PHYS 3015/3039/3042/3043/3044

First Semester 2017

Quantum Physics Module

Dr. Michael Schmidt
[email protected]

School
c of Physics, University of Sydney 2017

The course is based on the textbook Quantum Mechanics by David H. McIntyre. References to the
book are denoted by M: x.y, where x.y is the chapter number. These notes evolved from the lecture
notes of Associate Professor Brian James.

Contents
1 Quantum physics in one dimension 4
1.1 Review of Basic Concepts in Quantum Physics . . . . . . . . . . . . . . . . . . . . . . 4
1.1.1 Postulates of Quantum Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1.2 Orthogonality and Completeness . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1.3 Time evolution and Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.4 Wave function or quantum physics in position space . . . . . . . . . . . . . . . 5
1.2 Square well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.1 Infinite square well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.2 Inversion symmetry and parity . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.3 Finite square well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3 Quantum harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.1 Ladder operator method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.2 Application: molecular vibrational energy levels . . . . . . . . . . . . . . . . . . 14

2 Quantum Physics of Central Potentials 16

2.1 Separation of Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2 Solution to the Center of Mass equation . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 Classical Angular Momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4 Angular momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4.1 Spin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4.2 Orbital angular momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 The vector model for orbital angular momentum . . . . . . . . . . . . . . . . . . . . . 21
2.6 Application: molecular rotational energy levels . . . . . . . . . . . . . . . . . . . . . . 21
2.6.1 Rotational spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.6.2 Vibrational-rotational spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

1
 2.7 The Hamiltonian of a spherically symmetric potential . . . . . . . . . . . . . . . . . . 23
2.8 Separation of variables of radial and spherical part . . . . . . . . . . . . . . . . . . . . 23
2.9 Angular momentum eigenfunctions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.10 Inversion symmetry: parity of spherical harmonics . . . . . . . . . . . . . . . . . . . . 25
2.11 Visualisation of Spherical Harmonics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.12 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

3 Hydrogen Atom 28
3.1 Solution of the radial equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.1.1 Asymptotic Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.1.2 Series solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2 Full hydrogen wave functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3 Radial Probability Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.4 Hydrogen Energy Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.5 Degeneracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.6 Superposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.7 Emission Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

4 Transition probability 34
4.1 Radiative lifetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.2 Absorption and stimulated emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.3 Allowed & forbidden transitions: selection rules . . . . . . . . . . . . . . . . . . . . . . 35
4.4 Metastable levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

5 Magnetic moments, gyromagnetic ratio, ESR, NMR 37

5.1 Electron spin resonance (ESR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.2 Nuclear magnetic resonance (NMR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.3 Fine and hyperfine structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

6 Perturbation Theory 40
6.1 Non-degenerate Perturbation Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
6.2 Degenerate Perturbation Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

7 Fine structure, spin-orbit coupling, Dirac theory, Lamb shift 42

7.1 Relativistic correction to H energy levels . . . . . . . . . . . . . . . . . . . . . . . . . . 42
7.2 Spin-orbit coupling correction to H energy levels . . . . . . . . . . . . . . . . . . . . . 43
7.3 Dirac’s relativistic theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
7.4 Spectroscopic notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
7.5 The Lamb shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

8 Hyperfine structure; Addition of angular momenta 47

8.1 Hyperfine structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
8.2 Addition of angular momenta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

2
9 Zeeman and Paschen-Back effect 49
9.1 Zeeman effect without spin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
9.2 Zeeman effect with spin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
9.2.1 Weak magnetic field - anomalous Zeeman effect . . . . . . . . . . . . . . . . . . 51
9.3 Strong magnetic field - Paschen-Back effect . . . . . . . . . . . . . . . . . . . . . . . . 53
9.4 Arbitrary magnetic field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
9.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

10 Identical particles, symmetry requirements, fermions and bosons 55

10.1 Two spin- 21 particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
10.2 Symmetric or antisymmetric? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
10.3 Two identical particles in one dimension . . . . . . . . . . . . . . . . . . . . . . . . . . 56
10.4 Interacting Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
10.5 Helium atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
10.5.1 Ground state of helium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
10.5.2 First excited state of helium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
10.5.3 Spin-orbit coupling in helium . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
10.6 Pauli exclusion principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
10.7 Bose-Einstein condensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
10.8 Multi-electron atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
10.8.1 Alkali atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
10.8.2 The helium atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
10.8.3 LS coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

11 Time-dependent Perturbation Theory 66

11.1 Heisenberg and Interaction Picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
11.2 Transition probabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
11.3 Selection Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
11.4 Fermis Golden Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
11.4.1 Selection Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

A 3D infinitely deep potential well 73

B Bohr’s model of the atom 73

B.1 Bohr’s postulates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
B.2 Emission and absorption spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
B.3 Finite nuclear mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
B.4 The hydrogen spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

C Spherical Coordinates 76

D The Dirac equation 77

E Einstein relations 78

3
1 Quantum physics in one dimension
1.1 Review of Basic Concepts in Quantum Physics
1.1.1 Postulates of Quantum Mechanics
The Copenhagen interpretation of quantum mechanics can be summarised in the following postulates
1. The state of a quantum mechanical system (containing all information you know about it) is
mathematically represented by a (normalised) |ψi.

2. A physical observable is mathematically represented by an operator A acting on the kets

3. The only possible result of a measurement of an observable is one of the eigenvalues an of the
corresponding operator A.

4. The probability to obtain the eigenvalue an is given by

Pan = | han |ψi |2 (1.1)

where |an i is the normalised eigenvector of A with eigenvalue an .

5. After the measurement of A, the quantum system system is in the new (normalised) state

ψ = p Pn |ψi
 0
(1.2)
hψ|Pn |ψi
where Pn ≡ |an i han | is the projection operator onto the eigenstate |an i

6. The time evolution of a quantum mechanical system is described by the Schrödinger equation
∂
i~ |ψi = H(t) |ψi . (1.3)
∂t

1.1.2 Orthogonality and Completeness

Examples of operators are the momentum operator p, position operator x, and the Hamiltonian H,
whose eigenvalues are the energies of the different states. Eigenstates |an i of any hermitean operator
A are orthogonal1

han |am i = δnm (1.5)

and form a complete basis, i.e. any state can be expressed as a superposition
X
|ψi = cn |an i (1.6)
n

and in particular I can write the identity

X
|an i han | = 1 . (1.7)
n
1
The Kronecker-delta δnm is defined as (
1 for n = m
δnm ≡ . (1.4)
0 otherwise

4
The prime example is the Hamiltonian, i.e. the energy eigenstates form a complete orthonormal basis
X
hEn |Em i = δnm |En i hEn | = 1 . (1.8)
n

Thus quantum physics can be described by linear algebra, i.e. in terms of matrices and vectors.

1.1.3 Time evolution and Hamiltonian

The time evolution can be formally solved for a time-independent Hamiltonian2

|ψ(t)i = e−iH(t−t0 )/~ |ψ0 i (1.9)

with the quantum state at time t0 defined as |ψ(t0 )i ≡ |ψ0 i. It remains to solve the time-independent
Schrödinger equation, i.e. the eigenvalue equation of H: in terms of its eigenstates |Ei

H |Ei = E |Ei , (1.10)

i.e. find the eigenstates |Ei of the Hamiltonian H and determine their respective energies E. We will
be mostly dealing with the time-independent Schrödinger equation in this course.
The Hamiltonian is obtained from the Hamiltonian in classical physics by replacing all classical
quantities by operators. In classical mechanics the Hamiltonian determines the energy of a system
and it can be generally written as the sum of kinetic (T ) and potential (V ) energy

H =T +V → Ĥ = T̂ + V̂ . (1.11)

For example the kinetic energy T of a point particle with mass m is given by
p2 p̂2
T = → T̂ = (1.12)
2m 2m
and a position-dependent potential energy V would be represented by

V (x) → V̂ ≡ V (x̂) . (1.13)

1.1.4 Wave function or quantum physics in position space

We will now choose a (convenient) representation of the abstract operators and kets and work in
position space, i.e. kets will be represented by functions of position and operators will be act on those
functions. Concretely, the position operator x̂ and the momentum operator p̂ are represented by
∂
hx|x̂|xi ≡ x hx|p̂|xi ≡ −i~ (1.14)
∂x
and we will describe states by a wave function

ψ(x) = hx|ψi , (1.15)

i.e. the projection of the ket |ψi onto the eigenstates of the position operator x̂. The eigenstates of
the position operator form an orthogonal and complete set of states, i.e.
Z
hx|yi = δ(x − y) dx |xi hx| = 1 . (1.16)

2
A very similar formal solution exists for a time-dependent Hamiltonian.

5
Note that the sum in the completeness relation becomes an integral and the discrete Kronecker delta
becomes a delta function (delta distribution), which is defined as follows
Z
dxf (x)δ(x − y) = f (y) . (1.17)

The two operators x̂ and p̂ do not commute, but satisfy the commutation relation3

[p̂, x̂] = −i~ (1.20)

Following Eqs. (1.11,1.12,1.13), the Hamiltonian of a point particle in one dimension in a potential
V (x) is given by
~2 d2
H=− + V (x) (1.21)
2m dx2
and the energy eigenvalue equation becomes a differential equation
~2 d2
 
− + V (x) ϕE (x) = EϕE (x) (1.22)
2m dx2
with the eigenfunctions ϕE (x) ≡ hx|Ei.
The generalisation to more dimensions is straightforward. In three dimensions with, the Hamilto-
nian reads
p̂21 + p̂22 + p̂23
Ĥ = + V (x̂1 , x̂2 , x̂3 ) (1.23)
2m
~2 ∂2 ∂2 ∂2
 
=− + + + V (x1 , x2 , x3 ) (1.24)
2m ∂x21 ∂x22 ∂x23
and the commutation relations between the momentum operators p̂i and the position operators x̂j are
given by
[p̂i , x̂j ] = −i~δij . (1.25)
The probability to find a particle in the infinitesimal interval [x, x + dx] is given by

P(x) = |ψ(x)|2 dx . (1.26)

where |ψ(x)|2 can be interpreted as probability density. The probability to detect the quantum state
in the finite interval a < x < b is simply given by the integral over the probability density with the
boundaries a and b. Z b
Pa≤x≤b = |ψ(x)|2 dx . (1.27)
a
In particular the normalisation condition becomes
Z Z
1 = hψ|ψi = dx hψ|xi hx|ψi = ψ(x)∗ ψ(x)dx (1.28)

3
It can be easily derived by evaluating the commutator in position space for an arbitrary wave function
 
d d d
[−i~ , x]ψ(x) = −i~ xψ(x) − x ψ(x) (1.18)
dx dx dx
 
d d
= −i~ ψ(x) + x ψ(x) − x ψ(x) = −i~ψ(x) (1.19)
dx dx

6
and the probability amplitude to find a quantum state |ψi in the state |ϕi is given by
Z
hϕ|ψi = ϕ∗ (x)ψ(x)dx . (1.29)

The probability is simply | hϕ|ψi |2 . Finally, an expectation value of a local operator  [hx|A|yi =
hx|A|xi δ(x − y) ≡ A(x)δ(x − y)] is given by
D E D E Z Z D E Z
 = ψ|Â|ψ = dy dx hψ|xi x|Â|y hy|ψi = ψ ∗ (x)A(x)ψ(x)dx , (1.30)

where we used the completeness of position eigenstates and the definition of the delta-function. The
condition for the orthogonality of eigenstates of a hermitean operator A can be written as
Z Z
δnm = hn|mi = dx hn|xi hx|mi = dxϕn (x)∗ ϕm (x) = δnm (1.31)

and any wave function can be expressed as a superposition of the eigen wave functions ϕn (x) = hx|ni
of the operator A X X
ψ(x) = hx|ψi = cn hx|ni = cn ϕn (x) . (1.32)
n n

Summarising the correspondence between the bra-ket formalism and quantum mechanics in posi-
tion space, i.e. the wave function formalism, is given by
Z
∗
|ψi ↔ ψ(x) hψ| ↔ ψ (x) h|i ↔ dx  ↔ A(x) (1.33)

After recapitulating the basic notions of quantum mechanics, we apply them to simple one-dimensional
quantum mechanical systems.

1.2 Square well

A well studied problem in quantum physics is the square well. It serves as a prototype for more
complicated problems.

1.2.1 Infinite square well

The potential of an infinitely deep potential well of size L is given by

∞,
 x<0
V (x) = 0 0<x<L (1.34)

∞, x>L.


We solve the time-independent Schrödinger equation (1.22) in the three different regions separately.
Outside the box the potential energy is infinite and thus the only solution is ϕE (x) ≡ 0. Inside the
box the potential energy is zero and it can be rewritten as
 2 
d 2mE
+ 2 ϕE (x) = 0 . (1.35)
dx2 ~

7
It is a linear second order ordinary differential equation (ODE) with constant coefficients and can be
solved using the ansatz
ϕE (x) = Aeikx . (1.36)
The characteristic equation is given by
2mE
− k2 + =0 (1.37)
~2
and the general solution inside the box is the linear superposition of the different possible solutions
ϕE (x) = Aeikx + Be−ikx (1.38)
√
with k = 2mE/~ and A, B two complex numbers, which are determined by the boundary conditions
and the normalisation. The general solution is thus given by

0,
 x<0
ikx
ϕE (x) = Ae + Be −ikx 0<x<L (1.39)

0, x>L.


We have to require that the solution is continuous everywhere, i.e.

ϕE (0) = ϕE (L) = 0 ⇒ A + B = 0 and AeikL + Be−ikL = 0 . (1.40)
Hence there is a discrete set of solutions of the form
π
ϕn (x) = A0 sin(kn x) with kn = n , n = 1, 2, 3, . . . (1.41)
L
with energies
n2 π 2 ~2
En = . (1.42)
2mL2
The constant A0 is fixed by correctly normalising the wave function
Z L
1= dx|A0 |2 sin2 (kn x)dx (1.43)
0
and the correctly normalised energy eigenstates are
r
2 nπx
ϕn (x) = sin( ), n = 1, 2, 3, . . . (1.44)
L L

1.2.2 Inversion symmetry and parity

Inspection of the wave functions for the states of a particle in an infinite well show that, with respect
to the middle of the well, the wave functions are alternately symmetric (n=1,3,5,. . . ) or antisymmetric
(n=2,4,6,. . . ). This is even clearer if we change the x-axis origin to the middle of the well, and place
the walls at x = ±a so that the width of the well is 2a, and the wave functions become
r r
1 nπx 1 nπx
ϕn (x) = cos n = 1, 3, 5, . . . ϕn (x) = sin n = 2, 4, 6, . . . (1.45)
a 2a a 2a
Clearly state with odd n, are symmetric with respect to inversion (x → −x); those with even n
are antisymmetric. Symmetric states where ϕn (−x) = +ϕn (x) are said to have positive parity;
antisymmetric states where ϕn (−x) = −ϕn (x) are said to have negative parity. We will see later that
allowed transitions can only occur between states of opposite parity.

8
1.2.3 Finite square well
If the sides of the well are not infinite, so that the potential well of width 2a is defined by4

V0 ,
 x < −a
V (x) = 0 −a < x < a (1.46)

V0 , x>a.


Inside the box the Schrödinger equation is the same as for the infinitely deep potential well. Outside
the box the Schrödinger equation can be rewritten as
 2 
d 2m(V0 − E)
− ϕE (x) = 0 . (1.47)
dx2 ~2
Assuming a bound state with 0 < E < V0 we can similarly make an ansatz ϕE (x) = Aeqx and obtain
the general solutions with
2m(V0 − E)
q2 = (1.48)
~2
and the general solution is given by

qx −qx
Ae + Be , x < −a

ϕE (x) = Ceikx + De−ikx −a < x < a (1.49)

Ee + F e−qx , x > a
 qx

In order to interpret the wave function, we have to normalise it, this automatically forces B = E = 0.
Furthermore demanding that the solution is
1. continuous
2. and continuously differentiable ( dϕdx
E (x)
being differentiable) [if V0 6= ∞]
everywhere fixes three of the other coefficients. The last coefficient is fixed by the normalisation
condition.
Because of the finite height of the barriers, the particle can tunnel into the region beyond the edges
of the well, where on physical grounds we would expect the wave functions to decay with distance
beyond the edge of the well. Within the well, where the particle is free, we expect the wave functions
to be sinusoidal as in the case of the infinite well. The solution of the energy eigenvalue equation
in this case is not trivial (in fact it can be solved by numerical means only). For solutions to be
physically acceptable, the wave functions must approach zero with distance away from the well, and
be continuous and smooth across the boundaries (i.e. ϕE (x) and dϕE (x)/dx are continuous at the
walls of the well). Such solutions occur for discrete values of E only.
As the equation can be solved by numerical means only, it is not possible to write down a formula
for the energy levels. However, as expected, the energies approach those of the infinite case as V0 → ∞.
Figure 1 is a reproduction of Figure 5.18 from McIntyre showing energy levels for an infinite square
well and finite square well of the same width. Note that for the finite square well the wave function
penetrates the walls of the well, beyond which it decays, and that its value and gradient are continuous
across the well boundary, as required. In the case of the infinite well the wave function vanishes at
the walls of the well and in this limit the gradient is discontinuous at the edge of the well.
4
The width of the well is given as L for the infinite case and 2a for the finite case in order to be consistent with the
notation used by McIntyre

9
Figure 1: Energy levels and wave functions for (a) an infinite square well of width 2a and (b) a finite
square well of width 2a. Note that as the finite well has reduced confinement, the energy levels are
lower than those for an infinite well of the same width. Based on figure 5.18 of Quantum Mechanics
by D.H. McIntyre.

1.3 Quantum harmonic oscillator

The last 1D example which we are studying is the harmonic oscillator. It is very important approxi-
mation to many physical phenomena such as the vibrational modes of a diatomic molecule as shown
in Fig. 2. A particle with mass m is subject to a restoring force −kx, where x is the displacement
from the equilibrium position. The potential energy of the particle is
1 1
V (x) = kx2 = mω 2 x2 (1.50)
2 2
p
where ω = k/m is the angular oscillation frequency. Thus its Hamiltonian is
p̂2 1 ~2 d2 1
H= + mω 2 x̂2 = − 2
+ mω 2 x2 (1.51)
2m 2 2m dx 2
and the time-independent Schrödinger equation is given by
~2 d2
 
1 2 2
− + mω x ϕE (x) = EϕE (x) . (1.52)
2m dx2 2
”The career of a young theoretical physicist consists of treating the harmonic oscillator in ever-
increasing abstraction.” – Sidney Coleman
Now we will be going one step and solve the quantum harmonic oscillator using the ladder operator
method.

1.3.1 Ladder operator method

In order to simplify our following discussion, we define the variable
r r
mω d ~ d
ξ= x = (1.53)
~ dξ mω dx

10
Figure 2: First few vibrational modes of a diatomic molecule are well described by a harmonic oscil-
lator. Figure taken from https://en.wikipedia.org/wiki/File:Morse-potential.png.

and can rewrite our Hamiltonian in Eq. (1.51) as follows

d2
 
1 2
H = ~ω − 2 + ξ (1.54)
2 dξ
In the next step we want to factorise the Hamiltonian in analogy to the identity
u2 − v 2 = (u − v)(u + v) . (1.55)
Hence we form two new operators as linear combination of the old ones
  r  
1 d mω p̂
a= √ ξ+ = x̂ + i (1.56)
2 dξ 2~ mω
  r  
† 1 d mω p̂
a =√ ξ− = x̂ − i . (1.57)
2 dξ 2~ mω
The operators satisfy the following commutation relation
[a, a† ] = 1 . (1.58)
Looking at the product of the two operators
   
† 1 d 1 d
a a= √ ξ− √ ξ+ (1.59)
2 dξ 2 dξ
2
 
1 2 d d d
= ξ − 2 +ξ − ξ (1.60)
2 dξ dξ dξ
2
 
1 2 d d
= ξ − 2 + [ξ, ] (1.61)
2 dξ dξ
2
 
1 2 d
= ξ − 2 −1 (1.62)
2 dξ

11
we observe that we almost obtain the result which we wanted to obtain. Hence, the Hamiltonian can
be rewritten as  
† 1
H = ~ω a a + . (1.63)
2
Before interpreting the operators â and ↠physically, we have to show two more commutation
relations:

[H, a] = Ha − aH (1.64)
   
1 1
= ~ω a† a + a − a~ω a† a + (1.65)
2 2
 
= ~ω a† aa − aa† a (1.66)
   
= ~ω a† aa − a† a + [a, a† ] a (1.67)
= −~ωa (1.68)

Now given an energy eigenstate |Ei with a given energy E, we can calculate the energy eigenvalue of
the states a |Ei as follows

H(a |Ei) = Ha |Ei (1.69)

= (aH + [H, a]) |Ei (1.70)
= (aE − ~ωa) |Ei (1.71)
= (E − ~ω)(a |Ei) (1.72)

Similarly for the operator a†

[H, a† ] = +~ωa† . (1.73)

and

H(a† |Ei) = (E + ~ω)(a† |Ei) . (1.74)

Hence the states a |Ei, a† |Ei are also energy eigenstates with energies E ± ~ω, respectively. The
operators a and a† transform a state with energy E into a state with energy E ± ~ω. They are
denoted ladder operators, more specifically a† is denoted raising operator and a lowering operator.
Next we have to find the lowest energy eigenstate or ground state. Classically we observe that
there is a minimum energy of the harmonic oscillator. Hence there has to be a lowest energy eigenstate

a |Elowest i = 0 . (1.75)

This is called the ladder termination condition. The energy of this lowest energy eigenstate is given
by  
† 1 1
H |Elowest i = ~ω a a + |Elowest i = ~ω |Elowest i . (1.76)
2 2
Note that lowest energy level is not zero as it would be for a classical harmonic oscillator, but 21 ~ω.
It is known as zero point energy and ultimately due to the non-vanishing commutator of the ladder

12
 ↠â
↠â
↠â
↠â
↠â
↠â
↠â

Figure 3: Energy levels of harmonic oscillator. Raising operator a† increases energy by ~ω and
lowering operator a lowers it. Figure taken from https://commons.wikimedia.org/w/index.php?
curid=11623546.

operators [a, a† ] = 1. The lowest energy eigenstate is commonly denoted |0i. The energy of the nth
state |ni is given by  
1
En = ~ω n + , (1.77)
2
because applying the raising operator a† n times increases the energy with respect to the lowest energy
eigenstate by n × ~ω. In addition to the ladder operators it is convenient to introduce the number
operator,
N̂ = ↠â , (1.78)
which counts the energy quanta. It fulfils the following eigenvalue equation

N̂ |ni = n |ni , (1.79)

where the n in |ni denotes the number of energy quanta. We can rewrite the Hamiltonian as
 
1
Ĥ = ~ω N̂ + . (1.80)
2

See Fig. 3 for an illustration of the action of the ladder operator on the energy eigenstates. All other
energy eigenstates can be constructed from the lowest energy eigenstate using the raising operator.
By demanding that all states |ni are properly normalised,

hn|ni = 1 , (1.81)

it is possible to show5 that the raising and lowering operators act on a state |ni
√
a |ni = n |n − 1i (1.82)
√
a† |ni = n + 1 |n + 1i . (1.83)
5
See the discussion in McIntyre Chap.9.

13
Thus we can write the state |ni as follows
1  † n
|ni ≡ √ a |0i , (1.84)
n!
√
where we denoted the lowest energy eigenstate by |0i. The factor 1/ n! ensures that the states are
correctly normalised.
The wave function of the lowest energy eigenstate φ0 (ξ) can be determined from the ladder termi-
nation condition in Eq. (1.75)
 
1 d
0 = aφ0 (ξ) = √ ξ+ φ0 (ξ) . (1.85)
2 dξ
It is an ODE, which can be solved using standard techniques
 mω 1/4 2
φ0 (ξ) = e−ξ /2 . (1.86)
π~
The explicit form of the wave function involves a well studied special function, the Hermite polynomial,
such that the normalised wave functions can be written as
 mω 1/4 1 r
−ξ 2 /2 mω
ϕE (x) = √ Hn (ξ)e with ξ = x and n = 0, 1, 2, 3, . . . (1.87)
π~ n
2 n! ~
where Hn is a Hermite polynomial of order n

H0 (x) = 1, H1 (x) = 2x, H2 (x) = 4x2 − 2 H3 (x) = 8x3 − 12x . (1.88)

Figure 4 shows wave functions and probability densities for the first four energy levels of a harmonic
oscillator.

Figure 4: (a) Wave functions and (b) the probability density for the first four energy levels of a
harmonic oscillator.

1.3.2 Application: molecular vibrational energy levels

The two nuclei of a diatomic molecule can vibrate about their equilibrium separation. For sufficiently
small amplitude vibrations the motion can be treated as a harmonic oscillation in a parabolic potential.
The quantised energy levels would then be given by

14
Figure 5: Potential curve for a typical diatomic molecule showing vibrational energy levels. At lower
levels where the potential approximates a parabola the energy levels are almost equally spaced. At
higher levels where the potential profile becomes wider than a parabola, the spacing decreases.

1 1
En = (n + )~ω0 = (n + )hν0 where n = 0, 1, 2, 3, . . . (1.89)
2 2
and ν0 = ω0 /2π is the classical vibration frequency (in Hz). Transitions between adjacent vibrational
energy levels are in the infrared part of the electromagnetic spectrum.6 Figure 5 shows a typical
potential curve (energy as a function of nuclear separation) for a diatomic molecule. At low energy
levels, where a parabolic approximation to the potential curve is a good fit, the actual energy levels
are equally spaced, but at higher levels where the parabolic approximation is a poorer fit the spacing
between levels gradually decreases.

6
In fact transitions for which |∆n| > 1 are much more unlikely than those for which |∆n| = 1, and are called forbidden
transitions. The condition |∆n| = 1 is called a selection rule and transitions which satisfy this condition are said to be
allowed transition.

15
2 Quantum Physics of Central Potentials
One of the main goals of this course is to solve the hydrogen atom. It is a two-body problem in 3
dimensions, with a positively charged proton and a negatively charged electron. They interaction via
Coulomb interaction which is described by the Coulomb potential

Ze2 Zα~c
V (|r1 − r2 |) = − =− . (2.1)
4π0 |r1 − r2 | |r1 − r2 |

It is a central potential, which only depends on the relative distance between the two particles.

2.1 Separation of Variables

We will actually solve the slightly more general problem of an arbitrary two-body system with a central
potential. It is described by the Hamiltonian

p̂21 p̂2
Ĥsys = + 2 + V (|r̂1 − r̂2 |) , (2.2)
2m1 2m2
where mi , p̂i , and r̂i are the mass, momentum and position of particle i.
Similarly to classical mechanics, the Hamiltonian can be separated in the Hamiltonian of the centre
of mass motion and the relative motion about the centre. We define the centre of mass coordinate
m1 r̂1 + m2 r̂2
R̂ = (2.3)
m1 + m2
and the relative position vector
r̂ = r̂2 − r̂1 . (2.4)
Similarly we define the momentum in the centre of mass frame

P̂ = p̂1 + p̂2 (2.5)

and the relative velocity in terms of the relative momentum p̂rel

p̂rel p̂2 p̂1
= − (2.6)
µ m2 m1
with the reduced mass
1 1 1
= + . (2.7)
µ m1 m2
Rewriting the Hamiltonian in these new coordinates, it separates in a Hamiltonian describing the
centre of mass motion and the relative motion, Ĥ = ĤCM + Ĥrel with

P̂2
ĤCM = (2.8)
2M
p̂2
Ĥrel = rel + V (r) (2.9)
2µ
and M = m1 + m2 . It is a valid assumption that the quantum state also separates

ψsys (R, r) = ψCM (R)ψrel (r) , (2.10)

16
because the potential does not depend on the centre of mass coordinate R̂ and did not depend on
it previously, but it only depends on the relative coordinate r̂. Explicitly we can show using the
Schrödinger equation
Eψsys (R, r) = Ĥψsys (R, r) (2.11)
 
= ĤCM + Ĥrel ψCM (R)ψrel (r) (2.12)
 
= ψrel (r)ĤCM ψCM (R) + ψCM (R)Ĥrel ψrel (r) (2.13)
(2.14)
We divide by ψsys
−1 −1
E = ψCM (R)ĤCM ψCM (R) + ψrel (r)Ĥrel ψrel (r) (2.15)
and bringing one of the ratios to the other side of the equation
−1 −1
E − ψCM (R)ĤCM ψCM (R) = ψrel (r)Ĥrel ψrel (r) , (2.16)
we see that the left-hand side of the equation is independent of r and the right-hand side independent
of R. Hence both of them have to be constant, which we denote by Erel , the energy of the subsystem
of the relative coordinate. Thus we can write
ĤCM ψCM (R) = ECM ψCM (R) (2.17)
Ĥrel ψrel (r) = Erel ψrel (r) (2.18)
where ECM = E − Erel is the energy of for the subsystem describing the centre of mass coordinates.

2.2 Solution to the Center of Mass equation

The Hamiltonian for the centre of mass motion is given by
~2
 2
∂2 ∂2

∂
ĤCM = − + + (2.19)
2M ∂X 2 ∂Y 2 ∂Z 2
with R = (X, Y, Z)T . It corresponds to the Hamiltonian of a free particle in three dimensions, which
can be solved in terms of free particle eigenstates
1 1
ψCM (X, Y, Z) = 3/2
ei(PX X+PY Y +PZ Z)/~) = eiP·R/~ (2.20)
(2π~) (2π~)3/2
with P = (PX , Py , Pz )T . The energy eigenvalues are given by
1
PX2 + PY2 + PZ2 .


ECM = (2.21)
2M
Note that the momentum P is not discrete, but a continuous variable. Using Eq. (1.9), the time-
dependent wave-function is then given by
1
ψCM (t, X, Y, Z) = e−i(Et−P·R)/~ , (2.22)
(2π~)3/2
where we set t0 = 0. The solution are plane waves travelling forward with phase velocity |vph | = E/|P|
and group velocity vg = ∇p E. The average momentum is given by the expectation value of the
momentum operator
D E Z
∗
P̂ = d3 xψCM (X, Y, Z) (−i~∇) ψCM (X, Y, Z) = P . (2.23)

17
2.3 Classical Angular Momentum
In classical mechanics, the angular momentum is conserved for central forces, like the Coulomb po-
tential in the hydrogen atom
dL dr × p
= (2.24)
dt dt
=v×p+r×F (2.25)
= −r × ∇V (r) (2.26)
r
= −r × V 0 (r) = 0 . (2.27)
r
We will make use of that fact to simplify the eigenvalue equation for the relative motion of the hydrogen
atom.

2.4 Angular momentum

We are already familiar with the fact that an electron has intrinsic angular momentum, usually called
spin. This is a property of the electron, just like its mass or charge. An electron in an atom, however,
can also have angular momentum due to its motion, usually called orbital angular moment. Before
considering orbital angular momentum, let us summarise what we know about electron spin.

2.4.1 Spin
An electron has spin characterised by a spin quantum number s = 1/2 and magnetic quantum numbers
ms = ±1/2, with spin magnitude and component values quantised according to

3
Ŝ2 |sms i = s(s + 1)~2 |sms i = ~ |sms i Ŝz |sms i = ms ~ |sms i = ± 12 ~ |sms i (2.28)
4
In the Sz representation the spin eigenstates |sms i are
 
1 1 1
|+i =  ,
2 2 = (2.29)
0
 
0
=  12 , − 21
 
|−i = (2.30)
1

The various spin operators are given by

 
~ 0 1 ~
Sˆx = = σ1 (2.31)
2 1 0 2
 
~ 0 −i ~
Sˆy = = σ2 (2.32)
2 i 0 2
 
~ 1 0 ~
Sˆz = = σ3 (2.33)
2 0 −1 2
 
3 2 1 0
Sˆ2 = ~ , (2.34)
4 0 1

18
 (b) The vector model of angular momentum for
(a) The vector model for electron spin. l = 3. Angular momentum values are given in
units of ~.

Figure 6: The vector model

where σi denotes the three Pauli spin matrices. The vector model for spin in Figure 6a conveniently
summarises the quantisation properties for spin. It is straightforward to show7

[Ŝx , Ŝy ] = i~Ŝz (2.35)

[Ŝy , Ŝz ] = i~Ŝx (2.36)
[Ŝz , Ŝx ] = i~Ŝy (2.37)
2 2 2
[Ŝ , Ŝx ] = [Ŝ , Ŝy ] = [Ŝ , Ŝz ] = 0 . (2.38)

2.4.2 Orbital angular momentum

The angular momentum of a particle is given by L = r×p. As the momentum operator in the position
representation is given by  
∂ ∂ ∂
p̂ = −i~ , , = −i~∇ (2.39)
∂x ∂y ∂z
the operator for angular momentum is therefore

L̂ = r × (−i~∇) (2.40)

The operators for the components and the magnitude of angular momentum are
7
The three spin operators form the Lie algebra of SU(2) and the square of the vector of spin operators, Ŝ2 is
called Casimir operator, which commutes with all three spin operators. The Casimir operator plus any of the three
spin operators form the maximal commuting set of operators and can be used to classify quantum states with angular
momentum.

19
 • Cartesian coordinates

 
∂ ∂
L̂x = −i~ y −z (2.41)
∂z ∂y
 
∂ ∂
L̂y = −i~ z −x (2.42)
∂x ∂z
 
∂ ∂
L̂z = −i~ x −y (2.43)
∂y ∂x
L̂2 = L̂2x + L̂2y + L̂2z (2.44)

• Spherical polar coordinates (see Figure 28 in Sec. C)

 
∂ ∂
L̂x = i~ sin φ + cot θ cos φ (2.45)
∂θ ∂φ
 
∂ ∂
L̂y = i~ − cos φ + cot θ sin φ (2.46)
∂θ ∂φ
∂
L̂z = −i~ (2.47)
∂φ
1 ∂2
   
1 ∂ ∂
L̂2 = −~2 sin θ + (2.48)
sin θ ∂θ ∂θ sin2 θ ∂φ2

It is straightforward to show that the commutation relations for the angular momentum components
are
h i h i h i
L̂x , L̂y = i~L̂z , L̂y , L̂z = i~L̂x , L̂z , L̂x = i~L̂y . (2.49)

With a little more algebraic manipulation (see McIntyre, p211), it is possible to show that Lˆ2 commutes
with each of the component operators
h i h i h i
L̂2 , L̂x = L̂2 , L̂y = L̂2 , L̂z = 0 (2.50)

using the identity [AB, C] = A[B, C] + [A, C]B. We note that these are the same commutation
relations satisfied by spin angular momentum. By analogy, therefore we can conclude it is possible
to find states that are simultaneously eigenstates of L̂2 and one of the component operators, which
according to convention we choose to be L̂z . Recalling that the spin eigenvalue equations are

Ŝ 2 |sms i = s(s + 1)~2 |sms i (2.51)

Ŝz |sms i = ms ~ |sms i (2.52)

where ms = s, s − 1, . . . , −s, the equivalent equations for orbital angular momentum will be

L̂2 |lml i = l(l + 1)~2 |lml i (2.53)

L̂z |lml i = ml ~ |lml i (2.54)

where ml = l, l − 1, . . . , −l.

20
 We are now using the ladder operator method to understand the restriction ot |ml | ≤ l. Analo-
gously to the ladder operators for the harmonic oscillator, we can define raising and lowering operators
for the angular momentum operators
L̂± = L̂x ± iL̂y , (2.55)
which satisfy the following properties

(L̂± )† = L̂∓ (2.56)

[L̂z , L̂± ] = i~L̂y ± ~L̂x = ±~L̂± (2.57)
[L̂2 , L̂± ] = 0 . (2.58)

The operator L̂± raises (lowers) the magnetic quantum number by ~

   
L̂z (L̂± |lmi) = L̂± L̂z + [L̂z , L̂± ] |lmi = L̂± m~ ± ~L̂± |lmi = (m ± 1) ~L̂± |lmi , (2.59)

but preserves the total angular momentum because L̂± commutes with the total angular momentum
operator. The normalization of the states hlm|lmi = 1
D E D   E
lm|L̂†± L̂± |lm = lm| L̂2 − L̂2z ∓ ~L̂z |lm = ~2 l(l + 1) − m2 ∓ m


(2.60)

leads to p
L̂± |lmi = ~ l(l + 1) − m(m ± 1) |lm ± 1i (2.61)
and thus restricts |m| ≤ l.
Although we are mostly concerned with spin 21 systems, the spin quantum number s can in principle
be half integer or integer. In contrast we will see soon that the orbital angular quantum number l can
only be an integer; for the moment we will assume that this is the case.
In order to simplify the notation we will often drop the hat when writing operators, i.e. denote
operators by A instead of Â.

2.5 The vector model for orbital angular momentum

The quantisation of angular momentum can be summarised by the vector model which shows the
possible orientations of the angular momentum vector. The fact that the vector is limited to having
a discrete number of orientations with respect to the z axis is referred to as space quantisation.
√
Figure 6b depicts the vector model of angular momentum for l = 3, for which |L| = 2 3~ and
Lz = (3, 2, 1, 0, −1, −2, −3)~.

2.6 Application: molecular rotational energy levels

The moment of inertia of a diatomic molecule about its centre of mass is

I = µR02 (2.62)
where µ is the reduced mass of the molecule and R0 is the nuclear separation. The energy associated
with rotation can be expressed in terms of the angular momentum L as

L2
Erot = (2.63)
2I

21
Noting that angular momentum is quantised according to L2 = l(l + 1)~2 , where l is an integer
quantum number, leads to quantised rotational energy levels given by

~2
Erot = l(l + 1) where l = 0, 1, 2, 3, . . . (2.64)
2I

2.6.1 Rotational spectra

The selection rule8 for allowed transitions is : ∆l = ±1. Therefore,

hc ~2
hν = = hcσ = El+1 − El = (l + 1) (2.65)
λ I
where σ = 1/λ is the wave number.9 Rotational spectra are in the far infrared and microwave regions
between of the electromagnetic spectrum (λ ∼ 0.1 mm - 10 mm)10 . Only molecules with a permanent
dipole moment will have a pure rotational spectrum. Thus molecules with identical nuclei (e.g. H2 ,
C2 ) do not have pure rotational spectra.

2.6.2 Vibrational-rotational spectra

There will be rotational levels associated with each vibrational level. Thus a transition between vibra-
tional states may also involve a change in rotational state, and we speak more generally of vibrational-
rotational spectra for which the selection rules are: ∆l = ±1 and ∆n = ±1. Vibrational-rotational
spectra are typically in the infrared. As molecules are predominantly in the ground vibrational levels
at room temperature, vibrational-rotational spectra are usually observed as absorption spectra.

Example
For absorption transitions between vibrational levels n and n + 1, there are two groups of absorption
lines corresponding to ∆l = ±1. A small amount of algebra yields the following:

~2
hν = hν0 ± l l = 1, 2, 3, . . . (2.66)
I
Thus the lines are in two groups on either side of ν0 , with equal spacing of
~
∆ν = (2.67)
2πI
Note that there is no line at ν0 . Measurement of the spacing between lines allows the moment of inertia
of the molecule to be calculated. Figure 7 shows the rotational-vibrational absorption spectrum of
HBr. The spacing between the absorption lines is slightly non-uniform due to increase of the inter-
atomic separation (i.e. stretching of the bond) with increasing angular momentum (i.e. with increasing
value of the quantum number l).
8
Selection rules are the result of conservation of angular momentum when a photon is emitted or absorbed. In most
cases they are expressed in terms of the change in quantum numbers between the initial and final states. Transitions
that satisfy selection rules are refereed to as allowed transitions. Transitions that do not satisfy selection rules are called
forbidden transition. They can still occur but at much lower transition rates.
9
For vibrational-rotational spectra, wave number is often used instead of either wavelength or frequency. Although
the SI unit is m−1 , cm−1 is commonly used. The symbol ν is often used in older texts.
10
The terms sub-millimetre wave and millimetre wave are also used to describe a region of the electromagnetic spectrum
between the far-infrared and microwave regions.

22
 "l = #1 "l = +1
!0

! !
Figure 7: The vibrational-rotational absorption spectrum for HBr. (based on Figure 8-5, Basic
Principles of Spectroscopy, R. Chang, McGraw-Hill, New York, 1971)

2.7 The Hamiltonian of a spherically symmetric potential

We now move on to the energy eigenvalue equation of the relative motion for a spherically symmetric
potential, Eq. (2.9) and Eq. (2.18). Its energy eigenvalue equation can be written as
 2 2 
~ ∇
− + V (r) ψ(r, θ, φ) = Eψ(r, θ, φ) (2.68)
2µ
As it is spherically symmetric, it is convenient to express the Laplace operator in terms of spherical
coordinates, which is given in Eq. (C.3). Noting that the θ and φ parts are proportional to the L̂2
operator (Eq. (2.48)), the Hamiltonian of the relative motion can be written as
~2 1 ∂ L2
   
2 ∂
Hrel = − r − 2 2 + V (r) . (2.69)
2µ r2 ∂r ∂r ~ r
As the angular part of H is completely contained in L2 , it is obvious that

[H, L2 ] = [H, Lz ] = 0 (2.70)

and thus the magnitude of the orbital angular momentum and Lz are good quantum numbers. The
energy eigenvalue equation can be written as
~2 1 ∂
   
2 ∂ 1 2
− r − 2 2 L̂ ϕE (r, θ, φ) + V (r)ϕE (r, θ, φ) = EϕE (r, θ, φ) . (2.71)
2µ r2 ∂r ∂r ~ r

2.8 Separation of variables of radial and spherical part

The solutions of this partial differential equation (PDE) will be ϕE (r, θ, φ), wave functions for the
energy eigenstates. As each term of the PDE involves derivatives of only one independent variable, it
can be solved using the separation of variables technique. We begin by letting

ϕ(r, θ, φ) = R(r)Y (θ, φ) (2.72)

23
and substituting into Eq. (2.71). The resulting equation is then divided through by R(r)Y (θ, φ) and
the result can be rearranged to give:
 
1 d dR(r) 2µ 1 1
r2 + 2 (E − V (r))r2 = 2 L̂2 Y (θ, φ) (2.73)
R(r) dr dr ~ ~ Y (θ, φ)

Note that the left-hand side is a function of r only while the right-hand side is a function of θ and φ
only. The two sides can be equal for all values of the independent variables only if they are equal to
a constant, A. We have therefore an ODE involving r (which we will deal with shortly) and a PDE
involving θ and φ:
L̂2 Y (θ, φ) = A~2 Y (θ, φ) (2.74)

2.9 Angular momentum eigenfunctions

Writing Y (θ, φ) = Θ(θ)Φ(φ), another stage of separation, with the introduction of another constant
B, leads to two ODEs, one for each of the independent variables θ, φ:
   
1 d d 1
sin θ −B 2 Θ(θ) = −AΘ(θ) (2.75)
sin θ dθ dθ sin θ
d2 Φ(φ)
= −BΦ(φ) . (2.76)
dφ2

The solution to Eq. (2.76) is √

Φ(φ) ∝ e±i Bφ
. (2.77)
It follows that √ √
L̂z Y (θ, φ) = −i~(±i B)Y (θ, φ) = ±~ BY (θ, φ) (2.78)
Comparing Eqs. (2.74) and (2.78) with Eqs. (2.53) and (2.54) we can make the following identifications:

A = l(l + 1) B = m2l (2.79)

and
hθφ|lml i ∝ Y (θ, φ) . (2.80)
As wave function represent physical reality, they must be single valued, so Φ(φ + 2π) = Φ(φ), which
requires that ml be an integer, and by implication that l is also an integer.
To summarise, the θ and φ functions are the solutions, respectively, to the equations
   
1 d d 2 1
sin θ − ml Θ(θ) = −l(l + 1)Θ(θ) (2.81)
sin θ dθ dθ sin2 θ
d2 Φ(φ)
= −m2l Φ(φ) (2.82)
dφ2
where l is an integer and ml = l, l − 1, . . . , −l. The solutions of the θ equation are called associated
Legendre functions, and the solution to the φ equation is an exponential function. The orbital an-
gular momentum wave functions are product of these, and when appropriately normalised are called
spherical harmonics:
hθφ|lml i = Ylml (θ, φ) , (2.83)

24
where the dependence on l and ml is explicitly indicated by subscript and superscript respectively;
the normalisation condition is Z 2π Z π
|Ylml |2 sin θdθdφ = 1 (2.84)
0 0
The spherical harmonics can be expressed in terms of associated Legendre polynomials
s
(2l + 1) (l − |m|)! m
Ylm (θ, φ) = (−1)(m+|m|)/2 P (cos θ)eimφ (2.85)
4π (l + |m|)! l

The sign convention leads to

Yl−m (θ, φ) = (−1)m Ylm∗ (θ, φ) (2.86)
Several spherical harmonics are shown in Table 1. The spherical harmonics form a complete basis of

l ml Ylml (θ, φ) parity: (−1)l

q
1
0 0 Y00 = 4π +1
q
3
1 0 Y10 = 4π cos θ -1
q
±1 Y1±1 = ∓ 8π 3
sin θe±iφ -1
q
5
2 0 Y20 = 16π (3 cos2 θ − 1) +1
q
±1 Y2±1 = ∓ 8π 15
sin θ cos θe±iφ +1
q
±2 Y2±2 = 32π 15
sin2 θe±i2φ +1

Table 1: Several spherical harmonics. More can be found in Table 7.3 of Quantum Mechanics by D.H.
McIntyre, p238.

smooth (wave) functions on a sphere and satisfy the following orthonormality condition
Z π Z 2π
1∗
hl1 m1 |l2 m2 i = sin θdθ dφYlm
1
(θ, φ)Ylm
2
2
(θ, φ) = δl1 l2 δm1 m2 (2.87)
0 0

and the completeness relation

∞ X
X l
ψ(θ, φ) = clm Ylm (θ, φ) (2.88)
l=0 m=−l

where the coefficients are the projections of the smooth (wave) function onto the |lmi eigenstates
Z π Z 2π
clm = hlm|ψi = sin θdθ dφYlm∗ (θ, φ)ψ(θ, φ) . (2.89)
0 0

2.10 Inversion symmetry: parity of spherical harmonics

In three dimensions, inversion corresponds to, r → −r. In terms of Cartesian coordinates inversion
corresponds to x → −x; y → −y; z → −z. In terms of spherical polar coordinates inversion corresponds

25
to

r → r (2.90)
θ → π−θ (2.91)
φ → φ+π . (2.92)

The hydrogen wave functions either remain unchanged (even parity), or change sign (odd parity) under
inversion11 . The parity is determined by the value of (−1)l : +1 for even parity; −1 for odd parity. It
follows that if l is even, parity is even; if l is odd, parity is odd.
The parity of a wave functions depends on the inversion property of the θ, φ part, i.e. the spherical
harmonic. Noting that for the inversion indicated above,

eimφ → (−1)m eimφ (2.93)

cos θ → − cos θ (2.94)
l
Pl (cos θ) → Pl (− cos θ) = (−1) Pl (cos θ) (2.95)
Plm (cos θ) → Plm (− cos θ) = (−1)l+m Plm (cos θ) (2.96)
Ylm (θ, φ) → (−1)l Ylm (θ, φ) . (2.97)

2.11 Visualisation of Spherical Harmonics

The standard convention is to label the spherical harmonics, sometimes denoted orbitals, with a letter
corresponding to the orbital angular momentum quantum number l.

l = 0, 1, 2, 3, 4, 5, . . . (2.98)
= s, p, d, f, g, h, . . . (2.99)

The spherical harmonics are complex valued functions depending on two variables and hence difficult
to plot. We thus show the probability density |Ylm (θ, φ)|2 in Fig. 8.

2.12 Summary
Let us briefly recapitulate the main steps towards the solution of two particles in a central potential
before applying it to the hydrogen atom.

1. We started with the Hamiltonian, Eq. (2.2)

p̂21 p̂2
Ĥsys = + 2 + V (|r̂1 − r̂2 |) , (2.100)
2m1 2m2
and chose a more convenient coordinate system going to centre of mass and relative coordinates.

2. The centre of mass energy eigenvalue was straightforward to solve in terms of a free-particle
wave function (See Eq. (2.20)).
1
ψCM (R) = 3/2
eiP·R/~ (2.101)
(2π~)
11
The usefulness of this symmetry of the hydrogen wave functions will become apparent later.

26
 (a) l = 0(s), m = 0
(b) l = 1 (p), m = 0, ±1

(c) l = 2 (d), m = 0, ±1, ±2

Figure 8: Spherical Harmonics

3. The Hamiltonian of the relative motion is given by Eq. (2.9)

~2 2
Ĥrel = − ∇ + V (r) (2.102)
2µ
with the reduced mass µ. We then studied orbital angular momentum and showed in Eq. (2.69)
that the Hamiltonian can be written as
~2 1 ∂ L2
   
2 ∂
Ĥrel = − r − 2 2 + V (r) (2.103)
2µ r2 ∂r ∂r ~ r
and the wave function can thus be further separated in a radial and an angular part
ϕ(r, θ, φ) = R(r)Y (θ, φ) . (2.104)
The solution of the angular part led us to spherical harmonics Ylm (θ, φ) with discrete values for
l and m = −l, −l + 1, . . . , l − 1, l.
4. We are left with the energy eigenvalue equation for the radial component R(r), which can be
obtained from Eq. (2.69) by replacing L̂2 with its eigenvalue l(l + 1)~2
~2 d ~2
   
2 d
− r + V (r) + l(l + 1) R(r) = ER(r) (2.105)
2µr2 dr dr 2µr2
where l = 0, 1, 2, . . . . Note that the energy eigenvalue equation resembles a one-dimensional
eigenvalue equation with an effective potential
~2
Vef f (r) = V (r) + l(l + 1) , (2.106)
2µr2
where the second term is called the centrifugal barrier.

27
3 Hydrogen Atom
After our discussion of a general central potential in the last section, we now study the hydrogen atom
and apply our general result for the radial eigenvalue equation in Eq. (2.105) to the hydrogen atom.
The potential V (r) is given by Coulomb potential defined in Eq. (2.1) with Z = 1

Ze2 −Zα~c
V (r) = − = . (3.1)
4π0 r r

3.1 Solution of the radial equation

We will outline the solution to the radial equation. We first rewrite Eq. (2.105) in terms of dimen-
sionless quantities
r
ρ= . (3.2)
a
and obtain
d2 R 2 dR 2µa2 µZe2 2a l(l + 1)
 
+ + E+ − R=0. (3.3)
dρ2 ρ dρ ~2 4π0 ~2 ρ ρ2
Thus the characteristic length and energy scales are given by

a 4π0 ~2 1
≡ 2
= (3.4)
~c µZe ~c Zαµc2
E  a 2
−γ 2 ≡  2  = 2µc2 E (3.5)
~ ~c
2µa2

and the energy eigenvalue equation reads

d2 R 2 dR
 
2 2 l(l + 1)
+ + −γ + − R=0. (3.6)
dρ2 ρ dρ ρ ρ2

3.1.1 Asymptotic Solution

We first look at the two limiting cases for small and large ρ. In the limit of large ρ → ∞ we can
neglect all terms proportional to ρ−1 and ρ−2 and Eq. (3.6) becomes

d2 R
− γ2R = 0 . (3.7)
dρ2

There are two possible solutions R(ρ) = e±γρ . The requirement that the probability should be
normalised to 1 only allows
R(ρ) ∼ e−γρ (3.8)
for large ρ. In the limit of small ρ we can neglect −γ 2 + 2/ρ vs. the centrifugal barrier term l(l + 1)/ρ2
and study
d2 R 2 dR l(l + 1)
+ − R=0. (3.9)
dρ2 ρ dρ ρ2
This equation can be solved by a polynomial ansatz R ∼ ρq with the characteristic equation

0 = q(q − 1) + 2q − l(l + 1) = q(q + 1) − l(l + 1) . (3.10)

28
There are two solutions q = l and q = −l − 1. We can again discard the solution with negative q
because it diverges for ρ → 0. In order to obtain the full solution of the radial eigenvalue equation,
we will look for solutions with
R(ρ) = ρl e−γρ H(ρ) (3.11)
with some arbitrary function H.

3.1.2 Series solution

Our ansatz leads to the following equation for H

d2 H dH
ρ + 2(l + 1 − γρ) + 2(1 − γ − γl)H(ρ) = 0 . (3.12)
dρ2 dρ
This equation can be solved by a power series, which is generally possible if the solutions are smooth
∞
X
H(ρ) = cj ρj . (3.13)
j=0

Inserting it in Eq. (3.12) and comparing the coefficients for each power of ρ, we obtain a recursion
relation
2γ(1 + j + l) − 2 j→∞ 2γ
cj+1 = cj −→ cj . (3.14)
(j + 1)(j + 2l + 2) j
For large ρ, the function H(ρ) behaves like
X1
H(ρ) ∼ (2γρ)j = e2γρ (3.15)
j!
j

and thus diverges for ρ → ∞. Thus the series has to be finite and there is a jmax such that cj = 0 for
j > jmax ,
2γ (1 + jmax + l) − 2 = 0 (3.16)
and we define the new principal quantum number
1
n ≡ jmax + l + 1 = . (3.17)
γ
As jmax and l are non-negative we find that n is a positive integer n = 1, 2, 3, . . . . Turning the
argument around, for a given n, the angular momentum quantum number has to satisfy

0 ≤ l = n − jmax − 1 ≤ n − 1 . (3.18)

The energy of the different quantum states does not depend on l or ml and is given by
2
Ze2 Z 2 α2 µc2 1

1 µ
En = − 2 = − (3.19)
2n 4π0 ~2 2 n2
and the radial solutions are
( 3 )1/2  l
2Z (n − l − 1)! −Zr/na0 2Zr
Rnl (r) = − e L2l+1
n+l (2Zr/na0 ) (3.20)
na0 2n[(n + l)!]3 na0

29
 Radial wave function
 3/2
R10 (r) = 2 aZ0 e−Zr/a0
 3/2 h i
R20 (r) = 2 2aZ0 1 − 2aZr
0
e−Zr/2a0
 3/2
Zr −Zr/2a0
R21 (r) = √13 2aZ0 a0 e

Table 2: Several radial wave functions. More can be found in Table 8.1
of Quantum Mechanics by D.H. McIntyre, p262.

where L are polynomials (called associated Laguerre polynomials) and a0 is the Bohr radius. These
radial functions are normalised according to the condition
Z ∞
r2 dr|Rnl (r)|2 = 1 (3.21)
0

allowing us to identify
P (r) = r2 |Rnl (r)|2 (3.22)
as the radial probability density. Examples of low order radial wave functions are given Table 2.

3.2 Full hydrogen wave functions

The full normalised hydrogen wave functions are the product of a (normalised) radial function and a
(normalised) spherical harmonic,

hrθφ|nlml i = ψnlml (r, θ, φ) = Rnl (r)Ylml (θ, φ) (3.23)

Wave functions for several low hydrogen levels are given in Table 3.

Full hydrogen wave functions

 3/2
ψ100 (r, θ, φ) = √1π aZ0 e−Zr/a0
 3/2 h i
ψ200 (r, θ, φ) = √1π 2aZ0 1 − 2aZr
0
e−Zr/2a0
 3/2
Zr −Zr/2a0
ψ210 (r, θ, φ) = 2√1 π 2aZ0 a0 e cos θ
 3/2
Zr −Zr/2a0
ψ21±1 (r, θ, φ) = ∓ 2√12π 2aZ0 a0 e sin θe±iφ

Table 3: Full wave functions for several low hydrogen levels. More can be found in Table 8.2 of
Quantum Mechanics by D.H. McIntyre, p264.

30
3.3 Radial Probability Density
Looking at the full wave function, we can again recover the radial probability density by integrating
over the angular part of the probability density
Z Z π Z 2π
2 m 2 2 2
P (r) = r dΩ|Rnl (r)Yl (θ, φ)| = r |Rnl (r)| sin θdθ dφdφ|Ylm (θ, φ)|2 = r2 |Rnl (r)|2 (3.24)
0 0

where we used the orthonormality of the spherical harmonics, which yields the same result as Eq.
(3.22). The value of r for which the radial probability density is a maximum i.e. the value of r for
which12
d d
P (r) = (r2 [Rnl (r)]2 ) = 0 (3.25)
dr dr
The average radial position is given by the expectation value of r, given by
Z ∞ Z ∞
hri = rP (r)dr = r3 dr[Rnl (r)]2 (3.26)
0 0

3.4 Hydrogen Energy Levels

The hydrogen energy levels are given by Eq. (3.19)

α2 µc2 1
En = − for n = 1, 2, 3, . . . . (3.27)
2 n2
where we again make use of the fine structure constant

e2 1
α= ≈ (at low energy) . (3.28)
4π0 ~c 137
The energy levels are negative and the zero energy is defined as the energy when the electron and the
nucleus are infinitely separated. The generalisation to different atoms with Z protons is straightforward

Z 2 α2 µc2 1
En = − . (3.29)
2 n2
The reduced mass of the hydrogen atom is given by
me mp
µ= ≈ me = 511keV/c2 (3.30)
me + mp

and the energy levels are numerically given by

1 1
En = − 2
13.6eV = − 2 Ryd (3.31)
n n
where we defined the unit of one Rydberg, Ryd= 13.6eV. The other characteristic quantity is the
Bohr radius
~c
a0 = = 0.0529nm = 0.529Å (3.32)
αµc2
12
Of course there may be more than one maximum, and this condition will also give positions of minima.

31
3.5 Degeneracy
A state of a hydrogen-like atom is characterised by three quantum numbers: n, l, ml . The energy
of a state depends, however, upon the value of n only (Equation 3.19). For a given value of n, the
quantum number l can have the values n − 1, n − 2, . . . , 0 and for each l there are 2l + 1 possible values
of ml : l, l − 1, l − 2, . . . , −(l − 1), −l. When we take into account the two possible spin states for the
electron (|+i and |−i) there are 2 different quantum states with the same orbital angular momentum.
The number of states with the same energy is therefore
n−1
X
2 (2l + 1) = 2n2 (3.33)
l=0

Thus energy levels are degenerate with a degeneracy of 2n2 . We will see soon that this is only
approximately true, as may of these “degenerate” levels differ in energy by small amounts (called fine
structure).

3.6 Superposition
If an atom is in energy level n will it necessarily be in one of the 2n2 states with this energy? The
answer is no - it will be in a random superposition of all states with that same energy. If an experiment
is designed to measure the component of angular momentum in a particular direction, a value for
that component will be obtained indicating that the atom is then in the state corresponding to that
particular value of ml . The act of measurement is said to cause the wave function to “collapse” from
the random superposition to one with a definite value of ml . This scenario is known as the Copenhagen
interpretation of quantum mechanics.

3.7 Emission Spectrum

The hydrogen spectrum is characterised by the energy differences between the different energy levels
 
α2 µc2  1 1 
∆Ef i = |Ef − Ei | =  − 2 . (3.34)
2  n2i nf 

The energy of the photon which is emitted or absorbed in transitions between these energy levels is
given by
hc
Eγ = ~ω = hf = = ∆Ef i . (3.35)
λ
The different spectral lines fall into different bands with closely spaced lines depending on the final
energy level. See Fig. 9. Transitions to the ground state constitute the Lyman series, where the
transition from the second energy level to the ground state is denoted as Lyman-α, the one from the
3rd energy level to the ground state Lyman-β, etc. The series from higher levels down to n = 2 are
denoted Balmer series and the ones to the third level Paschen series.

32
Figure 9: Hydrogen energy levels and emission spectrum

33
4 Transition probability
Spontaneous emission is the most familiar process involving radiation and atoms: an atom in an excited
state |2i undergoes a transition to a lower energy state |1i by emitting a photon. The frequency of
the radiation emitted ν21 is given by
hν21 = E2 − E1 (4.1)
Such a transition is characterised by a transition probability, A21 , which is the probability per unit
time of an atom in state |2i undergoing the transition to state |1i. If the excited state density is N2 ,
then the spontaneous emission power per unit volume is

P21 = N2 A21 hν21 (4.2)

4.1 Radiative lifetime

As a result of spontaneous emission, the population N2 of the upper level |2i as a function of time is
governed by the equation
dN2
= −A21 N2 (4.3)
dt
for which the solution is
N2 (t) = N2 (0)e−A21 t (4.4)
where N2 (0) is the population at t = 0. Thus the upper level decays due to emission of radiation with
a lifetime given by
1
τ21 = (4.5)
A21
For example, the transition probabilities and radiative lifetimes for the first few lines of the Balmer
series are given in Table 4.13

line λ (nm) A21 (s−1 ) τ21 (ns)

Hα (n = 3 → 2) 656 4.41 × 107 23
Hβ (n = 4 → 2) 486 8.42 × 106 119
Hγ (n = 5 → 2) 434 2.53 × 106 395
Hδ (n = 6 → 2) 410 9.73 × 105 1030

Table 4: Transition probabilities and upper level lifetimes for the first 4 lines of the hydrogen Balmer
series

Note that if transitions to more than one lower energy level are possible the radiative lifetime of
the excited state will be determined by the sum of the relevant transition rates,
1
τ2 = . (4.6)
ΣA2i
13
These values are for allowed transitions, in contrast to forbidden transitions for which values of A21 are many orders
of magnitude lower. We will soon see what determines whether a transition is allowed or forbidden.

34
4.2 Absorption and stimulated emission
Absorption and stimulated emission differ from spontaneous emission in that they occur only in the
presence of radiation at the transition frequency ν21 , and their transition rates are dependent upon
the intensity of the radiation (See Figure 10). These processes are characterised by the coefficients
B12 (absorption) and B21 (stimulated emission), where the corresponding transition rates are given
by B12 ρ(ν) and B21 ρ(ν), where ρ(ν) is the energy density of the radiation at the transition frequency
ν21 = (E2 − E1 )/h. The coefficients A21 , B21 and B12 are known as Einstein coefficients. They are

Figure 10: (a) absorption; (b) stimulated emission; (c) spontaneous emission

related via the Einstein relations, which are given by

B12 A21 8πhν 3
=1 = . (4.7)
B21 B21 c3
As eigenfunctions contain all the information we can know about a state, it is to be expected that
the Einstein coefficients will depend upon the eigenfunctions of the upper and lower levels. This will
be discussed in a later section. Note however that we need find this relationship for only one of the
coefficients and the other two can be found using the Einstein relations. The derivation of the Einstein
relations is given in Appendix E.

4.3 Allowed & forbidden transitions: selection rules

Transitions are called allowed or forbidden depending on whether or not they satisfy selection rules.
Selections rules are conditions on the change in quantum numbers between the two levels in question,
and are a consequence of the conservation of angular momentum during emission or absorption of a
photon. If the section rules are obeyed the transition is allowed; if not, the transition is forbidden.
The probability for a transition between two energy levels is given by the matrix element

hnf lf mf |Hint |ni li mi i (4.8)

where Hint is some Hamiltonian which describes the interaction leading to the transition. The interac-
tion Hamiltonian of an electric dipole interaction describes the interaction of atoms, molecules,. . . with
light, like lasers. It is given by
Hint = −d · E (4.9)
where E is the electric field and d = −er is the atom’s dipole moment. Absorption and emission of
light can be described in this way.
Selection rules are a consequence of the fact that angular momentum (in addition to energy) must
be conserved during absorption or emission of a photon. Photons have spin angular momentum (s,

35
quantum number 1), so when a photon is absorbed or emitted by an atom the angular momentum of
the atom (L, quantum number l) will change. Ignoring electron spin,

L → L0 = L + s (4.10)

The spin angular momentum quantum number of the photon is s = 1, and thus ms = 0, ±1. The
final angular momentum quantum numbers are l0 = l + 1, l, l − 1 and m0 = m + 1, m , m − 1. See
the discussion of addition of angular momentum in Sec. 8.2. From this we would conclude that when
a photon is absorbed or emitted the change in l is ∆l = 0, ±1. For an electric dipole transition
the transition probability for the case ∆l = 0 is, however, identically zero. We can understand this
in terms of the parity of the upper and lower states. The dipole transition probability involves the
integral
h2|r|1i (4.11)
As r has odd parity, the integral will be zero if the upper and lower states have the same parity. As
the parity of a state is determined by (−1)l , ∆l = 0 corresponds to a forbidden transition. So we
conclude that for electric dipole radiation the selection rule for orbital angular momentum quantum
number is
∆l = ±1 (4.12)
It should be noted that the concept of a forbidden transition applies to dipole transitions. Emission
and absorption of photons can still occur due to higher multipole interactions (with much smaller
transition probabilities). Examples of such transitions are associated with upper and lower states of
the same parity and involve emission and absorption of two photons.
Summarising the selection rules for transitions between hydrogen states, characterised by the
quantum numbers are n, l, ml as discussed in the last section, are

∆l = ±1 (4.13)
∆ml = 0, ±1 (4.14)

with no restriction on ∆n.

Allowed transitions typically have transition probabilities ∼ 10−7 –10−8 s−1 . For forbidden transi-
tions, transition probabilities are usually many orders of magnitude smaller. Forbidden transitions are
not necessarily unimportant. For example, our knowledge of the distribution of atomic hydrogen in
our and nearby galaxies, is based on emission from a forbidden transition in hydrogen, the 21cm line in
hydrogen, which we will discuss in one of the following sections. In this case, intensities are measurable
as the large amount of neutral hydrogen compensates for the very low transition probability.

4.4 Metastable levels

If a state has no allowed transitions to lower levels, the lifetime of that state will be much longer than
would be the case if there were one or more allowed transitions to lower levels. Such states are called
metastable states. In discharges, for example, the populations of such states can build up to levels
comparable to that of the ground state.14 We shall see later that metastable states play a crucial role
in the population inversion mechanism of the He-Ne laser.

14
Electron-atom collisions still provide a means of transition between states and ultimately limit how large the
metastable state population becomes.

36
5 Magnetic moments, gyromagnetic ratio, ESR, NMR
An electron has magnetic dipole moments associated with both its orbital and spin angular momenta.
For orbital motion we can find the relationship between magnetic moment and orbital angular mo-
mentum by considering an electron in a circular orbit of radius r. The angular momentum is given
by

L = mvr (5.1)
The motion of the electron constitutes a current loop of area πr2 . The charge e completing an
orbit in a time 2πr/v corresponds to an equivalent current of I = ev/(2πr). The magnetic moment
of the current loop is, therefore:
ev e µB
µl = IA = πr2 = L= L (5.2)
2πr 2m ~
where µB = e~/2m = 9.27 × 10−24 amp m2 is the Bohr magneton. In terms of vectors this relationship
is
gl µB
µl = − L (5.3)
~
where gl , the gyromagnetic ratio, is equal to unity.
As you are aware, the electron also have a magnetic moment associated with its spin, with a similar
relationship,
ge µB
µs = − S (5.4)
~
where the gyromagnetic ratio in this case is an intrinsic property of the electron and has a value very
close to 2.15 . As |L|/~ ∼ 1 and |S|/~ ∼ 1, for both orbital angular momentum and spin µ ∼ µB .
We note for future reference that the proton also has an intrinsic magnetic moment,
gp µN
µp = I (5.5)
~
where I is the spin angular momentum of the proton (I = 1/2), µN = (me /mp )µB = 5.05×10−27 JT−1
is the nuclear magneton, and gp = 5.585. The neutron also has spin of I = 1/2 and a gyromagnetic
ratio of gn = −3.826.16

5.1 Electron spin resonance (ESR)

A free electron in a magnetic field has two energy levels due to the two possible orientations of its
spin, and hence of its magnetic moment, with respect to the magnetic field. The energies of the levels
are given by
µB µB
E = −µs .B = ge S · B = ge Sz B = ge µB ms B ≈ ±µB B (5.6)
~ ~
since ms = ±1/2, and ge ≈ 2.
Radiation that satisfies the condition
15
We will see later that ge = 2.00232 . . .
16
It may surprise you that the chargeless neutron has a magnetic moment. Like the proton, the neutron consist of
three quarks: their charges add to zero, but their magnetic moments do not cancel.

37
 hν = 2µB B (5.7)
will cause transitions in both directions between the two levels. This is called electron spin resonance
(ESR). Electrons which are effectively free can be found in radicals.17 In contrast to molecules where
electrons usually have paired electron spins, radicals are characterised by unpaired electrons. For a
field of 1 Tesla, ν = 28 GHz (microwave); for a field of 1 mT ν = 28 MHz (radiofrequency). The
width of the resonance depends upon the internal magnetic field distribution, which depends upon the
atomic environment of the electron; thus ESR spectra provide information about molecular structure
in situations where there is unpaired electron spin to act as a probe.

5.2 Nuclear magnetic resonance (NMR)

As noted earlier, protons have spin I = 1/2 and as a consequence they also have a magnetic moment
given by
µN
µp = gp
I (5.8)
~
In a magnetic field the proton spin has two orientations such that the components of the spin
parallel to B are mI ~ where mI = ±1/2. The energies of the two states are therefore given by
µN µN
E = −µp .B = −gp I · B = −gp Iz B = −gp µN mI B = ∓2.79µN B (5.9)
~ ~
Nuclear magnetic resonance occurs when electromagnetic radiation causes transitions between
these two levels. The condition for NMR in hydrogen is, therefore,

hν = 5.58µN B (5.10)
For magnetic fields of around 1 Tesla, the resonance frequency is in the radiofrequency range (42.5
MHz).18 Hydrogen atoms in different chemical environments are subject to different internal magnetic
fields, producing resonances at slightly different frequencies. Thus NMR in hydrogen is particularly
useful for studies of the structure of organic molecules. It is also the basis of the medical imaging
technique Magnetic Resonance Imaging (MRI) which produces images where the image contrast is
determined by proton density or various relaxation times. As these vary with the chemical environment
of the protons, MRI produces images with good contrast between different soft tissues.

5.3 Fine and hyperfine structure

The energy levels obtained by solving the energy eigenvalue equation for a Coulomb potential agree
with those given by Bohr’s theory. We have seen, however, that the levels are degenerate: there
are two or more different quantum states corresponding to each energy level. However, the electron
experiences a magnetic field due to its motion relative to the nucleus, and as a result there will be
additional energy associated with its magnetic moment in this magnetic field. The additional energy
17
Ionising radiation can rupture molecules to produce radicals. The use of ESR to detect enhanced radical content is
the basis of a dating technique for artefacts that have been exposed to environmental ionising radiation. When applied
to tooth enamel ESR can be used to measure human exposure to ionising radiation after the event.
18
As µB ∼ 103 µN , for a given magnetic field the resonant frequency for ESR is about 1000× that for NMR.

38
can vary among the degenerate states so that some of the degeneracy is removed. The resulting
different energy levels are referred to as the fine structure of the hydrogen energy levels.19
A similar, but much smaller effect occurs due to the magnetic field of the magnetic moment of
the hydrogen nucleus. The resulting different energy levels in this case are referred to as hyperfine
structure.

19
The term fine structure is also used to refer to the splitting of spectral lines into several closely-spaced lines as a
consequence of the fine structure of the energy levels.

39
6 Perturbation Theory
Most of the times we can not solve problems in quantum physics exactly, like for the hydrogen atom
and the harmonic oscillator. Thus we have to develop a technique to approximately solve problems.  
Suppose we know the exact energy eigenvalues and eigenstates for a Hamiltonian Ĥ0 : En0 and n0
respectively. We can obtain approximate solutions for the perturbed Hamiltonian Ĥ0 + Ĥ 0 provided
the effects of the extra term are small.
We have to separately discuss the case of quantum states with degenerate energies. In the following
section we will focus on non-degenerate energies and only briefly comment on the case with with
degenerate energies.

6.1 Non-degenerate Perturbation Theory

We write our Hamiltonian as a sum
Ĥ = Ĥ0 + λĤ 0 , (6.1)
where we introduced a parameter λ to keep track of the size of the perturbation. The energy eigenvalue
equation is given by
En |ni = Ĥ |ni = Ĥ0 |ni + λĤ 0 |ni . (6.2)
We can expect that the eigenstates and eigenvalues will change by small amounts only. Thus we
can systematically expand the eigenstates |ni and eigenvalues En around the eigenstates n
 (0) and
(0)
eigenvalues En of the Hamiltonian Ĥ0 .
  
|ni = n(0) + λ n(1) + λ2 n(2) + . . .
  
(6.3)
(0) (1) (2)
En = En + λEn + λ2 En + ... , (6.4)

where  E  E
En(0) n(0) = Ĥ0 n(0) . (6.5)
 

Writing down the eigenvalue equation for the perturbed situation,

   E  E     E  E 
Ĥ0 + λĤ 0
 (0)
+ λ n(1) + . . . En(0) + λEn(1) + . . .
 (0)
n = n + λ n(1) + . . . (6.6)
 

Collecting all terms of order λ0 , we find Eq. (6.5). At order λ1 , we find the equation which determines
the leading correction due to the Hamiltonian Ĥ 0
 E  E  E  E
Ĥ0 n(1) + Ĥ 0 n(0) = En(0) n(1) + En(1) n(0) (6.7)
   


As the eigenstates of the Hamiltonian Ĥ0 form a complete set of states {m(0) }, i.e. a basis, we

 (1)   (0) 
can express the state n in terms of the m
 E X  E
 (1)
n = am m(0) (6.8)

m

and thus X  E  E X  E  E
(0)
Em am m(0) + Ĥ 0 n(0) = En(0) am m(0) + En(1) n(0) . (6.9)
   
m m

40
 
Multiplying Eq. (6.9) by n(0) , we find

D E
En(1) = n(0) |Ĥ 0 |n(0) (6.10)

That is, the change in the energy of an eigenstate due to the perturbation is equal to the expectation
value of the perturbation with respect to that eigenstate. We can use this result from perturbation
theory to find the effects on energy levels of an atom of interactions associated with angular momentum
which were left out of the operator in Schrödinger’s equation. It can also be used to find the effects
on energy levels due to external electric and magnetic fields.  
On the other hand, if we multiply Eq. (6.9) by an arbitrary eigenstate p(0) unequal to n(0) of
 

the Hamiltonian Ĥ0 ,

X D E D E X D E D E
Em(0)
am p(0) |m(0) + p(0) |Ĥ 0 |n(0) = En(0) am p(0) |m(0) + En(1) p(0) |n(0) (6.11)
m m
D E
0
Ep(0) ap + p (0)
|Ĥ |n (0)
= En(0) ap (6.12)

we can determine the coefficients in the expansion (6.8)

D E
m(0) |Ĥ 0 |n(0)
am = (0) (0)
. (6.13)
En − Em

Note that we did not determine an , the correction in direction n(0) . It is not determined by Eq.


(6.8), but the condition that the perturbed state |ni is correctly normalised, hn|ni = 1. To this
order in perturbation theory, we find an = 0. Hence the approximate eigenstate |ni to the energy
(0) (1)
En = En + En is given by  E X  E
|ni = n(0) + am m(0) . (6.14)
 
m6=n

Above we have considered the first order corrections to the energy and wave function when a system
is perturbed. The theory can, of course, be extended to higher order corrections. Intuitively, if the
(1) (0) 
perturbation is small we expect En  En and that n(0) is the dominant term in the perturbed


wave function.
This procedure can be be continued to an arbitrary order in λ, i.e. the perturbation Ĥ 0 .

6.2 Degenerate Perturbation Theory

The theory outlined above applies to non-degenerate eigenstates. For degenerate eigenstates the
theory is more complicated and we will not discuss it during this lecture course, but only state that
for degenerate energy eigenvalues, one has to consider the whole degenerate subspace and diagonalise
the perturbation Hamiltonian Ĥ 0 in the degenerate subspace. In practice, one has to determine the
matrix elements of the perturbation Hamiltonian in the degenerate subspace spanned by {|ni}, Hmn 0 =
D E
0
m|Ĥ |n and diagonalise this matrix.
The effect of the perturbation may be different for the various degenerate eigenstates and generally
lifts the degeneracy. Thus a common effect of a perturbation is to ‘remove the degeneracy’, partially
or completely.

41
7 Fine structure, spin-orbit coupling, Dirac theory, Lamb shift
The Schrödinger equation is non-relativistic, i.e. it applies for small velocities v  c. Relating the
Bohr energy (i.e. the energy levels of the H-atom) to the kinetic energy of an electron, we obtain
1 2 2 1 v p
α mc ∼ mv 2 ⇒ α∼ = . (7.1)
2 2 c mc
As α = 1/137 is small we find that the non-relativistic treatment of the hydrogen atom is justified,
but relativistic corrections will play a role in precision studies. They contribute to the fine structure
of the hydrogen energy levels. We can identify three contributions to the fine structure of the H levels:

1. relativistic correction

2. spin-orbit coupling

3. Lamb shift (contributes to fine structure, but not of relativistic origin)

The first two contributions are of similar order of magnitude and strictly speaking are not really
independent corrections. In 1928 Dirac deduced the relativistic equation for a electron in a Coulomb
field in a way that included the effect of the electron spin. This comprehensive relativistic treatment
incorporates the first two corrections in the list above. We will consider here simplified estimates of
the corrections (detailed treatments are given in McIntyre §12.2).

7.1 Relativistic correction to H energy levels

The total energy (kinetic + rest) for a relativistic particle is given by
r  p 2
p
2 4 2 2 2
E = m c + p c = mc 1 + (7.2)
mc
If the relativistic effect is small (as it is in this case), we can expand this equation to give
 
2 1  p 2 1  p 4
E = mc 1 + − + ··· (7.3)
2 mc 8 mc

Subtracting the rest mass energy we obtain for the kinetic energy up to the first order relativistic
correction,
p2 1  p 4
− mc2 (7.4)
2m 8 mc
We can make a rough estimate of the relative magnitude of the first order relativistic correction to
the ground state energy (n = 1) using p/(mc) ∼ α and ignoring numerical factors

1  p 4 1 1
mc2 · ∼ α4 mc2 2 2 ∼ α2 ∼ 10−4 (7.5)
8 mc |E1 | α mc

42
 Figure 11: Proton rotating around electron. (Taken from McIntyre pg. 389)

7.2 Spin-orbit coupling correction to H energy levels

In order to understand the spin-orbit coupling, we will first consider the classical problem of a proton
orbiting an electron as shown in Fig. 11. The magnetic field induced at the centre of the loop is given
by Biot-Savart law
µ0 I
B= (7.6)
2r
with the current I = ev/2πr, where v denotes the speed of the proton, which can be related to the
angular momentum of the electron (by considering the electron moving around the proton)
L
L = me vr ⇒ v= . (7.7)
me r
Hence the magnetic field at the centre of the loop is given by
µ0 eL e
B= = L. (7.8)
2r 2πme r2 4π0 me c2 r3
As the magnetic field B and the angular momentum point in the same direction, we can write it as a
vectorial equation
e
B= L. (7.9)
4π0 me c2 r3
Hence the energy of the spin magnetic moment of the electron in the magnetic field of the proton is
given by
∆E = −µs · B (7.10)
with µs = − ge~µB S. Thus the energy is given by

µB e2
∆E = ge S·B= S·L (7.11)
~ 4π0 m2e c2 r3
and the corresponding Hamiltonian in quantum physics
e2
Ĥso = S·L, (7.12)
4π0 m2e c2 r3
This term couples the spin S to the orbital angular momentum L.
The z-component of the spin and the orbital angular momentum do not commute with Ĥso , i.e.
[Ĥso , L̂z ] 6= 0, and thus are not preserved. They are no longer good quantum numbers which can be
used to describe the system.
However, the total angular momentum

J=L+S (7.13)

43
 Figure 12: LS coupling (taken from Wikipedia)

is a constant of the motion and commutes with the Hamiltonian.

Thus we expect the state of an atom to be characterised by its value of total angular momentum.
See Fig. 12 for an illustration of the addition of spin and orbital angular momentum. As is the case
for any angular momentum in quantum mechanics, J is quantised according to

J 2 = j(j + 1)~2 (7.14)

Jz = mj ~ (7.15)

where mj = j, j − 1, . . . , −j.
L and S always add to give J. As the magnetic field of L will induce a torque

τ =µ×B (7.16)

on the spin S, S will precess and vice versa. In summary, both L and S must be precessing about J.
As a result their magnitudes remain constant but not their z components. It follows that when the
interaction between the orbital and spin angular momenta is taken into account, a stationary state of
the hydrogen atom is characterised by the quantum numbers n, l, s, j, mj . Such states are referred to
as coupled states, represented in ket notation as20

|nljmj i (7.17)

in contrast to the uncoupled states |nlsml ms i. Given that J = L + S and s = 12 , the possible values
of j are
j = l + 1/2, l − 1/2 (7.18)
except if l = 0 in which case j = 1/2 only.
We can use
J2 = (L + S)2 = L2 + S2 + 2L · S (7.19)
to rewrite
e2 2 2 2


Ĥso = J − L − S . (7.20)
8π0 m2e c2 r3
Thus the energy correction depends upon the quantum numbers j, l and s. If we take r ∼ a0 and
L.S ∼ ~2 , a small amount of algebra shows that
20
We do not include the quantum number s as it is 1/2 for all states

44
 ∆E/E1 ∼ α2 ∼ 10−4 (7.21)
Therefore the relative shifts in energy levels due to spin-orbit interaction are of the order α2 ∼ 10−4 .
As the shifts vary with quantum number, not all the states with the same n will have the same energy,
i.e. the spin-orbit interaction will remove some of the degeneracy.

7.3 Dirac’s relativistic theory

We saw above that the relativistic and spin-orbit corrections to the hydrogen energy levels are of order
α2 . Strictly speaking, the spin-orbit correction is also a relativistic correction as it depends upon the
magnetic field experienced by the moving electron (the internal magnetic field B → 0 as v/c → 0).
The complete relativistic treatment of the hydrogen atom, incorporating the spin of the electron
was done by Dirac21 in 1928 (for more detail see Appendix D). His theory gives the fine structure
energy levels for hydrogen, and also a value of 2 for ge . The resulting energy levels are

α2 µc2 α2
  
1 3
Enj = − 1+ − (7.22)
2n2 n j + 1/2 4n
Thus we see that the energy levels depend upon the quantum numbers j and n but not l. The
degeneracy of these levels is equal to the number of different mj values for a given j (which is 2j + 1)
summed over the relevant l values.

7.4 Spectroscopic notation

The electron configuration for a particular state of hydrogen is given by nx where the letter x represents
the value of l where x = s, p, d, f, g, . . . corresponds to l = 0, 1, 2, 3, 4, . . .. The spectroscopic notation
for a state has the form

n2s+1 Xj (7.23)
where X indicates the value of l: S, P, D, F, G, . . . correspond to l = 0, 1, 2, 3, 4, . . .
The ground state of hydrogen has an electron configuration of 1s and its spectroscopic notation
is 12 S1/2 . The configuration 2p corresponds to two states 22 P1/2 and 22 P3/2 . For hydrogen the
superscript is always 2; this indicates that there are two values of j for each value of l (except l = 0,
for which there is only one value).22

7.5 The Lamb shift

Although Dirac’s theory indicates that the 22 S1/2 and 22 P1/2 states have the same energy, an exper-
iment by Lamb and Retherford in 1947 confirmed speculation that these states do differ in energy,
by an amount now known as the Lamb shift and equal to 4.37462 × 10−6 eV (1057.77 MHz), which
is about one tenth of the fine structure splitting between the Dirac fine structure levels 22 P3/2 and
22 P1/2 . This experiment was an important stimulus to the development of quantum electrodynamics
(QED), which explains the Lamb shift in terms of the effect on the electron of the zero point energy of
21
See the recent biography: Graham Farmelo, The Strangest Man: The Hidden Life of Paul Dirac, Quantum Genius,
Faber & Faber, London 2009.
22
Without the n value, spectroscopic notation can also used for multi-electron atoms, where the quantum numbers
now refer to angular momenta of all the electrons.

45
 2D
5/2!
2D
3/2!
2P
n=3 3/2!
! 2S
1/2!
2P
1/2!

2P
3/2!

n=2
! Lamb shift
! 2S
1/2!
2P
1/2!

Figure 13: Fine structure of the n = 2, 3 hydrogen levels. The spacing between the fine structure
levels is greatly exaggerated compared to the spacing between principal levels. The allowed transitions
between n = 3 and n = 2, i.e. the fine structure Hα line, are also shown

the quantised electromagnetic field.23 QED also leads to a value for ge (= 2.00232) slightly different
from the Dirac value (2).
In his Nobel Prize lecture, Lamb quoted the following qualitative picture of the level shift.

The fluctuating zero-point electric field of the quantised vacuum acts on an electron bound
in a hydrogen atom. As a result, the electron is caused to move about its unperturbed
position in a rapid and highly erratic manner. The point electron effectively becomes a
sphere of a radius almost 10−12 cm. Such an electron in a hydrogen atom is not so strongly
attracted to the nucleus at short distances as would be a point electron. States of zero
orbital angular momentum like 22 S1/2 are therefore raised in energy relative to other states
like 22 P1/2 in which the electron has a smaller probability of being found near the nucleus.

The final picture then is that for each unique combination of the quantum numbers n, l, j there
is a distinct energy level with a degeneracy of 2j + 1. For a given n the sublevels are close together
with separations of the order of 10−4 of the separation between levels with different values of n. This
fine structure of the energy levels means that the spectral lines resulting from the transitions between
states of different n consist of several closely spaced lines. The fine structure of the n = 2 and n = 3
hydrogen levels and the transitions between them, which are the closely spaced components of the Hα
transition, are shown in Figure 13.

23
The 1955 Nobel Prize for Physics was divided equally between Willis Eugene Lamb (Stanford University) for his
discoveries concerning the fine structure of the hydrogen spectrum and Polykarp Kusch (Columbia University) for his
precision determination of the magnetic moment of the electron.

46
8 Hyperfine structure; Addition of angular momenta
8.1 Hyperfine structure
An electron in the ground state of hydrogen (n = 1) has zero orbital angular momentum (l = 0).
The total angular momentum of the electron is due, therefore, to its spin only. There will be energy
associated with the interaction between the magnetic moments µe of the electron and nucleus (proton)
µp .
0 2 ge µB gp µN
Ĥhf,1s = µ0 δ(r)S · I (8.1)
3 ~2
with the spin S (I) of the electron (proton). Each magnetic moment will experience a torque due to
the magnetic field of the other. There is no external torque on the atom, which in classical mechanics
would mean that the total angular momentum of the system is conserved. The corresponding state-
ment in quantum mechanics is that the system is characterised by a quantised value of total angular
momentum. If we use the symbol F for the total angular momentum we have

F=S+I (8.2)
where, using the usual quantisation rules for angular momenta,

F2 |F MF i = F (F + 1)~2 |F MF i (8.3)
Fz |F MF i = MF ~ |F MF i where MF = F, F − 1, . . . , −F (8.4)

where F is the quantum number associated with total angular momentum of the atom.
Using the coupled basis with the total spin F , we obtain for the correction to the energy of the
1s (n = 1, l = 0, ml = 0) state (Note that there is no orbital angular momentum (thus j = 12 ) and
we can use the uncoupled basis for spin and orbital angular momentum, i.e.|nlml F ISMF i.) due to
hyperfine splitting
D E
0
Ehf,1s = 100F ISMF |Ĥhf,1s |100F ISMF (8.5)
 
2 ge µB gp µN
= 100F ISMF | µ0 δ(r)S · I|100F ISMF (8.6)
3 ~2
 
2 ge µB gp µN 2 2 2
= 100F ISMF | µ0 (F − S − I )δ(r)|100F ISMF (8.7)
6 ~2
1
= µ0 ge µB gp µN (F (F + 1) − S(S + 1) − I(I + 1))|ψ1s (0)|2 , (8.8)
3
with the square of the spatial wave function |ψ1s (0)|2 ≡ h100|δ(r)|100i evaluated at the origin (the
position of the nucleus), i.e. the hyperfine splitting is proportional to the probability of the electron
to be in the nucleus.

8.2 Addition of angular momenta

What is the value for F ? The simplest way to answer this is to consider the component values,

Fz = Sz + Iz hence (8.9)
MF = ms + mI (8.10)

47
As ms and mI both have values of ± 21 the possible values of MF are

1 1 1 1 1 1 1 1
+ = 1; − = 0; − + = 0; − − = −1 (8.11)
2 2 2 2 2 2 2 2
Grouping these results as (1, 0, −1) and 0, we can see that they constitute all the MF values of F = 1
and F = 0 respectively. Thus the spin angular momenta of the electron and proton can add in two
possible ways: one gives F = 1 i.e. 12 + 12 , and the other F = 0 i.e. 12 − 12 . We are seeing here
a particular case of the addition of two arbitrary angular momenta. In general the total angular
momentum J of a coupled state |j1 mj1 i |j2 mj2 i can take values

J = j1 + j2 , j1 + j2 − 1, . . . , |j1 − j2 | (8.12)

and its z-component mJ is directly given by the sum of the z-components of the two coupled states

mJ = mj1 + mj2 (8.13)

Figure 14: The spin-flip’ transition between the hyperfine levels of the hydrogen ground state that
leads to emission at 21 cm (1420 Hz).

The F = 1 (upper) state corresponds to “parallel spins”24 ; the F = 0 (lower) state to “anti-parallel
spins”. As a consequence the transition between these two states is often called a “spin flip” transition.
The difference in energy between these states corresponds to 1420 MHz or in terms of wavelength, 21
cm. Emission on this line has allowed radio telescopes to map the neutral hydrogen distributions in
the Milky Way galaxy and other nearby galaxies. Note that as l = 0 for both upper and lower levels of
this transition, it does not satisfy the selection rule ∆l = ±1, and is therefore a forbidden transition.
Its transition probability is 2.9 × 10−15 s−1 , corresponding to a lifetime of 1.1 × 107 years25 .

24
This applies in the case of MF = 1. There are in total three different states MF = 0, ±1, which are symmetric under
the exchange of the spins.
25
Compare this with the allowed Hα transition, for which A = 4.4 × 108 s−1 and τ = 23 ns.

48
9 Zeeman and Paschen-Back effect
The Zeeman effect is the splitting of atomic energy levels by an externally applied magnetic field. As
a result spectral lines are split into several lines whose spacing increases with increasing magnitude of
the field. We will discuss the effect in detail for two limiting circumstances: when the applied field is
small relative to the internal magnetic field of the atom (∼ 1T ), and when it is much greater than the
internal magnetic field. The intermediate case, where the external and internal fields are comparable,
is much more complicated and will not be considered.
The quantitative treatment of the Zeeman effect is an example of time-independent perturbation
theory. Suppose the unperturbed atom (i.e. B = 0) has a Hamiltonian operator Ĥ0 , with eigenstates
 (0) 
n (0)
and corresponding eigenvalues En , where the zero superscript signifies the unperturbed atom.
Suppose the atom experiences a small perturbation (due to the applied magnetic field, i.e. B 6= 0),
represented by the new Hamiltonian Ĥ 0 + Ĥ 0 . This different Hamiltonian will, in general, have different
eigenstates and eigenvalues.
The essence of perturbation theory is the assumption that the perturbation is sufficiently small that
we can use the unperturbed eigenstates to estimate the small changes in the eigenvalues. Typically the
eigenvalues are shifted by small amounts, often in such a way that any degeneracy in the unperturbed
system is reduced. The simplest situation is when the eigenfunctions of Ĥ 0 are also eigenfunctions for
Ĥ 0 . In this case, which, as we will see is true for the Zeeman effect, the energy shift is given by the
expectation value of the perturbation Hamiltonian Ĥ 0 according to the unperturbed eigenstates. The
(0)
first-order correction to the unperturbed energy En is therefore given by26
D E
En(1) = n(0) |Ĥ 0 |n(0) (9.1)
We will consider three situations:

1. The Zeeman effect for a state without spin, i.e. a singlet state, for which S = 0. This is known
as the normal Zeeman effect as it can be explained classically.

2. The Zeeman effect for a state with spin in a weak magnetic field. Although this is called the
anomalous Zeeman effect it is in fact the general case of the effect of a weak external magnetic
field where in general S 6= 0. The normal Zeeman effect is a special case where S = 0.

3. The Zeeman effect for a state with spin in a strong magnetic field. This is called the Paschen-
Back effect.

9.1 Zeeman effect without spin

The perturbation to the Hamiltonian is the energy of the magnetic moment due to orbital angular
momentum in the external magnetic field. The additional energy is given by

µB B
E = −µL .B = Lz (9.2)
~
where we have used
µB
µL = − L (9.3)
~
26
In the terminology of perturbation theory this is called the zeroth approximation to the solution of the perturbed
problem. We will consider the first order correction only, but perturbation theory allows for higher order corrections.

49
 The perturbation Hamiltonian is therefore

µB B ˆ
Ĥ 0 =
Lz (9.4)
~
As the unperturbed eigenstates are |nlml i, the energy shifts are given by
 
(1) µB B ˆ
E = nlml | Lz |nlml = µB Bml hnlml |nlml i (9.5)
~
where we have used the fact that Lˆz |nlml i = ml ~ |nlml i. Noting that hnlml |nlml i = 1, the energy
shifts are

E (1) = µB Bml (9.6)

Thus the external magnetic field removes the degeneracy with respect to the magnetic quantum
number ml . Figure 15 shows the effect on a level corresponding to l = 1, resulting in separate energies
for ml = 1, 0, −1.

Figure 15: The effect of a weak external magnetic field on singlet states with l = 1 and l = 0. The
splitting of the allowed transition between the levels is also shown on the right.

In the absence of an external magnetic field, the transition between levels E2 and E1 produces
radiation at frequency ν0 = (E2 − E1 )/h. In the presence of an external field, the energy difference
becomes

(E2 − E1 ) → (E2 − E1 ) + µB B∆ml (9.7)

The selection rules are ∆ml = 0, ±1, resulting in the spectral line splitting into three: two lines
equally space about an undeviated line (see Figure 15):

µB B µB B
ν0 → ν0 − , ν0 ν0 + (9.8)
h h

9.2 Zeeman effect with spin

In this case the magnetic moment is given by27

S L
µ = µS + µL = −2µB − µB (9.9)
~ ~
and the perturbation Hamiltonian is given by

µB B
Ĥ 0 = −µ · B = −µz B = (2Ŝz + L̂z ) (9.10)
~
27
For simplicity we use ge = 2 instead of the more precise value ge = 2.00232 . . .

50
9.2.1 Weak magnetic field - anomalous Zeeman effect
The unperturbed eigenstates are |nljmj i, which are degenerate with respect to mj . The fact that,
due to ge ≈ 2, µ is not parallel to J = S + L is a complication. However the assumption of a weak
magnetic field allows the following simplification.
If the external field is weak compared to the internal magnetic field of the atom, µ precesses around
J much faster than it does about B. Thus the component of µ in the direction of B is to a good
approximation equal to its component parallel to J multiplied by the cosine of the angle between J
and B - see Figure 16. Hence the additional energy due to the external field is

Figure 16: Precession about an external magnetic field of total angular momentum J and total mag-
netic moment µ for the case of a weak magnetic field.

E = −µz B (9.11)

where
µ·J J·B
µz = × (9.12)
J JB
µB (L + 2S) · (L + S) Jz B
= − (9.13)
~ J JB
µB (L2 + 2S 2 + 3L · S)
= − Jz (9.14)
~ J2
With a little algebraic manipulation we get (using J 2 = L2 + S 2 + 2L · S)

µB B (3Jˆ2 + Ŝ 2 − L̂2 ) ˆ ĝL µB

Ĥ 0 = Jz = BJz (9.15)
~ 2Jˆ 2 ~
and following the same procedure used for the normal Zeeman effect, the first order correction to the
energy level is
 
(1) ĝL µB B ˆ
E = nljsmj | Jz |nljsmj = gL µB Bmj (9.16)
~

51
where gL evaluates to
j(j + 1) + s(s + 1) − l(l + 1)
gL = 1 + (9.17)
2j(j + 1)
is called the Landé g factor.
When a magnetic field is present energy levels that were degenerate with respect to mj split into
2j + 1 different levels. In addition to the selection rules already discussed there is now a selection rule
for mj :

∆mj = 0, ±1 (9.18)
The discussion above obviously applies to the hydrogen atom but it is also applicable to any atom
where the energy levels are determined by LS-coupling, i.e. the energy levels of a multi-electron atom
that are characterised by quantum numbers S, L, J, MJ 28 . See the discussion in Sec. 10.8.3.

Example
Consider the Zeeman effect for the 2 P1/2 states and the 2 S1/2 states in an alkali atom like sodium.
The gL values are 32 and 2 respectively. In the presence of a weak magnetic field, the energy levels
will split according to:

2
E2 → E2 + µB Bmj (9.19)
3
E1 → E1 + 2µB Bmj (9.20)

where in both cases MJ = ±1/2. Thus the split energy levels are:

E2 ± µB B/3 (9.21)
E1 ± µB B (9.22)

Applying the selection rules, there will be four transitions, from each of the two excited states to
each of the two ground states, and because the gL values are different there will be four different lines,
given by

2 µB B
ν0 ± (9.23)
3 h
4 µB B
ν0 ± (9.24)
3 h
Thus we see in Figure 17 that the Zeeman components consist of four lines in two groups of two,
symmetrically spaced on either side of the unperturbed line, ν0 , but note that there is no line at ν0 .
28
The convention is to use upper case for angular momenta quantum numbers of multi-electron atoms where the
angular momenta are the result of summation over two or more electrons. Lower case is used where only one electron is
involved - hydrogen and alkali atoms

52
Figure 17: The Zeeman splitting of the 2 S1/2 and 2 P1/2 levels of an alkali atom. Also shown are the
four allowed transitions,

9.3 Strong magnetic field - Paschen-Back effect

If the external magnetic field is stronger than the internal magnetic field of the atom µS and µL
precess about B much more rapidly than they do about J. Thus their interactions with the external
field are much greater than the effect of the spin-orbit interaction, which can be ignored as a first
approximation. We take, therefore, the unperturbed eigenstates to be the uncoupled states |nlsml ms i.
The perturbation Hamiltonian is given by,

µB
Ĥ 0 = −µ̂ · B = (L̂ + 2Ŝ) · B (9.25)
~
µB B
= (L̂z + 2Ŝz ) (9.26)
~
In the presence of a strong magnetic field, the first order correction to the unperturbed energy
level is
 
(1) µB B
E = nlsml ms | (L̂z + 2Ŝz )|nlsml ms = µB B(ml + 2ms ) (9.27)
~
The selection rules are ∆ms = 0, ∆ml = 0, ±1. This leads to spectral lines being split into three
components as in the normal Zeeman effect:

µB B µB B
ν = ν0 − , ν0 , ν0 + (9.28)
h h

9.4 Arbitrary magnetic field

For the first excited state of the optical electron of an alkali atom (np), Figure 18 shows how the
energy level spitting varies as the magnetic field increases from zero to large values. The energy level
structure varies from that of the anomalous Zeeman effect at low fields to that of the Paschen-Back
effect at high fields. The intermediate region where the external field is of comparable magnitude to
the internal field cannot be treated simply.

53
Figure 18: The energy levels of the np first excited state of the optical electron of an alkali atom as
a function of magnetic field. The two levels have spectroscopic notations 2 P3/2 and 2 P1/2 . (based
on Figure 93, Introduction to Modern Physics, F.K. Richtmeyer, E.H. Kennard and T. Lauritsen
(McGraw-Hill, New York, 1955)

9.5 Summary
If an external magnetic field causes a spectral line to split into three lines, two of which are equally
spaced about an undeviated line, it is either a case of the normal Zeeman effect (weak field, transitions
between singlet states) or the Paschen-Back effect (strong field). In either case (we do not need to
know which), the line spacing ∆ν is related to the magnetic field by

µB B
|∆ν| = (9.29)
h
In such circumstances, the line spacing can be used to measure the magnitude of the magnetic field.29
If the line is split into an even number of components, displaced symmetrically about the position of the
undeviated line (and there is no line in that undeviated position), it is a case of the anomalous Zeeman
effect, and the relation between the spacing between lines and the magnetic field is proportional to
µB B/h, where the constant of proportionality is of the order of unity but dependent upon the quantum
numbers of the energy levels involved.

29
This is how magnetic fields on the surface of the Sun in the vicinity of sunspots have been measured - it is a strong
field case. Information about the direction of the magnetic field is also available as the polarisation of the Zeeman lines
is determined by the relative direction between the line of observation and the direction of the field.

54
10 Identical particles, symmetry requirements, fermions and bosons
We now turn our attention to the helium atom. The two electrons are identical and as their wave
functions will, in general, overlap it is not possible to distinguish one electron from the other. The
wave functions which describe the helium atom must be consistent with the fact that identical particles
are indistinguishable. To see the consequences of this suppose the wave function ψ(r1 , r2 ) represents a
quantum system consisting of two identical particles whose positions are specified by r1 and r2 . If we
exchange the particles the system is now described by the wave function ψ(r2 , r1 ) . As the particles
are identical there can be no change in the physical reality of the system. Therefore, it must be true
that

|ψ(r1 , r2 )|2 ≡ |ψ(r2 , r1 )|2 (10.1)

From this we can conclude that
ψ(r2 , r1 ) = ±ψ(r1 , r2 ) (10.2)
i.e. the wave function must be either symmetric or antisymmetric with respect to interchange of the
particles. This requirement is a consequence of the indistinguishability of the particles. We will begin
our discussion of helium by considering first the description of the possible spin states for a system of
two spin- 21 particles.

10.1 Two spin- 12 particles

We can represent the two particle state by the ket |s1 s2 m1 m2 i. As the state of each electron is
indicated separately, we refer to this representations as using an uncoupled basis. However, as s1 and
s2 are always equal to 12 , we need only indicate the signs of the ms values ± 12 . Therefore, possible
configurations of the two particle system are,

|++i , |+−i , |−+i , |−−i (10.3)

The first and last of these are clearly symmetric, but the middle two are neither symmetric nor
antisymmetric. Note also that for these four states the total magnetic quantum numbers will be
MS = ms1 + ms2 = 1, 0, 0 and -1 respectively.
We can form symmetric and antisymmetric superposition states from |+−i and |−+i as follows,30

1
symmetric : √ [|+−i + |−+i] (10.4)
2
1
antisymmetric : √ [|+−i − |−+i] (10.5)
2
We now have three symmetric states (with MS = 1, 0, −1) and one antisymmetric state (with
MS = 0). The former are called triplet states, the latter a singlet state. This is not unexpected
since adding the individual spin angular momenta (s = 21 in each case) we obtain total spin quantum
numbers S = 1 (MS = 1, 0, −1) and S = 0 (MS = 0).
30
√
If each ket is normalised,the 1/ 2 factor ensures that the linear combination is also normalised.

55
 We can represent the three symmetric and one antisymmetric states in terms of total spin quantum
numbers (called coupled basis states) |SMS i as follows

|1 1i = |++i (10.6)
1
|1 0i = √ [|+−i + |−+i] (10.7)
2
|1 − 1i = |−−i (10.8)
1
|0 0i = √ [(|+−i − |−+i] . (10.9)
2
The state |1MS i is denoted triplet state and the state |00i singlet state. We can summarise the
symmetries associated with exchange of particles by introducing an exchange operator, P12 ,

P12 |1MS i = + |1MS i (10.10)

P12 |00i = − |00i (10.11)

To complete the description of the helium atom we need to include the spatial part of the wave
function |ψspatial i, such that the complete wave function is given by

|ψi = |ψspatial i |ψspin i (10.12)

We now consider the effect of symmetry requirements on the complete wave function.

10.2 Symmetric or antisymmetric?

As we have seen, the fact that particles are identical requires that systems of the particles be described
by symmetric or antisymmetric wave functions, but does not specify which. It is however a fact of
nature that particles can be divided into two groups:

bosons: particles with integer spin that have symmetric wave functions.

fermions: particles with half-integer spin that have antisymmetric wave functions.

As electrons, with spin 12 , are fermions31 the states of helium must be antisymmetric. As helium
states are the product of a spatial and spin part, as shown in Equation (10.12), antisymmetry may be
achieved in two ways:
 SA   S  A 
ψ = ψspatial ψspin (10.13)
helium
 AS   A  S 
ψ = ψ ψspin (10.14)
helium spatial

10.3 Two identical particles in one dimension

We can gain some insight into systems of two identical particles by considering a simple one-dimensional
system, an infinite square well of width L (x = 0 → L), which contains two non-interacting particles
31
Protons and neutrons, both with spin 12 are also fermions; photons with spin 1 are bosons, as is the Higgs boson
with spin 0; atoms may be bosons or fermions depending the total spin of their composite particles: for example the
hydrogen atom is a boson, j = 21 ,I = 21 → F = 1, 0.

56
with spin 12 . Suppose one particle is in energy eigenstate ϕna (energy: Ena ) and the other in energy
eigenstate ϕnb (energy: Enb ). Note that the combined states

ϕna (x1 )ϕnb (x2 ) and ϕna (x2 )ϕnb (x1 ) (10.15)

both have the same energy (Ena + Enb ) but neither satisfies either the symmetric or antisymmetric
requirement. However we can form linear combinations which do:

S 1
= ψnSa nb (x1 , x2 ) =


x1 x2 |ψspace √ [ϕna (x1 )ϕnb (x2 ) + ϕna (x2 )ϕnb (x1 )] (10.16)
2
A 1
= ψnAa nb (x1 , x2 ) =


x1 x2 |ψspace √ [ϕna (x1 )ϕnb (x2 ) − ϕna (x2 )ϕnb (x1 )] (10.17)
2

Specific example: ground state

 A 
If both particles are in the ground state i.e. na = nb = 1, it is obvious that ψspace is identically zero,
so the ground state can only be the product of a symmetric spatial function and an antisymmetric
spin function:  S
ψ11 |00i (10.18)
For an infinite well where the energy eigenfunctions are given by
r
2  nπx 
hx|ni = ϕn (x) = sin (10.19)
L L
S (x , x )|2 is shown in Figure 19.
the probability density function |ψ11 1 2

Figure 19: Probability density distribution for the ground state

Specific example: excited state

Suppose one electron is in the ground state and the other in the first excited state. The wave functions
for these states are, respectively

p
ϕ1 (x) = 2/L sin(πx/L) (10.20)
p
ϕ2 (x) = 2/L sin(2πx/L) (10.21)

57
 We now form the symmetric and antisymmetric spatial functions,

S 1
ψ12 (x1 , x2 ) = √ [ϕ1 (x1 )ϕ2 (x2 ) + ϕ2 (x1 )ϕ1 (x2 )] (10.22)
2
A 1
ψ12 (x1 , x2 ) = √ [ϕ1 (x1 )ϕ2 (x2 ) − ϕ2 (x1 )ϕ1 (x2 )] (10.23)
2
S (x , x )|2 and |ψ A (x , x )|2 are shown in Fig. 20. We can
The probability density functions |ψ12 1 2 12 1 2
see that, on average, the particles are closer together for the symmetric combination than for the
antisymmetric combination. This effect is called the exchange 32 We can quantify this by

interaction.
2

calculating the average distance between the two particles (x1 − x2 ) . In our specific example, we
find

p 0.20L symmetric spatial wave function

h(x1 − x2 )2 i = 0.32L distinguishable particles . (10.24)

0.41L antisymmetric spatial wave function


See McIntyre Sec. 3.2.4 for more details.

(a) symmetric excited state (b) antisymmetric excited state

Figure 20: Probability density distributions

10.4 Interacting Particles

So far we did not consider any interactions between the identical particles. However, in interesting
realistic situations, e.g. the two electrons in a helium atom, there are interactions between the identical
particles. Let us assume that the interaction only depends on the particle separation
Ĥ 0 = Vint (|x1 − x2 |) (10.25)
and
 (0)that
 it can be treated in perturbation theory. Hence the correction to the energy of the state
ψ is described by D E
E (1) = ψ (0) |Ĥ 0 |ψ (0) . (10.26)
32
The effect is also often said to be the result of an exchange force, but we need to realise that this can only be
an apparent force. It is a purely quantum consequence of the symmetry requirements and not related to any of the
fundamental force interactions.

58
As we argued previously the quantum state is a product of the spin and spatial part of the wave
function, i.e. D E
E (1) = ψspatial |Ĥ 0 |ψspatial hψspin |ψspin i . (10.27)

The spatial part of the ground state wave function has to be symmetric irrespective of the spin of the
particles. Hence we find
D E Z
(1)
E11 = ψ11 |Ĥ |ψ11 = dx1 dx2 ϕ∗1 (x1 )ϕ∗1 (x2 )Vint (|x1 − x2 |)ϕ1 (x1 )ϕ1 (x2 ) ≡ J11 .
S 0 S
(10.28)

In case of the first excited state, the spatial part of the wave function can be either symmetric of
antisymmetric. Defining the direct integral Jnm and the exchange integral Knm
Z
Jnm ≡ dx1 dx2 ϕ∗n (x1 )ϕ∗m (x2 )Vint (|x1 − x2 |)ϕn (x1 )ϕm (x2 ) (10.29)
Z
Knm ≡ dx1 dx2 ϕ∗n (x1 )ϕ∗m (x2 )Vint (|x1 − x2 |)ϕn (x2 )ϕm (x1 ) , (10.30)

we find for spin-0 bosons

(1)
E12 = J12 + K12 (10.31)

and for fermions

(1)
E12 = J12 ± K12 (10.32)

with a + (−) sign for a (anti-)symmetric spatial wave function.

We now apply these concepts to the helium atom.

10.5 Helium atom

The Hamiltonian operator for the relative motion of a helium atom is

~2 2 2e2 ~2 2 2e2 e2
Ĥ = − ∇1 − − ∇2 − + (10.33)
2m 4π0 r1 2m 4π0 r2 4π0 r12
where r12 = |r1 − r2 |. The last term is due to the electrostatic interaction between the two electrons
and is not a small correction to the non-interacting Hamiltonian.
Suppose we initially ignore the interaction term. In this case each electron has energy levels equal
to those of a “hydrogen atom with Z = 2”. Thus the zeroth order energy of the helium atom will be
 
(0) 1 1
Ena ,nb = −4Ryd + . (10.34)
n2a n2b

One of the electrons in helium is always in the ground state. The state with two electrons in the
(0)
excited state has energy E22 = −2Ryd, which is larger than the ground state energy of singly-ionised
helium E1∞ = −4Ryd. Hence the state with two excited electrons has a very short lifetime.
The interaction between the electrons will correspond to an increase in the energy. As this de-
pends upon the relative distance between the electrons it will depend upon the electron probability
distributions of each electron, which themselves depend upon their n and l quantum numbers. We
will follow McIntyre and call this increase

59
Figure 21: Energies and state vectors for the ground and first excited states of two identical particles
(taken from McIntyre Sec. 13.3 pg 426)

Jna la ,nb lb (10.35)

Noting that each state specified by na la , nb lb consists of a singlet (S = 0) and three triplet states
(S = 1), we now take account of the exchange interaction, which leads to the average distance between
the electrons in the singlet state being less than that for triplet states. Because of the electron-electron
electrostatic interaction this means that the singlet state energy will be higher than that of the triplet
states. Again following McIntyre we designate the energy change due to the exchange interaction as
(“+” refers to the singlet state; “−” refers to the triplet state),

± Kna la ,nb lb (10.36)

10.5.1 Ground state of helium

For both electrons n = 1, l = 0 and ml = 0 with an electron configuration 1s2 . As we have seen for
A
the 1D examples, it follows that ψspace = 0, i.e. there is no triplet state: the ground state of helium
consists of a singlet state only.

10.5.2 First excited state of helium

One electron remains in its ground state (n = 1, l = 0), the other in the first excited state (n = 2,
l = 1, 0) with corresponding electron configurations 1s2s and 1s2p. The effects of electron interaction
and the exchange interaction on the energy levels are summarised in Figure 22.

60
Figure 22: The effect of the electron-electron interaction and the exchange interaction on helium
energy levels. Reproduced from Figure 13.9, McIntyre p432.

10.5.3 Spin-orbit coupling in helium

When spin-orbit coupling is taken into account we find the triplet states have different energies. This
is discussed further in Section 10.8.2.

10.6 Pauli exclusion principle

Consider the possibility of the two electrons in a helium atom having the same set of quantum numbers,
i.e. of being in the same quantum state. As the ms values are the same, the spin wave function must
be symmetric, and therefore the spatial wave function must be antisymmetric, ψspace A . However if
A
the quantum numbers n, l, and ml are the same ψspace = 0, and as a result the total wave function
vanishes. It follows that it is impossible for the two electrons to be in the same quantum state. This
is in fact a statement of the Pauli Exclusion Principle:
No two fermions can be in the same quantum state33 (with the same quantum numbers).
The quantum states of fermions always have to differ in at least one quantum number.
If two particles are in the same quantum state, it is impossible to construct a two particle state, which
is anti-symmetric under the exchange of the two particles as it is required for fermions.
The fact that the ground state of helium consists of a singlet state only is an example of the Pauli
exclusion principle. As both electrons have n = 1, l = 0 and ml = 0, they must have different values
of ms : one + 21 , the other − 12 . The only possible value of MS is therefore zero. It follows that S can
only be zero, i.e. the ground state can only be a singlet state.
The Pauli Exclusion Principle is a consequence of the fact that electrons are fermions. There is
no equivalent principle for a system of bosons: e.g. a laser is a system of bosons (photons) all in the
same quantum state; Bose-Einstein condensation involves placing a collection of atoms with integer
spin (i.e. bosons) into the same quantum state.
33
This, in particular, applies to electrons in a multi-electron atom.

61
10.7 Bose-Einstein condensation
Hydrogen atoms in their ground state are bosons as they have integer spin: the proton has spin
I = 1/2, the electron has spin s = 1/2, so the total spin quantum number is F = 0, 1. For atoms at
temperature T , the de Broglie wavelength is given by
1/2
h2

h h
λ= ∼ ∼ (10.37)
p mvth 3mkT
where the root mean square speed has been used for vth . If the atoms are cooled until λ is greater
than the inter-atom spacing the wave functions of the atoms will overlap and the collection of atoms
must be treated as a single quantum system of indistinguishable bosons.
When this situation is reached by cooling the atoms, the atoms undergo a quantum mechanical
phase transition and form a Bose-Einstein condensate: a coherent cloud of atoms all occupying the
same quantum state; they can occupy the same quantum state because they are bosons.
Experiments since 1995 have achieved Bose-Einstein condensation with different atoms34 . For
hydrogen, temperatures ∼ µK are required. Atoms in the gaseous state can be trapped and cooled to
such temperatures using the techniques of laser cooling35 .

10.8 Multi-electron atoms

Consider an atom with atomic number Z: the nuclear charge is +Ze and there are Z electrons.
According to the Pauli exclusion principle, each quantum state can contain one electron only. For the
moment we will ignore spin-orbit coupling and specify a quantum state by the quantum numbers n,
l, s, ml and ms . Each value of n specifies a shell; each value of l specifies a subshell. Taking account
of the two possible values for ms (± 12 ), each full subshell contains 2(2l + 1) electrons and each shell
contains 2n2 electrons, as summarised in Table 5.

shell subshells 2n2

n=1 l = 0; two 2s electrons 2
n=2 l = 0; two 2s electrons 8
l = 1; six 2p electrons
n=3 l = 0; two 2s electrons 18
l = 1; six 2p electrons
l = 2; ten 2d electrons
.. .. ..
. . .

Table 5: Shell and subshell structure for a multi-electron atom

The total angular momentum of a full subshell is zero as it contains electrons with all possible
values of ml (l, l − 1, . . . , −l) and ms (± 21 ) the sum of which is clearly zero. As the component value
of the total angular momentum can only have the value of zero, the total angular momentum itself
must be zero. It follows that the total angular momentum of a multi-electron atom will be due to only
those electrons in unfilled subshells. We will look at a simple multi-electron example: alkali atoms.
34
See ”The Theory of Bose-Einstein Condensation of Dilute Gases”, K. Burnett, M. Edwards and C.W. Clark, Phys.
Today, 52 37 (Dec 1999)
35
See ”Cooling and Trapping Atoms”, W.D. Phillips and H.J. Metcalf, Sci. Am. 256 36 (Mar 1987)

62
10.8.1 Alkali atoms
The ground state of an alkali atom has one electron in an s state with all lower energy electrons in
full subshells. This s electron is called the optically active electron as the emission spectrum of the
atoms is almost entirely due to excitation of this electron and its subsequent radiative decay, while
the electrons in full subshells remain unexcited. For example. the ground state electron configuration
for sodium (Z = 11) is 1s2 2s2 2p6 3s; the first excited configuration is 1s2 2s2 2p6 3p.
As the total angular momentum of a full subshell is zero, the total angular momentum of the
electrons in an alkali atom is due to the optically active electron only. Unlike the electron in the
hydrogen atom, however, this electron does not experience a Coulomb potential, as the Coulomb
potential of the nucleus is modified due the combined effect of all the other electrons. Thus the energy
levels depend upon the values of the quantum number l, as well as n and j. 36
When spin-orbit coupling is taken into account the states are characterised by the quantum num-
bers n, l, j, mj ; the energy levels correspond to each unique combination of n, l, j, and their degeneracy
is 2j + 1, the number of different values for mj . The selection rules for allowed transitions are:

∆l = ±1 (10.38)
∆j = 0, ±1 (10.39)

Example
The ground state of sodium has an electron configuration 1s2 2s2 2p6 3s, for which the spectroscopic
notation is 2 S1/2 . For the first excited state the electron configuration is 1s2 2s2 2p6 3p, for which there
are two states with spectroscopic notations of 2 P3/2 and 2 P1/2 , as shown in Figure 23. These two
excited states have different energies (values of n and l are the same but j values are different), and
transitions from both of them to the 2 S1/2 ground state are allowed - they satisfy the selection rules.
The resulting two closely spaced spectral lines are the yellow doublet which is responsible for the
strong yellow light from sodium lamps.

Figure 23: The energy levels of the sodium corresponding to the ground and first excited electron
configurations. The two allowed transitions shown are the well-known yellow doublet of sodium:
λ = 589.0 and 589.6 nm

36
If the Lamb shift is ignored, it is only for a Coulomb potential that energy levels do not depend upon the quantum
number l. Irrespective of the functional form of V (r) however, the θ and φ equations are the same. Thus for multi-
electron atoms we can describe electron configurations using the same notation as used for for hydrogen, e.g.1s ,2s ,2p,
etc.

63
10.8.2 The helium atom
Helium, with two electrons, is another simple example of a multi-electron atom. Its ground state
configuration is 1s2 and exited states usually involve only one excited electron: 1snl. As we have seen,
adding the spin of the electrons leads to either S = 0, a singlet state, or S = 1, a triplet state. If we
now add the orbital angular momenta, noting that l = 0 for the 1s electron, we find that L is equal to
the value of l for the excited electron. We then obtain total angular momentum values J by adding
total spin and total orbital angular momenta.
For example, consider the 1s3d excited state of helium (s1 = 1/2, l1 = 0; s2 = 1/2, l2 = 2).
Combining spins gives S = 0, 1; combining orbital angular momenta gives L = 2. This results in
a singlet state with J = 2 (1 D2 ), and a triplet state with J = 3, 2, 1 (3 D3,2,1 ). When spin-orbit
interaction is taken into account, singlet states remain single energy levels, while triplet states split
into three different energy levels corresponding to the three values of J . This is illustrated in Figure 24.
The selection rule
∆S = 0 (10.40)
means that allowed transitions occur are only between singlet states or between triplet states, i.e. a
transition between a singlet and a triplet state is forbidden.

Figure 24: The energy levels of helium corresponding to the excited electron configuration 1s3d.

10.8.3 LS coupling
The determination of the total angular momentum of helium states is an example of LS coupling.37
It is applicable for atoms of small or intermediate atomic number where the spin-orbit interaction
is weaker than the interactions between the spin and orbital angular momenta themselves. Thus to
determine the total angular momentum we first add the spin angular momenta of the electrons to get
a total spin angular momentum, and similarly add the orbital angular momenta of all the electrons
to get the total orbital angular momentum38

S = S1 + S2 + S3 + . . . (10.41)
L = L1 + L2 + L3 + . . . (10.42)
37
Also called Russell-Saunders coupling.
38
For high Z atoms a better description is given by the jj coupling scheme: for each electron the spin and orbital
angular momenta are added to get a total angular momentum j. These are then added to obtain the total angular
momentum J for the atom.

64
Figure 25: Energy levels and allowed transitions for helium. Singlet and triplet states are shown
separately. Note that the lowest singlet and triplet states are metastable. Based on Figure 13-5, An
Introduction to Quantum Mechanics, A.P. French & E.F. Taylor (Thomas Nelson 1979).

The total angular momentum is then given by

J=L+S (10.43)

Selection rules for LS coupling

For the energy levels resulting from LS coupling allowed transitions obey the following rules.

• Only one electron undergoes a transition between sublevels

• For this electron ∆l = ±1
• The changes in S, L and J must satisfy
∆S = 0 (10.44)
∆L = ±1 (10.45)
∆J = 0, ±1 but not J = 0 → J = 0 (10.46)

Returning again to helium, Figure 25 is an energy level diagram for helium with singlet and triplet
levels grouped separately, and it shows allowed transitions. Due to the ∆S = 0 selection rule there
is no allowed transition from the lowest triplet state to the only lower state, which is a singlet state
(this transition would also violate the ∆L = ±1 selection rule). Also there is no allowed transition
from the first excited singlet state: a transition to the lower singlet state would violate ∆L = ±1 and
a transition to the lower triplet state would also violate ∆S = 0. The lowest singlet and triplet states
of helium are, therefore, metastable states.

65
11 Time-dependent Perturbation Theory
We have already argued that as the energy eigenstate contains all the information we can know about a
stationary state of an atom, the transition probability for a transition between two states must depend
upon the wave functions of those states. We will now seek this relationship, focusing on absorption
and stimulated emission. Once we have the Einstein coefficients for these processes, the spontaneous
emission transition probability can be obtained using the the Einstein relations.
The rigorous treatment is an example of time-dependent perturbation theory: the perturbing
Hamiltonian is a function of time, Ĥ 0 (t).39 In Sec. 6 we discussed how we can obtain an approximate
solution to the energy eigenvalue equation of an Hamiltonian Ĥ = Ĥ0 + Ĥ 0 , if we know the solution
to Ĥ0 and can consider Ĥ 0 a small perturbation. We however assumed that the Hamiltonian does
not depend on time. In the following we will consider perturbations which depend on time, like the
absorption or emission of a photon.

11.1 Heisenberg and Interaction Picture

As we are dealing with a time-dependent Hamiltonian, we have to return to the Schrödinger equation
(1.3) in its original form. So far we always considered the Hamiltonian to be time-independent, while
the quantum states evolved with time. This is commonly denoted as Schrödinger picture. If we
however consider the states40
 2 !
i 1 i
|ψiH ≡ eiĤt/~ |ψ(t)i = 1 + Ĥt + Ĥt |ψ(t)i (11.1)
~ 2 ~

we find that
 
∂ ∂ iĤt/~ iĤt/~ ∂
i~ |ψiH = i~ e |ψ(t)i = e −Ĥ |ψ(t)i + i~ |ψ(t)i = 0 (11.2)
∂t ∂t ∂t

This is commonly denoted as Heisenberg picture. In this picture, all operators

ÂH (t) = eiĤt/~ Âe−iĤt/~ , (11.3)

e.g. momentum operator, will be time-dependent and satisfy

!
d i ∂ Â
ÂH (t) = [ĤH (t), ÂH (t)] + . (11.4)
dt ~ ∂t
H

Thus the eigenvalues of any operator, which commutes with the Hamiltonian, are contants and provide
good quantum numbers to describe the system. We will now consider the interaction picture, in which
both the Hamiltonian and the quantum state will depend on time. It is useful for cases, where we can
split the Hamiltonian Ĥ(t) = Ĥ0 + Ĥ 0 (t) in a time-independent part Ĥ0 and a time-dependent one
Ĥ 0 . It is defined by

|ψ(t)iI = eiĤ0 t/~ |ψ(t)i ÂI (t) = eiĤ0 t/~ Âe−iĤ0 t/~ . (11.5)
39
See McIntyre, §14, p445.
40
Note that e = ∞ 1 n
P
n=0 n! Â .

66
11.2 Transition probabilities
We rewrite the Schrödinger equation for the Hamiltonian Ĥ = Ĥ0 + Ĥ 0 (t) in the interaction picture

∂ ∂
i~ |ψ(t)iI = i~ eiĤ0 t/~ |ψ(t)i (11.6)
∂t ∂t
∂
= −Ĥ0 |ψ(t)iI + eiĤ0 t/~ i~ |ψ(t)i (11.7)
 ∂t 
= −Ĥ0 |ψ(t)iI + e iĤ0 t/~
Ĥ0 + Ĥ 0 (t) |ψ(t)i (11.8)
 
= −Ĥ0 |ψ(t)iI + eiĤ0 t/~ Ĥ0 + Ĥ 0 (t) e−iĤ0 t/~ |ψ(t)iI (11.9)

= eiĤ0 t/~ Ĥ 0 (t)e−iĤ0 t/~ |ψ(t)iI (11.10)

= ĤI0 (t) |ψ(t)iI . (11.11)

We used that the exponential exp(iH0 t/~) commutes with H0 . Finally, we can now formally integrate
the equation and obtain the equivalent integral equation

1 t 0 0  0  0
Z
|ψ(t)iI = |ψ(t0 )iI + Ĥ (t ) ψ(t ) I dt . (11.12)
i~ t0 I

This integral equation can be iteratively solved by plugging the solution back in on the right-hand
side of the equation, i.e.
Z t  2 Z t Z t0
1 1
|ψ(t)iI = |ψ(t0 )iI + ĤI0 (t0 ) |ψ(t0 )iI dt +0
dt 0
dt00 ĤI0 (t0 )ĤI0 (t00 ) |ψ(t0 )iI + . . .
i~ t0 i~ t0 t0
(11.13)

We introduce the time-ordered product of operators to simplify the expression further. The time-
ordered product of two operators A and B is defined as
(
A(t1 )B(t2 ) for t1 ≥ t2
T (A(t1 )B(t2 )) ≡ . (11.14)
B(t2 )A(t1 ) otherwise

This allows us to rewrite the quadratic term as

!
Z t Z t0 Z t Z t0   Z t Z t0
dt0 dt00 ĤI0 (t0 )ĤI0 (t00 ) = dt0 dt00 T ĤI0 (t0 )ĤI0 (t00 ) = T dt0 dt00 ĤI0 (t0 )ĤI0 (t00 )
t0 t0 t0 t0 t0 t0
(11.15)
! !
Z t Z t00 Z t Z t00
=T dt00 dt0 ĤI0 (t00 )ĤI0 (t0 ) =T dt00 dt0 ĤI0 (t0 )ĤI0 (t00 )
t0 t0 t0 t0
(11.16)
Z t Z t  Z t 2
1 1
= T dt0 dt00 ĤI0 (t0 )ĤI0 (t00 ) = T 0 0 0
dt ĤI (t ) , (11.17)
2 t0 t0 2 t0

where we introduced the time-ordered product in the first line, relabelled the integration variables
t0 ↔ t00 in the second line, and noticed that summing the last terms of the first and second line and

67
dividing by 2 can be rewritten as the first term of the third line. Hence we can write the quadratic
term as the square of the integral and
2
1 t 0 0 0
Z  Z t
1 1 0 0 0
|ψ(t)iI = |ψ(t0 )iI + Ĥ (t )dt |ψ(t0 )iI + T dt HI (t ) |ψ(t0 )iI + . . . (11.18)
i~ t0 I 2 i~ t0
 iR 
− ~ tt ĤI0 (t0 )dt0
=T e 0 |ψ(to )iI . (11.19)

We will only consider the leading order transitions and neglect terms with more than one Ĥ 0 . If
we have the set of eigenstates |ni to the Hamiltonian Ĥ0 with eigenvalue En , we can calculate the
transition amplitude to find our particle in state |ni at time t if we start in state |mi at time t0 . The
system is initially in the state

|m(t)i = e−iH0 t/~ |mi = e−iEm t/~ |mi , (11.20)

where |m(t)i denotes the state |mi which has been evolved to time t with the Hamiltonian H0 . The
probability amplitude that there is a transition to the state

|n(t)i = e−iH0 t/~ |ni = e−iEn t/~ |ni (11.21)

is given by D E
hn(t)|ψ(t)i = n|eiH0 t/~ |ψ(t) = hn|ψ(t)iI (11.22)

where the subscript I indicates the interaction picture. The initial state |ψ(t0 )iI = eiH0 t/~ |m(t)i =
|mi. Inserting the states in Eq. (11.13), we find

1 tD
Z E
hn|ψ(t)iI = hn|mi + n|ĤI0 (t0 )|m dt0 (11.23)
i~ t0
1 t i(En −Em )t0 /~ D
Z E
= δnm + e n|Ĥ 0 (t0 )|m dt0 ,
i~ t0

where we used the definition of an operator in the interaction picture in Eq. (11.5). Note that all
expressions in the last line are given in the usual Schrödinger picture, which we worked with in the
previous sections. The probability for a transition from the state |mi to the state |ni 6= |mi is given
by
 Z t
1 D E2
0 i(En −Em )t0 /~ 0 0
Pmn (t) = 
 dt e n|Ĥ (t )|m  . (11.24)
i~ t0
The first term in Eq. (11.23) corresponds to an unchanged quantum state |mi, which has not been
affected by any interaction. The second term describes one transition between two states induced by
the Hamiltonian Ĥ 0 :

1. We start with a quantum system in the eigenstate |mi of the Hamiltonian H0 .

2. Then we evolve the quantum system in time from t0 to t0 using the Hamiltonian H0 . Thus
eigenstates |mi of H0 do not change and only pick up a phase factor exp(iEm (t0 − t0 )/~).

3. At time t0 , the Hamiltonian Ĥ 0 affects the

D quantum
E system and leads to transitions between
0
states, depending on the matrix elements n|Ĥ |m .

68
 4. After the interaction, the system is in a new state |ni and we evolve it with the Hamiltonian H0
in time from t0 to t and the quantum state picks up a phase factor exp(iEn (t − t0 )/~).

5. The transition probability is given by projecting this state at time t onto the eigenstates of H0 .
We explicitly include the phase factors from time evolution.

Higher order terms in Eq. (11.13) can be interpreted in the same way. The only difference is that
there will be multiple transitions at different times t0 , t00 , . . . , and the quantum states between the
transitions are evolved in time using the Hamiltonian H0 leading to additional phase factors.

11.3 Selection Rules

The expression for the transition probability at leading order given in Eq. (11.24) shows that transitions
are only possible, if the transition matrix element
D E
n|Ĥ 0 (t)|m 6= 0 (11.25)

does not vanish. The same applies for higher order transitions. If the matrix element vanishes, there
won’t be any transitions at any order in perturbation theory and thus the transition is forbidden. If
on the contrary, the matrix element takes a finite non-zero value, the transition is allowed.
Thus the selection rules originate from the transition matrix element: A vanishing matrix element
implies that the transition is forbidden, while transitions with a non-zero matrix element are allowed.

11.4 Fermis Golden Rule

We will now apply our general result to the interaction of an electromagnetic wave with an atom.
The electric field of the wave will in general induce an electric dipole d = −er in the atom. The
energy associated with the dipole in the electric field given by

E = −d · E (11.26)

where the electric field due to the electromagnetic wave can be represented as41

E(t) = 2E0 ˆ cos ωt (11.27)

where ˆ is a unit vector specifying the polarisation of the radiation. The perturbing Hamiltonian is,
therefore
Ĥ 0 = 2eE0 ˆ · r cos ωt . (11.28)
We will more generally consider an Hamiltonian of the form

Ĥ 0 (t) = 2V (r) cos ωt = V (r) eiωt + e−iωt .

(11.29)
41
The factor of 2 may not seem necessary. It is introduced in order to remain consistent with the way McIntyre
represents the electromagnetic field. It has no effect, of course, on the results for the Einstein coefficients . See the
discussion on p. 454 of McIntyre.

69
The amplitude for the corresponding transition probability from an initial state |ii given at time t = 0
to a final state |f i at time t is given by
1 tD
Z   E
0 0 0
Af i = f |V (r) eiωt + e−iωt |i ei(Ef −Ei )t /~ dt0 (11.30)
i~ 0
= ... (11.31)
   
ωf i +ω ωf i −ω

1 ωf i +ω sin 2 t ω −ω sin 2 t
i f i2 t
= Vf i ei 2 t ωf i +ω + e ωf i −ω
 (11.32)
i~
2 2

where we defined the characteristic frequency ~ωf i = Ef − Ei and Vf i = hf |V (r)|ii. See Sec. 14.2
in McIntyre for the intermediate steps. The transition probability Pf i is given by the square of the
amplitude Af i . We will now only consider the dominant term and take the limit t → ∞. The transition
amplitude is dominated by frequencies ω ≈ ±ωf i . The first term is relevant for stimulated emission,
because Ef < Ei and thus ωf i < 0, while the second one is relevant for absorption with Ef > Ei . We
will focus on absorption and only keep the second term. The transition probability for absorption of
a photon is given by  
2 ωf i −ω
|Vf i |2 sin 2 t
Pf i = , (11.33)
2 ωf i −ω 2
 
~
2

which is an oscillating function of t as shown in Fig. 26 with a period of ∆t = 2π(ωf i − ω). As a

function of ω, it is peaked at ωf i with a characteristic width of ∆ω = 4π/t. In the limit of ∆ω → 0,

Pf i Pf i

t ω

Figure 26: Probability as a function of time (left-hand side) and frequency (right-hand side)

i.e. t → ∞, we obtain a Dirac delta function,

 
ω −ω
sin2 f i2 t
lim  2 = 2πtδ(ωf i − ω) (11.34)
t→∞ ωf i −ω
2

and thus the transition probability is given by

2πt
Pf i = |Vf i |2 δ(ωf i − ω) (11.35)
~2

70
 dP
in the limit t → ∞. This allows us to define the transfer rate R ≡ dt

2π
Rf i = |Vf i |2 δ(ωf i − ω) , (11.36)
~2
which is called Fermi’s Golden Rule. The delta function imposes energy conservation.
We now specialise to the interaction Hamiltonian (11.28) and find

Vf i = eE0 ˆ · hf |r|ii . (11.37)

The energy density per unit volume in an electromagnetic field is given by

0 2 1 2
u= E + B (11.38)
2 2µ0
and thus our light source has an energy density

u = 40 E02 cos2 ωt , (11.39)

where we used that the energy density is the same in the electric and magnetic field. Taking a
time-average over one cycle, we obtain the average energy density

ū = 20 E02 . (11.40)

The energy density of a broadband light source is described by the field energy per unit volume and
angular frequency ω
ū = ρ(ω)dω (11.41)
and thus E0 can be expressed as
ρ(ω)
E02 = dω (11.42)
20
and we obtain for the transition rate of a broadband light source after integrating over angular fre-
quency ω
πe2
Rf i =  · hf |r|ii|2 .
ρ(ωf i ) |ˆ (11.43)
0 ~2
For a gas of atoms in thermal equilibrium
with black-body
 1 radiation, the polarisation vector is random.
42
Averaging over the direction , we find |ˆ 2
 · r̂| = 3 and thus the transition rate is given by

πe2
Rf i = ρ(ωf i ) |hf |r|ii|2 = Bif ρ(ωf i ) . (11.44)
30 ~2
with the Einstein coefficient Bif for absorption. The transition rate is proportional to the radiation
density ρ(ωf i ). The calculation for stimulated emission leads to the same result and thus we find for
a two-level system
πe2
B21 ≡ B12 = | h2|r|1i |2 (11.45)
30 ~2
and using the Einstein relations we obtain an expression for the spontaneous emission transition
probability for |2i to |1i,
e2 ω21
3
A21 = | h2|r|1i |2 (11.46)
3π0 ~c3
42
See Sec. 14.3.1 in McIntyre.

71
We see that when radiation of frequency ω21 is incident upon an atom, the probabilities of absorption
(∝ B12 ) and stimulated emission (∝ B21 ) are equal. The actual rates will depend upon the population
densities of the lower and upper states respectively. If the density of the lower |1i exceeds that of
the upper |2i, absorption will dominate. If, however, the density of the upper |2i exceeds that of the
lower |1i, there is a population inversion and stimulated emission dominates; the medium will be able
to amplify radiation at frequency ω21 . With the addition of an optical cavity and an output coupling
mechanism (e.g. a partially transmitting mirror at one end), we have a laser.

11.4.1 Selection Rules

Let us finally have a look at the selection rules. A transition between |2i and |1i is allowed, if the
matrix element does not vanish
h2|r|1i =
6 0. (11.47)
These conditions are usually specified in terms of the difference in quantum numbers between states |2i
and |1i. For example for the states |nlml i of hydrogen, the transition will be allowed, if the selection
rules ∆l = ±1 and ∆ml = 0, ±1 are satisfied for a transition. On the other hand a transition will be
forbidden if
h2|r|1i = 0 (11.48)
This does not necessarily mean that there is no interaction with electromagnetic radiation or sponta-
neous emission. Rather it means that there is no dipole interaction: radiation incident on the atom
will not induce an electric dipole moment. It may however induce a quadrupole moment, resulting in
a much weaker interaction between the atom and the radiation. In this case the Einstein coefficients
will not be zero, but many orders of magnitude smaller than for an allowed transition.

72
A 3D infinitely deep potential well
The potential of an infinitely deep potential well of size L1 × L2 × L3 is given by
(
0 0 < x < L 1 , 0 < y < L2 , 0 < z < L 3
V (x, y, z) = (A.1)
∞, otherwise .
Similarly to the 1D infinitely deep potential well, the potential energy is infinite outside the well and
thus the only solution is ϕE (x, y, z) ≡ 0. Inside the well the Hamiltonian is given by
p̂2 ~2 2 ~2
 2
∂2 ∂2

∂
Ĥ = =− ∇ =− + + , (A.2)
2m 2m 2m ∂x2 ∂y 2 ∂z 2
where we used the definition of the momentum operator p̂ = −i~∇. It can be written as a sum
of three Hamiltonians Hx,y,z for each of the three different coordinates and it is separable. As the
Hamiltonian separates into three different subsystems, we expect that the quantum state vector can
also be separated. This is not always the case and the different states might be entangled, but it is
a valid assumption for the 3D infinitely deep potential well, since the potential is constant and does
not depend on either of the three coordinates inside the box. Hence we make the following ansatz for
the quantum state
ϕE (x, y, z) = ϕx (x)ϕy (y)ϕz (z) . (A.3)
Plugging the ansatz in the Schrödinger equation, we obtain
EϕE (x, y, z) = ĤϕE (x, y, z)
~2 ∂2 ∂2 ∂2
 
=− ϕy (y)ϕz (z) 2 ϕx (x) + ϕx (x)ϕz (z) 2 ϕy (y) + ϕx (x)ϕy (y) 2 ϕz (z) . (A.4)
2m ∂x ∂y ∂z
Formally rewriting it as
1 ∂ 2 ϕx (x) 2mE 1 ∂ 2 ϕy (y) 1 ∂ 2 ϕz (z)
= − − − , (A.5)
ϕx (x) ∂x2 ~2 ϕy (y) ∂y 2 ϕz (z) ∂z 2
the left hand side only depends on x and the right-hand side does not depend on x. Thus the left-hand
side has to be constant and the equation separates
d2 ϕx (x) 2mEx
+ ϕx (x) = 0 (A.6)
dx2 ~2
with a undetermined constant Ex . Analogously for the other parts. The three introduced constants
satisfy Ex + Ey + Ez = E. Hence we showed that the problem of a 3D infinitely deep potential well
can be separated in three 1D infinitely deep potential wells, which we solved in the last lecture. The
energy levels are discrete and depend on 3 quantum numbers nx , ny , nz .
!
π 2 ~2 n2x n2y n2z
Enx ny nz = + + . (A.7)
2m Lx Ly Lz

B Bohr’s model of the atom

Bohr put forward the first quantum model of the atom in 1913. It explained some features of atomic
spectra, and although it has been superseded by modern quantum mechanics, it is instructive to
review its successes and deficiencies. Bohr assumed Rutherford’s model of the atom in which the
atom consisted of a very small positively charged nucleus at the centre with the electrons occupying
the full volume of the atom. To this he added the following postulates.

73
B.1 Bohr’s postulates
1. an electron moves in a circular orbit about the nucleus according to classical mechanics

2. only those orbits for which the orbital angular momentum of the electron is an integral multiple
of ~ are possible (~ = h/2π)

3. an electron in such an allowed orbit does not radiate electromagnetic radiation - its energy
remains constant

4. electromagnetic energy is emitted (or absorbed) when an electron changes from one allowed orbit
to another

The equation of motion for an hydrogen-like atom (Z = 1 → H, Z = 2 → He+ , etc) in which a

single electron orbits a nucleus of charge +Ze is

1 Ze2 mv 2
= (B.1)
4π0 r2 r
The quantisation of orbital angular momentum gives

L = mvr = n~ (B.2)
where n = 1, 2, 3, . . . is an integer. Solving these two equations simultaneously we get expressions for
the two unknowns, the velocity and the radius of the orbiting electron:

1 Ze2 v
v= → = Zαn (B.3)
4π0 n~ c
where α = e2 /(4π0 ~c) = 7.30 × 10−3 ≈ 1/137 is called the fine structure constant43 , and.

n2 ~2 n2
r = 4π0 2
= a0 (B.4)
mZe Z
where

4π0 ~2
a0 = = 5.29 × 10−11 m (B.5)
me2
is called the Bohr radius. It is the radius of the orbit of the ground state (n = 1) of hydrogen (Z = 1).
The energy of an orbiting electron is the sum of its kinetic and potential energies.

1 Ze2 1 Ze2 1 Ze2

E =K +V = − =− (B.6)
4π0 2r 4π0 r 4π0 2r
Substituting for r from eqn B.4,
2
Ze2 −Z 2

m
E=− 2 2 = Ryd (B.7)
2n ~ 4π0 n2
where
43
We will encounter the fine structure constant at various times during this course. Note that it is a dimensionless
ratio of fundamental constants, and its value is an indicator of the extent to which the motion of the electron in the
hydrogen atom is relativistic.

74
 2
e2

m
Ryd = − (B.8)
2~2 4π0
Ryd has the units of energy and its value is 13.6 eV.
For hydrogen (Z = 1),

v/c = α/n (B.9)

2
r = n a0 (B.10)
2
E = −Ryd/n (B.11)

Figure 27 shows low energy levels for hydrogen and transitions to the ground state.

Figure 27: Energy level diagram for hydrogen showing the first four transitions of the Lyman series of
spectral lines - transitions from higher levels to level n = 1: Lα (122 nm), Lβ (102 nm), Lγ (97 nm).

B.2 Emission and absorption spectra

If the hydrogen atom (Z = 1) undergoes a transition from state ni to state nf where ni > nf , the
frequency of the emitted radiation is given by ν = (Ei − Ef )/h. The wave number κ and wavelength
λ are given by
!
1 1 1
κ ≡ = R∞ − 2 (B.12)
λ n2f ni
where
2
me4

1 Ryd
R∞ = = (B.13)
4π0 4π~3 c hc

B.3 Finite nuclear mass

The equations above assume that the nucleus is infinitely massive. When the mass of the nucleus is
taken into account m is replaced by the reduced mass of the atom, µ, where

75
 mM
µ= (B.14)
m+M
where M is the mass of the hydrogen nucleus. The constant in Equation B.12 becomes

M
RH = R∞ (B.15)
m+M
and is called the Rydberg constant.

R∞ = 1.09737 × 107 m−1 (B.16)

7 −1
RH = 1.09681 × 10 m

B.4 The hydrogen spectrum

The wavelengths of emission spectral lines of the hydrogen Balmer series (nf = 2) are given by
 
1 1 1
= RH − (B.17)
λ 4 n2i
where ni = 3, 4, 5, . . .. The resulting spectral lines are in the visible and near ultraviolet, and are
known respectively as Hα (656 nm), Hβ (486 nm), Hγ (434 nm), . . . . Several transitions of the Lyman
series (nf = 1) are shown in Figure 27, corresponding to Lα (122 nm), Lβ (102 nm), Lγ (97 nm).

C Spherical Coordinates

Figure 28: The spherical polar coordinate system.

The appendix is a short summary of spherical coordinates which are most suitable for any problem
which is spherically symmetric. The coordinates are defined according to Fig. 28 as follows

x = r sin θ cos φ
y = r sin θ sin φ (C.1)
z = r cos θ .

76
The spherical volume element can be obtained by a change of variables from (x, y, z) to (r, θ, φ)
 
 ∂(x, y, z) 
dxdydz =   drdθdφ = r2 sin θdrdθdφ (C.2)
∂(r, θ, φ) 

and the Laplace operator

∂2
   
2 1 ∂ 2 ∂ 1 ∂ ∂ 1
∇ = 2 r + 2 sin θ + 2 2 . (C.3)
r ∂r ∂r r sin θ ∂θ ∂θ r sin θ ∂φ2

D The Dirac equation

The (time dependent) Schrödinger equation can be written as

∂ψ
i~ = Hψ (D.1)
∂t
where

~2 2
H=− ∇ + V (r) (D.2)
2m
Paul Dirac’s aim was to find a relativistic version of this that was also specific to a spin- 12 particle.
This required finding an expression for H that made the equation Lorentz invariant. We also know
that spin operators are matrices and their eigenfunctions are vectors. So it should be no surprise that
the equation Dirac arrived at is not simply a scalar equation. His equation is

(E/c + α · p + α4 m0 c)ψ = 0 (D.3)

where αi is a 4 × 4 matrix, p is the momentum 4-vector and ψ = (ψ1 , ψ2 , ψ3 , ψ4 ) is a four component
wave function. The αi are given by
 
0 σi
αi = (D.4)
σi 0
where i = x, y, z and σi are the Pauli matrices, and
 
1 0 0 0
 0 1 0 0 
α4 = 
 0 0 −1
 (D.5)
0 
0 0 0 −1
This is Dirac’s equation for a free electron with spin 12 . When Dirac modified this equation for an
electron in a magnetic field, E → E − eφ and p → p − e/cA he found that the solution included an
energy term that corresponded to the energy of a magnetic moment in the magnetic field (B = ∇ × A)
associated with a particle of spin 12 , provided g = 2.
Dirac’s equation also predicted a particle with the mass of an electron and charge +e, now called a
positron (the anti-particle of an electron) and first detected experimentally by Carl Anderson in 1932
(for which he received the Nobel Prize for Physics in 1936).

77
E Einstein relations
Consider a two-level atomic system that is in thermal equilibrium with radiation, in which case the
transition rate 2 → 1 (spontaneous emission plus stimulated emission) must equal the transition rate
1 → 2 (absorption):

n2 (A21 + B21 ρ(ν)) = n1 B12 ρ(ν) (E.1)

Because of thermal equilibrium the densities n1 and n2 are related by the Boltzmann distribution
n2
= exp(−hν/kT ) (E.2)
n1
and the radiation density is given by the Planck’s law (for radiation of a black body)

8πhν 3 1 ~ ω3
ρ(ν) = = (E.3)
c3 ehν/kT − 1 π 2 c3 ehν/kT − 1
From Equation (E.1), and substituting from Equation (E.2),

(A21 /B21 )(n2 /n1 ) A21 /B21

ρ(ν) = = (E.4)
B12 /B21 − n2 /n1 B12 /B21 ehν/kT − 1

Requiring that this corresponds to the black body formula leads to the Einstein relations
B12
= 1 (E.5)
B21
A21 8πhν 3
= (E.6)
B21 c3

Revised on June 9, 2017

78