Chapter 1

Introduction

1.1 Underlying concepts

Spectroscopy is the study of absorption and emission of electromagnetic radiation by mat-

ter. Its goal is to measure the structural and dynamical properties of atoms, molecules
and condensed matter with atomic to sub-atomic resolution. Over the past two centuries,
spectroscopy has greatly contributed to our knowledge of the atomic-scale details of mat-
ter. Spectroscopy has revealed the equilibrium geometry of most small molecules, including
chemical reaction intermediates and unstable species, such as radicals. Gerhard Herzberg
(1904-1999) has been one of the leading figures in this area of science and was awarded the
Nobel Prize in 1971. One representative achievement was the determination of the structure
of methylene (CH2 ) from the study of its absorption spectrum. The work of Herzberg is
summarized in a three-volume book that will serve as the main reference for this lecture (see
Appendix A and the website www.atto.ethz.ch/education/spectroscopy).
Long before the work of Herzberg, spectroscopy has had an even deeper impact on modern
science. It has driven the development of quantum mechanics. The measurement of the
emission spectrum of heated objects (so-called “black-body radiation”) has led Max Planck
(1858-1947) to his original quantum theory of radiation that we will discuss in detail be-
low. This theory established the quantum nature of light, which is nowadays of greatest
importance in areas as diverse as quantum information science and nano-photonics. Early
measurements of atomic transitions lead to the breakthrough of the atomic model proposed
by Niels Bohr (1885-1962) which initiated the development of quantum mechanics. Spectro-
scopic measurements also played a role in the discovery of the photoelectric effect by Albert

5
6 CHAPTER 1. INTRODUCTION

Einstein (1879-1955) which provided the foundations for photoelectron spectroscopy. The
latter has been developed into a remarkably powerful tool to determine the nature and en-
ergetic sequence of molecular orbitals and to identify the elemental composition of matter
(X-ray photoelectron spectroscopy).

   !"#

$"  %

*+,$
  

    
  
-.  "/
0 & ' ' ()' ' ' '

$$1
 2
-.  "

 

          
 
 

3 4 ,/ 0

        

+
$  
56 7""
""

 
7.  "
        
'89: '8&9: ; 889: <  9:

Figure 1.1: Illustration of the electromagnetic spectrum (taken from

en.wikipedia.org/wiki/Electromagnetic spectrum)

Analytical chemistry is an important sub-field that relies almost entirely on spectroscopy.

As a reminder of previous lectures, we start with the study of the electromagnetic spectrum
of light (Figure 1.1). Spectroscopy in the visible and ultraviolet regions probes transitions
between different electronic states in atoms or molecules. This will be the main focus of
the present lecture. Functional groups in molecules often possess electronic transitions at
characteristic wavelengths, e. g. around 280 nm in the case of carbonyl (–C=O) groups, that
allow their identification from spectral data. Transitions in the infrared region of the elec-
tromagnetic spectrum correspond to vibrational excitations in molecules which also have a
great importance in analytical applications. Radiation in the microwave region drives tran-
sitions between rotational energy levels of molecules which provide accurate information on
the equilibrium structure of molecules in the gas phase. Transitions in the radiofrequency
range form the basis of magnetic resonance methods like nuclear magnetic resonance (NMR)
and electron spin resonance (ESR) that have been studied in detail in the lecture Physical

PCV - Spectroscopy of atoms and molecules

1.2. THERMAL RADIATION AND PLANCK’S LAW 7

Chemistry IV.
In the following sections of this first chapter, we will introduce the basic concepts and prin-
ciples that underlie the whole field of spectroscopy. As is often the case in science, the major
part of a field can be cast in a few leading concepts from which all the important knowledge
can be derived.

1.2 Thermal radiation and Planck’s law

We consider a cubic cavity of side length L held in equilibrium at a temperature T . The

walls of the cavity emit and absorb radiation because of their thermal excitation. We would
like to know the total energy density as well as the frequency distribution of the radiation
inside the cavity.
Since the cavity is in thermal equilibrium, the emitted power Pe (ν) has to be equal to the
absorbed power Pa (ν) for all frequencies ν. Inside the cavity, the radiation field is thus
stationary and can be written as a superposition of plane waves
  
E(r, t) = Ap exp i(ωp t − kp · r) + compl. conj. (1.1)
p

The electric field strength has to be zero on the walls of the cavity which imposes the boundary
condition

E(x = 0, y, z, t) = E(x = L, y, z, t) = 0 with 0 ≤ y ≤ L and 0 ≤ z ≤ L (1.2)

and similar conditions for the y and z dimensions. Consequently, the electric field takes the
form
E(r, t) = A sin(kx x) sin(ky y) sin(kz z), (1.3)

which imposes the boundary conditions on the wave vector

π π π
kx = n1 , k y = n2 , kz = n3 , (1.4)
L L L
i. e. the cavity length must be an integer multiple of the half-wavelength L = ni λ2i , a situation
that is illustrated in Figure 1.2 in one dimension.
The wave vector of the radiation field can thus be written as

k = π (n1 , n2 , n3 ) (1.5)
L
with positive integers ni (ni ∈ N∗ ). The magnitude of the allowed wave vectors is thus
  π
 
k  = n21 + n22 + n23 (1.6)
L

PCV - Spectroscopy of atoms and molecules

8 CHAPTER 1. INTRODUCTION





Figure 1.2: Allowed electro-magnetic waves in a cavity of length L.
   
   
which can be written in terms of the wavelength λ = 2π/ k , the frequency ν = c k  /2π or
 
 
the angular frequency ω = c k .
The modes with angular frequency between 0 and ω fulfils
ω 2 L2
n21 + n22 + n23 ≤ . (1.7)
c2 π 2
In a Cartesian system with the coordinates (π/L)(n1 , n2 , n3 ) (see Figure 1.3), each triple
(n1 , n2 , n3 ) represents a point in a three-dimensional structure. If the radius is large com-
pared to π/L, i. e. the wavelength is much smaller than the dimensions L of the cavity,
Equation (1.7) represents the interior of a sphere with the radius

ωL
r = n21 + n22 + n23 = . (1.8)
cπ





Figure 1.3: Diagram of the possible (n1 , n2 , n3 ) combinations for occupation numbers of the
modes of an optical cavity. The possible combinations are contained within an octant of a
ωL
sphere of radius r = cπ .

In this limit, the number of modes can be estimated from the volume of the octant of the
sphere shown in Figure 1.3. This volume is
 3
1 4π ωL
Voct = . (1.9)
8 3 cπ

PCV - Spectroscopy of atoms and molecules

1.2. THERMAL RADIATION AND PLANCK’S LAW 9

Since two polarization states are possible for each mode, this leads to the mode number
 
π ωL 3
Nω (ω) = 2Voct = (1.10)
3 cπ
ω
and, expressed in frequencies ν = 2π ,

8πν 3 L3
N (ν) = . (1.11)
3c3

In the following sections, we will often refer to the density of modes per unit volume within
a frequency interval dω or dν, which is obtained by differentiating the number of modes per
volume Nω (ω)/V (here V = L3 is the volume of the cavity) with respect to ω (or with respect
to ν when expressed in frequencies)

1 dNω (ω) ω2
gω (ω) = = 3 2, (1.12)
V dω c π
1 dN (ν) 8πν 2
g(ν) = = 3 . (1.13)
V dν c

Each mode of the cavity described in the last setion is in equilibrium with the matter of the
cavity walls. This matter can be described as classical mechanical oscillators in thermody-
namic equilibrium at temperature T with a total mean energy of kT (see lecture Physical
Chemistry II), where k is the Boltzmann constant (k = 1.3806488(13) × 10−23 J K−1 ). This
implies that the mean energy in each mode is also kT and thus the energy content in the
frequency interval ν to ν+dν of a radiation field in a volume V at a temperature T is given
by
8πν 2
U (ν)dν = g(ν)V kT dν = V kT dν. (1.14)
c3
and the energy density is given by

U (ν) 8πν 2
ρ(ν) = = g(ν)kT = 3 kT. (1.15)
V c

This relation is also known as the Rayleigh-Jeans law and correctly predicts the radiation
density for low frequencies (hν  kT ) and long wavelengths, i. e. in the infrared region. How-
ever, it strongly disagrees with experimental observations in the high-frequency (hν  kT )
or short wavelength region because it predicts that the energy density becomes infinite. This
constituted a major puzzle in the late 1800’s and was called the “UV catastrophy”.
To resolve this discrepancy, Max Planck proposed that each mode of the radiation field can
only emit or absorb energy in integer multiples q of a fundamental energy quantum hν, where
h is Planck’s constant (h = 6.62606957(29) × 10−34 J s). This quantum is nowadays called the

PCV - Spectroscopy of atoms and molecules

10 CHAPTER 1. INTRODUCTION

“photon” and each mode of the cavity can thus only contain an integer number of photons
and thus carry the energy qhν.
The energy E (in J, usually in eV) carried by one photon can be expressed as a function of
its frequency ν (in Hz, i. e. s−1 ), its circular frequency ω (in s−1 ), its wavelength λ (in m,
usually nm) or its wavenumber ν̃ (in m−1 , usually cm−1 ) as

hc
E = hν = ω = = hcν̃. (1.16)
λ

It is very useful to remember approximate conversion factors between these units which are
as follows
1.000 eV ↔ 2.418 × 1014 Hz ↔ 8066 cm−1 . (1.17)

The partitioning of energy into the different possible modes in thermal equilibrium is governed
by the Maxwell-Boltzmann distribution stating that the probability p(q) to find the energy
qhν in a specific mode is  
1 qhν
p(q) = exp − , (1.18)
Z kT
where  
 qhν
Z= exp − (1.19)
q
kT

is the so-called partition function (see Lecture Physical Chemistry VI - Statistical thermody-

namics). The role of Z is to ensure that the sum of all probabilities is one, i. e. p(q) = 1.
q
One can thus define the mean energy per mode as

∞ qhν exp (−qhν/(kT ))
 q
E(ν) = p(q)qhν = , (1.20)
exp (−qhν/(kT ))
q=0 q

which leads to
hν
E(ν) = . (1.21)
exp (hν/(kT )) − 1
The energy density of the radiation field within the frequency interval ν to ν + dν is given
by the product of the mode density g(ν) (Equation (1.13)) per unit volume and the mean
energy E(ν) per mode, which constitutes Planck’s law

8πν 2 hν
ρ(ν)dν = g(ν)E(ν)dν = 3
dν. (1.22)
c exp (hν/(kT )) − 1

This relation characterizes the spectral energy distribution of a radiation field that is in
thermal equilibrium with its environment at a temperature T . Equation (1.22) is represented
graphically in Figure 1.4 for the temperatures T =293 K, 2500 K and 3500 K. This graph

PCV - Spectroscopy of atoms and molecules

1.2. THERMAL RADIATION AND PLANCK’S LAW 11

shows why a solid metallic body at room temperature does not emit any visible radiation,
whereas the same body emits red light when heated to 2500 K and eventually the whole visible
spectrum when heated to 3500 K, which makes it look white. Other examples of radiation
sources that closely follow Planck’s law are the sun, the tungsten wire of a light bulb and
flash lamps.









Figure 1.4: Spectral distribution of the energy density ρ(ν) for different temperatures
(Planck’s law).

Table 1.1 lists the frequencies and wavenumbers corresponding to the maximum of the energy
density ρ(ν) for different temperatures T . The cosmic background radiation (T =2.7 K) has
its maximum in the far-infrared. The thermal emission of the earth’s surface peaks in the
mid-infrared, close to the bending frequency of CO2 (667 cm−1 ), explaining why this molecule
is the main carrier of global warming. The maximum of the sun’s emission spectrum lies close
to the visible range.

T /K νmax /Hz ν̃max /cm−1 λ(νmax )/μm region

3 1.76 × 1011 5.88 1700 cosmic background
300 1.76 × 1013 588 17 surface of the earth
6000 3.53 × 1014 11768 0.85 surface of the sun

Table 1.1: Frequencies νmax , wavenumbers ν̃max and wavelengths λ(νmax ) describing the
maximum of the energy density distribution ρ(ν) for different temperatures T .

Other sources of radiation, like gas-discharge lamps (neon tubes) and lasers do not follow

PCV - Spectroscopy of atoms and molecules

12 CHAPTER 1. INTRODUCTION

Planck’s law and are also known as “non-thermal” radiation sources.

1.3 Absorption and emission of radiation

This section discusses the absorption and emission of thermal radiation by a two-level system
in the framework of a kinetic model, introduced by A. Einstein (A. Einstein, Physik Z. 18,
121 (1917)). We consider a system with two states, labelled 1 and 2, of energies E1 and E2
with E1 < E2 , in contact with thermal radiation. The system is initially in the lower state
1. The possible processes resulting from the interaction of the system with radiation are
illustrated in Figure 1.5. The system can be excited to state 2 by absorption of radiation
at frequency
E2 − E1
ν12 = , (1.23)
h
a transition that will be designated with the notation 2 ← 1. In a kinetic formalism, one can
thus write down the rates of change of the populations Ni in the two states as
(abs) (abs)
dN1 dN2
− = = B12 ρ(ν12 )N1 , (1.24)
dt dt

where ρ(ν12 ) is the photon density and B12 depends on the properties of the system, in
particular the wave functions of the states 1 and 2, but not on the temperature.

 




 
   
 

Figure 1.5: Interaction of a two-level system with a radiation field. The black dot labels the
state of the system prior to the radiative process.

The excited system can spontaneously decay from state 2 to state 1 under emission of light
at the same frequency ν12 defined above. This process is written 2 → 1 and can again be
described kinetically
(sp. em.) (sp. em.)
dN1 dN2
=− = A21 N2 . (1.25)
dt dt

PCV - Spectroscopy of atoms and molecules

1.3. ABSORPTION AND EMISSION OF RADIATION 13

This process is called spontaneous emission because it is independent of the radiation field.
A21 depends on the properties of the system but is temperature-independent. Spontaneous
emission is characterized by an arbitrary direction of emission.
Finally, the radiation field can also induce a transition from state 2 to state 1,
(stim. em.) (stim. em.)
dN1 dN2
=− = B21 ρ(ν12 )N2 . (1.26)
dt dt

This process is called stimulated emission. An important property of stimulated emission

is that it occurs along the direction imposed be the incoming radiation. In the following
paragraphs, the relation between the three coefficients B12 , B21 and A21 will be derived.
We now consider the system with the two states, 1 and 2, to be in equilibrium with a thermal
radiation field. Once the thermal equilibrium is established, the rate of absorption and the
total rate of emission (spontaneous and stimulated) must balance each other, a situation that
is described by the following equation:
 
dN1 (eq) (eq) (eq)
= −B12 ρ(ν12 )N1 + A21 N2 + B21 ρ(ν12 )N2 = 0 (1.27)
dt eq

or
(eq) (eq)
B12 ρ(ν12 )N1 = [A21 + B21 ρ(ν12 )] N2 . (1.28)

Since the system is simultaneously in thermal equilibrium with the environment at temper-
(eq) (eq)
ature T , the ratio of populations N1 and N2 has to fulfill
(eq)    
N2 E2 − E1 hν12
(eq)
= exp − = exp − , (1.29)
N1 kT kT

giving
A21 /B21
ρ(ν12 ) =

 . (1.30)
B12
B21 exp hν 12
kT − 1

Since Equations (1.30) and (1.22) must be simultaneously fulfilled at the same temperature,
equating the coefficients in the two equations leads to
3
8πhν12
B12 = B21 and A21 = B21 . (1.31)
c3

These two relations have many important fundamental implications. First, the Einstein
coefficients for absorption and stimulated emission are equal. This result will become obvious
in the next paragraph and in later chapters where we will calculate these quantities from first
3 . This means that the
principles. Second, the rate of spontaneous emission scales with ν12
lifetime of excited states is much longer in the case of pure vibrational excitation (typically

PCV - Spectroscopy of atoms and molecules

14 CHAPTER 1. INTRODUCTION

milliseconds) than in the case of pure electronic excitation (typically nanoseconds). This
scaling law also explains why it is much more difficult to build lasers in the ultraviolet to X-
ray wavelength range than in the visible or infrared range. The rate of spontaneous emission
is so large that it becomes difficult or impossible to maintain a population inversion as will
be discussed in the next sections.
Since all three Einstein coefficients are temperature-independent, one can choose one of them
(e. g. B12 ) to express the others. In later sections, we will show that B12 can be expressed
within the dipole approximation, in terms of the wave functions of the two states

8π 3  
 ˆ 2
B12 =  2| μ
 |1  , (1.32)
3h2 (4πε0 )

where |1 = Ψ1 represents the wave function of state 1, 2| = Ψ2∗ represents the complex
ˆ is the dipole operator and ε0 is the vacuum
conjugate of the wave function of state 2, μ
permeability (ε0 = 8.854187817 × 10−12 F m−1 ).
Finally, the relation (1.30) can also be used to calculate the average number of photons that
occupy each mode of the electromagnetic field at a certain temperature T . Each mode can
only be occupied by an integer number of photons, but the average number of photons per
mode is given by:
E(ν) 1
n(ν, T ) = = . (1.33)
hν exp (hν/(kT )) − 1
At room temperature, in the visible range (λ=600 nm), one finds exp(hνvis /(kTamb )) ≈
4 × 1035 and n(νvis , Tamb ) ≈ 2.5 × 10−36 . Consequently, the spontaneous emission per mode
exceeds by far the thermally stimulated emission. However, it is possible to concentrate
the radiation energy into a small number of modes, e. g. in a laser, in which case the oc-
cupation number n(ν, T ) can become much larger than one. In the high-temperature limit
exp (hν/(kT )) → 1 and the number of photons per mode can also become very large.

1.4 Measurable quantities in spectroscopy

The goal of this section is to establish a link between the quantities that are typically measured
in frequency-resolved experiments and the Einstein coefficients. The simplest spectroscopic
experiment consists in monitoring the absorption of a monochromatic beam of light by a
sample of atoms or molecules as illustrated in Figure 1.6. The particles (atoms or molecules),
at density C are initially all in the lower state, labelled “1”, and are contained in a transparent

PCV - Spectroscopy of atoms and molecules

1.4. MEASURABLE QUANTITIES IN SPECTROSCOPY 15

glass cell. They can absorb the radiation of frequency ν12 and undergo a transition to an
upper state labelled “2”.




  


 


 




Figure 1.6: Experimental scheme of an absorption measurement

The intensity I of the beam of light can be expressed as the flow of photons per unit surface
and unit time
I = cCγ , (1.34)

where c is the speed of light and Cγ is the photon concentration. We note that this differs
from the traditional definition of intensity IE which is however recovered by multiplying I
with the photon energy
IE = Ihν. (1.35)

Absorption of light across the cell of length can be expressed through the differential decrease
of the intensity dI over the distance dx. dI is proportional to the intensity at position x, the
concentration of molecules C and the length dx

− dI  (x) = σ12 I  (x)Cdx, (1.36)

where σ12 has the dimension of a surface and is called the absorption cross section. At
the position of the detector the measured intensity is
 I   
dI
− 
= σ12 Cdx, (1.37)
I0 I 0

which leads to the Lambert-Beer law

 
I0
ln = σ12 C . (1.38)
I

Another form of this law is

 
I0
A10 = log10 = ε12 Cmol , (1.39)
I

PCV - Spectroscopy of atoms and molecules

16 CHAPTER 1. INTRODUCTION

where A10 is the decadic absorbance, ε12 is the molar extinction coefficient in m2 mol−1 and
Cmol is the molar concentration in mol m−3 .
Absorption can also be described as a bimolecular reaction between particles of concentration
C in state i and photons γ. One can define a bimolecular rate coefficient for the decrease of
the photon concentration Cγ
dCγ
− = kbi Cγ C. (1.40)
dt
Using the photon speed c, one can transform from position to time

dx
c= (1.41)
dt
dI(x) dCγ c dCγ dx dCγ
− =− =− =− . (1.42)
dx dx dx dt dt
The absorbing species can be considered to be at rest relative to the photons, which gives

dCγ
− = kbi Cγ C = σ12 cCγ C. (1.43)
dt

The bimolecular rate constant is thus

kbi = σ12 c, (1.44)

which leads to the definition of a pseudo first-order rate constant (assuming that Cγ is
constant)
kfirst = σ12 cCγ . (1.45)

At this point, we would like to compare kfirst with the pseudo-first order rate coefficients
defined on the basis of the Einstein coefficients K12 = B12 ρ(ν12 ) and K21 = B21 ρ(ν12 ). Just
as in the case of a damped classical oscillator, a transition between two quantum mechanical
states has a finite linewidth that results from the lifetime of the excited state. An experimen-
tally measured line is subject to additional broadening mechanisms, like Doppler broadening
and power broadening that will be discussed in exercises or in later sections. This requires
the consideration of the photon density (or concentration) in the frequency interval between
ν and ν + dν
dCγ = Cγ (ν)dν (1.46)

with
 ∞
Cγ = Cγ (ν)dν. (1.47)
0

PCV - Spectroscopy of atoms and molecules

1.4. MEASURABLE QUANTITIES IN SPECTROSCOPY 17

Considering their spectral dependence, the rate constants for absorption K12 and for stim-
ulated emission K21 introduced previously are obtained by integrating over all frequencies
contributing to the transition 2 ↔ 1
 ∞  ∞
K12 = K12 (ν)dν = cσ12 (ν)Cγ (ν)dν. (1.48)
0 0

We note that σ12 (ν) depends on the frequency ν (it peaks strongly at the resonance fre-
quency), whereas Cγ (ν) depends on the light source. Therefore, it is sufficient to evaluate the
integrals in the vicinity of the resonance frequency, which is expressed by the substitution
 ∞ 
→ . (1.49)
0 line
ρ(ν)
Cγ (ν) is related to the energy density ρ(ν) through Cγ (ν) = hν and therefore,

c σ12 (ν)
K12 = ρ(ν)dν. (1.50)
h line ν
When the photon density of the excitation source varies little with its frequency (so-called
“white-light” excitation), ρ(ν) ≈ ρ(ν12 ) and

c σ12 (ν)
K12 ≈ ρ(ν12 ) dν. (1.51)
h line ν

Figure 1.7: Line strength S12 and integrated absorption cross section G12 of an absorption
line centered at νmax .

With the definition of the integrated absorption cross section G12 (see Figure 1.7)

σ12 (ν)
G12 = dν, (1.52)
line ν
one obtains
c
K12 ≈ ρ(ν12 )G12 (1.53)
h
c
and B12 ≈ G12 .
h

PCV - Spectroscopy of atoms and molecules

18 CHAPTER 1. INTRODUCTION

If the photon density of the light source varies strongly with its frequency, i. e. ρ(ν) is not
approximately constant, K12 must be calculated using Equation (1.50).
The line strength is defined as

S12 = σ12 (ν)dν, (1.54)
line

as illustrated in Figure 1.7. Using this definition, the integrated absorption cross section can
be approximated as
1
G12 ≈ max S12 (1.55)
ν12
and
c2
G12 ≈ 3 A21 . (1.56)
8πν12
We recall that the Einstein coefficient B12 can be expressed as a function of the transition
ˆ|1
moment M12 = 2|μ
8π 3
B12 = |M12 |2 (1.57)
3h2 (4πε 0)

and for the integrated absorption cross section

8π 3
G12 ≈ |M12 |2 . (1.58)
3hc(4πε0 )

1.5 Population inversion and laser operation

A beam of light of frequency ν12 incident on a sample of atoms or molecules in the excited
state 2 (i. e. C1 = 0) will be amplified according to the Lambert-Beer law

I = I0 exp (σ12 C2 ) . (1.59)

More generally, light will undergo net absorption if C1 > C2 but net amplification if
C1 < C2 . The net gain can be expressed as

I
= exp [σ12 (C2 − C1 ) ] . (1.60)
I0

The situation C1 < C2 is called population inversion and underlies the operation of the
laser. Laser stands for Light Amplification by Stimulated Emission of Radiation. Histori-
cally, the laser was preceded by the maser, an amplification device operating in the microwave
region (hence maser), based on population inversion in ammonia molecules.
Amplification of light alone is not sufficient for the construction of a laser, but one also needs
a feedback structure that will make light pass through the gain medium several times, as

PCV - Spectroscopy of atoms and molecules

1.5. POPULATION INVERSION AND LASER OPERATION 19

illustrated in Figure 1.8. This is most commonly realized using total reflection on a mirror
on one end of the cavity and a partially reflective mirror on the other end. The major part of
the radiation is reflected back into the cavity, where it is amplified further, whereas a small
part is allowed to escape. This fraction forms the laser beam that is characterized by a very
low divergence and a high degree of temporal and spatial coherence.

 

 








 

 

Figure 1.8: Experimental setup of a laser

Many different materials and processes can be used in the construction of a laser. Based on
the operation principle, they can be grouped into three categories illustrated in Figure 1.9.
The two-level laser relies exclusively on the principles of absorption and stimulated emission
discussed in the last sections. An example of this type of device is the original ammonia
maser. The problem with this type of laser is that it tends to turn itself off because the am-
plification is suppressed when strong stimulated emission occurs. In the ammonia maser, this
problem is solved by pumping away the ground state molecules, but this approach is limited
to gas lasers. In general, a two-level laser is thus very unlikely to work. The 3-level laser
avoids this problem by adding one step of radiationless transition into the cycle. The excita-
tion process which is also called “pumping” is spectrally distinct from the emitted radiation.
The radiationless transition, shown as a wavy line, can either serve to populate the upper
level or to depopulate the lower level of the laser transition. In both cases, strong stimulated
emission will not counteract the effect of pumping which allows the laser to run with much
higher yield. An example of the 3-level laser is the ruby laser, the first ever demonstrated
by Theodore Maiman in 1960. Pumping is achieved by optical excitation of the ruby crystal
with flashlamps and lasing occurs at 694.3 nm. Finally, in the 4-level laser, the lasing tran-
sition is decoupled form the pump transition through two steps of non-radiative transitions.
Examples of this kind include most practically useful lasers, including the high-power CO2
lasers and the titanium-sapphire (Ti:Sa) lasers. The CO2 lasers are pumped by high-voltage

PCV - Spectroscopy of atoms and molecules

20 CHAPTER 1. INTRODUCTION

discharge, whereas Ti:Sa lasers are optically pumped by another laser.







Figure 1.9: Operation principles of the three most common types of lasers

The following paragraphs will present a derivation of the equations underlying laser oper-
ation. In the preceding sections, we have calculated the occupation numbers of radiation
modes in thermal equilibrium. We will now calculate these numbers in the non-equilibrium
conditions associated with population inversion. The rate of photon production in an atomic
or molecular sample with population numbers N1 and N2 in states 1 and 2 is

dNγ
= (N2 − N1 ) B21 ρ (ν12 ) + A21 N2 , (1.61)
dt

where ρ (ν12 ) is the radiation density. Photons are produced within a bandwidth Δν around
the resonance frequency ν12
Nγ = g(ν12 )Δνn(ν12 ), (1.62)

PCV - Spectroscopy of atoms and molecules

1.5. POPULATION INVERSION AND LASER OPERATION 21

where g(ν12 ) is the mode density as defined in Equation (1.13) and n(ν12 ) is the number of
photons per mode. The total number of photons Nγ and n(ν) are time-dependent, whereas
g(ν) and Δν are time-independent. Using Equation (1.61) one thus obtains

dn(ν12 ) B21 ρ (ν12 ) A21

= (N2 − N1 ) + N2 . (1.63)
dt g(ν12 )Δν g(ν12 )Δν

Since V ρ (ν12 ) = g(ν12 )hν12 n(ν12 ), using g(ν12 ) from Equation (1.13), it follows that

8πV 2
V ρ (ν12 ) = ν hν12 n(ν12 ) (1.64)
c3 12

and with the expressions for A21 and B21 from Equation (1.31), we obtain

A21 c3 
B21 ρ (ν12 ) = 3 ρ (ν12 ) = A21 n(ν12 ). (1.65)
8πhν12

With the definition

A21
W = , (1.66)
g(ν12 )Δν
we obtain the laser equation

dn(ν12 )
= (N2 − N1 ) W n(ν12 , t) + W N2 , (1.67)
dt

where (N2 − N1 )W represents the gain. When the pump process is fast enough, one can
assume that N1 and N2 reach a stationary regime, which allows one to solve the differential
equation (1.67) with the boundary condition n(ν12 ) = 0 at t = 0:

N2
n(ν12 , t) = (exp (W (N2 − N1 )t) − 1) . (1.68)
N2 − N1

In the following, we will analyze the regions of stability of the laser equation.

• if N2 < N1 , n(ν12 , t → ∞) = − N2N−N

2
1
>0
The number of photons per mode is fixed, which corresponds to a stable condition.
However, this situation does not corresponds to net amplification of light.

• if N2 ≈ N1 , a series expansion of the exponential term gives exp (W (N2 − N1 ) t) ≈

1 + W (N2 − N1 ) t + ..., i. e. n(ν12 , t) ≈ N2 W t, corresponding to a linear growth of the
photon population with time.

N2
• if N2 > N1 , n(ν12 , t → ∞) = N2 −N1 exp (W (N2 − N1 ) t), the photon number diverges
with time and even the fastest pump process can not maintain the population difference.

PCV - Spectroscopy of atoms and molecules

22 CHAPTER 1. INTRODUCTION

Losses from the cavity by, e. g. scattering or a partially transmitting mirror lead to an
additional term in the laser equation

dn(ν12 )
= (N2 − N1 ) W n(ν12 ) + W N2 − k1 n(ν12 ), (1.69)
dt

where t1 = 1/k1 is the decay time of the laser resonator and k1 n(ν12 ) represents the losses.
For large values of n(ν12 , t) the contribution from spontaneous emission is often negligible,
leading to the condition of laser operation

dn(ν12 )
> 0 or (N2 − N1 ) W > k1 , (1.70)
dt

i. e. the gain has to be larger than the losses, which provides the condition of operation for
a laser
N2 − N1 c3 A21
> k1 . (1.71)
V 8πν 2 Δν
To summarize, laser operation needs population inversion and a gain larger than the losses.

PCV - Spectroscopy of atoms and molecules

1.6. INTERACTION OF LIGHT AND MATTER 23

1.6 Interaction of light and matter

An electromagnetic wave can be represented by a number of harmonic oscillators correspond-

ing to the modes of the field. In the case of laser radiation, the number of quanta per field
mode is extremely large, on the order of 1010 − 1020 . In such cases, electromagnetic fields can
be treated classically for all practical purposes and are well described by Maxwell’s equations.
In this lecture, atoms and molecules will be treated quantum mechanically, but the radiation
field will be treated classically. The interaction of light with particles (atom or molecules) is
greatly simplified when the wavelength of the radiation λ is much larger then the extension
d of the particle
λ  d, (1.72)

which is typically the case in the ultraviolet to infrared (100 nm−100 μm) range of the elec-
tromagnetic spectrum. However, this is no longer true for X-rays and γ-rays.
The electrical dipole interaction of a monochromatic linearly polarized wave of circular fre-
quency ω with a system of charged particles is


V̂el (t) = − qiri E (z) cos(ωt + φ), (1.73)
0
i

 (z) is the peak electric field and ri is the position of the particle i of charge qi .
where E 0

Writing the electric dipole moment μ  el = i qiri

(z)
 cos(ωt + φ),
V̂el (t) = −
μel E 0 (1.74)

and by analogy for the interaction with the magnetic field

V̂mag (t) = −  (x) cos(ωt + φ).

μmag B (1.75)
0

In the optical domain, the electric dipole interaction dominates. However, important meth-
ods, like nuclear magnetic resonance (NMR) and electron paramagnetic resonance (EPR)
rely on the magnetic dipole interaction (see lecture Physical Chemistry IV).

1.7 Coherent excitation of a two-level system

Many phenomena in nature are time-dependent and are described by the time-dependent
Schrödinger equation that has been discussed in the lecture Physical Chemistry III.

PCV - Spectroscopy of atoms and molecules

24 CHAPTER 1. INTRODUCTION

Reminder: The time-dependent Schrödinger equation of a n-particle system can

be written as follows:

∂Ψ (r1 , r2 , ..., rn , t)

i = ĤΨ (r1 , r2 , ..., rn , t), (1.76)
∂t

with coordinates rj of particle j. The general solution of this equation is

Ψ (t) = Û (t, t0 )Ψ (t0 ), (1.77)

where Ψ (t0 ) describes the initial state of the system.

The time-evolution operator Û is obtained from

∂ Û
i = Ĥ Û (t, t0 ). (1.78)
∂t

(i) Stationary (time-independent) states can be written in the form

 
iEt
Ψ (r, t) = Ψ (r) exp − (1.79)


where E is the energy of the state described by Ψ (r) and r is used collectively to represent
all coordinates (r1 , r2 , ..., rn ). Inserting Equation (1.79) into Equation (1.76) using the
relation    
d iEt iE iEt
exp − = − exp − (1.80)
dt   
gives
EΨ (r) = ĤΨ (r), (1.81)

which is the time-independent Schrödinger equation.

Using the index k for the eigenfunction Ψk with energy Ek , one can thus write all
stationary solutions of the time-dependent Schrödinger equation in the form
 
iEk t
Ψk (r, t) = Ψk (r) exp − . (1.82)


The probability density associated with such a state is given by

Ψk (r, t)Ψk∗ (r, t) = |Ψk (r, t)|2 = |Ψk (r)|2 , (1.83)

which is also time independent.

(ii) General time-dependent states: In this case, it is useful to remember the superpo-
sition principle. If Ψ1 (r, t) and Ψ2 (r, t) are two dynamical states of a system, then

PCV - Spectroscopy of atoms and molecules

1.7. COHERENT EXCITATION OF A TWO-LEVEL SYSTEM 25

Ψ (r, t) = c1 Ψ1 + c2 Ψ2 and in general, with complex ck ,

  
iEk t
Ψ (r, t) = ck Ψk (r, t) exp − (1.84)

k

are also possible dynamical states of the system. In this case Ψ (r, t)Ψ ∗ (r, t) is time
dependent. The probability of measuring the energy Ek is ck c∗k = |ck |2 . In a time-
dependent state, the energy itself is not well-defined, only its probability distribution.

In the following, we discuss the quantum mechanical description of the excitation of a N -level
system. The coupling of a quantum mechanical system with radiation can be described by
the time-dependent Schrödinger equation
∂Ψ
i = Ĥ(t)Ψ , (1.85)
∂t
where
Ĥ(t) = Ĥmol − μ  (z) cos(ωt + φ).
 el E (1.86)
0

Because Ĥ is time-dependent, the solution of the Schrödinger equation is often not available
in closed form. It is often convenient to expand Ψ (t) in eigenfunctions of the system, e. g.
into molecular eigenfunctions
Ĥmol φk = Ek φk (1.87)

i. e.

Ψ (t) = bk (t)φk (r1 , ..., rn ), (1.88)
k
where bk (t) are time-dependent coefficients. Inserting this ansatz into the Equation (1.85)
∂  
i bk (t)φk = Ĥ bk (t)φk , (1.89)
∂t
k k

multiplying from the left with φ∗j and integrating over all coordinates ri one obtains
dbj (t) 
i = Hjk bk (t). (1.90)
dt
k

To obtain this result, we have used the following properties:

1. the orthonormality of the molecular eigenfunctions

  
... φ∗j φk dr1 ...drn = φj |φk  = Δjk (1.91)

2. the notation for matrix elements of the Hamiltonian

  
Hjk = φj |Ĥ|φk  = ... φ∗j Ĥφk dr1 ...drn . (1.92)

PCV - Spectroscopy of atoms and molecules

26 CHAPTER 1. INTRODUCTION

This system of equations can be written as a matrix equation

d
i b(t) = Hb(t) (1.93)
dt

where b(t)=(b1 , b2 , ..., bN )T is a vector of coefficients and H={Hjk } is the matrix represen-
tation of the Hamiltonian Ĥ.
We now discuss the special case of Rabi oscillations in a strongly driven two-level
system. We again consider the two-level system described previously, which now interacts
with a coherent electromagnetic field E = E0 cos(ωt + φ). In this case, the quantities in
Equation (1.93) are H11 = φ1 |Ĥmol |φ1  = E1 = ω1 and H22 = E2 = ω2 . The coupling
matrix element for electric dipole coupling is

H12 = φ1 |V̂el (t)|φ2  (1.94)

and we define
ˆz |φ2  (z)
φ1 |μ
V12 = − E0 . (1.95)

ˆz is real, the case of real φ1 and φ2 leads to V12 = V21 = V . The time-dependent
Since μ
coefficients of the total wave function Ψ (t) = b1 (t)φ1 + b2 (t)φ2 must fulfill the system of
coupled differential equations

db1
i = ω1 b1 + V cos(ωt + φ)b2 (1.96)
dt
db2
i = V cos(ωt + φ)b1 + ω2 b2 . (1.97)
dt

This system of coupled differential equations has no simple closed-form solution but an ap-
proximate solution has been given by Rabi (Phys. Rev. 51, 652 (1937)).
For the case of near-resonant excitation with the detuning Δ = (ω2 − ω1 ) − ω = ω12 − ω
the population of state 2 is
   2
2 V2 t 2 2
p2 (t) = |b2 (t)| = 2 sin V +Δ = 1 − p1 (t), (1.98)
V + Δ2 2

i. e. the population undergoes periodic oscillations with the period

2π
T =√ . (1.99)
V + Δ2
2

Two cases of Rabi oscillations are illustrated in Figure 1.10. On resonance, i. e. for Δ = 0
and ω = ω2 − ω1 , T = 2π/V . In this case p2 (t) oscillates between 0 and 1. Off resonance, the

PCV - Spectroscopy of atoms and molecules

1.8. TIME- AND FREQUENCY-RESOLVED SPECTROSCOPIES 27

  
  






Figure 1.10: Temporal evolution of the excited state population p2 (t) of a two-level system
excited by radiation of circular frequency ω, detuned from resonance by Δ = ω12 − ω. The
time is given in units of the Rabi period 2π/V .

oscillation is faster and p2 (t) never reaches 1. The time-averaged excitation fraction amounts
to
1 V2
p2 (ω)t = , (1.100)
2 V 2 + Δ2
i. e. a Lorentz function with fwhm (Full Width at Half Maximum) of 2V and also corresponds
to the mean absorbed energy for excitation at circular frequency ω. Since V 2 ∝ E02 ∝ IE ,
this behavior is also known as intensity- or power broadening of spectral transitions that is
frequently observed in laser-based spectroscopic experiments.

1.8 Time- and frequency-resolved spectroscopies

Two fundamentally different but complementary approaches to spectroscopy exist. The first
and most widely used is frequency-domain spectroscopy. It relies on tunable sources of ra-
diation that are designed to be as narrow as possible in frequency. These sources are used
to measure energy intervals in atoms or molecules to the highest possible precision as illus-
trated in Figure 1.11a. The measured energy intervals are then used to construct or refine
the Hamiltonian Ĥmol of the system. Once Ĥmol is known to the desired level of accuracy,
all properties of the system, including time-dependent properties can be predicted. However,
the complete determination of Ĥmol from spectroscopic measurements becomes increasingly
difficult for larger molecules.
Time-domain spectroscopy relies on ultrashort laser pulses which, by virtue of the uncertainty
relation, have a very broad spectrum. The large coherent bandwidth of the excitation pulse

PCV - Spectroscopy of atoms and molecules

28 CHAPTER 1. INTRODUCTION


 
    
 






Figure 1.11: Principles of frequency-domain (a) and time-domain (b) spectroscopies. In

frequency-domain spectroscopy, a narrow-band light source is tuned across the transitions of
the molecule under study to determine the energy intervals. In time-domain spectroscopy, a
short pulse with a broad spectrum is used to create a coherent superposition of several levels
and the ensuing dynamics is studied.

is used to prepare a coherent superposition of molecular eigenstates whose time evolution

is studied (see Figure 1.11b). This evolution provides information on Ĥmol that is comple-
mentary to the frequency-domain approach and may be simpler to interpret. An important
example is the analysis of the mechanism of photostability of DNA which has been studied
with time-domain measurements (see e. g. H. Satzger et al., Proc. Nat. Acad. Sci. USA,
103 10196 (2006)).
In the following sections, two simple quantum mechanical systems will be discussed to
illustrate frequency-domain and time-domain measurements.

1.8.1 The harmonic oscillator

Reminder: we consider a one-dimensional quantum mechanical system of two masses
m1 and m2 held together by a potential V (r) = 12 kr2 that is a quadratic function of their
distance r. The Hamiltonian of this system can be written

2 ∂ 2 1
Ĥ = − 2
+ kr2 (1.101)
2μ ∂r 2

where μ = m1 m2 /(m1 + m2 ) is the reduced mass of the system and k is a force constant

PCV - Spectroscopy of atoms and molecules

1.8. TIME- AND FREQUENCY-RESOLVED SPECTROSCOPIES 29

characterizing the potential. The eigenvalues and eigenfunctions of this Hamiltonian are
 
Ev 1
= ωe v + (1.102)
hc 2
1 1 2
φv (r) = √ Hv (r)e− 2 r , (1.103)
πv!2 v

where v = 0, 1, 2, ... is the vibrational quantum number, ωe is the vibrational frequency

expressed in cm−1 and Hv (r) are the Hermite polynomials

H0 (r) = 1 (1.104)

H1 (r) = 2r (1.105)

H2 (r) = 4r2 − 2 (1.106)

etc.

Frequency-resolved spectroscopy measures the energy intervals between E0 and E1 , E1 and

E2 etc. Pure vibrational transitions in a harmonic oscillator can only be observed if Δv = ±1
and if the molecular dipole moment varies with r. In time-resolved spectroscopy, one can
use an ultrashort laser pulse to prepare a coherent superposition of states φ0 and φ1 (through
stimulated Raman scattering) and study its time evolution. We consider the coherent super-
position, which is also called a wave packet,
1  iE0 t iE1 t

Ψ (t) = √ φ 0 e−  + φ 1 e− 
2
1 − iE0 t  i(E1 −E0 )t

= √ e  φ 0 + φ 1 e−  . (1.107)
2
For t0 = 0, |Ψ (t0 = 0)|2 = 1 2
2 (φ0 + φ1 ) .
(E1 −E0 )t1
For t1 = π
(E1 −E0 ) , i. e.  = π, |Ψ (t1 )|2 = 1
2 (φ0 − φ1 )2 .
2π 2 2
For t2 = (E1 −E0 ) , |Ψ (t2 )| = |Ψ (t0 )| .

The period of oscillation of Ψ (t) is thus

h
T = , (1.108)
ΔE
which is a universal relation valid for all wave packets consisting of two levels.
When many levels of the harmonic oscillator are coherently populated, the wave packet still
undergoes a periodic motion. In the case of strong resonant excitation with monochromatic
light E = E0 cos(ωt), the time evolution of the wave packet is given by
1 2
|Ψ (q, t)|2 = √ e−(q−q0 (t)) (1.109)
π

PCV - Spectroscopy of atoms and molecules

30 CHAPTER 1. INTRODUCTION

as shown in the work of Marquardt and Quack (J. Chem. Phys. 90, 6320 (1989)) where q0 (t)
represents the center of the wave packet and is given by
 
V01 sin(ωt)
q0 (t) = − √ , (1.110)
2 ω

where V01 = − μ01 E0 .

Equation (1.109) shows that a quantum mechanical wave packet consisting of a large number
of levels can behave in an essentially classical manner. This fact shows how classical and
quantum mechanical systems are related.

1.8.2 The rigid rotor

Reminder: the Hamiltonian for a linear rigid rotor in three dimensions is

Jˆ2
Ĥrot = , (1.111)
2I

where Jˆ is the angular momentum operator (see Lecture Physical Chemistry III) and I
is the moment of inertia of the rotor. The eigenfunctions ΨJ,M fulfill the Schrödinger
equation

Ĥrot ΨJ,M = EJ ΨJ,M (1.112)

1 ˆ2 2
J ΨJ,M = J(J + 1)ΨJ,M
2I 2I
2 2
with the eigenvalues EJ = 2I J(J + 1) = hcBJ(J + 1), where B = 2hcI is the rotational
constant of the rotor in cm−1 .

A frequency domain measurement measures the interval EJ  − EJ  between the initial state
J  and the final state J  . We will show in later chapters that pure rotational transitions
can only be observed if ΔJ = J  − J  = 0, ±1 and if the molecule possesses a permanent
dipole moment. In a time-domain measurement, one could use a linearly polarized ultrashort
intense laser pulse to prepare a coherent superposition of rotational levels (through repeated
stimulated Raman scattering). Each level |J  , M   that was populated before the interaction
with the laser pulse contributes a subset of coherently populated levels |J  , M  

Ψ= aJ  ,J  |J  , M  = M  , (1.113)
J

where the aJ  ,J  are obtained by solving the time-dependent Schrödinger equation for the
interaction of the laser field with the molecule. A commonly used observable is the ”degree

PCV - Spectroscopy of atoms and molecules

1.8. TIME- AND FREQUENCY-RESOLVED SPECTROSCOPIES 31

of axis alignment” cos2 θ given by

 
cos2 θ(t) = P (J  ) a∗J,J  (t)aJ  ,J  (t)J, M | cos2 θ|J  , M  , (1.114)
J  J,J 

where P (J  ) is the initial thermal population of the rotational levels and the inner sum runs
over all levels populated by the ultrashort laser pulse.
Figure 1.12 shows cos2 θ(t) for N2 molecules exposed to a 100-fs laser pulse. The interaction
with the laser pulse induces a “prompt alignment” followed by a fast decay towards near-
random alignment (random alignment is characterized by cos2 (θ) = 1/3) and revivals at
h h 1
every half of the classical rotation period T = ΔE = hc(2B) = 2cB = 8.3 ps.








  











 


  


Figure 1.12: Evolution of the degree of axis alignment cos2 θ of N2 molecules with an
initial rotation temperature of T = 30 K that have been exposed to a 100 fs pulse of intensity
3×1013 W/cm2 .

PCV - Spectroscopy of atoms and molecules