
calculation will
is used in the radium dials commonly found on wrist watches. Individual
can be seen by a dark-adapted eye with the aid of a magnifying glass.
t This effect
light flashes
The
PREDICTIONS OF THOMSON'S MODEL
Sec. 4]
use classical mechanics, which
is justifiable
93
since vjc c^ 1/20.
The
de-
assumed to be due to the Coulomb forces acting between the
charged alpha particle and the charges of the various parts of the atom.
According to Coulomb's law, the force due to some region of the atom at
a distance r from the instantaneous position of the alpha particle, and
flection is
containing charge dq,
where
(r/r) is
is
a vector of unit length directed from the location of the
charge dq to the position of the alpha particle, and where 2e
of the alpha particle.
Let us begin by evaluating the
maximum deflection
the forces due to the electrons in the atom.
would seem
At
first
is
the charge
to be expected
from
glance, equation (4-1)
to imply that the electrons could give very large deflections to
the alpha particle since
dV -*
oo as r
* 0,
so that the alpha particle
would
experience a very large force and consequently suffer a large deflection
if it
happened to pass very
close to
case because the mass of an electron
an
electron.
This
is
actually not the
compared to the mass
of the alpha particle. We leave it as an exercise for the reader to show that,
in the collision of a very massive body initially moving at velocity v,
with a stationary body of small mass, the velocity of the light body after
the collision cannot be greater than 2u.| This result follows strictly from
the conservation of momentum and energy and does not depend on the
nature of the force between the two bodies. Consequently the maximum
momentum which can be transferred to an electron is p = m(2v), where
m is the mass of an electron. This must be equal to the momentum lost
by the alpha particle in the collision. Thus Apx the maximum change in
the momentum of the alpha particle due to a collision with an electron,
so very small
is
is
To
equal to 2mv.
deflection angle
evaluate the order of magnitude of the
assume that the
max
perpendicular to the initial
in figure
mass of
<
is
change
maximum
is
directed
~ 10 -4 radians
=
4(1836)m
4(1836)
the order of magnitude of the
maximum
collision with a single electron in the
We
momentum
momentum p of the alpha particle as illustrated
2mv/Mv, where
is the
Apjpa
(4-6). Then we have < max
an alpha particle which is equal to 4 X 1836m. Consequently
<max
r
This
atom;
deflection angle
but
it
is
due to a
clear that the
assume the binding energy of the electron to the atom to be small compared
t
to the energy of the alpha particle. Then in the collision the electron will be essentially
free.
THE DISCOVERY OF THE ATOMIC NUCLEUS
94
maximum
value of the total deflection angle (due to
electrons in the atom) will be of the
same order
all
[Ch.
the collisions with
since the chance of having
a very close collision with more than one or two of the electrons
is
negligible.
To
estimate
msLX
<f>
due to the forces arising from the positive charge, we
integrate equation (4-1) over the positive charge distribution to obtain
"^max
Figure 4-6.
Illustrating
momentum
of
p a looses
the evaluation of the
momentum Apa
the total force acting on
inside the atom.
the.
maximum
deflection angle
when
a particle
alpha particle at some instant when
=m)
where
it is
This gives
(4-2)
from the position of the alpha particle at that
volume element containing the positive charge dq, and (r/r)
a vector of unit length directed from the volume element to the position
r is the distance
instant to a
is
of the alpha particle.
It is clear
that in evaluating the integral there will
be cancellation from volume elements symmetrically disposed with respect
to the position of the alpha particle. However, to estimate the order of
magnitude of the maximum total force, we ignore such cancellation and
take
Tmax
The order of magnitude of
~ a#
I
the integral can be evaluated
by writing
it
as
FmaK ~2e^jdq=2eq(^)
where
(1/r 2 ) is the
dimensions, and q
(1/r 2 )
value of 1/r 2 averaged over a region of space of atomic
= Ze
is
the total positive charge of the atom.
must be of the order of l/R2 where
,
is
Now
the radius of an atom.
Consequently
Fmax ~2Ze*/R*
To
estimate the
maximum
deflection angle
(4-3)
<
max
we
evaluate A/?a the
PREDICTIONS OF THOMSON'S MODEL
Sec. 4]
95
given to the alpha particle by the force Fmax acting for a time
of the order of At, the transit time through the atom. The time At is
momentum
approximately equal to Rjv, where v
is
the velocity of the alpha particle, so
Apx ~ .Fm3x At
~ 2Ze IRv
2
and we have
~ApJpx ~2Ze*lRMv 2
m!lx
<j>
(4-4)
yth
atom
Figure 4-7.
The
velocity vectors before and after an alpha particle traverses the
j'th
atom.
For atoms containing the
largest positive charge,
Zci
100.
The other
quantities entering into equation (4-4) have the values
e^5
10- 1C esu
<=
X 10-24 gm
C:i
R~
10- 8
10- 9 cm/sec
cm
Conseqilently
<Pmax
2
10
We
x 10 a x 2 x 10"-19
x 8 x KT 21 x 4 x 1018
10
_ 4
radians
see that the positive charge can produce only a very small deflection
+Ze is distributed uniformly over a sphere of
cm, the very high velocity alpha particles never get close
enough to a large enough quantity of charge that they experience a
Coulomb force of the intensity required to produce a large deflection.
Now that we know something about the angle through which an alpha
because, since the charge
radius
~10~8
particle is deflected in the traversal of
a single atom,
learn about the angle through which the particle
is
let
us see what
we can
deflected in the process
of traversing the large number of atoms which lie along its path through the
Clearly, summing the results of the individual deflections is involved,
foil.
sum we must
recognize the three dimensional character of
the individual deflections.
Consider figure (4-7), which indicates typical
and
in this
THE DISCOVERY OF THE ATOMIC NUCLEUS
96
[Ch. 4
v,- and vj, the velocity vectors before and after traversing
and a coordinate system useful in describing these vectors.
orientations of
the yth atom,
The coordinate system is chosen such that the x axis is in the direction of
the beam of alpha particles incident upon the foil. Because the change in
the
momentum
vector of the alpha particle in any single deflection
extremely small, the angles between
all
the velocity vectors
velocity vector remain small and, for any/, both y,
Figure 4-8.
A two
dimensional diagram representing the
y'th
and
is
and the
initial
v' are
nearly
deflection of an alpha
particle.
perpendicular to the yz plane.
Consequently, the nature of the deflection
can be specified by the angle 3 measured between the two vectors in the
plane which they define, and the angle y,- between that plane and the
z axis. This allows us to represent the deflection in terms of a two dimen</>
diagram consisting of the y and z axes and a vector 8,., as shown in
The length of the vector is equal to the magnitude of the
angle <,, and the angle between the vector and the z axis is equal to the
angle ipj. A moment's consideration will show that the relative orientation
between the velocity vector after the
deflection and the initial velocity
vector can be described in the same representation by a vector 8 where
sional
figure (4-8).
Mh
N
3=1
Thus the
entire deflection process can be represented by a two dimensional
diagram, a typical example of which is shown in figure (4-9). The net
angle of deflection resulting from
individual deflections is equal to the
4
et 10~
radians,
<f> s
the magnitude of the individual deflection angles <f> as well as the oriens
tation of the deflection planes specified by \p jt depends in an essentially
length of the vector 8.
Except for the constraint,
,
manner on both the exact trajectory of the alpha particle
through the atom and on the exact location of the charges within the atom
fortuitous
at the time the particle passes through.
random
directions
Thus the vectors 8 will have
and a distribution of lengths, as indicated in the figure.
3
PREDICTIONS OF THOMSON'S MODEL
Sec. 4]
97
The random nature of the &,-, and the fact that there are a large number
we apply the mathematical theory of statistics to the
of them, suggest that
problem of determining the length of the vector 8. Now the diagram in
figure (4-9) represents the deflections suffered by one alpha particle in its
traversal of the foil. If we were to observe the traversals of a number of
alpha particles and make similar diagrams for each of them, we could
evaluate two characteristic properties of these diagrams: the average
Figure 4-9. A two dimensional diagram representing the entire deflection process
experienced by an alpha particle in passing through a foil.
length of the vector 8, and the distribution of the lengths of this vector.
The theory of
statistics makes predictions concerning both of these
These predictions are particularly easy to obtain because we
have formulated the problem in such a way that it is identical with a well-
properties.
known
statistical
problem called the "random walk."t According to the
theory (d 2 YA the root-mean-square average length of the net deflection
vector 8, is related to the total number of individual deflections
and to
,
(dj)
lA
,
the root-mean-square average length of the individual deflection
vectors 8 3 by the equation
,
(J*f
It is interesting to
sjN{8f)
note that the average length of the vector 8 increases
because of the random directions of the 8
on the other hand, the 8 were all pointing in the same direction, which
just the opposite of being randomly directed, the average length of 8
only as the square root of
(If,
is
t See
any textbook on the theory of
statistics.
THE DISCOVERY OF THE ATOMIC NUCLEUS
98
would
be proportional to the
clearly
first
power of N.) The
[Ch. 4
theoretical
predictions of the distribution of the lengths of the vector 8 can be stated as
JT(6) dd
number of
e-**'
dd
The quantity ^V{d) dd
is
net deflection vector
within the range of values d to d
is
the
cases in which the length of the
dd;
d2
average value of the square of the length of the net deflection vectors
Jf
is
is
;
the
and
number of cases considered.
and 8, are equal, respectively, to the
angle O and the individual deflection angle p the equations
the total
Since the lengths of the vectors 8
net deflection
<f>
above can be immediately written in terms of the angles
jf(O)
rfO
(O a
The quantity
N(<&)
the angular range
is
JV
the
is
the
JO
to
is
the
= ^^ -''* dD
f = y/Nitfj*
(4-6)
number of alpha
O + JO O2 is
;
the
particles scattered within
mean
square scattering angle
mean square scattering angle in a deflection from a single atom;
number of alpha particles passing through the foil; and N\% the
number of atoms
foil.
(4-5)
traversed by an alpha particle in
These equations, plus the restriction
4>.j
<
its
passage through the
10~ 4 radians
(4-7)
that the angle of deflection in passing through a single
atom
is
very small,
comprise the predictions of Thomson's model of the atom for the scattering
of alpha particles traversing a thin foil.
5.
Comparison with Experiment
These predictions were tested by Rutherford and his collaborators. In
a typical experiment (Geiger and Marsden, 1909) alpha particles were
scattered by an Au foil 10~ 4 cm thick. The average scattering angle
was found to be about 1
2 x 10 -2 radians, and more than
99 percent of the alpha particles were scattered at angles less than about 3.
Now, the number of atoms traversed by the alpha particle is approximately
equal to the thickness of the foil divided by the diameter of an atom.
2
(4> )'^
Thus
#~ 10-
cm/10"8
cm
104
Knowing
N and
(<P) M ,
we can use
RUTHERFORD'S MODEL OF THE ATOM
Sec. 6]
99
equation (4-6) to evaluate the average deflection angle in traversing a
single atom. We obtain
,-HM
(<i?p
~2
2
= (W*
,__ ~
VN
* 10
->
2
10 *
x m-4
aradians
10
in reasonable agreement with the restriction stated in equation
Accurate measurements of ^T(O) d were difficult to obtain
because of the small angles involved. However, the available data were
This
is
(4-7).
in
agreement with equation (4-5), using
1 for
( 2) M
for angles less than
2 or 3.
On
the other hand, the angular distribution of the small
particles scattered at angles large
compared to
marked disagreement with equation
in
This equation predicts that
(4-5).
^V(<b) rfO will decrease very rapidly with increasing
we
number of
was observed to be
<J>.
For
instance, if
evaluate the predicted fraction of alpha particles scattered at angles
greater than 90,
we
find
fiso
1
JTifb
>
90)
Using the experimental value of
small
number jV{Q>
>
90)/^
>
J 90
J90"
1 for
e"*
(O 2) ^, we obtain the impressively
10" 3500 But experimentally it
90 * 2
In general, the number of
scattered alpha particles was observed to be very much larger than the
predicted number for all scattering angles greater than a few degrees.
was found that ^T(0
The
90)/^T
10-*.
existence of a small, but non-zero, probability for scattering at large
was totally inexplicable in terms of Thomson's model of the atom,
which could only lead to the multiple small angle scattering process
angles
described in section
6.
4.
Rutherford's Model of the
Atom
Experiments utilizing foils of various thicknesses showed that the
number of large angle scatterings was proportional to N, the number of
atoms traversed by the alpha particle. This is just the dependence on
that would arise if there was a small probability that an alpha particle
could be scattered through a large angle in traversing a single atom. Since
this cannot happen in Thomson's model of the atom, Rutherford proposed
a new model (1911).
In Rutherford's model of the structure of the atom, all the positive charge
THE DISCOVERY OF THE ATOMIC NUCLEUS
100
of the atom, and consequently essentially all its mass,
in a small region called the nucleus.
is
[Ch. 4
assumed
to
be
concentrated
From symmetry
considerations the nucleus was assumed to be located at
the center of thelitom, but
its
exact location played
no
role in Rutherford's
work.
The arguments of section 4
indicate that a large angle scattering
would
indeed be possible in the traversal of a single atom of such structure if the
alpha particle happened to pass very near the nucleus, provided that the
dimensions of the nucleus are small enough.
We
can get a preliminary
idea of the dimensions required from equation (4-4) since, as a review of
its
derivation will demonstrate,
it
can be used to estimate the scattering
angle of an alpha particle passing near (or through) the nucleus
interpret
to be the nuclear radius.
if
we
In the computation directly below
10~ 8 cm and obtained <
10~* radians.
equation (4-4), we used R
Consequently, to obtain a large angle scattering, </> <~ 1 radian, the value of
R must be ~104 times smaller; that is, R 10~ 12 cm. As will be indi-
cated later in this chapter, and discussed in detail in Chapter 16, 10~ 12
is, in fact, a good estimate of the radius of the atomic nucleus.
7.
cm
Predictions of Rutherford's Model
Rutherford made a detailed calculation of the angular distribution to be
expected for the scattering of alpha particles from atoms of the type
proposed in his model. The calculation was concerned only with scattering
than several degrees. Consequently, as we have seen
above, the scattering due to the atomic electrons can be ignored. The
at angles greater
scattering
is
then due to the repulsive
positively charged alpha particle
Coulomb
and the
force acting between the
positively charged nucleus.
Furthermore, the calculation considered only the scattering from heavy
atoms and thus permitted the assumption that the mass of the nucleus
is so large compared to the mass of the alpha particle that the nucleus does
not recoil (remains fixed in space) during the scattering process. It was
also assumed that the alpha particle does not actually penetrate the nuclear
region, so that the alpha particle and the nucleus act like point charges as
Coulomb force is concerned. We shall see later that all these
assumptions are quite valid except for the scattering of alpha particles
from the lighter nuclei. The calculation employs classical mechanics since
far as the
y/c~l/20.
+ze
+Ze. The nucleus is
Figure (4-10) illustrates the scattering of a particle, of charge
and mass M,
in passing near a nucleus of charge
PREDICTIONS OF RUTHERFORD'S MODEL
Sec. 7]
When
fixed at the origin of the coordinate system.
from the nucleus, the Coulomb force on
approaches the nucleus along a straight
move
After the scattering, the particle will
with constant velocity
radial coordinate r
drawn
v'.
The
line
very far
particle
with constant velocity
v.
off finally along a straight line
is specified by the
measured from an axis
position of the particle
and the polar angle
6,
which
is
parallel to the initial direction of motion.
+ ze
is
and so the
the particle
will vanish,
it
101
The perpendicular
"
1
Figure 4-10.
distance
The
+Ze
scattering of a positively charged particle in passing near a nucleus.
from that
axis to the line of initial
motion
is
specified
by the
impact parameter b. The scattering angle
is just the angle between the
axis and a line drawn through the origin parallel to the line of final motion,
and the perpendicular distance between these two lines is b'. We resolve
the velocity of the particle at any point along its trajectory into a radial
component of magnitude drjdt and a tangential component of magnitude
<f>
r (dd[dt).
Now,
the force acting on the particle
Consequently
That is,
its
angular
is
always in the radial direction.
has the constant value L.
momentum Mr 2 (ddjdt)
Mr 2 = L
(4-8)
dt
Specifically, the initial angular
momentum,
Mvb
Of
momentum
is
equal to the final angular
=L
(4-8')
or
Mv'b'
course, the kinetic energy of the particle does not remain constant
during the scattering.
However, the
energy must be equal
assumed to remain stationary.
initial kinetic
to the final kinetic energy since the nucleus
is
Thus
\Mv*>
\Mv' 2
THE DISCOVERY OF THE ATOMIC NUCLEUS
102
From
the last two equations
we
see that v
and b
v'
b',
[Ch. 4
as indicated
in figure (4-10).
Next, applying Newton's law to the radial component of motion,
we
have
,\2~|
The
left-hand term
Idt
Coulomb
the
is
(4-9)
\dtl.
force.
The
term in the bracket
first
and the second term
the radial acceleration due to the change in
r,
(radially directed) centrifugal acceleration.
To
is
the
effect the fastest solution
of this equation we transform from the coordinates
u, 6,
is
r,
6 to the coordinates
where
r
\\u
(4-10)
Then
_drdd
dr
dr du dd
dd dt'
du dd dt
_
'
dt
dr_
dt~
d
dd
dt
dd
'dr'
dd 2
we have
-dJ
The constant D,
l(Lu*\2
u\M/
dd 2
d u
_Jh?dSL _
2
d 2 u Lu 2
M dd M
d 2u
du
M dd
dd
L2 u 2
Substituting into equation (4-9),
~~
\-dt\ dt
d*r
dt
du Lu 2
M
-~ir
zZe
zZe 2u 2
M
M
zZe2
'
Mvb
2 2
(4-11)
defined by
D=
zZe 2
(4-12)
\Mv2
a convenient parameter equal to the distance of closest approach to the
nucleus in a head-on collision (b
is the distance at which the
0), since
is
potential energy zZe2 /D
is
point the particle would
equal to the
come
initial kinetic
energy
to a stop and then reverse
%Mv2
its
at this
direction of
motion. In terms of this parameter, equation (4-11) reads
d2 u\dd 2
= -
D\lb2
(4-13)
Sec. 7]
This
is
PREDICTIONS OF RUTHERFORD'S MODEL
a second order differential equation
vative) for u as
a function of
=A
6.
cos 6
is,
involving a second deri-
The general
solution to this equation is
+ B sin 6 -
D/2b*
A and
which contains two arbitrary constants,
equation (4-14)
(i.e.,
103
B.
(4-14)
We may
prove that
in fact, the solution to (4-13) by evaluating
= -A
du/dd
+ B cos 6
sin 6
and
= -A cos 6 - B sin
dhi/dd 2
and
substituting into (4-13). This gives
-A cos 6 - B sin +
To determine A and
initial conditions:
we
5,
+ B sin 6 -
cos 6
.4
-*
=i.
= - D/262
require that equation (4-14)
as r -> oo,
2>/2Z>
and
/lcosO
</r/</r
->
BsinO
Q.E.D.
conform to the
oo. Thus
v as r
>-
2b 2
or
A=
2b*
and
*!
= _ k iH = _ = _
k.
(-a
sin
+ B cos 0)
or
_ Mv _ Mv _
~ L ~ Mvb ~
Therefore the particular solution of the differential equation pertinent to
the problem at
hand
is
D=
cos a J*
+ - sin
2b
D
2b 2
or
This
is
sin 6
+ . (cos 0-1)
(4-15)
2b
the equation of a hyperbola in polar coordinates.
We
see that the
trajectory of the particle moving under the influence of a repulsive inverse
square Coulomb force is one of the conic sections. The trajectory of a
particle moving under an attractive inverse square force (e.g., a satellite
in the gravitational field of the earth) could be obtained
from the derivation
THE DISCOVERY OF THE ATOMIC NUCLEUS
104
[Ch. 4
above by simply reversing the sign of the term on the left side of equation
(4-9). The trajectory would then turn out to be one of the conic sections
exactly which one depending on the values of the parameters.
From
by
<f>
Irinfl'
sin 6'
letting r
-=--
solution to this equation
r before scattering.
The
is 0'
cos
is
figure (4-10) that for this solution
1)
the polar angle for large
is
6'
tan
sin 6'
which gives the polar angle for large
r after scattering.
<f>
6'.
It is
apparent from
Consequently
h=) = 2*
tan
is
6'
Then
oo.
This
0.
interesting solution
2b
=
which
(cos 6'
2b 2
One
we may
the equation (4-15) of the trajectory of the particle,
evaluate the scattering angle
equivalent to
=^
2D
cot^
It will
(4-16)
also be useful to evaluate R, the distance of closest approach of
the particle to the center of the nucleus (the origin).
it is
apparent that the radial coordinate
polar angle
is
for this angle,
equal to d'/2
(tt
will
</>)/2.
assume
From
figure (4-10)
this value
when
the
Evaluating equation (4-15)
and using (4-16), we obtain
1
(4-17)
sin (</2).
-> n, and that R -> b as
-> 0, as
easy to verify that R - D as
would be expected.
We see from equation (4-16) that, in the scattering of an alpha particle
by a single nucleus, if the impact parameter is in the range b to b + db
the scattering angle is in the range
to
+ dj>, where the relation between
b and
is given by the equation.
Thus the problem of calculating the
It is
(f>
</>
<f>
<f>
</>
number of alpha
particles scattered in the angular range
<f>
to
<f>
d<f>
in traversing the entire foil is equivalent to the
problem of calculating the
with impact parameter from b to b + db,
number which are incident,
upon the nuclei in the foil. Let
the number of nuclei (i.e., the number of
atoms) per cm 3 in the foil be p, and let the foil be t cm thick. Consider a
segment of the foil with a cross-sectional area of 1 cm2 as shown in figure
,
Sec. 7]
PREDICTIONS OF RUTHERFORD'S MODEL
105
This diagram indicates a ring, of inner radius b and outer radius
around an axis passing through each nucleus. The area of
drawn
b + db,
2-nb
db. The number of such rings in the segment is pt, where
is
each ring
equal to the volume of the segment. The probability
numerically
t is
P(b) db that an alpha particle will pass through one of these rings is equal
(4-11).
Incident
alpha
particles
Figure 4-11.
are very
much
A beam
upon a
of alpha particles incident
smaller and there are very
to the total area obscured
many more
cm 2
rings, as seen
by the
particles, divided by the total area of the segment.
P(b) db
pt2irb
foil.
In actuality
the rings
of them.
by the incident alpha
That is,
db
Now
u
b
D
=
cot 2
so
D
db= - ~2
\d$
sin
(#2)
and
>
bdb
cos
-~Y
(<ft/2)
S in
d<j>
= _
-g
sin
16sin
(#2)
</>
d<j>
(</2)
Thus
will
particles
But -P(b) db is equal to the probability that the incident
sign compenminus
(The
to
+
range
d<f>.
angular
the
in
scattered
be
opposite sign.) Using the
sates for the fact that db and d<f> are of the
notation employed in equation (4-5), this is
tf>
<f>
^m^
= _
jV
P{b)db
- ptD
sin 3)
sin (0/2)
106
THE DISCOVERY OF THE ATOMIC NUCLEUS
Evaluating D,
we have
8^
This
is
\1mW
[Ch. 4
sin (<t>/2)
the angular distribution derived by Rutherford for the scattering
composed of atoms with the
The quantity Jf is the number
of alpha particles incident on the foil of thickness t and p nuclei per cm3
^(O) d<& is the number scattered at angles from $ to O + c/O ze is the
charge of the alpha particle;
is its mass; v is its velocity; and Ze is
of alpha particles in passing through a
foil
nuclear structure proposed by his model.
the charge of the nuclei.
We note
that, although the angular factor in this
equation decreases rapidly with increasing angle, nevertheless the decrease
is
very
much
less rapid
than the decrease predicted by the multiple small
angle scattering angular distribution (equation 4-5).
8.
Experimental Verification and the Determination of
Detailed experimental tests of equation (4-18) were performed by Geiger
and Marsden within a few months
after its derivation (1911).
They investi-
gated the following points.
tested, using foils of Ag and Au, over
Although JV{<&) d<& varies by a factor of
about 10s over this range, the experimental data remained proportional
to the theoretical angular distribution to within a few percent.
2. The quantity jV{<&) afO was found indeed to be proportional to t
for a range of thickness of about 10 for all the elements investigated.
3. Equation (4-18) predicts that the number of scattered alpha particles
1.
The angular dependence was
the angular range 5 to 150.
will
be inversely proportional to the square of their kinetic energy,
This was tested by using alpha particles from several different
\m&.
radioactive elements.
The predicted energy dependence was observed
over an energy variation of about a factor of
4.
Finally,
the equation predicts JV($>)
(Ze) 2 the square of the nuclear charge.
,
3.
At the
to
be proportional to
time,
Z was not known for
the various atoms, except that, as mentioned in section
1,
X-ray scattering
was roughly equal to A/2, where A is the
chemical atomic weight of the atom. The validity of equation (4-18)
experiments had shown that
being assumed, the experimental data could be used to determine
was known. The resulting values agreed with the law
Z A/2. Furthermore it was observed that, to within the accuracy of
the experiment, the measured values of Z were equal to the chemical
atomic number of the atoms in question, i.e., equal to the ordering number
since everything else
THE
Sec. 9]
OF THE NUCLEUS
SIZE
of the atoms in the periodic
This implies that the
table.
in the periodic table contains
107
2 electrons, the third atom, Li, contains 3 electrons,
number of
first
atom, H,
electron, the second atom, He, contains
etc.,'
since
is
the
This was shortly confirmed by very
using X-ray techniques, which will be
electrons in the atom.
accurate determinations of
described in Chapter 14.
The Size
9.
By
of the Nucleus
evaluating R, the distance of closest approach, Rutherford was able
on the
to place limits
Ra
of
it
of the nucleus. As an example, in the scattering
size
alpha particles
(v
10 9 cm/sec)
1.6
was observed that the Rutherford
largest scattering angle investigated,
we
see that
takes
R^an
on
from Cu (A = 63, Z = 29)
law was obeyed to the
scattering
<1>
From
180.
smallest value at that angle.
its
= D=
equation (4-17)
So
zZe2
2 1V1U
*Mftn
180
V=
x 29 x
x 1CT
|(4
1-7
X 10" 12 cm
1.67
(4.8
24
)
x IP" 10)2
x
(1.6
x 109)2
Rutherford's derivation depends on the assumption that the force acting
particle is always strictly a Coulomb force between two point
on the alpha
charges.
This would not be true
region at
its
if
the particle penetrated the nuclear
point of closest approach.
the radius of the
Cu
nucleus
is
<
.7
The obvious implication
is
that
10 -12 cm.f
The equation above shows that R decreases as Z decreases. The question
arises
How much can R decrease before R <
nuclear radius ? Departures
light (low Z)
were actually observed in the scattering
nuclei, of
very
light
violation,
for
the
was
due
to
nuclei. Part of this
a
the assumption that the nuclear mass is large compared to the alpha
particle mass. However, deviations remained even after the finite nuclear
mass was taken into account in the theory. Figure (4-12) shows some data
for the scattering of alpha particles, of various velocities, at a fixed large
from the very
We
may also conclude that the use of classical mechanics to calculate the angular
t
distribution for scattering from a Coulomb field is valid, even for collisions in which
is
as small as JO
for the case of a
give only
-12
cm. In Chapter 15 we
Coulomb
shall find that
force, but that for
an approximation
quantum theory
any other force
in such collisions.
classical
verifies this
mechanics would
THE DISCOVERY OF THE ATOMIC NUCLEUS
108
[Ch. 4
angle from an Al
foil. The ordinate is the ratio of the observed number
of scattered particles to the number predicted by the Rutherford theory
(corrected for the finite nuclear mass). The abscissa is the distance of
approach calculated from equation (4-17).
12
is about 10~
cm.
closest
These data imply
that the radius of the Al nucleus
(J
1
*'-'
2?
>"x
CD
sz
ye,/
g,
0.5
X
1
R (in
Figure 4-12.
10.
Some
10
units of
12
14
16
10" 13
18
cm)
data obtained in the scattering of alpha particles from aluminum.
Problem
The detailed experimental verification of the predictions of the nuclear
model of the atom left little room for doubt concerning the validity of the
model. However, the question of the stability of such an atom presented a
serious problem.
we assume
atom are stationary, it is easy to
no stable arrangement of the electrons surrounding the
nucleus which would prevent the electrons from falling into the nucleus
under the influence of its Coulomb attraction. We cannot allow the atom
to collapse, because then its radius would be of the order of a nuclear
radius, which is four orders of magnitude smaller than what we know the
radius of an atom to be.
At first glance it appears that we can simply allow the electrons to
If
that the electrons in the
see that there exists
about the nucleus in orbits similar to the orbits of the planets
Such a system can be stable mechanically
because the centrifugal force can be made to balance the Coulomb
circulate
circulating about the sun.
attraction just as
system.
But a
it
balances the gravitational attraction in the planetary
difficulty arises in trying to carry
planetary system to the atomic system.
electrons
over this idea from the
The problem
would be constantly accelerating
in their
is
that the charged
motion around the
Ch.
EXERCISES
4]
109
nucleus and, according to the classical electromagnetic theory,
ating charged bodies radiate energy in the
all acceler-
form of electromagnetic
radia-
The energy would be emitted at the expense of the mechanical
energy of the electron, and the electron would spiral into the nucleus.
Again we have an atom which would rapidly collapse to nuclear dimention.
sions.
Furthermore, the continuous spectrum of the radiation which
is not in agreement with the discrete spectrum which
would be emitted
known
is
to be emitted
by atoms.
BIBLIOGRAPHY
Evans, R. D., The Atomic Nucleus, McGraw-Hill Book Co., New York, 1955.
Rutherford, E., J. Chadwick, and C. D. Ellis, Radiations from Radioactive
Substances, Cambridge University Press, Cambridge, England, 1930.
EXERCISES
1. Use classical mechanics to prove the statement, made in section 4, that in
a collision of a very massive body, initially moving at velocity v, with a free
stationary body of small mass, the velocity of the light body after the collision
can be no greater than
2v.
Hint:
Make
a Galilean transformation to the
system, treat the collision, then transform back to the
2.
Derive equation (4-8) from Newton's law.
3.
Adapt the
an
LAB
calculation of section 7 to the case of a particle
attractive inverse square force.
Use the
CM
system.
moving under
motion of a
results to discuss the
satellite.
4.
-<<f>
Verify from equation (4-17) that
0.
-*
as ^ -> n,
Give physical arguments explaining these
and that
-> b as
results.
5. Adapt the calculation of section 7 to the case of an electron moving in a
hydrogen atom, under the assumption that it radiates energy at the rate given
by equation (2-7) because it is accelerated. If the radius of its orbit is initially
10~ 8 cm, how long would it take for it to spiral into the nucleus of radius 1 -12 cm ?
CHAPTER
Bohr's Theory of
Atomic Structure
I.
Atomic Spectra
The
difficult
problem of the
theory of atomic structure.
of atoms actually led to a simple
key feature of this very successful theory,
stability
proposed in 1913 by Bohr, was the prediction of the spectrum of electromagnetic radiation emitted by certain atoms. Because of this prediction,
and because we have mentioned atomic spectra on two occasions in the
previous chapter,
it is
necessary at this point to describe
complex features of such
is,
spectra.
A detailed
however, best deferred until after
some of the
less
discussion of atomic spectra
we have developed
the theory of
quantum mechanics.
typical apparatus used in the
measurement of atomic spectra
is
The source consists of an electric discharge
passing through a region containing a monatomic gas. Owing to collisions
with electrons, and with each other, some of the atoms in the discharge are
indicated in figure (5-1).
put into a state in which their total energy is greater than it is in a normal
atom. In returning to their normal energy state, the atoms give up their
by emitting electromagnetic radiation. The radiation is
slit, and then it passes through a prism (or diffraction
consequently breaking up into its spectrum, which is recorded
excess energy
collimated by the
grating),
on the photographic plate.
The nature of the observed
spectra is indicated by the front view of the
photographic plate. In contrast to the continuous spectrum of electromagnetic radiation emitted, for instance, from the surface of solids at
high temperature, the electromagnetic radiation emitted by a free atom is
110
Sec.
ATOMIC SPECTRA
I]
III
concentrated at a number of discrete wavelengths. Each of these wavelength
components
is
photographic
kinds of atoms,
istic
spectrum,
which
called a line because of the line
Upon investigating the
plate.
it is
i.e.,
it
produces on the
spectra emitted
from
different
own character-
observed that each kind of atom has
its
characteristic set of wavelengths at
which the
the spectrum are found.
This feature
is
lines
of
of great practical importance
Source
Slit
Front view of
photographic plate
An
Figure 5-1.
because
it
apparatus used to measure atomic spectra.
makes spectroscopy a very
niques of chemical analysis.
useful addition to the usual tech-
Primarily as a result of this motivation,
considerable effort was devoted in the nineteenth century to the accurate
measurement of atomic
spectra.
And,
was
and in
in fact, considerable effort
required because the spectra consist of
many hundreds of
lines
general are very complicated.
Designation
of line
HR
Hy
in
X(A)
5fi
CO
"Color"
Red
Blue
H$ HgHg
Hoe
o do
* en
m
m 00
3
to
7-i
Violet
Near
ultraviolet
Figure 5-2.
The
visible part of
the hydrogen spectrum.
However, the spectrum of hydrogen is relatively simple. This is perhaps
not surprising since hydrogen, which contains just one electron, is itself
the simplest atom. The
spectrum is of considerable interest, not only
because of its simplicity, but also because of historical and theoretical
reasons which will become apparent later. Figure (5-2) represents that part
BOHR'S THEORY OF ATOMIC STRUCTURE
112
of the
H spectrum which falls approximately within
[Ch. 5
the wavelength range
The wavelength A of the lines is measured in units of 1
Angstrom = 1 A = 10-8 cm; the unit is named after a physicist who was
responsible for some of the first precise spectroscopic measurements.
of visible
light.
We
see that the spacing, in wavelengths, between adjacent lines of the
spectrum continuously decreases with decreasing wavelength of the lines,
so that the series of lines converges to the so-called series limit at 3645.6 A.
The short wavelength lines, including the series limit, are hard to observe
experimentally because of their close spacing and because they are in the
ultraviolet.
The obvious regularity of the
spectrum tempted several people to
look for an empirical formula which would represent the wavelength of
the lines. Such a formula was discovered in 1885 by Balmer. He found
that the simple equation
A
3646
where n
3 for
Ha
the wavelength of the
known
4 for
p,
(in
units)
(5-1)
5 for
etc.,
was able to predict
nine lines of the series, which were
first
at the time, to better than
all
that were
part in 1000. This discovery initiated
a search for similar empirical formulas that would apply to series of lines
which can sometimes be
identified in the complicated distribution of lines
Most of this work was done
convenient to deal with the wave
that constitute the spectra of other elements.
by Rydberg
mmber
(ca.
1890),
who found
'
it
B MX
of the lines, instead of their wavelength.
(5-2)
In terms of this quantity, the
Balmer formula can be rewritten
k
where
RK
is
2? H (l/2 2
l/ 2 ),
3, 4, 5,
(5-3)
the so-called Rydberg constant for hydrogen.
spectroscopic data,
its
value
Ru =
is
known
109677.576
From
recent
to be
0.012 cm-
(5-4)
This
is indicative of the accuracy possible in spectroscopic measurements.
Formulas of this type were found for a number of series. For instance,
we now know of the
existence of five series of lines in the
shown in table (5-1). For
same general structure:
where
is
the
alkali
H spectrum,
as
elements the series formulas are of the
Rydberg constant for the
particular element, a
and b are
Sec. 2]
BOHR'S POSTULATES
||3
TABLE
(5-1)
Names, Wavelength Ranges, and Formulas for the Hydrogen Series
Lyman
Ultraviolet
= i?H /p - -\,
n =2,3,4,..
Balmer
Near ultraviolet
and visible
= RK (
Paschen
Infrared
=RK (^--A,
n=4,5,6,..
Brackett
Infrared
= Ra (
n= 5,6,1,..
= Rs ( ^ - -% \
Pfund
Infrared
constants for the particular
particular series,
and n
is
series,
is
2%
-^
3, 4, 5,
'
- -^ \
6, 7, 8,
an integer which is fixed for the
To within about 0.05 percent
a variable integer.
the Rydberg constant has the same value for
all elements, although it does
with increasing atomic weight.
We have been discussing the emission spectrum of an atom.
closely
related property is the absorption spectrum. This may be measured with
apparatus similar to that shown in figure (5-1) except that a source
show a very
slight systematic increase
emitting a continuous spectrum
is used and a glass-walled cell, containing
monatomic gas to be investigated, is inserted somewhere between the
source and the prism. After exposure and development, the photographic
plate is found to be darkened everywhere except for a number of unexposed lines. These lines represent a set of discrete wavelength components which were missing from the otherwise continuous spectrum
incident upon the prism, and which must have been absorbed by the atoms
the
in the gas
cell. It is observed that for every line in the absorption spectrum
of an element there is a corresponding (same wavelength) line in its emission
spectrum. However, the reverse is not true. Only certain emission lines
show up
in the absorption spectrum. For hydrogen gas, normally only
corresponding to the Lyman series appear in the absorption spectrum;
but, when the gas is at very high temperatures, e.g., at the surface of a star,
lines corresponding to the Balmer series are found.
lines
2.
Bohr's Postulates
All these features of atomic spectra,
discussed,
must be explained by any
and many more which we have not
successful theory of atomic structure.
Furthermore, the very great precision which characterizes spectroscopic
BOHR'S THEORY OF ATOMIC STRUCTURE
114
[Ch. 5
measurements imposes particularly severe requirements on the accuracy
with which such a theory must be able to predict the quantitative features
of the spectra.
Nevertheless, in 1913
Bohr developed a theory which was
in accurate
quantitative agreement with certain of the spectroscopic data (e.g., the
hydrogen spectrum). This theory had the additional attraction that the
mathematics involved was very easy to understand. However, the postulates upon which it was based were not so transparent. These postulates
were:
1.
An
electron in an
atom moves
in
a circular orbit about the nucleus
under the influence of the Coulomb attraction between the electron and the
nucleus, and obeying the laws of classical mechanics.
2.
But, instead of the infinity of orbits which would be possible in classical
mechanics,
its
h, divided
only possible for an electron to
momentum L
an
is
move
an orbit for which
of Planck's constant
in
integral multiple
by 2n.
Despite the fact that
3.
in
it is
orbital angular
it is
constantly accelerating, an electron moving
such an allowed orbit does not radiate electromagnetic energy.
total energy
E remains
Thus
its
constant.
Electromagnetic radiation is emitted if an electron, initially moving
an orbit of total energy E t discontinuously changes its motion so that
it moves in an orbit of total energy E . The frequency of the emitted radiation
f
4.
in
is
equal to the quantity (Et
The
Ef
divided by Planck's constant h.
on the existence of the atomic
embodies some of the ideas concerning the stability of a
nuclear atom discussed at the end of the preceding chapter. The second
postulate introduces quantization. The reader should note the difference
between Bohr's quantization of the orbital angular momentum of an atomic
electron moving under the influence of an inverse square (Coulomb)
first
nucleus.
postulate bases Bohr's theory
It
force,
L=
nh\2iT,
1, 2, 3,
(5-6)
and Planck's quantization of the total energy of a particle, such as an
executing simple harmonic motion under the influence of a
electron,
harmonic restoring force: E
nhv, n
We shall see in the
0,1,2, ...
next section that the quantization of the orbital angular momentum of the
atomic electron does lead to the quantization of its total energy, but with
.
an energy quantization equation which is different from Planck's equation.
The third postulate removes the problem of the stability of an electron
moving in a circular orbit, due to of the emission of the electromagnetic
radiation required of the electron by classical theory, by simply postulating
BOHR'S THEORY OF THE ONE-ELECTRON
Sec. 3]
that this particular feature of the classical theory
is
ATOM
115
not valid for the case
of an atomic electron. The postulate was based on the fact that atoms are
observed by experiment to be stable
by the
classical theory.
The fourth
v
even though
this is
not predicted
postulate,
= Ei~ Et
(5 _7)
h
is
obviously closely related to Einstein's postulate (equation 3-7) that
the frequency of a
quantum of electromagnetic
radiation
is
equal to the
energy carried by the quantum divided by Planck's constant.
These postulates do a thorough job of mixing
classical and non-classical
moving in a circular orbit is assumed to obey
classical mechanics, and yet the non-classical idea of quantization of
orbital angular momentum is included. The electron is assumed to obey
one feature of classical electromagnetic theory (Coulomb's law), and yet
not to obey another feature (emission of radiation by an accelerated
physics.
The
electron
we should not be surprised if the laws of classical
which are based on our experience with macroscopic systems, are
charged body). However,
physics,
not completely valid when dealing with microscopic systems such as the
atom.
3.
Bohr's Theory of the One-Electron
Atom
The justification of Bohr's postulates, or of any set of postulates, can
be found only by comparing the predictions that can be derived from the
postulates with the results of experiment. In this section we derive some of
and compare them with the data of section 1
Consider an atom consisting of a nucleus of charge +Ze and mass M,
and a single electron of charge e and mass m. For a neutral hydrogen
atom Z = 1, for a singly ionized helium atom Z = 2, for a doubly
ionized lithium atom Z = 3, etc. We assume that the electron rotates in a
circular orbit about the nucleus. Initially we suppose the mass of the
these predictions
compared to the mass of the nucleus,
and consequently assume that the nucleus remains fixed in space. Thus
we consider at first the situation depicted in figure (5-3). The condition of
electron to be completely negligible
mechanical
stability
of the electron
is
= m2
where
(5-8)
v is the velocity of the electron in its orbit,
and
r is
the radius of the
BOHR'S THEORY OF ATOMIC STRUCTURE
116
[Ch. 5
The left side of equation (5-8) is the Coulomb force acting on the
and the right side is the centrifugal force required to keep it in
orbit.
electron,
its
Now,
circular orbit.
Figure 5-3.
An
the orbital angular
momentum
of the electron
electron moving in a circular orbit about a fixed nucleus.
must be a constant because the force acting on the electron
the radial direction.
It is
is
entirely in
equal to
L = mvr
(5-9)
Applying the quantization condition (5-6) to L, we have
mvr
or, in
nhjln,
1, 2, 3,
terms of the convenient symbol
h
mvr
h/2n
n
nh,
(5-10)
1,
2, 3,
Solving for v and substituting into equation (5-8),
Ze2
mrv2
mrn 2h2\m 2r 2
(5-11)
we obtain
= rpfpjmr
So
r
nWjmZe*,
1, 2, 3,
(5-12)
and
v
nhmZe2 lmn2 FP
= Ze \nh,
2
n=
1, 2, 3,
(5-13)
The application of the angular momentum quantization condition has
restricted the possible circular orbits to those
Note that these
of radii given by equation
of the quantum
Let us evaluate the radius of the smallest orbit (n
1) for a
hydrogen atom (Z
Inserting the known values of H, m, and e, we
1).
obtain r = 5.3 x lO" 9 cm. It is shown below that the electron has its
minimum total energy when in the orbit corresponding to n
1.
Consequently we may interpret the radius of this orbit as a measure of the
(5-12).
number
radii are proportional to the square
n.
radius of a hydrogen atom in its normal state. It is in good agreement
with the estimate, mentioned previously, that the order of magnitude of
an atomic radius is 10 -8 cm. Evaluating the orbital velocity of an electron
BOHR'S THEORY OF THE ONE-ELECTRON
Sec. 3]
ATOM
117
hydrogen atom from equation (5-13), we find
apparent from the equation that this is the
largest velocity possible for an electron in a hydrogen atom. The fact
that this velocity is less than 1 percent of the velocity of light is the justification for using classical mechanics instead of relativistic mechanics in
the Bohr theory. On the other hand, equation (5-13) shows that for large
values of Z the electron velocity becomes relativistic. The theory could not
be applied in such cases.
Next we calculate the total energy of an atomic electron moving in
one of the allowed orbits. Let us define the potential energy to be zero
when the electron is infinitely distant from the nucleus. Then the potential
energy V at any finite distance r can be obtained by integrating the energy
imparted to the electron by the Coulomb force acting from oo to r. Thus
in the smallest orbit of a
2.2
108 cm/sec.f
It is
_
-j: r
Ze2
dr
Ze2
>
The
The
potential energy
is
negative because the
Coulomb
force
attractive.
is
kinetic energy of the electron, T, can be evaluated, with the aid of
equation (5-8), to be
T=
The
= Ze*llr
total energy of the electron, E, is then
E= T+
V=
-Zc*l2r
= -T
Using equation (5-12), we have
We
-Z(*mZe*l2n*?i 2
-/wZ*e*/22 ft2
1, 2, 3, 4,
see that the quantization of the orbital angular
electron leads to a quantization of
contained in equation (5-14)
figure (5-4).
is
is
The energy of each
shown on the
left in
its
total energy.
momentum
(5-14)
of the
The information
presented as an energy level diagram in
level, as
evaluated from equation (5-14),
terms of ergs and also in terms of electron volts
5, and the quantum
shown on the right. The diagram is so constructed
that the distance from any level to the zero energy level is proportional
to the energy of that level. Note that the lowest (most negative) allowed
value of total energy occurs for the smallest quantum number n = 1.
As n increases, the total energy of the quantum state becomes less negative,
(ev),
a unit of energy which will be explained in section
number of the
with
level is
E approaching
est total
energy
is,
zero as n approaches infinity. Since the state of lowof course, the most stable state for the electron, we see
t Equation (5-13) explains why Bohr could not allow the quantum number n ever to
assume the value n = 0, as it may in Planck's quantization equation.
BOHR'S THEORY OF ATOMIC STRUCTURE
118
[Ch. 5
that the normal state of the electron in a one-electron atom is the state for
which n = 1.
Next we calculate the frequency v of the electromagnetic radiation
emitted when the electron makes a transition from the quantum state
n i to the quantum state nf that is, when an electron initially moving in an
orbit characterized by the quantum number n t discontinuously changes its
,
1.36
x 10
-12
-12
erg
= - 0.85
erg
ev
4
3
2.41
x 10
= -
1.51 ev
5.42
x 10 -12 erg = -
3.39 ev
- 21.7 x 10" 12 erg = -
13.6 ev
An energy
Figure 5-4.
diagram for the hydrogen atom.
level
motion so that it moves in an orbit characterized by quantum number nf
Using Bohr's fourth postulate (equation 5-7), and equation (5-14), we
have
.
E -E,
t
4ttH
h
In terms of the wave
number k
k
mZ (1.1)
3
1/A
vie, this is
= JL3 z *(l - 1)
\n 2f
4nch
ni
or
\n 2f
where n f and nf are
integers.
ny
4vcfi
ATOM
BOHR'S THEORY OF THE ONE-ELECTRON
Sec. 3]
119
Bohr theory are contained in equations
and (5-15). Let us first discuss the emission of electromagnetic
radiation by a one-electron Bohr atom in terms of these equations.
1. The normal state of the atom will be the state in which the electron
The
essential predictions of the
(5-14)
has the lowest energy,
the state n
i.e.,
This
1.
called the
is
ground
state.
2.
etc.
In an electric discharge the atom receives energy due to collisions,
make a transition to a state of higher
This means that the electron must
>
energy, or excited state, in which n
common
1.
all physical systems, the atom will
and return to the ground state. This is accomplished
by a series of transitions in which the electron drops to states of successively
lower energy, finally reaching the ground state. In each transition electromagnetic radiation is emitted with a wave number which depends on the
energy lost by the electron, i.e., upon the initial and final quantum
3.
emit
Obeying the
its
tendency of
excess energy
numbers. In a typical case, the electron might be excited into state n 1
and drop successively through the states n = 4 and n = 3 to the ground
state n \. Three lines of the atomic spectrum are emitted with wave
numbers given by equation (5-15) for n i = 1 and nf 4, n t = 4 and
3,
4.
In the very large
nf
and n i
3>
and nf
1.
number of
excitation
and de-excitation processes
which take place during a measurement of an atomic spectrum, all possible
transitions occur and the complete spectrum is emitted. The wave numbers
of the set of lines which constitute the spectrum are given by equation
(5-15), where we allow n t and n t to take on all possible integral values
subject only to the restriction that n i
>
nf
For hydrogen (Z = 1) let us consider the subset of
arises from transitions in which n t = 2. According
the wave numbers for these lines are
k
= RMIn 2, -
1/nf),
n,
2,n t
spectral lines
which
to equation (5-15)
>
nt
or
2
RJ1/2
1/n
2
),
3, 4, 5, 6,
This is identical with the series formula for the Balmer series of the hydrogen
spectrum (equation 5-3), if R x is equal to i? H According to the Bohr
tn^lAnct?. Although the numerical values of some of the
theory, R
.
quantities entering into this equation
were not very accurately known at the
Bohr evaluated R x in terms of these quantities and found that the
resulting number was in quite good agreement with the experimental
time,
In the next section we shall make a detailed comparison,
using recent data, between the experimental value of i? H and the prediction
of Bohr's theory, and show that the two agree almost perfectly.
value of
Ru
BOHR'S THEORY OF ATOMIC STRUCTURE
120
[Ch. S
According to the Bohr theory, each of the five known series of the
hydrogen spectrum arises from a subset of transitions in which the electron
For the Lyman series n t = 1
Balmer n t = 2; for the Paschen n, = 3; for the Brackett n f = 4;
and for the Pfund nf = 5. These series are conveniently illustrated in
terms of the energy level diagram of figure (5-5). The transition giving
rise to a particular line of a series is indicated in this diagram by an arrow
goes to a certain final quantum state n t
for the
Balmer
-n =
Lyman
Figure 5-5.
Illustrating the
production of the
five
known
series of the
hydrogen
spectrum.
going from the initial quantum state n t to the final quantum state n
t
Only the arrows corresponding to the first few lines of each series are
shown. Since the distance between any two energy levels in such a diagram
.
proportional to the difference between the energy of the two
levels, and
wave number k is
proportional to the energy difference, the length of any arrow is proportional to the frequency or wave number of the corresponding spectral
is
since equation (5-7) states that the frequency v or
line.
The wave numbers of the lines of all these series are fitted very accurately
by equation (5-15) by using the appropriate value of n f This was a great
triumph for Bohr's theory. The success of the theory was particularly
impressive because the Lyman, Brackett, and Pfund series had not been
discovered at the time the theory was developed by Bohr. The existence of
these series was predicted by the theory, and the series were soon observed
experimentally by the gentlemen they are named after.
The theory worked equally well when applied to the case of oneelectron atoms with Z = 2, i.e., singly ionized helium atoms He+. Such
atoms can be produced by passing a particularly violent electric discharge
.
BOHR'S THEORY OF THE ONE-ELECTRON
Sec. 3]
ATOM
121
through normal helium gas. They make their presence apparent
by emitting a simpler spectrum than that emitted by normal helium atoms.
In fact, the atomic spectrum of He+ is exactly the same as the hydrogen
spectrum except that the wave numbers of all the lines are almost exactly
four times as great. This is explained very easily, in terms of the Bohr
(a spark)
theory,
The
by
setting
Z2 =
4 in equation (5-15).
properties of the absorption spectrum of one-electron atoms are
Bohr theory. Since the atomic
of one of the
the atom can only absorb discrete amounts of
also easy to understand in terms of the
electron
must have a
allowed energy
total energy exactly equal to the energy
states,
energy from the incident electromagnetic radiation. This fact leads to the
idea that we consider the incident radiation to be a beam of quanta, and
that only those quanta can be absorbed whose frequency is given by
E=
where E is one of the discrete amounts of energy which can be
absorbed by the atom. The process of absorbing electromagnetic radiation
is then just the inverse of the normal emission process, and the lines of the
absorption spectrum will have exactly the same wave numbers as the lines
of the emission spectrum. Normally the atom is always initially in the
ground state n = 1, so that only absorption processes from n = 1 to
n > 1 can occur. Thus only the absorption lines which correspond (for
hydrogen) to the Lyman series will normally be observed. However, if the
hv,
gas containing the absorbing atoms
owing
to collisions,
at a very high temperature, then,
is
some of the atoms
will initially
be in the
first
excited
corresponding to the Balmer series will
2,
be observed. The required temperature can be estimated from the Boltzmann probability distribution (equation 2-23), which shows that the ratio
state
and absorption
lines
of the probability of finding an atom in the state n
of finding it in the ground state n = 1 is
2 to the probability
p -En = zlkT
En = dkT
Boltzmann's constant = 1.38 x 10" 16 ergs/K. For hydrogen
the energies of these two states are given in the energy level diagram (figure
where k
5_4):
is
En=1
21.7
pw~2 =
x 10- 12
ergs,
En=2 =
-( -5.42 + 21.7) xlo~
(1.38 x
la
-5.42 X 1(H2
Thus
-1.18 x 10 "K
ergs
NT 18 ergs/'-Kjr
ergs.
Pn=l
apparent that a significant fraction of the hydrogen atoms will
initially be in the n = 2 state only when T is of the order of, or greater
than, 10 5 K. As mentioned before, Balmer absorption lines are actually
observed in the hydrogen gas of some stellar atmospheres. Thus we have
It is
one way of estimating the temperature of the surface of a
star.
BOHR'S THEORY OF ATOMIC STRUCTURE
122
4.
[Ch. 5
Correction for Finite Nuclear Mass
We have assumed the mass of the atomic nucleus to be infinitely large
compared to the mass of the atomic electron, so that the nucleus remains
fixed in space. This is a good approximation even for hydrogen, which
contains the lightest nucleus, since the mass of that nucleus is about 2000
times larger than the electron mass. However, the spectroscopic data are
CM
The center of mass of an electron-nucleus system.
Figure 5-6.
so very accurate that before
we make a detailed numerical comparison
we must take into account the fact that
of these data with the Bohr theory
the nuclear mass
is
actually finite.
Consider the electron-nucleus system shown in figure (5-6). The electron
of mass m and the nucleus of mass
must both move so that the center
of mass (CM) remains fixed in space. Let r be the distance between the
nucleus and the electron.
CM
is
Then the
m+
If the electron rotates
must also
about the
rotate about the
and
from the nucleus to the
distance x
determined by the equation m{r
x)
Mx;
so
m+
(5-16)
CM with angular velocity
co,
the nucleus
CM
with the same angular velocity. Let us
calculate the total orbital angular momentum L of such a system. We
have
L = m(r
2
= mM r co
(m + Mf
2
Marco
Mm r
2 2
xfco
o)
(-
+ Mf
(m
So
mM(m
,
+ M)
i
L ro^(o =
L=
(m
As
Mjm -
+ Mf
mM
m+
L -> {mMjM)r 2 m and
._
._.
(5-17)
x -- 0. Thus, for an infinitely
heavy nucleus the electron rotates about the stationary nucleus with orbital
angular momentum L = mr 2 co. This is just what we had before in
oo,
equation (5-9), since
rco is
equal to the orbital velocity
v.
CORRECTION FOR
Sec. 4]
To handle
NUCLEAR MASS
FINITE
Bohr modified
the case at hand,
his
123
second postulate to read
Instead of the infinity of orbits which would be possible in classical
it is only possible for an electron to move in an orbit for which
mechanics,
the total orbital angular
Planck's constant
h,
momentum of the atom,
divided by
L,
is
an integral multiple of
2tt.
This obviously reduces to the original statement of the postulate for
an infinitely heavy nucleus discussed in the previous section. Applying
we have
this postulate to equation (5-17),
mM
m+
Comparing
mM
this
m+
vr
with equation (5-1
fi
we
1),
replace the mass of the electron, m,
nh,
1, 2, 3,
see that the net effect has been to
by the
so-called reduced mass,
-f
(5-18)
m by the factor 1/(1 + m/M).
go through the rest of Bohr's derivation for the case
of finite nuclear mass. It is found that all the equations are identical with
those derived before except that the electron mass m is replaced by the
reduced electron mass (x. In particular, the formula for the wave numbers
of the spectral lines becomes
which
It is
is
than
slightly less
not
difficult to
lc
mM
^s
~m --j^
+ MAirc&
>(L-L)
-= R MZ*[9
9
?/'
RM -
),
\nf
(5-19)
R M is the Rydberg constant for a nucleus of mass M. As M/m * co,
apparent that R M > R^, the Rydberg constant for an infinitely heavy
where
it is
nucleus which appears in equation (5-15).
constant
RM
than
is less
R^ by
extreme case of hydrogen,
than
R x by
we
about
In general, the Rydberg
the factor 1/(1
M/m =
1836
(see
mjM). For the most
3) and R H is less
Chapter
part in 2000.
Rn
from equation (5-19), using the currently accepted
-1
values of the quantities m, M, e, c, and h, we find R H = 109681 cm
R
given
in
equation
this
with
the
experimental
value
of
(5-4),
Comparing
H
we see that the Bohr theory, corrected for finite nuclear mass, agrees with
If
evaluate
the spectroscopic data to within 3 parts in 100,000!
worth while to present another demonboth the spectroscopic data and the
Bohr theory. This involves a comparison of the spectrum of hydrogen,
H, and the spectrum of singly ionized helium, He+. As mentioned above,
the lines of these two spectra are observed to differ in wave number by a
Before closing this section,
it is
stration of the numerical accuracy of
BOHR'S THEORY OF ATOMIC STRUCTURE
124
factor
F which has
a value very close to
4.
[Ch. 5
According to equation (5-19),
the factor should be
mMse+
MHe+ _ 4 MHe + (m + MH
mAfH
MH (m + MHe+)
m + MH
m
F _ ^2 +
The mass of the helium nucleus is just the mass of the helium atom less two
electron masses.f That
is,
MHe+ = MHe
and the mass of the hydrogen nucleus
less one electron mass.f So
atom
is
- 2W
equal to the mass of the hydrogen
atom
MB MH
From
the study of chemistry
atom
we have an
accurate value of the ratio
Mm atom _
4.004
Mh atom
1.0081
The four equations above, plus the experimental value of the factor F,
allow an evaluation of
n /m. The result is
n jm = 1837, in very good
agreement with the value 1836 obtained in the experiments discussed in
Chapter 3.
Atomic Energy States
5.
The Bohr theory predicts that the total energy of an atomic electron is
For example, equation (5-14) gives the allowed energy values
quantized.
for the electron in a one-electron atom.
Although we have not attempted
to derive similar expressions for the electrons in a multi-electron atom,
clear that the total energy of each of the electrons will also be quantized
and, consequently, that the total energy content of the atom must be
it is
restricted to certain discrete values.
Thus the theory
predicts that the
atomic energy states are quantized. The correctness of this prediction was
confirmed in a simple experiment performed by Franck and Hertz in 1914,
less than one year after the announcement of Bohr's theory.
Apparatus of the type used by these investigators is indicated in figure
(5-7).
t
We
Electrons are emitted thermally at low energy from the heated
ignore the relativistic contribution to the atomic mass, which arises from the
have a negative total energy of the order of 10- 11 ergs, since this
10- u /10"
10" 32 gm is small compared to an electron mass.
/c 2
fact that the electrons
mass
M=
ATOMIC ENERGY STATES
Sec. 5]
125
cathode C.f They are accelerated to the anode A by a potential V applied
between the two electrodes. Some of the electrons pass through holes in
A and travel to plate P, providing their kinetic energy upon leaving A is
enough to overcome a small retarding potential Vr applied between P
and A
atoms
The
entire tube
current reaching
at a low pressure with a gas or vapor of the
The experiment involves measuring the electron
is filled
to be investigated.
P (indicated
by the current / flowing through the meter)
as a function of the accelerating voltage
-V
Heater
An
Figure 5-7.
V.
apparatus used to prove that atomic energy states are quantized.
The first experiment was performed with the tube containing Hg vapor.
The results are indicated in figure (5-8). At low accelerating voltage, the
When V
is observed to increase with increasing voltage V.
reaches 4.9 volts, the current abruptly drops. This was interpreted as
indicating that some interaction between the electrons and the Hg atoms
suddenly commences when the electrons attain a kinetic energy of 4.9
current /
10~ 12 ergs, as a result of falling through a potential difference
of 4.9 volts.+ Apparently a significant fraction of the electrons of this
energy excite the Hg atoms and in so doing entirely lose their kinetic
energy. If V is only slightly more than 4.9 volts, the excitation process
ev
7.85
must occur just
in front of the
anode A, and
after the process the electrons
point at which the thermal agitation
t By raising the temperature of the cathode to the
required to overcome the
kinetic energy of its electrons is comparable to the energy
attractive forces acting on the electrons at the surface of the cathode, some of the
electrons will be spontaneously emitted.
%
An
electron falling
kinetic energy
unimpeded through a potential difference of 1 volt attains a
19
joules =
1.60 X 10" 19 coulomb x 1 volt = 1.60 x 10~
E = eV =
10~ 12 ergs
ergs. In the present case the electron energy is 4.9 x 1.60 x
12
volt (ev)
electron
define
the
energy,
we
10~
unit
of
convenient
For
a
very
ergs.
7.85 x
1.60 x 10~ 12 ergs. This unit is used extensively in atomic and nuclear
such that 1 ev
physics. The kinetic energy of the electron discussed above can be specified as 4.9 ev or
12
as 7.85 x 10~ ergs.
1.60
x 10" 12
BOHR'S THEORY OF ATOMIC STRUCTURE
126
[Ch. 5
cannot gain enough kinetic energy in falling toward A to overcome the
retarding potential Vr and reach plate P. At somewhat larger V, the
electrons can gain enough kinetic energy after the excitation process to
overcome Vr and reach P. The sharpness of the break in the curve indicates that electrons of energy less than 4.9 ev are not able to transfer their
energy to an Hg atom. This interpretation is consistent with the existence
of discrete energy states for the Hg atom, as predicted by the Bohr theory.
10
V (volts)
The voltage dependence
Figure 5-8.
of the current measured
the Franck-Hertz
in
experiment.
Assuming the first excited state of Hg to be 4.9 ev higher in energy than
the ground state, an Hg atom would simply not be able to accept energy
from the bombarding electrons unless these electrons had at least 4.9 ev.
Now, if the separation between the ground state and the first excited state
actually 4.9 ev
7.85 X 10~ 12
spectrum of wave numberf
is
ch
7.85
3.00
10
10
ergs, there
should be a line in the
x 1(T 12
6.62
KT 27
3.95
10*
Hg
cm -1
corresponding to the transition from the
first excited state to the ground
Franck and Hertz found a line in the spectrum of very nearly this
wave number. Furthermore, they found that when the energy of the
bombarding electrons is less than 4.9 ev no spectral lines at all are emitted
from the Hg vapor in the tube, and when the energy is not more than a
few electron volts greater than this value only the single line k = 3.95
state.
104
cm-1
is
seen in the spectrum.
what would be expected
if
All these observations are precisely
the energy of the
Hg atom
is
quantized.
The
Franck-Hertz experiment provided a striking confirmation of the predictions of the Bohr theory. It also provided a method for the direct
measurement of the energy differences between the quantum states of an
t In evaluating this
we use
quantities in the cgs system
E=
7.85
x 10-12
ergs since
and want the answer
we have
in that system.
expressed
all
other
ATOMIC ENERGY STATES
Sec. 5]
127
the answers appear on the dial of a voltmeter
atom
/ versus
Some
V is
When
the curve of
extended to higher voltages, additional breaks are found.
are due to electrons exciting the
first
excited state of the
C to A
atoms on
but some are due to
excitation of the higher excited states and, from the position of these
breaks, the energy differences between the higher excited states and the
several separate occasions in their trip
ground
state
from
can be directly measured.
EHgjnna
2nd excited
)
state
1st excited state
.
>
>
a)
a*
r**
^J-
to
'
'
'
Ground
Figure 5-9.
state
considerably simplified energy level diagram for mercury.
Another experimental method of determining the separations between
atom is to measure its atomic spectrum and then
empirically construct a set of energy states which would lead to such a
spectrum. In practice this is often quite difficult to do since the set of
the energy states of an
lines constituting the spectrum, as well as the set
very complicated. However, in
it is
of energy
states, is often
common with all spectroscopic techniques,
a very accurate method. In
all
cases in which determinations of the
separations between the energy states of a certain
atom have been made
using both this technique and the Franck-Hertz technique, the results
have been found to be in excellent agreement.
In order to illustrate the preceding discussion, we show in figure (5-9)
a considerably simplified representation of the energy states of Hg in
terms of an energy level diagram. The separations between the ground
and the first and second excited states are known, from the FranckHertz experiment, to be 4.9 ev and 6.7 ev. These numbers can be confirmed,
and in fact determined with much higher accuracy, by measuring the wave
numbers of the two spectral lines corresponding to transitions of an
state
electron in the
Hg atom from
these
two
states to the
ground
state.
The
BOHR'S THEORY OF ATOMIC STRUCTURE
128
[Ch. 5
= 10.4 ev,
of the ground state relative to a state of zero total
not determined by the Franck-Hertz experiment. However, it
can be found by measuring the wave number of the line corresponding to a
transition of an atomic electron from a state of zero total energy to the
ground state. This is the series limit of the series terminating on the ground
state. The energy e can also be measured by measuring the energy which
energy
energy,
is
must be supplied to an Hg atom in order to send one of its electrons from
the ground state to a state of zero total energy. Since an electron of zero
total energy is no longer bound to the atom, s is the energy required to
ionize the atom and is therefore called the ionization energy.
Lying above the highest discrete state at E =
are the energy states
of the system consisting of an unbound electron plus an ionized Hg atom.
The total energy of an unbound electron (a free electron with E > 0) is not
quantized. Thus any energy E >
is possible for the electron, and the
energy states form a continuum. The electron can be excited from its
ground state to a continuum state if the Hg atom receives an energy greater
than 10.4 ev. Conversely, it is possible for an ionized Hg atom to capture
a free electron into one of the quantized energy states of the neutral atom.
In this process, radiation of wave number greater than the series limit
corresponding to that state will be emitted. The exact value of the wave
number depends on the initial energy E of the free electron. Since E
can have any value, the spectrum of Hg should have a continuum extending
beyond every series limit in the direction of increasing wave number. This
can actually be seen experimentally, although with some difficulty. These
comments concerning the continuum of energy states for E > 0, and its
consequences, have been
equally true for
6.
all
made
in reference to the
Hg
atom, but they are
atoms.
The Wilson-Sommerfeld Quantization Rules
The
success of the
Bohr
theory, as measured
by its agreement with
But it only accentuated the
mysterious nature of the postulates on which the theory was based. One
of the biggest of these mysteries was the question of the relation between
Bohr's quantization of the angular momentum of an electron moving in a
circular orbit and Planck's quantization of the total energy of an entity,
such as an electron, executing simple harmonic motion. In 1916 some light
was shed upon this by Wilson and Sommerfeld, who enunciated a set of
rules for the quantization of any physical system for which the coordinates
are periodic functions of time. These rules included both the Planck and
the Bohr quantization as special cases. They were also of considerable
experiment, was certainly very striking.
THE WILSON-SOMMERFELD QUANTIZATION RULES
Sec. 6]
use in broadening the range of applicability of the
rules
quantum
129
theory. These
can be stated as follows
For any physical system in which the coordinates are periodic functions
of time, there exists a quantum condition for each coordinate. These
quantum conditions are
j>pt dq
= nji
(5-20)
one of the coordinates, pa is the momentum associated with that
coordinate, n q is a quantum number which takes on integral values, and
means that the integration is taken over one period of the coordinate q.
where q
is
Figure 5-10.
The time dependence of the coordinates of an electron
The meaning of these
rules
can best be
in
illustrated in terms of
Bohr
some
orbit.
specific
examples.
Consider a particle of mass
m moving with constant angular velocity a>
Bohr
in a circular orbit of radius r (an atomic electron in a
orbit).
The
position of the particle can be specified by the polar coordinates r and
0.
The behavior of these two coordinates is shown in figure (5-10) as
functions of time.
r
They
are both periodic functions of time, if we consider
to be a limiting case of this behavior.
The momentum
associated
mr2 dd/dt.
with the angular coordinate 6 is the angular momentum L
The momentum associated with the radial coordinate r is the radial
momentum pT
dd/dt
= m dr/dt.
w, a constant.
In the present case r
Thus L
mrgco and
pr
=
0.
= 0, and
do not need to
dr/dt
We
apply equation (5-20) to the radial coordinate r in the limiting case in
which the coordinate is constant. The application of the equation to the
is easy to carry through for the present example.
angular coordinate
n; then
andp
L, a constant. Write n a
We have q d
Q
P, dq
= i Ld6 =
L$>dd
So the condition
Pa
dq
n ah
L\
"dd
2ttL
BOHR'S THEORY OF ATOMIC STRUCTURE
130
[Ch. 5
becomes
2ttL
nh
or
L=
= nh
nhjltr
which is identical with Bohr's quantization law.
Next let us apply the Wilson-Sommerfeld quantization rules to a particle
of mass m executing simple harmonic motion with frequency v. The
-*o
Figure 5-1
The time dependence
1.
of the coordinate of a simple harmonic oscillator.
position of the particle can be specified by the single linear coordinate x.
The behavior of this coordinate with time
and can be expressed as
=x
where x
is
sin
2-TTVt
sin
is
illustrated in figure (5-11)
m=
cot,
2ttv
The momentum of the particle
the amplitude of the oscillation.
is
dx
= m
= mx a> cos cot
dt
In this case
pa dq
<p
To
evaluate this integral
it is
<p
p dx
(p
mx
a)
cos
cot
dx
convenient to express cos
x^sin
cot
(l
cos
cot
in terms of x:
cot)
Then
cos^
cot
cos
cot
2
= V* -
xa
and
p dx
tnx
v*o
co
dx
into
PV*
Jo
The
last step
depends upon the fact that
'O
'ito
Jo
J -x
Jo
x?
dx
SOMMERFELD'S RELATIVISTIC THEORY
Sec. 7]
because the integrand
<J>
an even function of x. This gives
is
p dx
131
mco 4
sin
2
(p
p dx
mco 2{x% sin
p dx
mx\ am
-1
We would like to write this in terms
it is
is
easy to do
equal to
its
if
we
E = -m\
since
when
when x
dth=o
0.
x\ sin
a-o-J o
-1
0)
of the total energy
recall that the total
kinetic energy
m(x
0.
co
E of the particle;
energy of a harmonic oscillator
So
cos
ci>0?=
- mx\co %
Thus we have
p dx
2tt
= E
= ^1 mxlcoi2
,
and
Ejv
which
7.
is
nQh
nh,
E=
nhv
identical with Planck's quantization law.
Sommerfeld's Reiativistic Theory
One of
the important applications of the Wilson-Sommerfeld quanti-
zation rules
was to the case of a hydrogen atom
in
which
it
was assumed
This was done by
Sommerfeld in an attempt to explain the fine structure of the hydrogen
spectrum. The fine structure is a splitting of the spectral lines, into several
distinct components, which is found in all atomic spectra. It can be
that the electron could
move
in elliptical orbits.
observed only by using equipment of very high resolution since the sepanumber, between adjacent components of a single
4
spectral line is of the order of 10~ times the separation between adjacent
ration, in terms of wave
lines. According to the Bohr theory, this must mean that what we had
thought was a single energy state of the hydrogen atom actually consists
of several states which are very close together in energy.
Sommerfeld first evaluated the size and shape of the allowed elliptical
BOHR'S THEORY OF ATOMIC STRUCTURE
132
orbits, as well as the total
[Ch. S
energy of an electron moving in such an orbit,
using the formulas of classical mechanics. Describing the motion in terms
of the polar coordinates r and
0,
The first condition
momentum,
yields the
L=
he applied the two quantum conditions
LdB =
nh
pr dr
nr h
same
restriction
n e h,
ng
1, 2, 3,
on the
orbital angular
that it does for the circular orbit theory. The second condition (which was
not applicable in the limiting case of purely circular orbits) leads to the
following relation between
and
ajb, the ratio
of the semimajor axis to
the semiminor axis of the ellipse
L{a\b
By applying
1)
nr
nft,
0, 1, 2, 3,
the condition of mechanical stability analogous to equation
From
(5-8), a third equation is obtained.
these equations
Sommerfeld
evaluated a and b, which give the size and shape of the elliptical orbits,
and
also the total energy
E of an electron in such an orbit.
n A
The results
2
<.
fiZe
= a^
are
(5-21)
2t?h%
where
/j,
the reduced mass of the electron, and where the
is
number n
n
Since n e
quantum
defined by
is
1, 2, 3,
and n r
n
=
=
nB
nr
0, 1, 2, 3,
1, 2, 3, 4,
n can take on the values
For a given value of n, n e can assume only the values
n
1, 2, 3,
SOMMERFELD'S RELATIVISTIC THEORY
Sec. 7]
The
integer n
is
called the principal
133
quantum number, and ng
is
called the
azimuthal quantum number.
(5-21) shows that the shape of the orbit (the
semimajor to the semiminor axes) is determined by the ratio of
The second of equations
ratio of the
n, the orbits are circles of radius a. Note that the
n e to n. For n e
equation giving a in terms of n is identical with equation (5-12), the equation giving the radius of the circular Bohr orbits.f Figure (5-12) shows,
,n
ne =
{2Zis- n
=l
re
n=2
Figure 512.
re
Some
elliptical
Bohr-Sommerfeld
to scale, the possible orbits corresponding to the
first
orbits.
three values of the
quantum number. Corresponding to each value of the principal
quantum number n there are n different allowed orbits. One of these, the
circular orbit, is just the orbit described by the original Bohr theory. The
others are elliptical. But, despite the very different paths followed by an
electron moving in the different possible orbits for a given n, the third of
principal
equations (5-21)
us that the total energy of the electron
tells
is
the same.
of the electron depends only on n. The several orbits
characterized by a common value of n are said to be degenerate.
This degeneracy in the total energy of an electron, following the orbits
of very different shape but common n, is the result of a very delicate balance
between potential and kinetic energy which is characteristic of treating an
inverse square law of force by the methods of classical mechanics.
The
total energy
Sommerfeld removed the degeneracy by treating the problem with the
theory of
In the discussion following equation
-2
an electron in a hydrogen atom, vjc ca 10 or
mechanics.
relativistic
we showed
Thus we would expect the
that, for
(5-13)
less.
due to the
relativistic variation
relativistic corrections to the total energy,
of the electron mass, which will be of the
order of (v/c) 2 to be only of the order of IQr*. However, this is just the
order of magnitude of the splitting in the energy states of hydrogen that
would be needed to explain the fine structure of the hydrogen spectrum.
,
The
actual size of the correction depends
calculation which
t
is
much
on the average
velocity of the
After a
too tedious to reproduce here, Sommerfeld
electron which, in turn, depends
on
the ellipticity of the orbit.
Remember that equation (5-12) will have m replaced by p if proper account is taken
of the
finite
nuclear mass.
BOHR'S THEORY OF ATOMIC STRUCTURE
134
showed that the total energy of an electron in an
quantum numbers n and ne is equal to
-&.
+ ^(i
Irftf
The quantity a = e%\hc is a pure number
The numerical value of this constant is
a
*\hc
\n B
orbit characterized
(5 _22)
1)1
4n/J
(5-23)
1/137
^n =
1
1
-n =
-n =
"*
Figure 5-13.
The
splitting
The
is
<
=4
n=
n=
r-l
n=
fine structure splitting of
greatly exaggerated.
The
by the
called the fine structure constant.
X 10"s ~
7.297
[Ch. 5
some energy
3, rig
3,
ne = 2
3, tig
2,
ne = 2
n = 1
1,
n =
2,
ft
fl
hydrogen atom.
between the levels are
levels of the
possible transitions
indicated by solid arrows.
In figure (5-13)
atom
in terms of
we
several levels with a
the sake of clarity.
states
represent the
an energy
level
common
Arrows
which produce the
first
few energy
diagram.
states
of the hydrogen
The separation between
the
value of n has been greatly exaggerated for
indicate transitions between the various energy
lines
of the atomic spectrum. Lines correspond-
ing to the transitions represented by the solid arrows are observed in the
hydrogen spectrum. The wave numbers of these lines are in very good
agreement with the predictions derived from equation (5-22). However,
the lines corresponding to the transitions represented by dashed arrows
are not found in the spectrum. The transitions concerned do not take
place. Inspection of the figure will demonstrate that transitions only
occur if
n 6i
This
is
called a selection rule.
-,,= 1
(5-24)
THE CORRESPONDENCE PRINCIPLE
Sec. 8]
8.
The Correspondence
The reason
Principle
for the existence of selection rules
we have described so far. However, with
postulate known as the correspondence principle, a
theory
of these rules could sometimes be found.
Bohr
in 1923, consists of
135
is
not contained in the
the aid of an auxiliary
theoretical justification
This principle, enunciated by
two parts:
1. The predictions of the quantum theory for the behavior of any physical
system must correspond to the predictions of classical physics in the limit
in which the quantum numbers specifying the state of the system become very
large.
A selection rule holds true over the entire range of the quantum number
2.
Thus any selection rules which are necessary
concerned.
to obtain the
required correspondence in the classical limit (large n) also apply in the
quantum
limit (small n).
Concerning the first part, it is obvious that the quantum theory must
correspond to the classical theory in the limit in which the system behaves
classically. The only question is: Where is the classical limit? Bohr's
assumption is that the classical limit is always to be found in the limit of
large quantum numbers. In making this assumption he was guided by
certain evidence available at the time.
For instance, the
classical Rayleigh-
Jeans theory of the black body spectrum agrees with experiment in the
limit of small v. Since Planck's quantum theory agrees with experiment
between the quantum and classical
v. But it is easy to see
from equation (2-30) that as v becomes small the average value n, of the
quantum number specifying the energy state of black body electromagnetic
waves of frequency v, will become large.f The second part of the corre-
everywhere,
theories
is
we
see that correspondence
found, in this case, in the limit of small
spondence principle was purely an assumption, but certainly a reasonable
one.
correspondence principle by applying it to a simple
such
harmonic oscillator,
as a pendulum oscillating at frequency v. One
of the predictions of the quantum theory for this system is that the allowed
nhv. In the discussion of
energy states are given by the equation E
Let us
illustrate the
section 10, Chapter 2,
we have
seen that, in the limit of large
n, this
not in disagreement with what we actually know about the
energy states of a classical pendulum. In this sense, the quantum and
classical theories do correspond for n * co in so far as the energy states of
a pendulum are concerned. Next assume that the pendulum bob carries
prediction
f e
nhv
is
nhv, so e
kT, which
=
is
nhv. Equation (2-30) shows that e ->
a constant. Thus n -* co as v -* 0.
kTas
->- 0,
so in this limit
BOHR'S THEORY OF ATOMIC STRUCTURE
136
[Ch. 5
we can compare the predictions of the two
concerning the emission and absorption of electromagnetic
radiation by such a system. Classically the system would emit radiation
an
electric charge, so that
theories
due to the accelerated motion of the charge, and the frequency of the
v. According to the quantum theory,
radiation is emitted as a result of the system making a transition from
quantum state n { to quantum state nf The energy emitted in such a transiemitted radiation would be exactly
tion
is
E Ef =
equal to
(n t
a quantum of frequency (Et
nf)hv. This energy
Ef)/h = (n nt)v.
4
obtain correspondence between the classical and
the frequency of the emitted radiation,
rule n t
nf
be valid in the
is
we must
carried
away by
Thus, in order to
quantum
predictions of
require that the selection
classical limit.
similar
argument
concerning the absorption of radiation by the charged pendulum shows
that in the classical limit there is also the possibility of a transition in which
n { nf
1
The validity of these selection rules in the quantum limit
can be tested by investigating the spectrum of radiation emitted by a
The
vibrating diatomic molecule.
vibrational energy states for such a
system are just those of a simple harmonic
leads to the equilibrium separation of the
oscillator, since the force which
two atoms has the same form as
a harmonic restoring force. From the vibrational spectrum it can be
determined that the selection rule n t n f 1 actually is in operation
in the limit of small quantum numbers.
number of selection rules were discovered empirically in the analysis
of atomic and molecular spectra. It was often possible to understand these
selection rules in terms of a correspondence principle argument, but
some-
times ambiguities arose.
9.
A Critique of the
In Chapters
2, 3,
ment of what
is
and
now
Old Quantum Theory
5,
we have described the high points in the develop-
referred to as the old
quantum
theory.
In
even
many
more so than may be
apparent to the reader since we have not mentioned a number of successful
applications of the theory to phenomena, such as the heat capacity of solids
respects this theory
was very
successful
low temperatures, which were inexplicable in terms of the classical
However, the old quantum theory was certainly not free of
criticism. To complete our discussion of this theory we must indicate some
at
theories.
of
its
1.
undesirable aspects.
The theory only
there are
many
tells
us
how
to treat systems
which are
periodic, but
systems of physical interest which are not periodic.
Ch.
5]
BIBLIOGRAPHY
137
Although it does tell us how to calculate the energies of the allowed
of a system, and the frequency of the quanta emitted or absorbed
when the system makes a transition between allowed states, the theory does
not tell us how to calculate the rate at which such transitions take place.
3. The theory is only really applicable to a one-electron atom. The
alkali elements (Li, Na, K, Rb, Cs) can be treated approximately, but only
because they are in many respects similar to a one-electron atom. The
theory fails badly when applied to the neutral He atom, which contains
two electrons.
4. Finally we might mention the subjective criticism that the entire
theory seems somehow to lack coherence to be intellectually unsatisfying.
2.
states
That some of these objections are really of a very fundamental nature
was realized by everyone concerned, and much effort was expended in
attempts to develop a quantum theory which would be free of these and
other objections. The effort was well rewarded. In 1925, following an
idea proposed in the preceding year by de Broglie, Schroedinger developed
his theory of quantum mechanics. This theory is now almost universally
considered completely acceptable.
As we shall see,
the Schroedinger theory
is
in
many respects very different
from the old quantum theory. For instance, the picture of atomic structure
provided by the theory of quantum mechanics is the antithesis of the
picture, used in the old quantum theory, of electrons moving in welldefined orbits. Nevertheless, the old quantum theory is still often employed
as a first approximation to the more accurate description of quantum
phenomena provided by quantum mechanics. The reasons are that the
old quantum theory is often capable of giving numerically correct results
with mathematical procedures which are considerably less complicated
than those used in quantum mechanics, and that the old quantum theory
is
often helpful in visualizing processes which are difficult to visualize in
terms of the somewhat abstract language of the theory of quantum
mechanics. Therefore the reader would be well advised to become thor-
oughly familiar with the content of the old quantum theory, particularly
the
Bohr
theory.
BIBLIOGRAPHY
Born, M., Atomic Physics, Blackie and Son, London, 1957.
Ruark, A. E., and H. C. Urey, Atoms, Molecules, and Quanta, McGraw-Hill
Book Co., New York, 1930.
Richtmyer, F. K., E. H. Kennard, and T. Lauritsen, Introduction to Modern
Physics, McGraw-Hill Book Co., New York, 1955.
BOHR'S THEORY OF ATOMIC STRUCTURE
138
[Ch. 5
EXERCISES
1.
Carry through the details of the Bohr calculation for a nucleus of
mass and obtain equation
finite
(5-19).
2. Consider a particle of mass m moving back and forth along the x axis
between two points x = and x = a from which it rebounds elastically. Apply
the Wilson-Sommerfeld quantization rules to find the predicted possible values
of the total energy of the particle.
3.
Consider a body rotating about an
quantization rules, and
show
axis.
Apply the Wilson-Sommerfeld
that the possible values of
its total
energy are
predicted to be
E=
where / is
4.
its
moment
AW/2/,
0, 1, 2, 3,
of inertia about the axis of rotation.
Apply the Wilson-Sommerfeld quantization rules to the results of exercise
4, and obtain equations (5-21).
3,
Chapter
5.
Prove the statement, made in section
classical expression
of
(v/c)
E = p2/2m + V is
7, that for
smaller than
small v/c the error in the
E by
a factor of the order
2
.
6. Derive a relation connecting the frequency of the electromagnetic radiation,
emitted in a transition between two states of a Bohr atom, and the orbital fre-
quencies of the electron in these states. Study this relation in the limit of large
quantum numbers, and comment on
classical physics.
its
correspondence with the predictions of
CHAPTER
Particles
and
I.
De
Waves
Broglie's Postulate
In his doctoral dissertation (1924) de Broglie put forth a simple but
extremely important idea which initiated the development of the theory of
quantum mechanics. De
following:
Broglie's line of thought ran something like the
In classical physics electromagnetic radiation had been con-
wave propagation phenomenon. However the
and Compton has demonstrated that under certain
sidered to be purely a
work of
Einstein
circumstances
displays properties characteristic of particles (quanta).
it
it also be true that physical entities which we
normally think of as particles (electrons, alpha particles, billiard balls,
etc.) will under certain circumstances display properties characteristic of
waves ?
The particle-like aspects of electromagnetic radiation become apparent
when the interaction of the radiation with matter is investigated, while the
This being the case, could
wave-like aspects
propagates
is
become apparent when the manner in which the radiation
investigated.
It
is
immaterial whether the situation
is
described by saying that electromagnetic radiation actually consists of
waves which upon interacting with matter are able to manifest a particleby saying that it actually consists of particles whose
motion is governed by the wave propagation properties of certain
associated waves. In fact, it is somewhat naive even to imply that there
is a choice to be made.
However, by temporarily adopting the latter
statement, de Broglie was guided by analogy to look for the proposed
wave-like aspects of particles in terms of some wave propagation
like behavior, or
139
PARTICLES
140
AND WAVES
[Ch. 6
Thus he was led to investigate the idea that the
governed by the wave propagation properties of
certain pilot waves (to use his terminology) which are associated with the
character in their motion.
motion of a
particle is
particle.
At
the time of de Broglie's proposal, wave-like behavior in the motion
of a particle had never been observed although the motion of particles had
been extensively investigated. But
this
does not prove anything because
Source
Photographic
Slit
plate
Figure 6-1.
An
apparatus used to demonstrate the wave-like nature of electromagnetic
radiation.
is so small, compared to
had been used to investigate
particle motion, that the wave-like properties would not have been observable. The meaning of this statement can be illustrated by considering a
single-slit diffraction experiment typical of those which can be used to
demonstrate the wave-like nature of electromagnetic radiation. The
it
might be that the wavelength of the pilot waves
the dimensions of the physical systems which
experiment
is
indicated in figure (6-1). It
an experiment depend
of the radiation,
X,
in a drastic
to the width of the
However, for 1
observed.
its
Under
<d
known
slit,
that the results of such
the ratio of the wavelength
d.
For A
d,
a diffraction
phenomena is recorded on the photographic
pattern characteristic of wave
plate.
is
manner on
a quite sharply defined image of the
these circumstances the radiation betrays
wave-like character, and
little
slit is
sign of
propagation can be very adequately
its
described in terms of the laws of ray optics, a particle-like theory.
It is clear that
de Broglie's
first
task was to specify
determining the wavelength of the pilot waves.
He
some procedure
for
did this by making the
reasonable assumption that the wavelength of pilot waves, and also their
frequency, could be evaluated from the same equations which are used to
evaluate these parameters for electromagnetic radiation. Consider a
beam
of electromagnetic radiation containing quanta of total relativistic energy E.
According to Einstein's equation
(3-7), the frequency v of the radiation is
v
Ejh
(6-1)
SOME PROPERTIES OF THE PILOT WAVES
Sec. 2]
141
The wavelength A of the radiation can be evaluated from the usual relation
between
A, v,
and w, the velocity of propagation of the wave.
= w/v
For electromagnetic radiation w = c, so
X = cjv = hcjE
According to equation (3-10), E\c = p, the momentum
This
Thus A can
also be evaluated
(6-3)
of the quantum.
h\p
(6-4)
Broglie postulated that the wavelength A and the frequency v of the
a particle of momentum
by the equations
p and
pilot waves associated with
energy
E are given
l
v
and
(6-2)
from the equation
A
De
is
that the motion
=
=
of the particle
total relativistic
hlp
E\h
is
governed by the wave propagation
properties of the pilot waves.
His reason for choosing equations analogous to (6-4) and (6-1), instead
of to (6-3) and (6-1), will become apparent shortly.
2.
Some
Properties of the Pilot
Waves
Let us calculate the propagation velocity w of the pilot waves associated
with a particle, using equation (6-2) and evaluating A and v from equations
(6-5).
This gives
vk
hp
(6-6)
Using equation (1-25) to evaluate the total relativistic energy of the particle,
we have
2 2
VcV+^mpC
)
cVp 2
(m c?
p
or
w
Note
that
w is greater than
c.
the velocity v of the particle
cVl
+ (m
c//>)
This seems, at
must
"7
C6
first,
quite disturbing because
necessarily be less than
appear that the particle could not keep up with
its
own
c,
and
it
would
pilot waves.
PARTICLES
142
Actually there
is
no
difficulty, as
that a free particle (experiencing
are
moving
coordinate
AND WAVES
[Ch. 6
the following argument shows. Imagine
no
and
forces)
its
associated pilot waves
way that they can be described by the single spatial
Assume that we have distributed at a large number of
in such a
x.
points along the x axis a set of hypothetical instruments which are capable
of measuring the intensity of the pilot waves. At a certain time t we
record the readings of these instruments. The results of the experiment
,
ii
Figure 6-2.
group of
pilot
waves.
could be presented as a plot of the instantaneous value of the pilot waves,
which we designate by the symbol *(x, i)t=t as a function of a;. Although
,
we know
about the pilot waves at present, it is evident that the plot
must look qualitatively like the one shown in figure (6-2). The amplitude
of the pilot waves must be modulated in such a way that their value is
non-zero only over a finite region of space in the vicinity of the particle.
This is necessary because the pilot waves must somehow be associated
spatially with the particle whose motion they control. The pilot waves
form a group of waves and, as a function of time, the group must move
along the x axis with the same velocity as the particle. The reader may
recall,
little
from
his study of
wave motion
a moving group of waves
g of
it is
the group and the velocity
waves.
Furthermore,
is
in elementary physics, that for such
necessary to distinguish between the velocity
of the individual oscillations of the
in general less than w.
This
is
encouraging,
but of course we must prove that g is equal to the velocity of the particle.
To do this, we develop a relation between g and the quantities v and
X comparable to the relation (6-2) between w and these two quantities.
We
start by considering the simplest type of wave motion, a sinusoidal
wave of frequency v and wavelength X, which is of constant amplitude
from -co to +oo, but which is moving with uniform velocity in the
direction of increasing x. Such a wave can be represented by the function
T(x,
i)
sin 2t7
(M
or
Y(x,
t)
sin
2n(kx
vi),
where k
1/A
(6-8)
Sec. 2]
SOME PROPERTIES OF THE PILOT WAVES
That
does represent the wave can be seen from the following con-
this
siderations
1.
Holding x fixed, we see that the function
with frequency
2.
on
x,
3.
it
143
Holding
oscillates in
time sinusoidally
v.
fixed,
we
see that the function has a sinusoidal
dependence
with wavelength A or wave number k.
The
zeros of the function, which correspond to the nodes of the
represents, are
wave
found at positions x n for which
2n(kx n
vi)
vn,
0,
1, 2,
or
xn
Thus these nodes, and
2k
in fact all points
on the wave, are moving
in the
direction of increasing x with velocity
w
which
is
dxjdt
equal to
w =
vjk
Note that this is identical with equation (6-2) since k = 1/A.
Next we discuss the case in which the amplitude of the waves is modulated to form a group. We can obtain mathematically one group of
waves moving in the direction of increasing x [something similar to the
group of pilot waves pictured in figure (6-2)] by adding together an
infinitely large number of waves of the form (6-8), each with infinitesimally
differing frequencies v and wave numbers k. However, the mathematical
techniques become a little involved and, for the purposes of the present
argument, it will suffice to consider what happens when we add together
only two such waves. Thus we take
W(x,
t)
= Y (x, t) + Y2 (z, t)
(6-9)
where
i)
sin 2ir\kx
2 (x, i)
sin 2n[{k
Yj(x,
vi\
(y
B)
sin
\(A
{2k
dk)
dk)x
dv)i\
Now
sin
Applying
T(x,
A +
sin
B=
this to the case at
2 cos
2n\
2 cos \(A
E)
hand, we have
1 sin 2n
{2v
dv)
/]
PARTICLES
144
Since dv
< 2v and dk < 2k, this
T(a;,
()
= 2 cos
2tt\
AND WAVES
[Ch. 6
(6-10)
is
sin 2ir{kx
vi)
A plot otW(x, t) as a function of x for a fixed value of t =
/ is shown in
The second term of Y(a;, /) is a wave of the same form as
equation (6-8), but this wave is modulated by the first term so that the
figure (6-3).
-ir*"
two
waves of
&
The sum
Figure 6-3.
of
sinusoidal
slightly different frequencies
and wave
numbers.
oscillations of
Y(x,
t)
fall
Two waves of
within an envelope of periodically varying
frequency and wave number
and reinforce in such a way as to produce an infinite
succession of groups. These groups, and the individual waves which they
contain, are both moving in the direction of increasing x. The velocity
w of the individual waves can be evaluated by considering the second
term of Y(a;, t), and the velocity g of the groups can be evaluated from the
amplitude.
slightly different
alternately interfere
first
Proceeding as in 3 above,
term.
w
8
_
~
we
find
vjk
dv/2
dkj2
_
~
(6-11)
dv
Tk
It can be shown that for an infinitely large number of waves combining to
form one moving group, the dependence of the wave velocity w, and
the group velocity g, on v, k, and dvjdk is exactly the same as for the
we have considered. Equations (6-1 1) have a general validity.
we are in a position to calculate the group velocity g of the
simple case
Finally
group of pilot waves associated with the moving particle. From equations
(6-5), we have
v = Ejh and k = \\'k=p\h
So
dv
dEjh
and
dk
dp/h
SOME PROPERTIES OF THE PILOT WAVES
Sec. 2]
Thus the group
velocity
145
is
dvjdk
dE\dp
ing to equation (1-25),
Kc
E 2 == c2p2 +
2EdE =
dE
2p dp
?
E
dp
>re
2 2
(6-12)
According to equations (1-24) and (1-17),
E = mc 2
where
m m /Vl
of the
particle.
mv
v 2 /c 2 , the total relativistic mass,
Thus we have the
g
The
and p
and
is
the velocity
satisfying result that
=
mc
(6-13)
group of pilot waves is just equal to the velocity of the
whose motion they govern, and de Broglie's postulate is internally
velocity of the
particle
consistent.
It is
easy to see
why de
Broglie did not take a pair of equations
k=\jX =
v
E\hc
E\h
analogous to equations (6-3) and (6-1) to specify the frequency and wave
the pilot waves associated with a particle. Applying (6-11),
number of
these equations lead to a
wave
velocity
w
and to a group
vjk
velocity
dv
dk
dEjh
'
dEjhc
=c
quantum which moves with a velocity which is always
but certainly not correct for a particle which moves with a
velocity v which is necessarily different from c.
This
is
correct for a
equal to
c,
From
the
AND WAVES
PARTICLES
146
equations (6-6) and (6-12)
wave
velocity
w and
we
find the following relation
[Ch. 6
between
the group velocity g:
w
According to equation (6-13),
this
c*/g
can be written in terms of the velocity
of the particle v:
c % \v
(6-14)
Since w is larger than g, the individual waves are constantly moving
through the group from the rear to the front. The same situation occurs
in a group of water waves.
Experimental Confirmation of de Broglie's Postulate
3.
As mentioned
in section 1, comparison with the propagation characterof electromagnetic radiation predicts that wave-like behavior in the
motion of a particle will only be apparent when the de Broglie wavelength
istics
hjp is of the order of, or larger than, the characteristic dimensions
of the system used to investigate the motion. Since Planck's constant h
has the very small value 6.62 x 10~ 27 erg-sec, it is immediately apparent
that, unless the
momentum p
also very small, X will be so small as to
is
preclude any hope of observing the postulated
p and X for some specific cases.
effects.
Let us evaluate
Consider a very small dust particle of radius r = 10~ 4 cm, density
-1
10, moving with the low velocity v = 1 cm-sec
Since v
c, we
p
may evaluate p from the classical formula
p = mv =
The de
is
~4
10 -11 gm-cm-sec -1
Broglie wavelength of this particle
This
7rr 3 pv
6.6
x 10~ 27
10~ u
<
is
1.6
__ 16
18
cm
X 10
extremely small compared to the dimensions of any physical system,
we have taken the momentum of the particle to be very much
although
less
than that of the particles typically discussed in classical mechanics.
Thus we would not expect to be able to either prove or disprove the validity
of de Broglie's postulate by investigating the motion of macroscopic
particles.
Next consider an electron with the
10 ev
1.6
10 _11 ergs.
This
is,
relatively
low
kinetic energy
T=
for example, approximately equal to
the kinetic energy of an electron in a hydrogen atom.
The
velocity of the
EXPERIMENTAL CONFIRMATION OF THE POSTULATE
Sec. 3]
electron
is
small compared to
c,
so
we may
evaluate
its
147
momentum from
the classical equation
p=
VlmT=V2 x 9.1 x 10- X 1.6
~ 1.7 x 10~ 19 gm-cm-sec-1
28
KH
/?
The corresponding de
Broglie wavelength
is
= 6.6xl0-^
39xl0_8cm
10- 19
1.7
Although this is still a very small wavelength, it is of the order of the
size of an atom and, consequently, of the interatomic spacing of the
o+
Surface of crystal
-J U
Figure 6-4.
An
by the electrons
atoms in a
apparatus used to verify de Eroglie's postulate.
is
The region traversed
evacuated.
crystal.
This suggests the possibility of looking for diffraction
or transmission of a beam of electrons by a crystal.
effects in the reflection
Such
effects
were
first
found by Davisson and Germer.
In the Davisson-Germer experiment (1927), illustrated schematically
electrons
in figure (6-4), a parallel beam of monoenergetic low energy
a
was produced by accelerating electrons, thermally emitted from cathode
C, through a voltage drop V.
surface of a
Ni
crystal.
Some
the surface of the crystal.
This
beam was
incident
normal to the
back from
of the electrons were scattered
The number
scattered at angle
(defined in
behind
the figure) was measured with a detector consisting of a plate P
allowed
V,
than
less
slightly
V
voltage
retarding
a diaphragm D.
r,
PARTICLES
148
only electrons scattered from the crystal with
AND WAVES
little
P and produce a current indicated by meter M. Low
[Ch. 6
energy loss to reach
energy electrons lose
energy rapidly in traversing a solid, so the detected electrons must have
essentially from the surface of the crystal. The periodic
distribution of the atoms of that surface is indicated in the figure. For a
been scattered
Ni
d of the periodic structure is 2.15 x 10~8
comparable to the de Broglie wavelength of the incident
Since the essential feature of a diffraction grating is its periodi-
crystal the repetition length
cm,f which
electrons.
city,
the pattern of electrons scattered from the crystal surface should
Figure 6-5.
is
The angular
distribution of electrons scattered from a nickel
show
crystal.
54.0 volts.
diffraction effects if the
de Broglie postulate
is
correct.
Plotted in figure
form of Davisson and Germer's data for the number of
scattered electrons N(6) as a function of 0, measured for an accelerating
voltage V of 54.0 volts. The angular distribution JV(0) becomes very large
for small 0; that is, most of the scattered electrons are specularly reflected
(6-5)
is
the
(angle of reflection
angle of incidence). This
is not particularly informof both particle motion and
wave motion. However, at an angle of 50 a peak is observed in N(6).
The existence of this peak proves the qualitative validity of the de Broglie
postulate because such a peak can only be explained as a constructive
interference of waves scattered by the periodically placed atoms
of the
crystal surface. This is not an interference between waves associated
with one electron and waves associated with another. Instead, it is an
ative because such behavior is characteristic
interference between waves, associated with a single electron, that have
been scattered from various parts of the crystal. This conclusion can be
demonstrated experimentally by making measurements with an electron
t This
is
known from measurements of the diffraction pattern produced in
known wavelength by a Ni crystal. Such measurements
scattering of X-rays of
be described in Chapter
14.
the
will
EXPERIMENTAL CONFIRMATION OF THE POSTULATE
Sec. 3]
beam of such low
intensity that the electrons
149
go through the apparatus
one at a time, and by showing that the pattern of the scattered electrons
remains the same.
The data of Davisson and Germer can also be used to show that de
Broglie's postulate is quantitatively correct. If we evaluate the de Broglie
wavelength of a 54 ev electron from the equation X = hjp, as above, we
find
1.67
10- 8
cm =
1.67
The wavelength of the waves which are interfering constructively can
be evaluated from the well-known grating equation,^
nX
Assuming
that the
d sin
6,
peak corresponds to a
1, 2, 3,
first
also
(6-15)
order diffraction (n
1),
we have
The two
2.15
10-8
cm x
sin 50
1.65
values of X agree to within the accuracy of the experiment.
parable agreement was found in the data obtained at a
accelerating voltages.
(n
2)
was also
At somewhat higher
Com-
number of different
voltages a second order peak
seen.
Diffraction effects in transmission through crystals were observed
by
Thomson. His experiment (1928) involved sending a collimated
beam of electrons through a polycrystalline foil. The foil thickness was
only about 10~ s cm, but even so it was necessary to use an electron
energy of about 10* ev so that the energy loss in traversing the foil would
G.
P.
not be too large. In passing through the randomly oriented crystals of the
foil, the electrons will find certain ones for which the set of scattering angles
wu^ satisfy the equation analogous to (6-15) which applies
0i 6 2 %>
,
to this case.
The
from these crystals
The angles are all small since Xjd is small
however, by placing a photographic plate some
electrons will be strongly scattered
at the angles indicated.
(A~5
x 10~2 A);
distance behind the scattering foil a series of concentric rings corresponding
shown
The photograph on the left shows a ring pattern resulting
from the scattering of electrons by a thin layer of Au crystals. The photograph on the right shows the completely similar pattern observed in the
scattering of electromagnetic radiation (X-rays) by a thin layer of ZrO a
to the allowed scattering angles can be detected. Typical results are
in figure (6-6).
crystals. This comparison gives another convincing demonstration of
wave-like behavior in the motion of electrons. Furthermore the values of
the scattering angles measured from the electron ring pattern agree with
t See any textbook
on elementary
physics.
PARTICLES
150
AND WAVES
[Ch. 6
those of the scattering angles calculated from the wavelength predicted
by the de Broglie
postulate.
Several years after the initial experiments with electrons, Estermann,
Frisch, and Stern demonstrated the existence of diffraction effects in the
scattering of a beam of He atoms from the surface of a LiF crystal.
The beam was obtained from an enclosure heated to a temperature of
400K. Emerging from a channel communicating with the interior of the
oxide crystals.
The ring pattern produced in the scattering of electrons by gold
The ring pattern produced in the scattering of X-rays by zirconium
From U. Fano and L. Fano, Basic Physics of Atoms and Molecules, John
Wiley and Sons,
New
Figure 6-6.
crystals.
Left:
Right:
York, 1959.
beam of atoms of average energy E = \kT, where k is
Boltzmann's constant. The de Broglie wavelength of an atom of average
energy was
enclosure was a
J2ME
j3MkT
6.6
V3 x
6.7
x 1(T 27
X 1(T 24 x
1.4
~ 0.62 A
X 1(T 16 x 4 x 102
The scattered atoms were detected by a very sensitive pressure gauge.
The pattern of scattered He atoms was found to be similar to that observed
by Davisson and Germer for the scattering of electrons. The angular
location of the diffraction peak was in good agreement with what would be
expected for the wavelength predicted by de Broglie's postulate.
We have described three of the original experiments verifying that there
are
wave propagation
effects in the
motion of particles, as well as verifying
INTERPRETATION OF THE BOHR QUANTIZATION RULE
Sec. 4]
Since then very
the wavelength predicted by de Broglie's postulate.
more examples of
151
many
these effects have been observed experimentally,
and
the validity of de Broglie's postulate has been undeniably confirmed.
4.
Interpretation of the Bohr Quantization Rule
We
have seen above that the de Broglie wavelength of an electron of
comparable to that of an electron in an H atom is about
10~ 8 cm. According to the Bohr theory, the radius of the ground state
kinetic energy
4x
atom is about 0.5 x 10-8 cm. Since X ;> r, we would
wave propagation properties of the pilot waves to play an
orbit of the
expect the
important part in determining the motion of the electron in the atom.
However, there is a significant difference between the type of particle
motion we have previously discussed in this chapter and the motion of an
electron in
an atom.
trajectories just once
In the examples above the particles travel their
they are unbound particles, and their associated pilot
waves are traveling waves (moving nodes). An electron in an atom is
bound to the atomic nucleus. It travels repeatedly through the same orbit
Consequently its associated pilot waves would be expected to be standing
waves (fixed nodes).
Standing waves are one of the most striking phenomena of wave propagation. In 1924 de Broglie showed how a simple consideration of their
properties could lead to an interpretation of the mysterious Bohr angular
momentum quantization rule. According to equation (5-11), this rule
can be written
mvr
where
radius
r.
is
the linear
If
we
= pr =
n=
nh\2-n,
momentum
of an electron in an allowed orbit of
substitute into this equation the expression
p=
for
1, 2, 3,
hjX
in terms of the corresponding de Broglie wavelength, the equation
becomes
hr/X
This
nhjln
is
Itrr
nX,
1,
2, 3,
(6-16)
orbits are those in which the circumference of the orbit
can contain exactly an integral number of de Broglie wavelengths. If
we think of an electron traveling repeatedly around the same orbit, it
becomes apparent that equation (6-16) is just the necessary condition that
Thus the allowed
PARTICLES
152
AND WAVES
[Ch. 6
the pilot waves associated with the particle in a particular traversal of the
orbit will
combine coherently with the
pilot
waves associated with the
up a standing wave. However, if this equation is
violated, then in a large number of traversals the pilot waves will interfere
with each other in such a way that their average intensity will vanish.
Since the intensity of the pilot waves must be some sort of a measure of
where the particle is located, we interpret this as meaning that an electron
previous traversals to set
could not be found in such an orbit. Figure (6-7)
Figure 6-7.
the
first
is
an attempt to illustrate
schematic representation of the patterns of standing waves set up
in
three allowed Bohr orbits.
the
Tj =(
the
first
some
waves set up in
the magnitude
and sign of the amplitudes of these oscillatory patterns might be different,
but the locations of the nodes would be the same since the nodes of any
standing wave are fixed.
It is not difficult to show that the requirement that the pilot waves
associated with a particle undergoing any sort of periodic motion be a set
,
at
particular value of
three allowed
of standing waves
is
Bohr
orbits.
for the standing
At some other value of t
equivalent to the requirement that the motion of the
particle satisfy the Wilson-Sommerfeld quantization rules (5-20).
see in the following chapters that the properties of standing
We
shall
waves also
play an important role in the Schroedinger theory of quantum mechanics.
For
instance, the time independent features of the standing
ated with an electron in one of
possible to understand
why
its
waves associ-
allowed states in an atom will
the motion described
make
it
by the standing wave
does not cause the electron to emit electromagnetic radiation.
THE UNCERTAINTY PRINCIPLE
Sec. 5]
5.
The Uncertainty
An
153
Principle
inescapable result of de Broglie's standing
electron in a
Bohr
orbit
is
wave description of an
and the orienta-
that the position of the electron,
its linear momentum, cannot be exactly
a given instant of time since the azimuthal symmetry of the
tion of the vector indicating
specified at
pattern of standing waves, pictured in figure (6-7) at the instant
indicates that at that instant the electron in a particular orbit could be
located essentially anywhere in the orbit.
specification of position
and
linear
This
is
in contrast to the exact
momentum which is
possible in Bohr's
motion of an atomic electron, just as it is
possible in Newton's description of the motion of a planet. This uncertainty
in the position and momentum of a particle at any instant is, as we shall
see, a general feature of de Broglie's description of the motion of a particle
particle-like description of the
its associated pilot waves.
Let us consider a freely moving particle and its associated group, or
groups, of pilot waves. We take first the mathematically simplified case
in terms of the propagation of
of the pilot waves, described by equation (6-9), obtained by combining
two unmodulated waves of slightly different k and v. This is
Y = sin 2n[kx where we use
Ak and Av
at the instant
vt]
sin 2-n-p
instead of
dk and
in figure (6-3).
It
Ak)x
dv.
(v
Av)f]
This function
represents
an
is
(6-17)
plotted,
infinite succession
of groups of waves which are traveling in the direction of increasing
Now,
in consideration of the connection that
we have made on
x.
several
previous occasions between the intensity of the pilot waves and the location
of the particle,
it is
apparent that we should interpret
this particular pilot
which it is associated has equal
probability of being located within any one of the groups at the time
/ = t
Considering momentarily a single group, our interpretation of
indicates that even the location of the particle within that group is undetermined to within a distanee comparable to the length A* of the group.
From equation (6-10), or from figure (6-3), we obtain the following
relation between Aa; and the difference AA: between the wave numbers
wave as indicating
that the particle with
of the two constituent waves
Aa; AA:
A
2, in
similar relation can be obtained for the case,
which an
infinitely large
infinitesimally different
k and
(6-18)
mentioned in section
number of unmodulated waves, each with
v,
combine to form a
single traveling group.
PARTICLES
154
AND WAVES
[Ch. 6
Such a group is pictured at t = t in figure (6-8). A single group is formed
by adjusting the phases of all the unmodulated waves so that at one value
of x (the center of the group) they are all in phase and combine to give
large net values of <==i
Proceeding away from this value of x, in either
direction, the unmodulated waves begin to get out of phase with each other
because their wave numbers differ. Beyond certain points they are completely out of phase, and never again get back into phase. Everywhere in
these regions there are, on the average, as many unmodulated waves with
positive values as with negative values, and the waves combine to give a
^= whose value is zero at all x. It is clear that, the larger the range
Figure 6-8.
group of
pilot waves.
of the wave numbers of the unmodulated waves, the smaller is the distance to the points at which they become completely out of phase.
In fact, the distance Ax between these two points is just inversely pro-
Ak of the wave numbers, as in equation (6-18).
However, the proportionality constant is different. From the theory of
Fourier analysis,! it can be shown that a combination of unmodulated
waves with wave numbers covering the range Ak will form a group of
length Ax, where
portional to the range
Ax Ak ~
1/2tt
(6-19)
The exact value of the constant depends on the relative amplitudes of the
different unmodulated waves. The relative amplitudes also determine the
exact shape of the group.
Equation (6-19) describes a property which is common to all wave
motion. Let us now consider the pilot waves which govern the motion of
a particle moving along the x axis. For these waves de Broglie postulates
Jc
1/2.
=pjh
t The "Fourier integral" is used to sum the contributions from the infinitely
number of unmodulated waves with infinitesimally differing wave numbers.
large
THE UNCERTAINTY PRINCIPLE
Sec. 5]
where px
tum
is
is
the
in the
momentum
x direction).
155
of the particle (the x
We
signifies that the
momen-
then have
Ak
Apjh
(6-20)
where Ak is the range of wave numbers involved in the composition of a
group of pilot waves having, at some instant, a length Ax. Npw, we may
say that Ak represents the uncertainty in the wave number which should
be associated with the group at that instant. Therefore we may also say
that Apx represents the uncertainty in the momentum which should be
associated with the particle at the same instant. Furthermore, the discussion above indicates that, for pilot waves, the length Ax of the group
represents the uncertainty in the instantaneous position of the particle.
Thus from equations (6-19) and (6-20) we have the following
relation
between the uncertainties in the instantaneous values of the position and
of the momentum of the particle
Ax APx ~H,
(6-21)
hjlir
is one statement of the uncertainty principle, which was first enunciated
by Heisenberg in the year 1927.
This relation sets a fundamental limit on the ultimate precision with
which we can simultaneously know both the x coordinate of a particle and
We may arbitrarily decrease the
its momentum in the x direction.
uncertainty in one of these quantities, but only at the expense of increasing
the uncertainty in the other. For example, consider a particle whose
associated pilot waves form a group of length ten times less than the group
This
The instantaneous uncertainty in the position
of the particle is then less by a factor of 10. However, equation (6-19)
says that the range of wave numbers present in the new group will be
pictured in figure (6-8).
increased by a factor of 1 0. Consequently the uncertainty in the
momentum
of the particle at that instant will also be increased by a factor of 10, and
the product Aa; Ap x remains approximately equal to h. If we demand that
the uncertainty in our knowledge of the position of the particle at
some
instant approach zero, then equation (6-21) requires that the uncertainty
knowledge of the momentum of the particle at that instant approach
Going to the other limit, we see that if at any time the momentum
of the particle is precisely known (Apx = 0), then at that time the position
must be completely unknown (Aa; oo). A pilot wave associated with a
in our
infinity.
particle in this situation is given in equation (6-8).
Y=
sin
2n(kx
Here px always has the precise value px
of the particle
is
vt)
hk, but at
This
is
(6-22)
any time the position
entirely unknown since Y _^ extends from oo to + oo.
t
PARTICLES
56
It
AND WAVES
[Ch. 6
should be pointed out that the uncertainty principle gives a lower
product Ax Apx That is, it is possible to find a group of waves
limit to the
which the relation corresponding to equation (6-19) is Ax Ak = y,
where y > \\2tt. For a particle whose pilot waves form such a group, the
product of the instantaneous uncertainties in the position and momentum
of the particle would be larger than h. As a limiting example of this,
consider a particle whose associated pilot wave is represented by the
for
function
T = sin 2Tr[kx -
vt]
sin 2n[(k
Ak)x
(r
Av)t]
(6-23)
Ax -- oo,
which Ax Apx is
given in equation (6-9), with a large value of AA:. For this particle
and Apx
= h Ak
situation in
be large.f
than h means simply that the experimentalist has not
achieved the ultimate limit of precision set by the uncertainty principle.
In fact, it is very difficult to obtain Ax Ap x comparable to h in an actual
will also
significantly larger
measurement because of the very small size of this constant. This is why
the uncertainty principle was never noticed in the experiments of classical
10~27 erg-sec, the unBecause of the very small size of h
but it is
is not important in classical mechanics;
extremely important when we are dealing with the very small distances and
momenta involved in atomic and nuclear systems.
If we are considering a three dimensional problem then, according to
mechanics.
certainty principle
Heisenberg:
There are three independently operating uncertainty relations, one for
For rectangular coordinates x, y, z and corresponding
each coordinate.
momentum components px py p z
,
these relations are
Ax Apx > h
Ay Apy > h
Az Ap z % h
(6-24)
If some of the coordinates are angular coordinates, then for each coordinate
6 and corresponding angular momentum L e the uncertainty relation is
A0AL >
fl
(6-25)
f It may seem that the use of functions (6-22) and (6-23) in these examples weakens
the arguments of the second paragraph of section 2, but this is not actually so. Consider
a 1 cm long beam of alpha particles which are being scattered by an atom. From the
point of view of the atom, and in terms of distances of the order of the atomic radius R,
the x position of some particular particle in the beam may be entirely unknown; Aa; ^> R.
on the accuracy with which the momentum is
known, could provide a. good approximation in the vicinity of the atom to the pilot waves
associated with a particle of the beam. However, \x is really 1 cm and not infinity,
and the arguments of section 2 remain valid.
If so, (6-22) or (6-23), depending
Sec. 5]
THE UNCERTAINTY PRINCIPLE
The symbol > has been used
157
to indicate specifically that the uncertainty
Note that there is no restriction at
on products such as Ax &py
The "physical reason" for the existence of a limit in the maximum
precision with which we can simultaneously know the position and the
momentum of a particle is quite easy to understand. What happens is
that a measurement performed to find the value of one of these quantities
principle specifies only a lower limit.
all
Observer
Microscope
Particle
Bohr's microscope experiment.
Figure 6-9.
will
always disturb the particle in such a way as to leave the value of the
good example of this was pointed out by
other quantity uncertain.
Consider a measurement, illustrated in figure (6-9), made
to determine the instantaneous location of a particle by means of a
microscope. In such a measurement the particle must be illuminated,
because it is actually the light quanta scattered by the particle that the
microscopist sees. The resolving power of the microscope determines the
Bohr
in 1928.
ultimate accuracy with which the particle can be located. The resolving
power is known to be approximately A/sin a, where X is the wavelength of
and a is the half-angle subtended by the
of the microscope. So the accuracy of the measurement is
the scattered light
Acc~
objective lens
A
sin
Let us assume that the microscopist needs to see only one scattered light
quantum in order to complete his measurement. The magnitude of the
PARTICLES
158
AND WAVES
[Ch. 6
hjk. But the quantum may have been
anywhere within the angle a, since there is no way of measuring
the exact scattering angle. Consequently the x component of the momentum of the quantum after scattering is uncertain by an amount
momentum
of that quantum
is
scattered
2p
sin
2h
sin
component of the momentum of the quantum can be known
exactly before the scattering (there being no need to know its x coordinate),
conservation of momentum requires that the particle receive an amount of
recoil momentum in the x direction which is uncertain by the same amount.
Thus
,
As
the x
2h sin a
&Px
.
and the product of the uncertainties in the knowledge of the x position
and the x component of momentum of the particle at the instant of
measurement is
Ax Ap^ c=L2h> h
By
using light of shorter wavelength, the microscopist can increase the
accuracy of the position measurement, but
the uncertainty of the
We
is
momentum
this will lead to
an increase in
particle.
see that the state of affairs described
by the uncertainty principle
The quantization of electromagnetic
least one "unit" of light (a quantum of
a direct result of quantization.
radiation requires that either at
momentum p
no
of the
light at all.
hjX)
If
reason in principle
it
must be scattered from the
were not for
why
particle, or else absolutely
would be no
would not be able to see the
this requirement, there
the microscopist
good resolution using an illumination of arbitrarily small
scattered light quantum provides the interaction
which must exist between the measuring instrument and the particle.
This interaction disturbs the particle in a way which is uncontrollable and
unpredictable. As a result, the coordinates and momentum of the particle
cannot be completely known after the measurement. The relation
Ax Apx > h shows that Planck's constant is a measure of the (minimum)
particle with
momentum content. The
magnitude of this uncontrollable disturbance.
There is another form of the uncertainty principle which again follows
from the mathematical properties of waves plus the de Broglie postulate,
in just the same way that equation (6-21) follows from (6-19) and (6-20).
From the theory of Fourier analysis it can be shown that the frequency
range Av, which must be involved in the composition of a single group
of waves of length Ax and group velocity g, is given by the relation
At Av >
1/277-
(6-26)
SOME CONSEQUENCES OF THE UNCERTAINTY PRINCIPLE
Sec. 6]
where At
point;
i.e.,
Ax/g
At
is
is
59
the time required for the group to pass any given
the duration of the pulse of waves.f
Now consider a group of pilot waves of duration At. This quantity
would be a measure of the uncertainty in the time at which the particle,
whose motion is governed by the waves, would pass by a given point.
Furthermore, the frequency of the pilot waves is related to the energy of
the particle by de Broglie's postulate
v
E/h
Av
AE/h
Consequently
The uncertainty Av
AE
in the frequency of the
(6-27)
waves leads to an uncertainty
in the energy associated with the particle,
and
that uncertainty
related to the uncertainty in the time associated with the particle
is
by the
relation
ArAEyh
(6-28)
Heisenberg's interpretation of this uncertainty relation was, however,
than that implied by the discussion above. He stated
much broader
measurement of the energy of a particle (or any system) performed
At must be uncertain by an amount AE, where the
during a time interval
relation between these two quantities
We
shall discuss several
is
At AE y
examples of
h.
this uncertainty relation in sub-
sequent chapters.
6.
Some Consequences
of the Uncertainty Principle
Consider a particle moving freely along the x
instant
the position of the particle
is
axis.
Assume that at the
amount Ax
particle at some later
uncertain by the
Let us calculate the uncertainty in the position of the
time t. According to equations (6-24), the momentum of the particle at
h/Ax Consequently the velocity
is uncertain by the amount Apx
t
t The property of wave motion described by equation (6-26) makes it necessary to
use very high frequencies in the transmission of television programs. The signal from a
10 -6 sec. Thus the range of
television station consists of pulses of duration At
-1
6
10 sec , and the entire broadcast band
frequencies which must be present is Av
-1
-1
6
1.5 x 10" sec ) would be able to accommodate only a
to v
(y ~0.5 x 10 sec
single television "channel." At the frequencies used in television transmission (y
108 sec -1 )
many channels can
fit
into a reasonable portion of the spectrum.
PARTICLES
160
AND WAVES
[Ch. 6
is uncertain by the amount Av = Apjm
and the distance traveled by the particle in the time t
uncertain by the amount
of the particle at that instant
hjm Ax
Ax
tAv
~
is
(6-29)
m Ax
If
by a measurement
range
Ax
at
we have
then in a measurement of
localized the particle within the
position at time / the particle
could be found anywhere within the range A*. Since Ax is inversely
proportional to Ax we see that, the more carefully we localize the particle
,
its
at the initial instant, the less
Furthermore,
we
shall
know about
this uncertainty increases linearly
with
its final
f.f
position.
The behavior of
a system, involving distances and momenta small enough that the uncertainty principle is important, is in complete contrast to the behavior of a
system described by classical mechanics.
In classical mechanics, if the position and momentum of each particle
in an isolated system are known exactly at any instant, the exact behavior
of the particles of the system can be predicted for all future time. The
uncertainty principle shows us that
involving small distances and
it is
impossible to do this for systems
momenta because
it is impossible to know,
with the required accuracy, the instantaneous positions and momenta of
the particles of such a system.
As a
result,
we
be able to
shall
dictions only of the probable behavior of these particles.
make pre-
This idea has
already crept into our discussion through the interpretation of the intensity
of pilot waves, at a particular position, as some sort of measure of the
probability that the associated particle
is
located at that position.
interpretation will be put into a quantitative
form
This
in the next chapter.
In the same chapter we shall develop the differential equation (Schroedinger 's
equation) to which the pilot waves are a solution.
equation
we
In discussing this
behavior of the pilot waves is exactly predictable. But the pilot waves only give information about the probable
behavior of the associated particle.
shall see that the
The uncertainty principle provides us with a deep insight into our ability
to describe the behavior of small scale physical entities.
It also
helps us
to reconcile conflicts which appear to arise
nature of these
entities.
from the dual (particle-wave)
Let us consider, as an example, a double-slit
t This statement implies that the group of pilot waves associated with the particle
spreads in time, so that its length at any instant is essentially the Aa; of equation (6-29).
This can be understood as a consequence of the fact that the unmodulated waves which
form the group have a range of wave velocities Ah> because they have a range of wave
numbers A and a range of frequencies Av. As these waves move with different velocities,
they cannot maintain the phase relations required to form the group and, in time, the
group spreads.
Sec. 6]
SOME CONSEQUENCES OF THE UNCERTAINTY PRINCIPLE
diffraction experiment, using the photoelectric effect as a detector.
161
This
The source emits electromagnetic radiation.
From a study of the energy and time distribution of the photoelectrons
ejected from the photocathode, we conclude that the radiation must
possess particle-like aspects. However, when we measure the number of
is
shown
in figure (6-10).
photoelectrons ejected as a function of the y coordinate, the periodic
pattern which we observe (characteristic double-slit intensity pattern)
forces us to conclude that the radiation
aspects.
If
we
must also possess wave-like
think of the radiation as quanta whose motion
is
governed
Photoelectrons
Source
Slits
Figure 6-10.
Photocathode
An experiment
illustrating the particle-wave duality.
by the wave propagation properties of certain associated waves, then we
are faced with the following problem: Each quantum must pass through
either one slit or the other; if so, how can its motion in the region
to the right of the slits be influenced by the interaction of its associated
waves with a slit through which it did not pass ?
Let us analyze this problem in the spirit of Einstein's analysis of
The statement that a particular
quantum must pass through either one slit or the other is really meaningless unless some provision is made to determine through which slit the
quantum actually passed. One way of doing this would be to place a
number of very small particles in the region immediately to the right of the
slits. After its passage through a slit, the quantum would strike one of the
particles, causing it to recoil. By observing the recoiling particle, we
could tell through which slit the quantum passed, provided the uncertainty
in the y coordinate of the particle after the collision is small compared
the Galilean transformation equations.
to the distance between the
slits;
that
by
is,
<d
provided that
AND WAVES
PARTICLES
162
[Ch. 6
quantum exchanges some y momentum with the
The exact amount exchanged depends on details of the collision
which we have no way of determining. Thus the y momentum exchange
is uncertain by an amount which we call Ap
In order not to destroy the
y
double-slit diffraction pattern produced in the passage of a number of
quanta through the system, A/> must always obey the restriction
Now,
in colliding the
particle.
4ft/P.
<
where 6 is the angle between adjacent minima and maxima of the pattern,
and px is the x component of the momentum of the quantum. If this
condition is not satisfied, a quantum could receive so much y momentum
in the collision that it would end up in what had been a minimum of the
diffraction pattern. Since we can assume that the y momentum of the
particle was precisely known before the collision because there was no
need to know its y position at that time, and since momentum must be
conserved in the collision, b.py
y momentum of the
angle
is
is
also equal to the uncertainty in the
particle after the collision.
Now it is known that the
related to the wavelength of the radiation
between the
slits
Expressing this in terms of the
equation (6-5),
this
Xjld
momentum
of the quantum by means of
we have
6
Combining
and to the distance
by the equation
h/2dp x
with the second inequality above gives
4P.//.
< 6 = h 2dPx
l
so
A/>v
Multiplying this into the
first
<^h/2d
inequality,
Ay Apv
we have
< h/2
is a condition on the uncertainty in the y position and y
of the particle which must always be satisfied in order that
This
able to determine through which
slit
momentum
we may be
each quantum passed without
destroying the double-slit diffraction pattern.
But the uncertainty princannot be satisfied. If we do demonstrate
that each quantum actually did pass through one slit or the other, we shall
ciple tells us that this condition
no longer observe
the double-slit diffraction pattern.
The uncertainty
shows us that the problem we have drawn attention to is actually
illusory. In Chapter 8 we shall see additional examples of the way in which
the uncertainty principle resolves apparent conflicts between the wavelike and particle-like aspects of the same entity.
principle
163
EXERCISES
Ch. 6]
BIBLIOGRAPHY
Bohm,
D.,
Quantum Theory, Prentice-Hall, Englewood Cliffs, N.J., 1951.
and Human Knowledge, John Wiley and Sons,
Bohr, N., Atomic Physics
New
York, 1958.
EXERCISES
1
Add graphically six unmodulated waves of 1 cm amplitudes and wavelengths
1.5, 1.6, 1.7, 1.8, 1.9,
and 2.0 cm, adjusting the phases so
that, at
0,
Carry out the addition to x = 10 cm,
and compare the length Aa; of the group formed, and the range Mc of the wave
numbers used, with equation (6-19). Note that, for a finite number of components with finite differences in wave numbers, a single group is not obtained
because there are always many values of x at which all the components will be
^(x, 0t=t
= + 1 cm for aU tne waves.
in phase again.
Find the possible values of the momentum of the particle, treated in
from the possible values of energy predicted by the WilsonSommerfeld quantization rules. Use these values to evaluate the predicted
2.
exercise 2, Chapter 5,
Show
possible wavelengths of the pilot waves.
that,
with the exception of the
lowest energy state where the momentum is zero, it is just for these wavelengths
that the waves can combine to form standing waves, as stated in section 4.
3.
Discuss the exception mentioned in exercise 2 in terms of the uncertainty
principle.
4.
Consider a
single-slit diffraction
experiment for
particles, similar to that
illustrated in figure (6-1) for electromagnetic radiation. First treat the diffraction
of the pilot waves, and show by a Huygens wavelet construction that the angular
X/d. Next treat the deflection of the
width of the diffraction pattern is
by using the uncertainty principle to estimate the transverse momentum
introduced in passing through the slit by the measurement of transverse position
X/d. Comment
that the slit provides, and show that the deflection angle is d
on why both treatments lead to the same predictions.
is dropping marbles of mass m to the
5. A boy on top of a ladder of height
floor and is trying to hit a crack in the floor. To aim, he is using equipment of
particles
the highest possible precision.
Show
that, despite his great care, the
marbles
will miss the crack by an average distance of the order of
where g
is
the acceleration due to gravity. Using reasonable values of
evaluate this distance.
H and m,
CHAPTER
Schroedinger's
Theory of
Quantum
Mechanics
I.
Introduction
In this chapter we begin our development of Schroedinger's theory of
quantum mechanics. This theory, and its applications to a number of
occupy our attention for much of the remainder of
Furthermore, the discussion of atomic and nuclear physics
typical problems, will
the book.
which follows the discussion of quantum mechanics will be largely based
on the Schroedinger theory. For these reasons our treatment of the
theory, particularly of the mathematical details, will be considerably
more thorough than the treatments we have given previously. This is
necessary because it is often easier to think in terms of the mathematics of
quantum mechanics than it is to think in terms of the somewhat elusive
physical pictures which the mathematics represents and the mathematics
is sufficiently involved that a cursory treatment could not help being
confusing. However, we shall, as usual, be perfectly willing to sacrifice
rigor for clarity when it seems advisable.
Before embarking on Schroedinger's theory, we should mention that an
alternative approach was developed at about the same time by Heisenberg.
His theory did not discuss pilot waves. Instead he dealt only with dynamical
quantities such as x, px E, which were represented by matrices. The
164
Sec. 2]
THE SCHROEDINGER EQUATION
quantum
aspects were introduced into the theory via the uncertainty
principle,
and applied
uncertainty principle
as
is
165
commutation conditions on the matrices. The
Con-
equivalent to the de Broglie postulate.
sequently the Heisenberg and Schroedinger theories are actually identical
in content, although very different in form. Since the Schroedinger theory
is
better adapted to
an introductory treatment of the
subject,
we
consider
here only that theory.
2.
The Schroedinger Equation
should be apparent to the reader that de Broglie's postulate is essenticorrect.
It should also be apparent that the postulate does not
provide us with a complete theory for the behavior of a particle, but only
It
ally
first step toward such a theory. The postulate says that the motion of
a particle is governed by the propagation of its associated pilot waves,
but it does not tell us the way in which these waves propagate. Only for
the simple case of a free particle have we been able to learn some of the
features of the propagation of pilot waves.f To handle the case of a
particle moving under the influence of forces, we must have an equation
the
which
will tell
circumstances.
how
the pilot waves propagate under these more general
we must also have a quantitative connection
In addition,
waves and the associated particle; that is, we must know
exactly how these waves "govern" the motion of the particle. The required
propagation equation for pilot waves, called the Schroedinger equation,
was developed in 1925, and the quantitative connection between these
between the
pilot
waves and the associated
With
particle
was developed
in the following year.
several exceptions, Schroedinger directly followed de Broglie's
He did not use the picturesque term "pilot waves" but instead
denoted both the waves, and the mathematical function T(x, t) which
represents them, by the term wave function. We shall adopt this terminology ourselves. A more serious change was that Schroedinger attempted
only to develop a theory which would be valid in the non-relativistic
range of velocities, although, as we have seen in section 2, Chapter 6,
the de Broglie postulate was consistent with the theory of relativity.
ideas.
Schroedinger adopted de Broglie's two equations
A
and
= h/p
= E\h
,,
(7-1)
._
_.
(7-2)
associated with a free particle
t We know that the velocity of a group of pilot waves
equal to the velocity of the particle (equation 6-13); and also that a group spreads
with increasing time, its length Ax being proportional to / (equation 6-29).
is
SCHROEDINGER'S THEORY OF
166
QUANTUM MECHANICS
but he did not adopt de Broglie's definition of
energy.
E as
Now, from the theory presented in Chapter
[Ch. 7
the total relativistic
equation 1-22"),
apparent that, in the case in which there is a potential energy V, the
total relativistic energy approaches the expression
1 (cf.
it is
+ V+ m
2
^total relativistic ->-/> /2m
C2
in
the non-relativistic limit v/c-+0.
In his non-relativistic theory
Schroedinger did not bother to carry along the constant m c 2 , and he
took for E the classical definition of total energy
E = film + V
m=m
where
(7-3)
We see from equation (7-2) that a change
E will change the value of v. But recall that the experi-
the rest mass.
in the definition of
ments described in the previous chapter were
validity of equation (7-1), not equation (7-2).
all
such as to check the
We
shall see later that the
is of no significance in Schroedinger's theory. However,
one point should be investigated immediately. Assuming equation (7-3),
can we still obtain the necessary result that the group velocity g of the
actual value of v
wave function
Take
for a free particle
equal to the velocity v of the particle?
is
= E\h= pt/lmh +
Vjh
and
k
For a
\\X=p\h
free particle the potential energy
dv
V is
a constant
(cf.
3 below), so
= 2pdp
2mh
Also
we have
From
dk
dp/h
dvjdk
equation (6-11),
Thus we obtain the acceptable
result
g-P d P
mh dp
We
begin our development of the Schroedinger equation by
we know a priori it must satisfy.
listing
three requirements which
1.
It
2. It
must be consistent with equations
must be linear in Y(ar, /). That is,
and (7-3).
and 2(ar, if) are
(7-1), (7-2),
if "x (x, t)
tions to the equation, a linear combination of these functions T(x,
ajTjix,
t)
a^Y^x,
t)
must
also be a solution for
solui)
any values of the
THE SCHROEDINGER EQUATION
Sec. 2]
constants ax and a 2
167
This requirement ensures that we shall be able to add
interference phenomena which
Davisson-Germer experiments. In the
previous chapter we have already assumed that the wave functions can be
added (cf. equation 6-9).
3. The potential energy V is, in general, a function of x and t. In the
together
wave functions to produce the
are observed, for example, in the
special case V(x,
t)
= V a constant, the momentum
= F, and the force
dV _
dV(x,t) _
Q
F=
,
of the particle will
be constant since dpjdt
dx
dx
Also the total energy E is constant in this case. This is the situation of
a free particle with constant values of k = pjh and v = Ejh discussed in
the previous chapter. For this case we require the Schroedinger equation
to have oscillatory traveling wave solutions of constant wave number
and frequency, similar to the wave functions used in that chapter.
Rewriting equations (7-1) and (7-2) in terms of the convenient parameters,
K=
2rrk,
co
(7-4)
2ttv
we have
p =
Combining equations
To
satisfy
(7-3)
requirement
E=hco
HK,
and
(7-5),
we
(7-5)
obtain
H2
K l2m +
the Schroedinger equation must be consistent
V(x,
t)
(7-6)
hco
with equation (7-6).
Requirement 2 demands that every term in the equation be linear in
itself, or a
Y(x, t). Consequently every term must contain either T(a;,
derivative of *(x,
3W(x,
dx
t)
such asf
d 2x(x,
Q
'
dx*
BVjx,
t)
or
""
dt
d^jx, Q
Q
'
at
The equation cannot contain terms independent of Y(x,
i),
'""
or terms such
as \T(x, OFWe shall use requirement 3 as an aid in finding the form of the Schroetake one of the free particle wave functions 'F/s, t),
dinger equation.
We
derivative of
T, such
as dY/dx,
is
linear
inT
T = <rfF + a^
x
then
because
if
SCHROEDINGER'S THEORY OF
168
and then
QUANTUM MECHANICS
[Ch. 7
differential
an equation involving only terms in this function and its
which is consistent with (7-6). This equation will give us a
equation which has the free particle wave function as a solution
when
t)
find
derivatives
V(x,
= V
We
a constant.
shall then postulate that,
general case in which the potential V(x,
t) is
even for the
not a constant, the solutions
of this differential equation are the correct wave functions to be associated
with a particle moving under the influence of such a potential.
Let us take the simplest free particle wave function discussed in the
Y = sin 2n(kx -
previous chapter,
Y,
sin
In terms of
vt).
(Kx
we have assumed
-'
V(x,
= K cos {Kx -
t)
cos
(Kx
=
2-y
2
cot),
(7-7)
must
-K
- = -co
f
cot),
dt
We
this is
K and
to
are constants
dx
co
co,
= V We find
ox
and
cot)
Evaluate some of its derivatives, remembering that
because
sin
(Kx
sin
(Kx
cot)
(7-8)
2
cot)
dt*
find
an equation consistent with
K
2m
+ V =
hco
(7-9)
The fact that differentiating twice with respect to x brings in a factor of
and differentiating once with respect to t brings in a" factor of co
suggests that we try the equation
K2
d 2xV
lit
where a and
/?
?W
**,->%
are constants to be determined.
into this equation,
-a sin (Kx -
we
cot)K 2
o-m
Inserting (7-7)
and (7-8)
find
sin
(Kx
cot)V
-/S cos (Kx
coi)co
(7-11)
Even though the constants a and /9 are at our disposal, we obviously
cannot make this equation agree with (7-9) for all possible values of x
and t, and so this attempt has failed.
The observation
into cos (Kx
cot),
that the differentiation process changes sin
and
vice versa, suggests that
we
(Kx
cot)
try again to construct
the differential equation by starting with a combination of two functions
of the form (7-7), one being a sine and the other a cosine. That is, we try
>
cos (Kx
cot)
sin
(Kx
cot)
(7-12)
Sec. 2]
THE SCHROEDINGER EQUATION
169
where y is a constant to be determined that gives us some freedom which,
might be guessed, we shall need. Evaluating some derivatives, we find
it
^ = -A
sin
(Kx
+ Ky cos (Kx -
cot)
cot)
dx
^-' = -K 2 cos {Kx 3T
*-
co sin
(Kx
- K2y sin (Kx -
cot)
cot)
coy cos (Aa;
cot)
(7-13)
wf)
dt
We
again assume the form of the differential equation to be given by
equation (7-10) and substitute these derivatives into that equation. This
gives
xK y sin (Kx - cot) + V cos (Kx - cot)
+ V y sin (Kx cot) = fico sin (Kx cot) /3coy cos (Kx cot)
[-a.K2 +V + ficoy] cos (Kx - cot)
sin (Ax - cot) =
+ [-<x.K y + V y olK2 cos (Kx
cot)
fico]
If this is to
be true for
all
values of x and
cosine and the sine must vanish. That
(7-15),
and three
the coefficients of both the
is,
-K + V =
-Py<o
(7-14)
-aA2 + V =
&
V
(7-15)
This looks encouraging;
/,
we have
co
three equations to satisfy, (7-9), (7-14),
free constants, a,
/S,
Subtracting (7-15) from (7-14)
y.
gives
jiyco
= -1/y
/ = -1
_
y = V-1 =
co
(7-16)
Substituting this into equation (7-14) yields
-ocA2
Comparing
this
+ V =
equation with (7-9),
we
Ttfco
see that
= -h2j2m
= n, p = ^hji =
a
^tp
ih
(7-17)
There are two possible
QUANTUM MECHANICS
SCHROEDINGER'S THEORY OF
170
+i
then
ft
or y
= + ih,
/.
and the
sets
of solutions, depending on whether
Following
common
differential
equation (7-10)
fc2
o2uy
we
we
take
+i;
take y
is
W""?
3 vp
-r7t
2m
car
This equation
practice,
[Ch. 7
(7-18)
3t
consistent with the three a priori requirements.
is
Equation (7-18) has been obtained for the special case of a constant
potential V(x,
= V
t)
from a
starting
free particle
wave function t
We now assume
similar to those used in the previous chapter for this case.
which V(x, t) is not a constant, the differential
equation controlling the propagation of the wave function is of the same
that, for the general case in
form
That
as (7-18).
is
It is
equation to be
-we postulate the
oxr2
2m
This
is,
T(x,
V(x,
t)
ih
(7-19)
ct
the famous Schroedinger equation.
a second order partial differential equation
partial derivatives) for
"^*'
T as a function of x and
similar to other partial differential equations
However, there
(i.e.,
In
t.
which
involving second
many
respects
it is
arise in the theories of
one feature of the Schroedinger
way from the equations of
classical physics.
The Schroedinger equation contains the imaginary
number i. As a consequence, its solutions are necessarily complex
functions of x and t. We have already seen one example of this; we were
forced to take a complex function
classical physics.f
equation which
sets it off in a
is
very striking
Y = cos (Kx
r
cot)
sin
(Kx
cot)
(7-20)
for the case of the free particle.
3.
Interpretation of the
Wave
Function
f) is an inherently complex function.
We did
Chapter 6 when we developed some of the qualitative
properties of the free particle wave functions. However, each of the arguments of that chapter could be stated again, always replacing the phrase
"wave function" by, for example, the phrase "real part of the wave
The wave function Y(x,
not
know
this in
t See the equation (7-58) for the propagation of transverse
waves in a stretched
string.
Sec. 3]
INTERPRETATION OF THE
function."!
WAVE FUNCTION
171
In every case the argument would remain valid and the
Recall also that in the discussion of
conclusion would remain the same.
the diiference between group
and wave
velocities, in section 2,
Chapter
6,
described the results of a hypothetical measurement of the value of
This
the wave function for various coordinates a; at a certain instant t
was done simply to provide something concrete to think about. If we
we
were to repeat that discussion, we would now have sufficient acquaintance
with the idea of a wave function to render such an artifice unnecessary.
We would also realize that the hypothetical instruments for measuring the
value of a wave function cannot possibly
exist, since the
value of an in-
herently complex quantity cannot be measured with an actual physical
instrument.
wave functions are complex functions should not be
considered a weak point of the quantum mechanical theory. Actually,
it is a desirable feature because it makes it immediately obvious that we
should not attempt to attribute to wave functions a physical existence in
the same sense that water waves have a physical existence. That is, we
should not try to answer, or even pose, the questions: Exactly what is
waving, and what is it waving in ? The reader will recall that consideration
The
fact that
of just such questions concerning the nature of electromagnetic waves
led nineteenth century physicists to the fallacious concept of the ether.
the wave functions are complex, there is no temptation to make the
same mistake again. Instead, it is apparent from the outset that the wave
functions are computational devices which have an existence or at least
a significance only in the context of the Schroedinger theory of which
they are a part. This point is emphasized by the fact that wave functions
never enter in the Heisenberg theory. Yet that theory is completely
equivalent, in terms of its end results, to the Schroedinger theory. These
comments should not be construed as implying that the wave functions
are lacking in physical interest. We shall see in this section, and in section
8, that a wave function actually contains all the information which the
uncertainty principle allows us to know about the associated particle.
In order to obtain this information we must have a quantitative connection between T(x, i) and the dynamical quantities describing the
associated particle. Now, it has been indicated in the previous chapter
that there must be a relation between some measure of the intensity of
Wipe, t), in the region of the coordinate x at the time t, and the probability
As
of finding the associated
apparent that T(x, t) cannot
because the probability density is a real
per unit length, or probability density P(x,
particle in that region at that instant.
simply be set equal to P{x,
t
The real
(7-22).
part of the
t)
wave function Y(x,
t),
It is
t) is
the function R(x,
t)
denned by equation
QUANTUM MECHANICS
SCHROEDINGER'S THEORY OF
172
[Ch. 7
wave function is complex. However, a way to make
was proposed in 1926 by Born in the form of the following
quantity whereas the
this association
postulate:
If,
at the instant
a measurement
t,
with the wave function Y(a;,
particle will be found at
made
is
to locate the particle associated
then the probability P(x,
r),
a coordinate between x and x
P(x,t)dx
dx
i)
dx that the
is
= xF*(x,t)Y(x,t)dx
(7-21)
The symbol W*(x, t) represents the complex conjugate of T(a;,
meaning of this term, consider the complex function
i).
To
illustrate the
*{x,
t)
R(x,
t)
(7-22)
I(x, i)
where R(x, f) and I(x, t) are real functions. (A complex function can
always be decomposed in this manner.) Then the complex conjugate of
this function is defined as
T*(*,
An
example we
i)
come
shall often
R(x,
across
Y(z,
This provides a convenient
(7-20) because e iz
(7-23),
it is
easy to
I(x, t)
(7-23)
is
i(-
Kx -<ot)
(7-24)
way of writing the free particle wave function
From this relation and the definition
i sin z.f
cos z
show
t)
that
Y*(x,
t)
<.-*<**-<><>
Now let us evaluate T*T from equations
T*T = (R Y*Y = R*-
(7-22)
(7-25)
and
(7-23).
This gives
+ il)
I =R? + P
U)(R
i
= I. Thus we see that Y* ?" is always a real function, and Born
i
not inconsistent in equating T*(a;, O^fo
to P{x, i).
Of course, this does not prove that these two quantities must be equated
because there are other functions of T(a;, t), such as the absolute value
|T(a;, t)\, which are also always real.
Some justification for Born's
since
is
choice of "*(#,
t)
T(a;,
f)
to give the probability density can be found
t This equality can easily be proved
these functions:
e( '
from a consideration of the
(izy
(fc)*
(fe)
+
+ (te)
- 4+
+
+ 2!
1!
3!
4!
1 4-
z2
-I-
z4
=l--+_-- +
sinz=z __ + _-_ +
z
co S2
z6
z>
--.
...
series
expansions of
Sec. 3]
WAVE FUNCTION
INTERPRETATION OF THE
from the following argument. Consider
173
T and Y* which are, respectively,
solutions to the Schroedinger equation,
aw
- + V =ih
t2 22ur
2m
(7-26)
da?
dt
and the complex conjugate of the Schroedinger equation,!
t2
2m dx
avp*
+ VY*
ih
3*2
3x2
(7-26')
dt
Multiply equation (7-26) by T* and (7-26')
'2m \
g2w*
JL _!_
byT and subtract.
\
This gives
dt J
dt
which reduces to
o
fc2
32W
\p* " T
dx*
2m\
_ _2_
and further
T
_ ip v3 *F*\
=
)
dx2
j^
_3 vp*ip
dt
to
2mdx\
dx
dx
dt
Integrating both sides over x between the limits x x to x2
2m
Jx!
dx \
dx
dx
we
find
J Xl dt
which gives
ih T
dW
dW*!**
+ Jl Y* 1 _ Y 2i-
2m L
3x
3a;
J Xl
=-3
C x*
Y*Wdx
(7-27)
dt J Xl
Take the complex conjugate of every term of equation
(7-26).
This gives
3Y\
easy to show from the definition (7-23) that the complex conjugate of a product
equal to the product of the complex conjugates. Thus we have
It is
(-)W)
is
="0*
According to equation (7-23), the complex conjugate of a purely real quantity, such as
(h 2 l2m) or V, is equal to that quantity. The complex conjugate of a purely imaginary
quantity, such as (ih), is equal to the negative of the quantity. Furthermore, it is easy
to show from (7-23) and the mathematical definition of a derivative that the complex
conjugate of the derivative of a function is equal to the derivative of the complex
conjugate of the function. From these properties we obtain immediately equation
(7-260.
QUANTUM MECHANICS
SCHROEDINGER'S THEORY OF
174
Now let us
take for
vp
Y the free particle wave function (7-24).
HKx-iot)
\p*
an(j
[Ch. 7
Then
-i(Kx-af)
Differentiation yields
7)W
dx
and
? XV*
-iKe~ Kx at)
i<-
= -iKY*
dx
Substituting into equation (7-27),
we have
>2
KV*^
is
v,
where v
the velocity of the
is
Therefore
(Y*Y)
This
= pjm =
equation (7-5), hK/rn
particle.
w*Ydx
ut Jxi
x1
From
r*2
=-
really
(vW**F)x=Xi
= |-
"*
Y*Y r
(7-28)
not a very interesting equation for the case of a free particle,
where V(x, t) = constant, because Y*Y = e -w-<t)S Kx m = 1, r =
= 0. However, if we consider a
constant, and the equation amounts to
case in which V(x, t) is a very slowly varying function of x and / (almost a
constant), then, at least for a restricted range of the x axis, the wave
function can be written
*p
_ A eHKx-wt)
where A, K, and co are very slowly varying functions of x and t. In this
= A 2 and also t> is not constant, but we shall still obtain equation
case Y*
as
we
(7-28)
can ignore derivatives of A, K, and co in evaluating dWjdx
and 3Y*/dx. Under these circumstances we can use equation (7-28) to
confirm the identification ofY*Y with the probability density by comparing
this equation with the one dimensional conservation equation of flowing
fluids,
{vp) x = Xl
(vp) x=Xt
= d
C Xi
p dx
(7-29)
Ot Jx,
(S) x=Xl
This equation
p
is
is
is
% dx
- (S)._ = f
at Jxt
mass
flux
The quantity
The quantity vp = S
illustrated schematically in figure (7-1).
the mass density of a fluid, and v
the
(7-29')
of the
fluid,
which
is its
is
the
velocity.
mass crossing a given point x
WAVE FUNCTION
INTERPRETATION OF THE
Sec. 3]
per unit time, and the
left side
of the equation
is
175
equal to the mass of fluid
flowing into a region per unit time at x1; less the mass flowing out per unit
The integral is the total mass contained in the region xx to
x 2 so the right side is just the change in mass in that region per unit time.
Thus the equation simply states that in the region xx to x2 the fluid is
time at x %
neither created nor destroyed
We
conserved.
is
it
show
in the next
paragraph that the same kind of conservation equation must apply to
the quantum mechanical probability. By comparing equations (7-28)
and (7-29), we see that this will be true if we accept Born's postulate.
S * = *2
*1
Figure 7-1.
Illustrating
*2
the one dimensional conservation equation.
The reason why we must have conservation of probability is easy
" T*(x, t)W{x,
Consider the integral
oo
and
r
and that
**(x,
must be true
this
conserved. Unless
we
+ oo
at the time
t)
for all
This
(x,t)dx
It is clear
t.
Thus the
t.
are willing to
is
to see.
just the total prob-
with the wave function T(x,
ability that the particle associated
somewhere between
must have
i)
dx.
make some
t)
be
then that
we
(7-30)
total probability
must be
very special assumptions,
can only be achieved by conserving probability in every region x L to x 2
we must have a conservation equation of the form (7-29).
Before leaving this argument, we take the opportunity to obtain from it
the formula for the probability flux S(x, t). This quantity is the probability per unit time that the particle associated with the wave function
this
In other words,
T(a;,
i)
will cross the point x, in the direction
By comparing
coordinate.
in the case where V(x,
v**(x,
(x,
i)
t).
An
i)
of increasing values of that
equations (7-28) and (7-29')
is
we
a slowly varying function, S(x,
expression for S(x,
t)
see that,
i)
is
just
of general validity can be
obtained by considering equation (7-27), which was derived without any
approximations. If
we postulate
the probability flux to be
"
S(x
t)
= -
J* \v*(x,
?*&<> - T(x,
2mL
3W * (X
dx
'
f)
(7-31)
dx
then equation (7-27) 'becomes a quantum mechanical probability conservation equation of general validity.
postulate.
It is
easy to
show
that S(x,
We take this as justification for the
i) is
always
real.
SCHROEDINGER'S THEORY OF
176
4.
QUANTUM MECHANICS
[Ch. 7
The Time Independent Schroedinger Equation
The Schroedinger equation
Y as a function of x and
(7-19)
is
a partial differential equation for
A standard technique for the solution of such
t.
an equation is to look for solutions T(, t) which are products of a function
of x times a function of t, that is, solutions of the form
Y(*,
We
xp(x)
such solutions exist
(7-32)
<f>(t)
the potential energy is a function
procedure reduces the problem of
solving the partial differential equation to the problem of solving two
ordinary differential equations.
shall see that
of x alone.
We
if
shall also see that this
Let us substitute (7-32) into (7-19), assuming V(x,
y (x)
-T-f-%
2m oxr
#*>
F(a:
>
*<*> #*>
ih
t)
V(x).
This gives
**) *
f
at
Now
ax2
dx2
dx2
the notation di ip(x)ldx2 being redundant with
function of x only.
d2y(x)\dx2
since
xp
is
we have
Similarly
dt
dt
and the equation becomes
2m
+ n*) **)
dx*
Dividing both sides by y(x)
i**) 4
dt
<f>(i),
tf.x)
<Kt)
2m dx2
we have
V(X) W(X)}
r
ill
4>{i)
dt
The
left side of the equation depends only on the variable x; the right
depends only on the variable t. Since x and t are independent variables, both sides must be equal to a quantity which depends on neither
x nor t in order that the equality of the left and right sides of the equation
can be true for arbitrary values of the variables. Thus both sides must be
equal to the same separation constant C. That is,
side
-M-f^ +
xp(x)
2m
dxr
**>*>}
1
(7 - 33)
THE TIME INDEPENDENT SCHROEDINGER EQUATION
Sec. 4]
177
and
f^ =
(7-33')
'
dt
<Kt)
Equation (7-33') is a simple first order ordinary
^ as a function of t. It has the solution
#t)
differential
equation for
- iCt/H
(7-34)
This solution can be verified by differentiating and substituting into
equation (7-33')
dt
which shows that
-^L
in-
#0 a
The function
<f>(t)
in
Q.E.D.
C;
#r)L
a complex oscillatory function of time,
is
tfi)
e~
iCtin
cos (Ctjh)
with frequency v given by 2ttv
Thus
Cjh.
sin (Cf///)
Cjlirh
C//i.
But,
according to equations (7-2) and (7-3), we must have v
E\h, where E
is the total energy of the particle, because <(f) gives W(x, i) its time
dependence. Thus C must be equal to the total energy E. Knowing this,
we may
write equation (7-33) as
^ +
f
2m
V(x)
y>(x)
Ey>(x)
(7-35)
ax*
C=E
Setting
we can
in equation (7-34),
write the solution to the
Schroedinger equation as
Y(x,
where
%p(x) is
y{x)e- iEtin
(7-36)
a solution to equation (7-35), which
is
called the time
independent Schroedinger equation.
an ordinary second order differential equation for y> as a function
contains no imaginary numbers and, consequently, its solutions
are not necessarily complex functions.! In section 7 we shall show
It is
of
x.
y>(x)
Of
It
course, they can be complex
linear in y>(x) so
we can always
if
we
wish, because the differential equation
take a solution which
is
the
sum of a
is
real solution plus
times a second real solution, provided the two solutions correspond to the same value
of E. This is proved in the footnote on page 189. An example is the wave function
(7-24), which is a solution for the potential V(x, t)
0. This can be written as
"{x, t)
y>(x) <(f),
and
y>(x)
where
eiKx
cos
=
#<) =
Kx
e~ imt
sin
Kx
-*/
SCHROEDINGER'S THEORY OF
178
that this equation
is
QUANTUM MECHANICS
[Ch. 7
very closely related to the time independent differ-
wave motion. The functions f(x) are called
The reader is cautioned to keep clearly in mind the
between the eigenfunctions ip{x) and the wave functions Y(a;, t),
ential equation for classical
the eigenfunctions.
difference
and
also
the
difference
between the time independent Schroedinger
equation and the Schroedinger equation.
5.
Energy Quantization
in
the Schroedinger Theory
Energy quantization appears in a very natural way in the Schroedinger
theory. In this section
of
how
we
comes about.
this
sented in Chapter
present, in
some
detail,
a qualitative description
A number of quantitative examples will be pre-
8.
Consider a particle of mass m moving under the influence of a force
F(x, t), Corresponding to this force is a potential energy V(x, t). The
two are related by the equation
ey
F(x,
a
t)=
dV{x,
t)
(7-37)
ox
To determine
the motion of the particle under the influence of this
from the Schroedinger theory, we must solve the Schroedinger
equation with the potential V{x, t). Now consider the class of problems
for which the potential is a function of x only. (The more complicated
potential
case of a time dependent potential will be discussed in a subsequent
For a time independent potential, we know that there are
form *(x, t) = f(x) <j>(t). Since <f>(i) has already been
evaluated (equation 7-34), the problem of finding these solutions reduces
to the problem of finding the solutions ip(x) to the time independent
chapter.)
solutions of the
Schroedinger equation for the potential V(x) pertinent to the problem.
The behavior of the particle is then specified by the functions W{x, t)
through the probability postulates (7-21) and (7-31). Energy quantiarises because, in solving the time independent Schroedinger
equation, we shall find that acceptable solutions exist only for certain values
zation
of the total energy E.
To be an
Iy)(x)/dx
acceptable solution, an eigenfunction
must obey the following
dwi iC)
la. y>(x)
must be
finite.
y>(x)
and
its
restrictions for all values of
lb.
derivative
a;:
must be finite.
dx
(7-38)
dwi
r CC must be continuous.
2b.
I
2a. w(x)
must be continuous.
dx
ENERGY QUANTIZATION
Sec. 5]
If f(x) or drp(x)jdx
Y(x,
t)
179
were to violate requirement
e-
lEt/n
y>(x)
dY(x,
and
la or lb,
t)
then
-iEt/n dip(x)
dx
dx
would
also.
This would
mean
that the probability density
^,0 = ^,0^,0
and/or the probability flux
2m
were not well defined
all
dx
that
is,
values of the variable s.f
dx
did not have a finite and definite value for
i)
and
between potential and
total
This cannot be allowed since P(x,
*o*
Figure 7-2.
energies
S(x,
t)
relation
time independent Schroedinger equation.
describe the behavior of a real physical particle
defined.
2a.
An energy diagram showing the
in a typical
In order to have requirement
The need
for requirement
lb,
we must
2b can be demonstrated by considering the
time independent Schroedinger equation (7-35), which
dx2
and must be well
also have requirement
we
write
J1P
(7-39)
tf
For finite V(x), E, and ip(x), we see that tPy^/dx2 must be finite. This
demands that we impose requirement 2b.
Keeping in mind these restrictions on the acceptable form of y(x), let
us mentally solve the time independent Schroedinger equation for a
potential energy V(x) of the
solve the equation"
in figure (7-2). By "mentally
we imagine going through exactly the
form indicated
we mean
that
same type of integration procedure which would be used
in finding
t All the eigenfunctions we shall deal with in the next few chapters are single valued,
so the requirement that P(pc, t) and S(x, t) have a definite value for all values of x is
automatically satisfied. However, for the
atom eigenfunctions treated in Chapter 10
it will be necessary to consider this requirement explicitly.
180
SCHROEDINGER'S THEORY OF
QUANTUM MECHANICS
[Ch. 7
solutions to the differential equation with the aid of a high speed electronic
computing machine. This technique
of the function V(x)
is
used in situations in which the form
such that the standard mathematical methods for
obtaining an explicit integration of the differential equation (which will
is
be presented in the following chapters) are not applicable.
In order to integrate the differential equation (7-39), we must know the
value of the constant total energy E of the particle. But E is not known
initially
and
see if
choose
Energy
we know is the potential V(x). Consequently we must
choose a value for this constant, put this value in the equation,
we can find an acceptable solution. The value of E which we
all
arbitrarily
on the
indicated
is
We
E.
plot in figure (7-2)
note that, for the value of
=E
divide the x axis into three regions
first
and
third
second region
line:
chosen, there are two
and the line Energy = V(x), which
x < x', x'
x ^ x", x > x". In
regions the quantity [V(x) E] is positive; in the
Energy
intersections of the line
the
by the horizontal
this quantity is negative.
In integrating the differential equation on a computing machine we must
start at some point x
which we shall take such that x'
x
x", and
initial
equation
linear in ip(x), the value
is
value for the function
Let us choose [v(x)L
<
<
assume an
We
xp{x) at
we choose
must
for
Since the
that point.
[fix)],,
is
immaterial.
assume an initial value of
Then, knowing the values of
the function and its derivative at the point x we can calculate their values
at the nearby point x as follows. For (x x x ) very small, [ip(x)] x is
given simply by
+l-t
the derivative [df(x)/dx] x
also
at that point.
ix*)]*,
[**)].
A similar formula for [dii>(x)ldx]x
equation (7-39), which
r^
1
Oi
l ax ^ Xo
*o)
can be found by using the Schroedinger
is
dx2
Lv(x)
\V{x)
VV{X)
~E
tf
*)
or
J tfy(x) = 2m
"|
E\y{x) dx
dx J
So
dy(x)}
dy>(x)
-
dx
provided (x 1
) is
dx*]
A?
^ ^^ ~
v
*o }
very small.
t As y>(x) is not necessarily a
that it is real.
complex function, we simplify
this
argument by assuming
ENERGY QUANTIZATION
Sec. 5]
181
Repeating the procedure from the starting point x x , we can find y>(x)
dy)(x)/dx at x2
Continuing in this manner, we can trace out the
function and its derivative over the entire x axis. Now note that in
x
x", where [V(x) E] < 0, the sign of the change in the
the region x'
and
^ ^
derivative for increasing values of
if
is
This means that the function
itself.
the value of the function
is
is positive,
opposite to the sign of the function
concave downwards in that region
and concave upwards if its value is
+1
A'
'
i
t.
Figure 7-3.
Illustrating three unsuccessful attempts at numerical integration of a typical
time independent Schroedinger equation.
In the regions x
negative.
function
concave downwards
On curve
x
< x'
concave upwards
is
if
and x"
< x,
where [V(x)
the value of the function
the value
of figure (7-3)
which traced
if
is
E]
>
the
0,
positive,
and
negative.
is
we record
the results of a calculation, starting
and dxpjdx in the direction of increasing x. For
x < x" the function was concave downwards, but on passing x" it became
concave upwards. Although the derivative was negative at x = x", it
soon became zero, and then positive. The function ip then started increasing, and matters rapidly went from bad to worse because, according to
at
ip
Schroedinger's equation, the rate of change of the derivative,
is
proportional to
ip.
we found
In this calculation
ip
i.e.
cPipjdx2 ,
oo as x
oo.
not acceptable behavior for an eigenfunction, we made a
second attempt in which we tried to avoid this divergent behavior by
Since this
is
changing our choice of the
y>
value of the derivative
initial
[dip/dx] x
The
We
were not quite successful because
became negative in the region where the function is concave downwards
results are indicated in curve 2.
if its
value
is
negative.
function should
now be
The
difficulty in obtaining
apparent.
It
an acceptable eigen-
should also be apparent that, by
making exactly the right choice of [dxpjdx]x we shall be able to find a function whose acceptable behavior with increasing x is as shown in curve 3.
,
SCHROEDINGER'S THEORY OF
182
This
QUANTUM MECHANICS
asymptotically approaches the x axis.
ip
concave upwards
In figure (7-3)
it
becomes, so
it
The
closer
it
[Ch. 7
gets the less
never turns up.
we also indicate with a dashed curve the results of extend-
ing the function of curve 3 in the direction of decreasing x.
From the dis-
cussion above we must expect that the function will, in general, diverge when
we extendit to small x. We cannot adjust \dy\dx\ as that would disturb the
situation at large x. Nor can we attempt to prevent divergence at both large
,
7-4.
Figure
The
first
three
acceptable
solutions
of a
time independent
typical
Schroedinger equation.
and small x by joining two functions of different derivative at x = x This
ruled out by the requirement that the derivative of an eigenfunction be
.
is
everywhere continuous.
value of the total energy
We are forced to conclude that, for the particular
E which we initially chose, there is no acceptable
solution to the time independent Schroedinger equation. The relation between a function y) and its second derivative d^ipjch?, imposed by the equation for that value of E, is such that the function must diverge at either
large x or small x (or both).
However, by repeating this procedure for many different choices of the
energy E, we shall eventually find a value Ex for which the time independent
Schroedinger equation has an acceptable solution, y>x(x). In fact, there
will, in general, be a number of allowed values of total energy, Ex E2
E3
for which the time independent Schroedinger equation has
,
acceptable solutions
the form of the
for both small
shown
%()
y) x (x), y> 2 (x), y>3 (x),
first
....
and large x
is
the
same
For x
essentially similar to the behavior of yt^x).
derivative
is relatively
we
indicate
The behavior of
>
the behavior of
But, since
its
second
larger in magnitude, y^fx) crosses the axis at
value of x less than x but greater than
tp^x)
as the behavior of the function
in curve 3 of figure (7-3) for large x.
is
In figure (7-4)
three acceptable solutions.
x'.
When this happens,
some
the sign of
the second derivative reverses and the function becomes concave upwards.
At x
x'
the second derivative reverses again and, for x
asymptotically approaches the axis.
<
x',
the function
ENERGY QUANTIZATION
Sec. 5]
183
From figure (7-4), and from the time independent Schroedinger equation,
&? = ^ \y
- em*)
{x)
k2
dx2
see that the allowed energy E2 is greater than the allowed energy
Consider the point x where both ip x (x) and y2 (x) have the same value.
is apparent from figure (7-4) that at this point
we can
Ev
It
>
dx2
dx2
Thus
- E2 >
- E]\,
E% > E
From a similar argument we can show that E3 > E It is also apparent
(E3 EJ, etc., are not infinitesimals since,
that the quantities {E2 E
|
V(x)
V(x)
and
x),
for example, the quantity
d\x)
dx
d2y>i(x)
dx2
not an infinitesimal. Thus the allowed values of energy are well
separated and form a discrete set of energies. For a particle moving under
the influence of a time independent potential V(x), acceptable solutions to
the Schroedinger equation exist only if the total energy of the particle is
quantized to be precisely equal to one of the discrete set of energies
is
E\,
E2 E3
,
....
This statement
is
true as long as the relation between the potential
E is essentially as
shown in figure (7-2);
and an *" such that [V(x) E] >
for all x < x' and all x > x". For a potential of the type shown in that
figure, in which V(x) has a finite limiting value V as x becomes very large,
there is generally room only for a finite number of discrete allowed energy
values which satisfy that condition^ This is illustrated in figure (7-5).
For E > Fj, the situation changes. There are now only two regions of the
x axis: x < x' and x > x'. In the second region [V(x) E] will be
negative for all values of x out to infinity. But, when [V(x) E] is
negative, y>(x) is concave downwards if its value is positive, and concave
upwards if its value is negative. It always tends to return to the axis and
energy V(x) and the total energy
that
is,
as long as there exists an *'
t Potentials which approach the limiting value Vx very slowly can be an exception,
because the separation between the allowed energies En and E +1 becomes very small
as V t En becomes very small. In such a case it is possible for there to be an infinite
number of discrete values of E which satisfy the condition Vi > En An important
_1
which very slowly approaches the
example is the Coulomb potential V(f) oc
.
limiting value V%
= 0.
184
SCHROEDINGER'S THEORY OF
QUANTUM MECHANICS
[Ch. 7
is, therefore, an oscillatory function.
Consequently there will be no
problem of %p(x) diverging for large values of a;. Since we can always make
ip(x) asymptotically approach the axis for small values of a; by a proper
choice of [dy)(x)/dx] x we shall be able to find an acceptable eigenfunction
for any value of E > V
Thus the allowed energy values for
are
continuously distributed. They are said to form a continuum. It is evident
,
E>V
Continuum
Figure 7-5. The discrete and continuum values of allowed energy for a typical time
independent Schroedinger equation.
that, if the potential V(x) is limited for small values
of x, or for both large
and small values of a;, then the allowed energy values will form a continuum
for all energies greater than the smallest limit.
We
note that
when
the relation between the potential energy V(x) and
such that classically the particle would be bound to a
finite region of the x axis, in order to satisfy the classical requirement that
the total energy E can never be less than the potential energy V(x), then
according to the Schroedinger theory the total energy of the particle can
the total energy
E is
only be one of a discrete set of energies. When the relation between
V(x) and E is such that classically the particle is unbound and could move
in an infinite region of the x axis, then according to the Schroedinger
theory the total energy can have any value and the allowed energies form a
continuum. All the conclusions we have obtained from "mentally solving"
the time independent Schroedinger equation will be verified in the following chapters by explicit integrations of this equation.
6.
Mathematical Properties of the Wave Functions and
Eigenfunctions
We
(a)
have found that for a particular time independent potential
V(x) acceptable solutions to the Schroedinger equation exist only for
MATHEMATICAL PROPERTIES
Sec. 6]
which we
certain values of the energy,
185
in order of increasing energy as
list
E* E2 ,...,En ,...
(7-40)
These energies are called the eigenvalues of the potential. The eigenvalues
early in the list are discrete. However, unless the potential V(x) increases
without limit for both very large and very small values of a;, the eigenvalues
become continuous beyond a certain point. Corresponding to each eigenvalue
is
an eigenfunction
V>i(x)>
%(*)>
C7 41 )
Vn(x),
>
a solution to the time independent Schroedinger equation for the
potential V(x). For each eigenvalue there is also a corresponding wave
which
is
function
^(x,
t),
Y (z, 0,
Yn
(7-42)
(*, /),...
which is a solution to the Schroedinger equation for the potential.
According to equation (7-36), we know that these wave functions are
iE ^ /n
e-
The index
n,
Vl (x), e
- iE * t/n
y> 2 (x),
..., e-
iE " tfn
which takes on integral values from
used to designate a particular eigenvalue and
function and wave function,
(b)
called the
is
Each of the wave functions
Tn
wave
to oo,
(7-43)
and which
quantum number.
is a particular solution to the
Since that equation
functions, any linear combination of these functions
Let us check this for the simple case
also a solution.
W(x,
t)
ajYjfx,
t)
ajV^x,
Substituting this function into the Schroedinger equation,
3Y
F
+ VW - =
2m
h 2 d 2x
ih
dx*
dt
we have
h2 (
T\
2
+ V[a&x +
r
is
corresponding eigen-
(a;, t)
Schroedinger equation for the potential V{x).
linear in the
its
f n(x),
T - in{a^ + a,^ ) =
a2
2)
is
is
SCHROEDINGER'S THEORY OF
186
Since
and
2 satisfy
QUANTUM MECHANICS
[Ch. 7
the Schroedinger equation, both brackets vanish.
Thus
1
2m dx2
dt
+
It is
obvious that
this
fli
dx2
2m
+ FY2 -i/^l =0;
Q.E.D.
dt J
argument may be extended to show that the function
= 2 aY K (z,
=
Y(z,
(7-44)
where the a are arbitrary constants, is a solution to the Schroedinger
equation. In fact, it is the most general form of the solution for the
potential V(x).
(c)
Let us calculate the probability density ~V*(x, t^ix, t) for the
wave function is of the form of
general case of a particle whose associated
the
wave function
(7-44).
We
write
= 1 a/-*"VW
Y(s,
Then
the complex conjugate
Y*(z,
We
it
(7-45)
is
0=1 a?e
+OBrt
(7-45')
"tyi*(*)
have used a different symbol to designate the index of summation
two series to remind ourselves that these two series are independent
for the
and that
summation
their
series together,
Multiplying the two
indices are independent.
we havef
Y*(*, t)Y(x,
= | a*H a nV Z(x) n (x)
+ IIa?"nV>X*)Vn(x)e- i(En El)t/n
y>
t For the benefit of readers not proficient in handling series,
writing out a few terms.
[afe^i'/V
Iaie- i
ale iE i t l>hp*
i'"tyi
(7-46)
a$eiE Jl h y*
a&- iBttl'hp
we
illustrate this
a 3 e- iEa t l' y! 3
i
-\
[fl*aiV*Vi + a*aaV>*V>2 + a? "aVJVs +
+ [(.alche-Wz-EiMy)*^ + a*a ae-^8-iWty*Y>, +
+ K^e-^i-^WfysVi + aJflse-^^s-^X/VJVs +
+ (a?a e- B i--B8)/'V* Vl + *a,e-* -*>/>?% +
+(
) +
) +
i <-
fl
by
MATHEMATICAL PROPERTIES
Sec. 6]
We
187
see that in general the probability density
depends on time. Thus the
measurement made
relative probabilities of the various possible results of a
to locate the particle depend
on the time
at
which the measurement
is
made.
Next consider the special case of a particle whose associated wave
is one of the wave functions Y(a;, i) corresponding to a single
(d)
function
eigenvalue
En
Then
V*n(x, t)Y %{z,
= JW'e-WriLx) Vn(x) =
y>* (x) y> (x)
n
n
(7-47)
and we see that the probability density is independent of time. Even
though the wave function is time dependent, the relative probabilities of
the possible results of a measurement of the location of the particle do not
depend on the time at which the measurement is made. In such a case the
particle is said to
be in a stationary
state,
or eigenstate, of the potential
K(*).t
According to equation (7-30), the probability interpretation of
t) demands that any wave function *F(a;, i) must be normalized
(e)
x
*(x, t)W(x,
so that
it satisfies
the relation
E
If
we
*(*,
t)dx=l
(7-48)
consider the general form of the wave function
Y(*.
it is
HT{x,
= t a nW n (x,
n=l
t)
apparent that normalization of the wave function can be achieved,
any given instant of time, by a proper choice of the arbitrary
at least at
constants a n .% Of course, equation (7-48) must be satisfied at all times.
Later we shall derive a time independent relation between the constants
a n which, when
satisfied, will effect the
normalization of
*F(a:, i)
at all
times.
(f)
Consider the special case of a particle whose wave function ^(x, t)
i).
Then we have
simply one of the eigenstate wave functions ^Jx,
from equation (7-47)
is
W*(x,t)V(x,t)=y>*n(x) Wn(x)
t In section 12, Chapter 1 3, it will be seen that an atom does not emit electromagnetic
radiation when the atomic electrons are in eigenstates because its charge distribution,
which is proportional to yif * xF, is time independent.
J There are certain difficulties connected with normalizing the wave functions for an
unbound particle, for instance the free particle wave function ^(x, t) = ex( Kx ~'t \ However, these difficulties are not fundamental; they will be discussed in the first section of
-
the next chapter.
SCHROEDINGER'S THEORY OF
188
QUANTUM MECHANICS
[Ch. 7
In this case equation (7-48) requires that
V*(*)
V0) dx
(7-49)
Thus the eigenfunctions must also be normalized. This can always be
accomplished as follows. Assume that we have obtained an eigenfunction
V( x) by solving the time independent Schroedinger equation. We evaluate
the integral
r
Then,
if
we
construct a
"
J
oo
a;)
V'*(
(g)
The
V>J.X) dx
set
y>'
new
A,
a constant
function y>Jx)
and that
\,
dx
n (x)
y>'*(x)
y> n {x) is
A~^tp^(x),
we
see that
a normalized eigenfunction.
of eigenfunctions of the time independent Schroedinger
equation for a particular potential V(x) have an interesting and very useful
property expressed by the equation
xp*{x)
/:
The
integral, over all x,
dx
y) n (x)
l^n
0,
(7-50)
of the product of one eigenfunction of the set
times the complex conjugate of a different eigenfunction of the set always
vanishes. Because of this property eigenfunctions are said to be orthogonal.
The proof of equation
eigenfunctions yjz) and
is not difficult.
Consider two different
which are, respectively, solutions to the time
(7-50)
y> t (x)
independent Schroedinger equations
h*
A". + VWn = E
nWn
2m dx2
(7-51)
and
+ ^.=^.
~f
2m 7T
(7-51')
dx'
Taking the complex conjugate of the second equation, we obtain the
equation for ff(x),
fc2
J2
+ Wt = E ^*
f
2m TT
dx*
since everything in that equation
Then we multiply (7-51) by
y>*
is
< 7 -52)
real except possibly the eigenfunction.
and (7-52) by
ip
and
subtract. This gives
MATHEMATICAL PROPERTIES
Sec. 6]
Next we
integrate with respect to
2m
-E
( E,
189
oo to + oo, and find
x from
jjy r * _
.
vt
&. - ,. M) dI
which reduces to
Integrating the right side gives
- * jf%f
(*,
En
If the eigenvalues
and y
and
V> n
dx
=
[
Vf
^-
M] "_
(7-53)
are discrete, thus corresponding to eigen-
a bound particle, then we have seen
not only remain finite for very
large and very small values of x, but actually go to zero at these limits.
In this case the right side of equation (7-53) vanishes at both limits and,
since (E E)
0, we obtain the desired proof. For eigenvalues in the
continuum, the corresponding eigenfunctions remain finite for very large or
functions
fn(x)
(x) representing
t
in the previous section that tpjx)
and
y>j(#)
very small values of x (or both).
way
things in such a
Sometimes
happens that
it
However,
possible to arrange
it is still
that the right side of equation (7-53) vanishes.f
En E
for
^ n.
That
the eigenvalues
is,
corresponding to two different eigenfunctions are numerically equal. In
this case the two different eigenstates of the particle are said to be
degenerate. (Recall the degeneracy discussed in the Sommerfeld treatment
of the hydrogen atom.)
It is,
eigenfunctions can always be
t This
unbound
is
however, easy to show that degenerate
made
orthogonal. %
associated with the question of normalizing the
particle;
it is
discussed in the
wave functions for an
section of the next chapter.
first
% Consider y> and y> t which are solutions to the time independent Schroedinger
equation for the potential V(x) and for the same eigenvalue E = E t = E. Then the
linear combination
V>
is
also a solution.
To
y>,
We have
+ " lV " + K ^"Vn + "W^ ~ E(a<W + <Wi) =
a "Vn
check, substitute into the equation.
~2~dx* ^
But
this is
h* dhpn
X^n^
^Incidentally this
**
makes
h* dhp,
\
" **) + a \~ 2^ I* +
it
Vy"
~ Em '
)
QED
obvious that a linear combination of eigenfunctions
is
"
not a
SCHROEDINGER'S THEORY OF
190
QUANTUM MECHANICS
The normality and orthogonality of the
(h)
particular potential V(x) can be
of eigenfunctions for a
single equation
set
summarized in the
V>*( x) V>ri x)
[Ch. 7
dx
1,
n
(7-54)
l=n
0,
These properties often allow a considerable simplification of quantum
mechanical calculations.
(i) We now derive the relation between the arbitrary constants a
n of
the general form for a wave function
= I a nW n(x,
Y(x,
which must be satisfied in order that the wave function
satisfies
the normali-
zation requirement
"*(*,
JI
t)
(x,t)dx
CO
Integrating both sides of equation (7-46) over
we have
all x,
*(*, t)W(x, t)dx
JJ
oo
= X a*n a n f
v
n
Wn{x) dx
y>* (x)
oo
"
afaj-***-'* f v *(x)
11
J
In
y> n (x)
dx
oo
solution to the time independent Schroedinger equation in the normal case in which
Ei.)
Now choose
JftVidx
-co
Oi
y>*yi n
dx
J CO
Then
ipa is
orthogonal to
yt n
since
/*O0
/*0O
W*Wa dx
an
/*0O
y>*y>
dx
a,\
J 00
J CO
y*y>
dx
J CO
VnVl
a
Of
J 00
By applying the condition
pletely determined.
The
J_ 0O
VJV dx
J ~ oo
that y be normalized, the values of a
and a can be comy> n and y> are
pair of orthogonal degenerate eigenfunctions
used instead of the original pair,
y>
and
y> t .
MATHEMATICAL PROPERTIES
Sec. 6]
Since the double
sum
191
contains only terms for which
^ n, we may apply
n) to each integral appearing in that sum. We may
equation (7-54) (/
also apply equation (7-54) (/
n) to each integral in the single sum. Then
we
obtain
J:
V*{x,t)W{x,i)dx=^a*na n
n=\
Thus normalization of the wave function
I a*na n =
achieved
is
we have
if
(7-55)
n=l
Since this condition does not involve time,
we
always remain normalized.
(j) Consider a general wave function Y(a;,
see that the
wave function
will
t)
which
Schroedinger equation for the potential V(x). If we
wave function
for
is
a solution to the
the form of this
know
any particular time, say / = 0, then it is easy (at least
wave function at any other time t. At
formally) to evaluate explicitly the
the time
0,
we
write the function Y(a;,
as a
t)
sum of
wave functions ^Jx, /) of the Schroedinger equation
V(x). That is, we take equation (7-44), as evaluated
instant
0.
the eigenstate
for the potential
in (7-45), at the
This gives'
= fa nfn(x)
V(x,0)
(7-56)
n=l
(This can be thought of as a generalized type of Fourier analysis.) Multi-
plying both sides by the complex conjugate of one of the eigenfunctions,
and integrating over
f"
J qo
From
all x,
we
V ?(x)Y(x,
equation (7-54)
0)
obtain
dx
= f an f"
n=l
**
we
will vanish, except for the
v *(x)
tp n {x)
dx
oo
term of the sum the integral
which the integral equals unity.
see that in every
term n
in
Thus we have
J
This
tells
oo
us the value of the constant a v The constant a n can be evaluated
if we replace / by n throughout. Furthermore, the
from the same equation
value of a definite integral does not depend on what
integration. Let us then replace x by *', and write
an
J oo
W n(x')W(x',0)dx'
we
call
the variable of
192
To
SCHROEDINGER'S THEORY OF
wave function T(,
evaluate the
these values of
a,n
=2
K=
Y(x,
7.
The
Classical
In this section
t)
in equation (7-45)
f"
l
V-oo
QUANTUM MECHANICS
for an arbitrary value of t, we
and obtain
rp*
n {x')W{x',
-iBnt/H,
0)dx'
Theory of Transverse Waves
we
[Ch. 7
Wn (x)
in
insert
(7-57)
a Stretched String
shall develop the classical theory
of transverse waves
At each step we shall compare and contrast the
wave theory with the quantum mechanical wave theory that we
in a stretched string.
classical
Figure 7-6.
A small
element of a string at an instant when a transverse wave
have developed in the previous sections.
As we
is
shall see, the
matical properties of these two theories are often very similar.
we
passing.
mathe-
Con-
be able to illustrate the significance of much of the
quantum mechanical wave theory in terms of the familiar phenomena
sequently
shall
described by the classical wave theory.
Consider a string which is under tension. If the string is given an up
and down displacement at one end, the displacement will travel down the
string in the manner characteristic of a moving wave. The propagation of
this wave can be completely described by the wave function *{x, t), the
value of which is numerically equal to the transverse displacement at time
t of a point on the string at longitudinal coordinate x. We note that in the
classical theory the wave function is a real, measurable function with an
immediate physical interpretation.
The quantum mechanical wave
function is a complex function and is, therefore, not measurable. Furthermore, a physical interpretation exists only for the product of the quantum
mechanical wave function times its complex conjugate.
To discuss the propagation of either the classical or the quantum
mechanical waves, we must know the differential equation which the wave
function obeys. The quantum mechanical wave equation cannot be
TRANSVERSE WAVES
Sec. 7]
A STRETCHED STRING
IN
derived from classical physics and
The
classical
193
by postulate.
from Newton's laws
obtained, essentially,
is
wave equation can be derived
directly
as follows.
Consider a small element of the string located between the coordinates
+ dx. In figure (7-6) we show that element at an instant t when
the displacement is passing the point x. The figure can also be thought of
x and x
as part of a plot of the function Y(x, t) versus x, at the instant t. Now let
us impose the restriction
1
Our classical wave equation will apply
<
<
only to waves on the string of small amplitude. If
sin
cos
0=1
tan
length of element of string
the tension
The
and the net
force in the
Tsin
Now
t)
(0
tan
(0
dd)
Tcos
y direction
dd)
Tsin
9Y(a;, t)jdx,
with respect to
x,
is
is
then
= T- T=
is
which
T(6
is
dd)
T6
Tdd
the derivative of the function
evaluated at the point x and the time
dx
which
dx
a constant
net force on the element in the x direction
Tcos
Y(x,
T is
1,
t.
Thus
dx
the change in the derivative, per unit change in x, times the change
in x. Therefore
d
TdO=T ^^dx
dx 2
According to Newton's laws, this net force in the y direction must equal the
mass of the element of string times its acceleration in the y direction. The
mass is p dx, where p is the mass per unit length of the string. Since there
is no net x force, the element moves only in the y direction and its acceleration is just <PF(x, t)jdt 2 Consequently we have
.
dx2
or
~dx^~ ~ T
H
dt
~~
(?_58)
SCHROEDINGER'S THEORY OF
194
This
is
the classical
wave equation.
(of small amplitude) in a string.
QUANTUM MECHANICS
It
[Ch. 7
governs the propagation of waves
same equation governs
Essentially the
the propagation of electromagnetic waves (of any amplitude) through
space.
The
Y(a;,
to
wave equation relates the second space derivative of
second
time
derivative. All the constants in the equation are
t)
The quantum mechanical wave equation
real.
classical
its
2m
relates the
oar
ot
second space derivative of T(*,
t)
to
its first
time derivative,
The constants in the quantum mechanical wave
equation are partly real and partly imaginary. Nevertheless, the procedures
involved in solving these two partial differential equations are very similar.
Let us consider an example of the solution of the classical wave equation.
A string of length a is fixed at both ends and under tension T. The
string is given a small but arbitrary initial displacement, and we are
and
to the function
itself.
its subsequent motion. We define an x axis extending
along the length of the string with an origin at its center. Then we must
required to evaluate
solve the
wave equation,
T(z,
which
state that the
subject to the conditions
for
0,
a/2, for
(7-59)
all t
ends of the string are fixed. Similar conditions obtain
quantum mechanical wave function when it describes the motion
of a particle which is bound to a certain region of space. Such quantum
mechanical wave functions always go to zero at both large and small x.
for the
It is
obviously appropriate in this problem to look for solutions of the
form
W(x,
t)
y(z) 4(f)
we
Substituting this into equation (7-58),
d>(^)
^
Dividing both sides by
1
y(z)
Since x and
ip(x)
dx2
<j>(t),
rfXaQ
dx2
obtain
dW
dt
we have
p
T<f>(t)
rfV(t)
dt
= -K 2
must be
Thus we have
are independent variables, both sides of this equation
equal to the same separation constant, which
we
call
K2
the two second order ordinary differential equations
^- + *>(*) =
(7-60)
Sec. 7]
TRANSVERSE WAVES
A STRETCHED
IN
STRING
195
and
2 + tfItfO-0
at
(7-61)
Consider equation (7-60). It is identical in form with the time independent Schroedinger equation for the case V(x) O.t The general
solution of the equation is
ip(x)
where
A and B are
= A sin Kx + B cos Kx
arbitrary constants.
dip(z)
To prove
(7-62)
this,
evaluate
= AK cos Kx BK sin Kx
dx
and
dPipix)
= -AK* sin Kx - BK2 cos Kx =
->(:*)
dxr
and
+ Khj>(x) =
-Khp(x)
To
This gives
substitute into equation (7-60).
Q.E.D.
Q-,
determine the allowed values of the separation constant,
we apply
the
conditions (7-59), which read
T(a/2,
/)
y(a/2)
<Hf)
0,
for all
So
V(a/2)
(7-63)
Then we have
.
A sin
Ka
B cos
Ka =
and
A sm Ka
.
B cos
Substracting and adding,
we
Ka
2
obtain
=
Ka
Bcos =
A sm
Both of these equations must be
satisfied.
t The analogy is even closer than this. If we consider a string of mass per unit length
p(x) which is a function of x, we shall obtain an equation identical in form with the time
independent Schroedinger equation for a general potential V(x).
196
SCHROEDINGER'S THEORY OF
QUANTUM MECHANICS
[Ch. 7
no value of K for which both sin (Ka/2) and cos (Ka/2)
However, we can satisfy these equations either
by choosing K such that cos (Ka/2) vanishes and setting A equal to zero,
or by choosing K such that sin (Ka/2) vanishes and setting B equal to zero.
That is, we take
Now
there
is
vanish simultaneously.
0; cos
0,
which implies
with n
1, 3, 5,
or
B=
0; sin
(7-64)
0,
which implies
-n
Figure 7-7.
The
Thus there are two
first
For the second
=3
For the first
= B cos YITTX
1, 3, 5,
class,
a
(7-65)
class,
.
2, 4, 6,
three eigenfunctions for a string fixed at both ends.
classes of solutions.
y> n (x)
with n
xp n (x)
= A sin
.
JITTX
2, 4, 6,
These are the eigenfunctions for the string of length a fixed at botn ends.
In figure (7-7) we plot the first three eigenfunctions. Note the essential
similarity of these functions to the three quantum mechanical eigenfunctions for a bound particle plotted in figure (7-4). Also note the difference in their physical significance. In the classical theory, y>Jx) gives the
spatial dependence of the amplitude of the vibration. In the quantum
mechanical theory, y>*(x) ip n (x) gives the spatial dependence of the probable
location of the associated particle.
Next consider equation (7-61). By comparison with equations (7-60)
and (7-62), it is apparent that the general solution to this equation is
rf(0
sin
(KVTfr
t)
p cos (kVt/^ t)
(7-66)
..
TRANSVERSE WAVES
Sec. 7]
where a and
time,
/?
we may
A STRETCHED
IN
are arbitrary constants.
set
0,
By proper
STRING
choice of the zero of
and simplify equation (7-66)
<f>(t)
fi
cos
197
to
(kVTTp t)
(7-66')
no complex constants, it has a real solution
The corresponding equation in the quantum mechanical case
contains the imaginary number i, and its solution (7-34) is
Since equation (7-61) involves
(7-66').
(7-33')
necessarily complex.
Combining equations (7-59), (7-64), (7-65), and (7-66'), we find the
wave functions for a string of length a, density per unit length p, and tension
T, which is fixed at both ends:
Y(a>,
a n cos (27rr0
(7-67)
y, n (x)
where
mrx
1, 3, 5,
2, 4, 6,
= K [r =
cos
mtx
sin
and where
"n
The a n
2n
are arbitrary constants,
2a
/I
and the r are the vibration frequencies
This discrete set of frequencies constitutes the familiar
"fundamental," "second harmonic," "third harmonic," etc., produced by
any stringed musical instrument. In the classical theory the frequency is
of the string.
quantized. In the
is
quantum theory the energy
related to the frequency by the equation
is
E=
quantized, but the energy
hv.
An interesting difference between the two theories
is
that in the classical
theory the wave functions are not required to satisfy the normalization
condition (7-48); the integral P
T*T dx can assume any value because
the value of the constants a n which determine the amplitude of the vibrations, can assume any value that is not too large to violate the restriction
,
<
1.
However, the eigenfunctions of the
string are
orthogonal. The orthogonality
a/2
/:
This
is
Vi(*) ?(*)
dx
classical theory
of a vibrating
relation reads
= 0,
l^n
(7-68)
a/2
completely equivalent to equation (7-50) because, in the present
SCHROEDINGER'S THEORY OF
198
case, the range of
y> t (x).
is
QUANTUM MECHANICS
only from a/2 to +a/2, and because
The reader should
verify equation (7-68)
some of the integrals.
The general form of the
classical
W(x,
t)
=J
n=
tf
^(x)
by actually evaluating
wave function
combination of the functions (7-67). That
[Ch. 7
is
an arbitrary linear
y n(x)
(7-69)
is,
cos (IttvJ)
Normally, the form of the x dependence of the function W(x, i), at some
t, depends on the value of /.
This is analogous to the
situation described by equation (7-46). However, if we set the string into
particular time
^-* *,o) = (f-H)^
(
-/2
o/2
Figure 7-8.
string plucked at
center.
its
by giving each segment an initial displacement which is in the
form of one of the eigenfunctions, the form of the x dependence of Y(#, t)
will be constant in time. For instance, let us give the string a displacement
at t =
of the form Y(a;, 0) = C cos (ttx/o). Then, in equation (7-69),
we have
vibration
Ccos
In this case
the
all
an
wave function
=
at
0,
a = n=l
%a
except a x
n ip n (x)
C. Since the a n are time independent,
is given by
any instant of time
Y(x,
= C cos
t)
/.
\2tt
\
nx
IT
t) cos
la's
\
is analogous to a quantum mechanical eigenstate.
Consider a string which, at the instant t = 0, is given a displacement
of a form different from that of an eigenfunction; say we pluck the string
at its center, so that T(x, 0) = (a/2 \x\) d/(a/2). This is shown in
figure (7-8). To determine the subsequent motion of the string, we use
its initial displacement to evaluate the coefficients a
n At / = 0, equation
This situation
(7-69) reads
Y(*,0)
= f ?(*)
(7-70)
..
TRANSVERSE WAVES
Sec. 7]
To
A STRETCHED
IN
evaluate a particular constant, say a
equation by
ipi(x)
and
This gives
|*o/2
oo
ViO) ^O* 0)
199
multiply both sides of this
integrate over the range of x.
fa/2
STRING
= 2 a
V>l x) VnO) dx
J -a/2
n=l
da;
J-a/Z
Because of the orthogonality condition (7-68), every term of the summation
I.
We consequently have
vanishes except the term for n
f"% (ar)'F(a:,0)te
I
2
l> (*)] dx
f
/-a/2
which
is
a/2
f
a,
mWj
J
\2
J-/i
'
7 a/2
/a/a
a/2
J -a/2
Evaluating the eigenfunctions from equation (7-67), and performing the
integrations,
we
find
MlifirP,
0,
where we have replaced
by
n.
= 1, 3, 5,
= 2, 4, 6,
The
coefficients
of fjx) in the
(7-70) are non-zero only for odd n because of the symmetry of Y(x,
series
0).
The
series is
Y(*,0)=
-^sv
n = l,3,5,...7r'n
From
equation (7-67) this
is
Y(, 0)
2
=i,3,5,...
cos
Tl
7rn*
The coefficients of the cosine functions decrease rapidly with increasing
Thus a good approximation to
n, and the series rapidly converges.
W(x, 0) can be obtained by taking only the first few terms of the series.
This is illustrated in figure (7-9) by plotting each of the first three terms of
the series, their sum, and also the function
T(*,o)=(fto which the series converges.
W )A
/2
SCHROEDINGER'S THEORY OF
200
What we have done
of a
series
and
cosines, this
is
QUANTUM MECHANICS
[Ch. 7
to write the initial displacement Y(x, 0) in terms
of the eigenfunctions
amounts
to
y> n (x).
[As the fjx) happen to be sines
analysis' of Yfo 0).] This
making a Fourier
allows us to evaluate the displacement of the string at any instant of time
by simply inserting, into the series for the general form of the wave function
(7-69), the values of
^0*.
a n which we have determined. That
0=2
cos 2^0 c s
-n
n
=i,3,5,...77
is,
?22 +M-^
+ -mcos
2 2
2
ir
Figure 7-9.
Illustrating
^5
"
the Fourier anal/sis of a string plucked at
its
cos See
center.
A good approximation to T(a;, t) can be obtained from the first few terms
of the
The physical
series.
significance of this series can be interpreted as
we simultaneously set into vibration a number of
different eigenstates. As time progresses, these eigenstates vibrate independently, each at its own characteristic frequency v. Then at any
instant the net displacement at x can be obtained by summing the
follows.
At
0,
instantaneous displacements of the eigenstates at that value of x.
Using the general expression (7-71) for the a n we can write a general
,
solution to the problem of a string with fixed ends that
result
of an arbitrary given
initial
displacement
is
*F(a;, 0).
vibrating as a
The
solution
is
fa/2
Y(*.
0=2
^%
y, n
(x')W(x',0)dx'
cos (2nv n t)
y> n (x)
J-aji
This equation
is
completely analogous to equation (7-57), which gives the
quantum mechanical wave function for a particle moving in a
potential in terms of a given initial form of the wave function.
certain
8.
AND
EXPECTATION VALUES
Sec. 8]
DIFFERENTIAL OPERATORS
201
Expectation Values and Differential Operators
we compared some of
In the previous section
perties
the mathematical pro-
of the classical wave theory with certain similar properties which we
have developed for the quantum mechanical wave theory. We return now
to the task of continuing the development of the latter theory.
Consider a particle and its associated wave function Y(cb, t). In a
measurement of the position of the particle, there would be a finite
probability of finding it at any x coordinate in the interval x to x + dx,
as long as the
wave function
wave function
is
is non-zero in that interval. In general, the
non-zero over a certain range of the x axis. Thus we are
generally not able to state that the x coordinate of the particle has a
certain definite value. However, it is possible to specify some sort of an
average position of the particle in the following way. Let us imagine
making a measurement of the position of the particle at the instant t.
The
probability of finding
between x and x
it
dx
is,
according to the
postulate (7-21),
P(x,
t)
dx
= Y*(x, t)Y(x, i) dx
(7-72)
this measurement a number of times, always at the
same value of /, and recording the observed values of a;.f We can then use
Imagine repeating
the average of the observed values to characterize the position of the particle
at the instant
t.
This
we
call x, the expectation value
of the particle at the instant
It is
t.
-r xP(x,
J
of the x coordinate
evident that x will be given by
t)
dx
is just the value of the x coordinate weighted by the probaof observing that value and, upon integrating, we obtain the
average of the observed values. From equation (7-72) this can be written!
The integrand
bility
Y*(x,
v
xY(x,
t)
dx
(7-73)
(.
easy to see that an expression of the same form as (7-73) would be
appropriate for the evaluation of the expectation value of any function of
It is
x.
That
is,
~~
<
We
Y*(x,
x ix(x,
t)
dx
can visualize doing this either by making a number of measurements on identical
t
particles in identical environments, each particle being in the state associated with the
t), or by making a number of repeated measurements on the same
always with the particle in the state associated with the wave function ^(x, t).
t The terms of the integrand are written in this order to preserve symmetry with a
notation which will be used later.
wave function 'Vix,
particle,
SCHROEDINGER'S THEORY OF
202
QUANTUM MECHANICS
[Ch. 7
and
X
f(xj=! W*(x,t)f(x)Y(x,t)dx
J oo
where f(x) is any function of x. Even for a function which
depends on the time, such as the potential energy V(x, t), we
explicitly
may
still
write
P(iT0
because
all
value of
t,
V(x,
T*(*.
W(x,
t)
t)
dx
(7-74)
CO
measurements made <o evaluate V(x, i) are made at the same
and so the arguments of the previous paragraph would still
hold.
The coordinate x and the potential energy V(x, i) are two examples of
the dynamical quantities which can be used to characterize the state of a
Examples of other dynamical quantities are the momentum p
and the total energy E. The expectation value of these quantities is always
given by the same type of expression. For instance, for the momentum
particle.
p the
expression
is
W*(x,
pY{x,
dx
(7-75)
J 00
In order to evaluate the integral in equation (7-75), the integrand
Y*(x, t)pY(x, i) must be expressed as a function of the variables x and
In a problem in classical mechanics, p can always be written as a
/.
function of the variables x and/or t after the problem has been solved. For
instance, for a particle moving in a time independent potential, p can be
written as a function of a; since, after the classical problem has been solved,
the momentum is precisely known at every point of the trajectory. But
quantum mechanics this cannot
be done because p and x cannot be simultaneously known with complete
precision. We must find some other way of expressing the integrand in
terms of x and t.
the uncertainty principle shows us that in
clue can be
(7-24),
which
found by considering the
W(x,
Differentiate with respect to x.
^j*'
t)
iiKx -'ot)
iKe*'-~t
>
K = pjh,
ox
This gives
ox
Since
free particle
is
iK{x,
t)
wave function
AND
EXPECTATION VALUES
Sec. 8]
DIFFERENTIAL OPERATORS
203
which can be written
*[(*. 0]
= -tt!- [TOM)]
ox
an association between the dynamical quantity
ih djdx. A similar association between
the dynamical quantity E and the differential operator ih djdt can be
found by differentiating the free particle wave function with respect to t:
This indicates that there
p and
is
the differential operator
8T(a;
'
f)
-itue*^--"*)
-icoW(x,
V
t)
'
dt
Since
co
E/h, this can be written
at
It
may seem
particle
wave
of showing
that these associations apply only to the case of a free
function, but this
this;
actually not so.
is
There are several ways
perhaps the most convincing one
Consider the equation (7-3) defining the
film +
V(x,
classical total
t)
Let us replace the dynamical quantities
differential operators.
is
the following.
energy E:
=E
p and E by
their associated
Then we have
-L(-it\v(x,t) =
2m \
ih?-
dx!
dt
or
V ^^ =
-ff-2+
2m dx
ih
This is an operator equation.
To
(7
l
dt
obtain a differential equation,
"76 )
we multiply
the equation into the equality
TO,
= W(x,
and obtain
2m
This
is
dx2
dt
just the Schroedinger equation (7-19).
We
see that postulating the
associations
p-^-ih
dX
(7-77)
<> ih
dt
is
QUANTUM MECHANICS
SCHROEDINGER'S THEORY OF
204
equivalent to postulating the Schroedinger equation.
This
is
[Ch. 7
essentially
procedure originally followed by Schroedinger. The procedure
provides us with a powerful method for obtaining Schroedinger equations
the
cases than we have discussed in this chapter. We
Chapter 10 to obtain Schroedinger equations for three
dimensional problems, and for problems involving more than one
for
more complicated
shall use
in
it
particle.
Let us use (7-77) in the equation (7-75) for the expectation value of the
momentum. Then
r W*(x,
p*(x,
t)
dx
00
becomes
Y*(x, t)(-ih
J - oo
dx!
Y(x,
t)
dx
or
p=
We
-ih\ Y*(a;,
J-x
J^hJl dx
t)
(7-78)
dx
thus obtain an expression which can be immediately evaluated
know W(x,
f).f
we can
Similarly,
if
we
evaluate the expectation value of the
energy as follows
E=\
E=
T*(a;, i)[ih
*{x,i)E x{x,f)dx
E=ih)
-)
^fo
Y*(x,
J -oo
From
W(x,
f)
t)
dx
dx
(7-79)
dt
equation (7-76), this can also be written
f " **{x,
./-oo
i)(-
%- +
^
2m
ox'
V(x, o) T(z,
dx
t We see now the reason for the ordering of the terms in the integrand of p.
not be possible to have
-ih
T*^,
f *F*(*, /)(,
|
J 00
since this
is
meaningless.
-'*
I
I
J 00
Nor
for would
4ax
it
to have
be possible
pos
^ *(*, /)Y(iB, t)dx=
because this always vanishes.
yx dx
-ih [(*,
O^x, t)f
It
would
EXPECTATION VALUES
Sec. 8]
In general,
that
is,
AND
DIFFERENTIAL OPERATORS
205
we can obtain the expectation value of any dynamical quantity
a quantity which
a function ofx,p, and possibly
is
by evaluating
the integral
fix, P.
J"
^*(*. 0/op (*,
- it
t)
dx
(*,
(7- 80)
where the operator fQV (x, ih 3/ dx, i) is obtained from the function
t) specifying the dynamical quantity, by everywhere replacing p
by ih d/dz.f
We see that the wave function W(x, t) contains more information than
f(x,p,
just the probability density
P{x,
and the probability
i)=Y*(x, t)Y(x,t)
flux
2mL
The wave function
dx
ox
also contains, through equation (7-80), the expectation
values of the coordinate x, the potential energy V(x,
t),
the
momentum p,
the total energy E, and, in general, of any dynamical quantity f(x, p,
In fact, the wave function Y(a;,
t)
contains
all
t).
the information which, in
we can ever expect to learn
about the particle associated with that wave function.
Let us calculate the expectation value E of the total energy of a particle
consideration of the uncertainty principle,
whose associated wave function
Y(x,
is
of the general form
= | a n e- iE^ K Wn (x)
= | a nW n {x,
(7-81)
n=l
n-l
According to equation (7-79),
J -oo
E=\
dt
fateiE t/n y>?(x)ih
>
dx
2 a n (- ^Ae^^f^x)
n
=i
J-ooi=i
E = 2 2 Erfaj*-* Jf"
n li =
l
From
y> n (x)
dx
oo
equation (7-54)
t Cf. exercise 8 at
tf(x)
we
end of
see that the integral vanishes except
this chapter.
when
SCHROEDINGER'S THEORY OF
206
I
n, for
QUANTUM MECHANICS
which the integral equals
contributes only the single term
n,
[Ch. 7
So the summation over
and we have
unity.
n=l
or
E = I EaU n
(7-82)
is quantized and can only be equal to
one of the eigenvalues En any single measurement of this quantity can
only yield one of these values. However, as the wave function (7-81) is
a sum of a number of different wave functions ^(a;, t), each corresponding
to a different eigenvalue, any one of these eigenvalues could be found in
the measurement. In light of this fact, it is apparent that we should
interpret equation (7-82) as meaning that the quantity a*a n is equal to the
probability that a single measurement of the total energy will yield the value
En ; i.e., it is equal to the probability that the particle would be found in the
Since the total energy of the particle
,
eigenstate n.
Then
that equation will just give the average of the values
found in a large number of measurements of the energy of a particle in a
state described by the wave function (7-81), which is the definition of the
expectation value of the total energy. In general, E will not happen to
be equal to one of the eigenvalues. However, if the particle is known to be
in single eigenstate, sayY (a;, /), then in equation (7-81) a n = 1 if n = k,
otherwise. In this case equation (7-82) gives E = Ek as
and an =
would be expected.
fc
9.
The
Classical Limit of
Quantum Mechanics
Consider a particle moving under the influence of a potential V(x,
In the limit in which the characteristic distances and
momenta
t).
involved in
describing the motion of the particle are so large that the uncertainty
be ignored, we shall prove that the predictions of quantum
mechanics coincide with Newton's laws:f
principle can
dp
__
dVixJ)
(7g3)
dt
where
dx
= m
dt
The
first
equation
is
simply dpjdt
= F, where F
is
the force. Cf. equation (7-37).
THE CLASSICAL
Sec. 9]
LIMIT
OF QUANTUM MECHANICS
207
Such a proof was first given by Ehrenfest in 1927. It is considerably more
involved than showing that the laws of relativistic mechanics approach the
laws of Newtonian mechanics in the limit in which the characteristic
velocities are small
compared to the
velocity of light.
obviously of the greatest importance to be sure that
However,
it
is
quantum mechanics
agrees with classical mechanics in the appropriate limit.
In Chapter 6 we saw that, for large characteristic distances and momenta,
possible to associate with the particle a wave function W(x, t) in the
it is
form of a group involving uncertainties in these quantities which will
always be relatively small. Now for any wave function, and in particular
for ^(x,
i),
the expectation value of the coordinate x
is
T*xxF dx
J 00
we obtain
Evaluating the time derivative of this quantity,
m
^-\
Wis -[*
fa, + **!#?)**
J-,\
dtJ-oc
dt
dt
Substituting for
(7-26)
dt
and
dt
dY/dt and dW*jdt from the Schroedinger equations
(7-26') gives
HJ-S
The terms involving
2m
dx2
V cancel
out,
2m dx2
and we have
-*[ fa,** -**!&) i*
dx
dx
2m J- A
2
dt
Separate out the second term of the integrand, and integrate
it
by parts
as follows
dx =
xy ^.
2
dx
J-oo
Since
_ p w*& dx
[WMdx
dx
dx
L
J -oo
dv
J-oo
du
Y must be in the form of a group of limited spatial extent in order
that the uncertainty in the x coordinate be relatively small, both the
function and
grated term
x >
equal to zero, and we have
its
is
derivatives vanish as
Consequently the
oo.
xdvn
rx^dx=-(
-^dx
dx
dx
dx
d
J-oo
J-oo
wave
inte-
208
QUANTUM MECHANICS
SCHROEDINGER'S THEORY OF
by parts
Integrating
Again
this
again,
dx2
J-oo
[Ch. 7
L dx
J-oo
dx2
J-oo
reduces to
dx2
J-oo
dx2
J-oo
Putting this back into the equation for dxfdt,
2m
dt
Consider the bracket in the integrand.
d 2xF
dx
2
d (xW)y
dx
dx2 J
dx2
J-oo
can be written
It
dx2
dx
_ <PY_
dx
we have
dx
\
2*?
dx
3T
9Y
dx
dx
_2^
dx
Consequently
dx
_ih^^dX dx
m
dt
J-oo
3a;
According to equation (7-78), the expectation value of the
the particle
of
dx
gvp
Too
p=-ih\
W*
(7-84)
dx
J-oo
Thus we
momentum
is
see that
dx
dt
or
= m^-
(7-85)
at
Next
(7-84),
let
us evaluate the time derivative of p.
Differentiating equation
we have
dt
Again we
substitute for
J-A
dt
dx
dxdtl
dYjdt and dW*jdt from the Schroedinger equation.
THE CLASSICAL
Sec. 9]
OF QUANTUM MECHANICS
LIMIT
209
Then
2m
dt
J-<x>\
dx2
dx
J-ao\
This
dx9 /
dx
dx
is
2m
dt
Then we
term on the
first
= _Jf_
dt
2m
dx2
dx dx
'
W*
Ydx
dx
-|
/:
Integrating the
J- oo dx
dw* av
dx
right,
we have
2
_ TN! 3 T'
dx
dx
dV
r
vp* v
J-ao
xrf
dx
dx
see that this term vanishes because of the behavior of the
function as x -*
wave
Thus
oo.
&
T* (
J -oo
dt
\dxf
Y dx
(7-86)
dV/dx = dV(x, t)jdx is simply some function of x and t. Consequently, on comparing equation (7-86) with the general equation for the
expectation value of a dynamical quantity (7-80), we see that
Now
dp
= _ dvpt^
dx
dt
where dV(x,
t)jdx
is
the expectation value of the spatial derivative of the
potential energy.
Equations (7-85) and (7-87) are identical with Newton's laws (7-83),
In the limit in which the
except that they involve expectation values.
and momenta involved in describing the motion of the particle
compared to the uncertainties in these quantities, the
dynamical quantities all have precise values and it is no longer necessary
distances
are very large
dV/dx and dV/dx. Then the
become
completely identical with the laws of Newtonian mechanics. Quantum
mechanics does indeed agree with Newtonian mechanics in the limit in
which the uncertainty principle is not important. However, we shall see
to distinguish between x
equations
we have
x,
p and
when we
number of extremely
in the next chapter that,
predicts a
and
p, or
derived from the theory of quantum mechanics
quantum mechanics
phenomena which have no
are not at this limit,
interesting
counterpart in Newtonian mechanics.
QUANTUM MECHANICS
SCHROEDINGER'S THEORY OF
210
[Ch. 7
BIBLIOGRAPHY
Bohm, D., Quantum
Pauling, L.,
Book
Hill
Theory, Prentice-Hall, Englewood Cliffs, N.J., 1951.
and E. B. Wilson, Introduction to Quantum Mechanics, McGraw-
Co.,
New
York, 1935.
Quantum Mechanics, McGraw-Hill Book Co., New York, 1955.
Sherwin, C. W., Introduction to Quantum Mechanics, Henry Holt and Co.,
Schiff, L. I.,
New
York, 1959.
EXERCISES
1.
Show
quantity
2.
is
that the probability flux (equation 7-31)
real if
it
equals
its
own complex
always
real.
Hint:
Consider a particle of mass m in the state of lowest energy E1 of the potential
v
'
* < -all, x >
-all <x <a/l
Pi).
,,, v
0,
V , and a are such that E,
eigenfunction y^x) for this particle.
where m,
of
is
conjugate.
all
= 0.5 V Make a sketch of the form of the
Now assume that V is increased by a factor
.
What
effect does this have on the shape of y,(a:) in the regions where
,? What effect does this have on the shape of y x {x) in the region where
V(x)
,? What effect does this have on ,? These questions are to be
answered qualitatively by a direct consideration of the behavior of the time
3.
V(x)
>
<
independent Schroedinger equations for the potentials concerned.
3.
4,
Take a potential of the form indicated in exercise 2, except
and use the formulas for [v>(x)\
- Mx)~\Xj and
Xj+l
+
that
^mVtfP/U?
[dy(x)/dx]x
developed in section 5, to make a numerical integration of the time
independent Schroedinger equation and find the first eigenfunction y^x) and
\dy>(x)ldx\Xj ,
first eigenvalue E
v Compare these with the results of the analytical treatment
of this potential presented in section 4, Chapter 8. Hint: Start the integration
at x = 0, where it is apparent from symmetry that dyx)/dx = 0, and take
the
(xi+1
4.
Xj )
0.1a/2.
Use numerical
integration to find the
oo,
V(x)
<
a/1,
= 0,
-a/1
a/4
<x <
<x <
first
>
eigenvalue
Ex for the potential
a/2
-a/4,
a/4
<
<
a/2
a/4
where
v
Tfiffl/Sma 2
Employ the conditions y>j(x) = 0, x < -a/2, x > a/2, and yi^x) need not be
continuous at x = a/2, which are justified in section 5, Chapter 8, for a
potential of this type. One of the exercises of that chapter consists of an
analytical treatment of this potential.
Ch.
211
EXERCISES
7]
Derive the classical wave equation for a string of variable density per unit
Separate it into two ordinary differential equations, and compare
the equation in x with the time independent Schroedinger equation for the
5.
length p(x).
general potential V(x).
6. Verify
equation (7-68) by evaluating some of the integrals.
solve the problem of the motion of a string fixed at both ends,
which is plucked at a point at distance b from its center. Check the solution
by seeing that when b -* it reduces to the solution of the problem treated in
7. Set
up and
section 7.
8.
In calculating the expectation value of the product of position times
momentum, an ambiguity
arises
because
it is
not apparent which of the two
expressions
should be used.
Show
that neither of these
the obvious requirement that xp be real.
/*00
xp
**
{- m Tx)
is
acceptable because they violate
Then show
that the expression
+ {~ ih
Tx} Vdx
J 00
is
acceptable because
integrate
by
parts.
it
does satisfy this requirement.
Hint:
Feel free to
CHAPTER
Solutions
of Schroedinger's
Equation
I.
The Free
Particle
In this chapter we shall demonstrate some of the features of the quantum
mechanical theory and, in particular, some of the interesting phenomena
predicted by the theory,
specific
by solving Schroedinger's equation for several
t) and finding the eigenfunctions
forms of the potential energy V(x,
and eigenvalues.
Let us begin by considering the simplest case
V(x,
t)
For
constant.
a particle under the influence of such a potential the force F(x, i) =
dV(x, t)/dx will vanish; it is a free particle. Since this is true regardless
of the value of the constant, we do not lose generality by defining the
potential energy such that V(x, t) = 0. We know that in classical mechanics
a free particle may be either at rest or moving with constant momentum
p. In either case its total energy E is a constant.
To find the behavior predicted by quantum mechanics for a free particle,
we solve the Schroedinger equation for V(x, t) = 0. Since this potential
energy is not a function of t, the problem reduces to the problem of
solving the
time independent Schroedinger equation.
that equation reads
_*<*V*)_
2m dx2
The
.,
(8-1)
f( *)
solutions to the equation are the eigenfunctions
functions are
For V(x)
y>(x).
The wave
._,,
W(x,t)=e-* Et" w(x)
l
212
(8-2)
Sec.
THE FREE PARTICLE
I]
where
213
E is the total energy of the particle. From the discussion of section
we know
5 of the last chapter,
any value of E
(8-1) exists for
that an acceptable solution to equation
^ 0.
A solution to the differential equation (8-1)
tfx)
where
and
AeiKx
can be written
Be~ iKx
(8-3)
K=
and where
demonstrated by evaluating
are arbitrary constants,
validity of this solution is
V2mE/h. The
^> = i*K Ae iKx + K Be~ iKx = -K y>(z) = - ^- f(z)
2
dx2
and
ti>
substituting into equation (8-1)
-(-^)^)s ^
);
QED
Since there are two arbitrary constants, equation (8-3)
is
the general form
of the solution to the second order ordinary differential equation (8-1).
Thus the wave function
for a free particle
T(z,
may be
e- iEtin f(x)
AeiKx
written
(S-A)
where
y>(x)
Be~ iKx
(8-4')
and
K=
^2mE\h
We consider first the case in which one
to zero.
Then
the
wave function
W(x,
of the constants, say B,
is
equal
is
Ae~ imn e iKx
Aei(Kx
mm
(8-5)
which can also be written
Yix,
t)
= A cos (Kx -
Et/h)
iA sin {Kx
Etjh)
Its
is a traveling wave, oscillating at angular frequency co = Ejh.
nodes move to the right with velocity w = vjk = wjK = EjhK. Let us
This
evaluate
its
probability density P(x,
t)
and
its
probability flux S(x,
t).
These are
P(x,
A*
(8-6)
SOLUTIONS OF SCHROEDINGER'S EQUATION
214
[Ch. 8
and
2m\
=-
S(x,
S(x,
t)
= -
2m
dx
dx
[A*e- iiKx m/ AiKe Kx Etm
~ Et/
AeiiKx
A*( iK)e~ i( Kx
-
~ miK} ~\
A*A2iK = A*
2m
m
(8-7)
According to the de Broglie postulate (equation 7-5),
HK = p
where
v is the velocity
of the
= mv
Thus the probability
particle.
S(x,
vA*A
vP(x,
flux is
(8-8)
t)
which is the same as the familiar classical relation between flux, velocity,
and density. For the wave function (8-5), both the probability density
and the probability flux are constant in time and independent of position.
Furthermore, the wave function involves only the single
= HK and the
function
is
momentum
Consequently we realize that this wave
associated with a particle which is traveling along the x axis
with perfectly
single energy E.
known and
momentum p, but with completely
the physical situation of a particle of
constant
unknown x
coordinate.
momentum
that
an
long beam of particles. For the wave function obtained
infinitely
the constant
is
is set
This
known
to be exactly p,
equal to zero, which
Y(x,
we
is
t)
and moving somewhere
in
when
is
Be- iiKx+Et/n)
(8-9)
find
P(x,
= B*B
(8-10)
-
B*B
m
(8-11)
t)
and
htr
S(x,
The
t)=
sign of the probability flux S(x,
motion of the
particle.
vector quantity.)
moving
in a
beam
The
indicates, of course, the direction of
in the direction
is
is
wave function (8-5) is
of increasing x, and the particle associmoving in a beam in the direction of
particle associated with
ated with wave function (8-9)
decreasing x.
t)
(In three dimensions the probability flux
Sec.
THE FREE PARTICLE
I]
215
Let us evaluate the integral of the probability density for the wave
function (8-5). This
is
foo
/"oo
foo
V* x'dx=\
A* A dx
J 00
A* A
00
We see that it diverges unless A =
the wave function of an
unbound
0.
This
is
particle,
dx
00
the difficulty of normalizing
which we have already come
across in the previous chapter. But the difficulty
is
not really fundamental.
This wave function represents the highly idealized situation of a particle
moving in a beam of infinite length and whose x coordinate is, consequently,
completely unknown.
instead
case;
it is
But
known
in
any
real physical situation this
that the particle
would
is
not the
certainly not be
found
some region of the* axis of finite length. In an experiment the beam
A realistic wave function describing
is actually always of finite length.
the actual situation would be of essentially constant amplitude over the
length of the beam but would vanish at very large and very small values of
a;.
It would be in the form of a group of length A* which is long, but
Furthermore, we know from the uncertainty principle that the
finite.
realistic wave function must involve not only the single momentum
outside
p = hK
but also a distribution of momenta spread over the range
h/ Ax which is centered about the momentum p = hK.
h A.K
The wave function ^8 5) can be thought of as representing the behavior of
the realistic wave function in the limit Aa; -> oo. There is never any diffiA/?
wave function as long as Ax is finite. In the limit
the
sense that the wave function can be normalized
does
arise
in
a
only if the multiplicative constant A vanishes. Even so, the limiting form of
the wave function can be used to calculate quantities which do not depend
culty in normalizing the
difficulty
on the value of the multiplicative constant. In subsequent sections we shall
perform a number of calculations using wave functions of the limiting
form but they will always be used in such a way that a physical significance
is attached only to the ratio of the multiplicative constant of one wave
function to the multiplicative constant of another wave function. This
;
procedure bypasses the normalization
In certain situations
it is
difficulty.
necessary to use a wave function in the form
of a group of finite length. Such a wave function can be constructed by
adding together a number of wave functions of the form (8-5) if the group
is moving in the direction of increasing x, or of the form (8-9) otherwise,
and E. That is, we take
each with a different value of
W(x,
t)
- f A n Wn (x,
n=l
t)
where
Yn(x,t) = e- iE
"t
" eiK"x
t
(8-12)
SOLUTIONS OF SCHROEDINGER'S EQUATION
216
[Ch. 8
and where
Kn =
Tn
Since each of the
(a;, t)
y/2mEJh
a solution to the Schroedinger equation,
is
equation (7-44) shows that (8-12) will also be a solution. The values of the
constants A n can be calculated as in equation (7-57) in terms of the form
of the wave function at some particular instant of time, say / = 0. As an
example, consider a free particle moving in the potential V{x) 0, in
the direction of increasing x, where at / =
the particle was known to have
an x coordinate within the range Ax centered about the expectation value
x and momentum
Furthermore,
let
in the range
Ap centered about the expectation value p.
Ax and Ap be equal to
the minimum
the product of
ft,
allowed by the uncertainty principle. Then the normalized wave function
for this particle at any time
YOr,
0=2=
m
can be shown to bef
!
J
iE " tlh
Wn (x')V(x',0)dx'
(8-13)
Vn (x)
CO
where
(z'
) 2
ipx'-i
[
and
Vn (x)
iK *
[For a particle moving with the same
initial
potential V{x), the expression for Y(x,
;) is essentially
the eigenfunctions
equation (8-13)
for that potential
y> n (x)
strictly for the
more general
same except that
conditions in a
the
must be
used.]
Fortunately
sake of interest.
We
it is
present
almost
always possible to formulate quantum mechanical problems in such a
way
wave function for a free particle moving in the potential V(x) =
can be represented by the wave function (8-5) or (8-9), or by a simple
that the
combination of the two, instead of something as complicated as equation
(8-13). [The same is true for a particle moving in a more general potential
V(x).] Only such problems will be considered in this book.
Consider the two wave functions, (8-5)
W(x,
and
Ae~ iEtin e iKx
Be- iEtin e~ iKx
(8-9)
W{x,
t)
which are solutions to the Schroedinger equation for a free particle.
Choose the ratio of the arbitrary constants such that A = B, and add
the two wave functions. The result is a new wave function,
W(x,
To be
t)
Ae- iEt/n (eiKx
e~
iKx
)
must be written as
integrals since
n is a
continuous variable for the case of a free particle. Equation (8-13) has been written
here as a sum to bring out its analogy with equation (7-57).
t
correct, (8-13),
and also
(8-12),
Sec.
THE FREE PARTICLE
I]
217
which can be written
Y(x,
Ae~ iEt,n [cos
Kx +
Y(x,
Y(x,
t)
Kx -
sin
(-Kx)
cos
sin
(-Kx)]
= lAie-'^sin Kx
= A'e- iEt/K sm Kx
(8-14)
an arbitrary constant which is equal to 2i times the arbitrary
Next choose the ratio of the constants such that A = B,
and add the two wave functions. The result is a second new wave
where A'
is
constant A.
function,
Be~ im \eiKx
Y(x,
e~
iKx
)
which can be written
*(x,
t)
Be- iml \co%
Kx +
Y(x,t)
TO,
sin
Kx +
cos
(-Kx)
sin
(-Kx)]
= 2Be- iEt/n cosKx
= B'e- iEtlH cos Kx
(8-15)
where B' is an arbitrary constant equal to 2 times B, and which is independent of A'. Since the Schroedinger equation is a linear differential
equation, the wave functions (8-14) and (8-15) are also solutions, and so
is their sum,
T(x,
- iEtl
\A'
sin
Kx +
B' cos
Kx)
which can be written
Y(x,
t)
e-
iEt/K
(8-16)
y>(x)
where
y>(x)
= A' sin Kx +
B' cos
Kx
(8-16')
and
K = VlmE/h
We recognize from the form of equation (8-16) that (8-16') must also be a
solution to the time independent Schroedinger equation (8-1) for a free
particle.
Since
it
contains two arbitrary constants,
the solutionjust as general as (8-4').
The
it is
the general
form of
solution (8-16') could have
been written down directly as was (8^4'). However, the procedure we have
followed has the advantage of emphasizing the relation between the two
general forms of the solution.
Furthermore, it helps us to interpret the meaning of the wave function
First of all, we note that the wave function was obtained by
adding and subtracting two traveling waves of equal amplitudes and must,
therefore, be a standing wave. This, of course, can be seen directly from
the form of (8-16), since the nodes of T(a;, t) are fixed in space at values
(8-16).
SOLUTIONS OF SCHROEDINGER'S EQUATION
218
[Ch. 8
of x for which A' sin Kx + B' cos Kx = 0. Next we recall that each of the
two traveling waves have been associated with a particle of precisely known
momentum moving in an infinitely long beam, with the momentum being
equal in magnitude but oppositely directed in the two cases. Consequently
we must associate the standing wave (8-16) with a particle which is moving
in an infinitely long beam in such a way that the magnitude of its momentum
p is precisely known but the direction of the momentum is not known.
That is, the wave function is to be associated with a particle which could
be moving in the direction of either increasing or decreasing x, with
momentum of magnitude precisely equal to p, and with a completely
unknown * coordinate.
Let us check our interpretation of equation (8-16) by evaluating the
probability density P(x, i) and the probability flux S(x, i). For simplicity,
set one of the constants, say B', equal to zero. Then
Y(x,
i)
A'e- iEt ' n sin
Kx
(8-17)
and
P(x,
= Y*Y = A'*eiEtlK sin KxA'e- imi1i sin Kx
= A'* A' sin 2 Kx
P(x,
t)
(8-18)
and also
S(*,
=-^F*^_T^)
2m
dx
dx
\
S(x,
t)
= -
1m
iEtin
[A'*e
sm Kx A'e-
A'e- imin
S(x,
The
sin
mtlK
Kcos Kx
KxA*eimin K cos Kx~]
(8-19)
agreement with
According to that interpretation, the probability per unit time that the particle associated with the wave function
will cross the point x in the direction of increasing x will be equal to the
probability per unit time that it will cross in the direction of decreasing x.
These will balance each other out, and the total probability flux S(x, i)
will be equal to zero. Next consider the probability density P(x, t), which
is plotted in figure (8-1). Also plotted in the same figure is the constant
probability density which would be expected classically for a particle in
the situation which we have described as being associated with the wave
function (8-17). The fact that the quantum mechanical P(x, t) extends to
infinitely large and small values of x agrees with our interpretation of the
wave function and with the classical behavior of the particle, but the
presence of points along the x axis at which the particle would never be
fact that the probability flux vanishes is certainly in
our interpretation of ^(x,
t).
Sec.
THE FREE PARTICLE
I]
found
in contrast to
is
219
what would be expected
This
classically.
is
manifestation of the wave-like aspects of a particle in the situation
described by this
wave
because the distance d
Such behavior
function.
length of the particle.
is
not observed classically
= hjlp is only one-half the de Broglie wave-
tt/K
In the classical limit this distance becomes ext) pattern is so compressed in the x direction
tremely small, and the P(x,
that only the average (constant) behavior of P(x,
t)
can be resolved
experimentally.
P(x,t)
oc sin jRj:
P(x,t)
Figure 8-1.
The
probability density for a free particle standing
For the standing wave function (8-16) there
concerning normalization as there
is
will
wave
constant
function.
be the same problem
for the traveling
wave
For
function.
a particularly simple way to remove the difficulty. The normalization problem arises because of the unrealistic way
the wave function extends to infinitely large and small values of the cothe standing wave there
ordinate x.
is
Consequently the problem can be removed by completely
but finite, region
suppressing the wave function outside the large,
L/2
We
<; x <: L/2.
function which
that region.
is
That
do
by constructing a more
this
realistic
wave
the function (8-16) inside the region, and zero outside
is,
we
take
(*,
_
=
-iEtlK
a
ip(x)
e
(8-20)
where
rw =
, ,
A' sin
Kx
0,
w(x)
It is clear
(8-20).
<
that there will be
B' cos Kx,
-L/2,
no
difficulty in
Furthermore, the eigenfunctions
-L/2
< x < L/2
> L/2
,D
^,
(8-20
normalizing the wave function
now
vanish as x
-*
oo,
and
immediately apparent from equation (7-53) that there will be no
difficulty in proving that such eigenfunctions are orthogonal. We shall
see in section 5 that the procedure indicated in equation (8-20') is
mathematically equivalent to saying that the potential energy V(x)
becomes infinitely large for x
L/2 and x
L/2. This is as if the
it is
<
SOLUTIONS OF SCHROEDINGER'S EQUATION
220
region of space
Z./2 < x < L/2
[Ch. 8
were enclosed in a (one dimensional)
box with completely impenetrable walls. The procedure is known as
box normalization. As long as the walls of the box are very far
away from the region of interest, their presence will have no physical
effect of importance, but they will remove the mathematical difficulties
associated with normality and orthogonality for a free particle The values
of the constants A' and B' required to normalize the wave function will
depend on the length of the box, L; however, these constants will not
enter into the results of a calculation of any quantity of physical interest.
Box normalization will also change the eigenvalues E from continuous to
discrete. But we shall see that the separation between adjacent eigenvalues
2
is proportional to Lr
With a very large value of L the eigenfunctions
will, for all practical purposes, remain continuous.
Normalization of the traveling wave function (8-4) in a box with
impenetrable walls is not possible because, as will be shown in section 5,
the wave function must always vanish at the walls of the box. This
boundary condition can be satisfied by a standing wave with fixed nodes,
but it cannot be satisfied by a traveling wave with moving nodes. However,
we shall see in Chapter 15 that a form of box normalization can be used
for the traveling wave function, if we assume that the walls of the box are
penetrable so that the wave function need not vanish at the walls. In
this form of box normalization we impose the so-called periodic boundary
conditions by demanding that at any instant the wave function and its
spatial derivative have the same values at both walls. It is immediately
apparent from equation (7-53) that with these boundary conditions the
eigenfunctions are orthogonal. These boundary conditions also lead to the
convenient result that both forms of box normalization produce the same
discreteness of the eigenvalues. With normalization in a box of penetrable
walls, we are interested only in what happens in the box, and we restrict
the domain of the variable x to the region between its walls. The wave
.
The value of the normalization constant
depends on the length of the box, but this always drops out of all calculations of physically measurable quantities.
function can then be normalized.
The
problem of normality and orthogonality
wave functions and eigenfunctions, both traveling
wave and standing wave, may be summarized briefly by saying that it is
usually possible simply to ignore the problem. The mathematical consistency of the theory is ensured by the fact that a wave function in the
form of a group may be used, or that one of the forms of box normalizasituation relative to the
for the free particle
tion
may be employed.
to do either, and the
is."
In practice,
it is
wave functions
not normally necessary actually
(8-4)
and (8-16) can be used "as
STEP POTENTIALS
Sec. 2]
2.
221
Step Potentials
we shall calculate the quantum mechanical
whose potential energy can be represented by a
In the next few sections
motion of a
particle
function V(x) which has a different constant value in each of several
adjacent ranges of the x axis, and which changes discontinuously in going
V(x) =
V(x)
=
*
Figure 8-2.
step potential.
from one range to the next. Of course, the situations represented by
potentials which are discontinuous functions of x do not really exist in
nature. However, such potentials are often used in quantum mechanics
to approximate real situations because, being constant in each range, they
are easy to treat mathematically. With such potentials we shall be able to
illustrate
a number of characteristic quantum mechanical phenomena.
Figure 8-3.
Let us consider
figure (8-2).
physical system approximating a step potential.
the simplest example, a step potential, as shown in
choose the origin of the x axis to be at the step. Then
first
We
V(x) can be written
V(x)
= V
0,
x>0
x<0
(8-21)
V is a constant We may think of V(x) as a limiting representation
of the potential energy function for a charged particle moving along the
where
axis of a system of
two electrodes which are held
illustrated in figure (8-3).
at different voltages, as
SOLUTIONS OF SCHROEDINGER'S EQUATION
222
mass m and total energy E is in the region
moving toward the point at which V(z) changes. According
Assume
that a particle of
<
is
and
[Ch. 8
move
to classical mechanics the particle will
freely in that region until it
an impulsive force F(x) =
dV(x)/dx acting in the direction of decreasing x. The subsequent
motion of the particle depends, classically, on the relation between E
and V
this is also true in quantum mechanics. We take first the case:
reaches x
where
0,
it
subjected to
is
E<V
This
is
Since the total energy
illustrated in figure (8-4).
is
-
EV(x)
The
V(x)
=V
-E
x
Figure 8-4.
a constant,
case in which classical mechanics predicts that the particle
is
reflected
at the discontinuity of the potential.
classical
mechanics says that the particle cannot enter the region x
as in that region
^<^
>
0,
f\lm <
change the momentum
The impulsive force will
way that it will exactly reverse
its
of the particle in such a
motion, traveling off in the direction of
momentum equal in magnitude (so that the total energy
remains constant) but opposite in direction to its initial momentum.
To determine the motion of the particle quantum mechanically, we must
decreasing x with
find the
wave function which
is
our discussion of section
5,
E< V
We know from
a solution, for the total energy
to the Schroedinger equation for the potential (8-21).
Chapter
7,
that acceptable solutions to the
E^
Schroedinger equation for the potential V(x) exist for any value of
0.
Since we are dealing with a time independent potential^ the actual problem
is
to solve the time independent Schroedinger equation
eigenfunction.
With
and
find the
x axis breaks up into two regions.
a solution to the simple time inde-
this potential the
In one region the eigenfunction
pendent Schroedinger equation
is
_^_d>^) =
2m dx2
In the other region
it
is
YK
a solution to the simple time independent
Schroedinger equation
h*
dV*) + V y>(x) =
2m dx2
Ey>(x),
x>0
(8-23)
223
STEP POTENTIALS
Sec. 2]
Then an eigenfunction valid
x is constructed by joining the two solutions
in such a way as to satisfy the conditions (7-38), which
and dy>(x)ldx be everywhere finite and continuous.
The two equations
are solved separately.
for the entire range of
together at x
require that
y>(x)
Consider equation (8-22). This is precisely the time independent
Schroedinger equation (8-1) for a free particle. The general solution to
equation can be written either in the traveling wave form (8-4')
or in the standing wave form (8-16'). Since the two are mathematically
equivalent, either may be used. However, by using the form most approthis
priate to the
problem
at hand,
it is
often possible to simplify the calculation
When
or the interpretation of the results.
unbound
particle
it is
usually
most
convenient to use the traveling
bound, the standing wave form
convenient. For the problem at hand we take
form. If the particle
is
f(x)
Ae iKlX
an
wave
usually the most
discussing the motion of
Be~ iKlX
is
x<0
(8-24)
where
K
x
Next consider equation
(8-23).
VlmEjh
The
general solution to this equation
= Ce K x + De~ K x
x>0
yix)
is
(8-25)
where
K = V2m(V - E)\h
2
To
E< V
and
verify this, calculate
^^ = CK\eK
*x
dx2
and
+ D(-K2fe- K x
*
substitute into equation (8-23). This gives
_Jt2m
2m
The
^s
fr
arbitrary constants A, B, C,
eigenfunction satisfies
and
the conditions (7-38).
yKx)
QED
must be so chosen that the
Consider
first
the behavior
+ co.
In this region of the x axis
the solution to the time independent Schroedinger equation is given by
equation (8-25). It will, in general, diverge as x -* + oo because of the
of the eigenfunction in the limit
first
term. In order to prevent
this,
->-
we must
set the coefficient
of the
first
term equal to zero. Thus
C=
(8-26)
SOLUTIONS OF SCHROEDINGER'S EQUATION
224
Now
consider the eigenfunction at the point x
solutions (8-24)
are continuous.
Continuity of
D(e- K >*) x=0
is satisfied.
At
0.
and (8-25) must join in such a way that
ip(x) is
obtained
A(eiK*) x=0
if
[Ch. 8
this point the
y>(x)
and
dy>(x)jdx
the relation
B{ e iK *) x=0
This yields
D=A+B
(8-27)
Continuity of the derivative of the solutions
dy>(x)
= -K2 De- K x
x>0
dx
and
dip(x)
iK^Ae*^*
dx
is
obtained
if
-K
is satisfied.
iKx Be- iKlX
<
the relation
2 D(e~
K*% =0 =
iKxA(e iK n*=o
~ iK^***'),x =
This yields
i^ D =
Adding equations (8-27) and (8-28)
A-B
(8-28)
gives
(8-29)
Subtracting gives
JB
_ ^^1
iEi)
R(i
=
ll
2\
Thus the eigenfunction
(1 +
w(x)
De~ KiX
for this potential,
iK2 IK^e iK * x
(8-30)
Kj
+ - (1 -
and for the energy E,
is
<
iKJKJe-***',
x^O
(8-31)
The arbitrary constant D has not been determined because we have not
normalized the eigenfunction. This could be done by the techniques
described in section 1, but it is not necessary.
The wave function
,
f)
^ {X A
'
is
Ae -iBtih e KlX +
~ De-We-x*,
i
Be -iEtm e -iK ^
1
^
x^O
*~ 5Z)
STEP POTENTIALS
Sec. 2]
In the region x
<
it
increasing x, plus a
the region x
function,
<
>
it is
225
consists of a
wave
traveling in the direction of
traveling in the direction of decreasing x;
wave
in
In interpreting this wave
consists of a standing wave.
it
useful to calculate the probability flux S(x,
t)
in the region
equations (8-7) and (8-11) and the wave
functions from which these expressions were derived, it is apparent that
By comparison with
0.
in the present case
S(x,
where
hKjm,
t)
= vA*A-
vB*B,
<
(8-33)
the magnitude of the velocity of the particle.
The
first
term in the wave function, is a probability flux flowing in the direction of increasing x. The second term,
which comes from the second term in the wave function, is a probability
term, which comes from the
first
flux flowing in the opposite direction.
Consequently,
we
associate the
first
term in Y(a;, i) or S(x, t), in the region x < 0, with the incidence of the
particle upon the point at which the potential energy changes, and the
second term with the reflection of the particle from the change in potential.
Let us calculate the ratio of the intensity of the reflected probability flux
to the intensity of the incident probability flux. We obtain
vB*B
vA*A
(1
(1
+
-
B*B
A* A
(1
(1
- jKJKt)*(1 - iKjKJ
+ iK,IKJ*(\ + iKjKJ
Kx)(l - iKJKJ _
iKJKtfL + iKjKJ
tK/ 2
The quantity B*BjA*A is also the ratio of the intensity of the reflected
traveling wave to the intensity of the incident traveling wave in the region
x < 0. That these ratios are unity means that a particle incident upon the
change in potential, with total energy E < VQ has unity probability of
being reflected. This is in complete agreement with the classical pre,
dictions.
Kxx,
<
x^O
(8-34)
iK
Consider the eigenfunction (8-31). Writing e ^ x = cos xx + / sin
etc., it is easy to show that the eigenfunction can be expressed as
y(x)=
D - sin K,x,
D cos K,x
*
Kj.
De~ K *x
,
The wave function corresponding to this eigenfunction is actually a
standing wave for all x. In this problem the incident and reflected traveling
waves combine to form a standing wave because they are of equal intensity.
Let us plot (8-34), which is a real function of a; if we take D real. The
SOLUTIONS OF SCHROEDINGER'S EQUATION
226
plot
is
shown
[Ch. 8
Here we find a feature which is in sharp
Although in the region x >
the
in figure (8-5).
contrast to the classical predictions.
probability density
P(x,
= Y*(a;, t)W(sc,
t)
t)
D*De~ iK*x
(8-35)
decreases rapidly with increasing x, there is a finite probability of finding
the particle in this region. In classical mechanics it would be absolutely
impossible to find the particle in the region x
>
*
Figure 8-5.
An
because there the total
=o
eigenfunction for a step potential in the case in which classical mechanics
predicts that the particle
is
reflected.
energy
is less than the potential energy.
Penetration of the classically
excluded region is one of the more striking predictions of quantum
mechanics.
We shall discuss later certain experiments which confirm this prediction.
Here we would like to point out that equation (8-35) is not in disagreement
with the experiments of classical mechanics. It is apparent from that
equation that probability of finding the particle with a coordinate x >
is only appreciable in a region starting at x =
and of length Ax of the
order of 1/A"2 Thus
.
Ax
~ 1/K
hlV2m(V
E)
E)
In the classical limit the product of
and
to h2 , that A* is immeasurably small.
(V
is
so large, compared
Furthermore, the uncertainty principle demonstrates that the wave-like
by an entity in penetrating the classically excluded
region are not really in conflict with its particle-like properties. Consider
an experiment capable of showing that the particle was located somewhere
properties manifested
>
>
x
is appreciable only in a region
0. Since P(x, t) for x
of length Ax, the experiment would amount to localizing the particle within
in the region
STEP POTENTIALS
Sec. 2]
that region.
Ap
in the
In doing
momentum
this,
227
the experiment
would lead to an uncertainty
of the particle, where
A/>
~ HIAx ~ V2m(V
- E)
Consequently the energy of the particle would be uncertain by an amount
A =
(Ap?/2m
~V
-E
would no longer be possible to say that the total energy of the
E was definitely less than the potential energy V
Next let us consider the motion of a particle in a step potential when:
and
it
particle
E>V
This
in figure (8-6).
is illustrated
traveling in the region
< 0,
Classically,
a particle of total energy
E
V(x)
V(x)
= 0x
Figure 8-6.
The case
in
which
in the direction of increasing x, will suffer
= V
classical
mechanics predicts that the particle
is
not reflected
at the discontinuity of the potential.
an impulsive retarding force F{x)
dV(x)/dx at the' point x = 0. But
and it will enter the region x > 0,
the impulse will only slow the particle
motion in the direction of increasing x. Its total energy E
its momentum in the region x <
is px where E =
pWlm; its momentum in the region x > is p2 where E = p\J2m + V
Quantum mechanically, the motion of the particle is described by the
wave function Y(a;, i) = e-iEtlHy(x), where the eigenfunction y(x) is a
continuing
its
remains constant;
solution to
-?-E,).
2m dar
x<0
(8-36)
x>o
(8 - 37)
and
-^
=(E - v
xi
and where the eigenfunction satisfies the conditions (7-38) at the point
* = 0. The first equation describes the motion of a free particle of momentum/^. The traveling wave form of its general solution is
ip{x)
Ae iKlX
Be~ iKlX
where
K = JlmE/H = pJH
t
x<0
(8-38)
SOLUTIONS OF SCHROEDINGER'S EQUATION
228
[Ch. 8
The second equation describes the motion of a free particle of momentum
p2 The traveling wave form of its general solution is
.
f(x)
Ce iK **
De~ iK
x>0
*x
.
(8-39)
where
K 2 = 72m( - V )/h =
and
p2 /h
E> V
The wave function
specified by equations (8-38) and (8-39) consists of
waves of de Broglie wavelength Ax = hjp1 = 2tt\Kx in the region
x < 0, and of longer de Broglie wavelength A2 = hjp2 = 2tt/K2 in the
region x > 0.
traveling
Classically,
the particle has unity probability of passing the point
and entering the region x
cally.
>
0. This is not true quantum mechaniBecause of the wave-like properties of the particle, there is a
certain probability that the particle will be reflected at the point x = 0,
where there is a discontinuous change in the de Broglie wavelength.
Thus we need to take both terms of the general solution (8-38) to describe
the incident and reflected traveling waves in the region x < 0. However,
we do not need to take the second term of the general solution (8-39).
This term describes a wave traveling in the direction of decreasing x in
the region x > 0. Since the particle is incident in the direction of increasing x, such a wave could only arise from a reflection at some point with
a large positive x coordinate. As there is nothing out there to cause a
reflection, we know that there is only a transmitted traveling wave in the
region x > 0, and so we set the arbitrary constant
The
arbitrary constants A, B,
dfix)\dx continuous at x
Ae
iKlX
)*=o
D=
and C must
The
0.
Bie-
be chosen to make fix) and
requirement
first
iK
(8-40)
n*=o
Cie
is satisfied if
iK **)
x=0
or
A+B=C
The second requirement
ilM(e*a=o
(8-41)
is satisfied if
iKx Bie- iK
x =o
iK2 CieiK**) x=0
or
K iA - B) = K C
x
From
we
equations (8-41) and (8-42)
b =
(8-42)
find
-k
and
(8-43)
C=
2Kl
K +K
1
A
2
STEP POTENTIALS
Sec. 2]
Thus the eigenfunction
is
K K
+ A i~ * <T**i",
K + K2
AeiK i*
w(x)
arbitrary constant
<
x^O
iK",
e
(8-44)
IF
A
The
229
can be chosen to
satisfy the
normalization
condition, but
we do not bother to do
satisfying the
two continuity conditions and the normalization condition
could not be found if we had
wave equal to zero, since then
down
wave
the
initially set the coefficient
we would have
an eigenfunction
B of the
reflected
only two arbitrary constants
We shall not plot the complex function (8-44).
to satisfy three conditions.
Writing
It is clear that
this.
function,
and evaluating the probability
flux at
< 0, we find
some point x
S(x,
t)
= v A*A - v B*B
x
where
= hKjm = px/m
vx
The first term is the incident flux, and the second term is the reflected flux.
The transmitted probability flux can be evaluated by calculating S(x, t)
at some point x > 0. This gives
S(x,
i)
= v 2 C*C
where
v2
The
ratio
HK^jm
= pjm
of the intensity of the reflected flux to the intensity of the
the probability that the particle will be reflected back into
incident flux
is
the region x
< 0.
This
is
R= viE B = (K -K f
2
vx
The
ratio of the intensity of the transmitted flux to the intensity of the
incident flux
region x
>
is the. probability
0.
This
v 2 C*C
Vl
show
_ K2
A*A
T
easy to
that the particle will be transmitted into the
is
T=
It is
(Kx+KJ2
A*A
{IK,)2
(Kx
+ K2 f
4KX K2
(K t
+Kf
(8-46)
that
R+
T=
(8^7)
SOLUTIONS OF SCHROEDINGER'S EQUATION
230
The
we
upon
probability flux incident
and a
into a transmitted flux
[Ch. 8
the potential discontinuity
reflected flux, but
is split
from equation (8-47)
see that the total probability flux is conserved.
that either the particle will be reflected or
will
it
This statement says
be transmitted, but it will
not vanish. Of course, the particle is never split at the discontinuity.
In any particular trial the particle will go one way or the other. For a large
number of trials, the average probability of going in the direction of
decreasing x is R, and the average probability of going in the direction of
increasing x is T. Note that R and T are both unchanged if x and
2
are exchanged.
A moment's
consideration will
show
that this
means the
same values of R and T would be obtained if the particle were incident
upon the potential discontinuity in the direction of decreasing x from the
region x > 0. The wave function describing the motion of the particle,
and consequently the probability flux, is partially reflected simply because
there is a discontinuity in V(x), and not because V(x) becomes larger in the
direction in which the particle is going.
It is interesting to investigate the value of RforE~ V and also for
,
E > V In the rst case, K2 = V2m(E - V )/h - 0,
V2mElH-*V2mV IH. Thus
\Kt
and the probability
second case,
the particle
while
fi
+K
reflected
becomes
Thus
XkJ
that the particle is reflected approaches unity.
K2 -> VlmEjh = Kx
is
K =
In the
R ->
negligible.
and the probability that
However, the case E > V
at which we know there will
not necessarily the case of the classical limit
be no reflection. This becomes apparent if we evaluate equation (8-45)
in terms of E and V , giving
is
R-
(l-J^m
Vi
+ vi -
(M8)
VJEl
Planck's constant does not enter into the equation; the value of R depends
only on the ratio of
to E.
Keeping
this ratio constant,
(8-48) predicts that there will be reflection even
when
and
that the predictions of classical mechanics should be valid.
paradoxical until
we
equation
E are so large
This seems
realize that, in the limit of large energies,
the
assumption that the change in V(x) is perfectly sharp can no longer be
even an approximation to a real physical situation. The point is that the
reflection is the result of an abrupt change in the de Broglie wavelength;
"abrupt" means that the wavelength changes by an appreciable amount in a
distance of 1 wavelength. If the change in wavelength in 1 wavelength
becomes very small, because the change in the potential in that distance is
BARRIER POTENTIALS
Sec. 3]
231
very small, the reflection becomes negligible.
This can be demonstrated
by solving the Schroedinger equation for a potential of the form shown in
figure (8-7). These statements, which follow from the characteristic
properties of wave motion, should be familiar to the reader from his study
of the behavior of water waves or light waves. In the classical limit the
V(x)
Figure 8-7.
A smoothed
step potential.
de Broglie wavelength is so very short that any physically realistic potential
V(x) changes only by a negligible amount in 1 wavelength, and no reflection
However, for particles in atomic or nuclear systems, the de Broglie
wavelength can be long relative to the distance in which the potential
changes significantly, and reflection phenomena can become very
results.
important.
3.
Barrier Potentials
We
consider next a barrier potential, illustrated in figure (8-8).
potential
This
is
V(x)
=F
'
0,
<x<a
< 0, x >
(8-49)
vo~
(
x=Q
x=a
-x
Figure 8-8.
Classically,
incident
barrier potential.
a particle of total energy
upon the
transmitted if
in the region
< 0,
which
is
barrier in the direction of increasing x, will have unity
E> V
<
E V , and unity probability of being
Neither of these conclusions obtains in quantum
probability of being reflected if
mechanics.
SOLUTIONS OF SCHROEDINGER'S EQUATION
232
For
this potential acceptable solutions to the time independent Schroe-
dinger equation exist for all
E^
Also the equation breaks into three
x < 0,
< x < a, and x > a.
0.
separate equations for the three regions
and
energy E. The
In the
[Ch. 8
first
third regions the equation is that of a free particle of total
traveling
w(x)
f(x)
wave
=
=
solutions are
AeiK x
Ce
iK x
Be~ iK x
'
De~
4-
'
iK
<
>a
'
(8-50)
where
K\
^jlmElh
In the second region the form of the equation and of the solutions depends
E< F
on whether
or
E> V
the previous section. In the
y>(x)
Both of these cases have been discussed in
first
Fe~ K x
case,
Ge K x
0<x<a
(8-51)
where
x =
V 2m (K> - E)l h
and
E<
In the second case,
v(x)
Fe iK "i x
Ge~ iK i" x
0<x<a
(8-52)
where
K m = ^2m{E - V )lh
and
E> V
In evaluating the motion of a particle incident in the direction of increasing
x,
we may
set
D=
we know
that there can only be a transmitted wave in the region
However, in the present situation we cannot set G in equation
< x < a. Nor can we set
(8-51) since x remains finite in the region
G = in equation (8-52) since there will be a reflection from the discontinuity at x = a.
since
>
a.
Consider
first
the case
<V
In matching
ip(x)
and df(x)/dx
at the points
equations in the constants A, B, C, F, and
it
as an exercise for the reader to set
evaluate B, C, F, and
up
G will
and x
be obtained.
these equations
a,
four
We
leave
and use them
to
G in terms of A. The value of A would be determined
were applied. The form of the eigenThis figure is meant to be
< x < a,
only an indication, since the exact form of f(x) in the region
as well as the ratio of its amplitude in the region x > a to its amplitude
by the normalization condition,
function obtained
is
if it
indicated in figure (8-9).
BARRIER POTENTIALS
Sec. 3]
<
in the region x
y>(x)
>
0,
depends on the values of E, V a, and m. Furthermore,
complex function. [For example, ip(x)
CelKlX for
,
actually a
is
In the figure
a.]
The most
233
we have
plotted the real part of
ip(x).
interesting result of the calculation is the ratio
intensity of the probability flux transmitted into the region
the intensity of the incident probability flux.
T=
A*A
which
sinh
(4/F )(l
The value of
K^a
T
x
of the
>
a to
T is
E<V
E/V )
is
sinh
T=
(i
n2
Figure &-9.
-l
2mV a 2
(4/F )(l
The
- to)
/F )
E<V
(8-53)
real part of a barrier penetration eigenfunction.
If the argument of the hyperbolic sine is large compared to
a very small value which is given approximately by
T~ 16
V
(1
EIV )e- 2K " a
1,
Twill have
(8-54)
These equations make the remarkable (from the point of view of classical
mechanics) prediction that a particle of mass m and total energy E, which
is
incident
on a potential
on the other
course,
T is
the quantity
is
side.
This
V > E and finite thickness a,
T of penetrating the barrier and appearing
barrier of height
actually has a certain probablity
phenomenon
is
called barrier penetration.
Of
vanishingly small in the classical limit because in that limit
2mV a2/H2
which
is
a measure of the "opacity" of the barrier,
extremely large.
Next consider the case
>v
In this case the wave function
is a traveling wave in all three regions,
but of longer wavelength in the region
x
a. Evaluation of the
< <
SOLUTIONS OF SCHROEDINGER'S EQUATION
234
constants B, C, F, and
and x
G by the application of the continuity conditions at
a, leads to the
value
2
T=
vx
which
[Ch. 8
A*A
sin JCjija
(4/F )(/F
E>V
1)
is
T=
The forms of
2wF a2
2
//2
E>V
(4/F )(/F
(8-55)
1)
and (8-55) depend on the value of
the functions (8-53)
Figure (8-10) shows a plot of
as a function of
/F
for
1.0
0.9
0.8
2mV.a %
1
0.7
0.6
~ q9
ft*
0.5
0.4
0.3
0.2
0.1
/III
1
Figure 8-10.
The transmission
10
coefficient for a barrier potential.
an electron incident upon a barrier of height V = 10 eV and thickness
a = 1.85 X 10- 8 cm; (2mV a/fP = 9). This is typical of the values for
the potentials encountered, for instance, by an electron moving through
the atoms of a crystal. For EjV < 1, T is very small. But, when E/V
Q
is only somewhat smaller than unity, Tis not at all negligible. It is apparent
that barrier penetration can be very important in atomic systems. For
EjV
1, T is in general less than 1, owing to reflection at the discon-
>
tinuities in the potential.
that
T=
whenever
However, from equation (8-55)
Km a =
n, 2n, 3n,
dition that the length a of the barrier region
integral
number of wavelengths A |N
particular barrier, electrons of energy
This
is
is
it can be seen
simply the con-
equal to an integral or half-
2irlKm in that region.
E~ 21 ev,
53 ev,
For
etc., satisfy
this
the
Sec. 3]
BARRIER POTENTIALS
condition
Kw a =
any
reflection.
reflections at
-rr,
and so pass
2tt, etc.,
This effect
into the region x
>
a without
a result of destructive interference between
is
and x
235
a.
It is closely related
to the
Ramsauer
effect
observed in the scattering of electrons by atoms, which will be discussed in
Chapter
15.
Barrier penetration
a manifestation of wave-like behavior in the
is
motion of a particle. The same phenomenon is also observed in other types
of wave motion. We have seen that the time independent differential
equation governing the propagation of waves in a string is essentially the
same as the time independent Schroedinger equation. This is also true of
the time independent differential equation for electromagnetic radiation.
For radiation of frequency
refraction
\i
this
equation
d?y>(x)
z
dx
When we compare
this
propagating through a
^V)
(
y(*)
with equation (7-35),
independent Schroedinger equation
gation of light through a
medium of index of
is
medium
is
in
(8-56)
it is
apparent that the time
identical with that for the propa-
which the index of refraction
is
the following function of position
M *)= 2i;v^ C In
fact, there are optical
phenomena which
F(*)]
~
(8 57)
are exactly analogous to each
of the quantum mechanical phenomena that arise in considering the
motion of an unbound
particle.
For
instance,
an optical phenomenon,
completely analogous to the quantum mechanical
phenomenon of
unity
transmission coefficient for barriers of length equal to an integral or halfintegral
number of wavelengths,
obtain very high
is
light transmission.
used in the "coating" of lenses to
In particular, the situation
E<
V(x)
encountered in quantum mechanical barrier penetration is equivalent to
the optical situation for an imaginary index of refraction. This occurs in
the case of total internal reflection.
Consider a ray of light incident upon a glass to air interface at an angle
This situation, which is called total
internal reflection, is illustrated in figure (8-1 1). A detailed treatment of
greater than the critical angle 6 C
this process in
refraction,
terms of electromagnetic theory shows that the index of
line ABC, is real in the region AB but
measured along the
imaginary in the region BC. Furthermore, it shows that there are electromagnetic vibrations in the region BC of exactly the same form as the
exponentially decreasing standing wave of equation (8-34) for the region
E < V(x). The flux of energy is zero in this electromagnetic standing wave,
SOLUTIONS OF SCHROEDINGER'S EQUATION
236
just as the flux of probability
wave.
first
However,
if
is
zero in the
a second block of glass
[Ch. 8
quantum mechanical standing
is placed near enough to the
block to be in the region in which the intensity of the electromagnetic
vibrations
is
still
appreciable, these vibrations are picked
up and they
propagate through the second block, and the electromagnetic vibrations
in the air gap
now
carry a flux of energy through to the second block.
Glass
Figure 8-1
This situation, which
Illustrating total internal reflection.
1.
is
called frustrated
total
internal reflection,
is
same thing happens in the
quantum mechanical case when the region in which E < V(x) is reduced
from infinite thickness (step potential) to finite thickness (barrier potential).
The transmission of light through an air gap, at an angle of incidence
greater than 6 C was first observed by Newton (ca. 1700). The equation
illustrated in figure (8-12).
Essentially the
Glass
Glass
Figure 8-12.
Illustrating frustrated total internal reflection.
relating the intensity of the transmitted
gap, and the other parameters,
and
it
is
beam
identical in
to the thickness of the air
form with equation
(8-53),
has been verified experimentally.
Barrier penetration of particles was first discussed in connection with
a long-standing problem concerning the emission of alpha particles by
radioactive elements.
As a
typical example, consider a nucleus of the
BARRIER POTENTIALS
Sec. 3]
U238
237
The potential energy V(r) of an alpha particle
from the nucleus had been investigated by means of Rutherford scattering experiments. Using as a probe the 8.8 X 10 6 ev = 8.8
Mevf alpha particles emitted from the radioactive element Po 212 it was
radioactive element
at a distance r
observed that the probability of scattering agreed with the predictions of
Rutherford's formula over the entire angular range. From this it was
concluded
that, for the
= zZe^jr at
V(r') = 8.8 Mev.
V(r)
U238 nucleus,
least in to
It
was
also
scattering of alpha particles
V{r) exactly follows the
a distance
r'
known from
from the
3.0
10~ 12
Coulomb law
cm
at
which
the experiments involving the
nuclei of light
atoms that
V(r)
must
u*
N
3 8.8
UJ
V(r)
ss-E
4.2
-Kinetic energy at large r
Figure 8-13.
the relation between potential and total
An energy diagram showing
energies in alpha particle emission.
eventually depart from a 1/r law
at that time the exact value of
nuclei.
by
U 238
when
R was
<
not
R, the nuclear radius, although
known
for the
heavy radioactive
Furthermore, since alpha particles are (very infrequently) emitted
nuclei, it was assumed that they must exist inside such nuclei
to which they are normally bound by the potential V(r). From these
arguments it was concluded that the form of V(r) in the region r < r'
must be qualitatively as depicted in figure (8-13). (This assumption has
been verified by recent experiments involving the scattering of alpha particles of energies high enough to investigate the potential over the entire
range of r.) It was also known that the kinetic energy of alpha particles
emitted by U 238 was 4.2 Mev. The kinetic energy was, of course, measured
and the kinetic
at a very large distance from the nucleus, at which V(r) =
energy equals the total energy.
also
This value of the constant total energy
this situation is
completely paradoxical.
E is initially in the region r <
An alpha particle
R. This region
space by a potential barrier of height which is
t
is
shown in figure (8-1 3). From the point of view of classical mechanics,
Mev =
10 6 ev
nuclear physics.
1.602
x 10 -6
ergs.
is
of total energy
separated from the rest of
known to be at least twice E.
This unit of energy
is
used extensively in
SOLUTIONS OF SCHROEDINGER'S EQUATION
238
[Ch. 8
Yet it is observed that on occasion the alpha particle penetrates the barrier
and moves off to large values of r.
In 1928 Gamow, Condon, and Gurney treated alpha particle emission as
a quantum mechanical barrier penetration problem. They assumed that
V(r) = zZe*\r for r > R, and V(r) < E for r < R, as shown in figure
(8-13). Equation (8-54) was used to evaluate the transmission factor T
since K^a > 1. In fact, this quantity is so large that the exponential
completely dominates the behavior of T, and it was sufficient to take
This expression was derived for a rectangular barrier of height
width
and
but in the limit that the expression is valid it can be applied to the
barrier V(f) by considering it to consist of a set of adjacent rectangular
barriers of height V(r?) and width Ar{ The reason is that in the limit rea,
flections
from the
discontinuities in the rectangular barriers
important contributions to the transmission factor, and
pletely
do not make
is com-
value
dominated by the exponential decrease in the eigenfunction in the
The transmission factor for
regions within the barriers.
barrier
its
the
z'th
rectangular
is
J*
If reflections
-2V'(2m/S 2 )[F(r < )-S]Ar i
can be ignored so that there
is
no interaction between the
individual barriers, the total transmission factor will be the product of
the individual transmission factors because independent probabilities are
Then
multiplicative.
_ e - V (2m/S )[F(r - El An g - V(2m/B )[F(r -J2] Ar2g -2 V(2m/B )F[(r - E] Ar
_ e -2(\ (2m/S )tF(r )-S] Ar + V(2m/B )[F(r )- E] Ar + V (2m/B )[F(r - E] Ar +
_ e -2^V (2m/ )[F(r,)-]Ar,/
j-
Ar become very
s
J_
s)
Letting the
2)
1)
small, this
becomes
-2j\/ (2m/S 2 )[F(r)- E]dr
J
(8-58)
rigorous treatment of this problem shows that this procedure
justified, in that it gives the correct
answer,
if
T is
very small, as
it is
is
for
the case of alpha particle emission.
The quantity
T gives
the probability that in one trial an alpha particle
will penetrate the barrier.
The number of trials per second can be estimated
as
N~
v/2R
(8-59)
Sec. 4]
if it is
SQUARE WELL POTENTIALS
assumed that an alpha
particle
239
bouncing back and forth with
is
Then
velocity v inside the nucleus of diameter 2R.
second that the nucleus will emit an alpha
designated by the symbol
particle,
the probability per
which
is
conventionally
X, is
~ e -2^V(2W )[(,zeV)-]*
2
2R
(8
v
_6Q)'
In applying this expression to a particular radioactive nucleus, all the
quantities in this expression except v and R are known. Assuming v to
be comparable to the velocity of the alpha particle after emission (i.e.,
mv*
~9
E), then X
is
a function only of the nuclear radius R.
x 10-13 cm, which was
Using
certainly in line with the values obtained
from Rutherford's analysis of the alpha particle scattering experiments
from light elements, Gamow, Condon, and Gurney obtained values of X
which were in good agreement with those experimentally measured,
although they vary over a tremendously large range. For U 238 X =
5 X 10~ 18 sec" 1
For Po212 X = 2 X 106 sec" 1 This variation is due
primarily to the variation, from one radioactive element to the next, of
the energy E of the emitted alpha particles. The height and breadth of the
barrier (Z and R) do not change significantly for nuclei in the limited
range of the periodic table in which the radioactive elements are found.
,
The successful application of the theory of quantum mechanics to the
process of alpha particle emission provides one of the most convincing
verifications of the theory.
Furthermore,
illustration of the wave-particle duality.
this
On
process gives a striking
either side of the barrier the
alpha "particle" could be localized and shown to behave like a particle.
In penetrating the barrier it behaves purely like a wave. Any attempt to
localize it in the barrier would, according to the uncertainty principle,
necessarily introduce such a large uncertainty in the energy that it would
no longer be
possible to say definitely that
it
was not going over the top of
the barrier.
4.
Square Well Potentials
In this section we shall discuss the simplest potential which is capable
of binding a particle to a limited region of space, a square well potential,
indicated in figure (8-14).
V(x)
This potential
VQ
0,
<
is
a/2,
a/2
<x<
>
+aj2
(8-61)
+a/2
SOLUTIONS OF SCHROEDINGER'S EQUATION
240
Consider
>
[Ch. 8
the case
first
(traveling waves)
under the influence of this potential with total
energy E > V would be able to move over the entire range of the x axis.
Upon entering the region aj2<x< +a/2, it would receive an accelerating impulse and travel through that region with momentum higher than
Upon leaving, it would receive a compensating
its initial momentum.
Classically, a particle
decelerating impulse. Its final
We
tum.
shall
momentum would
equal
its initial
momen-
not develop the traveling wave solutions to the quantum
= -
a/2
A square
Figure 8-14.
mechanical problem, which exist for
x = a/2
well potential.
E > V since
E> V
all
they are essentially
the same as those for a barrier potential with
In fact, the trans-
can be obtained directly from equation (8-55) if the
terms are appropriately redefined. This factor is, as before, less than unity
except for energies at which a is just equal to an integral or half-integral
number of de Broglie wavelengths in the region a/2 < x < +a/2.
mission factor
We
return later to discuss the standing
potential with
We
<
E> V
wave
solutions for the square well
consider instead the case:
V (standing waves)
If the particle has total energy
in the region
a/2 < x <
E< V
+a/2. The
then classically
particle
it
could only be
would be bound
to that
region and would bounce back and forth between the ends of the region
momentum
of constant magnitude but alternating direction. Classiany value of the total energy E is possible for a bound state. In
section 5, Chapter 7, we have seen qualitatively that in quantum mechanics
with
cally,
only certain values of the total energy
are possible. In this section we shall
evaluate quantitatively the discrete eigenvalues, and the corresponding
eigenfunctions, for a square well potential.
Such a potential
is
very often used in
quantum mechanics
to represent
SQUARE WELL POTENTIALS
Sec. 4]
a situation in which a particle
moves
the influence offerees which hold
some
potential loses
it
details of the
241
in a restricted region of space under
Although this simplified
in that region.
motion,
it
retains the essential feature
of binding the particle by forces of a certain strength to a region of a
As an example, many aspects of the motion of a nuclear
certain size.
can be explained by assuming that the particle is
about 50
ev and linear
dimensions of about 10 -12 cm.
The description of the classical motion of particle bound by the square
well suggests that it would be most appropriate to look for solutions to the
Schroedinger equation in the form of standing waves. Thus we take, as a
solution to the time independent Schroedinger equation in the region
a/2 < x
+a/2, in which V(x) = 0, the free particle standing wave
particle in a nucleus
bound
in a square well potential with a depth of
<
This
eigenfunction (8-16').
y>(x)
=A
KyC
sin
is
+ B cos K^x,
a/2
<
<
+a/2
(8-62)
where
= VlmElh
> +a/2 the time independent Schroedinger
K,
In the regions x
<
equation
same
is
the
a/2 and x
According to equation (8-25),
as equation (8-23).
the solutions are
y(x)
Ce K " x
De~ K " x
f{x)
Fe K x
Ge~ K " x
<
> + aj2
-a/2
(8-63)
and
,
where
(8-64)
<
X M = yj2m(V To determine
E)/n
the arbitrary constants
the eigenfunctions remain finite for
the limit x ->
oo.
It is
and
first
all x.
apparent that
this
<V
impose the requirement that
Consider equation (8-63) in
requirement demands
D=
Similarly,
it is
(8-65)
necessary to set
F=
(8-66)
remain finite in the limit
+ oo. Next
impose the requirement that the eigenfunctions and their first derivatives
be continuous at x = a/2 and x = +a/2. Four equations are obtained.
x -*
in order that equation (8-64)
They
are
-A sin (K a/2) + B cos (X,a/2) = Ce~ K a
AK\ cos CK,a/2) + BK sin (X,a/2) = CK^e-*""
A sin (K,a/2) + B cos (X,a/2) = Ge _JC
AK\ cos (X,a/2) - BK\ sin (K,a/2) = -GK^e~ K m
'<
'2
(8-67)
12
(8-68)
a/2
"
(8-69)
(8-70)
SOLUTIONS OF SCHROEDINGER'S EQUATION
242
[Ch. 8
Subtracting equation (8-67) from (8-69) yields
2A
sin (K|a/2)
Adding equation (8-67)
IB
(G
- Qe'^"' 2
(8-71)
Qe-^ii^
(8-72)
to (8-69) yields
cos (K,/2)
(G
Subtracting equation (8-70) from (8-68) yields
2BK
sin (.K,a/2)
+ QKnc-^'^2
(G
Adding equation (8-70)
2AK
(8-73)
to (8-68) yields
cos (X,o/2)
= -(G -
C)Ke- x o/2
(8-74)
Provided 5
and (G
C)
# 0, we may divide equation (8-73) by (8-72)
and obtain
= Ku
(G + C) #
K, tan (K,a/2)
# 0and(G -
C)
if
Provided A
and
(8-75)
# 0, we may divide equation (8-74) by (8-71)
and obtain
X, cot (X,a/2)
if
It is
and
(G
-K|,
C)
(8-76)
easy to see that both equations (8-75) and (8-76) cannot be satisfied
simultaneously. If they could, the equation obtained by adding these two,
X, tan (K,a/2)
would be
becomes
valid.
+K
cot (K,a/2)
Multiply through by tan(AT|o/2).
2C|
tan 2 (X,a/2)
K,
Then
the equation
or
tan 2 (K,a/2)
But
this
= -1
cannot be valid as both K\ and a/2 are
real.
Thus
it is
only
possible either to satisfy (8-75) but not (8-76) or to satisfy (8-76) but not
(8-75).
Just as for the eigenfunctions of the vibrating string, the eigen-
functions of the square well potential form
two
classes.
For the
first
class,
K|tan(lS:|/2>=K||
G-
A=
C=
(8-77)
SQUARE WELL POTENTIALS
Sec. 4]
Then equation
243
(8-69) reads
cos (2^/2)
Ge~ K " al2
G = 5cos(X|a/2)e K
a/2
i'
=C
and the eigenfunctions are
K a/2 K x
[ cos (K\al2)e ~\e
y,(x)
[B] cos
(K
<
-a/2
-a/2
x),
<x<
a/2 (8-77')
\B cos
^>
(A' a/2)e^M/ ] e -^n^
2
a \2
For the second
class,
X| cot
Then equation
= K"n
B=
G+ C =
(A'|a/2)
(8-78)
(8-69) reads
= Ge^"
G = A sin (X|a/2)e K n 0/2 = -C
yl sin
(K
'2
a/2)
and the eigenfunctions are
[-4
sin
(K
al2)e
K " al2 ~\e K " x
<
-a/2
y,()
[A] sin
[A
Consider the
sin
first
-a/2
(A", a),
(X|a/2> K a/2 Je- K M^
a;
>
< x < a/2
(8-78')
a/2
Evaluating K\ and K^, and
of equations (8-77).
multiplying through by a/2, the equation becomes
(8-79)
VmEa /2h2 tan (VmEa2 l2h 2) = Vm(K - E)a2 \2h
For a given particle of mass m and a given potential well of depth V and
2
width a, this is an equation in the single unknown E. Its solutions are the
allowed values of the total energy of the particle the eigenvalues for
eigenfunctions of the first class. Solutions of this transcendental equation
can be obtained only by numerical or graphical methods. We present a
simple graphical method which will illustrate the important features of the
equation.
Let us
make
the change of variable
e
= VmEa l2H
2
j2h2
(8-80)
so the equation becomes
e tan
If
we
= VmV a
plot the function
p(e)
tan e
and the function
q(e)
= VmV a
/2h 2
(8-81)
SOLUTIONS OF SCHROEDINGEFVS EQUATION
244
[Ch. 8
the intersections specify values of s which are solutions to equation
(8-81).
at e
Such a plot
0,
it, 277,
function q(e)
is
shown in figure (8-15). The function p(e) has zeros
and has asymptotes at e = irj2, ~h-n\2, 5ttJ2, .... The
is
VmV a l2H
2
a quarter-circle of radius
number of
figure that the
It is clear
from the
solutions which exist for equation (8-81)
depends on the radius of the quarter-circle. Each solution gives an
E < V corresponding to an eigenfunction of the first
eigenvalue for
Figure 8-15.
graphical solution of the equation for eigenvalues of the first class of a
particular square well potential.
cl ass.
There
one such eigenvalue
exists
VfflVffl < 2n;
VrnV^fih 2
solutions: e
[Solution of e tan e
three
is
if
if
= VmV^jlh* e
VmV a j2h <
2
< VmV^fth < 3tt;
2
2tt
illustrated in the figure.
~ 1.25 and e c 3.60.
For
tt;
etc.
or
p(s)
two
= q{e).]
if
The
this case there are
From equation (8-80),
<
case
two
the eigenvalues
are
mV a 2
ma 2
4 /
and
2tf
-,v'
mVaa
$f)v. = <mv.
The eigenvalues corresponding
to eigenfunctions of the second class are
found from the solutions of an analogous equation obtained from
(8-78), which is
-e cot e =
VmV a l2h 2
(8-82)
SQUARE WELL POTENTIALS
Sec. 4]
245
Figure (8-16) illustrates the solution of this equation.
there will be
no eigenvalues
of the second class
Vm
if
E< V
for
< n/2; there will be one if 7r/2 ^
lit\2 < V mF q /2A < 5ir/2;
etc.
The
two
if
the
case
VmV cPI2fP = 4.
figure
illustrates
equation (8-82)
E=
Figure 8-16.
e*
mV a
graphical solution of the
class of a particular
r(e)
The
ci 2.45, and the eigenvalue
is e
apparent that
a2l2fi 2
VmV a l2h < 3tt/2;
2
It is
corresponding to eigenfunctions
single
solution
to
is
equation for eigenvalues of the second
square well potential. [Solution of
cot s
= VmV
o 2 /2A a
e2
or
q(e).]
We see that for a given potential well there are only a restricted number
E for E < F These are the discrete
of allowed values of total energy
eigenvalues for the
know
bound
that any value of
E is
On
of the particle.
states
allowed for
E> F
the other hand,
we
the eigenvalues for the
form a continuum. For a potential well which is very
shallow or very narrow or both, only a single eigenstate of the first class
unbound
states
VwF
a2/2/z 2 an eigenstate of the
be bound. With increasing values of
second class will be bound. For even larger values of this parameter an
will
additional eigenstate of the
first
consider the case
VmF^/W =
continuum eigenvalues are
bound. Next, an additional
bound, etc. As an example
class will be
eigenstate of the second class will be
4.
The
potential
and the
discrete
illustrated to scale in figure (8-17).
We
and
have
SOLUTIONS OF SCHROEDINGER'S EQUATION
246
used the quantum numbers n
1, 2, 3, 4, 5,
For
in order of increasing energy.
[Ch. 8
to label the eigenvalues
this potential
only the
first
three
eigenstates are bound.
Continuum
E2 = Q.37V
(2nd class)
(1st class)
Figure 8-17.
The eigenvalues
of a particular square well potential.
-Jatyx)
-^aip (x)
3
The eigenfunctions
Figure 8-18.
for the
bound eigenstates of
a particular square well
potential.
From the
VmV a
may
l2fi
solutions
2
,
of equations (8-81) and (8-82) for a given value of
the explicit forms of the eigenfunctions (8-77') and (8-78')
e,
be evaluated. The required relations are
*l| =
and
K | = s/mV a 2J2h2 -
(8-83)
SQUARE WELL POTENTIALS
Sec. 4]
The value of
function
4,
the eigenvalues
Et E2
must be adjusted so that each eigenFor the case
the three normalized eigenfunctions corresponding to
,
and
4= e
17.9
w,(x)
or
normalization condition (7-49).
the
Vm V a l2h =
2
the constant
satisfies
247
E3
are
3 80
x^-a/2
^,
'
1.26 -4= cos (1.25
Va
'
17.9
4= e
-380
-18.6-|=e
-a/2
),
< x < a/2
a/2/
^,
316
a;>a/2
*<-a/2
^,
va
Was)
.23
4=
sin
w3(x)
1.13
3 16
"
4= cos
,/a
-5.80
The
'
- a/2 < x < a/2
J3.60
-1
eigenfunctions, multiplied
a;^a/2
x^-a\2
^,
4= e
^,
1 74
4= e
a/2/
18.6 -j= e"
-5.80
2.45
,ya
74
J,
a/2/
^,
-a/2
< x < a/2
a;>a/2
figure (8-18) as a
by \/a, are plotted in
function of /(a/2).
This figure makes quite apparent the essential difference between the
The eigenfunctions of the first class,
classes of eigenfunctions.
two
ip x (x)
and
y> s (x),
are even functions of
rp{-x)
a;;
that
is,
(8-84)
+y>(x)
In quantum mechanics these functions are said to be of even parity.
The eigenfunction of the second class, tp2 (x), is an oddfunction of a;; that is
ip(x)
and
is
said to be of
odd
parity.
(8-78') will demonstrate that, for
(8-85)
xp{x)
Inspection of equations (8-77') and
any value of
VwK a /2S
2
2
,
the functions
SOLUTIONS OF SCHROEDINGER'S EQUATION
248
of the
first class
class are all of
are
odd
[Ch. 8
of even parity and the functions of the second
all
Furthermore, the continuum standing wave
parity.
eigenfunctions also form two classes, one of even parity and the other of
odd
That all the standing wave eigenfunctions have a definite parity,
is a direct result of the fact that we have so chosen the
origin of the x axis that the square well potential V(x) is an even function
of ie.f For such a choice it is intuitively obvious that any measurable
quantity describing the motion of the particle in a bound eigenstate (or
in a standing wave continuum eigenstate) will be an even function of x.
parity.
either even or odd,
We
note that
P(-x,
t)
density function P(x,
this is true for the probability
for both even* parity
and odd parity
y*{-x) y,(-x)
t)
eigenstates since
[yj*(x)][f(x)]
y>*(x) y(x)
= P(x, t)
(8-86)
This
is
not true for the wave function
eigenstate ; such a
wave function
is
itself in
the case of an
odd parity
an odd function of x. Of course,
this
not a contradiction as the wave function is not measurable. The traveling wave eigenfunctions do not have a definite parity since they are linear
combinations of standing wave eigenfunctions of opposite parities. For
is
a one dimensional problem the fact that the standing wave eigenfunctions
have a definite parity, if V(x) = V(x), is of importance largely because
it leads to a simplification in the mathematics of certain quantum mechanical calculations. But in a three dimensional problem this property takes
on a deeper significance that will become apparent later.
The probability densities multiplied by a,
of
VmV a l2h = 4,
2
for the three
bound
eigenstates
are plotted in figure (8-19) as a function of x/(a/2).
Also
illustrated in the figure is the normalized probability density which
would be predicted classically for a bound particle bouncing back and forth
between a/2 and +a/2. Since classically the particle would spend an
equal amount of time in any element of the x axis in that region, it would
be equally likely to be found in any such element. The quantum mechanical
probability densities
show
that, in
each eigenstate, there
is
finite
prob-
of finding the particle in the classically excluded region. This
effect is largest for n
3, which corresponds to the smallest value of
ability
En
(V
). Also there are points
quantum mechanical probability
ability
of finding the
in the classically allowed region at
densities
show
that there
is
which
zero prob-
particle.
ymV
For a potential in which
a2 l2tP is large enough that there are
many bound eigenstates, the quantum mechanical probability density
t In figure (8-18), if we redefine the origin of the x axis to be, for instance, at the
point x
a/2, it becomes apparent that the eigenfunctions will no longer have a
definite parity.
SQUARE WELL POTENTIALS
Sec. 4]
more and more
oscillates
potential, consider
249
rapidly with increasing values of n.
ip%(%) y(),
where n
is
For such a
large but not so large as to
violate the condition
y/mV a 2l2h2
-2.0
-1.6
-1.2
-0.8
- mE n a 2j2h2 =
-0.4
0.4
jLa/2
0.8
1.2
>1
(8-87)
2.0
1.6
x
aj2
Figure 8-19.
The
probability densities for the
bound eigenstates of a
particular square
well potential.
The
probability density has the
form indicated in
figure (8-20).
The
function a f*{x) y>J%) oscillates rapidly, and because of condition (8-87)
there is a negligible chance of finding the particle in the classically excluded
(I
(I
#;w*w-
All
aP(x),classical-
-o/2
11 111
a/2
Figure 8-20. Illustrating the approach to the classical limit of the probability density for
a square well potential.
In the limit n -* oo, the oscillations in ay)*(x) y) n (x) are so rapid
that only the average behavior can be observed experimentally and the
region.
results
of the quantum mechanical theory approach those of classical
mechanics.
SOLUTIONS OF SCHROEDINGER'S EQUATION
250
It is interesting
[Ch. 8
quantum
quantum numbers, as
of the old quantum theory.
to note that in this case the classical limit of the
mechanical theory
is'
found in the
limit of large
predicted by the correspondence principle
However, it should also be noted that it is necessary to apply the restriction
(8-87), which was not anticipated by the correspondence principle. For
Figure 8-21.
An even
a square well potential and a
unbound eigenfunction for
parity
typical value of total energy.
we have
seen that the approach to the classical limit
unbound
eigenstates,
has
relation to the technique of letting n -> oo.
little
Finally, let us consider the case:
>
V (standing waves)
E>
V
value of total energy
wave eigenfunctions are not the
Any
allowed, but the features of the standing
is
same
for all values of E.
We
shall
not
bother actually to solve the time independent Schroedinger equation for
this case
(it is
qualitative
easier to
do than for
arguments and
E< V
but shall only present certain
results.
In each of the three regions of the x axis the eigenfunctions are sinuIn the region
+ a/2, the momentum
a/2<x<
soidal standing waves.
of the particle
is
p = V2mE,
so the wavelength of the eigenfunctions
is
A,
is
hlVlmE. In
the regions x
p u = Vlm(E V ), and
<
a/2 and x
the wavelength
Figure (8-21) depicts the qualitative
for a typical value of E.
It is
>
is
momentum
= hfV2m{E V ).
+o/2, th e
kn
form of an even parity eigenfunction
easy to see that, because A N
>
\,
it
is
possible to join smoothly the sinusoidal solutions of the three regions only
< <
x
+fl/2 is less than it is
such that exactly an integral number
of wavelengths fit in the region a/2 to -fa/2, the slope of the interior
solution is zero at the boundaries and the interior and exterior solutions
have the same amplitude. This is shown in figure (8-22), which depicts
if
the amplitude inside the region
outside.
However,
if
the value of E
a/2
is
THE
Sec. 5]
INFINITE
SQUARE WELL
the eigenfunction for a value of
For odd
figure.
is less
251
slightly larger
than for the previous
parity eigenfunctions, the amplitude in the interior region
than the amplitude in the exterior regions except for values of E such
number of wavelengths fit in the region.
that exactly a half-integral
we
Considering both classes of eigenfunctions,
Figure 8-22.
An even
the total energy
is
at
parity
unbound eigenfunction for
see that the probability
a square well potential
when
resonance.
of finding the particle in the interior region, which
is
equal to
f/2
ip*(x) ip(x)
/:-a/2
is
dx
generally less than the probability of finding
it
in an exterior region of
the same length. This is analogous to the classical statement that the
particle moves faster in the interior region and so it would be less likely
to find the particle in that region. However, when the length of the region
is equal to an integral or half-integral number of wavelengths in the region,
the probability of finding
it
in the interior region equals that of finding
in a comparable exterior region.
This
is
it
precisely the condition (which
arose in the discussion of the traveling wave eigenfunctions) that the
region x
lead to
5.
The
An
limit
T will
be unity for the transmission of the particle from the
a/2 to the region x > +a/2. These so-called resonances
some interesting situations in atomic and nuclear physics.
probability
<
Infinite
Square Well
important special case of a square well potential is found in the
oo, which is illustrated in figure (8-23). This potential is
V(x)
0,
a/2
oo,
<
<
<
a/2
(8-88)
a/2,
>
a/2
SOLUTIONS OF SCHROEDINGER'S EQUATION
252
We
shall see that
[Ch. 8
easy to find simple closed-form expressions for the
For small values of
it is
eigenvalues and eigenfunctions of such a potential.
the
quantum number
n, these
eigenvalues and eigenfunctions can often
be used to approximate the corresponding (same ri) eigenvalues and eigenAn infinite
functions of a square well potential with large but finite V
.
-a/2
Figure 8-23.
square well potential
a/2
An
infinite
also used in discussing the
is
=*-
square well potential.
quantum mechanical
properties of a system of gas molecules (or other particles) which are
strictly
confined within a box of certain dimensions. In addition, such a
potential
is
used in the technique for the impenetrable walled box normali-
zation of free particle eigenfunctions.
Let us consider the
on the eigenfunctions of
effect
Shown in figure (8-24) is
potential. As V -> oo,
the eigenfunction for n
Ex
will
change; but
it
letting
->
oo.
in a finite square well
will
change very slowly
compared to the change in V Thus the coefficient of x in the exponential
will become infinitely large and the value of y>y(x) will decrease infinitely
.
- a/2
Figure 8-24.
The
a/2
first
eigenfunction for a finite square well potential.
x for x > a/2. In the limit, y>i(#) must be zero for
Then the requirement that an eigenfunction be everywhere
continuous demands that ip^ix) =
at x = a/2. For an infinite square
well potential, y>i(x) must have the form shown in figure (8-25). It is
rapidly with increasing
all
>
a/2.
apparent that
this
argument holds for
tpjx)
= 0,
all
x^-a/2,
the eigenfunctions.
x^ a/2
That
is,
(8-89)
THE
Sec. 5]
SQUARE WELL
INFINITE
253
for all n. The condition (8-89) can only be satisfied by violating the requirement of (7-38) that the derivative d\p n {x)jdx of an eigenfunction be con-
tinuous everywhere. However, if the reader will review the argument
which was presented to justify this requirement, he will find that dy)Jx)/dx
must be continuous only when V is finite.
For the infinite square well potential, we must solve the time independent
Schroedinger equation
d 2 f n(x)
subject to the condition (8-89).
2m _
(8-90)
Comparing with the
0C
-a/2
Figure 8-25.
The
first
COS
differential
equation
(^S *)
a/2
eigenfunction for an infinite square well potential.
(7-60) and the condition (7-63),
we
problem
see that the present
is
mathe-
matically identical with the problem of a vibrating string fixed at both ends.
The
solution of the present problem could be obtained directly from
equation (7-65) by making the appropriate substitutions. It could also be
obtained from the equations of the preceding section by letting V ~* oo.
However, we shall solve the problem again in order to illustrate how the
knowledge that the eigenfunctions must have a definite parity, because
V{ x) = V(x), can be used to simplify the calculation.
From equation (8-16') the standing wave form of the general solution
to (8-90)
is
y) n (x)
sin
(Knx)
B cos
-a/2
(Kx),
< x < a/2
where
Kn = V2mEjh
Since cos
of
x,
(Knx)
is
an even function of x and
sin
(Knx)
is
an odd function
the even parity eigenfunctions are
y, n (x)
= Bn cos (Knx),
-a/2
< x < a/2
(8-91)
-a/2
< x < a/2
(8-92)
and the odd parity eigenfunctions are
y> n (x)
= A n sin (Knx),
SOLUTIONS OF SCHROEDINGER'S EQUATION
254
[Ch. 8
In both cases the values of n must be such as to satisfy condition (8-89).
Because of the evenness or oddness of the eigenfunctions, it is only
necessary to apply condition (8-89) at x = +a/2. From equation (8-91)
we have
= Bn cos (Kn a/2)
This demands
Kna\2 =
which
tt/2, 377/2, 5tt/2,
Kn = mr/a,
Thus the even
n
1, 3, 5,
is
n=
(8-93)
parity eigenfunctions are those for
Next apply condition (8-89)
1, 3, 5,
at
quantum numbers
= +a/2 to (8-92)
and obtain
= A n sin (Kna/2)
This demands
Knaj2 =
which
tt, 2tt, 3tt,
is
Kn = nirja,
The odd
4, 6, ...
quantum numbers n
parity eigenfunctions are those for
To
2,
normalize the eigenfunctions, evaluate the integral
r
For even parity eigenfunctions
r0) Wn(X) dx
this is
Bz\ '\o{K n x)dx =
'-a/2
Since this
(8-94)
2,4,6,...f
must equal
unity,
B^2
we have
B n = V2fa
Similarly
it is
found that
A n = V2ja
Thus the normalized eigenfunctions are
V2/a
V>n(x)
cos (mrx/a),
= V 2/a sin {mrxja),
=
=
1, 3, 5,
2, 4, 6,
* < a/2,
0,
...
< x < a/2
a/2 < x < a/2
x ^ a/2
a/2
(8-95)
t The value n
ing solution Vote)
satisfies
condition (8-89), but
for all x.
it
leads to the physically uninterest-
THE INFINITE SQUARE WELL
Sec. 5]
It is
hardly necessary to plot
that there
is
or f*(x) y n (x). Except for the fact
excluded region for an infinite
y> n (x)
no penetration of the
255
classically
square well potential, the eigenfunctions and probability densities are
similar to those for a finite square well potential.
n -> oo
The approach
to the
same as for a finite square well except that
there is no auxiliary condition on n comparable to (8-87).
From equations (8-93) and (8-94), the eigenvalues are
classical limit as
is
the
En = K^jlm =
The
and the
potential V(x)
(8-26).
Of course,
TT
all
k=
2 2 2
h n l2ma 2
first
(8-96)
1,2,3,4,5,
few eigenvalues are
illustrated in figure
the eigenvalues are discrete for
an
infinite
square
^ = 16 i^
E, = 9
E, = A
Et =
-a/2
Figure 8-26.
The
first
is
the
first
is
a/2
to the region
< <
a;
is
bound
infinite
for
square well potential.
any
finite eigenvalue.
Of
7T
h 2 l2ma 2
It is the
(8-97)
lowest possible total energy
bound by the infinite square well potential
+a/2. The particle cannot have zero total
a manifestation of the uncertainty principle. If the particle
bound by the potential, then we know its x coordinate to within an
energy. This
is
if it is
eigenvalue,
called the zero point energy.
the particle can have
a/2
Ex =
This
2roa 2
2ma2
few eigenvalues for an
well potential since the particle
particular interest
7T
2ma 2
uncertainty
is
Ax
must be Apx
>
a.
hjt\x
Consequently the uncertainty in
hja.
The uncertainty
its
momentum
principle cannot allow the
particle to be bound by the potential with zero total energy since that
would mean the uncertainty in its x momentum would be zero. For the
eigenstate
the magnitude of the
momentum
is
px =
\2mEx =
-nhja.
SOLUTIONS OF SCHROEDINGER'S EQUATION
256
Since the particle whose motion
is
[Ch. 8
described by the standing wave eigen-
function can be moving in either direction, the actual value of the x
momentum
is
uncertain by an
agreement with the lower limit
point energy
is
amount &px = 2px = 2-nhja. This is in
by the uncertainty principle. The zero
set
responsible for several interesting phenomena, particularly
in connection with the behavior of matter at very
6.
low temperatures.
The Simple Harmonic Oscillator
We have
discussed
most of the
interesting cases of potentials
which are
discontinuous functions of position with constant values in adjacent
regions. Let us
now consider potentials which are
position. It turns out that there are only a limited
for which
it is
closed form.
continuous functions of
number of such potentials
possible to obtain a solution to the Schroedinger equation in
But, fortunately, these potentials include
some of
the
most
important cases: for example, the Coulomb potential V(f) oc r~ x which
will be discussed in Chapter 10, and the simple harmonic oscillator potential
V(x) oc x2 , which is discussed in this section. Techniques for handling the
,
more
recalcitrant continuous potentials will be presented in the next
chapter.
The simple harmonic oscillator potential is of great importance because
can be used to describe the broad class of problems in which a particle
is executing small vibrations about a position of stable equilibrium. At
a position of stable equilibrium, V(x) must have a minimum. For any
V(x) which is continuous, the shape of the curve in the region near the
it
minimum, which is all that is of interest for small vibrations, can be
approximated by a parabola. Thus, if we choose the zeros of the x
and the energy axis to be at the minimum, we can write
V(x)
C*\l
well
axis
(8-98)
a constant. This potential is illustrated in figure (8-27). A
influence of such a potential experiences a
dVjdx = Cx. The theory of
simple harmonic restoring force
classical mechanics predicts that, if the particle is displaced by an amount
x from the equilibrium position, and then let go, it will oscillate sinu-
where
is
moving under the
particle
F=
soidally
about the equilibrium position with frequency
v
fiL
2ir 'V
where
m is
energy
the mass of the particle.
of the particle
is
(8-99)
According to that theory, the total
2
and can have any value,
proportional to x
THE SIMPLE HARMONIC OSCILLATOR
Sec. 6]
since
is
arbitrary.
predicts that
257
However, the old quantum theory (Planck's postulate)
only one of the values:
E can have
En =
nhv,
0, 1, 2, 3,
(8-100)
What
are the predictions of quantum mechanics?
Let us specify the simple harmonic oscillator potential (8-98) in terms
of the classical oscillation frequency (8-99). From these two equations
we have
= l^mvV
V(x)
Figure 8-27.
A simple
harmonic
(8-101)
oscillator potential.
The corresponding time independent Schroedinger equation
2m
dx*
or
r?5
is.
is
(^yi
(8-102)
Define
Then
(8-103)
the equation can be written
cPipjdx2
It is
ImE/H2
iTrmvjh,
(fi
- oAr> =
convenient to express this in terms of a
|
Vac
new
_dy> d
dx~~dldx~
dx2
/-dip
yJa
d\dx)dx
variable,
(8-105)
a;
Then
djp
(8-104)
'
d$
d?
SOLUTIONS OF SCHROEDINGER'S EQUATION
258
[Ch. 8
and the time independent Schroedinger equation becomes
or
We
must find solutions
from oo to +oo. The
the solutions
which are continuous, and finite, for all I
condition will be automatically satisfied by
ip(g)
first
we shall obtain. However, it
consideration of the requirement that
this
purpose
values of
it is
useful
ip()
be necessary to take explicit
remain finite as ||[ > oo. For
will
to consider the
first
form of y(|) for very
large
|||
For any finite value of the total energy E, the quantity /8/oc becomes
compared to I2 for very large values of || Thus we may write
negligible
dhp/d?
The
= IV
general solution to this equation
xp
To check
Ae?'
|||
(8-107)
is
-* oo
Be-?1*
(8-108)
the solution, calculate
= Mefl2 + B(-)e-s
tPyld? = A?e?l + Ae^ 2 + BlV^ - Be~^ 2
= A( 2 + l)e?l* + B(2 - l)e~^ 2
2/2
dy>ldS
But for
|||
-> oo
this is essentially
2
A?e?'2
d 2 y\de
e(Ae^ 12
y>[d
B2e~&2
or
Now
III
->
Be'^' 2 )
|>;
Q.E.D.
apply the condition that the eigenfunctions must remain finite as
oo. It is apparent from equation (8-108) that this demands we set
>
.4
Thus the form of the eigenfunctions for very
y<|)
= Be-& 2
III
large
^ oo
|||
must be
(8-109)
This suggests looking for solutions to equation (8-106), which will be
valid for all |, in the
form
V<)
e"
|2/2
H(|)
(8-110)
Sec. 6]
THE SIMPLE HARMONIC OSCILLATOR
The H(i)
are functions which must, at
pared to e
_f2/2
.
To
^=
-e-t 2/2H
d\ =
e~ n2
d?
Then
substitute
-172
>
?e-!' <*H
-l /2
dH _
f /8
dH_
d
2
,dH
f ff
dH
2| =il
d2
+
.
df
(8-106).
dH
This gives
H + ^ e -l H -| e-^ H =
2
/2
/2
1~
<PH
d$ 2
d|
Dividing by e~^ 12 and canceling the terms involving
d 2H
/2
df2 J
'
e _f
<?H
and d2 xpjd2 into equation
xp
be slowly varying com-
oo,
evaluate these functions, calculate
-H + H
-H +
|f|
259
2
,
H, we have
g-l)*-0
(8-111)
This equation determines the functions H().
Let us recapitulate.
As
equation (8-106).
We started with the time independent Schroedinger
stands, this equation cannot be solved in closed
However, by writing the solutions to the equation as products of
_|2/2
the function e
which is the form of the solution for ||| -> oo, times the
functions H(g), we transform the problem to one of solving equation
it
form.
This equation is soluable in closed form. In fact, it is identical
with an equation called the Hermite differential equation whose solutions
(8-1 1 1).
are functions which are well
known
in the mathematical literature.
However, we shall solve it ourselves.
To do this we must make use of the most general procedure available
for the purpose of obtaining an explicit solution to a differential equation.
This consists in assuming a solution in the form of a power series in the
independent variable. That is, we assume
H (0 = 1 a** =
fc
The
+ aJ +
a.f
"s?
(8-112)
(8-1 12) into (8-1 1 1)
are then determined by substituting equation
and demanding that the resulting equation be satisfied
for any value of f
Calculate
coefficients
ax a 2
,
= 2 ka^' 1 =
^T
a?
*=i
lfl!
2a 2
3a 3 | 2
SOLUTIONS OF SCHROEDINGER'S EQUATION
260
[Ch. 8
and
^=
Then
|(fc
l)fca t f*-
2a 2
2a2
-2
3a3 |
4a4 ^
5a5 3
- 2 3c3 |
+ 08/a - IK + (|9/a - lKf + (/S/a - IK** + (/J/a
l^f
2a 2 f
4a 4 | 2
This gives
substitute into the differential equation.
1
3a 3 |
l)a3 3
Since this is to be true for all values of , the coefficients of each power of
I must vanish individually so that the validity of the equation will not
depend on the value of . Gathering the coefficients together and equating
them to
zero,
we have
+ (PI* - IK =
-2- 1K =
2-3fl3 + (iS/a- 2 2)c2 =
3 4a4 + (0/a 4-5c5 + (^/a- -2-3)a3 =
?:
I
2o 2
For the kth power of the
+ 2K+.+ (--l-2fc)a t =
(fe+lXfc
?:
relation is
so
(fe
This
is
called the recursion relation.
+
It
l)(fc
2)
allows us to calculate successively
and the coefficients a3 a5
ay The coefficients a and a1 are not specified by
equation (8-113). But this is as it should be; the second order differential
equation for #(!) should have a general solution containing two arbitrary
constants. Thus the general solution splits up into two independent series,
which we write as
the coefficients a 2 a4 , a6
,
a~,
JJ(|)
in terms of a
in terms of
= ajl + -2 e + -4 -
+ 4* + Tj3 +
+ -6 -4 -
%%? + %%%?+)
The ratios a]c+ Jak are given by the recursion
an even function of I, and the second series
variable.
relation.
is
The
(8-lW)
first series is
an odd function of that
THE SIMPLE HARMONIC OSCILLATOR
Sec. 6]
For an arbitrary value of
contain an infinite
both the even and the odd
/?/a,
number of terms. As we
Consider either
acceptable eigenfunctions.
261
shall see, this will
and evaluate the
series
of the coefficients of successive powers of f for large k. This
_ Q3/q-l-2fc) _
(fc + l)(fc + 2)
a
iSsa
ak
Let us compare
e
|2
,
which
e?
with the ratio for the
this
2fc
2
fc
fc
expansion of the function
=i+
t
+L+L+
fc+2
tie
+ __ + _
3!
(fc/2)!
(fc/2
large k, the ratio of the coefficients of successive
l/(fc/2
ratio
is
is
2!
For
series
series will
not lead to
1)!
(k/2)!
l/(k/2)!
(fc/2
(fc/2
k/2
l)(fc/2)!
1)!
powers of
(k/2)!
1)!
+1
is
^J_ = 2
fc/2
means that the terms of high power in f
e^ can only differ from the corresponding terms in the
even series of i/(f) by a constant K. They can only differ from the terms
in the odd series of //() by I times another constant K'. But, for ||| <x>,
the terms of low power in I are not important in determining the value of
any of these series. Consequently we conclude that
The two ratios
are the same. This
in the series for
tf(f)
Ke?
Now, according to equation
ax
K'^\
|||
-+ oo
(8-1 10), the solutions to the time independent
Schroedinger equation are
V()
Thus,
if the series
of H(J-) contain an infinite number of terms, the behavior
of these solutions for
e-*
But
2/2
= e-^H{)
||
H(!)
this diverges as
- oo
is
= a Ke& +
2
||
-*
oo,
aJCSt?*,
which
is
|||
-> oo
not acceptable behavior for an
eigenfunction.
However, acceptable eigenfunctions can be obtained for certain values
Set either the arbitrary constant a or the arbitrary constant ax
/?/<x.
equal to zero. Force the remaining series of if() to terminate by setting
of
/?/
= 2n+
(8-115)
where
n
n
= 3, 5,
= 0, 2, 4, 6,
1,
iffl
==
if ax ==
SOLUTIONS OF SCHROEDINGER'S EQUATION
262
It is clear
[Ch. 8
from equation (8-113) that such a choice of /?/oc will cause the
nth term since we shall have for k = n
series to terminate at the
(/?/(n
The
coefficients
l-2n)
lXn
a+4 ,
portional to a n+i
(2n
2)
a+8
fl +6 ,
The
+ 1 - 1 - 2n)'a =
(n + lXn + 2)
_
.
will also
resulting solutions
0-a
be zero since they are pro-
Hn () are polynomials of order
Hermite polynomials. For polynomial solutions to the Hermite
n, called
equation, the corresponding eigenfunctions
r(f)
= e-^H H(S)
(8-116)
always have the acceptable behavior of going to zero as ||| - oo,
|2/2
varies so much more rapidly than
||[, the function e~
the polynomial //() that it completely dominates the behavior of the
will
because, for large
eigenfunctions.
In
fact,
y({) oc e-?'\
oo
||
in agreement with equation (8-109).
Acceptable solutions to the Schroedinger equation exist only for the
/J/a specified by equation (8-115). According to (8-103), this
values of
quantity
is
2mE
/?
Thus equation (8-115)
2,
_2E
hv
Ittttiv
specifies the
allowed values of the total energy E,
which are
2Ejhv
In
or
En =
(n
\)hv,
0, 1, 2, 3,
(8-117)
where v is the classical oscillation frequency of the particle, and where
we have used the quantum number n to label the eigenvalues. The potential
and the eigenvalues are illustrated in figure (8-28). All eigenvalues are
discrete for the simple harmonic oscillator potential, since the particle is
bound for any finite eigenvalue. Comparing equation (8-117) with the
energy quantization equation postulated by Planck (8-100), we see that in
quantum mechanics all the eigenvalues are shifted up by an amount
\hv. As a consequence,
bound to the potential
this
the
is
minimum
EQ =
potential, the existence of
principle. It
is
\hv.
possible total energy for a particle
This
which
is
is
the zero point energy for
required by the uncertainty
interesting to note that Planck's postulated quantization of
Sec. 6]
THE SIMPLE HARMONIC OSCILLATOR
the simple harmonic oscillator
\hv.
263
was actually in error by the additive constant
This constant canceled out in most applications of the Planck
postulate because they involved only calculating the difference between two
(cf. Einstein's arguments leading to the idea of quanta).
However, discrepancies were noticed when attempts were made to explain
certain low temperature phenomena in terms of the old quantum theory.
Each Hermite polynomial HJ) can be evaluated from equation (8-114)
eigenvalues
_
4 -
3 _
_
E2 _
Ei -
9
2
7
2
5
by
The
first
hv
3
2
hv
The
first
HJ =
#itf) =
Htf) =
H3 () =
Ht(g) =
H (i) =
H (S) =
5
6
few eigenvalues for a simple harmonic
calculating the coefficients
for that value of n.
from (8-113) with
few are
listed
hv
Eq ~
Figure 8-28.
hv
hv
oscillator potential.
/?/a
given by (8-115)
below.
21
2
4f
- 8f3
12 - 48| 2 + 16| 4
1201 - 160|3 + 32f*
120 - 720I2 + 480f* 12
(8-118)
64f
In each case the arbitrary constant a or ax has been so chosen that the
normalized eigenfunctions are given by
*c-(jkrJ e ~'"HM
The
eigenfunctions can be written in terms of x
= Va x,
by making the
(8 " 119)
substitution
more convenient to leave them in terms of
an even function, the parity of the eigenfunctions is the
same as that of the Hermite polynomials even for even n, and odd for
odd n. The eigenfunction corresponding to the eigenvalue of lowest
energy is of even parity, just as in the case of a square potential well.
I
Since e
but
_|2/2
it is
usually
is
SOLUTIONS OF SCHROEDINGER'S EQUATION
264
[Ch. 8
few eigenfunctions are plotted in figure (8-29). The vertical lines
of. the classically allowed region in each case. In figure
(8-30) we show the quantum mechanical probability density for n = 10,
and also the probability density which would be predicted classically for a
The
first
indicate the limits
f
i
-5
ra
An=
\li
+5
1
Vi
+5
ra
iii
-5
if
Jol
ill
/ +5
'
i-
s.
Figure 8-29.
+5
The
first
-6
-5
+5
-5
few eigenfunctions for a simple harmonic
-4
oscillator potential.
-2
Figure 8-30. Illustrating the approach to the classical limit of the probability density for
a simple harmonic oscillator potential.
From
Quantum Mechanics, McGraw-Hill Book Co.,
particle of total energy equal to
Ew
L.
Pauling and E. B. Wilson, Introduction to
New
York, 1935.
The
velocity of a particle classically
executing simple harmonic oscillations goes to zero at the extremes of
its
motion. Consequently the particle spends most of the time near the limits
of the classically allowed region, and the classical probability of finding the
particle in
an element of the axis behaves as shown. The figure makes
Sec. 6]
THE SIMPLE HARMONIC OSCILLATOR
265
quite apparent the nature of the correspondence between the
mechanical and
classical probability densities in the limit
n -
quantum
oo.
To close our discussion of the simple harmonic oscillator, let us calculate
T and V of the
the expectation values
particle in the state n
0.
kinetic and
The wave function is
= w^y- iE
fed,
" tin
potential energies of a
4i e_|2/2
"**"*
77
Now
2m
Thus, according to equation (7-80),
Change
to a derivative with respect to
,-d
r d
d = Jo.
v =- = J
v
a.
djxx
dx
3f
Then
The
derivative
is
SX =
ai
The complex conjugate
1
tt
is
\J-*
_ J_ e -^l2 e +iE
t/A
IT
So
Im^JTT J -oo
The
definite integrals
can be found in any
h2x
"V^
[
4 J
we have
T=
They
_ ViH _ a^
m^TT L 2
Evaluating a,
table.
fplirmv
4mh
hv
4m
give
SOLUTIONS OF SCHROEDINGER'S EQUATION
266
According to equation (8-117)
[Ch. 8
this is
f= /2
From
(8-101), (8-103),
and (8-105) we may write the potential energy as
hnv?
Then, according to (7-80),
00
yjTT
F=
>
-oo
fV*
Ivh^TT
2vhir\\
hvjA
/2
Jo
can be shown that the relation
It
T=P=EJ2
true for all values of the
is
relation
is
quantum number. Furthermore,
true classically if T and
T and V averaged
of
(8-120)
the
same
V are understood to represent the values
over one period of the oscillation.
BIBLIOGRAPHY
Bohm,
D.,
Quantum Theory, Prentice-Hall, Englewood Cliffs, N.J., 1951.
and E. B. Wilson, Introduction to Quantum Mechanics, McGraw-
Pauling, L.,
Hill
Book
Schiff, L. I.,
Co.,
New
York, 1935.
Quantum Mechanics, McGraw-Hill Book Co.,
New
York, 1955.
EXERCISES
1
Derive equation (8-53) and verify (8-54).
2.
Derive equation (8-55).
3.
Consider the potential
0,
<0
<
>
Vi,
y,
<
where
<
and a
particle of total energy
increasing x.
Show
that
its
Vi
E > V2
< v2
approaching x
in the direction of
probability of continuing into the region x
>
is
Ch.
a
8]
EXERCISES
maximum
if
in the region
4.
is
<
267
an integral or
x < a.
half-integral
number of de
Broglie wavelengths
Obtain the transmission factor for a square well potential by redefining
the terms of equation (8-55), as suggested near the beginning of section 4.
5.
Evaluate quantitatively the standing wave solutions for the square well
potential of section 4 in the case
E> V
qualitative conclusions reached at the
end of
Use
these solutions to verify the
that section.
Obtain an analytical solution to the time independent Schroedinger equation
and determine the first eigenvalue Ex
Compare with the results of the numerical integration performed in that exercise.
6.
for the potential of exercise 4, Chapter 7,
7.
n ($) by using equations (8-113), (8-114), and
Check these against the forms presented in (8-118), and also verify
Evaluate several of the
(8-115).
that several of the eigenfunctions (8-119) are properly normalized.
8. Verify
equation (8-120) forn
1.
CHAPTER
Perturbation
Theory
I.
Introduction
The technique which we employed
in solving the Schroedinger equation
for the simple harmonic oscillator potential will not, in general, be of use
form V(x). What happens is that
found to involve more than two coefficients,
in the case of a potential of arbitrary
the recursion relation
is
making it impossible to find
solutions to the differential equation in closed
In such cases the equation can always be solved by numerical
integration in the manner described in section 5, Chapter 7. This method
form.
can be tedious if the calculation must be done "manually." But, with the
aid of modern high speed computing machines, numerical integration is
completely feasible and is widely used. In addition, there are approximation
techniques that are very useful for treating certain potentials for which
the Schroedinger equation cannot be solved in closed form.
The study
of one of these techniques forms the subject of perturbation theory, to
which
2.
this chapter is devoted.
Time Independent Perturbations
Consider a potential
V'(x), for
which
it is
either difficult or impossible
to solve the Schroedinger equation in closed form, but which can be
decomposed as follows
V'(x)
where V(x)
is
V(x)
v(x)
(9-1)
a potential for which the Schroedinger equation has been
268
TIME INDEPENDENT PERTURBATIONS
Sec. 2]
269
a potential that is small compared to V(x). We
from which it will be easy to obtain good approximations to the eigenvalues and eigenfunctions of the perturbed potential
V'(x), in terms of the perturbation v(x) and the known eigenvalues and
eigenfunctions of the unperturbed potential V(x). An example of equation
(9-1) is illustrated in figure (9-1). The potential V'(x) has been decomposed into a square well potential V(x), plus a perturbation v(x) which is
small compared to V(x).
solved,
and where
v(x) is
shall develop expressions
Co ntinu um
Co ntinu um
~4~
-3
-2
3
2
V'(x)
V(x)
v(x)
Figure 9-1. Illustrating the decomposition of a perturbed potential into an unperturbed
potential plus a perturbation.
On several occasions in Chapter 7 we wrote a function of
certain set of eigenfunctions (cf equations 7-56
.
for
some
and
That
particular perturbed eigenfunction Y^(a;).
as a sum of a
Let us do this
a;
7-70).
is,
we
writef
(9-2)
The
fi(x) are the
specify
how much
tion runs over
unperturbed eigenfunctions, and the coefficients a nl
is contained in y'n(x)- The summa-
of each of the ^(z)
the values of the quantum number /, including those in
The unperturbed eigenfunctions are solutions to the time
all
the continuum.
independent Schroedinger equation for the potential V, which
^dSpi
2m dx2
The perturbed eigenfunctions
potential V, which is
Vxpi
=E
l y> l
are solutions to the
2m dx2
V'rp'n
is
(9-3)
same equation for the
E'n y>'n
t It can be shown that every set of eigenfunctions y>i(x) forms a "complete set" in
terms of which almost any function of x can be expanded (i.e., written as in equation
9-2).
PERTURBATION THEORY
270
Using equation (9-1)
can be written
this
d2 y>'n
+ VW H +
values.
'
2m dx 2
where
vrp'n
>'
E and E'n are, respectively, the unperturbed
Now substitute equation (9-2) into (9-4)
l
2a
h 2 d 2 fl
i
[Ch. 9
+ 2 a nl v
2m dx
According to equation (9-3) the bracket
is
(9-4)
and perturbed
= J a nl E'n
y> t
equal to
eigen-
xp x
Thus we have
i?,y>,.
2 anlElVl + 2 a nl V Vl = 2 a nlE'n Wl
or
2 am(E'n ~ EdVi = 2 "mm
i
Multiplying through by the complex conjugate of a certain unperturbed
eigenfunction
y>
m and
,
we have
integrating over all x,
Too
/"oo
2 a m(E'n -
E)
ip*y> t
dx
00
= J a nl
J
I
The unperturbed eigenfunctions
ip*py dx
00
are, of course, orthogonal.
Assuming
that they have also been normalized,! equation (7-54) shows that the
on the left will be equal to zero if /
m, and equal to unity if
m. Thus there will be only one non-vanishing term in the summation
integral
1
on the
left side
of the equation, and
a nm (E'n
- EJ = 2 ai
wlpfi dx
'-OO
(9-5)
Let us define the symbol
/*00
vmi
Then equation
=J
"*(*)
V>t
(*) 1P00
dx
(9-6)
00
(9-5) can be written
a n JE'n
- EJ = 2 a nl v ml
(9-7)
This equation
that
is
is
small
useful,
is
we
exact, but
shall
it is
not very useful. In order to obtain one
employ the condition that the perturbation
compared to the unperturbed potential
t This involves
box normalization for the continuum eigenfunctions. In
the continuum eigenvalues actually
become
doing; this,
discrete (although very closely spaced).
This removes any difficulty of interpreting the summation
of/.
v(x)
V(x), so that the perturbed
^ fr the continuum values
I
TIME INDEPENDENT PERTURBATIONS
Sec. 2]
from
potential V'(x) differs only slightly
271
For such a
V(x).
situation
reasonable to assume
from the unperturbed eigenfunctions. In terms of equation
means that we assume
slightly
this
< 1,
a nni
'c=: 1,
If
we
it is
it is
that the perturbed eigenfunctions will differ only
^n
(9-2),
,
.
Q
y."-)
compared to the eigenvalues of V(x),
be small compared to the unperturbed
also require that v(x) be small
clear that the v ml
must then
all
eigenvalues because, according to (9-6), these quantities are just certain
Now
averages of v (x).
unperturbed eigenvalue
us divide both sides of equation (9-7) by the
let
Em We have
.
~ h m)
(E n
a nm
v mt
-rp
Z. "nl
Em
hm
Every term in the summation, except the term / = n, is the product of
two small quantities a nl and v ml \Em We shall neglect such terms, keeping
only the term for I = n. Then we have
.
Em
Em
or
a nm\Pn
Now take m =
We
n.
~ Em
)
(~9)
a nn v mn
obtain
SO
If
we
m # n,
take
E'n
we
~a
a
Setting a nn
-E ~ v nn
(9-10)
obtain
^ _
because of (9-8), this becomes
a
"mm
r^
Ehz>
"n
(9-11)
c 'm
Using (9-10) to evaluate ^, we find that
_
"mil.
En
Em +
V<m.<n.
(E n
- Ejll +
\
E n Em \
%n
Ej
Vnn
En ~
En E m \
EJ)
En
Ej
PERTURBATION THEORY
272
We
have taken the
first
Em
quantity v n J(En
).
[Ch. 9
term in the binomial expansion of 1 plus the
shall drop the term involving the product
Next we
- EJ and v n J(En - EJ. The validity of
two steps depends on the additional requirement that v(x) be small
even compared to the difference between En and any other eigenvalue
Em which enters into our calculations. We have finally
of the two quantities v m J(En
these
(9-12)
^n
t'n
Equations (9-10) and (9-12) are the expressions which provide good
approximations to the eigenvalues and eigenfunctions of the perturbed
potential V'(x). Consider (9-10), and evaluate v nn from (9-6). This yields
E'n
-E n ~ v nn = Jr
vC(*) v(x) Wn(x) dx
(9-13)
oo
This gives an approximation to the nth perturbed eigenvalue in terms of the
nth unperturbed eigenvalue and a certain integral involving the corre-
sponding unperturbed eigenfunction and the perturbation
v(x).
This
integral is the expectation value of v(x) for the nth unperturbed eigenstate.
To
see this, consider (7-74), with V(x,
e -iEn tm y,jx).
t)
v(x)
and T(a;,
t)
= WJx, t) =
The equation reads
v(x)
=J
iEt/n
v *^
-iE n t/n
vj^ dx
or
iff)
=
J
V>Z( X)
< X) VnW ^
oo
Thus perturbation theory
gives the very reasonable result that the shift
in the energy of the nth eigenvalue, due to the presence of the perturbing
potential v(x),
is
approximately equal to the value of v{x) averaged over
the nth unperturbed eigenstate with a weighting factor equal to the probability density y>*{x) y>Jx) for that eigenstate.
Next consider (9-1 1), and evaluate the symbol v mn
a nm
1
==:
yt(x) v{x)
xp
n {x) dx,
This yields
m^n
(9-14)
This equation gives the approximate value of the coefficients a nm which
specify how much of each of the unperturbed eigenfunctions y> m (x) is
mixed in with the dominant unperturbed eigenfunction y>Jx) to form the
TIME INDEPENDENT PERTURBATIONS
Sec. 2]
273
perturbed eigenfunction y^O**). Then in the series (9-2), with /replaced by m,
= 2 anmV>m(x)
W'( x)
(9-15)
we may use (9-14) to evaluate all the coefficients except ann From
(9-8) we know o ~ 1. Its exact value can be determined by requiring
.
that fn(x) be normalized. Note that a nm is proportional to \f(E
m).
Thus the perturbation v(x) will mix in with the unperturbed eigenfunction
amount of any unperturbed eigenfunction
ipjx) only a negligibly small
Em is very different from the
eigenvalue En This
has the important consequence that a good approximation to the series
(9-15) may be obtained by taking only the term for
n, plus a few
terms for
not very different from n. The coefficient a nm is also pro-
ip
whose eigenvalue
m(x)
m=
portional to the quantity
v mn
This
is
ft lx) v{x)
a certain average of
v(x),
y> n {x)
dx
with a weighting factor ip%(x)
y>(x)
which depends on the eigenfunction for the mth unperturbed eigenstate as
well as the eigenfunction for the nth unperturbed eigenstate.
The
quantities v mn , for
m~
n as well as
m / n,
are called the matrix
elements of the perturbation v taken between the state n and the state m.
This terminology is used because in advanced treatments of quantum
mechanics it is convenient to consider a matrix in which each element
one of the quantities v mn Such a matrix,
is
^12
"13
"22
"23
"m2
contains
all
possible information concerning the application of a perturba-
tion v(x) to a system
whose unperturbed eigenfunctions are
ipxix), yi^x),
PERTURBATION THEORY
274
3.
[Ch. 9
An Example
Let us
illustrate the
tion calculation.
We
use of (9-13) and (9-14) by doing a simple perturbashall evaluate the first eigenvalue
for the potential indicated in figure (9-2)
a
V'(x)
-a/2
We
which
is
an
2
(9-16)
a/2
V-bottom
^=
0,
oo,
<
*-
potential.
<x<
-a/2
>
sum of an unperturbed
consider this as the
<-a
x
a
<
oo,
and eigenfunction
by the equation
specified
a/2'
Figure 9-2.
<
and
potential,
a/2
>
a/2, x
a/2
square well, plus a perturbation
infinite
v(x)
lasl
a/2
According to (8-95), the normalized unperturbed eigenfunctions are
Wm( x)
yjlja cos (rrnrx/a),
m=
1, 3, 5,
yj2ja sin (m-n-x/a),
m=
2, 4, 6,
According to (8-96), the unperturbed eigenvalues are
Em =
where we use
the
first
TT
h 2 m 2 l2Ma2
M for the mass of the
m=
1, 2, 3, 4,
particle.
If d is small
eigenvalue,
E1 = TrW/lMa2
the perturbation technique should be applicable.
compared to
AN EXAMPLE
Sec. 3]
To
evaluate
which
k2
7T
m) J -a/2
(1
-m
mr)) J -a/2
7T /i
m=
2, 4, 6,
C\m
w=
3, 5, 7,
16M6
o^o2
_2fc2
ir
Z=
sin
m/1
dx,
)dx,
l|x|cosl
the integrand
Thus we have
Let
ViO)
ax,
77x/a;
a-im
1
m=
0,
the integral
(1
-m
a/2
3, 5, 7,
m=
2, 4, 6,
(ttx\
^ cos
x.
Since the
the integral will
2,4,6,...
(mirx\
cos
^7
2
0,
an even function of x;
is
m=
an odd function of
is
taken over a range symmetrical about x
vanish.
im
v ( x)
\\x\cos\
\
is
fl
wt( x)
integral
For
cos
(1
SM6
For
This gives
in (9-14).
is
a lm
take n
rp'^x),
im
275
it
gives
_
, _
m = 3, 5, 1,
dx,
...
then this becomes
-^
2
7T
C*
-m
(1
,z
cos
;r
2
) Jo
(mZ) Z cos
Z dZ
where we have introduced the convenient dimensionless ratio 6jEx
2Ma2 6JTrzh2 The integral can be evaluated easily by writing cos (mZ)
.
l( e +imZ
=
~
8_
Ulm
The
77
first
e-imZ^ etc
_d_
Ex (1 - m
The
/
result j s
+ 1M2] 2(m + l)
cos [(m
) I
few non-vanishing
cos [(m
have the values
coefficients
13
32
15
77
77
1728
a 19
864
17
n E1
2
(5
arw\ ~2 F
1)tt/2]
- l)
m = 3,
2(m
5, 7,
PERTURBATION THEORY
276
[Ch. 9
not surprising that alm =
for m = 2, 4, 6,
The perturbed
is symmetrical about the origin, and so its first eigenstate
must be of even parity. Consequently there can be no odd parity unperturbed eigenfunctions mixed into the first perturbed eigenfunction,
It is
potential V'(x)
and the odd parity unperturbed eigenfunctions are
The perturbed eigenfunction ^[(x)
2, 4, 6,
m=
precisely those for
is
obtained by sub-
alm in the series (9-15). Since the alm decrease rapidly with
increasing m [owing partly to the l/CE^ Em ) term and partly to the
v ml term], it is apparent that we can get a very good approximation to the
stituting the
series
by taking only the terms
m=
for
m=
and
3.
Thus
8 d
%0)
^ aiiViC^) +
~l
32
E
1
Vi(x)
tt
must be adjusted so that yfa)
Finally the coefficient an
but we leave
an
this as
(9-17)
is
normalized,
exercise for the reader.
Figure (9-3) illustrates equation (9-17). The relative amount of y>3 (z)
has been exaggerated for the sake of clarity. Fixing our attention on
ipfo) and fifr), we see that the second derivative of the perturbed eigen-
ends of the region a/2 to +a/2,
compared to the second derivative of
the unperturbed eigenfunction. Consideration of the form of the time
independent Schroedinger equation for the perturbed and unperturbed
function
and
is
relatively small near the
relatively large near the center,
potentials will
Next
let
make
it
clear
us evaluate E{
why
happens.
this
Taking n
E{.
in equation (9-13),
inserting the appropriate unperturbed eigenfunction,
E'1
-E = -^
1
m
\x\
cos
a J-a/2
E'x
E =
x
x cos 2
a Jo
E1
-E =
1
2
TT
which
dx
a \dx
)
Zi
JO
77
( )
and
we have
\16
4/
is
E'1
-E = 0.297(5
1
Figure (9-4) shows the perturbed eigenvalue E{ in terms of the dimensionless ratio (E[ E^jE^, plotted as a function of the dimensionless
ratio d/E^ Perturbation theory predicts the straight line of slope 0.297.
The
points are the correct answer.
They were
calculated
from the
eigen-
values E[ obtained by an exact (numerical integration) solution of the
AN EXAMPLE
Sec. 3]
277
Schroedinger equation for the four potentials V'(x) corresponding to the
The shift in the energy of the first eigenvalue, as
values of 6jEx indicated.
oc
Wx)
oc
Figure 9-3.
Illustrating
the composition of the
first
4/Jx)
eigenfunction for a V-bottom
potential.
predicted by perturbation theory,
for
{E[
is
seen to be in error by about 10 percent
~ 0.9, which corresponds to (E[ E-^jE^ ~ 0.25. For
E )IE ~ 0.05, the error about 0.5 percent. Now
apparent
d/E1
1
it is
is
that the error in the perturbation theory
we have developed
is
of the order
of the square of a small quantity since, throughout the development, the
Figure 9-4. A comparison between the first eigenvalues for several V-bottom potentials
obtained from perturbation theory and from exact solutions of the Schroedinger equation.
squares of small quantities were always neglected.
The numbers just quoted
an approximate measure of the size of
this small quantity is the ratio (E{ E )/E
Note also that the eigenvalue
1
1
E{ calculated by perturbation theory is always too large. It can be shown
indicate that, in the present case,
PERTURBATION THEORY
278
that this
why
it
is
to
tp-jix)
true for
happens.
any form of the perturbation
and
E[
E =
of ip[(x),
we
y>*(x) v(x)
y^x)
is largest,
eigenvalue E[
it is
easy to see
Comparing
dx.
see that this procedure gives too
to the values of v(x) near the ends of the region.
region v(x)
and
Perturbation theory uses the unperturbed eigenvalue
evaluate
plots of ipi(x)
v{x),
[Ch. 9
the
much weight
But near the ends of the
and therefore the contribution of v(x)
to the perturbed
overestimated.
is
comparison of the exact form of the eigenfunction y>[(x) of the
form (9-17) predicted by perturbation theory shows
potential V'(x) with the
that the error in the coefficient a13
quantity {E[
If
is
also of the order of the square of the
E^)jEv
more accurate
estimates of E'n
and
y'Jx) are needed,
it is
possible to
extend the perturbation theory to obtain expressions in which the error
is
of the order of the cube, or even of a higher power, of the appropriate
small quantity. However, in practice equations (9-13) and (9-14) are
normally adequate.
The Treatment
4.
of Degeneracies
Consider the case of two different unperturbed eigenfunctions, which
label ip^x) and f 2 (x), whose corresponding unperturbed eigenvalues
we
Ex
and
E2
happen
to be exactly equal.
These eigenfunctions are said to
We shall see later that a number of important examples
be degenerate.
this situation actually arise in the
of
study of atomic and nuclear physics^
For instance, many of the eigenfunctions are degenerate for an electron
bound in the \jr Coulomb potential of a hydrogen atom. When eigenfunctions are degenerate, we shall often be interested in studying the effect
of a small perturbation which changes the potential in such a way as to
remove the degeneracy.
However, to apply perturbation theory in a case involving degenerate
we must make certain modifications in the theory. This
need is clearly indicated by equation (9-11), which makes the prediction
1 and a
1 for the case Ex = 2 .t This really tells us only that
a 12
21
the theory we have developed breaks down in this case. But it also provides
some clue to the nature of the difficulty by showing that, when E1 = E2
/ (equation 9-8) is not consistent with the
the assumption a nl < 1 for n
results obtained from that assumption in the cases n = 1, 2 and / = 1,2.
eigenfunctions,
~
~
Equation (9-11) states a 12
v 21 l(E{ E2 ). Taking E1 = E2 and using (9-10),
becomes a 12
v^KE^ JS,)
1, in general. Note that this result does
%/
not depend on the "additional requirement" that v(x) be small compared to the
difference between two unperturbed eigenvalues.
t
this
THE TREATMENT OF DEGENERACIES
Sec. 4]
The
when we
difficulty is resolved
279
realize that there is certainly
<
priori basis for the assumption that a 12
and a21
<
1,
no a
when fi(x)
Under such
correspond to eigenvalues which are exactly equal.
circumstances it might very well be that, in contrast to the assumption, a
small perturbation could have a big effect and thoroughly mix up the two
and
y> 2 (x)
degenerate eigenfunctions.
To
we
account for
we modify
this situation
the procedure as follows. First
consider only the mixing, due to the presence of the perturbation
of the two degenerate unperturbed eigenfunctions y^x) and
we ignore the mixing with ip^x) and
v(x),
with each other. In doing this
of any of the non-degenerate unperturbed eigenfunctions
y> b (%),
....
From
we
this calculation
y?(aO
(x)
ip
2
=
=
y> 3 (x),
y> 2 (x)
ip
2 (x)
%(x),
shall find the eigenfunctions
aiiViO*)
a 2 np x {x)
+
+
OiaVaOe)
a 22 y> 2 (x)
and the eigenvalues
In
but exceptional cases it will turn out that El = E\, i.e., the perturbaremoves the degeneracy. The eigenfunctions y\ and yfy are in error by
all
tion
first powers of small quantities (the a nl for n = 1 2 and / = 3, 4, 5, ...
which have been neglected), but the eigenvalues E\ and E% are in error only
by the squares of small quantities. Since it is usually necessary only to
know the eigenvalues, the calculation is often completed when E\ and E%
are obtained. However, if it is also necessary to know the eigenfunctions,
we may take the set of non-degenerate eigenfunctions and eigenvalues
the
yRx), y&r), rp%x),
rp\(x),
po pO pO pO
^"2' ^3' ^4)
^1'
where
Vh( x)
y/fcx)
=
=
x)
and
E%
f^x)
and
E\
V>3(
=E
= 4
and apply the normal non-degenerate perturbation theory to obtain the
set of eigenfunctions and eigenvalues
ViO0>
V2O), Vs(aD, viC*).
t'Xi
^2 "3> ^4
El
and
E'2
>
where
E[
El
PERTURBATION THEORY
280
All these eigenfunctions
and eigenvalues are
in error only
[Ch. 9
by the square of
a small quantity.
Now
us study the calculation which evaluates
let
Assume
that y>i(aO
the other
and
ip3 (x), y> (#), y> (x)
4
5
y>, ip 2 ,
E\, and E%.
are orthogonal to each other, as well as to
y> 2 (x)
.
Assume
.f
all
also that all the eigenfunctions
have been normalized. Then it is apparent that we may write an equation
identical with (9-7), but E'n replaced by E%. That is,
a nm\E n
Em =
)
2, a nl v ml
I
Now
set all
a nl
obtained, one for
except for n
m=
a n i(K
= 2
=
Ei)
a^iK - E2) =
2 and
1,
m=
and one for
Two
1,2.
equations are
2:
a nl v u
1,
1,2
n=
a n iVn,
1,2
1=1,2
Evaluating the summations explicitly, these equations give
-E) =
n=l,2
(9-21)
+ a n2 v12
(9-22)
n = 1, 2
a n2 (En - E) = a nl v 21 + a n2 v 22
where we have set E = E2 = E. Write these equations in the form
n = 1, 2
(El - E - v n )a nl + (-v 12 )a n2 = 0,
n = 1, 2
(-%)m + (E- E - v22)a n2 = 0,
"ni(En
a nl uu
Then
it
becomes apparent that they constitute a pair of homogeneous
unknowns a nl and a n2 In order that a solution
linear equations in the
such a set of equations, it is necessary that the determinant of the
coefficients of the unknowns be equal to zero. That is, we must have
exist for
- Vl2
l(K-E- Vll)
\
-v21
(En
Evaluating the determinant,
(
- E - vn)(En
we
= h2
--
(9 _23)
v 22 )J
find
-E- v
22 )
v 21 v 12
0,
= 1, 2 (9-24)
v n = v 22 and that
n
In many cases of physical interest it turns out that
v 12 and t>2i are real. To simplify the discussion, we shall assume that both
hold true. Since the definition (9-6) shows that in all cases v 21 = yf2
we shall then have v21 = v12 as well as vn = v22 and equation (9-24)
,
becomes
(En
t
-E- vnf -v\
The footnote on page 189 proves
that this
= 0,
is
possible.
1,
THE TREATMENT OF DEGENERACIES
Sec. 4]
This
a quadratic in
is
2s.
one for each value of
the
sign for n
has two solutions,
It
- E - vn = v12
En
n.
281
We
1,
arbitrarily take the
2
sign for
and
Then
2.
= E + v11 + va
E% = E + vn vXi
Now consider equation (9-21) with n = 1. It is
E1
ii(-E?
Substitute
-E) = a n vu +
E\ from (9-25) and
fliiO>u
transpose.
iz
u ii)
au
a 12
o 12 i;12
The
(9-25)
result is
iai2
Therefore
Thus the perturbed eigenfunction
eigenvalue E\,
is
= a u \%px{x) +
xp\(x)
The
coefficient
ip 2 {x)~\
an must be adjusted to normalize
that this requires an
ip%x).
It is
easy to see
1/V2. Then
>?(*)
and we
(9-18), corresponding to perturbed
equal to
= -4 [>i(*) +
%(*)]
(9-26)
also have
= E+v11 + vu
Consider equation (9-21) with n = 2. It is
a 2i( 2 E) = a 21 u + a 22
E1
t>
Evaluating
fijj,
(9-260
t;
12
we have
a 2l(>ll
^12
V ll)
"22^12
So
a 21 =
a 22
and
Normalization gives
#*)
and we
= -4 Oi(*> -
V2 (*)]
(9-27)
also have
Ez
= E + vn -
i>
12
(9-27')
PERTURBATION THEORY
282
Except for the unusual case in which the matrix element
i>
12 is
[Ch. 9
equal to
and (9-27') show that the perturbation v{x) has
removed the degeneracy. The two perturbed eigenvalues E\ and E\ are
separated by the energy
zero, equations (9-26')
E\-El =
Then
y>i(x), tp%(x),
tions
y3 (x),
2v 12
(9-28)
E\ and E\, plus the original non-degenerate eigenfuncand eigenvalues E3 Et
provide the completely
non-degenerate set of eigenfunctions and eigenvalues (9-19) to which we
can apply the normal perturbation theory and obtain (9-20). Note that
equations (9-26) and (9-27) certainly confirm our suspicion that even an
y>i(pc),
arbitrarily small perturbation will
produce a large mixing of the degenerate
eigenfunctions fi{x) and y> 2 (x), because both y>%x) and y>$(x) contain
equal amounts of these eigenfunctions, independent of the magnitude
of
v(x).
For a case in which v12
v 21 and/or v n ^ v 22 the procedure is
the same in essence but the details are more complicated. For a case in
which there are a. degenerate eigenfunctions, the procedure is the same
except that equation (9-23) becomes a determinant of order <x. This leads
to an equation analogous to (9-24) of order a in E%- For large values of a,
such an equation can be very difficult to solve, but it is often possible to
perform certain manipulations on the determinant so that the equation for
E comes to a form for which its roots are easy to find.
Furthermore,
it is
often possible to eliminate the entire procedure of
degenerate perturbation theory, and use instead non-degenerate perturbation theory, if physical arguments can be used to find the particular linear
combinations of degenerate eigenfunctions that do not mix under the
Consider the problem that we have been
and assume that y> and xp\ are known a priori. Then E\ and E\,
which are usually what we are really interested in finding, can be evaluated
simply by applying the equation (9-13) of non-degenerate perturbation
theory. That is, equation (9-13) applied to ip gives
application of the perturbation.
treating,
El
- E = JV vfl dx = j^= [ft +
=
f 11
y>*t>yi
*>12
dx
f*vyj 2
Wtlv
dx
J|
[% + %]
y*fVi dx
dx
y)*c^ dx
THE TREATMENT OF DEGENERACIES
Sec. 4]
283
and the same equation applied
in agreement with equations (9-25),
to
ft g ives
E2
-E = vSVS dx =
J
J
=
f*vrp x dx
= iKl - 12 = v n - v 12
-j= [y*
- y*> -1= [vi - vJ
f*vy 2 dx
21
ip* v/,Pi
dx
d*
ip*vrp 2
dx
U 2 2]
also in agreement with equations (9-25).
Non-degenerate perturbation theory can be applied to the particular
linear combinations yl and y>1 of the degenerate eigenfunctions y> x and
f2 because those particular combinations are not mixed by the application
of the perturbation. We can see this by evaluating the integrals that appear
in the coefficients which, according to the equation (9-14) of non-degenerate
,
perturbation theory, determine the mixing.
rpfvxpl
dx
Of +
i[^ii
wt vxPi dx
"is
y>f\v -j= |>i
ip*vrp 2
"21
dx
For
instance,
V2]
dx
y*vfi dx
y>*vip 2
dx
^22]
Similarly,
[xpfvipl
dx
So the perturbation does not mix y\ and
tion theory can
=
ip
2,
and non-degenerate perturba-
be applied directly to these particular linear combinations,
even though they are degenerate before the application of the perturbation.
In subsequent chapters we shall consider a number of examples of the
physical arguments which are used in
quantum mechanical systems
to find
the particular linear combinations of degenerate eigenfunctions that are
not mixed by the application of a perturbation. (Cf. Chapters 11, 12, and
13.) We consider here an example of a physical argument which can be
used in a classical system. In one of the higher frequency modes of a
circular drum head, the drum head vibrates with a nodal line lying along
a diameter. This mode is degenerate because the same frequency is
obtained for all orientations of the nodal line. But there is only a twofold degeneracy because there are only
vibrations whose nodal
lines
are
two independent vibrations
perpendicular.
the
These independent
PERTURBATION THEORY
284
[Ch. 9
degenerate vibrations are indicated in figure (9-5). All other vibrations
at this frequency can be obtained by linear combinations of these two.
In particular, other sets of two independent degenerate vibrations, with
perpendicular nodal lines of different orientation, can be obtained by
appropriate linear combinations. In the absence of a perturbation, all
these sets are equivalent.
Node
Node
Node
Node
Two
Figure 9-5.
independent degenerate vibrations of a circular drum head.
Figure 9-6. Applying a perturbation to a circular
Figure 9-7.
Now
The
drum
head.
results of applying a perturbation to a circular
drum
head.
imagine applying a perturbation by fixing a small weight to the
at some position other than its center, as indicated in figure
drum head
(9-6).
Because of the asymmetry introduced by the perturbation, the two
new
The forms of these new vibrations, as well as their frequencies,
can be found by a calculation using the procedure of degenerate classical
perturbation theory. The vibrations which are found are indicated in
previously independent vibrations are mixed together to form two
vibrations.
figure (9-7).
It is also
found that the perturbation removes the degeneracy
because the weight lies along the nodal line for one vibration and therefore
has no effect on the frequency of that vibration, while it does effect the
frequency of the other vibration.
TIME DEPENDENT PERTURBATION
Sec. 5]
THEORY
285
After gaining some experience with these problems, it is possible to tell
from a priori physical arguments what the form of the perturbed vibrations
must be. This makes it possible to eliminate the complicated procedure of
degenerate classical perturbation theory by taking the set of independent
degenerate unperturbed vibrations to be the particular set for which one
nodal line runs through the weight. Then the application of the perturbation does not mix the vibrations because they have the same form both
before and after its application, and the non-degenerate classical perturbation theory can be used. This considerably simplifies the calculation of the
frequency
shift
produced by the perturbation, which
is
the quantity that
is
usually of interest.
5.
Time Dependent
Perturbation Theory
We now develop a theory for treating perturbations which are functions
of both position and time. This is an important case for several reasons,
one being that time dependent perturbation theory provides the only useful
method
for solving the Schroedinger equation for a time dependent
r).f Thus we consider a time dependent potential
which can be decomposed as follows
potential V(x,
V'(x,
where V(x)
is
i)
V(x)
equation for V(x) are the
set
The
t)
(9-29)
v(x, i)
a time independent unperturbed potential and
small time dependent perturbation.
V'(x,
v(x,
t) is
solutions to the Schroedinger
of unperturbed wave functions
Yn(x, = e~ m^ Wn{x)
i)
(9-30)
where the En and ip n (x) are the unperturbed eigenvalues and eigenfunctions.
Assume that a solution to the Schroedinger equation for V'(x, t) can be
written
= 2>(01 (*,
r
'H*.
(9-31)
where the coefficients aJJ) are functions of time. Different solutions will
have different sets of coefficients, but here we shall not use a second subscript to indicate this explicitly.
02xp
2M
dx%
fixir,
+ V'W
=
-ih
dt
exception is a time dependent potential of the form V(x, t) = Vt(x) + F2 (f).
form only, the Schroedinger equation can be separated in the manner of section
Chapter 7, by assuming a solution 'Ffo t) = y>(x) <j>(t).
t
One
For
this
4,
fc2
Substitute (9-31) into the equation
PERTURBATION THEORY
286
which
it is
supposed to
[Ch. 9
This gives
satisfy.
"
2, a n
9 F
3T
2m
ax2
at
2v
The bracket vanishes because
complex
conjugate
T m = e~
lEmt/n
ip
of some
m and
,
the
equation for the potential V.
at
are solutions to the Schroedinger
Multiply the remaining terms by the
wave function
unperturbed
particular
Then, evaluating the Y, we
integrate over all x.
have
J oo
Since the
y> n
are orthogonal
J go
ut
and normalized,
= - 7 2 "-(^-^--^""fm.
(9-32)
where we have extended the definition of the matrix element v mn (equation
9-6) to include time dependent perturbations. This is a set of coupled
first order ordinary differential equations, one for each m, which determine
the a n (t). The details of the solution of these equations depend on the
details of the particular problem at hand. We consider here a simple but
illustrative case.
Assume a perturbation of the form
v(x,
t)
t<0
>
0,
v(x),
This
is
a perturbation v(x) which
is
(9-33)
"switched on" at
0.
For
this
case the set of unperturbed wave functions (9-30) are exact solutions for
<
t
0. Next assume that the wave function for the particle is known to
be equal to a single one of these wave functions, say
k (x, t), for t < 0.
This amounts to assuming that the total energy of the particle is known to
be precisely Ek for t < 0. This does not conflict with the uncertainty
principle
At AE >
because in the
infinite
time before
h
it
(9-34)
would be possible
to
measure
In terms of equation
(9-31), this assumption provides the following set of initial conditions for
the energy of the particle with perfect precision.
the a(t) att
0.
.(0)
0,
t
1,
n^k
n
(9-35)
,
THEORY
TIME DEPENDENT PERTURBATION
Sec. 5]
[We assume that the ajf) do not change discontinuously at t
assumption will be justified by the results of the calculation.]
like to find the perturbed
wave function T'(z,
t)
287
0.
We
This
would
for the particle at a time
a n (t) for t > 0.
Let us require that the perturbation v(x) be small enough, or that the
time t be short enough, that
t
>
0.
To do
this
we
shall evaluate the
a n (t)
Then we may
for n
k.
neglect
all
< 1,
^ k,
1,
>
(9-36)
k,
terms in the right side of equation (9-32) except
This gives
~ --a
daji)
-i(E h
k (t)e
-Em)tlh V
(9-37)
dt
To
m=
evaluate ak (t), set
Then
k.
<tafc(0
dt
a*(0t>**
or
da k(t)
fl*(0
Integrate both sides from
independent of
for
>
to
t'
>
all
- tW
In a k (t)
which
remembering that the v mle are
0,
This gives
0.
is
a k (t')
In
^ - 7 Vat',
According to (9-35), ak (0)
=1. So we
a k (t)
>0
_%(0)J
~ e - iVatin
find
t
>
(9-38)
where we have dropped the primes to simplify the notation.
Next evaluate the ajf), n = k, by setting m = n in (9-37) and by making
the additional approximation that ak {i) = 1. We have
da(t)
_ _ 1 e -nB -B.)tiH v
t
dt
n=k
or
da n(t)
~ - - v nk e- i(E*- E wn dt,
n^k
PERTURBATION THEORY
288
Integrate
from
to
>
t'
to obtain
-i(.E k -E)t/h
(0
Ejs.
From
(9-35),
an (0)
0,
[Ch. 9
so
y-nEh -Bn )tm _
a n (t)
in>
B _^ k
(9-39)
-C<i
&Tf.
where we have again dropped the primes.
(9-31), (9-38), and (9-39), we find
+ 1 E.-E,
[e
Evaluating T'(,
-X>-BMH_
l]g-./ y^a;)
t)
from
(9_4Q)
Note that the energy Ek + vkk appearing in the exponential of the first
term is exactly the perturbed energy E'k = Ek + vkk which would be
predicted for a completely time independent perturbation equal to v(x).
It is of interest to consider the quantity a*{i) a n {t). This real function of
Multiplying
t is the square of the magnitude of the coefficient ajf).
conjugate,
we find
(9-39) into its complex
,
sin
]%Lp*
2h
al(t)a n {t)~?*&*-
(9-41)
&
(^J
This quantity oscillates in time between zero and ^v*k v nlc \{En
with frequency v
sin 2 [(En
fixed t.
(En
Ek)/h. We plot in figure
- Ek)j2hf as a function of
Ek)2
(9-8) the factor
- Ek)tl2h]l[(En
(En - Ek)\2h for
Now the wave function describing the particle initially contained
Wk
(x, t). The perturbation v(x, t) has
from other eigenstates over a whole
range of the quantum number n. However, we see that the most important
contributions come from those n which correspond to eigenvalues En
lying within a range centered about Ek and of width AE, where
only the eigenstate wave function
the effect of mixing in contributions
AE/in
it it
or
AE~2nhlt
In section
instant
t is
8,
Chapter
7,
we showed
(9-42)
that the value of a*(t) a(t) at
any
equal to the probability of finding the particle in the eigenstate
Sec. 5]
TIME DEPENDENT PERTURBATION
THEORY
289
n at that instant. (The argument actually considered the a n to be constants,
but inspection will demonstrate that it is equally valid when the a n are
functions of t .) Thus at any time t there is a certain probability of finding
the particle in final eigenstate n which is different from the initial eigenstate
k, and with total energy E different from the initial total energy Ek This
appears to be a violation of the law of conservation of energy by an
amount En Ek which may be large compared to the energy vkk supplied
.
2ir
IT
3t
En~ Ek
2%
Figure 9-8.
The
plot of a function
which
by the perturbation. However,
arises in
time dependent perturbation theory.
in the time interval
to
the probability
of finding the particle with energy En is important only when EH Ek is
at most equal to about A.E, where t and A.E are related by equation (9-42).
According to the uncertainty principle (9-34) any measurement of the total
energy of the particle which
uncertain by an
A.E. This
amount of
removes the
is
must be
which is comparable to
and provides an example of this statement
carried out in this time interval
the order of
difficulty
hjt,
of the uncertainty principle.
Consider equation (9-41) for small values of
>
0.
The equation
says that the probability of finding the particle in a particular eigenstate
is
proportional to the square of
linear dependence
t.
This statement
is
in contrast to the
that might be expected intuitively.
However,
always based on our experience with systems in the
In that limit the resolution of any experimental apparatus
physical intuition
classical limit.
on
is
PERTURBATION THEORY
290
[Ch. 9
so large compared to the separation of the eigenvalues, or even to the
width of the range AE, that it is not possible to measure a*(t) a n (t) for a
single value of n. All that can be measured classically is the total probability of finding that the particle has made a transition from the initial
is
eigenstate
to
some other
the transition probability
which
Pk =
To
evaluate this,
we assume
is
dNn
dNn
dEn
per energy interval
that" there are
is
express this in terms of
defined as
2n *(0 (0
spaced final eigenstates in the range
Then
We
final eigenstate n.
Pk
AE;
the
(9-43)
a large number of closely
number of final eigenstates
fc
a*n (t)a n (t)dN n
=\
J -oo
75ifJ-
00
Owing
{t)^dE n
J -oo
Evaluating a*(t) a n (t) from (9-41),
ft
at{i)a n
dhi n
we have
oo
-^
r( K
sin"
*
V nk v nkPn-
2H
dE r
(^J
to the factor sin 2 [(En
tribution to the integral
dNjdEn .
integral over
the density of final states p n
summation over n can be approximated by an
That is,
the
- Ek)tj2h]j[(En - Ek)j2hf, most of the con-
comes from the range AE.
If
we assume
that the
matrix element v nk and the density of final states p n are both slowly varying
functions of n in that range, we can write
~ir
Pn
i-
The quantum number n now
dE n
2h
refers to
00
V n!e V nlc
j f= f
a typical final eigenstate in the
initial eigenstate k.
X - eA
2h
(
neighborhood of the
- Ek
\E n
sin"
Let
sin
Z=
Z
(En
Ek)tj2fr;
then
dZ
which gives
D
**
^n
7r
"7" v nk v nkPrf
(9-44)
EXERCISES
291
Ch.
9]
The
transition probability
rate
Rk =
dPjdt
is
proportional to
independent of
is
Rk
t,
t,
as expected.
The
transition
since
~T V*nlPnlcPn
(9-45)
This important formula
is
often called Golden Rule No. 2.
It is
very
of very general applicability.! In any situation
in which transitions are made to an essentially continuous range of final
states under the influence of a constant perturbation, the transition rate
widely used because
it is
can be evaluated from this formula. Note that we have here a good example
of the use of quantum mechanics in the evaluation of transition rates.
The ability to do this is one of its most important advantages over the
quantum
old
theory.
BIBLIOGRAPHY
Bohm,
D.,
Quantum Theory, Prentice-Hall, Englewood Cliffs, N.J., 1951.
and E. B. Wilson, Introduction to Quantum Mechanics, McGraw-
Pauling, L.,
New York, 1935.
Quantum" Mechanics, McGraw-Hill Book Co., New York, 1955.
Sherwin, C. W., Introduction to Quantum Mechanics, Henry Holt and Co.,
New York, 1959.
Hill
Book
Co.,
Schiff, L. I.,
EXERCISES
1.
Use perturbation theory
eigenfunction
y> x (x)
to calculate the first eigenvalue
E1
and the
first
of the potential
oo,
V(x)
<
-a/2
a/2, x
<
<
>
a/2
a/2
a/2
where
8 is small relative to
calculation of section
Ev
Compare with
the results obtained in the
3.
Use perturbation theory to calculate the first eigenvalue Ex for the potential
4, Chapter 7. Compare with the results of the numerical integration
performed in that exercise, and also with the results of the analytical treatment
2.
of exercise
of exercise
6,
Chapter
8.
Except for certain pathological cases, no degeneracies arise in problems
involving one particle moving in one dimension. In order to obtain a simple
3.
We shall use it in Chapters 1 5 and 1 6, and use a closely related formula in Chapter 1 3
PERTURBATION THEORY
292
[Ch. 9
example of the application of degenerate perturbation theory, consider one
particle
moving
in the
V{x,y)
T
two dimensional
x
<
>
-a/2, x
-a/2
0,
infinite
<
<
square well potential
a/2,
<
-a/2
a/2,
-a/2, y
<
<
>
a/2
a/2
first section of the next chapter, set up the time independent
Schroedinger equation for this potential. Separate this partial differential
After consulting the
equation into two ordinary differential equations by the usual method, making
use of the fact that V(x, y) can be written V(x) + V(y). Since these equations,
and the conditions on y> at the edges of the well, have the same form as for a
one dimensional infinite square well, their solutions can be written immediately.
Note
that there are degeneracies in almost
all
the eigenstates.
Now
consider
the application of the perturbation
Make
effect
>
d,
0,
otherwise
a/4,
>
a/2,
a routine degenerate perturbation theory calculation to investigate the
first two degenerate eigenstates.
of this perturbation on the
4. Take the linear combinations of the unperturbed degenerate eigenfunctions
found in exercise 3, and show that if these linear combinations are used in
non-degenerate perturbation theory the energy shifts which are obtained are
the same as those obtained in that exercise. Show also that in non-degenerate
perturbation theory these linear combinations of the unperturbed eigenfunctions
are not mixed by the perturbation, even though they are degenerate before the
perturbation
is
<
0,
At
5.
dimensional
a/2.
At
applied.
an electron
infinite
t
0,
is
known
to
be
in the n
eigenstate of a
one
square well potential which extends from x = a/2 to
a uniform electric field E is applied in the direction of
The
electric field is left on for a short time t and then removed.
Use time dependent perturbation theory to calculate the probability that the
increasing x.
electron will be in the n
2, 3,
probabilities as a function of
used.
t.
or 4 eigenstates at
Hint :
Some
>
t.
Make
plots of these
of the results of exercise
can be
CHAPTER
10
One-Electron
Atoms
I.
Quantum Mechanics for Several Dimensions
and Several Particles
Up to this point we have restricted our treatment of quantum meqhanics
to a single particle
moving
in a one dimensional space.
tion for the discussion of atoms,
we must
Now,
in prepara-
generalize to the case of a system
of particles moving in a three dimensional space.
Let us consider
N particles
rectangular coordinates, x, y,
equation for this case, using the
whose positions are defined by a
z.
We
first
set
of
develop the Schroedinger
same technique as
in section 8,
Chapter
7.
We write the classical expression defining the total energy of the system:
+ Pi + p!)
i^-iPi
2m
+ V(x y %,..., x
i=i
x,
x,
},
y s , z jt ...,x N y N
,
t)
=E
(10-1)
The quantity^ is the x component of the momentum of the/th particle,
py and pZj are its y and z components, and rrij is the mass of that
particle. Thus the/th term of the summation is the kinetic energy of the
y'th particle, and the value of the sum is the kinetic energy of the system.
The potential energy of the system depends, in general, on the coordinates
of each particle, as well as on the time t. The quantities x jt y } z j are the
,
three coordinates of the
quantities
(cf.
px
7-77).
i=i
,pz
y'th particle.
and E, by
Now we
replace the dynamical
their associated differential operators
This gives*
+ -^
+ I +
d
dzy
2mAdx*
y)
293
V(x,,
y x ...,z N ,i)
}
ihdt
(10-2)}
v
ONE-ELECTRON ATOMS
294
[Ch. 10
Multiplying this operator equation into the equality
Y(xly ...,z N ,t)=
we obtain
(*!, ...,z N ,t)
(10-3)
the Schroedinger equation for the system,
2 3=i
v;y + vv = m
(io-4)
9(
2m.,-
where we use the symbol
V ? s ll + |l + Jl
3x^
a^ 9z^
which
( 10
-5)
called the Laplacian operator in rectangular coordinates of the
is
y'th particle.
The wave function
3iV~
variables
the time.
The
T^,
is
a function of
3 spatial coordinates for each of the N
particles, plus
for the system,
significance of this
wave function
t),
specified
is
by a natural
For
extension of the interpretive postulates (7-21), (7-31), and (7-80).
instance,
If,
we postulate
at the instant
that
a measurement
t,
is
made
to locate the particles in the
z N t), the probability
system associated with the wave function ^F^i,
x
P(xlt
,z N t) dxt
dzN that the
coordinate of the first particle will
be found between x 1 and x x + dx x etc., and finally that the z coordinate
.
of the Nth particle
P{xx
will
..,z N ,t)dx 1
be found between zN and
N+
dz N ,
is
dz N
= T*^,
,z N ,i)
T(xx
dxx
dz N
and
extensions
the
obtained
by
obvious
of
wave
function
be
the
(10-3) can
properties developed in Chapters 7, 8, and 9. As an example, it is easy to
show by the usual method that, when the potential V does not depend on
time, there are solutions to the Schroedinger equation which can be
Essentially all the properties of the Schroedinger equation (10-4)
written
Y(xx ...,z N ,t)=
,
xp( Xl ,
...,z N)
<Kt)
(10-6)
where
tft)
and where y(x2
z ) is
equation
e~ imin
a solution to the time independent Schroedinger
2 - 2m, V;fy +
3=i
(10-7)
Vy>
Erp
(10-8)
THE ONE-ELECTRON ATOM
Sec. 2]
295
Note that the time independent Schroedinger equation
case
we
are
now
wave function Yfo,
z N i) is denned
more than three dimensions. This makes it
the
function
is
in the general
is,
Also note that
discussing, a partial differential equation.
.
in a non-physical space of
quite clear that the
wave
only a computational device.
The One-Electron Atom
2.
The procedures involved in solving the time independent Schroedinger
equation (10-8) are best illustrated by taking a specific example. In this
we
our treatment of one of the simplest the onemost interesting example because
of the unique importance of the one-electron atom in atomic theory.
section
shall begin
electron atom.
The reader
This
is
also perhaps the
will recall that
feature of the old
Bohr's one-electron atom
quantum
We
theory.
electron atom is equally important in
Above and beyond its own intrinsic
system
is
fundamental because
mechanical treatment of
all
it
is
the
most
shall see that the role
the theory of
interest, the
significant
of the one-
quantum mechanics.
theory of this simple
forms the foundation of the quantum
multi-electron atoms.
The one-electron atom is a three dimensional system containing two
a nucleus and an electron, moving under the influence of a mutual
Coulomb attraction and bound together by that attraction. Let the mass
of the nucleus be m 1 and its charge be +Ze (Z = 1 for neutral hydrogen,
Z = 2 for singly ionized helium, etc.). Designate the mass of the electron
by w 2 and its charge by e. Then the (Coulomb) potential energy of the
particles,
system
is
V(xlt
=.
z2 )
Ze2
V(z 2
where xlt yx
zx
xxf
(y2
Vl f
(z 2
-zf
(10-9)
are the rectangular coordinates of the nucleus
and
.r
2,
y 2 z 2 are the rectangular coordinates of the electron. The denominator
of equation (10-9) is the distance between the two particles. The time
,
independent Schroedinger equation for
/9
2m x \
yj T
dx\
y>T
dy\
9Vr\
dz\ /
this case is
d ipi)
dx\
dy\2
d4-
V{xx ...,
z 2 ) VT
2m 2
d fi
ft
(9
v
fi
= E T yr
The reason for using the subscript T will become apparent
By making a change of variables, this equation goes over
(10-10)
later.
into a
form
in
ONE-ELECTRON ATOMS
296
[Ch. 10
can be separated into two parts. One part represents the transmotion of the center of mass of the system, and the other part
represents the motion of the two particles relative to each other. The
which
it
lational
new
variables are
= m1x + m
1
x2
= m 1 y1 + m z y2
m1 + m
(10-11)
_ m xzx + m 2z
m1 + m2
*-y
Figure 10 I.
The one-electron atom described
in
rectangular and spherical coordinates.
the rectangular coordinates of the center of
mass of the system, and
r,0,4>
the spherical coordinates of the electron relative to the nucleus.
These coordinates are
easy to see that
r, 8,
<f>
illustrated in figure (10-1).
are related to xx
r sin 6
cos
r sin 6 sin
r
cos 6
d>
<f>
z2
#2
/ 2
zx
z2
From
this figure it is
by the equations
xx
yx
(10-12)
Using equations (10-11) and (10-12), equation (10-10) can be transformed
THE ONE-ELECTRON ATOM
Sec. 2]
to the
new
This
variables.
is
2( mi
dy?
3f
fi
+ m %)\3x2
actually quite a tedious task,
3z
dz
'
T^ |5 + -T^r-M**
r
sin
3<
30
r sin
where
and so we
They are
present here only the results.
(d
297
J
/
+V
I?))
30/1
= _2MbL_
m +m
lu, \r
2/*
a nd
F(r)
dr
dr I
- ***
(10-13)
= -
The quantity jM is the reduced mass.
The separation of this time independent Schroedinger equation
effected
by
looking for solutions of the
ip
T (x, y,
z, r, 0,
<j>)
y>
Substitute this into equation (10-13),
is
form
CM (x,
y, z) ip{r, 0,
(10-14)
<f>)
and divide through by Vt
VcuV-
This gives
h
[_
L
2(m,
pVcM + g>cM + 3>cm V
2
2
I.
w OM \ dx2
m,)
3w
2
2/if\r dr\
dz
dr'
'
3> +
H
r
sin
3<f>
dfp\\,
- }
sin a
-r
2
36/
r sin 6 36\
3 (
The first term on the left is a function only of x,
ET
variables
r, 0,
constant.
y, z
V(r)
JBi
and the second term is
sum
of the two terms is equal to a
even though the variables x, y, z are independent of the
it is clear that each term must separately be equal to a
(f>,
a function only of
constant
r, 0,
Since the
<f>.
Thus we obtain two equations,
2(m x
+ m )y> cu \
dx2
By 2
dz
and
2i*y>\i*dr\
3r/
sin
6 3 </> 2
sin636\
where
-Ecm
+ E = ET
3d!)
ONE-ELECTRON ATOMS
298
Consider the
first
+m
V
2)
5ck
write
ycM
2(m x
we
equation, which
3y
[Ch. 10
"
ycM
dz
5
mass
whose
coordinates
are
those
of
the
center
of
mass
of
the
atom,
2>
i
and which is moving with total energy ECM in a region of zero potential.
The equation describes the translational motion of the atom. Consider
next the second equation, which we write
This
is
the time independent Schroedinger equation for a particle of
m +w
r
2
dr\
2fj.\r
Br)
sin
(sin
The curly bracket is V 2 ^j, where V
r, 8,
2
B<f>
sineae\
coordinates
is
This equation
<f>.
0^)1+
del)
= Ew
V(r)w
y,v
(10-15)
y
'
the Laplacian operator in the spherical
is
the time independent Schroedinger
equation describing the motion of the electron relative to the nucleus.
E = ET ECM where ET is the total energy of the atom and
CM is its energy of translational motion, E is the total energy of relative
motion that is, the total energy of the atom in a frame of reference in
which the center of mass is stationary. Equation (10-15) is formally
Since
"
an electron
identical with the time independent Schroedinger equation for
of mass
[i,
infinitely
instead of mass
x,
moving with coordinates
r, 8,
<f>
about an
massive nucleus which, consequently, remains stationary
section 4, Chapter
It is
(cf.
5).
apparent that in equation (10-14) we have decomposed the total
eigenfunction ^t mto tne product of an eigenfunction y CM specifying
the motion of the center of mass of the atom, times an eigenfunction y>,
,
specifying the
position
is
motion of the electron
of one particle relative
3.
relative to the nucleus.
possible whenever the potential
V depends only on
This decom-
the coordinates
to the other.
Separation and Solution of the Equation of Relative Motion
The
translational
motion of the atom
Consequently, we shall assume that
we
is
of no interest to us here.
atom from a frame
Then we need consider
are viewing the
of reference in which the center of mass
is
at rest.
only the time independent Schroedinger equation (10-15).
V = Ze2 jr
of a
Coulomb
any function of the radial coordinate
is
potential
For the case
and, in fact, whenever the potential
r only,
it is
possible to separate
THE EQUATION OF RELATIVE MOTION
Sec. 3]
299
To do
the equation into three ordinary differential equations.
for solutions of the
this,
we look
form
y{r, 6,
</>)
(10-16)
J?(r)0(0)O>(</>)
and dividing through by
Inserting this into equation (10-15)
R@Q>h
l2fx,
we obtain
,
l
r
r
Rdr\
>/M
J?\
d 2 <t>
sin
0<P d<4
sin 6
00 d6\
sin
d&\
del
+ |[^-nr)] =
Next multiply through by
l#<b
dd>
sin
=
~
d d /
sin 2 d
dR\
and transpose. This gives
d
sin 6
d0\
/
fl
dr\
dBJ
dd\
dr)
sin
0[E
F(r)]
and the right side only on r and 6,
Since the left side depends only on
both must equal a constant, which we shall call m 2 Thus we obtain two
<f>
equations,
= -m
d<
(10-17)
<P
and
1
dUdR \
sin
we
transposing,
l(
Rdr\
d0\
sin00d0\
By
dl.
d_R\
dr)
2^ 2rE
- r\E
h
</0/
.-.
F(r)J
2
m
sn 2
rewrite the latter as
=Jn^
2j^
h
sin
1_ .d
d&\
sin00d0\
dO
According to the usual argument, both sides of the equation must equal
a constant a. Consequently,
1
d (
sin a
.
sin0d0\
d@\
dO/
m 2 = n
a0
2
(10-18)
sin
and
2
+
-M*t)
r* dr\
dr I
The problem has been reduced
%n
-v
^R -
4
r
<
io - i9 >
to that of solving the three ordinary
differential equations (10-17), (10-18),
and
(10-19).
We
shall find that
ONE-ELECTRON ATOMS
300
[Ch. 10
equation (10-17) has acceptable solutions only for certain values of m.
Using these values of m in equation (10-18),
turns out that the equation
it
a. With these values of
a in (10-19), the equation is found to have acceptable solutions only for
certain values of the total energy E. Thus the energy is quantized.
Consider equation (10-17). The reader can easily verify that it has a
has acceptable solutions only for certain values of
solution
O(^)
Now we
must, for the
im *
time, explicitly consider the requirement that
first
the probability density
and probability
flux
have
single-valued, for all values of their variables.
definite values,
i.e.,
be
This must be the case
because they are measurable quantities describing the behavior of a real
We may ensure that it will be so by demanding that the
must be single-valued. This requires the function $(<)
to be single-valued, and it must be considered explicitly because the
azimuthal angles and 2v are actually the same angle. Thus we must be
physical particle.
eigenfunction
ip
sure that
<D(0)
This
= 0(2tt)
is
or
1
This
is
true only
cos miTT
|m|
that
is,
if
sin m2rr
if
0, 1, 2, 3,
has integral values.f
Thus the
(10-20)
set
of functions which are
acceptable solutions to equation (10-17) are
<D m
where
is
m assumes the values
(# = e**
specified
by
(10-20).
(10-21)
The quantum number
used as a subscript to identify the form of a particular solution.
In working with equation (10-18), it is convenient to change variables
from
8 to
where
I
t In
a more rigorous treatment
it is
cos
(10-22)
not necessary to demand that y be single-valued,
but only that the measurable quantity v>*V be single-valued. It is evident that this would
admit either integral or half-integral values of m. However, in the present context the
half-integral values can be ruled out because they lead to solutions of the 0(0) equation
On the other hand,
a quantum corresponding to m do arise in connection with
functions, related to the <b(<f>), that are used to describe "spin." This will be discussed
for which the probability flux has a physically impossible behavior.
half-integral values of
in the next chapter.
THE EQUATION OF RELATIVE MOTION
Sec. 3]
Then
301
the equation reads
+
a-a^)
m
d_
This differential equation
m
-
0(f)
(10-23)
e)
not directly solvable by the power series
relation involves more than two terms
is
method because the recursion
(cf.
But, by considering the behavior of the differ1, the extreme values of the variable, it is found
solution can be obtained for the function F() denned by
= (1 - 2 W/2 F()
section 6, Chapter
that a
power series
8).
ential equation at f
0(f)
After solving the differential equation for F(f),
found that solutions
it is
for 0(f), which have the acceptable behavior of being everywhere finite,
exist only when F(g) is a polynomial. This happens only when a has the
value
where
/ is
1(1
1)
1,
\m\
2, |m|
(10-24)
one of the integers
I
\m\, \m\
3,
(10-25)
an acceptable
Corresponding to each value of the quantum number I there is
solution
To
.Fi(f).
identify the particular solutions to equation (10-18),
which are written
*,(!)
it is
= (l-a W/2 ^(D
(10-26)
quantum numbers
necessary to use as a label the two
and m. The
functions 0[ m (f) are known in the mathematical literature as associated
Legendre functions. Some examples of these functions will be presented in
section
5.
Setting a
\_d_
r
/(/
1) in
^^ +
equation (10-19),
{_l(Ll)
;
dr^
In terms of the
new
we have
F(r)]
i?
(10-27)
variable
(10-28)
2/Sr
where
2
fl2
/3
and
_
=-_ IjxE
(10-29)
also using
= fiZe
(10-30)
h2p
the equation becomes
dR(p) \
I U im)
+
t
2
p dp\
dp
L i _ Sl+Ji + rU) ,
14
p)
awl)
ONE-ELECTRON ATOMS
302
[Ch. 10
Again it turns out that the differential equation is not directly amenable
to a power series solution. However, by considering the behavior of the
equation for p - oo, it is found that a series solution can be obtained for
the function G(p) defined by
R( P)
e- pl2 p G(p)
l
In order that G(p) be a polynomial, which
from diverging
where n
is
as p -- oo,
it is
is
necessary to prevent R(p)
found that y must have the value
1./
2,
(10-32)
one of the integers
n
The functions
= l+
/+
3,'...
(10-33)
that are acceptable solutions to equation (10-19),
which are
written
Rnlp)
e- pl2 p
G n (p)
carry as labels the quantum number n and the
(10-34)
quantum number /. Examples
of these associated Laguerre functions are given in section
5.|
The reader will note that the procedures involved in solving equations
(10-18) and (10-19) are essentially the same as those which were employed
in the solution of the simple harmonic oscillator equation. Thus he should
not have too
much
difficulty in filling in the details himself,
using the
outline presented above.
Quantum Numbers,
4.
Eigenvalues, and Degeneracy
According to equation (10-16), the eigenfunctions are
VmJT,
All three
6,
quantum numbers
<f>)
= R nl (r) @ [m (6) <B m (0)
(10-35)
are required to identify the eigenfunctions
because their mathematical form depends on the values of each of the
quantum numbers. That three quantum numbers
arise is
a consequence of
the fact that the time independent Schroedinger equation contains three
t
Note
that for a given set of
quantum numbers
Ri(p) to equation (10-31), even though this
is
there
is
only one acceptable solution
a second order differential equation and
two particular solutions. The reason is that one of the particular solutions
not acceptable as it becomes infinite within the range of the variable p. The same
situation obtains concerning the single acceptable solution & im () to (10-23).
therefore has
is
QUANTUM NUMBERS,
Sec. 4]
303
Gathering together the conditions which the
independent variables.
quantum numbers
EIGENVALUES, DEGENERACY
satisfy,
\m\
/
we have
= 0, 1, 2, 3,
= \m\, \m\ + 1, \m\ + 2,
= + 1,1+2, 1+3, ...
.
(10-36)
These conditions are more conveniently expressed in the form
n
/=
1, 2, 3, 4,
m =-/,-/+
It is
...,-
0, 1, 2,
1,
(10-37)
..., 0,
...,+/-
1,
+/
easy to see that equations (10-36) and (10-37) are equivalent.!
Now
let us combine equations (10-29), (10-30), and (10-32) to obtain
an expression for the allowed values of total energy for the atom containing
a single bound electron. This gives
E=
which
vzv =
h p _
=
_n^
2/j,
2fi
flyzv
hY
2/*W
is
pZ 2 e*
=-^-i>
1.2,3,4,...
(10-38)
2n n
Comparing this equation with (5-14), we see that the eigenvalues predicted
by quantum mechanics are identical with those predicted by the Bohr
theory. J
Schroedinger's derivation of this equation (1926) constituted the
convincing verification of the validity of the theory of quantum
mechanics.
first
It is important to note that, whereas the form of the eigenfunctions
depends on the values of all three quantum numbers n, 1, m, the eigenvalues
depend only on the quantum number n. Since for a given value of n there
are generally a
t
number of different possible values of / and m,
According to equations (10-36), the
minimum
value of \m\
is 0.
1, 2,
value
/ is
/ is
= + 1,
1.
the one satisfying the relation n
I
possible values of / are /
0, 1, 2, ...,
is
apparent
equal to \m\, and the
and the minimum value
= 1. Since n increases by integers without limit,
For a given n, the maximum value of /
3, 4, ...
value of
Thus the minimum value of
of , which is equal to /+1, is
the possible values of n are n =
which
minimum
it is
that
is,
1.
Finally, for a given
Consequently the
/,
the largest value
can assume is \m\ = I. Thus the maximum value of m is +/ and the minimum
/, and it can assume only the values m =/,/+ 1, ..., 0, ...,+/ 1,
\m\
is
+/.
%
the
Remember
finite
that in (5-14)
nuclear mass.
m will be replaced by
fi
when proper account
is
taken of
ONE-ELECTRON ATOMS
304
that there will be situations in which
two or more completely
[Ch. 10
different
eigenfunctions are degenerate because they correspond to exactly the same
The degeneracy with respect to m arises whenever the potential
depends only upon r, while the / degeneracy is a consequence of the particular form of the r dependence of the Coulomb potential V = Ze^jr
(cf. section 7, Chapter 5). From equations (10-37) it is easy to see how many
eigenvalue.
degenerate eigenfunctions there are which correspond to a particular
TABLE
Possible Values of
1,2,3
for n
Number
(10-1)
and
-1.0.+1 -2, -1,0, +1, +2
-1,0, +1
of
degenerate
eigenfunctions
for each
Number
of
degenerate
eigenfunctions
for each n
eigenvalue
and
3 are
(10-37),
1.
2.
3.
5.
The possible values of the quantum numbers for n = 1,2,
shown in table (10-1). From this table, or from equations
En
it is
apparent that:
For each value of n, there are n possible values of /.
For each value of /, there are (21 + 1) possible values of m.
For each value of n, there are a total of n2 degenerate eigenfunctions.
Eigenfunctions and Probability Densities
The eigenfunctions
are
WmJr,
8,
</>)
= R nl(r) e im (0) <D
TO (<)
and we know that
im(0)
R nl (r)
=
=
Jm4'
(polynomial in cos 6)
e-te<**rt* (polynomial in r)
AND
EIGENFUNCTIONS
Sec. 5]
All the
bound
PROBABILITY DENSITIES
state eigenfunctions
305
have basically the same structure,
/ the polynomials in r and in
except that with increasing values of n and
cos 6
become
increasingly
eigenfunctions for
more complicated.
unbound
In addition, there exist
an ionized hydrogen atom plus
which have a somewhat different
states (e.g.,
E>
an unbound electron in which
0)
we shall not be concerned at present with situations in
which unbound state eigenfunctions are needed. We list below the bound
structure, but
state eigenfunctions for the first three values
?ioo
/Z\'
4yj2TT\a
Vaii
-Zr/a
<>
cos
'
a,
= j/_zfZr e -Zr/2a sJn g e i4.
8V
^300
n.
-7=|-J
yJiT \a /
l/Z^Zi
Zr - Zr/Za
e
210
of
77"
_L/ZJV
18 +
-Zr/3a
(10-39)
e~
aly/TT \a
olyjir \a
V-320
-^
81^077
V32i
e
- ^f
\a
al
= rrV (-f
81^/77 \a
77^7=
1627"-
-ZH3a<,
cos
sin Q e
V C S2 6
i*
~D
^322
' Zrl3a
ZrlZa '
~ Zrl2a
sin
cos d ei *
aj
/Zf
ZV
i~ e -^r/3o sin 2 0e 2i*
These are written in terms of the parameter
a
= Ar =
0.529
X l(T 8 cm
(10-40)
lie*
which the reader will recognize as the radius of the smallest orbit of a
Bohr hydrogen atom.
ONE-ELECTRON ATOMS
306
Figure 10-2.
The
radial probability densities of
the one-electron atom for n
[Ch. 10
I,
2, 3.
AND
EIGENFUNCTIONS
Sec. 5]
The eigenfunctions (10-39)
/so
That
are normalized.
307
is,
f2ir
'"lm
J Jo
J
Jo
J
>o
where
/V
PROBABILITY DENSITIES
sin 6 dr
dd
(r, 0,
d<f>
<f>)
Vnlm (r,
the unit of
is
0)r sin 6 dr dB
6,
d<f>
(10-41)
volume in the system of spherical
Also the
all space.
coordinates, and where the integral extends over
eigenfunctions-are orthogonal, so that
Cn f 2ir
/*oo
Wtvm-ir,
Jo
unless ri
An
6,
<j>)
Wnl Jr,
<f>y sin 6 dr dd
6,
Jo Jo
n,V
and m'
I,
d<]>
(10-42)
m.
is best obtained by considering
form of the corresponding probability density functions ^*imim =
the
interpretation of the eigenfunctions
R%iimmRni imm- As these are functions of three variables, we cannot
plot them directly. However, an adequate understanding of their three
dimensional behavior can be had by discussing separately their dependence
on each variable. We consider first the r dependence in terms of the
Pn i(r),
radial probability density
defined by
n27T W*m m Wm m r2 sin
dr dd
d<f>
n2ir
The
R nl
integrals over 6
0, m and
Om
and
is
<f>
are equal to unity because each of the functions
separately normalized.
(They are also separately
orthogonal.) Thus
P nl (r) dr =
Now
r*R*nl (r)
R nl (r) dr
(10-43)
ytimWnnJ sm dr dd d<j> is the probability of finding the electron
volume element r 2 sin 6 dr dd dj>. Since P nl (r) dr is obtained from
quantity by integrating over the entire range of 6 and
Pn i(r) dr
2,
in the
this
is
<f>,
the probability of finding the electron anywhere with a radial coordinate
between
and
In figure (10-2),
dr.
as a function of rj{a jZ) for the cases n
/
1,
2,
is
plotted in units of Z/a
and
3,
and
all
the possible
Pn i(r)
does
see that the radial probability densities for each eigenstate
have
The radial
quantum number m.
values for these values of n.
not depend on the
We
Pn i(r)
probability density
appreciable values only in reasonably restricted ranges of the r axis.
Thus, when the atom
is
in
an
eigenstate, there is a high probability that
the radial coordinate of the electron will be found within a reasonably
well-defined range;
that
is,
the electron
would quite probably be found
within a certain shell limited by two concentric spherical surfaces centered
ONE-ELECTRON ATOMS
308
A study of these figures will demonstrate that the character-
on the nucleus.
of these shells
istic radii
although there
n,
quantitative
[Ch. 10
is
is
determined primarily by the quantum number
/ dependence.
This can be seen in a more
a small
way by using
the expectation value of the radial coordinate
of the electron to characterize the radius of the
shell.
The expectation value
Wmnf
sin 6
obtained by evaluating the integral
is
Too
*\i
/*oo
rP nl (r) dr
Jo
result
Civ
y*nlm
Jo Jo
Jo
dr dd
d<f>
is
r nl
The
r-ir
values of
apparent that
Kl
^{l+J
+
n
1)1
(10-44)
T^ are indicated in figure (10-2) with small arrows.
T^ depends primarily on n, since the / dependence is
l
It is
sup-
pressed by the factor J and the factor l/ 2 It is interesting to compare this
expression with equation (5-12), which gives the radii of the circular
.
orbits of a
Bohr atom. The equation
r
'Bohr
is
n2a
Quantum mechanics shows that the T\ are of approximately the same size,
and have approximately the same n dependence, as the radii of the circular
Bohr orbits.
x
All electrons in eigenstates with
radial probability densities
common
n values have roughly similar
and approximately the same expectation value
for the radial coordinate, independent of the values of
to the
commonly used
or m. According
terminology, such electrons are said to be in the
Each shell also has the property that the associated eigenhave exactly the same value En
Some insight into the reason why the eigenvalues depend only upon n
may be obtained by computing the expectation value of the potential
energy of the atom. This is
same
values
shell. f
all
y=\
Jo Jo Jo
)y
Wtimi
\
nlm r sin 6 drddd<]>
-zWi
Evaluating (1/r) leads to the result
,Ze
t Since at present we are considering one-electron atoms, this terminology refers to
different atoms in each of which the electron is in the same shell. However,
two or more
is also used for multi-electron atoms
be in the same shell.
this
terminology
can
literally
in
which two or more electrons
AND
EIGENFUNCTIONS
Sec. 5]
Since this does not depend
PROBABILITY DENSITIES
on / or m, the expectation value of the
309
potential
energy,
/rV
does not
because
This is reflected in the behavior of the total energy E,
a characteristic property of any potential of the form
that the total energy of a particle moving in the potential is
either.
it
V(r) oc \\r
is
always equal to exactly one-half
we
Finally,
call attention to
its
average potential energy.
a point which will later be seen to be of
considerable importance.
Inspection of the eigenfunctions (10-39) will
verify that for values of r
which are small compared to a [Z (where the
exponential term
functions
is
slowly varying), the radial dependence of the eigen-
is
Vmm^r
r^O
1
,
(1(M5)
Therefore the radial dependence of the probability density for small r
VtlmVnlmCCr
As a consequence,
21
,
-*
is
(10-45')
f* lm f nlm in a volume element near r = is
and decreases very rapidly with increasing /.
the value of
relatively large only for
0,
Consider again the probability density,
From
equation (10-21),
we have
Thus the probability density does not depend on the coordinate
<j>.
The
three dimensional behavior of y)* lm ip n i m is therefore completely specified
by the product of the quantity R*i(r) Rnl {r)
Pnl (r)/r 2 and the quantity
m (0)
factor.
lm (0), which plays the role of a directionally dependent modulation
The form of 0?m (0) ;m (0) is conveniently presented in terms of a
polar diagram as
the point
shown
The origin of the diagram is at
and the z axis is taken along the direction
measured. The distance from the origin to the
in figure (10-3).
(the nucleus),
from which the angle d is
curve, measured at an angle 6, is equal to the value of @?ro (0) &lm (0) for
that angle. Such a diagram can also be thought of as representing the
complete directional dependence of if*i m if nlm by visualizing the three
dimensional surface obtained by rotating the diagram about the z axis
through the 360 range of the angle
The distance, measured in the
direction specified by the angles 6 and
from the origin to a point on the
surface, is equal to 0?m (0) !m (0) 0*(<) O m (> for those values of 6 and
<f>.
</>
<f>.
ONE-ELECTRON ATOMS
310
In figure (10-4)
we
illustrate
form of the function *m (Q)
Figure 10-3.
a typical example of the dependence of the
!m (0) on
1=3, and
of polar diagrams for
[Ch. 10
the
quantum number m, by a
the seven possible values of
set
for this
polar diagram of the directionally dependent modulation factor of the
probability density.
,,Z
3,
Figure 10-4.
m=
Polar diagrams of the directionally dependent modulation factors of the
From L. Pauling and E. B. Wilson,
I. 2, 3.
/
0,
3; m
probability densities for
Introduction to
Quantum Mechanics, McGraw-Hill Book Co.,
value of
i.e.,
/,
\m\
0, 1, 2, 3.
Note
the
way
New
in
York, 1935.
which the region of
?m 0j m (and therefore of vJ, m VniJ snifts from the z
axis to the plane perpendicular to the z axis as the absolute value of
increases. Some important features of the dependence of 0*m (0) ; m (0)
concentration of
AND
EIGENFUNCTIONS
Sec. 5]
PROBABILITY DENSITIES
on the quantum number / can be indicated by presenting a
plots with \m\ = / and / = 0, 1, 2, 3, 4, as illustrated in figure
=m=
311
set
of polar
(10-5).
For
which is the ground state of the
and the probability density is
atom, iptimWmm depends on neither 6 nor
spherically symmetric For other states, the concentration of probability
density in the plane perpendicular to the z axis, when \m\ = I, becomes more
the eigenstate with n
1,
0,
<f>
and more pronounced with increasing
Figure 10-5.
m=
= o,
3,
The
set
1,
m=
+3
= 4,
2,
m=
m=
+4
Polar diagrams of the directionally dependent modulation factors of the
probability densities for
Introduction to
m=
/.
0,
I,
2, 3, 4;
\m\
/.
From
Quantum Mechanics, McGraw-Hill Book Co.,
L.
Pauling and
New
E. B.
Wilson,
York, 1935.
of quantum numbers nlm determines the complete three
dimensional behavior of y* im y n;m with the behavior as a function of the
,
direction relative to the z axis depending only
set
of quantum numbers,
and *m (0)
!m (0)
R*
{r)
R n i(r)
on
has zeros for several values of
density for an electron
and m. For a
typical
has zeros for several values of
bound in the Coulomb
and conical nodal
6.
r,
Thus the probability
potential of a nucleus has,
on which its value
These surfaces are analogous to the nodal points in the probability
density of a particle bound in a one dimensional potential, and they are a
consequence of the fact that the wave functions for a bound particle must
be standing waves. Of course, it is not possible to perform a sequence of
measurements on the location of an electron in a particular eigenstate
with quantum numbers nlm, and thereby locate the nodal surfaces and
in general, a set of spherical
is
surfaces
zero.
determine the direction of the
z axis.
This would amount to finding a
preferred direction in a space which should be spherically symmetric
because the potential V = Ze?/r is spherically symmetric. It cannot be
done because any eigenstate, except the spherically symmetric eigenstate
1,
=m=
0, is
degenerate with several other eigenstates with the
ONE-ELECTRON ATOMS
312
[Ch. 10
same value of n and so cannot be investigated separately. All that can be
measured experimentally is the average probability density for the whole
set of states which are degenerate with each other. For instance, it would
be possible to make measurements on the probability density for atoms
whose energy is E2 The results would be
.
il>200V200
ftl -1V21-1
= T^(-)3e
12877 \a
" Zr/a
r(
L \
VzloWilO
- "I
+
a /
V21lV21l]
(%
a
sin2 6
* sin2 d
cos2 e )
(10-46)
128t;
2877
\a /
This is a spherically symmetric distribution, and it cannot be used to
determine the direction of the z axis. Thus there is no contradiction of the
fact that this direction
On
was
initially
chosen in a completely arbitrary way.
the other hand, consider a situation in which the orientation of the
z axis is
not arbitrary because there
is
a preferred direction defined by an
external magnetic or electric field applied to the atom.
as
we
shall see later, the eigenstates are
In such a field,
not degenerate and measurements
of the probability density of a particular eigenstate could be performed.
In fact, such measurements could be used to determine the direction of the
external
We
field.
shall
show
related to the
and to
its z
quantum numbers
shortly that the
component
L by
z,
and
are also
momentum L of the electron,
magnitude of the orbital angular
the equations
L = J 1(1 + l)h
(10-47)
L z = mh
We mention this now because it is
an important clue to the interpretation
I an d m.
Consider the case m = I.
of the dependence of wtimWnim on
Then L z
angular
momentum
Ih,
which
is
almost equal to
For a Bohr atom
axis.
would
lie
L=
Vl(l
1) h.
In this case the
vector must point nearly in the direction of the z
this
would mean that the
orbit of the electron
nearly in the plane perpendicular to the z axis.
With increasing
> Vl(l
values of/, lh
+ 1) h and L z -^ L, so the angle between the angular
momentum vector and the z axis decreases. In terms of the Bohr picture,
this
demands
the z axis.
that the orbit
An
lie
more nearly
in the plane perpendicular to
inspection of the polar diagrams will demonstrate the
correspondence between these features of fnimWnim an ^ tne picture of a
Bohr orbit. For
have L z = 0, and the angular momentum
m=0we
ANGULAR MOMENTUM OPERATORS
Sec. 6]
In a Bohr
vector must be perpendicular to the z axis.
mean
313
atom
Some
that the plane of the orbit contained the z axis.
would
this
indication of
behavior can be seen in the polar diagram for / = 3, m = 0.
Although there are many points at which the quantum mechanical
theory of the one-electron atom corresponds quite closely to the Bohr
theory, there are certain striking differences. In both theories the ground
state corresponds to the quantum number n = 1
But in the Bohr theory
the orbital angular momentum for this state is L = nh = h, whereas in
this
quantum mechanics L
V/(/
1)
since
when n
1.
There is an overwhelming accumulation of experimental evidence which
shows that the quantum mechanical prediction is the correct one. It
would be very difficult to modify the Bohr theory in a way which would
momentum
allow for zero angular
such a state
states, since the orbit for
would be a radial oscillation passing directly through the nucleus. In
quantum mechanics the electron does not move in a well-defined orbit, and
so no problem arises.
6.
Angular
Momentum Operators
In the next sections
we
shall discuss the orbital angular
of the eigenstates of a one-electron atom.
We
shall start
lating the expectation values for this quantity, but
possible to obtain a very
much more
we
the theory of
shall find that
precise evaluation.
momentum
In classical mechanics the angular
the origin of a certain coordinate system,
is
it is
In the course of
new and broader insight into the general
quantum mechanics will be obtained.
this discussion a
momentum
by simply calcu-
properties of
of a particle, relative to
a vector quantity
which
is
given by the equation
L=
The vector
r is the position
vector p is the linear
in figure (10-6).
vector
is
of the particle relative to the origin, and the
The
"cross product"
do the
x, y, z
r, p,
pis defined to mean that the
p, and
have the same relative spatial
and L
is
defined so that the magnitude of the
is
L=
r
rX
axes of a right-handed coordinate system.
Furthermore, the cross product
where
(10-48)
perpendicular to the plane formed by the vectors r and
orientation as
momentum of the particle. These vectors are indicated
points in the direction so that
vector
rp sin 6
and p are the magnitudes of
and
(10-49)
p,
and
is
the angle between
ONE-ELECTRON ATOMS
314
these
two
From the
vectors.
three rectangular
definition of r
components of
p, it is
easy to
show
x, y, z are the
that the
are
Lx = yp z - zpv
Lv = zPx xPz
Lz = XPV - VPx
where
of p.
[Ch. 10
components of
r,
(10-50)
and px p y p % are the components
,
*-y
Figure 10-6.
moving
The
position, linear
momentum, and
angular
momentum
vectors of a
particle.
In order to discuss the dynamical quantity angular
momentum
in
quantum mechanics, we construct the associated operators by replacing
px py p z with their quantum mechanical equivalents ih djdx, ih djdy,
ihdjdz (cf. 7-77). Thus the operators for the three components of
,
angular
momentum
are
"*
LVos
= in\z
L,
= in\x
V
It is
r, d,
By/
dz
dx
(10-51)
dz!
y
dy
dx)
desirable to transform these expressions into the spherical coordinates
(f>.
When
this is
L*
done,
it is
= ^( sin
found that
4>
jig
l.op = ^(-cos $
L ** = - ih
cot 6 cos
^ + cot
4>
sin
^7/
4>
2\
(10-52)
ANGULAR MOMENTUM OPERATORS
Sec. 6]
We
shall also
L, which
315
be interested in the square of the magnitude of the vector
is
L2 =
I?B
4 + L\
(10-53)
In spherical coordinates the associated operator
a (
-sin0 30
sin
s
dd!
is
'
sin
(10-54)
2
0d< J
Let us calculate the expectation values of the components of L, and of
the square of
its
magnitude, for an electron in the eigenstate nlm of a
According to equation (7-80), extended to three
one-electron atom.f
dimensions, they are given by
-J7J
-///e
Tr2 sin
1"
6 dr dd dd>
iE tl!i
iE tl!i y>*
rZ sin 6 dr
y, nlm
nlm L Xo e-
dd
d<f>
So
Lx
=
J J J
2
'*
v imL *op^''" sin 6 dr dd d ^
Similarly,
L~v
2
vti m Ly 09 W n m r sin 6 dr dd dj>
i
J J J
-J7M
nL z
-HI*
w nlm r2
L2oV Wm m r2
sin 6 dr
sin 6 dr
dd
dd
(10-55)
d<f>
dcf>
where the integrations are over the entire range of the variables.
evaluate these integrals, we must first evaluate the quantities
L Xov y> n m
i
In doing
this,
we
Consider
first
the quantity
LVop ip nlm
L^rpnim,
shall find that the last
Lt Zov ,
y>ni m
L,
To
ov %p nlm
two lead to an
interesting result.
dyw
= ~ lh ~^T
.-fc
of a
t Strictly speaking, we shall calculate the components and squared magnitude
which is the orbital angular momentum of the electron plus the orbital angular
vector
of the nucleus, relative to the center of mass of the atom. This is because
y> nlm actually describe the motion of an electron of reduced mass n,
which is moving about a fixed nucleus in such a way that the orbital angular momentum
of the reduced mass electron is equal to the total orbital angular momentum of the atom.
momentum
the eigenfunctions
Cf. section 2.
ONE-ELECTRON ATOMS
316
[Ch. 10
Since
Vnlm
Rnl(r) l m (Q)
m(<t>)
we have
- ih dJ = R
nl (r)
lm (Q)
-ih
d m (4>)
dcj>
According to equation (10-21),
0>
m (<)
im*
So
dO n (4>)
ime
im<D m (>
dcp
Thus
-fc
Pynlm
d<j>
:mM
Bj
(r)0 ;m(0)O m()
and
Lz
nr>
V>nlm
mkip nl
(10-56)
Next evaluate
~
= R nl (r)(-h
L%y> nlm
2
)
m (4>) d_( sin
sin9 d6\
Differentiating equation (10-21),
dQtJO) )
dd
i m (8) d
sin
(<f>)
dtf
we have
= -m m
2
(<f>)
d<f>
So
L 2 p y> nlm = R nl(r)<& m(<l>)(-ti )
Now
This
dO
is
(6)
dd
a solution to the
0,
I
sin
differential
m]
equation
is
d (
Lsin
a function which
Jm (0) is
(10-18).
dL ne d& lm
sin
d lm (6)\
sindddX
sin
dV \
dd
sin
lm(0) =-/(/+
l) m (0)
where we have evaluated a from (10-24). Therefore
8
i< opVi
= RnMQj&X-Pti-Ki + i)0(fl)]
= /(Z+l)A w! (r)0, m(0)(D m (^
2
and we have
i?opVi
Kl
(10-57)
l)*Vi
we see that LZo y> nlm is equal to mhy nlm Thus the
of operating on the one-electron eigenfunction y nlm with the
is simply to multiply that eigenfunction by the
differential operator L
In equation (10-56),
result
2op
ANGULAR MOMENTUM OPERATORS
Sec. 6]
317
constant mh. Similarly, equation (10-57) shows that the result of operating
on the one-electron eigenfunction y> n[m with the differential operator L\
v
is
multiply that eigenfunction by the constant
to
results are certainly
/(/
\)h 2
These
not typical of the behavior of differential operators.
we operate on the function /(x) = x2 with the differential
operator djdx, we obtain a very different function/'(z) = 2x. As another
example, the reader may verify that the result of operating on y nlm with
the operators Lx
or L Vo is to produce new functions of r, 6,
in which
these variables enter in a different manner from the way they enter in the
For
instance, if
cj>
function
That
y> nlm .
is,
L Xop y>nim
Lv ot,Wmm
# (constant)y) Blm
^ (constant)^
Using equations (10-56) and (10-57),
L2
from equations
(10-58)
trivial to
it is
evaluate
T and
z
These values are
(10-55).
n^z^Wmmf
//J' nlrn^Zo
Lz = mh\
v*nlm y nlm r
sin 6
dr dd
d<f>
sin 6
dr dd
d<f>
T = mh
(10-59)
and
2
=
J J J
ftinfiovVnimr* sin 6 dr dd
L2 =l(l +
l)k
!?=/(/+
l)h
jjjy>*nlmWnlm r
d<f>
sin d
dr dd
d<f>
(10-60)
where we have employed the normalization condition (10-41). Let us
compare these results with the relations (10-47) which were stated without
proof in the previous section. These relations are
Lz = mh
L =
2
It is
/(/
(10-61)
\)h
(10-62)
apparent that equations (10-59) and (10-60) are consistent with
do not prove their validity, as the latter pair make
(10-61) and (10-62) but
very
much
stronger statements about the values of
do the former. In order
of equations (10-61) and (10-62), it
eigenstate nlm, than
Lz and
I?, for the
to complete the proof of the
will be necessary to make a
short digression into the general properties of the theory of quantum
validity
mechanics.
ONE-ELECTRON ATOMS
318
[Ch. 10
Eigenvalue Equations
7.
Consider a particle moving under the influence of a time independent
For simplicity, consider the motion to be one dimensional.
Let / be any dynamical quantity useful in describing the motion of the
particle. Then the average value of/, obtained in a series of measurements
made when the particle is in a state of motion described by the wave
function W, is the expectation value /. According to equation (7-80),
potential.
it
has the value
f^yVfeJVdx
=JV
where the integral is taken over the entire range of the variable. The
quantity /provides a partial specification of the behavior of/ To obtain
a more complete specification, we must learn something about the
fluctuations
of/ This can be done by
studying the quantity
A/ 2 which is
,
the expectation value of the square of the difference between the results of
a particular measurement
That
of/ and
the average of these measurements /.
is,
A/ 2 = (/-/)2
(10-63)
It is clear
that information relating to the fluctuations in /is provided
A/
instance, consider a case in
2
.
(e.g.,
For
by
which /can have only a single value
then/ will always be
the total energy of a particle in an eigenstate);
observed to be exactly equal to the value
However, if/
fluctuates
greater than zero.
To
about
evaluate
/,
and (/ ff
its
average value
A/
2
,
/,
will
then (/ /)
we apply equation
be zero.
2
will
be
(7-80) to (10-63)
and obtain
Y^
A/ 2 =jV*[(/-/)2] op
Since
/ is
a constant, the operator [(/
/)
2
] op
is
simply (/op
/)2
we have
A75 =JV*(/op-/fFiz
vp*
'*(i1p-2/op/+/)Ydx
= JV*/VF dx -
2/jVi/opY dx
+ f JV*T dx
and
EIGENVALUE EQUATIONS
Sec. 7]
319
Using equation (7-80) on the first two terms of the right
on the third term, we obtain
side,
and using
the normalization condition
W=P-w+f
or
&p=p-f
Thus the quantity A/2
is
(10-64)
equal to the difference between the expectation
value of the square of/, and the square of the expectation value of/.
Now let us assume that the particle is in a certain eigenstate so that
Yix,
t)
e~ iEi ' n y(x)
Then we have
je
iEt/n
y>*fov e-
iEtin
y>dx
According to equation (7-80), /op never contains derivatives with respect
to time. Consequently the term er iEtltl can pass through op as if it were a
constant, and we have
f=L
iEt!n
e-
iEil 'l
f*fov xp dx
or
f=\f*fo P fdx
Similarly
we can show
(10-65)
that
r*=\xp*flvW dx
(10-66)
when
the particle is in an eigenstate.
For a given dynamical quantity/ and a given eigenfunction
evaluate
A/2 >
A/2 =
0.
0.
A/2 from
the equations above.
In general,
However, as mentioned previously, situations
is
we can
ip,
found that
exist in
which
We shall now show that this happens when the relation between
the dynamical quantity /and the eigenfunction
/opV
where
have
it
F is
a constant.
/=
dx
y)*Fy)
dx
so
f=
such that
Fy>
(10-67)
If this is the case,
ip*fovW
ip is
i72
then in equation (10-65)
=F
rp*y>
dx
=F
we
ONE-ELECTRON ATOMS
320
[Ch. 10
we have
In equation (10-66)
f=
V^fopf dx
W*fopFf dx
f*fopfopf dx
=F
ip*fopf
dx
= FF
so
Then, from equation (10-64), we find
A/1 =/2 - f = F 2 - F 2 =
(10-68)
2
Thus, when the relation (10-67) is true, the quantity A/ which is a measure
of the fluctuations in the dynamical quantity /, is equal to zero. In any
normal case this is adequate to prove that/can have only the single value
,
f= /=
F. In exceptional cases equation (10-68) alone does not constitute
a proof of this statement, but
p -f =
However,
it is
it is
also necessary to
for any integer
0,
easy to see that
when equation
show
that
(10-69)
(10-67)
is
true the steps
immediately preceding equation (10-68) can be repeated k times to show
that the requirement (10-69) is satisfied. Thus we conclude that when
the relation between
f and rp
the dynamical quantity
f can
is
foW =
(10-70)
Fy>
have only the
definite, precise value
f=F
(10-70')
In the mathematical literature an equation of the form (10-70) is
an eigenvalue equation, the function y> is called an eigenfunction of
called
the operator fop,
and the constant Fis
called
an eigenvalue of the operator.
We have come across most of this terminology before in connection with a
Consider the time independent Schroe-
particular eigenvalue equation.
dinger equation (10-8), and rewrite
it
fc2
"J
i=i
Comparing the
2m,-
curly bracket with the classical definition of the total energy
of the system (10-1),
we
see that this equation
e op y
Ey>
may be
written
(10-71)
EIGENVALUE EQUATIONS
Sec. 7]
321
where is a constant numerically equal to the total energy of the system,
and where the total energy operator e op is obtained from the corresponding
dynamical quantity by replacing^, with ihdjdxj, etc. This operator
is often called the Hamiltonian.
For one particle in one dimension it is
simply
e ov
+ V
2m i
h2
= -
d*
T/
dx'
In the process of studying the particular eigenvalue equation (10-71),
we have
learned
common to
y>
and
its
finiteness,
all
much about mathematical
properties which are actually
such equations. For instance,
we know that,
and single-valuedness, there
will
(10-71) only for a certain set of eigenvalues
eigenfunctions
rp
x,
ip
ip 3 ,
2,
eigenvalue equation (10-70).
(10-71),
if
the function
derivative are required to satisfy conditions of continuity,
we concluded
ip
n,
be solutions to equation
Ex E2 E3
,
The same
....
is
total energy are the eigenvalues
of the subject,
is
En
and
In interpreting the significance of equation
operator appears in the equation.
we
that the set of eigenvalues comprise the only
physically possible values of the dynamical quantity
eigenvalue equation,
true of the general
(That
Ex E
2,
is,
.)
whose corresponding
the allowed values of the
If we
extend this to the general
obtain a statement which, in advanced treatments
usually taken as the fundamental postulate of the theory
of quantum mechanics:
A measurement of the value of a dynamical quantity f can yield only one
of the eigenvalues Fn of the corresponding operator /op where fop is obtained
from f by replacing x with ih djdx, etc., and where the relation between
,
Fn and fop
is
fovVn
in
which the eigenfunction
and
continuity, finiteness,
ip
= Fn y n
and
its
derivative satisfy
conditions of
single-valuedness.
This one-sentence postulate, and the postulates which provide a physical
interpretation
of the eigenfunctions, contain the complete theory of
quantum mechanics
for time independent
phenomena.
The theory
is
extended to include time dependent phenomena by introducing the
operator
ih djdt.
We may illustrate the theory developed in this
examples.
Consider
first
the traveling
y>
We know
that this
is
section with
some simple
wave eigenfunction
iKx
an eigenfunction of
(10-72)
e op , as the function (10-72)
ONE-ELECTRON ATOMS
322
was obtained by solving (10-71)
is
V=
for the case
associated with this eigenfunction
is
E=
h2
0,
and that the eigenvalue
K2 j2m.
The
function (10-72)
an eigenfunction of the momentum operator pop
also
[Ch. 10
= ih d/dx
since
P opW
ih
dx
V>
ih
dx
iKx
= ihiKeiKx = hKy
or
= pf
PotW
where the eigenvalue is P = hK. We see that this function is simultaneously
an eigenfunction of two operators. Therefore, according to equation
(10-70'), a particle whose motion is described by this function must have a
definite value of total energy E = h 2 K 2 \2m and a definite value of momen-
tum
function
E=
hK.
y>
K 2 /2m, and
h2
value
we may show
Similarly,
that the traveling
wave
eigen-
= e~~iKx is an eigenfunction of eop associated with the eigenvalue
P = hK.
also
is
function has the definite
definite value
an eigenfunction of pov associated with the eigenis described by this
value E = h2 K 2 j2m for its total energy, and the
Therefore a particle whose motion
P = hK for
its
momentum.
Next consider the standing wave eigenfunction
We know
that this
eigenvalue
E=
h2
is
also
K /2m.
2
sin
Kx
(10-73)
an eigenfunction of e p associated with the
However, it is not an eigenfunction of pop
since
Por,y>
ih
sm Kx =
dx
Although the energy of a
particle
function has the definite value
value.
ihK cos Kx ^ (constant)y
E=
h2
whose motion
is
described by this
K2/2m, the momentum has no definite
According to the postulate stated above, a measurement of the
momentum of the particle can yield only one of the eigenvalues of the
momentum operator. For a given value K there are just two eigenvalues of
= hK. A measurement of the momentum of
jp op namely P = hK and P
,
the particle
would
yield either of these
two values and,
in a large
number
of measurements, they would occur with equal probability.! It is evident
that these properties of the standing wave eigenfunctions and traveling
wave eigenfunctions are in complete agreement with the interpretation
presented in the
t This
may
is
verify.
first
section of Chapter
a consequence of the fact that p
8.
for this eigenfunction, as the reader
ANGULAR MOMENTUM OF ONE-ELECTRON ATOMS
Sec. 8]
Angular
8.
Momentum
323
of One-Electron-Atom Eigenfunctions
We return now to the functions y nlm which arise in treating the problem
of a one-electron atom.
Being solutions to the time independent Schroe-
dinger equation for this problem, the
y nlm
are necessarily eigenfunctions
of the total energy operator. According to equation (10-38), the associated
eigenvalues are
En = aZ 2 e4/2^2 2 From
.
we
2
equations (10-56) and (10-57),
see that these functions are also eigenfunctions of the operators for the
component of the
orbital angular
momentum L Zo
and for
its
square
L\ v with associated eigenvalues mh and /(/ + l)/j2 However, equations
(1058) state that they are not eigenfunctions of the operators for the
x and y components of orbital angular momentum. Also, it is easy to
show that they are not eigenfunctions of any of the linear momentum
.
operators.
Therefore,
the total energy of the
orbital angular
when an
atom is
momentum
is
electron
definitely
its z
component
is
by y n!m
the magnitude of the total
in a state described
En
Vl(l
1)
(10-75)
is
no
(10-74)
definitely
L = mh
but there
definitely
L=
and
is
definite value of the
x or y components of
orbital angular
momentum, or of any component of linear momentum.
The fact that rp nlm does not describe a state of definite linear momentum
any physical picture of the motion of an
and z component of orbital
angular momentum calls to mind some sort of rotation in which linear
momentum could not be constant. However, the fact that y nlrn does not
describe a state with a definite x and y component of orbital angular
is
certainly reasonable, since
electron in a state having a definite magnitude
momentum
somewhat mysterious because, in classical mechanics, the
momentum vector of an isolated system moving under the
influence of a spherically symmetrical potential V(r) would be fixed in
space, and all three components would have definite values since there
would be no torques acting on such a system. The quantum mechanical
result is a manifestation of an uncertainty principle relation which can be
shown to exist between any two components of orbital angular momentum.
Because the z component of orbital angular momentum has the precise
value mh, this relation requires that the values of the x and y components
be indefinite. Upon evaluating J7X and L~ it is found that both of these
v
quantities are equal to zero. Thus the orientation of the orbital angular
is
orbital angular
ONE-ELECTRON ATOMS
324
[Ch. 10
momentum
vector must be constantly changing in such a way that its
x and y components rapidly fluctuate about an average value of zero.
Many of the properties of the orbital angular momentum can be conveniently represented in terms of vector diagrams. Consider the set of
states
having a
common value of the quantum number For each of these
momentum vector, in units of
/.
states the length of the orbital angular
It,
= V + 1). In the same units, the z component of this vector is
LJh = m which, according to equations (10-37), may assume any integral
value from Ljh = to LJh = + depending on the value of m. The
is
L\h
/(/
1,
Figure 10-7. Illustrating the angular
the five possible states for /
2.
momentum
vectors precessing about the z axis in
case of
momentum
values of
is illustrated in figure (10-7). The figure depicts the angular
vectors of the five states corresponding to the five possible
for this value of /. Since L x and L fluctuate about their average
values of zero, these vectors must precess randomly in the conical surfaces
surrounding the z axis. The actual spatial orientation of the angular
momentum vector is known with the greatest precision for the states with
m=
1;
but even for these states there
is
some uncertainty
since the
vector can be anywhere on a cone of half-angle cos-^ljV 1(1
In
1)].
the classical limit / -> oo, and this angle becomes vanishingly small.
Thus, in
this limit, the
constrained to
angular
momentum
vector for the states
m = 1
almost along the z axis and is therefore essentially
fixed in space in agreement with the behavior predicted by the classical
is
lie
theory.
Since
Lz = mh
was chosen
component
with
arbitrarily,
m
it
an
integer,
and
follows that
since the direction of the z axis
quantum mechanics
requires the
any direction of the orbital angular momentum to be an
integral multiple of h. This is known as spatial quantization. If the reader
will review the derivation of the eigenvalue equations (10-56) and (10-57),
in
Ch.
EXERCISES
10]
325
will be evident that both spatial quantization and the quantization of the
magnitude of the orbital angular momentum according to the rule
it
L = V /(/ +
1)
h occur not only for the
Coulomb
potential V(r)
= Ze
but for any potential which depends only on the radial coordinate
/r,
r.
BIBLIOGRAPHY
Leighton, R. B., Principles of Modern Physics, McGraw-Hill
Book
New
Co.,
York, 1959.
Pauling, L.,
Hill
Book
Schiff, L.
I.,
and E. B. Wilson, Introduction
Co.,
New
to
Quantum Mechanics, McGraw-
York, 1935.
Quantum Mechanics, McGraw-Hill Book Co.,
New York,
1955.
EXERCISES
1
one
2.
Write an expression for the probability flux vector for a system containing
a three dimensional space.
particle in
Separate (10-10) into an equation in the rectangular coordinates of the
center of
mass of the atom,
x, y, z, defined
by (10-11), and an equation
in the
rectangular coordinates of the electron relative to the nucleus, X, Y, Z, defined
by
X = x - xu Y = y - yu Z = z - z v
2
3.
Transform the equation
and thereby
in X, Y,
Hint:
verify (10-15).
Z'mto an equation in the r, 6, j> of (10-12),
Consult some reference on intermediate
calculus for an explanation of the use of Jacobians.
4.
Solve equation (10-23) by the procedure indicated in the paragraph
containing that equation.
5.
Solve equation (10-31) by the procedure indicated in the paragraph
containing that equation.
Consider a rigid rod of length R and negligible mass, with one end fixed
way that it can rotate freely and the other end carrying
a particle of mass M. Adapt the time independent Schroedinger equation (10-1 5)
to this system. Solve the equation and show that the eigenvalues are
6.
at the origin in such a
hH{l
1)
I
2/
where /
MR
is
the
moment
0, 1, 2, 3,
of inertia about whatever axis the rod
may
Note that there is no zero point energy for this system because all the
coordinates are free to assume all possible values. Compare these results with
rotate.
the results of exercise
is
3,
Chapter
5.
Evaluate the expression analogous to (10-46) for n
spherically symmetric.
7.
3,
and show that
it
ONE-ELECTRON ATOMS
326
Verify equations (10-50).
8.
Use some of the equations obtained in exercise
9.
and
3 to verify equations (10-52)
(10-54).
10.
on
[Ch. 10
their values.
and Ap 2 for the eigenfunction y
Hint: Use box normalization.
Evaluate/,
sin
Kx, and comment
II
CHAPTER
Magnetic Moments,
Spin, and
Relativistic Effects
Orbital Magnetic
I.
In the
first
Moments
part of this chapter
we
shall discuss
some experiments which
provide very convincing evidence for spatial quantization.
momentum
related quantity
we
Actually
measure the z component of the angular
vector L, but instead measure the z component of a closely
the experiments
do not
(t,
directly
the magnetic
moment vector. To begin the discussion,
moment (*j of an atomic electron.
shall evaluate the orbital magnetic
This will be done by using a combination of simple electromagnetic theory,
the Bohr theory, and quantum mechanics. A completely quantum mechan-
not be presented because such a treatment would
more advanced knowledge of electromagnetic theory than has
ical calculation will
require a
been assumed in this book, and also because, owing to the conceptual
simplicity of the description of the atom provided by the Bohr theory,
there are advantages in using that theory when introducing something
new. Of course, the Bohr theory does not always yield quantitatively
correct results but for the calculations we shall perform here it does.
Consider, then, an electron of mass m and charge e moving with
Bohr orbit of radius
loop constitutes a current of magnitude
velocity v in a circular
r.
e
= -f~
r
c 2nr
c r
e
-
This charge circulating in a
327
(11-D
MAGNETIC MOMENTS,
328
RELATIVISTIC EFFECTS
SPIN,
[Ch.
where r
is the orbital period of the electron.
It is well known from
elementary electromagnetic theory that such a current loop produces a
magnetic field which is the same at large distances as that of a magnetic
dipole located at the center of the loop and oriented perpendicular to its
For a current / in a loop of area A, the magnitude of the magnetic
plane.
moment
of the equivalent dipole
is
P-i
iA
(11-2)
L
Figure
l-l.
The
quantities associated with an electron
The magnetic moment is a measure of the
dipole and is defined as
moving
in a circular
Bohr orbit
strength and orientation of the
V-^qh
(11-3)
where q
is the magnetic strength of each of the poles and 8 is a vector
extending from the negative (south) pole to the positive (north) pole.
For a current loop produced by a negatively charged particle, the direction
of {Aj is opposite the direction of the angular momentum vector L, whose
orientation and magnitude are given by the equation
L=
rXp = wrXv
(11-4)
These quantities are illustrated in figure (11-1). Evaluating / from equation
(11-1) and A for the circular Bohr orbit, we find from (11-2) that the
magnitude of the magnetic moment is
Vi
e
~
w =
v
evr
,.
(11-5)
2c
c 2-nr
Since r and v are perpendicular in a circular orbit, the magnitude of the
angular
momentum
is
simply
L=
mrv
(1
1-6)
Sec.
ORBITAL MAGNETIC MOMENTS
I]
From
these
magnetic
two equations we
329
see that the ratio
of the magnitude of the
moment to the magnitude of the angular momentum is a constant,
This constant
fi t
evr
2cmrv
(11-7)
line
usually written gifi B jfi, where
is
[i
ehj2mc
0.927
X lO" 20 ergs/gauss
(11-8)
and
gi=l
(H-9)
The quantity (iu forms a natural unit for the measurement of atomic
magnetic moments and is called the Bohr magneton. The quantity g
is called the orbital
g factor. It is introduced, even though it appears
redundant, because we shall later come across a g factor which is not unity.
t
In terms of these quantities,
we may
rewrite equation (11-7) as a vector
equation specifying both the magnitude of
(x,
and
its
orientation relative
to L:
|t,=
-^L
(11-10)
The
ratio of
fi
an
L does not depend on the size of the orbit or on the
By making a calculation similar to the one above for
it can be shown that (i \L is independent of the shape
to
orbital frequency.
elliptical orbit,
That
of the orbit.
completely independent of the details of
this ratio is
the orbit suggests that
its
value might not depend on the details of the
mechanical theory used to evaluate
evaluating
dividing by the
ratio of
(i l
to
it,
and
this is actually the case.
quantum mechanically (which
fx l
is
will
Upon
not be done here), and
quantum mechanical expression L = Vl(l+ 1) ft,
found to have the same value that we obtained from
the
the
Bohr theory. Granting this, the reader will accept that the correct
quantum mechanical expressions for the magnitude and the z component
of the magnetic
(i x
moment
vector are
^L=_ ^
gl/*B T
SiPb
V/(/
+ \)h= g^v-JIV +
1)
(H-ll)
and
K.
= -
^L = - S-^mh = 2
gl!a
Bm
(The minus sign follows from the minus sign in equation 11-10.)
(11-12)
MAGNETIC MOMENTS,
330
2. Effects
SPIN,
RELATIVISTIC EFFECTS
[Ch.
of an External Magnetic Field
Using the definition
with magnetic
(1 1-3), let
moment
us evaluate the forces acting on a system
placed in an external magnetic
field H, which
uniform in the region of the equivalent dipole. This is illustrated in
figure (11-2). In this field there is a force parallel to the direction of
and of magnitude qH acting on the positive pole, and a force of the same
|jl,
is
H
Figure
1-2.
magnetic
moment
in a
uniform external magnetic
field.
magnitude, acting in the opposite direction on the negative pole.
is no net translational force, but there is a torque of magnitude
T = qH - sin
where
to both
There
+ qH - sin = qd H sin = ^H sin 6
|ij
[L and H. The torque vector is perpendicular
and H, and it points in the direction which would tend to align
with H. Its magnitude and direction can thus be expressed by the
the angle between
is
(jl
equation
T=fijXH
(11-13)
H, the system with magnetic moment jx,;
AE which depends on the angle
The energy A.E can be evaluated by summing the
In an external magnetic
field
has an orientational potential energy
between the two vectors.
potential energy of each of the poles of the dipole, if we define the zero of
potential energy to be at its center. For a magnetic field which is uniform
in the region of the dipole, this
is
AE = -zqH - (~z)(-q)H = -2 - cos qH = -^Hcos
(11-14)
EFFECTS OF
Sec. 2]
where
z is the
AN EXTERNAL MAGNETIC
coordinate of the positive pole, and
FIELD
is
331
the coordinate
of the negative pole, both measured along an axis parallel to H.
can be written in terms of the "dot product" notation
AE=
Equation
(1
-|a,-H
(11-15)
1-14) can be taken as the definition of this notation.
AE
orientational potential energy
minimized when
is
This
as
The
and H are
mechanism, the
(a (
However, in the absence of some dissipative
cannot obey the natural tendency to align itself with the vector
H. Instead, (i will precess around H in such a way that 6 and A remain
aligned.
vector
(jL
constant.
This precessional motion
is
a consequence of the fact that the torque
is
always perpendicular to the angular
momentum
vector
since,
from
we have
equations (11-10) and (11-13),
_^ LXH
T=
h
Substituting this into Newton's law of motion for angular
momentum,
dt
we
obtain the equation of motion of the vector L:
-=-^LXH
From
the equation
we
(11-16)
dt
see that the
change
in
in the time dt
is
dL= _?LxH<il
h
These vectors are illustrated in figure (ll-3).f Since dL is perpendicular
to L, its magnitude L remains constant. In the time dt the tip of the vector
L moves through an angle
dL
gll*B
L sin
L, and also
magnitude
Thus
|a ; ,
LH sin 6 dt _
HL sin
precess about
(11-1),
with an angular frequency of
H = b h
H
dt
t This figure, which shows
H dt
gl!*B
L precessing about H,
which shows the electron circulating
in
should not be confused with figure
its orbit.
MAGNETIC MOMENTS,
332
As
the sense of the precession
is
RELATIVISTIC EFFECTS
SPfN,
in the direction of
[Ch.
H, we may write
Wi = &H
(11-17)
phenomenon is known as the Larmor precession, and w L is called the
Larmor frequency. It is quite analogous to the precession of the spin
This
Figure 11-3.
lustrating a magnetic
moment
precessing about an external magnetic
field.
angular
(11-16)
momentum
is
of a top in a gravitational
identical in
field.
form with the equation of a
An experiment capable
In fact, equation
top.
of detecting spatial quantization of the magnetic
moment of an atom must be able to measure /u, or some quantity which
depends on /x, Equation (11-15) suggests placing the atom in a magnetic
field oriented along the direction of the z axis and measuring the potential
energy A =
H = ^ H by observing, for instance, the effect of
the magnetic field on the optical spectrum of the atom. Such an experiment
will be described in Chapter 13 under the heading Zeeman effect. Here
we shall consider instead a different technique which involves sending a
,
(x,
AN EXTERNAL MAGNETIC
EFFECTS OF
Sec. 2]
beam of atoms through a magnetic
amount which depends on fi
FIELD
field that deflects
333
each atom by an
We
have seen above that in a uniform magnetic field there is no net
on the magnetic moment. However, if the
is not uniform there will be a translational force. Consider a magnetic
which is non-uniform, but does not vary with y and is symmetric
translational force acting
field
field
with respect to the yz plane.f In the yz plane the
field
then has a com-
ponent only along the z axis. Assume that a magnetic dipole producing
a magnetic moment (x is straddling the yz plane with \l at an angle 6
paper
axis into
-<7
Figure
1-4.
magnetic
moment
a non-uniform external magnetic field.
in
from the
z axis, as indicated in figure (1 1-4).
the field
is
which
is
negative poles the field in the z direction
,
(H z) +q
x
(H z )_ g
where x
(<5/2) sin
At
the center of the dipole
entirely in the z direction.
,,
=H
+
.
oH,
^ x
is
oH
^
z
+
,
ox
At
the positive
and
approximately
z
oz
d
d
= H + -^(-x)
+ -^(-z)
ox
and
oz
cos
(<5/2)
6.
Thus
dH,
Fz = q(H z) +a - q{H z )_ q = -?qd sin
dx
the net z force
is
dH,
+ ?qd cos
dz
or
dH z
Since
\}. l
is
not fixed in space, we must evaluate the average of
dH
ox
t
Such a
dH z
field is
shown
in figure (11-5).
dU z
oz
z.
It is
MAGNETIC MOMENTS,
334
In the classical theory
quantum mechanics
theories /If
RELATIVISTIC EFFECTS
= because of
= {g^^h) Lx =
7*7"
p~i
since
[i t
SPIN,
it
the
Larmor
because
0.
In
precession.
Lx =
has a fixed value. Consequently
[Ch.
In both
we have
- dH*u
(11-18)
similar calculation shows the average net x component of force to be
F~x = (dHjdz) n but this vanishes because dHJdz = near the yz plane
owing to the symmetry of the field.
%
The Stern-Gerlach Experiment and Electron Spin
3.
In 1922 Stern and Gerlach measured the possible values of /i^ for Ag
atoms by sending a beam of these atoms through a non-uniform magnetic
Figure
1-5.
An
apparatus used to measure the z component of the magnetic
moment
of an atom.
A schematic diagram of their apparatus is shown in figure (11-5).
beam of neutral atoms is formed by evaporating Ag in an oven O.
The beam is collimated by the diaphragms D and enters a magnet N-S.
The cross sectional view of the magnet shows that it produces a field of the
field.
type described in the last section.
Since the atoms are neutral, the only
by equation (11-18).| The force
on the value of p h
acting on each atom of the beam
according to the
components
into
analyzed
for that atom, and the beam is
metallic
plate P,
a
atoms
strike
deflected
The
various values of /i
trace.
visible
leave
a
and
upon which they condense
According to classical theory, p h can have any value from p to
The predictions of quantum mechanics, as summarized by equation
(11-12), are that (i can have only the discrete values
force acting
is
the force
specified
therefore depends
fj, t
/^
-gif*B>n
ionized atoms because they would
Hevjc, acting directly on the charge, which is very much larger
fThe experiment cannot be performed with
experience a force
F=
(11-19)
than the force acting on the magnetic moment.
STERN-GERLACH EXPERIMENT
Sec. 3]
where
is
SPIN
335
one of the integers
m = -1,-1 +
Thus
AND ELECTRON
1,
0,
...,+/-
1,
+/
(11-19')
beam would be spread
quantum mechanics predicts that the
up into several discrete components.
classical theory predicts that the deflected
into a continuous band, while
beam would be split
Furthermore, quantum mechanics predicts that this should happen for
all orientations of the magnet, i.e., for any choice of the direction of the
deflected
Stern and Gerlach found that the beam of Ag atoms is split into
two discrete components one component is bent in the positive z direction,
and the other component is bent in the negative z direction. The experiment was repeated using different orientations of the magnet and several
other kinds of atoms, and it was always found that the deflected beam is
2 axis.
split into
two, or sometimes more, discrete components. These results are,
qualitatively, very direct experimental
proof of the existence of
spatial
quantization.
However, they are not in quantitative agreement with equation (11-19).
According to that equation the number of possible values of n is equal to
the number of possible values of m, which is (21 +1). Since / is an integer,
this is always an odd number. Also, for any value of / one of the possible
values of m is zero. Thus the fact that the beam of Ag atoms is split into
l
only two components, both of which are deflected, implies either that
something
is
wrong with the Schroedinger theory of quantum mechanics,
or that the theory
is
incomplete.
The theory is not wrong, but, as it stands, it is incomplete. This is shown
most clearly by an experiment performed in 1927 by Phipps and Taylor,
who used the Stern-Gerlach technique on a beam of hydrogen atoms.
is particularly significant because the atoms contain a
and the theory we have developed makes the unambiguous
prediction that, the atoms being in the ground state, the quantum number
/ will have the value / = 0.
Then there is only one possible value of m,
namely m = 0, and we expect that the beam will be unaffected by the
magnetic field since //j will be equal to zero. However, Phipps and Taylor
found that the beam is split into two symmetrically deflected components.
Thus there is certainly some magnetic moment in the atom which we have
This experiment
single electron
not hitherto considered.
One possibility would be a magnetic moment due to a spinning motion
of the charged nucleus about its own axis. The magnitude of such a
magnetic moment would be of the order of ehjlMc, where
is the mass
of the proton. But the magnetic moment measured experimentally (from
the size of the splitting) is of the order of \i B = ehjlmc, which is a thousand
MAGNETIC MOMENTS,
336
RELATIVISTIC EFFECTS
SPIN,
[Ch.
times larger. Therefore the nucleus cannot be responsible for the observed
magnetic
moment
its
source must be the electron.
This leads us to the reasonable assumption, which
supported by
is
other evidence to be discussed later, that an electron has an intrinsic
magnetic moment
[i s
We
called the spin.
Sz
s
due
of the spin angular
and
to the existence
of an
angular momentum
S and 2 component
two quantum numbers
intrinsic
also assume that the magnitude
momentum,
are related to
by the equations
Sz
=m
Vs(s
s
+\)h
(11-20)
(11-21)
which are identical with those for orbital angular momentum, and that
the relation between the spin magnetic moment and the spin angular
momentum is of the same form as the relation for the orbital case; i.e.,
_gs^B s
iLs=
(u _22)
K=
The quantity gs
is
cally deflected
g factor. Now, from
called the spin
observation that the
beam
of hydrogen atoms
components,
(11-23)
-gsf^B^s
is split
apparent that
it is
/a,
the experimental
into
two symmetri-
can assume just two
values which are equal in magnitude but opposite in sign. If
we make
the
assumption that the possible values of m s differ by unity and range
from s to +s, as is true of the quantum numbers for orbital angular
momentum, then we conclude that the two possible values of m s are
final
m
and that
=-h+\
(11-24)
s has the single value
By measuring
the splitting of the
to evaluate the magnetic force
in traversing the magnetic field.
(H-25)
beam of hydrogen atoms, it is possible
which the hydrogen atoms experience
From equations (11-18) and (11-23),
this is
Fz= -
~T-l*Bgsm s
dz
Since
ix
is
known and dHjdz can be measured,
evaluation of the quantity
is
gm
s
this
equation allows the
Within the accuracy of the experiment
found that
gs m s
it
STERN-GERLACH EXPERIMENT AND ELECTRON SPIN
Sec. 3]
Since
we have concluded
that
m =
s
gs
\,
Thus the spin g factor for an electron
O n lne other hand, \i s and S
are,
this implies
factor.j
337
(11-26)
twice the value of
is
because the relative orientation of either
orbital
its
and L
pair of vectors depends only
are antiparallel, just as
{*,
on the fact that an electron is negatively charged.
The addition of electron spin to the quantum mechanical theory requires
the addition of the quantum number m s to the set n, /, m, in order to
identify completely the mathematical form of an eigenfunction. However,
it is not necessary to indicate specifically the value of the quantum
number s since it always has the same value, J. Thus an eigenfunction
must carry four subscripts, and it is written
(11-27)
Wnlm.m,
where we use the symbol m for the symbol we have previously written as
m in order to have a symmetrical notation. Such an eigenfunction contains
the information that the z component of the spin is +f h when m = +,
and is \h when m = \. The eigenfunctions y m m can be written
t
as a product of a spatial eigenfunction
y nlmi times a
spin eigenfunction
However, the mathematical structure of the spin eigenfunctions is
m>
quite different from that of the spatial eigenfunctions. The xp nlm are
<T
the continuous functions of the continuous variables
r, 6, <j>, which we
have studied; but the "spin variable" can only have two values corresponding to the two possible values of m s so the a m are not continuous
functions. There are several ways of representing these functions, the
most commonly used being the Pauli matrix representation in which they
are written as two row, single column matrices. This representation is
,
very closely related to the Heisenberg matrix formulation of
quantum
mechanics mentioned in section 1, Chapter 7, and is treated in advanced
books on quantum mechanics. Since for our purposes it is possible to
avoid situations in which it is necessary to know the explicit form of the
spin, eigenfunctions,
As
is
implied by
they will not be treated in this book.
its
name, the spin angular
of as analogous to the angular
momentum
momentum S
can be thought
of a particle spinning about an
axis through its center. Using this in the Bohr theory, we can make an
analogy between the motion of a spinning electron around its nucleus
and the motion of a spinning planet around its sun. However, this analogy
must not be taken too seriously as electron spin is an essentially non-classical
t
is
Very accurate experiments by
actually very slightly different
Lamb
from
(1947), using different techniques,
2.
Its precise value
is^ s
show
2.00232.
that
338
MAGNETIC MOMENTS,
SPIN,
RELATIVISTIC EFFECTS
[Ch.
The reason is that the quantum number s which specifies the
magnitude of the spin angular momentum has the fixed value \. Therefore
we cannot take S to the classical limit by letting s oo, as we did in the
last chapter for L by letting / > oo. Another way of saying the same thing
quantity.
>-
magnitude of the spin S
V\{\
is
that in the classical limit the
is
completely negligible, and so spin must be essentially non-classical.
Although electron spin
is
1)
completely compatible with the Schroedinger
theory of quantum mechanics, the theory does not predict
it
must be introduced
associated with the fact that the
which ignores
developed a
its existence and
The reason for this is
Schroedinger theory is an approximation
as a separate postulate.
relativistic effects (cf. section 2,
relativistic
Chapter
7).
In 1928 Dirac
theory of quantum mechanics. Utilizing essentially
same postulates as the Schroedinger theory, plus the requirement
quantum mechanics conform with the theory of relativity, Dirac
showed that an electron must have an inherent angular momentum and
the
that
magnetic
moment
with exactly the properties described above.
This put
on a firm theoretical foundation. A quantitative treatment
of the Dirac theory would not be appropriate in this book, and we shall
continue discussing spin as an ad hoc addition to the Schroedinger theory.
However, some of the more interesting features of the Dirac theory will be
electron spin
described qualitatively in the following chapters.
4.
The Spin-Orbit
Interaction
The existence of electron spin actually was postulated first by Goudsmit
and Uhlenbeck in 1925. These gentlemen were trying to understand why
certain lines of the optical spectra of hydrogen and the alkali elements are
composed of a closely spaced pair of lines. This is the fine structure,
which had been treated by Sommerfeld in the old quantum theory as due
to a splitting of the atomic energy levels because of a small contribution to
the total energy resulting
from the
relativistic variation
The
of electron mass
Sommerfeld
by equation (5-22), were in good numerical agreement
with the fine structure of the hydrogen spectrum. But they were also in
good agreement with the fine structure of the alkali element spectra.
This was perplexing because in the old quantum theory the spectroscopically active electron of an alkali element (an element with a single valence
electron) would be moving in an orbit of large radius at relatively low
with velocity
(cf.
section 7, Chapter
5).
results of the
theory, as expressed
be negligible. Consequently
Sommerfeld's explanation of the
velocity, so that relativistic corrections should
doubt arose concerning the
validity of
origin of the fine structure.
In considering other possibilities, Goudsmit
THE SPIN-ORBIT INTERACTION
Sec. 4]
339
and Uhlenbeck proposed that an electron has an inherent angular
momentum and magnetic moment, whose z components are specified by a
quantum number m s which can assume the two values \ and +\.
The splitting of the atomic energy levels could then be understood as due
to an orientational potential energy of the inherent electron magnetic
moment in the atom's own magnetic field, which would be either positive
or negative depending on the sign of m s
.
we
In this section
shall evaluate the orientational potential energy,
+Ze
Figure
in
1-6.
a circular
An
electron moving
Bohr
orbit,
as,
Figure
seen
1-7.
in a circular
An
electron moving
Bohr
orbit, as seen
by the electron.
by the nucleus.
by a calculation which involves a judicious
mixture of the Bohr theory and the theory of relativity. The justification
of this calculation is found in the fact that its results agree with those of a
called the spin-orbit energy,
rigorous calculation using the Dirac theory.
Consider an electron of charge
nucleus of charge
+Ze,
as
shown
moving
electron with respect to the nucleus be
nucleus be
around
it
r.
From the point of view
with velocity v. This
nucleus moving with velocity
According to Ampere's law,
position of the electron,
is
on
is
and
this
its
around a
position relative to the
shown
is moving
The charged
in figure (11-7).
a current element
Zev/c
produces a magnetic
field
Ze
vX
which, at the
convenient to express this in terms of the electric
the electron.
orbit
is
H== iXj
It
v,
Bohr
Let the velocity of the
of the electron, the nucleus
v constitutes
j
in a
in figure (1 1-6).
According to Coulomb's law,
E= +Ze
field
acting
MAGNETIC MOMENTS,
340
From
the last two equations
RELATIVISTIC EFFECTS
SPIN,
[Ch.
we have
H=--vXE
(11-28)
From
a study of the behavior of Maxwell's equations under the appli-
cation of a Lorentz transformation,
general validity; any particle
electric field
Figure
1-8.
E will
it
can be shown that
moving with
experience the magnetic field
The frames of reference used
moment
equation
is
of
an arbitrary
H given by the equation.
in calculating
According to equation (11-17), the magnetic
precession of the spin magnetic
this
velocity v through
the Thomas precession.
field will
cause a
Larmor
with angular frequency
(11-29)
where we have replaced g by gs to obtain the correct value of the spin
magnetic moment. The magnetic field will, according to equation (1 1-15),
also produce an orientational potential energy
t
^E =
-{*s
gst*B
S-H
(11-30)
Both of these quantities have been evaluated in a frame of reference in
which the electron is at rest, but what we are really interested in is their
values in the normal frame of reference in which the nucleus is at rest.
Although it is not obvious that transforming back to the normal frame of
reference will change the apparent values of w L or AE, this really happens
due to a kinematical effect of the theory of relativity called the Thomas
precession.
Consider the situation as seen by an observer in the frame of reference
which the nucleus is at rest. Figure (11-8) illustrates this. The
electron is momentarily at rest in the frame
at the instant tx and
xy, in
x^
THE SPIN-ORBIT INTERACTION
Sec. 4]
momentarily
at rest in the
341
frame x 2 y2 at the
slightly later instant
t2 .
The axes of both xy and x2y 2 have been constructed parallel to the axes
of x x y x as seen by an observer instantaneously moving with the electron
in x x y x Nevertheless, we shall show that the observer in xy sees the axes of
,
own axes. He
frame
which the
electron is instantaneously at rest are precessing, relative to his own set of
axes, as the electron goes around the nucleus
even though the observers
instantaneously at rest relative to the electron contend that each set of
axes x n+1 y n+1 is parallel to the preceding set x n y n
x 2 y2 rotated
slightly relative to his
xjy3 rotated even more,
Thus he
etc.
sees the axes of the
sees that the set of axes in
Figure
in
1-9.
The frames
of reference used in calculating the
Thomas
precession, as seen
the xjrj frame.
Figure (11-9) shows xy, xx yx and x 2y2 from the point of view of the
observer in x t yv Since the electron is moving with velocity v relative to
,
the nucleus, the axes xy are
moving with velocity
negative x x axis relative to x x y x
As
v in the direction of the
seen in x-$ x the electron
,
is
accelerating
toward the nucleus with acceleration a in the direction of the positive
tx is very small, the change in velocity of
axis. If the time interval t
yx
the electron in that interval
is
dv
a(t 2
t-j)
a dt
(11-31)
and this will be the velocity of x$2 as seen by xxyv Now let us use equations
(1-16) to evaluate the components of VOJ the velocity of x 2 y 2 as seen by
xy.
These give
V
dv x
vdv
1
dvjl V
=
1
vx
%=
VndV.,
- -v-0
dvjl
342
MAGNETIC MOMENTS,
SPIN,
RELATIVISTIC EFFECTS
[Ch.
Using equations (1-16) to evaluate the components of V, the velocity of xy
by x$2 we have
as seen
dvl
vhx =
dvy vy
c
"
vv
n =
1-10.
The reader may
of
V;,,
dv
c
= -Jl-
dv 2
c
dv
dvv vy
Figure
dvv
dv 2
1.
An
exaggerated illustration of the Thomas precession.
verify that the
magnitude of V
is
equal to the magnitude
in conformity with Einstein's postulate of the equivalence of
inertial frames.
and the x
Now
axis of the
let
us calculate the angle between the vector V
xy frame.
It is
*y-3
The angle between the vector
and the x
e.-
axis of the
x$2
frame
is
dv
I.
dv 2
Figure (1 1-10) shows the x^y2 and xy frames from the point of view of xy.
According to the postulate of equivalence, V and V6 must be exactly
opposite in direction. Since the angles between the x axes and the relative
THE SPIN-ORBIT INTERACTION
Sec. 4]
343
x$2 frame appears to be rotated
xy frame. The angle of rotation is
velocity vectors are not the same, the
relative to the
dd
6h
dv
ft,
-fA
d
dv<
As dv
is
differential,
we may
d6
Now
first
we may
Then
-tV"
v represents the velocity of
Therefore
the
neglect dv 2 jc 2
an electron in an atom, so v2 /c 2
two terms in the binomial expansion of
d6
<
10~ 4
obtain a very good approximation to dd by taking only
-M
dv
v
dvv2
2vc
dv
~2#
v
Vl
v 2 \c 2
We
have
4-4
dv
v
va dt
where we have evaluated dv from equation (1 1-31). The axes in which the
electron is instantaneously at rest appear to precess, relative to the nucleus,
with angular frequency
oT =
This frequency
is
called the
dd/dt
va\2c 2
Thomas frequency. Inspection of the figures
by the vector equation
will verify that the sense of the precession is given
Ui
T=~^**
Relative to frames in which the electron
moment
is
(11-32)
at rest, its spin magnetic
precesses with angular frequency u> L
But these frames are
themselves precessing with angular frequency (o T relative to the normal
frame in which the nucleus is at rest. Consequently the spin magnetic
moment is seen in the normal frame to precess with angular frequency
(11-33)
According to equations (11-28) and (11-29),
Wr =
gs/*B
ch
vx E
MAGNETIC MOMENTS,
344
Evaluating
pi
2eA
evaluate <a T
(11-34)
mc 2
we may use Newton's law
[Ch.
we have
(11-8) and (11-26),
vxE=--^-vXE
chime
To
RELATIVISTIC EFFECTS
SPIN,
B and gs from equations
to express the acceleration of
the electron in terms of the electric field
= eE/w
With
this,
equation (11-32) becomes
<o r
Thus the precessional frequency
o>=
-^vXE
2mc*
in the
(11-35)
normal frame of reference
--^VXE
+ -^VXE= --i-vXE
mc
2mc
is
(11-36)
2mc
first derived in 1926 by Thomas.
Comparing equations (11-36) and (11-34), we see that transforming the
spin precession frequency from the frame in which the electron is at rest
to the normal frame reduces its magnitude by exactly a factor of 2. The
same is true of the orientational potential energy since, as can be seen from
equations (11-29) and (11-30), the magnitude of that quantity is pro-
This equation was
portional to the magnitude of the precession frequency.
Therefore the
value of the orientational potential energy in the normal frame of reference,
which we
shall designate
by the symbol
A s
It is
we
AEa
L is
,
i^S-H
(11-37)
convenient to express this in terms of the quantity
use equation
(1
L.
To
this
end
1-28) to write
H=
- ivx E
c
and express
in terms of the force
acting
on the
electron, using
-eE = F
In the
Coulomb
potential of the nucleus or,
spherically symmetric potential V(r), the force
dV(r) r
dr
more
is
generally, in
any
THE SPIN-ORBIT INTERACTION
Sec. 4]
Combining
we have
these three equations,
H=
345
v X
dV(r)
dr
ec r
Multiplying the numerator and denominator by m, and using equation
(11-4),
we
obtainf
H = J_ldKr) L
emc
(U38)
dr
Thus equation (11-37) can be written
^-i^S'
=
A SL
s L
.
gs and
Evaluating
[x
B we
,
2,
lemch
dr
obtain
i^H^S-L
= _LAESL
S L
2 2
(11-39)
2m
dr
Although our calculation of the spin-orbit energy was essentially based on
Bohr theory, the results, as expressed by equation (11-39), are in
exact agreement with the results of a relativistically correct quantum
the
mechanical treatment.
Let us make an estimate of AZs s L for a typical state of a one-electron
atom, to check whether it is of the same order of magnitude as the observed
fine structure splitting of the corresponding energy level.
Consider
hydrogen in the n = 2, / = 1 state. The potential V(r) is
.
V(r)
= -e
2 /- 1
so
dr
and
A s
-^-S
2m
1_
'
2 2
The magnitude of S L
for the n
Since
approximately h 2 The average value of
approximately 1/2 3 Aq, where a = fr 2 /me 2 So
2 state
is
As-lI
~ ^^
S L can be
is
yyPp^
~ 10-
16
ergs
either positive or negative,
From
~ lO^ev
m = +\ or J, the energy level is split by about 2
t
1/r
p2
the definition of the cross product
r
it is
= v X
depending on whether
10~ 4 ev. Comparing
apparent that
r
MAGNETIC MOMENTS,
346
with the energy of the n
this
we
ev,
RELATIVISTIC EFFECTS
SPIN,
2, /
Ch.
level of hydrogen, E2 = 3.4
about one part in 104 in good
see that the predicted splitting is
agreement with the splitting required to explain the fine structure of the
lines of the hydrogen spectrum associated with this level.
estimate the magnitude of the magnetic field
It is also interesting to
acting
state
on the spin magnetic moment of an electron
From
of hydrogen.
where
Fs
equation (11-30)
~ Pb ~ 10
-20
in the n
2, /
H
1
we have
erg-gauss
-1
Therefore
H ~ wl(T
10
This
is
a strong magnetic
Total Angular
5.
16
ergs
erg-gauss
lrv4
10*
gauss
field.
Momentum
In the absence of the spin-orbit interaction, and of any external torques,
the magnitudes and z components of the spin and the orbital angular
momentum
vectors
quantum numbers
L and S would have the fixed values specified by the
m m .| But in actuality there is a strong magnetic
/,
determined by L, that produces a torque
orientation of which is determined by
S. We have seen in section 2 that such a torque will not change the
magnitude of the vector S. Nor will the reaction torque acting on L
change its magnitude. But this torque does provide a coupling between
L and S which makes the orientation of each vector dependent on the
orientation of the other. As a result, neither of the vectors will have a
field,
on
the orientation of which
the spin magnetic
fixed z
is
moment, the
component because both
will precess
J
which
is
their vectorial
momentum
sum.
=L+
about the vector
The vector J
(11-40)
is
called the total angular
vector.
magnitude and all three components
of the total angular momentum vector would be fixed if there were no
external torques acting on the atom. This is only partly true in quantum
mechanics because, as the reader might guess, this theory predicts that
According to
classical physics, the
t It is not necessary to specify the value of the quantum number
have the value s = i.
s since it
can only
TOTAL ANGULAR MOMENTUM
Sec. 5]
the magnitude and z
component of J
those
we used
vector,
it
angular
are fixed, whereas
By
ponents have no definite value.
347
its
x and y com-
using techniques closely related to
in studying the properties of the orbital angular
momentum
can be shown that the magnitude and z component of the
momentum
total
quantum numbers j and
vector are specified by two
according to the equations!
j=Vj(j+i)h
Jz
(11-41)
mfi
where
m,
-j
-j,
... ,j
1,
(1
1,7
1-42)
Let us determine the possible values of the quantum number j.
Taking the z component of equation (11-40), we have
/,
which
=L +S
z
is
m-h
mfi
mfi
or
m =m +m
(11-43)
the maximum possible value of m is
and the maximum possible
of m is s = \, the maximum possible value of m, is
j
Since
value
/,
K)max = +
According to equation (11-42),
of/
this is also
equal to the
maximum possible
common
with the other quantum numbers we have been
discussing, the possible values of j differ by integers. Therefore these
value
In
values must be
members of the
J
To determine
series
25*
2>'
'
'
2>
5
2>
the point at which this series terminates,
we may
use the
vector inequality
|L
the validity of which
may
S|
easily
>
||L|
|S||
be demonstrated by constructing some
simple vector diagrams. This inequality gives
|J|>||L|-|S||
or
Vj(j
l)
> |v7(/ +\)h - Vm +
i) h\
t In the Heisenberg matrix formulation of quantum mechanics it is proved that all
angular momentum vectors (orbital, spin, and all combinations) obey quantization
equations of exactly the same form.
MAGNETIC MOMENTS,
348
We
leave
the two
it
SPIN,
RELATIVISTIC EFFECTS
an exercise for the reader to show that when
as
members of the
of j
only
series
= /+*,/-*
When = 0, L = 0, and
(n-44)
satisfy this inequality.
[Ch.
the only possible value
is
j=h
The content of equations
(11-42), (11-44),
(11-44')
and (11-44') can be formally
represented in terms of the rules of vector addition,
if
we
construct a
II
1
'
CO
OJCVJ
II
set
of vectors whose lengths are proportional to the values of the quantum
As an example, consider 1 = 2. Then the two possible
/, s, and/'.
numbers
values of j are f and
For/ =
f.
For/
m = f, \,
+\, +f.
ntj
= f, f, i,
+i, +f, +f.
Vector addition diagrams for the
two values of/ are shown in figure (11-11). The interpretation of these
diagrams needs no discussion. The diagrams represent only the rules for
adding the quantum numbers / and s to obtain the possible values of the
quantum numbers j and w If the relation between the magnitude of an
angular momentum vector, such as L, and its associated quantum number
.
were
L=
Ih
1 = Vl(l + 1) H, these diagrams would also
of L and S to obtain J and Jz Although this is not
instead of
represent the addition
true, such diagrams are often presented in elementary discussions of
atomic structure as a simplified representation of the addition of the
vectors. This is a useful representation to keep in mind, but only if it is
TOTAL ANGULAR MOMENTUM
Sec. 5]
also kept in
mind
that
349
an approximation, and that a more accurate
it is
representation of the addition of these vectors would have an appearance
similar to that
The
shown
in figure (11-12) for the case
figure attempts to
vector J, which in turn
When the
is
the vectors
Illustrating
and
the
L and S
2,j
f, ntj
f.
precessing about the
precessing about the z axis.
spin-orbit interaction
quantum numbers
Figure 11-12.
show
is
considered,
we may no
to identify the function
L and S
y>
longer use the
because
it is
no
vectors precessing about the J vector, and the
J vector precessing about the z axis.
longer an eigenfunction of the operators associated with the dynamical
Lz and Sz That is to say, L z and Sz do not have definite values
and so we cannot characterize y> by definite values of m and m s But
in losing m and m s as good quantum numbers, we gain an appreciation of
the fact that the quantum numbers j and m s can be used instead. As
quantities
the definite values vj(j + 1) h and m s h, it is apparent that
must be an eigenfunction of the operators associated with these dynamical
J and Jz have
ip
quantities.
Therefore the eigenfunctions can be written
Vniim,
(H-45)
where the four quantum numbers identify the form of the function.
Furthermore, these quantum numbers can be used and the functions (1 1-45)
are proper solutions to the Schroedinger equation, even if there were no
spin-orbit interaction, since we know that / and Jz would certainly still
have definite values because there are no external torques acting on the
atom.
This brings up the interesting point that there is a redundancy of
quantum numbers if the spin-orbit interaction is neglected. That is,
n, I, m m s ,j, ntj would all be good quantum numbers, but, since there are
only three spatial coordinates and one spin coordinate, only four quantum
t,
MAGNETIC MOMENTS,
350
RELATIVISTIC EFFECTS
SPIN,
[Ch.
numbers are needed. In other words, if we ignore the spin-orbit interaction, we know from (11-27) that there are solutions to the Schroedinger
equation, written
(11-46)
Wnlm.m,
which are eigenfunctions of the operators associated with: the total
energy, the magnitude of the orbital angular momentum, the z component
of the orbital angular momentum, and the
component of the spin angular
momentum.
However, there are also solutions to the Schroedinger
equation, written as (11-45), which are eigenfunctions of the operators
associated with
the total energy, the magnitude of the orbital angular
momentum, the magnitude of the total angular momentum, and the z
component of the total angular momentum. The relation between the
y> nlrn m and the f nUm can best be understood by considering the procedures involved in evaluating the spin-orbit energy AEa t by the methods
:
of perturbation theory.
In using these methods
we
first
solve the
Schroedinger equation,
and obtain a set of eigenfunctions,
say the xp nlm m which we found first. Then these unperturbed eigenHowever, the y>m mim are
functions can be used to evaluate Ais s L
degenerate since the total energy of the state specified by the quantum
numbers nlm m s depends only on the quantum number n. Therefore it
neglecting the spin-orbit interaction,
is
necessary to follow the procedures of degenerate perturbation theory,
which require a preliminary investigation of the way that the perturbing
spin-orbit interaction mixes those eigenfunctions that are degenerate with
each other.
From this we
obtain a
new
of certain linear combinations of the
it is
found that the new
shall not bother to write
of the
know
yj nlm
set
of eigenfunctions which consist
In carrying this through,
nlmim
set are precisely the eigenfunctions f nUmj
set
ip
down
the
y> nljm
since they are quite complicated
their explicit form.
It is
more important
and we
that
shall
we make
that the set of degenerate unperturbed eigenfunctions
fn
not need to
sure that the
reader understands the physical significance of these results.
is
We
in terms of linear combinations
The point
mi m 3
cannot
be used directly in a perturbation calculation of the spin-orbit energy,
because in the presence of this small perturbation L z and S z have no
definite values. Thus the effect of the perturbation is to mix completely
m and m s forming a new set of functions.
new set of functions are just the rp n im which describe the fact that J
and/z have definite values even in the presence of the spin-orbit interaction.
functions with different values of
This
.,
This discussion suggests that the evaluation of the spin-orbit energy
might be considerably simplified by
degenerate unperturbed eigenfunctions
initially
ip
nUm
..
starting with the set of
Since J and
Jz have
definite
TOTAL ANGULAR MOMENTUM
Sec. 5]
351
values whether or not the spin-orbit interaction
is
present,
it is
apparent
that the application of this perturbation cannot change their values.
Consequently the perturbation cannot produce a large mixing of the
Vnijm' even though they are degenerate. In this case non-degenerate
perturbation theory can be used directly (cf. section 4, Chapter 9).
According to that theory (equation 9-13 and the following discussion
extended to three dimensions), the approximate value of the shift in the
energy of the state specified by the quantum numbers nljrrij, due to the
presence of the spin-orbit interaction energy Ais g L is
,
A s -l
which
&ES
H
Using
L in that state.
we have
aeTl
This
vIi*m,(AEi.L)opV'w/ asin e dr dd
equal to the expectation value of
is
(1 1-39),
=J J J
JjJ^ (dv 7 ^r s
L^-<
r2 sin d
dr de d +
is
As
'
=
2rf?
J J J
f *" mi
~r
~d?
(S
8
L)opV '""""'" sin 6 dr d6 d t
'
(11-47)
For a Coulomb potential V(r)
becomes
= -Ze /r,
2
so dV(r)jdr
= Ze^jr 2
and
itself,
and
this
AE,. L
To
Ze 2
evaluate (S
}))
Vtiim,
(S
L) op y Wmi r 2
sin
dr dd dj>
L) op consider the equation
,
=L+
Taking the dot product of this equation times
employing the fact that L S = S L, we have
defining J.
JJ = LL + S-S + 2S-L
So
\(3
S-L =
K-/
SL
This
-L L-
S S)
is
2
- -S
2
(11-48)
Therefore
(S
L) 0P
W -L*- S%
2
i(J"ov
- L% -
S%)
MAGNETIC MOMENTS,
352
RELATIVISTIC EFFECTS
SPIN,
[Ch.
Then we have
Ze 2
L
Now
4m 2 c 2
to the eigenvalues j(j
2
(J o P
( J2
P
-L
W+
*)&,
S 20V) Wnlimj
L2P
an eigenfunction of yp
is
v'nwm.
Vllimh
J J J
^ dd H
L^, and Sfp corresponding
and |(| + l)^2 .f Therefore
,
!)#*>
[j(;
S Op)VWm/ Sin
1)
/(/
sin
1)
*]Vh,
and we have
As
L=
2
Ze,2fc2
/r
"
''
D +1) -
/(/+1)
f]
x jjjvtum,
The
and
Thus
integrals over
properties of the
y>,
^^ = ri
IJO'
Am^c*
The
1)
^+
1)
I]
Note that when
"'
this
(11-49)
- Rm(ry dr
Jf**^) r
It gives
Rtir)RJrydr=r-*=
3
dr dd dj>
both give unity owing to the normalization
<
can be evaluated.
integral over r
V>m im
'"
aln%l
equation says r~ 3 -
oo.
)(/
For
1)
this value
of
/,
- /(/+ 1) |] = 0. Thus AEa L appears to be
= 0. However, this is merely the result of evaluating
r~ s in perturbation theory. A more accurate evaluation of this quantity
shows that is very large, but finite, for = 0. Since [j(j + 1) /(/ + 1)
] is strictly zero in this case, we have
=
| and [j(j+ 1)
indeterminate for /
it
AS
For other values of
/ it is
^ =
S'
0,
z
[j(j
4m 2 c2 a 3
+
n
l)
l(l
=
+
(11-50)
+ l) \){l + 1)
i(l
In terms of thefine structure constant a
e^\hc
value of the unperturbed energy of the state,
t All these functions are eigenfunctions of 5 p since
,
'
Hi +
DA*.
1/137,
f]
and the absolute
En = mZ2e4/2/j
S 1 can have only
2
,
this
the single value
Sec. 6]
RELATIVISTIC
can be written
^ = Z
*'
The value of
|E W | [j(j
2n
353
- /(/ +
+ |)(/ +
1)
1(1
1)
= +
I
the situations in which
AES
AS
\ or j
is
f]
1)
on the value of/ The two
which correspond respectively to
=
/
\,
roughly parallel to
L for these possibilities,
+
L
=
_
we
+^
1)
L"Jf
1)(/
or antiparallel to L.
find
Z2|J
,
n(2l
= + i'#0
= _
'
(11-51)
7 2 IF
l2
z
^"^
n/(2/+l)
6.
the spin-orbit energy depends
possibilities are
Evaluating
CORRECTIONS
Relativistic Corrections for
In section 4 of this chapter
|,
#o
One-Electron Atoms
we estimated
the shift in the energy of a
hydrogen atom, due to the spin-orbit interaction, to
be about 10~4 times the energy of the level. In section 7 of Chapter 5
we estimated that the shift in the energy of such a level, due to the relativistic dependence of mass on velocity, is also about 10~ 4 times the energy
of the level. The energy shifts due to relativistic effects other than the spin
are thus quite comparable to those due to the spin-orbit interaction for
the hydrogen atom, and we cannot expect to be able to compare the
typical level of the
predictions of equation (1 1-50) or
(1
1-51) with the experimental data until
A complete treatment of the
can only be given in terms of the Dirac theory; but
which are almost complete can be obtained with the Schroedinger
we have taken
these effects into account.
relativistic effects
results
theory by treating the additional relativistic corrections as a perturbation.
The
starting point of Schroedinger
quantum mechanics
writing the classical expression for E, which
kinetic energy
T and
the potential energy V.
is
defined as the
That
consists in
sum of
the
is,
E=T +V = ^-+V
(H-52)
relativistically correct since,
according to equations
2m
This equation
(1-22')
and
is
not
(1-25),
where we use
for the rest mass.
Let us write the
expression for E, and then evaluate
it
relativistically correct
in the limit of small values of p.
MAGNETIC MOMENTS,
354
RELATIVISTIC EFFECTS
SPIN,
[Ch.
This procedure gives
+V = (cV + mV) H - mc
nei
uei
= mc 2
[(' +
Re
=i
Enel
Tnei
;&
+ F
2mV
8mV
2m
+ V
\\4
,2
m&
+V
8m 4c 4
+ K
We have taken the first three terms in the binomial expansion.
this
for
with equation (11-52),
E is
we
Comparing
see that the error in the classical expression
approximately
(E
AE Rel
8mV
E 2 + V 2 - 2EV
- Vf
2mc 2
2mc
2mc
(11-53)
If we use the eigenfunctions y> ljm .,
we can use non-degenerate
perturbation
E of the corresponding state
theory to evaluate the shift in the total energy
due to the addition of the term AE^, which may be thought of as an
addition to V. According to the theory, the approximate value of the
energy
shift is
AReI
"wJJ
Now
y=
wt lims(E
-Ze 2 jr
+V
2EV) 0V y> nljm r2
E = En
and
a constant
So we obtain
zv
El
A Be = 2
i
rcc
2mc 2 JJJ
2mc
2T
Tn " mt
'" Jmi
J J JvXw, ; V /
sin
rfr
d6
d<f>
Evaluation of the integrals yields
Aaei
z2
31^
|.
...
P
i
A)
(11 _54)
Comparison of equations (11-51) and (11-54) shows that the energy shift
calculated is, in fact, comparable in magnitude to the energy
we have just
RELATIVISTIC
Sec. 6]
CORRECTIONS
355
due to the spin-orbit interaction. Since both quantities have the
same Z dependence, this is true for all one-electron atoms.
Adding the two energy shifts to the unperturbed energy, we obtain an
shift
expression for the total energy,
E=En + AE
+ ABel
of the state specified by the quantum numbers
(21
!)(/
+1)
It is
nljrrij.
21
An
j=l + hJ^0
-*4^(rh-h)\
+
and
For both values of j
For
0,
Ag
this
can be written
and
E = En + ABel
which
is
^Z 2e 4
'-
"Sbl'+^-s)}2/rV
<u - 56)
where we have evaluated En using the reduced electron mass fi.
Let us compare these results with the results of Sommerfeld's relativistic
Bohr theory of the energy levels of a one-electron atom, given in equation
,
(5-22).
This
is
Shi'
gZV
E_
2h
J,
ri
t
+ vii-s;)}
(11 - 57>
The quantum number n e appearing in this expression can assume the values
n e = 1, 2, 3,
n.
Since the quantum number j can assume the
.
values j
n
\,\ the quantity j
\ can assume the values
n. Thus, for /
1, 2, 3,
\
0, the values of total energy predicted by our calculations are in exact agreement with the values predicted
i,
-f ,
f
.
by Sommerfeld. For
For /
0,y
value of j
is
there
i; otherwise y
is
=/
and the maximum value
is
a discrepancy.
J.
Since
But
0, 1, 2,
!+=
\.
this
,
discrepancy
1,
the
is
minimum
MAGNETIC MOMENTS,
356
SPIN,
RELATIVISTIC EFFECTS
[Ch.
removed by the Dirac theory, which shows that there is one more relativistic correction to the energy, whose existence we could not have guessed
because it has no classical analog. This correction is zero for / ^ 0. For
its value is such that, when it is added to the right side of (1 1-56),
/ =
the resulting equation has the same form as equation (11-55). Therefore
the results of a complete relativistic treatment by the Dirac theory can be
expressed by the single equation
=
These
/.ZVf.
-fcl
results are in exact
theory.
+
,
ZV/
(jTi-J)
1
3\\
, co ,
(11 - 58)
agreement with the predictions of Sommerfeld's
Since the Sommerfeld theory was based on the
Bohr
theory,
it is
only a very rough approximation to physical reality. In contrast, the Dirac
theory represents an extremely refined expression of our understanding
of physical reality. From this point of view the agreement between equa-
and (11-58) is one of the most amazing coincidences to be
found in the study of physics.
The energy levels of the hydrogen atom, as predicted by the Bohr,
Sommerfeld, and Dirac theories, are shown in figure (11-13) for the first
tions (11-57)
three values of n.
In order to
make
the energy level splittings visible, the
displacements of the Sommerfeld and Dirac energy levels from those given
by the Bohr theory have been exaggerated by a factor of (137) 2 = 1.88 x
104 Thus the diagrams would be completely to scale if the value of the
fine structure constant were 1 instead of 1/137. The values of the quantum
number w which specifies the spatial orientation of the total angular
momentum vector, are not shown on the Dirac energy level diagram
because the energy of the atom does not depend on this quantum number
(unless the atom is in an external electric or magnetic field). Although we
have not mentioned it before, a quantum number analogous to w is
required in the Sommerfeld theory to specify the spatial orientation of the
plane of the orbit. Since the energy of the atom does not depend on this
quantum number (if the atom is not in an external field), its values are not
indicated on the Sommerfeld diagram. Also not shown are the energy
levels of hydrogen measured experimentally by optical spectroscopy,
which are in extremely good agreement with the levels of the Sommerfeld
and Dirac theories.
The only essential difference between the results of these two theories
is that the Dirac theory predicts that for most levels there is a degeneracy,
.
in addition to the trivial degeneracy with respect to spatial orientation
mentioned above, because the energy depends on the quantum numbers n
and j but not on the quantum number /. Since there are generally two
values of / corresponding to the same value of/, the Dirac theory predicts
RELATIVISTIC
Sec. 6]
CORRECTIONS
that most levels are actually double.
mentally in 1947 by Lamb,
357
This prediction was verified experi-
who showed
that for n
are two levels, which actually do not quite coincide!-
above the
and the n
2 and j
The / =
\ there
level lies
= level by about one-tenth the separation between that level
= 1,1 = 1, j = f level. This is called the Lamb shift. It can
Bohr
Sommerfeld
Dirac
n = 3,i=2,/ = 5/2
= 3, i = 1, > = 3/2 and n = 3, 1 = 2, = 3/2
^n = 3, i = 0,/=l/2and n = l, 1= l,j= 1/2
j'
n = 2,
= l,;=3/2
n = 2,
= 0,/ = l/2and n = 2,
(=1, J
= 1/2
-10
-13.6ev^
-15
n=U = 0,/ =
Figure
Dirac.
l/2
1-13. Energy levels of the hydrogen atom according to Bohr, Sommerfeld, and
The displacements from the Bohr energy levels are exaggerated by a factor of (1 37) 2
1
be understood quantitatively in terms of the theory of quantum electrodynamics, as can the slight departure of g s from 2 mentioned in the footnote
on page 337, but we shall not be able to discuss this theory here.
We close our treatment of the one-electron atom by mentioning one
which influences the atomic energy levels. This is the hyperfine
which is due to an interaction between the magnetic field produced by the moving electron and an inherent magnetic moment of the
final effect
splitting,
Lamb's experiment involved measuring the frequency of quanta absorbed in
between the n 2, I = 0,j = i and n = 2, /=l,y = J levels. The energy
difference between these levels is so small that the frequency is in the microwave radio
range. Since measurements of radio frequencies can be made very accurately, it was
possible to measure the energy difference to five significant figures!
t
transitions
MAGNETIC MOMENTS,
358
SPIN,
RELATIVISTIC EFFECTS
[Ch.
As nuclear magnetic moments are smaller than atomic magnetic
moments by three orders of magnitude, the hyperfine splitting is smaller
than the spin-orbit splitting by the same factor. Nevertheless, we shall
nucleus.
see in Chapter 13 that this effect can be described quantitatively in terms
of Schroedinger quantum mechanics. In fact, every aspect of the behavior
of a one-electron atom can be explained in detail by the theories of modern
physics. As we have seen, these explanations can be quite complicated.
Fortunately the degree of complication does not increase in proportion
number of electrons in the atom.
to the
BIBLIOGRAPHY
Leighton, R. B., Principles of
Modem
Physics,
McGraw-Hill Book Co.,
New
York, 1959.
Ruark, A. E., and H. C. Urey, Atoms, Molecules, and Quanta, McGraw-Hill
Book Co., New York, 1930.
White, H. E., An Introduction to Atomic Spectra, McGraw-Hill Book Co.,
New York, 1934.
EXERCISES
1.
Use Ampere's law
to evaluate the magnetic field produced
by a
circular
current loop at a point on the axis of symmetry far from the loop. Then evaluate
the magnetic field produced at the same point by a magnetic dipole located at
the center of the loop and lying along the axis of symmetry.
fields are the
magnetic
to
show
same
moment
that
if this
if
the current in the loop
and
its
Show
that the
area are related to the
of the dipole by equation (11-2). Next extend the argument
relation
is
obeyed the fields are the same at all points
the loop or dipole. Finally, extend the argument to
show
that this
is
far
from
true for a
plane current loop of arbitrary shape.
2. Evaluate ji^L for an electron moving in the n = 2, n = 1 elliptical orbit
e
of an atom of the type treated in the Sommerfeld theory. Compare with equation
(1 1-7). Hint: Use some of the equations derived in exercise 3, Chapter 4, and
exercise 4, Chapter 5.
3. The first footnote of section 3 states that a Stern-Gerlach experiment
cannot be performed with ionized atoms. It might be thought, though, that the
experiment can be done if the axis of the magnetic field is bent to follow the
curvature produced by the Hev/c force, and if a beam is used which has such
small transverse dimensions that the particles never get far from the axis, because
then all the particles of the beam will feel essentially the same Hev/c force.
However, the uncertainty principle shows that it is impossible to produce a
beam in which the transverse dimensions remain sufficiently small. Prove" this.
Ch.
EXERCISES
II]
359
4. The Thomas precession can also be described in terms of a time dilation,
between the reference frame in which the nucleus is at rest and the reference
frames in which the electron is instantaneously at rest, which leads to a disagreement between an observer at the nucleus and the observers at the electron
concerning the time required for each to make a complete revolution about the
other.
Work
equation
5.
out the details of this description, and compare the results with
(1 1-32).
Verify equation
to that equation.
(1 1-44),
and
also verify the vector inequality
which leads
CHAPTER
12
Identical Particles
I.
Quantum Mechanical
One
final topic
discussed before
of the theory of quantum mechanics remains to be
we can study
how
the structure of multi-electron atoms.
It
quantum mechanical
of a system containing two or more identical particles. The
concerns the question of
description
Description of Identical Particles
to give a proper
nature of the question can best be illustrated by a specific example.
Consider a box containing two electrons. These two identical particles
move around
scattering
in the box,
from each
bouncing from the walls and occasionally
In a classical description of this system, the
other.
electrons travel in sharply defined trajectories so that constant observation
of the system would allow us, in principle, to distinguish between the two
That is, if at some instant
and the other electron 2, we could constantly
observe the motion of the electrons without disturbing the system and
would be able to say which electron is 1 and which electron is 2 at any
later instant. In a quantum mechanical description this could not be done
because the uncertainty principle would not allow us to observe constantly
the motion of the electrons without completely changing the behavior of
electrons even though they are identical particles.
we label one of the electrons
An
is that in quantum mechanics the
wave function associated with each particle will lead
to an overlapping of these wave functions which would make it impossible
to tell which wave function was associated with which particle. A good
example is the two electrons in an He atom. When this system is in its
ground state the wave functions of the two electrons overlap completely.
Of course, there is also overlap of the wave functions associated with the
electron and proton of an H atom, but this does not lead to any problems
in distinguishing one particle from the other because the particles are not
the system.
finite
equivalent statement
extent of the
identical.
360
Sec.
DESCRIPTION OF IDENTICAL PARTICLES
I]
We
see that there
361
a fundamental difference between the classical and
is
quantum mechanical description of a system containing identical particles.
A correct quantum mechanical theory of such systems must be so formulated as to take explicit account of the indistinguishability of identical
particles.
When
done, the theory leads to the prediction of some
this is
very interesting
These
effects.
indistinguishability
have no
effects
analogy because
classical
a purely quantum mechanical situation.
is
way of
Let us look for a
wave functions describing the two
writing
box which
identical particles in the
To
will contain a
mathematical expression
we shall assume
two particles. Then the particles
will bounce between the walls of the box and their wave functions will
certainly overlap, but they will not scatter from each other. We shall show
of the ideas developed above.
that there are
no
simplify the argument,
interactions between the
later that the results of the following discussion are of quite general validity
The time independent Schroedinger equation
despite this simplification.
for the system
is
&_ / dVr
2m\dx\
yr
ip
T\
dz\)
dy\
tf_ l d
yT
2m\dx\
yr
~dyj
+ VT
tp
tp
T\
~dz\)
= ET y T
(12-1)
where
xi>
x2
We
have written
and m 1
=m =
2
with the labels
m=
=
Vi> z
z =
t/
2,
the
ine coordinates of particle
the coordinates of particle 2
this directly
and
either particle
m. Since
1
mass of
from equation (10-8)
for the case
N=
equation specifies the identity of the particles
in using it we stand a chance of violating the
this
2,
requirements of indistinguishability. This will be discussed more carefully
later.
In equation (12-1),
y T (*1;
VT (x,
Since
=
=
z
ET =
the eigenfunction for the whole system
2 2)
the potential energy for the whole system
2)
the total energy for the whole system
we have assumed
that there
is
no
the potential energies of each particle in
the box.
Each
its
is
simply the
two
sum of
interaction with the walls of
depend only on the coordinates of one
are identical, the two potential functions
potential energy will
particle and, since the particles
are the same.
interaction between the
energy of the whole system
particles, the potential
Thus
VT (xx
z 2)
V(xx yx zj
,
V(x 2 y 2 z^
,
(12-2)
IDENTICAL PARTICLES
362
We leave it
as
[Ch. 12
an exercise for the reader to show that for the potential
(12-2) solutions exist to equation (12-1) of the form
frixi, ... ,z 2 )
where
y)(x lt
y lt
and y{x2 y2
z-J
y2
(12-3)
2)
z 2 ) satisfy identical time independent
These equations have three spatial coordinates
,
Schroedinger equations.
v(*i, Vi, i) y(x2 ,
as independent variables.
Consequently, each of the eigenfunctions
quantum numbers to specify the mathematical form of its
dependence on the three spatial coordinates. In addition, one more
quantum number is required to specify the orientation of the spin angular
momentum vector of the particle.f However, we shall shorten the notation
to designate
by using a single quantum number from the series a, /?, y, d,
a particular set of the four quantum numbers required to specify the state
of a particle. Then a particular eigenfunction for particle 1 would be
requires three
written
We
by writing
further shorten the notation
this
V(l)
This eigenfunction contains the information that particle
described by a.
/?
An
eigenfunction indicating that particle 2
is
in the state
is
in the state
would be written
Then
1 is
the total eigenfunction
and
in the state a,
rp
We write
ip
T (xx
particle 2
T{xx
is
z 2 ) for the case in
in the state
z2 )
/?,
which
y>(l) V/,(2)
this
V = %(1) V&)
An
is
eigenfunction indicating that particle
in state
particle
is
(12-4)
is
in state
/?
and
particle 2
a would be
Vft
It is interesting to
Ytfl)
(12-5)
%(2)
evaluate the probability density functions for systems
described by the two eigenfunctions (12-4) and (12-5). These are
V> =
V?(l) W*( 2 ) V*(l) W(2)
(12-6)
tftt) V.*(2) V,(l) V(2)
(12-7)
and
v;V/,
Now,
since the
two
identical particles are indistinguishable,
we should
be able to exchange their labels without changing a physically measurable
t Particles other than electrons also
have
spin.
SYMMETRIC
Sec. 2]
AND ANTISYMMETRIC EIGENFUNCTIONS
quantity such as the probability density.
on
From
V*(l) Wp( 2 ) V(!)
V/>(2)
->
y(l)
ip*(2)
y> a (2)
^(1)
1-2
2-1
equations (12-6) and (12-7)
and so a
Let us carry out this operation
We find
<3W
vZvVrt
363
it
is
apparent that y>*pya p
y^w,.
^ W*aWfia
->
relabeling of the particles actually does change the probability
density function calculated
true for (12-5). Therefore
from the eigenfunction
we must conclude
The same
(12-4).
is
that these are not acceptable
eigenfunctions for the description of a system containing two identical
particles.
has been
2.
The suspicion which we expressed
after writing equation (12-1)
justified.
Symmetric and Antisymmetric Eigenfunctions
It is possible to construct an eigenfunction which satisfies the time
independent Schroedinger equation (12-1), and yet has the acceptable
property that its probability density function is not changed by a relabeling
of the particles.
two
In fact, there are two ways of doing
linear combinations of
y^
Consider the
this.
and y
fi(X ,
V>s
V/.J
(12-8)
Va
= -j= [V -
VVJ
(12-9)
-t=
[>*
and
done the exercise suggested below equation (12-2), it
be apparent that y^ and yfia are degenerate. If not, it should still be
apparent since the total energy of a system containing a particle in state a
If the reader has
will
and a
if
particle in state
fi
will
not depend on which particle
the particles are identical.
This
is
is
in
which
called exchange degeneracy.
state,
In the
we have shown that any linear combination of
degenerate eigenfunctions will be a solution to the time independent
Schroedinger equation. Therefore the two new eigenfunctions ys and y>A
footnote on page 189
are both solutions to equation (12-1).
The
eigenfunctions
s and
rp
rp
are,
of course, also degenerate. The probability densities for these two eigenfunctions are
YsWs
= KviVM +
vAVa.)
vlv>)
(12-10)
vIwa
wtv>pJ
- Kvw* +
vivn>
(i 2 - 11 )
Kyvf
and
i(y*P y> x p
IDENTICAL PARTICLES
364
From
y>p x
these
it is
[Ch. 12
show that y>s and %pA are normalized if y)ap and
Furthermore, since (12-4) and (12-5) show that
~* Vajs when we exchange the labels of the particles, we
easy to
are normalized.
Vxp - Vpa an<i V>pa
see by inspection that both
tion.
This
fsVs an d WaVa
Thus the eigenfunctions
ability flux.
are
unchanged by
this
opera-
also true for other measurable quantities such as the prob-
is
ip
s and
xp
as well as the
wave
functions associated with these eigenfunctions, provide an acceptable
description of a system containing two identical particles.
Only region
in
1/^(1) or i/^(2)
is
Figure 12-1.
which
non-zero
Only region
in
\p (1) or \p (2) is
Illustrating the classical limit for a
Although the
which
non-zero
system of two indistinguishable
particles.
numbers 1 and 2 do appear when we write in full the expressions for ys
and tpA this labeling does not violate the requirements of indistinguishability because the value of any physically measurable quantity, which can
be obtained from the eigenfunctions, is independent of the assignment of
,
As far as the eigenfunctions themselves are concerned, the
operation of exchanging the labels of the particles leaves one unchanged
the labels.
and
multiplies the other
by 1. By inspection we
see that
(12-12)
\Z\
Va-^-Wa
The eigenfunction
s is called a symmetric eigenfunction, and the eigenan antisymmetric eigenfunction. The behavior of y>A
not objectionable since an eigenfunction is not physically measurable.
An instructive interpretation of the role played by the first and second
function
is
y>
is
y>
called
parentheses in the expressions (12-10) and (12-11) for
fsVs
obtained by considering these expressions in the classical
aiK*
limit.
WaWa
IS
In this
one particle can be sufficiently localized that
do not overlap. That is, there
will exist two non-overlapping regions of space A and B, as indicated in
figure (12-1), such that y>J,l) or y>J2) is non-zero only when the coordinates
of particle 1 or particle 2 are in A, and such that y/1) or ^(2) is non-zero
only when the coordinates of particle 1 or particle 2 is in B. Then the
limit, the eigenfunctions for
two
different one-particle eigenfunctions
THE EXCLUSION PRINCIPLE
Sec. 3]
two-particle eigenfunction yi^
ordinates of particle
But,
if this is so, y>p a
are in
if>p(l)
ya(l)
365
y>p(2)
will vanish unless the co-
and the coordinates of particle 2 are in B.
ya(2) must vanish. Thus, if there is no region
which both of the one-particle eigenfunctions have non-zero values,
either one or the other of the two particle eigenfunctions ya/3 or y>^ will
always be zero. Then the second parentheses in the expressions for
ipsfs and WaVa w iH always vanish, and we have
in
Wsfs
We
VaWa
Kv*/>V*/j
see that in the classical limit the
(12-13)
V*Vp)
quantum mechanical
expressions for
the probability densities of the symmetric and antisymmetric eigenfunctions
are equal.
for
y>*ip
moment's consideration should demonstrate to the reader
common
that their
s and
value (the
y>*ipA ) is
first
parentheses in the general expressions
completely analogous to the classical expression
for the probability density of a system containing two identical particles,
and the other in the state /?, but it is not known which
which state. The quantity which is responsible for the
difference between f*y) s and f*yA when the one-particle eigenfunctions
if
one
is
in the state a
particle is in
overlap (the second parentheses in their general expression) contains terms
which correspond to some sort of interference between the two-particle
eigenfunctions xp^ and y>^. The presence of this quantity is a result of the
quantum mechanical requirements of indistinguishability.
3.
The Exclusion
We
T^,
now prove that,
N t) describing
shall
.
Principle
if at
some
certain instant the
a system of
wave function
identical particles
is
either
symmetric or antisymmetric with respect to exchanging the labels of any
two of the particles, then under any circumstances the wave function must
always preserve the same symmetry. To do this we consider the Schroedinger equation as a differential equation governing the time dependence
of *F(x lt
, z
N i), which can be written
.
eop^K, ...,zN ,t)
ih
d^ (*i.--
**.*)
at
This equation allows us to evaluate the wave function at time
we know it at time /, because
^i, .-.,%, +
t
dt)
= Y(x
t,
...,zN ,t)
+ gyfa '--'^
dt, if
t)
dt
dt
= TK, ...,zN ,t)+ ^T^, .... zN
in
t)
dt
IDENTICAL PARTICLES
366
Now the
[Ch. 12
derivative terms of the total energy operator
are obviously symmetric with respect to exchange of the labels of any
particles since such
This
on
is
will leave the
also true of the potential energy term since
the labeling of the identical particles, even
between the
(e
an exchange
j>lih)
^fa,
,z N ,t)
^(a^,
if
be symmetric and,
will
+
+
if
if
there are interactions
Since eop is a symmetric operator, the term
,z N t).
has the same symmetry as Yfo,
particles.
Consequently,
two
summation unchanged.
its value cannot depend
z N , t)
W (xv
is
,
symmetric,
z N , i) is
Y^,
antisymmetric,
zN,
Y^,
t
.
+
.
di)
,z N
be antisymmetric. By continuing this integration process to
2 dt, etc., we prove that the symmetry character of the wave function
t
must be preserved. If we consider the particular case of an eigenstate
t
di) will
wave function
W(xx ...,zN ,t)
,
e-
imlh
vte, ..-,%)
symmetry character of the eigenfunctions is the same
character
symmetry
of the wave functions,
as the
For a system of two non-interacting identical particles, we know that
the eigenfunction must be either symmetric or antisymmetric. We have
it is
just
clear that the
shown
that, in the
more general
case of a system of several identical
which are interacting with each other, the eigenfunction may
be either symmetric or antisymmetric. That is, we have shown that it is
consistent for the eigenfunction always to be symmetric or always to be
particles
The question then
antisymmetric.
real
arises:
Do
eigenfunctions describing
systems of several interacting identical particles have a definite
symmetry, either symmetric or antisymmetric, and, if so, which one?
The answer to this question is known. However, it is provided not by the
theory of quantum mechanics, but by experiment.
As a result of an analysis of experimental information concerning the
energy levels of certain atoms, in 1925 Pauli enunciated the exclusion
principle:
In a multi-electron
the
same quantum
atom
there can never be
more than one
electron in
state.
was soon established from the analysis of other experimental data that
and not, specithe exclusion principle operates in any system containing
fically, of atoms
It
the exclusion principle represents a property of electrons
;
electrons.
Now
consider the two-particle antisymmetric eigenfunction
THE EXCLUSION PRINCIPLE
Sec. 3]
367
(12-9) for the case in which both particles
and 2 are
in the
same
state a.
It is
Wa
The eigenfunction vanishes
~(^
v>
identically;
if
J=
two
(12-14)
Thus the property of electrons expressed by the exclusion
principle
exactly the property of the antisymmetric eigenfunctions,
and we
state.
is
by
same quantum
particles are described
the antisymmetric eigenfunction, they cannot both be in the
conclude that:
system containing several electrons must be described by an antisym-
metric eigenfunction.
We know
taining
the form of the antisymmetric eigenfunction for a system contwo non-interacting particles. Note that it can be written as a
determinant
= J_/%(1)
Wa
VJ>
(12-15)
V2"Wa) w(2>
since,
upon evaluating the determinant,
V>a
= j= IvJX) fi>(2) -
?/,(!)
this is
%(2)]
-i=
[VW
y,J
For a system containing
non-interacting particles the antisymmetric
eigenfunction can be written, in a completely analogous form, as
%(2)
^.(1)
Wa
(12-16)
\/W
W(l)
This
is
called a Slater determinant.
consists of
sums and
V>XN)j
Vv (2)
differences of
Since the value of this determinant
Nl terms comprised of products of
one-particle eigenfunctions, such as if>x (l) y> (2)
y>v (N) and similar
terms in which the particle labels are permuted, rpA is simply a linear
combination of degenerate eigenfunctions. Therefore %pA will also be a
fi
solution to the time independent Schroedinger equation.
IjVNl
is
such that
y>
particle eigenfunctions is
The constant
be normalized if each of the products of onenormalized. Because a determinant vanishes if
will
IDENTICAL PARTICLES
368
any two rows are
of vanishing
It is
such a
if
identical,
particle
possible to formulate
way
apparent that
it is
more than one
A has
same
y>
in the
is
[Ch. 12
the required property
state (e.g., if
most quantum mechanical
/?
a).
calculations in
that the interactions between the identical particles are con-
sidered perturbations.
Then, for unperturbed eigenfunctions, functions
However, if the interactions between
of the form (12-16) can be used.
the particles cannot be ignored, the antisymmetric eigenfunctions cannot
be written as (12-16), since y(l) y^(2)
y> u (N), and all the other terms,
will not be solutions to the time independent Schroedinger equation.
In this case antisymmetric eigenfunctions which vanish identically
same
if
two
be obtained by taking sums and
differences of the N\ degenerate solutions to the time independent Schroeidentical
dinger equation which will always exist for a system with
particles are in the
state
can
still
particles.
In this and the following chapters
we
shall study the multi-electron
systems whose analysis led Pauli and others to the conclusion that such
systems must be described by antisymmetric eigenfunctions. We shall
see that this property of electrons plays an extremely important role in
Atoms,
and therefore the entire universe, would be radically different if electrons
were such as to require a description in terms of symmetric instead of
governing the behavior of systems containing these particles.
antisymmetric eigenfunctions.
The symmetry character of particles other than electrons
is
also a
question which can be settled only by experiment. It is found that systems
atoms), or neutrons (uncharged particles of
of protons (the nuclei of
approximately proton mass), or certain other particles, must also be
described by antisymmetric eigenfunctions. Thus an eigenfunction describing a system of non-interacting protons must be of the form (12-16).
On
it is found that systems of quanta, or certain other
must be described in terms of symmetric eigenfunctions. Also,
systems containing particles which are tightly bound aggregates of an
even number of antisymmetric particles (e.g., a system containing alpha
particles which consist of two neutrons and two protons) must be described
by eigenfunctions which are symmetric with respect to exchange of the
labels of any two of the aggregate particles, because such an exchange is
equivalent to an even number of exchanges of the labels of antisymmetric
particles, and the net effect is to multiply the eigenfunction for the system
by an even power of 1 A symmetric eigenfunction for a system of N
non-interacting particles can be obtained by taking the sum of the Nl
the other hand,
particles,
y,(N) and the
terms that can be formed from the product y> a (l) %Pp(2)
similar terms obtained by permuting the particle labels, and then dividing
by the normalization constant
VWl
If the particles are interacting,
PROPERTIES OF ANTISYMMETRIC EIGENFUNCTIONS
Sec. 4]
symmetric eigenfunction can
still
be constructed by taking the
369
sum of
the TV! degenerate solutions to the time independent Schroedinger equation.
It is interesting to
note that there must be some connection between the
symmetry character of a
particle
and
its intrinsic
spin angular
momentum,
since all the antisymmetric particles also have half-integral spin just as the
However,
electron has, while
all
the reason for this
not completely understood at present. Consequently,
is
the others have zero or integral spin.
the symmetry character of a particle
like
4.
must be considered a basic property,
mass, charge, and spin, which can be determined only by experiment.
Additional Properties of Antisymmetric Eigenfunctions
As mentioned
in section 3,
Chapter
11,
it is
often convenient to write
the complete eigenfunction containing both the spatial
and spin coordin-
ates of a particle as a
function.
This
product of a spatial eigenfunction and a spin eigenpossible providing there are no interactions explicitly
is
involving the spin
(e.g.,
the spin-orbit interaction), so that
remains constant and m s is a good quantum number.
example, the one-electron-atom eigenfunctions f nlmim
its z
component
As a particular
may always be
written
Ynlmim s
since, if the
Wm mi m
are valid,
s
T nlmi
the general case in which this restriction
is
eigenfunction for a particle carrying the label
V(l)
must be a good quantum number. In
V0)
satisfied,
1
we may
write the
as
(12-17)
*..(!)
where the single quantum number a stands for a set of three spatial
quantum numbers. Now consider a system containing two particles in
which there are no explicit interactions, spin or otherwise, between the
particles, Then, according to (12-4) and (12-17), we may write the eigenfunctions describing these two particles as the product
V>i,
y.(l) y*(2)
%(1)
v(2)
a ms
(\)
a ms p.)
Using a notation analogous to that defined by (12-4),
Wa0
y>a b <y msa m, b
this
can be written
(12-18)
not an acceptable eigenfunction for a system of two electrons
an antisymmetric character) because it is
not antisymmetric with respect to an exchange of the numbers 1 and 2
This
is
(or other identical particles of
which label the
electrons. However, it is possible to construct a properly
antisymmetrized eigenfunction by taking a symmetric or antisymmetric
IDENTICAL PARTICLES
370
linear
[Ch. 12
combination of spatial eigenfunctions times a linear combination
of spin eigenfunctions of the opposite symmetry, so that the product which
forms the complete eigenfunction is antisymmetric. Consider the spin
eigenfunctions. Since for an electron m s can assume just two values
and
From
+ |,
these
there are four possibilities
we can
construct four different spin eigenfunctions,
-j=( a +K-K
a -A+vd
singlet
a -H+\d
triplet
a +X+H
(12-19)
j=(o+ii-H
All four are normalized.
The
one, which
first
is
called the singlet spin
obviously antisymmetric with respect to an exchange of
the particle labels. The last three form the so-called triplet of spin eigenfunctions. They are all obviously symmetric with respect to an exchange
eigenfunction,
of the labels.
we
is
To
obtain complete eigenfunctions which are antisymmetric,
multiply the antisymmetric singlet spin eigenfunction by a symmetric
spatial eigenfunction,
4(Va,+
and multiply each of the symmetric
(12-20)
V>
triplet spin eigenfunctions
by an
antisymmetric spatial eigenfunction,
j=(V,*-YO
(12-21)
So there are the following four different antisymmetric complete eigenfunctions for a system containing two non-interacting electrons
1
"/|(Va&
Va
Vba)
-]=(? A- A
l
ff-H + M)
=V
1
-^(Vai.
(12-22)
-friWab-
ViJ-7=( tf +H-W
V>ba)<*-X-<4
a -H + %)
PROPERTIES OF ANTISYMMETRIC EIGEMFUNCTIONS
Sec. 4]
The
371
and the others describe the three
and degenerate.
A physical interpretation of the singlet and triplet states can be obtained
by evaluating, for each state, the magnitude and z component of the total
spin angular momentum vector S', which is
first
describes the singlet state,
All four are normalized, orthogonal,
triplet states.
S'
sum
the
do
Sj
S2
(12-23)
momenta of the two electrons. We cannot
because we have not studied the properties of the
we shall merely state the results and give some
of the spin angular
this in detail here
spin eigenfunctions ;
s'=l
S
= ^||*2=
s'
Figure 12-2.
arguments for
s,
=H
o
Vector addition diagrams for the quantum numbers
their validity.
The reader
st
i, s 2
should, however, have no
|.
diffi-
culty in accepting the results, as they are completely analogous to those
governing the addition of the spin and orbital angular momentum vectors
to form the total angular momentum vector. It is found that the magnitude
and z component of the
by the equations
total spin
S'
=
=
m,
s'
S'
angular
vs'(s'
momentum
vector are given
l)h
(12-24)
m'h
where
s'
and where
0,
(12-25)
and triplet states, respectively. Equation (12-25) can be
proved by the same kind of arguments as were used in proving equation
(11-44). These rules can formally be represented by diagrams in which
for the singlet
two vectors of length
momentum
This
is
illustrated in
\, which must not be interpreted as angular
added to form a vector of length s =
or 1.
figure (12-2). For the singlet state, m' = 0. The
s
vectors, are
372
IDENTICAL PARTICLES
[Ch. 12
quantum number for the three triplet states is shown in
The properties of the states in which the spin eigenfunction
is <y
+ yt+ A or O-14-14 are easy to interpret. In these states, the z components
of the spin vectors of the two electrons are of the same sign. Thus the
spins combine into a total spin with z component for m' = 1 which is
s
twice that of a single electron. The quantum number specifying the
magnitude of the total spin vector for these two states would then
necessarily be s' = 1. For the states in which the spin eigenfunction is
value of this
table (12-1).
i
(1/V2)(e7
+H_ H
a_ H+ii ) or (l/V2)(o- _ K
+K
TABLE
Quantum Numbers
Spin eigenfunction
a_,
A+}4 ),
it
is
clear that
(12-1)
for the Singlet and Triplet States
s'
m's
triplet
+1
triplet
triplet
Designation
singlet
c +14+14
1
-'A-'A
we must have m' s
number
0.
Although
it is
not so apparent
-1
why
specifying the magnitude of the total spin vector
the antisymmetric state
and
s'
the
quantum
is s'
for
for the symmetric state, these results
The triplet and
which the spin vectors
of the two electrons are parallel and antiparallel, respectively. Although
it is not literally true, this description certainly contains an essence of
truth and provides a useful physical picture.
If the spin vectors of the two electrons are "parallel," the spin eigenfunction is symmetric and, to make the complete eigenfunction antisymmetric, the spatial eigenfunction must be antisymmetric. Let us consider
such a situation for a case in which the spatial coordinates of the two
x2 y1
z 2 Then y> a {\)
electrons are very similar, i.e., x l
y2 z x
fa {2)
and y 6(l)
Consequently
y(2).
follow directly from the theory of spin eigenfunctions.t
singlet states are often
spoken of as the
states in
Wat
These
Vo(l) Vk(2)
results are certainly also reasonable
from an
intuitive point of
view because
mean that all three symmetric spin eigenfunctions correspond to the same value of
quantum number s' = 1, i.e., to the same physical situation in which the electron
they
the
^ Vo(2) V&0) = Vi.0) Va(2) = fta
spin vectors are "parallel."
THE HELIUM ATOM
Sec. 5]
373
In this case the value of the antisymmetric spatial eigenfunction
~j= (fab
As a
result,
Vba)
^ ~j= {fab ~
fab)
the probability density will be very small
electrons are close together.
Since there
is
is
when
the two
only a very small chance of
finding the two electrons close together, the two particles act as
repel each other.
because
This has nothing to do with a
we assumed
Coulomb
if
they
repulsion
at the very beginning of this section that there
is
no
between the two electrons. Furthermore, it is easy to
the spatial eigenfunction describing the two electrons is
explicit interaction
show
that, if
symmetric, the probability density
they are close together.
Thus,
if
twice as large as average
is
when
the spin vectors of the two electrons
are "antiparallel," the spin eigenfunction
is
antisymmetric, the spatial
eigenfunction must be symmetric, and the two particles act as
if
they
a large chance of finding them close
together. We see that the requirement that the complete eigenfunction
be antisymmetric with respect to an exchange of the labels of the two
attract each other because there
is
and spatial coordinates.
They act as if they move under the influence of a force whose sign depends
on the relative orientation of their spins. This is called an exchange force.
It is a purely quantum mechanical effect and has no classical analog.
However, it does lead to directly observable results, such as those described
electrons leads to a coupling between their spin
in the next section.
The Helium Atom
5.
The
He
simplest real system containing
atom.
show
two
identical particles
is
the neutral
On the right side of the energy level diagram in figure (12-3), we
the energies of the ground state
and the
first
four excited states of
He. The energy of the ground state is obtained experimentally by measuring the energy required to ionize He to form He+. The sum of this ionization energy and the ground state energy of the one-electron atom He+,
which is 54.1 ev, is just equal to the energy of the ground state of He.f
The energies of the excited states relative to the ground state are obtained
from an
To
analysis of the
treat
He
He
spectrum.
with the theory of quantum mechanics,
we assume
first
t The magnitude of the energy of the ground state is thus denned as the energy required
to ionize completely the He atom to form He 2+ In spectroscopy it is, however, conventional to define the energy of the ground state as the energy required to ionize singly
.
the atom; a spectroscopist
would put
E=
at the value
we
label
E=
54.1
ev.
IDENTICAL PARTICLES
374
[Ch. 12
that the two electrons are
Coulomb
potential,
moving under the influence of the same nuclear
but that they are non-interacting. The interaction
between the two electrons is then taken into account by means of perturbation theory. Thus, we consider an unperturbed system whose energy
u
-10
-20
-30
-40
-50
.N
-60
-70
-80
-90
-100
-110
Figure 12-3.
Right:
An energy
level
diagram showing the ground state and the first
Left: The energy levels predicted by ignoring
four excited states of the helium atom.
the electron-electron interaction.
states are given
we
by the sum of two expressions of the form (10-38). That is,
take
E = - Z 2e4/2^ 2f - aZ e4/2A 2l
= 1, 2, 3,
Z = 2. In
2
2
/M
(12-26)
where n x
1, 2, 3,
rt
the
ground
state
375
THE HELIUM ATOM
Sec. 5]
quantum numbers n x and n 2 are both equal to 1. In
and the
one of the quantum numbers is equal to
the
approxithis
in
states,
these
for
predicted
energies
2.
The
other is equal to
mation, are shown on the left side of the energy level diagram of figure
of this system, the
first excited state,
poor descripis apparent that the unperturbed system provides a
that
the intermeans
This
system.
actual
the
of
levels
energy
of
the
tion
behavior
the
in
role
important
an
play
electrons
the
two
between
actions
(12-3). It
of the
He
atom.
The interactions consist primarily of the Coulomb repulsion which the
two negatively charged electrons exert on each other. It produces a
positive potential energy -which raises the total energy of the atom. In
addition, there are magnetic interactions involving the spins of the elec-
trons; but they are very small
and can be
neglected.
to the interaction of their charges
compared
Thus we take
for the perturbing potential the
quantity
v
(- eX~ e >
i!
(12-27)
r 12
r 12
between the two electrons. For unperturbed
use the antisymmetric eigenfunctions (12-22)
and, for each energy state of the unperturbed system, there will, in general,
be a number of degenerate unperturbed eigenfunctions. The degeneracy
of the energy
is due to exchange degeneracy, and also to the degeneracy
where
r 12 is the distance
we must
eigenfunctions
states of the individual electrons
and w r
For the ground
with respect to the quantum numbers
unperturbed system, there is, however,
only a single eigenfunction. The lm degeneracy is absent because when
n x = 2 = 1 the other spatial quantum numbers can only have a single
Furthermore, since all the spatial
value /x = m = /2 = m, =0.
quantum numbers for both electrons are the same, we see from equation
state of the
(12-22) that exchange degeneracy
is
absent as there
is
only one possible
antisymmetric eigenfunction
The ground
state of the
unperturbed system can only be the singlet
state
because the spatial part of the antisymmetric eigenfunctions, the use of
which is required by the exclusion principle, vanishes identically for the
triplet states
are the same.
the
when all the spatial quantum numbers for the two electrons
Of course, the fact that all the spatial quantum numbers are
same does not contradict the exclusion
quantum numbers
are different.
We
principle because the spin
see that the properties of spin enter
376
IDENTICAL PARTICLES
[Ch. 12
even in the properties of the spatial part of the unperturbed
However, when we use the eigenfunction to calculate the
implicitly,
eigenfunction.
from the perturbation, it is not necessary to consider
Because the perturbation does not involve spin, the spin
eigenfunctions enter only as a product in the term tp*vy> which must be
energy
shift resulting
spin explicitly.
evaluated in the calculation. Since the spin eigenfunctions are normalized,
this
product
is
function,
Consequently in the present calculation
equal to unity.
we may simply ignore
the spin part of the complete antisymmetric eigen-
and take for the properly normalized unperturbed eigenfunction
Waa
= U^e-^-e-*""*
Vioo(l) Vioo(2)
where we have evaluated
ip
lm from (10-39). Then, according to the three
dimensional extension of equation (9-13), the energy shift due to the
perturbation is
E'
- E~
J J
v4>(!)
Vi*oo( 2 )
ViooC 1 ) Vioo( 2 )
dri dr2
r 12
where
drx
dr2
=
=
r\ sin 6 1
rjj
sin
This integral has been evaluated.
dr x dd 1
dr 2 dd 2 dj> 2
It gives
E '-E~
5
-Z>^
2
Combining
this result
(-2Z 2 +
\
The value of E'
2/j
with (12-26) for n x
E' =
calculated
from
dj>
this
n2
1,
we have
Z)^
(12-28)
/2/7 2
equation for
Z=
2 agrees with the
energy of the ground state of He to within 5 percent. Perturbation theory
gives a reasonably accurate result, even though the perturbation is ob-
Much more accurate results can be obtained by using
a modification of perturbation theory in which the error is of the order of
the cube of a small quantity. Expression (12-28) also predicts the ground
viously not small.
state energy of a two-electron ion with Z > 2, such as Li+, Be 2+ B 3+
The agreement improves with increasing Z because the Coulomb interaction between the nucleus and each electron increases more rapidly than
the Coulomb interaction between the two electrons, so the perturbation is
smaller and the perturbation calculation more accurate.
,
Next consider the
first
excited state of the unperturbed system for He.
THE HELIUM ATOM
Sec. 5]
This
is
377
a highly degenerate state because there are four degenerate one2, and there are also exchange degeneracies.
electron states for n
However, much of the degeneracy is removed when the interaction between
the two electrons is applied as a perturbation. Before considering this
quantitatively, it is worth while to describe what happens qualitatively.
The increase in the energy of the atom produced by the Coulomb
repulsion between the two electrons depends on the average distance
between the two electrons. This, in turn, depends on the quantum numbers
of the spatial part of the eigenfunction, and also on its symmetry character.
Let us consider first the dependence on the spatial quantum numbers,
ignoring temporarily the symmetry properties and the requirements of
Then we may
indistinguishability.
numbers
for the
two electrons
list
in the first
the possible sets of
quantum
excited state of the unperturbed
system as follows:
!
nx
"i
"i
=
=
=
=
1,
1,
1,
1>
= 0, m h = 0;
= 0, m h = 0;
= 0, m h = 0;
= 0, m h = 0;
=
=
n2
n2
=
=
n2
n2
2,
12
2,
2,
2,
Since the probability density of electron
=
=
=
=
is
0,
1,
1,
1,
mh
mh
mh
mh
=
= -1
=
spherically symmetrical,
the average distance between the two electrons depends only
probability densities for the
two
particles.
on the
radial
These quantities, denned in
equation (10-43), are specified completely by the quantum numbers n
and
By
/.
inspecting figure (10-2), the reader will see that the average
distance between the two electrons
tum numbers
0;
1,
2, /2
1.
/jj
0; n 2
is
somewhat
2, /2
larger for the set of quan-
than
it is
for the set n 1
Since the perturbing potential
is
I,
inversely pro-
portional to the distance between the two electrons, the total energy of the
atom
2
will
= 2,
= 0.
/2
be increased more for the state specified by n 1 = 1, 4
1 than for the state specified by Hj = 1, lx = 0;
n2
= 0;
= 2,
l2
Thus the degenerate energy level for the first excited state of the
unperturbed system is split into two components by the perturbation. If
we now take into account the requirements of indistinguishability, we shall
see that each of these components is split again into two components.
This is a result of the fact that each of the states, such as the one specified
by the quantum numbers nr
1, lx
0; n 2
2, l2
0, is
comprised of
four exchange degenerate states corresponding to the four antisymmetric
complete eigenfunctions (12-22) which must be used to describe the system
of two electrons. The splitting occurs because the average distance between
the two electrons is smaller in the singlet state than it is in the triplet states,
so the increase in the energy of the atom is larger for the singlet state than
378
it is
IDENTICAL PARTICLES
The net
for the triplet states.
increase the energy of the
first
effect
of the perturbation
[Ch. 12
is
thus to
excited state of the unperturbed system
and split the energy level into four components. In the energy level
diagram of figure (12-3), the reader will see that this conclusion is qualitaagreement with the experimentally observed energy levels of He.
Let us use perturbation theory to evaluate quantitatively the effect
tively in
of the perturbation. This will not be so
difficult as the
reader might guess,
for the following reasons:
1
the
As in the ground state calculation, we may neglect the spin part of
antisymmetric complete eigenfunctions and use for unperturbed
eigenfunctions only the spatial part.
2. Since the total angular momentum of the atom (the sum of the orbital
and spin angular momenta of the two electrons) cannot change if there are
no external torques, and since the perturbation does not cause interactions
between the spin angular momenta and the orbital angular momenta, it
momentum nor the total
can change as a result of the application of the
Therefore the perturbation cannot induce appreciable
follows that neither the total spin angular
momentum
orbital angular
perturbation.
mixing between the degenerate states specified by n x = 1, /x = 0; n 2 = 2,
/2 = 0, for which the magnitude of the total orbital angular momentum
and the degenerate states specified by n x = 1, /x = 0; 2 = 2, /2 =
for which the magnitude of the total orbital angular momentum
is 0,
V 1(1 +
these
3.
1) h.
two
For
sets
Consequently
it is
1,
is
possible to treat the degenerate states for
of quantum numbers separately.
simplicity,
we
shall actually carry
for the degenerate states specified
by n 1
through the calculation only
\,
/x
0;
n2
2,
l2
0.
For these states there is no spatial quantum number degeneracy since
m and m can have only the single value m = m u = 0. There is,
however, an exchange degeneracy which arises because the spatial part
of the singlet and triplet eigenfunctions,
l
~j=
[Wai>
+ W] =
-j= [Vioo(l) %oo( 2 )
VaooC 1 ) Vioo(2)]
(12-29)
L>ioo(l) Wwoi 2 )
^ooC 1 )
Vioo( 2 )]
(12-29')
and
-j= [fab
Wba]
-ft
correspond to the same total energy for the unperturbed system.
4. As stated in 2, the total spin angular momentum of the atom cannot
change as a result of the application of the perturbation. For this reason
the perturbation cannot induce appreciable mixing between the degenerate
eigenfunctions (12-29) and (12-29') because the first corresponds to the
THE HELIUM ATOM
Sec. 5]
379
which the two spin vectors are "antiparallel" and the
magnitude of the total spin angular momentum is 0, while the second
corresponds to the triplet states in which the two spin vectors are "parallel"
singlet state in
and the magnitude of the
Therefore
it is
total spin angular
momentum
is
^1(1
1)
fr.
due to the perturbation
possible to evaluate the energy shift
by applying ordinary perturbation theory to each of the eigenfunctions,
even though they are degenerate, just as we did for the degenerate eigenfunctions
y nl j m when
calculating the spin-orbit energy shift.
Thus we use
the three dimensional extension of (9-13) with the pertur-
bation (12-27) and the unperturbed eigenfunctions (12-29) and (12-29'),
and obtain for the energy
E'
- E~i
due to the perturbation the expression
shift
f f [ Vl*oo(l) %*oo(2)
2J
X
corresponds to the
~E+
V2*oo(l) Vi*oo(2)]
r 12
where E'+ corresponds to the
Ol(K)(l) ^20o( 2 )
VW
1)
VlOo( 2 )]
drl dr2
by (12-29) and E'_
described by (12-29'). This
triplet state
singlet state described
is
flooil) V20o(2)
r iWV^ %oo(2) ~
rv
2 J J L
1 00
'
'
""""
t.
'
'
12
V2*oo(l)
Vlooi 2 )
V200C 1 ) Vioo(2)
dr x di
'12
2
Vioo(l) V2oo(2)
J/J[-
^2oo(l) Vioo(2)
r 12
VJ 2oo( 1 ) Vioo( 2 )
^100(1) VaooC 2 )
^12
d 7l dT2
(12-30)
The reader should check
these results by performing a routine degenerate
perturbation theory calculation, as described in section 4, Chapter 9, for
the case of two degenerate eigenfunctions
This
is
y100 (l) V2oo(2) and V2oo(0 Vioo(2)easy to do because the calculation carried out in that section
can be immediately applied to the present problem. Equations (9-26)
and (9-27) show that (12-29) and (12-29') are exactly the right linear
combinations of y100 (l) V2oo( 2) and
1)
yioo(2), and that (12-30) is the
vW
correct expression for the perturbed energy.
Let us compare the results of our calculation with equations (12-10)
and (12-1 l),and with the interpretation which was given to these equations.
This comparison will identify the first integral in equation (12-30) as an
expression which, from classical considerations, would be written for the
expectation value of the perturbation in a system containing two identical
particles in the one-particle states n = 1, / =
and n = 2, / = 0. The
same comparison identifies the second integral as a manifestation of the
380
IDENTICAL PARTICLES
quantum mechanical requirements of
indistinguishability.
[Ch. 12
This integral
takes into account the effective attraction or repulsion in the singlet or
triplet states
it
by increasing the energy
for the latter.
The second
shift for the
integral
n,
I:
is
n,
former and decreasing
called the exchange integral,
and
Spin
singlet
1, 0; 2, 1
"
triplet
singlet
1, 0; 2,
triplet
1;2
//
1,
i;i
singlet
0; 1,
Unperturbed
Perturbed
Perturbed
(ordinary
(ordinary
plus
integral
exchange
only)
integrals)
Figure 12-4.
and the
first
Illustrating
the effects which contribute to the energies of the ground state
four excited states of the helium atom.
the
first
we
evaluate
integral
y100
is
called the ordinary integral.
and
y^ from (10-39),
To
evaluate these integrals
which gives
-Zr/a
Vioo
By
V2O0
J7r\aj
substituting these into (12-30),
4J277W
Zr/a
and carrying through the
integration,
values for E'_ and E'+ are found. These values agree with the experimentally
observed energies of the
first
two excited
states
of
He
to about
percent.
Sec. 6]
THE FERMI GAS
The perturbation
381
calculation
is
more
accurate for the excited states than
ground state because the average distance between the electrons
is larger and the perturbation is weaker.
A perturbation calculation for the unperturbed states, in which one set
of quantum numbers is n = 1, / =
and the other set is n = 2, I = 1,
leads to results which are similar to (12-30) except that the numerical
value of the ordinary integral turns out to be somewhat larger. The
important features and results of the theoretical description of the first
few energy levels of He are represented schematically by the set of energy
level diagrams shown in figure (12-4).
Not shown in the diagram is a fine structure splitting of the triplet
energy levels, the existence of which is inferred from the spectrum of He.
It is due to the very small interactions involving the spins of the electrons,
and other relativistic effects, that split each of the triplet levels (except
those in which the total orbital angular momentum is zero) into three
levels with a separation about 10~ 4 times the separation between associated
singlet and triplet levels. A theoretical treatment of this effect can be
given in terms of the equations developed in the previous chapter. There
is also a hyperfine splitting of the levels, due to extremely small interactions
involving the spin of the nucleus, which produces a separation of about
10~ 3 times the separation produced by the fine structure splitting, and
which can be treated by methods to be developed in Chapter 13. But we
for the
shall
Of
not treat either of these
particular interest
is
effects here.
the absence of a triplet energy level associated
with the singlet ground state
level,
which
in
our treatment we have shown
This is quite true, but
to be a consequence of the exclusion principle.
historically the
argument was made in the opposite order. The experi-
mental fact that the He spectrum shows this triplet to be absent provided
the primary evidence that led Pauli to the discovery of the exclusion
principle.
6.
The Fermi Gas
All particles which
are given the generic
must be described by antisymmetric eigenfunctions
in honor of a man prominent
name Fermi particles,
in developing the theory of their properties. In the last section
we presented
a detailed discussion of a system containing two Fermi particles.
we shall discuss, in somewhat
number of these particles.
less detail,
Here
a system containing a large
Consider a system in which a "gas" of freely moving, non-interacting,
Fermi particles is confined to a certain region. Such a system, which is
IDENTICAL PARTICLES
382
[Ch. 12
Fermi gas, can be used to describe some of the important features
of the behavior of the electrons confined to a multi-electron atom. A
Fermi gas can also be used to describe certain features of the motion of
the protons and neutrons confined to a nucleus. Possibly the best physical
example of a Fermi gas is the conduction electrons in a metal. These
called a
which are responsible for the electrical conductivity of the metal,
it quite freely because the forces exerted by the various
through
move
atoms tend to cancel, and the forces which the electrons exert on each
other are small. However, the conduction electrons are confined to the
metal by non-canceling attractive forces at the surfaces. In this section
electrons,
shall develop some of the properties of a Fermi gas, such as its total
energy and the energy distribution of the individual particles, and shall
illustrate their significance for the conduction electrons. We shall see that
we
by the exclusion principle.
moving independently in a potential
we define as zero) inside a certain region, and
these properties are very directly influenced
Each
particle of a
Fermi gas
is
which is a constant (that
which becomes larger than the total energy at the boundaries of the region.
In most cases it is an adequate approximation to assume that the potential
becomes infinitely large at the boundaries. Then the identical Schroedinger
equations describing the motion of each of the identical particles assume
a particularly simple form,
+ & + )_*,
_(
2m\dx
dy
2
in
which y(x,
y, z)
y,(x, y, z)
2-31>
at the boundaries of the region.
The reader may
equation can be written
easily verify that solutions to this
in
dz
X{x) Y(y) Z(z)
(12-32)
which
_tf_#J^ =
2m
and
similarly for Y(y)
and
Z(),
(12_ 32
ExX{x)
dxr
and where
E=E
+ Ey + E
not restrict the generality of our calculation to assume that the
the Fermi gas is a cube of side a. Then, defining the
containing
region
coordinates
at the center of the cube and the axes parallel to
of
the
origin
It will
its
edges, the condition
X{x)
and
similarly for Y(y)
on X(x)
0,
is
<
-a/2,
>
a/2
(12-33)
and Z(z). Equation (12-32') and condition (12-33)
and (8-90), which describe the problem of a one
are identical with (8-89)
.:
THE FERMI GAS
Sec. 6]
dimensional
values of
Ex
383
2fc2
2,
1,
3, 4,
The
1, 2, 3,
three spatial
4,
Ey and E
energy of one of the identical particles
total
+; + !)
^("2
2ma
=
where n x
(12-34)
Similar expressions obtain for
Thus the allowed values of
is given by the equation
= ~^i
Ex
where nx
Consequently the allowed
square well potential.
infinite
can be written directly from (8-96), and we have
(12-35)
ny
1,
2, 3, 4,
quantum numbers n x n y n z
,
nz
specify the
1,
2, 3, 4,
quantum
states
of the particle.
It is
useful to calculate
number of quantum
within the limits
section 6, Chapter
in
from
this
equation the quantity N(E) dE, the
which the
states for
energy of the particle
total
is
E + dE. The procedure is similar to that used in
2.
We imagine a uniform cubic lattice constructed
to
one octant of a rectangular coordinate system in such a way that the
three coordinates of each point of the lattice are equal to a possible set of
quantum numbers n x ny n z Then the number of possible sets of quantum
numbers is equal to N(r) dr, the number of points contained between
concentric shells of radii r and r + dr, where
,
^r
E =
By
construction, N(r) dr
is
just equal to the
N(r) dr
- 477^ dr
(12-36)
volume enclosed by the
shells
According to equation (12-36),
dE
==
dr
ma
and
E'
ma
so
'
2
Since N(r) dr
irH
7T
N(E) dE, we have
JV()
dE
-g-il E* dE
n
(12-37)
IDENTICAL PARTICLES
384
where we write
particle of mass
V=
equation (12-37)
states in the
cP for the
volume of the cubical region
m and kinetic energy E is confined.!
is
the correct expression for the
E to E + dE for
energy range
It
in
[Ch. 12
which the
can be shown that
number of quantum
the general case of a particle
V and any arbitrary shape.
The eigenfunction (12-32) describing the motion of one of the identical
particles is a standing wave with nodes at the boundaries of the region.
The eigenfunction describing the entire system of particles is composed of
products of such standing wave eigenfunctions and, to satisfy the exclusion
principle, it must be written in the antisymmetrized form (12-16). If we
confined to a region of volume
discuss only the energetics of a
Fermi
gas,
it
will
not be necessary to deal
&
Without exclusion
With exclusion
principle
principle
Figure 12-5.
Illustrating
fat
the effect of the exclusion principle.
with this eigenfunction. However,
it
will
be very necessary to take account
This can be done simply by imposing the
of the exclusion principle.
no more than two particles have the same spatial
quantum numbers n x nv n z The exclusion principle will be satisfied
because one particle can have the spin quantum number m s = \ and
the other can have m s = +|.
auxiliary requirement that
,
It is
easy to see that the exclusion principle plays an important role in
determining the
minimum
total
energy of the Fermi gas. Consider the
two diagrams of figure (12-5).
situation illustrated schematically in the
The diagram on
the left represents the occupied
quantum
states
when a
system of eight identical particles, which do not obey the exclusion
principle, has the
minimum
possible total energy.
All the particles
would
be in the state of lowest energy. The other diagram represents the occupied
quantum states for the minimum possible total energy of a system in which
do obey the exclusion principle. Only two particles can be in
same spatial state, and so the total energy of the system is considerably
larger. The diagrams also illustrate the way in which the number of
For a system
quantum states per unit energy increases according to E
of N Fermi particles, the lowest total energy is achieved when there are
and E = Et
two particles in each of the energy states between E =
the particles
the
l/i
t Since the potential energy
kinetic energy.
is
zero in the region, the total energy
E is
equal to the
THE FERMI GAS
Sec. 6]
The
particles
fill
385
up
the states
to the
Fermi energy
to the
number of particles
That
in the system.f
Ef
This energy
dE from
evaluated by requiring that the integral of 2N(E)
to
may be
Et be equal
is,
N=
2N(E) dE
Jo
The energy
Figure 12-6.
The
distribution for a system of Fermi particles at zero temperature.
factor of 2 enters because there are
state.
Evaluating N(E)
dE from
two
(12-37),
particles per spatial
2 H m 3A vC E<i
T
quantum
we have
,
Jo
Thus
Ef =
2m \F/
or
E f = 3V*i
where p
is
the
number of
particles with energies
gas
is
just
(12-38)
particles per unit
volume.
The number of
E and E + dE in a minimum energy Fermi
E < Er As shown in figure (12-6), there are no
between
2N(E) dE for
particles with
2m
_2_ JA
E > Ef We leave it
.
as an exercise for the reader to prove
that the average energy of the particles in this energy distribution
= \Et
Thus the
total
energy of the Fermi gas
is
(12-39)
is
E = NE = %NEt
t
(12-40)
t The use of an integral is justified if we are treating a system containing a large
number of particles. Then the integral is a good approximation to the sum over the
discrete energy states which, strictly speaking, should be used.
IDENTICAL PARTICLES
386
[Ch. 12
should be noted that, for Fermi particles of a given mass, Ef and E
depend only on the density of the Fermi gas.
Let us evaluate some of these quantities for the Fermi gas of conduction
electrons in a metal. Consider Cu, which has one conduction electron
per atom. In this case p is equal to the density of atoms, which is Avogadro's
number times the density of Cu divided by its atomic weight. Thus
It
Figure 12-7.
states
The
10 23
x 9/64
10 22
cm"3
potential experienced by the conduction electrons in copper, and the
which these electrons
Using
this value
of
p,
fill.
and
setting
equal to the mass of the electron,
equation (12-38) gives
>
This
is
10
x 10" 12
ergs
~ 7 ev
the energy of the highest filled state relative to the
potential.
the metal
measured
The energy of
this state relative to the potential
just the quantity
is
W, appearing in equation
For Cu it is
bottom of the
energy outside
(3-8), that
is
in the photoelectric effect.
PK~4ev
Knowing Ef and W, we may draw
a diagram representing the potential
energy in which the conduction electrons move, and also the energy states
which they fill, as illustrated in figure (12-7). For a 1 cm3 sample of Cu
there are
~4
10 22
filled
energy states!
This fact
justifies the
approxi-
mation discussed in the footnote on page 385. The approximation of
assuming that the potential is infinitely large at the boundaries is justified
by the fact that the highest filled state is far below the top of the potential.
The conduction electrons of Cu will have a minimum total energy when
the total energy content of the metal
temperature
is
at absolute zero.
of the conduction electrons
is
itself is
But even
at
minimum,
T=
E = 3Ef/5 ~ 4 ev.
This
i.e.,
when
the
the average energy
is
a direct result of
Sec. 6]
THE FERMI GAS
the exclusion principle.
obey
387
For a system composed of particles which do not
gas of distinguishable particles),
this principle (e.g., a classical
the particles will be in the ground state
when
For such
their energy will be essentially zero.
the system
is
particles to
at
T=
0,
all
and
have an average
energy of 4 ev, the energy must be supplied by thermal agitation and the
required temperature satisfies the equation
%kT
= f Ef ~
4 ev
where k is Boltzmann's constant. This equation gives the high temperature
T~
10 4
which can be taken as a measure of the temperature equivalent of the
average energy of conduction electrons.
The fact that Fermi energy Ef is very much larger than thermal agitation
energy at normal temperatures,
(ffcZW*
leads to
some
~ 0.025 ev
For a Fermi gas
interesting consequences.
the states below the Fermi energy are completely
of the metal containing the Fermi gas
is
filled.
raised, there
the electrons to be excited to higher energy states
thermally agitated atoms.
by
at
T=
is
a tendency for
collisions with the
But, according to the exclusion principle,
make a
all
As the temperature
it is
which is not
filled.
Since at normal temperatures the thermal energy which can be
supplied to an electron is quite small compared to the Fermi energy, only
the electrons whose energies are very near Ef can be given enough energy
to allow a transition to an unfilled state at E > Ef Electrons with energies
appreciably less than Ef cannot be excited because there is not enough
only possible for an electron to
transition to a state
thermal energy available. Consequently, only a small fraction of the total
number of conduction
electrons can take part in the thermal excitation,
and the contribution of the electrons to the heat capacity of the metal
will be very small. For this reason the heat capacity of a metal at normal
temperatures is observed experimentally to be just what would be calculated
from the thermal motion of the atoms alone. When this explanation was
first proposed, it answered a long standing puzzle because, in classical
physics, the heat capacity arising from the thermal motion of the conduction
electrons should be quite comparable to fhat arising from the thermal
motion of the atoms.
Figure (12-6) shows the energy distribution for the particles of a Fermi
gas at T = 0. Energy distributions for a Fermi gas at several other values
of Tare shown in figure (12-8). Note that at each temperature the drop
to zero occurs in an energy range of width about equal to kT, because
388
IDENTICAL PARTICLES
[Ch. 12
particles of energy much less than Ef
kT cannot be excited so the
energy distribution of these particles cannot be different from the T =
distribution.
The curves
in this figure are plotted
Jf(E) dE
The quantity N(E)is
proportional to
l/i
.
2N(E) F(E) dE
(12-41)
the density of states for Fermi particles (12-37)
The quantity F{E) is
The energy
Figure 12-8.
from the equation
and is
the Fermi probability distribution
distributions for a system of Fermi particles at several tem-
peratures.
which
specifies the probability that a Fermi particle, in a system in
thermal equilibrium at temperature T, will be in a state with energy E.
It is
F
where
is
W-
eiE - E !)lkT
a constant which depends on
when &r< Ef
The role of
(12-12)
1
T in
such a way that
E = Ef
the Fermi probability distribution in the description of a
system of Fermi particles
is
analogous to that of the Boltzmann probability
distribution (2-23) in the description of a system of classical distinguishable
particles.
The
latter probability distribution
P()
,E/kT
(12-43)
specifies the probability that a particle of such a system, which is in thermal
equilibrium at temperature T, will be in a state with energy E. The
difference
reflects
between the probability distributions (12-42) and (12-43)
behavior of Fermi particles and
the difference between the
THE FERMI GAS
Sec. 6]
classical particles
and
a Fermi gas with
389
primarily an effect of the exclusion principle.
is
kT > Ef
the thermal excitation
is
very large,
all
For
the
particles are highly excited, and the states are very sparsely occupied.
Then
the chance that a Fermi particle would be prevented from
was already
no longer important. But
transition to a state, because that state
the exclusion principle
the constant
is
is
in equation (12-42) is large
filled, is
making a
so small that
in the limit
kT > Ef
and negative and the term
compared to the exponential. Then the probability distribuF{E) and P{E) are identical except for a multiplicative constant
negligible
tions
which only affects their normalization.
To complete this discussion, we mention that particles which must be
described in quantum mechanics by symmetric eigenfunctions are given
the generic name Bose particles. Even though they are indistinguishable,
these particles do not obey the exclusion principle because a symmetric
eigenfunction, such as (12-8), does not vanish if all the quantum numbers
of two or more particles are the same. The probability that a particle
in a Bose gas in thermal equilibrium at temperature T will be in a state
with energy E is given by the Bose probability distribution,
B
In general, the constant
W-
e,
E - Eo
L_
depends on
in
(12-44)
t
such a way that the Bose
probability distribution becomes identical, within a multiplicative con-
Boltzmann probability distribution at high temperatures.
However, the two probability distributions differ at low temperatures
where a large number of particles occupy the same state since, even though
they both describe particles which do not obey the exclusion principle,
the particles are indistinguishable in the Bose case and distinguishable
in the Boltzmann case. The He atom is an example of a Bose particle
because it is a tightly bound aggregate containing an even number of
stant, with the
antisymmetric particles. The extremely interesting properties of a system
of He atoms at low temperatures (liquid He) are due primarily to the
symmetry character of these atoms. The quantum is another example of a
Bose particle. In this particular case the constant in equation (12-44)
is identically zero, and the probability distribution is
B() quan ta
By setting E
J_
(12-45)
^
equation can be used to evaluate the average energy
v in an enclosure containing a large number
of quanta in equilibrium at temperature T. The resulting expression is
identical with Planck's equation (2-31) for the average energy contained
carried
hv, this
by quanta of frequency
IDENTICAL PARTICLES
390
in electromagnetic standing
density of a black
body
waves
in such
[Ch. 12
an enclosure. Thus the energy
cavity can be evaluated
by thinking of
it
as con-
taining a system of standing waves, or as containing a Bose gas of quanta.
In Chapter 2
we showed with a simple system how
probability distribution
ics.
is
developed from the theory of
The Fermi and Bose
the Boltzmann
statistical
mechan-
probability distributions are developed from
a variant of that theory, called quantum statistical mechanics, which differs
from the
classical theory only in the definition
of an experimentally
dis-
tinguishable division of the total energy of the system and, for Fermi
particles, in the application
of the exclusion principle. This
is
an important
quantum mechanics which we shall not treat here only
would divert us from its even more important application to
application of
because
it
the study of multi-electron atoms.
BIBLIOGRAPHY
Bohm,
Quantum Theory, Prentice-Hall, Englewood Cliffs, N.J., 1951.
Quantum Mechanics, McGraw-Hill Book Co., New York, 1955.
Sherwin, C. W., Introduction to Quantum Mechanics, Henry Holt and Co..
D.,
Schiff, L.
New
I.,
York, 1959.
EXERCISES
1.
2.
Show
that equation (12-3)
is
a solution to (12-1).
Using the procedure suggested immediately below (12-30), show that the
expression agrees with the results of a routine degenerate perturbation theory
calculation.
Consider two electrons moving in a narrow impenetrable walled tube of
Write proper eigenfunctions for the ground and first excited states
of this system, assuming that the electrons are non-interacting. Evaluate the
energy of the ground state and first excited state of the system in this approximation, and discuss the degeneracy of these states. Now consider the perturbation due to the Coulomb interaction between the electrons. Write equations
3.
length a.
for the energy shifts of the ground state
shifts.
and
first
excited state,
Explain the physical origin of the splitting of the
and evaluate these
excited state by
first
the application of the perturbation.
4.
Verify equations (12-31) and (12-32).
5.
Verify equation (12-39).
Repeat the calculation of exercise 3, Chapter 2 for a system of Fermi
and also for a system of Bose particles. Compare the first calculation
with equation (12-42) and the second with (12-44), and compare both with
6.
particles,
(12-43).
CHAPTER
13
Multi-electron
Atoms
Introduction
I.
A multi-electron atom
Z electrons of charge
by
contains a nucleus of charge
+Ze
surrounded
Each electron moves under the influence
of an attractive Coulomb force exerted by the nucleus and repulsive
Coulomb forces exerted by all the other Z 1 electrons, as well as certain
other forces involving the spin angular momenta which are usually weaker
than those involving the charges. To find an exact solution of the extremely
complicated
impossible.
e.
Schroedinger equation for such a system
is
essentially
Nevertheless, there exists a completely adequate theoretical
description of multi-electron atoms.
In section
by
first
5,
Chapter
12,
we
treated a multi-electron
atom with
Z=
considering an unperturbed system describing a simplified
atom
in which the interaction between the electrons is ignored, and then improving the accuracy of the description by using perturbation theory to
evaluate the effect of this interaction.
larger values of
atom
The modus operandi
for
atoms with
also consists in taking a simplified description of the
an unperturbed system, and then using perturbation theory to
However, if there are a number of electrons,
their interactions cannot be completely ignored in the unperturbed system
because the effect of these interactions is much too large to be treated as a
perturbation. A useful unperturbed system must take into account the
main effects of the Coulomb interactions between the electrons. On the
other hand, the unperturbed system must not be so complicated that the
as
evaluate the corrections.
Schroedinger equation describing the system is insolvable. In effect, the
requirement means that in the unperturbed system the electrons
latter
391
MULTI-ELECTRON ATOMS
392
only
must be non-interacting
[Ch. 13
then can the Schroedinger equation for
Z equations which can
because each of them involves the
the unperturbed system be separated into a set of
be solved without too
much
difficulty
coordinates of a single electron.
An unperturbed system with the required properties can be obtained
by assuming that each electron moves independently in a spherically
symmetrical net potential V(r), where r is the radial coordinate of the
electron with respect to the nucleus.
the spherically symmetrical attractive
This net potential
Coulomb
is
the
sum of
due to the
nucleus and a spherically symmetrical repulsive potential which represents
the average effect of the repulsive Coulomb forces between the electron
and its Z 1 colleagues. The average repulsive potential depends on
potential
the average radial probability densities of the electrons.
probability densities are not
known
Of
course, these
until after the equations
the behavior of the unperturbed system have been solved.
governing
This might
seem to be an intractable situation, but actually it is not. It can be handled
by demanding that the average repulsive potential be self-consistent. That
if
is,
we calculate
the probability densities for the correct average repulsive
and then evaluate an average repulsive potential from these
probability densities, we demand that the potential with which we end up
must be identical with the potential with which we started. This condition
potential,
is sufficient
Two
2,
to determine the correct average repulsive potential.
different self-consistent treatments will be presented.
In section
the electrons in the unperturbed system will be treated as a Fermi gas.
Then
it will not be necessary to solve the Schroedinger equations for the
unperturbed system explicitly, because many quantum mechanical properties of a system of non-interacting electrons have already been included in
the derivation of the equations specifying the properties of a Fermi gas. In
section 3, we shall describe a more accurate treatment of the unperturbed
system which involves solving the Schroedinger equations for the spherically symmetrical potentials. The results of this treatment will provide the
basis for a discussion of the
ground
state of multi-electron
atoms, and of
Then we shall use perturbation theory
effect of the Coulomb interaction between
the periodic table of the elements.
to correct for the fact that the
the electrons
is
only approximately represented by the spherically sym-
metrical repulsive potential, and also to correct for other interactions
such as those involving spin. This will allow us to give a description of
Finally, we shall use time dependent perturbation theory to evaluate the probability that an atom will make a
the excited states of atoms.
from one excited state to another. From all these considerations,
be able to understand the optical and X-ray spectra of multielectron atoms.
transition
we
shall
Sec. 2]
2.
THE THOMAS-FERMI THEORY
393
The Thomas-Fermi Theory
Thomas and Fermi independently developed a simple
The theory is based on
treating the electrons as a minimum energy Fermi gas confined to a region
by a spherically symmetrical net potential energy V(f). Then equation
About
1928,
theory of the ground state of multi-electron atoms.
(12-38) can be used to relate the depth of the potential, at any value of
r,
Fermi gas for that value of r. This relation is obtained
by assuming that the depth of the potential is everywhere such that the
energy levels are filled to the very top. That is,
to the density of the
Et =-V{r)
(13-1)
This condition assures that the total energy of the atom
Solving (12-38) for the density p
p
This procedure
= JH-
is
minimum.
\_-V(r)-p
(13-2)
the distance in which the value of V(r) does
is justified if
not change significantly
is
we then have
large enough,
compared
to the de Broglie
wavelengths of the electrons, that a number of electrons can be localized
volume in which V(f) is essentially constant. The condition is satisfied
atoms of large Z, except very near the center and very far from the
center. The theory does not apply for atoms of very small Z because the
in a
for
condition
is
not
satisfied
anywhere.
obtained by requiring that the electron density be
such as to reproduce the net potential V{r). The potential is evaluated in
terms of p by using Gauss' law. This law states that the integral over any
Self-consistency
is
component of the electric field E times the
element of surface area dS is equal to 4w times the charge contained within
the surface. Consider a spherical surface of radius r centered on the
closed surface of the normal
Then we have a completely spherically symmetrical situation
which Gauss' law has the simple form
nucleus.
in
477r
The
left side
= 4n Ze +
of the equation
is
-ep{r')A7Tr' dr'\
(13-3)
the area of the surface times the electric
which is constant in magnitude and everywhere normal
to it. The first term in the bracket is the nuclear charge. The integral
is the total charge due to the electrons contained within the surface;
2
ep(r') is the charge per unit volume, and 4nr' dr' is the volume element.
field at the surface,
MULTI-ELECTRON ATOMS
394
[Ch. 13
Now the electric field is related to the force acting on an electron, and thus
on the
to the net potential V(r) acting
_ e =
by the equations
electron,
= _^(!)
/r
(13
V
dr
When we
_ 4)
write this as
rf[-Hr)]
dr
equation (13-3) becomes
r2
4zW] = _ Ze2 + 47re2
dr
p(r')r'
dr'
Jo
Differentiating both sides with respect to r gives
d
I=im=A^
U
y
dr\
dr
p{r
dr
'
Using equation (13-2) to express
p(r) in terms of
L-^)])
hT^Y
dr\
r
It is
dr
4^
V(r),
we have
L-V(r)T
(13-5)
5ttFi
convenient to rewrite this in terms of the variable
x,
defined by the
equation
and
to
make
the substitution
-V= X
(13-7)
In terms of these quantities, the differential equation (13-5) becomes
ud
**!&-&)
'
(13-8)
dx2
To
know
obtain an explicit solution to this differential equation, we must
"boundary conditions" for the function %(x). They are obtained
the
from the following considerations. For r>*0, it is clear that the net
potential energy of an electron approaches the Coulomb potential energy
due to the nucleus alone. That is,
V(r)
From
->
-Ze2 /r,
equations (13-6) and (13-7),
we
X (x)^l,
r-+0
see this
x^O
means
(13-9)
THE THOMAS-FERMI THEORY
Sec. 2]
For
->
charge
co,
+Ze
395
an electron experiences a potential energy due to the nuclear
shielded by the charge (Z \)e of the other electrons.
Thus
e?jr,
F(r)
>-
r~>oo
or
rF(r)
-e\
CO
1.0
0.8
0.6
0.4
-\
0.2
o.i
-L
-0.08
0.06
0.04
0.02
nm
The
Figures 13-1.
We make
plot of a function
which
14
12
16
arises in the
the approximation that this condition
rV(r)>-0,
>Nl
10
r->
oo
Thomas-Fermi theory.
is
(13-10)
This
which amounts to assuming that e2 is very small compared to Ze2
approximation is another reason why the Thomas-Fermi theory is not
accurate far from the center of an atom. Using equations (13-6) and
.
(13-7) to write (13-10) in terms of %(x),
x (x)-+0,
we have
*^co
(13-11)
The
solution of equation (13-8), with the boundary conditions (13-9)
and (13-11), has been carried through by several people. Numerical
methods are required, and so the results must be displayed as a table or
a plot. Such a plot is shown in figure (13-1).
The function %(x) determines the net potential V(r) and the electron
density p(r).
In the next section these quantities will be plotted for a
particular atom,
theory.
It will
and compared with the predictions of a more accurate
that, even for an atom of fairly small Z, the
be found
MULTI-ELECTRON ATOMS
396
Thomas-Fermi theory
good
gives a
[Ch. 13
description of the average radial
we
behavior of the net potential and the electron density. Here
shall
only
equation (13-6) says the factor connecting the
and the dimensionless variable x is proportional to
call attention to the fact that
radial coordinate r
Z~ ^.
Since the same dimensionless variable
means
Z~
to
that for increasing
The Thomas-Fermi theory
X/i
.
is
used for
atoms, this
all
the radial scale of p(r) contracts according
predicts that the "size" of
atoms
decreases slowly with increasing Z.
The Hartree Theory
3.
A more accurate theory of the unperturbed system for the ground state
atom was developed by Hartree
of a multi-electron
following years.
system of
moving independently
The equation is
=^~
2m
where Vf
electron
fr
is
we
ET
is
is
the potential energy of
the total energy of the atom,
y>
WriTi* #1,
<i,
rv
&,
9 t>
rz , 6 Z ,
ignored, except that the exclusion principle
will
(13-12)
and
T is the eigenfunction describing the motion of all
That is,
The eigenvalue
Z electrons.
which
E T rpT
equation 10-15), and where F/^)
the eigenfunction
Spin
1 VJLrfrpr =
I= V?Vr + i=i
the Laplacian operator in spherical coordinates for the rth
is
(cf.
the rth electron.
its
and the
in spherically symmetrical
electrons
net potentials.
in 1928
solving the Schroedinger equation for a
It consists in
be described
expect that there
Wt
In fact there
is
is
<f>
(13-13)
z)
is satisfied
in a
way
Since the electrons are non-interacting,
later.
a solution to equation (13-12) of the form
v( r i
0i> i>i) V>( r 2>
such a solution.
e 2,
<t>z)
The equation
V>(
r z>
e z>
(13-14)
<t>z)
splits into
one-particle
Schroedinger equations such as
^
2m
VM'*
0*
&)
Kfo) Vfo.
6i +<)
= EM*,
Oi,
4>d
(13-15)
where
ET
=iE
=
t
still
(13-16)
Vt (rt) are not known. However, the problem
be solved by the following self-consistent treatment.
Initially the potentials
can
THE HARTREE THEORY
Sec. 3]
397
A reasonable guess about the form of the V^r,) is obtained by setting
for all
VAr ) = V(r ),
(13-17)
1.
where V(r^
for the
from the Thomas-Fermi theory
With these common values of V^rJ, all
the net potential evaluated
is
atom with
electrons.
the one-particle Schroedinger equations (13-15) are identical.
These equations are solved by numerical integration, and the one-
2.
y> a (r 6
<^), tpfa, Q t <f> ( ), ..., yfc, it &), ... are
found. They are listed in order of increasing energy of the corresponding
particle eigenfunctions
t,
Each of the quantum numbers a, fi,
i,
stands for
of spatial and spin quantum numbers for one electron.
3. To obtain the ground state of the atom, the one-particle states are
filled in such a way as to minimize the total energy (13-16) and yet satisfy
the exclusion principle. That is, the one-particle states are filled in order
of increasing energy with one electron in each state. Then the oneeigenvalues.
a complete
set
particle eigenfunctions describing the
v(i> 0i,
<M, Vnfo,
K 4*1
Z electrons
win,
#i,
&),
are
V;( r z. e z, 4>z)
4. Next a more accurate estimate of the potentials Vt (r ) is calculated
by integrating the equation obtained by combining Gauss's law (equation
13-3) and the relation (13-4) between the electric field and the potential
energy. In calculating the potential acting on a particular electron, say
(
the A:th electron, the quantity p(r) in equation (13-3)
is set
equal to the
sum
of the quantum mechanical probability densities for all the filled states,
except the state k occupied by the kth electron, averaged over a spherical
surface of radius
p(r)
r.
That
-4
Artr Jo
-Jo
is,
2>.*(i-. e .#V.(-,0,#r
i=x
sin0d0ty
(13-18)
Thus the potential acting on the kth electron is evaluated from a spherically
symmetrical charge distribution due to the nucleus plus the average radial
dependence of charge carried by all the electrons except the A:th. The
Vii r i) which are obtained will, in general, be somewhat different from
the initial estimates taken in step
2 and using the
1.
If they are different, the entire
5.
new set of Vfr,).
procedure
is
repeated, starting at step
After several cycles (2>-3>4>-2->-3-*
is found that the K,-(/" -) obtained at the end of a cycle are essensame as those used in the beginning. Then this set of Vi(rt) is
the set of self-consistent potentials, and the eigenfunctions, calculated
from these potentials are the set of one-particle eigenfunctions which
describe the motion of the electrons in the multi-electron atom.
tially
) it
the
MULTI-ELECTRON ATOMS
398
is
[Ch. 13
There are several obvious approximations in the Hartree theory. One
that the charge distribution due to each electron is effectively treated as
were spherically symmetrical. The justification of this will be discussed shortly. Another approximation is that antisymmetric eigenfunctions, such as (12-16), are not used to describe the system of Fermi
if it
particles.
satisfied
This
is
not a serious
by means of the
electron can be in each
fault,
because the exclusion principle
is
auxiliary requirement of step 3 that only one
quantum
state.
modification of the Hartree
theory to include antisymmetric eigenfunctions has been given by Fock
(1930).
In comparing the results of the two theories,
found that the
it is
error introduced by not using antisymmetric eigenfunctions
Consequently, most calculations have been
theory because
the original Hartree
Even
so, the calculations
relatively easier to handle.
it is
small.
is
made with
are very lengthy and, as yet, results have been obtained only for a limited
number of atoms. However,
tion of multi-electron
It is
found that
all
the essential features of the Hartree descrip-
atoms are quite well understood.
V(r) in
which
independently.
V (r )
the self-consistent potentials
so that, to a good approximation,
we may speak
atom
the electrons of the multi-electron
all
The
are very similar
of a single net potential
are
moving
one-particle eigenfunctions for these electrons in this
spherically symmetrical potential are closely related to the one-electron-
atom eigenfunctions
discussed in Chapter 10.
V-im.m.fo
The eigenfunctions
m m
t
0. 4>)
are labeled
In fact, they can be written
Rm(r) , mi (0) <M#f..
by the same
set
(13-19)
of quantum numbers
as are used for the one-electron-atom eigenfunctions,
quantum numbers are related to each
(10-37). The spin eigenfunction a m
n,
I,
and these
other just as specified by equations
is
same function
exactly the
as in
the case of a one-electron atom. Furthermore, the functions @ lm (6)
and Q> m (4>) are al so exactly the same. The reason is the Schroedinger
equation (13-15) for an electron in a spherically symmetrical net potential
is just equation (10-15). This equation leads directly to equations (10-17)
and (10-18), and
to their solutions
Q lmi (9)
and
<J>
mi (<).
and
Consequently,
dependence of the
one-electron-atom eigenfunctions applies directly to the 6 and dependence
all
the discussion in Chapter 10 concerning the
<j>
<f>
of the multi-electron
atom
As an example, equation
one-particle eigenfunctions.
(10-46) shows that the
sum of the
all
possible values of
is
spherically symmetrical.
certainly also true for the case of n
true for any given n
the
same statement
and
/.
From
2,
0,
and
it
probability
and
2,1
1
This statement is
densities for the one-electron-atom eigenfunctions with n
can be shown to be
we conclude that
the discussion above,
also applies to the one-particle eigenfunctions for a
THE HARTREE THEORY
Sec. 3]
multi-electron atom.
state, the
means
with
Now, when
a multi-electron
atom
is
in
lowest energy one-particle states are completely
that for almost all values of n
all
399
possible values of
for these electrons
is
and
Since the
its
ground
filled.
This
there are electrons in states
sum of
the probability densities
spherically symmetrical, their total charge distribution
Figure 13-2. The radial probability densities
atom according to the Hartree theory.
for the filled
quantum
states of the argon
also. At most, only a few electrons in the highest energy
which the states with all possible values of m might not be filled,
can contribute to any asymmetries in the charge distribution. This is the
is
symmetrical
states, for
justification for treating the charge distributions as spherically symmetrical
in the Hartree theory.
The functions R nl {r) determining the ""radial behavior of the eigenfunctions are not the same as for the case of a one-electron atom because
the net potential V{r), which enters in the differential equation for these
same r dependence as the Coulomb potential
Typical examples of the radial behavior of the
one-particle eigenfunctions are shown in figure (13-2). In this figure we
functions, does not have the
of a one-electron atom.
MULTI-ELECTRON ATOMS
400
[Ch. 13
A atom (Z = 18)
+ l)Pnl {r), where
plot the results of a Hartree theory calculation for the
in
terms
P nl (r)
of the
quantities
2(2/
Y)r
R2Jr) =
2(2/
by equation (10-43). The
one-particle quantum
completely
fill
the
18
possible
atom
electrons of
there are 2/ + 1
figure.
Since
values
shown
in
the
for
and
/
states
the n
possible values of m for each /, and for each of these there are 2 possible
is
the radial probability density defined
this
values of
m
=
s,
there are 2 states with n
=
=
1,
0;
2 with n
2, /
.0;
and 6 with n = 3, / = 1.
2,
1
These states are the ones which are filled in the ground state of the atom
because, as we shall see later, they are the states of lowest energy. The
6 with n
quantity 2(2/
2 with n
X)P nl {r)
is,
3,
0;
therefore, the radial probability density for
the state with
quantum numbers n and
in that state.
It is
/,
times the
number of
electrons
the total radial probability density for the electrons
and the sum of these quantities gives P(r), the total radial
which is the probability of finding some
electron with a radial coordinate in the region of r. Thus
in that state,
probability density for the atom,
P(r)
= Z2(2l+l)P nl(r)
(13-20)
In the same
plotted for the A atom in figure (13-3) as P(r) H
we plot P(r) T_ F which is the total radial probability density for
the A atom evaluated from the Thomas-Fermi theory. We see that the
Thomas-Fermi theory gives a good description of the average behavior
This
is
figure
of the total radial probability density predicted by the Hartree theory,
except that for large values of r it does not decrease rapidly enough.
The Thomas-Fermi theory is also inaccurate for small r, but it is not
apparent in a plot of P(r ) because this
small
r.
It is
is
dominated by the
factor for
measurement
electrons from the
possible to obtain a fairly direct experimental
of the total radial probability density by scattering
atom. The angular distribution of the scattered electrons depends, in a
manner which will be described in Chapter 15, on the total radial prob-
atomic electrons.! It can also be measured by
from the atom, using a technique which will be explained
in Chapter 14. Although neither of these measurements is accurate enough
to resolve the oscillatory structure of P(r) H they do confirm its average
ability density of the
scattering X-rays
behavior.
potential V{r) in
we
dependence of the net
which, according to the Hartree theory, each electron
In figure (13-3)
also indicate the radial
t The reader will recall that this is quite different from the information provided by
scattering alpha particles from an atom. The reason is that in the electron experiments
the mass of the scattered particle equals the mass of an atomic electron, whereas in the
alpha particle experiments the mass of the scattered particle is so large compared to the
mass of an atomic electron that it essentially cannot be scattered by these electrons.
THE HARTREE THEORY
Sec. 3]
401
of the A atom is moving. This is given by plotting the radial dependence
of the effective charge Z(r) defined by the equation
V(r)
The behavior of
Z(r)
is
energy of an electron
is
= -
e Z(r)
(13-21)
For r -> 0, the potential
dominated by the Coulomb attraction of the
easy to understand.
22 i
Figure 13-3. The total radial probability densities of the argon atom according to the
Hartree and Thomas-Fermi theories, and the effective charge of that atom according to
the Hartree theory.
nucleus of charge
increasing
+Ze. So V(r)^ e*Z/r, and Z(r)->-Z
18.
With
Z(r) decreases because the magnitude of V(r) decreases more
rapidly than the Coulomb potential due to the nucleus, as the electrons
r,
with radial coordinates
less than r shield part of the nuclear charge. For
V(r)-> e 2 /r, and Z(r) 1, because the nuclear charge +Ze is
shielded by the charge (Z l)e of the other electrons in the atom.
r^-
co,
By inspecting the
plots of
P nl (r) we
see that, for all the electrons in states
MULTI-ELECTRON ATOMS
402
common
with
values of the
[Ch. 13
quantum number n, the probability densities
same range of r. All these electrons are
terminology we have used before in connection
are large only in essentially the
said to be in the
same shell
Furthermore, the range of
with one-electron atoms.
which the
in
probability densities are large (the "thickness" of each shell)
restricted
is
enough that Z(r) has a reasonably well-defined value in that range. These
circumstances form the basis of a very approximate description of the
motion of the electrons in a multi-electron atom, in which all the electrons
in a particular shell are considered to be moving in a Coulomb potential,
VJr)
Z=
e2
(13-22)
where Z
is
a constant equal to Z(r) evaluated at the average value of r
As an example, we
for the shell (the "radius" of the shell).
figures (13-2)
and (13-3) for the
atom:
In the approximation (13-22), the one-electron
the expectation value of
Z by Z.
This
r,
the total energy,
to within a factor of 2 or 3 for the n
the approximation
many
picture of
is
3 shell of the
useful because
it
can be used
shell,
16,
find
but
it is
A atom.
if
from the
Z3 ~
8,
atom equations
etc.,
quite accurate for the n
is
2.5.
specifying
we
replace
only accurate
Nevertheless,
provides a readily understandable
of the basic results of the Hartree theory.
In the Hartree theory, shells for small n are of very small radii because
for these shells there
Coulomb
is
is
little
smaller than that of the n
1/Z X
shielding,
attraction of the nucleus.
and the electrons
feel the full
= shell
H atom by about a factor of
In fact, the radius of the n
shell
of the
This can be understood in our approximate description from
The electrons in these inner shells are in a region of
equation (10-44).
large negative potential energy, so their total energies are correspondingly
large
and
The
negative.
results of the
Hartree theory show that the magni-
tude of the total energy of an electron in the n
of an electron in the n
shell
of the
shell
atom by a
is
larger than that
factor of about Z\.
This can be understood by referring to equation (10-38).
On
the other
hand, electrons in shells of large n are highly shielded from the nuclear
The
electrons in
these outer shells feel only a small net attraction to the nucleus.
The radius
Coulomb
attraction
by the electrons
in the inner shells.
of atoms of different atomic number can be estimated from equation (10-44), using the appropriate values of Z n This
equation predicts an increase in radius, with increasing n value of the
of the outermost
shell
outermost
less
shell for
rapid than 2
atoms of increasing atomic number, which is somewhat
because Z n slowly becomes larger with increasing
atomic number since the shielding of the nuclear charge is not perfect.
Actually the Hartree theory does predict an increase in the radius of the
THE HARTREE THEORY
Sec. 3]
outermost
shell,
but one which
is
403
much
less
rapid than 2 ,t because the
electrons in the outermost shell do have a small probability, which we have
ignored in our approximate description, of being found in the region near
the center of the atom. Although the probability is small, it is important
because near the center the electrons
feel a
to decrease the radius of the shell.
This
very strong attraction that tends
effect
becomes more important
with increasing atomic number and, along with the effect of the increase
in Z n leads to the important consequence that according to the Hartree
,
theory the radius of the outermost shell is roughly the same for all atoms.
Finally, we can see from our approximate description why the Hartree
theory predicts that the total energy of the electrons in the outermost shell
comparable to that of an electron in the ground state of the
H atom. It is because in all atoms Zn for the outermost shell is always of
the order of unity, so the potential energy is comparable to the potential
of all atoms
is
energy acting on the electron in the
the Hartree theory mentioned
in this
atom. Each of the predictions of
paragraph are borne out by the
observed properties of atoms.
In a one-electron atom,
the
same
all
the electrons in a certain shell have exactly
total energy, if the very small effects like spin-orbit interaction
are ignored. J The reason is that in the one-electron atom the total energy
depends only on the quantum number n, and so is the same for all electrons
in the
same
shell,
even though they
may have
different values of the
In a multi-electron atom this is not true. This
statement should not be surprising because we have mentioned on several
previous occasions that the fact that the total energy of a one-electron
quantum number
/.
atom does not depend on
/ is
a consequence of the fact that the potential
In a multi-electron atom the electrons are
moving in a net potential V{r) which is not exactly proportional to 1/r,
and the total energy Enl of these electrons depends on / as well as on n.
However, since we are ignoring spin-orbit interactions and certain other
effects, the total energy does not depend on the quantum number m s which
determines the spatial orientation of the spin, nor on the quantum number
m which determines the spatial orientation of the "orbit." The important
features of the dependence of E nl on n were described approximately
is
exactly proportional to
1/r.
paragraph in terms of the energy equation for a one-electron
replaced by Z. That is, we took
in the previous
atom with
E nl ~ -ftZie*l2hW
(13-23)
of a
t This statement does not contradict the Thomas-Fermi theory prediction
gradual decrease in the overall radius with increasing atomic number.
number of different
X By this we mean that the total energies of the electrons in a
one-electron atoms are the same if the electrons have the same values of the quantum
number
n.
MULTI-ELECTRON ATOMS
404
For small
becomes
n, this
with increasing n much more
atom because of the dependence
less negative
rapidly than for the case of a one-electron
of Z n on n. For large
n dependence similar
Now,
Zn does not change rapidly,
the difference
increasing
Enl has a less rapid
show
that
Enl
is
indicated by the approximation (13-23).
is
is
largest for
This difference
/.
so
to that of the total energy of a one-electron atom.
the results of the Hartree theory
negative than
of
n,
[Ch. 13
the dependence of
is
7=0
actually
more
Furthermore,
and diminishes progressively with
quite appreciable.
In fact, for large values
Enl
on / can be stronger than it is on n.
The reason for the dependence of the total energy on / is easy to understand in terms of the behavior of Z(r) and the probability density
n,
Wtim im ,Wnim im! in the region near r
0.
In (10-45')
we pointed out
that
V>%imi
<* r \
m V>mm m
t
-*
(13-24)
This was with reference to one-electron-atom eigenfunctions, but it is
equally true for multi-electron-atom eigenfunctions because the term
l)/r 2 in the differential equation (10-27), which governs the
/(/
radial behavior of the eigenfunctions, completely dominates the term
(2^/A 2 ) [E
V(r)] for small r.f Consequently, in the region near r
the
exact form of V(r) is unimportant and all eigenfunctions for spherically sym-
metrical potentials have the r l radial behavior of (10-45).
Now
consider
two electrons in the same shell of a multi-electron atom, one with 1 =
and the other with / = 1. According to (13-24), there is some chance of
electron in the region of small r, but only a very small
/ =
chance of finding the / = 1 electron in that region, since r
1
and
r2
for small r. Although the /
electron will spend only part of
its time near r = 0, this makes an appreciable difference in the average
finding the
value of
its
potential energy because Z(r)
becomes very
so the net potential energy V(r) becomes very negative.
potential energy of the
An
more
electron will be
average potential energy of the
electron.
instructive interpretation of the first of these terms
the substitution u
rR
large near r
Thus
the average
negative than the
Similarly, the average
can be obtained by making
in equation (10-27). This gives
d'u
1(1
1)
2(i
or
which is seen to be equivalent to the Schroedinger equation for motion in one dimension,
with the term /(/ + l)h 2 /2fir 2
L*/2fir 2 added to the potential V(r). This term is often
called the centrifugalpotential, for reasons
which can be seen by considering the classical
energy conservation equation for a particle of mass ji moving under the influence of a
potential V(r). As the particle will move in a plane containing the origin, it can be
THE HARTREE THEORY
Sec. 3]
potential energy of the
405
electron will be
more
negative than that for
>
4
2
r for small r. On the
2 electron in the same shell because r
electron is larger than
other hand, the average kinetic energy of the /
1 electron, since it spends more time in the region of
that for the /
an
larger negative potential energy.
It
happens that these two
balance out for a potential in which Z(r)
\\r
Coulomb
dent of
on
n.
/.
is
effects exactly
a constant; thus, for the exact
potential of a one-electron atom, the total energy
But
in all other cases the total
In a multi-electron atom, the
energy depends on
is
indepen-
as well as
dependence of the average potential
dependence of the average kinetic energy, and the
total energy Enl which is the sum of the average potential and kinetic
than it is for /= 1, more negative
energies, is more negative for / =
=
high Z multi-electron atoms, in
/
is
1=2,
etc.
For
than
it
for
for
1
which Z(r) increases particularly rapidly with decreasing r in the region
near r = 0, and for outer shells of large n in which the n dependence of Enl
is particularly weak, the / dependence of Enl is actually stronger than the
n dependence. We shall see that this plays an important role in the strucenergy dominates the
ture of multi-electron atoms.
All the electrons in a particular shell have probability densities which
same form in the region where their values are
However, the probability densities are significantly different in the
region near r = 0. As we have seen, the result is that the total energies of
the electrons in the same shell are not all the same. For this reason, it is
are of approximately the
large.
convenient to speak of each shell as being composed of a number of
one for each value of /. All the electrons in the same subshell
have the same quantum numbers n and /. Therefore, they have exactly
the same total energy (in the approximation in which we neglect spin-orbit
interactions and other effects). Also they have exactly the same radial
subshells,
probability densities
Pnl (r).
described by the coordinates
r, 6,
=
Also the orbital angular
and
the equation is
^J +i4fJ-\+V(r)
momentum
of the particle
L = *r
is
a constant,
dd
dl
so the equation can be written
which
is
seen to be the classical energy conservation equation for motion in one dimenterm L 2 /2(ir 2 added to the potential V(r). This positive term acts like a
sion, with the
repulsive potential, tending to keep the particle
value of L, the stronger
is this effect.
away from
the origin.
The higher the
MULTI-ELECTRON ATOMS
406
tu
cn
0)
<
a s
00
to
h-4
PO-
CO
CO
in
US
CU
CO
CO
02
(2
00
W3
S
lO
CO
S-,
J w
s
3
S3
CO
CO
T3
00
ft
U5
PQ
[Ch. 13
CO
00
>.
CO
t>.
a
>i
$3
a
-4
t-
to
PQ
CO
Ph
co
<^
to
f^
B ^
O
to
00
^8
ac-
t's
"W
r_
fl
in
,n<
O-
(>
e>
**-.
88
*-.
00
CO
uj
-i
n
&
lO
eg
.e
to
fl
CO
c^-
CO
1
9
r2
<u
a.
**.
25
>
in
e2
CO
S3
s
(M
<M
U5
C
GQ
be
CM
U3
c8
(3
ffl
CO
88
3
us
CO
Ph
Oi
S3
(8
CO
>
00
in
CO
THE PERIODIC TABLE
Sec. 4]
407
The Periodic Table
4.
of the most important properties of the elements are periodic
which specifies the number of electrons
functions of the atomic number
Some
in
This can be demonstrated by constructing
an atom of the element.
a periodic table of the elements such as the one presented in table (13-1).
Each element is represented in the table by its chemical symbol, and also
by
(The entries sometimes appearing below the
quantum numbers for the ground state of an
atomic number.
its
chemical symbol
specify the
atom of the element in a way which will be explained later.) Elements
with similar chemical and physical properties are in the same column.
For instance, all elements in the first column are alkalis and have a valence
of
all
elements in the last column are noble gases and have a valence
We
assume that the reader is at least partially familiar with the
periodic properties of the elements from his study of elementary chemistry.
of
0.
Our
task here
is
to interpret these properties in terms of the theory of
multi-electron atoms.
This interpretation
filled
is
based upon knowing the way in which the outer
atoms are ordered according to energy. In
subshells of multi-electron
be obtained entirely from the Hartree
principle, this information could
theory calculations described in the last section.
However, these calcula-
number of elements.
we need must be obtained from
experiment. This can be done by studying the spectrum emitted when
excited atoms of an element make transitions to their ground states.
In such transitions an electron drops from some high energy one-particle
tions have been carried through for only a limited
Consequently,
much of
state to the lowest
highest energy
From an
energy state which
quantum numbers n and
ground
on a large
which will be described
filled
then the
of
state
of the atom.
found
it is
later, the
it is
possible to
is
This
the highest energy
By making such an
that, except for
analysis
minor deviations
ordering according to energy of the
shown in table (13-2). The first column identifies
quantum numbers n and /. The second column does
subshells
the subshell by the
is
this one-particle state.
to identifying an outer subshell which
number of elements,
the
unoccupied. This state
one-particle state for the ground state of the atom.
filled
subshell filled in the
outer
is
analysis of the spectral lines which are emitted,
identify the
amounts
the information
is
same thing except
as
that n
and
are given by the spectroscopic notation.
commonly used in discussing the spectrum and energy
levels of atoms. The number gives the value of n, and the letter gives the
value of /according to the scheme shown in table (13-3). The third column
This notation
of table (13-2)
is
is
2(2/
1),
which
is,
as mentioned in the last section, the
MULTI-ELECTRON ATOMS
408
[Ch. 13
number of possible quantum states for the value of / characteristic of the
subshell. Thus this column gives the maximum number of electrons that
can occupy the subshell without violating the exclusion principle.
TABLE
(13-2)
The E nergy Ordering of the Outer
Filled Siibshells
Quantum
Designation
Capacity
numbers
of
of subshell
nj
subshell
6,2
5,3
7,0
6d
10
5/
14
7.$
6,1
6p
5,2
4,3
5d
10
4/
14
6,0
6s
5,1
5f
6
10
2
2(2/
4,2
5,0
4d
4,1
3,2
Ap
3d
10
4,0
4s
3,1
lp
3,0
3s
2,1
2/>
2,0
2s
1,0
Is
5s
1)
t
Increasing energy
(less negative)
Lowest energy
(most negative)
TAB LE H3-3}
Spectroscopic
s
notation
Many
of the features of this experimentally observed ordering of the
according to energy have also been explicitly obtained
by Hartree theory calculations. For instance, calculations for the
atom
outer
filled subshells
show
that the energy of an electron in the 4s subshell should, in fact, be
THE PERIODIC TABLE
Sec. 4]
409
lower than the energy of an electron in the 3c? subshell of this atom. Further-
more,
the features which are observed experimentally can be understood
all
at least qualitatively in terms of the description of multi-electron
atoms
provided by the Hartree theory. In our discussion in the last section we
found that this theory predicts that the energy of the subshells becomes
more negative with decreasing values of n and with decreasing values of /.
We see this immediately in table (13-2).
subshell in the n
shell,
The
\s subshell,
which
The two
has the lowest energy.
is
the only
subshells of
2 shell are both of higher energy and, of these, the 2s subshell is
of lower energy than the 2p subshell. In the n = 3 shell the subshells
3s, 3p, 3d are also ordered in energy according to the predictions of the
the n
Hartree theory. However, the energy of the 4s subshell is actually lower
than the energy of the 3d subshell because, for reasons described in the
last section, the / dependence of the energy E nl of the subshells becomes
more important than the n dependence for outer subshells with large
Continuing up the
values of n.
shells
always
satisfies
list,
we
see that the ordering of the sub-
the rule that, for a given n, the subshell with the
lowest /has the lowest energy and that, for a given
lowest n has the lowest energy. Near the top of the
Enl
is
so
subshell
much
is
/,
the subshell with the
list
the
dependence of
stronger than the n dependence that the energy of the Is
lower than the energy of the 5/ subshell.
It
should be emphasized
that table (13-2) does not necessarily give the energy ordering of all the
any particular atom, but only the energy ordering of the
which happen to be the outer subshells for that atom. For
instance, the energy of the As subshell is lower than that of the 3d subshell
for K atoms and the next few atoms of the periodic table, but for atoms
subshells in
subshells
near the end of the periodic table the 3d subshell
is
of lower energy than the
As subshell because for these atoms they are inner subshells
dependence of
Enl is
so strong that
it
dominates the
and the n
dependence.
The characteristics of an atom depend on the motion of its electrons.
The motion of an electron is specified by the set of four quantum numbers
which specify its quantum state. However, in the approximation represented by the Hartree theory, only the quantum numbers n and / are
important. Therefore, in this approximation an atom can be characterized
by specifying the n and / quantum numbers of all the electrons, i.e., by
specifying the subshells occupied
by the various
electrons.
This
is
called
the configuration. The ordering according to energy of the outer filled
subshells being known, it is trivial to determine the configuration of any
atom
in
its
ground
subshells in such a
state
way
because in this state the electrons must
as to minimize the total energy of the
fill
all
the
atom and yet
not exceed the capacity 2(2/ + 1) of any subshell. The subshells will be
Consider
filled in order of increasing energy, as listed in table (13-2).
MULTI-ELECTRON ATOMS
410
first
the
H atom.
atom both
The
single electron occupies the Is subshell.
He atom
the ground state of the
figuration of
is
written
The configuration of He
which
it
Chapter
He:
Is
number of
on the chemical symbol
the atom. In the 3 Li atom one of the electrons
for
the superscript
The 4 Be atom completes the
In the six elements from
10
Ne
the additional electrons
The configurations of B and
5
10
will
table (13-1)
is
and has the configuration
ls 2 2s 2
Be:
to
Is subshell is only 2.
lsW
Li:
2s subshell
4
The reader
The con-
must be in the 2s subshell because the capacity of the
The configuration of this atom is
subshell.
12).
subshell designation specifies the
contains;
specifies the value of
5,
For the He
in our discussion of
written
is
on the
superscript
electrons
in section
The
we saw
electrons are in the Is subshell (as
[Ch. 13
10
Ne
fill
the 2p
are
B:
ls^s^p 1
Ne:
ls 2 2s 2 2p e
note that the periodic table of the elements presented in
divided vertically into a series of blocks with each
row
labeled by the subshell which, according to table (13-2), the elements of
the
row are
tion of any
With this information,
atom using a procedure that
filling.
it is
easy to write the configura-
will
become apparent by com-
parison with the configurations listed above. The configuration of certain
atoms, for which the
last
few electrons are observed to be in different
would be predicted by this scheme, are indicated by entries
below the chemical symbol of the element. To illustrate the way this
information is used to write the configuration of an atom, we present
subshells than
several typical examples
ls22s 2 2p*3s 2 3p e4s 1
23y
24
Cr
Tc
"Ru
46p d
2
2
l5 2i 2/j 6 35 2 3/45 2 3rf 3
2
\s 2s 2p*3s 2 2pHs 1 3d h
1
1s
1
2s 2 2/3 s 2 3/> 8 4s 2 3d 104/5s 2 4</ 6
2s 2 2/3s 2 3/Ms 2 3d 104/> 6 5.s 1 4rf 7
2s 2 2p s 3s 2 3p s 4s 2 3d 10 4p 6 4d 10
57
La
l5 25 2 2/35 2 3/j 6 4j 2 3c/ 104/5j 2 4rf 10 5/65 2 5</ 1
58
Ce
lj 2 2s 2 2/3s 2 3/45 2 3rf 10 4/5s 2 4rf 10 5/7 6 6.y 2 5d 1 4/' 1
59 Pr
s 2s*2p3 s 2 3pHs 2 3d
w4p & 5s 24d la 5p%s 2
4f
THE PERIODIC TABLE
Sec. 4]
We
411
see that in certain cases the actual configurations observed for the
elements do not
strictly
adhere to the predictions of table (13-2).
instance, this table says that the energy of the
the energy of the 4? subshell
24
these subshells are
is
For
greater than
filling.
Yet
in
29
Cu, one of the electrons that could be in the 4s subshell
actually in the 3d subshell. Similar situations are observed to occur
Cr, and also in
is
when
3d subshell
4d and 5s subshells. In 43 Tc the 5s subshell is filled in the normal
manner. But in 45 Rb there is only one electron in the 55 subshell; in
46
Pd both electrons have left the 5s subshell and moved to the Ad subshell.
The 78 Pt and 79 Au configurations show that the same kind of thing can
happen for the 5d and 6s subshells. From these circumstances we conclude
that the energy differences between the 3d and 4s, the 4d and 5s, and the
5d and 6s subshells must be so small while they are being filled that,
although generally the ordering of these subshells is as shown in table
(13-2), in certain cases the ordering can actually be reversed. Configura57
La and in the
tions which disagree with table (13-2) are also observed in
lanthanides (Z = 58 to 71) more commonly called the rare earths.
The table predicts that after the completion of the 6s subshell the 4/'
subshell should fill. But in 57 La, and several rare earths, there is one 5d
electron. A similar situation occurs in the group of elements following
89 Ac, which are called the actinides (Z = 90 to 103).
From the same
mean that the
observations
to
argument as above we interpret these
between the
subshells,
and
energy differences between the 5d and 4/
are being
these
subshells
while
6d and 5/ subshells, are very small
for the
filled.
On
the other hand, certain predictions of table (13-2) are always obeyed.
is exceptional for elements in the first two
columns of the periodic table (13-1), we conclude that
Since none of the configurations
and the
every
last six
subshell
is
always of higher energy than the preceding
subshell while these subshells are being
filled,
and that
s or
in these circum-
is always of higher energy than the preceding p
Therefore there must be large energy differences between the
stances every s subshell
subshell.
subshells concerned while they are being filled. In fact, the energy differences
between every 5 subshell and the preceding p subshell are particularly
large, and it is easy to understand why. Since for a given n the energy of a
subshell becomes higher with increasing /, an s subshell is always the first
subshell to be occupied in a new shell. -.Consequently, when an electron
is added to a configuration with a completed p subshell and goes into the
subshell of next highest energy,
an
s subshell,
which according to table (13-2) is always
first one in a new shell. Compared
the electron will be the
to the electrons in the preceding subshell,
will
be considerably larger,
its
its
average radial coordinate
average potential energy will be considerably
MULTI-ELECTRON ATOMS
412
[Ch. 13
and its total energy will be considerably higher much
higher than for the usual increase in total energy in going from one subshell
less negative,
to the next.
The fact
is a particularly large energy difference between every
and the preceding p subshell has some very important consequences. Consider atoms of the elements 10 Ne, 18 A, 36 Kr, 54 Xe, 86 Rn
in which ap subshell is just completed. Because of the very large difference
between the energy of an electron in the p subshell and the energy it
would have if it were in the s subshell, the first excited state of these atoms
is unusually far above the ground state.
As a result, these atoms are
particularly difficult to excite and are very inert.
Furthermore, they
produce no external electric field because they consist of sets of completely
filled subshells and, therefore, have charge distributions which are spherically symmetrical.! Also we shall see later that atoms with completely
filled subshells produce no external magnetic fields. Such atoms cannot
interact with other atoms to produce chemical compounds, and they
that there
s subshell
constitute the noble gas elements.
for this
atom
the
does not contain a
In addition,
He
is
a noble gas because
an 5 subshell (even though it
has an unusually high first excited
first
unfilled subshell is
filled
p subshell) so it
atom consists of completely filled subshells
and so produces no external electric or magnetic fields. That 2 He is a
noble gas is indicated by its being listed in the last column of the periodic
state,
and
also because the
table instead of the second column. An element such as 20 Ca is not a
noble gas, even though it consists of completely filled subshells, because
in
is
excited state an electron goes to a 3d subshell so the excited state
not far above the ground state and the atom is not particularly
its first
inert.
Another aspect of the particular inertness of the noble gases can be
obtained by plotting, for the various elements, the measured values of the
magnitude of the total energy of an electron in the highest energy filled
subshell.
This
is
the atom, which
such a plot.
value which
equal to the energy required to remove the electron from
is
the ionization energy of the atom.
Figure (13-4) shows
We see that the ionization energy oscillates about an average
is
essentially
independent of Z, in agreement with our earlier
conclusion that the total energy of the electrons in the outer shells
is
f There is obviously no external monopole electric field (the field produced by a single
charge) for any neutral atom. Also, there is no external dipole electric field (the field
produced by two displaced charges of opposite sign) because of the symmetry of the
However, there is an external quadrupole electric field (the field
produced by two identical dipoles arranged along a line with one charge of the same
sign from each dipole coinciding) unless the charge distribution of the atom is spherically
charge distribution.
symmetrical.
Cf. equation (10-46).
THE PERIODIC TABLE
Sec. 4]
413
roughly the same throughout the periodic table. However, the oscillations
are quite pronounced, and it is apparent that the total energy of an electron
in the highest energy filled subshell of a noble gas
atoms are very
We
Li,
25
r?
20
considerably
more
difficult to ionize.
also see that the ionization energy
elements
is
These electrons are very tightly bound, and the
negative than average.
u Na,
19
K,
37
55
Rb, and
is
87
Cs
particularly small for the
Fr has not been measured).
15
Kr
Xe
<D
0)
% 10
Na
Li
Rb
20
10
L
30
Cs
40
J
60
50
70
80
90
100
ZFigure 13-4.
These are the
alkalis.
The
ionization energy of the atoms.
They contain a
single
weakly bound electron in an
Alkali elements are very active chemically because they are
s subshell.
quite eager to get rid of the weakly
bound
electron
and
revert to the
more
arrangement obtained with completely filled subshells. These
elements have a valence of +1. At the other extreme are the halogens,
stable
F,
17
C1,
35
Br,
53
I,
85
At, which have one less electron than
is
required
These elements are very prone to capture an
p
electron, and they have a valence of 1
For the first three rows of the periodic table (13-1), the properties of
the elements, such as valence and ionization energy, change uniformly
from the alkali element with which the row begins to the noble gas with
which it ends. In the fourth row of the periodic table this situation is no
longer always true. The elements 21 Sc through 28 Ni, which are called the
to
fill
their
subshell.
have quite similar chemical properties and also almost
same ionization energies. These elements occur during the filling of
the 3d subshell. The radius of this subshell is considerably less than
that of the 4* subshell, which is completely filled for all the first transition
first transition group,
the
MULTI-ELECTRON ATOMS
414
24
group except
from outside
Cr.
The
[Ch. 13
4s subshell tends to shield the 3d electrons
and so the properties of these elements are all
quite similar, independent of exactly how many 3d electrons they contain.
The chemical properties of 29 Cu are somewhat different from those of
the first transition group because it has only a single 4? electron in the
filled
influences,
outermost subshell. To a lesser extent this is also true for 24 Cr. The
element 30 Zn consists of a set of completely filled subshells and so is
somewhat more inert, as can be seen from its ionization energy. Similar
transition groups arise in the filling of the Ad and 5d subshells.
An extreme example of the same situation is found in the rare earths,
58
Ce through
filling.
T1
Lu.
This subshell
These are the elements in which the 4/' subshell is
lies deep within the 6s subshell, which is completely
the rare earths. The 4/ electrons are so well shielded from the
outside that the chemical properties of these elements are almost identical.
filled in all
The same thing happens
the 5/ subshell
We
finish
is filling
in the actinides, 90 Th
through
103
?.
In this group
inside the filled Is subshell.
our discussion of the ground state of multi-electron atoms
and the periodic
table by calling attention to two points. First, the study
of the electronic structure of atoms gives no indication of why only 92
elements are found in nature. This implies that the instability of atoms
with Z
92 must be due to an instability of the atomic nuclei. We shall
>
Chapter 16 that this is indeed the case. However, it is possible to
produce such nuclei artificially in certain nuclear reactions. It is found
that they have absolutely no difficulty in collecting electrons and forming
atoms which are perfectly stable as long as the nuclei remain stable.
The members of the actinide group with Z > 92, which are called the
transuranic elements, are formed in just this way.
The second point we emphasize is the importance of the antisymmetric
see in
symmetry character of electrons the property of electrons which causes
them to^obey the exclusion principle. If they did not obey this principle,
all the electrons of a multi-electron atom would be in the \s subshell
because it is the subshell of lowest energy. If this were the case, all atoms
would have very small radii, very high first excited states and ionization
energies, and^no external electric or magnetic fields. All elements would
be extremely inert and there would be no chemical compounds, no life,
and no readers or writers of books on modern physics
5.
Excited States of
Atoms
In the previous sections
we have
described qualitatively the energies
of electrons in the subshells that are being
filled in
various atoms.
Now
Sec. 5]
let
EXCITED STATES OF ATOMS
415
us consider a quantitative example of the energies of electrons in
the subshells of a particular atom.
Consider the typical atom
26
Fe,
all
whose
shown in the energy level diagram of figure (13-5).
The diagram is plotted on an energy scale which is linear from to 10 ev,
and logarithmic for more negative energies. The measured values of the
electron energies are
total
26 Fe
energy
atom
in
Enl
its
of an electron in the various occupied subshells of the
ground
state are plotted,
and are also written to the
right
-5
4s
3d
-10
-50
-c
>
-7.9 (-6.58)
-8.2 (-10.3)
3p
56 (-62.5)
-100 -
3s
94 (-94.8)
-500
2p
-1,000 _
2s
-710 (-721)
-850 (-827)
-5,000
Is
7130 (-7115)
-10,000
Figure 13-5.
An energy
level
diagram for the
filled
subshells of the iron atom.
levels. These values are obtained from an analysis of the optical
and X-ray spectra produced by the atom. The procedures involved in
such analyses will become apparent later. Shown in parentheses are the
values of E nl obtained from a Hartree theory calculation for the 26 Fe
atom. The reader will note that the agreement between theory and experiment is generally quite good. He will also note that these data confirm
many of our previous statements regarding the energies of the various
subshells. For instance, the energy difference between each p subshell
and the s subshell of higher energy is seen to be particularly large. Also
the energy difference between the 3d and As subshells is seen to be very
small. It is also interesting to note that the energy of a 3d electron in the
26
Fe atom is actually slightly more negative than that of a 4* electron,
despite the fact that the observed configurations of this atom and the others
in the fourth row of the periodic table show that generally the 4s subshell
of the
MULTI-ELECTRON ATOMS
416
[Ch. 13
before the 3d subshell. This is not at all paradoxical, and it is predicted
by the Hartree theory. Because of repulsive effects, the presence of an
electron in a particular subshell changes somewhat the electric field produced by the other electrons in that subshell. When the energy difference
between two subshells is very small, this can lead to the situation found
in the 26 Fe atom.
Figure (13-5) shows only the subshells that is, the one-particle energy
levels
which are occupied in the ground state of the atom. There are
fills
also
many unoccupied
energy levels
one-particle
of higher energy.
Just as for the H atom, there are an infinite number of discrete one-particle
levels with E nl < 0, and a continuum of levels with Enl > 0.
In an
1
excited state of the atom,
some of
its
electrons go to the higher energy
one-particle levels.
Fortunately, the
tions
is
number of electrons in an atom which make such
transi-
small, because of the nature of the processes responsible for the
an atom. For instance, in a discharge tube atoms are excited
by the bombardment of energetic electrons. One of the energetic electrons
collides with an atom, knocking one or two atomic electrons into levels
excitation of
improbable that many atomic
Another typical process involves
the excitation of atoms by the absorption of quanta. At most, one quantum is absorbed by any particular atom and, particularly for higher energy
quanta, the quantum tends to interact with only one or two electrons.
Thus the excited states of an atom are not so complicated as they could be.
The configuration of an atom in an excited state differs from its ground
state configuration only in that one or a very few electrons are in higher
energy subshells, and there are corresponding vacancies in the subshells
from which they came. For simplicity, we often speak as if only one
electron in an atom can be excited. Even so, it is apparent that there are
still a very large number of different kinds of configurations for an atom
In such a process
of higher energy.
it is
electrons are simultaneously excited.
in
its
excited states.
However, it is possible to divide the different kinds of excitations into
two types. The first type corresponds to excitations in which an electron
in one of the higher energy subshells occupied in the ground state of the
atom
is
excited.
Since the subshells of higher energy are also the outer
subshells, this type corresponds to the excitation of
electrons of the atom.
one of the outer
The second type corresponds to the excitation of an
is one of the inner
examples of these two types of excitations are illustrated
26
in figure (13-6) for a Fe atom. [These diagrams employ the same pseudologarithmic energy scale as figure (13-5).] In an excitation of the first type,
the excited electron may make a transition to any of the discrete or
electron in one of the lower energy subshells, which
subshells. Typical
Sec. 5]
EXCITED STATES OF ATOMS
417
continuum one-particle
levels of energy higher than the energy of its
not true of an excitation of the second type. With
the exception of the 3d level, all the levels through the 4s level contain
their complete complement of electrons.
Consequently the exclusion
original level.
This
is
demands that the excited electron go to one of the one-particle
above the As level, or possibly to the 3d level. This means that the
energy required to produce an excitation of the second type is much higher
principle
levels
3p
3s
2p
2s
It
Figure 13-6.
Illustrating excitations of
the
first
type (optical excitations) and excitations
of the second type (X-ray excitations).
than that required to produce an excitation of the first type. The latter
is typically only a few electron volts, but an excitation of the
second type for the case illustrated in the diagram requires more than 800
energy
ev.
After an excitation of the second type, the very highly excited atom
will eventually return to its
ground state. It does this by emitting quanta.
These are the very high energy quanta which form the X-ray spectrum
of the atom. The much lower energy quanta which are emitted when
an atom returns to its ground state after an excitation of the first type
form the
optical spectrum of the atom (the spectrum in and near the visible
X-rays and X-ray spectra constitute the subject of the next
chapter. In this chapter we discuss only optical spectra. Consequently
range).
we
consider here only excitations of the
optical excitations.
first type, which are often called
These excitations are produced, for instance, by
exposing atoms to low voltage electrical discharges or flames.
MULTI-ELECTRON ATOMS
418
6.
Atoms
Alkali
In an optical excitation, usually
filled
[Ch. 13
it is
only the electrons in a partially
subshell that are excited, because the energy of these electrons
is
usually considerably less negative than the energy of the electrons of any
completely
There are exceptions to
filled subshell.
of an exception
is
26
Fe, for which the energy of
pletely filled 4s subshell
is,
as
we have
In an
is
example
in the
com-
filled
3d
subshell.
However,
normally obeyed.
atom of an
alkali
element the rule
contain a set of completely
An
seen, actually slightly less negative
than the energy of an electron in the partially
the rule
this rule.
an electron
filled subshells,
is
always obeyed. These atoms
the one of highest energy being
subshell, plus a single additional electron in the next s subshell.
As
mentioned before, the energy of the electrons in a filled p subshell is
particularly negative with respect to the energy of an electron in the next
s subshell. Consequently these electrons are not excited in any of the low
energy processes which lead to optical excitations. In essence, an alkali
atom consists of an inert noble gas core plus a single electron moving in an
external shell. The analysis of the optical spectrum of an alkali atom in
terms of
its
excited states
is
relatively simple
because the excited states
can be described completely by describing the single optically active
electron, and the core containing completely filled subshells can be ignored.
The total energy of the
atom is a constant plus
core does not change, so the total energy of the
the total energy of the optically active electron.
convenient in discussing the excited states of an alkali atom to define
the zero of total energy in such a way that the total energy of the atom is
It is
equal to that of the optically active electron.
Using
this definition,
we
present in figure (13-7) diagrams showing the energies of the ground
3
n Na,
state and the first few excited states of the alkali elements Li and
obtained from an analysis of the spectra of these elements, and also the
energy levels of 1 H for n = 2, 3, 4, 5, 6. Each energy level is labeled by
the
quantum numbers n and
configuration.
which
of the optically active electron,
These diagrams do not show a
fine structure
i.e.,
by
its
splitting,
be discussed later.
level diagrams also do not show the results of Hartree
calculations which have been made for these atoms, since the discrepancies
between theory and experiment are generally too small to be indicated.
The Hartree theory works particularly well in calculating the energy levels
will
The energy
of the optically active electron of an alkali element because the net potenV(r), due to the nucleus plus the electrons of the core, actually is
spherically symmetrical, as assumed in the theory. In terms of the Hartree
tial
ALKALI ATOMS
Sec. 6]
theory,
and
it is
419
easy to understand the structure of these energy level diagrams
diagram for 1 H. The dependence of the energy
its quantum numbers n and / is just as
we have described on previous occasions. For a given n, the energy is
their relation to the
of the optically active electron on
/-*^
-1
6
5
_ZeT
__
_. 6d_ _6f_
-t
~4
*Na
Li
5P
5d
5/
6d
5d
4p
4d
4/
5P
Ad
4s
-2
2.
5/
5s
3d
IF
_6.
3d
17
-3 -
IF
2
"2p"
-4 -
-5"5T
2s
-6
Some energy
Figure 13-7.
most negative
levels of
hydrogen, lithium, and sodium atoms.
for the smallest value of
because the electron spends more
time near the center of the atom, where
it
feels the full
nuclear charge.
In the ground state of the 3 Li atom, the optically active electron
2s subshell and
electron in a
electron
is
its
shells
is
in the 2p subshell
more
is
in the
about 2 ev more negative than an n
atom. But, in the
negative than an n
As level
energy
and
first
its
energy
2 electron in 1 H. In
negative than the 3d level.
with large values of n, the
excited state, the optically active
is
only about 0.2 ev
more
u Na the / dependence makes the
However, for the large radii subdependence becomes less important and
MULTI-ELECTRON ATOMS
420
the energy levels of the optically active electron
energy levels of an electron in a X
nuclear charge +Ze by the charge
become very
[Ch. 13
close to the
H atom, because the shielding of the
(Z l)e of the electrons in the core
of the alkali atom becomes essentially complete for an electron in a subshell
of radius large compared to the radius of the core. This occurs for smaller
u Na because the radius of the core is smaller for
the former atom. Putting it another way, although y*ip oc r 21 near r =
for all n, (10-39) shows that the proportionality constant decreases with
values of n in 3 Li than in
increasing n so that y>*%p
the
becomes smaller near
dependence becomes
less
0.
The
result is that
important when n becomes very
TABLE
large.
(13-4)
The Quantum Defects for the Sodium Atom
Note
i"
1.35
0.86
0.01
~0
going from the inner filled subshells, through the outer filled
and then through the unfilled subshells of a particular atom, the
that, in
subshells,
/
Subshell
dependence of EnX first increases in relative importance, is of maximum
importance in the outermost filled subshells, and then decreases in
relative
relative importance.
The energy
levels
of the
atom are given by
E=
the equation
-Rch\n 2
where R is the Rydberg constant. For an
can obviously be written
alkali
atom a
similar equation
E=-Rch/n' 2
if
the
n',
which are not
integers, are given the
quantities are related to the
(13-25)
proper values.
quantum numbers n by
ri
These
the equation
(13-26)
By fitting equation (13-25) to the
ft are called quantum defects.
observed energy levels of various alkali elements, it is found that for a
The
given element the
quantum defects
with the same quantum number
listed in table (13-4).
These
are essentially the
/.
The quantum
same
results explain the basis
formula (5-5) for the alkali elements.
for all subshells
defects for
u Na
are
of Rydberg's series
The
ATOMS
ALKALI
Sec. 6]
421
of the optical spectra emitted by alkali elements show a fine
which indicates that all energy levels are double, except
lines
structure splitting
due to a spin-orbit interaction acting on the
relativistic effects, which are just as
important as the spin-orbit interaction in the case of a one-electron atom,
are generally quite negligible. We can see this by using equation (5-13) to
those for
This
0.
is
The other
optically active electron.
estimate the average velocity of an optically active electron.
equation
we must
replace
ZM
by the quantity
In this
defined by (13-22).
As
roughly equal to unity for the optically active electrons of all atoms,
(5-13) shows that the average value of vjc is roughly equal to l/ times
the average value of this ratio for an electron in the ground state of the
is
atom; that
is,
u/ c
(l/)
lO
-2
.
Since the
quantum number n
for
an atom increases with increasing Z,
relativistic effects other than the spin-orbit effect, which are of the order
of (vjcf, are negligible for the optically active electrons of all atoms except
the optically active electron of
those of very small Z.|
The
of the energy levels of an alkali element due to a spin-orbit
shift
considering equation (11-47).
be understood by
immediately
to an
That equation leads
equation analogous to
This
interaction acting
AETi =
on the
(1
optically active electron can
1-49).
DO +
TTi
4m c
1)
"
Kl
is
1)
*(s
D] [\r
-^\
dr
(13-27)
where
Just as for a one-electron atom,
when
the spin-orbit interaction
is
included
no longer good quantum numbers and the eigenfunctions
describing the behavior of the optically active electron must be those
which are specified by the quantum numbers n, I, j, m^ These quantum
numbers obey the same rules as before. Specifically,
l
and
are
*=*
= /+ior/-l,
=o
j = h
/^0
(13-29)
For
/,
AS
0,
f These
For a
equation (13-27) shows that
&E8
0.
For other values of
and the other negative,
L assumes two different values, one positive
relativistic effects are
Is electron in the
82
not so negligible for electrons in the inner subshells.
U atom, the average value of vjc is about 0.5.
excitations
it
However,
this
energy of the core, and therefore in optical
affects only the definition of the zero of total energy.
affects only the fixed value of the total
MULTI-ELECTRON ATOMS
422
[Ch. 13
according to whether j = I + J or j = / \. Except for / = 0, each
energy level is split into two components, one of slightly higher energy
for the spin
and
of the
momenta "parallel" and one of slightly
momenta "antiparallel." The magnitude
orbital angular
lower energy for these angular
splitting is proportional to the expectation value of (1/r) (dV/dr).
Since both 1/r and the derivative of the net potential V(r)
for small
r,
the expectation value
is
become
large
very sensitive to the behavior of
0. In fact, it is so sensitive that it is not possible
y>i im and V(r) near r
to obtain accurate values of the splitting from Hartree theory calculations.
Nevertheless, the Hartree theory certainly predicts the general features of
the experimentally observed values of fine structure splitting.
TABLE
Spin-Orbit Splittings
Subshell
(13-5)
in
the Sodium
Ap
3P
Atom
5p
6p
Spin-orbit
splitting
21.3
For a given
x 10-* ev
x 10-*
7.63
3.05
x 10~ 4
1.74
x 10~ 4
alkali element, the theory predicts that the splitting will
decrease with increasing n because, with increasing values of this
quantum
number, the eigenfunction ip nUm becomes smaller in the region near
r =
where (1/r) (dVjdr) is large, so the expectation value of this quantity
.
decreases.
This behavior
measured energy
levels of
is
seen,
n Na
for instance, in the experimentally
associated with an electron excited to the
The splittings of these energy levels are listed in table (13-5).
we showed that according to the Hartree theory the dependence
of the net potential V(r) on r, near r = 0, becomes more pronounced with
increasing Z. Thus the Hartree theory predicts that dVjdr increases
subshells.
In section 3
with increasing
Z near r
0,
increase with increasing Z.
and therefore that the spin-orbit
splitting will
This behavior can also be seen in the experi-
mental data. In table (13-6) we list for various alkali elements the observed
splittings of the energy levels for an electron excited to the first p subshell.
Note
that the spin-orbit splitting
becomes quite
large for large values of Z.
Figure (13-8) shows a few of the lowest energy levels of the
Levels for which
quantum numbers except n
n Na
atom.
same are
grouped in columns, and the value of n for each level is labeled by the
configuration of the optically active electron. The quantum numbers
/ and j are given according to the standard spectroscopic notation by a
symbol such as 2P^A which is read "doublet P one-half." The letter
specifies the value of / according to the scheme we have become familiar
the
all
are the
ALKALI ATOMS
Sec. 6]
423
TABLE
Spin-Orbit Splittings
Element
Li
n Na
Subshell
2f
2p
(13-6)
Number
in a
19
Atoms
of Alkali
37
55
Rb
Cs
Ap
5p
6P
72 x 10" 4
295 x 10- 4
687 x 10"4
Spin-orbit
splitting 0.42
x 10" 4 ev 21 x 10" 4
with, except that
it is
conventional to use capitals.
of/ The superscript is equal to (2s + 1)
equal to the number of components into which
value
Figure 13-8.
between these
Some energy
P%
(2
subscript gives the
1),
and
the levels are split
is
also
by the
sodiunvatom, and the possible transitions
levels of the
X/i
levels that are
not
split.
Since
= +
/
\ and negative fory
\,
level should be slightly higher than the associated 2 P^ level, and
the spin-orbit energy
2
The
levels.
spin-orbit interaction, except for
each
similarly for the
is
D 5A
positive fory
and
and
3/i
for the
2
A and F
F-,
h/i
levels.
This
MULTI-ELECTRON ATOMS
424
splitting
is,
of course,
much too
however, we have indicated
on the scale of the diagram;
and 3p 2P,A levels.
small to be seen
for the 3p
it
We have also indicated in figure
3/i
(13-8) the possible transitions between
u Na
energy levels which produce the spectral lines of
with the few levels shown.
[Ch. 13
that are associated
If the reader visualizes the
complete
set
energy levels and transitions, he will certainly be impressed with
complexity, and with the tremendous
of
its
amount of work expended by
Rydberg, and others, in obtaining energy level diagrams from an analysis
of the observed spectra. The analysis became possible only after Rydberg
discovered that the spectral lines could be grouped into various series.
For alkali elements there are four important types of series. These arise
from the
transitions indicated below, f
"Sharp"
n^A ~+ n
series:
"Principal" series:
"Diffuse" series:
"Fundamental"
In a particular series, n
doublet of the
for
3
X/i
3.
and
u Na
is
is
VzM
"*
~* n o 2p
*AM
^AM
-+ n *
the
first
is
doublets of lines
doublets of lines
"o S14
and n
fixed
spectrum
2p
PHM
series:
D WA
D WA
triplets
of lines
triplets
of lines
variable.
member
The
intense yellow
of the principal series
That is, the two lines are due to the two transitions 3 2P,A -*
2
P - 3 2 S, All the lines of the sharp and principal series
lyi
are doublets since they involve transitions from a single
closely spaced pair of 2 P, and 2 P\A
A
levels,
or vice versa.
S,
The
level to
lines
of the
and fundamental series might be expected to be quartets since the
and final energy levels are always both members of a pair of levels.
However, it is observed that the transitions n 2 D > n 2P,A and n 2 FA >
n 2 D do not occur. From an analysis of many spectra, it is found that
diffuse
initial
byi
3/i
these are particular examples of a general selection rule for the
number j. Transitions only occur
A/=0, 1
Because of
triplets.
this rule the lines
In figure (13-8)
on the quantum number
(13-30)
of the diffuse and fundamental series are
we can
/.
quantum
for which
also see the operation of a selection rule
Transitions are indicated in the diagram only
when
A/=l
From
a study of the spectra of the alkali elements,
(13-31)
it is
t The series were named in the early days of spectroscopy for
reasons. These names are the origin of the scheme by which the
/
0, 1, 2, 3,
are represented by s,p, d,f.
found that
this
somewhat obscure
quantum numbers
ATOMS WITH SEVERAL OPTICAL ELECTRONS
Sec. 7]
425
on / is obeyed whenever there is a single optically
Thus equations (13-30) and (13-31) are both also obeyed
selection rule
active
electron.
in the
case of the
atom.
We
of selection rules in
shall discuss the origin
section 12.
Atoms with
7.
An atom
Several Optically Active Electrons
of a typical element contains a core of completely
filled
subshells surrounding the nucleus, plus several electrons in a partially
Since any of the electrons in a partially
filled subshell.
may
an optical
participate in
active.
It is
filled
subshell
excitation, all such electrons are optically
apparent, then, that a description of the excited states that
give rise to the optical spectrum of such
the behavior of
an atom requires a description of
the optically active electrons.
all
number of
Despite the fact that
which determine the behavior of
these electrons, it is possible to understand their influence on the energy of
the excited states by treating them one at a time, in order of decreasing
there are a
importance, with each
already treated.
As
different effects
new
effect
considered a perturbation on the effects
we consider each of
moving independently in the Hartree
a starting point of this perturbation treatment,
the optically active electrons to be
net potential V(r) due to the core plus the other optically active electrons.
In this Hartree approximation the energy of each optically active electron
quantum numbers n and /, and the dependence
quantum numbers is very similar to that of
a single optically active electron in an alkali atom with the same core.f
The total energy of the atom is a constant plus the total energy of the
optically active electrons. Consequently, the energy of the atom is deteris
completely specified by
of this energy
Enl on
mined completely
its
the two
in this approximation
by the configuration of the
by the n and / quantum numbers of these
electrons. Now, associated with each configuration of the optically active
electrons are generally a number of different possible sets of quantum
numbers m and m s Since the energy does not depend on these quantum
numbers, there is usually a quantum number degeneracy for each conoptically active electrons,
i.e.,
Also, there is generally an exchange degeneracy because the
energy does not depend on which of the identical particles has a particular
set of quantum numbers. We see that in this approximation a number of
degenerate energy levels are associated with each configuration. However,
many of these degeneracies are removed when the perturbations are taken
figuration.
The reason
is
that V(r)
is
very similar to the net potential due to the core alone.
MULTI-ELECTRON ATOMS
426
into account.
The
situation
is
[Ch. 13
analogous to the case of an alkali atom in
which some of the degeneracies associated with each configuration of the
optically active electron are
removed when the
spin-orbit interaction
is
treated as a perturbation.
The perturbations which we must include correct for the effects that are
The two most important correc-
ignored in the Hartree approximation.
tions are for:
1.
the fact that the Hartree potential acting
on each
optically active
Coulomb
electron describes only the average effect of the
interactions
between that electron and all the other optically active electrons, and
2. the magnetic interactions between the spin and orbital angular
momenta
of each optically active electron.
There are also
relativistic corrections, corrections for interactions
between
the spins of the optically active electrons because of magnetic interactions
between the associated magnetic moments, and certain other corrections;
of them are very small and can generally be neglected.
have seen one manifestation of the residual Coulomb interaction
2
effect (correction 1) in our treatment of the He atom. In this atom the
energy level for the \s2s configuration is split into two levels, one corresponding to the triplet states for which the energy is lowered, and one
but
all
We
corresponding to the singlet state for which the energy
is
true of the \slp configuration.
The reader
is
raised.
The same
will recall that this
happens
because the requirement that the complete eigenfunction be antisymmetric
introduces an interaction between the relative orientation of the spins of
The average distance
when the spins are
the electrons and their relative spatial coordinates.
between the two electrons
"parallel1 than it is in the
'
is
larger in the triplet states
singlet state
when
the spins are "antiparallel."
Consequently the positive Coulomb repulsion energy is smaller in the
triplet states, for which the magnitude of the total spin has the constant
value 5"
Vl(l
1) h,
than
it
is
in the singlet state, for
which the
magnitude of the total spin has the constant value S' = 0. Of course,
the magnitudes of the individual spin angular momentum vectors are not
changed by this effect of the residual Coulomb interaction. It is shown by
an analysis of the experimentally observed spectra, and can also be under2
stood theoretically from calculations similar to those we made for He,
active,
optically
that this effect is important in all atoms with two or more
electrons. That is, for such atoms there is a tendency for the spin angular
momenta
of the optically active electrons to couple in such a
magnitude of the total spin angular
energy
The
is
momentum
S'
usually smallest for the state in which 5"
residual
Coulomb
is
is
way
constant,
that the
and the
largest.
interaction effect also produces a tendency for
ATOMS WITH SEVERAL OPTICAL ELECTRONS
Sec. 7]
the orbital angular
in such a
L'
is
the
way
constant.
momenta of
427
the optically active electrons to couple
that the magnitude of the total orbital angular
We
momentum
did not see an example of this in the excited states of
He atom because
we studied one electron is a
most only one non-zero orbital
in all the configurations
Is electron and, consequently, there is at
momentum present. However, it is easy to understand the origin
of the forces tending to couple the orbital angular momenta by considering
angular
the polar plots of the probability density for an electron
spherically symmetrical potential
[cf.
the z axis so that the orbital angular
Figure 13-9.
maximum
Illustrating
and
momentum has
why two optically active
momentum vector.
its
moving in a
Choosing
(10-5)].
maximum possible
electrons tend to couple to give a
total orbital angular
component along
we
figures (10-4)
that axis,
and considering the corresponding polar
see that the probability density of the electron
is
plot,
concentrated near the
plane perpendicular to the orbital angular momentum vector. Thus the
charge density produced by the electron is largest near this plane. Now,
considering two or more electrons, itis apparent thatthe repulsions between
their charge densities will
their orbital angular
have the
momentum
effect
of producing interactions between
vectors.
acting between the orbital angular
These interactions are torques
vectors which do not tend
momentum
to change the magnitudes of the individual orbital angular momentum
vectors but only to make them precess about the total orbital angular
momentum
way that its magnitude L' remains constant.
Which of the possible values of L' corresponds
vector in such a
The question then
arises:
to the state of smallest energy?
not so easy to answer because
However, the physical basis of
the one which is usually most important can be understood by considering
two electrons in a Bohr atom, as illustrated in figure (13-9). Because of the
Coulomb repulsion between the electrons, the most stable arrangement is
This
is
there are several opposing tendencies.
obtained when the electrons stay at the opposite ends of a diameter.
In this state of minimum energy, the electrons rotate together with individual orbital angular momentum vectors parallel, and with the magnitude
of the total orbital angular momentum vector L' a maximum. This
MULTI-ELECTRON ATOMS
428
[Ch. 13
confirmed by an analysis of the spectra produced by atoms
is, for such atoms there is a
tendency for the orbital angular momenta of the optically active electrons
conclusion
is
with several optically active electrons. That
to couple in such a
momentum
in
which 11
L'
is
way
that the magnitude of the total orbital angular
constant,
and the energy
is
usually smallest for the state
is largest.
In contrast to the tendencies produced by the residual
Coulomb
inter-
action, the spin-orbit interaction (correction 2) produces a tendency for
the spin angular
couple with
its
momentum
own
vector of each optically active electron to
orbital angular
momentum
vector in such a
way
as to
leave the magnitudes of these vectors constant while they precess about
angular
their resultant total
magnitude
/.
We know
momentum
that this
is
due
vector,
which
is
to torques arising
of constant
from the
inter-
moment connected with the spin angular momentum
magnetic field connected with the orbital angular momentum.
know that the energy is smallest for the state in which / is
action of the magnetic
and the
We
also
smallest.
8.
LS Coupling
Because the residual Coulomb and spin-orbit interactions tend to
produce effects which are in opposition to each other, atoms in which both
interactions are of comparable importance are very difficult to treat.
Fortunately, such atoms are in the minority. For atoms of small and
intermediate Z, the effects of the residual
Coulomb interactions
are usually
compared to those of the spin-orbit interactions. We can
see evidence of this by comparing the energy difference (~1 ev) associated
with the residual Coulomb interaction coupling of the two spin angular
quite large
in the first excited states of 2 He with the energy difference
(/~10~ 4 ev) associated with the spin-orbit interaction coupling of the spin
momenta
and
orbital angular
momenta
small and intermediate
is
sometimes
who
first
in the first excited states of 8 Li.
are said to be examples of
also called Russell-Saunders coupling after
LS
Atoms of
which
two spectroscopists
coupling,
explained the structure of the energy levels of these atoms.
In studying coupling of this type, the perturbation due to the residual
Coulomb interactions is treated first since it is the most important, and
the perturbation due to the spin-orbit interactions is temporarily ignored.
Then
the individual spin angular
electrons couple to
form a
S'
Sx
total
S2
momenta
S, of the optically active
momentum
spin angular
S',
where
(13-32)
COUPLING
LS
Sec. 8]
429
satisfying the quantization equation
and where S' has a constant magnitude
S'
Vs\s'
1)
(13-33)
Also the individual orbital angular momenta L 4 of the optically active
electrons couple to form a total orbital angular momentum L', where
L'
=L +L +
2
and where L' has a constant magnitude
L'
L,
(13-34)
satisfying the quantization equation
+l)h
Vl'(l'
(13-35)
These vectors couple in such a way that all their magnitudes S{ and L t
also remain constant. Because of the perturbation, the energy of the atom
depends on S' and L'. Then the quantum states of the same configuration,
but associated with different values of S' and L', no longer have the same
energy.
has the
The
state
minimum
The dominant
account as a
included in
with the
residual
first
LS
maximum
possible values of S'
and L' usually
energy.
Coulomb
interactions having been taken into
perturbation, the smaller spin-orbit interactions are
coupling as an additional perturbation.
This
considering a spin-orbit interaction between the angular
vectors S'
and
L'.
done by
This interaction couples these two vectors in such a
that the magnitude /' of the total angular
J'
is
is
momentum
constant, and S'
L'
way
momentum
(13-36)
S'
and L' remain constant. The magnitude
J'
is
also
quantized according to the usual equation
J'
Vj'ij'
+l)h
(13-37)
As a
result of this additional perturbation, the energy of the
also
on
J'.
minimum
The
state
minimum
with the
atom depends
possible value of J' has the
energy.
Figure (13-10) illustrates the
LS coupling
way
the various angular
momentum
normally the one of
minimum energy for two optically active electrons with quantum numbers
d = 1, .si = i and /2 = 2, sa = . The spin angular momenta S2 and S2v
precess about their sum S', and this vector has its maximum possible
The orbital angular momenta Lx
magnitude (corresponding to s'
1).
vectors
combine in
in the state
which
is
and L2 precess about their sum
magnitude (corresponding to /'
sum J',
to j'
which also has its maximum possible
Also S' and L' precess about their
but this vector has its minimum possible magnitude (corresponding
Finally, J' precesses about the z axis in such a way that its
2).
L',
3).
MULTI-ELECTRON ATOMS
430
component J't along that
axis
is
a constant which
satisfies the
[Ch. 13
quantization
equation
J'z
(13-38)
m'fi
where
m;=-j',-j'+l,...,+j'-h+j'
Figure (13-10)
is
drawn
of the total angular
for
m\
momentum
=/. The
J'
and
(13-39)
quantization of the magnitude
its z
component
J'
z
is
a necessary
consequence of the absence of external torques acting on the atom
Figure 13-10.
in
Illustrating
the coupling between the various angular
a typical IS coupling state of
momentum
vectors
minimum energy.
Figure (13-10) illustrates only one of the quantum states that can be
formed in LS coupling by two optically active electrons with quantum
numbers lx = 1, sx = \ and /2 = 2, s2 = |. In fact, there are twelve
different sets of states, with different quantum numbers s', I',]', that can be
formed by these two electrons. And each of these sets consists of 2/" + 1
quantum states corresponding to the 2j" + 1 different possible
values of m'j. The rule specifying the possible values of m,' is expressed by
equation (13-39). The rules specifying the possible values of s', l',j' are
different
conveniently expressed with reference to vector addition diagrams employ-
ing vectors whose lengths are proportional to the values of the
quantum
LS
Sec. 8]
COUPLING
431
numbers. For the two electrons in question, these diagrams have the form
shown
s\
in figure (13-1
l',j'
shown
1).
The reader may
verify that the possible values of
in the vector diagrams agree with those obtained
from the
equations
=
=
V
j'
s2
\k
+
+
hi \li-k\
\s'-i'\,\s'-r\
= H\{.
h=2
s'=l
+k
+
'i
i,...., s
'
(13-40)
i'
h=2
'->
Jl=l
U=2
s'=0
f=2
l'=3
r=\
s'=l
s'=\
s'=l
f=3
r=3
I'
S '=l
1=3
l'=2
;=>t>
i'=2
/=4
/=3
/=3
j'=2
/=2
/=1
j'=2
\_
-/
'-l, /'=3
i'=l,
f=3,/=3
Figure
13-11.
/2
2, s 2
Since sx
s'=0,
/=1
/=o
\-
f-2
i'=2
s"=0,
-f|-
C=2,/=2
s'=0,
i'=l,/=l
Vector addition diagrams for the quantum numbers
I,
st
J.
s2
$,
the
first
equation gives
s'
0, 1
(13-40')
These equations can be proved by the same kind of arguments as were
used in proving equation (11-44). The equation for s' is, of course,
the
same
as (12-25).
Equations (13-40) are the rules specifying the possible values of s', l',f
MULTI-ELECTRON ATOMS
432
LS
for the general case of
coupling of any two electrons.
[Ch. 13
For any
electrons, the rules can be written
s'
+s +
jsj
+ sN min
\
|si
s2
+ 1,
s + s +
+ lA'Imin + *>
k + k +
+
.
I'
|li
+h+
'
'
Iwlmin'
/'|
'
+U+
I'l
'
'
f=
s'
/'|, Is'
1,
2.
s'
sN
In
(13-41)
Since the magnitudes of the vectors
= 0,1,2,3, ..., N/2,
= \,%,i,l,...,Nj2,
s'
equation gives
s N are all , the first
s'
N=
which reduce to (13-40) for
s1; s 2 ,
sN min
N even
for N odd
for
(13-41')
As an example of the second equation, consider three electrons with
quantum numbers lt = 1, /2 = 2, /3 = 4. For this case the minimum value
of
/'
is
|l
12
l3
7.
|li
+ h + Islmin =
As
min
to k
and the maximum value of
1,
another example,
if
= 1, 2 = 1, 3 = 1,
+ h = 3- We shall
/
lx
/' is l
x
then
/'
/2
+h
ranges from
soon see
that, for
two or more electrons with the same n
and / quantum numbers, the exclusion principle puts certain restrictions on
the possible values of s', l',f in addition to those indicated by equations
configurations in which there are
(13^0) and (13-41).
In the Hartree approximation the eigenfunctions y> T describing the
system of optically active electrons are labeled by the quantum numbers
of n, I, m m s of each of the electrons. In this approximation, all the %p T
associated with a particular configuration of these electrons are degenerate.
x
In the next approximation, the residual Coulomb and spin-orbit interactions of the optically active electrons are treated in LS coupling as
perturbations. These perturbations thoroughly mix the degenerate ip T
form an equal number of new eigenfunctions ip'T, which are certain linear
combinations of the y> T .-\ Each of the %p'T describes a state in which the
optically active electrons couple so as to yield a particular set of S', L',
Consequently the f'T are labeled by specifying the configuration of
J', J'z
to
quantum numbers s', V, j', m-.
LS coupling on the values of
on the configuration, most of the degeneracy is
the optically active electrons plus the
Since the energy of the atom depends in
S', L', J', as well as
removed.
t This
is
The
eigenfunctions
T have only a
tp'
trivial
degeneracy with
completely analogous to the mixing of degenerate one-electron eigenfunctions
(cf. section 5, Chapter 11). As in that case, it will not be
by a spin-orbit perturbation
necessary for us to
know
the exact form of these linear combinations.
Sec. 8]
COUPLING
LS
433
quantum number m'j which specifies the spatial orientation of
momentum vector. The energy levels of the atom are
respect to the
the total angular
LS coupling by the configuration plus s', I', ;", using the spectroscopic notation introduced previously. If we are not interested in the value
of m'j, the same notation can also be used to label the quantum states.
labeled in
For
instance, the energy level
and quantum
/
^-
states
of an atom in which
*=2
,'=2
l=l
j'=3
^F
i
/
/
ndn'p
y'=?,l,0
.'-1
y^Pl
-J^JLQ^
T=l
_L^i__ _H_L
)'=
3,2,1
^0
3l) 3
2
\\
/=
r=
4,3,2
%
Figure 13-12.
The
splitting of the
energy levels
in a typical
LS coupling configuration.
have the quantum numbers = 1,
and couple in such a way that s' = 1, /' = 1,
2, are identified by the symbol 1 sip 3P2
The first part of the symbol
gives the configuration, and the last part gives the other quantum numbers
according to the scheme described in section 6.
The splitting of the single degenerate energy level of a particular
configuration, say the nd n'p configuration of an atom with two optically
active electrons, due to the residual Coulomb and spin-orbit perturthe
=
/=
/
two optically
and n = 2,
active electrons
1,
bations
is illustrated in figure (13-12). For obvious reasons
we cannot
present explicit equations from which the energies of all these levels can
be evaluated. However, we can write an equation which gives the
dependence of the spin-orbit interaction energy. This is the effect which
produces the fine structure splitting of the levels for s' = 1, and a given
/', into triplets of levels.
Consider equation (13-27), which is
AEg.r.
'
2
4mb(r^r)
c
'
1)
/(/
1)
s(s
1) ^
This equation was developed for an atom with one optically active electron,
but the same equation written with primes gives the interaction energy of
434
MULTI-ELECTRON ATOMS
the total spin and orbital angular
momentum
and L' of any
atom
in which
any
energy of the atom
vectors S'
which these vectors can be defined, that is,
is valid. Thus in LS coupling the total
contains a term
atom
LS
in
coupling
AEs
..
K[j'(j'
1)
/'(/'
[Ch. 13
1)
s'(s'
(13-42)
1)]
where the quantity
K=
dV{r)
4m V
dr
has the same value for all energy levels of a configuration with common
values of s' and /'. Let us calculate from this equation the separation in
TABLE
(13-7)
Fine Structure Splittings in the Calcium
Atom
Ratio
Con-
4s4p
4s3d
P1; 3P
P1 3P
3d3d
D 2 3 D1
3
D 2 3 D
3d4p
Levels
Separation
Levels
figuration
4
16.7 x 10" ev
4
64.9 x 10" ev
x 10" 4 ev
4
33.1 x l(T ev
16.9
VPi
P2> 3P1
*D 3 3 D 2
D3 3D
,
Separation
Exp.
Theo.
33.3 x l(Hev
131.2 x 10" 4 ev
1.99
2/1
2.02
2/1
26.9 x 10" 4 ev
1.59
3/2
49.6 x 10~ 4 ev
1.50
3/2
energy between the adjacent levels of a configuration with common values
of s' and /', which are called energy levels of the same multiplet. If the
quantum number associated with the level of lower energy is/, the quantum
number
associated with the level of higher energy
separation
K[()
is j"
1,
and the
is
- s'(s' +
1)0"
2)
/'(/'
1)]
+ 1) - s'(s' + 1)] - K[j'(j' + 1) = K[(j' + l)(j' + 2) -/(/' + 1)] =
/'(/'
2K(f
1)
1)
(13-43)
separation in energy of adjacent levels of the same multiplet is
proportional to the total angular momentum quantum number of the
level of higher energy. This is called the Lande interval rule. In table
(13-7) we present some energy separations between adjacent levels of
The
20
Ca atom, and
multiplets which are observed in the
of these separations with the ratios predicted from the
compare the ratios
Lande interval rule.
Since we compare ratios, the quantity 2K cancels out. The excellent
agreement between the experimentally observed and theoretically predicted ratios, which is typical for atoms of small and intermediate Z,
provides convincing evidence of
LS coupling in
these atoms.
Sec. 8]
COUPLING
LS
435
In discussing the coupling of the angular
active electrons,
we have not
momenta
of several optically
yet considered the effect of the exclusion
But, for configurations in which two or more of these electrons
have the same n and I quantum numbers (are in the same subshell), the
exclusion principle cannot be ignored because it imposes important
restrictions on the way in which their angular momenta can couple.
To determine these restrictions, we consider the Hartree approximation
principle.
in
which
electrons.
n,
I,
m m
Then
f,
are
good quantum numbers
for the optically active
the exclusion principle states simply that
can have the same
set
of four quantum numbers.
We
no two electrons
shall see that this
/', w ',y", wj which
components of S', L', J', and which also
are all good quantum numbers in the Hartree approximation. Now,
although in LS coupling the 2 components of S' and L' are changed by
the application of the residual Coulomb and spin-orbit perturbations,
S', L', J', J'
are not changed.! Consequently, any restrictions we find in
z
the Hartree approximation concerning the quantum numbers s', l',j", m'j
limits the possible values of the
quantum numbers s', m's
determine the magnitudes and
will
be equally applicable in
Consider
first
LS
coupling.
a configuration ns 2
same quantum number n and so are
in
which two
s electrons
have the
same 5 subshell. Since for
it is only possible that m
0, these two electrons actually have the
/ =
same values of the three quantum numbers n, I, m To satisfy the exclusion
principle, their m s quantum numbers must be different; one electron
must have m s = + and the other must have m s = J. According to
(12-23), (12-24), and (11-21), m's is equal to the sum of the m s quantum
numbers of the two electrons. Thus the only possible value for this
quantum number is m's = + J \ = 0. Similarly, m[ is equal to the sum
of the m, quantum numbers of the two electrons, and so has only the
single possible value m\ =
+ = 0. According to (1 1-43), m- = m[ +
m'g so the only possible value of this quantum number is m'- =
+ = 0.
Since the three quantum numbers m's m{, m'j can have only the single
value 0, it is apparent from equations such as (12-24) that the quantum
numbers /, /',/ also can only have the single value 0. Thus the two electrons can couple to form only a single quantum state. This is the state
X
SQ in which the spin vectors of the two electrons are "antiparallel."
If it were not for the exclusion principle the m s quantum numbers of the two
electrons could be equal, and there would be the three additional S S
1
states in which the spin vectors are "parallel." These conclusions should be
in the
t In the presence of the perturbations, S'z and L' have no definite values and m', and
z
ml are not good quantum numbers. However, since S', L', J', J'z have definite values
whether or not the perturbations are present, it is apparent that the application of the
perturbations cannot change their values.
MULTI-ELECTRON ATOMS
436
familiar to the reader because
Is 2
ground
we have already used them
state configuration of the
He
Consider next the configuration np 2
[Ch. 13
in the case of the
atom.
in which two optically active
same p subshell. This case is more complicated, but
it can be handled by the same procedures used above.
In table (13-8) we
list all the possible sets of values of m and m for the two electrons which
s
,
electrons are in the
Possible
Number
1
2
3
4
5
6
7
8
TABLE (13-8)
Quantum Numbers for an np 2
m,
m.
m,
m.
+1
+1
+1
+1
+1
+1
+1
+1
+2-
+1
~"21
+h
+i
+i
+i
2
1
2"
+1
+*
15
do not
+2
+1
+1
+2
+1
+1
+1
-1
+1
-1
2
1
+1
-1
+1
-1
-1
-1
-1
-1
-1
-1
+i
+i
+i
-1
-1
-1
14
+1
+2
>
11
13
-1
-1
+1
mi
2
1
2
12
+1
+i
10
mi
m'
2-
-l
-l
"2
Configuration
+1
+1
2
1
_1
-1
_i
-2
For each
violate the exclusion principle.
sponding values of the quantum numbers m'K
-1
-1
set
we
also
m[, m'}
list
-2
-2
the corre-
There are 15
different sets of m and m s quantum numbers for the two electrons which
satisfy the exclusion principle, and a number of others, such asm, = +1,
m s = +J; i= +1, m s = +J, which are ruled out because they violate
this principle.
The problem now is to identify the allowed quantum
states, specified in Table (13-8) in terms of the quantum numbers m'
s
,
m'} with the
numbers s', l',j".
m'j,
specification of these states in terms of the
First
we
use equations (13-40), with
lx
l2
quantum
= 1, and
find that the various possible combinations of*', l',j' values are only those
which are
specified in the following list by the spectroscopic notation:
3
3
3
PV 3P2 *D lt 3 D 2 3 D3 The *Da quantum states are
^o, !/>!, *D S 5l7 P
immediately ruled out because, for these states with j" = 3, equation
(13-39) shows that there would necessarily be wj values of +3 and 3,
but there are none in the table. Since there are no 3 > 3 states, there can be
,
no
D 2 or 3 Z)
states.
All these states correspond to S'
and L' vectors of
Sec. 8]
COUPLING
IS
the same magnitude
437
they are states of the same multiplet, and they stand
must
and
be quantum states with s'
and
m's
s',
=
the
be
satisfied
only
by
These
requirements
can
m{
V
V
or
fall
Now,
together.
entry
number
quantum
values
states
s'
in table (13-8) says that there
2, since
s'
D2
1, 2.
such states corresponding to the five
Entry number 2 says that there must be
This requires the presence of the states
There are
= 2, 1,
m'j
states with
3
/'
0,
and
/'
five
1.
PX 3 P2 For 3P there is one quantum state corresponding to wj = 0,
S
P1 there are three states corresponding to m- = 1, 0, 1. For 3P2
3
,
For
= 2, 1, 0, 1, 2, The number of
+ 1+3 + 5=14. Only a single state
because all the other m'^ values
is left, and this must be a state with wj =
there are five corresponding to
states
we have
identified so far
of the table have been used.
quantum
quantum
then that this must be the single
S 1 D 2 3P 1 2 Since
quantum numbers s', l',j',
expressed in terms of the
LS
It is clear
Thus, in the Hartree approximation the possible
2
states for two electrons with the configuration np are only those
state
associated with the symbols
in
m,'
is
these restrictions are
they are equally valid
coupling.
In table (13-9)
we
list
the
quantum
states
which are allowed by the
exclusion principle for configurations containing several electrons in the
same subshell. Each symbol gives the values of the quantum numbers
and /' of an allowed multiplet of quantum states. The possible values
j'
of
and m'j for the states of any multiplet can be determined in terms of
Entries are given for cons' and /' from equations (13-39) and (13-41).
ranging
from
no
electrons
in
the
subshell
up to the maximum
figurations
number of electrons consistent with the exclusion principle. For no
1
For
electrons, s' = /' =j' = 0, which is described by the symbol S
one electron in any subshell, /' = / and, since s' = \, the allowed states
comprise a doublet. The allowed states for the other configurations are
determined by calculations similar to those we used in determining the
allowed states of the ns 2 and np 2 configurations, or by more elegant
calculations based on the mathematical theory of groups.
s'
It is particularly
about the
half-filled
interesting to note the symmetries in table (13-9)
subshell configurations.
greatest for this configuration,
a subshell
is filled
and the
The number of
states is
states for a configuration in
which
except for k electrons are exactly the same as the states
for the configuration in
which there are
just
k electrons in the subshell.
a striking demonstration of the effect of the exclusion principle
because, if it were not for this principle, the number of states would increase
This
is
monotonically as the number of electrons in the subshell increased.
There are several interesting consequences of this property. The most
important is that when a subshell is completely filled the only allowed
MULTI-ELECTRON ATOMS
438
TABLE
Possible
Quantum
(13-9)
States for Configurations Containing
Several Electrons in the
ns 1
3
l
np"
S,
tip
P,
tip*
tip
Subshell
ns 2
tip
Same
*S
ns
tip
[Ch. 13
S,
*S
nd"
S,
nd 1
nd 2
nd 3
nd 1
*S
3
np e
D, *G
2
>,
P,
D, F,
G,
P,
*P,
*F
3
D, 1 G, 1 S, 1 D, 1 G, 1 H, H
P, 3 F, 3 P, 3 D, 3 F, 3 G, 3
2
i
D, 2 P, 2 D, 2 F, 2 G, 2 H, 2 S, 2 D, 2 F, 2 G, 2 I
P, i F, i D,
1
3
S, 1 D, >G, 1 S, 1 D, 1 G, 1 H, H
P, 3 F, 3 P, 3 D, 3 F, 3 G, 3
2
D, 2P, 2 D, 2 F, 2 G, 2
*P, i F
1
3
S, 1 D, G
P, 3 F
X
nd*
S,
S
i
nd e
nd 1
nd 3
Z>
nd'
nd 1 "
Z)
*S
same as when the subshell is completely empty. This is the
1
5 in which S', L', and J' all vanish. We proved this for
the s subshell, and it is easy to prove for any other subshell. Let us
consider the np 6 configuration and construct table (13-10), which is
state is the
quantum
state
Possible
Number
Wj
TABLE (13-10)
Quantum Numbers for an np 6
mi
similar to the table
we used
mi
completely
filled
that the total spin
and
is
m m
t
tn's
m\
in',
+1-1-40
up 2 configuration. The table
obviously the single state
orbital angular
momentum
of a
is confirmed by analysis of the optical
Furthermore, if there is no net spin or orbital
subshell are zero
spectra of noble gas atoms.
angular
m m
in studying the
has only a single entry, and this entry
The conclusion
-i -1
+|
+1 +| +1 -i
Configuration
momentum,
no net magnetic moment, so this conby Stern-Gerlach experiments on noble
there can be
clusion can also be confirmed
gas atoms
ground
439
COUPLING
LS
Sec. 8]
or by the fact that the magnetic moment of
47
Ag, which in the
a single 5s electron outside a set of completely filled
measured in these experiments to be the same as the magnetic
state contains
subshells,
is
moment
We
of 1 H, which in the ground state contains a single Is electron.
are now in a position to understand the various features of a diagram
showing the optically excited energy
levels
rH
a,"
iH
of
CO
r5^ ,P
JP
of
CO
atom of small or
of a typical
ja.
CO
jp
CO
r-l
CO
j
CO
r\
U
1
1 _
~2p5s~~
-2 _
2p4s~
2p4d_
2p4p
2p3d
-3
~2p3p
-4 -
2p3s~~
-5
-6
-7
-8
-9
-10
-11
2>
-12
Figure 13-13.
Some energy
levels of
the carbon atom.
Consider the energy level diagram for 6 C presented in
state of this atom has the configuration
ls*2s 2 2p 2 so there are two/? electrons which are optically active. The zero
of the energy scale is so defined that the magnitude of the total energy of
intermediate Z.
figure (13-13).
The ground
atom
ground state is equal to the energy required to ionize singly
Consequently, the diagram is^directly comparable with energy
X
level diagrams for alkali atoms and H, in which the zero of energy is
the
in its
the atom.
defined in the same way.
The energy levels
are labeled by the configuration
of the two optically active electrons, and by the spectroscopic symbol
specifying s', /', /". The fine structure splitting according to the value of
y"
is
not shown because
it is
much too
small to be seen on the scale of the
MULTI-ELECTRON ATOMS
440
diagram.
Consider
[Ch. 13
the average energy of the levels of the various
first
configurations.
In the configuration of lowest average energy, 2p 2 both
electrons remain in the same subshell that they are in for the ground state
,
of the atom.
subshell
In the other configurations, one electron remains in that
is in a subshell of higher energy. Note that the average
and one
energies of the configurations
depend on the n and
quantum numbers of
same way as if
the electron in the higher energy subshell in essentially the
this electron
were the single optically active electron in an
alkali
atom.
The
configuration of lowest average energy provides an example of an np 2
configuration for which the exclusion principle allows only the 1 5 1 2
and
,i,2
The
states.
-P
,i,2
states are
of lower energy than the
S and
because they correspond to a larger value of s', and the 1 D 2
states are of lower energy than the
state because they correspond to a
larger value of /'. Note that the s' dependence is stronger than the /'
1
Z> 2 states
dependence.
almost always found that the energy associated with
interaction coupling of the spin angular momenta
larger than the energy associated with the residual Coulomb interaction
the residual
is
It is
Coulomb
coupling of the orbital angular momenta.
energy levels for the
the one for the
-P
,i,2
Of
the three closely spaced
would be seen on a
states that
of lowest energy because
state is
smallest value of7".
The exclusion
Thus the ground
state of the
larger diagram,
corresponds to the
it
atom
is
the state 2p 2
principle does not apply to the configurations other than
the one of lowest average energy, because in these configurations the
electrons are in different subshells.
Consequently,
all
the states predicted
by equation (13-40) are found. In the 2p3s configuration the energy
maximum
corresponding to
figuration.
/'
s' is
level
lowest in energy, just as in the 2p 2 con-
Deviations from this rule, and from the rule that the
maximum
gives the lowest energy, are seen in the configurations of higher average
energy, but there are
gives the
minimum
no deviations from the rule that the minimum j"
Not shown in the diagram are a few energy
energy.
which have been identified as belonging to the configuration 2s2/j3
The optical spectrum of the 6 C atom can be constructed from its energy
level diagram by evaluating the energy and frequency of quanta emitted
in all possible transitions which do not violate the following selection rules
levels
1.
Transitions can occur only between configurations which differ in
quantum numbers of a
the n
and
more
electrons cannot simultaneously
2.
Furthermore, the
configurations
must
single electron.
differ
by one unit
A/=l
This
is
the
same
This means that two or
make transitions between subshells.
quantum number which is different in the two
rule as (13-31).
(13-44)
Sec. 9]
JJ
COUPLING
441
Transitions can occur only between states in these configurations
for which the differences in the s', l',j' quantum numbers are
3.
As'
A/'=0, 1
A/'
0,
(13-45)
(but not;"
->;"
0)
The first equation prohibits transitions between singlet and triplet states,
and vice versa. Nevertheless, transitions are observed between the
because all excitations of atoms to
population of the 2p 2 D 2 states and,
when these states are highly populated, the total number of transitions
3
per second to the 2p 2 -P ,i,2 states becomes appreciable even though the
probability that any single atom will make this transition is extremely
2p
2 1
Z) 2 states
and the
2/>
2 3
,i,2
states
singlet states eventually lead to the
small because
above
is
the
which
9. JJ
is
it
The second of the equations
The third equation
selection rules apply in any atom
violates the selection rule.
similar to (13-31) but not quite so restrictive.
same rule as (13-30). All these
is an example of LS coupling.
Coupling
Atoms of
small Z, such as
C, are
known
to be
good examples of
LS
coupling because the energy levels are grouped in the characteristic LS
multiplets with ay" dependence of the energy that is very small compared to
LS
JJ coupling
coupling
Qi,H) 2
/
s-
3p
/
/
_
6
u Si
2p3s
3pAs
32
Ge
4p5s
\
j0
Sn
in fie
6p7s
Figure 13-14.
from LS to
McGraw-Hill Book Co.,
Illustrating the transition
Introduction to Atomic Spectra,
J) coupling.
New
From H.
E.
White,
York, 1934.
and /', and because the Lande interval rule (13-43)
However, with increasing Z the j' dependence
is accurately obeyed.
rapidly becomes larger, and the Lande interval rule is obeyed less accurately. This is shown schematically in figure (13-14), which indicates the
the dependence
on
s'
MULTI-ELECTRON ATOMS
442
[Ch. 13
disposition of energy levels in the 2p3s configuration of 6 C, and of
the levels in corresponding configurations of higher
atoms that are in
same column of the periodic table. Significant departures from LS
coupling are often seen in atoms of large Z because, as we saw in section
6, the spin-orbit interactions become large in these atoms.
In fact, for
some of the largest Z atoms the perturbation due to the spin-orbit interactions becomes appreciably larger than the perturbation due to the residual
Coulomb interaction. These atoms are said to be examples of JJ coupling.
In coupling of this type, we assume first that the dominant perturbation
due to the spin-orbit interactions causes the spin and orbital angular
the
momentum
vectors of each optically active electron to couple, forming a
momentum vector
L and S, couple to form
total angular
electron
for that electron.
Thus, for the
/th
J*
The magnitude of this vector
is
Ji
The energy of
L,
(13-46)
S,.
quantized according to the usual equation
VjAji
I)
(13-47)
ft
atom then depends on the quantum numbers j of all
is smallest when these quantum numbers have their
smallest possible values. Next, the weaker perturbation due to the Coulomb
interactions
the
and
the electrons
taken into account. This perturbation couples the individual
is
momentum
total angular
vectors to
form the
total angular
momentum
vector
J'
31
J2
J,
(13-48)
which obeys the same quantization equations (13-37), (13-38), and
LS coupling. The
energy of the atom then also depends, but
on the quantum number/. However, it is not possible
to make a general statement about the values of the /' which yield the
smallest energy. The possible values of the/ and of/" are given by obvious
adaptations of equations (13-41). These quantum numbers must be used
to identify the quantum states and energy levels. t In figure (13-14),
the four levels of the 6p7s configuration of 82 Pb are labeled with the
spectroscopic notation for these quantum numbers. This atom is a fairly
good example of JJ coupling because the (f, %) 12 levels, for which j\ f
and j2 = \, are of nearly the same energy and are well separated from the
(4> l)i,o levels, for which j\ = j2 = \. The subscript gives the value of/'.
Note that the smallest/' gives the smallest energy in one pair of levels and
the largest energy in the other pair. The exclusion principle operates in
(13-39), as in
to a lesser extent,
Of course,
of //coupling.
s'
and
/'
are not
good quantum numbers
in
an atom which
is
an example
THE ZEEMAN EFFECT
Sec. 10]
443
JJ coupling, for configurations in which two or more electrons are in
the same subshell, in much the same way as in LS coupling. Selection
rules for JJ coupling are the same as those for LS coupling except that the
restrictions on As' and AV are replaced by the restriction
A/
for the electron
all
(13-49)
which makes a transition between
A/,
for
0,
subshells,
and
(13-49')
the other electrons.
We do not give a detailed discussion of JJ coupling here because
examples of it are not very frequently found in the behavior of the optically
active electrons in atoms. Most atoms are either good or fair examples
of LS coupling However, examples of JJ coupling are very often found
in the behavior of the protons and neutrons in nuclei. We shall study this
Chapter 16, which concerns the nuclear shell model. The
model explains many properties of nuclei by assuming that the motion of
the protons and neutrons is the same as the motion of the electrons in an
in section 8,
atom, except that the net potential V(r) has a different
dependence, and
that these particles experience a very large spin-orbit interaction.
10.
The Zeeman
Effect
was observed by Zeeman that placing an atom in a magnetic
lines which it emits into several components.
This is called the Zeeman effect. For fields up to a few thousand gauss,
the splitting is proportional to the strength of the magnetic field and is
small compared to the fine structure splitting of the spectral lines of the
atom. This means that in the presence of a magnetic field the energy levels
of the atom which are involved in the emission of the spectral lines must
In 1896
field splits
be
it
each of the spectral
components. In certain special cases it was possible
splitting of the energy levels in terms of a
classical theory developed by Lorentz. But, in the general case, even a
qualitative explanation of the observed splittings could not be given until
the development of quantum mechanics and the introduction of electron
split into several
to understand the
Zeeman
spin.
In terms of the modern theory the Zeeman effect is easy to understand.
Except when it is in a 1 S state, an atom will have a net magnetic moment
(x due to the spin and orbital magnetic moments of its optically active
electrons.!
t
The other
According to equation
(1
1-15),
when
electrons are in completely filled subshells
this
magnetic
moment
which do not contribute
to
is
|i.
MULTI-ELECTRON ATOMS
444
magnetic
in the external
field
will
it
[Ch. 13
have an orientational potential
energy
A=
-(ji-H
or
AE=-/j, k H
where
/j,
the
is
be
level will
component of p along the
/n
To
direction of H.
Each energy
energy levels corresponding to the several
split into several
possible values of
(13-50)
calculate these values,
we
evaluate
(11-10) and (11-22) for each optically active electron.
fl
That
y.
is,
by using
we
take
2ii + 2Zs
where the summation goes over all the optically active electrons, and
where (11-9) and (1 1-26) are used to evaluate g and gs Now let us assume
that the atom in question is an example of LS coupling. Then
.
H
Note that
(jl
is
= _
(13-51)
2S']
not antiparallel to
J'
unless 1/
[L'
L'
S'
or S' =0, because the orbital and spin g factors are different.
coupling, L' and S' precess about J' because of the Larmor pre-
In LS
cession of S' in the atomic magnetic field associated with L', with a
precessional frequency m l which is proportional to the strength of
It is apparent from equation (13-51)
with the same precessional frequency.
In the presence of the external magnetic field H, there will also be a
precession of \l about the direction of this field with a precessional frequency which is proportional to its strength. Thus the motion of ^ is
the atomic magnetic field
that
(x
must
also precess
quite complicated.
(cf.
1-29).
about
However,
J'
if
tne external field
is
weak compared
to
about the external field will be slow
compared to its precession about J', and it is not so complicated to
evaluate the orientational potential energy &E. We have seen in section 4,
Chapter 11, that the strength of an atomic magnetic field is typically of
the atomic
field,
the precession of
4
the order of 10 gauss.
[>.
which is weak
Consider, then, the case of an external magnetic field
4
about J'
rapidly
more
much
precesses
(x.
~10
Since
gauss.
compared to
Sec. 10]
THE ZEEMAN EFFECT
than about H, we
component of
fij,
jji
may
^ H by
evaluate
in the direction of J',
in the direction of
445
H. That
finding /n r which is the average
and then taking the component of
,
is,
-V-- 3 '-
/uJ
(L'
2S')-(L'+S')
J'
and
r g-J?
j'//
rr
where we have chosen the
the dot product gives
gg (L'
j'
S'
J'
IS' 2
(i/
+ sy
31/
S')
1
J'
Writing equation (11-48) with primes,
2S>)
be in the direction of H. Evaluating
z axis to
m = - h (U
31/
|(J'
we have
L'
S'
2
)
So
B= -Pg (L' +
2
ft
2S' a
^ H=
Then according
^B
(3J'
+S' 2
"T
fJ'
-L
2}s
fL'
|-5' 2) li-
j' 2
2
)
JT ,
to equation (13-50), the orientational potential energy is
a
=
AE
i7
S' - L
+
is
^
L_ OJ'
^
= VbH
H
2
~f* B
2
)
>
j>
(13 _52)
The value of
to
this term is of the order of
p B H, which is small compared
the other effects that contribute to the total energy of the atom.
instance, the spin-orbit interaction energy is large compared to
this
all
For
term because it is of the order of ju times the strength of the atomic
b
magnetic field, which is large compared to H. Consequently, we may use
perturbation theory to calculate AE, the contribution of AE to the total
energy of the atom. Using the LS coupling eigenfunctions y' which
are
T
labeled by the quantum numbers s', l',j', m', we havej
,
AE =
y>'*
J
t
Although the eigenfunctions
we may use
yi'
T
AEoWt dr
are degenerate with respect to the
(13-53)
quantum number
the ordinary perturbation theory expression (13-53) to evaluate
because J', has a definite value whether or not the perturbation is present,
so the perturbation cannot produce a large mixing of the eigenfunctions.
m-j,
MULTI-ELECTRON ATOMS
446
[Ch. 13
where the integration is taken over all the coordinates of all the optically
active electrons, and where AEop is the operator formed from AE by replacing S' 2 L' 2 J' 2 J'z with their associated operators. Since
,
function of
/(/'
1)#
2
,
all
these operators with eigenvalues
m'jh, respectively,
s'(s'
T is an eigen-
tp'
\)h 2 ,
/'(/'
l)^
2
,
equations (13-52) and (13-53) immediately
give
pbH
\mr +
m
mr m
+
+
As'
vn
- iv +
"V
+%
2
p%(g=%)-
-#:
-%
-H
+H
*S H (g
Figure 13-15.
transitions
It is
2).
No magnetic
Magnetic
field
field
The Zeeman
between these
splitting of a typical pair of
energy
levels,
and the possible
levels.
convenient to write this as
A =
(13-54)
ftBHgm',
where
j'(j'
1)
+ 1) mr + 1)
s'(s'
I'd'
1)
(13-54')
Lande g factor. Note that g = 1 = g for
= 0, as would be expected. From equation
(13-54), we see that in the presence of an external field of strength
each energy level will be split into 2/' + 1 components, one for each of
the values of wj, and that the magnitude of the splitting will be different
for levels with different Lande g factors.
We illustrate equation (13-54) by the energy level diagram of figure
2
2
(13-15), which shows the splitting of P,AM and S,A energy levels in a
The quantity g is
s' = 0, and g = 2
called the
=g
for
/'
THE ZEEMAN EFFECT
Sec. 10]
447
weak magnetic field. Also shown are the possible transitions between the
P ?/i and the 2 S,A states (which, of course, are in a different configuration).
Transitions do not occur if they violate the selection rule
lyi
Am;.
Even with
0,
(but not m}
-> m;
if
A/ =
0)
(13-55)
Zeeman
effect splits each of the
which can be quite complicated
for lines arising from transitions to or from states of large y". However,
all spectral lines arising from transitions between singlet states are split
into a simple pattern of two components symmetrically disposed about
a third component which has the same frequency as the single zero field
for singlet states so all the g factors have the same
line, because s' =
value, g = 1. By drawing a diagram similar to the one in figure (13-15)
for transitions between singlet states, the reader should show for himself
that all the transitions which satisfy (13-55) lead to only three spectral
lines, even though there may be more than three possible transitions,
this
simplification, the
spectral lines into a pattern of components
because there are only three possible energy differences for these transitions.
We mentioned before that in certain special cases it was possible to explain
the
Zeeman
effect in
terms of a classical theory. The reader can probably
guess that these were the cases in which the spectral lines are split into
three components, because these cases correspond to transitions between
states in
role as s'
which the non-classical phenomena of spin does not play its usual
= 0. These cases are called the normal Zeeman effect; the other,
and more
general, cases are called the anomalous Zeeman effect. This
terminology was introduced long before the modern theory provided a
complete understanding of the Zeeman effect and, from the point of view
of this theory, it is not very appropriate.
An
H, which is weak compared to the atomic
couple S' and L' to form J', cannot disturb this
coupling and only causes a relatively slow precession of J' about the
external magnetic field
magnetic
fields that
direction of H.
However,
if
overpowers these
is
strong compared to the atomic magnetic
and destroys the coupling of S' to L'.
In this case S' and L' precess independently about the direction of H.
This is the so-called Paschen-Bach effect, which is observed for external
fields, it
fields that are
fields
comparable to or larger than
= _ M L +
/
(jl
~104
2S')
and
/,
=_?(l; +
n
2s;)
gauss.
In this case,
MULTI-ELECTRON ATOMS
448
where we have chosen the
z axis in the direction
AE = &(L', +
[Ch. 13
of H. Then
2S'.)
From
we immediately
this
obtain
A = fiB H(ml +
external field
the selection
(13-56)
good quantum numbers for an atom in a strong
and
because L'z
S'z have definite values. It is observed that
rules for these two quantum numbers are
Of course, m' and m's
l
2m's )
are
Am;
Am;
=
=
(13-57)
0,
All spectral lines are split by the Paschen-Bach effect into three
ponents, just as in the normal
Zeeman
com-
effect.
The Zeeman effect plays an important role in experimental spectroscopy. By analyzing the Zeeman splitting of the spectral lines of an atom,
the spectroscopist determines the Zeeman splitting of the energy levels of
the atom. This can be used to identify the quantum number j' of each
level because 2/" + 1 is equal to the number of components into which the
Furthermore, the magnitude of the splitting between any
being known,
two components gives the value of fiB Hg and, \i B and
the
of g depends
Since
value
level.
energy
of
for
the
gives
value
this
the
g
on s', /', j", it can be used to confirm the assignment of these quantum
level is split.
numbers. The quantum number s' is assigned by counting the number
of levels of the multiplet of which the level in question is a member. This
number
number
is 2s'
/'
(except that
when
s'
for the levels of a multiplet
>
is
/' it is
21'
1).
The quantum
assigned, after assigning
s',
by
analyzing the separations between adjacent levels in terms of the Lande
interval rule (13-43), and finding the value of/' which, in consideration of
the restrictions imposed by the third of equations (13-41), leads to the
values of /" that give the observed separations.! This procedure also
evidently leads to the assignment of the
numbers
/'
quantum numbers/. The quantum
for all the levels of a particular configuration being
known
can be grouped into configurations by the similarity of their
energies), the / quantum numbers of that configuration are identified by
using the second of equations (13-41). Identification of the n quantum
numbers associated with the various / quantum numbers is not difficult,
(the levels
if
the n
quantum numbers of
the ground state configuration are
f This is applicable only in the usual case in which the
coupling.
atom
is
known,
an example of
LS
HYPERFINE STRUCTURE
Sec. II]
by making use of the
values of
449
energy of the subshells with
fact that the
increases monotonically with increasing n.
The
common
identification
of the n quantum numbers of the ground state configurations of the atoms
is
based on the same
fact.
Hyperfine Structure
II.
When an
interferometer
atom with extremely high
is
used to measure the spectrum of a typical
resolution,!
many of
the spectral lines are
found to consist of several components. The separation between these
components is typically three orders of magnitude smaller than the separation characteristic of the fine structure of spectral lines. These very closely
spaced components constitute the hyperfine structure of the spectral lines.
They imply the existence of a hyperfine splitting of the energy levels of the
atom.J
Hyperfine structure was discovered in 1891 by Michelson. It was first
explained in 1924 by Pauli as a result of the orientational potential energy
of a magnetic dipole moment associated with the atomic nucleus, in the
magnetic field associated with the motion of the atomic electrons. The
nuclear magnetic dipole moment (/.; is due to a nuclear spin angular momentum I. The magnitude and z component of I are quantized according to
the equations
/= Vi(i +
L = mfi
1)
h
(13-58)
where
m = +
t
/,
1,
+i
1,
+/
These are the same as the quantization equations for the electron spin S.
However, the equations relating the nuclear magnetic dipole moment to
the nuclear spin,
Pit
are not quite the
same
= giP-X m
as (11-22)
and
t Interferometers are described in section 3,
We
(13-59)
H
i
(11-23),
Chapter
which
relate the electron
1.
all atoms have
same atomic weight, as well as the same atomic number. For a source containing
atoms of the same atomic number but a variety of atomic weights (a variety of "isotopes"),
are considering only the spectrum emitted by a source in which
the
there
is
an additional multiplicity of the spectral lines, but this is primarily a trivial
from the slight dependence of the reduced electron mass ji on the atomic
effect arising
weight.
MULTI-ELECTRON ATOMS
450
magnetic dipole
of unity, but
quantity
ft
moment to
N which
fi
some
nuclei
called the nuclear magneton,
is
The factor gt is of the order
and negative for others. The
the electron spin.
positive for
it is
_
=
[Ch. 13
eh
eh
2Mc
1836 2mc
is
1836
defined as
...
uB
(13-60)
mass of the nucleus of a X H atom, and m is the mass of an
electron. Since the nuclear magneton ft N is three orders of magnitude
smaller than the Bohr magneton fi B nuclear magnetic moments are smaller
than electron magnetic moments by the same factor. Consequently the
where
is
the
moment
orientational potential energy of the nuclear magnetic
structure)
is
potential energy of the atomic magnetic
moment
(fine structure),
both magnetic moments experience comparable magnetic
The
interaction of the nuclear magnetic dipole
magnetic
(hyperfine
three orders of magnitude smaller than the orientational
field
couples the nuclear spin angular
the total angular
momentum
vector
J of
because
fields.
moment with the atomic
momentum vector I to
Because of
the electrons.!
this
coupling, both vectors precess about the resultant grand total angular
momentum
vector
=I+
(13-61)
The magnitudes of I and J remain constant, but their z components do not.
Consequently, when the hyperfine structure interaction is considered,
m and m s are not good quantum numbers. However, the two quantum
numbers which specify the magnitude and z component of F according to
t
the equations
F=Vf(f+
J J
\)h
(13-62)
F = mf h
z
where
mt =
-/,
-/+1,...,
+/-1.+/
good quantum numbers and can be used to identify the quantum
and the energy levels. For instance, the term in the total energy of
the atom which gives rise to the hyperfine splitting of the energy levels
can be written
are
states
A,
This
is
,
.
K[f(f +
1)
- j(j +
1)
i(i
completely analogous to equation (13-42);
is
1)]
(1
3-63)
equation (13-63)
obtained by evaluating the orientational potential energy of the nuclear
magnetic
t
To
moment
in the atomic magnetic field,
simplify the notation, in this section
associated
quantum numbers.
we
shall not
which
is
proportional
put primes on J or on the
HYPERFINE STRUCTURE
Sec. II]
to the expectation value of I
J,
451
and equation (13-42)
is
obtained by
evaluating the orientational potential energy of the electron magnetic
moment
in the atomic magnetic field,
tion value of
which
is
proportional to the expecta-
L.
The term Ax j assumes different values according to the value of the
quantum number/. The possible values off are determined by an equation
analogous to the third of equations (13-41), and there are 2/ + 1 of these
values (except that when i > j there are 2/ + 1 values). Thus the energy
levels will generally be composed of 2i + 1 hyperfine structure components.
Consequently, the quantum number specifying the magnitude of the spin
of an atomic nucleus can be determined by simply measuring the hyperfine
structure of the spectrum of the atom, analyzing the spectrum into an
energy level diagram, and then counting the number of hyperfine structure
components of the energy levels. It is found that the quantum number i is
integral (0, 1 2,
.) for nuclei of atoms in which the atomic weight A
is even, with i =
when the atomic number Z is also even i is half-integral
) for nuclei of atoms in which the atomic weight A is odd.f
(I, f f
The quantity K in equation (13-63) depends on the magnetic field at
the nucleus and on the nuclear magnetic moment. If the Hartree eigen.
functions describing the motion of the atomic electrons are available,
the magnetic field can be calculated to an accuracy of about 10 percent.
measured value of Aj. s can be used to
moment.
As stated above, it is found that for all nuclei the magnitude of the dipole
magnetic moment vector is of the order of one nuclear magneton (i.e.,
Then, within
this accuracy, the
evaluate the magnitude and sign of the nuclear magnetic dipole
1), but that the vector is parallel to the spin vector in some nuclei
|^|
and antiparallel in others (i.e., g is + or ).
Measurements of hyperfine structure also give information about the
electric quadrupole moment Q of the nucleus of the atom producing the
spectrum. This quantity specifies the extent to which the charge distribution of the nucleus departs from spherical symmetry. In all cases, the
charge distribution is that of an ellipsoid with an axis of symmetry along
the direction about which the nuclear spin angular momentum vector
t
precesses
Q<
for
for
it is
Q>
the ellipsoid
is
elongated in the direction of the axis,
flattened in the direction of the axis,
charge distribution has spherical symmetry.J
will
For
and
have an orientational potential energy in an atomic
=fc
for
0,
Q=
the
the nucleus
electric field
which
t We are refering to the integer A which is closest to the atomic weight of a particular
"isotope" of an atom. With this stipulation, the atomic weight is always very close to
an
%
integer.
A precise definition of Q
is
given in section
9,
Chapter
16.
452
MULTI-ELECTRON ATOMS
[Ch. 13
not spherically symmetrical. In a typical case, the magnitude of this
energy is comparable to the magnitude of the orientational potential energy
of the nuclear magnetic dipole moment in the atomic magnetic field.
is
However, the dependence of the energy on the quantum numbers specifying
the orientation of the nucleus in the
atom
is
quadrupole and magnetic dipole interactions, so
these
is
two
effects experimentally.
different for the electric
it is
present, the hyperfine splitting of the energy levels
from the predictions of equation
dipole interaction.
atomic
possible to separate
When the electric quadrupole interaction
(13-63),
which
is
observed to depart
treats only the
By analyzing these departures, and then
magnetic
calculating the
from Hartree eigenfunctions, it has been
and the approximate magnitude of the electric
quadrupole moment of a number of nuclei. It is found that Q =
for
all nuclei with
= or i = J. This can be understood from symmetry
considerations; for instance, when
= there is nothing to define an
axis of symmetry about which an electric quadrupole moment can be
measured. For nuclei with i > 1, Q is usually measurably different from
occurs most often. The nucleus of the 71 Lu atom has one of
0, and Q >
the largest known quadrupole moments. The charge distribution of this
nucleus is elongated in the direction of the axis of symmetry, and the ratio
of the semimajor axis of the ellipsoid to its semiminor axis is 1.20 0.05.
Techniques for measuring the magnetic dipole moments of nuclei from
an analysis of hyperfine structure are limited in their accuracy because
they depend upon calculating the strength of the atomic magnetic field.
In the years following 1933, Rabi and others developed several different
techniques which are extremely accurate. All of them involve measuring
the orientational potential energy of the dipole in an external magnetic
field of known strength. For atoms in which the total magnetic dipole
moment produced by the atomic electrons is zero (an atom in a 1 S state),
it is possible to measure the nuclear magnetic dipole moment by sending a
beam of atoms through an apparatus similar to that used by Stern and
Gerlach. In other cases the energy of a nuclear magnetic moment in an
external magnetic field can be investigated by using a field which is strong
enough, compared to the atomic magnetic field, to destroy the coupling of
I and J. In the most accurate of these techniques the magnetic moment is
measured, not in terms of the deflection of an atomic beam, but in terms of
the energy of the radiofrequency quanta which produce transitions between
two energy levels that are split by hyperfine splitting in an external magnetic field of known strength. These measurements, and the quantum
mechanics upon which they are based, form a very interesting subject which
we reluctantly pass over in order to continue the discussion of atomic structure and spectra.
electric field at the nucleus
possible to evaluate the sign
('
/'
TRANSITION RATES
Sec. 12]
12.
AND
SELECTION RULES
453
Transition Rates and Selection Rules
Consider an isolated atom in an excited state of energy E v The atom
make a spontaneous transition from that state to a state of
lower energy Ek emitting a quantum of electromagnetic radiation. We
would like to calculate a quantity called the transition rate for spontaneous
will eventually
which is the probability per unit time that the atom will make
such a transition. This calculation will enable us to understand the selection
rules governing transitions between any two energy states, as well as the
lifetimes of these states and the widths of the spectral lines emitted in such
emission,
transitions.
The theory
how
that has been developed so far in this
to treat spontaneous emission, but
it
does
related process of absorption. In this process,
magnetic radiation makes
state of higher energy E
book does not
tell
us
how
tell
us
to treat the
an atom acted on by
a transition from the state of energy
electro-
Ek
to the
absorbing a quatum. The electromagnetic
radiation acts as a small perturbation on the atom, and time dependent
perturbation theory can be used to calculate the probability per unit time
t
that this perturbation will produce such a transition.
per unit time
is
We
the transition rate for absorption.
this transition rate
and then
establish a relation
This probability
shall first calculate
between
and the
it
transi-
tion rate for spontaneous emission.
An atom
between
its
its
in a field of electromagnetic radiation experiences interactions
magnetic moments and the magnetic
electric charges
and the
field,
and
The magnetic
electric field.
also between
interaction
is
compared to the electric interaction, and we
shall consider here only the latter. Although the electric field is described
by a vector E, it is possible to treat separately the interaction produced by
each component of this vector. Each component, such as Ex is an oscillatory function of time and also of position. However, the wavelengths
which are characteristic of the absorption or emission optical spectrum of
an atom (A
10 3 A) are so large, compared to the dimensions of the atom
(r
1 A), that at any instant Ex is essentially constant over any region of
space occupied by the atom. Consequently, we can ignore the dependence
on position and describe Ex at a given atom by the function
usually completely negligible
Ex =
2Ex (v) cos
where the amplitude 2Ex\v) depends,
is
(13-64)
2-nvt
on
in general,
the frequency
v.
It
convenient to write this as|
Ex =
Ex(v)(e2 " in
- 2 " in
(13-64')
t The procedure for proving the equality of (13-64) and (13-64')
footnote on page 172.
is
indicated in the
454
MULTI-ELECTRON ATOMS
The
[Ch. 13
of the atom have a potential energy in the electric
Let us define the zero of potential energy and the origin of the x
axis to be at the center of the atom, and calculate the total potential
energy of the atom due to the x component of the electric field by summing
field
electric charges
Ex
the potential energy of each charge.
vx
This gives
= -Ex 2
-ex,
(13-65)
The summation
is
taken over each electron of charge
x but the charged nucleus does not contribute as
Evaluating Ex we have the perturbing potential
its
e and
coordinate
x coordinate
is 0.
vx
which
We
is
is
an
(e3 <
e-**)
^ ex,
(13-66)
oscillatory function of time.
on an atom, which at t =
by solving equation (9-32), which is
find the effect of this perturbation
in the
quantum
state
of energy
^
at
with the
initial
Ek
a
=-{l
n
n (t)e-
i(EMI *v Xmn
conditions for a n (t) at
a n (0)
n^k
0,
1,
The quantity
is the matrix element of
the states n and m. That is,
t?,.
v Xmn
where the integration
element dr.
The
(13-67)
the perturbation v x between
=jwtv x f n di
taken over
is
(13-68)
all
the coordinates of the
volume
quantities a n {t) are the coefficients in the expansion
^'{x,i)
= ^a n{t)W{x,t)
n
of the perturbed wave function in terms of the unperturbed wave functions,
and the combination a*(t) a n (t) is equal to the probability of finding the
atom in
the
quantum
state
of energy
En at time
t.
Now,
if
we
require that
the perturbation be small enough that
a n {t)
we may
for n
neglect
k.
all
< 1,
^ k,
1,
>
(13-69)
k,
terms on the right side of equation (13-67) except
This gives
da m(t)
~- a (/)e-B
t
dt
*- B
->"i;,
Sec. 12]
TRANSITION RATES
By
m=
setting
I,
we
AND
SELECTION RULES
455
obtain the differential equation for the coefficient
which we are particularly interested. In this equation we may make
the approximation that a k {t) = 1. Then it reads
(t),
in
daM ^
Upon
evaluating v x
*B&1 ~
daM ^ _
_l
-nE*-Ei)t/h
dt
and doing a
rearranging, this
little
L* ex jfk dT [ e--B* + w +
hJi
dt
becomes
*aBi--B*-*W]0( v
or
ry(E,-.Et + Av)t/
_j_
HE,-Ek -hv)tlH-\0v
d?
where
/*.
=
J
V*
Integrating the differential equation
conditions (13-68),
*('')
2 ex % dT
(13-70)
>
from
to
t',
with the
initial
we obtain
e
/"*
i(E,-Et +ht)t'Jll
- +
ft
hv
_ ai(Ei-E -hv)t'/h'
E,-Ek - hv
(13-71)
Let us consider the behavior of this equation. Since the perturbation is
small, the quantity px E%(v) will be small. As the absolute values of the
numerator of each of the two terms in the bracket are always in the range
to
2,
the entire right side
is
small except for frequencies for which the
denominator of one of the terms becomes small. Thus
a,(t') is
appreciable
when
only
,-^11^0
Since
E > Ek
t
this
can be
satisfied
only for the
radiation satisfies approximately the relation
precisely Einstein's relation
quantum absorbed
From
(13-71)
may
E Ek =
h
hv.
This
is
between the frequency and the energy of the
in the transition.
these arguments
be neglected.
we
see that the first term in the bracket of
Doing
this,
then multiplying the remaining
complex conjugate, and finally using the relation discussed
the footnote on page 172, we find
equation into
in
sign.
Thus the proba-
of finding the atom in the state of energy E t which is at {t') a (t'),
appreciable only when the frequency of the perturbing electromagnetic
bility
is
(13-72)
its
~(E t
a*(t) ai (t)
-E - hv)
k
2h
= 4n*u x
(,
-E - hvf
k
E x(vf
(13-73)
4 56
MULTI-ELECTRON ATOMS
[Ch. 13
where we have dropped the primes to simplify the notation. The reader
should compare this with equation (9-41), which expresses the result of
a perturbation that is "switched on" at / =
and remains constant
afterwards. Both equations have a t 2 dependence for small t. In equation
(9-41) this goes over to a
intuitively,
(13-73).
when
dependence, which
the equation
To correspond
is
is what would be expected
The same is true for equation
integrated.
with physical
integrated over v because an
atom
reality, this equation must be
never subjected to electromagnetic
is
radiation of perfectly sharply defined frequency, but instead to electro-
magnetic radiation which covers a band of frequencies. The integration
can be taken from + oo to oo, since the integrand is small except for a
restricted range of v. The calculation is the same as that preceding
equation (9-44), and the results are
1 *
P^-Ti^^M^ft
h
(13-74)
PXa is the probability that an atom initially in the state of
and subjected to the x component of the electric field of a
continuous spectrum of electromagnetic radiation, will be found in the
state of energy E v The quantity E%v lk) 2 is the square of the value of E%v)
The quantity
energy
Ek
for a frequency satisfying the relation
-E
hv lk
The contribution of the x component of the
rate for absorption
is
(13-75)
electric field to the transition
thus
dP
Rx *
^~ir~J aO-^w
1
(b-76)
Similar expressions are obtained for the contributions from the y and z
components. The complete transition rate for absorption
* -
J*
tej*. a
E>J + < ^M*i? +
k
Let us assume that the radiation
is
It is
^ E%v
lk
)\nl ikfi Xa
+ fi*
all
lkl Vik
equal,
K{v lkf]
p,
the amplitudes of
(13-78)
the time average of
volume of the electromagnetic
Taking the time average of both sides of equation
P
= -E
An
(13-77)
and
+ ^.J
convenient to express this result in terms of
the energy content per unit
lk
Then
unpolarized.
the three components of the electric field are
* ^
is
(2-4),
we
radiation.
obtain
(13-79)
where
AND
TRANSITION RATES
Sec. 12]
~E
is
457
SELECTION RULES
the time average of the electric field, and where
deleted the subscript for transverse.
components of the
electric field are
assumed
same is true
That is,
to be all equal, the
of the time averages of the squares of these components.
pi
p2
we have
Since the amplitudes of the three
p2
Then, since
H + E\ + El
2 =
we have
By squaring both
(13-80)
3E]
sides of equation (13-64),
obtain a relation between | and E%(v lk )
El
and averaging over
/,
we
2
:
(13-81)
2Ex (v lkf
Then, from equations (13-78) through (13-81), we obtain
- ft>n)[M2j*x a + <* Pv a + rfjtij
Itt
Ru
where we have
(13-82)
explicitly indicated that the
at the frequency v lk
energy density
is
to be evaluated
This equation gives the desired expression for the
transition rate for absorption.
We
have seen that there is an appreciable probability of a transition
from the state of energy Ek to the state of higher energy E lf with the
absorption of a quantum, only when equation (13-72) is satisfied with
the sign. However, the equation can also be satisfied with the + sign
for a transition from the state of energy E to the state of lower energy
Ek if a quantum is emitted. This phenomenon, in which the perturbation produced by an electromagnetic field induces a transition where
x
a quantum is emitted, is called induced emission.
for induced emission R kl can be evaluated by starting
and interchanging k and / everywhere except in the
and in the symbol v lk In this case the first term in the
.
(13-71)
is
the one which contributes;
The
at
transition rate
equation (13-69)
condition
E > Ek
l
bracket of equation
but otherwise everything
is
the
same and we immediately obtain
2tt
R ki-Jp #")l>i!> + ft*A, +
It is
easy to see from equation (13-70) that
ftj>.
(13-83)
458
MULTI-ELECTRON ATOMS
So, comparing (13-83)
and (13-82), we
Rm
[Ch. 13
see that
= R
(13-84)
The
transition rates for absorption and induced emission are equal. This
an example of the principle of microscopic reversibility, which is very
important in the theory of statistical mechanics.
is
The process of induced emission must not be confused with the process
of spontaneous emission. According to equation (13-83), the transition
rate for induced emission
is proportional to the energy density p~ of the
electromagnetic radiation applied to the atom. If no electromagnetic
radiation is applied, there will be no induced emission. However, there
be spontaneous emission, whether or not electromagnetic radiation is
applied to the atom, because the transition rate for this process does not
involve p. The transition rate for spontaneous emission Skl is an inherent
will
atom and
characteristic of the
which the atom
evaluation of
is
Skl
is
not influenced by the environment in
now to our original purpose the
placed. Let us return
This can be done in two quite different ways.
ment
is
The more rigorous
based on the theory of quantum electrodynamics.^
some years before
chanics was developed, which leads to the same results and is
the other treatment, given by Einstein
The
idea
treat-
We present here
quantum me-
much simpler.
to establish a relation between the transition rates for absorp-
is
induced emission, and spontaneous emission, from considerations of
thermodynamic equilibrium. Imagine a cavity containing a large number
of atoms which are emitting and absorbing electromagnetic radiation.
tion,
When
1.
this
system
is
in equilibrium at temperature T:
The electromagnetic
radiation must have the black
body
spectral
distribution characteristic of that temperature.
2.
The
total
number of atoms
in each energy state
must be distributed
according to the Boltzmann probability distribution for that temperature.
3. The total number of atoms in each energy
of the second condition, remain constant.
state must, as
a consequence
Condition 3 requires that the total number of transitions per unit time
the state of lower energy Ek to the state of higher energy E equal
from
the total
number of
number of
transitions in the opposite direction.
transitions per unit time
higher energy state
k is
the total
t Cf. footnote
from the lower energy
the total
state to the
is
*"*
where
Now
= Nk R
number of atoms
on page 462.
lh
in the lower energy state,
(13-85)
and where
AND
TRANSITION RATES
Sec. 12]
the transition rate for absorption
that one of these
atoms
will
According to (13-82),
lk
R lk
make a
SELECTION RULES
459
gives the probability per unit time
transition to the higher energy state.
can be written
R lk = B lk p(v lk)
(13-86)
This defines the proportionality constant B lk between the transition rate
for absorption and the energy density, which is called Einstein's coefficient
of absorption.^ Using
this quantity,
equation (13-85) becomes
^b = N B
k
The
total
number of transitions per
to the lower energy state
lk p{v lk)
(13-87)
from the higher energy
unit time
^m = #,[* + S
(13-88)
kl ]
where
is
the total
number of atoms
transition rate for induced emission,
From
spontaneous emission.
(13-83)
Einstein's coefficient
Skl
which defines
coefficients
Einstein's
B lk Bkl
,
and
Rkl is the
in the higher energy state,
and Skl
we can
Rkl = Bkl p(v
which defines
state
is
is
the transition rate for
write
(13-89)
lk )
of induced emission
=A
Bkl We
.
also write
(13-90)
kl
The
of spontaneous emission A kl
depend only on the inherent characteristics
coefficient
A kl
all
of the atoms, and equation (13-90) emphasizes the fact that we assume
the transition rate for spontaneous emission to be independent of the
intensity of the electromagnetic radiation applied to the atoms. From
the last
two equations, we may write equation
(1
^m = WmpCv*) + A
Then we
satisfy condition 3
3-88) as
by equating Ykl and jV
N* _
N
t
B kl p(vik)
+A
(13-91)
kl ]
11c ,
and obtain
kl
(13-92)
B lk p(v lk )
The quantity Nk jN can be evaluated by applying condition 2. The
Boltzmann distribution (2-23) says that at temperature T the probability
that an atom will have energy Ek is proportional to e - E* IK r and the
probability that an atom will have energy E is proportional to e - E IKT
l
'
'
At
the time he developed this treatment, Einstein wrote equation (13-86) as a
know equation (13-82).
postulate because he did not
MULTI-ELECTRON ATOMS
460
K is Boltzmann's constant.
where
of atoms with these two energies
According
to
e-
(E,
(13-93)
'
equation (13-75), this
this
-Ek )IKT
E /KT
Hk
Combining
Consequently, the ratio of the number
is
-E k /KT
:t
ly k
[Ch. 13
is
hv lk /KT
(13-94)
with equation (13-92) gives
fJivit/KT
Bklf>( v lk)
+ Am
Bi k p{v lk)
Solving for
p~(v lk),
we obtain
pXv lk )
Finally,
we apply condition
tral distribution of
f^
(13-95)
by equating
p~(v lk)
to the black
body
spec-
energy density (2-31). In terms of the present notation
this is
Then a comparison of equations (13-95) and (13-96) shows that
Einstein coefficients must be related to each other as follows
B a = Bkl
and
An =
the three
(13-97)
^B
(13-98)
kl
Equation (13-97) shows that this treatment is consistent with equation
(13-84). Equation (13-98) provides the desired connection between the
transition rates for absorption and spontaneous emission. Using (13-83),
(13-89), and (13-90), we obtain at last the transition rate for spontaneous
emission
Su
The
~ Ifo?
^^ + ^A
zJ
(13 " 99)
is seen to be proportional
cube of the energy) of the emitted quanta.
with the transition rates for induced emission and absorption,
transition rate for spontaneous emission
to the cube of the frequency (or
In
common
it is
also proportional to the
fi x
and
their
sum
of the products of quantities such as
complex conjugates. The quantity
[x
Xui
is
called the matrix
AND
TRANSITION RATES
Sec. 12]
SELECTION RULES
461
element of the x component of the electric dipole moment of the atom.
The quantities fi and /e are given corresponding names. This nomenclature
is
used because the term
2 ex
that occurs in the integrand of
is the x component of the electric dipole
fi x
The matrix element of J exi * s its average value
equation (13-70) defining
moment
of the atom.
calculated with a weighting factor ip*ip t
the probability density y)*y> l of the initial
bility density y>*y k
of the
final
quantum
This weighting factor
quantum
state,
state,
is
neither
nor the proba-
but instead some sort of
"mixed probability density" which involves the eigenfunctions of both the
initial and the final states. The probability of an atom radiating a quantum
depends, then, not only on what the atom does in the initial quantum state
before it radiates, but also on what it is going to do in the final quantum
state after it radiates. The reason is that when the atom radiates it is
actually not in either the initial state or the final state, but in some linear
combination of the two.
This point of view can be used to give a physical picture of what happens
in a radiation process. Consider an induced transition of an atom from
the initial state of energy E to the final state of lower energy Ek accompanied by the emission of a quantum. According to (7-45), the wave
function describing the atom can always be written
,
Y = 2n a n e~
iEnt/n
Before the application of the electromagnetic
atom
non-zero, and
transition, the
is
is
in the eigenstate of energy
Y=
The corresponding
a e-
field
which induces the
Only the coefficient
iB ' t "'y>
l
probability density
Y*Y =
(13-100)
Wn
is
ata lV Vl
This quantity specifies the charge distribution of the atom.
(13-101)
It is
constant
Consequently the electric dipole moment (as well as every other
feature of the charge distribution) is constant in time. According to
classical physics, the atom can radiate only if its electric dipole moment
in time.
some other feature of the charge distribution) changes in time. Thus
we can understand, in a sense even classically, why the atom cannot
radiate as long as it is in an eigenstate. Now, when the electromagnetic
(or
produces a perturbation which mixes contributions from
wave function for the eigenstate of energy Ek into the wave function
the atom. The coefficient ak increases with time from its initial value
field is applied, it
the
for
MULTI-ELECTRON ATOMS
462
of zero, and the coefficient a decreases.
Consider the wave function for
It has the form
the
atom when both ak and a
The corresponding
W*W =
a*a
The
last
probability density
^p*rp l
are non-zero.
a e -uw fy
Y=
+
e-
iE " tlh
xp
(13-102)
is
*
fl
aiC
-(,-^),/
Etc
(13-104)
reduced to zero. Then the probability density
Y*F =
(13 _ 103)
Vi
with the frequency
oscillate
Ei
is
aic
a?a k y>tfk
two terms of F* F
Eventually a l
,V*rwy ft +
,
[Ch. 13
is
a*
k a k y>: Wk
With such a probability density the atom cannot
But it is apparent
by the wave function
(13-102). In this state the probability density (13-103), and consequently
the charge distribution, have components which oscillate with frequency
that
it
can radiate when
(13-104) that
is
it is still
radiate.
in the state described
exactly equal to the frequency of the
magnetic radiation which
is
quantum of electro-
emitted during the transition.
These arguments provide a good physical picture of the process of
induced emission and absorption. It would appear that they do not apply
to the process of spontaneous emission because in this process there is
apparently no perturbing electromagnetic field to produce the mixing of
wave functions that leads to an oscillating probability density. Howquantum electrodynamics removes this difficulty. In
this theory both the electromagnetic field and the atom are quantized.f
Then it is found that there are interactions between these two quantized
systems, even when the electromagnetic field is in the state of lowest
energy, because some electromagnetic radiation must be present in this
the
ever, the theory of
has a zero point energy like any other
quantized system executing simple harmonic oscillations.
state as the electromagnetic field
The electromagnetic
field
was quantized
in the very simple theory of
Planck and
comprised of quanta of energy E = hv. But a more
complete theory is needed because it is necessary to know how an electromagnetic field
in the quantum limit (containing few quanta) interacts with an external system of
Einstein,
which
charges, just as
states that
it is
it is
necessary to
know how an
electromagnetic
field in
the classical
quanta) interacts with an external system. The classical limit
is described by Maxwell's theory; the quantum limit is described by the theory of
quantum electrodynamics. In the latter theory, Schroedinger's quantum mechanics is
applied to a system of quanta, but the quantization of this system is considerably more
limit (containing
many
involved than the quantization of a system of particles since the number of quanta
in a system can change when it interacts with an external system.
AND
TRANSITION RATES
Sec. 12]
If the eigenfunctions for
elements between
all
SELECTION RULES
an atom are known, the
electric dipole
pairs of excited states can be obtained
Then
integrals such as (13-70).
463
matrix
by evaluating
the transition rates for spontaneous
emission between these states can be calculated from equation (13-99).
These transition rates immediately lead to theoretical predictions of the
relative intensities of the spectral lines emitted in transitions
between the
found that the agreement between these predictions
and the experimental data is quite good. The transition rate for spontaneous emission varies widely from one case to the next. For the transition
of the X H atom from its first excited state to its ground state, this transition
rate has the value Skl
10 8 sec -1 This is typical of the orders of magnitude encountered, except that the transition rates between certain pairs of
states are essentially zero. These are the transitions for which the spectral
lines are observed to be absent or extremely weak. The transition rates
are predicted to be zero in these cases because the integrals involved in
various states.
It is
the electric dipole matrix elements vanish identically.
Thus
the selection
rules are a set
of conditions on the quantum numbers of the eigenfunctions
of the initial andfinal states such that the electric dipole matrix elements are
zero when calculated with a pair of eigenfunctions whose quantum numbers
violate these conditions.
As a
we shall carry out part of the calculation of the
matrix elements for an atom with one optically active
simple example,
electric dipole
electron.
In this case the core of
and
filled subshells
does not change in the
necessary only to evaluate the matrix elements of e
times the three rectangular coordinates of the optically active electron.
transition,
it is
Furthermore, in evaluating these matrix elements
we need
only to write
the eigenfunctions for the optically active electron. All the other factors in
atom immediately integrate to unity
because of the normalization conditions. The rectangular coordinates are
the complete eigenfunction of the
related to the spherical coordinates, in terms of
which the eigenfunctions
are written, by the equations
r sin 6 cos
y = r sin 6 sin
z = r cos 6
x
<f>
(13-105)
<f>
Using these coordinates, we have
-"*'',;,.,
) J J
Vti-mf
sin 6 cos
nlm
f sin 6 dr dd
d<j>
e f
Jo
R:. v R nl r 3 dr f*^-,,,, sin 2 6 dd
Jo
P $: <D mi cos
Jo
<f>
d<f>
MULTI-ELECTRON ATOMS
464
[Ch. 13
Consider the third integral,
2
r^mi
-..
$d
cos
<f>
Jo
f 'e-".'V"*cos
d<f>
<>0
C 2n
-i(m',-mi + l)*
_|_
-i(m,-ro! -1)0-1
JJ
2 Jo
We leave it as an exercise for the reader to
show
that this integral vanishes
unless
m
If
it
does vanish, so does
evaluating
upon
o
X
u,.
'
y n j'm,,nlm(
-m
'
For *n
'
Since the
^^./^
-T
is satisfied.
Thus
m[
obtained
we have
f2ir
'- Sin0COS0de
*-.i
'0
^m,^
-m =
(13-107)
is
the complete selection rule for
-m =
l
Am,
0,
transitions that satisfy equation (13-107) are
direction.
The
spectral lines
from
The
is
(13-108)
in agreement with the second of equations (13-57),
for one or several optically active electrons.
is
transition rate 5. rm;w!mi will be non-zero only if either (13-106)
or (13-107)
This
Imj^Imj
selection rule
are orthogonal, the third integral vanishes unless
mj
The
(13-106)
fir
foo
/W..= Jo
= 1
The same
>
which
is
spectral lines
valid
from
linearly polarized in the
transitions that satisfy equation
(13-106) have components, which are linearly polarized in both the x
and y directions, that combine to form circularly polarized radiation.
These results are verified by measurements of the polarization of the
radiation emitted in the Paschen-Bach effect by the different spectral lines
corresponding to different values of A/j.
The selection rules for / can be obtained by a straightforward evaluation
of the integrals over dd. However, this is quite tedious, and part of the
can be obtained in a simpler way from considerations of the
In section 4, Chapter 8, we defined the
parity of a one dimensional eigenfunction. This is the quantity which
describes the behavior of the eigenfunction when the sign of the coordinate
is changed.
The definition can be immediately extended to three
dimensions. That is, eigenfunctions which satisfy the equation
results
parities of the eigenfunctions.
f{-x, -y,
z) =
+ip(x, y, z)
(13-109)
AND
TRANSITION RATES
Sec. 12]
SELECTION RULES
465
are said to be of even parity, and eigenfunctions which satisfy the equation
y(-x, -y, -2)
are said to be of odd parity.
not
It is
-ip(x, y, z)
difficult to
(13-110)
show
that
most
eigen-
functions which are solutions to Schroedinger equations with potentials
whose values are unchanged when the
signs of the coordinates are changed
even or odd).f
It is convenient to express equations (13-109) and (13-110) in terms
of spherical coordinates. By considering the diagrams in figure (13-16)
we see that, when the signs of the rectangular coordinates are changed,
the behavior of the spherical coordinates is
will
have a
definite parity (either
r~>r,
O-^tt-9,
<f>~+TT
(13-111)
<f>
t Write the time independent Schroedinger equation as
^op'*. y,
Change the
z )v(x , V, 2 )
Exp(x, y, z)
signs of the coordinates:
^opC
-x
ll>
z)y( x
y,
~ z) =
Ey>(x, y,
z)
Since e op contains only V(x, y, z) plus terms like d z /dx 2 ,e ,(x, y,
z) = e op (x, y, z)
0T
z) V{x, y, z). If this relation holds, the previous equation reads
when V(x, y,
^opfo
Since both
y(x, y,
y,
z)v( *, y,
z) and
- z) =
y)(x, y, z) satisfy
Ey>(x, y,
the
same
differential equation,
it
must
be that
z) =
y(x, y,
where
AT is
a constant.
Now
Ky>(x, y, z)
change the signs of the coordinates
in this
equation and
obtain
xp(x, y, z)
Kyi(x, y,
Multiplying this equation into the previous equation,
K"
z)
we
obtain,
or
K=
so
V( *. y, -2)
This proof breaks
down
if
there are
w(x,y,z);
Q.E.D.
two or more linearly independent eigenfunctions
same value of E, i.e., if there is degeneracy.
that satisfy the differential equation for the
But even in this case the eigenfunctions will have definite parities if they are approchosen linear combinations. The y>mm are examples of degenerate eigenfunctions with definite parities, as shown by equation (13-112). An example of a
degenerate eigenfunction without a definite parity is the traveling wave (15-64).
priately
MULTI-ELECTRON ATOMS
466
Now,
if
[Ch. 13
the reader will inspect the eigenfunctions for the hydrogen
(10-39), he can verify that for those
and
which are
atom
listed
<M* +
4>)
(-i) |mil <i> mi (<)
^im^ -
e)
(-i) l+M lmi (d)
Thus
Rm(r)
.,(*
~ ^mfr +
ft
VW."" ~
>"
(-l) l+lmi K-l)
MR
nl (r)
@ lmi (0) $(<)
or
This
true for all the hydrogen
is
transformation (13-1
1)
= (-V V>m mi(r,
l
ft
atom
6,
(13-112)
In fact, since the
eigenfunctions.
does not affect the coordinate
0)
r,
equation (13-112)
(*,y,z)
(-x, -y, -z)
Figure 13-16.
is
Illustrating the parity operation.
true for all eigenfunctions for spherically symmetrical potentials because
R nl (r) Q tmi (0)
^>
m i(ft where
Thus the
atom.
(c) are the same as for the hydrogen
{m (6) and O mi
determined
potentials
is
parity of all eigenfunctions for spherically symmetric
all
such
eigenfunctions
by ( 1)'; the parity
Now
dipole
can
even if I
is
be
is
written
even,
and odd
if
I is
odd.
consider the matrix element of the x component of the
moment, which
electric
is
^ m
The x component of the
;, te;
V-'i'mj
electric dipole
2 ex, Wnlmi dr
moment
is
of odd parity since
I(-^)= -lex,
y and z components. If the eigenfunctions
and ip nlm are of the same parity (both even or both odd), the
This
is
also true of the
y>
rm
entire
TRANSITION RATES
Sec. 12]
AND
SELECTION RULES
467
uy
integrands
of m_
and u,*n' V m i,nlmi will be of odd xparity.
J
''n'l'm^nlmi
n'l'mi,nlmi
,
'
'
'
it is obvious that the integrals will vanish identically
because the contribution from the point (r, 6, <f>) will be canceled by
If this is the case,
the contribution from the point
rates will
(r,
be zero. Therefore there
relative parity of the initial
and
is
d,
</>),
and the
transition
the following selection rule for the
final eigenfunctions in
an
electric dipole
transition
The
Equation (13-112) allows
the
quantum number
/,
parity
this to
must change.
(13-113)
be expressed, in terms of the change in
as
A/=
1, 3, 5,...
(13-113')
but not all, of the selection rule (13-31). The rest of it can be
by considering (13-108), which is &m = 0, 1. This selection
rule makes it possible to understand why transitions are not allowed when
A/ = 3, 5,
since in some of these transitions Am, > 1.
As
mentioned before, the selection rule (13-31) can also be obtained by a
This
is
part,
justified
straightforward evaluation of the matrix elements.
To calculate the selection rules for the quantum number/, it is necessary
know the exact form of the eigenfunctions in the presence of the spinorbit interaction. Therefore we can only assure the reader that calculations
to
with these eigenfunctions do verify the selection rule (13-30). The same
true of the selection rules (13-45) for the quantum numbers of an atom
with several optically active electrons that is an example of LS coupling.
is
But one of these, As' = 0, has such a simple interpretation that we must
mention it. The selection rule states that the coupling of the spin angular
momentum vectors cannot change in the transition. The reason is
simply that the interaction between electromagnetic radiation and spins
takes place through the interaction of the magnetic field and the magnetic
moments. These interactions are so very weak that the coupling of the
spins is normally not affected.
However, faint spectral lines that violate the selection rule for s' can
be seen in certain special circumstances (cf. the discussion at end of
Thus
this selection rule does not absolutely prohibit transitions
but only makes these transitions very unlikely. If the
transition cannot take place by the normal means of an interaction in-
section
8).
that violate
it,
volving the electric dipole
an
moments and
the electric fields, which
is
called
electric dipole interaction, there is
a very small probability that it can
take place through an interaction of the magnetic dipole moments and the
magnetic fields that are also present in electromagnetic radiation. In
fact,
there
is
a very small probability of violating, by means of such
468
MULTI-ELECTRON ATOMS
[Ch. 13
magnetic dipole interactions, any of the selection rules quoted in this
chapter. This can also be done by means of electric quadrupole interactions,
which represent violations of our assumption that the wavelength
is so long compared to the dimensions of the atom that
of the radiation
any variations of the
electric fields over the atom can be neglected. Nevergoing through electric dipole interactions are by far
the most important for atoms.
theless, transitions
Lifetimes and Line Widths
13.
Consider a system of atoms in which N(0) have been excited at time
by some process to a state of energy E Let us calculate the number
N(t) remaining in that state at some later time t, assuming that the transition rates for spontaneous emission Skl are known for all k for which
t
Ejc
<E
t,
and that these
much larger than the
because the average density p~ of
transition rates are very
Rkl
transition rates for induced emission
electromagnetic radiation
is
very small.
Then
the probability per second
atoms in the state of energy E will make a transition out of
that state into any of the states of lower energy E is the sum of the S
k
kl for
all k for which E < E
Call
this
probability
X;
then
k
that one of the
= 2s
In a time interval dt at
the total
t,
(13-H4)
number of such
transitions will be,
on
the average,
N(i)X dt
since during that interval there are approximately N(t)
and each one has a transition probability 1
the decrease in N(t) during that interval. Thus
state
dN(t)
-N(t)X
dt.
atoms
in the initial
This quantity
is
just
dt
or
-^=-Xdt
(13-115)
jv(o
Integrating
from
to
/',
we have
rd
Jo
-x!"dt
N{t)
Jo
which gives
In \_N(t')
iV(0)]
-At'
Sec. 13]
LIFETIMES
AND
LINE
WIDTHS
469
or
N(0)
So
N(t')
e~ u
'
N{0)
or
N(t)
N(0) e
-U
(13-116)
where we have dropped the primes to simplify the notation. The number
of atoms remaining in the initial state decreases exponentially from the
original value JV(0) and, at the time t = 1/A, the number remaining is
reduced by a factor e~ x The lifetime t of this state is defined as the average
time an atom spends in the state before making a transition. This is,
.
/*00
00
tN(i)dt
J*
_o
N(0)te'
xt
dt
Jo
N(0)e- Xi dt
N(t)dt
Jo
Jo
Since this ratio of integrals has exactly the same form as (2-19), we may
use the results of the calculation following that equation to write im-
mediately
t=1/A
which
(13-117)
is
IS J
(13-118)
- k
Because r generally involves a sum over several Skl the variations in
those quantities tend to average out and yield lifetimes which are of the
10" 9 sec for many states. However, there are a large number
order of t
,
of exceptions.
There is a very interesting relation between the lifetime of an excited
of an atom and the width of the spectral lines arising from transitions
involving that state. Spectral lines are not perfectly sharp; instead they
have finite widths that can be measured, under favorable circumstances,
by using equipment of the highest resolution. Usually the width is due
state
primarily to a
random Doppler shift of the frequencies of the emitted
random thermal motion of the atoms in the
quanta, resulting from the
Doppler broadening can be reduced by
using low temperature sources, and when this is done it is found that for
lines originating from states of short lifetime a measurable width remains.
This is a manifestation of the uncertainty principle relation (6-28), which
spectral source.
However,
this
MULTI-ELECTRON ATOMS
470
says that a measurement of the energy of a system that
time At must be uncertain by an amount AE, where
[Ch. 13
carried out in a
is
~h
AE At
Consider a transition from some excited state to the ground state of the
atom. A measurement of the frequency of the quantum emitted in this
transition constitutes a measurement of the energy of the atom in the
excited state. Furthermore, the measurement obviously must be carried
out within a time comparable to t, the lifetime of the excited state. Thus
the uncertainty principle
demands
that the energy of the
quantum can
have any value within the range
AE~hJT
Then
its
(13-119)
frequency can have any value within the range
Av
^^l = Jhr
and the associated
(13-119')
2-nr
spectral line will have a corresponding width.
natural broadening can be observed experimentally
if
This
the pressure in the
At high pressures the time between collisions of
is low.
atom with other atoms becomes smaller than the lifetime of
the excited state and, since it is no longer possible to carry out an undisturbed measurement for a time comparable to t, the width of the line will
spectral source
the emitting
be larger than
its
natural width.
This
is
called collision broadening.
If a spectral line is emitted in a transition
final state
of very
much
width of the spectral
longer lifetime
(e.g.,
line emitted in the transition is
lifetime of the initial state.
The
initial state to
due
entirely to the
In these circumstances the natural width of
the spectral line, expressed in energy units,
of the
from some
the ground state), the natural
is
said to be equal to the width
idea that a state has a non-zero width
if it has a
comes into quantum theory directly through the uncertainty
principle, as we have seen, and certainly is not in conflict with the statement
made in Chapter 7 that the energy of a system of bound particles, such as
an atom, can be equal only to one of a particular set of discrete (i.e., zero
initial state.
finite lifetime
width) energies.
We
could
possibility of transitions
mation,
all states
have
make
between
this
statement because
states in
infinite lifetimes or
we ignored
the
Chapter 7 and, in that approxizero widths.
The idea of a non-zero width for a finite lifetime even comes into classical
theory. Consider any resonant system, such as an electrical circuit containing inductance and capacitance. If it contains absolutely no resistance,
the theory of electrical circuits predicts that the circuit has a perfectly
sharp resonant frequency v and that,
frequency,
it
will accept
if driven by some source of variable
power only when the source frequency is exactly
LIFETIMES
Sec. 13]
equal to v
AND
But, of course,
LINE
WIDTHS
all circuits
471
have some resistance. In
the theory predicts that the resonant frequency
that the
this case,
not so well defined, and
is
power P accepted from the source depends on the source frequency
v according to the relation
^~
P(v) c
(v
The quantity
P(v),
-vf+
(13-120)
(Av/2) 2
which is plotted in figure (13-17), is the familiar
full width at half maximum Av is a simple function of
resonance curve. The
Figure 13-17.
resonance curve.
the resistance that increases with increasing resistance.
accepts a certain
turned
off,
amount of energy from
the energy content
Now,
if the circuit
and the source
is
then
of the circuit will decrease in time by
The theory of
dissipation in the resistance.
this decrease follows the
the source
electrical circuits
shows that
law
E(t)
= E(0)e-
tlT
(13-121)
where
T
(13-121')
2-rr
Of
course,
all
Av
these equations for electrical circuits have been verified
experimentally.
Now
consider an
atom
in
its
electromagnetic radiation which
ground
is
state,
exposed to a source of
emitting quanta of
The atom can absorb a quantum only if its energy is equal
one of
its
frequencies.
excited states but, since these states have non-zero widths,
quantum
possible for a
if its
all
to the energy of
energy
is
it is
to be absorbed in a transition to a particular state
within a range equal to the width of the state.
In fact,
quantum theory shows
that the transition rate
particular state depends
on the energy hv of the quantum according
for absorption to a
to
the relation
R(hv) oc
'
(hv-hv f + (hAvl2f
(13-122)'
K
MULTI-ELECTRON ATOMS
472
[Ch. 13
is the mean energy of the state and h Av is its full width at half
maximum. Furthermore, the theory shows that the width h Av is related
where hv
to the lifetime r of the state
by the equation
r
The width h Av of a
(13-122)
oc
is
the energy of the state
Eo f
and V
r
is
by the symbol T, and
state is usually designated
(e
(13-123)'
usually written in terms of the energy
is
R(E)
where
2nh Av
(r/2)
is its
of the quantum as
(13-124);
l
width.
Then
(13-125)
the relation between the lifetime t of a state
the transition rate for spontaneous emission
is
and
its
width V.
Since
proportional to the transition
former also has the functional dependence of
Therefore the natural width of a spectral line emitted in a
rate for absorption, the
(13-124).
transition
width
is
Note
from an
initial state
of width
to a final state of much
narrower
r, as has already been mentioned.
that equation (13-125)
is
in agreement with the limit set
uncertainty principle relation (6-28).
Also note that the
by the
results (13-122)
and (13-123) of the quantum theory for resonant absorption of quanta
by a state, which subsequently radiates, are essentially identical with the
results (13-120) and (13-121') of the classical theory for the resonant
absorption of energy by a system, which subsequently dissipates. The
natural spectral line shape predicted by (13-124) has been verified experimentally. Equation (13-125) cannot be tested by experiments measuring
both t and T because h is so very small that only one or the other is
measurable. Nevertheless, it is assumed to be valid and is often used to
predict the value of V from the measured value of r, or vice versa.
BIBLIOGRAPHY
Herzberg, G., Atomic Spectra and Atomic Structure, Dover Publications,
New
York, 1944.
Leighton, R. B., Principles of Modern Physics, McGraw-Hill Book Co., New
York, 1959.
Richtmyer, F. K., E. H. Kennard, and T. Lauritsen, Introduction to Modern
Physics, McGraw-Hill Book Co., New York, 1955.
White, H. E., An Introduction to Atomic Spectra, McGraw-Hill Book Co.,
New York, 1934.
Ch.
EXERCISES
13]
473
EXERCISES
Evaluate V(r) and p(r) in the Thomas-Fermi theory for the
and compare with
Use the approximation (13-22), the Z
atom equation (10-44), to estimate the
of the A atom. Compare with figure (13-3).
2.
electron
3.
A atom (Z =
18),
figure (13-3).
for the
atom, and the one-
radii of the n
1, 2,
and
3 shells
Verify that the vector diagrams of figure (13-1 1) are equivalent to equations
Also construct a similar set of vector diagrams for / x = 2, s 1 = |;
(13-40).
'2
==
A.
-^
^2
==
Z'
Carry through a calculation as in table (13-8) for the configuration np z
and thereby
5.
verify the results for that configuration listed in table (13-9).
is an example of LS coupling, and an nd n'd
atom in which both the ordering of the energy levels
and the relative strengths of the s' and /' dependence of the
Consider an atom which
configuration of that
according to
s', l',j',
Draw a schematic energy level diagram for this configuration,
using the same (exaggerated) scale for the fine structure splitting of all the levels
energy, are normal.
within a given multiplet. Label each level with the spectroscopic notation.
6.
Indicate which levels of the energy level diagram of exercise 5
absent
7.
if n'
On
the energy level diagram of exercise
exaggerated) scale the
Zeeman
effect splitting
6,
Count the
quantum states
total
number of components,
draw
to the
same
of each level into
components under the influence of a weak magnetic
8.
would be
n.
i.e.,
(highly
its
various
number of
different
field.
the total
in the configuration, obtained in exercise 7.
Show
that this
equals the degeneracy of the configuration before the application of the perturbawhich is the product of degeneracy factors 2(2/ + 1) for each of the two
tion,
optically active electrons of the configuration.
9. In an atom which is an example of LS coupling, the relative separations
between adjacent levels of a particular multiplet are 4:3:2:1. Use the procedure
outlined at the end of section 10 to assign the quantum numbers s', l',j' for
these levels. Repeat for multiplets with relative separations 7:5:3 and 9:7:5:3.
10.
Verify equation (13-74).
11.
Verify equation (13-112).
By a straightforward evaluation of the electric dipole matrix elements for
the eigenfunctions (10-39), show that the selection rule (13-31) is valid for
n = 2 -* n = 1 transitions in the
atom.
12.
Evaluate a few of the electric dipole transition matrix elements for a one
dimensional charged simple harmonic oscillator by using the eigenfunctions
13.
Show that their values agree with the selection rule An
pare with the discussion of section 8, Chapter 5.
(8-118).
1. Com-
14. Consider a gas of H molecules at room temperature and pressure, in
2
which the H atoms are making spontaneous transitions from the first excited
MULTI-ELECTRON ATOMS
474
state to the
is
10
sec
ground
-1
Use the
state.
Use the
[Ch. 13
fact that the transition rate for that transition
to estimate the natural broadening of the resulting spectral line.
relation of elementary kinetic theory connecting the average distance
between collisions to the density of the gas and the projected geometrical cross
sectional area of the molecules, the perfect gas law, and the law of equipartition
of energy to estimate the collision broadening. Use the law of equipartition
of energy and the classical Doppler shift equation to estimate the Doppler
broadening.
CHAPTER
14
X-rays
I.
The Discovery
of X-rays
Except for a brief description of the Compton effect, and a few other
we have postponed the discussion of X-rays until the present
chapter because it is particularly convenient to treat X-ray spectra after
remarks,
Although this ordering may have given the reader
a distorted impression of the historical importance of X-rays, this impression will be corrected shortly as we describe the crucial role played by
treating optical spectra.
X-rays in the development of modern physics.
X-rays were discovered in 1895 by Roentgen while studying the phenomena of gaseous discharge. Using a cathode ray tube (cf. Chapter 3)
with a high voltage of several tens of kilovolts, he noticed that salts of
barium would fluoresce when brought near the tube, although nothing
visible was emitted by the tube. This effect persisted when the tube was
wrapped with a layer of black cardboard. Roentgen soon established
that the agency responsible for the fluorescence originated at the point at
which the stream of energetic electrons struck the glass wall of the tube.
Because of its unknown nature, he gave this agency the name X-rays.
He found that X-rays could manifest themselves by darkening wrapped
photographic plates, discharging charged electroscopes, as well as by
causing fluorescence in a
number of
different substances.
He
also
found
low
that X-rays can penetrate considerable thicknesses of materials of
atomic number, whereas substances of high atomic number are relatively
opaque.f Roentgen took the first steps in identifying the nature of X-rays
t This property
tissue,
is the basis of the technique of X-ray photography of bones in living
a technique which was used by surgeons only a few months after Roentgen's
discovery.
475
X-RAYS
476
by using a system of
slits
to
show that
(1) they travel in straight lines,
that (2) they are uncharged, because they are not deflected
magnetic
[Ch. 14
by
and
electric
or
fields.
The discovery of X-rays aroused the
performed a
and many
Haga and Wind
interest of all physicists,
joined in the investigation of their properties.
In 1899
diffraction experiment with X-rays
which showed
from the size of the
diffraction pattern, their wavelength could be estimated to be X -~ 10~ 8 cm.
In 1906 Barkla proved that (4) the waves are transverse by showing that
they can be polarized by scattering from many materials.
There is, of course, no longer anything unknown about the nature of
X-rays. They are electromagnetic radiation of exactly the same nature as
visible light, except that their wavelength is several orders of magnitude
shorter. This conclusion follows from comparing properties 1 through 4
with the similar properties of visible light, but it was actually postulated
by Thomson several years before all these properties were known. Thomsingle
slit
that (3) X-rays are a wave motion phenomenon, and,
son argued that X-rays are electromagnetic radiation because such radiation would be expected to be emitted from the point at which the electrons
strike the wall of a
cathode ray tube.
At
this point, the electrons suffer
very violent accelerations in coming to a stop and, according to classical
electromagnetic theory,
magnetic radiations.
tion of X-rays
In
common
is
all
accelerated charged particles emit electro-
We shall see later that this explanation of the produc-
at least partially correct.
with
other
electromagnetic radiations,
X-rays exhibit
The reader will recall
that the Compton effect (cf. Chapter 3), which is one of the most convincing
demonstrations of the existence of quanta, was originally observed with
particle-like aspects as well as wave-like aspects.
electromagnetic radiation in the X-ray region of wavelengths.
2.
Measurement
of X-ray Spectra
The problem of measuring the wavelength spectrum of the radiation
emitted from an X-ray tube seemed, at
first, a very difficult one indeed.
Roentgen had shown that the index of refraction of all substances for
X-rays is very nearly unity. Consequently, a prism spectrometer is not
a useful instrument for X-rays.
An
alternative
is
to use a diffraction
However, the wavelength of X-rays is so small compared to the
line spacing of the best ruled gratings which can be produced that these
devices would also appear to be useless. The problem was solved in 1912
when Laue suggested that a crystal might be used as a diffraction grating
for X-rays because the atoms of a crystal are arranged in a regular lattice,
grating.
MEASUREMENT OF X-RAY SPECTRA
Sec. 2]
with a separation of the lattice planes that
wavelengths of X-rays.
is
This suggestion was
477
quite comparable to the
put into practice by
and Knipping, and the technique was perfected by the Braggs.
Consider a crystal of NaCl. The macroscopic characteristics of such
a crystal strongly suggest that on a microscopic scale the atoms are
arranged in a regular cubic lattice. A two dimensional section of this
lattice is indicated in figure (14-1). The figure also attempts to indicate
first
Friedrich
the density distribution of the electrons within each atom.
study of multi-electron atoms,
we know
that there
is
From our
a region of very high
##*#
## + #
##**#
A two
Figure 14-1.
dimensional section of a sodium chloride crystal.
electron density quite near the center of the atom,
in this region are very tightly
bound
and
that the electrons
Now
consider a beam
of X-rays incident upon the crystal. The beam penetrates into the interior
of the crystal; a fraction of it goes all the way through; but another
fraction is scattered by the atoms of the crystal. There are several mechanisms which produce this scattering. All of these mechanisms will be
discussed later, but at present
the X-rays are scattered with
we
to the atom.
are interested only in the one in which
no change
in wavelength and no change in
phase from the tightly bound electrons near the center of the atom. Each
atom scatters X-rays by this mechanism into all angles, although the
scattered intensity
is largest in the direction of the incident beam.
In the region exterior to any atom, the intensity pattern of the X-rays
which
center.
it
scatters
For
is
the
same as
this reason,
it is
if all
the scattering occurs exactly at
its
possible to analyze the scattering by the
model crystal consisting of a set of point scattering
same relative locations as the centers of the atoms of the
crystal in terms of a
centers with the
real crystal.
spaced
These scattering centers consequently lie on a
lattice planes.
set
of uniformly
X-RAYS
478
[Ch. 14
Imagine that a beam of monochromatic X-rays is incident on the model
an angle 6 measured from its surface. Part of the beam is
scattered from the scattering centers of each lattice plane. Let us consider
one such lattice plane. In the language of wave motion, the incident beam
of X-rays can be described as a succession of wave fronts (surfaces of
crystal at
constant phase) moving over the scattering centers of the
lattice plane.
Each time a wave front crosses a scattering center the scattering center
re-emits a spherical wave, with no change in phase, that moves out in all
directions. Figure (14-2) shows several incident wave fronts and one set
Figure 14-2.
beam from
Huygens' wavelet construction for the
partial reflection of an
X-ray
a lattice plane of a crystal.
Huygens' wavelet construction that the re-emitted waves combine to produce a set of reflected wave
fronts, several of which are shown in the figure. These wave fronts constitute a beam reflected from the lattice plane. As can be seen from the
of re-emitted waves.
It is
apparent from
figure^the angle of reflection
is
equal to the angle of incidence.
This result allows us to deal with
sisting
of the
ing surfaces.
this
an even simpler model
crystal con-
of lattice planes which are assumed to be partially reflectFigure (14-3) illustrates the incident beam passing through
set
the crystal and being partially reflected from each lattice plane.
figure the incident
and
reflected
beams
In this
Emerging
beams. These beams inter-
are represented as rays.
crystal is a set of reflected
each other, and the interference is destructive unless each beam is
in phase with all the others. For a given lattice plane spacing d, this
from the face of the
fere with
condition
is
reflection 6.
satisfied
only for certain values of the ang^e of incidence and
these angles, consider a typical pair of reflected
To determine
and r + 1. Since there is no phase change on reflection, these
will be in phase if the difference between the path length of the
radiation following beam r + 1 and the path length of the radiation
beams
beams
Sec. 2]
MEASUREMENT OF X-RAY SPECTRA
following
beam
r is
equal to an integral
they will be in phase
479
number of wavelengths. That
is
if
nX,
2, 3,
1,
Since
djd
sin 6
and
eld
cos<f>
cos (180
r+
s\
<A
Figure 14-3.
IS
r\r
20)
1
/
Illustrating the partial reflection of an
X-ray beam from each
lattice plane
of a crystal.
this
condition can be written
A
sin 6
This
cos (180
2d)
nX
sin 6
is
-4-n (1 -
cos 26)
= -A- (1 -
sin 6
2 sin 2 d)
nX
sin 6
or
2d
This condition
nX,
1, 2, 3,
(14-1)
Bragg 's law. For an incident beam of X-rays of
beam will be found only if 6 is one of the solutions
condition. The solution corresponding to n = 1 gives the so-called
wavelength
to this
sin d
X,
is
called
a reflected
first order reflection;
2 gives the second order
intensity of the first order reflected
beam is
reflection, etc.
The
greater than that of the second
order, the second greater than that of the third, etc., because the intensity
of the radiation scattered by atoms decreases with the increasing scattering
X-RAYS
480
angles corresponding to increasing values of n.
necessary to use the
first
For
this
reason
[Ch. 14
it is
often
order reflection although the higher orders give
better spectral resolution.
Figure (14-4) shows an X-ray tube and crystal spectrometer. Following
modern X-ray tubes are operated at high
vacuum, and the electron beam is produced by a heated filament F.
A high voltage applied between the filament and the anode accelerates
the electrons into the anode A, which usually must be water cooled. For
maximum X-ray yield, a heavy element such as tungsten is used for the
the design of Coolidge (1913),
An
Figure 14-4.
anode
however,
it is
apparatus used to measure X-ray spectra.
possible to use almost
emitted from the anode are defined into a
and
reflected
from the
crystal C.
any
solid substance.
beam by
For a given angle
X-rays
lead diaphragms D,
6,
only those wave-
The reflected beam
is defined by a second set of diaphragms D' and is detected by an ionization
chamber /. The X-rays enter through a thin window and pass between two
lengths are reflected for which Bragg's law
is satisfied.
producing ionization of the gas contained in this region (ionization
will be discussed later). Ions are attracted to the insulated plate
because of the electric field produced by the battery B, a current consequently flows through the galvanometer G, and the strength of the current
is proportional to the intensity of the reflected X-ray beam. By measuring
the galvanometer current as a function of the angle 0,t the spectrum of
plates,
by X-rays
by the anode is measured as a function of that parspacing d is known, this can be converted into a
a function of wavelength by using Bragg's law (14-1), since
the X-rays produced
ameter.
If the lattice
spectrum as
it is
easy to identify the order of the reflection,
n.
The lattice spacing d can be obtained for a simple cubic crystal like
NaCl from a knowledge of Avogadro's number N, the density p of the
is the molecular weight of NaCl,
crystal, and its molecular weight. If
t It
is
necessary to rotate both the ionization chamber and the crystal to keep both
angles equal to
6.
MEASUREMENT OF X-RAY SPECTRA
Sec. 2]
W grams.
then vV molecules weigh
and
in that
volume there
will
be
They
IN
will
atoms.
arranged at the corners of a cube of length
p
Substituting the
known
d
2.820
occupy a volume W\p cm3
For NaCl, these atoms are
Thus
,
2N
values of N,
d.
481
W, and p
x 10- 8 cm
for
gives
NaCl
This value has been verified by using Bragg's law to determine d in terms
of the known wavelengths of certain X-rays. These wavelengths were
from measurements using a ruled diffraction grating of
line spacing. Although the line spacing of any ruled
grating is very large compared to X-ray wavelengths, ruled gratings can
still be used if the X-rays are incident at a grazing angle because, under
actually obtained
directly
measurable
these circumstances, the effective
This technique
(i.e.,
projected) line spacing
is
very small.
and it is not appropriate for the routine
measurement of X-ray spectra. However, it is valuable for calibration
purposes, and it forms the basis of one of the most accurate experimental
determinations of the value of Avogadro's number.
is
very
difficult,
X-rays are extremely useful in studying the properties of crystals.
By bombarding
a crystal with a beam of X-rays of known wavelength
composition, and determining the angles at which Bragg reflections occur,
it
is
possible to determine the structure
and dimensions of the
crystal
As an example, such measurements confirm our supposition
that the atoms of NaCl are arranged in a regular cubic lattice. However,
we shall not describe the results of such experiments because we are
more interested in describing the results of the converse experiments,
in which the wavelength composition of an X-ray beam is measured by
using a spectrometer with a crystal of known properties. Typical results
lattice.
are indicated in figure (14-5), which represents the energy content per unit
wavelength interval emitted by an X-ray tube with a 74
anode. The
wavelength spectrum is shown for two different values of accelerating
two different values of the kinetic energy E eV o$
upon the anode. It is apparent that the observed
spectrum can be decomposed into a set of sharp lines superimposed on a
continuum. This decomposition is convenient because measurements
using a variety of anode materials and accelerating voltages demonstrate
that the properties of the continuum spectra are quite different from those
voltage V,
i.e.,
for
the electrons incident
of the line spectra.
Some
of these properties are described below.
X-RAYS
482
[Ch. 14
V=80kv
40 kv
0.2
0.4
0.8
0.6
1.0
1.2
1.4
1.6
X(xl0~ 8 cm)
Figure 14-5.
Typical X-ray spectra.
Continuum Spectra
(I)
The shape of the continuum spectrum depends only on the energy of
electrons incident on the anode, and not on the nature of the anode.
the
In particular, the short wavelength cut off Xq is inversely proportional
cjl
to the incident electron kinetic energy E. The cut-off frequency v
2.
follows the empirical law
hvv
where h
is
=E
(14-2)
in excellent numerical agreement with the value of Planck's
constant obtained from the photoelectric effect, etc.
3. The total X-ray energy per incident electron, W, which is proportional
to the integral over X of the continuum part of the spectrum, obeys the
empirical equation
W = kZE
where
~ 0.7
10~ 4
for
(14-3)
W and E in Mev
and
Z is
The
the atomic
number of the anode
material
fraction of electron kinetic energy converted into X-ray energy
W\E
= kZE
is
(14-3')
X-RAY LINE SPECTRA
Sec. 3]
For the
typical case
Z=
483
E=
90 and
Mev, WjE
0.05
is
only about 0.3
percent.
Line Spectra
(II)
1.
The wavelengths of the lines depend only on the nature of the anode
on its atomic number Z), and not on the energy of the incident
(in fact,
electrons.
2.
However, a
line
cannot appear in the spectrum unless the incident
electron kinetic energy
E satisfies
the relation
E^?ET =
where
3.
is
The
hv
hc\l
(14-4)
the frequency and X the wavelength of the line in question.
total
X-ray energy emitted in a particular
line increase
with
increasing incident electron kinetic energy according to the empirical
equation
Ioz{E- ET) n
(14-5)
where
n
The
origins of the
properties.
3.
For
this
r+j 1.5
continuum and
reason
it
is
line spectra are as different as their
convenient to discuss them separately.
X-ray Line Spectra
The
origin of the X-ray line spectra
may
be understood
if
we
a particular one of the processes by which an energetic electron
down
as
it
passes into the anode of an X-ray tube.
consider
is
This process
slowed
is
the
from the incident electron to the atoms of the anode by
Coulomb interaction collisions between the incident electron and the
transfer of energy
electrons of the atoms.
An
upon an atom can transfer an appreciable
an atomic electron only in a close collision
This statement may be justified and made quantitative,
energetic electron incident
fraction of
its
kinetic energy to
with that electron.
by considering the Coulomb interaction potential energy between two
electrons
V(r)
e 2 /r
Evaluating from this equation the separation distance r at which the
is V(r)
10 2 ev, we find r
10 -9 cm. Since the energy
potential energy
from the incident to the atomic electron cannot be bigger than
the maximum potential energy which they experience, we see that energy
transfer
X-RAYS
484
transfers greater than 100 ev occur only in collisions in
of closest approach between the two electrons
distance
is
is less
[Ch. 14
which the distance
than 10 -9 cm. This
about 10 percent of an atomic "diameter." Such close collisions
are not very frequent because the atomic electrons in the outer shells can
move out of
the
way of
the incident electron and, although the atomic
electrons in the inner shells cannot
do
this as
bound to the
come near
Almost all the
they are tightly
center of the atom, the probability that the incident electron will
these electrons
is
small because the inner shells are small.
an energy transfer of a few electron volts or,
few tens of electron volts. In such collisions, it is energetically
possible to excite only an atomic electron in one of the outer shells. The
electron can be excited into a bound state, or into a continuum state
collisions are associated with
at most, a
atom
leaving the
ionized.
some other
In either case, the electron, or
drop back down through the various excited states
to the ground state, and the atom will emit lines of its optical spectrum.
In a small energy transfer collision there is not enough energy available to
excite an atomic electron from an inner shell. These electrons are in
highly negative energy levelsf and, if they are to be excited, they must be
excited all the way up to one of the slightly negative bound energy levels,
or to one of the positive continuum energy levels, because all the intervening
energy levels already contain their complete complement of electrons
(cf. section 5, Chapter 13).
However, there are occasional collisions in which the energy transfer
is a big fraction of the energy of the incident electron.
Providing this
energy is large enough (cf. equation 14-4), it is possible to excite an atomic
electron of an inner shell from its highly negative energy level to one of
electron, will eventually
the slightly negative or positive energy levels.
very highly excited state.
state,
and
in this process
members of
it
The atom
This leaves the atom in a
will eventually return to its
ground
emits a set of very high energy quanta which are
spectrum. As an example, let us assume that
removed from the n = 1 shell. In the first step of
the de-excitation process an electron from one of the shells of higher n
drops into the hole in the n = 1 shell; for instance, an n = 2 shell
electron could drop into the hole. This would leave a hole in the n = 2
shell, but the excitation energy of the atom would be considerably reduced
because an electron in the n = 2 shell is much less tightly bound than an
electron in the n = 1 shell. Energy is conserved by the emission of a
quantum which has an energy equal to the decrease in the excitation energy
of the atom. In the next step of the de-excitation process an electron from
a shell of n > 2 drops into the hole in the n = 2 shell, a hole is produced
in the shell from which it came, and a quantum of the appropriate energy is
an electron
its
X-ray
line
is initially
For example, the energy of an electron
in the
innermost shell of
74
is
7 x
10 4 ev.
X-RAY LINE SPECTRA
Sec. 3]
485
is that a hole jumps
These steps continue until the hole has worked its
way to the outermost shell, where it is filled by the electron initially ejected
emitted.
Note
that the net effect of each of these steps
to a shell of higher
from the n =
The energy
n.
some other
or
shell,
electron.
an atom which are involved in the emission of its
X-ray spectrum can be conveniently represented in terms of an energy
level diagram. We present in figure (14-6) a diagram for the 92 U atom
showing all its X-ray energy levels through n = 4. In this diagram the
levels of
K series
10
*I
series
.L,
11= -
-L !r
-illl
'
'
/Mi
series
Mm
Af
'My
'
* (- .
"-I
10'
'-
'
'
uU
'
'
3
3
3
3
3
_5 i
-\ - -\
.*-JV,
'-
'
'
'i
'
'
'V
10'
Figure 14-6.
transitions
The higher energy X-ray
between these
levels for the
uranium atom, and the possible
levels.
energy of the atom in
its ground state is denned as zero. Each level gives,
on a logarithmic scale, the energy of the atom when one electron of the
indicated quantum numbers n, I, j is missing.f This can be thought of,
then, as an energy level diagram for a hole with quantum numbers n, l,j.
Since an electron of negative energy is missing, the energies of all these
levels are positive.
The energy levels are also identified by a notation
commonly used
in
X-ray spectroscopy.
In this notation the value of
t Note that the quantum numbers n, I, j, which completely specify the energy of an
atom with one electron in a subshell, also completely specify the energy of an atom with
one hole in the subshell. Because of the coupling of the individual spin and orbital
angular momenta required by the exclusion principle for the electrons in a subshell, the
magnitude of the
hole
is
the
the hole.
angular
same
and spin angular momenta of a subshell which has one
the subshell contained only the single electron corresponding to
total, orbital,
as
if
In either case the
momentum
is
quantum number
not written as
it
specifying the magnitude of the spin
can only be
J.
Cf. discussion of table (13-9).
X-RAYS
486
[Ch. 14
quantum number n is given according to the scheme shown in table
The Roman numeral subscripts in figure (14-6) label the levels
according to decreasing energy, but do not specify the values of / and j.
the
(14-1).
Because an X-ray energy
level
diagram gives the energy
the energy ordering of the levels with respect to the
n,
I,
is
levels
of a hole,
quantum numbers
the reverse of that found in an optical energy level diagram giving
the energy levels of an electron.
atom with one
electron of
This
is
true because,
quantum numbers
atom with a corresponding hole
more energy must be given to the atom in order
the energy of an
is
as
to
TABLE
if
the energy of an
n, /,/is particularly negative,
particularly positive,
remove the
electron.
(14-1)
rhe Spe ctroscopic Nota tion for n
n
Spectroscopic
notation
Keeping
this difference in mind, all the familiar features of the dependence
of atomic energy levels on the quantum numbers n, I, j can be found in
an X-ray energy
level
diagram.
In fact, they are
more pronounced,
particularly the splitting of the energy levels according to the value of j.
In the
of
shell
92
this splitting is
more than 2000
and
ev,
it is
larger
than the splitting according to the value of /. In such a case, it is hardly
appropriate to describe they dependence of the energy levels by the optical
energy level terminology "fine structure splitting." This strong j depend-
atoms except those of
which
is
characteristic of the inner shells of all
very low Z,
is
partly due to the usual increase in spin-orbit splitting with
ence,
decreasing
n,
but primarily due to other
relativistic effects
which become
very large for the high velocity electrons that populate these shells
(cf.
section 6, Chapter 13).
As we have
X-ray
indicated,
it is
convenient to think of the production of
terms of the creation of a hole in one of its higher
and the subsequent jumping of the hole through its lower
line spectra in
energy
levels,
energy
levels,
with the emission at each
carrying the excess energy.
jump
of an X-ray quantum
The frequency v of the quantum bears
the usual
which it carries; E = hv. However, not all
transitions are possible. There are the following set of selection rules for
the change in quantum numbers of the hole
relation to the energy
A/= 1
X-RAY LINE SPECTRA
Sec. 3]
487
Note that these are the same as the selection rules (13-30) and (13-31)
which govern the one optically active electron in an alkali atom.f In the
X-ray energy level diagram for 92 U, we have shown the transitions that
obey (14-6). The totality of X-rays which are emitted in these transitions
(plus a few which are observed to be emitted very infrequently in violation
of these rules) constitute the X-ray line spectrum of the atom. It is conventional to group those lines produced by transitions from the K shell
into the so-called K series, from the L shell into the L series, etc. X-ray
energy level diagrams are obtained experimentally from an analysis of
measured X-ray spectra, and the analysis is greatly facilitated by this
grouping.
One
between X-ray and optical spectra is that X-ray
from element to element; there are none of the
abrupt changes from one element to the next which occur in optical spectra.
The reason is that the characteristics of X-ray spectra depend essentially on
the binding energies of the electrons in inner shells. With increasing atomic
number Z, these binding energies simply increase uniformly, owing to the
higher nuclear charge, and are not affected by the periodic changes in
the number of electrons in the outer shell. The regularity of X-ray spectra
was first observed by Moseley. In 1913, shortly after the invention of the
crystal spectrometer, he made a survey of X-ray spectra and obtained data
for a number of elements on the wavelengths of two prominent lines,
Kx and Lx The first line is emitted in transitions from the K shell to the L
shell, and the second is emitted in transitions from the L shell to the
shell. (Actually, Ka and L are each comprised of two lines, but Moseley
x
could not resolve this structure.) Moseley found that these wavelengths
could be fitted to within the experimental accuracy by the simple empirical
striking difference
spectra vary smoothly
formulas
For
For L a
1/A
= CK J,Z -
1/A
= Cia(Z -
(14-7)
l)
7.4)
(14-7')
CKa and Cia are constants having values of the order of the Rydberg
constant for an infinitely massive nucleus, R x defined by (5-15). Moseley
where
on the basis of the Bohr theory of the atom, which
had been proposed in the same year as his experiments. According to
equation (5-14), the Bohr theory predict* that the energy of an electron
in the shell of quantum number n is
interpreted these results
En
=-R x hc
rr
(14-8)
t This is a manifestation of the symmetry, described in the previous footnote, between
the properties of a subshell with one electron and a subshell with a corresponding hole.
X-RAYS
488
[Ch. 14
where we have introduced R^. As the excitation energy of an atom with a
electron missing from the shell of quantum number n is equal to \En \,
the energy E of the X-ray quantum emitted when the hole jumps from
the n t shell to the n t shell
E
This energy
is
RJicZ\\\n\
1/n
2
)
quantum by
related to the frequency v of the
is
E=
hv
the equation
hcjk
so
hcjk
RJicZ\\\n\
1/A
RZ^l/nf
1/n
2
)
or
For
nt
a,
and n t
2; for
Lx
nt
1/n ,)
2 and n t
(14-9)
3.
Therefore the Bohr
theory predicts
For
For
Lx
\\l
1/A
=
=
[R^l/1 2
1/2 )]Z
(14-10)
(14-10')
[KJ1/2
1/3 )]Z
In each case the value of the constant in square brackets agrees well with
the constant
in equations (14-7)
equations contain (Z
1)
and (Z
and
(14-7').
7.4),
instead of Z, was explained by
That these empirical
Moseley as due to shielding of the nuclear charge by atomic electrons
(which was not accounted for in Bohr's derivation of equation 14-8).
According to Moseley, the one electron remaining in the K shell during
the transition of the hole from the K to L shells shields the nucleus so that
the effective Z is reduced to (Z 1) for the emission of the Kx line.
Similarly, the nine electrons remaining in the K and L shells shield the
shells. However,
nucleus during the transition of a hole from the L to
this shielding is not perfect, and the effective Z for the emission of the Lx
line is reduced only to (Z 7.4). Moseley's description of shielding is
closely related to the description connected with equation (13-22), and
both are only approximations to the more accurate description provided
by the Hartree theory. Furthermore, later measurements have shown that
the empirical formulas (14*7) and (14-7') are also only approximations.
Nevertheless, Moseley's work was an important step in the development
of modern physics. His simple and successful application of the Bohr
theory to X-ray line spectra provided one of the earliest confirmations of
that theory.
By using
the empirical formulas (14-7)
and
(14-7')
to
determine Z, Moseley established unambiguously the correlation of the
nuclear charge of an
elements.
He
atom with
its
ordering in the periodic table of the
found, for instance, that the atomic number of
27
Co
is
less
X-RAY CONTINUUM SPECTRA
Sec. 4]
489
than that of 28 Ni even though its atomic weight is greater. He also showed
that there were gaps in the periodic table, as it was then known, at Z = 43,
61, 72,
and
Elements of these atomic numbers have
75.
all
subsequently
been discovered.
4.
X-ray Continuum Spectra
The transfer of energy from an incident electron to the electrons in the
atoms of an X-ray tube anode is the process which is primarily responsible
for the slowing down of the incident electron, and it also leads (in the
infrequent case of a large energy transfer) to the emission of the X-ray line
spectra. But this process is not the only one which the incident electrons
undergo. Another is the Coulomb interaction scattering of these electrons
atoms in the anode. In a single
an incident electron suffers large accelerations and
can be scattered through a large angle. Although the electron does not
in close collisions with nuclei of the
collision of this type
lose energy to the nucleus because nuclei are so massive that they
do not
sometimes does lose energy by electroThis radiation constitutes the continuum X-ray
recoil appreciably, the electron
magnetic radiation.
spectrum.f
Both classical theory and quantum theory predict the emission of electromagnetic radiation by a scattered electron. Classically, small amounts of
radiation are emitted at
all times in every scattering because Maxwell's
equations lead to the conclusion that charged particles radiate whenever
they accelerate. In the quantum theory this is not what happens. Instead,
large
amounts of radiation (quanta) are emitted occasionally in a scattering
The situation can be compared to that which exists for the atom
process.
the classical theory predicts the constant emission of radiation by the
accelerated electron circling the nucleus, while the quantum theory
when the electron changes
In the case of the bound atomic electron, we know that
predicts the occasional emission of radiation
quantum
states.
conclusions drawn from the
drawn from
quantum theory
are always correct
the classical theory are generally incorrect.
and those
In the case of
the unbound scattered electron,
it is also true that the quantum theory
always leads to the correct conclusions. However, in this case the classical
theory does not lead to conclusions which are generally incorrect. In
theory is capable of describing quite well certain features
of the production of continuum spectrum X-rays. The contrast between
fact, the classical
t X-rays of the
continuum spectrum are often
word means braking
radiation
(i.e.,
called Bremsstrahlen.
deceleration radiation)
the theoretical explanation of their origin.
and
is
name
This
German
derived from
X-RAYS
490
[Ch. 14
the two cases reflects the fact that the differences between a quantized
and
a non-quantized theory are not so fundamental in the case of an unbound
particle as they are in the case of a
bound
particle.
According to equation (2-7) of the classical theory, a particle of charge
ze, suffering an acceleration a, will emit energy in the form of electromagnetic radiation at the rate
R=
l*
3
When
(14-11)
by a nucleus of charge Ze, it experiences a
and, if the mass of the particle
2
Consequently the rate
is m, its acceleration is proportional to zZe jm.
of emisssion of energy in electromagnetic radiation can be written
this particle is scattered
Coulomb force which is proportional to zZe2
oc s2 a 2 oc 4
/w 2
(14-12)
factor 1/m 2 predicts that the emission of electromagnetic radiation
The
very
much more important
heavy
particles.
This
is
for electrons than
true;
it is
is
for protons or other
the energy emitted in continuum spectrum
X-rays by a heavy charged particle in traversing matter is almost immeasurably small. The factor Z 2 predicts that anodes of high atomic
much more
number
are
verified
by experiment.
efficient
By
producers of these X-rays.| This
is
also
evaluating the proportionality constant in
and then integrating over the (average) trajectory of the incident
coming to a stop in the anode, it is possible to estimate the
total continuum spectrum X-ray energy JFthat is radiated. The results are
in reasonable agreement with the empirical equation (14-3).
The classical theory is also capable of giving a good account of the
polarization of the continuum spectrum X-rays. This theory predicts that
the X-rays, emitted in the direction r during an acceleration a, are polarized
with the electric vector perpendicular to r in the plane defined by r and a.
Now the higher energy X-rays are all emitted in "head-on" collisions in
(14-12),
electron in
always approximately antiparallel to i, the direction of the
incident beam of electrons. Consequently, such X-rays have a large net
polarization. In the collisions which lead to the emission of lower energy
which a
is
X-rays, the direction of a
polarization
is
is
not so constrained and, as a
result, the net
small.
Classical electromagnetic theory even gives
an accurate prediction of
continuum
the angular distribution of the X-ray energy emitted in the
The
dependence also shows that the radiation emitted
atomic electron
(i.e.,
for
Z=
1) is negligible.
in scattering
from an
X-RAY CONTINUUM SPECTRA
Sec. 4]
491
For incident electrons of non-relativistic
by equation (2-6) is
spectrum.
energies, the angular
distribution predicted
2
1(6) oc sin 6
where
which
is
(14-13)
the angle between r and the average direction of acceleration,
i.|
For incident electrons of relativistic energies, certain
is
relativistic
transformation effects shift the angular distribution in the
direction of
i.
The one aspect of the emission of continuum spectrum X-rays that
cannot be explained by classical theory is the shape of the spectrum; this
can be explained only by quantum theory. Consider the cut-off frequency
(equation 14-2), which is certainly the most striking feature of the spectrum.
v
Figure 14-7.
thin
>-
=E/h
anode X-ray continuum frequency spectrum.
This cut-off exists because the highest energy quantum that can be emitted
by an electron of kinetic energy E has the energy hv = E, where v is
its frequency.
In fact, this was one of the earliest pieces of evidence in
support of Einstein's theory of quanta. Unfortunately, an explanation
of the other features of the spectrum cannot be given by such elementary
quantum
subject
results
theory, but only in terms of
quantum electrodynamics. This
beyond the level of this book, and we can only describe the
of the theory. For electrons of energy E incident upon an anode
far
is
which is so thin that their energy remains essentially constant in passing
through it, the spectrum is predicted by the theory to have the simple
shape shown in figure (14-7). The quantity
unit frequency interval.
I(v) is the
X-ray energy per
For comparison with the data presented
in
we convert this to I(X), the X-ray energy per unit wavelength
The relation between two quantities such as these has been derived
section 2,
interval.
before
(cf.
equation 2-27).
It is
IW =
I(v)
t This angular distribution has the same form as the angular distribution of the
energy emitted in radiofrequency radiation by electrons accelerated back and forth
along a linear antenna. From the point of view of the classical theory, this process is
essentially the same as that involved in X-ray emission.
X-RAYS
492
Now,
since
I(v)
I(X)
[Ch. 14
~ constant,
<
>
cjv
E/h
can be written
I(X)
This spectrum
is
~ constant/A
ch/E
The predicted wavelength
good agreement with measurements
plotted in figure (14-8).
spectrum for a "thin" anode
is
X = ch/E
Figure 14-8.
A thin
in
anode X-ray continuum wavelength spectrum.
made for thin anodes. It is also in agreement with the more
measurements using "thick" anodes in which the incident electrons
are brought to a stop. We show this in figure (14-9) by generating a
thick anode spectrum from the sum of several thin anode spectra. Four
that have been
typical
Figure 14-9.
Illustrating
the composition of a thick anode X-ray continuum wavelength
spectrum.
values of
are used to represent the range of electron energies present
The dotted line shows the results of this approximate
The dashed line indicates the results of an accurate calculainvolving integration over the electron energy. Agreement with
in the anode.
calculation.
tion
experimental data, such as presented in section
2, is excellent.
THE SCATTERING OF X-RAYS
Sec. 5]
The Scattering
5.
493
of X-rays
Having discussed the nature and production of X-rays, we now turn
our attention to their interaction with matter. Consider a beam of X-rays
passing through a foil and into an X-ray detector, as indicated schematically
in figure (14-10). A quantitative measurement of the interaction of the
X-rays with the material contained in the foil is obtained by measuring
the intensity of the attenuated beam striking the detector, and then, by
removing the
measuring the intensity of the incident beam. The
of the attenuation of the foil for these X-rays,
foil,
results are expressed in terms
which
is
defined as the fractional decrease in the intensity of the beam.
Incident
Attenuated
beam
beam
Figure 14-10.
An
Detector
jl
apparatus used to measure the scattering and absorption of a beam
of X-rays.
In passing through the
foil, X-rays can be completely absorbed from the
beam, and they can also be scattered. If the angle subtended at the foil
by the detector is very small, then essentially all the scattered X-rays will
miss the detector and the measurement will give the attenuation due to
both absorption and scattering.! In such measurements it is found that
the attenuation increases with the thickness of the foil and, generally,
with the atomic number of the material which it contains. The attenuation
also depends
on the energy of the X-rays, but the dependence
easy to describe.
the absorption
In
this
Several very interesting
and
section
scattering processes
we
shall
discuss
phenomena
is
not so
are involved in
which produce the attenuation.
those
involved
in
the
scattering
process.
Consider the interaction of a
atoms of the
foil.
We make
beam of X-rays with
1. The wavelength of the X-rays
dimensions of the atoms
By
varying this angle
the electrons in the
the following assumptions.
is
small compared to the characteristic
< 10"
cm
(14-14)
it is possible to separate experimentally the attenuation
due
from the attenuation due to scattering, and measure these quantities
independently. The attenuation due to scattering can also be measured separately by
an experiment in which the scattered X-rays are detected.
to absorption
X-RAYS
494
2.
The energy of
quantum of
the X-rays
is
large
[Ch. 14
compared
to the
binding energy of the atomic electrons
/n>>B.E.
3.
rest
The energy of a quantum of the X-rays
mass energy of an electron
hv
The
(14-15)
let
small compared to the
<^mc 2
(14-16)
be discussed
of these assumptions will
validity
moment
is
us explore their consequences.
The
first
later.
For the
assumption implies
by the electrons in
same atom, will be essentially random, or incoherent.
It rules out interference effects between different atoms (such as those
discussed in section 2), as well as interference effects within the same atom.
If this assumption is valid, the interaction of any electron with the X-rays
is independent of the interactions of all the other electrons in the foil, and
we therefore need to consider only one of these interactions. The second
assumption allows us to ignore the binding of the electrons in the atoms and
consider them to be free. The third assumption allows us to ignore the
quantum aspects of X-rays which manifest themselves in Compton
that the phase relations between the X-rays scattered
different atoms, or the
Making these assumptions, we
give a classical electromagnetic
treatment of the scattering process due to
Thomson (ca. 1900). In this
beam of electromagnetic
scattering.!
treatment the X-rays are considered to be a
waves incident upon a
with the electron
is
single free electron.
The
sinusoidally oscillating electric field incident
with the magnetic
strength
and
interaction of the X-rays
then due to the interaction of
field is negligible.)
We
upon
its
it.
charge with the
(The interaction
describe the instantaneous
direction of the electric field at the location of the electron
by the vector E. This electric field acting on the electron produces a
force F = eE, and an acceleration
a
where E, F, and a
F//M
all oscillate
incident electromagnetic waves.
= -eE/w
(14-17)
sinusoidally with the frequency v of the
Under
the influence of this acceleration,
v, and
same frequency as,
and in phase with, the incident waves. The energy emitted in these waves
Thus the electron
is supplied by the incident electromagnetic waves.
absorbs energy from the beam of X-rays and scatters it in all directions.
The rate of emission of energy by the accelerated electron, that is, the
the position of the electron will oscillate sinusoidally with frequency
the electron will emit electromagnetic waves of the
t Equation (3-17) shows that if (14-16) is valid the
scattering of X-rays from a free electron is negligible.
change of wavelength in the
THE SCATTERING OF X-RAYS
Sec. 5]
rate at
2=
which the electron scatters energy,
given by equation (14-1 1) with
-1:
S
where here we write
5 for
E changes
to deal with
Using (14-17),
this
|e*/if^c
constantly in time,
its
feV/c3
scattered.
S
As
is
495
does
also,
becomes
(14-18)
and
it is
most convenient
time average,
S = fe^/mV3
= fe^/m2
*:
(14-19)
The quantity S is the average energy emitted in all directions per second
by the electron in the form of electromagnetic radiation, and we define
be the intensity of the scattered radiation.
/, the intensity of the incident radiation. We define
this quantity to be the average energy carried by the beam per second
across a 1 cm2 area oriented normal to the direction of the beam.f
Equation (2-5) shows that the relation between / and the average energy
this quantity to
Next
density
let
p~
us evaluate
is
I=pc
According to equation
(2-4),
p~
(14-20)
is
12/477
where we delete the subscript for transverse. Thus / may be written
I=cE?\bn
Now
(14-21)
comparing equations (14-19) and (14-21), we see that both
and /are proportional
other. That is
to
E2
Sand /are
Therefore,
proportional to each
S = a TI
(14-22)
where a T is the proportionality constant. Evaluating a T from equations
(14-19) and (14-21), we have
8ir
= -t (e2/wc2 =
2
6.66
x 10" 25 cm 2
(14-23)
This quantity, called the Thomson cross section of the electron, describes
a characteristic property of a free electron. Through equation (14-22),
aT tells what fraction of the intensity of a beam of X-rays is scattered into
all directions by a free electron due to the Thomson scattering process.
t
Note that the dimensions of S are
cm~*-sec _1
ergs-sec
-1
,
while the dimensions of I are ergs-
X-RAYS
496
It is
of
cm3
[Ch. 14
reasonable to call a T a "cross section" because it has the dimensions
But the name has a deeper significance that can be understood if
.
we imagine
the electron at the center of an imaginary disk oriented normal
beam, and of area equal to a T cm2 Then, if all the radiation in the
beam hitting the disk were scattered, the ratio of the scattered intensity 5
to the incident intensity I would be just a T in agreement with equation
(14-22). This is easy to see by referring to figure (141 1) (in which the
to the
Figure 14-1
1.
The geometrical interpretation of
a cross section.
area of a T is exaggerated by about a factor of 1023 !). The average energy
incident per second on the 1 cm2 area is equal, by definition, to the incident
intensity
/.
The
fraction of this energy that hits the imaginary disk
cm2 /l cm2
is
a T If all this radiation were scattered,
the energy scattered per second, which is the scattered intensity S, would
be a T I. This proves our point. The geometrical interpretation of a cross
section is convenient for visualization, and even calculation, but it should
precisely the ratio
Figure
14-12.
aT
Illustrating
the relation between the vectors which determine the
polarization of scattered X-rays.
literally as to lead to the conclusion that there actually
2
a well-defined distance r, satisfying the equation 27rr = a T such
electron
are
scattered,
distance
of
the
within
this
that all X-rays passing
not
scattered.
X-rays
are
whereas all other
Classical electromagnetic theory also makes predictions concerning the
not be taken so
exists
angular distribution of the X-rays scattered by an electron in the Thomson
scattering process. Consider figure (14-12). The direction of the incident
Sec. 5]
THE SCATTERING OF X-RAYS
497
beam of X-rays
is the vector i. The vector E, which is perpendicular
to i,
the direction of the electric vector of a typical polarization component
of the beam, and the acceleration a which this component gives to the
is
electron
lies in the opposite direction.
The angular distribution of the
radiation emitted by the electron suffering this acceleration is given by
(14-13), where d is the angle between the direction of acceleration a and
the direction of emission of the radiation r. This angular distribution may
Figure 14-13.
The element
of solid angle between
and
d.
also be written in terms of the angle
between r and i, and the angle
on the figure) specifying the azimuthal orientation of E.
Doing this, and then averaging over all possible azimuthal orientations
to
treat the case of an unpolarized incident beam,
the angular distribution is
(not indicated
obtained as a function of the angle 0.
cc (2
5(0)
where S(0)
is
It is
sin 2
0)
(14^24)
the average energy emitted per second per unit solid angle
The integral of S(@) over all solid angles is equal to the
at the angle 0.f
average total energy emitted per second, or the intensity
of the emitted
Let us write (14-24) as
radiation S.
S(0)
and evaluate the
K(2
sin 2
0)
(14-24')
integral:
S=
J
The element of solid
cones of half-angles
the sphere. This
is
5(0) dQ.
angled is equal to the area intercepted on a sphere by
and
d, divided by the square of the radius of
illustrated in figure (14-13).
The area
is 27rr 2
sin
'
so
,
n=
dil
f Since solid angle
is
2ttt
sin
277 sin
dB
dimensionless, the dimensions of S(@) are ergs-sec
(14-25)
1
.
X-RAYS
498
[Ch. 14
Consequently
S=
2ttK f*(2
sin
0)
sin
d&
(14-26)
Jo
According to equations (14-22) and (14-23),
{^Imcfl
S=
By
(14-27)
carrying out the integration of (14-26), and equating the resulting
we
expression to (14-27),
S{&)
evaluate K.
With
i(e 2 /wc 2 ) 2 (2
this value, (14-24')
sin2
0)/o
becomes
(14-28)
Denning the symbol
da T jdQ.
=
=
i(e
/wc 2) 2 (2
3.98
10- 26
(2
sin 2
equation (14-28) can be written
S(@)
0)
sin 2
0)
cm2 /unit
solid angle
(14-29)
(14-30)
dQ,
The quantity da T \dl, which also has the dimensions cm2 is called the
differential Thomson cross section of the electron. It tells, through equation
(14-30), what fraction of the intensity of a beam of X-rays is scattered in a
Thomson scattering process by a free electron into a unit solid angle at
the angle 0just as a T specifies, through equation (14-22), what fraction
of the intensity is scattered in that process by a free electron into all solid
,
angles.
It is possible to give
a geometrical interpretation to a differential
which we gave above for a cross section.
However, we again caution the reader against taking these interpretations
cross section similar to that
too
literally.
Having completed our discussion of the Thomson scattering of X-rays
by a free electron, let us return to the more realistic problem of the
scattering of X-rays by an atom. If assumptions (14-14) and (14-15) are
satisfied, this problem is trivial because each of the Z electrons of the
atom scatters X-rays independently, and the total intensity scattered by
the atom is Z times the intensity scattered by a single electron. Evaluating
this intensity from equation (14-22), we obtain for the total .?
S = Za T I
Defining
= Za T
(14-31)
S=as I
(14-32)
aSa
this
can be written
THE SCATTERING OF X-RAYS
Sec. 5]
The quantity
<r
Sn
is
499
the scattering cross section per atom.
given by (14-31) only
if
It
has the value
the assumptions mentioned above are satisfied.
If they are, the differential scattering cross section
per atom
is
also simply
do?
^darp
~
= Z
di\
This
related to the total
is
S(Q) by the equation
S(&)
From
(14-33)'
dQ.
1
^f
dil
these scattering cross sections per
atom
(14-34)
it is
easy to calculate the
predicted attenuation, due to scattering, of a foil containing a
number of atoms per
unit volume.!
I"
comparing
this prediction
known
with the
vresults of the experiments described
at the beginning of this section,
agreement to within 10 or 20 percent is found for atoms with Z < 15
and for X-rays of wavelength near I = 0.2 x 10~ 8 cm. This formed the
basis of experiments performed by Barkla in 1909, which constituted the
first measurements of Z, the number of electrons in an atom.
The quantum energy of X-rays of wavelength X = 0.2 x 10~ 8 cm is
hv = 6 x 10 4 ev. As this wavelength and energy satisfy both assumptions (14-14) and (14-15) for atoms of small Z (and also satisfy assumption 14-16), we can understand why the cross sections per atom measured
with these X-rays agree, approximately, with the predicted values.
We
can also understand why they do not agree for measurements made with
X-rays of longer wavelength and atoms of larger Z. For these cases, the
X-ray wavelength is comparable to, or larger than, the characteristic
dimensions of the atoms, and also the electrons do not act as if they were
For both these reasons the electrons of the atom do not interact
beam of X-rays. In the limit of very long
wavelengths and very low quantum energies, the interactions are the very
opposite of being independent because, under these circumstances, the
bound electrons of the atom are all oscillating under the influence of
free.
independently with the incident
the incident electromagnetic waves with exactly the same phase, and the
electromagnetic waves emitted by any one electron are completely in
phase with those emitted by all the others. Because of this phase coherence
the scattered intensity will be very
much
larger.
It is
easy to evaluate this
by thinking of all the Z electrons oscillating together as one "big electron"
of charge Ze and mass Zm. According to equation (14-17), the acceleration of this object will be the
to (14-1
1)
the energy which
single electron
t
same
it
by a factor of
Such a calculation
will
as that of a single electron, but according
radiates will be larger than that radiated
Z2
Thus,
by a
in this limit the actual scattering
be carried out at the end of section
7.
500
X-RAYS
cross section per atom,
the electron,
The
will
(x,^,
a T andZtimes
,
be
Z 2 times the Thomson cross
[Ch. 14
section of
larger than the prediction of equation (14-31).
actual differential scattering cross section per atom, dosJdQ., will also
times larger than the differential Thomson cross section of the
Z2
be
da T jdQ,, and Z times larger than the prediction of (14-33).
For X-rays whose wavelengths and quantum energies are such that the
scattering from the various atomic electrons is neither completely coherent
nor completely incoherent, the situation is much more complicated, but it
can still be treated by classical electrodynamics. Assuming that the distribution of electrons in the atom is spherically symmetrical, the amplitude
of the electromagnetic wave scattered by each volume element is evaluated,
and then the contributions from all the volume elements are summed by
adding amplitudes with relative phase factors that depend on the location
of the volume element, the angle of scattering, and the wavelength. By
electron,
squaring the net amplitude, the total intensity scattered at the angle in
is found.
Dividing this by the incident intensity, the differential
question
scattering cross section per
atom
is
obtained.
^=^|F(x)|
dQ.
It
can be written
(14-35)
dQ.
where
F(X)
"pWCsin xrlxr)4*r* dr
(14-35')
with
= 2K sin (0/2)
and
K=
In these expressions X
and
is
277/2
the X-ray wavelength,
p(r) is the electron density
charge.
is
the scattering angle,
of the atom in units of the electronic
Since
p(r)4Trr
dr
Jo
we
see that the function F{%),
takes on
its
maximum
This condition
is
which
is
value, the atomic
satisfied
when % >
0,
called the atomic
number Z, when
form
factor,
sin %r\%r
which happens for
all
1.
when
K 0,
> 0.
and which also happens for
The behavior of F(%) for K-+0, or A
>-
of the next to
last
paragraph.
Its
confirms the conclusions
oo,
behavior for
--
shows the reason for
the small angle peaking of the scattered intensity mentioned in Section 2.
Figure (14-14) compares the differential scattering cross section per
0.71 x 10~ 8 cm incident
atom, measured with X-rays of wavelength A
THE SCATTERING OF X-RAYS
Sec. 5]
501
upon 6 C atoms, with the prediction of equation (14-33). The discrepancy at
small angles between the measured values of das jdQ. and the predictions
of equation (14-33) provide convincing evidence of coherent effects in
To compare
the scattering process.
das
the measured values
jdl with
the predictions of equation (14-35), which takes these coherent effects
into account,
in the
it
necessary to
is
form factor
Hartree
theory of the atom
obtained.
know
the electron density that enters
If the electron density
(14-35').
is
used, quite
d<Tc
for
do
x = 71 * 10
agreement
is
and (14-35') can be used to
Alternatively, equations (14-35)
obtained from the
satisfactory
cm z = 6
'
1
1
4-
.2
I
\
ail
-^is.
i
60
30
90
120
150
180
e
The differential scattering cross section per atom for A =
upon carbon. From A. H. Compton and S. K. Allison, X-rays
Experiment, D. Van Nostrand Co., Princeton, N.J., 1935.
Figure 14-14.
0.71
incident
in
X-rays
Theory and
an atom, p(r), from the measurements of
das jdO.. X-rays allow us to "see"
structure of bones in living tissue, but also the
extract the electron density of
its
differential scattering cross section
not only the invisible
invisible structure of atoms.
By
integrating
das
jdQ.
from equation (14-35) over
all solid
angles, the
predicted value of the scattering cross section per atom, a s is obtained as
an expression involving the electron density p{r). For cases in which this
,
expression can be evaluated because p(r)
is
known, the
results are in
reasonably good agreement with experiment, as long as X is greater than
about 0.2 x 10~ 8 cm. Experimental data are shown in figure (14-15).
The ratio of the observed
scattering cross section per
atom to
the prediction
of equation (14-31) is plotted as a function of the X-ray wavelength. The
curves are labeled with the atomic number of the atom. Note that these
data provide examples of the fact, mentioned above, that the observed
scattering cross sections per
0.2 x 10~ 8
cm
owing to the
atom agree with equation (14-31)
for X
but become larger than this prediction with increasing X
effect
of coherences.
Also note that the observed scattering cross sections per atom become
X-RAYS
502
[Ch. 14
smaller than the predictions of classical electromagnetic theory (with or
without coherences) when X becomes smaller than ~0.2 x 10~ 8 cm.
For X-rays of these wavelengths, the quantum energy hv is no longer very
small compared to the rest mass energy of an electron,
0.4
0.6
X
Figure 14-15.
wavelengths.
Compton and
The
is
1.0
atom
labeled by the atomic
K. Allison, X-rays
in
1.2
5.
1.4
10 s ev.
1.6
10~ 8 cm)
scattering cross sections per
Each curve
S.
0.8
(in
mc 2
for several
number
atoms and
of the atom.
a range of
From A. H.
Theory and Experiment, D. Van Nostrand Co.,
Princeton, N.J., 1935.
This violates assumption (14-16), and so the scattering can no longer be
treated by classical theory; instead, quantum theory must be used. We
have already discussed the energetics of the quantum theory for scattering
of X-rays by free electrons. The reader will find it under the heading
The Compton Effect in section 8, Chapter 3, and it would be appropriate
him to review the discussion there.
The wavelength spectra presented in Chapter 3 are for the case of
scattering from 6 C. For this small Z atom, the data show that the probability of scattering with an increase in the X-ray wavelength (Compton
scattering) is quite comparable to the probability of scattering with no
change in wavelength (Thomson scattering), even for incident wavelengths
at this point for
THE SCATTERING OF X-RAYS
Sec. 5]
x 10~ 8 cm
as large as 0.7
10 4 ev.
This
503
quantum
that correspond to
energies as
energies.
Thomson
scattering is very large compared to its value for
of the coherences, but the cross section for Compton scattering
same
as
low as
not very typical of the situation for such low X-ray
For atoms of intermediate or large Z, the cross section for
is
for
it is
is
because
much
the
C since Compton scattering takes place only on essentially
For X-rays of low energy,
free electrons.
this
means
that
Compton
scattering can involve only the electrons of relatively small binding energy
in outer shells of the
atom. As the energy of the X-rays increases, more
and more electrons of the atom become available for Compton scattering.
However, for intermediate energies, Thomson scattering can still occur
from the very tightly bound electrons in inner shells even from the
quantum point of view
because these electrons are inseparably bound to
a very massive nucleus and cannot absorb energy by recoiling, so the
quantum is scattered with no change in energy or wavelength. For
X-rays of energy large compared to the binding energy of the electrons in
atom (~105 ev for 92 U), all the electrons of
the innermost shells of the
the
atom
only
are essentially free in interactions with the X-rays.
Compton
In this limit
scattering can take place.
apparent that the dependence of the scattering cross section for a
atom on the X-ray energy would be very difficult to calculate in
the energy range we are now discussing.! In fact, calculations of the cross
It is
typical
section for
Compton
scattering have been carried through only for a
single completely free electron.
who
1928 by Klein and Nishina,
These calculations were performed in
relativistic theory of quantum
used Dirac's
mechanics, and even the results of the calculations are complicated. They
are
4a2
dan
la
da
cr.t.
a) 2
4a 2
.1
(3(1
a,
-2
O
2a sin'
(14-36)
+
+
sin
(1
cos 2 0)1
2a
sin
"
2a
In (1
2a
2a)
a
1
.(1
+ 3a
+ 2a)
In (1
2
2a)"
(14-37)
2a
where
hv
mc
f
hv
2
0.51
Mev
quantum energy
electron rest mass energy
However, the scattering cross section can be measured. Some data
in section 7.
will
be presented
504
The
X-RAYS
[Ch. 14
Compton scattering from a free electron are plotted
and (14-17) from these two equations. For Compton
of X-rays of quantum energy large compared to the binding
cross sections for
in figures (14-16)
scattering
Figure 14-16.
The
differential scattering cross section for
Compton
scattering from a
free electron.
energy of electrons in the innermost shell of the atom, an approximate
atom can be obtained by setting
estimate of the scattering cross sections per
do Sa
= z da s
(14-38)
(14-39)
JD
dQ.
and
Sm
Za s
where das/dQ, and as are given by equations (14-36) and (14-37).
0.01
0.1
10
100
1000
a
Figure 14-17.
The
As Compton
scattering cross section for
scattering
is
Compton
scattering
from a free electron.
described in terms of quanta,
it is
convenient
in considering this process to express the intensities of the incident
scattered X-rays as fluxes of quanta.
Thus we
and
redefine /to be the average
number of quanta
cm2
AND
PHOTOELECTRIC EFFECT
Sec. 6]
PRODUCTION
505
beam per second
across a
PAIR
carried by the incident
to be the average
number of quanta
scattered per second in all directions,
and redefine 5(@)
to be the average
number of quanta
area oriented normal to the beam, redefine
scattered per second per unit solid
measured from the direction of the incident beam.
from the previous ones only by a factor equal
the energy carried per quantum. Since for Thomson scattering
angle at the angle
These definitions
to hv,
differ
the energy of the scattered
quantum,
may
all
quantum is equal to the energy of the incident
we have written for Thomson scattering
equations involving the new definitions simply
the equations which
be transformed into
by dividing both sides by the factor hv. Of course,
the form of the equations, and we still have
this
does not change
S = o sJ
5(0)
(14-40)
^/
(14-41)
all
where for Thomson scattering the cross sections have exactly the same
values as before. For Compton scattering the cross sections of the previous
paragraph also satisfy equations (14-40) and (14-41) when the new definitions are used.
new definitions allow all scattering
absorption processes) to be discussed in the same
Consequently, these
processes (and also
language and, for
all
this reason,
we
shall henceforth use these definitions, or
equivalent ones, exclusively.
We
i.e.,
close this section
by pointing out that the Klein- Nishina equations,
(14-36) and (14-37),
reduce to the free electron
show
the free electron
Thomson
Compton
cross sections
cross sections in the limit of very low
X-ray quantum energies. Furthermore, equation (3-17) shows the wavelength increase in Compton scattering from a free electron becomes
negligibly small for very long wavelength X-rays because the wavelength
change cannot be larger than 0.048 x 10~ 8 cm. Thus X-rays of very low
quantum energies do not lose energy in Compton scattering. From both
points of view,
Thomson
scattering acts as the
scattering for X-rays incident
6.
The Photoelectric
Effect
upon a
low energy limit of Compton
free electron.
and Pair Production
In section 7, Chapter 3, we presented an elementary discussion of the
photoelectric effect for electromagnetic radiation in the optical frequency
range.
The reader should review
photoelectric effect
is
that discussion because the optical
the limit of the X-ray photoelectric effect as the
X-RAYS
506
quantum energy becomes very small compared
energy,
in the limit hv
i.e.,
In this limit hv
< mc
[Ch. 14
to the electron rest
mass
2
.
comparable to the binding energy of atomic electrons
is
in outer shells, but
it
is
much
very
electrons in inner shells.
Thus the
smaller than the binding energy of
optical photoelectric effect can lead
only to the ejection of optically active electrons.
In contrast, the X-ray
photoelectric effect for very large hv can lead to the ejection of any electron.
The fundamental
Compton
difference between the photoelectric effect
and the
former process the incident quantum is
completely absorbed by the atom, whereas in the latter process the
quantum scatters from an essentially free electron and goes off suffering
only a reduction in energy. We leave it as an exercise for the reader to
demonstrate that in the Compton effect it is not possible to conserve both
total relativistic energy and momentum if an essentially free electron
absorbs all the energy of a quantum. On the other hand, the complete
effect is that in the
absorption of the incident quantum in the photoelectric effect does not
violate
energy-momentum conservation because
electron concerned
static forces to the
is
not essentially
it
plays
the recoiling nucleus can take
enough
the
quantum and
From
is
Since the nuclear mass
ing
in this process the
but instead
is
bound by
to recoil in such a
is
way
as to conserve
very large, and since/?
= y 2ME,
up the required momentum without absorb-
kinetic energy to affect significantly the energy balance
this
atomic
electro-
massive nucleus. This allows the nucleus to participate
in the process, and the role
momentum.
free,
between
the electron.
argument we can
see that the probability of a photoelectric
absorption on a particular atomic electron should increase with decreasing
quantum energy, since for low energy quanta the electron is relatively
more tightly bound to the nucleus and, therefore, it is easier for the nucleus
to absorb the required momentum. However, when the quantum energy
becomes smaller than the binding energy of the electron, the probability
of photoelectric absorption on that electron abruptly drops to zero
because there is no longer enough energy available to eject the electron
from the atom. This behavior can be seen in the measured cross section
for photoelectric absorption by 82 Pb atoms, which is presented in figure
(14-18). The probability for photoelectric absorption is expressed in terms
of a photoelectric cross section per atom, a PE defined such that A, the
average number of quanta absorbed by this process per second per atom in
a beam of / quanta per second per square centimeter, is
,
A = o pE i
(14-42)
an example of the use of a cross section to describe absorption
processes. For the reasons mentioned above, the photoelectric absorption
This
is
PHOTOELECTRIC EFFECT
Sec. 6]
AND
PRODUCTION
PAIR
507
cross section increases rapidly with decreasing
quantum energy but shows
sharp breaks at quantum energies equal to the binding energies of the
electrons in each of the subshells of the atom. These energies are labeled
with the X-ray subshell designations introduced in section
tending to smaller
show
quantum
3.
that the photoelectric absorption cross section continues
tJ
10
mL
-a
1^
-18.
ex-
its
general
Ij
i
10
hv
Figure
Data
energies than those presented in figure (14-18)
10
(ev)
The photoelectric absorption
cross section per
atom
for lead.
with decreasing energy, except for breaks of diminishing severity at
and
subshells. Such data provide a direct measure
of the binding energies of electrons in the various subshells of an atom
rise
the energies of the
and can be used to confirm or supplement the binding energy values obtained from an analysis of the X-ray line spectrum of the atom.
We
not be able to calculate the cross section for photoelectric
Such a calculation requires the theory of quantum electrodynamics, and we can only present some results. The predicted cross
section per atom for photoelectric absorption on a A' shell electron is
shall
absorption.
a pK =<r T 4V'2(l/137) 4
for hv
that,
>
binding energy of
K electron,
in the energy region indicated,
Z s (#wc*/A>') H
but hv
<
shell
mc
2
.
(14-43)
Experiment shows
absorption accounts for
X-RAYS
508
[Ch. 14
and that equation
The very strong Z
dependence of a PEa reflects the fact that a K shell electron becomes
much more tightly coupled to the nucleus with increasing Z, and so the
nucleus finds it much easier to absorb the momentum which it must
almost 90 percent of
(14-43)
in quite
is
all
photoelectric absorptions,
good agreement with the
data.
absorb in a photoelectric process.
For X-rays of quantum energies in excess of two times the electron
rest
mass energy, there is a mechanism in addition to the photoelectric effect
through which the X-rays can be completely absorbed in interaction with
matter. This is the process of pair production, which consists of the complete disappearance of a quantum and the subsequent materialization of
part of its energy hv into the rest mass energy of a pair of electrons. If
hv = 2mc 2 the materialization of the two electrons uses up all the available
energy. In the general case hv > 2mc 2 and the excess energy hv 2mc 2
appears in the form of kinetic energy of the two electrons. One of these
electrons is of the usual variety with mass m and charge e. The other has
the same mass m but has the positive charge +e and is called a. positron.
Thus charge is conserved in the pair production process because the quantum is uncharged and the total charge of the electron and positron is e
,
+e =
0.
Pair production cannot take place in free space because, as the
reader can easily show, the disappearance of a quantum, followed by the
appearance of an electron and positron, cannot conserve both total relativistic energy and momentum. However, the process can take place in the
Coulomb field of an atom because there can be interactions between the pair
of charged particles and the nucleus. These interactions allow the massive
nucleus to absorb whatever momentum
is
required to conserve
without affecting the energy balance. In the next section
more of
we
momentum,
shall describe
and some of the theory, of this extremely interesting
But it seems best to complete our discussion of the attenuation
of X-rays by matter before going into these subjects.
The cross section per atom for the absorption of X-rays by the pair
production process can be calculated only with the aid of the Dirac theory.
the details,
process.
The
results are
pr
FR"
=o T /(a)
8tt
(14-44)
137
where a = hv/mc2 The Z 2 dependence of the pair production cross section
per atom, a PR is a manifestation of the fact that the required momentum
transferring interaction between the nucleus and the pair of charged
.
particles
/(a) has
is
more
much
the
readily achieved for
same form
a graph of /(a) for the
82
atoms of large Z. The function
for all atoms.
We
Pb atom. This shows
present in figure (14-19)
that
aPRa rises slowly from
Sec. 7]
CROSS SECTIONS
AND ATTENUATION
COEFFICIENTS
509
2
1.02 Mev and, for
a value of zero at the threshold energy hv = 2mc
very large hv, approaches a constant value as /(a) approaches
.,
/(oo)}
JK
28, 183
In
A
(14-45)
T7
27
We
cannot write an equation for f(x) itself, as this function is obtained
from a numerical integration. In an incident beam of / quanta per second
12
^^-
10
$.6
*-
^~
rt~)
4
2
i^i
ill
il
1000
100
10
10,000
a
Figure 14-19.
From W.
The
which arises in the theory
Quantum Theory of Radiation, Oxford University
plot of a function
Heitler, The
of pair production.
Press,
London, 1944.
per square centimeter, the average number A of quanta absorbed per
second per atom by the pair production process is related to the cross
section
a PR by the usual
relation
A = o PR
7.
(14-46)
Total Cross Sections and Attenuation Coefficients
We have now discussed all the scattering and absorption processes
which contribute in a significant way to the attenuation of a beam of
X-rays passing through matter. Let us next see how these processes
combine to produce the total attenuation observed in an experiment, such
5. The average number A of
quanta which are either scattered or absorbed per second per atom, in a
beam of / incident quanta per second per square centimeter, can be
evaluated by adding equations (14-40), (14-42), and (14-46):
as that described at the beginning of section
A=
<*sj
where
<*a
peJ + OprJ = oj
as
+
,
,..
<*
PEa
o-pie,
(14-47)
510
X-RAYS
The quantity aa
is
called the total cross section per atom.
fraction of the incident
[Ch. 14
It specifies
beam of X-rays undergoes any of
the
what
beam
attenuating interactions with a given atom.
cross sections
Figure (14-20) shows the
aSa aPEa a PRa and aa for the 82 Pb atom. For 82 Pb, the energy
,
hv
Figure 14-20.
The
(ev)
scattering, photoelectric, pair, and total cross sections per
atom
for lead.
ranges in which each of the three processes makes the most important
contribution to aa are approximately
Photoelectric effect:
Scattering:
Pair production:
Because of the different
hv
5
5
<
< hv <
ev < hv
10 s ev
10
10 5 ev
5
10 6 ev
dependencies of the cross sections for these
which they dominate aa are quite different
processes, the energy ranges in
for
atoms of low atomic number. For 13 A1 these energy ranges are approxi-
mately
Photoelectric effect:
Scattering:
Pair production
hv
<
5
4
10 4 ev
< hv <
ev < hv
10 ev
10 7
10 7 ev
CROSS SECTIONS AND ATTENUATION COEFFICIENTS
Sec. 7]
Using the
/
beam of X-rays
cm, containing
figure (14-21).
The
number of quanta
the
flux
foil.
and
atom aa
total cross section per
attenuation of a
flux (or intensity) of the
crossing
cm2
beam, which
cm
I(x)
the attenuated flux emerging
then
is
the average
beam
per sec, decreases as the
Call the flux after penetrating x
I(t) is
us calculate the total
through a foil of thickness
Consider the situation shown in
in passing
cm3
atoms per
let
511
penetrates
1(0) is the incident
from the
Consider a
foil.
tern
A beam
Figure 14-21.
of X-rays passing through a
foil.
in the lamina is N
where S is the area of the foil
in cm 2 The probability of a quantum being removed from the beam by
one of these atoms is equal to the probability of the quantum striking an
imaginary disk of area aa cm 2 This is equal to aa cm2 /5 cm 2 the ratio of
lamina of thickness dx at
The number of atoms
x.
times the volume of the lamina;
it is
NS dx,
the area of the disk to the area of the lamina.
The
net probability of a
quantum being removed from the beam by any of the atoms in the lamina
is equal to this ratio times the number of atoms; i.e., (ajS)NS dx = a Ndx.
a
As an average of I(x) quanta are incident upon 1 cm 2 of the lamina in 1 sec,
the average number of quanta removed per cm2 per sec from the beam by
the lamina is I(x)aa N dx. The number of quanta per cm 2 per sec emerging
from the lamina, I(x + dx), is equal to the number incident minus the
number removed so
;
I(x
dx)
I(x
dx)
I(x)
I(x)aa Ndx
or
dl(x)
I(x)
~I(x)aa Ndx
(14-48)
X-RAYS
512
We
find I(t)
by integrating
this expression
dl(x)
[Ch. 14
over x:
= aNdx
I(x)
Jo l(x)
[in /(*)]'
Jo
= -aa Nt
7(0
In
-oNt
1(0)
7(0)
7(0
2.0
1.8
1.6
1.4
/^
l(0)e-"
^^_
S*^
1.2
^0.8
b
**>/
0.4
0.2
\\
A1
n
10 5
io 6
The attenuation
io 7
hv
Figure 14-22.
V/v
^^
^^^-^1
\ \\
\
0.6
(14-49)
i. 1.0
10 8
10 s
(ev)
coefficients for several
atoms and
a range of energies.
The flux of the attenuated beam decreases exponentially with increasing
The quantity aa N, which is called the attenuation coefficient, has the
dimensions cm -1 and is the reciprocal of the thickness of foil required to
attenuate the beam by a factor e. This thickness is called the attenuation
t.
length A.
That
is
A=
Of
course, the attenuation coefficient has the
the total cross section.
of
\jaa
82
Pb,
50
Sn,
and
13
(14-50)
same energy dependence as
Figure (14-22) shows the attenuation coefficients
A1 for X-rays of
relatively high energy.
8.
AND OTHER
POSITRONS
Sec. 8]
ANTIPARTICLES
513
Positrons and Other Antiparticles
The
first
experimental evidence for the pair production process, and the
was obtained in 1933 by Anderson during an
existence of positrons,
investigation of the cosmic radiation.
This radiation consists of a flux
of very high energy X-rays and charged particles incident upon the earth
from extra-terrestrial sources. Anderson was using a cloud chamber
(a device filled with supersaturated
formed by charged
vapor that condenses on the ions
particles traversing
it
and makes the paths of these
a
particles visible) containing a thin lead plate, with the entire apparatus in
magnetic
field.
Upon
exposing this apparatus to the cosmic radiation,
it
was found that very infrequently a pair of charged particles were ejected
from some point in the lead plate. These events were assumed to be the
result of the interaction of an X-ray quantum in the lead because no charged
particle was seen to strike the point of ejection, whereas a quantum, being
uncharged, could strike the point of ejection without being seen. The two
charged particles ejected in these events were bent in opposite directions
by the magnetic field. Therefore their charges were of the opposite sign.
From other considerations it could be shown that the magnitudes of these
charges were equal to one electronic charge and that the masses of the
particles
were at
least
the discovery of these
approximately equal to one electronic mass. Since
few electron-positron pairs, countless others have
first
been observed in many different experiments. In these experiments
has been established with great accuracy that the rest mass, charge, spin,
symmetry character, and magnetic moment of the positron and electron
are equal, except that the signs of the charge and magnetic moment of the
it
positron are opposite those of the electron.
The discovery of the pair production process explained the origin of
a discrepancy between the then current theory of X-ray attenuation and
the measured attenuation coefficients of several materials for 2.6 Mev
X-rays
(gamma
rays obtained from a radioactive source).
originally did not include pair production,
coefficients
were too small;
As
the theory
the predicted attenuation
with the inclusion of the pair production
good agreement is now obtained between experiment and theory.
However, the real importance of Anderson's discovery was in the beautiful
confirmation which it provided for Dirac's relativistic theory of quantum
process,
mechanics.
The Dirac theory
leads to the prediction that the allowed values of
total relativistic energy
E for
a free electron are
E = Vc p z +
2
where
is
the electron rest mass.
(mc2f
(14-51)
These are simply the solutions for
X-RAYS
514
[Ch. 14
of equation (1-25). But the solution with the minus sign corresponds to
a negative total relativistic energy a concept just as foreign to relativistic
mechanics as a negative total energy is to classical mechanics. Instead
of just throwing away the negative part of equation (14-51) on the
grounds that it is not physically realistic, Dirac pursued the consequences
of the entire equation. In doing this he was led to some very interesting
conclusions. Consider figure (14-23), which is an energy level diagram
+mc 2
Higher
levels,
Lowest
level,
corresponding to
corresponding to
>
p =
Highest
Lower
Figure 14-23.
The energy
make
level,
levels,
corresponding to
corresponding to
p =
>
levels of a free electron according to Dirac.
representing equation (14-51).
energy levels
If the indicated
exists, all free electrons
transitions into these levels,
continuum of negative
of positive energy should be able to
accompanied by the emission of quanta
of the appropriate energies. This obviously disagrees with experiment
because free electrons are not generally observed to emit spontaneously
2mc 2 However, Dirac pointed out that this diffiquanta of energy hv
culty can be removed by assuming that all the negative energy levels are
normally occupied at all points in space. Then the exclusion principle
would prevent a free electron from dropping into any of these levels.
According to this assumption, a vacuum consists of a sea of electrons in
negative energy levels.
This does not disagree with experiment.
For
assumed
to be
instance, the negative charge could not be detected, as
it is
uniformly distributed and therefore exerts no force on a charged body.
Similar considerations will demonstrate to the reader that all the "usual"
properties of a sea of negative energy electrons are such that
would not be apparent
in
any of the "usual" experiments.
its
presence
However,
POSITRONS
Sec. 8]
Dirac's theory of the
certain
new
AND OTHER
ANTIPARTICLES
515
vacuum is not completely vacuous because it predicts
properties which can be tested
The energy
by experiment.
diagram for a free electron suggests the
level
possibility of
an electron in a negative energy level by the absorption of a
quantum. Since all the negative energy levels are assumed to be occupied,
the electron must be excited to one of the unoccupied positive energy
exciting
The minimum quantum energy required for this process
= 2mc 2 and the process results in the production of an
levels.
ously hv
in a positive energy level plus a hole in a negative energy level.
strate
below that a hole
obvi-
is
electron
We demon-
in a negative energy level has the properties of a
positron of positive energy.
Consequently this is the pair production
process observed experimentally by Anderson three years after its theoretical prediction by Dirac.t
Consider the excitation of an electron from a level of total relativistic
to a level of total relativistic energy
The energy of
1
2
energy
the
+E
quantum required
hv
If
we
= +E - (-EJ = E + E
2
energy
(14-52)
use the symbol e~ to designate the electron
its total relativistic
is
left in
the level
+E
2,
is
E.-
= E2
and, according to equations (1-22') and (1-24),
its
kinetic energy
T - = E - - mc 2
e
is
(14-53)
where m is its rest mass. These relations are illustrated in the energy
level diagram of figure (14-24). Now, there will be a positive total relativistic energy, which we call E +, associated with the absence of an electron
e
in the level of total relativistic energy
two quantities
Ev
The
relation
between these
is
E e+ = -(-,) = E x
From
figure (14-24)
we
E can be decomposed as follows:
E = mc T +
see that
which may be taken as a definition of the positive energy
T +. Thus
e
E e+ = -{-mc 2 - 7~= mc 2 + 7>
or
7>
t In section 6
we mentioned
that
= E e+ - mc
(14-54)
momentum
to take place in the presence of a nucleus.
affect the
conservation requires pair production
However, the massive nucleus does not
energy balance, so an energy level diagram for a free electron can be used to
describe the process.
516
X-RAYS
[Ch. 14
By comparing this equation with (14-53), we identify Te+ as a kinetic
and mc 2 as a positive rest mass energy. There are, then, a positive
energy,
mass
E +,
Te+,
and a
positive
associated with the hole in a negative energy level.
Conse-
total relativistic energy
rest
quently the hole has
all
a positive kinetic energy
the mechanical properties (rest mass, inertia, etc.)
of a particle of rest mass equal to one electronic mass.
there will be a positive charge
+e
apparent that
I i
It is
associated with the absence of an
Eh"
+
i
u
E
o
B
1
1
'
+
'
'
An energy
Figure 14-24.
electron of charge e.
magnetic
viously
is
the
moment
filled
same
level
'
diagram illustrating pair production.
There are also a spin angular
momentum and
associated with the absence of an electron in a pre-
energy
level.
The magnitude of the
spin angular
as for a level containing only a single electron
(cf.
momentum
footnote on
page 485). The magnetic moment is of the same magnitude as for an
electron, but the sign is opposite because the charge associated with the
hole
is
opposite that of an electron.
We
see that in all respects the pro-
perties associated with a hole in a negative energy level are identical with
the experimentally observed properties of a positron, which
we have
designated by the symbol e+.
A positron is a completely stable particle as long as
it is kept away from
However, a positron passing through matter will eventually
collide with an electron and the two particles will annihilate. In terms of
the Dirac theory, this can be thought of as a transition in which the electron
drops into the hole that constitutes the positron, with the liberated energy
emitted in the form of quanta. Annihilation normally occurs when the
electrons.
Ch.
EXERCISES
14]
517
is moving very slowly since then the probability of colliding with
an electron is higher. If annihilation occurs on a free electron, two quanta
must be emitted in opposite directions in order to conserve momentum.
If the electron is bound to a nucleus, one quantum annihilation is possible.
positron
The average time required
for a slow positron to annihilate in
82
Pb
is
~10-10 sec.
The
and positron are but one example of a particle and its
The existence of a number of other examples is predicted
by theory and has been verified by experiment. For instance, the antiproton, a particle of charge e and mass equal to the proton mass M,
was observed in 1955 by Segre and his co-workers. Since the proton rest
mass energy Mc 2 = 931 Mev is very large, antiprotons can be generated
only in the cosmic radiation or in high energy particle accelerators. The
electron
antiparticle.
Segre experiment produced antiprotons in collisions between stationary
protons and protons of 6200 Mev kinetic energy obtained from a large
particle accelerator.
BIBLIOGRAPHY
Compton, A. H., and
S. K. Allison, X-rays in Theory and Experiment, D. Van
Nostrand Co., Princeton, N.J., 1935.
Evans, R. D., The Atomic Nucleus, McGraw-Hill Book Co., New York, 1955.
Richtmyer, F. K., E. H. Kennard, and T. Lauritsen, Introduction to Modern
Physics, McGraw-Hill Book Co., New York, 1955.
EXERCISES
1.
Verify equation (14-24).
2. Verify
3.
equation (14-28).
Prove that a
stated in section
4.
free electron
cannot absorb
all
the energy of a quantum, as
6.
Prove that pair production cannot take place in free space, as stated in
section 6.
5.
Using the data of
to attenuate a
6.
Show
quantum
beam of
figure (14-20), calculate the thickness of
10 4 ev X-rays by a factor of 100.
that the attenuation length
is
Pb required
just equal to the average distance a
will travel before being scattered or absorbed.
7. A slow positron moving through matter can be captured by a free electron,
forming an atomic-like system of transient existence called positronium. Evaluate
the spectrum of electromagnetic radiation emitted in processes of this type.
CHAPTER
15
Collision
Theory
I.
Introduction
We
have been able to go through a
fairly
complete discussion of the
atom in terms of the quantum mechanical theory of the motion of particles
bound in a potential, because most of the understanding of atoms has come
from interpreting experimental measurements of the spectra which they
emit in terms of the energy levels of their bound atomic electrons. For
the nucleus this
is
not the case.
From the very first the understanding of its
come largely from interpreting the results of experiments
in which unbound particles collide with nuclei. As an example, the reader
will recall that the very discovery of the nucleus came from the alpha parproperties has
ticle
scattering experiments.
nucleus,
we must now extend
In preparation for our discussion of the
quantum mechanical
unbound particle with a potential.f
previous chapter we described the processes which occur in the
to three dimensions the
theory of the collision of an
In the
quanta with atoms. This description was
probably not very satisfying to the reader because we were generally able
to state only results. A rigorous treatment of the scattering and absorption
of quanta is not possible at the level of this book as it requires quantum
electrodynamics and relativistic quantum mechanics. In this chapter
we shall, however, consider only the scattering and absorption of particles
and shall ignore spin and other relativistic effects. This will be adequate for
most of our discussion of the nucleus and will allow us to treat the problem
rigorously with the "old familiar" Schroedinger theory. At first we shall
collision of electromagnetic
t Collisions in
one dimension were treated
518
in the first sections of
Chapter
8.
LABORATORY-CENTER OF MASS TRANSFORMATIONS
Sec. 2]
519
we shall give an approxiand then an exact treatment. Finally we
consider scattering. After certain preliminaries,
mate treatment of
this process,
shall treat absorption.
2.
Laboratory-Center of Mass Transformations
In working with the Schroedinger equation for a system of two particles
of mass
and
interacting with each other through a
mutual potential
problem can be reduced from one
involving six spatial coordinates to one involving three by transforming to
a frame of reference in which the center of mass of the system is at rest.
In this center of mass (CM) frame the system can be described in terms of
the motion of a single fictitious particle of reduced mass
1
energy V(r),
we have found
that the
A*
-^^-
(15-1)
ml + m 2
interacting with a center of force fixed at the origin.
The
distance
this particle to the origin is equal to the separation of the
from
particles
x and
2 and the potential energy V(r) of the fictitious particle
equal to the mutual potential energy of the two particles. Furthermore,
of mass
is
two
the total (and kinetic) energy of the fictitious particle in the
CM frame is
equal to the total (and kinetic) energy in that frame of the system of two
particles.f In the
frame the system obtained by replacing one of the
CM
by a particle of mass (x, replacing the other by a particle of such
large mass that it must remain fixed at the origin, but otherwise making no
particles
changes in the system or the equations governing
equivalent to the original system.
work with because one of
its
The
its
is
is
completely
much
easier to
particles remains fixed and, consequently,
can be ignored except as a source of the potential
Now,
behavior,
equivalent system
V(r).
we have never
had to bother with the fact that the
frame is generally moving, since
this motion only adds a constant and non-quantized translational kinetic
energy to the system which does not affect the energy differences that are
observed in measurements of spectra. But, in treating scattering problems,
we cannot ignore the motion of the CM frame because it affects the
observed scattering angles and cross sections. In almost all experiments
these quantities are measured in a frame of reference, called the laboratory
(LAB) frame, which is moving with respect to the CM frame. The reason
is that in almost all scattering experiments one of the particles, and not
in evaluating the energy levels of such a system,
CM
t Cf. section 2,
Chapter
10,
and section
4,
Chapter
5.
COLLISION THEORY
520
the center of mass,
[Ch. I5_
stationary in the laboratory. Let us develop
is initially
the relations between the description of a scattering process in these
frames.
we
In doing this
shall deal
with
momentum
two
vectors of precise
magnitudes and directions however, since the locations of the particles
need not be specified, this will not violate the uncertainty principle.
;
Assume
that the particle of
mass
2 is
initially stationary in the
LAB
of m 1 with respect to m 2 is \ 1L as shown
in figure (15-1). After the collision, m 1 moves off with velocity \'1L at the
frame, and that the
initial velocity
V 1L
Before
Figure 15-1.
LAB
and
between two
collision
particles, as
seen
in
the
LAB
frame.
scattering angle 6 L measured from its initial direction of motion,
In the
frame
2 recoils in some direction with some velocity.
CM
has the different appearance shown in figure (15-2), because
frame is moving with velocity V with respect to the LAB frame.
this process
the
CM
By
differentiating equations (5-16) with respect to time,
and multiplying
each of the resulting velocities by the mass of the associated
CM
frame
mj
vie
particle,
it is
V2c "*2
Before
Figure 15-2.
between two
collision
particles, as
seen
in
the
CM
frame.
CM
frame the total momentum must vanish.f
and w 2 must therefore always be equal and
opposite in this frame. Furthermore, the magnitudes of the momenta
cannot change because energy is conserved. Therefore the result of a
collision is simply to rotate the line along which these vectors are directed
scattering angle 6 C without changing their magnitudes.
through the
easy to see that in the
The momentum
vectors of
CM
Thus
v ic
t This
is
ic
v zc
also true relativistically. Cf. exercise
8,
(15-2)
V 2C
Chapter
1.
LABORATORY-CENTER OF MASS TRANSFORMATIONS
Sec. 2]
An
521
CM
frame is obtained by
by fi, with the initial velocity of fi equal to the relative
velocity of the two particles (because the distance from fi to the origin is
always equal to the distance between m 1 and m), and by letting m 2 -> oo.
equivalent description of the process in the
replacing
This
is
We
indicated in figure (15-3).
v*c
and
it is
have
(15-3)
v il
apparent that
(1^4)
CM
frame
"
v" c
\_
Before
Figure 15-3.
the
CM
To
to the
An
After
equivalent description of a collision between
two
CM
determine the value of V, the velocity of the
frame with respect
LAB frame, we use the classical velocity transformation equations
Vic
v il
~ v
= V
v ic
>
Their use, and the use of velocity independent masses,
we
particles, as seen in
frame.
are not treating relativistic problems.
momentum
vanish in the
CM frame
(15-5)
is
justified
is
m lVlc = m 2V2C
This
because
Now the condition that the total
(15_6)
is
m l vlL = ( m l + m d V
or
V=
To
mi
mx + m 2
find the relation between 6 C
after the collision
form of the
from the
is
(15-7)
and 6L we transform the velocity of
The vectorial
,
CM frame to the LAB frame.
classical velocity transformation
ii = V +
This
v 1L
illustrated in figure (15-4).
equation which
we
use
is
(15-8)
vi c
Taking transverse components, we have
v il sin e L
v ic sin
0c
(15-9)
Longitudinal components give
v'1L
cos 6 L
= V + V1C cos 6 C
(15-10)
COLLISION THEORY
522
[Ch. 15
Dividing equation (15-9) by (15-10), we find
y,V sin
v'ic
sin 6 r
cos
^/"ic
cos 6c
But
1C
or
W"ic
Figure 15-4.
the
LAB
=m
lm 2
Illustrating the transformation of a velocity
vector from the
CM
frame to
frame.
Therefore
sin d c
tan0,
mjrriz
Now
let
(15-11)
cos 6 C
us find the relations between the scattering cross sections and
two frames of reference.
and of the incident and scattered
fluxes, may be directly adapted from equations (14-40) and (14-41) and
the definitions immediately preceding those equations, if we rewrite them
differential
The
scattering cross
sections for the
definitions of the cross sections,
in terms of a single incident particle instead of
That
is,
we
the incident particle will be carried by the
to the
many
incident quanta.
define the incident flux / to be the probability per second that
beam
direction,
and
beam across a
define the scattered fluxes
cm2
S and
area normal
S(0) to be,
respectively, the probability per second that the incident particle will be
and the probability per second that it will be
an angle 6 from the direction of the
quantities
are
related by the scattering cross section
beam.
These
incident
a and the differential scattering cross section dajdD. as follows
scattered into
all
angles,
scattered into a unit solid angle at
al
(15-12)
(15-13)
ail
Sec. 2]
LABORATORY-CENTER OF MASS TRANSFORMATIONS
523
In the present context we simplify the notation by dropping the bar
denoting average, and the subscript S for scattering. We also drop the
subscript a denoting cross sections per atom, as it is no longer necessarily
appropriate.
However, we must add a subscript
C for center
of mass, or
for laboratory, to all these quantities in order to designate the frame of
reference in which they are defined.
do not need to use some other
We
symbol to distinguish between the quantities defined in the two alternative
frame descriptions of the scattering process, since henceforth we shall
CM
be concerned only with the description of a particle of mass
n scattering
from the potential V(r) associated with the center of force fixed at the
origin.
From
the Schroedinger theory
cross sections for this description in the
we
shall calculate the scattering
CM frame.
These cross sections
can then be transformed to the LAB frame, for comparison with experiment, by noting that the probability per second for the incident particle
to be scattered must be the same whether measured in the
frame or the
LAB frame. That is,
CM
Using equation (15-12),
this
becomes
a L IL
a C IC
II
Ic
But
(15-14)
because the motion of the particle /j, relative to the origin before scattering
(which determines Ic ) is identical with the motion of the particle m
x
relative to the particle m 2 before scattering (which determines I ). ThereL
fore
oc
(15-15)
Similarly, the probability per second for the incident particle to be scattered
into the solid angle dD. L
dQ. c measured in the
LAB frame must be the same as
be scattered into the solid angle
measured in the
the probability per second that
it
will
CM frame, if the same physical range of directions
described by both solid angles.
That
S L (8 L) dQL
is,
= S C (6 C) dQ c
so
dtt L
L
dl c
and
da L
dac dQ.n
dl L
dQ. c dQ. L
is
COLLISION THEORY
524
[Ch. 15
This can also be obtained by formally differentiating equation (15-15).
The elements of solid angle have been evaluated in equation (14-25).
They are
dQ. c
dd c
2tt sin d
and
dQ L =
L dd L
2tt sin
so
d&c
sin6 c d0 o
dCl L
As an
sin 6 L
(15-161
dd L
should use (15-11) to evaluate (15-16), and
exercise, the reader
obtain
da L
dQ L
3.
doc
2(m 1 /m 2) cos dgf4
(m 1 /m 2 ) cos C
+ (mjmtf +
[1
dQ c
n 5 _ 17 )
The Born Approximation
In this section
we
shall
develop a method, due to Born, for calculating
the approximate values of dajdD.
and
(The subscript
a.
C is
to be under-
stood because here, and throughout the remainder of this chapter,
sections are defined in the
CM frame.)
The
first
step
is
to give a
all
cross
quantum
mechanical description for the motion of a particle of the incident beam.
Equation (8-4), and the subsequent discussion, show that in one dimension
ah eigenfunction describing a
mass n traveling with velocity
which we call here the z axis, is
free particle of
v in the positive direction along
an
y>(z)
axis,
AeiKz
(15-18)
where
K=
IwjX
lirplh
(15-18')
nv\h
and where A is a constant. The reader may show by substitution that the
wave eigenfunction (15-18) is also a solution to the three dimensional time independent Schroedinger equation for a free particle,
traveling
_^!
1(i
.dx
dV =
<Py)
ip
2
E%p
(15-19)
a?.
~dy~
where
E = />2 /2^ =
h 2 K 2 l2/j,
(15-19')
In three dimensions, equation (15-18) describes a particle which is definitely
known
to be
moving
parallel to the z axis with velocity
coordinates are entirely
of x and
y,
and whose
unknown since rp*{z) ip{z) is
z coordinate is also entirely
y>*(z) y>(z)
A*e-
iKz
iKz
Ae
v,
whose x and y
obviously independent
unknown
A*
since
(15-20)
Sec. 3]
THE BORN APPROXIMATION
Thus the
particle is
525
moving somewhere in a beam, parallel to the z axis,
and longitudinal dimensions. Of course, this is
not physically realistic since all beams are always limited in their transverse dimensions by diaphragms of finite aperture and in their longitudinal
dimensions by the finite length of the apparatus. On the other hand, the
dimensions of real beams are extremely large compared to the characteristic
atomic or nuclear dimensions. Therefore equation (15-18) provides an
of
infinite
transverse
accurate description of the incident particle in the region of importance
where the atomic or nuclear potential which produces the scattering has
any appreciable value.
Figure 155.
The
spatial
dependence of the
real part of an eigenfunction in
box normali-
zation with periodic boundary conditions.
The
unrealistic aspects of equation (15-18) are, however, the origin of
certain problems concerning the normalization of the eigenfunction.
In
Chapter 8, we showed that these problems can always be handled
and can usually be ignored. The present calculation provides an example
of a case in which they cannot be ignored we must use a three dimensional
extension of the technique of box normalization with periodic boundary
section
1,
conditions.
where
is
We
set
A =
(15-21)
the edge length of a very large cubical box surrounding the
region of the scattering potential,
variables to
Ir**
lie
within the box.
and we restrict the range of the spatial
Then the eigenfunction is normalized
because
ip*ip
A* A
= A2 = L
(15-22)
and
ip*ip
dr
=L
\dr
IT 31*
where dris the volume element, and where the integration is now taken only
over the volume of the box. We furthermore demand that the eigenfunction
and its spatial derivative in the direction normal to the wall have the same
values at corresponding points of the opposing walls of the box. The real
(or imaginary) part of y> would then typically have the behavior plotted in
figure (15-5) as a function of
one of the
spatial variables, holding the other
COLLISION THEORY
526
[Ch. IS
two constant. Using periodic boundary conditions, the eigenfunction
be completely periodic, with period L, in
all
three directions.
Its
will
behavior
repeats indefinitely in the adjacent boxes (just as a scene observed within
a cube with mirror walls repeats indefinitely), and we are justified in considering what happens only within a single box that is, in restricting
the ranges of the variables to the box.f
In most cases of physical interest the scattering potential
symmetrical function V(r).
necessary restriction.
It
We assume this
is
the rectangular coordinates x, y,
where
which
K
is
is
Lr ZA eiKz
beam can be
L- 3AeiKrcose
K directed
a vector of magnitude
the direction of the z axis, and where r
to the point
(r,
d,
<f>).
free particle traveling in
is
r,
6,
<j>
instead of
z axis.
This means
written
Z,-e iK-r
(15-23)
along the beam direction,
is a vector from the origin
In this form the normalized eigenfunction for a
some other
y,
where K'
not a
cos
for a free particle of the incident
it is
Define the origin to be at the center
z.
of the potential and the polar axis to be along the
y>
a spherically
then obviously convenient to describe the
incident particle in terms of the spherical coordinates
and f
is
to be true, although
direction can be written
z,-?V K
''
(15-24)
a vector in the direction in question of magnitude equal to the
value of K' appropriate to the reduced mass and velocity of the particle.
The
validity of equation (15-24)
can be
verified
by the same arguments
as were used for (15-23).
Now
consider a particle in the incident
potential V(r).
We
want
the particle will be scattered in
can treat
this as
beam impinging upon
the
to calculate the probability per second that
some direction.
a perturbation problem:
If V(r) is not too strong,
What
is
we
the rate at which a
constant (in time) perturbation V(r) induces transitions from the
initial
eigenstate associated with the free particle eigenfunction (15-23) to a
final
eigenstate associated with a free particle eigenfunction (15-24)?
Since the final eigenstate
is
in
an
essentially continuous range of final
eigenstates because the possible eigenvalues E'
fr
K'2 /2[i are almost
t Another consideration, which serves to justify box normalization with periodic
boundary conditions, is that the set of eigenfunctions satisfying these conditions forms
a complete set of orthogonal functions in terms of which almost any function can be
expanded. Cf. section 1, Chapter 8.
Sec. 3]
THE BORN APPROXIMATION
527
continuously distributed even with box normalization, the answer is
given, approximately, by the three dimensional extension of Golden Rule
No. 2 (equation 9-45). This
is
2n
R K-~^
(15-25)
^'K^K'K/OK'
where we have used the vectors K and K', instead of quantum numbers,
to label the initial and final states, and where v
K K is the matrix element of
the potential taken between these states. That is,
.
k-k
= jiL-X/r-'rWirWdT =
- '>L- 3 jV(ry (K K r rfT
= L~ VKK
3
(15-25')
with
FK' K
=jV(ry'(K K
'
d7
The quantity p K of (1 5-25) is the number of possible eigenstates per unit
energy interval for the particle associated with the final eigenfunction.
As we have employed box normalization with periodic boundary condi,
tions (required because the eigenfunctions appearing in v
K
K must be
normalized), the density of final states p
K will have some finite value
since the boundary conditions impose restrictions on the possible
de Broglie wavelengths. As an example, consider K' parallel to one edge of
,
Then the real (or imaginary) part of y would typically have the
appearance shown in figure (15-5), with the distance d equal to the
de Broglie wavelength X = 2-n\K' The periodic boundary conditions can be
the box.
propagation parallel to one edge of the box only if L contains
exactly an integral number of wavelengths of the traveling waves. Compare this with the case of free particle standing wave eigenfunctions in a
box with impenetrable walls, which we treated in section 6, Chapter 12.
In that case, the boundary conditions demand that y> have nodes at the
satisfied for
For the propagation direction parallel to one edge of
the box, the condition can be satisfied if L contains either an integral
number of wavelengths or a half-integral number of wavelengths. Conwalls of the box.
sequently, in every wavelength or energy interval there are
two times as
allowed wavelengths in the standing wave case as there are in the
traveling wave case. However, for each possible wavelength there are two
separate traveling waves, one propagating in one direction and another
propagating in the opposite direction. The factors of two cancel out, not
many
only for propagation in directions parallel to the edges of the box but also
and the number of possible eigenstates
for propagation in all directions,
COLLISION THEORY
528
per unit energy interval
therefore the
is
same
in
both
use equation (12-37) which, in our present notation,
dE'
cases.
[Ch. 15
Thus we may
is
u 3A T 3 F' 1A dF'
f*
(15-26)
,-f
where
I?
volume of the box
the
V,
and
E'
h 2 K' 2 l2p
Therefore
^L
PK
This
is
'
not quite what we want because
we want p K
within some
ciated with K', whereas
when
hK'
~ 2V/i3 2
that vector lies
>,
f,L*K'
it is
2-n*P
the density of
all states
asso"
the density of states associated with
Now
certain range of directions.
it is
clear that for a spherically symmetrical potential V(r) the scattering angular
distribution will not
depend on the azimuthal angle
appropriate to consider together
all final states
<j>.
Consequently,
it is
associated with vectors
K' whose directions lie anywhere within the angular range 6 to 6 + dd.
The density p K of these states is smaller than p K by a factor equal to
the ratio of the solid angle dQ.
8 to 6
6.
That
2tt sin 6
d6 contained within the range
d6 to the total solid angle 4n contained within the entire range of
is,
Pk'
dQ.
Pk-
^
477
so
Pr
_ fiL K'
=
S7T
Using
this in (15-25),
we have
R K =-
(15-28)
ti
^k l-Kk^k^F
3
877
Now let us
( 15
- 29
/P
calculate the probability per second that the particle in the
initial eigenstate
associated with the vector
normal to the direction of K. This
is
will cross
the incident flux
/,
and
cm2
it
area
can be
evaluated as in the steps preceding equation (8-7) by using the three
dimensional extension of the definition (7-31) of probability flux. Alternatively, we can use (8-8) and equate / to the product of the velocity of
the particle and the probability density. That
I=vy>*y>
is,
(15-30)
THE BORN APPROXIMATION
Sec. 3]
With equations
(15-18')
and (15-22),
Next, divide
RK
becomes
this
Kb
L~
529
(15-31)
by the element of solid angle dQ, to obtain the rate of
transitions per unit solid angle into the final states associated with the
vector K'.
Then we have
the probability per second of scattering into a
unit solid angle at the angle
which
d,
S(d), the scattered flux.
is
S(0)~^^Vk k Fk K
*.
According to equation (15-13), the
da
S(6)_
dQ
But K'
pv'jh
fxv\h
juL~
KhL~
47T
because
K,
(15-32)
differential scattering cross section is
ju
Thus
K'
KK KK
2
/*
equation
(15-4)
shows
v'
v.
Therefore
where
,
ilK - K yt
'
^K K
! V(r)e
dr
(15-33')
with the integration taken over a very large box surrounding the scattering
This is the Born approximation for dafdQ.. Note that the size
potential.
of the box has dropped out since
does not appear in (15-33), and since
come only from the small
region in which V(r) has any appreciable value and therefore the value of
the integral is independent of its limits.t
contributions to the integral in (15-33') will
It is possible to
carry out part of the integration of (15-33') immediately.
Set
X = K
K'
(15-34)
define a set of spherical coordinates r, 0, O with an origin at the center
of the potential and polar axis along the direction of %. (They should not
and
be confused with the spherical coordinates
along the direction of K.) Then
(K
K')
=X
r,
6,
%r cos
and
dr
t
sin
We use this fact in equation (15-35).
dr
d@ dO
<f>
whose polar
axis lies
COLLISION THEORY
530
[Ch. 15
so
>K'K
V(r)e ixrcos
dr
sin
d@
(15-35)
d<D
Jo Jo Jo
VKK =
V(ry* ,C0B
2nr 2
dr
sin
d0
Jo Jo
The
integral can be evaluated
iyj cos
0. The result
by making the change of variable
is
_r no
Jo
^k-k=
K'K
which
Z=
2nr 2 dr
(15-36)
'X r
is
Vkk
K(r)
^^47rr
Figure 15-6.
dr
(15-37)
Jo
xr
illustrating the relation
between the vectors which enter
in
the Born
approximation.
% in terms of the scattering angle
Finally, let us express
6.
Consider the
vector diagram of figure (15-6), which illustrates the relation (15-34).
From
this figure
apparent that
it is
Now
compare the
= 2K sin ((9/2)
results of
(15-38)
our calculation, (15-33), (15-37), and
(15-38), with the equation (14-35) for the differential cross section for
The similarity is no accident
an alternative derivation of the Born approximation
which leads to exactly the same results as we have obtained, but which is
coherent scattering of X-rays by an atom.
because there
is
completely analogous to the derivation of (14-35). This derivation
involves calculating the amplitude of the wave scattered in a given direction
by every volume element in the region of the
phases into account,
added, and the result
4.
Some
all
is
potential.
Taking the
relative
these contributions to the total amplitude are
squared to obtain the total scattering intensity.
Applications of the Born Approximation
Let us evaluate dajdQ. in the Born approximation for two illustrative
cases.
Consider an attractive Gaussian potential
V(r)
= - F e-
(r/is >
(15-39)
SOME APPLICATIONS OF THE BORN APPROXIMATION
Sec. 4]
which
is
illustrated in figure (15-7).
531
We have
= -V
Vkk.
Upon
f
Jo
'e-W*?^4Trr 2 dr
%r
integration this gives
VKK =
A V R 3 e- lA
-{27rf
-v.
Figure 15-7.
An
attractive Gaussian potential.
Substituting into equation (15-33),
we
obtain
da^jjtJ {2n?VlR*e-* R
\2
da
\2ttH
'
or
(15-40)
US'
Figure
15-8.
The
h4
differential
\R
scattering
cross
section
an
for
attractive
Gaussian
potential.
The differential
This
scattering cross section has the
a Gaussian-like function with a
-1
e
times its maximum value at an angle
is
4K 2 R2
form shown in figure
maximum
at 6
0' satisfying
sin2 (6' 12)
0.
(15-8).
It falls
the relation
to
COLLISION THEORY
532
For energies high enough that
KR =
(V'2i,E\K)R
>
1, 0' will
[Ch. 15
be small and
we have
AK2R\d'j2f
or
d'
(15-41)
KR
-V
Figure
An
59.
attractive square well potential.
Consider next an attractive square well potential,
V(r)
-V
r<R
(15-42)
>R
0,
Here
illustrated in figure (15-9).
sin vr
4-
Airr*
dr
Xr
and we obtain upon integration
(xRfi
So
da^,
dQ
V \
\2ttW
16jt2j/2q6
[si n
X*
XR cos xRf
'-*
(xRf
or
{sin
[2KR
sin (0/2)]
- 2KR sin (6/2) cos
[2KR
sin (0/2)]
\2KR
sin (0/2)] }
(15-43)
The form of
this
differential
scattering cross section
figure (15-10).
At
but [ sin
0,
%R
X*
is
indicated
- X R c os %Rfl(x RT =
in
1/9.
SOME APPLICATIONS OF THE BORN APPROXIMATION
Sec. 4]
Consequently da/dQ. has a
increasing angle, reaching
has
its first
maximum
finite
non-zero root. This
at 6
when
zero
first
its
sin
=
%R
0.
It
533
drops with
%R cos %R =
is
%R
4.49
or
2KR sin (0'/2) =
4.49
^-Vb 6 1
Figure 15-10.
The
differential scattering cross section for an attractive
square well
potential.
For high
da/dQ is
energies,
KR >
1, 6'
< 1,
and the value of
6 at the first zero of
4 49
0'~
(15-44)
KR
For a sharp edge
potential, such as (15-42), da/dQ. has the characteristic
behavior of an optical diffraction pattern.
For a potential with a smooth
edge, such as (15-39), the similarity between da/dQ, and an optical diffraction pattern
is,
perhaps, not so obvious because the secondary
are not present (or at least are weak).
wave
maxima
Nevertheless, the scattering of a
function, in passing through a small region in which the potential
changes in any way,
is
completely analogous to the scattering of an
electromagnetic wave, in passing through a small region in which the index
of refraction changes in an equivalent way
cases an angular distribution
concentrated at angles
less
is
(cf.
equation 8-56).
In both
obtained with most of the scattered intensity
than
0,
L = J_A
~ KR
2ttR
This angle decreases with increasing
K (increasing energy of the particle
or increasing frequency of the electromagnetic quanta), and the angular
distribution
The
becomes more strongly peaked forward.
two potentials we have considered
scattering cross sections for the
COLLISION THEORY
534
can be evaluated from their
scattering cross
differential
[Ch. 15
sections
by
calculating
(15-45)
dl
where the
integral
is
taken over
da
all solid
angle.
(This obvious equality
follows from the definitions of the quantities involved.)
We
shall
not
actually carry this out because the important characteristics of the scattering cross sections are easy to see qualitatively.
For both
potentials,
a decreases with increasing K because the angular region in which dajdQ.
has an appreciable value becomes smaller.
In closing,
we must discuss the range of applicability of the Born
The condition of validity of the perturbation theory
approximation.
underlying the approximation
is
(9-36),
which
states that the
amplitude
of the wave function for the scattered particle is small compared to the
amplitude of the wave function for the incident particle. It is also necessary that the free particle wave function for (15-23) be a reasonable
wave function and that the free particle
wave function for (15-24) be a reasonable representation of the scattered
wave function, in the region of the scattering potential where VKK is
representation of the incident
evaluated.
These conditions
incident particle
is
potential, that
if
is,
large
will generally
compared
be true if the energy E of the
magnitude of the scattering
to the
E>\V(r)\,
(15-46)
for all r
is a small perturbation which can
However, (15-46) is neither necessary
nor sufficient. For instance, the Born approximation turns out to be
fairly accurate for the scattering of low energy electrons from atoms for
reasons to be discussed later, and the Born approximation is not accurate
because then the scattering potential
usually produce only a small effect.
in certain cases for the scattering of high energy particles.
to give criteria
more
useful than (15-46), but this
is
It is
possible
complicated and will
not be done here.
5.
Partial
Wave Analysis
In the previous section we developed an approximation method useful
mainly for scattering at high energies. In this section we shall develop a
method which is exact at all energies, but useful mainly for scattering at
low energies. This is the method ofpartial wave analysis, originally applied
in the nineteenth century by Rayleigh to the scattering of sound waves,
PARTIAL
Sec. 5]
WAVE
ANALYSIS
535
and adapted
in 1927 by Faxen and Holtsmark to the scattering of quantum
mechanical waves.
Consider the scattering, by a potential
beam which
V(r),
of the wave function
upon the potential.
The incident wave passes over the potential, and scattered from it is an
outward moving wave. In the immediate vicinity of the potential both
the incident and scattered waves generally are strongly distorted by the
associated with a particle in a
is
incident
O
Figure 15-11.
Illustrating incident
and scattered waves as seen on a scale large compared
to the size of the scattering potential.
potential.
make
In fact, the waves
may become
so tangled that
it is difficult
to
a distinction between the incident and scattered waves.
However,
at large distances from the region in which the potential has any appreciable value these complications are no longer present, and the scattering
has the simple appearance represented in figure (15-11) in terms of a set
of nodal surfaces. If we do not look too closely, we see a plane wave
incident
it.
An
upon
the potential
and a spherical wave scattered outward from
eigenfunction for this can be written
iKx
00
(15^7)
where we have introduced spherical coordinates r, 6,
with origin at the
center of V(r), and z axis in the propagation direction of the incident wave.
The first term is the eigenfunction (15-18) for a plane wave moving in the
direction of increasing z, which represents the incident particle. (Here it
(f>
not be necessary to normalize the eigenfunction.) The second term is
an eigenfunction for a spherical wave moving in the direction of increasing
r, which represents the possible ways in which the scattered particle can
move. The directional dependence of its amplitude is given by f(d);
there can be no
dependence because of the symmetry of the scattering
potential. The 1/V provides for the necessary inverse square law decrease
in the intensity of the scattered wave. Later we shall verify that the second
term is a solution to the time independent Schroedinger equation in the
region r -> oo where V{r) = 0. For the present we shall assume this and
will
<t>
COLLISION THEORY
536
[Ch. 15
evaluate the incident flux, the scattered flux, and the differential scattering
cross section for the eigenfunction (15-47).
In evaluating the fluxes,
we may treat the two terms
of (1 5-47) separately
between the incident and scattered waves
in the regions where the fluxes are measured. This is true because the
incident flux is measured at large negative values of z where the second
term is negligible owing to its 1/r dependence, and because the scattered
flux is measured outside the transverse limits of the incident beam (which
since there are
no
interferences
Thus the incident
are always finite in a real scattering experiment).
is,
where
bility
to
its
hKjfx
is
v(e
iKz
iKz
)*(e
)
ve- iK2eiKz
(15-48)
This
the velocity of the incident particle.
is
the proba-
2
per second that the incident particle crosses a 1 cm area normal
direction of motion. Similarly, the probability per second that the
scattered particle crosses a
m
e
cm2
area normal to
iKr
iKr
its
direction of
r J
motion
is
f(6)\=vf*(d)f(d)
The
flux
in analogy to equation (15-30),
probability per second that
it
crosses
dA cm2
is
dA
But dA/r 2
is
equal,
by
definition, to the
element of solid angle dD., so the
probability per second for scattering into the solid angle dD.
is
vf*{6)f{6)dD
Dividing by dD,
we have
unit solid angle.
This
is
the probability per second for scattering into
S(d), the scattered flux.
S(d)
The
Thus
vf*(6)f{d)
scattered flux can also be evaluated
(15-49)
from the three dimensional extenmethod we have
sion of the definition (7-31) of probability flux, but the
used
is
simpler.
Now
dajdD
so
S(6)II
we have
do
dQ.
vr(8)f(&)
= p{Q)m
(15 _ 50 )
and the calculation of da\dO. is seen to be a matter of calculating f(6).
This is done by matching the incident wave with the scattered wave in
such a
way
as to obtain a solution to the Schroedinger equation, with the
PARTIAL
Sec. 5]
WAVE
scattering potential
preliminary step,
ANALYSIS
V(r), that
we must
537
behaves
find out
first
As a
like (15-47) for r->-oo.
how to decompose
the plane
wave
described by
e
and associated with the
linear
iK
iKr<x*e
momentum
(15-51)
Kh, into a
set
each associated with the orbital angular momentum Vl(l
do
of partial waves,
1) h.
We must
because the Schroedinger equation for a spherically symmetrical
potential yields most easily a set of solutions of definite orbital angular
this
momentum.
Now the
free particle eigenfunction (15-51) is a solution to
the time independent Schroedinger equation (10-15) for the case V(r)
We
have seen that solutions to
y(r, 0,
But, since (15-51) has no
(f>
<f>)
R(r) 0(0) <D(0
dependence,
solutions which are independent of
equation (10-17) for
m = 0,
which
0.
equation have the general form
this
is
it
suffices here to consider
Therefore
<f>.
(<f>)
(15-52)
1.
only
we take the solution to
Then 0(0) is an accept-
able (never divergent) solution to equation (10-23), with the condition
m=
These are the functions 0, o(0). They are called
0.
Legendre polynomials and are written Pj(cos 0). The first few are
(10-24), for
(cos 0)
P^cos
0)
=
=
cos
2
_
P2(cos 6) =
3 cos
_
.
=
P3 (cos 0)
5 cos
0-1
(15-53)
3 cos
The reader should compare these functions with the
dependence of
The Legendre polynomials are normalized to
(10-39) for m = 0.
2/(2/ + 1). This is represented by the first of the equations
cos0=l
fcose=-i
Pi-(cos 0) P,(cos 0) d(cos 0)
97
_l_
,^
(15-54)
/ft
0>
The second equation represents the orthogonality of these functions.
The function R(r) is an acceptable solution to equation (10-27) with
V(r)
0.
That equation
is
H("S) + (-^^}
COLLISION THEORY
538
[Ch. 15
E > 0, so E is not quantized and the quantum number
These solutions are written j\(Kr) and are called spherical
Bessel functions. The first few are listed below.
In the present case
n does not
arise.
UKr)
h(Kr)
sin
Kr
Kr
sin
Kr
cos
Kr
(Krf
Kr
(15-55)
h(Kr)
sin
.(Krf
J s (Kr)
cos
Kr
(Krf
15
(Krf
verify that for r
15
Kr
sin
_(Krf
The reader can
Kr
Kr
.(Krf
J_" cos
Kr
they have the same
Kr
dependence as
(10-45), in agreement with the remarks following (13-24). We shall
Finally,
verify their behavior for r -> oo in the following paragraph.
we see that for a free particle a particular one of the solutions (15-52)
can be written
(15-56)
;,(*>) itfcosfl)
This solution
is
associated with a definite orbital angular
the origin of coordinates, of magnitude Vl(l
in the
form (15-52)
for this case
1) fi.f
about
A general solution
is
aJiKr)
where the a are arbitrary constants.
t
free particle eigenfunction
momentum
P,(cos 6)
Now
(15-57)
the particle described by the
(15-51) has the definite linear
momentum
Kh. But it can have any orbital angular momentum about the origin since
its impact parameter with respect to the origin is completely unknown.
Therefore we certainly cannot describe the particle by a single one of the
eigenfunctions (15-56). However, we can describe it by the linear combination (15-57), since this contains all possible orbital angular momenta,
a general solution to the Schroedinger equation for the case
of a free particle. Thus we can write
and
since
it is
rco.
flJ,(K>) P,(cos 0)
1=0
t
The reason
is
that
it
satisfies the Z, p
eigenvalue equation (10-57).
(15-58)
WAVE
PARTIAL
Sec. 5]
To determine
P r (cos
the a lt
ANALYSIS
we
539
multiply both sides of this
and integrate over the
d) (/(cos 6)
P,.(cos 6)e iKrmse d(cos 6)
J -I
= f
=
aJ^Kr)
According to (15-54),
upon
this picks
P,.(cos 6)
P^cos
0) d(cos 6)
J-l
yielding,
equation by
entire range of cos 6
sum
out of the
the single term for
/',
transposition,
21'
a rj r (Kr)
P,.(cos 6)e
iKr cos '
d(cos 6)
dv
Integrating by parts,
we have
J-K.T cos
21'
a vJv(Kr)
P,.(cos 6)
0'
iKr
-l
ir cos 6
/.
dP r (cos
iKr
6)
d(cos 6)
d(cosd)
du
Integrating by parts again,
21'
avJviKr)
we
obtain
J.K.T cos 6'
'dP^cos
iY(cosfl)-
iKr
d) e
d(cosd)
l
iKr
cos 0'
(iKrf J-i
d(cos 6)
J-i (iKr)
(iKrf2
d(o
d(cos
6f
In this form the determination of a can easily be effected by taking
v
r -* co. Since the a are constants independent of r, the values obtained
l
For r - co, the last two terms of the equation
above will be negligibly small compared to the first term, since they are
both proportional to r~ 2 while the first term is proportional to r~K Neglectwill
be valid everywhere.
ing these terms,
2V
To
we have
a UlK
vj v (Kr)
P .(cosd)eiKrcose
iKr
evaluate this expression,
P,.(cos 6)
which can be
verified
we use
-> co
(15-59)
COS0= -1
the relations
8=1
1,
cos
(-If,
COS0
= -1
by comparison with equations (15-53). Sincef
(-1)''
t This relation
depends upon the equality e"
iv "
cos z
sin
z.
COLLISION THEORY
540
[Ch. 15
equation (15-59) gives
21'
a vj v (Kr)
[e
iKr
iKr
il
e '"e-
-'ir/2
,i(Kr-lV/2)
r-+co
],
iKr
-l(Kr-Vi,im
_> oO
Butf
(15-60)
andf
Ur - )
sin
(15-61)
2i
Therefore this can be written
2
21'
a vj v (Kr)
Kr
v sin(jK>
-
2i'
is
consistent with the behavior as r -* oo of
Note that equation
(1
5-62)
theyV(^") listed in (15-55) since, for
.
Jv(Kr)
Z'tt/2)
sin(Kr
^
-+ co
(15-62)
these functions,
all
J'77/2)
'-
-+ oo
(15-63)
Kr
by neglecting the terms of (15-55) which vanish more
and writing sin {Kr /V/2) = sin {Kr) cos {I'-njl)
(/V/2). Finally, we substitute (15-63) into (15-62) and solve
as can be verified
-1
rapidly than r
cos (Kr) sin
for a v
We
obtain
av
This
true for all
is
Consequently we
and
r.
It is also
may
use
it
(2V
'
l)i
true for all values of the
quantum number.
to evaluate the constants in equation (15-58)
find
00
jKrcose
This
is
= ^ (21 +
1=0
i)i jJ(Kr)
P (cos
the desired decomposition of the plane
(15-64)
6)
wave associated with the
momentum Kr into a set of partial waves,
orbital angular momentum Vl(l + 1) h. The
each associated with
linear
terms j\(Kr) P^cos 6)
specify the radial and angular dependence of the /th partial wave. The
plane wave elkrcos6 is obtained by mixing together the totality of partial
the
waves, giving each an amplitude (21
An
and a phase factor
term (21 + 1) can be
argument. Consider an imaginary
instructive interpretation of the amplitude
obtained from the following classical
t
1)
These relations depend upon the equality e iz
cos z
/'
sin z.
PARTIAL
Sec. 5]
WAVE
ANALYSIS
541
surface, perpendicular to the direction of propagation of the plane wave,
and containing a
set
of circles of radii
d
centered
on the
intercept of the z axis
in figure (15-12).
(15-65)
II
and the
surface.
Classically, the orbital angular
origin of coordinates of a particle
This
is
illustrated
momentum about
moving parallel to the z axis
is
the
equal to
its
Figure 15-12.
in
Illustrating the classical interpretation of the amplitude term which arises
the decomposition of a plane wave into partial waves.
impact parameter d times its linear momentum/) = KH. Thus all particles
passing through a ring of inner radius d and outer radius dl+1 will have
angular momentum between (l(K)Kh
IH and [(/ + \)jK\Kh = (/ + \)H.
In the classical limit, / is large compared to 1 so all particles passing through
t
the ring can be characterized as having the angular
momentum Ih. Now a
which is a member of a uniform beam (one which would be
described by eikrcose) is equally likely to have any impact parameter.
Therefore the probability that it passes through the ring is just proporparticle
tional to
its
area A. This
is
^=^ i-fl = J[(i + D
+
-!2]
A.
A = -k V 2 +
K
2l
[2/ + 1]
n=
K
2
(15-66)
or
a:
(21+
1)
So (2/ + 1) is the relative probability that a particle moving in a uniform beam will, in the classical limit, have an orbital angular momentum
about the origin of IH, which is the orbital angular momentum associated
COLLISION THEORY
542
quantum number
with the partial wave of
V7(/
1)
[Ch. 15
in the classical limit
where
h -> m.
Let us continue with the problem of determining the function f(8) in
the expression
rp(r,
6)
iKrmse
+f(B) -
iKr
-* oo
(15-67)
which, according to (15-47), describes the eigenfunction for the scattering
of a particle by the potential V(r), at positions far from the region where
V(r) has
any appreciable value. Now,
having no
(f>
From
for the potential V(r).
form
is
similar to the
the discussion above,
form of the solution (15-57)
V(r, 6)
The angular dependence
= |
of constants
because the
is
R (Kr)
t
different,
b.RlKr) P
exactly the
is
equation (10-23) which governs
set
at all positions,
a solution,
evident that
0.
and the
its
is,
(15-68)
(cos 6)
same because the
dependence
radial
That
differential
is
But the
different too
satisfy (10-27), i.e.,
dr!
dr \
it is
for V(r)
never contains the potential.
it
12 Uf) + [- !H!>
+
2
r
8) is
rp(r,
dependence, to the time independent Schroedinger equation
| [ - vtdR =
ft
(15-69)
= which is satisfied by the j\(Kr).
However, the Ri(Kr) are essentially different from the j\(Kr) only for
0. For large r, V(r) = 0, and both R^Kr) and
small r, where V{r)
ji(Kr) satisfy the same differential equation. Thus, for large r these functions are-essentially the same and, in particular, for r co the R {Kr)
have the form
instead of this equation with V(r)
>-
R
->
, .
(Kr)
sin
(Kr
\n\2
'
d,)
-,
r -- oo
/1C _.
(15-70)
Kr
R (Kr)
can at most differ from the corresponding
by a phase shift d which arises from the different r dependence of the two functions in the region where V(r) ^ 0.
It is not difficult to prove this by showing that the functions (15-70) are
For
oo,
each of the
ji{Kr) of equation (15-63)
solutions to (15-69) for
r-*
with increasing r faster than
co,
\jr,
providing only that V(r) approaches zero
but we shall not do
this here.
Instead,
shortly we shall study some graphs of j (Kr) and R {Kr), for typical potentials V(r), which will make it apparent that (15-70) is correct, and will
t
also
show the
significance of the phase shifts.
WAVE
PARTIAL
Sec. 5]
By equating
ANALYSIS
543
we
obtain
a iKr
2 (21 +
l)i
and (15-68), and
the right sides of equations (15-67)
writing eikrcose in terms of (15-64),
MKr) P
(cos 0)
oo
= MjC&) /'.(cos 0),
=
+f(6)
oo
Evaluating ^(Ar) for r-^ oo from (15-63), and .R/Ar) in the same limit
from
(15-70), gives
| (2/ +
sin(Xr-M2)
i),.
Kr
+ f{Q)
Q)
Kei
Kr
^^ ^(Kr-Kr
1.12
6Q
oo
This equation determines both/(0), and the entire set of b lf in terms of the
phase shifts. To see this, we use (15-61), and a similar equation, to write
the sine functions as complex exponentials.
Then, after multiplying
through by Kr, we have
~
2 (2i +
1=0
iy
i(.Kr-ln/2)
-i(Kr-lirl2Y
P,(cos0)+/(0)KViKr
2x
'gHKr-lirlZ + dd
lb
1=0
(21
l)i
P (cos
- n!r / 2
coefficients
bi
oo
0),
2i
+ 2 _
L
-HKr-lvl2 + diY
li
Grouping together the
t
.
of eiKr and e~ iKr gives
P (cos6)+f(d)K\e iKr
-ii7r/2 iSi
2i
(21
l)i
bi
eillr/2
2i
ii,i2
-i,
P (cos
;
2i
0) } e
- iKr
0,
oo
This can be satisfied for arbitrary r only if the coefficients of both eiEr
and e~ iKr vanish separately. The latter condition gives
P,(cos 0)
1
24
Since this must be true independent of the value of
each P^cos
0)
must be zero. For a
(21
2x
1)
typical
jj^/g
_ i en W
li
0,
the term multiplying
this requires
j2
-u
li
or
= (21 +
1)<V*'
(15-71)
COLLISION THEORY
544
Substituting this into the coefficient of e iEr ,
coefficient vanish,
and demanding that the
we obtain
P (cos
1
[Ch. 15
2i
0)
-f(6)K
2i
or
ffl
which
= r^
2(2/
2<a
l)(e
l)e"' sin 3,P,(cos 6)
'
l)P,(cos 6)
(15-72)
is
/(0)
= 7:
2(2/
(15-72')
where we have used (15-60) and an equation similar to (15-61). Since
the b are constants, they have the values specified by (15-71) everywhere,
even though that equation was obtained at r -* 00. A similar statement
-* 00.
is not necessary for (15-72') as the function
f(6) is defined only at r
t
Having the expression (15-72') for f(d), the scattering cross
can be obtained immediately. According to (15-50), da\di
sections
is
^=/*(0)/(0)
ail
= 4 2 (2/' +
K U'=o
l)e~
iS >-
| (2/ +
sin d v P,.(cos
0)]f
'
l)e
iS
>
sin
<5
P (cos 0)]
u=o
(15-73)
where we have used
to be
summed
a
/'
and
to indicate specifically that the
independently. According to (15-45), a
f
dQ
dQ.
!
J
dQ
2tt sin 6
dB
2n
*
\
J-i
two
series are
is
dQ
d(cos 0)
is evaluated by multiplying the two series of (15-73) and integrating
each of the resulting terms over cos 0. By applying (15-54), the reader
This
may
verify that all terms for
S
K 2
j
We
(2/
=o
I)
/'
sin
vanish, and that the result
=I
-41=0
2/
*,
see that the scattering cross sections
(2/
1) sin
is
3, (15-74)
da/dQ and a are completely
determined by the value of
and the values of the phase shifts d
The phase shifts are evaluated by solving equation (15-69) for each
t
and comparing the phase of R^Kr), at some large value of r, with the phase
of j\(Kr) at the same value of r. This is illustrated in figure (15-13) for a
typical / and three different forms of the scattering potential V(r). In
these curves the function
u,(Kr)
is
plotted versus
r.
This
is
r^
DfI
rR
(Kr)
t
,
,n
(15-75)
convenient because the amplitude of u (Kr)
t
PARTIAL
Sec. 5]
behaves like
including
rr
= 0.
WAVE
l+1
,
ANALYSIS
545
as r -* 0, so that
The reason
is
R (Kr)
t
it
goes to zero at r
cc r\ as r
0, in
for all
common
/,
with
the radial dependence of the eigenfunctions for any spherically symmetrical
potential
[cf.
(10-45)
and the remarks following
(13-24)].
In addition,
the function u (A>) satisfies a differential equation
(
dr 2
No
{-^+![*-nr)]}-0
(15-76)
potential
Attractive potential
<-*,
-1
/?
R
Repulsive potential
*|
r*
The
Figure 15-13.
radial
dependence of an eigenfunction for three different forms of
scattering potential.
which has a simpler form than the equation (15-69) for R {Kr). [The
equivalence of (15-69) and (15-76) can be verified by substituting (15-75)
t
into the latter.] In figure (15-13) the top curve
no
potential: V(r)
any
0, all
r.
Here
shows u (Kr) for the case of
t
R (Kr) = j (Kr), there is no phase
t
shift
and the phase of u (Kr) at large r provides the zero of phase
with respect to which the phase shifts that arise when V(r) ^
can be
measured. The middle curve shows u (Kr) for the case of a potential:
V{r) < 0, r < R and V(r) = 0, r > R. With such an attractive potential,
for
/,
[E
V(r)]
> E in the region of the potential and,
according to equation
be larger in that region than
it is in the case of zero potential.
Consequently, u^Kr) oscillates more
rapidly for r < R. For r
R, its behavior is unchanged except that its
phase is shifted. We see from the curve that: With an attractive potential
u (Kr) is "pulled in" by its behavior at r
R, its phase is advanced, and
(15-76), the
magnitude of (Pu^Kr^dr 2
will
>
<
COLLISION THEORY
546
[Ch. 15
The bottom curve shows u (Kr) for the case
> 0, r < R and V{r) = 0, r > R. Here
[E V(r)] < E in the region of the potential, and the magnitude of
d^u^Kr^jdr 2 is decreased in this region. The result is that: With a repulsive
potential u (Kr) is "pushed out," its phase is retarded, and the phase shift
the phase shift
is
positive.^
of a repulsive potential
V{r)
negative.^
is
Some
6.
Applications of Partial
In this section
we
Wave Analysis
shall calculate the scattering cross sections for
some examples of
quite different potentials in order to provide
wave
and to
analysis,
illustrate further the significance
An
of phase
shifts.
AAA
Mil
i/u V
Figure 15-14.
two
partial
DvUI
V u
eigenfunction and the scattering potential for the scattering of hard
spheres.
Consider
by
first
the scattering, in the classical limit, of a small hard sphere
a large hard sphere (the scattering of a ball bearing
The
by a
potential V(r) can be represented as an infinite repulsion
billiard ball).
commencing
R, where the interaction radius R is the sum of the radii of the two
R is the same as at
spheres. The boundary condition for the u (Kr) at r
at r
the edge of an infinite square well potential
it
demands
that each u t (Kr) vanish at r
(cf.
section
R.
The
typical u t (Kr), are plotted in figure (15-14).
interaction radius
length X
is
5,
Chapter
and
and a
8),
potential V(r),
In the classical limit, the
extremely large compared to the de Broglie wave-
2tt/K, so
R > IttJK
t
The phase
potential, only
shift greater
an attractive potential, and negative for a repulsive
not so strong as to make the magnitude of the phase
shift is positive for
if
the potential
is
than 180. Cf. section
6.
SOME APPLICATIONS OF PARTIAL WAVE ANALYSIS
Sec. 6]
547
or
KR >
and m (A>) has been pushed out by a
;
clear, then, that the resulting
value in
its
phase
(15-77)
large
number of wavelengths.
It is
of u (Kr) might well have any
shift d t
range of definition 180 to +180, and that for similar
randomly distributed over this range.
However, the statement is not true for all u (Kr). For values of / greater
than a certain /max u (Kr) is not significantly changed at any r by the
presence of the potential V{r), and the phase shift d is essentially zero.
The reason is u (Kr) = rj\(Kr) in the absence of the potential, and it is a
property of the spherical Bessel functions ji{Kr) that their values are very
u,(Kr) the values of d l will be
small for
r less
than r max
the location of their
Ijk,
first
maximum.
[This
property can be verified for the j,(Kr) of (15-55).] For sufficiently large
r max
its
/
>
-K>
presence cannot produce any significant change.
raax
/,
= rj\(Kr) is negligible in the region of the potential, and
so u i(Kr)
good estimate of
can be obtained by setting
~R
IK~
r
'max
lmax
~KR
/
'max/ JV
"
or
The same
(15-78)
estimate can be obtained from the classical argument connected
with equation (15-65) by saying that a particle of angular
momentum
greater than /max /r will have an impact parameter greater than lma,jK, and
thus, if Ima
jR, its impact parameter will be greater than the radius
jK =
Classically, such a particle would not hit the potential
motion could not be affected by its presence. Quantum mechanically, the potential also has no appreciable effect and the phase shift is
negligible for the function u^Kr) of corresponding angular momentum.
For the problem at hand, it happens to be particularly easy to calculate
an accurate approximation to the scattering cross section,
of the potential.
and
its
<7
We
sum
set
<5;
for
cuts off at
>
= ^2(2>+l)sin
max and
/max ,
be random for
let d,
and we may replace
sin 2
<5,
by
<
its
/max .
Then
the
average value
J.
This gives
4_
Jmax
o~
T(2/+l)
K*2i=o
Furthermore, since / max
we may approximate
a
~ 2H P
K
Jo
is
large
by an
it
many
maX
(2 /
terms contribute to the sum, and
integral over
1)
di
~ "kL
K
dl.
Thus
[""21
Jo
dl
^I^t
K2
COLLISION THEORY
548
Using (15-78),
this
[Ch. 15
becomes
a 2i 2ttK2 R 2 /K 2
or
The
~ 2nR
(15-79)
differential scattering cross section is,
calculate as
however, not so easy to
necessary to determine the exact behavior of the d for
it is
values of
in the
neighborhood of /max
This
is
because dajdQ, involves
.SIS
8-
Figure 15-15.
The
differential scattering cross section for the scattering of hard spheres.
interferences between terms of different
sensitive to the values of d t
We
calculation in figure (15-15).
/ (cf.
equation 15-73) and
The
differential
larger angles dajdQ.
is
very
scattering cross section
contains a strong peak at very small angles, with a width
At
is
therefore present only the results of the
The
isotropic.
6'
~ IjKR.
integral (do/dQ) dQ. over the
region of the peak has the value ttR 2 and the integral over the isotropic
,
region has the same value.
The peak
represents a diffraction effect which
It cannot be
immeasurably small;
the theory predicts to be present, even in the classical limit.
observed experimentally in
this limit
because
6' is
measurement of the scattering cross section yields the projected
geometrical cross sectional area ttR 2 and not 2ttR 2 because the diffraction
scattering is missed. However, the diffraction scattering can be observed
in experiments involving the scattering of nuclear particles by nuclei,
for which 6' is not so small. Another example is the scattering of electromagnetic radiation from a reflecting sphere of radius R large compared to
XjR is too small. The differenthe wavelength, but not so large that 8'
tial scattering cross section is found to have the form shown in figure
2
Half
(15-15) and the scattering cross section is found to be a = 2ttR
of this arises from the reflection of radiation hitting the sphere, and half
a
classical
SOME APPLICATIONS OF PARTIAL WAVE ANALYSIS
Sec. 6]
549
from the diffraction of radiation, passing near the sphere, into the
shadow on its dark side.
Partial wave analysis provides an exact procedure for solving scattering
problems at all energies. However, its application at high energies is
arises
K is
usually a very formidable task because
is
the value of r
large, so
KR >
beyond which the scattering potential V(r)
1,
is
where
negligible.
This means that the expression (15-73) for dajdQ. will contain contributions
KR there will be an
/ through /max
At high energies it is generally much easier to use
the Born approximation, and this approximation is often of sufficient
accuracy. The real utility of partial wave analysis is found at low energies
where KR is not large. In this region the Born approximation normally
cannot be applied because it is very inaccurate, but partial wave analysis
is both exact and easy to apply. We shall illustrate this point by calculating
the scattering cross sections for a low energy particle, incident upon a
from many terms,
appreciable phase
since for all
shift.
simple potential, in the limit
Equation (15-78)
KR <
appreciable in the limit
spectroscopy, this
is
KR <
us that only the phase shift for
tells
called
will
be
In terminology adopted from atomic
S-wave scattering. For this case, the expressions
(15-73) and (15-74) assume the simple forms
lZ = hi
dQ
K*
{'" sin 8
o
po(cos
B)}{e
id
sin d
(cos 0)}
or
da
d~n~
sin
2
<5
KR < 1
(15-80)
KR<1
(15-81)
'
and
a
because sin d
1.
We
is
^Jo,
K
essentially zero for
see that
da/dQ
times dajdQ, in the limit
is
isotropic,
1, 2, 3,
and a
...
is
and because
(cos 6)
the total solid angle
4n
KR < 1
Let us calculate, in this
limit,
a for the attractive square well potential,
V(r)=-
V>
<R
(15-82)
which is plotted in figure (15-16). The phase shift <5 that this potential
produces is evaluated by finding a solution to equation (15-76) for / =
which goes to zero at r = O.f The equation is
i}L
dr*
t Cf. equation (15-75).
It [
2t
h
v{ry\u
(15-83)
COLLISION THEORY
550
In the region r
<
R,
[Ch. 15
can be written
it
d 2u
dr 2
+ Ku =
(15-84)
where
K = V2M (E + V
The general
solution to this familiar equation
= A sin KgT + B cos K
(15-84')
)/ti
is
r,
<R
Figure 15-16.
To
An
attractive square well potential.
obtain the required behavior at
u
In the region r
>
=A
sin
0,
we
r
r,
B=
set
and take
<R
(15-85)
R, equation (15-83) can be written
^L + K 2 u=0
2
(15-86)
dr
where
K=
The general
solution to this
is
= C sin Kr + D cos Kr,
Both terms are needed
dujdr be continuous at
(15-86')
y/2/lElh
r> R
(15-87)
in order to satisfy the requirement that u
r
R.
Before applying these requirements,
and
it is
convenient to rewrite (15-87) as
u
/"sin (Kr
),
r> R
(15-88)
The equivalence of (15-87) and (15-88) can be seen by expanding the
latter to obtain u = Fsin Kr cos d -\- F cos Kr sin d
and setting F cos
= C, F sin = D. Furthermore, by considering the function R (Kr)
= u (Kr)/Kr, where u (Kr) is given by (15-88), and comparing this at
of (15-88)
r-> oo with (15-70) for / = 0, it is seen that the quantity
(5
<5
<5
is
precisely the
S-wave phase
shift.
Now
apply the requirement r that
SOME APPLICATIONS OF PARTIAL WAVE ANALYSIS
Sec. 6]
(15-85) and (15-88) give the same value of u at
A
The requirement
at r
cos
K R = FK cos (KR +
Dividing (15-89) by (15-90),
R. This yields
(1
5-89)
and (15-88) give the same value of du/dr
that (15-85)
R yields
AK
K R = F sin (KR +
sin
551
(1
5-90)
we obtain
tan
KR=
tan
(KR
<5
or
K
IS
tan"
Ka R = KR +
tan
so
d
Knowing
the phase
tan
shift,
~K
tan
we can
KR
(15-91)
calculate the scattering cross section
from equation (15-81). The calculation is simple if we use the
conditions
KR < 1 and K/K < 1, in the limit of very small E, and assume that
K R does not happen to have such a value that tan K R > 1. Then
(15-91) becomes
S
~ tan K R - KR~ sin
Kn
(S
or
;mS
^KRhKnKR R -l)
and (15-81)
gives
This approximate equation predicts a -> oo
for K R = tt/2, ln\2,
where tan K R -> oo. Of course, the equation
cannot be trusted for these
K R since they violate the assumption from which it was derived. The exact
equation (15-91) shows that for these
R the phase shifts <5 = K R and,
from
(15-81),
very large but
Note
very
we
see that for these
finite
values
that this can be written a
much
larger than
nR\
<5
the scattering cross sections have the
[4/(KRf]nR^ so
47r/X 2 -
(15-93)
that, since
KR <
1,
is
the projected geometrical cross sectional area
of the scattering potential. When a takes on
its maximum possible value
for S-wave scattering (15-93), the
scattering cross section is said to be at
an S-wave resonance. Resonances for other
partial
waves occur
if
E is
COLLISION THEORY
552
high enough that there are large phase
when dt
7r/2, 3tt/2,
shifts for /
>
0.
For
[Ch. 15
instance,
the scattering cross section takes on a particularly
and is said to be at a P-wave resonance.
The approximate equation (15-92) does correctly predict
large value
KqR which
that a -*
for
K R = K R.
For these
KgR, the S-wave phase shift is d = n,2ir,
,t and we find directly from
(15-81) that a 0. This behavior of the scattering cross section can be
seen in the Ramsauer effect, which is the term used to describe the extremely
are solutions to the equation tan
.
Figure 15-17.
The
radial
dependence of an eigenfunction for two attractive square well
potentials.
small value of a measured in the scattering of electrons from atoms of a
E of about 1 ev. At low energies only the S-wave
by the quite sharply defined attractive potential that a noble
gas atom exerts on a passing electron, and at about 1 ev the S-wave is
pulled in by just such an amount that d c n and a
0.
Figure (15-17) shows plots of the function u (Kr) for an attractive
square well potential of radius R which produces a phase shift <5 = tt/2,
and for a deeper attractive square well potential of the same radius which
produces a phase shift d = n. Note that this figure, and figure (15-13),
show that the amplitude of u^Kr) is generally smaller in the region r < R
than in the region r > R, but that the amplitude is equal in the two regions
when the partial wave is in resonance. It would be instructive for the
reader to compare this, and the other properties of three dimensional
scattering potentials, with the properties of one dimensional scattering
potentials discussed in sections 3 and 4 of Chapter 8. It would also be
instructive to compare the condition K R = tt\2, for the lowest energy
S-wave resonance in the potential (15-82), with the condition that a new
noble gas at an energy
is
affected
7=0 state is just bound in that potential.
t 8
is
not of physical interest because
it
This can be done by using the
corresponds to no scattering potential.
SOME APPLICATIONS OF PARTIAL WAVE ANALYSIS
Sec. 6]
553
solution to the Schroedinger equation for the potential (8-61), since
same form
exactly the
/
state in the potential (15-82).
show
it
as the Schroedinger equation (15-83) for a
when
that
also has
higher
The
the potential binds an
results of this
has
comparison
will
near zero total energy
state
an S-wave resonance near that energy. This
values and for potentials of any shape.
it
bound
is
also true for
The reason
for this
At a low energy resonance the eigenfunction
has a particularly large value in the region of the potential, and this is
exactly the condition that characterizes a bound state.
Furthermore,
relation
is
easy to understand.
since for a high energy resonance the eigenfunction also has a particularly
large value in the region of the potential, the eigenfunction for such a case
has a certain analogy to the eigenfunction for a bound
reason
it is
state.
For
this
often said that a potential has virtual states, or virtual levels,
which it exhibits resonances, even though these energies
and correspond to situations in which the particle is actually
at all energies for
are
all
positive
unbound.
For KR < 1, only the S-wave is affected by the scattering potential,
and do/dCl is isotropic. With increasing KR, partial waves of higher
angular momentum are affected; for KR = 1 the P-wave will be significantly affected and the D-wave will be slightly affected. When these
partial waves begin to contribute to the scattering, equation (15-73)
shows that do/dO. will no longer be isotropic. As more and more partial
waves contribute to da/dQ, its structure becomes more and more pronounced. This can be seen by comparing the isotropic do/dO., which
we have just obtained for an attractive square well in the limit KR < 1,
with the highly non-isotropic da/dD. for such a potential in the high energy
limit
4.
KR >
The
1,
which we obtained by the Born approximation in section
differential scattering cross section do/dO,
KR, but
not only exhibits more
becomes more sensitive to the
exact form of the scattering potential V(r). As an example, compare the
complete independence of the structure of the isotropic da/dQ. on the form
of V(r) in the limit KR < 1 with the strong dependence shown in the limit
KR > 1 for the two potentials treated in section 4.
This situation simply reflects the fact that the details of the form of the
structure with increasing
it
also
scattering potential can be important only if the de Broglie wavelength of
the scattered particle, X
size
2tt]k,
is
small enough, in comparison with the
of the potential, R, to allow the particle to resolve these details. These
conclusions are of great practical importance in nuclear physics where
form of V{r) is not known a priori but must be inferred from analyzing
measurements of dajdD.. To study the details of V(r), it is necessary to
measure da/dQ for high energy particles. The analysis of the measurements is usually quite difficult (unless the energy is so high that the Born
the
COLLISION THEORY
554
[Ch. 15
because many partial waves enter
However, it is not impossiblethanks to high
speed electronic computing machines.
We end this discussion by mentioning a very important scattering
potential which cannot directly be treated by the method we havedeveloped.
approximation
is
sufficiently accurate),
into equation (15-73).
This
is
the
Coulomb
potential V{r) oc
\jr.
As
this potential
does not
approach zero with increasing r faster than 1/r, the expression (15-70)
is not valid.
It is possible to modify the method in such a way that the
potential can be treated, but because this is complicated we shall only
present the results. It is found that for a particle of reduced mass //
and initial velocity v, scattering from a Coulomb potential V(r) = zZe2 jr,
the differential scattering cross section
do
dQ.
is
(*%-*
\2uvV smUC'^
(,5-94)
y
'
\2/uvV sin 4 (6/2)
Comparison with equation (4-18), and a moment's consideration, will
demonstrate that these results of the quantum mechanical theory are
identical with the results of Rutherford's classical theory of the scattering
by a Coulomb
potential.
This remarkable situation
is
true only for a
potential corresponding to an inverse square law of force,
in the following
way.
From
dimensional analysis,
it
and
it
arises
can be shown that
if
on a particle varies according to r n then the scattering
cross section must vary according to h i+2n
For the inverse square law
n = 2, the scattering cross section is independent of the value of
Planck's constant h, and this requires correspondence between the results
of the classical and quantum mechanical calculations of the scattering
cross section. In fact, for the Coulomb potential even the Born approximation yields a dajdD. which is identical with (15-94), as the reader may
verify with little difficulty. This is why the Born approximation is fairly
accurate for the scattering of electrons from atoms at energies as low as
^100 ev. The approximation is good in the outer shells of the atom
because there the energy of the scattering electron is large compared to the
non-Coulombic potential; it is good in the inner shells, although the
energy is not large compared to the potential, because the potential is
the force exerted
essentially
7.
Coulombic.
Absorption
The theory we have developed up to this point treats only the scattering
produced by the potential V{r) representing the interaction between the
incident particle and the target particle. But there are also processes in
ABSORPTION
Sec. 7]
555
which the incident particle is absorbed by the target particle. An example in
atomic physics would be the capture of an incident electron by a target
atom, leading to the formation of a stable negative ion or the re-emission
of an electron of reduced energy. In nuclear physics there are many
examples of the capture of an incident nuclear particle by a target nucleus,
some nuclear reaction. In this section we shall outline the
procedures used for treating absorption processes in the Schroedinger
theory. It is to be understood that we use the word "absorption" to mean
leading to
all
no energy
processes except scattering with
in the
loss in the
CM frame (elastic
in particular, absorption includes scattering with energy loss
scattering) ;
CM frame (inelastic scattering).
as throughout this chapter
The word
"scattering"
is
used here,
and the remainder of the book, to mean only
CM
no energy loss in the
frame.f
have seen in equation (15-47) that the amplitude of the radially
outward traveling waves, scattered at the angle 6, is proportional to the
scattering with
We
quantity f(d). According to (15-72), this
M=^I
is
(2/
2iK i=o
lXf,
- W(cos 0)
(15-95)
where we write
r\%
2i6 >
(15-96)
Equations (15-68) and (15-70) show that the total eigenfunction, describing
the incident plane waves plus the scattered waves, can be written as radial
standing waves. These radial standing waves can be decomposed into
radially inward and radially outward traveling waves. In doing this, it is
found that the rj of the relation (1 5-96) specify the amplitudes and phases
of the radially outward traveling waves relative to the radially inward
traveling waves. For instance, when r] = 1 or d =
for all /, the inward
and outward traveling waves have the same amplitudes and phases, and
they combine to give a total eigenfunction describing incident plane waves
plus no scattered waves. If r] = e 2ie = 1 for some /, because d
0,
the inward and outward traveling waves have the same amplitudes but
different phases, and they combine to give a total eigenfunction describing
incident plane waves plus some scattered waves.
l
'
To
allow for absorption as well as scattering, the relation (15-96)
is
broadened to read
= A ie \
A, <
(15-97)
If A <
for some the amplitude of the associated
2i
tj,
where the A are real.
1
/,
outward traveling wave, which is proportional to
t
x,
will
be smaller than
Note that in the LAB frame there is always at least a small energy loss in
due to the
recoil
of the struck
particle.
'
'scattering"
COLLISION THEORY
556
[Ch. 15
the amplitude of the associated inward traveling wave, which is proporThen there is a net inward probability flux, proportional to
tional to 1
.
the difference between the inward and outward intensity, i.e., proportional
1
This implies absorption. The absorption cross
r\*r\i.
to 1
Af
a A can be evaluated in terms of the r) by using the three dimensional
extension of the definition of probability flux (7-31) to calculate the net
section
inward flux integrated over all angles. The scattering cross section as
can be evaluated just as we have done before. The results are
It is
TiI&+W-riivd
1=0
OA
= ir,lW +
- vd
tfXi
easy to verify that (15-99) reduces to (15-74),
total cross section
aT
is
Using (15-98), (15-99), and a
oT
=f*(6)f(d)
little
I (21
fj
K i=o
The differential scattering
from (15-50) and (15-95).
~dh2
if
~98 )
(15-99)
A =
x
for all
/.
The
defined as
aT
do5
iXi
( 15
as
aA
we
algebra,
(15-100)
1) (l
find
dasldQ.
cross section
(15-101)
LI
is
obtained immediately
It is
= -L ( | (2 +
4K \i'=o
/'
1)07*
l)P,.(cos 0))
I
X (f (21
l)(m
- IWcos 6))
(15-102)
Differential absorption or total cross sections are not defined.
These equations show that the absorption and scattering cross sections
are not independent.
If there is absorption there
because, in order to have
for
all
not
true.
it is
/,
and
if
no
scattering,
so there can be
possible to have r?*?^
1,
and
must be
scattering
However, the converse
is
necessary that
no absorption.
If there is scattering there is
it is
rft
r\ x
not necessarily absorption because
1.
have rft
1, rj t
still
the interaction in the classical limit
target sphere, and assume that the
large
sphere
with
a
incident
of a small
R, where the interaction radius
incident sphere is absorbed whenever r
As a very simple example, consider
R is the sum of the radii of the two
problem of the perfectly
absorbing sphere can be treated similarly to the problem of the perfectly
reflecting sphere given in the previous section. In the present case, the
spheres. This
ABSORPTION
Sec. 7]
total eigenfunction
KR, and
557
can contain only inward traveling waves for /
/max
normal mixture of inward and outward
will contain the
traveling waves for
>
Thus
/max .
Tjt
Tjt
=
=
0,
<
1,
>
max
max
= KR
'max
In both equations (15-98) and (15-99) the sums have contributions only
for /
/max , and
<
'max
"a
As
in the
the sum.
work preceding
The result is
(15-79),
<*a-<*b=
and we
(21
we may
1)
use an integral to approximate
= jp- = wH
Ki
(15-103)
also have
aT
The absorption cross
area nR2 as would be
,
section
oA
is
a8
is
(15-104)
the projected geometrical cross sectional
expected. But there
of equal magnitude. This
2-nR?
is
also a scattering cross section
the diffraction scattering of incident particles
passing near the absorbing sphere into the shadow on
its far side, and it is
found in the case of the reflecting
sphere. As we have said before, the diffraction scattering cannot be
observed in the classical domain because the angle d'
1/KR, within
which it occurs, is much too small. However, diffraction scattering is very
observable and very important in the nuclear domain.
In more typical examples of problems involving absorption, it is not
possible to evaluate the r) by qualitative arguments, and these quantities
must come instead from the solution of an appropriate Schroedinger
exactly the
same phenomenon that
is
equation.
write a Schroedinger equation for a
which quantum mechanical probability is absorbed, even
though we have so far always found it to be conserved. This is done by
using, as an interaction potential, the complex function
It actually is possible to
situation in
V(r)
iW(r)
(15-105)
<
where V(r) and W(r) are both real, and for absorption W(r)
everywhere [for emission W(r) > 0]. There are several ways of demonstrating
that the presence of an imaginary term in the interaction potential leads
to absorption [or emission]. For instance, by starting at equation (7-26),
and carrying through the subsequent calculation with a complex potential
COLLISION THEORY
558
[Ch. 15
V(x) + iW{x) and a small region x x to x2 the reader may show that the
quantum mechanical probability absorbed per unit time in this region is
-(2Winy*Y(x2 - xj), where
and Y*T are evaluated at some typical
point in the region. The rate of absorption of probability per unit length is
,
then
R= -
2W
T*T
(15-106)
In three dimensions the same term gives the rate of absorption of probability per unit volume. Within recent years complex potentials have been
widely used in describing the interactions between nuclear particles and
nuclei.
BIBLIOGRAPHY
Bohm,
D.,
Schiff, L.
Quantum Theory, Prentice-Hall, Englewood Cliffs, N.J., 1951
Quantum Mechanics, McGraw-Hill Book Co., New York, 1955.
I.,
EXERCISES
1.
Carry through the calculation leading to equation (15-17).
2.
Verify that (15-18)
3.
Use the Born approximation
potential V(r)
is
a solution to equation (15-19).
to calculate do/dCl for the screened Coulomb
(zZe 2 /r)e~ r ' d , which provides a useful approximation to the
potential between a charged particle
radius
Then
of the atom.
the normal
Coulomb
let
and a neutral atom if d is set equal to the
<x>, and
show that da/dQ approaches
differential scattering cross section (15-94)
when
V(r)
approaches the normal Coulomb potential.
For the Bessel functions listed in (15-56), verify that, as stated in
and 6: (a), y,(A>) * r r -0; (b), j (Kr) = sin (Kr - h/2)/Kr,
4.
(c),
the
first
maximum
of jiiKr)
5.
Verify equation (15-74).
6.
By
is
using the procedure indicated in section
the lowest energy S-wave resonance
just
7.
located at rmax
bind an
Use
partial
hard spheres with
6,
II
oo;
K.
prove that the condition for
the same as the condition that the potential
near zero total energy.
state
wave
is
sections
and a for the scattering of
=0.1, considering only S-waves, and also with KR = 0.5,
and P-waves. Plot the da/dQ. obtained in both calculations.
analysis to calculate da/dQ.
KR
considering only SHint: For r > R it
is
necessary to use the general solution to the differential
(Kr) + P,,(Ar), where each n t (Kr) can be generated
j
from the corresponding j (Kr) by everywhere replacing cos by sin and replacing
sin by cos.
equation in
r.
This
is
8.
Derive equation (15-106).
CHAPTER
The Nucleus
I.
Introduction
It is fair to say that the properties of the atom are completely understood.
The nature of the forces acting between its various parts is known, and
quantum mechanics provides a complete technique for evaluating the
effects of the forces. For the nucleus this is not the case. Despite great
progress in recent years, the nature of the forces acting between
parts
is
not completely known.
quantum mechanics
will
Furthermore,
all its
exist
its
various
not even certain that
provide a completely adequate technique for
evaluating the effects of these forces
does not yet
it is
when
they finally are known. There
a well-established theory of the nucleus through which
properties can be understood. However, there
do
exist a
number of
models, or rudimentary theories of restricted validity, each of which can
explain a certain limited range of
the study of the nucleus
is
able consequence of this situation
lack
much
its
properties.
At
the present stage,
largely the study of these models.
is
An unavoid-
that a discussion of the nucleus
must
of the coherence which characterizes a discussion of the atom.
On the other hand, it is just the fact that everything about the nucleus is not
yet understood that
makes the
subject particularly interesting.
Let us introduce our study of the nucleus by recapitulating the pertinent
information which
and by
1.
we have
filling in certain
We
have
learned incidental to our study of the atom,
other information.
briefly described
(Chapter 4) Becquerel's discovery of the
and the identification of the alpha particles
as doubly ionized He atoms. The beta particles were identified as fast
electrons from a measurement of their charge to mass ratio. The gamma
rays were assumed to be electromagnetic radiation, and this was confirmed
radioactivity of heavy atoms,
559
THE NUCLEUS
560
by showing that they can be
that radiation.
diffracted
from a
[Ch. 16
crystal like other
forms of
After Rutherford's discovery of the nucleus from an
it became apparent that
heavy atoms. Later in this
study alpha, beta, and gamma emission by such nuclei,
analysis of alpha particle scattering (Chapter 4),
radioactivity involves the unstable nuclei of
chapter
but
first
ground
we shall
we shall
concentrate on the properties of stable nuclei in their
states.
We
have learned (Chapter 4) that the mass of a nucleus is only
than the mass of the corresponding atom. Thus the nuclear
mass is approximately equal to the integer A times the mass of an H atom,
or approximately equal to A times the mass of a proton, the nucleus of
an H atom. The integer A is the integer closest to the atomic weight of
2.
slightly less
We
have also learned that the charge of a nucleus is exactly
number Z times the negative of the charge of an electron, or exactly Z times the charge of a proton. The periodic table of the
elements shows that A is roughly equal to 2Z, except for the proton for
which A = Z = 1.
3. Analysis of alpha particle scattering from nuclei of low A (Chapter 4)
showed that the radii of such nuclei are of the order of 10-12 cm, where the
the atom.t
equal to the atomic
radius
is
from the center of the nucleus at which
on the alpha particle first deviates from a Coulomb
Analysis of the rate of alpha emission from radioactive nuclei
defined as the distance
the potential acting
potential.
(Chapter 8) indicated that the radii of these nuclei, defined in
same way, are ~9 x 10~13 cm. Note that the ratio of the density of a
nucleus to the density of bulk matter in the solid state is
of high
the
[volume of nucleus]
density of nucleus
density of solid matter
[volume of atom]
-1
[(10
-12
3
) ]
-8
-1
[(10
-1
nl2
-1
-3
So the density of a nucleus is ~10 gm-cm
4. The word "radius" implies that the nucleus has a spherical shape.
12
We
have seen that the study of atomic hyperfine structure (Chapter 13)
shows the charge distribution of nuclei is generally not spherical, but
instead forms an ellipsoid of revolution with an axis of symmetry along
the direction about which the nuclear spin angular momentum vector
precesses. As a consequence, nuclei have a static electric quadrupole
moment Q. We have learned that for nuclei of spin angular momentum
quantum number
>
1,
there are cases of
Q>
(ellipsoid elongated in
the direction of the symmetry axis) as well as cases of
flattened in the direction of the
t
symmetry
axis),
but for
For any particular isotope of an atom, the atomic weight
integer.
The term "isotope"
will be explained in section 4.
is
Q<
or
(ellipsoid
\,
Q=
0.
always very close to an
THE COMPOSITION OF NUCLEI
Sec. 2]
561
However, the departures from sphericity are never severe;
in extreme
cases the ratio of the semimajor axis to the semiminor axis
Thus
it is
is
~1.2.
often possible to ignore these departures and assume a spherical
Nuclei do not have static electric dipole moments, for this would
correspond to a constant displacement of the center of the charge distribution from the center of the mass distribution, which cannot happen.
shape.
5.
Nuclei do have
static
magnetic dipole moments. This
is
also
shown
in
We
have seen that
the magnitudes of these magnetic dipole moments are ~jMjv = eh/2Mc <~
10~ 3
is the proton mass, and fi B
i"b> where fi N is the nuclear magneton,
is the Bohr magneton. Hyperfine structure also shows that the signs of the
magnetic dipole moments (the relative orientation of the magnetic dipole
the study of atomic hyperfine structure (Chapter 13).
moment
vector and the nuclear spin angular
positive in
6.
The
some
cases
spin angular
and negative
momentum quantum number
able from atomic hyperfine structure.
i is
momentum
7.
is
given in terms of
vector) are
is
directly
measur-
As we have learned (Chapter 13),
when Z is also even, and i is half-
when A is even, with / =
when A is odd. The magnitude
integral
integral
momentum
in others.
/ of the nuclear spin angular
by the usual
relation /
Closely related to the question of nuclear spin
is
Vi(i
1) h.
the question of the
symmetry character of the eigenfunction for a system containing two or
nuclei of the same species (Chapter 12). This is studied by analyzing
the spectra of diatomic molecules containing two identical nuclei and, as
we have stated, it is found that nuclei of integral spin quantum number
(even A) are of the symmetrical type and nuclei of half-integral spin
quantum number (odd A) are of the antisymmetrical type.
more
2.
The Composition of Nuclei
With one exception, A <~ 2Z for all nuclei. (The exception is the proton
which A Z = 1 .) From this it is immediately apparent that a nucleus
with a given value of A cannot be composed of A protons alone, because
A/2. Before 1932 it was generally
its Z would be Z = A instead of Z
assumed that a nucleus is composed of A protons and (A Z) electrons.
This gives the correct charge and also allows the mass to be correct.
However, it was realized that the assumption that electrons are contained
for
in nuclei presents serious difficulties.
Although the assumption made it easy to understand how electrons
could be ejected from nuclei in the beta emission process, there were problems with the magnitude of the zero point energy necessary to confine a
particle of
mass
as small as
an electron in a region of space as small as
THE NUCLEUS
562
By
the nucleus.
directly applying the uncertainty principle,
momentum from
the equation
E=
[Ch. 16
and evaluating
cp which obtains for highly relativistic
particles, the reader may show that the kinetic energy of an electron
confined to a region of dimensions 10~ 12 cm must be > 100 Mev. It was
with the fact that the typical kinetic energy of
difficult to reconcile this
an
electron ejected in the beta emission process
It
was
is
~1
Mev.
also difficult to reconcile the observation that all nuclear magnetic
dipole
moments
are three orders of magnitude smaller than the magnetic
dipole
moment
of an electron, under the assumption that electrons are
contained in nuclei.
The insurmountable
difficulty
sideration of nuclei in which
the
N atom with A =
spin angular
14,
Z=
with the assumption came from a con-
is
7.
even and
Z is
odd,
e.g.,
the nucleus of
In conformity with the usual rules, the
momentum quantum number
of this nucleus
observed to
is
1), and the symmetry character is observed
to be symmetrical. But, if it contains 14 protons and 7 electrons, its spin
quantum number would necessarily be half-integral and its symmetry
character would necessarily be antisymmetrical, because both the proton
and the electron have spin quantum number i = \ and are of antisymmetrical symmetry character, and because 14 + 7 = 21, an odd number. It is
easy to see from the rules for the addition of spin and orbital angular
momentum, presented in Chapter 13, that a system containing an odd
number of particles with half-integral spin quantum number can only have
be integral
(specifically,
momentum (the nuclear spin angular momentum) corresponding to a half-integral quantum number. It is also easy to see, from
the discussion in Chapter 12, that the symmetry character of a system
containing an odd number of particles of antisymmetrical symmetry
character must be antisymmetrical.
Some years before its discovery in 1932, Rutherford suggested the
total angular
existence of a particle having the
same mass, symmetry
spin quantum number
we now call the neutron. One motivation
character,
as the proton, but zero charge. This
is
for this suggestion
and
the particle
was
that the
problems outlined above would not arise if nuclei were assumed to be
composed of protons and neutrons. In particular, the spin and symmetry
character of the
nucleus would agree with observation if it contained
7 protons and 7 neutrons. This would also, of course, be in agreement
with the mass and charge required for A = 14 and Z = 7. A number of
people tried to devise experiments which would detect the neutron. But
this was difficult because, being uncharged, the neutron cannot directly
produce ionization, and most devices for detecting particles depend upon
ionization.
We shall describe the sequence of experiments and interpretations leading
THE COMPOSITION OF NUCLEI
Sec. 2]
to the discovery of the neutron, as
and
it
is
563
of great historical importance
good examples of some of the early techniques of
nuclear investigation. In 1919 Rutherford bombarded a target of N nuclei
with 7.7-Mev alpha particles from a radioactive source and found that
charged particles of long range were emitted from the target. The experialso provides
mental apparatus, employing a ZnS scintillation screen (cf. section 3,
Chapter 4) as a charged particle detector, is indicated schematically in
figure (16-1). The range of a charged particle is the thickness of some
standard absorbing material which it can traverse before losing all its
kinetic energy by ionizing the atoms of the absorbing material, and coming
gas
An apparatus used
Figure 16-1.
ZnS
Absorber
Source
in
the
first artificially
produced nuclear reaction.
This quantity increases monotonically with the velocity of the
to a stop.
and decreases monotonically with its charge. The range of the
particles emitted from the N nuclei of the target was so long that they could
unambiguously be identified as protons, since these particles have much
particle
longer ranges than alpha particles of comparable energies because they
have higher velocities and smaller charges. Rutherford's experiment
constituted the
first artificially
2
The equation
2
He4
A =
A =
14;
1
+ N 14 -* 8
7
represents the reaction
17
2,
8
A =
17
is
4;
N 14 is
(16-1)
and introduces some
the bombarding particle, an alpha particle which
Z=
with
and
is
He4
produced nuclear reaction
useful notation:
is
the
He
nucleus
N nucleus with Z = 7,
O nucleus with Z = 8, A = 17;
the target nucleus, an
the residual nucleus,
an
H 1 is the product particle, a proton which is the H nucleus with Z =
I.
Several other reactions of essentially the same type were investigated in
succeeding years, but in 1930 a reaction of a different type was discovered
by Bothe and Becker. They used an arrangement
in figure (16-1), except that the
ZnS
one shown
was replaced by a
similar to the
scintillation screen
Geiger counter. This is a chamber filled with gas and containing a fine
wire held at a high positive voltage with respect to the walls. Any ionizawithin the chamber liberate one or more
which are accelerated so strongly in the direction of the wire
that they produce additional ionization by collisions with gas atoms.
This process repeats, and an avalanche of electrons rapidly builds up.
tion processes occurring
electrons,
THE NUCLEUS
564
Enough charge
is
collected
by the
wire, as the result of a single initial
ionization, to produce an easily detectable electric pulse.
particles of kinetic energy 5.3
bombardment of Be or B
very thick absorbers with
assumed these
rectly
Using alpha
Mev, Bothe and Becker found
uncharged
nuclei
little
[Ch. 16
particles,
that in the
capable of traversing
attenuation, were emitted.
to be very high energy quanta
They
(gamma
incor-
rays).
Geiger counter can detect quanta since they can produce electrons in
the filling gas by the photoelectric effect, the
Compton
effect,
or pair
production.
BeorB
Source
Hydrogenous
Absorber
Figure 16-2.
Further work on
They
An
apparatus used
this reaction
in
Geiger
counter
layer
the discovery of the neutron.
was done
in
1932 by
I.
Curie and Joliot.
discovered, quite accidentally, that placing a layer of hydrogenous
material in front of the Geiger counter caused the counting rate to increase
was soon
was due to the ejection of protons from
its bombardment by the uncharged
particles. Protons, being charged, are counted much more efficiently by a
Geiger counter than quanta or other uncharged particles. The apparatus
used by Curie and Joliot is indicated in figure (16-2). By measuring
the range in the absorber of protons ejected from the hydrogenous layer,
It
realized that the effect
the hydrogenous layer as a result of
the velocities or energies of these protons could be determined from the
known
Ev
range-velocity relation.
~ 4.5 Mev
for
Be
Curie and Joliot found the proton energy
Following Bothe and Becker, they
target nuclei.
assumed that quanta incident upon the hydrogenous layer ejected protons
by a process analogous to Compton scattering. They also evaluated E,
the minimum energy of the quanta required to eject a 4.5-Mev proton.
This gave Ey
50 Mev, as the reader may verify by using the equations of
section 8, Chapter 3 (with the electron mass replaced by the proton mass).
Shortly after this experimental development, Chadwick demonstrated
that its interpretation was untenable, and that the uncharged particles
emitted from the Be or B nuclei under alpha particle bombardment were
not quanta but the long-sought neutrons. He showed first that the number
of protons ejected from the hydrogenous layer was several orders of
magnitude smaller than would be predicted by the Klein-Nishina equation
(14-36 with the electron mass replaced by the proton mass), if the incident
NUCLEAR
Sec. 3]
AND THE
SIZES
OPTICAL MODEL
He showed
uncharged particles were 50-Mev quanta.
50-Mev quantum
energetically possible for a
in
which a 5.3-Mev alpha
This calculation
nucleus.
particle
is
is
565
next that
it is
not
to be liberated in a reaction
captured by a Be nucleus to form a
based upon a knowledge of the mass of the
C nucleus and
alpha particle, the mass of the Be nucleus, the mass of the
mass and energy. We
For the present, suffice
Einstein's relation (1-23) between
shall
such calculations in section
it
4.
do several
to say that a
showed that all the experimental results were consistent
was assumed that the reaction involved in the case of a Be target
similar calculation
if it
nucleus
is
He4 + 4 Be 9 -> 6 C 12 +
A =
1.
explain the experiments performed with a
where
is
the neutron with
Z=
0,
(16-2)
would
With this
similar reaction
target nucleus.
assumption, the ejection of protons from the hydrogenous layer could be
understood as a simple collision between an incident particle, the neutron,
and a stationary
free particle of equal mass, the proton.t
minimum neutron
energy required to eject a 4.5-Mev proton
Mev, and
this is in
masses of the
good agreement with the
entities
is
Then
the
En =
4.5
imposed by the
restrictions
appearing in (16-2) plus the mass-energy relation.
Since the discovery of the neutron, there has never been any doubt that
a nucleus
composed of Z protons and {A Z) neutrons. Thus a nucleus
(A Z) = A nucleons, a word used for both protons and
is
Z+
contains
neutrons.
Nuclear Sizes and the Optical Model
3.
Some of the
first
came from alpha
evidence for the sizes of nuclei, other than that which
particle scattering
and emission, was obtained from the
Two nuclei are
common values of A,
differences in the total binding energies of mirror nuclei.
said to
and
if
form a
Z=
{A
Some examples
set
- Z) +
In each
set,
N 15
for one
if
they have
and {A
and 2 He3
and
Li 7 and
u Na23 and
15
one nucleus
3
,
is
12
Be 7
5
,
Mg23
17
,
kinetic
C1 35 and
18
A35
unstable and eventually decays by the emission
differences in the total binding energies of the
The
for the other.
B U and 6 C n
of an electron or a positron, forming the stable
=Z+
Z)
are
W
7
of mirror nuclei
and binding energies of a proton
completely negligible compared to 4.5 Mev.
two
in a
member of
the
set.
The
nuclei of each set can be
hydrogenous compound are
THE NUCLEUS
566
[Ch. 16
obtained directly from measurements of the energies of the electrons or
positrons which are emitted in the decay, and can also be obtained in-
some cases, from measurements of the masses of the nuclei
and the mass-energy relation.
directly, in
Now it is clear
that, in addition to the repulsive
Coulomb
forces acting
between each pair of protons in a nucleus, there must be attractive forces
of some other type acting between all the nucleons. If it were not for these
so-called nuclear forces, neutrons would not be bound in nuclei and, in
addition, all nuclei for Z
2 would break up by Coulomb repulsion.f
It was assumed at an early date, and has subsequently been verified by
several different kinds of evidence to be presented later, that the nuclear
forces acting between pairs of nucleons are independent of whether the
nucleons are protons or neutrons. With this assumption, the total binding
energies of two mirror nuclei will differ only
by the additional Co'ijlpmb
repulsion energy A.EC of the nucleus with the extra proton.
assumed that
all
extra proton,
is
A c =
is
this
it is
also
uniformly distributed over a sphere of radius R, the addi-
Coulomb repulsion energy can be
statics. The result is
tional
where
Using
If
the charge due to the protons, including that due to the
the atomic
evaluated from classical electro-
7e
-
6
(16-3)
number of the nucleus without
the extra proton.
equation to analyze the observed differences in total binding
number of sets of mirror nuclei with A ranging from 3 to
was found that values of R so determined could be represented by
energies of a
38,
it
the equation
R=
where
recent
r is a constant.
work on
this
The
(16-4)
A*>
~ 1.4 X
analysis gave r
10~ 13 cm. However,
problem has shown that quantum mechanical cor-
rections to this simple classical calculation reduce r somewhat.
currently quoted
The quantum mechanical
12-30).
decrease in
1.2
10" 13
calculation uses a
tion for the extra proton
(cf.
The value
is
and takes
cm
more
(16-4')
realistic
charge distribu-
into account the exchange integral
Both factors tend to lower AEC thus requiring a compensating
,
The nuclear forces are definitely not gravitational in nature. From the data presented
1 on the charge and mass of nuclei, the reader may easily show that gravitational forces are too weak to overcome the Coulomb forces by about 35 orders of
t
in section
magnitude.
NUCLEAR
Sec. 3]
SIZES
AND THE OPTICAL MODEL
567
There are several other methods of measuring the size of the nuclear
We shall consider the one which is the most straightforward and accurate. It is the scattering of electrons from nuclei at
charge distribution.
hundred Mev. As nuclear forces do not act between
is due
to the Coulomb interaction between the electron and the nuclear charge
distribution.
Therefore measurements of electron scattering should
provide information about the nuclear charge distribution. The method
can be thought of as the use of an "electron microscope" of extremely high
resolution to "look at" the charge distribution. The resolution is extremely
energies of several
nucleons and electrons, the scattering of an electron from a nucleus
Target
foil
Electron
beam
'
Collimator
^\
Co
or
Magnet
Figure 16-3.
An
apparatus used to measure the scattering of high energy electrons by
nuclei.
high because the de Broglie wavelength X of the electrons
is
extremely
Using the expression p = Ejc to evaluate the momentum of the
electrons, which are highly relativistic because their energy is very large
compared to their rest mass energy m c 2 = 0.51 Mev, we find for 500-Mev
10~ 12 cm as an estimate of
electrons I = 2.4 x 10~ 13 cm. Taking R
the size of the charge distribution, we see that for such electrons 1 <~ RjA
and KR = (2ttIX)R <~ 20. According to the conclusions drawn in the
short.
preceding chapter, the differential cross section for the scattering of such
from the nuclear charge distribution should be very sensitive to
and even to the details of its shape.
shall not attempt to describe the machines which are used to accel-
electrons
its size,
We
erate electrons (or other particles) to the high energies required in these
experiments since, in the space available, we could not do justice to the
subject. f
However,
it
is
worth while to describe the other aspects of
the electron scattering experiments.
Hofstadter and colleagues
electron
beam
is
The apparatus used
passes through a thin
the nuclei to be investigated,J and
it is
foil,
the
work by
The collimated
in recent
indicated in figure (16-3).
atoms of which contain
stopped in an insulated metallic cup.
t Descriptions can be found in a number of elementary textbooks.
An electron of several hundred Mev has such a high relativistic mass that the
scattering
produced by the atomic electrons
is
completely negligible.
THE NUCLEUS
568
The
total charge collected
by the cup
[Ch. 16
a measure of the incident
is
flux.
Electrons scattered into a certain angular range at 6 enter a magnet, which
bends them in a semicircle in the plane perpendicular to the paper. Upon
leaving the magnet, they enter a scintillation detector similar to the old
ZnS
human
eye detector, but employing a
more
efficient crystal
and a
10"
\
10"
10"
10"
.Sic
\A
10"
10' -34
30
40
50
60
80
70
90
Figure 16-4.
upon carbon.
The differential scattering cross section for 420-Mev electrons incident
From R. Hofstadter, Annual Review of Nuclear Science, Vol. 7, Annual
Reviews, Stanford, 1957.
photoelectric
magnetic
cell.
field is a
The radius of curvature of the electron path in the known
measure of the
momentum
or energy of the scattered
electrons.
It is
found that by
energy loss in the
measured for 420-Mev
6
C 12
is
t This
shown
is
far the
CM
LAB
likely process
The
frame and the
LAB
is
scattering with
no
differential scattering cross section
electrons, incident
in figure (16-4).f
actually dajdQ. in the
between the
most
frame.
The
on the
typical small
nucleus
points with accuracy estimates are
frame, but for electron scattering the difference
CM frame
is
quite negligible.
NUCLEAR
Sec. 3]
SIZES
AND THE
OPTICAL MODEL
569
They are evaluated from the
number of full energy electrons
each angle, and the number of nuclei per
the experimental measurements of da/dQ..
measured incident
flux,
the measured
scattered per unit solid angle at
cm2
in the target foil.
That absorption processes are observed to be improbable verifies our
idea that the scattering should be explicable on the basis of a real scattering
potential representing the Coulomb interaction between the electron and
the nuclear charge distribution. That the electron energies are high
suggests the additional idea that the Born approximation should be useful
in the analysis of these experiments. It actually is found that for nuclei
of small A (and therefore small nuclear charge and scattering potential)
the Born approximation is capable of giving a reasonably good account of
the measured da/dQ if cognizance is taken of the fact that the electrons
are highly relativistic. This is done by using free particle eigenfunctions
obtained from the Dirac equation, instead of the Schroedinger equation,
to evaluate the matrix elements entering in the Born approximation.
The results of this relativistic Born approximation are
where
E is
da
(Ze2 \2 cos2
dD
\2e)
\2E)
(0/2) \F( X )\
si
sin*
(6/2)
The form
the electron energy.
(16-5)
Z2
factor F{%)
is
"
F(X)
r
f p(
Jo
)!!L2T Airr2 dr
(16-5')
with
where
= 2K sin (0/2),
scattering
is
p(r) is the nuclear
as in the case of the
(cf.
* = 7^4
charge density in units of proton charges.
form factor that
arises in the discussion of
equation 14-35), at small angles F(%)
essentially unity at small scattering angles.
writing
2E
en
~ 2mc
for the highly relativistic
~ Z.
Just
X-ray
Also, cos 2 (0/2)
Keeping these points in mind,
electrons, and then comparing
equation (16-5) with equation (15-94), we see that at small scattering
angles da/dO. of the former equation is equal to the Coulomb differential
scattering cross section for a point charge nucleus.
angles,
dajdQ.
2
is
reduced below this value because
<
The
At
large scattering
\F(%)\
< Z2
and
term describes the effect of the finite
extension of the nuclear charge; the second term describes a relativistic
effect which may be attributed to the electron spin.
The behavior of dajdQ, as calculated from equation (16-5), depends in a
quite sensitive manner on the assumed form of the nuclear charge density
because cos (0/2)
first
THE NUCLEUS
570
[Ch. 16
Thus a comparison of the calculated and measured values of dajdQ
allows a determination of this quantity. The dashed curve in figure
(16-4) shows the best fit obtained for dajdQ calculated in the Born
p{r).
The
approximation.
from a phase
shows dajdQ. calculated
using the same p(r) as was used in the Born
solid curve in the figure
shift analysis
approximation calculation.
Of
course, the phase shift analysis
upon the Dirac equation, and not upon
2
r in 10
Figure 16-5.
In figure (16-5)
same
p(r) is
we
7,
based
cm
The nuclear charge density of carbon. From
of Nuclear Science, Vol.
the
3
_13
is
the Schroedinger equation.
R. Hofstadter, Annual Review
Annual Reviews, Stanford, 1957.
plot the p(r) used in these calculations.
found
in the analysis of dajdQ.
measured
of electrons of several different energies from the
As
essentially
in the scattering
and as the
nucleus,
analyses are sensitive enough to distinguish variations in, for example,
p(0) of perhaps
10 percent, there
is
considerable confidence that this
must be quite close to the actual charge density of the nucleus.
For a given electron energy, the differential scattering cross sections
measured for nuclei of larger A are found to be similar to those measured
p(r)
for C, except for a decrease with increasing
first
minimum
This
means the
is
partial
in the angle at
found, and a development of subsidiary
The experience gained
larger angles.
this
is
size
which the
maxima
at
in the previous chapter tells us that
of the charge distribution increases with increasing A.
seen in figure (16-6), which plots charge densities obtained by a
wave
analysis of the differential scattering cross sections
for electrons of several energies
The forms of
and
for a
number of
measured
different nuclei.
For inand heavier nuclei,
is known to be constant only to within 10 or 20 percent.
However, it is
assumed that the central density is constant, and that the form of p(r)
p(r) given in the figure are
not completely unique.
stance, the charge density in the central region for Ca,
NUCLEAR
Sec. 3]
AND THE
SIZES
OPTICAL MODEL
571
can be described by the empirical equation
With this assumption, the parameters a and b are accurately determined
by the analysis of the scattering measurements to be
a=
b
l.01A 1A
10" 13
cm
(16-7)
0.55
x 10" 13 cm
0.10
E 0.08
a>
0.06
0.04
0.02
rin 10
Figure 16-6.
The nuclear charge
Review of Nuclear Science, Vol.
7,
"cm
density of several nuclei.
From
R. Hofstadter, Annual
Annual Reviews, Stanford, 1957.
The parameter
it
p(0) is not determined very accurately by this analysis, but
can be simply by demanding that the integral of the charge density over
volume equal Z. Figure (16-6) consists of plots of equation
(16-6) using the values of the parameters so determined.
the nuclear
We
1.
draw
the following conclusions
The charge
density of nuclei, which
tion of the protons in the nuclei,
region and
2.
The
falls fairly
The
must be essentially the distribuapproximately constant in the interior
rapidly to zero at the surface.
half-density radius a increases with increasing
A'A law.
3.
is
fall-off distance
is
according to an
approximately the same for
all nuclei.
THE NUCLEUS
572
[Ch. 16
The interior charge density p(0) decreases slowly with increasing A.
If we assume that the distribution of protons in nuclei is approximately the same as the distribution of neutrons (there is good evidence for
4.
5.
this
assumption), the charge density p(r) of a nucleus
density p
M {r)
is
related to
its
- PM (r)
p(r) cc
(16-8)
slow decrease of p(0) with increasing A
Z/A with increasing A (Z/A
If so, the
with the decrease in
1/2.5 for
density p
M (0)
mass
by the equation
completely consistent
is
1/2 for
~ 240),
is
and with the assumption that the
approximately the same for all nuclei.
~ 40;
interior
ZjA
mass
coefficient of A Vz in the expression for a is 1.07, and
equations (16-6) and (16-7) are actually consistent with (16-4)
and (16-4') because, within the accuracy of the analyses, both lead to
Although the
6.
not
the
1.2,
same root-mean-square
criterion for comparison.
radius,
which can be shown to be the proper
Consequently,
this provides support for the
assumption of the equality of the forces between various nucleons, which
is the basis of the mirror nuclei analysis that leads to (16-4) and (16-4').
Let us turn now to measurements of nuclear sizes from experiments
which use nucleons as probes instead of electrons. Certain contrasts
between these experiments and the electron experiments immediately
One
arise.
forces,
is that a nucleon interacts with a nucleus by means of nuclear
and not by means of Coulomb forces (although there is also a
Coulomb
interaction if the nucleon
is
a proton).
of nucleons with nuclei involve a different
measure a
set
Since the interactions
of forces, these interactions
different set of attributes of the nuclei,
and
it
should not be
the measurements yield nuclear sizes which are not the
same
measured in the electron experiments. Another point is that the
interaction between a nucleon and a nucleus must be described by a
complex interaction potential. A real potential is not adequate because
surprising
if
as those
there
is
appreciable absorption
(cf.
section
experiments show that, when a nucleon
is
7,
Chapter
incident
upon
15)
probability for absorption, including scattering with energy loss,
quite
since the
a nucleus, the
is
comparable to the probability for scattering with no energy
generally
loss.
In fact, the interpretation of the earliest experiments of this type was
based on the assumption that, in its interaction with a nucleon, a nucleus
behaves like a sphere of radius R which is completely absorbing. The
experiments were measurements of the total cross sections a T of nuclei
for neutrons of energies near 20 Mev. For a neutron of this energy, the
de Broglie wavelength is X
6 x 10 -13 cm. Using i!~5x 10~ 13 cm
as a reasonable estimate of the radius of a nucleus of intermediate A,
Sec. 3]
NUCLEAR
it
OPTICAL MODEL
573
find KR = {^.7tJX)R
5.
We have then the case of a sphere which
assumed to be completely absorbing, with large KR. Consequently
should be possible to apply equation (15-104), which is
we
is
AND THE
SIZES
aT
The
total cross section
aT
attenuation technique that
is
= 2nR
(16-9)
measured by
used for X-rays
essentially the
same beam
Chapter 14),
with a neutron detector consisting of a mixture of some hydrogenous
compound and some compound that scintillates. Neutrons entering the
Figure 16-7.
The
is
12
16
20
24
28
(cf.
32
section
40
36
total cross sections of several nuclei for
detector collide with protons of the hydrogenous
5,
26-Mev neutrons.
compound,
transfer
energy to the protons, and these charged particles subsequently produce
scintillations.
A number of different measurements have been made;
shows some data for 26-Mev neutrons. The solid curve is
equation (16-9), using R
1.55^ x 10" 13 cm. However, this is no
longer accepted as a reliable estimate of the nuclear size. More sensitive
figure (16-7)
measurements of the interaction between nucleons and nuclei show that
the analysis based
KR ->
oo is
much
upon a completely absorbing sharp-edged sphere with
too crude to yield accurate results.
These more sensitive measurements are of the differential scattering
cross sections of nuclei for nucleons. In the past few years many accurate
measurements have been made for both neutrons and protons, and for a
and nuclei. The experimental arrangements used are
basically the same as that used in the electron scattering experiments.
For protons, the energy sensitive detector can be a magnet plus scintillavariety of energies
tion counter, or simply a scintillation counter using a crystal such as Nal
is a linear function of the proton energy. Some data for
whose response
THE NUCLEUS
574
0"
Figure 16-8.
protons.
The
The
25"
50
75"
[Ch. 16
100 125 150 175
differential scattering cross sections of several
nuclei for
points are experimental and the curves are theoretical
Glassgold, Revs. Mod. Phys., 30, 419, (1958).
fits.
I7-Mev
From A.
E.
NUCLEAR
Sec. 3]
AND THE
SIZES
OPTICAL MODEL
17-Mev protons are shown by the points
575
Following
frame of the observed
in figure (16-8).
convention, this figure plots the ratio in the
CM
differential scattering cross section (scattering
with no energy loss in the
CM frame) to the Coulomb differential scattering cross section for a point
charge nucleus (equation 1 5-94).
da/dQ
It
should be noted, then, that the observed
actually decreases rapidly with increasing 6
owing
-4
to the sin
0/2
dependence of the Coulomb da/dQ.
| -9
Imaginary
s -55
CL
Figure 16-9.
The
real
Real
and imaginary parts of the optical model potential for the scatter-
ing of 17-Mev protons.
The solid curves are the best fits to the data obtained from a partial
wave analysis using equation (15-102). In this analysis the interaction
potential is taken to be of the form
The first term is a complex
(r-a)/ b
ydr)
(16-10)
interaction potential representing the refracting
and absorbing properties of the nucleus for the incident proton. The
second term is the real Coulomb potential acting upon the proton and
is evaluated from the nuclear charge density determined by electron
scattering. For the analysis of the 17-Mev proton data, the following
parameters were used in the first term
V = -55Mev
W = -9 Mev
a=
1.25
0.65
The
real
,A
x 10"13 cm
(16-10')
x 10" 13 cm
and imaginary parts of the interaction potential are indicated
in
figure (16-9).
It is
evident that a negative
the nucleus
tial
is
for
an unbound proton moving through
reasonable in light of the fact that the real part of the poten-
must, by definition, be attractive for a bound proton moving in the
It is even more evident that a negative
Q is reasonable since, as
nucleus.
THE NUCLEUS
576
[Ch. 16
indicated in (15-106), to have absorption the imaginary part of the
must be negative. But the very form of the interaction potential
it is only an approximation to the true potential
acting on the proton moving through the nucleus. This is because the
true potential must certainly be a function of all the distances between
the proton and the nucleons of the nucleus (or at least a number of these
distances), and not just a function of the single distance between the
proton and the center of the nucleus. Thus the approximation replaces
the very complicated true potential by a much simpler potential which,
potential
makes
it
evident also that
50 100 200
500
(Mev)
Figure
16-10.
The approximate energy dependences of the depths of the
real
and
imaginary parts of the optical model potential.
presumably, represents the average
effect
of the true potential
a procedure
quite similar to that used in the Hartree treatment of the atom.
This
approximation is called the optical model of the nucleus. It was originally
used around 1930, with a purely real potential. This was not adequate,
and so the optical model fell into disfavor. However, with a complex
potential it has experienced a very successful revival at the hands of
Feshbach, Porter, Weisskopf and Saxon in the years after 1952.
As is true for any model, the justification of the optical model is that
it
provides a simple picture of certain processes and enables a large
number
of experiments involving these processes to be explained in terms of a
small number of adjustable parameters. For instance, the set of parameters
which provide an acceptable fit to the 17-Mev proton differential scattering
cross sections also provides an acceptable fit to neutron total cross sections
at a similar energy. In fact, the single set of parameters a and b quoted in
(16-10') is found to give an acceptable fit to the scattering and total cross
sections for both protons and neutrons over a wide range of nuclei and
At a given energy, a single set of parameters V and
Q also is
adequate to explain both cross sections for both particles and all nuclei.
do exhibit a dependence on the energy E of the
However V and
proton or neutron, as indicated in figure (16-10). There are still certain
energies.
NUCLEAR
Sec. 3]
AND THE OPTICAL MODEL
SIZES
ambiguities in the values of
V and
some
since, in
577
cases, different values
of these parameters can give equally acceptable fits to the data. But the
general behavior of their energy dependence is fairly well established, as
For
are their values at low values of E.
E is
instance,
known
it is
that
when
2 or 3 Mev,
can only be 1 or 2 Mev, because very pronounced
resonances are observed in the scattering and total cross sections at these
values of E. Resonances (cf. section 6, Chapter 1 5) involve constructive
interferences between the radially inward traveling and radially outward
traveling components of some partial wave. Therefore they can occur only
if the outward traveling component is not absorbed and, as shown by
equation (15-106), small absorption means small
and the rate of absorption
of particles in the complex optical potential, which is of a form different
from (15-106). Consider a particle of mass
moving with total energy
E along the x axis through a medium of constant optical potential
(V + iWo)- The reader may verify by substitution that a solution to
There
is
an
between
interesting relation
the time independent Schroedinger equation for this situation
ip(x)
-S/2A
e~
is
(16-11)
where
(3-
AM
2
\W
//
(E+ini)
+ 1-1
(16-11')
and
K 2 = ^( +
Note
that in the limit
V,\)
\W
where
E+\V
\I(E
|K
|)
(16-11")
is
negligible
compared
to
1,
equation (16-11") reduces to the usual expression for K. When this
quantity is only small compared to 1, equation (16-11') reduces to
2(E+\V
A:
2|wg
To
identify
\)
(16-12)
A, calculate the probability density
y>*(x) yi(x)
-iKx -x/2A iKx -x/2A __ -x/A
a
e~
We see that the probability density dies out exponentially with an attenuation length equal to A. Now let us evaluate A for E = 20 Mev, using the
values of V and W taken from figure (16-10). We have: E = 20 Mev,
v = 45 Mev, W = 10 Mev, M = one proton mass. For these values
|
(16-12) gives
A=
This
is
3.3
lO" 13 cm,
E=
20
Mev
(16-13)
about one-half the average distance across a nucleus of average
THE NUCLEUS
578
radius.
Thus the probability density associated with a
energy following such a path through a nucleus
is
[Ch. 16
particle of this
attenuated by about
a factor of e 2 and the particle has only about a 10 percent chance of not
being absorbed. Nuclei are quite opaque to nucleons of energies in the
range 20 Mev (and higher). | The situation is quite different at E = 2 Mev.
,
At
this
energy
F =
|
50 Mev,
A=
20 x
|I
=
rr
WA
0]
10" 13
1.5
Mev, and we obtain
E=
cm,
a^l.07A*x
10
-13
Mev
(16-14)
cm
a^l.25A H x 10- 13 cm
Figure 15-11.
comparison between the nuclear mass distribution and the nuclear
interaction potential.
This
about three times the average distance across a nucleus of average
The probability density for a particle of this energy following such
a path is attenuated only by a factor of about e /s and the particle has
about a 75 percent chance of not being absorbed. Nuclei are fairly
transparent to nucleons of energies in the range 2 Mev (and lower).
Let us compare the results of the experiments that measure the charge
or mass distribution with the results of the experiments that measure the
distribution of the real or imaginary interaction potentials. This is done,
for a typical nucleus, in figure (16-1 1). The figure plots the mass distribuis
radius.
tion,
and the absolute value of the
real
interaction potential.
It
is
extremely interesting to note that the spatial distribution of the nuclear
interaction potential exerted
t This
is
by a nucleus on a passing nucleon extends
part of the explanation for the large values of
20-Mev neutron
total cross sections with equation (16-9);
found by analyzing the
is high enough
the opacity
that there is appreciable probability of absorption even when the neutron passes through
the outer nuclear surface. The rest of the explanation is that (16-9) assumes that the
de Broglie wavelength of the neutron is negligibly small. Actually it is not, and, since
the neutron can be absorbed if it passes within a de Broglie wavelength of the nucleus,
this also increases the
from
(16-9).
absorption probability and, therefore, the value of
R calculated
NUCLEAR
Sec. 3]
beyond the
<5
SIZES
AND THE OPTICAL MODEL
579
spatial distribution of the nucleons in the nucleus
which, for a typical nucleus,
is
only about
by a distance
x 10~ 13 cm.t This means
that nuclear forces must be of very short range. Thus nuclear forces are
certainly very different from the long range gravitational and electric
forces of
common experience. An electron passing 2
the surface of a nucleus feels an appreciable
passing 2 or 3
essentially
x 10~ 12 cm from
no force
at
Figure 16-12.
Illustrating the
and a proton.
Of
Coulomb
the surface of the
x 10~8 cm from
force ;
a neutron
same nucleus
feels
all.
typical nucleus
composition of the total
course, a proton passing 2 or 3
nucleus does feel a
or 3
Coulomb
force.
10~ 12
This force
real potential acting
cm from
is
between a
the surface of the
strong, but not so strong
would feel on passing a little closer. An
comparison of the relative strengths of nuclear forces and
Coulomb forces can be obtained by plotting the real nuclear potential,
and the Coulomb potential acting on a proton, for a typical nucleus and a
typical proton energy. This is done in figure (16-12). Because of the
as the nuclear force the proton
instructive
extension of the nuclear charge distribution, the Coulomb potential
does not continue its Ze2 /r behavior for
less than the nuclear radius.
For 20-Mev protons incident upon 92
nuclei, the maximum positive
finite
jr,
113
but more probably i!~l x 10~ 13 cm oc A". This
;
t The figure implies 6 oc A
not inconsistent since the values of a for the mass and (particularly) for the potential
distributions are not known with too much certainty, so that S is the relatively small
is
difference between
two large uncertain values.
THE NUCLEUS
580
value of the total real potential
is
about
Mev and
the
[Ch. 16
minimum
nega-
example of a large Z
nucleus, we see that when r is less than the nuclear radius the magnitude
of the real nuclear potential is somewhat larger than the magnitude of the
tive
value
is
about
45 Mev. Even
for this extreme
Coulomb potential.
4.
Nuclear Masses and Abundances
The detailed determinations of nuclear density distributions obtained
from the recent electron scattering experiments show that the density in
the interior of a nucleus is approximately the same for all nuclei. However,
this was known at an early date because all the analyses of nuclear sizes,
based on the assumption that the nucleus is a sharp-edged sphere of radius
R and uniform interior density, lead to the result R oc A!/z for which the
density p = mass/volume oc AI(A'A ) 3 is independent of A. In this section
we shall study, primarily, the measurements of nuclear masses. Among
other things, these measurements show that over a very broad range the
binding energy per nucleon is approximately the same for all nuclei.
This was also known at an early date and, along with the approximate
equality of the nuclear density,
it
led to the assumption that all nuclei
contain essentially the same material in essentially the same state, and differ
from each other only in the amount of this material they contain. The
assumption formed the basis of the very successful liquid drop model of
the nucleus, which we shall discuss in the next section.
We know that the mass of a nucleus is almost equal to the mass of the
corresponding atom. The masses of atoms of a particular Z, but possibly
a mixture of A, can be obtained to an accuracy of several significant figures
by the chemical techniques and a knowledge of Avogadro's number.
But, for the more accurate determinations needed in the study of nuclei, it
is necessary to use the physical techniques of mass spectrometry or energy
balance in nuclear reactions. Both give information about the masses of
atoms of a particular Z and A. From these masses, the masses of the
corresponding nuclei can be evaluated by subtracting Z times the electron
mass. In doing this, the mass equivalent of the electron binding energy is
neglected. However, even in extreme cases, the error thereby introduced
is less than 20 percent of the experimental uncertainties in the mass
measurements.!
The first mass spectrometer was built in 1911 by Thomson. It used the
same configuration of crossed electric and magnetic fields as the electron
t
For
binding energy is 2 x 10 5 ev
quoted as 238.1234 0.0010 mass units.
S2JJ23S t h e tota j e i ec tron
The mass of
the
atom
is
~ 0.0002
mass
unit.
NUCLEAR MASSES AND ABUNDANCES
Sec. 4]
charge to mass ratio measurements
(cf.
581
section 2, Chapter
3).
beam of
atoms, ionized by electron bombardment, was passed through these fields
which analyzed it into components according to the different charge to
mass ratios present in the beam. It was usually possible to determine the
degree of ionization of a component and thus determine
somewhat low accuracy of the apparatus.
the
Thomson
its
With
mass to within
this
apparatus
discovered the existence of isotopes. Using a source containing
Source
Region of
parallel
Boundary of
E
_
to'
region of
perpendicular
paper
to
paper
Photo
plate
Figure 16-13.
An
apparatus used to measure atomic masses.
component with mass corresponding
component with A = 22. A number
weaker
associated
and
an
20,
of tests proved that these were both due to a noble gas, and this could only
be Ne, which has a chemical atomic weight of 20.18. Thomson interpreted
his results to mean that there are two chemically indistinguishable species
of Ne atoms, called isotopes, one with A = 20 and relative abundance of
about 91 percent, and one with A = 22 and relative abundance of about
a mixture of noble gases, he found a
to A =
9 percent.
exactly the
They are chemically indistinguishable because they have
same structure of atomic electrons since their nuclei have the
same Z, but they
are physically distinguishable because they have different
have different A. The nuclei of the Ne isotopes
the second occurs with relative abundance of
about 0.3 percent and could not be detected with Thomson's apparatus.
A typical example of a more recent mass spectrometer is the type designed
masses since
are
10
Ne 20
their nuclei
10
Ne 21
10
Ne22
by Bainbridge (1933) and illustrated in figure (16-13). The source produces ionized atoms with charge +Ze,mass M, and a spectrum of velocities.
These atoms pass through a region of crossed electric and magnetic fields
which act as a velocity filter, passing only those with velocity v for which
ZeE
The terms on
the left
and
HZevjc
right are the electric
and magnetic
forces,
THE NUCLEUS
582
Atoms of velocity
respectively.
magnetic
EcjH
then enter a region of uniform
are bent in a semicircle of radius R,
field,
The
graphic plate where they produce an image.
image
is
2R, where
Solving for
the right
the
is
and
on a photofrom S2 to the
fall
distance
the equation
satisfies
HZevjc
The term on
[Ch. 16
Mv*/R
mass times the
centrifugal acceleration.
M, we have
= RZeH = RZeH*
M
c
Ec
The mass can be determined from absolute measurements of the
on the right side of the last equation. In addition, much use is
quantities
made
of hydrocarbon molecules to calibrate the apparatus over a wide
range of masses in terms of the carefully standardized masses of 1 1 and
6 12
With these techniques, extremely accurate measurements can be
C
.
A40 is quoted as
= 39.975022 0.000029 amu
made. As an example, the mass of
18A 4o
An amu
is
one atomic mass
unit.
80 i6
18
It is
defined such that
16.000000
amu
(16-15)
(Note that this differs slightly from the chemical mass scale which uses the
normally occurring mixture of s O isotopes as a standard.)
Mass spectrometers, using detectors which are more linear than photographic plates, can provide accurate determinations of the relative abundances of the various isotopes. As an example, the abundances of the
normally occurring mixture of
i6
17
isotopes are
known
to be
= 99.759%
= 0.037%
18 =
0.204%
Another technique of mass determination, which provides an accurate
supplement and check for the technique of mass spectrometry, is the study
of the energy balance in nuclear reactions. Consider a nuclear reaction
such as (16-1),
2
the one
first
He4
+ N 14 ->7
observed by Rutherford.
a
where a is the bombarding
17
is
The general
+ A-+B +
particle,
+ H
case
the target nucleus,
may
be written
(16-16)
B is the residual
NUCLEAR MASSES AND ABUNDANCES
Sec. 4]
583
and b is the product particle. As mentioned in section 9, Chapter 1
found that in these reactions mass and kinetic energy are not separately
conserved, but instead in any frame of reference there is conservation of
total relativistic energy, E = T + mc 2 where T is kinetic energy and m
is rest mass. For the reaction (16-16), the conservation of total relativistic
energy in the LAB frame of reference reads
nucleus,
it is
(J.
where
where
+ ma c*) + m A c* =
+ mB c*) +
(TB
+m
<Jb
c*)
(16-17)
Ta and ma are the kinetic energy and rest mass of a, and so forth, and
TA = since A is stationary in the LAB frame. Because there
o
Before
Figure 16-14.
After
Illustrating
the kinematics of a nuclear reaction.
can be an exchange of energy between kinetic energy and
in these reactions,
be greater, or
it is
less,
than the
TB + T Ta is called
b
the
initial
kinetic energy
of the reaction. That
Q = TB
From
rest
possible for the final kinetic energy
Ta
mass energy
TB + T
The
to
difference
is,
+T-T
(16-18)
(16-17) this can also be written
Q=
+ m A - mB - w )c2
{ma
(16-19)
We see that a measurement of the Q of a reaction gives information about
The Q can be
However, for TB this is usually
difficult and, fortunately, can be avoided by using a relation that comes
from the conservation of momentum. Consider figure (16-14), which
the rest masses of the entities involved in the reaction.
measured by measuring
Ta Th
,
and
TB
LAB frame. By
momentum, an equation is obtained
eliminate TB from (16-17). This is quite easy
represents the kinematics of the reaction as seen in the
applying the conservation of linear
which makes
it
possible to
in the limit
Tjma c 2
where the
<
TJm.c 2
classical expressions
<
such as
1,
TB \mB c* <
Ta = \ma t%
and pa
= ma va can be
THE NUCLEUS
584
we
Consequently,
used.
shall leave
it
as
[Ch. 16
an exercise for the reader to show
that in this classical limit
=T
(l
\
This
is
- Ta (l
+ ^)
m B>
\
Tm m
-^]m Bi mB (T
a
)y>
cos 6
(16-20)
of sufficient accuracy for the analysis of nuclear reactions at the
most experiments. Equation (16-20)
can be determined by a measurement of Ta and Tb
energies which have been used in
shows that the
[Even
if
the reaction
is
being studied in order to find the value of
terms of the other masses, in practice this quantity would already be
with sufficient accuracy for use in (16-20)
and
mJm B <
in typical cases.]
particularly since
The measurements of Ta and
mB
known
b /m B
in
<
use the
techniques employed in the scattering experiments.
In equation (16-19), the masses refer to the rest masses of the nuclei
and B, and to the rest masses of the completely ionized nuclear particles
a and b. However, to the accuracy of the approximation in which the mass
equivalent of the electron binding energy
is
ignored, this equation can also
be considered to read
Q = (Ma +
where the large
refer to the
M A - MB - M
b )c
rest
mass energy of an
+ Z )mc
+ Z A )mc
from the
last
This procedure
Za + Z A
must be
electron.
(16-21)
masses of the neutral atoms.
obtained from equation (16-19) by adding (Za
terms and subtracting (Z B
=ZB + Z
to the
two, where
This
first
mc 2
is
is
two
the
valid since the relation
is
(16-22)
true in any nuclear reaction in order to have conservation of
charge. Rutherford's reaction (16-1) provides a
good example of the use
of (16-20) and (16-21) to supply a connection between the atomic
masses involved in a reaction. The measurements show that for this
reaction
Q =
equivalent,
units, or
1.18 Mev. This energy can be converted to its rest mass
0.00126 amu, by dividing by c 2 and converting to the proper
by using the
relation
1
amu <->
which comes from evaluating the
1 amu.
Then we have
0.00126
rest
Mev
(16-23)
mass energy of a
Mmu M
atomic masses, say M
amu
This allows one of the
931.14
terms of the other three.
2He 4
particle of rest
SQl ,
ao n, to
The
Mm
mass
be determined in
analysis of the energy balance in a large
NUCLEAR MASSES AND ABUNDANCES
Sec. 4]
585
number of reactions has provided measurements of masses which accurately check the measurements by mass spectrometry. Furthermore, the
agreement between these two methods provides excellent confirmation of
the relativistic theory of mass and energy, upon which the energy balance
analysis
is
based.
many atomic masses that have
been measured, and also the mass of the neutron. Note that
z A the
In table (16-1) are listed a few of the
TABLE
(16-1)
Atomic Masses and Binding Energies
Binding Energy (Mev)
V
w
H
Mass (amu)
1.008982 (3)
1.008142 (3)
2.014732 (4)
3.016977 (11)
He3
2
He4
Be 9
8Q16
16
16.00000
29CU 63
29
63
62.94826
50
50
74
92
120
119.94012
184
184.0052
238
238.1234
Sn 128
74\\tt84
92IJ238
Total
Per Nucleon
(AE)
(AEIA)
2.22
1.11
7.72
2.57
4.003860 (12)
9.01494 (16)
28.3
7.07
58.0
6.45
(0)
( 20)
(72)
(11)
(10)
127.5
7.97
552.1
8.75
mass of an atom of given Z and A,
1020
8.50
1476
8.02
1803
7.58
is always quite close to A amu. This is a
= 16.00000. If the amu
consequence of defining the amu such that
a 16
=
which
might
1.00000,
at first seem more
were defined such that
1X
close
A
amu
for
atoms of large A.
would
not
be
to
natural, then
zA
The reader can demonstrate this by doing a little arithmetic.
The same procedure will also demonstrate the much more fundamental
point that the mass of an atom is less than the mass of its constituent
parts.
For example,
2A = 4.003860 amu, whereas the mass of its
constituent parts is the mass of two 1 1 atoms plus the mass of two neutrons, i.e., 2Mia + 2M01 = 4.034248 amu. This mass deficiency of the
atom
is
due to a mass deficiency of its nucleusf resulting directly from the
We can see this by considering
equivalence between energy and mass.
t In the comparison just made, two electron rest masses are included in both
2i4
and 2M + 2M0jl and the mass equivalent of the electron binding energy is negligible.
Thus the mass deficiency can only be due to the nucleus.
THE NUCLEUS
586
[Ch. 16
any one of the four nucleons in the 2 He 4 nucleus. Since the nucleon is
stably bound to the nucleus, it must be moving in some sort of an attractive
real potential representing the net attraction of the other three nucleons.
Furthermore, to be bound
situation
is
it
must have a negative energy E < 0. The
The minimum energy required to
depicted in figure (16-15).
remove the nucleon from the nucleus, leaving it a free nucleon at r ->- oo,
is \E\. Conversely, if a free nucleon comes in from r -> oo and combines
with the other nucleons to form the nucleus, its energy must decrease by
Attractive potential
Figure
nucleon
the
16-15.
in
schematic representation of the potential and total energies of a
a helium nucleus.
amount
\E\.
(The excess energy
or by some other mechanism.)
will
be carried off by a
gamma
ray,
Since the same situation obtains for the
other nucleons of the nucleus, we see that when a dispersed system of free
nucleons combines to form a nucleus the total energy of the system must
decrease by an
amount AE,
the binding energy of the nucleus. According
in the total energy of the system must
to equation (1-24), the decrease
AE
AM in mass, where
AM = AE/c
(16-24)
deficiency AM
4.034248 - 4.003860 = 0.030388
be accompanied by a decrease
its
For 2 He4 the mass
amu. Therefore the binding energy
,
we have used
(16-23).
This figure
is
is
is
AE = AMc 2 =
listed in the
28.3
Mev, where
next to last column of
The last column lists AEjA, the binding energy of the nucleus
number of nucleons it contains. For 2 He 4 this average
binding energy per nucleon is 28.3/4 = 7.07 Mev.
One of the most important features of a nucleus is its average binding
table (16-1).
divided by the
energy per nucleon. This quantity
is
plotted as a function of
in figure
The points are the data obtained from the measured masses in
the manner just described. The smooth curve is obtained from an equation
to be described later. Note that AE/A at first rises rapidly with increasing
(16-16).
A, and then becomes reasonably constant at a value
AE/A
~ 8 Mev
(16-25)
Sec. 4]
NUCLEAR MASSES AND ABUNDANCES
However,
about 8.7
Mev
587
not completely constant at this value;
it maximizes at
and then drops slowly to about 7.5 Mev for
A
240. One consequence of this slow drop in AE/A is the famous
phenomenon of nuclear fission, in which a large A nucleus, such as 92 U 238
splits into two intermediate A nuclei.
This happens because the final
state is more stable than the initial state, since the average binding energy
it
is
for
~ 60
9r-
20
40
60
80
100
120
140
160
180
200
220
240
A
Figure 16-16. The average binding energy per nucleon for a number of nuclei. The
smooth curve is from the semi-empirical mass formula. From R. B. Leighton, Principles
of Modern Physics, McGraw-Hill Book Co., New York, 1959.
per nucleon increases from the value
A
A
nucleus to the value
~8.5 Mev
~7.5 Mev
characteristic of the large
two intermediate
The energy liberated is about 1 Mev per nucleon, or about 1 Mev
per nucleon x 200 nucleons = 200 Mev in total.t Alpha particle emission
is a special case of fission in which one of the final nuclei is 2 He 4 and it
takes place for the same reason. The phenomenon of nuclear fusion
consists of the combination of two or more nuclei of very small A to form
characteristic of the
nuclei.
a larger nucleus which has a higher average binding energy per nucleon,
and therefore is more stable, because its value of A is nearer the value
14
Btu, which does not seem like much energy in terms of this
t 200 Mev = 3 x 10
macroscopic unit. However, the mass, 200^14^ = 200 x 1.7 x 10~ 24 gm = 7 x 10~ 25
lb, of the system emitting this energy is not very large either. The energy emitted per
pound, 3 x 10" 14 /7 x 10- 25 = 4 x 10 10 Btu/lb, is phenomenal; it is about 10" times
larger than the energy emitted per
pound from burning
coal.
THE NUCLEUS
588
A =
AE/A
60 for which
maximizes.
that only a few nuclei near
A =
It
might seem from
60 would be
this discussion
This
stable.
[Ch. 16
not true
is
because there are other factors which tend to inhibit fission and fusion.
Except for A 5 and A
beyond 200, which are
may
lifetimes that they
A from 1 to
have such immeasurably long
there are nuclei for all values of
8,
either stable or
be considered
stable.
100
90
.jr
80
pip
70
60
IIBM^I
50
^
-^
40
30
20
10
\?
10
40
30
20
60
50
80
70
90
100
110
120
130
140
N = (A - Z)
The
Figure 16-17.
distribution of stable nuclei.
Z and (A Z)
have seen that this information can be
obtained from the mass spectrometer measurements. For instance, these
measurements show that for Z = 8 there are stable nuclei for A = 16, 17,
It is
interesting to consider the distribution of the
We
values of these stable nuclei.
The data are plotted in figure (16-17). Each
whose abscissa is (A Z) = N,
the number of neutrons in the nucleus, and whose ordinate is Z, the
number of protons in the nucleus. Note that:
18 (the isotopes of
stable nucleus
1.
for
is
For each
Z=
isotopes
there are generally several values of N.
10 there are stable nuclei for
10
Ne20
10
,
For small
2b. For large
2a.
O).
indicated by a square
Ne 21
Z
Z
10
,
there
Ne 22
is
N=
10, 11,
and
12;
For
instance,
these are the
Z = N.
due to Coulomb
a tendency for stable nuclei to have
stable nuclei
have
Z<
N.
This
is
THE LIQUID DROP MODEL
Sec. 5]
589
repulsions between protons which produce a positive energy proportional
Z2
Coulomb effect were operating, the most stable nucleus
would be obtained for Z = 0, N = A. However, the data
show that for small Z, where the Coulomb effect is small, the tendency
for Z = N dominates and stability is obtained for Z
N. For large
Z the Coulomb effect becomes important and stability is obtained for
to
If
only this
with a given
Z < N.
At
Z=
Z/N =
82,
0.65.
There is also a tendency for stable nuclei to have even Z and even N.
This can be seen from table (16-2), which lists the number of stable nuclei
3.
TABLE
(16-2)
The Distribution of
Stable Nuclei
Number of
Stable
Nuclei
Even
Even
Odd
Odd
Even
Odd
57
Odd
Even
53
166
Even
Odd
of various types. The stable nuclei of odd
7
Nu
i9
40
so
57
La
have even Z
be explained in section
The preceding
free to
i3 8;
Z and
odd
N are H
1
2
,
Li 6 ,
B 10
i76
7i
and Lu
The reasons for tne tendency to
and even N, and also for the tendency to have Z = N, will
23
7.
discussion has been written as
make any adjustments
in
if
a nucleus of given
is
and N, consistent with the condition
Z+ N=
A, which are required to achieve the condition of stability.
not poetic license. Such adjustments can occur by the emission of
electrons or positrons from the nucleus (beta emission), or by the capture
This
is
of atomic electrons in the nucleus. These processes, which will be discussed
in section 11, allow for changes in
constant; that
5.
is,
and
such that
Z+N
remains
they convert protons to neutrons or vice versa.
The Liquid Drop Model and the Semi-empirical
Mass Formula
We shall now use the liquid drop model of the nucleus, and the information
obtained from the study of the distribution of Z and
values for stable
THE NUCLEUS
590
[Ch. 16
nuclei, to obtain a formula for the masses of these nuclei. This model is
based upon the facts that for all nuclei, except those of very small A, the
interior densities are approximately the same and the binding energies
are approximately proportional to the masses of the nuclei {iS.EjA
Both
can be compared with certain others
concerning macroscopic drops composed of some incompressible liquid.
For all such drops the interior densities are the same and the heats of
constant, so Ais oc A).
facts
vaporization are proportional to the masses of the drops.
comparison
is
required to disperse the drop into
it is
The
latter
significant because the heat of vaporization is the energy
its
constituent molecules,
analogous to the binding energy of the nucleus.
and therefore
In developing the
mass formula, we shall use the model to suggest other analogies between
a nucleus and a liquid drop.
The formula contains the sum of six terms,
MZ A =ZMZ,A)
=
(16-26)
The
first
term
is
the
mass of the constituent parts of the atom,
(16-27)
/ (Z, A) = 1.008142 Z + 1.008982 (A - Z)
where the coefficient of Z is the mass of the 1 H 1 atom in amu and the
coefficient of {A Z) is the mass of the neutron n 1 in amu. The remaining
terms correct for the mass equivalents of several
to the total nuclear binding energy.
effects
which contribute
is the term
Of most importance
f (Z, A) =
-M
(16-28)
which accounts for a binding energy essentially proportional to the mass,
or volume, of the nucleus. A similar term would be present for a liquid
drop. The next term is
(16-29)
f2 (Z, A) = +a*A*
a positive correction proportional to the surface area of the nucleus.
In a liquid drop this term would represent the effect of the surface tension
energy; it arises from the fact that a molecule at the surface of the drop
experiences attractive forces only from one side, so its binding energy is
less than the binding energy of a molecule in the interior.
Therefore
simply setting the total binding energy proportional to the volume of the
drop overestimates the binding energy of the surface molecules, and a
correction must be made which is proportional to the number of such
molecules, or to the surface area. The term
f3(Z,A)=+a 3
(16-30)
THE LIQUID DROP MODEL
Sec. 5]
591
Coulomb energy of the charged nucleus, which is
assumed to be a sphere of radius proportional to A y%
A similar term
would be present for a charged liquid drop. The last two terms bring in
accounts for the positive
properties specific to a nucleus.
have
Z=
For one thing, there
N. This can be accounted for by the term
UZ,
A)
is
the tendency to
= +a 4 (Z ~f l2f
A
- Z), or 2Z = A, but
(16-31)
zero for Z = N = {A
which is otherwise
and increases with increasing departures from that condition.
The form used in equation (16-31) is about the simplest one having these
properties, but there is also some theoretical justification that will be presented in section 7. The tendency to have even Z and even N is accounted
for by the term
which
is
positive
- Z) = N even
[Z even, {A - Z) = N odd
\Z odd, (A-Z) = N even
Z odd, (A-Z) = N odd
-f(A),
f5(Z,A)
0,
+f(A),
even,
{A
(16-32)
which decreases the mass if both Z and
are even, and increases it if
both Z and
are odd. The form of the function/^) is determined by
fitting the data. It is found that the best fit for a simple power law is
obtained with
f(A)
a5 A-*
(16-32')
Gathering together equations (16-26) through (16-32'), we have
MZ A = 1.008142 Z + 1.008982 (A - Z) - a A + a^A
x
a s Z 2A~ 1A
ai (Z
AfrfA- 1
\a 5 A-'A
(amu)
(16-33)
This
is called the semi-empirical mass formula because the parameters a
x
through a5 are obtained by empirically fitting the measured masses.
A formula of this type was first developed by Weizsacker (1935). Determinations of the parameters have since been made on several occasions.
Green (1954) found
= 0.01692
= 0.01912
a3 = 0.000763
a4 = 0.10178
a5 = 0.012
ax
a2
(all in
amu)
(16-33')
THE NUCLEUS
592
Using these parameters, quite a good
fit
[Ch. 16
obtained for the measured
is
masses of all stable nuclei except those of very small A the discrepancy is
generally less than several thousandths of an amu. A comparison is shown
;
smooth curve shows the average bindfrom equation (16-33), and the points
show the same quantity evaluated directly from the measured masses.
The semi-empirical mass formula is extremely useful because it describes
fairly accurately the masses of several hundred stable nuclei, and many
more unstable nuclei, in terms of only five parameters.
in figure (16-16).
In that figure the
ing energy per nucleon evaluated
6.
Magic Numbers
The
drop model gives a good account of the average behavior of
and therefore in regard to stability. However,
show significant departures from
nuclei with certain values of Z and/or
this average behavior by being unusually stable. These values of Z and/or
are the magic numbers
liquid
nuclei in regard to mass,
2, 8, 20, 28, 50, 82,
The
situation
is
analogous to the unusual
126
stability
(16-34)
of the electronic systems
of the noble gas atoms containing a magic number
But
electrons.
stability are
2, 10, 18, 36, 54,
86 of
in the nuclear case the effects indicating the unusual
not so pronounced as in the atomic case, and it is necessary
them in order to demonstrate conclusively the "magic"
to consider all of
character of the numbers (16-34).
Mayer
These considerations, largely due to
(1948), run as follows:
1
There is a tendency for nuclei to prefer magic Z and/or N. This can
be seen from consulting figure (16-17). For example, there are 6 stable
isotopes for Z = 20, whereas the average number of stable isotopes in
that region of the periodic table
is
about
stable isotopes, although the average
N=
There are 7 stable nuclei having
in that region of
is only about
3.
For
2.
number
Z=
50 there are 10
is about 4.
though the average number
number of similar examples can
in that region
82, even
be seen.
2.
Figure (16-16) shows that the average binding energy per nucleon
equal to 2 and 8, than
with Z and/or
significantly higher, for nuclei
is
for neighboring nuclei.
Z=N=
stability
These
2.
more
effects
sensitive
is
it
The outstanding example is 2 He4 for which
are even more pronounced if a measure of
,
than the average binding energy per nucleon
is
energy of the "last" neutron or proton
b the binding
in the nucleus, which is the minimum energy required to separate a
considered.
This
is
MAGIC NUMBERS
Sec. 6]
593
neutron or a proton from the nucleus. As an example, the binding energy
of the last neutron in 2 He 4 (the energy required to produce the reaction
2
He 4 -* 2 He 3 + n 1) is 20.6 Mev. The binding energy of the last proton
in
He 4
19.8
is
Mev. These are abnormally
high.
Figure (16-18) shows
a plot of the binding energy of the last neutron for a
The
nuclei.
ordinate
is
abscissa
number of neutrons
the
number of
heavier
the
in the nucleus;
the difference between the actual value of the binding energy of
and the value predicted by the semi-empirical mass
is a smooth function of N (except for the
the last neutron
The
formula.
is
predicted value
+3
+2
3
+1
T
bq
.*
BS
_2
-3
20
28
40
50
1
1
60
80 82
100
120
126
140
N
Figure 16-18.
J.
The binding energy
of the last neutron for a
number
of nuclei.
From
A. Harvey, Phys. Rev., 81, 353 (1951).
effects
of term 16-32), and
intermediate
nuclei with
to
N=
around
28, 50, 82,
large energy required to
3.
it
decreases slowly from around 7
Mev
126
remove
for large N.
is
The unusual
36
17
,
Kr87 and 54 Xe 137
,
of
their last neutron.
less
evidence, such as the fact that for three of the four
8
for
demonstrated by the exceptionally
There are a number of other somewhat
neutron emitters,
Mev
stability
convincing pieces of
known spontaneous
N equals a magic number plus
1.
This implies an unusually small affinity for the extra neutron.
The analogy between nuclear and atomic magic numbers prompted
people to look for an explanation of the nuclear phenomenon
analogous to the explanation of the atomic phenomenon. The reader will
many
key point in that explanation is the formation of closed
by the electrons moving independently in the atomic potential. But,
when the nuclear magic numbers were first being discussed seriously
(1948), it seemed very difficult to understand how any explanation based
upon independent particle motion could be valid for the nucleus. The
reason was that the very successful liquid drop model had been dominant
for a number of years, and it seemed basic to this model that a nucleon in a
recall that the
shells
THE NUCLEUS
594
[Ch. 16
its neighbors. With strong interactions,
would be constantly scattered in traveling through the nucleus
and would follow an erratic path, resembling Brownian motion much
more closely than the motion of an electron moving independently
nucleus interacts very strongly with
the nucleon
through
its
orbit in a closed shell.
From our present
point of view there
is,
however, no problem in under-
standing the independent particle motion of nucleons in a nucleus.
In
a very reasonable extrapolation of the optical model (1952).
The calculations recorded in equations (16-13) and (16-14) show that
fact, it is
A of a nucleon moving through a nucleus increases
by about a factor of 6 as the total energy of the nucleon decreases from
E = + 20 Mev to E = +2 Mev, and that A is several times nuclear
the attenuation length
dimensions at the lower energy.
through E
to E < 0, predicts
sponding to
E<
0,
An
extrapolation of this behavior,
that, for the
the attenuation length
is
that the
bound nucleons can move
nucleus.
Putting this in other words, a nucleon
with
E>
must have a
corre-
quite independently through the
moving through a nucleus
certain probability of suffering collisions, which
involve energy loss in the frame in which the
moving
bound nucleons
very long and, therefore,
CM of the nucleus
is
at rest,
complex potential. However, the energy dependence for E >
of the real and imaginary parts of this potential, indicated
by figure (16-10), makes it easy to believe that for E <
there is no
imaginary part to the potential, so that there can be no collisions with
energy loss in the
frame of the nucleus. But, since essentially all
collisions between a nucleon and another nucleon involve energy loss in
this frame, it follows that if the potential is purely real for the bound
nucleons there can be no such collisions between these nucleons, and that
they must therefore move independently through the nucleus.
because
it is
in a
CM
7.
The Fermi Gas Model
It is comforting to realize that the optical model analysis, of the experimental data concerning the motion through the nucleus of nucleons with
should move independE > 0, strongly suggests that nucleons with E
ently through the nucleus. But it is even more comforting to realize that
<
why this should happen. The
based on the Fermi gas model of the nucleus, which had been
in existence for a number of years, but which had not been held in very
high regard during the period of dominance of the liquid drop model.
This model is essentially the same as the Fermi gas model of the conduction
there
is
a simple theoretical explanation of
explanation
is
electrons in a metal considered in section 6, Chapter 12.
It
assumes that
Sec. 7]
THE FERMI GAS MODEL
move
the neutrons and protons
595
in
an
attractive potential of nuclear
dimensions, and that in the ground state these nucleons
fill the energy
such a way as to minimize the total energy but not violate the
exclusion principle. (Remember that both neutrons and protons are Fermi
particles with spin quantum numbers i
\, just like electrons.) Figure
levels in
(16-19) indicates the energy levels filled by the neutrons in the ground state
of a nucleus. Since a proton
is
we must imagine
energy levels
from a neutron, the excluon the two types of nucleons, and
distinguishable
sion principle operates independently
a separate and independent diagram representing the
filled
by the protons in the ground
state
Figure 16-19. A schematic representation of the energy levels
the ground state of a nucleus.
filled
of the nucleus.
by the neutrons
in
immediately apparent from these diagrams why there can be essentially
between the nucleons when the nucleus is in its ground state.
The point is that all the levels which are energetically accessible are already
It is
no
collisions
and so there can be no collisions except those in which two nucleons
of the same type exactly exchange energy. But the net effect of such an
filled,
exchange of two indistinguishable particles is the same as if there had been
no collision at all. We can understand, then, why for the ground state
of the nucleus the potential is real, and the nucleons are moving independ-
We can also understand how the exclusion principle can be effective
even in inhibiting collisions between a nucleon of low positive energy and
the bound nucleons of a nucleus. In fact, calculations based on the Fermi
gas model have shown that the decrease in the imaginary part of the optical
ently.
model potential with decreasing energy (the energy dependence of
in figure 16-10) is due primarily to the exclusion principle.
Furthermore, the Fermi gas model provides a good estimate of the depth
of the real potential for the ground state of a nucleus. This is obtained
by evaluating the Fermi energy (12-38), which is
in terms of the nucleon mass
and the nucleon density p. Let us consider
the Fermi gas of neutrons in a uniform spherical nucleus of radius
R=
A lA
THE NUCLEUS
596
For a
number of neutrons
typical nucleus the
N=
[Ch. 16
is
0.6/4
Thus
N
iirr^A
gives
0.6A
0.45
1.33ttt$A
nr\
* A
Eh
Ef
-vn
Figure 16-20.
and
the
between the depth of the
Illustrating the relation
energy, and the binding energy of the
Fermi energy
last
potential, the Fermi
neutron.
is
Ef =
A7TYs
h\0A5f
(16-35)
2M; .M_3
Using a radius constant r = 1.25 x 10~ 13 cm consistent with the optical
model potential radius of (16-10'), and evaluating the other parameters,
we
obtain
Ef
~ 34 Mev
relations between the depth of the potential V the Fermi energy Et
and the binding energy Eb of the last neutron are shown in figure (16-20).
As mentioned above, Eb is measured to be roughly 6 Mev for a wide range
of nuclei. Thus the Fermi gas model predicts
The
in satisfactory
V = Ef + Eb
\
^ 34 + 6 = 40 Mev
agreement with the value of about 50
by extrapolating to
(extrapolating the
E<
the values obtained
Mev
that
from the
is
obtained
optical
model
of figure 16-10).
= finds a simple explanation in
The tendency for nuclei to have
terms of the Fermi gas model. Consider a nucleus of very small Z, where
this effect
dominates because the Coulomb potential is very small. In
two independent Fermi gases, the neutrons and the
this nucleus, there are
THE FERMI GAS MODEL
Sec. 7]
moving
protons,
597
same nuclear
in the
The energy
potential.
levels
of these
systems are indicated in figure (16-21). Except for the effect of a very small
positive
Coulomb
potential acting
on the proton system, the energy
levels
of the two systems are identical because for both systems the nuclear
potential
is
the same. It
the nucleus
is
clear, then, that for
is
minimized
if
a given
the levels are filled with
the total energy of
Z=N
since, if this
condition were violated, nucleons would occupy levels which are of higher
Neutrons
Figure 16-21.
protons
'
'
Protons
schematic representation of independent Fermi gases of neutrons and
in a nucleus.
energy than necessary.
quantitative treatment of this argument leads
to the term (16-31) used in the semi-empirical
mass formula to express
According to the equations preceding (16-35),
the Fermi energy for the neutron and proton systems can be written
the t tendency for
Z=
N.
Efn =
C(N\A)K,
Eu =
C(ZjAfA
where C is a constant. Now use (12-40) to evaluate the total energy of the
neutron and proton systems, as measured from the bottom of the potential,
in terms of E and E
This gives
f
f
.
ZA
E =C'^where C"
terms, which
3C/5.
is
The energy of the nucleus
a function only of Z and
C'A-K[N iA
already know, the
condition (16-37),
minimum
is
the
sum
of these two
is
Z*]
minimum of the
found for
(16-36)
since
N+Z=
As we
is
E(Z, A)
This
'
(16-37)
function (16-36), subject to the
N = Z = A/2.
In departing from this
the energy of the nucleus increases, and the corresponding
THE NUCLEUS
598
correction in the semi-empirical
increase.
mass formula
will
[Ch. 16
be proportional to
this
Thus
/4(Z, A)
oc E(Z,
A)
(Z, A) min
C'A~ 2A {N
iA
2[A/2]*}
Z*
2|>4/2]}
Now let
D^*^ = N-* = -Z
2
Then
/4(Z, A)
Use Taylor's
oc
A~ 2A {[AI2 + DfA +
series to
expand the
keeping terms through
Z>
2
.
(-D)]*
two quantities
in the curly bracket,
This gives
*A
+ D
first
[All
[C
-H
5
+ -D
L2.
3 3 2 L
^HMfr+f^
L2
2.
5 2
(-D)2
33
so
'
/4(Z, A)
oc
A-\2
5 2
+ --D
33
cc
A -H
%
2
^4
L2J
- lA
3 3
oc
2
Z2
^-Z) 2 = ^/ ~ >
in agreement with (16-31).
The Fermi gas model
of nuclei to have even
also gives a partial explanation for the tendency
Z and even N.
Consider the even Z, even
nucleus
represented in figure (16-21), and imagine constructing the nucleus of
next higher A by adding one nucleon. To minimize the total energy, the
go into the lowest available energy level. This might be either
level, but let us assume that it is a neutron
level, as shown in the figure. Then this nucleus will contain one additional
neutron. Now construct the next nucleus in the series by adding yet
another nucleon. The nucleon will also be a neutron, since there is still one
vacancy in the neutron level and this level is still of lower energy than the
unfilled proton level. We see that for only one of the three nuclei involved
in this dicussion is
or Z odd, and for none of them are both N and Z odd.
This explanation is indicative of what is happening, but it is not the whole
story. The mass measurements, which demand the term (16-32) in the
semi-empirical mass formula, show that there is also a pairing energy that
nucleon
will
a neutron level or a proton
THE SHELL MODEL
Sec. 8]
599
increases the binding energy of a nucleus
when
NorZ becomes even.
The
origin of the pairing energy lies in the fact that the force between
two
an energy
level, the attractive force which it exerts on the nucleon that was already
there produces a negative potential that lowers the energy of both these
nucleons
nucleons,
is
attractive.
i.e.,
When
the second nucleon
added
to
lowers the energy level in question.
Neutrons
Figure 16-22.
is
Illustrating
why
there are
Protons
'
more neutrons than protons
in
a nucleus of
large Z.
we present figure (16-22), which
form of the potentials acting on the protons and neutrons in
Before leaving the Fermi gas model,
indicates the
a nucleus of intermediate or large Z.
Because of the Coulomb potential,
on the protons is elevated. However, the highest
essentially the same energy for both the neutrons and
the total potential acting
filled level
must be
the protons
if
at
the nucleus
explanation of the reason
is
to be stable.
why
there are
This provides a very graphic
more neutrons than protons
in a
nucleus of large Z.
8.
The
Shell
Model
The Fermi gas model establishes the validity of treating the motion
of the bound nucleons in a nucleus in terms of the independent motion of
each nucleon in a common net potential V(r), which represents the average
attraction of all the other nucleons. In this model, certain quantum
mechanical properties of a system of non-interacting Fermi particles
are brought in by the use of equation (12-38), and the Schroedinger equais not solved explicitly. This is adequate for the description
of certain average properties of the nucleus but, for a more detailed
description capable of accounting for the magic numbers, it is necessary
to bring in more of the important quantum mechanical properties of such
tion for V(r)
a system by actually solving the Schroedinger equation for
V(r).
The
THE NUCLEUS
600
[Ch. 16
treatment of the nucleus employing this procedure is called the shell
model of the nucleus. The reader will note that the shell model of the
nucleus is related to the Fermi gas model of the nucleus in much the same
way
that the Hartree theory of the
theory of the atom.
He
will also
atom
is
related to the
Thomas-Fermi
note that these nuclear models are
cruder than the corresponding atomic theories.
much
In the self-consistent
atomic theories the net potential V(r) is completely determined by the
theory; in the nuclear models the radius and exact form of the net potenV(r) must be inserted ad hoc.
The procedure of the shell model
tial
is
to find the neutron
and proton
energy levels for the potential V(r) of a particular nucleus by solving the
Schroedinger equation for this potential, and then to construct the nucleus
by
these levels in order of increasing energy with
filling
Z protons.
N neutrons
and
depend
Just as in the Hartree theory, the energies of the levels
on the quantum numbers n and
/,
and each
corresponding to the 2 possible values of
level has a capacity 2(2/
and the
(21
1)
1)
possible
was hoped that a form for the potentials V(r) of the
various nuclei could be found in which the ordering and spacing of the
energy levels would be such that an unusually tightly bound level would
completely fill in those nuclei having magic N or Z. However, it was
found that there is no reasonable form for the V(r) which leads even to the
values of
It
ffij.f
proper ordering of the energy
The
levels.
was provided by Mayer, and independently by Haxel,
Jensen, and Suess (1949). These workers proposed that, in addition to
solution
the potential V(r), there
portional to
this interaction
is
a strong inverted spin-orbit interaction, pro-
means that
would be predicted
the net nuclear potential and
L, acting on each nucleon in a nucleus. Strong
much (about 20
is
times) larger than
by using equation (13-27), equating V(r) to
m to the nucleon mass. Inverted means that the energy of the nucleon is
decreased when S L is positive and increased when it is negative, so that
the sign of the spin-orbit interaction is inverted from the sign of the
magnetic spin-orbit interaction experienced by an electron in an atom.
However, as the magnitude of the spin-orbit interaction is proportional
to S L just as it is for an electron, the magnitude of the spin-orbit splitting
will
be approximately proportional to the value of the quantum number
just as
it is
for
The quantum numbers
momentum
/,
an electron.}
vectors,
n,
/,
mu m
s,
as well as the other
quantum numbers, angular
and associated terminology, have the same
significance as in the
theory of the electronic structure of atoms.
t
This
is
the
the potential
quantity y
Lande
is
1
is
interval rule (13-43),
which depends only on the assumption that
L, plus the statement that since j
approximately proportional to /.
proportional to
=
/
\ the
Sec. 8]
THE SHELL MODEL
601
S>
CI
4
?
14
2
6
4
8
10
Figure 16-23.
184
126
126
112
110
104
100
92
70
68
64
58
10
2
50
40
38
32
28
28
20
20
16
14
2Pi/s-
2p
3d
Is
,/2
Without
With
S-L
S-L
The ordering
-3d 5
shell
c
o
2sv
Is
1:
184
168
164
162
154
142
=cr~
+
16
Id
r-t
50
of the highest filled nucleon energy levels according to the
model.
The net
effect
figure (16-23).
of the proposed spin-orbit interaction
The part on
the
left
is
indicated in
shows the ordering and approximate
spacing of the energy levels which nucleons are filling in nuclei with
form of square wells with rounded edges, such as are
potentials V(r) in the
used in the optical model
the part
on the
right
shows the way these levels
THE NUCLEUS
602
[Ch. 16
by the spin-orbit interaction. This is not an energy level diagram
any particular nucleus; it is a diagram giving the order in which levels
appear as the radius of the potential increases with increasing A. Thus
it gives the order in which the highest energy levels of the nucleus are filled.
The diagram is analogous to table (13-2) giving the ordering of the highest
are split
for
filled electronic
energy levels of the atoms, except that
it
also indicates
the relative magnitudes of the separation between adjacent levels.
the presence of the spin-orbit interaction,
and
quantum numbers, and n, /, j, m s must be used.
the levels depend on j as well as on n and /, the
are
In
no longer good
Also, the energies of
larger j corresponding
Each of these levels
number of possible values of m,. This
is shown in the first column on the right in the figure. The second column
shows the total capacity of the levels up to and including the level in
question. The third column shows the same thing for each level which
to the smaller energy since the interaction
has a capacity
lies
(2/
1)
will
inverted.
equal to the
unusually far below the next higher
which
is
level.
Since these are the levels
be unusually tightly bound, we see that the
shell
model with
strong inverted spin-orbit interaction predicts precisely the magic numbers
(16-34).f
Within the past few years, completely independent verification of the
existence of a spin-orbit interaction has been obtained
from experiments
It is found that
involving the scattering of low energy nucleons by nuclei.
a scattered nucleon can have
its
spin angular
momentum
polarized as a result of interacting with the nucleus,
vector partially
and
this
explained only by the presence of a spin-orbit term in the
describing the interaction.
can be
potential
Furthermore, the sign of the polarization
shows that the sign of this term is the same as that required by the shell
model, and the amount of polarization shows that its strength (the spinorbit potential
is
the shell model.
for nucleons
is
several
It
Mev
deep) agrees with the strength required for
should be emphasized that the spin-orbit interaction
not electromagnetic in origin. Instead
nuclear forces, whose origin
is
it is
an attribute of
not well understood.
The spin-orbit shell model can do much more than explain the magic
numbers and their consequences. For instance, it can also explain the
spin angular momentum of the ground states of almost all the nuclei.
t It should be mentioned that the eigenfunctions obtained in solving the Schroedinger
equation for the potential V(r) show that the probability densities for nucleons in the
eigenstates of this potential do not have the property of being large only in a fairly
restricted range of r. Thus in the nuclear shell model there are no "shells" of fairly
which the nucleons are moving. The contrast in this point
between the behavior of the nucleus and of the atom is due to the difference in the form
of V(f) for the two systems.
restricted thickness in
THE SHELL MODEL
Sec. 8]
603
Consider nuclei for which both
and Z are magic, such as 8 16 ^Ca40 ,
208
Pb
According to the model, they will contain only completely filled
subshells of neutrons and protons, and the exclusion principle therefore
,
82
demands
and
that for both the neutron
orbital angular
and
momenta couple
the proton systems the total spin
agreement with the
For nuclei such as 7 N 15 8 17
19
K 39 82 Pb 207 and 83 Bi 209 which contain a magic number of nucleons of
one type and a magic number plus or minus one of nucleons of the other
observation that
,
to zero,| in
for these nuclei.
type, the exclusion principle
that the total angular momentum
momentum of the extra nucleon or hole.|
demands
of the system be the total angular
shows that these nuclei should have i = \, f, f, |, and
agreement with observation. Now consider nuclei
f
for which
and/or Z are further away from the magic numbers. These
nuclei will contain subshells with more than one nucleon or hole, and the
problem of how the spin and orbital angular momenta of the nucleons in
these subshells couple is analogous to the same problem for the electronic
structure of an atom. However, there are important differences. Most
atoms seem to be examples of LS coupling, but most nuclei seem to be
examples of JJ coupling. Of course, this is simply a consequence of the
strong spin-orbit interaction in nuclei. It appears that another consequence
Thus
,
figure (16-23)
respectively, also in
that the coupling in nuclei is much simpler than in atoms. Consider
any nucleus with even N and even Z. As mentioned in section 1, all such
nuclei are found to have i = 0. It is clear that this means that, whenever
there are an even number of nucleons in a subshell, the total angular
is
momenta of these nucleons
the JJ coupling
simply couple to zero. If it
is
now assumed that
strong enough that the addition of one
more nucleon to
the subshell does not destroy the coupling of the nucleons that were already
is
there, the total angular
momentum
of the whole subshell will be that of
"odd" nucleon. With this assumption, the entire angular momentum
of an odd A nucleus will be due to the total angular momentum of the single
odd nucleon in the highest occupied energy level, and i will be equal to the
value ofj for that level. With only one or two exceptions, this assumption
allows the observed values of i for all odd A nuclei to be explained in
the
terms of figure (16-23). In doing this
it is, however, necessary to allow for
occasional interchanges of the ordering of certain closely spaced energy
levels,!
The
because of the
effect
of the pairir^g energy.
spin-orbit shell model, plus the additional assumption just
mensomewhat less satisfactory explanation of the magnetic
moments of odd A nuclei. The prediction is that the magnetic moment of
the nucleus is due to the magnetic moment of the single odd nucleon
tioned, provides a
t Cf. discussion associated with table (13-9).
X Recall that this also happens for the atomic energy levels.
THE NUCLEUS
604
[Ch. 16
because the contributions of the other nucleons cancel one another. Now it
is easy to calculate the magnetic moment of a single nucleon of given j in
terms of its orbital g factor (which is 1 for the proton and for the neutron),
and in terms of its spin g factor (which is such as to give z components of
intrinsic magnetic moment
2.79,m y for the proton and
1 .91^
v for the
neutron, where the sign gives the orientation relative to the spin). For
protons, two values are found corresponding to the two possibilities
= +
I
I
f r neutrons, two other values are found.
I an d j
i
plot of these two values of magnetic moment as a function of j forms the
'>
When
so-called Schmidt lines.
neutron Schmidt
it
lines are
N nuclei,
compared with those of even Z, odd
observed values, but that the agreement
N nuclei,
is
far
from
perfect.
of the
This shows
only an approximation to assume that an even number of nucleons
it is
always couple to give zero total angular
is
Z, even
seen that the predictions agree with the general trend
is
that
compared
and the
the proton Schmidt lines are
moments of odd
with the observed magnetic
good enough
spin angular
momentum
because this quantity
is
quantized
so, if there is
number of nucleons occasionally have a
angular momentum, there must also be a corresponding
some probability
non-zero total
momentum. The approximation
to lead to the prediction of correct values of the nuclear
that the even
odd nucleon have
compensate and keep the
probability that the
exactly the right total angular
momentum
total
to
system constant. But,
which
not quantized, can be appreciably different from the predicted
Apparently it does happen.
The
The
momentum of the
moment of the system,
angular
happens, the magnetic
is
value.
9.
if this
Collective Model
model
shell
nucleus
move
is
based upon the idea that the constituent parts of a
The
independently.
liquid
drop model implies
just the
opposite, since in a drop of incompressible liquid the motion of any
constituent part depends on the motion of
The
all
the neighboring parts.
between these ideas is indicative of the current state of the
theoretical understanding of the nucleus. It also emphasizes that a model
provides a unified description of only a limited set of phenomena, completely without regard to the existence of conflicting models used for the
conflict
description of other sets.
Contrast this with a theory, such as relativity
theory or quantum theory.
large set of
sets
phenomena and,
of phenomena,
it
theory provides a description of a very
border lines between this set and other
at the
fuses without conflict into the theories used for
the description of the other
sets.
THE COLLECTIVE MODEL
Sec. 9]
As
in every scientific field,
physics proceeds
on two
605
the theoretical development of nuclear
levels.
On
one
level,
people develop models,
which correlate experimental results and provide simple descriptions of the
phenomena involved. Then they constantly attempt to broaden the range
of applicability of each model, to remove any apparent conflicts between
the various models, and finally to fuse these models into a single unified
model. On another level, people attempt to develop theories, which
explain the properties of the models from the fundamental laws of nature.
a theory of the nucleus would be able to explain the
of the parameters used in the semi-empirical mass
values
numerical
the laws of force between nucleons.
terms
of
in
formula,
As an example,
parallel development of models and theories was not very evident
our discussion of atomic physics, because that is a field in which there
are now acceptable theories. At this stage, the models lose much of their
The
in
importance and so are used in a discussion only if they are of particular
historical interest, or are of pedagogical value, or provide convenient
calculational tools. An example is Bohr's description of atomic structure,
a model which
we discussed at length for precisely these three reasons.
we shall describe some of the attempts which have been
In later sections
and theories of nuclear forces. In
the present section we shall describe an example of one of the attempts to
fuse two apparently contradictory models into a unified model. This
been
is the collective model of A. Bohr and Mottelson (1953), which has
the
and
model
shell
features
of
the
quite successful in combining certain
made
to develop theories of the nucleus
The model assumes
liquid
drop model.
move
independently in a real potential.
that the nucleons in a nucleus
However, the potential
is
not
the static spherically symmetrical potential V(r) used in the shell model;
instead it is a potential capable of undergoing deformations in shape.
These deformations represent the collective motion of the nucleons in the
nucleus
the type of motion associated with a liquid drop.
As
in the shell
which are split
by the same spin-orbit interaction, and form a core containing an even
number of protons and an even number of neutrons, plus (for odd A
model, the nucleons
fill
the energy levels of the potential,
odd nucleon. As is also true in the shell model, the odd
nucleon plays an important role in determining the properties of the
entire nucleus. However, in the collective model the odd nucleon is not
nuclei) a single
entirely responsible for these properties because the core is not inert
and
can have angular momentum, etc. The instantaneous shape of the core
Its departure
is described by the instantaneous shape of the potential.
nucleon
and, in
from spherical symmetry affects the motion of the odd
latter
The
core.
of
the
turn, the motion of this nucleon affects the shape
coupling
is
due to what can be described as a centrifugal reaction exerted
606
THE NUCLEUS
[Ch. 16
by the odd nucleon on the wall of the potential. The net effect is that a
"tidal wave" circulates around the surface of the core, following the
motion of the odd nucleon. As the reader might guess, calculations based
on this model are very complicated. Consequently we shall only describe
the results.
It is found that the collective model preserves all the features of the shell
model which agree with experiment, and also provides new features which
allow this agreement to be considerably extended. For instance, in this
model part of the total angular momentum of the nucleus is carried in
the form of orbital angular momentum by the tidal wave circulating around
the surface of the core.
and
This moving deformation constitutes a current
moment proportional to the associated orbital
gives rise to a magnetic
angular momentum. The proportionality constant is, however, smaller
than the usual expression (11-7 with the sign changed and the electron
mass replaced by the proton mass) by a factor ZjA, which is the ratio of
the number of charged nucleons to the total number of nucleons. Consequently, that part of the spin angular momentum carried by the core
does not contribute the same magnetic moment as it otherwise would,
and the change is exactly what is required to explain the discrepancy
between the Schmidt lines and the measured nuclear magnetic moments.
Another feature which can be explained quite nicely in terms of the
model is the nuclear electric quadrupole moment Q. We have
collective
given a qualitative definition of this quantity on previous occasions
(cf.
moment about
the
section
Quantitatively, the electric quadrupole
1).
z axis is defined
Q =jp(x,
by the equationf
y, z)[3z
(x 2
z )]
dr
Z[_3z~
(x
?)]
(16-38)
where p(x, y, z) is the nuclear charge density in units of proton charges,
and where the integral is taken over the nuclear volume. The value of
Q is Z, the number of protons in the nucleus, times the average over their
2
charge distribution of the quantity [3z 2
(x 2
z 2 )].
It is clear
y
then
that
Q=
that case x 2
must
at least
if
p(x, y, z)
spherically
symmetrical,
since
in
2
2
If p(x, y, z) is not spherically symmetrical, it
y = z
have an axis of symmetry along the direction about which
.
the nuclear spin angular
taken as the
is
z axis,
momentum
vector precesses.
In (16-38) this
and the equation immediately shows
p(x, y, z) is elongated in the z direction so that z 2
t See also the footnote
on page
646.
>
x2
that
2
,
Q>
is
if
and that
Sec. 9]
Q<
THE COLLECTIVE MODEL
flattened in the z direction so that z 5
if p(x, y, z) is
Figure (16-24) shows a plot of
odd A
stable
607
known
the presently
all
< a? =
measured under circumstances in which
nuclei,
+ 14
71
is
175
Lu
+ 13
y*.
values of Q, for
+ 12 i
+ 11
+ 10 +9 -
+8
/
+7 -
t
I
+6
o +5
<
11
T
O-
CO
+4
+3
+2
'I
X
/
*J\1
v v
*"xJ
XI
'!
\l
-1
'
+1 ~
1
i
-2
1*
v
1
X
X
-3
-4
fill
2
20 28
50
82
126
40
50
60
70
80
90
1
1
10
20
30
ZorN
Figure 16-24.
The quadrupole moments
100 110 120 130 140
of stable odd
nuclei.
maximized because the nuclear spin angular momentum vector has
maximum possible component along the z axis (m t = i). The abscissa
is Z for odd Z, even JV nuclei, and the values of Q for these nuclei are
its
indicated with crosses;
and the values of
the abscissa
is
for even Z,
odd N nuclei,
The ordinate
for these nuclei are indicated with dots.
THE NUCLEUS
608
is
the ratio of
Some
to the square of the nuclear charge distribution radius.
of the features of this plot agree qualitatively with what would
be expected from the
Q<
[Ch. 16
shell
an odd Z, even
for
The reason
model. For example,
this
model would predict
N nucleus with Z equal to a magic number plus
that such a nucleus contains only completely filled proton
which the charge distribution is spherically symmetrical,
plus one proton moving in an "orbit" near the x-y plane, for which
1
is
subshells, for
<
x2
even
nucleus with
z2
quadrupole
2
.
The
shell
model would
moment would be due
orbit near the x-y plane.
also predict
equal to a magic
entirely to a
Thus the
fact that
positive to negative values each time the
Q>
for
number minus
an odd Z,
one, since the
proton hole moving
in
an
goes through zero from
number of odd protons goes
through a magic number, and also the concomitant fact that Q goes
through zero in the opposite direction somewhere between each pair of
magic proton numbers, can be understood on the basis of the shell
model.| However, the magnitudes of Q cannot be understood in terms of
this model. According to the shell model, the quadrupole moment of
nucleus is entirely due to the odd proton since the core
any odd Z, even
containing an even number of nucleons has zero total angular momentum
and must, therefore, have a charge distribution which appears to be
spherically symmetric because there is no way to define a symmetry axis.
Now
is
it
apparent from equation (16-38) that the
maximum
possible
due to a single odd proton is about Q = R 2 where R is the
charge distribution radius. But figure (16-24) shows that Q is observed
nuclei. Also,
to be very much larger than this for many odd Z, even
nuclei
in the shell model the quadrupole moment of all even Z, odd
value of
would be very small because
it
could arise only from the very small
displacement of the center of the charged core from the center of the
entire nucleus. However, the figure shows that the value of Q for odd
nuclei
Z, even
is
observed not to depend significantly on whether they are odd
The collective model provides an explanation
N or even Z, odd N.
It leads to large enough values of Q between the
magic numbers simply because the core can be deformed so that many
protons contribute to the total quadrupole moment. As the deformation
can be due to either an odd proton or an odd neutron, the collective
model explains why the values of Q are similar in both cases.
for all these features.
t Note that Q does not behave this way at 20. This, and other recent evidence, has
been interpreted to mean that 20 is somewhat less magic than the other magic numbers.
Also note that at 40 (corresponding to the filling of the ip\$ level) and 14 or 16 (corresponding to the filling of the 3<% or 3si^ levels) Q behaves as it does at magic numbers.
However, 40 and 14 or 16 are not considered magic because none of the other usual
effects is observed.
ALPHA DECAY AND
Sec. 10]
FISSION
609
we have confined our
attention to nuclei which are
Alpha Decay and
10.
Up
stable
to this point
at
Fission
have immeasurably long
least in the sense that they
The
qualification
with
A much
is
A
AEjA maximizes,
~ 60,
larger than the value
energy per nucleon
we saw
necessary because, as
which the average binding
at
are energetically unstable to fission.
Let us
now
decay.
In this particularly important special case of
discuss this interesting decay process, starting with alpha
2
spontaneously emits an alpha particle
greater than the
is
mass
m 2A
sum of
its final
He 4
mass
The energy equivalent of
because
fission,
its initial
mz _ 2A _ i and the
the mass difference
it
Mev
- (M z _
\_M Z A
,
If the semi-empirical
this equation,
or 6
carried
away
can be written
atomic masses as|
mz
alpha particle
is
the electron binding energies, this alpha decay energy
of
a nucleus
mass
Ignoring the mass equivalents of
as kinetic energy of the alpha particle.
in terms of
lifetimes.
in section 4, all nuclei
mass formula
predicts that
2 , 4 )]c
(16-39)
in
slowly with A, reaching values
However,
for the heaviest nuclei.
used to estimate the masses
is
E increases
accurate enough to be useful because
2 ,^_ 4
this prediction is
ignores magic
it
number
not
effects,
which are very important for nuclei of particular interest in alpha decay.
Figure (16-25) shows the values of E for the nuclei in the alpha emitting
range, obtained from direct measurements of the kinetic energy of the
alpha particle with range or momentum analysis techniques, and/or from
We see that the
E rises to a peak about 4 Mev above the general trend,
general trend slowly increases with A as predicted by the
the use of equation (16-39) with measured masses.
alpha decay energy
and that the
The peak occurs
semi-empirical mass formula.
is
easy to understand why.
N=
126.
The
is
82
because
Pb208 which has magic
,
about 4
[Figure (16-18) shows that the increase
Mev
and
it
Z=
82
increase in alpha decay energy arises because
the binding energy of the daughter
and about 2
In emitting an alpha particle, this parent
nucleus decays to the daughter nucleus
and magic
just at 84 Po 212
is
is
We
magic]
Mev higher than average.
Mev because Z is magic
about 2
also see
from
figure (16-25)
that the energies of the known alpha emitters range from 8.9 Mev for
84p 2i2 t0 4i ]yj ev for 90^^232
te, then, that all these energies are far
below the maximum energy of the Coulomb barrier acting on the alpha
particle, which is about 30 Mev. Therefore the alpha particle can escape
t
There
is
no spontaneous emission of
"n 1 ,
1
,
2
,
or
He 3 from
the naturally
occurring nuclei because the values of E, calculated from equations similar to (16-39),
are negative for
all
these processes.
THE NUCLEUS
610
[Ch. 16
the nucleus only by the process of barrier penetration discussed in section
3,
Chapter
8.
In equation (8-58) of that section,
T=
_2
its
the expression
v'ewrtw-^J*
an alpha
for the probability that
potential barrier V(r), if
we obtained
particle of
kinetic energy far
(16^10)
mass
will penetrate the
from the nucleus is E. This
10i
Cf
Fm
260
The alpha decay energy for the nuclei in the alpha emitting range.
Perlman, A. Ghiorso, and G. T. Seaborg, Phys. Rev., 77, 26 (1950).
Figure 16-25.
I.
equation
is
From
actually valid only in the case of emission of alpha particles
of zero orbital angular momentum because only for that case is the time
independent Schroedinger equation (15-76), in the radial coordinate r,
the same as the time independent Schroedinger equation, in the rectangular
coordinate x, which was used in the derivation of (16-40). For the emission
of alpha particles of non-zero orbital angular
of u in equation (15-76)
is
more
negative;
momentum,
this
the coefficient
has the same
effect as
'
ALPHA DECAY AND
Sec. 10]
making the
FISSION
611
and therefore the probability
potential barrier V(r) higher,
T is reduced. However, the reduction is negligible if
momentum is not too large, and usually it can be ignored.
of penetration
angular
the
Let us evaluate equation (16-40) by using the slightly simplified form
In the alpha decay of a parent nucleus
for V(r) indicated in figure (16-26).
of atomic number Z, the potential
is
2(Z
2)e
V(r)
\vw
E
,
Figure 16-26.
The
integral of (16-40)
radius
r'
r'
slightly simplified alpha particle-nucleus potential.
is
taken from the nuclear potential radius
to the
satisfying the equation
2(Z
2>2
Thus
T=
Changing
to the variable
2V2
\/iM
n
fr
J
the integrand becomes sin 2 p
Since Rjr'
may
is
fairly
2(Z
and the
_
2)e
integral immediately gives
-^4-- (?)-(f)(-f) i
,
-2
e
2(.Z-2)e'
I:
p defined by
COS p
T=
'
RV
small compared to
be expanded. The result
T~ e
1,
the terms in the square bracket
is
V 2MB
fM-
H
(f) ]
THE NUCLEUS
612
Evaluating
r',
becomes
this
[Ch. 16
e<-
which gives
1
when
As in
is
~g
2.97(Z -1)
written in units of
(8-59),
we
shall
A li A -3.95(2 -2)E~ A
Mev and R
now assume
is
(16-41)
written in units of 10~ 13 cm.
that there
is
one alpha particle with
energy E, which makes vj2R attempts at the barrier per second, where
There is little on which to base this
assumption because little is known about what an alpha particle is doing
in a nucleus before it is emitted. However, it does allow us to write for
the decay rate X, which is the probability per second that the nucleus will
decay by alpha emission, the simple expression
is
its
velocity inside the nucleus.
(16-42)
2R
Substituting for
In X
J from
(16-41),
and taking logarithms, we have
~ In + 2.97(Z - 2JA R Vi
3.95(Z
2)E~ H
(16-43)
2/v
This makes
it
apparent that X
the coefficient of
is
in (16-42).
form of
show the
actually insensitive to the exact
In figure (16-27), the points
measured values of In X for all the naturally occurring alpha emitters,
and the curve shows the values of this quantity predicted by equation
Equation (16-43)
(16-43), using R = 8 x 10~ 13 cm and Mv 2 /2 = E.
can be plotted, approximately, as a function of (Z 2)E^A alone because
there is essentially no change in In (v/2R), and very little change in
A R'A over the restricted range of nuclei involved. The only
change comes from 3.95(Z 2)E~ A which varies because E varies
from about 4 to about 9. We see that this leads to a variation of about
2.97(Z
2)'
55 in In
X,
corresponding to a variation
in X
of about a factor of e 55
10 25
The correlation between the extremely rapid variation of X and the slow
variation of E was first pointed out by Geiger and Nuttall (1911) from a
compilation of the experimental data. They devised an empirical formula
giving the essential features of the energy dependence of the theoretical
equation (16-43), which is due to Gamow, Condon, and Gurney (1928).
Now consider a system containing N(0) unstable nuclei of the same
species at time t = 0. If the probability per second that any one of the
ALPHA DECAY AND
Sec. 10]
nuclei will decay
time
is
FISSION
the decay rate
A,
the
613
number of
nuclei remaining at
t is
N(t)
and the
N(0)e' u
(16-44)
lifetime r, or average time required for
one of the nuclei to decay,
is
(16-45)
1/1
These relations are (13-116) and (13-117), which were obtained for the
decay of a system of excited atoms by the emission of quanta. But a
30
40
35
(Z-2)E-% {Em
Figure 16-27.
An series;
The
relation
4n
between the alpha decay
2 series;
of Modern Physics, McGraw-Hill
A=
Book
Co.,
4n
New
45
50
Mev)
3 series.
rates and alpha decay energies.
From
R. B. Leighton, Principles
York, 1959.
review of their derivation will show that they apply equally well to any
system where a constant probability per second of decay can be defined.
more
In a
typical system, there are several different nuclear species
For instance, 92 U 234 decays by
alpha emission into
Th
this nucleus decays by alpha emission into
88
Ra 226 this nucleus, in turn, decays by alpha emission into 86 Em222
and this nucleus decays by alpha emission into 84 Po 218 Consequently,
which decay successively into each other.
90
230
a system, initially
filled
with
92
U 234
will eventually contain a
mixture of
all
The decay of such a radioactive series is indicated schematifigure (16-28). The symbol N (t) is the number of parent nuclei
these nuclei.
cally in
is the probability per second for
t, and the symbol X
decay of these nuclei. Similar symbols are used for the numbers and decay
in the system at time
rates of the first daughter nuclei, the second daughter nuclei,
and so
forth.
THE NUCLEUS
614
[Ch. 16
In this system, the relation (16-44)
is valid for the parent only. However,
analogous relations for the daughters can be obtained from the following
set of equations
dN
(t)
= -N (OAo
dt
dNjft)
=N
(t)^
(16-46)
NjHOXj,
dt
^ = Nmi ~ N
dJ
2 (t)X,
dt
(t)
-M>
Figure 16-28.
Ntft)
Illustrating
-^->(
2 (t)
the decay of a radioactive series.
The negative term on the
right of each equation represents the loss of
due to decay into the next nuclei of the series; the
positive term represents the gain of nuclei per second due to decay from
nuclei per second
A number
the previous nuclei of the series.
of interesting solutions exist
for this set of coupled differential equations, depending
conditions and the constants
One
solution of particular importance
small compared to
very
much more
all
the other X.
slowly than
left sides
of
initial
for the case
when
is
very
the parent of the series decays
daughters and, after sufficient time,
all its
all
is
Then
a condition of equilibrium will obtain.
constant, so the
on the
X.
In equilibrium,
all
the N(t) are
the equations (16-46) go to zero.
(This
can be exactly true for all the equations except the first. For the first it
can obviously be only an approximation; but it is a very good approximation if X is very small as assumed.) Applying this condition, we have
= N (t)X = ^(OAi -
NjUt)^.
JV2 (0A 2
which gives
N (t)X =
The equilibrium obtains
independent of the
after a time
initial
= N (t)X =
N^t)^
(16-47)
>
such that Xt
1 for all Xt except X t,
This case is important because the
conditions.
dependence of E, and therefore of
X,
on
is
usually such as to satisfy the
ALPHA DECAY AND
Sec. 10]
FISSION
615
requirement that X be very small compared to all the other X [cf. the part
of figure (16-25) between A = 240 and A = 210]. Note that the relation
(16-47) shows that in equilibrium the number of nuclei of a particular
species
relation
is
is
inversely proportional to the decay rate of that species.
often used in determining the values of the X
of the N(t) and one
We
now
known
This
from measurements
X.
understand how alpha emitting nuclei with
very short lifetimes can be found in nature. For example, 84 Po 212 with
lifetime of about 10~ 6 sec, can be extracted from certain naturally occurare
in a position to
ring minerals
which presumably have been in existence for millions of
Of course, the reason is simply that the short lifetime alpha emitters
members of radioactive series with parents of long lifetime, and are in
years.
are
equilibrium in these
There are three such series which occur
whose parent is 9 Th 232 with lifetime r = 2.01 x
10 10 years, the An + 2 series whose parent is 92 U 238 with lifetime t =
6.52 x 10 9 years, and the An + 3 series whose parent is 92 U 235 with lifetime
t = 1.02 x 10 9 years. The names characterize the values of A for the
nuclei which are members of the series. For instance, the parent of the
4/3 + 3 series has A equal to 4 times an integer plus 3, where the integer is
58. Since each alpha decay reduces A by 4, all the daughters of this series
will also have A equal to 4 times some smaller integer plus 3.
In addition to the An, An + 2, and An + 3 series, there is evidently also
room for a An + 1 series. Actually there is such a series, and its parent is
93
Np 237 with lifetime t = 3.25 x 106 years. The series can be produced
naturally:
series.
the An series
artificially by using a nuclear reaction to make the parent, but it is not
found in nature since the lifetime of the parent is very short compared to
the age of the earth (estimated from certain geological and cosmological
evidence to be of the order of 10 9 years), and therefore any parent nuclei
initially present have decayed away. In this connection note that figure
(16-25) shows that the alpha decay energies of the parents of the three
naturally occurring series
than
Mev
happen
to be particularly low.
If they
were
less
would have lifetimes short compared
to the age of the earth, all the alpha emitters would have long since decayed
away, and the naturally occurring elements would stop at Z = 82 instead
of Z = 92. The same figure makes it clear why the naturally occurring
elements do stop at Z = 92: it is because the alpha decay energies of all
nuclei with Z > 93 are large enough to le.ad to lifetimes short compared
1
higher, these parents too
to the age of the earth.
Finally, a mental extrapolation of figure (16-25)
an actual calculation of the alpha decay energies from the
measured masses, shows that the corresponding elements are apparently
to
Z<
82, or
stable to alpha decay because the alpha decay energies are so small that
the lifetimes are immeasurably long.
Strictly speaking, a
number of
THE NUCLEUS
616
[Ch. 16
Z 82 are unstable, but their lifetimes are so long that decay
cannot be observed.
Alpha decay occurs essentially because the total Coulomb repulsion
nuclei with
energy of the protons in the nucleus
decreased by the emission of a
is
fragment containing two of these protons (the alpha
larger decrease in this energy
is
containing a larger fraction of the protons.
99
A
Figure 16-29.
particle).
But an even
obtained by the emission of a fragment
If the nucleus splits into
w9
schematic representation of the steps involved
two
in a fission process.
Coulomb repulsion energy of the system is obviously
minimized when the parts contain equal numbers of protons. The calculation performed in section 4 shows that, when a nucleus of large Z
undergoes such a fission process, the decrease in the energy equivalent of
the total mass of the system is about 200 Mev. This is almost entirely due
parts, the total
6 Mev
V(0)
\Surface
I i
tension
\
~2200 Mev
^W
Coulomb,
repulsion
V(=o)
Figure 16-30.
An energy diagram
to the decrease in the total
of this energy
is
carried
Coulomb
away by
for a fissionable nucleus.
repulsion energy of the system.
Most
the two fission fragments as kinetic energy.
steps which are involved in a fission process are indicated schematiby the set of drawings in figure (16-29). These drawings define
(somewhat unprecisely) a parameter s characterizing the progress of the
The
cally
process.
It is
convenient to use this parameter and, in particular, tplot
on s. This is shown in
that part of the energy of the system which depends
by the function V(s). Starting at small s, there is relatively
change in the Coulomb repulsion energy with increasing s, but the
figure (16-30)
little
ALPHA DECAY AND
Sec. 10]
FISSION
617
surface area of the nucleus increases rapidly. From the clearly appropriate
point of view of the liquid drop model, the increase in surface area produces an increase in the surface tension energy. Thus V(s) increases with
increasing s for small
As
s.
s continues to increase, the surface tension
"neck down" and eventually to split. After it
the surface tension energy no longer changes with s, and V(s) de-
effect causes the nucleus to
splits,
creases with increasing
s, following the decrease in the Coulomb repulsion
energy of the two fragments. Since V(s) first goes up and later comes
down, it necessarily must pass through a maximum. Calculations based
on the
drop model show that for a
liquid
maximum
typical nucleus of large
this
about 6 Mev above K(0). We already know that K(0) is
about 200 Mev above K(oo).
Figure (16-30) shows that nuclei are normally stable to decay by fission
since they are sitting, with total energy E = K(0), at the bottom of the
is
depression in the potential V(s).
penetration but, because the mass
The process can take place by
M entering
(16-40) for the barrier penetrability
penetration
extremely small.
this spontaneous fission process,
theless,
energy
is
spontaneous
is
fission
is
in the
barrier
exponent of expression
very large, the probability of barrier
For instance, if 9 2 U 238 decayed only by
its lifetime would be ~10 16 years.
Never-
can actually be observed, because so much
it is very easy to detect.
One
liberated in the fission process that
technique often used involves collecting the large number of ions produced
fission fragments as they come to a stop in the gas filling of a device
by the
called an ionization
process of
chamber.
much more importance
is induced fission.
Usually this is
brought about by the nucleus capturing a low energy neutron. As the
binding energy Eb of the last neutron in a nucleus of large Z is around
5 or 6
Mev,
energy to put
in favorable cases the capturing nucleus receives
it
over the top of the fission barrier.
Very often
enough
this
energy
actually does go into vibrations of the type which lead to fission. Induced
fission is perhaps the best example of the collective motions that
are
implied by the liquid drop model. The process is illustrated in terms of
an energy diagram in figure (16-31). For 92 U 233 and ^U 235 the binding
Eh is enough to make the total energy E higher than the top of
the barrier. For 92 U 238 it is not quite enough, and the neutron must also
carry in about 1 Mev of kinetic energy. The difference between the two
,
energy
cases reflects the fact that
capture a neutron,
when
is
about
b,
1
for the nuclei
Mev
formed when
greater than
92
U 233 and
for the nucleus
92
u 235
formed
92(j238
(16-32)
is
ca ptures a neutron. The reason is that the pairing energy
negative in the first case and zero in the second case. Fission
can also be induced by the capture of a gamma ray of sufficient energy, or
by the capture of a high energy proton (a low energy proton will not be
THE NUCLEUS
618
captured because
about 15
The
arises
Mev
it is
repelled
from the nucleus by a Coulomb
[Ch. |6
barrier
high).
produce power in a chain reaction
3 neutrons are emitted, on the
process. This can be understood by considering
possibility of using fission to
from the circumstance that 2 or
average, in each fission
figure (16-32).
The
figure plots the
An energy diagram
Figure 16-31.
Figure 16-32.
are the
most
Z and N
Illustrating
values of the nuclei which
illustrating induced fission.
why neutron
emission occurs
in fission.
stable for each value of A, as in figure (16-17).
These nuclei
are represented by the shaded line of stability. The large dot indicates the
fissioning nucleus. The two small dots indicate the two fragments which
result
from the
fission.
These fragments have the same
fissioning nucleus but, since their values of
ZjN
are smaller
ratio as the
by a factor of
ZjN ratios are smaller than those of stable nuclei with these values
A the fission fragments have too many neutrons. Most of the necessary adjustment in the ZjN ratio takes place by the beta decay process to
2, their
of
be discussed in the next section, but part of it is achieved simply by emitting
a few neutrons. These neutrons typically have kinetic energies of several
Mev. However, they rapidly lose energy by processes involving scattering
ALPHA DECAY AND
Sec. 10]
FISSION
619
with energy loss in the material containing the fissioning nuclei and are no
longer able to induce fission in 92 U 238 This is why that isotope is usually
.
not directly useful in a chain reaction. The slowing down of the neutrons
emitted in fission is, however, not undesirable since, when their energy
becomes so low that
of the
92
de Broglie wavelength
their
is
larger than the radius
by these nuclei increases
with increasing wavelength or decreasing energy.| For a chain reaction
nuclei, the cross section for capture
10
t lO"
10~ 2
10"
10"
10"
60
80
100
120
A
Figure 16-33.
The
140
160
180
distribution of the masses of fragments emitted in fission by thermal
neutrons.
it is
important to capture as
many of the
neutrons as possible; the reaction
one of the neutrons emitted in each fission
is later captured. In reactors, advantage is taken of the energy dependence
of the capture cross section by scattering the neutrons from some material
of low atomic weight so that they lose energy by recoil and eventually
come into thermal equilibrium at the very low kinetic energy %kT~ 0.025
ev. This is not done in a bomb because it takes too long (~10~ 6 sec).
is
self-sustaining only if at least
amount of
92
U 238
can be used indirectly to
These nuclei capture neutrons and, instead of fissioning,
the resulting unstable nuclei 92 U 239 undergo two successive beta decays,
turning into the stable nuclei 94 Pu 239 which have the same fission properties
In breeder reactors a certain
produce
as
fissions.
92
U 233 and 92 U235
We have simplified our discussion of fission by speaking as if the process
.
always leads to two equal fragments.
infrequently.
Actually this happens only very
Figure (16-33) shows the distribution in the
t This should be apparent from the discussion of the previous chapter.
values of
THE NUCLEUS
620
from
distribution
typical.
to
N=
is
U 235 by
fission of 92
the fragments
neutrons of thermal energy. This
Since one peak occurs near
50, while the other occurs
near
[Ch. 16
values corresponding
values corresponding to TV
82,
thought that these bimodal distributions arise from the influence of
the magic numbers. However, the details of their origin are not well
it is
understood.
II.
Beta Decay
We
have also simplified our discussion of the radioactive series by
fact, an important
feature of these series. This is illustrated by the processes occurring in the
An series, plotted in figure (16-34). In addition to alpha decay processes,
ignoring the processes of beta decay, which are, in
90^h 228
90
d-
^R a 224
88
/ \J
89^ c 228
86
E m 220
85
At
N
84
84
*;
82
82
j>b
208
212
&
\j/1
j
.j
208
/ X? >/\
/ \}/ ^
/ Jy \V
81
216
pj*
M/
Th 232
88jla 228
*/
86
90
B AW
82
pb 212
*Bi 212
80
78
124
126
128
130
132
134
N
Figure 16-34.
138
136
The decay processes occurring
in
is
142
144
the 4n series.
there are beta decay processes in which a nucleus Z,
charged electron and
140
*-
transformed into the nucleus
emits a negatively
Z+
1,
A.
Thus,
by 1, N decreases by 1, and the nucleus
becomes the next nucleus lying on a line of slope 135 in the Z versus
plot. (There are also other types of beta decay processes that will be
described later.) In alpha decay processes, Z decreases by 2,
decreases
in these processes
increases
Sec.
by
2,
BETA DECAY
1]
621
and the nucleus becomes the next but one nucleus
slope 45 on the
The reason why
versus
on a
line
of
these beta decay processes occur in the An series, as well
as in the other series, can be understood
Z and
on a plot of the
lying
N plot.
by superimposing figure (16-34)
most stable for
JV values of the nuclei that are the
each value of A, as shown in figure (16-35). As the slope of the line of
stability is appreciably less than 45, beta decay processes must occur in
100
80
Line of stability
~\ rf
vim
An
r-
series
60
*
f
is!
40
20
20
J_
40
60
80
N
Figure 16-35.
Illustrating
why
beta decay occurs
120
100
140
160
*-
in
the decay of a radioactive series.
order that the average slopes of the radioactive series be essentially the
same -as the slope of the line of stability. The necessity of this simply
reflects the
requirement that each
member of
the radioactive series be
"as stable as possible."
The decay of a radioactive series involves an interplay between the two
competing processes of alpha decay and beta decay. Each member of the
series will decay by the process having the largest decay rate. (The last
statement is essentially correct if one decay rate is very much larger than
the other. This is usually the case, but sometimes the decay rates are
comparable, both processes occur, and the series branches. Cf. 84 Po 216
and 83 Bi212 in the An series.) We have seen that the decay rate in alpha
decay depends strongly on the energy available for the decay.
see that this
To
Mz A
is
discuss the energy available in beta decay,
,
We
shall
also true in beta decay.
the mass of
atoms of a given
Z and A,
it is
convenient to plot
as a function of
for fixed
A. Typical plots are indicated in figure (16-36) for odd A, and in figure
and/or magic Z, the masses are
(16-37) for even A. Except near magic
described quite well by the semi-empirical mass formula (16-33). For odd
arises
ZiA lie on a slightly skew parabola. The parabolic shape
from the Z dependence of the term ( 1 6-3 1 ) the skewness arises from
A, the values of
THE NUCLEUS
622
[Ch. 16
the Z dependence of the terms (16-27) and (16-30). For even A, there are
two parabolas corresponding to the two possible values of the term (16-32).
The
odd Z, odd N;
elevated parabola corresponds to
Odd
the depressed
A
=Q
f6 (Z,A)
Beta stable^
Figure 16-36.
L_l
(for
L_J
all
Z)
L_^
*-
The masses of atoms with
a given
odd value of A.
Even
f(Z,A)= +f(A)
(for
odd Z)
f5 (Z,A)=-f(A>
'
(for
even Z)
Beta stable
X
Figure 16-37.
The masses of atoms with
a given
even value of
A.
parabola corresponds to even Z, even N. From the plot for odd A, we see
that for a given odd value of A there is only one nucleus which is stable
to beta decay the nucleus of the atom of smallest mass. (Very rarely an
apparent exception to
this rule is
found when the difference between the
Sec.
BETA DECAY
1]
623
masses for two adjacent values of Z at the bottom of the odd A parabola
happens to be extremely small.) From the even A plot, we see that there
are generally two nuclei with a given even value of
A which are stable to
beta decay, and that these two stable nuclei have even values of
differ
by
Figure (16-38) indicates a plot of
example
A =
found for
is
Z which
Occasionally there are three beta stable nuclei for even A.
2.
MZiA representing
this situation.
An
96.
stable
Figure 16-38.
even value of
Illustrating a case in
which there are three beta stable nuclei for a given
A.
Nuclei of given values of A, whose values of Z are such that they are not
stable,
radioactive series.
electron
by
1.
Z values
can change their
decay processes. One
is
is
As
to attain stability
by three
different beta
the process of electron emission that occurs in the
stated above, in this process a negatively charged
1 and N decreases
and positron emission.
emitted by the nucleus, so that Z increases by
The other two processes
are electron capture
In the former the nucleus captures a negatively charged atomic electron,
and in the latter it emits a positively charged positron. In both, Z decreases
by 1 and N increases by 1 The energy available for electron emission by
.
the nucleus Z,
is
E = [MZ A ,
The energy
ZA
,
is the electron rest mass.
fraction of
1
Mev. Of
by that nucleus
Mev
to
course, in
MZ_
(16-49)
by the nucleus
is
2m]c*
(16-50)
X A
The energy E ranges from a small
,
more than 10 Mev and
all
is
MZ_ A ^
available for positron emission
E = [MZ A where
(16-48)
]c*
available for electron capture
E=
The energy
Mz+1 A
is
typically of the order of
these processes the available energy arises
from
the differences in the masses of the nuclei involved. These three equations
are written in terms of the masses of the atoms, such as
because
it is
convenient and conventional to do
so.
Mz A
simply
A moment's reflection
THE NUCLEUS
624
will
[Ch. 16
demonstrate to the reader that, although these equations employ the
usual approximation of neglecting the mass equivalents of the binding
energies of the atomic electrons, the rest
mass energies of all the
electrons
involved in the processes are properly taken into account. Note that there
is
a range in which the difference in masses between two atoms is such
is possible but positron emission is energetically
that electron capture
However, if
ZA is greater than Z_ 1A by more than 2m,
both electron capture and positron emission can take place. In these
circumstances positron emission is usually the more important process
forbidden.
Electron
Z,A
Z + l.A
Initial
Figure
16-39.
state
Final state
The beta decay process under the assumption
that there are
two
particles in the final state.
for nuclei of small Z.
For nuclei of intermediate and
large Z, the
wave
functions of the atomic electrons are pulled in to the positively charged
and the wave function of a positron
nucleus,
effects
tend to
make
electron capture relatively
cularly for the case of a
Now
let
is
repelled.
us take up the interesting question of what happens to the
these processes, electron emission.
Consider the most
A nucleus Z,A, which we
stationary in the initial state, emits an electron
in figure (16-39).
and
fact, since the
all
little
common
of
assume to be
recoils, as indicated
Since there are just two particles in the final state, there
should be only one way in which the total available energy
carries
parti-
K shell electron, electron capture usually dominates.
available energy in beta decay processes.
In
Both of these
more probable and,
E can be shared.
very low and it
Thus the electron should carry away essentially
the form of kinetic energy. However, measurements
nucleus
is
so massive,
its
recoil velocity
is
kinetic energy.
the energy
E in
an early stage of the study of radioactivity, using momentum
analyzing magnets and other techniques, showed that the electrons are
emitted with a spectrum of kinetic energies Te as indicated in figure
made
at
(16-40).f
For a number of
years, the fact that electrons are emitted in the beta
some cases there are also a few groups of monoenergetic electrons in the measured
They come from a process, called internal conversion, which has nothing to do
with beta decay, and which will be explained in the next section.
t In
spectra.
Sec. II]
BETA DECAY
625
decay process with a spectrum of energies was very mysterious and very
disturbing. Electrons emitted at the end point T ax of the spectrum carry
away all the available energy E, because Tf ax is observed to be equal to
within experimental accuracy. But the more typical electrons carry
away much less energy than the energy E which, the mass differences
show, must be released in the process, and it would appear that some of
this energy has vanished
A number of suggestions aimed at resolving
this dilemma were made. For instance, it was suggested that the electrons
!
emitted from the nucleus could lose energy by collisions with the atomic
Figure 16-40.
typical beta decay electron kinetic
energy spectrum.
electrons in emerging from the atom. If this were the case,
and if the
electron emitting atoms were enclosed in an isolated system,
the lost
energy would eventually be degraded into the form of heat,
as would
the energy actually carried
energy
E would
performed by
electrons, so that the entire
But experiments
and Wooster (1927), using a calorimeter with very
showed that the energy deposited in the calorimeter per
in.
the system.
Ellis
thick lead walls,
electron emission
less
away by the emitted
eventually be deposited
is
equal to the average energy of the spectrum, which
is
than half of E.
Another suggestion was that the electrons of kinetic energy lower than
are emitted in processes m which the final nucleus Z
+ I, A is left
with excess energy in one of its excited states, and that these excited
states
are so close together that the spectrum of emitted electrons
appears
continuous. The excited nucleus would subsequently re-emit this energy
in the form of a gamma ray. However, this was not
tenable because there
are a number of cases in which electrons are emitted with
a spectrum of
energies but no gamma rays are emitted, and also because other
evidence
shows that the excited states are not so closely spaced as would be required.
The situation was grave enough that people were beginning seriously to
consider abandoning the law of conservation of energy, until Pauli
(1931)
proposed a less repugnant alternative.
l
THE NUCLEUS
626
[Ch. 16
Pauli postulated that there is a particle called the neutrino which is
emitted along with the electron in each electron emission process, but
which
1
not observed because
is
Neutrinos interact extremely weakly with matter.
This property allows the neutrinos, and the energy they carry, to pass
through the thick lead walls of the
Ellis
and Wooster calorimeter. Pauli
also postulated that:
3.
Neutrinos are uncharged.
Neutrinos have spin angular
4.
Neutrinos have zero rest mass.
2.
The second property
is
momentum quantum number
J.
required in order that charge be conserved in the
The third property allows total angular
emission process.
electron
Consider a nucleus Z,A
become the nucleus Z + I, A and assume, for
example, that A is even. Then the spin angular momentum quantum
numbers i of both the initial and the final nuclei are integral (cf. item 6 of
section 1). If only the electron with s = \ is emitted, it would be impossible
momentum
to be conserved in this process.
emitting an electron to
momentum
for the final total angular
initial integral total
angular
of the system to be equal to the
the sum of a half-integral
momentum, because
an integral angular momentum (the
However, if a neutrino with
final nucleus) can only be half-integral.
s = | is also emitted, there are two half-integral angular momenta which
can combine with the integral angular momentum to form an integral
angular
momentum
total angular
(the electron) plus
momentum equal to
the initial value.
The fourth property of
neutrinos was postulated by Pauli to agree with the observation that the
end point J ax of the electron kinetic energy spectrum is equal to the available energy E. When an electron happens to be emitted at the end point,
it
carries
away
all
the available energy,
and none
is left
for kinetic or rest
mass energy of the neutrino.
In general, however, the electron will be emitted with kinetic energy Te
than the available energy E, and the neutrino will be emitted with
less
non-zero kinetic energy
T,
With
such that
+T = E = TT X
(16-51)
Pauli's postulate, there are three particles in the final state of the
emission process, very little of the available energy is taken by the recoil
of the massive nucleus, and therefore essentially all of it is divided between
the electron and neutrino. As there are an infinite number of ways in
which
this division
can be made,
it is
evident that the postulate allows
the distribution of electron kinetic energies
spectrum.
Te
to
form a continuous
BETA DECAY
Sec. II]
The
627
postulate also forms the basis of the theory of the shape of this
spectrum. In
its
most elementary form due to Fermi
(1934), the theory
is
Consider a process in which a system makes a transition
from an initial state, consisting of the nucleus Z,A, to a final state, consisting of the nucleus Z + l,A plus an electron moving off with momenquite simple.
tum p e and a
electrons
momentum
neutrino moving off with
do not
exist inside nuclei,
neutrinos do either.
and there
transition of
p.
We know
no reason
we must imagine
Therefore
neutrino to be created in the transition
an optical
is
just as a
that
to believe that
the electron and the
quantum
is
created in
an atom. Then the eigenfunction for the
initial
state is
Wi
(16-52)
Vz.a
the eigenfunction describing the nucleus Z,A,
the final state
and the eigenfunction
for
is
YV
(16-53)
Vz+i.AWeWv
where we assume that in the final state there are no interactions between
the nucleus, the electron, and the neutrino, so that the eigenfunction for
the system may be written as a product of the eigenfunction y> z+i,A f r tne
nucleus Z + I, A, the eigenfunction xp e for the electron, and the eigenfunction
y> v
for the neutrino. If there are
for the electron
(15-23).
That
and the neutrino are
is,
VK
L-?
K =
jtjh
xp e
no interactions, the eigenfunctions
waves of the form
free particle plane
(16-54)
where
e
(16-54')
and
y> v
L-K**x
(16-55)
where
K =
v
pjh
(16-55')
from the origin of coordinates to the
Z,A and Z + 1 ,Aj These eigenfunctions are normalized, using periodic boundary conditions, in a box of edge length L.
We next use the three dimensional extension of Golden Rule No. 2 (equation 9-45) to calculate the rate at which the system will make transitions
from the initial state to the final states. This says that the transition rate is
and where
r is the position vector
center of the nuclei
approximately
^v*vfiPf
(16-56)
We assume the nucleus Z,A
Z+ \,A.
to be stationary,
and ignore the
recoil
of the nucleus
THE NUCLEUS
628
The quantity pf
the quantity vfi
is
the energy density of final states of the system,
That
and
the matrix element of the interaction potential V, the
is
perturbation which causes the transitions, taken between the
final states.
[Ch. 16
initial
and
is,
vmj v *V Vi dx
(16-56')
As we know nothing about this interaction potential, we are at liberty
make the simplest possible assumption concerning its form. Following
Fermi, we assume that V is a universal constant g, which is called the
beta decay coupling constant. Then
to
/<
gj VtVt dr
gL- 3
iK
Jy>Z + i,Ae~
'"e~
iKv ' t
Vz,A dr
(16-57)
There
will be contributions to the integral only from values of r of the
order of the nuclear radii, since for larger values both y>z+i.A anc* Vz.a
But for electron emission processes, except those of
show that Ke and Kv are
both always small enough that K r < 0.1 and K v r < 0.1, for r of the
order of or smaller than the nuclear radii. To a fairly good approximation,
we may therefore set both exponentials in equation (16-57) equal to unity
and obtain
obviously vanish.
unusually high energy, a quick calculation will
% - gL~ ] Vz+i,aV>z,a dr
3
or
v fi
~gir 3M
(16-58)
where the quantity
M=]y>z+
i,
(16-58')
A yz ,AdT
Considering (16-56), we see that,
independent of the electron and neutrino momenta, these
quantities enter into the transition rate, or into the shape of the electron
is
called the nuclear matrix element.
since
is
spectrum, only through the density of final states p f
As the transitions are assumed to lead to final states of the system in
.
which the stationary nucleus Z + I, A is in a single state (usually the
ground state), pf contains only the density of states of the electron and
the neutrino. In fact, p f is essentially the product of the densities of states
of these particles. According to our discussion of section 3, Chapter 15,
both particles do have finite densities of states because of the restrictions
imposed by the periodic boundary conditions used to normalize the
eigenfunctions (16-54) and (16-55). As in that discussion, we may immediately take
over the results (12-37) of the calculation of the density
Sec.
1]
BETA DECAY
629
Te
of states for an electron of kinetic energy
present notation, this
is
N(Tt)dTe
=m
in a
Let us write (16-59) in terms of the electron
box of length L. In our
l'
(16-59)
momentum pe by
,
using the
equation
=T
^
2m
(16-60)
Differentiation yields
P^Es
dT,
Multiplying this by the square root of (16-60),
p2 dp
we
obtain
=T?dT.
2*m A
which allows us to write (16-59) as
N(T )dT
t
^^
(16-61)
can be shown that (16-61), which involves only the momentum pe ,
relativistically correct expression, even though (16-59) comes from
the non-relativistic Schroedinger equation and (16-60), which is nonrelativistic.t
Consequently, equation (16-61) may be applied to the
It
is
relativistic electrons,
transitions
we
and also
to the relativistic neutrinos, emitted in the
are dealing with. For the latter particles the equation reads
3
= l pUP
^^
N(Tv )dTy
2tt
Now
(16-62)
!!
/j
equation (16-51) shows that in a given transition T, is determined
Te In fact, by differentiating that equation we obtain
by the value of
dTv
Therefore
= -dT
(16-63)
suffices to specify the final states
number of final states,
Te and Te + dTe is
which
in
of the transition.
this electron kinetic
energy
lies
The
between
d*N
t
The reason
is
= -N(T ) dT N(T ) dT = V
that the periodic
boundary condition
L6 tf
dP*P* dPe
4 6
47T h
restrictions
which lead
to the
of states are essentially restrictions on the allowed values of the de Broglie
wavelength or, through the relativistically as well as classically valid expression /> = h/X,
finite density
are restrictions
on the allowed values of the momentum.
THE NUCLEUS
630
[Ch. 16
fact that dTv and dT, are intrinof the opposite sign. The number of these states per unit electron
kinetic energy is
where the minus sign compensates for the
sically
d*N
LpUp vPUPe
To
simplify this
energy
is
the
we note
same as the
that for the zero rest
mass neutrino the
kinetic
total relativistic energy and, as in (3-10),
cp Y
Differentiating,
(16-64)
InW dZ
dT,
=T
(16-65)
we obtain
c dp,
= dT = -dT
v
(16-66)
where we have used equation (16-63).
have
Substituting this in (16-64),
d*N_LPtf
d%-^^c dP
we
(16 ~ 67)
'
The quantity /f may be eliminated by using equations (16-65) and
(16-51),
which give
__v
K1 e
I e)
Thus equation (16-67) may be written
max ~ T
- f ,
_ L8(r
^p!dp
d2N
4ttW
dTe
(16-68)
Taking (16-68) as the density of final states for which the electron
is between p, and
pe + dpe and using (16-58), we have an
expression for R(pe) dp e the rate of transitions into these states. This is
momentum
R(P.) d Pe
^^ (rr - T
x
Ztt
The function
n c
fp\ dp,
(16-69)
momentum
spectrum of the electrons emitted in
appear in a convenient form since most of
the experiments yield the momentum spectrum directly. They predict
that a plot of the quantity [R(p^lpl\A versus (Tf Te ), or simply
versus T should yield a straight line. Figure (16-41) shows such a Kurie
plot for the simplest of all electron emission processes,
R(p,)
the transitions.
is
the
These
results
V
1
the decay of the neutron "w into
H1 +
+v
a proton H
e~
(16-70)
plus an electron e~ and a
The neutron decays this way because [M01
+0.78 Mev. The lifetime t of the decay is about 1 100 sec.
neutrino
v.
M^c2 =
Sec.
BETA DECAY
1]
631
These data provide a fairly typical example of the kind of agreement
obtained between the simple Fermi theory and experiment for the beta
decay of low nuclei. The deviations at the low energy end of the spectrum
are often found to be a result of experimental difficulties such as selfabsorption of the low energy electrons in the source. However, for high
nuclei there are definite deviations between experiment and the simple
theory at the low energy end of the spectrum which are more pronounced,
02
0.1
0.3
0.4
Te
Figure 16-41.
0.5
0.6
0.7
0.8" - 0.9
1.0
(Mev)
Kurie plot for the beta decay of the neutron.
From
J.
M. Robson,
Phys. Rev., 83, 349(1951).
and are of the opposite
deviations for high
compared to those of figure (16-41). These
from the use of plane waves (equation
sense,
nuclei arise
16-54) for the electron eigenfunctions. This
is
an excellent approximation
for the neutrino eigenfunctions because neutrinos interact so weakly that
they really do behave like free particles.
Coulomb
to make %
strongly with the
field
interaction tends
larger than
of the
But electrons interact rather
final nucleus.
This attractive
a plane wave in the region near
and thus increases the transition rate. As the Coulomb
more important for small Te the effect is to enhance the low
energy part of the spectrum. The simple Fermi theory also applies directly
to positron emission, but the Coulomb effect has a different sign because
the interaction is repulsive. This makes the positron eigenfunction smaller
than a plane wave near the nucleus and leads to a depletion of the low
the nucleus
attraction
is
energy part of the spectrum.
are included in the theory,
When
its
corrections for these
Coulomb
effects
predictions are usually in
with accurate experiments, even for
good agreement
electrons or positrons of quite low
energy.
Of
course, the question of the shape of the spectrum does not arise in
THE NUCLEUS
632
electron capture.
the nucleus
state,
away
carries
In this process there are only two particles in the final
Z I, A
and the neutrino. The
essentially all the available energy
Although only a neutrino
so massive.
[Ch. 16
is
always
latter particle
since the nucleus is
emitted in the primary process
of electron capture, the process can be observed experimentally by detecting the X-rays which are subsequently emitted in the filling of the hole left
in the electronic system of the atom.
By
integrating (16-69) over the entire spectrum, the transition rate into
all final states
may be
As
calculated.
in equation (13-118), this total
probability per second that the initial nucleus will decay
reciprocal of
its
lifetime t.
is
equal to the
Thus
^sr^-^*t
To
2tt
convert the integrand completely to a function of pe
(T
where
mc2
max
TJ2
(i6-7i >
Jo
<r
we
write
^ ^ + ^^
_
= [T +
= (fir" " Eef
max
the rest mass energy and Ee is the total relativistic energy of
and where we have used equations (1-22') and (1-24). But,
is
the electron,
according to (1-25),
E e = Vc2^ +
mV
So
i
t
^ f^Wr" + V - Vc^ +
2irn<r Jo
rnV/p! dA
Integrating this gives
;"S^
(16-72)
F(a)
where
F(a)
= -Ja - Aa3 + 3^a5 + jVl +
-
a 2 sinh" 1
Vl +
a2
(16-72')
with
= p**jmc
a.
The function
F(<x)
has the behavior
-
F()~
ifsoc
7
,
<x
and an intermediate behavior between these
>5
< 0.5
limits.
(16-72*)
Sec.
BETA DECAY
1]
When
the
Coulomb
An
function F(a, Z).
633
corrections are included, F(a)
graphs and tables can be found in several
f(a,
Z)
~ F(a).
For
is replaced by a
Z) cannot be given, but
references. For small Z,
analytic expression for F(oc,
Z=
emission, and F(x,Z)
100, F(a,
~ 0.1 F(a) in
Z)
100F(a) in the case of electron
the case of positron emission,
when
a has a value corresponding to the typical available beta decay energy
E = 1 Mev.
Equation (16-72) shows that the lifetime t of a beta decaying nucleus
depends strbngly on />ax and therefore strongly on the available energy E.
,
The observed lifetimes for the naturally occurring beta unstable nuclei
range from values of the order of 1 sec for an available energy of several
Mev to values of the order of 10' sec for available energies of several
hundredths of an Mev. (Note, however, that the lifetime of an alpha
decaying nucleus is an even stronger function of the energy available in
the alpha decay.)
Equation (16-72) also shows that the quantity
F(a z>~^Tl7^r;
>
g'Tn^c*
M*M
(16-73)
depends only on a collection of universal constants, and on the value of
the nuclear matrix element
M =JVzi,aVz.a dr
where we
now
(16-73')
indicate that the theory applies equally well to electron or
The product Ft is inversely proportional to M*M and
when it takes on its maximum possible value M*M = 1. This
occurs when the eigenfunction y> zi.A is identical with the eigenfunction
positron emission.
is
smallest
Vz.a because, in these circumstances, the normalization condition for
eigenfunctions requires that (16-73') yield
1.
If the eigenfunctions
1, and this quantity becomes
Vzi.a and Vz.a ar e not identical,
M=
M*M <
smaller as the eigenfunctions
become less similar. In fact, M, and therefore
M*M, are exactly zero if y>zi,A and Vz.a are so dissimilar as to correspond
to different values of nuclear spin
i,
or to different nuclear
parities.
The
because the eigenfunctions y> z\.A an d Vz.a-> like
any other eigenfunctions, are orthogonal with respect to all their quantum
numbers, including the quantum number i. The second restriction arises
first restriction arises
because, if the parities of these two eigenfunctions differ, the entire
integrand will be of odd parity.t and the contribution to the integral
(16-73')
t That
from each point
is,
(even)(odd)
\x, y, z) will
(odd),
be canceled by the contribution
and (odd)(even)
(odd).
THE NUCLEUS
634
from the point
(-
x, y, ). These two
[Ch. 16
restrictions constitute the
Fermi
selection rules:
M=
The nuclear
on the
that give conditions
nuclei Z,
element
and
M to vanish
parity
(16-74)
must not change.
relative spins
and
parities of the states of the
1,A which, if violated, cause the nuclear matrix
Transitions which satisfy the selection
identically.
rules are called allowed transitions.
'
!
Neutrons Protons
Neutrons
Figure 16-42.
A schematic
Protons
representation of the ground states of a typical set of mirror
nuclei.
It
immediately follows from the shell model that beta transitions between
the ground states of any set of mirror nuclei are always allowed transitions.
As mentioned
in section 3, the two nuclei of a set of mirror nuclei consist
of an odd number of protons and the same odd number of neutrons, plus
an extra proton for one of the nuclei and an extra neutron for the other.
The neutron and proton
eigenstates filled
by the nucleons of a
of mirror nuclei are indicated in figure (16-42).
nuclear spin
is
equal to the total angular
nucleon and, since
this
nucleon
is
in the
momentum
same
typical set
In the shell model, the
of the single odd
eigenstate in each nucleus
set, the spins of these two nuclei will be the same. Also, in the shell
model the complete eigenfunctions yzi,A and Wz.a can De written as
products of the eigenfunctions for each of the A independently moving
nucleons, and the nuclear parity will be the product of the parities of each
of these eigenfunctions. Since there are two nucleons in each eigenstate,
except for the eigenstate containing the single odd nucleon, and since the
product of the parities of two eigenfunctions of the same parity is always
of the
even,f the nuclear parity
single
odd nucleon. As
nucleus of the
same.
t That
set, it
is
equal to the parity of the eigenfunction of the
this
nucleon
is
in the
same
eigenstate in each
follows that the parities of these two nuclei are the
These arguments lead us not only to the conclusion that the
is,
(even)(even)
(even),
and (odd)(odd)
= (even).
Sec.
BETA DECAY
1]
complete eigenfunctions
similar
enough
635
ip z1
A and y z<A
for a set of mirror nuclei are
to satisfy the selection rules (16-74), but also to the
much
stronger conclusion that these two complete eigenfunctions should be
They differ only in that the eigenfunction for one
odd nucleon) is a proton eigenfunction for one of the
essentially identical.
nucleon (the single
nuclei and a neutron eigenfunction for the other and, since these eigenfunctions are for the same eigenstate, they are identical except for
Coulomb
M*M ~
1. Thus
f zi,A V>z,a>
we conclude that for all beta transitions between the ground states of sets of
mirror nuclei the product Ft should assume its smallest possible value.
This conclusion is verified by the experiments, which show that for all these
35 sec, and that
transitions the Ft values are clustered around Ft ~ 10
the Ft values for these transitions are the smallest which are ever observed.
By using these experimental values of Ft in (16-73), and making the
effects.
But we have seen
M*M =
assumption that
tnen
that, if
for these cases,
constant can be evaluated. The result
gil.4x
the beta decay coupling
is
10- 49 erg-cm3
(16-75)
Of course, there are many allowed transitions which do not take place
between the ground states of mirror nuclei, since there are many other
Although
possibilities of satisfying the selection rules (16-74).
is greater than zero for all these transitions, it is less than unity because
M*M
y z1 A and y Z-A lead to cancellations which reduce
the value of the integral (16-73'). With increasing A, these differences
the differences between
should become greater because the nuclei become larger and more complicated.
This
is
seen in the experiments since the measured
these allowed transitions increase with increasing A.
55
10
sec.
for allowed transitions are around Ft
The
Ft values for
Ft values
largest
If either part of the selection rules
However, for such forbidden
become
transitions
infinitely large, as predicted
only by about a factor of 103
not exact because
it is
(16-74)
it is
is
M*M = 0.
violated,
observed that Ft does not
by equation (16-73), but increases
This simply
reflects the fact that
based upon several approximations.
(16-73)
One of
is
these
e~ iK '* equal to unity in (16-57). A correction to this approximation is found by using the more accurate expression
of (16-57) obtained by expanding e~' K '* in a power series and taking the
first two terms. This gives for the beta decay interaction potential matrix
is
the setting of e~' Ke-r e~' K,,
'
element
gL~
J
V>zi,aVz,a dT
- K
Vzi,A r V>z,A d i
<
(16-76)
'J
Situations can obviously arise in which the
first
term vanishes but the
THE NUCLEUS
636
second does not. As
J ip%i,A*y>z,A </t
<K
of normal energy, this second term makes
Ft values
transitions as long as the
it
<
10
-1
[Ch. 16
for beta decays
possible to have forbidden
are larger than those for allowed
by a factor >(10 -1 )' 2 = 10 2 A similar possibility arises from
_1
a term of order v/c < 10
which should actually appear in the beta decay
transitions
interaction potential matrix element because of relativistic effects.
The
situation
further complicated by the experimental observation
is
that there are certain beta transitions that violate the
(16-74) but nevertheless have
that are allowed
by these
Ft values
assumed, but
of a
is
V in
(16-56')
is
not a constant, as Fermi
more complicated type
interaction potential matrix element
which gives
show that
Theoretical investigations
Fermi selection rules
any of the transitions
This implies that in these transitions the
rules.
beta decay interaction potential
initially
as small as for
that leads to
an
different selection rules.
there are several types of matrix
elements which should be possible, and that certain of them lead to the
Fermi selection
rules, while others lead to the
Ai
0,
(but not
Gamow -Teller selection
->
rules:
0)
(16-77)
The nuclear
By
parity
must not change.
evaluating the consequences of the various possible types of matrix
elements, and comparing these consequences with experiment,
it is
found
that the beta decay interaction potential matrix element actually existing
in nature contains a mixture of
approximately equal amounts of the types
leading to the Fermi selection rules and the types leading to the
Gamow-
These comparisons are not based on the shape
of the beta decay spectrum, because this is always dominated by the same
density of states factor for all types of matrix elements. We shall not go
into the theory underlying these comparisons because these points are
really intelligible only in the language of Dirac's relativistic theory of
quantum mechanics. Similar motivation leads us to avoid making a
Teller selection rules.
between neutrinos and anti-neutrinos.
However, we shall consider one recently discovered property of the beta
decay interaction potential matrix element that is extremely interesting
and not difficult to describe. It is found that this matrix element is not a
distinction
scalar but instead
is
at least partly a pseudo-scalar.
when
The
distinction
is
that
performed that is,
when the coordinates of each point (x, y, z) are changed to (%, y, z)
whereas a pseudo-scalar changes sign in this operation. An example
of a scalar is the Coulomb interaction potential, or any other interaction
potential that can be written V(r). As the argument r is the magnitude of
a scalar does not change
the parity operation
is
a vector
r,
V(r) does not
change in the parity operation of
reflecting r
BETA DECAY
Sec. II]
637
through the origin, even though r -> r in this operation. Since the
matrix element of V(r) is a certain average of this quantity, the matrix
element is a scalar. In fact, with the exception of the beta decay interaction
potential matrix element, the matrix elements of
all
the
known
interaction
of classical, atomic, and nuclear physics are scalars. For
instance, consider the spin-orbit interaction potential matrix element,
potentials
L. We investigate what happens to S L
by writing L = r X P and realizing that in this
operation both r and p change sign.f Thus L does not change sign. To
bring out the distinction between their behavior in the parity operation,
vectors which change sign, such as r and p, are called polar vectors, and
vectors which do not change sign, such as L, are called axial vectors.
which
is
proportional to
in the parity operation
All angular
we
momentum
vectors, including S, are axial vectors.
see that the spin-orbit interaction potential matrix element
Therefore
is
a scalar
does not change sign in the parity operation. An example
of a pseudo-scalar is the quantity P A, where P is a polar vector and A
is an axial vector.
This is a pseudo-scalar since it changes sign in the
parity operation because P changes sign and A does not. The beta decay
because S
interaction potential matrix element contains quantities of this type.
the parity operation
is
equivalent to leaving the point
As
and
the operation amounts to
(x, y, z)
fixed
through the origin,
changing from a right-handed to a left-handed coordinate system and,
as at least part of the beta decay interaction potential matrix element
changes sign in this operation, the description of the beta decay processes
which this matrix element produces will depend on whether a right-handed
or a left-handed coordinate system is used. When this was first discovered,
it was extremely surprising because it had been a fundamental assumption
reflecting the coordinate axes
that the description of all physical processes must be independent of
whether a right-handed or left-handed coordinate system is used.
This assumption was proved incorrect for beta decay in experiments
performed by Wu and collaborators (1957), at the suggestion of Lee and
Yang (1956). In these experiments, which we describe first in a right-
handed coordinate system, the spin angular momentum vectors I of
electron emitting 27 Co 60 nuclei were aligned with respect to an external
magnetic field by using the hyperfine structure interaction at temperatures
so low that the thermal equilibrium energy kTis comparable to the hyperfine splitting energy. In these circumstances, a majority of atoms will be
in the lowest energy state, and in this state I is essentially aligned with
respect to the applied magnetic field. The angle between the direction of
I and the direction of the momentum vector p was then measured for
tp
other
changes sign since
m dxjdt -* m d{-x)jdt =
momentum components.
-mdxjdt, and
similarly for the
THE NUCLEUS
638
[Ch. 16,
each emitted electron, and the distribution of these angles was plotted.
It was found that this distribution is asymmetric, with about a 30 percent
preference for the emission of electrons whose
momentum
vectors p are
from the spin vector I. So in the average
emission process p I < 0. But p I is a pseudo-scalar since p is a polar
vector and I is an axial vector. Thus in any left-handed coordinate
in the opposite hemisphere
system the sign of
process p
>
0,
this
pseudo-scalar will change, in the average emission
and the description of
depends
this process therefore
on whether a right-handed or a left-handed coordinate system
is
used.
must be a
were purely
This proves the beta decay interaction potential matrix element
pseudo-scalar, or at least contain a pseudo-scalar part.
scalar, the electron distribution
in the average emission process
If
it
necessarily be symmetric so that
would
0,
and the description of
this
process would be independent of whether a right-handed or a left-handed
coordinate system is used, even though the pseudo-scalar p I changes sign,
because
+0.
Since the beta decay interaction potential matrix element is not a
z)
it is not true that for the interaction potential V(x, y,
scalar,
V(x, y,
Thus the eigenfunctions
z).
for the entire system
do not
necessarily
have a definite parity (cf. footnote on page 465), and it is not necessary
that the entire system conserve parity. Experiments indicate that it does
not. The parity of the initial state consisting of the nucleus Z, A is not
the
same
as the parity of the final state consisting of the nucleus
Z \, A
plus the emitted neutrino and electron or positron.
We close this
section with brief descriptions of
ments involving beta decay and
In the period since 1950 a
Allen, Sherwin, and others
on the
recoil velocity of the nucleus
As
the nucleus
is
remaining in
very massive
and neutrino, the recoil velocity is low enough
measured
by measuring electronically the time (several
can be
compared
it
recent experi-
number of measurements have been made by
the final state of a beta decay process.
that
some other
neutrinos that are of particular interest.
to the electron
microseconds) required for the nucleus to travel a given distance (several
centimeters) in a given direction through a field free vacuum. This
measurement determines the momentum vector of the recoiling nucleus.
By
momentum vector of the emitted electron, it is
momentum can be conserved only if a neutrino
process. Furthermore, the required neutrino momentum
also measuring the
possible to
show
that linear
emitted in the
agrees within the experimental accuracy (10 percent) with the neutrino
energy required to conserve energy, if the neutrino rest mass is zero as
is
assumed by Pauli.
Even after these
recoil
a strong feeling in
some quarters
experiments had been performed, there was still
that the neutrino is not a particle but
BETA DECAY
Sec. II]
only a
639
myth cleverly designed to provide for the conservation of energy,
momentum, and linear momentum. There was a certain merit in
angular
this
point of view as long as the neutrino remained undetectable. However,
Cowan and Reines in 1956 succeeded in detecting the
experiments by
neutrino, and thereby gave
tremely
it
the status of any other particle.
->
V + e+
the capture of a neutrino v by a proton
positron e +
In these ex-
experiments they observed the reaction
difficult
This
(16-78)
leading to a neutron n 1 and a
the inverse of the reaction
is
V + e+ - W +
(16-79)
which is an alternative form of the neutron decay process (16-70). A point
which will be of interest shortly is that the reaction (16-79) should have
10 3 sec^ as the reaction (16-70) because the
about the same lifetime, t
matrix elements and densities of states governing the lifetimes contain
essentially the same factors, albeit in a different order. The CowanReines reaction (16-78) took place on the protons contained in a very large
tank of hydrogenous scintillator exposed to the enormous flux of neutrinos
emitted from the fission induced beta decays in a nuclear reactor, and the
positrons were detected by the scintillations they produced in the same
tank. Elaborate methods were used to minimize background scintillations.
This was necessary because less than one reaction per minute was expected,
despite the magnitude of the neutrino flux and the size of the hydrogenous
target, since the interaction of neutrinos with
To show
we make
this,
Mev
incident
upon a proton, and assume
interaction between these
two
particles
their relative separation is
maximum
~10~ 10
that there
lifetime
~103
justified
cm. As their relative velocity
lifetime for the reaction (16-78)
sec for
its
a possibility of
This will occur
when
is
is
~10 10 cm-sec -1 the
20
is ~10~
sec. Now
,
about the same as the
this assumption is
inverse reaction (16-79);
by the principle of microscopic
equation (13-84).
is
comparable to the de Broglie wavelength of
time during which there can be overlap
assume that the
extremely weak.
whenever they are close enough to
allow appreciable overlap of their wave functions.
the neutrino
is
Consider a neutrino of the typical energy
reaction (16-78) as follows.
1
matter
a rough estimate of the cross section for the
reversibility
mentioned
after
Then
the probability that the reaction (16-78) will
10~ 23 and the cross section
occur in the time ~10~ 20 sec is ~10~~ 20 /10 3
10~ 20 cm 2 for
a for this interaction is the cross section <~(10~ 10 cm) 2
with overlapping wave functions times the probability just
10~ 43 cm 2 An accurate
evaluated. This gives <r~ 10~ 20 cm 2 x 10~ 23
collisions
THE NUCLEUS
640
on the method we used
calculation, based
[Ch. 16
in evaluating the lifetimes of
the processes leading to neutrino emission, gives
(16-80)
rk*C
For a positron of momentum p e and velocity v e corresponding to the
1 Mev, (16-80) predicts an interaction cross section
10~ 44 cm 2 The measurements of Cowan and Reines agree with this
a
typical kinetic energy
value, within the large experimental uncertainty.
Note
that equation (16-80)
shows
clearly that the extreme smallness of
the neutrino interaction cross section
is a result of the extreme smallness
10~ 49 erg-cm 3 The cross section
of the beta decay coupling constant g
small
that
a
neutrino
is so
emitted in a nuclear reaction in the interior of
the sun can traverse
its entire bulk with little probability of being absorbed.
Beta decay interactions are very weak compared to nuclear interactions.
A numerical comparison can be made by dividing the beta decaycoupling
constant, which has the dimensions of energy times volume, by the
volume
of a typical nucleus. This gives a characteristic energy ~10~ 49 ergcm3 /^ x 10~ 13) 3 cm 3 -~ 10~ 12 ergs <~ 1 ev, which is smaller by a factor of
~10~ 7
than the energy
~10 Mev
that characterizes nuclear interactions.
Since the square of the beta decay coupling constant enters into measurable
quantities, such as transition rates
and cross
sections,
it
is
appropriate
to say that beta decay interactions are weaker than nuclear interactions by a
factor of <~10~ 14 The reader will recall that the comparison of figure
(16-12) indicates that electromagnetic interactions are only somewhat
.
weaker than nuclear interactions;
will
show
in section 16 a quantitative
actions by a factor of'
TO -3
This conclusion, plus the conclusion drawn
in the first footnote of section 3,
shows that gravitational
weaker than nuclear interactions by a factor o/"~10~ 38
12.
comparison
that electromagnetic interactions are weaker than nuclear inter-
Gamma
interactions are
Decay and Internal Conversion
The nucleus
a system of nucleons
is
bound to a certain region of space.
quantum mechanics, it should
Since this system seems to obey the laws of
have a
set of
quantized energy states.
we have considered
we turn to the
Heretofore
only the lowest energy ground state, but in this section
higher energy excited states.
excited states has
gamma
rays,
state of
energy
come from
Much
of what has been learned about the
a study of the high energy quanta, called
which are emitted in transitions between some initial excited
E and a final state of energy Et < E These spontaneous
t
GAMMA DECAY AND INTERNAL CONVERSION
Sec. 12]
gamma
641
decay transitions are completely analogous to the optical transi-
between the quantized energy states of an atom. In
both cases, the transitions take place through the interaction of the charge
moments and magnetic moments of the system with electromagnetic
fields, and energy is conserved because the energy of the emitted quantum
tions that occur
is
hv
=E -E
t
(16-81)
There are a number of processes through which the nucleus can be
its higher energy states.
For instance, it can happen that
an alpha decay or beta decay process leaves the final nucleus in an excited
state instead of its ground state. The possibilities are indicated schemaexcited to one of
diagram of figure (16-43).
decay is, of course, the available beta decay
energy E, equal to the total energy carried away by the neutrino and the
electron or positron. Usually only a few states of the final nucleus are
tically,
for beta decay, in the energy level
The energy change
in the beta
excited because beta decays to states of certain spins and parities are
forbidden by the beta decay selection rules, and because the strong
dependence of the beta decay transition rate on E makes the transition
rate negligible to states higher
than several
In excitation following alpha decay,
Mev above
the
ground
state.
the even stronger dependence of
on the alpha decay energy usually limits
somewhat less than 1 Mev
above the ground state.f States of energies up to 6 or 8 Mev above the
ground state can be excited when this binding energy is liberated by the
capture of a low energy nucleon in a nucleus. States of even higher
the alpha decay transition rate
the excitation of the final nucleus to states
energy can be excited by the capture of a higher energy nucleon, processes
in which a high energy nucleon is inelastically scattered from a nucleus,
and by a number of other processes involving nuclear reactions.
After the excitation of a nucleus, it can decay by the emission of gamma
rays.
If
nucleon
the excitation energy
in the nucleus,
it
is less than the binding energy of the last
must decay by gamma emission, or by the related
process of internal conversion that will be explained later in this section.
t
Note
that,
when alpha decay between two
nuclei leads to excited states of the final
nucleus, the alpha particle energy spectrum will contain
of lower energy.
of the
initial
It is
nucleus,
of higher energy.
a. fine structure of components
also possible for the alpha decay to originate from excited states
and
The
in this
case the spectrum will contain long range components
similar situations which occur in beta decay result in several
values of the available beta decay energy E, and electron or positron energy spectra
which are a superposition of spectra corresponding to each value of E. Note, then, that
information about the energies of the excited states can be obtained directly from a study
of the energies of the particles emitted in alpha or beta decay. For these processes,
and particularly the latter, a study of the lifetimes can also give information about the
spins and parities of the excited states.
THE NUCLEUS
642
[Ch. 16
For excitation energies higher than the binding energy of the last nucleon*
decay by nucleon emission is possible and, for reasons to be explained in
the next section, with increasing excitation energy nucleon emission
rapidly becomes the dominant decay mode. As a consequence, most
gamma
rays are emitted with energies less than about 6 or 8 Mev. In fact,
their energies are generally
much
less
than
this
because the de-excitation
often proceeds in a cascade through a succession of low energy states,
instead of by the emission of a single
gamma
ray which carries
away
all
Excited states
'. ',
Zl,A
The
Figure 16-43.
the excitation energy.
is
set
'
Ground
state
excitation of a nucleus as a result of beta decay.
The lower
limit for the energy of these
gamma
rays
by the minimum energy difference between the low-lying nuclear
energy
states.
It is
of the order of 10 3 ev.
Most of the methods for measuring the energies of gamma rays, or
quanta in the energy range 10 3 ev to 6 or 8 Mev and higher, involve
transferring their energy content to electrons by means of the photoelectric
and Compton processes described in Chapter 14, or to electrons and
positrons by means of the pair production process described in the same
chapter. The energies of these charged particles can then be measured by
standard techniques. In perhaps the most widely used method, the energy
is transferred to electrons of the scintillating crystal Nal, and the electron
energies are measured in the same crystal by measuring the intensity
of the light pulse which it emits, due to scintillations produced by the
electrons. For hv < 0.3 Mev, it is most probable that all the energy of a
gamma ray will be transferred to a single electron of the 1 atoms by the
photoelectric process. For 0.3 Mev et hv < 7 Mev, the Compton effect
is the most probable, and only a fraction of the gamma ray energy is
transferred to an electron in a single collision. In this range, it is necessary
to use a large Nal crystal in order to absorb all the energy. Forhv > 7 Mev,
the pair production process dominates, and the gamma ray energy is
GAMMA DECAY AND INTERNAL CONVERSION
Sec. 12]
643
divided between the total energy of an electron and a positron. At even
larger hv, the ranges of these particles are too long to allow their energies
to be
measured
an Nal
in
called a pair spectrometer
foil
of some high
crystal of convenient size and, instead, a device
is
In this device the pair is produced in a
and the energies of the emitted electron and
used.
Z material,
momenta in a magnetic field.
The methods just described have gamma ray energy resolutions of several
positron are measured by analyzing their
percent at best, but other methods of different types and much better
resolutions are also available. The one with the best resolution involves
gamma rays from a crystal lattice of known lattice spacing d.
Although the gamma ray wavelengths X = cjv are as much as several
orders of magnitude smaller than d, diffraction can be used by making
the distances between the source, crystal lattice, and detector very large.
This gives a large linear displacement even though the angular displacement
diffracting the
~ Xjd
However, as all the solid angles involved are very small,
the diffraction method has low efficiency and can be used only in very
favorable circumstances. The other high resolution methods also suffer
6
is
small.
from low efficiency.
The measured energy spectrum of gamma rays, emitted in transitions
between the energy states of a nucleus, is used to determine the energies
of these states the connection being made by the same trial and error
procedures that were used in the very early days of the study of atomic
spectra and energy states. Of course, the energies of the nuclear states
give important information about nuclear structure.
Another source of important information is the lifetimes of these states.
If the lifetime t
>
10~ 10 sec,
it
can be measured by electronically timing
the delay between the instant the state was excited (e.g., by detecting the
beta decay which led to the excitation) and the instant it decayed (by
detecting the emitted gamma ray). In extreme cases, lifetimes as long as
t
10 8 sec are found. All gamma ray transitions with directly measurable
lifetimes are called isomeric transitions to distinguish
them from
the
more
which have lifetimes t ^ 10~ 10 sec that are not directly
-16
sec, it can be measured indirectly by
measurable. However, if t < 10
measuring the natural width of the gamma ray spectral line emitted in the
transition, and then employing the relation (13-125). This technique can
be used for nuclear transitions only with the aid of an ingenious method,
called resonance absorption, that employs absorption in a target, contypical transitions,
taining
some
species of nuclei, as a spectrometer of extremely high resolu-
tion to measure the width of the spectral lines emitted
containing nuclei of the same species.
10~ 17 sec have been observed.
Gamma
decay
by a source,
lifetimes as short
asT~
The tremendous
variation which
is
found
in the lifetimes for
gamma
THE NUCLEUS
644
decay
due, in part, to the dependence of t on the energy of the emitted
is
gamma
[Ch. 16
ray.
But an equally important part of this variation
and parities of the initial and
influence of the relative spins
is
due
to the
final nuclear
states.
Therefore analysis of the experimentally measured values of
t gives information about the spins and parities of the states. These
quantities are just as important in characterizing the states as their energies.
The analysis is based on the theory of gamma decay transitions in nuclei.
As the theory is basically the same as the theory of optical transitions in
atoms which we treated in section 12, Chapter 13, it would be appropriate
for the reader to review that section.
In our treatment of optical transitions in atoms
we considered only
the electric dipole transitions, which arise
electric
from periodic oscillations in the
dipole moment of the charge distribution, and we found that the
electric dipole transition rate for
spontaneous emission
is
proportional
to the "square of the matrix element of the interaction potential for the
moment,"f taken between the states involved in the transition.
completely rigorous treatment would also consider the electric quadru-
associated
pole, electric octupole,% etc., transitions arising
in the associated
dipole,
from periodic
moments of the charge distribution's well as
magnetic quadrupole, magnetic octupole,%
from periodic
oscillations in the associated
tion (or of the magnetic
transitions arising
moments of the current distribu-
moments produced by
would be found that the
etc.,
oscillations
the magnetic
electron spin). In each case
transition rate for spontaneous emission
by a
proportional to the square of the matrix element of the
interaction potential for that moment. We did not give the completely rigorit
particular moment
is
ous treatment because
The
it
is
really
not necessary for optical transitions.
magnetic optical transitions is
justification for neglecting all the
that the current distribution
is
pvjc,
of the moving charges, and this
by the factor
moments
v/c,
which
is
where
is
the characteristic velocity
smaller than the charge distribution p
typically ~10~ 3
As a result, the magnetic
is
moments by this
(The magnetic moments produced by electron spin are of the same
order as those produced by the current distribution.) Thus the square
of the matrix element of the interaction potential, and therefore the
are
all
smaller than the corresponding electric
factor.
We
use this somewhat loose terminology as an abbreviation. The transition rate
t
for spontaneous emission is proportional to the transition rate for absorption, and
(13-77) shows that the latter quantity is actually proportional to a sum of terms each of
which is the matrix element of the interaction (perturbation) potential for a component
of the electric dipole moment times its own complex conjugate.
J Except for a special process involving the entire atom, there are no electric mono-
pole transitions because nuclear charge conservation requires the associated
to be constant.
moment
is
zero.
moment
There are no magnetic monopole transitions because the associated
GAMMA DECAY AND INTERNAL CONVERSION
Sec. 12]
transition rate,
smaller for each of the magnetic
is
moment by
corresponding electric
The
etc.,
the factor (v/c) 2
moments than
~ 10~
is
that the matrix element of the interaction
moment
of the charge distribution
moment by
smaller than the matrix element for the preceding
2-nRjl,
for the
justification for neglecting electric quadrupole, electric octupole,
optical transitions
potential for each successively higher
is
645
where
is
the radius of the charge distribution and X
length of the radiation.
This factor
is
is
the factor
the wave-
~10~3 The matrix
successively higher moment
also typically
element of the interaction potential for each
of the current distribution decreases by the same factor. The factor arises
because the interaction potential of a dipole
field involves
products of the
field
moment
in a spatially varying
times distances of the order of R, the
interaction potential of a quadrupole
moment
involves products of
first
R2
the
spatial derivatives of the field times distances of the order of
interaction potential of an octupole
moment
involves products of second
spatial derivatives of the field times distances of the order of
t
Equation (13-65) shows that
potential,
and
it
electric potential
is
this is true for the electric dipole
y, z)
and the charge distribution be
S<7j.
R3
moment
easy to verify for the higher electric moments.
be V(x,
etc.t But,
interaction
Let the external
where
(x h
yH
z,)
are
the coordinates of charge q s and where the sum is taken over all the charges. Then the
total energy of the distribution is S^Ktx,, Vi, z,). Make a Taylor expansion of V(x, y, z)
,
about the origin
(0, 0, 0)
V(x,y,z)
of the charge distribution. This gives
[V
"l/a K\
2
+
+
Sx dy/
Use
this to evaluate the total energy.
l/a K\
2
xy
v
().*
d2
2\
ft?
\ dx dz/
\ By dzf
It is
where we employ the relations Ex = dVjdx,Ev = dVISy,E z = ft V/dz to introduce
the components of the electric field. The first square bracket is the electric monopole
interaction potential, and the sum it contains is the total charge; the second square
bracket is the electric dipole interaction potential, and the three sums it contains are
THE NUCLEUS
646
[Ch. 16
which is a sinusoidal function of position with wavelength X,
such as sin (Inx/X), each spatial derivative brings in a factor of lirjX.
in a field
Thus the interaction potential of each successively higher moment contains
an extra factor of 2-ttR/A. The square of the matrix element, and thus the
transition rate,
smaller for electric quadrupole transitions than for
10~ 6 and the transition
by the factor (IttRJI) 2
is
electric dipole transitions
moments
rates for the higher
are smaller by successively higher even
powers
of this factor. All optical transitions in atoms, other than electric dipole
transitions, have very small transition rates, or very long lifetimes. These
transitions
can be neglected because in the long lifetimes the atom
will
almost always be de-excited by collisions, or some other non-radiative
process, before it gets a chance to radiate.
These simplifications do not obtain for gamma decay transitions in
-4
10~ 2 and (2ttR/X) 2
For one thing, (v/c) 2
10
in the typical
case A = 200 and hv = 1 Mev. Therefore the factor suppressing the rates
for magnetic transitions is not so small, and neither is the factor suppressing the rates for transitions associated with the higher moments. Even more
nuclei.
the components of the electric dipole moment the third square bracket is the electric
quadrupole interaction potential, and the six sums it contains are the components of
the electric quadrupole moment. This proves our point.
It is also interesting to develop the relation between the components of the electric
quadrupole moment and the quantity Q defined by equation (16-38). In the case to
which that definition applies, there is an axis of symmetry along the z axis, each of the
2
last three components vanishes, and ~5Lq s x = YfliV^Then the electric quadrupole
;
interaction potential
is
/3 2 k\1
(d 2 V\
l/^W
where we have changed back to the external electric potential
potential, Laplace's theorem of electrostatics demands
d2
d2
d2
dx
So
By
V.
But, for an external
dz 2
the electric quadrupole interaction potential can be written
where
KSM^-H-^"
Q=
2>,[2z?
2x 2 ]
J^q&z
-x - y
2
2
]
= 2*p*j -
(*?
y*
*?)]
Equation (16-38) is obtained by
in units of proton charges.
letting the
sum go
to an integral
and expressing
the q s
GAMMA DECAY AND INTERNAL CONVERSION
Sec. 12]
important
no
the fact that there are
is
collisions
between nuclei, or other
non-radiative processes (except internal conversion)
if
the excitation energy
Thus all states with excitation
than the nucleon binding energy must eventually decay by
than the nucleon binding energy.
is less
energy
less
gamma
ray emission (or internal conversion), even
is
647
if
the lifetime for decay
extremely long.
Although
we
a task, which
it is
shall avoid,
even to write expressions for
the matrix elements of the higher electric
and magnetic moments, the real
difficulty in evaluating these matrix elements arises from the lack of
information concerning the nuclear eigenfunctions which they contain.
This is the same difficulty that arises in evaluating the matrix elements for
beta decay transitions. However, if some nuclear model is assumed, the
eigenfunctions for that model can be found, the matrix elements can be
evaluated, and predictions for the gamma decay transition rates for spontaneous emission can be obtained. The accuracy of these predictions will,
of course, be only as good as the accuracy of the model. Using the shell
model with a nucleus of radius R, Blatt and Weisskopf have obtained the
following expressions for approximate values of the transition rates S
for spontaneous emission in electric and magnetic transitions.
For electric transitions:
^ TJfiL(W"(f
W W
L[l'3-5---(2L+l)]\L
For magnetic
_ 82)
(1(i
3/
transitions:
+
s^ L[l-3-5---(2L+
./r-....
i^j(r'(r'^
+
1.9(L
1)
. r
where
L =
l)]
for dipole transitions,
3 for octupole transitions, etc.,
=
R =
E =
S
with
the energy of a
gives values of t
several values of hv,
l.2A 1A
gamma
10~ 13
cm
gamma
\L
L=
2 for quadrupole transitions,
and where
10 21 sec" 1
10- 13
197
cm
Mev
ray for which 2ttR JX
1/5, in seconds, calculated
and for
with
A =
R =
200.
ray transition indicated,
value of L, as
1.
Table (16-3)
from these equations for
cm corresponding to R =
can decay only by the single
10~ 13
If a state
is equal to this t. Note that
and magnetic transitions with any
well as the relative magnitudes of S for electric
the relative magnitudes of
common
its
lifetime
for electric
648
THE NUCLEUS
[Ch. 16
L and the relative magnitudes of S for
magnetic transitions with various values of L, agree with the implications
of the previous paragraph. Also note that for each value of L the dependence of S on the energy hv of the emitted gamma ray agrees with the
transitions with various values of
implications of that paragraph since X
dependence before
cjv.
We
have seen
energy
this
in the case of electric dipole transitions in
atoms
(the
v 3 factor in equation 13-99).
TABLE
The Reciprocal
Transition
10
Mev
Elec. dipole
Mag. dipole
4 x lO" 19
4 x 10" 17
Elec. quadrupole
Mag. quadrupole
Elec. octupole
Mag. octupole
Elec. sixteenpole
Mag.
sixteenpole
(16-3)
Model Transition Rates for
of the Shell
x 10 -16
x 10" 14
x
x
x
x
Mev
0.1
6 x 10" 18
6 x 10" 14
1
lO" 13
10" 11
10~ 10
10~ 8
Gamma
Mev
0.01
4 x 10~ 13
4 x 10" u
x 10" 11
x 10" 9
x 10 -6
x
x
x
x
x
x
1
1
1
x 10~ 4
x 10" 1
x 10 1
1
1
1
Decay
Mev
10" 6
4 x 10" 10
4 x 10~ 8
2 x 10~ 2
10" 4
2 x 10
10 1
10 3
10 8
x 10 10
x 10"
10 10
x 10 19
x 10 8
Since the electric dipole transition has the largest transition rate for
a given value of
hv, the question
immediately
Why
arises:
do not
all
gamma
is
ray decays proceed by an electric dipole transition? The answer
that in a gamma decay between initial and final states of given spin
quantum numbers
sitions are allowed
For
and if and given parities, only certain types of
by the following selection rules.
i(
tran-
electric transitions:
\i
A/
<L< +
i{
(but not
it
-+
i,
0)
(16-83)
The nuclear parity must change if L is
odd and must not change if L is even.
For magnetic
\it
transitions:
Ai
^L^ +L
(but not L
i{
= > =
i,
0)
(16-83')
The nuclear parity must change
even and must not change if L
The gamma decay
will
if
is
odd.
is
always proceed by the allowed transition having
Because of the strong L dependence of S, this
the largest transition rate.
means a
transition with
L=
Ai.
electric transition if the initial
If this value of
and
final states are
is
odd,
it
will
be an
of the opposite parity,
GAMMA DECAY AND INTERNAL CONVERSION
Sec. 12]
and a magnetic
value of
same
is
parity,
transition
even,
it
these states are of the
if
be an electric transition
will
and a magnetic transition
if
same
649
parity.
If this
these states are of the
they are of the opposite parity.
if
always the case, the selection rules for the various transitions arise
from the form of the matrix elements of the interaction potentials which
induce these transitions. The part of the rules relating the possible values
As
of
is
to the values of
and
i(
i,
that all these matrix elements
tum
in the transition,
momentum
angular
subject
it is
shown
can be understood most easily if we realize
as to conserve angular momen-
must be such
and that the
gamma
ray
must therefore carry away
In advanced treatments of this
as well as energy.
magnetic transitions associated
that, in the electric or
with a certain value of L, the quantum number specifying the magnitude
of the angular momentum carried by the gamma ray is L, and also that
ray cannot carry zero angular momentum. Granting these
statements, and writing the first part of the selection rules as
gamma
L=
we
see,
\h
i \,
f
\h
if\
1.
( but
this part
by comparison with (13-40), that
not
'i
simply represents the
requirements of conservation of angular momentum.
To understand the second part of the selection rules,
we must actually
potentials.
interaction
of
the
some
write
We do this for the simplest casesthe electric and magnetic dipole interaction potential matrix elements, ignoring the motion of the CM of the
down
the matrix elements for
These
nucleus.
are, respectively,
E-j"v/*2iWlT
H
The parameter a which
jy*
is
I {bjL, +
+e
(16-84)
(16-84')
c,S,M dr
for protons
and
for neutrons, relates the
of the nucleons to their electric dipole moments the parameters
bj and c which are the orbital and spin g factors for protons and neutrons,
relate the orbital and spin angular momenta L 3 and S,- of the nucleons to
positions
r,
their
magnetic dipole moments
the sums, taken over
all
the nucleons of
the nucleus, give the total nuclear electric and magnetic dipole moments.f
The matrix elements of these dipole interaction potentials initially contain
inside the integral signs the dot product of the
moments with
the electric
and magnetic fields E and H. However, both fields have a negligible
spatial dependence over the nucleus as far as the dipole moments are
t
The
effect
of the
matrix elements.
CM
motion
is
to increase the contribution of neutrons to these
650
THE NUCLEUS
concerned, and can be factored out.
section, r 3
is
[Ch. 16
Now, as we mentioned in the previous
S are axial vectors. By inspecting
a polar vector and L, and
Coulomb's law and Ampere's law, the reader can immediately verify that
E is a polar vector and H is an axial vector. Thus the electric dipole
interaction potential matrix element
is a dot product of two polar vectors,
the magnetic dipole interaction potential matrix element is a dot product
of two axial vectors, and so we see that both are scalars, in agreement with
a statement
made
in the previous section.
vanishes identically, unless
the polar vector o 3 r 3
y> t
and
y> f
We
also see that (16-84)
are of the opposite parity, because
of odd parity and,
if the entire integrand is of odd
from each point (x, y, z) will be
canceled by the contribution from the point ( x, y, ). We further see
that (16-84') vanishes identically, unless ip and rp are of the same parity,
t
as the axial vector 6,L, + c,S, is of even parity. Both of these conis
parity, the contribution to the integral
clusions are in agreement with the second part of the selection rules for
L
imagine evaluating the dot product in the electric dipole
1.
Now
interaction potential matrix element (16-84).
integral containing
This gives the
sum of an
x j in the integrand, an integral containing y
}
integrand, and an integral containing
in the integrand.!
in the
As we know,
the terms x t , y t z } are of odd parity. The electric quadrupole interaction potential matrix element consists of a similar set of integrals, each
all
containing a term such as xj, x{y t yj, etc., in the integrand. | As each of
is of even parity, it follows that all the integrals vanish iden,
these terms
%p t and y> f are of the same parity.
This is in agreement
with the selection rule quoted in (16-83) for L
2.
For L
3 the
integrals contain terms of odd parity such as xf, x-y-z^, etc., and this leads
tically unless
to the selection rules quoted in (16-83) for
L=
3.
Similar things happen
in the integrals that comprise the magnetic transition matrix elements,
and these lead
to the selection rules
quoted in (16-83') for the higher
values of L.
In
many gamma decay transitions,
and hv
is
along with the
cally to
particularly those in
which L
is
large
small, several groups of monoenergetic electrons are emitted
gamma rays. The energies E of the groups are found empirigamma rays by the equation
be related to the energy hv of the
E= hv - W
(16-85)
where the value of Wiox the most prominent group
energy of a
K shell electron
of the
gamma
is equal to the binding
decaying atom, and the values
t In this form the interaction potential matrix element (16-84) times its own complex
conjugate can easily be identified with the expression (13-77) that we derived for the
electric dipole transition rate
% Cf.
of atoms.
footnote on page 645.
GAMMA DECAY AND INTERNAL CONVERSION
Sec. 12]
651
of W-'for the other groups are equal to the binding energies of the electrons
in the L, M, N, etc., shells. This makes it quite evident that the various
groups of electrons are emitted from the various shells of the atom, so
that the energy of one of these electrons upon leaving the atom equals the
energy of the gamma rays less its binding energy. The similarity between
(16-85) and the photoelectric effect equation (3-8) suggested to
coverers, Ellis
and Meitner
(1921), that the electrons are ejected
its dis-
through
gamma ray delivers
energy to an atomic electron by the photoelectric effect before
leaving the atom.
a process of internal conversion, in which an emitted
all
its
However, calculations of Taylor and Mott (1933) show that the probainternal conversion by the photoelectric effect is very small
compared to the probability for internal conversion by the direct transfer
of energy from the nucleus to the electron through the Coulomb interaction between the two. In this process the nucleus is de-excited and the
electron is ejected. Except for the differences in the interaction potential
and in the initial and final eigenfunctions for the system, the calculations
are quite similar to those involved in beta decay theory. In recent years
Rose has used this picture of direct energy transfer to make a number of
bility for
calculations of the ratio of the probability for emitting an electron
given shell to the probability for emitting a
called the interna! conversion coefficient a.
The
gamma
ray.
from a
This ratio
is
calculations of a should be
very accurate because the factors involving the
eigenfunctions cancel
known Coulomb
accurately
known
initial and final nuclear
each other, leaving only factors involving the
interaction
eigenfunctions for the
electron.
initial
potential
bound
and the accurately
and the final free
electron
Figure (16-44) shows a 7C the value of a for the emission of a AT
,
shell electron, as calculated for the 40 Zr
atom. Because the probability
of internal conversion increases as the value at the nucleus of the bound
electron eigenfunction increases, a /c rapidly becomes larger as
becomes
For the same reason, at a given Z and hv, ol k is usually larger than
y.
L or a u or a Y etc. Furthermore, at a given Z and hv, the quantity
yK jy. L depends quite sensitively on the character of the gamma ray
transition, that is, on the value of L and on whether the transition is
electric or magnetic. Thus accurate measurements of a. ja.
K L which are
relatively easy to make, provide a very good method of determining the
larger.
.
character of a
gamma
ray transition and, therefore, the relative spins and
parities of the nuclear states involved in the transition.
compete with gamma ray emission in the
The processes are independent,
and the probabilities for the two processes are additive. Thus the total
rate S for transitions between the initial and final nuclear states is the sum
Internal conversion does not
sense that one process inhibits the other.
THE NUCLEUS
652
[Ch. 16
10>
10
ia
io
10
g io~
10
-3
i
10"
*
10"
0.4
0.2
0.1
Nuclear transition energy (Mev)
Figure
16-44.
The K
shell
internal
conversion coefficients.
broken curves, magnetic transitions. From
McGraw-Hill Book Co., New York, 1955.
transitions;
of the transition rate
Sic
SY
for
gamma
That
for internal conversion.
Solid
R. D. Evans,
curves,
electric
The Atomic Nucleus,
ray emission plus the transition rate
is,
Sy
Sic
(16-86)
This can be written in terms of the internal conversion coefficients since
sic
where
<x
is
( k a-
ai
<*-m
-)S y
If the initial state
S,(l
can decay only to a single
==
S
final state, as is usually true
shows that the
lifetime r
initial state is
Thus
for isomeric transitions, equation (13-118)
As
a<S y
the totai internal conversion coefficient.
of the
S 7 (l
(16-87)
()
can be accurately calculated, the experimental values of t can be
compared with the values predicted by the Sy of (16-82) and
(16-82').
Sec. 12]
GAMMA DECAY AND INTERNAL CONVERSION
653
Figure (16-45) shows such a comparison for a group of isomeric transitions
that have been identified as magnetic sixteenpole (A/
4, parity change).
The dots are experimental and the curve is theoretical. For these transitions, we see that the predictions of (16-82) and (16-82'), which are based
on the shell model, are reasonably accurate. But inspection of figure
(16-23) will demonstrate that in the shell model all A/ = 4, parity change
100
200
Energy (kev)
Figure 16-45.
The
lifetimes for a
M. Goldhaber and A. Sunyar, Phys.
500
1000
group of magnetic sixteenpole transitions.
906 (1951).
From
Rev., 83,
transitions are between states quite near those filled at the
magic numbers,!
where the model should be accurate. For transitions of
a different character, the shell model predictions are not so accurate.
However, the shell model does explain one striking feature of gamma
and
this is exactly
ray transitions from nuclear states of excitation energy of several Mev or
less. In transitions from these states, which are the states excited by beta
decay, electric dipole transitions are almost never observed. The selection
show that electric dipole transitions require A/ = 1, parity change.
But figure (16-23) shows that there is a particularly large energy difference
between any two states for which the transitions satisfy this condition.
rules
The
parity of a shell
example of a A<
and 3p,A
.
model
4, parity
state is even if / is even [cf. equation (13-112)]. An
change transition would be the transition between 5g%
THE NUCLEUS
654
[Ch. 16
A^
CO"!
^ 3 =*X
c
as
"?o
^ o
O 3
E<
I =
I-
LU?
!5<
txo
E X
suo)OJd )0 JsquinN
PROPERTIES OF EXCITED STATES
Sec. 13]
This energy
65S
almost always greater than several Mev, and so it is enerwhich can decay by an electric dipole
is
getically impossible for a state
transition to be excited
energy
is
by beta decay. However, when enough excitation
and they have
available, electric dipole transitions are found,
transition rates
which are not
in disagreement with theory.
Properties of Excited States
13.
In addition to the information about the excited states of nuclei obtained
from the study of internal conversion, gamma ray emission, and alpha
and beta decay to excited states, much information is obtained from the
study of nuclear reactions. Consider the general case (16-16), which is
written
a
If the
A^B + b
energy of the bombarding particles a
distribution of the product particles b
is
usually found to consist of a
is
is
held constant, and the energy
measured,
this
energy distribution
number of discrete groups. Figure (16-46)
shows an example the energy spectrum of protons produced in the
reaction 1 H 2 + 13 A1 27 -- 13 A1 28 + 1 H 1 for a bombarding deuteron energy
of 2.1 Mev. (A nucleus of the hydrogen isotope 1 H 2 is called a deuteron.)
The labeled groups can be identified as originating from impurities in the
target foil, but all others are proton groups from reactions on the target
nucleus 13 A1 27 When a proton is emitted in the highest energy group, the
residual nucleus 13 A1 28 is left in its ground state; when a proton is emitted
in a lower energy group, the 13 A128 is left in one of its excited states. Thus
the spectrum of excited states in the residual nucleus is immediately
obtained from the energy spectrum of the product particles [after using
.
equation (16-20) to
This
is
shown
make
small corrections for the recoil of this nucleus].
in figure (16-47).
Excited states can also be located by measuring the cross section for
reactions between a
bombarding
and a target nucleus, while
With low A nuclei and
low bombarding energies, the cross section is usually found to exhibit
resonances, or certain energies at which the probability of reactions taking
place is unusually large. As an example, figure (16-48) shows the relative
cross section for producing reactions by the bombardment of 13 A1 27
particle
varying the energy of the bombarding particle.
with
W.
Resonances occur at energies that excite the system consisting of the
bombarding particle plus the target nucleus, which is called the compound
nucleus, to one of its excited states. This is because at these energies the
THE NUCLEUS
656
[Ch. 16
4~
i
>
0)
c
LlJ
2-
Figure 16-47.
The energy
and A. Sperduto, Phys.
levels of
the nucleus
13
AI 28
From H.A. Enge, W. W. Buechner,
Rev., 88, 961 (1952).
eigenfunction for the bombarding particle has a particularly large value
in the region of the nucleus,
and thus the cross section
is
greatly increased.
These resonances are analogous to the resonances discussed in section
6, Chapter 15, and the excited states of the compound nucleus are analogous
to the virtual states mentioned there. This is a very interesting situation.
Since the
compound
nucleus
is
obviously energetically unstable to the
some other particle, it might
bombarding
excited
states
would be observed if
well-defined
no
unbound, its allowed energies form a continuum. We shall
re-emission of the
particle, or
be expected that
a particle
is
can actually exist
though the total excitation
see in the next section that well-defined excited states
for sufficiently long times to be observable, even
PROPERTIES OF EXCITED STATES
Sec. 13]
657
15.5
0.500
0.600
uy U uu
0.700
0.800
0.900
1.000
Proton energy (Mev)
Figure 16-48.
Resonances
in
1.100
1.200
1.300
1.400
the cross section for producing reactions by the bombardK. J. Brostrom, T. Huus, and R. Tangen, Phys.
ment of aluminum with protons. From
71,661 (1947).
Rev.,
energy
is
particle,
the
great
enough
to liberate a particle if it
because the energy
compound
nucleus,
and
is
usually divided
is all
concentrated on that
among many
particles of
also because reflection at the nuclear surface
inhibits particle emission.f
The data of
figure (16-48) locate states of the compound nucleus
a range of excitation energies starting at 0.5 Mev above the
11.59-Mev binding energy of a proton to that nucleus. Lower energy
protons cannot be used because they have negligible probability of getting
through the Coulomb barrier of the target nucleus. However, in reactions
14
Si 28 in
t These very sharp many particle resonances are definitely not the same as the very
broad single particle resonances, mentioned in section 3, that are observed in the cross
sections of nuclei for several Mev nucleons. The distinction will be made apparent in
the next section. In atoms the broad single particle resonances are just the Ramsauer
resonances in the cross section for scattering electrons, but the sharp many
have never been observed.
This is due, presumably, to difficulties in obtaining adequate electron energy resolution
(~10- 6 ev?). However, sharp many particle resonances probably do exist because the
c?-o^"potential for the interaction of a neutral atom and a free electron is qualitatively
similar to that in the nuclear case, in that reflection from the "surface" should inhibit
effect
particle resonances in the cross section for capturing electrons
electron emission.
THE NUCLEUS
658
compound
induced by neutrons states of the
the
way down
[Ch. 16
nucleus can be located
all
to excitation energies equal to the binding energy of the
neutron to that nucleus, which is typically 6 to 8 Mev.
By measuring the full width at half maximum T of the resonances
observed in these experiments, the lifetimes of the states may be obtained
from the
relation
(16-88)
HIT
which applies to the widths and
lifetimes of states excited by the resonant absorption of quanta (including
gamma rays). Its applicability to states excited by the resonant absorption
of particles has been justified by Breit and Wigner. This will be discussed
This
is
same
the
relation as (13-125),
in the next section.
proton or neutron experiments using bombarding energies
than several Mev, only states of the compound nucleus with spin
For
less
either
quantum numbers very
close to the spin
nucleus will be excited.
The reason
particle is so small that
momentum (KR
1)
it
is
quantum number of the target
bombarding
that the energy of the
must interact in a
and, therefore, angular
state
of low orbital angular
momentum can be conserved
the spin change is low. In higher energy nuclear reactions, more
can be excited. Furthermore, experimental and theoretical studies
of the angular distributions of the product particles can sometimes give
information about the spins and parities of the states of the residual
only
if
states
nucleus that are
excited after the emission of these particles.
left
some feeling for the procedures used in studying the
and parities of the excited states, let us consider
studies. Briefly put, there seems to be little rhyme or
Now that we have
energies, lifetimes, spins,
the results of these
reason in the
results.
That
is,
few of the regularities, which are so
an atom, are found in the excited
characteristic of the excited states of
states of
1.
a nucleus. There are only three well-established exceptions:
In the
first
few excited
states
of nuclei, particularly those with
N or
Z near magic numbers, certain shell model characteristics are often found.
An
example, which
we
discussed in the previous section,
is
the absence of
between these states.
2. In the first few excited states of even N, even Z nuclei, particularly
and Z far from magic numbers, certain characteristics of the
those with
rotational motion predicted by the collective model have been found.
For a rigid body rotating about a point at the center of its axis of symelectric dipole transitions
metry, the eigenvalues are
E =
21
i(i
1),
0, 2, 4, 6,
(16-89)
PROPERTIES OF EXCITED STATES
Sec. 13]
659
where /is the moment of inertia about the axis of rotation, and the parities
of all the eigenfunctions are even.j Every even N, even Z nucleus has
an i = 0, even parity ground state. In many of them the first excited
= 2, even parity, and the second excited state is i = 4, even
state is
;'
parity.
some
In
cases the third excited state
more, the energies of these
able values of
3.
even parity. Further-
6,
equation (16-89) for reason-
/.
In certain sets of light nuclei having
few excited
is
states agree with
After corrections are
made
of the different nuclei of a
approximately equal.
common
values of A, the
found to have corresponding spins and
states are
for the different
set,
This
is
Coulomb
first
parities.
repulsion energies
the energies of the corresponding states are
particularly true
if the' set
consists of
two
mirror nuclei.
Beyond
first few excited states, even these modest beginnings of
no longer observed, and the energies, lifetimes, spins, and
seem to be essentially random. But this should not be surprising
the
regularity are
parities
the nucleus
is
much more complicated than
theory of the atom, the net potential V(r)
mining the behavior of the electrons, and
perturbations. In the shell
V(r),
but the other
because
is
it
the atom.
the
is
all
in deter-
the other effects are small
model of the nucleus there
effects are
In the Hartree
dominant factor
is
also a net potential
not small perturbations.
Furthermore,
only a model instead of a theory, there are additional
and probably a number of others) that the
and these effects are also not small perturbations.
effects (the collective effects,
shell
model
leaves out,
Apparently in a real nucleus there are a number of different effects of
comparable importance that govern the properties of the excited states
and, as a result, it is not possible to give a description of the individual
properties of every excited state, except
However,
it is
by
them
listing
in a table.
possible to give a simple description of certain average
With respect to their energies, this is done
by specifying the average energy spacing D between adjacent states. Some
properties of the excited states.
values of
are given in table (16-4).
from one case
to the next,
average spacings must be taken
indicate the general trends,
Because of the large fluctuations
and also because of large
cum grano
uncertainties, these
Nevertheless, they do
salis.
and these trends are easy
to understand.
In the shell model, at very low excitation energies only one of the nucleons carries the total excitation energy of the nucleus, and so the spacing
Chapter 10. The / = 1, 3, 5,
states are absent because
odd, whereas only even parity states are possible since the
symmetrical system to which equation (16-89) applies is transformed into itself in the
parity operation; that is, the system is completely symmetrical with respect to reflection
t Cf. results of exercise 6,
in these states the parity
in the origin
is
and must therefore be described by eigenfunctions of even
parity.
THE NUCLEUS
660
TABLE
[Ch. 16
(16-4)
Average Spacings between Excited States of Nuclei
Excitation
energy
~10 6
~103
~102
~0 Mev
~6 to 8 Mev
~16 Mev
of the excited states
the shell model.
is
~ 20
~ 200
ev
~10 5
ev
~10 ev
ev
ev
the relatively large spacing of the low lying states of
In the collective model, the low lying excited states are
mainly rotational, and
excitation energies
their spacing
no known model
is
At higher
also relatively large.
adequate, and the nucleus
is
best thought of as a complicated system of
many
particles all of
is
perhaps
which can
ways in the excitation of the system. As there are a
number of divisions of excitation energy between the many particles
lead to approximately the same total excitation energy, it is evident
participate in various
large
that
that the spacing between the higher excited states of the system will be
small.
It is
also evident that this spacing will decrease with increasing
and with increasing A.
approximate description of the energy dependence of the average
spacings that are observed can be obtained by using the Fermi gas model
total excitation energy E,
An
Simple thermodynamic arguments, starting
for the highly excited nucleus.
from the
relation
E = aikTf
between the
energy
total excitation
of a Fermi gas and
T, lead to the following expression for the density
Po
(16-90)
e^VE
its
temperature
of excited states p
l/D:
(16-91)
The quantity a = tj^jV lAEt where jV is the number of Fermi particles in
the system and Ef is its Fermi energy (12-38). Treating the neutron gas and
the proton gas separately, and evaluating yT and Ef in terms of N, Z, and
the nuclear volume as in equation (16-35), we find that
,
where a
The
is
measured
in
~ A/10
(16-91')
-1
Mev
lifetimes of the excited states are usually expressed in terms of
As we have indicated, the relation (16-88) is valid for all
Thus any excited state with lifetime t has a full width at
half maximum Y, where
their widths V.
excited states.
HIT
(16-92)
PROPERTIES OF EXCITED STATES
Sec. 13]
661
We
have seen that for excitation energies up to several Mev the excited
can decay only by gamma ray emission and internal conversion,
and that the lifetimes range from about 10~ 15 sec to as long as 10 8 sec.
states
Since h
to
less
is
~ 10~
10~ 23
15
ev.
means that Y ranges from about T
lO" 1 ev
For excitation energies somewhat above this range, but
ev-sec, this
>
than the binding energy of a nucleon, the evidence indicates that T
E or A. It varies from r <~ 1 ev near the
not strongly dependent on
neutron binding energy in nuclei with
in nuclei with
neutron binding energy
in this energy range there are very
20,
<
~ 200.
many
to
T~
10"" 1
ev near the
This comes about because
lower energy states to which
the excited state can decay, so there will always be states available
and
whose
such as to allow decay of the excited state by electric
dipole transitions with lifetimes of the order of 10~ 15 sec.
spins
parities are
As the excitation energy goes above the binding energy of the last
nucleon in the nucleus, typically 6 to 8 Mev, decay by the emission of a
nucleon becomes energetically possible. In some nuclei, proton emission
becomes energetically possible first, and
possible
but in
first,
all,
in others
neutron emission becomes
except those of very small Z, the
Coulomb
barrier
proton emission until the excitation energy is quite
high. When nucleon emission is energetically possible, the total rate for
transitions from the excited state will be larger than that due to transitions
involving gamma ray emission and internal conversion alone. This
effectively inhibits
decreases the lifetime and increases the width.
inversely proportional to lifetimes
and
In fact, since widths are
lifetimes are inversely proportional
to transition rates, widths are proportional to transition rates
wnte
= ry + r B
because transition rates are additive.
width due to
all
and T n means
gamma
(16-93)
In this expression,
Y y means
the
ray and internal conversion transitions,
the width due to neutron emission (or nucleon emission if
proton emission
It is
possible
and we can
is
important).
easy to see what the general form of the energy dependence of the
neutron width
T n must
be.
Imagine using Golden Rule No. 2 to calculate
initial state consisting of the nucleus in an
the rate for transitions
from an
excited state of energy
toa
final state consisting
of the nucleus in
its
ground state plus a nucleon moving off with kinetic energy (E Eb ),
where Eh is its binding energy. The matrix element for the transition will
be a slowly varying function of E, and the energy dependence of the transition rate comes in mainly through the density of states for the nucleon.
From
to
equation (12-37), we see that this density of states
)'A
Therefore we expect
b
(E E
is
proportional
Tn
x(E- E
)'
(16-94)
THE NUCLEUS
662
at least for a restricted energy range.
This
[Ch. 16
the energy dependence
is
When r > T y
and these measurements simply involve finding the widths
exhibited by the average results of the measurements.
then
r,
of the resonances in data such as those given in figure (16-48).
Ty
oc constant,
compared
to
nuclei with
~
~
Tn
while
T y when (E
~ 20,
this
E)
(E
cc
Yn
Vi
,
will
Since
always become large
becomes larger than some value. For
occurs for (E Eb )
1 ev, and at this value
b)
10" 1 ev at
T = T y
1 ev.
For nuclei with A
200, F n = T y
3
(E - Eb)
10 ev.
When (E Eb) is not too large, there is an interesting empirical relation
between the average neutron widths r of the excited states of a nucleus
and the average spacings D between these states. It is
~ CK D
(16-95)
where
K =p
2
The quantity p 2 is
mass. For almost
the
2 \h
= V2M(E-E )lh
b
momentum
all nuclei,
C=
of the emitted nucleon, and
the constant
0.5
10" 13
is
is its
within a factor of 5 of
cm
(16-95')
T n vary by many orders of magnitude.
from (16-91) and (16-94), we see that TJD increases
with increasing excitation energy. At the point where TJD = 1, the
excited states begin to overlap and form a continuum. As an indication of
the general trend, it might be said that for nuclei with A <~ 20 the continuum
begins at (E Eb )
10 Mev, and for nuclei with A
200 it begins at
10" 1 Mev.
(E ~ Eb)
although the values of
From
(16-95), or
14.
Nuclear Reactions
Nuclear reactions provide
of excited
states.
much
useful information about the properties
In turn, the properties of excited states provide the key
to the understanding of nuclear reactions. In this section
we
shall use this
point of view to give a simplified version of the present theoretical description of nuclear reactions.
and Wigner
This description
is
due to Bohr (1936), Breit
and many others.
(1936), Weisskopf (ca. 1950), Butler (ca. 1953)
Before the discovery (1936) of very sharp resonances in the cross sections
by low energy neutrons, it had been assumed that
for reactions induced
nuclear reactions involve a direct transition from an
of the bombarding particle
initial state
and the target nucleus to a
consisting
final state consisting
NUCLEAR REACTIONS
Sec. 14]
663
of the product particle and the residual nucleus.
However, the discovery
of resonances replaced this direct interaction theory by the compound
nucleus theory. Bohr proposed the latter theory because he realized that
in reactions exhibiting resonances there
compound
long lifetime, which he called the
relatively
is
must be an intermediate
nucleus.
state of
This state
necessary because such reactions are essentially measurements of the
incident neutron energy, since the cross sections are very rapid functions
of energy. Therefore the uncertainty principle demands that the duration
hjT, where the resonance widths
of these measurements be at least t
The compound nucleus theory was
are their energy resolutions.
number of years
quite
was thought that it could explain
Recently, however, it has been found that some
all nuclear reactions.
reactions actually are most easily explained on the basis of a direct interand
successful,
for a
it
action theory.
We
shall describe first the
to distinguish
compound
nucleus
separated since
nucleus
rjD >
is
(a),
nucleus theory.
E)
(E
is
excited into a region in which
TJD <
(b),
(E
b ) is
large
It is
convenient
small enough that the
its
states are well
enough that the compound
which its states strongly overlap since
both cases the initial step in the nuclear reaction
This is the step described by the optical model.
excited into a region in
is
1
compound
between two cases:
However,
in
essentially the same.
Consider a beam of particles bombarding a target of nuclei. According
to the optical model, a particle striking a nucleus will either be scattered
is
beam or removed from that beam by absorption. If it
absorbed it will initiate a nuclear reaction, but the optical model cannot
tell us anything about the details of the reaction because it only describes
the particle moving through the absorptive nucleus in a single particle
state of the complex potential that represents the initial particle-nucleus
interaction. Because the potential is fairly absorptive for all positive
values of (E ),t this single particle state will have a very short lifetime,
or very large width. In fact, since it immediately follows from (15-106)
out of the incident
is
that, if the
W,
imaginary part of the complex potential has a constant value
the width of the single particle state
(16-10) shows that
the lifetime
ev-sec/10 7 ev
The very
2W ~
10
Mev
of the single particle
10~ 22 sec.
many
consisting of the
2W, and
state
is
(E
typically
since figure
- E ~ 10 Mev,
t = hjT ~ 10^ 15
b)
particle states that correspond, within the limits of
bombarding
E ) for the
E of figure (16-10).
Here we use (E
equals the
short lifetime single particle state rapidly decays into the
the uncertainty principle, to the
longer lifetime
It
is
for the typical case
same
total excitation energy
of the system
This is the
particle plus the target nucleus.
external kinetic energy of the
bombarding nucleon.
THE NUCLEUS
664
system we
compound
call the
These many particle
section.
One
many particle states
we were discussing in the
nucleus, and the
precisely the higher excited states that
particle state.
[Ch. 16
states are certainly different
difference
is
that in the cases
we
from the
are
last
single
are discussing the
measured in Mev while the widths of
measured in ev. Another is that the single
particle state and the many particle states do not describe the same physical
situation, even though they correspond to the same total excitation
energy. In the single particle state all the excitation energy is carried by a
single particle, but in any of the many particle states it is distributed over
a large number of particles. The excitation of the many particle states by
the single particle state takes place through a process in which energy is
transferred from one to the other in multiple collisions of the bombarding
width of the single particle
the
many
state is
particle states are
particle with the particles of the nucleus.!
Of
course, the
many
particle states
of the
eventually decay with an average lifetime r
take
~ 10 ev as a typical many
compound nucleus
hjY <~ 10~16
particle state width.
will
where we
sec,
We know that,
in
must be by nucleon emission since, except at small
(E Eh), r > T y for many particle states. But it is perhaps not apparent
how this can happen. The point is that in the many particle states the total
excitation energy E is distributed among many nucleons, and in order for
any nucleon to escape from the nucleus the nucleon must have an excitation energy at least equal to its binding energy Eb Thus a large fraction
of the excitation energy must somehow become concentrated again on a
single nucleon if it is to escape. Some idea of the way this comes about,
and also some idea of the process in which the many particle states are
initially excited, can be obtained from the following argument.
Let *F be the wave function for the compound nucleus, and y> k be the
general, this decay
eigenfunctions for the
Assume
many
particle states that are excited in that nucleus.
that the spacing between
particle states
is
all
adjacent eigenvalues
Ek of these many
a constant equal to their average spacing D, so that
Ek = E +
kD, where E is a constant and k = 1, 2, 3, ...
=
wave functions Tj. for the many particle states are
k
-*>c:Dt f
ip e-iEo l> e
linear
of
the
Next
write
as
a
combination
k
t
Y=2
k
Then the
tp
^=
k
e-^l* J
a kWk e
~ iWtin
lE" tin
That
is,
k e-
(16-96)
where the summation is taken over the many particle states that are excited.
In this expression t =
is the time at which they are initially excited, and
t This gives rise to
a simple relation between the sharp
the broad single particle resonances
strength,
TJD,
of the
many
the
many particle
resonances and
single particle resonances define the average
particle resonances.
NUCLEAR REACTIONS
Sec. 14]
= A ke'
ak
l4k
constants
same
<f>
zero.
where the
665
real quantities
Ak
are their amplitudes
are phase factors that allow the
Now
and the real
have the
in all the ^Yk to
note that the summation appears to be a perfectly
t, with repetition period
periodic function of
T=2TThjD
(16-97)
At t = 0, *F actually describes a single particle state, since at that time the
compound nucleus is just being formed and all the excitation energy is
carried by the bombarding particle. If the summation is perfectly periodic
with period T, then
(and also IT, 3T,
again describe a single particle state at
will
In fact,
etc.)
it
will describe exactly the
same
= T
single
was initially formed, and all the total excitation energy
be concentrated on the single particle. This particle will then escape
the nucleus, providing it is not reflected by the change in potential at
particle state that
will
the surface of the nucleus.
Let us consider case
which the quantity (E
Here
b ) is small.
For a proton, the
Coulomb barrier makes the probability of escape negligible. For a neutron,
the probability of escape, P, can be estimated from equation (8-46). It is
in
(a),
reflection at the nuclear surface
is
very important.
P = 4A 1 A 2 /(A 1 + K2f
(16-98)
where
K = pjh ~ V2MVJH
x
and
K = P% \h = VlM(E - E )lh
2
The quantity p1
the nucleus.
the
is
We
momentum
of the neutron of mass
equate this to the
momentum
it
M when inside
would have, approxi-
when moving in an optical model potential with real part of depth
The quantity p 2 is the momentum of the neutron after escaping.
mately,
(E
Equation
b ) as shown.
based on the approximation that the potential at the surface of
the nucleus changes rapidly in a distance of the order of 1 de Broglie
wavelength. This approximation is not bad for small (E Eb ). FurtherIt is
related to the external kinetic energy
(16-98)
is
more, with small (E
E K >K
b ),
and
so
P ~ 4Kx K2 lKf = 4KJK!
Now
(16-99)
the probability per second that a neutron will be emitted
nucleus
is
equal to
T" 1
number of times per second
the
that
from the
T repeats
its
by the probability P that any one
escape. Thus the lifetime for neutron
original single particle form, multiplied
of these times the neutron
emission
will
is
1/r-ip
T/P
(16-100)
THE NUCLEUS
666
hjr
H4K D
_ h4K
__
and the neutron width r
(16-99) this
Tn
is
HPjT.
From
[Ch. 16
(16-97) and
is
2
2K 2 D
Ink
*i
ttK
or
2h
Tn
;J1 K 2 D
(16-101)
where
K = ^2M( - ,)/*
V = 50 Mev that is indicated
2
Taking the value
2h/W2MV =
the coefficient
(16-101)
is
in excellent
0.39
probability
repeats
its
is
small
by the
optical model,
cm, and we see that equation
Note from equation
agreement with (16-95).
(16-100) that the lifetime t of the
the repetition period T,
10" 13
when (E
many particle
b ) is
states is
much
longer than
small, simply because the escape
the wave function Y for the compound nucleus
original single particle
form many times before the
particle is
emitted.
not true for case (b), in which (E Eb ) is large, because in these
1. To handle this case we must
circumstances the escape probability P
The quantities A k which are
argument.
correct a weak point in the
are not constants, but on
equation
(16-96),
the amplitudes of the a k in
This
is
the average decrease in time according to the lawf
A k (t) = A k(0)e~ mT*
(16-102)
where rk = hjT k is the lifetime of the many particle state of width Y k
This can be ignored for the purpose of evaluating the repetition period T
T and the A k (t) are very nearly constant over
in case (a), since all the T
any repetition period. But it cannot be ignored in case (b), since the
T does not obtain. Now, if all the rk had exactly the same
condition rk
tl2r
this term would
value t, and all the A k (t) had the same t dependence e~
outside
as a term
on
the
and
appear
(16-96)
factor out of the summation
leave the
would
nucleus.
This
compound
the
the
decay
of
describing
not
all have
r
do
But
the
of
t.
function
periodic
perfectly
a
summation
k
exactly the same values. Instead they have a spread of values, and thus
.
fc
>
>
not a perfectly periodic function of t. Even so, it will be
if the spread in the rk is not too large.
for the compound nucleus should, within a cerThen the wave function
tain time, return to a form that has an approximately single particle
the
summation
is
an approximately periodic function
character,
t
and the
particle
is
emitted on the
first
occasion on which this
the nucleus in the many particle state k
simply the usual exponential decay law.
As the probability of finding
equation (16-102)
is
is
a?a k
Al(t),
NUCLEAR REACTIONS
Sec. 14]
667
However, the particular approximately single particle wave
returns should have no necessary connection with the
exact form of. the initial single particle wave function. This will certainly
happens.
function to which
be true
if a. large
nucleus,
and
number of overlapping
if the
states are excited in the
width of each of these
states is
compound
not strongly correlated
In these circumstances, the decay of the
be independent of the exact details of its formation.
This fundamental result of the theory of compound nucleus reactions
with the widths of the others.
compound nucleus
will
also holds for case (a).
fact that the values of
period, so
Thus
actually change slightly during one repetition
T deviates slightly from being a perfectly periodic function of
W returns repeatedly to a single particle wave function that
different
from the previous
number of
all
This can be thought of as a consequence of the
A k (t)
single particle
wave function and,
cycles that occur before the particle
memory
of the exact details of the
is
in the large
emitted,
is finally
initial single particle
t.
slightly
wave
W loses
function.
In both cases, this result is only strengthened by deviations from the
assumption that the eigenvalues Ek of the many particle states are evenly
spaced. But, in any circumstances, the decay of the compound nucleus
must be independent of the exact details of its formation for case (a).
The reason is that energy conservation within the limits of the uncertainty
principle demands that only many particle states of excitation energy
t = hjY n when the
within T of the energy E be excited by the time t
compound nucleus has existed long enough to decay.f For case (a),
FJD < 1, so there can only be one many particle state within these
limits. Therefore, in case (a) the decay of the compound nucleus must
necessarily depend only on the properties of that many particle state and
cannot depend on the details of how it was excited.
After this somewhat esoteric description of compound nucleus reactions,
it might be well to recapitulate in the kind of language used in the original
papers of Bohr.
When
a particle enters a nucleus,
collides with the
it
particles of the nucleus and very rapidly gives up its kinetic and binding
energy to those particles. The bombarding particle is thereby captured, and
an excited compound nucleus is formed in a state, somewhat reminiscent
of a boiling liquid, in which the excitation energy is distributed at random
among all its particles. Each particle then has an excitation energy con-
siderably less than
its
binding energy. Occasionally,
statistical fluctuations
cause enough energy to be concentrated on a single particle to make it
If the circumstances are such that
energetically possible to escape.
f Of course, at an earlier time many particle states whose excitation energies fall
within broader limits can be excited, as we have assumed. For example, when only
10
-22
sec has elapsed (the lifetime of the initial single particle state),
over a range of
10 Mev can be excited.
Cf. section
5,
Chapter
many particle states
9.
THE NUCLEUS
668
reflection at the nuclear surface is important, this
of times before a particle
compound
lifetime
that
it
Thus the
finally emitted.
must happen a number
But, in any circumstances, the
nucleus has a relatively long lifetime and during this long
forgets everything about
must remember
it
is
[Ch. 16
how
it
was
originally
decay process
formed (except
momentum,
to conserve energy, angular
etc.).
independent of the exact details of the formation process. This means, for example, that the ratio of the probability
that the
that
it
final
is
compound nucleus will decay by nucleon emission to the probability
will
decay by gamma ray emission and internal conversion is equal
r n /T y (since transition rates are proportional to widths),
to the ratio
where T n and T y are evaluated for the state or states excited in the compound nucleus, and where this is true independent of exactly how these
states were excited. If in a certain reaction a particular compound nucleus
is formed with a particular excitation energy, and in a different reaction
(different bombarding particle and different target nucleus) the same
compound nucleus is formed at the same excitation energy, both compound
nuclei should decay in the same way. At small values of (E E), T can
y
be comparable to r, and there can be an appreciable probability that
the compound nucleus decays by gamma ray emission and internal con-
At higher values of (E Eb ),
F > F y and the compound nucleus always decays by particle emission.
If enough excitation energy is available, the compound nucleus can decay
either by neutron emission or by proton emission. For quite high excitation
energies, more than one particle can be emitted.
Figure (16-49) shows the measured cross sections for the set of reactions
version to form a completely stable nucleus.
,
+ 01
+ 01 + 01
29
Cu62 + 01 + H
30
1 _|_ 29
Cu
63
_+
30
Zn 63
Zn62
and
also for the set
He
28
Ni
60
so
Zn 63
oi
30
62
01
_|_
01
+ R
+
_+ Zn +
29
Cu62 +
01
1
bombarding energies. The abscissa is the kinetic energy of
bombarding proton for the first set of reactions, and the kinetic energy
of the bombarding alpha particle in the second set of reactions; the
relative displacement of the two abscissas is equal to the difference in Et
for the two bombarding particles. The ordinate is the cross section for
the various reactions, and each reaction is labeled according to a common
at quite high
the
notation.
compound
This figure provides a test of the prediction that the decay of a
nucleus is independent of its formation, in case (b), where
NUCLEAR REACTIONS
Sec. 14]
(E
b)
is
large
669
enough that the compound nucleus
region in which there are presumably
many
widths are not strongly correlated, because for both
compound
nucleus
^Zn64 and
,
12
sets there is the
same
13
17
21
25
16
20
24
28
32
Energy of alpha particle (Mev)
Figure 16-49.
excited into a
because the abscissas have been adjusted
Energy of protons (Mev)
is
overlapping states whose
36
40
Cross sections for a set of reactions which test the predictions of the
theory. From S. N. Ghoshal, Phys. Rev., 80, 939 (1950).
compound nucleus
to equalize
its
We see that the prediction is well verified
excitation energy.
same relative probabilities for the
modes of the compound nucleus.
It has not been possible to make such a test for case (a), where (E Eb)
is small enough that the states of the compound nucleus are well separated
and only a single state is excited and such a test is hardly necessary. But
other predictions of the compound nucleus theory can be tested for this
case. For instance, Breit and Wigner have shown that the energy dependence of the cross section for absorbing an S-wave neutron (K2 R < 1) to
because for both
sets there are the
various decay
THE NUCLEUS
670
[Ch. 16
form a compound nucleus excited to a state of energy E and width
r = r + r which subsequently decays by gamma ray emission and
internal conversion, is essentially identical with the energy dependence
(13-124) of the probability for absorbing a quantum to form an excited
state of an atom, which subsequently decays by emitting another quantum.
They found
,
o(E)
KJ
^-
K\(E
-Ef+
(r/2)
(16-103)
where
X = V2M( - E
2
The term
ir\K\
S-wave neutron
is
the
(cf.
possible cross section for absorbing an
equation 15-98 with
The term YJ?\\(E
contains a factor
maximum
- Ef +
Tn
b )lfi
rj
0,
r\
rj
r\
(r/2) 2 ] gives the resonance
1).
shape;
it
its
which, being proportional to the probability for
emitting a neutron, must also be proportional to the probability for
absorbing one (cf. equation 13-84 and the associated discussion of microscopic reversibility), and
it
also contains a factor
The term r /r
compound nucleus will decay by gamma
normalization.
excited
that
comes from
is the ratio of the probability that the
conversion to the probability that
it
will
ray emission and internal
decay by any process.
The
energy dependence of (16-103) has been accurately verified in many
different experiments. Granting its validity, it may be used to obtain
values of T, F n and T y from measurements of a{E). Breit and Wigner also
found that equation (16-88), which is identical with (13-125), applies to
the resonant absorption of particles in any situation covered by case (a).
Although the predictions of the compound nucleus theory seem to be
verified for reactions that proceed through a compound nucleus process,
it has been found in recent years that there are certain situations in which
,
reactions proceed through a direct interaction process.
Figure (16-50)
These are energy
distributions of protons inelastically scattered (scattered with energy loss)
50
Sn nuclei
at five different angles in the LAB system from a target of
bombarded by a beam of protons of external kinetic energy 31 Mev.
The abscissa is the external kinetic energy of the inelastically scattered
protons, and the ordinate is the cross section for scattering a proton into a
presents
some data
illustrating
one of these
situations.
Thus the data also give the angular
Note that in the energy
section
does not depend very
cross
the
and
20
Mev
range between 10
decreasing
angle. On both
with
rapidly
increases
energy
but
on
strongly
points, this is exactly the opposite of what would be expected from the
compound nucleus theory. The theory predicts that in the long-lived
unit energy
and
solid angle range.
distribution of the inelastically scattered protons.
NUCLEAR REACTIONS
Sec. 14]
compound
nucleus essentially
all
671
sense of direction
would be
lost,
and that
the angular distribution of the protons which are finally boiled off should
be isotropic.
high
It also
Coulomb
Z=
is their external kinetic
E=
(31
N(e)
<
Mev, and N(e)
0,
10
compound
energy and
(16-90) with
51
above the 10-Mev
nucleus with a strongly
elkT
Maxwellian-like energy distribution N(e) cc ee~
energy dependent
where
predicts that the protons boil off
barrier of the
+ E ) Mev =
b
kT is
38 Mev. That
L7Mev
oc e-
obtained from equation
10
is,
the theory predicts
Mev
<e<
31
Mev.
6i
>h
135
10
15
g(Mev)
Figure 16-50.
20
25
30
35
Energy and angular distributions for a reaction which proceeds through
From R. M. Eisberg and G. Igo, Phys. Rev., 93, 1039 (1954).
a direct interaction process.
is no Coulomb barrier, and their predicted energy
N(e) cc ce -/i-7Mev over the ent i re ran g e f i n t ^ c com .
pound nucleus theory almost all the emitted nucleons would be neutrons,
since only a few protons in the tail of the Maxwellian with
10 Mev
For neutrons, there
distribution
is
>
can escape, and it would be expected that considerably less than 1 percent
of the reactions induced by the incident 31 -Mev protons should lead to
the emission of inelastically scattered protons. But, on this third point,
the data also contradict the predictions, because the magnitude of the
cross section indicates that about
1 5 percent of the reactions lead to the
emission of inelastically scattered protons.
This does not mean that the compound nucleus theory is wrong.
Instead,
it
and high
means
that inelastic proton scattering reactions
on intermediate
Z nuclei proceed through some process, other than the compound
nucleus process, which has a cross section for these reactions so much larger
than the small compound nucleus cross section that the compound
THE NUCLEUS
672
nucleus effects are completely hidden.
The process
that
[Ch. 16
involved
is
is
thought to consist of a direct interaction in which a bombarding proton
collides with a nucleon in the diffuse rim of the target nucleus, and either
the
bombarding proton or the struck nucleon
(or both) immediately
escapes.
Consider a 31-Mev proton incident upon a 50 Sn nucleus. As the de
Broglie wavelength of the proton is several times smaller than the nuclear
radius, it is legitimate to think of the collision as being well enough localized that a distinction can be
sions with the diffuse rim.
made between head-on
Now the
optical
collisions and collimodel shows that attenuation
Mev are also several times smaller than the
nuclear radius. Thus, in any head-on collision with the nucleus, the bomlengths in nuclear matter at 31
barding proton will soon collide with the nucleons of the nucleus, these
nucleons will in turn collide with other nucleons, and so the kinetic energy
of the bombarding nucleon will rapidly be shared with
all the nucleons of
nucleus process that will
eventually lead to the emission of low energy neutrons. However, in a
the nucleus. This
is
the
first
step in the
compound
rim of the nucleus, the bombarding proton could
nucleon in such a manner that either it or the struck nucleon
(or both) would immediately escape, carrying away a large fraction of the
collision with the diffuse
collide with a
As the optical model shows that the thickness of the diffuse
rim in the 50 Sn nucleus is about 10 percent of its radius, its projected area
is about 20 percent of the projected area of the entire nucleus.
This
kinetic energy.
indicates that the cross section for this direct interaction process could be
of the order of magnitude observed experimentally.
The other
character-
of this process are such that they could also be in agreement with
experiment. The process would lead to a slowly varying energy distribution
istics
of high energy inelastically scattered protons, and to an angular distribuwhich would increase rapidly with decreasing angle
tion of these particles
(if it
were not for the
initial
momentum
of the struck nucleon, and refrac-
tion in the nuclear potential, conservation of momentum
and energy would
require the angular distribution to be concentrated entirely in the forward
90 of the LAB system).
A number
of semi-quantum mechanical calculations based on this direct
shown that it is capable of a very adequate fit
proton scattering reaction, and to many similar
reactions. Butler (ca. 1953) and others have made completely quantum
mechanical calculations for this process for reactions in which it is possible
to measure the angular distribution of a single group of inelastically scattered
protons that leave the residual nucleus excited in a particular low lying
interaction process have
to the
31-Mev
excited state.
inelastic
The
calculations are basically the
mation calculations for
same
elastic scattering studied in
as the
Born approxi-
Chapter
15,
but the
Sec. 14]
NUCLEAR REACTIONS
details are
somewhat more complicated
if
673
the scattered particle imparts angular
nucleus. Instead of giving
them
in the
very much more complicated
momentum as well as energy to the
form of a long equation, the
results
of these calculations for a particular case are shown as the solid line in
figure (16-51). The points are experimental data. This is the angular
distribution for the inelastic scattering of alpha particles, of external
kinetic energy 31
Mev, from a
target of
12
Mg24
The alpha
particles
50
.4?
40
30
20
10
V^V
40
20"
Scattering angle
60
80
(CM)
Experimental and theoretical angular distributions for a direct inter-
Figure 16-51.
action leaving the residual nucleus in a single excited state.
From
S.
T. Butler, Phys. Rev.,
106,272(1957).
measured are the highest energy
leave the residual nucleus in
excited state at 1.37
in addition to the general increase in the
angle, there are very
elastic scattering
these cases.
The
pronounced
angular distribution with decreasing
oscillations that are reminiscent of the
angular distributions. This
is
typical of
from the
what
is
found
in
locations of the oscillations in the theoretical results turn
out to be a very sensitive function of the orbital angular
ferred
and they
Mev. Note that,
inelastically scattered group,
its first
momentum
inelastically scattered particle to the nucleus,
trans-
and thus
provide a powerful tool for studying the spins and parities of the excited
states of the nuclei involved.
There may be direct interaction processes other than those involving
nucleon-nucleon collisions in the rim of the nucleus. In fact, recent evidence indicates that in some cases the highest energy inelastically scattered
THE NUCLEUS
674
particles
come from a
[Ch. 16
process in which the incident particle collides with
the rim of the nucleus but, instead of interacting with a single nucleon,
the particle interacts coherently with the entire nucleus
of
low lying rotational
The
and puts
it
into one
immediately moves
having suffered only a small energy loss and angular deflection.
its
states.
particle
Direct interactions seem to play an important role in
which a
fairly
high energy product particle
reactions where the product particle
is
all
emitted;
is
different
off,
reactions in
this
includes
from the bombarding
Important examples are reactions where a deuteron 1 H 2 is the
bombarding particle, and a proton 1 H 1 or a neutron n 1 is the product
particle. These reactions apparently go through a process in which one
particle.
of the nucleons of the deuteron is captured, while the other continues on.
Recently there have been attempts to use the formalism of the compound
nucleus theory to describe direct interactions. This
is done by dispensing
with the condition that the widths of the overlapping states excited in the
intermediate system are not strongly correlated. By adjusting these cor-
which is of course permissible since the widths are not directly
measurable when the states overlap, it is possible to make the intermediate
system do almost anything from decaying immediately in a manner
dependent upon its formation (direct interaction) to decaying after a long
relations,
time in a manner independent of
its
formation (compound nucleus).
With regard
to direct interactions, this might be criticized as
complicated
way of
describing a simple process.
However,
an overly
has the
advantage of providing a single unified theory of all nuclear reactions,
including those which are compound nucleus processes but are in the
awkward
15.
range, lying between our cases (a) and (b), where
Fn
it
~ D.
Nuclear Forces
In the preceding sections
we have concentrated on
describing the
properties of the system of nucleons called the nucleus and have largely
avoided discussing the properties of the forces acting between the nucleons
of the system, even though these properties are clearly the origin of the
properties of the system.
Nevertheless,
we have
learned something about
nuclear forces.
Nuclear forces must be attractive, at least in regard to their net
on the average, because otherwise nuclei would not exist.
2. Nuclear forces must be of short range because the spatial distribution
of the nuclear interaction potential exerted by a nucleus on a passing
nucleon extends beyond the spatial distribution of the nucleons in the
nucleus by a distance which is only of the order of 10~ 13 cm.
1.
effect
NUCLEAR FORCES
Sec. 15]
675
3. Nuclear forces between two neutrons in particular quantum states
must be nearly the same as nuclear forces between two protons in the
same quantum states, not counting the Coulomb repulsion of the protons.
This follows from the fact that the total ground state binding energies of
each
member
of a set of mirror nuclei are equal, after correction
for the different total
Coulomb
is
made
repulsion energies of the two members,
and from the fact that for one member the number of neutron-neutron
interactions between nucleons in various quantum states is equal to the
number of proton-proton interactions between nucleons in those quantum
states for the other member, and vice versa (cf. figure 16-42). As the
number of neutron-proton interactions in various quantum states is the
same for both members of a set, mirror nuclei do not allow a comparison
between neutron-proton interactions and neutron-neutron or protonproton interactions.
4. Nuclear forces cannot be such that
attractive interaction potential between all
states. This follows from the fact that
energy A is approximately proportional
there
is
nearly the same purely
pairs of nucleons in all
quantum
the total ground state binding
to
for all nuclei with
A^A,
and from the following argument due to Heisenberg (1933). Consider
some nucleus and fix attention on a particular nucleon in its interior, say
the z'th nucleon. Now assume that there is a purely attractive interaction
potential, of range a, acting between this nucleon and all the other nucleons
Then the average potential energy V of that nucleon is
number of other nucleons found within a sphere of
centered on that nucleon, and this energy is negative. That is,
of the nucleus.
proportional to the
radius
a,
where p
is
the density of nucleons.
oc
-p
(16-104)
If this were the only effect, the radius
of the nucleus would collapse to the range a of the interaction potential,
since this
would make
t,
as well as the average potential energies of all
the other nucleons, as negative as possible and, therefore, lead to the
and the most stable arrangement. But there
which tends to oppose such a collapse. It is the increase in
the kinetic energy of the nucleons demanded by the uncertainty principle
if the volume in which they are confined decreases.
For a nucleus of
radius R, the uncertainty Ap in any component of the linear momentum
of the /th nucleon must be at least
largest total binding energy
is
an
effect
APi ~ h/R
The magnitude of the
linear
momentum
Pi
vector
~ &Pi
is
THE NUCLEUS
676
and the uncertainty
because
arises
average kinetic energy of the
its
direction
nucleon
fth
unknown.
is
[Ch. 16
Thus
the
is
Ti-pZjlMKiAptfKllR 2
But
3
p oc ljK
so|
From
we
(16-104) and (16-105)
binding energy of that nucleon
+ P^
oc
(16-105)
find that the p dependence of the total
is
AE = -E = -(T + V ) = -ap +
i
where
j8
is
the negative total energy of the ith nucleon,
are constants.
The
total binding
A = 2
pP
and where a and
energy of the nucleus
A,
= -yp V3 +
is
dp
(16-106)
=i
where y and <5 are other constants. If this equation is valid, the nuclear
radius R would collapse to the range a of the potential because it would
lead to a large p and maximize A.E.% In these circumstances, all the
nucleons of the nucleus are within the range of each other's interaction
potential, and the average potential energy of the /th nucleon is proportional to the total number of nucleons in the nucleus. That is,
JAcc
As
the
same
is
true for
energy of the nucleus
-A
any other nucleon, the
total average potential
is
V=
i
K <* ~ A
2
=
(16-107)
But, as the average kinetic energy of any of the nucleons
depend on A, the
total average kinetic energy
T= 2
With a collapsed nucleus,
true for large A. Thus
T, oc
V is much
would not
is
+A
(16-108)
larger than T,
A.E= (T + K)~ -Foe
and
+A
this is particularly
(16-109)
t Note that equations (12-38) and (12-39) show that the same result is obtained for a
Fermi gas in which the exclusion principle is explicitly taken into account.
t The second term of equation (16-106) must be larger than the first term, as A
must be positive. Therefore A increases with increasing p since the second term varies
faster
than the
first
term.
NUCLEAR FORCES
Sec. 15]
We
677
see that the total binding energy
is proportional to A 2 despite the
of the uncertainty principle, under the assumption that there is the
same purely attractive interaction potential between all pairs of nucleons
,
effect
in a nucleus.
is
the
a.
But this
same radius R =
not true, and neither is it true that all nuclei have
Therefore the assumption is false. It must be
that the interaction potentials are different between pairs
of nucleons in
different quantum states, being repulsive in some states,
or it must be that
there is some kind of impenetrable repulsive core in the interaction
poten-
tials,
or else
it
must be that there
is
some combination of both
these effects.
One of these alternatives is absolutely necessary in order to keep the
nucleons far enough apart that at any instant one nucleon interacts only
with a limited number of other nucleons. Then the binding energy of
each
nucleon will be independent of the total number of nucleons in the nucleus
and the total binding energy of the nucleus will be proportional to A,
as observed. This requirement is described by saying that
the nuclear
forces
must lead
to saturation.
Most of the information about nuclear
forces that can be obtained from
complex as a saturated nucleus {A
4) is contained
>
the study of systems as
in the four items listed above.
More
detailed information
is obtained by
studying simpler systems where the effects of the nuclear forces can be
analyzed without the complications that characterize a saturated nucleus.
The
simplest of these
is
the deuterium nucleus
or deuteron, in its
This system consists of a neutron and a proton bound
together by the interaction potential for the neutron-proton force.
Its
ground
state.
observed properties are
1H2
A =
2.014732
2.22
amu
Mev
*'=!
(16-110)
= +0.8574^
=
Q +2.74 x 10"27 cm
p iM
The deuteron has a small quadrupole moment Q, which means
its
probability density
immediately shows
that
not completely spherically symmetrical. This
that the neutron-proton interaction potential is
is
complicated because the ground state would be a spherically symmetrical
S statet if the interaction potential could be written V(r), where r is the
magnitude of the vector from one nucleon to the other. [We have seen
that this
is
true in the case of the
atom and
its
Coulomb
potential
The symbols, and the quantum numbers they represent, are taken over directly
from the theory of the electronic structure of atoms, except that i is often used for; as
t
in the shell
model.
THE NUCLEUS
678
V(r) oc
V(r)
by
(10-27)
not
It is
\\r.
difficult to
prove
it
for
any
[Ch. 16
attractive potential
showing that the lowest eigenvalue of the Schroedinger equation
is
always found for
0.]
Thus
not simply of the form V(r).
the existence of a quadrupole moment is
potential
is
V(x, y, z)
V(r)
V'{r) {3(S a
the neutron-proton interaction
form which
r)(S 2
r)/r
is
consistent with
Sx S2 }
(16-1
1)
The
first term is the dominant spherically symmetrical potential.
second term gives rise to small departures from spherical symmetry because
its value depends on the angles between the vector r and the nucleon spin
The
angular
momentum
vectors Sj
and S 2
the quantity
so that the average of the entire second term over
S x S 2 is
subtracted
all
directions of r
vanishes.!
conventional to call the second term the tensor potential.
It is
nomenclature
just like the
is
first
somewhat unfortunate
term.
as
it is
This
not a tensor but a scalar,
Neither term changes sign in the parity operation,
so both are scalars and
V(x, y, z)
V(x, y,
z).
spherically symmetrical term mixes a state of higher
The small nonvalue in with the
dominant S state to form the ground state of the deuteron. But, since
V(x, y, z) = V(x, y, z) and the ground state is not degenerate, the
eigenfunction for the system must have a definite parity (cf. footnote on
page 465), and the state which is mixed in must have the same parity as
the S state. As the parities of these states are even if / is even and odd if
/ is
The
odd
(cf.
state
equation 13-112), this state can be a
is
the only one of these which
is
state,
state, etc.
possible because for
all
the
1
Thus the
too large to lead to the spin quantum number i
ground state is assumed to be a mixture of S and D states. In order to
3
3
Calculations of probobtain i
1, these must be the S1 and
1 states.
others
/ is
and corresponding values of quadrupole moments, show
that the observed quadrupole moment is obtained with the mixture
ability densities,
96% 3 S1
4%W
(16-112)
where these percentages give the probability of finding the deuteron
in
either of the states.
confirmed by comparison with the observed magnetic moment
3
If the ground state of the deuteron were a pure S1 state, then
This
Hi
is
/"<,
(H-i)v
(f*i)n
(16-113)
t This term has the same directional dependence as the magnetic interaction potential
for two magnetic moments. However, it must not be thought of as due to magnetic
interactions between the magnetic moments associated with the spins, because the
strength of those interactions is several orders of magnitude smaller than what
required to explain the quadrupole moment.
is
NUCLEAR FORCES
Sec. 15]
679
components of intrinsic magnetic
from their spins. There
would be no orbital contribution from the motion of the charged proton
since / = 0, and the z components of the intrinsic magnetic moments
would add since the two nucleons are in the triplet state with their spins
"parallel."
Evaluating (fa) v and ((*() in equation (16-113) gives
H iz = +2.7896/^ - 1.9103^ = +0.8793^, which is 2.6 percent
greater than what is observed.! The general agreement confirms the
where
(/i t
moments of
and (fa) n are the
the proton and neutron that arise
Figure 16-52.
An
attractive square well potential.
conclusion that the ground state
is
almost a pure
S1
state
and the
dis-
crepancy confirms the conclusion that it does have a bit of some other
state mixed in. When the magnetic moment is calculated for the mixture
(16-113), agreement with the observed value
better than the accuracy
is
of the observation. The fact that the neutron-proton interaction potential
is
not completely spherically symmetrical
is
interesting but, to a first
can be ignored at low energies because the asymmetry
is small. In most of our discussion we shall ignore it and, therefore, take
the ground state of the deuteron to be a pure 3 51 state.
approximation,
it
Consider, then, the spherically symmetrical interaction potential V(r),
and assume that
it is
in the
as in equation (16-114)
form of a square well of radius
and figure (16-52).
-V
V{r)
0,
We
would
like to establish
it
will turn
<R
(16-114)
AE =
VQ
and the measured
As might be
we shall not be able to evaluate separately
single known quantity Ais, and we shall certainly
2.22 Mev.
out that
both R and V from the
not be able to evaluate the shape of the potential well from
t
The
signs give the relative orientations of the magnetic
momentum
vectors.
depth
r>R
a relation between R,
deuteron ground state binding energy
guessed,
R and
moment and
this single
spin angular
THE NUCLEUS
680
Therefore any shape
quantity.
we might
is
[Ch. 16
equally appropriate and, to simplify the
The time independent
Schroedinger equation that we must solve for this problem is (10-27),
calculation,
with
since
we
as well take the square well.
are considering
r dr\
where
/x is
the reduced
dr
mass
11
with
M the nucleon mass.
an S
state.
It is
which
is
(15-76) for
make
convenient to
^h [ -
(16-115)
u(r),
K(r(r)
Equation (16-116)
0.
the substitution
rR{r)
This leads to the differential equation for
dr
is,
MM _M
M + M~ 2
u{r)
two nucleons. That
for the system of
_
~
It is
we
the one dimensional square well equation
=
is
(16-116)
identical in
solved in Chapter
form with
8, and we
could adapt that solution to our present purposes by redefining the terms.
But
it is
just as easy to solve
^42 +
it
Write
again.
Kfu
it
<R
0,
as
(16-117)
dr
where
K\
j2/i(V
AE)/H
(16-117')
and
K\u
0,
dr 2
>R
(16-118)
where
K\\
The general
y/2/tAEIh
solution to equation (16-117)
(16-118')
is
r<R
= A sin K\r + B cos K\r,
(16-119)
Since the eigenfunction R(r) never diverges, equation (16-115) shows that
at r
0.
This requires setting
u
The general
B=
= A sin K\r,
solution to equation (16-118)
in (16-119).
= Ce~ K " r + De K " r
<R
Then
(16-120)
is
r>R
(16-121)
NUCLEAR FORCES
Sec. 15]
To
681
>
prevent this from diverging as r
we must
co,
D=
set
0.
Then
equation (16-121) becomes
Ce~ K " r
r>R
(16-122)
Now we must equate u and dujdr for both (16-120) and (16-122) at r =
R.
This gives
KR =
sin
Ce~ K " R
(16-123)
and
AK\
cos
K\R
= -CK
n e-
K" B
(16-124)
Dividing equation (16-124) by (16-123) gives
or
K, cot
~ A) cot
is
essentially (8-82),
method used
= -X
NWo ~ A)
This
K^R
and
it
(16-125)
VgMg
can be solved by the same graphical
The
solution only establishes a relation
between those values of V and R that will lead to the observed value of
AE. However, we shall soon see that there is evidence from scattering
experiments that can be used to evaluate R separately, so that V can be
AE. This being the case, it is a good
determined. It is found that V
for that equation.
>
approximation to neglect AZs in the term (V
AE) and
write equation
(16-125') as
AE
h
h
h
or
cot
The
first
N v
root (corresponding to the ground state)
is
J 2t*K R = Z
h
so
V R2 ~
R=
2
TT h*l8/j,
2
tt
/4M
(16-126)
x 10^ 13 cm, a value which is indicated by the scattering
experiments, we obtain V ~ 25 Mev and confirm the consistency of our
If
we
take
approximation.
From
this
information concerning the
potential that
fits
V R2
value for the interaction
the ground state of the deuteron,
it is
not
difficult to
THE NUCLEUS
682
[Ch. 16
The calculation predicts that
and this is verified by experiment.!
the assumed square well potential V(r) with its
calculate the locations of the excited states.
there are
no bound
excited states,
Figure (16-53) indicates
bound state, and also indicates the associated eigenfunction. There
no bound excited states of the neutron-proton system because the
interaction potential is only barely strong enough to bind the ground state.
For the same reason, the ground state eigenfunction has a long exponential
tail extending into the region r > R, and there is a probability higher than
50 percent that the separation between the two nucleons of a deuteron is
single
are
greater than the range of their interaction potential.
Figure
16-53.
The assumed
potential and the single
bound eigenfunction for the
deuteron.
<
We turn now from the bound state' (E 0) of the neutron-proton
system to the unbound states (E
0) of that system. They are investigated
experimentally by measuring the cross section for scattering neutrons
>
from protons, using techniques similar to those used for scattering nucleons
from nuclei, and are analyzed theoretically by the procedures of Chapter
15. As we saw in equations (15-73) and (15-74) of that chapter, the
Furthermore,
scattering cross section is specified by the phase shifts d
equation (15-78) showed us that the <5j will have appreciable values only
for values of / through / = /max where
.
max
~ KR
J2/J.E
(16-127)
-13
cm, and
2 x 10
Taking the radius of the interaction potential as R
10 Mev.
evaluating the total energy E which gives /max = 1, we find E
Thus, at energies small compared to this value, only d should have an
has an infinite
t Compare this with the H atom. This simplest of atomic systems
number of bound excited states whose properties provide almost all the information
that was needed in developing the theory of atomic systems. One of the reasons why
the development of the theory of nuclear systems
little
information because
it
is difficult is
has no bound excited states.
that the deuteron provides
NUCLEAR FORCES
Sec. 15]
683
appreciable value, and isotropic S-wave scattering should be observed.!
As isotropic angular distributions actually are observed up to energies of
about 5 Mev,
it is
certainly safe in treating scattering in this energy range
and (15-81) that apply to S-wave
to use the equations (15-80)
The
latter
scattering.
equation reads
= Lpo
K
(16 _ 128)
where
K=
(16-128')
yJljIEjh
assuming the square well potential (16-1 14), we
Still
may immediately take
over any useful results from the calculation for the identical potential
(15-82). However, we shall not just write down the expression (15-91)
for
(5
because
it
in terms of
gives d
terms of the known quantity A.E.
An approximate expression of
and we prefer an expression in
this type is easy to
value
is
if
we
realize
not going to
total energy
goes from the small negative
the small positive values
we
are considering here since, for
change very much as the
AE to
obtain
<R
that the form of the eigenfunction in the region r
<
V , and the form of the eigenfunction in the region
these values, ||
Thus we use
of the well is determined almost entirely by its depth V
all
equation (16-120) inside the potential well. That
u
~ A sin K\r,
is,
<R
(16-129)
Outside the potential well we use equation (15-88), which
= F sin (Kr +
u and du\dr at r = R gives
u
Matching
sin
Kr~
{
sin
),
(KR
is
r> R
(16-130)
(16-131)
<5)
and
AKi
cos K\r
Dividing (16-132) by (16-131),
X, cot
~ FK cos (KR +
(16-132)
we obtain
K R ~ K cot (KR +
{
But equation (16-125) shows that the
where
K =
n
<5
left side
c5
(16-133)
of (16-133) equals K^,
j2ftbEIH
(16-134)
Therefore
K cot (KR +
t It is
-K||
and that all energies are measured in that system.
E and the LAB energy is E = JElab-
(16-135)
CM system,
The relation between the CM energy
to be understood that in this section all statements refer to the
THE NUCLEUS
684
we
Since
<5
are going to use the S-wave scattering equations that are valid
KR <
only for
in (16-135)
1,
and write the equation
we may
little
work with trigonometric
(16-136)
drop
as well
KR in comparison with
as
K cot
A
[Ch. 16
~ -X
<5
(16-136)
identities will convince the reader that
identical with
is
Evaluating (16-128) from (16-137), and using (16-128') and (16-134), we
obtain an expression for the low energy neutron-proton scattering cross
section,
4t7
K*
4tt
Ki\
2fxElh
2ttH
2/iAEIh
E + A
(16-138)
E and the known deuteron binding energy
Mev. This expression was first obtained by Wigner in 1933.
It is in fairly good agreement with experiment for E from about 1 Mev up
to 5 Mev. However, the expression considerably underestimates the observed values of a when E is small compared to 1 Mev. For example,
(16-138) predicts that a approaches the constant value a
2nh 2 lfj, AE
24
2
2.4 x lfT cm = 2.4 bn, for E
O.j But the experimental value of
a for E
1 ev is a = 20.4 bn.
Although (16-138) was obtained from
in terms of the scattering energy
AE =
2.22
certain approximations, calculations in
which these approximations are not
can hardly be off by as much as a factor of 2. Thus the
factor of 10 discrepancy between the predicted and measured values of a
shows that something is definitely wrong with the theory.
used show that
Wigner suggested that the point
In 1935
wrong
is
it
at
which the theory might be
the assumption that the interaction potential used to
fit
the
deuteron binding energy can also be used to explain the scattering of a
free neutron by a proton. In the deuteron, the neutron and proton must
always interact in the 3 S 1 state because the spin quantum number is i = 1
free neutron and proton can interact in the 1 S state as well as
in the 3 St state. In fact, for an unpolarized beam of neutrons incident
whereas a
upon an unpolarized
target of free protons, J of the scatterings will take
state where the spins are "antiparallel," and f of
the scatterings will take place in Jhe triplet 3 S1 state where the spins are
place in the singlet
"parallel."
The
to 3 ratio occurs because there are
triplet states (cf. section 4,
t
The barn
10- 24
cm 2
(bn)
is
Chapter
12).
a unit of cross section
singlet state
and
If the target contains free protons,
commonly used
in nuclear physics
bn
NUCLEAR FORCES
Sec. 15]
685
all the others, and the observed
an average of the singlet scattering cross
section a s and the triplet scattering cross section a t weighted according
to the relative probabilities J and f That is,
each scattering
be incoherent with
will
scattering cross section a will be
The
The
= to + to
(16-139)
what was
triplet scattering cross section is
really evaluated in (16-138).
have the same general form as
have a different value if the neutron-
singlet scattering cross section will
(16-138), but the quantity AE will
proton interaction potential in the singlet state is different from what
in the triplet state. Allowing for this possibility, Wigner wrote
4(
fX
+ A
+
4(
s)
+ AE
it is
(16-140)
t ).
AE = 2.22 Mev. By equating this to the observed value a = 20.4
E ~ ev, AE may be determined. The result is AE ~ 0.06 Mev.
this is very different from A
it is apparent that the interaction
where
bn
As
at
potential
is
different in the singlet
and
triplet states.
corresponds to the spins "antiparallel" while the
the spins "parallel,"
we
This
is
corresponds to
see that the interaction potential of a neutron
and proton depends on the
nucleons.
Since the former
latter
relative orientation of the spins
of the two
described by saying that the interaction potential
is
spin
dependent.
The experimental value of a
because, for
at
E~
ev does not determine the sign of
<< |A|, the expression
~ 2rrh
ft
E+
|A|
can be shown to be as valid if A < 0, as (16-138) is if A > 0. Thephase
shift changes sign if AE changes sign, but the scattering cross section
(16-128) is independent of the sign of the phase shift. Thus the singlet
state could be either a stable bound state with binding energy AES > 0,
or it could be an unstable virtual state with binding energy AES <
(cf.
section 6, Chapter 15). The observation that there are no stable systems
containing two neutrons or two protons, for both of which the exclusion
principle
demands
that the lowest energy state be the singlet state
suggests that the singlet state of the system containing a neutron
proton
is
also unstable, or virtual.
The
and a
fact that this state actually is
proved in measurements of the scattering of low energy neutrons
from a target of molecular hydrogen at very low temperatures. The
virtual
is
THE NUCLEUS
686
[Ch. 16
hydrogen molecule can either have the two protons "parallel" (orthohydrogen) or "anti-parallel" (para-hydrogen). The latter type has the
lowest energy state, and so at very low temperatures essentially all the
molecules will eventually be converted to that type by collisions and other
processes. For neutrons of de Broglie wavelength comparable to the
internuclear spacing of the molecule, the total wave scattered from the
molecule will be a coherent combination of the waves scattered from
20.0
10.0
5.0
2.0-
1.0
0.5
,-5
10
10"
10"
10
_/!
10"
10"
10
CM (Mev)
Figure 16-54.
Experimental cross sections and theoretical
neutrons from free protons
in
fits
for the scattering of
the S-wave energy range.
two protons. Since in one of these scattering processes the neutron
and proton are in the singlet state while in the other they are in the triplet
the
state, the total scattering cross section is sensitive to the relative signs
the singlet and triplet phase shifts.
An
analysis
by
of
Teller (1937) of the
measurements, which were made at his suggestion, shows that these
phase shifts are of the opposite sign and, therefore, that the singlet state
of the neutron-proton system is virtual.
The points of figure (16-54) shows some measured values of the cross
section for scattering neutrons from free protons in the S-wave energy
To within their accuracy, these data can be fitted perfectly by an
equation basically the same as (16-140), but including corrections for the
approximations we made in its derivation. This is indicated by the solid
range.
These corrections make the exact equation more dependent on the
characteristics of the neutron-proton interaction potential than (16-140)
and thus allow the radius and depth of both the triplet and singlet interaction
line.
NUCLEAR FORCES
Sec. 15]
potentials to be determined
687
by
fitting
the data.f
(It is
also necessary to
bring in the data obtained in the scattering of neutrons from low temperature molecular hydrogen.)
The values so determined
R
VQ
Rs
Vn
'
=
=
=
=
2.05
x 10"13 cm
33.9
Mev
2.9
11.7
Note that the interaction potential
the singlet potential
are
(16-141)
x 10- 13 cm
Mev
certainly is spin dependent,
and that
considerably weaker than the triplet potential for
is
the neutron-proton interaction.
Let us turn
now
particles with s
to the proton-proton interaction.
\,
For these
identical
the exclusion principle plays an important part in
ruling out certain states because
it
demands that the complete eigenfunction
for the system of two particles be antisymmetric with respect to an exchange
of their labels. Thus the spin part of the eigenfunction must be antisymmetric if the spatial part is symmetric, and vice versa (cf. section 4,
Chapter
12).
The symmetry of
realized that, for this system of
the
same
is
is
easily evaluated if
it is
exchange of their labels
particles,
CM of the two particles.
at the
the eigenfunction
is
two
is
as changing the signs of the coordinates because the origin of
coordinates
parity
the spatial part
is
symmetric
odd, and the parity
if its
is
parity
even
if
is
is
Therefore the spatial part of
even and antisymmetric
even and odd
if
/ is
if its
odd
(cf.
states correspond to symmetric
equation 13-112). Thus the S, D, G,
antisymmetric
triplet spin parts. The opposite
require
the
spatial parts and
possible
states.
The
states in which two protons
true
for
the
P,F,
H,
is
.
states only.
can interact are, then, the 1 5', a P, 1 D S F, 1 G, 3 H,
In scattering experiments at energies somewhat less than 10 Mev,
KR is small enough that the two protons interact only in the 1 S state,
which is the 1 S state. The exclusion of the 35x state simplifies the analysis
of the proton-proton scattering experiments, and this certainly helps
because the analysis is complicated by the existence of a Coulomb as well
as a nuclear interaction between the two protons. Because of the Coulomb
interaction, the overall angular distribution is not isotropic in the
.
CM
system, even though
S-wave scattering as far as the nuclear interaction
is concerned. Figure (16-55) shows the differential scattering cross sections
for proton-proton scattering measured at several different low energies,
and also shows theoretical fits to these measurements. Consideration of
figure (15-2) shows that do/dCl must be symmetric about 90 because, if
it is
t For example, the radius R appears separately in the term which was dropped between
equations (16^135) and (16-136). Cf. exercise 20 at the end of this chapter.
THE NUCLEUS
688
one proton
[Ch. 16
must be scattered at
and it is sometimes used as an
experimental check. At angles smaller than about 30, dajdD. rises very
rapidly. In this region it is following the rapid angular dependence of
Coulomb interaction scattering (cf. equation 15-94) which dominates
the nuclear interaction scattering. Near 90, the S-wave nuclear interaction
180
6.
is
scattered at the angle
This
is
6,
the other one
verified experimentally,
0.18N\eM
0.17
0-16
0.15
0.14
0.13
0.12
EcM =
2.101 Mev
0.11
0.10
20
30
50
40
60
70
80
90
"cm
Experimental differential cross sections and theoretical fits for protonin the S-wave energy range.
From G. Breit and R. L. Gluckstern,
Annual Review of Nuclear Science, Vol. 2, Annual Reviews, Stanford, 1953.
Figure 16-55.
proton scattering
scattering dominates
and dajdQ.
is
region in which the nuclear and
tively.
almost isotropic. In between, there
Coulomb
is
interactions interfere destruc-
This interference allows a determination of the sign as well as the
magnitude of the nuclear interaction phase shift. It is found that, for
energies up to about 5 Mev, the phase shift can be fit with a square well
proton-proton interaction potential with radius and depth exactly the same
as R s and V of (16-141), the radius and depth that are found for the
1
S state neutron-proton interaction potential. Thus at low energies the
neutron-proton and proton-proton interaction potentials appear to be the
same in the 1 S state, which is the only state in which a comparison can be
made. Also recall that mirror nuclei indicate that the neutron-neutron
NUCLEAR FORCES
Sec. 15]
and proton-proton interaction
689
same
potentials are the
in the
low energy
range characterizing the relative energies of two nucleons in a nucleus.
From all of this it is concluded that at low energies the interaction potentials
between
all
pairs of nucleons are the same.
however, that
this
It
mean
does not necessarily
should be remembered,
that the shapes of the inter-
action potentials are the same, but only that the phase shifts they produce
are the same.
None
15
of the low energy measurements can give detailed
E CM = 45
Mev
L
/
10
7^
-X
"^x
*-jf
x^^^
.X""
<
b
C
20
40
60
80 100 120 140 160 180
Figure 16-56.
Experimental differential cross sections and theoretical fits for neutronproton scattering at a high energy. From R. Christian and E. W. Hart, Phys. Rev., 77>
441 (1950).
information about the shapes of the interaction potentials since, when
KR 1, the shapes are not important in determining the scattering cross
<
sections.
more about the interaction potentials it is necessary
measure and analyze nucleon-nucleon scattering at high energies,
where KR > 1. According to statements made in the previous chapter,
it would be expected that such measurements should be able to give
information about the shapes of the interaction potentials. Since, for
KR > 1, partial waves with several / values will be involved, the measurements should also be able to tell whether or not the interaction potentials
have an explicit dependence on /.
Figure (16-56) shows the neutron-proton differential scattering cross
section measured at E = 45 Mev, and a theoretical fit to these data.
At other energies from about 15 to 150 Mev, dajdQ. has much the same
In order to learn
to
THE NUCLEUS
690
[Ch. 16
appearance except that
its magnitude has an approximately l/E behavior.
form of this da/dQ indicates that there is an / dependence in the neutron-proton interaction potential, as can be seen from the
following argument. If there were no / dependence the potential could be
written as V(r), and its high energy differential scattering cross section
could be evaluated, at least qualitatively, by the Born approximation.
But, in section 4, Chapter 15, we saw that for KR^> 1 it seems to be a
general characteristic of all potentials V{r) that dajdQ is a highly asymmetric
Even the
qualitative
function of
6,
much
with
larger values for
0~O
than for d
180.
symmetry about 90 that the dajdQ of
figure (16-56) evidently possesses, and it shows that the interaction
potential cannot be written as V(r). Serber (1948) has proposed an /
dependent interaction potential which is capable of fitting the experimental
data for the neutron-proton interaction, at least up to energies of about
This Serber potential can be written
1 50 Mev.
This
is
in strong contrast to the
F (r) =
L"
+C-
1 )']
F(r)
Since
V (r)
t
vanishes for
all
odd
/,
it
(16-142)
gives zero phase shift for all these
which has symmetry about 90,
because only even / terms will remain in the general phase shift expression
(15-73) for dajdQ, and because all the Legendre polynomials P^cosd)
for even / have symmetry about 90 (cf. 15-53). Note that the Serber
potential helps the nuclear forces lead to saturation by removing the
values.
This, in turn, gives a da/dD,
nucleon-nucleon interaction potential in half of the possible quantum
states. However, it does not do enough. For nuclear forces actually to
lead to saturation,
it
is
necessary that there actually be repulsions in
certain circumstances.
The study of high energy proton-proton
the repulsions arise.
scattering indicates
where
Figure (16-57) shows dajdQ. for proton-proton
measured at E = 170 Mev. In the energy range from about 50
Mev, dajdQ. has almost the same form a long flat region where
10~ 3 bn per unit solid angle, and an abrupt rise at small
dajdQ.
angles to follow the Coulomb interaction dajdQ that dominates at these
angles. When these measurements were first performed, it was thought
that, because dajdQ is isotropic in the angular region where nuclear
interaction scattering dominates, this was somehow a case of pure S-wave
scattering even though KR > 1. However, the maximum possible value
of dajdQ for S-wave scattering of two protons isf
scattering
to 200
~4x
dajdQ
(2//Q
2
1. The factor of 2
t Except for the factor of 2, this is equation (15-80) for sin <5
because there are two identical particles in proton-proton scattering.
arises
NUCLEAR FORCES
Sec. 15]
At
E=
170
Mev
this gives the
unit solid angle, which
that several partial
together in such a
691
is less
upper
limit da/dD.
than what
is
x 10"3 bn per
Thus it must be
and happen to mix
waves contribute to the scattering
as to yield an isotropic da/dD,.
way
E CM =
Figure 16-57.
~ 2.5
observed.
30
The measured
60
170 Mev
90
differential cross section for
proton-proton scattering at
a high energy.
Jastrow (1951) has shown that this implies the existence of a strong
repulsive core in the interaction potential,
form of the type indicated
led to this conclusion
is
Figure 16-58.
i.e.,
in figure (16-58).
that the potential has a
The
detailed analysis
which
complicated, but the essential points can be seen
potential with a strong repulsive core.
from a simple argument. Consider proton-proton scattering at E = 170
Mev, and assume that, as for neutron-proton scattering, the interaction is
described by the Serber potential. Since
at this energy, only
S-, P-, and D-wave phase shifts can be appreciable.
But the P-wave
KR~2
phase
shift is
zero for the Serber potential, so the differential scattering
THE NUCLEUS
692
cross section can be obtained
zero except S and
=\
X
=
K
dil
As
2 .|
{e'
{e
=
K
dO.
<5
iS
id
(sin
l
sin d
sin d
5e
5e~
id
25 sin
10 sin S sin
{sin
l
from (15-73) by
(cos 8)
id
sin
(5
sin
^(cos
P2 (cos
P2 (cos
(5
phase
setting all
shifts
this gives
1,
[Ch. 16
0)}
0)}
0)
d^ "'^ + e-^-'^PJcos
5
25 sin
2
t5
2 Pl(cos
0)}
0)
(16-143)
P (cos 0)}
+ 20 sin sin cos
Now P (cos 0) = (3 cos - l)/2, so P2(cos 0) = P (cos 180) = and
P (cos 90) = |. Thus the second term of (16-143) is larger at 0 and
<5
<3
<5
(<5
2)
1,
180 than at 90 and, if both 5 and <5 2 are positive, the third term will
be positive at 0 and 180 and negative at 90. This will make da/dQ. much
smaller at 90 than at 0 and 180,
experimentally.
opposite signs.
to
the low energy data, both d
fit
Chapter
<5
will
15).
and
certainly not isotropic as observed
To achieve isotropy it is necessary for d and 6 2 to have
Now, for a purely attractive potential of the strength used
and
<5
will
be positive
(cf.
section
5,
But, for a potential of the form indicated in figure (16-58),
become negative
at energies high
compared
to the depth of the
attractive region since the effect of the attractive region
becomes
negligible
and only the repulsive core, which produces a negative phase shift, is
important. The repulsive core is, however, not important in determining
l
the phase shift <5 2 until the energy becomes extremely high because the r
dependence of the eigenfunctions, for
r - 0,
makes
the
2 partial wave
smaller in the region of the repulsive core than in the attractive region
1 70 Mev, and similar energies, d 2 will be positive. By fitting the high
at E
energy proton-proton scattering data, it is found that the radius of the
-13
cm. The attractive region can then be
repulsive core is about 0.5 x 10
adjusted to give agreement with the low energy data. If its strength is
made somewhat greater than that of the simple attractive square well
by the parameters (16-141), the correct positive value
can be achieved at low energies because the positive phase shift
produced by the attractive region becomes larger than the negative phase
shift produced by the repulsive core at these energies.
potential described
for S
As
is
1
states are possible. Thereproton-proton scattering, only the 1 S and
and one phase shift for 1 = 2.
only one phase shift for / =
this is
fore there
NUCLEAR FORCES
Sec. 15]
693
Recent work of a number of investigators shows that by using a repulsive
core in a spin dependent potential, which
in the sense that
is
similar to the Serber potential
weak for odd
/, an interaction potential can be found
which gives a good fit to both neutron-proton and proton-proton data at
all energies up to more than 100 Mev, providing the tensor potential and a
spin-orbit potential are included. The spin-orbit potential is proportional
to S L and is attractive if this quantity is positive and repulsive if it is
negative.f One reason for its inclusion is to help fit experiments which
find that the spin angular momentum vectors of two nucleons are partially
polarized by the process of scattering from each other. We avoid quoting
it is
the parameters for the very complicated interaction potential that fits
all the data, because these parameters have not been determined uniquely.
In fact,
felt by many that it will not be possible to find a unique
and that it is better to express what is known about nuclear
it is
potential,
forces in terms of sets of phase shifts describing the nucleon-nucleon
interactions in various quantum states at various energies. There are
even ambiguities concerning the values of these phase shifts.
However, there is, at present, little doubt concerning the existence of
repulsive cores (at least in the singlet states,
which are the ones most
important in the proton-proton interaction). For one thing, repulsive
cores obviously produce nuclear forces that lead to saturation, although it
was
would not lead to saturation at the right
~0.5 x 10-13 cm of the repulsive cores is conmean center-to-center spacing ~ 1.2 x 10~ 13 cm
originally thought that they
density because the radius
siderably smaller than the
of the nucleons in a saturated nucleus. Recent calculations indicate,
though, that repulsive cores should actually produce nuclear forces that
lead to saturation at densities and binding energies close to those observed
in nuclei.
These calculations are based on a theory of nuclear structure due to
Brueckner (ca. 1953) and others. The theory starts from the nucleonnucleon interaction potential described in the next to last paragraph,
and then goes through a
self-consistent procedure similar to that of the
Hartree theory of the atom. However, for the nucleus the procedure is
very much more difficult than for the atom, because there is no single
dominant interaction analogous to the electron-nucleus interaction of the
atom, and because the nuclear forces which must be treated in the theory
are so complicated. Although the theory is still in its infancy, some results
have been obtained, and they are quite encouraging. In addition to finding
nuclear densities and binding energies which are close to what is observed,
t
Note that the
probably related.
two nucleons is similar to the spinand a nucleus (cf. section 8). The two effects are
spin-orbit interaction between
orbit interaction between a nucleon
THE NUCLEUS
694
[Ch. 16
the real and imaginary parts of
good values are found for V and
Furthermore,
the theory gives some feeling
the optical model potential.
instance, the self-consistent
nucleus.
For
actually
happening
in
a
for what is
procedure indicates that the nucleons interact primarily two at a time, and
not three or more at a time. This is the justification for assuming that the
nucleon-nucleon potential, obtained from analysis of experiments involving
the interaction of two free nucleons, applies to the interactions between
fairly
the nucleons of a nucleus.
It is
action potential which are the
structure are:
the
(a),
also found that the features of the inter-
most important
there are repulsive cores;
in determining nuclear
(b), the attractive
regions of
S state interaction potentials are just barely strong enough to overcome
the repulsive cores and give a net binding for the neutron-proton system;
and
Of
(c),
the entire
state interaction potentials vanish or are very
course, the exclusion principle
is
also extremely important.
weak.t
Despite
the uncertainties in the present state of knowledge concerning nuclear
been made in the theory of nuclear structure because
which are apparently the most important
are fortunately those which are the least uncertain.
forces, progress has
the features of nuclear forces
16.
Mesons
In the preceding section
we presented
a description of certain features
of nuclear forces that are observed in experiment. Although theory was
used in the description, it was used only to correlate the experimental
observations, and not to explain them. There is, however, a theory that
attempts to explain how the properties of nuclear forces are based on
more fundamental
attributes of nature.
originated with the
work of Yukawa
This
is
the
meson
theory,
which
in 1935.
Let us imagine two nucleons in interaction.
Yukawa proposed
that
there is associated with this system a "field" which describes the interaction,
and that the actual mechanism of the interaction involves the exchange
of a "quantum" of the field between the two nucleons (i.e., in interacting,
one nucleon emits a "quantum" and the other subsequently absorbs it).
This is pictured schematically in figure (16-59). In such an interaction
the momentum carried by the "quantum" is transferred from one nucleon
to the other, and this is the origin of the force acting between the
nucleons.
In
making
of the
this proposal,
Coulomb
Yukawa was
guided by analogy with the case
two charged particles.
interaction between a system of
t States with / values higher than those of the P state are not very important at the
energies characterizing the motion of nucleons in nuclei.
MESONS
Sec. 16]
695
According to the well-developed and highly successful theory of quantum
electrodynamics (cf. footnote on page 462), associated with this system
there is an electromagnetic field which describes the interaction, and the
interaction actually arises from the exchange of a quantum of that field
between the two charged particles. The quanta of the electromagnetic
field are the zero rest mass entities which throughout this book we have
called simply quanta.
Quantum
interaction
electrodynamics shows that the long range of the
is
in the interaction,
Figure 16-59.
that quanta
Coulomb
a result of the zero rest mass of the quanta that are exchanged
had
and that the range would not be long
if it
had happened
schematic representation of a meson exchange process.
finite rest
Yukawa adapted the theory to the case of
mass.
a system of nucleons, interacting with a short range nuclear force, by
assuming that the "quanta" of the "field" describing the interaction have
mass. These "quanta" are called mesons, and the "field" is
called the meson field. We shall not attempt to go beyond the first steps of a
quantitative development of the meson theory because that would rapidly
carry us into some difficult mathematics. However, there are certain
features of the theory which are not at all difficult to describe.
finite rest
For
instance,
it is
easy to estimate the
lead to the observed range
meson
rest
mass m^ required to
Consider the process
represented in figure (16-59). In this process one nucleon emits a meson
of rest mass w, the meson travels a distance of the order of R, and finally
the meson is absorbed by the other nucleon. While the meson is in flight,
of nuclear forces.
the conservation of total energy
is violated because the total energy of the
system equals two nucleon rest mass energies in the initial and final
state, and equals two nucleon rest mass energies plus at least one meson
mass energy in the intermediate state. But the uncertainty principle
shows that violation of energy conservation by an amount
rest
AE ~
is
not impossible
if it
mc*"
does not happen for a time longer than At, where
At A
~h
THE NUCLEUS
696
[Ch. 16
because such a violation would be immeasurable. The velocity of the
meson can be no greater than c, so its flight time At is at least
At
~ Rjc
These three equations give
m^c 2
hc/R
/j/At
m,
~ hjRc
'
or
Setting
R~2 X
10~ 13 cm, this predicts
tron rest mass. It
is
worth while to
Figure 16-60.
way.
meson of
rest
mass
)T
(16-144)
m^
~ 200m, where m
restate the
The Yukawa
argument
is
the elec-
in the following
potential.
~ hjRc leads to an interaction of range
by a distance much larger than
R, there could be no interaction as its flight time would be so long that
the uncertainty principle would allow an accurate enough measurement
r~"R because,
if
the nucleons are separated
of total energy to detect the violation of energy conservation.
A detailed treatment, using relativistic quantum mechanics, leads to
a detailed prediction of the interaction potential arising from
exchange. This
is
the
meson
Yukawa potential,
V(r)
-rlR
(16-145)
where
R = hjm w c
which
is
plotted in figure (16-60).
leading to this potential.
not
It is
relativistic
(16-145')
difficult to
wave equation
obtained by writing (1-25),
2 =
y+m
?c
where
Vl
p!
follow the steps
for the
mesons
is
Sec. 16]
MESONS
making the
substitutions (7-77),
697
Px^ in
dx
Py~^
d
*
in
Pz~+
d
in
dy
t,
dz
->
in
d
dt
and then multiplying the
resulting operator equation into the equality
y>
rp
This gives
-**?? = -c 2 n
v* w
or
+ my v
or
V-^ =
c
which
is
m\ c
-$r<P
dPtp __
J_
06-146)
~dr~~~li
called the Klein-Gordon equation.
For m^
it
reduces to the
wave equation
T72
V*W
^
= 12 <P"fc
for the quanta of the electromagnetic
the
This has a
field.
static solution
of
form
= e /r,
2
f
as can easily be verified
by
ip(r).
For m^
>
substitution, using the relation
for
3f
dr \
dr'
the Klein-Gordon equation has a static solution
of the form
y>
= - q- e~ rlB
>
where
-R
h/m n c
as can also easily be verified by substitution.
wave equation
for zero rest
Since the solution to the
mass quanta gives the Coulomb interaction
THE NUCLEUS
698
[Ch. 16
potential for the electromagnetic field, the solution for non-zero rest
assumed to be the interaction potential for the meson
quanta
is
that
the
is,
Yukawa
field,
potential (16-145).
The constant q2 of (16-145) determines the
Although
mass
this potential is
strength of the potential.
obviously not capable of explaining the existence
of the short range repulsive core in the nucleon-nucleon interaction,
Or~-
Intermediate
Initial
Figure 16-61.
it
Final
The exchange
of a
n~ meson.
can explain the attractive region outside the core. Fitting it to this region
2
In terms of the dimensionless constant q 2 jhc, it is
fixes the value of q
.
\hc~
q
In the original form of the
mesons carry the
electric
meson
charge
(16-147)
15
theory,
or +e.
it was assumed
The theory then
that all
predicts
that a typical interaction process between a proton and a neutron
lead to the situation pictured schematically in figure (16-61).
would
In this
p represents a proton, and -n~ represents a
meson of charge e. We see that, as a result of the interaction, the
neutron is changed into a proton and the proton is changed into a neutron.
figure n represents a neutron,
7T~
_3i~o
^0
Final
Initial
Figure 16-62.
small angle scattering.
This might seem like an unreasonable prediction, but there is very direct
experimental evidence that it really can happen. Consider the high
energy neutron-proton differential scattering cross section curve of figure
(16-56). We have interpreted its symmetry about 90 as evidence for the
Serber
potential
(equation
16-142)
would lead primarily
(16-62). However, we could
because
an ordinary interaction
potential V(r)
to scattering at 8
figure
also interpret the
as evidence for the conclusion that there
is
0, as indicated in
symmetry about 90
a probability of \ that in the
MESONS
Sec. 16]
699
scattering the neutron is changed into a proton, and vice versa, by the
exchange of a tt~~ meson emitted from the neutron, or by the exchange of a
7T+ meson of charge +e emitted from the proton. Figure (16-63) indicates
such a scattering for a case in which a n~ meson is exchanged. An inter-
~<Z)
>
Intermediate
Initial
A small
Figure 16-63.
action potential, that
Final
angle scattering with the exchange of a it- meson.
would lead to
scatterings in
which there was a
probability of \ that such processes occur, can be written formally as
[1+*]
where
V(r)
(16-148)
an exchange operator that exchanges the labels of the two
This potential always comes into formulas for the cross section
is
particles.
through the matrix element
-
y>*
so
it is
V(r)ip
of interest to consider the quantity
[1
f]
V(r)y> t
iV(r)V
+ iVWVi
(16-149)
where we have specified the eigenfunction by its quantum number /. In the
paragraph following (16-141), we found that exchanging the labels of the
particles is the same as changing the signs of the coordinates. In this
operation,
yi
odd, because
will
its
be unchanged if / is even, and will change sign if / is
is even in the first case and odd in the second.
parity
Therefore
Py> t
ip t ,
p Vi = Yi>
even
lodd
or
FVi=(-l)'v.
THE NUCLEUS
700
[Ch. 16
and equation (16-149) can be written
U v(r)fi =
mr)xpi
i F(r)(
_ 1)V; =
[1
+^- 1 >'J V (r) Wl
We see, then, that (16-148) is completely equivalent to the Serber potential
(16-142),
and
that the
meson theory provides
a qualitative explanation of
the origin of that potential.
The high energy neutron-proton
differential scattering cross section
shows, however, that the original assumption that
cannot be correct.
If
it
(16-63) would take place with a probability of
all
mesons are charged
were, processes of the type indicated in figure
Initial
1,
instead of with the
Intermediate
Figure 16-64.
Final
The exchange
of a
w meson.
probability of \ that
is required to fit the data. Thus there must also be
uncharged mesons which allow interactions of the type pictured schematically in figure (16-64), where 77
represents an uncharged n meson.
Furthermore, the v mesons are necessary in order to allow nuclear
interactions between two nucleons of the same charge.f
Meson theory also provides a qualitative explanation of how the
neutron can have an intrinsic magnetic moment even though its net
charge is zero. According to the theory, a neutron can spontaneously emit
_
a 7T meson and become a proton, if the proton reabsorbs the meson within
a time At
h/m^c 2 and becomes a neutron again. As a result of such
processes, there are internal currents associated with the neutron,
and
moment. The theory
predicts that protons also spontaneously emit and reabsorb mesons. Thus
meson theory pictures nucleons as being surrounded by a "cloud" of
mesons, which have a transient existence and are bound to the nucleon
"core" by the requirements of energy conservation.
At the time of Yukawa's proposal, there were no known particles of
rest mass between the electron rest mass m and the proton rest mass
1836am. The n + and it~ mesons were first observed by Powell and collaborators in 1947 as a component of the cosmic radiation. Shortly after,
these currents produce the intrinsic neutron magnetic
t At the time of Yukawa's original proposal
only between neutrons and protons.
it
was thought that nuclear
forces acted
MESONS
Sec. 16]
701
mesons were produced artificially at a large particle accelerator in
between high energy (~300 Mev) nucleons and nucleons in a
target. Presumably the cosmic radiation mesons are also initially produced
in high energy collisions. Measurements show that the tt+ and ir mesons
these
collisions
have the same
rest
mass,
m,
This
is
274m
(16-150)
The
enough to Yukawa's prediction (16-144).
certainly close
mesons were first observed by Moyer and collaborators in 1950 as
products of high energy collisions. Their rest mass is found to be
77
m^o
264m
(16-151)
free n mesons, which are observed in these
from the clouds surrounding the colliding
According to the theory, the
experiments, are liberated
nucleons by the energy made available in the collision.
that these are the
same
as the
mesons discussed
It is
in the
assumed, then,
meson
theory.
Of
course, their discovery constituted a striking verification of the qualitative
validity of the theory.
Unfortunately, there has been
little
opportunity to
test the quantitative
validity of the theory. The problem is that the usual calculational techniques cannot be applied to meson theory because the dimensionless
constant (16-147) is so large. This constant, q*\hc <= 15, plays a part in
meson theory which
constant, a
e2 jhc
is
analogous to that played by the fine structure
1/137, in
quantum
electrodynamics.
It is
possible
very accurate calculations in the latter theory by the use of perturbation techniques, because it turns out that the perturbations associated
with electromagnetic interactions are always proportional to powers of
to
make
etjhc
and are therefore very small.t Perturbation techniques cannot be
used in meson theory because the perturbations are proportional to powers
of q 2 lhc and are therefore very large. Quantitative calculations in meson
theory await the development of more powerful techniques.
Mesons are unstable. The n+ and it mesons decay spontaneously in
free space according to the
scheme
77*
-^/X*
+V
(16-152)
ith lifetime
tt =2.54
x 10" 8
sec
(16-153)
The symbol /j,+ represents a /*+ meson of charge +e; (x represents a
pr meson of charge e; and v represents a neutrino. The a mesons
perturbation (11-51)
t As a particular example, note that the spin-orbit
(e^/ic) 2 .
tional to a 2
is
propor-
702
THE NUCLEUS
[Ch. 16
decay according to the scheme
sy +
N
y>
*%
+e-+y,
(16 _ 154)
\,
with lifetime
V>~l0 _15 sec
(16-155)
represents a gamma ray; e+ represents a positron; and
er represents an electron. The fi+ and fjr mesons also decay. Their decay
The symbol y
scheme
is
M ->e
+v +
and
(16-156)
their lifetime is
x 10" 6
2.22
sec
(16-157)
There are no uncharged
mesons. All values quoted for lifetimes are
values in frames of reference in which the mesons are at rest. In a frame
in which a meson is moving, its lifetime appears to be lengthened because
fj,
of the relativistic time dilation effect described by equation (1-7). Measurements of the observed values of t and T , made in both rest frames
and moving frames, give direct experimental verification of time dilation.
The fi mesons have no part in Yukawa's meson theory of nuclear forces,
although this was not appreciated until some time after their discovery in
1936 by Anderson and Neddermeyer. These investigators found the
particles as components of the cosmic radiation and showed that their
rest mass is intermediate between the rest mass m of an electron and the
rest mass 1836m of a proton. We now know that they are probably
produced in the cosmic radiation from the decay of it mesons. But, in
1936, it mesons had not been discovered, and it was naturally assumed
that the fi mesons were Yukawa's mesons. An ever increasing accumulation of evidence showed, however, that the interaction of
mesons with
matter is very weak. For instance, the fi mesons of the cosmic radiation
can penetrate great thicknesses of solid matter with little attenuation since
they can be detected in deep mines. This being the case,
mesons can
hardly be the particles responsible for the strong interactions of nuclear
forces, despite the fact that their rest mass
ft
/j,
/j,
m
is
quite close to the value predicted
207m
(16-158)
by Yukawa.
This situation was the source of considerable confusion in the ten years
before the discovery of the tt mesons. But, after their discovery, it was
immediately assumed that the
tt
mesons are Yukawa's mesons
early evidence indicated that their interaction with matter
was confirmed by
later
is
since the
strong. This
measurements which showed that the cross section
Ch.
BIBLIOGRAPHY
16]
for interaction of a
it
703
meson with a nucleus
close to
is
its
maximum
possible value, the projected geometrical cross sectional area itR*
equation 15-103). The interaction
meson
interacts with a nucleus,
splitting the nucleus into several
able energy.
It is
now known
is
also particularly violent;
most of
that
mass energy goes into
rest
its
fragments which
n mesons
(cf.
when a n
fly
apart with consider-
are closely associated with
nucleons and interact with the strong nuclear interaction, while fi mesons
are closely associated with electrons and interact with the very weak beta
decay interaction. One of the pieces of evidence for the association between
mesons and electrons is that they both have spin quantum number
i, antisymmetric symmetry character, and magnetic moments corresponding to the spin g factor g s = 2. The n mesons have i = 0, symmetric symmetry character, and no magnetic moments.
Within the past few years, more than a dozen other unstable particles
have been discovered. These particles fall into two classes the K mesons
with rest masses around 1000m, and the hyperons with rest masses around
fi
i
2500/w.
At
present, not too
much
is
known about
However, some of them seem to have strong
contribute to nuclear forces.
interactions,
their properties.
and so probably
In fact, the relation (16-144) makes
it
tempting to associate those that interact strongly with the short range
repulsive core of the nucleon-nucleon potential, because the relation shows
that the range of the force is inversely proportional to the
mass of the meson.
probably also true that part of the effects
leading to the core arise from the simultaneous exchange of several tt
mesons. We shall not even attempt to list the
mesons and hyperons that
are presently known, since such a list would undoubtedly soon be obselete.
For up to date information concerning this rapidly developing field, and
the important part which it surely will play in the fundamentals of modern
This
may
be true, but
it is
physics, the reader
must consult current research
literature
and review
articles.
BIBLIOGRAPHY
Bethe, H. A., and P. Morrison, Elementary Nuclear Theory, John Wiley and
Sons, New York, 1956.
M., and V. F. Weisskopf, Theoretical Nuclear Physics, John Wiley and
York, 1952.
Elton, L. R. B., Introductory Nuclear Theory, Isaac Pitman and Sons, London,
Blatt, J.
Sons,
New
1959.
Evans, R. D., The Atomic Nucleus, McGraw-Hill Book Co., New York, 1955.
Halliday, D., Introductory Nuclear Physics, John Wiley and Sons, New York,
1955.
THE NUCLEUS
704
[Ch. 16
EXERCISES
1.
Show
that the
minimum
from a thin hydrogenous
2. Evaluate, in
Coulomb
energy quantum that will eject a 4.5-Mev proton
is E c 50 Mev, as stated in section 2.
y
layer
Mev, the energy of gravitational attraction, and the energy of
two protons separated by a distance of 5 x 10- 13 cm.
repulsion, for
3. Prove equation (16-3). Hint: Use Gauss' law, which
Chapter 13.
4. Verify
is
stated in section 2,
equations (16-11), (16-11'), and (16-11"). Also show that (16-11')
reduces to (16-12).
5.
Use the appropriate
classical expressions to derive (16-20).
Use (16-20) to calculate the energy in the LAB frame of the protons emitted
in the forward direction, and at 30 from the forward direction, in the reaction
2
He 4 + 7 N 14 8 17 +
induced by 30-Mev alpha particles. For this reaction
Q = -1.18 Mev.
6.
Using the data of table (16-1), evaluate, in Mev, the energy released in
two
nuclei to form a 2 He 4 nucleus. Also evaluate, in Mev,
the height of the Coulomb repulsion barrier which must be overcome before there
is a large probability that the two nuclei can get close enough together for fusion
to take place, assuming that the radius of 1 H 2 is 2 x 10-13 cm.
8. Evaluate the terms of the semi-empirical mass formula for the 92 U M8
7.
the fusion of
Use the results to calculate the predicted values of total binding energy
and average binding energy per nucleon AE/A, and compare with table
(16-1). Calculate the fraction which each of the terms /i through
5 contributes
to AE.
nucleus.
AE
9. Look up the ground state spin quantum numbers of all the odd A
and compare with the predictions of the shell model and figure (16-23).
nuclei,
Assuming that 71 Lu173 is an ellipsoid of revolution, find the lengths of
the semiminor and semimajor axes that lead to the value of quadrupole moment
given in figure (16-24), and also to a volume consistent with normal nuclear
10.
densities.
1
is
1.
Show
V, 1 H
1
,
H2
or
He3
but that
it is
9 2rj 238
unstable to the
2
He 4 Explain why this is so. Hint: Take the masses of the emitted
from table (16-1) because the semi-empirical mass formula is not
emission of
particles
mass formula correctly predicts that
that the semi-empirical
stable to the emission of
.
accurate for very small A.
An
unstable nucleus whose probability per second of decay is A is being
artificial means in a particle accelerator at a constant rate of /
nuclei per second. If the production process commences at / = 0, calculate
the number of these nuclei which will be present at f = f by setting up and
12.
produced by
solving an equation similar to one of the equations (16-46).
13.
Use the semi-empirical mass formula to estimate how the data of
(16-25) would extrapolate to
A =
190, 180, 170, 160.
figure
Ch.
705
EXERCISES
16]
Use the semi-empirical mass formula
14.
similar to (16-36) for
A =
77.
to evaluate points
on a
figure
From it predict the energies available for electron
and positron decay. Then look up the data and compare.
15. Use the shell model to predict the spins and parities of the ground states
of several sets of odd A nuclei in the region N = 50 to N = 82. Then predict
which of the beta transitions between these ground states will satisfy the Fermi
selection rules (16-74), and which will satisfy the Gamow-Teller selection rules
(16-77).
Evaluate the attenuation length A for neutrinos passing through matter
2
gm/cm 3 using the typical interaction cross section a = 10" cm
16.
of density 10
predicted by (16-80).
Taking the predictions of the
17.
shell
model and
figure (16-23) for the spin
of the ground and first excited states of odd A
nuclei in the regions of the magic numbers, where the model should be valid,
use the selection rules (16-83) and (16-83') to predict the character (L value and
quantum numbers and
whether
electric
parities
or magnetic) of the
gamma
decay transitions from the first
Also determine the predictions
nuclei in the regions between magic
excited state to the ground state in these nuclei.
of the collective model for even N, even Z
numbers, where the model should be valid, by consulting section
13.
29
Cu63 and 25-Mev
18. Compare the value of the a{p,pri) cross section for
protons, quoted in figure (16-49), with the projected geometrical area for a
1
13
2
Discuss the implications of this
nucleus of radius R = 1.25 -* x 10~ cm
.
comparison in terms of the optical model. Also calculate the probability that
a 25-Mev proton would make such a reaction in traversing a foil 0.001 inch
29
thick and containing the normal mixture of Cu isotopes.
19.
R =
Solve equation (16-125') graphically, but without approximation, for
2 x 10~ 13 cm. Compare with the predictions of (16-126).
20. Derive the low energy neutron-proton scattering cross section formula
without making the approximation of using the bound state eigenfunction for
r < R in (16-129), and without dropping the KR in (16-135). Show that this
13
R) instead of (16-138). Taking/? = 2 x lO"
gives a = [4t7/(A: 2 + Ap(l +
cm, evaluate the error in (16-138) at
21.
section
is
dajdQ. in the
E=
ev.
ev the isotropic neutron-proton scattering cross
frame, use equation (15-17) to evaluate
20.4 bn in the
Given the
fact that at
=
LAB
CM
frame. Plot the results.
22. A tt meson comes to rest in the LAB frame and then decays into a n meson
and a neutrino according to the scheme (16-152). Show that the kinetic energy
of the n meson in the LAB frame is (m^ m^^llm^, where w and m^ are
the -n and ji meson rest masses, respectively. Evaluate this kinetic energy, in
Mev, from the rest mass values quoted in (16-150) and (16-158).
OT
Ind ex
Absorption, of electromagnetic radiation, 453-7, 505-12
Einstein's coefficient for,
Alpha decay,
459
and magic numbers, 609
and semi-empirical mass formula,
609
rate,
physical picture of, 461-2
transition rate for,
457
of nucleons by nuclei, 577-8
with several optical electrons, 425-8
of particles, 555
Alpha
discovery and nature
89-91
emission through barrier, 236-9,
610-2
Alpha particle scattering, 91-2
and atomic charge, 106-7
and discovery of nuclei, 99-100
and nuclear radii, 107-8
by Rutherford atom, 100-6
by Thomson atom, 92-8
Ampere's law, 339
Angular momentum, 313-4
by absorbing sphere, 556-7, 573
in complex potential, 557-8
in Schroedinger theory, 554-8
Absorption spectra, 113, 121
Actinide elements, 141, 411
Age of earth, 615
Alkali atoms, 418-25
418-24
421-4
quantum numbers for, 421
selection rules for, 424-5
spectra of, 112, 424-5
fine
structure
of,
in,
addition of, in JJ coupling, 442
in LS coupling, 430-2
Alkali elements, 413
in Paschen-Bach
grand total, 450
operators, 314-5
Allowed transitions, 634
Alpha decay, 89-90, 236-9, 609-16
barrier penetration
probability,
238,
610-2
energy, 609-10
447
momentum
in,
structure
641
in,
effect,
orbital, 323; see also Orbital angular
336-9, 449; see also Spin angu-
spin,
equilibrium decay
fine
particles,
of,
Absorptivity, 47
energy levels
potential, 237, 611
radioactive series and, 613-6, 620-1
614-5
lar
total,
Geiger-Nuttall law for, 612-3
momentum
346-9, 429-30, 442
total orbital, 429; see also Orbital
lifetime, 613
angular
long range components in, 641
motion of alpha particle before, 612
nuclear radii and, 239, 612
707
momentum
total spin, 428; see also Spin
momentum
Annihilation, 516-7
angular
708
INDEX
Anomalous Zeeman
effect,
447
Atomic theory,
Antiparticle, 517
Antiproton, 517
Antisymmetric eigenfunctions, 363-73
and exchange force, 373
and exclusion principle, 366-7
for interacting identical particles,
368
for non-interacting identical particles,
363-4, 367
370
369
triplet, 370
Associated Laguerre functions, 302,
304-5
Associated Legendre functions, 301,
304-5
Atomic bombs, 619
Atomic energy states, 124-8
Atomic form factor, 500
Atomic mass formula, 590-2
Atomic mass unit, 582
energy equivalent of, 584
Atomic model, of Rutherford, see
Rutherford's model of atoms
of Thomson, see Thomson's model of
atoms
Atomic number (Z), 87, 560
Atomic radii, in Bohr theory, 116, 133
in Hartree theory, 403
macroscopic estimate, 78-9
in Thomas-Fermi theory, 396
Atomic reactor, 619
Atomic spectra, absorption, 113, 121
of alkali atom, 112, 424-5
singlet,
and
spin,
emission, 113
structure of, see Fine structure
of hydrogen, see One-electron atom
hyperfine splitting of, 357-8, 449
of JJ coupling atom, 443
fine
lines
of
111
of,
LS coupling atom, 440-1
measurement
of,
110-1
of one-electron atom, 111, 119, 425
optical,
417
series of,
112
series limit of, 112
spark,
120-1
Zeeman, 447
of
Bohr,
see
Bohr's
theory of atoms
of Hartree, see Hartree's theory of
atoms
of
Schroedinger,
see
One-electron
atom
of
Thomas and Fermi,
see ThomasFermi theory of atoms
Atomic weight (A), 78, 560
Atoms, charge of, 87-8, 106-7, 488-9
charge density of, 395-6, 400-1
energy levels of, see Energy levels,
of atoms
excitation of, 119, 416-8
mass of, 78, 585
stability of,
108-9, 114-5, 152, 187,
414, 467
Attenuation, exponential law for, 511-2
of nucleons, 577-8
of X-rays, 493-512
Attenuation coefficient, 517
Attenuation length, for nucleons, 577-8
for X-rays, 512
Avogadro's number, 78
determination of, 481
Axial vectors, 637
Balmer series, 113, 120
Barn (bn), 684
Barrier penetration, and alpha particle
emission, 236-9, 610-2
for barrier potential, 233-9
by electromagnetic radiation, 235-6
for step potential, 226-7
and uncertainty principle, 239
and particle-wave duality, 235, 239
Barrier penetration probability, for bar-
233-4
610-2
in three dimensions, 610-1
Barrier potentials, 231-9
rier potential,
for nuclear potential,
^alpha particle emission and, 236-9
boundary conditions
classical
behavior
232
231
for,
of,
eigenfunctions for, 232
energy levels
Ramsauer
for,
232
234-5
effect and,
transmission coefficients for, 233-4
transmission resonances for, 234-5
709
INDEX
Bohr's theory of atoms, radii of orbits,
Bessel functions, spherical, 538
compared
Beta decay, 89-90, 620-40
coupling constant, 628, 635, 640
Coulomb
effects in,
308
Sommerfeld's
631
density of final states for, 628-30
selection rules for,
for,
in,
in,
relation to
636
638-40
638
neutrinos and, 626
Born's interpretation of wave functions,
172-5, 294-5
Bose probability distribution, 389-90
Boundary conditions, for eigenfunctions,
412-3
585-6
of one-electron atom, 300-2
590
and fission, 587
and fusion, 587-8
and saturation of nuclear
675-7
Black body radiation, 47-9
Planck theory of, 63-6
of simple harmonic oscillator po-
forces,
E>
compared
to
quantum mechanics, 313
of, compared to quantum mechanics, 303
and magnetic moments, 327-9
energy levels
postulates of, 114-5, 123, 128-30
postulates of, de Broglie's interpreta-
151-2
predictions of, 115-24
tion,
<
V, 223-4
K, 228
K,
periodic,
and spontaneous emission, 458, 460
Bohr magneton, 329
Bohr's microscope, 157-8
Bohr's theory of atoms, 1 10-37
of,
E>
of square well potential,
Black body spectrum, 48-9
momentum
258-9, 261
tential,
of step potential,
Rayleigh-Jeans theory of, 51-7, 62-3
angular
178, 300
of barrier potential, 232
373,
of, 586,
Boltzmann's constant, 61
Bombarding particle, 563
Born approximation, 524-30
applicability of, 534, 554
for Coulomb potential, 554
for Gaussian potential, 530-2
for relativistic electrons,
Binding energies, of electrons in atoms,
constancy
probability dis-
388-90
for nuclear direct interactions,
636-8
pseudo-scalar matrix element for,
636-8
shell model and, 634-5
spectra, 624
Beta particles, 89-90, 599
parity non-conservation in,
of nuclei,
quantum
tributions,
672-3
569
for square well potential, 532-3
origin of, 621
128,
458-60
derivation of, 57-61
matrix element, 628, 633-8
neutrino recoil
distribution,
applications of, 121,
630
neutrino induced reactions
modification
61
in,
632-3
lifetime,
relativistic
Boltzmann probability
624-5
Fermi selection rules for, 633-4
Fermi theory of, 627-33
Ft values for, 633-6
Kurie plots
quantum mechan-
131-4, 355-6
of,
energy, 621-4
energy conservation
Gamow-Teller
to
ics,
E< K
241
250
.set'
Periodic boundary condi-
tions
for Thomas-Fermi
Box normalization, 220
theory,
394-5
applications of, 252, 270, 525-6, 629
Brackett series,
113,
120
Bragg's law, 479
Breit-Wigner formula, 670
Bremsstrahlen, 489; see also X-ray con-
tinuum spectra
Broadening of spectral
lines,
see Line
widths
Brueckner theory of nuclear structure,
693-4
Bucherer's experiment, 37, 75-6
710
INDEX
Cascades in gamma decay, 642
Cathode rays, 71
Cathode ray tube, 71
Compound
nucleus reactions, independence of formation and decay in,
667
Center of mass, 122
random width assumption
separation of time independent
Schroedinger equation for,
295-8
Center of mass
(CM
frame, 39, 519
relation to laboratory frame, 519-22
Centrifugal potential, 404-5
)
Chain reactions, 618-9
Charge, of atoms, 87-8,
106-7,
406,
488-9
of electrons, 74
measurement
of,
73-5
of nuclei, 106, 560
Charge
density, of atoms, 339-400
of nuclei, 570-2
Charge independence of nuclear forces,
675, 688-9
Classical limit of quantum theory, 206-9
and correspondence principle, 135-6
and identical particles, 364-5
and orbital angular momentum, 324
and simple harmonic oscillators, 68,
264-5
and spin angular momentum, 337-8
and square well potentials, 248-50
and time dependent perturbation
theory, 289-90
Classical physics,
Cloud chamber, 513
Collective model of nuclei, 605-8
and electric quadrupole moments,
606-8
and low lying energy levels, 658-9
and magnetic dipole moments, 606
and liquid drop model, 605
and shell model, 605
Collision broadening of spectral lines,
470
Compound
667
of,
uncertainty principle and, 663
nucleus, 655
and resonances, 655-7
Compound
nucleus reactions, 662-71
angular and energy distributions of,
670-1
Breit-Wigner formula for, 670
excitation and decay of, 664-7
experimental evidence for, 668-70
Complex
Complex
Compton
conjugate, 172
potentials, 557-8,
scattering, 81-5,
572-80
502-5
cross section for, 503-5
from protons, 564
quantum theory of, 82-5, 503
Compton wavelength, 85
Conduction electrons, 382, 386-7
Configuration, 409
Conservation, of charge, 584, 644
of energy, in beta decay, 624-5
and limitations of uncertainty principle, 289, 667, 695-6, 700
in relativity theory, 35-6
and nuclear reactions, 582-5
of probability, 175; see also Normalization, of
wave functions and of
eigenfunctions
and complex potentials, 557-8
Conservation equation, 174-5
Continuity of eigenfunctions and their
derivatives, 178, 321
Continuum of energy
128,
levels,
atomic,
184
nuclear, 662
Contraction of length, see Lorentz contraction
Correspondence
135-6;
principle,
also Classical limit of
see
quantum
theory
Cosmic
radiation, 513
Coulomb
Coulomb
cross section, 554
potential, 295; see also
electron
Coulomb
One-
atom
scattering,
100-6
angular distribution, 106
cross section, 554
Coulomb's law, 42
and quantum electrodynamics, 694-5,
696-8
Coulomb
Coupling
repulsion energy of nuclei, 566
constant,
635, 640
beta
decay,
628,
INDEX
711
Coupling constant, comparison between
magnitudes of, 640
Coulomb field, 701
meson
field,
Degeneracy, and
698, 701
Coupling of angular momenta,
LS, 428-41
orbital, 426-7
of perturbations,
identical
particles,
363
J],
441-3
Sommerfeld theory, 133
in
Degenerate
perturbation
theory,
Perturbation theory
see
degen-
for
restrictions
of exclusion principle,
435-9, 442
spin, 426
erate case
Density of energy levels for free par-
spin and orbital, 428
Density, of atoms, 79
strong
field,
447
ticle,
527-8
of nuclei, 560, 572, 580
Cross product, 313
Cross section, 495, 504-5, 522
absorbing sphere, 557
Born approximation formulas
529-30
Breit-Wigner, 670
CM and LAB, 522-4
Coulomb, 554
differential, 498, 504
Gaussian potential, 531
of final states, in beta decay, 628-30
in
in
for,
nucleon emission, 661
time dependent perturbation
theory, 290
Deuterium, see Deuteron
Deuteron, 677-82
eigenfunction for, 682
electric
quadrupole moment
of,
677-9
excited states of, 682, 685
ground
state of,
678
geometrical interpretation of, 496
hard sphere, 548
potential for,
Klein-Nishina, 503-4
time independent Schroedinger equa-
phase
shift
formulas
for, 544,
556
510
proton-nucleus, 574
proton-proton, 688-91
square well, 532, 551, 684
S-wave, 549
Thomson, 495, 498
X-ray, 509-10
Current loops, 327-8
Daughter nucleus, 609, 613
Davisson-Germer experiment, 147-9
De Broglie's postulate, 139-41
experimental confirmation of, 146-51
to
Schroedinger's theory,
165-6
Differential cross section, 498, 522; see
also Cross section
Differential operators,
for angular
201-5
momentum, 314-5
and associated dynamical quantities,
203
and development of Schroedinger
equations, 293-4
and eigenvalue equations, 318-22
and relation to Schroedinger's theory,
204, 321
and
relativistic
Diffraction,
tion,
of
wave equations, 697
electromagnetic
140,
148,
149-50,
radia-
160-2,
467-80
of electrons, 147-50, 567-8
relation to uncertainty principle, 153
Broglie wavelength, 146
typical values of, 146-7, 150
Degeneracy, 278
in atoms wjth several optical electrons,
680
685
tion for,
virtual state of,
photoelectric, 507,
relation
679, 687
properties of, 677
neutron-proton, 685-6, 689
pair production, 508-9, 510
De
effect
278-9
and exchange of
4^5-6
of helium atoms, 150
of protons, 573-5
of quantum mechanical waves, 533,
548-9, 557
Diffraction
gratings,
467-80
110,
148,
149,
712
INDEX
Dilation of time, see
Time
Eigenfunctions, linear combinations of,
dilation
Dirac theory, of quantum mechanics,
338
and fine structure, 356-7
and pair production, 513-7
of vacuum, 514-6
189-90, 191, 269, 272-3, 276-7,
538
for LS coupling, 432-3
mathematical properties of, 184-92
for multi-electron atoms, 397-8
Direct interactions in nuclear reactions,
for non-interacting identical particles,
662-3, 670-4
angular and energy distributions for,
670-1, 673
normalization of, see Normalization,
363-4, 367-8
of eigenfunctions
Born approximation and, 672-3
excitation and decay of, 672
for one-electron atom, see One-elec-
experimental evidence for, 670-1,
orthogonality of, see Orthogonality of
tron-atom eigenfunctions
672-3
eigenfunctions
for simple harmonic oscillator poten-
Discrete energy levels, 183
finite widths of, 470
Doppler broadening of spectral
469
tial,
lines,
Dot product, 331
for
Doublets, of energy levels, see Fine
square
424
Dynamical quantities, 202
and associated differential operators,
203
fluctuations in, 318-21
lines,
well
potential,
E< V
<
224, 225
243, 247
E> V
structure
of spectral
263-4
370-1
for spin angular momentum, 370
singlet,
250-1
for step potential,
E> V
229
for string, 196
symmetric, 363-4, 368-9
triplet, 370-1
Eigenstates, 187
and emission of radiation by atoms,
461-2
Eigenvalue equations, 318-22
for angular momentum, 323, 538
for energy, 320-1
for linear momentum, 321-2
Eigenvalues, 185, 320; see also Energy
Earth, age of, 615
Effective charge, 401,
Ehrenfest's
theorem,
488
206-9
Eigenf unctions, 185, 320
antisymmetric, 363-4, 367-8
barrier
for
potential,
232
beta decay, 627
in
conditions of acceptability, 178, 300,
levels
Einstein's coefficients,
321
459
Einstein's photoelectric theory,
for deuteron, 682
Einstein's relativity postulates,
finiteness of,
178, 321
for free particle, standing wave,
217
three dimensional, 526, 540
traveling wave, 213
and
gamma
decay transitions, 647
375-8
for helium atom,
for incident and scattered waves, 535
for infinite square well potential, 253,
254
for
interacting
368-9
identical
particles,
79-81
15-6
Einstein's mass-energy relation, 34
of, 585-6
and nuclear binding energy, 585-6
and nuclear reactions, 582-5
Einstein's quantum postulate, 80
and theory of radiating atoms, 455,
462
Einstein's spontaneous emission theory,
458-60
experimental verification
Electric
dipole
460-1
matrix elements, 455,
INDEX
713
Electric
dipole
transitions,
454,
467,
Energy
644
Electric quadrupole moment, 606, 646
and collective model, 606-8
of deuteron, 677-9
of nuclei, 451-2, 607
and shell model, 608
Electric quadrupole transitions, 468,
644
Electromagnetic
levels,
of atoms,
for hydrogen, see One-electron
atom
in JJ coupling,
LS
in
441-2
coupling, 433-4, 439-40
for
multi-electron atoms, 403-5,
408-9, 41 1-2, 415, 425-8, 433-4,
439-40, 441-2
absorption
for one-electron atoms,
453-7, 461-2
classical theory of, 42-6
dual nature of, 85-6, 139, 161-2
energy density of, 43
energy flux of, 45
induced emission of, 457, 461-2
rate of energy emission, 46
spontaneous emission of, 453, 460,
303-4, 308-9, 403-5
radiation,
of,
461-2
Electromagnetic wave equation, 697
Electron, capture in beta decay, 623-4
diffraction,
Zeeman
of
of infinite square well potential, 255
of nuclei, average spacing, 659-60
model, 658
662
density of, 660
excitation of, 641
in Fermi gas model, 595
measurement of, 655-8; see also
Beta decay; gamma decay
in shell model, 601-2, 658-9
widths of, 660-2
in collective
continuum
by nuclei, 567-70
Electron charge, 74
73-5
Electron mass, 74
measurement
of,
73-5
Electron-proton mass ratio, 124
Electron spin, see Spin angular momen-
446-7
particle,
of free electron, 128, 513-7
of free particle, 312
emission in beta decay, 623-4
of,
splitting of,
134,
of Fermi gas, 383
147-50, 567-8
measurement
bound
117,
184
continuum, 128, 184
discrete, 183
discovery of, 70-3
scattering
helium,
for
374, 376
of,
of rigid rotator, 138, 325
of
tum
simple
harmonic
oscillator,
64,
131, 262
Electron volt (ev), 125
of step potential, 222
of square well potential,
Emission spectra, 1 1
Emission theories, 15
<V
244-5
E > V
Emissivity, 46
240, 250
unbound particle, 184
X-ray, 485-6
(
e/m
ratio for electrons, 73
Energy, equipartition
relativistic kinetic,
rest mass,
of,
61
33
Equilibrium decay, 614-5
Equipartition of energy, 61
34
total relativistic, 34
Energy conservation,
tions, 582-5
of
Ether, 7
in
nuclear reac-
in relativity theory, 35-6
and uncertainty principle limits, 289,
667, 695-6, 700
Energy level diagram, 64
Energy levels, of atoms, for alkali
atoms, 418-24
Ether drag, 14
Ether frame, 8
Even-odd effects in nuclei, 589, 591
and Fermi gas model, 596-9
Exchange, of labels for identical particles, 362-3
of mesons, 694-5
and repulsive core, 703
714
INDEX
Exchange, of mesons, and Serber potential,
698-700
Exchange degeneracy, 363
helium atom, 375
Exchange force, 372-3
in helium atom, 377-8, 380
Exchange integral, 380
and nuclear Coulomb energy, 566
Exchange operator, 699
Excitation, of atoms, 119, 416-8
in
of nuclei, 641
Excited
states,
119;
see
also
Energy
Fermi
Fermi
and
and
and
and
gas, exclusion principle and,
depth of nuclear potential, 595-6
596-9
595-6
semi-empirical mass formula,
596-8
and shell model, 595, 599
Fermi probability distribution, 388-90
Fermi selection rules, 633-4
Field, electromagnetic, 694-5
meson, 694-5
even-odd
effects,
optical model,
Fine structure, 131, 334-5, 338-9,
levels
Exclusion principle, 366-8
and angular momentum of holes, 485
and antisymmetric eigenfunctions, 367
and capacity of subshells, 408
and conduction electrons, 386-7
and experimental evidence, 381
and Dirac theory of pair production,
514
and Fermi gas, 384, 386-7, 595
and Fermi probability distribution,
390
and independent particle motion in
nucleus, 595
and JJ coupling, 442
and LS coupling, 435-9
and properties of atoms, 414
and proton-proton scattering, 687
and theory of nuclear structure, 694
Expectation values, 201-5
general formula, 205
order of terms in integrals, 204, 211
relation to perturbation theory, 272
Exponential attenuation, 511-2
Exponential decay, 468-9
of compound nucleus states, 666
355-7
atoms, 421-4
in alkali
in
JJ coupling, 441
in LS coupling, 433-4, 441
Fine structure constant, 134
and quantum electrodynamics, 701
and spin-orbit energy, 352
Finitenesse of eigenfunctions and their
Fermi beta decay theory, 627-33
Fermi energy, 385, 387
of conduction electrons, 386
of nuclei, 596
Fermi gas, 381-9
applications of, 382
conduction electrons and, 386
energy distribution of, 383, 388
energy level density of, 383
of,
383
178, 321
derivatives,
Fission of nuclei, 616-20
chain reactions
in,
energy diagram
618-9
for, 616,
energy emitted
618
587
in,
fragment distribution, 619-20
fragments, 616
induced, 617-8
neutron emission
origin of,
in,
618
616
spontaneous, 617
Fluctuations
in
dynamical
quantities,
318-21
Flux, classical, 174-5
incident, 504-5,
522
probability, 175; see also Probability
flux
scattered, 505,
energy levels
384
gas model of nuclei, 594-9
Forbidden
Form
522
transitions,
factor,
635
500
Fourier analysis, 154, 158, 191, 200
Franck-Hertz experiment, 124-8
Free particle, 212-20
classical behavior of, 212
eigenfunctions, standing wave, 217
three dimensional, 526
traveling wave, 213
energy levels, 128, 213
INDEX
715
Free particle,
normalization
wave
of
Groups of wave
functions for, 215, 219
functions,
142-5,
153-4, 159-60, 163, 215-6
probability density, standing wave,
Group
velocity, 142, 144, 166
218
traveling wave, 213
Halogen elements, 413
probability flux, standing wave, 218
traveling wave, 214
Hamiltonian, 321
time independent Schroedinger equa-
see
tion,
wave
212
in
functions, standing wave, 217
traveling wave, 213
Ft
values,
Hartree's
633-6
of atoms,
theory
also
ments, 400, 501
theory of
Heisenberg's
Gamma
cascades, 642
and magnetic
644-8
energy of
gamma
rays
transitions,
642
650-3
in,
internal conversion and,
isomeric transitions, 643
lifetimes,
652-3
matrix elements, 649-50
measurement of energy, 642-3
measurement of lifetime, 643
rays,
Gamow-Teller
splitting, 357-8, 449
Hyperfine structure, 449-52
and
647-8
89-90, 559-60
selection rules,
Hermite equation, 259
solution of, 259-62
Hermite polynomials, 263
Holes, and pair production, 515-7
properties of, 437, 516
quantum numbers for, 485
selection rules for, 486
subshells with, 437
X-ray spectra and, 484-5
Hydrogen, see One-electron atom
Hyperfine
selection rules, 648
transition rates in shell model,
Gamma
quadrupole interaction,
electric
451-2
and magnetic dipole interaction, 448-
636
Gaussian potential, 531
Gauss' law, 393
Geiger counter, 563-4
51
Hyperons, 703
Identical
Geiger-Nuttall law, 612-13
General theory of relativity, 38
g factor, Lande, 446
meson, 703
nuclear, 450
orbital,
quantum me-
chanics, 164-5, 337
decay, 640-55
electric
329
particles
ability,
and
and beta decay, 627-30
and Born approximation, 527-9
and widths of nuclear energy levels,
661-2
Independent particle motion in nucleus,
593-5
360-1
Induced emission, 457
Indistinguishability,
459
physical picture of, 461-2
transition rate for,
Induced
fission,
Inertial frame,
Infinite
457
617-8
square well potential, 251-6
applications of, 252
boundary conditions
Gravitational forces compared to nuclear forces, 566
eigenfunctions, 253, 254
Ground
energy
119
indistinguish-
360-1
Einstein's coefficient for,
336-7, 357
Golden Rule No. 2, 291
spin,
state,
atoms,
Hartree theory
and nuclear shell model, 600
and theory of nuclear structure, 693
and verification by X-ray measure-
Fusion of nuclei, 587-8
Galilean transformation, 5
and mechanics, 3-9
396-405;
Multi-electron
levels,
255
for,
252-3
716
INDEX
Infinite
square well potential, normalization of eigenfunctions for, 254
parity of eigenfunctions for,
253-4
time independent Schroedinger equation,
253
zero point energy
Interaction
255-6
of,
strengths,
Legendre functions, associated, 301,
304-5
Legendre polynomials, 537
Lifetime of quantum states, 469
and widths of energy levels, 472
Limiting velocity in relativity,
comparison
and range of nuclear
tween, 640
Interference of eigenfunctions and
wave
17,
27,
29, 38, 44
be-
Line, spectral,
forces,
696
111
153-4,
of stability, 618
Linear combinations, see Superposition
Internal conversion, coefficients, 651-2
of eigenfunctions, 189-90, 191, 269,
272-3, 276-7, 538
functions,
151-2,
148,
365
Interferometer,
of wave functions
energetics of, 650-1
gamma
of wave functions, 186, 192, 285
decay and, 650-3
Interval rule of Lande, 434, 441
Linearity of Schroedinger equation,
Ionization energy, 128, 413
166-7, 185-6
Line widths, 469-72
Isomeric transitions, 643
Isotopes, 581
abundance
of, 582,
588-9
hyperfine structure and, 449
JJ coupling, in atoms, 441-3
in nuclei,
603
Kirchhoff's law, 47
Klein-Gordon equation, 697
Klein-Nishina cross sections, 503-5
K mesons,
Kurie
703
630
plots,
Laboratory (LAB) frame, 39, 519
relation to center of mass frame,
519-22
Laguerre functions, associated, 302,
433-4, 441
of orbital angular momenta, 429
fine structure splitting,
304-5
Lamb
shift,
and collision broadening, 470
and Doppler broadening, 469
and natural broadening, 469-70
and uncertainty principle, 469-70,
472
Liquid drop model of nuclei, 589-91
and collective model, 605-8
and fission, 617
and semi-empirical mass formula,
590-1
and shell model, 605
Lorentz contraction, 14, 18, 21-2
Lorentz transformation, 26
derivation of, 24-7
LS coupling, 428-41
energy levels, 433-4, 439-40, 441
and exclusion principle, 435-9
357
quantum numbers, 429-33, 435
Lande g factor, 446
Lande interval rule, in atoms, 434, 441
selection rules,
in nuclei, 600
Lanthanide elements, 411, 414
Laplacian operator, in rectangular co-
of spin angular momenta, 428
of total angular momenta, 429
spectra,
Lyman
440-1
440-1
series,
113, 120
ordinates, 294
in spherical coordinates,
298
Larmor frequency, 332
Larmor precession, 331-2
of net magnetic moment, 444-5
Last nucleon in nucleus, 592-3
Magic numbers, 592-4
and alpha decay energy, 609
and electric quadrupole moments, 608
and fission fragment distribution,
620
INDEX
717
Magnetic dipole moment, of deuteron,
677-9
of electron, 336-7
of filled subshell, 438-9
net, 443
of neutron, 679
and meson theory, 700
in non-uniform magnetic field, 333-4
orbital, of electron, 327-9
of positron, 516
and precessional motions, 331-2,
444-5
of proton, 679
quantization, 329, 336
spin, of electron, 336-9
Magnetic dipole moments of nuclei,
335-6, 357-8, 449
and collective model, 606
and electrons in nuclei, 562
and shell model, 603-4
Magnetic dipole transitions, 468
Magneton, Bohr, 329
nuclear, 450
Mass, of deuteron, 677
of electron, 74
measurement of, 73-5
of n meson, 702
of v meson, 701
of neutrino, 626
of neutron, 585
of nuclei, 585
measurement of, 580-5
of positron, 516
of proton, 585
relativistic,
rest,
Mass
Mass
32
32
deficiency of nuclei, 585-6
density of nuclei, 572
Mass-energy relation, 35-6
and nuclear reactions, 565
Mass spectrometry, 580-2
Mass unit, atomic, 582
energy equivalent of, 584
Matrix element, 273, 286
for beta decay, 628, 633-8
electric dipole, 455, 460-1
for
gamma
decay, 649-50
nuclear, see Beta decay
Maxwell's equations, 7
Mechanics,
classical,
3-4
30-7
Mesons, 694-703
charge of m, 701
relativistic,
charge of *-, 698, 700
decay of /*, 702
decay of w, 701-2
discovery of
/t,
discovery of
tt,
702
700-1
hyperons, 703
interaction of m, 702
interaction of v, 702-3
K mesons,
703
lifetime of
/i,
lifetime of
tt,
702
701-2
and time dilation, 702
magnetic dipole moment of fi, 703
magnetic dipole moment of ir, 703
mass of /i, 702
mass of tt, 696, 701
and range of nuclear forces, 695-6
spin of ,u, 703
spin of tt, 703
lifetime
symmetry character of /i, 703
symmetry character of *, 703
Meson theory, 696-700
and role of perturbation theory, 701
Michelson-Morley experiment, 9-13
Microscopic reversibility, 458
Mirror nuclei, 565
beta decay of, 634-5
binding
energies
and nuclear
radii,
565-6
Models, 88, 559
and theories, 604-5
Modern
physics,
Multi-electron atoms, in Hartree theory,
396-405; see also Alkali atoms;
Atoms, with several optical
elec-
trons; Periodic table of elements
effective
charge
for,
401
eigenfunctions for, 397-8
net potentials for, 397,
probability
density
400-1
for,
397-401,
404-5
quantum numbers
for,
self-consistency in,
shells
in,
402-3
398
397
718
INDEX
Multi-electron atoms, in Hartree theory,
411-
subshells in, 403-5, 408-9,
2,
415
time independent Schroedinger
equation for, 396
in
gas model of nuclei
Thomas-Fermi theory, 393-6
boundary conditions for, 394-5
net potential for, 394-5
probability density for, 394-5, 401
self-consistency in, 393
Multiple scattering, 92-8
angular distribution for, 98
Multiplet, 434
Nal
scintillation detector, 537
Natural broadening of spectral
lines,
469-70
Natural width of spectral
Net
470
lines,
potential, in Hartree theory, 397,
400-1
in multi-electron
in
atoms, 392
Thomas-Fermi theory, 395
Neutrino, beta decay and emission
626-33
meson decay and emission
induced reactions, 638-40
properties of, 626
recoil in beta decay, 638
Neutron, 562
beta decay of, 630-1
discovery of, 562-5
magnetic dipole
moment
of,
of,
of,
701-2
679
and meson theory, 700
mass of, 585
spin angular
momentum
symmetry character
of,
of,
562
562
Newton's laws, 4
for angular
momentum, 331
187-8,
of eigenfunctions,
190
examples, 254, 367-8, 525-6
of probability density, 175
wave functions,
219-20
Nuclear bombs, 619
of
187,
190-1, 215,
Nuclear collective model, see Collective
model of nuclei
Nuclear fission, see Fission of nuclei
Nuclear forces, 674-94
angular momentum dependence of,
690
attractive character of, 556, 674
charge independence of, 675, 688-9
and deuteron, 677-82
and eigenfunction for deuteron, 682
and meson theory, 694-703
and neutron-molecular hydrogen
scattering, 685-6
and neutron-proton scattering, high
energy, 689-90
low energy, 682-7
and nuclear structure, 693-4
and potential for deuteron, 679, 687
and proton-proton scattering, high
energy, 690-3
low energy, 687-8
range of, 579, 674, 687, 692
compared to other forces, 579
and meson theory, 696, 697
repulsive cores in, 691-3
and meson theory, 703
and saturation of, 693-4
saturation of, 675-7, 690, 693
and Serber potential, 690, 691-2,
693, 699-700
spin dependence of, 685, 693
and spin-orbit potential, 693
and stability of nuclei, 566
strength of,
compared
to other
forces, 556, 579-80,
Noble gas elements, 412-3
Normalization,
Nuclear Coulomb repulsion energy,
590-1
Nuclear even-odd effects, 589, 591
and Fermi gas model, 596-9
Nuclear Fermi gas model, see Fermi
640
and tensor potential, 678-9, 693
and virtual state of neutron-proton
system, 685-6
Nuclear fusion, see Fusion of nuclei
Nuclear liquid drop model, see Liquid
drop model of nuclei
Nuclear magic numbers, see Magic
numbers
Nuclear magneton, 450
INDEX
719
Nuclear mass formula, 590-2
and alpha decay energy, 609
and beta decay energy, 621-2
Nuclear matrix element, 628, 633-6
Nuclear model of atoms, see Rutherford's model of atoms
Nuclear optical model, see Optical
model of nuclei
Nuclear potential, 572-80; see also Shell
model of nuclei
composition of total real part, 57980
Nuclear radii, for charge distribution,
565-72
comparison with potential
tion radii, 578-9
electron
of,
scattering
Nuclei, composition of, 561-5
electric
quadrupole moments
Electric quadrupole
energy levels
of nuclei
Energy
see
of,
measurements
567-72
levels,
magnetic dipole moments of, see
Magnetic dipole moments of
nuclei
mass of, 585
mass deficiency of, 585-6
mass density of, 560, 572, 580
number of protons and neutrons
in,
565
spin of, see Spin angular
distribu-
of, see
moments
momentum,
of nuclei
stability of,
587-8, 609, 615-6, 622
of, 561
symmetry character
Nucleons, 565
mirror nuclei binding energy measurements of, 565-6
Numerical
integration
equations,
of differential
180-2, 268, 276-7
values of, 571
for potential distribution,
572-80
alpha particle scattering and decay
measurements, 100, 107-8, 239,
612
comparison with charge distribution radii, 578-9
neutron cross section measurements
of, 572-3
proton scattering measurements of,
573-9
values of, 575-6
nucleus,
see
Compound
nucleus reactions
direct, see Direct interactions in nuclear reactions
history of, 563-5
136-7
angular momenta,
expectation values, 317
precessional motion of, 323-4
quantization of, 323
One-electron-atom
eigenfunctions,
be-
cause, 308-9, 404-5
normalization of, 307
orthogonality of, 307
466
quantum numbers
for,
302-4, 337,
349-50
One-electron-atom energy levels, degeneracy of, 303-4, 350-1
notation for, 563
Q value of, 583
Nuclear reactors, 619
Nuclear shell model, see Shell model
of nuclei
Nuclear structure, theory
of,
693-4
Nuclei, binding energies of, see Binding energies, of nuclei
of,
of,
One-electron-atom
parity of,
energy balance in, 582-5
energy level measurements and, 655-8
charge density
critique
One-electron atom, 295-325, 350-8
and role in theory of multi-electron
atoms, 402-3
havior near origin, 309
degeneracy of, 303-4, 350-1
Nuclear reactions, 662-74
compound
Old quantum theory, 136
570-2
cause, 308-9, 404-5
with fine structure, 355-7
in old quantum theory, 117, 134
in Schroedinger theory, 303
One-electron-atom probability densities,
angular dependence of, 309-13
behavior near origin, 309
measurability of, 311-2
INDEX
720
One-electron-atom probability densities,
radial
relation to
306-9
angular momenta, 312-3
dependence
of,
One-electron-atom time independent
Schroedinger
equation,
com-
295
plete,
464-6
and atomic transition selection rules,
467
and beta decay selection rules, 634,
636
of deuteron ground state, 678
Parity, 247,
of eigenfunctions, for infinite square
for relative motion, 298
295-8
253-4
well potential,
separation of complete equation,
for one-electron atom,
separation of relative equation, 298-
for simple
harmonic
466
oscillator,
263
for spherically symmetrical poten-
9
solution of relative equation,
299-
tial,
466
247-8
for square well potential,
302
Optically active (optical) electron, 418
and form of potential, 248
and gamma decay selection
648, 650
Optical model of nuclei, 572-80
non-conservation in beta decay, 636-
Operators, see Differential operators
Optical excitation of atoms, 417, 484
and Fermi gas model, 594-6
and initial stage of nuclear reaction,
663
and shell model, 594
values of parameters, 575-7
Orbital angular
of
momentum, 323
filled subshell,
438
of nuclear energy levels, 653, 658-9
measurement
decay,
origin of,
quantization of, 114, 129-30, 323
429
in atoms with several optical electrons, 426-7
quantization of, 429
Orbital g factor, 329
Orbital magnetic dipole moment, 327-
Partial
Ordinary integral, 380
magmoment, 330-1
magnetic dipole moment, 445
Orientational potential energy, of
netic dipole
of net
Orthogonality of eigenfunctions, 188-90
in
continuum, 189
when
degenerate, 189-90
decay
253-4
wave
equations,
248,
554-6
556-
analysis, 534-44,
for absorbing sphere scattering,
total,
quantization of, 329
658; see also Beta
465
Schroedinger
atom
precessional motion of, 323-4
of,
Gamma
and perturbation theory, 276.
and solution of time independent
of one-electron atom, see One-electron
rules,
applicability of, 542, 549,
554
cross section formulas for, 544, 556
for hard sphere scattering,
546-9
for neutron-proton scattering, 682-4
691-2
549-52
Partial waves, 537-8, 555
orbital angular momentum of, 538
and plane waves, 540-2
Particle-wave duality, 85-6
and barrier penetration, 235, 239
and uncertainty principle, 160-2,
226-7
Paschen-Bach effect, 447-8
for proton-proton scattering,
for S-wave scattering,
Pauli principle, see Exclusion principle
Periodic boundary conditions, for elecPairing energy of nuclei, 598
Pair production, 508-9, 513-7
cross section for, 508-9, 510
Parent nucleus, 609, 613
tromagnetic radiation, 52
in relativistic range,
for
629
wave functions and eigenfunctions, 220, 525-6
INDEX
721
Periodic table of elements, 406
actinide elements, 411,
Phase
414
elements, 413
halogen elements, 413
Hartree theory interpretation
and cross
alkali
of,
407-
interpretation of,
for
noble gas elements, 412-3
for
rare earth elements, 411, 414
group elements, 413-4
transuranic elements, 414
Perturbation, 269
and effect on degeneracies, 278-9
Perturbation theory, and meson theory,
701
and quantum electrodynamics, 701
transition
Perturbation theory for degenerate case,
278-83
and atoms with several optical electrons, 425-8
classical example of, 283-5
formulas for, 279-81
and helium atom, 373-81
and JJ coupling, 442
and LS coupling, 428-9, 432-4
and relativistic corrections to Schroedinger's equation,
353-4
simplified treatment of, 282-3, 350-1
and spin-orbit energy, 350-3
and Zeeman effect, 445-6
Perturbation theory for non-degenerate
268-73
277-8
formulas for, 272
case,
of,
use in degenerate case, 282-3
and helium atom, 373-6
and V-bottom potential, 274-8
Perturbation theory for time dependent
285-91; see also Golden
Rule No. 2
and absorption of electromagnetic
radiation, 453-7
and Einstein's quantum postulate, 455
formulas
286
and Golden Rule No. 2, 286-91
and induced emission of electromagnetic radiation, 457
Pfund series, 113, 120
for,
556
sections, 544,
553
of,
for hard sphere scattering, 547
lanthanide elements, 411, 414
case,
542, 555-6
energy dependence
14
accuracy
shifts,
for absorbing sphere scattering, 557
544-6
>
KR, 547
Ramsauer effect, 552
/
551-2
for resonance,
for S-wave scattering,
Photoelectric
cell,
551
76
Photoelectric effect, classical theory of,
78-9
cross sections for, 507-8, 510
early experiments on, 76-8, 81
quantum theory
X-ray
of,
line spectra
79-81, 507
and, 507
Photoelectrons, 77
Pilot waves,
140; see also
Wave
func-
tions
Planck's black body spectrum theory,
63-6
Planck's constant, 66
and uncertainty principle, 158
and X-ray continuum spectra, 482
Planck's quantization postulate, 64
comments on, 66-8, 128-31, 262-3
Polarization, of scattered nucleons, 602,
693
of scattered X-rays, 476
Polar vectors, 637
Positronium, 517
Positrons,
513-7
annihilation of, 516-7
Dirac theory of, 514-7
discovery of, 513
emission of, in beta decay, 623-4
holes and, 516
properties of, 516
Potential,
alpha
particle-nucleus,
see
Alpha decay
barrier, see Barrier potentials
constant, see Free particle
Coulomb,
see
Coulomb
atom
potential;
One-electron
deuteron, see Nuclear forces
Gaussian, see Gaussian potential
INDEX
722
Potential,
infinite
square well, see In-
nucleon-nucleon, see Nuclear forces
nucleon-nucleus, see Optical model of
nuclei; Shell
model of nuclei
harmonic
oscillator
square well, see Square well potential
step, see
Step potential
V-bottom, see V-bottom potential
Yukawa,
Potentials,
Yukawa
see
and
potential
general treatment,
coordinate
dependent,
separation of center of mass
of energy in Schroedinger theory,
of grand total angular
of nuclear spin angular
Lar-
294-
for atom emitting radiation, 461-2
for free particle, standing wave, 218
traveling wave, 213
of total orbital angular
and Wilson-Sommerfeld
for multi-electron atom, 397, 401,
404-5
momentum,
rules,
Quantum, of electromagnetic
atom
atom, 401
atom, 306-9
in multi-electron
well
oscillator,
potential,
264
<
248
251
for step potential,
E< V
two non-interacting
ticles, 363-5
128-
field,
80,
694-5
of meson field, 694-5
Quantum defects, 420
Quantum electrodynamics, 462
and origin of Coulomb interaction,
307
harmonic
347,
32
for one-electron atom, see One-elec-
momentum,
429-30
of total spin angular momentum, 429
and uncertainty principle, 157-8
wave function, 186-7
eigenstate wave function, 187
square
moment, 329
429
for general
in one-electron
114,
of simple harmonic oscillator, 64
of total angular
for simple
momentum,
of orbital magnetic
Probability density, 171-2, 172-5,
momentum,
323
Probability conservation, 175
for
momentum,
449
of orbital angular
mor, 331-2, 444-5
and uncertainty principle, 324, 430
E>V
117,
450
general treatment, 176, 294
tron
124-8
of atomic energy states,
178-84
Precession of angular momenta,
for
moment
and
285
radial,
Quadrupole moment, see Electric quad-
mo-
time dependent, general treatment,
for
quantum mechanics, 160
Quantization, 64; see also Energy levels
298
tion,
Probability in
Product particle, 563
Proper length, 22
Proper time, 21
Proton, 560
Pseudo-scalars, 636
and beta decay, 636-8
rupole
298-9
and spatial quantization, 324-5
relative
178
forces,
radial dependent,
for incident
simple harmonic oscillator, see Simple
214
and scattered waves, 536
for step potential, E < V
225
E > V 229
traveling wave,
Net potential
net, see
Probability flux, 175
for free particle, standing wave, 218
square well potential
finite
226
identical par-
694-5, 696-8
and perturbation theory, 701
and spin g factor, 357
and spontaneous emission of radiation, 458, 462
Quantum numbers, 64, 185
assignment of, 448-9
INDEX
723
Quantum numbers,
for atoms, with //
coupling, 442
LS
with
coupling, 429-33, 435
with one electron, 302^1
with one optical electron, 421
with several optical electrons, 425
for nuclear shell model, 600
spectroscopic notation, for
408
486
422-3
for n,
for
/,
s,
and energy balance in nuclear reactions, 565, 582-5
experimental verification
585
of,
30-7
Relativistic mechanics,
momen-
Relativistic transformation, of
tum and energy, 36-7
for X-ray levels, 485
Quantum
mass-energy relation, and
binding energy of atomic electrons, 421, 580
and binding energy of nuclei, 585-6
Relativistic
64
lifetime of, 469,. 472
Quantum statistical mechanics, 390
states,
of position and time, 26
of velocity, 29
Relativistic
wave equation, 697
Relativity theory, general theory, 38
Quantum
experimental verification
special theory,
theory, 41
value of nuclear reaction, 583
of,
37-8
3-38
Relativity theory applications, see Rela-
Radial probability density, 307
in multi-electron atoms, 401
in one-electron atoms, 306-9
tivistic
Compton
mass-energy relation
83
effect,
corrections
to
Born approximation,
Radiation by accelerated charges, 42-6
Radii, of atoms, see Atomic radii
of nuclei, see Nuclear radii
corrections to Schroedinger equation,
Radioactive decay, 89; see also Alpha
decay; Beta decay; Gamma decay
non-radiation of uniformly moving
charge, 43
Radioactive series, 613-6, 620-1
branching in, 621
pair production, 513-7
equilibrium
Ramsauer
Random
569
353-4
Sommerfeld theory of
614-5
in,
234-5, 552, 657
effect,
walk, 97
Range, of charged
563
compared to other forces, 579
and meson theory, 696, 697
Rare earth elements, 411, 414
theory
radiation, 52-7,
of black
and meson theory, 703
and saturation, 693-4
Residual
body
62-3
spectral distribution of, 62
Recursion relation, 260, 268
Reduced mass, 123, 519
in
time independent Schroedinger
equation for
motion, 297
CM
Reflection coefficient, for step potential,
E < V 225
V 229-31
,
33-4
34-5
Relativistic energy,
Relativistic mass,
Relativistic
mass-energy relation, 34-5
Coulomb
interactions,
426-8
Residual nucleus, 563
Resonances, 551-2
many
particle vs. single particle,
657, 664
in nuclear reactions,
in nuclei,
in
655-7
577
square well potentials, 251
for transmission through barrier potentials,
E>
precession, 340-4, 359
Repulsive cores in nuclear forces, 691
of nuclear forces, 579, 674, 687, 692
Rayleigh-Jeans
momentum, 338
spin angular
Thomas
particles,
fine structure,
133
and
234-5
552-3
virtual states,
Rest mass, 32
Rest mass energy, 34
Rigid rotator, energy levels
of, 138,
325
724
INDEX
Russell-Saunders coupling, see
LS
Selection rules, for alkali spectra, 424
for beta spectra, 634, 636
coupling
Rutherford scattering experiments, 912, 98-9, 106-8; see also Cou-
lomb
scattering
Rutherford's model of atoms, 99-100
and Bohr's theory, 114
predictions of, 100-6
stability of,
108-9, 114-5, 152, 187,
461
verification of,
106-7
Rydberg constant, for hydrogen, 112
for infinitely heavy nucleus, 119, 123
for nucleus of mass M, 123
Rydberg series, for alkalis, 112, 420,
424
Saturation of nuclear forces,
675-7,
690, 693-4
in
693-4
Thomas-Fermi theory of atom,
393
and alpha decay energy, 609
and beta decay energy, 621-2
and meson theory, 698-700
of alpha particles by nuclei, 91-108
of electrons, by atoms, 400
by nuclei, 567-70
of neutrons, by molecular hydrogen,
685-6
atomic, 112
Balmer, 113, 120
Brackett, 113, 120
Series,
Lyman,
113, 120
Pfund, 113, 120
by protons, 682-7, 689-90
and nuclear physics, 518
of protons, by nuclei, 572-80
by protons, 687-8, 690-3
of X-rays, 493-505
Schmidt lines, 604
Schroedinger's equation, 170, 294
and classical wave equation, 193
complex character of, 170
development of, 165-70, 293-4
and differential operators, 204
linearity of, 166-7, 185-6
and Newton's laws, 206-9
independent,
see
Time
Rydberg, 112, 420, 424
Series limit,
postulates of, 165-6, 321
counter,
Nal,
573
112
Series solution of differential equations,
259-60, 301-2
Shadow
scattering, 548-9,
557
599-604
and beta decay, 634-5
and collective model, 605-8
and electric quadrupole moments,
Shell
inde-
pendent Schroedinger equation
Schroedinger's theory of quantum mechanics, and de Broglie's postulate, 165-6
and energy quantization, 178-84
ZnS, 92
Hartree theory of atom, 397
in nuclear structure theory,
Separation constant, 176
Serber potential, 690, 691-2, 693
636
Scattering, 555
Scalars,
Scintillation
in
Semi-empirical mass formula, 590-2
for hydrogen, 112
time
example of evaluation of, 463-7
for gamma spectra, 648
initial and final eigenfunctions and,
463
for JJ coupling spectra, 443
for LS coupling spectra, 440-1
in old quantum theory, 134, 135-6
for X-ray spectra, 486
for Zeeman spectra, 447
Self-consistency, 392
model of
nuclei,
608
and Fermi gas model, 594-5
and gamma decay, 647
and Hartree theory, 600
and JJ coupling, 603
and liquid drop model, 605
and magnetic dipole moments, 6034
and optical model, 594
and spin angular momentum, 603
and spin-orbit interactions, 600-2
theoretical basis of, 593-5
INDEX
725
Shells, in multi-electron atoms,
402-3
602
in nuclei,
Spectroscopic notation, for
307-8
in one-electron atoms,
Spin angular
Shielding, see Effective charge
Simple harmonic
oscillator,
boundary conditions
for,
classical behavior of, 68,
of,
258-9
256
438
meson, 703
of T meson, 703
of neutrino, 626
of neutron, 562
of nuclei, and beta decay selection
rules, 634, 636
and electrons in nuclei, 562
and gamma decay selection rules,
648, 650
measurement of, 451, 658; see
theory, 64, 131
parity of eigenfunctions for, 263
probability densities for, 264
quantization of, in old
theory, 64, 130-1
quantum
selection rules for, 136
time independent Schroedinger equation for,
also Beta decay;
370-1
and exchange force, 372-3
quantum numbers
for,
372
684
Single-valuedness of eigenfunctions and
their derivatives,
Slater determinant,
178, 300, 321
theory of
and fine structure, 131, 338
and quantum mechanics, 355-7
Spatial quantization, 324
experimental verification
of,
335
and form of potential, 324-5
Spectra, analysis of, 448-9
atomic, see Atomic spectra
black body, 48-9
624-33
417
simple harmonic oscillator, 136
X-ray, continuum, 482-3, 489-92
line, 483-9
beta,
optical,
111
widths of, see Line widths
Spectroscopic notation, for /, 408
origin of,
for n,
486
424
quantum numbers
for, for s
X,
and symmetry character, 369
428
in atoms with several optical electrons, 426
quantization of, 371, 429
Spin dependence of nuclear forces, 685,
693
total, 371,
atoms, 131-4
Spectral line,
decay
336
step potential, 231
relativistic
Gamma
and shell model, 603
and symmetry character, 369
values of, 451
of positron, 516
of proton, 562
quantization of, for s = I, 336
367
Solid angle, 497
Sommerfeld's
filled subshell,
ij.
quantization of, 449
Singlet eigenfunctions,
Smoothed
of
of
257-8
zero point energy of, 262
Simultaneity, 16-8
Singlet state of deuteron,
of deuteron,
eigenfunctions for, 300, 337
of electron, 336-9
262
quantum
momentum,
256-66
eigenfunctions for, 263-4
in old
422-3
677
applications of, 256
energy levels
s,
Spherical Bessel functions, 538
Spin g factor, 336-7, 357
Spin-orbit interaction, in atoms, 339-46,
351-3
for alkali
atoms, 421-4
formulas for, 345, 353
for // coupling, 442
LS coupling, 433-4
magnitude of, 345
in nuclei, 600-3
and JJ coupling, 603
magnitude of, 602
and nuclear properties, 600-4
and nucleon-nucleon scattering,
693
and nucleon-nucleus scattering, 602
for
726
INDEX
Spontaneous emission, by atoms, 453,
Step potential, probability
E< V
458-60
459
Einstein's coefficient for,
E > Vu
dipole interactions, 468, 644-
physical picture of, 461-2
transition rate for, 460, 463
by nuclei, see Gamma decay
Spontaneous fission, 617
Square well potential, 239-51
applications of, 240-1
boundary conditions for, E < V Q
241
E > V 250
classical behavior of, 240
eigenfunctions, E
V 243, 247
E > V 250-1
energy levels, E
V 244-5
E> V
wave
,
240, 250
<
247-8
248
Stokes' law, 74
Strength function, 664
Subshells, 405
in alkali atoms,
419-20
405
ordering according to energy, 408-9,
spherical
symmetry of
filled,
398-9,
412
E> V
240,
144,
251
148,
143,
167; see also Linear
combinations
of atoms,
114-5,
108-9,
152, 187, 414, 467
of nuclei, 587-8, 609, 615-6, 622-3
Standing waves in cavity, 52-7
number of allowed frequencies
for,
57
Surface tension energy of nuclei, 590
S-wave
and
187; see also Eigen-
states
46
Stefan's law,
barrier penetration of,
E< V
boundary conditions,
223-4
E > V 228
E< V
E > Vn
E <C V
222
E<V
224-5
229
227
Symmetric eigenfunctions, 363-4,
368-9
E< V
222
226-
363-4, 368
and spin angular momentum, 369
Symmetry character of
227
levels,
E> V
682-7
proton-proton scattering, 687-8
eigenfunctions,
classical behavior,
,
neutron-proton scattering,
for non-interacting identical particles,
applications of, 221
E> V
549-52
proton-proton scattering
for interacting identical particles, 369
Step potential, 221-31
eigenfunctions,
scattering,
limits to
cross section, 690
in
in
Stationary state,
energy
224
Superposition of wave functions,
251
transmission coefficient,
Stability,
E< V
411-2, 415
probability density,
,
function,
and electron spin, 335
on hydrogen atoms, 335
tential
Stern-Gerlach experiment, 334-5
in multi-electron atoms,
parity of eigenfunctions for,
E>
E> V
229-3
see Infinite square well po-
infinite,
transmission coefficient,
()
<
E V 225
229-3
time independent Schroedinger equaV 222
tion, E
E > V 227
,
<
<
225
<
E< K
229
reflection coefficient,
E> K
density,
226
probability flux,
justification for treating only electric
for alpha particles, 368
and behavior of atoms, 368, 414
for electrons, 367
and exclusion principle, 367
for m mesons, 703
for tv mesons, 703
for neutrons, 368
for protons, 368
INDEX
727
Symmetry character of
and
369
spin
eigenfunctions,
momentum,
angular
Total relativistic energy, 34
Transformation, of cross sections,
to
Synchronization of clocks, 16-7, 23-4,
27
Galilean, 5
Lorentz, 26
of
Target nucleus, 563
relativistic,
Thermal emission of electrons, 124-5
Thermal radiation, 42
and quantum theory, 41
Thomas-Fermi theory of atoms, 393-6
Thomas frequency, 343
Thomas precession, 340-4, 359
Thomson scattering, 493-8, 505
cross sections, 495, 498, 505
Thomson's model of atoms, 88
and alpha particle scattering, 92-8
Time dependent perturbation
19,
20-1
176-7, 294
acceptable solutions of, 178, 300
classical
37
521
29
Transition group elements, 413-4
classical,
relativistic,
Transition probability, 290
Transition rate, 291
for absorption, atomic,
453-7
for induced emission,
atomic, 457
and old quantum theory, 137, 291
for spontaneous emission, atomic,
453, 458-60, 463
nuclear, 647-8
Transmission
coefficient,
potential,
for barrier
233-4
E>
Time independent Schroedinger equa-
and
energy, classical,
for square well potential, 240, 251
for step potential,
V 229-31
for mesons, 38, 702
tion,
of velocity,
theory,
see Perturbation theory for time
dependent case
dilation,
momentum and
36
Tensor potential, 678-9, 693
Time
CM
LAB, 523-4
wave equation, 194-5,
235
Transmission probability of nucleons
through nuclear surface, 665
Transmission resonances, for barrier
potential,
in
Ramsauer
234-5
552
Transuranic elements, 414
for deuteron, 680
and energy quantization, 178-84
for Fermi gas, 382
for free particle, 212
for infinite square well potential, 253
for multi-electron atom, 396
for one-electron
atom, see Oneelectron-atom time independent Schroedinger equation
real character of, 177
effect,
Triplet eigenfunctions, 370-1
and exchange
force,
quantum numbers
Triplets, of
energy
372-3
for,
372
levels, see
Fine
structure
of spectral
lines,
424
Triplet state, absence of, in
state of helium,
ground
381
of deuteron, 684
for simple harmonic oscillator, 257-8
by numerical integration,
180-2, 268, 276-7
solution,
by
method, 259-60, 301-2
for step potential, E
Va 222
E > Kn 227
series
<
Total angular
30,
momentum, 346-9, 429-
442
Total cross section, 510, 556; see also
Cross section
Total internal reflection, 235-6
Ultraviolet catastrophe, 63
Uncertainty principle, 155, 156, 159
and barrier penetration, 226-7, 239
and compound nucleus reactions, 663
and de Broglie's postulate, 153-5
and energy
tions,
definition in nuclear reac-
667
and indistinguishability, 360
and line widths, 469-70, 472
and meson rest mass, 695-6
INDEX
728
Uncertainty
wave
principle,
and
duality, 160-2,
particle-
Wave
321-2
physical basis of, 157-8
and quantization of angular momen-
general, 170-5
linear combinations of, 186, 192, 285,
tum, 323
and quantization of electromagnetic
radiation, 157-8
and saturation of nuclear forces,
675-7
and spontaneous emission and absorption of mesons, 700
and spreading of wave groups, 15960
and time dependent perturbation
theory, 288-9
and zero point energy, 255-6
Uniform translation, 4
Unsold's theorem, 312
288
mathematical properties of, 141-6,
184-92
normalization of, 187, 190-1, 206
V 224
for step potential, E
<
274-8
eigenfunctions, 276-7
energy levels, 276-7
potential,
Vector diagrams for addition of angular
momenta, 348-9, 43
Velocity of light,
Velocity transformation, classical, 521
27-9
relativistic,
Virtual level, 553
Virtual state, 553
of deuteron, 685-6
of neutron-proton system, 685-6
Wave
equation, electromagnetic, 235,
697
quantum mechanical, see Schroedinger's equation
relativistic,
697
for string, 193
Wave
functions, 165, 185; see also Pilot
waves
complex character
for
of,
for free particle,
170
664
standing wave, 217
compound nucleus
reactions,
traveling wave, 213
groups
of,
163,
142-5, 153-4, 159-60,
215-6
for identical particles, 364
interference of, 148, 151-2, 153-4,
167
superposition of, 143, 144, 148, 167;
see also Linear combinations, of
wave functions
time dependence
Wave number, 112
of,
191-2
Wave-particle duality, see Particle-wave
duality
Waves
Wave
V-bottom
functions, interpretation of, 205,
206, 214, 217, 294, 295, 318-21,
226-7
in string,
192-200
velocity, 141,
142, 144
Widths, of nuclear energy levels, 660-2
and Breit-Wigner formula, 670
measurement of, 658; see also
Alpha decay; Beta decay;
Gamma
decay
and predictions of compound nucleus theory, 666
and random width assumption of
compound nucleus theory, 667
of quantum states, 470-2
and lifetimes, 472
and uncertainty principle, 470, 472
of spectral lines, see Line widths
Wien's law, 50-1
Weizsacker's atomic mass formula, see
Semi-empirical mass formula
Wilson-Sommerfeld quantization rules,
128-9
applications of, 129-31
de Broglie's interpretation
of,
152
X-ray attenuation, 493
attenuation coefficient, 512
attenuation length, 512
cross section for, 509-11
exponential law for, 511-2
X-ray energy levels, 485-6
X-ray excitation of atoms, 417, 483r 4
X-ray continuum spectra, classical
theory of, 490-1
INDEX
729
continuum
489-92
X-ray
origin
spectra,
of,
X-ray
scattering,
Hartree theory and,
400, 501
Planck's constant and, 482
incoherent, 498-9
properties of, 482-3
Klein-Nishina cross sections for,
quantum theory
503-5
measurement of atomic structure by,
X-ray
line
491-2
and Bohr theory,
of,
spectra,
488
and determination of atomic charge,
488-9
and holes, 484-5
and Moseley experiments, 487-9
origin of, 483-7
400, 501
quantum theory of, 501-4
Thomson, 493-8, 505
Thomson
cross sections for, 495, 498,
505
total cross sections for,
510
photoelectric effect and, 507
Yukawa
properties of, 483
quantum numbers
for,
selection rules for, 486
X-ray pair production, 508-9, 513-7
cross sections for, 508-9, 510
X-ray photoelectric
cross
sections
effect,
for,
505-8
507-8, 510
X-rays, diffraction of, 476-80
discovery of, 475-6
measurement of
spectra, 476,
480-1
properties of, 476
X-ray scattering, 493-505
493-501
coherent, 476-80, 499-501
classical theory of,
coherent cross sections for, 499-501
Compton,
8 1-5,
502-5
potential,
696-8
485
Zeeman
effect, 332, 443-7
anomalous, 447
energy level splitting in, 446-7
Lande g factor and, 446
net magnetic moment in, 444-5
norma], 447
selection rules, 447
spectra, 447-8
Zero point energy, 255
and electrons in nuclei, 561-2
for simple harmonic oscillator, 262
and spontaneous emission of radiation, 462
and uncertainty principle, 255-6
ZnS scintillation detector, 92
You might also like
- Marcelo Alonso, Edward J. Finn - Physics-Addison-Wesley Publishing Company (1992)100% (7)Marcelo Alonso, Edward J. Finn - Physics-Addison-Wesley Publishing Company (1992)1,159 pages
- Introductory Quantum Mechanics by Richard L. Liboff PDF92% (13)Introductory Quantum Mechanics by Richard L. Liboff PDF903 pages
- (G. Kane) Modern Elementary Particle Physics100% (5)(G. Kane) Modern Elementary Particle Physics367 pages
- (Oxford Graduate Texts) Milonni, Peter W. - An Introduction To Quantum Optics and Quantum Fluctuations (2019, Oxford University Press) PDFNo ratings yet(Oxford Graduate Texts) Milonni, Peter W. - An Introduction To Quantum Optics and Quantum Fluctuations (2019, Oxford University Press) PDF543 pages
- Great Experiments in Physics: Firsthand Accounts from Galileo to EinsteinFrom EverandGreat Experiments in Physics: Firsthand Accounts from Galileo to Einstein4.5/5 (2)
- Smith W. F. - Waves and Oscillations. A Prelude to Quantum Mechanics (Волны и колебания. Введение в квантовую механику) - 2010 PDF100% (1)Smith W. F. - Waves and Oscillations. A Prelude to Quantum Mechanics (Волны и колебания. Введение в квантовую механику) - 2010 PDF416 pages
- Modern Nuclear Physics - From Fundamentals To Frontiers by Alexandre Obertelli, Hiroyuki Sagawa100% (9)Modern Nuclear Physics - From Fundamentals To Frontiers by Alexandre Obertelli, Hiroyuki Sagawa739 pages
- (Major American Universities Ph.D. Qualifying Questions and Solutions) Yung-Kuo Lim-Problems and Solutions on Electromagnetism_ Major American University Ph.D. Qualifying Questions and Solutions-World100% (2)(Major American Universities Ph.D. Qualifying Questions and Solutions) Yung-Kuo Lim-Problems and Solutions on Electromagnetism_ Major American University Ph.D. Qualifying Questions and Solutions-World677 pages
- Princeton Problems in Physics With Solutions PDFNo ratings yetPrinceton Problems in Physics With Solutions PDF270 pages
- Oleg D Jefimenko Electricity and Magnetism An Introduction To The Theory of Electric and Magnetic Fields PDF100% (3)Oleg D Jefimenko Electricity and Magnetism An Introduction To The Theory of Electric and Magnetic Fields PDF618 pages
- (David J. Griffiths) Introduction To Quantum Mecha (BookFi) PDFNo ratings yet(David J. Griffiths) Introduction To Quantum Mecha (BookFi) PDF408 pages
- Modern Physics: The Quantum Physics of Atoms, Solids, and Nuclei: Third EditionFrom EverandModern Physics: The Quantum Physics of Atoms, Solids, and Nuclei: Third Edition4.5/5 (7)
- Curriculum Vitae: April 2011 Mujahid Kamran Vice Chancellor, Punjab University (Wef Jan 3, 2008)No ratings yetCurriculum Vitae: April 2011 Mujahid Kamran Vice Chancellor, Punjab University (Wef Jan 3, 2008)23 pages
- Chang K. Mathematical Structures of Quantum Mechanics 2011100% (1)Chang K. Mathematical Structures of Quantum Mechanics 2011208 pages
- University Physics I - Classical Mechanics100% (1)University Physics I - Classical Mechanics342 pages
- M - SC - Physics - 345 11 - Classical Mechanics100% (3)M - SC - Physics - 345 11 - Classical Mechanics290 pages
- P. C. Deshmukh - Foundations of Classical Mechanics-Cambridge University Press (2019)100% (1)P. C. Deshmukh - Foundations of Classical Mechanics-Cambridge University Press (2019)592 pages
- I. P. Gurskii-Elementary Physics - Problems and Solutions-Imported Pubn (1989) PDF100% (4)I. P. Gurskii-Elementary Physics - Problems and Solutions-Imported Pubn (1989) PDF520 pages
- (Addison-Wesley Series in Physics) Marcelo Alonso, Edward J. Finn - Fundamental University Physics I Mechanics (1967, Addison-Wesley Publishing Company) PDF100% (2)(Addison-Wesley Series in Physics) Marcelo Alonso, Edward J. Finn - Fundamental University Physics I Mechanics (1967, Addison-Wesley Publishing Company) PDF487 pages
- Greiner - Classical Electrodynamics (1998)100% (10)Greiner - Classical Electrodynamics (1998)569 pages
- Spiegel-SchaumsTheoryAndProblemsOfTheoreticalMechanics Text PDFNo ratings yetSpiegel-SchaumsTheoryAndProblemsOfTheoreticalMechanics Text PDF375 pages
- Bellantoni L Modern Physics The Scenic Route100% (1)Bellantoni L Modern Physics The Scenic Route205 pages
- Morris Edgar Rose Elementary Theory of Angular Momentum 1957100% (4)Morris Edgar Rose Elementary Theory of Angular Momentum 1957251 pages
- (Science For Everyone) G. I. Kopylov-Elementary Kinematics of Elementary Particles (Science For Everyone) - Mir Publishers (1983)100% (1)(Science For Everyone) G. I. Kopylov-Elementary Kinematics of Elementary Particles (Science For Everyone) - Mir Publishers (1983)273 pages
- Ginzburg Key Problems of Physics and AstrophysicsNo ratings yetGinzburg Key Problems of Physics and Astrophysics174 pages
- I V Savelyev Fundametals of Theoretical Physics Vol 1 PDF100% (6)I V Savelyev Fundametals of Theoretical Physics Vol 1 PDF425 pages
- Rodney Loudon-The Quantum Theory of Light, 3rd Ed. (Oxford Science Publications) - Oxford University Press, USA (2000)100% (1)Rodney Loudon-The Quantum Theory of Light, 3rd Ed. (Oxford Science Publications) - Oxford University Press, USA (2000)448 pages
- Thermal Physics by CHARLES KITTEL and HE PDF100% (2)Thermal Physics by CHARLES KITTEL and HE PDF495 pages
- A Course in Classical Physics 1-Mechanics PDF100% (7)A Course in Classical Physics 1-Mechanics PDF400 pages
- 50th Anniversary of The Judd Ofelt TheoryNo ratings yet50th Anniversary of The Judd Ofelt Theory19 pages
- A Ratiometric and NonEnzymatic Luminescence Assay For Uric Acid Differential Quenching of Lanthanide Excited States by AntiOxidantsNo ratings yetA Ratiometric and NonEnzymatic Luminescence Assay For Uric Acid Differential Quenching of Lanthanide Excited States by AntiOxidants3 pages
- Nuclei and Particles: Introduction To Nuclear and Subnuclear Physics-2Ed - Emilio Segre100% (2)Nuclei and Particles: Introduction To Nuclear and Subnuclear Physics-2Ed - Emilio Segre990 pages
- Form 4 Physics Sir Hazeeq 05.02.2024 - 2No ratings yetForm 4 Physics Sir Hazeeq 05.02.2024 - 214 pages
- Physical Sciences Grade 12 Feb March 2018 P1 and MemoNo ratings yetPhysical Sciences Grade 12 Feb March 2018 P1 and Memo33 pages
- Science9 - Q4 - M3 Impulse and MomentumNo ratings yetScience9 - Q4 - M3 Impulse and Momentum14 pages
- Phy 101 Lecture10 (Momentum and Collisions)No ratings yetPhy 101 Lecture10 (Momentum and Collisions)23 pages
- Physics 2000 and Calculus 2000 - Modern College-Level Calculus-Based Physics PDFNo ratings yetPhysics 2000 and Calculus 2000 - Modern College-Level Calculus-Based Physics PDF1,416 pages
- Abraham A. Ungar - On The Origin of The Dark Matter/Energy in The Universe and The Pioneer AnomalyNo ratings yetAbraham A. Ungar - On The Origin of The Dark Matter/Energy in The Universe and The Pioneer Anomaly6 pages
- BIU-ALLIED HEALTH SCIENCES-B.N. SC Nursing Science Integrated - Rosemary O. ObasiNo ratings yetBIU-ALLIED HEALTH SCIENCES-B.N. SC Nursing Science Integrated - Rosemary O. Obasi81 pages
- Conservation of Momentum Using PASCO TM Carts and Track To Study Collisions in One DimensionNo ratings yetConservation of Momentum Using PASCO TM Carts and Track To Study Collisions in One Dimension24 pages
- B.Sc. (Hons.) Mthematics - 21 - 8 - 18 IGU UniversityNo ratings yetB.Sc. (Hons.) Mthematics - 21 - 8 - 18 IGU University75 pages
- 2024 Version - AP Physics 1 Review PacketNo ratings yet2024 Version - AP Physics 1 Review Packet12 pages
