Physics of Atoms and Molecules

Prof. K. Heyne
Inhalt i

Inhalt
1 Introduction 1
1.1 Planck's energy distribution law . . . . . . . . . . . . . . . . . . . 2
1.2 Radiative transitions in a two level system . . . . . . . . . . . . . 10
1.2.1 Spontaneus Emission . . . . . . . . . . . . . . . . . . . . . 10
1.2.2 Stimulated Emission . . . . . . . . . . . . . . . . . . . . . 12
1.2.3 Stimulated Absorption . . . . . . . . . . . . . . . . . . . . 14
1.3 Relation between Einstein coecients . . . . . . . . . . . . . . . . 14
1.3.1 Level Degeneracy . . . . . . . . . . . . . . . . . . . . . . . 15
1.4 Fouriertransformation . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5 Homogeneous Line Broadening . . . . . . . . . . . . . . . . . . . . 31

2 Waves and particles 37

2.1 Photons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.2 Compton eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.3 Plane waves, free particles . . . . . . . . . . . . . . . . . . . . . . 42

3 Schrödinger equation 52
3.1 Expectation values . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.2 Time independent SE . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.3 Some properties of operators . . . . . . . . . . . . . . . . . . . . . 55

4 Particle in a box 59

5 Ground state of the hydrogen atom 61

6 General remarks to solve the SE 62

7 Harmonic oscillator 1-D 64

7.1 Conserved quantities . . . . . . . . . . . . . . . . . . . . . . . . . 70
7.2 Variational method . . . . . . . . . . . . . . . . . . . . . . . . . . 72
7.3 Angular momentum and spherical potentials . . . . . . . . . . . . 73

8 Hydrogen atom and central forces 87

8.1 Central forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
8.2 Radial functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
8.3 Hydrogen atom wave functions . . . . . . . . . . . . . . . . . . . . 93
8.4 Angular momentum . . . . . . . . . . . . . . . . . . . . . . . . . . 96
8.5 Spin angular momentum . . . . . . . . . . . . . . . . . . . . . . . 97
8.6 Total angular momentum . . . . . . . . . . . . . . . . . . . . . . . 99

Physics of Atoms and Molecules

ii Inhalt

9 Interaction of atoms with electromagnetic radiation 101

9.1 Time-dependent perturbation theory . . . . . . . . . . . . . . . . 106
9.1.1 Time-independent perturbation . . . . . . . . . . . . . . . 108
9.1.2 Periodic perturbation . . . . . . . . . . . . . . . . . . . . . 110
9.2 Absorption and emission . . . . . . . . . . . . . . . . . . . . . . . 111
9.3 Selection rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

10 Spin of the electron 119

10.1 Spin-orbit and additional couplings . . . . . . . . . . . . . . . . . 122
10.2 Spin of a nucleus . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
10.2.1 Hyperne splitting in an external magnetic eld . . . . . . 134
10.2.2 Electron spin resonance (EPR) . . . . . . . . . . . . . . . 135
10.2.3 Nuclear magnetic resonance (NMR) . . . . . . . . . . . . . 138
10.3 Lifetimes of excited states . . . . . . . . . . . . . . . . . . . . . . 144
10.3.1 Atomic lifetimes . . . . . . . . . . . . . . . . . . . . . . . . 144
10.4 Density operator . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

Physics of Atoms and Molecules

1 Introduction
The rst speculations on the atomic nature of matter was made by Greek philo-
sophers such as Anaxagoras (500-428 BC), Empedocles (484-424 BC), Leucippus
( 450 BC) and Democritus (460-370 BC) arguing that the universe consists of
empty space and indivisible atoms. Aristotele (384-322 BC), however rejected
the atomic hypothesis and strongly supported the concept of the continuity of
matter.
This question was re-opened following the experimental discovery of gas laws
by R. Boyle in 1662, the interpretation of these laws by D. Bernoulli in 1738, and
the development of the kinetic theory of gases throughout the nineteenth century
by R. Clausius, J.C. Maxwell and L. Boltzmann. In parallel laws of chemical
combination were being discovered, such as the law of denite proportions (by
J.L. Proust in 1801) and the law of multiple proportions (by J. Dalton in 1807),
supporting the concept of the atomic nature of matter. During the following years
elementary units and physical constants were determined strongly related to the
atomic nature of matter:

ˆ Avogadro's Number NA = 6.02214 · 1023 mol 1

is the number of atoms or
molecules in one mole of any substance. A mole is dened as the quantity
of 12 C weighing µ = 12 grams, where µ is the atomic or molecular weight
of the substance.

ˆ Faraday's constant F = 9.64853 · 104 mol C

denes the existence of the
elementary unit of electricity. Faraday's laws of electrolysis, which can be
summarized by the formula,

Qµ
M=
Fυ
where Q is the quantity of electricity and υ the valency, demonstrate that
it takes 96484.3 C to liberate for example 1.008 g of hydrogen, 35.5 g of
chlorine and 8 g of oxygen.

ˆ Specic charge mee = 1.75881962 · 1011 kg C

of an cathode ray measured in
Thomson's experiment (in 1897), where e is the natural unit of electricity
and me is its mass. The specic charge for the lightest known positive ion
(the hydrogen ion) is smaller by a factor of approximately 1840, so either
the cathode ray particles are much lighter or they carry a very large charge.
2 1 Introduction

ˆ The charge of the electron was measured in Millikan's experiment to

be e = 1.6021773 · 10−19 C . Combining these results with the value of mee , a
value for the mass of the electron is obtained me = 9.1093897 · 10−31 kg .
During the later part of the nineteenth century, and in the early years of the
twentieth century, evidence accumulated that classical physics, i.e. Newton's laws
of motion and Maxwell's electromagnetic equations, are inadequate to describe
atomic phenomena. Most striking is the atomic or molecular picture of a number
of electrons moving and accelerating around an attracting nucleus, similar to
the planets circling around the sun in the solar system. This picture is in direct
contradiction to the stability of atoms and molecules, since accelerating charges
would continuously loose energy by emitting electromagnetic radiation, as a result
of Maxwell's equations:
−
→ − →
∇•B =0 (1.1)
−
→ − → ∂− →
∇×E + B =0 (1.2)
∂t
−
→ − →
∇•D =% (1.3)
−
→ − → ∂− → − →
∇×H− D = j (1.4)
∂t
with relations in homogeneous and linear media:
−
→ −
→ − → −
→
B = µ0 ( H + M ) → µr µ0 H (1.5)
−
→ −
→ − → −
→
D = ε0 E + P → εr ε0 E (1.6)
Electromagnetic radiation from decelerated electrons can be observed for example
in the bremsstrahlung and one of the most powerful light sources available make
use of highly accelerated charged particles to generate broadband electromagnetic
radiation from the ultraviolet to the near infrared spectral range (Free electron
laser in Fig. 1, BESSY in, Fig. 1).

Abbildung 1.1. Light generation in a free electron laser. Note, that the lasing process is due
to stimulated emission in the direction of the moving electrons.

1.1 Planck's energy distribution law

The rst clues to a new physics, based on the quantisation of energy, came from
a study of the properties of radiation from hot bodies. In 1879, J. Stefan showed

Physics of Atoms and Molecules

1.1 Planck's energy distribution law 3

Abbildung 1.2. Top view of the electron synchrotron BESSY in Berlin-Adlershof. The elec-
trons stored on circular orbits move with relativistic velocities, emitting continuously synchro-
tron radiation.

that the total power emitted per unit area, R, called the total emissive power (or
total emittance) from a body at the absolute temperature T could be represented
by the empirical law
R = ²σT 4 (1.7)
where ² is called the emissivity with ² ≤ 1. The emissivity, ², varies with the
nature of the surface, and σ = 5.67 · 10−8 m2W·K 4 is known as Stefan's constant. The
spectral distribution of the emitted light depends strongly on the temperature
(Fig. 1.1) and exhibits a maximum value at wavelength λmax following the Wien's
displacement law, with b = 2.898 · 10−3 mK :

λmax T = b (1.8)

When a body is in thermal equilibrium with its surroundings, and therefore is

at constant temperature, it must absorb and emit the same amount of radiant
energy per time. Analogue to the emissivity, absorptivity, a, is dened as the
fraction of the absorbed radiant energy falling on the surface. A body having
an absorptivity equal to unity is called black body. A close approximation to
a black body is an enclosed cavity with blackened interior walls containing a

Physics of Atoms and Molecules

4 1 Introduction

small hole. Radiation that enters the hole has very little chance of escaping. If
the inside of this cavity is in thermal equilibrium it must emit as much energy
as it absorbs (on every time scale) and the emission from the hole is therefore
characteristic of the equilibrium temperature T inside the cavity. This type of
radiation is called thermal radiation. In the latter part of the nineteenth century

6
Temperature T:

5 5000 K
1000 K
4 373 K
105 Jm 4)
-

293 K
3
2000
( ) (

2 x 200000
x 500000
1

0
0 2 4 6 8 10
Wavelength ( m)

Abbildung 1.3. Energy density per unit volume and wavelength as a function of wavelength
in µm for dierent temperatures T.

experimental measurements of this spectral prole had already been obtained

and tted to an empirical formula. Attempts to explain the data were based on
treating the electromagnetic radiation as a collection of oscillators, each oscillator
with its own characteristic frequency. The problem was to determine how many
oscillations at a given frequency could be tted inside a cavity. For convenience
purposes, we choose cartesian coordinates and a cubical cavity with sides of length
−
→ −→
L. A plane electromagnetic standing wave ei k • r has to satisfy the (periodic)
boundary conditions inside this cavity

eikx x = eikx (x+L) ; and A(0) = A(L) = 0 (1.9)

with similar equations for the y and z components of the wave. Equation 1.9 is
satised if
2π`
k = (1.10)
L
−
→
where  = {x, y, z} and k are the components of the wave vector k of the
oscillation and ` are integers.
−
→ −
→ ω 2π
k = | k |k = k = k (1.11)
c λ
This means for a single component the wavelength has to be λ = L` or k = 2π` L
.
→
−
In the three-dimensional k -space, whose axes are kx , ky , and kz the possible k

Physics of Atoms and Molecules

1.1 Planck's energy distribution law 5

values in the volume L3 satisfying equation 1.10 form a lattice with size of a
unit cell of ( 2π
L
)3 . The spacing of adjacent modes in the k direction is 2π L
and
2π 4π
the permitted k values are 0, L , L , etc., correspond to oscillation wavelengths
−
→ −
→
∞, L, L2 , etc. The total number of modes of oscillation with | k | ≤ k (| k |-radius
in a sphere) and two distinct polarizations of the radiation eld is
−
→
volume of k − space
Nk = 2×
volume of unit cell
4 L
= 2 × πk 3 × ( )3 (1.12)
3 2π
3 3
k L
= (1.13)
3π 2
2πν
Since k = c
the total number of frequencies ≤ ν in the volume V = L3 is

8πν 3 L3 8πν 3 V
Nν = = (1.14)
3c3 3c3
The mode-density, which is the number of modes per unit volume and unit fre-
quency interval is
1 dNν (ν) 8πν 2
pb(ν) = = 3 (1.15)
V dν c
Rayleigh and Jeans used this type of mode-density to describe the spectral com-
position of black-body radiation by calculating the energy density
8πν 2
p(ν) = ε (1.16)
c3
where ε is the average energy in the mode with frequency ν . The energy ε of
each oscillator can take any value, independently of the frequency ν , but since
the system is in thermal equilibrium, the average energy ε can be obtained by
weighting each value of ε with the Boltzmann factor exp (−ε/(kB T )), where kB =
1.380658 × 10−23 JK −1 is Boltzmann's constant. With β = kB1T we have
R ∞ −βε
εe dε
ε = R0 ∞ −βε (1.17)
0
e dε
Z ∞
d 1
= − [ln e−βε dε] = = kB T (1.18)
dβ 0 β
Inserting this value of ε into (1.16) gives the Rayleigh-Jeans distribution law

8πν 2
p(ν) = kB T (1.19)
c3
In the limit of long wavelengths the Rayleigh-Jeans result approaches the experi-
mental results. However, p(ν) does not show the observed maximum, and diverges

Physics of Atoms and Molecules

6 1 Introduction

as ν → ∞. This behaviour at short wavelengths is known as the 'ultra-violet ca-

tastrophe'. Planck resolved these diculties with his quantum hypothesis. He
proposed that each oscillation mode could only take certain quantized energies

1
Env = (n + )hν, n = 0, 1, 2, 3 . . . , (1.20)
2

where the contribution 21 hν is called the zero point energy, with the Planck's
constant h = 6.6260755 × 10−34 Js. The probability of nding energy En in a
particular mode of oscillation is given by classical Maxwell-Boltzmann statistics
(integration over the energy is not longer possible), i.e.,

−En
P (n) e kB T −nhv
= −E0 = e kB T (1.21)
P (0) e kB T

where P(n)and P(0) are the probabilities of nding the energy En and the lowest
energy E0 in the mode. The average energy Eν of a mode ν is now given by:

∞
X ∞
X −nhν 1
Eν = P (n)En = P (0)e kB T (n + )hν (1.22)
n=0 n=0
2

Now, if a particular oscillation is excited it must be in one of the quantized states,

therefore the normalization relation holds

∞
X
P (n) = 1 (1.23)
n=0
∞
X −nhν
P (0)e kB T = 1 (1.24)
n=0
P (n)
⇒ P (0) = −nhν (1.25)
e kB T
∞
X X∞
1 −nhν
P (n) = e kB T (1.26)
P (0) n=0 n=0
P∞
n=0 P (n)
⇒ P (0) = P∞ −nhν (1.27)
n=0 e
kB T

1
⇒ P (0) = P −nhν (1.28)
∞
n=0 e
kB T

Physics of Atoms and Molecules

1.1 Planck's energy distribution law 7

This gives
P∞ −nhν

n=0 (n + 12 )hνe kB T
Eν = P∞ −nhν (1.29)
n=0 e
kB T

−hν −2hν −nhν

1 hνe kB T + 2hνe kB T + · · · + nhνe kB T
= hν + −hν −2hν −nhν (1.30)
2 1 + e kB T + e kB T + · · · + e kB T
−hν −2hν
d
1 d(1/kB T )
(e kB T + e kB T + · · · )
= hν − 1 (1.31)
2 −hν
(1−e kB T )
d 1
d(1/kB T )
[ −hν ]
1 1−e kB T
= hν − 1 (1.32)
2 −hν
(1−e kB T )
d −βhν −1
1 dβ
(1 −e )
= hν − (1.33)
2 (1 − e−βhν )−1
1 (1 − e−βhν )−2 × hνe−βhν
= hν + (1.34)
2 (1 − e−βhν )−1
1 hνe−βhν
= hν + (1.35)
2 (1 − e−βhν )
1 hν
Eν = hν + βhν
(1.36)
2 e −1
Here, we used the mathematical 'trick' that the series in the numerator is the
derivative of a geometric series that we can sum easily. Another way to derive
the formula is:
P∞ ∞
1 −βnε0 X
n=0 (n + 2 )ε0 e 1 d
ε = P∞ −βnε = ε0 − [ln e−βnε0 ] (1.37)
n=0 e 0 2 dβ n=0
1 d 1 1 1
= ε0 − [ln( −βε
)] = ε0 [ + βε0 ] (1.38)
2 dβ 1−e 0 2 e −1

Here, we used ε0 = hν . Substituting the value of ε in (1.16), we nd

8πν 2 1 1
p(ν) = 3
(hν[ + βhν ]) (1.39)
c 2 e −1

The factor 12 comes from zero point energy that cannot be released, so the availa-
ble stored energy in the eld is given by Planck's distribution law:

8πhν 3 1
p(ν) = 3
[ hν ] (1.40)
c e kB T − 1

Physics of Atoms and Molecules

8 1 Introduction

This is (the number of modes per volume per frequency interval) x (photon ener-
gy) x 1/(ehν/kB T −1). The quantity 1/(ehν/kB T −1) represents the average number
of photons in each mode. This is called the occupation number of the modes of
the eld.

Physics of Atoms and Molecules

1.1 Planck's energy distribution law 9

Abbildung 1.4. Frequency ranges in dierent units.

Physics of Atoms and Molecules

10 1 Introduction

1.2 Radiative transitions in a two level system

Consider molecules with discrete energy levels shown schematically in Fig. (1.2).
Transitions of electrons from one level to the other can occur in three ways:
Spontaneous emission, stimulated emission and stimulated absorption.

Ei νij is the frequency of the

emitted photon

hνij
Ej
Ek
E0 with the lowest energy
E0 is the ground state

Abbildung 1.5. Energy levels of a particle (atom, molecule, etc.). Radiative transitions can
occur between dierent energy levels upon absorbing or emitting photons.

1.2.1 Spontaneus Emission

An electron spontaneously undergoes a transition from a higher (occupied) energy
level Ei to a lower (unoccupied) one Ej as shown in Fig. (1.2.2). The released
energy is emitted by a photon of frequency

Ei − Ej
νij = (1.41)
h
This photon is emitted in a random direction with arbitrary polarization in the
absence of magnetic and electric elds. The photon carries away momentum
−
→
→
− hk −
→ h
p = =~k = k (1.42)
2π λ
h hν
|p| = = (1.43)
λ c
and the emitting particle (atom, molecule, ion, etc.) recoils in the opposite direc-
tion. The probability of the spontaneous emission process is given quantitatively
by the Einstein coecient Aij dened as the probability per second of a sponta-
neous transition from level i to level j . If the population density per unit volume
in the level i is Ni then Ni Aij is the probability per second and unit volume for
a transition from level i to level j (i → j ). The total rate at which spontaneous
radiative transitions are made between the two levels is
dNi
= −Ni Aij (1.44)
dt

Physics of Atoms and Molecules

1.2 Radiative transitions in a two level system 11

Ei Ni Population density

hνij
Ej Nj
Abbildung 1.6. Transition from a higher energy i level to a lower energy level j by light
emission.

The negative sign indicates that the population of the upper level is decreasing.
Generally the transition of an electron can occur to more than one unoccupied
lower level, unless it is in the rst (lowest) excited level. The total transition rate
for transitions from level i is Ai s−1 where
X
Ai = Aij (1.45)
j

The summation runs over all levels j lower in energy than the level i and the
total rate at which the population of level i changes by spontaneous emission is
dNi
= −Ni Ai ⇒ Ni = constant × e−Ai t (1.46)
dt
with t = 0 and Ni = Ni0 it is
Ni = Ni0 e−Ai t . (1.47)
The population Ni of the level i falls exponentially with time by spontaneous
emission. The time in which the population falls to 1/e of its initial value at
t = 0 is called the natural or radiative lifetime τi of level i, where τi = 1/Ai . The
magnitude of this lifetime is determined by the actual probabilities of transitions
from level i by spontaneous emission.
Transitions which are likely to occur are called allowed transitions, those which
are unlikely are said to be forbidden. Allowed transitions in the visible spectral
range have Aij coecients in the range of above 106 s−1 , whereas forbidden transi-
tions in this spectral range have Aij coecients below 104 s−1 . These probabilites
decrease as the wavelength of the transition increases. It turns out that there
are no transitions to be absolutely forbidden, but some transitions are so unli-
kely that populated levels are very long lived. They are said to be metastable.
Levels with lifetimes in excess of one hour have been observed under laboratory
conditions.
Real transitions are not innitely sharp, they are smeared out or broadened.
A particle in a given energy level can have any energy within a nite range.
The frequency spectrum of the spontaneous emitted radiation is described by the
lineshape function, g(ν). This function is usually normalized for a single transition
so that Z ∞ Z ∞
g(ν)dν = 1 ⇒ g(ν)dν = 1 (1.48)
0 −∞

Physics of Atoms and Molecules

12 1 Introduction

0,006
g( )

0,003

0,000
14800 15200 15600
-1
Frequency (cm )

Abbildung 1.7. Lorentzian line shape with central frequency ν0 .

g(ν)dν represents the spectral density of photons emitted spontaneously in the

frequency range ν + dν . The lineshape function g(ν) is usually sharply peaked
near some frequency ν0 , as shown in Fig.(1.2.1), and is frequently written g(ν0 , ν)
to highlight this. The integration limits ranging from −∞ to ∞ because then g(ν)
can be viewed as the Fourier transform of a real function of time. For a plane
wave we introduce the concept of intensity, I , with units W m−2 . The intensity is
the average amount of energy per second transported across the unit area in the
direction of the wave. The spectral distribution of the intensity, I(ν), is related
to the total intensity, I0 , by
I(ν) = I0 g(ν) (1.49)
Note, that real plane waves do not exist. Similarly, the spectral distribution of
the power, W (ν), is related to the total power, W0 by

W (ν) = W0 g(ν) (1.50)

For a collection of identical particles with the population density Ni the total
spontaneously emitted power per frequency interval is

W (ν) = Ni Ai hνg(ν) (1.51)

and is decreasing with time if the population density decreases.

1.2.2 Stimulated Emission

In addition to spontaneous emission, transitions can be stimulated by the action
of an external radiation eld, as shown in Fig. (1.2.2). Let the energy density
of the externally applied radiation eld at frequency ν be p(ν) and [p(ν)] =
Jsm−3 . The rate at which stimulated emission occurs is N2 B21
0 0
(ν) where B21 is a
function specic to the transition between the two level 2 → 1 and its frequency

Physics of Atoms and Molecules

1.2 Radiative transitions in a two level system 13

E2 N2
hν12
hν12 hν12
E1 N1
Abbildung 1.8. Process of stimulated emission in a two level system.

dependence is the same as the lineshape function

0
B21 (ν) = B21 g(ν0 , ν). (1.52)

B21 is called the Einstein coecient for stimulated emission. The total rate of
change of population density by stimulated emission is
Z ∞
dN2 0
= −N2 B21 p(ν)dν (1.53)
dt −∞
Z ∞
= −N2 B21 g(ν0 , ν)p(ν)dν (1.54)
−∞

with units of [B21 ] = m3 J −1 s−2 . The energy of a radiation eld p(ν) is related to
the intensity I(ν) by
I(ν)
p(ν) = . (1.55)
c
and the number of photons at given intensity is

I(ν)
Nphotons = . (1.56)
hν
The ideal monochromatic radiation eld at frequency ν21 has an innitely narrow
energy density prole at ν21 and the dierential equation in (1.54) is then
Z ∞
dN2
= −N2 B21 g(ν0 , ν)p21 δ(ν − ν21 )dν = −N2 B21 p21 g(ν0 , ν21 ) (1.57)
dt −∞

Note, the rate of stimulated emissions produced by the input monochromatic

radiation is directly proportional to the value of the lineshape function at the
input frequency. If the stimulating radiation eld has a spectrum that is broad,
and we can assume the energy density p(ν) = p(ν0 ) as constant where g(ν0 , ν) is
signicant, the dierential equation becomes
Z ∞
dN2
= −N2 B21 g(ν0 , ν)p(ν0 )dν = −N2 B21 p(ν). (1.58)
dt −∞

Physics of Atoms and Molecules

14 1 Introduction

1.2.3 Stimulated Absorption

As well as stimulating transitions in a downward direction (2 → 1), transitions
can be stimulated from a lower to a higher level (1 → 2). The rate of such
absorptions is
N1 p(ν)B12 g(ν0 , ν) (1.59)

which gives a result similar to Eq. (1.54)

Z ∞
dN1
= −N1 B12 g(ν0 , ν)p(ν)dν. (1.60)
dt −∞

B12 is a constant specic to the transition and is called the Einstein coecient for
stimulated absorption, with p(ν) the energy density of the stimulating radiation
eld. There is no analog in the absorption process to spontaneous emission, since
a particle cannot spontaneously gain energy without an external energy supply.
In stimulated emission the emitted photon has exactly the same frequency as the
stimulating photon. In absorption the incident photon disappears, as shown in
Fig. (1.2.3). In both processes the particle (atom, molecule, ion, etc.) recoils to
conserve linear momentum.

E2 N2 E2 N'2
hν12
E1 N1 E1 N'1
hν12 = E 2 - E1

Abbildung 1.9. Stimulated absorption in a two level system. The photon absorbed after the
interaction.

1.3 Relation between Einstein coecients

In thermal equilibrium the populations of the two levels in our two level system
N1 and N2 are constant

dN2 dN1 d(N1 + N2 )

= = =0 (1.61)
dt dt dt

and the rates of the transitions between the levels are equal. Since the energy
density of the black-body radiation eld is nearly constant over the range of the
lineshape function where the transitions between level 2 and 1 takes place, we
calculate the dierential equations in thermal equilibrium with (1.44), (1.54) and

Physics of Atoms and Molecules

1.3 Relation between Einstein coecients 15

(1.60) to be
dN2
= −N2 B21 p(ν) − A21 N2 + N1 B12 p(ν) (1.62)
dt
dN1
= −N1 B12 p(ν) + A21 N2 + N2 B21 p(ν) (1.63)
dt
N2 B12 p(ν)
⇒ = (1.64)
N1 B21 p(ν) + A21
The ratio of the population densities of two levels with energy dierence hν is
N2
= e−hν/kB T (1.65)
N1
B12 p(ν)
⇒ e−hν/kB T = (1.66)
B21 p(ν) + A21
−hν/kB T
⇒ p(ν)(B21 e − B12 ) = −A21 e−hν/kB T (1.67)
A21
⇒ p(ν) = hν/k
(1.68)
B12 e BT − B
21
3
8πhν A21
⇒ 3 hν/k T
= hν/k
(1.69)
c (e B − 1) B12 e BT − B
21

The last equality can only be satised if

B12 = B21 (1.70)
and
8πhν 3
A21 = B21 . (1.71)
c3
Equations (1.70) and (1.71) are called Einstein relations. The spontaneous emis-
sion rate, A21 , is independent of external radiation eld and is important for the
energy density p(ν), because with A21 = 0 it follows from (1.68)that p(ν) ≡ 0. No-
te, that contributions of the spontaneous emission increase with ν 3 with respect
to contributions of stimulated emission. This is very important for the develop-
ment of lasers where a population inversion between two levels induce a strong
stimulated emission signal (LASER: Light amplication by stimulated emission
of radiation).

1.3.1 Level Degeneracy

In real systems it frequently happens that a given energy level corresponds to a
number g of dierent arrangements, where g is the degeneracy of the level. The
separate states of the system with the same energy are called sub-levels. Assume
the levels 2 and 1 consist of a number of sub-levels g2 and g1 . For each of the
sub-levels of level 1 and 2 with populations n1 and n2 the ratio of populations is
n2
= e−hν/kB T (1.72)
n1

Physics of Atoms and Molecules

16 1 Introduction

and N1 = g1 n1 , N2 = g2 n2 . Therefore
n2 g1 N2
= (1.73)
n1 g2 N1
and
N2 g2
= e−hν/kB T (1.74)
N1 g1
and from (1.64) it gives

N2 B12 p(ν) g2
= = e−hν/kB T (1.75)
N1 B21 p(ν) + A21 g1
−hν/kB T −hν/kB T
⇒ p(ν)[g2 B21 e − g1 B12 ] = −A21 g2 e (1.76)
g2 A21
⇒ p(ν) = hν/k
(1.77)
g1 B12 e BT − g B
2 21
8πhν 3 1 A21
⇒ = g1 (1.78)
c3 ehν/kB T − 1 B ehν/kB T − B21
g2 12
g1
⇒ B12 = B21 (1.79)
g2
and
A21 8πhν 3
= . (1.80)
B21 c3
It is instructive to examine the relative rates at which spontaneous and stimulated
processes occur in a system at temperature T. The ratio is
A21 1
R= = ehν/kB T − 1 = . (1.81)
B21 p(ν) n(ν)

Here, p(ν) is the black-body radiation eld and the average number of photons
in a mode in the case of black-body radiation is
1
n(ν) = . (1.82)
ehν/kB T −1
For T=300 K and λ = 500 nm or ν = 6 × 1014 s−1 the ratio hν/kB T is

hν 6.626 × 10−34 Js × 6 × 1014 s−1

= ≈ 96. (1.83)
kB T 1.38 × 10−23 JK −1 × 300 K
and so the ratio, R, of spontaneous and stimulated processes

R = e96 − 1 ≈ 5 × 1041 ⇒ n(ν) = 2 × 10−42 (1.84)

is very high. In the visible and near-infrared region spontaneous emission is ge-
nerally dominates unless there are several photons in a mode (n(ν) > 1).

Physics of Atoms and Molecules

1.3 Relation between Einstein coecients 17

In a disperse medium the refractive index varies with frequency. There the
mode density pb(ν) of a black-body cavity has to be modied to

8πν 2 n2 ng
pb(ν) = (1.85)
c30

with
dn
ng = n + ν . (1.86)
dν
The ratio of A21 /B21 changes without changing the ratio of spontaneous and
stimulated processes.
A21 8πhν 3 n2 ng
= . (1.87)
B21 c30

Physics of Atoms and Molecules

18 1 Introduction

1.4 Fouriertransformation
In a multitude of physical and mathematical problems it is necessary to develop
a periodic function f (t) with period of 2π by a sum of trigonometric functions.
∞
a0 X
f (t) = + (ak cos(kt) + bk sin(kt) (1.88)
2 k=1
Periodic processes are known from dynamic properties of music and sounds, the
movement of planets, a pendulum, etc. The trigonometric functions to descri-
be periodic processes are the sine and cosine functions, which have very nice
properties:
sin(ωt) = − sin(−ωt) (1.89)
cos(ωt) = cos(−ωt) (1.90)
π
cos(ωt) = sin(ωt + ) (1.91)
2
sin(ω1 t ± ω2 t) = sin(ω1 t) cos(ω2 t) ± cos(ω1 t) sin(ω2 t) (1.92)
cos(ω1 t ± ω2 t) = cos(ω1 t) cos(ω2 t) ∓ sin(ω1 t) sin(ω2 t) (1.93)
1 = sin2 (ω1 t) + cos2 (ω1 t) (1.94)
sin(2ω1 t) = 2 sin(ω1 t) cos(ω1 t) (1.95)
cos(2ω1 t) = cos2 (ω1 t) − sin2 (ω1 t) (1.96)
1
sin2 (ω1 t) = (1 − cos(2ω1 t)) (1.97)
2r
ω1 t 1 − cos(ω1 t)
sin( ) = ± (1.98)
2 r 2
ω1 t 1 + cos(ω1 t)
cos( ) = ± (1.99)
2 2
ω1 t + ω2 t ω1 t − ω2 t
sin(ω1 t) + sin(ω2 t) = 2 sin( ) cos( ) (1.100)
2 2
and properties of the derivatives:
d
sin(ω1 t) = ω1 cos(ω1 t) (1.101)
dt
d
cos(ω1 t) = −ω1 sin(ω1 t) (1.102)
dt
−
→ −
→ → −
→ −
→ →
5 sin( k · − x ) = k cos( k · − x) (1.103)
iω1 t
e = cos(ω1 t) + i sin(ω1 t) (1.104)
1 £ iω1 t ¤
cos(ω1 t) = e + e−iω1 t (1.105)
2
1 £ iω1 t ¤
sin(ω1 t) = e − e−iω1 t (1.106)
Z 2i
1 1
cos2 (ω1 t)dt = t+ sin(2ω1 t). (1.107)
2 4ω1

Physics of Atoms and Molecules

1.4 Fouriertransformation 19

In 2
n Hilbert space (L ) with period 2π the functions o
√1 , √1 sin(ω1 t), √1 cos(ω1 t), √1 sin(2ω1 t), √1 cos(2ω1 t), . . . are eigenvectors
2π π π π π
and span the whole Hilbert space. They form an orthogonal and normalized
basis, since

ZT
1
sin(ω1 t) sin(ω2 t)dt = δ(ω1 − ω2 ). (1.108)
T
−T

For every periodic function g(t), which is a member of L2 with period 2π , the
deviation between g(t) and the trigonometric representation fn (t) becomes

Zπ
lim (g(t) − fn (t))2 dt = 0 (1.109)
n→∞
−π

Equation
P (1.109) describes the convergence in the quadratic average. The Fourier
sum n fn (t) converges only for every single point limn→∞ fn (t) = f (t) if the
function g(t) is continous. If the function g(t) has points of discontinouity at tu ,
then the Fourier sum becomes at tu (Dirichlet):

g(tu + h) + g(tu − h)
f (tu ) = lim (1.110)
h→0 2

An example for this is plotted in Fig. 1.4h. If every periodic continous function
can be expressed by the Fourier sum, the question is how to get the Fourier
coecients ak and bk . In Fig. 1.4 we tried to guess the Fourier coecients. To

0,5 a = - 0.4
Amplitude

0
1,0
a = 0.25 b = - 0.15
1 1
0,5

0,0 0,0

-0,5

-1,0
-0,5
0 1 2 3 4 5 6 0 1 2 3 4 5 6

Time in s

Abbildung 1.10. First steps of representing an arbitrary periodic function (black), as a sum
of trigonometric functions.

calculate the Fourier coecients we can make use of the orthogonality relation.
We take the function g(t) multiply it with the individual eigenfunctions, and

Physics of Atoms and Molecules

20 1 Introduction

integrate over the period:

Zπ
1
ak = g(t) cos(kt)dt, k = 0, 1, 2, . . . (1.111)
π
−π
Zπ
1
bk = g(t) sin(kt)dt, k = 1, 2, . . . (1.112)
π
−π

This is nothing else than projecting out the contribution of the eigenfunction
analog to (a, b, c) • ex = a.

The Fourier coecients of periodic function over the interval (−T, . . . , T ) is given
by
ZT
1 kπt0 0
ak = g(t0 ) cos( )dt , k = 0, 1, 2, . . . (1.113)
T T
−T
ZT
1 0 kπt0 0
bk = g(t ) sin( )dt , k = 1, 2, . . . (1.114)
T T
−T

Summing up the contributions of the Fourier coecients gives smooth functions

with negligible oscillatory features (Fig. 1.4). Note, if the Fourier coecients are
ak and bk , the phase ϕk is given by:
A2k = a2k + b2k (1.115)
ak = Ak cos ϕk (1.116)
bk = Ak sin ϕk (1.117)
µ ¶
bk
ϕk = tan (1.118)
ak
Note, the Fourier sum can be written as
X 2π 2π
f (t) = a0 + Ak cos ϕk cos( kt) + Ak sin ϕk sin( kt) (1.119)
k
T T
X 2π
= a0 + Ak cos( kt − ϕk ) (1.120)
k
T

Fourier sums can be expanded to nonperiodic functions if the following relation

holds:
Z∞
|g(t)|dt < ∞. (1.121)
−∞

Physics of Atoms and Molecules

1.4 Fouriertransformation 21

a) b)
1,0 2

0,5 1

0,0 0

-0,5 -1

-1,0 -2

0 2 4 6 8 10 12 0 2 4 6 8 10 12

c) d) 2

1
Amplitude

0 0

-1

-1

-2

0 2 4 6 8 10 12 0 2 4 6 8 10 12

4 e) f) 2

2 1

0 0

-1

-2 -1

-3

-4 -2

0 2 4 6 8 10 12 0 2 4 6 8 10 12

10
g) h) 2
8

1
4

0 0

-2

-4
-1

-6

-8
-2

-10

0 2 4 6 8 10 12 0 2 4 6 8 10 12

Time in s

Abbildung 1.11. a) Sine functions with frequencies 1 (black) and

P1
2 (red); b) sum of sine
functions of a) (blue) and weighted sum of sine functions of a) with k sin(kt) (black); c) Sine
k
functions with frequencies 3 and 4; d) sum of functions in c) and weighted sum of functions in
c); e) sum of functions with frequencies 1 to 4; f) weighted sum of functions with frequencies 1
P12 P
12
1
to 4; g) same as e) but with 12 frequencies sin(kt); h) k sin(kt).
k=1 k=1

This means, that the function becomes zero for t → ∞ and t → −∞. These
points are the analogue points to the points at −T and T for periodic functions.
The continous Fourier transformation fb(t) of f (t) is given by:

Z∞
1
f (t) = √ fb(ω)e−iωt dω (1.122)
2π
−∞
Z∞
1
fb(ω) = √ f (t)eiωt dt (1.123)
2π
−∞

The Fourier transformation fb(ω) is given by the Cosine Fourier transformation

Physics of Atoms and Molecules

22 1 Introduction

1,0

ak
0,8

/2
1
]
2

k
0,6

+b bk
2
Y

k
a
=[
0,4
A
k

0,2

0,0
0,0 0,2 0,4 0,6 0,8 1,0

Abbildung 1.12. Representation of the phase ϕ.

and the Sine Fourier transformation:

r Z∞
2
fc (t) = fbc (ω) cos(ωt)dω (1.124)
π
0
r Z∞
2
fbc (ω) = fc (t) cos(ωt)dt (1.125)
π
0
r Z∞
2
fs (t) = fbs (ω) sin(ωt)dω (1.126)
π
0
r Z∞
2
fbs (ω) = fs (t) sin(ωt)dt (1.127)
π
0
(1.128)

For an even function f(t) the Fourier transformation is given by fb(t) = fbc (t),
for an odd function the Fourier transformation is given by fb(t) = ifbs (t). Every
function f (t) can be written as a sum of an even and an odd function:

f (t) = g(t) + h(t) (1.129)

1
g(t) = [f (t) + f (−t)] (1.130)
2
1
h(t) = [f (t) − f (−t)] (1.131)
2
fb(ω) = gbc (ω) + ibhs (ω) (1.132)

Physics of Atoms and Molecules

1.4 Fouriertransformation 23

Therefore, we can concentrate on the Cosine and Sine Fourier transformation.

The Fourier transformation has additional importance for the folding technique
of functions:
Z∞
Cf g (t) = f (τ )g(t − τ )dτ = (f ? g)(t) (1.133)
−∞
Z∞
1
fb(ω)b
g (ω) = (f ? g)(t)eiωt dt (1.134)
2π
−∞

Some examples of Fourier transformations are:

1 ω2
⇒ fb(ω) = √ a− 2 e− 4a
2 1
f (t) = e−at (1.135)
2
r
2 a
f (t) = e−at ⇒ fb(ω) = (1.136)
π a + ω2
2
½ ¾ r
1 , |t| < a 2 sin(aω)
f (t) = ⇒ fb(ω) = 2 (1.137)
0 , else π ω

Numeric Fourier transformations are presented in the following gures. Note, that
the intensity amplitudes in the time domain are given by the power amplitudes
A(ω)A∗ (ω) of the Fourier transformation. Information on the timing of processes
are reected in the phase, given by the real and imaginary part of the amplitudes.
Short processes in time result in broad features in frequency and vice versa.

Physics of Atoms and Molecules

24 1 Introduction

1
A(t)

−1
0 0.2 0.4 0.6 0.8 1
Time (s)
1
A(w)

−1
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0.5

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.13. Cosine function with frequency of 50 Hz and amplitude 1.

A phase shift of π/2 introduce a sign change in the imaginary amplitude and
a vanishing of the real amplitude (Fig. 1.4)

1
A(t)

−1
0 0.2 0.4 0.6 0.8 1
Time (s)
1
A(w)

−1
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0.5

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.14. Cosine function with frequency of 50 Hz, amplitude 1 and phase π/2.

Physics of Atoms and Molecules

1.4 Fouriertransformation 25

For a beating you need at least two frequencies (Fig. 1.4).

2
A(t)

−2
0 0.2 0.4 0.6 0.8 1
Time (s)
2
A(w)

−2
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.15. Cosine functions with frequencies of 48 and 50 Hz and amplitudes of 1 and
1, respectively.

2
A(t)

−2
0 0.2 0.4 0.6 0.8 1
Time (s)
1
A(w)

−1
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0.5

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.16. Cosine functions with frequencies of 40 and 50 Hz and amplitudes of 0.7 and
1, respectively.

Physics of Atoms and Molecules

26 1 Introduction

A phase shift of π introduce a sign change in the real amplitude and a change
in the imaginary amplitude (Fig. 1.4).

2
A(t)

−2
0 0.2 0.4 0.6 0.8 1
Time (s)
1
A(w)

−1
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.17. Cosine functions with frequencies of 40 and 50 Hz and amplitudes of 0.7 and
1, respectively. The 40 Hz Cosine oscillation is phase shifted by π .

A phase shift of π/2 introduce a strong change in the imaginary amplitude and
a vanishing in the real amplitude (Fig. 1.4).

2
A(t)

−2
0 0.2 0.4 0.6 0.8 1
Time (s)
1
A(w)

−1
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0.5

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.18. Cosine functions with frequencies of 40 and 50 Hz and amplitudes of 0.7 and
1, respectively. The 40 Hz Cosine oscillation is phase shifted by π/2.

Physics of Atoms and Molecules

1.4 Fouriertransformation 27

If the frequencies are multiples of each other the information of the amplitude
can be represented by the imaginary amplitude and the real amplitude of the
multiple frequency. Note, a quarter of the higher frequency is the lower one, and
a quarter of 2π is π/2 (Fig. 1.4). A phase shift of π/2 reverses again the real and

2
A(t)

−2
0 0.2 0.4 0.6 0.8 1
Time (s)
1
A(w)

−1
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0.5

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.19. Cosine functions with frequencies of 12.5 and 50 Hz and amplitudes of 1 and
1, respectively.

imaginary part of the 12.5 Hz amplitudes (Fig. 1.4).

2
A(t)

−2
0 0.2 0.4 0.6 0.8 1
Time (s)
1
A(w)

−1
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.20. Cosine functions with frequencies of 12.5 and 50 Hz and amplitudes of 1 and
1, respectively. The 12.5 Hz Cosine oscillation is phase shifted by π/2.

Physics of Atoms and Molecules

28 1 Introduction

An exponential rise and decay time of an oscillatory signal induce frequency

broadening of the 50 Hz carrier frequency (Fig. 1.4).

0.5
A(t)

−0.5
0 0.2 0.4 0.6 0.8 1
Time (s)
0.1
A(w)

−0.1
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0.01

0.005

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.21. Carrier frequency of 50 Hz with an exponential rise time of 0.1 s and a decay
time of 0.13 s.

The timing of where the Gaussian envelope function has its maximum is given
by the imaginary amplitude (Fig. 1.4).

1
A(t)

−1
0 0.2 0.4 0.6 0.8 1
Time (s)
0.5
A(w)

−0.5
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0.2

0.1

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.22. One Gaussian envelope function with a width of σ = 0.2 and carrier fre-
quency of 50 Hz, centered around 0.7 s.

Physics of Atoms and Molecules

1.4 Fouriertransformation 29

The imaginary amplitude changes if the timing changes (Fig. 1.4).

1
A(t)

−1
0 0.2 0.4 0.6 0.8 1
Time (s)
0.5
A(w)

−0.5
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0.4

0.2

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.23. One Gaussian envelope function with a width of σ = 0.2 and carrier fre-
quency of 50 Hz, centered around 0.2 s.

The sum of both dierently timed Gaussian functions gives the sum of all con-
tribution, including the imaginary part (Fig. 1.4). It is not trivial to see by the

2
A(t)

−2
0 0.2 0.4 0.6 0.8 1
Time (s)
1
A(w)

−1
0 10 20 30 40 50 60 70
Frequency (1/s)
Re(A(w))*Im(A(w))

0.5

0
0 10 20 30 40 50 60 70
Frequency (1/s)

Abbildung 1.24. Sum of two Gaussian envelope functions centered around 0.2 s and 0.7 s;
widths of σ = 0.2 and carrier frequency of 50 Hz.

real and imaginary part how the timing looks like. On top of that in real Fourier
transformations, which are limited in accuracy, there is a big problem for long
time signals. If for example, in a symphony a single frequency is played all the
time and it should be o for only one second after 20 minutes, this aords enor-
mous accuracy. Here, small phase shifts would not reduce but increase the single
frequency, so that other ways of storing, analyzing, and compressing are used.
Wavelet transformations are very useful for such problems. They form the basis
sets for JPG and PNG formats.

Folding integrals are important in spectroscopy. They describe the signals mea-

Physics of Atoms and Molecules

30 1 Introduction

sured in experiments. For example an exponential function g(t) is folded with the
system response function of the experimental set up r(t) (Fig. 1.4). The measured
signal S(t) is dened by the convolution function:
Z∞
S(t) = g(τ )r(t − τ )dτ (1.138)
−∞

1,0
Amplitude

0,8

0,6

0,4

0,2

0,0

0,0 0,2 0,4 0,6 0,8 1,0

60

50
Amplitude of

the folding

40

30

20

10

-0,1 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 1,1

Time

Abbildung 1.25. Upper panel: Exponentials g(t) and a Gaussian function (black) r(t) for
dierent displacement times ti . Lower panel: Folding function S(t) of the Gaussian with an
exponential. For each displacement in time ti between the exponential and Gaussian function,
the integral over the product is calculated and gives rise to S(ti ). Note, the time scale is reversed
and goes from rihght to left.

Physics of Atoms and Molecules

1.5 Homogeneous Line Broadening 31

1.5 Homogeneous Line Broadening

As shown before the lifetimes of an excited state is given by the natural lifetime
τi = 1/Ai . If we consider many particles with the same lifetime excited simul-
taneously with a coherent light source, they would emit an electric eld of the
decaying excited particles (for t ≥ 0) as

E(t) = E0 e−t/τc cosω0 t. (1.139)

In the case of incoherent excitation the electric eld of a single atom is given by

E(t) = E0 e−t/τc cos(ω0 t + ϕ). (1.140)

The time constant τc describes the damping of the electric eld. The intensity
I(t) emitted by an individual excited particle is

I(t) ∝ |E(t)|2 = E02 e−2t/τc cos2 (ω0 t). (1.141)

For many such particles the total observed intensity is

X X
I(t) = Ik (t) ∝ E02 e−2t/τc cos2 (ω0 t + ϕk ) (1.142)
k k
X E2
0 −2t/τc
= e [1 + cos2(ω0 t + ϕk )] (1.143)
k
2
E02 −2t/τc E02 −t/τi
= k e =k e (1.144)
2 2

We see that the lifetime τi is two times faster than the dephasing time τc = 2τi .
This is because the dephasing time is related to the electric eld amplitude and
the lifetime to the electric eld intensity. The electric eld is now (for t ≥ 0)

E0 −t/2τi iω0 t
E(t) = E0 e−t/2τi cos(ω0 t) = e [e + e−iω0 t ] (1.145)
2
E0 i(ω0 +i/2τi )t
= [e + e−i(ω0 −i/2τi )t ] (1.146)
2

Physics of Atoms and Molecules

32 1 Introduction

The frequency distribution of the signal is given by its Fourier transform E(ω):
Z ∞
1
E(ω) = E(t)e−iωt dt (1.147)
2π −∞
Z
1 ∞ E0 i(ω0 +i/2τi )t
= [e + e−i(ω0 −i/2τi )t ]e−iωt dt (1.148)
π 0 2
Z
E0 ∞ i(ω0 −ω+i/2τi )t
= [e + e−i(ω0 +ω−i/2τi )t ]dt (1.149)
2π 0
E0 1 1
= [ (0 − 1) + (0 − 1)](1.150)
2π i(ω0 − ω + i/2τi ) −i(ω0 + ω − i/2τi )
E0 i i
= [ − ] (1.151)
2π (ω0 − ω + i/2τi ) (ω0 + ω − i/2τi )
E0 i
= [ ] (1.152)
2π (ω0 − ω + i/2τi )
Only frequencies with ω ≥ 0 are physical. The intensity of emitted radiation is
then given by

I(ω) ∝ |E(ω)|2 = E(ω)E ∗ (ω) (1.153)

E02 i(ω0 − ω) + 1/2τi −i(ω0 − ω) + 1/2τi
I(ω) ∝ [ × ] (1.154)
4π 2 (ω0 − ω)2 + (1/2τi )2 (ω0 − ω)2 + (1/2τi )2
E02 1
∝ 2
× (1.155)
4π (ω0 − ω) + (1/2τi )2
2

E02 1
I(ν) ∝ 2
× (1.156)
4π (ν0 − ν) + (1/4πτi )2
2

This type of function is called a Lorentzian. Since natural broadening is the same

1,0

0,8
g( )

0,6
FWHM

0,4

0,2

0,0
14800 15000 15200
-1
Frequency (cm )

Abbildung 1.26. Lorentzian line shape with FWHM at ∆ν = 1

2πτi .

for each particle, this broadening mechanism is called homogeneous broadening.

Physics of Atoms and Molecules

1.5 Homogeneous Line Broadening 33

The full width at half maximum height (FWHM) of this function is found from
half intensity points of I(ν) that occur at frequencies ν± 1 as shown in Fig. (1.5),
2
where
1 2
( ) = (ν± 1 − ν0 )2 . (1.157)
4πτi 2

The FWHM is given by

P
1 Aij
j
∆ν = = with ∆ν = ν+ 1 − ν− 1 (1.158)
2πτi 2π 2 2

The normalized form of the Lorentzian lineshape function for natural broadening
(Fig. 1.5) is given by

(2/π∆ν)
g(ν)LN = . (1.159)
1 + [2(ν − ν0 )/∆ν]2

Other mechanisms of homogeneous broadening are for example:

ˆ Soft collisions of phonons with the crystal particles, which does not abruptly
terminate the lifetime

ˆ Pressure broadening in gaseous and liquid phase by soft collisions with

neutral or charged particles

ˆ Stark broadening by a varying external eld perturbing the energy levels of

the particle

ˆ Resonance interactions reducing the lifetime by energy exchange with neigh-

boring particles.

The eects mention above induce phase shifts resulting in a destruction of the
macroscopic coherence of the ensemble, without reducing the population density
of the involved levels. If Xij = 1/τ2∗ is the rate per particle per unit volume
by which collisions disturb the macroscopic coherence of the ensemble of level i
and Aij = 1/τi is the intensity rate per particle per unit volume for spontaneous
emission to state j , we can deduce the decoherence rate τc or dephasing time τ2
by
1 1 X 1 1 1
= = ( Aij + Xij ) = + ∗. (1.160)
τc τ2 j
2 2τi τ2

The rate τ2∗ is called the pure dephasing time.

When the environment or properties of particles in an emitting sample are not

Physics of Atoms and Molecules

34 1 Introduction

identical, inhomogeneous broadening occurs. The normalized lineshape function

of an inhomogeneous broadened transition is given by
r
2 ln 2 2
g(ν)GN = × e−ln 2×[2(ν−ν0 )/∆ν] (1.161)
∆ν π
This is a Gaussian function with FWHM of ∆ν . In a gaseous phase an inhomoge-
neous broadening mechanism is the Doppler broadening. In the condensed phase
dierent interactions to the environment in the non-emitting ground state and
the emitting state can lead to inhomogeneous broadening.

Generally in spectroscopic experiments, the investigator disturbs the material

system with an external perturbation (force), such as exciting a particle into an
emitting state. How long a given property of the system persists until it is avera-
ged out by the microscopic motion of the particles in the system is qualitatively
described by a time-correlation function. The time-correlation function CAB (t, t0 )
is the product of the pair of dynamic variables A(t) and B(t0 ) averaged over an
equilibrium canonical ensemble:

CAB (t, t0 ) = hA(t)B(t0 )iE (1.162)

The autocorrelation function CA (t) of a dynamic variable A(t) is dened by

CA (t) = hA(t)A(0)iE (1.163)

The time dependence of the dynamic variable A(t) is determined by the pertur-
bations that arise from thermal motion and molecular interactions. These per-
turbations induce uctuations of positions and momenta of particles resulting in
phase changes of A(t) over time. The time dependence of the dynamic variable
A(t) is very complex, because the environment, consisting of a large number of
atoms and molecules, uctuates in a complex fashion. The time dependence of
A(t) will generally resemble a stochastic noise pattern generated by the inter-
action of various degrees of freedom (Fig. 1.5). A simple example of treating a
stochastic process is given by the equation
∂A(t)
= iΩ(t)A(t) (1.164)
∂t
where the time dependence of the frequency Ω(t) consists of an average part
ω0 = hΩ(t)iE and a uctuating part δΩ(t). If

Ω(t) = ω0 + δΩ(t) (1.165)

then the average of the uctuations is

hδΩ(t)iE = 0. (1.166)

Physics of Atoms and Molecules

1.5 Homogeneous Line Broadening 35

Abbildung 1.27. Random uctuations in time of the dynamic variable A.

Equations (1.164) and (1.165) dene the time dependence of dynamic variable
A(t) and Ω(t) in terms of ω0 and δΩ(t). The dynamics characterizing A(t) are
now incorporated in the random function δΩ(t). If δΩ(t) is known, it is possible
to extract dynamic information from A(t) and vice versa. The random variable
δΩ(t) is characterized by the variance, dened as
p p
∆ = hδΩ2 (0)iE = C2 (δΩ(0)) (1.167)
and by the correlation time τc , given by
Z ∞ Z ∞
hδΩ(t)δΩ(0)iE 1
τc = dt = 2 hδΩ(t)δΩ(0)iE dt (1.168)
0 hδΩ(0)δΩ(0)iE ∆ 0
The correlation time τc is directly related to the random frequency or energy
uctuations of the system and the time-correlation function CA (t) is related to the
line-shape function IA (ω). Furthermore, ∆ is a measure of the coupling strength
to the perturbers.

Up to now we focussed on ensemble average time-correlation functions, but in a

measurement the dynamic variable A(t) is averaged over an interval of time from
t0 to t0 + T . In an equilibrium system the average is independent of the initial
time t0 at which the measurement started. Thus, the property associated with
A(t) can be calculated to
Z Z
1 t0 +T 1 T
A = lim A(t)dt = lim A(t)dt. (1.169)
T →∞ T t T →∞ T 0
0

Temporal coherence exists if the value of A(t + τ ) is separated from that of A(t)
by small values of τ . The coherence is lost, however, if τ is much longer than the
correlation or coherence time. A measure of this correlation for the property A(t)
separated by a time t is given by the time-averaged time-correlation function,
which is dened as
Z
1 T
A(t)A(0) = A(t)A = lim A(t + τ )A(τ )dτ (1.170)
T →∞ T 0

Physics of Atoms and Molecules

36 1 Introduction

In spectroscopic experiments the time-averaged time-correlation function is mea-

sured. The ensemble averaged time-correlation function and the time-averaged
time correlation function of a single particle are identical for ergodic systems

CA (t) = hA(t)A(0)iE = A(t)A(0) ≡ hA(t)A(0)ic . (1.171)

The time-correlation function (TCF) has the following properties:

hA2 ic ≥ hA(t)Aic (1.172)

lim hA(t)Aic = hAi2c (1.173)
t→∞

Equation (1.173) follows because for very long separation times the values of
A(t) are no longer correlated, and the average of products becomes the product
of averages. That means, the time-correlation function CA (t) decays in time from
hA2 ic initially to hAi2c for large values of time. The details of the time decay
depend on the dynamical nature of the physical system. When the average hAic
does not vanish, the uctuations, δA, in A can be dened by

δA(t) = A(t) − hA(t)ic (1.174)

and the time-correlation function of δA is then given by

hδA(t)δA(0)ic = hA(t)A(0)ic − hAi2c (1.175)

From this it is evident that the time-correlation function of δA(t) always vanishes
at long waiting times. The probable range of time where the δA(t) values are
correlated can be measured by the correlation or coherence time τc , which is
dened as Z ∞
hδA(t)δA(0)ic
τc = dt (1.176)
0 h|δA(0)|2 ic
If one assumes a markovian process, i.e. no memory of the system, all time-
correlation functions higher than CA (t) are zero. Then the time-correlation func-
tion of δA(t) decays as a single exponential with relaxation time τr , so that

hδA(t)δA(0)ic = h|δA(0)|2 ic × e−t/τr (1.177)

then τc = τr and equation (1.176) provides a proper method of determining the

correlation time, which rapidly vanishes for t > τc .
According to the Wiener-Khintchine Theorem the spectral power density
for the stationary random variable A(t) is dened as the Fourier transform of the
time-correlation function CA (t):
Z ∞
1
IA (ω) = e−iωτ hA(τ )A∗ (0)ic dτ (1.178)
2π −∞

Physics of Atoms and Molecules

2 Waves and particles 37

2 Waves and particles

2.1 Photons
Electromagnetic radiation (light) behaves in interference experiments as if it were
a wave and in other circumstances, such as the photoelectric eect or Compton
eect, it behaves if it were composed of particles. These particles, called photons,
carry the discrete amount of energy E and other properties associated with the
wave.
hc
Energy E = hν = (2.1)
λ
−
→ E hν h
Momentum − →p = ~k = k= k= k (2.2)
c c λ
Rest mass m = 0 (2.3)
h
Spin 1 ~ = (2.4)
2π
There are dierent energy and wavelength units used in the eld of atomic and
molecular processes, such as electron volts [eV ] and wavenumbers [cm−1 ].
12398
E[eV] = ◦ (2.5)
λ[A]
E[eV] = 2.418 × 1014 ν[Hz] (2.6)
104
ν̄[cm−1 ] = (2.7)
λ[µm]
keV
1 me = 9, 11 × 10−31 kg = 511 (2.8)
c2
1 eV = 1.602 × 10−19 J (2.9)

2.2 Compton eect

The Compton eect (A.H. Compton 1923) describes the wavelength shift of in-
cident radiation after the 'collision' of photons or electromagnetic waves with
loosely bound (nearly free) electrons. In the experimental set-up a nearly mo-
nochromatic electromagnetic x-ray beam with (λ0 ≈ 1Å) was used to irradiate
a graphite target. The intensity of the scattered radiation was measured as a
function of scattering angle and wavelength. The results, illustrated in Fig. (2.2),
showed that although part of the scattered radiation had the same wavelength
λ0 as the incident radiation, there was also a second component of wavelength
λ1 , where λ1 > λ0 . The shift in wavelength between the incident and scattered
radiation, ∆λ = λ0 − λ1 called Compton shift could not be explained by the clas-
sical models. Classical models would predict the induced dipoles to oscillate and
emit with the frequency of the incident radiation and thus no wavelength shifts

Physics of Atoms and Molecules

38 2 Waves and particles

Abbildung 2.1. Set up and scattering pattern of photons as a function of wavelength. The
data were taken for graphite crystals.

should be observed ∆λ = 0. The observed shift in wavelength was found to vary

with the angle of scattering and to be proportional to sin2 (θ/2) where θ is the
scattering angle. Moreover, the Compton shift was independent of both the ma-
terial used as a scatterer and λ0 , and the value of the constant of proportionality
was 0.048 × 10−10 m. The results were interpreted with relativistic kinematics and
the use of momentum and energy conservation, with − →
p0 , the momentum of the
−
→ −
→
photon before interaction, p1 and p2 the momentum of the photon and electron,
respectively, after interaction, E0 the energy of the photon, mc2 the rest mass of
the electron (here we assume a free electron with momentum zero), E1 and E2

Physics of Atoms and Molecules

2.2 Compton eect 39

the photon and electron energy, respectively after interaction:

−
→
p0 = −
→
p1 + −
→
p2 (2.10)
q
hc
E0 + me c2 = E1 + E2 = + m2e c4 + p22 c2 (2.11)
λ1
−
→
p2 = −
→ −
→
p0 − p1 (2.12)
p22 = −
→
p2 • −
→
p2 = p20 + p21 − 2p0 p1 cos(θ) (2.13)
q
E0 − E1 = c(p0 − p1 ) = m2e c4 + p22 c2 − me c2 (2.14)
m2e c4 + p22 c2 = m2e c4 + 2me c3 (p0 − p1 ) + c2 (p0 − p1 )2 (2.15)
p22 = (p0 − p1 )2 + 2me c(p0 − p1 ) (2.16)
2me c(p0 − p1 ) = −p20 − p21 + 2p0 p1 + p20 + p21 − 2p0 p1 cos(θ) (2.17)
θ
me c(p0 − p1 ) = p0 p1 (1 − cos(θ)) = 2p0 p1 sin2 ( ) (2.18)
2
1 1 θ
me c( − ) = 2 sin2 ( ) (2.19)
p1 p0 2
me c θ
(λ1 − λ0 ) = 2 sin2 ( ) (2.20)
h 2
h θ θ
∆λ = λ1 − λ0 = 2 sin2 ( ) = 2λc sin2 ( ) ≥ 0. (2.21)
me c 2 2
The wavelength λc = mhe c is called the Compton wavelength, and a photon with
Compton wavelength λc has the energy of the rest mass of the electron me c2 =
511 keV
hc hc · me c
Eph = = = me c2 . (2.22)
λc h
The value of λc = 0.02426 × 10−10 m ≈ 0.024Å is independent of λ0 and the
material of the scatterer and in very good agreement with the observed value.
The existence of the unmodied component of the scattered radiation, which has
the same wavelength λ0 as the incident radiation, can be explained by scattering
from electrons so tightly bound that the entire atom (M À me ) recoils and the
Compton shift is negligible.

The Compton eect is a nice experiment to demonstrate the particle character

of electromagnetic waves. On the other hand the wave behaviour of particles
has also been demonstrated, i.e. by observing the diraction of electrons from
edges of slits (Fig. 2.2) and from scattering by a crystal (Fig. 2.2). According
to de Broglie (1924) all material particles possess wave-like characteristics. The
wavelength associated with particles is given by
h
λ= . (2.23)
p

Physics of Atoms and Molecules

40 2 Waves and particles

Abbildung 2.2. Build up of an interfernce pattern by accumulating single electrons in a two-

slit experiment of Tonomura et al. (Tonomura, J. et al. American Journal of Physics 57, 117,
1989). Number of electrons: (a) 10; (b) 100; (c) 3000; (d) 20000; (e) 70000.

The wave aspect is also reected in Bohr's atomic model, characterized by Bohr's
postulate

2πr = nλ, n = 1, 2, 3, . . . (2.24)

L = rp (2.25)

that the orbital angular momentum L is quantised.

nλp nhp
L= = = n~ (2.26)
2π 2πp

This mechanistic picture, which is an extension to the classical planetary model,

gives reasonable values for the energy levels of atoms, but is limited for deriving
properties.

Abbildung 2.3. Set up and diraction pattern of an electron diraction experiment on crystal
powder (from L. Germer).

me Ze2 2 1 1
En − Em = h(νn − νm ) = − 2
( ) ( 2 − 2) (2.27)
2~ 4πε0 n m
(4πε0 )~2
a0 = = 5.29177 × 10−11 m, Bohr0 s radius (2.28)
me e 2

Physics of Atoms and Molecules

2.2 Compton eect 41

Davisson and Germer conrmed (1925) equation (2.23) by measuring the con-
structive interference of electrons. The electrons were accelerated by a potential
of voltage V, resulting in an electron kinetic energy of E = eV .
p2
= eV (2.29)
2me
p
p = 2me eV (2.30)
h
λ = √ (2.31)
2me eV
◦ 12.3
λ[A] = p (2.32)
E[eV]
At a potential of 100 V the resulting wavelength is λ = 1.23Å, and diraction
eects should appear at d ≈ λ. Davisson and Germer observed the reected low
energy electrons (< 400 eV) from a Ni crystal to show constructive interference
at specic angles θn (LEED: Low energy electron diraction)
nλ = d sin(θn ). (2.33)
At energies of 54 eV the observed angle for n = 1 was found to be θ1 = 50°. With
d = 2.15Å, known from x-ray diraction experiments on Ni one can calculate the
wavelengths to be
◦ ◦
λ = 2.15(sin(50◦ )) A= 1.65 A (2.34)
h 12.3 ◦ ◦
λ= √ = √ A= 1.67 A (2.35)
2me eV 54
Electron diraction experiments in transmission on thin Au and Pt lms were
performed by G.P. Thomson (1927). He used powder samples of very small cry-
stals, so that the orientations of crystal axes were isotropically distributed. The
measured transmission powder spectrum has axial symmetry and shows Debye-
Scherrer rings (Fig. 2.2), similar x-ray diraction experiments. The electron dif-
fraction pattern can be explained with the Bragg condition (2.33). Since the
spatial resolution of experiments depend on the wavelength of the detecting par-
ticles, the short wavelength of electrons, which can be in the range of 1 Å, are
used for electron diraction and electron microscopy experiments resolving struc-
tural details on atomic or molecular scales.

The properties of electrons were always an important issue in physics: J.J. Thom-
son was awarded the Nobel Prize for showing that the electron is a particle. G.P.
Thomson, his son, was awarded the Prize for showing that the electron is a wave.

Neutral particles, such as neutrons and He-atoms, show also wave properties

Physics of Atoms and Molecules

42 2 Waves and particles

with dierent wavelengths, depending on their rest mass and velocity. The ki-
netic energy of neutrons can easily be moderated by scattering in a medium of
specic temperature. Water is an ecient neutron scatterer and is often used to
reduce the average kinetic energy of neutrons to room temperature
3
E = kB T (2.36)
2
1
kB T ≈ eV, at T = 293 K (2.37)
40
A crystal lattice can be used to select a monochromatic wavelength
h h
λn = =q (2.38)
pn 2mn 32 kB T

Since the mass of neutrons is much bigger than the electron mass m n
me
≈ 1800,
thermal neutrons have the appropriate wavelength for structures with distances
◦
or lattice constants of d ≈ 1 . . . 10 A. Neutrons are especially useful for scattering
experiments with hydrogen atoms.

2.3 Plane waves, free particles

For an electromagnetic plane wave or photons the following equations hold
−
→ −
→
E(x, t) = ei( k • x −ωt) (2.39)
−
→
−
→ h hk →
−
p = p= =~k (2.40)
λ 2π
ω
E = hν = h = ~ω (2.41)
2π
i −→ − →
E(x, t) = e ~ ( p • x −Et) (2.42)
E = pc (2.43)
ω = kc (2.44)
Equation (2.39) is a solution of the homogeneous wave equation
∂ 2 E(x, t) 1 ∂ 2 E(x, t)
− = 0, (2.45)
∂x2 c2 ∂t2
since inserting (2.39) gives
1
−k 2 − (−ω 2 ) = 0 (2.46)
c2
and this is the correct dependence of E(p) and ω(k). Note, that for the wave
equation eigenvalues exist for E , p, and p2 , but not for x and t.

Physics of Atoms and Molecules

2.3 Plane waves, free particles 43

For plane material waves or a free particle with non vanishing rest mass we have

−
→ h −
→
p = p=~k (2.47)
λ
E = hν = ~ω (2.48)
p2
E = (2.49)
2m
~k 2
ω = (2.50)
2m
Now we end up in analogy with a material wave function ψ(x, t)
→
− −
→
ψ(x, t) = ei( k • x −ωt) (2.51)
i −
(→
p •−
→
x −Et)
ψ(x, t) = e ~ (2.52)
p2
To guess a wave equation like (2.45) for a free particle in analogy to E = 2m
we
choose the second derivative in x and a rst derivative in t:
∂ψ(x, t) i−
→
p
= ψ(x, t) (2.53)
∂x ~
∂ 2 ψ(x, t) p2
= − ψ(x, t) (2.54)
∂x2 ~2
∂ψ(x, t) iE
= − ψ(x, t). (2.55)
∂t ~
Since we have a relation between energy and momentum (2.49) the guessed wave
equation for a particle, which fulll this equation is

~2 ∂ 2 ψ(x, t) ∂ψ(x, t)
− 2
= i~ . (2.56)
2m ∂x ∂t
Here, also Eigenvalues exist for values of E , p, and p2 , but not for x and t.
−
→
Generalization for a bound particle with a potential V and the force F with
−
→ −
→
F = − ∇V (2.57)

and
p2
+ V (x) = E (2.58)
2m
gives

~2 ∂ 2 ψ(x, t) ∂ψ(x, t)
− 2
+ V (x)ψ(x, t) = i~ , 1−D (2.59)
2m ∂x ∂t
~2 −
→ −
→ →
− ∂ψ(−
→r , t)
− 4ψ( r , t) + V ( r )ψ( r , t) = i~ , 3 − D. (2.60)
2m ∂t

Physics of Atoms and Molecules

44 2 Waves and particles

This equation is called Schrödinger equation. It is a linear and homogeneous

partial dierential equation, and in analogy to the wave equation the superposition
principle holds.

ψ1 , ψ2 solutions ⇒ c1 ψ1 ± c2 ψ2 solutions (2.61)

If ψ1 and ψ2 are solutions of the Schrödinger equation (SE), then the sum and
the dierence of both ψ1 ± ψ2 are also solutions of the SE. This is the fundamen-
tal basis for interference eects. Solutions of the SE are found in L2 : quadratic
integrable functions. Moreover, they should vanish for plus and minus innity.

−
→
A plane wave has momentum − →
p ( k ), but it is not localized. A plane wave has an
innite length, but a precise momentum. Therefore it is not possible to calculate
a probability of nding a plane wave at a specic position (no normalization).
Nevertheless, taken into account the superposition principle every function of L2
can be represented by a weighted sum of plane waves. For convenience purposes,
we go on calculating in one dimensional without limiting the generality:
Z∞
1
ψ(x, t) = √ A(k, ω)ei(kx−ωt) dk. (2.62)
2π
−∞

Every function ψ(x, t) can be described as an innite sum of plane waves (re-
member Fourier transformation), and the amplitude function of the wave vectors
A(k) gives the weighting of the individual plane waves. The amplitude function
can be calculated by the inverse Fourier transformation
Z∞
1
A(k, ω) = √ ψ(x, t)e−i(kx−ωt) dx. (2.63)
2π
−∞

A simple example is the sum of two cosine functions of the same amplitude as
shown in Fig. (2.3).

ψ(x, t) = ei(k1 x−ω1 t) + ei(k2 x−ω2 t) (2.64)

= ei[(
k1 +k2
2 )x−( ω1 +ω2
2 )t] × 2 cos 1 (∆kx − ∆ωt) (2.65)
2
By increasing the number of waves a more localized wavepacket with less spikes
2
can be generated (Fig. 2.3). Since ω = ω(k) with ω(k) = ~k 2m
for a free particle,
a distribution A(k) (see Fig. 2.3) is equivalent to a distribution of frequencies. If
we develop ω(k) around k0 , where ω(k0 ) = ω0 :
dω
ω(k) = ω0 + (k − k0 ) |k + . . . (2.66)
dk 0

Physics of Atoms and Molecules

2.3 Plane waves, free particles 45

Amplitude

-400 -200 0 200 400

Time

Abbildung 2.4. Sum of two cosine waves of the same amplitude and ω1 = 0.1745s−1 and
ω1 = 0.2094s−1 . The amplitude is plotted versus time.

and insert this into (2.62) it becomes

Z∞
1 dω
ψ(x, t) = √ ei(k0 x−ω0 t) A(k, ω)ei[(k−k0 )x−(k−k0 ) dk |k0 t] dk (2.67)
2π
−∞
Z∞
1 dω
ψ(x, t) = √ ei(k0 x−ω0 t) A(k, ω)ei(k−k0 )f (x− dk |k0 t) dk. (2.68)
2π
−∞

This means a wavepacket can be described as a product of a plane wave at k0

and carrier frequency ω0 , which is moving with the phase velocity υph and an
envelope f (x − dω | t), which is moving with the group velocity υgr .
dk k0
ω0
υph = (2.69)
k0

dω
υgr = |k (2.70)
dk 0
− −→
υ→
gr = ∇k ω(k) (2.71)

We can characterize the velocity of the particle or wavepacket by the group ve-
locity υgr :
~k 2 2~k ~k p0 p0
ω(k) = ⇒ υgr = = = ⇒ υgr = (2.72)
2m 2m m m m

Physics of Atoms and Molecules

46 2 Waves and particles

Abbildung 2.5. Left: Sum of several cosine waves of same (blue) and dierent (black) ampli-
tudes. Right: Amplitude distributions.

k0
1,0

0,8

0,6
A(k)

0,4

0,2 k0- k0+

0,0

6000 6500 7000 7500 8000

wave number k

Abbildung 2.6. Fourier components centered around k0 .

That is exactly what we expect for the velocity of a free particle. If we have an
amplitude function of wave vectors A(k) constant in the range from k0 − ∆k to
k0 + ∆k

½ 1
2∆b
, k0 − ∆k < k < k0 + ∆k
A(k) = k (2.73)
0 , else

Physics of Atoms and Molecules

2.3 Plane waves, free particles 47

Then we get
k0Z+∆b
k
1 dω
ψ(x, t) = √ ei(k0 x−ω0 t) A(k)ei(k−k0 )[x− dk |k0 t] dk (2.74)
2π
k0 −∆b
k

Z∆bk
k0 =k−k 1 1 0 dω
= 0
√ ei(k0 x−ω0 t) eik [x− dk0 |0 t] dk 0 (2.75)
2π 2∆b
k
−∆b
k
dω
α=x− dk 0t 1 h 0
i∆bk
= ei(k0 x−ω0 t) √ eik α (2.76)
2∆bkiα 2π −∆b
k
1 h i
b b
= ei(k0 x−ω0 t) √ ei∆kα − e−i∆kα (2.77)
2∆bkiα 2π
1 sin(∆kα)
= ei(k0 x−ω0 t) √ (2.78)
2π ∆kα
1 dω
ψ(x, t) = ei(k0 x−ω0 t) √ sinc(∆k[x − t]) (2.79)
2π dk
Equation (2.79) describes the wavepacket of a particle. As a direct conclusion
2π
from the Fourier transformation taking ∆k as the wave vector width we get:
2π
∆x = ⇒ ∆x · ∆k = 2π ⇒ ∆x · ∆px = 2π~ (2.80)
∆k
This reciprocal relation between the spread in space and momentum (or time

vector

Abbildung 2.7. Left: Amplitude of a sinc function (black) and its intensity (blue). Right: A(k)
of the distribution.

and energy) is fundamental and varies with the distribution functions of A(k).
For all distribution functions it is
~
∆x · ∆px ≥
2
~
∆t · ∆E ≥ . (2.81)
2

Physics of Atoms and Molecules

48 2 Waves and particles

These relations are examples of the uncertainty principle and gives an upper
limit for the localization of space and momentum or time and energy. Note,
space and time are not eigenvalues of the SE, and they are directly connected
with momentum and energy via the dierential equations.

In equation (2.72) the group velocity of a particle is determined by its rest mass.
For photons the rest mass is zero, and the group velocity is given by
dω dkc
υgr =|k 0 = |k = c. (2.82)
dk dk 0
The wavepacket consisting of photons has a group velocity c, and is not sprea-
ding out in time. This situation changes if the photon wavepacket, which is a
temporally short and spectrally broad pulse, is travelling through a dispersive
medium with n = n(λ) and the dielectric constant ². If that is the case, the wave
vector becomes dependent on the frequency
ω2 ω2 2
k 2 (ω) = ²(ω) = n (ω) (2.83)
c2 c2
dk 1 d2 k
k(ω) = k0 + |ω0 (ω − ω0 ) + |ω (ω − ω0 )2 + · · · = k0 + δk (2.84)
dω 2 dω 2 0
µ ¶−1
dk
υgr = |ω (2.85)
dω 0
The term
d2 k 1 dυgr
| ω0 = − 2 dω ω0
| = GVD(ω) (2.86)
dω 2 υgr
is the group velocity dispersion (GVD) parameter and is useful for calculating
the change of transit time for a broadband pulse as a function of frequency
through a medium or an optical setup. Alternatively, it is also possible to look at
the complex phase in the frequency domain and dene the GVD as the second
2
derivative of the phase with respect to frequency. The units of the GVD is [ sm ] or
time
[ length·bandwidth ] = [ cm·ff ss−1 ]. It is useful to represent the properties of the complex
electric eld E + (t) of short light pulses by the product of a complex envelope
function E(t) and a phase term:
1 1
E + (t) = E(t)eiΦ(t) = E(t)eiφ(t) eiω0 t (2.87)
2 2
Here, the plus at the electric eld indicate only positive frequencies, φ(t) is the
time dependent phase and ω0 the frequency expectation value of the electric eld.
In most practical cases of interest the spectral amplitude is centered around ω0
and has non negligible amplitudes only in a small frequency range ∆ω , with
∆ω
¿ 1. (2.88)
ω0

Physics of Atoms and Molecules

2.3 Plane waves, free particles 49

For inequality (2.88) to be satised, the temporal variation of E(t) and φ(t) within
an optical cycle T = ω2π0 has to be small (T ≈ 2 f s for visible radiation). The

Abbildung 2.8. Electric eld, time dependent carrier frequency and spectral amplitude of an
upchriped pulse.

physical meaning of the phase function Γ(t) is the following. The rst derivative
of the phase funtion Γ(t) is the time dependent carrier frequency choosen such
as to minimize the variation of phase φ(t).
d
ω(t) = ω0 + ϕ(t). (2.89)
dt
For dφ
dt
= b = const, a non-zero value of b just means a correction of the carrier
frequency, which is now ω = ω0 + b. For dφdt
= f (t), the carrier frequency varies

Physics of Atoms and Molecules

50 2 Waves and particles

with time and the corresponding pulse is said to be frequency modulated or chir-
2
ped. For ddt2φ > 0, the carrier frequency increases along the pulse, which is then
called up chirped (Fig. 2.3). If the carrier frequency decreases along the pulse it
is called down chirped.

The pulse duration and spectral width of a short light pulse is essential to follow
short (femtosecond) atomic and molecular dynamics. For very short and spec-
trally broad light pulses it is dicult to determine the exact pulse shape of a
light pulse. For single pulses, the typical representative function that is readily
accessible to the experimentalist is the intensity autocorrelation:

Z∞
Aint (τ ) = I(t)I(t − τ )dt. (2.90)
−∞

This is by denition a symmetric function Aint (τ ) = Aint (−τ ) and its Fourier
transform is given by the real function:

e Ie∗ (Ω).
Aint (Ω) = I(Ω) (2.91)

The most commonly cited pulse proles are the Gaussian, for which the temporal
dependence of the eld is:
2
E(t) = E0 e−(t/τG ) (2.92)
and the secant hyperbolic

2
E(t) = E0 sech(t/τs ) = E0 (2.93)
e(t/τs ) + e−(t/τs )
1 1
with the parameters τG = τp √2ln2 and τs = τp 1.76 . Since, the coherent temporal
and spectral characteristics of the elds (and not intensities) are related to each
other through Fourier transforms, the bandwidth ∆ωp and pulse duration τp
cannot vary independently of each other. There is a minimum duration-bandwidth
product:
∆ωp τp = 2π∆νp τp ≥ 2πcB (2.94)
The numerical constant cB is called the duration-bandwidth product and is on
the order of 1, depending on the actual pulse shape. The values of ∆ωp and τp
are given by the FWHM (Full Width at Half Maximum) values. If the pulse
shape is complex one can introduce dierent denitions as for example τp2 =
(ht2 I(t)i − htI(t)i2 )2 , using the second moment.

Physics of Atoms and Molecules

2.3 Plane waves, free particles 51

For some pulse shapes the duration-bandwidth product becomes

Gaussian pulse shapes
2
I(t) ∝ e−1/2(t/τG ) (2.95)
√
τp = τG 2 ln2 (2.96)
)2 /2
I(ω) ∝ e−(ωτG (2.97)
√
2 ln2
∆ωp = (2.98)
τG
cB ≈ 0.441 (2.99)

Sech pulse shapes

I(t) ∝ sech2 (t/τs ) (2.100)
τp ≈ 1.763τs (2.101)
I(ω) ∝ sech2 (πωτs /2) (2.102)
1.122
∆ωp ≈ (2.103)
τs
cB ≈ 0.315 (2.104)

Lorentian pulse shapes

1
I(t) ∝ (2.105)
(1 + (t/τL )2 )2
τp ≈ 1.287τL (2.106)
I(ω) ∝ e−2(|ω|τL ) (2.107)
0.693
∆ωp ≈ (2.108)
τL
cB ≈ 0.142 (2.109)
Note, this equations hold for the intensity proles of a bandwidth limited
√
pulse, and the duration-bandwidth product is given for the elds (factor 2). If
we introduce a linear chirp to a gaussian prole we get:
2
E(t) = E0 e−(1+ia)(t/τG ) (2.110)
dϕ −2at
= (2.111)
dt τG2
2ln2 √
⇒ ∆νp τp = 1 + a2 (2.112)
2π
ω 2 τG
2
e E02 τG2 − 2(1+a
I(ω) ∝ √ e 2)
. (2.113)
1 + a2
Upon introducing linear chirp to a bandwidth limited pulse the pulse duration
will become longer.

Physics of Atoms and Molecules

52 3 Schrödinger equation

3 Schrödinger equation
The SE (3.1) is a partial dierential equation describing the dynamics of the wave
function ψ(−
→
r , t).

~2 −
→ −
→ −
→ ∂ψ(→
−r , t)
− 4ψ( r , t) + V ( r )ψ( r , t) = i~ (3.1)
2m ∂t

To interpret the meaning of the wave function ψ(− →r , t) we must remember that
ψ(−→r , t) is in general a complex function and |ψ(−
→
r , t)| is large where the particle is
likely to be found. The diraction pattern made by light depend on the intensity
of the electric eld, which is ∝ |E(ω, t)|2 . In similar way, M. Born made the
fundamental postulate (1926), that if a particle is described by a wave function
ψ(−→r , t) the probability of nding the particle within the volume element d− →r =
−
→
dxdydz about the point r at time t is given by:

P (−
→
r , t) = |ψ(−
→
r , t)|2 d−
→
r = ψ ∗ (→
−
r , t)ψ(→
−
r , t)d−
→
r. (3.2)

Here P (−
→r , t) is the (position) probability density. Since the probability of nding
a particle somewhere must be unity the normalization relation must hold:

Z∞
ψ ∗ (−
→
r , t)ψ(−
→
r , t)d−
→
r = 1. (3.3)
−∞

Note, that |ψ(−

→
r , t)|2 is the physical quantity that is directly related to the expe-
rimental observable quantities.

3.1 Expectation values

If we consider a normalized wave function ψ(−
→
r , t), then the average or expectation
−
→
value of any function f ( r ) is dened as:

Z∞
hf (−
→
r )i = hψ ∗ (−
→
r , t)|f (−
→
r )ψ(−
→
r , t)i = ψ ∗ (→
−
r , t)f (−
→
r )ψ(−
→
r , t)d→
−
r. (3.4)
−∞

If we want to calculate the probability to nd a particle described by ψ(−

→
r , t) at
the position −
→
r , we have to set f (−
→
r)=− →r , and then the expectation value for
nding the particle at position −
→
r is given by:
Z∞
h−
→
r i = hψ ∗ (−
→
r , t)|−
→
r ψ(−
→
r , t)i = ψ ∗ (→
−
r , t)−
→
r ψ(−
→
r , t)d−
→
r. (3.5)
−∞

Physics of Atoms and Molecules

3.2 Time independent SE 53

Since in experiments all observed quantities are expectation values, the integral
or scalar product hf (−
→
r )i is necessary to connect the SE with physical quantities.
Z
ψ ∗ (−
→
r , t)Φ(−
→
r , t)dV = hψ|Φi (3.6)

The square integrable functions ψ ∈ X and Φ ∈ X together with the following

properties of the scalar product h | i dene a Hilbert space:

hΦ1 + Φ2 |ψi = hΦ1 |ψi + hΦ2 |ψi, ∀Φi , ψ ∈ X (3.7)

hλΦ|ψi = λ∗ hΦ|ψi, ∀Φ, ψ ∈ X, λ ∈ C (3.8)
hΦ|ψi = hψ|Φi∗ , ∀Φ, ψ ∈ X, (3.9)
hψ|ψi ≥ 0, ∀ψ ∈ X (3.10)
hψ|ψi = 0 ⇔ ψ = 0. (3.11)

The property of becoming complex conjugated upon exchanging the arguments

is called sesqi linear. From these properties the Schwarz relation follows:

|hΦ|ψi|2 ≤ hΦ|Φihψ|ψi (3.12)

|hΦ|ψi|2 = hΦ|Φihψ|ψi ⇔ Φ = λψ. (3.13)

Note, that the scalar product can also be written as a norm kψk, which is directly
connected to a metric. A metric describes specic properties of the space such as
calculating and comparing distances.
p
kψk ≡ |hψ|ψi| (3.14)
kΦ + ψk2 = hΦ + ψ|Φ + ψi (3.15)
2 2
kΦ + ψk ≤ (kΦk + kψk) (3.16)

Note, the scalar product hψ|Aϕi = hψ|eAeϕi

e is invariant to the Fourier transforma-
e
tion, with ψ , ϕ e
e, and A the Fourier transformations of ψ , ϕ, and A, respectively.

3.2 Time independent SE

The SE is very often abbreviated by

∂ →
r , t) = i~ ψ(−
Hψ(−
→ r , t) (3.17)
∂t
Hψ(−
→
r , t) = E(− →r , t)ψ(−→r , t) (3.18)
2
~
H = − 4 + V (− →r , t) (3.19)
2m
−
→ ~− →
p = ∇ (3.20)
i

Physics of Atoms and Molecules

54 3 Schrödinger equation

where H is called the Hamilton operator with the kinetic energy operator (rst
term) and the potential operator (second term). The Hamilton operator is known
from classical mechanics to describe the energy, and now it is used to describe
the quantum mechanical energy by going from variables to operators.
~−→ →
H(−→
p ,−→q ) 99K H( ∇, − r ). (3.21)
i
If the Potential operator V (−
→
r , t) = V (→−
r ) does not depend on time, we can use
the product solution method to solve the SE:
ψ(→
−
r , t) = f (t)ϕ(−
→
r) (3.22)
∂
Hψ(→ −r , t) = f (t)Hϕ(− → r ) = ϕ(− →r )i~ f (t) (3.23)
∂t
1 →
− 1 ∂
−
→ Hϕ( r ) = i~ f (t) ∀ r , t; ϕ(−
−
→ →r ), f (t) 6= 0 (3.24)
ϕ( r ) f (t) ∂t
Equation (3.24) is only true for all t and ϕ(− →
r ) if the following holds
1 1 ∂
−
→ Hϕ(− →r) = i~ f (t) = const. = E (3.25)
ϕ( r ) f (t) ∂t
1
⇒ − Hϕ(− →r) = E (3.26)
ϕ(→ r)
1 ∂
⇒ i~ f (t) = E (3.27)
f (t) ∂t
df (t) E
⇒ = dt (3.28)
f (t) i~
iE
⇒ f (t) = f (0)e− ~ t (3.29)
⇒ HϕE (−
→
r ) = EϕE (−→r) (3.30)
With the product solution method we get separate solutions for the space and
time coordinates. As a direct consequence of this the wave function is the di-
rect product of the space wave function ϕE (−→r ) for a specic energy E and the
time wave function or phase function f (t). The eigenfunctions for the space wa-
ve functions are given by equation (3.29) and the time dependence by equati-
on (3.30). Thus, the resulting solution of the SE for a Hamilton operator with
V (→
−
r , t) = V (−
→
r ) is given by

r , t) = e− ~ t ϕE (−
ψ(−
→ →
iE
r ). (3.31)
Equation (3.31) is the solution of the stationary SE with E representing the
energy of the state and ϕE (→
−r ) the time independent wave function. The phase
do not contribute to the probability density and therefore, ϕE has very similar
properties as ψ :
|ψ(−→r , t)|2 = |ϕE (−
→
r , t)|2 (3.32)

Physics of Atoms and Molecules

3.3 Some properties of operators 55

3.3 Some properties of operators

→
−
In the Hilbert space an operator, such as the momentum operator → −p = ~i ∇ ,
denes an operation on the wave function. Some operators U are called unitary,
if

U ∗U = U U ∗ = 1 (3.33)

some operators N are called normal if

N ∗N = N N ∗ (3.34)

and most importantly operators H called hermitian or self adjoint if

H ∗ = H. (3.35)

For two hermitian operators S and T the following rules hold

(S + T )∗ = S∗ + T ∗ (3.36)
(λS)∗ = λ∗ S ∗ (3.37)
(T S)∗ = S ∗T ∗ (3.38)
(T ∗ )∗ = T (3.39)
hT ψ|Φi = hψ|T ∗ Φi (3.40)

The normalization condition is physical meaningful, because the probability to

nd an existing particle in the universe should be one. This probability should
not change with time, and the time derivative of the normalization condition has
to vanish:

d
hψ|ψi = 0 (3.41)
dt
hψ̇|ψi + hψ|ψ̇i = 0 (3.42)
1 1
h Hψ|ψi + hψ| Hψi = 0 (3.43)
i~ i~
1
(−hHψ|ψi + hψ|Hψi) = 0 (3.44)
i~
⇒ hHψ|ψi = hψ|Hψi (3.45)
⇒ H ∗ = H. (3.46)

Physics of Atoms and Molecules

56 3 Schrödinger equation

It follows that the Hamilton operator is hermitian. In addition one can follow
from the time independent normalization condition:

Z
|ψ(→
−
r , t)|2 dV = 1 (3.47)
V
Z
d
|ψ(→ −
r , t)|2 dV = 0 (3.48)
dt
ZV
ψ˙∗ ψ + ψ ∗ ψ̇dV = 0 (3.49)
V
Z
1
[−(Hψ)∗ ψ + ψ ∗ Hψ] dV = 0 (3.50)
i~
V
Z · ¸
1 ~2 −
→
− − 4 + V ( r , t) ψ ∗ ψ +
i~ 2m
V
· ¸
∗ ~2 −
→
ψ − 4 + V ( r , t) ψdV = 0 (3.51)
2m
Z 2
1 ~
[ψ4ψ ∗ − ψ ∗ 4ψ] dV = 0 (3.52)
i~ 2m
V
Z
i~
[ψ ∗ 4ψ − ψ4ψ ∗ ] dV = 0 (3.53)
2m
V
Z
i~ − h ∗−
→ → → i
−
∇ • ψ ∇ψ − ψ ∇ψ ∗ dV = 0 (3.54)
2m
V
Z I
−
→ − →− −
→− −
→
with Gauss ∇ • E (→r )dV = E (→
r )•df (3.55)
V ∂V
Z Z
→
− − −
→ −
→
and Stokes A (→
r ) • d−
→
r = ∇ ×df (3.56)
∂F F
I
i~ h ∗ −
→ → ∗i −
− →
ψ ∇ψ − ψ ∇ψ • d f = 0 (3.57)
2m
∂V
I
→
− − −
→
j (→
r , t) • d f = 0 (3.58)
∂V
i~ h ∗ −
→ → i
− −
→→
ψ ∇ψ − ψ ∇ψ ∗ = j (− r , t) (3.59)
2m
∂% → − →−
⇒ + ∇ • ( j) = 0 (3.60)
∂t

Physics of Atoms and Molecules

3.3 Some properties of operators 57

Equation (3.60) is the continuity equation, describing that the change in density
−
→
% is given by the loss of probability density j .

As mentioned before the expectation value of an operator − →r , given by h−

→
r i or
−
→
f
h r i is an observable, which can be measured in experiments. The uncertainty of
an observable κ is given by its variance (or second weight) ∆κ:
p
∆κ = hκ2 i − hκi2 (3.61)

It is easy to show that for a hermitian operator A

p
∆A ≡ hψ|A2 ψi − hψ|Aψi2 ≥ 0 (3.62)
(∆A)2 = hψ|(A − hAi)2 ψi (3.63)
= hψ|(A2 − 2AhAi + hAi2 )ψi (3.64)
= hψ|A2 ψi − 2hAihAi + hAi2 (3.65)
= hψ|A2 ψi − hψ|Aψi2 (3.66)
= h(A − hAi)ψ|(A − hAi)ψi = k(A − hAik2 ≥ 0. (3.67)

the variance is always positive and zero if ψ is eigenfunction of A and Aψ = aψ

and therefore hAi = a.

Commutators are dened by

[A, B] ≡ AB − BA (3.68)

and if [A, B] = 0 the commutators commute, they are independent of each other,
and have the same eigenfunctions. For the variance of two operators A and B the
following holds:
1
(∆A)(∆B) ≥ |hϕ|[A, B]ϕi| (3.69)
2
To proof equation (3.69) we introduce two operators A' and B' with

A0 ≡ A − hϕ|Aϕi (3.70)
B 0 ≡ B − hϕ|Bϕi (3.71)
⇒ hϕ|A0 ϕi = hϕ|(A − hAi)ϕi (3.72)
= hAi − hAi (3.73)
hϕ|A0 ϕi = 0 (3.74)
hϕ|B 0 ϕi = 0 (3.75)

Physics of Atoms and Molecules

58 3 Schrödinger equation

The variance and commutator of A' and B', which are hermitian has the following
properties:
[A0 , B 0 ] = [A, B] (3.76)
hA0 i=0
(∆A0 )2 = hϕ|(A0 )2 ϕi (3.77)
= hϕ|(A − hAi)2 ϕi (3.78)
= (∆A)2 (3.79)
(∆B 0 )2 = (∆B)2 (3.80)
With this one can show the general uncertainty relation:
(∆A0 )2 (∆B 0 )2 = hϕ|(A0 )2 ϕihϕ|(B 0 )2 ϕi (3.81)
0 2 0 2
= kA ϕk kB ϕk (3.82)
≥ |laA0 ϕ|B 0 ϕi|2 (3.83)
0 0 2
≥ |=(hA ϕ|B ϕi)| (3.84)
1
= | (hA0 ϕ|B 0 ϕi − hA0 ϕ|B 0 ϕi∗ )|2 (3.85)
2i
1
= |(hϕ|A0 B 0 ϕi − hB 0 A0 ϕ|ϕi∗ )|2 (3.86)
4
1
= |(hϕ|A0 B 0 ϕi − hϕ|B 0 A0 ϕi)|2 (3.87)
4
1
= |hϕ|[A0 , B 0 ]ϕi|2 (3.88)
4
1
= |hϕ|[A, B]ϕi|2 (3.89)
4
1
⇒ (∆A0 )2 (∆B 0 )2 ≥ |hϕ|[A, B]ϕi|2 (3.90)
4
1
⇒ (∆A0 )(∆B 0 ) ≥ |hϕ|[A, B]ϕi| (3.91)
2
→
−
For the example of the space operator →−r and the momentum operator − →
p = ~i ∇
equation (3.69) becomes:
~→ −
[−
→
r ,−
→p ]i Φ = [−
→ r , ∇]i Φ (3.92)
i
~ − −
→ − →−
= (→ r ∇ − ∇→ r )i Φ (3.93)
i
~ − −
→ −
→ −
→→
= (→ ri ∇i Φ − Φ∇i −
r i ∇i Φ − −
→ ri ) (3.94)
i
~
= − Φri (3.95)
i
~
⇒ [−
→p ,−
→r ]i = (3.96)
i
−
→ →
− ~
⇒ ∆ pi ∆ ri ≥ (3.97)
2

Physics of Atoms and Molecules

4 Particle in a box 59

So we see that it is impossible to measure the exact location and momentum of

a particle at the same time.

4 Particle in a box
Assume the most simplest way of localizing a particle in a one-dimensional (x-
axis) box of length 2a, with boundary conditions, so that the potential jumps at
the positions x = −a and x = a from zero to innite (Fig. 4).

½ ¾
0 , |x| < a
V (x) = (4.1)
∞ , else

Since at space positions |x| ≥ a the potential is innite, the probability of the
ergy
En

- a 0 +a
Space x (nm)

Abbildung 4.1. Potential function V(x) for a particle in a box.

wave function has to vanish for those positions, i.e. the particle is trapped between
the potential walls. This leads us to the boundary conditions

Ψ(a) = Ψ(−a) = 0 (4.2)

of the SE for a particle in a box with V (x) = 0 inside of the box:

~2 d2
− Ψ(x) = EΨ(x) (4.3)
2m dx2

Physics of Atoms and Molecules

60 4 Particle in a box

This is the SE of a free particle with symmetric and periodic boundary conditions.
We can try to solve the SE equation with the following wave functions:
Ψg = cos(kn x) (4.4)
2 2
~k
⇒ Ψg = EΨg (4.5)
2m
Ψg (a) = cos(kn a) = 0 (4.6)
Ψg (−a) = cos(kn a) = 0 (4.7)
π
⇒ kn a = n , n odd (4.8)
2
Ψu = sin(kn x) (4.9)
Ψu (a) = sin(kn a) = 0 (4.10)
Ψu (−a) = − sin(kn a) = 0 (4.11)
π
⇒ kn a = n , n even (4.12)
2
nπ
⇒ kn = (4.13)
2a
~2 kn2 π 2 ~2 n 2
⇒ En = = , n = 1, 2, . . . (4.14)
2m 8ma2
The solutions of the wave functions for the particle in a box and the energies
have the following properties:
ˆ discrete eigenvalues for the energy En (discrete energy spectrum)
ˆ En ∝ n2 and En ∝ a−2
π 2 ~2
ˆ the lowest energy is E1 = 8ma2
and not zero
ˆ with increasing n, Ψn the number of nodes increases (n-1)
ˆ the stationary states are either even {cos(kn x)} or odd {sin(kn x)}. This
follows from the parity symmetry V (x) = V (−x).
ˆ The eigenfunctions are orthogonal and complete hΨn |Ψm i = κδn,m .
Ra
ˆ the normalization cn has to be chosen such as c2n cos2 ( nπx
2a
)dx = 1
−a

ˆ the position expectation value is zero hxi = 0

Ra
ˆ the momentum expectation value is zero hpx i = 0 = sin( nπx ) ~ d sin( nπx
2a i dx 2a
)dx
−a

ˆ in a classical picture the probability P (x) would be equal for every space
1
position P (x) = 2a , whereas the quantum mechanical description exhibits
positions where the probability is zero (nodes) and positions where the
probability is higher.

Physics of Atoms and Molecules

5 Ground state of the hydrogen atom 61

E =9E
3 1

n=3

Energy
n=2

E =4E
2 1

n=1

E
1

-a 0 +a
Space x (nm)

Abbildung 4.2. Wave functions Ψn with n = 1, 2, 3 of a particle in a box with energies

π 2 ~2
E1 = 8ma2 , E2 , and E3 .

5 Ground state of the hydrogen atom

To test whether the SE gives reasonable answers to physical problems we calculate
the ground state of a hydrogen atom, which is experimentally well characterized.
The potential operator of the hydrogen atom V (− →
r ) = V (r) has radial symmetry,
so that the eigenfunction in the lowest ground state should depend only on r.
With this assumption the part in the 4 operator depending on φ and θ can be
neglected. The eigenfunction corresponding to the lowest energy E1 is ϕ1 (r) and
it should solve the SE:
· ¸
~2
− 4 + V (r) ϕ1 (r) = E1 ϕ1 (r) (5.1)
2m
· µ ¶ ¸
~2 1 ∂ 2∂
− r + V (r) ϕ1 (r) = E1 ϕ1 (r) (5.2)
2m r2 ∂r ∂r
µ 2 ¶
~2 ∂ 2 ∂ Ze2
− + ϕ 1 (r) − ϕ1 (r) = E1 ϕ1 (r) (5.3)
2m ∂r2 r ∂r 4π²0 r
If we try now the wave function ϕ1 (r) ∝ e−r/a , which is square integrable, we get:
· ¸
~2 1 2 Ze2
− − ϕ 1 (r) − ϕ1 (r) = E1 ϕ1 (r) (5.4)
2m a2 ra 4π²0 r
1
The wave function can only be a solution if the r
terms cancel each other:
2~2 Ze2 4π²0 ~2
= ⇒a= (5.5)
2ma 4π²0 Ze2 m
With Z = 1 this is exactly the Bohr's radius a0 , and the energy E1 is given by:
~2 ~2 Z 2 e4 m2 Z 2 e4 m
E1 = − = − = − . (5.6)
2ma2 2m 16π 2 ²20 ~4 32π 2 ²20 ~2

Physics of Atoms and Molecules

62 6 General remarks to solve the SE

The energy E1 = −13.6 eV is the same as calculated by Bohr's model. The wave

1,0

0,8

Amplitude
0,6

0,4

0,2

0,0
0,0 2,0 4,0 6,0

Radial distance r (a )
0

Abbildung 5.1. Wave functions ϕ1 (r) of a hydrogen atom (black) and the probability to nd
an electron between r and r + dr, given by |ϕ1 (r)|2 r2 dr (blue).

function is maximal at r = 0, but due to the polar coordinates, integration over

the volume element dV = r2 drd(cos θ)dϕ introduces an additional r2 dependency,
so that the probability of nding the electron at the nucleus is zero (Fig. 5). The
probability of nding an electron between r and r + dr is given by

P (r, r + dr) ∝ |ϕ1 (r)|2 r2 dr ∝ r2 e−2r/a0 dr. (5.7)

Here, the normalization is missing! To nd the maximum of the probability (not
the expectation value) we have to dierentiate:
d 2 − a2r − 2r 2r2 − a2r
(r e 0 ) = 2re a0 − e 0 =0 (5.8)
dr a0
2r2
⇒ 2r − = 0 (5.9)
a0
⇒ rmax = a0 (5.10)

The energy value E1 and the dimension parameters a0 are still valid, but there is
no electron moving, now we have a stationary system with the angular momentum
start at zero.

6 General remarks to solve the SE

For one dimensional problems we can write the SE as follows:
d2 Ψ(x) 2m
= (V (x) − E)Ψ(x) (6.1)
dx2 ~2
d2 Ψ(x)
The sign of dx2
determines the curvature of Ψ(x) and is given by V (x) − E . In

Physics of Atoms and Molecules

6 General remarks to solve the SE 63

Energy
V(x)-E

V(x)
II

I III

x x
1 2 Space x

Abbildung 6.1. Arbitrary potential V (x) (black) with a given energy (red) and the dierence
of V (x) − E .

a classical potential V (x) (1-D) the probability of nding the particle between x1
and x2 is proportional to p1 = √ 1
(Fig. 6). It is maximal for the points
2m(E−V (x))
x1 and x2 and minimal for the minimum of V (x). In the space regions of I and
III, p1 is imaginary and classically forbidden. For the wave functions the regions
I and III are not forbidden (tunneling eect), but there the term V (x) − E ≥ 0
and the wave function shows exponential character
1p
Ψ ∝ e±κx , with κ = 2m(V (x) − E) (6.2)
~
for slowly varying V(x). In the region II V (x) − E ≤ 0 and the wave function has
oscillatory character.
1p
Ψ ∝ e±iκx , with κ = 2m(E − V (x)) (6.3)
~
d2 Ψ
In region I and III x2
has the same sign as Ψ, so that the wave functions shape

2
is concave with respect to the x-axis. In region II dx2Ψ has the opposite sign as
Ψ, so that the wave functions shape is convex with respect to the x-axis (Fig.

Physics of Atoms and Molecules

64 7 Harmonic oscillator 1-D

6). Convex shapes lead to oscillatory signals. At the positions x1 and x2 where
V (x) = E the second derivative is zero and the sign changes. This holds also
for classical models. Upon introducing the boundary conditions the spectrum
becomes discrete with a non zero lowest energy. Discrete energy eigenvalues next
to each other have eigenfunctions diering in one node, with Ψn has (n-1) nodes.
With increasing n the envelope function of |Ψn |2 gets more and more similar to
the classical probability function p1 . As a result of those dierences in quantum
mechanical versus classical behavior, tunneling eects are possible and in the
lowest state the probability is highest in the minimum of the potential.

7 Harmonic oscillator 1-D

The harmonic oscillator is well known from
q the mechanical spring with spring
→
−
constant k and oscillation frequency ω = m . With Hook's law F = −k −
k →r it is
easy to calculate the potential energy and the Hamilton function of a harmonic
oscillator to be
1 2
V (x) = kx (7.1)
2
~2 d2 Ψ 1 2
H(x) = − + kx Ψ (7.2)
2m dx2 2
HΨ = EΨ (7.3)

We perform a variable transformation

r
mω
y = x (7.4)
~
2E
λ = (7.5)
~ω
With this transformation the SE becomes
µ 2 ¶
d 2
− y Ψλ = −λΨλ (7.6)
dy 2
Now, we calculate the following operators:
µ ¶µ ¶ µ 2 ¶
d d d 2
+y − y Ψλ = − y − 1 Ψλ (7.7)
dy dy dy 2
µ ¶µ ¶ µ 2 ¶
d d d 2
−y + y Ψλ = − y + 1 Ψλ (7.8)
dy dy dy 2
and dene the lowering operator
µ ¶
d
a= +y (7.9)
dy

Physics of Atoms and Molecules

7 Harmonic oscillator 1-D 65

Energy

0,0 0,5 1,0 1,5 2,0

Space or Distance x

Abbildung 7.1. Potential of an one dimensional harmonic oscillator.

and the rising operator µ

¶
+d
a = −y (7.10)
dy
and inserting (7.9) and (7.10) into equations (7.7)and (7.8) we get
µ 2 ¶
d
− y Ψλ = (aa+ + 1)Ψλ
2
(7.11)
dy 2
µ 2 ¶
d
− y Ψλ = (a+ a − 1)Ψλ
2
(7.12)
dy 2
and we can write the SE as follows
aa+ Ψλ = −(λ + 1)Ψλ (7.13)
a+ aΨλ = −(λ − 1)Ψλ (7.14)
(⇒ a+ a − aa+ )Ψλ = 2Ψλ , ∀Ψλ (7.15)
⇒ a+ a − aa+ = 2 (7.16)
If we take the following equation
a+ a(aΨλ ) = (2 + aa+ )(aΨλ ) (7.17)
= 2aΨλ + a(a+ aΨλ ) (7.18)
= 2aΨλ + a(−(λ − 1))Ψλ (7.19)
= −(λ − 3)aΨλ (7.20)
+
a aΦ = −(λ − 3)Φ (7.21)
Thus, the equation for the eigenvalue (λ − 3) is given by
a+ aΨλ−2 = (λ − 3)Ψλ−2 (7.22)
aΨλ = Φ = Ψλ−2 (7.23)
⇒ aΨλ = Ψλ−2 (7.24)

Physics of Atoms and Molecules

66 7 Harmonic oscillator 1-D

Thus, the eigenfunctions of Ψλ−2 can be generated by applying the operator

d
a = dy + y on the eigenfunction Ψλ . With this process one can calculate the
eigenfunctions for eigenvalues of λ−2, λ−4, λ−6, . . . if we have the eigenfunction
Ψλ . In addition, this process stops at the minimal eigenvalue of λ = 2E~ω
, because
the energy of the harmonic oscillator is positive, as shown below.
µ ¶
~2 d2 1 2
− + kx ΨE = EΨE (7.25)
2m dx2 2
Z∞ µ ¶
∗ ~2 d2
⇒ ΨE (x) − ΨE (x)dx + (7.26)
2m dx2
−∞
Z∞ µ ¶ Z∞
1 2
Ψ∗E (x) kx ΨE (x)dx = E Ψ∗E (x)ΨE (x)dx
2
−∞ −∞
¯∞ Z∞
~2 ∗ dΨE (x) ¯¯ dΨ∗E (x) dΨE (x)
− ΨE (x) + dx + (7.27)
2m dx ¯−∞ dx dx
−∞
Z∞ Z∞
1
k Ψ∗E (x)x2 ΨE (x)dx = E Ψ∗E (x)ΨE (x)dx
2
−∞ −∞

In (7.28) the rst term is zero, because the wave function has to vanish for
lim Ψ(x) → 0. The second integral h dΨ | dΨ i ≥ 0 is positive, because of the
dx dx
x→∞
properties of the scalar product. The last integral hΨ|x2 Ψi = hxΨ|xΨi ≥ 0 is also
positive, so that the energy has to be full E ≥ 0. The energy is zero only for the
trivial solution Ψ(x) ≡ 0. So we can conclude that there exists a minimal value
for λ
∃λM (7.28)
aΨλM = 0 (7.29)
+
a aΨλM = −(λM − 1)ΨλM (7.30)
= a+ 0 = 0 (7.31)
ΨλM 6=0
⇒ λM = 1 (7.32)
So the spectrum of the eigenvalues is given by
λ = {1, 3, 5, 7, . . .} = 2n + 1, n = {0, 1, 2, . . .} (7.33)
and the eigenvalues of the energies are calculated to be
2E
λ =
~ω
~ω 1
En = (2n + 1) = (n + )~ω, n = 0, 1, 2, . . . (7.34)
2 2

Physics of Atoms and Molecules

7 Harmonic oscillator 1-D 67

We determined the eigenvalues for the harmonic oscillator without knowing the
eigenfunctions.

In order to determine the eigenfunctions of the harmonic oscillator we make use

of the properties of the a+ operator

a+ Ψλ ∝ Ψλ+2 (7.35)
a+ Ψn ∝ Ψn+1 . (7.36)

If we have identied Ψ0 we can calculate Ψn iteratively, using

µ ¶ µ ¶
+ n d d
Ψn ∝ (a ) Ψ0 ∝ − y ··· − y Ψ0 . (7.37)
dy dy
| {z }
n factors

The rst eigenfunction Ψ0 we can determine as follows

aΨ = 0 (7.38)
µ ¶ 0
d
+ y Ψ0 = 0 (7.39)
dy
dΨ0
⇒ = −ydy (7.40)
Ψ0
y2
⇒ Ψ0 = N0 e− 2 . (7.41)

Here, N0 is the normalization constant to guarantee that hΨ0 |Ψ0 i = 1. The

function Ψ0 is clearly square integrable. Now, we can derive the Ψn
y2
Ψ1 ∝ a+ Ψ0 ∝ 2ye− 2 (7.42)
y2
Ψ2 ∝ (a+ )2 Ψ0 ∝ (4y 2 − 2)e− 2 (7.43)
y2
The eigenfunctions Ψn are products of a gaussian type of function e− 2 with
a Hermite polynomial. The gaussian function is even with respect to y and the
Hermite polynomial is either even or odd, so that the eigenfunctions are also even
or odd. This is a direct consequence of the potentials symmetry V (−x) = V (x).
The general description to derive Hermite polynomials is given by
µ n¶
(−1)n y2 d 2
Hn (y) = p √ e n
e−y (7.44)
n
2 n! π dy
We can determine the unknown factor by make use of the normalization condition
Z∞
2
e−y Hn2 (y)dy = 1 (7.45)
−∞

Physics of Atoms and Molecules

68 7 Harmonic oscillator 1-D

If we transform y back we get Ψn (x)

r r
mω − mωx2 mω
Ψn (x) = 4
e 2~ Hn (x ) (7.46)
~ ~

~ω
As a result the eigenfunction Ψ0 (x) of the lowest energy eigenvalue E0 = 2
has
the following form
r
mω − mωx2
Ψ0 (x) = 4
e 2~ . (7.47)
~

This is a gaussian function with the maximum of |Ψ0 (x)|2 at x = 0. At this

point the space probability density is highest. In a classical picture the space
probability density would be minimal at x = 0. Hermite polynomials

q µ n¶
√ d
n b n y2
Hn (y) 2 n! π = Hn (y) = (−1) e e−y
2
(7.48)
dy n

Hb 0 (y) = 1 (7.49)
b 1 (y) = 2y
H (7.50)
b 2 (y) = 4y 2 − 2
H (7.51)
b 3 (y) = 8y 3 − 12y
H (7.52)
b 4 (y) = 16y − 48y + 12
H 4 2
(7.53)
b 5 (y) = 32y 5 − 160y 3 + 120y
H (7.54)
b 6 (y) = 64y 6 − 480y 4 + 720y 2 − 120
H (7.55)
b 7 (y) = 128y − 1344y + 3360y − 1680y
H 7 5 3
(7.56)
b 8 (y) = 256y 8 − 3584y 6 + 13440y 4 − 13440y 2 + 1680 (7.57)
H

The Hermite polynomials have the following properties:

Ḧn − 2y Ḣn + 2nHn = 0 (7.58)

2yHn − 2nHn−1 = Hn+1 (7.59)
Z∞
2
Hn (y)Hm (y)ey dy = δn,m (7.60)
−∞

Physics of Atoms and Molecules

7 Harmonic oscillator 1-D 69

Using equation (7.58) we obtain for a+ aΨn the equation

µ ¶
+ d2 2
a aΨn = 2
− y + 1 Hn e−y /2
2
(7.61)
dy
d2 2 2 2
= 2
Hn e−y /2 − y 2 Hn e−y /2 + Hn e−y /2
dy
d ³ 2 2
´ 2 2
= Ḣn e−y /2 − yHn e−y /2 − y 2 Hn e−y /2 + Hn e−y /2
dy
³ 2 2 2 2 2
´
= Ḧn e−y /2 − y Ḣn e−y /2 − y Ḣn e−y /2 − Hn e−y /2 + y 2 Hn e−y /2
2 2
−y 2 Hn e−y /2 + Hn e−y /2
2 2
= Ḧn e−y /2 − 2y Ḣn e−y /2
= −2nHn (7.62)

For further information see M.Abramowitz and I.A. Stegun, Handbook of ma-
thematical functions, Dover (1965), Chapter 22. The solutions of the wave func-
tions for a particle in a box and for the harmonic oscillator (Fig. 7) have some
similarities:

n =8

-6 -3 3 6
n=5
E
3
=9E1 n=4

n =3
n =3
Energy
Energy

n =2
E =4E1
2
n =2

n =1 n =1

E
1
n =0

-a 0 +a -4 -3 -2 -1 0 1 2 3 4
Space x (nm) Space x

Abbildung 7.2. Comparison of wave functions of the particle in a box and the harmonic
oscillator.

ˆ Discrete solutions and eigenenergies.

ˆ The n th wave function has n nodes.

Physics of Atoms and Molecules

70 7 Harmonic oscillator 1-D

~ω
ˆ The lowest energy has a nite positive value E0 = 2
≥ 0. Result of the
uncertainty relation.

ˆ The eigenfunctions of dierent eigenvalues are orthogonal. General property

of the Hamilton operator.

7.1 Conserved quantities

In general physical quantities, which can be measured, give real values and are
therefore related to hermitian operators:
µZ ¶∗ Z
∗
Φ AΦdV = Φ∗ AΦdV (7.63)

ˆ Operator hermitian

ˆ Eigenfunctions orthonormal AΦn = aΦn . RFor degenerate energies the eigen-

functions can be chosen as orthonormal. Ψ∗m Ψn dV = δmn .

ˆ The eigenfunctions form a complete set of orthonormal

P functions if every
function of the space can be developed as Ψ(−
→
r ) cn ψn (−
→r ).
n

The closure relation (7.69)follows from the complete set of the eigenfunctions
ψn ( →
−
r)
X
Ψ(−→r) = c n ψn ( →
−
r) (7.64)
n
X
hψm (−
→
r )|Ψ(−
→
r )i = cn hψm (−
→
r )|ψn (−
→
r )i (7.65)
n
X
= cn δm,n = cm=n (7.66)
n
X −
→ −
→
⇒ Ψ(−
→ hψk ( r0 )|Ψ( r0 )iψk (−
→
!
r) = r) (7.67)
k
XZ −
→ −
→ − →
= ψk∗ ( r0 )ψk (−
→
r )Ψ( r0 )d r0 (7.68)
k
X −
→ →
−
⇒ ψk ( r0 )ψk (−
→
r ) = δ(→
−
r − r0 ) (7.69)
k

A very important issue in physics is the identication of conserved quantities.

Since we are interested in expectation values of a given operator u, the time
dependence of hui is given by:

dhui 1 ∂u
= h[u, H]i + h i (7.70)
dt i~ ∂t

Physics of Atoms and Molecules

7.1 Conserved quantities 71

du
If dt
= 0 then hui is conserved, and it is a conserved quantity if u and H commute:
[u, H] = 0 (7.71)
Especially, if at time t = 0 the system is in an eigenstate of u with eigenvalue
ui , the system will stay in that state. The subscript i of the eigenvalue ui is
called good quantum number, because it is a conserved quantity and describes
the quantum system. The eigenvalues ui are possible observable quantities. If we
want to indicate that the eigenfunctions of the energy En are at the same time
eigenfunctions of the operator u with eigenvalue um , we have to introduce two
indices:
HΨn,m = En Ψn,m (7.72)
uΨn,m = um Ψn,m (7.73)
On the other hand, if there are eigenfunctions Ψn,m of u and H , then it follows:
uHΨn,m = uEn Ψn,m = En uΨn,m = En um Ψn,m (7.74)
HuΨn,m = Hum Ψn,m = um HΨn,m = um En Ψn,m (7.75)
⇒ (uH − Hu)Ψn,m = 0 (7.76)
⇒ [u, H] = 0 (7.77)
If equation (7.77) is given it is easy to show that there are eigenfunctions Ψn,m
of u and H .

The same holds for every set of operators Ai . If [Ai , Aj ] = 0 then Ai and Aj have
the same eigenfunctions. If additionally, [Ai , H] = 0 holds, and therefore hAi i is
constant, the operator Ai has the same eigenfunctions as the energy operator. To
characterize the energy eigenfunctions in the best possible way, we are looking
for the maximal set of operators Aj , which commute with the Hamilton operator
and with themselves:
[Ai , Aj ] = 0, ∀i, j (7.78)
[Ai , H] = 0, ∀i (7.79)
That means there are eigenfunctions of H, A1 , A2 , . . . at the same time and the
quantum numbers of those operators are good quantum numbers describing the
system.

A system with a potential V (−→r ) = V (r) only dependent on the length of the radi-
−
→
us | r |, such as the hydrogen atom, there are three pairwise commuting operators
describing the system:
−
→
H, L 2 , and Lz (7.80)

Physics of Atoms and Molecules

72 7 Harmonic oscillator 1-D

→
−
Thus, h L 2 i and hLz i are conserved and there are simultaneous eigenfunctions
−
→
of H , L 2 and Lz . The indices or quantum numbers are n, l, and ml , and the
eigenfunctions are described as

Ψn,l,ml (7.81)
HΨn,l,ml = En Ψn,l,ml (7.82)
−
→2
L Ψn,l,ml = l(l + 1)~2 Ψn,l,ml (7.83)
Lz Ψn,l,ml = m~Ψn,l,ml (7.84)

7.2 Variational method

The variational method is a very useful method to calculate or approximate wave

functions by minimizing parameters and the energy of the system. This technique
is especially useful for LCAO (Linear Combination of Atomic Orbitals) methods
in molecules. Assume we have to solve the eigenvalue equation

HΨn = En Ψn multiply with Ψ∗n , and integrate :

R ∗
Ψ HΨn d− → r
⇒ En = R n ∗ −
→ (7.85)
Ψn Ψn d r

We get the energy eigenvalues if we know the eigenfunctions, but what can we
do, if they are unknown? We can start to solve the equation by using a test or
trial wave function ψtr :

R ∗
ψtr Hψtr d→− r
²= R ∗ →
− . (7.86)
ψtr ψtr d r

For every solution it is ² ≥ E0 , with E0 is the lowest energy eigenvalue of the

system. To prove that we make use of the complete set and write the trial wave

Physics of Atoms and Molecules

7.3 Angular momentum and spherical potentials 73

function as a sum of the orthonormal wave functions of the system:

X
ψtr (−
→
r) = cn Ψn (−
→
r) (7.87)
n
Z XX Z
∗
⇒ ψtr (H − E0 )ψtr dV = c∗n cn0 Ψ∗n (H − E0 )Ψn0 dV (7.88)
n n0
XX Z
= c∗n cn0 (En0 − E0 ) Ψ∗n Ψn0 dV (7.89)
n n0
XX
= c∗n cn0 (En0 − E0 )δnn0 (7.90)
n n0
X
= |cn |2 (En − E0 ) ≥ 0 (7.91)
n
Z
∗
⇒ ψtr (H − E0 )ψtr dV ≥ 0, hH − E0 i = 0 if ψtr = Ψ0 (7.92)
R ∗
ψtr Hψtr d→− r
² = R ∗ →
− ≥ E0 (7.93)
ψtr ψtr d r
In order to use this variational method one chooses a trial function ψtr depending
on several parameters {a, b, . . .} and minimize the energy
∂² ∂²
²(a, b, . . .), = 0, = 0, . . . (7.94)
∂a ∂b
for these parameters. This gives the lowest upper limit of E0 for wave functions
of type ψtr (a, b, . . .). With a well chosen eigenfunction it is possible to come very
close to E0 . As an example one can chose the trial function
∂²
ψtr = e−αr , =0 (7.95)
∂α
as a test wave function for the hydrogen atom. Upon optimizing the parameter α,
it becomes the reciprocal of Bohr's radius α = a10 and the energy ² = E0 is exactly
the eigenvalue of the hydrogen atom. Here, we get the exact wave solution.

7.3 Angular momentum and spherical potentials

The angular momentum is classically and quantum mechanically dened by:
¯ ¯
¯ ex ey ez ¯
→
− ¯ ¯
L =− →
r ×− →
p = ¯¯ x y z ¯¯ (7.96)
¯ px py pz ¯
¯ ¯
¯ ex ey ez ¯
−
→ ~¯ ¯ ¯
L = x y z ¯ (7.97)
i ¯¯ ∂ ∂ ∂ ¯¯
∂x ∂y ∂z

Physics of Atoms and Molecules

74 7 Harmonic oscillator 1-D

That means

µ ¶
~ ∂ ∂
Lx = y −z (7.98)
i ∂z ∂y
µ ¶
~ ∂ ∂
Ly = z −x (7.99)
i ∂x ∂z
µ ¶
~ ∂ ∂
Lz = x −y (7.100)
i ∂y ∂x

−
→
The components of L are hermitian, but they do not commute.

µ ¶2 ½µ ¶µ ¶
~ ∂ ∂ ∂ ∂
[Lx , Ly ]ψ = y −z z −x −
i ∂z ∂y ∂x ∂z
µ ¶µ ¶¾
∂ ∂ ∂ ∂
z −x y −z ψ (7.101)
∂x ∂z ∂z ∂y
µ ¶2 ½
~ ∂ ∂2 ∂2 ∂2 ∂2
= y + yz − yx 2 − z 2 + zx
i ∂x ∂z∂x ∂z ∂y∂x ∂y∂z
¾
∂2 2 ∂
2
∂2 ∂ ∂2
−zy +z + xy 2 − x − xz ψ (7.102)
∂x∂z ∂x∂y ∂z ∂y ∂z∂y
½ ¾
~~ ∂ ∂ ~
= y −x ψ = − Lz ψ = i~Lz ψ ∀ψ (7.103)
i i ∂x ∂y i

Similar calculations can be performed for the other coordinates, leading to

[Lx , Ly ] = i~Lz (7.104)

[Ly , Lz ] = i~Lx (7.105)
[Lz , Lx ] = i~Ly (7.106)
[Li , Lj ] = i~Lk with{i, j, k} cyclic permutations (7.107)

The hermitian operators Lx , Ly , and Lz are called the angular momentum ope-
rators. Since, they do not commutate, there is no common eigenfunction of these
−
→
operators, but a single component Li commutes with the operator L 2 . For con-

Physics of Atoms and Molecules

7.3 Angular momentum and spherical potentials 75

venience purposes, we chose the operator Lz to describe the systems properties:

h − i
→
Lz , L 2 = [Lz , L2x + L2y + L2z ] = [Lz , L2x ] + [Lz , L2y ] (7.108)
= Lz L2x − L2x Lz + Lz L2y − L2y Lz (7.109)
= (Lz Lx )Lx − Lx (Lx Lz ) − Ly (Ly Lz ) + (Lz Ly )Ly (7.110)
= Lx (Lz Lx ) − Lx (Lx Lz ) + (Lz Lx )Lx − (Lx Lz )Lx
| {z } | {z }
1 2
+ Ly (Lz Ly ) − Ly (Ly Lz ) + (Lz Ly )Ly − (Ly Lz )Ly (7.111)
| {z } | {z }
3 4
= Lx [Lz , Lx ] + [Lz , Lx ]Lx + Ly [Lz , Ly ] + [Lz , Ly ]Ly (7.112)
| {z } | {z } | {z } | {z }
1 2 3 4
= i~Lx Ly + i~Ly Lx − i~Ly Lx − i~Lx Ly = 0 (7.113)
−
→
This can be shown faster by using commutator properties. The operator L 2 and
Lz share the same eigenfunctions. The raising and lowering operators L+ and L− ,
respectively, are rather useful and dened as

L+ = Lx + iLy (7.114)
L− = Lx − iLy (7.115)
L+ L− = (Lx + iLy )(Lx − iLy ) = L2x + L2y − i[Lx , Ly ] (7.116)
−
→
= L 2 − L2z + ~Lz (7.117)
−
→2
L = L+ L− + L2z − ~Lz (7.118)
−
→2
L = L− L+ + L2z + ~Lz (7.119)
[L± , Lz ] = ∓~L± (7.120)
[L+ , L− ] = 2~Lz (7.121)

Now, we have to prove that

−
→
[H, L 2 ] = [H, Lz ] = 0 (7.122)
−
→
and the operators H , L 2 , and Lz share the same eigenfunctions, for V (−
→
r) =
V (r). To show that we rst introduce spherical coordinates and transform the
operators to spherical coordinates r, ϑ, ϕ with the transformation:

x = r sin ϑ cos ϕ (7.123)

y = r sin ϑ sin ϕ (7.124)
z = r cos ϑ (7.125)

The derivatives in spherical coordinates are given by

Physics of Atoms and Molecules

76 7 Harmonic oscillator 1-D

z
P=(x,y,z)

r= ϑ

z=r cosϑ
1
y

s ϕ
ϕ

co
ϑ
y=r sinϑ sin ϕ

in
rs
x=
x

Abbildung 7.3. Spherical coordinates r, ϑ, ϕ. The coordinates can take the following values:
r ∈ [0, ∞), ϑ ∈ [−π, π], ϕ ∈ [0, 2π).

µ ¶ µ ¶ µ ¶
∂ ∂r ∂ ∂ϑ ∂ ∂ϕ ∂
= + + , for xi ∈ {x, y, z} (7.126)
∂xi ∂xi ∂r ∂xi ∂ϑ ∂xi ∂ϕ

This leads to 9 derivatives:

∂ϕ
∂r
∂x
= sin ϑ cos ϕ ∂ϑ
∂x
= 1r cos ϑ cos ϕ ∂x
= − rsin ϕ
sin ϑ
∂r ∂ϑ ∂ϕ cos ϕ
∂y
= sin ϑ sin ϕ ∂y
= 1r cos ϑ sin ϕ ∂y
= − r sin ϑ (7.127)
∂r ∂ϑ ∂ϕ
∂z
= cos ϑ ∂z
= − sinr ϑ ∂z
=0

⇒ (7.128)
∂ ∂ 1 ∂ sin ϕ ∂
= sin ϑ cos ϕ + cos ϑ cos ϕ − (7.129)
∂x ∂r r ∂ϑ r sin ϑ ∂ϕ
∂ ∂ 1 ∂ cos ϕ ∂
= sin ϑ sin ϕ + cos ϑ sin ϕ + (7.130)
∂y ∂r r ∂ϑ r sin ϑ ∂ϕ
∂ ∂ sin ϑ ∂
= cos ϑ − (7.131)
∂z ∂r r ∂ϑ

If we insert (7.129, 7.130, 7.131) into the denition of Li we get

µ ¶
~ ∂ ∂
Lx = y −z (7.132)
i ∂z ∂y

Physics of Atoms and Molecules

7.3 Angular momentum and spherical potentials 77

½ µ ¶
~ ∂ sin ϑ ∂
= r sin ϑ sin ϕ cos ϑ − −
i ∂r r ∂ϑ
µ ¶¾
∂ 1 ∂ cos ϕ ∂
r cos ϑ sin ϑ sin ϕ + cos ϑ sin ϕ +
∂r r ∂ϑ r sin ϑ ∂ϕ
µ ¶
~ 2 2 ∂ cos ϕ cos ϑ ∂
= (− sin ϑ sin ϕ − cos ϑ sin ϕ) −
i ∂ϑ sin ϑ ∂ϕ
µ ¶
~ ∂ ∂
Lx = − sin ϕ − cos ϕ cot ϑ
i ∂ϑ ∂ϕ

In analogy to this transformation we get for the other operators

µ ¶
~ ∂ ∂
Lx = − sin ϕ − cos ϕ cot ϑ (7.133)
i ∂ϑ ∂ϕ
µ ¶
~ ∂ ∂
Ly = cos ϕ − sin ϕ cot ϑ (7.134)
i ∂ϑ ∂ϕ
~ ∂
Lz = (7.135)
i ∂ϕ
µ ¶
iϕ ∂ ∂
L+ = ~ e + i cot ϑ (7.136)
∂ϑ ∂ϕ
µ ¶
−iϕ ∂ ∂
L− = ~ e − + i cot ϑ (7.137)
∂ϑ ∂ϕ

If we insert (7.136), (7.137) and (7.135) into (7.118) we get

−
→2
L = L+ L− + L2z − ~Lz (7.138)
µ ¶ µ ¶
∂ ∂ ∂ ∂
with L+ L− = ~2 eiϕ + i cot ϑ e−iϕ − + i cot ϑ
∂ϑ ∂ϕ ∂ϑ ∂ϕ
µ 2 2
¶
∂ ∂ ∂ ∂
= −~2 + cot ϑ + cot 2
ϑ + i (7.139)
∂ϑ2 ∂ϑ ∂ϕ2 ∂ϕ
~ ∂ ∂2
Lz = ⇒ L2z = −~2 2 (7.140)
i ∂ϕ ∂ϕ

With the following

cot2 ϑ + 1 = sin−2 ϑ (7.141)

1 ∂ ∂ ∂2 ∂
sin ϑ = 2
+ cot ϑ (7.142)
sin ϑ ∂ϑ ∂ϑ ∂ϑ ∂ϑ
we get nally
µ ¶
→
−2 1 ∂ ∂ 1 ∂2
L = −~2 sin ϑ + . (7.143)
sin ϑ ∂ϑ ∂ϑ sin2 ϑ ∂ϕ2

Physics of Atoms and Molecules

78 7 Harmonic oscillator 1-D

−
→ ∂
The operators L 2 and Lz do not depend on r or ∂r . Thus, they commute with
V (r). This is the result of V (r) being invariant under rotations around the x-, y-
→
− → −
or z-axis. Now, we have to transform H into spherical coordinates with 4 = ∇ • ∇
−
→
and ∇ in spherical coordinates
½ ¾
−
→ ∂ 1 ∂ 1 ∂
∇ = {∇r , ∇ϑ , ∇ϕ } = , , . (7.144)
∂r r ∂ϑ r sin ϑ ∂ϕ
−
→ −
→ −
→ − →
For a vector F with the components F = {Fr , Fϑ , Fϕ }, ∇ • F is given by
½ ¾
−
→ − → 1 ∂ 2 ∂ ∂
∇•F = 2 (r (sin ϑ)Fr ) + (r(sin ϑ)Fϑ ) + (rFϕ ) . (7.145)
r sin ϑ ∂r ∂ϑ ∂ϕ
−
→ − →
Using F = ∇ we have
→
− → −
4 = ∇•∇
½ ¾
1 ∂ 2 ∂ ∂ ∂ ∂ 1 ∂
= 2 (r sin ϑ ) + sin ϑ + (7.146)
r sin ϑ ∂r ∂r ∂ϑ ∂ϑ ∂ϕ sin ϑ ∂ϕ
µ ¶
1 ∂ 2∂ 1 1 ∂ ∂ 1 ∂2
4 = 2 r + sin ϑ + . (7.147)
r ∂r ∂r r2 sin ϑ ∂ϑ ∂ϑ sin2 ϑ ∂ϕ2

The Hamilton operator is

~2
H = − 4 + V (r) (7.148)
2m µ ¶
~2 1 ∂ 2 ∂ ~2 1 ∂ ∂ 1 ∂2
= − r − sin ϑ + + V (r)
2m r2 ∂r ∂r 2mr2 sin ϑ ∂ϑ ∂ϑ sin2 ϑ ∂ϕ2
−
→2
~2 1 ∂ 2 ∂ L
H = − r + + V (r). (7.149)
2m r ∂r ∂r 2mr2
2

→
− ∂ ∂ ∂2 ∂2
Since L 2 and Lz depend on (ϑ, ϕ, ∂ϑ , ∂ϕ , ∂ϑ 2 , ∂ϕ2 ) and V (r) and the other terms
∂ 2
depend on (r, , ∂ )
∂r ∂r2
it is
−→
[H, L 2 ] = 0 [H, Lz ] = 0 if V (−
→
r ) = V (r). (7.150)

The eigenfunctions Φ(ϕ) of Lz satisfy the eigenfunction equation

Lz Φ(ϕ) = m~Φ(ϕ) (7.151)

where the eigenvalues have been written as m~ for convenience. The normalized
solutions of (7.151) are
1
Φm (ϕ) = √ eimϕ . (7.152)
2π

Physics of Atoms and Molecules

7.3 Angular momentum and spherical potentials 79

Since the functions Φm (ϕ) must be single-valued, we have Φm (2π) = Φm (0), and
m is restricted to positive or negative integers or zero m = {0, ±1, ±2, . . .}. The
integer m is called the magnetic quantum number with orthonormal eigenfuncti-
ons
Z2π
Φ∗m0 (ϕ)Φm (ϕ)dϕ = δm,m0 . (7.153)
0
−
→ →
−
The simultaneous eigenfunctions of L 2 and Lz (remember [ L 2 , Lz ] = 0) are called
spherical harmonics and are denoted by Y` m (ϑ, ϕ). They satisfy the eigenvalue
equations
−
→2
L Y` m (ϑ, ϕ) = ` (` + 1)~2 Y` m (ϑ, ϕ) (7.154)
Lz Y` m (ϑ, ϕ) = m~Y` m (ϑ, ϕ) (7.155)
→
−
where the eigenvalues of L 2 have been written as ` (` + 1)~2 . The quantum
number ` is known as the orbital angular momentum quantum number. The
number ` (` + 1) is dimensionless and has to be positive or zero, because it is the
eigenvalue of the squared hermitian operator. Possible values for ` are given by

L2z Y` m (ϑ, ϕ) = m2 ~2 Y` m (ϑ, ϕ)

→
−
(L2x + L2y )Y` m (ϑ, ϕ) = ( L 2 − L2z )Y` m (ϑ, ϕ)
= [` (` + 1) − m2 ]~2 Y` m (ϑ, ϕ)
Z Z
Y` m (ϑ, ϕ)(Lx + Ly )Y` m (ϑ, ϕ)d r = ~ [` (` + 1) − m ] Y`∗m (ϑ, ϕ)Y` m (ϑ, ϕ)d−
∗ 2 2 −
→ 2 2 →
r

= hL2x + L2z i ≥ 0 (7.156)

2
⇒m ≤ ` (` + 1) = ` 2 + ` (7.157)

In equation (7.156) we used that the expectation value of a squared hermitian

−
→
operator is positive or zero. In equation (7.138) we expressed L 2 with the lowering
→
−
and rising operators L+ and L− . Since L± commute with L 2 (L± commute with
Lz see exercise) it satisfy the equation
−
→2 −
→
L L± Y` m (ϑ, ϕ) = L± L 2 Y` m (ϑ, ϕ) = L± ` (` + 1)~2 Y` m (ϑ, ϕ)
= ` (` + 1)~2 L± Y` m (ϑ, ϕ) (7.158)
→
−
L± and L 2 have the same eigenfunctions and if the lowering or rising operator
−
→
acts on Y` lm (ϑ, ϕ) it does not change the eigenvalues ` (` + 1)~2 of L 2 . But the
operator L± changes the eigenvalues of Lz , because with [L± , Lz ] = ∓~L± it is

Lz (L± Y` m (ϑ, ϕ)) = L± Lz Y` m (ϑ, ϕ) ± ~L± Y` m (ϑ, ϕ)

= (m ± 1)~(L± Y` m (ϑ, ϕ)). (7.159)

Physics of Atoms and Molecules

80 7 Harmonic oscillator 1-D

That means L± Y` m (ϑ, ϕ) are eigenfunctions of Lz to the eigenvalue (m+1)~. The

operator L± increases (+) or decreases (-) the value of the magnetic quantum
number by one:
L± Y` m (ϑ, ϕ) ∝ Y` (m±1) (ϑ, ϕ) (7.160)
The rising and lowering of m is limited by ` (` + 1), so that there is a maximum
value of m called mmax and a minimum value of m called mmin which satisfy the
equations

L+ Y` mmax (ϑ, ϕ) = 0 (7.161)

L− Y` mmin (ϑ, ϕ) = 0. (7.162)

Using the operator L− , L+ and equations (7.161, 7.162) we nd that

→
−
L− L+ Y` mmax (ϑ, ϕ) = ( L 2 − L2z − ~Lz )Y` mmax (ϑ, ϕ) (7.163)
2 2
= ~ [` (` + 1) − mmax − mmax ]Y` mmax (ϑ, ϕ) (7.164)
L− L+ Y` mmax (ϑ, ϕ) = 0 (7.165)
→
−2
L+ L− Y` mmin (ϑ, ϕ) = ( L − L2z + ~Lz )Y` mmin (ϑ, ϕ) (7.166)
2 2
= ~ [` (` + 1) − mmin + mmin ]Y` mmin (ϑ, ϕ) (7.167)
L+ L− Y` mmin (ϑ, ϕ) = 0 (7.168)

Since the eigenfunctions Y` m (ϑ, ϕ) are not identical zero it is

mmax (mmax + 1) = ` (` + 1) ≥ 0 (7.169)

mmin (mmin − 1) = ` (` + 1) ≥ 0 (7.170)
mmin = −`
⇒ mmin (mmin − 1) = −` (−` − 1) = ` (` + 1) (7.171)

We showed already that the magnetic quantum numbers m are integers and from
equations (7.169, 7.170) we can conclude that the orbital angular momentum
quantum numbers ` are also integers. Since the rising and lowering operators L+
and L− increases and decreases m by one, the values of the magnetic quantum
number are
−` ≤ m ≤ ` . (7.172)
Thus, for a given orbital angular momentum quantum number ` the magnetic
quantum number m can take (2` + 1) values

m = {0, ±1, ±2, . . . , ±` }. (7.173)

The eigenvalue equations are

−
→2
L Y` m (ϑ, ϕ) = ~2 ` (` + 1) Y` m (ϑ, ϕ) (7.174)
Lz Y` m (ϑ, ϕ) = ~ m Y` m (ϑ, ϕ), −` ≤ m ≤ +` (7.175)

Physics of Atoms and Molecules

7.3 Angular momentum and spherical potentials 81

If Y` ,−` (ϑ, ϕ) or Y` ` (ϑ, ϕ) are known, we can use the operators L+ and L− to
generate all other eigenfunctions Y` m (ϑ, ϕ), in analogy to the harmonic oscillator
eigenfunctions. For the maximum value of m = ` one has

L+ Y` ` (ϑ, ϕ) = 0
· ¸
iϕ ∂ ∂
⇒ ~e + i cot ϑ Y` ` (ϑ, ϕ) = 0 (7.176)
∂ϑ ∂ϕ
~ ∂
Lz Y` ` (ϑ, ϕ) = Y` ` (ϑ, ϕ) = ~ ` Y` ` (ϑ, ϕ) (7.177)
i ∂ϕ
∂
We know from equation (7.177) that ∂ϕ Y` m (ϑ, ϕ) does not depend on ϑ, and we
solve the (7.176) by using the product solution method. Setting

Y` m (ϑ, ϕ) = P`m (ϑ)Φm (ϕ) (7.178)

m=`
⇒ Y` ` (ϑ, ϕ) = P`` (ϑ)Φ` (ϕ) (7.179)
· ¸
∂ ∂
~eiϕ + i cot ϑ P ` (ϑ)Φ` (ϕ) = 0 (7.180)
∂ϑ ∂ϕ `
∂ ∂
Φ` (ϕ) P`` (ϑ) + i cot ϑP`` (ϑ) Φ` (ϕ) = 0 (7.181)
∂ϑ ∂ϕ
∂ ∂
Φ` (ϕ) tan ϑ P`` (ϑ) = −iP`` (ϑ) Φ` (ϕ) (7.182)
∂ϑ ∂ϕ
tan ϑ ∂ ` −i ∂
P (ϑ) = Φ` (ϕ) = c (7.183)
P`` (ϑ) ∂ϑ ` Φ` (ϕ) ∂ϕ
∂ ` cos ϑ `
⇒ P` (ϑ) = c P (ϑ) (7.184)
∂ϑ sin ϑ `
∂
⇒ Φ` (ϕ) = ic Φ` (ϕ) (7.185)
∂ϕ

With equation (7.177) the constant c has to be c = `.

Φ` (ϕ) ∝ ei`ϕ
P`` (ϑ) ∝ sin` ϑ
⇒ Y`` (ϑ, ϕ) ∝ sin` ϑei`ϕ (7.186)
⇒ Y`m (ϑ, ϕ) ∝ P`m (ϑ)eimϕ (7.187)

Using L− iteratively we get the complete set of 2` + 1 eigenfunctions {Y`m (ϑ, ϕ)}.
The functions P`0 (cos ϑ) are called the Legendre polynomials of degree `. They
are dened by the relation

1 d`
P` (cos ϑ) = ((cos ϑ)2 − 1)` . (7.188)
2` `! d(cos ϑ)`

Physics of Atoms and Molecules

82 7 Harmonic oscillator 1-D

The Legendre polynomials (m = 0) satisfy the dierential equation

· ¸
2 d2 d
(1 − cos ϑ) − 2 cos ϑ + `(` + 1) P` (cos ϑ) = 0 (7.189)
d(cos ϑ)2 d cos ϑ

with the recurrence relation

(2` + 1) cos ϑP` − (` + 1)P`+1 − `P`−1 = 0 (7.190)

which is also valid for ` = 0 if one denes P−` = 0. The orthogonality relations
read
Z+1
2
P` (cos ϑ)P`0 (cos ϑ)d cos ϑ = δ``0 . (7.191)
2` + 1
−1

One has the closure relation

∞
1X
(2` + 1)P` (cos ϑ)P` (cos ϑ0 ) = δ(cos ϑ − cos ϑ0 ) (7.192)
2 `=0

Important particular values of the Legendre polynomials are

cos ϑ = 1 : P` (1) = 1, cos ϑ = −1 : P` (−1) = (−1)` (7.193)

For the lowest values of ` one has explicitly

P0 (cos ϑ) = 1 (7.194)
P1 (cos ϑ) = cos ϑ (7.195)
1¡ ¢
P2 (cos ϑ) = 3 cos2 ϑ − 1 (7.196)
2
1¡ ¢
P3 (cos ϑ) = 5 cos3 ϑ − 3 cos ϑ (7.197)
2
1¡ ¢
P4 (cos ϑ) = 35 cos4 ϑ − 30 cos2 ϑ + 3 (7.198)
8
1¡ ¢
P5 (cos ϑ) = 63 cos5 ϑ − 70 cos3 ϑ + 15 cos ϑ (7.199)
8
The associated Legendre functions P`m (cos ϑ) are now dened by the relations

dm
P`m (cos ϑ) = (1 − cos ϑ) 2 m/2
P` (cos ϑ), m = 0, 1, 2, . . . , `. (7.200)
d(cos ϑ)m

They satisfy the recurrence relations

m
(2` + 1) cos ϑ P`m = (` + 1 − m)P`+1 m
+ (` + m)P`−1 (7.201)
p
(2` + 1) (1 − cos2 ϑ)P`m−1 = P`+1
m m
− P`−1 (7.202)

Physics of Atoms and Molecules

7.3 Angular momentum and spherical potentials 83

and the orthogonality relations

Z1
2 (l + m)!
P`m (cos ϑ)P`m0 (cos ϑ)d cos ϑ = δ` `0 (7.203)
2` + 1 (l − m)!
−1

The rst few associated Legendre functions are given explicitly by

p
P11 (cos ϑ) = (1 − cos2 ϑ) (7.204)
p
P21 (cos ϑ) = 3 (1 − cos2 ϑ) cos ϑ (7.205)
P22 (cos ϑ) = 3(1 − cos2 ϑ) (7.206)
3p
P31 (cos ϑ) = (1 − cos2 ϑ)(5 cos2 ϑ − 1) (7.207)
2
P32 (cos ϑ) = 15 cos ϑ(1 − cos2 ϑ) (7.208)
√
P33 (cos ϑ) = 15 1 − cos2 ϑ(1 − cos2 ϑ) (7.209)

Thus, the spherical harmonics Y`m (ϑ, ϕ) with m ≥ 0 are given by

s· ¸
m (2` + 1)(l − m)! m
Y`m (ϑ, ϕ) = (−1) P` (cos ϑ)eimϕ (7.210)
4π(l + m)!
Y`−m (ϑ, ϕ) = (−1)m Y`m
∗
(ϑ, ϕ) (7.211)

The rst few spherical harmonics are given explicitly by

(2` + 1) Y`m (ϑ, ϕ)

1
1 Y0 0 (ϑ, ϕ) = √
4π
r r
3 3 z
3 Y1 0 (ϑ, ϕ) = cos ϑ =
4π 4π r
r r
3 ±iϕ 3 x ± iy
3 Y1 ±1 (ϑ, ϕ) = ∓ sin ϑ e =∓
8π 8π r
r µ ¶ r
5 3 1 1 5 2z 2 − x2 − y 2
5 Y2 0 (ϑ, ϕ) = cos2 ϑ − =
4π 2 2 2 4π r2
r r
1 15 1 15 (x ± iy)z
5 Y2 ±1 (ϑ, ϕ) = ∓ sin ϑ cos ϑ e±iϕ = ∓
2 2π 2 2π r2
r r µ ¶2
1 15 1 15 x ± iy
5 Y2 ±2 (ϑ, ϕ) = sin2 ϑ e±2iϕ =
4 2π 4 2π r

Physics of Atoms and Molecules

84 7 Harmonic oscillator 1-D

The rising and lowering operators are

· ¸
−iϕ ∂
L− Y`m (ϑ, ϕ) = −~e − + m cot ϑ Y`m (ϑ, ϕ) (7.212)
∂ϑ
· ¸
iϕ ∂
L+ Y`m (ϑ, ϕ) = −~e + m cot ϑ Y`m (ϑ, ϕ) (7.213)
∂ϑ
p
L± Y`m (ϑ, ϕ) = ~ `(` + 1) − m(m ± 1) Y`m±1 (ϑ, ϕ) (7.214)
p
L± |`mi = ~ `(` + 1) − m(m ± 1) |`m ± 1i (7.215)

and they change the magnetic quantum number (Fig. 7.3)

· ¸
−iϕ ∂
L− Y11 (ϑ, ϕ) ∝ ~e − + m cot ϑ sin ϑeiϕ (7.216)
∂ϑ
∝ Y10 (7.217)
L− Y10 ∝ Y1−1 (7.218)

The absolute values of the spherical harmonics are given by the associated Le-

l=1 l=2

+2

+1 +1
L-
0 m 0
L+
-1 -1

-2
Abbildung 7.4. Operations of the rising and lowering operator on the magnetic quantum
number m.

gendre functions |Plm (ϑ)| which are presented in Fig. 7.3. The polar plots give
the length of |Plm (ϑ)| for every angle ϑ. Since ϑ varies from 0 to π the polar
plots have axial symmetry with respect to the z-axis. The set of eigenfuncti-
→
−
ons {Y`m (ϑ, ϕ)} are eigenfunctions of the hermitian operator L 2 . They form a
orthonormal complete set of functions with
Z Z
∗
Y`m (ϑ, ϕ)Y`0 m0 (ϑ, ϕ) sin ϑdϑdϕ = δ``0 δmm0 . (7.219)

Physics of Atoms and Molecules

7.3 Angular momentum and spherical potentials 85

M = 2l +1

1: l=0

m=0

3: l=1

m=0 m=+1, -1

5: l=2

m=0 m=+1, -1 m=+2, -2

7: l=3

m=0 m=+1, -1 m=+2, -2 m=+3, -3

Abbildung 7.5. Polar plots of the Legendre polynomials |Y`m (ϑ, ϕ)| = 1 m 2
2π |P` | . For every
orbital momentum quantum number ` there are 2` + 1 eigenfunctions.

Every square integrable function ψ(ϑ, ϕ) can be developed by {Y`m (ϑ, ϕ)}

∞
X +X̀
ψ(ϑ, ϕ) = c`m Y`m (ϑ, ϕ) (7.220)
`=0 m=−`

Some properties of the eigenfunctions are

∗
Y`m (ϑ, ϕ) = (−1)m Y`−m (ϑ, ϕ) (7.221)
P`−m = P`m by definition m ≥ 0 (7.222)
−
→
r → −−
→
r Y`m (π − ϑ, ϕ + π) = (−1)` Y`m (ϑ, ϕ) (7.223)
P`0 (cos(π − ϑ)) = (−1)` P`0 (cos ϑ) (7.224)
Y`m (ϑ, ϕ + 2π) = Y`m (ϑ, ϕ)eimϕ = Y`m (ϑ, ϕ) (7.225)

The Y`m (ϑ, ϕ) are complex functions for m 6= 0. It is useful for chemical bond
orbitals to create a linear combination of Y`m (ϑ, ϕ) and Y`−m (ϑ, ϕ) which are real

Physics of Atoms and Molecules

86 7 Harmonic oscillator 1-D

functions. Note, they are no longer eigenfunctions of Lz . Examples are

r
1 3 1 ¡ ¢
√ (Y1−1 (ϑ, ϕ) − Y11 (ϑ, ϕ)) = sin ϑ e−iϕ + eiϕ (7.226)
2 4π 2
r r
3 3 x
= sin ϑ cos ϕ = (7.227)
4π 4π r
r
1 3 1 ¡ ¢
√ i(Y1−1 (ϑ, ϕ) + Y11 (ϑ, ϕ)) = i sin ϑ e−iϕ − eiϕ (7.228)
2 4π 2
r r
3 3 y
= sin ϑ sin ϕ = (7.229)
4π 4π r
r r
3 3 z
Y10 (ϑ, ϕ) = cos ϑ = (7.230)
4π 4π r

These three linear combinations form the three orbitals in the directions of the
x-axis (px ), the y-axis (py ) and the z-axis (pz ) as shown in Fig. (7.3). The few

Abbildung 7.6. Polar plots of the real spherical harmonics s, px , py , pz .

Physics of Atoms and Molecules

8 Hydrogen atom and central forces 87

rst spherical harmonics in real form r Y`m (ϑ, ϕ) are

` |m| r Y`m (ϑ, ϕ) (7.231)

1
0 0 s: √ (7.232)
4π
r
3
1 0 pz : cos ϑ (7.233)
4π
r
3
1 1 px : sin ϑ cos ϕ (7.234)
4π
r
3
1 1 py : sin ϑ sin ϕ (7.235)
4π
r
5
2 0 d3z2 −r2 : (3 cos2 ϕ − 1) (7.236)
16π
r
15
2 1 dxz : sin ϑ cos ϑ cos ϕ (7.237)
4π
r
15
2 1 dyz : sin ϑ cos ϑ sin ϕ (7.238)
4π
r
15
2 2 dx2 −y2 : sin2 ϑ cos 2ϕ (7.239)
4π
r
15
2 2 dxy : sin2 ϑ sin 2ϕ (7.240)
4π
Here, we have used the so-called 'spectroscopic' notation, in which the value of the
orbital angular momentum quantum number ` is indicated by a letter, according
to the correspondence

Value of ` 0 1 2 3 4 5 ... (7.241)

l l l l l l ... (7.242)
Code letter s p d f g h, . . . (7.243)

These code letters are remnants of the spectroscopist's description of various

series of spectral lines. The letters s, p, d and f being the rst letters of the
adjectives 'sharp', 'principal', 'diuse' and 'fundamental', respectively. For values
of ` greater than three the letters follow in alphabetical order.

8 Hydrogen atom and central forces

8.1 Central forces
For central forces the potential V (→
−
r ) = V (r) depends only on |−
→
r | and the
−
→2
eigenfunctions are simultaneous eigenfunctions of H, L and Lz . We can solve

Physics of Atoms and Molecules

88 8 Hydrogen atom and central forces

the SE with the product solution method. Setting

ψn`m (r, ϑ, ϕ) = Rn` (r)Y`m (ϑ, ϕ) (8.1)

and substituting into (7.149) gives

" →
−2 #
~2 1 ∂ 2 ∂ L
− r + + V (r) Rn` (r)Y`m (ϑ, ϕ) = En`m Rn` (r)Y`m (ϑ, ϕ)
2m r2 ∂r ∂r 2mr2
(8.2)
−
→
From L 2 Y`m = `(` + 1)~2 Y`m we deduce
· ¸
~2 1 ∂ 2 ∂ `(` + 1)~2
− r + + V (r) Rn` (r) = En`m Rn` (r) (8.3)
2m r2 ∂r ∂r 2mr2

Equation (8.3) does not depend on the magnetic quantum number m. Setting

un` (r) = rRn` (r) (8.4)

we nd · ¸
~2 ∂ 2 `(` + 1)~2
− + + V (r) un` (r) = En` un` (r) (8.5)
2m ∂r2 2mr2
The radial SE is a one dimensional dierential equation in r, r ∈ (0, ∞). We
dene the eective potential Vef f

`(` + 1)
Vef f (r) = V (r) + (8.6)
2mr2
The second term is known as the 'centrifugal barrier' potential (with 2mr2 as
Ze
the moment of inertia), from the classical mechanics. With V (r) = − 4πε 0r
the
attractive potential of the Coulomb interaction is stronger than the 'centrifugal
barrier' potential for large r → ∞. For small r the 'centrifugal barrier' potential
dominates resulting in a minimum shifting to larger r upon increasing the orbital
angular momentum quantum number ` (see Fig. 8.1). The probability of nding
an electron in the vicinity of the nucleus decreases with increasing `.

8.2 Radial functions

The asymptotic properties of the radial functions Rn` are

ˆ for r → ∞ it is Vef f → 0 and

q
2|En` |m
un` (r) ∝ e−r ~2 (8.7)

Physics of Atoms and Molecules

8.2 Radial functions 89

V l=1
eff

Energy (arb.u.) V l=2

5 eff

Coulomb potential

Centrifugal barrier l=1

Centrifugal barrier l=2

-5

0
Radial component r

Abbildung 8.1. Potential components contributing to the Hamilton operator.

`(`+1)~2
ˆ for r → 0 it is Vef f → 2mr2
and with En` ≈ 0
un` (r) ∝ r`+1 (8.8)

Thus, we nd the asymptotic correct solution

q
un` 2|En` m|
Rn` = ∝ r` e−r ~2 (8.9)
r
For a hydrogen atom (Z = 1) and similar atom ions the Coulomb potential is
given by
Ze
V (r) = − (8.10)
4πε0 r
There are two ways of solving the partial dierential equation
1) describing H as a product of a and a+
2) using un` and its power series one nds the dierential equation
d2 w dw
z+ (c − z) − aw = 0 (8.11)
dz 2 dz
which is called the Kummer-Laplace dierential equation and can be solved
for c = 1 and a = 0. The functions which satisfy the dierential equation
are called the associated Laguerre polynomials L2`+1
n+` (αn` r) given explicitly
by
n−`−1
X [(n + `)!]2 rk
L2`+1
n+` (r) = (−1)k+1 . (8.12)
k=0
(n − ` − 1 − k)!(2` + 1 + k)!k!

Physics of Atoms and Molecules

90 8 Hydrogen atom and central forces

The Laguerre polynomials are special solutions only for hydrogen atoms and
similar atomic ions and the radial functions are given explicitly by

Rn` (ρ) = Nbn` · (ρ)` L2`+1 (ρ) e−ρ/2 (8.13)

n+`
mM a0 m 2Zr Z
with µ = , aµ = ρ= and αn = (8.14)
m+M µ naµ na0
` 2`+1 −αn r
Rn` (r) = Nn` · (αn r) Ln+` (2αn r) e (8.15)
s
mZ 2 e4 ~2 αn2
with En = − = − (8.16)
2~2 (4πε0 )2 n2 2m
1 na0 ◦
= , characteristic length a0 = 0.529 A (8.17)
αn Z
µ ¶ 23
Z − Zr
R10 (r) = 2 e a0 (8.18)
a0
µ ¶3 µ ¶
Z 2 Zr − Zr
R20 (r) = 2 1− e 2a0 (8.19)
2a0 2a0
µ ¶ 32 µ ¶
2 Z Zr − Zr
R21 (r) = √ e 2a0 (8.20)
3 2a0 2a0
µ ¶ 32 µ ¶
Z 2Zr 2Z 2 r2 − 3a Zr
R30 (r) = 2 1− + e 0 (8.21)
3a0 3a0 27a20
√ µ ¶3 µ µ ¶¶
4 2 Z 2 Zr Zr − Zr
R31 (r) = 1− e 3a0 (8.22)
3 3a0 3a0 6a0
√ µ ¶ 32 µ ¶2
2 2 Z Zr − Zr
R32 (r) = √ e 3a0 (8.23)
3 5 3a0 3a0

Because the spherical harmonics are normalized on the unit sphere the norma-
lization condition implies that
Z∞
Rn` (r) r2 dr = 1 (8.24)
0

The radial functions has the properties

ˆ The Laguerre polynomial is of the power (n − ` − 1) (see (8.12) and has

(n − ` − 1) radial nodes (without r = 0).

ˆ The probability Pn` (r) = r2 |Rn`

2
| has (n − `) maxima.
2Zr
− na
ˆ For (` = n − 1) the radial functions Pn,n−1 (r) ∝ r2n e 0 show only one
2
maximum at rmax = n Za0 .

Physics of Atoms and Molecules

8.2 Radial functions 91

2,0 0,5
n = 1, l = 0 n = 1, l = 0
0,4
1,5

0,3
1,0

0,2

0,5
0,1

0,0
0,0
0 5 10 15 20 25 0 5 10 15 20 25
3/2

0
2,0

(r) in units of Z/a

R (r) in units of (Z/a )
0

1,5

1,5 n = 2, l = 0, l = 1 n = 2, l = 0, l = 1

1,0 1,0

0,5
0,5

0,0
nl
2
r R

0,0
nl

0 5 10 15 20 25 0 5 10 15 20 25
2

2,0

n = 3, l = 0, l = 1, l = 2 3
n = 3, l = 0, l = 1, l = 2
1,5

1,0 2

0,5

0,0

0
0 5 10 15 20 25 0 5 10 15 20 25

Radius r in a Radius r in a
0 0

Abbildung 8.2. Radial functions Rn` (r) and the probability r2 Rn`
2
(r) of nding an electron
in (r + ∆r) − r.

ˆ For a given n the orbital angular momentum quantum numbers ` can only
range from {0, 1, . . . , (n − 1)} (see 8.12) and for a given ` there are (2` + 1)
magnetic quantum numbers m. En does not depend on ` and m, which is a
consequence of the symmetry of V (r) = − 1r and the independence of Rn` (r)
of m, respectively.

ˆ The degeneracy is given by

n−1
X
(2` + 1) = n2 (8.25)
`=0

Physics of Atoms and Molecules

92 8 Hydrogen atom and central forces

ˆ The expectation value for r, hri is

Z∞
k
hr in`m = rk Rn`
2
(r)r2 dr (8.26)
0
a0
hrin`m = [3n2 − `(` + 1)] (8.27)
2Z

ˆ For r → 0 the radial functions are Rn` (r) ∝ r` . For a given n larger values
of Rn` (r) occur for smaller ` at a given r (see Fig. 8.2).

ˆ Only radial functions Rn` (r) with ` = 0 (s-orbitals) are non zero at r = 0.
This is important for the Mössbauer eect where the orbital overlap of
the electronic wave function with the nucleus is responsible for the signal.
1 3
|ψn00 |2 = 4π |Rn0 (0)|2 = πnZ3 a3
0

P
30
3/2

1,0
P
P (r) in units of (Z/a )
0

31

P
32

0,5
nl

0,0
0 1 2 3 4

R adius r in units of a
0

Abbildung 8.3. Probability r2 Rn`

2
(r) of nding an electron in (r + ∆r) − r for very small r.

Physics of Atoms and Molecules

8.3 Hydrogen atom wave functions 93

8.3 Hydrogen atom wave functions

Abbildung 8.4. First wave functions of the hydrogen atom (and similar atomic ions) with
m = 0.

Physics of Atoms and Molecules

94 8 Hydrogen atom and central forces

n=1

n=2 0,0 1,0 1,+/-1

n=3 0,0 1,0 1,+/-1

2,0 2,+/-1 2,+/-2

n=4 0,0 1,0 1,+/-1

2,0 2,+/-1 2,+/-2

3,0 3,+/-1 3,+/-2 3,+/-3

Abbildung 8.5. First wave functions of the hydrogen atom (and similar atomic ions), inserted:
quantum numbers.

Physics of Atoms and Molecules

8.3 Hydrogen atom wave functions 95

n = 1; l = 0 ; m = 0 n = 2; l = 1 ; m = 0 n = 2; l = 1 ; m = +1 / -1

x 60 x 40 x 40

x 60 x 40 x 40

n = 3; l = 0 ; m = 0 n = 3; l = 1 ; m = 0 n = 3; l = 2 ; m = +1 / -1

x 60 x1 x1

x1 x1 x1
n = 3; l = 1 ; m = +1 / -1

Abbildung 8.6. Some wave functions of the hydrogen atom (and similar atomic ions) plotted
together with the quantum numbers. The solid black line represents the radial function, the
plots with a xed radius the spherical harmonics.

Physics of Atoms and Molecules

96 8 Hydrogen atom and central forces

8.4 Angular momentum

−
→
The functions Y`m (ϑ, ϕ) are simultaneously eigenfunctions of L 2 and Lz with
eigenvalues
p `(` + 1)~2 and m~, respectively. The angular momentum value
`(` + 1)~ > `~ is always greater than the maximum eigenvalue of Lz = `~.
−
→
This means the direction of the vector L is never parallel to the z-axis. Moreo-
ver, the Y`m (ϑ, ϕ) are no eigenfunctions of hLx i and hLy i and it is
hY`m |Lx Y`m i = hY`m |Ly Y`m i = 0 (8.28)
1
hY`m |L2x Y`m i = hY`m |L2y Y`m i = [`(` + 1) − m2 ]~2 . (8.29)
2
We note that the expectation values vanish, but the squares do not vanish (for
−
→
` 6= 0). The angular momentum L 'precesses' around the z-axis in the presence
ofpa magnetic eld with its projection on the z-axis to be ~m and its length
~ `(` + 1) (Fig. 8.4). With increasing ` and m = l the relative dierence between

Orbital angular momentum Spin angular momentum

Abbildung 8.7. Projections of the orbital angular momentum and spin angular momentum
on the z-axis.
p
`(` + 1) and ` becomes smaller approximating the classical case, where the
angular momentum is parallel to the z-axis and the particle moves in the x-y
plane.

The summation over all absolute squared spherical harmonics for a given orbital
angular momentum quantum number ` results in a value independent of ϑ and ϕ
X̀ 2` + 1
|Y`m (ϑ, ϕ)|2 = . (8.30)
−`
4π

This means that addition of all 2` + 1 functions |Y`m |2 leads to an isotropic

function. Since the radial eigenfunctions depend only of n and `, the charge
distribution of a lled or closed shell (`-shell) has radial symmetry.

Physics of Atoms and Molecules

8.5 Spin angular momentum 97

8.5 Spin angular momentum

Electrons possess an intrinsic degree of freedom, the spin, which behaves like an
angular momentum in the way it couples with a magnetic eld (see Stern-Gerlach
experiment). The z component of this spin angular momentum can only take on
two values ms ~, where ms = ± 12 . Therefore, the electron spin cannot be described
−
→ −
→
by the orbital angular momentum operator L , since the z component of L only
takes on the values m~, with m = −`, −` + 1, . . . , ` and ` = 0, 1, 2, . . .. We
assume that all angular momentum operators, whether orbital or spin, satisfy
commutation relations of the form (7.107). Thus, if Sx , Sy and Sz are the three
−
→
Cartesian components of the electron spin angular momentum operator S , they
must satisfy the commutation relations

i~Sz = [Sx , Sy ] (8.31)

i~Sx = [Sy , Sz ] (8.32)
i~Sy = [Sz , Sx ]. (8.33)

The properties of the spin angular momentum and the simultaneous spin eigen-
−
→
functions χs,ms of S 2 and Sz are
→
−2
S χs,ms = s(s + 1)~2 χs,ms (8.34)
Sz χs,ms = ms ~χs,ms . (8.35)

Since ms = ± 12 for an electron we must have s = 12 , and we say the electron spin
is one-half. There are two dierent normalized spin eigenfunctions χs,ms , namely

α = χ1/2,1/2 β = χ1/2,−1/2 . (8.36)

Using these equations we get

−
→2 3 2
S α = ~α (8.37)
4
−
→2 3 2
S β = ~β (8.38)
4
~
Sz α = α (8.39)
2
~
Sz β = − β (8.40)
2
The spin eigenfunctions α and β are said to correspond respectively to the spin
up (↑) and spin down (↓) states. General spin-1/2 functions χ are linear super-
positions of the spin states α and β . That is

χ = χ+ α + χ− β (8.41)

Physics of Atoms and Molecules

98 8 Hydrogen atom and central forces

where χ+ and χ− are the complex coecients and |χ± |2 the probability of nding
an electron in the 'spin up' (+) α or 'spin down' (-) β state. Some important
properties are

hχ|χi = 1 (8.42)
|χ+ | + |χ− |2
2
= 1 (8.43)
hα|αi = hβ|βi = 1 (8.44)
hβ|αi = hβ|αi = 0 (8.45)
−
→2 3 2
S = ~ (8.46)
4
S± = Sx ± iSy (8.47)
S+ α = S− β = 0 (8.48)
S− α = ~β (8.49)
S+ β = ~α (8.50)
~
Sx α = β (8.51)
2
~
Sx β = α (8.52)
2
i~
Sy α = β (8.53)
2
i~
Sy β = − α (8.54)
2
~
Sz α = α (8.55)
2
~
Sx β = β (8.56)
2
~2
Sx2 = Sy2 = Sz2 = (8.57)
4

Since there are only two basic spin eigenfunctions α and β , they exist in a two-
dimensional 'spin space'. The normalized spin 1/2 eigenfunctions α and β can be
considered as basis vectors given by two-component column vectors (also called

Physics of Atoms and Molecules

8.6 Total angular momentum 99

spinors)
µ ¶ µ ¶
1 0
α = ; β= (8.58)
0 1
α† α = β † β = 1 (8.59)
α† β = β † α = 0 (8.60)
µ ¶
→
−2 3 2 1 0
S = ~ (8.61)
4 0 1
µ ¶
1 1 0
Sz = ~ (8.62)
2 0 −1
µ ¶
0 1
S+ = ~ (8.63)
0 0
µ ¶
0 0
S− = ~ (8.64)
1 0
µ ¶
~ 0 1
Sx = (8.65)
2 1 0
µ ¶
~ 0 −i
Sy = (8.66)
2 i 0

The matrices can also be dened by Si = ~2 σi , where the matrices σi are called
the Pauli spin matrices. Using the explicit form (8.58) of the basic spinor α and
β , an arbitary spin-1/2 function may be written as the spinor
µ ¶
χ+
χ= ; χ† = (χ∗+ χ∗− ). (8.67)
χ−

It is worth noting that if the electron is in a pure 'spin up' state α or 'spin down'
state β , the expectation values of hSx2 i and hSy2 i do not vanish (see vector model
Fig. 8.4). The quantum number ms can take on 2s+1 values {−s, −s+1, . . . , s−
1, s}.

8.6 Total angular momentum

The total angular momentum can be written as
−
→ − → −
→
J = L+S (8.68)
−
→ → →
The orbital angular momentum L = − r ×− p operates only in 'ordinary' space and
satises the commutation relations (7.107). On the other hand the spin angular
→
−
momentum S satises the commutation relations (8.33) and operates only in the
'spin space'. All its components therefore commute with those of − →r and −→p , and

Physics of Atoms and Molecules

100 8 Hydrogen atom and central forces

→
− →
−
hence with all those of L . As a result, the total angular momentum J satises
the commutation relations

i~Jk = [Ji , Jj ] with {i, j, k} cyclic permutating. (8.69)

−
→
This characterizes an angular momentum operator J satisfying the eigenvalue
equations
−
→2
J ψjmj = j(j + 1)~2 ψjmj (8.70)
Jz ψjmj = mj ~ψjmj (8.71)

where j is an integer (including zero) or a half-odd integer (j = 0, 21 , 23 , . . .) and

−
→
mj = −j, −j + 1, . . . , j − 1, j . Since all the components of L commute with
→
− −
→ −
→
all those of S , the operators L 2 , Lz , S 2 , and Sz mutually commute, and have
simultaneous eigenfunctions

ψ`sm` ms = Y`m` (ϑ, ϕ)χsms . (8.72)

−
→ − → − →
The simultaneous eigenfunctions of L 2 , S 2 , J 2 and Jz are linear combinations
of the functions ψ`sm` ms . For a given value of ` and s the possible values of j are
given by j = {|` − s|, |` − s| + 1, . . . , ` + s} and the quantum number mj can
take on 2j + 1 values (see above).

In general, the wave functions depend on the radial coordinate, the angular mo-
mentum, the spin and on the time t. For example, in the case of a spin-1/2 particle
(for example, an electron), a general expression for the wave function is
µ ¶
→
− −
→ Ψ+
Ψ(q, t) = Ψ+ ( r , t)α + Ψ− ( r , t)β = (8.73)
Ψ−

where q denotes the ensemble of the continuous spatial variable − →

r and the discrete
1
spin variable (ms = ± 2 ) of the particle. The probability density for nding at time
t the particle at −
→
r with 'spin up' is |Ψ+ (−
→r , t)|2 (and for 'spin down' |Ψ− (→
−
r , t)|2 ).

Physics of Atoms and Molecules

9 Interaction of atoms with electromagnetic radiation 101

9 Interaction of atoms with electromagnetic

radiation
−
→ −
→
The electromagnetic eld E and magnetic eld B can be generated from scalar
−
→
and vector potentials φ and A by

−
→− −
→ − ∂− →→
E (→
r , t) = − ∇φ(→
r , t) − A (− r , t) (9.1)
∂t
−
→− −
→ − →→
B (→
r , t) = ∇ × A (−
r , t). (9.2)
→
−
The potentials are not completely dened by (9.1, 9.2), since the elds E and
−
→ −
→ −
→ → −
B are invariant under the (classical) gauge transformations A → A + ∇Θ and
Φ → Φ − ∂Θ
∂t
, where Θ is any real, dierentiable function of −
→
r and t. The freedom
implied by this gauge invariance allows us to choose
−
→ → −
∇ · A = 0, Coulomb gauge (9.3)
−
→ −
→ ∂Φ
∇·A+ = 0, Lorentz gauge. (9.4)
∂t
Using the Coulomb gauge is convenient when no sources are present and therefore
−
→ →
−
%(t) = %. When A satises the Coulomb gauge, one may take Φ = 0, and A
satises the wave equation
→
−
−
→2 → − 1 ∂ A2
0 = ∇ A− 2 2 (9.5)
c0 ∂t
→
− − →
− →
A (→
r , t) = 2 A0 (ω) εb cos( k · − r − ωt + ϕ) (9.6)
→
− − h − →− −
→− i
A (→
→ →
r , t) = A0 (ω) εb ei( k · r −ωt+ϕ) + e−i( k · r −ωt+ϕ) (9.7)
−
→ → − −
→
0 = ∇ · A ⇒ εb · k = 0, wave is transverse (9.8)
−
→
→
− − k × εb −
→ −
B (→
r , t) = E0 (ω) sin( k · →r − ωt + ϕ), with E0 (ω) = ωA0(9.9)
(ω)
ω
1 |B|2 −
→ →
Energy density : (ε0 |E|2 + ) = ε0 E02 sin2 ( k · − r − ωt + ϕ) (9.10)
2 µ0
1 c0
I(ω) = ρ(ω)c0 = ε0 c0 E02 (ω) = ~ωN (ω) (9.11)
2 V
Z∞
→
− − −
→−
A (→
→
r , t) = εb A0 (ω)ei( k · r −ωt+ϕ) dω (9.12)
−∞
Z∞
→
− − −
→ →
A (→
r , t) = εb A0 (ω) cos( k · −
r − ωt + ϕ)dω (9.13)
0

Physics of Atoms and Molecules

102 9 Interaction of atoms with electromagnetic radiation

Here, ϕ is a random phase and for a given xed direction of εb the electromagnetic
eld is linear polarized.

The Hamiltonian of a spinless particle of charge q and mass m in an electroma-

gnetic eld is
1 − −
→
H= (→p − q A )2 + qΦ (9.14)
2m
Ignoring small spin-dependent terms the time-dependent SE is
" µ ¶2 #
1 ~→
− →
− − Ze 2
∂ψ(−
→r , t)
∇ + e A (→r , t) + eΦ(−→
r , t) − ψ(−
→
r , t) = i~ (9.15)
2m i 4πε0 r ∂t

−
→→ −
→→
Assuming time independent elds A (−
r , t) = A (−
r ) and Φ = Φ(−
→
r ) the Hamilton
operator becomes time independent and the wave function is

r , t) = e−i ~ ψ(−
ψ(−
→ →
Et
r) (9.16)
−
→ → − −
→ − → −
→ − →
and the SE becomes with ∇ · ( A ψ) = ( ∇ · A )ψ + A · ∇ψ
" #
~2 i~e −
→ −
→ i~e − → − → e2 − → → − Ze2 −
→
− 4− A·∇− ∇·A+ (A · A) − + eΦ( r ) ψ = Eψ
2m m 2m | {z } 2m 4πε0 r
=0

· ¸
Φ=0 ~2 i~e →
− −
→ e2 −→ − → Ze2
⇒ − 4− A·∇+ (A · A) − ψ = Eψ (9.17)
2m m 2m 4πε0 r

In a constant magnetic eld we can write

−
→ 1 →− −
A = (B × →r) (9.18)
2
i~e →
− −
→ i~e −→ → − →
− A·∇ = − (B × −r )· ∇ (9.19)
m 2m
i~e −
→ − −
→ e →
− − →
= − B · (→
r × ∇) = B · L. (9.20)
2m 2m
−
→ →
−
Here we used the denition L = →
−
r ×− →p = −i~→−
r × ∇ of the orbital angular
→
− − →
momentum operator. The quadratic term A · A is
→
− − → −
→ →2
e2 A · A e2 ( B × −r)
= (9.21)
2m 8m
2 −
→ 2 2
e 2 2 −
→ − → 2 B =Bz e B
= (B r − ( B · r ) ) = (x2 + y 2 ) (9.22)
8m 8m

Physics of Atoms and Molecules

9 Interaction of atoms with electromagnetic radiation 103

Here we choose the axis of the magnetic eld to be parallel to the z-axis. If we
−
→
compare the terms linear and quadratic in A we nd
e −
→ → − e~
B·L ≈ B, (1) (9.23)
2m 2m
e2 B 2 2 e2 B 2 2
(x + y 2 ) ≈ a , (2) (9.24)
8m 4m 0
(2) ea2 B
⇒ = 0 ≈ 10−6 B, with B in T (9.25)
(1) 2~

Therefore, the second term is negligible and term (1) describes the potential
−
→
energy of a magnetic dipole moment − →µ in a magnetic eld B . The magnetic
dipole moment −→µ is dened by
−
→
−
→ e −
→ L
µ =− L = −µB . (9.26)
2m ~
With Bohr's magneton µB :
e~
µB = ≈ 9.27408 × 10−24 Am2 , (Joule/Tesla). (9.27)
2m
Now we dene this small interaction to be a perturbation H 0 to the system
−
→
0 −
→ − → −
→ L
H = − B · µ ≈ 0.4669 B · cm−1 , B in Tesla (9.28)
~
and the SE becomes
· ¸
~2 Ze2 0
− 4− + H ψ = Eψ. (9.29)
2m 4πε0 r

For H 0 = 0 we know the eigenfunctions ψn`m` of the hydrogen atom, that are
−
→
also eigenfunctions of L 2 and Lz . If we choose the magnetic eld parallel to the
z-axis we have
e
H0 = BLz . (9.30)
2m
With this setting the SE is readily solved without perturbation theory, since the
eigenfunctions ψn`m` are eigenfunctions of the perturbation Lz also

mZ 2 e4 eB mZ 2 e4
Enm` = − + ~m` = − + ~ωL m` . (9.31)
2~2 (4πε0 )2 n2 2m 2~2 (4πε0 )2 n2
The magnetic quantum numbers can take on values −`, . . . , `, and ωL is called
Larmor frequency.
eB
ωL = (9.32)
2m

Physics of Atoms and Molecules

104 9 Interaction of atoms with electromagnetic radiation

ml = 2
E=hω
1
l=2
0
-1
-2

1
l=1 0
-1
∆m l = +1
∆m l = 0
∆m l = -1

σ π σ
Abbildung 9.1. In the presence of a magnetic eld the 2` + 1 states are not degenerate
eB~
any more. The energy spacing is ∆E = 2m and the transitions are connected to dierent
polarizations for π and σ .

We observe a term splitting for the 2` + 1 terms, that are no longer degenerate.
Note, only in hydrogen atoms the degeneracy of states with the same n and
m` is still valid. For a magnetic eld of 1 Tesla (Lab magnet) the splitting is
approximately 9.27 × 10−24 J ≈ 0.58 × 10−4 eV, which is small compared to the
ground state energy of the hydrogen atom (13.6 eV). Possible transitions between
the energy levels obey selection rules. Selection rules state which transitions are
allowed (other transitions are forbidden). Here, the selection rules are

∆m` = 0 (π) (9.33)

∆m` = ±1 (σ). (9.34)

Thus, three transitions are allowed. They are called Lorentz triplet (1896). They
were observed in an experiment called the normal Zeeman eect that is shown in
Fig. 9. Moreover, the number of observable transitions depend on the direction of
observation. For observation directions perpendicular to the magnetic eld, three
optical transitions are detectable with linear polarizations along the z-axis (π )
and the y-axis (σ ). Detection along the z-axis (longitudinal) result in two optical
transitions absorbing circular polarized light (σ − and σ + ).

Physics of Atoms and Molecules

9 Interaction of atoms with electromagnetic radiation 105

z
B

y L
B=0:
z T
T

x B>0: σ π σ
L
r l
- +
σ σ
Energy

Abbildung 9.2. Normal Zeeman eect: If no magnetic eld (B = 0) is applied to the set-
up, only a single absorption line is observed along the z-axis (L) and x-axis (T). When a
constant magnetic eld is directed along the z-axis, two absorption signals are observed along
the longitudinal direction (L) and three absorption signals along the transversal direction (T).
The polarization of the absorbed elds are circular (σ − and σ + ) for the longitudinal direction
and parallel (π and σ ) along the transversal direction. The energy splitting between the three
levels is e~B
2m each.

If we would take into account the additional interaction of the spin with the
magnetic moment → −
µs
−
→
−
→ S
µs = −gs µB (9.35)
~
−
→ −
we would nd an additional term in the SE with − B · → µs . In such experiments
−
→ −→
with a weak magnetic eld the L · S coupling introduces an additional term into
−
→
the SE leading to the anomalous Zeeman eect. For | S | = 0 the normal Zeeman
eect will be observed.

The transitions between specic energy levels are due to the interaction of the
atoms with the electromagnetic eld. Since, the electromagnetic eld is described
−
→
by a time-dependent vector potential A (t) the perturbation H 0 is no longer time-

Physics of Atoms and Molecules

106 9 Interaction of atoms with electromagnetic radiation

independent and the SE becomes

∂ψ
[H0 + λH 0 (t)] ψ = i~ (9.36)
∂t

9.1 Time-dependent perturbation theory

The time-dependent perturbation theory is used for situations where the per-
turbation to the system λH 0 (t) is small and time dependent. The unperturbed
system has the eigenfunctions {ψn } of the SE H0 ψn = En0 ψn . The perturbation
is time-dependent and in our case given by
~e −
→ −
→
λH 0 (t) = −i A (t) · ∇ (9.37)
m
→ h i(−
~e − →−
k ·→
r −ωt)
−
→→
−i( k ·−
r −ωt)
i −
→
= −i A 0 e +e ·∇ (9.38)
m
Since the eigenfunctions {ψn } are orthonormal and dene a complete set, the
general solution ψ(−
→
r , t) of
∂ψ(−
→r , t)
[H0 + λH 0 (t)] ψ(−
→
r , t) = i~ (9.39)
∂t
can be expanded by
X
ψ(−
→ ψn (−
→
(0)
r , t) = cn (t)e−iEn t/~
r) (9.40)
n

where cn (t) are time-dependent coecients describing the amplitude changes

changes of the eigenfunctions with time. To nd the coecients, equation (9.40)
is inserted into the SE (9.39), and we have
X (0) £ ¤ X µ dcn (0)
−iEn t/~ 0
cn (t)e H0 ψn + λH (t)ψn = i~ e−iEn t/~ ψn (9.41)
n n
dt
¶
iEn0 (0)
−iEn t/~
− cn (t)e ψn
~
X (0) X dcn (0)
λ cn (t)H 0 (t)e−iEn t/~ ψn = i~ e−iEn t/~ ψn (9.42)
n n
dt
R ∗ X
ψ [...] (0) dcm −iEm (0)
⇒ λ cn (t)e−iEn t/~ hψm |H 0 (t)ψn i = i~ e t/~
(9.43)
n
dt
ωmn =
0 −E 0
Em n
dcm X
⇒ ~
i~ = λ cn (t)eiωmn t Hmn
0
(t) (9.44)
dt n

The system of coupled dierential equations (9.44) is completely equivalent to the

−
→
original SE i~ ∂ψ(∂tr ,t) = Hψ(→
−
r , t), and no approximations has been made thus

Physics of Atoms and Molecules

9.1 Time-dependent perturbation theory 107

far. However, if the perturbation λH 0 is weak, we can expand the coecients cn

and cm in powers of the parameter λ as

cn (t) = c(0) (1) 2 (2)

n (t) + λcn (t) + λ cn (t) + . . . . (9.45)

Substituting this expansion into the system (9.44) and equating the coecients
of equal powers of λ, we nd that

ċ(0)
m = 0 (9.46)
1 X (0)
ċ(1)
m = c (t)Hmn 0
(t)eiωmn t (9.47)
i~ n n
1 X (1)
ċ(2)
m = c (t)Hmn 0
(t)eiωmn t (9.48)
i~ n n
.. ..
. .
1 X
ċ(s+1)
m = c(s) (t)Hmn
0
(t)eiωmn t (9.49)
i~ n n

Thus, the original system (9.44) has been decoupled in such a way that the equa-
tions can now in principle be integrated successively to any given order. The rst
(0)
equation (9.46) simply conrms that the coecients cm are time-independent.
Since we want to describe transitions from an initial state ψi to a nal state
ψf , we assume that for t ≤ t0 the system is in the initial state ψ(t ≤ t0 ) = ψi
with energy Ei0 and the perturbation H 0 (t) is switched on at time t0 = 0. The
probability of nding the system in the state m is |cm (t)|2
(1)
dcm 1 iωmi t 0
⇒ =e Hmi (t) (9.50)
dt i~
Zt
(1) (1) 1
ci (t) = ci (t0 ) + Hii0 (t0 )dt0 (9.51)
i~
t0
Zt
1 0
c(1)
m (t) = c(1)
m (t0 ) + eiωmi t Hmi
0
(t0 )dt0 (9.52)
i~
t0

where the integration constants in (9.52 and 9.51) has be chosen in such a way
(1) (1)
that cm (t) = ci (t) vanish at t = t0 , before the perturbation is applied. To
rst order in the perturbation, the transition probability corresponding to the
transition i → m is therefore given by
¯ t ¯2
¯Z ¯
(1) (1) 2
¯
1 ¯ 0
¯
iωmi t0 0 ¯
Pmi (t) = |cm (t)| = 2 ¯ Hmi (t)e dt ¯ , i 6= m (9.53)
~ ¯ ¯
t0

Physics of Atoms and Molecules

108 9 Interaction of atoms with electromagnetic radiation

The coecient ci of the state i is given to rst order in perturbation by

(0) (1)
ci (t) = ci + ci (t) (9.54)
Zt
1
≈ 1+ Hii0 (t0 )dt0 (9.55)
i~
t0
Rt
− ~i 0 (t0 )dt0
Hii
≈ e t0
(9.56)
so that |ci (t)|2 ≈ 1 and the main eect of the perturbation is to change the phase
of the initial state.

9.1.1 Time-independent perturbation

If H 0 is time-independent, except for being switched on suddenly at a given time
t0 = 0. We then have
(1) 1 0
ci (t) = H t (9.57)
i~ ii
0
Hmi ¡ ¢
c(1)
m (t) = 1 − eiωmi t , m 6= i (9.58)
ωmi ~
0 2
(1) 2|Hmi |
Pmi (t) = |c(1)
m (t)| 2
= 2
F (t, ωmi ) (9.59)
~
1 − cos(ωmi t) 2 sin2 (ωmi t/2)
F (t, ωmi ) = 2
= 2
(9.60)
ωmi ωmi
The height of the peak of the function F (t, ω) is proportional to t2 , while its
width is approximately 2π/t. Setting ωt/2 = x we have
Z∞ Z∞
sin2 x
F (t, ω)dω = t dx = πt, (9.61)
x2
−∞ −∞
¡ t ¢2
and for t → ∞ we have F (t, ω) ≈ πtδ(ω) and for ω → 0 we have F (t, ω) → 2
.
From that we can conclude for the transition probability
ˆ At a xed time t, F (t, ω) has a sharp peak of width 2π/t about the value
ωmi = 0. It is clear that transitions to m for which ωmi does not deviate
form zero by more than δωmi ≈ 2π/t will be strongly favoured. Therefore
the transitions i → m will occur mainly towards those nal states m whose
energy is located in a band of width δE ≈ 2π~/t
ˆ If the transition is such that ωmi = 0, Em = Ei , and the states are degene-
rate, then the transition probability is given by
0 2
(1) |Hmi | 2
Pmi (t) = 2
t (9.62)
~

Physics of Atoms and Molecules

9.1 Time-dependent perturbation theory 109

(1)
and at sucient length of time the quantity Pmi (t) will no longer satisfy
(1)
the inequality Pmi (t) ¿ 1 required by the perturbation approach. Hence,
the perturbation method cannot be applied to degenerate systems which
are perturbed over long periods of time.
ˆ If ωmi 6= 0 we have
0 2
(1) 4|Hmi | ωmi t
Pmi (t) = 2 2 sin2 ( ) (9.63)
~ ωmi 2
0 |2
2|Hmi
(1)
and Pmi (t) oscillates with a period 2π/|ωmi | about the average value 2 2
~ ωmi
.
(1)
If the perturbation H 0 is sucient weak, the inequality Pmi (t) ¿ 1 can
always be satised. Note, for times t small with respect to the period of
oscillation one has
0 2
µ ¶2 2
(1) 4|Hmi | ωmi t 0 2t
t small : Pmi (t) ≈ 2 2 ≈ |Hmi | 2 (9.64)
~ ωmi 2 ~

ˆ If we consider transitions involving a group of states n whose energy En lies

(0) (0) (0)
within (Em − η, Em + η ) centered about Em and we denote ρn (En ) the
density of levels on the energy scale, so that ρn (En )dEn is the number of
nal states n in an interval dEn containing the energy En . The rst-order
(1)
transition probability Pni (t) from the initial state i to the group of nal
states n is then given by
X Z∞
(1)
0 2
4|Hni | sin2 ( ωni t
)
Pni (t) = |c(1)
n (t)|
2
= 2
2
ρn (En )~dωn (9.65)
n
~2 ωni
−∞
(0)
Z
Em +η
2 0 2
= 2 |Hni | F (t, ωni )ρn (En )dEn (9.66)
~
(0)
Em −η
Z∞
2 0 2
≈ ρn (En )|Hni | F (t, ωni )dωn (9.67)
~
−∞
2π 0 2
= ρn (En )|Hni |t (9.68)
~
If t is large enough so that η À 2π~/t the dominating part of the integral
arises from transitions which conserve the energy (within δE = 2π~/t), and
we have
(1) 2π 0 2 2π 0 2
Pni (t) = |Hni | ρn (E)t ⇒ Wni = |H | ρn (E) (9.69)
~ ~ ni
with Wni is the transition probability per unit time or transition rate. This
is called 'the Golden Rule' of perturbation theory.

Physics of Atoms and Molecules

110 9 Interaction of atoms with electromagnetic radiation

9.1.2 Periodic perturbation

Let us now consider a periodic perturbation
H 0 (t) = V eiωt + V ∗ e−iωt . (9.70)
With this denition H 0 (t) is hermitian, and the system is initially (for t ≤ 0) in
(0) (0)
the unperturbed bound state ψi , of energy Ei , so that the initial conditions
are ci (t ≤ 0) = 1 and cm (t ≤ 0) = 0 for m 6= i. To calculate cm (t) we substitute
(9.70) into (9.52) and use the fact that t0 = 0. This gives
 
Zt Zt
1  0 0
c(1)
m (t) = Vmi ei(ωmi +ω)t dt0 + Vmi ∗
ei(ωmi −ω)t dt0 
i~
0 0
¯
(0) (0) ¯ 1−e
(0) (0)
i(Em −Ei +~ω)t/~
~ωmi =Em −Ei (1) ¯
⇒ Pmi (t) = ¯Vmi +
¯ (0)
Em − Ei + ~ω
(0)

¯2
i(Em −Ei −~ω)t/~ ¯
(0) (0)
∗ 1−e ¯
Vmi ¯ (9.71)
Em − E − ~ω ¯
(0) (0)
i

(0)
Only one of the two terms contribute to the transition probability. For Em =
(0)
Ei + ~ω the second term will have an appreciable magnitude, and the corre-
sponding transition probability being given by
(1) 2 ∗ 2
Pmi (t) = |V | F (t, ωmi − ω). (9.72)
~2 mi
The two terms describe the physical processes of
(0) (0)
~ω = Em − Ei , Absorption (9.73)
(0) (0)
~ω + Em = Ei , Emission (9.74)
(0) (0)
and for discrete transitions the density of states becomes ρn (E) = δ(Em − Ei ∓
~ω). The main dierence to the time-independent case is that ωmi is replaced
by ωmi − ω . We see that the transition probability will only be signicant if
Em = Ei + ~ω and the system has absorbed an amount of energy given by ~ω . If
we take into account that the electric eld is a vector we can write
e→− →
V = − E ·− r (9.75)
2
e−→ → ¡ iωt ¢
= − E0 · − r e + e−iωt (9.76)
2
1−→¡ ¢
0
Hmi (t) = − E0 eiωt + e−iωt hψm |e− →r ψi i (9.77)
2
→
−µ mi = hψm |e− →r ψi i (9.78)
1−→ → ¡ iωt ¢
0
Hmi (t) = − E0 · − µ mi e + e−iωt . (9.79)
2

Physics of Atoms and Molecules

9.2 Absorption and emission 111

The vector −
→
µ mi is called transition dipole moment for the transition i → m. The
(1)
coecient |cm (t)|2 is given for long times as
1 − → → 2
|c(1) 2
m (t)| = 2
|E 0 · −
µ mi | (πδ(ω + ωmi ) + πδ(ω − ωmi ))t. (9.80)
~
The energy conservation is fullled by the delta functions for absorption and
emission. The Probability is given by
¯−→ − ¯2
¯E · →µ i(ωmi −ω)t ¯
(1) 2 ¯ 0 mi e − 1 ¯
|cm (t)| = ¯ ¯ (9.81)
¯ 2~ ωmi − ω ¯
−
→ − 2
|E0 · → µ mi | 2 sin2 ((ωmi − ω)t/2)
= (9.82)
2~2 (ωmi − ω)2
−
→ → 2 2
|E0 |2 |−
µ mi | sin ((ωmi − ω)t/2)
= (9.83)
3~2 (ωmi − ω)2
Here, we used an isotropic distribution of transition dipole moments. For conti-
nuous distribution of states with nearly constant density of states about ωmi = ω
we have the transition probability integrated over the frequencies
Z∞
1 − → → 2
Pm (t) = 2
|E 0 |2 |−
µ mi | ρ(ω) F (t, ωmi − ω)dω (9.84)
6~
−∞
π − → → 2
= 2
|E 0 |2 |−
µ mi | ρ(ω)t (9.85)
6~
and we can nally calculate the transition rate
Pm (t) π −→ → 2
Wmi = = 2 |E0 |2 |−
µ mi | ρ(ω) (9.86)
t 6~
that is constant and given by the transition dipole moment and the strength of
the electric eld. ThePtransition rates for absorption and emission are equal since
the dipole operator k e− →
rk is hermitian hψm |e→−
r ψi i = hψi |e→
−
r ψ m i.

9.2 Absorption and emission

Using a perturbation of the form
0 e−→ −
→
Hmi (t) = −i~ A·∇ (9.87)
m
the equation for the coecient becomes
Zt
1 −
→ → − 0
c(1)
m (t) = hψm | A · ∇ψi iei(ωmi )t dt0 . (9.88)
i~
0

Physics of Atoms and Molecules

112 9 Interaction of atoms with electromagnetic radiation

Inserting (9.13) we have


Z∞ Zt
e −
→→
i k ·− −
→ 0
c(1)
m (t) = − A0 (ω)dω e hψm |e
iϕ r
εb · ∇ψi i ei(ωmi −ω)t dt0
2m
0 0

Zt
−
→− → −
→ 0
+e−iϕ hψm |e−i k · r εb · ∇ψi i ei(ωmi +ω)t dt0  . (9.89)
0

The rst term describes absorption, the second one emission. The probability for
the system to be in the m state at time t (with initial conditions |cm (t ≤ 0)|2 = 0
and |ci (t ≤ 0)|2 = 1) is

πe2 2
|cm (t)|2 = A (ωmi )|Mmi (ωmi )|2 t. (9.90)
2m2 0
Thus the probability increases linearly with time and the transition rate for ab-
sorption (integrated over ω ) can be dened in rst-order perturbation theory
as
4π 2 e2 I(ωmi )
Wmi = 2
|Mmi (ωmi )|2 (9.91)
m2 c0 4πε0 ωmi
4π 2 α~2
σmi = |Mmi (ωmi )|2 (9.92)
m2 ωmi
where α = e2 /(4πε0 ~c0 ) ' 1/137 is the ne structure constant. The term σmi is
called the absorption cross-section which is the rate of absorption of energy (per
atom) divided by I(ωmi ). It has the dimensions area divided by time.

Evaluating the second term gives the transition rate for emission (integrated over
ω ):
∗
M̄im = −Mmi (9.93)
2 2
4π e I(ωmi )
W̄im = 2
|M̄im (ωmi )|2 (9.94)
m2 c0 4πε0 ωmi
W̄im = Wmi (9.95)
σ̄im = σmi (9.96)
Wmi 4π 2 1 X −
Bmi = = 2 | −e→
r n |2 (9.97)
ρ 3~ 4πε0 n

Here, we see that the absorption rate is the same as the emission rate as found
before, and the absorption and emission cross-sections are also the same. This is
consistent with the principle of detailed balancing, which says that in an enclosure

Physics of Atoms and Molecules

9.3 Selection rules 113

containing atoms and radiation in equilibrium, the transition from i to f is the

same as that from f to i, where f and i are any pair of states. The Einstein
coecients Bmi depend on the matrix element Mmi which determines the strength
of a transition.

9.3 Selection rules

−
→
Having neglected the term A 2 in the interaction of matter with light, the SE
may be written as

∂ →
i~ ψ(−
r , t) = [H0 + H 0 (t)]ψ(−
→
r , t) (9.98)
∂t
If we assume a time-dependent perturbation caused by an oscillating electroma-
gnetic eld (light wave) with polarization direction ε̂, the perturbation becomes

i~e −
→− → →
−
H 0 (t) = − A0 ei( k · r −ωt) ε̂ · ∇ + c.c. (9.99)
m

The time-dependent term with eiωt is analogous to the periodic perturbation,

which we already calculated in equation (9.86). Now we concentrate on the other
part Z −
→− −
→ →
H ∝ ψ ∗ (− →r )ei( k · r ) ε̂ · ∇ψ (−
0 →
mi m r )dV.
i (9.100)
−
→−→
For the phase factor ei( k · r ) it is
−
→ →n
(i)n ( k · −
∞
X
−
→→
i( k ·−
r) r)
e = . (9.101)
n=0
n!

For most of the experiments the atoms are much smaller than the wavelength
−
→ −
(except for x-rays) and the phase is constant over the atom (| k · → r | ¿ 1 and
−
→ − →
⇒ λ À a0 ). We can set i( k · r ) = 0 and the phase term is unity. This is called
−
→−→
the dipole approximation (ei( k · r ) = 1), and the next higher approximation the
−
→− −
→ →
quadrupole approximation (ei( k · r ) = i k · −
→
r ). With the dipole approximation
the integeral becomes
Z Z
∗ −
→ →
− →
− ∗ − →
−
ψm ( r )ε̂ · ∇ψi ( r )dV = ε̂ · ψm (→
r ) ∇ψi (−
→
r )dV (9.102)

The Heisenberg equation of motion reads

−
→ im −
→ → i−
− →
p m
p = [H0 , r ] ⇒ ∇ = = − 2 [H0 , →
−
r ]. (9.103)
~ ~ ~

Physics of Atoms and Molecules

114 9 Interaction of atoms with electromagnetic radiation

Inserting (9.103) into (9.102), using H0 ψk = E0 ψk we have

Z ³ m´Z
∗ →− −
→ − → ∗ −
ε̂ · ψm ( r ) ∇ψi ( r )dV = ε̂ · − 2 ψm (→ r )[H0 , −
→
r ]ψi (→
−r )dV (9.104)
~
³ m ´ Z
∗ −
(→r )−
→r ψi (−
→
(0) (0)
= − 2 (Em − Ei ) ε̂ · ψm r )dV
~
³ mω ´ Z
∗ −
(→r )−
→r ψi (−
→
mi
= − ε̂ · ψm r )dV (9.105)
~
(0) (0) −
→
Here we used (Em − Ei ) = ωmi ~.If we insert the vector potential A we have
Z
i~e ³ mωmi ´ ∗ →
0
Hmi = − A0 − ε̂ · ψm (−
r )−
→r ψi (−
→r )dV (9.106)
m ~
Z
−
→ ∗ −
→
= e E (ωmi ) · ψm r ψi dV (9.107)
−
→
= − E (ωmi ) · →
−
µ mi (9.108)
where 9.108 describes the interaction of the electromagnetic eld with the tran-
sition dipole moment. For (for i = m) it is the potential energy of an electric
dipole in an electric eld. The transition dipole moment is given by
Z
−
→µ mi = −e ψm ∗ −
→
r ψi dV (9.109)
Z Ã n N
!
X X −
→
= ψ ∗ −e −
→
r +e Z R ψ dV . (9.110)
m ` k k i
`=1 k=1

The intensities in dipole approximation are found

Z Ã n N
!
X X −→
I(ωmi ) ≈ |ε̂ · ψm ∗
−e →
−r` + e Zk Rk ψi dV |2 = |ε̂ · →
−
µ mi |2 (9.111)
`=1 k=1

where ε̂ is the polarization vector of the electromagnetic eld. The selection rules
say that transitions are forbidden if → −µ mi ≡ 0. To test if transitions are forbid-
den symmetry arguments are used and the term ε̂ · − →r is expressed by spherical
harmonics with ` = 1 and m = 0, ±1:
ε̂ · −
→
r = εx sin ϑ cos ϕ + εy sin ϑ sin ϕ + εz cos ϑ (9.112)
r ½µ ¶ µ ¶ ¾
4π εx + iεy −εx + iεy
= √ Y1,−1 + √ Y1,+1 + εz Y1,0 (. 9.113)
3 2 2
Now we investigate the transition from the initial state ψi = ψni ,`i ,mi to the nal
state ψf = ψnf ,`f ,mf :
Z
−
→
hε̂ · r i = ψf∗ ε̂ · →
−r ψi dV (9.114)
Z
= r Rni ,`i (r)Y`i ,m`i (ϑ, ϕ)−
Rn∗ f ,`f (r)Y`∗f ,m` (ϑ, ϕ)ε̂ · −
→ →
r dV (9.115)
f

Physics of Atoms and Molecules

9.3 Selection rules 115

Inserting (9.113) we nd

Z Z
Y`∗f ,m` (ϑ, ϕ)Y1,m (ϑ, ϕ)Y`i ,m`i (ϑ, ϕ).d cos ϑdϕ (9.116)
f

Z2π
e−im`f ϕ eimϕ eim`i ϕ dϕ = δm+m`i −m`f ,0 (9.117)
0

The integral (9.117) does not vanish if m`f − m`i = m = 0, ±1. Thus, our rst
selection rule in dipole approximation is
∆m` = 0, ±1. (9.118)

The next selection rule follows from inversion symmetry. The integral I(−
→
r ) should
→
−
not vanish going from r → − r : →
−
Z Z
I( r ) = e ψf r ψi d r ∝ Y`∗f ,m` −
−
→ ∗−→ −
→ →r Y`i ,m`i d cos ϑdϕ. (9.119)
f

−
→
Since Y`,m (−
→
r ) → (−1)` Y`,m (−r) the integral is not zero if
−
→
I(−r) = (−1)`f (−1)(−1)`i I(−
→
r ) > 0. (9.120)

This is only the case if `f + `i + 1 = 0, ±2, . . . and we can directly conclude that

∆` 6= 0. (9.121)

To specify the selection rule for the quantum number ` further we remember,
that we can write the product of two spherical harmonics with `1 and `2 as a sum
of spherical harmonics with quantum numbers ` ranging from |`2 − `1 | to |`2 + `1 |.

Y1,m Y`i ,m`i = αY`i +1,m`i +m + βY`i ,m`i +m + γY`i −1,m`i +m (9.122)

Inserting this into (9.116) we have

Z Z
Ylf∗ ,m` (αY`i +1,m`i +m + βY`i ,m`i +m + γY`i −1,m`i +m )d cos ϑdϕ. (9.123)
f

Using the orthonormality of {Y`,m } the integral does not vanish for

`f = {`i + 1, `i , `i − 1} (9.124)
⇒ ∆` = ±1 (9.125)

In summary we have selection rules for dipole approximation

∆m = 0, ±1 (9.126)
∆` 6= 0 (9.127)
∆` = ±1. (9.128)

Physics of Atoms and Molecules

116 9 Interaction of atoms with electromagnetic radiation

If we apply these rules to the normal Zeeman eect we have to analyze the integral
Z ½µ ¶ µ ¶ ¾
∗ εx + iεy −εx + iεy
Y`f ,m` √ Y1,−1 + √ Y1,+1 + εz Y1,0 Y`i ,m`i d cos ϑdϕ
f
2 2
(9.129)
−
→
with the magnetic eld B is parallel to the z-axis (see experimental setup).

For longitudinal (L) observation direction we have εz = 0, since the electroma-

gnetic eld is a transverse wave.
√
ˆ m`f = m`i +1: Polarization vector is given by 1/ 2(εx −iεy ), which describes
(right handed) circular polarized light σ− .
√
ˆ m`f = m`i −1: Polarization vector is given by 1/ 2(εx +iεy ), which describes
(left handed) circular polarized light σ+ .
Note, the denition of left handed and right handed circular polarized light is
sometimes used the other way around (σ+ for right handed circular polarization
and vice versa).

Along the z-axis the observed transitions are circular polarized. In this expe-
→
−
rimental geometry the angular momentum of the photon is parallel to the k
direction (in z-direction). We have conservation of angular momentum or spin
with the spin of the photon to be ms = ±~. Thus, the case m = 0 is not possi-
ble, because upon absorption/emission the angular momentum will change. For
m l = -1
p-Orbital m l = +1
k
k
B
h
s-Orbital
S
h
k S

σ+

S helicity + h σ−
h
left handed helicity - h
circular polarized light right handed
circular polarized light

Abbildung 9.3. Interaction of right and left handed circular polarized light with matter.
Since we have conservation of angular momentum or spin only such transitions are allowed
which result in spin changes of ∆~ms,ph .

photon absorption of ms = 1 we have ∆m` = 1, for photon emission ∆m` = −1.

Physics of Atoms and Molecules

9.3 Selection rules 117

For photon absorption of ms = −1 we have ∆m` = −1, for photon emission

∆m` = +1.

For transverse (T) observation direction we have εx = 0, since the electroma-

gnetic eld is a transverse wave. Now we have three terms resulting in transitions:

ˆ m`f = m`i , here only the term with εz 6= 0 is relevant; the light is linearly
polarized in z-direction.

ˆ m`f = m`i ± 1 the contributing term is Y1,±1 and the light is linearly pola-
rized in y-direction.

Another way to argue is that if one applies light polarized in z-direction (εy = 0),
only transitions with ∆m = 0 (with Y1,m ) are possible. Light polarized in
y-direction (εz = 0) can result in transitions to ∆m = ±1 (depending on the
frequencies).

Y1,0
Y1,1
p-Orbital

∆ ml = 1

Y1,-1
∆ ml = 0
∆ m l = -1
s-Orbital

Abbildung 9.4. Interaction of light polarized in εy and εz direction with a s-orbital.

ˆ The vector − →µ mi indicates a charge redistribution going from |ψi |2 to |ψf |2

described by the transition dipole moment. Upon absorption the atomic
'size' is growing.
−
→− →
− −
ˆ The developing parameter in ei k · r is k · →
→
r , which is approximately of the
size
−
→ − ωa0 ~ωa0
k ·→ r ≈ ka0 = = . (9.130)
c0 ~c

Physics of Atoms and Molecules

118 9 Interaction of atoms with electromagnetic radiation

~2
With the transition energy of ∆E ≈ 2ma20
in a hydrogen atom we nd

−
→ − ~2 a 0
k ·→
r ≈ (9.131)
2ma20 ~c0
~ ~ me2 e2 1
≈ = 2
= =α' (9.132)
ma0 c0 mc0 4πε0 ~ 4πε0 ~c0 137
with α the ne structure constant, which is a fundamental constant for the
coupling strengths between charged particles and the electromagnetic eld.
−
→− →
In the development of ei k · r every next term will result in contributions of
order
O(α2 ) ≈ 10−4 . (9.133)

ˆ The next term after the dipole approximation contains the electric quadru-
pole and magnetic dipole contributions, which have dierent selection rules
than the electric dipole transitions, e.g. ∆` = 2. But the intensities of those
transitions are weaker by a factor of ≈ 10−4 :
Z
∗ − →
− → −
→
ψm (→
r )i( k · −r )(ε̂ · ∇)ψi d−
→
r (9.134)

−
→
inserting −
→
p ∝ ∇ we nd
−
→ → 1h −→ → → →i
−
(k ·− r )(ε̂ · →
−
p) = (k ·− r )(ε̂ · →
−
p ) + (ε̂ · −
→
r )( k · −p)
2
1h −→ − → →i
−
+ (k ·→ r )(ε̂ · −
→
p ) − (ε̂ · −
→r )( k · −
p ) (9.135)
2
= I + II
1 −→
= I + ( k × ε̂) · (− →r ×− →
p ). (9.136)
2
−
→ −
→ −
→
Since ( k × ε̂) ∝ B , and (→ −
r ×→ −
p ) ∝ L the second term (II ) gives
Z −
→ Z
−
→ − → −
→ L →
− −
→ e −→
− B · µ f i = − B · ψf (−µB )ψi d r = B · ψf∗ (
∗
r ×− →p )ψi d−
→
r.
~ 2m
(9.137)
This is the interaction of the magnetic eld with the magnetic transition
dipole moment − →µ f i , leading to the splitting of the degenerate levels of m
in a hydrogen atom in the Zeeman experiment. The rst term (I ) can also
−
→
be transformed using − →
p = [H0 , →−
r ] im
~
, H0 ψk = E0 ψk , and k · ε̂ = 0 and we
have
3
X
I ∝ ki ε̂j Qij (9.138)
i,j=1
Z
Qfijm ≡ −e ψf∗ (3xi xj − r2 δij )ψm d−
→
r. (9.139)

Physics of Atoms and Molecules

10 Spin of the electron 119

Q is the electronic quadrupole transition moment for transitions from m →

−
→
f with ki ε̂j being the components of the gradient of E (x1 = x, x2 =
y, x3 = z . Both terms I and II result in new selection rules for electronic
quadrupole transitions and magnetic dipole transitions.

ˆ The parity operator P is also a very useful operator, that commutes with
→
−
V , H , L 2 , and Lz . The quantum number α of the parity operator, here
acting on the spherical harmonics is

P Yl,m = αYl,m = (−1)` Ylm . (9.140)

Eigenvalues of the parity operator are α = ±1.

10 Spin of the electron

→
−
We already introduced the Spin of an electron to be s = 12 . The Spin operator S
operates in a two dimensional spin space an exhibits the properties of an angu-
lar momentum operator with quantum numbers ms = ± 12 . Several experiments
result in the hypothesis of the electron spin. The Stern Gerlach experiment is
shown in Fig. 10, the anomalous Zeeman eect (with the Na D doublet) and the
general doublet structure of spectral lines. The new eects found in those expe-

Inhomogeneous
magnetic field
B
Atomic beam (Ag)

Abbildung 10.1. Stern Gerlach experiment. An atomic beam splits in two parts in an in-
homogeneous magnetic eld along the z-axis which is parallel to the average magnetic eld
direction.

riments were explained by an additional interaction to the magnetic moment of

the electron spin. In Stern Gerlach experiment silver atoms were sent through an
inhomogeneous magnetic eld resulting in a splitting of two spots, with a split-
ting proportional to the eld gradient. The explanation for this observation is the
−
→
interaction of the magnetic eld B (here in z-direction) with a magnetic moment
−
→µs
−
→
V = −− →µs· B (10.1)
−
→ −
→ →
− → − →
F = − ∇V = ∇(− µ s · B) (10.2)
∂
F z = µs z B z . (10.3)
∂z

Physics of Atoms and Molecules

120 10 Spin of the electron

In a classical picture all possible magnetic dipole directions would be allowed and
no distinct spots would be visible in the experiment. Quantum mechanically the
magnetic moment − →µ is correlated with the orbital angular momentum quantum
number ` which can take on 2(` + 1) distinct values (odd number). The two
spots observed in the Stern Gerlach experiment could not be explained with
the orbital angular momentum. Additionally, experiments with hydrogen (` = 0)
show similar results, so that a magnetic moment − →µ s due to the spin of an electron
was introduced
−
→
−
→ S
µ s = −gs µB (10.4)
~
α α2
gs = 2(1 + − 0.328 2 + . . .) = 2.002319310(6) Q.E.D (10.5)
2π π
gs = 2.002319314(7) Experiment (1971) (10.6)
−
→
−
→ L
µ = −gL µB (10.7)
~
gL = 1. (10.8)

The g -factor is the gyromagnetic ratio of the magnetic moment (in µB ) to the
angular momentum (in ~). The Stern Gerlach experiment shows that −→
µ s ≈ ±µB
−
→ − z
→
and thus gs ≈ 2. As a result the dierent angular momenta L and S add to a
→
−
general angular momentum J
−
→ → − − →
J = L + S. (10.9)

The magnetic moments add up to −

→
µJ
−
→ µB −
→ →
−
µJ = →
−
µL+→
−
µ s ≈ − [ L + 2.0023 S ]. (10.10)
~
−
→
Note, J and − →
µ J are not parallel (if there is an eective spin). All atoms with
a single outermost electron in the s-orbital show the same splitting in the Stern
Gerlach experiment. Therefore, the magnetic moments of the other spins in a
P̀
closed shell cancel each other out. This is in accordance with |Y`m |2 = c.
m=−`

The spin introduces two additional terms to the SE

−
→
0
Hmag = −− →
µ s · B ext (10.11)
−
→ 1 1 dV −→ − → −
→ → −
HSL0
= −− →
µ s · B orbit = 2 2
L · S = ξ(r) L · S (10.12)
2m c r dr
−
→
The term (10.12) is the potential energy of −
→
µ s in the internal magnetic eld B orbit
introduced by the 'moving' electron itself. In the system where the electron is at

Physics of Atoms and Molecules

10 Spin of the electron 121

the origin with velocity zero, the nucleus is 'moving' and creates a magnetic eld
−
→ −
→ Ze2
of B ∝ 2m12 c2 e 1r dV
dr
L with the potential V = − 4πε 0r
. This term is called spin-orbit
interaction, explaining the doublet structure of spectral lines.

Alkali metals have only one valence electron and if ` 6= 0 we have to take into
−
→ →−
account the L · S term. The electron spin can be oriented parallel or antiparallel
−
→
to L .

Example Na with 11 electrons (one valence electron):

[(1s)2 |(2s)2 (2px )2 (2py )2 (2pz )2 |3s]
Since the inner shells screen the nucleus charges the eective potential for the
valence electron is

α(r)
Vef f = − . (10.13)
r

For small distances the potential is the Coulomb potential with nucleus charge Z ,
and for large distances the eective nucleus charge is Z = 1. Assuming a transition
−
→− →
from 3p → 3s the L · S coupling introduces a splitting of the ` degenerated states
and we nd a doublet structure of transitions (see Fig. 10). In sodium atoms we

3p 3/2
3p
L S 3p 1/2

3s 3s

Abbildung 10.2. Screening of the nucleic charge by the inner electrons; green: Coulomb po-
tential with Z = 1; black: Coulomb potential with Z ; red: eective potential due to screening.
Degenerate transitions from the 3p → 3s orbital will split because the total angular momentum
quantum number is j = ` ± s = 1 ± 12 and allows two terms.

nd two transitions from 3p → 3s with D1 , 589.59 nm and D2 , 588.96 nm (3s do

Physics of Atoms and Molecules

122 10 Spin of the electron

not split). An estimation of the spin-orbit interaction strength for atoms is

e2 dV e2
V (r) = − ⇒ = (10.14)
4πε0 r dr 4πε0 r2
1 1 dV − → − → e2
L · S ≈ ~2
2m2 c2 r dr 8πε0 m2 c20 r3
1 1
with h 3 in=2,`=1 = (10.15)
r (3a0 )3
−
→
−−→µ ·B
s ≈ 0.4 10−23 J ≈ 0.25 10−4 eV ≈ 0.2 cm−1
orbit (10.16)
with µs ≈ µB ≈ 10−23 Am2 (10.17)
−
→ 10−23 J
⇒ B orbit ≈ ≈ 1 Tesla (10.18)
10−23 Am2
The internal magnetic eld leading to spin-orbit interaction is of the strength of
external magnetic elds (Zeeman eect).

10.1 Spin-orbit and additional couplings

→
−
The orbital angular momentum operator L acts on the spatial coordinates, while
−
→
the spin operator S acts only on the spin coordinates. Therefore both operators
−
→
commute and we can add them to the general angular momentum operator J
h→
− − →i
L, S = 0 (10.19)
−
→ −
→ − →
J = L+S (10.20)
[Ji , Jj ] = i~Jk , (i = x, y, z; j = y, z, x; k = z, x, y). (10.21)
−
→
There exist simultaneous eigenfunctions ψjmj of J 2 and Jz

→
−2 1 3
J ψjmj = j(j + 1)~2 ψjmj j = {0, , 1, , . . .} (10.22)
2 2
Jz ψjmj = mj ~ψjmj mj = {−j, −j + 1, . . . , j}. (10.23)
−
→ → −
Without coupling L · S = 0 the wave functions are

ψn`m` sms ∝ Rn` (r)Y`m (ϑ, ϕ)χsms (10.24)

µ ¶ µ ¶
1 0
χ1/2,1/2 = ; χ1/2,−1/2 = . (10.25)
0 1
−
→ → −
With L · S 6= 0 the quantum numbers m` and ms are no good quantum numbers
any more, since [H0 , Lz ] 6= 0 and [H0 , Sz ] 6= 0. We need a new basis of simulta-
−
→ − → → −
neous eigenfunctions of the operators L 2 , S 2 , J 2 , Jz , which commute with each

Physics of Atoms and Molecules

10.1 Spin-orbit and additional couplings 123

−
→ − →
other and with L · S , and H0 . The new eigenfunctions are of the type ψnjmj `s
and for a given ` the possible values for j are (vector model)

|` − s| ≤ j ≤ |` + s|. (10.26)

0
If we take HSL as a perturbation

0 −
→ − →
HSL = ξ(r) L · S (10.27)
1 −
→ −
→ −
→
= ξ(r) ( J 2 − L 2 − S 2 ) (10.28)
2

with ψnjmj `s as the eigenfunctions of the unperturbed system with eigenvalues of

En , j(j + 1)~2 , mj ~, `(` + 1)~2 , s(s + 1)~2 then we can calculate the energy shifts
in rst order perturbation theory for the hydrogen atom.

a) b)
l ζ nl
S E nl
l=0
-(l+1) ζ nl
L 2 j +1:
j = 1/2 2p 3/2 2 x 3/2 +1 = 4
j = l + 1/2 2p 1ζnl
j = l - 1/2
l=1
-2 ζ nl
2p 1/2 2 x 1/2 +1 = 2

Abbildung 10.3. Spin-orbit coupling in a hydrogen atom. a) possible relative orientations of

vectors L and S to each other; b) level splitting due to spin-orbit interaction and the number
of degeneracy (2j+1) for each level.

Z ½ ¾
1 −→2 − →2 − →2
∆ESL = ∗
ξ(r) ( J − L − S ) ψnjmj `s d−
ψnjm j `s
→
r (10.29)
2
· ¸
~2 3
= j(j + 1) − `(` + 1) − hξ(r)i. (10.30)
2 4

Physics of Atoms and Molecules

124 10 Spin of the electron

For a hydrogen atom hξ(r)i can be calculated exactly

¿ À
1 Z3
= (10.31)
r3 n`m a30 n3 `(` + 12 )(` + 1)
1 Ze2 Z3
hξ(r)i = (10.32)
2m2 c20 4πε0 a30 n3 `(` + 12 )(` + 1)
`=0 j=1/2
⇒ ∆ESL = 0 (10.33)
`6=0 j=`+1/2 3
⇒ j(j + 1) − `(` + 1) − = ` (10.34)
4
`6=0 j=`−1/2 3
⇒ j(j + 1) − `(` + 1) − = −` − 1 (10.35)
4
µ ¶3 ³ αmc ´3
1 e2
= , with α = (10.36)
a0 ~ 4πε0 ~c0
µ 2 4
¶ ½ 1
mc (Zα) `, j =`+
∆ESL = · 2
1 (10.37)
3 1
4n `(` + 2 )(` + 1) −` − 1, j =`− 2
| {z }
ζn`
1
∆ESL,`+1/2 = ` ζn` , j =`+ (10.38)
2
1
∆ESL,`−1/2 = −(` + 1) ζn` , j =`− (10.39)
2

The energy splitting is proportional to the 4th power of the nuclear charge and
4th power of the ne structure constant ∆ESL ∝ (Zα)4 . The term ζn` is always
positive so that we have a splitting in two levels, one is energetically higher (j =
` + 21 ) and the other one is energetically lower (j = ` − 12 ). Thus, the degeneracy
due to ` is reduced, but there is still some degeneracy left (2j +1). The splitting in
two levels was thought to explain the doublet line structure, but the calculations
do not t with the experimental results! This is because of the fact that the
SE does not contain any relativistic corrections.
p The relativistic expression for
the kinetic energy of an electron is Ekin = p2 c20 + m2 c40 − mc20 , which can be
developed in a Taylor expansion. Relativistic corrections are included in the Dirac

Physics of Atoms and Molecules

10.1 Spin-orbit and additional couplings 125

equation (next order correction terms)

·
0 ~2 −
→2 ~2 (E 0 − V (r)) →
−2
Eψ = − ∇ + V (r) + 2 2
∇
2m 4m c0
¸
1 dV → − − → ~2 dV ∂
+ 2 2 L·S − ψ (10.40)
2m c0 r dr 4m2 c20 dr ∂r
−
→p2 →
−p4
H = + V (r) − +
2m 8m3 c20
| {z }
I
1 dV − → − → π~2 Ze2
L · S + δ(→
−
r) (10.41)
2m2 c20 r dr 2m2 c20 4πε0
| {z } | {z }
II III
E = En` + ∆ELS + ∆Erel. + ∆EDar (10.42)
2 4
· ¸
mc (Zα) 1 3
∆Erel. = − 0 1 − (10.43)
2 n (` + 2 ) 4n4
3
½ mc2 (Zα)4
∆EDar = 2
0
n3
, `=0 (10.44)
0, else
· µ ¶¸
(Zα)2 1 3
Enj = En 1 + − (10.45)
n j + 1/2 4n
v 
à !2
u
 u Zα


exact
Enj = mc2 t 1+ p −1 .

 n − j − 1/2 + (j + 1/2)2 − Z 2 α2 


The term I is the second term of the Taylor expansion and the next term for
relativistic corrections to the kinetic energy and does not act on the spin varia-
ble. Term II is the spin-orbit coupling, introducing an explicit coupling between
−
→ −
→
L and S . Term III is called Darwin term and is only relevant for ` = 0 where
δ(−
→r ) is not vanishing. This term does not act on the spin variable.

exact
The exact solution Enj is obtained by solving the Dirac equation for the Cou-
lomb potential. Enj agrees with the exact solution up to order (Zα)2 . The energy
levels depend only on the principal quantum number n and the total angular
momentum quantum number j , with j = 1/2, 3/2, . . . , n − 1/2. To each value of
j correspond two possible values of ` given by ` = j ± 1/2, except for j = n − 1/2
where we nd only one state. For example the levels 2p, 2s are split by the spin-
orbit coupling and relativistic corrections in 2 levels (doublets) as shown in Fig.
10.1. However, experiments show that there is a (very small) additional splitting
of degenerate levels called Lamb shift. In Q.E.D. 'radiative corrections' to the
Dirac theory are obtained by taken into account the interaction of the electron
with the quantised electromagnetic eld. A qualitative explanation given by T.A.

Physics of Atoms and Molecules

126 10 Spin of the electron

2p3/2
2p, 2s 2s 1/2
2p3/2
l = 0, 1
n=2 2p1/2

2s 1/2
L. S 2p1/2 ; 2s 1/2
2p1/2

L . S + rel. Lamb shift

Abbildung 10.4. Spin-orbit coupling in an atom with electron spin 1/2. The spin - orbit
coupling alone would split the 2s and the 2p levels into three dierent levels, but since the total
corrections (including relativistic eects) depend only on n and j we end up with two levels
of dierent total angular momentum quantum number j. Additional splitting of the levels is
introduced due to the Lamb shift.

Welton (1948) is that a quantised radiation eld in its lowest energy state is not
zero, but there exist zero-point oscillations. This means that even in the vacuum
there are uctuations in this zero-point radiation eld which can act on the elec-
tron, causing it to execute rapid oscillatory motions so that the charge is 'smeared
out'. If the electron is bound by a non-uniform electric el, as in atomic systems,
it will therefore experience a potential which is slightly dierent from that corre-
sponding to its mean position. In particular, the electron in a one-electron atom
is not so strongly attached to the nucleus at short distances. As a result, s states
(which are most sensitive to short-distance modications because |ψ(0)|2 6= 0
for these states) are raised in energy with respect to other states, for which the
corresponding modications are much smaller. The Lamb shift decreases with
increasing `. The modications of the s1/2 level is roughly 10% of the energy
dierence of p3/2 − p1/2 , so that the s1/2 level lies higher than the p1/2 level. The
p3/2 energy levels are shifted about 0.2% of the energy dierence d5/2 − d3/2 of
the d5/2 and d3/2 levels, resulting in an upshift of the p3/2 level with respect to
the d3/2 level. The resulting Lamb shifts are of the order of 0.03 cm−1 .

→
−
Since the dipole operator D = −e− →
r does not depend on the spin, the selecti-
on rules derived above for the quantum number ` (in the dipole approximation)
remains

∆` = ±1 (10.46)
∆j = 0, ±1 (10.47)

For example we see from Fig. 10.1 that the multiplet np−n0 s has two components.
Thus each line of the Lyman series (hydrogen atom, lower state n = 1) is split
by the ne structure into a pair of lines called a doublet, corresponding to the

Physics of Atoms and Molecules

10.1 Spin-orbit and additional couplings 127

transitions
np1/2 − 1s1/2 , np3/2 − 1s1/2 . (10.48)
We see that in the case of the Balmer series (lower state n = 2) the mulitplet

nd 5/2 j = 5/2
np 3/2 j = 3/2
nd 3/2 j = 3/2
np 1/2 j = 1/2

n'p3/2 j = 3/2
n's 1/2 j = 1/2 n'p1/2 j = 1/2
Abbildung 10.5. Transitions of the Lyman series of hydrogen.

nd − n0 p has three components and we nd seven transitions in total. However,

since the levels ns1/2 and np1/2 coincide, as well as the levels np3/2 and nd3/2 ,
the Balmer line (see Fig. 10.1) contains ve distinct components (without Lamb
shift). Since the radial integrals are the same for both transitions np3/2 − n0 s1/2
and np1/2 − n0 s1/2 , it is easy to obtain from the angular parts of those integrals,
(that is from angular momentum considerations) the ratio of the two transition
probabilities, which is found to be equal to 2. More generally, the ratios of the
transition probabilities for the most important special cases are

s1/2 − p3/2
sp trans. : =2:1 (10.49)
s1/2 − p1/2
pd trans. : p3/2 − d5/2 : p3/2 − d3/2 : p1/2 − d3/2 = 9 : 1 : 5 (10.50)
df trans. : d5/2 − f7/2 : d5/2 − f5/2 : d3/2 − f5/2 = 20 : 1 : 14. (10.51)

Under most circumstances the initial states are excited in proportion to their
statistical weights, that is the (2j + 1) degenerate levels corresponding to an
initial state are equally populated. In this case the ratios of line intensities are
the same as those of the corresponding transition probabilities (for dierent values
of mj ).

In the case of the anomalous Zeeman eect the internal eld or the spin-orbit

Physics of Atoms and Molecules

128 10 Spin of the electron

Abbildung 10.6. Structure of the Hα lines of hydrogen and term scheme. The expected optical
spectrum is shown at the bottom of the picture, thicker lines correspond to stronger transitions.
Energies are given in wavenumbers.

coupling is dominant and the external magnetic eld is weak and is treated as
the perturbation. The unperturbed Hamiltonian is

~2 Ze2 −
→ − →
H0 = − 4− + ξ(r) L · S (10.52)
2m 4πε0 r

with ψnjmj `s eigenfunctions of the unperturbed system and eigenfunctions of

−
→ → − − →
{ J 2 , L 2 , S 2 , Jz }. The perturbation is

−
→ −
→ → → −
H 0 = −−→µ J · B = −−→
µL· B −− µS · B (10.53)
µB µB
= Bz (Lz + 2Sz ) = Bz (Jz + Sz ) (10.54)
~ ~

To rst order perturbation theory the energy shift is

Z
µBz
∆E = ∗
ψnjm j `s
(Jz + Sz )ψnjmj `s d→
−
r (10.55)
~

The rst term is easy to evaluate (mj ~) but the second term is problematic and
is a result of gS = 2.
Z
µBz
∆E = µBz mj + ∗
ψnjm j `s
(Sz )ψnjmj `s d→
−
r (10.56)
~

Physics of Atoms and Molecules

10.1 Spin-orbit and additional couplings 129

To analyze the second term we use the following operator identity and examine
its expectation value in the Dirac notation
−
→ −
→ → − − → − →
2i~ V = J × V + V × J (10.57)
h→
−2 − →2 − →i −
→ −
→ →
− −
→ −
→ →
− −
→
J , [ J , V ] = 2~2 ( J 2 V + V J 2 ) − 4~2 ( V · J ) J (10.58)
−
→− → −
→ → − −→
h`sjmj | V J 2 |`sjmj i = 2h`sjmj |( V · J ) J |`sjmj i (10.59)
−
→ −
→ →
− −
→
j(j + 1)~2 h`sjmj | S z |`sjmj i = 2h`sjmj |( S · J ) J z |`sjmj i (10.60)
−
→2 − →2 − →2
−
→ J −L +S
j(j + 1)~2 h`sjmj | S z |`sjmj i = 2~mj h`sjmj | |`sjmj i (10.61)
2
2 ~2
j(j + 1)~ hSz i = ~mj hjmj `s| [j(j + 1) − `(` + 1) + s(s + 1)]|jmj `si
2
(10.62)
and we can conclude
j(j + 1) − `(` + 1) + s(s + 1)
hjmj `s|Sz |jmj `si = ~mj . (10.63)
2j(j + 1)
Inserting (10.63) we nd the energy shift
µB B z
∆Emj = µB Bz mj + hjmj `s|Sz |jmj `si (10.64)
½ ~ ¾
j(j + 1) − `(` + 1) + s(s + 1)
= µB Bz mj 1 + (10.65)
2j(j + 1)
| {z }
gJ
= gJ µB Bz mj (10.66)
and the term gJ in (10.66) is called Landé factor. We found the interaction term
for perturbation theory which is
→
−
VB = −− →µ J,ef f · B (10.67)
−
→
−
→ J
µ J,ef f = −gJ µB (10.68)
~
µB −→ − →
VB = gJ J ·B (10.69)
~
= gJ µB Bz mj (10.70)
−
→ −
→
The vector −→
µ J is not parallel to J , but −
→µ J,ef f is parallel to J . In the case of
s = 0 the Landé factor becomes gJ = 1 = gL , and if ` = 0 then it is gJ = 2 = gS .
In general this is not the case and we have gJ 6= 1, 2. For a single electron we
have j = l ± 1/2 and the energy shifts are
 2`+2 1
 2`+1 µB Bz mj , j = ` + 2
∆Emj = (10.71)
 2` 1
µ B m
2`+1 B z j
, j = ` − 2

Physics of Atoms and Molecules

130 10 Spin of the electron

In contrast to the normal Zeeman eect (s = 0) the splitting of the levels are not
equidistant any more, because the g-factor is dierent for dierent levels. This
result in more dierent transitions.

ˆ ` = 0 (s-state), gJ = 2 = gS
 4 3
 3, j = 2
ˆ ` = 1 (p-state), gJ =
 2 1
3
, j= 2

The selection rules are (dipole approximation)

∆m` = 0, ±1 ⇒ ∆mj = 0, ±1 (10.72)

The scheme in Fig. (10.1) explains the observed additional splitting of the so-
dium D1 and D2 lines upon transitions from 3p1/2 → 3s1/2 and 3p3/2 → 3s1/2 ,
respectively.

−
→ →
The connection between J , − µ J and −→
µ J,ef f can be explained as follows: The vec-
→
− −
→ −
→ →
−
tors L and L are coupled via HSL to J and precesses about the vector J with
high velocity ω = ESL /~. The magnetic moment → −
µ J also precesses fast about
−
→ −
→ −
→
the J direction. J itself precesses slowly about the magnetic eld B (z-axis),
since the external eld is much weaker than the internal eld leading to spin-orbit
coupling. Averaged over the slow precession only the component of → −
µ J that is
−
→ →
−
parallel to J is not averaged out, because during the slow precession about B
−
→
we average over many round trips of − →µ J about J . Taking this model we can
−
→
calculate −
→µ J,ef f as the projection of −
→
µ J on J .
−
→
→
− −
→ J
| µ J,ef f | = µ J · (10.73)
J
→
− − →
→
− −
→ −
→ J ·B
EB = − µ J,ef f · B = −| µ J,ef f | (10.74)
à J
→! Ã−
− → − →!
J J ·B
= − − →µJ · (10.75)
J J
à →! Ã−
− → → −!
µB −
→ − → J J ·B
= − (J + S)· (10.76)
~ J J
−
→ −
→ − →
µB B z ( J 2 + S ) · J
= Jz −
→ (10.77)
~ | J 2|
(→ −2 1 − → →
− − )
→
µB B z J + 2 ( J 2 + S 2 − L 2)
= Jz −
→2 (10.78)
~ J

Physics of Atoms and Molecules

10.1 Spin-orbit and additional couplings 131

mj
Splitting: + l + 1/2
+ l -1 + 1/2

j = l + 1/2
- l - 1/2
l, s = 1/2

+ l - 1/2
j = l - 1/2 - l + 1/2

LS + rel. Zeeman

mj
Transitions:
3/2

1/2
p3/2 -1/2

-3/2

1/2

s1/2 -1/2
p s+ s-
Dm = 0 Dm = +1
Dm = -1

6 lines

Abbildung 10.7. Splitting of energy levels and transitions induced by the anomalous Zeeman
eect. The dierent energy levels split by dierent Landé factors. For example the p3/2 level
split in 4 new levels displaced by 4/3µB Bz . The s1/2 level split in two levels displaced by 2µB Bz .
As a result the 6 transitions have dierent energies and we observe 6 transitions in total.

−
→ −
→ − →
If we insert the eigenvalues of J 2 = ~2 j(j + 1) and of S 2 , L 2 and Jz we nd
½ ¾
µB B z ~2 j(j + 1) + 12 [j(j + 1) + s(s + 1) − `(` + 1)]
hEB i = ∆Emj = ~mj 2
~ ~ j(j + 1)
· ¸
j(j + 1) + s(s + 1) − `(` + 1)
= µB B z m j 1 + (10.79)
2j(j + 1)
= gJ µB Bz mj (10.80)

Physics of Atoms and Molecules

132 10 Spin of the electron

10.2 Spin of a nucleus

In addition to the magnetic moment of the spin of the electron, the protons and
→
−
neutrons of the nucleus have also a total angular momentum I , called nuclear
spin. The nuclear spin of a proton or neutron is I = 12 and I is the nuclear spin
−
→
quantum number. The eigenvalues of the operator I 2 are I(I +1)~2 . The eigenva-
lues of the operator Iz have possible values of MI ~, with MI = −I, −I +1, . . . , I .
A nucleus may possess 2k -pole moments, with k odd for magnetic moments and
k even for electric moments. Here we consider the nucleus as a point dipole with
−
→
a magnetic dipole moment − →
µI proportional to the nuclear spin I .
−
→
−
→ I
µ I = g I µK (10.81)
~
e~ me 1
µK = = µB = µB (10.82)
2Mp Mp 1836.15
µK = 5.05078 × 10−27 JT−1 (10.83)
(−
→
µ I )z = gI µK mI (10.84)

Values of the spin, Landé factor and magnetic moment of the nucleons and some
nuclei are

Nucleus Spin I Landé factor gI µI (in µB ) Natural abundance (%)

proton p 1/2 5.588 2.792 99.88
neutron n 1/2 -3.826 -1.913 -
deuteron 21 D 1 0.857 0.857 -
3
2 He 1/2 -4.255 -2.127 -
4
2 He 0 - 0 -
12
6 C 0 - 0 -
13
6 C 1/2 1.404 0.702 1.1
16
8 O 0 - 0 -
19
9 F 1/2 5.257 2.628 100
31
15 P 1/2 2.263 1.131 100
199
80 Hg 1/2 1.005 0.502 -
201
80 Hg 3/2 -0.371 -0.556 -

The contributions of the angular momenta were added to a total angular

→
−
momentum F
−
→ →
− −→
F = J + I (10.85)
−
→ p
| F | = ~ F (F + 1). (10.86)

Physics of Atoms and Molecules

10.2 Spin of a nucleus 133

The quantum number F can take on values F = j + I, j + I − 1, . . . , j − I . In

total 2I + 1 or 2j + 1 possibilities, depending whether I is bigger than j or vice
versa. Knowing one of the numbers (I or j ) one can determine the other one by
counting the number levels in the hyperne structure. The additional magnetic
interaction energy from the hyperne coupling is
→
−
∆EHF S = −−
→
µ I · BJ (10.87)
→
−
= −µI BJ cos(∠ − →
µ I, B J) (10.88)
F (F + 1) − I(I + 1) − J(J + 1)
= gI µK BJ p (10.89)
2 J(J + 1)
a
= [F (F + 1) − I(I + 1) − J(J + 1)] (10.90)
2
g µ B
a = pI K J (10.91)
J(J + 1)
1
= 2µ0 µB gI µK |ψn00 (0)|2 (10.92)
j(j + 1)
Z3
|ψn00 (0)|2 = (10.93)
πn3 a30
∆EF +1 − ∆EF = a(F + 1). (10.94)

The magnetic eld BJ is the magnetic eld induced by the internal electrons at
the nucleus. For some atoms the magnetic eld is (in Tesla)

n 2
S1/2 2
P1/2 2
P3/2
Na 3 45 4.2 2.5
K 4 63 7.9 4.6
Rb 5 130 16 8.6
Cs 6 210 28 13

ˆ The constant a is called hyperne separation constant and can be calculated

for j = 1/2 by equation (10.92). For s-orbitals the probability density at the
nucleus does not vanish and the magnetic eld induced by the s-electrons
at the nucleus interact with the nucleic magnetic moment. This interaction
is called Fermi contact interaction and is isotropic. In the case of ` > 0 the
probability density vanishes at the nucleus and the interacting magnetic
eld BJ results from dipole-dipole interaction between magnetic moment of
the electrons and magnetic moment of the nucleus. The hyperne separation
constant becomes anisotropic.

ˆ BJ and ∆EHF S are bigger the smaller the distant of the electrons form the
nucleus is. They increase with increasing Z , and decreasing n and `.

Physics of Atoms and Molecules

134 10 Spin of the electron

ˆ The hyperne splitting is zero for I = 0 or J = 0

ˆ For a hydrogen atom the hyperne separation constant is a = 1420 MHz =

0.0475 cm−1 = 5.9 · 10−6 eV

ˆ The hyperne splitting of spectral lines is small. For 2 S1/2 of Li the splitting
is 0.027 cm−1 .

ˆ Without external magnetic eld the degeneracy is (2F + 1), with magnetic
quantum number mF .

ˆ The selection rules for optical transitions are ∆F = 0, ±1.

F=1
a/4

n=1
l=0 0.0475 cm-1
j = 1/2 3a/4
(λ = 21 cm)

F=0
e- p
Abbildung 10.8. Hyperne separation of a hydrogen atom in the ground state. The spin of
electron and proton can be oriented parallel (F=1) or antiparallel (F=0). The splitting due to
hyperne interaction (1.420 GHz) is about six times smaller than the Lamb shift.

The energy dierence can be measured either directly upon absorption of the
matching high frequency radiation leading to a spin ip or by determining the
splitting of the spectral lines. By measuring of the energy dierence one can
calculate the gS value of the electron to be 2.0023 (magnetic moment of proton
and electron is known).

10.2.1 Hyperne splitting in an external magnetic eld

An important tool to investigate the hyperne structure is applying an external
→
−
magnetic eld B 0 in addition to the internal eld. The resulting splitting of states
depend on the relative strengths of both elds. The case of weak external elds
−
→ −
→
B 0 in comparison to B J is called Zeeman-eect of the hyperne structure. Here,
−
→ − → →
− −
→ −
→
the coupling of I , J , and F is maintained and the vectors of I and J precesses
→
− −
→ −
→ →
−
about the F vector around the B 0 direction. The frequency of I and J about

Physics of Atoms and Molecules

10.2 Spin of a nucleus 135

−
→ −
→ −
→
F is fast in comparison to the precession frequency of F about B 0 . Therefore,
−
→
the remaining components are the Fz = mF ~ components ( B 0 parallel to the
z-axis) with mF = F, F − 1, . . . , −F . The selection rules are ∆F = 0, ±1 and
∆mF = 0, ±1. The energy splitting of the Zeeman hyperne splitting is
→
−
∆EZHF S = −−→
µ F · B0 (10.95)
∆EZHF S = gF µB B0 mF (10.96)
F (F + 1) + J(J + 1) − I(I + 1)
gF = gJ
2F (F + 1)
µK F (F + 1) + I(I + 1) − J(J + 1)
−gI (10.97)
µB 2F (F + 1)
Since µB À µK the second term is negligible.

→
− −
→
If the external magnetic eld becomes stronger, the vectors I and J do not
→
− − →
couple any more. The L · S coupling is stronger (coupling of electrons) and
remains, while the coupling between the nucleic magnetic moment and the elec-
tronic magnetic moment is strongly reduced. This eect is called Paschen-Back
eect of the hyperne interaction.
gI µK BJ
a = p (10.98)
J(J + 1)
∆EP BHF S = gJ µB mJ B0 + amJ mI − gI µK B0 mI (10.99)

10.2.2 Electron spin resonance (EPR)

The technique of electron paramagnetic resonance (EPR) gives information on
structure and dynamics of paramagnetic atoms and molecules. Especially the
detection of triplet states (next chapter) is possible with EPR. Generally the
molecules are diamagnetic and show no EPR signal. Paramagnetic molecules are

ˆ Molecules with paramagnetic atoms such as Fe3+ or [Fe(CN)6 )]3− which are
still paramagnetic with paired valence electrons.

ˆ Radicals with an unpaired electron. There are stable radicals such as DPPH
(Diphenyl-Picryl-Hydracyl) and radicals which can be created upon illumi-
nation or chemical reactions.

ˆ Molecules in the triplet state. For some molecules the electronic ground
state is a triplet state (O2 , N O, N O2 ) and other molecules relax into a
metastable triplet state after light excitation, such as naphthaline, chlo-
rophylls, corroles, etc. The lifetime of the excited triplet states are in the
range of 10−6 s.

Physics of Atoms and Molecules

136 10 Spin of the electron

+a/4 ms mI
B0 -1/2gIµKB0
+1/2 +1/2

+1/2gIµKB0
1/2gµBB0 -a/4 +1/2 -1/2

+1/2gIµKB0
+a/4 -1/2 -1/2

-1/2gµBB0 -a/4 -1/2gIµKB0

-1/2 +1/2

Zeeman energy Hyperfine Zeeman energy Magnetic dipole

of the electron interaction of the nucleus transitions
gµBmSB0 amImS -gIµKmIB0
a

B0 B0

Abbildung 10.9. Hyperne structure of a hydrogen atom in strong magnetic eld and hy-
perne structure of the EPR. Lower panel: left: EPR spectrum of a free electron; right: EPR
spectrum of proton bound electron in a strong magnetic eld.

p
An electron with the magnetic moment |→−
µs | = µB gS s(s + 1) has two possible
−
→
orientations ms = ±1/2 in a magnetic eld B 0 with energy dierence ∆E

∆E = gs µB B0 (10.100)

Applying an electromagnetic eld perpendicular to the magnetic eld with the

frequency ν = 2.8026 · 1010 B0 Hz T
can induce transitions between the spin levels
of a 'free electron'. Selection rules for the magnetic dipole transitions are

∆ms = ±1. (10.101)

For magnetic elds in the range of 0.1T to 1T the frequencies are in the range
of microwaves (frequencies of GHz). The g-factor of the paramagnetic electron
can be determined by EPR. Generally the g-factor is a tensor and has dierent
values along dierent axes. This is because of the interaction with the magnetic
elds induced by the 'local currents' of the neighboring electrons (in chemical
bonds). Thus, dierent chemical bonds can produce dierent chemical shifts.
Nevertheless, the g-factor is often very similar to gS (free electron).

More important are the contributions from the hyperne interaction with the
−
→
nuclear spin I . The external and internal magnetic elds add up for the electron,
where dipole-dipole interactions between nucleic and electronic magnetic dioples

Physics of Atoms and Molecules

10.2 Spin of a nucleus 137

are averaged out in solution (rotation of the molecules). The remaining part is
the Fermi contact interaction

Bloc = B0 + amI (10.102)

³ a´
hν = gµB B ± (10.103)
2
∆ms = ±1, ∆mI = 0 (10.104)

In an experiment with N equivalent protons interacting with one electron (e.g.

mI
+1/2

B0 -1/2

-1/2

+1/2

B0

Abbildung 10.10. Hyperne interaction of the magnetic moments of an electron and a nucleus
(I = 1/2). In total there are four levels and two resulting transitions. Because of the experi-
mental setup the measured signals are derivatives of absorption signals. They are separated by
the hyperne separation constant.

benzene-radical) we nd N + 1 equivalent hyperne lines with intensity ratios

given by the Pascal triangle (1:6:15:20:15:6:1). Here we have to count all possible
orientations of the equivalent spins.

Interaction between two dierent electron spins introduce a dipole-dipole inter-

action term HD
(−→ − → −→ → −→ − )
S · S 3( S · −
r )( S · →
r )
1 2 1 12 2 12
HD = g 2 µ2B −
→ − −
→ (10.105)
r 312 r 512

Physics of Atoms and Molecules

138 10 Spin of the electron

which connects the distance between the two spins with its relative orientation
to the interaction strengths. This dipole-dipole interaction term shifts the ener-
gy levels and the energetic positions of the resonance frequencies. The shifts are
proportional to r13 and give information about the averaged distances of the in-
teracting electrons. This is especially useful in super-molecules such as proteins
(photosystems).

10.2.3 Nuclear magnetic resonance (NMR)

In NMR experiments the Lamor frequency of the nuclear spins were measu-
red. Since the magnetic moment of a nuclear spin is about 2000 times smal-
ler than the magnetic moment of an electron, the magnetic eld of the electron
has to vanish at the nuclei of the investigated atoms and molecules. This is
the case for atoms such as Hg, Cu, C, and S as well as for many molecules li-
ke H2 O, CaO, LiCl, CO2 , H2 , NH3 and most biological molecules. The potential
−
→
energy of the nucleic magnetic moment within the magnetic eld B 0 is
−
→
EN M R = −− →µ I · B0 (10.106)
= −gI µK B0 mI (10.107)
∆mI = ±1 (10.108)
|∆EN M R | = g I µK B 0 (10.109)
gI µK
νN M R = B0 (10.110)
h
gI µK
ωN M R = B0 = γB0 (10.111)
~
µI = gI Imax µK (10.112)
µI [Hz]
ν = 762.3 B0 . (10.113)
I [Gauss]
Equation (10.113) is the resonance condition for r.f. absorption in units of Hz /
Gauss (1 Tesla ∼= 10 kGauss) leading to spin ip processes. The ratio of absorption
(N1 → N2 ) and emission depends on the population ratio given by Boltzmann
statistics
N1 − N2 1 − e−gI µK B0 /(kB T ) gI µK B0
= −g µ B /(k T )
≈ . (10.114)
N1 + N 2 1+e I K 0 B 2 kB T
Protons in dierent electronic environments experience dierent amounts of shiel-
ding σ , and the resonance absorption of light energy will occur at dierent values
for the applied eld or irradiating light frequency. These changes are referred to
as chemical shifts.

Bef f = B0 − Binduced (10.115)

Bef f = B0 − σB0 (10.116)
hν = gI µK Blocal = gI µK B0 (1 − σ). (10.117)

Physics of Atoms and Molecules

10.2 Spin of a nucleus 139

r.f.

B1

N S

B0
Abbildung 10.11. NMR setup; The sample (green) is positioned in a homogeneous and con-
stant magnetic eld B0 ; perpendicular to B0 an additional magnetic eld B1 is generated with
a high frequency generator (r.f.); absorption of the r.f. eld result in nuclear spin ips and in
reducing the power of the induction coil, which is the measured quantity.

A possible NMR setup is presented in Fig. 10.2.3. In the static magnetic eld
along the z-axis, all nuclear magnetic spins precesses about the z-axis. Applying
a NMR pulse of duration τp and rotation angle β with β = −γB1 τp rotates the
equilibrium magnetization M0 about the direction of the applied r.f. eld B1 . For
an r.f. eld applied along the y-axis, the initial magnetization after the pulse is

Mx (0) = M0 sin β (10.118)

My (0) = 0 (10.119)
Mz (0) = M0 cos β. (10.120)

The subsequent free induction decay can be described in terms of two components

Mx (t) = M0 sin(β) cos(Ωt)e−t/τ2 (10.121)

My (t) = M0 sin(β) sin(Ωt)e−t/τ2 (10.122)
M (t) = Mx (t) + iMy (t) = M0 sin βeiΩt−t/τ2 . (10.123)

The complex signal s(t) obtained by simultaneous observation of both x- and

y-components by quadrature detection, is directly proportional to the complex
magnetization M (t). This signal may be subjected to a complex Fourier trans-

Physics of Atoms and Molecules

140 10 Spin of the electron

formation
Z∞
S(ω) = s(t)e−iωt dω (10.124)
0
S(ω) =
v(ω) + iu(ω) (10.125)
v(ω) =
M0 sin β a(∆ω) (10.126)
u(ω) =
−M0 sin β d(∆ω) (10.127)
∆ω =
ω−Ω (10.128)
1/τ2
a(∆ω) = (10.129)
(1/τ2 )2 + (∆ω)2
∆ω
d(∆ω) = (10.130)
(1/τ2 ) + (∆ω)2
2

The equations (10.129) and (10.130) represent absorption and dispersion signals,
respectively. Obviously, the maximum signal amplitude is obtained for a pulse
rotation angle β = π/2. The transverse relaxation time τ2 gives the width of the
absorption spectra.

Some properties of NMR spectroscopy are

ˆ The energy dierences are very small of about 10−4 to 10−8 eV.

ˆ In NMR magnetic dipole transitions were detected (not electric dipole tran-
sitions).

ˆ The wavelength of the used radiation is big with respect to the sample
dimensions. Thus, all nuclei can be excited coherently with the same phase.

As a result of the small energy dierences a reference standard - the molecule te-
tramethylsilane (CH3 )4 Si commonly abbreviated TMS - is measured in addition
to the sample of interest. The detected resonance frequencies are expressed as δ
in ppm
νi − νT M S
δi = × 106 ppm. (10.131)
ν0
In addition to the chemical shift of protons, one can also measure the chemical
shift of the carbons. In methane it is -2.1 ppm, and the chemical shift of the two
equivalent CH3 carbons in ethane is 5.9 ppm.

Direct magnetic dipole-dipole interaction between nuclei A and B change the

magnetic eld at the nucleus B. The magnetic eld from the nucleus A at the
position of nucleus B depend to the third power on the distance between the

Physics of Atoms and Molecules

10.2 Spin of a nucleus 141

nuclei, and depend on the orientation of the angle ϑ with respect to the vector
−
→r AB and is given by
µ0 1
BA = − gI µK mI 3 (1 − 3 cos2 ϑ). (10.132)
4π rAB

Assuming nucleus A to have spin 1/2, we have two orientations of the spin with
respect to the external magnetic eld B0 ± BA for parallel and antiparallel orien-
tations. The nucleus B can see two dierent magnetic elds B0 + BA or B0 − BA ,
leading to a resonance doublet of splitting 2BA . The same interaction is acting
on the nucleus A. The energy splitting of the doublet is called spin-spin interac-
tion J (see Fig. 10.2.3). In solid state samples the magnetic eld BA induced by
spin-spin coupling can be as big as 10−4 T (for distances of 0.2 nm). In liquid
samples the molecules rotate and move very fast, so that the term (1 − 3 cos2 ϑ) is
averaged out and becomes zero. In super-molecules, such as proteins the rotation
is very slow and the dipole-dipole interaction does not vanish. Therefore, in big
molecules we can make use of the dipole-dipole interaction to extract distances
and orientations from the spin-spin couplings. Indirect spin-spin coupling mecha-
nisms also exist. They are smaller than the direct coupling mechanisms and can
result from polarization of an electron magnetic moment by a nucleic magnetic
moment and vice versa. If the chemical shift is the same for several identical

∆δ

νA νB
J J

νAB νAB νBA νBA

Abbildung 10.12. Splitting of NMR lines by spin-spin coupling J. The magnetic eld of spin
A inuences the local eld of spin B and vice versa. Here, J < δ .

nuclei, then the nuclei are called equivalent. The two protons in a CH2 group
and the three protons in a CH3 group are equivalent. They have identical reso-
nance lines (single line in the spectrum). If both group are part of one molecule
(CH3 CH2 OH ), the dierent equivalent protons can interact with each other and
the 3 protons of the CH3 group split into 3 lines with intensity ratio 1:2:1, due
to the coupling to the two protons of CH2 . The coupling of the 3 CH3 protons

Physics of Atoms and Molecules

142 10 Spin of the electron

with one proton of the CH2 introduce a splitting into two lines and the coupling
with the other proton again into two lines with the same coupling constant J (→
three lines). The two protons of the CH2 group split into 4 lines with intensity
ration 1:3:3:1. The direct and indirect spin-spin interactions do not depend on the
external magnetic eld. The spin-spin interaction can be isolated by measuring
at dierent external eld strengths.

The nuclear magnetic resonance technique is very successful in visualizing the

consistence of a human body (Magnetic Resonance Tomography) and in re-
solving three dimensional structures of proteins on atomic resolution. The latter
uses coherence spectroscopy by exciting with two resonance elds ω1 and ω2 . For
short times (t < τ2 ) the two spins can interact coherently with each other resul-
ting in cross-peaks. The interaction strength gives information on the distance
and orientations of the spins.

Abbildung 10.13. Homonuclear 2D correlation spectrum of basic pancreatic trypsin inhibitor

(BPTI, with 58 amino-acid residues). Absolute mode.

Physics of Atoms and Molecules

10.2 Spin of a nucleus 143

Abbildung 10.14. Absolute-value 2D NOE spectrum of the protein seminal inhibitor II A.

Physics of Atoms and Molecules

144 10 Spin of the electron

10.3 Lifetimes of excited states

As we saw in the sections above, most experiments deal with dipole approximation
or dipole interactions. We learned that the intensity of a transition between a pair
of states f and i is proportional, in the dipole approximation, to the quantity
|−
→r f i |2 . The relative intensities of a series of transitions from a given initial state
i to various nal states f are determined by the quantity ff i called oscillator
strength
2mωf i −
ff i = |→
r f i |2 (10.133)
e~
with ωf i = (Ef −Ei )/~. Note, that fki > 0 for absorption and fki < 0 for emission
processes. The oscillator strengths obey the sum rule, due to Thomas, Reiche and
Kuhn X
ff i = 1 (10.134)
f

where the sum is over all states, including the continuum. The transition rates for
spontaneous emission in the dipole approximation are given in terms of oscillator
strengths by
2~α 2
Wfsi = ω |ff i |. (10.135)
mc2 f i
For hydrogenic atoms the oscillator strengths and transition probabilities decrease
as the principal quantum number n of the upper level increases.

10.3.1 Atomic lifetimes

If N (t) atoms are in an excited state i at a particular time t, the rate of change
of N (t) is X
Ṅ (t) = −N (t) Wfsi (10.136)
f

where Wfsi is the transition rate for spontaneous emission and the sum is over
all states f , of lower energy, to which decay is allowed by the selection rules. On
integration, N (t) can be expressed by

N (t) = N (t = 0)e−t/τi . (10.137)

where τi is called the lifetime of level i and is

1 X
= Wfsi . (10.138)
τi f

For example, the lifetime of the 2p level of a hydrogenic atom is seen to be

τ = 0.16
Z4
10−8 s. For higher nucleic charges is
1
τ (Z) = τZ=1 . (10.139)
Z4

Physics of Atoms and Molecules

10.4 Density operator 145

Average oscillator strengths for some transitions in hydrogenic atoms and ions
are shown below
Initial Final Continuum
P
∞
level level n = 1 n = 2 n = 3 n = 4 spectrum
n=5
1s np - 0.416 0.079 0.029 0.041 0.435
2s np - - 0.435 0.103 0.111 0.351
2p ns -0.139 - 0.014 0.003 0.003 0.008
2p nd - - 0.696 0.122 0.109 0.183

The lifetime of some levels of atomic hydrogen (in 10−8 s) are

Level 2p 3s 3p 3d 4s 4p 4d 4f
lifetime 0.16 16 0.54 1.56 23 1.24 3.65 7.3

10.4 Density operator

The density matrix of a pure quantum state |ψi is dened as:

ρ = |ψi|hψ|, (10.140)

when we expand ψ in a basis |ni, we nd

X
hψ| = cn hn| (10.141)
n
X
|ψi = c∗m |mi (10.142)
m
X
ρ = cn c∗m |nihm| (10.143)
n,m
ρn,m ≡ hn|ρ|mi = cn c∗m . (10.144)

The expectation value of an operator A is dened as:

X
hAi ≡ hψ|A|ψi = ρnm Anm . (10.145)
n,m

With the denition of the trace we have

X XX
T r(Aρ) = (Aρ)nn = ( Anm ρmn ) (10.146)
n n m
X
T r(A) = Ann (10.147)
n
hAi = T r(Aρ) (10.148)

Physics of Atoms and Molecules

146 10 Spin of the electron

The trace is invariant to unitary transformations (i.e. invariant to basis transfor-

mations). The time evolution of the density operator is given by
µ ¶ µ ¶
d d d d
ρ = (|ψihψ|) = |ψi hψ| + |ψi hψ| . (10.149)
dt dt dt dt

Inserting this into the SE we nd

d i
|ψi = − H|ψi (10.150)
dt ~
d i
hψ| = + Hhψ| (10.151)
dt ~
d i i
ρ = − H|ψihψ| + |ψihψ|H (10.152)
dt ~ ~
i i
= − Hρ + ρH (10.153)
~ ~
d i
⇒ ρ = − [H, ρ]. (10.154)
dt ~
Equation (10.154) is called Liouville von Neumann equation and describes the
time evolution of the system. Since the probability Pφ of nding the system in
the state φ is given by Pφ = hφ|ρ|φi (if φ is normalized), two systems are identical
if they have the same density operator.

The density operator can describe pure states ρ = |ψihψ|, but also mixed states.
Equations
d i d i
|ψi = − H|ψi ⇔ ρ = − [H, ρ] (10.155)
dt ~ dt ~
are identical, as long as ρ is the density matrix of a pure state. However, in general
we have statistical ensembles and there is no way to write down a wave function
of a statistical average, but we can write down the density matrix of a statistical
average. Let Pn be the probability of a system being in a pure state |ψi, then the
density matrix is dened as:
X
ρ = Pn |ψihψ| (10.156)
n
with Pn ≥ 0 (10.157)
X
Pn = 1 (10.158)
n
ρnm = ρ∗mn (10.159)
ρnn = ≥ 0 (10.160)
T r(ρ) = 1 (10.161)

Physics of Atoms and Molecules

10.4 Density operator 147

Let for example |ψi be one of the basis states

|ψi = |ai (10.162)
µ ¶
1 ¡ ¢
→ρ = 1 0 (10.163)
0
µ ¶
1 0
ρ = (10.164)
0 0
or |ψi = |bi (10.165)
µ ¶
0 ¡ ¢
→ρ = 0 1 (10.166)
1
µ ¶
0 0
ρ = (10.167)
0 1
and a coherent superposition state of both (which is still a pure state) is
1
|ψi = √ (|ai + |bi) (10.168)
2
µ ¶
1 1 1 ¡ ¢
⇒ ρnm = √ √ 1 1 (10.169)
2 1 2
µ ¶
∗ 1/2 1/2
= cn cm = . (10.170)
1/2 1/2
On the other hand, for a statistical average between both states with P1 = P2 =
0.5 we get µ ¶
1/2 0
ρ= . (10.171)
0 1/2
The diagonal elements are the same in both cases, but the o-diagonal elements
which describe the coherence between both states are dierent. Note, there is
no wave function |ψi which would give a density matrix as in equation (10.171).
Assume we have a Hamilton operator with eigenstate basis so that we have
µ ¶
²1 0
H = (10.172)
0 ²2
µ ¶ ·µ ¶µ ¶ µ ¶µ ¶¸
d ρ11 ρ12 i ²1 0 ρ11 ρ12 ρ11 ρ12 ²1 0
= − −
dt ρ21 ρ22 ~ 0 ²2 ρ21 ρ22 ρ21 ρ22 0 ²2
µ ¶
i 0 (²1 − ²2 )ρ12
= − (10.173)
~ (²2 − ²1 )ρ21 0
ρ̇11 = 0 ⇒ ρ11 (t) = ρ11 (0) (10.174)
ρ̇22 = 0 ⇒ ρ22 (t) = ρ22 (0) (10.175)
i ² 1 −² 2
ρ̇12 = − (²1 − ²2 )ρ12 ⇒ ρ12 (t) = e−i ~ t ρ12 (0) (10.176)
~
i ²2 −²1
ρ̇21 = − (²2 − ²1 )ρ21 ⇒ ρ21 (t) = e+i ~ t ρ21 (0). (10.177)
~

Physics of Atoms and Molecules

148 10 Spin of the electron

The diagonal elements are stationary in time (as expected) while the o-diagonal
elements oscillate with the frequency splitting ω = ∆²/~ = (²1 − ²2 )/~. In spec-
troscopic experiments the dephasing of an ensemble of molecules is observed. The
simplest approach to describe dephasing Γ (Γ real and positive) phenomenologi-
cally is:
i
ρ̇12 = − (²1 − ²2 )ρ12 − Γρ12 (10.178)
~
² −²
−i 1 ~ 2 t −Γt
ρ12 (t) = e e ρ12 (0) (10.179)
i
ρ̇21 = (²2 − ²1 )ρ21 − Γρ21 (10.180)
~
²2 −²1
ρ21 (t) = ei ~ t e−Γt ρ21 (0). (10.181)

In the wave function picture it is not possible to describe dephasing. If we intro-

duce an interaction with an optical light eld E(t):
−
→ → ¡
− ¢ →
H = H0 + E (t) · −
→
µ ' H0 + E 0 eiωt + e−iωt · −
µ, (10.182)

and we get
→
−
H = ²1 |aiha| + ²2 |bihb| − −
→
µ · E (t)(|aihb| + |biha|) (10.183)
à −
→ !
²1 −→ −
µ · E (t)
H = −
→ (10.184)
−→−µ · E (t) ²2

Nonlinear Optics:
−
→ −
→
The electric displacement D and the macroscopic polarization P is given by
−
→ −
→ − →
D = ε0 E + P (10.185)
−
→ ³ −
→ −
→→ − −
→− →−→ ´
P = ε0 χ(1) E + χ(2) E E + χ(3) E E E + . . . (10.186)
−
→
P (t) = T r(−
→
µ ρ(t)) ≡ h−
→
µ ρ(t)i, (10.187)

and for the example of a two level system we get

µ ¶
0 µ12
µ = (10.188)
µ21 0
hµρ(t)i = ρ12 µ21 + ρ21 µ12 (10.189)
(n) (n)
P (t) = hµρ (t)i (10.190)

Physics of Atoms and Molecules