<►;

$<E>oC=3<

LIBRARY §
Kashmiri Gate, Delhi-110006 |

Accession No.

Class No._ sj^yySi e

\oArt*> '• ■I
Book No._

*EOoa» ®00»«S)00> <S)005KHO«» oS)oO»

 DELHI COLLEGE OF ENGINEERING
Kashmiri Gate, Delhi-110006
LIBRARY

DATE DUE
For each day's delay after the due date a fine of
10 Pafse per Vol. shall be charged for the first week, and
50 Paise per Vol. per day for subsequent days.
 INTERNATIONAL SERIES IN
PURE AND APPLIED PHYSICS
G. P. HARNWELL, Consulting Editor

INTRODUCTION TO ATOMIC SPECTRA

INTERNATIONAL SERIES IN PURE AND APPLIED PHYSICS

Leonard I. SCHIFF, Consulting Editor

Allis and Herlin Thermodynamics and Statistical Mechanics

Becker Introduction to Theoretical Mechanics
Clark Applied X-rays
Evans The Atomic Nucleus
Finkelnburg Atomic Physics
Oinzton Microwave Measurements
Oreen Nuclear Physics
Oumey Introduction to Statistical Mechanics
Hall Introduction to Electron Microscopy
Hardy and Perrin The Principles of Optics
Harnwell Electricity and Electromagnetism
• Harnwell and Livingood Experimental Atomic Physics
Harnwell and Stephens Atomic Physics
Houston Principles of Mathematical Physics
Hund High-frequency Measurements
Kennard Kinetic Theory of Gases
Leighton Principles of Modern Physics
Middleton An Introduction to Statistical Communication Theory
Morse Vibration and Sound
Morse and Feshbach Methods of Theoretical Physics
Muskat Physical Principles of Oil Production
Present Kinetic Theory of Gases
Read Dislocations in Crystals
Richtmyer, Kennard, and Lauritsen Introduction to Modern Physics
Schiff Quantum Mechanics
Seitz The Modern Theory of Solids
Slater Introduction to Chemical Physics
Slater Quantum Theory of Atomic Structure, Volume I
Slater Quantum Theory of Atomic Structure, Volume II
Slater Quantum Theory of Matter
Slater and Frank Electromagnetism
Slater and Frank Introduction to Theoretical Physics
Slater and Frank Mechanics
Smythe Static and Dynamic Electricity
Stratton Electromagnetic Theory
Thorndike Mesons: A Summary of Experimental Pacts
Townes and Schawlow Microwave Spectroscopy
White Introduction to Atomic Spectra

The late P. K. Richtmyer was Consulting Editor of the series from its inception in
1929 to his death in 1939. Lee A. DuBridge was Consulting Editor from 1939 to
1946, and G. P. Harnwell from 1947 to 1954.
 INTRODUCTION
TO
ATOMIC SPECTRA

BY

HARVEY ELLIOTT WHITE, Ph.D,

Assistant Proffssor of Physics, at the
University of California

INTERNATIONAL STUDENT EDITION

McGRAW-HILL KOGAKUSHA, LTD.

TOKYO DUSSELDORF JOHANNESBURG LONDON MEXICO
NEW DELHI PANAMA RIO DE JANEIRO SINGAPORE SYDNEY
 INTRODUCTION lO ATOMIC SPECTRA

IN TERN A TIONAL S TV DEN T EDI TI ON

Exclusive rights by McGraw-Hill Kogakusha, Ltd., for manufacture and

export. This book cannot be re-exported from the countiy to which it
is consigned by McGraw-Hill.
XI

Copyright, 1934, by the McGiaw-Hill Book Company, Inc. All rights

reserved No part of this publication may be reproduced, stored in a
retrieval system, or transmitted, in any form or by any means, electronic,
mechanical, photocopying, recording, or otherwise, without the piior
written permission of the publisher.

TOSHO PRINTING CO., LTD.

TOKYO. JAPAN
TO MY WIFE
 PREFACE
During the past few years students have frequently come to the author
with this query, “ Where can I find an elementary treatment of atomic
spectra?” Further questioning has shown, in general, that, although
these students have had at least one year of calculus and one or more years
of college physics, they have found most treatments of atomic spectra
either too brief or too highly mathematical. The desire to meet the
situation has given the author the incentive and the encouragement
to write this book.
In preparing a manuscript that will in some measure meet the desires
of beginning students, as well as those already familiar with certain
phases of atomic spectra, the author set up the following three objectives:
first, to start as nearly as possible at the beginning of each subject;
second, to develop each new concept so that the student with a working
knowledge of elementary physics and elementary calculus should have
little difficulty in following; and third, clearly to illustrate each chapter, as
far as possible, with diagrams and photographs of actual spectra.
At the outset one is confronted with the problem of describing a field
of scientific investigation that has developed, and still is developing,
at a rapid rate. On the one hand we have the Bohr-Sommerfeld theory
of an orbital atom and on the other the newer and more satisfactory
theory of quantum mechanics. Believing that one can better under¬
stand the principles of quantum mechanics by first becoming well
acquainted with the observed spectra and with the vector model, the
orbital model of the atom is first treated in some detail. Once developed
the orbital model then furnishes a very easy step to the quantum-
mechanical atom as given by Schrodinger’s wave equation or by the
Dirac electron. At frequent intervals throughout the* book both the old
and new models, of orbits and probability density functions, are compared
with each other.
The treatment as it is given here starts with the historical background
of spectroscopy (Chap. I) and develops in the first ten chapters the old
and the new quantum theories of one-valence-electron atoms. The
second half of the book (Chaps. XI through XXI) deals with complex
atomic systems of two or more valence electrons and includes a brief
account of x-ray spectra. Believing that frequent comparison with the
observed spectral lines, by giving actual reproductions of spectrograms,
is all too often neglected, the author has spent a great deal of time in
photographing many of the spectra used as illustrations. That photo
vii
viii PREFACE

graphs are an extremely important feature of any book on atomic spectra

may be emphasized by pointing out that, of all the theories and knowledge
concerning atoms, the spectrum lines will remain the same for all time.
For the benefit of those readers who profit greatly by the working
out of examples, typical problems on each subject are given at the end of
most chapters. All sections, figures, tables, and equations are assigned
two numbers, the first number designating the chapter, and the second the
ordinal number in the chapter. The purpose of this notation is to facili¬
tate the finding of any section, figure, table, or equation referred to in the
text. In referring to Eq. (4.9), for example, one turns to Chap. IV, using
the headings at the top of each page, and then to the ninth equation in
that chapter.
The author wishes to take this opportunity especially to thank
Prof. Raymond T. Birge and Dr. E. Ramberg for their valuable criticism
of the entire manuscript. Sincere thanks are also to be extended to
Adeline Dally White for her assistance and patience in the reading of the
proof, Prof. F. K. Richtmyer and S. Allison for their reading of Chap. XVI
on x-ray spectra, Prof. H. N. Russell for reading Chaps. XI, XII, XIII,
XIV, XV, and XIX, on complex spectra, Prof. S. Goudsmit for reading
Chap. XVIII on hyperfine structure, Prof. P. S. Epstein for reading Chap.
XX on the Stark effect, Prof. J. R. Oppenheimer for the reading of
sections dealing with quantum-mechanical results, and Profs. R. B. Brode
and F. A. Jenkins for their reading of various sections and chapters.
Most sincere thanks are also extended to Drs. A. S. King and P. W.
Merrill of the Mount Wilson Observatory, and Prof. D. H. Menzel of the
Lick and Harvard observatories, for the original terrestrial, stellar, and
solar spectrograms, and to Mr. H. D. Babcock of the Mount Wilson
Observatory for the very fine Zeeman patterns of vanadium and chro¬
mium. I also wish to thank Prof. F. K. Richtmyer and Dr. F. Hirsch for
the original spectrograms reproduced in Figs. 16.5 and 16.7, Prof. M.
Siegbahn for the copies of the x-ray diffraction and interference patterns
reproduced in Figs. 16.2 and 16.3, Prof. J. S. Foster for the Stark-effect
patterns reproduced in Fig. 20.12, Profs. G. N. Lewis and C. D. Shane and
Drs. F. H. Spedding and N. S. Grace for the spectrograms of Hi and Hi
reproduced in Figs. 9.1 and 9.5. Finally I wish to thank Prof. R. B.
Brode and Miss Burmann for especially photographing the x-ray absorp¬
tion spectra shown in Fig. 16.9, S. S. Ballard for the photographing of the
Lyman series of H1 and H2, reproduced in Fig. 2.12, and Dr. N. S. Grace,
Dr. E. McMillan, and K. More for the original spectrograms reproduced
in Fig. 18.2.
Harvey Elliott White.
Berkeley, Calif.,
September, 1934.
 CONTENTS
Pag*
Preface.vii

CHAPTER I
Early Historical Developments in Atomic Spectra. 1
1.1. Kirchhoff's Law. 1.2. A New Era. 1.3. Balmer's Law. 1.4. Ryd¬
berg's Contributions. 1.5. The Rydberg-Schuster Law. 1.6. Series Nota¬
tion. 1.7. Satellites and Fine Structure. 1.8. The Lyman, Balmer,
Paschen, Brackett, and Pfund Series of Hydrogen. 1.9. The Ritz Com¬
bination Principle. 1.10. The Ritz Formula. 1.11. The Hicks Formula.
1.12. Series Formulas Applied to the Alkali Metals. 1.13. Neon with 130
Series. 1.14. Normal and Abnormal Series. 1.15. Hydrogen and the
Pickering Series. 1.16. Enhanced Lines.

CHAPTER II
Introduction to the Quantum Theory and the Bohr Atom.23
2.1. Planck's Simple Harmonic Oscillator. 2.2. The Bohr Atom. 2.3.
Bohr's First Assumption. 2.4. Bohr’s Second Assumption. 2.5. Bohr’s
Third Assumption. 2.6. Characteristics of the Bohr Circular Orbits. 2.7.
Bohr Orbits and the Hydrogen Series. 2.8. Series of Ionized Helium, He II
(IIe+). 2.9. Series of Doubly Ionized Lithium, Li III (Li++) and Triply
Ionized Beryllium, Be IV (Be+++). 2.10. Energy Level Diagrams. 2.11.
Unquantized States and Continuous Spectra. 2.12. The So-called “Reduced
Mass” of the Electron. 2.13. Variation of the Rydberg Constant. 2.14.
Bohr’s Correspondence Theorem.

CHAPTER III
Sommerfeld’s Elliptic Orbits and Space Quantization.42
3.1. Two Degrees of Freedom. 3.2. The Radial Quantum Number. 3.3.
The Total Energy W. 3.4. General Characteristics of Sommerfeld’s
Elliptic Orbits. 3.5. Space Quantization. 3.6. Larmor’s Theorem. 3.7.
Magnetic Moment and the Bohr Magneton.

CHAPTER IV
Quantum Mechanics and the SchrOdinoer Wave Equation.54
4.1. De Broglie's Corpuscular Wave Equation. 4.2. The Schrodinger Wave
Equation. 4.3. Schrodinger's Wave Equation Applied to Hydrogen.
4.4. Eigenfunctions. 4.5. The <p Factor <t»m of the Eigenfunction if/. 4.6.
The 0 Factor 6m,i of the Eigenfunction \J/. 4.7. Correlation of and
0»»,i9w,i with the Bohr-Sommerfeld Orbits. 4.8. The Radial Factor Rnti of
the Eigenfunction 4.9. Correlation of Rn,iRt,i with the Bohr-Sommerfeld
Orbits. 4.10. A General Interpretation of the Eigenfunction 4.11. Use¬
ful Atomic Models. 4.12. Spherical Symmetry.

CHAPTER V
The Alkali Metals and the Periodic Table.77
5.1. Energy Level Diagrams. 5.2. The Bohr-Stoner Scheme of the Building
Up of the Elements. 5.3. The First Period. 5.4. The Second Period. 5.5.
LX
X CONTENTS
Pagi
The Third Period. 5.6. The Fourth Period or First Long Period. 5.7. The
Fifth Period or Second Long Period. 5.8. The Sixth Period or Third Long
Period. 5.9. The Seventh and Last Period. 5.10. Energy Levels of the
Alkali Metals. 5.11. The Effective Quantum Number and the Quantum
Defect. 5.12. The Selection Principle.

CHAPTER VI
Excitation Potentials, Ionization Potentials, and the Spectra op Ionized
Atoms. .92
6.1. Critical Potentials. 6.2. The Spectra of Ionized Atoms.

CHAPTER VII
Penetrating and Nonpenetrating Orbits in the Alkali Metals.100
7.1. The Quantum-mechanical Model of the Alkali Metals. 7.2. Penetrat¬
ing and Nonpenetrating Orbits. 7.3. Nonpenetrating Orbits. 7.4. Pene¬
trating Orbits on the Classical Model. 7.5. Quantum-mechanical Model
for Penetrating Orbits.

CHAPTER VIII
Doublet Fine Structure and the Spinning Electron.114
8.1. Observed Doublet Fine Structure in the Alkali Metals and the Boron
Group of Elements. 8.2. Selection Rules for Doublets. 8.3. Intensity
Rules for Fine-structure Doublets. 8.4. The Spinning Electron and the
Vector Model. 8.5. The Normal Order of Fine-structure Doublets. 8.6.
Electron Spin-orbit Interaction. 8.7. Spin-orbit Interaction for Non¬
penetrating Orbits. 8.8. Spin-orbit Interaction for Penetrating Orbits.

CHAPTER IX
Hydrogen Fine Structure and the Dirac Electron.132
9.1. Sommerfeld Relativity Correction. 9.2. Fine Structure and the Spin¬
ning Electron. 9.3. Observed Hydrogen Fine Structure. 9.4. Fine Struc¬
ture of the Ionized Helium Line X4686. 9.5. The Dirac Electron and the
Hydrogen Atom. 9.6. The Angular Distribution of the Probability Den¬
sity P$. 9.7. The Radial Distribution of the Probability Density Pr. 9.8.
The Probability Density Distribution W*. 9.9. The Sommerfeld Formula
from Dirac’s Theory.

CHAPTER X
The Zeeman Eppect and the Paschen-back Effect.149
10.1. Early Discoveries and Developments. 10.2 The Vector Model of a
One-electron System in a Weak Magnetic Field. 10.3. The Magnetic
Moment of a Bound Electron. 10.4. Magnetic Interaction Energy. 10.5
Selection Rules. 10.6. Intensity Rules. 10.7. The Paschen-Back Effect.
10.8. Paschen-Back Effect of a Principal-series Doublet. 10.9. Selection
Rules forthe Paschen-Back Effect. 10.10. The Zeeman Effect, and Paschen-
Back Effect, of Hydrogen. 10.11. A Quantum-mechanical Model of the
Atom in a Strong Magnetic Field.

CHAPTER XI
Singlet and Triplet Series op Two-valence-electron Systems.171
11.1. General Series Relations. 11.2. Triplet Fine Structure. 11.3. The
Quantum Numbers n and l of Both Valence Electrons. 11.4. Penetrating
and Nonpenetrating Electrons for Two-electron Systems. 11.5. The Excita¬
tion of Both Valence Electrons.
 CONTENTS xi
Pag*
CHAPTER XII
The Atom Model fob Two Valence Electrons.184
12.1. U-coupling. 12.2. Spin-spin-, or 88-coupling. 12.3. L&-, or Russell-
Saunders Coupling. 12.4. The Pauli Exclusion Principle. 12.5. Triplet
Multiplets in Ionized Scandium, Sc II. 12.6. Coupling Schemes for Two
Electrons. 12.7. T Factors for L>S-coupling. 12.8. The Land6 Interval
Rule. 12.9. ^/-coupling. 12.10. ^-coupling in the Carbon Group of
Elements. 12.11. Term Series and Limits in Two-electron Systems. 12.12.
The Great Calcium Triad. 12.13. The Branching Rule. 12.14. Selection
Rules. 12.15. Intensity Relations. 12 16. Relative Intensities of Related
Multiplets. 12.17. Helium and Helium-like Atoms. 12.18. Quantum
mechanical Model of Helium. 12.19. Fine Structure of Helium-like Atoms.

CHAPTER XIII
Zeeman Effect, Paschen-Back Effect, and the Pauli Exclusion Principle,
for Two Electrons.215
13.1. The Magnetic Moment of the Atom. 13.2. The Zeeman Effect.
13.3. Intensity Rules for the Zeeman Effect. 13.4. The Calculation of
Zeeman Patterns. 13.5. LS- and jj-coupling and the g Sum Rule. 13.6.
Paschen-Back Effect. 13.7. //^-coupling and the Paschen-Back Effect.
13.8. jj-coupling and the Paschen-Back Effect. 13.9. Complete Paschen-
Back Effect. 13.10 Breit’s Scheme for the Derivation of Spectral Terms
from Magnetic Quantum Numbers. 13.11. The Pauli Exclusion Principle.
13 12. Pauli’s g Permanence Rule. 13.13. Pauli’s g Sum Rule for All Field
Strengths. 13.14. Land6’s T Permanence Rule. 13.15. Goudsmit’s T
Sum Rule. 13.16. The T Permanence and V Sum Rules Applied to Two
Equivalent Electrons.
CHAPTER XIV
Complex Spectra.248
14 1. The Displacement Law. 14.2. Alternation Law of Multiplicities.
14 3. The Vector Model for Three or More Valence Electrons. 14.4. Terms
Arising from Three or More Equivalent Electrons. 14.5. The Land6
Interval Rule. 14.6. Inverted Terms. 14.7. Hund’s Rule. 14.8. The
Nitrogen Atom. 14.9. The Scandium Atom. 14.10. The Oxygen Atom.
14.11. The Titanium Atom. 14.12. The Manganese Atom. 14.13. The
Rare-gas Atoms, Neon, Argon, Krypton, and Xenon. 14.14. The Normal
States of the Elements in the First and Second Long Periods. 14.15.
Houston’s Treatment of One s Electron and One Arbitrary Electron. 14.16.
Slater’s Multiplet Relations. 14.17. Multiplet Relations of Goudsmit and
Inglis. 14.18. Relative Intensities of Multiplet Lines.

CHAPTER XV
The Zeeman Effect and Magnetic Quantum Numbers in Complex Spectra 286
15.1. Magnetic Energy and the Land6 g Factor. 15.2. The Calculation of
Zeeman Patterns. 15.3. Intensity Rules and Zeeman Patterns for Quartets,
Quintets, and Sextets. 15.4. Paschen-Back Effect in Complex Spectra.
15.5. Derivation of Spectral Terms by Use of Magnetic Quantum Numbere.
15.6. Equivalent Electrons and the Pauli Exclusion Principle.

CHAPTER XVI
X-Ray Spectra.299
16.1. The Nature of X-rays. 16.2. X-ray Emission Spectra and the Moseley
Law. 16.3. AbsorDtion Spectra. 16.4. Energy Levels. 16.5. Seieeti<»*
xii CONTENTS
P\Q*
and Intensity Rules. 16.6. Fine Structure of X-rays. 16.7. Spin-relativ¬
ity Doublets (Regular Doublets). 16.8. The Regular-doublet Law. 16.9.
Screening Doublets and the Irregular-doublet Law. 16.10. A Predicted
Structure in X-rays. 16.11. X-ray Satellites. 16.12. Explanation of X-ray
Absorption Spectra.
CHAPTER XVII
ISOELECTRONIC SEQUENCES.331
17.1. Isoelectronic Sequences of Atoms Containing One Valence Electron.
17.2. Optical Doublets and the Irregular-doublet Law. 17.3. Optical
Doublets and the Regular-doublet Law. 17.4. Isoelectronic Sequences of
Atoms Containing Two Valence Electrons. 17.5. Isoelectronic Sequences
Containing Three or More Valence Electrons. 17.6. The Irregular-
doublet Law in Complex Spectra. 17.7. Energy Relations for the Same
Atom in Different Stages of Ionization. 17.8. Centroid Diagrams.

CHAPTER XVIII
Hyperfine Structure. .C.r2
18.1. Introduction. 18.2. Hyperfine Structure and the Land6 Interval
Rule. 18.3. Nuclear Interaction with One Valence Electron. 18.4.
Nuclear Interaction with a Penetrating Electron. 18.5. Classical Explana¬
tion of Normal and Inverted Hfs. 18.6. Hyperfine Structure in Atoms with
Two or more Valence Electrons. 18.7. Hyperfine Structure in Complex
Spectra. 18.8. Nuclear gt Factors. 18.9. Zeeman Effect in Hyperfine
Structure. 18.10. Back-Goudsmit Effect in Hyperfine Structure 18.11.
Isotope Structure. 18.12. Isotope Structure and Hyperfine Structure
Combined.
CHAPTER XIX
Series Perturbations and Autoionization. 386
19.1. Observed Abnormal Series. 19.2. Energy Level Perturbations. 19.3
The Nature of and Conditions for Term Perturbations. 19.4. The Anoma¬
lous Diffuse Series of Calcium. 19.5. The Anomalous Principal Series in
Copper. 19.6. The Inverted Alkali Doublets. 19.7. Autoionization. 19.8.
Autoionization in Copper. 19.9. Autoionization in Calcium, Strontium,
and Barium. 19.10. Hyperfine-structure Term Perturbations.

CHAPTER XX
The Stark Effect. .401
20.1. Discovery of the Stark Effect. 20.2. The Stark Effect of Hydrogen.
20.3. Early Orbital Model of Hydrogen in an Electric Field. 20.4. Weak-
field Stark Effect in Hydrogen. 20.5. Strong-field Stark Effect in Hydrogen.
20.6. Second-order Stark Effect in Hydrogen. 20 7 Stark Effect for More
than One Electron. 20.8. The Stark Effect in Helium.

CHAPTER XXI
The Breadth of Spectrum Lines.418
21.1. The Doppler Effect. 21.2. Natural Breadths from Classical Theory.
21.3. Natural Breadths and the Quantum Mechanics. 21.4. Observed
Natural Breadths and Doppler Broadening. 21.5. Collision Damping.
21.6. Asymmetry and Pressure Shift. 21.7 Stark Broadening.

Appendix.437

Index .443
 INTRODUCTION
TO ATOMIC SPECTRA
CHAPTER I

EARLY HISTORICAL DEVELOPMENTS IN ATOMIC SPECTRA

Spectroscopy as a field of experimental and theoretical research has
contributed much to our knowledge concerning the physical nature of
things—knowledge not only of our own earth but of the sun, of inter¬
stellar space, and of the distant stars. It may rightly be said that
spectroscopy had its beginning in the year 1666 with the discovery by
Sir Isaac Newton that different colored rays of light when allowed to
pass through a prism were refracted at different angles. The experiments
that Newton actually carried out are well known to everyone. Sunlight
confined to a small pencil of rays by means of a hole in a diaphragm and
then allowed to pass through a prism was spread out into a beautiful
band of color. Although it was known to the ancients that clear crystals
when placed in direct sunlight gave rise to spectral arrays, it remained
for Newton to show that the colors did not originate in the crystal but
were the necessary ingredients that go to make up sunlight. With a
lens in the optical path the band of colors falling on a screen became a
series of colored images of the hole in the diaphragm. This band Newton
called a spectrum.
Had Newton used a narrow slit as a secondary source of light and
examined carefully its image in the spectrum, he probably would have
discovered, as did Wollaston1 and Fraunhofer2 more than one hundred
years later, the dark absorption lines of the sun's spectrum. Fraunhofer
took it upon himself to map out several hundred of the newly found
lines of the solar spectrum and labeled eight of the most prominent ones
by the first eight letters of the alphabet (see Fig. 1.1). These lines are
now known as the Fraunhofer lines.
1.1. Kirchhoff’s Law.—More than half a century passed in the history
of spectroscopy before a satisfactory explanation of the Fraunhofer
lines was given. Foucault3 showed that, when light from a very power-
1 Wollaston, W. H., Phil. Trans. Roy. Soc., II, 365, 1802.
* Fraunhofer, J., Gilbert s, Ann., 56, 264, 1817.
* Foucault, L., Ann. chim. et phys., 68, 476, 1860.
1
 2 INTRODUCTION TO ATOMIC SPECTRA [Chap. I

ful arc was first allowed to pass through a sodium flame just in front
of the slit of a spectroscope, two black lines appeared in exactly the same
position of the spectrum as the two D lines of the sun’s spectrum. Not
many years passed before evidence of this kind proved beyond doubt
that many of the elements found on the earth were to be found also
in the sun. Kirchhoff1 was not long in coming forward with the theory
that the sun is surrounded by layers of gases acting as absorbing screens
for the bright lines emitted from the hot surfaces beneath.
Fraunhofer Lines
* D

Q0005 Q0006 0.0007 Millimeters

5000 6000 7000 Ahgstroms
Fia. 1.1.—Prominent Fraunhofer lines. Solar spectrum.

In the year 1859 Kirchhoff gave, in papers read before the Berlin
Academy of Sciences, a mathematical and experimental proof of the
following law: The ratio between the powers of emission and the powers of
absorption for rays of the same wave-length is constant for all bodies at the
same temperature. To this law, which goes under Kirchhoff’s name, the
following corollaries are to be added: (1) The rays emitted by a substance
excited in some way or another depend upon the substance and the tem¬
perature; and (2) every substance has a power of absorption which is a maxi¬
mum for the rays it tends to emit. The impetus KirchhofFs work gave to the
field of spectroscopy was soon felt, for it brought many investigators
into the field.
In 1868 Angstrom2 set about making accurate measurements of the
solar lines and published an elaborate map of the sun’s spectrum. Ang¬
strom’s map, covering the visible region of the spectrum, stood for a
number of years as a standard source of wave-lengths. Every line to be
used as standard was given to ten-millionths of a millimeter.
1.2. A New Era.—The year 1882 marks the beginning of a new era in
the analysis of spectra. Realizing that a good grating is essential to
accurate wave-length measurements, Rowland constructed a ruling
engine and began ruling good gratings. So successful was Rowland3 in
this undertaking that within a few years he published a photographic
map of the solar spectrum some fifty feet in length. Reproductions
from two sections of this map are given in Fig. 1.2, showing the sodium

1 Kirchhoff, G., Monatsber. Berl. Akad. Wiss., 1859, p. 662; Pogg. Ann.j 109,
148, 275, 1860; Ann. chim. et phys., 58, 254, 1860; 59, 124, I860.
* Angstrom, A. J., Uppsala, W. Schultz, 1868.
* Rowland, H. A., Johns Hopkins Univ. Circ. 17, 1882; Phil. Mag., 18,469,1882;
Nature, 26, 211, 1882.
Sec. 1.2] EARLY HISTORICAL DEVELOPMENTS 3
D lines, the iron E lines, and the ionized calcium H lines. With a wave¬
length scale above, the lines, as can be seen in the figure, were given to
ten millionths of a millimeter, a convenient unit of length introduced by
Angstrom and now called the Angstrom unit The Angstrom unit

Fig. 1.2.— Sections of Rowland’s solar map.

is abbreviated A, or just A, and, in terms of the standard meter, 1 m

= 1010 A.
Up to the time Balmer (1885) discovered the law of the hydrogen
series, many attempts had been made to discover the laws governing
the distribution of spectrum lines of any element. It was well known
that the spectra of many elements contained hundreds of lines, whereas
the spectra of others contained relatively few. In hydrogen, for example*
half a dozen lines apparently comprised its entire spectrum. These few
lines formed what is now called a series (see Fig. 1.3).
In 1871 Stoney,1 drawing an analogy between the harmonic over¬
tones of a fundamental frequency in sound and the series of lines in
Violet Blue Blue-Green Red

Hydrogen
-1---1-1_i_■
00004 0 0005 Q0006 Millimeters
4000 5000 6000 Angstroms
Fig. 1.3.—The Balmer series of hydrogen.

hydrogen, pointed out that the first, second, and fourth lines were the
twentieth, twenty-seventh, and thirty-second harmonics of a fundamental
vibration#whose wave-length in vacuo is 131, 274.14 A. Ten years later
Schuster2 discredited this hypothesis by showing that such a coincidence
is no more than would be expected by chance.
1 Stoney, G. J., Phil. Mag.} 41, 291, 1871.
* Schuster, A., Proc. Roy. Soc. Itmdon, 31, 337, 1881.
 4 INTRODUCTION TO ATOMIC SPECTRA [Chap. I

Liveing and Dewar,1 in a study of the absorption of spectrum lines,

made the outstanding discovery that most of the lines of sodium and
potassium could be arranged into series of groups of lines. A reproduc¬
tion of their diagram is given in Fig. 1.4. Excluding the D lines of
sodium, each successive group of four lines becomes fainter and more
diffuse as it approaches the violet end of the spectrum. Liveing and

■■»SHH■■HB
Not

hBH111 111111 ■1|| |II\mm

sd sd sd sd s d s d
j_|_
4500 5000 5500 6000 Angstroms

Fiq. 1.4.—Schematic representation of the sodium and potassium series. (After Liveing
and Dewar.)

Dewar say that while the wave-lengths of the fifth, seventh, and eleventh
doublets of sodium were very nearly as ^ the whole series cannot
be represented as harmonics of one fundamental. Somewhat similar
harmonic relations were found in potassium but again no more than would
be expected by chance.
Four years later Hartley2 discovered that the components of a
doublet or triplet series have the same separations when measured
in terms of frequencies instead of wave-lengths. This is now known as
Hartley’s law. This same year Liveing and Dewar3 announced their
discovery of series in thallium, zinc, and aluminum.4
1.3. Balmer’s Law.—By 1885 the hydrogen series, as observed in the
spectra of certain types of stars, had been extended to 14 lines. Photo¬
graphs of the hydrogen spectrum are given in Fig. 1.5. This year is
significant in the history of spectrum analysis for at this early date
Balmer announced the law of the entire hydrogen series. He showed
that, within the limits of experimental error, each line of the series is
given by the simple relation

(1.1)
where h = 3645.6 k and ni and n2 are small integers. The best agree¬
ment for the whole series was obtained by making nx = 2 throughout
1 Liveing, G. D., and J. Dewar, Proc. Roy. Soc. London, 29, 398, 1879.
* Hartley, W. N., Jour. Chem. Soc., 43, 390, 1883.
* Liveing, G. D., and J. Dewar, Phil. Trans. Roy. Soc.f 174, 187, 1883.
4 A more complete and general account of the early history of spectroscopy is to
be found in Kayser, “Handbuch der Spektroscopie, Vol. I, pp. 3-128, 1900.
Sec. 1.4] EARLY HISTORICAL DEVELOPMENTS 5
and n2 ® 3,4,5,6 • • • for the first, second, third, fourth, • • • members
of the series. The agreement between the calculated and observed
values of the first four lines is shown in the following table:
Table 1.1.—Balmer's Law for the Hydrogen Series

Angstrom's
Calculated wave-lengths Difference
observed values

Ha = P= 6562.08 A 6562.10 A +0 02 A
Hp = iJA = 4860 80 4860.74 -0.06
Hy = |*fc = 4340 00 4340.10 4-0 10
Hi = |p = 4101 30 4101.20 -0 10

While the differences between calculated and observed wave-lengths

for the next 10 lines are in some cases as large as 4 A, the agreement is as
good as could be expected from the existing measurements. The

Fig. 1.5.—Stellar and solar spectrograms showing the Balmer series of hydrogen.

accuracy with which lines were then known is seen from the following
measurements of five different investigators for the first member of the
hydrogen series:
Table 1.2.—Early Measures of the Hydrogen Line H«
6565 6 . Van der Willigen
6562 10 . Angstrom
6561.62... MendenhaU
6560 7. Mascart
6559 5 . Ditscheiner

1.4. Rydberg’s Contributions.—Rydberg’s early contributions to

atomic spectra consisted not only in finding the laws of a number of
series but also in showing that Hartley’s law of constant frequency
differences was applicable in cases where the doublet or triplet com¬
ponents were very far apart. For example, he showed that the first
member of the principal series of thallium is a doublet with the enormous
6 INTRODUCTION TO ATOMIC SPECTRA [Chap. I

separation of 1574 A, one line lying in the green at X5350 and the other
in the near ultrarviolet at X3776. This discovery proved to be of con¬
siderable importance, as it suggested the possibility that all series arising
from the same element were in some simple way connected with each
other.
Using Liveing* and Dewar’s data on the sodium and potassium
series, Rydberg made for the first time the distinction between what is
called a sharp series and a diffuse series. The Na and K series given in
Fig. 1.4 were each shown to be in reality two series, one a series of sharp
doublets (designated by s) and the other a series of diffuse doublets (des¬
ignated by d).
A third type of series found in many spectra is the so-called principal
series, involving as its first member the resonance or persistence line of the
entire spectrum. Resonance or persistence lines are those relatively
strong lines most easily excited. Such lines usually appear strong in a
Bunsen flame. The yellow D lines are a good example of this.
Still a fourth type of series was discovered by Bergmann. This type
of series is usually observed in the near infra-red region and is sometimes
called the Bergmann series. Since Bergmann did not discover all such
series, Hicks called them the fundamental series. Although this name is
perhaps less appropriate than the other and is really misleading, it has
after 20 years become attached to the series. New series were dis¬
covered so rapidly about this time that many different names and
systems of notation arose. Three of these systems commonly used are
given in the following table.

Table 1.3.—Series Notation Used by Different Investigators

Foote and Mohler Paschen Rydberg

Second subordinate . II. Nebenserie Sharp

Principal . . Hauptserie Principal
First subordinate. I. Nebenserie Diffuse
Bergmann. Bergmann Fundamental

In searching for a general series formula Rydberg discovered that

if the wave-lengths X or the frequencies v of a series be plotted against
consecutive whole numbers, a smooth curve which is approximately a
displaced rectangular hyperbola is the result.1 For a curve such as the
one shown in Fig. 1.6 he attempted a solution of the form

1 It is convenient in practice to express v numerically not as the actual frequency

but as the frequency divided by the velocity of light. Such quantities are called
wave numbers and, since c — v\, they are given by the reciprocal of the wave-length
measured in centimeters. In this way a spectrum line is described by the number of
waves per centimeter in vacuo. Wave numbers, therefore, are units with the dimen¬
sions of reciprocal centimeters, abbreviated cm*"1.
Sec. 1.4] EARLY HISTORICAL DEVELOPMENTS 7

— —t—’ n = 2, 3, 4, 5, • • • oo, (1.2)

where rn is the wave number of the given line n, and C, and m are
constants. In this equation, approaches v . as a limit, as n approaches
infinity. While this formula did not give the desired accuracy for an
entire series, Rydberg was of the opinion that the form of the equation
could not be far from correct.
The next equation investigated by Rydberg was of the form

"» = "» - (n + m)2’ n = 2, 3, 4, 5, • • • oe, (1.3)

where, as before, JV and ^ are constants, v«, is the limit of the series, and
n is the ordinal number of the line in the series. This formula proved

Fig. 1.6.—Frequency plot of the principal series of lithium.

to be so successful when applied to many series that Rydberg adopted

it in all of his succeeding work. The important fact about the formula
is that N takes the same value in all series in all elements. It is interest¬
ing to observe that, by placing n = 0, the equation reduces to Balmer’s
formula for hydrogen (see Sec. 1.8). Since n takes on integral values
only, the hydrogen series affords a direct means of calculating this
universal constant N, now known as the Rydberg constant R. Rydberg’s
equation is now written

Vn = V. - Kn = 2, 3, 4, 5, • • • ». (1.4)

If we now let v%, vt, and v{ represent vn for the sharp, principal,
diffuse, and fundamental series, respectively, then the four general
series may be represented by
Sharp series:
R
where n = 2, 3, 4, • • • (1.5)
(n + S)2’
8 INTRODUCTION TO ATOMIC SPECTRA [Chap. I

Principal series:
R
K = "£ (n + py> where n - 1, 2, 3, • •1 • CO . (1-6)

Diffuse series:
R
< = vt - (n + Dy> where n - 2, 3, 4, • - • 00. (1.7)

Fundamental series:
R
K - vl - ^ + Fy< where n - 3, 4, 5, • • • oo. (1.8)

Here S, P, D, and F represent the values of and v l, v&, vt, and the
limits of the different series. In applying the above equations to the
three chief series of lithium, Rydberg obtained the following expressions:

109721.6
K - 28601.6 - (1.9)
(ti + 0.5951) 2‘
109721.6
K = 43487.7 - (1.10)
(n + 0.9596)2
109721.6
28598.5 (1.11)
(n + 0.9974)2

These equations show a relation that might have been anticipated

from an examination of the sharp and diffuse series of Na and K in Fig.
1.4, viz., that the limits of the S and D series are probably the same,
i.e., vl> “ vi. It is to be noted that the first member of the sharp series,
Eq. (1.9), starts with n = 2. With n = 1 this formula gives the first
member of the principal series with a negative value.
1.5. The Rydberg-Schuster Law.—Rydberg observed that if n is
placed equal to unity in the sharp-series formula, Eq. (1.9), the right-
hand term became approximately the limit of the principal-series formula,
Eq. (1.10).
109721.6
43123.7 ^ v%. (1.12)
(1 + 0.5951)2

Similarly, if n is placed equal to unity in the principal-series formula,

the right-hand term becomes approximately the limit of the sharp
series:
109721.6
28573.1 vl. (1.13)
(1 + 0.9596)2

Rydberg came to the conclusion that the calculated differences were

due to experimental errors and that the sharp- and diffuse-series limits
were identical. He therefore could write Eqs. (1.9), (1.10), and (1.11)
in the following form:
Sec. 1.6] EARLY HISTORICAL DEVELOPMENTS 9
R R
(1.14)
(l+PY (» + SY■
R R
(1.15)
(1 + -S)* (n + PY'
R R
(1.16)
(l + PY (n + DY
It was not until 1896 that Rydberg1 discovered, as did Schuster2 in
the same year, that the first line of the principal series in the alkalies
is also the first member of the sharp series when taken with negative sign.
If n is placed equal to unity in Eqs. (1.9) and (1.10), then v\ = — v\. In
exactly this same way it was found that the difference between the limit
of the principal series and the common limit of the sharp and diffuse
series is equal to the first member of the principal series:
vpK — vl = v\ = (1.17)
These important relations are now known as the Rydberg-Schuster law.
Immediately upon the discovery by Bergmann, 12 years later, of the
fundamental series, Runge announced that the difference between the
limits of the diffuse and fundamental series is equal to the first line of
the diffuse series:
- vio = A' (1.18)

Hence, from Eqs. (1.7) and (1.18), Eq. 1.8 can be written

, _ R R
(1.18a)
'* (2 + DY (n + FY’

and this equation added to the group (1.14), (1.15), and (1.16) shows
that the frequency limit of every one of the four series has now been
expressed in terms of the constants of some other series. For the diffuse
series, n usually starts with two in place of unity.
The Rydberg-Schuster law as well as the* Runge law is shown in
Fig. 1.7 by plotting the spectral frequencies, in wave numbers v, against
the order number n, for the four chief series of sodium. To prevent
confusion among the lines belonging to the different series each one is
plotted separately. In order to show that the first member of the
principal series becomes the first member of the sharp series, when taken
with negative value, the scale has been extended to negative wave
numbers. Unfortunately the frequency scale is too small to show the
doublet nature of each line. The Rydberg-Schuster law is indicated
by the intervals X\ of the figure and the Runge law by the intervals Xz.
A study of the singlet and triplet series discovered in a number of
elements shows that similar laws are to be found. This may be illus-
1 Rydberg, J. R., Astrophye. Jour., 4, 91, 1896.
* Schuster, A., Nature, 55, 180, 200, 223, 1896.
10 INTRODUCTION TO ATOMIC SPECTRA [Chap. I

trated by the calcium series shown schematically in Fig. 1.8. The

upper four series represent the triplets and the lower four the singlets.
Intervals between certain series limits, shown at the bottom of the figure,
zrvinn 9nnnn tnnnn n -mnnn

Fla. 1.7.—Schematic plot of the four chief series of sodium doublets showing the Rydberg-
Schuster and Runge laws

are observed to be the same as the frequencies of certain radiated lines.

Figures such as these reveal many important relations and, as we shall
see later, facilitate the formation of what are called energy level diagrams.

1.6. Series Notation.—A somewhat abbreviated notation for Ryd¬

berg’s formulas was employed by Ritz. This abbreviated notation,
which follows directly from Eqs. (1.14), (1.15), (1.16), and (1.18a), for
the four chief types of series is written as follows:
Sec. 1.7J EARLY HISTORICAL DEVELOPMENTS 11

= IP — nSf v* =* IP — nD.
yp = IS — nP, j/£ = 2D — nP. (1.19)

In order to distinguish between singlet-, doublet-, and triplet-series

systems, various schemes vhave been proposed by different investigators.
Fowler,1 for example, used capital letters Sf P, D, etc., for singlets, Greek
letters <r, ir, 5, etc., for doublets, and small letters $, p, d, etc., for triplets.
Paschen and Gotze2 on the other hand adopted the scheme of small
letters s, p, d for both doublets and triplets, and capitals S, P, D for
singlets.
A more recent scheme of spectra notation, published by Russell, Shen-
stone, and Turner,3 has been accepted internationally by many investi¬
gators. In this new system capital letters are used for all series and small
superscripts in front of each letter give the multiplicity (see Table 1.4).
Table 1.4.—Series Notation

Series Fowler Paschen Adopted

Singlet. S P D F S P D F lS 'P 'D lF

Doublet. a ir 5 ip s p d f *S *P *D *F
Triplet. 8 p d f 8 V d f 3S *P >D V

1.7. Satellites and Fine Structure.—The appearance of faint lines,

or satellites, in some series was discovered by Rydberg and also by
Kayser and Runge. At first the satellites, which usually appear on the
long wave-length side of the diffuse series, were considered as a secondary
diffuse series until it was discovered that Hartley’s law of constant
frequency separations applied to the separation of the satellite and one
of the strong lines of the doublet. In most diffuse doublets the strongest
line lies between the satellite and the weaker line of the doublet.
This is illustrated in Fig. 1.9, where the first four members of the four
chief series of caesium are plotted schematically. Plotted to a frequency
scale it is seen that the outside separation of each member of the sharp
and diffuse series and the first member of the principal series is exactly
the same. It should also be noted that the first member of the principal
series is, when inverted, the first member of the sharp series. This mem¬
ber is shown dotted.
Unlike other 2F series, to be studied later, the strongest line in each
P-doublet in caesium lies on the low-frequency side of the respective
satellites. Like all 2F series, however, each doublet has one interval in
common with the first member of the diffuse series.
1 Fowler, A., “ Report on Series in Line Spectra,” Fleetway Press, 1922.
2 Paschen, F., and R. GcVtze, “Seriengesetze der Linienspektren,” Julius
Springer, 1922.
3 Russell, H. N., A. G. Shenstone, and L. A. Turner, Phys. Rev., 33, 900, 1929.
12 INTRODUCTION TO ATOMIC SPECTRA [Chap. I

Similar relations are found to exist among the triplet series of the
alkaline earths. The first four members of the four chief series of
calcium are shown schematically in Fig. 1.10. From the diagram it is
seen, first, that the common limit of sharp and diffuse series is a triple

Limits 4th 3rd 2nd 1st

Fia. 1.9.—Frequency plot of the doublet fine structure in the chief series of caesium.

limit with separations equal to the first member of the principal series;
second, that the limit of the fundamental series is a triple limit with
separations equal to the separations of the strongest line and satellites
of the first member of the diffuse series; and third, that the principal
series has a single limit. It is to be noted in the doublet series (Figs.

Limits 4th 3rd 2nd 1st

Fig. 1.10.—Frequency plot of the triplet fine structure in the chief series of calcium.

1.9 and 1.7) that, while the first principal doublet becomes the first
sharp doublet when inverted, the reverse is true for the triplet series of
calcium. By inverting the first sharp-series member (Fig. 1.10), the
lines fall in with the principal series in order of separations and intensities.
Sue. 1.7] EARLY HISTORICAL DEVELOPMENTS 13
In the development of Rydberg’s formula, each member of a series
was assumed to be a single line. In the case of a series where each
member is made up of two or more components, the constants vK and p
of Eq. (1.4) must be calculated for each component. Rydberg’s formulas
for the sharp series of triplets, for example, would be written, in the
accepted notation,
R R
i13r
P.s _ n
n3.Cf,
o i,
= a + sp»)2 ” (n + *Si)2
R R 13p,
iii _ n3Q,
n (1.20)
“a + ‘Pi)1 “ (n + ‘Si)1
R R
AFPn
X 0 — n3Si
AJlj
=a + SPo)J + ‘Si)1

where 3$i, 8Po, 3Pi, 3Pzf occurring in the denominators, are small con¬
stants. Symbolically 13P2 stands for the term R/{ 1 + 8P2)2 which is
one of the three limits of the sharp series. The subscripts 0, 1, and 2,
used here to distinguish between limits, are in accord with, and are part
of, the internationally adopted notation and are of importance in the
theory of atomic structure.
A spectral line is seen to be given by the difference between two
terms and a series of lines by the difference between one fixed term and a
series of running terms. The various components of the diffuse-triplet
series with three main lines and three satellites are designated
1SP2 — n3D3f first strong line; l3Pi — n3P2, second strong line.
13P2 — n8D2, satellite; l3Pi — n*Di, satellite. (1.21)
13P2 — n8£>i, satellite; l3Po ~ n3Z>i, third strong line.

The general abbreviated notation of series terms is given in the

following table along with the early schemes used by Fowler and Paschen.

Table 1.5.—Notation of Series Terms

Internationally
Series Fowler Paschen
adopted

Singlet. S P D F 8 P D F 'So 1Pi 'Dt 'Ft

Doublet. <r X2 6 <p s pt d% ft •Si 'Pi V,

xi S' <p' Pi di fi *jP, *i>| 'Fi

Triplet... s Pi d f 8 pi di fi *Si 'P, 'Di 'Fi

p, d’ r Pi di ft •Pi 'Dt 'F,
p. d" r Pt dt ft •P, 'Di 'Ft

While the new notation is somewhat complicated by the use of

superscripts and especially by half-integral subscripts, it will be seen
14 INTRODUCTION TO ATOMIC SPECTRA [Chap. I

later that each letter and number has a definite meaning, in the light of
present-day theories of atomic structure.
1.8. The Lyman, Balmer, Paschen, Brackett, and Pfund Series of
Hydrogen.—It is readily shown that Balmer's formula given by Eq. (1.1)
is obtained directly from Rydberg's more general formula

R_R
(1.22)
Vn (ni + mi) 2 (n2 + M2)2

by placing mi = 0, M2 ~ 0, m = 2, and n2 = 3, 4, 5, • • • . Inverting

Eq. (1.1),
1 J 1 _ 1 n? =
(1.23)
X a an\ Vnt

where a = 3645.6 A. Writing R = n\/ay

(1.24)

This is the well-known form of the hydrogen-series formula.

It was Ritz, as well as Rydberg, who made the suggestion that n2
might take running values just as well as ni. This predicts an entirely
different series for each value assigned to n 1. For example, with n 1 = 1,
2, 3, 4, and 5, the following formulas are obtained:

Lyman series:

» where n2 = 2, 3, 4, • • • . (1.25)
Balmer series:

> where n2 = 3, 4, 5, • • • . (1.26)

Paschen series:

> where n2 = 4, 5, 6, • • • . (1.27)

Brackett series:

» where n2 = 5, 6, 7, • • • . (1.28)
Pfund series:

vn = > where n2 * 6, 7, 8, • • • . (1.29)

Knowing the value of R from the well-known Balmer series the positions
of the lines in the other series are predicted with considerable accuracy.
The first series was discovered by Lyman in the extreme ultra-violet
region of the spectrum. This series has therefore become known as the
Lyman series. The third, fourth, and fifth series have been discovered
Sec. 1.9] EARLY HISTORICAL DEVELOPMENTS 15
where they were predicted in the infra-red region of the spectrum by
Paschen, Brackett, and Pfund, respectively. The first line of these five
series appears at X1215, X6563, X18751, X40500, and X74000 A, respec¬
tively. It is to be seen from the formulas that the fixed terms of the
second, third, fourth, and fifth equations are the first, second, third, and
fourth running terms of the Lyman series. Similarly, the fixed terms of
the third, fourth, and fifth equations are the first, second, and third
running terms of the Balmer series; etc. This is known as the Ritz
combination principle as it applies to hydrogen.
1.9. The Ritz Combination Principle.—Predictions by the Ritz
combination principle of new series in elements other than hydrogen
have been verified in many spectra. If the sharp and principal series of
the alkali metals are represented, in the abbreviated notation, by

Sharp: vn = VP — n2S, where n = 2, 3, 4, • • • , .

Principal: vu = 12S — n2P, where n = 2, 3, 4, • • * ,

the series predicted by Ritz are obtained by changing the fixed terms
12P and VS to 22P, 32P and 22£, 32£, etc. The resultant formulas are
of the following form:
Combination sharp series:

22P — n2Sy where n = 3, 4, 5, • • • .

(1.31)
32P — n2S, where n = 4, 5, 6, • • • .
etc.

Combination principal series:

22S — n2P, where n — 3, 4, 5, • • • .

(1.32)
32S — ?i2Py where n = 4, 5, 6, • • • .
etc.

In a similar fashion new diffuse and fundamental series are predicted

by changing the fixed 2P term of the diffuse series and the fixed 2D term
of the fundamental series.
Since ail fixed terms occurring in Eqs. (1.31) and (1.32) are included
in the running terms of Eqs. (1.30), the predicted series are simply com¬
binations, or differences, between terms of the chief series. Such series
have therefore been called combination series. Extensive investigations
of the infra-red spectrum of many elements, by Paschen, have led to the
identification of many of the combination lines and series.
In the spectra of the alkaline earth elements there are not only the
four chief series of triplets 8$, 3P, 3D, and 3P but also four chief series of
singlets lS, IP, 1Df and lF. Series and lines have been found not only
for triplet-triplet and singlet-singlet combinations but also triplet-singlet
and singlet-triplet combinations. These latter are called intercombination
lines or series.
16 INTRODUCTION TO ATOMIC SPECTRA [Chap. I

1.10. The Ritz Formula.—The work of Ritz on spectral series is

of considerable importance since it marks the development of a series
formula still employed by many investigators. Assuming that Rydberg's
formula for hydrogen was correct in form

’•-«(?- ?) <i-33>
and realizing that p and q must be functions involving the order number
n, Ritz obtained, from theoretical considerations, p and q in the form of
infinite series:
, , b\ , Ci , di ,
V = n1 + a1 + ^ + ^ + ^+ -- ‘
(1.34)
. . &2 , C2 . di
Q = tt2 + a2 + -2 + -^ + -6 + •

Using only the first two terms of p and q, Ritz’s equation becomes
identical with Rydberg’s general formula which is now to be considered
only a close approximation. In some cases, the first three terms of the
expansion for p and q are sufficient to represent a series of spectrum lines
to within the limits of experimental error.
1.11. The Hicks Formula.—The admirable work of Hicks1 in develop¬
ing an accurate formula to represent spectral series is worthy of mention
at this point. Like Ritz, Hicks starts with the assumption that Ryd¬
berg’s formula is fundamental in that it not only represents each series
separately but also gives the relations existing between the different
series. Quite independent of Ritz, Hicks expanded the denominator of
Rydberg’s Eq. (1.3) into a series of terms

M +
1
+
1
_a +. _
&
+
, c +
,
(1.35)

The final formula becomes

R
VH = Voo —
(1.36)
(n
+ '‘ + S + i? + £ + T
This formula, like Ritz’s, reduces to Rydberg’s formula when only
the first two members in the denominator are used.
1.12. Series Formulas Applied to the Alkali Metals.—The extension
of the principal series of sodium to the forty-seventh member by Wood2
(see Fig. 1. 11) and of the principal series of potassium, rubidium, and
caesium, to the twenty-third, twenty-fourth, and twentieth members,

1 Hicks, W. M., Phil. Tram. Roy. Soc., A, 210, 57, 1910; 212, 33, 1912; 213, 323,
1914;217, 361, 1917; 220, 335, 1919.
* Wood, R. W., Astrophys. Jour., 29, 97, 1909.
Sec. 1.12] EARLY HISTORICAL DEVELOPMENTS 17

respectively by Bevan1 furnishes the necessary data for testing the

accuracy of proposed series formulas. A careful investigation of these
series was carried out by Birge2 who found that the Ritz formula was to
be preferred and that with three undetermined constants it would
represent the series of the alkali metals of lower atomic weight with fair
accuracy. Birge shows that the number of terms that need be used in

Principal Series of Sociium

t ■wm 111111 i i r '■ r.*"T i r ' ..r i

Series Limit 64 75 2490 12 43 2593 —A—** 2680 A
Fig. 1.11.—Principal series of sodium in absorption. (After Wood.)

the denominator depends directly on the size of the coefficients of the

several terms, and that these coefficients increase with atomic weight.
This increase is shown in the following table:

Table 1.6.—Series Coefficients

(After Birge)1

1 Biroe, H. T., Aatrophya Jour , S2, 112. 1910.

To illustrate the accuracy ^ith which the Ritz formula represents

series in some cases, the principal series of sodium is given in the table
shown on page 18.
The Rydberg constant as calculated by Birge from the first five
members of the Balmer series of hydrogen and used by him for all series
formulas is R = 109678.6. It is to be noted that the maximum error
throughout the table is only 0.1 A.
This work greatly strengthened the idea that the Rydberg constant
was a universal constant and that it was of fundamental importance in
series relations. The Ritz equation has therefore been adopted by many
investigators as the most accurate formula, with the fewest constants,
for use in spectral series. A modified but equally satisfactory form of
the Ritz formula will be discussed in Sec. 1.14.

1 Bevan, P. V., Phil. Mag., 19, 195, 1910.

* Birge, R. T.f Astrophys. Jour., 32, 112, 1910.
18 INTRODUCTION TO ATOMIC SPECTRA [Chap. I

Table 1.7.—The Ritz Formula Applied to the Sodium Series Observed by Wood
(Calculations after Birge)'

where A = 41,450.083 cm"1, R = 109,678.6 cm-1, a = 0.144335, and 6 = -0.1130286-

1
Calc. Calc. Calc.
n Xvau Ohs. n Xyic obs. ! ft Xvar Obs.
diff. diff. diff.

2 5897 563 0.00 18 2432 08 -0 01 34 2418 03 -1 0 01

3 3303 895 0 00 19 30 07 -0 02 35 17 75 -t0 02
4 2853 63 -0 04 20 28 37 -0 01 36 17 45 0 00
5 2681 17 +0 09 21 26 93 +0 02 37 17 21 -t0 03
6 2594 67 +0 09 22 25 65 +0 03 38 16 98 -10 04
7 2544 49 -0 05 23 24 53 -to01 39 16 76 -t0 03
8 2512 90 +0 07 24 23 55 000 40 16 54 +0 02
9 2491 36 -0 04 25 22 69 000 41 16 35 -tO 03
10 76 26 +0 03 26 21 93 -to02 42 16 17 -1 0 02
11 65 18 40 10 27 , 21 25 -to01 43 16 02 4 0 03
12 56 67 +0 05 28 | 20 67 +0 04 44 15 86 4 0 04
13 50 11 -fO 04 j 29 20 15 +0 07 45 15 71 40 03
14 44 89 i -0 03 | 30 19 65 +0 06 46 15 59 4 0 05
15 40 71 +0 01 31 19 09 -o 05 ; 47 15 43 0 01
16 37 35 +0 06 32 18 74 0 00 48 15 29 -0 01
17 34 50 | +0 04 33 18 36 -0 01 49 15 15 -0 04

J Biikjb, It. T. Antrophi/H. Jour , 32, 112, 1910

1.13. Neon with 130 Series.—The first successful analysis of a xeally

complex spectrum was made by Paschen1 in the case of neon. Although
the neon spectrum was found to contain a great many lines, Paschen
able to arrange them into 130 different series. These series, classified
as 30 principal series, 30 sharp series, and 70 diffuse series, were found to
be combinations between 4 series of S terms, slf s2, s3) and s4; 10 series of
P terms, p\y p2} Pz, . • . , Pio) and 12 series of D terms, d\9 d2f d3f . . . ,
r/12. Paschen showed that, while many of the series were regular and
followed a Ritz formula, others were irregular and could not be fitted to
any formula. These abnormal series will be discussed in the following
section.
1.14. Normal and Abnormal Series.—Occasionally it is found that
certain members of a wrell-established series do not follow' the ordinary
Rydberg or Ritz formula to within the limits of experimental error.
Well-know'n series of this kind were pointed out by Saunders in the
singlet series of Ca, Sr, and Ba, and by Paschen in certain neon series.
A convenient method, employed by Paschen and others, for illustrating
deviations from a normal series is to plot /*, the residual constant in the
Rydberg denominator, of each term against n, the order number of the
1 Paschen, F., Ann. d. Phys., 60, 405, 1919; 63, 201, 1920.
Sec. 1.14] EARLY HISTORICAL DEVELOPMENTS 19
term. Several of the diffuse series of terms of neon as given by Paschen
are reproduced in Fig. 1.12. A normal series should show residuals that
follow a smooth curve like the first nine members of the d« series. The
curves for the di, d8, and d5 terms show no such smoothness, making it
very difficult to represent the series by any type of series formula.
A series follow ing a Rydberg formula is represented on such a graph
by a horizontal line, i.e., n constant. With a normal series like di of

Fio. 1.12.—Four diffuse series in neon showing normal and abnormal progression of the
residual m- (After Paachen.)

neon, a Ritz formula with at least one added term is necessary to ade¬
quately represent the series. If Tn represents the running term of a
Ritz formula, and Ti the fixed term,
R
vn Tl — Tn = Ti (1.37)
(n + a +

Another useful form of the Ritz formula is obtained by inserting the

running term itself as a correction in the denominator:
R
Tn (1.38)
(n + a + bTn)2
This term being large at the beginning of a series, the correction is
correspondingly large. Formulas representing abnormal series like
di, d3, and d5 in Fig. 1.12 will be treated in Chap. XIX.
Other anomalies that frequently occur in spectral series are the
irregular spacings of the fine structure in certain members of the series.
A good example of this type of anomaly is to be found in the diffuse
triplet series of calcium. In Fig. 1.13 a normal diffuse series, as is
observed in cadmium, is shown in contrast with the abnormal calcium
series. The three chief lines of each triplet are designated a, 6, and d,
and the three satellites c, e, and /. Experimentally it is the interva
20 INTRODUCTION TO ATOMIC SPECTRA (Chap. I

between the two satellites / and c and the interval between the satellite
c and the chief line a that follows Hartley’s law of equal separations in
both series. In cadmium it is seen that the main lines and satellites
converge toward the three series limits very early in the series. In
calcium, on the gther hand, the lines first converge in a normal fashion,
then spread out anomalously and converge a second time toward the
three series limits. These irregularities now have a very beautiful
explanation which will be given in detail in Chap. XIX.
.0 —

a d e f a be d e f
u -

_J_ 1 i , 1 J_Li_1 i ■
i_L_L_ 2 l L_ k
1 l II 3 I 1. Ii.
I l L 4 1 1 k
_I_ I. II 5 . 1 It It
-1 ii_Lu_ 6 i l . k
7 i L l
L 8 i L l
. I L Ii. 9 i k l
. i L 1 10 _i__l__1_
i l l 11 _l_ i

| Limits j Limits
l_ oo L_
Abnormal Series of Calcium Normal Series of Cadmium
Triplets Triplets

Fig. 1.13.—Diffuse series of triplets in cadmium and calcium.

1.16. Hydrogen and the Pickering Series.—In the hands of Balmer

and Rydberg the historical hydrogen series was well accounted for when
Pickering, in 1897, discovered in the spectrum of the star f-Puppis a
series of lines whose wave-lengths are closely related to the Balmer series
of hydrogen. Rydberg was the first to show that this new series could
be represented by allowing n2 in Balmer’s formula to take both half and
whole integral values. Balmer’s formula for the Pickering series may
therefore be written

where n2 = 2.5, 3, 3.5, 4, 4.5, • • • . (1.39)

The Balmer and Pickering series are both shown schematically in Fig.
1.14. So good was the agreement between calculated and observed
wave-lengths that the Pickering series was soon attributed to some new
form of hydrogen found in the stars but not on the earth.
Sue. 1.16] EARLY HISTORICAL DEVELOPMENTS 21

Since n» was allowed to take on half-integral values, Rydberg predicted

new series of lines by allowing »i to take half-integral values. One series,
for example, could be written

?» = where n2 = 2, 2.5, 3, • • • . (1.40)

All of the lines of this predicted series, except the first, are in the ultra¬
violet region of the spectrum. With the appearance of a line in the
spectrum of f-Puppis at X4688, the position of the first line of the pre¬
dicted series, Rydberg’s assumption was verified and the existence of a
new form of hydrogen was (erroneously) established.
He Hi Hr H* Ha

3500 4000 4500 5000 5500 4000 4500 A

Fia. 1.14.—Comparison of the Balmer series of hydrogen and the Pickering series.

Fowler in his experiments on helium brought out, with a tube con¬

taining helium and hydrogen, not only the first two members of the
Pickering series strongly but also a number of other lines observed by
Pickering in the stars. While all of these lines seemed to be in some way
connected with the Balmer formula for hydrogen, they did not seem to
be in any way connected with the known chief series of helium. The
whole matter was finally cleared up by Bohr1 in the extension of his
theory of the hydrogen atom to ionized helium. This is the subject taken
up in Chap. II.
1.16. Enhanced Lines.—Spectral lines which on passing from arc to
spark conditions become brighter, or more intense, were early defined
by Lockyer as enhanced lines. In the discovery of series relations
among the enhanced lines of the alkaline earths, Fowler2 made the dis¬
tinction between three classes of enhanced lines; (1) lines that are strong
in the arc but strengthened in the spark, (2) lines that are weak in the
arc but strengthened in the spark, and (3) lines that do not appear in the
arc at all but are brought out strongly in the spark.
Fowler discovered, in the enhanced spectra of Mg, Ca, and Sr, series
of doublet lines corresponding in type to the principal, sharp, and diffuse

1 Bohr, N., Phil. Mag., 26, 476, 1913.

’Fowler, A., Phil. Trans. Roy. Soc.y A, 214, 225, 1914.
22 INTRODUCTION TO ATOMIC SPBCTRa\^ ^ ' [Chap. I

doublets of the alkali metals. In an attempt to represent these series

by some sort of Rydberg or Ritz formula, it was found that n%, the order
number of the series, must take on half-integral as well as integral values.
The situation so resembled that of the Pickering series and the hydrogen
series that Fowler, knowing the conditions under which the enhanced
lines were produced, associated correctly the enhanced doublet series of
Mg, Ca, and Sr with the ionized atoms of the respective elements. For
such series we shall see in the next two chapters that the Rydberg con¬
stant R is to be replaced by 4R so that the enhanced series formula
becomes

* = 4iJ{(n, + mi )2 “ (n, + mi)2}' (L41)

where n* is integral valued only.

Problem
With the frequencies of the four chief series of spectrum lines as given for ionized
calcium by Fowler, “Series in Line Spectra,” construct a diagram similar to the one
shown in Fig. 1.7. Indicate clearly the intervals illustrating the Rydberg-Schuster
and Runge laws.
that has been made in atomic structure. Whether any proposed atomic
model is correct or not, it now appears that energy levels and energy
level diagrams will always remain with us.
77
78 INTRODUCTION TO ATOMIC SPECTRA [Chap. V

5.1. Energy Level Diagrams.—From a historical standpoint it is

of interest to note how nearly the energy level diagram came to being
discovered before the advent of the quantum theory. Returning to the
frequency plot of the four chief series of spectrum lines of sodium (Fig.
1.7), the Rydberg-Schuster law makes it possible to construct an energy
level diagram by turning the figure sideways and raising each series until
the limits fall together. Each spectrum line now becomes an energy
level and the diagram takes the form shown at the right in Fig. 5.1.
The lowest level is a sharp or S level, followed by a principal or P level,
then another sharp level S, a diffuse level D, etc. Thus the energy level
diagram of sodium consists of a series of sharp, a series of principal, a
series of diffuse and a series of fundamental energy levels approaching a
common limit.
The diagram at the left in Fig. 5.1 may be obtained in exactly the
same way from the four chief series of spectrum lines observed in lithium.
The superscript in the level notation indicates that the level in question
belongs to a doublet system and that it has certain doublet characteristics.'
Just as in hydrogen these energy levels represent certain possible
energy states of an electron, and transitions between them represent
spectrum lines. The energy levels n = 1, 2, 3, • • • of hydrogen are
given, for comparison, on the term-value scale in the center of the figure-
The words term or term value, often used in place of energy level or state,
have been retained from the early use of the terms in the Rydberg and
Ritz formulas treated in Chap. I. The fixed term of these series formulas
[see Eqs. (1.4) and (1.37)] corresponds to a final or lower energy level com¬
mon to a given series, and the running terms correspond to the different
initial levels, from which the series gets its name, sharp, principal, diffuse,
or fundamental. The first member of the principal series of lithium, for
example, is one in which the final state is the lowest level 2S with a term
value [see Eq. (1.10)] of 43487.7 cm"1, and the initial state, neglecting
the fine structure, is the first excited level 2P with a term value of 28582.5
cm-1.
Just as the energy level diagrams of Fig. 5.1, may be obtained directly
from the observed series of spectrum lines as shown in Fig. 1.7, so may
energy level diagrams for the alkaline earths be obtained from the
observed series of lines as given for calcium, for example, in Fig. 1.8.
Turning this figure sideways and raising each series until the limits fall
together, an energy level diagram is obtained with a singlet sharp level
at the bottom. These more complicated diagrams will be taken up in a
later chapter. With this brief introduction to energy level diagrams we
shall now turn to Bohr's Avfbauprinzip.
1 The doublet fine structure of certain of the energy levels and spectrum lines is
attributed to the spinning of the electron. This fine structure will be neglected in
this chapter but will be taken up in detail in Chap. VIII.
Sue. 5.2] THE ALKALI METALS AND THE PERIODIC TABLE 79

5.2. The Bohr-Stoner Scheme of the Building Up of the Elements.—

Before taking up in detail the well-known spectra of the alkali metals it is
important that we first consider the Bohr-Stoner scheme of the building
up of the elements in the periodic table. The periodic table, as it is now
frequently formulated, is given in Table 5.1.
The number above each element gives the atomic number Zf and the
symbols below give the lowest energy level for that atom. Referring
to Fig. 5.1, for example, the lowest level of lithium and sodium is a sharp
level designated as 2S. The meaning of the superscripts and the sub¬
scripts will be given in detail later.
Table 5.1

E9
First
Period PERIODIC TABLE

■IHI
m

n i m ■fl H El
Second
Period

n HEln H Hi
Third 18
Period Nil A
zSw ‘So

12gllrtllfSlgig 1
m
►

ii|iI I ft Ip El
Fourth 25 26 27 28 29 30 31 32 33
Period 2
Fe Co Ni Cu Zn Go Ac
*D4 4FVi ’So % *P#

T$i IIE053II
'5(i 4Stt
.37 38 39 40 41 42 43 48 49 50 51
Fifth
Period 1lb Sr
S* ’S.
155 56
y

57 72
Zr Cb
4 *Fi
73
Ii
*Dvk
Mo Mo

74 75 76 77 78 79
Cd
•Sb
80
In

81
Sn
%
82
Sb
4S*
83 84 85
m
86
Sixth
4 Ba
JeL £&
m1III I■■ ■■■■ 1■■■ 1 | I
Period ' Lfl “*rHf To W Jr Pt Au Pb Bi Pb Rn
s*
*svi lD *h. •Sn 5Pi
% _!5b_
Seventh i
Period J

wmmmm 2GZ3 □ fin'EDI

m nui mm
inn m Bin ■■■

HIEHll
60 61 62 63 64 65 66 67 70 71
Hd 11
at
•L iyjt
Sm
rHU
Eu
8Hyi
Tb
®HfW ft
Ho
i
Yb
H*
Lu
*Dyi

Making the assumption that the number of electrons around the

nucleus of a neutral atom is equal to the atomic number Z, and that each
of these electrons takes on a definite quantized state, Bohr put forward
his Aufbauprinzip of building up the elements. With quantum numbers
assigned to each electron each atom Z in the periodic table is, so far as
the extranuclear electrons are concerned, formed by adding one more
electron to the atom Z — 1. The quantum numbers to be assigned to
this last bound electron are such that the electron in this state is in the
most tightly bound state possible. With such a scheme the atom will
have the least amount of energy possible.
Rydberg was the first to show that the atomic numbers of the inert
gases He, Ne, A, Kr, Xe, and Rn, which are given in the last column of
Table 5.1, are given by the following formula:
80 INTRODUCTION TO ATOMIC SPECTRA [Chap. V

He Ne A Kr Xe Rn
Z = 2(12 + 22 + 22 + 32 + 32 + 42) (5.1)
2 10 18 36 54 86

With relations of this kind already known Bohr’s assignments of

electrons to the different elements are readily made. The various
electrons are classified under so-called shells of electrons. All electrons
belonging to the same shell are characterized by the same total quantum num¬
ber n. The first shell n — 1, the second shell n = 2, the third shell
n * 3, etc., is filled, or completed, when it contains 2, 8, 18, 32, etc.,
electrons, respectively. These numbers are given by the formula

N = 2n2.

The shells n = 1, 2, 3, 4, • • • are sometimes called (from x-ray

spectra) the K} L, My N, . . . shells, respectively.
The electrons in any shell n are further divided into subshells so that
electrons belonging to the same subshell have the same azimuthal quantum
number l. Electrons for which l = 0, 1, 2, 3, • • • are called s, p, d,
f, . . . electrons, respectively, where l = k — l (see Chap. IV). k
is the azimuthal quantum number of the Bohr-Sommerfeld orbits (see
Chap. III).
Bohr divided the number of electrons in each shell equally among the
subshells. Stoner,1 on the other hand, divided them in the manner
shown in Table 5.2.

Table 5.2

K L M N
n - 1 n = 2 n = 3 n = 4
Subshell System
Is 28 2p 3# 3p 3d 4s 4p 4d 4f

Number of electrons. 2 4 4 6 6 6 8 8 8 8 Bohr

Number of electrons. 2 2 6 2 6 10 2 6 10 14 Stoner
I

Stoner’s division, as we shall see later, follows directly from Pauli’s

exclusion principle and is in complete agreement with experimental
observations.
5.3. The First Period.—The one electron bound to hydrogen in its
normal state is a Is electron with an energy of binding of 109677 cm"1.
The energy level is designated 2S$. Helium in its normal state contains
two electrons both of which are bound to the atom in Is states. The
energy level of the system is designated as ^So. The electron configura-

1 Stoner, E. C., Phil. Mag.t 48, 719, 1924.

Sac. 5.5] THE ALKALI METALS AND THE PERIODIC TABLE 81

tions and the corresponding states for normal H and He atoms aae there¬
fore written:
H, Is, *Si
He, Is2, XS0
6.4* The Second Period.—With the K shell completed in helium the
third electron bound to the third element lithium, in its normal S state
(see Fig. 5.1), is a 2s electron. The fourth electron bound to the fourth
element beryllium, Z — 4, is also a 2s electron to complete the 2s sub-
shell. The normal state of the fifth element boron, Z = 5, is known
from spectroscopic data to be a P state. On the Bohr-Stoner scheme
the first electron added should be one with quantum numbers n = 2 and
l — 1. Such an electron is called a 2p electron. This, as well as other
examples to follow, indicates the origin of our present electron nomen¬
clature s, p, d, etc. The known spectra of the remaining elements
in the second period, carbon, nitrogen, oxygen, fluorine, and neon, show
that 2p electrons are added one after the other to complete not only the
2p subshell but also the main L shell, n = 2. The electron configurations
and the corresponding states for elements in the second period are
designated as follows:
Li, ls22$ , *Sh N , ls22$22p8,4S,
Be, ls22s2 , O , ls22s22p4, 8P*
B , ls22s22p , 2P* F , ls22s22p*, 2P,
C , ls22s22p2, 3P0 Ne, ls22s22p6, %
6.6. The Third Period.—The building up of the eight elements in the
third period of the periodic table is similar to that of the eight elements
in the second period. The normal level of the well-known sodium
diagram (see Fig. 5.1) is an S state, so that the eleventh electron to
be bound to sodium must be a 3s electron. Following sodium the
elements magnesium, aluminum, silicon, phosphorus, sulphur, chlorine,
and argon are built up by adding successively one more 3s electron and
then six 3p electrons. The normal states of the atoms and the corre¬
sponding configurations of the electrons are;
Na, ls22s22p63s , 2£* P , ls22s22p«3s23p*, 4£,
Mg, ls22s22p63s2 , lS0 S , ls22s22p63s23p4 8P2
A1 , ls22s22p63s23p , 2P* Cl, ls22s22p83s23p6 2P,
Si , ls22s22p63s23p2, 3P0 A, ls22s22p63s23p* X/S0
For brevity the first 10 electrons of these configurations are often
omitted, e.g., the configuration for A1 would be written 3s23p. The chief
reason for this is that only the last bound electrons are responsible for
the so-called optical spectra. In dealing with x-rays, however, the inner
electrons are involved in the process of emission or absorption and should
when necessary be given (see Chap. XVI).
82 INTRODUCTION TO ATOMIC SPECTRA [Chap. V

S.6. The Fourth Period or First Long Period.—The first long period
of elements, starting with potassium Z = 19, and ending with krypton
Z = 36, contains 18 elements. Since the subshells 3s and 3p have
been filled in argon Z = 18, and the 3d subshell is next in line, one might
well expect the normal state of potassium, Z = 19, to be a D state, i.e.,
one given by a 3d electron, n * 3, l = 2. The well-known spectrum of
potassium, on the other hand, is like that of sodium with an S state
at the bottom of the energy level diagram. It is therefore evident that
an 8 electron, and necessarily a 4s electron, must have been added in
place of the expected 3d electron, and this in turn must be due to the
fact that the 4s electron is more tightly bound than the 3d electron.
Table 5.3.—The Electron Configurations of the 92 Elements in Their Normal
Atomic States

K L M N Ncnnal
Shell
n- 1 »-2 ft — 3 »—4 state

Subehell 1-0 1-0 1-1 1-0 1-1 1-2 1-0 1-1 1=2 1-3

1 H la
2 He la* •So

3 Li la* 2a
4 Be 2a* •So

6 B la* 2a* 2p

■■
'P\
6 C “ 2p* *Po
7 N “ 2p» **3
8 0 14 2 p* *P2
0
10
F
Ne
M
1
“
“
2p*
2p* •So

11 N» la* 3a
12 Mg 1 3a* •So

18 At la* 2a* 2p* 3a* 3p 'P\

14
15
Si
P
mm ■ " 3p*
3p»
>Po
«S$
■
16
17
8
a H “
“
3p*
3p»
•Pt
*P3
18 A
mm 44 3p* •So

10 K la* 2a* 2p* 3a* 3p* 4a

44 44 44 44
20 Ce “ 4a* •So

21 So la* 2a* 2p* 3a* 3p* 34 4a* j *Dj

22 Vi “ “ “ “ •• 34* 4a* | *P>
66 66 66 4Fi
23 V 44 “ 34* 4a*
46 44 64
24 Cr •• " 34* 4a *S|
•• 66 14
25 Mn 44 44 34* 4a* ««
61 66 66 34*
26 Ft “ “ 4a* »Di
46 «« <4
27 Co “ 44 34* 4a* *1
44 M <«
28 Ni 44 44 34* 4a* »F«
66 M M
20 Cu “ 44 34“ 4a
M M 34“
80 Zn 44 •• 4a* •So
Sec. 5.6] THE ALKALI METALS AND THE PERIODIC TABLE 83

Table 5.3.—The Electron Configurations of the 92 Elements in Their NormaSl

Atomic States.—(Continued)
84 INTRODUCTION TO ATOMIC SPECTRA [Chap. V

Table 5.3.—The Electron Configurations of the 92 Elements in Their Normal

Atomic States.—(Concluded)

This anomaly will be discussed in Sec. 5.10. In calcium another 4s

electron is added to complete the 4s subshell. Following calcium the
next 10 elements, scandium, titanium, vanadium, chromium, man¬
ganese, iron, cobalt, nickel, copper, and zinc, are built up by the addition
of 3d electrons. The normal state of zinc has the complete electron
configuration
Zn, 1 s22s22p63s23p63d104$2, %.
The following six elements, Z = 31 to Z = 36, gallium, germanium,
arsenic, selenium, bromide, and krypton, are built up by the successive
addition of six 4p electrons and mark the end of the first long period.
The complete electron configuration for krypton in its normal state is
written
Kr, ls22s22p83s23p63<i104s24p8, »S0.

The building up of the entire periodic table may be given at this

point.
5.7. The Fifth Period or Second Long Period.—The second long
period, like the first, contains 18 elements starting with an alkali metal
and ending with an inert gas. Although the 4d and 4/ subshells are yet to
be filled, the thirty-seventh electron bound to rubidium, Z — 37, is a
Sac. 5.8] THE ALKALI METALS AND THE PERIODIC TABLE 85
5* electron. Another electron in a 5s state, added to form strontium,
Z = 38, completes the 5s subshell. In the next 10 elements, yttrium,
zirconium, columbium, molybdenum, masurium, ruthenium, rho¬
dium, palladium, silver, and cadmium, 10 4d electrons are added as
shown in Table 5.3 to complete the 4d subshell. The irregular way in
which these 4d electrons are added is indicative of an attempt to build
up one of the inner N subshells more quickly at the expense of one of
the already completed 0 subshells. Although these irregularities, which
are derived from experimental observations, were not predicted, they
are now well understood and will be treated in the following chapters.
With the next six elements, indium, tin, antimony, tellurium, iodine,
and xenon, six 5p electrons are added successively to complete the 5p
subshell. The normal state of xenon Z = 54, an S state, is given by
the complete electron configuration
Xe, ls22s22p83s23pa3d104s24p®4d105s25p6, »50
5.8. The Sixth Period or Third Long Period.—The third long period,
like the first and second, starts with an alkali metal and ends with
an inert gas. This period is unique, however, in that it is interrupted
at one point to allow the introduction of the 14 rare-earth elements.
Including these the period contains 32 elements. In the first two
elements, caesium and barium, 6s electrons are added to complete the
6s subshell. With the addition of one 5d electron in lanthanum, Z = 57,
a break occurs to allow the successive addition of 14 4/ electrons char¬
acteristic of the 14 rare-earth elements. The last element, Lu 71, marks
the completion of the N shell, n = 4. With the next element, Hf 72, the
addition of 5d electrons is resumed until the subshell is completed.
The building on of six 6p electrons in the next six elements, thallium 81
Table 5.4
S-ELECTRONS

Is* Is*
H H«
PERIODIC TABLE
2s* ~ 2s1*
Li Be NORMAL STATE ELECTRON
CONFIGURATIONS
F FTk
3s' * 3s*
Ha Mg 7m
86 INTRODUCTION TO ATOMIC SPECTRA [Chap. V

to radon 86, completes the third long period. The normal S state of
radon is given by the complete electron configuration
Rn, l«*2s22p63s23p83d104s24p64d104/145s25p65d106s26p6, ^So.
6.9. The Seventh and Last Period.—The seventh period, like all but
the first period, starts with an alkali metal, cornellium, followed by
radium, actinium, thorium, uranium X, and uranium. Though the
spectroscopic analysis for these elements is very meager, this period
very probably starts in just the same way as the first and second long
periods.
A very convenient and informing way of representing the addition
of the various electrons to the elements of the periodic table is shown
in Table 5.4. This arrangement not only retains most of the spectro¬
scopic and chemical relations between elements and groups of elements
but also places corresponding electron configurations and subshells
together in individual columns.
6.10. Energy Levels of the Alkali Metals.—Many attempts have
been made to calculate the energy levels arising from other than hydro¬
gen-like atoms. While calculations of the energy levels for the two and
three body problems, hydrogen, and helium have been reasonably
successful, many body problems still remain unsolved. In a few special
cases, however, some headway has been made with the more complex
atoms.
Next to hydrogen the alkali metals Li, Na, K, Rb, and Cs have
the simplest of the known spectra. Each alkali atom contains, in
addition to the completed subshells of electrons, one single electron in
an outer shell (see Table 5.3). The complete configurations are written
out in Table 5.5.

Table 5.5.—Normal Configurations for the Alkali Metals

L M N 0 P

Li Is* 2s
Na Is* 2s* 2 p* 3s
K Is* 2s* 2p* 3s* 3p* 4s
Rb Is* 2s* 2p* 3s* 3p* 3d10 4s* 4p* 6s
C8 Is* 2s* 2p* 3s* 3p* 3d10 4s* 4p* 4d10 5s* 5p6 6s

It is the single unbalanced electron that is responsible for the positive

valence of one of the alkali metals and also for the quite simple spectrum.
Referring to the simple hydrogen spectrum, the Lyman, Balmer,
Pascheh, Brackett, and Pfund series are given by Balmer’s formula

(5.2)
Sec. 5.10] THE ALKALI METALS AND THE PERIODIC TABLE 87
r
Rh is the Rydberg constant for hydrogen, and ni and n2 are integers.
The success of Bohr’s theory of the hydrogen atom lay in the fact that
he was able for the first time to obtain from theoretical considerations
the fundamental Rydberg constant and to show that ni and n* are the
necessary quantum numbers to be associated with each quantized
state of the atom. Neglecting small relativity and spin corrections, to
be treated in Chaps. VIII and IX, there is no difference in energy between
the Bohr circular orbits of hydrogen and the Sommerfeld elliptic orbits
with the same total quantum number n.

L. I I K I Rb 1 Cs I H
2S2P2D2F 2S2P2D2F 7S2P'D1? *S 2D 2F ^^^‘F *SPDF

Fiq. 5.2.—Energy level diagrams of the alkali metals, lithium, sodium, potassium, rub**
dium, and caesium, and hydrogen.

In complex atoms, on the other hand, there is a considerable energy

difference between the levels arising from s, p, d, and f electrons of the
same total quantum number. This difference in energy is attributed
to the various amounts of penetration of the different electron orbits
into the electron subshells.
Returning to the energy level diagram of lithium in Fig. 5.1, the
normal state 22S arises from the electron configuration ls22s. If now
the atom is to be excited, the last bound and most easily moved electron
2s has the possibility of going to any of the higher virtual states, t.e., to
2p, 3s, 3p, 3d, 4s, 4p, etc. When in one of these stationary states, the
energy level is designated, as in the diagram, by 22jP, 32S, 32P, 32Z), 42£,
42P, etc. If the excited electron is in the 3d state, for example, the
complete electron configuration of the atom is designated as ls^d and
the energy level as 32D.
88 INTRODUCTION TO ATOMIC SPECTRA [Chap. V

This same designation scheme leads to the level assignments shown

for sodium in Fig. 5.1. The K and L shells already being filled,\ the
various possible excited states for the last bound electron 3s are 3p, 3d,
4s, 4p, etc. If the excited electron is in a 4p state, the complete electron
configuration is designated ls22s22p84p and the energy level1 as 42P. To
save time and space this configuration is abbreviated 4p.
Energy-level diagrams of the five alkali metals are given in Fig. 5.2.
In this type of diagram the levels are shown as dots in place of straight
horizontal lines. The electron configuration of each level is given in
each case by the excited electron only. The energy levels of hydrogen
are given at the right in the figure for comparison.
Term values2 for the first five or six energy levels in each of the four
chief series of lithium and sodium are given in Tables 5.6 and 5.7. For
comparison purposes the hydrogen values are also given.
Table 5.6.—Term Values of Lithium
(After Paschen and Goetze)

Term Electron n = 2 n - 3 n = 4 n — 5 n = 6 n - 7

a, l = 0 43484.4 16280.5 8474.1 5186.9 3499.6 2535.3

’P p, i -1 28581.4 12559 9 7017 0 4472 8 3094.4 2268 9
*Z> d, l = 2 12202.5 6862.5 4389 2 3046 9 2239.4
*F j /, J = 3 6855.5 4381.2 3031.0

Hydrogen. 27419 4 12186.4 6854 8 4387 1 3046.6 2238.3

Table 5.7.—Term Values of Sodium

(After Paschen and Goetze)

Term Electron n = 3 n = 4 n — 5 n = 6 w - 7 n = 8

a, l ~ 0 41444.9 15706.5 8245.8 5073.7 3434 9 2481 9

*P p, i -1 24492 7 11181.9 6408 9 4152.9 2908 9 2150 7
*D d,l = 2 12274.4 6897 5 4411 6 3059 8 2245.0 1720.1
*F f, 1=3 6858.6 4388.6 3039 7 2231.0 1708.2

Hydrogen.. .. 12186.4 6854.8 4387 1 3046.6 2238.3 1713.7

6.11. The Effective Quantum Number and the Quantum Defect.—It

has already been pointed out in Chap. I that the principal series of sodium
has been observed as far as the forty-seventh member (see Fig. 1.11).
Although the different series belonging to the other alkali metals have
not been observed to quite as many members, the series limits and term
1 The figure 4 refers to the total quantum number of the electron and is only used
in one-electron systems as an abbreviation of the more explicit notation where the
electron configuration is given first, followed by the term type, t.e., 4p, *P.
* The terms, energy level, energy state, state, and term, are to be considered more
or less synonymous and are often used indiscriminately.
 Sac. 5.11] THE ALKALI METALS AND THE PERIODIC TABLE 89
values of the sharp, principal, diffuse, and fundamental series are known
with very great precision. We have also seen in Chap. I that the
frequency of a spectrum line is given by the difference between two
terms, and that a series of lines is given by the difference between a fixed
term and a series of running terms,
_ R R . .
(n i + Mi)2 (n>2 + M2)2* ' ’
where R is the Rydberg constant, ni and n2 are integers, and mi and m%
are small constants.
With somewhat definite knowledge concerning the quantum number
n of the valence electron in the alkali metals, the integers ni and n%
in Rydberg's formula may be replaced by the total quantum numbers
of the electron energy states. Since the quantum number n is in general
greater than Rydberg's denominator n + /x, the formula for a spectrum
line or series may, changing the sign of mi and /z2, be written

R R
V (ni - mi)2 (n, - M2)2 { }
and a spectral term by

7 ‘ a <5-5)
where n is now the total quantum number. The factor n — m is still
called the Rydberg denominator, m is the quantum defect, and nerr is
the so-called effective quantum number.
The bracketing together in Fig. 5.1 of the energy levels having
the same total quantum number reveals a number of interesting and
important relations. One important result is that in each group of
levels the S level lies deepest and is followed by P} D, and F levels in
this order. A second relation indicates that in going from the more
elliptic s orbits (S states) to the more nearly circular p, d, and / orbits
(P, D, and F states) the term values approach those of hydrogen. Using
the Eq. (5.5), the effective quantum numbers n«// for the term values of
lithium and sodium (see Tables 5.6 and 5.7) are calculated and given in
Tables 5.8 and 5.9.

Table 5.8.—Values of the Effective Quantum Number n*ti for Lithium

(Ru = 109729 cm-*)

Term Electron n — 2 n *= 4 n = 5 n - 7
s
II

2S s, l - 0 1.589 3 598 4.599 5.599 6.579

2P P, * “ 1 1.960 3.954 4 954 5.955 6.954
?Z) d,l- 2 3.999 5.000 6.001 7.000
ip /, i-3 4.000 5 004
90 INTRODUCTION TO ATOMIC SPECTRA [Chap. V

..
Table 5 9 —Values of the Effective Quantum Number n^t for Sodium
(An* * 109734 cm"1)

Term Electron n - 3 n = 4 n = 5 n - 6 n - 7 n « 8

2S 8,1- 0 1.627 2.643 3.648 4 651 5.652 6.649

2P p,l - 1 2.117 3.133 4 138 5 141 6.142 7.143
2D d,l - 2 2.990 3.989 4 987 5.989 6.991 7 987
2F /, i = 3 4 000 5.001 6 008 7.012 8 015

A correlation of n and n9tt shows that y is greatest for the S states,

and that, as we shall see in Chap. VII, it is a measure of the penetration
of the valence electron into the subshells of electrons. The picture
formed is that of a nucleus of charge Ze closely surrounded by a core
of Z — 1 electrons outside of which at some distance a single electron is
moving in a hydrogen-like state. The effect of the net core-nucleus
charge e is such as to make the system hydrogen-like, i.e., a singly
charged nucleus and one electron. For the s states the electron pene¬
trates down into the core where the effective nuclear charge is greater
and hence the binding is greater.
6.12. The Selection Principle.—Under ordinary conditions spectrum
lines corresponding to the transition of an electron from any one state
to another are not observed. In emission or absorption spectra a
selection rule is found to be in operation. In short, this selection rule,
which may be derived from Bohr’s correspondence principle as well as
the quantum mechanics, may be stated as follows: In any single electron
transition the quantum number l must change by +1 or — 1, i.e.t
Al — ± 1. (5.6)
In terms of the orbital model the orbital angular momentum must
change by one unit of h/2w only. With the law of the conservation of
angular momentum in mind this can only be interpreted to mean that
every photon hv has associated with it an angular momentum of h/2w.
In any transition the total quantum number n is not restricted in
any way and may change by any integral value, including zero. Let us
consider several examples.
If the sodium atom (see Fig. 5.1) is excited by some means or other
so that the single-valence electron is raised to one of the excited states
32P, 42<S, 32Z>, 42P, . . . , the electron may return to the normal state
32S by making only certain selected jumps. According to the selection
principle an electron in a 32D state must return to the normal state 32S
in two separate jumps 32D to 32P with the emission of the first line of the
diffuse series, written 32P-32D; then 32P to 32S with the emission of the
first line of the principal series, written 32<S-32P. If the electron is
in a 42P state, on the other hand, it has three possible routes open along
Sac. 5.12] THE ALKALI METALS AND THE PERIODIC TABLE 91

which it may return to the normal state: first, by the direct transition
4 2P to 3 2S; second, by the three transitions 4 2P to 3 2D to 3 2P to 32S; or
third, by the three transitions 42P to 42£ to 32P to 32S. There are some
exceptions to the selection rule for the quantum number J, but the
forbidden transitions occur so seldom that the resultant spectrum lines
are exceedingly faint. These exceptions are found experimentally and
should occur theoretically when high external fields are present (see
Chaps. X and XIII).
It should be pointed out in passing that the quantum numbers
assigned to any electron really belong to the atomic system as a whole.
It is only because we can approximate a complex atom by a set of inde¬
pendent one-electron systems with given values for the quantum numbers
n and Z, that we can associate the latter with the individual electrons.
Problems
1. If the valence electron in sodium is excited to the 42D state, what are the
different routes open for the electron in returning to the normal state ?
2. Calculate the Rydberg denominators for the first five term values of the sharp,
principal, diffuse, and fundamental series of potassium (see Fowler, “ Series in Line
Spectra”). Where the fine structure is known, use the larger of the two values,
R ~ 109736 cm"*1.
3. From the values of the effective quantum number calculated in Prob. 2, compute
the quantum defect ix.
4. The first five members of a certain series have wave-lengths as follows: X*ir -
7800.29, 4201.82, 3587.08, 3348.72, and 3228.05 A. Using Eq. (5.4), compute at
least four values of ^ and from their average determine m and the term value of the
fixed term. With the fixed term known, determine the five running terms. Assume
R - 109737 cm-1, and ni = 5 and n2 = 5, 6, 7, 8, and 9.
.
5 It will be found very useful to memorize the order of the elements in the
periodic table and the corresponding electron configurations of the normal state. This
task is greatly facilitated by taking one period at a time and writing them in the form
shown in Table 5.4.
 CHAPTER VI

EXCITATION POTENTIALS, IONIZATION POTENTIALS, AND THE

SPECTRA OF IONIZED ATOMS

One of the most direct proofs of the existence of discrete energy

states within an atom was first found in the experiments of Lenard,1
Franck and Hertz,2 and others, on critical potentials. The collision
of a neutral atom with a fast-moving atom, molecule, or electron will
under suitable conditions result in the excitation or ionization of the
atom. If in collision the energy exchange between, say, an electron
and atom is all energy of translation, the atom is not excited and the

Flo. 6.1.—Schematic diagram of Franck and Hertz’ critical potentials experiment.

collision is said to be elastic. If, on the other hand, the velocity of

the electron is sufficiently high, the collision may be inelastic. In this
case the electron gives up all or part of its energy in giving potential
energy to the atom, or, as we say, in exciting the atom. Inelastic col¬
lisions between atoms, molecules, or electrons, which by virtue of trans¬
lational motion result in the excitation or ionization of one of the atoms
or molecules, are called collisions of the first kind. An excited atom may
in turn by collision with another atom, molecule, or electron, return to
the normal state without the emission of a radiated frequency. The
potential energy of the excited atom is in this case transformed into
energy of translation and the collision is called a collision of the second
kind.
6.1. Critical Potentials.—In the Franck and Hertz experiments on
critical potentials a vapor like sodium or mercury is bombarded with

1 Lenard, Ann. d. Phys8, 140, 1902.

,
* Franck, J., and G. Hertz, Verh. d. Deutseh. Phys. Oes.f 16 12, 1914; Phys.
,
Z$Us.t 17 409, 1916.
92
Sac. 6.1] EXCITATION POTENTIALS, IONIZATION POTENTIALS 93

electrons of known velocity. A diagram of the apparatus in its simplest

schematic form is given in Fig. 6.1. Electrons from a filament F are
accelerated toward a grid G by applying a potential Vi between them.
An opposing potential Vi, much smaller than Vi, is applied between
the grid G and the plate P. The vapor pressure in the tube and the
distances between filament, grid, and plate are so adjusted that the mean
free path is considerably smaller than the filament-grid distance FG and
somewhat greater than the grid-plate distance GP.
If an electron starts from rest at the filament and reaches the grid G
without an encounter, its velocity v is given by the energy relation
Ve - \mv\ (6.1)

where e is the charge on the electron, m is its mass, and V is the acceler¬
ating potential in electrostatic units. Expressing V in volts Vi the
energy becomes1

me ~ *“*• <6-2>
In order for an electron to collide inelastically with an atom, it must
give up sufficient energy $mv2 to raise the valence electron from its
normal state to the first excited state. In the case of a sodium atom (see
Fig. 5.1) this amounts to raising the valence electron from the 32/S to the
3 2P state, an energy given by
Wd *P — W3*s = hv = hvc, (6.3)

where h is Planck’s constant, c the velocity of light, and v the frequency

in wave numbers of the first member of the principal series 32S — 32P.
In this example the kinetic energy given up by the bombarding electron
must be
he • 16956 ergs. (6.4)

It is customary to express the kinetic energy of an electron in terms

of volts. The general relation between electron velocity expressed in
volts and the excitation energy of the atom expressed in wave numbers
is, from Eqs. (6.2) and (6.3), given by the expression

= Vi = = 1.2336 X 10-4 X »>. (6.5)

An electron starting from rest must therefore fall through a differ¬

ence of potential Vi — 1.2336 X 10~4 X 16956 = 2.1 volts in order
that upon collision with a sodium atom at rest the valence electron will be
raised to the 32P state.

1 For a treatment of the early work on critical potentials see K. T. Compton and
F. L. Mohler, “Critical Potentials,” Nat. Research Council. Bull., Vol. 9, Pt. I, No. 48,
1924.
94 INTRODUCTION TO ATOMIC SPECTRA [Chap. VI

Characteristic of the early work on critical potentials are the experi¬

ments of Tate and Foote1 on sodium vapor in which an experimental
set-up similar to that of Franck and Hertz (Fig. 6.1) was used. The
curve reproduced in Fig. 6.2, showing the variation in plate current
with accelerating potential Vi, is characteristic of their results and may
be explained as follows. Initially, with Fi = 0, very few electrons
leave the filament with sufficient velocity to reach the plate against the
small retarding field Vif maintained in this case at about 0.8 volt. As
Vi is gradually increased, the speed and number of electrons reaching

c
ft)

fc
3
u
L
ft)
*
E
o
n
§
§

0 2 4 6 8
Accelerating Potential, Vol+3
Fig. 6.2.—Critical potential curve of sodium. (After Tate and Foote.)

the grid increase and the plate current rises. When the velocity has
increased sufficiently to excite sodium atoms, inelastic collisions occur.
With a further increase in Vi more electrons reach the critical velocity,
are stopped by inelastic collision, and not being able to reach the plate
cause a temporary drop in galvanometer current. This drop in current
continues until the critical speed is attained far enough in front of the
grid to collide inelastically and again reach the plate P. The current
therefore rises again and continues to rise until the electrons after one
inelastic collision attain the critical speed and make a second inelastic
collision. Except for the doubling of the third and fourth maxima, this
process and explanation repeats itself with each major rise and fall of the
curve.
The double peaks in Fig. 6.2 show that not only have collisions
occurred in which the valence electron has been excited but also collisions
in which the valence electron has been completely removed from the
atom, i.e., the atom has been ionized. Complete ejection of the electron
from sodium would, from Eq. (6.5), require an equivalent electron
velocity of 1.2336 X 41444.9 = 5.1 volts. In view of the initial electron
1 Tate, J. T., and P. D. Foote, Jour. Wash. Acad. Sci., 7, 517,1917.
Sec. 6.1] EXCITATION P0TENTIAL8} IONIZATION POTENTIALS 96

velocity at the filament the first ionization should occur at about Vi » 4.5
volts. With ionization taking place at this potential, it is to be expected
that the electrons starting from rest at this point will collide inelastically
at 2.1 volts further on, as observed.
Improved experimental technique on critical potentials has made
it possible to obtain not only one excitation potential for a given atom

Table 6.1.—Critical Potentials or the Alkali Metals

First excitation potential Ionization potential

Element
Computed Computed
Observed Observed
from spectra from spectra

Li 3 1 84 5 37
Na 11 2.12 2.09 5.13 5.12
K 19 1.55 1.60 4.1 4.32
Rb 37 1.6 1.55 4.1 4.16
Cs 66 1.48 1.38 3.9 3.88

but many. The additional critical potentials correspond to the raising

of the valence electron to excited states other than the first one dis¬
cussed above. Although in every element investigated the critical
potentials are in excellent agreement with the values calculated from
spectral terms, it should be mentioned that an occasional potential is
found which is difficult to explain. The first excitation potential and the
ionization potential of each of the alkali metals are given in Table 6.1.

These experimental determinations of critical potentials confirm

the values computed from spectroscopic data. With the physical con¬
stants e, h, and c known with considerable accuracy, spectroscopic
data furnish the most accurate means of determining excitation as well
as ionization potentials.
Known ionization potentials for the elements in the periodic table
are given in Fig. 6.3. The various electron subshells being built up
are indicated between the vertical lines. These symbols also indicate
06 INTRODUCTION TO ATOMIC SPECTRA [Chap. VI

the type of electron removed in ionization. It is to be noted that the

inert gases show the largest ionization potentials and the alkali metals
the smallest.
6.2* The Spectra of Ionized Atoms.—The elements beryllium, magne¬
sium, calcium, strontium, and barium contain, in addition to the closed
shells of electrons of the preceding inert gases, two rather loosely bound
e electrons (see Tables 5.3 and 5.4). The positive valence two, which is
characteristic of the chemical behavior of the alkaline earths, is attributed
to these two outermost, but deeply penetrating, electrons. The electron
configurations of these elements, excluding the inert gas cores, are

Be I Mg I Cal Sr I Ba I
2s2 3s2 4s2 5s2 6s2

If by some excitation process each of these atoms loses one of its

valence electrons, i.e., it becomes ionized, the electron configurations, so
far as the remaining electrons are concerned, become identical with
those of the preceding alkali metals:

Li I Na I K I Rb I Cs I
Bell Mg II Call Sr II Ball
28 Ss 4s 5s 6s

While the process of excitation and emission of radiation from

ionized atoms is similar to that for neutral atoms, the energies involved
are considerably greater. If the excitation is brought about by an
encounter between a fast-moving electron and a neutral calcium atom,
then, at impact, the interchange of energy must be sufficient to remove
completely one valence electron from the atom and to raise the other
to an excited state.
An energy level diagram of singly ionized calcium is given in Fig. 6.4.
The normal state of the atom is given by 4 2S and corresponds to the
valence electron in a 4s orbit. Unlike the alkali metals the first excited
state is a diffuse level rather than a principal level, however. Owing
to the fact that the electron in a 3 2D level cannot return to the normal
level 42S, without violating the selection rule (Al = ±1 only), this
32D level is called a metastable state. Once the electron is in this meta¬
stable state, it will remain there for some time and may return to the
normal state only by collision with another particle and the liberation
of its energy without radiation. Next following the metastable state
comes the first member of the principal series 42P, the second member
of the sharp series 52S, the second member of the diffuse series 42D,
etc. States having the same total quantum number n are bracketed
together. While the strong lines of the alkali metals (see Fig. 5.1)
are to be found in the visible and near infra-red spectral region, the
corresponding lines in the ionized alkaline earths are to be found in the
Sac. 6.2J EXCITATION POTENTIALS, IONIZATION POTENTIALS 97

visible and near ultra-violet region. In potassium, for example, the first
member of the principal series is in the far red between 7000 and 8000 A,
whereas the first line of the principal series of ionized calcium is in the

violet just below 4000 A. Photographs of these lines are given in

Fig. 17.1.
The atomic picture that may be formed of an alkaline-earth atom
after it is ionized is that of a nucleus of charge Ze closely surrounded
by a cloud, or core, of Z-2 electrons and, outside this, one valence electron.
The net core-nucleus charge 2e is such as to make the system similar
98 INTRODUCTION TO ATOMIC SPECTRA [Chap. VI

to that of ionized helium where the term values, according to Bohr's

theory, are given by

n2 n2
n = 1, 2, 3, • • • (see Chap. II) (6.6)
By analogy with the formulas of hydrogen and ionized helium it follow s
that for ionized atoms like those under discussion the expression for
term values may be written
4R 4R
T = (6.7)
nlti in - m)2>
where n is the total quantum number of the electron and m the quantum
defect. The well-known enhanced series of the alkaline-earth elements
are represented therefore [see Sec. 1.16 and Eq. (1.41)] by the formula

4R _4 R
(6.8)
(ni — m)2 (n2 — M2)2

Energy level diagrams for each of the ionized alkaline-earth elements

are given in Fig. 6.5. The term-value scale is one-fourth that of the

Fig. 6.5.—Energy level diagrams of singly ionized beryllium, magnesium, calcium, stron¬
tium, and barium, and hydrogen.

hydrogen scale shown at the right. With the relatively large energy
values for the different states, electron transitions between states will
in general be large. The difference between each term value and the
hydrogen term values of the same n, shown by horizontal lines, is a
measure of the quantum defect m. A comparison of these diagrams with
those of Fig. 5.2 brings out a number of interesting relations. The
relatively large quantum defect of s electrons in Li I and Be II, of a
Sac. 6.2] EXCITATION POTENTIALS, IONIZATION POTENTIALS 99

and p electrons in Na I and Mg II, and of s, p, and d electrons in K I,

Ca II, Rb I, Sr II, Cs I, and Ba II, which is a measure of the penetration
of the valence electron into the subshells of electrons, is of particular
interest and will be considered in detail in Chap. VII,
One outstanding difference between the diagrams is the relative
order of binding of s, p, and d electrons in the third, fourth, and fifth
elements. In potassium, for example, the order is 4s, 4p, 5s, 3d, 5p, etc.,
whereas in Ca II it is 4s, 3d, ip, 5s, 4d, etc. Noting this shift of the
3d electron from fourth place in K I to second place in Ca II, Bohr
predicted that in doubly ionized scandium, Sc III, the 32Z) state should
lie deepest. Later investigations, as will be shown in Chap. XVII,
confirm this.
Although spectra of many singly ionized atoms are well known,
the examples of the alkaline earths as treated here will serve as an
introduction to the general subject of the spectra of ionized atoms.
This subject will be resumed in Chap. XVII where multiply-ionized
atoms will also be treated.
 CHAPTER VII

PENETRATING AND NONPENETRATING ORBITS IN THE ALKALI

METALS

We shall now turn our attention to the formulation of an atomic

model which treats the interaction of a single valence electron with
the nucleus when it is screened by an intervening core of electrons,
t.e., by inner completed subshells of electrons. This must be done if

Fiq. 7.1.—Probability-denerity-distribution curves for the rubidium-atom core of 36 elec¬

trons. (After Hartree.)

we are to calculate from theoretical considerations the energy levels

of the alkali metals. On the classical picture of the atom the nucleus
is surrounded by various shells and subshells of electrons in orbits
resembling the various possible states of the Bohr-Sommerfeld hydrogen
atom. The quantum-mechanical picture, on the other hand, appears
in the form of a probability-density distribution for the same electrons.
7.1. The Quantum-mechanical Model of the Alkali Metals.—By
means of successive approximations to the so-called self-consistent field,
100
Sue. 7.11 PENETRATING AND NONPENETRATING ORBITS 101

Hartree1 has calculated the radial density distributions of each core

electron for the different alkali metals. Although a treatment of the
methods by which these calculations are made is out of place here, the
calculations are not difficult but are long and tedious. It will suffice
K

0 12 3 4 5 6 7a,
.K

Fig. 7.2.—Probability-density distribution curves for the neutral alkali atoms, lithium,
sodium, potassium, and rubidium. In each case the core is shown by one curve and the
valence electron by another.

to say, however, that the resultant electric field obtained for any atom
is such that the solutions of Schrddinger’s wave equation for all of the
core electrons in this field give a distribution of electrons which reproduces
the field.

1 Habtbxb, ,
D. B., Proe. Camb. Phil. Soc.t 24 89, 111, 1928.
102 INTRODUCTION TO ATOMIC SPECTRA [Chap. VII

As an example of Hartree’s results, probability-density-distribution

curves for the rubidium core are shown in Fig. 7.1. The radial curve
for each subshell is shown in the lower half of the figure. The heavy
curve represents the sum of all of the 36 core electrons. The four
loops, or humps, in the latter are taken to represent the K, L, M, and N
shells, even though every electron contributes something to each shell.
The dotted curves shown above for comparison purposes have been
calculated from hydrogen wave functions by Pauling1 by using approxi¬
mation methods and by Thomas2 from statistical considerations. In
comparing these curves with the corresponding hydrogen-like functions
shown in Fig. 4.6, it is observed that each density curve, due to a very
large nuclear charge, has been drawn in toward the nucleus by a con¬
siderable amount.
Probability-density-distribution curves for lithium, sodium, potas¬
sium, and rubidium are given together in Fig. 7.2. In each case the
core charge is shown by a shaded curve and the valence electron in its
normal state by another curve. Radially it should be noted that the
major shells in each atom lie well inside the first Bohr circular orbit of
hydrogen r = «i, and that the valence electron lies well inside the
corresponding hydrogen state. In hydrogen, for example, the density
distribution D for a 5s electron is appreciably large as far out as 50ai,
whereas in rubidium the nodes and loops have been pulled in to about
one-tenth of this.
The dots on the r axis represent the extremities of the classical orbits
based on model a (see Fig. 4.8). In these s states the kinetic energy
of the electron at the end of the orbit is zero, and the total energy is
all potential,

W = P. E. = --eV, (7.1)
r
where r is in centimeters. Expressing V in volts and r in Angstroms,
T/ _ 300e _ 300 X 4.77 X 10"10 _ 14.31 n 0,
r X 10“8 r X 10“8 r ^ '
Expressing r in units of ai (a2 = 0.528 A),
_ 27\1_27\1_(73)
rnu* V ionization potential in volts ^ * *
From the ionization potentials of the alkali metals given in Table 6.1
the following values of the orbital extremities are obtained:
Li Na K Rb Cs
r— = 5.0ai 5.3ai 6.3ai 6.5ai 7.0ai
1 Paulino, L., Proe. Roy. Soc., A, 114, 181, 1927.
* Thomas, L. H., Proc. Canib. Phil. Soc., 28, 542, 1927; see also Gaunt, Proc. Camb.
Phil. Soc., 24, 328, 1928; and Fermi, ZeiU. f. Phys., 48, 73, 1928.
Ssc. 7.2] PENETRATING AND NONPENETRATING ORBITS 103

The above derivation is only a close approximation, for we have assumed

a rigid core of unit charge.
7.2. Penetrating and Nonpenetrating Orbits.—The quantum-
mechanical model of the sodium atom is shown in Fig. 7.3. In addition
to the shaded curve for the 10 core electrons the three lowest possible
states for the one and only valence electron are also shown. The corre-

SODIUM
Penetrating Non* Penetrating

I.
3d

8 ? 10 II 12 13 14ci|

Fig. 7.3.—Comparison of the quantum-mechanical with the classical model of the

neutral sodium atom. Three of the lowest possible states for the single valence electron
are also shown.

sponding 3s, 3p, and 3d classical orbits based on model b (see Fig. 4.8),
are shown in the lower part of the figure. A comparison of these orbits
with the corresponding hydrogen orbits shows that, due to penetration
into the core, 3s and 3p are greatly reduced in size radially. The 3d
orbit, on the other hand, remains well outside the main part of the core
and is hydrogen-like. Corresponding to the penetration of the 3s and 3p
orbits the probability-density-distribution curves (above) have small
loops close to the nucleus.
104 INTRODUCTION TO ATOMIC SPECTRA [Chap. VII

Consider now the classical picture of a valence electron describing

any one of various types of orbits about the spherically symmetrical
sodium-atom core. In Fig. 7.4 six different orbits are shown representing
valence-electron states with the same total quantum number n (the
same major axis) but slightly different azimuthal quantum number l
(different minor axes). With a core-density distribution of charge
always finite but approaching zero as r —> », all valence-electron orbits
will be more or less penetrating.
In Fig. 7.4a the electron moves in a path well outside the major
part of the core. Since the field in this outside region does not deviate

Fio. 7A.—Showing the change in electron orbits with increasing penetration.

greatly from a Coulomb field, the orbit will be a Kepler ellipse precessing
slowly (due to small deviations from a Coulomb field) about the atom
center. In the remaining figures increased penetration is shown accom¬
panied by an increase in the precession at each turn of the orbit. As
the electron goes from aphelion (r^) to perihelion (rmln), it leaves
behind it more and more of the core charge. With the steady increase
in force field the electron is drawn from its original path into a more
and more eccentric path, with the result that at its closest approach
to the nucleus the electron has turned through somewhat more than
180 deg. Upon reaching again, there has been an advance, i.e., a
precession, of the aphelion. The increased force of attraction between
nucleus and electron at penetration increases the binding energy, the
kinetic energy, and the term values but decreases the total energy of the
atomic system [see Eq. (2.15)].
Bsc. 7.3] PBNBTRA TINQ AND NONPBNBTRA TINO ORBITS 105

7.3. Nonpenetrating Orbits.—Nonpenetrating orbits are defined

those orbits for which the observed energies are very nearly equal to
those of the corresponding hydrogen-like orbits. Such orbits on either
the classical or quantum-mechanical model do not appreciably penetrate
the atom core and are those states for which the azimuthal quantum
number l is more nearly equal to the total quantum number n. The /
orbits in all of the alkali metals are good examples of nonpenetrating
orbits. Term values for the 4/, 5/, and 6/ states given in Table 7.1 will
illustrate this.

Table 7.1.—Term Values op Nonpenetrating / Orbits in the Alkali Metals

Compared with Those of Hydrogen

Electron designation. 5.f 6/

Term designation . •. 4 *F 5V 6V

Hydrogen 6854.85 4387.11 3046.60

Li 6855.5 4381.2 3031.0
Na 6858.6 4388.6 3039.7
K 6879.2 4404.8 3057.6
Rb 6893.1 4413.7 3063 9
Cs 6935.2 4435.2 3076.9

With the exception of caesium the observed values are hydrogen-like

to 1 per cent or better.
Term values plotted as they are in Fig. 5.2 show, in general, that
p, d, and / orbits in lithium, d and / orbits in sodium and potassium,
and / orbits in rubidium and caesium
are nearly hydrogen-like. In the /
enhanced spectra of the ionized alka- /
line earths the term values are to f /
be compared with those of ionized
helium, or they are to be divided by 4
(t.e., by Z2) and compared with hydro¬
gen as in Fig. 6.5. In these energy
level diagrams it is observed that p, d, Fig. 7.5.—Schematic representation
and / orbits in Be II, d and/orbits in of th® polarisation of the atom core by
Mg II, and/orbits in Ca II, Sr II, and
Ba II are hydrogen-like and therefore correspond to nonpenetrating orbits.
Although nearly hydrogen-like, the term values of nonpenetrating
orbits (seeTable 7.1), in general, are greater than those of hydrogen. Born
and Heisenberg1 attributed these small differences to a polarization
of the core by the valence electron (see Fig. 7.5). In the field of the

1 Born. M., and W. Hbisxnbsbo, ZeiU. /. Phyt., S3, 388,1024.

106 INTRODUCTION TO ATOMIC SPECTRA [Chap. VII

valence electron the atom core is pushed away and the nucleus is pulled
toward the electron by virtue of the repulsion and attraction of like and
unlike charges, respectively. The effect of this polarization is to decrease
the total energy of the system. On an energy diagram this means a
lowering of the level, i.e,f an increase in the term value. Theoretical values
of the polarization energy calculated for the alkali metals with the aid
of the quantum mechanics are found to account for the major part of
these very small deviations from hydrogen-like terms.1
7.4. Penetrating Orbits on the Classical Model.—Although no sharp
line of demarkation can be drawn between penetrating and nonpene¬
trating orbits, the former may be defined as those orbits for which the
term values are appreciably different from those of hydrogen. Referring
to Fig. 5.2, the s orbits of Li, the s and p orbits of Na and K, and the
s, p, and d orbits of Rb and Cs come under this rough classification. Cer¬
tainly on the quantum-mechanical model all orbits are penetrating.
To effect a calculation of the term values for penetrating orbits on the
classical theory, one is led by necessity to simplify somewhat the atom-core
model given in Fig. 7.4. A suitable idealized model was first put forward
by Schrodinger2 in which the core elec¬
trons were thought of as being distrib¬
uted uniformly over the surface of one or
more concentric spheres. This same
model has been treated by Wentzel,3
Sommerfeld,4 * Van Urk,6 Pauling and
Goudsmit,® and others. Since the
classical treatment of penetrating orbits
is so closely analogous to the quantum-
mechanical treatment of the same
Fio 7.6 -Valence electron penetrat- orbitg to be taken up in the next
mg an ideal core where the core electrons * ^
ue distributed uniformly over the sur- section, Schrodinger’s simplified model
face of a sphere. will be considered here in some detail.

Consider the very simplest model in which the core electrons are
distributed uniformly over the surface of a sphere of radius p (see Fig. 7.6).
Let Z»e represent the effective nuclear charge inside the charge shell and
Z<# the effective nuclear charge outside the shell. Usually Z0 is 1 for the
alkali metals, 2 for the alkaline earths, etc. The potential energy

1For the quantum-mechanics formula giving the polarization energy see L.

Pauling and S. Goudsmit, “Structure of Line Spectra,” p. 45, 1930; and also J. H.
Van Vlbck and N. G. Whitelaw, Phya. Rev., 44, 551, 1933.
* Schrodinger, E., Zeita.f. Phys., 4, 347, 1921.
* Wentzel, G., Zeita. f. Phya., 19, 53, 1923.
4 Sommerfeld, A., “Atombau,” 5th German ed., p. 422, 1931.
•Van Urk, A. T., Zeita.f. Phya., 13, 268, 1923.
•Pauling, L., and S. Goudsmit, “Structure of Line Spectra,” p. 40, 1930.
Sec. 7.4] PENETRATING AND NONPENETRATING ORBITS 107

of the valence electron when outside the thin 'spherical shell of charge
(,Zi — ZQ)e will be
y_
(7.4)
Vt r
while inside it becomes
_Zte* Z<e2 _ Zof'
Vi (7.6)
r p p '
The total energy for an elliptic orbit, in polar coordinates r and <p, is
given by Eq. (3.26) as

w = T + V = ±(p'r + ^!) + v, (7.6)

where pr and pv are the radial and angular momenta, respectively.

From the results obtained in Chap. Ill it is seen that outside the charged
shell the motion will be that of an electron in a Coulomb field of charge
ZGe, and inside the shell the motion will be that of an electron in a Cou¬
lomb field of charge Zxe.
Applying the quantum conditions to the orbital angular momentum
Pv, which must be constant at all times throughout the motion,

= kh, Pv= k~ (7.7)

That part of the electron path which is outside the shell is a segment
of an ellipse determined by the azimuthal quantum number k and the
radial quantum number r0, whereas the path inside the shell is a segment
of an ellipse determined by the same azimuthal k (pv = constant) but a
different radial quantum number, rx. Substituting successively the
potential energies of Eqs. (7.4) and (7.5) in the total energy [Eq. (7.6)]
snd solving for pr) the radial quantum conditions can be written down as

m(w + - ^dr = <§RJr = rji, (7.8)

f^2m(w + ^f - — ~ - Vf^dr = §R<dr = rji. (7.9)

The total quantum numbers to be associated with r0 and r< will, as usual,
be given by
n0 = k + r0 and n< == k + r<. (7.10)

Since the electron does not complete either of the two ellipses in
one cycle, the integrals of Eqs. (7.8) and (7.9) are not to be evaluated
over a complete cycle as indicated but over only that part of the ellipse
actually traversed. The radial quantum number r for the actual path
traversed is therefore given by the sum of the two integrals
108 INTRODUCTION TO ATOMIC SPECTRA [Chap. VII

f
•fouteide
Rodr + f
•/inside
R4r = rh. (7.11)

The total energy in the outside region by Eqs. (2.14), (2.15), (2.30), (2.33),
and (7.4) is

f = -r = - = = Zl* (7.12)
2 2r 2 o,n*

where ai is the radius of the first Bohr circular orbit.

The total energy inside is
zy (Zj. - Z,)e*
(7.13)
2ain? p

Since the energy inside and outside must be the same,

zy = (Zt - Z0)e2
(7.14)
2a in2 2ainf p

Consider now the special case shown in Fig. 7.7, in which the two
partial Kepler ellipses are almost complete.1 If the outside orbit were
Penetrating Orbit
a complete Kepler ellipse the electron
would never penetrate the shell,
whereas if the inner orbit were com¬
plete the electron would always remain
inside. As the outer ellipse is made
less and less penetrating, the two
ellipses become more and more com¬
plete and the integrals of Eq. (7.11)
approach those of Eqs. (7.8) and (7.9).
Expressing this in terms of the radial
quantum numbers,
Flo. 7.7.—Special case where inner r = r0 + 7». (7.15)
and outer ellipses are almost complete.
Core charge distributed uniformly over In a similar fashion the perihelion
the surface of a sphere. distance of the outer orbit a0( 1 — Co)
approaches the aphelion of the inner ellipse a<( 1 + c<), both approaching
at the same time the radius of the spherical shell p. We write, therefore,

a.(l - t0) = a,(l + «,) = P, e» = -r — 1. (7.16)

In terms of the quantum numbers [Eqs. (3.23), (7.10), and (7.15)]

n — n0 = ni — k = p. (7.17)

This difference n — n0 is the so-called quantum defect and n0 is the Ryd¬

berg denominator nM. The semimajor axis and eccentricity of the
1 See Van Urk, A. T., Zeit*. f. Phya., IS, 268, 1923.
Sec. 7.4] PENETRATING AND NONPENETRATING ORBITS 109

inner ellipse are by Eqs. (3.32) and (3.22), respectively, ,

a» = avj£- and €? = 1 — ~ (7.18)

Substituting in Eq. (7.16) and squaring, there results

from which
m = pZx(2paiZi - a2fc2)~*. (7.20)

Replacing k2 and k by the corresponding quantum-mechanical values

1(1 + 1) and l + \y respectively [see Eq. (4.52)], and substituting in
Eq. (7.17), the quantum defect becomes

M - pZt{2paiZi - a\l(l + 1)}-* - Q + *). (7.21)

This expresses the experimental result, well known before the quantum
mechanics, that for a given atom the quantum defect n is a function
of the azimuthal quantum number and is independent of the total
quantum number (see Table 7.2).

Table 7.2.—Experimental Values of the Quantum Defect n - n, * /» foe

Penetrating Orbits in Lithium and Sodium

Element Term Electron n = 2 n = 3 n = 4

KB
Li s, l = 0 0.41 0.40 0.35 1 0.35 0.35

Na s
* s, l = 0 1 37 1.36 1 35 1.35
2p p, l — 1 0.88 0 87 0 86 0.86
1

In order to calculate quantum defects from Eq. (7.21), it is necessary

to determine values for the electron-nuclear distance p. Successful
attempts to calculate suitable values of p from quantum-mechanical
density-distribution curves similar to those shown in Figs. 7.1 and 7.2
have been made by Pauling.1 In the case of Na, Pauling obtains the
average value of the electron-nuclear distance pK for the two K electrons
as 0.132ai, and for the eight L electrons ph = 0.77ai. Assuming that
the valence electron penetrates only the outermost shell of eight electrons,
the effective nuclear charge Zxe will be 9e. Substituting Zx and pL
in Eq. (7.21), the values of p = 1.36 and p = 0.85 are obtained for the 8
and p orbits of sodium, in very good agreement with the observed values
given in Table 7.2. For lithium with Zt = 3 and pK = 0.53ai the value

1 Pauling, L., Proc. Roy. Soc.t A, 114, 181, 1927.

110 INTRODUCTION TO ATOMIC SPECTRA [Chap. VII

fi = 0.89 is computed for the $ orbits, also in good agreement with

experiment.
7.5. Quantum-mechanical Model for Penetrating Orbits.—The solu¬
tions of the Schrodinger wave equation for hydrogen-like atoms and
for the normal state of the helium-like and lithium-like atoms have been
carried out to a high degree of accuracy. The calculations for lithium¬
like atoms are of particular interest in that they appear to give the
quantum-mechanical analogue of the inner and outer orbit segments, so
well known on the classical theory and treated in the last section.
Numerous attempts have been made to calculate the energy of the
lithium atom in its normal state 2S. Perhaps the most recent and
accurate calculations are those made by Wilson.1 In previous deter¬
minations single-electron wave functions of the hydrogen type were
used instead of a wave function for the atom as a whole. The rather
brief treatment to be given here is that of Wilson based upon a principle
previously introduced by Slater.2
Slater has shown that a wave function representing an atom con¬
taining many electrons may be properly constructed by expressing it
in the form of a determinant the elements of which are built up out
of the Schrodinger wave equation. In the case of lithium, for example,
the wave function is written

1 A\ A2 A3
* = A1 A2 A31 (7.22)
V6 Bi Bz
where the A's and A*s are Is hydrogen-like wave functions for the
two K electrons and the IPs are 2s wave functions for the L electron.
From Eq. (4.55) the solutions of the wave equation for Is and 2s
electrons are
A - +u = Nitr*% (7.23)
/w Zr Zr\
B « - NsjfnT * - e~^y (7.24)

where N1 = — (Z3/^)*, N2 = (Z3/8tt)*, Z is the atomic number, and ah

the radius of the first Bohr circular orbit, has been chosen as unit of
length. Now the 2s function [Eq. (7.24)] for the valence electron is
orthogonal to the Is functions [Eq. (7.23)] for the K electrons. This
would no longer be so, if different values for Z should be substituted
in Eqs. (7.23) and (7.24). For this case Slater3 has shown that the
two functions may be orthogonalized by adding a fraction of one of the

1 Wilson, E. B., Jour. Chem. Phys., 1, 210, 1933. For other references see this
paper.
* Slater, J. C., Phys. Rev., 34, 1293, 1928.
* Slater, J. C., Phys. Rev., 42, 33, 1932.
Sec. 7.5] PENETRATING AND NONPENETRATING ORBITS 111

Is functions to the 2s function. Now a determinant possesses the prop*

erty that when any row is multiplied by any factor and added to any
other row the determinant remains unchanged in value. Wilson's
normalized wave function for the lithium atom therefore takes the form

1
Ai At At
* A,
\/6 Bi + bAi
At At > (7.25)
Bt + bAt Bz *+• bAt

where the A1s and Bfs now represent the simple hydrogen-like wave
functions with modified values for Z.
When determinants involving hydrogen-like functions with Z put
equal to the atomic number are used to calculate the energy, the result

is not so good as might be expected. Much better agreement is obtained

by considering the Z’s as parameters and adjusting them, by variation
methods, until a minimum energy is obtained. Wilson found the lowest
energy value with the wave functions and their parameters as follows:

Ai = ils = (7.26)
/y — (Z — m)rt — (Z—vi)r% — (Z—ai)rA

Bi + bA( - JV2|^-^r,e 2 - e 2 + be >, (7.27)

where <ri = 0.31, <r2 = 1.67, 0-3 = 0, and N1 and Nt are normalizing
factors. The probability-density-distribution curve obtained by plotting
4?rr2^2 against the electron-nuclear distance r is shown by the heavy
curve in Fig. 7.8. The dotted curve for the hydrogen 2s state does
not rightly belong here but is shown as a comparison with the 2s state of
lithium. The latter curve is obtained by plotting 4irrVi««
The pulling in of the inner loop of the Li 2s curve over the inner
loop of the hydrogen 2s curve is due to the lack of screening by the Is
electrons of the core and is to be compared with the deeper penetration
112 INTRODUCTION TO ATOMIC SPECTRA [Chap. VII

and speeding up of the electron in the inner part of the classical orbit.
The 2s electron, since it is most of the time well outside the core, is
screened from the nucleus by the two core electrons. This average
screening is well represented by the screening constant <r* = 1.67. The
screening of each Is electron from the nucleus by the other Is electron
should lie between 0 and 1. The value <ri = 0.31 is in good agreement
with this. When the 2s electron is inside the core (i.e., the smaller
loop), the screening by the outer Is electrons is practically negligible.
The value <r* = 0 is in good agreement with this. This same analogy
between the quantum-mechanical model and the orbital model should
extend to all elements.
The accuracy with which Eq. (7.25) represents the normal states of
lithium and singly ionized beryllium (see Figs. 5.2 and 6.5) is shown by
the following values:

Spectroscopic Calculated

Li I, 28 T - 43484 cm"1 43089 cm”1

Be II, 28 T = 146880 145984

Calculations of the quantum defect for the alkali metals from a

purely theoretical standpoint have also been made by Hartree.1 Har-
tree, employing his self-consistent field theory (see Sec. 7.1), has determined
the quantum defects and energy levels for a number of states in several
of the alkalis. After determining a probability-density-distribution
curve for the core of the atom, as in Fig. 7.1, the energy of the valence
electron moving in this field can be calculated. In rubidium, for example,
he obtained the following values of n:

Table 7.3.—Observed and Calculated Values op the Quantum Defect

n for Rubidium
(After Hartree)

Electron term m (obs.) n (calc.) Diff.

5« 5**5 3 195 3.008 0.187

6« 6**5 3 153 2.987 0.166
78 7**5 3.146 2 983 0.163
5p 5 *P 2.71 2.54 0.17
6p 6 *P 2.68 2.51 0.17
Ad 4*Z> 0.233 0.028 0 205

Since the quantum defect is a measure of the penetration of the

valence electron into the atom core, the orbital model as well as the
quantum-mechanical model would be expected to show that the pene-
1 Hartree, D. R.y Proc, Camb. Phil. Soc.t 24, 89, 111, 1928.
Sue. 7.5] PENSTRATI NO AND NONPEN STRATI NO ORBITS 113

tration is not greatly different for all states of the same series. The,
first seven of a series of d states, l * 2, for example, are shown in Fig.
7.9. The orbits given above are drawn according to model a* and
beneath them are drawn the hydrogen probability-density-distribution
curves of the same quantum numbers.
On the orbital model the perihelion distances are very nearly the
same. On the quantum-mechanical model the lengths of the first

Fig. 7.9.—Series of d orbits illustrating nearly equal penetration for all orbits with the same
l on either the classical or the quantum-mechanical model.

loop (indicated by the first nodal points), with the exception of 3d, are
nearly the same. Figure 7.9 brings out better than any other, perhaps,
the close analogies that may be drawn between the orbits of the early
quantum theory and the probability-density-distribution curves of the
newer quantum mechanics. This is one of the reasons why the terms
orbit, penetrating orbit, nonpenetrating orbit, etc., are still used in discussing
quantum-mechanical processes.

Problems
1. Assuming that the s and p orbits in potassium penetrate only the shell of eight
M electrons, compute the value of pM from Eq. (7.21). Compare these values of p with
the density-distribution curve in Fig. 7.2. If the values of the quantum defect are no*
known from Prob. 2, Chap. V, they are readily calculated from the term values
directly.
.2 Compute, on the classical theory, model a, the maximum electron-nuclear
distances attained by the valence electron of sodium in the first 10 8 orbits. Compare
these values with the corresponding s orbits of hydrogen by plotting a graph. Plot
Turn* against n for hydrogen and for sodium.
 CHAPTER VIII

DOUBLET FINE STRUCTURE AND THE SPINNING ELECTRON

From the very earliest observations of spectral series it has been

known that each member of certain general types of series shows fine
structure while those of others do not. Each member of some of the
series in the alkali metals, for example, is a close doublet (see Fig. 1.9),
whereas in the alkaline earths it is a close triplet (see Fig. 1.10). So
far as energy levels are concerned, term analyses of so-called doublet
spectra show that P, D, F,Gf . . . levels are probably double, whereas $
levels are always single (see Figs. 5.1, 5.2, 6.4, and 6.5). It is with this
fine structure that we are concerned in this chapter, for in its explanation
we are led to a new concept, the spinning electron.
8.1. Observed Doublet Fine Structure in the Alkali Metals and the
Boron Group of Elements.—Before taking up the question of the origin
of fine structure, it is important that we become more or less familiar
with the doublet spectra of the alkalies and another group of elements
not yet considered. These are the elements in the third group of the
periodiJf>‘tables (Table 5.1), boron, aluminum, gallium, indium, and
thallium. Like the alkali metals each of these atoms gives rise to four
chief series of spectrum lines: sharp, principal, diffuse, and fundamental.
From these observed series, energy level diagrams are finally constructed
as shown in Fig. 8.1. The chief difference between the alkali metals and
what we shall hereafter call the boron group of elements is best described
in terms of the energy levels rather than in terms of the spectrum lines.
In the alkali metals a 2S term lies deepest, followed by a 2P term
as the first excited state (see Fig. 5.2); whereas, in the boron group a 2P
term lies deepest, followed by 2S. This observed experimental fact
is the basis, in the Bohr-Stoner scheme of the building up of the elements,
for the addition of p electrons in these elements (see Table 5.1). The
complete electron configurations for the boron group are written as
follows:
B 5, Is2 2«2 2p
A1 13, Is2 2s2 2p* 3s2 Zp
Ga 31, Is2 2s2 23s* 3p* 3d10 4s2 4jfr
In 49, Is2 2s2 2p* 3s2 33d10 4s2 4p« 4d10 5s2 6p
T1 81, Is2 2s2 23s2 3p* 3d10 4s2 4p‘ 4d10 4/14 5s2 5p* 5d10 6s2 Zp
All other subshells being complete, it is the single unbalanced p electron
which, when the atom is excited, takes on the various possible energy
114
Sac. 8.1] DOUBLET FINE STRUCTURE 115

states shown in Fig. 8.1. In boron, for example, the first excited 2S,
state finds the electron in a 38 orbit, the Is and 2s subshell being already
filled. There being no d electrons in boron and no virtual d orbits
with a total quantum number n lower than 3, the excitation of the valence
electron to the first 2D state places the electron into a 3d orbit.
Neglecting for the moment the doublet nature of the different levels,
it should be noted that the 2F terms in all five elements, just as in the
alkali metals, are nearly hydrogen-like, indicating nonpenetrating

In
2S 2P 2d 2f 2s 2p 2d V 2s 2p 2d 2f £s 2p 2d 2f h 2p 2d 2f
t {Hi51

Fia. 8.1.—Energy level diagrams of the boron group of etments. *

X t
orbits. It is to be noted too that for a given total quantum*^^ n
the order of binding, which is also the order of increasing penetration, is
2P, ID, 2P, and 2S.
The largest fine-structure separations in each element of the boron
group are to be found in the normal states 2P. The very narrow 2P*-2Pg
interval in boron widens with each succeeding element until in thallium
it has become almost as large as the gross-structure interval 62P-72&
The first sharp series doublets given in Table 8.1 indicate the enormous
spreading out of the first 2P terms in going to higher atomic numbers. One
line of the thallium doublet is in the visible green region of the spectrum
and the other jnember is in the ultra-violet over 1500 k away.
Exactly this same spreading out of the fine structure is observed in
the alkali metals in going from lithium to caesium. This may be seen
directly from the principal series as they are plotted in Fig. 8.2. The
Bodium doublets are relatively much narrower than shown. The fine-
116 INTRODUCTION TO ATOMIC SPECTRA [Chap. VIII

Table 8.1.—Doublet Separations for the First Members of the Sharp Series

»PrVJ, >pr*5j AX Ap
X A cm~l

2497.82 2496.87 0.95 5.20

B
3961.68 3944 16 17.52 112.07
A1
4172 22 4033 18 139.04 826.10
Ga
4511 44 4101.87 409.57 2212.63
In
5350 65 3775.87 1574.78 7792.42
Tl

structure intervals for the first member of the principal series in each of
the alkalies and ionized alkaline earths are observed as follows:

Li I Na I K I Rb I Cs I
0.338 17.2 57.9 237.7 554.0 cm-1

Bell Mg II Ca II Sr II Ba II
6.61 91.5 223.0 800.0 1691.0 cm-1

In lithium the 22S-22P interval (see Fig. 5.1) is over 4000 times the
fine-structure interval 0.338 cm-1, whereas in caesium the corresponding
interval is only 20 times larger.
Principal Series

ill
n: i _ Na

■■■MM ill
— 1 1 7 .. Rb

mu_
-4-- Js =L= i 4-4-
Cs

4Q000 30,000 cm-1 20,000 10,000

-«—Radtated Frequency-
Fia. 8.2.—Illustrating fine structure in the principal series of the alkali metals. (Sodium
and potassium doublets are narrower than shown.)

A general survey of the energy level diagrams (Figs. 5.1, 5.2, 6.4, and
6.5) will enable certain general conclusions concerning fine-structure
intervals to be drawn: First, corresponding doublet separations increase
with atomic number. Second, doublet separations in the ionized
alkaline earths are larger than the corresponding doublets in the alkali
metals. Third, within each element doublet separations decrease in
going to higher members of a series. Fourth, within each element
 Sac. 8.2] DOUBLET FINE STRUCTURE 117

P doublets are wider than D doublets of the same n, and D doublets

are wider than F doublets of the same n. The last two statements
are illustrated schematically in Fig. 8.3 by giving the four lowest members
of the three chief term series. It should be mentioned in passing that
exceptions to these rules are well known. These exceptions, however,
are fully accounted for and will be treated as special cases in Chap. XIX.
*P-- ..- *D- 2F-

2P— 2d . — ■■■ 2f

*P ?D =-2F

F-O. 8.3.—Schematic representation of relative term separations in the different series

8.2. Selection Rules for Doublets.—In the doublet spectra of atomic

systems containing but one valence electron the small letters s, p, d,
/, . . . , for the different electron orbits are replaced by the corresponding
capitals S, P, D, F} . . . , for the terms. The small superscript 2 in
front of each term indicates that the level in question, including S levels,
has doublet properties and belongs to a doublet system. Although all
S levels are single, their doublet nature will later be seen to reveal itself
when the atom is placed in a magnetic field. In order to distinguish
between two fine-structure levels having the same n and l values, the
cumbrous but theoretically important half-integral subscripts are used.
This subscript to each term, first called the inner quantum number by
Sommerfeld, is of importance in atomic structure, for it gives the total
angular momentum of all the extranuclear electrons (see Sec. 8.4).
The inner quantum number is frequently referred to as the electron
quantum number j or the term quantum number J.
Observation shows that, for the transition of an electron from one
energy state to another, definite selection rules are in operation. This
is illustrated schematically in Fig. 8.4 by six different sets of com¬
binations. From these diagrams, which are based upon experimental
observations, selection rules for doublets may be summarized as follows:
In any electron transition1
l changes by +1, or —1 only,
and
j changes by 0, +1, or —1 only.

1 Violations of either of these selection rules are attributed to the presence of an

external electric or magnetic field (see Chaps. X and XX) or to quadripole radiation.
118 INTRODUCTION TO ATOMIC SPECTRA [Chap. VIII
The total quantum number n has no restrictions and may change by any
integral amount. The relative intensities of the radiated spectrum
lines are illustrated by the heights of the lines directly below each transi¬
tion arrow at the bottom of the figure. Combinations between 2P
and 2S always give rise to a fine-structure doublet, whereas all other
combinations give rise to a doublet and one satellite. In some doublet
spectra, HI and 2H terms are known. In designating any spectrum
line like X5890 of sodium (see Fig. 5.1), the lower state is written first
followed by the higher state thus, 32$$-32jPj. The reason for this order
goes back to the very earliest work in atomic spectra (see Chap. I). Spec¬
trum lines in absorption are written in the same way, the lowest level first.

aT ;%L_ AF
AP *AP ’off

$1 AP AP
AD
Fvr
± JL JL
AP AP AP aq-p—vaDap ADaf AFad
Fia. 8.4.—Illustrating selection and intensity rules for doublet combinations.

8.3. Intensity Rules for Fine-structure Doublets.—General observa¬

tions of line intensities in doublet spectra show that certain intensity
rules may be formulated. These intensity rules are best stated in terms
of the quantum numbers of the electron in the initial and final energy
states involved. The strongest lines in any doublet arise from transitions
in which j and l change in the same way. When there is more than one
such line in the same doublet, the line involving the largest j values is
strongest. For example, in the first principal-series doublet of Fig. 8.4. the
line is stronger than 2£j-2P* since in the former l changes by
— 1 (Z = 1 to l = 0) and j changes by — 1 (j = $ to j = £). As a second
example, consider a member of the diffuse series in which there are two
strong lines and one satellite. For the two strong lines 2P2-2Z>| and
2P\-~2Dh j and l both change by —1. The stronger of the two lines
2P|-2Dj involves the larger j values. For the faint satellite 2P%-2Dh
AZ = — 1 and Aj = 0.
Quantitative rules for the relative intensities of spectrum lines
were discovered by Burger, Dorgelo, and Ornstein.1 While these rules
,
1 Burger, H. C., and H. B. Dorgelo, Zeits. /. Pkya., 23 258, 1924; Ornstein,
, ,
L. S., and H. C. Burger, Zeite. f. Phys., 24 41, 1924; 22 170, 1924.
Sbc. 8.3] DOUBLET FINE STRUCTURE 119

apply to all spectra in general, they will be stated here for doublets
only, (a) The sum of the intensities of those lines of a doublet winch
come from a common initial level is proportional to the quantum weight

Table 8.2.—Intensity Measurements in the Principal Series op the Alkali

Metals
(After Sambursky)

Element Combination Wave-lengths Ratio

32S-32P 5890 : 5896 2 : 1

Na 32S-42P 3302 : 3303 2 : 1
32>S-52P 2852 : 2853 2 : 1

42S~42P 7665 : 7699 2 : 1

42S-52P 4044 : 4047 2.2 : 1
K 42S-62P 3446 : 3447 2.3 : 1
42S-72P 3217 : 3218 2 5:1

52S-52P 7800 : 7947 2 : 1

52S-62P 4201 : 4215 2.7 : 1
52/S-72P 3587 : 3591 3.5 : 1
Rb 52,S-82P 3348 : 3351 4.3 : 1
52jS-92P 3228 : 3229 5 : 1
52S-102P 3157 : 3158 3 : 1

62-S-62P 8521 : 8943 2 : 1

62<S-7!P 4555 : 4593 5 : 1
62<S-82P 3876 : 3888 10 : 1
62<S-92P 3611 : 3617 15 5 : 1
Cs 62£-102P 3476 : 3480 25.0 : 1
0VM12P 3398 : 3400 15.8 : 1
6-'<S-122P 3347 : 3348 5.7 : 1
62<S-132P 3313 : 3314 4 5:1

of that level. (6) The sum of the intensities of those lines of a doublet
which end on a common level is proportional to the quantum weight
of that level. The quantum weight of a level is given by 2j + 1 .
This,
it will be seen in Chap. X, is the number of Zeeman levels into which a
level j is split when the atom is placed in a magnetic field.
In applying these intensity rules, consider again the simple case
of a principal-series doublet. Here there are two lines starting from
the upper levels 2P\ and 2P\ and ending on the common lower level
2S$. The quantum weights of the 2P levels are 2(f) + 1 and 2(£) + 1,
giving as the intensity ratio 2:1. The same ratio results when the 2S
level is above and the 2P level below.
120 INTRODUCTION TO ATOMIC SPECTRA {Chap. VIII

Quantitative measurements of line intensities in some of the alkali

spectra are given in Table 8.2.1
The particular investigations of Sambursky on the principal series
of Na, K, Rb, and Cs show that, while the first members have, in agree¬
ment with observations made by others, the theoretical ratio 2:1, higher
members do not. This is expecially true in caesium where the observa¬
tions have been extended to the eighth member. In caesium the intensity
ratio starts with 2 and rises to a maximum of 25 in the fifth member, then
drops quite abruptly to in the eighth member.2 Consider next the
diffuse-series doublets which involve three spectrum lines. The following
combination scheme is found to be particularly useful in representing
all of the transitions between initial and final states. A diffuse-series
doublet is written

2P,
4 2
2D,6 X 0
2D,4 Y Z

The numbers directly below and to the right of the term symbols are
the quantum weights 2j + 1. Let X, F, and Z represent the unknown
intensities of the three allowed transitions and zero the forbidden transi¬
tion. From the summation rules (a) and (b) the following relations
are set up: The sum of the lines starting from 2Z>S is to the sum starting
X
from 2Z)j as 6 to 4, i.e., and, similarly, the sum of the lines
Y +Z
X + Y 4
ending on 2Pj is to the sum ending on 2P* as 4 is to 2, i.e.,
Z 2
The smallest whole numbers which satisfy these equations are X = 9,
F*l, and Z = 5. If the 2Z> terms are very close together so that the
observed lines do not resolve the satellite from the main line, as is usually
the case, the two lines observed will have the intensity ratio 9+1:5
or 2:1, the same as the principal-series or sharp-series doublets. Intensity
measurements of the diffuse series of the alkali metals by Dorgelo8 confirm
this.
A favorable spectrum in which the satellite of a diffuse-series doublet
can be easily resolved, with ordinary instruments, is that of caesium.
The first three members of this series are in the infra-red and are not
readily accessible to photography. The fourth member of the series,
composed of the three lines XX 6213, 6011, and 6218 has been observed

1 Sambursky, S., Zeits. f. Phys., 40, 731, 1928.

* The anomalous intensities observed in caesium have been given a satisfactory
explanation by E. Fermi, Zeits. f. Phys., 50, 680, 1930.
* Dorgelo, H. B., Zeits. /. Phys., 22, 170, 1924.
Sue. 8.4] DOUBLET FINE STRUCTURE 121
Table 8.3.—Intensity Measurements in The Diffuse Series of the Alkali
Metals
{After Dorgelo)

Element Combination Wave-lengths Ratio

32P-42D 5688 : 5682 100 : 50

Na
32P-52Z) 4982 : 4978 100 : 50

4»P-5*£> 5832 : 5812 100 : 51

K
4*P-62D 5359 : 5342 100 : 50

52P-62Z) 6298 : 6206 100 : 51

Rb ! 52P-72D 5724 : 5648 100 : 52
52P-82D 5431 : 5362 100 : 52

to haye the intensity ratios 9:5.05:1.17. Theoretical intensities for the

combination 2D-2F are given by the following formulations:
'Di
4 1 6 X 8 6 4
2F\ 8 X~' 0 Y + Z 6 2F, 8 20 0
*F% 6 Y Z X + Y 6 v, 6 1 14
! z 4
The smallest whole numbers satisfying the equations in the center are
X = 20, and Z = 14. The results given in Table 8.2 show that one
cannot always expect the intensity rules to hold. The theoretical
intensities are extremely useful, however, in making identifications in
spectra not yet analyzed.
8.4. The Spinning Electron and the Vector Model.—With the
co-development of complex spectrum analysis and the Land6 vector
model, it became necessary to ascribe to each atom an angular momentum
in addition to the orbital angular momentum of the valence electrons.
At first this new angular momentum was ascribed to the atom core
and assigned various values suitable for the proper explanation of the
various types of spectral lines: singlets, doublets, triplets, quartets,
quintets, etc. Due to the insight of two Dutch physicists, Uhlenbeck
and Goudsmit,1 and independently Bichowsky and Urey,2 this new
angular momentum was assigned to the valence electrons. In order
to account for doublet fine structure in the alkali metals, it is sufficient
to ascribe to the single valence electron a spin s of only one-half a quantum

unit of angular momentum, s— = ^ • — This half-integral spin is

not to be taken as a quantum number that takes different values like n

1 Uhlenbeck, G. E., and S. Goudsmit, Naturwissenschaften, 13, 953, 1925; Nature,
117, 264, 1926.
* Bichowsky, F. R., and H. C. Urey, Proc. Nat. Acad. Sci., 12, 80, 1926.
122 INTRODUCTION TO ATOMIC SPECTRA [Chap, VIII

and l but as an inherent and fixed property of the electron. The total
angular momentum contributed to any atom by a single valence electron
is therefore made up of two parts: one due to the motion of the center

Fig. 8.5.—Spin and orbital motion of the electron on the classical theory.

of mass of the electron around the nucleus in an orbit, and the other
due to the spin motion of the electron about an axis through its center of
mass (see Fig. 8.5). Disregarding nuclear spin the atom core, as we shall
see later, contributes nothing to the total angular momentum of the atom.
By analogy with the quantum-mechanical developments in Chap. IV,
we return now to the orbital models a, by c, and d (Fig. 4.8) to find a
suitable method for combining these two angular momenta. For this
Vector Vector

j-i+! j-t-i j-i+£ j'i~i

Fig. 8.6.—Vector models a and b for the composition of the electron spin and orbit.

purpose models a and b are both frequently used Of these models, a

is generally preferred since it gives in many cases, but not always, the
more accurate quantum-mechanical results. Models c and d have been
rejected because of the many fortuitous rules necessarily introduced
to fit the experimental data, and they are of historical interest only.
Sec. 8.5) DOUBLET FINE STRUCTURE 123

Vector diagrams for the composition of orbit and spin, on models a and 6,
are given in Fig. 8.6 for the two possible states of the d electron.
On model b the spin angular momentum 8 • h/2ir is added vectorially to
the orbital angular momentum Z • h/2r to form the resultant j • h/ 2vi where
j = l ± 8. On model a the spin angular momentum s* • h/2oc is added
vectorially to the orbital angular momentum Z* • h/2ir to form the resultant
j* • h/2ir, where s* = V*(* + 1), Z* = y/UJi + 1), j* = VJif+Tj, and
j = Z + 8. It should be noted that two is the maximum number of j
values, differing from each other by unity, that are possible on either
model and that for s electrons there is but one possibility.1 For s, p,
/, . . . orbits Z = 0, 1, 2, 3, • • • . The quantum conditions are that
j shall take all possible half-integral values only, i.e.9j = i, h %, * * • .
For a d electron Z = 2, s = j, l* = \/6, and s* — £\/3- The only
possible orientations for Z and s, or for l* and s*, are such that j = %
and j, and j* = and For a p electron l = 1, s = |,
Z* = \/2, and s* = |\/3. The only possible orientations for Z and s, or
for Z* and s*, are such that j = f and J, and j* — |\/15 and For
an s electron Z = 0, s = i, l* = 0, and s* = %\/3. The only possible
value for j is and j* = i\/3.
..
8 6 The Normal Order of Fine-structure Doublets.—In the doublet
energy levels of atomic systems containing but one valence electron
it is generally, but not always, observed that the fine-structure level
j — l — \ lies deeper than the corresponding level j — Z + J. For
example, in the case of p and d electrons, 2Pj lies deeper than 2P| and
2D} lies deeper than 2Dj. This result is to be expected on the classical
theory of a spinning electron and on the quantum mechanics. Classically
we may think of the electron as having an orbital angular momentum
and a spin angular momentum. Due to the charge on the electron
each of these two motions produces magnetic fields.
Due to the orbital motion of the electron, of charge 6, in the radial
electric field of the nucleus Ef there will be a magnetic field H at the
electron normal to the plane of the orbit.2 That this field is in the direc¬
tion of the orbital mechanical moment Z may readily be seen by imagining
the electron at rest and the positively charged nucleus moving in an
orbit around it. In this field the more stable state of a given doublet
will then be the one in which the spinning electron, thought of as a
small magnet of moment lines up in the direction of H. In Fig.
8.7 the electron spin moment is seen to be parallel to H in the state
j = Z — J, and antiparallel to H in the state j = Z + J. Of the two
1 The small letter 8 used for electron spin must not be confused with the small
letter for 8 electrons.
1 This field is not apparent to an observer at rest with the nucleus but would be
experienced by an observer on the electron.
124 INTRODUCTION TO ATOMIC SPECTRA [Chap. VIII

possible orientations the one j * l — \ is classically then the more

stable and therefore lies deeper on an energy level diagram. On the
vector model a (see Fig. 8.6) the same conclusion is reached and we can
say that of the two states the one for which the spin moment /*, is more
nearly parallel to H lies deeper.
Although most of the doublet spectra of one-electron systems are
in agreement with this, there are a few exceptions to the rule. In
caesium, for example, the 2P and 2D terms are normal, and the first 2F
term is inverted. By inverted is meant that the term j — l + i lies
deepest on an energy level diagram. In rubidium the 2P terms are
normal, and the 2D and 2F terms are inverted. In sodium and potassium

A j-l + s B j*l-s
Fig. 8.7.—Illustrating the mechanical and magnetic moments of the spinning electron
for the two fine-structure states j ** Z -f \ and j — l — J. The vectors are drawn accord¬
ing to the classical model (b).

the 2P terms are normal, and the 2D, and very probably the 2F, terms
are inverted. For a possible explanation of the inversions see Sec. 19.6.
It can be shown quantum mechanically that, neglecting disturbing
influences, doublet levels arising from a single valence electron will be
normal. Where resolved, all of the observed doublets of the boron
group of elements, the ionized alkaline earths, and the more highly
ionized atoms of the same type are in agreement with this.
..
8 6 Electron Spin-orbit Interaction.—The problem which next
presents itfeelf is that of calculating the magnitude of the doublet separa¬
tions. Experimentally we have seen that doublet-term separations, in
general, vary from element to element, from series to series, and from
member to member. Any expression for these separations will there¬
fore involve the atomic number Z, the quantum number l9 and the
quantum number n.
A calculation of the interaction energy due to the addition of an
electron spin to the atom model has been made on the quantum mechanics
by Pauli,1 Darwin,2 Dirac,8 Gordon,4 and others. By use of the vector

1 P\uli, W., Zeits.f. Phys., 43, 601, 1927.

* Dabwin, C. G., Proc. Roy. Soc., A, 116, 227, 1927; A, 113, 664, 1928.
* Dirac, P. A. M., Proc. Roy. Soc., A, 117, 610,1927; A, 118,351,1928.
* Gordon, W., Zeite. /. Phya., 48, 11, 1929.
Sue. 8.6) DOUBLET FINS STRUCTURE 125

model a semiclassical calculation of the interaction energy may also be

made which leads to the same result. Because of its simplicity this
treatment will be given here. Dirac's quantum-mechanical treatment
of the electron will be given in the next chapter.
On the classical model of hydrogen-like atoms the single electron
moves in a central force field with an orbital angular momentum

1*^k = mr X v (8.1)
where l* = \/l(l + 1), m is the mass of the electron, v its velocity, and
r the radius vector. According to classical electromagnetic theory, a
charge Ze on the nucleus gives rise to an electric field E at the electron
given by

Ze
E = -yr. (8.2)
r3

Mo^;ng in this field the electron experiences a magnetic field given by

h-EXv- (8.3)
c
From these two equations,

H = —t X v. (8.4)
crs

Applying Bohr's quantum assumption [Eq. (2.20)]

2rmi X v = Z* h, (8.5)
the field becomes

H = A.££.^. (8.6)
2t me r3

In this field the spinning electron, like a small magnetic top, under¬
goes a Larmor precession around the field direction. From Larmor's
theorem [Eq. (3.58)] the angular velocity of this precession should be
given by the product of the field strength H and the ratio between the
magnetic and mechanical moment of the spinning electron:1

e Ze . 1 . 2_e_
H * 2 (8.7)
2 me me r3 2mc

This is just twice the ordinary Larmor precession f given in Eq. (3.58)

1 The ratio between the magnetic and mechanical moment of a Bpinning electron
is just twice the corresponding ratio for the electron's orbital motion. This is in
agreement with results obtained on the quantum mechanics and accounts for the
anomalous Zeeman effect to be treated in Chap. X.
 126 INTRODUCTION TO ATOMIC SPECTRA [Chap. VIII

A relativistic treatment of this problem by Thomas1 has revealed, in

addition to the Larmor precession coL, a relativity precession o>r, one-half
as great and in the opposite direction. The resultant precession of the
spinning electron is therefore one-half t.e., just equal to the ordinary
Larmor precession:

= 0)L + <*T = *7*A Ze* 1 (8.8)

2 2tt m2c2 r*

Now, the interaction energy2 is just the product of the precessional

angular velocity and the projection of the spin angular momentum
on l*:

AW,,. = «' • 8* A COS

lir
(1*8*). (8.9)

With the value of from Eq. (8.8),

(8.10)

In this equation for the interaction energy the last two factors are
still to be evaluated. In general the electron-nuclear distance r is a
function of Z, n, and l and changes continually in any given state.
Because the interaction energy is small compared with the total energy
of the electron's motion the average energy AWi,a may be calculated by
means of perturbation theory. In doing this, only the average value

need be calculated. From perturbation theory and the quantum

mechanics (see Sec. 4.9),

(A & (8.11)
W a\nH{l + m+ 1)'

where ax is the radius of >the first Bohr circular orbit,

= h2 (8.12)
4w2me2

For the last factor of Eq. (8.10) we turn to the vector model of the
atom. In calculating the precessional frequency of s* around the field
produced by the orbital motion the vector l* was assumed fixed in space.

1 Thomas, L. H., Nature, 117, 514, 1926.

* The interaction energy here is just the kinetic energy of the electron's precession
around the field H. If W * i/to* represents the kinetic energy of the spin of the
electron in the absence of the field H, and W' * i/(to + to7)* the kinetic energy in the
presence of the field, then the change in energy is just AW * W' — W * i/to'* +
/toto'. Since I remains constant and to > > to', the first term is negligibly small and
the energy is given by the product of to' and the mechanical moment /to =* 8*hJ2r *■ p».
Ssc. 8.6] DOUBLET FINE STRUCTURE 127

One might equally well have calculated the precession of the orbit in tjie
field of the spinning electron. It is easily shown that this frequency is
just equal to the ordinary Larmor precession and to the precession of the
electron around Z*. In field-free space both orbit and spin are free to
move so that l* and s* will precess around their
mechanical resultant j*. By the law of conserva¬ .*
J
tion of angular momentum this resultant j* and
hence the angle between l* and s* must remain
invariant. The vector model therefore takes the
form shown in Fig. 8.8. With the angle fixed the
cosine does not need to be averaged and l*s* cos
(1*8*) is calculated by the use of the cosme law
j* = Z*2 + s*2 + 2Z*s* cos (1*8*) (8.13)
from which
j*2 - Z*2 - 8*
1*8* COS (1*8*) = (8.14)

Substituting Eqs. (8.11) and (8.14) in Eq. (8.10),

Ze2 h2 Z3
AWi,, = • f-2 Fig. 8.8.—Classical pre¬
2m2c2 4tt2 a\nH(l + $)(Z + 1) cession of electron spin »*
j** — l*2 — g*2 and orbit l* around their
-■=- (8.15) mechanical resultant/*.

Upon substituting the Rydberg constant R — 2r2me'l/c/i3 and the

square of the fine-structure constant a2 = 47rV/c2A2, the energy becomes

Ra*chZ* j** - l*2 - s*2

A Wi.. = (8.16)
n>l{l + m + D‘ 2
and, dividing by he, the term shift in wave numbers becomes

Ra*Z* j*l _ 1*2 _

AT,,. = — = -r. (8.17)
n'l(l + *)(J + 1)

This spin-orbit interaction energy, often referred to as the T factor,

is written in short as
(j** _ - ,»1)
= a ■ 1*8* COS (1*8*), (8.18)

where
RaW
cm- (8.19)
n*l(l + M+ 1)
Measured from the series limit down, the term value of any fine-
structure level will be given by

T = To - r, (8.20)
128 INTRODUCTION TO ATOMIC SPECTRA [Chap. VIII

where To is a hypothetical term value for the center of gravity of the

doublet in question and r gives the shift of each fine-structure level from
TV T values for 2P, 2D, and 2F terms are shown in Fig. 8.9. For an 8
orbit l = 0, j = 8, and AT = 0. This is in agreement with observation
that all S states are single. The two states of a doublet are seen by the
figure to be given by the difference between their T values. It is to be
noted that Z occurs in the numerator of Eq. (8.19), and n and l occur in
the denominator. This is in agreement with experimental observations

r
/

\
\

\
\
l Z

Fig. 8.9.—Illustrating fine-structure separations due to spin-orbit interaction.

that doublet-term separations (1) increase with increasing atomic number,

e.g.f in going from Na I to K I, or from Na I to Mg II; (2) decrease with
increasing n, t.e., in going to higher members of a given series; (3) decrease
with increasing Z, i.e., in going to different series p, d, /, etc. As the inter¬
action energy gets smaller and smaller in a given series, T will approach
zero and the levels will gradually come together at their center of gravity
To.
8.7. Spin-orbit Interaction for Nonpenetrating Orbits.—In applying
the spin-orbit-interaction energy formula [Eq. (8.17)], as just derived, to
the observed data, it is necessary that we first simplify the expression
and substitute the known physical constants. For any given doublet,
l and 8 have the same values whereas j = l + £ for the upper level and
Z — i for the lower level. The successive substitution of these values
for j in the last factor of Eq. (8.17) gives, by subtraction,

Ra2ZA Ra2ZA
Av cm-1. (8.21)
nH{l + \){l + 1) nH(l + 1)

Inserting the value of the Rydberg constant R — 109737 cm 1 and the

fine-structure constant a2 — 5.3 X 10“6,

(IU2)
For hydrogen-like systems the effective nuclear charge is given
simply by the atomic number Z.
Sic. 8.8) DOUBLET FINE STRUCTURE * 129

In the previous chapters we have seen hoW for other atomic systems
the deviations of the term values from those of hydrogen-like atoms are
attributed to a polarization of the atomic core or generally to a quantum
defect [see Eq. (5.4)],

T - HZ2 _ (8.23)
(n - aO* nht ’
where Z * 1 for neutral atoms, 2 for singly ionized atoms, etc. Instead
of attributing the increased binding of the electron to a defect in the
quantum number n, one may argue that it should be attributed to a
screening of the valence electron from the nucleus by the intervening
core of electrons, and that Z should be replaced by Z.ff where ZM = Z — <r,
Z is the atomic number, and a is a screening constant:
RZ2M = R{Z — a)2
(8.24)
n2 n2

By exactly the same reasoning one may write the fine-structure

doublet formula [Eq. (8.22)] as
Ra2(Z - •)« _ K QO (Z - *)«
(8.25)
H(l + 1) nH(l + 1)'

Most of the doublets to which this formula applies are known only for
singly and multiply ionized atoms. Although its general application
will be left to Chap. XVII on Isoelectronic Sequences, it should be
remarked here that this formula gives doublet intervals in remarkably
good agreement with experimental observations.
8.8. Spin-orbit Interaction for Penetrating Orbits.—In the preceding
chapter on penetrating and nonpenetrating orbits we have seen how
penetrating orbits may be considered as made up of two parts, an inside
segment of an ellipse and an outside segment. In attempting to apply
Eq. (8.25) to the doublets of penetrating orbits, much better agreement
with the observed values is obtained, especially for the heavier elements,
by again considering separately the inner and the outer part of the orbit
(see Fig. 7.6). Whatever atomic model is formulated, the electron in a
deeply penetrating orbit is by far the greater part of the time in an outer
region where the field is nearly hydrogen-like. If the electron remained
in an outer orbit like the outer segment, the doublet formula [Eq. (8.21)]
would be
R**zi
A r. (8.26)
n*l(l + 1)

whereas, if it remained in an inner orbit like the inner segment, the

formula would be
Ra2Z\
A - (8.27)
nW + 1)‘
 130 INTRODUCTION TO ATOMIC SPECTRA (Chap. VIII

To bring these two formulas together for the actual orbit, the motion
in each of these segments is weighted according to the time spent in each.
Now the time t required to traverse the whole path is so nearly equal to
the time ta required to traverse a complete outer ellipse that we may
write, to a first approximation,
f = njh8
(8.28)
" 4r2me4Zf

This equation for the period of an electron in a Kepler ellipse was left
as an exercise at the end of Chap. III. The time required to traverse a
completed inner ellipse is
n\h3
(8.29)
4ar2meAZ\

Now the resultant frequency separation Av for the actual orbit will
be A Vo times the fractional time tQ/t spent in the outer segment, plus
A Vi times the fractional time tx/t spent in the inner segment:

Av = + A,f (8.30)

Making the approximation that tQ = t and substituting the values of

Av0, Avif tf and tt from Eqs. (8.26), (8.27), (8.28), and (8.29), we get

Ra2Zl
Av —
<1(1 + 1) (Z\ + Zl) cm" (8.31)

For heavy elements the effective charge Zxe at deepest penetration is so

much greater than the effective charge Zae outside, that the formula may
be simplified to

Av =
n<x2zm (8.32)
nll(! + 1).
This equation was derived from the quantum theory and used by Land6l
before the advent of the spinning electron and the newer quantum
mechanics. In calculating Z, for a number of atoms Land£ showed that
the penetration in many cases is almost complete, being almost equal
to the atomic number Z (see Table 17.4A). It is to be noted that nQ is
the effective quantum number.
Inserting screening constants for each of the Z’s, in Eq. (8.32),

Ra\Z - s0)2(Z - sx)2

Av = cm-1. (8.33)
n*l{l + 1)

The application of this formula to observed doublets, in general, will be

set aside, to be taken up again in treating isoelectronic sequences of atoms
in Chap. XVII.
‘LANPifc, A.. Zeits f Phys , 25, 46, 1924.
Sec. 8.8] DOUBLET FINE STRUCTURE 131

Problems
1. Compute doublet-term separations for the nonpenetrating 2p states of lithium
and singly ionized beryllium. Assume complete screening by the core electrons.
Compare the calculated values with the observed values given in Sec. 8.1.
2. Determine theoretical intensity ratios for the doublet transitions %Ff-*Gj.
8. Construct vector-model diagrams for JFj, *Gj and *Gj states based on model
a, Fig. 8.6.
4. Determine the electron spin-orbit precession frequency «/2t for a 4f state in
potassium. Assume complete screening by the 18 core electrons.
6. Compute a theoretical doublet separation for the 6*P state in caesium. Assume
complete penetration when the electron is inside the core, t.e., «,* ** 0, and perfect
screening when it is outside, t.c., 8P = Z — 1. The effective quantum number n»
can be determined from the observed term values (use the center of gravity of the
doublet). All other factors remaining the same, what value of *,• will give the observed
doublet separation?
 CHAPTER IX

HYDROGEN FINE STRUCTURE AND THE DIRAC ELECTRON

Even the hydrogen spectrum, the simplest of all systems, is observed

to have a fine structure. At an early date Michelson studied the Balmer
lines with an interferometer and found that both Ha and Hp were close
doublets with separations of only 0.14 and 0.08 A, or 0.32 and 0.33 cm-1,
respectively. Many subsequent investigations by others have confirmed
these results (see Fig. 9.1).

Fig. 9.1. Photographs of the Ha line of both of the hydrogen isotopes H1 and H*. (After
Lewis and Spedding.)

The most informing observations that have been made on hydrogen¬

like atoms are those of Paschen1 on the singly ionized helium line X4686.
This line (see Fig. 2.8) corresponds to the first member of the so-called
Paschen series of hydrogen. Historically Paschen’s observations were
made and published at a most opportune time, for in the next issue of
the Annalen der Physik Sommerfeld independently predicted just such a
fine structure by an extension of the Bohr atom to include elliptic orbits
and the special theory of relativity.2 While the quantum mechanics
gives a more perfect account of the observed fine structure, the develop¬
ment of the orbital model is interesting in that it leads to the same energy
levels.
9.1. Sommerfeld Relativity Correction.—The extension of Bohr’s
atomic model by Sommerfeld to include elliptic orbits adds no new
energy levels to the hydrogen atom (see Sec. 3.3). For a given total
quantum number n, all elliptic orbits 8, p, d, . . . have just the same

1 Paschen, F., Ann. d. Phys.} 60, 901, 1916.

* Sommerfeld, A., Ann. d. Phys., 61, 1, 1916.
132
Sec. 9.11 HYDROGEN FINE STRUCTURE 133

energy as the Bohr circular orbit with the same n. This energy in waive
numbers is

l- m
where R is the Rydberg constant
2t2W64

ch*( i +

h is Planck's constant, c the velocity of light, m and e the mass and charge
of the electron, and M the mass of the nucleus with charge Ze.
Bohr pointed out in his earliest papers that the relativistic change in
mass of the orbital electron should be taken into account in computing
the energy levels. Introducing elliptic orbits, Sommerfeld applied the
special theory of relativity to the
electron mass. Due to the different
velocity of the electron in orbits of
the same n but differing azimuthal
quantum number, the mass of the
electron and hence the resultant
energy levels are all different. If the
rest mass of the electron is m0, its
mass when moving with velocity v is
given by the special theory of rela¬
tivity as

= ro{1-^)1
\ / Fig. 9.2.—Schematic representation of
the precession of an electron orbit due to
As a result of this change in mass, the relativity change in mass of the elec-
, . , • , , , , tron with velocity. {After Sommerfeld.)
which is greatest at perihelion and
greatest for the most elliptic orbits, there is an advance of the perihelion,
or a precession of the electron orbit, similar to that of a penetrating orbit
in the alkali metals (see Fig. 7.4), or to that of the planet Mercury moving
about the sun. This precession is shown schematically in Fig. 9.2.
While the derivation of Sommerfeld's equation for the change in energy
due to this precession is out of place here, we shall find use for it in making
comparisons with the quantum-mechanical results.1 According to the
Sommerfeld theory the term values of hydrogen-like atoms are given by

t= = maJ i a2Z2_
—~
Ac
-|-
h\l + (n - k + y/k2 — a2Z2)2 }"+£
1 For a derivation of Sommerfeld’s relativistic fine-structure formula see “Atomic
Structure and Spectral Lines,“ p. 467,1023; also A. E. Ruark and H. C. Urey, “ Atoms,
Molecules and Quanta/’ p. 132.
134 INTRODUCTION TO ATOMIC SPECTRA [Chap. IX

where a is the fine-structure constant

a2 * 2™. = 5.30 X 10“5 (9.5)

and
Mm
** * M + m

For convenience of evaluation, Eq. (9.4) has been expanded by Som-

merfeld into a converging series,

+ • • • • (9.6)

The first term of this expansion is the same as that derived by Bohr
for circular orbits, neglecting relativity, and gives the major part of the
energy. With n = 1, 2, 3, • • • , and with Z — 1 for hydrogen, Z = 2
for ionized helium, and Z = 3 for doubly ionized lithium, this term gives
the following values:

Table 9.1.—Term Values, Neglecting Fine-structure Corrections

Hydrogen Hydrogen Ionized Doubly

(isotope mass 1) (isotope mass 2) helium iomzed lithium
R * 109677.76 R = 109707.56 R - 109722.4 R = 109728.9

For » « 1 109677.76 109707.56 438889.6 987560.1

2 27419.4 27426.9 109722.4 246890.0
3 12186.4 12189.7 48765.5 109728.9
4 6854.8 6856.7 27430.6 61722.5
5 4387.1 4388.3 17555 5 39502.4
6 3046.6 3047.4 12191 3 27432.2
7 2238.3 2238.9 8956 9 20154.4

To each of these values corrections from Eq. (9.6) must be added.

For small values of Z the first term involving Z4 and a2 is the only one
of importance and the third and succeeding terms may be neglected.
In x-ray spectra, however, Z becomes large for the heavier elements
and terms in a4 and a6 must be taken into account (see Chap. XVI).
Sec. 0.2] HYDROGEN FINE STRUCTURE 135

The corrections to be added to each of the above given terms are there¬
fore given by

where k is Sommerfeld’s azimuthal quantum number 1, 2, 3, . . . for

s, p, d, . . . . For all allowed values of n and k the correction is positive
and is to be added to the terms in Table 9.1. For either of the two
hydrogen isotopes,

<U>

These corrections are shown graphically in Fig. 9.3. The straight

lines at the top of each of the four diagrams represent the first four terms

n-l n-2 n-3 n-4

Fiq. 9.3.—Fine structure of the hydrogen energy levels. A TV and AT*i,« represent
the relativity and the spin-orbit corrections respectively. The dashed linea represent
Sommerfeld’s relativity corrections.

of hydrogen given by Table 9.1. The shifted levels for each value of
n and k are shown by the dotted lines with the term value increasing
downward. The left-hand side of each diagram has to do with the
spinning-electron picture of the atom and will be taken up in the following
section. For ionized helium and doubly ionized lithium the intervals
given in Fig. 9.3 must be multiplied by 16 and 81, respectively.
9.2. Fine Structure and the Spinning Electron.—With the introduc¬
tion of the spinning electron and the quantum mechanics another account
of the hydrogen fine structure has been given. Heisenberg and Jordan1

1 Heisenberg, W.. and P. Jordan, ZeiU. /. Phys., ST, 268, 1926.

136 INTRODUCTION TO ATOMIC SPECTRA [Chap. IX

have shown from a quantum-mechanical treatment that Sommerfeld's

relativity correction for hydrogen-like atoms should be
Ra2ZV 1_3\
ATr (9.9)
n3 \Z + £ 4nJ

where Z = 0, 1, 2, • * • for $, p, d, • • • electrons. A general compari¬

son of all classical with quantum-mechanical results for the same phe¬
nomenon shows that the classical values ky k2y and fc3 may usually, but not
always, be replaced by l + £, 1(1 + 1), and 1(1 + £)(Z + 1), respectively,
to obtain the quantum-mechanical results. Sommerfeld’s relativity
equation is a good example of this; k in Eq. (9.7) replaced by Z + £ gives
Eq. (9.9).
To the quantum-mechanical relativity correction [Eq. (9.9)] a second
term due to the spin-orbit interaction must be added. This interaction
energy has already been calculated in Sec. 8.6 and shown to be given by
Eq. (8.17) :l
Ra2Z4 - I*2 - s*2
A Ti,. =
nH(l + £)(Z + 1) 2
Applying the first correction ATr to the hydrogen terms of Table 9.1,
each level n is split into n components as shown at the left of each of the
four diagrams in Fig. 9.3. Applying now the spin-orbit interaction
ATi.,, each of these terms, with the exception of $ terms, is split into two
parts just as in the alkali metals. In each case the level with.; = Z + £
has been shifted up, and the one with j = Z — £ has been shifted down,
to the nearest Sommerfeld level. In other words, levels with the same
j values come together at the older relativity levels kt where k = j + £.
The remarkable fact that Sommerfeld's formula derived from the relativity
theory alone should give the same result as the newer theory, where both
relativity and spin are taken into account, is a good example of how
two incorrect assumptions can lead to the correct result. While the
number of numerically different energy levels is the same on both theories,
there is an experimental method for showing that the first theory is not
correct and that the latter very probably is correct. This will become
apparent in Sec. 9.4.
Since the newer theory leads to Sommerfeld's equation, the sum of
Eqs. (9.9) and (9.10) should reduce to Eq. (9.7). Since j takes on values
Z + i or Z — i only, Eq. (9.10) is split into two equations:
Ra2Z4
A Tlt. = for j = Z + £, (9.11)
n32(Z + m + 1)
Ra2Z
ATU = for (9.12)
n32l(l + i)
1 Equation (9.10) was first derived on the quantum mechanics by W. Heisenberg
and P, Jordan, loc. cit.
Sec. 9.3] HYDROGEN FINE STRUCTURE 137
The sign in front of the right-hand side of each of these equations has been
changed to conform with Eq. (9.9), where a positive sign means an
increase in the term value, or a decrease in the energy. Adding Eq.
(9.9) to each of Eqs. (9.11) and (9.12),
Rot'Z*/ 1_j$\
AT ATi,9 + A Tr for 3 -1 + g»
n3 yZ + 1 4nj
(9.13)
Rot2ZA(l _ 3A
AT A Tl,a + A Tr for j
ns yZ 4n)

If k in Sommerfeld's equation is replaced by Z and Z + 1, respectively,

these equations will result. If again k is replaced by j + which is just

A 6563 A4861
Fig. 9.4.—Schematic diagrams of the lines Ha and H/g, in the Balmer series of hydrogen.

the same as replacing Z + 1 and Z of Eq. (9.13) by j + we obtain the

single equation

AT =
Ra2Z V 1_ r\ (9.14)
n3 \j + J 4nj

9.3. Observed Hydrogen Fine Structure.—Schematic diagrams of

the theoretical fine structure of the first two lines of the Balmer series
of hydrogen are shown in Fig. 9.4. Applying selection and intensity
rules, both Ha and H$ should be composed'of two strong components and
three weaker ones. Neither one of these patterns has ever been resolved
into more than two components. The best results to date are those of
Lewis, Spedding, Shane, and Grace,1 obtained from H3, the behavior of
1 Lewis, G. N., and F. H. Spedding, Phys. Rev., 48, 964, 1933; also Spedding,
F. H., C. D. Shane, and N. S. Grace, Phys. Rev., 44, 68, 1933.
138 INTRODUCTION TO A TOMTC SPECTRA [Chap. IX

the two known hydrogen isotopes. Using Fabry and Perot ^talons,
photographs similar to the one shown in the center of Fig. 9.5 have been
obtained. For this photograph the first order of a 30-ft. grating mount¬
ing (of the Littrow type) was used as the auxiliary dispersion instrument.
Microphotometer curves of both H1 and H2 are reproduced above and
below each pattern. It is to be noted that the components of H2 are
considerably sharper than Hi, and that a third component is beginning to
show up. The broadening is due to the Doppler effect and should be
greater for the lighter isotope.

Fio. 9.5.—Fine structure of Hi and Hi from the Balmer series of the two hydrogen
isotopes. Microphotometer curves above and below were made from the interference
patterns in the center. (After Spedding, Shane, and Grace.)

Theoretical intensities for the fine structure of hydrogen were first

calculated by Sommerfeld and Unsold1 in 1926. Experimentally it is
found that the relative intensities of the two main components, the only
ones resolved, depend largely upon the conditions of excitation. In
some instances the supposedly weaker of the two lines will be the stronger,
as it is in Fig. 9.5.
In going to higher members of the Balmer series the separation of the
two strong components of each member approaches theoretically and
experimentally the separation of the common lower state, 0.364 cm-1.
This interval occurs in each doublet between the fainter of the two strong
lines and the next to the weakest satellite.
9.4. Fine Structure of the Ionized Helium Line X4686.—A better
detailed agreement between observation and theory has been found in the
hydrogen-like spectrum of ionized helium. A microphotometer curve
of the line X4686 is given at the bottom of Fig. 9.6. This line corresponds
with the first line of the Paschen series in hydrogen (see Fig. 2.8). With

1 SoMMXSBniLD, A., and A. Unsold, Zeite. f. Phys., 36, 259, 1926; 38, 237, 1926;
a@e also ScrbOdxnoxr, E., Ann. d. Phyt., 80, 437, 1926.
Sec. 9.5] HYDROGEN FINE STRUCTURE 139
Z = 2 the fine-structure separations should bd 16 times as great as in
hydrogen [see Eqs. (9.13) and (9.14)]. The predicted fine structure
shown above in Fig. 9.6 was first given by Sommerfeld and Unsold. At
least four and possibly five of the predicted
components may be said to have been resolved
by Paschen.1 The appearance of certain com¬
ponent lines in this pattern, which are not
allowed on Sommerfeld’s original theory of
hydrogen fine structure, are strong points in
favor of the newer theory of the coincidence
of levels having the same j values.
9.5. The Dirac Electron and the Hydrogen
Atom.—On Diracs2 theory a single electron in
a central force field is specified by a set of
four wave functions ^i, ^3, and ^4, in place
of just one as in the case of the Schrodinger
theory. Each of these functions is a solution
of a wave equation. Although the setting up
of the equations is out of place here we shall
accept the solutions arrived at by Darwin
and Gordon3 and show in what way they 4686 1 0 9 8 7 6 .5 4 3
Fig. 9.6.—Diagram of the
correspond to the earlier theories and, at the
fine structure of the ionized
same time, get some picture of the new atom helium Une X4686. (After Som-
merfeld, Unsold, and Paschen.)

With each wave function \j/i, ^2, ^3, and ^4 properly normalized, the
probability density, just as in the case of Schrodinger's theory (see
Chap. IV), is given by
P = W, (9.15)
where
W* = Mi* + Mi + Mi + Ml (9.16)

For given values of the azimuthal quantum number l and the mag¬
netic quantum number m (m = u + J), there are two sets of solutions
corresponding to j = l + \ and j = l — $, respectively (j equals inner
quantum number).
j = 1+\
\fri = —iMePf+i • MrFi
^2 = —iMeP^ll • MrFi 17x

*3= (l + u + U-MeP'i MSh K J

*4 « -(Z ~ u) • MePp1 ‘ MrGi

1 Paschen, F., Ann. d. Phys., 82, 692, 1926.

* Dirac, P. A. M., Proc. Roy. Soc., A, 117, 610, 1928; A, 118, 351, 1928; see also
Darwin, C. G., Proc. Roy. Soc., A, 118, 654, 1928.
* Gordon, W., Zeits. f. Phys., 48, 11, 1928.
140 INTRODUCTION TO ATOMIC SPECTRA [Chap. IX

J-1-*
\j/i = —i(l + u) * MePi-i' MrF-i-1.
*2 = i(l -u- 1) • MaP?^1 * MJP-u.1.
(9.18)
= ilfflP1/ • MrG-i-1.
*4 = JI^P’/*1 • MrG-l-1.

In this form each wave function in polar coordinates is written as the

product of two functions, one of which gives ^ as a function of the angles
and 0 alone and the other as a function of r alone. Me and Mr are the
angular and radial normalizing factors, respectively. The spherical
harmonics P'iivyO) are defined by

P, - (-!)■<! + .)!

The radial functions F and G are functions of the electron-nuclear

distance alone and are given by

F\ = ~ ~ v ~ ”>i - <ti) and G\ = („, + <r2) (9.20)

where
ai — (N + l + 1) * rv~l • e~kr • iPi(—nr, 2p + 1, 2At), (9.21)
o-2 = —nr • • e~~kr • iPi( — /ir + 1, 2p + 1, 2fcr), (9.22)
and
nr == n — \l — 1| = radial quantum number, (9.23)
p - [(* + l)2 - «2P, (9.24)
N * [n* + (i + l)2 + 2pnr]*, (9.25)
A: = ax = 0.528 Angstroms, a2 = 5.30 X 10“5, (9.26)

where «i is the radius of the Bohr first circular orbit and a is the fine-
structure constant. The functions iPi are of the form of series

\F i(a,6,c) a c i Q(a + *) r2 l °(a + 1)(° + 2) . . .

1 +
6 • ir ^ 5(6 + 1) • 2r ^ 6(6 + 1)(6 + 2) *3! ^
(9.27)
which terminate for negative integer values of a.
9.6. The Angular Distribution of the Probability Density Pe.—With
two sets of four wave functions, the solutions and hence the probability
density must be divided into two parts, one for j — l + \ and one
for j = l — Corresponding to given n, l, and j, there are 2j + 1
magnetic states for which the magnetic quantum number m takes values
from m = +j to ra = —j. Hartree1 has shown that for given n, l} and
j the magnetic states with equal hut opposite m have the same probability-
density distribution. He has also shown that the angular distribution
1 Hartree, D. R., Proc. Camh. Phil. Soc., 25, 225, 1929.
Sbc. 9.6] HYDROGEN FINE STRUCTURE

for two 'electrons with the same n, j, and m, and l * j ± J is the same.
As an example of these two theorems, the angular charge distributions,
as they are called by Hartree, for the magnetic states m =s* ± f of a *Pj
term are not only the same but are also identical with the two magnetic
states m =* ±4 of the 2Dj term. It should be pointed out, however, that
the radial charge distributions of the 2P and 2D terms are different.
With this simplification of the problem the charged distributions need
be determined for j = l + \ and positive m only.

S9

Fio. 9.7.—The angular factor Pe of the probability density \plotted as a function

of the angle 6. The shaded areas indicate spherical symmetry presented by two or four p
electrons, or four or six d electrons.

For j — l + \ the probability density W* on the Dirac theory is

written
« pe. Pr = MJ[Pr+1PrA + P#}P#V] * M2[F\ + Of], (9.28)

The first part of the solution is a function of the angles ip and 0 alone.

Pe = M\[Pt+lFtt 1 + PR}P#n (9-29)

where the angular normalizing factor in the form given by Roess1 is

M\ = [47r(Z + u + l)!(i - u)\]~K (9.30)

From Eqs. (9.19) and (9.29) it may be seen that the angle tp occurs
in each polynomial in the form of an exponential eim* and is always

1 Robbs, L. C., Phys. Rev., 37, 532, 1931.

142 INTRODUCTION TO ATOMIC SPECTRA [Chap. IX

to be multiplied by its complex conjugate This gives a constant

for all m values. Exactly the same result is obtained on the Schrodinger
theory, which indicates that the probability density is symmetrical
about the <p (magnetic) axis for all states. Leaving out the factor
l/4ir in the normalizing factor M% the probability density Pg for a
number of states is given in Table 9.2 and is shown graphically in Fig.
9.7.1
Hartree has shown that for electrons with the same n, l, and j, the
probability density summed over the states m = +£tom = +j presents
spherical symmetry. In this case spherical symmetry is shown by
m — +y
^ Pg = constant, (9.31)
m« +1

given in the last column of Table 9.2 and graphically by the shaded

Table 9.2.—The Probability Density as a Function of the Angle 0

m— -fj

Energy states i m Pg X Pe

*Sh or *P* i \ 1 1

| sin* 0
*P| or 3 J(3 cos* 0 + 1) 2

V1 sin4 0
8
*£>, or V, 3 i | sin* 0(15 cos* 0 + 1) 3
1 h f(5 cos4 0 — 2 cos* 0 + 1)

i sin6 0
1 t5# sin4 0(35 cos* 0 + 1)
*Fj or *Crj I ! [8 sin* 0(21 cos4 0 — 6 cos* 0 + 1) 4
i 15(175 cos6 0 - 165 cos4 0+45
cos* 0 + 9)

areas under the straight lines in Fig. 9.7. Since the probability density
for the negative m states is the same as that for the corresponding
positive m states, the sum of the probability densities for all negative m
states is also a constant. This means that four electrons in 2P* states
or two electrons in 2P\ states will form a spherically symmetrical charge
distribution.2 Another and similar consequence of the Dirac theory
is that a single electron in a 2P\ state is two electrons in a 2Pj state
present spherical symmetry. Not only are all S states, formed from a

1 White, H. E., Phys. Rev., 38, 513, 1931.

* These correspond to the ^/-coupling states (f § § §)* and (} £)0 in Fig. 14.21.
 Fig. 9.8.—The angular factor P$ of the probability density plotted in angular
coordinates. Above and below the quantum-mechanical electron distributions the corre¬
sponding classical electron orbits are shown oriented in each case according to the model
l*, a*, j*, and m.

one-electron systems with but one valence electron and that in a

state. The normal states of B, Al, Ga, In, and T1 are examples of
this.
 144 INTRODUCTION TO ATOMIC SPECTRA [Chap. IX

If angular coordinates are used in plotting Pe a number of interesting

correlations with the classical orbits may be made. Such angular
distributions are shown in Fig. 9.8 for s, p> d, and / electrons. Above
and below each density figure the corresponding classical orbits are shown
oriented in each case according to model a (see Fig. 4.8). On this model
the angular momentum vectors Z* = y/l(l + 1) and s* = Vs(s + 1)
are combined to form the resultant j* = y/j(j + 1), j* in turn being so
oriented that its projection m on the <p axis takes half-integral values
m = ±4, ±f, ±|, • • • ±j. It may be seen from the figures that in
the precession of Z* about j* and the simultaneous precession of j* about
the <p (magnetic) axis the electron orbit fills out a figure in space not
greatly unlike that of the quantum mechanics. The density curves
are symmetrical about <p. The orbit normal Z* is shown for the four
positions it takes when in the same meridian plane (the plane of the
paper). In order to illustrate an orbit rather than its straight-line
projection, the orbit plane is tipped slightly out of the normal to l*.
While the classical models b, c, and d may also be compared with the
probability curves of Fig. 9.8, model a seems to be in the best general
agreement. It should be pointed out that the electron is not confined
to the shaded areas in each probability curve, but that the magnitude
of a line joining the center to any point on the curve is a measure of the
probability of the electron being found in the direction of that line.
From the physical standpoint it is interesting that Pe becomes zero
for 0 = 0 and tt only. The electron can pass therefore from any region
to another without going through a node of zero probability. This is
also true for the radial factor as will be shown.
9.7. The Radial Distribution of the Probability Density Pr—For
j = l + i the probability-density factor Pr is [see Eq. (9.28)],
Pr = M2(F2 + G?). (9.32)

The radial normalizing factors M2r have been given by Roess as

_ (N + p + nrWW+'V&p + nr + 1)_ ,Q ^
r nr\2N[Y(2p + l)]2[(Ar + Z + l)2 + nr(nr + 2p)f

which from tables of the gamma function are readily evaluated. The
radial function Fj, as compared with G], is extremely small and for
hydrogen is of the order of magnitude of the square of the fine-struckire
constant. If a is set equal to zero throughout the radial equations
the gamma functions become simple factorials, F\ vanishes, and Pr
reduces to the radial factor of the Schrodinger theory (Rn,i)2 (see Chap.
IV)
Since the radial densities on the Schrodinger and Dirac theories
are so nearly identical, it is difficult to show their differences graphically
Sac. 9.7] HYDROGEN FINE STRUCTURE 145

However, by splitting up Mf(F\ + Gf) into tfro parts M*Ff and MfGf,
curves may be given for each on different scales. In Fig. 9.9, for example,
curves for the 4p, 2P? state are drawn. The factors MfG\ and M*F\
are shown by the dotted lines in the top and middle figures, respectively.
Multiplying by 4xr2 the density-distribution curves (shaded areas) are
obtained. The main reason for representing the density distribution
in this way is to show that the zero points of each of the two upper

Fiq. 9.9.—The radial factor Pr of the probability density for the state 4p,*P|.
The spin correction (middle figure) added to the Schrodinger distribution (upper figure)
gives the Dirac distribution (lower figure).

curves, other than those points at r = 0 and r = °o, occur at different

values of r. While the resultant curve 4wr2M*(Ff -f- Gf) in the lower
figure is almost identical with the top figure for 4rr2MfG2l9 it does not
come to zero at the two points near r = 6 and r = 14. The difference
at these two points only has been exaggerated in the bottom figure. The
slight shift in the probability density, due chiefly to the addition of the
F function, produces the fine-structure shift of the hydrogen energy
levels. This will be given in Sec. 9.9. The classical 4p orbit shown in
the last figure is drawn according to model a. As in the case of the
Schrodinger theory, the density distribution differs greatly from zero
only within the electron-nuclear distance of the corresponding classical
orbits.
146 INTRODUCTION TO ATOMIC SPECTRA [Chap. IX
9.8. The Probability-density Distribution W*.—Attempts to bring
together the probability factors Pe and Pr into one single picture for
have been somewhat successful. With a spinning mechanical
model,1 photographs have been made which represent the Dirac electron
cloud for a number of the simpler hydrogen states. The photographs
given in Fig. 9.10 were taken from the equatorial plane of the model,

. i.it
m-+ior-£ 22P^ m-+j or-J [32D3/^] m•+ Jorj> 2gP% [3gDfo] m-+Jor-£

32D% [4^F%j m -+ ior 1

I •■•I Wm
32p3^ [4^F%] m-+ Jor-} 3VD^[42F^] m-+£ or f 42F% [52G#] mJ or

fPfjj [5fe%]m*+jor-f 42F%[52G%]m-+f or-j 2 22S^[3aP)jm^tor- £

Fio. 9.10.—Photographs representing the Dirac electron cloud for a number of the
simpler hydrogen-like states made by means of photographing a mechanical model. The <f>
(magnetic) axis is vertical and in the plane of the paper. The scale for each state is given
below each symbol in AngstrSm units.

so that the <p (magnetic) axis is vertical and in the plane of the paper.
While these photographs were made to represent the electron cloud
and not simply cross sections, the latitude lost in photograph copying
and printing makes them appear as cross sections. The states repre¬
sented are given beneath each figure. Since states with negative m
values are identical with the corresponding positive m states, only
one picture is given for both +m and — m. The electron cloud for each
state, given in brackets, is so nearly like the figure given for the unbrack¬
eted state that the difference is not distinguishable in such a photograph.
The scale in Angstrom units for each state is given beneath each symbol.
Graphical comparisons of the classical electron orbits, without
spin, have shown (see Chap. IV) that the electron path closely follows
i White, H. £., Phy*. RmH 88, 518, 1931.
Sec. 9.0] HYDROGEN FINE STRUCTURE 147

the probability density \frp* on Schrddinger’s theory. Introducing the

electron spin into the classical model, the orbit in a weak field no longer
precesses about the magnetic axis but about the spin-orbit resultant j*,
and j* in turn precesses about <p (or H). Corresponding to this change in
space quantization is the change in the 6 factor of the probability density
W* brought about by introducing the spin through Dirac’s theory.
Radially, however, little change has occurred in either the orbital or
the quantum-mechanical model M*. This might have been expected,
since the main part of the energy in hydrogen is given by the radial
factor in ^ and the introduction of the spin has added only the fine-
structure corrections to this energy.
9.9. The Sommerfeld Formula from Dirac’s Theory.—Gordon1 has
shown that Sommerfeld’s fine-structure formula [Eq. (9.4)] may be
derived from Dirac’s theory of the hydrogen atom. Gordon obtains
for the term values of Eq. (9.4),

0» - J
a2Z2
\ + VP a2Z*)
r
where j' is a new quantum number given for the various terips by the
+
nc
(9.34)

values given in Table 9.3.

Table 9.3.—Quantum Numbers for the Hydrogen Atom on Dirac’s Theory

’Pi >D, *Fl ’Pi

1 9 0 1 2 2 3 3
j * h 1 1 \ ! i
r -1 +1 -2 +2 —3 +3 -4

Since f occurs as the square only, the minus sign can be neglected.
Due to this fact f2 can be replaced by (j + $)2, and Eq. (9.34) becomes

T = -^jl +
a2Z2
+ Vj+hy-ct'Z*)2 r+r
(9.35)

Expanding, following Sommerfeld’s treatment,2 and dropping all but

the first two terms,

RZ2 Ra2ZA( 1_3\

(9.36)
n2 nz \j 4* $ 4n)

The first term gives the Balmer terms and the second the corrections
given by Eq. (9.14).

1 Gordon, W., Zeits. /. Phys., 48, 11, 1928.

2 See Sommerfeld, A., “ Atomic Structure and Line Spectra,” p. 477, 1923.
148 INTRODUCTION TO ATOMIC SPECTRA [Chap. IX

Problems
1. Calculate the fine-structure pattern for the electron transition n * 3 to n ■ 2
for ionized helium. Determine the wave-length at which this line is to be found, and
compare the fine-structure intervals in Angstroms with those of Ha in hydrogen.
2. Construct radial-density-distribution curves for a 4d,*D| and states
(see Fig. 9.9).
 CHAPTER X

ZEEMAN EFFECT AND THE PASCHEN-BACK EFFECT

10.1. Early Discoveries and Developments.—In the year 1896

Zeeman1 discovered that when a sodium flame is placed between the
poles of a powerful electromagnet the two lines of the first principal
doublet are considerably broadened. Lorentz pointed out that this
phenomenon is in harmony with the electron theory of matter and
radiation proposed by himself. He predicted from theoretical con¬
siderations that the light from these lines should be polarized by the
magnetic field, circularly polarized if viewed in a direction parallel to
the lines of force, and plane polarized if viewed at right angles to the field.
These predictions were later verified by Zeeman by means of Nicol
prisms as analyzers.
It has been shown from the simple classical theory of Lorentz that,
if a light source be placed in a magnetic field, the motions of the electrons
should be modified in such a way as to change their periods of motion.
In the simple case of an electron moving in a circular orbit, the plane
of which is normal to the field direction Hf the electron will be speeded
up or slowed down by an amount which depends upon the magnetic
field strength H, the mass and charge on the electron, and the velocity
of light. A classical treatment of this problem shows that if vo represents
the orbital frequency of the electron without field, the frequency in
the presence of a field will be given by v0 ± Av, where [see Eqs. (3.50)
and (3.59)]
eH
Av
4irmc
(10.1)

If the field is normal to and up from this page of the book, then
electrons moving in a counterclockwise direction in the plane of the
paper are speeded up by an amount Av and those moving in a clockwise
direction are slowed down by the same amount. It will now be shown
how these modified motions have been employed in giving a classical
explanation of the normal Zeeman effect.
In the following explanation we are concerned with an assembly
of electrons moving in orbits oriented at random in space. We start
by selecting one of these orbits and resolve the motion into three com¬
ponents along three mutually perpendicular axes (see Fig. 10.1a). The

1 Zeeman, P., Phil. Mag.} 5, 43, 226, 1897.

149
150 INTRODUCTION TO ATOMIC SPECTRA [Chap. X

motion of the electron is here pictured as consisting of three simple-

harmonic motions, one along the x axis, one along the y axis, and one
along the z axis. When this resolution is repeated for all of the electrons,
the average amplitude of all the motions along each axis will be the
same. If now, in the absence of a field, the electrons are emitting
light and we observe the radiation in the x direction, only the light
from the y and z motions will be observed. Since these motions are pro¬
jections from all orientations, this light will be unpolarized. Thus, in the
absence of a field, the light observed in any direction is unpolarized.

If now we return to Fig. 10.1a, which represents the three harmonic

components of a single orbital electron, and apply a magnetic field
in the direction of the z axis, the x and y motions will be modified and
the z motion will remain unchanged. Moving transversely to the field,
the x and y motions will take the form of rosettes as shown in Fig. 10.16.1
This is the same type of motion as one encounters in the Faraday effect.
These apparently complex motions can be described to better advan¬
tage in terms of circular motions somewhat as follows. The simple-
harmonic motion given by the y component of Fig. 10.1a, for example, is
equivalent to the resultant of two equal but opposite circular motions
y+ and y~ as shown in Fig. 10.1c. Similarly the x component can be
represented by two opposite circular motions x+ and x~. When the field
is applied, the x+ and y+ rotations will be speeded up by an amount
Av [see Eq. (10.1)] and the x~ and yr rotations will be slowed down by
the same amount. (The faster x+ motion combined with the slower x~
motion results in Fig. 10.16.) The x+ and y+ motions and the x~ and y~

1 The electrons are here moving at right angles to the field, and they therefore
experience a force at right angles to their motion the direction of which depends upon
the direction of the field and the direction of the motion.
Sec. 10.1] ZEEMAN EFFECT AND THE PASCHEN-BACK EFFECT 151

motions are now combined, as shown in Fig. 10. Id, to form plus an<J
minus resultants. Thus the motion of a single electron in a magnetic
field is represented by a linear motion along the field direction with
unchanged frequency vo and two circular motions at right angles to
this, one with the frequency v0 + Av and the other with frequency
vo — Av. On summing up such motions for all of the electrons, the
result will be the same as if one-third of the electrons are moving with
unchanged frequency along the z axis, one-third moving with a counter¬
clockwise circular motion normal to z of frequency vo + Av, and the other
third moving with clockwise circular motion normal to z of frequency
vo — Av (see Fig. 10.le).
We are now interested in the nature of the light that should classically
be radiated from these motions. When viewed in the direction of the
field, only the circular motions are observed and these as right- and
left-handed circularly polarized light (Fig. 10.IB). Since light is a
transverse wave motion, the z motions will not emit light in the field
direction. When viewed perpendicular to the field, the z motions are
observed as plane-polarized light with the electric vector vibrating
parallel to the field, and the circular motions, seen edge on, are observed
as plane-polarized light with the electric vector at right angles to the
field. A spectrum line viewed normal to H should therefore reveal
three plane-polarized components (see Fig. 10.1A), a center unshifted
line and two other lines equally displaced one on either side. This is
called a normal triplet. The abbreviation p stands for a vibration
parallel to the field and s (senkrecht) stands for a vibration normal to
the field. The experimental agreement with the direction of rotation
of the circularly polarized components is proof that the radiation is due
to moving negative electric charges.
In Zeeman’s early investigations he was not able to split any lines into
doublets or triplets, but he did find that they were widened and that
their outside edges were polarized as predicted. He was later able to
photograph the two outer components of lines in a number of the ele¬
ments, Zn, Cu, Cd, and Sn, by cutting out the p components with a
Nicol prism. Preston1 using greater dispersion and resolving power
was able to show not only that certain lines were split up into triplets
when viewed perpendicular to the field, but that others were split into
as many as four and even six components (see Fig. 10.2a). He also
pointed out that the pattern of all lines (usually called Zeeman patterns)
belonging to the same series of spectrum lines was the same and was
characteristic of that series. This is now known as Preston9s law. With
Preston’s law firmly established, the Zeeman effect has been, and still
is, a powerful tool in spectrum analysis.

,
1 Preston, T., Phil. Maa.< 46 326, 1898.
152 INTRODUCTION TO ATOMIC SPECTRA [Chap. X

From Lorentz’s classical treatment of the Zeeman effect (Fig. 10.1)

the shift Av of the s components from the unshifted p component (see
Eq. 10.1) is given by

Av = SwwT* = 4-67 X 10-6' H cm_1 = L cm_1' (101a)

where Av is in wave numbers, H is the field in gauss, c is the velocity

of light, and e is the charge on the electron in electrostatic units. Zee-

Zeeman Effect in the Chromium Spectrum

Fig. 10.2a.—Anomalous Zeeman effect as observed in the neutral spectrum of chromium.
{After Babcock.)

Zinc Single! Sodium Principal Doublet *

m mm
mi ini wig ~~
1-1 J

Normal Triplet
I 1

Anomalous
1 I—L-J

Patterns

Zinc Sharp Triplet

No field

Weak field

I—l—I » ■ ■ U-L-J

Anomalous Patterns
Fig. 10.26.—Normal and anomalous Zeeman effect. Viewed perpendicular to the magnetic
field.

man patterns showing just three lines with exactly these separations
are called normal triplets. All other line groups, as, for example, the
complex patterns observed in the chromium spectrum (see Fig. 10.2a),
Sac. 10.2] ZEEMAN EFFECT AND THE PASCHEN-BACK EFFECT 153

are said to exhibit the anomalous Zeeman effect. One of the niost impor¬
tant of the early investigations of the anomalous Zeeman effect was
carried out by Paschen and Runge.1 Each member of the principal
series of sodium, copper, and silver was observed to have 10 components
as shown in Fig. 10.26. The sharp-series triplets in mercury are still
more complicated, the strongest line in each triplet having nine com¬
ponents, the middle line six, and the weakest line only three. This
last line does not form a normal triplet since it has twice the norma)
separation.

Normal Triplet Normal Triplet

Fig. 10.3.—Schematic representation of anomalous Zeeman patterns viewed per¬
pendicular to the field and showing polarization (25-components above and 0-component?
below), relative intensities (heights of lines), and intervals (dotted lines).

In 1907 Runge made an important contribution to a theoretical

explanation of the anomalous Zeeman effect by announcing that all
known patterns could be expressed as rational multiples of the normal-
triplet separations. If, for example, L represents the shift of the 8
components from the unshifted line, as given by Lorentz’s formula,
then the principal- or sharp-series doublets (see Fig. 10.3) may be
expressed as
2S* - 2Pj, ±|Lf ±*L
2£* - 2Pj, ±JL, ±|L, ±*L
and the principal- or sharp-series triplets are expressed as

3Si ~ 3P2, 0, ±JL ± *L ± fL ± fL

*Si - 3Pi, ±*L ± fL ± fL
3£i - *p0, 0, ±fL.
this is now known as Runge1 s law.
10.2. The Vector Model of a One-electron System in a Weak Mag¬
netic Field.—Soon after the discovery of the anomalous Zeeman effect
1 Paschen, F., and C. Runge, Astrophys. Jour., 15, 235, 333, 1902; Phys. Zeils.,
3.441.1902.
154 INTRODUCTION TO ATOMIC SPECTRA [Chap. X

came the development of the LaruU vector model of the atom and the
calculation of the famous Landi g factor. The accuracy with which
this model, with its empirical rules, accounted for all observed Zeeman
patterns, and predicted others which were later verified, is one of the
No Field Weak Field .

-i

-f
Fia. 10.4.—Splitting up of an energy level in a weak magnetic field. This figure is
drawn for the case where j — f.

marvels of scientific history. With the advent of the spinning electron,

and later the quantum mechanics, Landes vector model gave way to a
more satisfactory and, at the same time, a simpler semiclassical model.
It is with this simplified model and its adequate account of the Zeeman
effect that we are concerned in this
H
chapter.
Experimentally it is observed that
in a weak field each spectrum line
is split up into a number of compon¬
ents forming a symmetrical Zeeman
pattern, and that, in general, the width
of any given pattern is not greatly
different from that of a normal triplet.
Theoretically this effect is attributed
to a splitting up of the energy levels
Fig. 10.5.—Classical precession of a into a number of predetermined
single valence electron around the field
direction //.
equally spaced levels (see Fig. 10.4).
Transitions between two sets of thej-e
levels, subject to certain selection and intensity rules, give rise to the
observed spectral frequencies.
Before attempting a calculation of these Zeeman levels it is well
that we formulate some picture of the atom in terms of the semiclassical
model. In Fig. 8.8, we have seen that the orbital angular-momentum
vector J* and the spin angular-momentum vector s* precess with uniform
speed around their resultant j*. When the atom is placed in a weak
magnetic field, the magnetic moment associated with the total mechan¬
ical moment p, = j*h/2ir causes the atom to precess like a top around
Sec. 10.3] ZEEMAN EFFECT AND THE PASCHEN-BACK EFFECT 155

the field direction H (see Fig. 10.5). The quantum conditions impose,d
upon this motion (see Sec. 9.6) are that the projection of the angular
momentum j*h/2r on the field direction H will take only those values
given by mh/2t, where m ±J, ±f, ±f • • • , ±j. In other words the
projection of j* on H takes half-integral values from +j to —j only.
The discrete orientations of the atom in space, and the small change
in energy due to the precession, give rise to the various discrete Zeeman
levels. While the number of these levels is determined by the mechanical
moment j*h/2iry the magnitude of j*
the separations is determined by the
field strength H and the magnetic // N\
moment /*. In field-free space an energy / N\
level is defined by the three quantum CTJ' *
numbers n, l, and j. In a weak magne- y7 ^ /
tic field an additional or fourth quantum * — -~jl^
number m is necessary to define the s -
state. \\ //'
10.3. The Magnetic Moment of a \\ //
Bound Electron.—To determine the ^ ^
magnitude of the separations between ///i\w
///t\\\^
Zeeman levels, it is essential that we //f✓V'l'W
/;jiv\ j^\\\
first determine the total magnetic / / / 1 \ \ \
moment of the atom. In the simplest A/L j \\\ _J\ \ M
case to be considered here the atom
core and nucleus will be assumed to -^_I___
have zero magnetic and mechanical 7 \]
' s
/ \i
I

\
I
\ / /
/

moments so that any moments attri-

buted to the atom must be assigned j
j
/ \ \ /
j\
\
^
\ \ /
/

to a single valence electron. Later it

will be shown where these assumptions " -jjy*—" ^Is
are justified. Fig. jog—Vector model
Fig. 10.6.—Vector model showing
showing
According to the classical theory, the magnetic and mechanical moments
the ratio between the magnetic and °f» valence electron,
mechanical moments of an electron in an orbit [see Eq. (3.52)] is given by

Just as observations of the fine structure of spectrum lines show

that the mechanical moment of the spinning electron is given by 8*h/2r,
where s* = \Zs(s -f 1) and s = £, so the anomalous Zeeman effect
shows that the ratio between the magnetic and mechanical moments
for the spinning electron is just twice that for the orbital motion, i.e.,

(10.3)
156 INTRODUCTION TO ATOMIC SPECTRA [Chap. X

This result has also been derived theoretically on the quantum mechanics
(see Chap. IX).
A schematic vector diagram of the magnetic and mechanical moments
is shown in Fig. 10.6. Here it is seen that the resultant magnetic
moment is not in line with the resultant mechanical moment j*h/2t.
Since the resultant mechanical moment is invariant, Z*, $*, m, and
M,g precess around j*. As a result of this precession, only the component
of parallel to j* contributes to the magnetic moment of the atom.
This may be seen by resolving mt9 into two components, one parallel to
j* and the other perpendicular. The perpendicular component, owing
to the continual change in direction, will average out to zero. The
parallel component may be evaluated as follows:
By Eqs. (10.2) and (10.3) m and are given as

i*h
e ergs , 0 * h e ergs
Ml = -and m* = 2 • «V • n— (10.4)
27r 2me gauss ( 27r 2mc gauss'

and their components along j* are given as

Component in = lcos (l*j*)

(10.5)
Component jus = 2 • scos (s*j*).

Adding these, we obtain

Mi = [l* cos (l*j*) + 2s* cos (s*j*)}^ ■ 2^- (10.6)

Since the last two factors in this equation are equivalent to one Bohr
magneton [see Eq. (3.57)], the quantity determined by the bracket
gives the total magnetic moment of the atom in Bohr magnetons. This
bracket term is readily evaluated by setting it equal to j* times a constant

j* • g — l* cos (l*j*) + 2s* cos (s*j*). (10.7)

Making use of the vector model and the cosine law that
S*2 = 1*2 + j*2 _ 21*j* cos (Z*j*), (10.8)
we obtain
<1*2 4_ 7*2 _ 0*2
l* COS (IT) = 1 2j* (10.9)

Similarly,
n*2 7*2 _L o*2
8* COS (s*j*) = 3 2j* (10.10)

Substituting these two cosines in Eq. (10.7), we get

- l*\
(10.11)
9 ~ 1 + 2j*2
Sec. 10.4] ZEEMAN EFFECT AND THE PASCHEN-BACK EFFECT 157

in terms of the quantum numbers l and j and the spin a,

_ , , id + i) + «(• + i) - id + l)
(10.12)
* = 1 +-MTT)-
The importance of this g factor cannot be overemphasized, for it
gives directly the relative separations of the Zeeman levels for the
different terms.1 We shall now see how this comes about.
10.4. Magnetic Interaction Energy.—By Eqs. (10.6) and (10.7) the
ratio between the total magnetic and mechanical moments of the atom,
y, and p„ is just
M/ __
(10.13)
p, 9 2me1
where p, = j*h/2ir.2 The precession of j* around H is the result of a
torque acting on both l* and s*. Due to the electron's anomalous spin
magnetic moment, s* tends to precess twice as fast around H as does Z*.
If the field is not too strong, the coupling between Z* and s* is sufficiently
strong to maintain a constant j*, so that this resultant precesses with a
compromise angular velocity, by Larmor's theorem [Eq. (3.58)], given
by g times the orbital precession angular velocity

■- - <ioi4>
The total energy of the precession is given by the precessional angular
velocity wL times the component of the resultant mechanical moment
j*h/2ir on the axis of rotation H :3 * * * * *

AW = cos (j*H) = H ■ COS (j*H). (10.15)

1 The values of g given by Eq. (10.12) are exactly the same as those given by
Land6’s model.
* In any experiment like the Stern-Gerlach experiment (Zeits. /. Phys8, 110,
1922), performed for the purpose of determining the magnetic and mechanical moment
of the atom, the moments and pj are oriented at some angle with the field just as
in the Zeeman effect (see Fig. 10.7). What one measures in this experiment is the
component y of the resultant magnetic moment along H. By theory we say the
component of /xy will be y *= m, cos (,j*H), and the component of j*h/2x along H will
be mJt/2v, where m takes values differing from each other by unity from m = +j
tom = — j.
9 The magnetic energy can be considered as the energy of a permanent magnet of
moment y, at an angle 9, in the field H, or as the added kinetic energy of the electron's
orbital motion. If, in the case of a circular orbit normal to the field, E represents
the kinetic energy before the field is applied, and E' = £/(<■> + «l)* the kinetic energy
after, then the change in energy is just AE ** E' — E = J/wJ, 4- /««£. Since the
added field does not change the size of the orbit, I remains constant. With <•> > >
the first term iB negligibly small and the energy change is given by the product of the
mechanical moment, /« « j*h/2*} and «l.
158 INTRODUCTION TO ATOMIC SPECTRA [Chap. X

In terms of the magnetic quantum number m, j*h/2r • cos (j*H) is just

equal to mh/2ir, so that1
TT e h rr eh
ATT * H • (fa— • m • jr- *
= m -• g -• H^l-.
H-.- (10.16)
v2 me 2ir v 4Tine
Dividing by he, the interaction energy in wave numbers becomes

AW Krp He
(10.17)

With 0 = 1 this equation reduces to Lorentz’s classical formula [Eq.

(10.1)].
Since the field H is the same for all levels of a given atom, it is con¬
venient to express the Zeeman splitting in terms of what may be called
the Lorentz unit, L = He/imnc2, and write simply

—AT7 =, m • 0 • L cm-1. (10.18)

It should be emphasized that AT is the change in energy for each m

level from the original level, and that the shift is proportional to the field
strength H. With m * ±J, ±f, ±J, • • • , ±j, the g factor is seen
to be of primary importance in the splitting of each level. Values of
the 0 factor for doublets are given in Table 10.1 and, along with them,
values of the corresponding splitting factors mg.

Table 10.1.—The Land£ g Factors and the Splitting Factors mg for Doublet
Terms

1 Term 9 mg

0 ! ±1

'Pi I ±i
1 'Pi I ±i, ±*

1 ±i, ±i
2 t ±1, ±1, ±¥
V, f ±4, ±1, ±¥
3 »F, 4 ±4, ±¥, ±¥, ±V
1 ±4, ±¥, ±¥. ±V
4 •0| ¥ ±t, ±¥. ±¥> ±¥. ±¥

Consider, for example, the splitting of a *Pj level in a weak magnetic

field. Withy = |, there are four magnetic levels m = |, i, —and f,
shifted from the field-free level by mg = $, f, —f, and — J. A graphical

1 Here m, the magnetic quantum number in the numerator, must not be confused
with m, the mass of the electron in the denominator.
Sec. 10.6] ZEEMAN EFFECT AND THE PASCHEN-BACK EFFECT 159

representation of this splitting is shown in Fig! 10.7, where the vector

j* is shown in the four allowed positions. Multiplying j* by g and pio-
jecting the product on the H axis, the displacements mg shown at the
left are obtained.
Referring now to the spin-orbit interaction energy [Eqs. (8.17) and
(8.20)], which gives rise to the doublet fine structure, and to the equation
H

Fig. 10.7.—Schematic orientation diagram of an atom in a state showing the resultant

Zeeman levels in a weak magnetic field.

just derived from the Zeeman splitting, the complete term value T of
any magnetic level may be written

T = To - T - mg • L. (10.19)

To is the term value of the hypothetical center of gravity of the field-free

doublet, T is the fine-structure shift, and mg • L is the magnetic shift.
10.5. Selection Rules.—As an example of the calculation of Zeeman
patterns, consider the simple case of a principal-series doublet like the
sodium yellow D lines XX5890 and 5896. The g factors for the initial
2P\ and 2Pj states and for the final 2S\ state (see Table 10.1) are §, $, and
2, respectively. The splitting of each of these levels is shown schemati¬
cally in Fig. 10.8, starting with the field-free levels at the left. The
dotted lines in each case represent the centers of gravity of the associated
levels.
The theoretical selection rules for transitions between levels, in
agreement with observations, may be stated as follows: In any transition
the magnetic quantum number m changes by +1, 0, or —1, i.e.,

Am = 0, ±1. (10.20)

For the stronger of the two field-free lines there are six allowed transi¬
tions and two forbidden transitions. For the other line there are four
160 INTRODUCTION TO ATOMIC SPECTRA [Chap. X

allowed transitions. The observed and calculated patterns are shown

at the bottom of the figure.
Polarization rules derived from the classical theory as well as the
quantum mechanics are found to hold experimentally. These rules
may be stated in the following way:
(Am = ±1; plane polarized J_ to H; s components
Viewed JL to field
)Am = 0; plane polarized || to H; p components
(Am —= ±1; circularly polarized; s components
Viewed II to field
1 Am — 0; forbidden; p components
10.6. Intensity Rules.—The intensity rules for field-free energy
levels, first derived from experimental observations by Burger, Dorgele,
mg
6

8 2
%
r~ -
V r- % —2—
w ■%
a % -%
V - - - ~ - - -
4
-4—1 -1
4
-
-'/j - ^

■a
r- o
- - - - ~

Fia. 10.8.—Zeeman effect of a principal-series doublet.

and Omstein (see Sec. 8.3) are readily shown to follow directly from
the intensity rules for the same levels in a weak magnetic field. In
short these rules may be stated as follows:
The sum of all the transitions starting from any initial Zeeman level
is equal to the sum of all transitions leaving any other level having the same
n and l values. The sum of all transitions arriving at any Zeeman level
is equal to the sum of all transitions arriving at any other level having the
same n and l values.
For any given field-free spectrum line these rules are better expressed
in terms of formulas which have been derived from the classical1 as well
1 For a classical derivation of the intensity rules, based on Bohr’s correspondence
principle, see J. H. Van Vleck, “Quantum Principles and Line Spectra,” Nat. Research
Council, Bull., Vol. 10, 1926.
Sec. 10.6] ZEEMAN EFFECT AND THE PASCHEN-BACK EFFECT 161

as the quantum-mechanical theories. These formulas may be abbrevi-'

ated as follows:

m —> m ± 1, I = A(j ± m + 1 )(j + m). gi)

Transition j —> j
m —> m , I = 4Am2. *
'm m ± 1,1 = B(j ± m + 1)0 ± m + 2).
Transition j -+j + m m , I = 4:B(J A) (j —
(10.22)

A and B are constants that need not be determined for relative intensities
within each Zeeman pattern. These formulas take into account the

. i
1 1
_ Zpeman Patterns _
For Doublets

\ Vi '1

■'l iDi , jT~ ** \\i „i

[ Vi
- i J «

Yi
•
1 «
1, ,

l1M
2c 2j\

fi ri r Vi 11 r,ir i •

Vi
i ii 1

i r * ,
[^
ft
V f | Vi
n 11

Fia. 10.9.— Zeeman patterns for all of the commonly observed doublet transitions. The
dots represent normal triplet separations.

fact that, when the radiation is observed perpendicular to the field

direction, only half of the light making up the s components of a pattern
is observed. Observations parallel to the field give the other half of the
8 components' intensity. The above given rules and formulas applied
to the principal doublet in Fig. 10.8 give the relative intensities shown
at the top of the arrows. The same values are indicated by the heights
of the lines shown at the bottom of the figure. The sum of all transitions
starting from any level in the figure is 12, and the sum arriving at either
of the lower levels is 36. (In obtaining these values the s components
must be multiplied by 2.) Experimentally, the p or s components may
be photographed separately by inserting a Nicol prism in the path of
the light coming from the source between the poles of an electromagnet.
162 INTRODUCTION TO ATOMIC SPECTRA [Chap. X

Zeeman patterns for a number of doublet transitions are given in

Fig. 10.9. These calculated patterns are in excellent agreement with
all experimental observations. It is particularly interesting to see how
closely the different lines group themselves around the normal-triplet
separations, which are indicated in the figure by dots. Normal Zeeman
triplets, as will be seen later, arise in special cases and in particular for all
lines belonging to a singlet series, where the g factor for both the initial
and final states is unity.
A method frequently employed for a rapid calculation of Zeeman
patterns will be given briefly as follows. The separation factors mg,
for both the initial and final states, are first written down in two rows
with equal values of m directly below or above each other. For a
2Z)|-2P| transition they are:

m “ \ i i ~~2 ~ £
mg initial state ^ 1 $ $ —| —$ —

mg final state
\JX1XIX!^
£ f —f —£

In this array, the vertical arrows indicate the p components, Am = 0,

and the diagonal arrows the s components, Am = ±1. The differences
expressed with a least common denominator are as follows:

Vertical Differences Diagonal Differences

p Components s Components
+-A, +*, -A, -A ±ih ±U, ±\b ±H
In short these may be abbreviated,

A„ = (±D, (±3), ±15, ±17, ±19, ±21l cm_1

15 ?

the four p components being set in parentheses, followed by the eight

s components (see Fig. 10.9).
A simple qualitative rule for the intensities has been given by Kiess
and Meggers as follows: If the j values of the two combining terms
are equal, the vertical differences at the end of the scheme, and the
diagonal differences at the center, give the strongest p and s components,
respectively. If the j values are not equal, as in the case shown above,
the vertical differences in the middle of the scheme and the diagonal
differences at the ends give the strongest p and 8 components, respectively.
10.7. The Paschen-Back Effect.—In deriving the interaction energy
between an atom containing one single valence electron and an external
magnetic field, it was assumed that the field was weak as compared
with the internal fields due to the spin and orbital motion of the elec¬
tron. When the external field becomes greater than these internal
Sue.-10.7J ZEEMAN EFFECT AND THE PASCHEN-BACK EFFECT 163

fields the internal motions are greatly perturbed and the atom gives
rise to the so-called Paschen-Back effect.
Just as the doublet fine-structure separations are a measure of the
classical frequency with which l* and 8* precess around their resultant
j* (see Sec. 10.4), so the Zeeman separations of the same energy states
in a weak magnetic field are a measure of the frequency with which j*
precesses around H. In calculating
the Zeeman separations in Sec. 10.4, it
was tacitly assumed that the precession _is*
of l* and s* around j* was much faster
than that of j* around H. This was
necessary in order that the components
of l* and s* normal to j* average to
zero and do not appreciably perturb
the other precession. If now the field
H is increased until the two precessions
are of the same order of magnitude,
then the Zeeman levels of the doublet P fiddu
will begin to overlap, there will be no so strong that 1+ and**precessindepend-
averaging to zero, and Eqs. (10.17) ently around the field direction H.
and (10.19) will not hold. Under these conditions the coupling between
l* and s* will be partially broken down, the classical motions of l* and
8* will become complicated, and j* will no longer be fixed in magnitude.
As the field H is still further increased, l* and s* will soon become

Fio. 10.11.—Space-quantisation diagrams for p and d electrons in a strong magnetic field,

Paschen-Back effect.

quantized separately and precess more or less independently around H

(see Fig. 10.10). This is the Paschen-Back effect.1
The quantum conditions in a strong Paschen-Back field are: (1)
The projection of l* on H takes integral values from mi = +1 to mi = —J.
(2) The projection of 8* on H takes one of the two values m, = +£, or
1 Paschbn, , ,
F., and E. Back, Ann, d. Phys., 80 897, 1912; 40 960,1913.
 164 INTRODUCTION TO ATOMIC SPECTRA [Chap. X

—£. For a p electron with l «■ 1, s = £, there are six possible states

mi = 1, 0, and —1, when m, = £ or —Space-quantization diagrams
of these cases are given in Fig. 10.11. Since for every electron there
are two values of m, and 21 + 1 values of mh there are 2(21 + 1) com¬
binations of the quantum numbers corresponding to 2(2Z + 1) different
states of the atom. As might be expected, this is exactly the number of
weak-field levels.
The total energy of the atom in a field strong enough to give the
Paschen-Back effect is made up of the three parts: (1) the energy due
to the precession of l* around H; (2) the energy due to the precession of s*
around H; (3) the interaction energy between l* and s*. By Larmor’s
theorem [Eq. (3.58)], the precessional angular velocities are given by H
times the ratio between the magnetic and mechanical moments:

Uf = H^- and w.* = 2H^~- (10.23)

Zmc i Jmc

Since the ratio between the magnetic and mechanical moment for the
spin of the electron is twice the orbital ratio, s* should, on the classical
picture, precess twice as fast as l*. Multiplying each of these angular
velocities by the projection of the angular momentum on H [see Eq.
(10.15)], one gets the first two terms of the energy:

“• <ra> - "sb-s1 <‘°-24>

* 2Hsr/‘s ”• <10-25>
The sum of these two energies accounts for the main energy shift
from the unperturbed energy level and is
pji
A WH = (mi + 2m.)H^~- (10.26)

Dividing by he, the term shift in wave numbers becomes

He
-ATh = (mt + 2m,)4-^ cm"1, (10.27)

or in Lorentz units of He/Airme2,

—ATh = (mt + 2mt)L cm-1. (10.28)

To this magnetic energy the small correction term due to the inter¬
action between l* and s* must be added. Although these two vectors
precess independently around H, each motion still produces a magnetic
field at the electron which perturbs the motion of the other. This
interaction energy, though small as compared with that due to the
external field, is of the same order of magnitude as the fine-structure
Sac. 10.8] ZEEMAN EFFECT AND THE PASCHEN-BACK EFFECT . 105

doublet separations in field-free space, which by Eqs. (8.14), (8.18), and

(8.19) are given by the r factor,
r = —ATi,9 = al*8* cos (l*s*)t (10.29)
where
Ra2Z* -
(10.30)
fl nH(l + m+ l)Cm •
In field-free space, the angle between l* and 8* is constant and the
cosine term cos (il*s*) is easily evaluated. In the present case, however,
the angle is continually changing, so that an average value of the cosine
must be calculated. From a well-known theorem in trigonometry it
may be shown that with s* and l* precessing independently with fixed
angles around a third direction H,

cos (l*s*) = cos (l*H) • cos (s*H). (10.31)

Making this substitution in (10.29),
r = —ATi,t = al* cos (l*H)s* cos (s*H). (10.32)
These are just the projections of l* and s* on H, so that
—ATi,t = amim, = r. (10.33)
Adding this term to Eq. (10.28), the total energy shift becomes
—A Tem-x = {mi + 2t?i#)L + amim8. (10.34)

We may now write down a general relation for the term value of any
strong field level,
Tem-1 = To - {mt + 2m.)L — amtm9i (10.35)

where To is the term value of the hypothetical center of gravity of the

fine-structure doublet.
10.8. Paschen-Back Effect of a Principal-series Doublet.—As an
example of the Paschen-Back effect, consider first the calculation of terms
and term separations involved in a principal-series doublet
The fine-structure separations due to the interaction of l* and 8* in
field-free space are given in Col. 2, Table 10.2 (see Fig. 8.9). In the
next three columns the weak-field energies are calculated (see Fig. 10.8).
In the last five columns the strong-field energies are calculated, using
Eq. (10.34).
The values tabulated are shown schematically in Fig. 10.12. At the
left the undisturbed fine-structure levels and the observed transitions
are shown. The weak-field Zeeman levels are next shown with the
observed Zeeman patterns below. In the strong field the Paschen-Back
levels are shown with, and without, the small 1*8* coupling correction
166 INTRODUCTION TO ATOMIC SPECTRA (Chap. X
amtma. The allowed transitions and the calculated pattern are shown
below.
In deriving the above equations for the Zeeman and Paschen-Back
effects, the atomic system was assumed to be in one of two ideal situations.
In the first case the field was assumed so weak that j*9 the resultant
of l* and $*, was invariant as regards magnitude and inclination to the
field axis. In the second case the field was assumed so strong that
l* and s* precess independently around H. The question of intermediate

Table 10.2.—Weak- and Strong-field Energies for a Principal-series Doublet

Weak field Strong field

No field
(Zeeman effect) (Paschen-Back effect)

m =
Term m 0 mg mi m. ' mi -f- 2m, am\m$
mi 4- rn.

i
+! +» +1 +* +» +2 -ha/2
-hi 0 +1 +* +1 0
Vi t
-i -i -1 +i -i 0 -a/2
-i -i +1 -i 0 -a/2

+1 +* 0 -i -i -1 0
—a i
-i -i -1 -i -! -2 -ha/2

m +i
2
+1 0 +i -hi +1 0

M -i -1 0 -i -i -1 0

fields, therefore, arises, and one asks, how does each weak-field level
go over to a corresponding strong-field level? Darwin’s treatment
of this problem, which will not be given here, answers this question in a
very simple manner.1 According to the classical law of the conservation
of angular momentum, the sum of the projections on H of the various
angular-momentum vectors must remain the same for all field strengths.
Since in weak field this sum is given by m and in strong field by mi + ra„
we may write, as part of the correlation rule, m = mi + m,. This
alone is not sufficient to correlate all weak- and strong-field levels, since
in most instances there will be more than one level with the same m
value. The more specific rule, in keeping with the quantum mechanics,
may be stated as follows: Levels with the same m never cross.
An ingenious method for obtaining the same correlation has been
given by Breit.2 An array of weak- and strong-field quantum numbers

1 Darwin, C. G., Proc. Roy. Soc., A, 115, 1 1927. ,

* Breit, G., Phys. Rev., 28, 334, 1926.
Sec. 10.8] ZEEMAN EFFECT AND THE PASCHBN-BACK EFFECT 167

is written down as follows (see Fig. 10.13). Values of wj are written

down in their regular order in a horizontal row and values of ma in a
vertical column. The array is next filled in with all possible sums of
Strong Field

Fiq. 10.12.—Energy levels for a principal-series doublet starting with no field at the
left and ending with a strong field (Paschen-Back effect) at the right. Allowed transitions
are shown below.

mi and m,. These sums are the weak-field quantum numbers, divided
into two parts by the dotted lines. Each weak-field level m is to be
correlated with the strong-field level given by the value of mi directly
above, and the value of mt directly to the right of the m value. The
2Pj, m = $ state, for example, goes to the state mi = 1, and m, = $.

P d
nr>| » 1 0 -1 nrije 2 1 0-1-2
m» 1 ± -± * m_-
l
m» m- f i-i -!]-!
* -j: -i
ms ms
*•» i%
Fia. 10.13.—Correlation of weak- and ig-field quantum numbers and energy levels.
Brett.)

It is obvious that there are two ways of drawing the L-shaped dotted
line. Of the two ways only the one shown will give the correct correla¬
tion for doublets from a single electron.
168 INTRODUCTION TO ATOMIC SPECTRA [Chap. X

10.9. Selection Rules for the Paschen-Back Effect.—Selection rules

for the Paschen-Back effect, derived from the classical theory may be
stated as follows: In any transition
A mi = 0, ± 1 and Am, = 0. (10.36)
When these rules are applied to a given doublet they are found to lead,
in a very strong field, to a pattern closely resembling a normal Zeeman
triplet. The radiated frequencies of the principal-series doublet con
sidered above are assembled again in Fig. 10.14. The restriction Am, = 0
implies the necessary condition that the polarization of any given
line be retained throughout all field strengths. The p and s components

Fig. 10.14.—Principal-series doublet in various magnetic field strengths. Paschen-Back

effect.

in weak field become p and s components, respectively, in the strong

field or else disappear as forbidden lines. For the s components, the
fine-structure separation is just two-thirds that of the field-free doublet.
It should be made clear that as the field strength increases and the
quantum number,/ disappears, and, asj and its projection m are replaced
by mi and m„ the selection rules for j can no longer be expected to
hold. This is in agreement with observations made by Paschen and
Back1 who were the first to observe experimentally the so-called normal
triplet of a principal-series doublet in a very strong field. Since the
strong-field levels must be widely separated as compared with the fine
structure, only the very narrow fine-structure doublets may be expected
to be carried over to the Paschen-Back normal triplet with the ordinarily
obtained field strengths. Using the principal-series doublet of lithium,
with only a separation of 0.34 wave numbers, Paschen and Back required
the very strong field of 43000 gauss to observe it as a magnetic triplet.
10.10. The Zeeman Effect, and Paschen-Back Effect, in Hydrogen.—
Although the hydrogen atom is the simplest of all atomic systems, the
1 Paschen, F., and E. Back, Physica, 1, 261, 1921.
Sec. 10.11) ZEEMAN EFFECT AND THE PASCHEN-BACK EFFECT 169

Zeeman effect in hydrogen is not very simple. In Sec. 9.3 we have

seen how each of the Balmer terms contains a fine structure which
i3 made up of doublets 2Sf 2Pf 2D, etc. When placed in a weak magnetic
field each of these doublet levels, as in sodium, should undergo an anoma¬
lous splitting (see Fig. 10.8). As a result of this splitting each fine-
structure component of a line like Ha, X6563, should reveal an anomalous
but symmetrical Zeeman pattern (see Fig. 10.9). With H« made up
of seven different transitions, there would be, in this
cace, seven patterns of the type shown in Fig. 10.9,
all lying within an interval of about half a wave
number (see Fig. 9.4). It is now easy to see why the
Zeeman effect of hydrogen has not been observed.
In a strong magnetic field, the magnetic levels
of each doublet begin to overlap each other, until
in a field strength of several thousand gauss the
Paschen-Back effect sets in. Under these condi¬
tions each doublet level, combining with another
doublet level, gives rise to approximately a normal
triplet like the one shown in Fig. 10.12. The H«
line, for example, will be made up of three normal
triplets 2S-2Py 2P-2Sf and 2P-2Dy superimposed
almost on top of each other. In a field of 32000
gauss, Paschen and Back1 observed just that; a Fjq 1015__Pagch
well-resolved triplet with practically the classical Back effect of hydrogen
separation. In such a field the pattern is some X6563. (a) Enlarged

six times as wide as the field-free line. A photo- published paper of Pas-
graphic reproduction of this triplet is given in Back. ()>) Same
Fig. 10.15. Normal triplets have also been observed paper moved parSiei to
by Paschen and Back for Hp and H7. lines during enlargement.
10.11. A Quantum-mechanical Model of the Atom in a Strong Mag¬
netic Field.—In this chapter the Zeeman effect, as well as the Paschen-
Back effect, has been treated chiefly from the standpoint of the semi-
classical vector model. Quantum-mechanical treatments of the same
problems have been made by different investigators and found to lead
to exactly the same formula. Chronologically, the more accurate
quantum mechanics led the way to a simpler formulation of Landd’s
vector model. We have seen in Chaps. IV and IX how, in field-free
space, this model is surprisingly similar to the quantum-mechanical
model of probability-density distributions for the electron. As would be
expected from energy relations, the weak- and strong-field distributions
for the electron on any model should be little different from each other
1 Paschen, F., and E. Back, Ann. d. Phys., 39, 897, 1912. The photograph in
Fig. 10.15 is a copy of the photograph given in Plate VIII of Ann. d. Phys., Vol. 39,
1912.
170 INTRODUCTION TO ATOMIC SPECTRA IChap. X

radially. Angularly, as shown by the vector model, they should change

considerably.
Space-quantized to a vanishing magnetic held the field-free states,
oriented as shown in Fig. 9.8, may be taken to represent the weak-field
states, the Zeeman effect, of the atom. Assuming separate angular
distribution curves for the spin of the electron, similar to the orbital
curves in Fig. 4.3, pictures analogous to the independent precession of
l* and 8* around the field direction H may be formulated for strong
fields, i.e., the Paschen-Back effect.
In passing, it should be mentioned that a given state of the atom
in a weak field is specified by the quantum numbers n, l, j, and m, and in
a strong field by n, l, mi, and mt. For intermediate fields either set
may be used, although, to use an expression introduced by Mulliken,
j, mi, and mB are not good quantum numbers.

Problems
1. Compute the Zeeman pattern (separations and intensities) for the doublet
transition
.
2 Find the total width in wave numbers of the Zeeman pattern of Prob. 1 in a
weak field of 5000 gauss.
3. Compute the weak- and strong-field energies for a diffuse-series doublet, and
tabulate them as in Table 10.2. Plot the initial and final states, as shown in Fig. 10.12,
and indicate the allowed transitions by arrows.
.
4 Plot, as in Fig. 10.14, the field-free lines, the weak-field lines, and the strong-
field lines of the above example. [Note:—Certain components of the forbidden
transition appear in strong fields and should be indicated (see Fig. 13.14)].
5. What field strength would be required to carry the first member of the principal
series of sodium over to the Paschen-Back effect where the separation of the resultant
normal triplet (see Fig. 10.14) is four times the fine-structure separation of the field-
free doublet?
 CHAPTER XI

SINGLET AND TRIPLET SERIES OF TWO-VALENCE-ELECTRON

SYSTEMS

In the preceding chapters only the simplest of the known atomic

systems have been treated in any detail. The simplest of elements,
classified as one-electron systems, always give rise to doublet energy
levels, transitions between which account for their observed doublet
series of spectrum lines. It is now well known that atoms like beryllium,
magnesium, zinc, cadmium, mercury, calcium, strontium, and barium, in
groups IIA and IIB of the periodic table, contain two valence electrons
and give rise to series of singlet and triplet energy levels and spectrum
lines. Some of these series were briefly introduced in Chap. I. Before
taking up any quantum-mechanical or vector-model treatment of two-
electron systems, it is important that we first become familiar with
the observed series, the derived energy level diagrams, and the associated
types of electron orbits.
11.1. General Series Relations.—Like the doublet spectra of the
alkali metals, the singlet and triplet series of the elements Be, Mg,
Zn, Cd, Hg, Ca, Sr, and Ba may be grouped into four chief classes:
sharp, principal, diffuse, and fundamental. In each of these atoms,
however, there are four chief series of singlets as well as four chief series
of triplets. If the chief series for each element are plotted to a frequency
scale, as in Fig. 1.8, a number of important relations are brought out.
In calcium, as well as in most spectra, the fine structure is very small
as compared with the other structure and for such a diagram can be
neglected. The schematic plot in the lower half of the figure brings
out the following relations: (1) The 3S and 3D series have one common
limit and the XS and 1D series another. (2) The frequency difference
between the common limit of the 3S and 3D series and the 3P series
limit Xi is equal to the frequency of the first member of the 3S series.
This is the Rydberg-Schuster law. (3) The frequency difference between
the common limit of the 35 and 3D series and the 3P series limit Xa
is equal to the frequency of the first member of the triplet diffuse series.
This is the Runge law. (4) The above relations are also observed to
hold when the fine structure is taken into account. (5) Relations
(1), (2), and (3) hold for the respective singlet series.
Taking into account the fine structure of the triplets in general it
is observed (see Fig. 1.10) that (1) all members of the sharp series are
171
172 INTRODUCTION TO ATOMIC SPECTRA [Chap. XI

composed of three lines with the same separations and approach a triple
limit; (2) all members of the principal series are composed of three lines
with decreasing separations and approach a single limit; (3) all members
of the diffuse and fundamental series contain six lines, three strong lines
and three satellites, and approach triple limits.

Fig. 11.1.—Energy level diagram for the four chief singlet and triplet series of the neutral
calcium atom. For complete diagram, see Fig 11.9.

As in the doublet series of the alkali metals nearly all of the observed
singlet or triplet series are closely represented by the general series
formula
R_R
(ni - juj)2 (n2 — m)2’
(11.1)

where vn is the radiated frequency of a line in wave numbers, R is the

Rydberg constant (= 109737 cm-1); and mi and m are constants. Thus
each spectrum series, or each fine-structure component of a series, is
given by the difference between a fixed term Ti and a running term Tn\

(11.2)

where ni is an integer and n2 takes successive integral values. With

these relations known to hold for each series, and the relations sum¬
marized above, term diagrams for each element may be constructed.
Sue. 11.1] SINGLET AND TRIPLET SERIES 173

Such a diagram is given for calcium in Fig. 11.1. / This same energy level
diagram may be arrived at in another way.
If all of the limits of the eight chief series of calcium (Fig. 1.8) are
brought together into one common limit, the spectrum lines, including

3P ===== 3D 1 *F -

3p
3D 3F

3P

3P 3D

Fig. 11.2.—Schematic representation of the triplet fine structure of the first four members
of the three chief 3P, 3Z), and 3F term series.

the ones plotted as negative, become the energy levels of Fig. 11.1.
Taking into account the fine structure of the triplet series, shown
in Fig. 1.10, this process transforms the triplet sharp-series lines into
single levels and the three remaining triplet series into series of con-

Table 11 1 —Term Values of the Chief Series in Calcium

Electron
Terms n = 3 71—4 71—5 n — 6 n = 7 n = 8
configuration

\Si 4s ns 17765 8830 5324 3566

3P* 4s np 33989 12730 6778 4343
*0. 4s nd 28933 11547 6557 4252 2998 2259
4s nf 7134 4541 3140 2298 1754

'Sc 4s ns 49305 15988 7518 5028 3417

1Pi 4s np 25652 12573 7626 5371 3880
t>2 4s nd 27455 12006 6385 4315 2995
'P3 4s nf 6961 4500 3122 2290 1750

Hydrogen. 12186 6855 4387 3047 2238 1714

verging triple levels (see Fig. 11.2). Although the so-called *S levels
are single, their triplet nature will be brought out when the atom is
placed in a magnetic field (see Chap. XIII). The assignment of the
total quantum numbers n to the valence electrons in the different levels
 7195 7392 7905
Fia. 11.3.—Photographs of triplets taken from original spectrograms kindly loaned to
the author by Dr. A S. King

11.2. Triplet Fine Structure.—Photographs showing a number of

triplets are given in Fig. 11.3. Lines having the same frequency separa¬
tions are enlarged to the same frequency scale. The sharp triplets,
all with the same separations, arise from transitions starting from the
series of 3S levels and ending on the lowest 3P levels. Similar transitions
from the series of 3L> levels into the lowest 3P levels give rise to the
observed diffuse triplets with satellites. Transitions from the series of
lP terms into the normal state lS give rise to principal-series singlets,
etc. In some spectra, lines arising from transitions between levels
Sec. 11.2) SINGLET AND TRIPLET SERIES 175
higher up on the energy level diagram are observed. These lines do*
not belong to the chief series and are called combination lines. Occasion¬
ally transitions between singlet and triplet levels are also observed. Such
lines are called intercombination lines.
Observations on the fine structure of triplets in general enable one
to formulate certain qualitative rules concerning triplet term intervals: (1)
In each series the triplet fine structures converge rapidly as they approach
a single common limit. (2) 3P intervals are wider than the corresponding
3Z> intervals. (3) 3D intervals are wider than the corresponding ZF
intervals. (4) Corresponding triplet intervals in corresponding elements
increase with atomic number. This latter rule is illustrated by the
values given in Table 11.2.
Table 11.2.—Triplet Fine-structure Intervals
(First member of the principal and diffuse series showing increase in wave-number
separation with atomic number)

Element Z 3Pi-8P, 3Dr-’D, *Di-*D2

1
Be 4 2 36 0 69 Unresolved
Mg 12 40 92 19.89 Unresolved
Zn 30 388.90 189 8 5 5 3.4
Cd 48 1171.00 541 8 18 2 11 7
Hg 80 4630.60 1767.3 35 1 60.0

Ca 20 105 9 52 30 21 7 13 9
Sr 38 394.4 187.1 100.3 60 0
Ba 56 878.2 370.6 381.1 181.5

A brief study of the diffuse and fundamental triplets will show the
reader that certain general selection rules are in operation. These
rules are most easily stated in terms of the quantum numbers or quantum
values of the different energy states. For various types of levels the
capital letter L is almost universally used and assigned the following
quantum values:
L = 0, 1, 2, 3, 4, 5,
for S, P, D, F, G, H, . . . terms.
The subscripts to the right of each term are the inner quantum numbers
J and always take the following values:
'So 'Pi 'D2 'F 3 'Ga 'Hi

*Si 3iVPi3P2 3Z>i3Z>23Z>3 3P23P33P4 *GJGSGb

The selection rules for triplet-triplet transitions or for singlet-singlet
transitions are1
AL = ± 1 and A J = 0, ±1.
1 Under certain conditions in complex spectra the transitions AL — 0 are allowed,
but not for the chief series being here discussed.
176 INTRODUCTION TO ATOMIC SPECTRA [Chap. XI

The allowed transitions for a number of triplets are shown in Fig. 11.4.
Relative separations and intensities are shown directly below the arrows
by the respective separations and heights of the lines. Qualitatively
the following rules for the intensities are found to hold experimentally:
The stronger lines arise where AL changes in the same direction as AJ.

_iL ±L_ ji- ^KA[) A[)^K

_jL 'Jl
^UAF

Fia. 11.4.—Triplet-triplet transitions showing selection rules and relative intensities.

These are shown by the heavier arrows. Of these the strongest line
arises from the transition involving the largest values of L and J. These
two well-known rules should be memorized, for they are quite generally
applicable in complex spectra. It should be pointed out in passing that
small letters s, Z, and j are used for the quantum values of individual
electrons, whereas capital letters S, L, and J are used to designate terms
in all atomic systems where there are one or more valence electrons.
11.3. The Quantum Numbers n and l of Both Valence Electrons.—In
order to correlate energy levels with the various types of electron orbits
we return to the Bohr-Stoner scheme of the building up of the elements
and the normal states of the atom. According to Table 5.2, the complete
normal electron configurations for the neutral atoms of Groups IIA and
[IB (Table 5.1) are as follows:

Table 11.3.—Electron Configurations

4 Be, Is* 2s*
12 Mg, Is* 2s* 2 p® 3s*
20 Ca, Is* 2s* 2 p« 3s* 3 p« 4s*
30 Zn, Is* 2s* 2 p® 3s* 3 P6 3d10 4s*
38 Sr, Is* 2s* 2p® 3s* 3 p® 3d10 4s* 4 p® 5s*
48 Cd, Is* 2s* 2p« 3s* 3p« 3d10 4s* 4p6 4d10 6s*
56 Ba, Is* 2s* 2p® 3s* 3p« 3d10 4s* 4p« 4d10 5s* 5 p® 6s*
80 Hg, Is* 2s* 2 p® 3s* 3 p® 3d10 4s* 4 p® 4d10 4/14 5s* 5p®

It is the last two equivalent electrons in each of these configurations (set

in heavy type) that are responsible for the chemical valence of two and
Sec. 11.3] SINGLET AND TRIPLET SERIES 177

are chiefly responsible for the general characteristics of their optical

spectra.
Just as in the alkali metals we may think of all but the last subshell
of electrons as comprising a core of completed subshells around which
the valence electrons move in various types of orbits, some penetrating
the core and coming close to the nucleus and others remaining well
outside the core. Thus in calcium, on the Bohr-Stoner scheme, there
are in the normal state two equivalent electrons in 4s orbits. If now
the calcium atom is excited, by some means or other, either or both of
these electrons may be excited to higher energy levels. It turns out
that all eight of the chief series of spectrum lines and energy levels result
from the excitation of only one of the valence electrons.1
Noting what subshells of electrons are already filled in calcium, one
can predict with reasonable certainty that the first excited states of
the atom should be D or P states, corresponding to the excited electron
being in a 3d or a 4p orbit, respectively. Experiment confirms this,
the levels 3Z), lD, 3P, and lP coming almost together (see Fig. 11.1). The
first 3P along with the lP close by is therefore attributed to one of
the valence electrons being in a 4s orbit and the other in a 4p orbit. The
abbreviated electron configuration for these states is written 4s 4p.
Similarly the first 3Z) and lD are attributed to the electron configuration
4s 3d. By the same process the second °P and lP are assigned the con¬
figuration 4s 5p. It is to be noted that close by each triplet a singlet
level with the same L value is found. In agreement with theory, the
single exception to this observed rule is the normal state 4s4s,1$o, where
no 3&i term has ever been found. Exceptions of this kind always occur
where the electrons are equivalent, i.e., have the same n and l values.
This rule of finding a singlet and triplet close together fits in with
the theory of the spinning electron and will be treated in the next chapter.
Continuing the assignments of the observed energy levels, successive
members of the 3P and lP series in calcium have the electron configura¬
tions 4$4p, 4s5p, 4s6p, . . . , 4s<x>p, or in short 4snp, where n — 4,
5, 6, . . . , oo (see Table 11.1). When n becomes infinite the p electron
is removed from the atom and the atom is said to be ionized. The diffuse
term series 3Z> and 1D are represented by the configurations 4snd, where
n = 3, 4, 5, • • • , oo, and the fundamental series by 4snf, where n = 4>
5, 6, . . . oo. Regardless of the series of orbits through which we may
imagine the electron to be carried in ionizing the atom, we are finally
left with one valence electron in a 4s state. This is as it should be, for
all eight term series approach a common limit. Hereafter the universally
adopted notation will be used in which the designation of all of the
valence electrons is given preceding the term symbols 4s5p,3P*; 4«3d;
1 When both electrons are raised to excited states the atom gives rise to a complex
system of energy levels. These will be treated in the next chapter.
178 INTRODUCTION TO ATOMIC SPECTRA [Chap. XI

zDi; 4s4/,1F8; etc. When a certain term is to be given frequently, the

configuration is written down the first time and subsequently omitted.
11.4. Penetrating and Nonpenetrating Electrons for Two-electron
Systems.—A quantum-mechanical picture of the neutral calcium atom
is shown in Fig. 11.5, with the corresponding classical model below.
Above, the probability-density distribution 4xr2^* is plotted as a
function of r, the electron-nuclear distance in units of the radius of the
first Bohr circular orbit (ai
= 0.528 A.)
(see Chaps. IV and IX). The

Fig. 11.5.—Quantum-mechanical model and the corresponding orbital model of the

neutral calcium atom. The two valence electrons are shown separately, one in a pene¬
trating 4a orbit and the other in a nonpenetrating 4/ orbit.

shaded area represents the calcium core of 18 electrons. The two

valence electrons are shown separately, one in a penetrating 4s orbit
and the other in a nonpenetrating 4/ orbit. On the classical model
the orbits are drawn according to model a (see Fig. 4.8).
As in the one-electron systems the penetrating of the various possible
electron orbits into the atom core is measured by the deviation of the
term values from those of hydrogen. From the energy level diagram
of calcium in Fig. 11.1, or from the term values in Table 11.1, it is observed
(1) that only the levels involving / electrons are hydrogen-like, and (2)
that the order of penetration is/, d, p, and s, e.g.> for the total quantum
number n = 5 the order is 4s5/, 4s5d, 46‘5p, and 4s5s.
Two methods have already been given by which this penetration
can be measured, one by modifying the nuclear charge in the hydrogen
formula T = RZ2/n2, and the other in modifying the total quantum
number. On the one hand there is the general term formula

T = 7 ff)2, (11.8)
n'1
where Z is the atomic number and a is a screening constant, and on the
ether hand the Rydberg formula

T
zi (11.4)
(n — m)*
Sec. 11.5] SINGLET AND TRIPLET SERIES 179

where Z0 is the effective nuclear charge when the'electron is well outside*

the core, n is the total quantum number of the excited electron, and /x
is the quantum defect. Values of Z — <r and of calculated from the
above Eqs. (11.3) and (11.4), and Table 11.1, are given in Table 11.4,
for calcium. For nonpenetrating orbits, and hydrogen, Z — <r = 1 and
/x = 0. The values of Z — c for the 4sn/ terms clearly show that the
/ electron orbits lie, for the most part, well outside the core of 18 electrons
as well as outside the orbit of the 4s electron and that they move in a
field which is nearly hydrogen-like. All of the other orbit types are
penetrating.

Fiu 11.6.—Energy level diagrams of beryllium, magnesium, zinc, cadmium, and mercury.
(Hydrogen comparison.)

Energy level diagrams of a complete group of elements are given

in Fig. 11.6. These diagrams show throughout that s orbits are pene¬
trating and / orbits are nonpenetrating.
11.5. The Excitation of Both Valence Electrons.—Not all of the
prominent lines in the spectra of the alkaline earths, calcium, stron¬
tium, and barium, can be attr buted to the four chief series of singlets
and triplets: sharp, principal, diffuse, and fundamental. Through
the identification and classification of these additional lines into what
are called muUipletsy Russell and Saunders,1 in 1925, were led to the key
to complex spectra. There are in the spectra of Ca, Sr, and Ba three
prominent groups of lines now recognized as forming a triad of triplet

1 Russell, H. N., and F. A. Saunders, Aatrophys. /our., 61, 38, 1925.

180 INTRODUCTION TO ATOMIC SPECTRA [Chap. XI

multiplets. These three great triplet multiplets, photographs of which

are given in Fig. 11.7, for calcium, arise from transitions which start
Table 11.4.—Penetrating and Nonpenetrating Orbits in Calcium
[Values of the effective nuclear charge Z—a (in heavy type) and values of the quantum
defect m (in italics)]

4* nd lDi
*Dt

. 1.0 0.0 1.0 0.1 1.0 0.1 1.0 0.1 Nonpenetrating

. 1.0 0.1 1.0 0.1 1.0 0.1 1.0 0.1

Hydrogen.. 1.0 0.0 1.0 0.0 1.0 0.0 1.0 0.0 1.0 0.0

from three triplet terms which have no place in the ordinary chief series
of terms and end on the lowest 3D 1,2,3 of the diffuse series. The 3Z>
intervals, common to all three multiplets, are shown by brackets in the
frequency plot. It is to be noted that each multiplet is composed of
Vv _3pO
,du/3f2°34 3DUjf D1°2.3 UU.5 r0,U

CD'V O) O lO CM
<7> 0> GO vo «£>
ID LO if) CM CM CM
ID ID lO if) if) UO

GREEN GREEN
Fig. 11.7.—Photographs and frequency plot of the great calcium triad of triplet multiplets
one in the red region of the spectrum and the other two in the green.

three relatively strong lines and three or four fainter lines. The observed
intensities indicate that the inner quantum numbers of the upper triad of
Sac. 11.5] SINGLET AND TRIPLET SERIES 181
triplet levels are those of sP0li.*, and */Ys,4 terms. The wave-'
lengths, frequencies, and estimated relative intensities are given in
multiplet form in Table 11.5.
Table 11.5.—Relative Intensities, Wave-lengths, and Frequencies of thb
Great Calcium Triad
(Given in multiplet form)

*Dt 21.77 *Dt IS. 89 >Di

150
*f< 6439.09
16626.87
78.15
40 125
6471.66 6462.58
16447.72 15469.44
88.28
1 30 80
6508.84 6499.65 6493.79
15859.40 15381.20 15395.08

80 25
5588.74 5581.97
17888.16 17909.85
40.01
30 60 20
*Di 5601.28 5594.46 5590.11
17848.11 17869.87 17883.79
26.73
i 20 50
3/>i 5602.83 5598.48
17843 18 17867.03

60 20 2
5270.27 5264.24 5260.38
18969.08 18990.89 19004.77
4.80
40 20
'Pi 5265.56 5261.70
18986.07 18999.99
1.94
26
*Po 5262.24
18998.05

In addition to this triad Russell and Saunders identified, in calcium,

a series of anomalous multiplets in the ultra-violet spectrum. At the
time of discovery this series was peculiarly different from the ordinary
series, for the upper terms combine like terms with the low 4s4p,
9Po.i.s terms of the principal series. One of the five members of this
 182 INTRODUCTION TO ATOMIC SPECTRA [Chap. XI

series is shown schematically in Fig. 11.8. The significance of this

5po series of multiplets is due to the fact that the
, 1 , 26 0 running terms have a negative series limit, i.e.,
cm-« a limit which lies above the limit of the chief
$, P, D, and F series. The last three members of
C* the series have negative term values. In such
negative states the atom contains more than
. -- enough energy to ionize it, yet it cannot be
1059 doubted that the atom is still neutral in the sense
^ ,. that it still contains all of its electrons. As
1 523 shown by Russell and Saunders this is the clue
5po-— to anomalous multiplets and to complex spectra.
A complete energy level diagram of calcium, as
it is now known, is given in Fig. 11.9.
If more than enough energy to ionize an atom
be put into the atom, and ionization does not
Fig. H.8.—A >F° - *Pmul- occur, the energy must be divided between two
tipiet observed m calcium, electrons, each one of which is excited to a higher
energy state.1 The series of 8P terms found by Russell and Saunders to

Vp'dVp 3d

Fig. 11.9.—Complete energy level diagram of calcium.

approach a negative limit corresponds to a set of energy states in which

1 The suggestion that two electrons are excited to higher energy states was first
made by Bohr and Wentzel, Phys. Zeits., 24,106,1923, to account for wide separations
in the anomalous spectral terms in calcium and strontium.
Sec. 11.5] SINGLET AND TRIPLET SERIES 183

one of the two valence electrons is in a 3d orbit, while the other occupies
successively the states 3d, 4d, 5d, 6d, etc. With one of the electrons tak¬
ing on larger and larger n values (n—> »), the atom will finally be left
ionized with one electron in a 3d state. The series limit of the anomalous
ZP terms therefore becomes an excited state of the first spark spectrum,
Ca II, and lies above the common limit of the chief series.
Returning to the energy level diagram of Ca II (Fig. 6.4), the normal
state is seen to be in the new notation 4s2S, followed by 3d2D and
4p2P. The first excited state (the metastable state 2D) lies 13711 cm""1
above the normal state 2S. The series limit of the anomalous SP terms
is calculated by Russell and Saunders to be 13961 cm""1 above 2S. The
agreement is so remarkable that there is little doubt that the two are
one and the same.
The question now arises as to how two electrons in d orbits give
rise to 3P terms, or how one electron in a 4p orbit and the other in a 3d
orbit give rise to a triad of triplets like the ones given above. The
answer to this question will serve as a starting point for the next chapter.

Problems

1. With the term values of the chief senes in strontium given, plot an energy level
diagram similar to the one shown in Fig. 11.1. Indicate the hydrogen levels with
dotted lines. For term values see Bacher and Goudsmit, “Atomic Energy States.”
2. Compute values of Z — a and the quantum defect m for the first four members
of the chief term series in strontium. Show which terms indicate penetrating and
nonpenetrating orbits (see Table 11.4).
 CHAPTER XII

THE ATOM MODEL FOR TWO VALENCE ELECTRONS

In the early development of spectrum analysis it was well known

to all spectroscopists that the elements in the center of the periodic
table gave rise to large numbers of spectrum lines yet revealed no appar¬
ent series. While equal wave-number differences and other regularities
were occasionally found in the spectra of these elements, complex spec¬
trum analysis undoubtedly had its beginning with the analysis of the
manganese spectrum by Catalan1 in 1922. Catalan was able to classify
large groups of spectrum lines into what are now called multiplets (see
Fig. 14.15). By 1923 the theory of atomic structure had developed to
such an extent that Land6, by an extension of a theory proposed by
Sommerfeld and Heisenberg, was able to account for the relative fine-
structure separations of the lines within these multiplets and to predict
the anomalous Zeeman pattern of each identified line.2
According to Landes model a multiple term was thought to be due
to the space quantization of a single valence electron with the atom core
(or the atom rumpf as it is sometimes called). To be more specific the
angular momentum of the single valence electron Kh/2iry was assumed
to take only certain quantized angular positions with respect to the core
angular momentum Rh/2ir. While this model, with suitable restrictions,
can be brought into excellent agreement with most experimental obser¬
vations, we now attribute the angular momentum Kh/2nr, and the
rumpf angular momentum Rhj2Tt to the resultant orbital angular
momentum L*h/2ir} and the resultant electron-spin momentum S*h/2irf of
all the extranuclear electrons, respectively.
We have already seen in the preceding chapters that in addition
to the orbital angular momentum Z*/i/27t, where l* ~ \/l(l + 1), and

l — 0, 1, 2, 3, 4, 5, 6, * * *
for s, p, d, ft g, h, i, . . . electrons, respectively,3

each electron is assigned a spin angular momentum s*h/2w, where

1 Catalan, M. A., Trans. Roy. Soc., A, 223, 127, 1922.

* For a complete account of the Land6 vector model, see E. Back and A. Land6,
“Zeemaneffekt.”
* In complex spectra small letters s, p, d, . . . , and s, l, j, and m, are used for
individual electrons and capital letters S, P, Z), . . . , and S, L, J, and AT, are used
for resultant termB.
184
Sac. 12.1] THE ATOM MODEL FOB TWO VALENCE ELECTRONS 185

8* «« y/s(s + 1) and a — J. The angular momentum of any atom con¬

taining n extranuclear electrons is therefore the vector sum of 2n angular
momenta. It will be shown in Sec. 15.6, that all electrons making up
completed subshells of electrons are so quantized with respect to each
other that their resultant angular momentum vanishes. For this reason,
only the electrons in incomplete subshells, in general the valence electrons,
contribute to the resultant angular momentum of the atom. In this
chapter the problem confronting us is that of accounting for the observed
fine structure arising from atomic systems containing two valence
electrons.

Fia. 12.1.—Vector diagrams of the space quantizat of the orbital motions of two
valence electrons.

12.1. //-coupling.—Consider first the orbital motions of two electrons

and let h and Z2 represent their respective orbital quantum numbers, and
l*h/2w and l*h/2t, or better If and Z*, their respective angular momenta.
Following Russell and Saunders, coupling scheme,1 modified accord¬
ing to the quantum-mechanical ideas, If and Z* are quantized with
respect to each other in such a way that they form a resultant L*, where
L* = \/L(L + 1) and
L = 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,
for S, Pf D, F, G, Hf 7, K, L, M, . . . terms, respectively.
Consider, for example, one electron in a p orbit and the other in
a d orbit (see Fig. 12.1 B). Here the two vectors If = \/2 and If = \/6
may orient themselves in any one of three positions, L* = \^2, \/6, and
y/12, corresponding to L = 1, 2, and 3, or to P, 7), and F terms, respec¬
tively. Vector diagrams of other electron configurations commonly
occurring in two-electron systems are also shown in Fig. 12.1.
The following method is very useful in making rapid calculations
of terms. With given values of h and Z2, where Z2 ^ Zi, all integral
1 Russell, H. N., and F. A. Saunders, Astrophys., Jour61, 38, 1925.
186 INTRODUCTION TO ATOMIC SPHCTRA [Chap. XII

values of L from l2 — h to l2 + h are allowed. Making use of this fact

the following term table is formulated.

Table 12.1.
S-8, S ps, P ds, D /*. F
sp, P P’Pf SPD dp, PDF fp, DFG
sd, D Pdy PDF dd, SPDFG fd, PDFGH
*/> F Pfl DFG df, PDFGH //, SPDFGHI

The dot between two electrons of the same type indicates that the
electrons have different total quantum numbers. If they are alike
certain of the terms shown are not allowed. (Such special cases will be
treated in Sec. 13.11.) Where l2 < h the roles of the two electrons are
interchanged.
Returning at this point to the discussion of calcium in Sec. 11.5, we
observe from Table 12.1 how it comes about that a P term may arise
from two electrons in d orbits. It is also observed how with-only two
valence electrons, one in a p orbit and the other in a d orbit, the atom
may be in a 3P, a 3D, or a SF state. In
SPIN-SPIN-COUPLING
calcium these three terms, combining with
another 3D term arising from the con¬

S,* S?
s‘ /•*
r7 figuration 4s3d, give rise to the three great
triplet multiplets of Fig. 11.7.
12.2. Spin-spin-, or ss-coupling.—
With two electrons, each having a spin
angular momentum of s*h/2tt, where
s* = \/s(s + 1) and s = J, there are two
S,«*2 s*-'/2 S|*k s2-J$
ways in which a spin resultant S*h/2ir may
s 0 S-l
Fig. 12.2. -Vector diagrams of
be formed. Let s* and s* represent the
the space quantisation of the spins of respective spin vectors of the two electrons,
two valence electrons. Quantizing these (see Fig. 12.2), we find,
with sf = £\/3 and sj = two resultants, one with S* = 0, and the
other with S* = y/2. These give the resultant quantum values S = 0
and S = 1. The resultant S* = 0 will now be shown to give rise to
singlet terms and S* = y/2 to triplet terms.
12.3. LS-, or Russell-Saunders Coupling.—With the orbital motions
of two electrons coupled together to give a resultant L*f and the spins
of the same electrons coupled together to form S*, both L* and S*
will in turn be coupled together to form «/*, which is a vector representing
J*h/2v the total angular momentum of the atom. The quantum con¬
ditions imposed upon this coupling are that J* = y/J(J + 1), and that
J take non-negative integral values. Consider, as a specific example, the
case of one electron in a p orbit and the other in a d orbit. From Secs.
12.1 and 12.2 we have the following possible values to work with; L -- 1
Sec. 12.3] THE ATOM MODEL FOR TWO VALENCE ELECTRONS 187
/

2, and 3 and S = 0 and 1. Taking first S = 0 and L = 1,2, and 3, wef

construct the vector diagrams at the left in Fig. 12.3. S* being zero,
J* is just equal to L*, and J = 1, 2, and 3. These correspond to three
singlet terms IPi, 1Z>2, and lF*. With S = 1 there are three possibilities
for each of the three Vs. These are shown at the right in Fig. 12.3 and
correspond to three triplets

3Po3Pi3P2 3Z>i3D23Z>3 3F23F33F4.

With these term symbols it may now be pointed out that the super¬
script, which we have already used, expressing the multiplicity of the
fine-structure system to which the term belongs, is always 2S + 1.

Singlets Doublets Triplets

S = 0 S = i S = 1
The term type S, P, D, etc., gives the value of L, and the subscript
gives the resultant value of J. The following method may conveniently

o
L-1S-0 L-2 S-0 L-3 S-0 L"1 S’l L-2 S*1 L-3 S-l
J-l J-2 J-3 J "0,1,2 J“ 1,2,3 J-2,3.4
Fiq. 12.3.—Vector diagrams of two valence electrons in LS- (Russell-Saunders) coupling.

be used for the rapid calculation of J values, when they are not remem¬
bered. With given values of L and S} (1) where L ^ S, all integral
values of J between L — S and L -f S are allowed, or, (2) where S ^ L,
all integral values of J between S — L and S + L are allowed. All
allowed terms arising from some of the more common electron con¬
figurations are given in Table 12.2.

Table 12.2.—Terms Arising from the Interaction of Two Electrons in LS-

coupling (Russell-Saunders Coupling)

8 •8, lS„ *S 1 P'P> 15olPi1D2 **Si3Po,!,J3Dl,*t*

8 P, 'P° 8Po.1.2 P d,
8 d, lDt 3Di,i,z P f> 'Di'Fl'Gi 'Ci.w'F i.i.i’Oi.i.i
8 f, lK *K,*, 4 d-d, ^Sq^P i^Dz^F3*C?4

Terms for which the sum h + U is odd are called odd terms, all others
are called even terms. Odd terms are distinguished by the small super¬
script “°” to the right of the term symbol S, P, D . . . . Thus we
have completed the explanation and meaning of the modem spectro-
188 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII

scopic nomenclature. It is to be noted that with S*1 and L » 0 there

is but one possibility, J * 1. This is in agreement with experiment
that all S levels are single. In the following chapter it will be shown
that in a magnetic field each S level will reveal its true characteristics,
i.e., a 3S level splits up into three components, a *S level into two, and a
*S level remains single.
12.4. The Pauli Exclusion Principle.—If two electrons have the
same total quantum number n, and the same azimuthal quantum number
lt they are called equivalent electrons. Observation shows that when two
electrons are equivalent, certain terms are forbidden. For example, the
normal state of each alkaline-earth element is given by two equivalent
s electrons. Of the two possible terms arising from the schemes devel¬
oped above, only the 1S0 term is observed. As soon as one of the electrons

Fig. 12.4.—Singlet and triplet levels observed in ionised scandium, Sc II.

is excited to an s orbit of different n the *Si term is also observed (see

Fig. 11.1). Simply stated the Pauli exclusion principle is that no two
electrons can have all quantum numbers the same. Although the calcula¬
tion of allowed spectral terms for equivalent electrons will be deferred
to Sec. 13.11, the results will be given here.

Table 12.3.—Terms Arising prom Two Equivalent Electrons

(Pauli exclusion principle in operation)
'S
p\ *S *P lZ>
d\ 'S *P lD *F HI
A lS *P lD *F HI *R ‘I
Bsc. 12.6] THE ATOM MODEL FOR TWO VALENCE ELECTRONS 189

The energy-level diagram of calcium (Fig. 11.9) shows two good

examples of the exclusion principle. The normal lS state is the only
term arising from 4s4s. At the extreme right lS, 3P, and lD arise from
4p4p. Of the terms predicted for 3d3d, only lD and 3P have been
observed. Violations of the Pauli exclusion principle have never been
found. Two equivalent electrons are written, for example, 4p2, p2, or
4p4p.
12.5. Triplet Multiplets in Ionized Scandium, Sc II.—Before going
further with theoretical developments let us consider some of the experi¬
mental observations confirming what has already been derived. For
this purpose a single example will be chosen from the spectrum of singly
ionized scandium, atomic number 21. In this spectrum a fairly complete
set of singlets and one of triplet multiplets have been photographed in
the ultra-violet. From these lines energy levels have been derived
which fall into two groups, one a complete group of initial states and the
other a complete group of final states (see Fig. 12.4). The electron
configurations that have been assigned these levels (see Table 12.2) are
as follows:

Initial states 3d4d ^So lPi lD2 XF3 lGi 3£i 3Po,i,2 3Z>i,2.3 3F2f3.4 ^3.4.5

Final states 3d4p IP?1Dl1F% 3P$,1(28 D\,2.3 zF°2tZA

The quantum values, i.e., the term types assigned each level are
determined by selection rules for complex spectra (see Sec. 12.14).
The upper set is composed of a pentad of singlet and triplet levels and the
lower set of a triad of singlet and triplet levels. It should be noted
that the selection rule for electrons is obeyed in that one electron goes
from a 4d to a 4p orbit, Al = 1, while the other remains in a 3d orbit.
12.6. Coupling Schemes for Two Electrons.—Attempts to calculate
the fine-structure separations of the various energy levels arising from
any given electron configuration have been made by many investigators.
Prior to the development of the quantum mechanics, the vector model,
qualitatively, accounted for the fine structure of all analyzed spectra.
Although quantitative agreement was also found in a few cases, it
remained for the quantum mechanics to bring about a more general
agreement. It should be pointed out that, even at this time, there is
much to be desired especially in the case of the more complex atomic
systems. We shall first consider, in detail, the vector model as it applies
to atoms containing two electrons and then formulate a quantum-
mechanical picture of the same atomic states.
In hydrogen and the alkali metals, the simplest of all atomic systems,
the interaction between the electron spin $* and the orbit l* of the
190 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII

single valence electron splits each energy level P, D, F, . . . into two

fine-structure components. We have already seen in Sec. 8.6, that
this interaction energy, t.e., the shift of each fine-structure level from
the hypothetical center of gravity, is given by
_ /-•*2 _ 7*2 _ «*2\
“•APi,* a • l*s* cos (1*8*) = a ^-^ = T, (12.1)
z
where
Ra*Z*
a cm-1. (12.2)
nH(l + *)(Z + 1)

In the case of two valence electrons there are four angular momenta
If, If, sf, and s% which give rise to the following six possible interactions:

sf with sj, l* with sf, If with sf, (12.3)

If with If, If with sf, If with $*.

Applying Eq. (12.1) to these six interactions, there will be six energy
relations:

ri = aisfsf cos (sfs*)! r3 = ajfsf cos (Ifsf), r& = abZfs? cos (Ifsf),
(12.4)
r2 = adflf cos (Iflf), r4 = ad*2sf cos (l*s%), r6 = a6l*s* cos (Ifsf).

In terms of the classical model of the precession of vectors, each spin

5*, and each orbit l*, produces a field around which the others tend to
carry out a Larmor precession. In some atomic systems and electron
configurations certain of the interactions are large as compared with
others, whereas in other systems and configurations different interactions
predominate. A predominating interaction between any two vectors
may be interpreted to mean that these two vectors precess more rapidly
around their mutual field than around any other. With this in mind
it is easy to see that any attempt to formulate classical models for all
possible coupling schemes would be very difficult. Fortunately, how¬
ever, two rather simple ideal models have been developed, one or the
other of which gives fair agreement with experiment in most cases.
These cases are known as LS-, or Russell-Saunders, coupling, on the one
hand, and jj-coupling, on the other. A large percentage of all known
spectrum lines and energy levels may be classified with one of these
coupling schemes, or with some intermediate but closely associated
scheme.
In LS-coupling, the interaction energies Ti and r2 [Eq. (12.4)] are
assumed to predominate over r3 and r4, while T& and r6 are assumed
negligibly small. In jj-coupling, r3 and r4 are assumed to predominate
over Ti and r2, while r6 and r6 are again assumed negligibly small.
Vector diagrams of these two coupling schemes are shown in Fig. 12.5.
Sec. 12.7] THE ATOM MODEL FOR TWO VALENCE ELECTRONS 191
With Ti and r2 predominant in Fig. 12.5A, sf and sf process rapidly
around their resultant S*f and l* and l* process rapidly around their
resultant L*. Due to the weaker interactions r8 and r4, L* and S*
precess more slowly around their resultant J*.
In Fig. 12.52?, with r3 and r4 predominant, If and sf precess around
their resultant j*, and If and sf precess around their resultant j%. Due
to the weaker interactions Ti and r2, jf and jf precess more slowly around
their mechanical resultant J*. We shall now take up these two ideal
models separately and consider some of the consequences of each.

J* J*

Fia. 12.5.—Ideal vector models for (A) LS-coupling, and (B) jy-coupling. Examples
drawn with lx * 1, U *= 2, / = 3.

12.7. T Factors for LS-coupling.—In deriving an expression for fine

structure, assuming the ideal case of L/S-coupling between the two
valence electrons, we have only to evaluate the first four interaction
energies
Ti = aisfsj cos (sfsf), V3 = ajfsf cos (Ifsf), f .
r2 = a2l*l* cos (1*1*), r4 = ajfsf cos (Ifs*). ' ’
Since sf and sj precess, with fixed angles of inclination, around their
resultant S*, which remains invariant in magnitude (see Fig. 12.5A),
Ti is readily evaluated by applying the cosine law1 for triangles, i.e.,
S*2 = sf2 + s** + 2s*sf cos (s*s*),
Ti = iai(S*2 - s** - s*2). (12.6)
With If and If precessing around their resultant L* with fixed angles
of inclination, r2 is readily evaluated in the same way,
r2 = la2(L*2 - If2 - If2). (12.7)
1 The cosine law as given above is used a great deal in the calculation of interaction
energies and will be referred to hereafter as the cosine law.
i92 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII
Now L* and S*, due to their respective fields, process around their
mechanical resultant in just the same way as l* and s* of a single electron
precess around their resultant j*. The interaction energy due to this
precession is attributed to the coupling between each electron’s l* and
its own 8*, T3 and r4. Here the average values of the cosines must
be evaluated since the angles between the vectors are continually chang¬
ing. Due to the constancy of certain angles throughout the various
precessions, and a well-known theorem in trigonometry (see Sec. 10.7),
this is readily accomplished. For example, the projection of If on
L*, then L* on S*f and finally S* on sf gives, for the average cosine,

cos (Zfsf) = cos (iIfL*) • cos (L*S*) • cos (&*sf).

Similarly, (12.8)
cos (Zfsf) = cos {IfL*) • cos (L*S*) • cos (S*af).

In these equations it is assumed that the individual precessions of sf

and If are much more rapid than that of L* and S* around J*, so that
their components normal to the respective axes of rotation will cancel
out.
Adding these two terms together, taking out the common factor
cos (L*S*), and employing the cosine law for triangles, we obtain the
following interaction energy:

r* + r4 (a3a3 + aAon)L*S* cos (L*S*) — a3 + clacu)

{J*2 - L*2 - S*2), (12.9)
where
sf2 - sf2 + S*2 If2 - If2 + L*2
2/S*2 ' 2L*2
(12.10)
and
Sf2 - sf2 + £*2 If2 - l*2 + L*2
OtA
2 S*2 2 L*2
(12.11)
For any given triplet sf, sf, Zf, If, S*, L*y nh and n% are fixed in
magnitude so that a3, a4, a3, and a4 are constants. Writing

A = dz<x3 *4" a4a4, (12.12)

Eq. (12.9) becomes

r3 + r4 = A •L*S* cos (.L*S*) = $A(J*2 “ L*2 - S*2). (12.13)

In the next section we shall see that this equation yields the Land6
interval rule.
We have now to evaluate the a coefficients to obtain complete expres¬
sions of the energies. If in each case the interactions are magnetic
in character, then formulas of the type shown in Eq. (12.2), can be
derived for each coefficient. A comparison of the calculated energies
with observed fine structure in general shows that, while spin-orbit
Sec. 12.7] THE ATOM MODEL FOR TWO VALENCE ELECTRONS 193

interactions are magnetic in character, orbit-orbit and spin-spin inter-*

actions may be attributed mainly to electrostatic effects. Due to the
general observation that singlet levels lie above the corresponding
triplet levels of the same electron configuration, the coefficient ay of Ty
may be considered negative. This has been explained by Heisenberg
as being due to an electrostatic resonance-interaction phenomenon and
will be considered briefly here, to be taken up again in Sec. 12.17. We
shall assume that Eq. (12.6) gives the singlet-triplet intervals in terms
of the quantum values slf $2, and S, and a constant ay which is negative,
even though it is questionable just what significance can be attached
to this equation.

r -
jf
T>, mmmm
i'/o*
SINGLET r -*Pi SINGLETS-
I
1
■31

-2A
Sp—+T0- pd-^T0-7^ “H
•* A
°A ♦A Ay #. —---A1-

TRIPLET'-^” TRIPLETS '4^— — -iA- — >0-

*3A"
-2 A ->F4
-3p’ —A-
0 -*3

Fig. 12.6.—Schematic representation of the interaction energies between two valence

electrons in LS-coupling.

Schematic representations of the fine structure for two different

electron configurations are shown in Fig. 12.6. The effect of each
interaction Ti, r2, r3, and r4 is shown separately, starting at the left
with only the energy attributed to the n and l values of the two electrons.
With Si = ^ Si — £, and S = 0 or 1, for singlets or triplets, respectively,
we obtain from Eq. (12.6), Ti = —3ai/4, or ai/4. This is the first
splitting shown.
Second comes r2, the W2 interaction which gives rise to the various
term types S, P, D, . . . (see Sec. 12.3). Classically it may be shown,
by the same arguments put forward in considering the spin-orbit inter¬
action of a single electron in Sec. 8.6, that the most stable state for two
orbital motions is one in which the two magnetic moments, and hence
the mechanical moments, are most nearly parallel to each other. In
other words, out of all the terms arising from a given configuration
and with given multiplicity the term with the largest L value should
194 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII

lie deepest. This is part of Hund's rule,1 which applies to LS-coupling

only and states that out of all the terms with the same Lvalue arising from
any given electron configuration (1) a term with the highest multiplicity, i.e.,
the highest S value, will lie deepest, and (2) of these the term with the highest
L value will lie deepest. As a consequence of these considerations the
coefficient a2 in Eq. (12.7) should be negative.
For a pd configuration (Fig. 12.6) with lx = 1 and 12 == 2, there
will be F, D, and P terms with L = 3,2, and 1, respectively. Substitut¬
ing these values in Eq. (12.7), r2 = 2a2, — a2, and —3a2, respectively.
For any configuration involving an s electron, h = 0 and r2 vanishes.
We have already seen in Sec. 8.6, that the spin-orbit interaction
leads to coefficients a3 and a4, given by Eq. (12.2), which are positive.
Substituting the various allowed values of s 1, s2, h, L, S, L, and J in
Eqs. (12.10), (12.11), (12.12), and (12.13), the last splitting r3 + r4,
shown at the right in the figure, is obtained. For singlet terms with
S = 0, the afs and hence the A}s vanish.
We may now write for any fine-structure term value the following
formula:
Tcm-i = To - - r2 - r3 - r4 (12.14)

where T0 is a hypothetical term value for the center of gravity of the

entire electron configuration. The horizontal dotted lines in the figure
represent centers of gravity. For each triplet they lie at points obtained
by assigning each level a weight 2J + 1. For the triplets, or singlets,
they lie at points obtained by assigning each term L a weight 2L + 1.
For the entire configuration in turn the weights 2S + 1 are assigned
the triplets below and the singlets above. The results obtained from the
preceding equations will now be compared with observed fine structures.
12.8. The Lande Interval Rule.—Of the equations derived in the
previous section, Eq. (12.13) is perhaps the most significant, for it
expresses symbolically the well-established Land6 interval rule. Intro¬
ducing the quantum values S, L, and J, this equation becomes

r3 + r4 = + i) - L(L + l) - S(S + i)j (12.15)

and states that for a given triplet, i.e., for given S and L, each fine-
structure term difference is proportional to the larger of the two J
values associated with it. In the following examples,
SP„ 8Pi 3P2 3£>i *D2 3Z>3 3F2 3F3 3F4
the ratios are
1:2 2:3 3:4
This result is directly obtained from Eq. (12.15) by taking the difference
between one term J and the adjacent term J + 1:
1 See Hund, F., “Linienspektren,” p. 124, 1927.
Sac. 12.9] THE ATOM MODEL FOR TWO VALENCE ELECTRONS 195

-AT - Ui(J + 1 )(J + 2) - J(J + 1)}'« A(/ + 1). (12.16)f

The intervals are thus proportional to the larger J values.
Observed separations of many triplets in a number of elements
are in excellent agreement with the interval rule. Several examples

Table 12.4.—Triplet Separations in Wave Numbers Illustrating the Land£

Interval Rule (L£-coupling)

Theoretical interval ratio 2 : 1 Theoretical interval ratio 3:2

Element Element
Obs. Obs.
Configura¬ 3Po - *Pi »Pl - *P! Configura¬ - *D* *D, - *D,
ratio ratio
tion tion

Ca, 3d3d 13.5 26.9 2.0 Ca, 3d48 13.6 21.7 1.6
Ca, 4s4p 52.3 105.9 2.0 Ca, 3d4p 26.7 40.0 1.6
Sr, 5«5p 187.0 394.6 2.1 Zn, 4s4d 3.4 4.6 1.4
Mg, 3s3p 20 0 40.9 2.0 Cd, 5s5d 11.7 18.2 1.6
Zn, 4s4p 190.0 389.0 2.0 Ca, 4s4d 3.8 5.6 1.6

taken from one group of elements in the periodic table are given in Table
12.4.
The agreement with theory clearly justifies the use of the vector
model and identifies the electron coupling as Russell-Saunders.
A graphical representation of the Land6 interval rule is shown in
Fig. 12.7 for a 3D term. With
S = landL = 2, S* is shown quan¬
tized with respect to L* in the three
allowed positions J = 1, 2, and 3.
The J vectors are not shown.
Projecting S* on L*, we obtain S*
cos (L *S *). The resultant intervals
have the ratio 2:3, and the central
dotted line is at the center of
gravity.
The chief characteristics by
which LS-coupling is readily recog¬
nized are: (1) The singlet-triplet
intervals are large compared with
the triplet fine structure. (2) The Fiq. 12.7.—Graphical representation of the
triplet intervals follow the Land6 Land6 interval rule for a *D term.

interval rule. Both of these conditions are observed when the coupling
is LS.
12.9. jj-Coupling.—We now turn to the ideal case of ^'-coupling for
two electrons. Before calculating general expressions for the inter¬
action energies on such a scheme, let us first determine all of the various
196 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII

spectral terms arising from any given electron configuration in ^'-coupling.

The spin 8* of each electron is quantized with respect to its own l* to
form a resultant jf, such that ji takes half-integral values only. The
two j*8 are in turn quantized with respect to each other to form a result¬
ant J*, such that J takes integral values only. Consider, for example, the
configuration pd, with $i = h = 1, s2 = £, and i2 = 2. For the p
electron, ji = i, or £; and for the d electron, j2 = £, or f. Combining
these four values in all possible ways (see Fig. 12.8), one obtains from
ji = £ andy2 = f, J = 1, and 2; from^i * £ and j% = £, / = 0,1, 2, and
3; from ji = £ and j2 = 4, J = 2, and 3; and from ji = | and j2 =

Jr!*Js*S J,-34 Jj-^2 j2- ^2

J-1.2 J-0.1.2,3 J-2,3 J’ 1,2,3,4
Fig. 12.8.—Vector diagrams of two valence electrons in jj-coupling.

J *» 1, 2, 3, and 4. This is exactly the same number of terms arrived

at on the L&-coupling scheme, and they have exactly the same set of
resultant J values.
The question arises as to how each term is to be designated when
the coupling is jj9 and L and S no longer are constants. Although no
scheme has been generally adopted, such terms will here be designated
by writing down the two j values in parentheses followed by the resultant
J value as a subscript, e.g., (| -f)4. When the electron configuration is
required, it will precede these values, e.g., 3d4d,(f |)4.
In setting up the ^’-coupling model in Sec. 12.6, it was assumed that
the interaction between the spin of each electron and its own orbit is
greater than the interactions between the two spins and the two orbits,
respectively. Proceeding in exactly the same way as for L&-coupling
the cosine terms in the T factors [Eq. (12.5)] must be evaluated. With
the angle between sf and If fixed, and the angle between sf and If fixed,
Ta and r4 are readily calculated from the cosine law for triangles:

Ta = W;f2 - IV - *V), (12.17)

r4 = ia4(i*2 - ip - 8P). (12.18)
Since the angle between s* and s$ and the angle between If and l*
are continually changing (see Fig. 12.52?), the cosines in Ti and r2
must be averaged. Projecting sf on jf, then jf on jf, and finally jj on 8$,
Sue. 12.9J TIIE ATOM MODEL FOR TWO VALENCE ELECTRONS 197

Ti * ojsfsjf cos (s?sj) = aisfs* cos (sfjf) cos (j*jt) cos (jjs?) (12.19)
and, similarly,
Ti = ajffi cos (1*1*) = OiZfZJ cos (Jfr'f) cos (j*j*) cos (jtl*). (12.20)
Each of these cosines is constant and readily evaluated with the cosine
law:
Ti + r2 = i(ai/3i + atEt){J*2 - jV - jV\, (12.21)
where
R ay + jv - iy a? + j? - iy .
(12.22)
Pl 2jV 2jt*
and
_ i? + jy - «p is* + jy - s*t\ (12.23)
Pi 2j? 2jV
Writing
A = Qifii + (I2P2, (12.24)
Eq. (12.21) may be written

r, + r, = Aj*j* cos (j*jt) = hA(J*> - j** - i«). (12.25)

In applying these equations, consider the configuration pa as an
example. Here with = \,h = 1, s2 = ), and h = 0, four values of J

Fiq. 12.9.—Vector diagrams for a ps electron configuration in ^’-coupling.

are possible. Vector diagrams are shown in Fig. 12.9. Substituting

in Eqs. (12.17) and (12.18), r3 = o3/2, or —a3, and r4 = 0. These
splittings are shown schematically at the left in Fig. 12.10. Substituting
in Eqs. (12.21), (12.22), and (12.23), the values Ti + r2 = —5oi/12,
+3ai/12, —ai/12, and +3ai/12 are calculated for the states (j £)i,
(I £)«> (i i)i, and (i i)0, respectively. In ^'-coupling the spin-spin
interaction is greatly reduced in magnitude (as compared with L5-cou¬
pling). The coupling, however is such that the coefficient ai is still
negative. For this reason the terms have the order shown in the figure.
A good example of a ps configuration revealing ^‘-coupling is to be
found in tin. The observed separations are given at the right in Fig.
12.10. Each observed separation has been divided into two parts to
show the magnitude of the different interactions. It is to be noted, as
198 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII

shown by the formulas, that the large interaction is due to r8, the spin-
orbit interaction of the p electron; the s electron, with h = 0, contributes
nothing (ri = 0). That this splitting is due to the p electron is evidenced
by the 5p,2P|,j separation, observed in the ionized spectrum where the
6s electron is absent. It is seen that the effect of the addition of a 6s
electron to Sn II, 6p, to form Sn I, 5p6s, is to split each doublet level
into two levels as calculated. This smaller interaction is due to the
spin-spin coupling IY If this latter splitting were due to the interaction
between h and s8 [see Eq. (12.4)], the two upper terms would be inverted.

Fig. 12.10.—Schematic representation of the interaction energies between two valence

electrons.

12.10. jj-coupling in the Carbon Group of Elements.—Good examples

of atomic systems revealing jj’-coupling between two valence electrons
are to be found in the carbon group of elements (groups IV and IV B of the
periodic table). Energy level diagrams of these elements, carbon, silicon,
germanium, tin, and lead, are so nearly alike, generally, that a study
of any one of them will illustrate the main features of all. Germanium,
atomic number 32, will serve as a typical example (see Fig. 12.11). The
normal state of this atom is given by the complete electron configuration
(see Table 5.3):
Ge, 32, Is2 2s2 23s2 3p6 3d10 4s2 4p2.

All of these electrons, with the exception of the last two similar p electrons,
form completed subshells. Two similar p electrons (see Table 12.3)
give rise to the terms ^So, *P<u,2, lDt. Applying Hund's rule, which is
always valid for normal electron configurations, the 3P0,i,2 terms should
lie deepest, followed by xDi and then %, as observed.
Exciting the atom now by raising one electron to the lowest possible
available state 5s, changes the configuration to 4p5s. From Table 12.2
Sue. 12.10] THE ATOM MODEL FOR TWO VALENCE ELECTRONS 199
these two electrons give rise to lPi and 3Po,i,s terms. On examining
the fine-structure intervals, it is observed that the coupling is not LS
but very closely jj. These four terms constitute the first member of a
series of terms arising from the series of configurations 4pns, where

Fia. 12.11.—Energy level diagram of germanium, Ge I below and Ge II above.

n — 5, 6, 7, • • • oo. When the limit of this series is reached, the atom

is ionized with but one 4p electron remaining. This is the normal *Pj.|
state of the ionized atom.
Starting at the normal state of Ge I again, the excitation of one
of the electrons to either a 5p or 4d state raises the atom to one of the
states shown in the figure. The circles represent states not yet identified
in germanium but known in other elements in this group. These terms
200 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII

likewise form series, which have as a limit the 4p,*P*.| state of the
ionized atom. Subsequent excitation of the ionized atom will raise the
one remaining electron to other doublet states, which as series have a
single limit, lSo. When this limit is reached, the atom is doubly ionized
and 30 electrons still remain in completed subshells.
In taking up each one of the carbon group of elements in detail,
good examples of intermediate coupling between LS- and ^-coupling
are found. Consider the first member of the 4pns series mentioned
above. In carbon this first member, due to the configuration 2p3s,
i,
shows good LjS-coupling; the 8P0 — 8P 3Pi — 3Pi intervals are 20 and 40
cm"1, with the theoretical ratio 1:2 and the lPi — 3P2 interval 1589 cm"1

Carbon Silicon Germanium Tin Lead Carbon Silicon Germanium Tin Lead

Jj
CoudmS
4p5s 5p6s \\ 6p7s

Fig. 12.12.—LS- to .//-coupling as shown by the normal and first excited states in the
carbon group of elements.

is correspondingly large. These levels are shown at the left of the right-
hand figure in Fig. 12.12. In lead the same four levels have gone over
to ^-coupling with two groups of two levels rather far apart. The
intermediate elements (Si, Ge, and Sn) furnish transition stages between
the two extreme cases. While jj-coupling for similar electrons has
not been treated above, the terms making up the normal states of these
atoms show LS-coupling in C and relatively good jj-coupling in Pb.
12.11. Term Series and Limits in Two-electron Systems.—The
fine structure of two series of singlet and triplet terms is shown graphically
in Fig. 12.13. Fine-structure intervals are plotted vertically, and the
term values, on a very much smaller scale, horizontally. The cadmium
series, it is observed, shows fairly good LS-coupling as far as the series
is observed. The silicon series, on the other hand, shows fairly good
L/S-coupling for the first series member and goes over to good ^-coupling
as the series approaches the double limit.
In the cadmium series 5snp, the 5s electron remains in the Same
tightly bound state while the p electron takes on one of the series of
more and more loosely bound states 5p, 6p, 7p, . . . The limit
Sec. 12.121 THE ATOM MODEL FOR TWO VALENCE ELECTRONS 201

of this series is consequently 5s,2£j. In silicon the p electron is in a

tightly bound 3p state and the s electron takes on one of a series of more
loosely bound states 4s, 5s, 6s, . . . oo. The limit of this series is
3p,2Pj.|. In more complex configuration series like the 4pnd series in
germanium (Fig. 12.11), ^-coupling is also to be expected near the series
limits.
Although a scheme for predicting the terms that go to each limit
has been developed by Hund1 for complex spectra in general, Shenstone2
has shown that the rules are often incorrect and cannot be relied upon.
The very simple rule that levels with the same J values never cross,

CADMIUM SILICON
Fig. 12.13.—Two types of term series and series limits.

along with the jj-coupling scheme developed in Fig. 12.8, will show what
levels go to each limit.
One interesting result brought out in Fig. 12.13 is that the over¬
all separation of each of the 3P terms in silicon is approximately the
same as that of the 2P limit. This quite frequent observation occurs
when one of the electrons is an s electron. For a ps configuration in
L£-coupiing (see Fig. 12.6) the 3P0 — interval is calculated to be
3A, or 3a3/2. In ^/-coupling the same interval (see Fig. 12.10) is again
3a3/2. For the ionized atom the 2P\ — 2P% interval is given by Eq.
(8.18) (see also Fig. 8.9) as 3a/2. In cases where the $ electron does not
appreciably change the effective nuclear charge for the p electron, the
a’s in each case should be nearly the same. These conditions are met
in a number of other atoms, and the same phenomenon is observed in
the following series, in C I, N II, and O III, 2pns, *P, where n = 3, 4,
5, . . . oo; in Si I and P II, 3pns, 3P, where n = 4, 5, 6, • • • oo; in
Sc II, Ti III, V, IV, and Cr V, 3dns, 3D, where n = 4, 5, 6, • • • oo.
12.12. The Great Calcium Triad.—Sufficient preparation has now
been made to enable us to return to the energy levels of the well-known
calcium triad and to calculate the width of each triplet in the upper set

1 Hund, F., Zeits. f. Phys., 52, 601, 1928, and previous papers.
,Shbnstonei A. G., Nature, 121, 619, 1928; 122, 727, 1928.
202 INTRODUCTION TO ATOMIC SPECTRA {Chap. XII

of levels. The relative positions of the 8P, *Df 8F, lP, lD, and ]F terms
arising from the electron configuration 3d4p may be seen in the energy
level diagram of calcium (Fig. 11.9). Assuming LS-coupling for this
configuration the triplet intervals may be calculated with a reasonable
degree of accuracy. With the treatment of LS-coupling in Sec. 12.7, we
have only to determine the values of two coefficients, <z8 and a4, in the
A’s at the bottom of Fig. 12.6:
—, 3a4#
*P, A = 4 1- 4 »
CLl . 5o4 #
*£, A' (12.26)
12 + T2;
*F, A" = ?i _L. —
6 3*

Since a8 and o4 are both due to the interaction between the spin
of each electron and its own orbit, they may be evaluated directly from
the electron configurations 4p4s and 3d4s, respectively. Here with one
electron in an 8 orbit the observed triplet separations 4p4s,3P and 3d4s,
lD are due to the 4p and 3d electrons, respectively. The following table
gives the observed intervals and the calculated coefficients [see Eq.
(12.9)].
Table 12.5.—Coefficients for 4p and 3d Electrons in Calcium

Configuration Term Triplet separations Coefficients

ioJ £
tu

4p4« 52 cm-1 106 cm-1 A 2A a3 = 105 cm-1

II
et<M

3449 14 21 2A 3A aa = 14
II

Using these values of 105 cm-1 and 14 cm-1, for a3 and a4, in the
preceding Eq. (12.26), the following values for the 8P, 3D, and 3F triad are
obtained:

Interval •F, - >Ft

Observed.
Calculated.
166 cm"1
156 HH
Although the intervals throughout this configuration do not indicate
ideal LS-coupling, the agreement between observed and calculated
intervals is quite good.
12.18. The Branching Rule.—The branching rule is a very simple
rule frequently used in the construction of energy level diagrams.
Suppose, for example, that the atom of germanium has been doubly
ionised. This state of the atom is represented in Fig. 12.11 by the
Sec. 12.14] THE ATOM MODEL FOR TWO VALENCE ELECTRONS 203

term at the top of the diagram. Now allow one electron to return to
the atom and end up in a 4p orbit. This state, 4p,2P, of the atom is
represented in the middle of the diagram as the lowest doublet state
of the singly ionized atom and at the same time the doublet limit of
the various series of the neutral atom. Suppose now that the second
electron is allowed to return to the atom in a 5s, or a 5p, or a 4d orbit.
The various possible states of the atom corresponding to these configure
tions may be written down as follows:

4p 4p 4p
Ionized atom
2P ip 2P

A
1P 3P 'S'P'D 3S3P3D lPhDlF 3P3D3F
Neutral atom
4p5s 4 p5p 4p4d

By adding the second electron, the multiplicity goes up one to

triplets and down one to singlets. If the added electron is in an s orbit
(l = 0), the L value is not changed. Adding a 5p electron (1=1),
the L values go up one and down one and include all integral values in
between. Adding a 4d electron to a 4p2P state gives the same resultant
terms as adding a 4p electron to a 4d2D state; the multiplicity goes
up and down one, and the L values go up and down one from 2Z>, the
larger of the two, and include all integral values in between.
Consider, as another example, the energy levels of calcium (see
Fig. 11.9). Starting with the ionized atom in a 3d,2D state, let the
second electron return to a 4s, a 4p, or a 4d orbit. The various possible
states arising from these configurations are:

3d 3d 3d
Ionized atom
2D 2D 2D

A
lD 3D
^/\
ipiD'F 3P3D3F 'S'P'D'F'G *S*P*D*F*G
Neutral atom
3dis 3dip 3did

The 4p2P and 3d2D terms in these two atoms are known as parent
terms to the ones below them. If the L values of a parent term is equal
to or greater than the l of the added electron, the allowed terms go from
L — l to L + 1. If l > L, they go from l — L to l + L (see Sec. 12.1
and Fig. 12.1).
12.14. Selection Rules.—An extension of the selection rules of
hydrogen and the alkali metals to two-electron systems introduces
new sets of rules, of which those for one-electron spectra may be thought
of as special cases. With two electrons taking part in producing
204 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII

the various types of terms, transitions may occur in which two electrons
jump simultaneously with the emission of a single radiated frequency.
Selection rules for two-electron systems in general may be written as

AZi = ±1 and Al2 = 0, ±2. (12.27)

If a single electron jumps, the l value of one changes by unity, and the
other by zero. If a double electron jump occurs, the l value of one
changes by unity and the other by zero or by two. There are no restric¬
tions on the total quantum number n of either electron. For the various
types of terms arising from all possible electron configurations the
further restrictions are divided into two parts.
A. For L/S-coupling the further restrictions are:

AS = 0
AL = 0, ±1 (12.28)
AJ = 0, ± 1 (0 —> 0 excluded)

The singlet and triplet transitions shown for ionized scandium in Fig.
12.4 will serve as good examples of these rules.
B. For .^/-coupling the further restrictions are:

Aji = 0 >
A . „ , i Vor vice versa (12.29)
Aj 2 = , ± )
0 1

AJ = 0, ± 1 (0—0 excluded)

According to these rules transitions are allowed which, under the rules
for LS-coupling, are not allowed. Intercombination lines, although
weak, are good indications that the coupling in either the initial or final
state is not good Uncoupling.
A more convenient way of expressing the selection rules for the
individual electrons [Eq. (12.27)] has been brought out by the quantum
mechanics. The rules are expressed in terms of the oddness or evenness
of the electron configurations and terms. All spectrum terms arising
from an electron configuration for which the sum of the l values is even
are called even terms. All terms for which the sum is odd are called odd
terms. Odd terms are designated by small exponents “ °.” For example
we write:

3d4d (even): 'Sc, 3Po3Pi3P2, */>*, ZF23F,3FA, 'G,

4s4p (odd): 'Pi, 3p°3p°3po

The selection rules for l values are that only even terms can combine with
odd terms, and only odd terms with even terms. This rule1 was dis¬
covered empirically by Laporte. The even terms in the above example
1 For a quantum-mechanical proof of this rule see H. Weyl, “The Theory of Groups
and Quantum Mechanics,” p. 201, 1931.
Sec. 12.151 THE ATOM MODEL FOR TWO VALENCE ELECTRONS 205

can combine with the lower set in what is called a double electron jump,'
Ad to 4p (AZ = 1), and 3d to 4s (AZ = 2), or 3d to 4p (Al * 1), and
4<Z to 4s (AZ = 2). In calcium the 3dnd,*P terms combining with
4s4p,3P° terms give rise to all but the first member of the anomalous
series observed by Russell and Saunders,1 (see Sec. 11.5). In jy-coupling
odd terms will be designated in analogous fashion, e.g., (? f)J.
12.15. Intensity Relations.—In calculating relative intensities of
spectrum lines in genera) it is not always sufficient to use the summation
rules of Ornstein, Burger, and Dorgelo (see Secs. 8.3 and 10.6). In
thermal excitation, for example, the intensity of a line will depend upon
the temperature of the source. At low temperature there will be very
few atoms in excited states. With an increase in temperature the numbei
in higher states will increase, and hence the possibility of jumping back
with the emission of radiation becomes more probable. Just as in the
classical theory of the emission of energy from an electric oscillator the
intensity of a spectrum line will also depend upon the frequency. The
ratio between the intensities of two emitted spectrum lines may be
written as
hv,n

The first term involves the well-known Boltzmann factor containing

the effective temperature T of the two initial states m and p and is
called the excitation function for each state. The second term is called
Einstein1 s v4 correction. The third term involves the ratios given by
the sum rules, which may be stated as follows: (1) The sum of the intensi¬
ties of all lines of a multiplet which start from a common initial level
is proportional to the quantum weight (2J + 1) of the initial level,
(2) The sum of the intensities of all lines of a multiplet which end on a
common final level is proportional to the quantum weight (2J + 1)
of the final level.2 These sum rules alone are not sufficient to determine
the relative intensities within a multiplet. Before the principles of
the quantum mechanics were introduced, correct intensity formulas
were derived by Kronig,3 Russell,4 and Sommerfeld and Honl,5 by means
of Bohr's correspondence principle and the classical model of the atom.

1 The first member of the series is now ascribed by Russell to the transition 4p* to
4*4p.
2 The quantum weight (2/ + 1) gives the number of magnetic levels into which a
given fine-structure level J will split when the atom is placed in a magnetic field.
3 Kronig, R. de L., Zeits. f. Phys., 31, 885, 1925; 33, 261, 1925.
4 Russell, H. N., Proc. Nat. Acad. Sci., 11, 314, 322, 1925; Nature, 115, 835, 1925.
6 Sommerfeld, A., and H. H6nl, Sitz-ber., Berl. AkacL Wiss., 9, 141, 1925.
206 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII

A Fourier analysis of precessing electron orbits in conjunction with the

sum rules leads to the following formulas:
For transitions L — 1 —► L
B(L +•/ + £ + 1 )(L + J + S){L +J - S)(L + J - S - 1)
J -1-J,/ J
Ij^J, 1 J(J + 1)
B{L — / + S)(L -J+S — 1)(L - J - S — 1 )(L - J - S - 2)

(/ + D
(12.31)

For transitions L —► L
A(L + ^ + S + 1)(L + / - S)(L - / + S + 1)(L - / - S)
J-l- J» / J
A[L(L + 1) + J(J 1) - S(S ± 1)]«(2J ± 1)
ij-;, /
J(J + 1)
4-

A(L + / +5 4- 2)(L + J - S + 1)(L - / +B)(L-J-S-1)

*j + i-* J,/
(/ + «
(12.32)

These equations have since been derived on the quantum mechanics

by Dirac.1 If the initial states of a multiplet are close together and the
final states are not far apart, the vK and temperature corrections will
be very small and the constants A and B need not be determined for
the calculation of relative intensities.
Consider, for example, the calculation of the relative intensities
for a SP — 3D multiplet. Writing down the terms in an array like the
one shown at the left in Table 12.6, and substituting the 5, L, and J
values in Eq. (12.31), the intensities given in the center of the array
are obtained.

Table 12.6.—Illustrating Relative Intensities and the Sum Rule for a zP — ZD

Multiplet

*D, »D, *D, *D, •Dt *Di ‘D, *D, *Di

168 30 2 200 5 *P* 100 17.9 1.2 *P, 100 17.5 1.6
90 30 53.6 17.9
120 t »Pi aPi 54.0 17.5
»Po 40 40 1
aP» 23.8 *P0 22.2
168 120 72
7:6:3

According to the sum rules, the sums of the intensities in the columns
have the ratios given by the integers 7:5:3 below and the sums of
the intensities in the rows have the ratios given by the integers 5:3:1
at the right in the array. The three strongest lines are those for which
L and J change in the same direction and they are called the diagonal
1 Dirac, P. A. M., Proc. Roy. Soc.t A, 111, 281, 1926.
Sac. 12.15] THE ATOM MODEL FOR TWO VALENCE ELECTRONS . 207
/ t
lines. Of these lines the one involving the largest L and J is the strong¬
est. The three weaker lines are called off-diagonal lines. At times
it is convenient to compare the intensities of the lines of each multiplet
with the strongest line in the multiplet designated as 100. These
reduced intensities are given in the center in Table 12.6. Relative
intensities expressed in this way are given in the Appendix for all triplet
combinations up to 3J. The relative intensities of the calcium *P — lD
multiplet (see Fig. 11.7), as measured by Burger and Dorgelo, are given
in the third array. Wherever LS-coupling is revealed by the fine
structure, the calculated intensities are in good agreement with those
observed. Where deviations occur in one they occur as expected in
the other. It should be noted that the sum rules make it possible to
interchange the r61es of the initial and final states in the above given
formulas.
For the relative intensities of lines arising from ^-coupling schemes,
Eqs. (12.31) and (12.32) may be employed by replacing S by ji and L
by j2. Here ji is taken as the quantum value which, in any given transi¬
tion, does not change. Applying the modified formulas to the intensities
of the lines arising from the configurations sp —> s-s, the values shown
in the first two arrays in Table 12.7 are obtained:

Table 12.7
sp, (ii) (if) sp
/ = 0 1 J = 1 2 J — 0 1 1 2

J * 0 s 1 J = 0 24 1 J - 0 12 24 1
*, (ii) 1 - * • «, (ii) •• 8 • 8 ••

/ = 1 3 6 3 J - 1 12 60 3 J - 1 12 24 12 60 3

1 :3 3 : 5 1 : 3 3 : 5

Assuming that the temperature and the vK corrections are negligibly

small, these two sets of three lines each are brought into agreement
with the sum rule by multiplying the first array by four. The two
are brought together at the right in Table 12.7 and are to be compared with
the values obtained for LS-coupling and given in Table 12.8.

Table 12.8

lPx *Po »Pi

\s. 36 1
••
’Si 12 36 60 S

3 1 : 3 : 6

The relative intensities given in this table, as calculated from Eqs.

(12.31), have been multiplied by three.
208 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII

Although no formulas have been derived for intermediate coupling

schemes, the summation rules may be applied in any case. As seen in
the above tables the transitions corresponding to 3£i — SP2, 3&i — zPo,
and lS0 — 8Po will have the same intensities (60,12, and 0) in all coupling
schemes. For the other transitions it is possible to determine only the
sum of the intensities of all transitions starting from, or ending on, levels
with the same J values.
12.16. Relative Intensities of Related Multiplets.—Formulas for
the relative intensities of multiplets arising from certain types of con¬
figurations have been derived by Kronig1 and checked experimentally
in the more complex spectra by Harrison and Engwicht.2 These formulas
have been derived for one-electron jumps where the electron configura¬
tions reveal L£-coupling. Consider, for example, the electron con¬
figurations d-d and dp, which in Sc II have been observed to give rise
to a number of triplet multiplets (see Fig. 12.4).
Let Lo represent the orbital value of the parent term, i.e., the electron
common to both configurations, Z the orbital value of the other electron,
and L their resultant. These three quantum values are coupled together
in just the same fashion as S, L, and J (see Figs. 12.1 and 12.3) and, as
shown by Kronig, lead to the same formulas. Making the substitutions
of L0 for S, Z for L, and L for J in Eqs. (12.31), and (12.32), formulas
for the relative intensities of related multiplets are obtained. It must
be remembered that these formulas apply only after the vA and the
temperature corrections are made. For multiplets lying close together
this is unnecessary.
In the scandium example d-d —> dp, given above, Lo = 2, 1 = 2,
and L = 4, 3, 2, 1, and 0. Substituting these values for S, L, and J
in Eqs. (12.31), and (12.32), the intensities shown in Table 12.9 are
found.

Table 12.9
*P 3F HI

3p° 24 64 42 3

w° 18 70 112 5

3jfp° 8 66 216 7

<! 1:3:5 : 7 : 9\

Although only visual estimates of the lines in all of these multiplets

have been made from photographic plates, the general agreement of
the observed values with those calculated is quite good. On the basis
1 Kronig, R. de L., Zeits. /. Phys., 33, 261, 1926.
2 Harrison, G. R., and H. Engwicht, Jour. Opt. Soc. Amer., 18, 287, 1929.
Sac. 12.17) THE ATOM MODEL FOR TWO VALENCE ELECTRONS 209

of 100 for the strongest multiplet, the calculated and observed valuesf
are as follows:
Table 12.10
*S *P »D *F *G *P «F *Q
tpo
11.1 25.0 19.4 apo 24.0 18.3 14.1
*D° 8.3 32.4 51.8 »D° 4.2 41.0 35.2
»p° 3.7 25.9 100 apo 9.9 22.5 100
calculated observed

Due to nonuniformity in photographic plates the multiplets at

shorter wave-lengths have been underestimated and those at longer
wave-lengths overestimated. Objective measurements of the lines will
doubtless lead to much better agreement.
12.17. Helium and Helium-like Atoms.—Like that of the alkaline
earths the helium spectrum is composed of four chief series of singlets called
parhelium and four chief series of triplets called orthohelium. That the

fine structure of the very narrow orthohelium series was triplets and
not doublets was first predicted by Slater1 and verified by Heisenberg.2
Similar series have been observed also in ionized lithium. The two sets
of series in each of the elements may be arranged into two apparently
noncombining sets of energy levels which, in the case of helium, have
come to be known as parhelium and orthohelium levels.
1 Slater, J. C., Proc. Nat. Acad. Sci11, 732, 1925.
* Heisenberg, *W., ZeiU. f. Phys., 39, 499, 1926.
210 INTRODUCTION TO ATOMIC SPECTRA [CHAJr. XII
Energy level diagrams of helium and singly ionizod lithium are
shown in Fig. 12.14. The enormous energy shift from the normal
l«VSo state to the first excited states 1s2s,1iS0 and 3£i is characteristic
of all the inert gases and similar atomic systems.
Many attempts to calculate the energy levels of helium, a three-body
problem, from a purely theoretical standpoint have been made, and with
reasonable success. The normal state of helium, for example, has been
calculated by Hylleraas1 to be 24.470 volts below the series limit, a
value differing only 0.003 volt from the spectroscopic value of 24.467
volts. For Li+ he obtained the value 75.272 volts, in excellent agreement
with the observed value 75.279 volts.
Heisenberg has shown that the enormous energy difference between
the first two or three triplet levels and their associated singlet levels
is due not to the magnetic interactions of the two electrons but to an
electrostatic resonance interaction. This phenomenon of resonance,
which so frequently arises in the quantum mechanics, is analogous
to the resonance between two mechanical oscillators. The analogy
has been drawn of two similar pendulums, connected to the same support,
and set swinging. If the support is not too rigid the energy of oscillation
is found to shift back and forth between one pendulum and the other.
At one instant one pendulum is at rest and the second vibrates with a
maximum amplitude. Some time later the second pendulum comes
to rest and the first vibrates with the maximum amplitude. It may be
said that the two pendulums have changed places.
For helium-like atoms this resonance phenomenon has been shown
to yield fair agreement with experiment by Heisenberg, Dirac, Slater,
Gaunt, Unsold, Breit, Hylleraas, and others. The first attempts to
solve the three-body problem of helium on the quantum mechanics led
to a wave equation, quite like Schrodinger’s equation for hydrogen, and
of the form,

V!* + Vi* + ^(w + Z-£ + ^ - £)* = 0, (12.33)

where the subscripts refer to the two electrons respectively. The first
term in parentheses represents the total energy of the system; the second
and third terms represent the potential energies of the two electrons,
due to their attraction by the nucleus; the last term represents the
potential energy due to the mutual repulsion of the two electrons. If
it were not for this last term, which accounts for the fine structure we
wish to measure, the equation could be solved by analytical methods.
Neglecting this term, the solution is of the form of the product of two
hydrogen-like functions,
i = la, m«; n, dx, fa) • tH(nb, lb) mb; rlf 6l} fa), (12.34)
1 Hylleraas, E. A., ZeiU. f. Phy54, 347, 1929; 65, 209, 1930.
Sec. 12.18] THE ATOM MODEL FOR TWO VALENCE ELECTRONS 211

where n, l, and m are the quantum numbers of the two electrons and.
r, 6, and <t> their polar coordinates. This solution is degenerate in that
the energy of any state of the atom is not altered by interchanging the
two electrons, i.e., interchanging the coordinates r, 6, and fa

V = ^«(n0, la, fna; r2, 02, fa) • (n&, lb, m&; r\, $i, fa). (12.35)

Treating the e2/ri2 term as a perturbation,1 two solutions may be obtained,

one of which is symmetric in the electrons; i.e., when the coordinates are
interchanged, the wave function is not altered. The other solution is
antisymmetric in that when the coordinates of the two electrons are
interchanged, the wave function changes sign. As the interaction is
made smaller and tends to zero, the limiting form of the solutions is
such that each electron has an equal probability of being found in either
of the two states. This is the austausch principle which has been inter¬
preted as meaning that the two electrons exchange places regularly
as time goes on, i.e., periodically. The rate of this interchange depends
upon the magnitude of the interaction and for the lower excited states
of atoms like helium is of the same order as the frequency of ordinary
light. The total energy is not altered by this interchange, and hence
no radiation takes place. Taking into account the spin of the electrons
it may be shown that all triplet states realized correspond to wave
functions which are antisymmetric in spin and coordinates, i.e., change
sign when both electron spins and electron coordinates are interchanged.
To the triplet system of terms belong functions antisymmetric in the
coordinates and symmetric in the spins. To the singlet system of terms
belong functions symmetric in the coordinates and antisymmetric in
the spins.
12.18. Quantum-mechanical Model of Helium.—A fairly accurate
quantum-mechanical picture of the helium atom in different energy
states may be formed by the superposition of hydrogen-like density-
distribution curves (see Figs. 4.3 and 4.6). In doing this for the excited
states the inner electron, in a Is state, takes on a distribution which is
like the Is of ionized helium, i.e., radially it is one-fourth that of hydrogen
in the normal state. The outer electron, due to almost perfect screening
by the inner electron, is during most of the time in a field not greatly
different from that of hydrogen (see Fig. 12.15). The term values of
helium and ionized lithium, as shown in Fig. 12.14, indicate quite clearly
that the outer electron in all of the excited states is almost hydrogen-like.
It is the overlapping with each other of the dense regions of the dis¬
tribution curves that makes for a greater electrostatic repulsion, or
interaction, i.e., greater singlet-triplet separations. This overlapping

1 For an elementary treatment of this problem the reader is referred to that given
by E. U. Condon and P. M. Morse, “Quantum Mechanics,” Chap. IV.
212 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII

decreases rapidly with an increase in the total quantum number of the

excited electron.
12.19. Fine Structure of Helium-like Atoms.—Taking into account
all of the interactions between two electrons, Heisenberg has calculated
the fine-structure triplet separations for helium and helium-like atoms.1
For many different electron configurations the energy of the singlet
term is very much greater than that of the corresponding triplet (the
inglet lies above the triplet). On the vector model the spins s* and s\

Fia. 12.15.—Probability-density distribution curves for the two electrons of helium in

the normal lsV<So fltate and several of the excited states.

are oppositely directed for the singlets, S = 0, and more nearly in the
same direction for the triplets, S — 1. It follows that the interaction
energy is positive when the angle between $* and s* is large, and negative
when the angle is small (see Sec. 12.7).
According to Heisenberg’s results the interaction energy between
S and L for the configuration ls2p is given by

—AT = - L** - S*2)

, „(/*» - l** - - [(j + 4)2 — {L + muz, + w - {J_ ~ l)8)
-r o 16£/*2

(12.36)

where a is a constant given by Eq. (8.19) and Z is the atomic number.

For the triplet terms 3Po.i,2 of helium, Z = 2; of singly ionized lithium,
Z = 3; of doubly ionized beryllium, Z = 4; etc.:
He I, —AT - a(Y, -J, -f); Be III, —AT = a(f, -f, +|); (12.37)
Li II, -AT - a(Y, -f, +i); B IV, -AT - a(-f, +f)-
1 Heisenberg, W., Zeits. f. Phys., 30, 499, 1926; see also Breit, G., Phys. Rev.,
36, 383, 1930; 30 616, 1932, for later improvements and references.
BBC. 12.19] THE ATOM MODEL FOR TWO VALENCE ELECTRONS 213

These intervals are shown in Fig. 12.16. The dash-dot line repre¬
sents the center of gravity of each triplet. As Z becomes large, the
second and third terms of Eq. (12.36) become negligibly small compared
with the first. Dropping them we obtain

J*2 - L*2 - S*2

—ATZ = A (12.38)
2

This is the Land6 interval rule which for sPo,i,a gives the interval ratio
1:2. The relative separations of the first two elements, He I and Li II,
have been observed to have, quite accurately, the intervals given by

Hei tin Bern m cs~~

3a
2a xnr-
0 V
a
0
-a
2a
vc —
£ __v

3a Z-1 5 4 5 6 X i —»oo

' ' ' ' - —3R

Fia. 12.16.—Relative separations for the l«2p, 8Po,i,j. states of helium-like atoms. (After
Heisenberg.)

Heisenberg’s formula, Eq. (12.37). The fact that the 3P intervals in

He I and Li II are far from the ratio given by the interval rule shows
that the interaction between the s electron spin and the p electron orbit
is of the same order of magnitude as the spin-orbit interaction of the p
electron alone.
Spectrum lines arising from transitions into and out of the ls2p,
8P levels of ionized lithium show that, in addition to the triplet fine
structure of the levels, there is a still finer structure known as hyperfine
structure (see Chap. XVIII). A discussion of the more recent quantum-
mechanical treatments of the interaction of two electrons in complex
spectra will be given at the end of Chap. XIV.

Problems

1. Derive all of the terms arising from the electron configuration fg.
2. Graphically represent the Land6 interval rule for a 9P term as in Fig. 12.
214 INTRODUCTION TO ATOMIC SPECTRA [Chap. XII

.
8 Show for a ds electron configuration that the total lD separation is the same in
both LS- and ^-coupling. Show also that the d electron alone gives the same %D
separation (see Sec. 12.11).
4. Using Eqs. (12.31) and (12.32), calculate the relative intensities in each of the
three triplets arising from the transition pp —► ps. Assume L5-coupling.
5* Calculate the relative multiplet intensities for the triad of triplets arising in
Prob. 4 (see Sec. 12.16).
6. Repeat Prob. 4 assuming jj-coupling between the electrons. Compare these
intensities, as in Tables 12.7 and 12.8, by applying the sum rules.
 CHAPTER XIII
ZEEMAN EFFECT, PASCHEN-BACK EFFECT, AND THE PAULI
EXCLUSION PRINCIPLE FOR TWO ELECTRONS

One of the outstanding successes of the vector model, developed

by Land6 before the advent of the quantum mechanics, is the amazing
accuracy with which it predicts the observed Zeeman patterns of many
spectrum lines. Modified according to the quantum mechanics this
model was employed in the last chapter in the treatment of LS- and
^’-coupling for two electrons. This treatment will be extended in this
chapter to the Zeeman effect, the Paschen-Back effect, and the Pauli
exclusion principle.
13.1. The Magnetic Moment of the Atom.—We have already seen, in
Chap. X, how the total magnetic moment of the atom and the Land6
g factor enter into the development and derivation of expressions for
the Zeeman effect. The same principles used in obtaining these two
factors for a single valence electron will now be extended to atomic
systems containing two valence electrons. With two electrons there
will be four individual magnetic moments to be taken into account;
these are associated with the mechanical moments sf, sf, Zf, and ZJ.
With the magnitude of each of these moments known, their vector sum
will depend upon their mutual orientations. If the coupling between
the four mechanical moments is L/S-coupling (Russell-Saunders coupling),
there will be one set of resultant magnetic moments, whereas, if the
coupling is jj-coupling, there will be a different set.
The magnetic moments of the spin and orbital motions of an electron
have already been given by Eq. (10.4) as

e ergs e ergs
2me gauss* 2me gauss (13.1)

In L&-coupling the two spins are coupled strongly together and

are quantized to form a resultant S* (see Fig. 12.5). Projecting each
spin moment on S* and adding, we obtain the vector sum of the two
magnetic moments,

f cos (sf S*) + s} <

Similarly the two Z’s are coupled together and quantized to form a
resultant L*. Projecting each orbital moment on L* and adding,
215
216 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

e ergs
Mi * K cos (J?L*) + £ cos (ZJL*)}A . • „ (13.3)
11 u ' n v3 /j2t 2mc 2t 2me gauss
Just as the spin and orbit of a single electron are coupled together to
form a resultant moment j*, so L* and S* are coupled together to form
a resultant moment «/* (see Sec. 10.3 and Fig. 10.6). Projecting ns
and nL on J* and adding, we get for the total magnetic moment of the
atom,

M, = [L* cos (L*J*) + 25* cos (5V*)]A • -^Si- (13.4)

1 ' J2ir 2mc gauss v

The terms within the brackets give hj in units of the Bohr magneton,
he/fame. Replacing these terms in the brackets by g times J*f i.e.,

gj* = L* cos (LV*) + 2S* cos (S*J*), (13.5)

we obtain
_ r» he ergs
Hj = gJ* • t-——-• (13.6)
47rmc gauss

Since the angles between L*, S*, and J* are fixed, in LiS-coupling,
the cosine terms in Eq. (13.5) are readily evaluated.

J*2 4- L*2 - S*2

L* cos =
2J* '
(13.7)
J*2 + £*2 - £*2
S* cos =
2J*

Substituting these two cosines in Eq. (13.5), we get for the g factor,

J*2 + S*2 - L*2

0 = 1 + (13.8)
2T*2

This is the Lande g factor in exactly the same form as Eq. (10.11)
for a single electron, except that l* is here replaced by L*, s* by S*f and
j* by J*. Values of the g factor calculated from this equation are given
in the Appendix.
In ^'-coupling the s* and Z* of each electron are coupled together
and quantized to form their own resultant j*. The magnetic moment
of a single electron has already been given in Eqs. (10.6) and (10.7).
With two electrons we may write for each one separately,

w = (13-9)
With j* and j* in turn coupled together and quantized to form a
resultant J*, these moments are projected on J* and added to give the
total magnetic moment of the atom,
Sac. 13.2] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 217

Pi = cos (if/*) + cos OT«f*) (13.10)

= fodf cos (jtJ*) + 9*3? cos ^ (13.11)

, Replacing the terms in the brackets by g times J*, we get

g J* - gd*i cos + g*J? cos 0?^*). (13.12)

A table of g factors for ^'-coupling is given in the Appendix.

LS-Coupling jj-Coup!ing
Fio. 13.1.—The vector model for LS- and /./-coupling in a weak magnetic field (Zeeman
effect).

13.2. The Zeeman Effect.—In a weak magnetic field the atom as a

whole is quantized with H in such a fashion that the projection of the
angular momentum J*h/2ir on H is equal to Mh/2w, where M takes values
differing from each other by unity, from M = +J to M = —J.1 In

1 For a quantum-mechanical proof of this relation see E. U. Condon and P. M.

Morse, “Quantum Mechanics,” Sec. 64.
218 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

terms of the vector model these quantum conditions are expressed as

follows:

J* cos (J*H) - M, where M = 0, ±1, ±2 • - • ±J (13.15)

for singlets and triplets. On the classical model of a precessing atom

(see Fig. 13.1), J* carries out a Larmor precession around H with an
angular velocity given by Eq. (10.4),

<1316>

Multiplying by the projection of the mechanical moment on //, the

energy of this precession (see footnote 3 on page 157)

AW = H ■ g • ^ J* • £ cos (13.17)

In terms of the magnetic quantum number [Eq. (13.15)],

AW, - M ■ , ■ „-£mc (13.18)

Dividing by he, the energy in wave numbers (see Sec. 10.4) is given by

He
—AT\t = M • g^~~c2 cm"1 = M • g • L cm’1. (13.19)

If T represents the term value of a field-free energy level, then in a

weak magnetic field the term value of each magnetic level will be given
by
He
Tm = T — M • cm_1 = T — M g • L cm”1. (13.20)

Before going further it should be pointed out just what assumptions

were made in deriving the preceding expressions for the g factor and the
magnetic interaction energy. On the classical model for LS-coupling
it was assumed (1) that s* and s* precess rapidly around $*, (2) that
l* and l* precess rapidly around L*, (3) that L* and S* precess moie
slowly around J*, and (4) ihat J* in turn precesscs still more slowly
around H. These assumptions, as well as similar ones for jj-coupling,
are necessary in order that the components of the different magnetic
moments normal to their respective axes of precession cancel out. The
quantum-mechanical analogue of these precessions is to be found in
the angular factors of the probability-density distributions (see Sec;.
4.3). The formula for the Land6 g factor has recently been derived
on the quantum mechanics.
As an example of the splitting up of an energy level in a magnetic
field, consider the case of a 3D3 level. With g = $ and J = 3, there will
Sec. 13.2] ,
ZEEMAN EFFECT PASCHEN-BACK EFFECT 219

be seven equally spaced magnetic levels M « 3, 2, 1, 0, —1, —2, and

—3, shifted from the field-free level by M - g — Y> f, |, 0, —J, —}, —Y*
H

Fig. 13.2.—Graphical representation of the splitting of a *D8 level in a weak magnetic field.

A graphical representation of this splitting is shown in Fig. 13.2, where

the vector J* is drawn in the seven allowed posi¬
M- + 2
tions. Multiplying J* by g and projecting on
+1
the H axis, the displacements Mg, shown at the
left, are obtained. If numbers in this figure are 0
multiplied by He/4rmc2, the level separations are -1
given directly in wave numbers. -2
From Eq. (13.19) it is observed that the
relative separations of the magnetic levels of one
term and those of another are determined by the
g factor alone. This fact greatly simplifies the
calculation of Zeeman patterns. As a simple +1
case consider the diffuse-series singlet IPi — 1Z)2. 0

For each of these levels the g factor is equal to -1

unity, so that the initial and final states with s p s
J = 2 and / = 1 are split into five and three
equally spaced levels, respectively (see Fig.
NORMAL TRIPLET
LU
13.3). Fio. 13.3.—Zeeman pat¬
The selection rules for the Zeeman effect tern of a diffuse-series
singlet, showing equal sepa¬
of two-electron systems are just the same as rations in both the initial
those for a single electron (Sec. 10.5). and final states of the atom.

AM = ±1; plane polarized ± to H; 8 components

Viewed _L to field
AM = 0; plane polarized || to H;p components
AM = ±1; circularly polarized; s components
Viewed || to field
AM =* 0: forbidden; p components
220 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

For the 1Pi~1Z>2 transition there are nine allowed combinations

falling into three equally spaced groups (see Fig. 13.3). Having just
the separations given by Lorentz's classical theory of the Zeeman
effect [Eq. (10.1)], this pattern is called a normal triplet. Since the g
factor for all singlet levels is unity, all singlet lines will show a normal-
triplet pattern in a magnetic field. For an observed pattern see Fig. 10.2.
Turning now to the triplet levels, let us consider the Zeeman patterns
arising from a principal-series triplet 3Si - 3P<U,2. Assuming L£-coupling,

the g factors for the initial states 3P0, 3Pi, and 3JP2, and for the final
state ZS\, are &, f, f, and 2, respectively. The resultant magnetic
levels are shown schematically in Fig. 13.4. At the bottom of the
figure the observed patterns are shown for a zinc triplet. With a g factor
of two for the lower state 3Si, the pattern 3Si- 3Po has twice the normal-
triplet separation.
13.3. Intensity Rules for the Zeeman Effect.—Intensity rules for the
Zeeman effect of atomic systems containing more than one valence
electron are independent of the type of coupling and depend only on
the quantum numbers M and J. These rules, first discovered empirically
by Ornstein and Burger,1 may be derived from the sum rules. The
1 Ornstein, L. S., and H. C. Burger, Zeits. f. Phys., 28, 135, 1924; 29, 241, 1924.
Sec. 13.4] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 221
sum of the intensities of all transitions starting from any initial Zeeman
level is equal to the sum starting from any other level. Similarly, the sum
of the intensities of all the transitions arriving at any final Zeeman level is
equal to the sum arriving at any other level. From the classical model
and these rules the following equations have been derived,
Transition J -+ J:

fM-*M ± 1, I = A(J ±M + 1)(/ T M);

\M ->M, I = 4AM2)

Transition J —» J + 1: (13.21)

(M-+M ± 1, 1 = B(J ± M + 1 )(J ± M + 2);

\M->M, I = 4B(J + M + 1)(/ - M + 1).

In deriving these equations account has been taken of the fact that
when observed perpendicular to the field only half the intensity of
the s components is observed. Observations parallel to the field give
the other half of the s components. In any direction the total light
emitted is unpolarized just as in field-free space. The constants A and B
in the above equations are proportionality constants and are not needed
for any given Zeeman pattern. Applying these equations to the principal
triplet of Fig. 13.4, the values shown at the arrow tips are obtained.
In order that the sum rules hold for all of the levels in the figure the
values derived for the first pattern *Si — 3Po have all been multiplied
by two and those of the second pattern 3Si — 3P± by three. Doubling
the s components the sum of the intensities of all transitions starting
from any upper level is 24 and the sum ending on any lower level is 72.
13.4. The Calculation of Zeeman Patterns.—A scheme for the rapid
calculation of Zeeman patterns has been given by Sommerfeld (see
Sec. 10.6). Consider, for example, the complex pattern arising from the
transition 3D3 — 3P2. Here the J values are 3 and 2, and the respective
g factors, from Eq. (13.8), are $ and f. The separation factors Mg
for both the initial and final states are first written down in two rows
with equal values of M directly below and above each other as follows:

M = 3 2 1 0 -1 -2 -3
Mg for initial state:

Mg for final state:

In this array the vertical differences AM — 0 give the p components |,

j, 0, — J, and and the diagonal differences AM — +1 give the s com-
 222 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

ponents ± j, ±J, ±$, ±§, and ±V-. These may be abbreviated to give
the separations (see Fig. 13.5 and Sec. 10.6):
(0), (1), (2), 6, 7, 8, 9, 10
-ATm = ± L cm”"1. (13.22)
6
The common denominator is called the Runge denominator and is
the least common multiple of the two denominators of the Mg factors.
This pattern is plotted in Fig. 13.5. The p components are plotted

V? Vi
i■
Anomalous Zeeman

’ 1 1 ‘1 _ Patterns
for
Triplets
r'i 1
®i& |
II' | l Ii.Ml| ’V+1 H |
5d,3ds 3d,5d, |

r 1 I 1 .1, ,1.

w L_ ill
i r %\
ir i l li, ,|iil ■ .'
3r 5C
'5 rl

r 1 1 f jJTr i 11

L_- ii Ml. • ,1 ■ ~rh"r i

"mL. 5D,Jlj

1 Y 1r
TP ''iii

«il yj 1L_ w
r i T | '1 W ■ "W
nili' 1 fi 1 1 i

Fig. 13.5.—Anomalous Zeeman patterns for triplet combinations. Dots show normal
triplet intervals (LS-coupling).

above the line in the usual way, and the s components are plotted below.
The heights of the lines represent the relative intensities as calculated
from the equations in Sec. 13.3.
13.5. LS- and jj-coupling and the g Sum Rule.—In classifying a
given spectrum line for the first time it is not always known to just
what type of coupling scheme the corresponding levels belong. Since
the g factors are in many instances different for ^/-coupling than they
are for L£-coupling, the Zeeman patterns will also be different. For
this reason it is sometimes difficult to assign a given pattern to a definite
transition. This uncertainty is often alleviated by Pauli's1 so-called
g sum rule. This rule states that out of all the states arising from a
given electron configuration the sum of the g factors for levels with the
same J value is a constant independent of the coupling scheme. As
an example, consider the four terms 8Po,i.2 and IPi, arising from the
configuration ps. The following table gives the g values of these four

1 Pauli, W., Zeita f. Phys., 16, 155, 1923.

Sec. 13.6) ZEEMAN EFFECT, FASCHEN-BACK EFFECT 223

levels, calculated for LS-coupling by Eq. (13.8) and for ^/-coupling by

Eq. (13.14).

Table 13.1.—Calculated g Factors Illustrating Pauli’s g Sum Rule for a pa

Electron Configuration

Terms J * 0 J * 1 Coupling scheme

>p. 17-1
LS
»Po.l,« 0 i i
I**?* r*C%

0 4
II II
II II

1 jj
**■»

0 a
1

When only one term occurs for a certain J value the g sum rule
slates that the g factor is the same in all coupling schemes. Observed
Zeeman patterns illustrating the g sum rule are to be found for many
< leinents. In Table 13.2 zinc, tin, and lead are given as examples.

Table 13.2.—Observed g Factors Illustrating Pauli’s g Sum Rule for a p*

Electron Configuration

Element Terms J - 0 J = 1 J — 2 Coupling scheme

g = 1.000
Zn LS
0,1,2 0 1 500 1.500

lPl 1.125
Sn* Intermediate
•Po.1.2 0 1 375 1.500

'Px 1.150
Pbt jj
0 1.350 1 500

Sp 0 1.500 1 500

* Grben, J. B., and R. A. Loring, Phya. Rev., 30, 574, 1927.

t Goudbmit, S., and E. Back, Zeita. /. Pkya., 40, 530, 1927.

The observed Zeeman patterns of the transition sSi — SPi in each

of these elements are shown in Fig. 13.6.
13.6. Paschen-Back Effect.—In the preceding sections of this
chapter we have seen how in a weak magnetic field the atom as a whole is
quantized with respect to the field direction. If the magnetic field
is continually increased there will come a time when the interaction
energy between J* and H becomes so great that the coupling between
£* and L* in the case of LS-coupling, or between j* and j\ in the case of
224 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

jj-coupling, is broken down. This will occur when the interaction energy,
given by Eq. (13.19), exceeds the L*S* interaction energy given by Eq.
(12.13) or the jfjj interaction energy given by Eq. (12.25). In other
words when the Zeeman levels of the different levels of a multiplet begin
to overlap each other, Eq. (13.19) will no longer hold. As the field

t S S P p 5 S

Fio. 13.6.—Observed Zeeman patterns for the transition 3Si — 3P1 in Zn, Sn, and Pb.

is still further increased, S* and L*, or j* and j*, will independently

become quantized with the field H. This is the Paschen-Back effect
in terms of the atom model.
The classical vector model of the Paschen-Back sta^e of the atom
is shown in Fig. 13.7. With L* and S* precessing independently around
the field direction, J* is no longer constant in magnitude and ceases
to be a quantum number. Similarly with jf and j* precessing inde-

H Strong field H

L S-Coupling jj-Coupling
Fig. 13.7.—Classical vector model for LS- and //-coupling schemes in a strong magnetic
field (the Paschen-Back effect).

pendently around H, their resultant «/* ceases to have any meaning. The
two special cases of LS- and ^'-coupling will now be treated separately.
13.7. LS-coupling and the Paschen-Back Effect.—The method used
in deriving an expression for the interaction energy when the Paschen-
Back effect sets in, for original LS-coupling, is exactly the same as that
Sec. 13.7] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 225

used for a single electron in Sec. 10.7. With L*' and S* independently ,
quantized with the field H, the quantum conditions are (1) that the
projection of L*h/2x on H is equal to MJi/2v, where Ml = 0, ±1,
±2, • • • ±L; and (2) that the projection of S*h/2ir on H is equal to
Msh/2x, where Ms = 0, ±1, ±2, • • • ±S. For singlets Ms = 0, and
for triplets Ms = +1, 0, and —1. On the classical model the Larmor
precessions of L* and S* will be given by H times the ratio between
the magnetic and mechanical moments,

“d "■-h'2'25V (13-23)

Multiplying each of these angular velocities by the corresponding

angular-momentum components along H (see footnote 3 on page 157) gives
the two interaction energies:

- »• 5Bt’s “» <L-H> - «• <>3M>

*W„ - 2H • cos (S*H) - 2H ■ mJL. (,3.25)

The sum of these two energies gives the main energy shift of each mag¬
netic level from the field-free level from which it sprang:

AW„ = (Ml + 2Ms) ■ (13.26)

Dividing by he, the energy in wave numbers becomes

-ATa = (Ml + 2Ms) ■ = (Ml + 2MS) ■ L cm*'. (13.27)

To this energy a small correction due to the interaction energy

between L* and >S* must be added. By exactly the same treatment
as that given for a single electron in Sec. 10.7 this energy is given by

-ATls = A • MlMs = r8tronK (13.28)

where A is given by Eq. (12.12). Adding this term to Eq. (13.26), the
total energy shift becomes

-AT*, = (Ml + 2MS)L + A • MsMl. (13.29)

If now Ti represents the term value of the center of gravity of a

multiple term, the term value of each magnetic level (see Fig. 12.6) will
be given by
Tm = Th - (Ml + 2Ms) • L - AMlMs. (13.30)
This splitting is shown graphically in Fig. 13.8, for zSi, 8Po,i,2, and
8Di,2,3 levels. In each case the field-free levels are shown at the left.
226 INTRODUCTION TO ATOMIC SPECTRA (Chap. XIII

Next come the weak-field Zeeman levels symmetrical about the field-
free levels followed by the Paschen-Back levels at the right. The

Fig. 13.8.—Strong-field energy levels for >5, 3P, and lD terms (Paschen-Back effect.
I/<S-coupling).

equally spaced dots in the strong-field levels represent the two inter¬
action energies due to the external field. Adding the L*S* interaction
term divides the 3Z>3 levels into three groups of five equally spaced
levels.
Table 13.3.—Computation of Magnetic Energies for LaS-coupling in Weak
and Strong Magnetic Fields

Weak field Strong field

1 strong
Term rWMk M Mg M Ms Ml 2Ms 2MS + Ml
A ■ MlMs

+2 +1 +2 +1 +1 +2 +3 +A

+1 +! +1 +1 0 +2 +2 0
3P2 +A 0 o 0 -1 +2
+1 +1 —A
~1 -2
+1 +1 +1
-2 -3
0 0 0
+1 +1
-1 -1 -1
*Pi -A 0 0
0 +1 -2 -1 -A
-1 0 0
-* I -1 “2 -2

*Po -2 A o -2 -1 -2 —3 +A

As an example of the use of Eq. (13.30) in calculating strong-field

levels, consider the triplet terms 3P0i,». Since L = 1 and S = 1 the
Sec. 13.7] ZEEMAN EFFECT, PA SC HEN-BA CK EFFECT 227

possible values of ML are 1, 0, and —1, and the possible values of Ma

are 1, 0, and —1. A tabulation of all possible combinations of these
values is given in Table 13.3.
Corresponding to each of the nine weak-field levels there is a strong-
field level with the same M. Actual calculations of the magnetic energy
for the stages between weak and strong fields have been made by Darwin.1
These calculations enable one to determine just what weak-field level
goes over to each strong-field level. A general rule for connecting weak-
and strong-field levels may be stated as follows: The sum of theprojections

Fig. 13 9.—Paschen-Back pattern of a principal-series triplet.

of the mechanical moments on H does not change. In other words, the

weak field M is equal to the strong field ML + Ms:

M = Ml 4- Ms. (13.31)

Conservation of angular momentum requires this to be true. In addition

to this rule the further restriction is necessary that no two levels with the
same M ever cross.
As an example of an observed Paschen-Back pattern, consider a
principal-series triplet ZSi — ZP0,1,2. The initial and final states for
this case are shown at the left in Fig. 13.8. The selection rules for
strong field are
A Ms = 0,
AM - l 0 for p components, (13.32)
L (± 1 for s components.
Applying these rules the pattern shown at the bottom of Fig. 13.9
is obtained. The bracket shows that, were it not for the small inter¬
action between L* and S*, the pattern would be a normal triplet. Exper¬
imental observations of such strong-field patterns are very few, owing
to the enormous fields required for producing the effect. A principal-
series triplet has been observed by Paschen and Back2 in oxygen as a

1 Darwin, K., Proc. Roy. Soc., A, 118, 264, 1928.

2 Paschen, F., and E. Back, Ann. d. Phys.} 39, 897, 1912; 40, 960, 1913. See
figures At end of each article in the Annalen.
228 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

fairly good normal triplet with the s components unresolved. There is

good evidence, however, that with suitably strong fields experiment will
check theory in almost every detail.
13.8. .//-coupling and the Paschen-Back Effect.—A calculation of
the magnetic energy for .//-coupling in a strong magnetic field is essentially
the same as the treatment given for L$-coupling in the previous section.
As seen in Fig. 13.7, j* and j* are quantized independently with the field
H. The conditions for this quantization are: (1) The projection of
j*h/2ir on H is equal to mnh/2t, where mh = ±£, + f, • • • , ±jx.
(2) The projection of j*h/2t on H is equal to raJS/i/2ir, where m,-, = ±i,
+ 3, * * * ) ±j*
The total magnetic energy will be given by the sum of three energies,
the interaction energy between (1) j* and //, (2) j* and //, and (3)
j* and j*. By Eq. (10.14) the Larmor precessions of j* around H} and
j* around Hy are given by

a>' = H' 9^ and Uj, = H-g^ (13.33)

where gi and p2 are the g factors for the two electrons, respectively.
Multiplying each of these angular velocities by the corresponding pro¬
jections of the angular momenta on H (see footnote 3 on page 157) gives
the two energies,

AW- " " c” ill“- ” H ’ »W”4' <13 !l4)

AW- - « ■ <'W'4 «“ B ■ »W”4 <13 35>

The sum of these energies is
t*h
A WH = {gw,, + • (13.36)

Dividing by he, this energy in wave numbers becomes

He
-ATh = (gimn + QWn)^-^ cm_1 = + ^m,,) - Lem"1. (13.37)

To this energy must be added the interaction energy due to the coupling
between and j*. By analogy with LS-coupling this energy is given
by [see Eq. (13.28)]

—AT7,, = A • mnmn cm-1 = r,tron* (13.38)

where A is given by Eq. (12.24). Adding this small term to Eq. (13.36),
the total magnetic energy becomes

-ATm = (jgimh + 9*mu) -'L + A- mixmit. (13.39)

Sec. 13.8] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 229

If now we let T\ represent the term value of the center of gravity of a1

multiple term (see Fig. 12.10), the term value of each magnetic levei
will be given by

Tm = Ti - (g<mh + g2mJt) • L - A • (13.40)

As an example of the calculation of strong-field energy levels, con¬

sider the terms arising from part of the electron configuration sp, where

Fio. 13.10.—Strong-field energy levels for the electron configurations sp and a-s (Paschen-
Back effect, //-coupling).

ji = i and j2 = f, and those arising from the configuration s-s, where

ji = \ and j% = A graphical representation of the resultant levels
J = 1 and 2, and / = 0 and 1, is shown in Fig. 13.10. The weak- and

Table 13.4.—Computation of Magnetic Energies for ^-coupling in Weak and

Strong Magnetic Fields. Configuration sp, ji - \ and j2 * \

Weak field Strong field

Term Tweak M Mg M mn 9imn Patrons

+2 +i +2 +i +! +1 +l +34/4
+1 +i +1 +1 +1 +! +4/4
(i 1)2 +34/4 0 0 +i -i +1 -! -4/4
-1 -! -1 +* -i +1 -i -34/4
_o
— U -t
+1 -i +1 -1 +1 -34/4
+1 +i 0 -J +1 -1 +! -4/4
(1 l)i -54/4 0 0 -1 -i -1 -1 -1 +4/4
-1 -I -* -! -1 -1 +3A /4
230 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

strong-field levels for the two terms at the left are calculated by tabulating
the various interaction energies as in Table 13.4.
It is to be noted that for each weak-field level Af, there is a strong-
field level with the same M value. Just as in the case of LS-coupling
the conservation of angular momentum requires that the sum of the
\projections of the mechanical moments on H does not change with changing
field. As an equation,
M — mn + mJt. (13.41)

Since there are in general several levels with the same M value,
the further restriction that no two levels with the same M cross each other

G+z ^ ^ ms

electron transition sp

must be added to the above rule. As an example of the behavior of

jj-coupling in a strong field, consider one of the two Paschen-Back
patterns arising from the electron transition sp —> S’S. With the set
of levels shown at the left in Fig. 13.10 as initial states, and the set at
the right as final states, the pattern in Fig. 13.11 is constructed. The
selection rules for jj-coupling in a strong field are

Amn « 0 or with subscripts (13.42)

j 0 for p components
Amh ”1+1 for s components) interchanged

These rules are in agreement with the polarization rules that the
polarization of a line is the same in all field strengths. In the pattern
it is seen that lines tending to go from p to s or from s to p components
fade out and should not be observed in a strong field. Unlike the
Paschen-Back patterns for L£-coupling the lines do not resemble a
normal triplet but possess an anomalous pattern. Were it not for the
fine structure, the pattern shown in Fig. 13.11 would be the same as
the pattern for a principal-series doublet line, — 2P§ (see Fig. 10.9).
Sxc. 13.0] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 231

13.9. Complete Paschen-Back Effect.—In the three previous sections

we have seen how, in a strong magnetic field, the coupling between
j* and j*, or between L* and S*, is broken down and each is quantized
separately with the field direction H. In calculating the energies
in such cases it was tacitly assumed that
Very Strong Field
the inner couplings between the individual
H
electron spins s* and orbits l* were not
broken down. On the classical model it ,M
S*
was assumed, for example, that the preces¬
sion cf s* and s* around S* and the pre¬
cession of Zf and 1% around L* are very
much faster than that of S* or L* around//. m _
In a very strong magnetic field the
coupling between all of the individual jK-
electron vectors may be broken down, N
regardless of the original coupling scheme
LS or jj, so that each part will quantize
separately with the field H. Under these
conditions we have what is called the
US orj^-coupling
complete Paschen-Back effect1 (see Fig.
13.12). The quantum conditions for this
Fia. 13.12.—Classical vector
state of the atom are: model for either LS- or jj-coupling
1. The projection of s%h/2ir on H in a very strong field (complete
Paschen-Back effect).
takes values mnh/2Tr, where = ±
2. The projection of s%h/2ir on H takes values m9th/2t, where
mn = ±h
3. The projection of l*h/2ir on H takes values mhh/2t where m= 0,
±1, ±2, • • • , ±1.
4. The projection of l*h/2Tr on H takes values mt h/2ir, where wij, ■» 0,
±1, ±2, • • • , ±Z*.
The complete Paschen-Back effect has never been observed for any
but one-electron systems because of the enormous fields needed to
produce it. For terms close to the series limit, however, it may be
possible in the near future to produce the effect with the fields now
available.
In a field strong enough to break down the quantization between
all of the individual vectors sf, Zf, and ZJ, the chief contribution
to the magnetic energy will arise from the independent couplings of
these vectors with H. To this energy must be added small correction
terms due to the different mutual interactions between the individual
vectors. While these latter corrections will be different for LS-coupling
than for ^'-coupling the chief contributions will be exactly the same.
1 Pauli, W., Zeits. f. Phya., 16, 155, 1923; 31, 765, 1925; see also L. Pauling and
S. Goudsmit, “Structure of Line Spectra,n p. 122.
 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

The Larmor precessions of each spin and orbit around H [see Eq.
(10.4)] are given by

<■>* * H • 2,2me n 2me 1 2 me H 2mc

(13.43)
Multiplying each of these angular velocities by the corresponding
projections of the angular momentum on H gives the energies (see footnote
3 on page 157)
A TXT Ck TT ^ ... ^ A TXT TT ® ... ^

(13.44)

The sum of these energies is

ph
ATTh = (2wi„ + 2m„ + mi, + ot,,) • II ■ (13.45''

Dividing by Ac, the energy in wave numbers becomes

Up
—ATa = (2m„ + 2m„ + mh + cm-1 (13.46)

To this energy must be added the small interaction energies [see

Eqs. (12.5)],
Fi = aisfs* cos (sfsj), Ta = azsfl* cos (s*lf),
r, = ajfl* cos (1*1*), r4 = ats^l’H cos (s*Z*)- (13.47)

Table 13.5.—Very Strong Field Energy Levels tor the Electron Configura¬
tion sp
(Complete Paschen-Back effect)

M=m.l+mn
Sec. 13.9] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 233

These average cosines are evaluated just as in Sec. 10.7 [see Eq. (10.33)]
to give

Ti = aimaimatf r2 = a2milmi„ Ts = azmgxmh, r4 = aAmatmi%.

(13.48)

If now To represents the hypothetical center of gravity of all the

terms arising from a given electron configuration, the term value of

Fia. 13.13.—Very strong field energy levels for an sp electron configuration (complete
Paschen-Back effect).

each very strong field level will be given by the addition of Eqs. (13.46)
and (13.48):

TM = To — (2mSl + 2mMS + mh + mti) • L — Ti — r2 — r3 — r4, (13.49)

where L = He/ fame2 cm-1.

As an example of the calculation of levels in a very strong magnetic
field, consider the electron configuration sp. The various terms of
Eq. (13.49) are tabulated in Table 13.5.
Diagrams of the energy levels for the sp configuration in LS- and
^'-coupling are shown in Fig. 13.13, starting with the field-free levels
on either side and ending with the very strong levels in the middle.
Except for the fine structure due to the T factors the very strong field
levels are the same for both coupling schemes. However, it is just
these differences that, as the field decreases, cause one system to go
over to LS-coupling and the other to jj-coupling. It is seen that through¬
out each diagram each level retains its total M value and further that
234 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

no two levels with the same M values cross. It should also be pointed
out that the coefficient Oi, due to the spin-spin interaction, is negative and

VP012 VP, 3Sf’P012 VP,

Fio. 13.14.—Complete Paschen-Buck effect for the transition sp —> (L/S-coupling).

large in LS-coupling and negative but small in ^-coupling. The reverse

is true for the o4 coefficient, being positive and small in LS-coupling
and positive aDd large in ^'-coupling.

Normal triplet
Fio. 13.15.—Complete Paschen-Back effect for the transition ep —♦ (//-coupling).

As an example of the spectrum lines, or pattern, to be observed

when the complete Paschen-Back effect sets in for both the initial and
final states of an atom, consider the electron transition sv —* s s. The
Bbc. 13.10j ZEEMAN EFFECT, PASCHEN-BA CK EFFECT 235

resultant patterns for all allpwed transitions in weak, strong, and vety
strong fields are shown in Figs. 13.14 and 13.15. The initial levels giving
rise to these patterns are the ones shown in Fig. 13.13. The selection
rules for very strong fields may be stated as follows:
Am* = Am* = 0
Amh =
or with subscripts
i 0 for p components/ (13.50)
Am,, = interchanged
[±1 ior s components)
Were it not for the fine structure arising from the T factors, each
very strong field pattern would be a normal triplet. This result is
arrived at with all electron configurations. It should be noted that
the two strong field patterns for ^‘-coupling, in Fig. 13.15, closely resemble
the anomalous patterns of a principal-series doublet in a weak field,
f P s
rrv 1 0 -1
msrl \ -J4
"V
t 0 “1
_ 20

Mls 1 0 -1 0 S m5= 1 0 mj2= \ \ - '/2 V2 mJ, 'i'i —

P
ml| Ms= °1:L "V VV,-v2
sing'+ripl
K m$2 ms,
Ml= 1 0 -1
= '4

i
■_
mj2
M» 2 1 0 1 Ml= 1 0 -i
M =2 1 \ M *1 0 %
M= i o']-1 0 M ■ t 0 -i _0_ \
M ■f 1 0 "m" B0l
-4
-I
M= *0] -1 j-2 -1 *P| MS
0 j -1 -<>!_-
\ -HX —KM
lP0'3P, 3P2 MS I??
m

Fig. 13.16.—Magnetic quantum numbers for an ap electron configuration. (LjS-ooupling

left and jj-coupling right.)

13.10. Breit’s Scheme for the Derivation of Spectral Terms from

Magnetic Quantum Numbers.—In the early development of complex
spectra a scheme for the calculation of spectral terms was put forward
by Breit.1 This scheme is of considerable importance in that it starts
with the very strong field quantum numbers of each valence electron,
develops successively the strong- and weak-field quantum numbers,
and ends up by giving the allowed field-free levels. Consider as an
example the electron configuration sp for which the field-free levels are
3jP0,1,2 and IPi.
In a very strong field each electron is designated by the four quantum
numbers n, l, mt, and m$. Since the total quantum number takes no
part in the types of terms we need only consider the quantum values
h = 0, mh = 0, m* = ±\, l2 = 1, = 0, ±1, and m* = ±£. If
the coupling -is to be LS, the values of m,, and mi, are written down in
1 Breit, G., Phys. Rev., 28, 334, 1926; see also Russell, H. N., Phya. Rev., 29,
782, 1927.
 236 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

a row and column as in the array in Fig. 13.16. To the left of raZl = 0
and below mlt = 1 the sum Mh = 1 is written. This process continued
gives the strong-field quantum numbers for a P term. Combining
mtl and m9t in the same way, the array shown at the top of the figure
is constructed. The values Ma — 0 and Ma = 0 and ±1 divided by
the L-shaped line are just the Ms quantum numbers for singlets and
triplets, respectively. Combining these two sets of Ms with the set of
Ml, the two arrays in the lower left of the figure are obtained. The
quantum sums divided by the dotted lines are just the weak-field quan¬
tum numbers for the 3P0, ®Pi, %P*, lP\ terms. In the case of ^‘-coupling,
shown at the right in the figure, the values of ma and mz are first combined
for each electron separately to give the strong-field quantum numbers
m,. These in turn are combined to form the array shown at the bottom
of the figure. These give the same runs of weak-field quantum numbers
and correspond to the same four levels obtained for L£-coupling.
13.11. The Pauli Exclusion Principle.—In 1925 Pauli put forward a
new principle which came to play an important role in the development
of complex spectra (see Sec. 12.4). In its simplest form the principle
may be stated as follows: No two electrons in the same atom can have all
of their quantum numbers the same. This principle is well known as the
Pauli exclusion principle. In order to assign quantum numbers to each
electron we go to very strong magnetic fields where the coupling between
the electrons is completely broken down. In such a field the four
quantum numbers for any electron are n, l, mi, and mt.
Let us now calculate the allowed spectral terms arising from two
equivalent electrons. The term equivalent refers to any two electrons
having the same n and l values. As a first example, consider two equiva¬
lent p electrons. First all possible combinations of mt and mi for a
single p electron are written down in two rows as follows:
™. = $ \ i -i -i -i
mi = 1 0 -1 1 0-1
(a) (b) (c) (d) (e) (f)
It is seen that there are six possible states, (a), (6), (c), (d), (e), and
(/), in which a single p electron may exist in an atom. With two equiva¬
lent p electrons Pauli's exclusion principle says that one of these quantum
values m, or mi must be different. We therefore obtain all possible
states for two electrons by writing down all combinations of the above
states taken two at a time, with no two alike. They are
ab
ac be
ad bd cd
ac be ce de
af bf cf df ef
Sec. 13.11] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 237

For each of these 15 combinations of very strong field quantum numbers

we add the two values of m, to obtain the strong-field values of Mg,
and the two values of mi to form Ml. This leads to the following
tabulation:

Ms « 1 1 0 0 0 1 0 0 0 0 0 0 -1 -1 -1
ML = 1 0 2 1 0-1 1 0-10-1-2 1 0-1

The largest value of Ml is two indicating part of a D term. Cor¬

responding values oi Ms show that there are strong-field quantum num¬
bers enough to form a lD term:

Ms = 0 0 0 0 0\
Ml = 2 1 0-1-2/

Taking these out we find just enough numbers left to form a 8P and a
l8:
Ms — i 1 1 0 0 0 -1 -1 -n Ms = o\w
Ml = 1 0-1 1 0-1 1 0-1/ Ml = 0/ ’

Two equivalent p electrons therefore give rise to 1S, *P, and lD

terms. Referring to Table 12.2 it is seen that two nonequivalent p
a
™l)c 2 1 0 -1 -2 _ ™Si» \ J/2
ML= Vv 3 2 10 2 Ms= ^ 0 \
1 ~1~o T-x-ife
2 I 's'I -I |-2 0d Sing.j+ri'p. m$2

^ o"|-l l-ij-3 "I

oVl 1-2 1-3 iH 2

S | P | D | F j G |ml2
Fio. 13.17.—Magnetic quantum numbers for two similar d electrons (LS-coupling).

electrons give rise to 1S, lP, lD, *S, 3P, and 3Z) terms. The Pauli exclu¬
sion principle thus excludes IP, 3Sy and 3Z>, when the two electrons have
equal n values.
The same terms are readily calculated from Breit’s scheme given
in the preceding section. Consider, for example, two d electrons in
LiS-coupling. We first write down all possible combinations of mtl
and mti in one array and those of mZl and mZa in another (see Fig. 13.17).
If the total quantum numbers are different, all possible combinations
of the magnetic quantum numbers are allowed. The runs ML = 4 to
—4, 3 to —3, 2 to —2, 1 to —1, and 0 correspond to G, F, D, P, and S
terms, respectively; and the two sets of Ms = 1, 0, —1, and 0 correspond
to triplets and singlets. These are just the terms given in Table 12.2
for d-d.
238 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

If the two d electrons are equivalent, i.e., equal values of n, certain

of these terms are forbidden. If we permit the mt values to be alike
(i.e., Ma = 1, or —1), the values of ML * 4,2,0, —2, and —4, crossed out
by the diagonal line, are forbidden. Since the values in the lower left
of the array are identical with, and are a mirror image of, those in the
upper right half, one of these groups must also be eliminated from the
calculations.1 Leaving out the lower left half of the array the remaining
values are seen to form two runs ML = 3 to —3, and 1 to —1:

Ms = 1, Ml- 3 2 1 0 -1 -2 -3

Ms = -1, Ml = 3 2 1 0 -1 -2 -3

Ms = 1, Ml = 1 0 -1
Ms = -1, Ml = 1 0 -1

If, on the other hand, we permit the ML values to be alike, the values
of M8 — 1, or —1, are forbidden. Since the two Ms = 0 values are
identical, one of these must also be excluded. The remaining com¬
binations are tabulated as follows:

Ms = 0, Ml = 4 3 2 1 0 -1 -2 -3 —4
Ms = 0, Ml = 3 2 1 0 -1 -2 -3
Ms = 0, Ml = 2 1 0 -1 -2
Ms « 0, Ml = 1 0 -1
Ms = 0, Ml - 0

The second and fourth rows go with the preceding tabulation to com¬
plete the quantum numbers for *F and 3P terms. The remaining rows
correspond to Hj, 1D, and lS terms.
Consider, as a third example, the calculation of terms arising from
two equivalent p electrons in jj-coupling. The first step is to write
down all possible combinations of m, and mi for each electron separately.
This forms the two arrays at the top of Fig. 13.18. Each of the two
runs of mn combined with each of the two runs of m3l will give all possible
combinations. Of the four resultant arrays there will be two like the
lower right one in the figure. Since the electrons are equivalent, one
of these must be entirely excluded. In the lower left-hand array, the
diagonal and either the upper right or lower left half of the array must
be excluded. The same is true for the lower center array. The remain¬
ing combinations are just sufficient to form the five terms (f |)0, (4 I)2,
(i £)o, (4 !)i> and (4 4)2. These correspond to the five terms bS0,
*Po, *Pi, and 8P2 in L£-coupling, derived above.
1 Each value of Ml in the lower left half of the array in Fig. 13.17 is identical with
a corresponding value in the upper right half. It is observed that each pair of values
arises from the same mi values. Although the mi subscripts are just interchanged in
each of the two combinations of a pair, the equivalence of the two electrons makes the
two states identical in all respects.
Sec. 13.12] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 239
13.12. Pauli’s g Permanence Rule.—In 1923 Pauli1 proposed a new
rule for the g factors of any given multiple term in weak and strong
fields. This rule states that the sum of the g factors for a given M is the
same in all field strengths. In a weak field the g factor of any state is
given by the ratio between the magnetic moment of the atom in Bohr

"Vi o -i "Vt o -i
1 m .S1 i .i 1
mj = \ 2 “2 2 J2 2 2 2 2
~r n 3 i
77}! "2l 2 "2 '2
mS2
ms,
Coupling

3 1.13
mjis2 2 2 2 _ mjfH_ - "vf ?4"i
i o l M* 2 1 0 4 T
2 X 0 |-l 4 o\\ 1 0 -1 }-2 "2i
1 o'l^K |-2 -1 mJz
-V 1 \
0 |-l |-2 |-3, -f
(fi)0 cii\ 1 mj2 (H), (41),
Fig. 13.18.—Magnetic quantum numbers for two similar p electrons 07-coupling).

magnetons, eh/4rmc, and the mechanical moment in units h/2r [see

Eqs. (10.13) and (13.6)]:

Qwmk
M£. (13.51)
P/
Here

"-sfe “d <13®>

where, as usual, pj is in ergs per gauss and pj is in erg seconds. In

these units,
fxj_ < h/2w _ m/ # 2me _ gjMj
(13.53)
Pj eh/Amnc pj e Mj

In weak or strong field the g factor is given by the ratio of the total
projection of the magnetic moments on the field direction to the total
projection of the mechanical moments.2
For L/S-coupling,
_ ML + 2MS
0*m" - Ml + Mb' (13.54)

1 Pauli, W., Zeits. /. Phys., 16, 155, 1923.

* The weak field gr-faetor of Eq. (13.53) is also expressed in terms of projections.
With pj and pj in line with each other, see Fig. 10.6, their ratio and the ratio of their
projections on any line will be the same.
240 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

For ^‘-coupling,
_ giinH + gtmh
{/strong — t (13.55)
mh + ^;S
For a given multiple term it often happens that there are several
magnetic levels with the same M value. Pauli’s g permanence rule
states that for given L, S, and M, or for given jh jt, and M the sum
of the weak-field g factors is equal to the sum of the strong-field g factors:
2^we*k ~ trons* (13.56)
Values of weak- and strong-field g factors are given in Table 13.6
for a 8P term.
Table 13.6.—Pauli’s g Permanence Rule for L£-coupling

s
Term M = 1 M = 0 M = -1

II
1
wm
•P* ! ! 1 ! 1
•Pi ! ! ! Weak field
»Po 8

•Pa 1 2 1 i
•Pi 1 2 Strong field
•Po

i 3 3 !
8
Similarly for the two upper states arising from the sp configuration
in ^-coupling we obtain the values given in Table 13.7.
Table 13.7.—Pauli’s g Permanence Rule for jj-coupling
<N

Term M 2 M - 1 M = 0 M = -1
II

=
1

(i I). l 3 3 Weak field

(i Da ! l 1 1 !

(H)i i l
Strong field
(1 i). i i i §

2* ! i (1) 1 i

One of the most important features of the g permanence rule is that

the weak-field g factors can be determined without the use of the Land6
g formula and the quantum-mechanical or classical model. For 3P2,
M = 2, in the above table patron* = •§. The weak-field g factor will
therefore be $ for all weak-field 3P2 levels. This leaves for 3Pi, M — 1
the value g = 3 — $ = $. For 3P0, M = 0, g is indeterminate in all
fields.
Sac. 13.13] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 241

Darwin has shown that the g permanence rule holds for all inter¬
mediate field strengths.1
13.13. Pauli’s g Sum Rule for All Field Strengths.—In a very strong
magnetic field the individual electron couplings are completely broken
down and each spin and orbit becomes quantized independently with
the field. Pauli showed that even under these extreme conditions
the sum of the g factors for a given M, of a given electron configuration is
constant.2 Starting with LS- or ^‘-coupling the g factor for any term
in a very strong magnetic field is given by the sum of the projections on
H of the magnetic moments in Bohr magnetons divided by the sum of the
projections of the mechanical moments in units h/2ir.
/2 m9l + 2 m,t + mh + mt\
(13.57)
y m81 + mtz + mh + mi% )

Consider, for example, the terms arising from the configuration

sp in L£-coupling. The g factors for weak, strong, and very strong
fields are given in Table 13.8.
Table 13.8.—Pauli’s g Sum Rule for the Electron Configuration sp in LS-
coupling

Consider again the terms arising from the configuration sp in ^-cou¬

pling. The g factors for weak, strong, and very strong fields are given
in Table 13.9.
In general the weak-field g factors cannot be calculated from the
very strong field g factors. The reason for this is that the coupling
scheme is not known. When the coupling scheme is given, the strong-
1 Darwin, K., Proc. Roy. Soc., A, 118, 264, 1928.
* For a discussion of the p Bum rule in weak magnetic fields, see Sec. 13.5.
242 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

field g factors can be determined from the g permanence rule and hence
the weak-field g factors can be evaluated (see the previous section).
13.14. Land€’s r Permanence Rule.—Analogous to the g permanence
rule we have what is known as the r permanence rule discovered by
LandA The rule may be stated as follows: For a given multiple term,
i.e.y given S and L, or given ji and the sum of all the T factors for terms
with the same magnetic quantum number M is a constant independent
of the field strength.
Table 13.9.—Pauli’s g Sum Rule por the Electron Configuration sp in
./[/-coupling

For LjS-coupling in field-free space, or in a weak magnetic field, the

T factor of a term is given by Eq. (12.13):
Twmk « r, + r4 - AL*S* cos (.L*S*) = U(J*2 - L*2 - S*2), (13.58)
where A is a coupling constant. In other words, a T factor, as we have
already seen in Sec. 12.6 and 12.7, is just the interaction energy of any
two quantum vectors. This particular Eq. (13.58) gives the Land6
interval rule for the fine structure (see Sec. 12.8). Now, in a strong
field, the Paschen-Back effect sets in, and L* and S* become inde¬
pendently quantized with the field H. In this case the average value
of the cosine in the above equation had to be evaluated to obtain the
interaction energy and the strong-field T factors. This average inter¬
action energy [see Sec. 13.6, Eq. (13.28), and Sec. 10.7, Eq. (10.33)1
is given by
r.tron, * r, + r4 - a-l*s* cos (L*s*) = a * mlms.. (13.59)
Consider now the T permanence rule applied to a ZP term. Since S = 1.
L « 1, and J * 0, 1, and 2, the following T factors are determined.
Sac. 13.14] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 243

In going from weak fields to strong fields the r factors always change/
if at all, so that ST * constant for each M. It is to be noted that the
weak-field T's can be determined from the strong-field T’s. The Land6
interval rule is thus obtained without the use of the quantum-mechanical
or classical model of the atom [Eq. (13.58)].

Table 13.10.—r Permanence Rule for a 3P Term (LS-coupling)

o
SI

SI
M = —1 M = -2
(M

Term M « 1

II
II

3P2 A A A A A
Weak field
3Pi -A -A -A
T factors
3Po -2 A

3P2 A 0 -A 0 A
Strong field
•Px 0 0 0
T factors
3Po —A

xr = A 0 -2 A 0 A

For ^'-coupling in field-free space or in a weak magnetic field the

T factor of a term is given by Eq. (12.25):

Tweak - T1 + r2 « Aj*j* cos (ifif) = -if2 - if2). (13.60)

In a strong field, the Paschen-Back effect sets in and if and if become

independently quantized with the field H. Here the average value
of the above given cosine term must be evaluated to obtain the inter¬
action energies and the strong-field T factors. This average interaction
energy, by analogy with L/S-coupling, is given by

Tatronx = A * fij* COS (jlji) = A * (13.61)

Consider, as an example, the two upper terms of the electron con¬

figuration sp in i/-coupling. Since ji = \} j2 = $, and J — 1, and 2,
the following weak- and strong-field T factors are obtained:

Table 13.11.—T Permanence Rule for the Two Terms (£ |)i and ($ §)*, Arising
from the Configuration sp (if-couPLiNG)

Term M = 2 M = 1 M = 0 M « -1 M = —2

-5A/4 -BA/4 -BA/4 Weak field

3A/4 BA/4 3A/4 BA/4 BA/4 T factors

MB -A/4
Tt< rfi

-BA/4 Strong field

CO

BA/4. -A/4 A/4 BA/4 I* factors

1

XT m BA/4 -A/2 -A/2 -A/2 BA/4

244 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII

13.16. Goudsmit’s r Sum Rule.—Analogous with the g sum rule for

magnetic levels in weak and strong fields is the r sum rule of Goudsmit.1
This rule may be stated as follows: For a given electron configuration
the sum of the T values corresponding to a given value of M is independent
of the field strength H. In LS- or ^-coupling there are four interactions
to be taken into account: Ti, r2, r3, and r4 (see Sec. 12.7). In a very
strong magnetic field, where each spin and each orbit l* are inde¬
pendently quantized with the field, these interaction energies are given
by Eq. (13.48).

Table 13.12.—r Sum Rule for sp Electron Configuration (ZmS-couplino)

Term M - 2 M = 1 M = 0 M = —1 M - -2

lPx 0 0 0
5P* a/2 o/2 a/2 a/2 a/2 Weak field
*Pi -o/2 -a/2 -a/2 T factors
*Po —a

*Pi 0
aP2 a/2 -a/2 a/2 Strong field
aPi | 0 r factors
aPo -a/2

lPx 0 -a/2 a/2

aP* a/2 a/2 0 -a/2 a/2 Very strong field
aPi -a/2 0 T factors
*Po -a/2

2r4 a/2 0 —a 0 a/2

In LS-coupling Ti gives the singlet-triplet intervals and is due prin¬

cipally to Heisenberg's resonance phenomena, r2 gives the separations
between multiple levels, and r3 and r4 give the Land6 interval rule for
the fine-structure separations (see Fig. 12.6). Goudsmit has applied
the T sum rule to the r3 and r4 interactions and has shown that they
are in general agreement with observation. Consider for example
the sp configuration in weak, strong, and very strong fields. The
T factors of the allowed terms in LS-coupling are computed from Eqs.
(12.9), (13.28), and (13.48), respectively, and are given in Table 13.12.
In this particular example, where h = 0, r3 = 0. In configurations
where both Vs are not zero, the T sum rule will apply to both r's separately
or collectively.
A similar table formed for the same configuration in jj-coupling
will lead to the same T sums shown here for L&-coupling. It should

1 Goudsmit, S., Phys. Rev., 31, 946, 1928; see also Paulino, L., and 8. Goudsmit,
“Structure of Line Spectra.”
Sue. 13.16] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 246

be pointed out that the r sum rule may also be applied and shown to'
hold for Ti and IV
13.16. The r Permanence and r Sum Rules Applied to Two Equiva¬
lent Electrons.—Consider the permanence and sum rules as they apply to
the case of two equivalent d electrons, i.e., to two d electrons having
the same total quantum number n. In this case the Pauli exclusion
principle is in operation and the allowed terms (see Sec. 13.11) are
X/S, 8P, 1DJ 8P, and Hr. In a very strong field there are 10 possible states
for each d electron:

m. = i i i \ i
mi — 2 1 0 —1 —2 2 } o -I -2 («•«»
Following the first scheme developed in Sec. 13.11 for the calculation
of terms arising from two equivalent electrons we combine these columns
two at a time, taking no two alike. The resultant combinations should
be tabulated by the reader as follows: (1) Make four columns of all
possible combinations of, and giving each value of, mtl, mtt, milf and
(2) Using the formulas for r3 and r4, Eq. (12.9), multiply each m,
by its own mt times a to form the fifth and sixth columns under r3 and r4.
(3) Since the a coefficients are identical in the two electrons, add each
r8 and r4 to form the seventh column T^ry .tron«. (4) Add the first
four columns to form the last and eighth column of total magnetic
quantum numbers M = mtl + + mh + m^. We now collect all
values of Tvery „tron, and tabulate them under corresponding values of M
as shown at the top of Table 13.13.
The strong-field r sums may now be used to calculate the weak-field
T factors, provided the coupling of the two electrons is specified. Sup¬
pose first that the coupling is LS. In LS-coupling the T factors for
singlet terms are always zero. We therefore write zero in all columns
for the 1S, 1Z>, and lG terms, as shown. Since 3P4 is the only remaining
term with M = 4, and the r sum for if = 4 is 3a/2, we write T = 3a/2
for all 8P4, M’s. This leaves —a/2 for 3P3, M = 3, which is next written in
all 8P3, M’s. With — 3a/2 for 3P4, the r permanence rule, which also
gives the Land6 interval rule, gives T = —a/2 and —2a for SP3 and *F2)
respectively. This in turn leaves only a/2 for 8P2, then —a/2 for 8Pi
and —a for 8P0.
If the coupling of the two equivalent d electrons is ^-coupling, the
individual j’s and the field-free terms are first determined by the scheme
given in Sec. 13.11. The resultant values are given in the lower left
hand column of Table 13.13. For each d electron, separately, ji = |
and |, for which the T factors are a and —3a/2, respectively. The
T factors for both electrons, taken together, will be just the sum of the two
individual T's. These inserted in the lower third of Table 13.13 agree
with Goudsmit’s T sum rule.
246 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIII
Tablb 13.13.—r Sum Rule for Two Equivalent d Electrons in Weak and in
Very Strong Magnetic Fields

M* ilf- M* M= M— M-
Term
4 3 2 1 0 -1 -2 -3 -4

3a J2 a —3a/2 3a/2 2a 3a/2 —3a/2 a 3a/2

0 a/2 a —3a/2 a —3a/2 a a/2 0
-a/2 0 —a 0 —a 0 —a/2
—a a/2 —a a/2 —a Very strong
a/2 —a/2 -2a -a/2 a/2 field
a/2 0 -a/2 0 a/2 T factors
0 -a/2 0
-a/2
-a/2

zr 3a/2 a -a/2 —a -2a -a/2 a 3a/2

K?4 0
3o/2 3a f 2 3a/2 3a/2 3a/2 3a/2 3a/2 3a/2 3a/2
—a/2 -a/2 -a/2 -a/2 -a/2 -a/2 —a/2
'Ft -2a -2a -2a -2a -2a Weak field
'Dt 0 0 0 0 0 T factors
'P* a/2 a/2 a/2 a/2 a/2 L$-coupling
*Pl -a/2 -a/2 -a/2
*Po —a
'So 0

III
ill
-a/2 -a/2
-a/2
-a/2
-a/2
— a/2
-a/2
-a/2
-a/2
—a/2
-a/2
-a/2 -a/2 -a/2
-a/2 -a/2
mim -a/2 -a/2
-a/2
-a/2
—a/2
-a/2
-a/2
-a/2
Weak field
(iDt
(H)« 2a 2a 2a 2a 2a 2a 2a 2a 2a T factors
(II). 2a 2a 2a 2a 2a //-coupling
(I f). 2a
(1 !). —3a -3a -3a -3a -3a
(ID. -3a

Table 13.14.—Triplet Separations for the Electron Configuration 3d2,

Showing Agreement with the Land£ Interval Rule and the Goudsmit
r Sum Rule

Interval Interval
Element *^4, — *Fj — 8F j *p. - «Pi - *p. a cm"1
ratio ratio

Sc II 104 81 4:3.1 63 27 2:1.0 52, 54, 58, 54

Ti III 288 184 4:3.1 118 67 2:1.1 119, 128, 118, 184
V IV 412 818 4:3.1 216 117 2:1.1 206, 212, 216, 284
Cr V 687 500 4:3.1 368 208 2:1.1 819, 888, 868, A16
Sic. 13.16] ZEEMAN EFFECT, PASCHEN-BACK EFFECT 247

Suitable comparisons of the calculated F factors with observed1

spectral terms are to be found in Sc II, Ti III, V, IV, and Cr V. The
observed intervals and calculated a’s are given in Table 13.14.
That the coupling is Russell-Saunders is shown by the interval
ratios given in the third and fifth columns of Table 13.14. That the T sum
rule holds is shown by the a coefficients in the last column. Perfect
agreement would find each of the a coefficients the same for each of
the four intervals. The same a coefficients calculated from the con¬
figuration 3d4s in the same elements are 49, 144, 240, and 365 cm-1>
respectively, in good agreement with those tabulated.

Problems

1. Calculate the relative separations and intensities for the Zeeman pattern arising
from the transition ZFA — *GA.
2. Calculate the Zeeman patterns for the transition 3Pi — 3D2 (sp-sd) in LS- and
in jy-coupling.
3. Show that Pauli's g sum rule holds for the terms arising from a pp configura¬
tion in LS- and in ^/-coupling.
4. Find the relative intensities of the lines comprising the Paschen-Back pattern
at the bottom of Fig. 13.9. See Fig. 13.8 for the levels, and use the sum rules.
6. Compute the terms arising from two similar / electrons.
6. Show that the T sum rule holds in weak, strong, and very strong fields for a pd
electron configuration (a) in L£-coupling and (6) in ^-coupling.
 CHAPTER XIV

COMPLEX SPECTRA

Prior to 1922 only singlet, doublet, and triplet series of spectrum

lines were recognized by experimental and theoretical physicists. It
was well known that these series had been found only in the groups of
elements at both sides or edges of the periodic table, and that the spectra
of the elements in the center of the table were very complex. The
titanium spectrum, or the iron spectrum, for example, was known to
contain hundreds of lines distributed apparently at random throughout
the visible, the near ultra-violet, and the near infra-red spectrum.
As early as 1901, however, a fairly complete analysis of the complex
spectrum of rhodium had been made by Snyder.1 In this spectrum
some 470 lines were grouped together in an array of equal frequency
differences tying together what is now known as the main energy level
structure of rhodium. In addition Snyder had been able to link together
several hundred lines in ruthenium, iridium, nickel, platinum, and
osmium. The significance of Snyder’s work was not realized until
some 25 years later.
In 1922 the complex spectrum of manganese was analyzed by Cata¬
lan,2 and a part of the chromium spectrum by Gieseler.8 The manganese
spectrum was shown to contain a number of triplet series, some members
of which formed groups of lines more complex than had ever been assigned
to triplets. In the hands of Sommerfeld these complex groups of lines
were shown to arise from multiple energy levels and were not to be
associated with triplets. This marked the beginning of complex-
spectrum analysis.
The chief characteristics of complex spectra may be briefly described
in terms of the energy levels. In the place of series of singlet, doublet,
and triplet levels, there are, in general, multiple levels composed of
four, five, six, seven, or eight regularly spaced levels. Such multiple
levels are frequently called quartet, quintet, sextet, septet, and octet levels.
It is found experimentally that the multiplicities of the levels belonging
to a given spectrum will either be all even or all odd. In the cobalt arc
spectrum, for example, the levels are doublets, quartets, and sextets,
whereas in the arc spectrum of iron they are triplets, quintets, and septets.

1 Snyder, C. P., Astrophys. Jour., 14, 179, 1901.

* Catalan, M. A., Phil. Trans. Roy. Soc., A, 223, 127, 1922.
* Giuselbr, H., Ann. d. Phys., 69, 147, 1922.
248
Sec. 14.2] COMPLEX SPECTRA 249
Before going further it should be pointed out that just as S terms are
always single, P terms never reveal more than three fine-structure levels,
D terms never more than five, F terms never more than seven, etc. This
anomaly, which is often confusing to the beginner, really simplifies
complex-spectrum analysis. This will become apparent in the following
treatment of complex spectra and the vector model. We shall find, for
example, that a 4P, a 6P, a 6P, etc., term will each have only three fine-
structure levels. That they really belong to quartet, quintet, sextet,
etc., systems is shown when the atom is placed in a weak magnetic
field and the Zeeman effect is observed. It is little wonder therefore
that the principal series discovered by Catalan in manganese were first
called triplets in place of sextets and octets as they are now shown to be.
14.1. The Displacement Law.—In 1919 Kossel and Sommerfeld1
promulgated the displacement law for the spectra of the neutral and
singly ionized atoms. This law states that the spectrum and energy
levels of any neutral atom of atomic number Z closely resemble the spectrum
and energy levels of the ionized atom Z + 1 succeeding it in the periodic
table. In the preceding chapters, for example, we have seen how the
doublet series of the alkali metals Li, Na, K, Rb, and Cs resemble the
doublet series of the ionized alkaline earths Be, Mg, Ca, Sr, and Ba
(see Figs. 5.2 and 6.5). This law will, in the following sections, be seen
to hold in complex spectra.
14.2. Alternation Law of Multiplicities.—Along with the displacement
law just discussed another important law has developed known as
the alternation law of multiplicities. In short this law may be stated
as follows: Spectral terms arising from successive elements in the periodic
table alternate between even and odd multiplicities. We shall give, for
example, the alternation of multiplicities as they are found in the elements
of the first long period (Table 14.1).
Table 14.1.—Alternation Law of Multiplicities for the Elements in the
First Long Period
KI Cal Sc I Til VI Cr I Mn I Fe I Col Nil Cul Znl Ga I Ge I
Sing. Sing. (Sing.) (Sing.) Sing. Sing. Sing.
Doub. Doub. Doub. (Doub.) Doub. Doub. Doub.
Trip. Trip. Trip. Trip. Trip. Trip. Trip.
Quar. Quar. Quar. Quar. Quar.
Quin. Quin. Quin. Quin.
Sex. Sex. Sex.
Sept. Sept.
Oct.

All but singlets in Cr and Fe, and doublets in Mn, have been observed.
Photographs of multiplets illustrating term multiplicities for these
elements are given in Fig. 14.1. The multiplicity and term designation

1 Kossel, W., and A. Sommerfeld, Verh. d. Devisch. Phys. Ges., 21, 240, 1919.
250 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

of each multiplet are given above each photograph. Wave-lengths

of the strongest lines in each multiplet, bracketed by the heavy lines,

*$ -*p POTASSIUM
ALTERNATION LAW
MULTIPLICITIES

Fig. 14.1.—Illustrating the alternation law of multiplicities for the elements in the
first long period. Photographs taken by the author with a 15-ft. Rowland grating and
mounting.

are given below each picture. The centers of the potassium-doublet,

the calcium-singlet, and the copper-doublet lines show self-reversal
due to absorption. It may be noted that a triplet multiplet may be
Sbc. 14.3] COMPLEX SPECTRA 251

made up of as many as seven lines whereas an octet multiplet may

contain as few as three. The faint lines symmetrically located on each
side of each of the zinc triplet lines are grating ghosts. Since both the
alternation and displacement laws are valid, a combination of the two
shows that a table similar to the one above may be formed from the
spectra of the same atoms singly ionized.
Table 14.2.—Alternation Law of Multiplicities for the Singly Ionized
Elements in the First Long Period
Call Sell Till VII Cr II Mali Fell Coll Ni II Cull Zn II Ga II Ge II As II
Sing. (Sing.) (Sing.) (Sing.) Sing. Sing. Sing.
Doub. Doub. Doub. (Doub.) Doub. Doub. Doub.
Trip. Trip. (Trip.) Trip. Trip. Trip. Trip.
Quar. Quar. Quar. Quar. Quar.
Quin. Quin. Quin. (Quin.)
Sex. Sex. (Sex.)
Sept. (Sept.)
(Oct.)
It is to be noted that the tables are approximately the same as far as
observed multiplicities are concerned, and that the elements have been
shifted one place to the left.
A table for doubly ionized atoms may be formed in exactly the same
way by shifting the elements one more column to the left. Observations
show that this displacement of multiplicities is valid as far as the eighth
and ninth stages of ionization, and there is no reason to doubt that
it will be valid until the atoms are completely stripped of all extranuclear
electrons.
14.3. The Vector Model for Three or More Valence Electrons.—It is
an easy step from the vector model for two electrons to the model for
three or more electrons. This step is made with the aid of the branching
rule (see Sec. 12.13). We shall start with an atom stripped of all but
one valence electron and return electrons to it, one by one, until the
atom is neutralized. Let the individual quantum numbers of the
electrons be sl} s2, s3, . . . , lh Z2, h, . . . , ji, j2, ja, ... If we
assume L<S-coupling among the electrons we begin by successively adding
the spins to form a resultant S. With each st = ?,

Sl
+
sa

+
S*

+
*4
252 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

In general the addition of an electron to any multiple term raises

and lowers the multiplicity by one. This in terms of the vector model
is shown in Fig. 14.2a, for the case where s = £ is added to a quartet

Ss-Coupling LI-Coupling LS-Coupling

SP'i,s-£,$-1,2 Lp-2,1-1,L-1.2.3
Fig. 14.2.—Vector diagrams of the addition of one electron to the parent term of an
ionized atom.

term with S = | to give triplets, (S = 1), and quintets, (S = 2). SP

represents a parent S to which a single electron spin s is added.1
Adding Vs in the same fashion, take h = 0, U = 1, U = 1, and U — 2:

The two p electrons are assumed to have different total quantum

numbers. Similar electrons will be treated in the next chapter. The
addition of Vs according to the vector model is shown in Fig. 14.2b, for
the case where a p electron, (l = 1), is added to a D term, (L = 2), to
form the resultant terms P, D, and F with L = 1, 2, and 3, respectively.
If we now add spins and orbits simultaneously the branching rule
gives the terms shown at the top of page 253.
Combining any resultant spin S with a resultant L will, just as in the
case of two electrons (see Sec. 12.3), give the resultant J values. Briefly
the J values of a given multiple term are just given by the sum and
1 The small letter s, used for the electron spin, and the capital letter S} used for
spin resultant, should not be confused with 8 for an electron with l ■* 0, and S for a
term with L * 0, respectively.
Sac. 14.3] COMPLEX SPECTRA 253

p *P *P

p *S *P !D *5 SP 2jD 4S 4P 4Z>

difference of L and S and all intervening values differing by unity.

Combining S = \ with L = 0, 1, 2, and 3, J values corresponding
to 4/Sj, 4Pj,i,i, and terms are obtained. Similarly
combinations of S = 3 with L = 0, 1, 2, and 3, lead to */S8, 1Pz.zAt
7£i.2.3.4.6, and 7F0, i,2, s, 4,6, «• & terms are never more than single, P
terms triple, D terms quintuple, etc. A term is called a triplet, quartet,
quintet, etc., because its spin resultant S = -f, * • ’ > regardless
of the number of fine-structure levels it may contain. It is only when
L ^ S that the term will contain the maximum number of fine-structure
levels 2S + 1, from which the whole system of terms with the same 8
derive their name quartett quintet, sextet, etc. This maximum multiplicity
for each given 8 is called the permanent multiplicity.
In deriving the spectral terms arising from //-coupling we need
consider only the addition of one electron to a parent system already
made up of one or more valence electrons. The parent system of terms
represents the energy levels of an ionized atom, to which we add one
electron with //-coupling to form energy levels for the neutral atom or
an atom in one lower stage of ionization. Consider as parent terms, for
example, those arising from the configuration sp. They are / = 1, 0, 1,
and 2. Adding now a p electron We obtain the following:

*P'P ii i ti i * M Hi 14 nil
These correspond exactly with the 18 terms obtained for the same
resultant electron configuration in L/S-coupling: 2£j, 2P*j, 2D|j, 2S^,
*M. 'DU> M..» 4*>M.I.I-
The processing vector models for LS- and //-coupling for complex
atomic systems are represented in Fig. 14.3. Just as for two-electron
systems, L% and I*, and and $*, precess rapidly around their respective
resultants L* and S* which in turn precess more slowly around their
resultant J* In //-coupling the spin and orbit of the added electron
are coupled together and precess rapidly around.their resultant j*
which in turn precesses more slowly with JP of the parent term around
J*. It should be pointed out that in //-coupling the parent term
254 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

JP may be formed by two or more electrons in LS-, jj-, or any other

coupling scheme.
14.4. Terms Arising from Three or More Equivalent Electrons.—The
terms arising from three or more equivalent electrons cannot be calculated

s
LS-Coupling Jj-Coupling
Fig. 14.3.—Processing vector models for complex atomic systems.

by the method outlined above. Here the Pauli exclusion principle must
be taken into account and the strong-field or very strong field quantum
numbers resorted to (see Sec. 13.11). Although this necessitates the
postponement of the calculation of terms until the Zeeman and Paschen-
Back effects have been treated in the next chapter, the results may be
given here.

Table 14.3.—Spectral Terms Arising from Equivalent p and d Electrons

V\ aP
p2, 'S 1D
p3, 2P 2D AS
pA, >£ lD 3P
p*, 2P
p®, 'S
dl, *D
d\ 'SlD*G 3P3F
*D s I APAF
dA, WZW *P3F WD'FKPI 'P*D'F*G'H
d*, 2£) J PS*D*F*G*I4PAF *AD*0 *S
d\ 'S'D'G *P*F WWW *P*D*F*G*H *D
<P, *D I APAF
W2W *P*F
<*,
d*

It is important that we note here the lSo term arising from six equiva¬
lent p electrons or from ten equivalent d electrons. Omitting these
configurations in the tables, the upper half of each table is identical
with the lower half.
Sec. 14.5] COMPLEX SPECTRA 255
14.5. The Land! Interval Rule.—The Landd interval rule, as already
shown in Sec. 12.8, applies to the special case of Russell-Saunders coupling
and gives the interval rule for the fine structure of each multiple level.

Fig. 14.4a.—Graphical representation of the Land6 interval rule for 4P, 4D, and 4F terms.

If, for example, the spins of three or more electrons are coupled together
to form a resultant S and the Vs coupled together to form a resultant L,
the derivation of the interval rule becomes identical with the derivation
cor two electrons. The interval rule is
written, therefore,
F = AL*S* cos (L*S*) =*
±A(J*2 -L*2 - S*2), (14.1)
where A is a constant for a given
multiple term and is given by 6A

A = asof3 + a 40(4. (14-2) 5

A comparison of Fig. 14.3 with Fig.

12.5 shows that a8 and cu are the
% A
/
4
coupling coefficients for the parent term
and the added electron, respectively, 5p3 -3A

and a8 and a4 are given by Eqs. (12.10) 3

and (12.11), where s* is replaced by S*} &p -6A
(XI

and Z* by LJ. In other words, A is a

CD
<
1
1
1

measure of the sum of the energies of 5fri 1

interaction between each electron spin Fig. 14.46.—Graphical representation
and its own orbit. When the constant ^he Land6 interval rale ,or a ‘F
A is expressed in wave numbers, Eq.
(14.1) gives the wave-number shift of each fine-structure level J from
the hypothetical center of gravity of the multiple level. Consider, for
example, a *F term, where S « f, L = 3, and J = ■}, |, -J, and f. Equa¬
tion (14.1) gives T = —6A, —|A, 0, +$A. A graphical representation
of these four levels is shown at the right in Fig. 14.4a. Since A, L*, and
S* are fixed in magnitude, the proiection of S* on L* is proportiona.
256 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

to cos (L*S*) and hence proportional to the interaction energy I\ The

differences have the ratio 5:7:9. By Eq. (14.1) the difference between
one level J and the next level J + 1 is given by

Ar - A(J + 1), (14.3)

i.e., by the larger of the two J*s. From Fig. 14.4a it may be seen that
the interval ratios for 4P and *D terms are 3:5 and 3:5:7, respectively.
For a hF term, shown in Fig. 14.46 with J = 1, 2, 3, 4, and 5, the interval
ratios are 2:3:4:5.
Of the many examples revealing good LS-coupling and the Land6
interval rule in complex spectra, only a few will be given as illustrations
(see Table 14.4).

Table 14.4.—Observed Fine-structure Intervals Illustrating the Land£

Interval Rule for L5-coupling

Observed intervals Observed Theoretical

Element Terms
in centimeters ratios ratios

Sc I

Ti I 58.00, 86.23, 113.89, 140.13 2.07, 3.08, 4.06 2:3:4:5

5.00

V I 'Fi.i.M.l.V 40.88, 68.57, 94.10, 118.59, 3.17, 5.31, 7.29, 3:5:7:9:11

141.98 9.19, 11.00

Fe I 7F o.l, 2,1,4,6.0 -25.55, -52.34, -81.56, 1.00, 2.05, 3.19, 1:2:3:4:5:6

-114.26, -150.81, -195.44 4.47, 5.90, 7.65

Mn I ,pu.i 129.18, 173.71 7.00, 9.41 7:9

14.6. Inverted Terms.—In some elements it is observed that nearly all

of the terms have inverted fine structure, whereas in others they are
nearly all normal. A normal term is defined as one in which the fine-
structure level with the smallest J value lies deepest on the energy
level diagram, and an inverted term is one in which the largest J value
lies deepest. The 7F term given for iron in Table 14.4 is a good example
of an inverted term. The others given there are normal.
A correlation of spectral terms as either normal or inverted shows
that (1) normal terms arise from electron configurations involving less
than half an incompleted subshell of electrons, and (2) inverted terms
arise from configurations involving more than half an incompleted
subshell of electrons. Since normal terms have already been treated
in the previous section and shown to follow the interval rule, only
inverted terms need be treated here.
Sue. 14.6] COMPLEX SPECTRA 257
First a vector-model treatment of one of the causes of inverted
terms will be given. Consider for this example the two multiplet
terms and 6P3,2,i as they arise from the addition of a p electron
to the ion term d5,6S*. This addition is shown graphically by vector
diagrams in Fig. 14.5. The spin 8* of the p electron is added to the
spin resultant S% of the parent term 6S to form the S* of a septet or a
quintet. With no orbit resultant L* of the parent term, the orbit l*
of the p electron becomes the resultant L*, i.e., L* = Z*. In combining
S* with L*f to form resultants J*, S* is held fixed and L* shown in the
allowed positions where J* (not shown) gives J = 2, 3, 4, and J = 3, 2,1.
Now the spread of both of these multiplets depends upon two coup¬
lings: first, the spin-orbit coupling of the p electron which is given by

r' = a's*L* cos (s*L*) (14.4)

and, second, the coupling between and the orbit L*, which is given by

T" = a"S*L* cos (S*L*). (14.4a)

Applying the cosine law for triangles,

_ ,S^_+j*2 - S*2
r = a‘ (J*2 - L*2 - S*2), (14.5)
4 S*2
„ _ _,„S” + sp - s**
r" = a - L*2 ~ S*2). (14.5a)

Adding these,
^r - i +, rr,„ -_ ^
r /-S*2 +igss
•** - S*2 ,S*2 + S*2 - s*2
+ o'
4 S*2 ].
CJ*2 - L*2 - S*2). (14.6)

Since for a given multiplet the values in the bracket are constant,
the fine structure will follow the Land6 interval rule. Substituting the
values of S, s, SP, L, and J in this formula, the following values are
obtained:
t4, r = +K + V«"; 6^3, r = -|a' + V-a";
Ts, r = -ia' - ia"; 6P2, r = +ia' - |al#;
7P2, r = -4a' - Va"; 5Pi, r = +«af -

Since a! and a" are magnetic in character, and therefore positive, the
7P term can only be normal. If, however, a' is greater than 7a", the
6P term will be inverted. This may be seen from the diagrams of
Fig. 14.5, for in the 5P diagram the component of s* along S* adds to
S* and in 5P it subtracts. If, therefore, the s*L* interaction is sufficiently
strong, the energy will be greatest when the component of 8* is most
nearly parallel to L*. This state corresponds to 7P4 on the one hand
and 6Pi on the other. A good example of this is observed in the chromium
258 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

spectrum where the two terms arising from 3d54p follow the Land6
interval rule and have the separations 112.5 and 81.4 cm-1 for 7P2.3.4
and —5.7 and —8.8 cm"1 for 6P3,2,i for which the coupling coefficients
a' = 104.2 cm"1 and a" = 12.4 cm"1.
The most common cause for the inversion of spectral terms is to
be attributed to more than half-filled, but incomplete, subshells of
electrons. Although the derivation of spectral terms arising from
more than two equivalent electrons will not be given until the next

Fig. 14.5.—Vector diagrams illustrating the addition of a p electron to a parent term

db'aS$ to form a normal 7P and an inverted bP term.

chapter, the question of inversion can be treated here. To treat the

interaction energies of more than a half shell of equivalent electrons
we resort to Goudsmit’s T sum rule (see Sec. 13.15). Consider, for
example, the electron configuration d8, which gives rise (see Table 14.3)
to the same types of terms as d2, viz., lS, IP, lD, 3F, and 1G. To show
that the 3P and 3F terms from the former are inverted, the procedure
described in Sec. 13.5 may be employed. The 10 possible states for
one d electron are first written down as in Eq. (13.02). We next com¬
bine these states taking eight at a time with no two alike. The resultant
45 combinations should be tabulated as there described. Each spin
w,, is multiplied by its own mit to obtain individual iYs. These are
summed over the eight electrons to obtain r for each of the 45 combina¬
tions and then tabulated under the corresponding values of M as shown
in Table 13.13.
T = 2I\ and M = 2ma. +
VyEen this table is completed the r sums will be:
M = 4 3 2 1 0 -1 -2 -3 -4
3a , a . , 0 , , a 3a
ST = ——a ~^~2 “t"2a +a —a —g-

Following the scheme already described in Sec. 13.15 these give for
the two triplet terms arising from d8,
Sec. 14.8] COMPLEX SPECTRA 259

•**, r - -y; »P„ r -

JF8, r = +|; ’Pi, r = +^;
’P2, r = +2o; ’Po, r = + a.

Where observed these terms are found to be inverted.

Finally we shall mention that large group of inverted terms that
arise from the branching rule. If to an inverted parent term of an
ion we add another arbitrary electron, the new terms will in general
be inverted. The reason for this is readily seen when we remember
that in LS-coupling the l* of the added electron couples with the parent
L% and the spin $* couples with the parent $*. If the spin-orbit inter¬
action of the added electron is small compared with the S*L* interaction,
all of the resultant terms will be inverted. If the former is large, some
of the terms may be normal and the others inverted. Because of the
many possibilities that come under this heading further discussion
will be left until specific examples come up in the following sections.
14.7. Hund’s Rule.—Hund’s rule, given for two electrons in Sec.
12.7, may be restated here to apply to complex spectra in general. Out
of all the terms with the same L value arising from a given electron con¬
figuration: (1) the term with the highest multiplicity, i.e., highest S value,
will in general lie deepest; and (2) of these the term with the highest L value
will lie deepest.1 Experimentally this rule is observed to hold for many
electron configurations, and especially for configurations involving the
normal state, or the so-called ground state, of many atoms. Very good
examples of Hund’s rule will be encountered in the following sections
of this chapter in the discussion of the energy levels arising from the
elements nitrogen, carbon, oxygen, scandium, titanium, manganese, etc.
It should be pointed out that the deepest lying term i.e., the normal state,
in all atoms in all observed stages of ionization is the one given by Hund's
rule.
14.8. The Nitrogen Atom.—As examples of three-valence-electron
systems, let us first consider the elements in Group \B of the periodic
table (see Table 5.1). According to the Bohr-Stoner scheme of the
building up of the elements the normal states of these elements are given
by the configurations*

N, Is2 2s2 2p3

P, Is2 2s2 2p6 8s2 3p3
As, Is2 2s2 2p6 3s2 3p6 3d10 4s2 4p8
Sb, Is2 2s2 2p6 3s2 3p6 3d10 4s2 4p6 4d10 5s2 5p*
Bi, Is2 2s2 2p6 3s2 3p6 3d10 4s2 4p« 4d10 4/14 5s2 5p6 5d10 6s2 6p*

1See Hu^d, F., “Linienspektren.” p. 124, 1927.

 260 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

The last three equivalent p electrons (in heavy type) represent the
three valence electrons of each atom and are responsible for the optical
spectra. As a typical example of these spectra we shall consider nitrogen
in some detail.

The electrons 2p3 form an incomplete subshell of electrons and

(see Table 14.3) give rise to the terms 4S, 2P, and 2D. Applying Hund's
rule, the lowest of these should be 4S, followed by 2D and 2P. As seen
by the energy level diagram in Fig. 14.6, these are exactly the terms
observed and in the correct order. When the nitrogen atom is excited,
one of the 2p electrons is raised to one of the unoccupied orbits 3s, 3p,
3d, 4s, 4p, 4d, . . . Excited to a 3s orbit the electron configuration
8»c. 14.8] COMPLEX SPECTRA 261
becomes l$22«22p23s, which, as shown by the branching rule, gives rise
to the following terms:

N II, 2p2, lS •P lD

N I, 2p23s, 2S 2P 4P 2D

All but the incompleted subshells of electrons are usually omitted.

The parent terms given above are the lowest levels of the singly
ionized atom, a two-valence-electron system, and are shown at the top of
the energy level diagram as series limits. From the displacement law,
N II is similar to C I, with the low states lS, 1Df and 3P arising from
2p2 in agreement with theory. Of the terms 2£, 2P, 4P, and 2D, all but
the 2S built upon lS of the ion N II have been observed by combinations
with the lower terms. These combinations involve the transition of
the 3s electron to a 2p state (Al = 1).
Excited to a 4s, 5s, 6s, . . . state the same term types given above
arise to form series of terms approaching as limits the ion terms from
which they are shown to branch.
Excited to a 3p state the electron configuration becomes ls22s22p23p,
which abbreviated is 2p23p, and by the branching rule gives rise to the
following terms:

NII, 2p2, 1S SP 1D

NI, 2p23p, 2P 2S2P2D 4S*PAD *P*D*F

The parent terms are again the low levels of the ion N II. Of the
resultant terms all but the 2S and 2F have been observed in combination
with those of 2p23s.
Similarly the excitation of an electron to a 3d state gives rise to the
terms:

N II, 2p2, l8 8P lD

N I, 2p*3d, 2D *P*D2F 4P4D4F 2S2P2D2F*G

Of these terms only those built upon *P of the ion are observed. In
this connection it is to be noted that no term above the 8P series limit
262 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

has been observed in nitrogen. The longest observed series of terms are
the 4P terms of 2p2nd, where n = 3, 4, 5, 6, and 7.
The observed spectrum lines of the neutral atom of nitrogen lie in
two widely separated regions of the spectrum: one group of lines is
in the extreme ultra-violet and the other is in the visible and near infra¬
red. This is due to the enormous amount of energy required to raise
a 2p electron to a 3s state, as compared with the difference between the
3s and other excited states (see Fig. 14.6). A schematic diagram of

2p23s,4P-2p23p*D 3d24s,4F-3dz4p,4F° 3d24s.4F-3d24p.4Gc

Fig. 14.7.—Quartet multiplets observed in nitrogen and scandium.

typical quarter multiplets is given in Fig. 14.7. The fine structure of

the multiple levels is spread out to show the Land6 interval rule. The
observed line3 are plotted below to a wave-number scale.
The low 2pZ2P and 2pz,2D levels in nitrogen are metastable in
that, once an atom is in one of these states, a transition into a lower state
is forbidden. Similarly the 2p2,lS and 2p2,1D states of the ion are
metastable. Bowen has pointed out, however, that transitions within
a configuration sometimes occur in gaseous nebulae where densities
are very low.1 Transitions from lD2 to 3P2 and 3Pi, shown in the figure,
account for two of the many nebular lines X6548 and X6583.
14.9. The Scandium Atom.—The scandium atom and its spectrum
will be taken up next, as an element characteristic of Group IIIA of the
periodic table (Table 5.1). The normal state of the neutral scandium
atom is given by the complete electron configuration,
Sc, ls22s22p63s23p64s23d.

With all but one electron forming completed subshells, the norma1
state, as observed, is given by the one 3d electron as 2D. Observation
shows that when the scandium atom is excited, it is easier to raise one
1 Bowen, I. S.f Asirophys. Jour., 67, 1, 1928.
Sec. 14.9J COMPLEX SPECTRA 263

of the electrons to an outer orbit than it is to raise the last bound

3d electron. It is for this reason that the neutral scandium atom is to
be classified as a three-electron system. In order to build up an energy

3d4s2
Fiq. 14.8.—Energy level diagram of scandium.

level diagram it is convenient to start with the atom stripped of the

three outer electrons 4s23d and then allow them to return one at a
264 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

The energy state of the scandium atom triply ionized is represented

by a horizontal line at the top of Fig. 14.8. Containing only closed sub¬
shells of electrons, this state is designated ^o. If now one electron
is allowed to return to the atom it may go into any one of the unoccupied
8, p, d, /, . . . orbits, i.e., any one of the 2S, 2P, 2D, 2Ff . . . states
shown in the top part of the diagram. Of all the doublet levels observed,
3dy2D lies deepest. This is in agreement with the prediction made by
Bohr and shows that a 3d electron is more tightly bound to the scandium
atom than a 4s electron (see Sec. 17.1).
With one electron in the lowest state 3d,2D, we now allow a second
electron to return to the atom in different allowed orbits and thus build
up the Sc II levels. If the second electron is returned to an s orbit,
the resultant configuration becomes 3dns giving rise to lD and 3D terms.
Taking n = 4, 5, 6, • • • , these form series of D terms approaching
3df2D as a limit.
Returning the second electron to a p orbit, and utilizing the branching
rule, the following terms are derived:

Sc III, 3d,

Sc II, 3d np,

Setting n = 4, 5, 6, • • • , these should all form series approaching

3d,2D of Sc III as a limit. Only the first members of this set of series
have been observed to combine with lD and 8D of 3d4s. Photographs
of two of the resultant triplet multiplets 3D — SD° and 3Z> — *F° are
given in Fig. 17.7.
Returning the second electron to a d state gives rise to the following
terms:

Sc III, 3d, 2D

Sc II, 3d nd,

These form five series of singlets and five series of triplets approaching
3d,2D as a limit. The second members of these series are the levels
shown in Fig. 12.4 in combination with the triads of 3d4p. The first
member of the configuration series, viz., 3d3d, finds both electrons
in the same orbits so that the Pauli exclusion principle comes in to
exclude three triplets and two singlets. All predicted terms and no
others are observed for 3d2. The higher series members shown in Fig.
14.8 in brackets are predicted but not observed.
Sec. 14.9] COMPLEX SPECTRA 265

In turning to the neutral scandium atom we/find that the main energy
level diagram is built upon two parent configurations 3d48 and 3d* of
the ion. Starting at the lower left in the figure we have the terms arising
from the addition of a 4d electron to 3d4s. These are obtained as
follows:
Sc II, 3d4s, 1D

Sc I, 3d4«4d,
A
2S2P2D2F2G 2S2P2D2F2GA SAPADAF*G
Of these terms five doublets and three quartets have been found.
The second group of terms shown are those attributed to 3d4s4p.
All predicted terms are observed for this first member of the series
but none of those for 5p, 6p, etc.
In the third set of series 3d4sns the first member 3d4s4$ comprises
the normal state of the neutral atom.
The fourth group of series 3d2ns is built upon 3d2 of Sc II. These
terms arise as follows:
Sc II, 3d2, 8P lD 8F Hi

Sc I, 3d2ns, 2S 2P 4P 2D 2P 4F *G
With the exception of 2S the first member of each of these series has been
identified.
Similarly the fifth group of series 3d2np are built upon 3d2 as follows:
Sc II, 3d2, 1S 8P 1D 8P

Sc I, 3d2np,
IAAAA
*P 2S2P2D 4£4P4D 2P2D2F% W'FKi ADAF*G 2FH}2H
Nearly all of these have been observed for the lowest configuration
3d*4p. The remaining terms, 8P and 8F, arise from the configuration
3d*.
Schematic diagrams of the fine structure of two quartet multiplets
are shown in Fig. 14.7. Photographs of the same two multiplets are
reproduced at the top in Fig. 17.11. Altogether about 160 lines have
been identified as belonging to the Sc II spectrum, and about 360 lines
have been identified as belonging to Sc I.1
1 Russell, H. N., and W. F. Meggers, U. S. Bur. Standards, Sei. Papers, 22,329,
1927.
266 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

The displacement law of multiplicities predicts that the Sc I, Ti II,

V III, . . . spectra should resemble each other in general structure
and multiplicity. The corresponding multiplets in Fig. 17.11 illustrate
the observed similarity.
14.10. The Oxygen Atom.—As examples of four-valence-electron
systems we shall first consider the simpler elements oxygen, sulphur,
selenium, tellurium, and polonium. According to the Bohr-Stoner
scheme of the building up of the elements, the last four electrons bound

Odd Even Odd

2p3rts 2p3np 2p3nd

to each of these atoms are equivalent p electrons (see Table 5.3). As a

typical example of these elements we shall consider oxygen in some detail.
The low levels in the oxygen atom (see Fig. 14.9) are given by the con¬
figuration ls22s22p4 (abbreviated 2p4). Four equivalent p electrons
give rise to exactly the same terms as two equivalent p electrons, 1S)
*P, and 1D. The only difference between the two is that in p4 the SP
term is inverted and in p2 it is normal. The energy level diagram of
carbon has been given in the same figure for comparison purposes.
Although the low set of terms is alike for both atoms the higher term
structures are quite different. For carbon the excitation of a 2p electron
to a 3s orbit gives rise to a lP and a 3P term, whereas in oxygen the terms
are as follows:
Sec. 14.11] COMPLEX SPECTRA 267

OII, 2p\ 4S 2D 2P

O I, 2p33s, 3S 5£ lD *D lP 3P

Both 3S and hS form series approaching 4$ as a limit; lD and 3D form

series approaching 2D as a limit; and *P and 3P form series approaching
2P as a limit. The terms arising from the next configuration in oxygen
are those of 2p3Sp built upon 2p3 of the ion. They are:

OII, 2p3, *S 2Z> 2P

0 I, 2p33p, 3P 5P lPlDlF 3P3D3F lSlP'D 3S3P3D

The one quintet, four of the seven triplets, and one of the six singlets
have been identified.
Wave-lengths of some of the strongest lines in oxygen are indicated
in the diagram. Special mention should be made of the auroral green
line X5577, and the red nebular lines X6300 and X6363. These lines
arise from what are commonly called forbidden jumps. The auroral
line has been reproduced in the laboratory by Kaplan, and others, and
its Zeeman effect has been studied by MacLennan and by Paschen.
The two nebular lines have been produced in the laboratory by Hop-
field. Other nebular lines arise from similar transitions between the
low states of O II. These are shown at the top and right in Fig. 14.9.
14.11. The Titanium Atom.—One of the most complete analyses of
a really complex spectrum has been made for the titanium atom.1 This
analysis, which has been carried out chiefly by Russell, includes 43
singlet levels, 65 triplet levels, and 34 quintet levels and involves about
1400 spectrum lines. In singly ionized titanium some 33 doublet and
17 quartet levels, involving about 500 so-called spark lines, have been
identified. These constitute practically all of the lines commonly
observed in the titanium arc or spark.
In the doubly ionized spectrum, Ti III, 7 singlet and 12 triplet
levels involving some 90 spectrum lines have been observed.2
In the triply ionized spectrum, Ti IV, 12 doublet levels involving
31 lines have been identified.3
1Kiess, C. C., and H. K. Kiess, Jour. Wash. Acad. Sci., 13, 270, 1923; Jour.
,
Opt. Soc. Amer., 8 607, 1924; Gieseler, H., and W. Grotrian, Zeits.f. Phys., 26,342.
1924; Russell, H. N., Astrophys. Jour., 66, 347, 1927.
2 Russell, H. N., and R. J. Lang, Astrophys. Jour., 66, 13, 1927.
* Gibbs, R. C., and H. E. White, Proc. Nat. Acad. Sci12, 448, 598, 1926; see
also Russell, H. N.. and R. J. Lang, Astrophys. Jour., 66, 13, 1927.
268 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

A great deal of information concerning the outer structure of the

titanium atom can be obtained by bringing together the energy levels
of Ti I, II, III, and IV, into a single diagram similar to the one given
for scandium in Fig. 14.8. Because of the complexity of the titanium
diagram only the part due to the neutral atom will be given (see Fig.
14.10). The multiple levels belonging to the different electron configura-

& •,% '?%% 'dU 'DVD -fTy 'gVg ’h’h’h 4ft, Vtt

• 3d*4s4s
• 3d*4s5s
• 3d*4s6s
• 3d*4t4p
- 3d*4i4d
- 3d*4»5d

Fia. 14.10.—Energy level diagram of titanium.

tions are given by the legend at the lower right in the figure. The
normal state of the atom is given by the abbreviated electron configura¬
tion 4s23d2. All of the predicted terms 3Pf lDy 3F, and lG are observed.
In agreement with Hund’s rule the 3F lies deepest. Not far above the
normal state lies a bF term arising from the configuration 3d34$. This
multiple level as well as all levels lying below the first odd bG are metas¬
table levels. Nearly all of the predicted terms arising from 3d34s
are found. As shown by the branching rule, these terms are:

Ti II, 3d3, 2D

Til, 3d34s, lD ZD *P 3P JD 3D l¥ 3F

The observed terms of Ti I and Ti II are given in heavy type.

The next higher terms in the figure arise from the odd configurations
3d84p and 3d24s3p and are frequently referred to as middle terms. These
combining with the low even terms give rise to the major part of the
observed titanium spectrum. According to the selection rules, odd terms
Sac. 14.11] COMPLEX SPECTRA 269

combine with even terms or even terms with odd. Expressed in terms
of the electron configurations:

Odd Odd
3d24s4p 3<234p

Even Even

The transitions 3d34p to 3d24s2 give rise to some of the strongest lines
in the titanium spectrum and constitute double electron jumps 4p to
4s (Al = 1) and 3d to 4s (Al = 2). The titanium triplet shown in Fig.
14.1 is a photograph of the double electron jump 3d24s2,3F — 3d34p,3F°.
A very plausible explanation of the great strength of such lines has
been given by Condon.1 By a quantum-mechanical treatment Condon
has shown that the terms of two configurations having the same sign

5F->6# *F-‘F# >F-*G*

Fia. 14.11.—Schematic diagram showing the fine structure of quintet-quintet combina¬

tions and triplet-quintet intercombinations in titanium.

(odd or even), and lying close together or overlapping each other, belong
in part to both configurations. Because of a sort of resonance or periodic
interchange of states, the atom may be thought of as jumping back
and forth between two states having the same quantum numbers,
without radiation. In this sense Condon has shown that double-electron
transitions may be reduced to single-electron transitions (see Sec. 19.3).
Intercombinations between singlet and triplet levels and between
triplet and quintet levels are also commonly observed. A schematic
diagram of an intercombination multiplet is shown in Fig. 14.11, along
with two quintet-quintet combination multiplets. A photograph of the
*F-*G° multiplet is reproduced at the top of Fig. 17.12 and at the left
1 Condon, E. U.t Phys. Rev36, 1121, 1930.
270 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

in Fig. 14.1. The wave-lengths, frequencies, and relative intensities

of these lines are given in Table 14.5. The photographs of the corre¬
sponding multiplet in singly ionized vanadium and doubly ionized chro¬
mium (see Fig. 17.12) illustrate the displacement law of multiplicities.

Table 14.5.—Quintet Multiplet Observed in Titanium (Fig. 17.12)

Hj\(137.7) *G\(115.6) Kj\(93.0) Hl\{70.1) <

100 8.8 0.3

4981.75 5016.17 5045.43
20067.7 19980.0 198144
{100.2)
75.2 12.2 0.6
4991.08 5020.04 5043.59
20030.2 19914.6 19821.6
{81.8)
56.2 11.9 0.5
4999.50 5022.87 5040.63
19996.4 19903.41 19833.3
062.2)
41.0 8.2
5007.22 5024.85
19965.6 19895.6
Ul.9)
29.2
*Fi 5014.28
19937.5

14.12. The Manganese Atom.—A general treatment of complex

spectra can hardly be complete without some mention of the well-known
arc spectrum of manganese. There are two reasons for this; (1) The
energy level diagram is made up of terms of high multiplicity, quartets,
sextets, and octets. (2) Some of the multiple levels are inverted and
others are normal. This latter is due chiefly, but not altogether, to
the fact that the system of normal multiplets arises from five equivalent
d electrons (z.c., half a subshell) and two other arbitrary valence electrons,
and the system of inverted multiplets arises from six equivalent d electrons
[■i.e., more than half a subshell) and one arbitrary electron (see Fig. 14.12).
The normal state of manganese is given by 3d64s2,6$$. Although
theory predicts other terms for the same configuration, only the 6S
has been observed. If one of the 4s electrons is raised to a 4p state,
the most probable of the predicted states are the 8P, 6P, 6P, and 4P states
observed. A schematic diagram of the origin of these levels is shown
at the left in Fig. 14.13. Starting at the extreme left with the term
3d*f6/S, of doubly ionized manganese, a 4s electron is added to form
and 7S3 terms in singly ionized manganese. The observed separation
of these two S terms, 9473 cm"1, is due to the interaction between S « $
of the parent term and s = £ of the added electron. These levels
Sac. 14. i 2] COMPLEX SPECTRA 271
are the series limits at the left in Fig. 14.12. The addition now of a
4p electron raises and lowers the multiplicity again and adds an l value

3d6ns _3d*np_
Even Odd
3d54sns 3d54sn ♦po 4qo 4p0 6po 6qo SpO

3d54s HWs

MANGANESE
3d54s4s*
Fig. 14.12.—Energy level diagram of manganese.

of one to each term to form P terms. It should be noted that the

fine structure is drawn to one scale and the multiple-term separations

-61
-103

Mn II Mn I
9473 cm"1 6199 cm-1

7/2 36 cm”1
i S/2 -
3/2 ■
25
■*/*
151
7‘
117
jl/2

Observed Calculated
Fia. 14.13.—Schematic representation of the branching rule for the p terms in manganese.
Configuration 3d64a 4p.

to another. The interaction between the spin of the 4p electron and

S, the spin resultant of 3d54«> separates the upper P terms by about
272 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

4700 cm”1 and the lower terms by about 6100 cm""1. The observed
order 8P, #P, 4P, ®P indicates that the interaction between the spins
of 4* and 4p is considerably greater than that between the spin of 4p
and the spin of 3d5. We shall return to the calculation of the fine structure
shown at the right in Fig. 14.13.
If one of the 4s electrons from the normal state of manganese is
excited to a 3d orbit, the most probable among the many terms predicted
for the new configuration 3d64s are the 6D and 4D terms observed. Being
built upon the inverted 3d6,6Z> term of Mn II, both of these D terms are
inverted.

11/2
9/2
6 DV2
V2
3/2

9/2
8r
7/2

3/2

JL Ui JJL ill
•p0-‘D d5sp-d5sd ‘D-6D° d‘s-d‘p ‘D-6F°

OCTET AND SEXTET MULTIPLETS IN MANGANESE

Fio. 14.14.—Schematic diagram of three multiplets observed in manganese.

Similarly the excitation of a 4s electron from the metastable configura¬

tion 3d64s of manganese to a 4p orbit gives a triad of inverted quartets
and a triad of inverted sextet terms, again built upon 3d*,6D of the ion.
Schematic diagrams of three manganese multiplets are shown in
Fig. 14.14. The octet multiplet at the left belongs to the normal set of
energy levels and is due to the configuration combination 3d64s4p-3d54s4d.
The two inverted sextet multiplets arise from the configuration com¬
bination 3d®4s-3d64p. Photographs of these latter two multiplets
are given in Fig. 14.15. By drawing the initial and final states to
scale and at an angle to each other the spectrum lines are seen to come
at the intersection of the respective initial and final levels. This is
only possible where the spectrum lines are on a linear frequency scale.
The spectrograms were taken with a 15-ft. Rowland grating and mounting
where the wave-length scale is practically linear. Over a short range
of wave-lengths the frequency scale is practically linear also. The
Sue. 14.12] COMPLEX SPECTRA 273
self-reversed triplet in the upper photograph is due to transitions from
the lower 6P term of 3d64«4p to the normal state of the atom 3#4«*,e&|.
We shall return now to the calculation of the fine structure of the
P terms shown in Fig. 14.13. The splitting in each of these four terms
is attributed to the interaction between the spin resultant S* and the

X4083.64 MULTIPLETS IN MANGANESE X40i8.ii

Fig. 14.15.—Photographs of two Bextet multiplets observed in the arc spectrum of

manganese.

orbit l* of the p electron. If this interaction were due only to the

coupling of the spin of the 4p electron and its own orbit it is readily
§een, either from Eq. (14.5) or from the vector-model diagrams, that
the octet and upper sextet terms should both be normal, whereas *the
quartet and lower sextet terms should be inverted. A study, however,
of the observed intervals as well as the four vector diagrams for each
of the four P terms shows that the spin of the 5s electron also couples
quite strongly with l. Further consideration shows that the spin
resultant of 3d5 also couples (but weakly) with l. The three interactions
can be written in the usual way by

r, = aps*l* cos (s*l*),

r, = a,s*l* cos (s*l*),
Td = a,dS$l* cos (SJiJ),
where sp, and S<* are the spins of the 4p, 4$, and 3d5 electrons, respec¬
tively, and lP is the orbit of the 4p electron. Although the exact coupling
276 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

J = i j- i-
Xe II, 5p*» *Pi •Pi

Xel, 5p66p,
/A\
(§ i)o.i (i 4)1.2
/A\
(1 2) 1*2 (f ■§■)(),1,2,3

The 10 terms in /./-coupling correspond to ^So, lPi, ^2, 3&i, SP0.1.2,

3Dit2,s in L/S-coupling. The 2P§.j interval being large, these 10 levels
are split into two groups of 4 and 6 as shown, and are the first members
of 10 series approaching the double limit given above. Because the
excited electron is in a p orbit the terms behave as p terms.
Similarly the excitation of one of the 5p electrons from the normal
state of the atom to a 5d gives rise to the following terms:

Xe II, 5p5, 2P* 2P,

Xel, 5p‘5d, (H)l.2 (H)2.3 (!!)o,U3 (11)1,2,3,4

These 12 terms in //-coupling correspond to lPi, 1Z)2, lF3, 3P<u,2, ZD 1.2.3,

and 3P2,3,4 in L<S-coupling. With the exception of one term these two
groups of terms have been observed in xenon as the first members of
series approaching the double limit.
The spectra of neon, argon, and krypton are so nearly like xenon
that their energy levels and diagrams may be closely represented by
lowering the upper limit 2P* of Xe, with all of the series approaching it,
until the 2P interval is 782, 1431, and 5371 cm-1, respectively.
Selection rules for .//-coupling (see Sec. 12.14), which also apply
here, show that transitions from the 10 even series of p terms 5p6wp
to the 4 lowest s terms, (5p66s), give rise to 30 principal series of spectrum
lines. These are just the 30 series observed first by Paschen in the
neon spectrum. Transitions from the 4 series of s terms, (5p5ns), to
the lowest p terms, (5p56p), give rise to 30 sharp series of lines also
observed by Paschen in neon.
Observed transitions from the odd f terms, (5p6n/), to the even s
terms, (Sp^s), were first thought to be forbidden transitions but are
now known to be allowed transitions on the double-electron-jump
hypothesis. An / electron jumps to a 5p orbit while a 5p electron
jumps to a 6s orbit. It should be mentioned that a majority of the lines
observed in neon have also been observed in A, Kr, Xe, and Rn.
The normal state ^0 of each of the rare gases lies far below the first
excited states and is determined by only six possible lines in the extreme
ultra-violet region of the spectrum. Three of these six lines, observed
Sic. 14.13] COMPLEX SPECTRA 277

in xenon, are shown in the diagram. It should be mentioned in passing

that no levels above the lower series limit 2P| have ever been observed.
The imaginary lowering of the 2P* limit to form the term diagrams for

Fia. 14.17.—Fine structure of the « and p terms in neon, argon, krypton, and xenon.

Kr, A, and then Ne brings successively higher series members below

2Pj, where they are subsequently observed. This disappearance of
spectral terms that lie above a series limit is due to the phenomenon of
autoionization (see Chap. XIX).
n*1 2 3 4 5 6 7 8 9 10 11 n-1 2 3 4 5 6 7 8 9 10 11

KI CaI ScI Til VI CrI MnI FeI Col Nil Cul Rat Sal YtI ZrI Cal Mol MaI RuI RhI Pol AgI
Fig. 14.18.—Deep lying electron configurations in the elements of the first and second long
periods, illustrating breaks in the building up of the s and d shells of electrons.

To show the general change taking place in the term fine structure
as we go from Ne to A to Kr and to Xe, this structure for 8 and p series
has been plotted to scale in Fig. 14.17. At the left the coupling is neither
LS nor Jj but something between the two.
278 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

14.14. The Normal States of the Elements in the First and Second
Long Periods.—It was pointed out in Chap. V that certain irregularities
in the Bohr-Stoner scheme of the building up of the elements occur in
the first and second long periods. Energy level diagrams of all but one
of the known elements in these two periods have made possible the
construction of Fig. 14.18.1 Here the deepest level in each of four of
the deepest lying configurations dn~2s2, dn~ls, dn~lp, and dn has been
plotted with respect to dn_1s of the same element as zero. It is observed
that, while there is a regular progression of the configurations dn~lp
across each period, there is an apparent breaking up of the dn~2s2 curve
into two similar parts. This is only apparent, for when the dn~2s2 curve
is used as the origin in plotting, the other two curves for dn_1s and dn~lp
appear as the broken curves. The curves plotted either way show,
however, a gradual increase in the binding of 3d or 4d electrons over
48 and 5s electrons as we progress from element to element. With the
addition of one, two, and three d electrons there is a strong tendency
to complete half a subshell of electrons so that with the addition of a
fourth d electron to form the normal state of Cr, or Mo, one of the s
electrons also goes over into a d state to yield 3d64s, or 4d55s. With
the shell half filled, the next electron added to form the next element
goes into a 4s or 5s orbit to again fill the s subshell. At the approach
of a complete subshell of d electrons we observe that additional d electrons
are again added at the expense of the s subshells. No such anomalies
are found to exist in the sequences of elements where p electrons are
added.
14.15. Houston’s Treatment of One s Electron and One Arbitrary
Electron.—A quantum-mechanical treatment of the fine structure due
to the interaction of one s electron with one arbitrary electron has
been made by Houston,2 by applying the Darwin-Pauli treatment of the
electron to the Schrodinger wave equation for two electrons. By
setting up suitable wave functions, based upon Schrodinger’s theory,
solutions for the spin-spin resonance interaction between the two electrons
and the spin-orbit interaction for the arbitrary electron are obtained.
The treatment is one involving perturbation methods in which only
the first-order perturbation to the energy is evaluated. Houston’s
results may be given most easily by his resultant equations;

Singlet = iA[(X - 1) + {(X + l)2 + 4L(L + 1)}*] (14.7a)

(»rL+i = A-L (14.76)
Triplet^ = Ul(X - 1) - {(X + l)2 + 4L(L + 1)}*] (14.7c)
(#rL-i - -A(L + 1) (14.7d)

1 Graps, R. C., and H. E. White, Nat. Acad. Sci., 14, 559, 1928.
* Houston, W. V., Phys. Rev., 33, 297, 1929.
Sue. 14.15] COMPLEX SPECTRA 279

where A is a measure of the spin-orbit interaction of the arbitrary*

electron and is given by a/2, a defined in Eq. (12.2), and X is a parameter
which measures the magnitude of the spin-spin interaction. The
second of these enters into the two terms having equal J values only,
and not into the two outer triplet terms at all. These four interaction
energies are plotted in Fig. 14.19 for a ps electron configuration. If
X > > A, we have LS-coupling with the singlet high above the triplet

Fig. 14.19.—Theoretical curves for the fine structure of a ps electron configura¬

tion Observed triplets and singlets are shown to be in good agreement with theory.
(After Houston,)

and the triplet fine structure following the interval rule with the ratio
2:1. If X << A we have ^‘-coupling with the same outside triplet
separation. This is just the result obtained from the classical theory
given in Secs. 12.7, 12.8, and 12.9 (see also Fig. 12.13). For negative
values of X the middle triplet level becomes the singlet, and the singlet
level becomes the triplet. Such cases of the singlet lying below the
triplet are observed in the diffuse series of Cd and Hg.
Observed triplet and singlet intervals for Si, Hg, Ge, Sn, and Pb
are seen (in Fig. 14.19) to fit the theoretical curves to a remarkable
degree of accuracy. In plotting the observed terms, the value of A
is first determined by dividing the observed intervals 3P0 — 5P* by three.
This gives a value of A in wave numbers which, with the observed inter¬
vals of *Pi and 3Pi substituted in Eqs. (14.7a) and (14.7c), respectively,
gives two values of X. These are averaged and the terms plotted at
that X. The configurations plotted are: Se, 3p4s; Hg, 6p6s; Ge, 4p5«;
Sn, 5p6s; Pb, 6p7s.
Many other configurations are found to fit these curves at large
values of X (not shown in the figure), where the singlet lies far above
280 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

the triplet and the triplets have the ratio 2:1. This figure is also to be
compared with the observed silicon series in Fig. 12.13.
In addition to the interaction energies Houston’s formulas include
the Landd g factors for the weak-field Zeeman effect. They are:

_X±1_
Singlet lgL
1 + 2L(L + 1) 2L(L + 1)[(X + 1)’ + 4L(L + 1)]»
(14.8a)
L + 2
(14.8b)
L+ 1
X + 1
Triplet ^ 2L(L + 1)
+ 2L(L + 1)[(X + 1)* + 4L(L + 1)]*
(14.8c)
L - 1
(14.8eJ)
L

For the outside triplet terms these formulas give the same g factors
as before and are independent of the coupling scheme. According to
the g sum rule, the sum of the g’s for the two terms with the same J
value is the same for all couplings and all field strengths. This is given
by the sum of Eqs. (14.8a) and (14.8c):

%9l = 2 + UjT+T) (14-9)

The X values obtained for Sn and Pb from Fig. 14.19 and substituted
in the above given equations yield g factors in excellent agreement with
experimental values obtained from the Zeeman effect. The following
calculated values were obtained by Houston.

Table 14.6.—Calculated and Observed g Factors for Sn and Pb

(After Houston)

Element Term Obs. Calc. Element Term Obs. Calc.

«

lPi 1.125 1.123 1.150 1.148

Sn Pb
3Pi 1.375 1.377 8Pi 1.350 1.352

20 2.500 2.500 20 2.500 2.500

The application of Houston’s formulas to configurations of the type

p5s and d?s has been carried out by Laporte and Inglis,1 and by Condon
and Shortley.2 This extension is possible for two reasons: first, the
terms are of the same type with the same J values, and, second, the
outside triplet separation is just the same as for one electron except that

Laporte, O., and D. R. Inglis, Phys. Rev.f 36, 1337, 1930.

* Condon, E. U., and G. H. Shortley, Phys. Rev., 36, 1342, 1930.
Sue. 14.16] COMPLEX SPECTRA 281

it is inverted. This latter result means only that the above given A
factor is negative instead of positive. In order that the singlet be above
the inverted triplet, as observed, X must also be negative.
Substituting —X' for X and —A' for A in Eqs. (14.7a, b, c, and d),
formulas for the configurations p5s, d?s, and are obtained. Curves
plotted from these new equations are given in Fig. 14.20, along with
the observed intervals for a number of elements. It will be noted
that, even for highly ionized atoms like In IV and Sn V, the observed
intervals are in excellent agreement with theory. The configurations

Fig. 14.20.—Fine-structure curves for the configurations ph8 and d*a. Observed terms are
shown to agree with the theoretical curves. (After Laporte and InQiis.)

plotted in the left-hand figure are: Ne I, Na II, and Mg III, 2p63s; A I,

K II, and Ca III, 3p64s; Kr I, 4p55s; Xe I, 5p66s. Those in the right-hand
figure are: Ni I, Cu II, Zn III, Ga IV, and Ge V, 3rf94s; Pd I, Ag II,
Cd III, In IV, and Sn V, 4d®5$; Pt I, 5^6s; Cu II, 3d®6s; Cu II, 3d»7s.
14.16. Slater’s Multiplet Relations.—A general first-order perturba¬
tion treatment of the interaction between two arbitrary electrons in a
central force field has been attempted by Slater. Again taking into
account the Heisenberg electrostatic resonance interaction, but neglect¬
ing spin-orbit interactions, he obtained equations for the relative intervals
between the different triplets and singlets belonging to the same electron
configuration. Although Slater’s calculations are beyond the scope
of this book, we may consider briefly some of his results.1 In titanium,
for example, the lowest configuration 3d24s2 gives rise to two triplets
and three singlets, 3P, 8F, lS, lD, and lG. Assuming certain values for
the coupling parameters in his equations, Slater obtains the following
interval ratios between levels:

1 Slater, J. C., Phys. Rev., 84, 1293, 1929.

282 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

Value »F - - HI W _ id *F - 1S

Observed. 8500 cm”1 12100 cm”1 7200 cm”1 15100 cm”1

Calculated. . 98 113 67 266

One of the important features of these calculations is the order of the

terms. Both the observed and calculated order starting with the lowest
term is 3P, lD, 3P, k-r, and 1S.
In general, Slater's results show that one cannot always rely upon
Hund's rule for excited configurations.
14.17. Multiplet Relations of Goudsmit and Inglis.—An extension of
Houston's and Slater's quantum-mechanical treatment of two electrons
has been made by Goudsmit1 and Inglis.2 Taking into account the
electrostatic spin-spin interaction, as well as the spin-orbit interaction,
equations have been derived for two or more arbitrary electrons. Though
their mathematical treatment will not be given here, their results are
of great importance and will be given. In Houston's beautiful account
of one 8 electron and one arbitrary electron, there were but two inter¬
actions to be taken into account. With two or more arbitrary electrons
the number of interactions is greatly increased and the problem becomes
very difficult. Solutions for each particular configuration can be
obtained from the observed intervals, however, by adjusting the param¬
eters in the equations until the best agreement is obtained. In the
special cases of equivalent electrons a number of the interactions become
identical and the equations are greatly simplified. These configurations
are of importance in that they involve the normal states of a number
of atoms. Theoretical curves for two, three, and four similar p electrons
are shown in Figs. 14.21 and 14.22. Observed terms for different
elements are given to show the excellent agreement with theory. The
configurations plotted are shown on the diagrams. X is a measure
of the electrostatic spin-spin interaction and A is a measure of the
magnetic spin-orbit interaction. One rather interesting case is that of
tellurium where the 3P term is partially inverted; 3P0 and 3Pi fall almost
together (Av = 44 cm-1) and about 4700 cm”1 above 3P2. The two
singlet terms in selenium have not yet been identified.
14.18. Relative Intensities of Multiplet Lines.—Extensive measure¬
ments of the relative intensities of spectrum lines have been made by
a number of investigators. In many cases the observed intensities
are found to be in excellent agreement with the well-known intensity
formulas (see Sec. 12.15). In the titanium multiplet of Fig. 17.12 and

1 Goudsmit, S., Phys. Rev., 36, 1326, 1030.

* Inglis, D. R., Phys. Rev., 38, 862, 1931.
Sec. 14.18] COMPLEX SPECTRA 283

Table 14.5, for example, the following experimental values have been
obtained by Harrison.1

Fig. 14.21.—Fine-8tructure curves for the configurations ps and p4. (After Goudsmit.)
Observed terms are shown to agree with the theoretical curves.

The calculated intensities (see table of intensities in the Appendix)

are given in italics and the observed intensities beneath them in heavy
type.

Table 14.7.—Observed and Calculated Intensities for the Titanium

Multiplet of Fig. 17.12

< < «g; < <

*Fb WO 8.5 0.8
100 8.8
bFA 76.2 12.2 0.6
78.2 12.2
*Ft 56.7 12.0 0.6
56.2 11.9
*F* 41.2 8.2
41.0 8.2
§Fi 29.7
29.2

In some cases where the observed intensities of a multiplet do not

agree with the formulas, the sum rules for the lines with a common
initial or final state nevertheless hold. Harrison has shown, for certain
multiplets in titanium, that while the sum rules do not always apply
to a given multiplet, they may apply to a triad of multiplets. In still

1 Harrison. G. R.. Jour. Opt. Soc. Amer., 17, 389 1928.

284 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIV

other cases it is necessary to take into account ail of the multiplets

associated with given initial and final electron configurations.

155)
*i!

-+-LS | Coupling

Fig. 14.22.—Fine-structure curves for the configuration p3. (After Inglis.) Observed
terms are shown to agree with the theoretical curves.

It has also been shown that the relative intensities between related
multiplets, taken as a whole, obey the sum rules (see Sec. 12.16). For
a titanium triad of quintet multiplets, for example, the following values
have been obtained by Harrison and Engwicht:1

90 70 50 calculated
81 70 49 measured (without and T corrections)
89 70 48 measured (with vK i

The measures taken from many multiplets show that better agreement
between theory and experiment is obtained when v4 and temperature
corrections are made.
Problems

1. The following list of observed lines form a multiplet. Plot the lines to a
frequency scale and, looking for equal frequency differences by marking them on a

1 Harrison, G. R.. and H. Engwicht, Jour. Opt. Soc. Amer., 18, 287, 1929.
Bsc. 14.18] COMPLEX SPECTRA 285

strip of paper, find the term intervals. Tabulate the lines in multiplet form, as in
Table 14.3. Applying the Land6 interval rule to the observed terms, determine the
J1S, and L values.
2. Using the values of X as read from Fig. 14.19 for Si, Ge, and Hg, compute the
g factors for each of the four terms, respectively [see Eqs. (14.8)]. Show that the
g sum rule holds.
.
8 Derive all the terms arising from the electron configurations rf1*, d*pt p*df p4s,
and d7p.
4. Carry out the calculation of the F, factors for d* as outlined in the middle of
Sec. 14.6, and tabulate them. From this table make another tabulation like that
given in the top third of Table 13.13. From these r sums determine the T values of
the terms as shown in the middle third of Table 13.13.
 CHAPTER XV

THE ZEEMAN EFFECT AND MAGNETIC QUANTUM NUMBERS

IN COMPLEX SPECTRA

It is well known to spectroscopists that the Zeeman effect plays

an important part in the making of any analysis of a complex spectrum.
This is due largely to the fact that the Zeeman patterns for all types
of transitions can be predicted with considerable accuracy. Let us
start therefore with a calculation of Zeeman patterns for various types
of transitions and then compare the results with the observed patterns.
To do this we must first obtain an expression for the magnetic energy
of an atom in any stationary state.

Fio. 15.1.—Classical models for the motion of complex atoms in a weak magnetic field.

'15.1. Magnetic Energy and the Land6 g Factor.—In the previous

chapter we have seen that the vector model for complex spectra is
essentially the same as the model for two valence electrons. The
extension into complex spectra was made by considering all but one
of the electrons as a parent term or configuration to which the remaining
electron is added (see Figs. 14.2 and 14.3).
In a weak magnetic field the atom as a whole becomes quantized and
oriented to the field Hf in such a way that the projection of J*h/2ir on H is
equal to Mh/2ir (see Fig. 15.1). For terms with whole-integral values
of J the magnetic quantum number M = 0, ±1, +2, • • • , ±J. For
terms with half-integral values of J, M = +$, ±f, +$, • • • , ±J.
The magnetic quantum numbers for a 5Z>3 term, for example, are M = 3,
286
Sec. 15.1] ZEEMAN EFFECT, MAGNETIC QUANTUM NUMBERS . 287

2, 1, 0, —1, —2, —3, and for a 4Pj term they are M » f, $, J, — \} —

—Just as in the case of one electron the magnetic energy is determined
by My Hy and the Land6 g factor [see Eq. (10.17)]:

-AT = M-g- j^-2 cm-1 = M ■ g ■ L cm"1. (15.1)

For a given field strength H the separations of consecutive magnetic

levels for one term, relative to those of another, will be given by the
g factor of each respective level. Just as in one- and two-electron
systems the g factor is the ratio between the total magnetic moment
of the atom in Bohr magnetons and the total mechanical moment in
units of h/2jr.
If the coupling is LS (i.e.y all spins coupled together to form S*
and all orbits coupled together to form L*, and S* and L* in turn coupled
together to form J*), the g factor is obtained in exactly the same manner
as it is for a single electron. S* is thus treated like the spin of a single
electron and L* like the corresponding orbital momentum vector. From
Eq. (10.11) we write
J*2 + S*2 - L*2
*7 = 1 + (15.2)
2 J*2

Should the coupling on the other hand be Jj (i.e., the spin and orbit
of the last bound electron coupled together to form jf, and jf in turn
coupled together with JJ of the parent term of the ion to form J*),
the g factor is obtained in exactly the same manner as it is for two
electrons in ./^-coupling [see Eq. (13.14)]. JJ of the parent term is
now treated like j2f the spin-orbit resultant of a single electron to be
coupled with the j* of the other electron, so that

j*2 + jt2 - jy , j*2 + j*2 - if2

g = 91 2J*2
+ gp-
2J*2
(15.3)

where gi is the g factor for the single added electron derived from Eq.
(10.11), and gP is the g factor for the parent term derived from Eq. (13.8)
or (13.14). It should Be pointed out that the parent term may arise
from an LS- or jj-coupling scheme even though the added electron j*
is coupled to it finally in Jj-coupling. Tables of g factors for LS-coupling
are given in the Appendix. For Jj-coupling there are so many possi
bilities that Eq. (15.3) must be used.
A graphical representation of the magnetic energy corresponding
to the vector model, given by Eq. (15.1), is shown in Fig. 15.2 for a 4Dj
term. The vector J* is quantized with respect to the field direction H
so that its projection M = f, $, i, —i, —— $, —J. Multiplying
J* by the g factor ^ and projecting on H we obtain the values of Mg.
These are proportional to the magnetic energy. The result is eight
288 INTRODUCTION TO ATOMIC SPECTRA [Chap. XV

equally spaced levels each separated from the next by g • L. If the Mg

values are multiplied by He/Armc2, the figure will give the intervals in
cm"1.
H

Fig. 15.2.—Schematic representation of Zeeman splitting for a 4Z)j term in a weak magnetic
field.

15.2. The Calculation of Zeeman Patterns.—As an example of the

calculation of Zeeman patterns we shall first consider the theoretical
patterns of7Si — 7P2,3,4, observed patterns for which appear in chromium
(see Fig. 14.1). Since S and P terms never have more than one and
three fine-structure levels, respectively, this multiplet contains but
three lines. Although the lines look just like those of a principal-series
triplet, the Zeeman patterns are entirely different from those shown in

Fig. 15.3.—Zeeman effect, theoretical and observed patterns, for a rSt — 7P*,«,4 multiplet.
(Observed patterns are from original spectrograms by H. D. Babcock.)

Fig. 13.4. The intervals between the three lines indicate L&-coupling
with a ratio approximately 4:3. The q factors for 7/Ss, 7Pj, TPs, and 7Pi
(see Appendix) are 2, !•}, and £, respectively. A graphical construe-
Sec. 15.3] ZEEMAN EFFECT, MAGNETIC QUANTUM NUMBERS 289

tion, such as that shown in Fig. 15.2, gives th6 intervals shown in Fig.
15.3.
With the selection rule that in any transition the magnetic quantum
number M changes by +1, 0, or —1 only, the Zeeman patterns shown
below in the same figure are obtained. Another method is to write
down the Mg values for both the initial and final states of a given line
with equal values of M directly above and below each other, as follows:

M =3 2 1 0 -1 —2 -3

Vertical differences, AM = 0, give the p components at

Av = 0, L, and ±£L cm-1.

Diagonal differences, AM = ±1, give the s components at

Av = ±|L, ±£L, ±#L, ± £L, and ±|L cm“l.

These fractions for the whole pattern are written together as

7S3 - 7P2, Av = ± (0). (!)> (2y> 5> 6> h ^ cm-1.

Similarly the patterns for the two other lines are calculated as

% - 7P3, A, = ± (1), (2), (3), 21, «, 23,24,25,2^ ^

7iS3 - 7P4, Av = ± 5’ Ml 8> 9’ _l cm-1.

The observed patterns of these three lines, reproduced at the bottom

of Fig. 15.3, are from original spectrograms taken by Babcock at the
Mt. Wilson Observatory. These are some of the earliest anomalous
Zeeman patterns to be photographed in any laboratory.
15.3. Intensity Rules and Zeeman Patterns for Quartets, Quintets,
and Sextets.—The intensity rules for the Zeeman effect in complex
spectra are the same as those for one- and two-electron systems. These
rules, derived from the classical orbital model of the atom and the sum
rules, are given in Sec. 13.3. Since these rules involve only the Jfs
and M’s of the initial and final states they are valid for all coupling
schemes. The relative intensities of the different components of a
given pattern are usually represented by the heights of the lines in the
schematic diagrams. It is to be noted that the component J * 0 to
290 INTRODUCTION TO ATOMIC SPECTRA [Chap. XV

J mm 0 of 7Si — 7P8 (Fig. 15.3) is zero as given by the intensity formulas,

i.e.y the transition is forbidden.
A number' of patterns typical of quartet and sextet multiplets in
general are given in Fig. 15.4 for vanadium. Patterns typical of quintet
multiplets are given in Fig. 15.5 for chromium. The calculated patterns
are given above or below each observed pattern. The five lines XX4352,
4344, 4339, 4337, and 4340 are the five diagonal lines of a 5D — bF
multiplet. It is to be noted that, as the number of components decreases
in going from one pattern to the next, the separations increase. The line
X4340, (6/>o — 6Fi), although partly masked by X4339, is of particular
interest in that it is one of the few spectrum lines known which remains
single in all field strengths. The reason for this is that the Mg values
for both the initial and final states are zero. Unresolved patterns like
those of X4352 and X4344 are called blends.
The pattern X4613 in chromium has 2.5 times the separation of a
normal triplet. In general the Zeeman patterns for different combina¬
tions are so different from one another that the assignment of a completely
resolved pattern to a definite multiplet is unambiguous. The importance
of resolved patterns in making the analysis of an uncharted spectrum, as
well as checking the analysis of a spectrum already made, cannot be
overemphasized.
16.4. Paschen-Back Effect in Complex Spectra.—So far as experi¬
mental observations are concerned the Paschen-Back effect in complex
spectra is practically unknown. This is due in general to the fact that
even with the strongest magnetic fields attainable, the multiplet separa¬
tions are many times the Zeeman separations of any one level. There
is little doubt, however, that if, and when, strong enough fields are
obtained the observed splitting will be very closely that predicted by
theory. There is some justification for this as we shall see in the study
of the Paschen-Back effect of hyperfine structure in Chap. XIX.
In a weak magnetic field a complex atom is quantized with respect
to H more or less as a rigid whole. When the field becomes strong
enough so that the Zeeman levels of each fine-structure level begin
to overlap the patterns of the neighboring levels of the same multiplet,
the strong-field Paschen-Back effect sets in and S* and L*, or JJ and
become quantized separately with the field direction H. The quan¬
tum conditions under these circumstances are the same as those for two
electrons: the projections M of S* and L*, or J* and j*, on H take integral
or half integral values only, depending upon whether S and L, or JP and
jf9 are whole or half integral respectively. When the field becomes
still stronger and the magnetic levels of one multiplet begin to overlap
those from the neighboring multiplets of the same configuration, the
complete Paschen-Back effect sets in and each s* and 1* becomes quan¬
tized independently with the field H. The quantum conditions again
Sec. 16.41 ZEEMAN EFFECT, MAGNETIC QUANTUM NUMBERS 291

BB 4F%-4f% HSS 4F%-4% >4421 M4& A4444 6D^-4P^1

m. *03.5.79) 553537M41454m JM5S81J41M267 ‘(7). 29.43 *2).(I2>.24,32.40

53 105 IT TJ
Fig. 15 4.—Zeeman patterns, observed and calculated, for typical quartet and sextet
multiplet lines in vanadium. (Observed patterns are from original spectrograms by H. D.
Babcock.)

ZEEMAN RATTERNS of CHROMIUM

® m \\ m vp, m Vf, m Wp. m W

*(010.1.3,5 tmi 7.911.13 J&M. 0_ mb i(Cm.3Ab.e.7
2 6 2 0 2 -3-

Fig. 15.5.—Zeeman patterns observed and calculated, for typical quintet multiplet lines
in chromium. {Observed patterns are from original spectrograms by H. D. Babcock.)
292 INTRODUCTION TO ATOMIC SPECTRA [Chap. XV

are just the same as those for two-electron systems. Each spin and
each orbit has a component along H. The sum of these components
-f Smz, gives the magnetic quantum number M of the entire
atomic system in that state.
When considering transitions between levels in weak, strong, and
very strong fields, each multiplet in L£-coupling in a strong field will
go over to a pattern resembling a normal triplet (see Fig. 13.9). In
a very strong field all the lines arising from the configuration of electrons
will spread out into a pattern again resembling a normal triplet (see
Fig. 13.14). Patterns similar to those of Figs. 13.11 and 13.15 will
result in the case of /./-coupling. We shall now make use of the weak-
and strong-field and very strong field quantum numbers to calculate
the allowed spectral terms arising from given electron configurations.

"V V* -/2 _ ^
= 1 0-1 ESS 0
Ms = 1 0 V4 Ms = 3/2 '/z -yz a ms = ’/2 Vz
; -Vi Ms= K -V4jj-H -Ji J/2 -Ji
Sing.'TripJ Doubt 'Quar. ms3 Doub.1 mS3

D
d :| 2 1 0 -1 -2 |
m, s 2 1 0-1-2 _ Ml; 3 2 10 -1 T
*2
2 1 0-1-2 0 S Ml = 2 1 6 -i ! -2 0 P
Mtp=
D mlt Ml. 1 0 -1 ;-2 -3 -i
■
P | D F
Fia. 16.6.—Combination of very strong field quantum numbers for the electron configura¬
tion dap.

16.6. Derivation of Spectral Terms by Use of Magnetic Quantum

Numbers.—In deriving spectral terms from magnetic quantum numbers,
one may start with the very strong field where the coupling between
all spins and orbits is broken down. Consider as a first example three
arbitrary electrons, no two of which are equivalent, such as s, d, and
p. For these electrons we shall have $i = s2 = h, $3 = i, li = 0, U = 2,
and U — 1, for which the magnetic quantum numbers are mtl « ±£,
mn = ±$, = ±i, mh « 0, mh = 0, ±1, ±2, and mh = 0, ±1.
Extending the scheme of Breit (see Sec. 13.10), we first combine the
spins Si and s2 as shown at the left in Fig. 15.6. The resultant values of
Map are cut apart by the L-shaped dotted line to form parent values for
the third electron. Combining each run of MSp values with m9i we
obtain three runs of Ms values. These correspond, as shown, to strong-
field quantum numbers for doublet, quartet, and doublet terms. In
a similar fashion the Vb are combined in the lower arrays. mh with m*.
forms the run MLp = 2, 1,0, —1, —2, to form the parent values for the
Shc. 15.6] ZEEMAN EFFECT, MAGNETIC QUANTUM NUMBERS 293

third electron. Combining these with mi the strong-field runs of

values Ml * 1 to —1, 2 to —2, and 3 to —3, corresponding to P, Z),
and F terms, respectively, are obtained. In terms of the branching
rule these resultant values would be written down from the parent
terms as follows:

Parent terms

Final terms

The first and fourth arrays give the first row; and the second, third,
and fifth arrays give the last row. If now each final run of Ms values
is combined with each run of ML values, all of the weak-field quantum
numbers of the above final terms are obtained. This has been done

F
ml= 3 2 10-1-2-3
M = % 7/2 5/2 34 1/, -1/2-3/2 %
M« i % % \ 4 -%l-%
M= % % 14 J/2

M = % V2 -Vfe-%1 -%!-%!-%
4p |4p 4p ,4p Ms
F3/! F5/l F%\ F%
Fig. 16.7.—Combination of the strong-field quantum numbers of a AF term showing the
formation of resultant weak-field quantum numbers.

for the 4F term in Fig. 15.7. The others will be left as exercises for the
reader.
15.6. Equivalent Electrons and the Pauli Exclusion Principle.—In
the calculation of spectral terms arising from two or more equivalent
electrons we must take the Pauli exclusion principle into account and
start with the very strong field quantum numbers (see Sec. 13.11).
Consider as an example three equivalent p electrons. We first write
down the six possible states for a single p electron in a very strong field.
They are:
m.= i i i ~i
mi = 1 0 -1 1 0-1 (15.4)
(o) (b) (c) (d) (e) (/)
Since the exclusion principle requires that no two electrons have
all quantum numbers n, l, m, or mi alike we collect all possible com-
294 INTRODUCTION TO ATOMIC SPECTRA IChap. XV

binations of the above states three at a time, with no two alike. They
are:
Table 15.1

abc acd adf bef ede

abd ace aef bde cdf
abe acf bed bdf cef
abf ade bee bef def

Writing Ms for 2m, and ML for 2m/, the results are tabulated as follows:

For Ms — h Ml = 0
For Ms = i = 0, 2, 1, 0, — 1, —2, 1, 0,-1
For Ms = -i ML = 0, 2, 1, 0,-1,-2, 1, 0,-1
For Ms - - Ml = 0 (15.5)
These are just the strong-field values for *S, 2D, and 2P terms (see Table
14.3).
In a similar calculation for four equivalent p electrons we write
down the same six possible states for one p electron [Eq. (15.4)] and
take all possible combinations of four states at a time with no two alike.
They are:
Table 15.2
abed abde aede adef beef
abce abdf aedf bede bdef
abef abef acef bedf edef

When m, and mi values are summed for each combination and tabulated
one finds:

For Ms = 1, Ml = 1, 0,-1

For Ms — 0, Ml = 1, 0,-1, 2, 1, 0,-1,-2, 0

For Ms = —1, Ml — 1, 0,-1
corresponding to 1S, ZP, and lD terms. These are exactly the terms
arising from two similar p electrons (see Table 14.3).
For five equivalent p electrons there are just six combinations, taking
five states at a time. They are

abode abcdf abcef abdef acdef bcdef

When summed and tabulated as before one finds:

For Ms « i, Ml = 1, 0,-1
For Ms = -i, Ml - 1, 0,-1

These are just the same values given in Eq. (15.4) for one electron and
give rise to a 2P term.
For six equivalent p electrons there is but one combination of the
very strong field quantum numbers, taking six states at a time, with no
Sec. 15.6] ZEEMAN EFFECT, MAGNETIC QUANTUM NUMBERS . 295

two alike, and that is dbcdef. The sum gives Ms = 0 and Ml = 0, *a

»S0 term. Here we see that six p electrons with the same total quantum
number, by Pauli’s exclusion principle, close a subshell and have zero
as a resultant angular momentum (see Table 14.3). The table of
p electrons is symmetrical about the center where the greatest multi¬
plicities and the greatest number of magnetic levels arise.
In the calculation of terms arising from equivalent electrons in
jj-coupling we need only go to the strong-field quantum numbers.
The reason for this is that here each electron is specified by four quantum
numbers and it is not necessary to go to very strong fields to separate
them as in LS-coupling. As a single example of the calculations let
us consider three equivalent p electrons. The six possible states for
a p electron in a strong field are

i- t t ! t i i
m, = | \ -4 -f 4 -4
(a) (b) (c) (d) (e) (f)

Since the Pauli exclusion principle requires that no two electrons have
all the quantum numbers n, l, j, and m, alike, we collect all possible
combinations of the above given states three at a time, with no two alike.
These will be just the combinations given in Table 15.1. Summing up
the values of m, for each of the 20 combinations the results may be
tabulated as follows:

For ji = i) J 2= |= % j wij==I A) 2) 4
Forji555^, ji j 3 2j 4* 2) 1) 4»4» 4> 4
Forji = |, j2 = i, j 3 = i m, = |,4,-4,“f
These correspond to the five terms (j f |)j, (f | i)4, (f f 4)*, ({ |J)t,
and (| 4 i).], which go over in LS-coupling to 2P3, 2P3, 2PS, 2D3, and
4£3 (see Fig. 14.22).
We shall now turn to the more complicated cases of equivalent d
electrons. Two equivalent d electrons have already been treated in
Sec. 13.11 and shown to give rise to lS9 3P, lD, 8P, and *G terms. For
three or more electrons we continue the same scheme by first writing
down the 10 possible states for one d electron:

m. = 4 4 4 4 4 —* —* —* —* —*
mi = 2 1 0-1-2 2 1 0-1-2 (15.6)
(a) (6) (c) (d) (e) (f) (g) (A) (i) (j)

For d8 we take these three at a time with no two alike and find 120
combinations that have to be collected and segregated. The resultant
sums will correspond exactly to the terms given in Table 14.3. For
d4, taking four at a time with no two alike, there are 210 possible Com¬
binations to be evaluated. Such calculations become tedious and
296 INTRODUCTION TO ATOMIC SPECTRA [Chap. XV

cumbersome, to say the least, but fortunately a shorthand method

has been discovered by Gibbs, Wilber, and White,1 and by Russell.*
We shall first apply this scheme to three equivalent d electrons.
Referring to Eq. (15.6) take only those combinations of numbers
three at a time where all three m, values are plus and those where all
three are minus. These in terms of the letters are
abc ace bde fgh fhj gij
abd ade cde fgi fij hij
abe bed fgj ghi
acd bee fhi ghj

Summing the corresponding values of the quantum numbers we

obtain for the plus values
Ms = Ml — 3, 2, 1, 0, — 1,—2,—3, 1, 0,-1
and for the minus values (15 7)
Ms = Ml = 3, 2, 1, 0,-1,-2,-3, 1, 0,-1.

These correspond to parts of 4P and AF terms, the rest of which

will be obtained from the following process. Of the 100 remaining
combinations to be made and summed each one will have two of the
m, values plus and the other one minus, or vice versa. If in each of
these 100 combinations the two m„ values that are alike are considered
alone, they will contain just the combinations that give triplets for two
equivalent d electrons: these are 8P and 3F. The third electron in all
combinations has an opposite sign and corresponds to the values for a
single d electron. Combining the triplets that arise from d2 with the
doublet that arises from d in all possible ways to form doublets, we
obtain just the 100 combinations desired. This is done quickly by the
branching rule as follows:
3P *F

*P
A *D 2F *P 2D 2F HI 2H

This final set of terms has Ms values of +£ and Striking out

one P and one F term to go with Eq. (15.7), we have left 2P, *D, *D, 2F,
*Gr, and 2H. The resultant terms for three equivalent d electrons are
d*, 2D, 2P, 2D, 2F, 2G, 2H} 4P, 4P.

Assume, as another example, that the correct terms have been derived
for d, d*, d*, and d4 and we wish to calculate the terms for d6. Referring
1 Gibbs, R. C., D. T. Wilber, and H. E. White, Phys. Rev., 29, 790, 1927.
1 Russell, H. N., Phys. Rev., 29, 782, 1927.
Sac. 15.6J ZEEMAN EFFECT, MAGNETIC QUANTUM NUMBERS 297

to Eq. (15.6) we take only those combinations five at a time where all
five m, values are plus, or all five minus. These are abode and fghij
Summing these we have for the plus values
Ms = 4, Ml = 0
and for the minus values
Ms - Ml - 0. (15.8)
These are just one-third the values needed to form a 6S term.
There will now be a set of combinations from Eq. (15.6) in which
four spins will be plus and the fifth one minus, or vice versa. That
part of the combinations in which four of the spins add gives quintets.
From Table 14.1 we find for d4 one quintet only, 8Z>. The fifth electron
with opposite sign corresponds to 2D. Combining these in all possible
ways we get

6Z>

Striking out the S term to go with Eq. (15.8), we have left parts
of 4P, 4Z>, 4F, and HI terms. It should be pointed out here that we now
have those parts of the *S term for which Ms = f, — f, and — f, and
those parts of each-quartet term for which Ms = $ and —f.
Again there will be a set of combinations from Eq. (15.6) in which
three spins will be plus and the two others minus, or vice versa. The
combinations of d3 have already been shown to give 4P and AF terms,
and the combinations of d2 to give 3P and 3F terms. Combining these
in all possible ways to form doublets,
4p 4p ip 4p

2s 2P 2Z> 2D 2F HI 2D 2F HI 2S 2P 2D 2F HI 2H 2I
Striking out one S term to complete the *S term above, and one
each of P, D, P, and G, to complete the 4P, 4D, 4P, and HI terms, respec¬
tively, we have the remaining terms as doublets. The resultant terms
arising from five equivalent d electrons are, therefore,

d6, 2Z>, 2P, 2Z>, 2P, 2G, 2H} 2S, 2D, 2P, 2G, 2/, 4P, 4P, 4Z>, HI, *S.
Since a subshell of d electrons lacking n electrons to complete it
will give rise to exactly the same terms as a configuration of n equivalent
298 INTRODUCTION TO ATOMIC SPECTRA [Chap. XV

electrons, written symbolically d10”* = dn, the lower half of Table 14.3
is symmetrical with the upper half.
A continuation of this process for equivalent f electrons leads to
the terms given in the tables in the Appendix. The number of possible
combinations for each configuration of equivalent electrons is given in
parentheses at the beginning of each row. These numbers are computed
from the well-known combination-theory formula for p things taken q
at a time.

Number of combinations = —ry• (15.9)

q\(p - q)\
This number is equal to the number of Zeeman levels for the entire
configuration. For d2, for example, the terms are 1S, 3P, lD, 3F, and
lG, for which there are 1+9 + 5 + 21 + 9 = 45 Zeeman levels.
Without the above presented shorthand process of making several
thousand combinations, the calculations for equivalent / electrons would
become very laborious.
The largest number of Zeeman levels and the highest multiplicity to be
derived from equivalent electrons are from half a subshell. Furthermore,
the term of highest multiplicity is an S term and, wherever observed,
always lies deepest. On the quantum mechanics the configuration for
an S term is spherically symmetrical about the nucleus (see Chap. IV).
This is part of the explanation of the increased binding of d electrons
at the expense of s electrons at Cr and Mo in the first and second long
periods (see Fig. 14.18). An s electron removed from a spherically
symmetrical shell s2 leaves a spherically symmetrical distribution.

Problems

1. Compute and plot the Zeeman patterns (intervals and relative intensities)
for the transitions ’^G^ — 7F6, 7U6 — 7Fe, 7Cr6 — 7F6, and *Pg — 8aSj.
2. Starting with the very strong field quantum* numbers, derive the terms arising
from the electron configurations 3d4p5p and 3dHp (see Secs. 15.5 and 13.11).
3. Using the shorthand method outlined in Sec. 15.6, calculate the spectral terms
arising from (o) four equivalent d electrons, (6) two equivalent/electrons.
 CHAPTER XVI

X-RAY SPECTRA

The story of how Rontgen1 while experimenting with a Crookes

vacuum tube in 1895 discovered a new kind of radiation now known as
x-rays, or Rontgen-rays, is well known. So great was the import of
Rontgen’s discovery that within 3 months x-rays were being used in
surgical operations in Vienna. This, along with other practical uses to
be made of a single scientific discovery within so short a time, is a good
example of the r61e played by modern science in the rapid advancement
of civilization. In this introductory treatment of atomic spectra we
are not so much concerned with the methods used in obtaining x-rays
or with the uses to be made of them as with the atomic processes giving
rise to x radiation2 of which a brief account is here given.
16.1. The Nature of X-rays.—Believing that a crystal is made up
of regularly spaced atoms, Laue in 1912 suggested the possibility of
using a crystal as a diffraction grating. This idea was verified, at
Laue’s suggestion, by Friedrich and Knipping.3 By passing a narrow
beam of x-rays through a crystal and then on to a photographic plate,
a characteristic diffraction pattern, composed chiefly of small sym¬
metrically placed spots, is observed (see Fig. 16.1). These spots are
called Laue spots and the entire pattern a Laue pattern. This experiment
proved for the first time the wave nature of x-rays, although Haga
and Wind4 5 had previously shown that a beam of x-rays passing through
a V-shaped split was broadened more at the narrow end than would
be expected from a corpuscular theory.
The two Braggs6 next succeeded in showing that x-ray diffraction
by a crystal could be treated as specular reflection from the crystal
planes, the layers of regularly spaced atoms acting as reflecting surfaces.
Knowing the distance d between successive layers of atoms in the crystal

1 Rontgen, Sitz-ber. Phys.-med. Ges., Wurzburg, 1895; English translation,

Electrician, Jan. 24, 1896, and April 24, 1896.
2 For a more complete treatment of the general subject of x-ray spectra and their
production the reader is referred to M. Siegbahn, “ Spektroskopie der Rontgen-
strahlcn,” 2d ed., 1931.
3 Laue, M., and W. Friedrich and P. Knipping, Silz-ber. Bayer. Akad. Wise.,
Math.-phys. Klasse, p. 303, 1912; Ann. d. Phys., 41, 971, 1913.
4 Haga, H., and C. H. Wind, Koninklijke Akad. Wctenschappin Amsterdam,
Versl., 11, 350, 1902.
5 Bragg, W. H., and W. L. Bragg, “X-rays and Crystal Structure.”
299
300 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

and the equal angles of incidence and reflection 6, the x-ray wave-length
X is given by the well-known Bragg equation

nX = 2d sin 6. (16.1)

Within the last decade Walter,1 Backlin,2 Larsson,3 Kellstrom,4 and

others have obtained single-slit diffraction patterns of x-rays by sending

Fig. 16.1.—Laue pattern.,

Fig. 2a. Fig. 26.

Fig. 16.2a.—Single-slit diffraction, 8.33 A. (After Larsson.)
Fig. 16.26.—Diffraction at a thin wire, 8.33 A. (After Kellstr&m.)

these extremely short waves through a narrow slit and then on to a

photographic plate (see Fig. 16.2a). Interference fringes formed by

1 Walter, B., Ann. d. Phys., 74, 661, 1924; 75, 189, 1924.
2 Backlin, E., Dissertation, Uppsala Univ. Arsskr., 1928.
* Larsson, A., Dissertation, Uppsala Univ. Arsskr., 1929.
4 Kellstrom, G., Dissertation, Uppsala Univ. Arsskr., 1932.
Sec. 16.1] X-RAY SPECTRA 301
the reflection of x-rays at a grazing angle from plane polished mirrors'
(Lloyd mirror experiment) have also been photographed by Linnik1
and Kellstrom (see Fig. 16.3a).
Bergen Davis and Slack,2 * and Siegbahn,8 have been able to refract
x-rays by passing them through prisms of various materials, and Comp¬
ton and Doan,4 * Siegbahn, and others have diffracted x-rays from
mechanically ruled gratings. It is with ruled gratings that Osgood,*
Thibaud,6 and Siegbahn and Magnusson7 and others have crossed
the last gap in the electromagnetic wave chart by photographing and

Fig. 3a. Fia. 36.

Fig. 16.3a.—Interference, Lloyd’s mirror experiment, 8.33 A. (After KellstrOm.)
Fig. 16.36.—Diffraction at a straight edge, 8.33 A. (After KdUtrbm.)

measuring the wave-lengths of spectrum lines in the region of 10 to

200 A.
Before the wave nature of x radiation was known with certainty,
Barkla and Sadler8 had shown by the absorption of x-rays in aluminum
that the penetrating radiation consisted of two groups of rays differing
in penetrating power superposed on a continuous background of radiation.
The more penetrating or “hard radiation” is called K radiation and the
less penetrating or “soft radiation” is called L radiation.
Photographs of x-ray spectra diffracted by crystals and the measure¬
ment of their wave-lengths now show that the K radiation characteristic
of a given element consists of short wave-lengths and the L radiation
of longer wave-lengths. From the heavier elements in the periodic
table “very soft radiations” can now be observed and their wave-lengths
measured. These are called M, N, and 0 radiations.

1 Linnik, W., Zeiis. f. Phys., 66, 107, 1930. For other photographs of single-slit
diffraction patterns Bee M. Siegbahn, “Spektroskopie der Rontgenstrahlen,” 2d ed.
1931.
* Davis Bergen, and C. M. Slack, Phys. Rev., 26, 881, 1926.
* Siegbahn, M., Jour. Phys., 6, 228, 1925.
4 Compton, A. H., and R. L. Doan, Proc. Nat. Acad. Set., 11, 598, 1925.
* Osgood, T. H., Nature, 119, 817, 1927; Phys. Rev., 30, 567, 1927.
* Thibaud, J., Jour, de phys. et radium, 8, 13, 1927; 8, 447, 1927; Phys. Zeiis. 29,
241, 1928.
1 Siegbahn, M., and T. Magnusson, Zeiis. f. Phys., 62, 435, 1930.
* Barkla, C. G., and C. A. Sadler, Phil. Mag., 17, 739, 1909.
302 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI
16.2. X-ray Emission Spectra and the Moseley Law.—In 1913
Moseley,1 from a systematic study of the K radiation of Ca 20, Sc
21, Ti 22, V 23, Cr 24, Mn 25, Fe 26, Co 27, Ni 28, Cu 29, and Zn 30,

i
Angstrom Units
0 1000 2000 3QG0,
X-Umts
Fig. 16.4.—Wave-lengths of K-series x-ray lines. (After Moseley.)

announced what is known as the Moseley law. In brief this law may be
stated as follows: The frequency of each corresponding x-ray line is
approximately proportional to the square of the atomic number of the

10 20 30 40 5j) ^60 70 60 90

Fig. 16.4a.—Moseley law for iT-series x-ray lines.

emitting element. A plot of the K lines observed and measured by

Moseley is given in Fig. 16.4 to show that for each element investigated
the K radiation consists of two strong spectrum lines. The gradual
1 Moseley, H. G. J., Phil. Mag., 26. 1024, 1913; 27, 703, 1914.
Sec. 16.2] X-RAY SPECTRA 303

change in separation of, and the stepwise shift of, corresponding wave¬
lengths in going from one element to the next are unmistakable in their
meaning. For the first time one could say with certainty that the order
of the chemical elements is as shown in Fig. 16.4, and also that the
elements in the periodic table are constructed with extreme regularity.
Prior to Moseley's discovery, cobalt and nickel, because of their atomic

Fia. 16.5.—Photographs of x-ray L series lines. (After Richtmyer.)

weights 58.9 and 58.7, respectively, were listed in the order nickel, cobalt.
Moseley's x-ray photographs show without doubt that so far as structure
is concerned, the order is cobalt, nickel. Optical spectra are also in com¬
plete agreement with this result. In Moseley's work the x-rays were
photographed in the second and third orders as diffracted from a crystal
of potassium ferrocyanide (grating constant d = 8.408 A).
A continuation of Moseley's early work by other investigators
has shown that in the heavier elements each of the Ka and K$ lines
(see Fig. 16.4) is in itself a close doublet. These four lines are usually
designated Ka, KatJ Kffi, and K$z. X-ray spectrograms of all available
elements now show that Moseley's law continues throughout the periodic
304 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI
table. From a graph similar to the one shown in Fig. 16.4a Moseley
arrived at the expression
Pcm-I = KR(Z - a)\ (16.2)
where R is the Rydberg constant = 109737 cm”1, and <r and K are
constants. For Ka, a = 1 and K = $. By analogy with hydrogen-like
atoms and Balmer’s formula [see Eq. (1.24)], Moseley wrote

V = R(Z - iy(± - i) (16.3)

In Moseley’s first work y/v was plotted against Z. The division of v

by the quantity R to form a dimensionless constant is due to Sommerfeld.
Like K radiation, the L radiation discovered by Barkla is also com¬
posed of a group of spectral lines. Photographs of some of the L lines
of a group of elements are shown in Fig. 16.5 and their wave-lengths
are given in the following table. These are the strong lines of the
L series.
Table 16.1.—Wave-lengths of L Series X-ray Lines
(X in X units)1

L line Cb 41 Mo 42 Ru 44 Rh 45 Pd 46 Ag 47 Cd 48 In 49 Sn 60

«i 5712 5395 4836 4588 4359 4145 3948 3764 3592

at 5718 5401 4844 4596 4367 4154 3956 3772 3601
fit 5480 5166 4611 4364 4137 3927 3730 3548 3378
fit 5226 4909 4362 4122 3901 3694 3506 3331 3168
fit 5297 5005 4476 4245 4026 3824 3636 3462 3299
fit 5331 5041 4513 4280 4062 3861 3674 3499 3336

11000 x units — 1 Angstrom unit — 10-10 meter.

The characteristic grouping of the observed lines for each element

and the regular stepwise shift in the wave-length with atomic number
suggest that the Moseley law also holds for L series lines. That it does
is clearly seen in Fig. 16.6.
In the photographs of Fig. 16.5 it is observed that Lat is in each
case a faint line on the long wave-length side of La. In addition it is
to be observed that Lpz and LpA are strong lines in Sn 50, In 49, Cd 48,
and Ag 47, whereas in the other elements they are faint. This is in
agreement with the general observation (see Fig. 16.6), that in going
to higher atomic numbers the number of observed lines increases. In
the heaviest elements the L series is composed of as many as 20 lines.
In plotting a graph similar to that given in Fig. 16.6, Moseley obtained
the same relation for L radiation that he had for K radiation (see Fig.
16.2). For the L& line, a = 7.4 and K = -fo, which, by analogy with
hydrogen-like orbits, is written
Sec. 16.2] X-RAY SPECTRA
306 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

v = R(Z — 7.4)’ ■ (16.4)

In 1916 Siegbahn1 discovered in the heavier elements a so-called

M radiation and 6 years later Dolejsek2 discovered a still softer radiation
called N radiation. Photographs3 of M series lines from a group of
six elements are shown in Fig. 16.7. Like the K and L series lines,
the M series lines follow nearly straight lines on a Moseley diagram
and may be represented by Eq. 16.2. It should be pointed out in passing
that the M and L radiations arising from the heavy elements are harder
(i.e., of shorter wave-length) than the K radiation of the light elements.

Direct Silver Bromine

Radiation Edge Edge
Fia. 16.8.—X-ray absorption edges of silver and bromine. (After de Broglie.)

16.3. Absorption Spectra.—In the first x-ray spectrograms taken

by de Broglie in 1916 a sudden change in blackening was always noted at
two discrete positions on the photographic plate. It is now known
that these abrupt changes are due to the selective absorption of the
silver and bromine in the photographic emulsion and that the wave¬
lengths of the absorption edges are independent of the x-ray source
used (see Fig. 16.8). Further investigation shows that, when a thin layer
of some metal like Cu, Ag, Fe, or Sn, is placed just in front of the photo¬
graphic plate, other absorption edges appear as characteristic of the
respective element used as an absorber. With successive elements
in the periodic table used as absorbing screens a regular stepwise shift
of the characteristic absorption edges is observed. This may be seen
in Fig. 16.9 for the elements Ag, Cd, In, Sn, Sb, Te, and I. The differ¬
ence between the silver edge, which shows up on the right at the same
point in each spectrum, and the other edges at the left is due to the silver
absorption in the photographic plate itself (causing increased darkening)
and the absorption of the radiation before reaching the plate, respectively.

1 Siegbahn, M., Verh. d. Deutsch. Phys. Ges., 18, 278, 1916; Compt. rend., 162,
787, 1916.
Zeits. f. Phys., 10, 129, 236, 1922.
1 Dolejsek, V.,
’These photographs are reproduced from original negatives kindly loaned by
F. R. Hirsch.
Sec. 16.3] X-RAY SPECTRA 307

A search for absorption edges over the large range of x-ray wave-'
lengths reveals in the heavier elements, in particular, several different
edges for each element. For a given element the absorption edge
occurring at the shortest wave-length is called the K absorption limit.
Beyond the K limit to longer wave-lengths three relatively close absorp¬
tion edges have been photographed for the elements starting at about
Rb 37. These three edges are called the L limits, Ltf Ln, and Lm, in
the order of their wave-lengths.1

Ag

Ag

Co
In

Sn

Sb

Te

I Ag
Fig. 16.9.—X-ray absorption edges of silver, cadmium, indium, tin, antimony, tellurium,
and iodine. (After Erode and Burmann.)

At still longer wave-lengths in the heaviest elements a group of five

absorption edges has been observed. These are called the M limits,
Mi, Mn, Mmy Miv, and M\.
Absorption limits are of very great importance in x-ray spectra,
for they furnish the clue to the origin of x-rays within the atom. A
very simple and plausible theory of this was first given by Kossel.2
A sharp absorption edge for a given element indicates that incident
x-rays having a frequency greater than a certain critical frequency
ve are largely absorbed, whereas those of lower frequency pass on through
the absorption screen with little diminution in intensity. The absorbed
energy hvC) or greater, taken from the x-ray beam is used up in ejecting
an electron. Such ejected electrons are called photoelectrons. The
several K, L, and M limits observed for each element correspond there¬
fore to the energies necessary to remove electrons from the different
shells and subshells in the atom. The absorption limits thus provide
a direct means by which the energy levels of an atom may be determined.
This is the subject of the following section.

1 For references concerning L absorption limits see M. Siegbahn,c “Spektroskopie

der Rdntgenstrahlen," 2d ed., p. 271, 1931.
, ,
* Kossel, W., Verh. d. Deutsch. Phys. Ges., 16 898, 963, 1914; 18 339, 1916;
Zeits. /. Phys., 1, 119, 1920.
308 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

16.4. Energy Levels.—In the preceding chapters of this book we

have been dealing with so-called optical spectra, i.e., with radiation
arising from electron transitions in the outermost part of the atom.
In this chapter on x-rays, however, we are dealing with high-frequency
radiations known to arise from electron transitions between the inner
electron shells of the atom.
The condition for the excitation of an atom prior to its emission
of x radiation consists in the complete removal of an electron from
one of the completed inner subshells of electrons. This removal of
an electron may be accomplished by the inelastic impact of a high-speed
particle (an electron, proton, or atom), or by the absorption of a photon
of energy hve or greater. With an electron removed from an inner shell
of the atom, an electron from one of the neighboring outer shells may
jump into the unoccupied orbit, giving rise simultaneously to x radiation
of energy hv given by the energy difference between the two stationary
states. Another electron from a shell still farther out may now jump
into the orbit just vacated giving rise to radiation of another frequency.
This process of emission will finally cease when the positively charged
ion captures an electron. With this explanation we see that the absorp¬
tion limits, briefly discussed in the preceding section, should theoretically
give the x-ray energy levels of the atom.
In order to formulate some comprehensive picture of the atomic
processes just described and the energies involved in x-ray emission
and absorption, let us consider a specific example, viz., the atomic system
of cadmium, Z = 48. Cadmium is here chosen for the reason that,
like He, Ne, A, Kr, Rn, Be, Mg, Ca, Sr, Ba, Zn, and Hg, it contains
in the normal state only completed subshells of electrons. The complete
electron configuration is given in Table 16.2.
Table 16.2.—Electron Configuration of Neutral Cadmium

Shells K L M N 0

Subshells 8 Is2 2s2 3s2 4«2 6a2

V 2 3 p* 4pa
d 3d10 4d10

The x-ray notation K, L, M, N, . . . represents the main shells

of electrons for which the total quantum number n = 1, 2, 3, 4, • • • ,
respectively. In cadmium there are five completed s subshells, three
completed p subsheils, and two completed d subshells. These shells
and subshells are represented schematically in Fig. 16.10. In terms of a
quantum-mechanical model, similar to that given for rubidium in Fig.
7.1, each electron orbit in cadmium interpenetrates all other electron
orbits. The circles may therefore be thought of as a sort of time average
of the position of the electron in the atom (see Fig. 4.6).
Bsc. 16.4] X-RAY SPECTRA 309
Of all the 48 extranuclear electrons in cadmium the two Is electrons
ere the most tightly bound to the atom, since for the greater part of the
time they are inside all the remaining 46 electrons and in the field of the
nucleus of charge 48e. Second in binding energy come the 2s electrons,
screened partially from the nucleus by the two Is electrons. Third in
binding energy come the six 2p electrons screened quite effectively from
the nucleus by the four inner electrons Is2 and 2s2. Following the
L electrons the order of binding, as seen from Fig. 16.10, is 3s, 3p, 3d;

Fiq. 16.10.—Cadmium shells and subshells of electrons.

4s, 4p, 4d; and 5s, respectively. Since these binding energies are meas¬
ured by the absorption edges, discussed in Sec. 16.3, an energy level
diagram is obtained by plotting just the frequencies of the absorption
limits.
At the bottom of Fig. 16.11 the normal state of the cadmium atom
with all of its electrons is represented by a state, where S *= 0, L = 0,
and J = 0. The removal from the atom of one of the 5s electrons
raises the atom to the lowest state, where £ = J, L = 0, and J —
This ionized state is the single limit of the regular optical series of singlets
and triplets (see Fig. 11.6). The notation 5s_1 means that a 5s electron
has been removed from the neutral atom. This leaves the atom with an
incompleted subshell, a 5s electron.
The removal of a Is electron from the neutral atom, on the other
hand, raises the atom to the highest energy state K shown at the top of
the figure. The removed electron is represented by Is"1. With but
one electron remaining in the Is subshell this energy state is represented
by 2Si, in the notation used in optical spectra. The term values given
at the left in the figure show that the binding energy of a Is electron is
about 3000 times that of a 5s valence electron.
310 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

In a similar fashion the removal of a 2s electron, designated 2s~1,

from the L shell raises the atom to the state Li. In this state the atom
with one 2s electron in an incompleted subshell is again in a 2S state.
Now the removal of a 2p electron, 2p_1, from the L shell leaves an
incomplete subshell 2p5, which, as we have seen in Sec. 15.0, gives rise
to an inverted 2P term,1 *P| and 2P\. These correspond to the two
absorption limits Lm and Lu, respectively.

2.16-10* K M- 2 1
l S^- s - ts-V
[2 5510*1

K'Series

3.25107 Lj- iMm. -22S^ 2s 2sj

a i
3.0I-107Ln-
28610' Lm- ?p5 v
[632-I07)

L-Series

63 106 M,
51 10°
. (Li J. 3s
3p?
33 10* My'
fe.81107*
IS
M-Seriei
3d9

9.0 10s |K*4 4s-'

-M&'* V
-M 4dJ
'•O'O’lKf N-Senes

7.3 104 0 •55Sv2-5s--5S-J-

Normal opfical levels
. Normal^
state Is2 2sV3sV3d"4^4dK)5s2
' ’So
Fig. 16.11.—X-ray energy level diagram of cadmium.

Again the removal of a 3s, 3p, or 3d electron leaves incomplete

subshells 3s, 3p5, or 3d9, giving rise to a 2S term, two 2P terms, or two
2D terms, respectively. In optical notation these are 2£j, 2P\, 2Pg,
2Z>|, and 2D|, and in x-ray notation they are Mi, Mu, Mm, MiV, and My.
A continuation of this process for the N and O shells leads to the remaining
doublet levels shown in the figure. It should be pointed out that in the
heavier elements of the periodic table 4/ electrons are present and give
rise to two more N levels, Nw and JVvn (2Ff and 2F{). With the exception
of a very close structure, not yet mentioned, S terms in x-ray spectra

1 Any subshell of electrons lacking but one electron to complete it gives rise to an
inverted doublet level having an L value equal to the l value of the missing electron.
Sec. 16.5] X-RAY SPECTRA 311
as in optical spectra are always single; other levels P, D, F, . . . are
double.
Attempts oy Moseley and his successors to find a simple law governing
the terms, or, what is the same thing, the absorption limits v, as a function
of the atomic number Z, has led to the so-called Moseley diagram shown

in Fig. 16.12.1 The regularities as well as the irregularities to be noted

in this figure can, for the most part, be explained. This explanation will
be undertaken in the following sections.
16.5. Selection and Intensity Rules.—The selection rules derived
from theoretical considerations and found to hold in optical spectra
also apply in x-ray spectra. These rules for the various quantum
numbers n, L, and J, are2

1 Bohr, N., and D. Coster, Zeits. f. Phys., 12, 342, 1923.

2 These selection rules apply only to atoms of the cadmium type, i.e., to atoms in
which the energy levels arise from the absence or presence of one electron in a subshell.
Although they may be applied to the chief x-ray lines of all atoms, they are only a
special case of a more general set of selection rules to be given in Sec. 16.10.
312 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

An arbitrary,
AL = +1 or -1, (16.5)
AJ = +1, 0, or —1.

Transitions giving rise to the K and L series lines in cadmium are

shown in Fig. 16.11. Photographs of some of the strongest L and M
lines are given in Figs. 16.5 and 16.7 for a few of the elements. At the
right in these figures the respective transitions are designated in x-ray
notation and also in optical notation. It should be noted that not all
transitions allowed by the above rules have been observed. In particular
the transitions Li to Lu> and Li to Lm, although looked for by many
investigators, have not been observed with certainty. These are the
transitions for which An = 0. In optical spectra of the alkali metals
such transitions give rise to the strongest lines in the neutral spectrum.
In sodium the transitions Mi to Mu and Mi to Mm give rise to the first
doublet of the principal series. A plausible explanation of the absence
of these lines with an appreciable intensity in x-rays is that in the intensity
formulas, derived from theoretical considerations by Einstein,1 there
occurs a factor v*. For the transitions An = 0 this vA factor is extremely
small, as compared with other allowed transitions starting from the same
initial states, and should therefore give rise to very weak lines. It must
be mentioned that, as in optical spectra, a few cases are known where
forbidden transitions are observed.2
It should be noted that when an electron jumps from a 3p to a 2s state
(for example, Mm —* La), the atom goes from the higher energy state Li
to the lower energy state Mm. For this reason the transitions on all
energy level diagrams are shown as double arrows. The arrows pointing
up represent electron transitions of the type given in the figure with a
minus sign prefixed, and the arrows pointing down represent energy
changes in the emission of radiation.
The same intensity rules derived and found to hold for optical spectra
are also found to hold in x-ray spectra. For observed lines not greatly
different in frequency the v4 term in the intensity formulas may be
neglected and the intensity rules applied to multiplets in Sec. 12.15
may be used. According to these formulas the theoretical intensities
for the Kat and Kai lines (in optical notation 2S* — 2Pj and 2Sj — 2P\)
have the ratio 2:1. Accurate intensity measurements on some 24 ele¬
ments are given in Table 16.3. The agreement between observation
and theory for these lines is excellent.

1 Einstein, A., Verh. d. Deutsch. Phys. Ges., 18, 318, 1916; Phys. Zeits., 18, 121,
1917; see also Dirac, Proc. Roy. Soc., 114, 243, 1927. There is now some question
whether the ** factor belongs in the intensity formulae for spectrum lines.
1 Coster, D., Phil. Mag., 48, 1070, 1922; see also Kaufman, S., Phys. Rev., 40,
116, 1932.
 Sbc. 16.5J X-RAY SPECTRA 313
Table 16.3.—Observed Relative Intensities for the K Series Lines
{After Meyer)1

Element Kax Kat Element Kai Kat Element Kai Kat

y 23 100 52.1 Ga 31 100 50.6 Zr 40 100 49.1

Cr 24 100 50.6 Ge 32 100 50.7 Cb 41 100 49 7
Mn 25 100 54.9 As 33 100 49.2 Mo 42 100 50.6
Fe 26 100 49.1 Se 34 100 50.3 Ru 44 100 51.1
Co 27 100 53 2 Br 35 100 50 9 Rh 45 100 51.2
Ni 28 100 47.6 Rb 37 100 49 3 Pd 46 100 52.3
Cu 29 100 46 0 Sr 38 100 48 6 Ag 47 100 51 7
Zn 30 roo 48 9 Yt 39 100 50 0 Cd 48 100 53 8

1 Meyer, H. T., Wise . . . . ,

Vcrh a d Siemens Konzern 7, 108, 1929; aee also Williams, J. H., Phye .
Rev., 44, 146, 1933

As a second example of the intensity rules consider the strongest

L series lines Laiy Lai, and Lfil.
These three lines arise from the transi¬
tions between the doublet levels 2p6,2Pg.j and 3dVD|.|. The predicted
relative intensities are

’Pi 'Pi

9 0 My
!0i 1 5 Miy

Lni Ln

The observed intensities for eight elements are given in Table 16.4.

Table 16.4.—Observed Relative Intensities for the L Series Lines

{After J&nsson1 and Allison*)
—
Element fii Element otl 02 01

U* 92 9 1 4.4 Ag* 47 9 1 5.3

Th» 90 9 1.1 5.6 Pd> 46 9 1 5.3
Pt‘ 78 9 1 4.6 Rh1 45 *9 1 2.8
W» 74 9 1 4.7 Mo1 42 9 1 2.8

. ., , ,
» JONsaoN, A., Zeits /. Phye 36 426, 1926; 46 383, 1928.
. .,
* Allison, S. K, Phyv Rev 80, 245, 1927; 81, 1, 1928.

Peculiarly enough when vx corrections are made for these lines they
are far from being in agreement with theory. Bethe1 explains this on the
basis of the fact that the measured intensity is not equal to the transition
probability but to the product of the life time of the initial state and the
• transition probability. The measurements thus indicate simply that the
life time of Li (due, e.g.y to a greater probability of autoionization) is

1 Bethe, H., Handbuch der Physik., Vol. 24, No. 1, p. 468, 1933.
314 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

shorter than that of Lu,m. It may be seen that the Lm lines in Fig. 16.5,
where resolved, are quite faint as compared with the strongest line Lai.
The Lp lines are taken from different plates and may not justly be
compared.
16.6. Fine Structure of X-ray Levels.—The three L levels, the five
M levels, the seven N levels, etc., may be referred to as the fine structure
of x-rays. Since a fairly satisfactory account of this fine structure may
be given, a treatment of the splitting of each main energy level into
2n — 1 sublevels is made the object of this and the three succeeding
sections.
It has already been pointed out that the L electrons move in an approx¬
imately Coulomb field, the two K electrons acting very much as if they
had coalesced with the nucleus. This screening of the nucleus by the
K electrons reduces the nuclear charge by about two, so that, neglecting
for the moment the fine structure, the energy should be given by the
hydrogen-like formula

T = ^ 7 g)-> (16.6)
n2 n2

where R is the Rydberg constant = 109737 cm-1, ZM the effective

nuclear charge, Z the atomic nymber, and a the screening constant
(equal to about two).
In a similar manner one can expect the M series to be given by a
formula of the same type. In this instance the M electrons are screened
from the nucleus by the eight L and two K electrons, thus reducing the
nuclear charge by roughly ten.
In addition to this internal screening effect there is also a correction
known as the external screening effect. This external screening is perhaps
most easily pictured by imagining any given outer shell of Q electrons
as distributed uniformly over the surface of a sphere of radius r. The
potential energy of an electron within this charged shell will thus be
Qe2/r. The sum of such energy terms taken over all external subshells
must be subtracted from the energy of the hydrogen-like orbit within.
Referring to Fig. 16.10, it is readily seen that the energy necessary to
earry an electron from the L shell through the other shells to infinity
is less than the energy required to do so when the outer electrons are
absent. In general this external screening is small compared with the
main energy, and, since the quantum-mechanical model has all orbits
interpenetrating all other orbits, it is customary to include the external
with the internal screening by means of the one screening constant <r.
This single constant <r will therefore represent collectively both types of
screening in the following treatment.
In anticipation of the fact that the x-ray energy levels will be almost
hydrogen-like, it is now an easy step from Sommerfeld’s fine-structure
Sec. 16.61 X-RAY SPECTRA 315

formula for hydrogen [see Eq. (9.6)] to his x-ray formula. Replacing
Z by Z — a, or Z — 8,

R(Z - <r)2 - sWn _ 3\

w2 n4 \fc 4/
RaA(Z — s)6/ In* 3n2 3n 5\
+ ^ \4p + 4F2 2fc + 8/
#a6(Z - 8)*(ln& 3n4 In* 15n2 15n 35\ , ,
+ n8 \8&6 + 8fc4 + 8fc3 8fc2 8fc 64/
(16.7)

where a2 = 4ir2e*/h2c2 — 5.30 X 10-6 is the square of Sommerfeld’s

fine-structure constant, R is the Rydberg constant, a and s are screening
constants, n is the total quantum number, and k is Sommerfeld^ original
azimuthal quantum number; fc = 1,2,3, • • • for s, p, d, • • • electrons,
respectively.
The application of this equation to observed x-ray energy levels,
to be given in the three following sections, shows that the screening
constants a and s are different from each other even though, a priori, they
might be expected to be the same. This equation, without the screening
constant, was first derived by Sommerfeld1 from the orbital model
of hydrogen. Starting with Dirac’s quantum-mechanical theory of
the electron (Sec. 9.9), Gordon has arrived at exactly the same formula.
With the screening constants included, the first term in the equation
gives the main energy of each electron shell n, and the second, third,
fourth, etc., terms give the fine-structure corrections. In Sommerfeld’s
early theory these latter corrections were due to the relativity change
in mass of the electron, whereas, according to quantum-mechanical
derivation from Dirac’s theory, they are due jointly to what has been
called the electron spin-orbit interaction and to relativity. It is for this
reason that the fine-structure term differences, accounted for by the
second and succeeding terms of this equation, are called spin-relativity
doublets. In x-ray notation these are the doublet intervals Ln — Lm,
Mn — Mm, Miv — Afv, Nu — Nmf • • • , for which Sommerfeld
assigned quantum numbers k = 1 for Ln, Mn, Nu, • • • , k = 2 for
Lm, Mill, Afiv, Nil I, Niy, 0 • • , k = 3 for My, Ny, Nyi, Oy, Oyi, * ■ * I
etc. It is now known that these are not the so-called azimuthal quantum
numbers of the electron orbits but new quantum numbers given by
the Dirac electron theory of hydrogen-like orbits. These new quantum
numbers (see Sec. 9.9) are assigned the values shown on page 316.
In Gordon’s derivation of the fine-structure formula, j'2 takes the
place of Sommerfeld’s k2 so that \j'\ becomes the analogue of the old k .
One may therefore proceed to use Sommerfeld’s formula, as it is giver

1 Soioodbfbld, A., “ Atombau,” 6th ed., p. 282.

316 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

in Eq. (16.7), for with his values of k we shall obtain the quantum-
mechanical term values (see Sec. 9.9).

Table 16.5

Screening Spin-relativity Screening Spin-relativity

Doublet Doublet Doublet Doublet

*s, 2Pl 2P| 2I>! Id,

S = * * * * *
L = 0 1 1 2 2
J = 1 * f i 1
f = -1 +1 —2 +2 -3

The effect of the screening constants <r and s in shifting the energy
levels from those of hydrogen-like atoms to the observed x-ray levels
is illustrated in Fig. 16.13. At the left the Bohr term values are given

Hydrogen Reduced Observed

by
-like terms terms Tn = (16.8)
terms (1^

Introducing the screening constant a

to account for the internal and external
screening these levels are each split into
n sublevels:

r= 7---' (i6.9)
where <r is different for the different
sublevels s,p,d, ... . Adding finally

with the exception of S levels, splits

into two sublevels. It should be noted

are often used in front of 2S, 2P, 2D,

etc., to distinguish them from each
Fig. 16.13.—The splitting up of other. The intervals Li — Ln, Mi —
energy levels according to Sommerfeld’s Mu, Mm — Mwj • • • are generally
formula.
called Screening doublets. In optical
notation (see Table 16.5 above) these are the levels having the same n,
S, and J values but different L values.
16.7. Spin-relativity Doublets (Regular Doublets).—The term spin-
relativity doublet refers in general to each pair of energy levels having the
same n, S, and L values but different J values. In x-ray notation
these are the intervals Ln — LlU, Mn — Mm, Mvr — My, Nn — NUi,
Sbc. 16.7J X-RAY SPECTRA 317

Niv — Ny, Nvt — Nyiif Ou — Om, Oiv — Ov, etc., which prior to the
quantum mechanics were called regular doublets. The intervals between
each of these doublets in the various elements in the periodic table are
known accurately from observed differences between a number of x-ray
spectrum lines (see Figs. 16.5 and 16.7).
The substitution in Sommerfeld’s formula of the known quantum
numDers n and k for a given doublet leads to the so-called regular doublet
law. Substituting in Eq. (16.7), one obtains two equations for each
doublet. For the L doublet (22P* — 22P8),

_ R(Z - <r)2 Z - s)4(5

Rol\Z s)U5 ,21 or
Tlh
22 24 \4
, 429 .. ,
(16.10)
+ 64 2^Z ~ 8) + }’
R{Z-cY Ra\Z-Sy{ l.lo' ,,
^Lm 22 + -2*-\4 + 8 2* * ~ S)

+ ^|>-*)4 + (16.11)
}•
Now if Ay represents the difference between tnese two equations,
then

Ay *=3|fLiL-ji + £(z _ (16.12)

Although only the first term of the analogous Eq. (9.6) is required
in describing the hydrogen fine structure, Sommerfeld1 has shown that
the first term of this equation does not suffice to determine s and that
higher terms are necessary.
From the observed x-ray lines, shown in Fig. 16.5 and tabulated in
Table 16.1, the following screening constants (Table 16.6) are computed:

Table 16.6.—Screening Constants s for the Spin-relativity Doublet Ln — Lui

(After Sommerfeld)

Element Z Av/R Z - s 8

Cb 41 ■m 37.53 3.47
Mo 42 38.50 3.50
Ru 44 9.49 40.54 3.46
Rh 45 10.48 41.53 3.47
Pd 46 11.57 42.55 3.45
Ag 47 12.69 43.52 3.48
Cd 48 13.96 44.52 3.48
In 49 15.29 45.50 3.50
Sn 50 16.72 46.51 3.49

1 Sommerfeld, A., “Atombau,” 5th ed., p. 300.

31S INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

It is observed that 8 is practically a constant independent of Z.

For some 40 elements from Z * 41 to Z * 92 Sommerfeld obtained an
average value
8 « 3.50 ± 0.05. (16.13)

In a similar fashion Eq. (16.7) gives for the Mu — Mm doublet

(32P* - 32P,),
Ra2(Z — s)4(3 279a2 13059a4
34 12 + 288
(Z - a)2 +
20736 K
-*)4 +
(16.14)

From the observed doublet separations of the elements Z = 37 to

Z = 92, Wentzel1 has shown that a is again constant throughout (see
Table 16.7) and has the average value

s = 8.5 ± 0.4. (16.15)

Table 16.7.—Screening Constants s for the Spin-relativity Doublet Mn — Mm
{After Wentzel)
Cb 41 Mo 42 Ru 44 Rh 45 Pd 46 Ag 47 Cd 48 In 49 Sn 50
8.50 8.56 8 52 8.50 8.52 8.58 8.28 8.56 8.48

Similar calculations carried out for other spin-relativity doublets

have yielded the following average values for a.

Table 16.8.—Screening Constants s for Spin-relativity Doublets

{After Sommerfeld and Wentzel)

2 *Si 2- 2aPj 3aS* 3aPj - 3aPj 3’Dj - 3*Z>, 4’S, 4aPj - 4aPj
Li Lu — Lm Mi Mn M in M iv My Ni Nu Nm
2 0 3 50 6 8 8.5 13 14 17

The method used in determining the screening constants for the

single Li, Mi, and Ni levels is given in Sec. 16.8. Attempts to calculate
a from theory alone have been made by Pauling.2 By a first-order
perturbation treatment of the problem Pauling has been able to calculate
values which, in general, are in quite good agreement with those given
above.
16.8. The Regular-doublet Law.—The regular-doublet law, well-
known in x-ray spectra, refers to the apparent law governing the spin-
relativity doublets Lu — Lm, Mn — Mm, Miy — My, Nu — Nm, etc.
Neglecting higher terms in Eq. (16.12), or Eq. (16.14), the law may
be expressed by the first term of the equation as follows: The doublet
separation, Av in wave numbers, for any spirt-relativity doublet is approxi¬
mately proportional to the fourth power of the effective atomic number

1 See Wentzel, G., Zeits. f. Phys., 16, 46, 1923.

* Pauling, L., Zeits. f. Phys., 40, 344, 1927.
SBC. 16.9] X-RAY SPECTRA 319

Z — s. As an example of the validity of this law consider the Ln — Lm

doublet. For this doublet (see Table 16.8) $ * 3.5, and Ay, to a first
approximation, is given by

Av = ~{Z - 3.5)4. (16.16)

Now the values of Ay for some 70 elements are known quite accurately
from the average of several observed x-ray doublet intervals Kai — Kat,
Lfa — La„ L„ — L\, Lfit — L«, and the absorption limits Ln — Lm.
The average values of Ay are plotted in Fig. 16.14 with four different sets
of ordinates y/Av/R, y/Av/R, yfAv/R, and y/Av/R. The y/Av/R

Fiq. 16.14.—The regular-doublet law for L doublets in x-ray spectra.

curve, with an intercept at 3.5, shows that the fourth power law [Eq.
(16.16)] gives the best agreement.
16.9. Screening Doublets and the Irregular-doublet Law.—Making a
systematic study of L series lines, Hertz1 in 1920 announced what is known
as the irregular-doublet law. This law applies to those pairs of energy
levels having the same n, S, and J values but different L values and
states that the difference between the square roots of the term values of a
given doublet is a constant independent of the atomic number Z. This
law found by Hertz to hold for the Li and Ln levels of the six elements
Cs, Ba, La, Ce, Pr, and Nd is now known to hold for all doublets la — Ln,
Mi — Mu, Mm — Miv, Ni — Nn9 Nin — Nw, and Ny — Nyi (see
Fig. 16.12). That the law is only approximately true is shown by the
values of Ay/T/R given in the last column of Table 16.9. A general
widening of the interval with increasing Z is observed.
In formulating the irregular-doublet law Hertz began with the
equation
1 Hertz, G., Zette. /. Phys.} S, 19, 1920.
320 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

r _*(*-*)».
(16.17)
n2
Transposing J? and taking the square root, this equation becomes

(16.18)

If now <r i and <r2 are the screening constants for the two levels of a doublet,
the difference is written

= —-— = — = constant. (16.19)

Referring to Fig. 16.16 the factor <n — a2 is seen to be practically

a constant for the two levels forming an irregular doublet. This is
Table 16.9.—The Irregular-doublet Law for the X-ray Levels Li and Lii

Element
y/T/R Vt/r A VrjR
(M (in)

Cs 55 20.538 19.873 0.665

Ba 66 21 022 20.355 0.667
La 57 21.514 20.837 0.677
Ce 68 21.984 21.310 0.674
Pr 59 22 456 21.785 0.671
Nd 60 22 939 22.260 0.679
Tb 65 25 349 24.664 0.685
Yb 70 27.830 27.120 0.710
Re 75 30.381 29.667 0.714
Hg 80 33.051 32 349 0.702
Th 90 38 844 38 055 0 789
U 92 40 053 39 256 0.797

the irregular-doublet law more recently referred to as the screening-

doublet law.
Sommerfeld has shown that a refinement of the irregular-doublet
law is brought about by the use of the so-called reduced terms [Eq. (16.9)]
rather than the observed terms (see Fig. 16.13). With values of the
screening constants s determined from the spin-relativity doublets
of the preceding section, the higher terms of Eq. (16.7) can be calculated
and the corrected, or reduced, terms

r = R(-?~ g)i (16.20)

evaluated. Dropping all but the first two terms in Eq. (16.7), and
substituting the value of T", we obtain

T, _ T _ R«*(Z - aWn _ 3\
(16.21)
1 “ 1 n4 \k i)
Sic. 16.0] X-RAY SPECTRA 321
J

Here T represents the observed term values given directly by the absorp¬
tion limits.

A Moseley diagram of reduced terms is shown in Fig. 16.15. For

given n the resultant curves are now quite parallel to each other. The
parallelism is more accurately shown by the following Table 16.10.
Table 16.10.—The Irregular-doublet Law for Reduced Terms
{After Sommerfeld)

Cs 55 La 57 Ce 58 Nd 60 Sa 62 Gd 64 Dy 66 Er 68

a^/W/r
2*S - 2*P 0.59 0 60 0.59 0 59 0 59 0.58 0.58 0 58

Ay/W/R
3‘S - 3*P 0.54 0.54 0.51 0.54 0.56 0 51 0.51 0.53 0.54 0.53

Ay/TjR
3*P - 3 *D 1.21 1.19 1.20 1.19 1.19 1.19 1.20 1.19 1.18 1.19
322 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

Although the values shown here are quite constant, some of the values,
for elements not given in the table, deviate slightly more from the
average value. It should be pointed out that there are no spin-relativity
doublets from which to calculate s and a for the 2S levels. Wentzel,1
however, making the assumption that the irregular-doublet law is valid
for the reduced terms, has adjusted the screening constants $ for these
levels until the curves for 22S, 3 2S,
and 42S (see Fig. 16.15) are parallel
NffNJv to the center curves as shown. These
NfNm values of S have been included in
Table 16.8.
In Fig. 16.15 the dashed lines
through the origin represent hydrogen-
Ml like atoms for which there is no
screening. The difference between
these ideal curves and the corres-
j-n Lm
Ll ‘ ponding reduced curves

multiplied by n, gives directly the

K screening constant a plotted in Fig.
16.16. The sudden change in the
slope of the M and N curves at approx-
Fig. 16.16.—Screening constants a. imately 7j — 57 and Z — 71 shows
{After Sommer/eld.) the increasec[ screening effect of the
4/ electrons added in the rare-earth elements. Their addition causes
an increase in the external screening discussed in Sec. 16.6. As Sommer-
feld2 shows, the corresponding increase in the screening constant for
a state of total quantum number n is directly proportional to n2 and
inversely proportional to the effective nuclear charge; hence the effect
is relatively slight for states of low quantum number, as is, indeed, borne
out by the figure. Similar screening effects are shown by the Moseley
diagram of Fig. 16.12 for the sequence of elements where 3d and 4d
electrons are added.3
The irregular-doublet law has an aspect other than that of the
parallelism of Li and Lu on a Moseley diagram, viz., the frequency
difference Av = TLl — TLji is very nearly a linear function of the atomic
number Z. Although the x-ray transitions Lu to Li have not yet been
observed, this law is of considerable importance in optical spectra where

1 Wentzel, G., Zeits. f. Phys., 16, 46, 1923.

2Sommeefeld, A., “Atombau,” 5th ed., p. 427.
3 These sequences of elements and the screening effect due to the addition of 3d
and 4d electrons have been treated theoretically by L. Pauling, Zeits. /. Phys.} 40,
344, 1927. See also L. Pauling and S. Goudsmit, “Structure of Line Spectra,” p. 187.
Sac. 16.10] X-RAY SPECTRA . 323

the corresponding transitions are observed. Let us start with Hertz’s

Eq. (18.17),

Av = AT = —J(Z - o-,)* - (Z - *,)*}• (16.22)

n*
Upon multiplying out, one obtains

| = 2^1){Z - = Ci(Z - c). (16.23)

Plots of Av/R for the intervals L i — Ln, Mi — Mu, Mm — Miv, Ni — Nn,

and Nin — Niy given in Fig. 16.17 show, at least over short ranges of
Z, how closely this law applies.
These curves are to be compared 50
with the <Tv law for the regular
doublet (Fig. 16.14). 40
16.10. A Predicted Multiplet
Structure in X-rays.—In Sec. 16.4
the cadmium atom was chosen as a 30
suitable example for the discussion f
of the origin of x-ray energy levels ^
for the reason that all subshells of 20
electrons are complete in the neutral
atom and the resultant angular
momentum of all the extranuclear 10
electrons is zero. The removal of an
inner electron from the atom in this
simple case would leave the atom 0
with the same angular momen¬
tum as that of the removed electron,
either l + £ or l — £. If, on the other hand, an atom with an incomplete
valence-electron subshell is considered, the resultant angular momentum
J1 of this outer shell must be combined with the resultant angular momen¬
tum J2 of the inner shell to give the resultant angular momentum J of
the atom. This, by analogy with optical spectra, may be accomplished
in such a way as to give several different J*b for the same x-ray level Jt.
The energy difference between the various possible J's will depend on the
type of coupling that may arise between the two //s. For heavy ele¬
ments, in general, intermediate shells of electrons will screen quite
effectively the inner shell from the outer shell, and the coupling will be
closely JJ. For the lighter elements, where the two shells are not
screened from each other, the electron spins may interact strongly and
the coupling will then be closely LS. X-ray energy levels for most
elements are therefore expected to show a multiplet fine structure
to that observed in optical spectra.
324 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

Attempts to observe this predicted structure in x-ray lines have

been made by a number of investigators. Coster and Druyvesteyn,1
and Van der Tuuk,2 for example, have observed complicated structures
in the N levels of the rare-earth elements. Because of the lack of
sufficient resolving power in x-ray apparatus detailed studies of these
multiplet structures have not yet been made. This multiplet structure
is not to be confused with the x-ray satellites to be discussed in the
following section. The selection and intensity rules for these more
complex atoms may be given here since they are the same as those used
in optical spectra (see Secs. 12.14 and 12.15).
The selection rules for x-rays are:

LS-coupling JJ-coupling
AS = 0 AJ i = 0
|or vice versa
AL = +1, 0, -1 AJ2 = +1, 0, -1
AJ = +1, 0, -1 AJ = +1, 0, -1
subject to the further restriction that odd terms combine only with even
terms.
Since the splitting of x-ray energy levels, due to a resultant angular
momentum of the valence electrons, is in general very small, a number
of the radiated lines will fall together as a single line. For this reason,
and by virtue of the fact that the sum rules are valid for multiplets, the
selection and intensity rules given in Sec. 16.5 will, when applied to the
one-electron jumping in the inner part of the atom, describe adequately
the observed lines.
16.11. X-ray Satellites.—X-ray satellites are those relatively weak
lines often observed close to and on the high-frequency side of the chief
x-ray diagram lines. These satellites, sometimes called non-diagram
lines, were first discovered by Siegbahn and Stenstrom3 on the high-
frequency side of the K series lines of the elements Na 11 through Zn 30.
Since their discovery an extensive investigation of the L series lines by
Richtmyer and Richtmyer4 and others has shown that a number of
satellites appear on the high-frequency side of the chief L lines of most of
the elements from Cu 29 through U 92. Similarly, Hirsch5 and others have
shown that M series satellites are to be found in the elements Yb 70
through U 92. The regions of a number of these satellites are marked
in Figs. 16.5 and 16.7 and may in some of the reproductions be traced
from one element to the next. Semi-Moseley plots of the satellites of

i Coster, D., and M. J. Druyvesteyn, Zeits. f. Phys., 40, 765, 1927.

* Van der Tuuk, J. H., Zeits. f. Phys., 44, 737, 1927.
* Siegbahn, M., and W. StenstrOm, Phys. Zeits., 17, 48, 318, 1916.
4 Richtmyer, F. K., and R. D. Richtmyer, Phys. Rev., 34, 574, 1929; see also
Richtmyer, F. K., and S. Kaufman, Phys, Rev., 44, 605, 1933.
* Hirsch, F. R., Phys. Rev., 38, 914, 1931.
Sac. 16.11] X-RAY SPECTRA 325
some of the more prominent x-ray lines are given in Fig. 16.18. The
square roots of A v/R (the difference in frequency between satellite
and nearest chief x-ray line, divided by the Rydberg constant R =
109737 cm-1) are plotted as ordinates against the atomic number Z.
The first attempt to explain x-ray satellites was made by Wentzel1
in 1921. Wentzel’s hypothesis requires essentially a double ionization
of the atom by high-speed electrons followed by a single electron
jump and the simultaneous emission of a radiated frequency vt. Because
of certain apparent defects in Wentzel’s theory as it was first proposed,

Fia. 1C. 18.—Semi-Moseley diagram of x-ray satellites. (After Richtmyer, Hirach, and
Ford.)

Richtmyer2 advanced the theory that in a doubly ionized atom double

electron jumps probably occur giving rise simultaneously to a single
radiated frequency vt. In view of recent developments in atomic spectra
it would appear that both theories are plausible and that together they
explain with some satisfaction the observed satellites.
Consider first Wentzel's theory, extended by Druyvesteyn,8 as it
has been applied to the satellites of the x-ray line Kpx in the elements
A1 13 to Fe 26. The parent line Kfil of the ordinary x-ray spectrum is
due to the transition K — Mu,m which, neglecting the fine structure
Mn — Mm, will be written K — M. The well-known satellite »
has been assigned by Druyvesteyn to the transition KL — LM. This
last notation indicates that in the initial state of the atom one K and oneL
electron is missing and that the transition of an M electron into th3
vacant K shell leaves the atom in a final state with one L and one M elec¬
tron missing. Now the term KL is somewhat greater than the sum of the

1 Wentzel, G., Ann. d. Phys., 66, 437, 1921; Zeits. /. Phys., 31, 445, 1925.
s Richtmyer, F. K., Phil. Mag., 6, 64, 1928; Jour. Franklin Inst., 203, 325, 1929.
* Druyvesteyn, M. J., ZeiU. f. Phys., 43, 707, 1927; Dissertation, Groningen,
1928.
 326 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

K and L terms of the ordinary x-ray-term scheme and the term LM is

greater than the sum of the L and M terms. This follows directly from
the assumption that the energy necessary to remove an L electron from
an atom in which a K electron is missing is greater than that necessary
to remove the same electron from a normal atom. When a K electron
is missing, the inner screening is smaller by unity than in the case of the
normal atom so that the binding energy of the L electron is approximately
that for the corresponding electron in the element with the next higher
atomic number. It therefore follows that
(KL)z — Kz + Lz+i (16.24)
and
(LM)z = Lz + Mz+1. (16.25)
The difference in frequency between the satellite (KL — LM) z and the
parent line (K — M) z becomes

Ap = (KLz — LMz) — (Kz Mz) — (Lz+i — Lz) — (Mz+i — Mz).

(16.26)
With these known differences1 on the right Druyvesteyn obtained for
Kfi— Kfr the following values of Av/R.

Table 16.11.—Observed and Calculated Positions of the Kp Satellites.

Values of A v/R
{After Druyvesteyn)

Calculated
Element Observed
Z A v/R
I'll.hi Li

A1 13 1.79 2 23 1.82 2 34
Si 14 2 07 2 74 2.11 2.63
P 15 2 38 2.97 2.40 2.92
S 16 2.97 2.70 3.22
a 17 3.27 2.97 3 49
K 19 3.59 3 55 4 07
Ca 20 3.83 3 84 4 36
Sc 21 3.77 3.53 4 05
Ti 22 3! 72 3 73 4 25
V 23 3.92 3 95 4 47
Cr 24 4.00 4 16 4.68
Mn 26 4.34 4.38 4.90
Fe 26 4.69 4 58 5.10

When one takes the fine structure of the L levels and neglects the
relatively smaller fine structure of the M levels, Eq. (16.26) predicts a
1The L differences being always greater than the M differences, Eq. (16.25)
indicates that the satellites will appear on the high-frequency side of the parent
x-ray line (see Fig. 16.5).
Sbc. 16.11] X-RAY SPECTRA 327
fine structure for Kp"’. Since the Ln — Lm interval is small as compared
with Li — Ln, the observed fine structure should first reveal itself as a
doublet. This splitting is observed in the first three elements, Al, Si, P.
It should be pointed out that, if the fine structure of both the initial
and final states is considered in the above example, many different
transitions become possible. The Kp satellites should therefore be
composed of a number of lines. Consider, for example, the possible
ways in which satellite transitions KL — LM may take place. All
of the possible terms and transitions are shown in the following scheme
(Table 16.12).
Table 16.12

Even Odd

K Li K Lii.ni
1«-1 28"1 l*"1 2p-'
lS *S xp ip

S' I
iP 3P ip »p ipiDlF*P*D*F 'S *S *D 'S'P'D'S'P'D
2s"1 3p~l 2p~l 3s*1 2p“l 3d"1 2s-1 3s"1 2s"1 3d"1 2p-» 3?-1
Li Mii.hi Lnjii Mi Lii'in AfIV,v Li Mi Li Afiv.v Ln,m AfntIXI

Odd Even

The configurations are given in terms of the missing electrons and the
levels in LS-coupling. Regardless of the type of coupling to be expected
between two adjacent incompleted subshells, the number of terms will be
the same as that given in the table above. Of the many allowed transi¬
tions it should be noted that those given by KLi — Ln,mMiv,y involve
the double electron jump 2p3d to 152s.1 All double electron jumps are
shown by dotted lines. In this diagram we see that both single and
double electron jumps are possible. Although the double jumps are
possible, they should give rise to weaker lines.2
Langer3 and Wolfe4 have suggested that the five well-known satellites
of Ka (see Fig. 16.18) are to be attributed to the transitions KL—>LL.
Following Wolfe’s assignments, based upon quantitative calculations
of the electron-interaction energies in a Hartree field for the potassium
atom (see Fig. 7.2), a schematic energy-level diagram has been con¬
structed in Fig. 16.19. The parent K and L levels are shown at the left
with the unresolved Kaiit doublet below. The removal of a second
electron from the atom splits these levels (neglecting valence electrons if
1 The selection rules for double electron jumps are Ah * 1, AU * 0, 2.
* A treatment of double electron jumps in x-ray spectra has been given by E. G.
Ramberg, Phys. Rev., 46, 389, 1934.
* Langer, R M., Phys. Rev., 37, 467, 1931.
4 Wolfe, H. C., Phys. Rev., 48, 221, 1933.
328 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

here are any in unclosed subshells) into singlets and triplets as shown.
As pointed out by Wolfe, LS-coupling is to be expected in these levels
since the spin-orbit interaction energy (shown by the Ln — Lm interval)
is small compared with the interaction between the two electrons (shown
by the satellite separations). Five of the six allowed transitions account
for the five observed satellites. The sixth allowed transition, shown
as a dotted line, is a double electron transition and should give rise to a
relatively weak line on the low-frequency side of KThe results show

Via. 16.19.—Splitting up of x-ray levels due to the removal of a second electron from the
atom. Illustrating a possible explanation of x-ray Ka satellites.

that Wolfe's assignments give a reasonable explanation of the K* satel¬

lites. The Kfi satellites may also be assigned in the same way to the
transitions shown in Table 16.12.
Although the existing data for checking this theory are somewhat
limited, double ionization followed by single, or less frequently by
double, electron transitions accompanied by the simultaneous emission
of a single radiated frequency appears to be a very plausible explanation
of x-ray satellites in general.
There appears to be, at first sight, two processes by which double
ionization of an atom may be brought about: (1) a single high-speed
particle may remove both electrons, and (2) two high-speed particles
may collide successively with the atom and remove one electron after
the other. Du Mond and Hoyt, and others, have shown experimentally
that the latter tandem process must be ruled out. From the theoretical
standpoint the probability that two collisions will occur within the life
Sac. 16.121 X-RAY SPECTRA 329

time of the excited state is far too small to account for the observed*
satellites.
Although few quantitative data are to be found on < the excita¬
tion potentials of x-ray satellites, observation supports the theory
of double ionization by a single impact. Druyvesteyn has shown for
example that while the parent x-ray lines Kaui in vanadium appear at
5.45 kv.,the satellites K*tA appear only at potentials above 6.09 ± 0.1 kv.
This is not in agreement with WentzePs early suggestion that two
K electrons have been removed from the atom. Assuming Wolfe’s
assignments are correct for Kaiti (see Fig. 16.19), the removal of one
K and one Ln,m electron from vanadium requires about the same
energy as that necessary to remove a K (Is) electron from vanadium,
Z = 23, and an Lu,m (2p) electron from chromium, Z = 24. The
sum of these two ionization potentials is 5.450 + 0.586 = 6.036 kv.,
in excellent agreement with observation. In a similar manner the Lp,
line (due to the transition Lm — N\) in Ag 47 has an excitation potential
of 3.34 kv., and the Lp, satellites should, if attributed to the allowed
transitions LmMx — MxNv, appear at about 20 per cent higher voltages.
i.e., 4.00 kv. This likewise is in good agreement with experiment.
16.12. Explanation of X-ray Absorption Spectra.—In photographing
x-ray absorption edges, it often happens that absorption lines are found
close to and on both sides of the absorption edge. Under high dispersion
and resolving power the lines on the low-frequency sides appear as
absorption lines, whereas those on the high-frequency side appear as
subsidiary absorption edges. In 1920 Kossel1 made the suggestion
that an inner electron, as the result of the absorption of a photon hv, need
not be completely removed from the atom but instead be simply excited
to one of the outer unoccupied states. The lowest possible frequency
to be absorbed therefore would be given by the energy necessary to
carry an inner electron to the most tightly bound but unoccupied electron
orbit. These transitions are governed by the same selection rules valid
in emission spectra and lie within only a few volts of the absorption
edges2 which would correspond to complete removal of the electron.
A detailed study by Kievit and Lindsey3 of the subsidiary absorption
edges, observed on the high-frequency side of the K edges of the elements
Ca 20 to Zn 30, appears to confirm the theory already postulated, that
they are due to the removal of more than one electron from the atom.
Kronig4 has, however, been able to explain this extended structure
more satisfactorily on the basis of the existence of prohibited zones
in the velocity spectrum of free electrons in crystals.

1 Kossel, W., ZeUs. f. Phys., 1, 124, 1920.

* Siegbahn, M., Zeits.f. Phys., 67, 567, 1931.
* Kievit, B., and G. Lindsey, Phys. Rev., 36, 648, 1930.
4 Kronig, R. db L., Zeits. f. Phys., 70, 317, 1931; 76, 191, 1932.
330 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI

It should be mentioned that for a given element the chief absorption

edges are not always found at the same wave-length but vary somewhat
by an amount that depends upon the chemical composition of the
substance used in producing the edge. This shifting of the edge has
been attributed by Pauling1 to the difference in the electrostatic fields
within the material used as a screen. As an example of the magnitude
of the shift, the K limit of sulphur, as measured from 50 different com¬
pounds, lies between X — 4987.3 and X = 5011.7 X units, a maximum
variation of 34.4 X units.2
1 Pauling, L , Phys. Rev., 34, 954, 1929.
* Siegbahn, M., “Spektroskopie der Rdntgenstrahlen,” 2d ed., p. 284, 1931.
 CHAPTER XVII

ISOELECTRONIC SEQUENCES

The term isoelectronic sequence as it is commonly used refers to a

sequence of atoms having the same number of extranuclear electrons.
In general such a sequence starts with any element in the periodic table
and is followed by other elements in the order of their atomic number.
Since each neutral element contains one more electron than the one
just preceding it in the periodic table, each atom must be stripped,
i.e., ionized, of just the right number of electrons to leave it isoelectronic
with the first element in the sequence. Suppose, for example, that a
sequence starts with potassium, Z = 19. The following elements,
calcium, Z = 20, scandium, Z — 21, titanium, Z = 22, etc., are all
made isoelectronic with neutral potassium (K I) by removing one
electron from calcium, yielding Ca II; two electrons from scandium,
yielding Sc III; three electrons from titanium, yielding Ti IV, etc.
Because each atom in such a sequence contains the same number of
extranuclear electrons the energy levels and the spectrum lines arising
from each atom will show remarkable similarities from element to
element. Moreover, the empirical rules discovered from a study of
these energy levels and spectrum lines are closely associated with the
well-known x-ray doublet laws treated in the preceding chapter.
In 1923 Paschen1 and Fowler2 identified series of sodium-like doublets
in Mg II, A1 III, and Si IV.
Not long afterward, Land6 pointed out the similarity between
the first doublet of the principal series of these elements and the spin
relativity or regular doublets Ln — Lm, Mn — Mm, • • • of x-ray
spectra.3 This correlation between x-ray and optical spectra in turn led
him to the correct assignment of the quantum numbers k and j to the
x-ray levels K, Li.n.m, Afi.n.m.iv.v, etc. Similarly, de Broglie and
Dauvillier4 pointed out that the Ln — Lm doublet is analogous to the
P doublet of the alkalies and that these two doublets therefore must
have the same azimuthal quantum numbers. At about the same time

1 Paschen, F., Ann. d. Phys., 71, 142, 1923.

* Fowler, A., Proc. Roy. Soc., A, 103, 413, 1923; Phil. Trans. Roy. Soc., 226, 1,
1926.
*Land£, A., Zeits.f. Phys., 16, 391, 1923.
4 De Broglie, L., and A. Dauvillier, Jour, dt phys. et radium, 6, 1, 1924.
331
332 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVII

Millikan and Bowen,1 quite independently, were able to go further and

show that the regular- and irregular-doublet laws of x-rays may be
applied to optical doublets. Somewhat later an extension of these same
laws to complex spectra was made by Gibbs and White.2
17.1. Isoelectronic Sequences of Atoms Containing One Valence
Electron.—As an example of the sim¬
Vp, plest type of isoelectronic sequence, let
us consider the group of alkali-like
atoms that are isoelectronic with neu¬
tral potassium, Z = 19, viz.: K I, Ca
II, Sc III, Ti IV, V V, Cr VI, . . . .
In this sequence each atom contains in
Col addition to the closed shells of electrons
one single valence electron. With a
similar core of 18 electrons (Is2 2s2 2p*
3s2 3p6) and one valence electron, the
Sen system of energy levels and the radiated
series of spectrum lines characteristic
of each element will consist of doublets.
Photographs of the first member of
Ti N the principal series of the first five
elements K I, Ca II, Sc III, Ti IV,
and V V are reproduced in Fig. 17.1.
The potassium doublet, in the far red
end of the visible spectrum, and the
ionized calcium doublet, in the violet,
1679.36 are the strongest fines in K I and Ca
Fig. 17.1.—Photographs of the prin¬ II. From these and other observed
cipal-series doublets 4s,5*S$ — 4p,
for the isoelectronic sequence K I, Ca II, doublets, energy level diagrams are
Sc III, Ti IV, and V V. easily formulated for each element in
the sequence. It is with these energy levels and the laws governing
them that we are now concerned.
By analogy with x-ray doublets [see Eq. (16.17)], term values are
given by Hertz’s formula
R(Z - <r)2
T = 07.1)

From the theory of penetrating orbits on the other hand they may
also be represented by (see Sec. 7.4)

r <17-2>
1 Millikan, R. A., and I. S. Bowen, Phys. Rev., 24, 209, 1924; 26, 296, 1926;
26, 160, 1926; 27, 144, 1926.
2 Gibbs, R. C., and H. E. White, Phys. Rev., 29,426,1927; 29,666,1927; Proc. Nat.
Acad. Set.. 13. 626. 1927.
Sec. 17.1] ISOELECTRONIC SEQUENCES 333
In the first formula, Z is the atomic number, n is the principal quan-,
turn number and a is a screening constant. In the second formula
Z0 is the effective nuclear charge outside the atom and fi is the quantum
defect. (Z0 takes the values, unity for neutral atoms, two for singly
ionized atoms, three for doubly ionized atoms, etc.) The differences
between the term values of all atoms as given by the first formula and
those of hydrogen-like atoms as given by

RZ2
T = (17.3)
n2

are accounted for by the screening constant cr, whereas in the second
formula the differences are attributed to the quantum defect /*. We

Fig. 17.2.—Moseley diagram of the isoelectronic sequence K I, Ca II, Sc III, . . . The

slopes are closely 1/n.

shall now see how both of these formulas apply separately to the doublets
of an isoelectronic sequence. Transposing R in the first formula [Eq.
(17.1)] and taking the square root,

(17.4)

In Table 17.1 and in Fig. 17.2 the observed term values for the potassium¬
like doublets are given, as is customary in x-rays, in terms of y/T/R.
It is observed from the difference columns in Table 17.1 and from
the Moseley diagram that levels with the same principal quantum
number (n = 4) run nearly parallel. It is to be noted further that
the slopes of the curves, as shown by the dotted lines, are in agreement
with Eq. (17.4), i.e.t they are very nearly equal to 1/n. If the dotted
lines are drawn through the zero-zero origin of coordinates, as they
334 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVII

should be for hydrogen-like atoms, the ordinate differences between

them and the plotted points will, when multiplied by n, give <r, the
screening value of the 18 extranuclear electrons. These screening

Table 17.1.—Values op y/T/R for the Potassium-like Isoelbctbonic Sequence

K I, Ca II, Sc III, Ti IV, ... 1

Term KI Diff. VV

3 *D 0.350 0.516 0.865 0.434 0.407 2.190

4 *S 0.565 0.369 0.934 0.326 1.260 0.306 1.565 0.292 1 857
4 *P 0.448 0.363 0.801 0.319 1.120 0.298 1 418 0 288 1.706
4 *D 0.263 0.334 0.597 0.296 0.893 0.284 1.177 0.279 1 456
4 *F 0.250 0.252 0.502 0.255 0.757 0.267 1.014 0.258 1.272

1 Term values T for a Moseley diagram are always computed with respect to the aeries limit as sero.

constants have been calculated and are given in Table 17.2. Since
the doublet fine-structure separations of 2P, 2D, and 2F are small as
compared with 2D — 2F or 2P — 2D separations, they are neglected
in the figures and in the tables.

Table 17.2.—Values of the Screening Constant a for the Isoelectronic

Sequence of Potassium-like Atoms

Electron
Term KI Ca II Sc III Ti IV V V
configuration

3d 3 'D 17.95 17.40 16.95 16.65 16 43

4s 4*S 16 74 16 26 15.96 15.74 15.58
4V 4 'P 17 21 16 80 16.52 16.33 16.18
4d 4 'D 17.95 17.61 17.43 17 29 17.18
4f 4V 18 00 17.99 17.97 17 94 17.91

In general it is observed, as would be expected from any of the

atomic models, that the screening by 18 core electrons is not so complete
for penetrating s and p electrons as it is for the nonpenetrating 4f elec¬
trons. In the particular sequence chosen it should be noted that after
K I the 32D states have considerably larger <r values (see Fig. 17.2). It
is this change in the binding of 3d electrons that brings about an inter¬
ruption at potassium in the building up of the elements of the periodic
table.
A more general study of the electron-binding energies and the applica¬
tion of the x-ray laws to the alkali-like doublets throughout the periodic
table is brought out in the Moseley diagram of Fig. 17.3. In this figure
of five isoelectronic sequences it is observed (1) that levels with the same
n value are closely parallel, (2) that the slopes of the curves for the first
sequence are quite accurately 1/n, (3) that the slopes of most curves
deviate more and more from 1/n in going to heavier and heavier elements,
Sue. 17.1] ISOELECTRONIC SEQUENCES 335

(4) that the 2F curves start at approximately the same value for all.
sequences and toward heavier elements curve slightly upward, (5)
that the 2S, 2P, and 2D curves for heavier elements curve downward,

Fig. 17.3.—Moseley diagrams of the doublet levels in five periods of the periodic table.
(After Gibbs and White.)

and (6) that the 3 2D curves change their slope abruptly between the
Na and K sequences.
Although little is known concerning the energy levels of the rare-
earth elements, it is suspected that the 2F curve in the last sequence

Fig. 17.4.—Energy level diagrams of the potassium-like atoms. A modified Moseley

diagram. (After While.)

should have a much steeper slope than that indicated in the figure.
This seems necessary from the standpoint of the building on of / electrons
in the 14 rare-earth elements starting with cerium. The figure as it is
336 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVII

now shown indicates that a 5d electron is more tightly bound to the

triply ionized cerium atom, Ce IV, than is a 4/ electron. A comparison
of Fig. 17.3 with the x-ray Moseley diagram of Fig. 16.15 brings out a
number of interesting similarities.
When values of y/T/R are plotted vertically upward as is done on a
Moseley diagram, the normal state of each atom becomes the highest
point on the curve. As an energy level diagram the figures are therefore
inverted. By plotting -y/T/R (or \/T) downward, instead of upward,
Fig. 17.2 may be transformed into a sequence of energy level diagrams

KI Can Sc 1 Tin VT H

and a Moseley diagram combined (see Fig. 17.4). The observed transi¬
tions may now be shown in the usual manner by arrows pointing down.
Let us now turn to the second formula [Eq. (17.2)] derived for pene¬
trating orbits and see how it applies to the energy levels of a typical
isoelectronic sequence. Transposing, Eq. (17.2) may be written

T_ R
(17.5)
Z\ (n - m)2

Dividing each term value by Z\ and plotting T/Z* in place of T} the

energy level diagram of each element will be reduced to about the same
order of magnitude. Consider, for example, the potassium-like sequence
of Fig. 17.2 and Table 17.1 as it is replotted in Fig. 17.5. In this new
figure the hydrogen-like states are represented by horizontal dotted
lines (ji = 0). The difference between these dotted lines and the plotted
observed terms now give, indirectly, the quantum defect p. Values of p
are given in Table 17.3.
‘Whether the differences between observed term values and those
calculated for hydrogen-like atoms are to be attributed to a screening
Sbc. 17.2] ISOELECTRONIC SEQUENCES 337

constant <r [Eq. (17.1)] or a quantum defect n [Eq. (17.2)] depends upon
the purpose for which these constants are to be used. In identifying'
higher members of a spectral series and in evaluating terms and series
limits, the second formula is to be preferred. From the theoretical
standpoint, on the other hand, the first formula is to be preferred in that

Table 17.3.—Values op the Quantum Defect ju for the Isoelectronic Sequence

of Potassium-like Atoms

Electron
Term K I Ca II Sc III Ti IV V V.
configuration

3d 32D 0.143 0.688 0.776 0.757 0 717

4s 4 2S 2.230 I 858 1 619 1.444 1 308
4p 42P 1 769 1 504 1 322 1 179 1.069
4d 4 2D 0 198 0.640 0 637 0 602 0 566
4/ 4 2F 000 0 016 0 037 0 055 0 069

quantum numbers should take integral values only. Moreover the x-ray
laws have been used so successfully in the analysis of the spectra of
highly ionized atoms and in the extension of isoelectronic sequences in
general that the screening formulas frequently can be used to greater
advantage.
17.2. Optical Doublets and the Irregular-doublet Law.—The irregu¬
lar-doublet law, extended from x-ray to isoelectronic sequences in optical

Fio. 17.6.—Illustrating the irregular-doublet law discovered in optical doublets by Millikan

and Bowen. The law holds only for transitions where An = 0.

spectra by Millikan and Bowen, may be stated in terms of the energy

levels as follows: the differ eare between the square roots of the term values
of the levels having the same principal quantum number n is independent oj
338 INTRODUCTION TO ATOMIC SPECTRA (Chap. XVII

the atomic number Z. In other words such levels on a Moseley diagram

run parallel to each other (see Fig. 16.15 and compare it with Figs. 17.2
and 17.3 for optical doublets).
As a consequence of the parallelism of any two curves on a Moseley
diagram, the irregular-doublet law has another and a more useful form.
In Sec. 16.9 it was shown (see Fig. 16.17) that the term differences
between two levels having the same n value are linear functions of the
atomic number Z. While corresponding transitions (An = 0) are not
observed in x-ray spectra, they are observed in optical spectra. As an
example of this form of the irregular-doublet law, a plot of the radiated
frequencies due to the transitions np to ns have been plotted for five
sequences of alkali-like atoms in Fig. 17.6. Such a graph shows at a
glance how remarkably exact this law holds in optical spectra. The
realization of this law furnishes a most reliable and accurate means for
predicting the frequencies and wave-lengths of unidentified spectrum
lines in multiply ionized atoms.
17.3. Optical Doublets and the Regular-doublet Law.—The regular-
doublet law, extended by Millikan and Bowen to optical doublets, is
given by the x-ray formula (see Secs. 16.6, 16.7, and 16.8)

Ra\Z - $)*
Avom-i — (17.6)
tj.3
where Av represents the fine-structure doublet separations in wave
numbers, Z the atomic number, n the principal quantum number,
R the Rydberg constant = 109737 cm-1, a2 = 5.3 X 10~6, and

1 = 1
(17.7)
k\ &2 ki fci *i" 1 k\(k\ -{- 1)

In this formula ki and k2 are Sommerfeld’s quantum numbers, as

used in x-ray formulas. The value of B fortunately remains the same
if ki is replaced by the newly assigned quantum number l, where 1 = 0,
1, 2, 3, • • • for s, p, d, /, • • • electrons, respectively:

Ra2(Z - s)4
Av cm”1. (17.8)
nH(l + 1)

This is the same formula as that derived in Chap. VIII from a different
treatment [Eq. (8.25)]. From the observed separations in the lithium¬
like sequence of nonpenetrating 2p orbits, screening constants s are
calculated and given in the last column of Table 17.4.
Assuming a complete screening by the two core electrons, i.e., s = 2,
Z - « - 1, 2, 3, 4, 5, and 6 for Li I, Be II, B III, C IV, N V, O VI, the
separations given in Col. 3 have been calculated. Still better calculated
values are obtained by first computing values of Z — v from the observed
Sec. 17.31 ISOELECTRONIC SEQUENCES 339

term values and Eq. (17.1), and substituting <r for 8 in Eq. (17.8). The
values computed in this way are given in Col. 4.
Table 17.4.—Doublet Separations for Lithium-like Atoms, 2*P$ — 2*P§

z -r, Z — st
Ay Ay Ay
z Atom from term from doublet 8
(obs.) (calc.) (calc.)
values separations
_ ... _

3 Li I 1.021 0.98 2.02

4 Be II 6 61 S'. 82 6 39 2.047 1.94
5 B III 34 4 29.4 32.1 3.064 3.12 1.88
6 C IV 107.4 93.0 100 4 4.076 4.14 1.86
7 N V 259 1 227.4 243.1 5.084 5.16 1.84
8 0 VI 533 8 471 0 500.8 6 091 6.18 1.82

Table 17.5.—Screening Constants st Calculated from the Observed Term

Valueb and Land^'s Formula for Penetrating Orbits
(After Gibbs and White)1

Atom Ay (obs.) Z Z0 So zx sx

Na I 17 18 11 1 10 7.66 3.34
Mb II 91 55 12 2 10 9.56 2 44
A1 III 234 0 13 3 10 10.91 2.09 Third period
Si IV 461.8 14 4 10 12.08 1.92 3*Pj - 3*P|
P V 794 8 15 5 10 13.15 1.85
s VI 1267 1 16 6 10 14 23 1.77
Cl VIII 1889 5 17 7 10 15 29 1 71

K I 57.7 19 1 18 14.85 4.15

Ca II 222 8 20 2 18 17 24 2 76
Sc III 474 3 21 3 18 18.29 2.71 Fourth period
Ti IV 817 5 22 4 18 18 99 3.01 4*Pj - 4*P,
V V 1264 7 23 5 18 19 54 3 46
Cr VI 1821 5 24 6 18 19 98 4 02
Mn VII 2464 7 25 7 18 20 15 4 85

Rb I 237 6 37 1 36 31 23 5.77
Sr II 801 5 38 2 36 34.81 3.19 Fifth period
Yt III 1553 7 39 3 36 36 29 2.71 6*P| - 6*P,
Zr IV 2484 9 40 4 36 37.10 2.90

Cs I 554 0 55 1 54 49.55 5.45

Ba II 1690 9 56 2 54 53 88 2.12 Sixth period
La III 3095 7 57 3 54 55 52 1.48 VP\ - 7*P§
Ce IV 4707 0 58 4 54 56 05 1.95

1 Gibbs, R. C , and H £. White, Phya. Rev., S3, 157, 1929.

A refinement of Eq. (17.6) for penetrating orbits has been given

by Land6 (see Sec. 8.7), in which Z4 is split up into two parts, Z\ and Z\
where Z> - Z — 8if Z0 =* Z — 80f and Z is the atomic number.
340 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVII

Ra'ZlZf
Av (17.9)
nll(l + 1)'

Here n0 is the effective quantum number. Values of Z* and s< have been
calculated for the first principal doublets of four isoelectronic sequences
in Table 17.5.
The large values of Z* indicate very deep penetration of the p orbits.
Regardless of what physical significance can be attached to these screen-
X6494 6462 6439 55989188

Fig. 17.7.—Photographs of triplet multiplets for the isoelectronic sequence Ca I, Sc II,

Ti III, and V IV.

ing constants sif Eq. (17.9) serves as an excellent guide in predicting new
doublet separations for spectra not yet photographed or analyzed.
17.4. Isoelectronic Sequences of Atoms Containing Two Valence
Electrons.—In the previous sections of this chapter we have seen how
the x-ray doublet laws apply to the optical doublets of atoms containing
one valence electron. In this and the following sections we are con¬
cerned with the extension of these laws to the complex spectra of atoms
containing a number of valence electrons. As a first step in this direc¬
tion, let us consider the isoelectronic sequence of semicomplex atoms,
viz,, neutral calcium (Ca I), singly ionized scandium (Sc II), doubly
ionized titanium (Ti III), triply ionized vanadium (V IV), etc. In this
intermediate case each atom contains, in addition to the closed subshells
Sue. 17.4] ISOELECTRONIC SEQUENCES 341

of 18 elections (Is2 2s2 2pe 3s2 3p6), two valence electrons. With two,
electrons in incomplete subshells the energy level diagram of each atom
is composed of a system of singlet and triplet levels and the spectrum
of each atom is made up of groups of lines called singlets, triplets, and
triplet muUiplets.
Photographs of two sets of these triplet multiplets are shown in Fig.
17.7 for the first four elements of the calcium-like sequence. A schematic
representation of the triplet terms and intervals is given below each
photograph. Upper states are shown on the left and lower states on the
right.

Fig. 17.8.—Modified Moseley diagram of the isoelectronic sequence Ca I, Sc II, Ti III, . . .

A modified Moseley diagram of the type constructed in Fig. 17.4

for the alkali-like atoms may also be constructed for the calcium-like
sequence. To illustrate, term values of the first six elements, calcium
to manganese, are plotted in Fig. 17.8. Due to the many terms arising
from the four configurations shown, 3d2, 3d4s, 3d4p, and 3d4d, only the
lowest and the highest terms for each configuration are drawn. Plotted
in this way all allowed terms (listed at the right in the figure) fall within
the upper and lower limits of each rectangle. Grouped as they are,
smooth curves can be drawn through the levels of corresponding con¬
figurations. It is observed that configurations for which the total
quantum numbers of the electrons are the same run nearly parallel to
one another. This is the irregular-doublet law in complex spectra.
A comparison of Fig. 17.8 and Fig. 17.4 indicates that the irregular-
doublet law, as applied to radiated frequencies in Fig. 17.7, will also apply
to two electron spectra. Briefly the irregular-doublet law in this, its
most useful form, means that, in any transition involving no change in
the principal quantum numbers of the electrons, the radiated frequencies
342 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVII

in going from one element to the next are displaced to higher and higher
frequencies by very nearly equal frequency intervals. In the calcium-like
sequence, two such sets of frequencies have been observed, first the set
of lines arising from the electron transition 3d4d to 3d4p and second the
set of lines arising from the transition 3d4p to SeMs.1 The allowed terms
for these transitions are as follows:

Initial even states 1S lP lD lF HJ *s IP 3D ZF Hj 3d4d

Final odd states! ipo ipo zp° 3jf)o ZJpo 3d4p

Initial odd states)

Final even states lD *D Sdis

Examples of the linear progression of v with Z are given in Table 17.6

for the line — Hh as representative of the upper set of transitions and
for the lines 3Z), — and lD% — lD°2 as representative of the second set
of transitions.

Table 17.6.—The Irregular-doublet Law Applied to Singlets and Triplets

3d4p -- Mid 3d4s -- 3d4p 3d4« -- 3d4p

Element
Diff. 3DS - 3F°a Diff. ^2- lD°2 Diff.

Ca I (13974) 15526 13986

(18640) 12137 9554
Sc II 32614 27663 23540
18700 12070 9964
Ti III 51314 39733 33494
18782 11838 10578
V IV 70096 51571 44072
18902 11733 10865
Cr V 88998 63304 54937

None of the terms arising from the electron configuration 3d4d hatf
as yet been determined for Ca I; however, an extrapolation by means
of the second and third columns of Table 17.6 locates the *F\ — 3(j6
transition not far from 13974 wave numbers.
The almost linear displacement of frequency with atomic number,
for all lines due to the transition 3d4p to 3d4$, is shown graphically
in Fig. 17.9. The frequency scale for each element starts 11000 wave
numbers higher than the preceding element. This type of diagram
shows not only the successive displacement of each singlet and triplet
group of lines toward the violet by about 11000 cm~l but also magnifies

1 The many spectrum lines arising from the transition 3<i4<i to 3d4p are
shown schematically in Fig. 12 4 for singly ionized scandium (Sc II).
Sec. 17.5J ISOELECTRONIC SEQUENCES 343

many times the variations in this displacement for each individual

spectrum line.
Similar studies of other two-electron isoelectronic sequences show
that the irregular-doublet law may always be expected to hold. In
every sequence investigated the results may be summarized as follows:
(1) Curves on a Moseley diagram are smooth and almost straight. (2)
Curves on a Moseley diagram having the same principal quantum num¬
bers in the electrons run nearly parallel to each other. (3) Radiated
frequencies arising from transitions involving no change in n increase
almost linearly with the atomic number Z.
sd-5f®

Fig. 17.9.—Illustrating the irregular-doublet law for triplet multipleta arising from
the configuration transitions 3d4a — 3d4p. The almost linear displacement of frequency
with atomic number holds only for An = 0. (After White.)

A number of interesting relations may be brought out by an extension

of the regular-doublet law to complex spectra. Since these relations
are seldom employed in the analysis of new spectra the regular-doublet
law will not be treated here.
17.5. Isoelectronic Sequences Containing Three or More Valence
Electrons.1—An extension of the x-ray doublet laws from the optical
spectra of one and two valence-electron atoms to systems containing
three or more valence electrons is now but an easy step. Consider,
for example, the isoelectronic sequence starting with neutral scandium,
Z = 21. Each atom in this sequence Sc I, Ti II, V III, Cr IV, Mn
V, . . . contains, in addition to the closed subshells of 18 electrons

1 Gibbs, R. C., and H. E. White, Phys. Rev., 29, 655, 1927; White, H. E., Phys.
Rev., S3, 672, 1929; 33, 914, 1929.
344 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVII

(Is* 2s* 2p6 3s* 3p8), three valence electrons. As seen by the energy level
diagram of neutral scandium in the lower part of Fig. 14.8, each atom
in this sequence will give rise to complex systems of doublet and quartet
levels.
Along with this group of elements we may also consider the iso-
electronic sequence starting with neutral titanium, Z = 22. Each atom
in this second sequence Ti I, Y II, Cr III, Mn IV, Fe V, . . . contains,
in addition to the same closed subshells of 18 electrons, four valence

Fig. 17.10.—Modified Moseley diagram for doublets, triplets, quartets, and quintets
in the first long period. (After White.)

electrons. As seen by the energy level diagram of neutral titanium

in Fig. 14.10, each atom in this sequence will give rise to complex systems
of singlet, triplet, and quintet levels.
The main features of the energy level diagram of each element
in both of these sequences being known, Moseley diagrams of the type
shown in Figs. 17.4 and 17.8 may be constructed. For this construction
four of the lowest and most prominent electron configurations in each
element have been selected:

3d8, 3d24s, 3d24p, 3d24d for Sc I, Ti II, . . . .

3d4, 3d84s, 3d34p, 3d84d for Ti I, V II, . . . .

Since each configuration gives rise to many levels, the deepest lying
level is chosen as representative of that configuration on the diagrams.
In accord with Hund's rule, these are the levels of largest S and L
values shown schematically in Fig. 17.10. The corresponding terms
Sec. 17 6] ISOELECTRONIC SEQUENCES 345

of the one- and two-electron sequences of Figs.' 17.4 and 17.8 are also,
given for comparison. It should be mentioned that if all possible
levels for each configuration are known, the center of gravity of the
configuration can be plotted in place of the lowest term. (Each level
is assigned its quantum weight 2J + 1.) In cases where the termB are
all known, the resultant diagrams are so little different from the type
here shown that the essential features of the diagram as a whole remain

Fia. 17.11.—Photographs of quartet multiplets for the isoelectronic sequence Sc I, Ti II,

and V III.

unchanged. For most purposes, one is interested in the deepest lying

term of a configuration as that represents the most stable state of the
atom in that electron configuration.
Photographs of two strong quartet multiplets, 4F — AF° and AF — AG°,
arising from the electron transition 3d24p to 3d24s in Sc I, Ti II, and
V III are shown in Fig. 17.11. Similar photographs of a quintet multiplet
&F — Ki0, arising from the electron transition 3d34p to 3d34s in Ti I,
V II, and Cr III are shown in Fig. 17.12. The initial and final states
and their intervals are represented schematically below each photograph.
Considering that these multiplets arise from different elements, they
show by their similarity the remarkable likeness of different atomic
systems, one with the other.
17.6. The Irregular-doublet Law in Complex Spectra.—A preliminary
study of the modified Moseley diagrams in Fig. 17.10 shows that the
curves for the configurations 3dn-14$, 3dn_14p, and 3dn-l4d run nearly
parallel to one another. This is the irregular-doublet law in complex
346 INTRODUCTION TO ATOMIC SPECTRA IChap. XVII

spectra. In contrast with this law, it is to be noted that the configura¬

tions 3dn cut down across the others and become the lowest levels in
the third and succeeding elements.1 As a consequence of the parallelism
of certain curves on the Moseley diagram it was shown in Secs. 17.2
and 17.4 that the related spectral frequencies increase linearly with the

(3d3 4s-3d34p) \5014 5007 4999 4991 4981

Sec. 17.6J /SOELECTRONIC SEQUENCES 347

triad of doublets and the AD°t AF°, *G° terms, combining with AF, give
rise to a strong triad of quartet multiplets. These two triads have been

Fig. 17.13.—Illustrating the irregular-doublet law for quartet multiplets arising

from the transitions 3d24s — 3dHp, in the isoelectronic sequence Sc I, Ti II, V III, . . .

Fig. 17.14.—Showing the contrast between the irregular-doublet law (b) and (c), where
An *= 0, and the transitions (a) where An ** 1.

identified and plotted to frequency scales in Fig. 17.13. It is to be noted

that each scale starts 12000 wave numbers farther along than that
of the preceding element. In this way the successive displacement of
348 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVII

many lines toward the violet by almost equal frequency intervals is

brought out most effectively. If any one particular line in this figure
is plotted as a function of Z, the resultant curve, as in the case of the
doublets in Fig. 17.6, is practically a straight line. Such curves are
shown in Fig. 17.14 for a number of strong characteristic lines belonging
to each of the four sequences starting with neutral potassium, calcium,
scandium, and titanium. For transitions involving a change in the
principal quantum number of one of the electrons, i.e., 3dn-14p to 3dnf
the curves are by no means linear but curve rapidly upward for the first
three or four elements and then begin to straighten out.

Fio. 17.15.—Modified Moseley diagram and the irregular-doublet law extended to the
energy levels and radiated frequencies of the same element, vanadium, in different stages
of ionization. (After White.)

17.7. Energy Relations for the Same Atom in Different Stages of

Ionization.—A further study of the modified Moseley diagram in Fig.
17.10 shows that smooth curves can be drawn through corresponding
states of the same atom in different stages of ionization. Consider,
for example, singly ionized vanadium (V II), doubly ionized vanadium
(V III), triply ionized vanadium (V IV), etc. The points for these
atomic systems when replotted as shown at the left in Fig. 17.15 reveal a
diagram similar to Fig. 17.10 for an isoelectronic sequence. With the
discovery that certain curves in this figure run nearly parallel to each
other, we now postulate that certain groups of spectrum lines arising
from the same atom in successive stages of ionization are displaced to
higher and higher frequencies by approximately equal frequency intervals.
This is verified experimentally by the plot of two different electron
transitions shown at the right in Fig. 17.15. Just as in the case of
isoelectronic sequences these nearly straight curves involve transitions
for which the total quantum numbers of the electrons do not change.
The third set of transitions shown in the figure involve combinations
for which the quantum number n of one of the electrons changes. In
Sec. 17.7] ISOELECTRONIC SEQUENCES 349

constructing Fig. 17.15 there are many terms And spectrum lines fron*
which to choose. From any one given electron configuration, for
example, there may be as many as 100 theoretically possible levels.
Since ail possible levels in complex spectra are seldom observed, the
deepest lying term is conveniently chosen as representative of that
configuration. This is quite general practice in predicting new terms
and in making an analysis of a new spectrum.
A V I Multiplet

A2530 V II V II VIH X2269

Fig. 17.10.—A. Ultra-violet spectrum of vanadium. Furnace spectra at (a) 2000° Abs.,
(b) 2300° Abs., (c) 2600° Abs. (<l) Arc spectra, (e) Spark spectra. (After King.) B.
Vanadium spectrum from a condensed spark discharge. (After White.)

A small section of the visible spectrum of vanadium is reproduced

in Fig. 17.16A. This set of five spectrograms, assembled by King,1 shows
how lines and multiplets arising from transitions between low levels
or* between relatively high levels may be distinguished from each other
as well as from lines arising from ionized atoms. At very low tempera¬
tures only the low-level lines of the neutral atom V I appear. As the
temperature increases, the low-level lines of the ionized spectrum appear,
first very weak, then stronger in the arc, reaching a maximum in the
spark. Figure 17.162? is another section of the vanadium spectrum
taken with a high-voltage spark mounted parallel to the slit of a stig-
matic spectrograph. From such a spectrogram lines arising from
different stages of ionization are distinguished by the region of the spark
from which they appear to come. Lines from the neutral and singly
ionized atoms V I and V II appear with nearly the same intensity at

1King, A. S., Astrophys. Jou*., 60, 282, 1924.

350 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVII

all points along the spark. Lines from higher stages of ionization, V III
and V IV, appear strong at the electrodes and very weak, if at all, in the
middle. These are but a few of the criteria that are often employed
in the unraveling of an unanalyzed spectrum.
17.8. Centroid Diagrams.—The centroid diagram, introduced into
atomic spectra by Mack, Laporte, and Lang,1 is often drawn for the
purpose of showing just how the energy levels of a given electron con¬
figuration change in going from one element to the next. As illustrations

Fio. 17.17.—Centroid diagrams for two different electron configurations in the isoelectronic
sequence Pd I, Ag II, Cd III, In IV, and Sn V. {After J. E. Mack.)

of this, two centroid diagrams for the levels arising from the electron
configurations 4dB5s and 4d95p in the sequence Pd I, Ag II, Cd III,
In IV, and Sn V are given, following Mack,2 in Fig. 17.17. Term
separations are plotted vertically against atomic number Z horizontally.
The straight horizontal line serves as an origin and represents the center
of gravity of each system of levels, each level being assigned the usual
weight 2J + 1. Since the intervals increase rapidly from element
to element the 4dB5«, iDi — *D8 interval is made the same for all elements,
and all levels of the same element are plotted to that scale. Since the
observed intervals show that the coupling is somewhere between LS-
and jj-coupling (see Fig. 14.20), the terms are given L£-coupling designa¬
tions on the left and ^‘-coupling designations on the right.
1 Mack, J. E., O. Laporte, and R. J. Lang, Phys. Rev., 31, 748, 1928.
,
* Mack J. E., Phye. Rev.t 34 17, 1929.
Sec. 17.8] ISOELECTROS 1C SEQUENCES 351

From the Zeeman effect of various levels in the Pd I and Ag II f

spectra, as well as from the relative trends of the different levels across
the diagram, one finds that the coupling is becoming more nearly of the
jj type with increasing Z. For good L/S-coupling the singlets should lie
far above the triplets, and the triplets should follow the interval rule.
For good jj-coupling the terms should form two groups for the first
diagram, and four groups for the second diagram, around the points
shown at the right. These points may be obtained from the spectra
of the once more ionized atoms of these same elements, as well as those
in the next lower state of ionization, where one can measure directly
the interaction energy of 4dp and 5p> as given by the doublet intervals
2D§,| and
Similar diagrams constructed for the corresponding sequences of
atoms in the first and third long periods of the periodic table show the
same type of trends as are shown in Fig. 17.16. A comparison of the
three periods shows that more nearly jj-coupling arises in the heavier
elements Pt I, Au II, Hg III, etc., whereas more nearly L&-coupling
arises in the lighter elements Ni I, Cu II, Zn III, etc. (see Fig. 14.20).
 CHAPTER XVIII

HYPERFINE STRUCTURE

18.1. Introduction.—It is indeed difficult to say just when, where,

and by whom hyperfine structure was first observed. For want of a
more exact starting point we may begin with Michelson,1 Fabry and
Perot,2 and Lummer and Gehrcke,3 who constructed and employed
various types of interferometers and successfully showed that many

Fia. 18.1.—Photographs of hyperfine structure typical of many of the lines in praseo¬

dymium, reproduced from original negatives taken in the fourth order of the 75-ft. spectro¬
graph on Mt. Wilson. (After White.)

spectrum lines are not single but made up of a number of components.

In 1909 Janicki4 published an extensive account of his observations
and discovery of hyperfine structure in the spectrum lines of a large
number of elements. This structure was later verified in detail by
Wali-Mohammed.5 As an example of the appearance of hyperfine-
structure patterns, photographs of a number of lines arising from the
rare-earth element praseodymium are reproduced in Fig. 18.1. These
1 Michelson, A. A., Phil Mag., 31, 338, 1891.
* Fabby, C., and A. Perot, Ann. de chim. et phys. (7), 12, 459, 1897.
8 Lummer, O., and E. Gehrcke, Ann. d. Phys., 10, 457, 1903.
4 Janicki, L., Ann. d. Phys., 29, 833, 1909.
8 Wali-Mohammed, Ch., Astrophys. Jour., 39, 185, 1914.
352
Sec. 18.1] HYPE MINE STRUCTURE 353

photograms represent one of the most commonly observed forms of

hyperfine structure, known as flag patterns, in which each line contains
several components degrading in intensity and interval toward longer
or shorter wave-lengths. The photographs are not to be interpreted
to mean that ail the lines of a given element are of the same flag-pattern
type nor that each pattern contains the same number of components.
On the contrary many line patterns in a given element are quite different
from one another, both in spacing and in the number of components.
In terms of energy level diagrams hyperfine structure in a spectrum
line is interpreted to mean that either the initial state, the final state,
or both the initial and final states of the atom are not single but contain
a definite structure. Early in the development of this field of investiga¬
tion it was believed that each observed component of a pattern belonged
to a different isotope of the same chemical clement. Just as soon as
hyperfine structure (hereafter to be abbreviated hfs) was discovered in

Hyperfine - Structure Ta nt a I u m A 5997

it i i i

ft ‘"Till iEMI

Isotope - Structure Tungsten A 5225

Fig. 18.2.—Photographs illustrating (a) hyperfine structure of a spectrum line as
observed in tantalum, and (b) isotope structure as observed in tungsten. (After Grace,
More, MacMillan, and White.)

elements known to have but one isotope this hypothesis \\as abandoned,
as is always done in science, for a newer and better theory.
Apparently the first suggestion that hfs is to be attributed to a small
magnetic moment associated with the nucleus was made by Pauli,1 and
later, but independently, by Russell.1 Experimentalists were not long
in showing that this hypothesis, in most cases, is apparently correct.
Since the time that this theory was accepted generally, an isotope effect
has been observed in some spectra so that now there are two types of hfs
to be distinguished from each other. First, there is a hfs due to a nuclear
magnetic and mechanical moment and, second, a hfs due to the different
isotopes of the same chemical element. The first of these will be referred
to as hfs and the second as isotope structure. In some elements hfs
alone is observed, in others isotope structure alone is observed, and in
still others both are observed.

1 Pauli, W., Naturwissenschaflen, 12, 741, 1924. The suggestion that hfs is due
probably to a spinning proton was suggested by H. N. Russell to W. F. Meggers and
K. Burns, Jour. Opt. Soc. Amer.f 14, 449, 1927.
354 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVIII

Photographs of two spectrum lines are reproduced in Fig. 18.2.

The tantulum line illustrates hfs of the flag-pattern type with eight
components due to a large nuclear moment, and the tungsten liDe illus¬
trates isotope structure due to the three even isotopes of tungsten.
Taken with Fabry-Perot 6talons these pictures differ from direct photo¬
graphs of the type given in Fig. 18.1 in that each pattern is repeated
several times for each line. Although there is yet much to be learned
from an experimental and a theoretical standpoint, considerable progress
has been made in the analysis and interpretation of hfs. We shall start
therefore by giving the simple treatments of the subject, comparing as
we go the predicted results with observation.
18.2. Hyperfine Structure and the Lande Interval Rule.—The very
excellent experimental observations by Back, of the hfs in the spectrum
lines of bismuth, and its beautiful interpretation by Goudsmit,1 revealed

Flo. 18.3.—Orbital and vector models illustrating the coupling of a nuclear moment I *
with an electron moment/* to form a resultant F.*

for the first time that a new quantum vector should be added to the
atom model, and that the Land6 interval rule for fine structure also
applies to hfs. The model proposed is essentially the one shown in
Fig. 18.3, where the quantum vector/* of one electron, or J* representing
the total mechanical moment of all the extranuclear electrons, is coupled
with the quantum vector /*, representing the total mechanical moment
of the nucleus (I*h/2w) to form a resultant F*. It is this resultant F*
that now represents F*h/2ic, the total mechanical moment of the atom,
in place of J* as previously stated.
Goudsmit and Back have shown that just as the interaction energy
between L* and S* is proportional to the cosine of the angle between
them [see (Eq. 8.18)], so the interaction energy between the nuclear
moment J* and the electron moment J* is given by

IV - A7*7* cos (/*«/*), (18.1)

1 Goudsmit, S., and E. Back, ,
Zeits. f. Phys43, 321, 1927: 47 174, 1928 .
Sec. 18.2] HYPERFINE STRUCTURE 355

where A'I*J* is constant for each given fine-structure level «/, and
A' is a measure of the strength of coupling between /* and J*. Just
as in fine structure the starred quantities are given by

J* = VJ{J + 1), I* = VTU + 1), F* = VFW + 1), (18.2)

where I takes one value only and is half or whole integral valued, and
F takes values differing from each other by unity from the value I — J
to I + J when I ^ J, and from J — I to J + I when J ^ 7. The cosine
term in Eq. (18.1) is evaluated by the use of Fig. 18.3 in exactly the same
fashion as the cosine in LS-coupling for fine structure (see Sec. 12.6) and
gives
IV = §A'(F*2 - 7*2 - J*2). (18.3)
This is the interval rule for hfs energy levels.

Fig. 18.4.—Vector diagram illustrating graphically the interval rule for hyperfine struc¬
ture. Specific example of a level, where J is taken as $.

As an example of the application of Eq. (18.3), consider the 6p3,

term in the bismuth atom where J is f and the observed value of
I is The allowed values of F are 2, 3, 4, 5, 6, and 7, for which the
IV values are —545-A', —^A', —\TA', — JA', -H^A', and +-\5-A'. A
graphical representation of this splitting is shown in Fig. 18.4. Drawn
to scale, the projection of J* on I* gives J* cos (J*/*), which is pro¬
portional to IV. The differences between levels are seen to be 3A',
4A', 5A', 6A', and 7A' values proportional to the larger of the F values.
This result may be derived in general by substituting first F + 1, then F,
in Eq. (18.3) and taking the difference. Carried out, this gives
Ar = A'(F -f 1). (18.4)
356 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVIII

The observed intervals in the 2D\ term of bismuth are

Av = 0.563, 0.491, 0.385, 0.312, and 0.256 cm"1

with ratios closely 7:6:5:4:3.

With each fine-structure term composed of several hfs terms, some
means of identification or designation is desired to distinguish between
the latter. Since the I value of a given atom is the same for all terms in
all states of ionization, it is suggested that the new quantum number F
be given by a small subscript to the left of the term type as shown in
Fig. 18.4. The complete designation of the top term in the figure would
therefore be 6p8,?DJ in LS-coupling and 6p3,7CHf)f *n jj-coupling.
Leaving off the electron configuration, a hfs line would be designated for
example by JPj — $D\ in L/8-coupling and by «CHf)i — in
^/-coupling. The importance of this subscript lies in the fact that F
represents the total angular momentum of the atom.
In the analysis of many spectrum lines the hfs of some levels is found
to be normal, i.e., smallest F value deepest, while in others it is inverted,
i.e., largest F value deepest. In the first instance the coupling constant
A' of Eq. (18.3) is positive and in the second instance negative. The ques¬
tion of how both positive and negative values of A' arise from the same
element and often in the same multiple term will be taken up in the
following sections.
As an example of normal and inverted hfs a schematic diagram of
the bismuth line X4722 is shown at the left in Fig. 18.5. The observed
pattern of lines arising from transitions between the two sets of levels
is shown below. The selection rules for F in hfs are just the same as
those for J in fine structure, viz.,

AF = 0, ± 1. (18.5)

Because of the ideal coupling between I and J in nearly all hfs,

the intensity rules and formulas derived for LS-coupling, and given
in Eqs. (12.31) and (12.32), may be applied to hfs by replacing the
electron-spin resultant S by the nuclear spin I, the orbital resultant
L by the electron resultant J, and the electron resultant J by the atom
resultant F.1
With a nuclear spin I = f and a J value of i, the initial state in Fig.
18.5 is split into two levels F = 4 and 5. Again with I = and J = £ the
final state is split into four levels F = 3, 4, 5, and 6. [The fine structure
in Bi does not show L>S-coupling but very good .^-coupling (see Fig. 14.21).
The transition 7s to 6p with J changing from ^ to f is an allowed transi¬
tion.] The observed line intensities are shown at the bottom of the

1 For a derivation of hfs intensity rules see E. L. Hill, Proc. Nat. Acad. Sci., 15 ,
779, 1929. Intensity tables are given in the Appendix.
Sec.'18.2] HYPERFINE STRUCTURE 357
figure by means of the microphotometer curve, for comparison with the
calculated intensities given just above.
The second hfs pattern in Fig. 18.5 is drawn for one of the two reson¬
ance lines of manganese, an intercombination line, X5394. With an
I value of 4, each level J = f and 4 is split into six components as shown.
Applying the relative intensity formulas to the allowed transitions, the
theoretical pattern shown near the bottom is obtained. For want of
greater resolving power this very narrow pattern is only partially resolved
as shown by the microphotometer curve at the bottom of the figure.
In making an analysis of the hfs of an uncharted spectrum, patterns
such as we have shown here are all one has to work with. Where all

6p27S,(H!)l ” .830
l A
4

iii i 1 1

3 J-
eps. Gil) 3 I.-1 - =;||
V

III 11

AABismuth A 4722 A 5394 Manganese

Fit;. 18.5.—Schematic diagrams of the hfs levels and patterns of (a) X4722 in bismuth
(after Back and Goudsmit), and (b) X5394 in manganese (after White and Ritschl).

components are completely resolved, as is easily accomplished in a simple

and wide pattern like Bi, equal frequency differences enable one to
construct hfs energy levels for the initial and final states. If there
are more than two levels, the interval rule may be applied and the F
values written down from the interval ratios. Knowing the J values
from the fine-structure analysis, the nuclear spin I is obtained directly.
These are often the steps taken in the determination of the nuclear
spin I.
It should be noted that, if I ^ /, the level will be broken up into
27 + 1 hfs levels, whereas, if / ^ J, the level will be broken up into
2J + 1 hfs levels. This rule shows that, if the J values of a line are
known to be large, the nuclear spin may be determined directly from the
number of lines or levels and in some cases from an incompletely resolved
pattern. A good example of this is to be found in praseodymium (see
Fig. 18.1) where all lines that have been resolved as shown are flag
358 INTRODUCTION TO ATOMIC SPECTRA (Chap. XVIII

patterns with six components. This can be interpreted to mean only

that for Pr, I = f.
18.3. Nuclear Interaction with One Valence Electron.—Calculations
of the interaction energy between a single arbitrary electron and a
nucleus with a mechanical and magnetic moment have been made by
Fermi,1 Casimir,2 Hargreaves,8 Breit,4 and Goudsmit.6 The treatment
to be given here is essentially the semiclassical treatment given by
Goudsmit which leads, with suitable modifications, to the quantum-
mechanical results obtained by the others. It should be pointed out at
the outset that the treatment does not apply to the special case of an
s electron. Suitable changes, however, will make the results applicable
even here.
The interaction of a single electron with the nucleus (see Fig. 18.3)
may be divided into two parts: (1) the interaction of the orbital motion l*t
with the nuclear moment /*, and (2) the interaction of the electron spin s*
with I*. Applying classical electromagnetic theory, as has been done
in Sec. 8.6 for the spin-orbit interaction, the electric field at the nucleus
due to the electron at a distance r is given by

Similarly the magnetic field at the nucleus due to the orbital motion
of the electron is

H = E X
c
Substituting Bohr's relation for the angular momentum
2xmr X v — l*h,
the magnetic field becomes

H - <18«

where m and e are the mass and charge on the electron, and r is the
electron-nuclear distance. Since r is not constant in any orbit, either
on the classical theory or on the quantum mechanics, (1/r3) must be
averaged. Now the nucleus with a mechanical moment I*h/2ir and a
magnetic moment /*/ tends to carry out a Larmor precession around this
field with an angular velocity wL given (see Secs. 3.7 and 8.6) by the
1 Fermi, E., Zeits. f. Phys., 60, 320, 1930.
* H. B. G. Casimir, quoted in L. Pauling and S. A. Goudsmit, “The Structure of
Line Spectra/1 p. 208.
* Hargreaves, J., Proc. Roy. Soc.f A, 127, 141, 1930.
4 Breit, G., Phys. Rev., 37, 51, 1931.
‘Goudsmit, S. A., Phys. Rev., 37, 663, 1931; see also Pauling, L., and S. A
Goudsmit, “The Structure of Line Spectra," p. 204.
Sec. 18.31 HYPERFINE STRUCTURE 35§

product of the field strength H and the ratio between the magnetic and
the mechanical moment. For this latter ratio we write
Mi _ „ e
(18.7)
I*h ^I2mc
2r

where c/2me is the classical ratio and gt is an unknown factor called the
nuclear g factor. If the nuclear moment were due to a proton in an orbit,
gi would be expected to be that of an orbital electron. If it is due
to a spinning proton, gi should be between and 7 times this.1
The precessional angular velocity becomes therefore

e2 l*h(l\ „oox
Ul 9'2mV 2jtVV 18'8)
The interaction energy is now given by the product of ^ and the
projection of the nuclear mechanical moment I*h/2ir on l*f
e2 l*hf 1
(18.9)

Since l* precesses around j*f and j* and I* in turn precess around

their resultant F*f the above cosine must be averaged. Since the
first precession is very much more rapid than the latter, we may, in the
usual manner, project l* on j*} then j* on I* and get for the average
cosine,
cos (1*1*) — cos (I*j*) cos (l*j*). (18.10)
Substituting in Eq. (18.9) this gives

th( 1
(18.11)
-W cos ( wcos
Since both of these angles are fixed, the cosines are readily evaluated in
terms of the quantum vectors s*, l*} j*, I*, and F*. This will be done
after we obtain the interaction energy for the electron spin s* and the
nuclear spin 7*.
According to classical electromagnetic theory the mutual energy
of two magnetic dipoles with moments hi and and at a distance r apart,
is equal to

Wi,8 = ^p{cos (y/v«) — 3 cos (y/r) cos (y.r)). (18.12)

r°
1 Recent Stem-Gerlach and Breit-Rabi experiments with hydrogen have shown
that the nuclear magnetic moment of hydrogen lies between 2\ and 31 nuclear
magnetons; one nuclear magneton being defined as eh/AxMc, where M is the mass
of the proton. Because the magnetic moment of a proton appears to be considerably
larger than one nuclear magneton the term proton magneton frequently used for this
same unit of magnetic moment is misleading.
360 INTRODUCTION TO ATOMIC SPECTRA [Chap. XV11X

The nuclear magnetic moment fit is given by Eq. (18.7), and the
magnetic moment of the spinning electron by

-2;(_3* (18.13)
2me 2t
The mean value of the term in braces, involving the electron-nuclear
distance r, is readily shown by the use of direction cosines to be1
—J cos cos (j*s*) — 3 cos (j*l*) cos (18.14)
Inserting these results in Eq. (18.12),

12—s*-(-V cos (/V*){cos (j*s*)

V
2ir 2mc 27r\r3/2
—3 cos cos (s*l*)}. (18.15)
Adding the two interaction energies [Eqs. (18.11) and (18.15)], the total
interaction for spin and orbit becomes
IV = Wlti + W,,, = afI*j* cos (I*j*) = Jo'(F*2 - Z*2 - J*2), (18.16)
where the coefficient o' is given by

a' = fir/(
cos (l*r) + w* 008 (s*r)

cos (j*/*) cos (s*Z*) (18.17)

4? }■
Upon substituting the quantum-mechanical value of (1/r3) (see
Secs. 4.9 and 8.6),
, _ Rhca.2Zz
(18.18)
° ~ 9,nH(l + m + 1)
For a given spectral term, i.e.9 given s, l, and j, a' is a constant.
Fermi, Hargreaves, and Breit have shown from a quantum-mechanical
treatment, and Goudsmit has shown from the classical theory of energy
sums, that the terms in the braces of Eq. (18.17) are to be replaced by
1*2/j*2. This changes the a' coefficient of Eq. (18.18) to
Rhca2Z3
(18.19)
a'
w+ii/cr+i)
Dividing by he to obtain o' in cm-1:
, _ Ra2Z3
(18.20)
° + W +1)'
According to Eq. (18.7), the nuclear g factor is given by the ratio
between the nuclear magnetic moment in Bohr magnetons, eA/4irmc, and
the mechanical moment in units of h/2r. If gt is to be expressed in
1 See Paulino, L., and S. A. Goudsmit, “The Structure of Line Spectra,” p. 206.
Sec. 18.4] HYPERF1NE STRUCTURE 361
nuclear magnetons (eh/AwMc), where M is the mass of a proton, gt in
Eq. (18.20) must be divided by 1838, the ratio of the mass of the proton
to the mass of the electron. The result is

Qi . Ra2Z3
cm- (18.21)
1838 n\l + i)j(j + 1)

This af is the A' of Eqs. (18.1) and (18.3) for hydrogen-like atoms. If
gr is positive, a' is positive and the hfs terms will be normal. If gj is
negative, a' is negative and the hfs terms will be inverted. If the nuclear
magnetic moment is due to a spinning proton, the gt factor would be
expected to be between 4| and 7.1
As already pointed out, the above equations have been derived
for hydrogen-like atoms. Since hfs has not been observed in these
simplest of atomic systems we must look to the more complex spectra
and the heavier elements for a comparison of theory with experiment.
18.4. Nuclear Interaction with a Penetrating Electron.—In a classical
treatment of fine structure Landd showed, by the use of an idealized
model, that Z4 in the resultant equation should be replaced by Z\ times
Z2. The alkali doublet separations, for example, are given by Eq. (8.32),

= Im) cm-1> <18-22>

where Zx is the effective nuclear charge inside the core of closed electron
shells, and ZQ the effective nuclear charge outside. By analogy with
this equation and its derivation, Z3 in Eq. (18.21) is broken up into
two parts so that the interaction energy of a single valence electron,
screened from the nucleus by a shell of electrons, is given by [see Eq. (18.3)]
Tf = W(F*2 - I*2 - J*2), (18.23)

where gj Ra2ZtZ2
(18.24)
1838 ' nl(l + $)j(j + 1)'
1 In any experiment performed for the purpose of determining the magnetic and
mechanical moments of the nucleus, m and I*h/2ir are so oriented with the field, as
in the Zeeman effect, that what one measures is the component of m along the field H.
From the experimental value of this component the moments and I*h/2ir are calcu¬
lated from theory. In Frisch, Stem, and Estermann’s experiments on the hydrogen
molecule a magnetic moment of approximately 21 ± 10 per cent nuclear magnetons
has been measured for the proton. By a different method Rabi, Kellogg, and
Zacharias find a value of 31 ± 10 per cent nuclear magnetons. The agreement
between these two values must be regarded as more important than their difference.
Dividing by / = J gives a value of qi from 41 to 7 for the proton. This comes from
Eq. (18.7), for, if the ratio jJfcjfiic g*ves ^ie 9* factor, then the ratio of their com¬
ponents along any line will also give the same ratio.
See Frisch, R., and O. Stern, Zeits. f. Phys., 86, 4, 1933; Estermann, I., and
O. Stern, Zeits. f. Phys., 86, 16, 1933. Rabi, I. I., J. M. B. Kellogg, and J. R.
ZXcharias, Phys. Rev., 46, 157, 1934.
362 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVIII

The latter equation is somewhat simplified by substituting the

observed fine-structure doublet separation [Eq. (18.22)]. This gives
, _ 9i A vl(l + 1)
(18.25)
1838 ’ Z%(1 + i)j(j + 1)’

A graphical comparison of this equation with the observed hfs of

the 7p,2Pj,j levels in doubly ionized thallium is made in Fig. 18.6. The
calculated hfs separations are shown in Fig. 18.6B, the observed separa¬
tions in Fig. 18.6D. The relative intervals are not in good agreement.
It should be noted that, since gt for the T1 nucleus is unknown from
theory, the lower level in each part of the figure has been taken arbi¬
trarily to have the same separation as the observed.

(A) (B) (C) (D)

Obs. Calc. Calc. Obs. 2P
-
-Z-
v2" -1 437cm-1 .
<>/2
-\\
TlDr’ 7P~Wcm“
-a
-r»Tt
.375 cm"1
-
Fia. 18.6.—Graphical comparison of calculated and observed hfs in thallium. (A)
Observed doublet separation. (B) Relative separations according to Eqs. (18.23) and
(18.25). (C) Relativity corrections added. (D) Observed hfs separations.

Breit1 and Racah2 have shown that relativity corrections are differ¬
ent for the two levels of a doublet and must be taken into account. The
relativity corrections to be added to Eq. (18.25) are obtained by multi¬
plying a' by k/X, where

* “ p2 = + *)2 ~(aZ,)!’ (18-26)

* = + V1 - WJ* - 1 - IP - W)»]. (18.27)

Equation (18.25) becomes therefore

, _ Qi AvlQ, + 1)k
(18.28)
0 1838 ’ Z,(l + \)j(j + 1)X

Substituting the known values in these equations and assuming the

same separation for the 2Py hfs, the 2P\ separation in Fig. 18.6C is
obtained. Although there is better agreement with observation, the
discrepancy is still rather large.

1G. Breit, see Goudsmit, 8. A., Phys. Rev., 43, 636, 1933.
* Racah, G., Zeits. f. Phys., 71, 431, 1931.
Sec. 18.5J HYPERFINE STRUCTURE 363

As stated above, the equations thus far developed should not apply
to s orbits. McLennan, McLay, and Crawford1 have shown, however,
that the equations do seem to be approximately correct even for this
special case. Goudsmit has suggested that we write for s electrons,

/ = Jl_ &R**Z%Zl (18.28a)

! 1838' 3n\ *

As examples of the application of Eqs. (18.28) and (18.28a), we

substitute observed hfs separations from a known spectrum and calculate
nuclear gr factors. Such calculations aie given in Table 18.1 for the
doubly ionized atom of thallium. The observed hfs separations are
given in Col. 2. With a nuclear moment I — the a: factors, given in
Col. 3, are obtained from Eq. (18.23). Many calculations of Zx from
fine-structure term values show that, throughout the periodic table,

Table 18.1.—Nuclear gf Factors Calculated from the Observed Hfs in Tl III

Term Av a' K X zx z* gi

6 s, 2Sh 6 4 cm-1 6 4 cm-1 2 31 81 3 3 7

7s, 2Sh 1 37 1 37 2 31 81 3 2.94
7Pi *P\ 0 375 0 375 2.08 1 17 77 3 2.96

Z% Z for s electrons and Z» ^ Z — 4 for p electrons. The lack of

agreement between the values obtained in the table as well as in other ele¬
ments in the periodic table indicates that the above presented theoretical
treatment of hfs is not complete and must still be modified further. For
this reason too much weight should not be given to such calculations.
It should be noted, however, that the magnitude of the calculated gt
factors should be about right and is about what one would expect if the
nuclear moments are to be attributed to positively charged particles in
the nucleus.
18.6. Classical Explanation of Normal and Inverted Hfs.—Classically
it is not difficult to show that for each level of a fine-structure doublet
arising from a single valence electron the orbital interaction of the electron
with the nucleus tends to produce a normal hfs splitting of both levels.
The electron-spin interaction with the nucleus, however, tends to invert
the hfs of the term j = l + \ and to widen that of j — l — This
may be seen from the classical orbital model in Fig. 18.3. The orbital
motion of the electron produces a magnetic field at the nucleus which
is opposite in direction to its mechanical moment l*h/2w. The most
stable state for the nucleus in this field, and therefore the state of least
1 McLennan, J. C., A. B. McLay, and M. F. Crawford, Proc. Roy. Soc., A, 138,
652, 1931.
364 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVIII

energy, will be the one for which the nuclear magnetic moment is most
nearly parallel to the field of the electron. If this magnetic moment
is due to something like an orbital or spinning proton, this will be the
state for which the mechanical moments I* and J* are oppositely
directed. These are the hfs states fP3 and §P* in Fig. 18.6.
If we now think of the spinning electron and the nucleus as small
magnets acting on each other at some distance from each other, the
most stable position of the two will be that for which their magnetic
moments are in opposite directions and the mechanical moments are
in the same direction. These correspond to the |P3 and \P$ states
in Fig. 18.6. Since the orbital interaction with the nucleus is greater
than the spin interaction, the hfs of both levels is normal. Although
the greater splitting is observed in the 7p,2P3 level in T1 III (see Table
18.1) the o' coefficient, which measures the total interaction with the
nucleus, is greatest for the 7p,2P* level.
Fermi has shown from a quantum-mechanical treatment of $ electrons
that the interaction with the nucleus is opposite in sign to that of the
spin of an arbitrary electron, i.e., it has the same sign as an orbital
interaction. Classically, therefore, an s electron may be thought of
as a cloud of negative charge, distributed as shown in Figs. 4.6 and 4.7
and rotating about the nucleus. A positive charge rotating at the
center of this spinning negative cloud is most stable when the magnetic
moments are parallel and the mechanical moments are opposite. This
shows that s electrons also give rise to normal hfsf as observed. Because
of the deep penetration of an s electron on the classical theory, or the
large probability density near the nucleus on the quantum-mechanical
model, the interaction between s* and I* and hence the hfs for 2&j states
should be very large.
18.6. Hyperfine Structure in Atoms with Two or More Valence
Electrons.—In 1927 Meggers and Burns1 published an account of the hfs
in the lanthanum spectrum, at which time they pointed out that the
largest structures were always observed for configurations involving a
single unbalanced s electron. They also pointed out at the same time
that, while certain levels revealed wide structures, others of the same
multiplet were either single or else very narrow and unresolved. In
the 685d,3Di,2,3 terms of La II, for example, the two outer levels 3Di
and 8Dj show a wide splitting, whereas the middle level 3Z)2 shows
practically none. An explanation of this anomaly in terms of the vector
model has been given by White,2 and equations for the splitting have
been given by Goudsmit and Bacher.3
1 Meggers, W. F., and K. Burns, Jour. Opt. Soc. Amer., 14, 449, 1927.
* White, H. E., Phys. Rev., 34, 1288, 1929; 34, 1397, 1404, 1929; 36, 441, 1930;
Proc. Nat. Acad. Set., 16, 68, 1930.
•Goudsmit, S. A., and R. F. Bacher, Phys. Rev., 34, 1499, 1601, 1929.
Sec. 18.6] HYPERFINE STRUCTURE 365
In Fig. 18.7 we picture, on both the classical model and on the
quantum-mechanical model, an atom with two valence electrons, one in
a 'penetrating s orbit and the other in a nonpenetrating d orbit. A vector
diagram for the same atom in LS-coupling is drawn schematically at
the left. From this picture it is easy to see how the penetrating electron
has an opportunity of coupling strongly with the outer electron when
it is outside the core and also with the nucleus when it is inside the core.
Thus the s electron may greatly strengthen the coupling between the
electron resultant J* and the inner nuclear resultant I*. Due to the
screening of the nucleus by the core electrons, the interaction between
the nucleus and the d electron is small compared with that between
sj and I*. For strong coupling with the nucleus the electron must
not only be deeply penetrating but also tightly hound. All s orbits, for

Fio. 18.7.—Vector and orbital model for the interaction of a penetrating s electron and
a nonpenetrating d electron with a nuclear moment.

example, will be deeply penetrating, but only those that spend a con¬
siderable amount of their time near the nucleus will couple strongly with
7* and give wide hfs.
Due to the large spin-spin interaction of the electrons in LS-coupling,
and si precess rapidly around their mechanical resultant S*. Due
to the somewhat weaker coupling between the spin sj and the orbit
ll of the d electron, S* and ll precess more slowly around their mechanical
resultant J*. Finally J* and I* are coupled weakly together and
precess slowly around their mechanical resultant F*. The relative
frequencies of these precessions are given by, and follow directly from,
the observed energy differences, i.e., by the singlet-triplet intervals, the
triplet fine-structure intervals, and the hfs intervals, respectively.
The well-known general relation that the energy of interaction
between any two magnetic moments is proportional to the cosine of the
angle between them (see Sec. 8.6) enables one to write

TF = afI*sl cos (/*s?) (18.29)

for the energy of interaction between an s electron and the nucleat
366 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVIII

moment. Owing to the above mentioned classical precessions, cos

(/*«*) must be averaged. Projecting the rapidly precessing sf on S*,
then the less rapidly precessing S* on J*, and finally the slowly precessing
J* on I*, we obtain

cos (I*sf) = cos (/*/*) cos (J*S*) cos (S*sf), (18.30)

giving for Eq. (18.29),
rF = a'I*s* cos (/V*) cos (J*S*) cos (£*$*)• (18.31)
This may be written in the generally observed form of Eq. (18.1):
Tf = A'I*J* cos (/V*), (18.32)
where

A' = a'^i cos (J*S*) cos (S*sf). (18.33)

Since each of the angles is constant, all three cosine terms are readily
evaluated in terms of the mechanical moment vectors to give for LS-
coupling
Tf = f A'(F*2 - 7*2 - J*2), (18.34)
,J*2 + S*2 - Ip S*2 + sl2-s*2
where (18.35)
A a 2 J*2 2£*2
One interesting result regarding this equation is that A', as shown by
Goudsmit and Bacher, may be simplified by expressing it in terms of the
electron g factor. Substituting f for Si and for s2 and the value of g as
given by Eq. (13.8), the A' factor for L/S-coupling reduces to
A' = W(9 - 1) (18.36)
If the coupling between the s electron and the outer electron is
jjf the average cosine is obtained in a similar fashion by projecting
I* on J*, then J* on sj, to give

cos (I*sf) = cos cos (J*s*). (18.37)

This again leads to Eqs. (18.32) and (18.34), where, for jj-coupling,

,J*2 + sf2 - jV
A' => a'yj cos (J*sf) = a (18.38)
2J*2

As an example of the application of these equations let us first con¬

sider an sd electron configuration in LS-coupling and calculate the relath e
separations of the hfs in each of the 3D terms. Graphically this has
been done in Fig. 18.8 for the case where I = J. The first step is to
calculate the A' coefficient for each of the three terms. The quantum
values to be substituted in Eq. (18.35) are: $i = s2 = i, S = 1, l = 2,
/ = 1, 2, and 3. These give the values A' = — ^a', and fa' for
*Di, 8D2, and SDS, respectively. The same values of A' can be calculated
Sec. 18.6] HYPERF1NE STRUCTURE 367
more quickly from the g factors J, £, and $, and Eq. (18.36) (see Table IV
in the Appendix). Substituting in Eq. (18.34), along with I — \ and
the J values 1, 2, and 3, the relative separations shown in the center of
Fig. 18.8 are obtained. The observed levels plotted at the right are

Fig. 18.8."- -Calculated and observed hfs intervals for an sd configuration *D term in
thallium.

those of the configuration 6s7d in singly ionized thallium (T1 II). The
observation of normal splitting in 3D3 and of inverted splitting in 3Z) i
substantiates here, as it does in many other cases, the supposition that
the penetrating s electron is responsible for the strong coupling of J*
with the nucleus.
Table 18.2.—Coupling Coefficient o' for a 6s Electron in Tl II; Calculated
from the Observed Separations

Configura¬ Configura¬
Terra Ai/(0b. ) A' a' Terra Av(.obm.) A' a'
tion tion

37>a +3 33 +0 95 +5 70 >F< +3 31 +0.735 +5.88

6s6d *d2 +0 55 +0.22 +2 64 6s5/ 3Ft -0.68 -0 194 +4.66
3/>i — 2 12 -1.41 +5 64 -2.49 -0.996 +5.97
,F>
3A, +3 39 +0.97 +5.82 3F4 +3 35 +0.745 +5.96
Qs7d +0 78 +0 31 +3 72 6»6/ -2 24 -0.640 +15.4
3Di -2 21 -1.47 +5 88 >f2 -2 58 -1 030 +6 18

+3 45 +0 99 +5 94
6;8 d 3Z>2 + 1 66 +0 66 +7 92
3/>i -2 25 -1 50 +6 00

Referring to Eq. (18.33) we find that for a given multiple term

like 3D the coefficient a', the spin sf, and cos (£*s?) are constants for
all values of J. The splitting factor A' is therefore proportional to
1/J* cos (,J*S*). The vector diagrams at the right in Fig. 18.8 are
drawn to show this relation. As the J* value decreases, the cosine of the
angle goes from a positive to a negative value. That the splitting in
Tl II is due chiefly to the 6s electron may be shown by substituting the
368 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVIII

observed separations of many different levels in Eqs. (18.34) and (18.35)

and calculating the coefficient a'. This has been done in Table 18.2.
It is seen that the lower level of each triplet shows inverted hfs.
At the same time the values of a' for the lower and upper level of each
triplet are closely the same. The center level in each case is in poor
agreement with the others. This latter is to be expected since the
fine-structure intervals of the triplet show departures from the Land6
interval rule for triplets. Because the two outside levels have the same
g factors in all coupling schemes, they should show agreement if our
assumptions are correct. Goudsmit has pointed out that by taking
all levels with the same J value into account the energy sum rules will
apply to those levels of the same configuration. This has been verified
experimentally by McLennan, McLay and Crawford1 for T1II. Taking
into account only the outside triplet separations in Table 18.1, an average
value o' = 5.90 cm*"1 is calculated as the coupling coefficient for a
6s electron in thallium.
If for a two-electron system both electrons are in s orbits, both
may couple strongly with the nucleus and contribute appreciably to the
splitting. In such cases the vector model of Fig. 18.7 is greatly simplified
and the interaction of each electron with the nucleus will be given by
Eq. (18.29). The two energies when added together and the cosines
averaged and evaluated in the usual fashion give again Eq. (18.34) as the
resultant interaction energy, where
A' - i« + <). (18.39)
Here oj, is the coupling coefficient for one of the s electrons and
a', the coefficient for the other. The derivation of this equation will
be left as an exercise for the reader. If the electrons are in s orbits with
the same total quantum number, there will be but one term, viz,, T$o,
and this cannot possibly show any splitting with J = 0. The only
possibility of observing hfs due to the coupling of two s electrons with I*
will arise in *Si terms where the total quantum numbers are different.
In T1 II the series of 8Si terms arising from the configurations 6sns,
where n = 7, 8, 9, • • • , has been observed to have hfs. Assuming
the value o' == 5.90 cm-1 as the coupling coefficient of the 6s electron
common to each configuration (see Table 18.2), the following values
are calculated for 7s and 9s electrons:

EB
/
Configuration Term
mem
Av

6s7s •Si mm 0.72

6«9s
mm MEM ■ 0.12

1 McLennan, J. C., A. B. McLay, and M. F. Crawford, Proc. Roy. Soc., A, 138,

,
632 1931 .
Sec. 18.7] HYPERFINE STRUCTURE 369

A comparison of the three values 5.90 for 6s, 0.72 for 7s, and 0.12
for 9s indicates the rapid decrease in the coupling with the nucleus
with increasing total quantum number. The conditions for wide hfs
therefore specify that at least one electron be not only deeply penetrating,
as all s orbits are, but that it also be tightly bound to the atom.
18.7. Hyperfine Structure in Complex Spectra.—Attempts to corre¬
late observed hfs in complex spectra with theory have in some respects
been more successful than in the simpler cases of one-electron systems.
As shown in the last section this is due largely to the general observation
that nearly all wide structures and patterns arise from configurations in
which there is a single tightly bound s electron. The classical and
quantum-mechanical model for a complex atom may be represented by
Fig. 18.7. This figure is generalized by letting the sy ly and j of the
nonpenetrating electron represent the resultant Sy Ly and J of several
nonpenetrating electrons. In such a system the penetrating $ electron
couples strongly with the outer electrons when it is outside the core
and with the nucleus when it is inside the core. The treatment for several
electrons will therefore be essentially the same as that given in the preced¬
ing section for two valence electrons. Replacing ZJ, sj, and j* in
Eqs. (18.35) and (18 38) is L£, S*t and J*y the interaction energy for
complex spectra becomes
TV = JA'(F*2 - I*2 - J*2)y (18.40)
where, for L$-coupling,
,J*2 + S*2 - L*2 S*2 + s*2 - S*2
(18.41)
A a 2 J*- 2/S*2
and, for .//-coupling,
j*2 + s*2 - j?2.
(18.42)
A 2J*2
Here SPy LPy and JP represent the resultants of the nonpenetrating
electrons, Si the spin of the penetrating s electron, and S and J the
total resultants of all the electrons together.
Let us now apply these equations to a specific example where hfs
separations have been observed. In the neutral spectrum of lanthanum
the first excited state is a AF term arising from 5eZ26s. Three of the
four levels in this term have quite wide hfs whereas the fourth is rela¬
tively narrow. The observed intervals are shown at the right in Fig. 18.9.
To calculate relative splittings for each level, we first evaluate A' coeffi¬
cients for each. The observed fine-structure intervals of 341.8, 484.6,
and 627.0 cm-1 have the ratios f: \ revealing good L£-coupling between
electrons. From the configuration and term 5d26s,4F we have the quan¬
tum values S = f, si » \y SP = 1, LP = 3, and J * f, $, J, and f.
Substituted in Eq. (18.41), we obtain A' * \a’y t^o', fya', and
370 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVIII

for 4Fg, 4Fj, 4Fj, and 4Fj, respectively. Substituting these in Eq. (18.40)
the over-all separations —a', ^a', 50aa', and ^a', shown at the left
in Fig. (18.9), are obtained. The calculated intervals in the first column
are obtained by computing a constant a' from 4F% and using this for all
four levels. Better agreement between calculated and observed intervals
is obtained by taking the orbital motions of the 5d electrons into account.1
Assuming that the interaction will be given to a first approximation by
a"L*I* cos (L*/*), the energy will be given by Eq. (18.40), where
Af = a"(J*2 + L*2 — S*2)/2J*2. Substituting the AF quantum num¬
bers, the energy contributions ^a", -Mpa", -^a", and ^a" are obtained.
The intervals in the middle column are obtained by calculating constants
o' and a" from 4F# and 4F? and using these for the other two levels.

Fig. 18.9.—Calculated and observed hfs intervals for d2s, 4F, in lanthanum.

Although agreement with experiment is not so good as one would like,

it is certain from the above considerations that the orbital motion of
the two bd electrons contributes considerably to the coupling of the
electron resultant J with the nuclear moment I. The vector diagrams
at the right in Fig. 18.9 show graphically how the stronger coupling of the
8 electron, acting through the spin resultant S*y inverts the hfs of 4F4.
Not all wide hfs patterns arise from configurations involving a single
tightly bound s electron. This fact is illustrated by the SdHs2f2Ds,»terms
of copper and the 6p3,($ $ \)u terms of bismuth. In ail such cases the
binding of the p,d, or / electrons, as the case may be, will be large for at
least one electron. This is shown by the fact that the terms lie deep
on the energy level diagram and the electrons contribute equally to the
coupling of J* with I*. Equations for the interaction of two or more
1 Anderson, 0. E., Phys. Rev., Vol. 46, 1934.
Sec. 18.8] HYPERFINE STRUCTURE 371
arbitrary electrons with a nucleus have been derived by Goudsmit using
the principle of energy sums.1 Although the calculations from these
equations are in good agreement with some hfs terms, Goudsmit points
out that there are unaccountable discrepancies in others. Breit and
Wills2 have brought some of these discrepancies into agreement with
experiment by a quantum-mechanical treatment of the same problem
for coupling intermediate between LS and jj. For these rather com¬
plicated cases of nuclear interaction the reader is referred to the original
articles.
18.8. Nuclear g7 Factors.—A nuclear factor is a number expressing
the ratio between the magnetic moment of the nucleus in nuclear magnetons
(eh/4wMc) and the mechanical moment of the nucleus in quantum
units of h/2ir. From Eqs. (18.28) and (18.28a), the following may be
written:
For arbitrary electron orbits,
= + !)(* + h) 1838 X.
(18.43)
°‘ A vl(l +1) k’
and for s electrons,3
_ 3a'rt31838 1
(18.44)
gi 8Ra*ZlZ2' k
As an example of the application of these equations we shall first
apply Eq. (18.44) to the hfs of T1 II, where the very large separations
are due chiefly to the penetrating s electron. The average value of a'
for the 6s electron is calculated for the values given in Table 18.2 Using
only the outside triplet separations a' = 5.90 cm-1. As pointed out in
Sec. 18.4, Z% for s electrons is about equal to the atomic number Z. ZOJ
the effective charge outside the core, is taken as one for T1 I, two for
T1 II, and three for T1 III. The effective quantum number n0 is given
by the term value of 6s,2£ in T1 III, where the 6s electron is the only
valence electron. The relation T = #Z2/n2, applied to this state, with
T = 240600 cm-1 and Z0 = 3, yields n0 = 2.02. Substituting these
values in Eq. (18.44),
3 X 5.9 X 8.3 X 1838 1
(18.45)
8 X 5.8 X 81 X 9 ’ 2.31
Although the exact value obtained here may not be correct, it is
significant that the value is of the right order of magnitude on the
assumption that nuclear moments arise from protons in the nucleus.
Values of gt computed for a number of elements by Breit, Wills, and
1 Goudsmit, ,
S., Phys. Rev., 37 663, 1931.
2 Breit, G., and L. A. Wills, Phys. Rev., 44, 470, 1933.
3 These equations have been given by S. A. Goudsmit as fair approximations for
,
the gr factors, Phys. Rev., 43 636, 1933.
372 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVIII

Goudsmit, and independently by Fermi and Segr6 by a somewhat

different method, are given in Table 18.3, along with values of nuclear
moments. The values of I determined from band-spectrum intensities
alone are given in parentheses. In many cases, only a lower limit for I
has been set. A negative gt factor suggests the possibility that the
nuclear spin is due to negative charges, although this is not necessary.

Table 18.3.—Table of Nuclear Moments

Element Element
/
and atomic Isotope I Qi M/ and atomic Isotope / ffi
number number

H 1 1 (1) 5.0? 2.5? Cb 41 93 I 0.8 3 5

2 (5) 0.7? 0.7? Rh 45 103
He 2 4 0 \ 110, 12, |( Q
Li 3 7 1 2.2 3.3 Cd 14, 16, r
48 1
C 6 12 (0) ( 111, 13 i -1 0 -0 5
N 7 14 (S) In 49 115 9 1 2 5 4
0 8 16 (0) Sn 50 117, 119 I -1 8| -0 9
F 9 19 1 i 121 1 1.1 2 7
Sb
51 1
Ne 10 20, 22 0 l 123 l 0 6 2 1
Na 11 23 8 1.4 2 1 I 53 127
2 l
A1 13 27 i 4.2 2.1 Cs 55 133 I? 0 7 2 6
P 15 31 (1) Jla ) 136, 38 0
S 16 32 0 135, 37 V 0 4 1 0
Cl 17 35 a) La 57 139 l
Ca 20 40 0 Pr 59 141 l
V 23 51 ki Eu 63 k'i7
Mn 25 55 l Tb 65 kl
Co 27 59 l 0.8 2.8 Ho 67 ki
Cu 29 63, 65 l 1.5 2 3 Lu 71 ki
Zn 30 67 1 Ta 73 181 1
8
Ga 31 < 69 2 1.4 2.1 w
w
182, 4, 6 0
74 J
} 71 l 1.8 2.7 74 1 183 ?
As 33 75 I Re 75 185, 87 1
Br 35 79, 81 I Au 79 197 •» 1 2 1 8
2
Kr 36 83 kl ( 198, 200
{ n
Eb 37 V 85 1 0 5 1 3 202, 204 r
QA
Hg oU <)
\ 87 1 1.9 2.8 199 i 1 1 0 5
Sr 38 87 V ( 201 3
-0 4 -0 6
T1 81 203, 05 I 2.8 1 4
Pb 82 207 i 1.0 0 5
Bi 83 209 0 8 3 5

Just how far the above equations can be expected to give correct gt
factors is a question that will be answered when we know more con¬
cerning the structure of the nucleus. From the recent experiments of
Sec. 18.9J HYPERFINE STRUCTURE 373
Frisch, Stern, and Estermann as well as Rabi, Kellogg, and Zacharias1
there is some evidence that the magnetic moment of the neutron is
negative. This should play an important rdle in any new theory of
nuclear structure.
18.9. Zeeman Effect in Hyperfine Structure.—We now turn to the
Zeeman effect of hfs, a subject which had its beginning with a study
of the bismuth spectrum by Back and Goudsmit. The paper published
by them in 1928 is considered a classic in the annals of spectroscopy.2
At this early stage in the development of hfs, in general, it was shown
by Back and Goudsmit that all of the simple classical derivations and
formulas for the anomalous Zeeman effect and the Paschen-Back effect
could be carried over to hfs by making the following substitutions:
I, J, F, gh gJy gF] Mh Mj, and MF
for S, L, J, gs, gl, gj, Ms, ML) and Mj, respectively.
Unfortunately the so-called Zeeman effect of hfs has not yet been
observed, chiefly because of the lack of sufficient resolving power in
the optical instruments thus far employed. What has been observed
by Back and Goudsmit, however, is an effect similar to the so-called
Paschen-Back effect in fine structure. Since there is good reason to
believe that the hfs Zeeman effect will be observed and studied in the
near future, a simple treatment of the subject
will be given here. H
In a very weak magnetic field the atom
may be thought of as precessing as a whole
around the field direction H subject to the
quantum conditions that the projection of
the mechanical resultant F*h/2w on H is equal
to MFh/2ir. The magnetic quantum number
Mf takes values differing from each other by
unity ranging from MF = —F to MF = +F.
A classical vector model for very weak field
is shown in Fig. 18.10. To calculate the mag¬
netic energy it is necessary that we first com¬ Fio. 18.10.—Vector model
pute the magnetic moment of the atom as of an atom with a nuclear
spin in a very weak magnetic
a whole. This moment will be made up of field (hfs Zeeman effect).
two parts or components: iij, the magnetic
moment of the extranuclear electrons, and /zz, the magnetic moment
of the nucleus. These moments [see Eqs. (13.1) and (13.6)] should be
given by
1Frisch, R., and O. Stern, Zeits. f. Phys., ,
85 4, 1933; Estermann, I., and O.
. ,
Stern, Zeits. f Phys., 85 17, 1933; Rabi, I. I., J. M. B. Kellogg, and J. R.
Zacharias, , ,
Phys. Rev., 45 769, 1934; 46 157, 1934.
* Back, E., and S. A. Goudsmit, ,
Zeits. f. Phys., 47 174, 1928.
 374 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVI11

and * = 9tI*£^ <18-46>

Since it is the mechanical moments 7* and J* that are quantized,
the total magnetic moment of the atom is just the projection of these
magnetic moments on F* (see Sec. 10.3). As hi and 7* are parallel to
each other, and hj and J* are oppositely directed, the resultant atomic
magnetic moment hf becomes
I*f = jMj cos (J*F*) — fij cos (I*F*). (18.47)
Upon substituting from Eq. (18.46),

IXF = [gjJ* COS (J*F*) - g,I* cos (18.48)

As in the Zeeman effect of fine structure, the term in brackets is placed

equal to F* times a constant gF, i.e.,
gFF* - gjJ* cos (J*F*) - gh7* cos (7*7’*). (18.49)
This gives
ph

^ = 9fF*A^Tc (18-50)

Referring to the vector model again, we may now evaluate the cosines
in terms of the mechanical vectors and obtain, for Eq. (18.49),
F*2 _|_ J*2 _ 1*2 f*2 _|_ J*2 __ J*2

gr gj 2F*2 ^ 2/^*2 * (18.51)

Because of the general observation that hfs separations are many times
smaller than the fine structure the gi factor for the nucleus must also
be very small compared with the gj factor for the extranuclear electrons.
If we make the assumption that gi << gJ) we may write
F*2 -b J*2 - 7*2
gF _ gj-2F*2-■ (18.52)

If the nuclear spin is due to one or more particles with the mass
of protons, the nuclear gi factors should be in the neighborhood of one
or two thousandths of those for electrons. By Eq. (18.50) the ratio
between the magnetic and mechanical moments of the entire atom is
therefore closely represented by

Hf e e F*2 + J*2 - 7*2

F*h 9f2me °J2mc' 2F** ' (18.53)
2r

According to Larmor's theorem, F* in a very weak field will carry

out a classical precession around the field direction, the angular velocity
of which ^should be given by H times the ratio between the magnetic
and mechanical moments:
Sec. 18.9] HYPERF1NE STRUCTURE 375

(*>F = H -g> (18.54)

2 me

The energy of this precession is given by oof times the component of the
mechanical resultant F*h/2w along H:

AW = cos (F*H) = H ■ gr^ ~ cos (F*H). (18 55)

In terms of the magnetic quantum number MF,

<18M)
Dividing by Ac, the energy in wave numbers becomes
Tip
— ATf = gF • Mft-cm-1 = gF - Mf * L cm-1. (18.57)
47rmc

As an example of the magnetic splitting given by these equations,

(onsider the transition 2£j — 2P$ in an atom where the nuclear spL*

Fio. 18.11.—Zeeman effect of hfs in a very weak magnetic field.

I = Specifically the fine structure arising from the transition 7s,

*Sk — 7p,2P* in doubly ionized thallium is shown at the left in Fig. 18.11.
The gj factor for 2P| is f, which substituted in Eq. (18.52) gives the
two values gF = 1 for |P3 and gF = f for ?P3. Similarly the gj factor
of two for 2£j gives g, = 1 for ?£* and gF = $ for 20Sj. Multiplying
each by the possible values of Mf [see Eq. (18.57)], the magnetic
splitting shown at the extreme right in the figure is obtained. For any
given field strength, L is a constant.
By analogy with fine structure the selection rules for hfs are

AMf = 0 for p components,

AMf = ±1 for s components. (18.58)
 376 INTRODUCTION TO ATOMIC SPECTRA (Chap. XVIII

Similarly the relative intensities should be given by the fine-structure

equations (13.21), by replacing J by F and M by Mr.1
In a very weak field the hfs lines each break up into symmetrical
patterns similar to the anomalous Zeeman patterns of fine structure
in a weak field. Because the complete resolution of hfs patterns in
zero field is accomplished only for the very widest and simplest structures,
it is now quite clear why the so-called Zeeman effect of hfs has not been
observed.
If now the field is increased to a point where the ordinary anomalous
patterns of fine structure would be observed, each of the hfs patterns,
like those in the figure, will begin to overlap and something analogous
to the Paschen-Back effect sets in. This is the subject to be treated
in the next section.
18.10. Back-Goudsmit Effect in Hyperfine Structure.—If the mag¬
netic field strengths commonly used in studying the Zeeman effect are
employed in a study of the spectrum of an
H element having a nuclear magnetic and mechani¬
cal moment, a phenomenon analogous to the
Paschen-Back effect in fine structure is observed.
Because this effect was first observed and ac¬
counted for by Back and Goudsmit2 the phe¬
nomena will here be referred to as the Back-
Goudsmit effect.
A magnetic field that is called a weak field
for fine structure is in reality a strong field for
hfs. In such a weak field the coupling between
the nuclear moment I* and the electron re¬
sultant J* will be broken down, and each
will be quantized separately with the field
direction. Classically each will precess sepa-
Fig. 18.12.—Independent rately around H as shown in Fig. 18.12. Under
mo^rv' amf eUctron the8e conditions the magnetic energy will be
moment J*, in a toeak mag- made up of the sum of the following energies;
effect).6^ (Baok-Goudsmit (1) the energy due to the interaction between
/* and H; (2) the energy due to the interac¬
tion between J* and H; (3) the energy due to the interaction between
I* and J*. By Larmor’s theorem the two precessional angular veloci¬
ties are given by H times the ratio between the magnetic and mechanical
moments [see Eqs. (3.58), (10.23), and (13.23)]:

a‘ = H g' ^’ «' = Hgj ‘ 2mc" (18-59)

1 For hfs intensity formulas derived by the quantum mechanics, see E. L. Hill,
Proe. Nat. Acad. Set., 15, 779, 1929.
* Back, E., and S. A. Goudsmit. Zrnts. f. Phys47, 174, 1928.
Sac. 18.10] HYPERFINE STRUCTURE 377
Multiplying each of these angular velocities by the corresponding
angular-momentum components along H, the following two energy
contributions are obtained:

- -a ■ „ ■ co. (/*H) <i8 ^

' J"S M'S'

The sum of these two energies gives the main energy shift from the
fine-structure term from which it sprang, i.e., from the center of gravity
of the hfs terms arising from a given fine-structure term. This sum
may be written

AW ir = + gsMJH-- (18.61)

Dividing by he, the energy in wave numbers becomes

He
-ATn = (-gjMr + = {-giMi + gjMj)L cm’1. (18.62)

The third interaction mentioned above has still to be added to this

equation. By exactly the same treatment as that given for a single
electron in Sec. 10.7, and for several electrons in Sec. 13.7, this energy
is given by
r = A'I*J* cos CT*J*) cm-1. (18.63)
Referring to Fig. 18.12 it is seen that the angle between I* and J*
is continually changing so that it must be averaged. In the usual
manner the average cosine is given by

cos (/*/*) = cos (/*//) cos CJ*H). (18.64)

Substituting in Eq. (18.63),
T = A'l* cos (J*H) • J* cos (,J*H). (18.65)
In terms of the magnetic quantum numbers these projections are
T = A'MiMj. (18.66)
Adding this to Eq. (18.62), the total magnetic energy in a weak field
becomes
-A T = (—giMi + gjMj) L + A'MjMj. (18.67)
Now the gi factor for the nucleus is so small compared with gj for
the external electrons that the first term in the parentheses may be
dropped. Neglecting this term,
—AT = gjMjL + A’MiMj cm"1. (18.68)

Let us now apply this equation to the specific example of one of the
resonance lines of thallium, X * 5349, 6p,2Pi — 7s,2Sj (see Fig. 8.1).
378 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVIII

For the lower level 2P,, g3 - $,Mj = f, -f, —and M, = % and —

The energies in very weak and in weak fields are readily calculated by
tabulating them as in Table 18.4. Starting at the left the values of
each column are written down in the order given.

Table 18.4.—Magnetic-energy Factors for the Terms 2Pj and 2S±, in an Atom
Where the Nuclear Shin I ~ \

Very weak field Weak field

(hfs Zeeman effect) (Back-Goudsmit effect)

Term IV Mf 9F QfMf Mi Mj Qj QjMj AMjMj

2 2 9, 6 IA"
\ a
j 1 1 \ •2 a \A"
+A]" 0 1 0 _1 2 -\A"
ip* \ 2 j
-1 -1 1 _a \ 0
2 a 3 -IA"
-2 -2 3 fi ~4 A"
-i J
— 1 2. -J A "
2 2 J
1 7 _1 _2 \A"
a 2 J
0
s
0 —2 —2 _0 IA"
\Pi -M" a A
_ 6
-1

1 1 1 1 1 \A'
i 2
+ \A' 0 1 o 1 1 -1
ISj 2 2 -M#
-1 -1 1 2 1 -\A'
2 2
1 _ J2 _1 1 A'
1 1 4/l

JSi -\A' 0
0
0
0 0

After the table is completed, the energies are plotted as shown in

Fig. 18.13 starting with the field-free levels on the left and ending with
weak-field levels on the right. Calculations of the interaction energies
for field strengths between the two extreme cases of weak and very
weak fields have been carried out by Goudsmit and Bacher,1 and by
Inglis.2 Their calculations show that the following rules can be applied
for the transitions between very weak and weak fields: (1) the total
projection of the mechanical moment of the entire atom on H remains
the same in all field strengths, i.e.f
Mf = AT/ + Mj
(18.69)
and (2) levels with the same value of MF never cross.
Selection rules for hfs transitions between levels in a weak field
are similar to those for fine structure in strong fields and may be written
A Mi ~ 0,
( 0 for p components, (18.70]
A Mj
~~ ( ± 1 for 5 components.
1 Goudsmit, S. A., and R. F. Bacher, Phys. Rev., 34, 1499, 1929.
* Inglis, D. R., Zeits. f. Phys., 84, 466, 1933.
Sec. 18.10] HYPERFINE STRUCTURE 379

1 0 -i
Fig. 18.14.—Schematic diagram of the change in the T1 hfs pattern X5349 from no
field to very weak field to weak field Back-Goudsmit effect). (Observed pattern after
Wulff). In reproducing the microphotometer curve the field-free lines have been left
out as shown by the break at the center of the figure.

the weak field where the coupling between I and J is broken down, the
lines have crossed each other to form a pattern resembling closely the
Zeeman pattern of the same transition in an atom where the nuclear
 380 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVIII

moment is zero (see Fig. 10.8). A comparison of Fig. 18.14 with Fig.
13.11 for the Paschen-Back effect of the fine structure in ^'-coupling
is of interest. With I = % it is to be noted that each line of the Zeeman
pattern should theoretically be double, and that the separations should
be of the same order of magnitude as the field-free hfs. The micro¬
photometer curve at the bottom of the figure has been obtained in a
field of 43350 gauss by Wulff.1
A similar calculation for any spectral line where 7 has an integral
or half-integral value will lead to a pattern resembling closely the Zeeman
pattern for the same line if 7 were zero. Due
to the interaction energy term A'MMj in Eq.
(18.68) each of the Zeeman terms will be split
up into 27 + 1 equally spaced levels, and
each line in the observed Zeeman pattern will
be split into 27 + 1 equally spaced lines. This
may be seen to better advantage from the
following description of an atom where 7 is
large. In weak field, 7* and J* precess inde¬
pendently around H. For each orientation of
J*, as shown in Fig. 18.15, there will be
27 + 1 orientations of the nuclear moment 7*.
In Fig. 18.15 the 7 value, as in bismuth, is
taken as -f. Since the coupling of 7* with
H, as measured by the energy factor giMIy
Fio. 18.15.—Vector dia¬
gram showing the 21 + 1
is negligibly small, 7* is coupled to, and pre-
orientations of the nuclear cesses around, the magnetic component of J*
moment/* for a fixed orienta- along JJ Like the ordinary
tion Of the electron moment —& “• Like the ordinary Zeeman effect
J*. Drawn for the special this energy of interaction is proportional to Mj.
GoudsmiTeffect).= f (Back‘ Since the field for the precession of I* is fur-
nished by J*, and its magnitude by Mj, the
energy of the splitting will be proportional to MiMj, or equal to A'MlMJ,
as already shown.
As an example of this effect a diagram of a weak-field pattern observed
by Back and Goudsmit in bismuth is reproduced in Fig. 18.16. Each
of the four middle groups has been resolved by Back and Goudsmit and
found to have exactly 10 components. The two center groups of lines
marked a have been reproduced and published by Zeeman, Back, and
Goudsmit.2 Figure 18.17 has been enlarged and reproduced from the
halftone reproduction in their published paper. To obtain the lower half
of the figure, the photographic paper was moved parallel to the lines

1 Wulfp, J., Zeits. /. Phys., 69, 74, 1931; see also Green, J. B., and J. Wulff,
Phys. Rev., 88, 2176, 1931; 38, 2186, 1931.
* Zeeman, P., E. Back, and S. A. Goudsmit, Zeits. f. Phys., 66, 11, 1930.
Sec. 18.10] HYPERFINE STRUCTURE 381
during exposure thus enabling the 10 components to be easily distin- 4
guished. The effect is much the same as a microphotometer trace.
That there are 10 components in each group furnishes conclusive evidence
that, for the Bi nucleus, J * f

Fia. 18.16.—Calculated pattern for the bismuth line X4722 in a weak field (Back-Goudsmxi
effect).

Fig. 18.17.—Observed patterns, by Back and Goudsmit, of the two center components
marked (a) in Fig. 18.16. (A) Enlarged from the halftone reproduction in the published
paper of Zeeman, Back, and Goudsmit. (B) Photographic paper moved parallel to the
lines during enlargement.

Intensity rules for hfs in weak fields are the same as those for the
Zeeman effect in weak fields and are given by Eq. (13.21). From the
assumption that Mi does not change in any transition the intensities
will be independent of the value of Mt. This means that for each
component of an ordinary Zeeman pattern the intensity will be divided
equally between the 2/ + 1 components due to the nuclear moment.
382 INTRODUCTION TO ATOMIC SPECTRA [CkAP. XVIII

This means that Eq. (13.21) should be valid and that Af, as given there,
is the Mj as given here. This is in fair agreement with the observations in
Tl. In Bi the appearance of so-called forbidden transitions (AAf j — ±1)
shows that the field is not strong enough for these simple rules to hold.
Calculated intervals and intensities have been shown by Goudsmit and
Bacher,1 and by Back and Wulff2 to be in excellent agreement with
observed patterns in Tl and Bi in different intermediate field strengths.
18.11. Isotope Structure.—It is now known with certainty that the
hfs observed in the spectrum lines of some elements is due to small
differences in the energy levels of two or more isotopes and not to a

Fig. 18.18.—Isotope structure observed in the spectrum lines of tungsten. {After Grace,
More, and White.)

nuclear spin of one isotope. A good example of such structure is to be

found in the spectrum of tungsten, one section of which, taken with
Fabry and Perot 6talons, is reproduced in Fig. 18.18. Each line, it is
noted, shows three components, each of which corresponds, respectively,
to one of the isotopes of even mass 182, 184, and 186. Although these
patterns are distinctly different from the very commonly observed
hfs flag patterns as shown in Fig. 18.1, one cannot always distinguish
isotope structure from hfs by appearances. Some hfs patterns and some
isotope patterns look much alike.
In tungsten as in other elements where isotope structure predom¬
inates, many lines appear to be sharp, i.e., the lines arising from the
different isotopes fall on top of each other. This fact indicates that
the relative shifts of a set of terms belonging to one isotope are practically
the same as the relative shifts of the same levels in another isotope.
Where isotope structure is observed the relative shifts are materially
different. This latter means that the nucleus by itself is not entirely
responsible for the structure and that the interaction between the
1 Goudsmit, S. A., and R. F. Bacher, Zeits. f. Phys., 66, 13, 1930.
* Back, E., and J. Wulff, Zeits. /. Phys.t 66, 31, 1930.
Sec. 18.12] HYPERFINE STRUCTURE 383

nucleus and the extranuclear electrons must be taken into account. ,

Because there are no combinations between the levels of one isotope and
those of another one cannot say how much one system has been shifted
from the other but only how levels shift relative to others. As shown by
the tungsten lines, as well as the lines in other elements, the relative
shifts, where observed, are found to be approximately proportional to the
mass differences.
The suggestion that isotope shift may be due to a change in the
Rydberg constant (see Sec. 2.13) is occasionally but not always in
agreement with observation. In singly ionized lithium, for example,
the observed shift of Li6 ls2p and l$2s from Li7 ls2p and ls2s is several
times that expected. Hughes and Eckart1 have shown, however, that
with two electrons a change in the nuclear mass may alter the energy
by a greater amount than in the case when there is but one electron, as
in hydrogen. Their calculations for Li6 and Li7 are in excellent agree¬
ment with experimental observations.
Attempts to calculate isotope displacements in heavy elements
have been made by Breit.2 These calculations, based upon a change
in the size of the nucleus, are in fair agreement with the relative shifts
observed in Hg, Tl, and Pb.
The slight differences in the relative intensities of the tungsten
isotope lines in each pattern of Fig. 18.18 are due to the presence of a
fourth isotope of odd mass 183. Having a nuclear spin, this isotope
gives rise to hfs which divides into components differently in different
patterns. This subject of isotope structure and hfs together is taken
up in the next section. As a matter of fact isotope structure is seldom
observed in an element alone.
18.12. Isotope Structure and Hyperfine Structure Combined.—Per¬
haps the most complicated line patterns in all the periodic table are
those observed in the spectrum of mercury. Since mercury is a heavy
element, and since the mercury arc is a convenient light source for the
adjustment and demonstration of interferometers, the mercury lines
have been investigated by many experimenters. In such elements with
high atomic weight, the Doppler broadening does not interfere in ordinary
light sources to the extent that some lines were known over two decades
ago to contain as many as 10 to 15 components.
In the very excellent hands of Schuler, Keyston and Jones,5 these
patterns have now been analyzed. Their analysis shows that of the six

1 Hughes, D. S., and C. Eckart, Phya. Rev., 36, 694, 1930.

8 Breit, G., Phya. Rev., 42, 348, 1933; see also Rosenthal, J. E., and G. Breit,
Phys. Rev., 41, 459, 1932; and Racah, G., Nature, 129, 723, 1932.
8SchOler, H., Naturwiaaenachqften, 19, 950, 1931; SchUler, H., and J. E. Key¬
ston, Zeiia.f. Phya., 72, 423, 1931; SchCler, H., and E. G. Jones, Zeita.f. Phya., 74,
631, 1932; 77, 801, 1932.
384 INTRODUCTION TO ATOMIC SPECTRA [Chap. XVIII

mercury isotopes,

Mass 196 198 199 200 201 202 204

Abundance 0.10 9.89 16.45 23.77 13.67 29.27 6.85 per cent,
the five with even mass have either a zero magnetic or a zero mechanical
moment,1 and the two with odd mass have mechanical moments as
follows: for Hg 199, I = \ and for Hg 201, I = j. The magnetic
moments of these two isotopes are opposite in sign and quite different
in magnitude. For Hg 199, gt = 1.1 and for Hg 201, gt ^ —0.41 nuclear
magnetons.

Flu. 18.19.—Hyperfine structure of the mercury isotopes as observed in the transition

S«6p, *Pa — 6s6d, lD\, X3662. (After Schiller, Westmeyer, and Jones.)

Hfs terms for two energy levels in mercury are shown in Fig. 18.19.
The heavy lines near the center of both the upper and lower levels
represent the even isotopes 196, 198, 200, 202, and 204. The very light
lines represent the levels of the odd isotope 201, and the remaining levels
belong to 199. The observed pattern is shown at the bottom, and the
corresponding transitions are given directly above. The relative
abundance of the isotopes as given by Aston’s mass-spectrograph meas¬
urements is in agreement with the observed hfs in that the center line
contains about 70 per cent of the total intensity. If there were an
1 Since we know relatively little about nuclear structure at the present time, it is
difficult to say whether one, or both, of the nuclear moments, magnetic or mechanical,
is zero. If a nucleus has a large magnetic moment, but no mechanical moment,
there can be no space quantization with J * and hence no hfs. If a mechanical moment
exists, but no magnetic moment, there will be quantization but no splitting. Such
cases of one or the other moment equal to zero are known to exist in extranuclear
electron structure.
Sue. 18.12] HYPERFINE STRUCTURE 385
appreciably different isotope shift in the initial dr final states, as there ,
is in the resonance line — zPi at X2536, this line would not be single
as observed but would reveal five components. The relative positions
of the levels belonging to different isotopes are arbitrary, for, if one system
of levels is moved up or down, the radiated lines remain unchanged.
Returning to Eqs. (18.34) and (18.35), the observed hfs in mercury
may be used to calculate the coupling coefficient for the deeply penetrat¬
ing and tightly bound 6s electron. For the ZP2 state the calculated
values are as follows:
Hg 199 Hg 201
A' = +0.303 Af = —0.114
o' « +1.212 a! = -0.456
Since a! is negative for Hg 201, the gt factor [by Eq. (18.24)] is negative.
For Hg 199, the factor is positive and by the ratio of the o' coefficients
is about 2.66 times as large. If the same calculations are made for the
upper term zDi, a smaller ratio is obtained. This may be attributed
to perturbations of certain of the hfs terms by those of a 1D2 term, only
three wave numbers away. When the perturbation corrections shown
in Fig. 19.15 are made, the ratio of the a’s, and hence the ratio of the
gi factors, comes out to be 2.79, in very good agreement with similar
calculations for six or more other terms.
In addition to mercury, line patterns of a number of other elements
have been sufficiently analyzed to show the presence of both isotope
structure and hfs. In Cd, for example, there are six isotopes of mass
110, 111, 112, 113, 114, and 116. Like mercury the even isotopes are
observed to have a nuclear moment of zero, and both of the odd isotopes
to have the same mechanical moment Z = i and the same magnetic
moment gt = —1.0.
Problems

1. Show that the interaction energy of two s electrons with a nuclear moment I
is given by Eq. (18.38).
.
2 Construct a diagram similar to Fig. 18.9 for the term d8s,6F. Assume a '—
2 cm-1 and 7 = f-
3. Construct a diagram similar to Fig. 18.13 for the Bi levels given in Fig. 18.5.
Assume gj factors of 2 and f for the upper and lower levels, respectively. Calcula¬
tions are most easily made by tabulating as in Table 18.4. Note that, for the lower
level, A1 is negative.
.
4 From the levels derived in Prob. 3, construct a diagram similrr to Fig. 18.14.
 CHAPTER XIX

SERIES PERTURBATIONS AND AUTOIONIZATION

It was pointed out in Chap. I that certain well-established spectral

series show abnormalities of one kind or another. Of particular impor¬
tance are the anomalies of the type found in one of the calcium series
shown in Fig. 1.13. In this one instance the diffuse series behaves in
the usual fashion by contracting after the first member to very narrow
patterns. This contraction is soon disturbed, however, and the patterns
expand anomalously reaching an apparent maximum at the sevehth
member; then decreasing again to narrow patterns, they approach the
series limit.

Fig. 19.1.—Normal cadmium series. Fig. 19.2.—Anomalous calcium senes.

Term values
Fig. 19.3.—Anomalous copper series. Fig. 19.4.—Anomalous aluminum series.

In terms of the energy levels, themselves, this anomaly is to be

attributed to the series of initial 3D 1,2.3 states and not in any way to the
common final 3P0.i,2 states. Of the surprisingly la”ge number of well-
known anomalous series found in other elements, those of aluminum
and copper are particularly notable.
19.1. Observed Abnormal Series.—To start with, let us consider
several observed series where the causes of the irregularities are now
well known. Although it is the anomalous progression of spectral
lines that is observed, we are here concerned with the derived energy
levels themselves. In Figs. 19.1, 19.2, 19.3, and 19.4, the fine structures
Sec. 19.2] SERIES PERTURBATIONS AND AUTOIONIZATION 387

of four series of terms are shown for cadmium, calcium, copper, and
aluminum, respectively. In each figure the fine structure is plotted
vertically and the term values horizontally. The dotted center lines
represent the center of gravity of each series member, each level being
assigned the quantum weight 2J + 1. The cadmium series in the first
figure represents a series of triplets approaching in a regular fashion
the series limit.
The 3F2,3,4 series in singly ionized aluminum, like the 3Di,2,s series
in calcium, expands abnormally and contracts again as it approaches
the single limit. The copper series discovered by Shenstone1 is of
particular interest: the first term is right side up (2P3 above 2Pj), the second
term is very narrow (both terms falling practically together), the third
term is inverted, the fourth term is right side up and very wide, and the
fifth term is inverted. The beautiful way in which each of these anoma¬
lous series has been explained by Shenstone and Russell2 is the subject
of the following sections of this chapter.
19.2. Energy Level Perturbations.—In the early chapters of this book
we have seen that the term values of hydrogen-like atoms are given by
the simple Bohr formula

where Z is the nuclear charge, R the Rydberg constant = 109737 cm-1,

and n the total quantum number. If Z is replaced by the effective
nuclear charge Z„ (see Sec. 7.4) and n by the effective quantum number
n*, one obtains Rydberg's general formula for series of many electron
atoms,
RZl = RZl
(19.2)
n*2 (n — n)2
Here the integral valued quantum number n is retained by writing
n* = n — /x, where u is the so-called quantum defect. For all neutral
atoms Z0 — 1, for singly ionized atoms Z0 — 2, etc. Using this formula,
three aluminum series are shown in Fig. 19.5 with values of u = n — n*
plotted vertically and the term values plotted horizontally. Series
following Rydberg's formula [Eq. (19.2)] are represented on such a
diagram by a straight horizontal line. A refinement of Rydberg's
formula by Ritz in which n* is replaced by n — n — aTn is necessary
for a series like lF2:
RZl = RZl
(19.3)
n*2 (n — n — aTn)2
On the diagram such a series is represented by a straight line with inter¬
cept fj, and a slope given by a.
1 Shenstone, ,
A. G., Phys. Rev., 34 1623, 1929.
2 Shenstone, ,
A. G., and H. N. Ritssell, Phys. Rev., 39 415, 1932.
388 INTRODUCTION TO ATOMIC SPECTRA {Chap. XIX

Now in the anomalous *F series of aluminum the fourth term is a

foreign term and rightfully belongs to another *F series.1 When this
term is omitted, the n values of the following terms will all be lowered
by one. A replot of the series with corrected n values is shown in Fig.
19.6, along with the anomalous 3D series of calcium in Fig. 19.7. The

Fio. 19.6.—Aluminum series showing per- Fig. 19.7.—Calcium series showing per¬
turbations. turbations.
resemblance between these curves and anomalous dispersion curves is
quite striking.
A generalized form of the Rydberg-Ritz formula, derived from
perturbation theory and the quantum mechanics, has been given by
Langer.2
RZl
Tn (19.4)

1 In a study of this series Schrodinger attributed the anomalous behavior to a sort

of polarisation of the atom and assigned the third and fourth members of the series
to the same total quantum number; Ann. d. Phys., 77, 43, 1925.
* Langer, R. M., Phya. Rev., 35, 649, 1930.
Sue. 19.2] SERIES PERTURBATIONS AND AUTOIONIZATION 389

Here T% represents as usual the successive terms of the series and a

foreign perturbing term not belonging to the series. Although Langer’s
formula does not adequately account for all series details, it does account
for the existing anomalies in many series and its success is much beyond
the expectations of a first approximation.
Langer's formula, modified by Shenstone and Russell to fit the
anomalous series taken up in this and the last section, takes the form

m _ _Ml_ (19.5)

(^.)i
where the effective quantum number is

n* = n — n acTn + (19-6)

This equation postulates that under certain conditions (to be specified

later) a term Ta foreign to the series in question will perturb the series,
causing it to deviate from the ordinary progression expected. Because
of the form of the perturbation term P/{Tn — Ta)f this formula is analo¬
gous to the formula used in anomalous dispersion theory. This analogy,
as seen in Fig. 19.6, is closer than might at first be expected.
As an example of the use of Eqs. (19.5) and (19.6), the observed
term values and the computed constants of the *Di series of calcium
are given in the following table.
Table 19.1.—Series Calculations for the Anomalous 4*nd,*Z)i, Series of
Calcium
(After Shenstone and Russell)

Term value Term value rcobs.) - r

Configuration n n* n — n*
observed calculation (calc.)

28969.1 4«3d 3 1 1 9463 1.0537 26465 2504

11556.4 4s4d 4 3 0815 0 9185 11556 0
6561 4 4s5d 5 4.0896 0 9104 6562 -1
4255 5 4»6rf 6 5.0780 0.9220 4252 +3
3002 4 4s7 d 7 6.0456 0 9544 3002 0
2268 3 4s8d 8 6.9552 1.0448 2287 -19

1848 9 M5s Foreign term

1551.3 4*9d 9 8.4105 0.5895 1551 1 0

1273.1 4sl0d 10 9.2840 0.7160 1276 -3
1045.6 Mid 11 10.2445 0.7555 1048 -2
869.8 As\2d 12 11.2323 0.7677 873 -3
734.0 4s 13d 13 12.2273 0 7727 737 -3
628.0 4sl4d 14 13.2190 0.7810 -2

M - 0.8705, a - 0.00000339, 0 - -85, T. - 1848.9.

 390 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIX

With the exception of the first series member, the agreement between
the observed and calculated term values is very good. The origin of
the foreign term is discussed in Sec. 19.5.
19.3. The Nature of and Conditions for Term Perturbations.—A
first-order perturbation treatment and calculation of atomic energy
levels have been given by Slater1 and by Condon.2 Including electron
spin in the calculations, Condon shows that under certain conditions
terms belonging to different electron configurations will perturb each
other. If two levels have the same characteristic quantum numbers,
and lie close together, second-order terms of the perturbation correction
become large and cannot be neglected in the calculation of the energy cf
either level. A large second-order correction implies that the eigen¬
functions for each of the respective levels contain components of the
eigenfunction of the other level. Thus both levels belong in part to
both electron configurations. The assignment of a given level to a
definite electron configuration is therefore only approximate and becomes
more exact when the levels are far apart.
The nature of the perturbation between the two levels in question
reveals itself essentially as a repulsion. This apparent repulsion may
be thought of as being due to a sort of resonance phenomenon, there
being a certain probability that the atom will jump back and forth
between the two characteristic states without the radiation of energy.
The frequency of this interchange, and hence the repulsion, increases
as the interval between the levels decreases.
The quantum conditions for term perturbations may be stated as
follows: Two levels belonging to the same or to different electron configurations
will perturb each other when (1) both electron configurations are of the same
parity, i.e., both even or both odd,3 and (2) both levels have the same J value.
In addition to these necessary conditions, observation shows that the
greatest effect is to be expected when the two levels have the same
L and S values in L&-coupling, or the same ji and values in ^’-coup¬
ling. For intermediate coupling schemes the more nearly alike the
coupling schemes the greater will be the effect.
Consider as a simple example two neighboring sets of 3Di,2,3 terms
arising from different electron configurations of different parity, i.e.,
one configuration odd and the other even. In the ideal case of LS-coup*
ling these levels will be represented as shown at the left in Fig. 19.8.
If now the configurations are of the same parity, i.e., both even or both
odd, a sort of resonance will be set up tending to repel corresponding
levels. These are shown at the right in the figure. The nearer two
1 Slater, J. C., Phys. Rev., 34, 1293, 1929.
* Condon, E. U., Phys. Rev., 36, 1121, 1930.
3 An electron configuration or term is said to be odd when the sum of the l values
of the electrons is odd, and even when the sum is even.
Sec. 19.5] SERIES PERTURBATIONS AND AUTOIONIZATION 391

corresponding levels are to each other, the greater is the repelling effect'
(see Figs. 19.6 and 19.7). In Fig.
19.8 the 3Z)3 terms, being the nearest
to each other, undergo the greatest
displacement, with the result that the
narrow triplet is widened and the wide
triplet is narrowed. This same result
is to be expected when the unperturbed
wide triplet is above the narrow triplet.
19.4. The Anomalous Diffuse
Series of Calcium.—We now return to
the anomalous 3D (4snd) series of
neutral calcium discussed in the
preceding section. Referring to the
calcium energy level diagram in Fig.
11.9, it is seen that the first member
of the diffuse series 4snd is also the
first member of another 3Z> series
3dns. Now the seventh member of
the anomalous series is without doubt Fig. 19.8.—Schematic representation of
term perturbations.
an extra member and should be
called the second member of the 3dns series. If this assignment is
correct the first two terms of the latter series should have as a limit the
first excited 2D state (3d) of the ionized atom. Term values taken with
respect to 2D as a limit give Rydberg denominators 1.60 and 2.64, in good
agreement with the assignment.
A more detailed study of the fine structure of the two 3D series is to
be seen in Fig. 19.9. At the extreme left and right in this figure the two
series are shown as they would occur without perturbation. In the
center the observed series is shown. It should be pointed out that the
unperturbed 3dns triplets should have a total separation of approxi¬
mately 60 wave numbers, the same as that of the series limit 2D (see
Sec. 11.5), and that the higher members of the 4snd series should be very
narrow. The effect of the perturbation is to widen the triplets on the
left and to narrow those on the right, as observed. T^ie third member
of the right-hand series, shown by dotted lines, because it lies above the
first series limit, should not be observed. This will be discussed in
Sec. 19.9.
19.6. The Anomalous Principal Series in Copper.—The fine structure
of the anomalous 2P series in copper is shown in Fig. 19.3. This very
irregular series, probably the most distorted series known, was discovered
by Shenstone and now has a very beautiful and simple explanation.
Although the first term of the series, 3dl04p,2Pj.|, in combination with
3dl04s,2£$, gives rise to the strongest lines in the copper arc spectrum,
392 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIX

Unperturbed ftrturtxd ftrturted Unperturbed Unperturbed fcrturbed Perturbed Unperturbed

Fig. 19.9.—Perturbed diffuse series in Fig. 19.10.—Perturbed principal series in

calcium. copper.

0 1QOOO 201000 3Q00Q

Term Values
Fig. 19.11*—Showing perturbations in the principal series of copper.
Sec. 19.6] SERIES PERTURBATIONS AND AUTOIONIZATION

it was not until the spectrum had been so thoroughly analyzed and every-’
thing else was accounted for, that the higher members of the series were
definitely identified.
It is now known that the irregularities in this series are due to the
perturbing influences of a foreign, widely spaced, inverted 2P term.
As shown in Fig. 19.10 these two fine-structure terms, 3952 cm-1 and
5973 cm”1, respectively, are due to the odd electron configuration 3d®4s4p,
and the perturbed series of terms are due to the odd configurations
3dlonp. Plotting the quantum defect n — n* against the term values,
Fig. 19.11 results. It is observed that, as the series approaches the
perturbing terms from either side, the mutual repulsions between cor¬
responding terms (i.e., terms with the same J values) increase slowly at
first, then more rapidly.
19.6. The Inverted Alkali Doublets.—It may be shown from the
quantum mechanics or from the classical model of the atom that the
doublet levels of the alkali metals should be normal, i.e., of the two levels
j = l + i and j = l — \ the level j = l — $ should lie deeper. While
this is observed to be generally true, it is well known that the 2D and 2F
terms of some of the alkalies are inverted. In caesium, for example, the
first 2P and 2D terms are normal and
the second 2F term is inverted.1 In
rubidium the 2P terms are normal and
the 2D and 2F terms are inverted.2
In potassium the 2P terms are normal
and the 2D terms are inverted.3 In
view of the anomalous turning over
of the normal copper doublets, dis¬
cussed in the preceding section, it
seems highly probable that the inver¬
sion of the narrow alkali doublets
might be due to the perturbing influ¬
ence of widely spaced inverted doub¬
let terms lying high above the series
limit.4 This inverting effect is shown
schematically in Fig. 19.12.
Inverted doublet terms arise in Fia. 19.12.—Possible explanation of the
inversion of alkali doublets.
general from electron shells lacking
but one electron to complete them. Consequently the excitation
of an electron from one of the completed electron shells to an outer
orbit may give rise to several of the desired doublets. Consider, for

i K. W., Ann. d. Phya., 50, 713, 1916; 65, 378, 1921.

Meissner,
* Ramb, R., Ann. d. Phya., 10, 311, 1931.
8 Meissner, K. W., and O. Masaki, Ann. d. Phya., 10, 325, 1931.
4 White, H. E., Phya. Rev., 40, 316, 1932.
394 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIX

example, the normal rubidium atom with the electron configuration

Rb 37, ls1 22s2 *2p63s23p63d104$24p65s. (19.7)

The excitation of the 5s valence electron to any of the unoccupied

d orbits 4d, 5d, 6d, etc., should give rise to a series of very narrow but
normal 2D levels. The excitation of a 4p electron from the completed
4p6 subshell to a 5p orbit, on the other hand, will result in the even
configuration (ileaving off the completed subshells) 4p55s5p, giving rise
among other terms to one inverted 2D term and one normal 2D term.
Similarly a double excitation of two electrons to the even configuration
4p64d5p gives rise to three inverted 2D terms and three normal 2D terms.
Due to the spin-orbit interaction of the 4p6 subshell these levels, whatever
the coupling scheme, will be very wide. Furthermore the inverted
doublets should lie deepest. Although these terms have not been
observed in the alkali spectra, they certainly do exist and their position
may be predicted fairly accurately from x-ray spectra to be high above
the regular series limit of the chief optical series. Even at so great
a distance, the inverted terms tend to invert the narrow doublet terms
far below and the normal terms tend to restore them. The net second-
order effect of these two opposing actions, although small, is great
enough to cause inversion.1
19.7. Autoionization.—The atomic process of autoionization put for¬
ward by Wentzel and Shenstone2 is of considerable importance in atomic
spectra, for it clears up a great many of the difficulties previously unex¬
plained. Above the limit of any series (see the calcium series in Fig. 19.9)
lies a continuum of possible energy states characterized by the same
quantum numbers as the series of discontinuous levels below the limit,
i.c., the same L, S, and J values and the same odd or even parity of
the electron configuration. An atom excited to a hypothetical energy
state like the even 3Z> (3ri6s) state of calcium, above the first series
limit and into a region accompanied by an even *I) continuum, is in
a condition for resonance of the type discussed in Sec. 19.3. Before
the atom has time to reach any stationary state and return to a lower
state giving rise to radiation hv, there will be a spontaneous transition
into the continuum equivalent to the ejection of an electron with kinetic
energy \mv2 and the return of the atom to the series limit 2S<r This
process of autoionization clearly shows why the third and succeeding
members of the calcium series 3dns are not observed and postulates

1 These actions at so great a distance have been calculated by M. Phillips, Phys.

Rev., 44, 644, 1933, and shown to be great enough to produce the observed inversion.
2 Autoionization has long been known for x-rays as the Auger effect. In giving a
quantum-mechanical treatment of the effect, G. Wentzel (Zeits. /. Phys., 43, 524,
1927), first suggested its possible application to optical spectra. Direct application to
optical spectra was made by Shenstone, Phys. Rev., 38, 873, 1931.
Sec. 19.8] SERIES PERTURBATIONS AND AUTOIONIZATION 395

that these hypothetical levels are not discrete in the usual sense but are
spread out over the continuum.
In some instances the quantum condition for resonance between a
given discrete level and continuum are only partly satisfied (z.e., there is
a weak resonance coupling between the discrete state and the continuum,
see Sec. 19.3) so that the process of either spectral emission or autoionizar
tion may take place. This is the subject of the following sections.
19.8. Autoionization in Copper.—It has long been known that the
copper arc spectrum, unlike most spectra, consists of many spectrum
Even Odd Even

lines half of which are sharp and the other half unusually broad and
diffuse. A thorough investigation of these lines by Allen1 has shown
that the diffuseness is not due to a pressure or temperature effect but
to a natural breadth of the energy levels. Such broad levels immediately
suggest that we are dealing with short mean lives and with the process
of autoionization.
A thorough analysis of the copper arc spectrum by Shenstone2 reveals,
in addition to the well-known series of normal doublets, a system of
inverted doublet and quartet levels. From the energy level diagram
of copper in Fig. 19.13 it is observed that several of the inverted doublet
1 Allen, C. W., Phys. Rev., 39, 42, 55, 1932.
* Shenstone, A. G. Phy*. Urn.. 28. 449 1926: 34, 1623, 1929.
396 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIX
and quartet levels lie above the series limit of the normal doublets
and in the continua of even 2S and 2Z> and odd *P and *F terms. Of
these high levels the inverted 4Z>i,t,f,i terms (3d94«5s) are of particular
interest. In combination with a number of lower levels it is observed
that all lines with AD\ or AD\ as initial levels are narrow sharp lines
whereas those with or 4Z>| as initial levels are very broad and diffuse.
Photographs of the combinations 4Fhit^ — 4Z>|,|,|,* (3d94$4p — 3d94s5s)
are shown in Fig. 19.14 and the wave-lengths are given in Table 19.2.

The intervals between the 4D terms, as well as the intercombinations

with lower doublet levels, show that the coupling is not Russell-Saunders
but something between LS- and jj-coupling. Now the two broad
levels 4D| and 4D| are not strictly quartet levels with S = $ but are
nearly enough like 2Z>§ and 2D§ terms to permit, at times, autoionization.
The 4Z)| and 4Z>j levels, on the other hand, are strictly quartet levels,
for with 3d94s5« the spins of each of the s electrons and the S of 3d9
must take the so-called parallel positions to produce, with the L value 2,
the J values $ and j, respectively. The observed and calculated
g factors of 0 and -J- from the Zeeman effect confirm this. In this latter
case autoionization is not to be expected, and the sharp lines arising
from these two levels indicate that it does not occur.
 Sec. 19.9] SERIES PERTURBATIONS AND AUTOIONIZATION 397

Table 19.2.—Quartet Multiplet in Copper

‘Dj 54S.0 «D, 636.2 *D, 887.8 •z>»

100
‘Ft 4651.13
21494.2
2U-S
22 67 U
‘Ft 4704.60 4586.97
21249.9 21794.8
Ifl9.lt
2 21 U SOU
‘Ft 4797.04 4674.76 4539.70
20840.4 21886.5 22021.7
739.7
20 13 U 12
‘Ft 4842.20 4697.49 4509.39
20646.0 21282.0 22169.8

For want of a picture of the spreading out of an energy level, the

following process is suggested. By the absorption of energy the atom
is raised toward an excited state, e.g., toward 4Dj of the above given
example. Because of the resonance phenomenon discussed in Sec. 19.3
there will be a certain probability that the atom will go over into the
continuum and self-ionize before it has time to reach any stationary
state. Thus in absorption the energy level in question will not be
discrete but will be broadened by an amount depending upon the proba¬
bility of self-ionization. The emission of radiation being the reverse
of absorption, the observed spectral lines are broad.
The other diffuse lines in the copper spectrum all start from initial
states that lie above the first series limit and have, without doubt, the
same explanation as that given above for the 4DJtj states of 3d94s5«.
19.9. Autoionization in Calcium, Strontium, and Barium.—In the
early development of complex-spectrum analysis, Russell and Saunders1
discovered the now famous series of p' terms. These terms which they
designated as npfi,2fa, where n = 2, 3, 4, and 5, are now designated as
3P0,i,2 and are attributed by Russell2 to the electron configurations

3d3d, 3d4d, 3d5d, and 3d6d. All but the first member of this series
(see Fig. 11.9) lie above the first series limit 4 2S of the four chief series of
singlets and triplets. The continuum above these four chief series
corresponds to even S and D terms and to odd P and F terms. Russell’s
ZP terms (3dmd), being characteristically even, enable the electron in any
one of these states to have a mean life sufficiently great to combine
normally with odd lower terms and give rise to the respectably sharp
1 Russell, H. N., and F. A. Saunders, A atrophys. Jour., 61, 38, 1926.
* Russell, H. N., A atrophy a. Jour., 66, 13, 1927.
398 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIX

spectrum lines observed. The even *S term Sd4d combining with *P 0,1,2,
4s4p in a double electron jump gives rise to three hazy lines at X 2757.40,
X 2749.34, and X 2745.49. The broadness of these lines is due to the
broad 3&i level and indicates that (1) the quantum conditions for auto¬
ionization are not far from satisfied (with respect to the zSi continuum),
(2) the mean life of the atom is long enough to permit some atoms to
return to lower levels giving rise to radiation, and (3) that some of
the atoms starting toward this state go over into the continuum and
self-ionize.
In strontium a ZF° term 4d6p [the corresponding term in calcium
(see Fig. 11.9) is ZF°, 3d5p] lies just above the 52S limit and, as in calcium,
is in a continuum of *F° terms 5smf. These negative 3F°2,3.4 terms
4d6p combine with lower 3D 1,2.3 terms 5s4d, giving rise to five very diffuse
lines (see Table 19.3).
Table 19.3

Strontium, hazy lines (short mean life) Barium, sharp lines (long mean life1)

Interval X Designation Interval X Designation

4n 3189 23 •d, - < 10 3586 50 zo3 -

3n 3189 93 3D2 - *F°{ 10 3561 94 *z>2 - v;
2n 3198 99 - *f\ 10 3566 66 2Z>i - *F\
2n 3200 22 sz>, - >Fl 15 3610 96 3z>,-
In 3205 13 ’Z>2 - »f°2 3 3589 95 sd2 -
2 3639 72 3D, - 3F\

,
» White, H. E., Fhya. Rev., 38 2016, 1931.

In barium these same 3F°2(3>4 terms 5d7p lie just below the series
limit 62*S so that autoionization is not possible. Combining with lower
3Di,2,3 terms 6s5d, six relatively sharp lines are expected and observed
(see Table 19.3). Where the mean life in a given state is long the lines
will have a very narrow natural breadth, and where the mean life is
relatively small the lines will have wide natural breadth. This question
of the breadths of spectral lines will be taken up in Chap. XXL
19.10. Hyperfine-structure Term Perturbations.—In the early
development of hfs and nuclear spin it was observed that the Land6
interval rule was always valid. This observation is in complete agree¬
ment with the present-day atomic models, and, were it not for one
fairly well-established case in mercury, found by Schuller and Jones,1
the rule might be laid down as infallible. Since the one known departure
from the interval rule is attributed by them to the same type of pertur¬
bation phenomena as those discussed in the first sections of this chapter,
a discussion of this single known example is not out of place here.
1 SchOlbr, H. and E. G. Jones, Zeits f Phys., 74, 631, 1932; 77, 801, 1932.
Sac. 19.10] SERIES PERTURBATIONS AND AUTOIONIZATION

The first member of the diffuse series in mercury consists of four

terms *Di, *Z)2, ZDS, and 1D2, arising from the electron configuration
6$6d (see Fig. 11.6). Contrary to what one would expect from L*S- or
jj-coupling, the 1Z)2 term lies about three wave numbers below *Z>i.
Since the triplet intervals 60 and 35 cm-1 have the ratio 5:3 in place of
2:3 for L/S-coupling, only the 8Di and 3D3 levels can be assigned definite
S and L values. Due to the close proximity of the two levels *Di and

I- W 1-0 l-Yz
Hg 199 Hg 198.200,202,204 Hg 201

Fig. 19.15.---Hyperfine-structure term perturbations in the arc spectrum of mercury.

VAfter Schiller and Jones )

1D2; their respective hfs terms perceptibly perturb each other. These
perturbations are shown schematically in Fig. 19.15.
Due to the relatively large number of mercury isotopes (198, 199,
200, 201, 202, and 204) the hfs patterns of a majority of the spectrum
lines are veiy complex. As stated in Sec. 18.12 the very interesting and
praiseworthy analysis of many of these patterns by Schuler and Jones has
led to the assignment of different nuclear spins to the different isotopes:

Isotope 198 199 200 201 202 204

Mercury^
Nuclear spin 0 i 0 f 0 0
400 INTRODUCTION TO ATOMIC SPECTRA [Chap. XIX

Starting with the even isotopes with zero spin in the center column
of Fig. 19.15, the two terms *Di and lDt will, for the 199 isotope, each
be split into two hyperfine levels. The same two terms for the 201 iso¬
tope will be split into three and four hyperfine levels, respectively. The
dotted lines represent the centers of gravity of each set of levels (each
level is assigned the quantum weight 2F + 1), and the arrows indicate
the levels that repel each other. It is assumed that there is no isotope
displacement.
By analogy with the perturbation rules given in Sec. 19.3 the quantum
conditions for hyperfine term perturbations may be postulated as follows:
Two levels belonging to the same or to different electron configurations
will perturb each other (1) when both electron configurations are of the
same sign, and (2) when both levels have the same F value. The I value
being always the same for every energy level of each atom, the greatest
effect is to be expected when the two levels have the same J value.
(In the case of mercury the J's are different.) Schuler and Jones point
out that the repulsion is not only a function of the distance between the
levels but also a function of the F and I values. Though the F = $
levels are farther apart than the F = £ levels, the shift is greater. Fur¬
thermore the shift of the F = $ levels for I = \ is twice as great as for
7 = |. A possible explanation lies in the relative magnitudes of the
nuclear factors.
It is to be noted that the levels for 7 = f are inverted with respect
to the levels for I = J. Whatever the coupling scheme may be for the
6s6d configuration, the *7>i term can only be L/S-coupling with S = 1,
L * 2, and J = 1. With a positive g factor and a strong coupling
between the 6s electron and the nucleus, the hfs terms of 3Di will be
large and inverted as observed for I = With a negative nuclear
g factor they will be normal as observed for I = f. The magnitude of
the hyperfine separations shows that the positive g factor is 2.7 times the
negative g factor.
 CHAPTER XX

THE STARK EFFECT

20.1. Discovery of the Stark Effect.—Although the splitting up of

spectral lines in a magnetic field was "discovered by Zeeman as early
as 1897, some 16 years elapsed before anyone succeeded in showing
that a similar effect is produced when a source of light is placed in an
electric field. In 1913 Stark1 * * demonstrated that every line of the Balmer
series of hydrogen, when excited in a strong electric field of at least
100000 volts per centimeter, is split into a number of components.
When viewed perpendicular to the field, some of the components of
each line pattern are observed to be plane-polarized with the electric
vector parallel to the field (p components), and the others polarized with
the electric vector normal to the field (s components). When viewed
parallel to the field only the s components appear, now unpolarized.
There is at least one good reason why early attempts to observe
the splitting of spectrum lines in an electric field were not successful,
viz., the production of very high potential gradients along a discharge
tube is a very difficult task. Soon after Stark's discovery, Lo Surdo*
observed the splitting of lines excited in the cathode dark space of a
discharge tube. These two sets of observations mark the beginning of a
very interesting, and at the same time complex, field of investigation.*
20.2. The Stark Effect of Hydrogen.—If one were to ask, What was
the most outstanding success of the early Bohr quantum theory?, the
reply would be, The first satisfactory treatment of the hydrogen atom
in an electric field, by Epstein4 * and by Schwarzschild.6 It is interesting
to point out that these early results, coming as they did at a time before
the spinning electron made its d6but, have been little altered by the
more satisfactory quantum mechanics as given by the Schrodinger
wave equation6 and by the Dirac theory of the electron.7

1 Stark, J., Berl Akad. Wiss., 40, 932, 1913; Ann. d. Phys., 43, 965,1919.
1 Lo Stjrdo, A., Accad. Lincei Alii, 22, 665, 1913; 23, 83, 256, 326, 717, 1914.
* A rather complete bibliography and account of the Stark effect is given by R.
Minkowski in H. Geiger and K. Scheel, Handbuch der Physik, Vol. 21, p. 389, 1929.
4 Epstein, P. S., Ann. d. Phys., 50, 489, 1916; also Phys. Zeits., 17,148, 1916.
1 Schwarzschild, K., Sitz-ber. Berl. Akad. Wiss., 1916, p. 548.
4 SchrOdinger, E., Ann. d. Phys., 80, 437, 1926; Epstein, P. S., Phys. Rev., 28,
695, 1926.
7 Schlapp, R., Proc. Roy. Soc., A, 119, 313, 1928.
401
402 INTRODUCTION TO ATOMIC SPECTRA [Chap. XX

We shall not attempt to derive any of the equations for the Stark
effect but shall begin by first writing down the general energy relations
for a hydrogen atom in an electric field and then proceed to interpret
them in terms of atomic models and the observed spectrum lines. The
interaction energy of a hydrogen-like atom in an electric field is given by
AT = AF + BF2 + CF* + • • • (20.1)
Here AT represents the change in the term value of the atom in
wave numbers, i.e.s the shift in the energy levels from the field-free states
to the states in the electric field, and F is the strength of the field in
electrostatic units. The coefficients A B, and C, in this equation, have
been calculated from classical and quantum-mechanical considerations by
Epstein, Wentzel, Waller, Van Vleck, S.Doi, Schrodinger, and others,
and are given by

A = JLn(n* -ni)’

B = - 3(n2 - fii) - 9m? + 19), (20.2)

C = 2u^ncn^23n2 ~ (n> ~ n+ 39f-

Here n is the usual total quantum number, and m, n2) and mt are electric
quantum numbers, subject to the condition

mi = n — m — 7ii — 1. (20.3)

The allowed! values are

n = 1, 2, 3, • • • «, n\ = 0, 1, 2, 3, • • • n — 1,
(20.4)
mi — 0, ±1, ±2, • ■ • ± (n — 1), n2 = 0, 1, 2, 3, • • • n — 1.

If the field is expressed in volts per centimeter the independent

constants preceding the quantum numbers in these expressions are
6.42 X 10~6, 5.22 X 10~16, and 1.53 X 10~28 for A, Bf and C, respectively.
The first term in Eq. (20.1), involving F to the first power, is called
the first-order Stark effect; the second term, involving F to the second
power, is called the second-order Stark effect; etc. The values of A, B,
and C indicate that, if the field is not too large (F < 100000 volts per
centimeter), the lower states of the hydrogen atom (n small) would be
expected to show only a first-order Stark effect. Such fields result in a
symmetrical splitting of the energy levels about their field-free positions.
The second-order effect, ^hich is always present and becomes large for
higher states and higher fields, results in a unidirectional displacement
of each line. One peculiar result should be pointed out before going
Sac. '20.31 THE STARK EFFECT 403

further, viz., that the first-order Stark effect is observed first in hydrogen¬
like atoms, whereas the second-order Stark effect predominates in most
others. The latter will be considered more in detail in Sec. 20.8.
20.3. Early Orbital Model of Hydrogen in an Electric Field.1—The
early treatment by Epstein and Schwarzschild of the hydrogen atom in
an electric field is shown schematically in Fig. 20.1. We have seen in
Chap. Ill how the Bohr-Sommerfeld orbits, in field-free space, are
quantized, and how the size and shape of each orbit are given by the
total quantum number n, the azimuthal quantum number l (or k), and the
radial quantum number r. In a magnetic field we have seen how

the orbit becomes oriented, or space-quantized, and how the magnetic

quantum number is introduced.
Now the action of an external electric field on an orbital electron
is quite different from the action due to a magnetic field. So different
in fact is the new motion that a different system of coordinates is required
for a solution of the problem. The required solutions are found in
parabolic coordinates, and the quantum numbers describing the motion
are those given in Eq. (20.2). The quantum number mh which gives the
projection of the orbital angular momentum (i.e., mih/2r), on the electric
axis, is .to be compared with the magnetic quantum number m{. It
should be pointed out that in an electric field the orbital angular momen¬
tum is not necessarily constant as it is in a magnetic field.
1 For a treatment of the Stark effect in terms of electron orbits, see A. Sommerfeld,
“Atomic Structure/1 p. 276, 1923; see also A. E. Ruark and H. C. Urey, “Atoms,
Molecules and Quanta/1 p. 149, 1930.
 404 INTRODUCTION TO ATOMIC SPECTRA [Chap. XX

The quantum numbers nx and n2| called the 'parabolic quantum

numbers, are closely analogous to and replace the radial and azimuthal
quantum numbers of field-free space. Just as r and l determine the
range of the electron-nuclear distance rmin (perihelion) to tw (aphelion),
so ni and n* determine the range of the electron’s motion in the electric-
field coordinates as shown in Fig. 20.1. While nx limits the motion to
the region between the two paraboloids of revolution £mhl and
n% limits the motion to the region between the two paraboloids of revolu¬
tion J7min and vm*x. The region intersected by the two pairs of parab¬
oloids is shown in cross section only. Confined to this region the
electron has three periodic motions, one around the field direction given
by mh and one each along the £ and rj coordinates given by nx and n2.
Since the last two periods are not necessarily the same, the electron will
in time have covered every point in the cross-sectional area shown.1
If mt 0, nx 7* 0, and n2= 0, then r/mm = and the motion is con¬
fined to move along a paraboloid given by 17 = constant between the
limits £min and If nx and n2 are both zero, ^mtn = tj^ and £mm =
and the electron moves in a circle with mt = n — 1.
Although quantum-mechanical models in the form of electron-cloud
pictures have not been made for hydrogen-like atoms in an electric
field, there is little doubt but what they closely resemble the orbital
model just as they do for the field-free states shown in Chaps. IV and IX.
It will be seen in Sect. 20.5 that in a strong electric field the spin of the
electron adds very little to the orbital picture presented above. This
is to be contrasted with the large contribution of the spin to the energy
in a magnetic field, as shown by the anomalous Zeeman effect.
20.4. Weak-field Stark Effect in Hydrogen.—A treatment of the
hydrogen atom in a weak electric field has been given for the orbital
model, neglecting electron spin, by Kramers,2 and for the quantum-
mechanic ateaodel, including spin, by Schlapp.3 This latter treatment
employs the Dirac electron theory and is therefore to be preferred.
Although Schlapp’s derivations will not be given here, we may neverthe¬
less form some picture of the atom in a weak field and see in a qualitative
way how the energy levels and spectrum lines should split up according
to his results.
By a weak electric field in hydrogen is meant one in which the inter¬
action energy between the electron resultant j* and the field F is con¬
siderably less than the magnetic interaction energy between l* and 8*.
In other words we have a weak field when the Stark splitting is small
1 Attempts to represent the various possible motions of an orbital electron in an
electric field have been made by R. W. Wood, Phys. Rev., 88, 346, 1931; see also A.
Sommerfeld, “Atombau.”
* Kramers, H. A., Zeits. f. Phys., 8, 199, 1920.
* Schlapp, R.f Proc. Roy. Soc., A, 119, 313, 1928.
Sbc. 20.4] THE STARK EFFECT 405
compared with the fine-structure splitting. In a weak field the spinning
electron, which may be thought of classically as a small magnet, does
not interact with the field so that the coupling of j* with F is due only to
the interaction of l* with F. On the classical picture of a precessing
atom the electron mechanical resultant j*h/2t precesses around the
field F, somewhat as shown in Fig. 20.2. The projection of j* on the
field direction F is given by m„ where m2 takes values differing from each
other by unity from + j to —j.
The important difference between the Zeeman effect and the Stark
effect is that each pair of levels +m, and — my arising from a given level
have the same energy when in an electric field but different energies

Fig. 20.2.—Classical vector model for the hydrogen atom in a weak electric field.

when in a magnetic field. The state m,• = § in Fig. 20.3, for example,
has the same energy as the state m} — — f. Similarly the states m, = +J
and my = have the same energy. Instead of a level j = $ being
split up into four components as in the Zeeman effect, there are but two
levels. The reason for this can best be understood by referring to the
classical orbital model, or to the quantum-mechanical model of electron
clouds (see Figs. 9.8 and 9.10). The nature of the forces acting on the
electron are purely electrostatic so that the energy of the electron in an
orbit of given n and l depends only on the inclination of the orbit plane
with respect to the electric field, or to the distribution of charge on the
quantum-mechanical model, and not on the direction of rotation or
motion of the electron in its orbit. The states with +my and — my
correspond to the same inclination of the orbital plane, or to the same
charge distributions, and therefore have the same distortion, or energy
change, due to the applied field. In a magnetic field, on the other hand,
406 INTRODUCTION TO ATOMIC SPECTRA [Chap. XX

the energy depends on the direction of rotation, and the energies change
sign when m,- changes sign. It shouid be noted in Fig. 9.8 that for given
n, l9 and j each pair of states ±m7- have the same charge distributions.
Again each pair of states with the same n, j7 and tn, but different l have the
same angular distribution but different radial distribution.
The energy levels of the hydrogen atom for the two states n — 2
and n = 3 are shown in Fig. 20.3. The field-free levels and theoretical
pattern for H« are given at the left and the weak-field levels and pattern
are given at the right. Theoretical treatments of the Stark effect show
that wherever two or more levels with the same n and j but different

Fig. 20.3.—Fine structure and weak-field Stark effect for hydrogen H«. (After Schlapp.)

I lie close together the first-order Stark effect predominates over the
second-order Stark effect. At the left in the figure it is to be noted that
states for which n and j are equal fall together (see Fig. 9.3), and that
these same levels at the right show a large first-order splitting. The
only unpaired state for each n is in each case the one at the top, 2p,
2P§, and 3d,I 22>i, and these show only second-order splittings too small
to show in the figure. The reason why a first-order effect occurs in
the double levels and not in the single may be explained as follows:
Owing to the overlapping of angular wave functions, and the mutual
energy shared by the two states, the atom in either one of the shared
states will possess a net electric moment. When a state is unpaired.
Sec. 20.5] THE STARK EFFECT 407

there can be no appreciable electric moment and hence only a second-

order effect.
Due to the space quantization of the mechanical moments, as shown
in Fig. 20.2, the electric moment interacts with F to give j + I equally
spaced levels. The quantum numbers for weak-field Stark effect are
n, j, and my. Although l is no longer a quantum number, Schlapp
arbitrarily assigns the upper levels of each parent j to the higher l value
l — j + 2» and the lower levels to l = j — £. All weak-field levels
arising from the same fine-structure term j retain the same j value.
The selection rules for the weak-field Stark effect are the same as
those for the Zeeman effect:
Amy = 0 gives p components,
Am, = ± 1 gives s components. (20.5)
The relative intensities of the various components shown at the lower
right in Fig. 20.3 are those calculated by Schlapp. Each fine-structure
component in H« should show a sym¬ F
metrical pattern. Since the fine structure
without field has never been completely
resolved, these patterns have never been
observed. Due to the falling together of
levels with the same m„ the s com¬
ponents when viewed parallel to F will
(since each arises from two opposite
circular polarizations) be unpolarized.
20.6. Strong-field Stark Effect in
Hydrogen.—A strong electric field for
hydrogen may be defined as one for which
the interaction energy between the elec¬
tron and the field F is greater than the
interaction energy between the electron Fio. 20.4.—Classical vector model
for the hydrogen atom in a strong
spin and orbit. In other words when the electric field. Space quantization for
splitting of the energy levels due to the the Stark effect.
field is greater than the fine-structure splitting, the field is said to be strong.
In such a field the magnetic coupling between l* and s* is broken down, and
l* is quantized with respect to (and on the classical model precesses
independently around) the field direction F as shown in Fig. 20.4. While
the spin is not acted on by the external electric field, it does interact with
the magnetic field of the orbital motion l*. This interaction takes place
even though s* is no longer coupled with the orbit to form j*. Resolv¬
ing the magnetic field due to l* into two components, one along F and
the other normal to F} the normal component will average to zero.
As a result of this, $* will precess around the component parallel to
F. As in a strong magnetic field the spin s* therefore precesses around
408 INTRODUCTION TO ATOMIC SPECTRA [Chap. XX

the electric field direction but for a different reason. We may, there¬
fore, ascribe quantum numbers mi and m, to these two motions where
mi takes values from +1 to — Z, and m. takes the values +§ and — £.
[This is the mi of Eq. (20.3).]
In such a strong field Epstein and Schwarzschild, and others, have
shown that the first-order Stark effect is many times larger than the
so-called second-order Stark effect (see Sec. 20.2), and that the inter¬
action energy of Z* with F is given by the first term of Eq. (20.1).

AT = AF = 6.42 X 10-*n(n2 - nJF. (20.6)

Although the coupling between l* and 5* has been broken down

in a strong field, there is still a magnetic interaction between the two
vectors which is of the same order of magnitude as the fine structure
in no field, and this must be added to Eq. (20.6). Just as in the Zeeman
effect this energy is given by Eq. (10.29) as

T = al*s* cos (1*8*) cm-1. (20.7)

Even though Z is not a quantum number, the cosine when averaged

should give [see Eq. (10.33)]

T = amm, cm-1. (20.8)

Adding this energy to Eq. (20.6), each strong-field term is shifted by

AT = 6.42 X 10~5n(n2 — ni)F + amtm9y (20.9)

where a is the l*s* coupling coefficient (see Sec. 8.6) determined from the
fine structure, and where F is measured in volts per centimeter.
Let us now apply Eq. (20.9) to the specific example of the first three
states of hydrogen and then the selection rules to the first member
of the Balmer series Ha. The energies for each quantum state are
easily calculated by tabulating as in Table 20.1.
The allowed values of n, ni, and n2 are first written down in columns
to insure obtaining all possible combinations. The quantities n(n2 — n0
are next computed. From Eq. (20.3) the values of mi are obtained.
For every possible state mi there are the two possibilities m, = +£
and m, = —The a coefficients in the last column are not all equal
as each is a measure of the spin-orbital interactions and there are several
different Z’s. We see from this table that there is exactly the same
number of possible states of the atom as in the Zeeman effect. In the
Stark effect, however, certain levels fall together. Plotting the energies
as given by Cols. 5 and 8, each weak-field level ra;* is connected to a
strong-field level with the same m,-, such that no two levels with the same
m, cross.
Sbc. 20.5] THE STARK EFFECT 409
Tablb 20.1.—Energy Factors for the Stark Effect in Hydrogen

n n% ni n2 - »i n(ns - ni) mi amim.

1 0 0 0 0 0 ±i 0

+1 +1 0 0 -1 To/2
o +1 0 +1 +2 0 0
0 +1 -1 -2 0 ±i 0
0 0 0 +1 ±* ±a/2

+2 +2 0 -2 ±*
Ta
+2 +1 +1 +3 -1 ±i To/2
+2 0 +2 . +6 0 ±* 0
+1 +2 -1 —3 -1 ±* To/2
3 +1 +1 0 0 ±* 0
+1 0 +1 +3 +1 ±i ±o/2
0 +2 -2 -6 0 ±\ 0
0 +1 -1 -3 +1 ±\ ±o/2
0 0 +2 ±h ±o

As in the Paschen-Back effect the selection rules for a strong electric

field may be stated as follows:
A m, = 0,
A mi — 0 for p components, (20.10)
Ami = 1 for s components.

Fxg. 20.5.—Stark effect of hydrogen, showing weak- and strong-field energy le\ els for » ■■ 2
and n « 3-
410 INTRODUCTION TO ATOMIC SPECTRA [Chap. XX

Intensity rules have been worked out from the quantum mechanics
by a number of investigators. Those shown in Fig. 20.6 have been
calculated by Schlapp using as a model the Dirac electron as applied
to hydrogen. Although the weak-field patterns have not been observed,
the strong-field pattern for each line has been observed and found to
be in good agreement with theory, in both the intensities and the relative
separations. It should be pointed out that under different electrical
conditions in different types of sources the relative intensities of the
components are different. The microphotometer curve shown for
Ha in Fig. 20.6, taken from the work of Mark and Wierl,1 is observed

Fig. 20.6.—Stark effect of hydrogen, showing calculated and observed structure for H«.
(Observed field-free -pattern after Spedding, Shane, and Grace; observed strong-held pattern
after Mark and Wierl.)

to be in excellent agreement. The fine-structure splitting due to the

magnetic l*s* interaction is not resolved. The lower figure if plotted
to the scale of the field-free line above would be many times wider
than shown. A reproduction of Mark and Wierl’s published photograph
of H« is given in Fig. 20.7. In enlarging this photograph the photo¬
graphic paper was moved in the direction of the lines during the exposure.
20.6 Second-order Stark Effect in Hydrogen.—In field strengths
of over 100000 volts per centimeter the Stark components of the hydrogen
line patterns show deviations from the shifts given by the formulas
for the first-order Stark effect. This effect, which is found experimentally
to shift different components to the red or violet by an amount pro¬
portional to the square of the field strength, is called the second-order
Stark effect. In other than hydrogen-like atoms the same phenomenon
is called the quadratic Stark effect.
The first observations of the second-order effect in hydrogen were
made by Takamine and Kokubu2 in the Stark pattern of HY when
1 Mark, H., and R. Wierl, Zeits. f. Phys.} 55, 156, 1920.
* Takamine, T., and N. Kokubu, Proc. Tokyo Math. Phys. Soe., 9, 394, 1918.
Sbc. 20.6] THE STARK EFFECT 411
the spectrum was produced in a field of 147000 volts per centimeter. In
this field the middle 8 component of H7 was observed to be shifted
toward the red by only 0.8 A. These experiments were followed up by
extensive investigations in fields of several hundred thousand volts per
centimeter by Kiuti,1 Foster,2 Ishida and Hiyama,8 Rausch von Trauben-
berg and Gebauer,4 and others.
The theoretical relation giving the second-order shift of the energy
levels in hydrogen-like atoms is given by the second equation in Eq.

Coilc.

Obs.

Fig. 20.7.—Observed Stark pattern for hydrogen Ha. (After Mark and Wierl.)

(20.2). Early derivations of this formula, using the orbital model

of the atom, did not include the constant term 19. Recent comparisons
of theory and experiment show better agreement with the newer formula
as it is given in Eq. (20.2). This may be shown by the observations
of Rausch von Traubenberg and Gebauer for two of the s components
of H7 in Fig. 20.8. The circles represent the observed shifts as measured
from the photographic plates, the solid lines the theoretical shifts accord¬
ing to Eq. (20.2), and the dotted lines the shifts according to the older
theory. Equally good agreement has been found in other components
and other line patterns in hydrogen.
In measuring the second-order shifts of various components of a
given Stark pattern better agreement with theory is found by taking
1 Kiuti, M., Japan. Jour. Physics, 4, 13, 1926.
* Foster, J. S., Astrophys. Jour., 63, 191, 1926.
* Ishida, Y., and S. Hiyama, Inst. Phys. Chem. Research, Tokyo, Sci. Paper 9,
Vol. 152, 1928.
4 von Traubenberg, H. Rausch, and R. Gebauer, Zeits. /. Phys., 64, 307, 1929;
.
66 2JU. 1929.
412 INTRODUCTION TO ATOMIC SPECTRA [Chap. XX

symmetrical pairs of lines and measuring the shift of their center of

gravity from the field-free lines. The wave-length shift is given by

AX = - Xo, (20.11)

where Xr is the wave-length of the red component, X, is the wave-length

of the violet component, and X0 the wave-length of the field-free line.
20.7. Stark Effect for More than One Electron.—As compared
with the Stark effect in hydrogen the effect of an electric field on atoms
containing more than one electron is in many cases quite complicated.
One of the most important features to be mentioned is the unexpected
appearance of so-called forbidden lines. These lines, involving what

Fio. 20.8.—Second-order Stark effect in hydrogen. (After Rausch von TraubenberQ

and Gebauer.) A. Wave-length shift of the middle a component of H7. B. Shift of the
a components, ±3, of H7.

in field-free space would correspond to a change in the azimuthal quantum

number of zero, two, three, four, etc., are often the strongest lines in a
pattern. The reason for this is that in an electric field the quantum
conditions are quite different and the selection rules apply to a new set
of quantum numbers.
Although the splitting up of the energy levels for different atoms is
different from level to level and from atom to atom, the formulas derived
from the quantum mechanics for hydrogen are found to apply in many
cases. To be more exact, the more nearly hydrogen-like a given term
happens to be, the more nearly the Stark splitting will follow the hydrogen
formulas [Eqs. (20.1) and (20.2)]. Hydrogen-like states in complex
atoms are those for which the effective quantum number n0 is nearly
an integer, and the quantum defect (see Sec. 7.4) is practically zero.
Such terms arise when the excited electron, or running electron as it is
sometimes called, is in a nonpenetrating orbit like 2p, 3d, 4/, 5/, 5g,
6/, fy, 6h, etc.
If the excited electron moves in a penetrating orbit, the state will
not show, in ordinary fields, a first-order Stark effect. A classical
explanation that has been given for this is as follows: Owing to the rapid
precession of a penetrating orbit when in a non-Coulomb force field
the electric moment will precess rapidly in the orbit plane so that its
Sec. 20.7] THE STARK EFFECT 413
time average in any direction will vanish. With no electric moment
there should be no first-order Stark effect. In an electric field, however,
the orbit will be displaced as a whole in one direction and this will give
rise to a quadratic, or second-order, Stark effect. On the quantum
mechanics one might say, just as in hydrogen, that, due to the overlapping
of wave functions and the mutual sharing of energy of two close neighbor¬
ing states, the atom in either of the two states will possess a net electric
moment. This moment, due to the space quantization of mechanical
moments, will interact with the field in such a way as to give rise to
equally spaced levels, the intervals between which will be proportional
to the first power of the field F. The farther removed a state is from
other hydrogen-like states, the smaller will be the net electric moment

Fig. 20.9.—Quadratic Stark effect for the potassium doublet — 5p,sP^|f X4047
and X4044. (After Grotrian and Ramsauer.)

and the smaller the first-order splitting. Just as in hydrogen we can

now say that, if the excited electron is in a hydrogen-like orbit and
there are no other hydrogen-like orbits with the same n, the state will
show no first-order effect.
As an example of the practical application of the above principles,
let us consider the sodium D lines arising from the transitions 3s,2$j-3p,
2Pjj. Here these states, arising from electron orbits with the same total
quantum number n, are thousands of wave numbers apart as well as a
long way from the almost hydrogen-like 3d,2D§.§ states. These lines
have been studied in strong electric fields by Ladenburg1 and found to
show only a quadratic Stark effect. In a field of 160000 volts cm-1 the
various components for the two lines were displaced to the red by a
small fraction of an Angstrom unit. Again in potassium, Grotrian and

1Ladenburq, R., Phya. Zeits., 22, 549, 1921; Zeits. f. Phys., 28, 51, 1924; Ann.
d. Phya., 78, 675, 1925.
414 INTRODUCTION TO ATOMIC SPECTRA [Chap. XX

Ramsauer1 have observed similar shifts, with slight splitting into com¬
ponents of the doublet 4— 5p,aP*,§. A reproduction of their measure¬
ments on this doublet is shown in Fig. 20.9.
If a given orbit is not far from hydrogen-like, the term will show
only a quadratic Stark effect in a weak held. As the field increases,
and the levels from the hydrogen-like orbits of the same n begin to overlap
the one in question, the term will begin to show a first-order Stark
effect. An electric field which, for a given level, gives only a quadratic
effect is sometimes called a weak field. A field strong enough to give a
first-order effect is also called a strong field. We therefore see that what
may be called a weak field for one energy level may be a strong field
for other levels of the same atom. (It should be noted that this reference
to weak and strong fields cannot be applied to hydrogen, for there the
first-order effect is observed first in weak fields and then the second-order
effect in strong fields.) A pure quadratic Stark effect arises when the
conditions for weak field are satisfied for both the initial and final states.
These are the conditions under which Fig. 20.9 was obtained. As the
field increases, the initial state usually goes over to the strong-field case
long before the final state does. When the field is strong for both the
initial and final states the Stark separations are given by AF + BF2
[see Eqs. (20.1) and (20.2)], and the resultant Stark patterns will resemble
those of hydrogen. We shall now apply these principles to the Stark
effect of helium.
20.8. The Stark Effect in Helium.—The appearance of forbidden
lines in helium when the spectrum is produced in an electric field was
first observed by Koch.1 Following Koch’s discovery, Stark3 and his
coworkers were able to produce long series of forbidden lines like ls2s,*S-
1 snst*S, l«2p,*P — 1 snp,zP, etc., where Al = 0 and ls2s,3& — 1 snd,*D,
l«2p,*P — 1 snf,zF, etc., where Al = 2. Later experiments by Foster4 and
others extended the observations and, under improved experimental
conditions, brought out the significance of these new lines. The correla¬
tion between the Stark patterns of helium with its two electrons and the
Stark patterns of hydrogen with its one electron is one of the most
interesting and remarkable achievements of our modern theory of
atomic spectra.
As a specific example of the Stark effect in helium let us consider
the spectrum line X3705, arising from the transition ls2p,3P — ls7d,*D,
and examine the various energy levels, transitions, and observed Stark
components associated with it. From the energy level diagram of

1 Gbotrian, W., and G. Ramsauer, Phys. Zeiis., 28, 846, 1927.

* Koch, J., Ann. d. Phys., 48, 98, 1915.
•Stark, J., Ann. d. Phys., 48, 210, 1915; 56, 577, 1918.
4 Foster, J. S., Phys. Rev., 28, 667, 1924; Proc. Roy. Soc., A, 114, 47, 1927; A, 11/,
137, 1927.
Sbc. 20.8] THE STARK EFFECT 415
helium in Fig. 12.14, it is seen that both of these terms are nearly hydro¬
gen-like. By hydrogen-like is meant that the term value is within a
few wave numbers of the corresponding term value in a hydrogen-like
atom. In such a state the excited electron is in an orbit not greatly
different from one of the hydrogen orbits. (Compare the He 2p curve
in Fig. 12.15 with that of H in Fig. 4.6.)
Ml Mf
0 a*i

*1 a«

n an
o

13
*1 an
*4
*2 an
0
*5
*3 an
*1
±6
*4
*2
an
0
±5
*3
11

14
12
0
13
U an

12 an
0

11 an

0 an

Fig. 20.10.—Theoretical Stark splitting for the 1«7Z states of helium, where l — 0, 1, 2, 3.
4, 5, and 6, for a, p, d, f, g, h, and i electrons respectively.

Turning first to the final state l«2p,8P, we find that the only other
near-by states are x8 and *S, arising from ls2s several thousand wave
numbers lower. From the discussion in the previous section one should
therefore expect only a quadratic Stark effect for both the lP and *P
terms of ls2p. Even the very highest fields yet attained would not be
strong enough to bring the ls2s and ls2p sets of levels near enough
together to give rise to a first-order Stark splitting. We may therefore
assume little or no splitting and a very small shift in the final state *P
for all ordinary fields.
416 INTRODUCTION TO ATOMIC SPECTRA [Chap. XX

Turning now to the 1«7d,*D initial state which is almost hydrogen¬

like, we find ls7p,3P 42 cm*"1 lower and 1$7$,*S 159 cm”1 still lower.
Above ls7d,*Z>, and within but a few wave numbers, must lie the states
arising from the configurations 1$7/, ls7g, 1«7A, and ls7i (see Fig. 20.10).
Although not observed in connection with the ordinary spectral series,
these levels certainly do exist as possible states of the helium atom.
Due to the close proximity of these latter states, an electric field will
not have to be very strong to produce a first-order Stark splitting of *D.
At first the 3P and 3S states will show a second-order Stark splitting.
As the field is increased, and the components of the upper levels approach
those of 3P, the levels will begin to show a first-order effect. Finally,

field strengths.

when the field and splitting are large enough, the 3S term will show a
first-order splitting. In these strong fields the spin-orbit interaction
SL is broken down and the quantum number L no longer has meaning.
Only the projections of the orbital and spin momenta on the field direction
are quantized, and we have as quantum numbers ML and Ms in place
of mi and m, as in hydrogen. For each quantum value of Mh there
will be three possible values of MS) viz.} 1, 0, and — 1. These are indi¬
cated at the right in Fig. 20.10. Although there is no coupling between
the spin resultant S and the field P, S will interact with the orbital
magnetic moment projected on F (see Sec. 20.5) and give rise to splittings
of the same order of magnitude as the fine structure of each triplet.
This splitting is too small to show in the figure. The small separations
appearing at the right are shown for the purpose of correlation with field-
free states.
Mention should be made at this point of the singlet terms arising
from each of the configurations shown in Fig. 20.10. These singlets in
field-free space lie just above each corresponding triplet, and these go
over in an electric field to an array of levels almost identical with the one
Sec. 20.8] THE STARK EFFECT 417

shown for the triplets. Due to the spin-spin interaction this set of levels
will be just above but almost superposed upon the triplet array.
In considering the allowed transitions between two sets of levels the
selection rules,
A Ms - 0,
AMl = 0 for p components, (20.12)
AM l = 1 for s components
must be applied. With the set of levels shown at the right in Fig. 20.10
as initial states, and the states Ml — 1, 0, and —1 (from ls2ptzP)
HELIUM
4120 4026 3867 3819 J733 3705

Fia. 20.12.—Stark effect in helium. (Enlarged from a section of a plate taken by J. S.

Foster.) The field-free spectrum is shown between the p components above and the *
components below.

as final states, transitions from the states ML = 0, ±1, and ±2 are the
only ones allowed. Since the final levels show only a quadratic effect,
the three lower levels will be superposed and the observed spectrum
lines should follow the dotted lines in Fig. 20.11. The observed patterns,
as photographed by Foster, are reproduced in Fig. 20.12. These observed
lines are drawn as heavy lines in Fig. 20.11. In very weak fields only
certain of the lines from levels with higher L values appear. Most of
these in zero field are forbidden transitions. As the field increases, the
2ZP — VP transition appears. Quantum-mechanical calculations for
the tracing of various levels from zero field to strong field have been carried
out by Foster1 and shown to be in good agreement with experiment.

1 Foster, J. S., Proc. Roy. Soc., A, 117, 137. 1927.

 CHAPTER XXI

THE BREADTH OF SPECTRUM LINES

It is well known from experiment that many spectrum lines have an

observable breadth that is independent of any optical system used to
observe them. In many cases narrow and broad lines are observed simul¬
taneously in the same spectrum. The sharp and diffuse series of each
of the alkali metals are good examples of this. Oftentimes it is found
that suitable methods of excitation will sharpen the lines to within the
limits of the resolving power of the best spectrographs. Other lines are
known, however, where all of the general methods used for sharpening
lines have failed to produce fine sharp lines. The copper arc lines dis¬
cussed in Chap. XIX, and reproduced in Fig. 19.14, are good examples of
this. In this chapter we are concerned with the general causes of, and
where possible the theoretical calculations for, the observed breadth of
spectrum lines. We shall start by writing down the known causes and
types of spectrum-line breadths and then proceed, in the following sec¬
tions of this chapter, to treat each one more or less independently:1
A. Doppler effect.
B. Natural breadth.
C. External effects:
a. Collision damping.
h. Asymmetry and pressure shift,
c. Stark effect.
The first of these, the Doppler effect, has to do with the distribution of
frequencies observed from many atoms moving at random as they do
according to the kinetic theory of gases. This simplest of all broadening
effects will be treated first in this chapter (see Sec. 21.1).
The second of these, natural breadth, treats of an inherent property
of the atom, independent of all external effects. The quantum mechanics
shows, for example, that the energy levels of an isolated atom are not
sharp, as given by Bohr's simple theory, but have a finite natural breadth.
This will be treated in Secs. 21.2 and 21.3.
The third and perhaps the least understood of the causes for the
broadening of spectrum lines includes all external effects produced by
(1) collisions between atoms and (2) fields of neighboring atoms and

1 For a more detailed account of, and references to the subject of, the breadth

,
of spectrum lines the reader is referred to the article published by V. Weisskopf,
Phys. Zeits., 84 1,1983, and also to Max Bom, “Optik,” pp. 421-455, 1933.
418
Sac. 21.1] THE BREADTH OF SPECTRUM LINES 419
t '

molecules. These effects will be treated in the last three sections of

this chapter.
21.1. The Doppler Effect.—Apparently one of the most classical
of all atomic phenomena is the effect of the Doppler principle as it
applies to the observed frequency of a radiated line. This is perhaps
best shown by the different stellar spectra and by the solar spectrum
as observed from the limbs of the sun (see Fig. 21.1). Here, where the
velocities of the stars and their emitting gases, with respect to the observer,
are high, the observed shifts of the spectrum lines are large. Quite
similar effects are well known to exist in the case of a gaseous discharge
where, due to thermal agitation, the atoms emitting light have relatively

Fig. 21.1.—Solar spectra of the east and west limbs, and the north and south poles,
of the sun. Arrows indicate solar lines showing the Doppler shift. Bracketed lines,
Bhowing no shift, are part of the oxygen A band produced by absorption in the earth’s
atmosphere.

high velocities. The random motions of the atoms or molecules in a

gas, however, produce a net broadening of the line with no apparent
shift of its central maximum. As one might well expect, this broadening
is found experimentally to (1) increase with temperature and (2) decrease
with increasing atomic weight.
For an appreciable Doppler effect the atom must have an appreciable
velocity at the time of radiation. If v is the velocity of the atom, and 0
is the angle between v and the direction of observation, the frequency
of the light will be changed by an amount Av, given simply by the classical
expression
Av _ v — vp v cos 0 __ u
Vo Vo c c
(21.1)
where v0 is the frequency of the line for v = 0, v the observed frequency,
u v cos 0 (21.2)
the component of the velocity v in the direction of observation, and c the
velocity of light. Unlike the closely analogous Compton effect this
relation does not involve h and is exactly the relation given by the
quantum mechanics. Assuming a Maxwellian distribution of the veloci-
420 INTRODUCTION TO ATOMIC SPECTRA [Chap. XXi

ties the probability that the velocity will lie between u and u + du is
given by

dw = ^e-Ko'du, (21.3)

where

* = 25* • (21-4)

Here p is the molecular weight, R the universal gas constant, and

T the absolute temperature. Substituting the value of u from Eq. (21.1)
we get for the relative intensity I as a function of the frequency v,
c*

I(p) = constants o) (21.5)

Fia. 21.2.—Intensity-frequency contour for the Doppler broadening of a spectrum line.

This general equation is plotted in Fig. 21.2 for the purpose of showing
the general shape of a spectrum line which has been broadened by the
so-called Doppler effect. To find the two frequencies at which the
intensity drops to half its maximum value1 the exponential term in
Eq. (21.5) is set equal to one-half. Solving this for v — and multiply¬
ing by two we get for the half-intensity breadth, in absolute frequency
units,

(21.6)
This equation shows that the Doppler broadening is (1) proportional to
the square root of the temperature, (2) proportional to the frequency vQf
and (3) inversely proportional to the square root of the molecular weight.
1The half-intensity breadth of a spectrum line is here defined as the interval
between two points where the intensity drops to half its maximum value. (Some
authors take half of this interval and call it the half breadth.)
Sec. 21.2] THE BREADTH OF SPECTRUM LINES 421

Remembering that Av/v » AX/X, the half-intensity breadth in terms of

absolute wave-length units is

(21.7)

As an example of the application of this equation, consider the sodium

D lines at X5893 A. For a temperature of 500° Abs., Eq. (21.6) gives a
half-intensity breadth of 0.056 cm”1 or 0.02 A, a value 200 times as large
as the natural half-intensity breadth (see the following section).
While Eq. (21.6) shows that the half-intensity breadth in frequency
units increases with frequency, Eq. (21.7) shows that, in wave-length units,

Fig. 21.3.—Amplitude-time curve of a single electric charge oscillating about a fixed point.
A damped oscillator.

it decreases. The latter would show that the x-ray region of the spectrum
might be suitable for the measurement of natural breadths (see Sec. 21.4).
21.2. Natural Breadths from Classical Theory.—According to classical
electromagnetic theory a vibrating electric charge is continually damped
by the radiation of energy. The energy E of such an oscillator decreases
exponentially by
E - E^r*1 (21.8)
and the amplitude (see Fig. 21.3) by

y T
422 INTRODUCTION TO ATOMIC SPECTRA [Chap. XXI

Here a>o is equal to 2r times the frequency of the oscillation p0. The
displacement z of the oscillator at any time t is given by

x = Aoe * cos (wot + ip). (21.11)

An ordinary Fourier analysis of the damped wave emitted by the
oscillator gives for the energy, or better the intensity which is proportional
to the energy radiated, as a function of the frequency,

I{v) (21.12)
2r 4rs0o - >)2 + (|)'

The general contour of such a distribution is shown in Fig. 21.4.

The half-intensity breadth 5* of this symmetrical distribution, like the

Fio. 21.4.—Intensity-frequency contour for the natural breadth of a spectrum line.

Doppler half-intensity breadth, is here defined as the interval between the

two points where the intensity drops to half its maximum value. As seen
from Eq. (21.12) the intensity will drop to half its maximum when the
two terms in the denominator are equal, i.e., when

Vo ~ ”= i ■ \6'«= (2L13)
which gives for the natural half-intensity breadth
y 2. 4re2v20
(21.14)
n 2r 3mc*'
Since Av/v = AX/X, and c = v\, the half-intensity breadth, in terms
of wave-length, is
c Awe2
AX 1.16 X 10-12 cm (21.15)
2TV** Zme2
which is constant and equal to 0.000116 A for all wave-lengths, a value
many times too small to be measured by ordinary spectroscopic methods.
21.8 Natural Breadths and the Quantum Mechanics.—According
to quantum-mechanical principles an energy level diagram of an atom
is not tb be thought of as a set of discrete levels but as a sort of continuous
Sec. 21.3] THE BREADTH OF SPECTRUM LINES 423
term spectrum in which the probability distribution is concentrated
in regions where the terms are observed (see Fig. 21.5). With such
probability distributions for the different levels of a given atom the
transitions between levels will not give rise to infinitely sharp lines.
From Heisenberg’s uncertainty principle, expressed in terms of energy
and time,
AEAt ~ h, (21.16)
it is not difficult to understand how it is that the energy levels are
not discrete. Writing for AE the uncertainty in the energy hAT, and
for the time At the mean life t, the uncertainty in
the breadth of an atomic state in absolute
frequency units is of the order

AT ~ ~ (21.17)

This may be interpreted to mean that the

greater the mean life of an atom in a given
state the smaller will be the half-intensity
breadth of the state. It follows therefore that
normal states, or metastable states, of atoms
will be sharp whereas other states will be broad.
It is not surprising to find that the quantum
mechanics leads to an expression for the half¬
intensity breadth of spectrum lines which,
although different for different levels, is of the
same order of magnitude as that given by the Fio. 21.5.—Schematic
classical theory of radiation damping. The ilJU8trating, natural
, _ , , T . re breadths of energy levels.
probability distribution I(T) of an energy state
for an atom on the quantum mechanics has been given by

I(T) = (21.18)
' ' 2x 4r*(Tn - T)2 + (7n/2)2
where Tn is the term value of the center of gravity of the term distri¬
bution, and
.
7»
_ XV
-
&***2(Tn
3mc, - Tmy 1
(21.19)
T

Here Tm is the term value of any lower level to which a transition may
take place. Tn — T represents small Av intervals in the initial term
and Tn — Tm the mean radiated frequency vn,m corresponding to transitions
between the two levels. With the above value of yn the term half¬
intensity breadth is given [see Eq. (21.14)] by
W(7\> - Tm)2
n’m 3mc*
(21.20)
424 INTRODUCTION TO ATOMIC SPECTRA [Chap. XXI

Now the coefficient Cn,m varies from level to level and depends for its
value on all other states of the atom lower than the one in question:

Cn,m = (21.21)

Here gn and gm are the quantum weights (2J + 1) of the levels for the
transition n —> m, and fn,m is the number of so-called dispersion electrons*
for the same transition. The sum is to be taken over all levels m which
lie deeper than the level n.
Although not easily determined the / factors in the above given
equation are a measure of the intensity of a spectrum line and in a few
simple cases have been calculated1 and measured.2
London has shown from Schrodinger’s theory that the number of
dispersion electrons to be associated with any state of the atom is equal
to unity. This means that, if we sum up the f’s for all transitions into
and out of a given state,
2/ = 1. 1.22)
The fa for transitions into a state are taken with a plus sign and those
out of a state with a minus sign. The table of / values given in Table
21.1 will serve as a simple example of this rule. Since these are all
transitions into the normal state, 3s,2S, and there are no transitions
out, the sum, including the rest of the series and the small sum con¬
i'able 21.1.—Dispersion Electrons, / Factors. Observed Values to Be Asso¬
ciated with the Principal Series of Sodium
(After Filippov and Prokofjew)1

Transition Xlir Transition Xilr fn,m

3s2S - 3 p2P 5893 0 9755 3.s, 2S — nP,*P 2476 0 0000384

3s,2S - 4 p2P 3303 0.01403 3 s,2S - \2p,*P 2464 0 0000284
3s,2S - 5p,2P 2853 0.00205 3s,2S - 13 r,*P 2456 0 0000217
3s,2S - dp,2P 2680 0.000631 3.v, 2S - 14 p,2P 2449 0 0000173
3s,2S - 7p‘P 2594 0 000256 3s,2S -
15 V,* 2444 0.0000140
3s,2S - 2544 0.000134 3 s,2S 16p,2P
- 2440 0 0000116
3s,2S - 9 V,'P 2512 0.0000811 3s,2S - 17 v2P 2437 0 0000092
3s,2S - 10 v,*P 2491 0 0000537 3s,2S - 1 Sp,2P 2434 0.0000075

1 Filippov, A , and W. Pbokofjew, Zeits /. Phys , 56, 458, 1929.

tributed by the continuum above the series limit, should add up to

unity. It is to be noted that almost the entire sum in this case comes
from the resonance lines X5890 and X5896 (average 5893) due to the
1 London, F., Zeits. f. Phys., 39, 322, 1926; Hargreaves, J., Proc. Camb. Phil.
Soc.t 25, 323, 1929; Trumpy, B., Zeits. f. Phys., 54, 372, 1929; 61, 54, 1930; 66, 720,
1930.
2 Filippov, A., Zeits. f. Phys., 69, 526, 1931; Minkowski, R., Zeits. f. Phys.. 36
839, 1926; Ladenburo, R., and E Thieie Zeits f Phys 72.697 1931
Sbc. 21.3] THE BREADTH OF SPECTRUM LINES 425
transitions 38,2S\ — 3p*P% and 3s, — 3p,2P*. Taking this fine struc¬
ture into account the / factors for each of these two lines separately
are 0.3252 and 0.6503, respectively. Using these values we shall now
return to the calculation of half-intensity breadths. Remembering that
the summation of the fs for a given state is to be taken over those
levels lying deeper than the one in question, the .
value of C for 3p,2P$ and for 3p,2P [Eq. (21.21)] J
is practically unity. For 2P*, 2fn,m — 0.3252, h,tu
gn - 2, and = 2. For 2P;, 2/n,m = 0.6503, **
gn — 4, and gm = 2. These levels, for these y
particular lines, have just the classical half¬
intensity breadth of 0.000116A. By Eq. (21.20)
this corresponds to a mean life of J

r = = 1.6 X 10~8 sec. (21.23) 'SS

It is not difficult to show that, if 5i = 7i/27r

and 82 = 72/2tt represent the respective half¬
intensity breadths of the initial and final states
for a given transition, the intensity distribution
of tne
01 the observed spectrum line willwill be
oe given by
oy Radiated line

Jf N _ 71
7i +
+ 72
72 1 Fio. 21.6.—Showing th
I\v) — 7y * T~ jT ~ \ 2* relation between the natur
T 4 tt2(vi 9 —
4tt2(vi(2 vv)2
)2 +-
4 ( '1 1 breadth of a spectrum lit
V 2
y 2
)
J
/ni Ai\
and the natural breadths
the corresponding energ
(21.24) levels.

From this relation we observe that the line drops to half its maximur
intensity when the first and second terms in the second denominator ar*
equal, i.ewhen

2jr(n.s - v) = ± ?-*• (21.25

From this we see that the half-intensity breadth of the radiated lin»

7i + 72
K = vi,2 — v = « *x + h

is just equal to the sum of the half-intensity breadths of the initial an

final states. This is shown schematically in Fig. 21.6, where both th
initial and final states have been drawn with relatively large breadth
For the sodium D lines, discussed above, the final state with an almot*
infinite mean life is sharp so that the natural breadths of the observe
lines will be due almost entirely to the spread of the initial states. Mea
urements of the half-intensity breadths of spectrum lines will be discuss*
in Sec. 21.4.
426 INTRODUCTION TO ATOMIC SPECTRA IChap. XXI

21.4. Observed Natural Breadths and Doppler Broadening.—In the

preceding section we have seen that the Doppler breadths of the sodium
D lines at normal temperatures and pressures are many times the natural
breadths as given by Eq. (21.26). In spite of this observed fact fruitful
attempts to measure the natural breadths of many lines have been
made. In order to see how this is possible the two types of curve are
superposed in Fig. 21.7. One represents the natural breadth of a
line and the other the Doppler breadth for the same total intensity.
Each curve assumes all other broadening effects are absent. It is
observed that, while the Doppler intensity drops off sharply and becomes
negligibly small in a very short frequency interval, the natural breadth
remains fairly large at considerable distances from the center. If the

same half-intensity breadth and the same total intensity.

Doppler breadth is not too large it is easy to see how one might measure
the intensity contours far out on the wings and from these calculate the
half-intensity breadth. This has been done for the sodium D lines by
Minkowski.1 In this particular instance almost perfect agreement
between theory and experiment was found. Many measurements of
other lines have been made in this indirect way with varying degrees
of agreement.
In hydrogen the Doppler effect has been the chief limiting factor
m all attempts to resolve the fine structure of the Balmer series lines
(see Fig. 9.5). For Ha, at temperatures between 250° and 300° Abs., the
Doppler half-intensity breadth is about 0.2 cm-1, or 0.09A, a value of the
same order of magnitude as the largest fine-structure interval. By oper¬
ating a hydrogen-discharge tube in a liquid air bath the Hi and Hi lines
have been considerably sharpened. Attempts to excite the lines with the
tube cooled in liquid hydrogen have been unsuccessful. The effect of
atomic mass on the Doppler breadth is very beautifully brought out by
the comparison of the H« lines reproduced in Fig. 9.5. The Hi lines

Minkowbki, R.. Zeits. /. Phys., 86, 839, 1926.

Sec. 21.6] THE BREADTH OF SPECTRUM LINES 427
» ' f
are 1.4 times as broad as those of H£. The measured contours of these
and other lines as well as their increase in breadth with temperature are
in excellent agreement with Eq. (21.24).
Mention should be made at this point of the several attempts that
have been made to measure the natural breadths of x-ray lines. The
results thus far obtained in this region of the spectrum show that, while
the shapes of the lines studied are not far from that expected, the half¬
intensity breadths are several times larger than is predicted by theory.
Experimental verification of the natural contour of a spectrum
line has been made in the case of the anomalously broad lines so well
known in the copper arc spectrum (see Fig. 19.14). The intensity

Fig. 21.8.—Observed natural breadth contour of one of the anomalous copper arc lines.
See Fi«. 19.14. (After Allen.)

contour of one of these lines, as measured by Allen,1 is reproduced in

Fig. 21.8. For this as well as many other broad lines in copper the
mean life of the atom in the upper state has been greatly reduced by the
probability of autoionization, a phenomenon discussed in Chap. XIX.
Although this particular type of broadening is not due to exactly the
same effect as a natural breadth, it is here classified as such in that it
takes the place of natural breadth and is an inherent property of the
atom independent of all external effects. The observed half-intensity
breadths show that the mean life of the atom in such broad upper states
is of the order of 10“13 sec.
With the exception of a few special cases, such as the broad copper
lines, the natural breadth of a spectrum line may be neglected in compari¬
son with the Doppler breadth at ordinary temperatures. The importance
of the subject of natural breadths lies in its application to the dispersion
of a gas and to dispersion theory in general, a field of investigation
which has developed largely in the hands of Ladenburg and Kopfermann.1
21.5. Collision Damping.—One of the important external causes
in the broadening of spectral lines is the process called collision damping.

1 Allen, C. W., Phys. Rev., 39, 42, 55, 1932.

* Ladenburg, R., Zeits. f. Phys., 65, 167, 189, 1930. For other references see H.
,
Kopfermann and R. Ladenburg, Zeits. f. Phys., 48 26, 1928.
428 INTRODUCTION TO ATOMIC SPECTRA [Chap. XXI

This is but one of the effects produced by the collision of two atoms, one
of which is in the process of emitting or absorbing radiation. Following
the early suggestion of Michelson1 in 1895, and Lorentz2 in 1905, the
phenomenon is based upon the following assumptions: //, during the time
an atom is emitting or absorbing radiation of frequency v0 it collides
elastically with another atom, the phase and amplitude of the radiation
have a chance of undergoing a considerable change.
If one assumes that the time two atoms are in collision is large com¬
pared with the mean time between collisions, the emitting or absorbing
atoms will most of the time be under the influence of strong atomic
fields. These conditions, which exist in a gas at relatively high pressures,
give rise in general to a red shift of the spectrum lines. This phenomenon
will be discussed in the following
section under the heading of Asym¬
metry and Pressure Shift.
In this section we are concerned
with the damping effect, or broaden¬
ing of a spectrum line, produced by
the sudden change in the atomic
radiation by collision. The assump¬
tions usually made are: (1) the
mean time between collisions is large
Fig. 21.8a.—Undamped oscillation in- compared with the collision time,
stantly stopped after a Bhort interval of and (2) with every impact the oscil-
time. Ideal model for collision damping. , ... . . . , ~
lations are either completely cut off,
or they are momentarily interrupted during impact only to resume the
same frequency again with possibly a phase and amplitude change
(see Fig. 21.10). That the phase of the oscillations may be changed
in the short collision time may be seen from the following rough calcula¬
tion. For normal temperatures and normal-effective-collision cross
sections the collision time is of the order of 10'13 sec. For ordinary
visible light with a period of about 10~15 sec. there will be some hundred
modified oscillations during impact.
In giving a classical treatment of the phenomenon of collision damp¬
ing, it is customary to assume the radiation to be an undamped oscillation
of the form shown in Fig. 21.8a, and given by
x — Ao cos (wot + <p) (21.27)
(where «o = 2tv0), which is instantly stopped after an interval of time r.
Now r is the interval between the time the atom begins radiating and
the time a collision occurs. A Fourier analysis of a single finite wave
1 Michelson, A. A., Astrophys. Jour., 2, 251, 1895.
* Lorentz, H. A., Koninklijke. Akad. Wetenschappen Amsterdam, Versl., 14*
518, 577, 1905-1906.
Sac. 21.5] THE BREADTH OF SPECTRUM LINES 420
t ’ i

like this leads, strangely enough, to an intensity distribution which has

the contour of a single-slit diffraction pattern. The shorter the wave
train, or the longer the wave-length, the broader the central maximum
of this pattern. Now for each atom of a gas, r will have a different
value so that for the gas as a whole there will be a certain probability
distribution of the t’s. If we assume the same distribution for r as is
usually assumed for the free paths of the same gas atoms, the probability
that r will lie between r and r + dr is given by
—r
dw - —eT* dr, (21.27a)
TO

where r0 is the mean flight time, i.e., the average time between collisions.
When a Fourier analysis of the set of wave trains has been made, the
relative intensity 7, as a function of the frequency v, is found to be given
by the relation

ZM = constant^ _ ^ + (1/to)3- (21.28)

This equation gives the same intensity contour as radiation damping

shown in Fig. 21.4. The intensity drops to half its value when

o>0 ~ = — = 2ir(j'o — v) (21.29)

To

and gives for the collision-damping half-intensity breadth,

= 7TT
~ o
(21.30)

This result is to be compared with Eq. (21.20). It should be noted that

in the derivation of this equation the small effect of natural breadth due
to radiation damping has been neglected. Further consideration of
this problem shows that, if the radiation starts again after impact with
a new phase and amplitude, and radiation during impact is neglected,
exactly the same intensity distribution is obtained as given above.
Let us now find the order of magnitude of the line broadening due
to collision damping. To do this we must first find a value of r0. From
the kinetic theory of gases the average velocity of an atom or molecule
is given by

(21.31)

where R is the universal gas constant = 8.3 X 107 ergs mole-1, T the
absolute temperature, and m the molecular weight. The mean free
path Z0 is just
Zo * t>oTo, (21.32)
 430 INTRODUCTION TO ATOMIC SPECTRA [Chap. XXI

so that the half-intensity breadth

1 l8ET (21.33)
7rlo\ Tfi

If a foreign gas is present, the molecular weight p is replaced by

> mM/(m 4- Af), where m is the molecular weight of the emitting gas and M
that of the foreign gas. The mean free path, as given by the Maxwellian
formula, is also written
kT = _RT_
(21.34)
y/ 2irp2p y/ 2Nirp2p
where p is the optical cross section, or collision diameter, i.e., the average
distance between nuclear centers at closest approach, and p is the pres¬
sure. This gives for the half-intensity breadth, in terms of the Avogadro
number N = 6.06 X 1023 mole-1, and the pressure p or density d,
4 N p2p
5 (21.35)
y/ RT par
The only unknown factor in this expression is the collision diameter p.
Instead of assuming the kinetic-theory values of p and attempting to
check theoretically the observed line breadths for different pressures,
it is customary to compute values of p from the observed lines. The
following values, for example, were obtained by Schutz for the sodium
D lines excited in an atmosphere of different foreign gases (see Table 21.2).
Table 21.2.—Optical Cross Section, or Collision Diameter p, for the Sodium
D Lines Absorbed in Mixtures with a Foreign Gas
(After Schutz)

Source h2 He No A N.

Schutz1. Pnpt 5 8 5 0 r> 15 9 0 8 3 A

Kinetic theory . Pkin 2 3 1 9 2 3 2 8 3 1

» SchCtz, ,
W., Zeits. /. Phy45 30, 1927.

Since the observed values povt are from two to three times those cal¬
culated from kinetic theory the term optical cross section has been applied
in the one case and kinetic cross section in the other.
In some instances the effect of a foreign gas on the spectrum lines
of a given element is to give very broad lines and large optical cross
sections. In favorable cases popt may be from 10 to 100 times pkll). Th}s
is interpreted to mean that even at great distances another gas atom
may act on the radiating atom strongly enough to change the phase of
the oscillation. Such large actions often arise when no foreign gases
are present.
A comparison of Doppler half-intensity breadths with collision half¬
intensity breadths [Eqs. (21.6) and (21.35)] shows that for visible light,
Sac. 21.6] THE BREADTH OF SPECTRUM LINES 431
and room temperature and pressure, the two effects are of the same order
of magnitude. In mercury, Fabry and Perot reduced the pressure to
1 mm in order to measure the Doppler broadening without large distor¬
tions from collision damping. With increased pressures, on the other
hand, collision damping becomes the predominant cause for the broaden¬
ing of lines. The effect of pressure or density [see Eq. (21.35)] on the
breadth of a line is shown in the specific case of the mercury resonance
line X2537 in Fig. 21.9. Here the half-intensity breadth, as it is observed
in absorption, is plotted against the relative density, the latter being
defined as the density of the gas relative to its density at normal tem-

Fig. 21.9.—Observed pressure broad- Fig. 21.10.—Schematic diagram of

cning of the mercury resonance line the change in frequency at the time of
X2537 in absorption. (After Fucht- collision between two atoms one of
baupr, J008, and Dmkelacker.) which is radiating.

perature and pressure. The largest readings correspond to values taken

at a pressure of 50 atm. The foreign gas mixed with the mercury vapor
to obtain the upper curve was C02, and to obtain the lower curve, N2.
21.6. Asymmetry and Pressure Shift.—In the previous section we
have seen how the broadening of a spectrum line due to pressure has
been attributed to the sudden change in phase of the emitted radiation
by collision, a phenomenon called collision damping. Since collision
damping takes into account only that part of the radiation given out
before and after collision, we have yet to consider the radiation that
takes place during collision. This is shown schematically in Fig. 21.10.
In this very ideal model it is assumed that the frequency of the oscillation
changes suddenly when the emitting atom first comes under the influence
of the colliding atom. After collision the original frequency is resumed
again. Asymmetry and pressure shift are attributed to the modified
radiation indicated at the center of the figure.
The explanation has been given that at close approach of a foreign
atom the energy levels of the excited or radiating atoms in question are
altered, due chiefly to polarization effects (see Fig. 7.5).1 It is convenient
1 This is quite similar to a‘ quadratic Stark effect and might well be classified as
such (see Sec. 21.71.
432 INTRODUCTION TO ATOMIC SPECTRA [Chap. XXI
in this explanation, following Jabtonski,1 to represent the energy levels
by means of potential curves for the two atoms considered as a molecule.
In Fig. 21.11 the initial and final states of the pseudo molecule are plotted

Fig. 21.11.—Potential curves for two states of an atom at close approach of another atom.
Illustrating the cause for asymmetry and red shift. (After Jabldnski and Weisskopf.)

as functions of the distance between the atomic centers, r. Due to the

polarization of one atom at close approach of another, an excited or outer
state will be lowered more than a tightly bound lower state. The
frequency distribution during this time of close approach is added to the
collision-damping distribution. The result is that the spectrum line

Fig. 21.12.—Asymmetry and broadening of the mercury resonance line X2536 with increased
pressure. (After Fdchtbauer, Joos, and Dinkelacker.)

observed is spread out more on the long wave-length side than it is on the
short (see Fig. 21.12).
At normal temperatures and pressures the collision times are small,
and the broadening and shift are small. With increasing pressure
the mean collision time increases and the time between collisions decreases

1 Jabl6nski, A., Zeil8.f. Phpe., 70, 723, 1931; Bee also Margenau, H., Phys. Rev.,
,
43 129, 1933.
Sec. 21.6] THE BREADTH OF SPECTRUM LINES 433
with the result that, as the line is shifted to the red, it is broadened
asymmetrically. The mercury resonance line X2537 is shown in Fig.
21.12, as observed in absorption by Fiichtbauer, Joos, and Dinkelacker,1
at a pressure of 10 and 50 atm., respectively. The foreign gas used in
these observations was nitrogen. To show that the pressure shift
depends upon the foreign gas used,
curves have been reproduced in Fig.
21.13. In general it is observed that
the shift of a line is very closely pro¬
portional to the relative density. The
relative density is here defined as the
density of the gas relative to the
density at normal temperature and
pressure.
That the lowering of an energy
level due to pressure effects is greater SO 40
Relative Density
for high levels than it is for the low Fig. 21.13.—Pressure shift of the
is confirmed by Babcock’s2 interfer- mercury resonance line X2537, in absorp-
ometer measurements on the iron F^bauer, Joo,. and

lines. The well-known spectrum of

iron was investigated at pressures below 1 atm. for the purpose of
determining how reliable the different lines may be as wave-length
standards. At low pressures the lines are very sharp, and the pressure

Fig. 21.14.—Relative shifts of the energy levels of iron as a function of their term values.
(After Babcock.)

shift is extremely small. Employing Fabry-Perot 6talons, however,

shifts accurate to thousandths of an Angstrom or better were measured-
Since an observed spectrum line gives only the difference between two
1 FttcHTBAUBR,C., G. Joos, and O. Dinkelacker, Ann. d. Phys., 71, 204, 1923.
* Babcock, H. D., Aatrophys. Jour,, 87, 240, 1928.
434 INTRODUCTION TO ATOMIC SPECTRA [Chap. XXI

term values, relative shifts are all that can be determined. The results
from over 100 lines are best shown by the graph reproduced in Fig. 21.14,
where each term is plotted relative to the ground state as zero. . Weight¬
ing each plotted point, Babcock expresses the belief that the quintet and
septet terms are depressed slightly more than the triplets. Whether
this is a real effect or not, the major effect is in good agreement with what
one would expect from polarization effects explained above.
21.7. Stark Broadening.—In an ordinary arc of high current density
many ions are produced which upon collision with other atoms give
rise to strong electric fields. The effect of these intermolecular fields
is to produce a Stark-effect broadening of the observed spectrum lines.
In addition to this effect by ion fields, dipole or quadrupole moments
of gas atoms or molecules may also produce relatively strong inter¬
molecular electric fields. The hydrogen atom and molecule are good
examples of a dipole and quadrupole, respectively. Being of the same
order of magnitude as collision damping and Doppler broadening (see
Secs. 21.5 and 21.1), the Stark effect plays a relatively important part
in the general subject of the breadth of spectrum lines. In attempting
to calculate Stark-effect broadening one is immediately confronted
with the problem of continually changing, inhomogeneous, electric
fields. At the approach of two atoms in collision the electric field at
each of the respective atoms increases, reaches a maximum at closest
approach, and then diminishes as the two atoms recede.
Since the Stark effect for an inhomogeneous electric field has never
been worked out, Holtsmark1 and Debye2 made the simplifying assump¬
tion that there will be an average intermolecular field F to which, one
can assume, the ordinary Stark-effect formulas apply. The problem
of calculating average fields has been divided into three classes, (1)
the field due to charged atoms or molecules, i.e.} ions, (2) the field due
to dipole moments, and (3) the field due to quadrupole moments. Although
the theoretical treatment of each of these three possibilities is out of
place here, we shall write down the derived formulas for the three average
field strengths:
(For ions) F = aicn*. (21.36)
(For dipoles) F = a*tin. (21.37)
(For quadrupoles) F = ci^qn*. (21.38)

Here the a’s are constants, e is the ionic charge, m the dipole moment,
q the quadrupole moment, and n the number of corresponding particles
per cubic centimeter.
Holtsmark made the next simplifying assumption that a spectrum
line is spread out symmetrically into a continuous frequency band, the
1 Holtsmark, J., Phys. Zeits., 20, 162, 1919; 26, 73, 1924.
* Debye. P.. Phvs. Zeits.. 20. 160. 1919.
Sec. 21.7] THE BREADTH OF SPECTRU 1ft LINES 435

total breadth of which is given by the outermost components of the first-

order Stark effect (see Chap. XX). The total spread of the levels will
be
A(21.39)

In general there will be a statistical distribution of the fields and

a resultant distribution of bands of different breadths. The summing
up of these distributions, in the case of dipole fields for example, leads
to an intensity distribution having the same form as that for natural
breadths.

For dipole moments, I = 70^ ■ ^ '(b~/2)*' (^1.40)

where 8a is the half-intensity breadth,1

(for dipole) 8a = AmmxF = 4.54Anilut/in. (21.41)

The intensity contours for ions and for quadrupole moments have
almost the same form as those for the dipole moments (see Fig. 21.4).

Fia. 21.15.—Observed half-intensity breadth of hydrogen Ha. X6563, in angstroms.

(After Michelaon.) Calculated half-intensity breadth due to Stark effect from inter-
molecular quadrupole fields. (After Holtamark.)

The essential features of the above given equations are the expressions
for the half-intensity breadths of the observed spectrum lines. For the
latter two types of field these are:
For ion 8t = 3.25A^enK (21.42)
For quadrupole 8, = 5.52A^qn*, (21.43)

As one might well expect, all half-intensity breadths are functions of the
density n.
Of the many attempts made to check these formulas with observa¬
tion those made by Holtsmark are perhaps the best. Of these his
comparison of Eq. (21.43) with Michelson’s observations on the hydrogen
1 For a derivation of the equations given here the reader is referred to M. Bom,
,
“Opfifc,” 1933; also V. Weisskopf, Phys. Zeits84 1, 1933.
436 INTRODUCTION TO ATOMIC SPECTRA [Chap. XXI

line H« has been pointed out by several investigators as giving the best
general agreements. The quadrupole moment of the hydrogen molecule
is known from theory (by Debye) to be 3.2 X 10-26 gm* cm* sec.-1.
This value substituted in Eq. (21.43) gives the half-intensity breadth as
a function of n. Am is calculated from Eq. (20.2). In Fig. 21.15 both
the calculated and observed half-intensity breadths have been plotted
for comparison. Subtracting the Doppler breadth the agreement, even
here, is not very good.
In general the first-order Stark effect will account for the broadening
of many spectrum lines. The second-order, or quadratic, Stark effect
which only becomes appreciably large in strong fields must also produce
an observable effect.
One of the best confirmations of St ark-effect broadening may be
found in the chief series of the alkali atoms where historically the terms
sharp and diffuse series originated. Due to the penetration of the s and
p orbits involved in the sharp-series lines, both in the initial S states
and in the final P states show no first-order Stark effect. As a result
the lines, despite strong intermolecular electric fields, are sharp. For the
diffuse series, however, the d and / orbits are nearly hydrogen-like
and, being not far removed from the hydrogen-like /, g, h, . . . orbits,
show a first-order Stark effect. Under suitable excitation conditions
even the diffuse and fundamental series have been observed as sharp
lines.
Finkelnburg1 has studied the Balmer series under pressures of from
1 to 30 atms. of hydrogen. Using a high-voltage condensed-spark
discharge, between metal electrodes placed very close together, each
member of the series is observed as a very broad line. At a pressure
of about 2 atms. the H«, H^, and HT lines are symmetrically broadened,
without an appreciable shift of the maximum, to the extent that the half¬
intensity breadths are about 25, 100, and 200 A, respectively. At from
10 to 30 atms. the lines are so broad that they overlap one another and
form a continuous spectrum. These enormous breadths are attributed
by Finkelnburg to a Stark effect, the enormous fields being produced
by the very high ion density in the spark discharge. That the Stark
effect is chiefly responsible for the broadening is confirmed by (1) the
general contour of the lines, (2) the symmetrical broadening without
appreciable shift, and (3) the increased broadening with higher members
of the series.
It should be pointed out that asymmetry and pressure shifty discussed
in Sec. 21.6, may well be classified as a second-order Stark effect. At
close approach of a foreign atom the associated electric field causes a
polarization and a displacement of the energy levels of the atom in
question (see Fig. 21.11).
1 Finkelnburg, W., Zeite. f. Phys , 70, 375, 1931.
 APPENDIX
Table I.—Values of the Physical Constants
(After Birge)1
Velocity of light. c - (2.99796 ± 0.00004) X 1010 cm sec.”
Mass of the electron (spectro-
scopic) . m© (9.036 ± 0.010) X 10”*g
Mass of the proton. Mp (1.6608 ± 0.0017) X 10”Mg
Charge on the electron. e (4.770 ± 0.006) X 10”10 abs-e.s.u.
Planck’s constant. h (6.647 ± 0.008) X IQ'*7 erg. sec.
Rydberg constant (hydrogen1). Rhx (109677.769 ± 0.06) cm”1
Rydberg constant (hydrogen2). Rh* (109707.56 ± 0.05) cm'1
Rydberg constant (helium) Rue (109722.403 ± 0.05) cm*1
Rydberg constant (infinite.
mass). R (109737.42 ± 0.06) cm*1
Ratio Mp/mo (spectroscopic).... 1838 ± 1
Wave number per absolute volt. (8106 ± 3) cm"1 abs-volt”1
Wave-length per absolute volt. (12336 ± 5) X 10”* cm”1 abs-volt“
Energy of one abs-volt electron (1.5910 ± 0.0016) X 10”1* erg
Speed of one abs-volt electron. . (5.9346 ± 0.0017) X 107 cm sec.^
Fine-structure constant.. . . a = 2ire*/hc (7.283 ± 0.006) X 10"*
a* (5.305 ± 0.008) X 10”»
1/a 137.29 ±0.11
Unit angular momentum. h/2ir (1.0419 ± 0.0013) X 10”27 erg. sec.
Magnetic moment (one Bohr
magneton). mi = (0.9174 ±0.0013) X 10”*° erg gauss”
Magnetic moment (one nuclear
magneton). mi/1838 = 4.991 X 10~84 erg gauss”1
Zet>man displacement per gauss. - (4.674 ± 0.003) X 10"# cm-1 gains”
Bohr magneton Mi
Ratio = (0.8805 ± 0.0005) X 107 gauss”1 sec”
Bohr mechanical moment h/2r
1 Biroe, R. T.t Phya. Rev., Supplement, X, 1, 1929.

Table II.—Equivalent Electrons

(After Gibbs, Wilber, and While)1
p Electrons
(6) pi - »P
(15) p* - >5 iD »P
(20) p* - *D
(15) P* - >5 iD *P
(6) Ps - *P
(1) p* - 'S
a Electrons
(10) - *(D)
(45) d* - HSDG) HPF)
(120) d» - *a» HPDFGH) HPF)
(210) d4 - HSDG) HPF) HSDFGI) HPDFGH) HD)
(252) d‘ - *(D) HPDFGH) HPF)HSDFGI) HDG) HS)
(210) d« - i OSDG) HPF) HSDFGI) HPDFGH) HD)
(120) d* - HD) HPDFGH) HPF)
(45) d» - HSDG) HPF)
(10) d» - HD)
(1) d* _ i(fi)
1 Gibbs, R. C., D. T. Wilbur, and H. E. Whitu, Phya. Ret., 19, 790, 1927.
437
438 INTRODUCTION TO ATOMIC SPECTRA

Table II.—Equivalent Electrons.—(Continued)

/ Electrons
(14) /' - *(F)
(91) /* - KSDGI) *(PFff)
(364) P - HPDtFiGtHdKL) 4(SZ>F<?I)
(1001) P - 1 (StDiFGiHtl iKLtN) *(PiDiFiOiHJiKiLM) HSDFGI)
(2002) p - *{PiDiFiG*HiIiKiLiMtNO) *(SPtDiFtGiHJtKiLM) *(PFH)
(3003) p - * (SiPDtFtGiHilyKtLtMtNiQ) '(PtDiFtGrHtftKdsMtNO) '(SPDiF&HiItKL) HF)
(3432) P - UStPiDiFi&GuiUtltKyLtMiNtOQ) JiKiLiMN) •(PDFGHI) HS)
(3003) P - '(SiPDtFiGtHJiKiLuMiNtQ) '{PiDiFtGilhUKtLiMiNO) KSPDiF&iHJtKL) 1(F)
(2002) /* - '{PtDtFTGMiIiKJjiMiHO) i(SPiD»FiG*HiIiKtLM) *{PFH)
(1001) /•» - 1 (StDiFGtH tl»KLtN) * (PaDtFtGiHtl sK tLM) '(SDFGI)
(364) /" ■ KPDiFiGtHdKL) *(SDFGI)
(91) /«* ■ '(SDGI) *{PFII)
(14) /»» ■ *(F)
(1) /'« ■ '(.S’)
The subscripts to the right of each term type in the table for / electrons are not inner quantum
numbers but indicate the number of such terms. For example P — *(StPi, etc ,) indicates that there
are two *8 terms, five 9P terms, etc , arising from seven / electrons.
Table III.—Relative Intensity Tables for Multiplets, for Fine Structure in LS orjj-Couplino,
Table III.—{Continued)
APPENDIX 441
442 INTRODUCTION TO ATOMIC SPECTRA

Table IV.—Land£ g Factors for LS-couplino

Table V.—Land£ g Factors for Two Electrons in .^-coupling

(Values in parenthesis are excluded if electrons are equivalent)

7 - 0 1 2 3 4 7 - 0 1 2 3 4 5 6 7

SB ;i - 4. h - 4 0/0 (2) 7i - 4,7*1 - 2 1 83

i\ - 4,71 - 4 0/0 1 7*1 - S. 7* - I W U

■P Pf
i\ - 1. it - 2 3 7i - I, 7* - 8 4 113 111 IS

71 - i, it - I 11 9 m 8
ji - 1, it - 2 4 14
sd 7*i - I* 7*i - 2 (8) 1 (S)
ii - 1. 71 - 1 18 I
7i - i, h - 3 2 I 12 14
dd
7i - |. 71 - 2 (2) (1) (13) (34)
8f ji - 4. it - 1 1 2!
7*1 - I. it - K jH (1) 1 (X) X (1)
>i - *. y* - i 21
E ■
- 4. it - 4 BE (2) 7i - 2. 7i - 1 A HI III IIS
it - 4.7* - 1 1 I df 71 - 2, 71 - I §3 9 491 if
PP
71 - 3.7* - 4 (2) <l> 7*i - t, 7*1 - I 0/0 it 93 IS 18 II
7l - I. 7* - 1 0 0 (D X (4) 7*i - I. 71 - I IS 278 II 948 111! I
71 - 4. it - 2 1 44 7*1 " *. 7. - 1 0/0
(9) 9 (9) 9 (9)
71 - It it - | S3 V 2 » - «. 71 - I i i 4! 14 14 SI
Pd
7*1 * 2t it - 2 IX IS 11 7i - h it - 1 (!) (1) (II) (14) (24) (SI)
71 - 3. 7. - « 14 w 111 S h - i. 71 - I 0/0 (f)| # (») 1 (9) f
 INDEX
A Aufbauprinzip, Bohr's, 78, 79
Auroral line, 266
Abnormal series, 18, 386, 388 Autoionization, 386, 394
Absorption, Kirchhoff’s law of, 1 in barium, 397
Absorption edges, x-rays, 306, 307 in calcium, 397
Absorption spectra, x-rays, 306, 329 in copper, 395
Actinium, 83 in strontium, 397
Allen, C. W., 396, 427 Azimuthal quantum number, 43, 73
Allison, S., 313 defined, 43
Alternation law of multiplicities, 249, 251
Aluminum, 81, 82 B
anomalous series in, 386, 388
energy level diagram for, 115 Babcock, H. D., 291, 433
nuclear moment of, 372 Bacher, R. F., 364, 378, 382
Anderson, O. E., 370 Back, E., 73, 163, 169, 184, 223, 227, 354.
Angstrom, A. J., 2 357, 376, 380, 382
unit of wave-length, defined, 2 Back-Goudsmit effect, of hyperfine
Angular factor, probability density, 60, structure, 376
64 Biicklin, E., 300
Anomalous series, 18, 386, 388 Ballard, S. S., 37
Anomalous Zeeman effect, 152 Balmer, 3
Antimony, 83, 259, 307 law, 4
nuclear moment of, 372 series in hydrogen, 3, 14, 30, 31, 33
Aphelion, of electron orbit, 44 fine structure of, 132, 135, 137
Arc spectra, photographs, 174, 250, 349 Paschen-Back effect of, 168
Argon, 81, 82, 274 photographs of, 5, 31
fine structure of, 277 fine structure, 132, 138
Arsenic, 83, 259 Stark effect, 410
nuclear moment of, 372 Barium, 83, 176
Asymmetry of spectrum lines, 418, 431, ionized, energy level diagram of, 98
432, 433 nuclear moment of, 372
Atom model, for complex spectra, 251 photograph of triplet in, 174
for doublets, 118 Barkla, C. G., 301
for hyperfine structure, 354, 365 Bergmann, 6, 9
Back-Goudsmit effect, 376, 380 series, discovery of, 6
Faschen-Back effect, 376, 380 Beryllium, 80, 81, 82, 176
for//-coupling, 191, 224 energy level diagram of, 179
for //-coupling, 254 ionized, energy level diagram of, 98
for L5-coupling, 191, 224, 254 Bethe, H., 313
for Paschen-Back effect, 163 Bevan, P. V., 17
complete, 231 Bichowsky, F. R., 121
for Stark effect, 405, 407 Birge, R. T., 17
for Zeeman effect, complex atoms, 286 series formulas, 17
hyperfine structure, 373 Bismuth, 84, 259
one-electron systems, 154 Back-Goudsmit effect, 381
443
444 INTRODUCTION TO ATOMIC SPECTRA

Bismuth, hyperfine structure, 381 Cadmium, triplet series in, 20

nuclear moment of, 372, 380, 381 Caesium, 83
Bohr, N., 21, 311 doublet fine structure, 116
circular orbits, 28, 30 energy level diagram, 87
correspondence theorem, 38, 41 nuclear moment of, 372
magneton, 52, 53 Calcium, 82, 176, 250, 332, 340
theory, assumptions in, 25, 28 abnormal series in, 20
Bohr-Stoner scheme of the elements, 78, anomalous series in, 20, 386, 388
86 energy level diagram of, 172, 182
Bom, M., 105, 418, 435 ionized, energy level diagram of, 97, 98
Boron, 80, 81, 82 nuclear moment of, 372
doublet fine structure of,. 115 photographs, of multiplets in, 180, 250
energy level diagram for, 115 of singlet in, 250
Bowen, I. S., 226, 332, 337 of triplets in, 174, 250
Brackett, series in hydrogen, 14, 30, 33 series relations, 10
Branching rule, 202 term values, 173
for jy-coupling, 253, 275, 276 Carbon, 81, 82, 266
for L6-coupling, 253, 261, 264, 268, 275 energy level diagram in, 266
for multiplicities, 251 fine structure, 200, 266
for terms, 252 Casimir, H. B. G., 358
Bragg, W. H., 299 Catalan, M. A., 184, 248
Bragg, W. L., 299 Centroid diagrams, 350
Breadth of spectrum lines, 418 Cerium, 83
due to collision damping, 418, 427 Charge density, electron, 59
due to Doppler effect, 419, 420, 426 Chlorine, 81, 82
due to external effects, 418, 428 nuclear moment of, 372
due to natural breadths of levels, 418, Chromium, 82, 250, 346
421, 422, 426 photographs of multiplets, 250
due to pressure shift, 418, 431, 433 of Zeeman patterns, 288, 291
due to Stark effect, 418, 434 Zeeman effect of, 152, 288, 291
Breit, G., 166, 212, 235, 358, 362, 371, 383 Cobalt, 82, 250
scheme, for correlation of Zeeman nuclear moment of, 372
levels, 166, 235 photographs of multiplets in, 250
for derivation of spectral terms, 235, Collision damping, 427
237, 239 formula for, 428
for equivalent electrons, jy-coupling, Collision diameter, of atoms, 430
239 Collisions, elastic, 92
LS-coupling, 237 of the first kind, 92
Brickwedde, F. G., 37 inelastic, 92
Broad lines, in copper, 396 of the second kind, 92
Bromine, 83 Columbium, 83
nuclear moment of, 372 nuclear moment of, 372
Burger, H. C., 119, 160, 220 Combination lines, 175
Bums, K., 353, 364 Combination principle, 15
Combination series, 15
C Complete Pasehen-Back effect, 231
patterns for, 233, 234
Cadmium, 176, 306 selection rules for, 235
complete electron configuration, 308 Complex spectra, 248
electron shells, 309 g factor for, 286
energy level diagram, 179 hyperfine structure in, 364, 369
x-ray energy level diagram, 310 irregular-doublet law for, 345, 347
nuclear moment of. 372 Land6 interval rule for, 255
 INDEX 445

Complex spectra, Paschen-Back effect, Dirac, P. A. M., 124, 139, 206, 312
290 electron, 139
Stark effect in, 412 angular distribution for, 140
Zeeman effect, 286 eigenfunctions for, 139
intensity rules, 289 for hydrogen, 139
photographs of, 288, 291 photographs representing, 146
selection rules, 289 probability-density distribution for,
Compton, A. H., 301 140, 143, 146
Compton, K. T., 93 radial distribution for, 144
Condon, E. U., 72, 211, 217, 269, 280, 390 Dispersion electrons, table of, 424
Continuous spectrum, in hydrogen, 33 Displacement law, complex spectra, 249
Continuum, 33, 392 Doan, R. L., 301
Copper, 82, 250 Dolejsek, V., 306
anomalous series in, 386 Doppler effect, 417, 418, 424
energy level diagram of, 395 formula for, 418, 419
nuclear moment of, 372 photographs of, 418
perturbations in, 391 Dorgelo, H. B., 119, 120, 160
Core, atom, 184 Double-electron excitation, 179
Correspondence theorem, Bohr, 38, 41 Double-electron transitions, optical
Coster, D., 311, 312, 324 spectra, 269
Coupling schemes, for two electrons, 189 x-ray spectra, 327
Crawford, M. F., 363, 368 Doublets, fine structure, 11, 114
Critical potentials, 92, 96 formulas for, 128
of the alkali metals, 95 intensity rules for, 118
selection rules for, 117
D separations for, 128
vector model for, 118
Damping, collision, 427 Druyvesteyn, M. J., 324, 325, 326
Darwin, C. G., 124, 139, 166 Dysprosium, 83
Darwin, K, 227, 241
Dauvillier, A., 331 E
Davis, Bergen, 301
De Broglie, L., 306, 331 Eckart, C., 383
corpuscular wave equation, 54 Edlen, B., 32
Debye, P., 434 Effective nuclear charge, 180
Derivation of spectral terms, from equiv¬ Effective quantum number, 89
alent electrons, 187, 292, 293, 295 Eigenfunctions, 59, 139
^/'-coupling, 196 defined, 57
LS-coupling, 187 Einstein, A., 23, 312
shorthand method, for equivalent Electron orbits, 25
electrons, 296 circular, 25, 29, 30, 34
two arbitrary electrons, 187 elliptic, 42, 47
two equivalent electrons, .^-coupling, in hydrogen, 25, 29, 30, 34, 42, 47, 72
238 in lithium, 103, 111
LiS-coupling, 236 in potassium, 101
Designation of terms, ^-coupling, 196 precession of, 104, 106, 108
L&-coupling, 187 quantum-mechanical picture of, 63,
Dewar, J., 4 68, 71, 100, 101, 103
Diagonal lines of multiplets, 206 rubidium, 100
Diffuse series, discovery of, 6 series of, 113
fine structure in, 12, 20 sodium, 101, 103
photographs of, 174 Electron spin-orbit interaction, 124, 128,
Dinkelacher, O., 433 129
446 INTRODUCTION TO ATOMIC SPECTRA

Electron shells, 309 Even terms, defined, 187, 204

subshells, 80, 85 Excitation potentials, 92
Elliptic orbits in hydrogen, 42, 47
Energy level diagrams, 32, 260 F
of the alkali metals, 86
beryllium, 179 Fermi, E., 102, 120, 358
cadmium, 179 Filippov, A., 424
carbon, 266 Fine structure, defined, 30
copper, 395 discovery of, 11, 12
germanium, neutral and ionized, 199 doublets, 114
helium, 209 hydrogen, 132, 137
ionized, 33 photographs of, 132, 138
hydrogen, 33 separations, 128
ionized atoms, 98 Sommerfeld’s formula for, 134
lithium, 77 triplets, 173
ionized, 209 x-rays, 314
magnesium, 179 Finkelnburg, W., 436
manganese, 271 First-order Stark effect, 402
mercury, 179 Fluorine, 81, 82
oxygen, 266 nuclear moment of, 372
perturbations in, 392 Foote, P. D., 94
potassium-like atoms, 336 Forbidden transitions, 267
scandium, 263 Foster, J. S., 411, 414, 417
sodium, 77 Foucault, L., 1
titanium, 268 Fowler, A., 11, 13, 21, 31, 331
x-rays, 308, 310 Franck, J., 92
xenon, 275 Fraunhofer, J., 1
zinc, 179 solar lines, 2, 3
Energy levels, hyperfine structure, 357, Friedrich, W., 229
367 Frisch, R., 361, 373
Back-Goudsmit effect, 379 Fuchtbauer, C., 433
Zeeman effect, 375 Fundamental series, discovery of, 6
perturbations of, 387 Furnace spectra, photographs, 349
Stark effect, in helium, 414
in hydrogen, 407 G
strong field, 409
weak field, 406 g factors, formula for, 157
Engwicht, H., 208 in complex spectra, 216, 287
Enhanced lines, 21 in jy-coupling, 217, 240, 242, 287
Epstein, P. S., 401, 408 in L&-coupling, 216, 240, 241, 287
Equivalent electrons, 236, 254, 293 one electron, 157
defined, 188 nuclear, 359, 363, 371
T permanence and T sum rules for, 245 strong field, 241, 242
terms arising from, 187, 196, 236, 238, tables of, 158, 442
293, 295, 296 very strong field, 241, 242
tables of, 254, 437 weak field, 240, 241, 242
Erbium, 83 Zeeman effect, 157, 216, 217, 286, 287
Ericson, A., 32 g permanence rule, 239
Estermann, 1., 361, 373 ^-coupling, 240
Europium, 83 LS-coupling, 240
nuclear moment of, 372 g sum rule, 222, 241
Evans, E. J., 31 jj-coupling, 222, 242
 INDEX 447

g sum rule, LiS-coupling, 222, 241 Half-intensity breadth of spectrum lines,

Pauli’s, 222 due to pressure, 418, 431, 433
Gadolinium, 83 due to Stark effect, 418, 434
Gallium, 83 Hargreaves, J., 358, 422
doublet fine structure in, 115 Harrison, G. R., 208, 283
energy level diagram for, 115 Hartley, W. N., 4
nuclear moment of, 372 law of doublet separations, 4
T factors, 190 Hartree, D. R., 101, 112, 140
for equivalent electrons, 246 Haas, A., 55
in .y-coupling, 246 Heisenberg, W., 105, 209, 212
in L5-coupling, 191, 246 matrix mechanics, 54, 135
in strong fields, 243, 244, 246 Helium, 80, 82
in very strong fields, 244, 246 energy level diagram, 209
in weak fields, 243, 244, 246 fine structure, 212
formulas for, 127, 191, 192, 196, 197 ionized atom, 29, 31, 32
for one electron, 127 energy level diagram, 33
T permanence rule, 242 fine structure of, 139
for equivalent electrons, 245 term values for, 134
for jj-couplmg, 243 -like atoms, 209
for LiS-coupling, 243 model of atom, 211
I1 sum rule, 244, 245 Stark effect of, 414
Gartlein, C. W., germanium spectrum wave equations for, 210
and energy level diagram, 199 Hertz, G., 92, 319
Gaunt, A., 102 Hicks, W. M., 16
Gobauer, R., 411 Hill, E. L., 356
Gerlach, W., 51 Hirsch, F. R., 305, 306, 324, 325
Germanium, 83 Hiyama, S., 411
energy level diagram, by Gartlein, 199 Holmium, 83
fine structure of, 200 nuclear moment of, 372
Gibbs, R. C., 267, 278, 296, 332, 335, Holtsmark, J., 434
339, 343 Honl, H., 205
Gicseler, H., 248, 267 Houston, W. V., 278
Goetzc, R., 11 Hughes, D. S., 383
Gold, 83 Hund, F., 194, 259
nuclear moment of, 372 rule, for complex spectra, 259
Gordon, W., 124, 139, 147 for L»S-coupling, 194
Goudsmit, S. A., 121, 123, 231, 244, 282, Hydrogen, 80, 132
283, 322, 354, 357, 358, 364, 371, Bohr theory of, 25, 31
373, 376, 378, 380, 382 continuous spectrum of, 33
T sum rule, 244, 245 fine structure of, 132
Grace, N. S., 137, 353, 382 photograph of, 132, 138
Green, J. B., 223, 380 schematic diagram for, 135. 137
Grotrian, W., 267, 414 Sommerfcld’s formula for, 134
isotope (deuterium), fine structure of,
H 132
Lyman series of, 37
Haga, H., 299 term values for, 134
Hafnium, 83 nuclear moment of, 372
Half-intensity breadth of spectrum lines, Paschen-Back effect, 168
defined, 422 Stark effect, 401
due to collision damping, 418, 427 strong field, 407
due to Doppler effect, 419, 420, 426 weak field, 406
aue to natural breadth, 418, 421, 426 term values for, 134
448 INTRODUCTION TO ATOMIC SPECTRA

Hylleraas, £. A., 210 Interaction energy, for complete Paschen-

Hyperbolic orbits, 33 Back effect, 232
Hyperfine structure, 352 for electron spin and orbit, 125
Back-Goudsmit effect of, 376, 379 for ^-coupling, 195
patterns for, 379 for LiSf-coupling, 191
selection rules for, 378 for magnetic field, one-electron sys¬
in complex spectra, 364, 369 tems, 157
coupling coefficients for, 367 two-electron systems. 217
energy levels, 357 for nucleus, with penetrating electron,
inverted, 363 361
isotope shift, 382, 383 with valence electron, 358
Land6 interval rule for, 354 for Paschen-Back effect, 225, 229, 232
in lanthanum, 370 for spin-orbit interaction, 125
in mercury, 384 for two electrons, 190, 212
normal, 363 Intercombination lines, 175, 269
notation of, ^'-coupling, 356 scries of, 15
LS-coupling, 356 Intermediate coupling, 278, 281, 283
orbit model for, 354 Zeeman effect, 223, 224, 280
Paschen-Back effect in, 376 Interval rule, Land6's, 194
perturbations, 398 Inverted alkali doublets, explanation of,
photographs of, 352, 353, 381 124, 393
quantum numbers for, 355 Inverted fine structure, 256
selection rules for, 356 Inverted terms, 124, 256
Back-Goudsmit effect, 378 Iodine, 307
splitting factors, 367 nuclear moment of, 372
term perturbations, 398 Ionization potentials, 90
vector model, 354, 355, 365, 373 Ionized atoms, 96
Zeeman effect, 373 Iridium, 83
energy levels for, 375 Iron, 82, 250
selection rules for, 375 Irregular doublet law, for complex spec¬
vector model for, 373 tra, 342, 343, 345, 347, 348
for optical doublets, 337
I for single atom, 348
for x-rays, 319
Illinium, 83 Ishida, Y., 411
Indium, 83 Isoelectronic sequences, complex spectra,
doublet fine structure, 115 340, 345
energy level diagram for, 115 defined, 331
nuclear moment of, 372 energy level diagram for, 335
Inglis, D. R., 280, 282, 378 irregular doublet law, 337
Inner quantum number, 117 Moseley diagram for, 341
Intensities of spectrum lines, in complex Moseley law for, 333
spectra, 282 one-electron systems, 332
formulas for, 205, 206 quantum defect for, 337
rules for doublets, 118 quartet multiplets for, 345
tables for, 439-441 quintet multiplets for, 346
x-rays, 301 regular doublet law for, 338
Zeeman effect, 220 screening constants for, 334, 339
complex spectra, 289 triplet multiplets for, 340
Intensity sum rule, 119, 205 two-electron systems of, 340
if-coupling, 207 Isotope shift, 353, 383
LS-coupling, 206 Isotope structure, 353, 383
 INDEX 449

J Krypton, 83, 274

fine structure of, 277
Jablonski, A., 432 nuclear moment of, 372
.y-coupling, 190, 195
g factor, 217 L
g permanence rule, 240
g sum rule, 222, 242 L -radiation defined, 301
r factor, 190, 191 L-series, x-rays, Moseley law for, 305
T permanence rule, 243 photographs of lines, 303
r sum rule, 246 wave-lengths, 304
intensities, 207 Ladenburg, R., 413, 424, 427
interaction energy for, 196 Laguerre polynomials, derivatives of, 67
magnetic moment of atom for, 217 LandC*, A., 73, 184, 331
notation for, 196 g factor, 154, 156, 286
Paschen-Back effect for, 224, 228, 233 complex spectra, 286
selection rules for, 203 formula for, 157, 216
sum rules for intensities in, 207 strong field, 240, 241, 242
vector model for, 191, 197 tables of, 158, 442
Zeeman effect, 217, 224 very strong field, 241, 242
//’-coupling, 275, 350 weak field, 240, 241, 242
isoelectronic sequence, 350 Zeeman effect, 157, 216, 286
in xenon, 275 r permanence rule, 242
Zeeman effect in complex spectra, 286 interval rule, 194, 246
Jones, G., 383, 398 in complex spectra, 255
Jonsson, A., 313 graphical representation, 195, 255
Joos, G., 432 in hyperfine structure, 354
Jordan, P., 135 vector model for, 185, 195, 255
Lang, R. J., 267, 350
K Langer, R. M., 327, 388
Lanthanum, 83
K-radiation, defined, 301 hyperfine structure of, 370
K-series x-ray lines, 302 nuclear moment of, 372
Moseley law, 302 Laporte, O., 280, 350
Kaplan, J., 267 Larmor precession, 52, 125
Kaufman, S., 312, 324 in the Zeeman effect, 157
Kellogg, J. M. B., 361, 373 Larmour theorem, 51, 157
Kellstrom, G., 300 Larsson, A., 300
Keyston, J. E., 383 Laue, M., 299
Kiess, C. C., 267 pattern, 300
Kiess, H. K., 267 Lead, 83
Kievit, B., 329 fine structure of, 200
King, A. S., 174, 349 nuclear moment of, 372
Kirchhofif, G., 1 Zeeman effect of, 223
Legendre polynomials, 61
Kiuti, M., 411
Lenard, 92
Knipping, P., 299
Lewis, G. N., 132, 137
Koch, J., 414
Lindsey, G., 329
Kokubu, N., 410
Line breadths, causes for, 418
Kopfermann, H., 427
Linnik, W., 301
Kossel, W., 249, 307, 329 Lithium, 80, 81, 82
Kramers, H. A., 404 doublet fine structure, 116
Kronig, R. de L., 205, 208, 329 doubly ionized, 29
Kruger, P. G., 32 term values for, 134
450 INTRODUCTION TO ATOMIC SPECTRA

Lithium, effective quantum number for, Magnetic energy, for Zeeman effect, 157,
89 217
energy level diagram for, 77, 87 Magnetic moment, of atoms, with one
-like atoms, 339 valence electron, 155
nuclear moments of, 372 with two valence electrons, 215
probability-density distributions for, ^'-coupling, 217
101, 110, 111 L*ST-coupling, 216
singly ionized, energy level diagram of nucleus, 359
for, 209 of orbital electron, 52, 155
term values for, 88 of proton, 361
Liveing, G., 4 of spinning electron, 155
^-coupling, 185 Magnetic quantum number, 158, 218
12-coupling, 252 complex spectra, 292
London, F., 424 Magnusson, T., 301
Lorentz, H. A., 152, 428 Manganese, 82, 250, 270
classical treatment of Zeeman effect, energy level diagram for, 271
152 fine structure in, 271, 273
unit, for Zeeman effect, 158 multiplets in, 272
Loring, R. A., 223 nuclear moment of, 372
Lo Surdo, A., 401 Margenau, IL, 432
JLS-coupling, 186, 190 Mark, H., 410
g factors for, 157, 216, 240, 241, 287 Masaki, O., 393
g permanence rule, 239, 240 Masurium, 83
g sum rule for, 241 Mechanical moment, defined, 52
T factors, 190, 191, 246 Meggers, W. F., 265, 353, 364
T permanence rule for, 243 Meissner, K. W., 393
r sum rule for, 244, 245 Mercury, 83, 176
energy level diagram, 179
intensities for, 206
hyperfine structure, 384
interaction energy for, 191
term perturbations in, 399
Pasehen-Back effect for, 224
isotope structure in, 384
selection rules for, 203
nuclear moments of, 372
sum rules, for intensities, 206 Metastable state, 96
to .//-coupling, 279, 281, 283, 284 Meyer, H. T., 313
vector model for, 191 Michelson, A., 132, 428
Zeeman effect for, 286 Minkowski, R., 401, 426
Lutecium, 83 Modified Moseley diagram, 335
nuclear moment of, 372 for isoelectronic sequence, 341, 344, 348
Lyman, T., series, in helium, He II, 31, 32 Moller, F. L., 93
in hydrogen, 14, 30, 33 Molybdenum, 83
More, K., 353, 382
M Morse, P. M., 72, 211, 217
Moseley, H. G. J., 302
flf-series x-ray lines, photographs of, 305 diagram, isoelectronic sequence, 333,
Mack, J. E., 350 341, 344, 348
modified, 335
McLay, A. B., 363, 368
for a single atom, 348
McLennan, J. C., 363, 368
x-rays, 311, 321
MacMillan, 353
A-series, 302
Magnesium, 81, 82, 176
L-scries, 305
energy level diagram for, 179 satellites, 325
ionized, energy level diagram for, 98 Multiplets, in calcium, 181
photographs of triplets in, 174 defined, 179
 INDEX 451
Multiplets, in manganese, 272, 273 Orthohelium, 209
photographs of, 260, 340, 346, 346 Osgood, T. H., 301
in titanium, 269 Osmium, 83
x-ray structure, 323 Oxygen, 81, 82, 266
Murphy, G. M., 37
P
N
Palladium, 83
Natural breadths, 420 Parhelium, 209
of energy levels, 423 Paschen, F., 11, 13, 18, 31, 132, 139, 153,
formulas for, 422 163, 169, 227, 267, 331
observed in copper, 427 Paschen-Back effect, complete, 231-235
of spectrum lines, 425, 426, 427 complex spectra, 290
Nebular lines, 266 hydrogen, 168
Neodymium, 83 hyperfine structure, 376
Neon, 81, 82, 274 interaction energy for, 225
fine structure of, 277 ^-coupling, 223, 228
series in, 18 LS-coupling, 223, 224
Nickel, 82, 250 magnetic energy for, 164
Nitrogen, 81, 82, 259 normal triplet, 168
energy level diagram for, 260 one electron, 162
fine structure in, 260 pattern for jj-coupling, 227
multiplets in, 262 for LS-coupling, 227
nuclear moment of, 372 photograph of, 169
Nonpenetrating orbits, 100, 103, 105 principal series doublet, 165
Normal doublets, 123 selection rules for, 168, 227
Normal states, of atoms, 277 vector model for, 163
Normal Zeeman triplet, 151, 152, 219 Pauli, W., 124, 222, 231, 239, 353
Notation, electron orbit, 48 exclusion principle, 188, 236, 293
for hyperfine structure terms, g permanence rule, 239
in jj-coupling, 356 g sum rule, 222, 241
in LN’-coupling, 356 Pauling, L., 102, 106, 109, 231, 318, 322,
jy-coupling, 196 330, 358
JW?-coupling, 185 Penetrating electron orbits, 100, 103, 104,
odd and even terms, 204 178, 339
series, 10, 13 Pentad, of multiplets, 189
two-electron systems, 185 Perihelion, of electron orbit, 44
x-rays, 308 Periodic table, 79, 81, 82, 83, 86
Nuclear g factors, 361, 363 Perturbations, conditions for, 390
calculation of, 371 in copper, 391
table of, 372 of energy levels, 387
Nuclear magneton defined, 359 nature of, 390
Nuclear moments, table of, 372 of terms, 386
Nuclear spin, 353, 361, 363, 365 Pfund series, in hydrogen, 14, 30, 33
Phase integral, 24, 27, 43, 49, 50
O Phase space diagram, 24, 27, 45
Phosphorus, 81, 82, 259
Odd terms, defined, 187, 204 nuclear moment of, 372
Optical cross-section of atoms, 430 Photographs, of absorption spectra, 17
Optical doublets, 332 of anomalous Zeeman effect, 152, 291
irregular doublet law, 337 of Back-Goudsmit effect, 381
regular doublet law, 338 of Balmer series of hydrogen, 531
Omstein, L. &, 119, 160, 220 of complex spectra, 250
 452 INTRODUCTION TO ATOMIC SPECTRA

Photographs, of Doppler effect, 419 Praseodymium, 83

of flash spectrum of sun, 5 nuclear moment of, 372
of hydrogen fine structure, 132, 137 Precession, orbital, 104, 133
of hyperfine structure, 352, 353 Pressure broadening of spectrum lines,
of interference of x-rays, 301 431-433
of isoelectronic sequences, 332, 340, Pressure shift of spectrum lines, 431-433
345, 346 Preston’s law, 151
of isotope structure, 353 Principal series, for alkali metals, 116
of Laue pattern, 300 anomalous, 391
of Lyman series of hydrogen, 32, 37 discovery, 6
of manganese multiplets, 273 Paschen-Back effect, 167
of multiplets, 250 triplet, 175
of octet multiplet, 250 Zeeman effect of, 160
of Paschen-Back effect, 169 Probability-density distributions for
of principal doublets, 332 electrons, Dirac theory of, 140
of principal series of sodium, 17 in helium, 212
of probability-density distribution for in hydrogen, 59, 61, 140
hydrogen, 71, 146 Schrodinger wave equation for, 59, 61
of quartet multiplets, 250 Prokofjew, W., 424
of quintet multiplets, 250 Proton, magnetic moment of, 361
of septet multiplets, 250
of sextet multiplets, 250, 273 Q
of solar spectrum, 3
of Stark effect, 411, 417 Quantum, defect, 89, 108, 109
of stellar spectrum of hydrogen, 5 in anomalous series, 388
of triplet multiplets, 174, 180, 250 for calcium, 180
of vanadium spectrum, 349 of energy states, 23
of x-rays, 300, 301, 303, 305 in isoelectronic sequences, 337
absorption edges, 306, 307 for rubidium, 112
diffraction, 300, 301 Quantum mechanics, 54
interference, 301 Quantum numbers, 26
L-series lines, 303 azimuthal, 43
M-series lines, 305 inner, 117
satellites, 303, 305 magnetic, 52, 158
of Zeeman effect, 152, 291 radial, 43
Pickering series, 20, 35 Stark effect, 402
Planck, constant h, 23, 24 total, 26, 45
harmonic oscillator, 23 Quantum theory, introduction to, 23
quantum theory, 23, 24 Quartets, Moseley diagram for, 344
Platinum, 83 photographs of, 345
Polarization, of atom core, 105, 106 Quintets, Moseley diagram for, 344
of spectrum lines, Zeeman effect, 151, photographs of, 346
160, 219
Polonium, 83 R
Potassium, 82, 250, 332
doublet fine structure, 116 Rabi, I. I., 361, 373
energy level diagram of, 87 Racah, G., 362, 383
-like atoms, 336 Radial factor, probability density, 64
probability-density distribution for, Radial quantum number, 43
101 Radium, 83
series in, 4 Radon, 83
Potential curves, for atoms in collision, Ramb, R., 393
432 Ramberg, E. G., 327
 INDEX 453

Ramsauer, G., 413 Rydberg, series formula, 7

Reduced mass of the electron, 34 Rydberg-Schuster law, 8, 10, 171
Reduced terms, x-rays, 316, 321
Regular doublet law, 318, 338 S
Regular doublets, x-rays, 316
Relative intensities, 205 Sadler, C. A., 301
formulas for, 206 Samarium, 83
for jf/’-coupling, 207, 439-441 Sambursky, S., 119, 120
for L/8-coupling, 206, 439-441 Satellites, in optical spectra, 11
for multiplets, 207, 439-441 photographs of, 174
for related multiplets, 208 in x-ray spectra, 324
tables of, 439-441 Moseley diagram for, 325
Relativity, doublet correction for, 133, photographs of, 303, 305
135, 136 Saunders, F. A., 179, 185, 397
x-rays, 315, 316 Scandium, 82, 250, 262, 332, 340, 345
orbital precession due to, 133 energy level diagram, 263
Residual, Rydberg denominator, 19 ionized atom, 188
Resonance lines, definition, 6 energy level diagram of, 263
Rhenium, 83 multiplets in, 262
nuclear moment of, 372 Schlapp, R., 401, 406
Rhodium, 83, 248 Schrodmger, E., 106, 138, 401
nuclear moment of, 372 wave equation, 54
Richtmycr, F. K., 303, 324, 325 solution of, 56
Richtmyer, R. D., 324 Schuler, H., 383, 398
Ritschl, R., 357 Schuster, A., 3, 9
Ritz combination principle, 15 Rydberg-Schuster law, 8, 10, 171
Ritz series formula, 16, 18 Schutz, W., 430
Rdntgen, 299 Schwartzchild, K., 401, 408
rays, discovery of, 299 Screening constants, x-rays, 317, 318
Rosenthal, J. E., 383 Screening doublets, x-rays, 316, 319
Rowland, H. A., map of solar spectrum, 3 Screening effect, internal, 314
Ruark, A. E., 70, 133, 403 external, 314
Rubidium, 83 x-rays, 314
doublet fine structure, 116 Second-order Stark effect, 402
electron-density distribution for, 100, Selection principle, 90
101 Selection rules, for Back-Goudsmit effect,
energy level diagram for, 87 378
nuclear moment of, 372 for complete Paschen-Back effect, 235
Rumpf, atom, 184 for doublets, 117
Runge, C., 153 for hyperfine structure, 356
law, 9, 10 Zeeman effect of, 375
denominator, for Zeeman effect, 153 for jj-eoupling, 203
Running terms, defined, 13 for L*S-coupIing, 203
Russell, H. N., 11,179,185, 205, 235, 265, for Paschen-Back effect, 168
^/-coupling, 230
267, 296, 353, 387, 397
L&-coupling, 227
Russell-Saunders coupling, 186
for Stark effect, complex spectra, 417
Ruthenium, 83
strong field, 409
Rutherford, L. M., 25
weak field, 407
Rydberg, J. R., 5, 9
for triplets, 176
constant, 7, 28 for x-rays, 301, 324
variation with mass, 35 for Zeeman effect, 219
denominator, 18, 89 complex spectra, 289
454 INTRODUCTION TO ATOMIC SPECTRA

Selenium, 83 Spedding, F. H., 132, 137

Self-consistent field, 100 Spherical symmetry, 75
Self-reversal of spectrum lines, photo¬ electron distribution for, 142
graphs, 250 Spinning electron, 114,121,124, 127,129,
Series, anomalous, 386 139
formula for, 89 magnetic moment of, 124
for ionized atoms, 98 Spin-relativity doublets, x-rays, 315, 316
limits, germanium, 199 Spin-spin coupling, 186
negative, 181 ^-coupling, 252
in two-electron systems, 200 Stark effect, 401
notation, 10 complex spectra, 412
perturbations, 386 energy levels for, in helium, 414
Shane, C. D., 137 in strong field, 409
Sharp series, discovery of, 6 first-order, defined, 402
Sharp triplets, photographs of, 174 forbidden lines, 412
Shells, electron, 79, 85 helium, 414
Shenstone, A. G., 11, 394, 395 hydrogen, 407
Shortley, G. H., 280 line breadths, due to, 418, 434
Siegbahn, M., 299, 301, 306, 329, 330 orbital model for, 403
Silicon, 81, 82 photographs of, in helium, 417
fine structure, 200 in hydrogen, 411
Silver, 83, 306, 307 second-order, defined, 402
Singlets, irregular doublet law for, 342 selection rules for, 407, 417
series of, 171 vector model for, 407
Slater, J. C., 70, 110, 209, 281, 390 weak field for, 404
Snyder, C. P., 248 energy levels for hydrogen, 406
Sodium, 81, 82 Stark, J., 401, 414
critical potentials, 94 Stenstrom, W., 324
dispersion electrons for, 424 Stern, O., 51, 361, 373
doublet fine structure of, 116 Stoner, E. C., 80
effective quantum numbers for, 90 Stoney, G. J., 3
energy level diagram for, 77, 87 Strong-field Zeeman effect, 162
nonpenetrating orbits for, 103 Strontium, 83, 176
nuclear moment of, 372 ionized, energy level diagram, 98
penetrating orbits for, 103 nuclear moment of, 372
principal series of, 117 photographs of triplets, 174
probability-density distribution for, Subordinate series, 6
101 Subshells of electrons, 80, 85
series relations in, 10 Sulphur, 81
discovery of, 4 Sum rules, for g factors, 222, 241, 242
term values of, 88 for T factors, 244, 245
Zeeman effect of, 152 for line intensities, 205-207
Sommerfeld, A., 42, 106, 147, 205, 249, for Zeeman effect, 220, 222, 241, 242
315, 321, 322, 404
elliptic orbits, 42, 132, 138 T
Spark lines, defined, 21
photographs of, 349 Takamine, T., 410
Space quantization, 48 Tantalum, 83
Paschen-Back effect, 163 hyperfine structure, 353
Zeeman effect, 159 nuclear moment of, 372
Spectral terms, derivation of, 187, 196, Tate, J. T., 94
236, 238, 296 Tellurium, 83, 307
notation, 185, 196, 204, 308, 356 Temarium, 83
 INDEX 455
Temperature, correction for intensities, Two-electvdn systems, T factors for, 1Q1
205 T sum rule for, 244
effect on spectrum lines, 419 Hund’s rule for, 194
Terbium, 83 intensity rules for, 206
nuclear moment of, 372 jy-coupling for, 191, 196, 200
Terms, defined, 13, 78 Landd interval rule for, 194
derivation of, 186, 187, 196, 292, 296 tf-coupling for, 185
perturbations of, 387, 391, 392 LS-coupling for, 186, 191, 200
hyperfinc structure, 398 odd terms for, 187, 204
quantum defect for, 89, 108, 112, 180, Paschen-Back effect for, 223
337, 388 Pauli exclusion principle for, 188, 236
Thallium, 83 quantum mechanical model of, 211
doublet fine structure in, 115 quantum numbers for, 184, 185, 187,
energy level diagram for, 115 218
hyperfine structure in, 367 Russel 1-Saunders coupling for, 186
observed, 362 spin-spin coupling for, 186
nuclear moment of, 372 terms arising from, 187, 188, 254
Thibaud, J., 301
Thiele, E., 424 U
Thomas, L. II., 102, 126
Thorium, 83 Uhlenbeck, G. E., 121
Tin, 83, 307 Unquantized states, 33
fine structure of, 198, 200 Unsold, A., 75, 138
nuclear moment of, 372 Uranium, 83
Titanium, 82, 250, 267, 332, 340, 345, 346 Urey, H. C., 37, 70, 121, 133, 403
Total quantum number, defined, 26, 45
Transitions, between energy levels, 33, 90 V
for fine structure, 77
between orbits, 29 Vanadium, 82, 250, 332, 340, 345, 346
Triad, great calcium triad, 180, 181, 201 arc spectrum of, 349
multiplet, 180 furnace spectrum of, 349
Triplet fine structure, 11, 173, 175 nuclear moment of, 372
defined, 11 photographs of spectrum, 349
irregular doublet law for, 342, 343 spark spectrum of, 349
Moseley diagram for, 344 Van der Tuuk, J. H., 324
multiplots, 189, 250 Van ITrk, A. T., 106
photographs of, 174, 250, 340 Van Vleck, J. H., 106, 160
selection rules for, 176 Vector model, for complete Paschen-
series, 171 Back effect, 231
Trumpy, II., 424 for complex spectra, 251
Tungsten, 83 for doublets, 118
isotope structure, 353, 382 for hyperfine structure, 354, 355, 365
nuclear moments of, 372 Back-Goudsmit effect, 376, 380
Turner, L. A., 11
strong-field Zeeman effect, 376, 380
Two-electron systems, 184
for ^-coupling, 191
branching rule for, 202
for J /-coupling, 254
complete Paschen-Back effect for, 231
for L£-coupling, 191, 254
coupling schemes for, 189
derivation of terms for, 187 for Paschen-Back effect, 163
even terms for, 187, 204 ^'-coupling, 224
fine structure for, 193, 201 L5-coupling, 224
g factors for, 216, 217 for Stark effect, 405
g sum rule for, 222, 241 strong-field, 407
456 INTRODUCTION TO ATOMIC SPECTRA

Vector model, for Zeeman effect, 154, X-rays, non-diagram lines, 324
286, 373 notation, 308
complex spectra, 286 photoelectrons, 307
hyperfine structure, 373 photographs, 303, 305
one-electron systems, 154 reduced terms, 316, 321
Von Traubenberg, H. Rausch, 411 regular doublet law, 316, 318
relativity doublets, 315, 316
W satellites, 324
energy level diagram for, 328
Waller, I., 69 explanation of, 326
Walter, B., 300 screening constants, 317, 318
Wave numbers, defined, 6 screening doublets, 319
Weak field, Stark effect, 404 selection rules, 301, 324
Zeeman effect, 149, 217 shells, electron, 307, 309
Weisskopf, V., 418, 435 soft radiation, defined, 301
Wentzel, G., 106, 318, 322, 325, 394 Sommerfeld’s fine-structure formula
Weyl, A., 72 for, 315
White, H. E., 36, 70, 142, 267, 278, 296, spectra, photographs, 303, 305
332, 335, 339, 343, 349, 357, 364, spin-relativity doublets, 315, 316
subshells, 310
Whitelaw, N. G., 106 wave-lengths, 304
Wierl, R., 410 Xenon, 83, 274
Wilber, D. T., 296 energy level diagram for, 275
Williams, J-H., 313 fine structure for, 277
Wills
WilBorif E. b7Tio44 . Y
Wilson, W„ %
Ytterbium, 83
Wind, C. H., % I
Wolfe, H. C., 327 *’
Z
Wollaston, W. H., 1
Wood, R. W., 404 Zacharias, J. R., 361, 373
Wulfe, J., 380, 382 Zeeman, P., 149, 380
Zeeman effect, anomalous, 152
X classical explanation, 149
complex spectra, 286
X-rays, absorption spectra, 306 hyperfine structure, 373
explanation of, 329 energy level diagram for, 373
discovery of, 299 selection rules for, 375
electron shells, 308, 309 vector model for, 373
subshells, 310 intensity rules, 160, 220
energy level diagram, 308, 310 complex spectra, 289
fine structure, 314 ^/-coupling, 217, 218
formula for, 315 .LS-coupling, 217, 218
irregular doublet law for, 319 normal Zeeman triplet, 151, 219
K radiation defined, 301 defined, 151
L-doublets, 319 one-electron systems, 149
L radiation defined, 301 p and s components, defined, 151
Moseley diagram for, 311 photographs of, 220
Moseley law, 321 polarization of lines, 151, 160
X-series, 302 Preston’s law, 151
L-series, 305 selection rules, 159, 219
Af-series photographs, 305 complex spectra, 289
multiplet structure, 323 two electrons, 215
 INDEX 457
Zeeman patterns, 161, 222 Zeeman patterns, triplets, 222
calculation of, 221 vanadium, 291
* chromium, 291 Zinc, 82
complex spectra, 291 energy level diagram, 179
doublets, 161 nuclear moment of, 372
jy-coupling, 223, 224 photographs of triplets, 174
LS-coupling, 223 Zeeman effect, 152, 176, 220
rapid calculation of, 221 Zirconium, 83