Molecular Photophysics and

Spectroscopy (Second Edition)

 Molecular Photophysics and
Spectroscopy (Second Edition)
David L Andrews
School of Chemistry, University of East Anglia, Norwich, UK

Robert H Lipson
Department of Chemistry, University of Victoria, Victoria, BC, Canada

IOP Publishing, Bristol, UK

ª IOP Publishing Ltd 2021

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system
or transmitted in any form or by any means, electronic, mechanical, photocopying, recording
or otherwise, without the prior permission of the publisher, or as expressly permitted by law or
under terms agreed with the appropriate rights organization. Multiple copying is permitted in
accordance with the terms of licences issued by the Copyright Licensing Agency, the Copyright
Clearance Centre and other reproduction rights organizations.

Certain images in this publication have been obtained by the author(s) from the Wikipedia/
Wikimedia website, where they were made available under a Creative Commons licence or stated
to be in the public domain. Please see individual figure captions in this publication for details. To
the extent that the law allows, IOP Publishing [and full partner name (if applicable)] disclaim any
liability that any person may suffer as a result of accessing, using or forwarding the image(s). Any
reuse rights should be checked and permission should be sought if necessary from Wikipedia/
Wikimedia and/or the copyright owner (as appropriate) before using or forwarding the image(s).

Permission to make use of IOP Publishing content other than as set out above may be sought
at [email protected].

David L Andrews and Robert H Lipson have asserted their right to be identified as the authors of
this work in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988.

ISBN 978-0-7503-3683-3 (ebook)

ISBN 978-0-7503-3681-9 (print)
ISBN 978-0-7503-3684-0 (myPrint)
ISBN 978-0-7503-3682-6 (mobi)

DOI 10.1088/978-0-7503-3683-3

Version: 20210701

IOP ebooks

British Library Cataloguing-in-Publication Data: A catalogue record for this book is available
from the British Library.

Published by IOP Publishing, wholly owned by The Institute of Physics, London

IOP Publishing, Temple Circus, Temple Way, Bristol, BS1 6HG, UK

US Office: IOP Publishing, Inc., 190 North Independence Mall West, Suite 601, Philadelphia,
PA 19106, USA
 Dedicated to Karen Ann
Ever held in loving memory
 Contents

Preface to the second edition x

Preface to the first edition xi
Author biography xii
List of symbols xiii

1 Introduction to molecular spectroscopy and photophysics 1-1

1.1 The distinctiveness of molecular interactions with light 1-1
1.2 Properties of molecules and their spectra 1-2
1.3 The electromagnetic spectrum 1-4
1.4 Photon properties: polarisation and spin 1-5

2 Atoms and molecules: developing principles 2-1

of electronic structure
2.1 Review of atomic orbitals, angular momentum and electron spin 2-1
2.2 Multielectron atoms 2-6
2.3 Diatomic molecules 2-9
2.4 Orbitals and bonding in molecules 2-10

3 Polyatomic molecules: orbitals, symmetry and group theory 3-1

3.1 Symmetry elements 3-1
3.2 Point groups and operations: Schoenflies notation 3-3
3.3 Matrix representations and character tables 3-4

4 Electronic and nuclear energy levels in molecules 4-1

4.1 The separation of electronic and nuclear motions 4-1
4.2 Types of nuclear motions and degrees of freedom 4-2
4.3 How far do the atoms move? 4-6

5 Small molecule rotational energy levels and spectra 5-1

5.1 Diatomic and linear polyatomic molecules 5-1
5.2 Nuclear spin effects 5-8
5.3 Interpreting rotational spectra 5-9

vii
 Molecular Photophysics and Spectroscopy (Second Edition)

5.4 Centrifugal distortion 5-10

5.5 Non-linear polyatomic molecules 5-11

6 Diatomics and triatomics: vibrational energy levels and spectra 6-1

6.1 Diatomic molecules: harmonic motion 6-1
6.2 Anharmonicity and dissociation 6-5
6.3 Vibration–rotation spectra of diatomic molecules 6-10
6.4 The vibrations of triatomic molecules 6-12

7 Large molecule infrared absorption spectroscopy 7-1

7.1 Group frequencies and skeletal modes 7-1
7.2 Infrared spectroscopy in the condensed phase 7-3
7.3 Near-infrared spectroscopy 7-4

8 Raman scattering and spectral interpretation 8-1

8.1 Rayleigh scattering 8-1
8.2 Vibrational Raman scattering 8-2
8.3 Depolarisation ratio 8-10
8.4 Resonance Raman spectroscopy 8-12

9 Electronic and vibrational states in large molecules 9-1

9.1 Electronic states, transitions, and molecular structure 9-1
9.2 Vibronic structure in electronic absorption spectra 9-2
9.3 Vibronic structure in electronic emission spectra 9-3
9.4 Transition metal complexes: vibronic coupling in electronic transitions 9-5

10 Electronic transitions, colours, and detection 10-1

10.1 The origins of colour 10-1
10.2 Photometry and Beer’s law 10-2
10.3 Organic molecules: conjugation and colour 10-5

11 After light is absorbed: photophysics in an excited 11-1

electronic state
11.1 Interplay of excitation and decay 11-1
11.2 States accessible to photoexcitation 11-2
11.3 Decay channels 11-5

viii
 Molecular Photophysics and Spectroscopy (Second Edition)

12 Molecular fluorescence 12-1

12.1 Quantum yields and fluorescence measurements 12-1
12.2 Transition dipole orientations 12-3
12.3 Photoselection and fluorescence anisotropy 12-4
12.4 Fluorophores and laser-induced fluorescence imaging 12-5
12.5 Quantum dots 12-6

13 Fluorescence resonance energy transfer 13-1

13.1 Mechanism for intermolecular energy transfer 13-1
13.2 Spectroscopic shift 13-3
13.3 Distance measurements 13-4

14 Chiral phenomena and optical activity 14-1

14.1 Criteria for chirality in matter and in light 14-1
14.2 Circular dichroism 14-2
14.3 Optical rotation 14-4

15 Multiphoton absorption in molecules 15-1

15.1 Two-photon absorption 15-1
15.2 Non-resonant two-photon absorption 15-1
15.3 Resonant two-photon absorption 15-3
15.4 Two-photon spectroscopy 15-4
15.5 Higher order processes 15-6
15.6 Multiphoton imaging and processing 15-8

Bibliography 16-1

Index I-1

ix
 Preface to the second edition

This new edition provides an extended photon‐based description of modern

molecular spectroscopy and photophysics, with additional examples drawn from
across the wide breadth of science embraced by this continuously developing area of
chemical physics. New foundational chapters have been added, covering the basic
electronic structures of atoms and molecules, the concept of molecular orbitals, and
an introduction to molecular symmetry and the formalism of point group theory.
These principles underpin the topics discussed in the subsequent chapters, laying the
groundwork for more detailed or advanced studies. The original content of the first
edition has also been somewhat reorganised to accommodate further new material
in logical sequence.
The concepts and applications of spectroscopy and their relationship to molecular
motions have been reordered to enhance the pedagogical delivery, starting with the
Born–Oppenheimer approximation, followed by rotational and vibrational (infrared
and Raman) spectra of diatomics, moving on to larger molecules in the condensed
phase. Electronic transitions are then considered with a discussion of Franck–
Condon factors, the different origins of colour, and the photophysical mechanisms
of excited state decay with an emphasis on fluorescence detection. The book
concludes with overviews of fluorescence resonance energy transfer, chirality and
optical activity, and multiphoton processes, each with expanded content.
As with the first edition, intentionally in contrast to most older textbooks and
monographs, this approach dwells less on high-level theory and instrumentation:
instead, the emphasis is on describing physical principles and mechanisms, supple-
mented by examples of real-world applications. Key equations are included, and
their origins are explained, but the mathematics is again kept to a minimum and
illustrated by worked examples; the required knowledge of university-level calculus
and linear algebra seldom extends beyond recognising the meaning of derivatives,
integrals, and matrices. As such, our intended readership includes senior under-
graduates and graduate students seeking an accessible overview of the major topics
in modern photophysics and molecular spectroscopy.
David L Andrews and Robert H Lipson
Norwich, UK and Toronto, CA, December 2020

x
 Preface to the first edition

This book aims to provide a fresh, photon‐based description of modern molecular

spectroscopy and photophysics, with applications that are drawn from across the
breadth of chemistry, biology, physics and materials science, including recent
developments. In contrast to many older, established textbooks, this approach
dwells less on detailed accounts of theory and instrumentation. Instead, the focus is
at the mechanisms that operate at the fundamental level, on how light absorption
and scattering occur in molecules, and what happens to the energy which the
molecules acquire. Mathematics is kept to a minimum, and the required knowledge
of calculus seldom extends beyond recognizing the meaning of derivatives and
integrals. Quantitative understanding is nevertheless put into practice in example
calculations which pepper the text; SI units are used throughout. With the aid of
extensive, purposely devised illustrations, this approach aims to foster a deeper
intuition for the photophysical processes involved in light–matter interaction,
aiming to consolidate the principles, and to exemplify how widely ranging
information can be derived from spectroscopic study.
I am very grateful to members of my research group who have read the
manuscript and offered me helpful advice. Particular thanks are due to David
Bradshaw, Jamie Leeder and Mathew Williams.
David L Andrews
Norwich, June 2014

xi
 Author biography

David L Andrews
David Andrews is known for advances in the theory of quantum and nonlinear optics,
fundamental photonics, energy transport, and optical vortices. He has published 400
papers and over twenty books, including as author or co-author Lasers in Chemistry,
Optical Harmonics in Molecular Systems, Optical Nanomanipulation, and an
Introduction to Photon Science and Technology. David has taught extensively at the
University of East Anglia and run numerous spectroscopy courses for industry: he has
twice been Chair of the Royal Society of Chemistry Molecular Spectroscopy Group.
Andrews is a Chartered Chemist and Chartered Physicist, and he is the 2021 President
of SPIE, the international society for optics and photonics.

Robert H Lipson
Rob Lipson has published extensively in the areas of laser spectroscopy, nano-
science, and novel material for photonics applications. He has taught a wide variety
of undergraduate and graduate level physical chemistry courses at Western
University and the University of Victoria (UVic). Administratively, he has served
as Chair of the Department of Chemistry at Western and two terms as Dean of
Science at UVic. In addition to his academic work, he was also a long-serving Senior
Editor of the Canadian Journal of Chemistry, and a panel member on numerous
funding and research integrity committees for the federal Government of Canada.
Lipson is a Fellow of the Chemical Institute of Canada.

xii
 List of symbols

Mostly of the symbols used in the following text are standard, including several
unfortunately overworked symbols that commonly have more than one meaning.
Usually, the correct meaning in any appearance is clearly given by the context, and
definitions are given in each of the relevant sections.

A absorbance (optical density); also Einstein A-coefficient

B rotational constant
B´ rotational constant for an upper vibrational level
B magnetic field
c speed of light in vacuum
c´ speed of light in a refractive medium
C concentration
Cn rotation about an axis by 2π/n radians
D centrifugal distortion constant
De dissociation energy relative to the classical equilibrium point
D0 dissociation energy relative to the vibrational ground state
e electron charge
E energy
E identity operation in group theory; also 3D irreducible representation in group theory
E electric field
F normalised fluorescence spectrum
g degeneracy factor; also dissymmetry factor; also gerade (even parity)
h Planck constant
ℏ reduced Planck constant = h/2π
Hint interaction Hamiltonian operator
I irradiance: also moment of inertia; also nuclear spin
Ii moment of inertia about the axis denoted by the subscript
J rotational quantum number; also total angular momentum
J’ rotational quantum number in an upper vibrational level
k force constant; also rate constant
k wave-vector
kB Boltzmann constant
ℓ azimuthal quantum number
L angular momentum; also Avogadro number
m atomic mass
mℓ magnetic quantum number
ms spin quantum number
m0 rest mass
m magnetic dipole
mr reduced mass
M molar mass
MS total spin component for multielectron atoms
ML azimuthal quantum number for multielectron atoms
Mfi transition dipole moment
n number of photons; also principal quantum number
N number of atoms in a molecule
p linear momentum

xiii
 Molecular Photophysics and Spectroscopy (Second Edition)

P degree of polarisation
q electron coordinate; also charge
Q generalised nuclear coordinate
Qe nuclear coordinate in equilibrium position
r bond length; also florescence anisotropy
r radial displacement vector
re equilibrium bond length
R distance
RH Rydberg constant
Ri rotation about the axis denoted by the subscript
R0 Förster distance
s spin
Sn improper rotation about an axis of 2π/n radians; also nth singlet state of a molecule
S total spin; also label for a singlet electronic state
t time
T absolute temperature; also kinetic energy; also transmittance
T tetrahedral; also 3D irreducible representation in group theory
Ti spatial translation along the Cartesian axis denoted by the subscript
u unified atomic mass unit
u ungerade (odd parity)
v velocity; also vibrational quantum number
V potential energy; also volume
VM molar volume
w beam radius
xe anharmonicity constant
z propagation distance
Z nuclear charge
Zeff effective nuclear charge
α polarizability; also absorption coefficient; also fine structure constant
α′ mean polarizability derivative
Δν full-width at half-maximum linewidth
ε extinction coefficient (molar absorption coefficient)
ε0 vacuum permittivity
γ′ polarizability anisotropy derivative
θ angle
Γ transition rate
κ orientation factor
λ wavelength
Λ axial component of angular momentum
μ electric dipole
ν frequency
ν´ scattered frequency
ν0 classical frequency of vibration; also frequency of peak resonance
ν wave-number
Π electronic state of a linear molecule with Λ = 1
π off-bond axis molecular orbital
ψ general wavefunction; also electronic wavefunction
Ψ vibrational wavefunction
Φ molecular wavefunction
ρ density; also density of states; also depolarization ratio

xiv
 Molecular Photophysics and Spectroscopy (Second Edition)

σ absorption cross-section
σ on-axis molecular orbital; also reflection operation in group theory
Σ axial projection of atomic spin; also diatomic or linear molecule state label when Λ = 0
τ decay lifetime
ω circular frequency
Ω axial component of total angular momentum

xv
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 1
Introduction to molecular spectroscopy and
photophysics

1.1 The distinctiveness of molecular interactions with light

Spectroscopy is all about determining information on the atom- and molecule-level
constitution of matter, primarily based on observing the interactions of that matter
with light. Most of the world that we observe below our feet is condensed phase
(solid and liquid) matter and, on and above the surface of our planet, almost all of it
is made of molecules. Our eyes and minds routinely perform a kind of basic
spectroscopy when we observe colours caused by the absorption of light, and from
this we can—and often do—make simple chemical deductions. For example, we
know that white hair usually signifies an absence of a pigment, melanin—a lack
typically associated with the process of ageing. We recognise the green colour of
many vegetables as denoting a content of chlorophyll, the dark red of some meats as
a particularly rich source of a compound containing iron. Developed into a
quantitative science that more accurately distinguishes the amount of absorption
as a function of optical wavelength, this affords the basis for spectroscopy based on
visible light. Extending the principle beyond the visible region leads us to visible/
ultraviolet, infrared and microwave spectroscopies. In all of these, for reasons that
we shall see, molecules display behaviour that is very different from the form of
atomic spectra.
The absorption of electromagnetic radiation represents the basic principle in a
wide range of spectroscopies. In each absorption process molecules gain energy
through capture of the quantised radiation, and spectroscopy exploits the fact that
there is a powerful specificity in the wavelengths that individual kinds of molecule
can absorb. Equally, the emission of light—by materials that contain excess energy
(usually resulting from prior absorption, or another excitation process)—also reveals
important information. So too do certain types of light scattering where light is not
absorbed but, by interaction with materials, changes its direction. In each case the

doi:10.1088/978-0-7503-3683-3ch1 1-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

observed dependence on optical wavelength provides the crucial basis for spectro-
scopic interpretation. The origin of this dependence is of course energy conservation:
the requirement that the balance of photons, delivered to and emerging from each
particle in the course of a single process, must match the transition energy lost or
acquired by that particle—and the latter energy change is itself constrained by the
restriction of moving between quantum energy levels.
The capacity to infer structural information from optical interactions thus owes
its origin to the fact that atoms and molecules have discrete levels, determined by
quantum mechanical principles, and given by the solutions of the Schrödinger
equation. Accordingly, since photon energies are related to frequency by
E = hv = hc/λ, (where h is Planck’s constant, c is the speed of light, ν is the optical
frequency and λ the wavelength), only certain frequencies of light can induce
changes from one level to another. Always, some quantum level other than the
stable (lowest energy) ground state is involved. The reason we can deduce so much
from molecular spectra is that the molecular transitions responsible are very strongly
linked with other characteristic properties of the matter—primarily their atomic
framework. This link between spectroscopic features and molecular constitution
provides the basis for a whole host of applications, familiar in our everyday world.
The field of photophysics is primarily concerned with the properties and
behaviour of molecular excited states. These are usually very short-lived—they
are almost always less than a millisecond in lifetime, and frequently present for times
on the scale of nanoseconds (10−9 s). What happens to the energy these states
contain, as they ‘relax’ to states of lower energy, is itself a rich field of study. By a
variety of mechanisms, energy that most often has been captured from light itself is
thereby available not only for subsequent re-emission as light; it can be transported
away over microscopic scales to initiate other processes. Most familiar and
important to us are the primary processes of vision, and of photosynthesis, each
of which entails a rich interplay of molecular photophysical mechanisms. We could
not find our way around the world, nor have any food to sustain our lives even if we
did, without photophysics.
In successive chapters, this book aims to build an increasingly rich and detailed
picture of how light is absorbed, what then happens to the energy acquired, and
what information the study of such processes can reveal. As indicated, we shall focus
on molecular matter which, with the exception of metals and minerals, accounts for
most of the matter we see in the everyday world. We shall start with the simplest
examples, including those few molecules we do not directly see—the molecular
gases. Once the general principles are established, we shall cover the field from the
lowest to the highest energy molecular transitions, from free rotations in a gas to
multiphoton absorption in complex biological media.

1.2 Properties of molecules and their spectra

First, it is instructive to consider why there is such an immediately evident difference
between the look of visible and ultraviolet spectra for gas phase and most condensed
phase samples: clearly, the nature of the phase is extremely important.

1-2
 Molecular Photophysics and Spectroscopy (Second Edition)

Under ambient conditions most gas-phase samples are comprised of only atoms or
small molecules. Larger molecules, whose standard state phase is seldom gaseous,
can usually be present only in low concentrations—which is why trace detection
techniques are required for their observation. At any instant, the atoms or molecules
in a gas are well separated in space, and their random motions produce a temper-
ature-dependent Doppler broadening (due to motion-dependent frequency shifts) in
their absorption and emission spectra.
Whatever the nature of the sample, transitions that occur at visible and ultraviolet
wavelengths generally involve transitions between electronic states—for most gases,
photons of sufficient energy in fact have wavelengths just outside the visible
spectrum, in the near-ultraviolet region. Atomic spectra may exhibit a multitude
of lines, but those lines appear clearly separated and, compared to molecular
spectra, much narrower. In contrast, under normal ambient conditions, the visible/
ultraviolet spectra of condensed phase materials will most often exhibit broad,
overlapping features with no simply describable shape. And of course, in any
molecule, the electron sharing involved in chemical bonding between the atoms
means that the electronic properties are entirely different from the constituent
atoms; absorbing light will not excite individual atoms.
In regular atomic or simple ionic crystals (metals, semiconductors and salts, for
example) there is a uniformity of multiply-repeated unit cell structures that allow
bulk properties to emerge. But our visible world is mostly made up of condensed
phase materials: amorphous solids, liquids, solutions etc. Such substances most often
comprise relatively large molecules, each of which is electronically distinct; it is only
the non-bonding forces of attraction between them that prevents their complete
spontaneous evaporation. The interactions experienced by individual molecules can
vary much more widely from site to site, however, according to their local electronic
environment. Moreover, whereas every atom comprises a single nucleus, polyatomic
molecules can rotate and have internal vibrations whose number grows linearly with
the number of atoms per molecule—and these can also be involved in electronic
transitions, as we shall see in a later chapter.
There is one other, fundamental difference that plays into the linewidth of
electronic transitions in the condensed phase, and that is the uncertainty effect of
excited state lifetimes that are often in the sub-nanosecond regime. In gases, in which
the atoms or molecules are essentially isolated and infrequently collide, radiative
emission is a primary means of excited state decay. However, in the condensed
phase, the availability of a host of other internal decay and energy transfer processes
can dramatically shorten excited state lifetimes. Heisenberg’s time-energy
Uncertainty Principle then decrees that the associated transition frequencies become
less well-defined: they acquire greater breadth.
For all these and other reasons that we shall look at in due course, the electronic
spectra of most molecular materials display very broad bands with only their peak
values precisely quantifiable. But there is a flip side to this seemingly problematic
complexity; unique possibilities then arise to derive additional information from
other measurements, such as optical polarisation effects and excited state lifetimes.
And even more importantly, spectroscopy in other regions of the electromagnetic

1-3
 Molecular Photophysics and Spectroscopy (Second Edition)

spectrum can directly identify individual molecules, and indeed reveal their atomic
composition and structure.

1.3 The electromagnetic spectrum

As we know, if an atom or molecule absorbs light it gains energy, causing specific
motions of the electrons and/or nuclei. Quantum mechanics provides a framework
for understanding how the energies associated with these motions can be used to
secure structural information about the material. A system once excited will
ultimately lose energy and generally return to its lowest energy state—a process
known as excited state decay, or relaxation. Alternatively, certain molecules may
undergo photochemical change, producing new species. As we shall see later, the
more highly energetic states populated by the absorption of light are usually those
that rapidly decay non-radiatively, and it is the absorption and dynamical
deactivation pathways following absorption that constitute the core of molecular
photophysics.
There are important connections between the different regions of the electro-
magnetic spectrum, and the types of internal molecular motions with which they are
broadly associated. Spectroscopy uses several different measures of photon energy E,
and it is helpful to first spell them out with reference to the units in which they are
usually reported;
E = hc λ = hν = hcν˜ = ℏω, (1.1)
where λ and ν are the photon wavelength and frequency. In SI units these are
expressed in m and Hz (≡ s−1) respectively—though in the ultraviolet and visible
ranges, wavelength is more commonly reported in nm (= 10−9 m). The symbol ν̃ is
called the wavenumber, which is the number of wavelengths that fit within a unit of
length. Traditionally spectroscopists use cm as the unit of length, and so wave-
numbers are commonly expressed in cm−1. The last identity in equation (1.1) casts
the photon energy in terms of an angular frequency ω (= 2πν with units of radians s−1),
while the reduced Planck’s constant (also known as the Dirac constant) is defined as
ħ = h/2π.
Table 1.1 shows the different wavelength regions that will be discussed in this
book. There are powerful spectroscopies associated with still shorter wavelengths
(for example photoelectron spectroscopy using extreme ultraviolet or x-ray radia-
tion) and also much longer wavelengths (e.g. nuclear magnetic resonance, based on
radio waves) but their uses fall outside the usual province of photophysics. Most of
the phenomena and applications that we shall encounter are based on laser
instrumentation, exploiting the unique monochromaticity and directional power
associated with such sources. Although there is no rigid boundary between the
different wavelength ranges, those indicated in table 1.1 are accepted as standard by
most laser spectroscopists.
A few details of the wavelength sub-divisions are worth noting. In the far-
ultraviolet region, the term ‘vacuum ultraviolet’ is commonly used to denote the fact
such radiation is absorbed by oxygen (and hence air itself), necessitating the

1-4
 Molecular Photophysics and Spectroscopy (Second Edition)

Table 1.1. Wavelength λ, frequency ν, wavenumber ν, and photon energy E, for selected spectral regions, and
the types of atomic and molecular motions with which they are associated.

Range λ (nm) ν (Hz) ν (cm−1) E (J) Motion (s)

−18
Vacuum 100–200 3.00 × 10 −
15
10 − 50,000
5
~ 2 × 10 – Electronic +
Ultraviolet 1.50 × 1015 ~ 1 × 10−18 vibrational,
(VUV) rotational
Ultraviolet 200–380 1.50 × 1015 − 50,000–26,300 ~ 1 × 10−18 – Electronic +
(UV) 7.89 × 1014 5.23 ×10−19 vibrational,
rotational
Visible 380–780 7.89 × 1014 − 26,300–12,800 5.23 × 10−19 – Electronic +
3.84 × 1014 2.54 × 10−19 vibrational,
rotational
Infrared (IR) 780–106 3.84 × 1014 − 12,800–10 2.54 × 10−19 – Vibrational +
3.00 × 1011 ~ 2 × 10−22 rotational
Microwave 106 − 109 3.00 × 1011 − 10–0.01 ~ 2 × 10−22 – Rotational
3.00 × 108 ~ 2 × 10−25

provision of vacuum conditions for spectroscopic studies. Photon energies in this

region are typically sufficient to ionise molecules. In the visible region, spanning
almost an octave of wavelengths to which the human retina responds, individual
photons usually do not have enough energy to cause molecular ionisation. They can
be sufficiently energetic to break the weakest chemical bonds in a few materials, but
most spectroscopy in this region causes no real chemical change; the whole aim of
these techniques is to non-destructively gain information on the composition of
materials.
Spectroscopists commonly sub-divide the infrared region into three: near-infrared
(wavelengths running from the red end of the visible spectrum at 780 nm up to
2.5 μm), mid-infrared (2.5–50 μm), and far-infrared (50 μm–1 mm). More recently
the additional term ‘terahertz radiation’ has been introduced to specifically identify
the range 0.1–20 THz (1 THz = 1012 Hz), corresponding to ~3–600 cm−1, which
encompasses parts of the IR and microwave regions. This spectral region, for which
few suitable sources existed until recently, proves to allow ready access to
information on not just individual molecules, but also their local interactions. We
shall look in more detail at the types of transition associated with each spectral
region as we discuss each of the rich variety of spectroscopic techniques.

1.4 Photon properties: polarisation and spin

Electromagnetic radiation conveys electric and magnetic fields that rapidly oscillate
in time at the optical frequency, and over space within each wavelength, as
illustrated in figure 1.1. All of the charges in a molecule—both the electrons and
the nuclei—respond to those fields, but it is the least bound electrons that respond
most strongly, and due to their relatively small mass they are able to oscillate at the

1-5
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 1.1. Variation of electric and magnetic fields in propagating light, the oscillations having a repeat
length that defines the wavelength. In free space the two fields are always mutually orthogonal: (a) plane
(linearly) polarised light, where the two fields are exactly in phase; (b) circularly polarised light (right-handed
case) where the fields have a π/2 radians phase difference.

frequency of the electric field. The form of the field oscillations is most often
represented as a plane wave, as shown in figure 1.1(a), but it may for example take
the form of a helix (of either left- or right-handedness) as in figure 1.1(b). These two
cases are known as plane polarisation and circular polarisation, respectively.
Circular polarisations convey angular momentum, ±ℏ per photon according to
handedness (the positive sign for left when looking down the propagation axis of the
incoming wave), thereby revealing another key property of photons; the quantisation
of spin. Many of the most important properties of light, especially in the realm of
spectroscopy, owe their origin to the fact that photons have an intrinsic unit spin,
S = 1. This categorises photons in the class of elementary particles known as bosons,
i.e. particles of integer spin whose collective behaviour is subject to Bose–Einstein
statistics. This contrasts strongly with the electrons with which they interact, whose
own half-integer spin renders them fermions (being subject to Fermi–Dirac statistics).
One immediately important aspect of spin is that a collection of bosons can exist
in the same quantum state—and photons can therefore produce highly coherent,
monochromatic and directional beams, which is of course a major feature of lasers.
(Conversely, not even two electrons can exist in the same quantum state, as the
periodic table makes very apparent.) We shall encounter numerous other evidences
of photon spin in the course of the following chapters, because it is a feature that
underpins almost every one of the selection rules in spectroscopy—rules that
determine whether or not a quantum transition can take place, even if the photon
has exactly the right energy to match a gap in molecular energy levels. These rules
still operate in the more common case where light is plane-, rather than circularly
polarised, since plane polarisation acts as a superposition of opposite circular
polarisations.
To round off the most significant properties of photons it is worth finally observing
that despite having zero mass, they do possess linear momentum, p = h/λ per photon.
This neatly exhibits both the quantum and relativistic nature of the particle, since
energy, momentum and ‘rest mass’ m0 are linked by the relation E 2 = p2 c 2 + m 02c 4 .
With m0 = 0 and relation (1.1), the result p = h/λ immediately follows. However, linear
momentum has very little direct influence in most forms of spectroscopy; it mostly
comes into its own in the sphere of optical manipulation of atoms and molecules; for
example, trapping samples in space and reducing their kinetic energies to near zero.

1-6
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 2
Atoms and molecules: developing principles of
electronic structure

2.1 Review of atomic orbitals, angular momentum and electron spin

Molecular photophysics and spectroscopy involves well-established, content-rich
designations for atomic and molecular states. A particularly useful starting point is
to concisely review the familiar labels for the electronic states of the hydrogen atom,
and their physical connotations—primarily because so much of molecular spectro-
scopy builds upon its foundational concepts.
The H atom consists of a nucleus containing one proton, and one electron bound
to it by a Coulomb potential. It should be noted that this element, as well as many
others, have isotopes: they have the same number of electrons, and nuclei containing
the same number of protons but a differing number of neutrons. For example,
deuterium, D, is an isotope of hydrogen (2H) found in ‘heavy water’ and having
mass of essentially 2u, where u is the fundamental unit of atomic mass which is close
to the mass of 1H hydrogen. Another is tritium, T (3H), which has a nucleus with two
neutrons as well as the one proton.
The non-relativistic quantum mechanics of such a system can be solved analyti-
cally using the Schrödinger wave equation, to determine the orbital wavefunctions
ψn,ℓ,m ℓ (r) and energy eigenvalues En, where r is the position vector of the electron. Let
us first focus on the three quantum numbers: n, ℓ, and mℓ.
The principal quantum number n establishes the energy En of the electron in the
H atom, which is given by;
RH
En = − , (2.1)
n2
where n is the principal quantum number, which can take only discrete positive
integer values ⩾1; that is, n = 1, 2, 3, …, ∞. The Rydberg constant, RH, signifies the
minimum energy to ionize the H atom from its lowest electronic state or ground

doi:10.1088/978-0-7503-3683-3ch2 2-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

state: that is the energy needed for ionization; removal of the electron leaving behind
a proton H+. This energy is often expressed in electron-volt units, ~13.6 eV, where
1 eV is around 8066 cm−1. It should be noted that this lower limit for ionization is
defined as where the kinetic energy of the freed electron is zero. The energies of all
H atom states are referenced to this zero, and therefore have negative (binding)
energies. In this regard the ground state is the most bound state of the system.
Spectroscopy does not involve measuring the absolute energies of states, rather
the energy differences for transitions between them. Hence, it makes more sense
to pick the ground state as a ‘zero’ of energy and measure excited state energies
relative to that level. Thus, when an atom (or molecule) absorbs light it goes from
the lower state (usually the ground state) to an excited state that is higher in
energy. Conversely, when an atom (or molecule) transitions from an excited state
to a lower state by emission or some other deactivation mechanism, it loses
energy. Regardless, the transition energies are reported as positive quantity,
ΔE = Eexcited state − Elower state. In spectroscopic studies of the light emitted in a
discharge lamp containing elemental hydrogen (H2) one observes a series of
atomic H-atom lines (called transitions) originating from high n-levels and
terminating on specific lower n-states. Here, the energy level differences between
H atom levels with quantum numbers labelled n1 and n2 can be written as:
⎛1 1 ⎞
ΔE = E n2 − E n1 = RH⎜ 2 − 2 ⎟ . (2.2)
⎝ n1 n2 ⎠

Transitions terminating on n1 = 1 form the Lyman series, which lies mainly in the
extreme ultraviolet (λ < 100 nm), while the group of transitions terminating on
n1 = 2 and n1 = 3 are called the Balmer and Paschen series, respectively. The former
lies in the visible range while the latter is observed in the infrared. There are further
series at even longer wavelengths. One important point here is that the energy levels
in hydrogen (and all other atoms as well) are not equally spaced. Almost half the
energy required to ionize H—to remove its electron completely—is required to excite
the atom from its ground state just to the first excited state, n = 1 to n = 2. This is
comparable to the energy to excite transitions from n = 2 up to the ionization limit,
even though there are an infinite number of n-states in this region.
Before delving deeper into the quantum features of the H atom, some comments
regarding spectral lines is warranted. In describing the precise electronic structure of
atoms and simple molecules, more intricate quantum calculations will deliver
extremely accurate results for the energy levels of the ground and most excited
states. Indeed, advanced computational techniques based on quantum mechanical
models can give very precise figures for the excited state energies of even quite large
molecules. However, it is usually only gas phase samples that exhibit sharp lines in
their absorption or emission spectra. Hydrogen atom spectra are indeed composed
of sharp (narrow linewidth) features, corresponding to transitions between different
electronic states—but even here, or for other atoms, how ‘sharp’ a line is will depend
on the sample conditions, the presence or absence of electric and magnetic fields, the
light source, and the detection system being used. Many details will depend on the

2-2
 Molecular Photophysics and Spectroscopy (Second Edition)

experimental setup, and of course this is one of the key reasons for the wide
application of laser instrumentation: to achieve the narrowest possible linewidth for
the light source.
In addition to the generic lineshape issues relating to the nature of the sample,
discussed in chapter 1, a host of other line-broadening mechanisms generally contribute
to an effective linewidth for any spectral feature. As we observed, the effects of
uncertainty line-broadening due to finite excited state lifetimes are significant; so too are
the broadening due to the Doppler effect of random particle motions, and through
stochastic perturbations produced by inter-particle collisions. These fall into two
categories of line-broadening relating to intrinsic and motion-dependent origins, known
as homogeneous and inhomogeneous respectively: each is generally associated with a
specific form of lineshape function. Homogeneous lifetime broadening, whose magni-
tude depends for each transition on the specific excited state, produces a frequency
response with a Lorentzian lineshape, f (ν ) ∼ 1/[(ν − ν0 )2 + ( 12 Δν )2], where v0 is the
resonance frequency and Δv is the FWHM (full-width at half-maximum) linewidth. On
the other hand, inhomogeneous collisional and/or Doppler broadening give rise to a
Gaussian shape, f (ν ) ∼ exp[−4 ln 2((ν − ν0 )/Δν )2]. When both phenomena are
present, the two lineshapes can be convoluted to yield a Voigt profile.
Returning now to the H atom, theory reveals that there is another quantum
number, symbol ℓ, known as the azimuthal quantum number, that signifies the
orbital angular momentum of the electron as it travels around the nucleus. Although
it is not determinable from any of the prominent frequencies in the spectrum—note
that it does not feature at all in equation (2.2)—it is spectroscopically very
important, as we shall see. Specifically, it is related to the square of quantised
orbital angular momentum of the electron in each nℓ orbital in units of ħ2, and to the
shape of the orbital itself. However, that shape, whose boundaries define a volume of
space around the nucleus where one is most likely to find the electron if measured,
does not tell us about the actual trajectory of the electron within that space. Indeed,
quantum mechanically the idea of a trajectory is a meaningless concept. It can be
shown that the value of the square of the angular momentum is = ℓ(ℓ + 1) ħ2.
The quantum number ℓ is also quantised and its possible values are related to n.
For a given n value, ℓ can take on the values 0, 1, …, n − 1, such that for n = 1, ℓ = 0
alone. Orbitals with ℓ = 0, 1, 2 and 3 are labelled s, p, d and f, respectively: the
electronic ground state is therefore labelled 1 s. Similarly, for n = 2, 3 and 4, there are
2, 3 and 4 possible states, as listed in table 2.1. In principle there are orbitals with
ℓ > 4 (g-orbitals, h-orbitals, etc.) but they are usually not considered because the
highest ground state orbitals in the periodic table are the 5f states of the actinides.
Since the hydrogen atom energy levels given by equation (2.1) are independent of
ℓ, i.e., different states for a given n have the same energy, they are called degenerate.
The basic shapes of the ℓ-orbitals, determined as the angular part of the wave-
functions in the solution of the Schrödinger equation, are similar for every n, except
that as n increases, they grow in volume. In contrast, as ℓ increases they exhibit an
increasing number of radial and angular nodes (spatial positions where one will
never measure the electron).

2-3
 Molecular Photophysics and Spectroscopy (Second Edition)

Table 2.1. Observed orbitals and n, ℓ quantum numbers for the first four energy
states of H.

N ℓ State labels nℓ

1 0 1s
2 0, 1 2s, 2p
3 0, 1, 2 3s, 3p, 3d
4 0, 1, 2, 3 4s, 4p, 4d, 4f

Figure 2.1. Shapes of the 1 s, 2p and 3d orbitals of the hydrogen atom. Shaded lobes indicate an opposite sign for
the angular wavefunction, compared to unshaded lobes. The subscripts dependent on (x, y, z) indicate the
directions in which the orbitals point. There is no directional preference for s-orbitals, which have a characteristic
round shape. The 4f-orbitals have more complex structures and are not presented. Note that s- and d-orbitals do
not change sign by inversion through the nucleus; that is going from (x, y, z) to (−x, −y, −z)—they are of even
spatial parity: however, the p-orbitals do—they have odd parity.

Several of the most important orbital shapes are shown in figure 2.1, in which
shading indicates the sign of the wavefunction—an aspect that proves to be
important in the rules governing the types of optical transition. Clearly, some are
symmetric about the nucleus, while others are not; this behaviour is known as spatial
parity. It can be shown that the parity eigenvalue for one-electron systems is given
by (−1)ℓ. It follows that s and d orbitals are even with respect to inversion (parity +1),

2-4
 Molecular Photophysics and Spectroscopy (Second Edition)

and are therefore labelled as g-states, where g stands for gerade, the German word
for even. Similarly, p and f states are odd with respect to inversion (parity −1) and
are labelled as u-states, deriving from ungerade, the German for odd. It is worth
recalling, however, that the modulus square of the wavefunction is proportional
to the charge density distribution in space—and this of course has a constant
negative sign.
The standard designation of Cartesian axes, used in describing the shape and
orientations of atomic orbitals in figure 2.1, invites more comment. There are no
unique directions in empty three-dimensional space, and atoms themselves have
spherical symmetry—so any such labelling has to be arbitrary. At least some of the
degeneracies therefore become obvious. However, physically imposing a space axis
by placing the atom within a constant magnetic field breaks those degeneracies and
leads to line splittings in the spectrum. We can then deduce that there is a third
quantum number mℓ, accordingly termed the magnetic quantum number, which tells
us for a given n the number of orbitals of a given ℓ. Formally, mℓ ranges from −ℓ to +ℓ
in integer steps, leading to a total of 2ℓ + 1 degenerate states. Thus, each n-level, for
n ⩾ 1 has one s-orbital, while for every n ⩾ 2 there are also three p-orbitals, and
so on.
There is one further, additional quantum number for the electron associated with its
spin s. It is related to the square of an intrinsic spin angular momentum of the electron.
Although the quantum behaviour of the electron spin resembles that of its orbital
motion, it does not in fact depend on spatial coordinates. Spin is something therefore
that is usually added on in an ad hoc way to the solutions described above (although it
does naturally emerge using a relativistic formulation of quantum mechanics). Similar
to ℓ, the square of the spin angular momentum is given by s(s + 1)ħ2 and every electron
has s = ½. The number of spin states for s = ½ is labelled by the spin quantum number
ms; every electron has only two allowed spin states, ms = + ½ (spin up) and ms = −½
(spin down). In many chemistry books the spin up function is denoted by α, and the spin
down function likewise β. While the use of the spin terms ‘up’ and ‘down’ is pervasive,
it is a little unfortunate because it implies some specific spatial orientation, which is not
the case.
There is one further spectroscopic feature to introduce before we move on from
hydrogen to more complex systems. In any light absorption or emission process it
does not, in general, simply suffice for the energy gap between the initial and final
states of the atom to be matched by the energy of a photon. In fact, additional
criteria have to be satisfied; it is necessary to conform to selection rules. This is an
aspect of special significance for not only for optical transitions in atoms but also in
molecules, where considerations of parity, the forms of charge distribution, and
angular momentum all come into play.
In both absorption and emission, the strongest transitions correspond to a dipolar
form of coupling with the electric field of the radiation. More specifically, the
property that determines the strength of such interactions transition electric dipole,
through which an atom or molecule interacts with the electric field of optical
radiation. It is important to emphasise the quantum nature of this parameter, a

2-5
 Molecular Photophysics and Spectroscopy (Second Edition)

characteristic of photon absorption and emission. The transition electric dipole

moment is a property specifically defined and associated with each particular
transition. It is not the same as a static or permanent dipole, nor is it a difference
between the dipole moments for the initial and final states—each of which would of
course be zero in any atom. Its definition entails a quantum integral engaging the
electric dipole operator, which in the case of hydrogen is given by μ = −er, where −e
is the electron charge and r is the instantaneous position vector of the electron
relative to the nucleus. The integral takes the form shown on the left-hand side of the
following equation: a transition in the H atom will be allowed if the result is non-
zero, i.e.

∫ ψn′⁎′,ℓ′,m ′(r )μψn,ℓ,m (r )dr ≠ 0,

ℓ ℓ
(2.3)

where the prime indicates an upper state. If the wavefunction contains the complex
number i = −1 then one must also use the complex conjugate of upper state
wavefunction (where i → −i ), as indicated by the asterisk. It can be shown that there
are no restrictions on transitions in H between different n-levels, but the change in
the azimuthal quantum number must satisfy Δℓ = ±1. This is sometimes referred to
as the Laporte selection rule. Ultimately, this is related to the angular momentum
properties of light. As we have seen, fundamental states of a photon are either left-
hand circularly polarised with angular momentum +ħ, or right-hand circularly
polarised with angular momentum −ħ. It follows that when an H atom absorbs one
photon the change in the angular momentum of the atom must either be Δℓ = +1 or −1
due to conservation of angular momentum. Hence, the first allowed transition in H is
from the 1 s → 2p. However, the transition 1 s →2 s is said to be forbidden. This
principle also satisfies parity, as the electric dipole operator has odd parity; the
integrand in equation (2.3) is therefore an even quantity when transitions are
allowed.
While light can interact with other moments such as the magnetic dipole and/or
the electric quadrupole, the resultant transitions are usually weaker than the electric
dipole transitions by several orders of magnitude and only become dominant in
spectral regions where electric dipole transitions are not observed.
We can now wrap up our consideration of hydrogen by noting that equation (2.3)
can be adapted to include electron spin, as;

∫ ψn′⁎′,ℓ′,m ′ (r )μψn,ℓ,m (r )dr ∫ ψm′ ψm dτs ≠ 0

ℓ ℓ s′ s
(2.4)

where τs denotes a non-spatial spin ‘coordinate’. Here an additional selection rules

arises: Δms = 0. This means that electron does not ‘flip’ its spin during a transition.
Again, this is a feature of fundamental importance—one that emerges with powerful
significance in the spectroscopy and photophysics of molecules.

2.2 Multielectron atoms

The electronic structure of multielectron atoms introduces several kinds of compli-
cation, compared to hydrogen. Beyond the simple multiplicity of charges, there are

2-6
 Molecular Photophysics and Spectroscopy (Second Edition)

now electron–electron interactions to take into account—including not just their

mutual repulsion, but also some issues of a purely quantum mechanical origin.
Moreover, it transpires that the Schrödinger equation is no longer analytically
solvable, although modern computer methods can still secure energies to arbitrary
degrees of precision.
For a multielectron atom the shapes of the individual electron orbitals resemble
those for the H atom and have the same nodal structure; however, their radial
distributions and size now depend on how strongly each electron ‘feels’ the charge of
the nucleus, Z. Specifically, outer electrons with different n and ℓ are subject to a
Coulombic pull from the nucleus that is reduced due to shielding by the inner
electrons; the effect is equivalent to their experiencing a lower, effective nuclear
n,ℓ
charges Zeff. One can write Zeff = Z − S , where Z is the full nuclear charge (equal
to the atomic number) and S is a screening constant that depends on the n and ℓ
values of the electron of interest. The latter can be interpreted as the average—
possibly non-integer—number of electrons between the nucleus and that electron.
Thus, to a first approximation the term energies of a multielectron atom can be
written as:
R
(
n,ℓ
En,ℓ = − Zeff )2 nH2 . (2.5)

The first thing to notice is the ℓ-dependence in Zeff

n,ℓ
which makes equation (2.5) seem
deceptively simple. Crucially, this means that states of a given n but different values
for the azimuthal quantum number no longer have quite the same energy—in
marked contrast to the H atom. Next, we observe that since Zeff is usually greater
than 1, the orbitals in a multielectron are more tightly bound than those for
hydrogen. Furthermore, as noted above, the degeneracy of a set of nℓ orbitals in H is
now broken. By examining the radial distribution plots for the wavefunctions of H,
one can deduce that the probability of finding an electron close to the nucleus is
largest for s electrons, followed by p electrons, then d and then f. It follows that the
effective nuclear charge felt by an outer nℓ electron follows the same trend. Thus, the
binding energies are largest for ns followed by np, then nd, and then nf. The exact
energy differences however will depend in large part on electron–electron repulsion
and correlation effects which require advanced computational techniques to
establish.
Unlike the H atom, where the electron configuration completely determines the
electronic state, the configuration for a multielectron atom can often lead to many
possible states. In electronic spectroscopy, the most relevant excited states arise from
promoting electrons in the ground state configuration to one or more unoccupied
excited state orbitals; this is a principle we shall see also extends to the electronic
spectroscopy of molecules. Therefore, an obvious starting point for our present
discussion is to establish the ground state electron configuration of the multielectron
atom. This can be done using the Aufbau Principle, which is covered in most
introductory chemistry textbooks. For example, the electronic configurations of the
first three elements H, He, and Li are 1s1, 1s2, and 1s22p1, where the exponents

2-7
 Molecular Photophysics and Spectroscopy (Second Edition)

indicate how many electrons are in the labelled nℓ shell. Essentially, as we move from
one atom to another, through the periodic table, electrons move into shells of
progressively lower binding energy.
Two other important principles arise, however, which the electron configurations
alone do not identify. The first is the Pauli Exclusion Principle, which specifies that
no two electrons in any orbital can have the same set of electronic quantum
numbers. Put another way, two electrons at most can occupy any specific (n, ℓ, mℓ)
orbital, and they must have opposite spin quantum numbers, ms; that is, one will be
spin up and the other, spin down. The second principle is Hund’s Rule, which states
that it is more energetically favourable to first fill a set of (n, ℓ, mℓ) orbitals singly
with electrons having the same spin orientation, before the accommodation of
further electrons results in any of those orbitals becoming doubly occupied and
spin-paired.
For a given ground or excited state electronic configuration the total orbital
angular momentum L of a many-electron system can be calculated from the
quantised vector sum L = Σ⃗i ℓi . As an example, consider two orbital angular
momenta ℓ1 and ℓ2. The permitted vector sums give rise to a resultant whose
magnitude lies in the range L = ℓ1 + ℓ2 → ℓ1 − ℓ2 in integer steps. Repeating the
procedure for the next electron, and so on, it is easy to see that a single multielectron
configuration can lead to many possible values of L. As each ℓ is an integer, the
resultant L values are also integers. States of different L are called terms, and they
are represented in analogy to H atom states with capitalised labels S, P, D, F, …, for
L = 0, 1, 2, 3, …, respectively. The ML components which range from –L to +L in
integers can be resolved in a magnetic field, which lifts their degeneracy.
Since each electron has a spin s = ½, one can also derive the total spin of the
system S using the vector sum S = Σ⃗i si whose permitted quantum magnitude is
similarly limited to the range s1 + s2 → s1 − s2 , again in integer steps. If the
number of electrons in the atom is even then the total S will be an integer; an odd
number of electrons leads to multiples of ½. The resultant MS components, ranging
from –S to +S in integer steps are resolvable in an applied magnetic field.
It is useful to define the spin multiplicity as the value of (2 s + 1). Two of the most
common cases of spin multiplicity also commonly arise for organic molecules. When
S = 0 (all electrons are spin-paired) the spin multiplicity = 1 and the term is referred
to as a singlet state. Similarly, if S =1 (two of the electrons in the same set of n, ℓ
orbitals have the same spin orientation), then (2 s + 1) = 3. The resultant term is
called a triplet state.
The orbital motion of the electrons in an atom (for L ≠ 0) creates a magnetic field
which can itself interact with magnetic moment of the total spin of the atom (again
provided S ≠ 0). This so-called spin–orbit interaction can couple the total L and S
angular momenta together to create different total angular momenta which are
labelled J. For light atoms with a relatively small number of electrons (typically < 40)
the L and S values considered above are summed vectorially to yield a total angular
momentum of magnitudes J = (L + S) → (L−S); a term with specific L and S will
often give rise to several values of J. In this approach to understanding the interaction

2-8
 Molecular Photophysics and Spectroscopy (Second Edition)

of spin and orbital angular momenta, called the Russell–Saunders (or L−S) coupling
scheme, the resultant terms are labelled by 2s+1LJ, and their individual energies differ.
Once again, the energies of individual MJ components, ranging from −J to +J in
integer steps, can be split by a magnetic field.
There is one further subtlety that should be noted: there may be restrictions on S
(and therefore J) due to the Pauli Exclusion Principle for terms derived from
electrons within an nℓ configuration versus terms derived from electrons in different
nℓ states. For example, a 1s2 configuration yields only a 1S0 term (the two electrons
must be spin-paired) while a 1s12s1 configuration allows both 1S0 and 3S1 terms (the
two electrons may have either the same or opposite spin orientations).
As we move on through the periodic table and examine atoms with increasingly
high atomic numbers, and large numbers of electrons, a different kind of behaviour
becomes apparent. One factor that becomes more and more important is a direct
coupling of spin and orbital angular momenta—a relativistic effect whose influence
grows with the fourth power of Z. As the atomic number increases, the resultant
spatial confinement of a large number of circulating electrons about the nucleus will
lead (by the Heisenberg Uncertainty Principle) to them having increasingly high
momenta—so it makes sense that relativistic effects become more significant.
When the number of electrons in an atom is large, the relatively larger spin–orbit
interaction will first couple the individual ℓi and si angular momenta to create
individual total angular momenta ji. Again, the ji values can be derived by adding
each ℓi and si vectorially. The total J angular momenta are found by first vectorially
adding individual values, say j1 and j2, to obtain intermediate J values. Those values
are then added vectorially to another, say j3, and so on successively until all the
electrons have been accounted for. Here, L and S are not well defined, and what are
now described as j–j coupling terms are labelled simply by J values alone—which can
have either integer or half-integer values, depending on whether there is an even or
odd total number of electrons. The number of J-states and their numerical labels
found this way will be the same as those derived using Russell–Saunders coupling,
but the actual energy ordering of the terms may be significantly different.
The parity of a multielectron atom is determined by the sum of individual electron
orbital angular momenta in the electronic configuration for a particular state; that is
by( −1)∑i ℓi . Accordingly, a term may be even (g) or odd (u) regardless of L. The
selection rules for transitions between Russell–Saunders terms, and between j–j
coupling terms are provided in table 2.2.
Again, the forbidden transitions reflect the conservation of angular momentum
requirement for the absorption (or emission) of a photon by the atom.

2.3 Diatomic molecules

With the insights and terminology gained from our review of atomic spectroscopy,
we can now begin looking at molecules, starting with the simplest case. We continue
focussing on electric states and transitions—though, of course, the inclusion of more
than one nucleus means that internal internuclear motions and also rotations of the

2-9
 Molecular Photophysics and Spectroscopy (Second Edition)

Table 2.2. Selection rules for electric dipole transitions between Russell–Saunders terms, and between j–j
coupling terms.

Russell–Saunders coupling j–j coupling

g↔u g↔u
ΔS = 0
ΔL = 0, ±1 but L = 0 → L = 0 forbidden
ΔJ = 0, ±1 but J = 0 → J = 0 forbidden ΔJ = 0, ±1 but J = 0 → J = 0 forbidden
ΔMJ = 0, ±1, but ΔMJ = 0 forbidden if ΔJ = 0

whole molecule can take place. However, the fundamental quantum transitions
associated with these very important forms of motion take place at much longer
wavelengths, and they have a very different character; we shall return to consider
them in later chapters.
An obvious difference between atoms and diatomic molecules is that the latter
have cylindrical symmetry, the internuclear bond introducing a unique, physically
distinguishable axis. Accordingly, the electronic states of diatomics are usually
labelled by the absolute magnitude of the component of orbital angular momentum
along their bond axis, Λ. By analogy with atoms, such states are labelled by the
corresponding capital Greek letters: Σ, Π, Δ, Φ, … when Λ = 0, 1, 2, 3…,
respectively. Here the orbital motion can be thought of as clockwise or counter-
clockwise rotation of the electron about the bond axis. Therefore, states with Λ ⩾ 1
are doubly degenerate since to a first approximation the two motions are indis-
tinguishable in energy. The case of Σ-orbitals (Λ = 0) requires special attention with
respect to their symmetry on reflection in any plane containing the molecular axis.
This reflection plane (and the operation of reflection) is denoted by σv—the first of
several molecular symmetry properties to be discussed in more detail in the next
chapter. Based on this mirror symmetry, Σ states that are symmetric on reflection are
labelled Σ+while those that change sign are denoted as Σ−. Such superscripts are not
required for states with Λ > 0 because of their double degeneracy.
Molecular terms are typically denoted by 2s+1ΛΩ, where Ω = (Λ + Σ) is the
component of total angular momentum along the molecular axis. Note the italics: Σ
does not mean Λ = 0: it is instead the standard symbol for the component of the
electron spin S along the bond axis, which ranges from +S to −S in integer steps.
The selection rules here are ΔΛ = 0, ±1, ΔS = 0, ± ↔ ±, g ↔ u ; the latter of
these applies only when both atoms comprising the molecule are the same—i.e. for
homonuclear diatomics. When the two atoms are different—heteronuclear diatomics—
the g and u labels are no longer meaningful.

2.4 Orbitals and bonding in molecules

Our consideration of diatomic molecules serves to introduce a feature of great
significance in the electronic properties and spectroscopy of larger molecules—for

2-10
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 2.2. (a) Molecular orbitals and energies (ascending on the vertical scale) for a homonuclear diatomic
A—A, formed from the constructive and destructive addition of the electron wavefunctions for two ns orbitals,
one on each atomic centre. (b) Molecular orbitals for a hetermuclear diatomic A—B from the constructive and
destructive addition of two ns orbitals, one on each atomic centre. Inversion symmetry denoted by g/u is only
meaningful for the homonuclear case.

almost all molecules are held together by chemical bonds between neighbouring
atoms. This does not mean that the features in a molecular spectrum can in general
be attributed to individual bonds, for both electronic and nuclear motions
commonly extend beyond individual bonds, sometimes to a very substantial effect.
But certain bond structures do exhibit characteristic features—the carbonyl unit
C=O where the C and O atoms are held together by a double bond (see below) is a
prime example—and there the principles that apply to simple diatomics do lend
useful insights. Later, we shall encounter molecules whose electronic orbitals are
more substantially delocalised.
A chemical bond is formed when the atomic orbitals (AOs) on the two atom
centres overlap to form molecular orbitals (MOs). A simplistic expectation that
sharing electrons between two nuclei in a bond will lower the energy, relative to that
for the separated atoms, which explains why most molecules are stable entities.
Forming linear combinations of separate atomic orbitals can be done in two ways,
with different energies. Consider figure 2.2(a), where ns orbitals on each atomic
centre are combined to form a homonuclear diatomic molecule A—A. Similarly, in
figure 2.2(b), the ns orbitals associated with the atoms making up a heteronuclear
diatomic A—B are combined. For convenience we take the horizontal axis between
the atoms to be the bond axis. In both cases, adding the orbitals constructively (that
is, combining them with the same +/− phase) produces significant electron density
between the atoms. This is a bonding MO and is labelled nsσg for A—A and nsσ for
A—B. Accordingly, these are termed sigma bonds. Recall that the g/u labels apply if
the two atoms making up the diatomic are the same. Otherwise, those labels are
omitted. The shared electron density between the atoms lowers the energy, as
expected.
When the orbitals are added destructively however, (i.e. with opposite phase) the
resultant electron density between the atoms necessarily exhibits a nodal plane that

2-11
 Molecular Photophysics and Spectroscopy (Second Edition)

bisects the bond axis. This is not conducive for bonding; it signifies that electrons
cannot move freely between the atoms. The resultant MO is in fact higher in energy
than that for the separated atoms, and is therefore termed anti-bonding. These states
are labelled nsσu* and nsσ* for A—A and A—B, respectively where the asterisk
superscript * denotes an anti-bonding configuration. The main difference for a
heteronuclear diatomic, compared to the homonuclear case, is that the relative
energies of the ns orbitals are different and therefore the shapes of the resultant MOs
are not symmetrical. As drawn in figure 2.2(b) the bonding MO has more B-atom
character, while the anti-bonding MO has more A-atom character—reflecting the
higher AO energy of the free atom A. An example of a molecule which only has
σ bonds is tetrahedrally-shaped methane, CH4 (figure 2.3). The sigma bond electron
density is localised between the C and H atoms in each C–H bond.
These principles can be extended to MOs formed from combinations of other
atomic orbitals having different n and ℓ. For example, in figure 2.4(a) a bonding
orbital is formed by constructively adding the 2pz orbitals on, say, two atoms to
form a 2pσg molecular orbital (it is conventional to assign the z-axis to the
internuclear bond). The other combination forms a 2pσu* anti-bonding orbital
with a nodal plane between the two atoms, as in figure 2.4(b).
Since there are three p-orbitals for every n ⩾ 2, MOs can also be formed using the
npx and/or npy orbitals on each atom. In figure 2.4(c) shows constructive formation
of electron density from either the 2px and or 2py atomic orbitals, on each side of the
bond axis (above and below, as shown) but not directly between the atoms. This MO
is represented as 2pπu; it has a bonding nature, termed a pi bond, though it is less
strong than 2pσg. The destructive combination in figure 2.4(d) with a nodal plane
between the atoms is anti-bonding in nature: the resultant MO is designated as
2pπg*.
A double bond is formed when the p-orbitals form one σ-bond and one π-bond.
The remaining atomic p-orbital not involved in molecular bonding is called a non-
bonding orbital, denoted by n. (This is the standard notation for a type of MO,
contrasting with σ and π; beware confusion with the principal quantum number n.)

Figure 2.3. Representations of methane. (a) Bonding structure in a conventional organic chemistry 3D
representation (dashed line indicating a bond heading ‘backwards’ relative to the plane of the page, the
thickened line a ‘forwards’ bond relative to that same plane, the straight line a bond in that plane); (b) atom
positioning, carbon at the centre of a regular tetrahedron of hydrogen atoms (bonds not shown).

2-12
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 2.4. (a) 2pσg bonding molecular orbital formed from two 2pz atomic orbitals; (b) 2pσu* anti-bonding
molecular orbital formed from 2pz orbitals; (c) 2pπu bonding molecular orbital from either two 2px or two 2py
atomic orbitals; (d) 2pπg* anti-bonding orbital formed from either two 2px or two 2py atomic orbitals. The red
dashed line indicates the anti-bonding nodal plane between the atoms.

Figure 2.5. Conventional depiction of methanal, the pair of parallel lines denoting a double bond.

Similarly, p-orbitals forming one σ-bond and two π-bonds yield a triple bond.
Double and triple bonds are commonly formed in many organic molecules.
An example of a planar molecule with a double bond, shown in figure 2.5, is
methanal (CH2O; informally known as formaldehyde). This molecule has a double
bond composed of one σ-bond and one π-bond located between the C and O atoms.
The π-system does not extend to the C–H bonds. While sigma and pi bonds can be
formed from combinations of s, p and d atomic orbitals, the combinations described
above are the most common in organic molecules—which represent by far the
largest number of molecules we encounter in the everyday world.
We can now begin to appreciate some of the key elements of electronic
spectroscopy, involving the absorption of photons in the UV/visible region.
Assuming selection rules allow (a matter we pursue in the next chapter), it is usually
the transition which bridges the smallest energy gap that are the most probable. As
shown in figure 2.6, the strongest electronic transitions in organic molecules will take
place from the highest occupied molecular orbital (HOMO; usually the π-system if
there are multiple bonds) to the lowest unoccupied molecular orbital (LUMO) which

2-13
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 2.6. Energy level diagram (energy ascending on the vertical scale) for a generic organic molecule,
indicating the possible types of electronic transitions, and their relative energy ordering.

are typically anti-bonding in nature (σ* or π*). However, the type of transitions
observed will depend on the exciting light source. In general the energies E of the most
common transitions are ordered: E(π* ← n) < E(π* ← π) < E(σ* ← n) < E(σ* ← σ).
In terms of the absorption wavelength, σ* ← σ and σ* ← n transitions are typically
strongest for λ < 200 nm, while π* ← π and π* ← n dominate in the UV/visible range.

2-14
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 3
Polyatomic molecules: orbitals, symmetry and
group theory

3.1 Symmetry elements

The strong connection that we have already seen between angular momentum,
symmetry and transition selection rules is just as evident when we move to
polyatomic molecules—those comprising three or more atoms. Principles of this
kind are powerfully important in determining spectroscopic and photophysical
properties. It may seem paradoxical but, in comparison to completely spherical
atoms, we shall see how the lower symmetry of most molecules—including diatomic
molecules—means that many additional elements of symmetry come into play.
The term labels for polyatomic molecules are determined by the symmetry
properties of both the molecule and its states, which is the bailiwick of point group
theory. The intent here is not to cover this topic in copious detail; many excellent
textbooks on this specific subject are well geared to students of the physical sciences.
Instead, enough details are presented to allow one to interpret the resultant term
labels, to be able to read a character table, and understand how the selection rules
for electronic and other transitions can be derived.
A useful starting point is to consider the concept of a symmetry operation. In the
current context this can be defined as a geometric operation that, when applied to a
three-dimensional molecule, leaves its resultant position and orientation indistin-
guishable from those before the operation was carried out. A symmetry element is a
geometric entity (point, line, plane) with respect to which one or more symmetry
operations may be carried out.
There are five types of 3-D symmetry operations. The first is the trivial
‘do-nothing’. This identity operation is represented as E, and while it may seem
irrelevant its inclusion is a necessary requirement of group theory. Another
operation is reflection through a plane, which has the symbol σ. The plane must
pass through and not be outside the molecular framework. Some molecules contain

doi:10.1088/978-0-7503-3683-3ch3 3-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

an inversion centre, i. As the name suggests the inversion operation involves

inverting the positions of all the atoms in the molecule through the molecular
centre; that is, change the coordinate of each atom in the molecule from (x, y, z) to
(−x, −y, −z).
There are two types of rotation operations about an internal axis. The first, which
involves rotating the molecule through an angle that renders it indistinguishable
from its original configuration, is called a proper rotation. It has the label Cn which
means ‘carry out a rotation through an angle 2π/n (radians)’. If the molecule is
linear, then any rotation angle about the molecular axis yields an indistinguishable
configuration. These axes are labelled C∞.
As an example, consider a C3-axis. There are three operations associated with this
symmetry element. The first is E, formally equivalent to C1, corresponding to a
rotation of 2π or 360° that brings the system back to its original configuration. The
next operation is C13, which is a single clockwise rotation of 2π/3 = 120°. The third
operation is C32 , which corresponds to a rotation of 4π/3 = 240°. However, the last
operation could equally have been achieved by a counterclockwise rotation of 2π/3
about the C3-axis, labelled as C−3 1 where the negative sign indicates rotation in the
opposite direction. It follows that C−3 1 ≡ C32 . Care must therefore be taken to ensure
that one does not overcount the unique number of symmetry operations associated
with each symmetry element. Another example is a C2, a ‘binary’ rotation
corresponding to 180°. A second C2 rotation will return the molecule to its original
configuration.
When more than one axis of rotational symmetry exists, the one with the largest n
value is called the principal axis. If n is odd, the existence of one C2-axis
perpendicular to Cn implies n–1 additional C2-axes labelled by a prime superscript;
that is, C2’. Such molecules will also have n vertical planes containing Cn which are
labelled σν. The horizontal plane of a molecule which is perpendicular to Cn, if it
exists, is labelled σh. In sufficiently large molecules, a distinction can be drawn
between differently positioned C2-axes perpendicular to the principal axis; if one set
passes through atoms while another does not, they are distinguished by labels C2’
and C2’ respectively. Similarly, there may be two sets of vertical planes containing
the principal axis. By convention, those planes which contain the atoms are again
called vertical, σν, while those that pass between atoms are called dihedral, σd.
The second type of rotation is called an improper rotation, represented as Sn. This
is a composite of two operations: a Cn rotation followed by reflection through the
plane σh perpendicular to that Cn axis. It should be noted that there is nothing
improper about such a symmetry operation at all despite its name! Clearly, Sn exists
if Cn and σh exist, but Sn can also exist is Cn and σh do not.
An example of a molecule which exhibits all these symmetry elements is the
molecule benzene, C6H6, shown in figure 3.1. Benzene is a planar molecule with a
hexagonal carbon framework. One hydrogen atom is bonded to each C-atom and
each C–H bond lies in the molecular plane. By convention of organic chemistry, the
H-atoms are usually not explicitly shown for benzene and many larger carbon-based
molecules, for simplicity’s sake. Benzene is an example of a molecule which exhibits

3-2
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 3.1. Resonance structures of benzene.

Figure 3.2. The left-hand depiction of benzene shows six p-orbitals, one per carbon atom, which combine to
form the molecular orbital shown on the right, with electron density above and below the ring plane.

resonance, in a specific sense unconnected to the spectroscopic concept of matching

optical frequencies. The figure on the left-hand side of figure 3.1 with alternating
single and double bonds is equivalent to the figure in the middle where the locations
of the double and single bonds have moved by π/3. But in either such form the single
and double bonds would have distinctly different lengths; in fact all of the C–C
bonds are identical. Thus, the molecule is often drawn with an imbedded circle
(right-hand side) to indicate this equivalence—which results from electron delocal-
isation around the ring.
A more detailed way of understanding this delocalisation feature is to recognise
that on each carbon, one p-orbital oriented perpendicular to the plane of the benzene
ring is not involved in the bonding that holds together the framework of the
molecule. As indicated in figure 3.2, in the ground state six p-orbitals (one per
carbon atom) overlap to form a molecular orbital with a ring-like electron density
above and below the plane of the molecule. The symmetry elements associated with
this molecule are shown in figure 3.3. Associated with these symmetry elements are
the following symmetry operations: 2C6, 2C3 (= C62 and C64 = C6−2),
C2 (= C63 = C6−3), 3C2’, 3C2’, i, 2S3, 2S6, σh, 3σd, 3σν. Note, the obvious S2
operation is in fact equivalent to inversion, i.

3.2 Point groups and operations: Schoenflies notation

The product, OiOj, of any two symmetry operations {Oi, Oj} associated with a
molecule will itself correspond to a symmetry element associated with the molecule,
Ok. The complete set of symmetry operations associated with the symmetry elements
described above, including E generated from all possible products OiOj, constitutes
the point group of the molecule. Note that the order of the multiplication can make
a difference; that is, OiOj may or may not equate to OjOi; that is, the symmetry
operations may not commute. However, if they do commute, such groups are called

3-3
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 3.3. Symmetry elements of the benzene molecule. The C6- and S6-axes coincide and are coming out of
the horizontal plane of the figure indicated by the green box (σh). The red lines which correspond to the three
C2’ axes perpendicular to C6 also indicate the edges of three σν planes coming out of the page, containing C6
and one of the C2’ axes. The blue lines correspond to three C2’ axes perpendicular to C6, and the edges of the
three σd planes coming out of the page, containing the C6 and one of the C2’ axes.

abelian. As an example, table 3.1 shows products of the symmetry operations for an
equilateral triangle (depicted in figure 3.4), i.e. {E, C31, C3−1, σA, σB, σC}.
It is important to note here that this group exemplifies four key properties that
apply to all proper groups, namely that they must satisfy the following:
1. Closure. No new symmetry elements are generated through multiplication.
2. The group contains E.
3. The elements are associative; that is (OiOj)Ok = Oi(OjOk).
4. Each element has an inverse such that the product of Oi and its inverse
yields E.

Each row and column of a group multiplication table contains every operation once,
similar to a Sudoku puzzle. Given this information, there are several algorithms one
can use to establish which unique combination of symmetry elements, i.e. which
point group a molecule belongs to. These are named using Schoenflies notation, and
are presented in table 3.2. The term ‘point group’ indicates that only the relative
point positions of atoms within the molecule are significant. For crystalline solids,
whose symmetry additionally involves the relative spatial translations of unit cells,
space groups are employed. For regular crystals (a great many of which are of
atomic or ionic rather than molecular constitution) there are 230 distinct types of
space group; their study is beyond out present remit, where we are primarily
concerned with the optical properties of individual molecules.

3.3 Matrix representations and character tables

A matrix O may be defined as a rectangular array of elements {oij} where element oij
is found in row i and column j. The mathematics of matrices constitutes a major part

3-4
 Table 3.1. Multiplication table for the group symmetry elements of an equilateral triangle.

Equilateral triangle group E C31 C3−1 σA σB σC

1 1
E EE = E EC3 = C3 EC3−1 = C3−1 EσA = σA EσB = σB Eσc = σc
C31

3-5
C31E = C31 C31C31 = C3−1 C31C3−1 = E C31σA = σC C31σB = σA C31σC = σB
C3–1 C3−1E = C3−1 C3−1C31 = E C3−1C3−1 = C31 C3−1σA = σB C3−1σB = σC C3−1σC = σA
σA σA E = σA σAC31 = σB σAC3−1 = σC σA σA = E σA σB = C31 σA σC = C3−1
σB σ BE = σ B σBC31 = σC σBC3−1 = σA σB σA = C3−1 σB σB = E σB σC = C31
σC σC E = σC σCC31 = σA σCC3−1 = σB σC σA = C31 σC σB = C3−1 σC σC = E
Molecular Photophysics and Spectroscopy (Second Edition)
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 3.4. Symmetry elements of an equilateral triangle with its C3-axis in the z-direction out of the plane of
the page. The red lines denote the vertical planes containing C3.

of the field of linear algebra, but it is sufficient here simply to note that matrices of
the correct dimensions can be added, subtracted and multiplied. Many square
matrices, but not all, have an inverse O−1 such that the product inverse
O−1O = OO−1 = I, where I is the Identity matrix whose diagonal elements equal
1 and all others are 0. The only definition we need to introduce is the character of
matrix, χ, which is the sum of diagonal elements: χ = ∑i oii . Another common name
for this character is the trace; in group theory the former designation is more
prevalent.
Matrices are introduced because symmetry operations can be written in matrix
form. A logical basis set to construct a matrix representation of a symmetry
operation Γ(O) comprises the unit vectors pointing in the x-, y-, and z-directions:
( )
iˆ, jˆ, kˆ , for example:
⎛1 0 0 ⎞ ⎛− 1 0 0 ⎞
Γ(E) = ⎜⎜ 0 1 0 ⎟⎟ ; Γ(i) = ⎜⎜ 0 − 1 0 ⎟⎟ .
⎝0 0 1 ⎠ ⎝ 0 0 − 1⎠
On the left here is the 3 × 3 identity matrix, and on the right the matrix for spatial
inversion; their characters are 3 and −3, respectively. Matrices for all other
operations can be similarly derived.
One of the powerful aspects of group theory is that unit vectors are not the only
basis functions that can be used to derive matrix representations of symmetry
operations. For example, one might use as bases the directions of the bonds in a
molecule, the inter-bond angles, the orbitals on each atom, or so on, depending on
the type of problem being considered. Some of these choices could in principle lead
to very large matrix representations indeed. Consider again, for example, placing
three Cartesian unit vectors at each vertex of the equilateral triangle shown in
figure 2.4. The dimension of each resultant matrix representation, describing how all
9 of those vectors ‘move’ at the same time under a symmetry operation, would
already be 9 × 9.

3-6
 Molecular Photophysics and Spectroscopy (Second Edition)

Table 3.2. Major point groups collected into categories, and the symmetry operations needful for each. Point
groups labelled ‘uncommon’ are not representative of any common molecule.

Point group Minimum symmetry Notes

Low symmetry groups

C1 E
CS E, σ
Ci E, i

Rotational C groups
Cn E, Cn Abelian
Cnh E, Cn, σh Even n: operations include E, Cn and
its repetition, σh, i, various Sn.
Odd n: operations include E, Cn and
its repetition, σh, various Sn.
Cnν E, Cn, nσν
Dihedral D groups
Dn E, Cn, nC2’
Dnh E, Cn, nC2’, σh, nσν Even n: the nσ planes divide into one
set of n/2 σν planes and one set of
n/2 σd planes.
Dnd E, Cn, nC2’, nσd Odd n: an inversion operation i will
exist.
For all n, several S2n symmetry
operations are possible.
Linear point groups
C∞v E, 2C∞, ∞σν
D∞h E, 2C∞, ∞σν, σh, ∞C2’, i
S groups
Sn E, Sn Uncommon
Cubic groups
Td E, 8C3, 3C2, 6S4, 6σd Tetrahedral point group
Oh E, 8C3, 6C2, 6C4, 3C2, i, 6S4, 8S6, Octahedral point group
3σh, 6σd
Ih E, 12C5, 12C52, 20C3 15C2, i, 12S101, Icosahedral point group
12S103, 20S6, 15σd
Additional cubic groups
T E, 4C3, 4C32, 3 C2 Uncommon
Th E, 4C3, 4C32, 3 C2, i, 4S6, 4S65, 3σh Uncommon
O E, 8C3, 6C2, 6C4, 3C2 = C42 Uncommon
I E, 12C5, 12C52, 20C3 15C2 Uncommon

3-7
 Molecular Photophysics and Spectroscopy (Second Edition)

The set of matrices for all of the symmetry operations applicable to a specific
molecule, regardless of their dimensions, also form a mathematical group. It can
hugely simplify applications if the matrix representation is expressed in terms of the
smallest dimensional matrix representations possible. These are known as irreducible
representations (IRs), because they can be reduced no further; any other representa-
tion is said to be reducible. The irreducible representations are 1 × 1 (which means its
character is the matrix itself!), 2 × 2 or 3 × 3, but no larger—except in the very rare
case of icosahedral symmetry. We can simplify further by asking how the characters
χ of these matrix IRs behave under symmetry operations. The advantage of
considering characters over the any explicit matrix representation is they are matrix
invariants; the individual matrix elements may vary according to the choice of
coordinate system, but the characters do not. The labels of the IRs are called
Mulliken symbols:
• A one-dimensional IR is labelled A if it is symmetric under a Cn rotation; that
is, χ(Cn) = +1. Otherwise, if χ(Cn) = −1 the IR is labelled B.
• Subscripts 1 and 2 are used depending on whether the IR is symmetric or anti-
symmetric, χ = +1 or −1, respectively, under any C2’ rotation or σν reflection.
• Prime or double prime superscripts indicate whether the IR is symmetric or
anti-symmetric under any σh reflection.
• g and u subscripts, (introduced in chapter 2), indicate whether the IR is
symmetric or anti-symmetric with respect to inversion, i, if an inversion centre
exists.
• Two-dimensional and three-dimensional representations are labelled E and T,
respectively.

In the last of these rules, the labels E and T should not be confused with the
‘do-nothing’ symmetry operation E, or with T groups. In each of these cases,
ancillary labels (primes, subscripts etc) may also be applied, just as with the A and B
representations.
Linear systems are often described with an alternative set of primary symbols.
One-dimensional A representations are in this case labelled Σ, and there is an
unlimited range of two-dimensional (i.e. doubly degenerate) E representations, now
labelled Π, Δ, Φ etc. The virtue of this system is that when applied to linear
molecules such as diatomics, these designations correspond with the magnitude of
the z-component of the angular momentum along the symmetry axis, Λ. As we
observed in section 2.3, terms labelled Σ, Π, Δ, and Φ signify Λ = 0, 1, 2, and 3.
However, the same Mulliken subscripts and superscripts as those described above
still apply, if appropriate.
Let us look in detail at a more typical example, the character table for the
C3v point group, shown in table 3.3. The ammonia molecule, NH3, is a common
example of this kind. The left-hand column of the table lists all the IRs associated
with the point group specified in the upper left-hand corner. The transformation of
the characters for each class of operations is listed in the right adjacent columns.
Commonly such tables include additional information on the right-hand side. Here,
for example, we are told for example how components of x, y and z vectors transform.

3-8
 Molecular Photophysics and Spectroscopy (Second Edition)

Table 3.3. Character table for the Schoenflies point group C3ν.

Linear functions,
C3v E 2C3(z) 3σν rotations Quadratic functions Cubic functions

A1 +1 +1 +1 z x2 + y2 , z 2 z 3, x(x 2 − 3y 2 ), z (x 2 + y 2 )
A2 +1 +1 −1 Rz − y(3x 2 − y 2 )
E +2 −1 0 (x , y ) (Rx , Ry) (x 2 − y 2 , xy ) (xz, yz ) (xz2, yz2 )[xyz , z (x 2 − y 2 )]
[x(x 2 + y 2 ), y(x 2 + y 2 )]

Figure 3.5. Rotations about the three Cartesian axes. In the point group C3v, Rz transforms as A2 while
(Rx, RY) are doubly degenerate and transform as E.

In C3v, z transforms as A1, while (x, y) form a doubly degenerate pair and transform
as E. This identification is useful for understanding the behaviour of the components
of the dipole moment, or p-orbitals, in this symmetry environment. The character
table shown also tells us how quadratic functions and cubic functions transform. The
former information is useful for d-orbitals while the latter describes the symmetry
properties of f-orbitals. It should be noted that s-orbitals have no directional
dependence and therefore they always transform under the totally symmetric IR;
for example, A1 for a C3v molecule.As will be discussed later, the quadratic functions
also describe the symmetry properties of the polarizability of a molecular system,
which is important for understanding Raman spectroscopy. Furthermore, the
character table always lists the symmetry behaviour of rotations about the x-, y-,
and z-axes (Rx, Ry, and Rz respectively), as shown in figure 3.5. A rotation vector will
be symmetric if, under a symmetry operation, its sense of rotation remains the same,
but anti-symmetric if it changes between left and right-handed.
To conclude our excursion into symmetry principles, we can now identify their
principal application in the realm of molecular spectroscopy, where they afford the
basis of selection rules for polyatomic molecules. As we established earlier, the most
common form of electromagnetic interaction between light and matter takes place
through electric dipole coupling. In chapter 2, equation (2.4), we saw that the key

3-9
 Molecular Photophysics and Spectroscopy (Second Edition)

requirement for an allowed transition in hydrogen is the following condition on the

spatial wavefunctions for the initial and final states:

∫ ψn′⁎′,ℓ′, m ′ (r )μψn, ℓ, m (r )dr

ℓ ℓ
≠0

This represents just one specific instance of a more general rule, expressible as;

∫ ψ ′* μψdτ ≠ 0, (3.1)

where the dipole operator μ sits between an initial wavefunction and the complex
conjugate of the final state wavefunction, in an integral over a generalised space τ —the
latter including every degree of freedom that undergoes change in the transition. For the
electronic transitions of hydrogen this was a simple integral over the three spatial
coordinates of the electron; for electronic and other transitions in polyatomic molecules
it is less straightforward. But here is where symmetry comes powerfully into play.
To ascertain whether the transition integral shown above may be non-zero, one
constructs the direct product Γ(ψ’)⊗Γ(μ)⊗ Γ(ψ) by multiplying the characters in each
column of the character table for the IRs involved. The resultant character set may
be irreducible (if it tallies with one of the representations listed in the table) or
reducible—in which case it can be reduced to a linear combination of IRs. In either
case the transition is only allowed if the direct product equals or contains the totally
symmetric representation. More simply, and equivalently, a transition will be
allowed if the direct product Γ(ψ’)⊗ Γ(ψ) ⊆ Γ(μ) = Γ(x, y, z). It should be noted
that since the dipole moment may have components represented by more than one
IR it follows that a transition between two states in a polyatomic molecule may be
allowed by some components and forbidden for others. We shall see the imple-
mentation of these principles, and many important conclusions that can be drawn
from it, in many of the chapters that follow.

3-10
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 4
Electronic and nuclear energy levels in molecules

4.1 The separation of electronic and nuclear motions

Most molecules conform to the Born–Oppenheimer principle, whose basis is the
much greater mass, and correspondingly slower motions of nuclei, compared to
electrons. A typical time scale for electronic transitions is ⩽10−15 s, while transitions
involving nuclear motion are at least 1000 times slower. This means that, in general,
we can to a good approximation separate a full molecular wavefunction Φ into
electronic and nuclear parts. So, we write:

Φ(q , Q ) ≈ ψ elec(q , [Q ])Ψ nuc(Q ) (4.1)

where q and Q are schematic labels indicating all the electron coordinates, and all
the nuclear coordinates, respectively. The square brackets around Q shown in the
parameters of the electronic wavefunction indicate an indirect, parametric depend-
ence on Q; physically this reflects the fact that, to a very good approximation, the
electrons in a molecule respond instantaneously to any nuclear motion. Thus, in
each electronic state the electrons move on the potential energy surface, V(Q),
formed by the nuclear motion in that electronic state. The nuclear wavefunction
itself factorises into two parts: vibrational and rotational. As we shall see, such a
factorisation of the full molecular wavefunction also enables the corresponding
identification of separate energy terms.
The wavefunctions in quantum terminology are all eigenstates of the electronic,
vibrational, and rotational parts of the molecular Hamiltonian, derived using the
Born–Oppenheimer approximation. This in turn means they are stationary states,
i.e. stable states whose measurable properties do not change with time. All this
changes of course when light (indeed, any electromagnetic field) is present, so
allowing transitions to take place—which is the fundamental basis of spectroscopy.
In principle there are other energies relating to whole-molecule linear motion
(translation) through space, and nuclear spin. The translational levels usually merge

doi:10.1088/978-0-7503-3683-3ch4 4-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

into a quasi-continuum, so that we cannot observe quantum transitions between

them. Although, on the other hand, quantum transitions are indeed observable
between nuclear spin levels, the corresponding energy separations are very small,
compared to electronic, vibrational or rotational energies—and the observation of
transitions also depends on the presence of a magnetic field. This principle is of
course widely exploited in nuclear magnetic resonance (NMR) spectroscopy.
However, except for some effects that manifest themselves in the high-resolution
analysis of rotational motions, which we shall encounter in chapter 5, nuclear spin is
not a dominant factor in optical spectroscopy.

4.2 Types of nuclear motions and degrees of freedom

Corresponding to the factorisation of the molecular wavefunction in equation (4.1),
a fairly accurate picture of molecular energies now emerges as a sum of three terms:
E = Eelec + E vib + E rot . (4.2)

The three contributions on the right of this equation relate to electronic, vibrational,
and rotational energies, respectively, of successively diminishing magnitude.
In terms of energy, the lowest electronic levels are usually very far apart, and
typically the frequencies for transitions between them lie in the visible or the
ultraviolet part of the spectrum. We can readily verify that at room temperature
almost every molecule is its electronic ground state; that is, the Boltzmann population
factors for electronically excited states are close to zero. For example, we can do the
calculation for a typical carbonyl group based on its π–π* transition. The correspond-
ing absorption typically occurs at an ultraviolet wavelength λ of around 200 nm, so we
can calculate exp(−ΔE/kBT) = exp(−hc/kBTλ), where kB is Boltzmann’s constant and
T is the absolute temperature. With kB = 1.381 × 10−23 J K−1, c = 2.998 × 108 m s−1
and h = 6.626 × 10−34 J s, then at T = 298 K we find the result of around 10−105, in
other words no molecules will be in the electronic excited state at room temperature—
in any volume of the world where the concept of ‘room’ could have meaning.
Numerous vibrational levels, with much closer spacing, are associated with each
electronic level. Since the corresponding energy gaps ΔE between adjacent levels are
that much smaller, so too are the frequencies at which vibrational transitions are seen.
The correspondingly longer wavelengths involved in these transitions generally lie in
the infrared part of the electromagnetic spectrum. We can calculate for example that
for the wavenumber ν = 1700 cm−1 we have exp(−ΔE/kBT) = exp(−hc ν /kBT), giving
a result of 2.73 × 10−4. In other words, about one molecule in three or four thousand
will be in its first vibrationally excited state at room temperature. This is still usually a
small enough number to regard as insignificant. So basically, it is safe to assume that,
at room temperature, a molecule undergoing an absorption transition starts from the
ground vibrational level in the ground electronic state. Equally, emission spectra
(associated with excited state decay) will not be observable without prior excitation.
Finally, we have rotational levels, and these are still more tightly packed together:
rotational transitions are therefore observed at yet lower frequencies and longer

4-2
 Molecular Photophysics and Spectroscopy (Second Edition)

wavelengths, in the far infrared or microwave region. For a diatomic molecule, for
example, figure 4.1 illustrates the varying spacings of the different types of
energy level.
The fact that the rotational levels are separated by the smallest amounts of energy
has two immediate consequences. First it is clear that at normal room temperatures,
thermal energies (typically corresponding to wavenumbers of magnitude kBT/hc ~
200 cm−1) will greatly exceed the rotational level spacing, which are typically
~1 cm−1 or less, unless these small molecules contain the exceptionally light atom
hydrogen. Consequently, there will usually be a substantial population of numerous
rotational states, not just the lowest one. But secondly, unlike internal electronic and
vibrational motions, rotations can be frustrated in the condensed phase by the steric
effect of neighbouring molecules, so giving rise to a heterogeneous broadening of the
energy levels. Hence, complete energy level quantisation based on free rotation can
only occur in a rarefied medium, and the observation and investigation of rotational
transitions is essentially a gas-phase technique.
For any given molecule, each electronic state of the molecule is usually identified
by some label, often based on the molecular point group and state symmetries, as
described in chapters 2 and 3. For every such state there is an equilibrium nuclear
geometry about which vibrations occur, signifying oscillatory motions about a
configuration of minimum potential energy. The form of each vibration is controlled
by restoring forces associated with the shape of each potential energy surface—and
the detailed form of the latter depends on the electronic state. Thus, each electronic
state has its own set of vibrational states. So too, each vibrational level has its own
set of rotational states.

Figure 4.1. Typical plots of potential energy V against the interatomic distance (bond length) r for a diatomic
molecule, schematically showing (not to scale) the positioning of the quantised electronic, vibrational and
rotational energy levels.

4-3
 Molecular Photophysics and Spectroscopy (Second Edition)

Considering in more detail the nuclear vibrations, it is clear that a diatomic

molecule has only one kind: the oscillatory stretching and compression of its single
bond. However, for polyatomic molecules we need to account for the fact that there
is more than one type of vibration that can take place. First let us see how many
distinct vibrations we may expect.
The instantaneous position of each atom in a polyatomic molecule can in
principle be specified by three Cartesian coordinates: an N-atom molecule therefore
requires 3N coordinates to specify all the atomic positions. We say that there are 3N
degrees of freedom associated with the molecule. In fact, these need not be Cartesian
coordinates—what matters is that 3N distinct pieces of information are required (to
be mathematically precise, these data are linearly independent). We can choose to
have some of this information expressed in terms of angles, for example.
Some of the degrees of freedom can be used to specify inter-nuclear distances—
this is what we shall be interested in when we look at vibrations. However, three of
the coordinates can be used to specify the position of the centre of mass; changes in
these coordinates will tell us about the motion of the molecule as a whole through
space. So, we say there are three translational degrees of freedom. Rotational degrees
of freedom are also accounted for when we specify the orientation of the molecule in
space; changes in these coordinates tell us about the rotation of the molecule.
Consider, for example, the molecule CO2, whose spatial translations and rotations
are illustrated in figure 4.2.
Evidently there are two rotational degrees of freedom for any such linear molecule.
We disregard on-axis rotations (Rx), essentially because they could involve only the
off-axis mass of the electron clouds around each nucleus—and all rotations of the
electrons are already accounted for in the corresponding electronic wavefunctions.
Generally, we shall have a remaining (3N − 5) degrees of freedom for the linear case,
which correspond to the number of internal vibrational motions. This is the number
of coordinates left to describe the internal configuration of atoms in the molecule,
and it is the changes in these intramolecular coordinates that must therefore tell us
about nuclear vibrations.

Figure 4.2. Independent translational and rotational motions of the linear carbon dioxide molecule.

4-4
 Molecular Photophysics and Spectroscopy (Second Edition)

Before considering the form of those vibrations, let us consider another triatomic
molecule, H2O, that is not linear. As shown in figure 4.3, here there are three
translational and also three rotational degrees of freedom for this (and any) non-
linear molecule.
Thus, for any non-linear molecule—we have six of our 3N coordinates accounted
for, leaving (3N − 6) degrees of freedom which correspond to the vibrational modes
of the molecule. In summary, we have:
• (3N − 5) vibrational degrees of freedom for a linear molecule:
• (3N − 6) vibrational degrees of freedom for a non-linear molecule.

Now it turns out that, just like the translations and rotations, there are certain
specific vibrations in terms of which all others can be expressed. These are known as
the normal modes of vibration. Just as a molecule moving in the x-direction cannot
begin to move in the y- or z-direction without some external force being applied, so
too a molecule oscillating in one of its normal modes cannot begin oscillating in a
different mode without being excited in some way. Moreover, each of these modes of
vibration is close to being simple harmonic in form—at least for the lowest energy
quantum levels (see chapter 5).
So, to summarise: the nuclear motion part of the molecular wavefunction
factorises into a product of rotational and wavefunctions for each of the 3N − 6
normal modes of vibration (3N − 5 if the molecule is linear), so that we effectively
now have:
3N − 6(5)
Φ(q , Q ) ≈ ψmelec(q , [Q ])ψ rot(Q ) ∏ ψvvib(
i
m)
(Qi ), (4.3)
i

where the capital pi, Π, signifies a product of the nuclear vibrational wavefunctions
for each normal mode, indicated by the index i.

Figure 4.3. Independent translational and rotational motions of the non-linear water molecule.

4-5
 Molecular Photophysics and Spectroscopy (Second Edition)

4.3 How far do the atoms move?

We understand from quantum mechanics that for a molecule in any particular
electronic state, the outermost electrons are in rapid motion within the broad extent
of the associated wavefunctions (the latter essentially signifying the space occupied
by the molecule, or a that of a particular molecular component). Although it is
understood that modes of nuclear vibration are also quantised and therefore also
have formal wavefunctions, the actual extent of the associated atomic motions are
much less often fully appreciated. Of course, the quantum nature of particles on this
scale means that one can only gauge an approximate extent; there is no absolute
amplitude of vibration.
Let us consider how to gauge the typical extent of the nuclear vibrational motion.
Take for example the well-known absorption at around 1700 cm−1, which character-
ises the stretching mode of carbonyl groups in a great many organic compounds; we
first encountered its electronic properties in chapter 2. Just how much does the C–O
bond vary in length, during such a vibration? For simplicity, let us suppose that most
of the motion is in the terminal oxygen atom: motion of the carbon will be tempered
by the anchorage of this atom to the rest of the molecule, which is usually very much
more massive. If there were no motion at all in the carbon, the natural frequency of
carbonyl vibration would be determined by just two parameters: the bond force
constant k, and a reduced mass that would simply equate to the mass m of the
oxygen atom. The circular frequency of vibration is then given by ω = k /m . (In
passing, it should be said that this notion of a group frequency, meaning the
vibration frequency for just one chemically meaningful part of a molecule. is
potentially misleading; its relation to normal modes of vibration will be addressed
in chapter 7.)
To derive the equations that we shall need, consider a general case concerning a
1
particular vibration with circular frequency ω, and ground state energy 2 ℏω (chapter 6).
Now it is a feature of simple harmonic motion that the mean kinetic energy T and
the mean potential energy V are equal. (This is a consequence of the Virial
Theorem: when a potential energy varies with the nth power of an inter-particle
distance, then T = (n /2) V . For simple harmonic motion, n = 2.) Here, then, each
average has a value of T = V = 14 ℏω. Although we recognise that quantum
theory allows vibrational compressions and extensions beyond the expected classical
limits, a classical interpretation of the potential energy will give us an estimate
(rather than a definite result) for the most likely extent of motion, which we can
express as:
1
V = 2 mω 2 x 2 ∴ x 2 = 2 V / mω 2 = ℏ/2mω

where x signifies the extent of displacement from the equilibrium configuration (at
the bottom of the quadratic potential well). Also, ω = 2πcν , where ν is the
wavenumber of the vibration, and ℏ = h/2π : hence, we find;
∴ x 2 = ℏ/4πmcν = h /8π 2mcν .

4-6
 Molecular Photophysics and Spectroscopy (Second Edition)

Exercise
Using the values for h and c and the data m(O) = 16.00 amu (1 amu = 1.6606 × 10−27 kg)
and ν = 1700 cm−1, evaluate <x2> for the carbonyl bond. (This establishes the
variance, since <x> = 0.)

Answer
The result, 6.197 × 10−24 m2, has a square root which signifies a mean (rms)
variation in bond length of about 2.49 pm. Compare this with the reported
equilibrium bond length which is around 123 pm. The fractional change, almost
exactly a fiftieth, is a fairly typical result.
So it is that in molecular vibrations, the atoms don’t generally move very much,
and therefore the Born–Oppenheimer approximation, separating electronic from
nuclear motions, works well.
We still have to remember that in most regular samples, we are not seeing the
behaviour of any individual molecule in isolation. Whether the molecules are held
together immobile in the static structure of a solid, or constantly buffeted by other
molecules in the mobile surroundings of a liquid, the proximity of those other
surrounding molecules slightly modifies the electronic environment of each mole-
cule, generally contributing to a level broadening. Taking account of these factors,
we can understand why it is that for most molecules the absorption of light, although
a quantum phenomenon, gives broad lines rather than the sharp, discrete lines found
in atomic spectra. As we shall see, this is a very important distinction; as a corollary,
there is a very high density of states (ρ) associated with any energy level within the
vibrational continuum of a given electronic state. (Note, the term density of states
signifies the number of quantum states to be found within a small energy interval—
this terminology has nothing to do with volume.) If we compare the successive
vibrational levels in even a simple series of increasingly large molecules, we can see—
even without taking the corresponding rotational levels into account—how the density
of states rapidly increases with the molecular size.

Exercise
Consider a molecule of formic acid HCOOH (methanoic acid), for which the
fundamental vibration wavenumbers are as given below (note the 3N − 6 rule). To
gauge the density of states, estimate how many vibrational states will generally be
found in a typical 1000 cm−1 energy interval. Note: the first 1000 cm−1 above the
ground vibrational state will not be typical, because only the single-quantum
vibrational levels for ν6 and ν 9 can feature below that threshold (table 4.1).

Answer
For formic acid, the average (1/3.570) + (1/2.943) + … = 7.76; say ~8, the nearest
integer (since one cannot have a fraction of an energy level!). However, any such
estimation ignores the huge number of higher energy levels, rapidly generated by the
so-called overtones such as 2ν6 = 2210 cm−1 or combination bands and overtones
such as ν4 + ν7 = 2012 cm−1; ν5 + 2ν7 + ν9 ≡ ν4 + ν6 + ν7 = 3117 cm−1. Even this is

4-7
 Molecular Photophysics and Spectroscopy (Second Edition)

Table 4.1. Vibrations of HCOOH. From: Shimanouchi T 1972 Tables of Molecular Vibrational
Frequencies Consolidated, Nat. Stand. Ref. Data Ser., N.B.S. (U.S.) 1, 39.

Vibration Type Wavenumber (cm−1)

ν1 OH stretch 3570
ν2 CH 2943
ν3 CO stretch 1770
ν4 CH bend 1387
ν5 OH bend 1229
ν6 CO stretch 1105
ν7 OCO deform 625
ν8 CH bend 1033
ν9 Torsion 638

Figure 4.4. Benzophenone.

without regard to Fermi resonance—an energy level splitting that occurs when two
vibration modes of the same symmetry are close in energy. Overtones and
combination bands will be discussed further in chapters 6 and 7.
Computer calculations show that the number of nuclear vibration levels rises
exponentially with N, the number of atoms in the molecule. It is clear that many
thousands of vibrational lines per 1000 cm−1 will emerge from a similar calculation
even for a relatively small molecule. For example, benzophenone, shown in
figure 4.4, has 24 atoms, properly counting all the carbon and hydrogen atoms as
well as the oxygen, and hence 66 separate individual vibrational modes—all of
which can exhibit combination bands and overtones.

4-8
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 5
Small molecule rotational energy levels
and spectra

5.1 Diatomic and linear polyatomic molecules

We shall begin with the rotational levels of a simple diatomic. Most conveniently, it
turns out that the rotational levels of any linear polyatomic molecule take the same
analytical form as those of a diatomic, so there is an expediency in dwelling on the
simplest case. Moreover, we can achieve a great deal by starting from a classical
description, before moving to a quantum mechanical analysis.
Consider a molecule with two atoms of masses m1 and m2, separated by a distance r
(the bond length). The key parameters are shown in figure 5.1. Since we are focusing
first on rotation, we shall for the present assume that the bond length is a fixed
quantity.
In general, the capacity of a rotating object to hold rotational energy is
determined by its moment of inertia, usually given the symbol I and defined as:

I= ∑mi ri2 , (5.1)

i

where the sum is taken over all component particles (in the case of molecules, the
individual atoms) of masses mi at equilibrium distances ri from the centre of mass—
the latter being the point about which rotation occurs. In the case of the diatomic
molecules, the centre of mass has a location defined such that:
m1r1 = m2r2 . (5.2)

Since r1 + r2 = r, we find:
m2 m1
r1 = r; r2 = r. (5.3)
m1 + m2 m1 + m2

doi:10.1088/978-0-7503-3683-3ch5 5-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 5.1. Key parameters for describing the rotations of a diatomic molecule.

From equation (5.1) it then follows that:

I = m1r12 + m2r22
⎛ m2 ⎞2 ⎛ m1 ⎞2
= m1⎜ ⎟ r 2 + m2⎜ ⎟ r2
⎝ m1 + m2 ⎠ ⎝ m1 + m2 ⎠ (5.4)
m m (m + m1)r 2 ⎛ m1m2 ⎞ 2
= 1 2 2 =⎜ ⎟r = mr r 2 .
(m1 + m2 )2 ⎝ m1 + m2 ⎠

Thus, the moment of inertia in this simple case reduces to the square of the bond
length multiplied by what is known as the reduced mass, mr, (note: some books use
the symbol μ), the latter defined as;
m1m2
mr = . (5.5)
m1 + m2

Now let us think about the rotation itself. The two atoms must have to move at
different speeds, ν1 and ν2, obviously proportional to their distances from the centre
of mass—or else one of them would catch up with the other one (figure 5.2). The
constant of proportionality, ω, is called the angular velocity. So, we can write
ν1/r1 = ν2/r2 = ω. It therefore follows that the kinetic energy of rotation is given by:

1 1 1 1 1 1 2
E= m1ν12 + m2ν 22 = m1r12ω 2 + m2r22ω 2 = (m1r12 + m2r22 )ω 2 = Iω . (5.6)
2 2 2 2 2 2

Note the similarity to the usual kinetic energy formula. Now the angular
momentum of the molecule is given by:

L = m1ν1r1 + m2ν2r2 = m1r12ω + m2r22ω ∴ L = Iω . (5.7)

Again, the result is like the linear motion momentum, p = mv. So classically, we can
write the rotational energy in terms of the angular momentum, using the earlier
equations (5.4) and (5.5), as follows:

L2
E= . (5.8)
2I

5-2
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 5.2. Motions of the individual atoms in the rotating diatomic.

This equation reflects an underlying connection between energy and angular

momentum which, since it reflects the basic symmetry of free rotation about a
point, remains true in a fully-fledged quantum mechanical development. The key
difference, of course, is that the energy levels are discrete in the latter case. We can in
fact take a shortcut to the quantum mechanical result without going through the
usual rigmarole of solving the Schrödinger equation. Quantum mechanics tells us
that the angular momentum itself can only take certain discrete values, here
given by:

L = ℏ J (J + 1) , J = 0,1,2,... (5.9)

where ℏ = h/2π and J is the rotational quantum number which as noted, can only
take on discrete integer values 0, 1, 2, …. The angular momentum is a signified by a
vector, directed along the axis perpendicular to the plane of rotation with the sense
of a right-handed screw, and with a magnitude L, as indicated in figure 5.3.
It follows that the energies are therefore expressible as:
E = BhJ (J + 1), (5.10)

h
B= , (5.11)
8π 2I
where B is the rotational constant of the molecule, with units of Hz. (Note that
several authors define B differently, introducing a dividing factor of c, the speed of
light, such that its units are cm−1. Others introduce another power of h such that the
units are Joules.) The first few rotational energy levels are shown in figure 5.4.
Because the spacing between rotational levels depends on B, which is inversely
proportional to the moment of inertia, the levels became more and more tightly
packed as the moment of inertia increases—this is true for non-linear molecules, too.
As derived below, the spacings between rotational levels (indicated by the arrows in
figure 5.4) are not constant but increase linearly with increasing J.
One other property of the rotational levels should be mentioned before we go any
further, namely the degeneracy. It is another feature of the quantum mechanical

5-3
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 5.3. Angular momentum vector for the rotating diatomic molecule.

Figure 5.4. First few energy levels of a perfect rotor in units of Bh (vertical arrows also showing allowed
transitions), and degeneracy factors gJ associated with each of the rotational quantum numbers J.

theory that the component of the angular momentum in a particular direction

(usually chosen as z) is also quantised, and can only take one of the following values:
Lz = M ℏ, where M = 0±1 …±J (5.12)
where M is called the azimuthal quantum number. For instance, with J = 2, the total
angular momentum is L = 6ℏ , and the projection along a given direction is limited
to one of the values 0, ±ℏ, ±2ℏ. We therefore find that there are only five distinct
orientations of axis about which rotation can occur in this case, as illustrated in
figure 5.5.

5-4
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 5.5. Permissible orientations of the angular momentum vector for the case J = 2.

If we apply an electric field, the angle that each angular momentum vector makes
with the field vector produces an offset in the energy—for example with J =2 there
are five states, all with a slightly different energy. But with no field, there is no
preferred direction and for a given J, all values of M give the same energy. In
general, there are (2J + 1) values of M, so we have (2J + 1) energetically equivalent
states: we therefore say that the rotational state with quantum number J has
(2J + 1)-fold degeneracy, or
gJ = (2J + 1). (5.13)
The degeneracy factor is important because it means there are more and more states
with the same energy as J increases—and this affects the thermal distribution of
molecules between the various rotational states.
Having found a formula for the rotational energy levels, we now have to
investigate what transitions can be observed in a spectroscopy experiment.
We have already seen in a general way that when a molecule is irradiated with
light of the right frequency, light absorption occurs if and only if there is a non-zero
transition dipole moment. Applying the general formula of equation (7), it follows
that for a transition from a state with rotational quantum number J to a state with
quantum number J´, this transition dipole is given by

MJ ′ J = ∫ ψJ*′Mˆ ψJdq. (5.14)

In fact, MJ ′ J is directly proportional to the permanent, static dipole moment of the

molecule, μ;
MJ ′ J ∝ μ (5.15)

5-5
 Molecular Photophysics and Spectroscopy (Second Edition)

because molecular rotation doesn’t influence or induce any dipole in the molecule.
The physical significance of the apparently innocuous relation (5.15) is that we can
only have a transition dipole moment if there is a permanent dipole in the molecule—
otherwise no rotational transitions can occur. We can summarise:

So, whereas we can certainly observe the rotational spectra of heteronuclear dia-
tomics CO, HCl etc, we can evidently see nothing from homonuclear molecules such
as O2, N2 which, being centrosymmetric, have no permanent electric dipole. In fact,
this is a general rule, true also for non-linear molecules. For example, benzene does
not give a pure rotation spectrum because it too has no permanent dipole moment.
Still, pure rotational spectra of centrosymmetric molecules can be obtained using a
light scattering phenomenon called the Raman effect, a subject that is covered in
chapter 8.
The other important thing that we find out when evaluating the transition dipole
MJ ′ J is that:
MJ ′J ≠ 0 if J ′ = J ± 1: MJ ′J = 0 if J ′ ≠ J ± 1.
Therefore, transitions do not occur (they are forbidden) unless the rotational
quantum number changes by one unit only. So, we have the selection rule for
rotational spectroscopy:
ΔJ = ±1. (5.16)
ΔJ = + 1 corresponds to an increase in energy by absorption; ΔJ = −1 corresponds
to emission of light. This is consistent with the conservation of angular momentum
where the total angular momentum of the system (light + molecule) is conserved.
So, if we irradiate molecules and observe absorption frequencies for transitions
J → J + 1, these frequencies are given by:

hν = ΔE = EJ +1 − EJ
= Bh(J + 1)(J + 2) − BhJ (J + 1)
(5.17)
= 2BhJ (J + 1)
∴ ν = 2B(J + 1)

and the spectrum accordingly shows absorption lines at 2B, 4B etc (see figure 5.4).
As noted in chapter 4, in each case the absorption intensity will depend significantly on
the population of the state from which the transition occurs. Since rotational levels are
closely spaced, it emerges that many of the states with low J are appreciably populated
at room temperature—the population is governed by the Boltzmann factor (accom-
modating the degeneracy factor pre-multiplier) NJ ~ (2J + 1) e−BhJ(J + 1)/kBT.
It is relatively straight forward to numerically estimate the most populated J-level,

5-6
 Molecular Photophysics and Spectroscopy (Second Edition)

Jmax, at a given T by treating J as a continuous variable, setting the derivative of the

Boltzmann distribution above with respect to J to zero, and solving for Jmax:

kBT
Jmax = − ½. (5.18)
2hB

Equation (5.14) shows the expected behaviour that as the moment of inertia
increases, B decreases, and more J levels become populated for a given T.
An idealised spectrum is shaped as shown in figure 5.6, with a frequency interval
νJ+1 − νJ = 2B between successive absorption lines. As T decreases the rotational
line intensity distribution favours lower J levels while at higher T, the distribution
broadens and shifts its maximum to states with higher J values.

Exercise
Using equation (5.14), find the most populated rotational levels of nitrogen gas at
room temperature, 298 K and at 1200 K. For the nitrogen molecule, the quantity
hB/kB (deceptively known as the characteristic temperature for rotation) is 2.87 K.

Answer
The answer has to be an integer, and the nearest values for Jmax are 7, at room
temperature; 14 at 1200 K. The answer reminds us that the air around us is
rotationally quite excited; whereas for nitrogen the most populated level is Jmax =7,
for oxygen at room temperature (characteristic temperature 1.45 K) Jmax = 10. The
big difference is due to the oxygen molecule having slightly more massive atoms, and
a slightly greater bond length, than diatomic nitrogen; hence its moment of inertia is
larger, and the rotational levels are closer together.

Figure 5.6. Idealised absorption spectrum of a rigid rotor heteronuclear diatomic molecule.

5-7
 Molecular Photophysics and Spectroscopy (Second Edition)

5.2 Nuclear spin effects

One often associates nuclear effects with extremely high energy phenomena.
However, even before the structure of the nucleus was well understood, the spin
of a nucleus was known to have a profound effect on the rotational spectrum of a
corresponding homonuclear diatomic molecule (experimentally measured using
Raman spectroscopy—see chapter 8).
Recall that nuclei with integer spins (0, 1, 2, …) are bosons, while those with half-
integer spins (1/2, 3/2, 5/2,…) are fermions. For the former kind of particles, the
wavefunction must be symmetric (i.e. retain sign) on exchange of any two equivalent
species; for the latter, fermions, such a particle exchange must change the wave-
function sign. Now the total wavefunction of a homonuclear wavefunction,
including a nuclear spin function ψ nuc , is expressible as Ψ = ψ elecψ vibψ rotψ nuc . The
action of the exchange operator on the total homonuclear molecular wavefunction
has to be such that P12Ψ = +Ψ for bosons, P12Ψ = −Ψ for fermions—but it
transpires that the signs of the ψ elec and ψ vib parts of the wavefunction are themselves
unaffected by nuclear interchange. So, in addition to the nuclear spin wavefunctions,
we need only consider the effect of P12 on ψ rot . In fact, that outcome of
interchanging nuclei depends on the odd or even value of the rotational quantum
number J. Specifically, P12, transforms ψ rot as ( −1)J ψ rot . Thus, states with even J
have even symmetry, while those with J odd have odd symmetry.
As an example, consider the hydrogen molecule, H2. The most common,
stable isotope of hydrogen,1H, has a nuclear spin I = ½. (Note: in this section
alone, I denotes nuclear spin. Elsewhere I means the moment of inertia.
Unfortunately both are the standard symbols, widely accepted.) If we label the
two hydrogen atoms in 1H2 as 1 and 2, there are three symmetric nuclear spin
wavefunctions: α(1)α(2), β(1)β(2), 1 (α(1)β(2)+α(2)β(1)), and just one that is anti-
2
symmetric: 1
(α(1)β(2)−α(2)β(1)). Since 1H is a fermion, the total wavefunction Ψ
2
must be antisymmetric overall for every level. As even J levels are symmetric under
P12, they must be combined with the antisymmetric nuclear spin function to ensure
Ψ is odd. Similarly, levels with odd J must combine with the even spin functions,
again ensuring that Ψ is odd.
By convention, the levels with the larger statistical spin weighting are labelled
ortho while the ones with the smaller weighting are labelled para. Thus, we find that
spectroscopic transitions from the ortho levels (here, those with odd values of J) will
be three times stronger than those from the para levels (even J), giving rise to an
intensity alternation of 3:1 in the rotational spectrum—superimposed on the
expected Boltzmann intensity distribution for rotational lines. The overall shape
of the spectrum changes according to the temperature of the sample, but the
alternation in relative intensities does not.
Nuclear spin effects can be generalised as follows: for a given nuclear spin I, the
number of symmetric spin states; that is, statistical weight is given by (2I+1)(I+1),
while the number of antisymmetric spin states equals (2I+1)I. Consider therefore, as
a second example, 16O which has a nuclear spin I = 0 and is therefore a boson.

5-8
 Molecular Photophysics and Spectroscopy (Second Edition)

The total molecular wavefunction of 16O2 must be symmetric overall. However, in

this case the statistical weight for the antisymmetric nuclear spin states which
combine with J odd becomes zero. Therefore, all rotational (necessarily Raman)
transitions involving these para J levels will have zero intensity; here, only transitions
from states with even J are allowed—an extreme example of intensity alternation.

5.3 Interpreting rotational spectra

What can we deduce from a rotational spectrum? The relative intensities of the lines
enables us to calculate the system temperature, based on the Boltzmann factor
mentioned earlier: there are various non-intrusive methods of temperature determi-
nation based on this principle. At the molecular level, from the line spacing we can
also find B and hence the moment of inertia I. From this, we can calculate the bond
length in the case of a diatomic molecule since I = mrr2.
For example, for H127I we find B = 196.4 GHz (note 1 GHz = 109 Hz), giving a
bond length of 0.1604 nm—observe the high accuracy of the result. Rotational
constants mostly lie in the 1–1000 GHz range, so that rotational transitions are
observed in the gigahertz, microwave region. (Often, rotational constants are expressed
in wavenumber units (cm−1), obscuring this connection.) Microwave spectroscopy
thus provides a means of obtaining the bond lengths of heteronuclear diatomics.
As mentioned at the outset, a similar form of result for the rotational levels emerges
for all linear molecules, provided the more general formula for moment of inertia,
equation (5.8), is applied. The results given so far therefore also apply to molecules
such as H–C≡N, or O=C=S. Here, however, whilst the rotational spectrum enables us
to find the moment of inertia, it cannot directly tell us the bond lengths because there
is more than one to be determined. The usual way to overcome this problem is to use
isotopic substitution. For example, if we find the moments of inertia of the isotopomers
O C S and 16O12C34S, then we can solve to find the O=C and C=S bond lengths.
16 12 32

This is, however, on the assumption that the bond lengths are unchanged on isotopic
substitution—in fact, this is not quite true, and consequently the results are not quite
as accurate as those for diatomic molecules.

Exercise
To determine both the H–C and C≡N bond lengths in hydrogen cyanide, rotation
spectra are collected from two isotopomers, H12C14N and D12C14N. Here deuterium
D is an isotope of H (2H) found for example in heavy water, having mass ~2 u. The
corresponding rotational spectra show adjacent line spacings of 88.632 and 72.416
GHz, respectively. Calculate each of the moments of inertia, IHCN and IDCN; then
using the following formula, work out the C≡N bond length, rCN;
⎛ 9 ⎞ ⎛1 ⎞
rCN = ⎜ ⎟IHCN − ⎜ ⎟IDCN .
⎝ 28u ⎠ ⎝ 6u ⎠

5-9
 Molecular Photophysics and Spectroscopy (Second Edition)

Answer
Since the separation between adjacent lines equals 2B, the B values are 44.316 × 109
Hz for HCN and 36.208 × 109 Hz for DCN. Also,
B = h 8π 2I ∴ I =h 8π 2B .
Hence, with h = 6.626 × 10−34 J s; u = 1.661 × 10−27 kg, we find IHCN = 1.8937 ×
10−46 kg m2 and IDCN = 2.3178 × 10−46 kg m2. Using the equation leads rCN = 1.157
× 10−10 m, or 0.1157 nm.

5.4 Centrifugal distortion

There is one other point to consider, which is again well illustrated by the example of
O=C=S. Table 5.1 compares the observed frequencies of selected transitions for
16
O=12C=32S with results calculated on the basis of a rigid molecule, as assumed so
far. As the results in the second and third columns of the table show, the agreement
is quite good for low J, but not so good for high J.
Better results can be obtained if we take into account the centrifugal stretching of
the molecule in states with high J, where it is rotating more rapidly. It is evident that
some such effect ought to occur: there has to be some degree of flexibility in the
bond itself, since other forms of spectroscopy reveal bond stretching vibrations.
(Note that the usage here is a correct application of the term ‘centrifugal’: during
rotation the centripetal attraction is insufficient to keep the bond length at a fixed
value.) This bond extension effectively increases the moment of inertia slightly ever
increasing with the rotational quantum number J, and so brings the rotational levels
a bit closer together. It then turns out that we have to add a correction term to our
previous energy expression, equation (5.6):
E = BhJ (J + 1) − DhJ 2(J + 1)2 (5.19)
where D is termed the centrifugal distortion constant. By applying this correction—
see the fourth column in table 5.1—the theory proves to give a remarkably good and
consistently close agreement with the observed frequencies. Occasionally, for very
light molecules containing H, one centrifugal distortion term may prove insufficient.

Table 5.1. Frequencies of selected rotational transitions in 16O=12C=32S; observed and calculated results
derived on the basis of fixed, or variable bond length (third and fourth columns, respectively).

Rotational Observed Rigid rotor result Distorted rotor

transition frequency (GHz) (GHz) result (GHz)

0→1 12.162 98 12.162 99 12.162 98

1→2 24.325 92 24.325 98 24.325 94
2→3 36.488 82 36.488 96 36.488 82
9→10 121.624 63 121.629 88 121.624 04
19→20 243.218 09 243.259 76 243.217 81

5-10
 Molecular Photophysics and Spectroscopy (Second Edition)

One then empirically adds additional centrifugal terms of order J n(J+1)n, n ⩾3, as
needed.

5.5 Non-linear polyatomic molecules

Considered as a three-dimensional object, a polyatomic molecule in general has
three distinct moments of inertia directed along notional x-, y- and z-axes. In certain
cases, two or more of these moments may be equal, and we can classify molecules
accordingly. There are four basic categories, illustrated with examples in figure 5.7.
Instead of using (x, y, z) axes, which implies a specific orientation of the molecule
in space, the axes are often labelled (a, b, and c) such that the following inequalities
hold: Ia ⩽ Ib ⩽ Ic. Thus, linear molecules have Ia = 0 and Ib = Ic. Spherical top
molecules with Ia = Ib = Ic have similar rotational levels to linear molecules but,
since they can have no dipole moment, it is not possible to obtain their pure rotation
microwave spectra. Symmetric tops come in two categories. If Ia < Ib = Ic the
molecule is a prolate symmetric top with a rugby football-like shaped contour, and if
Ia = Ib < Ic the molecule is an oblate symmetric top with a disk-like shaped contour.
Finally, asymmetric tops have Ia ≠ Ib ≠ Ic. Both symmetric top and asymmetric top
molecules do display rotational absorption spectra, but those spectra are much more
complicated because the energy levels depend on more than one quantum number;
they cannot be accurately described by any simple analytic formula. However, the
information about moments of inertia which we can derive from any such spectra
remains very useful, and by using isotopic substitution methods we can again derive
precise values for bond lengths and bond angles. On the whole, rotational
spectroscopy provides the most precise method of determining the structure of
free, reasonably small molecules. It has even proved possible to identify and
characterise weakly coupled van der Waals complexes such as the HCN…HCN
dimer by such methods.
Two main factors have ultimately prevented wider usage of these techniques with
large molecules. One is the difficulty of resolving and interpreting highly complex
spectra with necessarily much more tightly packed lines (recall how the line spacing
varies inversely with the moment of inertia). The other is the problem of getting
sufficient vapour pressure, given that the observation of free quantised rotation
requires a gaseous sample, and larger molecular species tend to be less volatile.

Figure 5.7. Examples of the four major classes of rotor in three dimensions.

5-11
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 6
Diatomics and triatomics: vibrational energy
levels and spectra

We now turn to the subject of vibrational energy levels and spectra, again starting
with the simplest case, the diatomic molecule. For this analysis, the key parameters
are defined as shown in figure 6.1.

6.1 Diatomic molecules: harmonic motion

As a starting point, it is convenient to treat the diatomic molecule as a simple
harmonic oscillator. Classically, this means that when the atoms are separated by a
distance r, there is a restoring force F = −k(r–re) pulling the molecule back to its
equilibrium bond length re. The force is therefore negative (attractive) when r > re,
and positive (repulsive) when r < re. The constant of proportionality k, so defined, is
the force constant of the bond. Note that this measures the elasticity, rather than the
breaking strength, of the bond; the larger the force constant, the stiffer or stronger
the bond. The force constant depends not only on the nature of the atoms, but also
on the type of bonding between them—and in larger structures this can vary
according to the type of molecule. As an example, look at the average force
constants for carbon-carbon bonds in a variety of organic compounds. One finds
that k for a C–C σ-bond is ∼500 N m−1. The average force constants for a C=C
double bond (1 σ-bond + 1 π-bond), and a C≡C triple bond (1 σ-bond + 2 π-bonds)
are approximately twice, and three times larger, at 1000 N m−1 and 1500 N m−1,
respectively. As a result, the average C–C bond length shortens as the bond order
increases; that is, re(C–C) > re (C=C) > re (C≡C).
The potential energy associated with the harmonic oscillator has a parabolic
form:
1
V= k (r − re )2 (6.1)
2

doi:10.1088/978-0-7503-3683-3ch6 6-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 6.1. Key parameters for a diatomic molecule. The left-hand atom is arbitrarily considered the
coordinate origin for displacement of the right-hand atom, so that its appearance as stationary is for
illustrative purposes only. The molecule experiences a restoring force F on extension (as shown) or
compression of the bond, where r < re.

Table 6.1. Molecular parameters deduced for the hydrogen halides from
rotational and vibrational spectra.

Rotational spectra bond Vibrational spectra force

length (nm) constant (N m−1)
1
H19F 0.0917 965.5
1
H35Cl 0.1275 515.7
1 79
H Br 0.1413 411.6
1 127
H I 0.1604 314.1

and we find that the classical frequency of vibration is

1 k
ν0 = (6.2)
2π mr
where mr is again the reduced mass of the molecule which we encountered earlier,
equation (5.5). Not unexpectedly, equation (6.1) shows that molecules with larger
reduced masses vibrate more slowly, as do species with weaker bonds (small k). One
should nevertheless keep in mind that some molecules that have a large total mass
can still have a small reduced mass. Hypothetically, an isolated uranium–hydride
(U–H) bond could have a relatively large vibrational frequency since its reduced
mass is approximately the mass of H alone.
Since the reduced mass mr of the diatomic molecule is easily worked out,
experimental determination of the vibrational frequency enables us to find a value
for the force constant k. Together with the bond length, which we find from the
rotational spectrum, we can thus obtain a fairly detailed picture of the diatomic
chemical bond. Consider the data for the hydrogen halides, shown in table 6.1. As
we might expect, the bond length increases with the size of the halogen atom, and the
bond also becomes more or less taut (remember that classically the restoring force
depends directly on the force constant k). However, this doesn’t tell us everything we
need to know; a very important property of the bond is its strength, as measured by
how easily the bond can be broken.

6-2
 Molecular Photophysics and Spectroscopy (Second Edition)

Whereas classically there is no restriction upon the value of the total energy
(potential plus kinetic energy) of the system, the quantum mechanical treatment of
the simple harmonic oscillator limits the energy to discrete amounts:
⎛ 1⎞
E = ⎜v + ⎟hν0 (6.3)
⎝ 2⎠
where v is the vibrational quantum number (often given the alternative symbol n.
Equally, the energy expression is commonly cast as E = n + 12 ℏω0 , where ( )
ω0 = 2πν0). The possible values of v are integers 0, 1, 2, …. So, we have equally
spaced levels with ΔE(v) ≡ E(v + 1) − E(v) = hνo. This system exhibits a zero-point
energy of E = ½ hν0 when v = 0. The origin of the latter can be considered a
quantum uncertainty effect, since a zero for both kinetic and potential energy would
contravene the position-momentum uncertainty principle—see the calculation in
chapter 4. The zero-point energy is not dependent on temperature, which means that
hypothetically even at a temperature of absolute zero, the oscillator would still be in
motion—a result that has no classical analogue. The first few levels and wave-
functions for the harmonic oscillator are shown in figure 6.2.
An examination of the quantum mechanical wavefunctions exhibited in figure 6.2
also shows that there is a finite probability that one can measure the bond length of
the quantum oscillator to be either larger or smaller than the values expected at the
classical extremes. This is strictly another quantum mechanical effect called
tunnelling, where one can find a system in regions of energy E < V which are
forbidden classically. One other especially striking quantum feature, exhibited by the

Figure 6.2. Potential energy curve for the simple harmonic oscillator model of a diatomic molecule, with the
potential energy V plotted against bond length r and centred on an equilibrium value re, also showing the
positioning of the first few quantum energy levels and their normalised wavefunctions.

6-3
 Molecular Photophysics and Spectroscopy (Second Edition)

ground state alone, is that the probability of finding a given separation between the
two atoms in the molecule is greatest near to the equilibrium bond length. (Recall
that large amplitude features in the normalised wavefunction ψ, of either positive or
negative sign, are associated with the largest values of probability per unit length,
∣ψ∣2.) This is in complete contrast to the classical behaviour.
Although it only becomes fully apparent for higher energy levels, it emerges that
as the vibrational quantum number v continues to increase, the probability of
finding the molecule does more closely approach classical limits. Modelling the
oscillator as a single particle with a reduced mass mr moving back and forth within
the potential well, the probability of finding the particle within a particular range of
positions would classically be equivalent to the fraction of time, within one cycle of
oscillation, spent passing through that region. The particle would accordingly be
most likely found closest to a turning point, where the total energy is entirely
potential energy—as compared to its equilibrium position, where the total energy is
all kinetic energy. Classically, these points correspond to the bond lengths of
maximum extension and compression—the two points where the parabolic potential
curve crosses the horizontal line for a particular energy. This is a manifestation of
the correspondence principle for large quantum numbers. (Note that the notion of
‘turning point’ in this classical sense, which we shall see in later chapters proves to be
a very useful concept in advanced spectroscopy, is not the same as the mathematical
sense of that term as a position where a derivative changes sign.)
For basic spectroscopy, our first question must once again be, what transitions are
allowed between vibrational levels? For the transition v → v′, our transition dipole
moment is given by

Mv ′ v = ∫ ψv*′Mˆ ψvdr. (6.4)

Evaluation of the integral reveals that, for the transition moment to be non-zero, the
transition needs to satisfy the selection rule v′ = v ± 1, i.e. any change in vibrational
quantum number is governed by:
Δv = ±1. (6.5)

At this stage we therefore conclude that for each energy level, absorption transitions
(upwards in energy, Δv = +1) can only occur to the level immediately above;
emission transitions (downwards, Δv = –1) go to the level one below. From the
detailed evaluation of equation (6.4) we can also discover the following
proportionality:

⎛ dμ ⎞
Mv ′ v ∝ ⎜ ⎟ . (6.6)
⎝ dr ⎠ r
e

So, the dipole moment of the molecule must change during the vibration for the
transition to be allowed:

6-4
 Molecular Photophysics and Spectroscopy (Second Edition)

In the case of diatomic molecules, it is clear that vibrational spectra can therefore
only be recorded for heteronuclear compounds such as CO, HCl, but not for
homonuclear species such as H2, N2. When we come to polyatomic molecules we
shall discover cases where a molecule lacks a permanent dipole moment, but it
acquires dipolar character in the course of a particular vibration—with the result
that the vibration is active in the absorption spectrum. Unlike purely rotational
spectra, it is not enough just to have a dipole moment: hence the necessarily lengthier
statement of the selection rule.
Now from equations (5.18) and (6.1) it is clear that the absorption frequency for
the v = 0 → 1 transition is given by hν = ΔE = 32 hν0 − 12 hν0 (i.e. photon
energy = difference between vibrational level energies). Hence,
ν = ν0. (6.7)
This is not as trivial as it appears: it means that the absorbed frequency of light
equals the classical frequency of molecular vibration, (1/2π ) k /mr from equation
(6.2). Vibrational frequencies are typically in the range 1012–1014 Hz, or hundreds to
thousands of wavenumbers (cm−1); hence absorption bands are measured in the
infrared. When we interpret infrared spectra, we often take it for granted that
absorption frequencies directly signify vibrational frequencies—but even this is true
only for simple harmonic vibrations. By way of contrast, as we have seen, no such
principle applies to rotations, for example; the absorption frequencies measured in
rotational spectroscopy cannot be directly interpreted as frequencies of rotation.

6.2 Anharmonicity and dissociation

Unfortunately, the harmonic oscillator model does not tell us everything we need to
know. It fails for a particularly important property of the chemical bond, namely, its
strength, as measured by how easily the bond can be broken. To deal with this we
shall have to refine our model. So far, we have assumed simple harmonic oscillator
behaviour, associated with a parabolic potential curve that has no upper energy
limit. However, we know that the molecule must ultimately dissociate if its bond is
sufficiently stretched; pure harmonic motion wrongly implies that no matter how
much energy is supplied to the system, the bond never breaks. Another inevitable,
and clearly incorrect consequence of a strictly harmonic potential—again if we go to
high enough energies—is that during the contraction part of the vibrational cycle the
atoms would eventually begin to overshoot, reaching a point where they would have
to pass through each other (as the potential curve crosses the vertical axis). This is
clearly not what happens. In recognition of these two failures, we need to adopt a
more realistic potential curve, as shown in figure 6.3.

6-5
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 6.3. Potential energy curves for an oscillator: simple harmonic (…..) and anharmonic (___).

It is relatively straightforward to understand the shape of the potential energy

curve in figure 6.3. Suppose we consider two atoms moving towards each other from
infinite separation. The initial condition corresponds to a molecule that is disso-
ciated (the bond is completely broken) and the system energy, set to zero as a
reference, corresponds to two separate ground state atoms. As the atoms approach
each other, bond formation begins and the potential energy drops, because the
bonded structure is more stable than the free atoms. As the separation between the
atoms decreases further, coulombic and Pauli repulsions arise. The nuclei repel each
other due to both having a positive charge, and the interpenetrating electron clouds
on the two atoms also repel each other—not only because of their common negative
charge, but because occupying the same region of space violates the Pauli exclusion
principle. The equilibrium bond distance between the atoms reflects a separation
where the attractive and repulsive interactions balance to zero. As the atoms are
pushed still closer together the potential energy increases rapidly due these repulsive
interactions. The result is a highly asymmetric potential energy curve of finite depth
corresponding to bond dissociation energy, De, or dissociation limit.
Fortunately, any such curve should closely approximate to a parabola near the
bottom of the potential well—so that the lowest vibrational levels will quite closely
agree with harmonic oscillator results.
A widely used function describing an anharmonic potential which leads to exact
solutions of the Schrödinger wave equation is the Morse potential:
V (r ) = De(1 − e−β(r−re ))2 (6.8)

6-6
 Molecular Photophysics and Spectroscopy (Second Edition)

which has the property that as r approaches ∞, V(r) approaches the value De. The
constant β is specific for the diatomic being considered. If we solve the Schrödinger
equation using a Morse potential we find that the energy levels are given by the
following equation:
⎡⎛ 1⎞ ⎛ 1 ⎞2 ⎤
E = ⎢⎜v + ⎟ − xe⎜v + ⎟ ⎥hν0 (6.9)
⎣⎝ 2⎠ ⎝ 2⎠ ⎦

where xe is a dimensionless anharmonicity constant (whose typical value is of the

order of 10−2) and, like νo, is specific to the molecule being studied. If more precision
is needed, in fitting the observed pattern of energy level spacings, the term featuring
xe can be considered a second-order correction of a Taylor series expansion whose
successive terms, scaling as (v + ½)n, where n is an integer >2, become progressively
smaller. As shown in figure 6.4, the effect of the xe term is to draw the vibrational
levels closer together as v increases, as opposed to the equal spacing of quantum
levels we had before. If we compare the observed and theoretical energies of the first
few vibrational levels of H35Cl, shown in table 6.2, it is evident that including
anharmonicity gives much better agreement.
One can easily show from equation (6.8) that ΔE(v) = hν0–2hνoxe(v + 1) which
implies that deviations from simple harmonic behaviour are more pronounced at
higher v. However, even though anharmonic effects are smallest for v = 0,
as described below they are usually not difficult to resolve experimentally using
infrared or Raman spectroscopy.

Figure 6.4. Energy levels shown on the potential curve for an anharmonic oscillator, showing the dissociation
energy D0 as it is experimentally measured, i.e. relative to the ground state, and also as defined with respect to
the bottom of the classical potential energy curve at the equilibrium position (De). Note how the average bond
length, signified by a position midway across the horizontal lines for successive energy levels, increases with
increasing v .

6-7
 Molecular Photophysics and Spectroscopy (Second Edition)

Table 6.2. Experimental results for the positioning of the first few vibrational levels of H35Cl, compared with
the results of calculations based on the harmonic oscillator and anharmonic oscillator models. Here the
superscript on the Cl atom symbol indicates which isotope of Cl is being considered.

Observed energy (cm−1) Harmonic oscillator Anharmonic oscillator

v = 0 1442.9 1442.9 1442.9

v = 1 4328.8 4328.8 4328.6
v = 2 7110.9 7214.7 7111.1
v = 3 9789.9 10 100.6 9790.4
v = 4 12 366.0 12 986.5 12 366.5

Figure 6.5. Schematic representation of the differences between the vibrational infrared spectrum of a
supposedly simple harmonic diatomic molecule, and one with anharmonic vibrations.

Specifically, anharmonicity relaxes the Δv = ±1 selection rule allowing weaker

features corresponding to previously forbidden transitions with Δv = ±2, ±3, etc to
be observed. Moreover, certain absorption lines that would overlap for the
harmonic case, appear at slightly different frequencies when anharmonicity is taken
into account. A comparison of the spectra for a harmonic oscillator and for an
anharmonic oscillator is shown in figure 6.5.
The 1 ← 0 transition is known as the fundamental; the transitions occurring at
optical frequencies around 2ν0 , 3ν0 , etc are known as overtones. Usually the first
overtone is typically ∼1/10 the intensity of the fundamental, the second overtone is

6-8
 Molecular Photophysics and Spectroscopy (Second Edition)

smaller again by another factor of ten. The other new features are transitions that
originate from vibrationally excited states, for which the absorption intensities are
governed by the Boltzmann factor exp(−Einitial/kBT). Since at room temperature
kBT ∼ 200 cm−1, the excited vibrational levels are generally poorly populated, so
that transitions from the vibrational ground rate are by far the strongest. Here we
see a marked contrast to the population of rotational levels, seen in the previous
chapter. However, since the excited state vibrational populations depend on
temperature, the intensity of such bands grows as temperature is increased—hence
the term hot bands is commonly applied. In the harmonic oscillator model, hot
bands would be obscured because their transition frequency is the same as the
fundamental.
From equation (6.9) we find the following expressions for the key transitions:

⎧⎡ 3 ⎛ 3 ⎞2 ⎤ ⎡ 1 ⎛ 1 ⎞2 ⎤⎫
⎪ ⎪

V1←0 = ⎨⎢ − ⎜ ⎟ xe ⎥ − ⎢ − ⎜ ⎟ xe ⎥⎬ν0 = vo(1 − 2xe ) Fundamental

⎩⎣ 2 ⎝ 2 ⎠ ⎦ ⎣ 2 ⎝ 2 ⎠ ⎦⎭
⎪ ⎪

⎧⎡ 5 ⎛ 5 ⎞2 ⎤ ⎡ 3 ⎛ 3 ⎞2 ⎤⎫
⎪ ⎪

V2←1 = ⎨⎢ − ⎜ ⎟ xe ⎥ − ⎢ − ⎜ ⎟ xe ⎥⎬ν0 = vo(1 − 4xe ) Hot band

⎩⎣ 2 ⎝ 2 ⎠ ⎦ ⎣ 2 ⎝ 2 ⎠ ⎦⎭
⎪ ⎪

⎧⎡ 5 ⎛ 5 ⎞2 ⎤ ⎡ 1 ⎛ 1 ⎞2 ⎤⎫
⎪ ⎪

V2←0 = ⎨⎢ − ⎜ ⎟ xe ⎥ − ⎢ − ⎜ ⎟ xe ⎥⎬ν0 = 2vo(1 − 3xe ) First Overtone

⎩⎣ 2 ⎝ 2 ⎠ ⎦ ⎣ 2 ⎝ 2 ⎠ ⎦⎭
⎪ ⎪

⎧⎡ 7 ⎛ 7 ⎞2 ⎤ ⎡ 1 ⎛ 1 ⎞2 ⎤⎫
⎪ ⎪

V3←0 = ⎨⎢ − ⎜ ⎟ xe ⎥ − ⎢ − ⎜ ⎟ xe ⎥⎬ν0 = 3vo(1 − 4xe ) Second Overtone.

⎩⎣ 2 ⎝ 2 ⎠ ⎦ ⎣ 2 ⎝ 2 ⎠ ⎦⎭
⎪ ⎪

Note that the fundamental frequency is not precisely ν0 , but slightly less; however if
we find the overtones or a hot band we have enough information to deduce xe, and
hence find the true value of ν0 .

Exercise
The ratio of the principal overtone and the fundamental absorption frequencies
satisfies the relation (as follows from the above results):

ν(1st overtone) ⎛ 1 − 3xe ⎞

=2⎜ ⎟.
ν(fundamental) ⎝ 1 − 2xe ⎠

Using this result, calculate a value for the anharmonicity constant of HCl, given that
the absorption spectrum shows a fundamental absorption at 2885.9 cm−1 with
another weak line, ascribed to the principal overtone, occurring at 5668.0 cm−1.

Answer
0.017 36.
At dissociation, the spacing between adjacent vibrational levels will close in to
zero. The highest v-level supported by the potential, nD, can then be determined by

6-9
 Molecular Photophysics and Spectroscopy (Second Edition)

assuming the relationship for ΔE(v) given in equation (6.5) holds. A graph of ΔE(v)
versus v to determine nD is called a Birge–Sponer plot. It emerges that De can be
calculated using:
hνo2
De = . (6.10)
4νoxe
(Note that for both the harmonic and anharmonic oscillators, the depth of the
potential well De is not the same as the measured dissociation energy D0, because of
1
the zero-point vibrational energy; to a good approximation De = D0 + 2 hν0.)

Exercise
For the molecular vibration of H35Cl the anharmonicity constant, determined by
comparing the absorption frequencies for the first few transitions, is evaluated at
0.018. Estimate the number of vibrational levels before a dissociation limit is
reached.

Answer
The dissociation limit should be the point at which the diminishing gap between successive
levels becomes zero. Thus, for this level nD we should have E(nD) = E(nD + 1).
From equation (6.4) we therefore have
⎛ 1⎞ ⎛ 1 ⎞2 ⎛ 3⎞ ⎛ 3 ⎞2
⎜n D + ⎟ − xe⎜n D + ⎟ = ⎜n D + ⎟ − xe⎜n D + ⎟
⎝ 2⎠ ⎝ 2⎠ ⎝ 2⎠ ⎝ 2⎠
1 1 3 9
∴ n D + − xen D2 − xen D − xe = n D + − xen D2 − 3xen D − xe
2 4 2 4
∴ 2xen D = 1 − 2xe
∴ n D = (1 − 2xe )/2xe = 0.964/0.036
= 27 (result for the next largest integer).

6.3 Vibration–rotation spectra of diatomic molecules

If we examine the vibrational spectrum of a diatomic molecule such as HCl at high
resolution, we find that there is a great deal of fine structure. We saw earlier that
there is a complete set of rotational levels belonging to each vibrational state—this
fine structure is due to transitions between rotational levels in the two vibrational
states. If the electronic ground state has angular momentum Λ = 0 (a Σ-state) the
selection rules are:
Δv = ±1, ΔJ = ±1. (6.11)
(In rare cases where the molecule has electronic angular momentum about the axis
(Λ ⩾ 1), ΔJ = 0 is also be allowed; for example, in NO). So, for the v = 1 ← 0

6-10
 Molecular Photophysics and Spectroscopy (Second Edition)

transition we have a large number of possible rotational transitions, as shown in

figure 6.6.
By reference to the ΔJ = 0 case, referred to as the Q-branch when it occurs,
transitions with ΔJ = −1 form the P-branch, while those with ΔJ = +1 make up the
R-branch. Note that for large polyatomic molecules, the rotational levels are usually
so close together that this fine structure cannot be resolved except with high
resolution instrumentation—that is why infrared absorption transitions are gener-
ally observed as wide bands; they embrace a spread of rotation–vibration
frequencies.
Now a noticeable feature of the vibration–rotation spectrum is that the lines draw
closer to form band edges at high frequencies, in the R-branch, as shown in
figure 6.6. The explanation for this is that the rotational levels do not have quite the
same spacing in the two vibrational states—indicating that the rotational constants
for those vibrational states are different. This difference arises because the average
bond length increases with increasing v in an anharmonic potential, resulting in
smaller Bv -values (which scale as 〈1/r 2〉).
The transition frequencies are thus given by the following (duly accommodating
the effects of vibrational anharmonicity, but ignoring the much smaller energy
contributions from centrifugal distortion):

Figure 6.6. Energy level representations of the rotation–vibration transitions in a heteronuclear diatomic
molecule, shown in order of increasing optical frequency and mapped to the corresponding lines in the
absorption spectrum. The rotational quantum numbers in the ground and first excited vibrational levels are
here designated J and J′ respectively. Note the distribution of intensities amongst the lines in the spectrum,
primarily determined by the thermal Boltzmann population of the initial rotational levels.

6-11
 Molecular Photophysics and Spectroscopy (Second Edition)

hν = ΔE = hν0(1 − 2xe ) + B′hJ ′(J ′ + 1) − BhJ (J + 1) (Δv = +1). (6.12)

Here, we use symbols J and J′ respectively for the rotational quantum numbers in
the ground and first excited vibrational levels. (It may be noted many traditional
texts use J″ for the former.) For the P- and R-branch transitions specifically, we
therefore have:
P−branch; J′ = J = 1 ∴ v = v0(1 − 2xe ) + B′(J − 1)J − BJ (J + 1)
∴ v = v0(1 − 2xe ) + (B ′ − B )J 2 − (B ′ + B )J
R−branch; J′ = J = 1 ∴ v = v0(1 − 2xe ) + B′(J + 1)(J + 2) − BJ (J + 1)
∴ v = v0(1 − 2xe ) + (B′ − B )J 2 + (3B′ − B )J + 2B′ .
By analysing the vibration–rotation spectrum we can evidently determine the
rotational constants (and hence moments of inertia and bond lengths) for both
the ground and first excited vibrational states, without resorting to microwave
spectroscopy, albeit usually with less precision.
It is worth remarking that such spectra can appear more complicated if the
elements involved have more than one significant isotope in their natural form of
abundance. The vibration–rotation spectrum of ordinary HCl, for example, in
contrast to the pure isotope spectrum shown on the previous page, reveals each line
to be split into a doublet, because there are two isotopomers, H35Cl and H37Cl; since
these have slightly different reduced masses, their vibrational and rotational energy
levels differ slightly. The intensities of the two resultant lines of each doublet reflect
the ∼3:1 natural abundance ratio of the chlorine isotopes.

6.4 The vibrations of triatomic molecules

Triatomic molecules present us with a first encounter of the much more complicated
vibrational energy levels and transitions that arise in polyatomic systems. The
primary difference from diatomics is that triatomics, as with all other molecules,
possess more than one vibrational mode. Immediately, we find that it is no longer
sensible to consider the molecule as a whole as either active or inactive in absorption
spectroscopy; each vibration has to be considered in its own right. As we observed in
chapter 4, the number of normal modes of vibration for a polyatomic molecule is
equal to the number of vibrational degrees of freedom. So, for example, there are
(3 × 3) − 5 = 4 vibrations for the linear molecule CO2, as shown in figure 6.7.
The mode descriptions and wavenumbers for CO2 are as follows: ν1: symmetric
stretch; ν2 antisymmetric stretch; ν3 bending mode. Note, the latter is doubly
degenerate; two independent modes of the same wavenumber, and hence the same
energy. One can use the arrows on the atoms in figure 6.7 to deduce the symmetry of
the vibrational motions. In the case of CO2, the D∞h character table is used to
establish how the arrows transform under each symmetry operation of that
molecular point group. If the sense of the arrows remains the same, then the motion
is symmetric under that operation. Conversely, if the arrows change direction, such
that a stretch becomes a compression or vice versa, then the motion is antisymmetric
under that operation. In this way one can show that for CO2 the irreducible

6-12
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 6.7. Normal modes of vibration of the carbon dioxide molecule, and their wavenumbers.

Figure 6.8. Normal modes of vibration of the water molecule, and their wavenumbers.

representation of ν1 has Σ g+ symmetry. Similarly, ν2 has Σ u+ symmetry, while the

doubly degenerate ν3 mode is Πu.
Now for H2O, as a bent molecule with C2v symmetry, there are (3 × 3) − 6 = 3
vibrations, depicted in figure 6.8:
Here, there is one symmetric stretch (ν1, A1 symmetry), one bending mode
(ν2, also A1 symmetry) and one antisymmetric stretch (ν3, B2 symmetry). Note, both
for CO2 and H2O, that the bending mode has a significantly lower frequency
(wavenumber) than the stretching modes, reflecting the fact that a bending of two
bonds usually needs less energy than for a bond stretch or compression. This is a
general feature, useful in interpreting the vibrational spectra of these and many other
molecules.
In fact, each normal mode of vibration—let us label each by an index i—behaves
like a simple harmonic oscillation, and the corresponding energy levels are given by:
⎛ 1⎞
Ei = ⎜vi + ⎟hνi (6.13)
⎝ 2⎠

Consequently, the selection rules are exactly as for the diatomic case:
Δvi = ±1 (6.14)

and
⎛ dμ ⎞
Mv ′ v ∝ ⎜ ⎟ (6.15)
⎝ dQi ⎠ Q
e

6-13
 Molecular Photophysics and Spectroscopy (Second Edition)

where Qi is the vibrational normal coordinate and Qe refers to the equilibrium

structure. Hence, we can affirm the following statement:

Water has a permanent dipole moment, and it changes in each of its three normal
vibrations—so all of them are active, giving three main bands in the infrared
spectrum. On the other hand, CO2 has no permanent moment—the dipole remains
zero throughout the symmetric stretch vibration ν1—yet the ν2 and ν3 vibrations
remove the centre of symmetry and allow a dipole moment to be formed. Using a
bond dipole model, the variation of molecular dipole moment during the course of
each of the vibrations of CO2 is shown in figure 6.9.
Consider ν2 for example. The dipole oscillates about zero, and the derivative of
the equilibrium position is non-zero—see figure 6.10. Hence, this mode is active in
the infrared spectrum. Similar arguments apply to ν3. The only mode that is not
active is ν1, the symmetric stretch, because for symmetry reasons the dipole remains
zero throughout the course of the vibration. So we notice that a difference in shape,

Figure 6.9. Normal modes of vibration of the carbon dioxide molecule, showing the variation in geometry in
the course of each vibration, and partial charge separations resulting from the greater electronegativity of
oxygen compared to carbon. At each step the resultant dipole μ from addition of the bond moments are also
shown. Note the modern physical chemistry convention that dipoles point in the direction that a free electron
would travel, i.e. pointing from more negative to more positive regions.

6-14
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 6.10. Variation in the electric dipole moment of the carbon dioxide with the coordinate Q2 signifying
the extent of its antisymmetric stretch vibration.

between these two triatomics, water and carbon dioxide, itself produces a clear
difference in the look of their infrared spectra, producing one less band for CO2.
A pictorial interpretation, while immensely insightful, becomes more challenging
although not impossible, as the number of atoms in the molecule increases. The
alternate, direct and reliable approach is to use character tables. The irreducible
representation of every ground vibrational wavefunction (v = 0) Γ(ψv ) is always
totally symmetric. Thus, the problem reduces to determining whether the irreducible
representation a vibrational mode, v′ = 1, Γ(ψv′) transforms as one of the irreducible
representations of the dipole moment operator Γ(μ). In the point group D∞h, the
representation Γ(μ) transforms as Σ u+ (z) ⊕ Πu(x,y). Thus, as shown before, ν2 and ν3

are infrared active but ν1 is not. Similarly, in C2v , Γ(μ ) transforms as A1 (z) ⊕ B1(x)
⊕ B2(z). Thus, all three vibrational modes of H2O have the correct symmetry to be
infrared active.

6-15
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 7
Large molecule infrared absorption spectroscopy

When we look at larger molecules, the vibrational spectrum quickly becomes much
more complex than the cases we have looked at so far. Consider benzene, for
example, which is still a relatively small molecule compared to most of the materials
in the living world. It has twelve atoms (recall the common schematic representation
in figure 3.1, which omits the positions of one carbon and one hydrogen atom at
each vertex), and hence (3 × 12) − 6 = 30 normal modes of vibration. In fact, ten of
these are non-degenerate and ten are doubly degenerate, so there are 20 distinct
vibrational frequencies. Although not all of these modes are active in the infrared
spectra, complications arise because of overtones and combination bands.
Fortunately, the appearance of group frequencies in the vibrations of complex
molecules is a redeeming feature which does allow some sense to be made of
the spectra.

7.1 Group frequencies and skeletal modes

Group frequencies arise in large molecules because, in certain normal modes, most of
the atomic displacement is localised in a small part of the system; the rest of the
structure undergoes very little motion. So the bulk of the molecule has little effect on
vibrations in small groups—for example the stretch of a carbonyl group—except to
act as a large mass attached to the carbon atom. The vibrational frequency
associated with each group thus remains much the same in quite dissimilar
molecules—see for example the striking comparison in figure 7.1. The span of
such spectra is typically depicted with the higher wavenumbers at the left, starting
somewhere near 4000 cm−1, running across to a few hundred cm−1. As we saw in the
previous chapter, water has absorption bands in the region approaching 4000 cm−1
(the modes and wavenumbers in figure 6.8) and these are amongst the strongest
observed; accordingly, any sample containing water is usually dominated by water
at this end of the spectrum, effectively closing off the spectral identification of other
species in this region.

doi:10.1088/978-0-7503-3683-3ch7 7-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 7.1. Infrared vibrational spectra of: (a) cyclohexanone, highlighting the C=O stretch at ∼1715 cm−1;
(b) ethanoic acid, with C=O stretch at ∼1739 cm−1. The two molecules are depicted in the usual schematic
chemical form.

Figure 7.2. Typical infrared absorption regions for chemical groups common in organic molecules.

The reason the characteristic band in an infrared spectrum associated with a

carbonyl stretch, for example, falls within a range of values, rather than one specific
value, reflects the fact that motion is never entirely restricted to the C=O bond. In
fact, what we see is one of the normal modes of vibration of the whole molecule, a
particular mode in which the majority of the motion is indeed within the carbonyl
group, but which will still have some degree of motion taking place throughout the
rest of the molecule. It is because of these other associated atomic motions, which
obviously must vary according to the particular chemical compound, that the
carbonyl stretch in various aldehydes and ketones can appear within a range of
wavenumbers.
This localisation of vibrational excitations can help enormously in assigning
peaks in infrared spectra to many functional groups. The ranges and typical
intensities (strong, medium, or weak) are often grouped together in correlation
tables or charts for easy reference. Figure 7.2 presents a broad guide to the features
that dominate the infrared spectra of many organic molecules. Recalling the simple
formula for the vibrations of a diatomic molecule, equation 6.2 back in chapter 6, it
is noticeable that even for polyatomic molecules there are trends explicable on a
similar basis. For example, the stretching frequency or wavenumber of a bond to
hydrogen is exceptionally high due to the low mass of the H atom. In the series of
groups with single, double or triple bonds to carbon, the frequency increases with the
bond order, as befits an increasing force constant. We can also observe an additional

7-2
 Molecular Photophysics and Spectroscopy (Second Edition)

feature that we encountered in looking at the triatomic molecules in chapter 7; the

frequencies of a bond stretch are generally higher than those of a bending mode; the
latter require less energy because the associated potential energy curves are more
shallow. The specific form of bending modes involving hydrogen bonds are generally
termed deformation modes, because these are involved in structural twists in the
molecular skeleton.
As an example, the peak assignments for methanoic acid given earlier, in table 4.1,
are based on such principles; for most molecules the peak values are also amenable
to calculation using dedicated software. In addition to the expected frequency
ranges, the shape of the absorption band can also provide evidence for a specific
functional group. For example, a carboxylic acid group in aqueous solution;
–COOH (where OH is bonded to the C of a C=O group), exhibits broad infrared
absorptions around 3400 cm−1 due to ion–water interactions. Specifically, –COOH
dissociates to form COO− + H+, satisfying the Brønsted–Lowry definition of an
acid. Similarly, the absorption band of an alcohol group (–OH) broadens due to
hydrogen-bonding, an unusually strong intermolecular force that owes its origin to
attraction between slightly positive (δ+) hydrogen atoms in one molecule and the
slightly negative (δ−) oxygen atom in a near-neighbour. The concept of group
frequencies ties in with the idea of bond moments—if a bond dipole changes during
vibration, the vibration should be active in the infrared spectrum. This also means
that the vibrations of strongly dipolar groups will often give the most prominent
absorption lines.
As a counterpart to the emerging group frequencies, the low-frequency part of the
infrared spectrum (typically between ∼500 and 1500 cm−1) is due to molecular
vibrational modes that correspond to motions associated with the nuclear frame-
work—the greater mass being responsible for the lower vibrational frequencies.
Accordingly, these ‘skeletal’ modes are more specifically characteristic of the
molecule as a whole. The main analytical value of this part of the spectrum is
that it serves as a fingerprint region, frequently enabling the rapid computerised
identification of unknown substances by comparison with results held in compre-
hensive databases of standard infrared spectra. In this way, for example, one can
distinguish isomers which are molecules with the same number of atoms of each
element, but having distinctly different structural arrangements of the atoms.

7.2 Infrared spectroscopy in the condensed phase

Most of the molecules we encounter are too large to be in the gas phase at normal
room temperature. Aside from the simple atomic, diatomic and triatomic constit-
uents of our atmosphere, few other substances are naturally gases at standard
temperature and pressure. Even amongst the very light alkanes—the hydrocarbons
CnH2n+2, only four of them (methane CH4, ethane C2H6, propane C3H8 and butane
C4H10) are gases under ambient conditions. Locally attractive intermolecular forces
between the molecules of most substances favour the condensed phase. The
heterogeneous electronic environments of individual molecules therefore broaden
spectral lines, as we discussed in chapter 1.

7-3
 Molecular Photophysics and Spectroscopy (Second Edition)

Water has unique properties, worth noting at this juncture. Most of the physical
properties of H2O, in the condensed phase (liquid water, solutions, water-containing
substances and ice) are all strongly affected by hydrogen-bonding. It is the strength
of this interaction that keeps water liquid at room temperature. Spectroscopically,
this effect very substantially broadens the peaks in the infrared spectrum of any
substance with water content. So the ν1 and ν3 vibrations shown in figure 6.8 are
together smeared over a broad range of wavenumbers, and their absorption bands
are so strong that they would usually obliterate the signals of other substances
absorbing in the same region. For this reason, infrared spectroscopy is not a very
helpful technique for intrinsically water-containing materials—which includes
almost all biological samples; this is indeed where Raman spectroscopy comes to
the fore, as we shall see in the next chapter.
Infrared spectroscopy is one of the most familiar workhorses in industrial
laboratories; it is relatively fast, highly amenable to computer control, and the
instrumentation is relatively cheap. Pure substances whose infrared molecular
spectra are required may be measured in films, oil-based suspensions, or vaporised
for the purpose. In the analysis of more complex, real-world samples, there may
often be a need for first separating components, by chromatographic techniques for
example. The exact methodology depends very much on the nature of the sample
and the analytical precision required—a topic which goes beyond the remit of this
book. One of the more recent developments, which has come to fruition as a result of
breakthroughs in source and detector technology, lies in security and defence
applications of terahertz imaging. This deploys sensors operating in the range
0.1–20 THz (3–600 cm−1) spanning the long-wavelength infrared and microwave
regions. Here, there is seldom a need to identify specific molecules, but more often
the contrast in spectral response from different materials.

7.3 Near-infrared spectroscopy

At the other end of the usual infrared spectral range, starting in the water-dominated
region and running from around 4000 cm−1 up to 12 500 cm−1, close to the visible
region, is a near-infrared region that has acquired many other kinds of spectroscopic
application, often circumventing the kinds of problem presented by complex, multi-
component samples. A striking example from medicine is in the non-intrusive
monitoring of brain activity, with surface probes using function near-infrared
spectroscopy (fNIRS). This application exploits the relatively low scatter in tissue
at near-infrared wavelengths, and the strong and highly differential absorption of
oxygenated and deoxygenated blood exhibited by two forms of haemoglobin, as
shown in figure 7.3.
In many applications of NIRS, the aim is to compare samples to a reference,
rather than to identify specific molecules present. Here, chemometric methods such
as principal component analysis, (PCA) or partial least squares (PLS) regression can
be deployed to good effect with near-infrared spectroscopy. Prominent examples of
this method are industrial applications in quality control, especially in foodstuffs
and allied materials. In such samples, a huge variety of complex organic molecules

7-4
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 7.3. Absorption (logarithmic scale) spectra of haemoglobin (Hb) and oxyhaemoglobin (HbO2) across a
window of wavelengths in the near-infrared.

may be present in widely varying concentrations, usually along with a substantial

water content that effectively rules out the identification of spectral features due to
individual components. Here, advanced methods of statistical analysis enable
correlations to be established between spectroscopic data and other salient phys-
icochemical attributes known to relate to purity, freshness, region of origin etc. Such
methods are also used for quality control in the pharmaceutical industry, as well as
in processed satellite imaging. The latter is an example of hyperspectral imaging,
where spectral information is secured for each individual pixel in an image. Needless
to say, such methods generate huge volumes of data, calling for advanced storage
and data-mining methods.

7-5
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 8
Raman scattering and spectral interpretation

In discussing molecular rotations, vibrations and electronic excitations, we thus far

have concentrated on the frequencies and selection rules responsible for absorption
spectra. We shall now look at a different kind of spectrum resulting from the
scattering, rather than the absorption of light—this is known as the Raman spectrum.
To understand its mechanism, it is first helpful to look at a more familiar light
scattering effect, with which it is closely related: Rayleigh scattering.

8.1 Rayleigh scattering

Let us begin by thinking about ordinary light scattering. When a beam of light is
passed through any gas or liquid, some light is scattered in various directions, as
depicted in figure 8.1. Most of the scattered light has the same frequency ν′ as the
incident light ν, and it is known as Rayleigh scattering: ν′ = ν. In fact Rayleigh
scattering is responsible for the brightness of the sky, as sunlight is scattered by
individual molecules in the atmosphere. (Note: this is not Tyndall scattering, which
is associated with microscopic particles: a perfectly clean atmosphere still scatters
light.) If it were not for Rayleigh scattering, the sky would be completely black even
in the daytime—as indeed it is from the surface of the Moon. It emerges that the
scattering intensity depends on the fourth power of the frequency; light at the blue
end of the spectrum is much more effectively scattered than the red end. This
accounts for the blue of the sky, and the red of dawn and sunset.
If we look in detail at the theory of Rayleigh scattering, we find that it involves the
molecular polarizability. This is easy enough to understand, since polarizability
generally signifies the ease of distorting a charge distribution. If the fluctuating
electric field of the electromagnetic radiation impinging on a given molecule is E ,
then there is a resultant induced electric dipole moment that is given by:
μind = αE . (8.1)

doi:10.1088/978-0-7503-3683-3ch8 8-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 8.1. Light scattering through a non-absorbing medium. Monochromatic input radiation with optical
frequency ν, encountering atoms or molecules, produces scattered light with frequencies ν′.

In the above equation, for introductory purposes, the polarizability is treated as a

scalar constant of proportionality. But, in fact, the molecular polarizability cannot in
general be represented by a single value. The electron distribution in molecules has a
shaping based on the chemical structure, anchored as it is by the nuclear arrangement,
and it will generally be easier in some directions compared to others for an applied
electric field to distort the arrangement of charge. This means that the direction of the
induced dipole moment will not usually be the same as that of the applied field.
To correctly describe the relationship a second rank tensor is required, represented by
a 3 × 3 matrix of components αij, where i and j are Cartesian coordinates:
μiind = ∑ αij Ej . (8.2)
j = x, y, z

The quantum mechanical expression for the polarizability tensor is beyond our
present scope, but it will be useful to observe that it has the properties of a product of
transition electric dipoles (in fact, a weighted sum of such products). On this issue it
is worth noting that a superficially appealing, traditional ‘polarizability ellipsoid’
description proves rather less than helpful; indeed it can be highly misleading if used
to describe inelastic scattering. Such a representation does not allow for more than
three independent polarizability components.
To continue: from equation (8.2) it follows that if the electric field is oscillating at
frequency ν, so too does the dipole oscillate at this frequency, giving rise to its own
electric field at frequency ν. In the language of classical electrodynamics, the
molecule acts like a secondary source; electromagnetic waves are emitted in all
directions with this same frequency, and Rayleigh scattering is therefore observed.
Rayleigh scattering using a monochromatic laser has a strong angular dependence.
The strongest scattering is in the forward direction, coincident with the direction of
propagation of the incident beam, while the weakest signals are in the opposite
direction. This angular effect however is not prominent when using an incoherent
lamp or in sunlight.

8.2 Vibrational Raman scattering

Raman scattering is an inelastic counterpart effect, which takes its name from
Chandrasekhara Venkata Raman—the scientist who pioneered experimental work on
the subject early in the last century. Raman and others examined the spectra of light
scattered from purified organic liquids, discovering weak frequencies significantly

8-2
 Molecular Photophysics and Spectroscopy (Second Edition)

shifted from the frequency of irradiation. These lines appeared shifted to both the high
and low frequency sides of ν by equal amounts. However, the intensity of lines on the
high-frequency side was observed to fall off rather sharply.
The frequency shifts measured in the scattered light were shown to be character-
istic of the sample, but not due to fluorescence—spontaneous light emission from an
excited state. (This will be covered in chapter 13.) For example, adding a substance
known to quench fluorescence would not prevent the unusual frequencies appearing
in the light emerging from the sample. Also, the shifts were found to have the same
values if a different input frequency was used—the entire spectrum shifts along with
the Rayleigh line. A model spectrum exhibiting the key features is shown in
figure 8.2, with features either side of a Rayleigh scattering line subsequently
designated as Stokes and anti-Stokes. It was Raman who first realised that such
observations could be accounted for by the scheme illustrated in figure 8.3.
In the mechanism responsible for the Stokes frequencies, the molecule scatters
light of a lower frequency, and hence lower energy than the absorbed photon. The
energy difference must promote the molecule to an excited (usually vibrational or
rovibrational) level in the ground electronic state: the difference in the initial and
final state energies is therefore small, compared to the photon energies. For energy
conservation we must have:
Stokes: hν′ = hν − ΔE , ∴ ν′ = ν − Δν. (8.3)

Figure 8.2. Schematic Raman spectrum, here showing the intensity of scattered light as a function of input
optical frequency. Lines typically appear in pairs of dissimilar intensity, with frequency shifts of equal
magnitude to the either side of a much more intense Rayleigh line (ν′ = ν). The high-frequency (anti-Stokes)
lines are less intense than their Stokes counterparts, and progressively diminish in intensity. In practice the
Rayleigh signal may be removed by a cut-off optical filter to save overwhelming the detector.

8-3
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 8.3. Illustration of Raman scattering processes connecting the ground state with just one selected upper
level, compared to Rayleigh scattering, showing the photon energies involved in each case.

In the anti-Stokes case, however, the molecule starts off in the excited vibrational
state and Raman scattering leaves the molecule in its ground state; we therefore have
a shift in the other direction, to a higher frequency:
anti−Stokes: hν′ = hν + ΔE , ∴ ν′ = ν + Δν . (8.4)
Hence the Stokes and anti-Stokes lines are shifted by the same amount Δν from the
Rayleigh line at frequency ν. Also, the intensity of anti-Stokes lines must fall off
rapidly with ΔE because the population of molecules initially in the excited level is
governed by the Boltzmann distribution e−ΔE /kT . As we saw in the previous chapters,
it is such vibrationally excited levels that are indeed responsible for the weak hot
bands in infrared absorption spectra.
It follows, from equations (8.3) and (8.4) that the various shifts in either a Stokes
or an anti-Stokes Raman spectrum can usually be attributed to transitions between
vibrational states, and directly identified with frequencies of molecular vibration.
However, we shall see that Raman spectroscopy is often capable of delivering
information that other spectroscopies such as infrared cannot.
Raman scattering is a very weak process—much weaker than Rayleigh scattering.
It should be noted that usually the input light is not resonant with an intermediate
excited state. Instead the scattering is said to proceed via a virtual state, a quantum
mechanical entity which is a superposition of all real eigenstates of the system. The
intensities of the Raman lines can be increased dramatically however, if the exciting
light frequency approaches an allowed transition in the sample; a process called
resonance Raman (section 8.4).

8-4
 Molecular Photophysics and Spectroscopy (Second Edition)

What revolutionised the Raman effect as a spectroscopic method was the advent
of the laser, the obvious choice for a source of intense, monochromatic radiation.
While in principle any input light frequency ν can be used, it can be shown that the
Raman scattering intensity, like Rayleigh scattering, scales as ν4. For this reason, the
blue-green 488 nm or green 514 nm lines of a powerful Ar+ ion laser, were and still
are often used. However, these sources can also generate a lot of background
fluorescence noise. Popular alternatives today are to use a near-infrared 785 nm
diode laser, or the 1064 nm output of a solid-state Nd (neodymium):YAG (yttrium
aluminium garnet) laser which have high powers but tend not to excite sample
fluorescence. Unlike most Ar+ ion lasers, these light sources can be made relatively
compact, leading to the development of hand-held Raman spectrometers for a
variety of applications outside the lab environment. These include the identification
of narcotics, unknown powders, liquids gels and mixtures, pharmaceutical quality
control, and medical anti-counterfeiting.
In principle, Raman signals can also be produced by transitions between
electronic levels—but in practice it is almost always used in connection with
molecular vibrations, so this is what we shall focus on. Evaluation of the transition
dipole moment connecting states with quantum numbers v and v′ reveals that:

⎛μ ⎞
Mv ′ v ∝ ⎜ induced ⎟ (8.5)
⎝ dQ ⎠ Q
e

(compare infrared absorption, where in the counterpart expression it is the change in

permanent moment with nuclear motion that appears). Hence, we have:

⎛ dα ⎞
Mv ′ v ∝ ⎜ ⎟ . (8.6)
⎝ dQ ⎠ Q
e

The derivative on the right-hand side of equation (8.6) is often written for shorthand
as α′. The key symmetry criterion for any specific vibration to generate a Raman
signal is thus as follows:

We also find that Mv ′ v is zero, and hence the transition is forbidden, unless
Δv = ±1 (Stokes, anti−Stokes) (8.7)

which, in this respect, is just as for infrared absorption. However, the dependence on
a change in polarizability, rather than dipole moment, is crucial, and it means that

8-5
 Molecular Photophysics and Spectroscopy (Second Edition)

quite different vibrations appear in the Raman spectrum. (For small polyatomic
molecules, rotational structure is also exhibited in the vibrational Raman spectrum,
where the selection rules are Δν = ±1, ΔJ = 0, ±2.) Notice especially the requirement
for vibration-induced variation in polarizability. In contrast to a dipole moment,
every molecule has a non-zero equilibrium value polarizability.
Consider first diatomic molecules, with only one vibration. This motion must
always change the polarizability. In visualising polarizability it is often most helpful
and realistic to consider it as something akin to the shape of and a magnitude
proportional to the volume of the electron cloud around the molecule itself.
Consequently, it is immediately evident that even non-polar, symmetric molecules
(e.g. homonuclear diatomics) generate Raman signals. As we saw earlier, the anti-
Stokes line should be much weaker because it requires the molecule to start in an
excited state, and as we have seen in earlier chapters the room-temperature
population of vibrational excited states is generally small.
With polyatomic molecules we have to decide whether or not each vibration is
Raman-active. Figures 8.4–8.6 illustrate how the polarizability varies in each of the
fundamental vibrations of CO2. Clearly the ν1 symmetric stretch is Raman-allowed,
whereas in the ν2 and ν3 vibrations (antisymmetric stretch and bending modes), since
the change in polarizability is symmetric about the equilibrium position, the
derivatives vanish at that point. Hence the latter modes (as well as the counterpart
to the degenerate bending mode) are forbidden in the Raman spectrum.
As with infrared spectroscopy, it is often much easier to use character tables for
more complicated molecules, to establish which vibrational modes are Raman-
active or forbidden. Specifically, a Raman transition will be allowed if the
irreducible representation of the normal mode involved transforms in the same
way as one of the components of the binary product of vectors: x2, y2, z2, xy, xz, yz.
This can be read directly off the right-hand side of the character table.

Figure 8.4. Variation of molecular polarizability in the course of the symmetric stretch vibration of CO2.. The
polarizability changes uniformly during the symmetric stretch and the slope is non-zero at the equilibrium
configuration.

8-6
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 8.5. Polarizability of CO2 during antisymmetric stretch vibration.

Figure 8.6. Polarizability of CO2 during one of its bending modes of vibration.

It was shown in section 6.4 that the ν2 (Πu) and ν3 (Σ u−) vibrations of CO2 are
infrared active but the ν1 (Σ g+ ) mode is not. Examination of the D∞h character
table shows, as did figures 8.4–8.6, that the opposite result holds; namely, ν1 is
Raman-active but ν2 and ν3 are not. With CO2 we therefore have a good illustration
of the mutual exclusion rule for centrosymmetric molecules, which states that
molecular vibrations that are allowed in the infrared spectrum are Raman-
forbidden, and vice versa. A word of caution, however: some vibrations may be
forbidden in both.
The mutual exclusion rule can be generalised as follows:
• Raman-active modes are gerade (g) and infrared inactive
• IR-active modes are of ungerade (u) and Raman inactive

In the present connection gerade means that the vibration looks exactly the same if
we perform a spatial inversion through the centre of symmetry. Applying this to the
vibrations of CO2 we can observe the symmetry principles illustrated in figure 8.7.

8-7
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 8.7. Effects of spatial inversion on the fundamental vibrations of CO2.

With a non-centrosymmetric molecule there is no mutual exclusion rule. Thus, in

the case of water where all three normal modes of vibration (section 6.4) are both
infrared and Raman-active. However, we will still often find vibrations in the
Raman spectrum that are forbidden, or at most appear only weakly, in the infrared.
So, these two kinds of spectroscopy can complement each other very usefully, as can
be seen by comparing the Raman, and infrared spectra of paracetamol (acetami-
nophen) shown in figure 8.8. Prominent differences are readily apparent: for
example, the stretch vibrations of the non-polar C–H groups, close to 3000 cm−1,
show up particularly well in the Raman spectrum. By contrast, in the infrared
spectrum this frequency region is dominated by stretching vibrations of the highly
polar O–H and N–H groups, broadened by association with hydrogen bonding.
Raman vibrational spectroscopy has several key advantages over infrared
absorption. First, the common usage of visible input light removes direct heating
problems associated with infrared input. Secondly, since the Raman signal from
water is not very strong, the spectra of substances in an aqueous environment can be
obtained without spectral domination by solvent bands that beset the infrared
spectra of aqueous solutions. Thirdly, conventional glass optics and cells can be
used, since glass does not significantly absorb visible radiation. For all these reasons,
Raman spectroscopy is much better suited than infrared for the study of many
biological materials.

Exercise
Three unknown substances, A, B and C, are available as samples: two are gases, one
a solid. It is known that each is a simple tetrahalide of the type MF4, where M
represents either carbon, sulfur or xenon. The Raman, infrared and microwave
absorption spectra of each are secured (in the case of the solid, by gently heating to
produce a vapour). Comparing the infrared and Raman vibrational spectra of A, it
is apparent that they contain no features in common. It also transpires that it is only
sample C that will produce a rotational spectrum.

8-8
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 8.8. Vibrational spectra of paracetamol (acetaminophen): (a) Stokes Raman scattering; (b) infrared
absorption. Both spectra courtesy of Nicolet Instruments Inc.

The origins of these differences in behaviour is that all three substances have
differently shaped molecules. CF4 is tetrahedral; SF4 is of trigonal bipyramidal
shape with one position (actually an equatorial one) occupied by a lone electron
pair, and XeF4 is square planar (octahedral with two opposing lone pairs, above and
below). Which of these are compounds A, B and C?

8-9
 Molecular Photophysics and Spectroscopy (Second Edition)

Answer
The mutual exclusion of vibrational modes between the infrared and Raman spectra
of A indicates that this is a centrosymmetric molecule, and hence it has to be XeF4
(in fact, this is the solid substance). This principle does not apply to the
non-centrosymmetric CF4 or SF4. Since only C gives a rotational spectrum (and
the other two compounds are therefore presumably non-polar) it has to be SF4.
Hence B is CF4.

8.3 Depolarisation ratio

Scattering generally produces radiation with a changed polarisation state. The effect
is typically characterised by the measurement of a depolarisation ratio, defined as the
intensity ratio of plane polarised components of the scattered light. For conventional
right-angled scattering, with input light polarised in the z-direction and directed
along the y-direction as shown in figure 8.9, the depolarisation ratio ρ of light
scattered in the x-direction is calculated as:
ρ = I (z → y )/ I (z → z ). (8.8)
For both Rayleigh and Raman scattering, the value of the depolarisation is directly
expressible in terms of polarizability parameters—in the latter case, polarizability
changes during vibration. The result is commonly expressed as:
3γ′2
ρ= . (8.9)
4γ′2 + 45α¯′2
Here the primes on the ‘mean’ and ‘anisotropy’ parameters, ᾱ′ and γ′, respectively,
denote values obtained from the polarizability derivatives defined in the sense of
equation (8.6): expressed in terms of Cartesian components, we have
1
α¯′ =
3( ′ + α yy
αxx ′ + α zz′ ) (8.10)

3
γ′ =
2 i, j (
∑ αij′ − α¯′δij . ) (8.11)

Figure 8.9. Scattering geometry for the usual measurement of depolarisation ratios. Against the lab-based
Cartesian frame the incident radiation is z-polarised and the light scattered at right-angles is resolved for
components with: (a) z- and (b) y- polarisation.

8-10
 Molecular Photophysics and Spectroscopy (Second Edition)

In the summation in the latter equation, the indices i and j are each taken over x, y
and z; δij is the Kronecker delta whose value is 1 if i = j, 0 if i ≠ j. Physically, ᾱ′ is
essentially a measure of the distortion in mean polarizability, produced in the course
of a particular molecular vibration, whilst γ′ measures the corresponding vibration-
induced distortion of the polarizability shape.
For gases and liquids, it emerges that the value of ρ usually lies between 0 and ¾,
for vibrations that are totally symmetric (vibrations during which the symmetry of
the molecule is unchanged—technically transforming under the totally symmetric
representation of the point group: see chapter 3), but equal to ¾ for other vibrations
that lower the molecular symmetry (since ᾱ′ is then zero). Using this principle,
extensive information on the symmetry properties each Raman-active molecular
vibration can be obtained by measurement of the depolarisation ratios for each line
in the Raman spectrum.

Figure 8.10. Ultraviolet/visible absorption spectrum of myoglobin. Raman spectra, measured with input laser
radiation corresponding to three different wavelengths from the absorbing regions, display selective resonance
enhancement of the signals from specific protein components, in each case those closest to the corresponding
absorption centres. Trp and Tyr refer to the amino acids tryptophan and tyrosine, respectively, and the inset
depiction uses the simplified motif representation of protein structures that is common in biochemistry. Data
courtesy of Sanford Asher (University of Pittsburgh).

8-11
 Molecular Photophysics and Spectroscopy (Second Edition)

8.4 Resonance Raman spectroscopy

There is an additional feature that can be exploited to good effect in Raman
spectroscopy. As noted above, although the Raman technique affords a means of
obtaining vibrational spectra without the use of resonant radiation (the laser light
used is not absorbed), the scattered intensity that needs measuring is at most about
one part in a thousand the strength of Rayleigh scattering—which is itself fairly
weak. However, it is possible to substantially enhance the Raman intensity by using
an irradiation wavelength that is specifically chosen to be close to that of an optical
absorption band; near resonance, in other words.
The resonance Raman spectrum obtained under such conditions is much more
intense than usual—often by a factor of 106 or more. The use of near-resonance input
wavelengths does, however, mean that sample fluorescence and heating can present
more of a problem; special instrumental measures such as sample spinning may be
required. Also, the optimum irradiation wavelength becomes a consideration driven
by the properties of the sample, so that some microprobe devices offer a selection of
different laser wavelengths. One other difference from the conventional Raman
technique is that molecular polarizability loses its index symmetry—for example
under these conditions we may have αxy ≠ αyx. In consequence, the depolarisation
ratio formula (8.5) no longer applies, and its values can often exceed ¾.
More importantly, considerable advantages arise for applications in the study of
large biomolecules, as it is the vibrations of parts of the molecule close to the optical
absorption centre that are most enhanced. As an example, specific vibrational bands
due the heme unit in the resonance Raman spectra of the highly complicated
molecule myoglobin can thus be amplified and readily identified against a back-
ground of much less intense peaks—see figure 8.10.

8-12
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 9
Electronic and vibrational states in
large molecules

9.1 Electronic states, transitions, and molecular structure

Electronic states differ between themselves far more than rotational or vibrational
states. In consequence of differently configured electronic orbitals, even the basic
structural parameters of the molecule—the equilibrium vales of the bond lengths
and the angles between different bonds—can substantially differ from one electronic
state to another. Accordingly, although the electronic energy levels are quantised,
there is no counterpart to the simple algebraic formulae that describe the positioning
of quantum energy levels for rotations and vibrations. This, in turn, means that there
is no general formula for the absorption frequencies. Nonetheless, one can often
observe trends and similarities between the electronic spectra of chemically similar
molecules.
As we saw in chapter 1, the much greater energy separation between electronic
levels, as compared to rotations or vibrations, means that single-photon absorption
transitions usually fall into the UV–visible region. The corresponding span of a
typical spectrum may encompass transitions to only a small number of electronic
excited states—in some cases only one or two. The possible changes in nuclear
configuration on moving from one electronic state to another mean that vibrations
should also be taken into account, when considering such transitions. Features of
this kind are generally referred to as vibronic (shorthand for vibrational-electronic).
Before proceeding further, it is important to recognise that, although vibronic
behaviour is intrinsic to each species of molecule, its spectroscopic resolution very
much depends on the nature of the sample. It is easiest to identify such features in the
gas phase, where molecules are effectively isolated from each other. This limits high-
resolution observations to relatively small molecules, although larger species—even
protein fragments—may be brought into a cold vapour phase, using molecular beam
research apparatus, for example. However, for most condensed-phase matter,

doi:10.1088/978-0-7503-3683-3ch9 9-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

spectra measured under ambient conditions exhibit the considerable impact of

inhomogeneous broadening. Moreover, this effect grows with the size of the
molecule, such that the individual absorption features typically merge into broad,
sometimes almost featureless bands—often broadened to the point of overlap.
Generally, the electronic states get closer together as we climb the energy scale,
where the possibility arises of potential curves of the same symmetry crossing each
other; that is, having the same energy at the same geometry. This leads to
perturbations to the expected vibrational energy structure, and even the possibility
of predissociation (bond breaking below the expected dissociation limit) if crossed by
a repulsive excited state potential energy curve from below.
Since the excited electronic states are higher in energy, less additional energy is
required for ionization. Very often, higher energy states often exhibit minima at
longer bond lengths (the higher energy orbitals generally extend over a larger region
of space; the charge density is a little less concentrated and in the course of vibration
the nuclei are on average less strongly attracted back in bond contraction).
However, there are cases where the opposite is observed. For example, if an
electronic excitation creates a state which is highly ionic in character, the resultant
Coulomb attraction can lead to excited states with shorter bond lengths than the
ground state. Another example is afforded by the excited electronic states of van der
Waals molecules (loosely bound dimers), which are usually considerably more bound
their weakly-bound ground states.

9.2 Vibronic structure in electronic absorption spectra

For simplicity let us consider an electronic transition 0 → f that also involves change
in one particular mode of vibration. Based on the general formulation given in
earlier chapters, equations (2.5) and (3.1), the transition dipole moment can now be
defined as;

∫ Φf *(q, Q ) μ^ (q, Q )Φi (q, Q )dq dQ

≈ ∫ ψ felec*(q , [Q ])Ψvvib ( f )*(Q )μ^ (q , Q )ψ0elec (q , [Q ])Ψvvib (0) (Q ) dq dQ
f i
(9.1)
≈ ∫ Ψvvib ( f )*(Q )Ψvvib (0) (Q )dQ × ∫ ψ felec*(q ) μ^ (q , [Q ])ψ0elec (q ) dq
f i

nuclear vibration integral electronic integral

where, since rotations are not resolved, the nuclear wavefunction is expressible as
simply its vibrational part. As we may recall, the above separation of nuclear and
electronic integrals, based on the Born–Oppenheimer approximation, rests on the
assumption that nuclear motions, being much slower than electronic motions, have
little influence on the electronic transition dipole moment. Our earlier calculations
(in chapter 4) about the extent of nuclear motion give a good physical basis for this
assumption.
We see that the transition dipole moment factorises into a form involving an
electronic transition integral, a convenient shorthand for which is μ f 0, and a nuclear

9-2
 Molecular Photophysics and Spectroscopy (Second Edition)

vibration integral that signifies the overlap between the vibrational wavefunctions
for the ground and excited state. The square of the latter overlap is called the
Franck–Condon (FC) factor. The electronic integral establishes the selection rules for
the electronic transition, and for this section we will assume that such transitions are
allowed. However, unlike infrared spectroscopy, where there are restrictions on Δv
within the ground state potential (chapter 6), in principle all vibronic transitions are
possible between the two electronic potential energy curves. The intensity of any (vf, vi)
vibronic band however is determined by the magnitude of its FC factor.
Consider figure 9.1(a) where, for generic diatomic potential energy curves the
absorption of light produces a transition from a ground electronic state vi = 0 to an
excited electronic state having a very similar equilibrium geometry. In this case the
largest FC factor will correspond to vf = 0 since both vi and vf have their maximum
vibrational wavefunction amplitude close to the equilibrium bond length, re. The
strongest transition will correspond to (vf,vi) = (0,0). All other transitions where
vf ≠0 will be weaker. On the other hand, if re of the excited state is less than that of
the ground state, or greater—as in figure 9.1(b) and (c) respectively, the strongest
vibronic absorption transitions will be to vf—levels whose outer turning points or
inner turning points, respectively, is closest to the ground state re. In these cases the
strongest transitions will be to (vf, 0) with vf > 0. The intensity of hot bands; that is,
absorptions from vi ⩾ 1, are still expected to be weak since the population of the
ground state vibrational levels is governed by Boltzmann statistics, as discussed
previously for infrared spectroscopy.

9.3 Vibronic structure in electronic emission spectra

The vibronic structure of electronic transitions in emission spectroscopy can be
considerably richer than that seen in absorption. Emission spectroscopy involves
studying allowed vibronic transitions that originate from an excited electronic state
and terminate on a lower electronic state by the emission of a photon. The lower
electronic state is often the ground electronic state, but in principle it could be any
other lower-lying excited state, if selection rules allow. Electronic excited states can
be directly populated in a mild electrical discharge where electrons collide with the
ground state molecule of interest to excite the species. Alternatively, one could use
light to promote the molecule into its excited state. In both instances the experiment
involves detection of the wavelengths of the emitted light using a dispersive
instrument such as a monochromator. Unlike absorption spectroscopy, which is
essentially a probe of the excited state, emission spectroscopy probes the ground
state, and it is a very useful methodology for extracting the ground state vibrational
and rotational parameters if rotational and/or infrared spectroscopy prove difficult.
Vibronic transitions can be divided appropriately into progressions and sequences.
A progression involves a series of vibronic transitions with a common vlower or vupper.
Each band will have a relative intensity governed by its FC factor. Again, if the two
potentials have very similar geometries, the strongest transition in emission will be
(vupper,vlower) = (0,0). However, if the geometries of the two electronic states differ, one

9-3
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 9.1. Electronic transitions (denoted by vertical arrows) occurring without change in instantaneous
nuclear configuration, for three cases: (a) an electronic excited state with similar equilibrium geometry to the
electronic ground state. The strongest transition (red arrow) connects the ground vibrational levels of the two
states (0,0). In principle a higher vibrational level of the upper state is accessible as a hot band such as (3,3)
(blue dotted arrow) from an initial level in the electronic ground state, itself vibrationally excited—though their
intensities will be weak due to the Boltzmann statistics. (b) For an electronic excited state with markedly
different geometry—a shorter equilibrium bond length in the upper state—upper vibrational levels are directly
accessible in transitions from the ground state, primarily landing at the far extreme of the vibration. (c) With
an upper electronic level having a longer equilibrium bond length, similar transitions occur, landing at the
most contracted point in the vibration.

9-4
 Molecular Photophysics and Spectroscopy (Second Edition)

can expect longer progressions. An example might be (vupper, vlower) = (n,0), (n,1),
(n,2), … where n could be the vibrational quantum number of several excited state
levels. When vupper > 0, there is also the possibility of observing progressions in
different wavelength regions: those originating from the inner turning points of the
excited levels differing from those originating from the outer turning points. This is
due to three factors. The first is that typically emission takes place on the nanosecond
(10−9 s) timescale, while vibrational motions are faster (~10−12 s). Thus, even if an
upper level vupper > 0 is excited specifically at one turning point (say by using
monochromatic laser excitation), the strongest emission bands will still involve both
turning points. The second factor is that the potential energy curves involved are in
general asymmetric; transitions from a specific excited vibrational level will thus probe
different lower state geometries, and hence potentially different lower vibrational
levels with substantial FC factors. Lastly, as will be described later in detail in chapter
12, excited state relaxation from high vibrational levels to those closer to the bottom of
the potential is extremely rapid for larger molecules. Thus, the emission usually comes
at significantly longer wavelengths than those required to excite an absorption.
A sequence corresponds to a group of transitions with a common value of
Δv = vupper − vlower. An example of a sequence might be the grouping corresponding
to Δv = 0; that is, (vupper,vlower) = (0,0), (1,1), (2,2), (3,3), …. The frequencies will be
similar, but usually not the same if the vibrational frequency and anharmonicity in
the upper state differs from those in the lower state. It follows that the frequency
separations between the members of a sequence could come to either higher or lower
relative values, depending on those vibrational constant differences. For larger
molecules—especially in the condensed phase—the vibronic structure, while present,
will be usually extremely difficult if not impossible to resolve due to the high density
of iso-energetic excited vibrational states inhomogeneously broadened by solvent
collisions, or other stochastic variations in the local electronic environment.
For small molecules—diatomics in particular—the paucity of vibrational modes
is such that measuring the decreasing frequency gap between successive absorption
lines in a progression is relatively easy to resolve. Again, in principle, the gap will
approach zero (due to anharmonicity), close to the excited state dissociation limit.
Graphical extrapolation based on this principle thus enables an approximate value
for the dissociation energy to be determined. This is the Birge–Sponer method, noted
in chapter 6.

9.4 Transition metal complexes: vibronic coupling in electronic

transitions
Transition metals are elements whose outer shell electron configurations contain
3d or 4d electrons. As discussed in chapter 2, there are five nd-orbitals for every n,
and these are essentially degenerate for a free metal atom or ion. However, this
degeneracy is usually lifted upon ligation—the formation of a spatially stable structure
where the transition metal, in one of its possible positively charged states, is
surrounded by a loosely bound chemical groups. Then, splitting of the d-orbital
set occurs, its form being determined by the symmetry of the resultant metal–ligand

9-5
 Molecular Photophysics and Spectroscopy (Second Edition)

complex, and so allowing electronic transitions to take place between the split levels.
In Ligand Field Theory one uses group theory to construct symmetry-adapted
molecular orbitals to explain the nature of the resultant molecular orbitals involving
the metal and ligands. Here, for simplicity, we will consider only the qualitative
Crystal-Field Theory, to obtain insight into the nature of the visible absorptions of
metal–ligand complexes.
Consider an ML6 octahedral complex (Oh point group) where M and L denote
the metal and ligand, respectively. In crystal field theory the ligands are approxi-
mated as point negative charges. When the metal ion is placed at the origin of a
coordinate system, its d z 2 and d x 2−y2 orbitals form a two-fold degenerate set (with eg
symmetry) that have lobes pointing along the axes directly at the ligands. This raises
the energy of these two orbitals, due to Coulombic repulsion. The remaining three
orbitals (dxy, dyz, dzx) form a degenerate set, with t2g symmetry, that lies lower in
energy because they are orientated between the axes and not directly at the ligands.
The d-orbital splitting patterns for complexes of lower symmetries can be deduced
by notionally removing specific ligands from the octahedral configuration, and
deciding if this yields a net stabilisation of some of the d-orbitals relative to others.
As shown in figure 9.2, for example, forming a square planar ML4 complex with D4h
symmetry, by removing the two ligands orientated along the z-axis, lowers the
energy of the d z 2 orbital in the eg set relative to d x 2−y2 , as well as the energy of the dyz
and dxz orbitals relative to dxy in the t2g set. Thus, the energy ordering of the
d-orbitals in such a case might be E(dxy, dyz) < E(d z 2 ) < E(dxy) < E(d x 2−y2 ). The exact
ordering, however, will depend on the metal ion and ligands involved.
The splitting between the eg and t2g set in an octahedral complex is denoted by Δo
and its magnitude is a strong function of the ligands making up the complex. This
splitting can be substantial. For example, Δo for the Ti(H2O)63+ ion has a magnitude
of 20 000 cm−1. Ligands that yield a large Δo value are called strong-field ligands

Figure 9.2. Crystal field splitting of the d-orbitals in an octahedral (Oh) and square planar (D4h) geometry. The
d-orbitals are labelled by their irreducible representations in the appropriate point group.

9-6
 Molecular Photophysics and Spectroscopy (Second Edition)

while those that result in smaller splittings are called weak-field ligands. The ligands
can be empirically ordered in this regard for a given metal and geometry in an
empirical spectrochemical series. A partial list from strong-field to weak-field is as
follows:
CO ≈ CN− > NO2− >NCS− >H2O > OH− >F− >SCN− >Cl− >Br − >I−.
It should be noted that such complexes can involve more than one type of ligand,
and that the splittings also depend on the metal cation and its charge. The resultant
colours of these species either in solution or in the solid phase are often striking. For
example, aqueous solutions of common cobalt, nickel and copper compounds,
which contain the solvated species [Co(H2O)6]2+, [Ni(H2O)6]2+, and [Cu(H2O)6]2+
respectively, are red, green and blue, respectively, indicative of a decreasing Δo.
Figure 9.3 shows these and some of the other colours that such octahedral complexes
can exhibit in aqueous solutions. The d-orbitals fill according to the Hund’s rule,
such that the terms of highest spin multiplicity lie lowest in energy. Electrons will
spin-pair in the t2g set before filling the eg orbitals if the magnitude of Δo is large
(strong-field ligands). Complexes of this sort are termed low-spin systems.
Conversely, electrons will fill both the t2g and eg orbitals before spin-pairing if the
ligands are weak-field: these complexes are designated as high-spin systems.
UV–visible transitions in a ML6 complex involve promoting a d-electron from t2g
to eg. Formally however, d–d transitions are electric-dipole Laporte forbidden.
However, this rule can be relaxed through vibronic coupling, as the transition
connects states whose total wavefunctions include not only an electronic component,
but also a vibrational normal mode. If that normal mode has the correct symmetry
the transition can become allowed—a phenomenon known as the Herzberg–Teller
effect. Physically, the vibrational motion breaks the inversion symmetry of the
molecule. In the case of the transition metal octahedral complexes, the normal mode
must have u-symmetry. These absorptions tend to be weak, even if the metal
complexes have no inversion centre such as those with tetrahedral or trigonal
bipyramidal geometries.

Figure 9.3. Octahedral ML6 complexes formed by transition metal ions in aqueous solution, where the ligand
L is H2O. The d-shell configurations are: Ti3+ (d1); V2+ (d3); V3+ (d2); Cr2+ (d4); Cr3+ (d3); Mn2+ (d5); Fe2+ (d6);
Co2+ (d7); Ni2+ (d8); Cu2+ (d9).

9-7
 Molecular Photophysics and Spectroscopy (Second Edition)

The strongest transition metal complex absorptions usually lie at higher energies,
in the UV, and they can be assigned to metal–ligand charge transfer (g ↔ u)
transitions. If an electron is transferred from an orbital lying principally on the
ligand to an orbital lying principally on the metal, or vice versa, the transitions are
termed a ligand-to-metal or metal-to-ligand charge transfer, respectively. These
transitions however can occur in the visible if the metal is easily oxidised (in the
redox sense of losing an electron) and the ligand easily reduced (gaining an electron);
if they do, they are often sufficiently intense enough to mask the crystal field
transitions. Notable in this regard are metal iodide complexes, because I− is readily
oxidised. Charge transfer transitions readily facilitate the analysis of ions such as Fe
and Mg in biologically important systems such as hemoglobin and chlorophyll,
respectively. The molecular structure and visible range absorption spectrum of
chlorophyll a—a key light-absorbing pigment in the process of photosynthesis—are
shown in figure 9.4.
The concept of vibronic coupling can be generalised. Let the irreducible
representation of the components of the electric-dipole moment operator be Γ(x,y,z).
Depending on the point group this could be a direct sum of two or more irreducible
representations. An electric-dipole forbidden transition between states ψi and ψf with
irreducible representations Γ(ψi) and Γ(ψf), respectively, will become allowed if
Γ(ψf ) ⊗ Γ(ψi ) ⊗ Γ(x , y, z ) ⊂ Γ(Qk ), where Γ(Qk) is the irreducible representation of
one of the normal modes of vibration of the molecule.
Vibronic coupling is not just important for transition metal complexes. It can also
be used to explain the absorption spectra of other types of molecules such as
benzene. Based solely on the excited state molecular orbitals of benzene, one might
expect only one absorption band near 184 nm in the vacuum ultraviolet region,
due to a 1E1u ← 1A1g transition. Experimentally, however, symmetry-forbidden
1
B2u ← 1Ag and 1B1u ← 1Ag transitions are observed near 200 nm and 260 nm in the

Figure 9.4. Chlorophyll a. (a) Molecular structure showing the complex coordination of the magnesium atom,
whose doubly charged state is indicated by the oxidation state II. Adapted from original by Yikrazuul <https://
commons.wikimedia.org/w/index.php?curid=7974375>. (b) Absorption spectrum in the visible range; absorp-
tion is principally in the blue and red regions, giving the compound its green colour (see chapter 10).

9-8
 Molecular Photophysics and Spectroscopy (Second Edition)

UV, due to vibronic coupling with B2g and E2g normal modes, and as a result they
exhibit resolved vibrational fine structure. To complete the picture for benzene it
should be noted that there is also a very weak 3E1u ← 1A1g transition near 350 nm,
partially allowed through spin–orbit coupling.

9-9
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 10
Electronic transitions, colours, and detection

10.1 The origins of colour

As noted in chapter 1, the visible region of the spectrum covers the wavelength
region between approximately 380 nm and 780 nm. It is, of course, of critical
importance from a physiological perspective that these wavelengths encompass the
colours humans can detect by eye. It has been known since the time of Isaac Newton
that white light, composed of all visible colours, can be dispersed; that is, separated,
in a prism to form a spectrum. That dispersion is due to a wavelength-dependent
variation in the index of refraction of the prism—which, away from any material
absorption lines, is a monotonically increasing function of the incident light
frequency. Thus, blue light emerges closer to the prism base than red light.
But, as we know, it is not only the absorption or refraction of light that can
generate colour; several other physical phenomena can be responsible. We already
observed, in chapter 8, how Rayleigh scattering, due to its intrinsic wavelength-
dependence, produces the blue of the sky—whose reflection is itself primarily
responsible for the colour of the sea. The source of our daylight is a well-known
example of incandescence from gas excitation. Stars such as our Sun are a
notable example because, as blackbodies their colours are determined entirely by
their temperatures—which can be as low as 2000 K and as high as >40 000 K. Red
stars, for example, have temperatures of around 3000 K, while blue stars are ~25 000 K.
Our sun as a yellow star has a temperature of ~6000 K.
Flame colours are produced from the complex reactions associated with com-
bustion, commonly involving hydrocarbons. For example, the yellow colour that
arises in flames under lower concentrations of oxygen is attributable to incandes-
cence from soot. On the other hand, the bluish colour observed by burning fuel in
higher oxygen concentrations, suppressing soot formation, arise from the generation
of electronically excited CH and C2 free radicals. If metal compounds are introduced
into such a flame, different colours (emission lines) are emitted that are specific to the
electronically excited metal atoms generated. The analytical technique of atomic

doi:10.1088/978-0-7503-3683-3ch10 10-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

emission spectroscopy takes advantage of this specificity by measuring the intensity

of light emitted from a flame to establish the quantity of elements in a sample.
Chemiluminescence is light observed from chemical reactions that, unlike combus-
tion reactions, do no not produce a large amount of heat. Bioluminescence, such as we
see in glow-worms and fireflies, is a specific kind; both arise from molecular
photochemical mechanisms. Depending on the reaction, such chemiluminescent
effects can generate emission in the UV, visible or infrared. One well-known forensic
application is the detection of blood using luminol (C8H7N3O2)–hydrogen peroxide
(H2O2) solutions. The hydrogen peroxide reacts with the iron in trace amounts of
blood to form O2 which then oxidises the luminol, leaving it in an excited state. This
state relaxes by emitting blue light.
From our everyday world, a last example is afforded by the colours often
observed in thin oil films and soap bubbles, which arise from interference effects.
When the dimensions of the films are on the order of the wavelength of light
(microns) certain colours from incident white light reflected from the inner and outer
surfaces of the films will add constructively while others will add destructively. The
net result is a colourful iridescent pattern since the films tend not to have uniform
thicknesses. The bright blue colour seen in many bird feathers and butterflies is also
due to this kind of surface effect; there are no blue pigments producing these colours.
Returning to our main spectroscopic theme, we will now focus on colours
resulting from electronic visible absorptions. When white light encounters an
absorbing medium, the colours reflected or transmitted, observed by eye, are
complementary to those absorbed. Thus, a material that absorbs red light strongly
for example will appear green while a substance that absorbs blue light strongly will
have a yellow to orange appearance. The spectrum of chlorophyll a shown in
figure 9.4(b) neatly illustrates the principle.
In a typical experiment, absorption spectra are obtained by measuring the
intensity of light transmitted through an absorbing sample; that is, one measures
the intensity of reflected light. The ‘missing’ light intensity as a function of
wavelength is a measure of the absorption taking place in the sample.

10.2 Photometry and Beer’s law

As noted above, an absorption spectrum is commonly found by recording a plot of
transmitted light intensity (sometimes called the irradiance; units are W m−2
(1 W = 1 J s−1) versus wavelength. Absorption is linearly dependent on the light
intensity I when I is low. It is also clear that the rate of loss of intensity on passing
through an absorbing medium is proportional to the instantaneous intensity, the
concentration of the absorbing species, and the distance the light travels through the
absorbing medium—thus we have the relation:
dI
− = α(λ)I [C ] (10.1)
dz
for the rate of change of irradiance along the propagation direction z. Here α is the
absorption coefficient, which is a sample-dependent function of wavelength λ, and [C ]

10-2
 Molecular Photophysics and Spectroscopy (Second Edition)

is the concentration of the absorbing species usually expressed in moles per litre, mol
L−1 or M. (The over-use of symbols such as α is ultimately unavoidable: in standard
usage the same symbol is used here as for polarizability in earlier chapters. However,
the context should always make it clear which meaning is intended.)
The key to the application of equation (10.1) is that the rate of intensity
diminution must equate to the rate at which absorption promotes molecules to
excited states. The simple measurement of intensity reduction as a quantitative
register of concentration for a specific analyte is known as photometry—the same
term may also be used for emission measurements. When the wavelength is varied so
that a spectral response can be recorded, the method is sometimes called spectro-
photometry. Here, we focus on absorption spectroscopy in the UV–visible range,
associated with electronic transitions.
Before proceeding further, it is worth noting that electronic absorption transitions
generally populate excited states that are, in general, no longer able to undergo
further resonant absorption of the incident radiation. Although this effectively leads
to a small reduction in the concentration of ground state molecules [C ], most of the
excited molecules rapidly return to the ground state by fluorescence (or less direct
methods, some of which are described in the next chapter): such emission occurs in
all directions. For each molecule, the mean interval between the arrival of successive
throughput photons is generally far longer than the excitation decay time, so the
ground state population is very little changed. Under high-intensity conditions, it is
possible that a sufficiently significant population difference could be established
between the ground and excited states—a phenomenon known as saturation, where
the effect of absorption no longer increases with the incident intensity. However,
while possible using high power lasers, this condition is almost never achieved using
UV–visible lamps, the usual light source in most common UV–visible spectrometers.
To continue: the solution to the differential equation (10.1) is a simple exponen-
tial decay function:
I = Ioe−αℓ[C ] (10.2)
where ℓ is the path length through the sample along the z-direction, usually
expressed in centimetres, cm. This result, generally known as Beer’s Law (sometimes
also the Beer–Lambert Law), expresses the intensity of light transmitted through the
sample in terms of the intensity I0 incident upon it. To facilitate expressions based on
logarithms to base 10, it is alternatively written as
I = Io10−εℓ[C ] (10.3)
where ε = α/2.303 is generally termed the extinction coefficient, or molar absorption
coefficient which has units of L M−1. Since the absorption usually covers a range of
wavelengths or frequencies, its value when integrated over the full extent of each
band is directly related to the square of the transition electric dipole for that band. If
we label the peak absorption frequency νf0 and the transition dipole μf0, we can
discover the following relation:
2π 2L(μf0)2 νf0 Lhνf0
∫ εdν = 3 ln 10hcε0
=
c ln 10
B. (10.4)

10-3
 Molecular Photophysics and Spectroscopy (Second Edition)

Here, L is Avogadro number, and εo is the vacuum permittivity = 8.8542×10−12 J−1 C2 m−1;
the second expression on the right introduces and serves to define the ‘Einstein
B-coefficient’, whose significance we shall pursue in the next chapter.
Returning to equation (10.3), the product εℓ[C] is often given the symbol A, and
referred to as the absorbance or optical density (OD) of the sample; it is evidently
related to the transmittance T = I/I0 by
A = −log10T . (10.5)

Exercise
Show that the fractional loss of intensity of the light passing through the sample,
ΔI /I0 = 1 − 10−A where ΔI = I0 − I.

Answer
ΔI /I0 = (I0 − I )/I0 = 1 − (I /I0 ) = 1 − T = 1 − 10−A , taking 10 to the power of
each side of (10.4) in the last step. The result is an important basis for quantitative
spectrophotometry.
For example, suppose that on passage through a certain sample, a beam of light
suffers a 14.2% diminution of intensity. The absorbance or optical density A is
defined by I = 10−AI0 , so when I/I0 = 100% − 14.2% = 85.8% ≡ 0.858, then –A = log
(0.858) = –0.0665. Hence the optical density = 6.65×10−2. If the sample cell is 0.1 cm
long and the sample is a solution of 10−3 M concentration, then since A = εℓ[C ], we
have an extinction coefficient for the solute given by ε = Α/ℓ[C] = 665 M−1 cm−1.

Exercise
Oxyhaemoglobin, HbO2, has a molar extinction coefficient in aqueous solution with
very small values for wavelengths at the long-wavelength, red end of the visible
spectrum, but a huge maximum value of 52 430 M−1 cm−1 at around 414 nm.
Calculate the transmittance of light with this wavelength through a 1.5×10−3 M
solution of HbO2, in a 0.3 mm cell.

Answer
Using ℓ = 0.03 cm and the other data as given, T = 10−A = 10−εℓ[C ] = 0.0044, i.e.
about 0.44% transmission. It is an optically very dense sample.
While direct measurement of absorption is common, there can be issues with
sensitivity because one is measuring small reductions in light intensity against a high-
intensity light background. This can be particularly problematic if the intensity of
the light source fluctuates. Although this can be ameliorated by a double beam
configuration where one records the ratio of the transmitted beam to the intensity of
the beam before entering the sample, there are more sensitive techniques to infer
absorptions.
Many molecules fluoresce after absorbing light (see chapters 11 and 12), the
emission often covering a range of wavelengths. Hence, one could measure the total
fluorescence yield as a function of incident wavelength, without needing to resolve

10-4
 Molecular Photophysics and Spectroscopy (Second Edition)

the spectrum of the output. This is called a fluorescence excitation spectrum.

Typically, one can only collect a small fraction of the emission—using for example
a photomultiplier, as the fluorescence goes in all directions. However, since the
measurement is against a zero-light background, such detection is usually much
more sensitive than direct absorption.
While the above method is well suited to condensed phase sample, other methods
are possible in the gas phase. Here, usually under short-wavelength excitation,
molecules can also often ionize to produce photoelectrons, which can be measured in
an electrostatic energy analyser, or molecular ions (parent ion or fragment ions)
which can be detected using a mass spectrometer. As these signals are also due to
absorptions, their measurements as a function of incident wavelength also constitute
excitation spectra. These are also sometimes referred to as action spectra.

10.3 Organic molecules: conjugation and colour

We have already observed that, although molecular orbitals involved in absorption
extend over the entire molecule, it is often possible to identify a specific atom or
group of atoms in the molecule as a chromophore; that is, as a functional group that
acts as the dominant light absorbing unit. In chapter 2, figure 2.6 showed some
prominent examples of such transitions in organic molecules, including π* ← π
transitions. Although these are not particularly intense, their transition wavelengths
can shift dramatically to longer wavelengths and grow in intensity (hyperchromism)
with increasing degree of conjugation; an effect readily observed in polyenes of the
form H(CH=CH)mH (table 10.1). The term ‘conjugation’ hare refers to the effect of
an alternation of single and double bonds between bonded carbon atoms in an
organic compound—a delocalisation feature that has the capacity to allow an
unusually low-resistance channel for electron conduction along the chain. In chapter
3 we saw an example of this in the case of benzene, where the associated resonance,
shown in figure 3.1, gives rise to characteristic ring currents.
The conjugation effect in polyenes can be understood quantum mechanically
using the well-known one-dimensional particle-in-a box model. Here, we consider an

Table 10.1. Absorption wavelength maxima, λmax, and molar absorption coefficients, εmax, for various
polyenes; for m > 1 the successive (CH=CH) units are linked by C–C single bonds.

m (H(CH=CH)mH) λmax/nm εmax/L mol−1 cm−1

1 162 10 000
2 217 21 000
3 258 35 000
4 296 52 000
5 335 118 000
8 415 210 000
11 470 185 000
15 547 150 000

10-5
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 10.1. The chemical structure of β-carotene.

electron confined inside a box of length d with zero potential inside but having
infinitely high potential walls at its boundaries. The energy of the electron in the box
is purely kinetic energy, but because the particle is confined, it is quantised with
energy eigenvalues En that depend on a quantum number, n:
n 2h 2
En = n = 1, 2, 3, … (10.6)
8me d 2
where me is the electron mass, and h is Planck’s constant. Such a system exhibits a
zero-point energy because n ≠ 0; that is, the particle never stops moving—a non-
classical result. The energy level separations ΔEn = En+1 − En scale as 2n + 1; that is,
they increase linearly with n.
Using these results, we can begin to understand the spectra of the polyenes by
modelling these compounds as a one-dimensional box whose length increases with
the number of double bonds, m. The box dimension, d, can be estimated by summing
the appropriate number of known average C–C and C=C bond lengths. Each of the
m double bonds in H(CH=CH)mH contributes two electrons to the energy levels of
the box up to n = m, taking the Pauli exclusion principle into account. The energy
level with n = m corresponds to the Highest Occupied Molecular Orbital (HOMO),
and the lowest energy transition will be from the HOMO to the Lowest Unoccupied
Molecular Orbital (LUMO) which has n = m + 1. As the lengths of the polyene
increases the peak absorption wavelength λmax shifts to the red, as expected for a
particle-in-a-box type system. The initially increasing values of the peak absorption
coefficients εmax, reflect the trend in a transition dipole that lies approximately along
the direction of the π-system. The simple model is remarkably successful in
identifying the main trends; the deviations evident from the values given in table 10.1
arise because the π-electrons of the polyene backbones are not completely delocal-
ised but still exhibit some influence of the alternating single and double bonds.
Furthermore, the box-length is determined more by the extent of the electron cloud
delocalisation than strictly by the dimension of nuclear framework of the molecule.
One naturally occurring conjugated molecule is β-carotene shown in figure 10.1
which is responsible for the distinctive orange colour of carrots. It consists of a long
chain of alternating single and double carbon–carbon bonds (9 double and 10 single
bonds), with a bulky cyclohexene ring on either end.
Substituents that increase the intensity and often the λmax of absorption peaks
when attached to a chromophore are called auxochromes. A set of empirical rules
exists for dienes that allow the absorption maximum of a compound to be

10-6
 Molecular Photophysics and Spectroscopy (Second Edition)

determined for different additional chromophores present and are useful for
structure determination. Similar rules have also been deduced for conjugated
carbonyl compounds, and mono- and di-substituted benzenes.
Solvent polarity can also play a role in determining the position of peak maxima
in an absorption spectrum. This effect, called solvatochromism, arises because the
HOMO and LUMO involved in a UV–visible absorption have different polarities,
and therefore interact differently with a polar solvent. Bathochromic (red) shifts
result when the LUMO is stabilised in the solvent relative to the ground state.; this
effect is commonly observed for π* ← π transitions. Conversely, hypsochromic (blue)
shifts, which arise when the HOMO is preferentially stabilised by the solvent, are
common in π* ← n transitions.

10-7
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 11
After light is absorbed: photophysics in an
excited electronic state

11.1 Interplay of excitation and decay

Most of the foregoing chapters have dealt with optical processes that lead to the
population of excited states. To look a little further into the theory, let us first
suppose we can model such processes as involving transitions in a two-level system,
comprising just the ground electronic state and one excited state. Labelling the
populations of the lower and upper states of such a two-level system as N1 and N2,
respectively, the change in the excited population N2 as a function of time can be
written as:
dN2
= B21 ρ(ν )N1 (11.1)
dt
where ρ(ν) is the energy density of the input light per unit frequency interval, and B21
is the so-called Einstein B-coefficient for (stimulated) absorption (SI units m2 J−1 s−2),
introduced in the previous chapter, in equation (10.4). The change in popula-
tion of the excited state after photoexcitation in general comprises two terms,
given by:
dN1
= B12 ρ(ν )N2 + A12 . (11.2)
dt
Here, the first term on the left represents the effect of stimulated emission, and the
second denotes spontaneous emission—i.e. radiative decay, whose photophysical
details we shall be considering later in this chapter. In equation (11.1), B12 is the
Einstein coefficient for stimulated emission and A12 is the Einstein A-coefficient for
spontaneous emission. By considering conditions of thermal equilibrium, equating
dN1/dt = dN2/dt one can show that B12 = B21 (which enabled us to drop the
subscripts previously) and that

doi:10.1088/978-0-7503-3683-3ch11 11-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

8πhν 3 16π 3ν 3μ122

A12 = B12 = (11.3)
c3 3hc 3ε0
where εo is the vacuum permittivity = 8.8542 × 10−12 J−1 C2 m−1 and μ12 is the
magnitude of the transition dipole moment for a transition between levels 2 and 1.
Absorption and its inverse, stimulated emission, are expected processes for atoms
and molecules in an external electromagnetic field. However, the N2 multiplier in
equation (11.2)—which is the excited state population—is usually sufficiently small
that spontaneous emission is the greatly more significant radiative route for excited
state decay. (Conversely, when thermal equilibrium is no longer present, as for
example in laser systems, stimulated emission becomes the dominant process.)
Spontaneous emission is a strictly quantum mechanical effect that occurs even in the
absence of an external light source. The inverse, A12−1 is called the excited state
lifetime, τ, with units of s.
In the more complicated case of real molecular systems, excited vibrational
(or rotation–vibration) levels in the electronic ground state, populated through
infrared absorption or Raman scattering, generally relax in a dissipative fashion, by
collisions or other close-contact interactions; the extent of decay by radiative
emission is relatively insignificant. When the much higher energy electronic levels
are populated, however, a wider variety of phenomena can ensue, prominently
including the radiative deactivation discussed above. This difference has much to do
with the fact that rate of radiative emission A12 is proportional to the cube of optical
frequency. On this basis, the mechanism of radiative emission can be considered
likely to be a million or more times more effective for the decay of electronic excited
states than for the decay of vibrational levels. This is, indeed, one reason why
emission spectroscopy is most commonly associated with visible light or shorter
wavelengths.

11.2 States accessible to photoexcitation

We have already seen that, in principle, the absorption of a photon in the
appropriate ultraviolet/visible frequency region can produce a transition from any
vibrational level in the ground electronic state, to any other vibrational level in the
excited electronic state, if the spectroscopic selection rules allow—as determined by
symmetry conditions. The intensity of the transition will also depend in part on
whether there is sufficient population in a ground state vibrational level based on its
Boltzmann distribution. As importantly, we showed that the intensity of any
individual transition is also dictated by the transition Franck–Condon factor. The
physical consequence is that if the ground and excited state potential energy surfaces
are similar, the most intense transitions connect states with the approximately same
vibrational quantum numbers. However, if the ground and excited state potential
energy surfaces differ—especially if they significantly differ in equilibrium bond
length—then transitions will take place between states of different and possibly
many vibrational quantum numbers.

11-2
 Molecular Photophysics and Spectroscopy (Second Edition)

In explaining the photophysics of excitation for any but the simplest molecules,
the widespread adaptation of potential energy diagrams designed for diatomic
species, with their single chemical bond, can lead to some misapprehensions—not
least being a failure to appreciate the huge number of bound electronic excited
states. Even Jablonski diagrams such as the example shown in figure 11.1 can be
misleading in this respect. Quantum mechanical calculations for most molecules
typically reveal tens or hundreds of electronic excited states, each with their own set
of vibrational level continua, all tightly packed together and overlapping on the
energy scale. (Moreover, if molecules are free to rotate then each vibrational level
will have its own set of much more closely spaced rotational levels—though in the
liquid phase, the proximity of other molecules means that steric effects usually wash
out the quantum structure, so that the rotational motions are essentially random.)
Fortunately, a degree of simplicity is afforded by the fact that usually only the lowest
energy and most widely spaced few states are involved in determining the most
important aspects of photophysical or photochemical behaviour, since these states
lie on the scale over which single-photon visible and ultraviolet photoexcitation
occurs.
The Jablonski diagram of figure 11.1 assumes the electronic structure of large
polyatomic organic compounds consists of two kinds of electronic states, singlets
and triplets. The singlet electronic states have closed shell electron configurations—
all electron spins are paired such that S = 0; thus we have (2S+1) = 1, and their S (for
singlet) designations are distinguished by subscripts, being labelled S0, S1, S2 etc.
When two electron spins are unpaired we find triplet states, (2S+1) = 3, thus labelled

Figure 11.1. Representative Jablonski diagram, its vertical scale denoting energy, showing singlet electronic
states on the left and triplets on the right, with vibrational levels represented as continua. The separation
between singlet and triplet states in this way are not meant to suggest they necessarily have different
geometries. In addition to labelled processes of absorption and fluorescence (spin-allowed) and phosphor-
escence (spin-forbidden), secondary absorption processes that can occur if the S1 or T1 excited states are
sufficiently long-lived are depicted by dotted blue wavy lines and arrows. The wavy horizontal lines indicate
optical input and output; dotted horizontal arrows denote intersystem crossing, and the wavy vertical lines
signify intramolecular vibrational relaxation.

11-3
 Molecular Photophysics and Spectroscopy (Second Edition)

T1, T2, etc. The subscript index is ordered in terms of increasing minimum energy of
the electronic state. Although this is standard notation, one should remember that S
in the spin multiplicity is the total spin quantum number and not the label of a
singlet state. As there is only one electronic ground state, S0, the index for the triplet
states begins at n = 1; that is, all triplet states are excited states.
It needs to be appreciated how it is that the energy of Tn is less than the energy of
the corresponding Sn state. The reasons are two-fold. First, due to Hund’s Rule
discussed in chapter 2, the two electrons in a triplet state occupy spatially distinct
orbitals which reduces the repulsive Coulomb interaction between the electrons.
Second, there is an important quantum mechanical symmetry consideration that
must be accounted for in multielectron systems. As noted above, electrons are
fermions and their wavefunctions are antisymmetric with respect to interchange of
labels. In multielectron systems no two electrons can be distinguished from one
another, and therefore it follows that if two electrons had exactly the same
coordinates (spatial and spin) the required anti-symmetry would yield a wave-
function that is identically zero. This is the basis of the Pauli Exclusion Principle,
also encountered earlier, dictating that no two electrons can occupy the same region
of space with the same spin orientation. Quantum mechanical calculations of the
electron–electron energy of an antisymmetric wavefunction yields not only the
expected Coulomb repulsion, but also a new energy term called exchange which has
no classical analogue. Roughly speaking it can be thought of as the energy resulting
from ‘sharing’ electrons simultaneously in two different orbitals. For singlet states
the repulsive nature of this exchange raises the energy, while for triplet states it has
an energy-lowering, stabilising effect. Hence, we can understand why T1 states are
lower in energy than S1. (Molecular oxygen, although a diatomic, is a rare but
important exception: with two unpaired ½ spins in its ground state configuration,
the total spin of S = 1 dictates a threefold degeneracy; i.e., a triplet ground state.)
The primary criterion that determines the accessibility of higher energy states by
photoexcitation from the ground state is the electron spin relationship between the
states so connected. Specifically, the strongest transitions are those that are spin-
allowed ΔS = 0, (chapter 2). Indeed, absorption from the ground state S0 as
described by Beer’s Law is governed primarily by spin selection rules. Hence as
shown in figure 11.1, the strongest transitions are between S0 and S1, where the
population of S0 is depleted and S1 increased. While shorter wavelength input can
sometimes directly populate an S2 level, direct transitions to T1 are not indicated
because to first order they are spin-forbidden, even though T1 is lower in energy
than S1.
At a more granular level it is helpful to recall from statistical thermodynamics
that after photoexcitation, if equilibrium were attained, the most probable
distribution of energy within the molecule would be among the highest number
of rotational-vibrational-electronic degrees of freedom. Put another way, this
principle will disfavour conditions in which all or most of the energy is concentrated
in one particular mode of any relatively large molecule. The actual processes
involved in distributing energy within a photoexcited molecule and deactivation are
considered next.

11-4
 Molecular Photophysics and Spectroscopy (Second Edition)

11.3 Decay channels

The process by which S1 decays by radiative emission is called fluorescence, which is
the correct terminology for spontaneous emission that is spin-allowed (lifetimes
typically in the 10−9 to 10−7 s range). For relatively small molecules in the gas phase,
fluorescence is expected to be a dominant deactivation process. However, other
processes listed in table 11.1 also become important for larger polyatomics.
Furthermore, these species tend to have low vapour pressures and therefore, are
usually studied in solution phase.
For the moment, we shall assume that there is only one species of fluorescent
molecule present (plus a possible quencher); in the chapter 13 we shall look at the
important differences that arise in the presence of any other, chemically different
molecule that can also fluoresce. All but one of the processes shown in table 11.1 are
kinetically first order; that is, the rate of change of the concentration (or population)
of the state involved over time depends linearly on the concentration (or population)
of that state, with a constant of proportionality known as the rate constant k in units
of s−1. It should be noted that the absorption rate constant ka is proportional to the
intensity (J m−2 s−1) of the exciting light source and can, therefore, be controlled
experimentally. The other rate constants are inherent and specific to the molecule
being studied. Concentrations, denoted by square brackets, with units of mol L−1
(molarity, M) are commonly used when dealing with liquid solutions.
The one exception to first order kinetics is collisional quenching whose kinetics is
second order, depending on the concentrations of both the excited state molecule
and the quencher. Here the rate constant has units of M−1 s−1.
One extremely important competitive process to fluorescence for large organic
molecules is intramolecular vibrational relaxation (IVR) where the energy deposited
into the molecule via absorption is rapidly converted non-radiatively (10−14 to 10−11 s)
into different vibrational motions associated with the excited state. This is
tantamount to converting the photoexcitation into heat. If, as is usually the case
for large organic molecules, the vibrational energy levels strongly overlap other
electronic energy levels, a non-radiative process called internal conversion (IC) can
transfer the vibrational energies in one electronic state to other vibration levels in a

Table 11.1. Photophysical processes and rates.

Process Reaction Rate

Absorption S0 → S1 ka[S0]
Fluorescence S1 → S0 + hν kf[S1]
IVR S1* → S1 kIVR[S1]
IC S1 → S0 kIC[S1]
ISC S1 → T 1 S
k ISC [S1]
Phosphorescence T1 → S0 + hν kp[T1]
ISC T 1 → S0 kISCT[T1]
Collisional quenching S1 + Q → S0 + Q* kQ[S1][Q]

11-5
 Molecular Photophysics and Spectroscopy (Second Edition)

lower electronic state, at approximately the same rate as IVR. Consequently, the
processes of IVR and IC are sometimes not clearly distinguished from each other.
In principle, an excited molecule could, as a result of effective IVR and IC decay
channels, relax to its ground state without emitting any light at all.
Consider the absorption of the laser dye Rhodamine 6G, in aqueous solution
(figure 11.2). This is a molecule that exhibits strong fluorescence, and whose
geometry slightly differs in its ground and excited states. For reasons that have
been established above, the release of electronic energy by fluorescence generally
occurs from the lowest vibrational levels of the S1 electronic excited state, due to
IVR and IC. Accordingly, we expect that the transition will ‘land’ on higher
vibrational levels of the S0 potential energy surface at a point that lies vertically
below its starting point on the upper surface. This means that the photon energy
released in the process of fluorescent decay is usually smaller—and hence the
wavelength is usually longer—than for the initial absorption: this is the explanation
for the well-known Stokes shift.
In the case of almost any organic dye such as Rhodamine 6G, its solution
absorption spectrum partly overlaps its fluorescence spectrum—but it also covers a
range of wavelengths shorter those found in emission. Irradiation with broadband
radiation, such as the output of a conventional white or ultraviolet lamp, is capable
of exciting a dye solution across its entire absorption band. As a result, fluorescence
will also occur right across its emission spectrum. For this reason, dye lasers which
employ molecules like Rhodamine 6G as their active (lasing) medium incorporate a
dispersive element such as a grating to generate a tuneable monochromatic output.

Figure 11.2. Absorption spectrum (solid line) and fluorescence spectrum (dotted line) of the laser dye
Rhodamine 6G, in aqueous solution.

11-6
 Molecular Photophysics and Spectroscopy (Second Edition)

Even if monochromatic laser radiation is to be used to excite the dye, any input
wavelength within the absorption band can be chosen; emission still occurs at a
range of wavelengths, but only those that are longer than the input—i.e., photons
with less energy than the input light.
Outside the laboratory, more familiar examples of the spectral shift associated
with fluorescence are afforded by the fluorescent pigments used in many paints,
marker pens etc, whose high visibility colours are a result of light absorbed across a
broad spectrum being re-emitted at longer wavelengths. The reason that orange and
yellow ‘Day-Glo’® colours appear brighter than any others is only in part due to
wavelength-variation in responsivity of the human eye; there is plenty of light with
shorter wavelengths in the daylight that produces the primary absorption—as
compared to fluorescent colours at the blue end of the spectrum which can capture
only a much smaller fraction of the energy in the daylight spectrum.
Returning to the mechanisms of excitation decay, in addition to fluorescence,
IVR and/or IC, after photoexcitation, the S1 population can often transfer into the
T1 triplet state by intersystem crossing (ISC). This is generally a slower process
(10−11−10−6 s range) compared to IVR and IC, due to the change in spin state
required—but it can still be competitive with fluorescence. The rate of ISC, kISC, is
governed by the Fermi’s Golden Rule, which enables it value to be estimated from the
following expression;
2π 2
kISC =
ℏ
∫ ψT⁎HˆSOψSdτ ρ(E T ) (11.4)

where ψS, ψT refer to the overlapping singlet and triplet states, ĤSO is the spin–orbit
coupling operator, and ρ(E T ) is the density of triplet states—i.e. the number of states
per unit energy interval. The magnitude of kISC increases when the vibrational levels
of S1 overlap those of T1, and also for molecules containing a ‘heavy’ (high atomic
number) atom such as Br or I, due to their stronger spin–orbit interactions that
weaken the distinction between singlet and triplet states.
Decay due to IVR relaxes the T1 state into its lowest vibrational levels—where
either non-radiative ISC to S0 could take place, or the molecule might emit light by a
spin-forbidden process called phosphorescence. The latter is relatively slow (10−4−10 s
or longer) because of the forbidden nature of the transition.
Finally, there is the possibility of an additional process called collisional quenching
(Q), which is decay due to interaction with a molecule of another type. For example
collisions between an excited molecule in state S1 and a quencher molecule Q, which
is usually in a singlet ground state, may lead to energy exchange, where the excited
molecule decays to S0 and Q acquires the spent energy in excitation to a typically
vibrationally excited state, labelled Q*. Q could be a solvent or solute molecule. This
deactivation pathway can be minimised by working at low temperatures or by using
viscous solutions. Notably, the presence of even minute traces of molecular oxygen,
O2, will often non-radiatively ‘quench’ (depopulate) T1 states, because their shared
triplet character allows them to strongly interact.

11-7
 Molecular Photophysics and Spectroscopy (Second Edition)

There are other complicating factors not readily represented in a Jablonski

diagram. For example, not all excited states are bound, in the sense of having a well-
defined potential energy minimum and corresponding equilibrium geometry. Some
states are associated with repulsive forces at all values of a particular bond length.
These electronic states—usually excited ones—are therefore termed dissociative:
molecules in such a state which spontaneously dissociate through bond fission.
Similarly, a molecule M could absorb one or more additional photons to produce
M+ + e− by photoionization. Then the energy initially deposited into the molecule is
used to create the ion M+ which may be vibrationally excited and impart kinetic
energy to the released electrons.

11-8
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 12
Molecular fluorescence

In the study of electronic excitation, by far the simplest direct measurement—to gain
insights into the photophysics—is molecular fluorescence. Even as a straightforward
measure of some prior absorption it has the virtue of presenting an essentially zero-
background signal. However, as we have seen, considerable intricacies are involved
in the processes of excited state decay and their interplay. Given the widely differing
time scales and rates associated with these excited state deactivation processes, it is
important to understand the kinetics involved more deeply and to establish their
influence on observing fluorescence after photoexcitation. In fact the fluorescence
spectrum, lifetime and decay, and polarisation all provide extremely valuable
information.

12.1 Quantum yields and fluorescence measurements

Let us first elicit the kinetics associated with the interplay of the processes discussed
in the previous chapter. The rate equation for the formation and deactivation of the
S1 state in the Jablonski diagram (figure 11.1), using the terms in table 11.1, is:
d[S1] S
= ka[S0] − kf [S1] − kIVR[S1] − kIC[S1] − k ISC [S1] − k Q[S1][Q]. (12.1)
dt
In this equation, terms with a negative sign imply deactivation, while the absorption
term with a positive sign is the only process that produces the excited state
molecules. Under steady-state conditions, where the rate of formation of S1 equals
the rate of its deactivation, it is straightforward to derive the following, steady-state
(SS) concentration of S1, [S1]SS:
ka
[S1]SS = S . (12.2)
kf + kIVR + kIC + k ISC + k Q[Q]
One can therefore define a fluorescence quantum yield, ϕf, which denotes the number
fraction of photons emitted in fluorescence relative to the number absorbed, as the

doi:10.1088/978-0-7503-3683-3ch12 12-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

rate of fluorescence/rate of absorption, or kf[S1]SS/ka. Using equation (12.2) this

yields:
kf
ϕf = S . (12.3)
kf + kIVR + kIC + k ISC + k Q[Q]

This, simply put, represents the fraction of [S1] giving rise to fluorescence. A
fluorescence yield that is large (approaching unity) suggests that the rate constants
for other deactivation processes are small; i.e. those processes are slow, relative to
fluorescence. Conversely, if the non-radiative processes have large rates relative to
fluorescence, the quantum efficiency will approach zero.
Collisional quenching fluorescence experiments allow kf to be determined. Here
we define the fluorescence lifetime, τf, as:
1 S
= kf + kIVR + kIC + k ISC + k Q[Q] (12.4)
τf
and kfτf = ϕf. Using equation (12.1), we therefore have:
d [S1] [S ]
= ka[S0] − 1 . (12.5)
dt τf
At steady-state the left-hand side of (12.5) vanishes, so that [S1] = ka[S0]τf. Then,
since the measured fluorescence intensity is If = kf[S1], it follows that:
If = ka[S0]kf τf . (12.6)
Inverting this equation, applying (12.4) now yields a result expressible as:

1 1 ⎛ S ⎞
kIVR + kIC + k ISC k Q[Q]
= ⎜1 + ⎟+ . (12.7)
If ka[S0] ⎝ kf ⎠ k ak f [S0 ]

In the absence of a quencher the second term in (12.7) disappears and the
fluorescence intensity secured under those conditions would be written as Ifo.
Thus, taking the ratio Ifo /If one obtains:
Ifo kQ
=1+ [Q]. (12.8)
If kf
A plot of the intensity ratio as a function of [Q] will yield a straight line with a slope
of kQ/kf and a y-axis intercept = 1. Such graphs are called Stern–Volmer plots. No
lifetime measurements are required for the adoption of this simple method.
By performing fluorescence lifetime measurements, the rate constants kQ and kf
can be separately determined to obtain τf, if fluorescence and collisional quenching
are the dominant processes taking place. Then, equation (12.4) simplifies to:
1
= kf + k Q[Q]. (12.9)
τf

12-2
 Molecular Photophysics and Spectroscopy (Second Edition)

Here, a plot of the inverse excited state lifetimes as a function of [Q] yields a straight
line with slope kQ and y-intercept kf.

12.2 Transition dipole orientations

As we have observed at several intervals, molecular excitation and fluorescent decay
mechanisms are commonly based on electric dipole transitions, in which the transition
dipole couples to the electric field of the incoming or outgoing radiation, E . The
transition dipole is a polar vector, and the explicit form of the coupling Hamiltonian is
a scalar product with the electric field vector, Hint = −μf0 · E . In the case of photon
absorption, this means that those molecules with transition dipoles parallel to the
electric field of the input light produce the greatest amount of absorption. For plane-
polarised light the coupling thus depends on the cosine of the angle between the
transition dipole and the polarisation vector.
Now when absorption leads to the process of emission, the decay transition dipole
moment, because of the Franck–Condon Principle, has precisely the same magni-
tude and direction as the absorption dipole (indeed this is why the Einstein
coefficients for absorption and emission are the same). This is assuming that
fluorescence results from the same electronic excited state as is initially excited—
usually the S1 state; in certain molecules such as naphthalene, where the primary
absorption is to an S2 state, emission comes from decay of the S1 state after internal
conversion, and the emission dipole therefore differs from the one involved in
excitation. Nonetheless, in either case the molecular fluorescence resulting from laser
excitation is usually polarised, because radiation is preferentially emitted with its
electric vector parallel to the decay transition dipole.
The first thing to note about transition dipole moments is that they need not point
in the same direction as any static, ‘permanent’ dipole. Indeed, cases are known
where a transition dipole is perpendicular to the ordinary dipole in a polar molecule.
The reason for this goes back to the principles we examined in chapter 3; the
transition dipole will commonly be determined by the symmetry of the excited state
it connects to. This is, itself, generally of lower symmetry than the molecule—and it
is the latter that decrees the internal orientation of any static dipole.
Even though the transition moments for excitation and decay commonly have the
same direction within each molecule, they may still be differently oriented with
respect to the outside world, as for example if the molecule rotates between the
absorption and emission events. Such rotations in a fluid are essentially random, so
that if the decay has a long lifetime the sample emission may become diffuse, losing a
specific polarisation. However, while such molecular rotations may be randomly
oriented, the speeds of rotation are governed by the viscosity of the fluid, and
molecular shape plays a significant role in the extent of viscous drag. Thus,
measurements of the extent of polarisation in the fluorescence, and its time-
dependence, can provide access to useful dynamic and structural information.

12-3
 Molecular Photophysics and Spectroscopy (Second Edition)

12.3 Photoselection and fluorescence anisotropy

The study of fluorescence polarisation requires initial excitation of the sample using
polarised light—and for time-dependent measurements, that input has to be in
pulses much shorter in duration than the excited state lifetime. Mode-locked lasers,
which can provide picosecond or femtosecond pulses, are ideally suited to this
purpose. With plane-polarised input, a randomly oriented distribution of molecules
cannot be uniformly excited; for reasons we have already established, the distribu-
tion of excited species will be weighted in favour of molecules whose transition
moments are most closely aligned to the direction of optical polarisation. This
phenomenon, known as photoselection, leads to anisotropy in the distribution and
polarisation of the ensuing fluorescence.
In contrast to Raman scattering, where in section 8.3 we encountered a significant
application of polarisation measurements, the polarisation effects in absorption and
fluorescence are more intricate, primarily because the photon input and output
processes in any Raman scattering event have to be essentially simultaneous.
Measurements of depolarisation in the case of fluorescence are usually cast in one
or other of two quantitative expressions in common usage; the anisotropy of
fluorescence, r, and the degree of polarisation, P. These are defined and related as
follows:
I∥ − I⊥ I∥ − I⊥ 2P
r= , P= ⇒ r= . (12.10)
I∥ + 2I⊥ I∥ + I⊥ 3−P

Here, I∥ and I⊥ are measured components of the fluorescence respectively parallel

and perpendicular to the polarisation of the incident excitation beam, resolved using
a polarisation filter. (These correspond to I(z → z) and I(z → y) as shown for the
Raman analogue in figure 8.9.)
The key factor determining the observed change in polarisation is the mean angle
α between the directions of the transition dipoles for excitation and decay. As
observed above, these will frequently have a parallel disposition within the
molecular frame, yet be differently oriented with respect to the laboratory-frame
axes because of the time-lag between absorption and emission. Assuming molecules
randomly oriented in a regular liquid, the angle θ is related to P by the following
formula:
3 cos2 θ − 1
P= . (12.11)
3 + cos2 θ
The maximum value of the polarisation anisotropy, r = 0.4 (P = 0.5) occurs when
the excitation and decay correlate with parallel transition dipole moments.
Measuring the macroscopic value of P (or r) from a given sample allows direct
calculation of the molecular parameter θ, and for fluids this can give information on
the excited state lifetime and the speed of rotational diffusion—the latter also
depending on viscosity. In the case of diffusion that is fast compared to the excited
state lifetime, the result for P averages out to zero—so the value at any given time is

12-4
 Molecular Photophysics and Spectroscopy (Second Edition)

a useful indicator of such processes. One rare exception is the case θ = 54.7°, the so-
called magic angle, when the fluorescence is fully unpolarised (r = P = 0) even before
any rotational motion. (A similar phenomenon arises in a spin-decoupling method
in NMR, and it too is associated with a detachment of two coupling parameters.)
In certain cases, it might be considered potentially problematic to differentiate
fluorescence from Raman scattering, since each can produce radiation with a red-
shifted wavelength. However, there are several well-known ways to overcome this
problem; for example use can be made of the fact that the fluorescence signal will not
generally exhibit any shift in frequency when the excitation frequency is changed,
whereas the entire spectrum of Raman signals will shift by the same uniform
amount. A more sophisticated method is to use pulsed lasers and time-gating
techniques to separate the time-delayed fluorescence from any spontaneous Raman
emission.
One other feature that can play into the emission characteristics in a multi-
component system is a degree of energy transfer between molecules, prior to
fluorescence. Here too, not only the emission wavelengths but also the temporal
characteristics of the donor and acceptor fluorescence can usually be distinguished,
as we shall see in the next chapter.

12.4 Fluorophores and laser-induced fluorescence imaging

Fluorophores are defined as molecules, or molecular components, that absorb light
and strongly fluoresce. Although the word chromophore could also be used, that
term is often used to identify specifically the part of a molecule which is responsible
for its colour. Many organic compounds containing highly conjugated aromatic
(benzene-like) rings and/or polyene chromophores have been synthesised that allow
emission wavelengths to be selected over the entire spectrum from the near ultra-
violet to the near infrared. The compound β-carotene is one example discussed
earlier—see figure 10.1: the Rhodamine 6G molecule (figure 12.1(a)), also discussed
above, is another. The latter molecule is a commercially available dye which can
be used to generate dye laser wavelengths between ~570 and 660 nm with an
output maximum at ~590 nm. One should keep in mind that this species exhibits a

Figure 12.1. (a) Chemical structure of Rhodamine 6G. (b) Solution of Rhodamine 6G in methanol excited by
a green laser (solid dye at the side). Adapted from https://exciton.luxottica.com/laser-dyes.html.

12-5
 Molecular Photophysics and Spectroscopy (Second Edition)

reddish-orange colour because it is absorbing green wavelengths or shorter; that

is, the colour of the dye is complementary to the strongest absorption wave-
lengths (figure 12.1(b)).
There are other factors that can affect the fluorescence output, such as the pH and
temperature of an aqueous solution; the emission spectrum itself may be shifted by
the chemical addition of specific electron-donating or electron-withdrawing sub-
stituents to the molecular framework. It is particularly important that these
molecules are rigid; there are no single bonds about which the molecular chromo-
phores could rotate away from each other and so potentially break the conjugation.
Photon absorption could involve longer wavelength π* ← n transitions (see
figure 2.6 in chapter 2), but those tend to be less fluorescent than π* ← π transitions
because they have smaller extinction coefficients and longer excited state lifetimes
(10−5–10−7 s), making competition from IVR and IC more significant. In addition,
the singlet-triplet energy gap is usually smaller for π* ← n transitions, leading to
enhanced ISC and phosphorescence.
Some fluorophores with strong and characteristic fluorescence signals are used in
imaging applications—especially in research biology, where information is often
sought on the location and local concentration of key biomolecules in complex
samples. The selective attachment of suitable fluorophores to such molecules of
interest enables scanning laser light in the blue–violet region to excite their
characteristic bright green fluorescence, and so secure high-definition micron-scale
maps. One of the most widely vaunted such fluorophores is green fluorescent protein
(GFP), originally derived from a species of jellyfish; genetically engineered variants
can also provide for the emission of other colours. Used in conjunction with fibre
optics, such materials also now find use in analyte-specific biosensors.
The adoption of ultrashort laser pulses as the excitation source, using the
appropriate detection instrumentation, also enables the lifetime of individual
fluorophores in complex samples to be discriminated. This method, known as
FLIM (fluorescence lifetime imaging), can add another dimension to the data
analysis afforded to imaging spectroscopy. Its significance is that even fluorophores
with similar, broadly overlapping emission spectra, whose spatial distributions
might otherwise be difficult to separately resolve, may nonetheless have substantially
different and readily distinguishable lifetimes. Images can thus be secured in which
the shorter or longer lifetime components are individually enhanced.

12.5 Quantum dots

Nanoscience deals with the synthesis, characterisation, and applications of materials
having sizes on the order of nanometres (10−9 m). The field has exploded over the
last 30 years because at these dimensions such materials’ optical, physical, and
chemical properties can be precisely controlled by simply changing their size. The
field is simply too broad to be adequately covered here, but we shall briefly identify
one important facet that once again relates to colour; we focus on the absorption
and resultant colours of another type of fluorophore: quantum dots, which are small
pieces of spherical semiconducting materials with radii ranging from ~2–10 nm.

12-6
 Molecular Photophysics and Spectroscopy (Second Edition)

A material made up of nano-sized quantum dots has a larger surface area to volume
ratio compared to the same volume made up of bigger particles. These particles are
neither atoms nor molecules, but in several respects they do behave rather like
atoms.
It is instructive to briefly consider the electronic structure of semiconducting
solids, to better understand the optical properties of quantum dots. If a solid
material has a periodic atomic arrangement, the Pauli exclusion principle leads to
the formation of electronic energy bands; specifically, a lower lying valence band
composed of a continuous density of orbitals that contain the outer electrons from
each atom. This valence band is separated from a usually empty higher-lying
conduction band, and the energy separation between the two bands is called the
band gap, Eg. From a general point of view, if Eg is zero, the material is classified as
a metal, electrons can easily move under an applied voltage. If Eg is relatively small,
typically 1–4 eV, the material is classified as a semiconductor because heat, light or
electricity can readily promote electrons from the valence band to the conduction
band, where they can then form a current under an applied voltage. This ability to
switch semiconductor current states on and off is at the heart of how computer chips
operate. Lastly, materials having very large band gaps (~15 eV) are classified as
insulators.
Semiconducting quantum dots are sufficiently small that quantisation of the
valence and conduction band becomes important (figure 12.2), and the actual energy
of an optical transition (in both absorption and emission) will be affected by the
radius of the quantum dot in a manner analogous to a three-dimensional particle in
a box.
When an electron is promoted to the quantised conduction valence band from the
quantised valence band it leaves behind a hole which behaves like a positive charge.
In fact, the electron–hole pair, which is bound together Coulombically like an

Figure 12.2. Left: cartoon of the continuum conduction (lower) and valence (upper) bands of a regular
semiconductor (left), separated by an energy gap Eg: Right: quantised energy levels for the same material in a
quantum dot of nanoscale radius d.

12-7
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 12.3. Emission from cadmium selenide (CdSe) quantum dots in solutions with differing diameters: blue
(2.0 nm), green (2.5 nm), yellow (3.0 nm), orange (3.9 nm), and red (4.2 nm). Photo: Courtesy Dr Michael
Wong, Rice University.

H-atom, is called an exciton. Overall, the optical transition energy ΔE for a

quantum dot with radius d can be estimated by the following expression:
hc h 2(n12 + n 22 + n 32 ) h 2(n12 + n 22 + n 32 ) e2
ΔE = = Eg + + − . (12.12)
λ 8m e⁎d 2 8m h⁎d 2 8rπε
The second term in equation (12.12) is the quantised particle-in-a-box energy of the
electron having an effective mass m e⁎; the third term is the corresponding energy of
the hole with effective mass m h⁎, and the last term is the energy of an exciton whose
electron–hole pair is separated by a distance r, in a material having a dielectric
constant ε. In solid-state physics one uses effective masses for the electrons and
holes, expressed in multiples of the mass of a free electron, to model their motions as
if they themselves were free entities. Their values are specific to the semiconducting
material of the quantum dot, but in general, m h⁎ > m e⁎. The three quantum numbers
(n1, n2, n3) specify the three-dimensional quantised energy levels for the hole and
electron involved in the transition.
The important point is that ΔE is strongly sensitive to the radius of the quantum
dot, d. The absorption spectrum of a quantum dot is quite broad, extending from
above its band gap into the ultraviolet. The excited state however, relaxes rapidly to
the bottom of the conduction band upon which the electron–hole recombines to
produce photoluminescence over a narrow wavelength range determined primarily
by Eg. As shown in figure 12.3, one can thereby generate any colour in the visible
simply by changing the radius of the quantum dot, in this case made from CdSe. The
material is chemically the same in each case; only the sizes of the particles differ.
Applications of quantum dots abound, ranging from cancer treatment to television
display pixels. When the surface of a quantum dot is functionalized to become water
soluble, its potential toxicity is obviated and it becomes particularly useful for medical
imaging. Compared to fluorescent dyes, for example, quantum dots have tunable,
intense emission outputs, they exhibit long fluorescence lifetimes and negligible
photobleaching—and they are readily conjugated (chemically added) to proteins.

12-8
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 13
Fluorescence resonance energy transfer

13.1 Mechanism for intermolecular energy transfer

In a variety of complex materials, a close positioning of chemically different chromo-
phore units leads to a remarkable phenomenon: the absorption of light by one molecular
species produces fluorescence from a neighbour. The effect occurs in a wide range of
materials, including laser crystals and multi-component solutions; it proves especially
important in samples of biological origin. In any such process, excitation passes from
chromophores initially excited by photon absorption—these acting as ‘donors’—to
‘acceptor’ chromophores with suitably positioned energy levels. The latter, acting as
fluorophores, generate the fluorescence emission that is then observed. The effect is most
readily apparent in systems comprising two or more chromophores with well-charac-
terised absorption and fluorescence bands at broadly similar, but experimentally
differentiable wavelengths. Figure 13.1 exhibits a typical spectral discrimination,
indicating the wavelength response of suitable optical bandpass filters.
The transfer of energy between molecules or chromophores most commonly
occurs by a mechanism known as resonance energy transfer (RET—also known as
electronic energy transfer, EET). The term ‘resonance’, which indicates overall
energy conservation, also conveys the fact that energy can transfer from any level
within the vibrational continuum of the donor excited state, to a corresponding level
in the excited state continuum of the acceptor. One other prominently used acronym
in such studies is FRET, in which the added prefix can be taken to denote either
‘fluorescence’ or ‘Förster’. A quite different mechanism, known as Dexter energy
transfer, only ever operates between units sufficiently close for significant spatial
overlap of their electronic wavefunctions.
Consider such a transfer of excitation between two typical chromophores, an
initially excited donor A and acceptor B. A typical RET process may be expressed
by the equation:
RET
A* + B ⎯⎯⎯⎯→ A† + B*† (13.1)

doi:10.1088/978-0-7503-3683-3ch13 13-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 13.1. Typical spectral discrimination between the fluorescent emission from donor and acceptor species
(notionally a cyan fluorescent protein donor and a yellow fluorescent protein acceptor): (a) the transmission
characteristics of a short-wavelength filter will ensure initial excitation of only the donor. Separating the
different wavelengths with a beam-splitter and another narrow emission filter ensures only the Stokes-shifted
fluorescence from the donor reaches a detector; (b) in the same system a longer-wavelength emission filter
ensures capture of only the acceptor fluorescence, following energy transfer.

where the asterisk denotes an electronic excited state and the dagger, vibrational
excitation. In systems where RET occurs, the donors and acceptors are usually
fluorophores, i.e. chromophores that with a capacity to exhibit significant fluores-
cence decay. Moreover, the transitions of donor decay and acceptor excitation are
generally electric dipole-allowed—although other possibilities occasionally arise.
Accordingly, the most common mechanism of energy transfer, for donor–acceptor
displacements beyond significant wavefunction overlap, is electrodynamical cou-
pling between transition dipoles. With an inverse cubic dependence on distance,
much like the form of dipole–dipole interaction between two static dipoles, the rate
of energy transfer depends on the square of this coupling, and hence an inverse sixth
power dependence on distance.
For any donor–acceptor separation, R, usually a distance substantially smaller than
the wavelengths of visible radiation—a distance regime known as the near-field—a
widely used Förster theory gives the following rate constant for RET:
9κ 2c′4
kRET =
64π 4τ A*R6
∫ FA(ν)σB(ν) dνν4 . (13.2)

Here, FA(ν) is the normalised fluorescence spectrum of the donor and τA* is the
associated radiative decay lifetime; σB(ν) is the absorption cross-section of the
acceptor, c′ the speed of light in the host material. The overall rate and efficiency of
RET thus involves a spectral overlap, the integrated frequency-weighted product of
donor emission and acceptor absorption profiles. The parameter κ2 is a numerical
orientation factor, its value often close to unity. As is apparent from the inverse sixth
power dependence on R, the likelihood for energy to transfer between chromophores

13-2
 Molecular Photophysics and Spectroscopy (Second Edition)

is severely restricted by separation; any longer-range transfer therefore usually takes the
form of a series of small hops between near-neighbours, as indeed happens in
photosynthetic light-harvesting complexes. It is known that the distance dependence
actually changes beyond the near-field; a unified theory shows that it eventually acquires
the inverse square dependence associated with radiative transfer. However, in the most
common applications, distances are seldom more than a few nanometres—sufficiently
small for the Förster theory to be an extremely good approximation.

13.2 Spectroscopic shift

For each chromophore involved in RET, the fluorescence peak is generally Stokes-
shifted to a lower frequency—a longer wavelength—than its absorption counterpart,
for reasons associated with intramolecular vibrational relaxation, as established in
previous chapters. This shift is exemplified for the donor (A) in figure 13.1(a).
Commonly the peak of the acceptor absorption curve lies at a lower frequency than
that of the donor emission; the relative energy levels are shown in figure 13.2.
In considering the preferred direction to the traffic between any two chromo-
phores involved in energy transfer, it will be immediately apparent that there is a
close similarity of form between the equations for ‘forward’ and ‘backward’ transfer
between any given pair of electronic levels. The distance aspects for forward and
backward transfer are obviously the same; the key to directedness is due to
differences in the relative spectral profile overlaps. This is exemplified in the very
small area of overlap shown in figure 13.1(b); the species A cannot be at all effective
as an acceptor for fluorescence emitted by B. Thus it emerges that, as a result of the
intramolecular vibrational losses that usually follow directly after electronic
excitation, ‘forward’ transfer, A to B, is generally favoured over back-transfer.
This is a principle that is powerfully effective in the chains of energy transfer events
responsible for the primary processes in photosynthesis, where energy initially
absorbed by chlorophyll molecules rapidly progresses by a sequence of RET hops
towards a reaction centre, where the necessary electron transfer processes begin.
Without the spectroscopic shift between successive components along each chain of
events, the remarkable efficiency of this natural mechanism would be lost.

Figure 13.2. Resonance energy transfer from donor (left) to acceptor (right). The shaded boxes indicate the
vibrational continua of ground electronic (singlet) states S0 and relevant excited singlet states Sn—most often in
practice the first electronic excited state S1.

13-3
 Molecular Photophysics and Spectroscopy (Second Edition)

13.3 Distance measurements

The equation for the rate of Förster energy transfer can be expressed in the following
simplified form. It is written in terms of a Förster distance, R0, defined as the donor–
acceptor distance at which the rates of fluorescence from the donor and competing
energy transfer to the acceptor are equal. Hence, we find:
3κ 2 1 ⎛ R 0 ⎞6
kRET = ⎜ ⎟ . (13.3)
2 τA* ⎝ R ⎠
For example, in a donor–acceptor pair for which R0 = 14 Å, the rate of energy
transfer over a distance of 7.5 Å would be (14/7.5)6 = 42 times larger than the
competing rate of direct fluorescent decay by the donor.

Exercise
Consider a certain donor–acceptor pair separated by the Förster distance. It is found
that when the energy transfer acceptor moves away from the donor, to a separation
of 19 Å from the donor, the extent of its fluorescence due to FRET (fluorescence
resonance energy transfer) is diminished to about 0.1% of its occurrence at its
previous location. Find the Förster distance.

Answer
Acceptor fluorescence depends on (R0/R)6, so (R0/19 Å)6 = 10−3(R0/R0)6, hence
(R0/19 Å)6 = 10−3, thus (R0)6 = 10−3 × (19 Å)6 and so R0 = 0.316 × 19 Å = 6.00 Å.
In fact, the detailed distance dependence of RET provides a basis for the
identification of relative motions in molecules, or parts of molecules. There are
many examples in structural biology, such as the molecular traffic across a cell
membrane, and the mechanisms of protein folding. These and other such processes
can be registered by selectively exciting one chromophore, and observing the
generally longer-wavelength fluorescence from the other chromophore as it adopts
the role of acceptor—see figure 13.3. In cases where the components of interest do

Figure 13.3. Energy transfer for the detection of protein–protein interactions. Chemical interactions between
proteins result in the attached chromophores becoming closer in proximity, allowing transfer of energy
between them. Emission wavelengths differ when energy transfer is present.

13-4
 Molecular Photophysics and Spectroscopy (Second Edition)

not have suitable overlapped energy levels, molecular tagging with site-specific
extrinsic (i.e. artificially attached) chromophores can solve the problem. Lanthanide
ions, with characteristically sharp absorption features, prove especially useful. In
some applications the actual distance between the chromophore groups is of specific
interest, and the scale of their separation can be quantitatively assessed by
comparing RET efficiencies. This technique is popularly known as a spectroscopic
ruler.
The polarisation features of FRET are also interesting to pursue. FRET that
occurs due to energy transfer between different units held at a fixed mutual
orientation within freely mobile molecules have polarisation features that conform
to the interpretations provided by equations (12.10) and (12.11) in the previous
chapter. If, however, the donor and acceptor are unattached and they can undergo
uncorrelated rotational motions, a degree of polarisation anisotropy can still persist
in the acceptor fluorescence, though its extent will be small and evident only at very
short times.

13-5
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 14
Chiral phenomena and optical activity

14.1 Criteria for chirality in matter and in light

The vast majority of the molecules in biology have the property of being chiral: they
differ from their mirror images, which would have identical chemical composition
and bonding. Since, like our two hands, such opposite forms are non-superimpos-
able, they are often designated either left- or right-handed. Usually the two forms,
known as enantiomers, do not readily interconvert. For example, most natural
amino acids are left-handed; most natural sugars are right-handed. Almost all of the
substances of which the plant and animal kingdom are made, and also our food and
many drugs, are chiral. This feature is crucial to our biology: some molecules of the
opposite handedness prove deleterious to our health. The drug thalidomide proved a
horrendously tragic example; one enantiomer was safe, but it needed to be separated
from its teratogenic counterpart.
For individual molecules, the property of being different from a mirror image is
identifiable by the absence of an inversion centre. This can be understood by
recalling the symmetry elements introduced in chapter 3. Inversion in all three
spatial dimensions, i, is equivalent to the improper rotation S2, which itself
comprises mirror reflection, σ, and C2 rotation about a perpendicular axis, as shown
in figure 14.1: the order of these operations is irrelevant. The C2 operation does not
change the enantiomeric form of the molecule. So, chiral molecules do not conform
to inversion symmetry; the inversion operator i is not one of the symmetry elements
of their molecular point group. Equally, by extension of the same arguments, since
S1 ≡ σ and S2 ≡ i, the full criterion for chirality can be succinctly expressed as a lack
of Sn (improper rotation) elements of any order n. Such molecules must belong to
one of the pure rotation groups.
Before considering any molecular example, it will be instructive to consider how
the concept of chirality features in the general scheme of physics, where all physical
laws conform to CPT symmetry. This means they are invariant under a combination
of charge conjugation C, spatial parity inversion P and time reversal T. For optical

doi:10.1088/978-0-7503-3683-3ch14 14-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 14.1. Equivalence of parity inversion (top) to successive reflection and rotation (bottom).

interactions, charge conjugation (exchanging positive and negative charges) is never

a realisable issue, so we can restrict attention to PT rules. Previous chapters have
considered only electric dipole interactions—which are, indeed, by far the most
prominent. Electric fields and dipoles associated with charge separation have a clear
direction, and so they are odd with respect to P, but being invariant in time they are
even under T. Conversely, magnetic fields and dipoles associated with charge
circulation are even under P, odd under T (think of each operation on a rotation).
So for chiral molecules which are neither even nor odd under P, breaking spatial
parity, their interactions with electromagnetic radiation permit transitions that are
both electric and magnetic dipole allowed.

14.2 Circular dichroism

The simplest manifestation of chirality occurs in optical absorption. In the explicit
form of the light–matter interaction operator, which we previously could express as a
scalar product of the electric transition dipole with the electric field vector,
Hint = −μf0 · E, we must now include a term that couples the magnetic dipole to the
magnetic field, i.e.
Hint = μf0 · E − mf0 · B. (14.1)
where mf0 is the magnetic transition dipole and B is the magnetic field.
The reason that both electric and magnetic dipoles are involved in chiral effects
can be understood by noticing that, in the course of a transition, both translational
and circular motions of charge can be involved—and these are exactly the forms of
electric and magnetic dipoles, as illustrated in figure 14.2.
In fact, it is the quantum interference of electric and magnetic couplings that
provides the basis for chiral discrimination, in optically active molecules. Since the
angular momentum associated with orbital motions of charge has quantum values in
multiples of ℏ, and electric dipoles typically separate electron charges e by
dimensions on the order of the atomic length unit a0, it emerges that the relative

14-2
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 14.2. In an optical transition of a (notional) chiral molecule, left, the magnetic dipole transition
moment (purple double-headed arrow) and its electric dipole counterpart (red arrow) may be co-aligned. The
former involves electronic angular momenta; the right-hand panel is a reminder that such rotational quantities
l = r x p are invariant to space inversion.

Figure 14.3. Depiction of the hexameric structure contained in a unit of human insulin protein, exhibiting the
threefold rotation symmetry. With no other symmetry elements present, this signifies C3 point group symmetry.

magnitude of the couplings due to the electric and magnetic terms in the above
expression is around the same value as the universal fine structure constant:
e2
α= (14.2)
4πε0ℏc
giving α ~ 1/137. In terms of contribution to an observed spectral line, the small
magnitude of the result means that magnetic coupling is usually unimportant, and
for most transitions in achiral (non-chiral) molecules it is not always allowed by the
selection rules, if the electric dipole coupling is allowed. In fact, due to its link with
charge circulation, transitions are only magnetic dipole allowed if the symmetry of
irreducible representation of the excited state is the same as that of a rotation, Rx, Ry
or Rz. But it turns out that for chiral molecules, x and Rx always appear in the same
representation, as do all other translation and rotation pairs: so for such molecules,
any electric dipole-allowed transition is also weakly magnetic dipole allowed.
An example of chirality for molecules in the pure rotation group C3, is the human
insulin hexamer shown in figure 14.3; the same symmetry class can also be found in
some inorganic complexes based on octahedral coordination. As is evident from the
corresponding character table, table 14.1, the rotations and translations behave in

14-3
 Molecular Photophysics and Spectroscopy (Second Edition)

Table 14.1. Character table for the Schoenflies point group C3; ε = exp(2πi/3).

Linear
functions, Quadratic
C3 E C3 (C3)2 rotations functions Cubic functions

A +1 +1 +1 z, Rz x +y,z
2 2 2
z3, y(3x2 − y2), x(x2 − 3y2),
z(x2 + y2)
E +1 +ε +ε* x + iy; Rx + iRy (x2 − y2, xy) (yz, xz) (xz2,yz2)[xyz,z(x2 − y2)]
+1 +ε* +ε x − iy; Rx − iRy [x(x2 + y2), y(x2 + y2)]

precisely the same way under each of the allowed symmetry operations. Each of the
irreducible representations supports chirality.
We find that the interaction term for magnetic coupling with light carries an
opposite sign for molecules of opposite handedness. Moreover, for molecules of one
particular handedness, this form of coupling term takes a different sign for left- and
right-handed circular polarisations. As a result, the slight difference in absorption
strengths for left- and right-handed light can be quantified by what is known as the
dissymmetry factor, usually defined in terms of extinction coefficients (molar
absorption coefficients) ε:
ε L − εR
g= 1 L (14.3)
2
(ε + ε R)
in which the denominator represents the average absorption. Typical values are in
the range 10-3–10−4. The exact value usually varies across the absorption band, and
can be plotted as a circular dichroism spectrum. Such methods can be very useful for
the conformational analysis of organic compounds, with great importance for the
drug industry.

14.3 Optical rotation

The dual involvement of electric and magnetic dipole interactions is also responsible
for optical rotation, the most familiar and earliest discovered manifestation of
molecular chirality—and the standard origin of the designation of chiral molecules
as being either left- or right-handed. The term describes observations of a rotation in
the plane of polarisation as polarised light propagates through a liquid—typically a
solution. The sense of that rotation, clockwise or anticlockwise, depends on which
molecular enantiomer is present—or in the case of a mixture, the degree of
enantiomeric purity (which can itself give information on the possible extent of
the interconversion process known as racemization). Again, the variation with
wavelength is also instructive, forming the basis for the technique known as optical
rotatory dispersion. Notably, however, the effect of optical rotation is not an
absorption phenomenon; it is a forward Rayleigh scattering process that engages
an interference involving both conventional polarizability-based forward scattering
and a chiral electric-magnetic counterpart.

14-4
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Chapter 15
Multiphoton absorption in molecules

Multiphoton processes involve the nonlinear concerted interaction of two or more

photons with individual atoms or molecules; the most widely studied cases involve
the pairwise absorption of photons from an intense beam.

15.1 Two-photon absorption

As illustrated in figure 15.1, there are several essential differences between (a) the
concerted process of two-photon absorption, and (b) two sequential single-photon
absorption events. The latter only takes place after a molecule has already been
promoted into an excited state by a single-photon absorption. The second photon
(with possibly a different wavelength) then populates a second higher excited state.
In genuine two-photon absorption, both photons are usually derived from the same
input beam and therefore have the same energy; moreover, only one excited state at
the two-photon level becomes populated.
The phenomenon of non-resonant two-photon absorption was predicted by
Goeppert Mayer in 1931, but first experimentally observed by Kaiser and Garrett
in a fluorite crystal only in 1961, soon after the invention of the laser. The key to the
discovery and proof was the observation of fluorescence emission at a wavelength
substantially shorter than the input light; as shown in figure 15.1 the decay of a two-
photon excited state releases a photon that is much more energetic than the input
photons. The long delay between the development of theory and its experimental
realisation reflects that fact that only the high intensities delivered by lasers produce
sufficient photon flux to allow the nearly simultaneous absorption of two photons.
Crucially, it emerges that the efficiency of any such process varies with the square of
the input intensity.

15.2 Non-resonant two-photon absorption

Away from any single-photon resonance, as indicated in figure 15.1(a), the
process of two-photon absorption is very weak. Consider the following estimate.

doi:10.1088/978-0-7503-3683-3ch15 15-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 15.1. (a) Two-photon absorption from a single input beam can populate a highly excited state that is
inaccessible by single-photon absorption, due to different section rules; the ensuing decay emission has a
wavelength shorter than the input; (b) consecutive absorptions of different wavelength photons also populates
a highly excited state, but the main channel of decay from the first excited state produces emission with a
wavelength longer than the input.

For two-photon absorption to occur, it is clearly necessary for two photons to pass
essentially simultaneously through the region of space occupied by one molecule (or
else the space occupied by a chromophore, in the case of localised absorption in a
large molecule). The likelihood of this depends on the intensity of light and the
volume, and can be estimated quite easily using the following formula:
n Iλ
= 2 (15.1)
V hc
where n is the mean number of photons in a volume V, given an irradiance I of light
with an optical wavelength λ. For any one molecule in a liquid the mean volume is
given by the molar volume divided by Avogadro’s number L, i.e. VM/L = M/Lρ,
where M is the molar mass, and ρ is the density. Hence, for the instantaneous mean
number of photons n̄ per molecular volume (which will be a number much less than
one) we have:
IMλ
n¯ = . (15.2)
Lhc 2ρ
Let us see, for example, how this formula might be applied to an experiment on a
sample of benzene, whose molar mass M = 7.811 × 10−2 kg mol−1, and STP
(standard temperature and pressure) density ρ = 8.765 × 102 kg m−3. Consider a
focused incoherent UV source delivering a continuous irradiance of 107 W m−2, at a
wavelength of 255.0 nm corresponding to the absorption maximum of the sample.

15-2
 Molecular Photophysics and Spectroscopy (Second Edition)

Using equation (15.2) we find that the ‘photon density’ is approximately 6.3 × 10−12
photons per molecule, or one molecule in around 1.6 × 1011. With this level of
intensity, few molecules experience the transit of even a single photon at any instant
of time, though this is enough to produce a strong and easily measured absorption
signal: clearly the measured absorption could even be 100% if every photon were to
be absorbed by one of the molecules it encounters. However, the probability of two
photons simultaneously traversing a molecule, which depends on the square of the
irradiance, is approximately 4 × 10−23, corresponding to only one in 2.5 × 1022—
equivalent to only a couple of dozen molecules per mole. As experiment verifies,
there is no chance of observing any two-photon absorption at this level of intensity.
But consider a pulsed laser source operating at the same wavelength, and whose
beam is focused into the sample. Now, peak intensities of 1014 W m−2 can easily be
delivered to the sample. Whilst this increase by a factor of 107 results in a roughly
proportional increase in the probability of finding one photon in a molecular
volume, the probability of finding two photons increases by a factor of approx-
imately 1014, signifying approximately one molecule in around 2.5 × 108; easily
enough for a two-photon absorption process to be detected. This illustrates the fact
that with pulsed lasers, we may indeed expect to be able to detect two-photon
absorption in most media. Indeed results abound on systems ranging from single
atoms and inorganic crystals through to numerous complex biochemical structures.
Non-resonant two-photon absorption is related to the energy–time uncertainty
1
principle ΔE Δt ≥ ℏ. Consider again the case where two long-wavelength photons
4
together provide the energy for an electronic transition at the two-photon level.
Bearing in mind molecular dimensions, we must assume both photons are located at
the molecule within a brief interval Δt. This leads to a virtual state with a large ΔE
that is badly energy non-conserving when there is no real state near the single-
photon level. Accordingly, the molecule can exist within that virtual state only if a
second photon absorption restores energy conservation within a very short time, as
dictated by the time–energy uncertainty principle. This is why high light intensities
are needed to achieve an appreciable two-photon absorption.

15.3 Resonant two-photon absorption

The converse case arises when the incident laser frequency is near one-photon
resonant with an excited state, as in figure 15.1(b). Since electronic states are seldom
equally spaced in energy, this may require the absorption of two photons of differing
wavelength, as for example under conditions of two-beam excitation. From an
energy–time uncertainty perspective, the satisfaction of near energy conservation in
single photon absorption renders a small ΔE in a state that can, therefore, can exist
for a much longer time. This greatly increases the probability that the next photon
will arrive within the necessary time window. As a result, the probability of a two-
photon transition becomes much larger even when using lower intensity laser
sources such as continuous wave Ar+ or Kr+ lasers. Then the process can be
completed by the concerted absorption of a second photon, without the need to fulfil

15-3
 Molecular Photophysics and Spectroscopy (Second Edition)

the exacting conditions for both photons to arrive at essentially the same instant. In
the limit where an exact one-photon resonance occurs and ΔE = 0, there is no longer
any temporal constraint, save for the decay lifetime of the resonant state, because
that state can be physically populated by single-photon absorption; however, doing
so will completely dominate and mask any two-photon absorption.
As with the original Kaiser and Garrett experiment, one observation that verifies
the population of an excited state through two-photon absorption is the subsequent
fluorescence at wavelengths much shorter than the input. To illustrate: two photons
of 700 nm visible laser light convey the same energy as one photon of 350 nm
ultraviolet radiation; fluorescence after IVR losses might for example occur at 450
nm, shorter than the input. With a filter used to cut out any fluorescence near to and
beyond the input wavelength, such substantially shorter wavelength fluorescence can
only result from multiphoton excitation.
Another sensitive detection method, particularly for gas-phase molecules, is to use
laser input of a wavelength such that the sum energy of three incident photons
exceeds the ionization limit of the molecule; one can then monitor the production of
electrons or molecular ions by non-resonant two-photon ionization. Detection of
ions using a mass spectrometer is particularly useful if there are more than one
absorbing species in the sample.
If a two-photon absorption populates an excited state, one can detect electrons or
ions in a process called resonance enhanced multiphoton ionization or REMPI. Two-
photon absorption followed by ionization is called one-colour (2 + 1) REMPI,
where the first number in the brackets indicates that two photons are required to
pump the molecule to the excited state, and the second number indicates that one
additional photon is required to ionize the molecule from the excited state. One-
colour simply means that all photons come from the same laser. In general, one can
have (n + m) REMPI depending on the lasers available and the molecule. Ionization
limits for typical organic molecules range from 7 to 13 eV. Therefore, one could use
a (1 + 1) or (2 + 1) scheme using UV radiation. On the other hand, a molecule such
as N2, having an ionization limit in excess of 15.5 eV would require (2 + 2) REMPI.

15.4 Two-photon spectroscopy

We now turn specifically to the spectroscopic properties of the two-photon process.
As illustrated in figure 15.1, it is not necessarily the case that the same transitions can
be induced by single-photon absorption from a beam of frequency 2ν. In centrosym-
metric molecules, for example, where the usual absorption selection rules permit
only transitions between states of opposite parity, i.e. gerade–ungerade, two-photon
selection rules permit only transitions between states of the same parity, i.e. gerade–
gerade or ungerade–ungerade. Indeed, two-photon absorption in the ultraviolet/
visible range stands in the same relationship of complementarity to conventional
photon absorption as Raman scattering does to infrared absorption. Even for
compounds with transitions which are both one- and two-photon allowed, two-
photon spectroscopy using visible light usefully provides access to states that would
otherwise require ultraviolet excitation.

15-4
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 15.2. Two-photon absorption spectra of 131Xe132Xe with a Σ+g ground state: (top) plane polarised light;
(bottom) circularly polarised light. The bands which are reduced in intensity in the lower spectrum involve Σ+g
excited states. Adapted from Dimov S S et al 1995 Chem. Phys. Lett. 239 332–38, with permission of Elsevier,
Copyright 1995.

A related observation is that the two-photon spectra, in striking contrast to

conventional single-photon absorption, exhibit band intensities that vary with laser
polarisation, as shown for the van der Waals molecule Xe2 in figure 15.2. One can
define the polarisation R0:
Icp
R0 = , (15.3)
Ilp

where Icp and Ilp are the two-photon signal intensities measured on excitation with
input light that is circularly polarised input (photon angular momentum = ±1), and
linearly polarised (photon mean angular momentum = 0), respectively. Detailed
analysis shows that the value of the ratio R0 is expected to be 1.5 for two-photon
transitions that take place between electronic states of different symmetry, but
significantly <1.5 for two-photon transitions between electronic states of the same
symmetry. This is yet again another manifestation of the conservation of angular
momentum. The benefit of polarisation studies is that insight into excited symme-
tries can be had in the absence of rotationally resolved vibronic spectra.
Since the rate of two-photon absorption depends quadratically on the irradiance,
but still linearly on the concentration [C ] of absorber, the associated rate of loss of
intensity of light passing through a distance ℓ in the absorbing medium along the
z-direction is given by a modified Beer’s Law equation:
dI
− = β(λ)I 2[C ] (15.4)
dz

15-5
 Molecular Photophysics and Spectroscopy (Second Edition)

where β is the wavelength dependent two-photon absorption coefficient. Given an

input irradiance I0, the solution takes the simple form
I = I0 /(1 + β ℓC ). (15.5)

Hence the usual Beer’s Law does not hold for this process, nor do the concepts of
absorbance and molar absorption coefficient as normally defined.
One additional feature of interest feature is the possibility of Doppler-free
spectroscopy in the gas phase. If a mirror is placed such that the laser beam
traverses the sample in both directions, then each molecule of the sample is
effectively irradiated by two counterpropagating beams of wave-vectors k and –k,
as illustrated in figure 15.3. The two Doppler-shifted frequencies ‘seen’ by a sample
molecule travelling with velocity v and a component vk in the direction of the left-
hand beam have equal and opposite ‘red’ and ‘blue’ shifts, so that two-photon
transitions in which counterpropagating photons are absorbed provide an excitation
in which the Doppler shifts essentially cancel; ΔE = hνk′ + hν−′ k = 2hν . The result is
a readily resolved, sharp and symmetrical Doppler-free absorption line, on top of a
broader band due to the two-photon absorption of photons travelling in the same
direction, for which the usual Doppler shift occurs.

15.5 Higher order processes

Multiphoton studies where more than two photons are absorbed are generally based
on a single beam of laser light, and transitions are subject to the condition
mhν = ΔE, where m is an integer. Once again, the selection rules differ from
conventional absorption, and indeed they are distinctive for each value of m. The
intensity of absorption, in the absence of resonance enhancement, depends on I m,
where I is the laser beam irradiance. Very few electronic spectra involving more than
three or four UV/visible photons in the excitation process have been recorded.
However, a distinction should be made before generalising. Whilst three and four-
photon absorption generally represent the limit for UV–visible spectroscopic
purposes (largely because the absorption of any greater number of photons in this

Figure 15.3. Two-photon absorption with counterpropagating beams, photons in each direction having
Doppler shifts of opposite sign.

15-6
 Molecular Photophysics and Spectroscopy (Second Edition)

frequency region would lead to ionization), transitions involving the absorption of

many more photons (20–30 for example) have been observed in the infrared laser
excitation of polyatomic molecules, although usually all within the ground electronic
state, and commonly leading to relatively non-selective dissociation (i.e. chemical
bind fission and molecular fragmentation).

Exercise
Using the fact that the process of (non-resonant) three-photon absorption displays a
cubic dependence on the instantaneous intensity of an excitation beam, derive a
relationship for the dependence of intensity on the distance of beam travel within a
medium that exhibits three-photon absorption.

Answer
Using k for a constant of proportionality, I for the intensity, [C ] the concentration of
absorber and z the distance, we can write
dI
−
dz
= kI 3[C ] ⟹ − ∫ dI
I3
1
= ∫ k [C ]dz ⟹ 2 =
2I
k [C ]ℓ + const.

where ℓ is the pathlength through the absorber in the z-direction. At z = 0 the input
intensity can be defined as I0, so substituting in we find the constant is 1/2I02. Hence:

1 1 2kIo2[C ]ℓ + 1 Io
= k [C ]ℓ + = ⇒I=
2I 2 2Io2 2Io2 2kIo2[C ]ℓ +1

where the positive square root is selected since light intensity cannot be negative.
As noted above (REMPI) spectroscopy allows multiphoton absorptions to be
detected at significantly lower laser intensities, and that ion detection is particularly
sensitive. As the laser wavelength becomes shorter, the number of photons required
to promote molecules from their ground state to the ionisation continuum drops,
and the ion current increases accordingly. However, by far the largest signals are
obtained under resonance conditions when the energy of one, two or three photons
coincides with that of a bound state, as shown in figure 15.4. Fragmentation patterns
also typically change according to the excitation wavelength.
REMPI is not limited to the use of one laser. If two lasers are involved the process
is called two-colour REMPI. Here the notation is (n + m′) REMPI, where the first
number indicates the number of photons n from the laser used to excite the first
transition while the m′ is the number of photons from a second laser required to
ionize the molecule, following excitation. Typically, the first laser is tuned through
an electronic state to excite rovibronic transitions while the second laser has a fixed
wavelength sufficient to ionize the molecule, regardless of the wavelength range of
the first laser.

15-7
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 15.4. In this example of REMPI, as the laser frequency is tuned between νl and ν2, multiphoton
resonances (encircled) are observed at νA (three-photon), νB (one-photon) and νC (two-photon). The orange
circle identifies the threshold for thee-photon ionization. For example, at frequency νC, three photons are
responsible for the ionization process itself, but it is the two-photon resonance that results in a peak in the
spectrum; this is commonly represented as (2 + 1)-ionization.

15.6 Multiphoton imaging and processing

Recall that the rate at which single-photon absorption takes place is directly
proportional to the intensity, so that over any given period of time the amount of
light energy absorbed will be a specific fraction of the amount delivered to the
sample. This is why conventional spectroscopy generally deals with parameters
related to the change in intensity of the optical input. In consequence, there is not
usually any advantage to be gained in delivering the input over shorter times, or
within a smaller volume (for example by pulsing or focusing).
With multiphoton absorption, however, the situation is different. As we have
seen, the amount of excitation depends on an integer power of the irradiance.
Accordingly, if a given amount of light energy can be compressed into a shorter
interval, the fact that it is present for a shorter time is more than compensated by the
much greater increase in absorption during that interval. This is why multiphoton
absorption is almost invariably studied with pulsed laser radiation
But equally important are the gains to be had by spatially focusing the input.
Although focusing into a sample means that less molecules are intercepted by the
beam, there is a much larger increase in the response from those that are. Moreover,
since the transverse spatial profile of the beam is usually something like a Gaussian
distribution of intensity, the amount of multiphoton absorption in the centre of the

15-8
 Molecular Photophysics and Spectroscopy (Second Edition)

Figure 15.5. (a) Narrowing profile of intensity and its square, across a typical laser beam. (b) Three-
dimensional nanoscale model of a bull fabricated by Satoshi Kawata’s team at Osaka University in Japan,
based on two-photon photo-polymerisation of an acrylic gel at the programme-scanned focus of a laser beam.
The result won for the team a Guinness Book of Records prize for microfabrication.

beam is disproportionately large (due to its nonlinear intensity dependence), compared

to that which occurs elsewhere across the beam edge—see figure 15.5(a). In
consequence, the fraction of energy absorbed at the beam centre is much larger
than the fraction absorbed in other areas. This, in turn, means that it is possible to
create excitation at exceptionally well-defined positions. Two of the many applications
of this principle are micro- and nano-fabrication by two-photon induced polymer-
isation, (figure 15.5(b)) and multiphoton in vivo fluorescence imaging.

Exercise
Consider a well-behaved laser beam whose radial distribution of intensity has a
2
Gaussian form, I (r ) = I (0)e−2(r /w) where 2w is defined as the beam waist (which is
the radial distance from the centre of the beam to where the intensity drops to 1/e2 of
its centre-beam value). For a focused beam with w = 720 nm, calculate by what
factor the rate of three-photon absorption is less at r = 400 nm, compared to the rate
at the beam centre.

Answer
The rate of three-photon absorption depends on the cube of intensity, i.e.
2 2
[I (0)e−2(r /w ) ]3 = I 3(0)e−6(r /w ) .
So, comparing three-photon at the beam centre and at a radial position r, the relative
rate is:
2 2 2
e−6(r /w ) / e−6(0/w ) = e−6(r /w ) /1.

15-9
 Molecular Photophysics and Spectroscopy (Second Edition)

Hence with the given data we find e−6(1/1.8)2 = e−1.85 = 0.16. But even this
figure represents an upper limit for a process such as polymerisation triggered by
actual three-photon absorption.
The innovations in nanofabrication are typified by setups in which a laser beam is
focused into a photopolymerisable liquid such as an acrylic gel, and the focus is
scanned under pre-programmed computer control. In this way, a three-dimensional
solid of nanoscale dimensions can be sculpted, usually by two-photon activation that
is effective only at the moving focus. The reason this method works so well is that
there is an effective threshold intensity for photo-polymerisation to occur; no
excitation occurs below a cut-off intensity well within the transverse profile, because
the photon flux outside those regions is just not high enough to initiate a multi-
photon process. Here, there are emerging applications in the fabrication of nano-
scale motor parts for optofluidics and for lab-on-a-chip analytical applications.
There are other advantages to the use of multiphoton absorption—especially
three-photon absorption—in addition to its sharp focus. For in vivo applications
based on fluorescence imaging there are good clinical reasons to prefer using long-
wavelength light to produce the necessary photoexcitation, as opposed to potentially
damaging short-wavelength UV light; moreover, longer input wavelengths suffer
less scattering. Thus, with a scanning laser and imaging equipment, surface and
deeper sub-surface layers of skin can be scanned, and the ensuing molecular
fluorescence mapped out. This enables the imaging of local concentrations of key
fluorophores, including photodynamic dyes that are selectively absorbed by cancer-
ous tissue.

15-10
 IOP Publishing

Molecular Photophysics and Spectroscopy (Second Edition)

David L Andrews and Robert H Lipson

Bibliography

For further reading on the majority of the topics covered in this book, most of the
standard physical chemistry textbooks provide decent coverage, save for some of the
more advanced material in the later chapters of the present volume. The choice is
huge, and individual books have their own slant—but most are considerably more
mathematical in their approach. Beyond these, there are many more specialized
books on specific optical techniques and interactions, most of them aimed at a
graduate readership. The following list is therefore only an indicative selection of
titles worth investigating.

Andrews D L and Bradshaw D S 2018 Introduction to Photon Science and Technology

(Bellingham, WA: SPIE Press)
Andrews D L 1997 Lasers in Chemistry 3rd edn (Berlin: Springer)
Atkins P W and de Paula J 2016 Elements of Physical Chemistry 7th edn (Oxford: Oxford
University Press)
Atkins P W and de Paula J 2017 Atkins’ Physical Chemistry 11th edn (Oxford: Oxford
University Press)
Bernath P F 2016 Spectra of Atoms and Molecules 3rd edn (Oxford: Oxford University Press)
Bishop D M 1993 Group Theory and Chemistry (Mineola, NY: Dover)
Brisdon A K 1998 Inorganic Spectroscopic Methods (Oxford: Oxford University Press)
Brown J M 1998 Molecular Spectroscopy (Oxford: Oxford University Press)
Carter R L 1997 Molecular Symmetry and Group Theory (New York: Wiley)
Cotton F A 1990 Chemical Applications of Group Theory 3rd edn (New York: Wiley)
Demtröder W 2019 Atoms, Molecules and Photons 3rd edn (Berlin: Springer)
Duckett S and Gilbert B 2000 Foundations of Spectroscopy (Oxford: Oxford University Press)
Dykstra C E 1991 Quantum Chemistry and Molecular Spectroscopy (New Jersey: Prentice Hall)
Engel T 2019 Quantum Chemistry and Spectroscopy 4th edn (New York: Pearson)
Engel T and Reid P 2013 Physical Chemistry 3rd edn (Boston, MA: Pearson Education)
Gil V 2000 Orbitals in Chemistry, A Modern Guide for Students (Cambridge: Cambridge
University Press)
Grinter R 2005 The Quantum in Chemistry: An Experimentalist’s View (New York: Wiley)

doi:10.1088/978-0-7503-3683-3ch16 16-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

Harris D C and Bertolucci M D 1990 Symmetry and Spectroscopy. An Introduction to Vibrational

and Electronic Spectroscopy (Mineola, NY: Dover)
Hollas J M 2004 Modern Spectroscopy 4th edn (New York: Wiley)
Hollas J M 2002 Basic Atomic and Molecular Spectroscopy (Cambridge: The Royal Society of
Chemistry)
Lakowicz J R 2006 Principles of Fluorescence Spectroscopy 3rd edn (Berlin: Springer)
Laidler K J, Meiser J H and Sanctuary B C 2003 Physical Chemistry 4th edn (Boston:
Houghton Mifflin)
Levine I N 2008 Physical Chemistry 6th edn (New York: McGraw-Hill)
Ogden J S 2001 Introduction to Molecular Symmetry (Oxford: Oxford University Press)
Silbey R J, Alberty R A and Bawendi M G 2005 Physical Chemistry 4th edn (New York: Wiley)
Valeur B and Berberan-Santos M N 2013 Molecular Fluorescence: Principles and Applications 2nd
edn (Weinheim: Wiley-VCH)
Vincent A 2001 Molecular Symmetry and Group Theory 2nd edn (New York: Wiley)
Wayne C E and Wayne R P 1996 Photochemistry (Oxford: Oxford University Press)

16-2
 Index

A Carbon–carbon bonds 6.1, 10.3

Abelian groups 3.2 Carbonyl group 2.4, 4.2, 4.3, 7.1
Absorbance 10.2 Carbonyl sulphide 5.3, 5.4
Absorption coefficient 10.2 Carboxylic acid group 7.1
Absorption coefficient, molar 10.2, 14.2 Carotene 10.3, 12.4
Acetaminophen 8.2 Centrifugal distortion 5.4
Action spectrum 10.2 Character tables 3.3, 14.2
Angular momentum, electron 2.1 Characteristic temperature for
Angular momentum, molecule 5.1 rotation 5.1
Angular momentum, photon 1.4, 2.2, Charge conjugation (symmetry) 14.1
14.2 Chemiluminescence 10.1
Anharmonicity 6.2, 9.3 Chirality 14.1–14.2
Anisotropy, fluorescence 12.3, 13.3 Chlorophyll 1.1, 9.4, 13.2
Atoms, multielectron 2.2 Circular dichroism 14.2
Aufbau Principle 2.2 Colour 9.4, 10.1–10.3
Auxochromes 10.3 Combination bands 4.3
Axis, principal 3.1 Commuting operators 3.2
Conjugation 10.3
B Correspondence principle 6.1
Band gap 12.5 Crystal field theory 9.4
Bathochromic shifts 10.3 Cyclohexanone 7.1
Beer’s Law 10.2
Benzene 3.1, 7.1, 9.4, 15.2 D
Benzophenone 4.3 Deformation modes 7.1
Bioluminescence 10.1 Degrees of freedom, nuclear
Birge–Sponer plot 6.2, 9.3 motions 4.2
Boltzmann distribution 4.2, 5.1, 6.2, 6.3, Density of states 4.3
8.2, 9.2, 9.3 Depolarisation ratio 8.3, 8.4 – see also:
Bond length determination 5.3 Polarisation, degree of
Born–Oppenheimer approximation Diatomic molecules 2.3–2.4
4.1, 9.2 Dichroism, circular 14.2
Bose–Einstein statistics 1.4 Diffusion, rotational 12.3
Boson 5.2 Dipole moment, electric 5.1, 6.1, 6.4,
Brackett series 2.1 8.1 – see also: Dipole, transition
Brønsted–Lowry acid 7.1 electric; Dipole, induced electric
Butane 7.2 Dipole, induced electric 8.1
Dipole, transition electric 2.1, 3.3, 5.1,
C 6.1, 9.2, 10.2, 12.2–12.3, 13.1, 14.2
Cadmium selenide 12.5 Dipole, transition magnetic 14.2
Carbon dioxide 4.2, 6.4, 8.2 Dissociation 6.2, 11.3
Carbon monoxide 5.1, 6.1 Dissymmetry factor 14.2

I-1 ª IOP Publishing Ltd 2021

 Molecular Photophysics and Spectroscopy (Second Edition)

Doppler effect, gases 1.2 Hot bands 6.2, 9.2

Doppler-free spectroscopy 15.4 Hund’s Rule 2.2, 11.2
Hydrocarbons 7.2
E Hydrogen atom 2.1
Einstein B-coefficient 10.2, 11.1 Hydrogen bonding 7.1, 7.2, 8.2
Electric field, optical 1.4, 8.1, 12.2, 14.2 Hydrogen chloride 5.1, 6.2–6.3
Electromagnetic spectrum 1.3 Hydrogen cyanide 5.3, 5.5
Enantiomers 14.1 Hydrogen molecule 5.2, 6.1
Energy transfer, resonance 13.1–13.3 Hyperchromism 10.3
Ethane 7.2 Hyperspectral imaging 7.3
Ethanoic acid 7.1 Hypsochromic shifts 10.3
Exchange interaction 11.2
Exciton 12.5 I
Extinction coefficient 10.2, 14.2 Identity operation 3.1
Infrared spectroscopy 6.1, 7.1–7.3
F Insulin 14.2
Fermi’s Golden Rule 11.3 Internal conversion 11.3
Fermion 5.2 Intersystem crossing 11.2, 11.3
Filters, optical 8.2, 13.1 Intramolecular vibrational relaxation
Fine structure constant 14.2 11.2, 11.3
Fingerprint region 7.1 Isotopomers 5.3
Fluorescence 11.2–11.3, 12.1–12.5,
13.1–13.3 J
Fluorescence excitation spectrum 10.2 Jablonski diagrams 11.2
Fluorescence imaging 12.4 j–j coupling 2.2, 2.3
Fluorophores 12.4
Force constant 6.1 K
Formaldehyde 2.4 Kronecker delta 8.3
Formic acid 4.3
Förster distance 13.3 L
Förster theory 13.1 Laporte selection rule 2.1
Franck–Condon factor 9.2, 9.3, 12.2 Lifetime, fluorescence, 12.1, 12.4
Ligands 9.4
G Line broadening 2.1, 4.2, 4.3, 7.2, 9.1, 9.3
Gaussian lineshape 2.1 Linear momentum, photon 1.4
Gerade and ungerade 2.1, 8.2, 15.4 Linewidth 1.2, 2.1 – see also; Line
Group frequencies 7.1 broadening
Lorentzian lineshape 2.1
H Luminol 10.1
Haem (heme) group 8.3 LUMO (lowest unoccupied molecular
Haemoglobin (haemoglobin) 7.3, 9.4 orbital) 2.4, 10.3
Hair, white 1.1 Lyman series 2.1
Herzberg–Teller effect 9.4
HOMO (highest occupied molecular M
orbital) 2.4, 10.3 Magic angle 12.3

I-2
 Molecular Photophysics and Spectroscopy (Second Edition)

Magnetic field, optical 1.4, 14.2 Paschen series 2.1

Methanal 2.4 Pauli Exclusion Principle 2.2, 11.2, 12.5
Methane 2.4 Phases of matter 1.2
Methane 7.2 Phosphorescence 11.2, 11.3
Methanoic acid 4.3, 7.1 Photometry 10.2
Microwave spectroscopy 5.3, 7.2 Photo-polymerisation 15.6
Moment of inertia 5.1, 5.3 Photoselection 12.3
Morse potential 6.2 Photosynthesis 13.2
Multiphoton absorption 15.1–15.6 Pi bonds 2.4
Multiphoton imaging 15.6 Point group 3.2
Multiphoton processing 15.6 Polarisation, circular 1.4, 14.2, 15.4
Mutual exclusion rule 8.2 Polarisation, degree of 12.3 – see also:
Myoglobin 8.3 Depolarisation ratio
Polarisation, photon 1.4
N Polarisation, plane 1.4, 8.3, 12.3, 15.4
Naphthalene 12.2 Polarizability 8.1–8.3
Near-infrared spectroscopy 7.3 Polyenes 10.3
Nitrogen molecule 5.1, 6.1 Predissociation 9.1
Normal modes of vibration 4.2 Protein interactions 13.3
Number of photons 15.2
Q
O Quantum dots 12.5
Operation, identity 3.1 Quantum number, azimuthal 2.1
Operation, reflection 3.1 Quantum number, azimuthal 5.1
Operation, rotation 3.1 Quantum number, magnetic 2.1
Optical activity 14.1–14.3 Quantum number, principal 2.1
Optical density 10.2 Quantum number, rotational 5.1
Optical rotation 14.3 Quantum number, spin 2.1
Optical rotatory dispersion 14.3 Quantum number, vibrational 6.1
Orbitals, atomic 2.1–2.2 Quantum yield, fluorescence 12.1
Orbitals, bonding and anti-bonding 2.4 Quenching, collisional 11.3
Orbitals, molecular 2.4
Ortho and para levels 5.2 R
Overtones, vibrational 4.3, 6.2 Racemization 14.3
Oxygen atom 5.2 Raman scattering 8.1–8.4
Oxygen molecule 5.1, 5.3 Raman scattering, resonance 8.4
Oxyhaemoglobin (oxyhemoglobin) 7.3, Rate constant 11.3
10.2 Rayleigh scattering 8.1, 14.3
Reduced mass 5.1
P Representation, irreducible 3.3
P, Q and R branches 6.3 Resonance energy transfer 13.1–13.3
Paracetamol 8.2 Resonance enhanced multiphoton
Parity, spatial 2.1, 3.1, 14.1 ionization 15.3, 15.5
Particle-in-a-box 10.3, 12.5 Rhodamine 6G 11.3, 12.4

I-3
 Molecular Photophysics and Spectroscopy (Second Edition)

Rotation, improper 3.1 T

Rotation, proper 3.1 Terahertz radiation 1.3, 7.2
Rotational constant 5.1, 5.3 Term symbols 2.2, 2.3
Rotational energy levels 4.2, 5.1–5.5 Three-photon absorption 15.5
Rotational spectroscopy 5.1–5.5 Time reversal (symmetry) 14.1
Russell–Saunders coupling 2.2, 2.3 Transition metal complexes 9.4
Rydberg constant 2.1 Transmittance 10.2
Triplet states 11.2, 11.3
S Tryptophan 8.3
Saturation 10.2 Tunnelling 6.1
Schoenflies notation 3.2–3.3 Two-level system 11.1
Screening 2.2 Two-photon absorption 15.1–15.4
Selection rule, Laporte 2.2 Tyrosine 8.3
Selection rule, rotational spectrum 5.1
Selection rule, vibrational spectrum U
6.1, 6.4 Uncertainty Principle, Heisenberg 1.2,
Sigma bonds 2.4, 6.1 2.2, 15.2
Simple harmonic motion 4.3, 6.1 Unified theory of energy transfer 13.1
Skeletal modes, 7.1
Sky colour 8.1 V
Solvatochromism 10.3 Vacuum ultraviolet 1.3
Spectral overlap 13.1 Van der Waals complexes 5.5, 9.1, 15.4
Spectrochemical series 9.4 Vibrational energy levels 4.2
Spectroscopic ruler 13.3 Vibrational spectroscopy 6.1–6.4
Spectroscopic shift in energy transfer Vibration–rotation spectra 6.3
13.2 Vibronic transitions 9.1–9.3
Spherical top, 5.5 Virial theorem 4.3
Spin, electron 2.1, 2.2, 9.4, 11.2 Voigt profile 2.1
Spin, nuclear 5.2
Spin, photon 1.4, 2.1 W
Spin–orbit interaction 2.2 Water 4.2, 6.4, 7.1, 7.2
Stern–Volmer plot 12.1 Wavenumber 1.3
Stimulated emission 11.1
Stokes and anti-Stokes Raman lines 8.2 X
Stokes shift, fluorescence 11.3, 13.1 Xenon dimer 15.4
Symmetric tops, oblate and prolate 5.5
Symmetry element 3.1 Z
Symmetry operation 3.1 Zero-point energy 6.1, 10.3

I-4