Seogjoo J.

Jang

Quantum Mechanics
for Chemistry
Quantum Mechanics for Chemistry
Seogjoo J. Jang

Quantum Mechanics
for Chemistry
Seogjoo J. Jang
Chemistry and Biochemistry
Queens College, CUNY
Queens, NY, USA

ISBN 978-3-031-30217-6 ISBN 978-3-031-30218-3 (eBook)

https://doi.org/10.1007/978-3-031-30218-3

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland
AG 2023
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
To my family
Preface

Modern chemistry was built on quantum mechanical foundations, but the amount
of time allowed for chemistry students to learn both fundamental principles of
quantum mechanics and details of its applications to molecular systems is very
limited. Therefore, instruction of quantum mechanics in typical undergraduate and
graduate chemistry curricula has been fairly limited in its depth. Rather, the focus
of conventional physical chemistry or quantum chemistry education has been more
on mastering facts and data resulting from applications of quantum mechanics that
are essential for developing key chemical intuition and reasoning ability. However,
in this age of unlimited accessibility to huge amount of data, the role of course
instruction as a means to deliver comprehensive information can be relieved to a
large extent. On the other hand, the need to lay a strong conceptual foundation
remains irreplaceable and may have become even more important considering the
demand for reasoning skills to assess information and data critically. This is the
main motivation for writing this book.
As the title Quantum Mechanics for Chemistry suggests, this book intends to
provide standard lessons on quantum mechanics as completely as possible, but also
focuses on helping readers understand implications and applications of those lessons
for molecular sciences. Thus, for example, standard lessons on quantum mechanical
vibration, rotation, and hydrogen-like systems are combined with key principles and
ideas of related spectroscopies in same chapters.
An important feature of this book is that the Dirac notation is introduced in a
rather abstract manner and as completely as possible from the outset. This is because
the Dirac notation serves as the most effective language to describe most quantum
mechanical ideas and is the major notation being used for research level publications
in physical chemistry and quantum chemistry.
Significant effort has also been made to provide clear quantum mechanical
understanding of subtle and qualitative details of atomic and molecular properties,
based on ideas and approximations that are as simple as possible. These lessons are
then complemented by presentations of more advanced theories and computational
methods in Appendices or later chapters of the book.

vii
viii Preface

A unique feature of this book is that lessons that are not treated in most quantum
chemistry textbooks such as general derivations of Fermi’s golden rule, path integral
approaches, and open system quantum dynamics theories are included while also
providing reasonable description of electronic structure calculation theories. This
is because new directions of research call for more balanced and comprehensive
understanding of the structure, the energetics, and the dynamics of molecular
systems altogether.
Intended readers of this book are upper level undergraduate chemistry students
or beginning graduate students. Chapters 1–9 can be used as a text or reference for
undergraduate level physical chemistry course. Chapters 1–11 can be used as a text
for the first year graduate level quantum chemistry course. Chapter 12 is to provide
preliminary lessons for those who plan to pursue theoretical and computational
research in physical chemistry and chemical physics. This book can also be a
useful reference for physics students and researchers who are interested in learning
molecular quantum mechanics and quantum dynamics methods.
This book is an outcome of notes that I have developed for 17 years while
teaching both undergraduate and graduate level quantum chemistry classes at
Queens College and the Graduate Center of the City University of New York
(CUNY), but major undertaking of writing this book started during the COVID-19
campus lockdown in 2020. Final completion of this book became possible through
supports from the Korea Institute for Advanced Study (KIAS) for summer visit
through its KIAS scholar program and from the Korea Advanced Institute of Science
and Technology (KAIST), in particular the KAIX program, for my sabbatical stay
in Fall 2022, during which I also taught a graduate level quantum chemistry course
for students at KAIST.
Hardly anything presented in this book is original. Most standard contents were
developed from textbooks and references that I used to learn and teach physical
chemistry and quantum chemistry. Key references that I should mention here are
Quantum Chemistry by McQuarrie [1], which was used for developing Chaps. 3–9;
Quantum Chemistry by Levine [2], which served as an amazing reference for all
the details of theories and methods that can be taught for quantum chemistry class;
and Quantum Mechanics in Chemistry by Schatz and Ratner [3], which helped me
learn broad subjects of quantum mechanics used in chemistry and also motivated
the choice of the title of this book. In addition, many ideas for lessons originated
from well-known books on quantum mechanics [4–8], group theory [9, 10], physical
chemistry [11–13] quantum/computational chemistry [14, 15], and some books on
advanced topics [16–18].
Completion of this book would not have been possible without assistances from
Dr. Murali Devi who read manuscripts carefully and offered numerous valuable
comments, Dr. Pablo Ramos who provided many key figures/illustrations and
suggestions for improving Chap. 11, and Taner Ture who also provided many
important figures and validated many equations. I also would like to thank all
the students at Queens College and the Graduate Center of CUNY and at KAIST
who took my classes and provided valuable feedbacks. In particular, YoungKyun
Preface ix

Ahn, Junsoo Lee, Sowon Kim at KAIST are appreciated for providing important
comments on earlier versions of the manuscript for this book.
Supports from the National Science Foundation, the Department of Energy,
and the Camille Dreyfus foundation for my research and teaching all have made
significant contributions to the development of this book. Encouragements and
comments from my colleagues and intellectual mentors also have played important
roles. In particular, I would like to thank Profs. George Schatz, Jeff Cina, Jae Woo
Park, Weita Yang, and Greg Voth, and Dr. Qin Wu for their positive feedback and
comments. Profs. Changbong Hyeon at KIAS and Young Min Rhee at KAIST
are also appreciated for hosting my visits and valuable comments. I also would
like to thank Sam Harrison of Springer Nature for generous consideration and
encouragement of this book. Most of all, my family’s support and love gave me
lasting power to work on this book project.

Queens, NY, USA Seogjoo Julian Jang

January 2023
Contents

1 Concepts and Assumptions of Quantum Mechanics. . . . . . . . . . . . . . . . . . . . 1

1.1 Assumptions of Classical Mechanics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Classical Point Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Wave: Classical View . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.3 Particle Versus Wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Concepts of Quantum Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.1 Blackbody Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.2 Photoelectric Effect and Photon . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.3 de Broglie’s Postulate of Material Wave . . . . . . . . . . . . . . . . . . 8
1.2.4 Heisenberg’s Uncertainty Principle . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.5 Wave-Particle Duality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3 Schrödinger Equation (in One Dimension)
and Probability Amplitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.1 Wavefunction and Time Dependent Schrödinger
Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.2 Measurement as Mathematical Operation
on Wavefunction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.3 Stationary States and Time Independent
Schrödinger Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.4 Eigenfunction and Eigenvalue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3.5 Linear and Hermitian Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.3.6 Results of Measurement and Expectation Value . . . . . . . . . 18
1.4 Quantum Particle in a One-Dimensional Box . . . . . . . . . . . . . . . . . . . . . . 20
1.4.1 Time Independent Schrödinger Equation. . . . . . . . . . . . . . . . . . 21
1.4.2 Time Dependent States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.4.3 Completeness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.5 Summary and Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Appendix: Dirac-Delta Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Exercise Problems with Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

xi
xii Contents

2 Dirac Notation and Principles of Quantum Mechanics . . . . . . . . . . . . . . . . 35

2.1 Formulation of Quantum Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.2 Ket, Bra, and Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.3 Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.3.1 Hermitian Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.3.2 One Dimensional Position Operator and Eigenket . . . . . . . . 42
2.3.3 One Dimensional Momentum Operator and Eigenket . . . . 44
2.3.4 Expressions for Momentum Operator . . . . . . . . . . . . . . . . . . . . . 45
2.3.5 Schrödinger Equations in the Dirac Notation . . . . . . . . . . . . . 46
2.3.6 Commutator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.3.7 Compatibility and Completeness . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.3.8 Measurement Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.3.9 Unitary Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.4 Particle in a One-Dimensional Box: Revisited with the
Dirac Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.5 Direct Product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.6 Summary and Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Appendix: Cauchy-Schwarz Inequality and a General
Uncertainty Relationship. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Exercise Problems with Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3 Harmonic Oscillator and Vibrational Spectroscopy . . . . . . . . . . . . . . . . . . . . 69
3.1 Classical Harmonic Oscillator and Hamiltonian . . . . . . . . . . . . . . . . . . . . 69
3.2 Schrödinger Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.2.1 Solution of Time Independent Schrödinger Equation. . . . . 72
3.2.2 Operator Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.2.3 General Time Dependent State. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.3 Vibrational Spectroscopy of Diatomic Molecules . . . . . . . . . . . . . . . . . . 83
3.3.1 Vibrational Absorption or Infrared (IR) Spectroscopy. . . . 85
3.3.2 Vibrational Raman Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.3.3 Anharmonic Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.4 Summary and Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
Exercise Problems with Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
4 Multidimensional Systems and Separation of Variables . . . . . . . . . . . . . . . 101
4.1 Three Dimensional System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
4.1.1 Position, Momentum, Hamiltonian, and
Schrödinger Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
4.1.2 Particle in a Three Dimensional Rectangular Box . . . . . . . . 106
4.1.3 Separation in Cartesian Coordinate System . . . . . . . . . . . . . . . 110
4.2 Many Particle Systems and the Center of Mass Coordinates . . . . . . 112
4.2.1 Two-Particle System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
4.2.2 Normal Modes and Vibrational Spectroscopy
of Polyatomic Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
Contents xiii

4.3 Summary and Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

Exercise Problems with Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5 Rotational States and Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
5.1 Rotation in Two Dimensional Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
5.2 Rotation in Three Dimensional Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
5.3 Angular Momentum Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
5.4 Spectroscopy of Rotational Transitions
for Diatomic Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
5.4.1 Microwave Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
5.4.2 Rotational Raman Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
5.4.3 Ro-Vibrational Transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
5.4.4 Centrifugal Correction and Ro-Vibrational Coupling. . . . . 155
5.5 Summary and Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Appendix: Associated Legendre Equations and Their Solutions . . . . . . . . . . 161
Exercise Problems with Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6 Hydrogen-Like Systems and Spin Orbit States of an Electron . . . . . . . . 171
6.1 Bohr’s Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
6.2 Solution of Schrödinger Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
6.3 Separation of Variables in Spherical Coordinate System . . . . . . . . . . . 175
6.3.1 Radial Equation and Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
6.3.2 Radial Probability Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
6.3.3 Eigenfunctions and Eigenstates in the Dirac Notation . . . . 180
6.3.4 Zeeman Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
6.3.5 Real-Valued Orbital Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
6.4 Spin States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
6.5 Electronic Transitions and Term Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . 188
6.6 Summary and Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
Appendix: Solutions of the Radial Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
Exercise Problems with Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
7 Approximation Methods for Time Independent Schrödinger
Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
7.1 Variational Principle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
7.1.1 General Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
7.1.2 Variational Principle for Trial States as Linear
Combinations of Basis States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
7.2 Time Independent Perturbation Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
7.2.1 Non-Degenerate Perturbation Theory . . . . . . . . . . . . . . . . . . . . . 212
7.2.2 Degenerate Perturbation Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
7.3 Summary and Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
xiv Contents

Exercise Problems with Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225

Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
8 Many Electron Systems and Atomic Spectroscopy . . . . . . . . . . . . . . . . . . . . . 241
8.1 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
8.2 Independent Electron Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
8.2.1 Major Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
8.2.2 Orbitals and Electronic Configuration . . . . . . . . . . . . . . . . . . . . . 244
8.2.3 Spin States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
8.2.4 Energy Levels of Spin-Orbit States . . . . . . . . . . . . . . . . . . . . . . . . 247
8.2.5 Examples of Energy Levels Based on
LS-Coupling Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
8.2.6 Atomic Spectroscopy: Selection Rules and
Simple Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
8.3 Case Study of Helium Atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
8.3.1 Hamiltonian and Schrödinger Equation . . . . . . . . . . . . . . . . . . . 254
8.3.2 Independent Electron Model with Variational
Optimization of Effective Charge . . . . . . . . . . . . . . . . . . . . . . . . . . 255
8.3.3 Self Consistent Field (SCF) Approximation
for Helium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
8.4 Self Consistent Field (SCF) Approximation for Many
Electron Atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
8.4.1 Hartree Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
8.4.2 Hartree-Fock Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
8.5 Summary and Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Exercise Problems with Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
9 Polyatomic Molecules and Molecular Spectroscopy . . . . . . . . . . . . . . . . . . . . 275
9.1 Born-Oppenheimer Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
9.2 Molecular Orbitals and Electronic Configurations
for Diatomic Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
9.2.1 Example of H2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
9.2.2 Molecular Orbitals and Electronic
Configurations of Diatomic Molecules . . . . . . . . . . . . . . . . . . . . 284
9.2.3 Molecular Electronic States of Diatomic Molecules . . . . . . 289
9.3 Conjugated Hydrocarbons and Hückel Approximation . . . . . . . . . . . . 292
9.3.1 Ethylene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
9.3.2 Butadiene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
9.3.3 π Orbital and Delocalization Energies . . . . . . . . . . . . . . . . . . . . 296
9.4 Molecular Symmetry and Group Theory. . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
9.4.1 Symmetry and Symmetry Operation . . . . . . . . . . . . . . . . . . . . . . 297
9.4.2 Group Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
9.4.3 Groups of Point Symmetry Operations . . . . . . . . . . . . . . . . . . . . 301
9.4.4 Matrix Representation of Point Symmetry
Group Elements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Contents xv

9.4.5 Application for Symmetry Adapted LCAO-MO . . . . . . . . . . 309

9.5 Spectroscopy of Polyatomic Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
9.5.1 Infrared and Raman Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . 311
9.5.2 Electronic Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
9.6 Summary and Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
Appendix: Important Theorems and Proofs in the Group Theory . . . . . . . . . 317
Exercise Problems with Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
10 Quantum Dynamics of Pure and Mixed States . . . . . . . . . . . . . . . . . . . . . . . . . . 331
10.1 Quantum Dynamics of Pure States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
10.1.1 Heisenberg Picture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
10.1.2 Interaction Picture and Time Dependent
Perturbation Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
10.1.3 Fermi’s Golden Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
10.2 Quantum Dynamics of Mixed Quantum States . . . . . . . . . . . . . . . . . . . . . 341
10.2.1 Density Operator and Quantum Liouville Equation . . . . . . 341
10.2.2 Time Dependent Perturbation Theory for Mixed
Quantum States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
10.2.3 FGR for Mixed States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
10.3 Summary and Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
Appendix: Interaction Hamiltonian in the Presence of Radiation . . . . . . . . . 355
Exercise Problems with Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
11 Theories for Electronic Structure Calculation of Polyatomic
Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
11.1 Hartree-Fock Approximation and Roothaan Equation . . . . . . . . . . . . . 364
11.1.1 General Single Determinant State . . . . . . . . . . . . . . . . . . . . . . . . . 364
11.1.2 Restricted HF Equation for Doubly Filled
Orbital States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
11.1.3 Linear Combination of Basis States . . . . . . . . . . . . . . . . . . . . . . . 376
11.1.4 Choice of Basis Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
11.1.5 Methods Beyond HF Approximation . . . . . . . . . . . . . . . . . . . . . . 380
11.2 Density Functional Theory. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380
11.3 Summary and Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
Exercise Problems with Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
12 Special Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
12.1 Path Integral Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
12.1.1 Real Time Propagator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392
12.1.2 Imaginary Time Propagator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
12.2 Quantum Master Equation for Open System
Quantum Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
xvi Contents

12.2.1 Projection Operator Formalism and Exact Time

Evolution Equations for a Projected Density Operator . . . 399
12.2.2 Quantum Master Equations for a Reduced
System Density Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404
12.3 Green’s Function Approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412
12.3.1 Second Quantization and Field Operators . . . . . . . . . . . . . . . . . 414
12.3.2 Ground State (Zero Temperature) Green’s Functions . . . . . 418
12.3.3 Nonequilibrium Green’s Functions . . . . . . . . . . . . . . . . . . . . . . . . 419

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427
Physical Constants and Abbreviations

Table 1 contains important physical constants as follows.

Table 1 Physical constants in SI units from the NIST reference data

Constant Symbol SI units
Planck constant h .6.62607015 × 10−34 J · s
Planck constant, reduced .h̄ .1.054571817 × 10
−34 J · s

Speed of light c 8
.2.99792458 × 10 m · s
−1

Boltzmann constant .kB .1.380649 × 10

−23 J·K −1

Electron mass .me .9.1093837015 × 10

−31 kg

Proton mass .mp .1.67262192369 × 10

−27 kg

Elementary charge e .1.602176634 × 10

−19 C
Rydberg constant .R∞ .1.0973731568160 × 10 m
7 −1

Bohr radius .a0 .5.29177210903 × 10

−11 m
Bohr magneton .βB .9.2740100783 × 10
−24 J · T−1

Vacuum permittivity .ǫ0 .8.8541878128 × 10

−12 C2 N−1 m−1

Hartree .Eh .4.3597447222071 × 10

−18 J

Unit Conventions

• Atomic units (a.u.) are in h̄ (action), e (electric charge), a0 (length), and me

(mass). Thus, the unit energy in a.u. is Eh = h̄2 /(me a0 ).
• eV is a unit of energy defined as electron charge e times volt (JC−1 ). Thus,
1 eV = 1.602176634 × 10−19 J.
• Wavenumber (cm−1 ) is energy divided by hc = 1.98644586×10−23 J·cm and is
used as a unit of energy in spectroscopy. Thus, 1 eV = 8.06554393 × 103 cm−1 .

xvii
xviii Physical Constants and Abbreviations

Abbreviations

BO Born-Oppenheimer
B.O. Bond Order
CI Configuration Interaction
DFT Density Functional Theory
FGR Fermi’s Golden Rule
HF Hartree-Fock
Irrep. Irreducible Representation
LCAO-MO Linear Combination of Atomic Orbitals as Molecular Orbital
MO Molecular Orbital
POVM Positive Operator Valued Measure
QME Quantum Master Equation
SCF Self-Consistent Field
TD-DFT Time-Dependent Density Functional Theory
Chapter 1
Concepts and Assumptions of Quantum
Mechanics

To exist is to be free from contradiction.

— Henri Poincare

Abstract This chapter introduces key concepts, major assumptions, and basic
formulations of quantum mechanics. For contextual understanding, assumptions
of classical mechanics are clarified first. Then, light quanta, material wave, and
uncertainty principles are explained. The wavefunction and Schrödinger equation
defined in a one dimensional space are introduced along with their major properties.
The chapter concludes with a solution of a quantum particle in a one dimensional
box in order to illustrate the major concepts and ideas of quantum mechanics.

1.1 Assumptions of Classical Mechanics

Before the development of quantum mechanics, classical mechanics was believed

to explain the state and the dynamics of all physical objects. The assumption behind
this belief was that it should be possible to represent any physical object as a
collection of classical point particles, whereas fields such as electromagnetic waves
act as agents for forces and are governed by laws that are also well understood.

1.1.1 Classical Point Particles

A point particle in classical mechanics is completely specified by its position

and momentum that can be determined simultaneously and precisely. Then, such
information can be used for solving the Newtonian equation of motion for the
particle (or Hamilton’s equations in more general terms), so as to predict or retrodict
its position and momentum at another time. In other words, according to classical
mechanics, the position and momentum of a point particle, once known completely
at a given time, can be determined fully at any past or future time. The rest of issue

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 1

S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_1
2 1 Concepts and Assumptions of Quantum Mechanics

in understanding the physical world then falls down to finding out the way a given
object can be represented as a collection of point particles and determining all the
forces (or potential energies) governing the dynamics of these particles.
For example, consider a point particle with mass m in one dimension. The
position and momentum of this particle at a given time t are .x(t) and .p(t),
respectively. Assuming that there are no other degrees of freedom interacting with
this particle and that the force on this particle depends only on the value of position,
they can be shown to satisfy the following coupled equations of motion:

dx(t) p(t)
. = ,. (1.1)
dt m

dp(t) dV (x) 
= F (x(t)) = − , (1.2)
dt dx x=x(t)

where .V (x) is the potential energy of the particle at position x. Combination of the
above two equations of motion results in the familiar Newtonian equation of motion
involving the acceleration, .a(x(t)) = F (x(t))/m.
Solving Eqs. (1.1) and (1.2) along with known values of position and momentum
.x0 and .p0 at a certain time .t = t0 , one can obtain the position and momentum at any

other time, .x(t; x0 , p0 ) and .p(t; x0 , p0 ). This results in a completely deterministic

trajectory of the particle in the abstract space of position and momentum, which
is called phase space. Along the trajectory of the particle in phase space, the
Newtonian equation of motion ensures that the following Hamiltonian of the particle
remains a conserved quantity.1

p2
H (x, p) =
. + V (x). (1.3)
2m
This can be proved directly using Eqs. (1.1) and (1.2) as follows:
 
d dx(t) ∂H (x, p) dp(t) ∂H (x, p)
. H (x(t), p(t)) = +
dt dt ∂x dt ∂p x=x(t),p=p(t)

p(t) dV (x(t)) dV (x(t)) p(t)

= − = 0. (1.4)
m dx(t) dx(t) m

Thus, the trajectory determined by the Newtonian equation of motion is the one
conserving the Hamiltonian, which in general can be considered as the energy of
the particle expressed as a function of position and momentum.
At this point, it is important to note that the point particle, the limit of an object
with negligible spatial extent, is a mathematical idealization in classical mechanics.

1 For more general and complete account of classical mechanics, refer to advanced textbooks on
classical mechanics, for example, by Goldstein [19].
1.1 Assumptions of Classical Mechanics 3

However, for the description of realistic objects with significant spatial extent, it is
possible to devise a limiting procedure that divides the body into as small volume
elements as possible so that each of the volume element can be treated like a point
particle. Thus, as long as any object can be represented by a collection of finite or
infinite number of interacting point particles with precisely determinable positions
and momenta, it should be possible to completely understand the state and dynamics
of the object on the basis of classical mechanics.
The advance of chemistry and physics in late nineteenth century revealed that all
materials consisted of interacting atoms and molecules that can serve as ultimate
building blocks of point particles. Thus, the reality of the physical world as
portrayed by classical mechanics appeared to be all but confirmed.2 An important
exception to such description was light, or more broadly, the electromagnetic field.
For this, Maxwell has already compiled and refined four distinct equations, now
known as Maxwell’s equations, which provide a complete description of an electro-
magnetic field in space and time and the manner of its interaction with materials.
However, the question of how electromagnetic waves interact with materials at the
atomic and molecular level was never clearly understood. Furthermore, detailed
properties of individual atoms and molecules and those of subatomic particles such
as the electron and the proton had yet to be understood well.

1.1.2 Wave: Classical View

What is a wave? A wave is characterized by a periodic motion of amplitude, in

general, both in time and space, which can be standing or traveling, regardless of
whether we can see them directly or not. One of the simplest wave equation in one
dimension is

1 ∂2 ∂2
 
. − 2 u(x, t) = 0, (1.5)
v 2 ∂t 2 ∂x

which is a partial differential equation with .f (x ± vt) as a general solution. A

well-known class of functions that can represent these solutions are sine and cosine
functions. For example, the following sine function can easily be shown to be the
solution of the above equation:
 x 
u(x, t) = A sin(k(x − vt)) = A sin 2π( − νt) .
. (1.6)
λ

2 Indeed, outcomes of many modern computational simulations have confirmed that much of
material systems we know of can be accurately described in terms of classical mechanics for
molecules represented as a set of interacting point particles as long as all forces are determined
accurately.
4 1 Concepts and Assumptions of Quantum Mechanics

Fig. 1.1 Illustration of an electromagnetic field propagating along the z-direction with .ϕ = 0

In the above expression, A is an amplitude, k is a wavevector, and v is the speed, or

the magnitude of the phase velocity, of the wave, .λ is wavelength, and .ν is frequency.
The above equation implies the following relationships between these quantities:

2π
. k= ,. (1.7)
λ
kv
ν= ,. (1.8)
2π
v = λν. (1.9)

The speed of the electromagnetic wave in vacuum, denoted as c, has the value
of .2.99792458 · · · × 1010 cm/s. If we define the direction of the propagation of the
electromagnetic wave as the z-direction, the electric and magnetic field components
of most commonly used forms3 in vacuum are (see Fig. 1.1)
 z 
E = ex E sin 2π( − νt) + ϕ , .
. (1.10)
λ
 z 
B = ey B sin 2π( − νt) + ϕ , (1.11)
λ
where .ex and .ey are directions along which the electric and magnetic field
components oscillate, .ϕ is an arbitrary phase factor, and the wavelength .λ and
frequency .ν are related to the speed of light c through the following relationship:

.c = νλ. (1.12)

3 These correspond to one particular polarization direction of traverse wave in Coulomb gauge.
1.1 Assumptions of Classical Mechanics 5

For an electromagnetic wave, its mass and position cannot be defined like a
classical point particle. However, a well-defined concept of the energy density for
the wave, which is consistent with experimental results, exists. The energy of a wave
is the energy associated with filling up the space with the wave. The exact expression
for the energy density depends on the specific units of choice, but it can always be
expressed as a linear combination of .E 2 and .B 2 . Thus, the energy of the (classical)
electromagnetic wave increases as the square of their amplitudes. For example, in
Gaussian units, the energy density of an electromagnetic wave is

1  2 
E=
. E + B2 . (1.13)
8π
This expression implies that the energy of an electromagnetic wave depends on
the square of its amplitude and can be made as small as possible. It turns out
that abandoning this definition of energy density has become the starting point of
quantum mechanics.

1.1.3 Particle Versus Wave

The dichotomy of the physical world into those of particles and waves, thus into
full locality and non-locality, appeared to be intact until the late nineteenth century.
However, with the progress of experimental capability, puzzling evidence that
could not be explained in terms of known physical laws and concepts started to
emerge. In retrospect, this is not surprising because the assumptions and concepts
of classical mechanics and electromagnetism relied on observations of macroscopic
phenomena, which then became idealized through mathematical abstraction. In fact,
there was no guarantee that the physical behavior at atomic and molecular scale
should obey the same law as those we observe with our senses directly.
Contrary to what was firmly believed, the key assumption of classical mechanics
that (i) both position and momentum of any object can be determined with arbitrary
precision had never been confirmed with sufficient accuracy for particles with very
small mass. Another important assumption implicit in both classical mechanics and
electromagnetism was that (ii) the magnitude of the energy of a particle or the energy
density of an electromagnetic wave can be made arbitrarily small in a continuous
manner, without restriction. Discoveries of subatomic particles such as the electron
and the proton proved the existence of indivisible units in mass and charge, but it
was still too difficult to test whether the two assumptions (i) and (ii) were valid
for such particles in the early days of those discoveries. On the other hand, solid
experimental findings for light and its interaction with electrons were recognized as
not being consistent with classical theories, which made it necessary to examine the
two assumptions (i) and (ii). It turned out that not only these assumptions but even
the classical concept of particle versus wave needed to be altered in a fundamental
manner, which then led to new concepts and ideas of quantum mechanics.
6 1 Concepts and Assumptions of Quantum Mechanics

1.2 Concepts of Quantum Mechanics

1.2.1 Blackbody Radiation

What is light? How does light interact with matter? Pursuing answers for these
questions has motivated the development of key concepts and principles of quantum
mechanics. The beginning of this was Max Planck’s success in explaining the energy
distribution of light radiated from a blackbody.
A blackbody is a perfect absorber. The light that comes out of it is in full
equilibrium with the body at a given temperature (see Fig. 1.2). The most puzzling
experimental result concerning the blackbody radiation was that the measured distri-
bution of light energy intensity in the high frequency region was drastically different
from theoretical prediction based on Eq. (1.13). Experimentally, the light energy
density reaches a maximum at certain frequency and decays to zero as the frequency
increases. On the other hand, according to Eq. (1.13), such behavior is impossible to
understand because the light wave should be packed more densely, namely with
more modes of wave, as the frequency increases (wavelength decreases). This
results in the energy density increasing as a quadratic function of its frequency, and
thus leads to a hypothetical catastrophic behavior called ultraviolet divergence. On
the other hand, the actual distribution of light energy coming from the blackbody
has vanishing density as its frequency increases.
In an effort to solve the mystery of the blackbody radiation, Max Planck came up
with an idea that the energy of light emitted from a blackbody is always quantized,
which turned out to be one of the greatest discoveries in physics. He showed that the
experimental data on blackbody radiation can be explained by a single assumption
that the energy of light is an integer multiple of its frequency times a universal
constant as follows:

Elight = nhν, n = 1, 2, . . . , (1.14)

Fig. 1.2 Illustration of a

blackbody. Light gets
absorbed into the body and
remitted numerous times
before they make way out
through the small hole
1.2 Concepts of Quantum Mechanics 7

where .h = 6.626 · · · × 10−34 J · s and .ν is the frequency of light. The fundamental

constant h is now known as the Planck constant and can be considered as the unit of
the action4 of light.
Equation (1.14) represented a drastic departure from known concepts and
principles that were believed to provide a complete description of light. It was
firmly believe that the energy of light, as an electromagnetic wave fully describable
by Maxwell’s equations, can have arbitrary magnitude according to Eq. (1.13).
However, this very assumption was simply an outcome of human perception of
waves observable at macroscopic scale. In fact, the energy of light turned out to
be a quantity that can be defined in a new way without affecting the validity of
Maxwell’s equations.
The acceptance of Eq. (1.14) even by Planck himself was a reluctant one. Ini-
tially, this was proposed as a remedy specific for fixing the “ultraviolet divergence”
of the blackbody radiation. It was indeed possible to explain the blackbody radiation
very accurately on the basis of Eq. (1.14), after accounting for some effects of
statistical and geometric factors. As yet, Planck himself at the time of this discovery
was not sure whether Eq. (1.14) was due to the specific nature of interaction between
heated oscillators within the blackbody and the light trapped inside or whether it
indeed represents an intrinsic property of light.

1.2.2 Photoelectric Effect and Photon

The simplest interpretation of Eq. (1.14) is that light is a collection of independent

particles, each with energy .hν. Albert Einstein used this concept of “packets of
light” energy, which were named as photons later, and successfully explained a
photoelectric effect that was difficult to understand with classical theories.
When light hits a metal surface, some electrons can be ejected from the surface.
If one assumes that such phenomenon occurs as a result of the activation of electrons
in metal surface by the driving of classical electromagnetic field, the ejection
of the electron should depended entirely on the intensity of light. However, the
experimental observation was that only the number of ejected electrons depended
on the intensity of light. Careful analyses of experimental results established that
(1) whether electrons are ejected or not and (2) the magnitude of the kinetic energy
of the ejected electron depended only on the frequency of light, not the intensity of
light. These can be summarized by the following relationship:

4 Action is a rather abstract physical quantity originally introduced in a reformulation of Newtonian

Mechanics and is defined as the product of momentum and position or that of energy and time. In
other words, it is a cumulation of the energy of a certain physical “action” over certain duration of
time.
8 1 Concepts and Assumptions of Quantum Mechanics

1
.hν = me v 2 + , (1.15)
2
where .ν is the frequency of the incident light, .me is the mass of electron, v is the
speed of ejected electron, and . is the work function of the metal.
Both the blackbody radiation and the photoelectric effect can be explained in
the simplest possible manner by assuming that light indeed consists of independent
photons, each with energy .hν. Then, what is the implication of this interpretation
with regard to all other existing experimental evidences supporting the wave-nature
of light? Since the two views of light as wave and particle are based on well-
established experimental evidences, one compromising solution is that light can
be viewed as either particle or wave depending on experimental condition and
measurement type. But is this logically possible? It turns out that the answer for
this can be yes as long as each photon retains the full characteristics of wave.
Once the notion of photon as a light particle is accepted,5 the natural question
that arises is what is its physical properties. First, the energy of a single photon is
simply given by

Eph = hν.
. (1.16)

Since light has zero rest mass and travels with a constant speed c in vacuum,
the same property should be shared by each photon. According to the theory of
relativity, the momentum of light with energy E is given by .E/c. Therefore, it is
easy to find that the momentum of a photon is given by

ν h
pph = h
. = , (1.17)
c λ
where .λ is the wavelength of the photon. Thus, there is an inverse relationship
between the momentum and the wavelength of a photon.

1.2.3 de Broglie’s Postulate of Material Wave

Considering that light can be viewed as either wave or particle depending on the
physical situation, it is natural to ask whether the same is true for objects that have
been established as particles. If that is the case, what would be the characteristics
of such a wave? Equation (1.17) shows that there is clear relationship between
momentum and wavelength for the case of a photon. Can this relationship also

5 However, the meaning of photon and the necessity to define it had been debated even after full
development of quantum theory because most interaction with light could be explained in terms of
assumption of classical wave before the advance of laser and modern quantum optics experiments
[20].
1.2 Concepts of Quantum Mechanics 9

be used for any particle? Indeed, de Broglie suggested that it can be used for
any material, and supposed the existence of material wave with the following
wavelength:

h
λdB =
. , (1.18)
p

where p is the magnitude of the momentum of the particle.

de Broglie’s suggestion of Eq. (1.18) was extraordinary in the sense that it
was made without any experimental backing at all. Soon it became a reality
through confirmation by the Davisson-Germer’s experiment [21], according to
which a beam of electrons passing through a nickel foil showed diffraction pattern
corresponding to a wave with the de Broglie wavelength of an electron. Even with
such confirmation, the full physical meaning of de Broglie’s material wave remained
unclear, except that it linked particle with some kind of wave properties. Genuine
understanding of this required development of a comprehensive theory.
As will be detailed later in this book, de Broglie’s material wave also serves as
a good justification for Bohr’s quantization rule for explaining energy levels of an
electron in a hydrogen atom. It can also serve as an essential component for proving
and demonstrating Heisenberg’s uncertainty principle. Most importantly, it can be
viewed as the precursor of Schrödinger’s theory of wave mechanics, now known as
quantum mechanics.

1.2.4 Heisenberg’s Uncertainty Principle

Upon further reflection on the nature of physical observables from a deeper

philosophical viewpoint, namely, from the view that any given physical observable
has its true meaning only to the extent of “definite experiment” by which it can
be determined, Heisenberg reached a conclusion that there has to be an intrinsic
limit in the accuracy of the position and the momentum that can be determined
simultaneously. His initial estimate was that the lower limit for the simultaneous
indeterminacy or uncertainty is .xp ∼ h, where .x and .p are uncertainties of
the values of the position and momentum. This relationship is also consistent with
de Broglie’s theory given that .x ∼ λdB and .p ∼ p.
Heisenberg recognized that the uncertainty is a fundamental limit that cannot
be overcome by any improvement of experimental precision. He viewed this as
the only way for the principle of causality, a fundamental cornerstone in any
scientific reasoning, not to be violated while accepting wave-like nature of particle.
Heisenberg also came up with a similar relationship between the energy and time,
namely, that the minimum of the product of uncertainties of the energy and time is
about .Et ∼ h. However, the physical origin and interpretation of this is different
from that between position and momentum. Following a suggestion by Niels Bohr,
10 1 Concepts and Assumptions of Quantum Mechanics

who strongly believed this as reflecting the dual nature of particle-like and wave-like
properties that quantum objects have, Heisenberg termed these relationships as the
uncertainty principle.
An exact statement of the uncertainty principle, which was clarified later through
a well-defined mathematical analysis, corresponds to the following inequality:

h̄
xp ≥
. , (1.19)
2

where .h̄ = h/(2π ) = 1.0551 · · · × 10−34 J s. In the above expression, .x and .p
are defined as standard deviations from average values of position and momentum
being measured at the same time. In fact, this relationship is not confined to only
between position and momentum. Similar relationships can be found for any pair of
observables that cannot be determined simultaneously.

1.2.5 Wave-Particle Duality

Niels Bohr’s contribution to quantum mechanics goes far beyond the model of a
hydrogen atom.6 He laid out a fundamental conceptual basis for quantum mechanics
along with Max Born by developing a key concept of the duality of particle and
wave nature. He viewed that the concepts of particle and wave, taken separately, are
not sufficient for describing physical phenomena in the quantum limit, but that they
rather have to be combined together for a full description because they in fact play
complementary roles.
The concept of wave-particle duality is easy to misunderstand or misinterpret.
Even what Bohr meant by this remains not fully understood because he never agreed
with others’ interpretation of his view on this concept [22]. It is however easy to say
what it is not. The concept of duality certainly does not imply that a small object
such as electron suddenly becomes particle or wave depending on physical situation
and the type of measurement. This would break the principle of causality, which
ensures that there is objective reality that proceeds by a well-defined cause and effect
relationship, a principle that serves as one of the two most important cornerstones
of physics along with that of energy conservation.
In order for the concept of wave-particle duality to be consistent with the
principle of causality, both properties of particle and wave should remain as intrinsic
possibilities that can show up when an appropriate projection (measurement) is
made while being governed by an objective causal relationship. It is important to
recognize that the concepts of particle and wave are born out of human perception
of the physical world, and are not guaranteed to be the best means to describe

6 We will go through more details of this when we study hydrogen-like systems.

1.3 Schrödinger Equation (in One Dimension) and Probability Amplitude 11

the quantum phenomena. Understanding what the duality really means remains an
important philosophical question that goes beyond the scope of this book.
At the minimum, the duality means that we should abandon the simple concept
of particle and wave that can be visualized easily as in classical theories. Rather, it is
much more convenient to assume that there is a certain quantum state with unknown
characteristics to be specified, which appear to have particle-like properties or
wave-like properties depending on the types of physical measurement being made.
Heisenberg assumed that such a quantum state can be represented by a complex
valued vector in an abstract space. On the other hand, Schrödinger introduced a
wavefunction, which turned out to embody the dual nature of particle and wave in
the most precise way. This will be described in more detail in the following section.

1.3 Schrödinger Equation (in One Dimension)

and Probability Amplitude

Shortly after Heisenberg developed the theory of quantum mechanics in terms

of matrix equations, Schrödinger completed an alternative mathematical theory
by introducing the concept of wavefunction, which represents a general quantum
mechanical state, and identifying a powerful partial differential equation that allows
its full specification.

1.3.1 Wavefunction and Time Dependent Schrödinger

Equation

Given a wavefunction .ψ(x, t), a complex valued function of position and time,
representing a quantum mechanical state of an object or system, the time dependent
Schrödinger equation that determines its time evolution in one dimension is given
by

∂ h̄2 ∂ 2
i h̄
. ψ(x, t) = − ψ(x, t) + V (x)ψ(x, t), (1.20)
∂t 2m ∂x 2

where m is the mass of the object that the wavefunction represents and .V (x) is
the potential energy it is subject to. This potential energy is assumed to be time
independent in the above expression but can be extended to time dependent cases.
The peculiar nature of .ψ(x, t) is that it is in general a complex valued function.
An example is .eikx−iωt = cos(kx − ωt) + i sin(kx − ωt). The physical meaning
of the wavefunction has been under intense debate during the formative period of
quantum mechanics. The standard interpretation that is now well-established is that
.ψ(x, t) gives the probability to find a quantum mechanical particle as prescribed

below.
12 1 Concepts and Assumptions of Quantum Mechanics

Implication and Properties of Wavefunction

For a system represented by a wavefunction with its value .ψ(x, t) at x and at
time t, the probability to find the system is as follows:

|ψ(x, t)|2 dx : Probability to find the system between x and x + dx at time t

.

It is important to note that the above interpretation is consistent with the

concept of the wave-particle duality and the principle of causality. The former
is ensured by the probabilistic nature of the wavefunction and the latter is
guaranteed by the fact that the time dependent Schrödinger equation governs
its time evolution deterministically.
There are two important properties a wavefunction should satisfy.
• A wavefunction for a bounded system is normalized7 as

. |ψ(x, t)|2 dx = 1, (1.21)

R

where R is the region where the wavefunction is defined. The normaliza-

tion condition exists because the total probability has to be equal to one.
• A wavefunction should be continuous in space. This is because the
probability density has to be defined uniquely at any point in space. It is
important to note that the wavefunction becomes zero in the region where
the potential becomes infinite.
7
Bounded system means that the potential energy renders the particle to be localized at a
region of finite volume. For unbounded systems, a different convention for normalization is
used.

1.3.2 Measurement as Mathematical Operation

on Wavefunction

An important assumption in Schrödinger’s formulation is that measuring a physical

property of a quantum mechanical state can be represented by a “mathematical”
operation on the wavefunction. For example, the operation corresponding to a
position measurement is represented by the following simple multiplication:

x̂ψ(x, t) = xψ(x, t).

. (1.22)

In the same way, measuring any function of position .A(x) simply amounts to
multiplication with this function. In other words, the operation of .Â(x̂) is defined as
1.3 Schrödinger Equation (in One Dimension) and Probability Amplitude 13

Â(x̂)ψ(x, t) = A(x)ψ(x, t).

. (1.23)

Note that .xψ(x, t) and .A(x)ψ(x, t) are not constant multiples of .ψ(x, t) in general.
Thus, measurement of this position dependent physical observable changes the
quantum state in general. The exception for this is when the wavefunction is fully
localized at one specific point, for which a special definition of function is needed
as will become clear later in this chapter.
On the other hand, the measurement of momentum turns out to be a differentia-
tion given by

h̄ ∂
p̂ψ(x, t) =
. ψ(x, t). (1.24)
i ∂x

For the case where .ψ(x, t) ∼ eikx−iωt , which represents a complex valued wave
with definite wavelength .λ = 2π/k, the above definition results in the following
value for the momentum:
h 2π h
. p = h̄k = = . (1.25)
2π λ λ
Thus, Eq. (1.24) is consistent with de Broglie’s postulate of a material wave.
Classically, the kinetic energy is given by .p2 /(2m). This definition can be used
to define quantum mechanical kinetic energy operator as

p̂2 h̄2 ∂ 2
K̂ =
. =− (1.26)
2m 2m ∂x 2

The combination of Eq. (1.23), with .V (x) for .A(x), and Eq. (1.26) leads to
a compact expression for the time dependent Schrödinger equation (1.20) as
summarized below.

Time-Dependent Schrödinger Equation and Hamiltonian

∂
i h̄
. ψ(x, t) = Ĥ ψ(x, t), (1.27)
∂t

where .Ĥ is the Hamiltonian operator defined as

p̂2 h̄2 ∂ 2
Ĥ =
. + V (x̂) = − + V (x). (1.28)
2m 2m ∂x 2
(continued)
14 1 Concepts and Assumptions of Quantum Mechanics

The implication of this is that the Hamiltonian operator generates the time
evolution of the wavefunction. Thus, information on the Hamiltonian and the
wavefunction at a given time .t = t0 allows determination of the quantum state
at any time.

1.3.3 Stationary States and Time Independent Schrödinger

Equation

Let us define a stationary quantum mechanical state .ψs (x, t) as the one satisfying
the following relationship:

∂
i h̄
. ψs (x, t) = Eψs (x, t). (1.29)
∂t
Let us also assume that the dependences of the wavefunction on position and time
can be decoupled from each other as follows: .ψs (x, t) = ψ(x)f (t). Inserting this
relation into the above equation, we obtain

df (t)
i h̄
. ψ(x) = Ef (t)ψ(x). (1.30)
dt

For nonzero .ψ(x), the above relation implies that

df (t) iE
. = − f (t), (1.31)
dt h̄

the solution of which is .f (t) = f (0)e−iEt/h̄ . Let us now assume that the time
independent part of the wavefunction is equal to the full time dependent wave
function at .t = 0. That is, .ψ(x) = ψs (x, 0). Then, .f (0) = 1 and

f (t) = e−iEt/h̄ .
. (1.32)

Inserting .ψs (x, t) = f (t)ψ(x), with the above expression for .f (t), into Eq. (1.29),
we obtain

Ĥ ψ(x) = Eψ(x).
. (1.33)

Thus, for a special kind of wavefunction .ψ(x) that satisfies the above equation,
the corresponding time dependent wavefunction that satisfies Schrödinger’s time
dependent equation, Eq. (1.27), is obtained by simply multiplying it with .e−iEt/h̄ .
1.3 Schrödinger Equation (in One Dimension) and Probability Amplitude 15

Time Independent Schrödinger Equation and Eigenvalue of Hamiltonian

Equation (1.33) is known as time independent Schrödinger equation and
can be used for complete specification of all the stationary states and
corresponding energies for a given Hamiltonian. Not all wavefunctions and
energy E satisfy this equation. In fact, this is an eigenvalue problem, for
which both E and .ψ(x) should be determined. Let us denote the wavefunction
satisfying Eq. (1.33) for a specific value of E as .ψE (x). Namely,

Ĥ ψE (x) = EψE (x).

. (1.34)

Mathematically, .ψE (x) is called an eigenfunction of the Hamiltonian operator

with eigenvalue E. Physically, what this means is that, if the energy is
measured for a system represented by .ψE (x), the state is not disturbed and the
result of the energy measurement is the value of E. For a given Hamiltonian,
there is a well-defined set of finite or infinite eigenfunctions and eigenvalues,
the determination of which amounts to fully solving the time independent
Schrödinger equation.

1.3.4 Eigenfunction and Eigenvalue

Eigenfunction and eigenvalue are general mathematical terms that are not necessar-
ily limited to quantum mechanics. For any mathematical operator .Ô satisfying the
following condition:

Ôψλ (x) = λψλ (x),

. (1.35)

one can call .ψλ (x) as an eigenfunction of .Ô with eigenvalue .λ. Solving this
eigenvalue problem is often the major step for finding a general solution involving
the operator .Ô.
Even within quantum mechanics, eigenfunction and eigenvalue are not limited to
the Hamiltonian operator but can also be defined for other physical observables. For
example, consider .φk (x) = eikx , where k is a real number. Then,

h̄ ∂ ikx
p̂φk (x) =
. e = h̄keikx = h̄kφk (x). (1.36)
i ∂x

The above relation shows that .φk (x) is an eigenfunction of the momentum operator
with eigenvalue .h̄k. In other words, .φk (x) represents a quantum mechanical state
with a definite momentum .h̄k.
16 1 Concepts and Assumptions of Quantum Mechanics

For a quantum particle moving freely in space without any potential energy, the
Hamiltonian operator is

p̂2 h̄2 ∂ 2
.Ĥfp = =− . (1.37)
2m 2m ∂x 2
Then,

h̄2 ∂ 2 ikx h̄2 k 2 ikx

Ĥfp φk (x) = −
. e = e . (1.38)
2m ∂x 2 2m

This shows that .φk (x) is also an eigenfunction of .Ĥfp with eigenvalue .h̄2 k 2 /(2m).
On the other hand, if there is nonzero potential .V (x), .φk (x) no longer becomes
the eigenfunction of the Hamiltonian as shown below.

h̄2 k 2
.Ĥ φk (x) = + V (x) φk (x) = λφk (x), for any λ (1.39)
2m

As the above equation indicates, .φk (x) is not an eigenfunction of .Ĥ for nonzero
potential because .V (x)φk (x) cannot be a constant multiple of .φk (x).

1.3.5 Linear and Hermitian Operator

For any physically measurable quantity, there should be a well-defined operator for
which eigenvalues and eigenfunctions can be identified. However, it is important to
note that not all kinds of mathematical operators can be used to represent a physical
observable.
First, it is assumed that any operator corresponding to physical observable is
linear. The definition of a linear operator is as follows.

Definition of Linear Operator

An operator acting on a set of complex functions, .F, is called linear, if for any
functions .f1 (x) and .f2 (x) belonging to .F and complex coefficients .c1 and .c2 ,
the following identity always holds:

Ô(c1 f1 (x) + c2 f2 (x)) = c1 Ôf1 (x) + c2 Ôf2 (x).

. (1.40)

The above definition implicitly assumes that any linear combination (with
complex coefficients) of functions belonging to .F still remains in .F. It is easy
to show that the position, momentum, and Hamiltonian operators are all linear
operators.
1.3 Schrödinger Equation (in One Dimension) and Probability Amplitude 17

Second, the eigenvalues of operators corresponding to physical observables have

to be real-valued because no physical measurement can produce complex numbers.
A class of operators called Hermitian operators guarantee this property. Therefore,
all operators corresponding to physical observables are assumed to be Hermitian.
The definition of a Hermitian operator is as follow.

Definition of Hermitian Operator

An operator . that acts on a complex valued function of one dimensional
variable x is called Hermitian if it satisfies the following identity for any
complex valued functions .φ(x) and .ψ(x):
 ∗
. dx φ ∗ (x)Âψ(x) = dx ψ ∗ (x)Âφ(x) , (1.41)
R R

where R represents the domain of x where the functions are defined (including
its boundary condition) and .(· · · )∗ denote complex conjugate. It is assumed
that the operator . acts on the function on its righthand side.8 The above
definition can be extended to functions of multi-variables by simply extending
the integration to corresponding multi-dimensional integration.
8
More generally, it is possible to define an operator acting on the function on the lefthand
side, for which the definition of a Hermitian operator remains the same.

It is easy to show that the position operator and any function of position operator
are Hermitian. The proof that momentum operator is a Hermitian operator can also
be proved employing integration by parts given that the boundary terms do not
contribute. The fact that the eigenvalue of a Hermitian operator is a real number
can be shown easily as follows.
Theorem 1.1 If .φλ (x) is an eigenfunction of a Hermitian operator .Â, its eigenvalue
λ is always a real number.
.

Proof By employing .ψ(x) = φ(x) = φλ (x) on the lefthand side of Eq. (1.41) and
using the fact that .Âφλ (x) = λφλ (x), we obtain

. dx φλ∗ (x)Âφλ (x) = λ dx φλ∗ (x)φλ (x) = λ dx |φλ (x)|2 (1.42)

R R R

On the other hand, using the same definitions and identity on the righthand side of
Eq. (1.41), we also obtain
 ∗  ∗
. dx φλ∗ (x)Âφλ (x) = λ dx φλ∗ (x)φλ (x) = λ∗ dx |φλ (x)|2 ,
R R R
(1.43)
18 1 Concepts and Assumptions of Quantum Mechanics

where the fact that . R dx φλ∗ (x)φλ (x) is a real number has been used. Since the fact
that . is Hermitian implies that Eqs. (1.42) and (1.43) are the same and because
∗ ∗
.
R dx φλ (x)φλ (x) is nonzero, we find that .λ = λ . Thus, .λ is a real number. ⊔
⊓

1.3.6 Results of Measurement and Expectation Value

When a quantum system is in an eigenstate of a physical observable .Â, the result

of the measurement is always the corresponding eigenvalue. What if the state is not
any of the eigenstates? Then, the result of the measurement is not certain. However,
even in this case, a possible value of individual measurement result is always one of
the known eigenvalues of .Â.
To be more specific, let us assume that the state of the system at a certain time
is .ψ(x) and the eigenstates of . are .φj (x)’s with eigenvalues .Aj ’s, where j is an
index labeling the eigenstates. Then, the statement given above can be represented
by the following probability for the result of the measurement made for the state
.ψ(x) to be the eigenvalue .Aj :

 2
φj∗ (x)ψ(x) (1.44)
 
.pj (ψ) =  dx .

R

In case .ψ(x) is .φj (x), this probability is equal to one because .φj (x) is normalized.
Thus, the measurement result is always .Aj . In case .ψ(x) is .φk (x) for different
.Ak = Aj , it is logically clear that this probability should be zero. This is possible

only if

. dx φj∗ (x)φk (x) = 0, for Aj = Ak . (1.45)

R

This can be proved using the fact that . is Hermitian as follows.
Theorem 1.2 Two eigenfunctions of a Hermitian operator with different eigenval-
ues are always orthogonal to each other.
Proof Equation (1.41) for the case of .φj (x) and .φk (x) becomes
 ∗
. dx φj∗ (x)Âφk (x) = dx φk∗ (x)Âφj (x) . (1.46)
R R

Since .Âφk (x) = Ak φk (x) and .Âφj (x) = Aj φj (x), the above relation implies that
 ∗
.Ak dx φj∗ (x)φk (x) = Aj dx φk∗ (x)φj (x) . (1.47)
R R
1.3 Schrödinger Equation (in One Dimension) and Probability Amplitude 19

Since .Aj is a real number and .(φk∗ (x)φj (x))∗ = φk (x)φj∗ (x), the above identity
implies that

(Ak − Aj )
. dx φj∗ (x)φk (x) = 0. (1.48)
R

Since .Ak − Aj = 0, the above identity is equivalent to Eq. (1.45). ⊔

⊓
Let us consider more general case where the wavefunction is given by a linear
combination of different .φk (x)’s for which all the eigenvalues .Ak ’s are different as
follows:

ψ(x) =
. Ck φk (x), (1.49)
k

where.Ck ’s are complex coefficients that can be arbitrary except for the constraint
that . k |Ck |2 = 1. The reason for this constraint will soon become clear.
Employing Eq. (1.44), we find that
 2  2
   
.pj (ψ) =  Ck ∗
dxφj (x)φk (x) =  Ck δj k  = |Cj |2 , (1.50)
   

kR   
k

where Theorem 1.2 has been used and .δj k is the Kronecker-delta symbol.9 This
means that the absolute square of each coefficient in Eq. (1.49) corresponds to the
probability. Therefore, . k |Ck |2 = 1 because the sum of all probabilities have to
become one.
For the case where some of the eigenvalues .Aj ’s are the same, the orthogonality
given by Theorem 1.2 is not always satisfied. However, even in this case, it is
always possible to identify orthogonal .φk (x)’s with respect to which a unique
linear combination given by Eq. (1.49) can be identified for any wavefunction .ψ(x).
Then, each .|Ck |2 can be interpreted as the probability for the state to exist in the
state represented by .φk (x). Thus, for such a general state, although the results of
measurement are not always certain, one can expect that the average over many
repeated measurements, so called expectation value can always be calculated as
described below.

Expectation Value
For a general normalized quantum state given by Eq. (1.49) the expectation
value for the measurement of an operator . is given by
 2
dx φk∗ (x)ψ(x) Ak . (1.51)
 
Â
. ψ = pk (ψ)Ak = 

k k R

(continued)

9 The Kronecker-delta symbol .δj k = 1 for .j = k and is zero otherwise.

20 1 Concepts and Assumptions of Quantum Mechanics

Alternatively, the expectation value can be expressed as

. = dx ψ ∗ (x)Âψ(x). (1.52)

R

The equivalence of Eqs. (1.51) and (1.52) can be proved by using Eq. (1.49)
and employing the orthogonality Theorem 1.2.

1.4 Quantum Particle in a One-Dimensional Box

Let us consider a particle with mass m contained in a one-dimensional box of length

l. That is, the potential .V (x) = 0 for .0 < x < l, and .V (x) = ∞ for .x ≤ 0 or
.x ≥ l. Figure 1.3 illustrates the system and an example of a classical trajectory that

bounces back and forth between two infinite walls at constant speed. The quantum
mechanical behavior of this particle is drastically different, as will be shown in this
section. The Hamiltonian operator within the box and the wavefunction outside the
box are given by

p̂2
. Ĥ = for 0 < x < l,
2m
ψ(x, t) = 0 for x ≤ 0 or x ≥ l.

Fig. 1.3 One dimensional

box, where the potential
energy is zero for .0 < x < L
and infinite outside of this
region. The ordinate
represents energy. The blue
circle represents a classical
particle, and the arrows
represent the direction of
their movement at different
positions within the box
1.4 Quantum Particle in a One-Dimensional Box 21

1.4.1 Time Independent Schrödinger Equation

Let us denote the eigenfunction with energy E as .ψE (x). Then, for .0 < x < l,

h̄2 ∂ 2
Ĥ ψE (x) = −
. ψE (x) = EψE (x), (1.53)
2m ∂x 2

with the following boundary conditions: .ψE (0) = ψE (l) = 0. A general solution
for Eq. (1.53) is given by

ψE (x) = C1 eikx + C2 e−ikx .

. (1.54)

By plugging this into Eq. (1.53), one can show that

h̄2 k 2
E=
. . (1.55)
2m
However, this does not complete the solution yet because the boundary conditions
need to be satisfied as well. Imposing the boundary conditions limits the possible
values of k (therefore E). In other words, the wavefunction should become zero
when .x = 0 and .x = l. Thus,

ψE (0) = C1 + C2 = 0,
. (1.56)

from which we get .C1 = −C2 . Inserting this into Eq. (1.54) and imposing that
ψ(l) = 0, we get
.

ψE (l) = C1 (eikl − e−ikl ) = 2iC1 sin(kl) = 0.

. (1.57)

This condition is satisfied for nonzero .C1 only if .k = nπ/ l with .n = 1, 2, . . . .

Inserting this into Eq. (1.55), we find that the energy is quantized as follows:

π 2 h̄2 n2
En =
. , n = 1, 2, . . . (1.58)
2ml 2
Up to now, the constant .C1 has not been determined yet. Let us introduce a new
constant .C = 2iC1 and label each eigenfunction with the integer quantum number
n. Then,
 nπ x 
.ψn (x) = C sin . (1.59)
l
22 1 Concepts and Assumptions of Quantum Mechanics

The normalization condition can be used to determine the constant C as follows:

l l  nπ x 
. 1= dx|ψn (x)|2 = |C|2 dx sin2
0 0 l
|C|2 l 2nπ x 2
  
|C| l
= dx 1 − cos = . (1.60)
2 0 l 2
√
Assuming that C is real and positive,10 the above equation shows that .C = 2/ l.
Therefore, the nth eigenfunction is given by

2  nπ x 
ψn (x) =
. sin . (1.61)
l l

One can directly confirm that .ψn (x) given above satisfies the time independent
Schrödinger equation, Eq. (1.53), for the energy given by Eq. (1.58).
It is clear that .ψn (x) given by Eq. (1.61) is not an eigenfunction of the position
or momentum operator. On the other hand, it is an eigenfunction of .p̂2 with an
eigenvalue .h̄2 π 2 n2 / l 2 . This is because the wavefunction is a combination of equal
weight of those with positive and negative values of the same magnitude of the
momentum, .h̄π n/ l. This value of momentum indeed satisfies the de Broglie’s
relationship given that the wavelength .λn = 2l/n of .ψn (x) is used. Thus, the de
Broglie wave is nothing but the Schrödinger’s wavefunction.
The expectation value of position for .ψn (x) can be calculated as

2 l  nπ x   nπ x 
x̂ =
. dx sin x sin
l 0 l l
2 l 1 − cos(2nπ x/ l)
 
l
= dx x= .
l 0 2 2

Thus, the average position is the center of the box and is independent of the quantum
number n. On the other hand, the expectation value of momentum is always zero as
follows:

2 h̄ nπ x  ∂ l
 nπ x 
. p̂ = sin dx sin
0 l i l ∂x l
2 h̄ nπ l  nπ x   nπ x 
= dx sin cos = 0. (1.62)
l i l 0 l l

10 C in fact can be a complex value with unit norm, which does not affect the outcome of any
physical observable. Thus, this assumption is more of convenience rather than necessity.
1.4 Quantum Particle in a One-Dimensional Box 23

In order to check the uncertainty relationship, let us also calculate the expectation
values of .x̂ 2 and .p̂2 . First, the expectation value of .x̂ 2 is given by

2 l  nπ x   nπ x 
.x̂ 2 = dx sin x 2 sin
l 0 l l
2 l 1 − cos(2nπ x/ l)
 
= dx x2
l 0 2
l2 1 2nπ x l
   
l
= + x − cos
3 nπ 2nπ l 
0

l2 l2
= − 2 2. (1.63)
3 2n π

On the other hand, since .ψn (x) is an eigenfunction of .p̂2 , the expectation value is
equal to the eigenvalue as follows:

h̄2 n2 π 2
p2 = 2mEn =
. . (1.64)
l2
Therefore, the uncertainties of position and momentum are as follows:

1/2  l 2 1/2 1/2

l2
 2 2

2 2 l n π
. x = x̂ − x̂ = − = −2 ,
12 2n2 π 2 2nπ 3
. (1.65)
 1/2 h̄nπ
p = p̂2 − p̂ 2 = . (1.66)
l
The product of the above two uncertainties satisfies Heisenberg’s uncertainty
principle as follows:
1/2 1/2
n2 π 2 h̄ π 2
 
h̄ h̄
xp =
. −2 ≥ −2 > . (1.67)
2 3 2 3 2

1.4.2 Time Dependent States

For each eigenfunction, one can define the following time dependent stationary
wavefunction.

−iEn t/h̄ 2  nπ x 
.ψn (x, t) = ψn (x)e = sin e−iEn t/h̄ , (1.68)
l l
24 1 Concepts and Assumptions of Quantum Mechanics

where .En is given by Eq. (1.58). This satisfies the time dependent Schrödinger
equation as shown below.

∂
i h̄
. ψn (x, t) = En ψn (x, t) = Ĥ ψn (x, t). (1.69)
∂t
Now let us consider the following linear combination:
∞ ∞ 
2  nπ x 
ψ(x, t) =
. Cn ψn (x, t) = Cn sin e−iEn t/h̄ . (1.70)
l l
n=1 n=1

Then, one can show that

∞
∂ ∂
. i h̄ ψ(x, t) = Cn i h̄ ψn (x, t)
∂t ∂t
n=1
∞ ∞
= Cn En ψn (x, t) = Cn Ĥ ψn (x, t) = Ĥ ψ(x, t), (1.71)
n=1 n=1

where the last equality holds because .Ĥ is a linear operator.

Equation (1.71) proves that any linear combination of the eigenstates with
appropriate time dependent factor .e−iEn t/h̄ satisfies the time dependent Schrödinger
equation. Although we have specifically assumed that .ψn (x) is the solution of the
time independent Schrödinger equation for a quantum particle in a box, Eq. (1.71)
is generally true for any Hamiltonian.

1.4.3 Completeness

The eigenfunctions of the Hamiltonian form a complete orthonormal basis set for all
quadratically integrable functions defined for .0 < x < l. The meaning of this will
be explained in more detail below. First, orthonormality11 can be shown explicitly
as
∞ 2 l  π nx   π mx 
. ψn∗ (x)ψm (x)dx = sin sin dx
0 l 0 l l
1 l     
π(n − m)x π(n + m)x
= cos − cos = δnm , (1.72)
l 0 l l

11 This means being orthogonal and normalized.

1.5 Summary and Questions 25

where .δnm is the Kronecker-delta symbol introduced in Eq. (1.50). The set of
eigenfunctions form a complete basis in the sense that any well-behaved function
can be expressed as
∞
f (x) =
. Cm ψm (x), (1.73)
m=1

where .Cn is uniquely determined by the following relation:

l ∞ l ∞
. ψn∗ (x)f (x)dx = Cm ψn∗ (x)ψm (x)dx = Cm δnm = Cn . (1.74)
0 m=1 0 m=1

Inserting the above relation into Eq. (1.73),

∞ l l ∞
∗ ′
f (x) =
. ψm (x) ψm (x )f (x ′ )dx ′ = ∗ ′
ψm (x)ψm (x )f (x ′ )dx ′ .
m=1 0 0 m=1
(1.75)

Because the above relation holds for any function .f (x), we come up with the
following identity:
∞
∗ ′
. ψm (x)ψm (x ) = δ(x − x ′ ), (1.76)
m=1

where .δ(x − x ′ ) is called Dirac-delta function, a singular “function” that is in fact

something called a “distribution” mathematically. More detailed description of this
is provided in Appendix.
Equation (1.76) is an example of the completeness relationship. Although this
was demonstrated here for the particular example of the particle in a box, it is in
fact possible to identify a set of eigenfunctions satisfying such an identity for any
Hamiltonian. Such eigenfunctions form a complete basis set.

1.5 Summary and Questions

The theory of quantum mechanics is based on two principles well established

through classical mechanics. These are fundamental principles that (i) the energy of
a closed isolated system remains conserved (the principle of energy conservation),
and that (ii) any physical processes are manifestations of clear relationships between
certain causes and effects that are both well defined and quantifiable (the principle
of causality).
26 1 Concepts and Assumptions of Quantum Mechanics

However, the theory of quantum mechanics is a complete departure from that of

classical mechanics in that simultaneous determination of position and momentum
of any physical object cannot be made with arbitrary precision. Rather, the precision
of one or the other has to be given up according to the uncertainty principle, as
shown by Eq. (1.19) for the case of one dimensional position and momentum.
Within the formalism of quantum mechanics, the classical notion that position
and momentum can be determine completely and simultaneously is an approxima-
tion that appears to be valid only when the action, another well-established concept
in classical mechanics implying “the outcome of applying force for certain time,” is
much larger than .h̄, the reduced Planck constant. This posed serious problem both
conceptually and practically because it meant that classical Newton’s equation or
classical Hamiltonian dynamics (a more general form of classical dynamics) can no
longer be used as rules to describe the dynamics or to understand the relationship of
causality.
The notion of the “dual nature of wave and particle,” as first expressed by
de Broglie for simple cases, Eq. (1.18), and articulated by Bohr, provided new
conceptual basis for understanding the uncertainty principle and motivated further
experimental discoveries. Then, the formulations by Schrödinger and Heisenberg
made it possible to describe states and dynamics of quantum particles while not
violating the two fundamental principles of energy conservation and causality, as
stated by (i) and (ii) above.
The success of Schrödinger equation, Eq. (1.27), and its time independent version
for stationary states, Eq. (1.34), led to the establishment of the theory of quantum
mechanics as an ultimate theory for all atomic and molecular properties and
phenomena. Although there have been and are still fundamental debates about the
physical meaning of the Schrödinger equation, the fact that actual states of a system
and physically measured quantities have to be considered separately is now well
established.
While the Schrödinger equation appears to provide a full description of a
quantum mechanical state, what is being described by the equation, namely a
wavefunction, cannot in fact be measured experimentally. Rather, a wavefunction
can provide only the probability of certain outcomes of the measurements in
general. The exception to this is when the state is in an eigenstate of the operator
corresponding to measurement, which often involves the measurement process itself
as ways to prepare such states. In this case, the outcome of the measurement is
certain and it is assumed that the complete information on the state of the system
can be deduced once we are able to determine the eigenstates of all other compatible
physical observables.
Another difficulty of quantum mechanics as a theory lies in the fact that the
solution of the Schrödinger equation is difficult to get exactly except for very
simple model systems. As yet, detailed study of such few examples, for which exact
solutions are available, helps understand clearly major concepts and phenomena
that are unique for quantum mechanics. These include superposition states as
linear combinations of other states and important rules concerning definitions and
interpretations of measurements. A short overview of the problem of quantum
particle in a one dimensional box, described in Sect. 1.4, serves as a good example.
Appendix: Dirac-Delta Function 27

Questions

• Can blackbody radiation and the photoelectric effect be explained only if

the energy of light is proportional to its frequency?
• Is it possible for both the position of a quantum particle and momentum
along the same direction to be determined with arbitrary precision?
• What happens to the de Broglie wave if a particle completely stops?
• Given that the dynamics of a closed classical system is energy conserving,
is the dynamics of the corresponding closed quantum system also energy
conserving?
• For any quantum system, does full information on its wavefunction allow
calculation of the outcome of any physical observable?
• Is it possible for the eigenvalue of a Hermitian operator to become a
complex number?
• What happens if a quantum mechanical operator is not linear?
• What is the relationship between time dependent and time independent
Schrödinger equations?

Appendix: Dirac-Delta Function

Dirac-delta function [23] is defined by the following two properties:

δ(x − x ′ ) = 0 for x = x ′ , .
. (1.77)
∞
dxf (x)δ(x − x ′ ) = f (x ′ ), (1.78)
−∞

where .f (x) is an arbitrary continuous and integrable function. In fact, the Dirac-
delta function is not a genuine function, but a limit of a series of functions, which
is known as distribution in mathematics. Thus, more rigorously, the Dirac-delta
function is defined as follows:

δ(x) = lim δ (ǫ) (x),

. (1.79)
ǫ→0

where .δ (ǫ) (x) is a well-defined function for finite .ǫ but becomes singular in the limit
of .ǫ = 0. There is no unique way to define the sequence of .δ (ǫ) (x). Well-known
forms of .δ (ǫ) (x) are as follows.

1/ǫ, −ǫ/2 < x < ǫ/2


(ǫ)
.δh (x) = ,. (1.80)
0, |x| > ǫ/2
28 1 Concepts and Assumptions of Quantum Mechanics

1 −|x|/ǫ
δe(ǫ) (x) = e ,. (1.81)
2ǫ
(ǫ) 1 ǫ
δl (x) = ,. (1.82)
π x2 + ǫ2
1 2 2
δg(ǫ) (x) = √ e−x /ǫ , . (1.83)
ǫ π
1 sin(x/ǫ) 1 1/ǫ
δs(ǫ) (x) = = eikx dk. (1.84)
π x 2π −1/ǫ

In the above equations, different subscripts denote different ways to define the
sequence such as “hard step function,” “exponential function,” “Lorentzian func-
tion,” “Gaussian function,” and “sine function.” Note that the second equality in
Eq. (1.84) can be shown by direct integration of the last expression. One of the
most frequently used form of Dirac-delta function in quantum mechanics employs
Eq. (1.84). When this is inserted into Eq. (1.79), the resulting Dirac-delta function
becomes

1 ∞
δ(x) =
. eikx dk. (1.85)
2π −∞

Exercise Problems with Solutions

1.1 How many photons per second are generated in a 100.0 watt beam of red light
with 700.0 nm wavelength?
Solution 1.1 The 100.0 watt beam of light corresponds to Elight = 100.0 J per
second. According to Eqs. (1.14) and (1.12), this corresponds to the following
number of photons per second:

100.0 7.000 × 10−7

n=
. = 3.523 × 1020 .
6.626 × 10−34 2.998 × 108

1.2 Light is incident on a metal with work function 1 eV (1.602 × 10−19 J), which
causes ejection of electrons with 1.00 nm de Broglie wavelength. Assuming that the
magnitude of the momentum of electron is given by me ve , where me and ve are the
mass and speed of electron, determine the frequency of the incident light?
Solution 1.2 According to Eq. (1.18) and the assumption that pe = me v, the speed
of the electron is given by

h 6.626 × 10−34 kgm2 s−1

ve =
. = .
λdB me (1.00 × 10−9 m) × (9.109 × 10−31 kg)
Exercise Problems with Solutions 29

Using the above value in Eq. (1.15), we find that the frequency of incident light is
given by

 me 2
ν=
. + v = 6.055 × 1014 s−1 .
h 2h e

1.3 A particle with mass 1 g is placed on a sharp tip of 1.00 mm width. The
particle falls off the tip if it goes outside of its width. According to the Heisenberg’s
uncertainty principle, what is the maximum time the particle can stay on top of the
tip?
Solution 1.3 For the particle to stay on top of the tip, the uncertainty of position has
to be equal to or less than 1 mm. According to the Heisenberg’s uncertainty relation,
Eq. (1.19), this implies that

h̄ 6.626 × 10−34 kgm2 s−1

.p ≥ = = 5.28 × 10−32 kg · m · s−1 .
x 4 × 3.14 × 1.00 × 10−3 m

If the particle stays on top of the tip for t, this means that the magnitude of the
momentum is approximately given by mx/t. Since p > p, this means that

mx mx
t ≈
. ≤
p p
(1.00 × 10−3 kg) × (1.00 × 10−3 m)
≤ = 1.89 × 1025 s.
5.26 × 10−32 kg · m · s −1

1.4 A proton is confined in a one dimensional box of 0.2 nm length. Use the
uncertainty principle to determine its minimum kinetic energy.
Solution 1.4 The minimum uncertainty in the momentum of the proton with the
uncertainty in position of 2 Å, according to Eq. (1.19), is given by

h̄
p =
. = 2.64 × 10−25 kg · ms−1 .
2x
For the above minimum uncertainty in momentum, the minimum kinetic energy is
given by

1 p2
Emin =
. = 2.08 × 10−23 J.
2 m

1.5 The wavefunction of a particle in one dimension is given by

2 /2
ψ(x) = N e−x
. .

Determine the normalization constant N. Prove that Heisenberg’s uncertainty

principle is satisfied for this wavefunction.
30 1 Concepts and Assumptions of Quantum Mechanics

Solution 1.5 For the wavefunction to be normalized,

∞ ∞ 2 √
1=
. dx|ψ(x)|2 = N 2 dxe−x = N 2 π .
−∞ −∞

Therefore, N = 1/π 1/4 . With this normalization constant, it is straightforward

to calculate the expectation values of position, position square, momentum, and
momentum square as follows:

1 ∞ 2
.x̂ = √ dxxe−x = 0,
π −∞
1 ∞ 2
x̂ 2 = √ dxx 2 e−x
π −∞

1 1 −x 2 ∞ 1 ∞
  
−x 2
= √ − xe  + dxe
π 2 −∞ 2 −∞

1 1√ 1
= √ π= ,
π2 2

1 h̄ ∞ 2 /2 d −x 2 /2 1 h̄ ∞ 2 /2 2 /2
p̂ = √
. dxe−x e =√ dxe−x xe−x = 0,
π i −∞ dx π i −∞

1 ∞ 2 /2 d 2 −x 2 /2
p̂2 = − √ h̄2 dxe−x e
π −∞ dx 2
h̄2 ∞  2 2

= −√ dx −e−x + x 2 e−x
π −∞

h̄2√ 1√ h̄2
 
= −√ − π+ π = .
π 2 2

Employing the above results, we can calculate the uncertainties of position and
momentum as follows:
1

.x = x̂ 2 − x̂ 2 = √ ,
2
h̄

p = p̂2 − p̂ 2 = √ .
2
The above results show that xp = h̄/2. Thus, the given wavefunction is a
minimum uncertainty state.
1.6 An operator Ô is defined by the following action on a given function f (x).
Ôf (x) = f (x)2 .
.

Prove that this operator is not linear.

Exercise Problems with Solutions 31

Solution 1.6 Given that f (x) = C1 f1 (x) + C2 f2 (x),

Ô(C1 f1 (x) + C2 f2 (x)) = (C1 f1 (x) + C2 f2 (x))2

.

= C12 f1 (x)2 + 2C1 C2 f1 (x)f2 (x) + C2 f2 (x)2 .

This is not in general equal to C1 Ôf1 (x) + C2 Ôf2 (x) = C1 f1 (x)2 + C2 f2 (x)2 .
Therefore, the given operator is not linear.
1.7 An operator Ô is defined by the following action on a given function f (x).

d
Ôf (x) =
. xf (x).
dx
Prove that this operator is linear.
Solution 1.7 Given that f (x) = C1 f1 (x) + C2 f2 (x),

d
Ô(C1 f1 (x) + C2 f2 (x)) =
. x (C1 f1 (x) + C2 f2 (x))
dx
d d
= C1 xf1 (x) + C2 xf2 (x)
dx dx
= C1 Ôf1 (x) + C2 Ôf2 (x).

Therefore, the given operator is linear.

1.8 Use the particle in a box problem, where the wavefunction is zero outside the
region of 0 < x < l, to prove that momentum operator is Hermitian. Hint: You need
to use integration by part.
Solution 1.8 For arbitrary wavefunctions ψ(x) and φ(x) for the particle in a box
of length l, integration by part results in the following identity:
l
l l ∂φ ∗ (x)
 
h̄ ∂ h̄ h̄
dxφ ∗ (x) ψ(x) = φ ∗ (x)ψ(x) −

. dx ψ(x).
0 i ∂x i 0 i 0 ∂x

Since ψ(x) and φ(x) are zero for x = 0, l, the first on the righthand side of the
above equation vanishes. Therefore,

l  l ∗
∗ h̄ ∂ ∗ h̄ ∂
. dxφ (x) ψ(x) = dxψ (x) φ(x) .
0 i ∂x 0 i ∂x

This proves that the momentum operator is Hermitian.

1.9 Consider a particle in a box of unit length (l = 1) with the following
wavefunction:

ψ(x) = sin(π x) + sin(2π x).

.
32 1 Concepts and Assumptions of Quantum Mechanics

If energy is measured for this particle, what are possible values of the measurement
and their respective probabilities? What is the probability for this particle to be
found in the region of 0 < x < 1/4?
Solution 1.9 For a particle in a box of unit length, the normalized eigenfunctions
are
√
φn (x) =
. 2 sin(nπ x), n = 1, 2, 3, . . . (1.86)

for which the eigenvalue of the Hamiltonian, En = h̄2 π 2 n2 /(2m). Therefore, the
given function can be expressed as

1 1
ψ(x) = √ φ1 (x) + √ φ2 (x).
.
2 2

This expression implies that possible values of the measurement of energy are E1 =
h̄2 π 2 /(2m) and E2 = 4h̄2 π 2 /(2m), both with probability 1/2. The probability for
the particle to be in the region between 0 and 1/4 can be calculated as follows:

1/4
P r[0 < x < 1/4] =
. |ψ(x)|2 dx
0
1/4  
= sin2 (π x) + 2 sin(π x) sin(2π x) + sin2 (2π x) dx
0
1/4  1
= (1 − cos(2π x)) + cos(π x) − cos(3π x)
0 2
1

+ (1 − cos(4π x)) dx
2
1 1 1

= x− sin(2π x) + sin(π x) − sin(3π x)
4π π 3π
1/4
1
− sin(4π x) = 0.32.
8π 0

1.10 A quantum particle with unit mass (m = 1) is confined in a box of length

l = 2, and is in a state represented by the following wavefunction: ψ(x) =
N (sin(π x) + sin(2π x)), where N is a normalization constant. Determine N and
calculate the expectation value of the energy.
Solution 1.10 The normalized eigenfunction and eigenvalue are respectively,

nπ x h̄2 π 2 n2
φn (x) = sin(
. ), En = ,
2 8
Problems 33

where n = 1, 2, · · · . Therefore,

.ψ(x) = N (φ2 (x) + φ4 (x)).

It is clear 2
√ or possible to show by direct integration that 2N = 1. As a result,
N = 1/ 2. Either by direct integration or using the above expression for ψ(x), we
find that

1 1 h̄2 π 2 5
.E = E2 + E4 = + h̄2 π 2 = h̄2 π 2 .
2 2 4 4
1.11 Consider a particle in a box with l = 2. Assume that the wavefunction at t = 0
is given by ψ(x) in Problem 1.9. Then, what is the wavefunction at time t = 1?
Solution 1.11 For particle in a box of length l = 2, the eigenfunctions of the
Hamiltonian are
 nπ x 
.φn (x) = sin , n = 1, 2, . . .
2

for which the eigenvalue En = h̄2 π 2 n2 /(8m). Thus, the given ψ(x) can be
expressed as follows:

ψ(x) = φ2 (x) + φ4 (x).

.

For the above wavefunction as that for ψ(x, t = 0), the wavefunction at arbitrary
time is then given by

ψ(x, t) = e−iE2 t/h̄ φ2 (x) + e−iE4 t/h̄ φ4 (x)

.

2 t/(2m) 2 t/m
= e−i h̄π φ2 (x) + e−i2h̄π φ4 (x).

Problems

1.12 Calculate de Broglie wavelength for each case described below.

(a) An electron moving at the speed of 106 m/s.
(b) A proton moving at the speed of 105 m/s.
(c) An oxygen molecule moving at the speed of 300 m/s.
(d) A human body of mass 70 kg moving at the speed of 10 m/s.
1.13 Prove that Eqs. (1.51) and (1.52) for the expectation value are equivalent.
34 1 Concepts and Assumptions of Quantum Mechanics

1.14 Use the example of particle in a box problem to prove that the following
operator is Hermitian:

1 
 =
. x̂ p̂ + p̂ x̂ .
2
1.15 Consider a particle of unit mass (m = 1) in a box of unit length (l = 1) with
the following wavefunction:

.ψ(x) = N (sin(π x) − 2 sin(2π x) + sin(3π x)).

(a) Determine the normalization constant N so that the wavefunction ψ(x) is

normalized.
(b) Determine the expectation value of the energy for this particle.
(c) If the energy is measured for this particle, what are actual values of the
individual measurement possible and their respective probabilities?
(d) Given that the wavefunction above is ψ(x, t = 0), determine the smallest time
t > 0 for which ψ(x, t) = ψ(x).
1.16 For a particle with unit mass (m = 1) confined in a box of unit length l = 1,
its wavefunction is given by ψ(x) = 4N x for 0 < x ≤ 1/4, N for 1/4 < x < 3/4,
and 4N (1 − x) for 3/4 ≤ x < 1. Answer the following questions.
(a) Determine N so that the wavefunction is normalized.
(b) Determine the expectation value of the energy for this particle.
(c) Calculate the probability for the particle to have energy h̄2 π 2 /2.
(d) Calculate the probability for the particle to have energy 2h̄2 π 2 .
(e) Calculate the uncertainty in the position of the particle. What is the expected
minimum of the uncertainty in the momentum of this particle?
Chapter 2
Dirac Notation and Principles
of Quantum Mechanics

The fundamental laws necessary for the mathematical treatment

of a large part of physics and the whole of chemistry are thus
completely known, . . . .
— Paul A. M. Dirac

Abstract This chapter introduces Dirac’s notations for kets and bras, which
represent quantum states, and the rules for inner and outer products between them.
Operators are defined as linear combinations of outer products between kets and
bras. The definition of a Hermitian operator in this Dirac notation and proofs of
its properties are provided. Based on expressions for position, momentum, and
Hamiltonian operators, time dependent and time independent Schrödinger equations
are also provided. After important identities concerning commutators and unitary
operators are detailed, the results for particle in a box problem are re-expressed in
the Dirac notation. The chapter concludes with introduction of direct products that
are used to expand the dimension of the state space.

2.1 Formulation of Quantum Mechanics

Schrödinger’s formulation of quantum mechanics, as introduced in Chap. 1, is based

on the assumption that quantum states can be represented by complex functions
of position and that all physical observables can be expressed as operators acting
on the wavefunction through the position variable. Schrödinger’s formulation has
been essential in obtaining practical solutions of quantum mechanical problems.
However, because it gives the position a special role, the distinctions between
position as a variable for describing quantum state, as an operator for position
measurement, and as an outcome of the measurement, are not always expressed
clearly. This can sometimes cause confusion and misunderstanding.
On the other hand, within Heisenberg’s formulation of quantum mechanics,
which in fact preceded Schrödinger’s formulation by a few months, states and
operators are represented as vectors and matrices, and position and momentum

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 35

S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_2
36 2 Dirac Notation and Principles of Quantum Mechanics

operators are treated in more equivalent manner. Another major difference of

Heisenberg’s formulation is that what changes with time are operators themselves,
rather than the state of a quantum system as assumed in Schrödinger’s formulation.
Heisenberg’s formulation makes the role of an operator clear. It is also convenient
if the eigenstates and eigenvalues are known or can be determined by other means.
However, dealing with all the matrix and vector elements often can be cumbersome
if the number of states is large or infinite.
Although the two formulations by Schrödinger and Heisenberg appear to be
completely different, they can in fact be related to each other through an appropriate
transformation from the function space to the vector space. Dirac facilitated
clarification of this relationship by inventing an elegant set of notations and rules,
which make such transformations look very natural.1 The efficacy of the Dirac
notations in exposing the major principles of quantum mechanics has made it the
most popular framework for describing quantum mechanics.
This chapter provides a concise description of the Dirac notation and presents the
principles of quantum mechanics embedded in that notation [5, 6, 8]. The approach
here is somewhat axiomatic in the sense that abstract mathematical definitions are
given first without offering clear physical justifications. However, once a reader
becomes familiar with them, understanding quantum mechanical principles will be
much more straightforward.

2.2 Ket, Bra, and Products

It is important to note that the state of a quantum system in Schrödinger’s

formulation is represented by the entirety of the wavefunction, not by its particular
value at certain position as the notation .ψ(x) may suggest. Likewise, a quantum
mechanical operator should be viewed as the entire mapping from one state (the full
set of values of wavefunction at all positions) to another state. Thus, .Â(x̂)ψ(x) in
Schrödinger’s formulation should be viewed as representing the full set of results of
“multiplying the value of function .A(x) with the value of the wavefunction at the
same position x.”
A quantum mechanical state, even when represented by Schrödinger’s wavefunc-
tion, can be viewed as a complex-valued vector in Hilbert space [24], with each
element of this vector representing the value of the wavefunction at specific position.
Once this notion of a vector in abstract function space is accepted, its components
can be transformed to different ones involving parameters other than position
variable through an appropriate transformation. In this sense, one can perceive a
quantum state as a vector in abstract function space, which exists independent of
how it is represented. Dirac represented such a vector by a ket or a bra equivalently.

1 Mathematically speaking, Dirac notation can simply be considered as an effective visualization

tool to introduce a complete and normed (or inner product) complex vector space, so called Hilbert
space, and mappings defined in the Hilbert space.
2.2 Ket, Bra, and Products 37

Fig. 2.1 A simple illustration of the relationship between a ket .|α, its Hermitian adjoin, bra .α|,
and the corresponding norm, .α|α, which corresponds to the projection on to the positive side of
the real axis

The relationship between a ket and its corresponding bra is like a complex number
and its conjugate (see Fig. 2.1). More detailed description is provided below.

Notation and Properties of Ket and Ket Space

A ket is denoted as .| · · · , where .· · · represents the parameters specifying the
ket, and has the following properties.
• Given two quantum mechanical states respectively represented by .|α and
.|β defined in the same vector space, it is possible to construct a different

ket as their linear combinations as follows:

|γ  = c1 |α + c2 |β,
. (2.1)

where .c1 and .c2 are arbitrary complex numbers. This linear combination,
indeed, represents a different quantum mechanical state as long as .|α and
.|β are independent, which means that the two are not related to each other

by a constant factor.
• Any two kets related by a constant factor is viewed as representing the
same state. In other words, one of the major assumptions of quantum
mechanics is that multiplying a ket with a nonzero complex number does
not alter the physical state. That is, .|α and .c|α (with .c = 0) represent the
same quantum state.2 For this reason, only the relative value of .c1 to .c2
does matter in Eq. (2.1).
• A complete set of all the kets that can be constructed through linear
combinations constitute an abstract vector space called ket space.
2
Multiplication of any ket or bra with zero results in the null ket or bra.
38 2 Dirac Notation and Principles of Quantum Mechanics

A bra space is a vector space dual to a ket space. Mathematically, these two
can define an inner product space as will be shown below. Using the terms of
matrix theory, a ket can be viewed as a column vector of complex numbers, and
bra as its complex conjugate and transpose, thus a row vector consisting of complex
conjugates. For every ket .|α, there exists a unique bra denoted as .α| and vice versa.
Given a ket defined as a linear combination of different kets, the corresponding bra
can be obtained according to the following rule.

Relationship Between Ket and Bra Space

A state expressed as a linear combination of kets in a ket space is equivalent
to the same linear combination of bras in the corresponding bra space but with
complex conjugates of the corresponding coefficients as follows.

c1 |α + c2 |β + · · · ⇔ α|c1∗ + β|c2∗ + · · · .

. (2.2)

Between a bra and a ket, two types of products can be defined. One is the inner
product, and the other is the outer product. The inner product is defined only
between a ket on the right hand side and a bra on the left hand side. More detailed
definitions and properties are provided below.

Definition and Properties of Inner Product

The inner product between a ket .|α and a bra .β| results in a complex number,
and is expressed as

. (β|) · (|α) = β|α, (2.3)

where .· is a temporary symbol used to denote the product. There are three
important properties involving inner products.
• Two states represented by .|α and .|β are called orthogonal to each other
if .α|β = β|α = 0.
• The complex conjugate of an inner product is equal to the inner product
between corresponding ket and bra interchanged as follows:

β|α = α|β∗ ,
. (2.4)

where .∗ represents the complex conjugate. As a result, .α|α is always real.

(continued)
2.2 Ket, Bra, and Products 39

• The inner product between a ket and bra of the same state is nonnegative.
Namely,

α|α ≥ 0.
. (2.5)

The equality above holds true only when .|α is a null ket that represents
a non-existent state. The inner product .α|α serves as the measure of the
size of the ket (or bra), and its square root is called norm. A normalized
state has a unit norm. As an outcome of Eq. (2.5), the magnitude of the
inner product between two normalized states is always less than or equal
to one (see Appendix of this chapter).

The outer product is defined only between a ket on the left hand side and a bra
on the right hand side, and is used to represent a mapping from a ket (bra) to a bra
(ket). Its notation and detailed properties are provided below.

Definition and Properties of Outer Product

The outer product between a ket .|α and a bra .β| is defined as

. (|α) · (β|) = |αβ|, (2.6)

where again .· is a temporary symbol used to denote the product. In relation

to inner products, outer products have two important properties as described
below.
• The combination of inner and outer products satisfy the following associa-
tive rule:

. (|αβ|) · |γ  = |α · (β|γ ) = |αβ|γ . (2.7)

That is, .|αβ| is nothing but an operator that turns an arbitrary state .|γ 
into .|α multiplied with a complex number .β|γ . Similarly, the following
associative rule always applies:

. (α|β) γ | = α| (|βγ |) = α|βγ |, (2.8)

which shows that .|βγ | is an operator that turns an arbitrary bra state .α|
into another bra state .γ | multiplied with a complex number .α|β.

(continued)
40 2 Dirac Notation and Principles of Quantum Mechanics

• Both inner and outer products satisfy distributive rules as follows.

α| (C1 |β + C2 |γ ) = C1 α|β + C2 α|γ , .

. (2.9)
(C1 |β + C2 |γ ) α| = C1 |βα| + C2 |γ α|. (2.10)

2.3 Operators

An operator acts on a ket (bra) and produces another ket (bra). It can always be
represented as a linear combination of outer products of kets and bras. Like the
relationship between ket and bra, for a given operator .Ô, there is an equivalent form
of the same operator that “mirrors” the action in ket space to the one in bra space,
and vice versa as describe below.

Hermitian Conjugate of an Operator

Given an operator .Ô, its Hermitian conjugate .Ô † satisfies the following dual
correspondence rule for an arbitrary ket .|α:

Ô|α ⇔ α|Ô † .
. (2.11)

Therefore, if .Ô = c1 |α1 β1 | + c2 |α2 β2 | + · · · ,

Ô † = c1∗ |β1 α1 | + c2∗ |β2 α2 | + · · · .

. (2.12)

For arbitrary .|α and .|β, one can show that

 ∗
α|Ô † |β = β|Ô|α .
. (2.13)

For a given operator .Ô, its eigenket3 is defined as the ket satisfying the following
relationship:

Ô|u = λu |u,
. (2.14)

where .λu is the eigenvalue of the operator .Ô corresponding to the eigenket .|u
and is a complex number in general. According to the definition of the Hermitian

3 The state represented by this eigenket is called eigenstate.

2.3 Operators 41

conjugate, Eq. (2.14) implies that

u|Ô † = u|λ∗u = λ∗u u|,

. (2.15)

where .λ∗u is the complex conjugate of .λu .

The operator that does not alter any ket or bra is called the identity operator, and
is denoted as .1̂. That is, any ket or bra is an eigenket or eigenbra of the identity
operator with its eigenvalue as one.

2.3.1 Hermitian Operator

Definition of Hermitian Operator

An operator  is Hermitian if it is identical to its own Hermitian conjugate
as follows:

 = † .
. (2.16)

Therefore, inserting the above identity into Eq. (2.13), one can show that the
Hermitian operator  satisfies the following identity:

α|Â|β = β|Â|α∗ ,
. (2.17)

where |α and |β are arbitrary states.

The definition of Hermitian operators given above makes it straightforward to prove

two theorems proven in Chap. 1 for operators being applied to wavefunctions in one-
dimensional space. Namely, the eigenvalues of a Hermitian operator are always real
numbers, and two eigenstates of a Hermitian operator with different eigenvalues are
always orthogonal. Full statements and proofs of these theorems are provided below.
Theorem 2.1 Given a Hermitian operator .Â, all of its eigenvalues are real
numbers.
Proof Given that .|a is an eigenstate of . with finite norm and eigenvalue .λa , the
use of .|α = |β = |a in Eq. (2.17) leads to the following identity:

λa a|a = a|Â|a = a|Â|a∗ = λ∗a a|a.

. (2.18)

Since .a|a is real and is assumed to be nonzero, this identity leads to .λa = λ∗a ,
which is true only when .λa is a real number. ⊓
⊔
42 2 Dirac Notation and Principles of Quantum Mechanics

Theorem 2.2 The eigenkets of a Hermitian operator . corresponding to different

eigenvalues are always orthogonal.
Proof Let us assume that .|a ′  and .|a ′′  are eigenkets of . with two different
eigenvalues, .λa ′ and .λa ′′ . Then,

a ′ |Â|a ′′  = λa ′ a ′ |a ′′  = λa ′′ a ′ |a ′′ ,
. (2.19)

where the first equality results from applying . to .a ′ | and the second equality
results from applying . to .|a ′′ . Since .λa ′ = λa ′′ , the only possibility that the above
equality is true is that .a ′ |a ′′  = 0. This proves the above statement regarding the
orthogonality. ⊔
⊓
All physical observables correspond to Hermitian operators. If not, one has
to accept that complex values can result from the measurement of the physical
observable, which is impossible because all the values we deal with are real
numbers. Well known examples of physical observables are position and momentum
operators, which will be described in more detail in the next subsection. Most
importantly, a Hamiltonian operator is Hermitian.

2.3.2 One Dimensional Position Operator and Eigenket

Let us consider a one dimensional position operator .x̂. Then, one can define a
position eigenket .|x such that

.x̂|x = x|x = |xx. (2.20)

Similarly, the action of the position operator on .x| is defined by

.x|x̂ = x|x = xx|. (2.21)

Then, the Schrödinger’s wavefunction, or its value at particular position x, can be

defined as an inner product of a state ket with the position bra , .x|. For example,
consider a state .|α. Then, the corresponding wavefunction can be defined as

. ψα (x) = x|α. (2.22)

Taking the complex conjugate of this,

ψα∗ (x) = α|x.

. (2.23)
2.3 Operators 43

Combining the above definitions, one can find out the effect of the operation of .x̂ on
an arbitrary state .|α as follows:

.x|x̂|α = (x|x̂)|α = xx|α. (2.24)

The set of position eigenkets is always complete as follows:

 ∞
. 1̂ = dx|xx|. (2.25)
−∞

This means that the values of wavefunction at all positions provide complete
information on the state. This becomes clear by applying the above identity to an
arbitrary state as follows:
 ∞  ∞  ∞
|α = 1̂|α =
. dx |xx|α = dx |xψα (x) = dx ψα (x)|x.
−∞ −∞ −∞
(2.26)

Note that the last equality in the above equation results from .ψα (x) being a complex
number and that its order with respect to the ket does not matter. The above identity
implies that the value of a Schrödinger’s wavefunction at each x can be viewed
as nothing but the coefficient (with the inclusion of integration element dx) in the
expansion of the state .|α in terms of the position eigenkets. This also ensures that
the normalization condition within the Dirac notation is the same as that involving
Schrödinger’s wavefunction as follows:
 ∞  ∞
α|α = α|1̂|α =
. dx α|xx|α = dx |ψα (x)|2 . (2.27)
−∞ −∞

Applying Eq. (2.25) to another position ket, we also find that

 ∞
|x ′  =
. dx |xx|x ′ . (2.28)
−∞

Since the above equation must be true for any value of .x ′ , this means that

x|x ′  = δ(x − x ′ ),
. (2.29)

where .δ(x − x ′ ) is the Dirac-delta function defined in Chap. 1, and is zero for .x = x ′
and .∞ for .x = x ′ while its integration is equal to one. Note that this Dirac-delta
function is not a genuine function, but can be defined only as a limiting value of a
sequence of functions that attain the stated properties. Thus, care should be taken
working with the delta function.
44 2 Dirac Notation and Principles of Quantum Mechanics

2.3.3 One Dimensional Momentum Operator and Eigenket

It was stated in Chap. 1 that measurement of momentum corresponds to taking the

first order derivative of the wavefunction, as shown by Eq. (1.24). Let us assume that
.φp (x) is an eigenfunction of the momentum operator with eigenvalue p. According

to Schrödinger’s formulation, this should satisfy

h̄ ∂
. φp (x) = pφp (x). (2.30)
i ∂x
A general solution of the above equation is

φp (x) = Np eipx/h̄ .
. (2.31)

where .Np is a normalization constant. In the Dirac notation, the above eigenfunction
of momentum operator can be viewed as the inner product between momentum
eigenket and position eigenbra as follows:

φp (x) = x|p,
. (2.32)

where .|p is an eigenstate of the momentum operator satisfying

p̂|p = p|p.
. (2.33)

As is the case of position ket and bra, the momentum ket and bra are assumed to
satisfy the following completeness relation:
 ∞
1̂ =
. dp |pp|. (2.34)
−∞

Then, the normalization constant .Np can be determined in the following way:
 ∞
. δ(x − x ′ ) = x|x ′  = x|1̂|x ′  = dpx|pp|x ′ 
−∞
 ∞ ′
 ∞
′
= Np2 dp eip(x−x )/h̄ = Np2 h̄ dk eik(x−x ) . (2.35)
−∞ −∞

This is consistent with the definition of the Dirac-delta function shown in Eq. (1.85)
if .Np2 = 1/(2π h̄). Therefore,

1
φp (x) = x|p = √
. eipx/h̄ . (2.36)
2π h̄

This identity can be used to find out the effect of the operation of .p̂ on the position
eigenket.
2.3 Operators 45

2.3.4 Expressions for Momentum Operator

One can show that

h̄ ∂
x|p̂|p = x|p|p = px|p =
. x|p. (2.37)
i ∂x

Taking complex conjugate of the above equation and noting that .p̂ is a Hermitian
operator, one can find that
 
h̄ ∂ h̄ ∂
p|p̂|x = −
. p|x = p| − |x , (2.38)
i ∂x i ∂x

Since the above relationship holds for any .p|, this implies that

h̄ ∂
p̂|x = −
. |x, (2.39)
i ∂x
where
h̄ ∂ h̄ |x + δx − |x
. − |x = − lim . (2.40)
i ∂x i δx→0 δx

Note that the sign on the righthand side of Eq. (2.39) is opposite to that for the
Schrödinger’s definition of momentum measurement given by Eq. (1.24). In fact,
the two is consistent with each other and can be shown to be equivalent if we use
a proper version of the Schrödinger’s definition for a momentum operation in the
Dirac notation as described below.
Taking the Hermitian conjugate of Eq. (2.39) leads to

h̄ ∂
x|p̂ =
. x|. (2.41)
i ∂x
Now taking the inner product of the above identity with an arbitrary state .|ψα , one
can obtain
h̄ ∂ h̄ ∂
x|p̂|ψα  =
. x|ψα  = ψα (x). (2.42)
i ∂x i ∂x
The above identity is equivalent to Eq. (1.24) if we interpret the left hand side as
an application of the momentum operation on a Schrödinger’s wavefunction. Note
that .p̂ψα (x) does not make sense in the Dirac notation because an operator can be
applied only to either a ket or bra, not to a complex number.
46 2 Dirac Notation and Principles of Quantum Mechanics

2.3.5 Schrödinger Equations in the Dirac Notation

Time Dependent Schrödinger Equation

The time dependent Schrödinger equation for |ψ; t, a ket representing
a quantum mechanical state labeled as ψ and dependent on time t as a
parameter, is as follows:

∂
i h̄
. |ψ; t = Ĥ |ψ; t. (2.43)
∂t
Taking the Hermitian conjugate of the above equation and noting that
Hamiltonian operator is Hermitian, the above equation for a ket is equivalent
to an equation for the following equation for the corresponding bra.

∂
. − i h̄ ψ; t| = ψ; t|Ĥ . (2.44)
∂t

On the other hand, for the case of a stationary state, namely, .|ψ; t = |ψe−iEt/h̄ ,
the time independent part of this state satisfies the time independent Schrödinger
equation as described below.

Time Independent Schrödinger Equation

The time independent Schrödinger equation for a stationary eigenket .|ψ of a
Hamiltonian .Ĥ with eigenvalue E is

Ĥ |ψ = E|ψ.
. (2.45)

The above equation for an eigenket is equivalent to the following equation for
the corresponding bra.

ψ|Ĥ = Eψ|.
. (2.46)

For a particle with mass m in one dimension subject to a potential energy .V (x),
the Hamiltonian operator is defined

p̂2
. Ĥ = + V̂ (x̂). (2.47)
2m
2.3 Operators 47

Note that the above expression is the same as the first form in Eq. (1.28). However,
note that the second form in this equation has its meaning only when projected to
the position state. In the Dirac notation, this is realized by taking the inner product
of the Schrödinger equation with the position state. For example, taking the inner
product of Eq. (2.45) with .x|, we find that

p̂2
 
.x|Ĥ |ψ = x| + V̂ (x̂) |ψ
2m
 
1 h̄ ∂ 2
 
= + V (x) x|ψ
2m i ∂x
 
h̄2 ∂ 2
= − + V (x) ψ(x) = Eψ(x), (2.48)
2m ∂x 2

where we have used the identity of Eq. (2.41) twice in obtaining the second equality
and used the definition of .ψ(x) = x|ψ in the last line. Note that the last line has
the same form as the original Schrödinger equation.

2.3.6 Commutator

Given two operators . and .B̂, there are two ways to order them in creating product
operations. The difference between them is called a commutator, which plays an
important role in quantum mechanics and is defined as follows.

[Â, B̂] = ÂB̂ − B̂ Â.

. (2.49)

When the two operators can be measured simultaneously, the above commutator
is zero. A good example for two commuting operators is the position operator and
any of its function. Namely, for any operator .Â(x̂) defined as a function of position
operator only, .[x̂, Â(x̂)] = 0.
If the commutator between two physical observables . and .B̂ is nonzero, it is
impossible to measure them simultaneously with arbitrary precision. Rather, the
product of uncertainties of the two observables has lower bound given by half the
absolute value of the expectation value of the commutator. Appendix of this chapter
provides a detailed derivation of this a general uncertainty relationship.
Expressing an operator as a sum of outer products in the Dirac notation makes
it straightforward to calculate the commutator. For example, consider two operators
.|αβ| and .|γ δ|. The commutator between the two is given by

[|αβ|, |γ δ|] = |αβ|γ δ| − |γ δ|αβ|

.

= (β|γ )|αδ| − (δ|α)|γ β|. (2.50)

48 2 Dirac Notation and Principles of Quantum Mechanics

It is easy to see that this commutator is zero (i) if .|α = |γ  and .β| = δ|, or (ii)
β|γ  = δ|α = 0.
.

The commutator with the most fundamental importance in quantum mechanics

is that between position and momentum, which is calculated below.

Commutator Between Position and Momentum Operator

Let us first apply .[x̂, p̂] to an arbitrary ket .|α. Taking the inner product of the
resulting ket with .x| leads to

.x|[x̂, p̂]|α = x|(x̂ p̂ − p̂ x̂)|α

= x|x̂ p̂|α − x|p̂ x̂|α
h̄ ∂
= xx|p̂|α − x|x̂|α
i ∂x
h̄ ∂ h̄ ∂
=x x|α − xx|α
i ∂x i ∂x
h̄ ∂ h̄ h̄ ∂
=x x|α − x|α − x x|α
i ∂x i i ∂x
= i h̄x|α = x|(i h̄)|α. (2.51)

Since the above identity holds true for arbitrary .x| and .|α, it implies the
following identity:

.[x̂, p̂] = i h̄. (2.52)

There are a few important rules that commutators satisfy. From the definition of
the commutator, it is straightforward to show the following identities:

. [a Â, B̂] = [Â, a B̂] = a[Â, B̂], . (2.53)

[Â, B̂] = −[B̂, Â], . (2.54)
[Â + B̂, Ĉ] = [Â, Ĉ] + [B̂, Ĉ], . (2.55)
[Â, B̂ + Ĉ] = [Â, B̂] + [Â, Ĉ], (2.56)

where .Â, .B̂, and .Ĉ are arbitrary operators, and a is an arbitrary complex number.
There is also an important identity involving products between three operators as
shown below.
Theorem 2.3 For any operator .Â, .B̂, and .Ĉ, the following identity is always true.

[ÂB̂, Ĉ] = Â[B̂, Ĉ] + [Â, Ĉ]B̂.

. (2.57)
2.3 Operators 49

Proof Employing the definition of a commutator given by Eq. (2.49), the lefthand
side of the above equation can be expressed as

[ÂB̂, Ĉ] = ÂB̂ Ĉ − ÂĈ B̂ + ÂĈ B̂ − Ĉ ÂB̂

.

= Â(B̂ Ĉ − Ĉ B̂) + (ÂĈ − Ĉ Â)B̂

= Â[B̂, Ĉ] + [Â, Ĉ]B̂. (2.58)

⊔
⊓
In a similar manner, one can also prove the following identity:

[Â, B̂ Ĉ] = B̂[Â, Ĉ] + [Â, B̂]Ĉ.

. (2.59)

Finally, it is also easy to prove the following Jacobi identity by directly expanding
expressions for commutators:

[Â, [B̂, Ĉ]] + [B̂, [Ĉ, Â]] + [Ĉ, [Â, B̂]] = 0.

. (2.60)

2.3.7 Compatibility and Completeness

Two operators . and .B̂ are called compatible when their commutator .[Â, B̂] = 0. If
two operators are compatible, one can always find the set of simultaneous eigenkets
of . and .B̂. A set of operators is called maximally compatible if all the operators
in this set are compatible and if there is no additional observable that are compatible
with all the operators within the set.
A basis is a set of linearly independent kets (or equivalently a set of linearly
independent bras). The basis is called orthonormal if all the kets in the set have the
unit norm and are orthogonal to other kets in the set. A basis set consisting of .|n’s
is called complete if it satisfies the following identity:

. |nn| = 1̂, (2.61)

n

where n represents the collection of all parameters need to specify each ket. The
sum has to become integration if n is a continuous real parameter as in the cases
of position or momentum. The set of simultaneous eigenkets of maximally com-
patible observables is orthonormal and complete. The number of such eigenkets is
called the dimension of the ket or bra space.
50 2 Dirac Notation and Principles of Quantum Mechanics

Let us assume that a set of states .|n’s forms a complete orthonormal basis in the
sense that Eq. (2.61) holds true and that .n|n′  = δnn′ . Then, for a given state .|α,

|α = 1̂|α =
. |nn|α = αn |n, (2.62)
n n

where .αn = n|α. Both associative and distributive rules for inner and outer
products have been used in deriving the above equality. Similarly,

α| = α|1̂ =
. α|nn| = αn∗ n|. (2.63)
n n

Taking the inner product between .α| and .|α shown above,

α|α =
. α|nn|mm|α
n m

= α|nδnm m|α = |αn |2 . (2.64)

n m n

If .|α is normalized, the above relation implies that . n |αn |2 = 1.

Given a complete orthonormal basis, it is easy to find an expression for any
operator . in terms of linear combinations of those involving outer products
between states within the basis. This expression can be found by applying Eq. (2.61)
on both sides of . as follows:

 = 1̂Â1̂ =
. |nn|Â|mm|
n m

= |nAnm m| = Anm |nm|, (2.65)

n m n m

where .Anm = n|Â|m. For the special case where .|n’s are eigenstates of .Â, the
above expression reduces to . = n An |nn|.

2.3.8 Measurement Operator

Measurement in quantum mechanics is in general a complicated process through

which an experimental device interacts with a quantum system and produces real
numbers as its record [25]. How to understand and define measurement process has
remained an important foundational issue of quantum mechanics. Although debates
on quantum measurement still continue, it has been shown that it is possible to
formulate measurement in a manner fully consistent with the theory of quantum
mechanics [26–28] unlike the old Copenhagen interpretation that required ad hoc
2.3 Operators 51

interaction with classical systems. In addition, there is good consensus that a large
class of measurements, at least theoretically, can be represented by well-defined
operators.4
The simplest operator representing measurement process is called projective
measurement. Consider a quantum state .|ψ and a physical observable .Â. Assume
that a measurement of . has produced a real number. If the same measurement of .Â
is made immediately afterwards, the same number has to be produced according to
the principle of causality. Therefore, the state right after the measurement of . has
to be an eigenstate of . and the outcome of the measurement has to be an eigenvalue
of .Â. Let us denote this eigenvalue as .λa and assume that we can identify a single
normalized eigenstate .|λa  corresponding to the state right after the measurement.
Then, the following projection operator can be used to represent the measurement
operation:

. P̂λa = |λa λa |. (2.66)

Thus, the process of measuring .λa for a given state .|ψ amounts to .P̂λa |ψ =
|λa λa |ψ, and the probability of having this outcome is given by .|λa |ψ|2 .
More generally, let us assume that it is possible to identify all observables
compatible with .Â. Then, denoting the set of all eigenvalues for the operators as .ξ ,
we can identify eigenstates, .|λa , ξ ’s, which are orthonormal and complete. Thus,

.1̂ = |λa , ξ λa , ξ |. (2.67)

λa ,ξ

Then, the least intrusive projective measurement of .λa is represented by

. P̂λa = |λa , ξ λa , ξ |. (2.68)

ξ

The outcome of this measurement is a linear combination of states with different

values of .ξ ,

P̂λa |ψ =
. |λa , ξ λa , ξ |ψ. (2.69)
ξ

The general projective measurement described above can be extended further

to a so called positive operator valued measure (POVM) measurement, which is

4 This does not yet mean that the issue of quantum measurement is fully settled because there is
still large gap between theoretical definition and experimental practice of quantum measurement.
It is often the case that reliable experimental realization of fully quantum mechanical measurement
is extremely difficult, whereas reliable theoretical modeling of experimentally viable measurement
can be challenging.
52 2 Dirac Notation and Principles of Quantum Mechanics

defined as a nonnegative Hermitian operator . ˆ a . The subscript a in this operator

collectively represents a set of parameters defining the outcome of the measurement,
and integration (or summation) of all the operators over all possible values of a leads
to the identity operator as follows:

1̂ =
. da ˆ a . (2.70)

It is assumed that different . ˆ a ’s do not necessarily commute with each other.

The only requirement is that . ˆ a is Hermitian and is positive, which means that
.ψ| ˆ a |ψ ≥ 0 for any .|ψ. The POVM measurement is more general than

projective measurement and can account for physical situations where the details
of post-measurement states are not completely specified [25].

2.3.9 Unitary Operator

Definition of Unitary Operator

A unitary operator, denoted as Û here, satisfies the following condition:

Û † Û = Û Û † = 1̂.
. (2.71)

In other words, the Hermitian conjugate of a unitary operator corresponds to

its inverse operator. This property allows the operator to preserve the norm of
any quantum state and also the orthogonality between two states.

Unitary operators are essential components in the formulation of quantum mechan-

ics. Let us consider an arbitrary ket .|α, which is not necessarily normalized. Then,
let us introduce another ket created by applying a unitary operator to this as follows:

|α ′  = Û |α.
. (2.72)

Note that the above equation does not correspond to any measurement process
because .Û is not Hermitian except for the trivial case of the identity operator. Rather,
this can simply be viewed as either transformation needed for better representation
of quantum states or transition that happens naturally.
The main reason why unitary operators are important is because they conserve
the norm and the orthogonality between states. In other words, while the new
2.3 Operators 53

transformed state .|α ′  is different from .|α in general, the two have the same norm
as follows:
    
′ ′ †
.α |α  = α|Û Û |α = α| Û † Û |α = α|1̂|α = α|α. (2.73)

When two orthogonal states are transformed by the same unitary operator, they
remain orthogonal as well. Let us consider another ket .|β and its unitary trans-
formation as follows:

|β ′  = Û |β.
. (2.74)

Then, it is also straightforward to show that the inner product between .|β ′  and .|α ′ 
remains the same as that between .|α and .|β as follows:
    
β ′ |α ′  = β|Û † Û |α = β| Û † Û |α = β|1̂|α = β|α.
. (2.75)

Thus, for the case where .β|α = 0, the above identity implies that .β ′ |α ′  = 0.
An important application of a unitary operator is a transformation from one basis
to another. Consider a set of orthonormal states, .|ϕn ’s with .n = 1, 2, · · · , satisfying
the following completeness relationship:
∞
1̂ =
. |ϕn ϕn |. (2.76)
n=1

Thus, these form a complete orthonormal basis. Then, the following set of states
created by a unitary transformation,

|ϕn′  = Û |ϕn , n = 1, 2, 3, · · ·
. (2.77)

remain orthonormal due to Eqs. (2.73) and (2.75). The fact that these satisfy the
completeness relationship can also be shown easily. Let us apply .Û on the lefthand
side and .Û † on the righthand side of Eq. (2.76). Then,

∞ ∞ ∞
 
1̂ = Û
. |ϕn ϕn | Û † = Û |ϕn ϕn |Û † = |ϕn′ ϕn′ |, (2.78)
n=1 n=1 n=1

where the first equality results from the definition of the unitary operator, Eq. (2.71).
The above identity indeed proves that .|ϕn′ ’s form a complete orthonormal basis.
54 2 Dirac Notation and Principles of Quantum Mechanics

Given a Hermitian operator .Â, the following operator5 is always a unitary

operator.

i
ÛA = exp − Â
.
h̄
1 −i 2 2 1 −i n n
   
i
= 1− Â + Â + · · · + Â + · · ·
h̄ 2! h̄ n! h̄
   
= cos Â/h̄ − i sin Â/h̄ , (2.79)

where the third equality comes from the fact that .e−ix = cos(x) − i sin(x). The
fact that .ÛA is a unitary operator can be proved as follows. Taking the Hermitian
conjugate of .ÛA , we find that
   
ÛA† = cos Â/h̄ + i sin Â/h̄ ,
. (2.80)
   
where .cos Â/h̄ and .sin Â/h̄ are Hermitian operators, since they are real
functions of a Hermitian operator. Then,
         
ÛA ÛA† = cos Â/h̄ − i sin Â/h̄
. cos Â/h̄ + i sin Â/h̄
   
= cos2 Â/h̄ + sin2 Â/h̄ = 1̂, (2.81)

where the last equality comes from the fact that .cos2 (x) + sin2 (x) = 1 and that this
identity remains true whether x is a number or an operator. In a similar manner, it is
possible to prove that .ÛA† ÛA = 1̂. Therefore, .ÛA is a unitary operator.
A well-known example of the type of the unitary operator given above is the time
evolution operator defined below.

Definition of a Time Evolution Operator

For any Hamiltonian operator .Ĥ , which is time independent, the correspond-
ing time evolution operator is defined as

it
ÛH (t) = exp − Ĥ .
. (2.82)
h̄

(continued)

5 This is an example of the definition for a function .f (Â) of an operator .Â, which implies
substituting . for x in the Taylor expansion of .f (x).
2.4 Particle in a One-Dimensional Box: Revisited with the Dirac Notation 55

If we known the complete set of the eigenstates and eigenvalues of the

Hamiltonian operator, namely, .|φn ’s and .En ’s with .n = 1, 2, 3, · · · such that
.Ĥ |φn  = En |φn , the above time evolution operator can also be expressed as

ÛH (t) =
. e−iEn t/h̄ |φn φn |. (2.83)
n

It is easy to show that the time evolution operator satisfies the time dependent
Schrödinger equation as follows:
 
∂ iEn −iEn t/h̄
i h̄
. ÛH (t) = i h̄ − e |φn φn |
∂t n
h̄

= e−iEn t/h̄ Ĥ |φn φn |

n

= Ĥ ÛH (t). (2.84)

Given an arbitrary state .|ψ at time .t = 0, the state at arbitrary time t can be
expressed as

|ψ; t = ÛH (t)|ψ.

. (2.85)

The above state satisfies the time dependent Schrödinger equation as a result of
Eq. (2.84).

2.4 Particle in a One-Dimensional Box: Revisited with the

Dirac Notation

In Chap. 1, we have determined the eigenvalue, Eq. (1.58), and eigenfunction,

Eq. (1.61), for a particle in a one dimensional box. Let us denote the eigenket of
the Hamiltonian with eigenvalue .En as .|n. Thus, these satisfy the following time
independent Schrödinger equation:

Ĥ |n = En |n, n = 1, 2, · · · ,
. (2.86)

where the full information on .|n can be obtained from

√
2/ l sin (nπ x/ l) , 0 < x < l
x|n =
. (2.87)
0 , x ≤ 0 or x ≥ l
56 2 Dirac Notation and Principles of Quantum Mechanics

Since the above function is real, .n|x = x|n. The set of eigenstates given above
is orthonormal as follows:
 ∞
.n|m = dxn|xx|m
−∞

2
 l  nπ x   mπ x 
= dx sin sin
l 0 l l
1
 l 
(n − m)π x
 
(n + m)π x

= dx cos − cos
l 0 l l
= δnm . (2.88)

These eigenstates form a complete orthonormal basis and can be used to represent
any quantum state as a linear combination. Let us consider a quantum state .|ψ
given by a linear combination of eigenstates as follows:
∞
|ψ =
. Cn |n. (2.89)
n=1

This state .|ψ is assumed to be normalized and therefore satisfies the following
normalization condition:
∞ ∞ ∞
ψ|ψ =
. Cn∗ Cn′ n|n′  = |Cn |2 = 1. (2.90)
n=1 n′ =1 n=1

Given that the state .|ψ is the state at time .t = 0, the state at another time t is fully
determined as follows:
∞ ∞
|ψ; t = ÛH (t)|ψ =
. Cn ÛH (t)|n = Cn e−iEn t/h̄ |n, (2.91)
n=1 n=1

where .En has been defined by Eq. (1.58).

2.5 Direct Product

The direct product is different from the inner or outer product in that it is not an
operation involving ket (bra) and bra (ket), but rather represents an expansion of the
vector space or its dimensionality.6 A direct product can be defined between any two

6 Note that the dimensionality of the vector space is different from that of the space, and can often
be infinite.
2.5 Direct Product 57

independent ket (bra) spaces, each with any dimension, resulting in a combined ket
(bra) space with higher dimension. For example, if a ket .|α is defined in a ket space
.A with dimension N and another ket .|β is defined in a ket space .B with dimension

M, one can define the following ket

|α, β ≡ |α ⊗ |β = |α|β,

. (2.92)

which is defined in the direct product space .A⊗B of dimension .M ×N. The second
equality in the above expression is often used as an abbreviation for the first form
and should not be confused with incorrect notation for inner or outer product.
A good example of the direct product is the three dimensional position state.
For a three-dimensional position vector expressed in Cartesian coordinate system as
.r = xex + yey + zez , one can define the corresponding position state employing

direct product as follows:

|r ≡ |x ⊗ |y ⊗ |z = |x|y|z.

. (2.93)

This is an eigenstate of the position operator .r̂ = x̂ex + ŷey + ẑez as follows:

r̂|r = r|r.
. (2.94)

Similarly, for a three dimensional momentum vector defined in Cartesian coordinate

system as .p = px ex +py ey +pz ez , one can define the three dimensional momentum
state employing the direct product as follows:

|p ≡ |px  ⊗ |py  ⊗ |pz  = |px |py |pz .

. (2.95)

This is an eigenstate of the momentum operator .p̂ = p̂x ex + p̂y ey + p̂z ez as follows:

p̂|p = p|p.
. (2.96)

The direct product can be used to define a state for many particle systems. For
example, one can define the position and momentum kets of two particles:

. |r1 , r2  ≡ |r1  ⊗ |r2  = |r1 |r2 , . (2.97)

|p1 , p2  ≡ |p1  ⊗ |p2  = |p1 |p2 . (2.98)

For a system with N particles, one can construct position and momentum vector
states by expanding the above direct products. It is assumed that the sets of 3N
dimensional position or momentum vector states constructed this way can be used
to completely specify the quantum state of the N particle system.
58 2 Dirac Notation and Principles of Quantum Mechanics

2.6 Summary and Questions

The Dirac notation provides an efficient and general way to represent states and
rules of quantum mechanics. In this notation, Schrödinger’s formulation of quantum
mechanics employing wavefunction and differential operators can be viewed as a
position representation of quantum mechanics.
In the Dirac notation, a quantum state is represented by either ket .| · · ·  or bra
.· · · |, where .· · · can be any index or symbol employed to label the state. Any

complex-valued linear combination of kets (bras) belonging to the same ket (bra)
space becomes a ket (bra) in the same ket (bra) space, which in general represents a
different quantum state. An inner product can always be defined between a bra and
a ket defined in these equivalent bra and ket spaces, as in Eq. (2.3), and is a complex
number in general.
If the ket and bra represent the same state, the inner product becomes a
nonnegative real number and is considered as the magnitude of the state. Quantum
mechanically, this magnitude does not have any physical significance. Thus, any
ket or bra normalized to have unit magnitude is a convenient choice to carry all the
necessary quantum mechanical information.
The inner product between normalized ket and bra can be viewed as indicating
the dependence on each other, and its absolute value is always smaller than one if
they represent two different states. If the inner product becomes zero, the two state
is called orthogonal. Orthonormal states are orthogonal and normalized states, and
serve as convenient choice to span either the ket or bra space. If a set of orthonormal
states can be used to represent any state as a linear combination, it is called a
complete orthonormal basis.
Outer products can also be defined between any ket in a ket space and another bra
in the equivalent bra space, as in Eq. (2.6), and can be used to represent an operator
or a mapping that transforms a ket (bra) to another ket (bra) in the same ket (bra)
space.
Given an operator, its Hermitian conjugate is defined by Eq. (2.11). An operator
is called Hermitian if it is the same as its Hermitian conjugate. Eigenvalues of
a Hermitian operator are always real and can be used to represent all physical
observables including the Hamiltonian. Two operators are called compatible if
the commutator between the two, as defined in Eq. (2.49), is zero. The position
and momentum operators are not compatible since their commutator is nonzero as
shown by Eq. (2.52). Eigenstates of maximally compatible physical observables can
be used to form a complete orthonormal basis. The measurement operator can be
defined as an outer product of a state with itself, sum of such outer products, or sum
of positive Hermitian operators.
The Hermitian conjugate of a unitary operator is its inverse and satisfies the
relationship of Eq. (2.71). The inner product between a ket and a bra remains the
same even after they are transformed by a unitary operator. All time evolution
operators, which satisfy the time dependent Schrödinger equation, are unitary.
Appendix: Cauchy-Schwarz Inequality and a General Uncertainty Relationship 59

Direct product can be formed between any kets (bras) from different and
independent ket (bra) spaces, and are used to expand the ket (bra) space. The new
space formed this way is called direct product space. Operators in this direct product
space are defined as outer products of corresponding kets and bras in the direct
product space.

Questions

• How is a ket or bra space defined?

• Is the inner product between a ket and its corresponding bra always
nonnegative?
• Can any linear combination of outer products of kets and bras in the Dirac
notation represent a physical observable?
• How is an eigenket or eigenbra defined for a given operator?
• Why is Hermitian operator important in quantum mechanics?
• When do two operators become compatible with each other?
• How can the identity operator be expressed in terms of complete orthonor-
mal basis states?
• What is the major difference between projective measurement and POVM
measurement?
• How is time evolution operator defined for a given time independent
Hamiltonian operator?
• When can a direct product space be formed?
• How can an operator defined in a given ket and bra space be extended to a
new direct product space?

Appendix: Cauchy-Schwarz Inequality and a General

Uncertainty Relationship

The fact that a ket or bra space forms a well defined normed vector space means that
the inner product between any two states satisfies the Cauchy-Schwartz inequality.
The statement of this inequality and its proof [6] are summarized below.
Theorem 2.4 Cauchy-Schwartz inequality: For any states .|α and .|β, which have
nonzero norms, the following inequality always holds:

|α|β|2 ≤ α|αβ|β
. (2.99)

Proof Consider the following linear combination of .|α and .|β:

|γ  = |α + C|β.
. (2.100)
60 2 Dirac Notation and Principles of Quantum Mechanics

Then, since the inner product of any ket with its corresponding bra is always
nonnegative,

γ |γ  = (α| + C ∗ β|)(|α + C|β)

.

= α|α + C ∗ β|α + Cα|β + |C|2 β|β ≥ 0. (2.101)

Let us choose
β|α
C=−
. . (2.102)
β|β

Employing the above expression in Eq. (2.101), we find that

α|β β|α |α|β|2 α|αβ|β − |α|β|2

α|α −
. β|α − α|β + β|β = ≥ 0.
β|β β|β β|β2 β|β
(2.103)

Since .β|β > 0, the above inequality implies that

α|αβ|β − |α|β|2 ≥ 0,
. (2.104)

which completes the proof. ⊓

⊔
For the case where .|α and .|β are normalized, the above theorem shows that the
magnitude of the inner product between the two cannot be greater than 1.
The above Cauchy-Schwarz inequality can be used to prove general uncertainty
relationship between two non-commuting operators . and .B̂ as stated below [6].
Theorem 2.5 For two Hermitian operators . and .B̂ and for any state .|ψ, the
following inequality holds.

1  2
ψ|Â2 |ψψ|B̂ 2 |ψ ≥ ψ|[Â, B̂]|ψ , (2.105)

.
4

where . =  − ψ|Â|ψ and .B̂ = B̂ − ψ|B̂|ψ.

Proof Let us introduce .|α = Â|ψ and .|β = B̂|ψ. Then, according to the
Cauchy-Schwarz inequality proven above,

ψ|Â2 |ψψ|B̂ 2 |ψ = α|αβ|β ≥ |α|β|2 .

. (2.106)

Employing the definitions of .|α and .|β, we find that

1 1
α|β = ψ|ÂB̂|ψ =
. ψ|[Â, B̂]|ψ + ψ|(ÂB̂ + B̂Â)|ψ
2 2
1 1
= ψ|[Â, B̂]|ψ + ψ|(ÂB̂ + B̂Â)|ψ, (2.107)
2 2
Exercise Problems with Solutions 61

where in the second equality, the fact that .[Â, B̂] = [Â, B̂] has been used. Note
that
 ∗
. ψ|[Â, B̂]|ψ = ψ|[B̂, Â]|ψ = −ψ|[Â, B̂]|ψ, (2.108)

which indicates that this is an imaginary number. On the other hand,

 ∗
. ψ|(ÂB̂ + B̂Â)|ψ = ψ|(ÂB̂ + B̂Â)|ψ, (2.109)

which indicates that this is a real number. Since the square of the absolute value of
a complex number is equal to the sum of those for real and imaginary parts,
 2 1  2 1  2
ψ|ÂB̂|ψ = ψ|[Â, B̂]|ψ + ψ|(ÂB̂ + B̂Â)|ψ
    
.
4 4
1  2
≥ ψ|[Â, B̂]|ψ . (2.110)

4
Combining the above inequality with Eqs. (2.106) and (2.107), we find that the
general uncertainty relationship of Eq. (2.105) is always satisfied. This completes
the proof. ⊔
⊓

 state .|ψ, let us define

For the two physical observables . and .B̂ and the given
uncertainties of the two physical observables as .A = ψ|Â2 |ψ and .B =

ψ|B̂ 2 |ψ. Then, Eq. (2.105) implies that

1  
AB ≥ ψ|[Â, B̂]|ψ , (2.111)

.
2
which is the general uncertainty relationship that holds between any two non-
commuting physical observables. For the case where . = x̂ and .B̂ = p̂, namely
position and momentum operators, .[x̂, p̂] = i h̄. Therefore, .xp ≥ h̄/2, which is
a precise statement of Heisenberg’s uncertainty principle.

Exercise Problems with Solutions

2.1 Two states |α and |β are normalized and orthogonal to each other. Given the
two states |a and |b defined as follows:

. |a = |α + 3i|β,

|b = 2i|α − |β,

calculate a|a, b|b, a|b, and b|a, which are all real or complex numbers.
62 2 Dirac Notation and Principles of Quantum Mechanics

Solution 2.1 The four inner products can be calculated using the fact that α|α =
β|β = 1 and α|β = β|α = 0 as follows:

a|a = (α − 3iβ|) (|α + 3i|β)

.

= 1 + 9 = 10,
b|b = (−2iα| − β|) (2i|α − |β)
= 4 + 1 = 5,
a|b = (α − 3iβ|) (2i|α − |β)
= 2i + 3i = 5i,
b|a = (−2iα| − β|) (|α + 3i|β)
= −2i − 3i = −5i.

2.2 Two states |α and |β are normalized and orthogonal to each other. A state |1
is given by the following linear combination.

1
|1 = √ (|α + |β) .
.
2

Show that the above state is normalized. Find another state |2 given by a different
linear combination of |α and |β, which is orthogonal to |1 and is normalized.
Solution 2.2 First, let us prove that |1 is normalized. Taking inner product of this
with itself, we find that

1
1|1 =
. (α| + β|) (|α + |β)
2
1
= = (1 + 1) = 1.
2
Second, let us assume a state |2 = Cα |α+Cβ |β. For this state to be orthogonal
to |1,

1  
1|2 = √ (α| + β|) Cα |α + Cβ |β
.
2
Cα Cβ 1  
= √ α|α + √ β|β = √ Cα + Cβ = 0.
2 2 2

For the above condition to be satisfied, Cβ = −Cα . Since |2 has to be normalized,

1 = 2|2 = |Cα |2 + |Cβ |2 = 2|Cα |2 .

.
Exercise Problems with Solutions 63

√
The above condition can be satisfied by choosing Cα = 1/ 2. Note that this can
be multiplied by any unit normal complex value but let us √
make the simplest choice
that Cα is real and positive as assumed. Then, Cβ = −1/ 2. As a result, the state
|2 is given by

1
. |2 = √ (|α − |β) .
2

2.3 Two states |α and |β are normalized and orthogonal to each other, in terms of
which two operators  and B̂ are defined as follows:

 = |αα| − |ββ|
.

B̂ = −i|αβ| + i|βα|

Then, express the following two operators, ÂB̂ + B̂ Â and i ÂB̂ − i B̂ Â in terms of
linear combinations of outer products, |αα|, |ββ|, |αβ|, and |βα|. Make sure
that you have the simplest possible forms. State whether each of the two operators
is Hermitian or not, and justify your answers.
Solution 2.3 The products ÂB̂ and B̂ Â can be calculated as follows:

ÂB̂ = (|αα| − |ββ|) (−i|αβ| + i|βα|)

.

= −i|αβ| − i|βα|,
B̂ Â = (−|αβ| + |βα|) (|αα| − i|ββ|)
= i|αβ| + i|βα|,

Summing the above two operators,

. ÂB̂ + B̂ Â = 0,
i ÂB̂ − i B̂ Â = 2 (|αβ| + |βα|) .

Therefore, both ÂB̂ + B̂ Â and i ÂB̂ − i B̂ Â are Hermitian.

2.4 The following operators are not Hermitian. Add additional (minimum number
of) terms and make these operators Hermitian.

. (a) 2|αα| + (3 + 2i)|αβ| (c) |12| + i|23| + |33|

(b) |11|22| + |23|33| (d) |αα| − |βα|

64 2 Dirac Notation and Principles of Quantum Mechanics

Solution 2.4 Corrected Hermitian operators with minimal number of terms added
are provided below.

(a) 2|αα| + (3 + 2i)|αβ| + (3 − 2i)|βα|

.

(b) |11|22| + |22|11| + |23|33| + |33|32|

(c) |12| + |21| + i|23| − i|32| + |33|
(d) |αα| − |βα| − |αβ|

2.5 For the following functions of operators, which are not necessarily Hermitian,
determine the Hermitian conjugate for each function of operators. The answer
should be expressed in terms of given operators and their Hermitian conjugates.

. (a) Ô 2 + Ô (c) Ô1 Ô2 + i(Ô1 + Ô2 )

(b) Ô 2 + i Ô (d) Ô1 Ô2 Ô3

Solution 2.5 Hermitian conjugates are provided below.

 2
(a)
. Ô † + Ô †
 2
(b) Ô † − i Ô †
 
(c) Ô2† Ô1† − i Ô1† + Ô2†

(d) Ô3† Ô2† Ô1†

2.6 Prove the following identities.

(a) [x̂, p̂2 ] = 2i h̄p̂

. (b) [p̂, x̂ 2 ] = −2i h̄x̂

Solution 2.6 The above identities can be proved by employing Eq. (2.57) as
follows.

(a) [x̂, p̂2 ] = p̂[x̂, p̂] + [x̂, p̂]p̂ = i h̄p̂ + i h̄p̂ = 2i h̄p̂
.

(b) [p̂, x̂ 2 ] = x̂[p̂, x̂] + [p̂, x̂]x̂ = −i h̄x̂ − i h̄x̂ = −2i h̄x̂

2.7 It is known that Û (θ ) defined below is unitary for any value of its parameter θ .

Û (θ ) = cos θ − i sin θ Â,

.

where  is a Hermitian operator. Find out the property that  should satisfy for the
above statement to be true.
Exercise Problems with Solutions 65

Solution 2.7 For the given operator to be unitary,

Û (θ )Û † (θ ) = (cos θ − i sin θ Â)(cos θ + i sin θ Â)

.

= cos2 θ − i sin θ cos θ Â + i sin θ cos θ Â + sin2 θ Â2

= cos2 θ + sin2 θ + sin2 θ (Â2 − 1) = 1 + sin2 θ (Â2 − 1) = 1.

For the above identity to be true, Â2 = 1. For this case, it is also easy to show that
Û † (θ )Û (θ ) = 1 as well.
2.8 A quantum particle of unit mass in one dimensional box of unit length is
prepared in the following state at time t = 0.

1 i
|ψ = √ |E1  − √ |E3 ,
.
2 2

where |E1  and |E3  are the first and third eigenstates of the Hamiltonian of the
particle. Answer the following questions.
(a) Find out the expression for x|ψ.
(b) Determine the state |ψ; t at time t = 1.
(c) Calculate the expression for the expectation value for x̂ of the particle for
arbitrary time t.
Solution 2.8 For particle in a box of unit length, the eigenvalue and eigenfunction
of the Hamiltonian are given by

h̄2 2 2
. En = π n ,
2
√
x|En  = 2 sin(nπ x),

where n = 1, 2, . . . . Answers for (a), (b), and (c) are provided below.
(a)

1 i
x|ψ = √ x|E1  − √ x|E3 
.
2 2
√ √
2 2
= √ sin(π x) − i √ sin(3π x)
2 2
= sin(π x) − i sin(3π x)

(b)

1 i
.|ψ; t = √ e−iE1 t/h̄ |E1  − √ e−iE3 t/h̄ |E3 
2 2
66 2 Dirac Notation and Principles of Quantum Mechanics

1 2 i 2
= √ e−i h̄π t/2 |E1  − √ e−i9h̄π t/2 |E3 
2 2

For t = 1, the above state becomes

1 2 i 2
|ψ; t = 1 = √ e−i h̄π /2 |E1  − √ e−i9h̄π /2 |E3 
.
2 2

(c)
 ∞
ψ; t|x̂|ψ; t =
. dxψ; t|xxx|ψ; t
−∞

1 1 2

i 2

= √ ei h̄π t/2 E1 |x + √ ei9h̄π t/2 E3 |x x
dx
0 2 2
 
1 −i h̄π 2 t/2 i −i9h̄π 2 t/2
× √ e x|E1  − √ e x|E3 
2 2
 1 
= dx sin2 (π x)x − 2 sin(4h̄2 π 2 t) sin(π x) sin(3π x)x
0

+ sin2 (3π x)x
 1 x
= dx (1 − cos(2π x)) − x sin(4h̄2 π 2 t)(cos(2π x)
0 2
x  1
− cos(4π x)) + (1 − cos(6π x)) = .
2 2

Problems

2.9 Given two states |α and |β, which are normalized and orthogonal to each
other, answer the following questions.
(a) Given two states |1 = C(|α + i|β) and |2 = C(|α − i|β), prove that these
two states are orthogonal to each other and determine a real positive number C
such that they are normalized.
(b) An operator  is defined as  = |12| + |21|. Find out the expression for Â
in terms of outer products involving |α and |β.
(c) Determine the eigenstates and eigenvalues of  given above.
2.10 Prove the following identity:

[x̂ p̂ + p̂ x̂, x̂], p̂ = 2h̄2 .

 
.
Problems 67

2.11 For arbitrary state |α,

x|(x̂ + p̂)2 |α = Ô(x)ψα (x),

.

where ψα (x) = x|α and Ô(x) is a mathematical operator (or an operator in

Schrödinger’s formulation) that involves functions of x and its derivative. Find out
the expression for Ô(x).
2.12 Prove the identity given by Eq. (2.59).
2.13 Prove the Jacobi identity given by Eq. (2.60).
2.14 Use some of the identities provided in the main text and simplify [p̂2 + x̂ 2 , p̂ x̂]
as much as possible, where x̂ is the position operator and p̂ is the momentum
operator along the same direction.
2.15 A quantum particle of unit mass in one dimensional box of unit length is
prepared in the following state at time t = 0.
√
3 i
.|ψ = |E1 | + |E2 ,
2 2
where |E1  and |E2  are the first and second eigenstates of the Hamiltonian of the
particle. Answer the following questions.
(a) Find out the expression for x|ψ.
(b) Determine the state |ψ; t at time t = 1.
(c) Calculate the expression for the expectation value for x̂ of the particle for
arbitrary time t.
Chapter 3
Harmonic Oscillator and Vibrational
Spectroscopy

We have confidence that the human spirit will always be able to

create ideas fitting somehow to the external objects that are
conveyed to us . . .
— Wolfgang Pauli

Abstract This chapter describes the solution of the time independent Schrödinger
equation of a harmonic oscillator and shows how the requirement of physically
well-behaving eigenfunction leads to the quantization of the harmonic oscillator
energy. The solution and its properties are then used for a quantum mechanical
description of vibrational states of a diatomic molecule that can be represented by a
harmonic oscillator Hamiltonian with reduced mass. Infrared absorption and Raman
spectroscopies are introduced, and their selection rules for the case of diatomic
molecules are explained. A brief account of how the effects going beyond the
harmonic oscillator model for diatomic molecule is provided.

The harmonic oscillator is one of the simplest but the most widely used models
in physics, chemistry, and engineering. An analytical solution of the Schrödinger
equation for the harmonic oscillator is well known and is described in this chapter.
Details of a harmonic oscillator system also clarify important features of quantum
mechanical states and energies distinct from those of classical mechanics. Solutions
of the quantum harmonic oscillator have contributed immensely to understanding
molecular vibrations and a correct quantum mechanical interpretation of spectro-
scopic data.

3.1 Classical Harmonic Oscillator and Hamiltonian

Let us consider a classical harmonic oscillator with mass m, as illustrated in Fig. 3.1,
with .l0 as the equilibrium length of the spring at which there is zero restoring force.
The position of the oscillator is defined as the displacement of the oscillator relative

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 69

S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_3
70 3 Harmonic Oscillator and Vibrational Spectroscopy

Fig. 3.1 Illustration of a

mass m attached to a wall of
infinite mass via an ideal
massless spring with linear
restoring force

to the equilibrium length, namely, .x = l − l0 . The force applied to the particle at x

is then given by

F (x) = −k(l − l0 ) = −kx,

. (3.1)

where k is the spring constant. The Newtonian equation for the mass attached to the
spring is then equal to

d2
m
. x(t) = −kx(t) = −mω2 x(t), (3.2)
dt 2
√
where .ω = k/m. It is straightforward to show that the solution of the above
equation is

p(0)
.x(t) = x(0) cos(ωt) + sin(ωt), (3.3)
mω

where .x(0) and .p(0) are position and momentum at time .t = 0. The classical
momentum of the particle can be calculated by taking the time derivative of the
above expression as follows:

d
p(t) = m
. x(t) = p(0) cos(ωt) − mωx(0) sin(ωt). (3.4)
dt
On the other hand, the Hamiltonian of the harmonic oscillator is

p2 mω2 2
H =
. + x , (3.5)
2m 2
where the second term is the potential energy of the harmonic oscillator at x. It is
straightforward to show that the time dependent position and momentum given by
Eqs. (3.3) and (3.4) conserve the above Hamiltonian.
3.2 Schrödinger Equation 71

Now, let us assume that the particle no longer behaves classically but should
be considered as a quantum mechanical particle. Then, determining the position
and momentum of the particle both simultaneously and precisely has no physical
meaning in quantum mechanics. As yet, the expression for the Hamiltonian,
Eq. (3.5), remains valid except that the position and momentum be replaced with
operators as follows:

p̂2 mω2 2
Ĥ =
. + x̂ . (3.6)
2m 2
Solving the time independent Schrödinger equation for the above Hamiltonian is the
first step for understanding how the quantum harmonic oscillator can be described
and probed. This is also essential for correct quantum mechanical understanding of
molecular vibrations.

3.2 Schrödinger Equation

The time independent Schrödinger equation (in the Dirac notation) for the harmonic
oscillator Hamiltonian, Eq. (3.6), is

p̂2
 
1 2 2
Ĥ |ψ =
. + mω x̂ |ψ = E|ψ. (3.7)
2m 2

Taking the inner product of the above equation with the position state bra .x|, we
obtain
 2 
p̂ 1
.x|Ĥ |ψ = x| + mω2 x̂ 2 |ψ
2m 2
 
h̄2 ∂ 2 mω2 2
= − + x x|ψ = Ex|ψ. (3.8)
2m ∂x 2 2

The second line in the above equation corresponds to the original Schrödinger
equation because .x|ψ = ψ(x). The Schrödinger equation for this wavefunction
can be expressed as

h̄2 d 2 ψ(x) mω2 2

 
. + E− x ψ(x) = 0, (3.9)
2m dx 2 2

where we have moved all the terms to the lefthand side and changed the sign so that
the second derivative becomes positive. Detailed mathematical steps for solving this
differential equation are provided below.
72 3 Harmonic Oscillator and Vibrational Spectroscopy

3.2.1 Solution of Time Independent Schrödinger Equation

As the first step for solving Eq. (3.9), let us introduce a new scaled coordinate .ξ =
√
γ x, where .γ is yet unknown real parameter with a unit of inverse length square,
√
and also define .φ(ξ ) = ψ(x). Then, .dξ = γ dx and

d2 d2
. ψ(x) = γ φ(ξ ). (3.10)
dx 2 dξ 2

Employing the above relationship in Eq. (3.9), one can obtain the following
equation:

h̄2 γ d 2 φ(ξ ) mω2 2

 
. + E − ξ φ(ξ ) = 0. (3.11)
2m dξ 2 2γ

Dividing the above equation with .h̄2 γ /(2m), we obtain

d 2 φ(ξ ) m2 ω 2
 
2mE 2
. + − ξ φ(ξ ) = 0. (3.12)
dξ 2 h̄2 γ h̄2 γ 2

The above equation has two parameters within the parenthesis, one as a constant
and the other as a coefficient of .ξ 2 . Now, the free parameter .γ can be determined to
remove the second one. Namely, by choosing .γ = mω/h̄, the above equation can
be simplified as follows:

d 2 φ(ξ )
. + (κ − ξ 2 )φ(ξ ) = 0, (3.13)
dξ 2

where
2mE 2E
κ=
. = . (3.14)
h̄2 γ h̄ω

While Eq. (3.13) appears to be simple, its conventional solution as a power series
expansion of .φ(ξ ) with respect to .ξ is not straightforward because the resulting
equation involves three different powers of .ξ . A well-known technique to overcome
this difficulty is to factor out a term that dominates for large .ξ limit, which then
helps identifying an equation that results in a simpler relationship involving only
two powers of .ξ .
Inspection of Eq. (3.13) shows that the term .−ξ 2 φ(ξ ) becomes one of dominant
terms in the large .ξ limit. For this term to be cancelled by the second derivative
2
term, .φ(ξ ) should have a term that contains .e−ξ /2 that produces a term with .ξ 2 as
3.2 Schrödinger Equation 73

a multiplicative factor when second derivative is taken. Thus, let us introduce a new
function .H (ξ ) such that
2 /2
φ(ξ ) = e−ξ
. H (ξ ). (3.15)

Then,

d 2 2
. φ(ξ ) = −ξ e−ξ /2 H (ξ ) + e−ξ /2 H ′ (ξ ), . (3.16)
dξ
d2 2 2 2 d 2 d2
2
φ(ξ ) = −e−ξ /2 H (ξ ) + ξ 2 e−ξ /2 H (ξ ) − 2ξ e−ξ /2 H (ξ ) + e−ξ /2 2 H (ξ )
dξ dξ dξ
2 /2 d 2 d2
= (ξ 2 − 1)φ(ξ ) − 2ξ e−ξ H (ξ ) + e−ξ /2 2 H (ξ ). (3.17)
dξ dξ

Inserting the above expression into Eq. (3.13), we obtain

d2
. φ(ξ ) + (κ − ξ 2 )φ(ξ )
dξ 2
2 /2 dH (ξ ) 2 /2
2 d2
= (κ − 1)e−ξ H (ξ ) − 2ξ e−ξ+ e−ξ /2 2 H (ξ )
dξ dξ
 2 
2 d d
= e−ξ /2 2
H (ξ ) − 2ξ H (ξ ) + (κ − 1)H (ξ ) = 0. (3.18)
dξ dξ
2 /2
Since .e−ξ is nonzero, the above equation implies that

d2 d
. H (ξ ) − 2ξ H (ξ ) + (κ − 1)H (ξ ) = 0. (3.19)
dξ 2 dξ

While the above differential equation has an additional term involving the first
derivative, it has the same order as the original function .H (ξ ) because of the
additional multiplicative factor .ξ . Thus, when a power series solution is assumed
for .H (ξ ), a simple recursion relationship involving terms with only two different
powers of .ξ appears.
Indeed, Eq. (3.19) is a well-known differential equation called the Hermite
equation. The solution of this equation and the condition for .κ that allows well-
behaving solution can be found as described below. Let us assume the following
power series expansion for .H (ξ ):

H (ξ ) = C0 + C1 ξ + C2 ξ 2 + · · · + Cn ξ n + Cn+1 ξ n+1 + Cn+2 ξ n+2 + . . .

.

(3.20)
74 3 Harmonic Oscillator and Vibrational Spectroscopy

Taking the first and second derivatives of the above expression, we find that

d
. H (ξ ) = C1 + 2C2 ξ + · · · + nCn ξ n−1
dξ
+(n + 1)Cn+1 ξ n + (n + 2)Cn+2 ξ n+1 + . . . , . (3.21)
d2
H (ξ ) = 2C2 + · · · + n(n − 1)Cn ξ n−2
dξ 2
+(n + 1)nCn+1 ξ n−1 + (n + 2)(n + 1)Cn+2 ξ n + . . . . (3.22)

Inserting Eqs. (3.20)–(3.22) into Eq. (3.19) and requiring that the coefficient of each
term .ξ n is zero lead to the following relation between coefficients:

(n + 2)(n + 1)Cn+2 − 2nCn + (κ − 1)Cn = 0.

. (3.23)

Solving the above equation with respect to .Cn+2 , we obtain the following recursion
relationship:

2n − κ + 1
Cn+2 =
. Cn . (3.24)
(n + 2)(n + 1)

It is easy to find that the above recursion relationship implies that .Cn+2 should be
zero for some finite n in order for the wave function to be physically acceptable.
This can be shown as follows.
For simplicity, let us assume that .C1 = 0. Then, we only need to consider even
terms. For large .n = 2n′ >> (κ − 1)/2 & 1, Eq. (3.24) approximately becomes

4n′
C2(n′ +1) ≈
. C2n′ . (3.25)
2(n′ + 1)2n′

The above relation implies that .C2n′ ≈ C0 /n′ ! for large .n′ . In this case, .H (ξ ) ∼
2
C0 eξ for large .ξ . When this expression is used in Eq. (3.15), the resulting
wavefunction behaves as follows:
2 /2 2 2 /2
φ(ξ ) ∼ C0 e−ξ
. e ξ = C0 e ξ as ξ → ∞. (3.26)

For nonzero .C1 , following a similar reasoning, the function can be shown to contain
2
a term that behaves as .C1 ξ eξ /2 for large .ξ as well.
The asymptotic behavior of Eq. (3.26) shows that .φ(ξ ) is not an acceptable
solution, if the series for .H (ξ ) continue up to an infinite order, because it means
that the probability to find the position to be at infinity is infinitely large, which is
unphysical. The only way out of this is for the series Eq. (3.20) to terminate at finite
2
order. Then, .H (ξ ) becomes a polynomial and .φ(ξ ) = H (ξ )e−ξ /2 becomes zero for
3.2 Schrödinger Equation 75

ξ → ∞. Inspection of Eq. (3.24) shows that the termination of the series is possible
.

if the following condition is satisfied.

κ = 2v + 1, v = 0, 1, . . .
. (3.27)

Employing the definition of .κ given by Eq. (3.14), we find that the above condition
implies that the energy should have the following form:
 
1
.Ev = h̄ω v + , v = 0, 1, . . . (3.28)
2
The corresponding solution for each v, .Hv (ξ ), is called a Hermite polynomial,
which is a polynomial of .ξ with maximum power of v.
Since the recursion relationship, Eq. (3.24), involves terms with two order
difference in .ξ , solutions should consist of terms with even or odd powers of .ξ
depending on the value of v. That is, for even v, solution for Eq. (3.24) exists
only for .C1 = 0. For odd v, the solution exists only for .C0 = 0. Examples of
these polynomials, known as Hermite polynomials, and a few terms for general
expressions are provided in Table 3.1.
On the other hand, it is also known that the above Hermite polynomials can be
generated from a Gaussian-type function as follows:
∞
−s 2 +2sξ
 1
e
. = Hv (ξ )s v , (3.29)
v!
v=0

where the left hand side is called the generating function of the Hermite polynomial.
Summing up the results obtained so far, the eigenfunction for each energy .Ev
can be expressed as
2 /2 √
ψv (x) = Nv e−γ x
. Hv ( γ x), (3.30)

Table 3.1 Examples of Hermite polynomials and general expressions. Note that the second and
third terms are shown assuming that .m ≥ 4. For .m = 1, only the first and the last terms exist. For
.m = 2, only the first, second, and last terms exist

n .Hn (ξ )

0 1
1 .2ξ

2 .4ξ −22

3 .8ξ
3
− 12ξ
4 4 2
.16ξ − 48ξ + 12

5 5 3
.32ξ − 160ξ + 120ξ

6 6 4 2
.64ξ − 480ξ + 720ξ − 120
(2m)! (2m)!
2m .(2ξ )
2m − (2m−2)!2! (2ξ )
2m−2 + (2m−4)!4! (2ξ )
2m−4 + · · · + (−1)m (2m)!
m!

(2m+1)! (2m+1)!
.2m +1 .(2ξ )
2m+1 − (2m−1)!2! (2ξ )
2m−1 + (2m−3)!4! (2ξ )
2m−3 + · · · + (−1)m (2m+1)!
m! (2ξ )
76 3 Harmonic Oscillator and Vibrational Spectroscopy

Fig. 3.2 Four eigenfunctions .ψv (x)’s with .v = 0, 1, 2, and 3 and corresponding probability
densities, .|ψv (x)|2 , for a harmonic oscillator

where .Nv is the normalization constant and can be determined by the following
normalization condition:
∞ ∞ 2 √
. ψv (x)2 dx = Nv2 e−γ x Hv ( γ x)2 dx = 1. (3.31)
−∞ −∞

The integration over the Hermite polynomial in the above equation can be done
explicitly, which leads to the following expression:

1 γ 1/4
Nv = √
. , (3.32)
2v v! π

where note that .0! = 1. Figure 3.2 shows .ψv (x) for .v = 0, · · · , 3 and the
corresponding probability densities.
There are three properties of Hermite polynomials that are important for
harmonic oscillators as described below.

Properties of Hermite Polynomials

• Orthogonality and normalization condition

∞ √
−ξ 2 γ
. dξ Hv (ξ )H (ξ )e
v′ =δ vv ′ , (3.33)
−∞ Nv2

(continued)
3.2 Schrödinger Equation 77

where .Nv is the normalization constant defined by Eq. (3.32). Also, note
√
that .ξ = γ x. For .v = v ′ , the above equation amounts to the orthogonality
condition for the eigenfunctions of the harmonic oscillator.
• Recursion relationship involving three consecutive Hermite polynomials:

.Hv+1 (ξ ) − 2ξ Hv (ξ ) + 2vHv−1 (ξ ) = 0. (3.34)

The above relationship allows generating all the higher order Hermite
polynomials from .H0 (ξ ) and .H1 (ξ ).
• Relationship involving the derivative of a Hermite polynomial:

d
. Hv (ξ ) = 2vHv−1 (ξ ). (3.35)
dξ

The combination of Eqs. (3.33) and (3.34) for .v ′ = v ± 1 can be used to show
that
∞
x̂v =
. dxψv (x)xψv (x) = 0. (3.36)
−∞

The combination of Eqs. (3.33)–(3.35) can be used to show that

h̄ ∞ ∂
p̂v =
. dxψv (x) ψv (x) = 0. (3.37)
i −∞ ∂x

The recursion relationship, Eq. (3.34), can also be used for evaluating the following
expectation value of .x̂ 2 with respect to the vth eigenstate.
∞ 1 1 ∞ 2
x̂ 2 v =
. dxψv∗ (x)x 2 ψv (x) = Nv2 √ dξ Hv2 (ξ )ξ 2 e−ξ . (3.38)
−∞ γ γ −∞

In the above expression, Eq. (3.34) can be used twice as follows:

 
1
. ξ 2 Hv (ξ ) = ξ Hv+1 (ξ ) + vHv−1 (ξ )
2
1 1 v
= Hv+2 (ξ ) + (v + 1)Hv (ξ ) + Hv (ξ ) + v(v − 1)Hv−2 (ξ ). (3.39)
4 2 2
When this expression is inserted into Eq. (3.38), the integrations involving products
of .Hv+2 (ξ ) and .Hv−2 (ξ ) with .Hv (ξ ) vanish due to the orthogonality condition, and
only the term involving .Hv2 (ξ ) survives. Thus,
78 3 Harmonic Oscillator and Vibrational Spectroscopy

1 (2v + 1) ∞ 2
x̂ 2 v = Nv2
. dξ Hv2 (ξ )e−ξ
γ 3/2 2 −∞
√
1 (2v + 1) γ (2v + 1)
= Nv2 =
γ 3/2 2 Nv2 2γ
 
h̄ 1
= v+ . (3.40)
mω 2

Since
 
1 1 1
Ĥ v =
. p̂ 2 v + mω2 x̂ 2 v = h̄ω v + , (3.41)
2m 2 2

Eq. (3.40) implies that

   
h̄ω 1 1
p̂2 v = 2m
. v+ = mh̄ω v + . (3.42)
2 2 2

From Eqs. (3.36), (3.37), (3.40), and (3.42), it is easy to show that the position
and the momentum have the following magnitudes of uncertainties for the ground
vibrational state:

1/2 1/2 h̄
. σx = x̂ 2 0 − x̂20 = x̂ 2 0 = ,. (3.43)
2mω
1/2 1/2 mh̄ω
σp = p̂2 0 − p̂20 = p̂2 0 = . (3.44)
2

Thus,

h̄
σx σp =
. . (3.45)
2
This means that the ground vibrational eigenstate is the minimum uncertainty state.
The recursion relationship, Eq. (3.34), can also be used for calculating the
following quantity:
∞ 1 ∞ 2
. dx ψv (x)xψv ′ (x) = Nv Nv ′ dξ Hv (ξ )ξ Hv ′ (ξ )e−ξ
−∞ γ −∞
1 ∞

1

2
= Nv Nv ′ dξ vHv−1 (ξ ) + Hv+1 (ξ ) Hv ′ (ξ )e−ξ
γ −∞ 2
 √ √ 
1 γ 1 γ
= Nv Nv ′ vδv−1v ′ ′2 + δv+1v ′ ′2 . (3.46)
γ Nv 2 Nv
3.2 Schrödinger Equation 79

Note that Eq. (3.36) is the special case of the above identity for .v = v ′ . The above
identity implies that the measurement of .x̂ causes the transition of vibrational state
from v to .v ± 1. This fact has important implications for the selection rule of the
spectroscopy detecting molecular vibrational states.

3.2.2 Operator Approach

The commutator between the Hamiltonian, Eq. (3.6), and the position operator is
given by

1 2 1  i h̄
[Ĥ , x̂] =
. [p̂ , x̂] = p̂[p̂, x] + [p̂, x̂]p̂ = − p̂. (3.47)
2m 2m m
Similarly,

mω2 2 mω2
x̂[x̂, p̂] + [x̂, p̂]x̂ = i h̄mω2 x̂.

[Ĥ , p̂] =
. [x̂ , p̂] = (3.48)
2 2
Combining Eqs. (3.47) and (3.48), we find that

. [Ĥ , mωx̂ + i p̂] = −i h̄ωp̂ − h̄mω2 x̂ = −h̄ω(mωx̂ + i p̂), . (3.49)

[Ĥ , mωx̂ − i p̂] = −i h̄ωp̂ + h̄mω2 x̂ = h̄ω(mωx̂ − i p̂). (3.50)

The above identities motivate definition of two important operators as described

below.
Definition 3.1 For the harmonic oscillator with mass m and angular frequency .ω,
its lowering and raising operators .b̂ and .b̂† are defined as follows:

1 mω p̂
. b̂ = √ (mωx̂ + i p̂) = x̂ + i √ ,. (3.51)
2mh̄ω 2h̄ 2mh̄ω
1 mω p̂
b̂† = √ (mωx̂ − i p̂) = x̂ − i √ . (3.52)
2mh̄ω 2h̄ 2mh̄ω

The lowering and raising operators defined above have the following important
properties.
80 3 Harmonic Oscillator and Vibrational Spectroscopy

Properties of Lowering and Raising Operators

• Raising and lowering operators satisfy the following commutation rela-
tions:

. [Ĥ , b̂] = −h̄ωb̂, . (3.53)

† †
[Ĥ , b̂ ] = h̄ωb̂ , . (3.54)
[b̂, b̂† ] = 1. (3.55)

• The product of .b̂† and .b̂ can be shown to be

  
mω p̂ mω p̂
b̂† b̂ =
. x̂ − i √ x̂ + i √ ,
2h̄ 2mh̄ω 2h̄ 2mh̄ω
mω 2 i i p̂2 1 1
= x̂ + x̂ p̂ − p̂ x̂ + = Ĥ − . (3.56)
2h̄ 2h̄ 2h̄ 2mh̄ω h̄ω 2

Therefore,
 
1
.Ĥ = h̄ω b̂† b̂ + . (3.57)
2

Equation (3.57) is equivalent to Eq. (3.6), and offers an efficient way to obtain
some of the key properties of the quantum harmonic oscillator. As will be shown
below, this expression allows to deduce the details of how the eigenvalues are
quantized even without solving the Schrödinger equation.
For now, let us assume that we have no knowledge of the solution of the
Schrödinger equation in the previous subsection. Thus, let us suppose an eigenstate
of the Hamiltonian .|E such that

Ĥ |E = E|E,
. (3.58)

where E is as yet unknown value. Then, one can show that .b̂|E and .b̂† |E are
eigenstates of .Ĥ with different eigenvalues using the commutation relations. For
example,

[Ĥ , b̂]|E = Ĥ (b̂|E) − b̂(Ĥ |E) = −h̄ωb̂|E,

. (3.59)

where Eq. (3.53) has been used. Using the assumption of Eq. (3.58) and sending
b̂Ĥ |E = E b̂|E to the righthand side, the above equation implies that
.

Ĥ (b̂|E) = (E − h̄ω)b̂|E.
. (3.60)
3.2 Schrödinger Equation 81

Thus .b̂|E is an eigenstate of .Ĥ but with an eigenvalue .E − h̄ω, which is lower than
the original value by .h̄ω. Similarly, one can show that

Ĥ (b̂† |E) = (E + h̄ω)b̂† |E.

. (3.61)

Thus, .b̂† is an eigenstate of .Ĥ but with a different eigenvalue .E + h̄ω, which is
higher than the original value by .h̄ω.
The Hamiltonian of the harmonic oscillator is positive definite. Therefore,

E|Ĥ |E = E ≥ 0.
. (3.62)

Let us denote the lowest energy state as .|E0 . Then, it is clear that .b̂|E0  = 0. If
this is not the case, .b̂|E0  becomes an eigenstate of .Ĥ with eigenvalue of .E0 − h̄ω,
which contradicts the assumption that .E0 is the lowest possible value. This in turn
implies that

1 h̄ω
Ĥ |E0  = h̄ω(b̂† b̂ + )|E0  =
. |E0 . (3.63)
2 2
Therefore,

h̄ω
E0 =
. . (3.64)
2
Now, let us denote .|E0  as .|0 and assume that it is normalized. Then, we can
generate eigenstates with eigenvalues .h̄ω(v + 1/2) by applying .b̂† on .|0 by v times
repeatedly. Let us denote the normalized eigenstates generated in such a way as .|v.
In other words,

b̂† |v = Av |v + 1, v = 0, 1, 2, · · · ,

. (3.65)

where we have introduced a constant factor .Av because .b̂† does not conserve
normalization. This can be shown by taking inner product of .b̂† |v with itself as
follows:
 
Ĥ 1
. v|b̂ b̂† |v = v| + |v = v + 1 = |Av |2 , (3.66)
h̄ω 2

where Eqs. (3.55) and (3.56) were used in the first equality, and the fact that .Ĥ |v =
h̄ω(v + 1/2) was used in the second equality. The last equality is from Eq. (3.65)
along with the fact that .|v is normalized. Therefore, we find that
√
b̂† |v =
. v + 1|v + 1. (3.67)
82 3 Harmonic Oscillator and Vibrational Spectroscopy

This implies that the vth eigenstate can be defined as

1
|v = √ (b̂† )v |0,
. (3.68)
v!
where .|v is normalized and
1
Ĥ |v = h̄ω(v + )|v.
. (3.69)
2
Note that the above equation is nothing but the time independent Schrödinger
equation for the eigenvalue, Eq. (3.28). It is remarkable that this result can be
obtained purely out of the commutator relationship between .Ĥ , .b̂, and .b̂† and that
there has to be a lower bound in the eigenvalue.
In a similar manner to obtaining Eq. (3.67), we can obtain the following identity
involving the operation of .b̂.
√
. b̂|v = v|v − 1, v = 1, 2, 3, . . . . (3.70)

This identity can be used to confirm that the eigenvalues .h̄ω(v + 1/2)’s are the only
eigenvalues possible for the harmonic oscillator.

3.2.3 General Time Dependent State

A general (normalized) quantum mechanical state for a harmonic oscillator can be

expressed as

.|ψ = Cv |v, (3.71)
v

where . v |Cv |2 = 1. The corresponding wavefunction (position representation) is


given by
 
.ψ(x) = x|ψ = Cv x|v = Cv ψv (x). (3.72)
v v

The normalization condition for .|v is as follows:

∞
. v|v ′  = v|1̂|v ′  = dxv|xx|v ′ 
−∞
∞
1 1 γ 1/2 2 √ √
=√ √ dx e−γ x Hv ( γ x)Hv ′ ( γ x)
2v v! v′′
2 v! π −∞
= δvv ′ , (3.73)

which can be proved employing the property of the Hermite polynomial.

3.3 Vibrational Spectroscopy of Diatomic Molecules 83

Let us assume that the quantum mechanical state at .t = 0 is given by


|ψ; t = 0 =
. Cv |v. (3.74)
v

Then, the state at time t is given by

 
|ψ; t =
. Cv e−iEv t/h̄ |v = Cv e−iω(v+1/2)t |v. (3.75)
v v

Taking time derivative of the above expression, we find that the state indeed satisfies
the time dependent Schrödinger equation as shown below.
 
∂  1 −iω(v+ 1 )t
i h̄
. |ψ; t = Cv h̄ω v + e 2 |v
∂t v
2
 1
= Cv Ĥ e−iω(v+ 2 )t |v
v

= Ĥ |ψ; t. (3.76)

3.3 Vibrational Spectroscopy of Diatomic Molecules

Let us consider a diatomic molecule consisting of two atomic masses .m1 and
m2 . At the simplest level, their bonding can be modeled by an ideal spring with
.

spring constant k. If we denote the two coordinates along the internuclear axis as .l1
and .l2 (see Fig. 3.3) and assume that their dynamics are governed by the classical
mechanics, we arrive at the following equations of motion:

d2 k
. l1 = − (r0 − l2 + l1 ), . (3.77)
dt 2 m1
d2 k
l2 = − (l2 − l1 − r0 ). (3.78)
dt 2 m2

In above equations, .r0 is the equilibrium bond distance where there is no restoring
force of the spring. From the difference of the above two equations, one can obtain
the following equation of motion for .l = l2 − l1 :

d2 m 1 + m2
. l = −k (l − r0 ). (3.79)
dt 2 m1 m2
84 3 Harmonic Oscillator and Vibrational Spectroscopy

Fig. 3.3 Illustration of two

masses .m1 and .m2 connected
by an ideal massless spring
with linear restoring force

Let us introduce the displacement from the equilibrium bond distance, .x = l − r0 ,

and a reduced mass .μ = m1 m2 /(m1 + m2 ). Then, the above equation simplifies to

d2 k
.
2
x = − x = −ω2 x, (3.80)
dt μ
√
where .ω = k/μ. Thus, within the classical mechanics, the dynamics of x is
equivalent to that of a harmonic oscillator with an effective mass .μ, for which the
classical Hamiltonian is given by

p2 μω2 2
H =
. + x . (3.81)
2μ 2

If the above assumption of the classical mechanics were true, it means that the
displacement .x(t) and its momentum .p(t) = μdx(t)/dt should go through the
following harmonic time evolution

p(0)
. x(t) = x(0) cos(ωt) + sin(ωt), . (3.82)
μω
p(t) = p(0) cos(ωt) − μωx(0) sin(ωt), (3.83)

for which the Hamiltonian .H = p(0)2 /(2μ) + μω2 x(0)2 /2 remains conserved and
can have an arbitrary value depending only on the initial value of the displacement
and the momentum.
Quantum mechanically, one has to consider the following Hamiltonian operator
corresponding to Eq. (3.81):

p̂ 2 1
Ĥ =
. + μω2 x̂ 2 , (3.84)
2μ 2
3.3 Vibrational Spectroscopy of Diatomic Molecules 85

where .x̂ and .p̂ are now operators that should obey the commutator relationship
[x̂, p̂] = i h̄. The solution for the Schrödinger equation for the above Hamiltonian
.

is the same as the one presented in Sect. 3.2 except that

μω
γ =
. , (3.85)
h̄

for the present case. Thus, the expressions for the eigenvalue, Eq. (3.28), and the
eigenfunction, Eq. (3.30), remain the same.
One of the important consequences is that the minimum vibrational
√ energy of
the diatomic molecule is no longer zero but becomes .h̄ω/2 = h̄ k/μ/2. This also
means that the internuclear distance and the corresponding momentum can never
become zero simultaneously. For the ground vibrational state, the uncertainties of
relative position and momentum of the two atoms within the diatomic molecules
can be calculated to be

h̄
. σx = ,. (3.86)
2μω

μh̄ω
σp = . (3.87)
2

These satisfy the minimum possible value of the Heisenberg’s uncertainty relation-
ship, .σx σp = h̄/2.

3.3.1 Vibrational Absorption or Infrared (IR) Spectroscopy

The quantum mechanical behavior of interatomic displacement in a diatomic

molecule can be confirmed from its interaction with light, viewed as electromagnetic
radiation. Detailed mathematical treatment of the interaction will be provided later.
For now, it is sufficient to accept that the lowest order interaction of light with the
displacement is through the dipole moment of the molecule, which is in general
defined as

.D = qi ri , (3.88)
i

where .ri is the coordinate of the ith particle constituting the molecule. For the
case of vibrational spectroscopy, each .ri can be viewed as the coordinate of each
nuclei and .qi the effective point charge of each nucleus, with the effects of electrons
incorporated. For the case of a diatomic molecule with their internuclear coordinate
.x + r0 , the dipole moment operator along the internuclear axis is given by

. D̂ = q(x̂ + r0 ), (3.89)
86 3 Harmonic Oscillator and Vibrational Spectroscopy

Fig. 3.4 Diatomic molecule

and their effective charge and
internuclear distance. The
arrow represents the dipole
moment

where .x̂ is now viewed as quantum mechanical position operator and q is the
magnitude of the effective charge (see Fig. 3.4).
The simplest vibrational spectroscopy is the one where molecular vibrational
states changes through the lowest order interaction with light. This is conventionally
known as Infrared (IR) spectroscopy because it turns out that the molecular
vibrational frequencies are typically in the range of IR. There are two important
selection rules as described below.

Selection Rules for IR Spectroscopy

• The first is called gross selection rule stated as follows: Only vibrational
modes that cause change of dipole moment are IR active. For diatomic
molecules, this means that the molecular has to be polar. Therefore, only
heteroatomic diatomic molecules are IR active.
• For those molecules that are IR active, the transition that is allowed (given
that the harmonic oscillator form of the Hamiltonian is exact) is the
following transition:

. v = ±1. (3.90)

This can be understood from that the IR spectroscopy amounts to applying

the dipole operator given by Eq. (3.89) to a given eigenstate, which then
makes transition to another eigenstate. Due to the identity of Eq. (3.46),
such measurement causes transitions to only nearby eigenstates satisfying
Eq. (3.90).

IR spectroscopy is typically an absorption spectroscopy where a photon is

absorbed to excite the vibrational state. Thus, within the assumption of the harmonic
oscillator model, the energy of the photon being absorbed in the IR spectroscopy is
given by
   
1 1
hνph
. = h̄ω v + 1 + − h̄ω v + = h̄ω, (3.91)
2 2

where .νph is the frequency of the photon being absorbed and .ω is the vibrational
angular frequency of the molecule that appears in the Hamiltonian, Eq. (3.84).
3.3 Vibrational Spectroscopy of Diatomic Molecules 87

The IR spectroscopy is most commonly described in terms of wavenumber,

ν̃ph = νph /c, where c is the speed of light in vacuum. Equation (3.91) can then
.

be expressed in terms of the wavenumber as follows:

 1/2
h̄ 1 1 k
. ν̃ph = ω= ω= . (3.92)
2π ch̄ 2π c 2π c μ

For the description of the IR spectroscopy, it is often convenient to introduce the

following vibrational term:
 
1
G(v) = ν̃vib v +
. , (3.93)
2

where
 1/2
1 k
. ν̃vib = . (3.94)
2π c μ

Then, the condition for the IR absorption, Eq. (3.92), reduces to

. ν̃ph = ν̃vib . (3.95)

3.3.2 Vibrational Raman Spectroscopy

The next order of interaction between light and molecule is through the polarizabil-
ity .α of the molecule, which is defined by

Dind = αE.
. (3.96)

This interaction term is quadratic with respect to the amplitude of the field because
it involves interaction of the above induced dipole with the electromagnetic field
again.
The polarizability .α is a matrix in general (second rank tensor) but can be repre-
sented by a number for its portion that depends on the vibrational displacement of
the diatomic molecule. This can again be expanded with respect to the displacement
x as follows:

∂α 
.α̂ = α(0) + x̂ + ··· . (3.97)
∂x x=0

This expression shows that the first order interaction term of the molecule with
radiation is proportional to the vibrational displacement. This fact leads to the
following selection rules for the vibrational Raman spectroscopy.
88 3 Harmonic Oscillator and Vibrational Spectroscopy

Selection Rules for Vibrational Raman Spectroscopy

• Any vibrational mode that causes change of polarizability is Raman active.
This means that even homonuclear diatomic molecule is in general Raman
active because the distance dependence of polarizability is nonzero.
• As in the case of IR spectroscopy, the following selection rule applies to
vibrational Raman.

. v = ±1. (3.98)

Unlike the case of IR, Raman Spectroscopy is a scattering process where

the frequency difference between in-coming radiation and out-going radiation (in
directions perpendicular to in-coming light) is measured. The case where the
outgoing light has lower energy corresponds to . v = 1 and is called Stokes
vibrational transition, which is similar to classical inelastic scattering processes
that involve loss of energy. The case where the outgoing light has higher energy
corresponds to the case of . v = −1 and is called anti-Stokes vibrational
transition, a counter-intuitive process that light extracts energy out of molecules.

3.3.3 Anharmonic Effects

For actual molecules, the potential energy in general deviates from the quadratic
dependence as shown in Fig. 3.5. Note that the actual potential energy becomes
stiffer (softer) than a quadratic function as the distance becomes smaller (larger).
The resulting deviation of the properties of the vibrational motion from those of a
harmonic oscillator is called anharmonicity. In general, the potential energy between
a diatomic molecule can be expanded as

mω2 2 V (3) (0) 3 V (4) (0) 4

V (x) = V (0) +
. x + x + x + ··· , (3.99)
2 6 24

where .V (n) (0) = ∂ n V (x)/∂x n |x=0 . Solving the Schrödinger equation exactly for
the general expansion of the potential energy as given above is a difficult task with
the exception of few models such as the Morse potential. For the calculation of
general anharmonic contributions, one has to rely on approximation methods that
you will learn in later chapters of this book, which are valid if the contribution of
anharmonic terms are small compared to the harmonic terms.
In actual analyses of spectroscopic data, the following empirical expressions have
been frequently used.

1 2 1 3
     
1
.Ev = h̄ωe v+ − xe h̄ωe v + + ye h̄ωe v + + ··· , (3.100)
2 2 2
3.3 Vibrational Spectroscopy of Diatomic Molecules 89

Fig. 3.5 Illustration of a typical diatomic potential energy curve and its parabolic approximation.
The quantum energy levels for actual potential and the harmonic oscillator levels for the parabolic
approximation are shown. The potential energy depth .De and the dissociation energy .D0 for actual
potential energy are shown as well

where .xe and .ye are called anharmonicity coefficients and are much smaller than
one. Or, in terms of vibrational term symbols,
   2  3
1 1 1
Gv = ν̃e
. v+ − xe ν̃e v+ + ye ν̃e v+ + ··· . (3.101)
2 2 2

Two major effects of anharmonicity are (1) non-uniform level spacing and
(2) finite dissociation energy (see Fig. 3.5). In general, the potential well-depth
is denoted as .De and the dissociation energy is denoted as .D0 . The difference
between these two is the zero point energy of the vibrational motion. According
to Eq. (3.100), the expression for this is as follows:

1 1 1
D e − D0 =
. h̄ωe − xe h̄ωe + ye h̄ωe + · · · , (3.102)
2 4 8
In terms of the wavenumber, the above quantity can be expressed as

1 1 1
. D̃e − D̃0 = ν̃e − xe ν̃e + ye ν̃e + · · · . (3.103)
2 4 8
The above expression shows that the anharmonic effect in general reduces the zero
point energy.
90 3 Harmonic Oscillator and Vibrational Spectroscopy

3.4 Summary and Questions

The time independent Schrödinger equation for a harmonic oscillator has exact
solutions. Eigenvalues of the Hamiltonian are given by Eq. (3.28), which are
quantized with an equal spacing. The corresponding eigenfunctions are given by
Eq. (3.30), which are given by Hermite polynomials (listed in Table 3.1) multiplied
with a common Gaussian function. The quantization of the eigenvalue arises from
the requirement that the eigenfunctions remain finite and that its squares are
integrable over the entire value of the position. The ground vibrational state of the
harmonic oscillator is a minimum uncertainty state that satisfies Eq. (3.45).
By introducing .b̂ and .b̂† , lowering and raising operators that are defined by
Eqs. (3.51) and (3.52), one can express the Hamiltonian of the harmonic oscillator
in the form of Eq. (3.57). The raising operator, when applied to an eigenstate of
the Hamiltonian, produces the next higher eigenstate according to Eq. (3.67). The
lowering operator, when applied to an eigenstate of the Hamiltonian, produces
the next lower eigenstate according to Eq. (3.70). Application of the lowering
operator to the ground vibrational state results in a null state, which means that
the state does not exist any more. It is possible to make a complete description of a
quantum harmonic oscillator, including the dynamics, solely in terms of expressions
involving .b̂ and .b̂† .
The vibrational motion of a diatomic molecule, given that the interatomic
potential energy is well approximated by a quadratic function of the relative
displacement, can be represented by a harmonic oscillator with a reduce mass .μ as
defined below Eq. (3.79). Thus, the eigenfunction and eigenvalue of the harmonic
oscillator with the reduced mass can be used for quantum mechanical understanding
of the molecular vibration and selection rules for IR and Raman spectroscopies as
given by Eqs. (3.90) and (3.98). For diatomic molecules with soft potential energy
and/or in high vibrational states, the harmonic oscillator assumption breaks down.
The resulting anharmonic effects can be effectively represented by the modification
of the expression for the vibration energy as in Eq. (3.100), although this expression
is empirical.

Questions
• In solving the time independent Schrödinger equation for a harmonic
oscillator, which requirement for the wavefunction causes eigenvalues of
the Hamiltonian to be quantized?
• The ground vibrational state of a harmonic oscillator is a minimum
uncertainty state. Is it the only minimum uncertainty state that can be found
for the harmonic oscillator?
(continued)
Exercise Problems with Solutions 91

• Time dependent expectation values for position and momentum operators

of a harmonic oscillator appear to satisfy the classical Newtonian equation
of motion. What is the reason for this apparent classical behavior?
• How are the expectation values of the kinetic and potential energies of a
quantum harmonic oscillator related with each other?
• Are the raising (.b̂† ) and lowering (.b̂) operators of a quantum harmonic
oscillator physical observables?
• What makes a diatomic molecule IR active and what is the transition being
involved in an IR absorption?
• What makes a diatomic molecule Raman active and what transitions are
involved in Stokes and anti-Stokes Raman transitions respectively?
• For a given diatomic molecule, how are the dissociation energy, the depth
of the potential energy, and the zero point energy of the vibration related?
• If anharmonic effects become significant for the vibration of a diatomic
molecule, do the selection rules for IR and Raman transitions, . v = ±1,
still remain the same?

Exercise Problems with Solutions

3.1 Prove that the time dependent position and momentum of the classical har-
monic oscillator given by Eqs. (3.3) and (3.4) conserve the classical Hamiltonian,
Eq. (3.5).
Solution 3.1 Hamiltonian at time t is shown to be independent of t and the same
as that for t = 0 as follows.

p(t)2 1 1 2
. + mω2 x(t)2 = p(0) cos(ωt) − mωx(0) sin(ωt)
2m 2 2m
1 p(0) 2
+ mω2 x(0) cos(ωt) + sin(ωt)
2 mω
p(0)2
= cos2 (ωt) + sin2 (ωt)
2m
1
+ mω2 x(0)2 sin2 (ωt) + cos2 (ωt)
2
p(0)2 1
= + mω2 x(0)2 .
2m 2
92 3 Harmonic Oscillator and Vibrational Spectroscopy

3.2 Confirm that the recursion relationship, Eq. (3.34), holds true for v = 1, 2 using
the expressions for the Hermite polynomial provided in Table 3.1.
Solution 3.2 For v = 1,

H2 (ξ ) − 2ξ H1 (ξ ) + 2H0 (ξ ) = 4ξ 2 − 2 − 2ξ · (2ξ ) + 2.
.

= 4ξ 2 − 2 − 4ξ 2 + 2 = 0.

For v = 2,

H3 (ξ ) − 2ξ H2 (ξ ) + 4H1 (ξ ) = 8ξ 3 − 12ξ − 2ξ(4ξ 2 − 2) + 4 · 2ξ

.

= 8ξ 3 − 12ξ − 8ξ 3 + 4ξ + 8ξ = 0.

3.3 Use the recursion relationship, Eq. (3.34), to find out the expression for H4 (ξ )
and H5 (ξ ) employing the expressions for H2 (ξ ) and H3 (ξ ) provided in Table 3.1.
Solution 3.3 For v = 4,

H4 (ξ ) = 2ξ H3 (ξ ) − 6H2 (ξ )
.

= 2ξ(8ξ 3 − 12ξ ) − 6(4ξ 2 − 2)

= 16ξ 4 − 24ξ 2 − 24ξ 2 + 12
= 16ξ 4 − 48ξ 2 + 12.

For v = 5,

.H5 (ξ ) = 2ξ H4 (ξ ) − 8H3 (ξ )
= 2ξ(16ξ 4 − 48ξ 2 + 12) − 8(8ξ 3 − 12ξ )
= 32ξ 5 − 96ξ 3 + 24ξ − 64ξ 3 + 96ξ
= 32ξ 5 − 160ξ 3 + 120ξ.

3.4 Prove that the expectation values of position x̂ and momentum p̂ for eigenstates
of the harmonic oscillator are zero.
Solution 3.4 For the position,
∞
x̂ =
. dxψv (x)xψv (x)
−∞
∞ 2 √ √
= Nv2 dxe−γ x Hv ( γ x)xHv ( γ x)
−∞

Nv2 ∞ 2
= dξ e−ξ Hv (ξ )ξ Hv (ξ )
γ −∞
Exercise Problems with Solutions 93

N2 ∞ 
1

−ξ 2
= v dξ e Hv (ξ ) Hv+1 (ξ ) + vHv−1 (ξ )
γ −∞ 2
Nv2 ∞ 2 v 2 ∞ 2
= e−ξ Hv (ξ )Hv+1 (ξ ) + N dξ e−ξ Hv (ξ )Hv−1 (ξ ) = 0,
2 −∞ γ v −∞

where the last equality is due to the orthogonality of different eigenfunctions of the
harmonic oscillator. For the momentum,
∞ h̄ ∂
p̂ =
. dxψv (x) ψv (x)
−∞ i ∂x
∞ 2 /2 √ h̄ ∂ −γ x 2 /2 √
= Nv2 dxe−γ x Hv ( γ x) e Hv ( γ x)
−∞ i ∂x
h̄ ∞ 2 √  2 √
= Nv2 dxe−γ x /2 Hv ( γ x) −γ xe−γ x /2 Hv ( γ x)
i −∞
√ 2 √ 
+ γ e−γ x /2 Hv′ ( γ x) .

2
Combining the two e−γ x /2 terms together and noting the fact that the first term in
the above integral becomes zero because x̂ = 0 as shown above, we find that

h̄ 2 √ ∞ 2 √ √
p̂ =
. N γ e−γ x Hv ( γ x)Hv ′ ( γ x)
i v −∞
h̄ 2 ∞ 2
= N dξ e−ξ Hv (ξ )Hv′ (ξ )
i v −∞
h̄ ∞ 2
= 2vNv2 dξ e−ξ Hv (ξ )Hv−1 (ξ ) = 0,
i −∞

where the identity involving the derivative of Hermite polynomial, Eq. (3.35) and
the orthogonality of the eigenstates of the vibrational Hamiltonian were used.
3.5 For the lowest two eigenstates of a harmonic oscillator |0 and |1, for which
mω = h̄, calculate 0|x̂|1 by explicitly evaluating the integration involving the
expressions for the corresponding wavefunctions.
Solution 3.5 Expressing |0 and |1 in the position space,
∞
0|x̂|1 =
. ψ0 (x)xψ1 (x)
−∞
∞ 2 /2 √ 2 √
= N0 N1 dxe−γ x H0 ( γ x)xe−γ x /2 H1 ( γ x)
−∞
94 3 Harmonic Oscillator and Vibrational Spectroscopy

∞ 2 √
= N0 N1 dxe−γ x x · (2 γ x)
−∞

2γ 2 ∞ 2 1 1
= dxe−γ x x 2 = √ =√ ,
π −∞ 2γ 2

where the fact that γ = mω/h̄ = 1 has been used.

3.6 Find out expressions for the position and momentum operators as linear
combinations of raising and lowering operators b̂† and b̂.
Solution 3.6 Summing Eqs. (3.51) and (3.52), we find that

2mω
b̂ + b̂† =
. x̂.
h̄

Therefore,

h
x̂ =
. b̂ + b̂† .
2mω

On the other hand, subtracting Eq. (3.52) from Eq. (3.51), we find that

2
b̂ − b̂† = i
. p̂.
mh̄ω

Therefore,

mh̄ω 1
p̂ =
. b̂ − b̂† .
2 i

3.7 For a harmonic oscillator with mass m and angular frequency ω, find out the
expression for p̂ 2 , the square of the momentum operator, in terms of lowering and
raising operators b̂ and b̂† . Use this expression to calculate 0|p̂2 |0, where |0 is
the ground eigenstate of the harmonic oscillator Hamiltonian.
Solution 3.7 The momentum operator p̂ can be expressed as

mh̄ω 1
p̂ =
. b̂ − b̂† .
2 i

Taking square of the above expression, we obtain

mh̄ω mh̄ω 2
.p̂ 2 = − (b̂ − b̂† )2 = − b̂ − b̂b̂† − b̂† b̂ + b̂†2 .
2 2
Exercise Problems with Solutions 95

Therefore,

mh̄ω mh̄ω mh̄ω

0|p̂2 |0 =
. 0| −b̂2 + b̂b̂† + b̂† b̂ − b̂†2 |0 = 0|b̂b̂† |0 = .
2 2 2
3.8 Given that the state of a harmonic oscillator with unit mass and unit angular
frequency, namely, m = ω = 1, is in the following state at time t=0:

1
|ψ = √ (|0 + |1) ,
.
2

find out expression for the time dependent expectation value of the position x̂(t) =
ψ; t|x̂|ψ; t.
Solution 3.8 The state at time t is given by

1
|ψ; t = √ e−iωt/2 |0 + e−3iωt/2 |1 .
.
2

Therefore,

1
x̂(t) =
. 0|x̂|0 + e−iωt 0|x̂|1 + eiωt 1|x̂|0 + 1|x̂|1 .
2
On the righthand side of the above expression,

. 0|x̂|0 = 1|x̂|1 = 0.

On the other hand,

√
1 γ 1 N0 1
0|x̂|1 = 1|x̂|0 = N0 N1
.
2
=√ =√ ,
γ N0 γ N1 2γ

where Eq. (3.46) has been used. As a result,

1
x̂(t) = √ cos(ωt).
.
2γ

3.9 For any eigenstate |v of a harmonic oscillator with angular frequency ω, prove
the following identities:

. I : ei Ĥ t/h̄ b̂e−i Ĥ t/h̄ |v = e−iωt b̂|v,

II : ei Ĥ t/h̄ b̂† e−i Ĥ t/h̄ |v = eiωt b̂† |v.

96 3 Harmonic Oscillator and Vibrational Spectroscopy

Solution 3.9 The identity I can be proved as follows. First, using

e−i Ĥ t/h̄ |v = e−iω(v+1/2)t |v,

.

one can show that

ei Ĥ t/h̄ b̂e−i Ĥ t/h̄ |v = ei Ĥ t/h̄ b̂|v e−iω(v+1/2)t .

.

Since b̂|v is an eigenstate of Ĥ with eigenvalue h̄ω(v − 1 + 1/2) = h̄ω(v − 1/2),

ei Ĥ t/h̄ b̂|v = eiω(v−1/2)t b̂|v.

.

Therefore,

I : ei Ĥ t/h̄ b̂e−i Ĥ t/h̄ |v = eiω(v−1/2)t e−iω(v+1/2)t b̂|v = e−iωt b̂|v.

.

Similarly, the identity II can be proved first recognizing that

ei Ĥ t/h̄ b̂† e−i Ĥ t/h̄ |v = ei Ĥ t/h̄ b̂† |ve−iω(v+1/2)t .

.

Since b̂† |v is an eigenstate of Ĥ with eigenvalue h̄ω(v + 1/2) = h̄ω(v + 3/2),

ei Ĥ t/h̄ b̂† |v = eiω(v+3/2)t b̂† |v.

.

Therefore,

II : ei Ĥ t/h̄ b̂† e−i Ĥ t/h̄ |v = eiω(v+3/2)t e−iω(v+1/2)t b̂† |v = eiωt b̂† |v.
.

3.10 For the following diatomic molecules with fundamental vibrational frequen-
cies (in wavenumber) as shown, calculate their respective force constants in the unit
of dyne cm−1 = g s−2 . Assuming that there are no anharmonic effects, determine
whether each of the molecules is IR and/or Raman active. Explain the reason for
your answer and also determine the frequency of transition for each case.
(a) H2 (ν̃e = 2321 cm−1 )
(b) 16 O2 (ν̃e = 1510 cm−1 )
(c) HCl (ν̃e = 1529 cm−1 )
(d) DCl (ν̃e = 1220 cm−1 )
√
Solution 3.10 Note that ν̃e = ω̃e /(2π c) = k/μ/(2π c), where k and μ are spring
constant and reduced mass. Thus, k = (2π cν̃e )2 μ. The table below shows these
data for the four molecules.
Exercise Problems with Solutions 97

Molecule μ (a.m.u.) μ (g) 2π cν̃e (s−1 ) k (gs−2 )

H2 0.5 8.31 × 10−25 4.38 × 1014 1.59 × 105
16 O
2 8 1.33 × 10−23 2.84 × 1014 1.07 × 106
HCl 0.97 1.61 × 10−24 2.88 × 1014 1.34 × 105
DCl 1.89 3.14 × 10−24 2.30 × 1014 1.66 × 105

Note that H2 and 16 O2 are only Raman active because they are symmetric while
HCl and DCl are both IR and Raman active.
3.11 For a certain heteroatomic diatomic molecule, its vibrational term can be
approximated well by the following expression:

1 2
   
1
.Gv = ν̃e v+ − 0.1ν̃e v + .
2 2

It is known that the molecule is either in the ground vibrational state v = 0 or the
first excited vibrational state v = 1. Assuming that the same selection rules for
harmonic oscillator apply, find out expressions for all possible frequencies of IR
absorption, Raman Stokes, and Raman Anti-Stokes frequencies.
Solution 3.11 The IR absorption and Raman stokes transitions are the same and
correspond to v: 0 → 1 or 1 → 2.
The wavenumber of the photon absorbed in IR or the red shift in Stokes Raman
can be calculated as follows.
For v: 0 → 1,
 
3 9 1 1
.G1 − G0 = ν̃e − 0.1ν̃e − ν̃e − 0.1ν̃e
2 4 2 4
 
3 9 1 1
= ν̃e − − + = 0.8ν̃e .
2 40 2 40

For v: 1 → 2,
 
5 25 3 9
G2 − G1 = ν̃e
. − 0.1ν̃e − ν̃e − 0.1ν̃e
2 4 2 4
 
5 25 3 9
= ν̃e − − + = 0.6ν̃e .
2 40 2 40

There is only one Raman Anti-Stokes, v: 1 → 0, for which G1 − G0 = 0.8ν̃ as

shown above.
98 3 Harmonic Oscillator and Vibrational Spectroscopy

Problems

3.12 A quantum harmonic oscillator with unit mass and angular frequency, i.e.,
m = ω = 1, is in the first excited eigenstate of the Hamiltonian, |1, for which the
eigenvalue is equal to 3h̄/2. Determine the uncertainties in position and momentum,
σx and σp , and show that they satisfy the uncertainty relationship.
3.13 Use the recursion relationship Eq. (3.34) and answers in Solution 3.3 to find
out expressions for H6 (ξ ) and H7 (ξ ).
3.14 For a quantum harmonic oscillator with mass m and angular frequency ω,
find out expression for x̂ 2 in terms of the lowering and raising operators, b̂ and b̂† .
Then, use this expression to calculate 2|x̂ 2 |2, where |2 is the eigenstate of the
Hamiltonian with eigenvalue E2 = 5h̄ω/2.
3.15 Given that the state of a harmonic oscillator with unit mass and angular
frequency, namely, m = ω = 1, is in the following state at time t = 0:

1
|ψ; 0 = √ (|0 + i|1) ,
.
2

where |0 and |1 are the ground and first excited states of the Hamiltonian, calculate
the expectation value of the energy for |ψ; t = π .
3.16 A quantum harmonic oscillator with unit mass (m = 1) and unit angular
frequency (ω = 1) at time t = 0 is in the linear superposition of its three normalized
eigenstates of the Hamiltonian as follows:

1 i 1
|ψ; 0 = √ |0 − √ |1 + √ |3,
.
3 3 3

where the eigenvalue for |v is h̄(v + 1/2). Calculate the expectation value of the
position for |ψ; t = π .
3.17 The vibrational term for HBr, where the atomic numbers of H and Br are
respectively 1 and 79, is given by

1 2
   
1
Gv = ν̃e
. v+ − xe ν̃e v + ,
2 2

with νe = 2649.67 cm−1 and xe ν̃e = 45.21 cm−1 . It is known that the molecule is
in one of the three vibrational states with v = 0, 1, and 2. Assuming that the same
selection rules for harmonic oscillator apply, find out expressions for all possible
frequencies of IR absorption, Raman Stokes, and Raman anti-Stokes frequencies.
Problems 99

3.18 For a certain homonuclear diatomic molecule, its vibrational term can be
approximated well by the following expression:
   2
1 1
Gv = ν̃e
. v+ − 0.1ν̃e v+ .
2 2

It is known that the molecule is in one of the two states with v = 0 and 1. Assuming
that the same selection rules for harmonic oscillator apply, find out expressions for
all possible Raman Stokes frequencies in the unit of ν̃e .
Chapter 4
Multidimensional Systems
and Separation of Variables

There’s no sense in being precise when you don’t even know

what you’re talking about.
— John von Neumann

Abstract This chapter provides time independent Schrödinger equations for a

quantum particle in three dimensional space and many interacting quantum particles
in multidimensional space. As the simplest example, a quantum particle in a three
dimensional box is considered. The corresponding eigenvalues and eigenfunctions
are provided employing the separation of variables method. These solutions are
used to illustrate the concept of degeneracy. Then, a more general description of the
separation of variables method is provided. As an application, quantum mechanical
description of the multidimensional vibrational motion of a polyatomic molecules,
within the harmonic oscillator approximation for each normal mode, is presented.

So far, we have considered one dimensional motion in order to understand the

principles and features of quantum mechanics as clearly as possible. However,
hardly anything is one dimensional. Typically, the motion of a single particle needs
to be considered in three dimensional physical space. If there are N particles
interacting in three dimensional space, we need to consider a 3N dimensional
mathematical space in order to provide complete description of the system even
for the description within the classical mechanics.
It turns out that extending quantum mechanical position and momentum oper-
ators and their corresponding eigenstates to multidimensional spaces is straight-
forward at least in Cartesian coordinate systems, where one dimensional position
and momentum states along each direction can be combined independently to form
the multidimensional space. The corresponding states can also be defined in a

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 101
S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_4
102 4 Multidimensional Systems and Separation of Variables

straightforward manner by direct products, as defined in Sect. 2.5. This results in

expansion of the dimension of the corresponding vector space of states to that of
the power of 3N. This chapter provides detailed consideration of these states and
introduces a simple separation of variables method for solving the multidimensional
Schrödinger equation.

4.1 Three Dimensional System

4.1.1 Position, Momentum, Hamiltonian, and Schrödinger

Equation

The position and momentum operator defined in one dimension can be extended to
three dimension through a direct product of those for x, y, and z as described in
Sect. 2.5. This then allows straightforward extension of Schrödinger’s equation to
three dimension.
Taking the Hermitian adjoint of the three dimensional position ket .|r defined by
Eq. (2.93), leads to the following three dimensional position bra:

.r| = x| ⊗ y| ⊗ z|. (4.1)

Then, for a quantum mechanical state .|ψ representing a particle in three dimension,
the value of the corresponding Schrödinger’s wave function at .r can be defined as
the following inner product:

ψ(r) = r|ψ.
. (4.2)

If .|ψ corresponds to another position ket .|r′ , the above identity leads to the three
dimensional Dirac-delta function as follows:1

δ(r − r′ ) = r|r′ .
. (4.3)

The above identities are equivalent to the following completeness relation that
expresses the identity operator in terms of outer products of positions states.

1 This three dimensional Dirac delta function, or distribution, satisfies the two conditions that
dr′ δ(r ′ (r′ ) = f (r) for any integrable function .f (r) and .δ(r − r′ ) = 0 for .r = r′ ,

.
 − r )f  
where . dr = dx dy dz in the Cartesian coordinate system.
4.1 Three Dimensional System 103

Completeness of Three Dimensional States

Three dimensional position states form a complete basis and integration of
its outer products over the entire space is equal to the identity operator as
expressed below.
 
1̂ =
. dr|rr| = lim δr|ri ri |. (4.4)
δr→0
i

The above identity is a three dimensional analogue of Eq. (2.25) and the
second equality simply represents the mathematical definition of the inte-
gration as the limit of the sum over smaller volume elements .δr becoming
infinitesimally small.

The above completeness relation implies that any quantum mechanical state .|ψ
describing a quantum particle in three dimensional space can be represented as a
linear combination of the position states. Thus,
  
|ψ =
. dr |rr|ψ = dr ψ(r)|r = lim δrψ(ri )|ri , (4.5)
δr→0
i

where Eq. (4.2) has been used. The second equality in the above equation shows
that .δrψ(ri ) is nothing but the coefficient for .|ri  in the representation of the state
as a linear combination of the position states.
Taking the Hermitian adjoint of Eq. (2.94), we obtain

.r|r̂ = r|r = rr|, (4.6)

where the fact that the position operator is a Hermitian operator and that .r is a real-
valued vector has been used. Applying the above identity to a state ket .|ψ, we then
obtain

r|r̂|ψ = rr|ψ = rψ(r),

. (4.7)

which is a three dimensional generalization of Eq. (2.24). Application of this

identity for the case where the state is .r̂|ψ leads to

r||r̂|2 |ψ = r|r̂ · r̂|ψ

.
 
= r|r̂ · r̂|ψ
= r · r|r̂|ψ
= r · rr|ψ = |r|2 r|ψ. (4.8)
104 4 Multidimensional Systems and Separation of Variables

For any function .V (r), one can use Eq. (4.7) repeatedly for a Taylor expansion of
V (r) and obtain
.

r|V (r̂)|ψ = V (r)r|ψ = V (r)ψ(r).

. (4.9)

In other words, applying an operator .V (r̂) to a state .|ψ simply amounts to

multiplying .ψ(r) (the position representation of .|ψ) with the function .V (r).
As the next step, let us consider the momentum operator in the three dimensional
space, .p̂ = p̂x ex + p̂y ey + p̂z ez , as introduced in Sect. 2.5. Here, each component
is defined by its effect on the corresponding position ket as follows:

h̄ ∂
. p̂x |x = − |x, . (4.10)
i ∂x
h̄ ∂
p̂y |y = − |y, . (4.11)
i ∂y
h̄ ∂
p̂z |z = − |z. (4.12)
i ∂z

On the other, each component does not have any effect on the position states of other
components. For example, .p̂x does not have any effect on .|y and .|z. Therefore,
application of .p̂ to .|r leads to

p̂|r = (p̂x ex + p̂y ey + p̂z ez )|x ⊗ |y ⊗ |z

.
   
= ex p̂x |x ⊗ |y ⊗ |z + ey |x ⊗ p̂y |y ⊗ |z + ez |x ⊗ |y ⊗ (pz |z)
h̄
= − ∇|r, (4.13)
i
where
∂ ∂ ∂
∇ = ex
. + ey + ez . (4.14)
∂x ∂y ∂z

The Hermitian adjoint of Eq. (4.13) is

h̄
r|p̂ =
. ∇r|. (4.15)
i
Taking inner product of the above identity with an arbitrary ket .|ψ, we find a
key identity that establishes the effect of momentum operation on arbitrary state
as described below.
4.1 Three Dimensional System 105

Implication of Three Dimensional Momentum Operation

For an arbitrary state .|ψ,

h̄ h̄
r|p̂|ψ =
. ∇r|ψ = ∇ψ(r). (4.16)
i i
The last term in the above equation is the definition of the momentum operator
in Schrödinger’s formulation of quantum mechanics. The complex conjugate
of the above identity is

∗ h̄ h̄
ψ|p̂|r = r|p̂|ψ = − ∇ψ ∗ (r) = − ∇ψ|r.

. (4.17)
i i
This identity can in fact be obtained more easily by taking the inner product
of Eq. (4.13) with .ψ| and employing the fact that .ψ|∇|r = ∇ψ|r.

In a way similar to Eq. (4.8), application of Eq. (4.16) twice leads to

h̄
r||p̂|2 |ψ = r|p̂·p̂|ψ =
. ∇·r|p̂|ψ = −h̄2 ∇ 2 r|ψ = −h̄2 ∇ 2 ψ(r), (4.18)
i

where note that .|p̂|2 = p̂ · p̂ = p̂x2 + p̂y2 + p̂z2 and

∂2 ∂2 ∂2
∇2 = ∇ · ∇ =
. + + . (4.19)
∂x 2 ∂y 2 ∂z2

Summing up all the definitions and identities provided above, now we can find a
general expression for the time independent Schrödinger equation for a particle in
three dimensional space as describe below.

Time Independent Schrödinger Equation in Three Dimension

The Hamiltonian operator for a quantum mechanical particle subject to a
potential energy .V (r) in a three dimensional space is

|p̂|2
.Ĥ = + V (r̂). (4.20)
2m

(continued)
106 4 Multidimensional Systems and Separation of Variables

Thus, the time independent Schrödinger equation for the particle in the Dirac
notation is
 2 
|p̂|
.Ĥ |ψE  ≡ + V (r̂) |ψE  = E|ψE , (4.21)
2m

where .|ψE  is the eigenket of .Ĥ with energy E. Taking inner product of the
above equation with .r| and using Eqs. (4.9) and (4.18), one can obtain the
original Schrödinger’s equation for time independent wavefunction in three-
dimensional space,

h̄2 2
r|Ĥ |ψE  ≡ −
. ∇ + V (r) ψE (r) = EψE (r), (4.22)
2m

where .ψE (r) ≡ r|ψE . The above equation is the position representation of
the Schrödinger equation in the Dirac notation. Solving this time independent
Schrödinger equation is an important first step for understanding the behavior
of a given quantum particle in a three dimensional space. While finding
an exact closed form solution of the above equation for a general three
dimensional potential is difficult, a simple method of solution exists for the
case where the potential energy can be separated into one dimensional terms.

4.1.2 Particle in a Three Dimensional Rectangular Box

As the simplest example, let us consider a quantum particle in the following three
dimensional rectangular box (see Fig. 4.1):

0 , 0 < x < a, 0 < y < b, 0 < z < c ← Region I

V3D−box (r) =
.
∞, otherwise ← Region II
(4.23)

For this potential, it is easy to find .ψE (r) = r|ψE  and eigenvalue E satisfying the
Schrödinger equation, Eq. (4.22).
For now, let us denote the wavefunction satisfying the time independent
Schrödinger equation for energy E as .ψ(r). First, for the region II in Eq. (4.23),
it is clear that .ψ(r) = 0. Otherwise, the energy becomes infinity, which is not an
acceptable solution. This imposes the following boundary condition to be satisfied
by the finite wave function in region I:
4.1 Three Dimensional System 107

Fig. 4.1 Three dimensional box with sides a, b, and c along x, y, and z directions

ψ(0, y, z) = ψ(a, y, z) = ψ(x, 0, z) = ψ(x, b, z)

.

= ψ(x, y, 0) = ψ(x, y, c) = 0. (4.24)

Along with the above boundary condition, Eq. (4.22) for region I reduces to

h̄2 2 h̄2 ∂2 ∂2 ∂2
 
. − ∇ ψ(r) = − + + ψ(r) = Eψ(r). (4.25)
2m 2m ∂x 2 ∂y 2 ∂z2

This equation can be solved by making the following assumptions:

. ψ(r) = X(x)Y (y)Z(z), . (4.26)

E = Ex + Ey + Ez . (4.27)

Inserting Eqs. (4.26) and (4.27) into Eq. (4.25), moving .Eψ(r) to the lefthand side,
and dividing the whole equation with .−h̄2 /(2m), we find that

d 2 X(x) 2mEx
 
. + X(x) Y (y)Z(z)
dx 2 h̄2
 2 
d Y (y) 2mEy
+ + Y (y) X(x)Z(z)
dy 2 h̄2
 2 
d Z(z) 2mEz
+ + Z(z) X(x)Y (y) = 0. (4.28)
dz2 h̄2
108 4 Multidimensional Systems and Separation of Variables

Let us divide the above equation with .X(x)Y (y)Z(z) (assuming that this is
nonzero). Then, it becomes

d 2 X(x) 2mEx
 
1
. + X(x)
X(x) dx 2 h̄2
 2 
1 d Y (y) 2mEy
+ + Y (y)
Y (y) dy 2 h̄2
 2 
1 d Z(z) 2mEz
+ + Z(z) = 0. (4.29)
Z(z) dz2 h̄2

The above equation shows that the sum of three independent functions of x, y,
and z, respectively, becomes zero. For this to be true in general, each term should
become zero. Thus, the above equation, along with the boundary condition, Eq.
(4.24), implies that

d 2 X(x) 2mEx
. + X(x) = 0, with X(0) = X(a) = 0, . (4.30)
dx 2 h̄2
d 2 Y (y) 2mEy
+ Y (y) = 0, with Y (0) = Y (b) = 0, . (4.31)
dy 2 h̄2
d 2 Z(z) 2mEz
+ Z(z) = 0, with Z(0) = Z(c) = 0. (4.32)
dz2 h̄2

Note that each of the above equation corresponds to that for the one dimensional
particle in a box problem that was considered in Sect. 1.4. Thus, the solutions for
the above three equations are

2 nx π x  h̄2 n2x π 2
. Xnx (x) = sin , Enx = , nx = 1, 2, . . . . (4.33)
a a 2ma 2
2 ny πy  h̄2 n2y π 2
Yny (y) = sin , Eny = , ny = 1, 2, . . . . (4.34)
b a 2mb2
2 nz π z  h̄2 n2z π 2
Znz (z) = sin , Enz = , nz = 1, 2, . . . (4.35)
c a 2mc2

Inserting these solutions into Eq. (4.26), we obtain the following eigenfunction

8 nx π x  ny πy  nz π z 
ψnx ny nz (r) =
. sin sin sin , (4.36)
abc a b c
4.1 Three Dimensional System 109

with the eigenvalue

h̄2 π 2 n2x n2y n2z

Enx ny nz =
. + + . (4.37)
2m a2 b2 c2

In the Dirac notation, the eigenstates can simply be represented by .|nx , ny , nz  =

|nx  ⊗ |ny  ⊗ |nz . The kets comprising this direct product are related to the
wavefunctions given above as follows: .Xnx (x) = x|nx ; .Yny (y) = y|ny ;
.Znz (z) = z|nz . Thus, .|nx , ny , nz  satisfies the following time independent

Schrödinger equation (in the Dirac notation):

|p̂|2
 
Ĥ |nx , ny , nz  =
. + V̂3D−box |nx , ny , nz  = Enx ny nz |nx , ny , nz . (4.38)
2m

In the above expression, .V̂3D−box corresponding to Eq. (4.23) is defined as

.V̂3D−box |r = V3D−box (r)|r. (4.39)

Although the eigenstates defined by Eq. (4.38) have been determined under the
assumption that the state is a product of those for each Cartesian coordinates, the
general quantum states need not necessarily be such product states. This is because
the state of a quantum particle in a box can be represented by the following linear
combination of eigenstates:

 ∞
∞ 
∞ 
|ψ =
. Cnx ,ny ,nz |nx , ny , nz . (4.40)
nx =1 ny =1 nz =1

Then, the corresponding wavefunction .ψ(r) = r|ψ is expressed as

 ∞
∞ 
∞ 
ψ(r) =
. Cnx ,ny ,nz ψnx ny nz (r). (4.41)
nx =1 ny =1 nz =1

It is well-known mathematically that the above expression can represent any well-
behaving function of .r. Thus, the solutions based on the assumption of separation of
variables can be used to represent any well-behaving non-separable wavefunctions
as long as the same boundary conditions are satisfied.
If Eq. (4.40) is the state at time .t = 0, then the state at time t becomes
∞  ∞
∞ 
Cnx ,ny ,nz e−i(Enx +Eny +EnZ )t/h̄ |nx , ny , nz .

|ψ; t =
. (4.42)
nx =1 ny =1 nz =1
110 4 Multidimensional Systems and Separation of Variables

The corresponding time dependent wavefunction .ψ(r; t) = r|ψ; t is expressed as

∞  ∞
∞ 
Cnx ,ny ,nz e−i(Enx +Eny +EnZ )t/h̄ ψnx ny nz (r).

ψ(r; t) =
. (4.43)
nx =1 ny =1 nz =1

Having described a simple three dimensional system, it is now appropriate to

explain the concept of degeneracy, which plays an important role in quantum
mechanics. For a general case where .a = b = c and none of them is an
integer multiple of the other, the energy eigenvalue .Enx ny nz is different for each
set of .(nx , ny , nz ). However, in case the box becomes more symmetrical, some of
the energies become the same, namely, “degenerate." In that case, that particular
energy is said to have degeneracy and the corresponding quantum states are called
degenerate.
As an example, let us consider the case where .a = b = c. In this case, Eq. (4.37)
reduces to

h̄2 π 2 2 2 2

Enx ny nz =
. n + n + nz , (4.44)
2ma 2 x y

where .nx , ny , nz = 1, 2, 3, . . . . While the ground state .|1, 1, 1 with energy .E111 is
non-degenerate, the first excited states .|2, 1, 1, .|1, 2, 1, and .|1, 1, 2 are degenerate
with a common eigenvalue .E211 . Thus, the degeneracy of this eigenvalue is 3. It is
important to note that any linear combination of degenerate eigenstates remains the
eigenstate. This is obvious because application of the Hamiltonian to such a state
results in the same eigenvalue that can be factored out. For example, any linear
combination of .|2, 1, 1, .|1, 2, 1, and .|1, 1, 2 is an eigenstate of .Ĥ with eigenvalue
.E211 .

4.1.3 Separation in Cartesian Coordinate System

The three dimensional particle in a box problem is a special case of a more general
situation where the three dimensional Schrödinger equation can be separated into
three one dimensional equations. As long as the potential is given by a sum of three
terms as

V (r) = Vx (x) + Vy (y) + Vz (z),

. (4.45)

and the boundary condition in each direction is independent of the other coordinates,
one can assume that the energy is given by the sum of three terms as

E = Ex + Ey + Ez ,
. (4.46)
4.1 Three Dimensional System 111

and the eigenfunction of the Hamiltonian can be assumed to have the following
factorized form:

ψ(r) = ψx (x)ψy (y)ψz (z).

. (4.47)

Then, employing the separation of variables method as was done for the case of the
particle in a three dimensional box, it is straightforward to show that the Schrödinger
equation for the particle can be satisfied through the following three one dimensional
equations:

h̄2 d 2
. − + Vx (x) ψx (x) = Ex ψx (x), . (4.48)
2m dx 2

h̄2 d 2
− + Vy (y) ψy (y) = Ey ψy (y), . (4.49)
2m dy 2

h̄2 d 2
− + Vz (z) ψz (z) = Ez ψz (z). (4.50)
2m dz2

As an example, consider a three dimensional harmonic oscillator, for which the

potential energy is given by
m 2 2
V (r) =
. (ω x + ωy2 y 2 + ωz2 z2 ). (4.51)
2 x
This potential is a sum of three independent terms along the three directions. Solving
one dimensional harmonic oscillator problem in each direction and combining the
solution, one finds that
      
1 1 1
.Enx ,ny ,nz = h̄ ωx nx + + ωy ny + + ωz nz + , (4.52)
2 2 2

where .nx , ny , nz = 0, 1, 2, . . . and

ψnx ny nz (r) = ψnx (x)ψny (y)ψnz (z),

. (4.53)

with
1 γx 1/4 −γx x 2 /2 √
. ψnx (x) = √ n
e Hnx ( γx x), . (4.54)
2 nx ! π
x

1 γy 1/4 −γy y 2 /2 √
ψny (y) =  n e Hny ( γy y), . (4.55)
2 y ny ! π
1 γz 1/4 −γz z2 /2 √
ψnz (z) =  e Hnz ( γz z), (4.56)
n
2 nz !
z π
112 4 Multidimensional Systems and Separation of Variables

where .γp = mωp /h̄ with .p = x, y, and z, and .Hn represents the nth order Hermite
polynomial.
In the Dirac notation, the eigenstate of the three dimensional Harmonic oscillator
can be expressed as

|nx , ny , nz  = |nx |ny |nz  ≡ |nx  ⊗ |ny  ⊗ |nz .

. (4.57)

The corresponding wavefunction in the position representation is obtained by taking

the inner product of this state with a position state, .r| = x|y|z| ≡ x| ⊗ y| ⊗ z|,
as follows:

. ψnx ny nz (r) = r|nx , ny , nz 

= x|nx y|ny z|nz  = ψnx (x)ψny (y)ψnz (z). (4.58)

4.2 Many Particle Systems and the Center of Mass

Coordinates

The Hamiltonian operator for a quantum system consisting of N particles can

be found by replacing their classical momentum and position with corresponding
operators as follows:

|p̂1 |2 |p̂ |2
Ĥ =
. + · · · + N + V (r̂1 , · · · , r̂N ), (4.59)
2m1 2mN

where .p̂k and .r̂k for each of .k = 1, · · · , N are momentum and position operators
of particle k with mass .mk . For this many-particle system, one can construct 3N
dimensional cartesian coordinates .R = (r1 , · · · , rN ), for which a position ket .|R
can be defined through direct products,

|R = |r1  ⊗ · · · ⊗ |rN .

. (4.60)

Let us denote a quantum state representing this N-particle system in the Dirac nota-
tion as .|N . Then, the corresponding wavefunction in the position representation is
obtained by taking the inner product of this state with the multidimensional position
bra as follows:
 
N (r1 , · · · , rN ) = R|N  = r1 | ⊗ · · · ⊗ rN | |N .
. (4.61)

Note that .|N , even if it were an eigenstate, in general cannot be expressed as a

direct product of the single particle states (unlike the position state) if the potential is
not separable. Thus, the right hand side of the above expression cannot be simplified
further unless there is some symmetry in the system.
4.2 Many Particle Systems and the Center of Mass Coordinates 113

One case of simplification that is fairly common is when the potential energy can
be separated into those for the center of mass coordinates and remaining relative
coordinates. Before considering this case in more detail, let us define the following
center of mass coordinates,

m1 r1 + · · · + mN rN
rcm =
. . (4.62)
m1 + · · · + mN

Then, let us also define new relative coordinates .r′1 , · · · , r′N−1 that are orthogonal
to each other and are independent of .rcm . Then, the position ket for the total
coordinates, Eq. (4.60), can alternatively be written as

|R = |rcm  ⊗ |r′1  ⊗ · · · ⊗ |r′N−1 .

. (4.63)

Now assuming that the potential energy is separable as follows:

V (r̂1 , · · · , r̂N ) = Vcm (r̂cm ) + Vrel (r̂′1 , · · · , r̂′N−1 ),

. (4.64)

the Hamiltonian operator for the N particle system, Eq. (4.60), can be expressed as

|p̂cm |2 |p̂′1 |2 |p̂′N−1 |2

Ĥ =
. + + · · · + + Vcm (r̂cm ) + Vrel (r̂′1 , · · · , r̂′N−1 ), (4.65)
2M 2m′1 2m′N−1

where .p̂cm and .p̂′k (for .k = 1, · · · , N − 1) are canonical momentum operators

conjugate to .r̂cm and .r̂′k , M is the total mass of the system and .m′k (for .k =
1, · · · , N − 1) is the effective mass for the kth coordinate .r′k . For the above
Hamiltonian, an eigenstate can be assumed to be a product form as follows:

|N  = |ψcm  ⊗ |ψrel ,

. (4.66)

where .|ψcm  and .|ψrel  are eigenstates respectively satisfying the following
Schrödinger equations:

|p̂cm |2
 
. + Vcm (r̂cm ) |ψcm  = Ecm |ψcm , . (4.67)
2M

|p̂′1 |2 |p̂′N−1 |2
+ ··· + + Vrel (r̂′1 , · · · , r̂′N −1 ) |ψrel  = Erel |ψrel . (4.68)
2m′1 2m′N−1

Then, the total wave function, the position representation of the state, can be
expressed as

N (R) = R|N  = (rcm |ψcm ) r′1 | ⊗ · · · ⊗ r′N−1 |ψrel 
.

= ψcm (rcm )ψrel (r′1 , · · · , r′N−1 ). (4.69)

114 4 Multidimensional Systems and Separation of Variables

4.2.1 Two-Particle System

As an example, let us consider a system of two particles, with masses .m1 and
.m2 . Within the classical mechanics, it is possible to completely specify the three
dimensional coordinates of two particles, .r1 and .r2 , and their momenta, .p1 and
.p2 . For the general case where the two particles interact with each other, as was

mentioned before, it is often best to separate the dynamics into those for the center
of mass and relative coordinates.
For the two particles, the center of mass coordinate is

m1 r1 + m2 r2
rcm =
. , (4.70)
M
where .M = m1 + m2 , and the relative coordinate is

rrel = r1 − r2 .
. (4.71)

The above two equations can be used to express .r1 and .r2 in terms of .rcm and .rrel
as follows:
m2
. r1 = rcm + rrel , . (4.72)
M
m1
r2 = rcm − rrel . (4.73)
M
Taking the time derivative of the above two equations leads to

dr1 drcm m2 drrel

. = + ,. (4.74)
dt dt M dt
dr2 drcm m1 drrel
= − . (4.75)
dt dt M dt
As a result,

p21 p2 dr1 2 m2 dr2 2

   
m1
. + 2 = +
2m1 2m2 2 dt 2 dt
 2 2 
m22
    
m1 drcm m2 drcm drrel drrel
= +2 · +
2 dt M dt dt M2 dt
 2 
drcm 2 m21
     
m2 m1 drcm drrel drrel
+ −2 · +
2 dt M dt dt M2 dt
4.2 Many Particle Systems and the Center of Mass Coordinates 115

 2  2
M drcm μ drrel
= +
2 dt 2 dt
p2cm p2
= + rel , (4.76)
2M 2μ

where .μ = m1 m2 /M, the reduced mass for the two particle system, and

drcm
pcm = M
. ,. (4.77)
dt
drrel
prel =μ . (4.78)
dt

Hamiltonian and Schrödinger Equation for Two Particles with Separable

Relative Potential
If the potential energy is a sum of those for .rcm and .rrel , namely, .V (r1 , r2 ) =
Vcm (rcm ) + Vrel (rrel ), the total classical Hamiltonian for the two particles can
be expressed as

p2cm p2
H = Hcm + Hrel =
. + Vcm (rcm ) + rel + Vrel (rrel ). (4.79)
2M 2μ

For the quantum mechanical case, the Hamiltonian operator is obtained by

replacing the position and momentum variables with operators as follows:

p̂2cm p̂2
Ĥ = Ĥcm + Ĥrel =
. + Vcm (r̂cm ) + rel + Vrel (r̂rel ). (4.80)
2M 2μ

Let us introduce .|ψcm,Ecm  and .|ψrel,Erel , which are respectively eigenstates

of .Ĥcm and .Ĥrel as follows:

Ĥcm |ψcm,Ecm  = Ecm |ψcm,Ecm , .

. (4.81)
Ĥrel |ψrel,Erel  = Erel |ψrel,Erel . (4.82)

Then, it is straightforward to show that .|ψ = |ψcm,Ecm  ⊗ |ψrel,Erel  is an

eigenstate of .Ĥ = Ĥcm + Ĥrel with eigenvalue .Ecm + Erel .
116 4 Multidimensional Systems and Separation of Variables

4.2.2 Normal Modes and Vibrational Spectroscopy

of Polyatomic Molecules

For a diatomic molecule in the absence of any external potential, the Hamiltonian
operator given by Eq. (4.79) reduces to

p̂2cm p̂2
Ĥ =
. + + Vrel (r̂), (4.83)
2M 2μ

where, note that the subscript rel used for the position operator in the previous
subsection has been dropped here for convenience. Thus, .r̂ is the operator measuring
the relative position vector of the two nuclei and .p̂ is the corresponding momentum
operator. As will become detailed in the next chapter, the kinetic energy operator for
the relative motion of the nuclei .p̂2 /(2μ) in the above equation can be divided into
one dimensional motion along the radius and rotational motions that are specified by
two independent angles. Thus, for the diatomic molecule represented by the above
Hamiltonian, which have total of 6 degrees of freedom, there is only one vibrational
degree of freedom.
For polyatomic molecules with more than two nuclei, there are more than one
vibrational degrees of freedom. For linear molecules with N atoms, the number
of vibrational degrees of freedom, .Nvib = 3N − 5. This is because there are
two rotational degrees of freedom in addition to three degrees of freedom for
the translational motion of the center of mass. On the other hand, for nonlinear
molecules, there are three rotational degrees of freedom. As a result, there are
.Nvib = 3N − 6 vibrational degrees of freedom.

Let us consider a polyatomic molecule with .Nvib vibrational degrees of freedom

and assume that it is possible2 to identify .Nvib orthogonal coordinates .xk ’s with
.k = 1, · · · , Nvib , which represent vibrational displacements from the minimum

potential energy. The kinetic energy term can still be expressed as a quadratic
form involving time derivatives of these coordinates. In other words, it is assumed
thatthe kinetic energy corresponding to vibrational motion can be expressed
as . k (mk /2)(dxk /dt)2 and that the difference of the potential energy from its
minimum value can be expanded as follows:

. V = V (x1 , · · · , xNvib ) − V (0, · · · , 0)

 2 
1  ∂ V
= xi xj + · · ·
2 ∂xi ∂xj
i j

1 
= fij xi xj + · · · , (4.84)
2
i j

2 Actualprocedure of identifying orthogonal vibrational coordinates for a general polyatomic

molecule is nontrivial. For general description, see [29].
4.2 Many Particle Systems and the Center of Mass Coordinates 117

Given that the displacements .xk are sufficiently small, the potential energy can be
approximated up to the second order and the classical Hamiltonian representing the
vibrational motion of the molecule can be approximated as
1 1 
Hvib =
. mj ẋj2 + fj k xj xk . (4.85)
2 2
j j k

√
Let us introduce mass-weighted coordinates .qj = mj xj . Then, the above
Hamiltonian can be expressed as
1 1  ˜
Hvib =
. q̇j2 + fj k qj qk , (4.86)
2 2
j j k

where
fj k
f˜j k = √
. . (4.87)
mj mk

In Eq. (4.86), the potential energy term can be expressed as

1  ˜ 1
. fj k qj qk = qT F̃q, (4.88)
2 2
j k

where .qT = (q1 · · · qNvib ), the transpose of a column vector .q with the same
elements, and
⎛ ˜
f˜12 f˜1Nvib
⎞
f11 ···
⎜ f˜21 f˜22 ··· f˜2Nvib ⎟
.F̃ = ⎜ . ⎟. (4.89)
⎜ ⎟
.. .. ..
⎝ .. . . . ⎠
f˜Nvib 1 f˜Nvib 2 · · · f˜Nvib Nvib

Note that the above matrix is real and symmetric. This is called Hessian matrix and
has .Nvib real eigenvalues. Thus, one can find out .Nvib eigenvectors of .F̃ such that

F̃uk = λk uk , k = 1, · · · , Nvib .
. (4.90)

The above eigenvectors can be chosen to be normalized. In addition, the real

symmetric nature of .F̃ makes the eigenvectors orthogonal to each other. Thus, the
following property is satisfied.

.uTk uj = δj k , j, k = 1, 2, · · · , Nvib . (4.91)

118 4 Multidimensional Systems and Separation of Variables

Now one can express .q in terms of these eigenvectors as follows:

q = Q1 u1 + Q2 u2 + · · · + QNvib uNvib .
. (4.92)

Then, due to the properties of Eqs. (4.90) and (4.91), Eq. (4.88) can be expressed as

Nvib
1 T 1
. q F̃q = λj Q2j . (4.93)
2 2
j =1

In the above expression, all .λj ’s are positive because the above term is a quadratic
expansion of the potential energy V with respect to its minimum. If, for example,
.λj is negative, the displacement of .Qj , while all others remain zero, can make the

potential energy lower than the minimum, which is contradictory. Therefore, we can
introduce .ωj such that

λj = ωj2 , j = 1, · · · , Nvib .
. (4.94)

Taking time derivative of Eq. (4.92), we find that

q̇ = Q̇1 u1 + Q̇2 u2 + · · · + Q̇Nvib uNvib .

. (4.95)

Then, due to the orthonormality condition, Eq. (4.91), the kinetic energy term in Eq.
(4.86) can be expressed as
1 1 T 1 2
. q̇j2 = q̇ · q̇ = Q̇j . (4.96)
2 2 2
j j

Therefore, combining Eqs. (4.93), (4.94), and (4.96), the Hamiltonian can be
expressed as
1  1 
Hvib =
. Q̇2j + ωj2 Q2j = Pj2 + ωj2 Q2j , (4.97)
2 2
j j

where .Pj = Q̇j . The displacement vector represented by each eigenvector .uk is
called the normal mode of vibration and .ωj is the corresponding normal mode
vibrational frequency. Thus, Eq. (4.97) shows that the quadratic expansion of the
deviation of the potential energy from its minimum can be expressed as a sum
of .Nvib one dimensional harmonic oscillator Hamiltonians of independent normal
modes.
4.2 Many Particle Systems and the Center of Mass Coordinates 119

Molecular Vibrational Hamiltonian as a Sum of Harmonic Oscillators for

Normal Modes
Due to the independence of amplitude along the direction of each normal
mode from others, one can view the vibrational motion of polyatomic
molecules as being consisting of one dimensional normal modes. Therefore,
the quantum mechanical Hamiltonian operator for the vibrational motion of
the polyatomic molecule can be expressed as
1 
Ĥvib =
. P̂j2 + ωj2 Q̂2j . (4.98)
2
j

Let us denote the eigenstate of the j th term of the above harmonic

oscillator Hamiltonian with eigenvalue .h̄ωj (vj + 1/2) as .|vj . Namely,
 
1  1
Ĥvib,j |vj  =
. P̂j2 + ωj2 Q̂2j |vj  = h̄ωj vj + |vj . (4.99)
2 2

Then, we can define the direct product of these states as

|ψv  = |v1  ⊗ · · · ⊗ |vNvib .

. (4.100)

It is easy to show that the above state is an eigenstate of the total vibrational
Hamiltonian as follows:

Ĥvib |ψv  = Ev1 ,··· ,vN

. |ψv , (4.101)
vib

where
N vib  
 1
.Ev1 ,··· ,vN = h̄ωj vj + . (4.102)
vib 2
j =1

At the same time, one can define the position state of normal coordinates as
direct product of those for individual modes,

. |Q = |Q1  ⊗ · · · ⊗ |QNvib . (4.103)

Then, the position representation of the eigenstate is given by the product of

those for individual modes,
N
 vib N
 vib
ψvib (Q) = Q|ψv  =
. Qj |vj  = ψvj (Qj ). (4.104)
j =1 j =1
120 4 Multidimensional Systems and Separation of Variables

Not all the vibrational normal modes can be detected by the IR spectroscopy
and/or the Raman spectroscopy. As was mentioned in Chap. 3, only those modes
for which dipole moment changes are IR active. On the other hand, vibrational
modes for which the polarizability tensor changes with displacements are Raman
active. There are vibrational modes that are not active either in IR or Raman, which
are called silent modes. A systematic consideration of these vibrational modes is
possible employing the group theory, which will be described in more detail in
Chap. 9 where we consider overall molecular spectroscopy.
For each of IR and Raman active modes, the allowed transitions within the
harmonic oscillator approximation correspond to

. vj = ±1, (4.105)

where .vj is the vibrational quantum number for either IR or Raman active mode.
For IR absorption, only . vj = 1 transition occurs. As was the case for diatomic
molecule, the majority of these transitions are from the ground vibrational state
(.vj = 0) to the first excited vibrational state (.vj = 1). For the Raman spectroscopy,
. vj = 1 corresponds to Stokes transition and . vj = −1 corresponds to anti-

Stokes transition. The Stokes transition is dominantly from the ground (.vj = 0) to
the first excited state (.vj = 1), whereas the anti-Stokes transition is dominantly from
the first excited vibration state (.vj = 1) to the ground vibrational state (.vj = 0).
Relative intensities of Stokes and anti-Stokes peaks are determined by populations
of initial states. Since thermal population of the first excited vibrational state is much
smaller than that of the ground vibrational state for typical small to intermediate size
molecules, intensities of Stokes peaks are much stronger than those of anti-Stokes
peaks for these molecules.

4.3 Summary and Questions

Three dimensional position and momentum states can be defined as direct products
of those for x, y, z components of the Cartesian coordinate system. Accordingly, all
quantum mechanical operators, including the Hamiltonian, for a quantum particle
in three dimensional space can be defined in the direct product space of those for
the three dimensional position or momentum coordinates. Similarly, for a system of
N interacting quantum particles, direct products of position and momentum states
can be used to construct position and momentum states for the corresponding 3N
dimensional position and momentum vectors. The resulting direct product space of
position or momentum in turn can be used to represent any physical observables,
including the Hamiltonian, of the many-particle quantum system for any kind of
interacting potentials.
On the other hand, when it comes to finding analytical solutions for a general
multidimensional Schrödinger equation, it is in general necessary to decompose
the equation into those for one dimensional coordinates. The symmetry of a given
4.3 Summary and Questions 121

system plays an important role in such decomposition. For the case of a particle
confined in a three dimensional rectangular box, it is possible to separate the
equation into the sum of three independent terms for different Cartesian components
and to assume that the wavefunction is given by a product form of those for the three
components. For a three dimensional harmonic oscillator, a similar decomposition
into each direction is possible. For these systems with multiple degrees of freedom,
it is often the case that certain eigenstates have the same eigenvalue. These states
are called degenerate and the number of states with the same eigenvalue is called
the degeneracy of the eigenvalue.
Many particle systems in general can be described more conveniently by
separating the center of mass coordinates and all other relative coordinates. This
is the case for quantum many particle systems as well. For polyatomic molecules
consisting of N atoms, the wavefunction representing the three dimensional center
of mass motion can factor out as a simple free motion if there are no external
potentials. After subtracting additional two (for linear molecule) or three degrees
of freedom representing rotation, the Hamiltonian describing the pure vibrational
motion can be identified.
Given that the potential energy for the multidimensional vibrational motion
can be approximated up to the quadratic terms of the vibrational displacements
around the potential minimum, it is possible to diagonalize the second derivative
matrix of the potential energy, which is called Hessian matrix, resulting in normal
vibrational modes. Each of these normal modes can be described as independent
one dimensional vibrational motion, making it possible to reduce the quantum
mechanical vibrational motion of polyatomic molecules as independent products
of those for effectively one dimensional harmonic oscillators. Out of these normal
modes, those that induce change in the dipole moment are IR active and those that
induce change in the polarizability are Raman active.

Questions
• How can a direct product space be constructed to represent states of many
quantum particles out of those for single particle states? Is the direct
product space complete even though the particles are interacting with each
other?
• Consider three dimensional position and momentum operators of a
quantum particle. Given the components of these along two orthogonal
directions, can they always be measured simultaneously with arbitrary
precision?
• What does the degeneracy inform about the symmetry of a quantum
system? Does the system have more degeneracy if it has more symmetry?
• What is the degeneracy of the first excited energy of a quantum particle
confined in a cube?
(continued)
122 4 Multidimensional Systems and Separation of Variables

• When can the Scrödinger equation for a multidimensional system be solved

by the separation of variables method?
• For a molecule with a nonzero net charge, can one use the center of mass
coordinate system to define the dipole moment of the molecule uniquely?
• Why introducing a mass weighted coordinates is convenient for describing
the vibrational motion of a polyatomic molecule in general?
• Is it possible for a certain vibrational normal mode of a polyatomic
molecule to be active in neither IR nor Raman spectroscopy?

Exercise Problems with Solutions

4.1 A quantum state |ψ represents a quantum particle in three dimensional space,
and the corresponding wave function is as follows:

1 1   
r|ψ = √
. exp {i (k · r)} = √ exp i kx x + ky y + kz z ,
2π h̄ 2π h̄

where k = kx ex + ky ey + kz ez is a constant three dimensional vector. Find out

expressions for r||r̂|2 |ψ, r||p̂|2 |ψ, r|(r̂ · p̂ + p̂ · r̂)|ψ.
Solution 4.1 First, for r||r̂|2 |ψ,

r||r̂|2 |ψ = (x 2 + y 2 + z2 )r|ψ

.

1
= (x 2 + y 2 + z2 )ei(kx x+ky y+kz z) .
2π h̄

For r||p̂|2 |ψ,

r||p̂|2 |ψ = r|(p̂x2 + p̂y2 + p̂z2 )|ψ

.

 2
∂2 ∂2

∂
= −h̄2 + + r|ψ
∂x 2 ∂y 2 ∂z2
 2
∂2 ∂2

∂ 1
= −h̄2 + + √ ei(kx x+ky y+kz z)
∂x 2 ∂y 2 ∂z2 2π h̄
1
= h̄2 (kx2 + ky2 + kz2 ) √ ei(kx x+ky y+kz z) .
2π h̄
Exercise Problems with Solutions 123

Finally, for r|(r̂ · p̂ + p̂ · r̂)|ψ,

. r|(r̂ · p̂ + p̂ · r̂)|ψ
 
h̄ ∂ ∂ ∂ ∂ ∂ ∂
= x +y +z + x+ y + z r|ψ
i ∂x ∂y ∂z ∂x ∂y ∂z
   
h̄ ∂ ∂ ∂ h̄
= 2 x +y +z +3 r|ψ
i ∂x ∂y ∂z i
 1
ei(kx x+ky y+kz z) .

= h̄ 2(xkx + yky + zkz ) − 3i √
2π h̄

4.2 A quantum state |ψ represents a quantum particle in three dimensional space,
and the corresponding wave function is as follows:

γ 3/4  γ 
r|ψ =
. exp − x 2 + y 2 + z2 ,
π 2

where γ is a constant real number. Find out expressions for r||r̂|2 |ψ, r||p̂|2 |ψ,
r|(r̂ · p̂ + p̂ · r̂)|ψ.
Solution 4.2 First, for r||r̂|2 |ψ,

.r||r̂|2 |ψ = (x 2 + y 2 + z2 )r|ψ

γ 3/4 2 2 2 2
= (x + y 2 + z2 )e−γ (x +y +z )/2 .
π

For r||p̂|2 |ψ,

∂2 ∂2 ∂2
 
2 2
. r||p̂| |ψ = −h̄ + + r|ψ
∂x 2 ∂y 2 ∂z2
 
γ 3/4 2 ∂ ∂ ∂ 2 2 2
=− h̄ (−γ x) + (−γ y) + (−γ z) e−γ (x +y +z )/2
π ∂x ∂y ∂z
γ 3/4 2  2 2 2
= h̄ 3γ − γ 2 (x 2 + y 2 + z2 ) e−γ (x +y +z )/2 .
π

Finally, for r|(r̂ · p̂ + p̂ · r̂)|ψ,

. r|(r̂ · p̂ + p̂ · r̂)|ψ
 
h̄ ∂ ∂ ∂ ∂ ∂ ∂
= x +y +z + x+ y + z r|ψ
i ∂x ∂y ∂z ∂x ∂y ∂z
124 4 Multidimensional Systems and Separation of Variables

   
γ 3/4 h̄ ∂ ∂ ∂ 2 2 2
= 2 x +y +z − 3h̄i e−γ (x +y +z )/2
π i ∂x ∂y ∂z
γ 3/4  2 2 2
= 2i h̄γ (x 2 + y 2 + z2 ) − 3i h̄ e−γ (x +y +z )/2 .
π

4.3 For an electron confined in a three dimensional box, for which the lengths of
three sides are 1.00, 2.00, and 3.00 nm, calculate the ground state, first excited, and
second excited energies. Determine the degeneracy for each state.
Solution 4.3 Let us introduce l = 1.00 nm, and denote the three sides of the box
as a = l, b = 2l, c = 3l. Let us also denote the mass of electron as me = 9.109 ×
10−31 kg. Then, the energy eigenvalue of the electron in the three dimensional box
is given by the following expression:

h̄2 π 2 n2y n2z

En =
. n2x + + , nx , ny , nz = 1, 2, 3, . . . .
2me l 2 4 9

Below is the table of four lowest possible values of the sum within the parenthesis
in the above equation.

nx ny nz n2x + n2y /4 + n2z /9

1 1 1 1 + 1/4 + 1/9 = 49/36
1 1 2 1 + 1/4 + 4/9 = 61/36
1 2 1 1 + 4/4 + 1/9 = 76/36
1 1 3 1 + 1/4 + 9/9 = 81/36

On the other hand,

h̄2 π 2 (1.055 × 10−34 )2 · 3.142

. = J = 6.02 × 10−20 J.
2me l 2 2 · 9.109 × 10−31 · (1.00 × 10−9 )2

Combining this value with the table, we find that Eg = 8.19 × 10−20 J = 0.511 eV,
Eex−1 = 10.20 × 10−20 J = 0.637 eV, and Eex−2 = 12.71 × 10−20 J = 0.7933 eV.
None of these three states is degenerate. So, the degeneracy of each state is 1.
4.4 For an electron confined in a three dimensional box, for which two sides have
the same length of 1.00 nm, and the remaining side has 2.00 nm, calculate the
ground state, first excited state, and the second excited energies. Determine the
degeneracy for each sate.
Exercise Problems with Solutions 125

Solution 4.4 Using the same definition of l and me as in Solution 4.3, the energy
eigenvalue for the present case becomes

h̄2 π 2 n2z
En =
. n2x + n2y + .
2me l 2 4

The table for the lowest possible four values of the sum within the parenthesis in the
above equation is as follows.

nx ny nz n2x + n2y /4 + n2z /9

1 1 1 1 + 1 + 1/4 = 9/4
1 1 2 1+1+1=3
1 1 3 1 + 1 + 9/4 = 17/4
2 1 1 4 + 1 + 1/4 = 21/4
1 2 1 1 + 4 + 1/4 = 21/4

Therefore, Eg = 13.5×10−20 J = 0.843 eV, Eex−1 = 18.1×10−20 J = 1.13 eV,

and Eex−2 = 25.6 × 10−20 J = 1.60 eV. None of these three states is degenerate.
So, the degeneracy of each state of the three states is 1. As can be seen from the
table above, the degeneracy appears in the third excited state.
4.5 A quantum particle with unit mass (m = 1) is subject to the following potential
energy.

1 2 2
2ω z , 0 < x < a, 0 < y < a
V (r) =
. ,
∞, otherwise

where the angular frequency ω and a are related by a 2 ω = 2π 2 h̄. What is the ground
state energy and wavefunction of this particle? What is the first excited state energy
of this particle and its degeneracy? Find out all the degenerate eigenfunctions for
the first excited state.
Solution 4.5 The Schrödinger equation for the wavefunction for 0 < x < a and
0 < y < a is

h̄2 ∂2 ∂2 ∂2
 
1
. − + + ψ(x, y, z) + mω2 z2 ψ(x, y, z) = Eψ(x, y, z).
2m ∂x 2 ∂y 2 ∂z2 2

For other regions, the wavefunction is zero. Through separation of variables, this
can be solved with
2 π nx x  π ny y 
ψnx ,ny ,v (x, y, z) =
. sin sin ψv (z),
a a a
126 4 Multidimensional Systems and Separation of Variables

where nx , ny = 1, 2, · · · , v = 0, 1, 2, · · · , and

2 /2 √
ψv (z) = Nv e−γ z
. Hv ( γ z),

with γ = mω/h̄. The corresponding eigenvalue is

π 2 h̄2 2  1
Enx ,ny ,v =
.
2
nx + n2y + h̄ω(v + ).
2a 2

Since a 2 = 2π 2 h̄/ω as assumed, the following relation holds:

π 2 h̄2 h̄ω
.
2
= .
2a 4
Therefore,

h̄ω 2 1
.Enx ,ny ,v = (nx + n2y ) + h̄ω(v + ).
4 2
The ground state energy and the wavefunction is for nx = ny = 1 and v = 0 as
follows:
h̄ω h̄ω
. E1,1,0 = (1 + 1) + = h̄ω,
4 2
2 πx  πy 
ψ110 (x, y, z) = sin sin ψ0 (z).
a a b
The first excited state energy is for nx = 2, ny = 1, v = 0 or for nx = 1, ny = 2,
v = 0 as follows:
5h̄ω h̄ω 7h̄ω
E2,1,0 = E1,2,0 =
. + = ,
4 2 4
for which the two degenerate wavefunctions are
 
2 2π x πy 
. ψ210 (x, y, z) = sin sin ψ0 (z),
a a b
 
2 πx  2πy
ψ120 (x, y, z) = sin sin ψ0 (z).
a a b

4.6 For each of the molecule, determine the number of vibrational degrees of
freedom, and draw pictures of all IR inactive modes.
Exercise Problems with Solutions 127

(a) Carbonyl-sulfide (OCS)

(b) Acetylene (C2 H2 , linear form in the ground state)
(c) Phosphine (PH3 )
(d) Cyclopropenylidene (C3 H2 )
Solution 4.6 Answers (except for the pictures of IR active modes) for each case are
provided below.
(a) This is a linear molecular. Therefore, there are 9 − 5 = 4 vibrational modes.
There are no IR inactive modes. In other words, all vibrational modes are IR
active. This is because all vibrational normal modes change the dipole moments
of the molecule in one way or the other.
(b) This is a linear molecule. Therefore, there are 12 − 5 = 7 vibrational modes.
Out of these, there are two symmetric stretching vibrational modes, which are
IR inactive.
(c) This is a nonlinear molecule. Therefore, there are 12−6 = 6 vibrational modes.
There are no IR inactive modes. In other words, all vibrational modes change
dipole moments and are IR active.
(d) There are 15 − 6 = 9 vibrational modes. All vibrational normal modes change
the dipole moment of the molecule. Therefore, there are no IR inactive modes.
4.7 A molecule has three vibrational degrees of freedom, and thus its vibration
involves three mass-weighted vibrational coordinates q1 , q2 , and q3 . The vibrational
potential energy is zero for q1 = q2 = q3 = 0 and the potential energy can be
expressed as a quadratic function of these coordinates. The mass-weighted Hessian
matrix for these coordinates is given by
⎛ ⎞
1
1 √ 0
⎜ 1 2 2
1
F̃ = ω2 ⎜
⎟
⎝2 2 1
. √ √ ⎟.
2 2 ⎠
1
0 √ 1
2 2

Answer the following questions.

(a) Express the total vibrational potential energy in terms of q1 , q2 , and q3 .
(b) Find out the three normal mode vibrational frequencies of the molecule
(expressed in terms of ω).
(c) Determine the three normal mode vectors.
Solution 4.7 Let us introduce uT1 = (1, 0, 0), uT2 = (0, 1, 0), and uT3 = (0, 0, 1),
where subscript T denotes transpose. Then, qT = (q1 , q2 , q3 ) = q1 uT1 + q2 uT2 +
q3 uT3 and q = q1 u1 + q2 u2 + q3 u3 .
128 4 Multidimensional Systems and Separation of Variables

(a) The vibrational potential energy can be expressed as follows:

1 T
V (q1 , q2 , q3 ) =
. q F̃q
2
ω2
 
q1 q2 q1 q2 q3 q2 q2 q3
= q12 + √ + √ + q22 + √ + √ + q32
2 2 2 2 2 2 2 2 2
ω 2 
1 1

2 2 2
= q1 + q2 + q3 + √ q1 q2 + √ q2 q3 .
2 2 2

(b) The normal mode frequencies can be obtained from the eigenvalues of F̃, which
can be obtained from the following equation:

1
1−λ √ 0
2 2   
1 1 3 1
. √ 1−λ √ = (1 − λ) −λ − λ = 0.
2 2 2 2 2 2
1
0 √ 1−λ
2 2

Thus, the solutions are λ = 1/2, 1, 3/2. Since the values of this multiplied by
ω2 are squares of the
√ normal mode
√ frequencies, the resulting three normal mode
frequencies are ω/ 2, ω, and ω 3/2, which will be denoted as ωa , ωb , and ωc
here.
(c) The normal modes can be determined by plugging the eigenvalues to the
eigenvalue equation. For λ = 1/2,
⎛ ⎞
1
1 √ 0 ⎛ ⎞ ⎛ ⎞
⎜ 1 2 2
1 ⎟ Q1 1
Q1
⎝2 2 1
.⎜ √ √ ⎟ ⎝ Q2 ⎠ = ⎝ Q2 ⎠ .
2 2⎠ 2
1 Q3 Q3
0 √ 1
2 2

√
Solving the above equation, we obtain Q1 = Q√ 3 = −Q2 / 2. Using the
√ condition, Q1 = 1/2, Q2 = −1/ 2, and Q3 = 1/2. Thus,
normalization
for ωa = ω/ 2, the normal mode is as follows:

1 1 1
Qa =
. u1 − √ u2 + u3 .
2 2 2

For λ = 1,
⎛ ⎞
1
1 √ 0 ⎛ ⎞ ⎛ ⎞
⎜ 1 2 2
1⎟ Q1 Q1
⎝2 2 1
.⎜ √ √⎟ ⎝ Q2 ⎠ = ⎝ Q2 ⎠.
2 2
⎠
1 Q3 Q3
0 √ 1
2 2
Problems 129

Solving the above equation, we obtain

√ Q1 = −Q3√and Q2 = 0. Using the
normalization condition, Q1 = 1/ 2 and Q3 = −1/ 2. Thus, for ωb = ω, the
normal mode is as follows:
1
Qb = √ (u1 − u3 ).
.
2

Finally, for λ = 3/2,

⎛ ⎞
1
1 √ 0 ⎛
Q1
⎞ ⎛
Q1
⎞
2 2
⎜ 1 1 ⎟⎝
⎟ 3
⎝2 2 1
.⎜ √ √ Q2 ⎠ = ⎝ Q2 ⎠ .
2 2⎠ 2
1 Q3 Q3
0 √ 1
2 2

√
Solving the above equation, we obtain Q1 = Q3 √= Q2 / 2. Using the normaliza-
√
tion condition, Q1 = Q3 = 1/2 and Q2 = 1/ 2. Thus, for ωc = ω 3/2, the
normal mode is as follows:
1 1 1
Qc =
. u1 + √ u2 + u3 .
2 2 2

Problems

4.8 A quantum state |ψ represents a quantum particle in three dimensional space,
and the corresponding wave function is

γ 3/4  γ  
r|ψ =
. exp − x 2 + y 2 + z2 + ik(x + y + z) ,
π 2

where γ is a constant real number. Find out expressions for r||p̂|2 |ψ and r|(r̂ ·
p̂ + p̂ · r̂)|ψ.
4.9 For a proton confined in a three dimensional box, for which one side has length
of 2.00 nm, and the remaining sides are 4.00 nm long, calculate the ground state,
first excited state, and the second excited energies. Determine the degeneracy for
each sate.
4.10 A quantum particle with unit mass (m = 1) is subject to the following
potential energy.

1 2 2
2 ω (y+ z2 ), 0 < x < a
V (r) =
. ,
∞, otherwise
130 4 Multidimensional Systems and Separation of Variables

Find out the general expression for the energy eigenvalue for this quantum particle.
Determine the relationship between a and ω for the first excited state to have
degeneracy of 3.
4.11 A quantum particle with unit mass (m = 1) is subject to the following
potential energy. V (r) = 2(x 2 + y 2 + 4z2 ). What is the ground state energy? What
is the second excited state energy and its degeneracy?
4.12 A quantum particle with unit mass (m = 1) is subject to the following
potential energy.

1 2 2
V (r) =
. ω (x + 4y 2 + 9z2 ),
2
The eigenstate is represented as |vx , vy , vz , where vx , vy , vz = 0, 1, 2, · · · .
(a) What is the ground state energy?
(b) What is the third excited state energy and its degeneracy? List all the corre-
sponding degenerate states.
4.13 A molecule has three vibrational degrees of freedom, and thus its vibration
involves three mass-weighted vibrational coordinates q1 , q2 , and q3 . The vibrational
potential energy is zero for q1 = q2 = q3 = 0 and the potential energy can be
expressed as a quadratic function of these coordinates. The mass-weighted Hessian
matrix for these coordinates is given by
⎛ ⎞
211
2
.F̃ = ω ⎝ 1 2 0 ⎠ .

102

Answer the following questions.

(a) Express the total vibrational potential energy in terms of q1 , q2 , and q3 .
(b) Find out the three normal mode vibrational frequencies of the molecule
(expressed in terms of ω).
(c) Determine the three normal mode vectors.
Chapter 5
Rotational States and Spectroscopy

What we observe is not nature itself but nature exposed to our

method of questioning.
– Werner Heisenberg

Abstract This chapter describes rotational motion in two and three dimensions and
solutions of time independent Schrödinger equations for corresponding rotational
Hamiltonians. Detailed description of coordinate transformation is provided. It
is shown that the periodic nature of rotational motion and the requirement that
wavefunctions are uniquely determined and remain bounded leads to quantization
of rotational energies. Well known spherical harmonics are shown to be eigen-
functions of angular momentum operators as well as three dimensional rotational
Hamiltonian. These solutions are then used for the description of rotational motion
of diatomic molecules as free rigid rotors. Important principles concerning pure
rotational transitions and those accompanying vibrational transitions are explained.
Corrections for rigid rotor states, incorporating the effects of centrifugal distortion
and ro-vibrational coupling, are provided as well.

Rotation is intrinsically periodic motion along a curved and closed loop in space,
and can be described best in coordinate systems different from Cartesian coordi-
nates. In two dimensions, this can be done by introducing an angle variable in polar
coordinate system. In three dimension, it takes consideration of two angle variables
in the spherical coordinate system.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 131
S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_5
132 5 Rotational States and Spectroscopy

5.1 Rotation in Two Dimensional Space

Let us consider a classical particle that moves with constant speed on a circle of
radius r centered at origin in a two dimensional xy-plane as shown in Fig. 5.1. The
classical Hamiltonian for such a motion is

p2 
 
1 2 2 

.HR = √ = (p x + p )
y  √ , (5.1)
2m r= x 2 +y 2
 2m r= x 2 +y 2


where .r = x 2 + y 2 in the subscript implies that it remains constant throughout
the motion. Without this constraint, the quantum mechanical Hamiltonian operator
corresponding to the above classical Hamiltonian is

1  2 
Ĥf =
. p̂x + p̂y2 , (5.2)
2m
which is in fact the Hamiltonian for free motion as the subscript f indicates. In order
to representthe rotational motion, it should be possible to impose the constraint of
fixed .r = x 2 + y 2 . For the specification of this constraint, it is important to use
an appropriate coordinate system as will become clear below.
Let us first denote the quantum mechanical rotational Hamiltonian as .ĤR and
start with the time independent Schrödinger equation in the Dirac notation.

ĤR |ψ = ER |ψ.

. (5.3)

The position representation of this is obtained by taking the inner product with
x, y| ≡ x| ⊗ y|, the bra representing the state with cartesian coordinates x and y.
.

Fig. 5.1 Illustration of a two

dimensional (in the xy-plane)
rotation (around the z axis)
5.1 Rotation in Two Dimensional Space 133

Namely, the Schrödinger equation for the waefunction .ψ(x, y) = x, y|ψ is given
by

h̄2 ∂2 ∂2
  
.x, y|ĤR |ψ = − + ψ(x, y) √

2m ∂x 2 ∂y 2 r= x 2 +y 2

= ER ψ(x, y) √ 2 , (5.4)

r= x +y 2

where .ER is the eigenvalue for the rotational energy to be determined.

Equation (5.4) appears to be complicated due to the constraint. However, convert-
ing .(x, y) to polar coordinates simplifies the mathematical problem significantly.
Let us consider polar coordinates in the two dimensional plane, .(r, φ), which are
defined by

. r= x2 + y2, . (5.5)
y
tan φ = . (5.6)
x
Given the above values, any cartesian coordinates can be determined by .x = r cos φ
and .y = r sin φ.
Employing the chain rule for differentiation, the partial derivatives with respect
x and y can be related to those of r and .φ as follows:

∂ ∂r ∂ ∂φ ∂
. = + ,. (5.7)
∂x ∂x ∂r ∂x ∂φ
∂ ∂r ∂ ∂φ ∂
= + . (5.8)
∂y ∂y ∂r ∂y ∂φ

From Eq. (5.5), we find that

∂r x x
. = = = cos φ, . (5.9)
∂x 2
x +y 2 r
∂r y y
= = = sin φ. (5.10)
∂y 2
x +y 2 r

From Eq. (5.6), we also find that

∂φ y
. sec2 φ = − 2 ,. (5.11)
∂x x
∂φ 1
sec2 φ = . (5.12)
∂y x
134 5 Rotational States and Spectroscopy

Note that .sec2 φ = 1 + tan2 φ = r 2 /x 2 . Thus, the above identities can be simplified
to
∂φ y sin φ
. =− 2 =− ,. (5.13)
∂x r r
∂φ x cos φ
= 2 = . (5.14)
∂y r r

Combining Eqs. (5.9), (5.10), (5.13), and (5.14) with Eqs. (5.7) and (5.8), we find
that
∂ ∂ sin φ ∂
. = cos φ − ,. (5.15)
∂x ∂r r ∂φ
∂ ∂ cos φ ∂
= sin φ + . (5.16)
∂y ∂r r ∂φ

The above two identities can be combined to express the sum of squared partial
derivatives in Eq. (5.4) as follows:

∂2 ∂2
  
∂ sin φ ∂ ∂ sin φ ∂
. + = cos φ − cos φ −
∂x 2 ∂y 2 ∂r r ∂φ ∂r r ∂φ
  
∂ cos φ ∂ ∂ cos φ ∂
+ sin φ + sin φ +
∂r r ∂φ ∂r r ∂φ
∂ 2 sin φ ∂

∂

= cos2 φ 2 − cos φ
∂r r ∂φ ∂r
   
∂ sin φ ∂ sin φ ∂ sin φ ∂
− cos φ +
∂r r ∂φ r ∂φ r ∂φ
∂2
 
cos φ ∂ ∂
+ sin2 φ 2 + sin φ
∂r r ∂φ ∂r
   
∂ cos φ ∂ cos φ ∂ cos φ ∂
+ sin φ +
∂r r ∂φ r ∂φ r ∂φ
∂2 1 ∂
= (cos2 φ + sin2 φ) + (sin2 φ + cos2 φ)
∂r 2 r ∂r
1   ∂2
+ 2 sin2 φ + cos2 φ
r ∂φ 2
∂2 1 ∂ 1 ∂2
= + +
∂r 2 r ∂r r 2 ∂φ 2
   
1 ∂ ∂ 1 ∂ 1 ∂
= r + . (5.17)
r ∂r ∂r r ∂φ r ∂φ
5.1 Rotation in Two Dimensional Space 135

The separation of differentials involving r and .φ in Eq. (5.17) now makes it easy
to impose constraint of constant r, which simply amounts to assuming that .∂/∂r =
0. Thus, Eq. (5.4) simplifies to

h̄2 1 ∂ 2 
− ψ(x, y) = ER ψ(x, y)|r=√x 2 +y 2 . (5.18)

.
2m r 2 ∂φ 2  √

r= x 2 +y 2

Now, let us define

. ψr (φ) = ψ(x, y)|r=√x 2 +y 2 , (5.19)

and introduce .I = mr 2 , which is the moment of inertia for the rotational motion.
Then, Eq. (5.18) becomes

h̄2 ∂ 2
. − ψr (φ) = ER ψr (φ). (5.20)
2I ∂φ 2

A general solution of the above equation can be expressed as

ψr (φ) = N eimφ ,
. (5.21)

which results in the following relationship:

h̄2 ∂ 2 h̄2 m2
. − 2
ψr (φ) = ψr (φ) = ER ψr (φ). (5.22)
2I ∂φ 2I

Equation (5.20) represents free motion along the .φ coordinate. However, there is
major difference of this rotational motion from that in infinite space. For the former,
the domain of .φ representing different physical space is confined only to a range of
.2π . In other words, the periodic boundary condition, .ψr (φ) = ψr (φ + 2π ), should

be satisfied. This condition implies that

eim2π = 1,
. (5.23)

which can be satisfied only for integer values of .m = 0, ±1, ±2, · · · . Therefore, we
obtain the following expression:

Eigenvalue of Two-Dimensional Rotation

h̄2 m2
ER,m =
. , m = 0, ±1, ±2, · · · . (5.24)
2I
136 5 Rotational States and Spectroscopy

The normalization constant of the rotational wavefunction in Eq. (5.21) can be

determined using the following condition:

2π
. dφ|ψr (φ)|2 = |N|2 2π = 1. (5.25)
0
√
A choice of .N = 1/ 2π satisfies the above condition. Thus, the normalized
eigenfunction of the 2-dimensional rotation with eigenvalue .mh̄ can be expressed
as follows:

Normalized Eigenfunction of Two-Dimensional Rotation

1
ψm (φ) = √ eimφ , m = 0, ±1, ±2, · · ·
. (5.26)
2π

5.2 Rotation in Three Dimensional Space

The approach of obtaining rotational kinetic energy term directly from the Cartesian
coordinate system for two-dimension can also be used for rotation in three dimen-
sional space. For this, the spherical coordinate system, .(r, θ, φ) (see Fig. 5.2), need
to be used. This will be demonstrated in detail in the next section, when we describe
angular momentum operators.

Fig. 5.2 Illustration of

spherical coordinates .(r, θ, φ)
5.2 Rotation in Three Dimensional Space 137

Here, let us use a well-established formula for converting the Laplacian (.∇ 2 ) in
the Cartesian coordinate system to that in a curvilinear coordinate system, where
coordinates are defined along a system of three curved lines that remain orthogonal
at each point. First, this can be illustrated for the two dimensional case given by Eq.
(5.17).
For the polar coordinate system specified by .(r, φ), the multiplicative factors used
to convert the change in the values of coordinates to appropriate length units, so
called the metric factors along r and .φ, are respectively defined as .hr = 1 and .hφ =
r. Then, the Laplacian in the two dimensional space, Eq. (5.17), can be expressed as

∂2 ∂2
   
2 1 ∂ hφ ∂ 1 ∂ hr ∂
.∇ = + = + . (5.27)
∂x 2 ∂y 2 hr hφ ∂r hr ∂r hr hφ ∂φ hφ ∂φ

In fact, this is a special case of much more general expression that is valid in
any dimension. For example, in three dimensional general coordinate system of
.(ξ1 , ξ2 , ξ3 ) with metric factor along the three directions, .h1 , .h2 , and .h3 , which

themselves can be considered as functions of .ξ1 , .ξ2 , and .ξ3 , there is a general
expression [30] for the Laplacian as follows:

∂2 ∂2 ∂2
∇2 =
. + +
∂x 2 ∂y 2 ∂z2
     
1 ∂ h2 h3 ∂ ∂ h1 h3 ∂ ∂ h1 h2 ∂
= + + .
h1 h2 h3 ∂ξ1 h1 ∂ξ1 ∂ξ2 h2 ∂ξ2 ∂ξ3 h3 ∂ξ3
(5.28)

In the above expression, .(1/ hi )(∂/∂ξi ) for each of .i = 1, 2, 3 corre-

sponds to the component of the mathematical gradient vector operation and
.(1/ h1 h2 h3 )(∂/∂ξi )hj hk with .j = k = i corresponds to the divergence along

each direction.
For the case of cartesian coordinate system, .hx = hy = hz = 1, which results
in the simplest expression for .∇ 2 . However, for spherical coordinates .(r, θ, φ) as
shown in Fig. 5.2, .h1 = 1, .h2 = r, and .h3 = r sin θ . Therefore,
     
2 1 ∂ 2 ∂ ∂ ∂ ∂ 1 ∂
.∇ = r sin θ + sin θ +
r 2 sin θ ∂r ∂r ∂θ ∂θ ∂φ sin θ ∂θ
∂2
   
1 ∂ ∂ 1 ∂ ∂ 1
= 2 r2 + 2 sin θ + . (5.29)
r ∂r ∂r r sin θ ∂θ ∂θ r 2 sin2 θ ∂φ 2

For any function f , one can show that the derivative with respect to r above can be
expressed as follows:

∂2 1 ∂2
   
1 ∂ 2 ∂ 2 ∂ 1 ∂ 1 ∂ ∂
. r f = f + 2f = f+ r f = rf.
r 2 ∂r ∂r r ∂r ∂r r ∂r r ∂r ∂r r ∂r 2
(5.30)
138 5 Rotational States and Spectroscopy

Employing this identity, Eq. (5.29) can often be expressed as

1 ∂2 ∂2
 
2 1 ∂ ∂ 1
.∇ = r+ 2 sin θ + . (5.31)
r ∂r 2 r sin θ ∂θ ∂θ r 2 sin2 θ ∂φ 2

The above equation, or Eq. (5.29), now makes it easy to impose the constraint of
fixed r and to represent the rotational motion of a particle with fixed r. In other
words,

−h̄2 2  h̄2 1 ∂2
   
1 ∂ ∂
. − ∇  =− sin θ +
2m  2mr 2 sin θ ∂θ ∂θ sin2 θ ∂φ 2
r

h̄2 1 ∂2
   
1 ∂ ∂
=− sin θ + , (5.32)
2I sin θ ∂θ ∂θ sin2 θ ∂φ 2

where .I = mr 2 has been defined as the moment of inertia for the case of two
dimensional rotation and remains the same.
Let us denote the eigenfunction of a general three dimensional rotation as
.Y (θ, φ). Then, the Schrödinger equation for this rotational motion is given by

h̄2 1 ∂2
   
1 ∂ ∂
. − sin θ + Y (θ, φ) = EY (θ, φ). (5.33)
2I sin θ ∂θ ∂θ sin2 θ ∂φ 2

Dividing the above equation with .−h̄2 /(2I ) and moving all terms to the lefthand
side, we find that the above equation can be expressed as

1 ∂2
 
1 ∂ ∂ 2I E
. sin θ Y (θ, φ) + 2 ∂φ 2
Y (θ, φ) + 2 Y (θ, φ) = 0. (5.34)
sin θ ∂θ ∂θ sin θ h̄
√
Let us assume that .Y (θ, φ) = (θ ) m (φ), where . m (φ) = eimφ / 2π , with .m =
0, ±1, ±2. · · · . Then,

d2
 
1 d d 1
. m (φ) sin θ (θ ) + 2
(θ ) 2 m (φ)
sin θ dθ dθ sin θ dφ
2I E
+ (θ ) m (φ) = 0. (5.35)
h̄2

d 2
Since . dφ 2 m (φ) = −m2 m (φ), the above equation is simplified to

m2
 
1 d d 2I E
. m (φ) sin θ (θ ) − 2
(θ ) + 2 (θ ) = 0. (5.36)
sin θ dθ dθ sin θ h̄
5.2 Rotation in Three Dimensional Space 139

This implies that

m2
   
1 d d
. sin θ (θ ) + β − (θ ) = 0, (5.37)
sin θ dθ dθ sin2 θ

where
2I E
β=
. . (5.38)
h̄2

Equation (5.37) is called associated Legendre equation, for which well-known

solutions exist. Appendix of this chapter provides a detailed consideration of this
equation and show that well-defined physical solutions exist for the following
conditions:

. β = J (J + 1), J = 0, 1, 2, · · · , . (5.39)
J ≥ |m|. (5.40)

The resulting solution for given J and m is called associated Legendre function
|m|
and is denoted as .PJ (cos θ ), for which detailed derivation is also provided in the
Appendix.
|m|
Table 5.1 provides expressions for .PJ (cos θ ) for small values of J and m.
Plots of some Legendre polynomials (.m = 0) are shown in Fig. 5.3, and plots of
associated Legendre polynomials for .J = 4 are shown in Fig. 5.4.

Table 5.1 Examples of associated Legendre functions

J m .PJ
m (x) .PJ
m (cos θ)

0 0 1 1
1 0 x .cos θ

1 1 2 1/2
.(1 − x ) .sin θ
1 2 1 2
2 0 . (3x − 1) − 1)
2 .
2 (3 cos θ
2 1 2 1/2
.3x(1 − x ) .3 cos θsin θ
2
2 2 .3(1− x2) .3 sin θ
1 3 1 3
3 0 . (5x − 3x) . (5 cos θ − 3 cos θ)
2 2
3 2 3
3 1 .
2 (5x − 1)(1 − x 2 )1/2 .
2
2 (5 cos θ − 1) sin θ
3 2 .15x(1 − x2) .15 cos θsin2 θ
2 3/2 3
3 3 .15(1 − x ) .15 sin θ
1 4 2 1 4 2
4 0 . (35x − 30x + 3) . (35 cos θ − 30 cos θ + 3)
8 8
5 3 2 1/2 5 3
4 1 .
2 (7x − 3x)(1 − x ) .
2 (7 cos θ − 3 cos θ) sin θ
15 2 2 15 2 2
4 2 . (7x − 1)(1 − x ) . (7 cos θ − 1) sin θ
2 2
2 3/2 3
4 3 .105x(1 − x ) .105 cos θ sin θ
4
4 4 .105(1 − x 2 )2 .105 sin θ
140 5 Rotational States and Spectroscopy

Fig. 5.3 Examples of four Legendre polynomials .PJ0 (x) = PJ0 (cos θ) for J =0, 1, 2, and 3.
The upper panel is shown with respect to x and the lowe panel is shown with respect to .θ.
Normalizations for these plots were chosen so that .PJ0 (x) = 1

Fig. 5.4 Examples of four associated Legendre polynomials for .J = 4, .P4m (x) = P4m (cos θ) with
.m = 0, 1, 2, and 3. The upper panel is with respect to .x = cos θ and the lower panel is with respect
to .θ
5.2 Rotation in Three Dimensional Space 141

|m|
Each of .PJ (θ ), as can be seen from the examples in Table 5.1, is not
normalized. Including normalization factor such that integration of the function over
.d cos θ is equal to 1 and also now employing the appropriate subscript denoting the

quantum numbers J and m, we can define

 1/2
m 2J + 1 (J − |m|)! |m|
.J (θ ) = PJ (cos θ ). (5.41)
2 (J + |m|)!

Combining this with . m (φ), one can define eigenfunctions of the rotational
Hamiltonian known as spherical Harmonics as follows.

Spherical Harmonics
 1/2
m m+|m| 2J + 1 (J − |m|)! |m|
.YJ (θ, φ) =i PJ (cos θ )eimφ , (5.42)
4π (J + |m|)!

where .i m+|m| is the most commonly used phase factor although different
phase factors are also used for convenience or other reasons. Expressions for
some of these functions are provided in Table 5.2. Figure 5.5 provides plots
of real and imaginary parts of these spherical harmonics for .J = 0, 1, and 2.

Table 5.2 Expressions for J m .YJ

m (θ, φ)
spherical harmonics for  1/2
.J = 0, · · · , 3 1
0 . 0 .
4π
 1/2
3
1 . 0 .
4π cos θ
 1/2
3
1 .±1 .∓
8π sin θe±iφ
 1/2
5
2 . 0 .
16π (3 cos2 θ − 1)
 1/2
15
2 .±1 .∓
8π sin θ cos θe±iφ
 1/2
15
2 .±2 .
32π sin2 θe±2iφ
 1/2
7
3 . 0 .
16π (5 cos3 θ − 3 cos θ)
 1/2
21
3 .±1 .∓
64π sin θ(5 cos2 θ − 1)e±iφ
 1/2
105
3 .±2 .
32π sin2 θ cos θe±2iφ
 1/2
35
3 .±3 .∓
64π sin3 θe±3iφ
142 5 Rotational States and Spectroscopy

Fig. 5.5 Examples of (a) real and (b) imaginary parts of spherical harmonics for .J = 0, 1 , and 2

Combining Eqs. (5.38) and (5.39), we obtain the eigenvalue and eigenfunction of
the three dimensional rotation as summarized below.

Eigenvalue and Eigenfunction of Three-Dimensional Rotation

The eigenvalue for a three-dimensional rotational motion with moment of
inertia I is

h̄2
EJ =
. J (J + 1), (5.43)
2I
(continued)
5.3 Angular Momentum Operators 143

where .J = 0, 1, 2, · · · , and the corresponding eigenfunction, spherical

harmonics, for each J and .m = 0, · · · , ±J , satisfy the following time
independent Schrödinger equation:

h̄2 1 ∂2 h̄2
   
1 ∂ ∂ m
.− sin θ + 2 Y J (θ, φ) = J (J +1)YJm (θ, φ).
2I sin θ ∂θ ∂θ sin θ ∂φ 2 2I
(5.44)

In addition, since the dependence of .YJm (θ, φ) on .φ is only through . m (φ) =

√
eimφ / 2π , it is easy to show that

h̄ ∂ m
. Y (θ, φ) = h̄mYJm (θ, φ). (5.45)
i ∂φ J

Note that, for a given value of J , there are .2J +1 possible values of m because

m = −J, · · · , J.
. (5.46)

The spherical harmonics defined by Eq. (5.42) satisfy the following normaliza-
tion and orthogonality (orthonormality) condition:

π 2π
. dθ sin θ d φYJm (θ, φ)∗ YJm′′ (θ, φ) = δJ J ′ δmm′ . (5.47)
0 0

The above identity in fact implies that .YJm (θ, φ) form a complete orthonormal basis
for any well-behaving functions depending on .θ and .φ. Namely, any function of .θ
and .φ can be expressed as a linear combination of these as follows:

f (θ, φ) =
. CJ,m YJm (θ, φ), (5.48)
J,m

where .CJ,m ’s can be determined uniquely using Eq. (5.47).

5.3 Angular Momentum Operators

Each of the spherical harmonics introduced in the previous section is an eigenfunc-

tion of the rotational part of the kinetic energy operator. In classical mechanics, such
rotational energy can be expressed in terms of angular momentum. For example,
consider a particle with mass m rotating with a constant speed v at a fixed distance
144 5 Rotational States and Spectroscopy

r from the origin. By introducing .L = mrv, the kinetic energy of this particle can
be expressed as

mv 2 L2 L2
. = = . (5.49)
2 2mr 2 2I
In fact, the last expression for the rotational kinetic energy in the above equation is
much more general and can be shown to be true if it is defined as the magnitude of
the following angular momentum vector:

L = r × p = Lx ex + Ly ey + Lz ez
.

= (ypz − zpy )ex + (zpx − xpz )ey + (xpy − ypx )ez . (5.50)

The square of the magnitude of this vector can be shown to be

|L|2 = L2x + L2y + L2z = r 2 p⊥

.
2
, (5.51)

where .p⊥ is the projection of a momentum .p vector to a plane perpendicular to .r.

In other words, classically, kinetic energy can be decomposed as

|p|2 p2 + p⊥
2 p2 2
r 2 p⊥ p2 L2
T =
. = = + = + , (5.52)
2m 2m 2m 2mr 2 2m 2I
where .p is the component of the momentum along the direction of .r. Thus,
T can always be considered as the sum of the kinetic energy for a linear one-
dimensional motion along the direction of .r and rotational kinetic energy, which
can be represented by the magnitude of the angular momentum vector .L as

L2 1  2 
TR =
. = Lx + L2y + L2z . (5.53)
2I 2I
Based on the above classical expression for the rotational kinetic energy, it is
possible to define quantum mechanical Hamiltonian operator for three dimensional
rotation as summarized below.

Hamiltonian for Three Dimensional Rotation

The Hamiltonian operator for a rotor with moment of inertia I can be
expressed as

L̂2 1  2 
ĤR =
. = L̂x + L̂2y + L̂2z , (5.54)
2I 2I
(continued)
5.3 Angular Momentum Operators 145

where .L̂k ’s, with .k = x, y, and z, are cartesian components of angular

momentum operators defined as

. L̂x = ŷ p̂z − ẑp̂y , . (5.55)

L̂y = ẑp̂x − x̂ p̂z , . (5.56)
L̂z = x̂ p̂y − ŷ p̂x . (5.57)

The three cartesian components of the angular momentum operators defined

above do not commute with each other, unlike the linear momentum operators.
Instead, the following relations hold.

. [L̂x , L̂y ] = i h̄L̂z , . (5.58)

[L̂y , L̂z ] = i h̄L̂x , . (5.59)
[L̂z , L̂x ] = i h̄L̂y . (5.60)

It is straightforward to prove the above identities for commutators of angular

momentum operators using the identity of Eq. (2.57) or calculating the commutator
directly. For example,

.[L̂x , L̂y ] = [ŷ p̂z − ẑp̂y , ẑp̂x − x̂ p̂z ]

= [ŷ p̂z , ẑp̂x ] − [ẑp̂y , ẑp̂x ] − [ŷ p̂z , x̂ p̂z ] + [ẑp̂y , x̂ p̂z ]
= ŷ[p̂z , ẑ]p̂x + x̂[ẑ, p̂z ]p̂y
= i h̄(x̂ p̂y − ŷ p̂x ) = i h̄L̂z . (5.61)

On the other hand, using the identities used above once again, one can prove that
L̂2 commute with individual components, .L̂x , .L̂y , and .L̂z . For example,
.

[L̂2 , L̂z ] = [L̂2x + L̂2y + L̂2z , L̂z ]

.

= L̂x [L̂x , L̂z ] + [L̂x , L̂z ]L̂x + L̂y [L̂y , L̂z ] + [L̂y , L̂z ]L̂y
= L̂x (−i h̄L̂y ) + (−i h̄L̂y )L̂x + L̂y (i h̄L̂x ) + (i h̄L̂x )L̂y = 0. (5.62)
146 5 Rotational States and Spectroscopy

Results of application of the angular momentum operator components to the

position bra .r| can be calculated by using those for position and momentum
components as follows:
 
h̄ ∂ ∂
. r|L̂x = y −z r|, . (5.63)
i ∂z ∂y
 
h̄ ∂ ∂
r|L̂y = z −x r|, . (5.64)
i ∂x ∂z
 
h̄ ∂ ∂
r|L̂z = x −y r|. (5.65)
i ∂y ∂x

In order to express these in terms of spherical coordinates, let us now consider

conversion of Cartesian coordinates into spherical coordinates in more detail. First,
the Cartesian components can be expressed in terms of spherical coordinates as
follows:

. x = r sin θ cos φ, . (5.66)

y = r sin θ sin φ, . (5.67)
z = r cos θ. (5.68)

From these, one can also obtain expressions for spherical coordinates in terms of
Cartesian coordinates as follows:

. r = x 2 + y 2 + z2 , . (5.69)

x2 + y2
tan θ = ,. (5.70)
z
y
tan φ = . (5.71)
x
The above expression can be used to obtain the following relations between partial
derivatives with respect to Cartesian coordinates and those with respect to spherical
coordinates. First, the partial derivatives of r are expressed as

∂r x r sin θ cos φ
. = = = sin θ cos φ, . (5.72)
∂x x 2 + y 2 + z2 r
∂r y r sin θ sin φ
= = = sin θ sin φ, . (5.73)
∂y x 2 + y 2 + z2 r
∂r z r cos θ
= = = cos θ. (5.74)
∂z x 2 + y 2 + z2 r
5.3 Angular Momentum Operators 147

The partial derivatives of .θ , which can be obtained from the fact that .d tan θ =
sec2 θ dθ , are given by

∂θ r sin θ cos φ cos θ cos φ

. = cos2 θ = ,. (5.75)
∂x r sin θ r cos θ r
∂θ r sin θ sin φ cos θ sin φ
= cos2 θ = ,. (5.76)
∂y r sin θ r cos θ r
∂θ r sin θ sin θ
= − cos2 θ 2 =− . (5.77)
∂z r cos2 θ r

Finally, the partial derivatives of .φ, which can be obtained from the fact that
d tan φ = sec2 φdφ, can be expressed as
.

∂φ r sin θ sin φ sin φ

. = − cos2 φ 2
=− ,. (5.78)
∂x 2
r sin θ cos φ 2 r sin θ
∂φ 1 cos φ
= cos2 φ = ,. (5.79)
∂y r sin θ cos φ r sin θ
∂φ
= 0. (5.80)
∂z

The above expressions can be used to convert the partial derivatives with respect to
x, y, and z to those with respect to r, .θ , and .φ. The resulting expressions can then
be used to obtain expressions for the components of angular momentum operator.
The expression for the angular momentum along the z-coordinate has the
simplest form and can be shown to be
 
h̄ ∂ ∂
.r|L̂z = x −y r|
i ∂y ∂x
  
h̄ ∂ cos θ sin φ ∂ cos φ ∂
= r sin θ cos φ sin θ sin φ + +
i ∂r r ∂θ r sin θ ∂φ
 
∂ cos θ cos φ ∂ sin φ ∂
−r sin θ sin φ sin θ cos φ + −
∂r r ∂θ r sin θ ∂φ
h̄ ∂
= r|. (5.81)
i ∂φ

Considering the position state in spherical coordinate system, one can express .r| as
follows:

r| = r| ⊗ θ | ⊗ φ|.

. (5.82)

The position ket .|r can also be written in a similar way. Note that the ranges of these
spherical coordinate variables are different from those of cartesian coordinates, and
148 5 Rotational States and Spectroscopy

are given by .0 ≤ r, .0 ≤ θ ≤ π , and .0 ≤ φ < 2π . Employing Eq. (5.82) in Eq. (5.81)

and using the fact that .r| and .θ | remain invariant after application of .L̂z , we find
that Eq. (5.81) is equivalent to the following expression:

h̄ ∂
φ|L̂z =
. φ|. (5.83)
i ∂φ

Similarly, .L̂x and .L̂y can be shown to be

 
h̄ ∂ ∂
θ, φ|L̂x =
. y −z θ, φ|
i ∂z ∂y
 
∂ ∂
= i h̄ sin φ + cot θ cos φ θ, φ|, . (5.84)
∂θ ∂φ
 
h̄ ∂ ∂
θ, φ|L̂y = z −x θ, φ|
i ∂x ∂z
 
∂ ∂
= i h̄ − cos φ + cot θ sin φ θ, φ|, (5.85)
∂θ ∂φ

where .θ, φ| = θ | ⊗ φ| and the fact that .L̂x and .L̂y do not affect .r| has been used.
Summing the squares of the above two operators, we find that
 2
2 ∂ ∂
.θ, φ|(L̂x + L̂2y ) = −h̄ 2
sin φ + cot θ cos φ
∂θ ∂φ
 
∂ 2

∂
+ − cos φ + cot θ sin φ θ, φ|
∂θ ∂φ
 2
∂2

2 ∂ ∂ 2
= −h̄ + cot θ + cot θ 2 θ, φ|. (5.86)
∂θ 2 ∂θ ∂φ

Summing this with the square of .L̂z , we find that

θ, φ|L̂2 = r|(L̂2x + L̂2y + L̂2z )

.
 
2 ∂2 ∂ cos2 θ + sin2 θ ∂ 2
= −h̄ + cot θ + θ, φ|
∂θ 2 ∂θ sin2 θ ∂φ 2

1 ∂2
 
2 1 ∂ ∂
= −h̄ sin θ + θ, φ|. (5.87)
sin θ ∂θ ∂θ sin2 θ ∂φ 2
5.3 Angular Momentum Operators 149

Now let us consider application of .L̂2 to an arbitrary quantum state .|ψ as follows:

1 ∂2
 
1 ∂ ∂
.θ, φ|L̂2 |ψ = −h̄2 sin θ + θ, φ|ψ. (5.88)
sin θ ∂θ ∂θ sin2 θ ∂φ 2

Note that this is the same as that representing the rotational kinetic energy except
for the factor of .1/(2I ). Therefore, .YJm (θ, φ) is an eigenfunction of the above
differential equation as well. Thus, the spherical harmonics are eigenfunctions of
squared angular momentum operator as summarized below.

Rotational Eigenstates and Eigenvalues

Let us define .|J, m such that

YJm (θ, φ) = θ, φ|J, m.

. (5.89)

Then, Eqs. (5.44) and (5.45) imply that

. L̂2 |J, m = h̄2 J (J + 1)|J.m, . (5.90)

L̂z |J, m = h̄m|J, m, (5.91)

where .J = 0, 1, 2, · · · and m are integers such that .|m| ≤ J . Note that the
reason why .|J, m can be a simultaneous eigenstate of .L̂2 and .L̂z is because
the two commute with each other.

So far, we have considered rotation of a particle around origin. In fact, the same
approach can be used to describe the rotation of a diatomic molecule around its
center of mass. For this, let us consider the Hamiltonian operator for the two particle
system, Eq. (4.83), in the absence of an external potential and removing the kinetic
energy term for the center of mass. Thus, let us consider the following Hamiltonian
operator:

p̂2
Ĥrel =
. + Vrel (r̂), (5.92)
2μ

where .μ = m1 m2 /(m1 + m2 ) is the reduced mass and the fact that the potential
energy is a function of r only has been used. Applying this operator on the position
bra .r| = r, θ, φ| on the lefthand side, we find that
150 5 Rotational States and Spectroscopy

 
h̄2 2
r, θ, φ|Ĥrel
. = − ∇ + Vrel (r) r, θ, φ|.
2μ
 
h̄2 1 ∂ 2
= − r + Vrel (r) r, θ, φ|
2μ r ∂r 2

h̄2 1 ∂2
   
1 ∂ ∂
− sin θ + r, θ, φ|. (5.93)
2μr 2 sin θ ∂θ ∂θ sin2 θ ∂φ 2

In the second equality of the above equation, the first component involving deriva-
tives of r represents vibration motion along the internuclear direction, whereas the
second part represents a rotational motion. Note that this part is the same as the
rotational Hamiltonian for a particle with effective mass .μ. Thus, all the results
provided so far for three dimensional rotation of a particle around origin can also be
used to describe the quantum states representing the rotation of a diatomic molecule
around its center of mass given that the following expression is used for the moment
of inertia:

I = μr 2 .
. (5.94)

5.4 Spectroscopy of Rotational Transitions for Diatomic

Molecules

We will consider three major transitions involving rotational motions of diatomic

molecules. The first is pure rotational transition normally known as microwave
spectroscopy. The second is rotational Raman spectroscopy. Finally, we will con-
sider ro-vibrational transitions, where rotational transitions accompany vibrational
transitions.

5.4.1 Microwave Spectroscopy

Microwave spectroscopy in general refers to the detection of transitions of rotational

states due to linear interaction of the molecular dipole with the electric field
component of light. The reason why this is called “microwave” is because the
energy level spacings involved are typically in the microwave region. The transition
in rotational states occurs because the dipole operator, if not zero, does not commute
with the rotational Hamiltonian. Selection rules for this can be understood by
considering how different rotational states are coupled by the dipole operator.
5.4 Spectroscopy of Rotational Transitions for Diatomic Molecules 151

Selection Rules for Microwave Spectroscopy of Diatomic Molecules

• If a molecule does not have a permanent dipole, light cannot distinguish
different rotational state at the level of the first order interaction. Thus,
an important gross selection rule for microwave spectroscopy is that the
molecule has to be polar to be microwave active.
• The rotational property of the permanent dipole moment, which is pro-
portional to the internuclear distance vector, has the same symmetry
property as a three dimensional position vector. Employing the properties
of spherical harmonics, it is possible to show that such dipole operator
can couple only neighboring rotational quantum numbers. In other words,
the following selection rule applies for the microwave spectroscopy of
diatomic molecules:

. J = ±1. (5.95)

Microwave spectroscopy is commonly described in terms of rotational constant

B defined by the following relationship:

h̄2
EJ =
. J (J + 1) = BJ (J + 1), (5.96)
2I
where the second equality serves as the definition of B. Typically, the rotational
energy is reported in the unit of wavenumber (.cm−1 ) and the following rotational
term is often used:

EJ h̄2
G(J ) =
. = J (J + 1) = B̃J (J + 1). (5.97)
hc 2I hc
The microwave spectroscopy is in general absorption spectroscopy. Thus, the
transition involved is .J → J + 1, for which the energy difference is given by

. E = B(J + 1)(J + 2) − BJ (J + 1) = 2B(J + 1). (5.98)

The corresponding wavenumber is given by

E
. ν̃ = = 2B̃(J + 1). (5.99)
hc
As Eq. (5.98) or (5.99) shows, the rotational transition energy or wavenumber
increases linearly with the initial total angular momentum quantum number J . This
is in contrast to the vibrational transition of a harmonic oscillator where there is
152 5 Rotational States and Spectroscopy

only one transition energy, within the harmonic oscillator approximation. For this
reason, the microwave spectroscopy appears as a packet of transition lines rather
than a single or few transitions.
Since the interaction strength between the dipole and the electric field does not
change significantly with the rotational state of the molecule, the intensity of each
rotational transition is proportional to the relative population of molecule in the
specific initial rotational state with quantum number J . For a given rotational state
with J , there are .2J + 1 degenerate states because any of .mJ = −J, · · · , J is
possible. In other words, a given angular momentum quantum number J has .2J +
1 degeneracy. Therefore, the probability of having the molecule in the J state is
proportional to this degeneracy multiplied by the Boltzmann factor as follows:
 
+1)
(2J + 1) exp − BJk(J
B T
.pJ =  . (5.100)
∞ BJ (J +1)
J =0 (2J + 1) exp − kB T

5.4.2 Rotational Raman Spectroscopy

Rotational Raman spectroscopy probes transition of rotational states due to the

interaction of the polarizability of a molecule with light, which is second order with
respect to the electric field amplitude of the light. Selection rules for the rotational
Raman spectroscopy can be understood considering this aspect.

Selection Rules for Rotational Raman Spectroscopy

• Any molecule that changes its polarizability with rotation is active in the
rotational Raman spectroscopy. This means that the molecule should have
anisotropic polarizability, which is always the case for diatomic molecules.
• Since polarizability is independent of the sign of rotation, it is easy to
see that it is an even function of angles and is invariant after rotation
by .180o degrees. This symmetry property makes the selection rule for
rotational Raman different from the microwave spectroscopy and results
in the following selection rule:

. J = ±2. (5.101)

• The positive case is Stokes Raman transition and the negative case is anti-
Stokes Raman transition. For the Stokes Raman transition from J to .J + 2,
the magnitude of the transition energy (in wave number) is

. ν̃ = B̃ {(J + 2)(J + 3) − J (J + 1)} = 2B̃(2J + 3). (5.102)

5.4 Spectroscopy of Rotational Transitions for Diatomic Molecules 153

5.4.3 Ro-Vibrational Transition

For the understanding of ro-vibrational transition, it is necessary to consider the

diatomic molecular Hamiltonian given by Eq. (5.92) in more detail. For .|ψE , an
eigenstate of this Hamiltonian with eigenvalue E, applying the Hamiltonian for the
relative motion part and then taking inner product with the position bra, we find that
 
h̄2 1 ∂ 2
.r, θ, φ|Ĥrel |ψE  = − r + Vrel (r)
2μ r ∂r 2

h̄2 1 ∂2
   
1 ∂ ∂
− sin θ + ψE (r, θ, φ)
2μr 2 sin θ ∂θ ∂θ sin2 θ ∂φ 2
= EψE (r, θ, φ), (5.103)

where .ψE (r, θ, φ) = r, θ, φ|ψE  and Eq. (5.93) has been used. Let us assume that

1
ψE (r, θ, φ) =
. ψv (r)YJ M (θ, φ). (5.104)
r
Inserting this expression into Eq. (5.103), and factoring out the common function
YJ M (θ, φ), we find that
.

 
h̄2 d 2 eff
. − + VJ (r) ψv (r) = Eψv (r), (5.105)
2μ dr 2

where

eff h̄2 J (J + 1)
VJ
. (r) = Vrel (r) + . (5.106)
2μr 2

Let .re be the equilibrium internuclear distance of the diatomic molecule. If the
internuclear potential energy is strong enough, the deviation, .s = r − re , is very
small compared to .re . Indeed, this is the case for most diatomic molecules consisting
of light atoms. Then, the moment of inertia in this case can be approximated as
2 2
.I = μr ≈ μre in the above equation and

eff
. VJ (r) ≈ Vrel (r) + Be J (J + 1), (5.107)

where .Be = h̄2 /(2Ie ) = h̄2 /(2μre2 ). Let us also assume that .Vrel (r) can be
approximated as a parabolic potential around .re with curvature .Vrel ′′ (r ) = μω2 .
e
Thus .Vrel (r) ≈ Vrel (re ) + μωe2 (r − re )2 /2. Then, Eq. (5.108) can be approximated
as
154 5 Rotational States and Spectroscopy

 
h̄2 d 2 1 2 2
. − + μωe s + Vrel (re ) + Be J (J + 1) ψv (re + s) = Eψv (re + s),
2μ ds 2 2
(5.108)

where we have introduced a new coordinate .s = r − re .

Equation (5.108) is the same as the Schrödinger equation for the harmonic
oscillator except for the constant terms and the addition of rotational energy, which
is independent of r. Therefore, the corresponding eigenfunction is the same as that
for the harmonic oscillator as follows:
2 /2 √
ψv (re + s) = Nv e−γ s
. Hv ( γ s), (5.109)
√ √
where .γ = μωe /h̄, .Nv is the normalization constant, and .Hv ( γ s) is the Hermite
polynomial. On the other hand, the eigenvalue is
 
1
Ev,J = h̄ωe v +
. + Be J (J + 1) + Vrel (re ). (5.110)
2

The above expression shows that the eigenstate of a diatomic molecule, in its
simplest approximation of .I ≈ Ie , can be viewed as the direct product of the
vibrational and rotational states, namely, .|v ⊗ |J, m. Now, note that the dipole
vector is given by
 
D = qr = q(re + s) ex sin θ cos φ + ey sin θ sin φ + ez cos θ ,
. (5.111)

where q is the partial nuclear charge of the diatomic molecule and .ex , .ey , and .ez are
unit vectors along x, y, and z coordinates. Thus, the measurement of dipole moment
in fact involves measuring both the internuclear distance and angle of the molecule.
If the energy of light is small enough not to excite any vibrational motion, it is
possible that only rotational states are altered. This is the case of the microwave
spectroscopy. However, for the case of IR spectroscopy that involves transition of
vibrational energies, it is inevitable that rotational states change as well because the
energy for rotational energy is readily available and there is nothing that prevents
the rotational transition from accompanying the vibrational transition. The selection
rule governing the rotational transition in this case is the same as the microwave
spectroscopy because the same dipole operator is involved. Since wavenumber
is used for the description of these ro-vibrational transitions, it is convenient to
introduce the corresponding ro-vibrational term as follows:

Ev,J 1
G(v, J ) =
. = ν̃e (v + ) + B̃e J (J + 1). (5.112)
hc 2
5.4 Spectroscopy of Rotational Transitions for Diatomic Molecules 155

Selection Rules for RO-Vibrational Transitions and P, Q, and R Branches

• Due to the availability of rotational transition during vibrational transition
as noted above, in any IR spectroscopy, not only the vibrational quantum
number changes according to .Δv = 1 but the rotational quantum number J
also changes in general. For the case of most diatomic molecules,1 . J =
±1.
• The transitions corresponding to . J = −1 form the P-branch and have
the following wavenumbers:

ν̃P (J ) = ν̃e + B̃e (J − 1)J − B̃e J (J + 1) = ν̃e − 2B̃e J.

. (5.113)

• The transitions corresponding to . J = 1 form the R-branch and have the

following wavenumbers:

.ν̃R (J ) = ν̃e +B̃e (J +1)(J +2)−B̃e J (J +1) = ν̃e +2B̃e (J +1). (5.114)

• The transitions corresponding to . J = 0, which is normally forbidden,

form the Q-branch and the corresponding wavenumber is .ν̃e .

1
There are few exceptions for this such as the molecule .NO, for which . J = 0 is also
possible. Such exception occurs due to the special nature of the electronic states that
contributes to the angular momentum.

The above selection rules are based on the approximation that the vibrational and
rotational motion are decoupled from each other and that there are no centrifugal
distortion. Complications that arise when these are included are described in the
next subsection.

5.4.4 Centrifugal Correction and Ro-Vibrational Coupling

In actual diatomic molecules, corrections need to be made for the distortion of

bond distance due to centrifugal distortion and the coupling between vibration and
rotation. As long as such effects are small, they can be added as small corrections.
For this, let us expand .Vrel (r) and .h̄2 J (J +1)/r 2 in Eq. (5.106) around .re as follows:
156 5 Rotational States and Spectroscopy

1 1
Vrel (re + s) ≈ Vrel (re ) + μω2 s 2 − κs 3 , (5.115)
2 3!
h̄2 J (J + 1) h̄2 J (J + 1) s −2
 
= 1 +
2μ(re + s)2 2μre2 re
h̄2 J (J + 1) s2
 
s
≈ 1 − 2 + 3 . (5.116)
2μre2 re re2

From these approximations, we find the following expressions for the first and the
eff
second derivatives of .VJ (r):

d eff 1 h̄2 J (J + 1) 3h̄2 J (J + 1)

. VJ (r) ≈ μωe2 s − κs 2 − + s, . (5.117)
ds 2 μre3 μre4
d 2 eff 3h̄2 J (J + 1)
2
VJ (r) ≈ μωe2 − κs + . (5.118)
ds μre4

In Eq. (5.117), let us consider only the constant term and the most dominant linear
term, .μωe2 s, in order to find a new minimum of the potential energy, .se . That is, for
.se , the following relation is assumed:

h̄2 J (J + 1)
.μωe2 se − ≈ 0. (5.119)
μre3

Solving this equation, we find that

h̄2 J (J + 1)
se ≈
. . (5.120)
μ2 ωe2 re3

This can be understood as the shift of the potential minimum due to the centrifugal
force. Inserting this value into Eq. (5.118),

d 2 eff κ h̄2 J (J + 1) 3h̄2 J (J + 1)


2

. V (r e + s)  ≈ μωe − +
ds 2 J 
s=se μ2 ωe2 re3 μre4

≡ μωJ2 . (5.121)

where
 1/2
κ h̄2 J (J + 1) 3h̄2 J (J + 1)
ωJ = ωe
. 1− + . (5.122)
μ3 ωe4 re3 μ2 ωe2 re4
5.4 Spectroscopy of Rotational Transitions for Diatomic Molecules 157

eff
Then, making quadratic expansion of .VJ around .re + se , we find that

eff 1 μωJ2 2 h̄2 J (J + 1)

VJ
. (re + s) ≈ Vrel (re ) + μωJ2 (s − se )2 − s +
2 2 e 2Ie
1 h̄2 h̄4
= V0 + μωJ2 (s − se )2 + J (J + 1) − J 2 (J + 1)2 .
2 2Ie 2ωe2 Ie3
(5.123)

Now inserting this new form of effective potential into Eq. (5.108), one can find the
following expression for the eigenvalue:

1 h̄2 h̄4
Ev,J = V0 + h̄ωJ (v + ) +
. J (J + 1) − J 2 (J + 1)2 , (5.124)
2 2Ie 2ωe2 Ie3

where .ωJ is defined by Eq. (5.122) and the last term corresponds to the centrifugal
distortion energy. Further approximation of Eq. (5.122) can be made. Let us
introduce
 
1 κ h̄3 3h̄3
.αe = − 2 2 4 . (5.125)
2 μ3 ωe4 re3 μ ωe re

In general, .αe << h̄. Thus,

J (J + 1) 1/2
 
αe ωe
ωJ = ωe 1 − 2αe
. ≈ω− J (J + 1). (5.126)
h̄ h̄

Inserting this into Eq. (5.124) and also introducing centrifugal distortion constant

h̄4
De =
. , (5.127)
2ωe2 Ie3

we find that
 
αe 1
Ev,J
. ≈ V0 + h̄ ωe − J (J + 1) (v + ) + Be J (J + 1) − De J 2 (J + 1)2
h̄ 2
1 v,J
= V0 + h̄ωe (v + ) + Be J (J + 1) + δErot , (5.128)
2
where

v,J 1
δErot
. = −αe (v + )J (J + 1) − De J 2 (J + 1)2 . (5.129)
2
158 5 Rotational States and Spectroscopy

The first term in the above correction term comes from the coupling between the
vibration and rotation and the second term comes from the centrifugal distortion.
Note that both terms reduce the rotational energy level spacings because they
effectively increase the moment of inertia.
Let us first consider the case of rotational states for vibrational ground state,
.v = 0. The rotational energy level (without including the constant term) is given by

EJ = B0 J (J + 1) − De J 2 (J + 1)2 ,
. (5.130)

where
αe
B0 = Be −
. . (5.131)
2
The corresponding rotational energy term is as follows:

.G(J ) = B̃0 J (J + 1) − D̃e J 2 (J + 1)2 , (5.132)

where .B̃0 = B0 /(hc) and .D̃e = De /(hc). With this expression, the correction for
the rotational transition wavenumber, for the case of .J → J + 1, can be shown to
be

. ν̃(J → J + 1) = B̃0 (J + 1)(J + 2) − D̃e (J + 1)2 (J + 2)2

−B̃0 J (J + 1) + D̃e J 2 (J + 1)2
= 2B̃0 (J + 1) − D̃e (J + 1)2 (4J + 4)
= 2B̃0 (J + 1) − 4D̃e (J + 1)3 . (5.133)

Note that the reduction of the level spacing due to the centrifugal distortion depends
on the third order of .(J + 1).
Next, let us consider the effects of the coupling of rotational and vibrational
states, but without including the effects of the centrifugal distortion. The corre-
sponding term for the state in this case is given by

1
G(v, J ) = ν̃e (v + ) + B̃v J (J + 1),
. (5.134)
2
where
1
B̃v = B̃e − α̃e (v + ),
. (5.135)
2

with .α̃e = αe /(hc). The above expression implies that the rotational constant
decreases as the vibrational quantum number increases. This is because larger
vibrational quantum number results in larger moment of inertia. Note that the
rotational constant is inversely proportional to the moment of inertia. Given the
5.5 Summary and Questions 159

above term, the P-branch, assuming that the vibrational transition is .v = 0 → 1,

now has the following expression:

ν̃P (J ) = ν̃e + B̃1 J (J − 1) − B̃0 J (J + 1)

.

= ν̃e + (B̃1 − B̃0 )J 2 − (B̃1 + B̃0 )J

= ν̃e − α̃e J 2 − 2(B̃e − α̃e )J. (5.136)

This P-branch has lower energy than the original pure vibrational transition, with
the red shift increasing (becoming more negative) as J increases. The level spacing
of this P-branch between consecutive values of J increases as J increases. On the
other hand, the R-branch has the following expression:

.ν̃R (J ) = ν̃e + B̃1 (J + 1)(J + 2) − B̃0 J (J + 1)

= ν̃e + (B̃1 − B̃0 )J 2 + (3B̃1 − B̃0 )J + 2B̃1
= ν̃e − α̃e J 2 + (2B̃e − 4α̃e )J + 2B̃e − 3α̃e . (5.137)

This R-branch has higher energy than the original vibrational transition (.J = 0),
and shows blue shift (at least for small values of J ) as J increases. However,
the difference of transition energies between consecutive values of J within this
R-branch decreases as J increases. For sufficiently large value of J , the above
expression predicts that .ν̃R (J ) decreases as J increases. This is called “reversal
of R-head” [31].

5.5 Summary and Questions

Eigenvalues and eigenfunctions of two dimensional rotation (or rotation around

the z-axis) are given by Eqs. (5.24) and (5.26), respectively. For three dimensional
rotation, separation of variables in spherical coordinate system leads to an equation
for the axial angle .θ , the solution of which is related to associated Legendre
functions provided in Table 5.1. Combination of these with eigenfunctions for
the two-dimensional rotation results in spherical harmonics defined by Eq. (5.42)
as eigenfunctions for the three dimensional rotation with eigenvalues given by
Eq. (5.43).
Alternatively, three dimensional rotational Hamiltonian can be expressed in
terms of the angular momentum square operator as shown in Eq. (5.54). Spherical
harmonics are simultaneous eigenfunctions of the angular momentum square
operator and the z-component of the angular momentum operator, and serve as
a complete basis for describing any three dimensional rotational motion. This is
also related with the fact that three different cartesian components of the angular
momentum operators do not commute with each other, as shown in Eqs. (5.58)–
(5.60), making it impossible to specify more than one of them.
160 5 Rotational States and Spectroscopy

The rotational spectroscopy of diatomic molecules, which are approximated as

rigid rotors, can be fully described in terms of the interaction of dipole operator
or polarization operator associated with the rotation and electric fields. Microwave
spectroscopy involves interaction with the dipole operator, for which the selection
rule is given by Eq. (5.95). Rotational Raman spectroscopy involves interaction with
the polarization operator, for which the selection rule is given by Eq. (5.101). It
is also important to note that change of rotational states accompany vibrational
transition in general, typically resulting in two major branches called .P and .R, as
expressed respectively by Eqs. (5.136) and (5.137). It has long been an experimental
convention to express all of rotational transition energies in the unit of wavenumber,
.cm
−1 , for which rotational constants serve as a natural unit of energy.

For flexible molecules or for molecules in highly excited rotational states,

the assumption of rigid rotor breaks down. Inclusion of the effect of centrifugal
distortion is possible by adding corrections for each rotational level as long as
its effect remains small. Combining this with the change of moment of inertia
associated with different vibrational level, it is possible to obtain an approximate
expression for the ro-vibrational level as shown in Eq. (5.128). This expression
provides more satisfactory description of ro-vibrational transitions of many diatomic
molecules.

Questions
• What are physical origins for the quantization of rotational energies?
• Why are quantum states for three dimensional rotation appear to be much
more complicated than those for two dimensional rotation?
• What is the degeneracy for a two dimensional rotational quantum number
m?
• What is the degeneracy for a three dimensional rotational quantum number
J?
• What is the physical meaning of the moment of inertia?
• Why is it impossible to determine eigenstates of more than one components
of the angular momentum vector operator?
• Is angular momentum a conserved quantity for a diatomic molecule in
space free of any external potentials?
• Can a molecule with zero permanent dipole moment still be active in
rotational Raman spectroscopy?
• Why do multiple transition lines exist in the rotational spectroscopy even
for the simplest case of a diatomic molecule?
• Why does centrifugal distortion tend to reduce rotational level spacings as
the rotational quantum number increases?
Appendix: Associated Legendre Equations and Their Solutions 161

Appendix: Associated Legendre Equations and Their

Solutions

Let us first consider the case where .m = 0 in Eq. (5.37) as follows:

 
1 d d
. sin θ (θ ) + β(θ ) = 0, (5.138)
sin θ dθ dθ

which is called Legendre equation (with appropriate choice of .β as will be clear

below). Let us introduce .x = cos θ and .(θ ) = f (x). Since .dx = − sin θ dθ ,
Eq. (5.138) can be expressed as
 
d 2 d
. (1 − x ) f (x) + βf (x) = 0. (5.139)
dx dx

Let us assume that .f (x) can be expressed as the following power series:
∞

f (x) =
. Ck x k+p , (5.140)
k=0

where p is the order of the first term and has to be determined by solving the
equation. Taking the derivative of .f (x) and multiplying with .(1 − x 2 ), and then
taking derivative again, we obtain
   ∞
d d 
. (1 − x 2 ) f (x) = Ck (k + p)(k + p − 1)x k+p−2
dx dx
k=0

−(k + p)(k + p + 1)x k+p

= C0 p(p − 1)x p−2 + C1 (p + 1)px p−1

∞

+ {Ck+2 (k + p + 2)(k + p + 1)
k=0

−Ck (k + p)(k + p + 1)} x k+p .

(5.141)

Combining this with Eq. (5.140) in Eq. (5.139), we obtain

. C0 p(p − 1)x p−2 + C1 (p + 1)px p−1

∞

+ {Ck+2 (k + p + 2)(k + p + 1) − Ck (k + p)(k + p + 1) + Ck β} x k+p = 0.
k=0
(5.142)
162 5 Rotational States and Spectroscopy

The first two terms are zero if .p = 0, for which one can assume either .C0 = 0 or
C1 = 0. The choice of .p = 1 with .C1 = 0 or .p = −1 with .C0 = 0 is equivalent to
.

one of these cases. Thus, without losing any generality, we can assume that .p = 0.
For Eq. (5.142) to be satisfied, each coefficient for .x k should be zero. This means
that
k(k + 1) − β
Ck+2 =
. Ck . (5.143)
(k + 2)(k + 1)

Now let us consider the large k limit, for which .Ck+2 ∼ Ck . This means that, as x
approaches .±1, .f (x) ∼ 1/(1−x 2 ) if .C1 = 0 and .f (x) ∼ x/(1−x 2 ) if .C0 = 0. The
resulting wavefunction in this case is not normalizable. Therefore, for the solution
to be physically acceptable, the series has to terminate at finite order of x. Thus, let
us assume that the series terminate at .x J . This is possible if

β = J (J + 1),
. (5.144)

where .J = 0, 1, · · · . The resulting solution of Eq. (5.139) for each case of J is

called Legendre polynomial of order J and is denoted as .PJ (x).
Let us now consider the case where .m = 0 in Eq. (5.37) for .β given by
Eq. (5.144). Employing the same .x = cos θ as variable, it can be expressed as

m2
   
d d
. (1 − x 2 ) f (x) + J (J + 1) − f (x) = 0. (5.145)
dx dx (1 − x 2 )

Note that the additional term .−m2 /(1 − x 2 ) multiplied to .f (x) becomes singular
for .x = ±1, for which, .θ = 0, π . This singular term has to be taken care of before
assuming power series for the solution. This can be achieved by introducing another
function .g(x) such that

f (x) = (1 − x 2 )|m|/2 g(x).

. (5.146)

Why the above form removes the singularity can be understood considering the
leading contribution from the second derivative of .f (x), as detailed below. Taking
the derivative of Eq. (5.146) with respect to x and multiplying it with .(1 − x 2 ), we
obtain,

d d
(1 − x 2 )
. f (x) = −|m|x(1 − x 2 )|m|/2 g(x) + (1 − x 2 )|m|/2+1 g(x). (5.147)
dx dx
Appendix: Associated Legendre Equations and Their Solutions 163

Taking derivative of the above equation one more time with respect to x, we obtain

d d
. (1 − x 2 ) f (x) = −|m|(1 − x 2 )|m|/2 g(x) + m2 x 2 (1 − x 2 )|m|/2−1 g(x)
dx dx
d d
−|m|x(1−x 2 )|m|/2 g(x)−(|m|+2)x(1−x 2 )|m|/2 g(x)
dx dx
d2
+(1 − x 2 )|m|/2+1 g(x)
dx 2
= −(|m| + m2 )(1 − x 2 )|m|/2 g(x) + m2 (1 − x 2 )|m|/2−1 g(x)
d
−(2|m| + 2)x(1 − x 2 )|m|/2 g(x)
dx
d2
+(1 − x 2 )|m|/2+1 g(x). (5.148)
dx 2
Using this expression and Eq. (5.146) in Eq. (5.149) and then rearranging terms, we
find that
 
2 |m|/2 d d d
. (1 − x ) (1 − x 2 ) g(x) − 2|m|x g(x)
dx dx dx
  
+ J (J + 1) − (m2 + |m|) g(x) = 0. (5.149)

Since this has to be satisfied for any value of x, .(1 − x 2 )|m|/2 can be factored out. In
addition, using the fact that .J (J + 1) − (m2 + |m|) = (J − |m|)(J + |m| + 1), we
obtain
 
d 2 d d
. (1 − x ) g(x) − 2|m|x g(x) + (J − |m|)(J + |m| + 1)g(x) = 0.
dx dx dx
(5.150)

Note that the above equation is similar to Eq. (5.139) except for additional first
derivative term. In fact, .g(x) can be related to .PJ (x) through derivatives as shown
below.
Let us consider the Legendre polynomial .PJ (x), which is the solution of the
following differential equation:

d2 d d
.(1 − x 2 ) 2
PJ (x) − 2x PJ (x) + J (J + 1) PJ (x) = 0, (5.151)
dx dx dx
where we have separated out the first and second derivative terms. Taking the
derivative of this equation with respect to x, we obtain

d3 d2 d
.(1 − x 2 ) PJ (x) − 4x PJ (x) + (J (J + 1) − 2) PJ (x) = 0. (5.152)
dx 3 dx 2 dx
164 5 Rotational States and Spectroscopy

Taking the derivative of the above equation once again, we also obtain

d4 d3 d2
(1 − x 2 )
. PJ (x) − 6x PJ (x) + (J (J + 1) − 2 − 4) PJ (x) = 0.
dx 4 dx 3 dx 2
(5.153)

Inspection of the above two equations shows that the following general relation,
which can be proved by taking derivatives of Eq. (5.153) for .|m| times, is satisfied.

d |m|+2 d |m|+1
. (1 − x 2 ) |m|+2
PJ (x) − 2(|m| + 1)x |m|+1 PJ (x)
dx dx
d |m|
+(J (J + 1) − |m| − m2 ) PJ (x)
dx |m|
  |m|   |m| 
d d d d d
= (1 − x 2 ) PJ (x) − 2|m|x PJ (x)
dx dx dx |m| dx dx |m|
 |m| 
d
+(J − |m|)(J + |m| + 1) PJ (x) = 0. (5.154)
dx |m|

This last equation is exactly the same form as Eq. (5.150). Thus, we find the
following solution:

d |m|
g(x) =
. PJ (x). (5.155)
dx |m|

Since .PJ (x) is a polynomial of order J , the following condition should hold

.|m| ≤ J. (5.156)

Combining this with Eq. (5.146), we obtain the following definition of associated
Legendre function:

|m| d |m|
PJ (x) = (1 − x 2 )|m|/2
. PJ (x). (5.157)
dx |m|
Whether there are other physically acceptable solutions of the associated Legendre
equation that can be defined for .−1 ≤ x ≤ 1 is a rather subtle issue, but can be
ruled out after more careful consideration [32].

Exercise Problems with Solutions

5.1 Prove that Y21 (θ, φ) in Table 5.2 satisfies Eq. (5.33) and determine the corre-
sponding eigenvalue E for this function.
Exercise Problems with Solutions 165

Solution 5.1 Y21 (θ, φ) = N sin θ cos θ eiφ , where N = −(15/8π )1/2 . Application
of the differentiation with respect to θ constituting L̂2 to Y21 (θ, φ) is thus propor-
tional to
 
1 ∂ ∂
. sin θ sin θ cos θ eiφ
sin θ ∂θ ∂θ
1  
= sin θ (cos2 θ − sin2 θ )eiφ
sin θ
1 ∂  
= sin θ cos(2θ )eiφ
sin θ ∂θ
1
= (cos θ cos(2θ ) − 2 sin θ sin(2θ )) eiφ
sin θ
1  
= cos θ (cos2 θ − sin2 θ ) − 4 sin2 θ cos θ eiφ
sin θ
 
1
= cos θ − 6 sin θ cos θ eiφ .
sin θ

On the other hand,

1 ∂2 1
.
2 ∂φ 2
sin θ cos θ eiφ = − cos θ eiφ .
sin θ sin θ

Therefore,

h̄2 1 ∂2
   
1 ∂ ∂
. − sin θ + Y21 (θ, φ)
2I sin θ ∂θ ∂θ sin2 θ ∂φ 2
h̄2 6h̄2 1
 
1 1 1
=− − 6 − Y 2 (θ, φ) = Y (θ, φ).
2I sin2 θ sin2 θ 2I 2

5.2 Prove that [L̂y , L̂z ] = i h̄Lx in a way similar to Eq. (5.61).
Solution 5.2 Employing the definitions of L̂y and L̂z ,

[L̂y , L̂z ] = [ẑp̂x − x̂ p̂z , x̂ p̂y − ŷ p̂x ]

.

= [ẑp̂x , x̂ p̂y ] − [x̂ p̂z , x̂ p̂y ] − [ẑp̂x , ŷ p̂x ] + [x̂ p̂z , ŷ p̂x ]
= ẑ[p̂x , x̂]p̂y + ŷ[x̂, p̂x ]p̂z
= −i h̄ẑp̂y + i h̄ŷ p̂z = i h̄L̂x .

5.3 Prove that [L̂2 , L̂x ] = 0 in a way similar to Eq. (5.62).

166 5 Rotational States and Spectroscopy

Solution 5.3 Employing the fact that L̂2 = L̂2x + L̂2y + L̂2z ,

. [L̂2 , L̂x ] = [L̂2y , L̂x ] + [L̂2z , L̂x ]

= L̂y [L̂y , L̂x ] + [L̂y , L̂x ]L̂y + L̂z [L̂z , L̂x ] + [L̂z , L̂x ]L̂z
= L̂y (−i h̄L̂z ) + (−i h̄L̂z )L̂y + L̂z (i h̄L̂y ) + (i h̄L̂y )L̂z = 0.

5.4 Prove that the following wave function is an eigenfunction of L̂2 and find the
eigenvalue. Is this also an eigenfunction of L̂z ? Provide clear basis for your answer.
 1/2   
5 3
ψ(θ, φ) =
. 3 cos2 θ + sin2 θ e2iφ − 1 .
32π 2

Solution 5.4 Let us first consider the following operation:

  
1 ∂ ∂ 2 3 2 2iφ
. sin θ 3 cos θ + sin θ e − 1
sin θ ∂θ ∂θ 2
  
1 ∂ 3 2iφ
= sin θ 6 cos θ (− sin θ ) + 2 sin θ cos θ e
sin θ ∂θ 2
  
1 ∂ 3 2iφ
= sin θ sin(2θ ) −3 + e
sin θ ∂θ 2
  
1 3 2iφ
= (cos θ sin(2θ ) + 2 sin θ cos(2θ )) −3 + e
sin θ 2
  
2 2 3 2iφ
= (4 cos θ − 2 sin θ ) −3 + e
2

2 3
= −6(3 cos θ − 1) + (4 − 6 sin2 θ )e2iφ .
2

On the other hand,

   
1 ∂2 3 2 2iφ 3 2iφ
. 3 cos2 θ + sin θ e − 1 = − 4e .
sin2 θ ∂φ 2 2 2
Exercise Problems with Solutions 167

Combining the above two results,

 1/2   
2 2 5 2 3 2 2iφ
L̂ ψ(θ, φ) = −h̄
. −6(3 cos θ − 1) − 6 sin θ e
32π 2

= 6h̄2 ψ(θ, φ).

Thus, ψ(θ, φ) is an eigenfunction of L̂2 with eigenvalue 6h̄2 . On the other hand,
 
1/2  
5
h ∂ 2 3 2 2iφ
.L̂z ψ(θ, φ) = 3 cos θ + sin θ e − 1
32π
i ∂φ 2

5 1/2 3
  
= 2h̄ sin2 θ e2iφ .
32π 2

This is not proportional to ψ(θ, φ). Therefore, ψ(θ, φ) is not an eigenfunction of

L̂z .
5.5 Diatomic molecule H2 and its deuterium isotope D2 have the same internuclear
bond distance 0.748 Å. Calculate the values of rotational constant, B̃, and the three
lowest rotational energy levels in the unit of wavenumber cm−1 .
Solution 5.5 The rotational constant B̃ is given by the following expressions:

h̄2 h̄2 h
B̃ =
. = = ,
2I hc 2μR 2 hc 8π 2 μR 2 c

where R = 0.748 Å is the internuclear distance and μ is the reduced mass. For H2 ,
μH2 = mH /2 = 1.0078 × 1.6605 × 10−27 /2 kg = 8.3673 × 10−25 g. For D2 ,
μD2 = mD /2 = 2.01410 × 1.6605 × 10−27 /2 kg = 1.6722 × 10−24 g. Therefore,
B̃H2 = 59.79 cm−1 and B̃D2 = 29.92 cm−1 . With these values, we obtain the
following values for the energies of three lowest rotational energy levels:

J =0 J =1 J =2
H2 0 1.196 × 102 cm−1 3.587 × 102 cm−1
D2 0 5.984 × 10 cm−1 1.795 × 102 cm−1

5.6 The equilibrium bond length for H127 I is 1.604 × 10−10 m. Calculate the
three lowest transition energies for the microwave (pure rotational) absorption
spectroscopy and (ii) the two lowest Stokes Raman transition energies for the
rotational Raman spectroscopy.
Solution 5.6 For this diatomic molecule, the reduced mass is given by

1.0078 × 126.90
μ=
. × 1.6605 × 10−24 g = 1.6604 × 10−24 g.
1.0078 + 126.90
168 5 Rotational States and Spectroscopy

For this reduced mass and the given internuclear distance, the rotational constant
becomes B̃ = 6.553 cm−1 = 1.302×10−22 J. Thus, the three microwave absorption
transition energies are as follows:

. J (0 → 1) : 2B̃ = 13.11 cm−1 = 2.604 × 10−22 J

J (1 → 2) : 4B̃ = 26.21 cm−1 = 5.206 × 10−22 J
J (2 → 3) : 6B̃ = 39.32 cm−1 = 7.810 × 10−22 J

On the other hand, the two Stokes Raman transition energies are as follows:

. J (0 → 2) : 6B̃ = 39.32 cm−1 = 7.810 × 10−22 J

J (1 → 3) : 10B̃ = 65.53 cm−1 = 1.302 × 10−21 J

5.7 For a diatomic molecule AlF, B̃0 = 0.55 cm−1 and D̃e = 1.05 × 10−6 cm−1 .
Calculate microwave absorption and Stokes Raman transition energies respectively
for each case that the molecule has initial rotational quantum numbers J = 1 and
J = 5.
Solution 5.7 This involves simple application of Eq. (5.132). Thus, the microwave
absorption transition energies starting from J = 1 and J = 5 are as follows:

. J (1 → 2) : B̃0 (6 − 2) − D̃e (36 − 4) = 4B̃0 − 32D̃e

= 2.2 cm−1 = 4.37 × 10−23 J
J (5 → 6) : B̃0 (42 − 30) − D̃e (1764 − 900) = 12B̃0 − 864D̃e
= 6.599 cm−1 = 1.311 × 10−22 J

On the other hand, the Stokes Raman transition energies starting from J = 1 and
J = 5 are as follows:

. J (1 → 3) : B̃0 (12 − 2) − D̃e (144 − 4) = 10B̃0 − 140D̃e

= 5.5 cm−1 = 1.09 × 10−23 J
J (5 → 7) : B̃0 (56 − 30) − D̃e (3136 − 900) = 26B̃0 − 2236D̃e
= 14.298 cm−1 = 2.84 × 10−22 J

5.8 For a diatomic molecule HCl, ν̃e = 2885.9775 cm−1 , B̃e = 10.59342 cm−1
and α̃e = 0.30718 cm−1 and assume that D̃e = 0. Then, calculate the transition
energies (in the unit of wavenumber cm−1 ) for the first three P -branches and the
first three R-branches.
Problems 169

Solution 5.8 This involves simple applications of Eqs. (5.136) and (5.137). Thus,
the three P -branches are as follows:

ν̃P (1) = ν̃e − α̃e − 2(B̃e − α̃e )

.

= ν̃e + α̃e − 2B̃e = 2865.09784 cm−1 ,

ν̃P (2) = ν̃e − 4α̃e − 4(B̃e − α̃e )
= ν̃e − 4B̃e = 2843.60382 cm−1 ,
ν̃P (3) = ν̃e − 9α̃e − 6(B̃e − α̃e )
= ν̃e − 3α̃e − 6B̃e = 2821.49544 cm−1 .

On the other hand, the three R-branches are as follows:

ν̃R (0) = ν̃e + 2B̃e − 3α̃e = 2906.2428 cm−1 ,

.

ν̃R (1) = ν̃e − α̃e + (2B̃e − 4α̃e ) + (2B̃e − 3α̃e )

= ν̃e + 4B̃e − 8α̃e = 2925.89374 cm−1 ,
ν̃R (2) = ν̃e − 4α̃e + 4B̃e − 8α̃e + 2B̃e − 3α̃e
= ν̃e − 15α̃e + 6B̃e = 2944.93032 cm−1 .

Problems

5.9 Prove that [L̂2x + L̂2y , L̂z ] = 0.

5.10 A three-dimensional rotor is in a rotational state |ψ with the following
wavefunction:
 
1 15
.ψ(θ, φ) = θ, φ|ψ = ψ(θ ) = A + cos θ + cos2 θ.
4π 16π

Determine A such that the above wavefunction is normalized. If L̂2 and L̂z are
measured for this state, what are possible outcomes and their probabilities?
5.11 A three-dimensional rotor is in a rotational state |ψ with the following
wavefunction:
 1/2  1/2
15 15
ψ(θ, φ) = θ, φ|ψ =
. sin θ cos θ cos φ + i sin2 θ sin(2φ).
8π 32π
170 5 Rotational States and Spectroscopy

If L̂2 and L̂z are measured for this state, what are possible outcomes and their
probabilities?
5.12 The ro-vibrational state of a diatomic molecule has the following term:

1
G(v, J ) = ν̃e (v + ) + B̃v J (J + 1) − D̃v J 2 (J + 1)2 ,
.
2
where
 
1
.B̃v = B̃e − α̃e v+ ,
2
 
1
D̃v = D̃e − δ̃e v + .
2

Find out expressions for P and R branches for the vibrational transition from v = 0
to v = 1.
Chapter 6
Hydrogen-Like Systems and Spin Orbit
States of an Electron

When it comes to atoms, language can be used only as in poetry.

The poet, too, is not nearly so concerned with describing facts
as with creating images.
— Niels Bohr

Abstract This chapter provides detailed description of energies, states, and quan-
tum numbers of the state of an electron in hydrogen-like systems, namely, hydrogen
atom or ions with a single electron. First, Bohr’s model is used for a simple deriva-
tion of energy levels. Then, the complete set of eigenfunctions and eigenvalues
for the Hamiltonian as solutions of the time independent Schrödinger equation
are provided. In particular, the equation for the radial part of the wavefunction
and corresponding solutions are described in detail. Spin states of electron are
then described, which in combination with spatial wavefunctions, namely, orbitals,
provide full specification of a quantum state of an electron in a hydrogen-like
system. Effects of a magnetic field on these states and consequences of spin-orbit
interactions are briefly explained.

The hydrogen atom or hydrogen-like ion consists of only one nucleus with charge
Z and an electron. The potential energy between the electron and the nucleus is the
following Coulomb potential:

Ze2
V (r) = −
. , (6.1)
4π ǫ0 r

where .r = |re − rn |, with .re and .rn being the positions of the electron and the
nucleus respectively, .e = 1.602176634 × 10−19 C, and .ǫ0 = 8.8541878128 ×
10−12 C2 /Nm, which is the electric permittivity of vacuum. Before we consider the
solution of the Schrödinger equation for this potential, let us go over the Bohr’s
model first.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 171
S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_6
172 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

6.1 Bohr’s Model

Let us consider an electron, assumed to rotate around the nucleus with charge Z at a
fixed distance r from the nucleus and with a constant speed v. This is possible only
if the speed v of the electron is such that the Coulomb force balances the centrifugal
force, which means that

Ze2 me v 2
. = . (6.2)
4π ǫ0 r 2 r

This results in the following relationship between r and v:

Ze2 Zme e2 Zme e2

r=
.
2
= 2
= , (6.3)
4π ǫ0 me v 4π ǫ0 (me v) 4π ǫ0 pe2

where .pe = me v is the magnitude of the linear momentum of the electron.

Information on r also determines the energy of the electron because

pe2 Ze2 Ze2 Ze2 Ze2

E=
. − = − =− . (6.4)
2me 4π ǫ0 r 8π ǫ0 r 4π ǫ0 r 8π ǫ0 r

On the other hand, according to de Broglie, .pe is inversely proportional to the

wavelength of the electron as follows:

h
pe =
. . (6.5)
λe

Having learned the principles and some details of the quantum mechanics and
wavefunction, we can now view .λe as the wavelength of electron’s wavefunction
defined along its circular trajectory. For the wave not to self-destruct itself along
the trajectory with a periodic boundary condition, the integer multiples of the
wavelength have to be equal to the circumference of the orbiting circle as follows:

nλe = 2π r, n = 1, 2, 3, · · · .
. (6.6)

In other words, if the above condition is not satisfied, there is no standing wave that
is consistent with stable circular motion of the electron. Combining Eq. (6.6) with
Eq. (6.5), we obtain the following relationship between .pe and r.

hn h̄n
pe =
. = , n = 1, 2, 3, · · · (6.7)
2π r r
6.1 Bohr’s Model 173

Employing this expression in Eqs. (6.3) and (6.4), we find that

4π ǫ0 h̄2 n2 n2
. rn = = a0 , . (6.8)
Zme e2 Z
me Z 2 e 4 Z 2 h̄2
En = − =− , (6.9)
32π 2 ǫ02 h̄2 n2 n2 2me a02

where .n = 1, 2, 3, · · · , and .a0 is the value of .rn for .Z = 1 and .n = 1. The expression
for .a0 is as follows:

4π ǫ0 h̄2
a0 =
. = 0.529 . . . Å. (6.10)
me e 2

This is an important atomic constant known as the Bohr radius and is approximately
equal to the size, namely, van der Waals radius, of a hydrogen atom. The constant
factor, except for charge Z, in Eq. (6.9) is related to another important constant
known as Rydberg constant given by

me e 4 1 me e 4 h̄
R∞ =
.
2
= 2
= = 1.0974 · · · × 105 cm−1 . (6.11)
2 2
32π ǫ0 h̄ hc 3
8ǫ0 h c 4π me a02 c

As a result, the energy of the electron within the Bohr’s model is expressed as

Z2
En = −R∞ hc
. . (6.12)
n2
Energy levels given above turned out to be fairly accurate and were able to explain
the electronic spectroscopic data of a hydrogen atom. It is even more amazing that
Bohr was able to come up with the quantization condition, Eq. (6.6), before de
Broglie’s theory was proposed.
Equation (6.12) is based on the assumption that the position of the nucleus is
fixed, which can be corrected by replacing the mass of an electron with the reduced
mass for the system. For the case of a hydrogen atom, the reduced mass is .μ =
me mH /(me + mH ). The resulting Rydberg constant for the electron in hydrogen
atom is thus given by
μ
RH =
. R∞ = 1.0968 · · · × 105 cm−1 . (6.13)
me

Therefore, the wavenumber of photons coming out of the (emission) transition

between different states of the electron, according to the Bohr’s model (with the
correction of reduced mass), can be expressed as
174 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

 
En2 − En1 1 1
ν̃n2 →n1
. = = RH − 2 . (6.14)
hc n21 n2

In fact, the above expression, known as Ritz combination principle, had already been
well established experimentally before Bohr’s model was developed. Specifically,
the transitions to .n1 = 1 from any value of .n2 > 1 were known as Lyman series,
those to .n1 = 2 from any value of .n2 > 2 Balmer series, and those to .n1 =
3 from any value of .n2 > 3 Paschen series. Bohr’s model was able to explain
these observations in a remarkably simple manner by magically combining particle
and wave properties of the electron as described above. While the idea engrained
in this Bohr’s model formed the conceptual foundation of quantum mechanics, as
will become clear, it is important to note that the success of Bohr’s model as an
accurate theory for energy levels of a hydrogen atom is due to the special nature of
the Coulomb potential involved.

6.2 Solution of Schrödinger Equation

Hydrogen atoms or hydrogen-like ions are examples of two-particle system with

their relative potential energy given by Eq. (6.1). Therefore, the Hamiltonian for
their states and energies can be expressed in terms of the Hamiltonian given
by Eq. (4.79). Let us assume that there is no external potential and that we are
considering the hydrogen-like system in the center-of-mass coordinate system.
Then, the Hamiltonian operator becomes

p̂2
Ĥ =
. + V̂ (r̂), (6.15)
2μ

where .μ = mn me /(mn + me ), with .mn being the mass of the nucleus, and .p̂ = p̂e −
p̂n is the momentum of electron relative to the nucleus. When the above operator is
applied to the position state represented as a bra in the Dirac notation, we obtain
 
h̄2 2
.r|Ĥ = − ∇ + V (r) r|. (6.16)
2μ

This implies that the Schrödinger equation for the wavefunction .ψ(r) = r|ψ is as
follows:
 
h̄2 2
.r|Ĥ |ψ = − ∇ + V (r) ψ(r) = Eψ(r). (6.17)
2μ
6.3 Separation of Variables in Spherical Coordinate System 175

From now one, for convenience, we will not strictly follow the convention of the
Dirac notation and often express the first term in the above equation, simply as,
.Ĥ ψ(r), following the original Schrödinger’s definition.

6.3 Separation of Variables in Spherical Coordinate System

As was shown in Chap. 5, the Laplacian .∇ 2 in the spherical coordinate system is

expressed as

∂2
   
1 ∂ ∂ 1 ∂ ∂ 1
∇2 =
. r2 + 2 sin θ +
r 2 ∂r ∂r r sin θ ∂θ ∂θ r 2 sin2 θ ∂φ 2
1 ∂2 ∂2
 
1 ∂ ∂ 1
= r + sin θ + . (6.18)
r ∂r 2 r 2 sin θ ∂θ ∂θ r 2 sin2 θ ∂φ 2

Therefore, the Schrödinger equation, Eq. (6.17), in spherical coordinate system is

expressed as
 
h̄2 1 ∂ 2 1 2
.Ĥ ψ(r, θ, φ) = − r+ L̂ + V (r) ψ(r, θ, φ) = Eψ(r, θ, φ),
2μ r ∂r 2 2μr 2
(6.19)

where .L̂2 is the angular momentum operator defined in Chap. 5 and has the
following expression:

∂2
   
2 2 1 ∂ ∂ 1
L̂ = −h̄
. sin θ + . (6.20)
sin θ ∂θ ∂θ r 2 sin2 θ ∂φ 2

This has the same form as that for the rotation of a diatomic molecule. However, in
the present case, it is the angular momentum of (mostly) electron with respect to the
center of mass, which is almost the same as the position of the nucleus.
We learned in Chap. 5 that the spherical harmonics are eigenfunctions of .L̂2 .
Thus, we may be able to simplify Eq. (6.19) by assuming that the eigenfunction can
be expressed as

.ψ(r, θ, φ) = R(r)Ylml (θ, φ). (6.21)

Note that we have used different labels for the quantum numbers of total angular
momentum and the z-component angular momentum, l and .ml , in order to
distinguish these from those for molecular rotations. Thus,

L̂2 Ylml (θ, φ) = h̄2 l(l + 1)Ylml (θ, φ).

. (6.22)
176 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

Employing the above identity in Eq. (6.19), we obtain


ml h̄2 1 ∂ 2 h̄2 l(l + 1)
.Y
l (θ, φ) − rR(r) + R(r) + V (r)R(r) − ER(r) = 0.
2μ r ∂r 2 2μr 2
(6.23)

The above equation is satisfied for arbitrary value of .θ and .φ if the equation within
the square bracket is zero. Dividing the resulting equation with .−h̄2 /(2μr), we thus
obtain the following equation for the radial part of the eigenfunction:

d2
 
2μE 2μV (r) l(l + 1)
. (rR(r)) + − − (rR(r)) = 0. (6.24)
dr 2 h̄2 h̄2 r2

The above equation is general and applicable to any potential energy .V (r). In the
next subsection, we will consider the solution of this equation for the Coulomb
potential.

6.3.1 Radial Equation and Solution

Employing the expression for the Coulomb potential, Eq. (6.1), in Eq. (6.24), we
obtain

d2 2μZe2
 
2μE l(l + 1)
. (rR(r)) + + − (rR(r)) = 0. (6.25)
dr 2 h̄2 4π ǫ0 h̄2 r r2

To simplify the above equation, let us introduce a scaled and dimensionless radial
coordinate .ρ = αr r, where .αr is yet unknown parameter with the unit of inverse
length. Then, .dρ = αr dr. Let us also introduce .F (ρ) = R(r). Then, Eq. (6.25) is
equivalent to

d2 2μZe2 1
 
2μE l(l + 1)
. (ρF (ρ)) + + − (ρF (ρ)) = 0. (6.26)
dρ 2 h̄2 αr2 4π ǫ0 h̄2 αr ρ ρ2

The above equation can be simplified further by choosing .αr such that

2μZe2 2
αr =
. = , (6.27)
4π ǫ0 h̄2 a

where a is a generalization of the Bohr radius. For .Z = 1 and .μ = me , this becomes

the Bohr radius .a0 defined by Eq. (6.10). Thus, employing this parameter, one can
express Eq. (6.26) as follows:
6.3 Separation of Variables in Spherical Coordinate System 177

d2 μa 2 E
 
1 l(l + 1)
. (ρF (ρ)) + + − (ρF (ρ)) = 0. (6.28)
dρ 2 2h̄2 ρ ρ2

Solutions for the above equation can be found by first factoring out a term that
becomes dominant at large .ρ and then expanding the remaining function as a power
series with respect to .ρ. As detailed in the Appendix of this chapter, from the
recursion relationship between coefficients of the power series expansion, one can
show that physically well-behaving solutions exist only for the following values of
the energy:

μa 2 E 1
. =− , (6.29)
2h̄2 4n2

where n is a positive integer satisfying the following condition:

.n ≥ l + 1. (6.30)

Appendix of this chapter also provides some examples of the resulting radial
functions up to .n = 3, with the lowest order coefficient .C0 as yet undetermined for
each case. By determining the coefficient .C0 such that the following normalization
constant is satisfied,
∞
. drr 2 Rnl (r)Rn′ l (r) = δnn′ , (6.31)
0

one can full specify the radial functions. Some examples of these expressions are
provided in Table 6.1. Figure 6.1 provides plots of radial functions for .n = 0, · · · , 4
and .l = 0, and Fig. 6.2 provides those for .n = 4 and .l = 0, · · · , 3.

Table 6.1 Expressions for n l .Rnl (r)

radial functions, .Rnl (r)’s 3/2
1
1 0 .2
a e−r/a
3/2
1 r
e−r/(2a)

2 0 .
2a 2− a
3/2
1 1 r −r/(2a)
2 .1 .√
2a ae
3
3/2
2 1 r 2
27 − 18 ar + 2 e−r/(3a)

3 0 .
27 3a a
3/2
1 2 r r
e−r/(3a)
 
3 1 .
27 3a a 6− a
3/2
4 1 r 2 −r/(3a)

3 2 . √
3a a e
27 10
178 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

Fig. 6.1 Examples of four radial functions .Rn0 (r) (upper panel) and .r 2 |Rn0 (r)|2 (lower panel),
for n=1, 2, 3 ,and 4, versus r in the unit of a

Fig. 6.2 Examples of four radial functions .R4l (r) (upper panel) and .r 2 |R4l (r)|2 (lower panel), for
l=0, 1, 2 ,and 3, versus r in the unit of a

While the solution of the radial equation provided in Appendix is straightforward,

for a historical reason, the radial functions have been expressed more commonly in
terms of associated Laguerre polynomials [23]. Thus, the radial functions are in
general expressed as follows:
 1/2  l+ 3  
(n − l − 1)! 2 2 2r
Rnl (r) =
. r l
L2l+1
n−l−1 e−r/(na) , (6.32)
2n[(n + l)!]3 na na
6.3 Separation of Variables in Spherical Coordinate System 179

2l+1 2r
where .Ln−l−1 na represents an associated Laguerre polynomial and is related to
a Laguerre polynomial as follows:

dk
Lkn (x) = (−1)k
. Ln+k (x). (6.33)
dx k

In the above expression, .Ln (x) is the Laguerre polynomial of order n defined as

ex d n
x n e−x ,

Ln (x) =
.
n
(6.34)
n! dx
and is a solution of the following Laguerre equation:

d2 d
x
. f (x) + (1 − x) f (x) + nf (x) = 0. (6.35)
dx 2 dx

6.3.2 Radial Probability Density

It is important to note that only the information on the radial function is needed to
calculate the probability density to find the electron at a specific distance from the
nucleus. The corresponding radial probability density is given by

. Pnl (r) = r 2 [Rnl (r)]2 , (6.36)

which satisfies the following normalization condition:

∞
. drPnl (r) = 1. (6.37)
0

One can show that the average value of radius for each radial function is as follows:
∞ 
1

l(l + 1)

rnlml =
. drPnl (r)r = an2 1 + 1− . (6.38)
0 2 n2

As an example, consider the case where .n = 1 and .l = 0, for which

4 2 −2r/a
P10 (r) =
. r e . (6.39)
a3
Taking the derivative of this with respect to r, we can determine the most likely
value of r by using the following condition:

2r 2 −2r/a
 
d 4
. P10 (r) = 3 2re−2r/a − e = 0. (6.40)
dr a a
180 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

The above equation is satisfied for .r = a. In other words, a is the value of the radius
for which the state with .n = 1 and .l = 0 has the maximum probability. The radial
probability can also be used to determine the average potential energy as follows:

∞ 
e2

V (r)10 =
. drP10 (r) −
0 4π ǫ0 r
e2 4 ∞ e2 1 h̄2 1
=− drre−2r/a = − =− = 2E1 .
4π ǫ0 a 3 0 4π ǫ0 a me a 2
(6.41)

The above result shows that the average potential energy is equal to the value of the
potential at the most probable value of r, an outcome specific for .1/r dependence
of the potential energy. This also explains why the simple Bohr’s model, which
relied on classical-like relationship employing the most probable value of r, was
successful in reproducing the experimental data.

6.3.3 Eigenfunctions and Eigenstates in the Dirac Notation

Combining the radial function with the spherical harmonics, we can now write down
the full eigenfunction of .Ĥ for each eigenvalue.

Eigenfunctions for a Hydrogen-Like System

The eigenfunction for the electron in a hydrogen atom or hydrogen-like ion
with nuclear charge Z is

ψnlml (r, θ, φ) = Rnl (r)Ylml (θ, φ).

. (6.42)

This satisfies the following Schrödinger equation:

 
h̄2 1 ∂ 2 1 2 Ze2
. − r + L̂ − ψnlml (r, θ, φ) = En ψnlml (r, θ, φ),
2μ r ∂r 2 2μr 2 4π ǫ0 r
(6.43)

where .L̂2 is the angular momentum operator given by Eq. (6.20) and .En is the
eigenvalue of the Hamiltonian given by

h̄2 1 μZ 2 e4 1
En = −
.
2 2
=− . (6.44)
2μa n 32π 2 ǫ02 h̄2 n2
(continued)
6.3 Separation of Variables in Spherical Coordinate System 181

All the eigenfunctions defined by Eq. (6.42) satisfy the following orthonor-
mality condition:

∞ π 2π
. drr 2 dθ sin θ dφ ψn′ l ′ ml ′ (r, θ, φ)ψnlml (r, θ, φ) = δnn′ δll ′ δml m′l .
0 0 0
(6.45)

Having identified all the eigenfunctions of .Ĥ , it is now convenient to introduce

the eigenstate in the Dirac notation, .|n, l, m, which is defined through the following
relationship:

ψnlml (r, θ, φ) = r|n, l, ml .

. (6.46)

Since . dr |rr| = 1̂, the identity operator, the above definition along with
Eq. (6.45) implies that

n′ , l ′ , ml ′ |n, l, ml  = δnn′ δll ′ δml ml ′ .

. (6.47)

Employing these, all the results obtained so far can be expressed succinctly in the
Dirac notation as summarized below.

Eigenstates in the Dirac Notation

For the eigenvalue .En given by Eq. (6.44), there is an eigenstate .|n, l, ml  that
satisfies the following identities:

p̂2 Ze2
 
. Ĥrel |n, l, ml  = − |n, l, ml  = En |n, l, ml , . (6.48)
2μ 4π ǫ0 r̂
L̂2 |n, l, ml  = h̄2 l(l + 1)|n, l, ml , . (6.49)
L̂z |n, l, ml  = h̄ml |n, l, ml , (6.50)

where .l = 0, · · · , n − 1 and .ml = −l, · · · , l.

The fact that .[Ĥrel , L̂2 ] = 0 and .[Ĥrel , L̂z ] = 0 explains why there is a
simultaneous eigenstate of the three operators, .Ĥrel , .L̂2 , and .L̂z , which is .|n, l, ml 
shown above. The set of numbers .(n, l, ml ) is called good quantum numbers, which
represent a set of compatible quantum numbers representing conserved quantities.
In this set, n is called the principal quantum number, l is called the angular
182 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

momentum quantum number, and .ml is the magnetic quantum number. The value
of l is also often represented by the symbols shown below.

Symbols for Angular Momentum Quantum Numbers

l 0 1 2 3 4

s p d f g

6.3.4 Zeeman Effect

In the absence of an external field, all the states with different values of l and .ml for
the same value of n are degenerate. On the other hand, states with different values
of .ml have different values of energy in the presence of a magnetic field, which is
typically assumed to be in the z-direction. This is called Zeeman effect and is the
reason why .ml is called magnetic quantum number.
In the presence of a magnetic field, the potential energy of system with magnetic
moment is given by

V = −mL · B.
. (6.51)

where .mL is the magnetic moment vector with its magnitude defined as follows:

|mL | = Current × Area surrounded by current

.

qv 1
= I · πr2 = π r 2 = qrv. (6.52)
2π r 2
In a vector notation, the magnetic moment is given by
q q
mL =
. r×p= L. (6.53)
2m 2m
For an electron,
e
mL = −
. L, (6.54)
2me

where .L is the angular momentum of the electron. Thus, in the presence of a

magnetic field .B = Bz ez , the potential energy of the electron due to the magnetic
6.3 Separation of Variables in Spherical Coordinate System 183

field is given by
 
e eBz βB
V =− −
. L ·B= Lz = B z Lz , (6.55)
2me 2me h̄

where .βB = eh̄/(2me ) and is called Bohr magneton. Thus, in the presence of the
magnetic field, the total Hamiltonian is given by

βB
ĤB = Ĥrel +
. Bz L̂z . (6.56)
h̄

Since .[Ĥrel , L̂z ] = 0, the eigenstate of .Ĥrel , .|n, l, ml  still remains the eigenstate of
ĤB . The only difference is that the eigenvalue now becomes
.

h̄2 1
En,ml = −
. + βB Bz ml . (6.57)
2μa 2 n2

This expression provides quantitative details of the Zeeman effect, and shows that
the energy splitting is proportional to both .ml and .Bz .

6.3.5 Real-Valued Orbital Functions

As mentioned above, .|n, l, ml  or .ψnlml (r, θ, φ) has the same energy in the absence
of external field as long as n remains the same. Thanks to this degeneracy, any linear
combination of .Ylml (θ, φ) for the same l can serve as simultaneous eigenstate of the
Hamiltonian .Ĥrel and .L̂2 . Namely, the following linear combination leads to another
eigenstate of the same eigenvalue.
 
ml
Cml Ylml (θ, φ)

.Rnl (r)Y
l (θ, φ) → Rnl (r) . (6.58)
ml

This fact can be used to create real functions out of complex valued .Ylml (θ, φ)’s. For
example, for .l = 1, out of

3 1/2

. Y11 (θ, φ)
=− sin θ eiφ , . (6.59)
8π
 1/2
−1 3
Y1 (θ, φ) = sin θ e−iφ , (6.60)
8π
184 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

one can create

 1/2
1 3
. px = √ Y1−1 (θ, φ) − Y11 (θ, φ) = sin θ cos φ, . (6.61)
2 4π
 1/2
i 3
py = √ Y1−1 (θ, φ) + Y11 (θ, φ) = sin θ sin φ. (6.62)
2 4π

Along with .pz defined as

 1/2
3
pz =
. Y10 (θ, φ) = cos θ, (6.63)
4π

.px and .py given by Eqs. (6.61) and (6.62) form orthonormal and real eigenfunctions
of .L̂2 . However, note that these are no longer eigenfunctions of .L̂z . Similarly, for
.l = 2, one can form the following five real valued angular functions:

 1/2
5
. dz2 = Y20 = 3 cos2 θ − 1 , . (6.64)
16π
 1/2
1 15
dxz = √ Y2−1 − Y21 = sin θ cos θ cos φ, . (6.65)
2 4π
15 1/2
 
i
dyz = √ Y2−1 + Y21 = sin θ cos θ sin φ, . (6.66)
2 4π
15 1/2 2
 
1
dx 2 −y 2 = √ Y22 + Y2−2 = sin θ cos 2φ, . (6.67)
2 16π
15 1/2 2
 
i −2 2
dyz = √ Y2 − Y2 = sin θ sin 2φ, (6.68)
2 16π

All of the above functions are orthonormal and are eigenfunctions of .L̂2 . However,
except for .dz2 , none is the eigenfunction of .L̂z any more.

6.4 Spin States

Imagine a particle circling around a circle of radius r with speed v. Assume that
r → 0 while .|v| ∝ 1/r. In this limit, the particle spins around itself with infinitely
.

fast speed, but with finite angular momentum. The angular momentum in this
limiting case can be called spin. However, one should be careful because this is
just an analogy. Quantum mechanical spin cannot be visualized like an orbital
angular momentum wavefunction. On the other hand, it still behaves almost like
6.4 Spin States 185

conventional angular momentum with respect to the rotational transformation of the

space. As a consequence, they satisfy the same commutation relations as angular
momentum operators.

Commutation Relations and Eigenvalues of Spin Operators

The three components of the spin operator satisfy the following commutator
relations:

. [Ŝx , Ŝy ] = i h̄Ŝz , . (6.69)

[Ŝy , Ŝz ] = i h̄Ŝx , . (6.70)
[Ŝz , Ŝx ] = i h̄Ŝy . (6.71)

Thus, the spin state can be specified in terms of .Ŝ 2 = Ŝx2 + Ŝy2 + Ŝz2 and .Ŝz .
The eigenvalues of these operators are given by .S(S + 1) and .h̄ms , where
.S ≥ |ms |. The Dirac notation is very convenient for describing these spin

states. Namely, the spin eigenstates are defined as follows:

. Ŝ 2 |S, ms  = h̄2 S(S + 1)|S, ms , . (6.72)

Ŝz |S, ms  = h̄ms |S, ms . (6.73)

There are two major differences between the spin and the spatial angular
momentum as stated below.
• Spin is a unique intrinsic property of a particle like its mass.
• The total spin quantum number S can be half integers as well as full integers.
Particles with a half integer value of S have properties distinctively different from
those with a full integer value of S. The former type of particle is called fermion
(named after Fermi) and the latter is called boson (named after Bose).
Well known examples of fermions are electron, proton, and neutron, which all
have .S = 1/2. Well known examples of bosons are photon and deuteron, which all
have .S = 1. Fermions cannot share a quantum state, which is specified by the all
compatible orbital quantum numbers and the spin state. On the other hand, there is
no limit in the number of bosons that can occupy the same state. Therefore, at zero
or near zero temperature, bosons tend to occupy the ground state, which is called
Bose-Einstein condensation.
Since the spin of an electron is .1/2, it is a fermion and has the following two spin
states:
1 1 1
. |α ≡  , =  , . (6.74)
2 2 2
186 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

1 1  1
|β ≡  , − =  − , (6.75)
2 2 2
where the second equality is an abbreviation typically used because .S = 1/2
remains constant for electron. By definition, .α|α = 1, .β|β = 1, and .α|β =
β|α = 0. Combining the above spin states of an electron with those for spatial
orbitals, we can define the following two states:

n, l, ml , 1 = |n, l, ml |α = |n, l, ml  ⊗ |α, .

 
. (6.76)
2
n, l, ml , − 1 = |n, l, ml |β = |n, l, ml  ⊗ |β,
 
(6.77)
2
where .⊗ represent the direct product, which amounts to expanding the state space as
explained in Chap. 2. Note that this kind of direct product is possible between two
sets of states corresponding to independent variables. Another example of direct
product is .|r = |x ⊗ |y ⊗ |z = |x|y|z. As can be seen from these definitions,
direct product symbols are often omitted unless there is any source of confusion.
Taking inner products of Eqs. (6.76) and (6.77) with a position state, we also
obtain
 1
. r|n, l, ml , = r|n, l, ml |α = ψnlml (r, θ, φ)|α, . (6.78)
2
 1
r|n, l, ml , − = r|n, l, ml |β = ψnlml (r, θ, φ)|β. (6.79)
2
The spin of an electron also interacts with a magnetic field through its magnetic
moment defined as follows:
ge
ms = −
. S, (6.80)
2me

where .g = 2.002319 . . . and is called anomalous spin factor. Therefore, in the

presence of a magnetic field along the z-direction, the potential of the electron is

e βB
V = −(mL + ms ) · B =
. (L + gS) · B = Bz (Lz + gSz ). (6.81)
2me h̄

Therefore, the total Hamiltonian for an electron in a hydrogen-like system subject a

magnetic field is

βB
Ĥ = Ĥrel +
. Bz (L̂z + g Ŝz ). (6.82)
h̄
6.4 Spin States 187

States given by Eqs. (6.76) and (6.77) (or Eqs. (6.78) and (6.79) in position
representation) can be shown to be eigenstates of the above Hamiltonian, and the
corresponding eigenvalues are

h̄2 1
En,ml ,ms = −
. + βB Bz (ml + gms ). (6.83)
2μa02 n2

While the above effect is easy to understand, the fact that electron is in a frame
that continuously accelerates causes another effect of spin on the energy of the
electron in hydrogen-like system even when there is no magnetic field. The reason
is that the electron sees the nucleus as accelerating and therefore experiences the
magnetic field caused by its motion. This magnetic field then interacts with the spin
of the electron. Thus, the true Hamiltonian of the electron in a hydrogen-like system
should be expressed as

Ĥ = Ĥrel + ξ(r)L̂ · Ŝ,

. (6.84)

where .ξ(r) is a function that depends on the distance between the nucleus and the
electron. Solving the Schrödinger equation for this Hamiltonian is more difficult if
not impossible. On the other hand, the above form gives insight into what should be
good quantum numbers when the spin-orbit interaction is included.
Let us define the sum of the orbital angular momentum and the spin as follows:

J = L + S.
. (6.85)

Then, it is clear that

1  1 
L·S=
. (L + S)2 − L2 − S2 = J2 − L2 − S2 . (6.86)
2 2
Thus,

1 2 
L̂ · Ŝ =
. Ĵ − L̂2 − Ŝ2 . (6.87)
2
Therefore, the spin-orbit coupling can be determined once we know all three
eigenvalues of .Ĵ2 , .L̂2 , and .Ŝ2 . For this, let us consider the z-component of .Ĵ given
by

Jˆz = L̂z + Ŝz .

. (6.88)

Since the eigenvalue of .Ŝz for an electron is .±1/2, the eigenvalue of .Jˆz is expected
to be
1
jz = m l ± ,
. (6.89)
2
188 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

where .ml = −l, · · · , l. This means that the maximum magnitude of .jz is .l − 1/2
or .l + 1/2 depending on the value of spin. Otherwise, the eigenvalue of .Ĵ2 has the
same structure as .L̂2 and .Ŝ2 as follows:
1
Ĵ2 → h̄2 j (j + 1), where j = l ± .
. (6.90)
2
Note that, for the case where .l = 1/2, only .j = 1/2 is possible.

6.5 Electronic Transitions and Term Symbols

In an electronic transition of a hydrogen-like system due to linear interaction with

light (single photon transition), the following selection rules apply.

Selection Rules of Electronic Transitions in Hydrogen-Like Systems

. l = ±1, . (6.91)
S = 0, . (6.92)
j = 0, ±1 (No 0 → 0 transition). (6.93)

The orbital angular momentum selection rule, Eq. (6.91), can be understood from
the fact that (1) the total angular momentum of the electron plus the spin of photon,
which is equal to one, has to be conserved and (2) that the transition is possible only
between states with different odd/even symmetry with respect to the inversion of
the space. This latter rule exists because the interaction with light occurs through
the dipole operator, which has odd spatial symmetry. Due to this constraint, .l = 0
is not possible.
The spin selection rule, Eq. (6.92), results from the fact that the spin of an electron
does not interact with the electric field of the radiation. Finally, Eq. (6.93) can be
understood from the fact that .J = L + S. Even though .l = 0 is not possible,
.j = 0 can occur through the contribution of the spin having a direction different

from that of the orbital angular momentum.

Although this chapter is focused on a hydrogen-like system with only one
electron, it is useful to introduce here more general term symbols used for describing
many electron atoms as described below.
6.5 Electronic Transitions and Term Symbols 189

Definition of Atomic Term Symbols

Atomic term symbols provide compact ways to represent the angular momen-
tum, spin, and sum of the two, for electrons in an atom, and are expressed
as
2S+1
. TJ , (6.94)

where .T = S (for .L = 0), .P (for .L = 1), .D (for .L = 2), . . . . Namely, these are
capital letters for the symbols used for hydrogen-like orbital functions. The
left superscript denotes the spin multiplicity .2S + 1, and the right subscript is
the value of J .

For a hydrogen-like system, .2S + 1 = 2. Thus, possible electronic term symbols

for a hydrogen-like system are as follows:
2
. S1/2 ,2 P1/2 ,2 P3/2 ,2 D3/2 ,2 D5/2 , · · · (6.95)

For the consideration of many-electron systems in the next chapter, it is also

useful to introduce the general rule of angular momentum addition here. Classically,
when two angular momenta, .L1 and .L2 , are added to become .L = L1 +L2 , possible
magnitudes of .L are in the following range:

||L1 | − |L2 || ≤ |L| ≤ |L1 | + |L2 |.

. (6.96)

Quantum mechanically, this condition remains true except that the quantization
condition has to be satisfied as well. That is, the eigenvalue of .L̂2 is as follows:

L̂2 → h̄2 L(L + 1), L = |L1 − L2 |, |L1 − L2 | + 1, · · · , L1 + L2 − 1, L1 + L2 ,

.

(6.97)

where .L1 is a quantum number for .L̂21 , for which the corresponding eigenvalue is
2 2
.h̄ L1 (L1 + 1), and .L2 is that for .L̂ , for which the eigenvalue is .h̄L2 (L2 + 1).
2
Because the above identity is related to the symmetry property of the angular
momentum operator, the same relation holds for spin and the total angular momen-
tum (the sum of spin and orbital angular momentum) as well. Namely, for .S =
S1 + S2 ,

Ŝ2 → h̄2 S(S + 1) , with S = |S1 − S2 |, · · · , S1 + S2 ,

. (6.98)
190 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

where .S1 and .S2 are quantum numbers for .Ŝ12 and .Ŝ22 , which respectively have
eigenvalues .h̄2 S1 (S1 +1) and .h̄S2 (S2 +1). The same relation also holds for .J = L+S
as follows:

Ĵ2 → h̄2 J (J + 1) , with J = |L − S|, · · · , L + S,

. (6.99)

where L and S are quantum numbers for .L̂2 and .Ŝ 2 , which respectively have
eigenvalues .h̄2 L(L + 1) and .h̄S(S + 1).

6.6 Summary and Questions

Quantized energy levels of an electron in a hydrogen-like system, according to

Bohr’s model, can be explained assuming a classical-like motion of the electron
orbiting around the nucleus, and requiring only one additional condition. It is to
assume that the circular orbit of the electron is an integer multiple of its de Broglie
wavelength, which in turn can be calculated by the value of the momentum that
maintains the balance between the centrifugal force and the Coulomb attraction by
the nucleus. The resulting expression is Eq. (6.12), which becomes exact if the mass
of an electron is replaced with the reduced mass for the system.
The success of Bohr’s model indicates that the physical origin of the quantized
energy levels of an electron is the periodic nature of its motion. Bohr’s model
captures this in the simplest manner by incorporating the duality of a particle-
like property and a wave-like nature of the electron into the requirement of energy
conservation described in a classical manner.
For more complete description of the electron, a complete solution of the time
independent Schrödinger equation is necessary. This becomes possible by using
a spherical coordinate system and employing the solution for the rotational part,
the spherical harmonics, to derive an equation for the radial part, Eq. (6.25). This
equation can be solved exactly, resulting in radial functions shown in Table 6.1
as examples. A general expression involving Laguerre function and an exponential
function is given by Eq. (6.32). Combination of this with the spherical harmonics
results in the full expression for the eigenfunction, Eq. (6.42), for which the
eigenvalue is given by Eq. (6.44). Alternatively, these can be expressed in the
Dirac notation as .|n, l, ml , which represents a simultaneous eigenstate of the the
Hamiltonian, the square of the angular momentum, and the z-component of the
angular momentum according to Eqs. (6.48)–(6.50).
Because an electron has a spin of magnitude .h̄/2, for which there can be two spin
states, up and down, as shown in Eqs. (6.74) and (6.75), a complete specification
of the quantum state of an electron requires consideration of both the spin and
spatial orbital states, as shown in Eqs. (6.76) and (6.77). Although many of these
states are degenerate, having the same energy for the same principal quantum
number, application of an external magnetic field can easily lift such degeneracy
6.6 Summary and Questions 191

since the corresponding eigenvalue is given by Eq. (6.83). At finer scale, it turns out
that further complications arise due to spin-orbit effects for example. A heuristic
description of the spin-orbit effects has been provided, which shows that the
information on the sum of the orbital angular momentum and the spin is needed
for complete specification of the eigenstate. In fact, the corresponding quantum
number, normally denoted as J turns out to be important in general and serves as
an important quantum number for fully specifying many electron states as well as
a single electron state. The atomic term symbol given by Eq. (6.94) conveys this
information succinctly.
Selection rules for electronic transitions involving interaction of an electron with
one photon are provided by Eqs. (6.91)–(6.93). Term symbols are introduced in
order to make the description of these electronic transitions in a succinct manner
and also for the consideration of more general many-electron states that will be
treated in the next chapter.

Questions
• Bohr’s model can predict exact energy levels of an electron in a hydrogen-
like system despite its simplicity. How is this possible?
• Energy levels of an electron in a hydrogen-like system are independent
of the orbital angular momentum quantum number? Why does such
independence occur despite that the radial equation in fact depends on the
orbital angular momentum quantum number?
• For eigenfunctions of the electron in a hydrogen-like system, what is the
sum of the number of radial nodes in the radial function and that of nodal
planes in the angular part of the eigenfunction? How does it depend on the
principal quantum number and angular momentum quantum number?
• For a given principal quantum number, does the increase of the orbital
angular momentum quantum number make the average distance between
the electron and the nucleus larger or smaller?
• For a given principal quantum number, does the increase of the orbital
angular momentum quantum number make the probability of finding the
electron near the nucleus higher or lower?
• What is the common property of the orbital angular momentum and the
spin of an electron and what are major differences?
• What causes the coupling between the spin and the orbit angular momen-
tum of an electron in a hydrogen-like system?
• In the transition of the electronic energy level in a hydrogen-like system
due to emission or absorption of one photon, the orbital angular momentum
quantum number does not remain the same. What is the reason for this?
192 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

Appendix: Solutions of the Radial Equation

We here consider in more detail the solution of Eq. (6.28) for the bound state .E < 0.
Let us introduce

μa 2 E
.γ = − , (6.100)
2h̄2

and another function .G(ρ) such that

ρF (ρ) = e−γρ G(ρ).

. (6.101)

Then,

d2 d2
 
−γρ 2 d
. (ρF (ρ)) = e γ G(ρ) − 2γ G(ρ) + G(ρ) . (6.102)
dρ 2 dρ dρ 2

Using Eqs. (6.101) and (6.102) in Eq. (6.28) and factoring out .e−γρ , we find that

d2
 
d 1 l(l + 1)
.
2
G(ρ) − 2γ G(ρ) + − G(ρ) = 0. (6.103)
dρ dρ ρ ρ2

Now let us assume that

∞

G(ρ) =
. Ck ρ k+p , (6.104)
k=0

where p and .Ck ’s need to be determined so as to satisfy Eq. (6.103). Using the above
expression and its first and second derivatives with respect to .ρ in Eq. (6.103), we
obtain
∞ 

. Ck (k + p)(k + p − 1)ρ k+p−2 − 2γ Ck (k + p)ρ k+p−1
k=0

+Ck ρ k+p−1 − Ck l(l + 1)ρ k+p−2 = 0. (6.105)

Rearranging terms, the above equation can be expressed as

. C0 (p(p − 1) − l(l + 1)) ρ p−2

∞

+ {Ck+1 ((k + p + 1)(k + p) − l(l + 1)) − Ck (2γ (k + p) − 1)} ρ k+p−1 = 0.
k=0
(6.106)
Appendix: Solutions of the Radial Equation 193

From the first term in the above equation, we obtain the following condition for p:

p(p − 1) − l(l + 1) = (p + l)(p − l − 1) = 0.

. (6.107)

Thus, .p = −l or .l + 1. For nonzero l, the former case results in a function that

cannot be normalized due to the singularity at .r = 0. For .l = 0, the solution with
.p = 0 still appears to be possible because it leads to a normalizable wavefunction.

However, when this function is included as one of the basis function, the resulting
kinetic energy operator is no longer Hermitian. Therefore, all cases with .p = −l
should be excluded and we can use the latter option of .p = l + 1 only.
On the other hand, for the summation in Eq. (6.106) to be zero, the following
recursion relation has to be satisfied.
2γ (k + p) − 1
Ck+1 =
. Ck . (6.108)
(k + p + 1)(k + p) − l(l + 1)

Let us first assume that the series do not terminate and .Ck ’s remain nonzero for all
k. Then, in the large k limit of .k >> l, p,

2γ
Ck+1 ≈
. Ck . (6.109)
(k + 1)

This means that .G(ρ) ∼ e2γρ for large .ρ, which again means that .ρF (ρ) ∼ eγρ .
This is an unphysical solution. Therefore, the series has to terminate at some finite
order, which is possible as long as the following condition for some .k ≥ 0 is
satisfied.
1 1
γ =
. = , k = 0, 1, 2, · · · (6.110)
2(k + p) 2(k + l + 1)

In the second equality of the above expression, we have used the fact that .p = l + 1.
Now let us introduce

n = k + l + 1,
. k = 0, 1, 2, · · · (6.111)

Then, .n ≥ l + 1 and Eq. (6.110) implies that

1
γ =
. . (6.112)
2n
Combining this with the definition of .γ given by Eq. (6.100), we obtain the
following expression for the energy:

h̄2 1
En = −
. , n = 1, 2, · · · (6.113)
2μa 2 n2

where n was also used as a subscript to label each energy.

194 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

For a given .n = kmax + l + 1 and .l = p − 1, the recursion relation, Eq. (6.108),

can be expressed as

n−k−l−1
Ck+1 = −
. Ck , k = 0, · · · , kmax (6.114)
n(k + 1)(k + 2l + 2)

Using this recursion relationship, it is straightforward to find expressions for all of

the .Gnl (ρ), some of which are listed below.

. G10 (ρ) = C0 ρ, . (6.115)

ρ
G20 (ρ) = C0 ρ 1 − ,. (6.116)
4
G21 (ρ) = C0 ρ 2 , . (6.117)
 
1 1 2
G30 (ρ) = C0 ρ 1 − ρ + ρ , . (6.118)
3 54
 
1
G31 (ρ) = C0 ρ 2 1 − ρ , . (6.119)
12
G32 (ρ) = C0 ρ 3 . (6.120)

Therefore, the corresponding expressions for .Fnl (ρ) are as follows:

. F10 (ρ) = C0 e−ρ/2 , . (6.121)

ρ −ρ/4
F20 (ρ) = C0 1 − e ,. (6.122)
4
F21 (ρ) = C0 ρe−ρ/4 , . (6.123)
 
1 1 2 −ρ/6
F30 (ρ) = C0 1 − ρ + ρ e ,. (6.124)
3 54
 
1
F31 (ρ) = C0 ρ 1 − ρ e−ρ/6 , . (6.125)
12
F32 (ρ) = C0 ρ 2 e−ρ/6 . (6.126)

Note that .ρ = 2r/a in above equations. Using an appropriate value of .C0 for each
case that makes the resulting radial wavefunction normalized, we obtain the results
listed in Table 6.1.
Exercise Problems with Solutions 195

Exercise Problems with Solutions

6.1 Calculate the first three transition energies (in the units of cm−1 ) for Lyman,
Balmer, and Paschen like series for the He+ ion.
Solution 6.1 The reduced mass for electron in helium is
mN me 4.002603254
μ=
. = me = 0.999863me .
mN + me 4.002603254 + 0.00054854

The nuclear charge of He+ is equal to 2. Therefore, the energy level of electron in
He+ is given by

1
En = −4RH e+
. , n = 1, 2, · · ·
n2

where RH e+ = 0.99863 × 1.0974 × 105 cm−1 = 1.0972 × 105 cm−1 . Thus, the
three Lyman series, which correspond to transitions from n > 1 to n = 1, are as
follows:
 
1
. 4RH e+ 1− = 3.2916 × 105 cm−1 ,
4
 
1
4RH e+ 1 − = 3.9012 × 105 cm−1 ,
9
 
1
4RH e 1 −
+ = 4.1144 × 105 cm−1 .
16

Next, the Balmer series, which correspond to transitions from n > 2 to n = 2, are
as follows:
 
1 1
. 4RH e+ − = 6.0956 × 104 cm−1 ,
4 9
 
1 1
4RH e+ − = 8.2292 × 104 cm−1 ,
4 16
 
1 1
4RH e+ − = 9.2164 × 104 cm−1 .
4 25
196 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

Finally, the Paschen series, which correspond to transitions from n > 3 to n = 3,

are as follows:
 
1 1
. 4RH e+ − = 2.1334 × 104 cm−1 ,
9 16
 
1 1
4RH e+ − = 3.1209 × 104 cm−1 ,
9 25
 
1 1
4RH e+ − = 3.6573 × 104 cm−1 .
9 36

6.2 Prove that R20 (r) and R21 (r) in Table 6.1 are solutions satisfying the differen-
tial equation, Eq. (6.28). You do not need to include the normalization constant in
proving this.
Solution 6.2 Let us use ρ = 2r/a. Then, without including the normalization
factor,
r −r/(2a) ρ −ρ/4
. R20 (r) = 2 − e = 2− e ,
a 2
r −r/(2a)
R21 (r) = e = ρe−ρ/4 .
a

The fact that R20 (r) is the solution of the radial equation is proved as follows.

d2
 
−ρ/4 1 1 ρ −ρ/4
. ρ(2 − ρ)e + − + ρ 2− e
dρ 2 16 ρ 2
ρ 2 −ρ/4
   
d 1
= (2 − ρ)e−ρ/4 − 2ρ − e
dρ 4 2
 
1 1 ρ −ρ/4
+ − + ρ 2− e
16 ρ 2
ρ2
  
1 1 1
= −1 − (2 − ρ) − (2 − ρ) + 2ρ − e−ρ/4
4 4 16 2
ρ2
   
1 ρ
+ − 2ρ − + 2− e−ρ/4 = 0.
16 2 2
Exercise Problems with Solutions 197

The fact that R21 (r) is the solution of the radial equation is proved as follows.

d2
 
2 −ρ/4 1 1 2
. ρ e + − + − 2 ρ 2 e−ρ/4
dρ 2 16 ρ ρ
ρ 2 −ρ/4
   
d −ρ/4 1 1 2
= 2ρe − e + − + − 2 ρ 2 e−ρ/4
dρ 4 16 ρ ρ
ρ2 ρ2
 
ρ ρ
= 2− − + − + ρ − 2 e−ρ/4 = 0.
2 2 16 16

6.3 Prove that R20 (r) and R21 (r) in Table 6.1 are not orthogonal to each other. You
do not need to include the normalization constant in proving this.
Solution 6.3 This can be shown by direct integration as follows.
∞ ∞ r −r/(2a) r −r/(2a)
. drr 2 R20 (r)R21 (r) = C drr 2 2 −
e e
0 0 a a
C ∞ r 4 −r/a
 
3
= dr 2r − e
a 0 a
2 1
= C( 3!a 4 − 2 4!a 5 ) = −12Ca 3 ,
a a
where C is a nonzero constant that comes from the normalization factors of radial
functions. Since this is non-zero, the above shows that R20 (r) and R21 (r) are non-
orthogonal.
6.4 For the radial wavefunction R32 (r) in Table 6.1, determine the most probable
value of r, and the average potential energy.
Solution 6.4 The radial probability density for this radial wavefunction is as
follows:

.P32 (r) = r 2 [R32 (r)]2 = Nr 2 · r 4 e−2r/(3a) ,

where N = 8/(5 · 36 a 7 ). Taking derivative of P32 (r) with respect to r,

 
d 2
. P32 (r) = N · 6r 5 e−2r/(3a) − N r 6 e−2t/(3a) −
dr 3a
 
2
= N 6− r e−2r/(3a) r 5 = 0.
3a
198 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

The above equation is satisfied for r = 9a0 . The average potential energy is
calculated as follows:
∞ 
e2

.V (r)32 = drP32 (r) −
0 4π ǫ0 r
2  ∞
e2
 
4 1
=− √ drr 5 e−2r/(3a)
4π ǫ0 27 10 27a 7 0
 2    6
e 4 1 3a
=− √ 5!
4π ǫ0 27 10 27a 7 2
e 1
=− .
4π ǫ0 9a

6.5 For an isolated hydrogen atom in space, the average distance of its electron
from the nucleus is known to be 1 mm. Then, what is the range of possible values
of its energy?
Solution 6.5 Let us first calculate the range of average distance for a given principal
quantum number n employing Eq. (6.38). If l = 0,

3 2
rn00 =
. an .
2
For l = n − 1,
   
3 2 2 1 2
rnn−1ml = a
. n −n +n =a n +n .
2 2

Thus, if we assume that l = 0, the average distance of 1 mm corresponds to the

following minimum principal quantum number:
1/2
2 1 × 10−3

nmin =
. ≈ 3550.
3 5.29 × 10−11

For this,

1.0968 × 105
Emin = −
. cm−1 = −8.703 × 10−3 cm−1 .
(3550)2

On the other hand, if we assume that l = n − 1, the average distance of 1 mm,

the maximum principal quantum number is the solution of the following quadratic
equation:

1 2 1 × 10−3
. n +n= .
2 5.29 × 10−11
Exercise Problems with Solutions 199

Solving this quadratic equation, we find that nmax ≈ 6148. For this,

1.0968 × 105
Emax = −
. cm−1 = −2.902 × 10−3 cm−1 .
(6148)2

Thus, the range of energies is −8.703 × 10−3 cm−1 ∼ −2.902 × 10−3 cm−1 or
−1.079 × 10−6 eV ∼ −3.598 × 10−7 eV.
6.6 Prove that px orbital is an eigenfunction of L̂x and determine its eigenvalue.
Solution 6.6 This can be shown by applying L̂x (in position representation)
directly to px (with normalization constant omitted) as follows:
 
∂ ∂
. L̂x px = i h̄ sin φ + cot θ cos φ sin θ cos φ
∂θ ∂φ
= i h̄ (sin φ cos θ cos φ + cot θ cos φ sin θ (− sin φ))
= i h̄ (cos θ sin φ cos φ − cos θ cos φ sin φ) = 0 = 0px .

This proves that px is an eigenfunction of L̂x with eigenvalue 0.

6.7 Prove that dxz orbital is not an eigenfunction of L̂z .

Solution 6.7 Let us apply L̂z to dxz orbital (with normalization constant omitted)
as follows:
h̄ ∂ h̄
. sin θ cos θ cos φ = − sin θ cos θ sin φ.
i ∂φ i

The final expression in the above expression is not proportional to dxz . Therefore,
this is not an eigenfunction of L̂z .

6.8 For two electrons 1 and 2 with L1 = 0 and L2 = 1, respectively, list all possible
values of L, S, and J , and write all the electronic terms representing states possible
for this system of two electrons.
Solution 6.8 The resulting values of L and S are as follows: L = 1, S = 0, 1. For
these values of L and S, J = 0, 1, 2. The resulting terms are as follows: 3 P0 , 3 P1 ,
3P , 1P .
2 1

6.9 For two electrons 1 and 2 with L1 = 1 and L2 = 1, respectively, list all possible
values of L, S, and J , and write all the electronic terms representing states possible
for this system of two electrons.
200 6 Hydrogen-Like Systems and Spin Orbit States of an Electron

Solution 6.9 Possible values of L and S, and the corresponding values of J and
terms are shown in the table below.

L S J Terms
0 0 0 1S
0
0 1 1 3S
1
1 0 1 1P
1
1 1 0, 1, 2 3P , 3P , 3P
0 1 2
2 0 2 1D
2
2 1 1, 2, 3 3D , 3D , 3D
1 2 3

6.10 For two electrons 1 and 2 with L1 = 2 and L2 = 3, respectively, list all
possible values of L, S, and J , and write all the electronic terms representing states
possible for this system of two electrons.
Solution 6.10 Possible values of L and S, and the corresponding values of J and
terms are shown in the table below.

L S J Terms
1 0 1 1P
1
1 1 0, 1, 2 3P , 3P , 3P
0 1 2
2 0 2 1D
2
2 1 1, 2, 3 3D , 3D , 3D
1 2 3
3 0 3 1F
3
3 1 2, 3, 4 3F , 3F , 3F
2 3 4
4 0 4 1G
4
4 1 3, 4, 5 3G , 3G , 3G
3 4 5
5 0 5 1H
5
5 1 4, 5, 6 3H , 3H , 3H
4 5 6

Problems

6.11 Calculate the first three transition energies (in the units of cm−1 ) for Lyman,
Balmer, and Paschen like series for the Li2+ ion.
6.12 For the radial wavefunction R31 (r) in Table 6.1, determine the most probable
value of r, and the average potential energy.
6.13 The electron of a hydrogen atom is in a state with principal quantum number
n = 2 and angular momentum quantum number l = 1. Calculate its average kinetic
and potential energies.
6.14 For all the radial functions listed in Table 6.1, confirm Eqs. (6.37) and (6.38).
Problems 201

6.15 For all the radial functions listed in Table 6.1, calculate r̂ = r̂ 2  − r̂2 .
6.16 An electron in a hydrogen atom is in a state represented by the following
wavefunction: ψ(r) = C 2e−r/a0 + (r/a0 )e−r/(2a0 ) . This is not an eigenstate of
the Hamiltonian. Explain why and calculate the expectation value of the energy?
6.17 The electron in a hydrogen atom is in a state represented by the following
wavefunction:
3/2
√

1 1 r
ψ(r, θ, φ) = √
. 2 + ( 3 sin θ cos φ − 1) e−r/(2a) .
24π 2a a

(a) Prove that this is an eigenstate of the Hamiltonian and determine its eigenvalue.
(b) Prove that this is not an eigenstate of L̂2 and determine its expectation value.
6.18 For two electrons 1 and 2 with L1 = 1 and L2 = 3, respectively, list all
possible values of L, S, and J , and write all the electronic terms representing states
possible for this system of two electrons.
6.19 For two electrons 1 and 2 with L1 = 2 and L2 = 2, respectively, list all
possible values of L, S, and J , and write all the electronic terms representing states
possible for this system of two electrons.
Chapter 7
Approximation Methods for Time
Independent Schrödinger Equation

The simplicities of natural laws arise through the complexities

of the language we use for their expression.
— Eugene Wigner

Abstract This chapter is focused on two standard methods of approximating the

solution of a time independent Schrödinger equation, the variational principle
and the time independent perturbation theory. The variational principle states that
the true ground state energy of a quantum system is always smaller than any
approximation, for which a simple quantum mechanical proof is provided. Then,
assuming a trial state given by a a linear combination of predetermined basis states,
which are not necessarily orthogonal, an eigenvalue equation for the coefficients of
the linear combination and the energy is derived. Solution of this equation results
in the best choice of the linear combination and the corresponding variational
approximation for the ground state energy. The perturbation theory is applicable for
the case where the Hamiltonian of a system is the sum of a zeroth order Hamiltonian
term with known solutions and a small perturbation term. First, for the case where
there is no degeneracy in the zeroth order Hamiltonian, general expressions for the
first and second order corrections for both energies and states are derived. Then, for
the case where the zeroth order Hamiltonian is degenerate, corrections of the energy
up to the second order and of the state up to the first order of the perturbation are
derived.

Analytic solutions for the time independent Schrödinger equation exist only for very
few systems. For most cases, one has to resort to either approximation methods or
numerical solutions. The variational principle and the time independent perturbation
theory serve as two major methods for such approximation. The objective of this
chapter is to explain the theoretical bases of these approximation methods and derive
important expressions that can be employed for actual applications.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 203
S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_7
204 7 Approximation Methods for Time Independent Schrödinger Equation

7.1 Variational Principle

The variational principle provides a general strategy of approximation for a broad

range of problems. It has long been used widely for solving any kind of eigenvalue
or optimization problems for the cases where the quantity of interest is bounded. The
idea behind the variational principle is simple. Let us consider any mathematical
operator .Ô defined with respect to a certain set of functions, .Sf , and that one can
identify a set of normalized and orthogonal eigenfunctions .φn (x)’s such that any
function in .Sf can be determined as a unique linear combination of .φn (x)’s. In
other words, let us assume that there is a unique expression for any function .f (x)
in .Sf as follows:

.f (x) = C1 φ1 (x) + C2 φ2 (x) + · · · + Ck φk (x) + · · · . (7.1)

Then,

. dxf ∗ (x)Ôf (x) = |C1 |2 λ1 + |C2 |2 λ2 + · · · + |Ck |2 λk + · · · , (7.2)

where ∗.λk is an eigenvalue for .φk (x) such that .Ôφk (x) = λk φk (x) and the fact that
. dxφk (x)φj (x) = δkj has been used. In case all .λk ’s are real valued and bounded
from below (or above), it is clear that Eq. (7.2) divided by a proper normalization
constant should also be bounded the same way. Although functions depending on
only one variable x were considered above, the same principle holds for any function
that depends on multiple variables. The Hamiltonian operator in quantum mechanics
has all the properties validating these assumptions, which can be utilized for finding
approximations in case exact solutions are difficult to find.

7.1.1 General Case

The most common use of the variational principle in quantum mechanics is

finding out the approximation for the ground state energy. The underlying theorem,
employing the Dirac notation, can be stated as follows.
Theorem 7.1 Consider a system represented by a Hamiltonian operator .Ĥ , which
is Hermitian. The expectation value of the Hamiltonian for any state .|φ, which is
not necessarily normalized, is always bounded from below by its true ground state
energy .Eg as follows:

φ|Ĥ |φ
Eφ =
. ≥ Eg . (7.3)
φ|φ
7.1 Variational Principle 205

Proof Let us denote the complete set of eigenstates of .Ĥ as .|ψk ’s, which are
normalized. Thus, any state can be represented in terms of linear combination of
these states as follows:

.|φ = ak |ψk , (7.4)
k

and its norm is given by the sum of absolute squares of .ak ’s as follows:
  
φ|φ =
. ak ak∗′ ψk ′ |ψk  = ak ak∗′ δkk ′ = |ak |2 . (7.5)
k k′ k k′ k

On the other hand,

  
.φ|Ĥ |φ = ak ak∗′ ψk ′ |Ĥ |ψk  = ak ak∗′ Ek δkk ′ = |ak |2 Ek .
k k′ k k′ k
(7.6)

Then, .Eφ defined by Eq. (7.3) can be calculated as follows:

2 Eg k |ak |2 + k (Ek − Eg )|ak |2

  
k |ak | Ek
.Eφ = = . (7.7)
2 2
 
k |ak | k |ak |

In the second equality of the above expression, the fact that .Ek = Eg + (Ek − Eg )
was used. Now splitting the numerator into two terms, we find that

− Eg )|ak |2

k (Ek
Eφ = Eg +
.  2
≥ Eg , (7.8)
k |ak |

where the fact that .Ek − Eg ≥ 0, which should be satisfied because no eigenvalue
can be lower than the ground state eigenvalue, has been used. The above equation
completes the proof of the variational principle. ⊔
⊓
The power of the variational principle is the boundedness of .Eφ . This guarantees
that any effort to lower .Eφ always results in a better approximation for the true
ground state energy. It is often the case that the variational principle sometimes
serves as the last reliable resort in quantum mechanics when finding good enough
approximations for solutions are extremely difficult. The fact that the choice of the
state .|φ, which is called a trial state, can be chosen without any restriction is also
a great advantage. However, in practice, it is impossible to search for all possible
states. Rather, the search is limited to a class of states that can be represented by a
set of known basis states.
206 7 Approximation Methods for Time Independent Schrödinger Equation

7.1.2 Variational Principle for Trial States as Linear

Combinations of Basis States

One systematic and mathematically well understood way to construct a trial state is
to express it as a set of linearly independent states: .|f1 , .. . . , .|fn . While these are
independent states, we do not necessarily have to assume that they are orthogonal,
which significantly expands the choice of basis states one can make. Thus, let us
assume that the trial state is given by

|φ = C1 |f1  + C2 |f2  + · · · + Cn |fn ,

. (7.9)

where .C1 , .. . . , .Cn are unknown constants. The components of a given Hamiltonian
with respect to these states are denoted as

Hj k = fj |Ĥ |fk ,

. (7.10)

and the overlap matrix elements of the basis states are given by

Sj k = fj |fk .
. (7.11)

Then, the statement of the variational principle for the present case reduces to the
following inequality:
  ∗
φ|Ĥ |φ j k Cj Ck Hj k
.Eφ = =   ∗ ≥ Eg . (7.12)
φ|φ j k Cj Ck Sj k

The best approximation of .Eφ for .Eg can be found by choosing .Ck ’s minimizing .Eφ
given above. A necessary condition for this is that the partial derivative of .Eφ with
respect to .Ck ’s and .Ck∗ ’s, which can be viewed as independent complex variables,
should be zero. Thus, first taking the partial derivative of .Eφ with respect to one
coefficient, for example, for .Ci , where i can be any of the indices,
 
∗C H ∗S ′
 
Cj∗ Hj i C C

∂Eφ j k j k j k j ′ j ′ j i
j
. =   ∗ −
( j k Cj∗ Ck Sj k )2
 
∂Ci j C C S
k j k jk
n  ∗ ∗

j =1 Hj i Cj − Eφ Cj Sj i
=   ∗ = 0 , i = 1, . . . , n, (7.13)
j k Cj Ck Sj k

where Eq. (7.12) has been used in the second equality.

7.1 Variational Principle 207

Note that the denominator in Eq. (7.13) is nonzero and that it is satisfied as long
as the numerator becomes zero. Since .Hj i = Hij∗ and .Sj i = Sij∗ , it is thus equivalent
to the following identity:
n 
 
. Hij∗ Cj∗ − Eφ Sij∗ Cj∗ = 0, i = 1, . . . , n. (7.14)
j =1

Similarly, taking the partial derivative of .Eφ given by Eq. (7.12) with respect to .Ci∗ ,
   ∗ 
∂Eφ j Cj Hij j k Cj Ck Hj k k ′ Ck ′ Sik ′
.
∗ =   ∗ −   ∗
∂Ci j k Cj Ck Sj k ( j k Cj Ck Sj k )2
n 
j =1 Hij Cj − Eφ Sij Cj
=   ∗ = 0 , i = 1, . . . , n. (7.15)
j k Cj Ck Sj k

This equation is satisfied if the numerator is zero, and is thus equivalent to the
following equation:
n
 
. Hij Cj − Eφ Sij Cj = 0, i = 1, . . . , n. (7.16)
j =1

Note that Eqs. (7.14) and (7.16) are complex conjugates of each other. For conve-
nience, the latter is used in general.

Variational Principle for Linear Combination of Basis States

Equation (7.16), after dropping the subscript in .Eφ , is expressed as

n
 
Hij . − ESij Cj = 0, , i = 1, · · · , n, (7.17)
j =1

which is a general linear equation to be solved for E and .Cj ’s. In a matrix
notation, the above equation can also be expressed as

⎞⎛C ⎞
H11 − ES11 H12 − ES12 H1n − ES1n
⎛
... 1
⎜ H21 − ES21 H22 − ES22 ⎜ C2 ⎟
. . . H2n − ES2n ⎟ ⎜
⎟
. ⎟ = 0.
⎟
.⎜
⎝ ... ... ... ... ⎠⎜⎝ .. ⎠
Hn1 − ESn1 ... Hnn−1 − ESnn−1 Hnn − ESnn Cn
(7.18)

(continued)
208 7 Approximation Methods for Time Independent Schrödinger Equation

For the solution of the above equation to exist for nonzero .Cj ’s, it is necessary
for the following determinant to be zero.
 
 H11 − ES11 H12 − ES12 ... H1n − ES1n 

 H21 − ES21 H22 − ES22 ... H2n − ES2n 
.
 ... ... ... ...  = 0.
 
 H − ES ... Hnn−1 − ESnn−1 Hnn − ESnn 
n1 n1

(7.19)

The above equation is called secular equation. Solving this secular equation,
one can find n solutions for E. Inserting each value of E into Eq. (7.18)
and also using the normalization condition, one can find the set of .Cj ’s
corresponding to each solution of E. This completes the solution of Eq. (7.18).
The lowest eigenvalue and the corresponding eigenstate are deemed as the
best approximations for the ground state energy and the ground eigenstate.

It is possible [2] to show that all the eigenstates of Eq. (7.18) are orthogonal as
is explained below. Let us denote the .αth eigenvalue as .Eα and assume that there
(α)
exists a unique set of coefficients .Cj ’s such that

n
(α)

. (Hij − Eα Sij )Cj = 0, i = 1, · · · , n. (7.20)
j =1

Then, we can define the .αth eigenstate as follows:

n
(α)

|φα  =
. Cj |fj . (7.21)
j =1

Theorem 7.2 Two states .|φα  and .|φβ  defined by Eq. (7.21) with coefficients
satisfying Eq. (7.20) for two different energies .Eα = Eβ are orthogonal to each
other.
Proof First, let us consider the following matrix elements:
n n
(α) (α)
 
. fi |Ĥ |φα  = Cj fi |Ĥ |fj  = Hij Cj , . (7.22)
j =1 j =1
n n
(α) (α)
 
fi |φα  = Cj fi |fj  = Cj Sij . (7.23)
j =1 j =1
7.1 Variational Principle 209

Multiplying Eq. (7.23) with .Eα , we find that

n
(α)

Eα fi |φα  =
. Eα Cj Sij . (7.24)
j =1

Combining Eqs. (7.22) and (7.24), we find that Eq. (7.20) can be expressed as

fi |(Ĥ − Eα )|φα  = 0, i = 1, · · · , n

. (7.25)

(β)
Now consider another eigenstate .|φβ  = Cj |fj , where .Eβ = Eα . Then, using
the above identity, it is straightforward to show that
n
(β)∗

φβ |(Ĥ − Eα )|φα  =
. Cj fj |(Ĥ − Eα )|φα  = 0. (7.26)
j =1

In a similar manner, or exchanging .α and .β, one can also show that

. φα |(Ĥ − Eβ )|φβ  = 0. (7.27)

Subtracting the complex conjugate of Eq. (7.27) from Eq. (7.26) and noting the fact
that .Eα and .Eβ are real valued numbers, we then find that

(Eα − Eβ )φα |φβ  = 0.

. (7.28)

For .Eα = Eβ , the above equation proves that .φα |φβ  = φβ |φα  = 0. ⊔
⊓
As is clear from the above proof, in case .Eα = Eβ , the two states .|φα  and
.|φβ  do not have to be orthogonal to each other. However, even in such case, it
is still possible to find out two orthogonal states out of the linear combination of
those states. In this way, it is possible to find a set of orthogonal states satisfying the
variational principle.
As a simple example, consider a trial state as a linear combination of two basis
states as follows: .|φ = C1 |f1  + C2 |f2 . In addition, let us assumed that .|f1  and
.|f2  are normalized. All the matrix elements are assumed to be real numbers, and

.H11 = H22 = Hd and .H12 = H21 = Hc . Then,

  
Hd − E Hc − ES C1
. = 0, (7.29)
Hc − ES Hd − E C2

where .S = f1 |f2  = f2 |f1 . For the above equation to be satisfied for nonzero
values of .C1 and .C2 ,
 
 Hd − E Hc − ES 
.
 Hc − ES Hd − E  = 0, (7.30)
 
210 7 Approximation Methods for Time Independent Schrödinger Equation

which is equivalent to the following quadratic equation:

(Hd −E)2 −(Hc −ES)2 = (Hd −E+Hc −ES)(Hd −E−Hc +ES) = 0.

. (7.31)

The two solutions of the above equation are as follows:

Hd + Hc
. E+ = ,. (7.32)
1+S
Hd − Hc
E− = . (7.33)
1−S

The eigenstates corresponding to the above two eigenvalues can be found by using
each eigenvalue in Eq. (7.18) and solving the linear equation with respect to .C1 and
.C2 and employing the normalization condition. The resulting expressions for the

eigenstates are as follows:

1
|φ+  = √
. (|f1  + |f2 ), . (7.34)
2(1 + S)
1
|φ−  = √ (|f1  − |f2 ). (7.35)
2(1 − S)

7.2 Time Independent Perturbation Theory

Perturbation theory is a general approximation method for solving a differential or

integral equation, and plays an important role in practical applications of quantum
mechanics because there are only few limited cases for which exact solutions of
the Schrödinger equation are available. The term perturbation refers to a small
deviation or disturbance from the original equation, and the theory allows finding
approximate solutions in a systematic manner on the basis of solutions for the
original unperturbed equation. Perturbation theory can be developed for both time
independent and time dependent Schrödinger equations. The former, known as time
independent perturbation theory, is presented here whereas the latter, known as time
dependent perturbation theory, will be introduced in Chap. 10.
The idea of time independent perturbation theory can be summarized in a general
way for any eigenvalue problem. Let us consider a zeroth order operator .Ô (0) and
its eigenfunction or vector .f (0) with eigenvalue .λ(0) as follows.

Ô (0) f (0) = λ(0) f (0) .

. (7.36)

Then, given that the operator is perturbed by an additional perturbation operator

Ô (1) , the main question is what will be the resulting corrections for .f (0) and .λ(0) .
.

To find the answer for this question, it is useful to introduce a small parameter .ǫ
7.2 Time Independent Perturbation Theory 211

that controls the magnitude of the perturbation and try to find out the solution for
Ô(ǫ) = Ô (0) +ǫ Ô (1) as power series with respect to .ǫ. In other words, let us assume
.

that

. f (ǫ) = f (0) + ǫf (1) + ǫ 2 f (2) + · · · , . (7.37)

(0) (1) 2 (2)
λ(ǫ) = λ + ǫλ +ǫ λ + ··· . (7.38)

Then, employing these expressions in the new eigenvalue equation, .Ô(ǫ)f (ǫ) =
λ(ǫ)f (ǫ) and expanding the resulting expression with respect to .ǫ on both sides, we
obtain the following equation:
  
. Ô (0) + ǫ Ô (1)f (0) + ǫf (1) + ǫ 2 f (2) + · · ·
   
= Ô (0) f (0) + ǫ Ô (0) f (1) + Ô (1) f (0) + ǫ 2 Ô (0) f (2) + Ô (1) f (1) + · · ·
 
= λ(0) f (0) + ǫ λ(0) f (1) + λ(1) f (0)
 
+ǫ 2 λ(0) f (2) + λ(1) f (1) + λ(2) f (0) + · · · , (7.39)

where both left and right sides of the equation were grouped into terms of the
same order of .ǫ. At the zeroth order of .ǫ, the above equation is already satisfied
by construction. In order for the above equation to be satisfied for any value of .ǫ
including the particular case where .ǫ = 1, the equation for each term of .ǫ n with
.n = 1, 2, 3, · · · has to be solved. This is a nontrivial task because corrections

for both eigenfunction (or eigenvector) and eigenvalue have to be calculated, and
solving the equation for the nth order in general requires solutions for all the lower
orders. However, outcomes of such a task, once completed, can serve as powerful
and general expressions applicable to a wide range of realistic situations.
In applying the perturbation theory for quantum mechanics, it is particularly
useful to employ the Dirac notation as described below. Let us first consider a zeroth
order Hamiltonian .Ĥ0 for which the eigenstates and the eigenvalues are known, and
(0) (0)
denote them as .|ψn ’s and .En ’s. Here, n is a label representing each eigenstate.
Thus, for each n,

Ĥ0 |ψn(0)  = En(0) |ψn(0) ,

. (7.40)

(0) (0)
where .ψn |ψm  = δnm . Then, let us introduce a perturbation Hamiltonian .Ĥ1 ,
which is assumed1 to be small compared to .Ĥ0 , and consider the following total

1 The smallness of .Ĥ1 implies that its effect on the change of .En(0) is small compared to the level
(0)
spacing of .En ’s. Note that this definition of smallness has a problem in case there is degeneracy,
for which a separate treatment is needed.
212 7 Approximation Methods for Time Independent Schrödinger Equation

Hamiltonian:

Ĥ = Ĥ0 + ǫ Ĥ1 ,
. (7.41)

where .ǫ does not necessarily have to be small and can simply be considered as a
parameter to keep track of the order of the perturbation approximation. Once the
solution is obtained for each order of .ǫ, one can then set .ǫ = 1 without losing
generality of the solution because it can be viewed that the small parameter has
already been absorbed into .Ĥ1 .
In short, the perturbation theory amounts to solving the following equation:

(Ĥ0 + ǫ Ĥ1 )|ψn  = En |ψn ,

. (7.42)

by expanding .|ψn  and .En as powers of .ǫ such that each term in the expansion
(0) (0)
can ultimately be expressed in terms of .|ψn  and .En . The development of the
(0)
perturbation theory differs depending on whether .En ’s are degenerate or non-
degenerate. The latter case is straightforward, and is described first.

7.2.1 Non-Degenerate Perturbation Theory

(0)
The starting point of the perturbation theory is the complete set of .|ψn ’s, which
are normalized eigenstates of .Ĥ0 , and their respective eigenvalues .En(0) ’s. All of
these eigenvalues are assumed to be different for the present case. First, let us
assume that the eigenstates and eigenvalues of the total Hamiltonian satisfying
Eq. (7.42) can be expanded as follows:

. |ψn  = |ψn(0)  + ǫ|ψn(1)  + ǫ 2 |ψn(2)  + . . . , . (7.43)

En = En(0) + ǫEn(1) + ǫ 2 En(2) + . . . . (7.44)

In Eq. (7.43), the second and the third terms can be expanded in terms of the
eigenstates of the zeroth order Hamiltonian as follows:

|ψn(1)  =
.
(1) (0)
Cm |ψm , . (7.45)
m=n

|ψn(2)  = (2) (0)
Cm |ψm , (7.46)
m=n

where the summation does not include the index n because it is already represented
by the zeroth order state. Including this zeroth order state in the perturbation terms
simply amounts to changing the normalization constant of the zeroth order state and
does not result in any actual difference.
7.2 Time Independent Perturbation Theory 213

Employing Eqs. (7.43) and (7.44), the Schrödinger equation for the perturbed
Hamiltonian, Eq. (7.42), can be expressed as
 
. (Ĥ0 + ǫ Ĥ1 ) |ψn(0)  + ǫ|ψn(1)  + ǫ 2 |ψn(2)  + . . .
 
= (En(0) + ǫEn(1) + ǫ 2 En(2) + . . . ) |ψn(0)  + ǫ|ψn(1)  + ǫ 2 |ψn(2)  + . . . .

(7.47)

The perturbation theory amounts to solving the above equation for each order of .ǫ
as described below. First, for the order of .ǫ 0 ,

Ĥ0 |ψn(0)  = En(0) |ψn(0) ,

. (7.48)

which is already satisfied by the starting assumption, Eq. (7.40). The next order
terms are considered below.

First Order Terms of the Perturbation Expansion

The terms corresponding to the first order of .ǫ in Eq. (7.47), when combined,
lead to the following first order equation:

Ĥ1 |ψn(0)  + Ĥ0 |ψn(1)  = En(1) |ψn(0)  + En(0) |ψn(1) ,

. (7.49)

where .En(1) and .|ψn(1)  are yet unknown and need to be determined. Note that
(1)
.|ψn  is assumed to be given by Eq. (7.45). Thus, determining this amounts
(1)
to determining all the coefficients, .Cm ’s.

(0)
Equation (7.49) can be solved as follows. First, taking inner product with .ψn |,
we obtain

ψn(0) |Ĥ1 |ψn(0)  + ψn(0) |Ĥ0 |ψn(1)  = En(1) ψn(0) |ψn(0)  + En(0) ψn(0) |ψn(1) ,
. (7.50)

(0) (1) (0) (1) (0) (0) (1)

where .ψn |ψn  = 0. In addition, .ψn |Ĥ0 |ψn  = En ψn |ψn  = 0 because
(1) (0)
.|ψn  is expanded with respect to states that are orthogonal to .|ψn  as can be seen
(0) (0)
from Eq. (7.45). On the righthand side of the above equation, .ψn |ψn  = 1. For
these reasons, we obtain a simple expression for the first order perturbation energy
as summarized below.
214 7 Approximation Methods for Time Independent Schrödinger Equation

First Order Perturbation Energy

Equation (7.50) simplifies to the following expression for the first order
perturbation correction of the energy.

En(1) = ψn(0) |Ĥ1 |ψn(0) .

. (7.51)

Note that the above expression is nothing but the expectation value of the
perturbation Hamiltonian .Ĥ1 with respect to the zeroth order eigenstate.

(1)
Next, in order to determine .|ψn , let us take inner product of Eq. (7.49) with
(0)
.ψm |,
for .m = n. This leads to
(0)
. ψm |Ĥ1 |ψn(0)  + ψm
(0)
|Ĥ0 |ψn(1)  = En(0) ψm
(0) (1)
|ψn . (7.52)

(0) (0) (0) (0) (1)

Note that .ψm |Ĥ0 = ψm |Em . From the definition of Eq. (7.45), .ψm |ψn  =
(1)
Cm . Therefore, the above equation is equivalent to
(0)
ψm
. |Ĥ1 |ψn(0)  + Em
(0) (1)
Cm = En(0) Cn(1) , (7.53)

(1)
This equation can be solved easily for .Cm and leads to

(0)
(1) ψm |Ĥ1 |ψn(0) 
Cm
. = (0) (0)
. (7.54)
En − Em

The above expression is well-defined for the present case because there is no
degeneracy in the zeroth order energies. This also shows why the present approach
cannot be used when there is degeneracy.

First Order Perturbation State

(1)
Employing the expression for .Cm in the definition of Eq. (7.45), we obtain
the following first order perturbation correction for the state .|ψn(0) :

(0)
 |ψm (0) (0)
 ψm |Ĥ1 |ψn 
|ψn(1)  =
.
(1) (0)
Cm |ψm  = (0) (0)
. (7.55)
m=n m=n En − Em

Next, let us consider the second order terms of Eq. (7.47) by collecting all the
terms that are second order with respect to .ǫ.
7.2 Time Independent Perturbation Theory 215

Second Order Terms of the Perturbation Expansion

The terms corresponding to the second order of .ǫ in Eq. (7.47), when
combined, result in

. Ĥ1 |ψn(1)  + Ĥ0 |ψn(2)  = En(2) |ψn(0)  + En(1) |ψn(1)  + En(0) |ψn(2) ,(7.56)

(2) (2)
where .|ψn  and .En are as yet unknown and need to be determined.

(0)
Taking inner product of Eq. (7.56) with .ψn |, we obtain

. ψn(0) |Ĥ1 |ψn(1)  + ψn(0) |Ĥ0 |ψn(2) 

= En(2) ψn(0) |ψn(0)  + En(1) ψn(0) |ψn(1)  + En(0) ψn(0) |ψn(2) . (7.57)

In the lefthand side of the above expression,

.ψn(0) |Ĥ0 |ψn(2)  = En(0) ψn(0) |ψn(2)  = 0, (7.58)

(2)
where the second equality comes from the definition of .|ψn , Eq. (7.46), which
(0)
is expanded only with respect to those states that are orthogonal to .|ψn . In
(0) (0) (0) (1)
addition, in the righthand side of Eq. (7.57), .ψn |ψn  = 1, .ψn |ψn  = 0,
(0) (2)
and .ψn |ψn  = 0. All of these, when combined, lead to a general expression for
the second order perturbation energy as summarized below.

Second Order Perturbation Energy

Equation (7.56) reduces to the following expression for the second order
perturbation correction for the energy:

En(2) = ψn(0) |Ĥ1 |ψn(1)  =
.
(1)
Cm ψn(0) |Ĥ1 |ψm
(0)

m=n
(0)
 ψm (0) (0) (0)
|Ĥ1 |ψn ψn |Ĥ1 |ψm 
= (0) (0)
. (7.59)
m=n En − Em

In each term of the summation in the above expression, the numerator is

nonnegative as follows:
 2
(0)
|Ĥ1 |ψn(0) ψn(0) |Ĥ1 |ψm
(0)  (0)
ψm  = ψm |Ĥ1 |ψn(0)  ≥ 0. (7.60)

.

(continued)
216 7 Approximation Methods for Time Independent Schrödinger Equation

An important implication of this is that the second order perturbation order

correction for the ground state energy always lowers its value because the
denominator of each term in the summation is always negative.

We can obtain the second order correction for the state by taking inner product
(0)
of Eq. (7.56) with .ψm | for .m = n. Thus, we obtain
(0)
ψm
.
(0)
|Ĥ1 |ψn(1)  + ψm (0) (1)
|Ĥ0 |ψn(2)  = En(1) ψm (0) (2)
|ψn  + En(0) ψm |ψn ,
(7.61)
(0) (0)
where the fact that .ψm |ψn  = 0 has been used in obtaining the righthand side.
(0) (2) (2) (0)
Now using Eq. (7.55) and the fact that .ψm |Ĥ0 |ψn  = Cm Em on the lefthand
side of the above equation, we find that

(1) (0)

(0) (2) (0)
. Cm′ ψm |Ĥ1 |ψm′  + Cm Em = En(1) Cm
(1) (2) (0)
+ Cm En . (7.62)
m′ =n

(2)
Solving this for .Cm , we obtain

(1) (0)

.(En(0) − Em
(0) (2)
)Cm = (0)
Cm′ ψm |Ĥ1 |ψm′  − En(1) Cm
(1)
. (7.63)
m′ =n

(1) (1)
Employing the expressions for .En and .Cm , the above expression can be expressed
as
(0)
 ψm (0) (0) (0)
(2) |Ĥ1 |ψm ′ ψm′ |Ĥ1 |ψn 
.Cm = (0) (0) (0) (0)
m′ =n (En − Em )(En − Em′ )
(0) (0) (0) (0)
ψn |Ĥ1 |ψn ψm |Ĥ1 |ψn 
− (0) (0)
. (7.64)
(En − Em )2

Second Order Perturbation State

(2)
Employing the expression for .Cm , Eq. (7.64), in the definition of .|ψn(2)  given
by Eq. (7.46), we obtain the following second order correction for the state
(0)
.|ψn :

(continued)
7.2 Time Independent Perturbation Theory 217

⎧
⎨  ψ (0) |Ĥ |ψ (0) ψ (0) |Ĥ |ψ (0) 
  m 1 m′ m′ 1 n
|ψn(2)  =
.
(2) (0)
Cm |ψm  = (0)
|ψm  (0) (0) (0) (0)
m=n m=n
⎩ ′ (En − Em )(En − E ′ )
m =n m

(0) (0) (0) (0)
ψm |Ĥ1 |ψn ψn |Ĥ1 |ψn 
− (0) (0)
. (7.65)
(En − Em )2

Summarizing all the results obtained above, second order perturbation correc-
tions of energies and states, for the case of .ǫ = 1, are

(0)
 ψm (0) (0) (0)
|Ĥ1 |ψn ψn |Ĥ1 |ψm 
. En = En(0) + ψn(0) |Ĥ1 |ψn(0)  + , (7.66)
.

m=n En(0) − Em
(0)

(0)
 |ψm (0) (0)
ψm |Ĥ1 |ψn 
|ψn  = |ψn(0)  +
m=n En(0) − Em
(0)

⎧
⎨  ψ (0) |Ĥ |ψ (0) ψ (0) |Ĥ |ψ (0) 

(0) m 1 m′ m′ 1 n
+ |ψm  (0) (0) (0) (0)
m=n
⎩ ′
m =n(En − Em )(En − E ′ ) m

(0) (0) (0) (0)
ψm |Ĥ1 |ψn ψn |Ĥ1 |ψn 
− (0) (0)
. (7.67)
(En − Em )2

7.2.2 Degenerate Perturbation Theory

When there is degeneracy, the expressions, Eqs. (7.66) and (7.67), cannot be used
(0) (0)
because the denominator becomes zero, i.e., .En − Em = 0, for degenerate states.
Thus, a different approach [4–8] should be used for these states.
(0)
Let us assume that there are d degenerate eigenstates of .Ĥ0 with energy .En as
shown below:
(0) (0)
|ψn,1 , . . . , |ψn,d .
. (7.68)

Then, any linear combination of these states can be an eigenstate with the same
(0)
energy .En . Among all possible linear combinations, there can be unique d linear
combinations that are least affected by the perturbation .Ĥ1 and thus serve as the best
(0)
starting point for applying the perturbation theory. Let us denote them as .|φn,k ’s
with .k = 1, · · · , d.
218 7 Approximation Methods for Time Independent Schrödinger Equation

(0) (0)
Each .|φn,k  can be expressed as a linear combination of .|ψn,k ’s as follows:

(0) (0) (0)

|φn,k
.  = C1 |ψn,1  + · · · + Cd |ψn,d  , k = 1, · · · , d. (7.69)

Then, let us assume that the eigenstate of .Ĥ that reduces to the one with eigenvalue
(0)
.En in the absence of perturbation (the limit of .ǫ → 0), can be expanded as

(0) (1) (2)

|φn,k  = |φn,k  + ǫ|φn,k  + ǫ 2 |φn,k  + . . . ,
. (7.70)

(1) (2)
where detailed expressions for .|φn,k  and .|φn,k  will be provided later in this section.
Employing Eq. (7.70) in Eq. (7.42), we obtain
 
(0) (1) (2)
. (Ĥ0 + ǫ Ĥ1 ) |φn,k  + ǫ|φn,k  + ǫ 2 |φn,k  + ...
(0) (1) (2)
= (En(0) + ǫEn(1) + ǫ 2 En(2) + . . . )(|φn,k  + ǫ|φn,k  + ǫ 2 |φn,k  + . . . ). (7.71)

(0) (0) (0)

The zeroth order term (.ǫ 0 ) of Eq. (7.71) is .Ĥ0 |φn,k  = En |φn,k , which is already
(0)
satisfied since .|φn,k  is a linear combination of all states with the same energy
eigenvalue .En . The first order terms are considered below.

First Order Term of the Degenerate Perturbation Theory

Collecting all the terms that are first order with respect to .ǫ in Eq. (7.71), we
obtain
(0) (1) (0) (1)
Ĥ1 |φn,k  + Ĥ0 |φn,k  = En(1) |φn,k  + En(0) |φn,k .
. (7.72)

(0)
Taking inner product of Eq. (7.72) with .ψn,j | for .j = 1, . . . , d, we obtain

(0) (0) (0) (1) (1) (0) (0) (0) (1)

ψn,j |Ĥ1 |φn,k  + ψn,j |Ĥ0 |φn,k  = En,k ψn,j |φn,k  + En(0) ψn,j |φn,k .
. (7.73)

(0) (1)
On the lefthand side of the above equation, .ψn,j |Ĥ0 |φn,k  = En(0) ψn,j
(0) (1)
|φn,k ,
which cancels with the same term on the righthand side. Inserting Eq. (7.69) into the
resulting equation leads to a general set of equations that can be solved to determine
all the coefficients of Eq. (7.69) as described below.
7.2 Time Independent Perturbation Theory 219

First Order Correction for the Eigenvalue and the Zeroth Order Correc-
tion for the Eigenstate
Employing Eq. (7.69) in Eq. (7.73) and canceling all terms that are identical
on both sides of the resulting equation, we obtain

H1n,j 1 C1 + H1n,j 2 C2 + · · · + H1n,j d Cd = En(1) Cj , j = 1, . . . , d,

. (7.74)

where
(0) (0)
.H1n,j k = ψn,j |Ĥ1 |ψn,k . (7.75)

Equation (7.74) for all j can be expressed as the following matrix equation:

(1)
⎛ ⎞⎛ ⎞
H1n,11 − En H1n,12 ... H1n,1d C1
(1) ⎟ ⎜ C2 ⎟
H1n,21 H1n,22 − En ... H1n,2d
⎜
⎟ ⎜ . ⎟ = 0.
⎜ ⎟⎜ ⎟
.⎜
⎝ ... ⎠ ⎝ .. ⎠
(1)
H1n,d1 ... H1n,dd−1 H1n,dd − En Cd
(7.76)

This matrix equation implies that the following determinant is zero:

 (1)

 1n,11 − En
H H1n,12 ... H1n,1d 

(1)
H1n,21 H1n,22 − En ... H1n,2d
 
 = 0, (7.77)
 
.
 ... 
 (1) 
 H1n,d1 ... H1n,dd−1 H1n,dd − En

the solution of which results in the first order degenerate perturbation

(1)
correction for the energy, .En,k for .k = 1, · · · , d.
(1)
For each solution of .En,k , one can find the corresponding set of coefficients
(0)
Ck ’s in Eq. (7.69), which results in the zeroth order eigenstate .|φn,k  for each
.
(1)
En,k as follows:
.

d
(0) (0) (0)

|φn,k  =
. Cn,kj |ψj , (7.78)
j =1
(0)
where .Cn,kj ’s are solutions of the matrix equation, Eq. (7.76), for the eigen-
(1)
value .En,k and are assumed to satisfy the normalization condition.
220 7 Approximation Methods for Time Independent Schrödinger Equation

(1)
If all the .En,k ’s are different, one can then apply the non-degenerate perturbation
(0)
theory employing .|φn,k ’s as the starting point. However, if there is degeneracy in
(1)
En,k ’s, one should again consider linear combination of those degenerate states in
.

going up to the next order in a manner similar to finding the degenerate first order
corrections for the eigenvalue and the zeroth order corrections for the states, which
results in additional complications.
(1)
Here, let us consider only the case where .En,k ’s are non-degenerate. So far, we
have considered the perturbation effects of degenerate states only. However, other
non-degenerate states can still make some contributions as well, which have to be
added to the results of the degenerate perturbation theory in an appropriate manner.
As the first step for calculating higher order terms, we need to find out the
(1)
first order correction for the eigenstate .|φn,k  in Eq. (7.72). Taking inner product
(0)
of Eq. (7.72) with .φn,j | for .j = k, we obtain

(0) (0) (0) (1) (0) (0) (0) (1)

φn,j |Ĥ1 |φn,k  + φn,j |Ĥ0 |φn,k  = En(1) φn,j |φn,k  + En(0) φn,j |φn,k .
. (7.79)

(0) (0)
In the above equation, .φn,j |Ĥ1 |φn,k  = 0 since .Ĥ1 is diagonal in the basis of new
(0) (0) (0)
zeroth order eigenstates, .|φn,j ’s. Similarly, .φn,j |φn,k  = 0. On the other hand,
(0) (1) (0) (0) (1)
φn,j |Ĥ0 |φn,k  = En φn,j |φn,k , which becomes the same as the second term on
.

the righthand side of the above equation. Thus, Eq. (7.79) is already satisfied.
(1) (0)
Therefore, the expansion coefficients of .|φn,k  with respect to .|φn,j  remain
undetermined at the first order level. On the other hand, the expansion coefficients
(0)
with respect to other non-degenerates states, .|ψm , where .m = n1 , · · · , nd , can
be determined in the same manner as the non-degenerate perturbation theory as
summarized below.

First Order Correction for the Degenerate Eigenstate (Incomplete)

(0)
Taking inner product of Eq. (7.72) with .ψm |, where .m = n, one can
(1) (0)
determine the expansion coefficients of .|φn,k  with respect to .|ψm ’s and
find the following expression:

d d (0) (0) (0)

(1) |ψm ψm |Ĥ1 |φn,k 
Dj(1) |φn,j
(0)
 
|φn,k
. = + (0) (0)
, (7.80)
j =k m=n k=1 En − Em

where .Dj(1) ’s are yet undetermined. Consideration of the second order pertur-
bation terms is necessary to determine these as will be shown below.
7.2 Time Independent Perturbation Theory 221

The second order corrections for the eigenvalue and eigenstate can be obtained
by considering all the second order terms in Eq. (7.71) as explained below.

Second Order Term of the Degenerate Perturbation Theory

Collecting all the terms that are second order with respect to .ǫ in Eq. (7.71),
we obtain
(2) (1) (2) (1) (1) (2) (0)
. Ĥ0 |φn,k  + Ĥ1 |φn,k  = En(0) |φn,k  + En,k |φn,k  + En,k |φn,k , (7.81)

(1) (0) (2)

where .|φn,k  is orthogonal to .|φn,k  according to Eq. (7.80) and .|φn,k  is also
(0)
assumed to be orthogonal to .|φn,k  by construction. In the above equation,
(2) (2) (1)
En,k and .|φn,k  as well as part of .|φn,k  are yet undetermined, for which the
.

equation needs to be solved.

(0)
Let us take inner product of Eq. (7.81) with .φn,k |. Then,

(0) (2) (0) (1) (2)

. φn,k |Ĥ0 |φn,k  + φn,k |Ĥ1 |φn,k  = En,k , (7.82)

(1) (2) (0)

where the fact that .|φn,k  and .|φn,k  are orthogonal to .|φn,k  has been used on
the righthand side. On the lefthand side of the above equation, the first term also
(0) (2) (0) (0) (2) (2)
becomes zero as follows: .φn,k |Ĥ0 |φn,k  = En φn,k |φn,k  = 0. Thus, .En,k simply
becomes the second term on the lefthand side, for which Eq. (7.80) can be used as
summarized below.

Second Order Correction for the Degenerate Eigenvalue

Employing Eq. (7.80) for the second term on the lefthand side of Eq. (7.82),
we obtain
d (0) (0) (0) (0)
(2)
 φn,k |Ĥ1 |ψm ψm |Ĥ1 |φn,k 
.E
n,k = (0) (0)
. (7.83)
m=n k=1 En − Em

Note that the sum involving degenerate states with the unknown coef-
(1)
ficients .Dj in Eq. (7.80) do not contribute to the above term because
(0) (0)
φn,k |Ĥ1 |φn,j  = 0 for .j = k.
.
222 7 Approximation Methods for Time Independent Schrödinger Equation

(0)
As the next step, let us take inner product of Eq. (7.81) with .φn,j | for .j = k.
Thus,
(0) (2) (0) (1) (0) (2) (1) (0) (1)
φn,j |Ĥ0 |φn,k  + φn,j |Ĥ1 |φn,k  = En(0) φn,j |φn,k  + En,k φn,j |φn,k ,
. (7.84)

(0) (0)
where the fact that .φn,j |φn,k  = 0 has been used. On the lefthand side of the above
(0) (2) (0) (0) (2)
equation, the first term, .φn,j |Ĥ0 |φn,k  = En φn,j |φn,k , cancels out with the same
term on the righthand side. Therefore, the above equation reduces to

(0) (1) (1) (0) (1)

φn,j |Ĥ1 |φn,k  = En,k φn,j |φn,k .
. (7.85)

(1)
This equation can be used to determine the .Dj in Eq. (7.80). Employing Eq. (7.80)
in the above equation, we obtain

d φ (0) |Ĥ |ψ (0) ψ (0) |Ĥ |φ (0) 

n,j 1 m m 1 n,k
Dj(1) En,j
(1)
= Dj(1) En,k
(1)

. + . (7.86)
m=n k=1 En(0) (0)
− Em

(1) (1) (1)

Assuming that .En,j = En,k , the above equation can be solved for .Dj as follows:

d φ (0) |Ĥ |ψ (0) ψ (0) |Ĥ |φ (0) 

(1)
 n,j 1 m m 1 n,k
Dj =
.
(1) (1)
. (7.87)
m=n k=1 (En,k − En,j )(En(0) − Em
(0)
)

This completes the first order perturbation correction for degenerate states as
summarized below.

First Order Correction for the Degenerate Eigenstate (Complete)

Using Eq. (7.87) in Eq. (7.80), we obtain the following expression for the
(1)
complete first order expansion of .|φn,k :

d  |φ (0) φ (0) |Ĥ |ψ (0) ψ (0) |Ĥ |φ (0) 

(1)
 n,j n,j 1 m m 1 n,k
.|φ
n,k  = (1) (1) (0) (0)
j =k m=n (En,k − En,j )(En − Em )
(0) (0) (0)
 |ψm ψm |Ĥ1 |φn,k 
+ (0) (0)
. (7.88)
m=n En − Em

(0)
Finally, let us take inner product of Eq. (7.86) with .ψm | for .m = n. Then,
(0) (2) (0) (0) (2) (0) (0)
employing the fact that .ψm |Ĥ0 |φn,k  = Em ψm |φn,k  and .ψm |φn,k  = 0, we
7.3 Summary and Questions 223

obtain
(0) (0) (2) (0) (1) (2) (1) (1)
Em
. ψm |φn,k  + ψm |Ĥ1 |φn,k  = En(0) ψm
(0) (0)
|φn,k  + En,k ψm |φn,k . (7.89)

(0) (2)
Solving this equation for .ψm |φn,k , we obtain

(0) (1) (1)

(0) (2)
ψm |(Ĥ1 − En,k )|φn,k 
.ψm |φ
n,k  = (0) (0)
En − Em
⎡
d  ψ (0) |Ĥ |φ (0) φ (0) |Ĥ |ψ (0) ψ (0) |Ĥ |φ (0) 
1  m 1 n,j n,j 1 m′ m′ 1 n,k
= (0) (0)
⎣
(1) (1) (0) (0)
En − Em j =k m′ =n (En,k − En,j )(En − Em′ )
(0) (0) (0) (0)
 ψm |Ĥ1 |ψm ′ ψm′ |Ĥ1 |φn,k 
+ (0) (0)
m′ =n En − Em′
(1) (0) (0) 
En,k ψm |Ĥ1 |φn,k 
− (0) (0)
. (7.90)
En − Em

(2)
The above expressions determines the expansion coefficients of .|φn,k  with respect
(0) (2)
to the .|ψm  for .m = n. However, the expansion coefficients of .|φn,k  with respect
(0)
to the .|φn,j  with .j = k remain yet unknown, which can be determined only by
solving the third order terms of the perturbation theory.

7.3 Summary and Questions

The variational principle results from the boundedness of the ground state energy for
a stable quantum system, and thus offers an important guideline for approximating
the ground state energy. Proving the variational principle is simple, and is based
on the completeness of the eigenstates of a Hamiltonian. However, finding out
an actual approximation is not always straightforward. Given that it is possible
to identify a reasonable basis states that can be used to represent a trial state as
a linear combination, it is possible to reduce the variational principle to a set of
linear equations involving coefficients and eigenvalues, Eq. (7.18). Eigenvalues of
this equation can be determined by solving Eq. (7.19). Employing these solutions
in Eq. (7.18), one can then determine corresponding eigenstates. The resulting
eigenstates, even when expanded in terms of nonorthogonal basis states, are
orthogonal for different eigenvalues and can also be made to be orthogonal between
states with the same eigenvalue.
The time independent perturbation theory provides general expressions for
approximate eigenvalues and eigenstates of a Hamiltonian, which is given by a
224 7 Approximation Methods for Time Independent Schrödinger Equation

sum of the zeroth order and small perturbation terms. The solutions are provided
in terms of those for the zeroth order Hamiltonian. The reason for why this is
always possible is because the eigenstates of the zeroth order Hamiltonian forms
a complete basis. For the case where there is no degeneracy in the zeroth order
Hamiltonian, perturbation corrections for the energies and states can be obtained by
comparison of terms of the same order of the perturbation and projecting them to
different eigenstates of the zeroth order Hamiltonian. The resulting corrections for
the eigenvalues and eigenstates up to the second order of the perturbation are given
by Eqs. (7.66) and (7.67).
However, for the case where there is degeneracy in the zeroth order Hamiltonian,
the expressions obtained for the non-degenerate case become singular, and the
theory should be developed in a slightly different manner. It turns out that
identifying the best zeroth order eigenstates that suit the type of the Hamiltonian
should precede determination of higher order terms. Diagonalizing the perturbation
Hamiltonian in the space of degenerate states of the zeroth order Hamiltonian
solves this issue. Thus, the first order corrections for the eigenvalues are given by
solutions of Eq. (7.76) and the zeroth eigenstates are given by Eq. (7.78). Further
consideration of higher order corrections including those of other non-degenerate
components leads to the expression for the second order correction for the energy,
Eq. (7.83), and the first correction for the state, Eq. (7.88).

Questions
• How general is the variational principle? Is there any possible exception
where it is inapplicable?
• In applying the variational principle, it is common to approximate the
trial state as a linear combination of predetermined basis states. What are
benefits and limitations of this approximation?
• Is it possible to extend the variational principle for the calculation of
excited state energies?
• For the case where the trial state is given by a linear combination of basis
states, solving an eigenvalue equation produces many states and energies
as solutions of the variational principle. Can all the higher energy states,
except for the lowest energy state, be also used as solutions of the starting
time independent Schrödinger equation?
• How can we examine the accuracy of a particular outcome of the varia-
tional theory?
• Expressions for the non-degenerate perturbation theory are inapplicable for
the case where there is degeneracy in the zeroth order Hamiltonian. What
is the reason for this?
• Are the states obtained from the perturbation theory normalized?
• How can we assess the accuracy of a particular application of the pertur-
bation theory?

(continued)
Exercise Problems with Solutions 225

• Let us assume that a hydrogen atom is subject to an external electric field.

How can we apply the perturbation theory to calculate the effect of the
electric field on the energy of the electron in this case?

Exercise Problems with Solutions

7.1 A certain Hamiltonian can be expressed as

Ĥ = −|ϕ1 ϕ1 | + |ϕ2 ϕ2 |,

.

where |ϕ1  and |ϕ2  are normalized but are not orthogonal as follows:

1
ϕ1 |ϕ2  = ϕ2 |ϕ1  =
. .
2
Use the variational theorem to determine the approximation for the ground state
energy and express the corresponding approximate normalized ground state as a
linear combination of |ϕ1  and |ϕ2 .
Solution 7.1 For the given Hamiltonian,

1 3
. ϕ1 |Ĥ |ϕ1  = −1 + =− ,
4 4
1 3
ϕ2 |Ĥ |ϕ2  = − +1= ,
4 4
1 1
ϕ1 |Ĥ |ϕ2  = − + = 0 = ϕ2 |Ĥ |ϕ1 .
2 2
The corresponding
 3 secular1 equation that results from the variational principle for
√

− − E − E 
this case is  4 1 2  = (E 2 − 3 ) = 0. This results in E = ± 3 .
3
3 4 4 2
−2E 4 − E 
√
Therefore, the approximation for the ground state energy is Eg = − 23 .
For Eg, we can√find the
√
approximate
   solution by solving the following matrix
3 3 3
− 4 √+ 2 4√ C1
equation: 3 3 3
= 0.
C2
4 4 + 2 √
Solving the above equation, we find that C2 = ( 3 − 2)C1 . Thus,
 √ 
|ψg  = C1 |ϕ1  + ( 3 − 2)|ϕ2  .
.
226 7 Approximation Methods for Time Independent Schrödinger Equation

Assuming the normalization condition,

√ √
 
1
ψg |ψg  = C12 1 + 2( 3 − 2) + 7 − 4 3 = 1,
.
2

we find that
1
C1 = 
. √ .
6−3 3

Therefore,
√
1 3−2
.|ψg  =  √ |ϕ1  +  √ |ϕ2 .
6−3 3 6−3 3

7.2 Consider the Hamiltonian of a hydrogen-like system for the case where l = 0.
In this case, the angular part becomes constant, and one can consider the radial
part only. In other words, the Hamiltonian effectively becomes one dimensional as
follows:

h̄2 d Ze2
 
2 d
.Ĥ = − r − .
2μr 2 dr dr 4π ǫ0 r
2
Then, assuming that a trial wave function, φ(r) = e−αr , where α is a free
parameter, use the variational principle to calculate approximate ground state
energy.
Solution 7.2 For the given trial wavefunction,
  ∞ 2
φ|φ =
.
2
dr|φ(r)| = 4π drr 2 e−2αr
0
√
1 π π 3/2
= 4π = √ .
2 2(2α)3/2 2 2α 3/2

On the other hand,


.φ|Ĥ |φ = drφ ∗ (r)Ĥ φ(r)
 
 ∞ h̄2 d
 
Ze2
2 −αr 2 2 d 2
= 4π drr e − 2
r − e−αr
0 2μr dr dr 4π ǫ0 r
 ∞  
2 −αr 2 h̄2 d  3 −αr 2
 Ze2 −αr 2
= 4π drr e − −2αr e − e
0 2μr 2 dr 4π ǫ0 r
Exercise Problems with Solutions 227

 
 ∞ 3α h̄2 2h̄2 α 2 2 Ze2 2
= 4π drr 2
− r − e−2αr
0 μ μ 4π ǫ0 r

3α h̄2 ∞ 2h̄2 α 2 ∞
 
2 −2αr 2 2
= 4π drr e − drr 4 e−2αr
μ 0 μ 0

Ze2
 ∞ 2

− drre−2αr .
4π ǫ0 0

Employing the following identities for the integrals:

 ∞  ∞ √
2 1 2 π
. drr 2 e−2αr = r 2 e−2αr = ,
0 2 −∞ 4(2α)3/2
 ∞  ∞ √
2 1 2 3 π
drr 4 e−2αr = r 4 e−2αr = ,
0 2 −∞ 8(2α)5/2
 ∞ 2 1
drre−2αr = ,
0 4α

φ|Ĥ |φ can be simplified as

√ √ 
3α h̄2 π 2h̄2 α 2 3 π Ze2 1
.φ|Ĥ |φ = 4π − −
μ 4(2α)3/2 μ 8(2α)5/2 4π ǫ0 4α

4π 3/2 h̄2 Ze2

 
3 3
= √ √ − √ −
2μ α 4 2 8 2 4ǫ0 α
3π 3/2 h̄2 Ze2
= √ √ − .
4 2μ α 4ǫ0 α

Therefore,
  √
3π 3/2 h̄2 Ze2 2 2α 3/2
.Eφ = E(α) = √ √ −
4 2μ α 4ǫ0 α π 3/2
√
3h̄2 α Ze2 α
= −√ .
2μ 2ǫ0 π 3/2

Now, taking the derivative of E(α) with respect to α and using the following
necessary condition for it to be a minimum:

dE(α) 3h̄2 Ze2 1

. = − √ √ = 0,
dα 2μ 2 2ǫ0 π 3/2 α
228 7 Approximation Methods for Time Independent Schrödinger Equation

we find that
2
Ze2 Z 2 e4 μ2

2μ
α=
. √ = .
2 2ǫ0 π 3/2 3h̄2 18π 3 h̄4 ǫ02

For this value of α, the energy given by Eφ becomes

3h̄2 Z 2 e4 μ2 Ze2 Ze2 μ

E=
.
4 2
−√ √
2μ 18π 3 h̄ ǫ0 2ǫ0 π 3/2 2ǫ0 π 3/2 3h̄2
μZ 2 e4 μZ 2 e4 μZ 2 e4 32 μZ 2 e4
= − =− =−
12π 3 h̄2 ǫ02 6π 3 h̄2 ǫ02 12π 3 h̄2 ǫ02 12π 32π 2 ǫ02 h̄2
= −0.849Eg ,

where Eg is the true ground state.

7.3 Prove that the states |φ+  and |φ−  given by Eqs. (7.34) and (7.35) are varia-
tional eigenstates corresponding to the variational energies, Eqs. (7.32) and (7.33),
respectively.
Solution 7.3 For E+ = (Hd + Hc )/(1 + S),
     
Hd − E+ Hc − E+ S C1 Hd S − Hc 1 −1 C1
. = = 0.
Hc − E+ S Hd − E+ C2 1+S −1 1 C2

This means that C1 = C2 . Thus,

|φ+  = C1 (|f1  + |f2 ).

.

The norm of this is as follows:

.φ+ |φ+  = |C1 |2 (1 + 1 + 2S) = 2|C1 |2 (1 + S).

√
Thus, C1 = 1/ 2(1 + S).
For E− = (Hd − Hc )/(1 − S),

C1′ C1′
     
Hd − E− Hc − E− S Hc − Hd S 11
. = = 0.
Hc − E− S Hd − E− C2′ 1−S 11 C2′

This means that C1′ = −C2′ . Thus,

|φ+  = C1′ (|f1  − |f2 ).

.
Exercise Problems with Solutions 229

The norm of this is as follows:

φ− |φ−  = |C1′ |2 (1 + 1 − 2S) = 2|C1′ |2 (1 − S).

.

√
Thus, C1′ = 1/ 2(1 − S).
7.4 Consider a harmonic oscillator with linear perturbation as follows:

p̂2 1
. Ĥ0 = + μω2 x̂ 2 , . (7.91)
2μ 2
Ĥ1 = C x̂. (7.92)

Use the non-degenerate perturbation theory to determine perturbation corrections of

the eigenvalue up to the second order and the eigenstate up to the first order.
Solution 7.4 Then, |ψv(0)  = |v, where

1
. Ĥ0 |v = h̄ω(v + )|v. (7.93)
2
It is straightforward to show that

′ h̄ √ √ 
v|Ĥ1 |v  = C
. v + 1δv ′ ,v+1 + vδv ′ ,v−1 . (7.94)
2μω

Therefore,

Ev(1) = v|Ĥ1 |v = 0, .

. (7.95)
 v ′ |Ĥ1 |vv|Ĥ1 |v ′ 
Ev(2) =
′
h̄ω(v − v ′ )
v =v
 √ √ 
( v)2 ( v + 1)2 h̄ C 2 h̄ C2
= − C2 =− = − . (7.96)
h̄ω h̄ω 2μω 2h̄ω2 μ 2μω2

Combining the above expressions, we find that

1 C2
Ev = Ev(0) (1) (2)
. + Ev + Ev = h̄(v + ) − . (7.97)
2 2μω2
230 7 Approximation Methods for Time Independent Schrödinger Equation

In fact, one can show that this is an exact value employing the following identity:

p̂2 1
Ĥ =
. + μω2 x̂ 2 + cx̂. (7.98)
2μ 2
p̂2 C2 1 C2
 
1 2C
= + μω2 x̂ 2 + 2
x̂ + 2 4
−
2μ 2 μω μ ω 2 μω2
p̂2 C 2 1 C2
 
1 2
= + μω x̂ + −
2μ 2 μω2 2 μω2
p̂2 1 1 C2
= + μX̂2 − , (7.99)
2μ 2 2 μω2

C
where X̂ = x̂ + μω2
. On the other hand, the eigenstate, up to the first order is given
by

C h̄ √ √ 
.|ψv  = |v + v|v − 1 − v + 1|v + 1 . (7.100)
h̄ω 2μω

7.5 The zeroth order Hamiltonian of a certain system is given by Ĥ0 = |φ1 φ1 | +
2|φ2 φ2 |, where |φ1  and |φ2  are orthonormal sates. A perturbation Hamiltonian
Ĥ1 is added to Ĥ0 , and the expression for Ĥ1 is as follows:

1 1
Ĥ1 =
. (|φ1 φ1 | + |φ2 φ2 |) + (|φ1 φ2 | + |φ2 φ1 |) .
2 4
Calculate the first order and second order perturbation energy corrections for the
two eigenvalues of Ĥ0 .
(0)
Solution 7.5 Note that |φ1  and |φ2  are eigenstates of Ĥ0 with eigenvalues E1 =
1 and E2(0) = 2 respectively. Therefore,

(1) 1 (1) 1
. E1 = φ1 |Ĥ1 |φ1  = & E2 = φ2 |Ĥ1 |φ2  =
2 2
(2) |φ1 |Ĥ1 |φ2 |2 1 (2) |φ2 |Ĥ1 |φ1 |2 1
E1 = =− & E2 = =
E1(0) − E2(0) 16 E2(0) − E1(0) 16

7.6 Consider a Hamiltonian Ĥ defined as follows:

.Ĥ = a|11| + 2a|22| + 3a|33|,

where |1, |2, and |3 are orthonormal states and a is a positive real number. Answer
the following questions.
Exercise Problems with Solutions 231

(a) For any state |φ = C1 |1 + C2 |2 + C3 |3, find out the expression for Eφ .
(b) Determine the minimum and maximum values of Eφ and corresponding state
|φ for each case.
(c) Assume |φ as a variational trial state, write down the matrix equation Eq. (7.18)
for the Hamiltonian Ĥ , determine three variational energies and states.
(d) A small perturbation term Ĥ1 is added to the Hamiltonian so that the total
Hamiltonian changes to Ĥ ′ = Ĥ + Ĥ1 , where
a
Ĥ1 =
. (|12| + |21|).
5
Find out the first order approximation for the ground eigenstate and second
order approximation for the ground state energy.
Solution 7.6 Let us first calculate φ|Ĥ |φ and φ|φ as follows:

φ|Ĥ |φ = C1∗ 1| + C2∗ 2| + C3∗ 3| (a|11| + 2a|22| + 3a|33|)

.

× (C1 |1 + C2 |2 + C3 |3)

= C1 1| + C2∗ 2| + C3∗ 3| (aC1 |1 + 2aC2 |2 + 3aC3 |3)
 ∗

= a|C1 |2 + 2a|C2 |2 + 3a|C3 |2 ,

φ|φ = |C1 |2 + |C2 |2 + |C3 |2 .

(a) The expectation value Eφ can be expressed as follows:

φ|Ĥ |φ |C1 |2 + 2|C2 |2 + 3|C3 |2

.Eφ = =a .
φ|φ |C1 |2 + |C2 |2 + |C3 |2

(b) Let us introduce the following normalized coefficients:

C1
. C̃1 =  ,
|C1 |2 + |C2 |2 + |C3 |2
C2
C̃2 =  ,
|C1 |2 + |C2 |2 + |C3 |2
C3
C̃3 =  .
|C1 |2 + |C2 |2 + |C3 |2
232 7 Approximation Methods for Time Independent Schrödinger Equation

Then,
 
.Eφ = a |C̃1 |2 + 2|C̃2 |2 + 3|C̃3 |2
 
= a 1 + |C̃2 |2 + 2|C̃3 |2
 
= a 3 − 2|C̃1 |2 − |C̃1 |2 ,

where the fact that |C̃1 |2 +|C̃2 |2 +|C̃3 |2 = 1 has been used. The second equality
of the above expressions shows that the minimum value Eφ is a, which occurs
when C2 = C3 = 0 and the third equality shows that the maximum value of Eφ
is 3a, which occurs when C1 = C2 = 0.
(c) For the given Hamiltonian, H11 = a, H22 = 2a, and H33 = 3a and all other
terms are zero. On the other hand, Sij = δij . Therefore,
⎛ ⎞⎛ ⎞ ⎛ ⎞
a−E 0 0 C1 0
.⎝ 0 2a − E 0 ⎠ ⎝ C2 ⎠ = ⎝ 0 ⎠ .
0 0 3a − E C3 0

The above matrix is already diagonal. Thus, the three values of E are a, 2a, and
3a. The states corresponding to these values are respectively, |1, |2, and |3.
(0)
(d) The ground state is |1 and its energy is E1 = a. The first order perturbation
energy is zero as can be seen below.

(1) a
E1 = 1|Ĥ1 |1 =
. (1|12|1 + 1|21|1) = 0.
5
On the other hand, the coefficients for first order correction for the ground state
are as follows:

(1) 2|Ĥ1 |1 a 2|21|1 1

. C2 = (0) (0)
= =− ,
E1 − E2 5 −a 5
(1)
C3 = 0.

This implies that

1
|ψ1(1)  = − |2.
.
5
Therefore, up to the first order,

1
|ψ1  = |1 − |2.
.
5
Exercise Problems with Solutions 233

The second order perturbation of the ground state energy is as follows:

(2) 2|Ĥ1 |11|Ĥ1 |2 (a/5)2 a

E1 =
.
(0) (0)
= =− .
E1 − E2 a − 2a 25

Therefore, up to the second order of perturbation, the ground state energy

becomes corrected to
a 24
E1 = a −
. = a.
25 25

7.7 For a quantum particle of unit mass confined in a box of unit length, the
eigenfunctions and eigenvalues are given by
√
. x|ψn  = ψn (x) = 2 sin(nπ x) for 0 < x < 1,
π 2 h̄2 n2
En = ,
2
where n = 1, 2, 3, · · · . Now assume that the following potential energy is added
inside the box:

h̄2 x/5 , 0 < x < 0.5

V (x) =
.
h̄2 (1 − x)/5 , 0.5 < x < 1

Therefore, the total Hamiltonian becomes

p̂2
. Ĥ = + V (x̂).
2

(a) Calculate the following matrix elements:

. H11 = ψ1 |Ĥ |ψ1 ,

H22 = ψ2 |Ĥ |ψ2 ,
H12 = ψ1 |Ĥ |ψ2  = ψ2 |Ĥ |ψ1  = H21 .

(b) Consider a trial state given by the following linear combination

|φ = C1 |ψ1  + C2 |ψ2 ,

.

where C1 and C2 are real coefficients. Use the variational principle to determine
the approximation for the ground state energy.
(c) Use V (x̂) as a perturbation Hamiltonian, and calculate the first order perturba-
tion approximation for the ground state energy.
234 7 Approximation Methods for Time Independent Schrödinger Equation

Solution 7.7
(a) The matrix elements of the Hamiltonian can be calculated as follows:

H11 = ψ1 |Ĥ |ψ1 

.
 
h̄2 ∂ 2
 1
=2 dx sin(π x) − + V (x) sin(π x)
0 2 ∂x 2
  1  1/2
2 2 2 x
= h̄ π dx sin (π x) + 2 dx sin2 (π x)
0 0 5
 1 
1−x
+2 sin2 (π x) .
1/2 5

In the above expression,

 1/2 1
 1/2
. dx sin2 (π x)x = dx (1 − cos(2π x)) x
0 2 0

1
 1/2 1 1
= dx (x − x cos(2π x)) = + .
2 0 16 4π 2
1 2 1−x
Due to symmetry, 1/2 sin (π x) 5 has the same value. Therefore,

π2 π2
    
4 1 1 1 1
H11 = h̄2
. + + = h̄2 + + .
2 5 16 4π 2 2 20 5π 2

Similarly,
 
1 h̄2 d 2

H22 =
. ψ2 |Ĥ |ψ2  = 2 dx sin(2π x) − + V (x) sin(2π x)
0 2 dx 2
 2  1 
2 4π
= h̄ +2 dx sin(2π x)V (x) sin(2π x)
2 0

4 1/2 1
  
2 2
= h̄ 2π + dx (1 − cos(4π x))x)
5 0 2
 
1
= h̄2 2π 2 + .
20
Exercise Problems with Solutions 235

On the other hand,

 1
H12 = ψ1 |Ĥ |ψ2  = 2
. dx sin(π x)V (x) sin(2π x)
0
 1
= dx (cos(π x) − cos(3π x)) V (x)
0

h̄2 1/2

= dx (cos(π x) − cos(3π x)) x
5 0

h̄2 1

+ dx (cos(π x) − cos(3π x)) (1 − x)
5 1/2

h̄ 2  1/2
= dx (cos(π x) − cos(3π x)) x
5 0

h̄2 1/2

+ dx (cos(π(1 − x)) − cos(3π(1 − x))) x
5 0
h̄2 1/2

= dx (cos(π x) − cos(3π x)) x
5 0
h̄2 1/2

− dx (cos(π x) − cos(3π x)) x = 0.
5 0

(b) For the given trial state,

    
H11 − E 0 C1 0
. = .
0 H22 − E C2 0

Since the matrix is already diagonal, the solutions are simply E = H11 and E22 ,
for which the corresponding states are |φ = |ψ1  and |ψ2 , respectively.
(c) As shown in the answer for part (a),

4h̄2
 
1 1
.ψ1 |V (x̂)|ψ1  = + .
5 16 4π 2

Therefore,

π2
 
(0) (1) 1 1
E1 + E1 = h̄2
. + + .
2 20 5π 2
236 7 Approximation Methods for Time Independent Schrödinger Equation

7.8 For a particle of unit mass confined in a box of unit length, assume that the
following potential energy is added.

h̄2
V (x) =
. cos(π x).
2
Provide answers for (a), (b), and (c) in Problem 7.3 for this potential energy.
Solution 7.8
(a) The matrix elements of V̂ can be calculated as follows:
 1 1
2
. ψ1 |V̂ |ψ1  = 2h̄ dx sin2 (π x) cos(π x) = 0,
0 2
 1 1
ψ2 |V̂ |ψ2  = 2h̄2 dx sin2 (2π x) cos(π x) = 0.
0 2

The fact that the above integrals are zero can be confirmed easily by noting
that cos(π x) is antisymmetric with respect to x = 1/2 whereas sin2 (π x) and
sin2 (2π x) are symmetric. On the other hand,
 1
2
ψ1 |V̂ |ψ2  = h̄
. dx sin(π x) sin(2π x) cos(π x)
0

h̄2 1
  
= dx cos2 (π x) − cos(π x) cos(3π x)
2 0

h̄2 1 h̄2

= dx (1 + cos(2π x) − cos(2π x) − cos(4π x)) = .
4 0 4

Combining the above results with the eigenvalues of the zeroth order Hamilto-
nian, H11 = h̄2 π 2 /2, H22 = 2h̄2 π 2 , and H12 = h̄2 /4.
(b) The coefficients C1 and C2 of the trial state satisfy the following matrix
equation:
    
H11 − E H12 C1 0
. = .
H12 H22 − E C2 0

This is satisfied for nonzero coefficients only if the determinant is zero, which
leads to the following equation for E:
2
(H11 − E)(H22 − E) − H12
. = 0.
Exercise Problems with Solutions 237

The solutions for the above quadratic equation are as follows:


H11 − H22 2
 
1 2
.E = (H11 + H22 ) ± + H12
2 2
 ! 
2 2 5 3 1
= h̄ π ± 1+ .
4 4 9π 4

Thus, the variational approximation for the ground state energy is as follows:
 ! 
2 25 3 1
.Eg = h̄ π − 1+
4 4 9π 4
 
2 2 1
≈ h̄ π 1 − ,
24π 4

where the last line is an approximation but is a useful one with fairly good
accuracy. The coefficients corresponding to the above ground state energy can
be found by using the above energy in the matrix equation. The resulting
relationship between C1 and C2 is as follows:
⎧  ⎫
H11 − H22 2
 
1 ⎨1 2
⎬
.C2 = − (H11 − H22 ) + + H12 C1
H12 ⎩ 2 2 ⎭
 ! 
2 1 1
= 3π 1− 1+ C1 ≈ − C1 .
9π 4 6π 2

Then, C1 can be determined by the normalization condition.

(c) Since ψ1 |V̂ |ψ1  = 0, there is no first order correction for the ground state
energy.
7.9 The zeroth order Hamiltonian of a certain system is given by

Ĥ0 = |φ1 φ1 | + |φ2 φ2 | + |φ3 φ3 |,

.

where |φ1 , |φ2 , and |φ3  are orthonormal sates. A perturbation Hamiltonian Ĥ1 is
added to Ĥ0 , and the expression for Ĥ1 is as follows:

1
Ĥ1 = √ (|φ1 φ2 | + |φ2 φ1 | + |φ2 φ3 | + |φ3 φ2 |).
.
2

Use the degenerate perturbation theory to determine the lowest value of the first
order correction of the energy and the corresponding zeroth order ground state.
238 7 Approximation Methods for Time Independent Schrödinger Equation

Solution 7.9 The equation for the determinant |Ĥ1 − E| = 0 is as follows:

 
 −E √1 0 
 2 
 1 1  3 1 1
 2 −E 2  = −E + 2 E + 2 E = −E(E − 1)(E + 1) = 0.
.  √ √

 0 √12 −E 
 

Therefore, the lowest value of the first order perturbation correction is −1.
To find the zeroth order state corresponding this value, we need to solve the
following matrix equation.
⎛ ⎞
0 √1 0 ⎛ ⎞ ⎛ ⎞
⎜ 1 2 C1 C1
√1 ⎟ ⎝ C ⎠ = − ⎝ C ⎠ .
⎟
⎝ 2 0
.⎜ √
2⎠ 2 2
0 √1 0 C3 C3
2

The resulting coefficients satisfying the normalization condition are C1 = C3 =

1/2 and C2 = − √1 . Thus, the zeroth order state corresponding to E (1) = −1 is
√ 2
(|φ1  − 2|φ2  + |φ3 )/2.
7.10 Consider a zeroth order Hamiltonian Ĥ0 defined as follows:

Ĥ = b|11| + b|22| + 2b|33|,

.

where |1, |2, and |3 are orthonormal states and b is a positive real number. A
small perturbation term Ĥ1 is added to the Hamiltonian so that the total Hamiltonian
changes to Ĥ = Ĥ0 + Ĥ1 , where

b
Ĥ1 =
. (|12| + |21|).
3
Use the degenerate perturbation theory to find out the first order correction for the
ground state energy and the zeroth order correction for the ground eigenstate.
Solution 7.10 In the space of |1 and |2, 1|Ĥ1 |1 = 2|Ĥ1 |2 = 0 and 1|Ĥ1 |2 =
b/3. Therefore, according to the degenerate perturbation theory,
 2
 −λ b 
 
3 2 b
.
 b −λ  = λ − 3 = 0.
 
3

The solution of this is that λ = ±b/3. Therefore, the energy of the state 1 that is
corrected up to the first order is as follows:

b 2
E1(0) + E1(1) = b −
. = b.
3 3
Problems 239

The zeroth order eigenstate corresponding to this energy can be determined by

solving the following matrix equation:
b b
   
3 3 C1 0
. b b = ,
3 3 C2 0

which results in C1 + C2 = 0. Thus, after normalization, the zeroth order state can
be expressed as follows:

(0) 1
|φ1  = √ (|1 − |2) .
.
2

Problems

7.11 For a particle with unit mass subject to the following potential:

9 2 4
V (x) =
. h̄ x .
16
2
Use the following trial wavefunction φ(x) = e−αx /2 and the variational principle
to determine the best value α approximating the ground state wavefunction and the
best approximation for the ground state energy.
7.12 A zeroth order Hamiltonian Ĥ0 is given by

Ĥ0 = |φ1 φ1 | + 2|φ2 φ2 | + 3|φ3 φ3 |,

.

where |φ1 , |φ2 , and |φ3  are normalized and orthogonal to each other. Given that
the following perturbation Hamiltonian is added,

1 1
Ĥ1 =
. |φ1 φ1 | + |φ2 φ2 |
4 2
1 1 1 1
+ |φ2 φ3 | + |φ3 φ2 | − |φ1 φ3 | − |φ3 φ1 |,
2 2 2 2
answer the following questions.
(a) Determine the first order perturbation corrections for all the eigenvalues and
eigenstates of Ĥ0 .
(b) Determine the second order perturbation corrections for all the eigenvalues
of Ĥ0 .
240 7 Approximation Methods for Time Independent Schrödinger Equation

7.13 A zeroth order Hamiltonian Ĥ0 is given by

Ĥ0 = |φ1 φ1 | + |φ2 φ2 | + |φ3 φ3 |,

.

which has degeneracy of 3. Note that |φ1 , |φ2 , and |φ3  are normalized and
orthogonal to each other. Given that the following perturbation Hamiltonian is
added,

.Ĥ1 = λ (|φ1 φ1 | + 2|φ2 φ2 | + 3|φ3 φ3 |)

λ
+ (|φ1 φ2 | + |φ2 φ1 | − |φ2 φ3 | − |φ3 φ2 |) .
2
Use the degenerate perturbation theory to calculate the first order corrections for the
eigenvalues of Ĥ0 and the corresponding zeroth order corrections for the eigenstates
of Ĥ0 .
Chapter 8
Many Electron Systems and Atomic
Spectroscopy

The mathematical framework of quantum theory has passed

countless successful tests and is now universally accepted as a
consistent and accurate description of all atomic phenomena.
– Erwin Schrödinger

Abstract This chapter explains approximate quantum mechanical description of

many electron (more than one) atoms and their electronic spectroscopy. A quick
overview of independent electron model revisits well-established concepts of
atomic orbitals and configurations, and clarifies approximations involved in such
description. Symmetry properties of spin-orbit states and simple methods for
determining atomic term symbols are described. Rules for determining relative
energies of atomic terms and selection rules for atomic spectroscopy are clarified.
The chapter then provides a more detailed mathematical account of independent
electron approximations using helium atom as an example and then provides a
general description of the self-consistent approximation for many electron atoms.

8.1 Hamiltonian

The Hamiltonian operator for atoms with many electrons is as follows:

Ne
 2  Ne 
P̂2c  p̂j 1
.Ĥ = + + Ven (|r̂j − R̂c |) + Vee (|r̂j − r̂k |), (8.1)
2mc 2me 2
j =1 j =1 k=j

where .P̂c and .R̂c represent momentum and position operators of the nucleus, and
p̂j and .r̂j the momentum and position operators of the j th electron. .mc is the mass
.

of the nucleus and .me is the mass of the electron. .Ne is the number of electrons, and
the last term involves double summation over j and k running from 1 to .Ne except
that .j = k. .Ven and .Vee represent electron-nucleus and electron-electron Coulomb

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 241
S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_8
242 8 Many Electron Systems and Atomic Spectroscopy

potential energy operators that are defined as

Ze2
. Ven (|r̂|) = − ,. (8.2)
4π ǫ0 |r̂|
e2
Vee (|r̂|) = . (8.3)
4π ǫ0 |r̂|

Although Eq. (8.1) is a simple extension of the Hamiltonian for hydrogen-like

systems, the fact that it involves interactions of three or more particles causes the
corresponding quantum mechanical problem much more challenging. In particular,
the last electron-electron repulsion term makes it impossible to solve the corre-
sponding Schrödinger equation exactly.
Assuming that the nucleus is fixed at the origin (or considering in such coordinate
system), one can reduce Eq. (8.1) to the following many electron Hamiltonian:

Ne
p̂2j N
 
 1e 
Ĥel =
. + Ven (r̂j ) + Vee (|r̂j − r̂k |), (8.4)
2me 2
j =1 j =1 k=j

where .r̂j = |r̂j |. Even with this simplification, there is no exact analytic solution of
the Schrödinger equation for this Hamiltonian for .Ne ≥ 2. Therefore, one has to rely
on approximation methods. Due to the fact that electrons are fermions satisfying the
Pauli exclusion principle, care should be taken in employing approximation meth-
ods. We will first start with the simplest approximation of assuming independent
electrons, which can be justified with a proper application of the variational principle
as will be detailed later in this chapter.

8.2 Independent Electron Model

8.2.1 Major Assumptions

At the simplest level, the effects of electron-electron interactions can be accounted

for by introducing an effective single-electron potential, which modifies the original
electron-nucleus Coulomb interaction while treating each electron independently.
This means assuming that an electron labelled j , for example, can be described by
the following effective Hamiltonian:

p̂2j Zeff,j e2
Ĥj,eff =
. − + δVj (r̂j ), (8.5)
2me 4π ǫ0 r̂j

where .Zeff,j is a screened nuclear charge that the electron labeled as j experiences
and .δVj (r̂j ) represents additional effects of other electrons that cannot be expressed
8.2 Independent Electron Model 243

in the form of the Coulomb potential. In general, the value of .Zeff,j and the
detailed form of .δVj (r̂j ) should depend on the states of other electrons. This way,
interdependency of different electrons are taken into consideration implicitly.
Care should be taken in devising an approximation as outlined above because
the physical reality dictates that it is not possible to distinguish one electron from
the other. For this reason, the labeling of electrons does not have genuine physical
meaning and should be viewed as an intermediate step for calculating physical
properties that are independent of the specific manner of the labeling. In addition,
any many-electron state should be antisymmetric with respect to the exchange of
any two electrons, which brings further complication. These issues will be addressed
later in this chapter. For now, let us assume that an effective single electron state can
be determined by solving the Schrödinger equation for the Hamiltonian of the type
given by Eq. (8.5).
Because of the fact that the potential energy in Eq. (8.5) depends only on
the magnitude operator of the position, .r̂j = |r̂j |, the effective Hamiltonian
commutes with the angular momentum operators .L̂2 and .L̂z . Therefore, within
this approximation, the angular momentum quantum numbers l and .ml still serve
as good quantum numbers. Another important fact is that the effective electron-
nucleus Coulomb term of the potential remains dominant compared to the additional
correction. This makes the principal quantum number of n determined for a
hydrogen-like system a nearly good quantum number for classifying the eigenstates
of the Hamiltonian. However, there is major difference from the hydrogen-like
system in that the degeneracy of the energy for different values of l, which is a
non-negative integer in the range of .[0, n − 1] for a given quantum number n, gets
lifted due to the additional term .δVj (r̂j ). Important qualitative trends and major
definitions of terms are provided below.

Summary of Independent Electron Model

• For many-electron atoms, .(n, l, ml ) serve as nearly good quantum numbers
to represent effective spatial single electron states, which are also termed
as orbitals.
• The collection of orbitals having the same value of n constitute a shell.
Shells with .n = 1, 2, 3, and 4, . . . are respectively denoted as K, L, M,
N, .. . . .
• The energy of each orbital depends on both n and l, which are used for the
labeling of different orbitals. Orbitals with .l = 0, 1, 2, 3, 4, .. . . are called
s, p, d, f , g, .· · · , respectively.
• For each value of l, there are .2l + 1 degenerate orbitals (in the absence of
external field) with different values of m.
• For a given shell, orbitals with smaller values of l have lower energies than
those with higher values of l. This is because electrons tend to have higher
probability density near the nucleus for smaller value of l, which causes
them to experience lower Coulomb potential energy.
244 8 Many Electron Systems and Atomic Spectroscopy

Table 8.1 Numbers of Orbital s p d f g h

orbitals and maximum
numbers of electrons allowed Number of orbitals 1 3 5 7 9 11
for different types of orbitals Number of electrons 2 6 10 14 18 22

Given that all the single electron states have been identified in a given atom, one
can then assign electrons to those states as if assigning one electron does not affect
the other except for satisfying the symmetry rule for electrons as fermions. For this,
first, it is important to note that a state of an electron is defined by specifying both
its orbital (spatial state) and the spin state. Since electrons are fermions, only one
electron can occupy one state. Since there are two spin states for each electron, this
means that a single orbital can be occupied by two electrons at most.

8.2.2 Orbitals and Electronic Configuration

Since the number of orbitals for a given l is (.2l + 1), the maximum number of
electrons allowed for orbitals with l are .2(2l + 1), as shown in Table 8.1.
Assuming that electrons for a given atom can be filled in one by one such that
each of them occupies the lowest possible orbital, which is called Aufbau principle
(building-up principle), it is possible to assign the relative order of orbitals.

The Order of Orbitals Based on Relative Energies

Given that all the orbitals with lower energies have already been filled, the
following order of energies of orbitals has been established empirically.

− Increasing relative energy .→

.

1s
2s 2p
3s 3p 3d
. 4s 4p 4d 4f
5s 5p 5d 5f
6s 6p 6d
7s 7p

In the above lists, each row represents different principal quantum number,
and the orbitals on the right have higher energies than those on the left. Note
that the intervals between different orbitals do not reflect the actual values of
energy difference. Rather, the differences of energies become smaller as the
principal quantum number increases.
8.2 Independent Electron Model 245

The main factor that determines the relative energies are principal quantum
numbers for s and p orbitals. However, for d and f orbitals, this is not the case.
For example, nd orbitals have higher energies than .(n + 1)s orbitals but lower
energies than .(n + 1)p orbitals. The atoms for which the d orbitals are partially
filled correspond to transition metal elements. Also note that nf orbital has higher
energy than .(n + 2)s orbital but lower energy than .(n + 1)d orbital. The atoms for
which the f orbitals are partially filled are known as lanthanides and actinides.
With these rules, it is now possible to determine all the ground state electron
configurations for all the many-electron atoms. Some examples of ground state
electronic configuration are shown below.

. Li :1s 2 2s 1
C :1s 2 2s 2 2p2
Fe :1s 2 2s 2 2p6 3s 2 3p6 4s 2 3d 6

On the other hand, the following configurations correspond to excited state elec-
tronic configurations for the same atoms.

. Li :1s 1 2s 2 ; 1s 2 2p1
C :1s 1 2s 1 2p3 ; 1s 1 2s 2 2p1 3s 1
Fe :1s 2 2s 2 2p6 3s 2 3p6 4s 2 3d 5 4p1 ; 1s 2 2s 2 2p6 3s 2 3p6 4s 1 3d 6 4p1

8.2.3 Spin States

The total spin quantum number of two electrons occupying the same orbital is zero
because it is possible only when their spins are opposite to each other. Likewise,
when any type of orbitals is fully filled, the sum of all the spins of electrons is
zero. Therefore, the total spin quantum number is zero for the following completely
filled configurations: .s 2 , .p6 , .d 10 , .f 14 , and .g 18 . For partially filled orbitals, there are
various ways for the spins of electrons to be aligned with respect to each other. As
a result, multiple spin states are possible.
Before detailed consideration of the spin states of partially filled orbitals, some
mathematical account of the symmetry properties of many electron quantum states
is needed. First of all, it is important to examine the properties of electrons as
fermions. For many electron systems, this means that the total state should change
its sign if two electrons are exchanged with each other. For example, consider the
following multi-electron state:

|Ne  = |1, 2, · · · , j, · · · , k, · · · , Ne ,
. (8.6)
246 8 Many Electron Systems and Atomic Spectroscopy

where j denotes the label of each electron and the position of each label represents
a single electron spin-orbit state that each electron is assigned to. For .|N  to be
an acceptable N electron state, it should change its sign if electrons j and k are
exchanged with each other as follows:

|Ne  = −|1, 2, · · · , k, · · · , j, · · · , Ne .
. (8.7)

For two electrons, this means that the spin states are antisymmetric when the orbital
states are symmetric and vice versa.
For example, let us consider the s orbital for which there is only one orbital
available for two electrons. The only possible two-electron orbital state in this case
is

|s (1, 2) = |φs (1) ⊗ |φs (2) = |φs (1)|φs (2),
. (8.8)

which is symmetric with respect to exchange of the two electrons because the order
of two states in a direct product does not make any difference. The only two-electron
spin state compatible with the above two-electron s orbital state is the following
antisymmetric spin state called singlet:

1
|χS (1, 2) = √ (|α(1)|β(2) − |α(2)|β(1)) ,
. (8.9)
2

which satisfies the antisymmetric property, .|χS (1, 2) = −|χS (2, 1). For this state,
the total spin of electron is zero. Thus, the only acceptable two-electron spin-orbit
state where two electrons occupy the same orbital is

|(1, 2) = |s (1, 2)|χS (1, 2),

. (8.10)

which satisfies the antisymmetric property, .|(1, 2) = −|(2, 1) and has the total
spin quantum number of .S = 0.
On the other hand, one can also construct symmetric spin states. For two
electrons, there are three such states as follows:
⎧
⎪
⎨ |α(1)|α(2)
1
.|χT (1, 2) = √ (|α(1)|β(2) + |β(1)|α(2)) (8.11)
⎪ 2
|β(1)|β(2)
⎩

These three states are collectively called triplet and have the total spin quantum
number .S = 1.
Let us now consider the case where two different orbitals .|φa  and .|φb  are
available for two electrons. There are two possible ways for the two electrons occupy
8.2 Independent Electron Model 247

these orbitals without violating the fact that they are indistinguishable as follows:

1
|sym (1, 2) = √ (|φa (1)|φb (2) + |φa (2)|φb (1)), .
. (8.12)
2
1
|asym (1.2) = √ (|φa (1)|φb (2) − |φa (2)|φb (1)). (8.13)
2

Then, the total spin-orbit states possible for the two electrons are as follows:

. |S (1, 2) = |sym (1, 2)|χS (1, 2), . (8.14)

|T (1, 2) = |asym (1, 2)|χT (1, 2). (8.15)

Note that both of above two states are antisymmetric with respect to the exchange of
two electrons. These two states show that fully decoupled orbital and spin parts can
be used to construct all possible and independent spin-orbit states for two electrons.
However, it is not always the case for more than two electrons.

8.2.4 Energy Levels of Spin-Orbit States

While it is possible to determine relative energies of different configurations, a given

electronic configuration does not provide full information on the energy of the many
electron system because there are still many ways for the total angular momenta
to be added up. For this, it is necessary to define all possible angular momentum
quantum numbers that can be defined for many electron systems.

Angular Momentum Quantum Numbers for Many Electron System

Let us denote the orbital angular momentum vector and spin vector of the ith
electron as .li and .si , respectively. Different types of total angular momenta
exist as described below.
• Total angular momentum quantum number of a single electron (j )—
This represents the sum of orbital angular momentum and spin of an
electron. Because .s = 1/2, for a given value of l, there are only two
possible values, .j = l + 1/2, l − 1/2.
• Total orbital angular momentum quantum number of all the electrons
(L)—This represents the quantum number for the sum of the orbital
angular momenta of all the electrons.
• Total spin quantum number of all the electrons (S) -This represents the
quantum number for the sum of the spins of all the electrons.

(continued)
248 8 Many Electron Systems and Atomic Spectroscopy

• Total angular momentum quantum number of all the electrons (J )—

This represents the sum of all the angular momenta of all the electrons.
• Each of the angular momentum quantum numbers listed above corre-
sponds to the quantum numbers of the squares of corresponding angular
momentum vector operators, .ĵ, .L̂, .Ŝ, and .Ĵ. In other words, .h̄2 j (j + 1),
2 2 2
.h̄ L(L + 1), .h̄ S(S + 1), and .h̄ J (J + 1) are respectively eigenvalues
2 2 2 2
of .ĵ , .L̂ , .Ŝ , and .Ĵ . With this understanding, for the determination of
the quantum numbers, one has to consider only the vector nature of the
orbital angular momentum operators. Therefore, we will omit the operator
symbols whenever possible, and consider only the corresponding vectors
in the consideration below.

Since the total angular momentum, namely, the sum of all orbital angular
momenta and spins of all electrons, .Ĵ, is a conserved quantity due to the spherical
symmetry of the system, its quantum number J is always a good quantum number
and can be included in any classification of the eigenstates of many-electron atoms.
There are two ways to find out this quantum number. First, we can add all the orbital
angular momentum vectors and spin vectors as follows:

L=
. li , . (8.16)
i

S= si . (8.17)
i

Then, the total sum of the angular momentum and the spin vectors is given by

J = L + S.
. (8.18)

This is called LS-coupling or Russell-Saunders coupling scheme, and works well

when spin-orbit coupling of an individual electron is weak as in most light atoms.
This is not the case for heavy atoms for which the large nuclear charge creates strong
spin-orbit coupling especially for those core electrons. For this case, the following
individual .j’s serve as better quantum numbers.

j i = li + s i .
. (8.19)

The total angular momentum vector in this case is given by


J=
. ji . (8.20)
i
8.2 Independent Electron Model 249

This is called jj -coupling scheme, which becomes complicated as the number of

electrons increases.

8.2.5 Examples of Energy Levels Based on LS-Coupling

Scheme

Within the LS-coupling scheme, one only needs to specify L, S, and J values. The
electronic term symbols introduced in Chap. 6 can be used for this purpose. Let us
consider the simplest example, two electrons in helium, .He. The ground electronic
configuration of .He is .1s 2 . Since this corresponds to a completely filled orbital,
.L = 0 and .S = 0. Therefore, .J = 0 as well. This configuration has only one term,
1
. S0 .

Next, let us consider an excited state configuration of .He given by .1s 1 2p1 . For
this, only .L = 1 is possible. On the other hand, for the total spin, .S = 0, 1 are
possible. Since two electrons occupy two different orbitals, either .S = 0 or .S = 1
is acceptable. When .S = 0, only .J = 1 is possible. The term symbol for this case is
1
. P1 . When .S = 1, three values of .J = 0, 1, 2 are possible. Term symbols for these

cases are as follows: .3 P0 , .3 P1 , and .3 P2 .

For .Li, which has three electrons, the ground electronic configuration is .1s 2 2s 1 .
This is effectively one electron system since two electrons are in a completely filled
orbital. The orbital angular momentum quantum number of this electron is .L = 0,
since it is in s orbital. For this one electron, .S = 1/2. The only possible combination
of this spin and the orbital angular momentum .L = 0 is such that .J = 1/2.
Therefore, the term symbol for this is .2 S1/2 .
For .Be, there are four electrons. However, the ground electronic configuration is
2 2
.1s 2s , where two s orbitals are completely filled. As in the case of .He, .L = 0,
1
.S = 0, and .J = 0 for this configuration. Therefore, the term symbol for this is . S0 .

For boron, .B, there are five electrons. Out of these, four electrons completely
fill in 1s and 2s orbitals and only one electron occupies a 2p orbital. Therefore,
the ground electronic configuration is .1s 2 2s 2 2p1 . For this, .L = 1 and .S = 1/2.
Possible values of J in this case are .3/2 and .1/2. Therefore, there are two terms,
2 2
. P1/2 and . P3/2 .

Determination of term symbols so far have been simple because orbitals are
completely filled or occupied with single electrons. When there are more than one
electrons occupying the same orbital, care should be taken. This is the case for
carbon, .C, for which the ground electronic configuration is .1s 2 2s 2 2p2 . Although
all the electrons are equivalent for a given atom, for practical reasons, it is
assumed that electrons assigned to different orbitals have unique labels and are
thus nonequivalent. This is certainly an approximation but a useful one. Within this
approximation, electrons in the filled orbitals need not be considered because their
net orbital and spin angular momenta are zero. Thus, two electrons in unfilled p
orbitals only need to be considered.
250 8 Many Electron Systems and Atomic Spectroscopy

Table 8.2 Possible L S J Term Symbols

combinations of L and S and 1
term symbols for two 2 0 2 . D2

1 1 2,1,0 3 3 3
. P2 , P1 , P0
equivalent p electrons
0 0 0 1
. S0

For the two electrons occupying 2p orbitals, the total angular momentum
quantum numbers of .L = 2, 1, 0 are possible. For the total spin, two quantum
numbers for the total spin .S = 0, 1 are possible. Due to the symmetry requirement,
not all the three values of L can be combined with the two values of S. Instead, only
the combinations given in Table 8.2 are possible.
A simple way of finding term symbols without going through full angular
momentum calculation as introduced by Landau and Lifshitz [7] is described below.
Note that each electron has .l = 1 and .s = 1/2. Therefore, there are six possible
combinations for the sets of .(lz , sz ) as follows:

1 1 1
a : 1, b : 0, c : −1,
.
2 2 2
1 1 1
a ′ : 1, − b′ : 0, − c′ : −1, −
2 2 2

The set of .(Lz , Sz ) can be found by summing two out of the above six states. In
enumerating all possible sums, one needs to consider only two rules. First, the sum
of the same set is not allowed because it is not possible to make it antisymmetric
with respect to the exchange of two electrons. Second, we only need to consider
nonnegative values of .Lz and .Sz because the numbers of positive and negative values
are always the same.
The resulting list of all possible distinctive combinations with non-negative
values of .Lz and .Sz are as follows:

a + a ′ : (2, 0) a + b : (1, 1) a + c : (0, 1)

a + b′ : (1, 0) a + c′ : (0, 0)
.
a ′ + b : (1, 0) a ′ + c : (0, 0)
b + b′ : (0, 0) ,

where each column corresponds to a different value of .Lz . For each column, those
with higher value of .Sz appear in upper rows. Based on the above combinations, it
is possible to deduce all possible term symbols.
First, the presence of .(2, 0) means that there has to be a state with .L = 2 and
.S = 0. This also means that there have to be two additional combinations: .(1, 0)

and .(0, 0), which correspond to different possible nonnegative values of .Lz for
.L = 2 and .S = 0. Thus, we have found three possible combinations of .Lz and
1
.Sz corresponding to the term . D. Removing these combinations from the list of
8.2 Independent Electron Model 251

combinations shown above, we are now left with the following combinations:

(1, 1) (0, 1)
. (1, 0) (0, 0)
(0, 0)

Note that the origin of each of the above combinations has been left out because
it does not matter in identifying term symbols. In the above combinations, the
presence of .(1, 1) implies that there has to be a state with .L = 1 and .S = 1. This
implies the existence of the following combinations: .Lz = 1, .Sz = 1, 0; .Lz = 0,
.Sz = 1, 0. All of these correspond to .(1, 1), .(1, 0), .(0, 1), .(0, 0), which constitute

the term .3 P. Subtracting these from the above combinations, we are now left with the
following single combination: .(0, 0). This single combination implies that .L = 0
and .S = 0, which corresponds to the term . 1 S. Thus, we have identified all three
terms that appear in Table 8.2, . 1 D, . 3 P, and . 1 S. For each case, all possible values
of J can be obtained following the rules for the addition of angular momentum
quantum numbers explained before.
The approach described above can be extended to the case of three electrons.
Consider the configuration .p3 . All the possible combinations are as follows:

1 1 3
a + a ′ + b : (2, ) a + a ′ + c : (1, ) a + b + c : (0, )
2 2 2
1 1
a + b + b′ : (1, ) a + b′ + c : (0, )
.
2 2
′ 1
a + b + c : (0, )
2
′ 1
a + b + c : (0, )
2
From the above combinations, following a similar procedure, we can find out the
following term symbols: . 4 S 3 , . 2 P 3 , . 2 P 1 , . 2 D 5 , and . 2 D 3 .
2 2 2 2 2
Having identified all possible terms, the next step is to determine the relative
energy levels. There are three well known rules [31] for this. First, two rules by
Hund prescribe relative energies based on the values of L and S as described below.

Hund’s Rules
According to Hund, relative energies of different terms can be determined
based on the following two rules:
(i) Of all the terms arising from the same electronic configuration, those with
the highest multiplicity (largest value of S) lie lowest in energy.

(continued)
252 8 Many Electron Systems and Atomic Spectroscopy

(ii) Of all the terms with the same value of S arising from the same electronic
configuration, the lowest is that with the highest value of L.
The justification for (i) is that the states with the largest multiplicity have the
lowest electron-electron repulsion. The justification for (ii) is that the highest
L states can penetrate toward the nucleus more easily, and thus experiences
larger effective charge and thus lower energy.

Of the terms with the same values of L and S that originate from the same
electronic configuration, Lande’s interval rule as described below then prescribes
relative energies depending on the value of J .

Lande’s Interval Rule

For less than half-filled orbitals, smaller J has lower energy (normal case).
For more than half filled orbitals, larger J has lower energy (inverted case).
This is due to the spin-orbit coupling, which results in the following form of
energy difference between different values of J :

EJ − EJ −1 = AJ.
. (8.21)

A is positive for normal case and is negative for inverted case. The above
prescription leaves the case with exactly half-filled case undetermined, for
which the rule for normal case seems to work in general.

8.2.6 Atomic Spectroscopy: Selection Rules and Simple

Examples

Atomic spectroscopy results from the transition between many electron states
due to interaction with a photon. The Russell-Saunders coupling scheme and
the corresponding electronic terms can be used for fairly accurate description of
the spectroscopy of light atoms and the relevant selection rules [31] are easy to
understand.
8.2 Independent Electron Model 253

Selection Rules for General Many-Electron Atoms

The general selection rules governing the transition between two electronic
states of many-electron atoms, .(L′ , S ′ , J ′ ) → (L′′ , S ′′ , J ′′ ), are as follows.

. S = 0. (8.22)
L′ 0
±1, 0 if =
L= . (8.23)
1 if L′ = 0
J = 0, ±1 ( no 0 ↔ 0 transition) (8.24)

The fact that the total spin of electrons does not change, Eq. (8.22), results
from the fact that electrons interact only with the electric field component of the
light which does not affect the spins of electrons. Of course, this selection rule
can be broken if there is significant spin-orbit coupling. The selection rule for
. L, Eq. (8.23), results from the fact that the sum of the total angular momentum

of electrons and the spin of photon, which is one, should remain conserved. The
selection rule concerning . J is an outcome of those for . S and . L.
Another important selection rule that is fundamental but can be overlooked easily
is the Laporte’s rule as described below.

Laporte’s Rule
The transition in atomic spectroscopy is allowed only between terms origi-
nating from two different electronic configurations that satisfies the following
condition:

. li : even ↔ odd (8.25)
i

This rule results from the fact that the dipole operator involved in the
interaction with the light has odd inversion symmetry and thus can connect
only two electronic configurations with different inversion symmetries, which
are determined by the sum of all individual angular momentum quantum
numbers. As a result of this rule, transitions between terms coming from the
same electronic configurations are not allowed.

As the simple examples of the atomic spectroscopy, some major results for
the spectroscopy of alkali metal and alkaline earth metal atoms [31] are briefly
described below.
254 8 Many Electron Systems and Atomic Spectroscopy

Alkali Metals The emission spectra of alkali metal atoms consist of four series: S
(sharp, .s → p), P (principal, .p → s), D (diffuse, .d → p), and F (fundamental,
.f → d). Only one electron is involved in these transitions and the selection rules

are similar to those for hydrogen atoms. That is, . n is unrestricted, . l = ±1, and
. J = 0, ±1 except that .J = 0 ↔ J = 0 is forbidden. Note that . J = 0 can be

consistent with . l = ±1 due to the interplay of the spin state except when .J = 0.
The principal series in the sodium atom result from the transitions, .n 2 P1/2 →
3 2 S1/2 and .n 2 P3/2 → 3 2 S1/2 , where .n ≥ 3. The lowest of these are called sodium
D lines, which have the wavelengths of .589.592 nm and .588.995 nm.
Helium and Alkaline Earth Metal Atoms In the ground state where the two electrons
are in the same s-type orbital, the total spin state is singlet. However, when one
electron is excited and thus there are two nonequivalent electrons, both singlet
and triplet states are possible. However, the triplet excited states are not optically
accessible from the ground state because of the following selection rules: . S = 0.
That is, only singlet-singlet and triplet-triplet transitions, with . l = ±1 for the
promoted electron, are allowed.

8.3 Case Study of Helium Atom

This section provides a detailed mathematical description of independent electron

model and then more advanced self-consistent field (SCF) approximation for the
case of helium atom.

8.3.1 Hamiltonian and Schrödinger Equation

The full Hamiltonian for a helium atom is given by

h̄2 2 h̄2 2 h̄2 2

ĤHe = −
. ∇c − ∇1 − ∇
2mc 2me 2me 2
2e2 2e2 e2
− − + . (8.26)
4π ǫ0 |r1 − Rc | 4π ǫ0 |r2 − Rc | 4π ǫ0 |r1 − r2 |

Working in terms of the center of mass and relative coordinates in this case is
mathematically complicated. Thus, using the fact that .mc >> me , let us assume
that the nucleus is fixed at the origin and that the kinetic energy of the nucleus can
be ignored. With this assumption, the following Hamiltonian for the electrons in
helium atom can be used:

h̄2 2 h̄2 2 2e2 2e2 e2

ĤHe,el = −
. ∇1 − ∇2 − − + , (8.27)
2me 2me 4π ǫ0 r1 4π ǫ0 r2 4π ǫ0 r12
8.3 Case Study of Helium Atom 255

where .r12 = |r1 − r2 |. Then, let us introduce .ψHe,e (r1 , r2 ) as the eigenfunction
of the above Hamiltonian that satisfies the following time independent Schrödinger
equation:

ĤHe,el ψHe,e (r1 , r2 ) = EψHe,e (r1 , r2 ).

. (8.28)

The two-electron wave function .ψHe,e (r1 , r2 ) is defined in the six dimensional
space of the two electron coordinates .r1 and .r2 , and provides full information on
the electrons within the helium atom under the assumption that the nuclear kinetic
energy, the contribution of which is extremely small, can be ignored. Unfortunately,
no exact solution has been found even for this simplest example of the multi-electron
system. Thus, one has to rely on approximation methods.

8.3.2 Independent Electron Model with Variational

Optimization of Effective Charge

Let us consider the simplest independent electron model that assumes .ψHe,e (r1 , r2 )
as the product of two eigenfunctions for hydrogen-like systems. For this, it is useful
to divide the Hamiltonian as follows [1]:

e2 1
.ĤHe,el = ĤH,Z=2 (1) + ĤH,Z=2 (2) + , (8.29)
4π ǫ0 r12

where .ĤH,Z=2 (1) and .ĤH,Z=2 (2) represent the Hamiltonian of a single electrons 1
and 2 around a nucleus of charge .Z = 2 at the origin, respectively, and are expressed
as

h̄2 2 2e2 1
ĤH,Z=2 (j ) = −
. ∇j − , j = 1, 2 (8.30)
2me 4π ǫ0 rj

On the other hand, the Hamiltonian of a single electron interacting with a nucleus
with charge Z at the origin is

h̄2 2 Ze2
ĤH,Z = −
. ∇ − . (8.31)
2me 4π ǫ0 r

Eigenvalues for the above Hamiltonian can be expressed as

Z 2 h̄2 1
En = −
. , for n = 1, 2, 3, · · · (8.32)
2me a02 n2
256 8 Many Electron Systems and Atomic Spectroscopy

where .a0 is the Bohr radius (the effective radius of the hydrogen atom) defined as

4π ǫ0 h̄2
a0 =
. . (8.33)
me e 2

Similarly, the eigenfunction of .ĤH,Z , which depends on the three quantum numbers
as detailed in Chap. 6, is given by

. ψn,l,ml (r, θ, φ) = Rnl (r)Ylml (θ, φ), (8.34)

where .Ylml (θ, φ) is the spherical harmonics and .Rnl (r) is the radial function defined
as

1/2 l+ 32
(n − l − 1)! 2Z 2l+1 2Zr
Rnl (r) =
. r l Ln−l−1 e−Zr/(na0 ) . (8.35)
2n[(n + l)!]3 na0 na0

2l+1 2Zr
In the above expression, .Ln−l−1 na0 is an associated Laguerre polynomial as
defined in Chap. 6.
Now let us consider the following trial wave function:

φ0 (r1 , r2 ) = ψ1s (Z, r1 )ψ1s (Z, r2 ) .

. (8.36)

where .ψ1s (Z, r) is the 1s eigenfunction for the hydrogen-like system with nuclear
charge Z, a parameter to be determined later, and is given by
 1/2
Z3
ψ1s (Z, r) = ψ100 (r) =
. e−Zr/a0 . (8.37)
π a03

Then, it is easy to show that

h̄2 2
 
Ze2 1 (2 − Z)e2 1
.ĤH,Z=2 (j )ψ1s (Z, rj ) = − ∇ − − ψ1s (Z, rj )
2me j 4π ǫ0 rj 4π ǫ0 rj

Z 2 h̄2
 
(2 − Z)e2 1
= − − ψ1s (Z, rj ). (8.38)
2me a02 4π ǫ0 rj

Therefore, application of the Hamiltonian, Eq. (8.29), to the trial wavefunction

φ0 (r1 , r2 ) results in
.

Z 2 h̄2
 
(2 − Z)e2 1 1 e2
.ĤHe,el φ0 (r1 , r2 ) = − − + + φ0 (r1 , r2 ).
me a0 2 4π ǫ0 r1 r2 4π ǫ0 r12
(8.39)
8.3 Case Study of Helium Atom 257

As is clear from Eq. (8.39), .φ0 (r1 , r2 ) cannot be an eigenfunction .ĤHe,e even
when .Z = 2 because of the last electron-electron repulsion term. In fact, .Z = 2 is
not expected to be the best choice considering that the electron-electron repulsion
term causes partial screening of the nuclear charge. According to the variational
theorem as described in Chap. 7, the best choice of Z is the one that minimizes the
expectation value of the Hamiltonian. Therefore, let us first calculate the following
expectation value:
 
Eφ =
. dr1 dr2 φ0 (r1 , r2 )ĤHe,e φ0 (r1 , r2 )

Z 2 h̄2 2(2 − Z)e2 1


=− − dr ψ1s (Z, r) ψ1s (Z, r)
me a02 4π ǫ0 r
e2 1
 
+ dr1 dr2 φ0 (r1 , r2 ) φ0 (r1 , r2 ). (8.40)
4π ǫ0 r12

In the above expression, the two integrals can be calculated explicitly as follows:

1

Z
. dr ψ1s (Z, r) ψ1s (Z, r) = , . (8.41)
r a0
1 5Z
 
dr1 dr2 φ0 (r1 , r2 ) φ0 (r1 , r2 ) = . (8.42)
r12 8a0

Thus, employing the above expressions and the definition of .a0 given by Eq. (8.33),
one can express the expectation value .Eφ as follows [1]:

h̄2 5 h̄2 27Z

 
2 2
.Eφ = − Z + 2Z(2 − Z) − Z = Z − . (8.43)
me a02 8 me a02 8

Equation (8.43) shows that the expectation value of the Hamiltonian is a

quadratic function of Z. Thus, it becomes minimum for the value of Z that satisfies
the following condition:

h̄2 27

∂E
. = 2
2Z − =0, (8.44)
∂Z me a0 8

which results in .Zopt = 27/16. This implies that the electron-electron repulsion
can be seen as screening the charge of the nucleus by .5/16. Using this value in
Eq. (8.43), we find that the best estimate for the ground state electronic energy of
the helium atom within the present approximation is

2
27 h̄2 h̄2
Emin = −
. = −2.8477 . (8.45)
16 me a02 me a02
258 8 Many Electron Systems and Atomic Spectroscopy

8.3.3 Self Consistent Field (SCF) Approximation for Helium

As a preliminary step for a general description of SCF approximations, let us use

the helium atom as an example and go through approximations involved in this
approach in more detail [1]. In order to clarify the main ideas in this approach, let
us now use the atomic and Gauss units where .h̄ = me = a0 = e = 1. Then, the
expression for the electronic Hamiltonian of the helium atom can be expressed as

1 1 2 2 1
ĤHe,el = − ∇12 − ∇22 −
. − + . (8.46)
2 2 r1 r2 r12

Let us assume that the two electron wavefunction can once again be expressed as
a product of single electron wavefunctions as follows: .ψ(r1 , r2 ) = ψ1 (r1 )ψ2 (r2 ),
which are assumed to be normalized. Then,
 
.E[ψ1 .ψ2 ] = dr1 dr2 ψ ∗ (r1 , r2 )Ĥ ψ(r1 , r2 ) = I1 [ψ1 ] + I2 [ψ2 ] + J12 [ψ1 , ψ2 ],
(8.47)

where the square brackets imply that they are functionals of the functions within
them. Thus,

1 2

I1 [ψ1 ] =
. dr1 ψ1∗ (r1 ) − ∇12 − ψ1 (r1 ), . (8.48)
2 r1
1 2

dr2 ψ2∗ (r2 ) − ∇22 −
I2 [ψ2 ] = ψ2 (r2 ), . (8.49)
2 r2
1
 
J12 [ψ1 , ψ2 ] = dr1 dr2 ψ1∗ (r1 )ψ2∗ (r2 ) ψ1 (r1 )ψ2 (r2 ). (8.50)
r12

Let us also define

1

eff
.V
12 (r1 )[ψ2 ] = dr2 ψ2∗ (r2 ) ψ2 (r2 ), (8.51)
r12

which is an effective potential energy that electron 1 experiences due to electron 2

and is thus the functional of .ψ2 (r). Then,

eff
.J12 [ψ1 , ψ2 ] = dr1 ψ1∗ (r1 )V12 (r1 )[ψ2 ]ψ1 (r1 ). (8.52)

Alternatively, we can also define

1

eff
V21 (r2 )[ψ1 ] =
. dr1 ψ1∗ (r1 ) ψ1 (r1 ), (8.53)
r12
8.3 Case Study of Helium Atom 259

which is an effective potential energy that electron 2 experiences due to electron 1

and is also a functional of .ψ(r1 ). Then,

eff
.J12 [ψ1 , ψ2 ] = dr2 ψ2∗ (r2 )V21 (r2 )[ψ1 ]ψ2 (r2 ). (8.54)

Finally, let us introduce an effective total potential for electron 1,

2 eff
V1 (r1 )[ψ2 ] = −
. + V12 (r1 )[ψ2 ], (8.55)
r1

and define an effective one-electron Hamiltonian for particle 1,

1
Ĥ1 [ψ2 ] = − ∇12 + V̂1 (r1 )[ψ2 ].
. (8.56)
2
Then, we can define an effective one-electron wavefunction that satisfies the
following Schrödinger equation:

Ĥ1 [ψ2 ]ψ1 (r1 ) = E1 [ψ1 , ψ2 ]ψ1 (r1 ),

. (8.57)

where the fact that .E1 [ψ1 , ψ2 ] is a functional of both .ψ1 and .ψ2 according to the
following identity has been used.

E1 [ψ1 , ψ2 ] =
. dr1 ψ1∗ (r1 )Ĥ1 [ψ2 ]ψ1 (r1 ) = I1 [ψ1 ] + J12 [ψ1 , ψ2 ]. (8.58)

Similarly, we can also introduce an effective total potential for electron 2 as follows:

2 eff
V2 (r2 )[ψ1 ] = −
. + V21 (r2 )[ψ1 ], (8.59)
r2

and define the Hamiltonian and effective one-electron Schrödinger equation for the
wavefunction for particle 2 as follows:

1
. Ĥ2 [ψ1 ] = − ∇22 + V̂2 (r2 )[ψ1 ], . (8.60)
2
Ĥ2 [ψ1 ]ψ2 (r2 ) = E2 [ψ1 , ψ2 ]ψ2 (r2 ). (8.61)

Then,

E2 [ψ1 , ψ2 ] =
. dr2 ψ2∗ (r2 )Ĥ2 ψ2 (r1 ) = I2 [ψ2 ] + J12 [ψ1 , ψ2 ]. (8.62)

Equations (8.57) and (8.61) are effective single electron equations, and constitute
the major equations to be solved. It is important to note that these are different
from conventional one particle Schrödinger equations because the terms constituting
260 8 Many Electron Systems and Atomic Spectroscopy

the Hamiltonian depend on the choice of as yet undetermined wavefunction. Thus,

these have to be solved in a self consistent manner. For example, one can start from
the first guess of the wavefunctions and then continue iterations. One important
point to recognize in conducting such iterations is that the two effective one-electron
wavefunctions are not independent but are coupled together as follows:

eff
. ψ2(0) (r2 ) → V12 (r1 )[ψ2(0) ] → Ĥ1 [ψ2(0) ] → ψ1(1) (r1 ) → · · · . (8.63)
eff
ψ1(0) (r1 ) → V21 (r2 )[ψ1(0) ] → Ĥ2 [ψ1(0) ] → ψ2(1) (r2 ) → · · · (8.64)

This procedure can continue until .E1 and .E2 being determined converge. Note that
the total energy is not equal to the sum of orbital energies. Rather, it is given by

E = I1 + I2 + J12 = E1 + E2 − J12 ,
. (8.65)

where the last term accounts for the electron-electron repulsion that is double
counted when the two orbital energies are added up.
The SCF approximation described above is known as the Hartree approximation
and in general produces a better solution than simply assuming an eigenfunction for
a hydrogen-like system for the single electron wavefunction. Of course, the actual
performance of this approach depends on the choice of basis functions used for
expanding the single electron wavefunction.
One important issue that was not considered so far is the symmetry requirement
for the final solution since the total two-electron wavefunction should be antisym-
metric with respect to the exchange of the two electrons. Assuming that the two
electrons occupy the same orbital, ensuring such symmetry requirement is easy. One
(k) (k)
simply needs to assume that .ψ1 (r) = ψ2 (r) throughout the iteration procedure
and use the singlet spin state. However, if more than two electrons and one orbital
are involved, such procedure can be cumbersome. Thus, it is best to devise a method
that satisfies the symmetry requirement.

8.4 Self Consistent Field (SCF) Approximation for Many

Electron Atoms

Consider many electron atoms or ions with nuclear charge of Z. The classical
Hamiltonian for this system in atomic units is

N N N
1e e
Z 1e 
1
p2j −

.Hel = +
2 rj 2 rj k
j =1 j =1 j =1 k=j

Ne N
 1e 
1
= hj + , (8.66)
2 rj k
j =1 j =1 k=j
8.4 Self Consistent Field (SCF) Approximation for Many Electron Atoms 261

where the second line defines .hj , the single electron term of the Hamiltonian
representing its kinetic energy and potential energy with the nucleus, and .rj k =
|rj − rk | with .rj and .rk respectively denoting the coordinates of electrons j and k.
The corresponding quantum mechanical Hamiltonian operator can be expressed as

Ne N
 1e 
1
Ĥel =
. ĥj + , (8.67)
2 r̂j k
j =1 j =1 k=j

where
p̂j2 Z
. ĥj = − ,. (8.68)
2 r̂j
1 1
= . (8.69)
r̂j k |r̂j − r̂k |

Let us denote an .Ne -electron eigenstate of .Ĥel with eigenvalue .ENe as .|Ne .1 Thus,

Ĥel |Ne  = ENe |Ne .

. (8.70)

It is straightforward to extend the independent electron model and the SCF

approximation that have been detailed for the helium atom to find approximations
for the above many-electron Schrödinger equation. In the description below, we go
back to the Dirac notation for more compact presentation of formulation.

8.4.1 Hartree Approximation

The Hartree approximation can easily be extended to the general n electron state.
For this, let us introduce single electron states, .|ψ1 (1), · · · , |ψNe (Ne ), where .ψk
denotes the kth spin orbit state and the argument inside of this is the label of each
electron. Thus, each of the single electron state represents the electron 1 in state
.ψ1 , the electron 2 in state .|ψ2 , and so on. Let us also assume for now that the .Ne -

electron state can be expressed as the direct product of the single electron states as
follows:2

|Ne  = |ψ1 (1) ⊗ |ψ2 (2) ⊗ · · · ⊗ |ψNe (Ne )

.

= |ψ1 (1)|ψ2 (2) · · · |ψNe (Ne ). (8.71)

1 Although there are many eigenstates, which all have to be labeled with appropriate indices, we
do not consider them now but focus only on one of them, particularly, the ground electronic state
here.
2 Apparently, this state is not consistent with the fact that all electrons are equivalent let alone it

does not satisfy the Pauli exclusion principle.

262 8 Many Electron Systems and Atomic Spectroscopy

Within this approximation, following a similar procedure as in the helium atom, the
Hartree approximation for the total energy can be expressed as follows:

. ENe [ψ1 , · · · , ψNe ]

⎛ ⎞
Ne Ne 
 1  1 ⎠
= ψ1 (1)| · · · ψNe (Ne )| ⎝ ĥj + |ψ1 (1) · · · |ψNe (Ne )
2 r̂j k
j =1 j =1 k=j
⎛ ⎞
Ne
⎝Ij [ψj ] + 1
 
= Jj k [ψj , ψk ]⎠, (8.72)
2
j =1 j =k

where

. Ij [ψj ] = ψj (j )|ĥj |ψj (j ), . (8.73)

1
Jj k [ψj , ψk ] = ψj (j )| ψk (k)| |ψj (j )|ψk (k). (8.74)
r̂j k

In above expressions, those in the square brackets denote their functional depen-
dences. In other words, .ENe depends on the choice of all of .|ψ1 (1), · · · , |ψNe (Ne ),
.Ij only on the choice of .|ψj (j ), and .Jj k on the choice of the two states .|ψj (j )

and .|ψk (k). From now on, for notational conveniences, these dependences will be
omitted.
On the other hand, let us introduce the following effective one-electron Hamilto-
nian operator:

Ĥj = ĥj +
. Jj k |ψj (j ) ψj (j )|, (8.75)
k=j

where the second term represents the potential of electron j due to others and is
based on the assumption that each of electron j is in the state .|ψj (j ), which is
yet undetermined. Now, let us assume that .|ψj (j ) can be determined to be the
eigenstate of .Ĥj with eigenvalue .Ej as follows:
⎛ ⎞

Ĥj |ψj (j ) = ⎝ĥj +
. Jj k ⎠ |ψj (j ) = Ej |ψj (j ). (8.76)
k=j

Then, taking the inner product of the above equation with . ψj (j )|, we obtain the
following expression for the energy of electron j :

Ej = Ij +
. Jj k , (8.77)
k=j
8.4 Self Consistent Field (SCF) Approximation for Many Electron Atoms 263

where the definition of Eq. (8.73) has been used. The total energy of the .Ne -
electron system, Eq. (8.72), can thus be expressed using the above definition of
single electron energy as follows:

Ne N
 1e 
ENe =
. Ej − Jj k . (8.78)
2
j =1 j =1 k=j

Equations (8.76)–(8.78) serve as the formal framework for the iteration proce-
dure in the Hartree approximation, and one can continue the following procedure
until convergence in .ENe is reached.
(0) (0)
1. Choose the initial guess states for .|ψ1 (1), .· · · , .|ψNe (Ne ).
2. Determine .ĥj and .Jj k using the states determined in step (1).
3. Solve the one electron Schrödinger equation, Eq. (8.76), with .ĥj and .Jj k
determined from step (2), which results in the next approximation for states,
(1) (1)
.|ψ
1 (1), .· · · , .|ψNe (Ne ).
4. Repeat the procedures (2) and (3), until convergence is reached in the energy
determined by Eq. (8.78).
Although the procedure of iteration within the Hartree approximation, as
sketched above, is a powerful method applicable to general many-body problems, in
practice, it is not applicable to actual many-electron systems because the resulting
solution is not guaranteed to satisfy the symmetry requirement. Unlike the case with
the helium atom that has only two electrons, it is not straightforward to check or
ensure the symmetry requirement separately from the iteration procedure.

8.4.2 Hartree-Fock Approximation

The Hartree-Fock (HF) approximation is similar to the Hartree approximation

except that it ensures that the symmetry property is satisfied by considering trial
states that meet the symmetry condition at the outset. A well-known expression
for the many-electron state as a linear combination of the products of single
electron states, which satisfies the antisymmetry property, is the following Slater
determinant:
 
 |ψ1 (1) |ψ2 (1) · · · |ψNe (1) 
 
1  |ψ1 (2) |ψ2 (2) · · · |ψNe (2) 
.|Ne  = √  .. .. .. .. . (8.79)
Ne !  . . . . 

 |ψ (N ) |ψ (N ) · · · |ψ (N ) 
1 e 2 e Ne e
264 8 Many Electron Systems and Atomic Spectroscopy

For the ground electronic state where there are .Ne = 2Nr electrons that completely
fill in .Nr orbitals, the above state can be expressed as
 
 |φ (1)|α(1) |φ1 (1)|β(1) · · · |φNr (1)|β(1) 
 1
1 |φ (2)|α(2) |φ 1 (2)|β(2) · · · |φNr (2)|β(2) 
 
 1
.|2Nr  = √  .. .. .. .. 
(2Nr )! 

. . . .


 
 |φ1 (2Nr )|α(2Nr ) |φ1 (2Nr )|β(2Nr ) · · · |φNr (2Nr )|β(2Nr ) 
(8.80)

Equivalently, this determinant state can also be expressed as follows:

1 
(−1)NP P̂ |φ1 (1)|α(1) · · · |φNr (2Nr )|β(2Nr ) ,
 
|2Nr  = √
.
(2Nr )!
P̂
(8.81)

where .P̂ represents a permutation operation of .1, · · · , 2Nr electron indices and
NP is the number of pairwise permutations constituting a given permutation.3 It
.

is straightforward to show that the above definition satisfies the antisymmetric

property with respect to any exchange of two electron indices. Since .φk ’s correspond
to orbitals for the same atom, the following orthonormality relation holds:

. φk (l)|φj (l) = δkj . (8.82)

Now, let us consider the following Hamiltonian for .2Nr electrons:

2Nr 2N
 1 r  1
Ĥel =
. ĥμ + . (8.83)
2 μ r̂μν
μ=1 ν=μ

Then, the total electronic energy within the HF approximation is obtained by taking
expectation value of the above Hamiltonian with respect to the antisymmetric state
of 2N electrons, Eq. (8.81).

Expression for the Total Energy Within the HF Approximation

Taking the expectation value of the Hamiltonian .Ĥel with respect to the single
determinant many-electron state, Eq. (8.81), one can obtain the following
expression for the HF energy:

(continued)

3 There is well established theorem that any permutation can be expressed as composite of pairwise

permutations.
8.4 Self Consistent Field (SCF) Approximation for Many Electron Atoms 265

2Nr 2N 2N
 1 r r 1
. H F = 2Nr |Ĥel |2Nr  =
E 2Nr |ĥμ |2Nr + 2Nr | |2Nr 
2 μ r̂μν
μ=1 ν=μ

Nr
 Nr 
 Nr
=2 Ip + (2Jpq − Kpq ), (8.84)
p=1 p=1 q=1

where

. Ip = φp (1)|ĥ1 |φp (1), . (8.85)

1
Jpq = φp (1)| φq (2)| |φp (1)|φq (2), . (8.86)
r̂12
1
Kpq = φp (1)| φq (2)| |φq (1)|φp (2). (8.87)
r̂12

In the above expressions, .Ip is the energy of the orbital p for a single electron,
. Jpq is called Coulomb integral, and .Kpq is called exchange integral. Detailed
derivation of Eq. (8.84) is provided below.

Let us consider detailed derivation of all the terms that appear in Eq. (8.84). First,
the expectation value of the single electron term .ĥμ is as follows:

2Nr 2N N
 1 r r
. 2Nr |ĥμ |2Nr  = (2Nr − 1)! 2 φp (μ)|ĥμ |φp (μ)
(2Nr )!
μ=1 μ=1 p=1

Nr Nr
1  
= 2 · 2Nr φp (1)|ĥ1 |φp (1) = 2 Ip , (8.88)
2Nr
p=1 p=1

where the fact that . φp (μ)|ĥμ |φp (μ) is independent of the particular choice of .μ
has been used in replacing all .2Nr expressions in terms of the one with .μ = 1. On
the other hand,

2Nr 
 1
. 2Nr | |2Nr 
r̂ μν
μ=1 ν=μ

2N 2N
1 r  r  1
= (2Nr − 2)! ψj (μ)| ψk (ν)| |ψj (μ)|ψk (ν)
(2Nr )! r̂ μν
μ=1 ν=μ j =1 k=j

1

− ψj (μ)| ψk (ν)| |ψk (μ)|ψj (ν)
r̂ μν
266 8 Many Electron Systems and Atomic Spectroscopy

2Nr 
 1
= ψj (1)| ψk (2)| |ψj (1)|ψk (2)
r̂ 12
j =1 k=j

1

− ψj (1)| ψk (2)| |ψk (1)|ψj (2) , (8.89)
r̂ 12

where the negative sign for the second term within the curly bracket reflects the
antisymmetric property of the Slater determinant, given by Eq. (8.81). In other
words, this terms results from the component of the .1/r̂μν operator between a
particular component of the .2Nr -electron product in the bra of Eq. (8.81) and
another term of the ket where two electrons .μ and .ν have switched their states.
The resulting ket should have an opposite sign from that of the bra state because
they are different by one pairwise exchange of two electrons. The second equality
in Eq. (8.89) reflects the fact that the two integral terms involving .μ and .ν are
independent of particular choice of these indices. Thus, .2Nr (2Nr − 1) of such terms
can all be replaced with the one with .μ = 1 and .ν = 2.
Note that each single electron state is a spin-orbit state. In other words, for .j =
2p − 1,

|ψ2p−1 (μ) = |φp (μ)|α(μ).

. (8.90)

On the other hand, for .j = 2p,

|ψ2p (μ) = |φp (μ)|β(μ).

. (8.91)

Now, the components in the summation in Eq. (8.89) can be simplified depending
on whether the indices are even or odd as follows.
For .j = 2p − 1, .k = 2p or .j = 2p, .k = 2p − 1,

1
. ψj (1)| ψk (2)| |ψj (1)|ψk (2)
r̂ 12
1
= φp (1)|φp (2)| |φp (1)|φp (2) = Jpp . (8.92)
r̂ 12

On the other hand, the second term that results from the exchange of two electrons
in this case is zero as follows:
1
. ψj (1)| ψk (2)| |ψk (1)|ψj (2) = 0. (8.93)
r̂ 12

This is because the spin .α and .β states are orthogonal to each other.
8.4 Self Consistent Field (SCF) Approximation for Many Electron Atoms 267

For .j = 2p − 1, .k = 2q − 1 or .j = 2p, .k = 2q with .p = q,

1
. ψj (1)| ψk (2)| |ψj (1)|ψk (2)
r̂ 12
1
= φp (1)|φq (2)| |φp (1)|φq (2) = Jpq . (8.94)
r̂ 12

On the other hand, for this case,

1
. ψj (1)| ψk (2)| |ψk (1)|ψj (2)
r̂ 12
1
= φp (1)| φq (2)| |φq (1)|φp (2) = Kpq , (8.95)
r̂ 12

which is nonzero in general.

For .j = 2p − 1, .k = 2q or .j = 2p, .k = 2q − 1 with .p = q,

1
. ψj (1)| ψk (2)| |ψj (1)|ψk (2)
r̂ 12
1
= φp (1)|φq (2)| |φp (1)|φq (2) = Jpq . (8.96)
r̂ 12

On the other hand, the second exchange term for this case is zero as follows:

1
. ψj (1)| ψk (2)| |ψk (1)|ψj (2) = 0. (8.97)
r̂ 12

This is because the spin .α and .β states are orthogonal to each other.
Combining all of the above cases, Eq. (8.89) can be expressed as

2Nr 
 1
. ψj (1)| ψk (2)| |ψj (1)|ψk (2)
r̂ 12
j =1 k=j

1

− ψj (1)| ψk (2)| |ψk (1)|ψj (2)
r̂ 12
⎧ ⎫
Nr ⎨
  ⎬
=2 Jpp + (2Jpq − Kpq )
⎩ ⎭
p=1 q=p

Nr 
 Nr
=2 2Jpq − Kpq , (8.98)
 

p=1 q=1

where, in the last equality, the fact that .Kpp = Jpp has been used. This completes
the proof of Eq. (8.84).
268 8 Many Electron Systems and Atomic Spectroscopy

Having demonstrated the validity of Eq. (8.84), let us now consider its relation-
ship with the single electron picture. For this, let us consider the following effective
single electron Hamiltonian operator:

Nr

Ĥp (1) = ĥ1 +
. 2Jˆpq (1) − K̂pq (1) , (8.99)
q=1

where .Jˆpq (1) and .K̂pq (1) are operators defined for electron 1 by the following
relations:

1
. Jˆpq (1)|φp (1) = φq (2)| |φq (2) |φp (1), . (8.100)
r̂ 12
1
K̂pq (1)|φp (1) = φq (2)| |φp (2) |φq (1). (8.101)
r̂ 12

Let us assume that .|φp (1) is the eigenstate of .Ĥp (1) with eigenvalue .Ep as follows:
⎛ ⎞
Nr

Ĥp (1)|φp (1) = ⎝ĥ1 +
. 2Jˆpq (1) − K̂pq (1) ⎠ |φp (1) = Ep |φp (1).
q=1
(8.102)
Then, taking inner product of the above equation with . φp (1)|, we obtain the
following expression:

Nr

.Ep = Ip + (2Jpq − Kpq ). (8.103)
q=1

This is the energy of each orbital. In this expression, 2.Jpq is the regular Coulomb
interaction term representing interactions with other electrons. The factor of 2 in
this term accounts for the contributions of both spin up and down states. On the
other hand, .−Kpq represents purely quantum mechanical interaction that comes
from exchange of two electrons. This is nonzero only when the two electrons are
in the same spin state. Note that the interaction between two electrons in the same
orbital, .Jpp , is also included in Eq. (8.103), for which .Kpp = Jpp . With the above
expression, the HF energy can be expressed as follows:

Nr
 Nr 
 Nr
EH F = 2
. Ep − (2Jpq − Kpq ). (8.104)
p=1 p=1 q=1

Equations (8.102)–(8.104) serve as the formal framework for the iteration procedure
in the HF approximation. Thus, one can continue the following procedure until
convergence in .EH F is reached.
8.5 Summary and Questions 269

(0) (0)
1. Choose the initial guess states for .|φ1 (1), .· · · , .|φNr (1), which are all
orthogonal to each other.
2. Determine .ĥ1 , .Jˆpq (1), and .K̂pq (1) using the states determined in step (1).
3. Solve the one electron Schrödinger equation, Eq. (8.102), with .ĥ1 , .Jˆpq (1), and
.K̂pq (1) determined from step (2), which results in the next approximation for the
(1) (1)
orbital states, .|φ1 (1), .· · · , .|φNr (1).
4. Repeat the procedures (2) and (3), until convergence is reached in the energy
determined by Eq. (8.104).
Although the above HF approximation accounts for the effect of many electrons
on average, it misses the effects of correlation between different electrons, which
requires going beyond the assumption of trial state as a single Slater-type deter-
minant. The approach of assuming the state as a linear combination of multiple
determinants is usually called multi-configuration methods. Another important
effect that is missing in the HF approximation described above is the spin-orbit
interaction and/or relativistic effect, which become important for heavy atoms.

8.5 Summary and Questions

For many-electron atoms, the corresponding time independent Schrödinger equa-

tions cannot be solved exactly. However, due to the spherical symmetry of the
system having only one nuclear center and the Pauli exclusion principle that leads
to substantial reduction of repulsive interactions between electrons, it is possible to
come up with an approximate scheme that is simple enough to offer good physical
insights into many-electron atoms. The simplest of this kind is the independent
electron model that assumes that the collection of many-electron states can be
viewed as filling up electrons one by one on top of others according to the Aufbau
(Building-up) principle that prescribes putting electrons in single electron orbitals
that are available and have the lowest possible energies.
Within the independent electron model, the degeneracy of the single electron
orbitals for the same value of the principal quantum number is lifted because the
effective electron-nuclear potential, even though it remains centrally symmetric,
is no longer in Coulombic form. In general, orbitals with smaller orbital angular
momentum quantum number l have lower energy than those with higher l for the
same principal quantum number n. This is because the former tends to penetrate
better into the region closer to the nucleus than the latter. While these energy
differences between different values of l are small compared to the principal energy
differences for small n, the two become comparable as n becomes large. This results
in unique properties of transition metal elements, lanthanides, and actinides.
For many electron atoms, the sum of total orbital angular momenta of electrons
and total spin of electrons remains conserved and the corresponding quantum
number, normally called J , serves as a good quantum number. For the case where
the spin-orbit interactions are small, which remains a good approximation except for
heavy atoms, the total orbital quantum number L and the total spin S serve as nearly
270 8 Many Electron Systems and Atomic Spectroscopy

good quantum number and can be effectively used to specify different electronic
states. With proper consideration of the Pauli exclusion principle that forbids having
two electrons in the exactly the same spin-orbit states, it is possible to come up with
all distinctive states that can be specified by the values of L, S, and J . Then, relative
orders of these states can be determined according to the application of Hund’s rules
and Lande’s interval rule.
Selection rules for the atomic spectroscopy of many atoms can be understood on
the basis of terms classified according to the values of L, S, and J and employing
the fact that the interaction with the radiation is through the dipole operator, which
causes the change of L at most one and leaves S to remain unchanged. The fact
that the dipole operator has odd inversion symmetry also leads to Laporte’s rule for
allowed transitions.
More quantitative description of many-electron states can be conducted by
applying the variational principle in a self-consistent way. The simplest of this
is called the Hartree approximation, which however does not take the symmetry
property of many electron states into consideration. A proper approach that also
accounts for the Pauli-exclusion principle is the HF approximation. This serves
as fairly accurate method of calculating many electron states although it does not
account for correlation effects of many electrons and other effects that require more
advanced theories.

Questions
• What is the physical basis for Aufbau principle? Can this be justified based
on the perturbation theory?
• What is the difference between a configuration and a state for many
electron systems?
• What is the major difference between LS-coupling and jj -coupling
schemes?
• Why are not all combinations of L and S possible for states for electrons
occupying the same orbital?
• How can Hund’s rules be understood? Can they be violated?
• What is the physical origin of Lande’s interval rule?
• What are selection rules for L, S, and J in electronic absorption or
emission involving electronic states of many electron atoms? How can they
be explained?
• Can one apply the Hartree approximation for a collection of identical
bosons?
• What is the origin of the exchange interaction in the HF approximation?

Exercise Problems with Solutions

8.1 Find out all the atomic term symbols for p5 d 1 , and list them in the order of
increasing energy (assuming Lande’s interval rule for more than half filled orbitals.).
Exercise Problems with Solutions 271

Solution 8.1 p5 d 1 has the same terms as p1 d 1 . Thus, we consider the latter. Since
there are two electrons with l1 = 1 and l2 = 2 occupying different orbitals, L = 1,
2, and 3, and for each case S = 0 and 1 are possible. Therefore, the following
combinations and terms are possible.

L S J Terms
1 0 1 1P
1
1 1 0,1,2 3P , 3P , 3P
0 1 2
2 0 2 1D
2
2 1 1,2, 3 3D , 3D , 3D
1 2 3
3 0 3 1F
3
3 1 2, 3, 4 3F , 3F , 3F
2 3 4

The order of the above terms from the lowest to the highest energies is as follows:
3
. F4 , 3 F3 , 3 F2 , 3 D3 , 3 D2 , 3 D1 , 3 P2 , 3 P1 , 3 P0 , 1 F3 , 1 D2 , 1 P1

8.2 Find all the atomic term symbols for d 2 configuration, and list them in the order
of increasing energy. Specify where Hund’s rule and Lande’s interval rule are used.
Solution 8.2 For the given configuration, there are the following ten states of
(lz , sz ) for one electron:

a : (2, 21 ) b : (1, 12 ) c : (0, 21 ) d : (−1, 12 ) e : (−2, 12 )

.

a ′ : (2, 21 ) b′ : (1, 12 ) c′ : (0, 21 ) d ′ : (−1, 12 ) e′ : (−2, 21 )

The combination of two different states among the above leads to the following pairs
for (Lz , Sz ):

a + a ′ : (4, 0) a + b : (3, 1) a + c : (2, 1) a + d : (1, 1) a + e : (0, 1)

a + b′ : (3, 0) a + c′ : (2, 0) b + c : (1, 1) b + d : (0, 1)
a ′ + b : (3, 0) a ′ + c : (2, 0) a + d ′ : (1, 0) a + e′ : (0, 0)
. b + b′ : (2, 0) a ′ + d : (1, 0) a ′ + e : (0, 0)
b + c′ : (1, 0) b + d ′ : (0, 0)
b′ + c : (1, 0) b′ + d : (0, 0)
c + c′ : (0, 0)

In the above list, the presence of (4, 0) means that the combination of L = 4 and
S = 0 exists, for which (3, 0), (2, 0), (1, 0), and (0, 0) can also be removed from
the list. The remaining list thus becomes
272 8 Many Electron Systems and Atomic Spectroscopy

(3, 1) (2, 1) (1, 1) (0, 1)

(3, 0) (2, 0) (1, 1) (0, 1)
(2, 0) (1, 0) (0, 0)
.
(1, 0) (0, 0)
(1, 0) (0, 0)
(0, 0)

where we have removed the combinations of labels how the above pairs are formed
because they are not important for determining the terms. In the above list, the
presence of (3, 1) means that the combination of L = 3 and S = 1 exists, for
which (3, 0), (2, 1), (2, 0), (1, 1), (1, 0), (0, 1), (0, 0) can also be removed from the
list. The remaining list thus becomes

(2, 0) (1, 1) (0, 1)

(1, 0) (0, 0)
.
(1, 0) (0, 0)
(0, 0)

In the above list, the presence of (2, 0) means that the combination of L = 2 and
S = 0 exists, for which (1, 0) and (0, 0) can also be removed from the list. The
remaining list thus becomes

(1, 1) (0, 1)
. (1, 0) (0, 0)
(0, 0)

In the above list, the presence of (1, 1) means that the combination of L = 1 and
S = 1 exists, for which (1, 0), (0, 1), and (0, 0) can also be removed from the list.
As a result, the only remaining list becomes (0, 0), which corresponds to L = 0 and
S = 0. Summing up all of those identified above, we find that

L S J Terms
4 0 4 1G
4
3 1 2, 3, 4 3 F2 , 3 F3 , 3 F4
2 0 2 1D
2
2 1 0, 1, 2 3 P0 , 3 P1 , 3 P2
0 0 0 1S
0

Applying Hund’s rules and Lande’s rule, terms from the lowest energy to the
highest energy can be ordered as follows:
3
. F2 , 3 F3 , 3 F4 , 3 P0 , 3 P1 , 3 P2 , 1 G4 , 1 D2 , 1 S0
Exercise Problems with Solutions 273

8.3 For an oxygen atom in the ground configuration [He]2s2 2p4 , find out the term
symbol for the ground electronic state. Assuming that the oxygen atom makes
transition to an excited state configuration [He]2s2 2p3 3d1 , find out term symbols
for all the excited states to which the transition from the ground electronic state is
allowed.
Solution 8.3 2s 2 p4 has the same terms as p 2 , for which, as explained in the main
text, the following terms can be identified:
1
. D2 , 3 P2 , 3 P1 , 3 P0 , 1 S0

Since the p orbital is more than half-filled, according to Lande’s interval rule,
3P is the ground state term. Now let us consider possible terms for the excited
2
configuration, which is equivalent to 2p3 3d 1 . As described in the main text, for p3 ,
the following three cases are possible.
1
L = 2, S =
.
2
1
L = 1, S =
2
3
L = 0, S =
2
For each case, adding l = 2 and s = 1/2 of the d orbital electron, we obtain the
following combinations:

. L = 0, 1, 2, 3, 4; S = 0, 1
L = 1, 2, 3; S = 0, 1
L = 2, ; S = 1, 2

Therefore, possible terms can be obtained as follows:

L S J Terms
0 0 0 1S
0
0 1 1 3S
1
1 0 1 1P
1
1 1 0,1,2 3P , 3P , 3P
0 1 2
2 0 2 1D
2
2 1 1, 2, 3 3D , 3D , 3D
1 2 3
2 2 0, 1, 2, 3, 4 5 D0 , 5 D1 , 5 D2 , 5 D3 , 5 D4
3 0 3 1F
3
3 1 2, 3, 4 3F , 3F , 3F
2 3 4
4 0 4 1G
4
4 1 3, 4, 5 3G , 3G , 3G
3 4 5
274 8 Many Electron Systems and Atomic Spectroscopy

Among the terms above, transition from 3 P2 to the following terms are allowed:
3S , 3P , 3P , 3D , 3D , 3D
1 1 2 1 2 3

Problems

8.4 Find out all the atomic term symbols for s 1 p5 d 1 , and list them in the order of
increasing energy (assuming Lande’s interval rule for more than half filled orbitals.).
8.5 For an excited configuration of C given by 1s 2 2s 1 2p3 , find out all possible
terms.
8.6 For the ground electronic state of Be with the configuration 1s 2 2s 2 , find
out total electronic energy within the Hartree-Fock approximation by employing
Eq. (8.84). Express your answer in terms of the following quantities:

. I1s = 1s(1)|ĥ1 |1s(1),

I2s = 2s(1)|ĥ1 |2s(1),
1
J1s1s = 1s(1)| 1s(2)| |1s(2)|1s(1),
r̂ 12
1
J2s2s = 2s(1)| 2s(2)| |2s(2)|2s(1),
r̂ 12
1
J1s2s = 1s(1)| 2s(2)| |2s(2)|1s(1),
r̂ 12
1
K1s2s = 1s(1)| 2s(2)| |2s(1)|1s(2).
r̂ 12
Chapter 9
Polyatomic Molecules and Molecular
Spectroscopy

We have sought for firm ground and found none. The deeper we
penetrate, the more restless becomes the universe; all is rushing
about and vibrating in a wild dance.
– Max Born

Abstract This chapter provides a general quantum mechanical account of poly-

atomic molecules, and describes the Born-Oppenheimer approximation that allows
calculation of adiabatic electronic states for fixed nuclear coordinates. Taking
diatomic molecules as examples, independent electron model and the approximation
of linear combination of atomic orbitals as molecular orbitals (LCAO-MO) are
described, and the molecular term symbols for corresponding electronic states are
explained. As the next example, the LCAO-MO approximation with additional
assumptions due to Hückel is used to describe electronic states of .π -conjugated
molecules. The chapter then provides a brief overview of the group theory and its
applications for constructing MOs and analyses of vibrational spectroscopy.

The Hamiltonian operator for a polyatomic molecule with .Nu nuclei and .Ne
electrons consists of two terms as follows:

ĤM = Ĥnu + Ĥen ,

. (9.1)

where .Ĥnu represents the nuclear kinetic and potential energies, and .Ĥen represents
the kinetic and potential energies of electrons. The potential energy of electrons
includes both electron-nuclei interactions and those between electrons. In atomic

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 275
S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_9
276 9 Polyatomic Molecules and Molecular Spectroscopy

units, these two terms are expressed as

Nu Nu 
 P̂2c 1 Zc Zc′
Ĥnu =
. + ,. (9.2)
2Mc 2 ′
|Rc − Rc′ |
c=1 c=1 c =c

Ne
 p̂2μ  Nu
Ne 
Zc e2 1
Ne 
1
Ĥen = − + . (9.3)
2 |r̂μ − R̂c | 2 |r̂μ − r̂ν |
μ=1 μ=1 c=1 μ=1 ν=μ

In Eq. (9.2), .Mc , .P̂c , .R̂c , and .Zc are the mass, momentum operator, position operator,
and charge of the nucleus labeled as c. In Eq. (9.3), .p̂μ and .r̂μ are the momentum
and position operators of the electron labeled as .μ.
Let us denote the eigenstate of the above molecular Hamiltonian .ĤM as .|M ,
for which the time independent Schrödinger equation is

ĤM |M  = (Ĥnu + Ĥen )|M  = EM |M .

. (9.4)

Solving the above equation is challenging even numerically. In particular, the large
disparity of the masses of electrons and nuclei make it difficult to come up with
common length and energy scales applicable to both of them. Thus, due to both
conceptual and numerical reasons, it is much more beneficial to treat them at
different levels. The standard approach for this is the Born-Oppenheimer (BO)
approximation that treats nuclear motions adiabatically and ignores the coupling
between electron and nuclear kinetic energies as described in detail below.

9.1 Born-Oppenheimer Approximation

Let us denote the whole set of nuclear coordinates as .R = (R1 , · · · , RNu ) and
the corresponding position state bra of the nuclear coordinates as .R|. Taking inner
product of this with Eq. (9.4), we obtain

.R|ĤM |M  = R|Ĥnu |M  + R|Ĥen |M  = EM R|M . (9.5)

In the above expression, the nuclear term involving .Ĥnu can be calculated using
Eq. (9.2) and is expressed as
⎧ ⎫
Nu Nu 
⎨  h̄2 2 1  Zc Zc′ ⎬
.R|Ĥnu |M  = − ∇ + R|M . (9.6)
⎩ 2Mc c 2 ′
|Rc − Rc′ | ⎭
c=1 c=1 c =c
9.1 Born-Oppenheimer Approximation 277

On the other hand, the term involving .Ĥen in Eq. (9.5) can be simply expressed as

R|Ĥen |M  = Ĥen (R)R|M ,

. (9.7)

where .Ĥen (R) is the same as Eq. (9.3) except that .R̂ is replaced with .R as follows:

Ne
 p̂2μ  Nu
Ne 
Zc e2
Ne 
1 1
Ĥen (R) =
. − + . (9.8)
2 |r̂μ − Rc | 2 |r̂μ − r̂ν |
μ=1 μ=1 c=1 μ=1 ν=μ

Equation (9.8) represents the Hamiltonian for electrons with fixed nuclear coor-
dinates as parameters. Let us now define adiabatic electronic states as eigenstates of
this Hamiltonian. Thus,

Ĥen (R)|ψe,a (R) = Ee,a (R)|ψe,a (R),

. (9.9)

where a denotes the index specifying different adiabatic electronic eigenstates for
the fixed nuclei.
Before moving to the next step, it is worthwhile to note the parametric depen-
dence of Eq. (9.9) on the nuclear coordinates .R = (R1 , · · · , RNu ). As indicated,
both the eigenstates and eigenvalues depend parametrically on .R. Although not
shown, in fact, even the index a depends implicitly on .R. As a consequence, two
different adiabatic states for two different nuclear coordinates are not in general
orthogonal as indicated below.

ψe,a ′ (R′ )|ψe,a (R) = 0.

. (9.10)

Once we have specified an adiabatic electronic state .|ψe,a (R), let us assume
that the inner product of .R| with the total molecular wavefunction .|M  can be
expressed as1

R|M  = ψnu,a (R)|ψe,a (R).

. (9.11)

The implication of the above expression is that the molecular state at .R is defined
by electrons in the particular adiabatic state .|ψe,a (R), whereas the wavefunction
for nuclei is .ψnu,a (R). Employing Eq. (9.11) in Eq. (9.7), we obtain

Ĥen (R)R|M  = Ĥen (R)ψnu,a (R)|ψe,a (R) = ψnu,a (R)Ĥen (R)|ψe,a (R),
.

(9.12)

where the second equality results from the fact that .Ĥen (R) does not involve
any derivative with respect to nuclear coordinates. On the other hand, employing

1 More generally, one can consider a linear combination of these.

278 9 Polyatomic Molecules and Molecular Spectroscopy

Eq. (9.11) in Eq. (9.6), we obtain

⎧ ⎫
Nu Nu 
⎨  h̄2 2 1  Zc Zc′ ⎬
.R|Ĥnu |M  = − ∇ + ψnu,a (R)|ψe,a (R).
⎩ 2Mc c 2 ′
|Rc − Rc′ | ⎭
c=1 c=1 c =c

(9.13)

In the above equation, due to the dependence of the electronic state .|ψe,a (R) on the
nuclear coordinates, derivatives with respect to nuclear coordinates entail couplings
between electronic and nuclear degrees of freedom as follows:

∇c2 (ψnu,a (R)|ψe,a (R) = |ψe,a (R)∇c2 ψnu,a (R) + 2 ∇c |ψe,a (R) · ∇c ψnu,a (R)
.

+ ∇c2 |ψe,a (R) ψnu,a (R). (9.14)

In principle, the molecular Schrödinger equation, Eq. (9.5), can be solved by

employing Eq. (9.13) including all the derivative coupling terms as indicated above.
However, this is computationally challenging, and is not necessary typically because
the first and second terms in the righthand side of the above equation are very small
compared to the first term. The BO approximation is to ignore these small terms2
and to employ the following simplification:

∇c2 (ψnu,a (R)|ψe,a (R) ≈ |ψe,a (R)∇c2 ψnu,a (R).

. (9.15)

Within the BO approximation as noted above, using Eq. (9.11) in Eq. (9.5) now
leads to
⎧ ⎫
Nu 2 Nu 
⎨  h̄ 1  Zc Zc ′
⎬
. |ψe,a (R) − ∇c2 + + Ee,a (R) ψn,a (R)
⎩ 2Mc 2 ′
|Rc − Rc′ | ⎭
c=1 c=1 c =c

= EM |ψe,a (R)ψnu,a (R). (9.16)

The above equation implies that the electronic Schrödinger equation, Eq. (9.11),
can be solved for all values (practically only for important region) of .R, from
which the electronic energy .Ee,a (R) for each of the eigenstate can be determined.
Combination of this with the nuclear potential energy leads to the following full BO

2 Incase two or more adiabatic electronic states are degenerate or close to each other for certain
value of .R, these terms cannot be ignored and result in nonadiabatic effects. Nonadiabatic effects
can be significant for bond forming or breaking situations, excited electronic states, and processes
involving proton or electron transfer.
9.2 Molecular Orbitals and Electronic Configurations for Diatomic Molecules 279

potential energy for the nuclear degrees of freedom:

N N
1u  u
Zc Zc′
VBO,a (R) =
. + Ee,a (R). (9.17)
2 ′
|Rc − Rc′ |
c=1 c =c

Taking inner product of Eq. (9.16) with .ψe (R)| and employing the above definition
of .VBO,a (R), we thus obtain the following nuclear Schrödinger equation (in
Schrödinger’s formulation):

Nu

 h̄2 2
. − ∇ + VBO,a (R) ψnu,a (R) = EM ψnu,a (R). (9.18)
2Mc c
c=1

Alternatively, introducing the nuclear state .|ψnu,a  such that .ψnu,a (R) = R|ψnu,a ,
the above Schrödinger equation can be expressed in the Dirac notation as follows:

Nu

 P̂2c
. + VBO,a (R̂) |ψnu,a  = EM |ψnu,a . (9.19)
2Mc
c=1

Solution of the above equation completes the calculation of the molecular eigenstate
within the BO approximation and the total molecular energy .EM .

9.2 Molecular Orbitals and Electronic Configurations

for Diatomic Molecules

This section provides more detailed consideration of diatomic molecules, as the

simplest examples of polyatomic molecules. The electronic Hamiltonian and
Schrödinger equation based on the BO approximation are presented in detail. A
simple molecular orbital theory for approximating these electronic states [1] is
summarized, which offers clear understanding of important qualitative features
observed for these systems.

9.2.1 Example of H2

For .H2 molecule, the nuclear and electronic components of the molecular Hamilto-
nian constituting Eq. (9.1) are

P̂2A P̂2 1
. Ĥnu = + B + ,. (9.20)
2M 2M |R̂A − R̂B |
280 9 Polyatomic Molecules and Molecular Spectroscopy

p̂21 p̂2 1 1 1 1
Ĥen = + 2− − − −
2 2 |r̂1 − R̂A | |r̂1 − R̂B | |r̂2 − R̂A | |r̂2 − R̂B |
1
+ . (9.21)
|r̂1 − r̂2 |

Thus, the electronic Hamiltonian for fixed values of .RA and .RB , defined by
Eq. (9.3), for the present case is

p̂21 p̂2 1 1
Ĥen (RA , RB ) =
. + 2− −
2 2 |r̂1 − RA | |r̂1 − RB |
1 1 1
− − +
|r̂2 − RA | |r̂2 − RB | |r̂1 − r̂2 |
p̂21 p̂2 1 1 1 1 1
= + 2− − − − + . (9.22)
2 2 r̂1A r̂1B r̂2A r̂2B r̂12

Although the above Hamiltonian corresponds to the simplest electronic Hamiltonian

of all polyatomic molecules, it is not amenable for an analytic solution. Thus,
approximations are needed. As in the case of many-electron atoms, where the
solution for hydrogen atom serves as a good basis for constructing states for many
electrons, one can use the electronic states for .H+ 2 , for which exact analytic or
reasonably good approximations exists, as a useful basis for constructing states for
two or many electrons. The simplest approximation of this nature is called the linear
combination of atomic orbitals as molecular orbital (LCAO-MO).
Within the LCAO-MO approximation, MOs are approximated as linear combi-
nations of atomic orbitals. For example, for the ground electronic state of .H+
2 , where
there is a single electron, one can assume the following form for the MO:

. | = CA |1SA  + CB |1SB , (9.23)

where .1SA and .1SA respectively represent the 1s orbital for a hydrogen atom
centered at .RA and .RB . Employing the variational principle, as described in Chap. 7,
we can obtain the following matrix equation for .CA and .CB :
    
HAA − E HAB − ES CA 0
. = , (9.24)
HBA − ES HBB − E CB 0

where

. HAA (R) = 1SA |Ĥen (R)|1SA , . (9.25)

HBB (R) = 1SB |Ĥen (R)|1SB , . (9.26)
HAB (R) = 1SA |Ĥen (R)|1SB  = HBA (R), . (9.27)
S(R) = 1SA |1SB  = 1SB |1SA . (9.28)
9.2 Molecular Orbitals and Electronic Configurations for Diatomic Molecules 281

Employing the expression for the 1s orbital explicitly in the above expressions, it is
possible to confirm the following relationship:

1
HAA (R) = HBB (R) = − + J (R),
. (9.29)
2
where
 
−2R 1 1
J (R) = e
. 1+ − . (9.30)
R R

Similarly,

R2
 
−R
S(R) = e
. 1+R+ ,. (9.31)
3
S(R)
HAB (R) = − + K(R), (9.32)
2

where3 .K(R) = −e−R (1 + R).

The matrix equation, Eq. (9.24), implies that
 
 HAA − E HAB − ES 
.
 HBA − ES HBB − E  = 0, (9.33)
 

which has the following solutions:

HAA ∓ HAB
E=
. . (9.34)
1∓S

Note that
1 1
HAA − HBB = − + J (R) + S(R) − K(R), .
. (9.35)
2 2
1 1
HAA + HAB = − + J (R) − S(R) + K(R). (9.36)
2 2

3 Note that .K(R) is not the exchange interaction term introduced in Chap. 8 although similar
notation is used.
282 9 Polyatomic Molecules and Molecular Spectroscopy

Therefore,

− 12 (1 − S(R)) + J (R) − K(R)

E− (R) =
.
1 − S(R)
1 J (R) K(R)
=− + − . (9.37)
2 1 − S(R) 1 − S(R)

Similarly,

1 J (R) K(R)
E+ (R) = − +
. + . (9.38)
2 1 + S(R) 1 + S(R)

In the above expressions, .J (R) represents the Coulomb interaction between the
electron in orbital .1SA (or .1SB ) and the nucleus B (or A).
From Eq. (9.30), it is clear that .J (R) + 1/R > 0 always. On the other hand,
.K(R) represents stabilization energy due to sharing of an electron between two

orbitals and is always negative. Therefore, .E+ (R) < E− (R) for all values of R.
For .E+ (R), Eq. (9.24) reduces to
    
SHAA − HAB HAB − SHAA CA 0
. = . (9.39)
HBA − SHAA SHAA − HAB CB 0

The solution for .CA and .CB satisfying the above equation is that .CA = CB .
Therefore, the corresponding state is .|ψ+  = CA (|1SA  + |1SB ). The remaining
coefficient .CA can be determined by the normalization condition, .ψ+ |ψ+  =
CA2 (2 + 2S) = 1. Thus, the state is given by

1
|ψ+  = √
. (|1SA  + |1SB ) . (9.40)
2(1 + S)

For .E− (R), Eq. (9.24) reduces to

    
−SHAA + HAB HAB − SHAA CA 0
. = . (9.41)
HBA − SHAA −SHAA + HAB CB 0

The solution for .CA and .CB satisfying the above equation is that .CA + CB = 0.
Therefore, the corresponding state is .|ψ−  = CA (|1SA  − |1SB ). The remaining
coefficient .CA can also be determined by the normalization condition, .ψ− |ψ−  =
CA2 (2 − 2S) = 1. Thus, the state is given by

1
|ψ−  = √
. (|1SA  − |1SB ) . (9.42)
2(1 − S)
9.2 Molecular Orbitals and Electronic Configurations for Diatomic Molecules 283

Fig. 9.1 Two LCAO-MO states formed by 1S orbitals in .H+

2

Fig. 9.2 Energy diagrams of two LCAO-MO states formed by 1S orbitals

Figure 9.1 illustrates the two orbitals .|ψ+  and .|ψ− . Because both of these
states are cylindrically symmetric around the internuclear axis, they are called .σ
orbitals. Note that .|ψ+  does not have any nodal plane in between A and B and
is thus symmetric with respect to inversion, whereas .|ψ−  has a nodal plane and
is antisymmetric. Therefore, the former is denoted as4 .|σg 1s while the latter as5
.|σu 1s.

Figure 9.2 shows the energy diagram of .|σg 1s and .|σu 1s states. As is alluded
from the figure, the lowering of .Eg = E+ (R) from that of the atomic state is less
than the relative increase of .Eu = E− (R). This can in fact be proved directly.

4 Subscript g represents gerade (even in German).

5 Subscript u represents ungerade (odd in German).
284 9 Polyatomic Molecules and Molecular Spectroscopy

Equations (9.37) and (9.38) can be re-expressed as

1 J (R) − K(R) 1
. Eu = − + + ,. (9.43)
2 1 − S(R) R
1 J (R) + K(R) 1
Eg = − + + , (9.44)
2 1 + S(R) R

where note that .E1s = −1/2. Therefore,

 
1 1
.(Eu − E1s ) − (E1s − Eg ) = − − K(R)
1 − S(R) 1 + S(R)
 
1 1 2
+ + J (R) +
1 − S(R) 1 + S(R) R
 
1
≥ 2 J (R) + ≥ 0. (9.45)
R

Thus, the amount of destabilization of energy in anti-bonding orbital is more than

the amount of stabilization in bonding orbital.

9.2.2 Molecular Orbitals and Electronic Configurations

of Diatomic Molecules

LCAO-MOs can be constructed for all other diatomic molecules in a manner similar
to those for .H2 described in the previous subsection. Within the independent electron
model, one can then construct the electronic configuration by filling electrons in
available MOs, up to two electrons per each orbital so as not to violate the Pauli
exclusion principle. An important quantity indicating the stability of each electronic
configuration is its bond order defined below.

Bond Order (B.O.)

Given the electronic configuration, the bond order (B.O.) can be calculated as
defined below.
1
Bond Order (B.O.) =
. {(# of electrons in bonding orbitals)
2
−(# of electrons in anti-bonding orbitals)}
9.2 Molecular Orbitals and Electronic Configurations for Diatomic Molecules 285

Let us consider the examples of diatomic molecules in the first row. For the case
of .H2 , the electronic configuration is .(σg 1s)2 and the B.O. is 1. On the other hand,
for .He2 , the electronic configuration is .(σg 1s)2 (σu 1s)2 and the B.O. is 0. Because
of Eq. (9.45), the total energy of electrons for .He2 is higher than the sum of .He, even
within the approximation of independent electrons. Therefore, .He2 is energetically
unfavorable and is not formed.
For the specification of electronic configurations of diatomic molecules with
more electrons than .He2 , it is necessary to consider additional MOs formed by other
types of atomic orbitals. For 2s atomic orbitals, the MOs that can be formed are
similar to those from 1s orbitals, and are denoted as .σg 2s and .σu 2s.
The MOs formed by the three 2p orbitals can be classified into two different
groups. The .2pz orbitals form .σ orbitals as follows:

. |σg 2pz  ∝ (|2pz,A  − |2pz,B ), . (9.46)

|σu 2pz  ∝ (|2pz,A  + |2pz,B ). (9.47)

Note that the bonding orbital in this case is formed by the difference (instead of
the sum) of the two .2pz atomic orbitals. This way, the two has the same phase in
the region between the two nuclei. Bonding and anti-bonding orbitals that can be
formed by the two atomic orbitals are illustrated in Fig. 9.3.
On the other hand, .px and .py orbitals form .π MOs as follows:

. |πu 2py  ∝ (|2py,A  + |2py,B ), . (9.48)

|πu 2px  ∝ (|2px,A  + |2px,B ), . (9.49)
|πg 2py  ∝ (|2py,A  − |2py,B ), . (9.50)
|πg 2px  ∝ (|2px,A  − |2px,B ). (9.51)

These .π MOs have one nodal planes that contain the internuclear axis as illustrated
in Fig. 9.4. The bonding orbital has u symmetry and the anti-bonding has g
symmetry. The axis perpendicular to the internuclear axis can be either x or y axis.
Thus, there are two .π MOs for each type of bonding and anti-bonding.
In general, .σ interactions are stronger than .π interactions because electrons are
more localized along the internuclear axis for the former. Therefore, it is expected

Fig. 9.3 Bonding molecular orbital, .σg 2pz (left), and anti-bonding molecular orbital, .σu 2pz
(right), formed by two .2pz atomic orbitals
286 9 Polyatomic Molecules and Molecular Spectroscopy

Fig. 9.4 Bonding .π molecular orbital, .2πu 2p (left), and anti-bonding .π molecular orbital, .2πg 2p
(right), formed by .2px or .2py atomic orbitals

Fig. 9.5 MO diagram corresponding to .Li2 through .N2

that .σg 2pz bonding orbital has lower energy than .πu 2px,y and that .σu 2pz has higher
energy than .πg 2px,y . However, for diatomic molecules ranging from .Li2 through
.N2 , the relative order of bonding orbitals is switched. Thus, .σg 2pz has higher energy

than .πu 2px,y for these molecules. This is an outcome of repulsive interactions from
electrons already occupying the .σu 2s orbital, which destabilizes the .σg 2pz orbital
because they have significant overlap along the internuclear axis. Figure 9.5 shows
the MO energy diagram reflecting this effect. The ground electronic configurations
that are obtained by filling up electrons according to the molecular building-up
principle are listed in Table 9.1.
9.2 Molecular Orbitals and Electronic Configurations for Diatomic Molecules 287

Table 9.1 Molecular configurations, bond order, and term symbols of ground electronic states for
stable diatomic molecules
# of elec. Configuration B. O. Ground st.
.H2
+
1 .(σg 1s)
1 .1/2
2+
. g
.H2 2 .(σg 1s)
2 1 .
1+
g
+
.He2 3 2
.(σg 1s) (σu 1s)
1 .1/2
2 +
. u

.He2 4 2
.(σg 1s) (σu 1s)
2 0
.Li2 6 .KK(σg 2s)
2 1 1+
. g
.Be2 8 2
.KK(σg 2s) (σu 2s)
2 0
.B2 10 2 2
.KK(σg 2s) (σu 2s) (πu 2px,y )
2 1 .
3−
g
.C2 12 2 2
.KK(σg 2s) (σu 2s) (πu 2px,y )
4 2 1 +
. g

.N2 14 2 2 4
.KK(σg 2s) (σu 2s) (πu 2px,y ) (σg 2pz )
2 3 1 +
. g

.O2 16 2 2 4 2
.KK(σg 2s) (σu 2s) (πu 2px,y ) (σg 2pz ) (πg 2px,y )
2 2 3 −
. g

.F2 18 2 2 4 2
.KK(σg 2s) (σu 2s) (πu 2px,y ) (σg 2pz ) (πg 2px,y )
4 1 1 +
. g

Fig. 9.6 MO diagram for .O2

and .F2 . The electrons filled in
represent the configuration of
.O2

On the other hand, for .O2 and .F2 , the original order is restored and the .σg 2pz
orbital has lower energy than .πu 2p orbitals. This is because large nuclear charges
of these molecules cause large enough separation of energy levels between the
.σu 2s and .σg 2pz orbitals compared to their electron-electron repulsion effect. The

resulting MO diagrams, with all electrons for .O2 filled, are shown in Fig. 9.6. For
the case of .F2 , the additional two electrons fill up the .πg 2p orbitals. The ground
electronic configurations for these are also listed in Table 9.1.
Having defined the MOs, as depicted in Figs. 9.5 and 9.6, and identified the
electronic configurations of diatomic molecules as summarized in Table 9.1, it is
important to assess the validity and accuracy of the approximations involved. To
this end, photoelectron spectroscopy and Koopmans’ theorem play important roles,
which are summarized below.
288 9 Polyatomic Molecules and Molecular Spectroscopy

Photoelectron Spectroscopy
Photoelectron spectroscopy is the molecular version of the photoelectric effect
where a photon of large enough energy is absorbed by a molecule and an
electron is released from the molecule. Due to the energy conservation, the
following relationship holds for a photoelectron spectroscopy:

1
h̄νph =
. me v 2 + Ii , (9.52)
2
where .νph is the frequency of the photon being absorbed by the molecule,
v is the speed of the ejected electron, and .Ii is the ionization energy of the
molecule.

Koopmans’ Theorem
According to Koopmans, the ionization energy can be approximated well by
the negative of the orbital energy as follows:

Ii ≈ −Ei ,
. (9.53)

where .Ei is the energy of the orbital the electron is ejected from.

Combination of Eqs. (9.52) and (9.53) provide direct means to experimentally

determine the energies of MOs. Indeed, these experimental results have confirmed
the validity of the MOs and configurations for diatomic molecules described above.
Ionization energies of molecules are typically several electron volts even for valence
electrons, for which photon with wavelength less than .200 nm is typically needed.
So far, only homonuclear diatomic molecules have been considered. For het-
eroatomic diatomic molecules, if atomic numbers are not significantly different,
similar MO levels can be used. However, even in this case, the notation g or u cannot
be used any more because the molecule no longer has the inversion symmetry. The
magnitudes of coefficients contributing to LCAO-MOs are different as well. Other
than that, the order of MO energy levels and the building-up principle remain intact.
For example, .CO and .CN− have the following ground electronic configuration:

KK(σ 2s)2 (σ ∗ 2s)2 (π 2p)4 (σ ∗ 2pz )2 ,

.

for which the B.O. is 1.

9.2 Molecular Orbitals and Electronic Configurations for Diatomic Molecules 289

For diatomic molecules formed by those with significantly different atomic

numbers, different types of MOs have to be used because only atomic orbitals
with similar energies and proper symmetries form significant molecular orbitals.
For example, .HF has the following electronic configuration:

.(1sF )2 (2sF )2 (1σ )2 (2pxF )2 (2pyF )2

While the LCAO-MO approximation still provides simple understanding that is

useful, in reality, actual MOs being formed become complicated once interactions
among all the electrons and perturbation by other MOs are taken into consideration.
Still, the notations for .σ , .π , .· · · can be used since the molecules remain cylindrically
symmetric. Thus, more generally, instead of specifying the atomic orbitals consti-
tuting MOs within the simplest LCAO-MO approximation, the MOs are labelled by
numbers of increasing order. For example, .F2 has the following configuration:

.(1σg )2 (1σu∗ )2 (2σg )2 (2σu∗ )2 (3σg )2 (1πu )4 (1πg∗ )4

9.2.3 Molecular Electronic States of Diatomic Molecules

All the diatomic molecules have cylindrical symmetry, for which .L̂z (with the z
axis defined as the direction of internuclear axis) commutes with the molecular
Hamiltonian. Therefore, the eigenvalue of .L̂z serves as a good quantum number,
and can be used to label the electronic states of all diatomic molecules. For this,
., the absolute value of the eigenvalue of .L̂z divided by .h̄, is used. In other words,

states with . have .±h̄ as eigenvalues of .L̂z . For . = 0, this is a single state.
For . = 0, this represents doubly degenerate states. These different states, along
with other important properties, can be represented compactly by molecular term
symbols that are introduced below.

Molecular Term Symbols

The following molecular term symbol is used to express the electronic states
of diatomic molecules:
2S+1
. M±
g(u) ,

where .M is one of capital Greek letters, ., , , Ŵ for . = 0, 1, 2, 3,

respectively. In the left superscript, S is the total spin quantum number of
electrons. In the right subscript, g is used for a state with even inversion
symmetry and u is used for a state with odd inversion symmetry. In the right

(continued)
290 9 Polyatomic Molecules and Molecular Spectroscopy

superscript, .+ means even reflection symmetry and .− means odd reflection

symmetry with respect to a plane containing the internuclear axis.

For the case of singly occupied MOs, it is easy to find the corresponding
term symbols because the state has the same symmetry as the MO and the spin
multiplicity is always 2. Below are examples:

. (σg )1 → 2 g+

(σu )1 → 2 u+
(πu )1 → 2
u
1 2
(πg ) → g

For the case of fully occupied MOs, the spin multiplicity is 1 and the electronic
state is totally symmetric. Thus, all of .(σg )2 , (.σu )2 , .(πg )4 , and .(πu )4 have the same
term, .1 g+ . For the case of one less than fully filled MOs, the term is the same as
singly occupied MOs. This is because, the former can be viewed as that of a single
hole (absence of electron) which has to obey the same property as the electron as
far as the molecular term symbols are concerned.
For the case where two electrons occupy the same orbitals, for example, as in
2 2
.(πg ) or .(πu ) , care should be taken in identifying all possible term symbols that

are consistent with the Pauli principle. A simple way to figure out all possible terms
is to consider the sum of the pair of single electron angular momentum and spin
values, .lz and .sz . For example, for an electron occupying the .π molecular orbital,
the following four pairs exist.
   
1 1
. a:1, b : −1,
2 2
   
1 1
a ′ : 1, − b′ : −1, −
2 2

Then, all possible values for the total .Lz and .Sz of the two electrons occupying the .π
orbitals can be found by summing all distinctive pairs of .(lz , sz ). Considering only
the positive values of these, we obtain the following four cases of .(, S):

a + a ′ : (2, 0)
. a + b : (0, 1)
a + b′ : (0, 0)
a ′ + b : (0, 0)
9.2 Molecular Orbitals and Electronic Configurations for Diatomic Molecules 291

In the above list, the case with . = 2 and .S = 0 corresponds to a term with
1
. g . Here, the subscript g is used because two electrons occupying the same orbital
always have even symmetry with respect to inversion regardless of the symmetry
of individual MOs. In the remaining three pairs, .(0, 1) and .(0, 0) represent the case
with . = 0 and .S = 1. These apparently represent triplet states. For this to be
true, the MO should be an antisymmetric combination of the product of two single
electron .π orbitals, which is expected to have odd reflection symmetry with respect
to a plane containing an internuclear axis. Therefore, these correspond to .3 g− .
The remaining pair corresponds to the case of . = 0 and .S = 0. Since the spin
state is antisymmetric, the orbital in this case has to be a symmetric combination
of two atomic .π orbitals, for which the reflection symmetry with respect to a plan
containing the internuclear axis is even. Therefore, this corresponds to .1 g+ .
Compiling all the results obtained so far, we find that both .(πg )2 and .(πu )2 have
the following three different terms:
3 1 1
. g− , g, g+ .

Note that the terms above are listed in the order of increasing energy according
to Hund’s rules.6 In other words, the states with higher spin multiplicity have lower
energies. Among the states with the same spin multiplicity, the states with the higher
value of . have lower energies.
For two electrons in two different orbitals, all possible combinations of . and
S can be used without restriction. For example, for an excited configuration of .H2 ,
1 1
.(1σg ) (1σu ) , the following two terms exist:

3
. u+ , 1 u+ ,

where the two terms are listed in the order of increasing order of energy according to
Hund’s rules. For an excited state configuration of .N2 , .(1πu )3 (3σg )2 (1πg∗ )1 , there
are four possible terms, and they can be written in the order of increasing energy
according to Hund’s rules as follows:
3 3 1 1
. u, u± , u, u± .

Transitions between the electronic states of diatomic molecules due to absorption

or emission of a single photon occur according to well known selection rules. More
generally, one can identify a set of selection rules for electronic states of linear
molecules as described below.

6 For molecular electronic states, Hund’s rules are less reliable than atomic electronic states.
292 9 Polyatomic Molecules and Molecular Spectroscopy

Selection Rules for Electronic Transitions of Linear Molecules

• . = 0, ±1: This is because the total orbital angular momentum number
of all electrons L can change only by .±1 or remain the same, for which .
can change only by .±1 at most or remain the same.
• . S = 0: This is because the spectroscopy involves interaction of the
electric filed part of the radiation, which does not interaction with spins
and thus leaves the total spin of electrons intact.
• There is no transition between . + and . − states: This is because the
transition dipole moment interacting with the radiation is always along
the direction of the internuclear axis due to the cylindrical symmetry. This
means that the transition dipole has positive reflection symmetry and thus
cannot couple any two states with different reflection symmetries.
• Only transitions between u and g symmetries are allowed: In other
words, transition between states with the same inversion symmetries is not
allowed. This is because the transition dipole has u symmetry and thus can
couple only states with different inversion symmetries.

9.3 Conjugated Hydrocarbons and Hückel Approximation

The simple LCAO-MO approximation can be useful for qualitative understanding

of .π -type electronic states formed by .pz orbitals in hydrocarbons with conjugated
double bonds. In fact, one can make further simplifications known as Hückel
approximation as described below.

Hückel Approximation
For .π -type molecular orbitals formed by .pz orbitals (p orbitals perpendicular
to .C − C bonding), the following three approximations can be made.
• All the overlap integrals between different .pz orbitals are zero.
• All the diagonal elements of the electronic Hamiltonian with respect to .pz
orbitals are assumed to be the same as follows:

Hii = 2pzi |Ĥen |2pzi  = α.

. (9.54)

• All the off-diagonal elements of the electronic Hamiltonian between

different .pz orbitals are assumed to be zero except for those due to the

(continued)
9.3 Conjugated Hydrocarbons and Hückel Approximation 293

nearest neighbors, which are also assumed to be the same as follows:

.Hii+1 = 2pzi |Ĥen |2pzi+1  = 2pzi+1 |Ĥen |2pzi  = β. (9.55)

9.3.1 Ethylene

As the simplest example, let us consider ethylene, for which the .π -type LCAO-MO
formed by the two .pz orbitals can be expressed as

.|ψπ  = C1 |2pz1  + C2 |2pz2 . (9.56)

Application of the variational principle described in Chap. 7 for the above state leads
to the following matrix equation:
    
H11 − E H12 − ES12 . C1 0
. = , (9.57)
H12 − ES12 H22 − E C2 0

where .H11 = pz1 |Ĥen |pz1 , .H22 = pz2 |Ĥen |pz2 , .Ĥ12 = pz1 |Ĥen |pz2 , and
S12 = pz1 |pz2 . Based on the Hückel approximation, .H11 = H22 = α and .H12 =
.

β, Thus, the above matrix equation implies that

 
α − E β  2 2
.
 β α − E  = (α − E) − β = (α − E − β)(α − E + β) = 0. (9.58)
 

Therefore, there are two possible solutions, .E = α ± β.

For the case of .E = α + β, solving Eq. (9.57) results in

. − βC1 + βC2 = 0. (9.59)

√
This, along with the normalization condition, means that .C1 = C2 = 1/ 2. For the
case of .E = α − β, solving Eq. (9.57) results in

βC1 + βC2 = 0.
. (9.60)
√
This, along with the normalization condition, means that .C1 = −C2 = 1/ 2. These
results for ethylene are summarized below.
294 9 Polyatomic Molecules and Molecular Spectroscopy

.π Molecular Orbitals and Energies for Ethylene

The .π bonding and .π ∗ anti-bonding states for ethylene and their energies are
as follows:
1
. |ψπ  = √ (|2pz1  + |2pz2 ) , Eπ = α + β. (9.61)
2
1
|ψπ ∗  = √ (|2pz1  − |2pz2 ) , Eπ ∗ = α − β (9.62)
2

9.3.2 Butadiene

As the next example, let us consider butadiene. For this case, the .π −MO constructed
from the linear combination of four .pz orbitals is expressed as

|ψπ  = C1 |2pz1  + C2 |2pz2  + C3 |2pz3  + C4 |2pz4 .

. (9.63)

Then, application of the variational principle and Hückel approximation leads to the
following equation for E and the .Ck :
⎛ ⎞⎛ ⎞ ⎛ ⎞
α−E β 0 0 C1 0
⎜ β α−E β 0 ⎟ ⎜ C2 ⎟ ⎜ 0 ⎟
.⎜
⎝ 0
⎟⎜ ⎟ = ⎜ ⎟. (9.64)
β α − E β ⎠ ⎝ C3 ⎠ ⎝ 0 ⎠
0 0 β α−E C4 0

Let us introduce
α−E
x=
. . (9.65)
β

Then, the above matrix equation can be equivalently expressed as

⎛ ⎞⎛ ⎞ ⎛ ⎞
x 1 0 0 C1 0
⎜1 x 1 0 ⎟ ⎜ C2 ⎟ ⎜ 0 ⎟
⎟ ⎜ ⎟ ⎜
.⎜
⎝0 = ⎟. (9.66)
1 x 1 ⎠ ⎝ C3 ⎠ ⎝ 0 ⎠
0 0 1 x C4 0
9.3 Conjugated Hydrocarbons and Hückel Approximation 295

This implies that

 
x
 1 0 0    
x 1 0 1 1 0

1 x 1 0     
.
0 = x  1 x 1  −  0 x 1  = x(x 3 −2x)−(x 2 −1) = x 4−3x 2+1 = 0. (9.67)
 1 x 1  0 1 x  0 1 x 
0 0 1 x

Solving the above algebraic equation, we obtain

√ √ 2
2 3± 5 5±1
.x = = , (9.68)
2 2

which correspond to the following solutions for x:

√ √ √ √
5+1 5−1 5−1 5+1
x=
. , ,− ,− . (9.69)
2 2 2 2
Since .x = (α − E)/β, the above solutions correspond to the following four values
for the energies:
√ √ √ √
5+1 5−1 5−1 5+1
E=α−
. β, α − β, α + β, α + β. (9.70)
2 2 2 2
Since .β is negative, the energies given above are in the decreasing order.
Employing the corresponding value of x for each case in Eq. (9.66), we can
obtain the corresponding MOs. For this, it is useful to rearrange the linear equations
for .Ck ’s. First, note that Eq. (9.64) is equivalent to the following equations:

. C1 x + C2 = 0, . (9.71)
C1 + xC2 + C3 = 0, . (9.72)
C2 + xC3 + C4 = 0, . (9.73)
C3 + C4 x = 0. (9.74)

Adding Eqs. (9.72) and (9.73), we obtain

C1 + (1 + x)C2 + (1 + x)C3 + C4 = 0.
. (9.75)

Subtracting Eq. (9.73) from Eq. (9.72), we also obtain

C1 + (x − 1)C2 + (1 − x)C3 − C4 = 0.
. (9.76)
296 9 Polyatomic Molecules and Molecular Spectroscopy

From Eqs. (9.71) and (9.74), we find that .C2 = −xC1 and .C3 = −xC4 . Substituting
these for .C2 and .C3 in Eqs. (9.75) and (9.76), we obtain the following two different
equations involving .C1 and .C4 .

. C1 (1 − x − x 2 ) + C4 (1 − x − x 2 ) = 0, . (9.77)
2 2
C1 (1 + x − x ) − C4 (1 + x − x ) = 0. (9.78)
√ √
For the cases of .x = ( 5 + 1)/2 and .x = −( 5 − 1)/2, .1 + x − x 2 = 0 and
Eq. (9.78) is already satisfied. Therefore, using
√ Eq. (9.77), we find that √
.C1 = −C4 .

On the other hand, for the cases of .x = ( 5 − 1)/2 and .x = −( 5 + 1)/2,

2
.1 − x − x = 0 and Eq. (9.77) is already satisfied. Therefore, using Eq. (9.78), we

find that .C1 = C4 . As a result, we obtain the following four MOs corresponding to
the four values of the energy for butadiene.

π -MO Energies and States for Butadiene

.

π -MO energies and states formed by linear combinations of .pz orbitals,

.

indexed in an increasing order of energy, are provided below.

Eπ |ψπ −MO 
√ √ √
5+1 5+1 5+1
1α+ 2 β C1 |2pz1  + 2 |2p z2  + 2 |2pz3  + |2pz4 
√ √ √
5−1 5−1 5−1
. 2α+ 2 β C1 |2pz1  + 2 |2pz2  − 2 |2pz3  − |2pz4 
√ √ √
5−1 5−1 5−1
3α− 2 β C1 |2pz1  − 2 |2pz2  − 2 |2pz3  + |2pz4 
√ √ √
5+1 5+1 5+1
4α− 2 β C1 |2pz1  − 2 |2pz2  + 2 |2pz3  − |2pz4 

The values of .C1 in above expressions can

be determined by the normalization
√
condition. For states 1 and 4, .C1 = 1/ 5 + 5. For states 2 and 3, .C1 =
 √
1/ 5 − 5.

9.3.3 π Orbital and Delocalization Energies

Having obtained the .π -MOs, let us now consider the .π orbital energies of electrons
within the independent electron model. The ground electronic state is obtained by
filling in two electrons from the lowest possible .π -MOs.
9.4 Molecular Symmetry and Group Theory 297

For the case of ethylene, the two valance electrons from .pz atomic orbitals can fill
the lowest energy .π -MO with energy .α + β. Thus, the .π orbital energy for ethylene
is

Eπ (ethylene) = 2(α + β) = 2α + 2β.

. (9.79)

On the other hand, for butadiene, there are four valance electrons coming from .pz
atomic orbitals. These can fill in the two lowest energy .π -MOs, for which the energy
is equal to
 √   √ 
5+1 5−1
Eπ (butadiene) = 2 α +
. β +2 α+ β
2 2
√
= 4α + 2 5β. (9.80)

Note that the .π orbital energy for butadiene is different from twice the .π electron
energies of ethylene. The difference between the two accounts for the additional
stabilization due to conjugation (or delocalization) of neighboring .π -MOs and
accounts for the stability of .π -conjugated molecules as summarized below.

π -Bonding Delocalization (or Conjugation) Energy

.

The .π bonding delocalization energy is the difference of the total .π -orbital

energy from the sum of those for isoelectronic ethylenes. For the case of
butadiene,
√ √
Eπ −del (butadiene) = 4α + 2 5β − 4(α + β) = 2( 5 − 2)β.
. (9.81)

9.4 Molecular Symmetry and Group Theory

9.4.1 Symmetry and Symmetry Operation

Symmetry plays an important role in quantum mechanics. It allows determination

of the nature of eigenstates of the Hamiltonian and classifying them without going
through detailed calculations. For molecular systems, consideration of symmetry
in particular enables constructing MOs in an efficient manner and identifying the
nature of vibrational modes, both with the help of the group theory [9–13]. We
begin this section with more exact definition of symmetry and related terms as listed
below.
298 9 Polyatomic Molecules and Molecular Spectroscopy

Symmetry and Related Terms

• Symmetry: An object is called to have symmetry if it remains equivalent
(appears to look the same) after certain operation.
• Symmetry operation: An operation that is intended to test whether an
object has symmetry or not is called symmetry operation.
• Symmetry element: A geometrical object with respect to which a symme-
try operation is performed.
The two definitions of symmetry and symmetry operation seem somewhat
circular, but it is important to note that the former is the property of a specific
object whereas the latter can be applied to any object. Thus, a symmetry
operation can be defined independent of objects it is being applied to. It only
requires existence of a symmetry element with respect which the operation is
performed.

Symmetry operation is also a physically well-defined operation, which can also

be viewed as a quantum mechanical operator with proper definition of its action on
either physical observables or states. The fact that a molecule has certain symmetry
is equivalent to stating that the Hamiltonian for the molecule commutes with the
corresponding quantum mechanical symmetry operator. This important fact will be
detailed further later. In many cases, the same symbol is used for symmetry operator
and symmetry element, which often becomes the source of confusion between the
two. Therefore, operator notation is used here for the symmetry operation.
Out of all possible symmetry operations, we will focus here only on point
symmetry operations, which leave at least one point of the object being operated
on unchanged. Point symmetry operations are the only kinds of operations needed
for the specification of intrinsic properties of molecules. Let us first review the basic
types of point symmetry operations as listed below.

Point Symmetry Operations

There are five types of symmetry operations [9] that form the basis for all
point symmetry operations as listed below.
• Identity operation, .Ê, is doing nothing, and does not have any symmetry
element associated with it.
• Inversion operation, .î, is an inversion with respect to an inversion center.
If the inversion center is origin, this operator transforms .(x, y, z) to
.(−x, −y, −z). Application of this operation twice becomes the identity

operation as follows: .î 2 = î · î = Ê.

(continued)
9.4 Molecular Symmetry and Group Theory 299

• Proper rotation, .Ĉn , is an operation that rotates around an axis (denoted as

.Cn axis) by .2π/n and is called n-fold proper rotation operation. Note that

applying this operator n times leads to the identity operation as follows:

n
. Ĉn = Ĉn · Ĉn · · · · Ĉn = Ê
  
n times

• Reflection operator, .σ̂ , reflects a point or body with respect to a plane

of reflection. For example, if the plane of reflection is the xy-plane,
this operation transforms a point .(x, y, z) to .(x, y, −z). If the plane
of reflection is yz-plane, the operation transforms a point .(x, y, z) to
.(−x, y, z). When both .Ĉn ’s and .σ̂ ’s coexist in a group of symmetry

operations, the reflection plane that contains .Cn axis with the highest value
of n is denoted as .σv . On the other hand, the plane perpendicular to .Cn is
denoted as .σh .
• Improper rotation, .Ŝn , is a combination of a proper rotation and a
reflection with respect to .σh , the plane perpendicular to the proper rotation
axis .Cn . Thus, .Ŝn = σ̂h Ĉn = Ĉn σ̂h . Note that the two operators .σ̂h and .Ĉn
commute with each other. The symmetry element of the improper rotation
is denoted as .Sn , which is in fact the same as .Cn . If .Sn is equal to the
z-axis, this operation transforms a point .(x, y, z) to .(x ′ , y ′ , −z), where
′ ′
.x = cos(2π/n)x − sin(2π/n)y and .y = sin(2π/n)x + cos(2π/n)y.

This operation transforms every point except the origin where .Cn axis
and .σh intersects. There are a few important properties of .Ŝn . Application
of improper rotation twice is equal to that of proper rotation as follows:
2 2 2
.Ŝn · Ŝn = σ̂ · Ĉn = Ĉn . This is because .Ĉn and .σ̂h commute with each
h
other. Therefore, if n is even, .Ŝnn = Ê. On the other hand, if n is odd,
.Ŝn = σ̂h . It is also important to note that .Ŝ2 = î. Therefore, the smallest
n

value of n for .Ŝn , which is distinctive from others, is .n = 3.

9.4.2 Group Theory

Mathematically, a group is defined by a set of group elements, which can be finite

or infinite, and a rule of mapping called group multiplication, which combines any
pair of group elements and produces another one [9, 10].
300 9 Polyatomic Molecules and Molecular Spectroscopy

Definition of Group
Let us denote group elements of a group .G as A, B, C, .· · · , and the
group multiplication as .◦. Then, the set of elements and the multiplication
constituting .G should satisfy the four properties described below.
1. If A and B are elements of .G, then .A ◦ B and .B ◦ A are also elements of
.G.

2. There is an identity element E in the group such that .A ◦ E = E ◦ A = A

for any A in .G.
3. The group multiplication .◦ is associative. In other words, .(A ◦ B) ◦ C =
A ◦ (B ◦ C) for any A, B, and C in .G. Note that the multiplication does
not have to be commutative. In other words, .A ◦ B does not the have to be
the same as .B ◦ A. In case the group multiplications are commutative for
all elements of a group, such a group is called an Abelian group.
4. For any group element A in the group .G, there is always a unique inverse
of A, which is denoted as .A−1 , such that .A ◦ A−1 = A−1 ◦ A = E.

There are various finite and infinite groups. For example, the set of integers can
be shown to be a group by defining addition .(+) as the group multiplication. A well-
known finite group is the set of integers .mod n (remainder after division), where n
is an arbitrary positive integer. The group multiplication in this case is defined as
the sum of two integers .mod n. For example, for .n = 3, the group elements are 0,
1, and 2, and the results of group multiplications of these elements are as follows:
.0 ◦ 1 = 1 ◦ 0 = 1, .0 ◦ 2 = 2 ◦ 0 = 2, .1 ◦ 2 = 2 ◦ 1 = 0. Thus, 0 is the identity

element for this group, and the inverse of 1 (2) is 2 (1). Both of these groups are
Abelian groups.

Important Terms and Concepts in Group Theory

• Order, is the total number of elements of a group, which will be denoted
as Q here, and can be either finite or infinite.
• Group multiplication table, is a .Q × Q table that specifies the results of
the group multiplication of all pairs of group elements.
• Conjugate elements: Two elements A and .A′ of a group are called
conjugate if they are related by the following similarity transformation:
′
.A = B ◦ A ◦ B
−1 , where B is any element in the group and .B −1 is its

inverse. If this relationship holds, .A′ ◦ B = B ◦ A.

• Class: The set of all conjugate elements in a group is called a class. For an
Abelian group, each element forms its own class since .A ◦ B = B ◦ A for
any A and B.

(continued)
9.4 Molecular Symmetry and Group Theory 301

• Representation, refers to any group composed of concrete mathematical

entities that are homomorphic (many-to-one mapping from represented
elements to the original group element) or isomorphic (one-to-one map-
ping) to the original abstract group that can be completely specified by the
group multiplication table.
• Matrix representation, is the representation where each group element is
represented by a square matrix and the group multiplication is the matrix
multiplication.
• Equivalent matrix representations: Two matrix representations .Ŵ ′ and .Ŵ
of a group are called equivalent if there is an invertible matrix .S relating
the two for all matrix representations of the group as follows:

.Ŵ ′ (Ai ) = S−1 Ŵ(Ai )S, for i = 1, · · · , Q , (9.82)

where .Ai ’s represent all the elements of the group with order Q.

9.4.3 Groups of Point Symmetry Operations

Certain sets of point symmetry operations can form group. In a group of point
symmetry operations, the group elements are symmetry operations and the group
multiplication of two “elements” is defined as successive application of the two
operations, with the convention that the one on the righthand side is always applied
first. These groups can be classified according to what symmetry operations generate
the entire elements of a group and are listed in the Table 9.2, employing Schönflies
notation. For example, the group .Cnv consists of .Ê, .Ĉn , and n different .σv ’s and all
others that can be generated from them such as .Ĉnk with .k = 2, · · · , n − 1.
In Table 9.2, note that there are no .S2n−1 points groups. The reason for this
is as follows [9]. If there were a group called .S2n−1 , it should be generated by the
corresponding improper rotation operation .Ŝ2n−1 . This means that such a group also
2n−1
should have .Ŝ2n−1 = σ̂h , which also implies that it should have .Ĉ2n−1 = σ̂h Ŝ2n−1 as
its group element. However, such a group has already been classified as the .C2n−1h
group. Since we do not want duplicate notations for the same group, the customary
practice is to assume that only .S2n groups exist.
The groups from .T to .Ih in Table 9.2 are called cubic point groups. Among these
point groups, .Td , .Th , .Oh , and .Ih exactly have the same symmetries of tetrahedron,
cube, octahedron, and icosahedron, respectively.
Typically, it is stated that a given molecule belongs to a certain point group if its
structure (in the ground electronic state) remains unchanged with respect to all the
symmetry operations constituting the point group. For example, it is easy to show
that .H2 O belongs to the .C2v point group and benzene .C6 H6 belongs to the point
302 9 Polyatomic Molecules and Molecular Spectroscopy

Table 9.2 Point groups and their respective generating operators. Except for the first three point
groups, where all the generators form a complete group, there are additional symmetry operators
being generated by those listed, which altogether form a complete set of symmetry operations
Point group Generating operators
.C1 .Ê (complete)
.Cs .Ê, .σ̂ (complete)

.Ci .Ê, .î (complete)

.Cn .Ê, .Ĉn
.Cnv .Ê, .Ĉn , n different .σ̂v ’s

.Cnh .Ê, .Ĉn , .σ̂h

.Dn .Ê, .Ĉn , n different .Cˆ2 ’s (perpendicular to .Ĉn )

.Dnh .Ê, .Ĉn , n different .Cˆ2 ’s (perpendicular to .Ĉn ), .σ̂h

.Dnd .Ê, .Ĉn , n different .Cˆ2 ’s (perpendicular to .Ĉn ), n .σ̂d ’s
.S2n .Ê, .Ŝ2n

.C∞v .Cnv in the limit of .n → ∞

.D∞h .Dnh in the limit of .n → ∞
.T .Ê, three perpendicular .Ĉ2 ’s, and four .C3 ’s

.Td (tetrahedron) .Ê, three perpendicular .Ĉ2 ’s, four .C3 ’s, six .σ̂d ’s
v .Th (cube) .Ê, three perpendicular .Ĉ2 ’s, four .C3 ’s, .î
.O .Ê, two sets of three perpendicular .Ĉ4 ’s, eight .C3 ’s

.Oh (octahedron) .Ê, two sets of three perpendicular .Ĉ4 ’s, eight .C3 ’s, .î
.I .Ê, four sets of three perpendicular .Ĉ5 ’s, twenty .C3 ’s
.Ih (icosahedron) .Ê, four sets of three perpendicular .Ĉ5 ’s, twenty .C3 ’s, .î

group .D6h . If this is the case, The importance of the point group of a given molecule
is that the Hamiltonian within the BO approximation, Eq. (9.18), commutes with
all the symmetry operations of the group, which also means that it is possible to
determine simultaneous eigenstates of both the Hamiltonian and all the symmetry
operators of the group. This can be used to identify symmetry adapted LCAO-MOs
and also for identifying IR and Raman activity of molecular vibrational modes.
If a molecule is polar, it belongs to only one of .Cn , .Cs , and .Cnv . This is because
a given polar molecule should have nonzero permanent dipole moment. Since the
permanent dipole should be invariant with respect to all the symmetry operations
constituting the group that the molecule belongs to, the dipole vector should be
contained in the intersection of all symmetry elements defining the group operations.
This is possible only if all symmetry operations share at least one line on which
the permanent dipole vector lies. For this to be true, there should be no inversion
center or planes perpendicular to .Cn . The three groups listed above are the only
such groups having this property.
9.4 Molecular Symmetry and Group Theory 303

9.4.4 Matrix Representation of Point Symmetry Group

Elements

For groups consisting of point symmetry operations, representation of each group

element (symmetry operation) by a square matrix is a convenient choice. Different
matrix representations are possible depending on the choice of the basis set, but
there are certain quantities that remain invariant and play important roles for
defining more universal properties. One such quantity is called character of the
matrix representation.

Character
The character of a matrix representation, which is denoted as .χ here, refers
to the trace, namely, the sum of all the diagonal matrix elements, of the matrix.
The characters of two matrix representations are always the same if they are
related by a unitary transformation.

For example, let us consider a basis consisting of three cartesian coordinates.

Then, the symmetry operations .Ê, .Ĉ2 , .σ̂v , and .σ̂v′ constituting the .C2v point group
can be represented as follows:
⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎛ ⎞
x x 100 x
Ê
→ ⎝y ⎠ = ⎝0 1 0⎠⎝y ⎠,
.⎝y ⎠ − (9.83)
z z 001 z
  
Ê

⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎛ ⎞
x −x −1 0 0 x
Ĉ2
.⎝y ⎠ −→ ⎝ −y ⎠ = ⎝ 0 −1 0 ⎠ ⎝ y ⎠ , (9.84)
z z 0 0 1 z
  
Ĉ2

⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎛ ⎞
x x 1 0 0 x
σ̂v
. ⎝ y −
⎠ → −y = 0 −1 0
⎝ ⎠ ⎝ ⎠ ⎝ y⎠, (9.85)
z z 0 0 1 z
  
σ̂v

⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎛ ⎞
x −x −1 0 0 x
σ̂v′
→ ⎝ y ⎠ = ⎝ 0 1 0⎠⎝y ⎠.
.⎝y ⎠ − (9.86)
z z 00 1 z
  
σ̂v
304 9 Polyatomic Molecules and Molecular Spectroscopy

For the matrix representations of the four symmetry operations constituting the .C2v
group, the characters are as follows: .χ (Ê) = 3, .χ (Ĉ2 ) = −1, .χ (σ̂v ) = 1, and
′
.χ (σ̂v ) = 1.

Note that all the four matrix representations for the above .C2v group are diagonal,
which means that they can be represented by the properties of each one dimensional
component. The matrices shown are examples of reducible representation and each
of x, y, and z components here are considered as irreducible representations. More
exact definitions of these are provided below.

Reducible and Irreducible Representations

• A given matrix representation of a group is called reducible if all the
matrices representing the symmetry operations of the group can be reduced
to smaller block diagonal or fully diagonal forms by the same similarity
transformation, which means that there exists an invertible matrix .Ŝ such
that for any . in the group, .Â′ = Ŝ ÂŜ −1 , where .Ŝ −1 is the inverse of .Ŝ,
becomes a matrix consisting of smaller diagonal blocks.
• A representation is called irreducible if all the matrices cannot be reduced
to smaller block diagonal or fully diagonal forms by any similarity
transformation.
• For a given point group, there is a unique set of all distinct irreducible
representations.

Each irreducible representation can be fully characterized by the set of characters

of symmetry operations constituting a group. For example, for the representation
of the .C2v group shown above, each component x, y, and z corresponds to an
irreducible representation with the following characters.

Ê Ĉ2 σ̂v σ̂v′ Irrep.

x 1 −1 1 −1 B1
.
y 1 −1 −1 1 B2
z 1 1 1 1 A1

In the above table, the right column shows the conventional notation for the
three one dimensional irreducible representations. A fairly complete description of
notations for irreducible representations is provided below.
9.4 Molecular Symmetry and Group Theory 305

Notations for Irreducible (Irrep.) Representations

The main symbols denoting the irreducible representations of finite point
groups are A, B, E, and T . Each of these symbols is described below.
• A: One dimensional and symmetric with respect to .Ĉn of the highest n.
• B: One dimensional and antisymmetric with respect to .Ĉn of the highest
n.
• E: Two dimensional.
• T : Three dimensional.
Each symbol can further classified with subscripts as described below.
• 1 (subscript): Symmetric with respect to the .Ĉ2 ’s around .C2 axes that are
perpendicular to .Cn of the highest n or .σv ’s.
• 2 (subscript): Antisymmetric with respect to the .Ĉ2 ’s around .C2 axes that
are perpendicular to .Cn of the highest n or .σv ’s.
• g (subscript): Even inversion symmetry when there is .î operator in the
group.
• u (subscript): Odd inversion symmetry when there is .î operator in the
group.
In case the group has .σ̂h operation, each Irrep. is labeled with ′
. or ′′
. as
follows.
• ′:
. Symmetric with respect to .σ̂h .
• ′′ :
. Antisymmetric with respect to .σ̂h .

Having introduced the notations for irreducible representations as detailed above,

it is now possible to appreciate the implications of a character table for a point group.
Let us consider, for example, the full character table for .C2v as follows:

C2v Ê Ĉ2 σ̂v σ̂v′

A1 1 1 1 1 z, x 2 , y 2 , z2
. A2 1 1 −1 −1 xy, Rz
B2 1 −1 1 −1 x, xz, Ry
B2 1 −1 −1 1 y, yz, Rx

Each row in the above table represents a one dimensional irreducible representation,
for which only .±1 is possible as the character. The value .1(−1) implies that the
representation is symmetric (antisymmetric) with respect to the symmetry operation.
The right column shows linear or quadratic functions of Cartesian coordinates that
have the same symmetry as the given irreducible representation. .Rx , .Ry , and .Rz
represent rotations around x, y, and z axes.
306 9 Polyatomic Molecules and Molecular Spectroscopy

In the above table for the .C2v point group, the number of irreducible represen-
tations is the same as the order of the group, .Q = 4. This is because each element
belongs to a different class and all irreducible representations are one dimensional.
This is not always the case. For example, let us consider the .C3v group, which
consists of the following symmetry operations:

Ê, Ĉ3 , Ĉ23 , σ̂v(1) , σ̂v(2) , σ̂v(3) .

.

In the above list, .Ĉ3 and .Ĉ23 are related to each other, forming a class, and so do the
(1) (2) (3)
three reflection operations .σ̂v , .σ̂v and .σ̂v , which forms another class. Symmetry
operations belonging to the same class have same values of characters for all the
irreducible representations and thus can be grouped together in the character table.
In other words, the character table for the .C3v group can be expressed as follows:

C3v Ê 2Ĉ3 3σ̂v

A1 1 1 1 z, x 2 , y 2 , z2
.
A2 1 1 −1 Rz
E 2 −1 0 2 2
(x, y), (x − y , xy), (xz, yz), (Rx , Ry )

In the above character table for .C3v point group, there is a two dimensional
irreducible representation, E. All the characters of this can be confirmed by
considering how the two dimensional x and y coordinates are transformed as
follows:
      
x Ĉ3 cos(2π/3)x − sin(2π/3)y cos(2π/3) − sin(2π/3) x
. −→ = ,
y sin(2π/3)x + cos(2π/3)y sin(2π/3) cos(2π/3) y
  
Ĉ3

  2     
x Ĉ3 cos(4π/3)x − sin(4π/3)y cos(4π/3) − sin(4π/3) x
. −→ = ,
y sin(4π/3)x + cos(4π/3)y sin(4π/3) cos(4π/3) y
  
Ĉ32

  (1)     
x σ̂v x 1 0 x
. −−→ = .
y −y 0 −1 y
  
(1)
σ̂v

Since .2 cos(2π/3) = 2 cos(4π/3) = −1, this confirms that the character for .2Ĉ3
is equal to -1. The character for .Ê is equal to 2 because this is a matrix in two
dimensional vector space. The character for .σ̂v is equal to zero as can be seen from
its matrix representation. Note that the number of classes is equal to the number
9.4 Molecular Symmetry and Group Theory 307

of irreducible representations. In addition, the order of .C3v group, .Q = 6, is also

equal to .12 + 12 + 22 , which is the sum of squares of the dimensions of all the
irreducible representations. In fact, these relationships are true for .C2v and for all
the point groups.

Relationships Between the Order, the Number of Classes, and the Num-
ber of Irreducible Representations
• The number of classes is equal to the number of irreducible representations.
• The order of a group Q is equal to the sum of squared dimensions of all
irreducible representations.

The importance of irreducible representations is that they can be regarded as an

orthogonal basis in defining any representation. In other words, given a representa-
tion in a certain basis, the following decomposition always exists uniquely:

P

Ŵ=
. ak Ŵk′ , (9.87)
k=1

where .ak is a nonnegative integer, .Ŵk′ is an irreducible representation .Ŵk embedded

in the full dimension of .Ŵ so that all other matrix elements except for the block of
.Ŵk are zero, and P is the total number of irreducible representations in the group.

Thus, both .Ŵk and .Ŵk′ have the same trace.

An important theorem concerning the characters of irreducible representations is
their orthogonality as stated below.

Orthogonality Theorem
Given two different irreducible representations .Ŵj and .Ŵk , the following
relation holds:
Q
1 
. χj [Ŝi ]∗ χk [Ŝi ] = δj k , (9.88)
Q
i=1

where .χj [Ŝi ] is the character of the irreducible representation .Ŵj for the
symmetry operation .Ŝi . This is a special case of the most general great
orthogonality theorem given by Theorem 9.6, provided in the Appendix of
this chapter, as described below.

(continued)
308 9 Polyatomic Molecules and Molecular Spectroscopy

For the present case, the group elements G’s in Eq. (9.135) of the Appendix
are symmetry operators, .Ŝj ’s. Since the trace is given by

lj

χj [Ŝi ] =
. [Ŵ (j ) (Ŝi )]mm , (9.89)
m=1

we find that

Q lj Q
lk 
  
∗
. χj [Ŝi ] χk [Ŝi ] = [Ŵ (j ) (Ŝi )]∗mm [Ŵ (k) (Ŝi )]nn
i=1 m=1 n=1 i=1
lj lk
  Q
= δj k δmn = Qδj k . (9.90)
lj
m=1 n=1

Dividing the above expression with Q, we obtain Eq. (9.88).

Given the decomposition of a (reducible) representation .Ŵ according to

Eq. (9.87), the same relationship should hold for the character of any symmetry
operation as follows:

P

χŴ [Ŝi ] =
. ak χk [Ŝi ], (9.91)
k=1

where .χŴ [Ŝi ] is the character of the reducible representation .Ŵ for the symmetry
operation .Ŝi . Multiplying the above equation with .χl [Ŝj ]∗ and summing over all the
symmetry operations of the group, .Ŝi , and dividing with Q, the order of the group,
we obtain
Q P P
1  
. χj [Ŝi ]∗ ak χk [Ŝi ] = ak δj k = aj , (9.92)
Q
i=1 k=1 k=1

where Eq. (9.88) has been used. The above expression confirms that each coefficient
ak in the expansion, Eq. (9.87), can be obtained only from the information on the
.

characters for all symmetry operations

9.4 Molecular Symmetry and Group Theory 309

9.4.5 Application for Symmetry Adapted LCAO-MO

Let us consider an example of water molecule, which belongs to the .C2v group, and
assume LCAO-MOs consisting of six atomic orbitals,

|ψMO  = CsA |SA  + CsB |SB  + csC |SC  + cpx |px  + cpy |py  + cpz |pz ,
. (9.93)

where .|SA  and .|SB  are 1s orbital states of two hydrogen atoms, respectively
denoted as A and B, .|SC  is the 2s orbital state of the central oxygen atom, .|px ,
.|py , and .|pz  are .px , .py , .pz orbital states of the oxygen atom. The molecule is

assumed to be in the yz-plane. Thus, xy-plane bisects the two hydrogen atoms.
All the representations of the symmetry operators in this basis can be identified
by finding out the outcomes of their actions on the vectors of six orbital states as
described below. First, the representation of .Ĉ2 operation is obtained by
⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎛ ⎞
SA SB 0 1 0 0 0 0 SA
⎜S ⎟ ⎜ S ⎟ ⎜1 0 0 0 0 0⎟ ⎜S ⎟
⎜ B⎟ ⎜ A ⎟ ⎜ ⎟⎜ B ⎟
⎜ SC ⎟ ⎜ SC ⎟ ⎜ 0 0 1 0 0 0 ⎟ ⎜ SC ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
.Ĉ2 ⎜ ⎟=⎜ ⎟=⎜ ⎟⎜ ⎟, (9.94)
⎜ px ⎟ ⎜ −px ⎟ ⎜ 0 0 0 −1 0 0 ⎟ ⎜ px ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎝ py ⎠ ⎝ −py ⎠ ⎝ 0 0 0 0 −1 0 ⎠ ⎝ py ⎠
pz pz 0 0 0 0 0 1 pz

where for convenience we have omitted the ket symbols for the atomic orbital states.
The actions of other operators in the .C2v group can be identified similarly. Thus,
⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎛ ⎞
SA SB 0 1 0 0 0 0 SA
⎜S ⎟ ⎜ S ⎟ ⎜1 0 0 0 0 0 ⎟ ⎜ SB ⎟
⎟ ⎜
⎜ B⎟ ⎜ A ⎟ ⎜ ⎟
⎜S ⎟ ⎜ S ⎟ ⎜0 0 1 0 0 0 ⎟ ⎜ SC ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
. σ̂v ⎜ C ⎟ = ⎜ C ⎟ = ⎜ ⎟ ⎜ ⎟ ,. (9.95)
⎜ px ⎟ ⎜ px ⎟ ⎜ 0 0 0 1 0 0 ⎟ ⎜ px ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎝ py ⎠ ⎝ −py ⎠ ⎝ 0 0 0 0 −1 0 ⎠ ⎝ py ⎠
pz pz 0 0 0 0 0 1 pz
⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎛ ⎞
SA SA 100 0 00 SA
⎜S ⎟ ⎜ S ⎟ ⎜0 1 0 0 0 0 ⎟ ⎜ SB ⎟
⎟ ⎜
⎜ B⎟ ⎜ B ⎟ ⎜ ⎟
⎜ S ⎟ ⎜0 0 1 0 0 0 ⎟ ⎜ SC ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
′ ⎜ SC ⎟
σ̂v ⎜ ⎟ = ⎜ C ⎟ = ⎜ ⎟⎜ ⎟. (9.96)
⎜ px ⎟ ⎜ −px ⎟ ⎜ 0 0 0 −1 0 0 ⎟ ⎜ px ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎝ py ⎠ ⎝ py ⎠ ⎝ 0 0 0 0 1 0 ⎠ ⎝ py ⎠
pz pz 000 0 01 pz
310 9 Polyatomic Molecules and Molecular Spectroscopy

Matrices shown above are matrix representations of the symmetry operations in

the basis of the six atomic orbitals. Thus, including the identity operation, which is
represented by a six dimensional unit matrix, one can calculate the characters for all
the symmetry operations as listed in the table below.

.C2v .Ê .Ĉ2 .σ̂v (xz)

′
.σ̂v (yz)

.Ŵ 6 0 2 4

Using the orthogonality theorem, we can identify the coefficients for irreducible
representations for the above reducible representation as follows:

1
. A1 : (1 · 6 + 1 · 0 + 1 · 2 + 1 · 4) = 3
4
1
A2 : (1 · 6 + 1 · 0 + (−1) · 2 + (−1) · 4) = 0
4
1
B1 : (1 · 6 + (−1) · 0 + 1 · 2 + (−1) · 4) = 1
4
1
A2 : (1 · 6 + (−1) · 0 + (−1) · 2 + 1 · 4) = 2
4
Therefore, we find that

Ŵ = 3A1 + B1 + 2B2 .
. (9.97)

This implies that the MOs that are simultaneous eigenstates of the symmetry
operations of the .C2v group, which can be constructed out of the six atomic orbitals
are three MOs with the symmetry of .A1 , one MO with the symmetry of .B1 , and two
MOs with the symmetry of .B2 . Due to the simplicity of water molecule, these MOs
can be identified easily by inspection.
√ it is clear that the three MOs with .A1 symmetry correspond to .(|SA  +
Indeed,
|SB )/ 2, .|SC , and .|pz . The MO with .B1 symmetry is the .px orbital of oxygen
√
and thus corresponds to .|px . Two MOs with .B2 symmetry are .(|SA  − |SB )/ 2
and .|pz . Note that these are not yet bonding or anti-bonding orbitals, but can
be used to construct them easily because only MOs belonging to the same
irreducible representations can be combined together. This implies that at least
three bonding/anti-bonding MOs can be constructed from the three MOs with .A1
symmetry and that bonding and anti-bonding MOs can be constructed from the two
MOs with .B2 symmetry. On the other hand, the single orbital with .B1 symmetry,
.|px , is expected to remain as nonbonding. Figure 9.7 shows four MOs obtained

from a HF calculation, and indicates that the simple analysis based on the group
theory can indeed explain qualitative patterns of the lowest unoccupied molecular
orbital (LUMO), the highest occupied molecular orbital (HOMO) and HOMO-1.
9.5 Spectroscopy of Polyatomic Molecules 311

Fig. 9.7 Four MOs of a

water molecule, highest
occupied MO (HOMO),
lowest unoccupied MO
(LUMO), HOMO-1, and
LUMO+1 obtained from a
Hartree-Fock calculation.
Their symmetry properties
are also shown

9.5 Spectroscopy of Polyatomic Molecules

Within the BO approximation, the eigenstates of the molecular Hamiltonian can

be decoupled to electronic and nuclear parts as described in Sect. 9.1. Given that
the potential energy surfaces .VBO,a (R) can be determined by solving the electronic
Schrödinger equation, Eq. (9.9), for all possible values of the nuclear coordinates,7
one can then obtain the Schrödinger equation for the nuclear degrees of freedom,
Eq. (9.19). This Schrödinger equation can again be decomposed into translation,
rotation, and vibrational components. In the absence of external potential, the
translational components can be ignored.

9.5.1 Infrared and Raman Spectroscopy

As in the case of diatomic molecules, the selection rules for the IR and Raman
spectroscopy are the same. In other words, . v = ±1 in both cases. However, for
general polyatomic molecules with many vibrational normal modes, determining
which modes are IR and Raman active is not a simple task. The group theory and
character table for each group is very helpful in this respect.
Given a molecule, we can identify a point group with the maximum number
of symmetry operations with respect to which the molecule remains invariant.
Then, we can determine the characters for the reducible representation of all the

7 While this is possible in principle, it is almost impossible in practice for most polyatomic
molecules. Thus, only partial information around local minima around the minimum of each
electronic state is typically available.
312 9 Polyatomic Molecules and Molecular Spectroscopy

coordinates for operations in the group. Application of the orthogonality theorem

then allows determining the linear combination of irreducible representations. Once
all the translational and rotational modes are subtracted, the IR active modes
correspond to those that behave like x, y, and z. The Raman active modes
correspond to those that behave like quadratic functions of the coordinates.
Let us once again consider the water molecule as an example. For this, consider
the representation of the symmetry operations in the nine dimensional cartesian
coordinates consisting of .x1 , .y1 , .z1 , .x2 , .y2 , .z2 , .x3 , .y3 , and .z3 , where the subscript
1 denotes the coordinates of the oxygen and 2 and 3 denote the coordinates of two
hydrogen atoms. In this basis, the character of the identity operator .χ (Ê) = 9. The
operation of .Ĉ2 can be represented according to the following relation:

⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎛ ⎞
x1 −x1 −1 0 0 0 0 0 0 0 0 x1
⎜ y1 ⎟ ⎜ −y1 ⎟ ⎜ 0 −1 0 0 0 0 0 0 0 ⎟ ⎜ y1 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜z ⎟ ⎜ z ⎟ ⎜ 0 0 1 0 0 0 0 0 0⎟⎜z ⎟
⎜ 1⎟ ⎜ 1 ⎟ ⎜ ⎟⎜ 1 ⎟
⎜ x ⎟ ⎜ −x ⎟ ⎜ 0 0 0 0 0 0 −1 0 0 ⎟ ⎜ x ⎟
⎜ 2⎟ ⎜ 3⎟ ⎜ ⎟⎜ 2⎟
.Ĉ2 ⎜ y2 ⎟ = ⎜ −y3 ⎟ = ⎜ 0 0 0 0 0 0 0 −1 0 ⎟ ⎜ y2 ⎟ , (9.98)
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜ z2 ⎟ ⎜ z3 ⎟ ⎜ 0 0 0 0 0 0 0 0 1 ⎟ ⎜ z2 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜ x3 ⎟ ⎜ −x2 ⎟ ⎜ 0 0 0 −1 0 0 0 0 0 ⎟ ⎜ x3 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎝ y3 ⎠ ⎝ −y2 ⎠ ⎝ 0 0 0 0 −1 0 0 0 0 ⎠ ⎝ y3 ⎠
z3 z2 0 00 0 01 0 00 z3

for which .χ (Ĉ2 ) = −1. The operation of .σ̂v can be represented according to the
following relation:

⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎛ ⎞
x1 x1 1 000 000 00 x1
⎜ y1 ⎟ ⎜ −y1 ⎟ ⎜ 0 −1 0 0 0 0 0 0 0 ⎟ ⎜ y1 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜z ⎟ ⎜ z ⎟ ⎜0 0 1 0 0 0 0 0 0⎟⎜z ⎟
⎜ 1⎟ ⎜ 1 ⎟ ⎜ ⎟⎜ 1 ⎟
⎜x ⎟ ⎜ x ⎟ ⎜0 0 0 0 0 0 1 0 0⎟⎜x ⎟
⎜ 2⎟ ⎜ 3 ⎟ ⎜ ⎟⎜ 2⎟
.σ̂v ⎜ y2 ⎟ = ⎜ −y3 ⎟ = ⎜ 0 0 0 0 0 0 0 −1 0 ⎟ ⎜ y2 ⎟ , (9.99)
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜ z2 ⎟ ⎜ z3 ⎟ ⎜ 0 0 0 0 0 0 0 0 1 ⎟ ⎜ z2 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜ x3 ⎟ ⎜ x2 ⎟ ⎜ 0 0 0 1 0 0 0 0 0 ⎟ ⎜ x3 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎝ y3 ⎠ ⎝ −y2 ⎠ ⎝ 0 0 0 0 −1 0 0 0 0 ⎠ ⎝ y3 ⎠
z3 z2 0 000 010 00 z3
9.5 Spectroscopy of Polyatomic Molecules 313

for which .χ (σv ) = 1. Finally, the operation of .σ̂v′ can be represented according to
the following relation:
⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎛ ⎞
x1 −x1 −1 0 0 0 0 0 0 0 0 x1
⎜ y1 ⎟ ⎜ y1 ⎟ ⎜ 0 1 0 0 0 0 0 0 0 ⎟ ⎜ y1 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜z ⎟ ⎜ z ⎟ ⎜ 0 0 1 0 0 0 0 0 0⎟⎜z ⎟
⎜ 1⎟ ⎜ 1 ⎟ ⎜ ⎟⎜ 1 ⎟
⎜ x ⎟ ⎜ −x ⎟ ⎜ 0 0 0 −1 0 0 0 0 0 ⎟ ⎜ x ⎟
⎜ 2⎟ ⎜ 2⎟ ⎜ ⎟⎜ 2⎟
′⎜
0 0 0 0 1 0 0 0 0 ⎟ ⎜ y2 ⎟ , (9.100)
⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
.σ̂v ⎜ y2 ⎟ = ⎜ y2 ⎟ = ⎜
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜ z2 ⎟ ⎜ z2 ⎟ ⎜ 0 0 0 0 0 1 0 0 0 ⎟ ⎜ z2 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜ x3 ⎟ ⎜ −x3 ⎟ ⎜ 0 0 0 0 0 0 −1 0 0 ⎟ ⎜ x3 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟
⎝ y3 ⎠ ⎝ y3 ⎠ ⎝ 0 0 0 0 0 0 0 1 0 ⎠ ⎝ y3 ⎠
z3 z3 000 000 001 z3

for which .χ (σ̂v′ ) = 3. Summing up all these results, we obtain all the characters for
the reducible representation as listed in the table below.

.C2v .Ê .Ĉ2 .σ̂v (xz)

′
.σ̂v (yz)

.Ŵ 9 .−1 1 3

Application of the orthogonality theorem then leads to

1
. A1 : (9 − 1 + 1 + 3) = 3
4
1
A2 : (9 − 1 − 1 − 3) = 1
4
1
B1 : (9 + 1 + 1 − 3) = 2
4
1
B2 : (9 + 1 − 1 + 3) = 3
4
Therefore, we find that

.Ŵ = 3A1 + A2 + 2B1 + 3B2 . (9.101)

Out of the above linear combinations, the translational motion of the molecule
corresponds to .Ŵtr = A1 + B1 + B2 and the rotational motion corresponds to
.Ŵrot = A2 + B1 + B2 according to the character table of .C2v . Thus, subtracting

these from .Ŵ given above, we obtain the following decomposition of the vibrational
reducible representation:

Ŵvib = 2A1 + B2 .
. (9.102)
314 9 Polyatomic Molecules and Molecular Spectroscopy

Fig. 9.8 Three vibrational modes of a water molecule

Thus, there are two vibrational modes with .A1 symmetry, which correspond to
symmetric stretching and bending, and one vibrational mode with .B2 symmetry,
which is asymmetric stretching (see Fig. 9.8). Both of these symmetries contain
those that are linear and quadratic in the cartesian coordinates. Therefore, they are
both IR and Raman active.

9.5.2 Electronic Spectroscopy

In general, the electronic spectroscopy involves transition between electronic states

due to the interaction between the electric field part of the radiation and the
following full dipole operator involving electrons,

Nu Ne

 
D̂ T (r̂, R̂) = e
. Zc R̂c − r̂ν , (9.103)
c=1 ν=1

where the .Zc and .R̂c are nuclear charges and position operators and the .r̂ν are
the position operators of the electron .ν. Thus, in general, the electronic transition
involving one photon is allowed as long as the following transition dipole vector is
nonzero.

D tr
. if = f |D̂T (r̂, R̂)|i , (9.104)

where .|i  and .|f  refer to initial and final molecular states including both
electronic and nuclear degrees of freedom. Let us assume that the initial and final
states can be approximated as the product of electronic states and nuclear wave
function in respective BO surfaces as follows:
 
|i  =
. dr dR |r, Rr, R|i 
 
= dr dR |r, Rψnui (R)ψei (r; R), . (9.105)
9.5 Spectroscopy of Polyatomic Molecules 315

 
|f  = dr dR |r, Rr, R|f 
 
= dr dR |r, Rψnuf (R)ψef (r; R). (9.106)

Then,
 
∗
tr
.D if = dR ψnuf
(R)ψnui (R) dr ψe∗f (r; R)D̂ T (r, R)ψei (r; R). (9.107)

In many electronic transitions, it turns out that the integration of the dipole
operator with respect to the electronic degrees of freedom can be approximated in
terms of the following electronic transition dipole vector calculated for a reference
nuclear coordinate:

e
.D if (R0 ) = dr ψe∗f (r; R0 )D̂T (r, R0 )ψei (r; R0 )
  

= −e dr ψe∗f (r; R0 ) rν ψei (r; R0 ), (9.108)
ν

where .R0 is the reference nuclear coordinates that are typically assumed to be
the coordinates for the minimum of the initial electronic state and the second
equality results from the fact that the two electronic states determined for the
nuclear coordinate .R0 are orthogonal to each other. This is so called Franck-Condon
approximation, and the resulting approximation for the total transition dipole vector
is as follows:

tr e ∗
.D if ≈ D if dR ψnu f
(R)ψnui (R), (9.109)

where the overlap integration of the nuclear wavefunction is nonzero in general

because they are defined with respect different structures and is known as the
Franck-Condon factor.
For electronic absorption spectroscopy, the initial state is the ground electronic
state (labeled as X) and the transition occurs to an electronic state with the same spin
multiplicity as the ground state (labeled as A, B, C, etc) and with nonzero value of
the electronic transition dipole vector given by Eq. (9.108). For small molecules
where all the vibrational frequencies are much larger than thermal energy, .ψnui (R)
can be approximated well by the product of ground state vibrational wave functions
of all the vibrational normal modes.
For the case of radiative electronic emission spectroscopy, normally called
fluorescence, the initial state is one of the excited electronic states (A, B, C, etc),
which has the same spin multiplicity as the ground electronic state (X). In this case
as well, if the vibrational frequencies in the excited electronic state are much larger
316 9 Polyatomic Molecules and Molecular Spectroscopy

than the thermal energy, the initial nuclear wave function can be approximated well
by the product of ground vibrational state wave functions of all normal modes in the
excited electronic state potential energy surface.
Although fluorescence is the dominant mechanism of transitions of excited
electronic state, there are other non-radiative transitions that can occur as well. If
the non-radiative transition is between excited states of the same spin multiplicity,
it is called internal conversion. If the transition occurs to states with different
spin multiplicity (labeled as a, b, c, etc), it is called intersystem crossing. The
mechanism of this intersystem crossing is spin-orbit coupling. Radiative transition
from excited states with different spin multiplicity to the ground electronic state can
occur although it is dipole forbidden. This process is called phosphorescence and
typically takes microseconds and beyond.

9.6 Summary and Questions

Within the BO approximation, quantum states of polyatomic molecules can be

treated in two stages. First, ignoring the kinetic energy of nuclei, the Schrödinger
equation for electrons for fixed nuclei can be solved. Then, energies obtained
from these calculations for all possible nuclear coordinates (in principle) can be
incorporated into the potential energy for nuclei so as to describe the vibrational
states of molecules. Although conceptually straightforward, even solving the equa-
tions for electrons alone involves challenging numerical issues, for which various
approximation methods have been developed.
One of the simplest approximations that can be used for obtaining electronic
states in polyatomic molecules is to use LCAO-MOs and assume that electrons
can be filled in independent of each other while satisfying the Pauli principle.
This simple treatment leads to qualitatively correct description of small or sim-
ple molecules, helping to understand important molecular quantum mechanical
details without going through numerical calculations. Representative examples are
diatomic molecules of the second row, for which Table 9.1 sums up major results.
Verification or further correction of the simple MO description can be made through
photoelectron spectroscopy in combination with Koopmans’ theorem given by
Eq. (9.53).
Another important application of the LCAO-MO approximation is approximat-
ing .π -orbital energies of conjugated hydrocarbons, for which additional Hückel
approximation can also be made. This simple model has helped to understand
the types of conjugated hydrocarbons that can be stabilized and also contributed
to developing semi-empirical methods during early stages of electronic structure
calculation method development.
For the understanding and description of both electronic and vibrational states
of polyatomic molecules, group theory has also provided important conceptual
framework for understanding the symmetry of states and relevant selection rules.
The success of the group theory relies on the fact that a set of symmetry
Appendix: Important Theorems and Proofs in the Group Theory 317

operations that leave a certain molecule invariant can form a mathematical group.
The properties of the group then can be employed to classify electronic and
vibrational states of molecules. The most important theorem in such application
is the orthogonality theorem, based on which efficient ways to construct MOs and
to identify types of vibrational modes have been developed. In particular, for the
vibrational spectroscopy, identification of irreducible representations and referral of
the character table make it possible to determine IR and Raman active vibrational
normal modes of a molecule.

Questions

• What are neglected in the BO approximation and when do they become

significant?
• Is it possible to use the LCAO-MO approximation to describe formation
and breaking of chemical bonds?
• The component of the total orbital angular momentum along the internu-
clear axis in a diatomic molecule is conserved. What is the reason for this?
• The relative order of bonding .σ and .π orbitals for .O2 and .F2 are different
from other diatomic molecules with smaller atomic numbers in the second
row. What is the reason for this?
• What are additional assumptions involved in the Hückel approximation in
addition to the LCAO-MO approximation?
• Koopmans’ theorem is known to be fairly accurate despite its simple
expression and even though it does not account for electron correlation
effects. What is the implication of this success?
• When does a molecule have vibrational modes that are both IR and Raman
active?
• If a molecule belongs to a certain point group, all the symmetry opera-
tions constituting the point group commute with the Hamiltonian of the
molecule. What is the reason for this?
• What is the difference between reducible and irreducible representations
in the group theory?

Appendix: Important Theorems and Proofs in the Group

Theory

This appendix provides a complete set of theorems and proofs for obtaining the
orthogonality theorem, Eq. (9.88). The theorems and proofs presented here are
based on those presented by Tinkham [10], with some alternation in notations and
additional clarifications.
318 9 Polyatomic Molecules and Molecular Spectroscopy

Theorem 9.1 Rearrangement theorem: For any group .G, the multiplication of any
element of the group with all the elements of the group produce a complete set
containing all the elements, where each element of the group appears only once.
Proof Consider a set .S consisting of all the group elements multiplied with X,
which is also an element of the group. For convenience, assume that each element
is multiplied with X on the right.
Let us choose an arbitrary element of the group, say A. Then, the group .G should
contain .B = A ◦ X−1 as its element because there should be an inverse of X in the
group and the group is closed with respect to multiplication. Since .B ◦ X = A, A
should appear at least once in the set .S. Thus, (i) any group element of .G should
exist in the set .S.
Since .S and the original set of .G have the same number of elements, each
element should appear only once in the set of products. Otherwise, there should
be an element that cannot appear, which is contradictory with the statement (i). This
completes the proof for the case where the elements are multiplied with X on the
right. Similar proof is possible for the case where X is multiplied on the left by
considering .X−1 ◦ A. ⊔
⊓
As a result of the above theorem, the following identity holds for any sum over
all the elements of the group:
  
. F (Ak ) = F (Ak ◦ X) = F (X ◦ Ak ), (9.110)
k k k

where F is an arbitrary function, .Ak ’s for all k constitute the complete set of element
of the group, and X is an arbitrary element of the group.
Theorem 9.2 Any matrix representation of a group is equivalent to a representation
by a unitary matrix as long as determinants of all the matrices are nonzero.
Proof Let us denote the matrix representing the ith element of the group as .Ai ,
where .i = 1, · · · , Q. Then, one can always construct the following Hermitian
matrix:
Q

. H= Ai A†i , (9.111)
i=1

where .A†i is the Hermitian conjugate of .Ai . Therefore, .H can be diagonalized to .H′
by a unitary matrix .U as follows:

Q
 Q
 †
H′ = U† HU =
. U† Ai A†i U = A′i A′i , (9.112)
i=1 i=1
Appendix: Important Theorems and Proofs in the Group Theory 319

where .A′i = U† Ai U. Note that

Q 
 Q 

†  ′ 2
.[H′ ]jj = [A′i ]j k [A′i ]kj = [A ]j k  .
i (9.113)
i=1 k i=1 k

Therefore, all the diagonal elements of .H′ are real and positive, and we can define
.H
′ 1/2 and .H′ −1/2 . Therefore, Eq. (9.112) can also be expressed as

⎛ ⎞
Q
′ −1/2 ⎝ † −1/2
I=H
. A′i A′i ⎠ H′ , (9.114)
i=1

where .I is the identity matrix. Now, let us consider a new transformed matrices
A′′j = H′ −1/2 A′j H′ 1/2 for all .j = 1, · · · , Q. Then, these can be shown to be unitary
.

as follows:
† −1/2 1/2 1/2 † −1/2
.A′′j A′′j = H′ A′j H′ IH′
A′j H′
⎛ ⎞
Q
−1/2 1/2 −1/2 † −1/2 ′ 1/2 ′ † ′ −1/2
= H′ A′j H′ H′ ⎝ A′i A′i ⎠ H′ H Aj H
i=1
⎛ ⎞
Q

−1/2 ⎝ † † −1/2
= H′ A′j A′i A′i A′j ⎠ H′
i=1
⎛ ⎞
Q

−1/2 ⎝ † −1/2
= H′ A′k A′k ⎠ H′ = I, (9.115)
k=1

where in the second and last equality, Eq. (9.114) has been used and Eq. (9.110) in
matrix representation has been used for .A′k = A′j A′i . Now, each .A′′j is related to .Aj
as follows:
1/2 1/2
U† Aj UH′
−1/2 −1/2
A′′j = H′
. = H′ U−1 Aj UH′
1/2 −1 1/2
= UH′ Aj UH′ . (9.116)

Thus, .A′′j is related to .Aj by a similarity transformation. This completes the proof.
⊔
⊓
Theorem 9.3 Any Hermitian matrix commuting with all matrices of an irreducible
representation must be a constant matrix.
320 9 Polyatomic Molecules and Molecular Spectroscopy

Proof Let us denote the Hermitian matrix that commutes with all matrices of a
representation of a group as .H. This implies that

. Ai H = HAi , for i = 1, · · · , Q (9.117)

Since .H is Hermitian, there should be a unitary matrix .U such that .H′ = U† HU is

diagonal. Then, applying .U† and .U on the left and right side of the above equation,
we obtain

A′i H′ = H′ A′i ,
. (9.118)

where .A′i = U† AU = U−1 AU. Taking the matrix element of the above relation and
using the fact that the .H′ has only diagonal elements, we obtain

[A′i ]nm [H′ ]mm = [H′ ]nn [A′i ]nm ,

. (9.119)

which means that

[A′i ]nm [H′ ]mm − [H′ ]nn = 0.

. (9.120)

If all the diagonal elements of .H are the same, the above identity is satisfied. On the
other hand, if .[H′ ]mm − [H′ ]nn = 0 for some .n = m, the above identity means that
′
.[A ]nm = 0 for all i. This implies that the unitary transformation of .Ai ’s to .A ’s
′
i i
resulted in smaller blocks. However, this contradicts the assumption that .Ai ’s are
irreducible, Therefore, all the diagonal matrix elements of .H′ should be the same.
Thus, .H′ = cI, where c is an arbitrary constant, and .H = cUIU† = cI. ⊔
⊓
Theorem 9.4 Schur’s Lemma: Any matrix commuting with all matrices of an
irreducible representation must be a constant matrix.
Proof Let us denote the matrix that commutes with all matrices of a representation
of a group as .M. This implies that

Ai M = MAi , for i = 1, · · · , Q
. (9.121)

Taking the Hermitian conjugate of the above relationship, we obtain

M† A†i = A†i M† .
. (9.122)

Then, by multiply .Ai and .Ai on the left and right side of the above expression, we
obtain

Ai M† A†i Ai = Ai A†i M† Ai .
. (9.123)

According to Theorem 9.2, we can always find out a representation where all .A’s
are unitary by similarity transformation. Thus, let us define such a transformation
Appendix: Important Theorems and Proofs in the Group Theory 321

matrix as .S, which has inverse, and let .A′i = S−1 Ai S. Then, multiplying .S−1 and
.S on the left and right hand side of the above expression, and inserting .1 = S
−1 S

between every operator, we obtain

† †
.A′i S−1 M† SA′i A′i = A′i A′i S−1 M† SA′i . (9.124)

Since .A′i ’s are unitary, the above relation implies that

† †
.A′i M′ = M′ A′i , (9.125)

where .M′ = S−1 M† S. On the other hand, applying the same similarity transforma-
tion to Eq. (9.121), we obtain

A′i M′ = M′ A′i ,
. (9.126)

Now, let us defined .H1 = M′ + M′ † and .H2 = i(M′ − M′ † ). Then, employing

Eqs. (9.125) and (9.126), we find that

A′i H1 = H1 A′i , .
. (9.127)
A′i H2 = H2 A′i , (9.128)

for all .i = 1, · · · , Q. According to Theorem 9.3, .H1 and .H2 are constant matrices.
Therefore, .M′ is constant matrix and .M = SM′ S−1 is also a constant matrix. ⊔
⊓
Theorem 9.5 There are two irreducible matrix representations of the same group,
Ŵ (1) (Ai ) and .Ŵ (2) (Ai ), each with dimensionality .l1 and .l2 . Let us assume that a
.

rectangular matrix .M exists such that

.MŴ (1) (Ai ) = Ŵ (2) (Ai )M, (9.129)

for all .i = 1, · · · , Q. Then, the following is true. For .l1 =

 l2 , .M = 0. For .l1 = l2 ,
either .M = 0 or the determinant of .M is nonzero, for which the two irreducible
representations are equivalent.
Proof Assume that all the irreducible matrices are unitary as well. If not, they
can be transformed to unitary matrices according to Theorem 9.2. Then, taking the
Hermitian conjugate of Eq. (9.129) and using the fact that the matrix representations
are unitary,
† † † † (2) † † (2)
Ŵ (1) (A−1
.
(1) −1
i )M = Ŵ (Ai ) M = M Ŵ (Ai ) = M Ŵ (Ai ). (9.130)
322 9 Polyatomic Molecules and Molecular Spectroscopy

Multiplying the above expression with .M on the lefthand side and using the fact that
MŴ (1) (A−1
.
−1
i ) = Ŵ (Ai )M, which is a specific case of Eq. (9.129) for .Ai , we
(2) −1

obtain the following expression:

† † † (2)
MŴ (1) (A−1
.
(2) −1 −1
i )M = Ŵ (Ai )MM = MM Ŵ (Ai ). (9.131)

Since the above relation holds for any .A−1

i in the group, according to Theorem 9.4,
the above relation implies that

MM† = cI,
. (9.132)

where .I is the identity matrix of dimension .l1 .

If .l1 = l2 , Eq. (9.134) implies that the absolute square of the determinant of .M
is equal to .cl1 . For the case where .c = 0, this implies that the determinant of .M is
nonzero. For the case where .c = 0, considering the diagonal elements of Eq. (9.134),
we find that
 
. [M]mn [M† ]nm = |[M]mn |2 = 0. (9.133)
n n

The above identity holds true only when all the matrix elements are zero.
If .l1 < l2 , we can create a square matrix of dimension .l2 by adding .(l2 − l1 )
columns of zeros to .M. Then, .NN† = MM† . Therefore,

NN† = cI.
. (9.134)

Since the determinant of .N is zero, the above identity implies that .c = 0. Then, the
proof for the case of .l1 = l2 above can again be applied to show that .N = 0.
If .l1 > l2 , similar proof is possible by constructing a square matrix of dimension
.l1 by inserting .l1 − l2 rows of zeros in .M. ⊔
⊓
Theorem 9.6 The great orthogonality theorem: For all possible irreducible and
unitary representations of a group,
 Q
. [Ŵ (j ) (G)]∗mm′ [Ŵ (k) (G)]nn′ = δj k δmn δm′ n′ , (9.135)
lj
G

where the summation index G represents all possible group elements and .lj is the
dimensionality of .Ŵ (j ) .
Proof First, consider the case where the two representations are inequivalent, for
which one can construct the following matrix:

M=
. Ŵ (k) (G)XŴ (j ) (G−1 ), (9.136)
G
Appendix: Important Theorems and Proofs in the Group Theory 323

where .Ŵ (k) (G) is a .lk × lk matrix, .Ŵ (j ) (G−1 ) is a .lj × lj matrix, and .X is an arbitrary
matrix with .lk row and .lj column. Then, for a certain element S of the group, it is
straightforward to show the following identities:

Ŵ (k) (S)M =
. Ŵ (k) (S)Ŵ (k) (G)XŴ (j ) (G−1 )
G

= Ŵ (k) (S)Ŵ (k) (G)XŴ (j ) (G−1 )Ŵ (j ) (S −1 )Ŵ (i) (S)
G

= Ŵ (k) (SG)XŴ (j ) (G−1 S −1 )Ŵ (k) (S)
G

= Ŵ (k) (SG)XŴ (j ) ((SG)−1 )Ŵ (k) (S)
SG

= Ŵ (k) (G)XŴ (j ) ((G)−1 )Ŵ (k) (S) = MŴ (k) (S). (9.137)
G

In the last equality of the above equation, Theorem 9.1 was used. According to the
Theorem 9.5, Eq. (9.137) implies that .M = 0 for .l1 = l2 or the two representations
are not equivalent for .l1 = l2 . Thus,

[M]nm′ = 0 =
. [Ŵ (k) (G)]np Xpq [Ŵ (j ) (G)]qm′ . (9.138)
G pq

Now, consider the case where the elements of the matrix .X are zero except for
the .n′ m element as follows: .Xpq = δpn′ δqm . Then, the above relation becomes the
following identity:

. [Ŵ (k) (G)]nn′ [Ŵ (j ) (G)]mm′ = 0, (9.139)
G

which completes the proof for the cases where .lj = lk or the two representations
are not equivalent while .lj = lk .
Now, let us consider the case where the two representations are equivalent. Thus,
let us drop the superscript j and k. Then,

M=
. Ŵ(G)XŴ(G−1 ) = cI. (9.140)
G

Therefore, considering the .mm′ element of the above identity, we obtain


. [Ŵ(G)]mp [X]pq [Ŵ(G−1 )]qm′ = cδmm′ . (9.141)
pq G
324 9 Polyatomic Molecules and Molecular Spectroscopy

Let us consider the case where .[X]pq = δpn δqn′ . Then,


. [Ŵ(G)]mn [Ŵ(G−1 )]n′ m′ = cδmm′ . (9.142)
G

Summing the above elements over .m = m′ , we thus obtain


. [Ŵ(G)]mn [Ŵ(G−1 )]n′ m = clj . (9.143)
m G

On the other hand, the lefthand side of the above is equal to .Qδnn′ as follows.
 
. [Ŵ(G)]mn [Ŵ(G−1 )]n′ m = [Ŵ(G−1 G)]n′ n = Qδn′ n . (9.144)
G m G G G

Therefore, .c = δnn′ Q/ lj . As a result,

  Q
. [Ŵ(G)]mn [Ŵ(G−1 )]n′ m′ = [Ŵ(G)]mn [Ŵ(G)]∗m′ n′ = δmm′ δnn′ , (9.145)
lj
G G

where the fact that .Ŵ is a unitary irreducible representation was used. ⊔

⊓

Exercise Problems with Solutions

9.1 Write down the ground electronic configuration for B− 2 , determine its bond
order, and the term for the ground electronic state. Use KK for the core shell.
Solution 9.1 For B−
2 , which has 11 electrons, the configuration is as follows:

KK(σg 2s)2 (σu 2s)2 (πu 2px,y )3 .

.

The bond order for this is 3/2. The molecular terms for this is the same as
(πu 2px,y )1 , for which the only and the ground state term is 2 u .
9.2 Answer the following questions for the electronic states of N+
2.
(a) Write down the ground electronic configuration for N+ 2 , determine its bond
order, and the term for ground electronic state.
(b) Assume that an electron in the πu 2px,y orbital of the ground electronic
configuration gets excited to a 2πg 2px,y orbital. Write down all possible term
symbols for the resulting excited state configuration. Among these, list all the
states to which the transition from the ground state is allowed. For each allowed
transition, state the appropriate selection rule clearly.
Exercise Problems with Solutions 325

Solution 9.2
(a) The ground electronic configuration of N+
2 is as follows:

KK(σg 2s)2 (σu 2s)2 (πu 2px,y )4 (σg 2pz )1

.

For this configuration, BO = 5/2. Since all MOs are fully filled except for
σg 2pz , which is singly filled, there is only one term for this as follows:2 g+ .
(b) The corresponding excited state configuration is as follows:

KK(σg 2s)2 (σu 2s)2 (πu 2px,y )3 (σg 2pz )1 (πg 2px,y )1
.

This is equivalent to the following:

(πu 2px,y )1 (σg 2pz )1 (πg 2px,y )1

.

Since there are three electrons occupying three different orbitals, all possible
combinations of  and S can be used. First, summing the states of two electrons
occupying πu 2px,y and σg 2pz respectively, we obtain

 = 1; S = 0, 1
.

Summing these with the state of the electron occupying πg 2px,y , we obtain

1 3
 = 0, 2; S =
. ,
2 2
All states have u symmetry. Therefore, the resulting molecular term symbols
are as follows:
2
. u± , 4 u±
2 4
u, u

Among the above terms, transition from the ground state 2 g+ to the following
state are allowed:
2
. u±

9.3 The π -type molecular orbital of a (noncyclic) propyl anion, CH2 CHCH2 − , is
given by the linear combination of the three carbon 2p-orbitals as follows:

|ψ = C1 |2pz1  + C2 |2pz2  + C3 |2pz3 ,

.
326 9 Polyatomic Molecules and Molecular Spectroscopy

where 1, 2, and 3 denote carbon atoms from the left to the right. For the molecular
Hamiltonian Ĥ ,

. 2pz1 |Ĥ |2pz1  = 2pz2 |Ĥ |2pz2  = 2pz3 |Ĥ |2pz3  = α,

2pz1 |Ĥ |2pz2  = 2pz2 |Ĥ |2pz3  = β.

Assume that 2pz1 |Ĥ |2pz3  = 0 and that all overlap integrals between different
orbitals are zero. Use the variational principle and determine three MOs and their
energies, and determine the total π -electron energy of propyl anion.
Solution 9.3 Applying the variational principle, we obtain the following matrix
equation:
⎛ ⎞⎛ ⎞ ⎛ ⎞
α−E β 0 C1 0
.⎝ β α − E β ⎠ ⎝ C2 ⎠ = ⎝ 0 ⎠ .
0 β α−E C3 0

This is possible for non-zero the Ck only if the determinant is zero as follows:
 
α − E β
 0 
3 2
 β α − E β  = (α − E) − 2β (α − E)
. 
 0 β α −E
√ √
= (α − E)(α − E − 2β)(α − E + 2β) = 0.
√
Thus,√there are the following three solutions for the energy: E = α, α − 2β, and
α + 2β. The coefficients for MO can be determined by plugging each value of E
into the above matrix equation and also using the normalization condition.
For E = α, βC2 = 0, β(C1 + C3 ) = 0. This means that C2 = 0 and C1 = −C3 .
Using the normalization condition, we find that

1
|ψn  = √ (|2pz1  − |2pz3 ) .
.
2
√ √
√ For E = α − 2β, two independent equations
√ are 2C1 +√C2 = 0 and C2 +
2C3 = 0, which result in C1 = −C2 / 2 and C3 = −C2 / 2. C2 can then be
determined by normalization condition. Thus,

1 1 1
|ψ−  = √ |2pz1  − √ |2pz2  + |2pz3 .
.
2 2 2
Exercise Problems with Solutions 327

√ √
For√E = α + 2β, two independent equations
√ are − 2C√1 + C2 = 0 and
C2 − 2C3 = 0, which result in C1 = C2 / 2 and C3 = C2 / 2. C2 can then be
determined by normalization condition. Thus,
1 1 1
.|ψ+  = √ |2pz1  + √ |2pz2  + |2pz3 .
2 2 2
The four π electrons occupy |ψ+  and |ψn . Thus the π-electron energy is as follows:
√ √
.Eπ = 2(α + 2β) + 2α = 4α + 2 2β.

9.4 For 1,3-butadienyl anion, [CH2 = CH − CH = CH2 ]−1 , the four LCAO-MO
π orbitals formed by pz orbitals of carbon
√ atoms, within the Hückel approximation,
have the following energies: α ± β( 5 ± 1)/2, where α is the energy of each pz
orbital and β is the resonance integral between neighboring pz orbitals. Find out the
expressions for the total π orbital energy and for the total π -bonding delocalization
energy.
Solution 9.4 There are five electrons occupying π MOs. Below are the three MO
energies from the lowest and the number of electrons in parenthesis.
√ √ √
5+1 5−1 5−1
.α + β (2); α + β (2); α − β (1)
2 2 2
Summing the above for all five electrons,
√
3 5+1
.Eπ = 5α + β .
2
The π orbital delocalization energy is given by
√
3 5−9
Eπ−deloc
. = Eπ − 5(α + β) = β .
2
The above quantity is positive, which means that there is no stabilization within the
Hückel approximation.
9.5 The molecule PCl3 shown below, where all the P − Cl distances are the same
and all the Cl − P − Cl angles are the same, belongs to C3v point group.
328 9 Polyatomic Molecules and Molecular Spectroscopy

Use the character table below to answer questions below.

C3v E 2C3 3σv

A1 1 1 1 z, x 2 + y 2 , z2
A2 1 1 −1 Rz
E 2 −1 0 (x, y), (Rx , Ry ), (x 2 − y 2 , xy), (xz, yz)

(a) The reducible representation of the 12-dimensional coordinates of all atomic

displacements of PCl3 is found to be Ŵ = 3A1 + A2 + 4E. Complete the values
of characters for this reducible representation in the table below.

C3v E 2C3 3σv

.
Ŵ

(b) Identify all the irreducible representations constituting the vibrational modes.
(c) How many IR active modes are and what are their irreducible representations?
(d) How many Raman active modes are and what are their irreducible representa-
tions?
Solution 9.5
(a) Making the linear combination of characters for each symmetry operation, we
obtain the following components for the table.

C3v E 2C3 3σv

.
Ŵ 12 0 2

The above elements can be confirmed to produce the given linear combina-
tion as follows:
1
. A1 : (12 · 1 + 2 · 0 · 1 + 3 · 2 · 1) = 3
6
1
A2 : (12 · 1 + 2 · 0 · 1 + 3 · 2 · (−1)) = 1
6
1
E : (12 · 2 + 2 · 0 · (−1) + 3 · 2 · 1) = 4
6
(b) From the character table, it is clear that

. Rtr = A1 + E,
Rrot = A2 + E.
Problems 329

Therefore,

Ŵvib = Ŵ − Ŵtr − Ŵrot = 3A1 + A2 + 4E − (A1 + A2 + 2E) = 2A1 + 2E.

.

(c) and (d): There are 6 vibrational modes, which are all IR and Raman active
according to their symmetry properties.
9.6 Express a reducible representation Ŵ, for which the values of characters
are shown in the following character table for D3 , in terms of all irreducible
representations and identify all possible candidates for irreducible representations
that are IR and Raman active vibrational modes.

D3 E 2C3 3C2
A1 1 1 1 z2 , x 2 + y 2
. A2 1 1 −1 z Rz
E 2 −1 0 (x, y), (x 2 − y 2 , xy), (xz, yz) (Rx , Ry )
Ŵ 12 0 −2

Solution 9.6 Each component of Ŵ is as follows:

1 1 1
A1 :
. (12 − 6) = 1; A2 : (12 + 6) = 3; E: (2 · 12) = 4
6 6 6
Therefore, Ŵ = A1 + 3A2 + 4E. Since Ŵtr = A2 + E and ŴR = A2 + E, Ŵvib =
Ŵ − Ŵtr − ŴR = A1 + A2 + 2E. Out of these five modes belonging to A2 and E are
IR active, and five modes belonging to A1 and E are Raman active.

Problems

9.7 Write down the ground electronic configuration for C2 + , corresponding bond
order, and all possible electronic terms. If an electron in σu 2s of this configuration
becomes excited to πg 2px,y , what are possible terms corresponding to the excited
electronic configuration?
9.8 For 1,3-butadiene cation, [C4 H6 ]+ , what is the π -orbital delocalization energy
within the Hückel approximation? Express your answer in terms of α = Hii and
β = Hi,i+1 .
9.9 The molecule allene (H2 C = C = CH2 ) belongs to a point group D2d . Draw
the molecule and indicate all the symmetry elements. Assume that hydrogen atoms
are rigidly fixed to carbon atoms. Then, find out reducible representation of the
molecule in the basis of 9 dimensional coordinates of carbon atoms only. Express
this as the sum of irreducible representations and identify all IR and Raman active
modes.
Chapter 10
Quantum Dynamics of Pure and Mixed
States

Whatever Nature has in store for mankind, unpleasant as it may

be, men must accept, for ignorance is never better than
knowledge.
— Enrico Fermi

Abstract This chapter provides an overview of quantum dynamics and its approxi-
mations. For pure states that can be represented by a linear combination of kets/bras
(or wavefunctions), this entails solving the time dependent Schrödinger equation.
Three major pictures of the dynamics, Schrödinger picture, Heisenberg picture, and
interaction picture are explained first. Then, the interaction picture is used for the
description of time dependent perturbation theory, and Fermi’s golden rule is derived
from the first order approximation. For the description of more general mixed states,
introduction of density operator and extension of the time dependent Schrödinger
equation to the quantum Liouville equation become necessary. Application of the
perturbation theory to this quantum Liouville equation is presented as well.

So far, the focus has been on stationary (time independent) properties of a closed
quantum system and a pure quantum state that can be expressed by a single ket (bra)
or wavefunction. While this has been enough for explaining the foundational aspects
of quantum mechanics and applications to well-isolated molecular properties, it is
limited for quantitative understanding of time dependent quantities and properties of
a large collection of molecules that are interacting with each other. Most chemical
phenomena involve understanding these dynamical processes in complex environ-
ments. For this, explicit consideration of quantum dynamics and general approaches
to represent mixed states are necessary. This chapter describes basic principles
governing the dynamics of pure and mixed states, and simple approximations.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 331
S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_10
332 10 Quantum Dynamics of Pure and Mixed States

10.1 Quantum Dynamics of Pure States

A pure quantum state refers to any single quantum mechanical state that can be
represented by a ket (bra) or a linear superposition of kets (bras). As described
briefly in Chap. 2, for a closed quantum system with a Hamiltonian operator .Ĥ , the
pure state at time t can be determined completely by solving the time dependent
Schrödinger equation, Eq. (2.43). Given that the initial state is .|ψ, the solution of
the time dependent Schrödinger equation can be expressed in terms of the time
evolution operator .ÛH (t) defined by Eq. (2.82). The resulting expression, Eq. (2.85),
can again be expressed in a number of different ways as follows:
 
i
.|ψ; t = ÛH (t)|ψ = exp − Ĥ t |ψ
h̄
∞ 
it n n
 1  
= − Ĥ |ψ. (10.1)
n! h̄
n=0

In the limit of .t = 0, the above expression approaches the limit of .|ψ; t = 0 = |ψ
as it should.
Given the time dependent state as expressed above, the corresponding time
dependent expectation value at time t of any physical observable .Â, which is
assumed to be time independent for now, is equal to

Ât = ψ; t|Â|ψ; t.

. (10.2)

While this view of the quantum dynamics is sufficient for describing all the time
dependent processes in quantum mechanics and is most widely used, it turns out to
be just one of many possible “pictures” of the dynamics known as the Schrödinger
picture. In this picture, the time evolution of the system is viewed as being entirely
due to the change of the state from .|ψ to .|ψ; t although there can still be an
additional time dependence of an operator . due to extrinsic factors. To make this
clear, let us denote the operator in the Schrd̈inger picture with subscript “S” and also
assume the possibility of additional time dependence that cannot be explained in
terms of time evolving quantum states in the Schrödinger picture. Thus, the general
time dependent expectation value of any time dependent physical observable can be
expressed as

Â(t)t = ψ; t|ÂS (t)|ψ; t.

. (10.3)
10.1 Quantum Dynamics of Pure States 333

Taking the time derivative, we find that the above time dependent expectation value
satisfies the following differential equation:
   
d ∂ ∂
. Â(t)t = ψ; t| ÂS (t)|ψ; t + ψ; t|ÂS (t) |ψ; t
dt ∂t ∂t
 
∂
+ψ; t| ÂS (t) |ψ; t
∂t
i i
= ψ; t|Ĥ ÂS (t)|ψ; t − ψ; t|ÂS (t)Ĥ |ψ; t
h̄ h̄
 
∂
+ψ; t| ÂS (t) |ψ; t
∂t
 
i ∂
= ψ; t| [Ĥ , ÂS (t)] + ÂS (t) |ψ; t. (10.4)
h̄ ∂t

10.1.1 Heisenberg Picture

Heisenberg viewed that the dynamics is entirely due to time evolving physical
observables whereas the state remains the same. This is known as the Heisenberg
picture and can easily be shown to be equivalent to the Schrödinger picture. For
any time dependent operator in the Schrödinger picture .ÂS (t), we can define the
following time dependent operator in the Heisenberg picture:

ÂH (t) = ÛH† (t)ÂS (t)ÛH (t).

. (10.5)

Then, the expectation value of .Â(t) at time t can be viewed as the expectation of the
above operator with respect to the time independent state .|ψ as follows:

.Â(t)t = ψ|ÂH (t)|ψ = ψ|ÛH† (t)ÂS (t)ÛH (t)|ψ = ψ; t|ÂS (t)|ψ; t,
(10.6)

where the second equality shows the equivalence of the two pictures.
In the Heisenberg picture, all the time dependent properties are directly encoded
into time evolving operators as if physical observables evolve in time in classical
mechanics. This is represented by the time evolution equation for the time dependent
334 10 Quantum Dynamics of Pure and Mixed States

operator as follows:

d i i
. ÂH (t) = Ĥ ÛH† (t)ÂS (t)ÛH (t) − ÛH† (t)ÂS (t)ÛH (t)Ĥ
dt h̄ h̄
 
∂
+ÛH† (t) ÂS (t) ÛH (t)
∂t
 
i ∂
= [Ĥ , ÂH (t)] + ÛH† (t) ÂS (t) ÛH (t). (10.7)
h̄ ∂t

For the case where .ÂS (t) commutes with the Hamiltonian .Ĥ , it also commutes with
the time evolution operator .ÛH (t). For this case, .ÂH (t) = ÂS (t), and the time
derivative of .ÂH (t) is the same as the partial derivative of .ÂS (t) with respect to
time.
In the Heisenberg picture, the relationship between time dependent position and
momentum operators are similar to those for classical mechanics. For this, let us
consider the case of the following one dimensional Hamiltonian:

p̂2
Ĥ =
. + V̂ (x̂).
2m
Then, the time derivatives of the position and momentum operators in the Heisen-
berg picture are as follows:

d i i
. x̂H (t) = [Ĥ , x̂H (t)] = ÛH† (t) [Ĥ , x̂]ÛH (t)
dt h̄ h̄
i p̂2
= ÛH† (t) [ , x̂]ÛH (t)
h̄ 2m
p̂H (t)
= ,. (10.8)
m
d i i
p̂H (t) = [Ĥ , p̂H (t)] = ÛH† (t) [Ĥ , p̂]ÛH (t)
dt h̄ h̄
i
= ÛH† (t) [V̂ (x̂), p̂]ÛH (t)
h̄
= ÛH† (t) −V̂ ′ (x̂) ÛH (t) = F̂H (t). (10.9)

In the above expression, .V̂ ′ (x̂) implies the derivative of the potential energy with its
argument as the position operator and the fact that .[p̂, f (x̂)] = −i h̄f ′ (x̂) for any
function of position .f (x) has been used. Thus, .F̂H (t) is the time dependent force
operator in the Heisenberg picture.
10.1 Quantum Dynamics of Pure States 335

For the case of a harmonic oscillator with .V (x) = mω2 x 2 /2, the force operator
is proportional to position as follows: .F̂H (t) = −mω2 x̂H (t). Therefore, Eq. (10.9)
reduces to
d
. p̂H (t) = −mω2 x̂H (t). (10.10)
dt
This is exactly the same as the classical equation of motion for the momentum.
Therefore, one can find out the solution for the time dependent position operators in
terms of that for the classical mechanics as follows:

p̂
x̂H (t) = x̂ cos(ωt) +
. sin(ωt), . (10.11)
mω
p̂H (t) = p̂ cos(ωt) − mωx̂ sin(ωt). (10.12)

10.1.2 Interaction Picture and Time Dependent Perturbation

Theory

Let us consider the case where the system is subject to additional perturbation
Hamiltonian, .Ĥ1 (t), which can be time dependent in general, in addition to the
original zeroth order Hamiltonian, .Ĥ0 . Thus, the total Hamiltonian describing the
system is given by

.Ĥ (t) = Ĥ0 + Ĥ1 (t). (10.13)

The time dependence of .Ĥ1 (t) may originate from any source, but the most typical
one is the application of time dependent field.
In the Schrödinger picture, the state evolves according to the following time
dependent Schrödinger equation:

∂
i h̄
. |ψ; t = Ĥ (t)|ψ; t = (Ĥ0 + Ĥ1 (t))|ψ; t. (10.14)
∂t

Given that the eigenstates and eigenvalues of .Ĥ0 are known and that we want to
develop perturbation theory with respect to .Ĥ1 (t), it is convenient to consider a
Heisenberg picture with respect to .Ĥ0 . This is known as the interaction picture.
Thus, let us introduce the time evolution operator for .Ĥ0 ,
 
i
Û0 (t) = exp − Ĥ0 t ,
. (10.15)
h̄
336 10 Quantum Dynamics of Pure and Mixed States

which represents the dynamics of a system subject to only the zeroth order
Hamiltonian .Ĥ0 . Note that .Û0 (t) is a unitary operator as follows:

Û0 (t)Û0† (t) = Û0† (t)Û0 (t) = 1̂.

. (10.16)

We can now define the state in the interaction picture by applying .Û0† (t) to .|ψ; t
as

|ψ; tI ≡ Û0† (t)|ψ; t.

. (10.17)

Then, the equation governing the time evolution of .|ψ; tI is

 
∂ ∂ † ∂
.i h̄ |ψ; tI = i h̄ Û (t) |ψ; t + Û0† (t)i h̄ |ψ; t
∂t ∂t 0 ∂t

= −Û0† (t)Ĥ0 |ψ; t + Û0† (t) Ĥ0 + Ĥ1 (t) |ψ; t

= Û0† (t)Ĥ1 (t)|ψ; t

= Ĥ1,I (t)|ψ; tI , (10.18)

where Eqs. (10.16) and (10.17) have been used and the following first order
Hamiltonian in the interaction picture has been introduced.

Ĥ1,I (t) ≡ Û0† (t)Ĥ1 (t)Û0 (t).

. (10.19)

Integrating Eq. (10.18) with respect to time, we find that

i t
|ψ; tI = |ψ; 0I −
. dt ′ Ĥ1,I (t ′ )|ψ; t ′ I . (10.20)
h̄ 0

The above expression is nothing but the integral equation form equivalent to
Eq. (10.18) but is amenable for approximate solution more easily as detailed below.
Both sides of Eq. (10.20) involve as yet unknown .|ψ; tI at two different times.
However, for the state in the integrand, we can substitute the definition given by the
lefthand side as follows:

i t
|ψ; tI = |ψ; 0I −
. dt ′ Ĥ1,I (t ′ )|ψ; 0I
h̄ 0

i 2
 t t′
+ − dt ′ dt ′′ Ĥ1,I (t ′ )Ĥ1,I (t ′′ )|ψ; t ′′ I . (10.21)
h̄ 0 0

The above expression does not look simple either and appears to make getting
the solution even more complicated because the unknown .|ψ; t ′′ I still exists on
the righthand side, which now involves double integration with respect to time.
10.1 Quantum Dynamics of Pure States 337

However, given that the magnitude of .Ĥ1 (t) is small enough so that its effect
becomes smaller and smaller as the order of integration increases, one can continue
applying the iteration procedure and obtain the following formal solution:

i t
|ψ; tI = |ψ; 0I −
. dt ′ Ĥ1,I (t ′ )|ψ; 0I
h̄ 0

i 2
 t t′
+ − dt ′ dt ′′ Ĥ1,I (t ′ )Ĥ1,I (t ′′ )|ψ; 0I
h̄ 0 0

i 3
 t t′ t ′′
+ − dt ′ dt ′′ dt ′′′ Ĥ1,I (t ′ )Ĥ1,I (t ′′ )Ĥ1,I (t ′′′ )|ψ; 0I
h̄ 0 0 0
+··· (10.22)

Equation (10.22) is a general formal solution for the state in the interaction
picture and is well defined as long as the infinite order limit of the sum of
multiple time integrations converges. Finding out the exact condition for such
convergence to be ensured is not a simple task in general. However, in many cases,
the convergence can be determined at least numerically given a specific form of
the Hamiltonian. Here, we consider only physically well-defined perturbations for
which the convergence of perturbation terms is guaranteed.
t
If the effect of . 0 dt ′ Ĥ1,I (t ′ )/h̄ is very small compared to the zeroth order term,
an approximation that keeps only the first order term in Eq. (10.22) is reasonable.
Thus, we obtain the following first order approximation:

i t
(1)
|ψ; tI = |ψ; 0I −
. dt ′ Ĥ1,I (t ′ )|ψ; 0I . (10.23)
h̄ 0

Employing the definitions of Eqs. (10.17) and (10.19), and returning to the original
Schrödinger picture, we find that the above expression is equivalent to

i t i ′
|ψ; t(1) = |ψ; t(0) −
. dt ′ e− h̄ Ĥ0 (t−t ) Ĥ1 (t ′ )|ψ; t ′ (0) , (10.24)
h̄ 0

where
i
|ψ; t(0) = Û0 (t, 0)|ψ; 0 = e− h̄ Ĥ0 t |ψ; 0.
. (10.25)

10.1.3 Fermi’s Golden Rule

Suppose we are interested in the rate of transition of the system from an eigenstate
of .Ĥ0 to another one due to the perturbation .Ĥ1 (t). Thus, let us consider the case
where the state of the system at .t = 0 is an eigenstate of .Ĥ0 with eigenvalue .Ej ,
338 10 Quantum Dynamics of Pure and Mixed States

which we denote as .|Ej . Then, the first order approximation for the state at time t,
Eq. (10.24), can be expressed as

i i t i ′ i ′
|ψ; t(1) = e− h̄ Ej t |Ej  −
. dt ′ e− h̄ Ĥ0 (t−t ) Ĥ1 (t ′ )e− h̄ Ej t |Ej . (10.26)
h̄ 0

Let us denote the probability for the system at time t to have made transition to
another eigenstate .|Ek  of .Ĥ0 as .Pj →k (t). Without losing generality, we can assume
that .Ej |Ek  = 0. Up to the lowest order of .Ĥ1 (t), this probability can be calculated
by projecting the above first order time dependent perturbation approximation onto
.|Ek  and taking its absolute square as follows:

t 2
2 i i ′ i ′
Pj →k (t) = Ek |ψ; t(1)
. = − dt ′ e− h̄ Ek (t−t ) Ek |Ĥ1 (t ′ )|Ej e− h̄ Ej t
h̄ 0

1 t t i ′ ′′
= dt ′ dt ′′ e h̄ (Ek −Ej )(t −t ) Ek |Ĥ1 (t ′ )|Ej Ej |Ĥ1 (t ′′ )|Ek .
h̄2 0 0
(10.27)

The time derivative of the above probability can be defined as the rate of transition,
which is in general time dependent and is given by

d
Ŵj →k (t) ≡
. Pj →k (t)
dt
 t
1 i ′′
= 2 dt ′′ e h̄ (Ek −Ej )(t−t ) Ek |Ĥ1 (t)|Ej Ej |Ĥ1 (t ′′ )|Ek 
h̄ 0
t 
′ h̄i (Ek −Ej )(t ′ −t) ′
+ dt e Ek |Ĥ1 (t )|Ej Ej |Ĥ1 (t)|Ek  . (10.28)
0

In the integrals of the above expression, let us replace .t ′′ with .t − τ in the first term
and .t ′ with .t − τ in the second term. Then,

1
 t i
Ŵj →k (t) =
. dτ e h̄ (Ek −Ej )τ Ek |Ĥ1 (t)|Ej Ej |Ĥ1 (t − τ )|Ek 
h̄2 0
t i

+ dτ e− h̄ (Ek −Ej )τ Ek |Ĥ1 (t − τ )|Ej Ej |Ĥ1 (t)|Ek 
0

1 t i

= 2Re dτ e h̄ (Ek −Ej )τ Ek |Ĥ1 (t)|Ej Ej |Ĥ1 (t − τ )|Ek  .
h̄2 0
(10.29)

In the above expression, the fact that the first and second integrals in the first equality
are complex conjugates of each other have been used and “Re” implies the real part.
10.1 Quantum Dynamics of Pure States 339

Let us now consider the case where the matrix element of the perturbation
Hamiltonian is given by

Ek |Ĥ1 (t)|Ej  = Jj k e−iωt ,

. (10.30)

where .ω is a real number. Then, Eq. (10.29) can be expressed as

2
 t i

Ŵj →k (t) =
. Re dτ e h̄ (Ek −Ej )τ Jj k e−iωt Jj∗k eiω(t−τ )
h̄2 0

|2 t
 
2|Jj k i
= Re dτ e h̄ (Ek −Ej )τ e−iωτ , (10.31)
h̄2 0

where the fact that .Ej |Ĥ1 (t − τ )|Ek  is the complex conjugate of Eq. (10.30) with
t − τ as time argument has been used. In the limit of .t → ∞, the above rate
.

approaches the following steady state limit:

2|Jj k |2
 ∞ 
Ek − Ej
 
Ŵj →k (∞) =
. Re dτ exp i −ω τ
h̄2 0 h̄
|Jj k |2 ∞  
Ek − Ej
 
= dτ exp i −ω τ
h̄2 −∞ h̄
2π |Jj k |2
= δ(Ek − Ej − h̄ω), (10.32)
h̄

where the Fourier integral definition of the delta function, Eq. (1.85), has been used.
Equation (10.32) is the simplest form of the Fermi’s golden rule (FGR), and
serves as the basis for calculating transition rates for many dynamical processes
and for calculating intensities of spectroscopic transitions. Although this expression
involves a delta function, which is singular, a physically meaningful answers can be
obtained from this if the implication of taking the mathematical limit of .t → ∞
is taken into consideration within the proper physical context. In addition, the final
states to which the transition occurs are hardly a single discrete state in many real
situations. This can also be used to remove the singularity.
Let us consider the case where .Ĥ1 (t) represents the interaction of an electric
field and a molecule with transition dipole .D j k . In this case, within the dipole
approximation1 and the rotating wave approximation,2 the perturbation Hamiltonian

1 This is to assume that there is no spatial variance of the filed at molecular length scale. See
Appendix of this chapter for more details.
2 This is to ignore highly oscillating non-resonant term between field and transition. See Appendix

of this chapter for more details.

340 10 Quantum Dynamics of Pure and Mixed States

can be expressed as

Ĥ1 (t) = A(D j k · ue )|Ek Ej |e−iωt + A∗ (D j k · ue )∗ |Ej Ek |eiωt ,

. (10.33)

where .ω is the frequency of the radiation, and .A and .ue are the amplitude and unit
polarization vector of its electric field component. For this case, Eq. (10.32) reduces
to

2π |A|2 |D j k · ue |2
Ŵjs k (ω) =
. δ(Ek − Ej − h̄ω), (10.34)
h̄

which gives the intensity of the transition at frequency .ω. In this expression, .δ(Ek −
Ej − h̄ω) can be viewed as an idealization of a narrow peak centered at .ω = (Ek −
Ej )/h̄.
Let us consider another common case where the FGR rate expression,
Eq. (10.32), leads to a more well-defined rate. Assume that the final states are
not in fact a single state but a set of states and assume also that .ω = 0 in Eq. (10.30)
for simplicity. The FGR rate in this case becomes the sum of all rates given by
Eq. (10.32) over all possible final states as follows:

j →S  2π |Jj k |2
ŴF GRf =
. δ(Ek − Ej ). (10.35)
h̄
k∈Sf

Let us also assume that the final states form a continuum with distribution
wf (Ef , ξf ) such that
.


. = dEk dξf wf (Ek , ξf ), (10.36)
k∈Sf

where .ξf collectively represents the set of additional quantum numbers needed
to completely specify each of the final states in the set .Sf . Employing the above
relationship in Eq. (10.35) and then conducting the integration over .Ek , we then
obtain the following rate expression:

j →S 2π |Jj k (ξf )|2

ŴF GRf =
. dEk dξf wf (Ek , ξf ) δ(Ek − Ej )
h̄
2π |Jj k (ξf )|2
= dξf wf (Ej , ξf ), (10.37)
h̄

where we have included the general case where .Jj k depends on .ξf . In a simple case
where .Jj k is independent of .ξf , which is now denoted as .Jjf , the above expression
10.2 Quantum Dynamics of Mixed Quantum States 341

can be simplified further to

j →S 2π |Jjf |2
ŴF GRf =
. ρf (Ej ), (10.38)
h̄

where .ρf (Ej ) = dξf wf (Ej , ξf ) is the final density of states at .Ej .

10.2 Quantum Dynamics of Mixed Quantum States

So far, we have assumed that a complete information on the quantum state of the
system is available. This is hardly the case in reality. In addition to the intrinsic
quantum uncertainty, there always exist additional uncertainties at practical level
due to imprecision, disorder, and lack of full information on conditions determining
the states. Systems of these characteristics are in mixed quantum states and need to
be treated at the level of a density operator.

10.2.1 Density Operator and Quantum Liouville Equation

The concept of the density operator is more general than quantum states we
considered so far and requires both ket and bra for its definition. Any pure quantum
state can be represented by a density operator. Consider a quantum state .|ψ. Then,
the density operator corresponding to this state is as follows:

ρ̂ψ = |ψψ|.
. (10.39)

If the system is governed by a Hamiltonian .Ĥ , then the state prepared in .|ψ at time
.t = 0 evolves to become .|ψ; t = ÛH (t)|ψ at time t. For the definition of the
density operator to be consistent at all time, given that it is defined as above at time
.t = 0, it should at time t become

.ρ̂ψ (t) = |ψ; tψ; t| = ÛH (t)|ψψ|ÛH† (t). (10.40)

Taking the time derivative of the above density operator, then we find that

∂ i i
. ρ̂ψ (t) = − Ĥ ÛH (t)|ψψ|ÛH† (t) + ÛH (t)|ψψ|ÛH† (t)Ĥ
∂t h̄ h̄
i
= − [Ĥ , ρ̂ψ (t)]. (10.41)
h̄
342 10 Quantum Dynamics of Pure and Mixed States

The above equation is an example of the quantum Liouville equation, which can be
shown to be valid for any density operator.
Now, let us consider an example of a mixed quantum state. Assume that we know
the system is either in state .|ψ1  or .|ψ2 , but do not know exactly which state the
system is in. The only information available is that their respective probabilities are
.p1 and .p2 . This kind of situation is completely different from the case where the

state is given by a linear combination of .|ψ1  and .|ψ2  that is still a pure state, and
can be represented by the following density operator:

ρ̂ = p1 |ψ1 ψ1 | + p2 |ψ2 ψ2 |,

. (10.42)

where it is assumed that .|ψ1  and .|ψ2  are normalized. In the above expression,
p1 + p2 = 1, while .p1 , p2 ≥ 0. Note that the two states .|ψ1  and .|ψ2  do not have
.

to be orthogonal to each other. Since the possibilities for the system to be in either
of these two states are completely uncorrelated, the above density operator should
at time t become

ρ̂(t) = p1 |ψ1 ; tψ1 ; t| + p2 |ψ2 ; tψ2 ; t|

.

= p1 ÛH (t)|ψ1 ψ1 |ÛH† (t) + p2 ÛH (t)|ψ2 ψ2 |ÛH† (t)

= ÛH (t)ρ̂ ÛH† (t). (10.43)

The above expression suggests that the manner of time evolution for the density
operator is independent of whether it represents the pure or mixed state as long as
the governing Hamiltonian is the same.
Extending the concept of the mixed state illustrated above to more general cases
that involve more than two states, we can now define a general density operator,

ρ̂ =
. pj |ψj ψj |, (10.44)
j

where each .|ψj  is normalized

 and .pj ≥ 0 represents the probability for the system
to be in state .|ψj . Thus, . j pj = 1. It is obvious that the density operator
defined in this way is always Hermitian. In addition, for any state .|ϕ, .ϕ|ρ̂|ϕ is
always nonnegative. This means that the diagonal elements of any density operator
in any basis, which are probabilities to find the system in those states, are always
nonnegative.
For the density operator given by Eq. (10.44) at time .t = 0, the density operator
at time t becomes

.ρ̂(t) = ÛH (t)ρ̂ ÛH† (t), (10.45)

10.2 Quantum Dynamics of Mixed Quantum States 343

which satisfies the following quantum Liouville equation:

∂ i  
. ρ̂(t) = − Ĥ , ρ̂(t) ≡ −i L̂ρ̂(t). (10.46)
∂t h̄

The second equality in the above equation serves as the definition of .L̂, a super-
operator3 called quantum Liouvillian.
Given the mixture of states .|ψj , each with probability .pj , the expectation value
of any physical observable . should be

 =
. pj ψj |Â|ψj . (10.47)
j

It turns out that the above quantity can be calculated in terms of the density operator
ρ̂ given by Eq. (10.44) as follows:
.

 
 = T r Âρ̂ ,
. (10.48)

where .T r{· · · } implies sum of all the diagonal elements of .· · · calculated for any
complete basis. More detailed description of the trace operation and its properties
are provided below.

Definition and Invariance of Trace

Given an operator . and a complete basis, .|φn , with respect to which the
operator can be represented, the trace of . is defined as the sum of all the
diagonal elements as follows:
  
T r  =
. φn |Â|φn . (10.49)
n

There are two important propertiesof trace operation that can be proved easily
using the completeness relation, . n |φn φn | = 1̂.
• The trace of the product of two operators, . and .B̂, is independent of the
order as follows.
   
.T r ÂB̂ = T r B̂ Â . (10.50)

(continued)

3 Super-operator refers to an operator that operates on quantum mechanical operators. In this case,
note that .L operates on the density operator.
344 10 Quantum Dynamics of Pure and Mixed States

The proof of this is as follows:

  
T r ÂB̂ =
. φn |ÂB̂|φn 
n

= φn |Â|φm φm |B̂|φn 
n m

= φm |B̂|φn φn |Â|φm 
n m
  
= φm |B̂ Â|φm  = T r B̂ Â , (10.51)
m

where the fact that .|φn  form a complete basis has been used in the second
and the fourth equalities.
• Given an operator . and its unitary transformation .Â′ = Û † ÂÛ , the traces
of . and .Â′ are always the same. This is a direct outcome of Eq. (10.50) as
shown below.

T r{Â′ } = T r{Û † ÂÛ } = T r{ÂÛ Û † } = T r{Â},

. (10.52)

where the fact that .Û Û † = 1̂ has been used.

Having defined the trace operation and its two invariance properties, we can
easily prove the equivalence of Eqs. (10.47) and (10.48) as follows:
  
T r Âρ̂ =
. φn |Âρ̂|φn 
n

= pj φn |Â|ψj ψj |φn 
n j

= pj ψj |φn φn |Â|ψj 
n j
  
= pj ψj |Â|ψj  = T r ρ̂ Â , (10.53)
j

   
where the last equality results from Eq. (10.50). Also note that .T r ρ̂(t) = T r ρ̂
for .ρ̂(t) given by Eq. (10.45).
In addition to the fact that the diagonal elements of a density operator are
always nonnegative, there are two important properties concerning traces of density
operators as described below.
10.2 Quantum Dynamics of Mixed Quantum States 345

Traces of Density and Squared Density Operators

For any density operator, .ρ̂, the following two properties hold true.
 
. T r ρ̂ = 1, . (10.54)
 
0 ≤ T r ρ̂ 2 ≤ 1. (10.55)

The proof of Eq. (10.54) is straightforward and results from that the sum of
all the probabilities is equal to one as follows.
   
T r ρ̂ =
. φn |ρ̂|φn  = pj φn |ψj ψj |φn 
n n j
 
  
= pj ψj | |φn φn | |ψj  = pj = 1, (10.56)
j n j

where the fact that .|ψj ’s are normalized and that .|φn ’s form a complete basis
has been used.
The proof of Eq. (10.55) is more involved but is straightforward as shown
below.
  
2
.T r ρ̂ = φn |ρ̂ ρ̂|φn 
n

= pj pk φn |ψj ψj |ψk ψk |φn 
n j k
 
 
= pj pk ψj |ψk  ψk |φn φn |ψj 
j k n

2
 
= pj pk ψj |ψk  ≤ pj pk = 1, (10.57)
j k j k

2
where the fact that . ψj |ψk  ≤ 1 has been used. Note that the equality is
satisfied only for pure states. The above expression is obviously nonnegative
since each term
 in the summation is nonnegative. This also proves the fact
that .0 ≤ T r ρ̂ 2 .
346 10 Quantum Dynamics of Pure and Mixed States

The necessity to introduce a mixed state goes beyond any experimental imper-
fection, but has also more fundamental quantum mechanical origin. Mixed states
can emerge for any system that cannot be completely decoupled from the rest of
the universe even though there is no direct coupling of the system Hamiltonian
and that of the rest [56]. For this, let us consider the direct product space of the
system, for which .|φn ’s form a complete basis, and the rest of the universe, for
which we assume the existence of a complete orthonormal basis denoted as .|Xγ ’s.
Since the entire universe is closed, any state defined in the direct product space can
be expressed as the following linear combination:

| =
. Cnγ |φn  ⊗ |Xγ . (10.58)
n γ

Let us assume that we are interested in calculating the expectation value of an

operator .Â, which is defined in the system space only. Then,

∗
 =
. Cnγ Cn′ γ ′ φn |Â|φn′ Xγ |Xγ ′ 
n n′ γ γ′

∗
= Cnγ Cn′ γ φn |Â|φn′ 
n n′ γ
⎞⎛
 
∗
= ⎝ Cnγ Cn′ γ ⎠ φn |Â|φn′ , (10.59)
n n′ γ

where .Xγ |Xγ ′  = δγ γ ′ has been used in the second equality. The above average in
general cannot be expressed in terms of a pure state defined as a linear combination
of .|φn ’s only. On the other hand, one can employ the following density operator:
⎛ ⎞
 
∗
.ρ̂ = ⎝ Cnγ Cn′ γ ⎠ |φn′ φn |. (10.60)
n′ n γ

The above operator satisfies all the properties of a density operator. It is obvious that
this operator is Hermitian and has nonnegative diagonal elements in any basis. The
trace of this is equal to one as follows:
⎛ ⎞
   
.T r ρ̂ = ⎝ |Cj γ |2 ⎠ = 1, (10.61)
n γ
10.2 Quantum Dynamics of Mixed Quantum States 347

which is satisfied because .| given by Eq. (10.58) is normalized. On the other hand,
⎧ ⎛ ⎞⎛ ⎞ ⎫
  ⎨     ⎬
∗
.T r ρ̂
2
= Tr ⎝ Cnγ Cn ′ γ ⎠ ⎝ Cn∗′′ γ ′ Cnγ ′ ⎠ |φn′ φn′′ |
⎩ ′ n ′′ γ ′
⎭
n n γ
2
 
∗
= Cnγ Cn ′ γ , (10.62)
n′ n γ

which can be shown to be always less than or equal to one.

Thus, .ρ̂ given by Eq. (10.60) in general represents a mixed state created due to
an entanglement of the system with other environmental degrees freedom, for which
no explicit control or specification through measurement is possible. On the other
  to show that the average value given by Eq. (10.59) is
hand, it is straightforward
indeed equal to .T r Âρ̂ . In other words, the definition of average, Eq. (10.48), is
general and can be used even when the average value cannot be expressed as an
expectation value with respect to pure system states.
An important class of density operators is called the canonical equilibrium
density operator, which is defined as follows:

e−β Ĥ
ρ̂β =
. , (10.63)
Zβ

where .Ĥ is the Hamiltonian of the system, .β = 1/(kB T ), with .kB being the
Boltzmann constant, and .Zβ is the partition function of the canonical ensemble
defined as

−β Ĥ
.Zβ = T r{e }= e−βEj . (10.64)
j

In the above expression, .Ej ’s are eigenvalues of the Hamiltonian .Ĥ . By definition,
T r{ρ̂β } = 1. In addition, it is stationary with respect to time since it commutes with
.

the time evolution operator. For any physical observable, .Â, the equilibrium average
over the canonical ensemble is given by
 
Âβ = T r Âρ̂β .
. (10.65)
348 10 Quantum Dynamics of Pure and Mixed States

10.2.2 Time Dependent Perturbation Theory for Mixed

Quantum States

Let us consider again the case where the total Hamiltonian .Ĥ (t) is given by the sum
of the zeroth order and time dependent first order terms such as Eq. (10.13), and
denote the time evolution operator for this time dependent Hamiltonian as .ÛH (t).
Given the density operator .ρ̂(0) at time .t = 0, the density operator at time t becomes

ρ̂(t) = ÛH (t)ρ̂(0)ÛH† (t).

. (10.66)

On the other hand, the time dependent state at time t in the interaction picture with
respect to .Ĥ0 is

|ψ; tI = Û0† (t)|ψ; t = Û0† (t)ÛH (t)|ψ; 0.

. (10.67)

Let us combine this with Eq. (10.22) and note the fact that the initial state, .|ψ; 0 =
|ψ; 0I , is general. Therefore, we obtain the following identity:

i t 
.ÛH (t) = Û0 (t) exp(+) − dt ′ Ĥ1,I (t ′ ) , (10.68)
h̄ 0

where the second term in the above expression is a time-ordered exponential

operator defined as

i t ′ i t ′
  
. exp(+) − dt Ĥ1,I (t ′ ) ≡ 1 − dt Ĥ1,I (t ′ )
h̄ 0 h̄ 0
′
i 2 t ′ t ′′
 
+ − dt dt Ĥ1,I (t ′ )Ĥ1,I (t ′′ )
h̄ 0 0
t′ t ′′
 3 t 
i ′ ′′ ′′′ ′ ′′ ′′′
+ − dt dt dt Ĥ1,I (t )Ĥ1,I (t )Ĥ1,I (t ) + · · · .
h̄ 0 0 0

(10.69)

The subscript .(+) in the above expression implies positive or chronological time
ordering.
Employing the definition of .Ĥ1,I (t) given by Eq. (10.19) in Eq. (10.68), we
obtain the following expression for the time evolution operator:

ÛH (t) = e−i Ĥ0 t/h̄

.

i t ′ ′
− dt ′ e−i Ĥ0 (t−t )/h̄ Ĥ1 (t ′ )e−i Ĥ0 t /h̄
h̄ 0
10.2 Quantum Dynamics of Mixed Quantum States 349


i 2
 t t′ ′ ′ ′′ )/h̄ ′′ /h̄
+ − dt ′ dt ′′ e−i Ĥ0 (t−t )/h̄ Ĥ1 (t ′ )e−i Ĥ0 (t −t Ĥ1 (t ′′ )e−i Ĥ0 t
h̄ 0 0
+··· . (10.70)

Employing the above expression and its Hermitian conjugate in Eq. (10.66) and
keeping only up to the second order terms with respect to .Ĥ1 (t), we thus obtain
the following second order approximation for the time dependent density operator:

ρ̂ (2) (t) = e−i Ĥ0 t/h̄ ρ̂(0)ei Ĥ0 t/h̄

.

i t ′ ′
− dt ′ e−i Ĥ0 (t−t )/h̄ Ĥ1 (t ′ )e−i Ĥ0 t /h̄ ρ̂(0)ei Ĥ0 t/h̄
h̄ 0
i t ′ ′
+ dt ′ e−i Ĥ0 t/h̄ ρ̂(0)ei Ĥ0 t /h̄ Ĥ1 (t ′ )ei Ĥ0 (t−t )/h̄
h̄ 0
1 t t ′ ′
+ dt ′ dt ′′ e−i Ĥ0 (t−t )/h̄ Ĥ1 (t ′ )e−i Ĥ0 t /h̄ ρ̂(0)
h̄2 0 0
′′ /h̄ ′′ )/h̄
×ei Ĥ0 t Ĥ1 (t ′′ )ei Ĥ0 (t−t
1 t t′ ′ ′ ′′ )/h̄
− dt ′ dt ′′ e−i Ĥ0 (t−t )/h̄ Ĥ1 (t ′ )e−i Ĥ0 (t −t
h̄2 0 0
′′ /h̄
×Ĥ1 (t ′′ )e−i Ĥ0 t ρ̂(0)ei Ĥ0 t/h̄
1 t t′ ′′ /h̄
− dt ′ dt ′′ e−i Ĥ0 t/h̄ ρ̂(0)ei Ĥ0 t Ĥ1 (t ′′ )
h̄2 0 0
′ ′′ )/h̄ ′
×ei Ĥ0 (t −t Ĥ1 (t ′ )ei Ĥ0 (t−t )/h̄ . (10.71)

For any physical observable .Â, then its average value at time t including the
effects of the perturbation up to the second order can be expressed as
   
T r Âρ̂ (2) (t) = T r ei Ĥ0 t/h̄ Âe−i Ĥ0 t/h̄ ρ̂(0)
.

i t  ′ ′

− dt ′ T r ei Ĥ0 t/h̄ Âe−i Ĥ0 (t−t )/h̄ Ĥ1 (t ′ )e−i Ĥ0 t /h̄ ρ̂(0)
h̄ 0
i t  ′ ′

+ dt ′ T r ei Ĥ0 t /h̄ Ĥ1 (t ′ )ei Ĥ0 (t−t )/h̄ Âe−i Ĥ0 t/h̄ ρ̂(0)
h̄ 0
1 t t  ′′ ′′
+ dt ′ dt ′′ T r ei Ĥ0 t /h̄ Ĥ1 (t ′′ )ei Ĥ0 (t−t )/h̄ Â
h̄2 0 0
′ ′

×e−i Ĥ0 (t−t )/h̄ Ĥ1 (t ′ )e−i Ĥ0 t /h̄ ρ̂(0)
350 10 Quantum Dynamics of Pure and Mixed States

1 t t′  ′
− dt ′ dt ′′ T r ei Ĥ0 t/h̄ Âe−i Ĥ0 (t−t )/h̄ Ĥ1 (t ′ )
h̄2 0 0
′ ′′ )/h̄ ′′ /h̄

×e−i Ĥ0 (t −t Ĥ1 (t ′′ )e−i Ĥ0 t ρ̂(0)

1 t t′  ′′ ′ ′′
′
− dt dt ′′ T r ei Ĥ0 t /h̄ Ĥ1 (t ′′ )ei Ĥ0 (t −t )/h̄ Ĥ1 (t ′ )
h̄2 0 0
′

×ei Ĥ0 (t−t )/h̄ Âe−i Ĥ0 t/h̄ ρ̂(0) , (10.72)

where we have used the identity, Eq. (10.50), repeatedly to send the density operator
at time .t = 0 to the rightmost place within the trace.

10.2.3 FGR for Mixed States

As an application of Eq. (10.72), let us consider a simple but widely used model.
Assume that there is a two-state system spanned completely by .|1 and .|2, which
form an orthonormal basis for the system. This system is coupled to all other degrees
of freedom denoted as “bath.” At the zeroth order level, this bath is independently
coupled to each system state and does not couple them. Thus, the zeroth order
Hamiltonian for the system plus bath is assumed to be

Ĥ0 = (E1 + B̂1 )|11| + (E2 + B̂2 )|22| + Ĥb ,

. (10.73)

where .B̂1 and .B̂2 are bath operators representing the coupling of each state, .|1 or
|2, to the bath and .Ĥb is the Hamiltonian for the bath only.
.

Examples of the two system states are spin states of any spin .1/2 particle or two
electronic states of interest in any molecular system. The total quantum state space
is a direct product of the two system state space and that of the bath. Thus, the trace
operation of an operator .Ô that depends on both system and bath can be expressed
as
   
.T r Ô = T rb 1|Ô|1 + 2|Ô|2
   
= 1|T rb Ô |1 + 2|T rb Ô |2. (10.74)

Let us assume that a small time dependent perturbation coupling the two system
states is turned on at time .t = 0. Thus, for .t ≥ 0,
†
. Ĥ1 (t) = Jˆ12 (t)|12| + Jˆ12 (t)|21|, (10.75)
10.2 Quantum Dynamics of Mixed Quantum States 351

where .Jˆ(t) is time dependent (in general) and can also depend on the bath degrees
of freedom. Thus, .Jˆ(t) is assumed to be an operator in the bath space.
For .t ≥ 0, the total Hamiltonian becomes .Ĥ (t) = Ĥ0 + Ĥ1 (t). Let us assume that
the total system plus the bath, before the perturbation Hamiltonian .Ĥ1 (t) is turned
on, is represented by the following density operator.

ρ̂(0) = |11|ρ̂b ,
. (10.76)

where .ρ̂b is a bath density operator that can be a canonical equilibrium density
operator or any other nonequilibrium density operator. Thus, the above density
operator represents the system prepared in the pure state .|1 and the bath in some
kind of mixed states as prescribed by .ρ̂b .
Suppose that the major quantity of interest is the population of the state .|2 of
system at time t. This can be calculated, up to the second order of perturbation,
by using . = |22| in Eq. (10.72). Although there are many terms in Eq. (10.72),
for the model and process considered here, most of the terms disappear because
.2|e
±i Ĥ0 t/h̄ |1 = 1|e±i Ĥ0 t/h̄ |2 = 0. Thus, Eq. (10.72) simplifies to

 
(2)
P2 (t) = T r |22|ρ̂ (2) (t)
.

1 t t  ′′ ′′
= dt ′ dt ′′ T r ei Ĥ0 t /h̄ Ĥ1 (t ′′ )ei Ĥ0 (t−t )/h̄ |22|
h̄2 0 0
′ ′

×e−i Ĥ0 (t−t )/h̄ Ĥ1 (t ′ )e−i Ĥ0 t /h̄ ρ̂(0)

1 t t  ′′ ′′
= dt ′ dt ′′ T rb 1|ei Ĥ0 t /h̄ Ĥ1 (t ′′ )ei Ĥ0 (t−t )/h̄ |2
h̄2 0 0
′ ′

×2|e−i Ĥ0 (t−t )/h̄ Ĥ1 (t ′ )e−i Ĥ0 t /h̄ |1ρ̂b .

(10.77)

Since .Ĥ0 given by Eq. (10.73) is diagonal with respect to the two state system
states, it is easy to confirm that

e±i Ĥ0 t/h̄ = e±i(E1 +B̂1 +ĤB )t/h̄ |11| + e±i(E2 +B̂2 +ĤB )t/h̄ |22|.
. (10.78)

Thus, employing the above identity in all the expressions having .Ĥ0 in the exponent
in Eq. (10.77), we obtain

1 t t  ′′ ′′
P2(2) (t) =
. dt ′ dt ′′ T rb ei(E1 +B̂1 +Ĥb )t /h̄ 1|Ĥ1 (t ′′ )|2ei(E2 +B̂2 +Ĥb )(t−t )/h̄
h̄2 0 0
′ ′

×e−i(E2 +B̂2 +Ĥb )(t−t )/h̄ 2|Ĥ1 (t ′ )|1e−i(E1 +B̂1 +Ĥb )t /h̄ ρ̂b
352 10 Quantum Dynamics of Pure and Mixed States

1 t t ′ ′′ )/h̄
 ′′
= dt ′ dt ′′ ei(E2 −E1 )(t −t T rb ei(B̂1 +Ĥb )t /h̄ Jˆ12 (t ′′ )
h̄2 0 0
′ ′′ )/h̄

† ′ −i(B̂1 +Ĥb )t ′ /h̄
×ei(B̂2 +Ĥb )(t −t Jˆ12 (t )e ρ̂b .

(10.79)

The second equality of the above expression is obtained by further rearrangement

of terms and using the definition of .Ĥ1 (t) given by Eq. (10.75) explicitly.
Now let us define the transition rate from state 1 to state 2 as the time derivative
of .P2(2) (t). Then,

d (2)
Ŵ1→2 (t) =
. P (t)
dt 2
1 t ′′ )/h̄
 ′′
= dt ′′ ei(E2 −E1 )(t−t T rb ei(B̂1 +Ĥb )t /h̄ Jˆ12 (t ′′ )
h̄2 0
′′ )/h̄

†
×ei(B̂2 +Ĥb )(t−t Jˆ12 (t)e−i(B̂1 +Ĥb )t/h̄ ρ̂b

1 t ′

+ dt ′ ei(E2 −E1 )(t −t)/h̄ T rb ei(B̂1 +Ĥb )t/h̄ Jˆ12 (t)
h̄2 0
′

† ′ −i(B̂1 +Ĥb )t ′ /h̄
×ei(B̂2 +Ĥb )(t −t)/h̄ Jˆ12 (t )e ρ̂b . (10.80)

Replacing the integrands .t − t ′′ and .t − t ′ to .τ respectively and noting the fact

that the two integrations above are complex conjugates of each other, we obtain the
following general expression:

2
 t 
Ŵ1→2 (t) =
.
2
Re dτ ei(E2 −E1 )τ/h̄ T rb ei(B̂1 +Ĥb )(t−τ )/h̄ Jˆ12 (t − τ )
h̄ 0

†
×ei(B̂2 +Ĥb )τ/h̄ Jˆ12 (t)e−i(B̂1 +Ĥb )t/h̄ ρ̂b . (10.81)

The rate expression given above is valid for any time and any form of .Jˆ(t) as long
as these are small enough to justify the application of the first order time dependent
perturbation theory. Thus, this can be used as the starting point for calculating rates
under various physical conditions.
First, let us assume that

Jˆ12 (t) = Jˆ12 eiωt ,

. (10.82)

where .Jˆ12 is time independent but is still assumed to be an operator in the bath
space. In addition, let us also assume that the bath density operator is the canonical
10.2 Quantum Dynamics of Mixed Quantum States 353

equilibrium density operator corresponding to the system state 1 as follows:

eq e−β(B̂1 +Ĥb )
. ρ̂b,1 =  , (10.83)
T rb e−β(B̂1 +Ĥb )

where .β = 1/(kB T ). Then, Eq. (10.81) simplifies further to

2
 t
Ŵ1→2 (t) =
.
2
Re dτ ei(E2 −E1 −ω)τ/h̄
h̄ 0
 
† eq
× T rb e−i(B̂1 +Ĥb )τ/h̄ Jˆ12 ei(B̂2 +Ĥb )τ/h̄ Jˆ12 ρ̂b,1 . (10.84)

Taking the limit of .t → ∞, the above rate becomes the following time independent
rate expression:

2
 ∞  
† eq
s
.Ŵ1→2 =
2
Re dt ei(E2 −E1 −ω)t/h̄ T rb e−i(B̂1 +Ĥb )t/h̄ Jˆ12 ei(B̂2 +Ĥb )t/h̄ Jˆ12 ρ̂b,1 ,
h̄ 0
(10.85)
where we have replaced the integration variable .τ with t .
Equation (10.85) is in fact a kind of the FGR rate expression. For the case
where the perturbation Hamiltonian corresponds to the interaction of two states of
a molecule with an electric field component of light and the Condon approximation
holds, .Jˆ12 = AD 12 · ue , where .D 12 is the corresponding transition dipole vector.
Equation (10.85) in this case reduces to the following lineshape expression for the
transition from state 1 to state 2:

s |A|2 |D 12 · ue |2
.I1→2 (ω) =
h̄2
∞  
eq
× dt ei(E2 −E1 −ω)t/h̄ T rb ei(B̂2 +Ĥb )t/h̄ e−i(B̂1 +Ĥb )t/h̄ ρ̂b,1 ,
−∞
(10.86)
eq
where we have used the fact that .ρb,1 commutes with .Ĥb + B̂1 and also made the
dependence of the rate on .ω explicit. For the case of absorption ( .E2 > E1 ), .ω
is positive. For the case of emission (.E2 < E1 ), .ω is negative, which represents
emission of photon with frequency .−ω. Equation (10.86) can be used for the
modeling of absorption and emission lineshapes of various electronic transitions
and is reliable as long as the two states involved in the transition are well separated
from others and assumptions of the model Hamiltonian are well justified.
354 10 Quantum Dynamics of Pure and Mixed States

Similarly, for the case where .ω = 0 and .Jˆ12 is not a vector but a time independent
constant, i.e., .J12 , Eq. (10.85) reduces to the following familiar FGR rate expression:

2
J12 ∞  
eq
1→2
.ŴF GR = dt ei(E2 −E1 )t/h̄ T rb ei(B̂2 +Ĥb )t/h̄ e−i(B̂1 +Ĥb )t/h̄ ρ̂b,1 . (10.87)
h̄2 −∞

Various rate processes involving transitions of quantum states such as the transfer
of electron, excitation energy, or proton can be described in terms of the above
FGR rate expression. Indeed, well known Marcus rate [33, 34] expression for the
electron transfer rate or the Förster rate [35] expression for excitation energy transfer
can be derived [36, 37] from this FGR expression with further approximations and
consideration of relevant physical conditions.

10.3 Summary and Questions

In the Schrödinger picture, the dynamics of a quantum system can be viewed as

that due to the time evolution of states. On the other hand, physical observables are
represented by time dependent operators unless explicit time dependence is needed
due to external factors. In the Heisenberg picture, the view is opposite and the time
dependence is carried through that of time dependent Heisenberg operators defined
by Eq. (10.5). Interaction picture is in-between the two pictures and assumes that
the operators evolve according to a zeroth or reference Hamiltonian while the states
evolve according to the remaining part that represents the key interaction term of
the Hamiltonian that is the major focus of the dynamics study. All of these pictures
are equivalent because of the unitarity of a time evolution operator.
Time dependent perturbation theory allows calculation of the effects of a
perturbation Hamiltonian on the time evolution of states and physical observables
and can be best expressed in the interaction picture where the state evolves in time
only due to the perturbation Hamiltonian term. Considering only the first order
term, one obtains a simple and general expression for the state in the interaction
picture, Eq. (10.23), and that in the Schrödinger picture, Eq. (10.24). Use of this
expression for the calculation of the transition probability between two states and
calculation of the transition rate as its time derivative in the stead state limit leads to
the FGR rate expression. This leads to the well-known expression for the intensity
of spectroscopic transition, Eq. (10.34). For the case of transitions to a collection
of final states with energy density .ρf (E), averaging of the FGR rate expression,
Eq. (10.35), leads to the final rate expression, Eq. (10.38).
A broad class of physical systems requires introduction of mixed states that
cannot be represented by a single ket (bra). This is because the system under study
is a collection or ensemble of states due to experimental limitations in creating
pure quantum states. More fundamentally, this can be due to the entanglement of
the system under study with other environmental degrees of freedom. This requires
introduction of a density operator, which can be expressed in general as Eq. (10.44),
Appendix: Interaction Hamiltonian in the Presence of Radiation 355

for which time evolution in both ket and bra space becomes necessary. The resulting
time evolution equation, Eq. (10.46), is known as quantum Liouville equation. The
average of any physical observable for the density operator is given by Eq. (10.48),
and any density operator is Hermitian and satisfies Eqs. (10.54) and (10.55).
Time dependent perturbation theory can be extended to the time evolution of
the density operator by employing perturbation expansions for both time dependent
kets and bras constituting the density operator. The resulting lowest order terms
representing transitions between states now become second order with respect to
the perturbation. Calculation of the time derivative of the probability for the system
to be in a certain final state of interest and consideration of its steady state limit,
results in a general expression for the FGR rate, Eq. (10.85). This expression is valid
for a general mixed state given by Eq. (10.76) as the initial condition. The lineshape
expression, Eq. (10.86), and a well-known FGR rate expression, Eq. (10.87), can
then be obtained with additional assumptions and approximations.

Questions

• What are major differences between Schrödinger and Heisenberg pictures and
why are the two pictures equivalent in describing quantum dynamics?
• What makes consideration in the interaction picture useful?
• Can one apply the time dependent perturbation theory even when the perturba-
tion is time independent?
• What is the consequence of applying the time dependent perturbation theory
when the initial states are degenerate?
• What are conditions and assumptions validating the use of the first order
perturbation theory?
• Is the Fermi’s golden rule rate always well defined and finite?
• Does an ensemble of quantum states always require description at the level of
a density operator?
• What are physical origins of mixed states?
• What is the Condon approximation in lineshape calculation? What is its utility
and what happens if the approximation breaks down?

Appendix: Interaction Hamiltonian in the Presence

of Radiation

This section provides a brief overview of the classical theory of electromagnetic

fields [38] and then outlines how the interaction between molecules and the
classical radiation field can be treated. Thus, the approach being described here
is semiclassical because only molecules are considered quantum mechanically
356 10 Quantum Dynamics of Pure and Mixed States

while the electromagnetic field is simply viewed as time dependent classical

external potential. This approach is valid and practical for the description of simple
spectroscopic experiments.
Classically, the Hamiltonian for a particle with mass m and charge q and its
interaction with electric and magnetic fields, .E(r, t) and .B(r, t), in Gaussian units,
can be expressed as [37, 38]

1 q 2
.H = p − A(r, t) + q(r, t) + V (r), (10.88)
2m c

where .V (r) is the potential applied to the particle in the absence of the electromag-
netic radiation and .A(r, t) and .(r, t) are vector and scalar potentials that are related
to the electric and magnetic fields, .E(r, t) and .B(r, t), as follows:

1 ∂A(r, t)
. E(r, t) + = −∇(r, t).. (10.89)
c ∂t
B(r, t) = ∇ × A(r, t). (10.90)

It is straightforward to show through vector calculus that Eq. (10.88) results in the
following equation of motion [37, 38]:

d 2r
 
1 ∂A(r, t) q dr
.m = q − − ∇(r, t) + × {∇ × A(r, t)}
dt 2 c ∂t c dt
− ∇V (r). (10.91)

Inserting Eqs. (10.89) and (10.90) into the above equation, one can show that it is
equivalent to the following equation of motion that is more well-known:

d 2r
 
1 dr
.m = q E(r, t) + × B(r, t) − ∇V (r) . (10.92)
dt 2 c dt

The definitions of electric and magnetic fields, Eqs. (10.89) and (10.90), leave
great degree of flexibility in choosing the vector and scalar potentials because, for an
arbitrary function .f (r, t), the following replacement does not cause any difference
in the electric and magnetic fields.

. A′ (r, t) = A(r, t) + ∇f (r, t), . (10.93)

1 ∂f (r, t)
′ (r, t) = (r, t) − . (10.94)
c ∂t
Appendix: Interaction Hamiltonian in the Presence of Radiation 357

In other words,

. B(r, t) = ∇ × A′ (r, t), . (10.95)

1 ∂A′ (r, t)
E(r, t) + = −∇′ (r, t). (10.96)
c ∂t

Equations (10.93) and (10.94) are called gauge transformation, and provide ways to
choose appropriate vector and scalar potentials that are easy to work with while not
changing the electric and magnetic fields. One important gauge is called Coulomb
gauge, which is characterized by the following condition:

.∇ · A(r, t) = 0. (10.97)

The gauge transformation, Eqs. (10.93) and (10.94), makes it always possible to
find a vector potential satisfying the above property. Let us assume that we happen
to know a pair of .A′ and .′ corresponding to given electric and magnetic fields,
but that .∇ · A′ = 0. Then, we can choose a function f in Eq. (10.93) such that
′ 2
.∇ · A = ∇ f . The resulting .A satisfies the Coulomb gauge condition, Eq. (10.97).

In a continuous medium satisfying linear constitutive relations with dielectric

constant .ǫ and magnetic permeability .μ, Maxwell’s equations can be shown to be
equivalent to the following partial differential equations for the vector and scalar
potentials:

1 ∂ 4πρ(r, t)
. − (∇ · A(r, t)) − ∇ 2 (r, t) = ,. (10.98)
c ∂t ǫ
 
ǫμ ∂ 1 ∂A(r, t)
∇ (∇ · A(r, t)) − ∇ 2 A(r, t) − − − ∇(r, t)
c ∂t c ∂t
4πμ
= J(r, t), (10.99)
c

where .ρ(r, t) and .J(r, t) are charge and current densities. In Coulomb gauge where
Eq. (10.97) is satisfied, the above equations are simplified further to

4πρ(r, t)
. −∇ 2 (r, t) = ,. (10.100)
ǫ
ǫμ ∂ 2 A(r, t) ǫμ ∂ 4πμ
−∇ 2 A(r, t) + 2 2
+ ∇(r, t) = J(r, t). (10.101)
c ∂t c ∂t c

Let us now consider the simplest case where there are no sources of charge and
current. Then, .ρ(r, t) = 0 and .J(r, t) = 0, and Eq. (10.100) reduces to

2
.∇ (r, t) = 0. (10.102)
358 10 Quantum Dynamics of Pure and Mixed States

A solution of this, which satisfies the boundary condition of the infinite space, is
that . = 0. That is, in free space, the scalar potential can be assumed to be zero (in
the Coulomb gauge). With this assumption, Eq. (10.101) now simplifies to

2 ǫμ ∂ 2 A(r, t)
.∇ A(r, t) − = 0. (10.103)
c2 ∂t 2

Many solutions are possible for this partial differential equation. One of the simplest
solutions is the following plane wave solution:

.A(r, t) = A0 ue cos(k · r − ωt), (10.104)

where

2 ǫμ 2
.k = |k|2 = ω , (10.105)
c2

Assuming that both .ǫ and .μ are real and positive (non-absorptive medium), the
refractive index is defined as
√
.n = ǫμ. (10.106)

Thus, in a medium with refractive index n, the plane wave solution travels with a
speed .v = ω/k = c/n.
Inserting Eq. (10.104) into the condition Eq. (10.97), one can find that

.k · ue = 0. (10.107)

That is, .ue is perpendicular to the direction of the propagation of the vector potential.
Inserting Eq. (10.104) into Eq. (10.90),

.B(r, t) = A0 (k × ue ) sin(ωt − k · r) = A0 kub sin(ωt − k · r), (10.108)

where .ub = k × ue /k , the direction of the magnetic field. Inserting Eq. (10.104) into
Eq. (10.96) with . = 0,

A0 ω A0 k
.E = ue sin(ωt − k · r) = ue sin(ωt − k · r). (10.109)
c n

For the case of a a quantum mechanical particle with mass m and charge q in the
presence of an electromagnetic field, which is still treated as a classical wave, the
Hamiltonian operator is obtained by simply replacing the physical observables in
Eq. (10.88) with quantum operators as follows:

1 q 2
.Ĥ = p̂ − A(r̂, t) + q(r̂, t) + V (r̂), (10.110)
2m c
Appendix: Interaction Hamiltonian in the Presence of Radiation 359

where .p̂ and .r̂ are momentum and position operators. Expanding the square in
Eq. (10.110),

q 2 q q q2
. p̂ − A(r̂, t) = p̂2 − p̂ · A(r̂, t) − A(r̂, t) · p̂ + 2 A(r̂, t)2 . (10.111)
c c c c

Employing the following identity:

h̄
.p̂ · A(r̂, t) = A(r̂, t) · p̂ + ∇ · A(r̂, t), (10.112)
i

the Hamiltonian of Eq. (10.110) can be expressed as

p̂2 q h̄q q2
.Ĥ = + V (r̂) + q(r̂, t) − A(r̂, t) · p̂ − ∇ · A(r̂, t) + A(r̂, t)2 .
2m mc ic 2mc2
(10.113)

In Coulomb gauge, .∇ · A(r̂, t) = 0. In addition, assuming weak enough field,

the quadratic term .A(r̂, t)2 can be neglected. As a result, Eq. (10.110) can be
approximated as
q
.Ĥ (t) = Ĥ0 + q(r̂, t) − A(r̂, t) · p̂, (10.114)
mc

where .Ĥ0 is the Hamiltonian in the absence of the electromagnetic radiation and is
given by

p̂2
.Ĥ0 = + V (r̂) . (10.115)
2m

Let us consider the case where the particle is subject to a monochromatic

radiation of frequency .ω. Then, using the fact .(r̂, t) = 0 and inserting Eq. (10.104)
into Eq. (10.114), we find the following time dependent Hamiltonian:

.Ĥ (t) = Ĥ0 + Ĥ1 (t), (10.116)

where
q
.Ĥ1 (t) =− A0 eik·r̂−iωt + e−ik·r̂+iωt ue · p̂. (10.117)
2mc

In a typical molecular spectroscopy, the wavelength of the radiation is much larger

than the molecular length scale. Therefore, one can make an approximation that
.k · r̂ ≈ 0, which is referred to as dipole approximation for a reason to be clear below.

Then, the matter radiation interaction can be approximated as

360 10 Quantum Dynamics of Pure and Mixed States

q
.Ĥ1 (t) ≈− A0 e−iωt + eiωt ue · p̂, (10.118)
2mc

where the momentum operator .p̂ can be expressed in terms of the position operator
employing the following relation:

im  
.p̂ = Ĥ0 , r̂ . (10.119)
h̄

Then, Eq. (10.118) can be expressed as

i  
.Ĥ1 (t) =− A0 e−iωt + eiωt H0 , ue · D̂ , (10.120)
2h̄c

where .D̂ = q r̂, the dipole operator. That is, the radiation interacts with the dipole
operator and this is the reason why the approximation of .k · r̂ ≈ 0 is called dipole
approximation.
Let us consider the matrix element of .Ĥ1 (t) between two orthonormal eigenstates
of .Ĥ0 with eigenvalues .Ej and .Ek as follows:

i  
.Ej |Ĥ1 (t)|Ek  =− A0 e−iωt + eiωt Ej | H0 , ue · D̂ |Ek 
2h̄c
i
=− A0 e−iωt + eiωt (Ej − Ek )Ej |ue · D̂|Ek 
2h̄c
i
=− A0 e−iωt + eiωt (Ej − Ek )ue · D j k , (10.121)
2h̄c

where .D j k is the transition dipole vector between states .|Ej  and .|Ek . The above
expression has two time dependent terms, one of which becomes almost non-
oscillatory when resonant with the energy difference between the final and initial
states. On the other hand, the other becomes highly oscillatory. Ignoring this
highly oscillatory term and keeping only the resonant term is called rotating wave
approximation.
All the expressions derived above can easily be extended to molecular or many
particle systems, and the resulting expressions similar to Eq. (10.121), along with
the rotating wave approximation, leads to the interaction Hamiltonian such as
Eq. (10.33).
The description of molecule-radiation interaction presented here is limited to lin-
ear interactions and semiclassical in nature since the electromagnetic wave is treated
classically. There are well-established theories of higher order molecule-radiation
interactions [39, 40] or fully quantum mechanical treatment of radiation [20, 41],
which can provide more modern perspective of molecule-radiation interactions.
Exercise Problems with Solutions 361

Exercise Problems with Solutions

10.1 A one dimensional harmonic oscillator with mass m and angular frequency ω is
subject to the following time dependent perturbation Hamiltonian:

.Ĥ1 (t) = Ae−t/τ (b̂ + b̂† ), for t > 0

where τ and A are positive real numbers, and b̂ and b̂† are lowering and raising operators
of the harmonic oscillator corresponding to the zeroth order Hamiltonian Ĥ0 . Given
that the state at time t = 0 is |j , the eigenstate of the zeroth order Hamiltonian
with eigenvalue Ej = h̄ω(j + 12 ), the first order time dependent perturbation theory
approximation for the state at time t is

i i t i ′ i ′
.|ψ; t
(1)
= e− h̄ Ej t |j  − dt ′ e− h̄ Ĥ0 (t−t ) Ĥ1 (t ′ )e− h̄ Ej t |j .
h̄ 0

Assume that j = 1. Then, calculate the probability for the state to be at |2 at time
t = π/ω.
Solution 10.1

i t i ′ ′ i ′
. 2|ψ; t = − dt ′ e− h̄ E2 (t−t ) Ae−t /τ 2|(b̂ + b̂† )|1e− h̄ E1 t
h̄ 0
1
i √ i
t
′ −( τ1 −iω)t ′ i √ i 1 − e−( τ −iω)t
= − A 2e− h̄ E2 t dt e = − A 2e− h̄ E2 t 1
.
h̄ 0 h̄ τ − iω

2|A|2 (1 + e−π/(ωτ ) )2
Probability: |2|ψ; t = π/ω|2 = .
h̄2 (1/τ )2 + ω2

10.2 Consider the following operator

⎛ ⎞
1
0C 4
⎜ 1 ⎟
.ρ̂ = ⎝ 0 ,
4 0⎠
1
C 0 2

where C is a real number. Find out the largest possible value of C for ρ̂ to be a density
operator.
Solution 10.2
⎛ ⎞⎛ ⎞ ⎛ ⎞
11 1 3C
0C 4 0 C + C2 0
2 ⎜ 1 ⎟ ⎜ 1 ⎟ ⎜ 16
4
1
4
.ρ̂ = ⎝ 0 =⎝ 0 ⎠.
4 0 ⎠⎝ 0 4 0 ⎠ 0
⎟
16
C 0 12 C 0 12 3C
4 0 1
4 + C2

Since T r{ρ̂ 2 } ≤ 1, 38 + 2C 2 ≤ 1. Therefore, C 2 ≤ 5

16 . As a result, the largest possible
√
value of C is 5/4.
362 10 Quantum Dynamics of Pure and Mixed States

Problems

10.3 Prove the following identity.

i
. [p̂, V̂ (x̂)] = V̂ ′ (x̂),
h̄

where V ′ (x) is the derivative of V (x).

10.4 For a harmonic oscillator with mass m and angular frequency ω, its lowering
and raising operators are defined by Eqs. (3.51) and (3.52). Prove that their Heisenberg
†
operators are expressed as b̂H (t) = b̂e−iωt and b̂H (t) = b̂† eiωt .
10.5 Calculate the average energy, square displacement, and kinetic energy for a canon-
ical ensemble of harmonic oscillators and examine how they depend on temperature.
10.6 A one dimensional harmonic oscillator with mass m and angular frequency ω is
subject to the following time dependent perturbation Hamiltonian:

.Ĥ1 (t) = Ae−t/τ (1 + ǫ x̂), for t > 0,

where τ , A, and ǫ are positive real numbers, and x̂ is the displacement of the harmonic
oscillator. Let us denote the eigenstate of the zeroth order Hamiltonian, i,e,, the harmonic
oscillator Hamiltonian, with energy h̄ω(n+1/2) as |n. Given that the state at time t = 0
is |ψ; 0 = |1 and assuming that A and ǫ are small enough, use the first order time
dependent perturbation theory to answer the following questions.
(a) Find the expression for the first order time dependent perturbation approximation at
time t, |ψ; t(1) (Hint: Express x̂ in terms of raising and lowering operators b̂ and
b̂† . Do the integration explicitly.)
(b) What is the probability for the oscillator to be in the state |2 at time t = 1/ω?
10.7 Prove that T r ρ̂ 2 given by Eq. (10.62) satisfies Eq. (10.55) using the Cauchy-
 

Schwarz inequality.
10.8 For a system where two orthonormal states |1 and |2, a state is represented by
the following density operator:

1 1
.ρ̂ = |11| + |22| + C|12| + C ∗ |21|.
2 2

(a) Find the exact condition for C in order for the above operator is a genuine
density operator.
(b) Find the exact condition for C in order for the above density operator to
represent a pure state.
Chapter 11
Theories for Electronic Structure
Calculation of Polyatomic Molecules

It is a profound and necessary truth that the deep things in

science are not found because they are useful; they are found
because it was possible to find them.
– J. Robert Oppenheimer

Abstract This chapter presents more detailed theoretical account of methods used
for the calculation of electronic states of polyatomic molecules. The main focus
is the Hartree-Fock (HF) approximation and its implementation employing finite
basis as formulated by Roothaan. Brief accounts of more advanced methods that
go beyond the HF approximation are provided. The basic principles underlying the
Density Functional Theory (DFT) for the calculation of the ground electronic state
of adiabatic electronic states are also explained.

In this chapter, we will go through in more detail how to find approximate solutions
for the electronic Schrödinger equation, Eq. (9.9), or its equivalent. For this, let us
rewrite the electronic Hamiltonian .Ĥel given by Eq. (9.8) as follows:

Ne Ne 
 1 1
Ĥel =
. ĥμ + , (11.1)
2 r̂μν
μ=1 μ=1 ν=μ

where

p̂2μ Nu
 Zc
. ĥμ = − ,. (11.2)
2 |r̂μ − Rc |
c=1
1 1
= . (11.3)
r̂μν |r̂μ − r̂ν |

In above expressions, the parametric dependences of .ĥμ and .Ĥel on the nuclear
coordinates are not shown explicitly for notational convenience.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 363
S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_11
364 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

11.1 Hartree-Fock Approximation and Roothaan Equation

A simple but limited derivation of the HF energy, for the case of atomic orbitals,
was provided in Chap. 8. This chapter provides a more detailed derivation for more
general cases [14, 42].

11.1.1 General Single Determinant State

Let us assume that the electronic state for the .Ne electrons can be expressed as a
single Slater determinant as follows:
 
 |ψ1 (1) |ψ2 (1) · · · |ψNe (1) 
 
1  |ψ1 (2) |ψ2 (2) · · · |ψNe (2) 
.|Ne  = √  .. .. .. .. 
Ne !  . . . . 

 |ψ (N ) |ψ (N ) · · · |ψ (N ) 
1 e 2 e Ne e

1 
(−1)NP P̂ |ψ1 (1)|ψ2 (2) · · · |ψNe (Ne ) , (11.4)
 
= √
Ne !
P̂

where .|ψk (μ) is the kth single electron spin-orbit state for electron .μ = 1, · · · , Ne ,
and .P̂ represents all distinct permutations of the .Ne electrons, with .NP being the
number of pairwise permutations equivalent to each permutation. It is assumed that
all .|ψk (μ)’s are orthogonal. If not, it should be possible to find a transformation
that makes them orthogonal since the starting spin-orbit states should be linearly
independent [42]. Such a transformation only results in the difference in the
prefactor of the Slater determinant. On the other hand, since the starting basis
states for molecular electronic states are nonorthogonal in general, it is important
to specify the condition for orthogonality in more detail as will become clear later.
First, let us calculate the expectation value of .Ĥel with respect to .|Ne  as
follows:
1 
(−1)NP +NP ′ P̂ ψ1 (1)|ψ2 (2)| · · · ψNe (Ne )|
 
. Ne |Ĥel |Ne  =
Ne !
P̂ P̂ ′
⎧ ⎫
Ne Ne 
⎨ 1 1 ⎬ ′
P̂ |ψ1 (1)|ψ2 (2) · · · |ψNe (Ne ) . (11.5)

× ĥμ +
⎩ 2 r̂μν ⎭
μ=1 μ=1 ν=μ
11.1 Hartree-Fock Approximation and Roothaan Equation 365

First, the term involving sum over single electron terms can be simplified to
⎛ ⎞
Ne
1  
(−1)NP +NP ′ P̂ ψ1 (1)|ψ2 (2)| · · · ψNe (Ne )| ⎝
 
. ĥμ ⎠
Ne !
P̂ P̂ ′ μ=1

×P̂ ′ |ψ1 (1)|ψ2 (2) · · · |ψNe (Ne )

 

Ne Ne
 1 
= (Ne − 1)! ψj (μ)|ĥμ |ψj (μ)
Ne !
μ=1 j =1

Ne

= ψj (1)|ĥ1 |ψj (1). (11.6)
j =1

Major steps leading to the final expression above are summarized below.
• The first equality results from the fact that there are .Ne different ways to assign
single electron state for the electron with index .μ by the permutation .P̂ .
• The single electron interaction term survives only when each realization of the
permutation .P̂ ′ is exactly the same as that of .P̂ . Otherwise, there are at least one
spin-orbital states that are orthogonal for those electrons with indices different
from .μ.
• The .(Ne − 1)! factor results from counting all possible ways of permuting the
.(Ne − 1) electrons except for the .μ electron.

• The second equality above results from the fact that the single electron term is
independent of .μ and can be replaced with 1.
• After this replacement, the summation over .μ results in a factor of .Ne , which is
divided by the same factor in the remaining denominator.
Therefore,

Ne
 Ne
 Ne

Ne |
. ĥμ |Ne  = ψj (1)|ĥ1 |ψj (1) = ψj |ĥ|ψj , (11.7)
μ=1 j =1 j =1

where we have dropped electron index 1 for simplicity in the last expression.
On the other hand, the two electron terms in Eq. (11.5) simplifies to

1 
(−1)NP +NP ′ P̂ ψ1 (1)|ψ2 (2)| · · · ψNe (Ne )|
 
.
Ne !
P̂ P̂ ′
⎧ ⎫
Ne 
⎨1  1 ⎬ ′ 
× P̂ |ψ1 (1)|ψ2 (2) · · · |ψNe (Ne )
⎩2 r̂μν ⎭
μ=1 ν=μ
366 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

Ne  Ne  
1 1  1
= ψj (μ)|ψk (ν)| |ψj (μ)|ψk (ν)
2 Ne (Ne − 1) r̂μν
μ=1 ν=μ j =1 k=j

1

−ψj (μ)|ψk (ν)| |ψk (μ)|ψj (ν)
r̂μν
Ne
Ne 
1 1

= ψj (1)|ψk (2)| |ψj (1)|ψk (2)
2 r̂12
j =1 k=1

1

−ψj (1)|ψk (2)| |ψk (1)|ψj (2) . (11.8)
r̂12

The final expression above is obtained through a procedure described below.

• In the first equality of the above identities, the summation over j and k represent
all distinct single electron spin-orbital states assigned to .μ and .ν electrons out of
the permutation .P̂ . There are .(Ne − 2)! ways of permuting electrons other than
these two. These, when divided with the denominator, result in .(Ne − 2)!/Ne ! =
1/Ne (Ne − 1).
• Out of all realizations of the permutation .P̂ ′ , only those that are exactly the same
as those from .P̂ and those for which the spin-orbital states for .μ and .ν electrons
are interchanged survive because, for other permutations, there are at least one
electron (other than .μ and .ν) that involves overlap of two orthogonal spin-orbit
states.
• The permutation of .P̂ ′ , which differs from .P̂ only by the exchange of two
spin-orbit states involves one additional exchange compared to .P̂ . Therefore,
.(−1) P
N +NP ′ = −1 for this case, which explains the negative sign of the second

term.
• The second equality results from the fact that the indices .μ and .ν, being dummy
indices, can be replaced with 1 and 2 without affecting the final outcome. The
summation over .μ and .ν thus results in .Ne (Ne − 1) factor, which is divided by
exactly the same factor in the remaining denominator.
• In the second equality, the case with .j = k was added since the two-electron
terms in this case cancel out and do not contribute to the summation.
For more compact expression of the two-electron interaction terms, let us use
simpler notations that are consistent with conventional definitions used in electronic
structure theory as described below.

Notations for Two Electron Terms

Since the two electron interaction terms commonly involve the same Coulomb
operator and electron indices 1 and 2 are dummy variables, we can express all
the two-electron interaction terms as follows:
(continued)
11.1 Hartree-Fock Approximation and Roothaan Equation 367

1
ψj ψk |ψj ′ ψk ′  = ψj (1)|ψk (2)|
. |ψj ′ (1)|ψk ′ (2)
r̂12
= (ψj ψj ′ |ψk ψk ′ ). (11.9)

Note that .ψj ψk |ψj ′ ψk ′  = ψk ψj |ψk ′ ψj ′  since the exchange of 1 and 2 for
both ket and bra does not cause any difference. The second equality above
introduces so called chemist’s notation, and is intuitively useful because bra
and kets for the same electron are placed together. Note that .(ψj ψj ′ |ψk ψk ′ ) =
(ψk ψk ′ |ψj ψj ′ ) for this case.
Let us also define the following difference term:

ψj ψk ||ψj ′ ψk ′  = ψj ψk |ψj ′ ψk ′  − ψj ψk |ψk ′ ψj ′ 

.

= ψj ψk |ψj ′ ψk ′  − ψk ψj |ψj ′ ψk ′ 

= ψk ψj |ψk ′ ψj ′  − ψk ψj |ψj ′ ψk ′ 
= ψk ψj |ψk ′ ψj ′  − ψj ψk |ψk ′ ψj ′ . (11.10)

Combining Eqs. (11.7) and (11.8) and employing the two-electron integrals
introduced above, we find that Eq. (11.5) can be expressed as

Ne Ne
Ne 
HF
 1  
Eel
. = ψj |ĥ|ψj + ψj ψk |ψj ψk −ψj ψk |ψk ψj 
2
j =1 j =1 k=1

Ne Ne
Ne 
 1
= ψj |ĥ|ψj + ψj ψk ||ψj ψk . (11.11)
2
j =1 j =1 k=1

The above expression has been derived under the assumption that all the single
electron states are orthogonal. Therefore, the following conditions have to be
ensured.

ψj |ψk  = δj k .
. (11.12)

Thus, employing the Lagrange multiplier method [19, 43] to incorporate the above
constraint, we can find out the best approximation for the energy with this constraint
employing a Lagrangian .L defined as follows:

Ne N N N N
 1e e e e
L=
. ψj |ĥ|ψj  + ψj ψk ||ψj ψk  − λj k (ψj |ψk  − δj k ).
2
j =1 j =1 k=1 j =1 k=1
(11.13)
368 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

where .λj k ’s are as yet unknown Lagrange multipliers to be determined. Now,

consider the variation of .L as a functional of all .|ψj ’s and .ψj |’s as follows:
 
δL =
. δψj |ĥ|ψj  + ψj |ĥ|δψj 
j

N N
1e e

+ δψj ψk ||ψj ψk  + ψj δψk ||ψj ψk 
2
j =1 k=1

+ψj ψk ||δψj ψk  + ψj ψk ||ψj δψk 
Ne 
 Ne
λj k δψj |ψk  + ψj |δψk  = 0. (11.14)
 
−
j =1 k=1

In the second double summation of the above expression, .ψj δψk ||ψj ψk  =
δψk ψj ||ψk ψj  and .ψj ψk ||ψj δψk  = ψk ψj ||δψk ψj . The exchange of the
dummy summation indices j and k in these terms thus make these identical to other
two electron terms. Thus, the above expression is equivalent to

Ne

   
δL =
. δψj |ĥ|ψj  + δψj ψk ||ψj ψk  − λj k δψj |ψk 
j k=1

Ne


ψj ψk ||δψj ψk  − λkj ψk |δψj  = 0,
 
+ψj |ĥ|δψj  +
k=1
(11.15)

where the indices j and k in the last term of Eq. (11.14) were interchanged, resulting
in .λkj ψk |δψj .
Let us consider two electron terms in Eq. (11.15) in more detail, which is given
by

1
. δψj ψk ||ψj ψk  = δψj (1)|ψk (2)| |ψj (1)|ψk (2)
r̂12
1
− δψj (1)|ψk (2)| |ψk (1)|ψj (2). (11.16)
r̂12

These can also be expressed in terms of effective single electron operators as

follows:

1
 
. Ĵ1 [ψk ]|ψj (1) = ψk (2)| |ψk (2) |ψj (1), . (11.17)
r̂ 12
11.1 Hartree-Fock Approximation and Roothaan Equation 369

1
 
K̂1 [ψk ]|ψj (1) = ψk (2)| |ψj (2) |ψk (1), (11.18)
r̂ 12

where .Ĵ1 [ψk ]|ψj (1) and .K̂1 [ψk ]|ψj (1) are Coulomb and exchange interaction
operators. These are functionals of as yet undetermined state .|ψk (2) for electron
2 while being operators on the state .|ψj (1) of the electron 1. Employing these,
Eq. (11.16) can be expressed as

δψj ψk ||ψj ψk  = δψj (1)|(Ĵ1 [ψk ] − K̂1 [ψk ])|ψj (1).

. (11.19)

Similarly,

ψj ψk ||δψj ψk  = ψj (1)|(Ĵ1 [ψk ] − K̂1 [ψk ])|δψj (1).

. (11.20)

Therefore, Eq. (11.15) can be expressed as

Ne Ne
  
  
. δL = δψj (1)| ĥ1 + (Ĵ1 [ψk ] − K̂1 [ψk ]) |ψj (1) − λj k |ψk (1)
j k=1 k=1

Ne Ne
   
  
+ ψj (1)| ĥ1 + (Ĵ1 [ψk ] − K̂1 [ψk ]) − ψk (1)|λkj |δψj (1)
j k=1 k=1

= 0.
(11.21)

If the states .|ψj  are those minimizing the energy under the constraint,
Eq. (11.21) has to be satisfied regardless of any variations of them, .|δψj  and
.δψj |, which can be viewed as being independent of each other. This means that the

following two effective single electron equations have to be satisfied.

Ne Ne
 
 
. ĥ1 + (Ĵ1 [ψk ] − K̂1 [ψk ]) |ψj (1) = λj k |ψk (1), . (11.22)
k=1 k=1
Ne Ne
 
 
ψj (1)| ĥ1 + (Ĵ1 [ψk ] − K̂1 [ψk ]) = ψk (1)|λkj . (11.23)
k=1 k=1

Note that .ĥ1 , .Ĵ1 [ψk ], and .K̂1 [ψk ] are Hermitian operators. Therefore, taking
Hermitian adjoint of Eq. (11.22), we find that

Ne Ne
 
 
.ψj (1)| ĥ1 + (Ĵ1 [ψk ] − K̂1 [ψk ]) = ψk (1)|λ∗j k . (11.24)
k=1 k=1
370 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

Comparing this with Eq. (11.23), we find that .λ∗j k = λkj . Thus, .λj k ’s are components
of a Hermitian matrix, which has real eigenvalues. Thus, solving Eq. (11.22) is
sufficient for finding the solution of the HF approximation.
Equation (11.22) is similar to an eigenvalue problem, but is not yet in a
conventional form because .λj k ’s form a non-diagonal matrix. However, since this
matrix is Hermitian, it is possible to find a unitary transformation such that it
becomes diagonal. For this, let us introduce the following Fock operator :

Ne 
 
.F̂1 [ψ] = ĥ1 + Ĵ1 [ψk ] − K̂1 [ψk ] . (11.25)
k=1

Thus, Eq. (11.22) can be expressed as

Ne

.F̂1 [ψ]|ψj (1) = λj k |ψk (1), for j = 1, · · · , Ne (11.26)
k=1

Since the matrix consisting of the .λj k as its elements, which we denote here as ., is
Hermitian, there always exist normalized eigenvectors .u1 , · · · , uNe with eigenvalues
.E1 , .· · · , .ENe such that

.uk = Ek uk , for k = 1, · · · , Ne (11.27)

Let us define a transformation matrix .U = (u1 , · · · , uNe ), for which .Uj k = (uk )j .
Then, .U† U = I , where I is the identity matrix with dimension .Ne , and transforms
. to diagonal form as follows:

†
.U U = U† UE = E, (11.28)

where .E is the diagonal matrix with the .Ek as its diagonal elements. It is also possible
to prove that .UU† = I . Thus, .U is a unitary matrix.
Now, let us consider the following vector notation for Eq.(11.26).
⎛ ⎞ ⎛ ⎞
|ψ1 (1) |ψ1 (1)
⎜ |ψ2 (1) ⎟ ⎜ |ψ2 (1) ⎟
⎜ ⎟ ⎜ ⎟
.F̂1 [ψ] ⎜ .
⎜ .
⎟ = ⎜ .
⎟ ⎜ .
⎟.
⎟ (11.29)
⎝ . ⎠ ⎝ . ⎠
|ψNe (1) |ψNe (1)

In the above expression, note that .F̂1 [ψ] is a linear quantum mechanical operator that
applies to each component state. Therefore, applying .U† on both sides, we obtain
11.1 Hartree-Fock Approximation and Roothaan Equation 371

⎛ ⎞ ⎛ ⎞ ⎛ ⎞
|ψ1 (1) |ψ1 (1) |ψ1 (1)
⎜ |ψ2 (1) ⎟ ⎜ |ψ2 (1) ⎟ ⎜ |ψ2 (1) ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
†⎜ † †⎜ †⎜
.F̂1 [ψ]U
⎜ .. ⎟ = U UU ⎜ ..
⎟
⎟ = EU ⎜ ..
⎟ ⎟.
⎟ (11.30)
⎝ . ⎠ ⎝ . ⎠ ⎝ . ⎠
|ψNe (1) |ψNe (1) |ψNe (1)

Note that we have used the fact that .F̂1 [ψ] and .U† commute with each other, since
the former is a quantum mechanical operator whereas the latter is a matrix being
applied to the vector of states. The identity of Eq. (11.28) has also been used.
Let us now define,

Ne  
U†

′
.|ψj (1) = |ψk (1). (11.31)
jk
k=1

Then, Eq. (11.30) implies that

′
.F̂1 [ψ]|ψj (1) = Ej |ψj′ (1), for j = 1, · · · , Ne (11.32)

In the above expression, note that .F̂1 is defined in terms of untransformed states.
However, it turns out that it is equivalent to one expressed in terms of transformed
states. This can be shown using the definition of .F̂1 directly. Since .ĥ1 is independent
of states, we only need to consider the two electron terms. Thus, for an arbitrary state
.|ψ (1), the following identity can be established.
′′

Ne

. (Ĵ1 [ψk ] − K̂1 [ψk ])|ψ ′′ (1)
k=1
Ne 
1 1
 
= ψk (2)| |ψk (2)|ψ ′′ (1) − ψk (2)| |ψ ′′ (2)|ψk (1)
r̂12 r̂12
k=1
Ne 
Ne  Ne
1
 
∗
= Ukj Ukj ′ ψj′ (2)| |ψj′ ′ (2)|ψ ′′ (1)
r̂12
j =1 j ′ =1 k=1

1

−ψj′ (2)| |ψ ′′ (2)|ψj′ ′ (1)
r̂12
Ne 
1 1
 
= ψj′ (2)| |ψ ′ (2)|ψ ′′ (1) − ψj′ (2)| |ψ ′′ (2)|ψj′ (1)
r̂12 j r̂12
j =1

Ne

= (Ĵ1 [ψk′ ] − K̂1 [ψk′ ])|ψ ′′ (1), (11.33)
k=1
372 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

where the fact that .|ψk (1) = kj Ukj |ψj′ (1) and the unitarity of .U have been used.
Therefore, .F̂1 [ψ] in Eq. (11.26) can be replaced with .F̂1 [ψ ′ ].
Note that the transformation from .|ψj (1) to .|ψj′ (1) by the unitary transforma-
tion changes the Slater determinant state, Eq. (11.4), only by a complex factor of unit
magnitude since the determinant of .U is a complex number with unit magnitude.
Therefore, the transformation does not affect the actual many electron states either
[14, 42]. Therefore, without losing generality, we can replace .|ψj′ (1) in Eq. (11.26)
with .|ψj (1). The resulting equation is known as the canonical form of the HF
equation and is summarized below.

Canonical Form of the HF Equation

The slater determinant state for many electrons, Eq. (11.4), which minimizes the
total electronic energy, consists of the .|ψj  that satisfy the following canonical
HF equation:

Ne

.F̂1 [ψ]|ψj (1) = (ĥ1 + (Ĵ1 [ψk ] − K̂1 [ψk ])|ψj (1) = Ej |ψj (1),
k=1
(11.34)

where .j = 1, · · · , Ne . Note that all the states are orthogonal and satisfy the
condition of Eq. (11.12).

Taking inner product of Eq. (11.34) with .ψj (1)|, we obtain the following
expression for the energy of the state .Ej :

.Ej = ψj (1)|F̂1 [ψ]|ψj (1)

Ne

= ψj (1)|ĥ1 |ψj (1) + ψj (1)|Ĵ1 [ψk ]|ψj (1) − ψj (1)|K̂1 [ψk ]|ψj (1)
k=1
Ne

= ψj (1)|ĥ1 |ψj (1) + ψj ψk |ψj ψk  − ψj ψk |ψk ψj 
k=1
Ne

= ψj (1)|ĥ1 |ψj (1) + ψj ψk ||ψj ψk . (11.35)
k=1

Thus, we find that the HF energy given by Eq. (11.11) is related to the sum of the
as follows:
.Ej

Ne N N
 1e e
.EH F = Ej − ψj ψk ||ψj ψk . (11.36)
2
j =1 j =1 k=1
11.1 Hartree-Fock Approximation and Roothaan Equation 373

In general, Eqs. (11.34) and (11.36) can be solved together in the following self-
consistent manner.
1. Assume a set of trial .|ψj .
2. Calculate .F̂1 [ψ] for the trial .|ψj .
3. Determine the new .|ψj  and .Ej satisfying Eq. (11.34).
4. Repeat steps 2 and 3 until the HF energy, Eq. (11.36), has converged sufficiently.

11.1.2 Restricted HF Equation for Doubly Filled Orbital States

Let us now consider the simplest case where there are .Ne = 2Nr electrons doubly
occupying .Nr orbitals. Thus, for .p = 1, · · · , Nr , the spin-orbit states are expressed
as

. |ψ2p−1 (1) = |φp (1)|α(1), . (11.37)

|ψ2p (1) = |φp (1)|β(1), (11.38)

where .|φp (1) represents the spatial-orbital state .|α(1) for electron labeled 1. On
the other hand, .|α(1) and .|β(1) represent spin up and down states, respectively,
also for electron 1. Then, the two electron interaction operators can be expressed
in terms of orbital states. First, the Coulomb interaction terms can be expressed as
follows:
1
.Ĵ1 [ψ2q−1 ]|φp (1)|α(1) = φq (2)| |φq (2)|φp (1)|α(1)
r̂12
 
= Jˆ1 [φq ]|φp (1) |α(1), . (11.39)
1
Ĵ1 [ψ2q ]|φp (1)|α(1) = φq (2)| |φq (2)|φp (1)|α(1)
r̂12
 
= Jˆ1 [φq ]|φp (1) |α(1), . (11.40)
1
Ĵ1 [ψ2q−1 ]|φp (1)|β(1) = φq (2)| |φq (2)|φp (1)|β(1)
r̂12
 
= Jˆ1 [φq ]|φp (1) |β(1), . (11.41)
1
Ĵ1 [ψ2q ]|φp (1)|β(1) = φq (2)| |φq (2)|φp (1)|β(1)
r̂12
 
= Jˆ1 [φq ]|φp (1) |β(1), (11.42)

where .Jˆ1 represents Coulomb interaction operator applied to spatial orbital only. On
the other hand, only two of the following exchange terms are nonzero in general.
374 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

1
.K̂1 [ψ2q−1 ]|φp (1)|α(1) = φq (2)| |φp (2)|φq (1)|α(1)
r̂12
 
= K̂1 [φq ]|φp (1) |α(1), . (11.43)
1
K̂1 [ψ2q ]|φp (1)|β(1) = φq (2)| |φp (2)|φq (1)|β(1)
r̂12
 
= K̂1 [φq ]|φq (1) |β(1), (11.44)

where .K̂1 represents exchange interaction operator applied to spatial orbital only.
It is easy to show that the following two cases of exchange interaction operation
vanish due to the orthogonality of spin states.

. K̂1 [ψ2q ]|φp (1)|α(1) = 0, . (11.45)

K̂1 [ψ2q−1 ]|φp (1)|β(1) = 0. (11.46)

Employing Eqs. (11.39)–(11.46) provided above, one can express all the two
electron interaction terms for spin-orbit states in terms of those for spatial orbital
states as follows:
Ne

. (Ĵ1 [ψk ] − K̂1 [ψk ])|φp (1)|α(1)
k=1
Nr 
 
= Ĵ1 [ψ2q−1 ] − K̂1 [ψ2q−1 ] + Ĵ1 [ψ2q ] − K̂1 [ψ2q ] |φp (1)|α(1)
q=1

Nr 
 
= |α(1) 2Jˆ1 [φq ] − K̂1 [φq ] |φp (1), . (11.47)
q=1

Ne

(Ĵ1 [ψk ] − K̂1 [ψk ])|φp (1)|β(1)
k=1
Nr 
 
= Ĵ1 [ψ2q−1 ] − K̂1 [ψ2q−1 ] + Ĵ1 [ψ2q ] − K̂1 [ψ2q ] |φp (1)|β(1)
q=1

Nr 
 
= |β(1) 2Jˆ1 [φq ] − K̂1 [φq ] |φp (1). (11.48)
q=1

Taking inner products of the above expressions with either .α(1)| or .β(1)| results
in the same expression for two electron operations involving spatial orbitals. Thus,
combining this with the remaining one electron term, one can obtain the canonical
HF equation for spatial orbitals as summarized below.
11.1 Hartree-Fock Approximation and Roothaan Equation 375

Canonical Form of the HF Equation for Doubly Occupied Spatial Orbital

States
Employing Eq. (11.47) (or Eq. (11.48)) in Eq. (11.34) and taking inner-product
with .α(1)| (or .β(1)|), we obtain

Nr

.F̂1 [φ]|φp (1) = (ĥ1 + (2Jˆ1 [φq ] − K̂1 [φq ])|φp (1) = Ep |φp (1),
q=1
(11.49)

where .p = 1, · · · , Nr . Note that all the orbital states are orthogonal to each other.

Taking inner product of Eq. (11.49) with .φp (1)|, we obtain the following
expression for the orbital energy:

.Ep = φp (1)|F̂1 [φ]|φp (1)

Nr

= φp (1)|ĥ1 |φp (1) + (2φp (1)|Jˆ1 [φq ]|φp (1) − φp (1)|K̂1 [φq ])|φp (1)
q=1

Nr

= φp (1)|ĥ1 |φp (1) + (2φp φq |φp φq  − φp φq |φq φp ). (11.50)
q=1

On the other hand, following a procedure similar to that used for deriving the above
HF equation, the HF energy, Eq. (11.11), for the present case can be expressed as

Nr
 Nr 
 Nr
.EH F =2 φp (1)|ĥ1 |φp (1) + (2φp φq |φp φq  − φp φq |φq φp ).
p=1 p=1 q=1
(11.51)

Therefore,

Nr
 Nr 
 Nr
.EH F =2 Ep − (2φp φq |φp φq  − φp φq |φq φp ). (11.52)
p=1 p=1 q=1

In general, Eqs. (11.49) and (11.52) can be solved together in the following self-
consistent manner.
1. Assume a set of trial .|φp .
2. Calculate .F̂ [φ] for the trial .|φp .
376 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

3. Determine new the .|φp  and .Ep satisfying the restricted HF equation, Eq. (11.49).
4. Repeat steps 2 and 3 until the HF energy, Eq. (11.52), has converged sufficiently.

11.1.3 Linear Combination of Basis States

The self-consistent procedure described in previous subsections for determining

spin orbital or orbital states is simple in principle, but actual implementation in its
most general form is a numerically challenging task. In order to address this issue,
Roothaan [42] employed an approximation for the states as linear combinations of
pre-defined basis states. The resulting equation, often called HF-Roothaan equation,
amounts to solving matrix equation, and is more amenable for computation. Here,
this approach is described for the case of doubly filled orbital states.
Let us assume that each orbital state can be approximated as a linear combination
of .Ns basis states .|ϕj (1), which are not necessarily orthogonal but are independent
of each other. Then, we can in general create .Ns orbital states out of them as follows:

Ns

.|φp (1) = Ckp |ϕk (1), p = 1, · · · , Ns . (11.53)
k=1

However, note that only .Nr lowest energy orbital states will be used for calculating
the Fock operator. Employing the above expression in Eq. (11.49), we obtain

Ns
 Ns

.F̂1 [φ]|φp (1) = Ckp F̂1 [φ]|ϕk (1) = Ep Ckp |ϕk (1), (11.54)
k=1 k=1

where .Ep is now one of .Ns possible orbital energies. This is due to the use of a finite
basis and is in contrast to the fact that the number of orbital energies for the original
HF equation can be infinite.
Taking inner product of Eq. (11.54) with .ϕj (1)|, we obtain

Ns

.ϕj (1)|F̂1 [φ]|φp (1) = Ckp ϕj (1)|F̂1 [φ]|ϕk (1)
k=1
Ns

= Ep Ckp ϕj (1)|ϕk (1). (11.55)
k=1

Let us define

. Fj k = ϕj (1)|F̂1 [φ]|ϕk (1), . (11.56)

Sj k = ϕj (1)|ϕk (1). (11.57)
11.1 Hartree-Fock Approximation and Roothaan Equation 377

Then, Eq. (11.55) can be expressed as follows:

Ns

Fj k Ckp − Sj k Ckp Ep = 0, (11.58)
 
.
k=1

where .j, p = 1, · · · , Ns . The above equation is called HF-Roothaan equation and

has to be solved self consistently because the .Fj k ’s depend on the .Ckp ’s. The above
equation can be written compactly in terms of a matrix equation as follows:

.FC − SCE = 0, (11.59)

where .E is a .Ns × Ns diagonal matrix with its elements .Ep ’s.

The matrix elements of the Fock operator, defined by Eq. (11.56), can be
calculated as follows:

. ϕj (1)|F̂1 [φ]|ϕk (1) = ϕj (1)|ĥ(1)|ϕk (1)

Nr 
 
+ 2ϕj (1)|Jˆ1 [φq ]|ϕk (1) − ϕj (1)|K̂1 [φq ]|ϕj (1) , (11.60)
q=1

where

.ϕj (1)|Jˆ1 [φq ]|ϕk (1) = ϕj φq |ϕk φq 

Ns 
 Ns
= Cj∗′ q Ck ′ q ϕj ϕj ′ |ϕk ϕk ′ , . (11.61)
j ′ =1 k ′ =1

ϕj (1)|K̂1 [φq ]|ϕk (1) = ϕj φq |φq ϕk 

Ns 
 Ns
= Cj∗′ q Ck ′ q ϕj ϕj ′ |ϕk ′ ϕk . (11.62)
j ′ =1 k ′ =1

Therefore,

. ϕj (1)|F̂1 [φ]|ϕk (1) = ϕj (1)|ĥ(1)|ϕk (1)

 Ns 
Ns  Nr
Cj∗′ q Ck ′ q 2ϕj ϕj ′ |ϕk ϕk ′  − ϕj ϕj ′ |ϕk ′ ϕk  . (11.63)
 
+
j ′ =1 k ′ =1 q=1

Let us introduce
Nr

.Pj ′ k ′ =2 Cj∗′ q Ck ′ q , (11.64)
q=1
378 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

which are related to the density1 of occupied electron states in the basis of .|ϕk ,
and employ the chemist’s notation for two electron integrals. Namely, let us use
the fact that .ϕj ϕj ′ |ϕk ϕk′  = (ϕj ϕk |ϕj ′ ϕk′ ) and .ϕj ϕj ′ |ϕk′ ϕk  = (ϕj ϕk′ |ϕj ′ ϕk ). Then,
Eq. (11.63) can also be expressed as

. ϕj (1)|F̂1 [φ]|ϕk (1) = ϕj (1)|ĥ(1)|ϕk (1)

Ns
Ns 
1
 ! "
+ Pj ′ k ′ (ϕj ϕk |ϕj ′ ϕk ′ ) − (ϕj ϕk ′ |ϕj ′ ϕk ) . (11.65)
2
j ′ =1 k ′ =1

Thus, Eq. (11.59), along with Eq. (11.65), and Eq. (11.52), expressed in terms of
the same basis orbital states, can be solved together in the following self-consistent
manner:
1. Assume a set of trial .Ckp .
2. Calculate .F̂1 for the trial .Ckp .
3. Determine the new .Ckp and .Ep satisfying the restricted HF equation, Eq. (11.59).
4. Repeat steps 2 and 3 until the HF energy, Eq. (11.52), has converged sufficiently.
Note that, among the .Ns orbitals obtained, only the .Nr lowest orbitals need to
be occupied for the calculation of the ground electronic state within the HF
approximation. Other .Ns − Nr orbitals are called virtual orbitals. The secular
equations are then solved and iteration continues until satisfactory convergence is
achieved.

11.1.4 Choice of Basis Functions

The choice of the best possible basis set is a key issue in actual numerical
implementation of the HF method. This is because the computational cost of the HF
method calculation increases as the fourth power of the number of basis functions
in general. Thus, unless judicious choice of basis functions is made, numerical
implementation of the HF method can become unwieldy quite easily. Decades of
theoretical and computational research have been made for identifying efficient and
accurate enough basis functions and for numerical verification of their accuracies
[2, 14, 15, 44–46]. A comprehensive presentation of this topic goes beyond the scope
of this book. Thus, only a few key points concerning the choice of basis functions
are summarized below.
• Although using a set of orthogonal basis functions is mathematically convenient
and is possible to some extent, using atom-centered orbital functions that are
non-orthogonal in general turned out to be much more convenient and reliable

1 The matrix of .Pj k ’s is known as the charge density matrix and plays an essential role in
determining various chemical properties such as bond order, partial charges, and dipole moments,
etc.
11.1 Hartree-Fock Approximation and Roothaan Equation 379

choice. For this, the approach of using the LCAO-MO approximation serves as
an important starting point except that the primitive basis functions used for the
linear combination are not necessarily the actual atomic orbital functions but
rather are viewed as closely related to them.
• For qualitatively correct description of short and long range Coulomb inter-
actions, using Slater-type basis functions , .r n−1 Ylm (θ, φ)e−ζ r , which decay
exponentially with r , is the best choice in reproducing qualitatively correct
behavior for large r . However, numerical evaluation of Coulomb interactions
using Slater-type orbitals is costly due to slow convergence.
2
• Using Gaussian type basis functions , .x p y q zr e−αr , is much more efficient for the
evaluation of Coulomb interactions in particular because the actual integrations
can be done by utilizing analytic expressions in most cases. However, using a
single Gaussian type function as a basis is not a good choice because it does
not reproduce appropriate limiting behavior. As a solution for this, the idea of
contraction was developed.
• A contracted Gaussian basis function is a linear combination of Gaussian-type
functions with different exponents so as to best mimic a Slater type orbital. It
turned out that combination of a few primitive Gaussian functions serves as a
reasonable representation for a Slater-type orbital as far as the calculation of
important physical quantities including the energy is concerned. The coefficients
for forming a contracted Gaussian function are predetermined so as to best
approximate a Slater-type orbital and remain fixed during a HF calculation.
• A typical and most widely used HF method is to approximate the MO as a linear
combination of primitive and contracted Gaussian functions and to determine the
coefficients of the linear combination for the determination of minimum energy.
• Application of the HF method is not limited to only calculation of the electronic
state and its energy but is in fact used for identifying the structure of the
molecules in the ground electronic state. This is because it is possible to include
the variation of the molecular structure in the HF calculation so that the lowest
possible energy can be obtained varying the structure of a molecule. This is
typically known as geometry optimization.
• Although not necessary for the geometry optimization calculation of the ground
state, second derivatives of the potential energy surface, which form the Hessian
matrix (see Eq. (4.89)), around the determined structure serve as important
verification of the validity of the calculation result. This is because existence
of negative eigenvalues for the Hessian matrix indicates that the determined
structure does not even correspond to a local minimum energy, let alone the
global energy minimum2 . Eigenvalues of the Hessian matrix for the optimized
structure also provide key information on vibrational states of the molecule.

2 We describe here only the ground state, but geometry optimization is also used for determining
transition states, for which the Hessian matrix has negative eigenvalues, and for excited states as
well.
380 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

11.1.5 Methods Beyond HF Approximation

Configuration interaction (CI) [2, 14, 15, 47] goes beyond the HF approximation
by adding new excited state Slater determinants (configurations) where some of the
virtual orbitals in the HF determinant have replaced lower occupied orbitals. The
simplest of this is called CI singles (CIS), where only single occupation of only
one virtual orbitals is considered. CI singles and doubles (CISD) on the other hand
include all the slater determinants where one or two virtual orbitals are occupied.
It is known that the CI method can account for electronic correlation effects that
are missing in HF reasonably well when their contributions are relatively small.
However, for strongly correlated systems, convergence becomes slower and the
calculation becomes unmanageable. Another fundamental issue of the CI methods
is that it is not size consistent unless full contributions of CI are included. There are
alternative approaches that are generally called many-body perturbation theories,
the description of which requires more advanced formulations [2, 14, 15, 48].
It is also possible to use more than one determinants as the starting point for HF-
like calculations, which is in general called multi-configuration (MC) SCF method.
Any method that uses MC-SCF solution as the starting reference state, instead of a
single determinant HF state, is called multi-reference method [15].
The CI methods and multi-reference methods based on MC-SCF can also be
used for the calculation of excited electronic states. In addition, it is also possible
to conduct further SCF procedure for the excited states as well. A well-known
method is called complete active space SCF (CAS-SCF) [15], which focuses on
active regions of orbitals. CAS perturbation theory (CAS-PT) conducts additional
perturbation correction for CAS-SCF and can improve the accuracy significantly as
well.
Among various electronic structure calculation methods that build on the HF
approximation, the best but the most demanding method is called coupled cluster
(CC) method [15, 48], which accounts for important many-electron interactions up
to the infinite order based on theorems of diagrammatic expansion. The CC method
can be based on either single or multi-reference. The CC method is often referred
to as the gold standard and its full implementation employing multi-references is
typically used for benchmarking other lower level approximations.

11.2 Density Functional Theory

Let us define .|r; μ as the state where the position of electron .μ is at .r. Then,
#
. dr|r; μr; μ| = 1̂μ , (11.66)
11.2 Density Functional Theory 381

where .1̂μ is the identity operator in the Hilbert space of the spatial orbital states
for electron .μ. Here, we assume a little more general case where there can be
external potential applied to each electron .μ, .Vext (r), in addition to the electron
nuclei potential. Thus, we introduce

Nu #
 Zc
.V (r) =− dr + Vext (r). (11.67)
|r − Rc |
c=1

It is straightforward to find position representations of the terms constituting the

electronic Hamiltonian, Eq. (11.1), with the electron-nuclei potential replaced with
the above single electron potential. First, the single electron term can be expressed
as

p̂2μ
# # #
.ĥμ = dr dr′ |r; μr; μ| |r′ ; μr′ ; μ| + dr V (r̂μ )|r; μr; μ|
2
1
# # #
=− dr dr′ |r; μ ∇r2 δ(r − r′ )r′ ; μ| + dr V (r)|r; μr; μ|.
2
(11.68)

Second, the two electron term can be expressed as

1 1
# #
. = dr dr′ |r, r′ ; μ, νr, r′ ; μ, ν|
r̂ μν r̂μν
1
# #
= dr dr′ |r, r′ ; μ, νr, r′ ; μ, ν|, (11.69)
|r − r′ |

where .|r, r′ ; μ, ν ≡ |r; μ ⊗ |r′ ; ν.

Then, for arbitrary many-electron state .|e ,

.Eel (e ) ≡ e |Ĥel |e 

Ne #
1
#
dr dr′ e |r; μ ∇r2 δ(r − r′ )r′ ; μ|e 

=−
2
μ=1

Ne #

+ dr V (r)e |r; μr; μ|e 
μ=1

Ne  #
1 1
#
+ dr dr′ e |r, r′ ; μ, νr, r′ ; μ, ν|e .
2 |r − r′ |
μ=1 ν=μ

(11.70)
382 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

Let us define
Ne

. ρ1 (r) = e |r; μr; μ|e , . (11.71)
μ=1

Ne 
1
ρ2 (r, r′ ) = e |r, r′ ; μ, νr, r′ ; μ, ν|e . (11.72)
2
μ=1 ν=μ

In above expressions, .ρ1 (r) is the density to find an electron (one-electron density)
at position .r and .ρ2 (r, r′ ) is the density to find an electron at position .r and the other
electron at .r′ , which is called two-electron density. As expected, these satisfy the
following normalization conditions:
#
. dr ρ1 (r) = Ne , . (11.73)

Ne (Ne − 1)
# #
dr dr′ ρ2 (r, r′ ) = . (11.74)
2

The above one and two electron densities can be viewed as diagonal elements of
corresponding one and two electron density operators with their matrix elements
defined as
Ne

. γ1 (r; r′ ) = e |r; μr′ ; μ|e , . (11.75)
μ=1

Ne 
1
γ2 (r, r′ ; r′′ , r′′′ ) = e |r, r′ ; μ, νr′′ , r′′′ ; μ, ν|e . (11.76)
2
μ=1 ν=μ

It is clear that .ρ1 (r) = γ1 (r; r) and .ρ2 (r, r′ ) = γ2 (r, r′ ; r, r′ ). It is not difficult to find
that .γ1 (r; r′ ) is a sum of matrix elements of the density operator, .|e e |, reduced
to a single electron position space and that .γ2 (r, r′ ; r′′ , r′′′ ) is that reduced to two
electron position space.
The average energy of the many-electron state, Eq. (11.70), can be expressed in
terms of electron densities and density matrix elements, as follows:

1
# #
.Eel (e ) =− dr′ γ1 (r; r′ ) ∇r2 δ(r − r′ )
dr
2
ρ2 (r, r′ )
# # #
+ dr V (r)ρ1 (r) + dr dr′ , (11.77)
|r − r′ |

which appears to be much simpler than the original expression. The above expres-
sion shows that the information on .γ1 (r; r′ ), .ρ1 (r), .ρ2 (r, r′ ) and .V (r) completely
11.2 Density Functional Theory 383

determines the energy. Since all the three densities are completely determined from
’s, which are in turn fully determined once .V (r) is known because all other
.|e 

terms are intrinsic properties of the many electron system, all possible values of .Eel
and corresponding states can be considered as different functionals of .V (r).
For the ground electronic state .|e,g  or its energy .Eel,g , which can be viewed as
a unique functional of .V (r), stronger statement is in fact possible. It turns out that
the ground electronic state or energy can be viewed as a unique functional of .ρ1 (r).
An elegant proof for this was first demonstrated by Hohenberg and Kohn [49].
Theorem 11.1 For any system with fixed number of electrons, the potential energy
.V (r)is a unique functional of the one-particle electron density, .ρ1,g (r), for the ground
electronic state .|e,g .
Proof If .V (r) is not a unique functional of .ρ1,g (r), the former cannot be determined
even if the latter is known. This means that there can be multiple .V (r)’s for the same
′
.ρ1,g (r). Thus, let us take two potentials .V (r) and .V (r), which differ by more than a

constant but correspond to the same .ρ1,g (r). Let us denote the electronic Hamiltonian
for the two potentials as .Ĥel and .Ĥel′ . Note that the following relationship between these
two Hamiltonians holds:

.Ĥel − Ĥel′ = V (r̂) − V ′ (r̂). (11.78)

Let us also denote the ground electronic states for the two Hamiltonians respectively
′ , which however corresponds to the same .ρ (r). Since the two
as .|e,g  and .|e,g 1,g
potentials differ by more than a constant, .|e,g  and .|e,g′  should be different.
′
Let us first consider the fact that .Eel,g is the eigenvalue of the ground electronic state
′
.|e,g  for .Ĥel′ . As shown in Chap. 7, this means the following inequality:

′ ′
.Eel,g = e,g |Ĥ ′ |e,g
′


< e,g |Ĥ ′ |e,g  = e,g |(Ĥ + V̂ ′ − V̂ )|e,g 

#
= Eel,g + dr ρ1,g (r)(V ′ (r) − V (r)), (11.79)

where the fact that the effect of the potential can be fully accounted for by the single
electron density has been used. Similarly, switching between the two states above, we
also find that

.Eel,g = e,g |Ĥ |e,g 

′ ′ ′
< e,g |Ĥ |e,g  = e,g |(Ĥ ′ − V̂ ′ + V̂ )|e,g
′

#
′
= Eel,g − dr ρ1,g (r)(V ′ (r) − V (r)). (11.80)
384 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

Adding the above two quantities, we find that

′ ′
.Eel,g + Eel,g < Eel,g + Eel,g , (11.81)

which is contradictory. This means that .V (r) is uniquely determined up to an additive

constant factor once .ρ1,g (r) is known. Thus, .V (r) is a unique functional of .ρ1,g (r). ⊓
⊔
Since .|e,g  or .Eel,g is a unique functional of .V (r), the above theorem implies
that it is also a unique functional of the ground state single electron density. This
theorem leads to the density functional theory (DFT) [50] as summarized below.

Density Functional Theory (DFT)

• According to Hohenberg and Kohn [49], for a given system of fixed number of
electrons, the ground state energy (up to an additive constant factor) is a unique
functional of the single electron density. This implies that, if the exact form of
the functional is known, only the information on the density is sufficient to
determine the energy. However, this theorem is limited to the case where the
energy can be represented as a unique functional of an external potential .V (r),
which does not necessarily account for all possible electron densities.
• Mel Levy developed a general theorem [51], which overcame the limitation
of the theorem by Hohenberg and Kohn and proved the existence of a
universal functional for the ground state energy as a constrained minimum.
This extended the applicability of DFT significantly.
• Kohn and Sham [52] used the variational theorem and derived the following
effective one-electron equation:

1 2
# ′ 
′ n(r ) δExc (n(r))
. − ∇ + V (r) + dr + ψi (r) = Ei ψi (r),
2 |r − r′ | δn(r))
(11.82)

where atomic units were used and .n(r) = Ne 2

i |ψi (r)| , and .Exc (n(r)) is
the functional of exchange interaction. The above equation can be solved
iteratively as long as the form of .Exc (n(r)) is known. This approach is known
as Kohn-Sham DFT and has served as the major approach in the application of
DFT.

DFT as stated above is a powerful theory that allows calculation of almost infinite
number of many electron systems by a simple set of equations if the exact functional
form is known. This is a beautiful theoretical statement, but the exact form of the
functional is impossible to know in practice. Over many decades, great advances
have been made in devising approximate but fairly accurate forms of functional,
which have established the DFT method as one of the most efficient and reliable
electronic structure calculation methods [50, 53, 54].
11.3 Summary and Questions 385

Later, further extension of the above proof for the cases of time dependent
potential was established, leading to time dependent DFT (TD-DFT) theory [55].
This theory states that any time dependent potential energy that can be expanded as
polynomials of time variable can be shown to be a functional of the single electron
density, which in turn allows calculation of the excited state properties in terms of
the single electron density as well. Actual proof of TD-DFT is more involved and
the implementation of the resulting TD-DFT method also requires additional set
of new assumptions and approximations. As yet, this method has also made great
progress and is now considered as one of the leading computational methods for the
calculation of excited state properties.

11.3 Summary and Questions

For a many-electron state given by a Slater determinant of orthogonal single

electron spin-orbit states such as Eq. (11.4), the expectation value of the electronic
Hamiltonian, Eq. (11.11), is expressed as a sum of single electron and two electron
terms. The latter two electron terms in turn consist of Coulomb interactions between
electrons, which have classical analogues, and exchange interaction terms that have
negative sign originating from the Pauli exclusion principle. Thus, these exchange
interaction terms have purely quantum mechanical origin and are responsible for
unique quantum mechanical properties of many-electron systems.
The simplification of the total electronic energy for many electron states, even
though the Slater determinant involves all electronic states, reflects that the potential
energy of the Hamiltonian has only one and two electron interactions. It is also
an outcome of the orthogonality of single electron states, which changes only
the coefficient of the Slater determinant and does not restrict the nature of many
electron Slater determinant state. On the other hand, it is important that the condition
of orthogonality has to be actually incorporated as constraints in applying the
variational principle.
Application of the Lagrange multiplier method to combine the minimum energy
requirement and the orthogonality condition, leads to a general form of the HF
equation, Eq. (11.26). Further simplification of this equation into a canonical form,
Eq. (11.34), can be made using the fact the a unitary transformation diagonalizing
the Lagrange multiplier matrix . can be applied to single electron states, which
does not alter the Slater determinant except for a numerical factor of unit magnitude.
The canonical HF equation has the appearance of standard eigenvalue problem, but
it in fact needs to be determined self consistently because both the Coulomb and
exchange interaction operators are functionals of one-electron spin-orbit states yet
to be determined.
Considering the types of spin states more explicitly, it is possible to reduce the
canonical HF into a form that involves only spatial orbitals. For the case of singlet
many-electron states where all the spatial orbitals are doubly occupied, the resulting
expression is given by Eq. (11.49). In this expression, Coulomb terms have a factor
386 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

of 2 multiplied, which reflect two possible cases of single electron spins up and
down that do not affect the spatial integral of the Coulomb potential energy. On the
other hand, the exchange interaction terms do not have that factor. This is because
the exchange interaction terms become zero when the two spin-orbit states being
exchanged have opposite, thus orthogonal, spin states.
For developing efficient and general methods for solving the HF equation, it is
convenient to introduce basis states that can represent single electron orbital states
as their linear combinations. For a general nonorthogonal basis, the HF equation
can thus be converted into a HF-Roothaan equation, Eq. (11.58), which are more
amenable for numerical iterative procedures. In combination with the development
of practical and efficient ways to construct a general class of basis functions, the
HF method has been established as one of the most successful approaches for
calculating the ground electronic state energy and determining the structure of
polyatomic molecules. The HF method also serves as the starting point for more
advanced calculations for general electronic states.
Alternatively, it is possible to express the electronic energy in terms of single
and two electron densities as in Eq. (11.77). According to the proof by Hohenberg
and Kohn, the ground electronic state energy can in fact be defined as a unique
functional of a single electron density. This reduces calculating the ground state
energy of many-electron system into determining the universal functional and the
electron density that minimizes the energy for a given system, leading to the DFT
method that serve as an important alternative to the direct wavefunction formulation.
Extension of this perspective to the case of time dependent external potential led to
the TD-DFT method, which now serves as one of the most widely approaches used
for calculating the excited states of many polyatomic molecules.

Questions

• Is Fock operator always Hermitian regardless of the nature of basis set? If yes,
does it also mean that it can be related to a physical observable?
• Under what conditions, a many-electron state cannot be represented by a single
Slater determinant but requires linear combination of many Slater determinants
at least?
• Under what conditions, a many-electron state cannot be represented even by a
linear combinations of Slater determinants but requires description at the level
of density operator?
• Can any electronic state energy of a general many-body system be uniquely
determined once its single electron density is known?
Exercise Problems with Solutions 387

Exercise Problems with Solutions

11.1 The Hamiltonian for a certain three-electron system is as follows:

1 1 1
.Ĥ = ĥ1 + ĥ2 + ĥ3 + + + ,
r̂12 r̂23 r̂31

where ĥj with j = 1, 2, 3 are one electron Hamiltonians that are identical except for the
index and rj k = |rj − rk | is the distance between electron j and electron k. The state of
the three electrons is represented by the following Slater determinant:
 
 |ψ (1) |ψ (1) |ψ (1) 
1 2 3
1   
.| = √  |ψ1 (2) |ψ2 (2) |ψ3 (2)  ,

6  
|ψ1 (3) |ψ2 (3) |ψ3 (3) 

where |ψk (j ) is the kth single electron spin-orbit state with electron j . Answer the
following questions.
(a) Calculate the determinant explicitly and express | as a linear combination of
direct products
 of the single
 electron spin-orbit states.
(b) Express | ĥ1 + ĥ2 + ĥ3 | in terms of the hk where hk = ψk (j )|ĥj |ψk (j )
 
(c) Express | r̂1 + r̂1 + r̂1 | in terms of all possible Coulomb and exchange
12 23 31
interaction terms.
Solution 11.1
(a)

1
.| = √ (|ψ1 (1)|ψ2 (2)|ψ3 (3) + |ψ2 (1)|ψ3 (2)|ψ1 (3)
6
+|ψ3 (1)|ψ1 (2)|ψ2 (3)
−|ψ3 (1)|ψ2 (2)|ψ1 (3) − |ψ1 (1)|ψ3 (2)|ψ2 (3)
−|ψ2 (1)|ψ1 (2)|ψ3 (3)) .

(b)

1 1
.|ĥ1 | = (h1 + h2 + h3 + h3 + h1 + h2 ) = (h1 + h2 + h3 ),
6 3
1 1
|ĥ2 | = (h2 + h3 + h1 + h2 + h3 + h1 ) = (h1 + h2 + h3 ),
6 3
1 1
|ĥ3 | = (h3 + h1 + h2 + h1 + h2 + h3 ) = (h1 + h2 + h3 ).
6 3
388 11 Theories for Electronic Structure Calculation of Polyatomic Molecules

Therefore, |(ĥ1 + ĥ2 + ĥ3 )| = h1 + h2 + h3 .

(c) Let us define

1
.Jj k = ψj (μ)|ψk (ν)| |ψj (μ)|ψk (ν) &
r̂μν
1
Kj k = ψj (μ)|ψk (ν)| |ψk (μ)|ψj (ν),
r̂μν

which are the same for any μ and ν. Then,

1 1
.| | = (J12 − K12 + J23 − K23 + J31 − K31
r̂12 6
+J32 − K32 + J13 − K13 + J21 − K21 ) ,
1 1
| | = (J23 − K23 + J31 − K31 + J12 − K12 + J21 − K21
r̂23 6
+J32 − K32 + J13 − K13 ) ,
1 1
| | = (J31 − K31 + J12 − K12 + J23 − K23 + J13 − K13
r̂31 6
+J21 − K21 + J32 − K32 ) .

Since all three lines above the same and Jj k = Jkj and Kj k = Kkj , we find that

1 1 1
 
. | + + | = J12 − K12 + J23 − K23 + J31 − K31 .
r̂12 r̂23 r̂23

Problems

11.2 Prove that the Coulomb and exchange operators defined by Eqs. (11.17)
and (11.18) are Hermitian.
11.3 Prove Eq. (11.51) explicitly.
11.4 The Hamiltonian for a certain four-electron system is as follows:

4 4
 1 1
.Ĥ = ĥμ + ,
2 r̂μν
μ=1 μ=1 ν=μ

where the ĥμ are one electron Hamiltonians that are identical except for the index μ and
rμν = |rμ − rν | is the distance between electron μ and electron ν. The state of the four
electrons is represented by the following slater determinant:
Problems 389

 
 |ψ1 (1) |ψ2 (1) |ψ3 (1) |ψ4 (1) 

1  |ψ1 (2) |ψ2 (2) |ψ3 (2) |ψ4 (2) 

.| = √ ,
24  |ψ1 (3)

|ψ2 (3) |ψ3 (3) |ψ4 (3) 

 |ψ1 (4) |ψ2 (4) |ψ3 (4) |ψ4 (4) 

where |ψk (μ) is the kth single electron spin-orbit state with electron μ. Answer the
following questions.
(a) Calculate the determinant explicitly and express | as a linear combination of
direct products
 of the single
 electron spin-orbit states.
4
(b) Express | μ=1 ĥμ | in terms of the hk where hk = ψk (μ)|ĥμ |ψk (μ).
(c). Express |Ĥ | in terms of the hk and all possible Coulomb and exchange
interaction terms.
11.5 Prove that the kinetic energy term in Eq. (11.77) satisfies the following identity:

1 1
# # # #
. dr dr′ γ1 (r; r′ ) ∇r2 δ(r − r′ ) = dr dr′ δ(r − r′ ) ∇r2 γ1 (r; r′ ).
2 2

11.6 Equation (11.76) provides the definition of γ2 (r; r′ ), which is related to a density
operator for an electron defined in the single electron position space. Find out an exact
form of such a density operator, which is normalized and defined in the basis of |r; 1,
the position state of an electron 1.
11.7 Equation (11.75) provides the definition of γ2 (r, r′ ; r′′ , r′′′ ), which is related to a
density operator for two electrons defined in the two electron position space. Find out
an exact form of such a density operator, which is normalized and defined in the basis of
|r; 1 ⊗ |r′ ; 2, the position state of two electrons 1 and 2.
11.8 Find out expressions for the single and two electron densities for the case of Ne
electron state represented by the Slater determinant, Eq. (11.4).
Chapter 12
Special Topics

No problem is too small or too trivial if we can really do

something about it.
– Richard P. Feynman

Abstract This chapter introduces three special topics as follows: (i) path integral
representation of quantum dynamics and equilibrium properties; (ii) open system
quantum dynamics and quantum master equation approaches; (iii) Green’s function
approach. These topics have played important roles in theoretical and computational
investigation of quantum processes in complex environments. Some of the core
concepts and relations will be derived and explained.

Three special topics being introduced in this chapter serve as important theoretical
tools for investigating quantum processes in complex environments and for manipu-
lating quantum effects beyond simple measurement processes. Each of these topics
has a long history of its own and has been developed for decades. Thus, even a
reasonable overview of these topics requires substantial effort that goes beyond the
scope of this book. Rather, a few key concepts, identities, and relations that can help
readers gain better understanding and develop reasonable perspective of these topics
will be presented. Due to informative nature and incompleteness of presentations
here, no summary or problems are provided in this chapter.

12.1 Path Integral Representation

Both the time evolution operator of a quantum system and the density operator of
a canonical ensemble can in general be expressed in forms of path integrals [16,
17, 56, 57]. This is because the latter can be obtained from the former by using
.t = −iβ h̄. Although development of a path integral expression is possible with

respect to any complete basis [17, 57], the most well-known and powerful basis is

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 391
S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3_12
392 12 Special Topics

that of position states. Thus, we will consider here only conventional path integrals
in position basis.

12.1.1 Real Time Propagator

Take for example the time evolution operator for a time independent Hamiltonian,
given by Eq. (2.47), in one dimension. For the corresponding time evolution operator
.ÛH (t) defined by Eq. (2.82), let us introduce the following real time propagator

between .xi and .xf :

P (xf , t; xi , 0) ≡ xf |ÛH (t)|xi  = xf |e−it Ĥ /h̄ |xi .

. (12.1)

The time evolution operator shown above can be expressed as a product of .Np
propagators for smaller time steps. Thus,

e−it Ĥ /h̄ = e−iδt Ĥ /h̄ · · · e−iδt Ĥ /h̄ ,

. (12.2)

where .δt = t/Np . Inserting the identity resolution in position space, Eq. (2.25),
Np − 1 times in between each time evolution operator for .δt, we can express
.

Eq. (12.1) as

  Np

.P (xf , t; xi , 0) = dx1 · · · dxNP −1 P (xk , tk ; xk−1 , tk−1 ), (12.3)
k=1

where .xNp = xf , .x0 = xi , and .tk = tδt for . k = 0, · · · , Np .

For small enough .δt, one can make the following Trotter factorization [58]
approximation for each time evolution operator for .δt.

e−iδt Ĥ /h̄ = e−iδt V̂ /(2h̄) e−iδt T̂ /h̄ e−iδt V̂ /(2h̄) + O(δt 3 ),

. (12.4)

where .O(δt 3 ) implies that the error for the above Trotter factorization is of order
3
.δt . Within this approximation, the corresponding propagator can be calculated

employing the identity resolution in the momentum space, Eq. (2.34), as follows:
 ∞
. P (xk , tk ; xk1 , tk−1 ) ≈ dp xk |e−iδt V̂ /(2h̄) |pp|e−iδt T̂ /h̄ e−iδt V̂ /(2h̄) |xk−1 
−∞
 ∞ 2 /(2mh̄)
= e−iδt (V (xk )+V (xk−1 ))/(2h̄) dp xk |pe−iδtp p|xk−1 
−∞
1
 ∞ 2 /(2mh̄)+i(x
−iδt (V (xk )+V (xk−1 ))/(2h̄)
=e dp e−iδtp k −xk−1 )p/h̄ . (12.5)
2π h̄ −∞
12.1 Path Integral Representation 393

In the last equality of the above expression, Eq. (2.36) and its complex conjugate
have been used.
Note that the integration over momentum in the above expression can be done
by completing the square in the exponent and conducting Gaussian integration as
follows:
 ∞
2
. dp e−iδtp /(2mh̄)+i(xk −xk−1 )p/h̄
−∞
2 /(2h̄δt)
 ∞ 2
= eim(xk −xk−1 ) dp e−iδt/(2mh̄)(p−m(xk −xk−1 )/δt)
−∞

2π mh̄ im(xk −xk−1 )2 /(2h̄δt)
= e . (12.6)
iδt

Employing the above identity in Eq. (12.5), we obtain


m 2
. P (xk , tk ; xk1 , tk−1 ) = eim(xk −xk−1 ) /(2h̄δt)−iδt (V (xk )+V (xk−1 ))/(2h̄) + O(δt 3 ).
2iπ h̄δt
(12.7)

As is clear from the second term, the error involved in using the first term in the
above expression is of order .δt 3 = (t/NP )3 . When this expression is used in
Eq. (12.3), this error is added .Np times and the net error of the resulting expression
is of order .t 3 /NP2 . Thus, Eq. (12.3) can be expressed as

 Np   
t3
 
mNp 2 i AN (xN )/h̄
P (xf , t; xi , 0) =
. dx1 · · · dxNP −1 e p p +O ,
2iπ h̄t Np2
(12.8)

where .xNp ≡ (x0 , x1 , x2 , · · · , xNp ), the vector of discrete position trajectories from
x0 = xi at .t = 0 to .xNP = xf at t, and .ANp (xNp ) is a function of this vector defined
.

as
Np
m(xk − xk−1 )2 V (xk ) + V (xk−1 )
ANp (xNp ) ≡
. − δt
2δt 2
k=1
⎡ ⎤
Np  2 
V (x0 ) + V (xNp )

m xk − xk−1
= δt ⎣ − V (xk ) − ⎦.
2 δt 2
k=1

(12.9)
394 12 Special Topics

Let us now consider the limit of .NP → ∞, for which .xNp becomes a continuous
trajectory of time .x(t) and Eq. (12.9) becomes the following functional of the
trajectory known as action functional:
 2 
 t m

dx(τ )
. lim ANp (xNp ) = A[x(·); t] ≡ dτ − V (x(τ )) .
NP →∞ 0 2 dτ
(12.10)

In the above expression, .x(·) represents a given function .x(τ ) as a whole defined for
the entire domain of time, .τ . The dot .· is used as an argument here to distinguish it
from a particular value of the function at a certain time t.
Let us also define the following path integral as the limit of the multidimensional
integration in Eq. (12.8) as follows:

  Np    x(t)=xf
mNp 2
. lim dx1 · · · dxNP −1 (· · · ) = Dx(·)(· · · ),
NP →∞ 2iπ h̄t x(0)=xi
(12.11)

where dots within parentheses imply any functional that is well defined and remains
integrable as the discrete trajectory becomes continuous. Whether this limit of path
integration is well defined is a nontrivial mathematical issue and also has practical
significance. For example, for a free particle or a particle subject to a potential
bounded from below, it is generally accepted that the path integral is well-defined as
the limit of the multidimensional integration. However, for singular potentials such
as Coulomb potential, it is not the case and more careful definition of the measure
of integration1 becomes necessary [17].
Since .t 3 /NP2 becomes zero in the limit of .NP → ∞ for any finite value of t, the
error term in Eq. (12.8) becomes zero in that limit. Thus, we obtain the following
path integral expression for the propagator:
 x(t)=xf
P (xf , t; xi , 0) =
. Dx(·)ei A[x(·);t]/h̄ . (12.12)
x(0)=xi

Noting the definition of the propagator, Eq. (12.1), the above expression can now be
used to obtain the following expression for the real time propagator:
 
  x(t)=xf
i A[x(·);t]/h̄
ÛH (t) =
. dxf dxi Dx(·)e |xf xi |. (12.13)
x(0)=xi

1 Even for integration of functions, this is an important issue because there are kinds of singular
functions for which conventional Riemann measure integration fails but more elaborate Lebesgue
measure produces convergent result.
12.1 Path Integral Representation 395

Employing the above expression and its Hermitian conjugate in Eq. (10.45), the time
dependent density operator can now be expressed as
   
ρ̂(t) =
. dxf dxi dxf′ dxi′ |xf xi |ρ̂|xi′ xf′ |

x ′ (t)=xf′
 
x(t)=xf 
′ i(A[x(·);t]−A[x ′ (·);t])/h̄
× Dx(·) Dx (·) e .
x(0)=xi x ′ (0)=xi′

(12.14)

Accordingly, the time dependent average for any operator .Ô, as defined by
Eq. (10.48), with .ρ̂ replaced by .ρ̂(t), can be expressed as
 
Ô(t) = T r Ô ρ̂(t)
.


= dx x|Ô ρ̂(t)|x
   
= dxf dxi dxf′ dxi′ xf′ |Ô|xf xi |ρ̂|xi′ 

x ′ (t)=xf′
 
x(t)=xf 
[x ′ (·);t])/h̄
× Dx(·) Dx ′ (·) ei(A[x(·);t]−A .
x(0)=xi x ′ (0)=xi′

(12.15)

In the second equality of the above expression, .xf′ |x = δ(xf′ − x) has been used.
Extension of expressions derived here to general multidimensional Cartesian
coordinates is straightforward [17, 57]. However, for curvilinear coordinates or
for cases with cyclic coordinates such as angles, care should be taken for proper
definition of path measures [17].

12.1.2 Imaginary Time Propagator

Let us consider the canonical equilibrium density operator, Eq. (10.63), which can
also be expressed as

e−i(−i h̄β)Ĥ /h̄

.ρ̂β =  . (12.16)
T r e−i(−i h̄β)Ĥ /h̄

Thus, the path integral expression for the canonical equilibrium density operator can
be obtained by simply replacing t with .−i h̄β. Let us introduce an imaginary time .tI
such that
396 12 Special Topics

.t = −itI , 0 ≤ tI ≤ h̄β (12.17)

Then, the definition of the real time propagator, Eq. (12.1), can be extended to an
imaginary time propagator as follows:

xf |e−β Ĥ |xi  = P (xf , −iβ h̄; xi , 0) = xf |e−i(−iβ h̄)Ĥ /h̄ |xi .
. (12.18)

Thus, the action given by Eq. (12.9) can be extended to the imaginary time at
t = β h̄, by simply replacing the time integration variable .τ with .−iτ and also
. I

introducing .xe (τ ) ≡ x(−iτ ), as follows:

 2 
 β h̄ m

dxe (τ )
A[x(·); β h̄] = −i
. dτ − V (xe (τ ))
0 2 −idτ
 2 
 β h̄ m

dxe (τ )
=i dτ + V (xe (τ )) ≡ iE[xe (·); β h̄],
0 2 dτ
(12.19)

where .E[xe (·); β h̄] is called Euclidean action and is equal to the average energy
along the path time leading to .β h̄. Note that .xe (τ ) and .x(−iτ ) are the same except
that the former is viewed as a function of a real valued parameter .τ whereas the latter
is viewed as that of an imaginary time .−iτ . Thus, the imaginary time propagator
defined by Eq. (12.18) can be expressed by the following path integral form:
 xe (β h̄)=xf
xf |e−β Ĥ |xi  =
. Dxe (·) e−E[xe (·);β h̄]/h̄ , (12.20)
xe (0)=xi

where the detailed form of the path integral measure can be obtained by using .t =
−iβ h̄ in Eq. (12.11) and is given by

 xe (β h̄)=xf   Np  
mNp 2
. Dxe (·)(· · · ) ≡ lim dx1 · · · dxNP −1 (· · · ).
xe (0)=xi NP →∞ 2πβ h̄2
(12.21)

Equations (12.20) and (12.21) complete the derivation of the path integral repre-
sentation for the imaginary time propagator. Alternatively, these expressions can be
obtained directly by expressing .e−β Ĥ as a product of those for higher temperature,
.e
−ǫ Ĥ , where .ǫ = β/N = 1/(k T N ). Application of the Trotter factorization for
P B P
each of .e −ǫ Ĥ as in Eq. (12.4), conducting Gaussian integration as in Eq. (12.6), but
for .δt = −iβ h̄/NP , and sending the limit of .NP → ∞ directly leads to Eqs. (12.20)
and (12.21).
12.2 Quantum Master Equation for Open System Quantum Dynamics 397

Employing Eq. (12.20) for .xf = x ′ and .xi = x ′′ , the unnormalized canonical
density operator can be expressed as
xe (β h̄)=x ′
   
−β Ĥ ′ ′′ ′ ′′ −E[xe (·);β h̄]/h̄
.e = dx dx |x x | Dxe (·) e . (12.22)
xe (0)=x ′′

Taking the trace of this, we also obtain the following path integral expression for the
partition function:
  
−β Ĥ
.Zβ = T r e = dx ′ x ′ |e−β Ĥ |x ′ 
  x(t)=x ′
= dx ′ Dxe (·) e−E[xe (·);β h̄]/h̄
x(0)=x ′

≡ Dx(·) e−E[xe (·);β h̄]/h̄ , (12.23)

where the path integration in the last line represents sum over all the cyclic paths that
end at the same position as the starting point. Combining Eqs. (12.22) and (12.23),
we thus obtain the following path integral expression for the canonical equilibrium
density operator:
 ′  ′′ ′ ′′  xe (β h̄)=x ′ −E[xe (·);β h̄]/h̄

e −β Ĥ dx dx |x x | xe (0)=x ′′ Dxe (·) e
.ρ̂β = = .
Zβ Dx(·) e−E[xe (·);β h̄]/h̄


(12.24)
Therefore, the average of any operator over the canonical ensemble can be expressed
as
x (β h̄)=x ′
 
dx dx x |Ô|x ′  xee(0)=x ′′ Dxe (·) e−E[xe (·);β h̄]/h̄
 ′  ′′ ′′
 
.T r Ô ρ̂β = .
Dx(·) e−E[xe (·);β h̄]/h̄


(12.25)

Extension of the expressions derived above to the case of multidimensional

Cartesian coordinate system is straightforward [17, 57]. As is the case with the
time evolution operator, mathematical difficulty arises in identifying correct path
measure for curvilinear or cyclic coordinates [17].

12.2 Quantum Master Equation for Open System Quantum

Dynamics

Consider a quantum system interacting with its environment. Assume that we are
interested only in the net effects of the environment on the system and are not
concerned about details of the environment. For these cases, it may be possible
398 12 Special Topics

to develop ways to consider a reduced quantum dynamics for the system only
while only accounting for the net effects of the environment on the system. To
what extent is this approach feasible? How can one identify the whole environment
and its interaction with the system of interest? These are some of difficult and
fundamental questions that still remain active issues of research in open system
quantum dynamics. However, in this section, we will bypass asking such questions
and simply assume that it is possible to identify the environment, which will be
called bath, and its interaction with the system of interest.
To be more specific, let us assume that it is possible to identify a well-defined
system Hamiltonian .Ĥs , a bath Hamiltonian .Ĥb , and a system-bath interaction
Hamiltonian .Ĥsb , which are all defined in the direct product space of the system
and the bath and can be added together to define a total Hamiltonian as follows:

Ĥ = Ĥs + Ĥsb + Ĥb = Ĥ0 + Ĥ1 ,

. (12.26)

where, in the second equality, we have also introduced the zeroth order reference
Hamiltonian, .Ĥ0 , and the first order term .Ĥ1 that is assumed to be small2 compared
to .Ĥ0 . Definitions of these can vary depending on the nature of the problem. If .Ĥsb
is small, defining .Ĥ0 = Ĥs + Ĥb and .Ĥ1 = Ĥsb is a natural choice. However, an
alternative choice is possible if .Ĥsb is not small enough.
Given that the system and bath as a whole is completely isolated from the rest
of the universe, the time evolution of the total density operator representing the
sum of system and bath at time t is determined according to the following quantum
Liouville equation:

d
. ρ̂(t) = −iLρ̂(t)
dt
= −i (Ls + Lsb + Lb ) ρ̂(t)
= −i (L0 + L1 ) ρ̂(t). (12.27)

In the above expression, .L(· · · ) = [Ĥ , (· · · )]/h̄ and .Lα (· · · ) = [Ĥα , (· · · )]/h̄, for
each of .α = s, sb, b, and 0, respectively.
There are two major approaches to reduce the dynamics of the total density
operator represented by Eq. (12.27) into that involving only the system degree
of freedom. One is to use the path integral influence functional approach, which
accounts for the net effect of the bath on the system dynamics in the path integral
representation of the density operator. This is a powerful approach that has widely
been used in the condensed matter physics and remains a promising approach for
simulating molecular systems, but is not described here. Readers can refer to an
excellent monograph [59] and a review paper [60] on this approach. The other

2 The meaning of “small" here is the extension of the concept used in the perturbation theory in
that the effects of .Ĥ1 on the eigenstates of .Ĥ0 are small.
12.2 Quantum Master Equation for Open System Quantum Dynamics 399

approach is to use a quantum master equation (QME) [37, 61–64], which has longer
history and has been well-established as one of the most general approaches for
investigating open system quantum dynamics.
One of the most systematic approaches to derive QMEs is to employ a projection
super-operator in the interaction picture with respect to the zeroth order Hamilto-
nian. For this, let us first introduce the density operator in the interaction picture
with respect to .Ĥ0 as follows:

ρ̂I (t) = ei Ĥ0 t/h̄ ρ̂(t)e−i Ĥ0 t/h̄ .

. (12.28)

The dynamics of this is governed by the following interaction picture quantum

Liouville equation:

d
. ρ̂I (t) = iL0 ρ̂I (t) − ei Ĥ0 t/h̄ (iL0 + iL1 ) ρ̂(t)e−i Ĥ0 t/h̄
dt
= −ei Ĥ0 t/h̄ iL1 ρ̂(t)e−i Ĥ0 t/h̄
i
= − [Ĥ1,I (t), ρI (t)] ≡ −iL1,I (t)ρ̂I (t), (12.29)
h̄

where .Ĥ1,I (t) = ei Ĥ0 t/h̄ Ĥ1 e−i Ĥ0 t/h̄ .

12.2.1 Projection Operator Formalism and Exact Time

Evolution Equations for a Projected Density Operator

If only a partial information on the density operator is needed, a projection super-

operator .P, which is assumed to be time independent, can be introduced such that
.Pρ̂I (t) contains all such information. The complement of the projection super-

operator is denoted as .Q = 1 − P. So, let us call .Pρ̂I (t) the projected part of the
density operator and .Qρ̂I (t) the unprojected part of the density operator. First, we
will show that a general exact time evolution equation involving only the projected
part of the density operator can be derived without detailed specification of the form
of the projection super-operator. In deriving this, it is convenient to assume that the
following identity holds:

.PL1,I (t)P = 0. (12.30)

It is important to note that the above condition is not so restrictive and can always
be satisfied by an appropriate definition of .Ĥ1 . As will be clear, this identity helps to
simplify final expression for the formal solution and its perturbative approximations.
400 12 Special Topics

Applying .P on the left-hand side of Eq. (12.29), inserting the identity .P + Q = 1̂

between .L1,I (t) and .ρ̂I (t), and employing the identity of Eq. (12.30), we obtain

d
. Pρ̂I (t) = −iPL1,I (t)Qρ̂I (t). (12.31)
dt

Similarly, applying .Q on the left-hand side of Eq. (12.29) and inserting the identity
of .P + Q = 1̂, we also obtain

d
. Qρ̂I (t) = −iQL1,I (t)Qρ̂I (t) − iQL1,I (t)Pρ̂I (t). (12.32)
dt
A formal solution of the above equation can be found easily. For this, it is convenient
to introduce time ordered exponential super-operators as described below.

Time Ordered Exponential Super-Operators

Given a certain time dependent super-operator .M(t) such as interaction
picture Liouville or projected/unprojected interaction picture Liouville super-
operator and for .t ≥ 0, a positively time ordered exponential super-operator
is defined as
t  t
−i 0 dτ M(τ )
.e
(+) ≡ 1−i dτ M(τ )
0
 t  τ
+(−i)2 dτ dτ ′ M(τ )M(τ ′ )
0 0
+···
 t  τ1  τn−1
+(−i)n dτ1 dτ2 · · · dτn M(τ1 ) · · · M(τn )
0 0 0
+··· , (12.33)

which satisfies the following time evolution equation:

d −i 0t dτ M(τ ) t
−i dτ M(τ )
 
. e(+) = −iM(t)e(+) 0 . (12.34)
dt
On the other hand, a negatively time ordered exponential super-operator is
defined as
t  t
i 0 dτ M(τ )
.e
(−) ≡ 1 + i dτ M(τ )
0
 t  τ
+i 2
dτ dτ ′ M(τ ′ )M(τ )
0 0
(continued)
12.2 Quantum Master Equation for Open System Quantum Dynamics 401

+···
 t  τ1  τn−1
+i n dτ1 · · · dτ2 · · · dτn M(τn ) · · · M(τ1 )
0 0 0
+··· , (12.35)

which satisfies the following time evolution equation:

d i 0t dτ M(τ ) t
i dτ M(τ )
 
. e(−) = ie(−)0 M(t). (12.36)
dt
The two time ordered exponential super-operators defined above form an
inverse relationship as follows:
t t t t
−i dτ M(τ ) i 0 dτ M(τ ) i dτ M(τ ) −i 0 dτ M(τ )
e(+) 0
. e(−) = e(−)0 e(+) = 1̂. (12.37)

Moving the first

t
term on the righthand side of Eq. (12.32) to the lefthand side,
i 0 dτ QL1,I (τ )
and applying .e(−) , which is defined by Eq. (12.35) for .M(t) = QL1,I (t)
on both sides of the equation, we find that
t t
i 0 dτ QL1,I (τ ) d i 0 dτ QL1,I (τ )
. e(−) Qρ̂I (t) + e(−) iQL1,I (t)Qρ̂I (t)
dt
 t  t
d i 0 dτ QL1,I (τ ) i dτ QL1,I (τ )
= e(−) Qρ̂I (t) = −ie(−)0 QL1,I (t)Pρ̂I (t),
dt
(12.38)

where the second equality is obtained by using Eq. (12.36) with .M(t) = QL1,I (t).
Integrating the above equation over time from 0 to t, we obtain
t  t τ
i dτ QL1,I (τ ) i dτ ′ QL1,I (τ ′ )
e(−)0
. ρ̂I (t) − ρ̂I (0) = −i dτ e(−)0 QL1,I (τ )Pρ̂I (τ ).
0
t
(12.39)
−i 0 dτ QL1,I (τ )
Applying .e
(+)which is defined by Eq. (12.33) for .M(t) = QL1,I (t),
,
on both sides of the equation, we find that
 t t t ˆ (τ )
−i dτ ′ QL1,I (τ ′ ) −i dτ QL 1,I
Qρ̂I (t) = −i
. dτ e(+) τ QL1,I (τ )Pρ̂I (τ )+e(+) 0 Qρ̂I (0),
0
(12.40)
402 12 Special Topics

where the following relationship has been used.

t τ t
−i dτ QL1,I (τ ) i 0 dτ ′ QL1,I (τ ′ ) −i dτ QL1,I (τ )
e(+) 0
. e(−) = e(+) τ . (12.41)

The validity of Eq. (12.40) as the solution of Eq. (12.32) can also be proven by direct
substitution.
Note that Eq. (12.40) expresses .Qρ̂I (t) in terms of .Pρ̂I (τ ) for .0 ≤ τ ≤ t and the
initial value of the unprojected part of the density operator at .t = 0. Thus, when this
expression is used in Eq. (12.31), we obtain a closed form of differential equation
for .Pρ̂I (t) as summarized below.

Exact Time-Nonlocal Evolution Equation for the Projected Part of the

Density Operator
Using Eq. (12.40) in Eq. (12.31), we obtain the following time evolution
equation for the projected part of the density operator:

d
 t t
−i dτ ′ QL1,I (τ ′ )
. Pρ̂I (t) = − dτ PL1,I (t)e(+) τ QLI (τ )Pρ̂I (τ )
dt 0
t
−i 0 dτ QL1,I (τ )
−iPL1,I (t)e(+) Qρ̂(0), (12.42)

where the inhomogeneous term (the second term) retains the information on
the initial preparation of the system and the bath at .t = 0. Note that the
above expression involves only the positively time ordered exponential super-
operator involving .QL1,I (t). Thus, as long as this is well defined, the above
solution is also well-defined and valid.

The time evolution equation, Eq. (12.42), is non-local in time and shows that
even calculating the time derivative of .Pρ̂I (t) requires full information on its past.
This is not unexpected since the effect of the bath on the dynamics of the system
builds up over the history of its time evolution. However, it turns out that it is also
possible to account for such memory effect employing a fully time local form. This
is because the full dynamics for .ρ̂I (t) is unitary and can be reversed backward. That
is, one can always find a formally exact solution for Eq. (12.29), which can relate
.ρ̂I (t) to .ρ̂I (τ ) for .τ ≤ t as follows:

t
−i dτ ′ L1,I (τ ′ )
ρ̂I (t) = e(+) τ
. ρ̂(τ ). (12.43)

Equivalently, one can also find the following relationship:

t
i dτ ′ L1,I (τ ′ )
ρ̂I (τ ) = e(−)τ
. ρ̂I (t). (12.44)
12.2 Quantum Master Equation for Open System Quantum Dynamics 403

Employing the above identity in Eq. (12.40), we find that

 t t t
−i dτ ′ QL1,I (τ ′ ) i dτ ′ L1,I (τ ′ )
.Qρ̂I (t) = −i dτ e(+) τ QL1,I (τ )Pe(−)τ ρ̂I (t)
0
t
−i 0 dτ QL1,I (τ )
+e(+) Qρ̂(0). (12.45)

Inserting .P + Q = 1̂ in front of .ρ̂I (t) on the righthand side of the above equation
results in two terms. One involves .Pρ̂I (t) and the other involves .Qρ̂I (t). Moving
the term involving the latter to the lefthand side of the equation and applying the
inverse3 of the entire operator being applied to .Qρ̂I (t) [65], we now obtain the
following key relationship:

.Qρ̂I (t) = (1 + iŴ1,I (t))−1

 t t t
−i τ dτ ′ QL1,I (τ ′ ) i τ dτ ′ L1,I (τ ′ )
× −i dτ e(+) QL1,I (τ )Pe(−) Pρ̂I (t)
0
t
−i dτ QL1,I (τ )
+e(+) 0 Qρ̂(0) , (12.46)

where
 t t t
−i τ dτ ′ QL1,I (τ ′ ) i τ dτ ′ L1,I (τ ′ )
Ŵ1,I (t) =
. dτ e(+) QL1,I (τ )Pe(−) . (12.47)
0

Equation (12.46) can be combined with Eq. (12.31), to find an exact time-local
equation governing the dynamics of .Pρ̂I (t) as summarized below.

Exact Time-Local Equation for the Projected Part of the Density

Operator
Using Eq. (12.46) in Eq. (12.31), we obtain the following time evolution
equation for the projected part of the density operator:

d
. Pρ̂I (t) = −PL1,I (t)(1 + i Ŵ̂1,I (t))−1
dt
 t t t ˆ (τ ′ )
−i dτ ′ QL1,I (τ ′ ) i dτ ′ L1,I
× dτ e(+) τ QL1,I (τ )Pe(−)τ Pρ̂I (t)
0
t
−i dτ QL1,I (τ )
−1 t0
−iPL1,I (t)(1 + iŴ1,I (t)) e(+) Qρ̂(0).
(12.48)

(continued)

3 It is assumed here that this inverse can always be found.

404 12 Special Topics

Note that the righthand side of the above equation involves the information
on .Pρ̂I (t) only at the same time as that for the lefthand side. This is achieved
by expressing the information on .ρ̂I (τ ) for .τ ≤ t in terms of .ρ̂I (t) by
evolving backward in time and then dividing their effects on projected and
unprojected parts again. As a result, the final forms of super-operators in the
above expression is more complicated than those in Eq. (12.42).

Equations (12.42) and (12.48) are both exact and general for any form of
the Liouville and projection super-operators. Alternative methods to derive these
equations exist as well. For example, one can employ the generalized cumulant
expansion method [66] for the total density operator and then apply chronological
time ordering or partial time ordering prescription [67, 68].

12.2.2 Quantum Master Equations for a Reduced System

Density Operator

If we are interested in finding out the general information on the system of interest,
we need to be able to determine at least the reduced system density operator defined
as

ρ̂s,I (t) ≡ T rb {ρ̂I (t)}.

. (12.49)

For this, we can introduce the following projection super-operator [37, 62–64]:

P(·) ≡ ρ̂b ⊗ T rb {(·)} = ρ̂b T rb {(·)} ,

. (12.50)

where .ρ̂b is a bath density operator, which is commonly assumed to be an

equilibrium density operator, and the second equality is a short notation for the
direct product form. In addition, let us define the zeroth order and the first order
Hamiltonians as follows:

. Ĥ0 = Ĥs + Ĥb , . (12.51)

Ĥ1 = Ĥsb . (12.52)

Then, the time evolution operator for .Ĥ0 can be expressed as

Û0 (t) = e−i Ĥs /h̄ e−i Ĥb t/h̄ .

. (12.53)
12.2 Quantum Master Equation for Open System Quantum Dynamics 405

Therefore,

. Ĥ1,I (t) = Ĥsb,I (t) = ei Ĥs t/h̄ ei Ĥb t/h̄ Ĥsb e−i Ĥb t/h̄ e−i Ĥs t/h̄ . (12.54)

For the above case, the condition of Eq. (12.30) implies that, for any operator .Ô that
depends on both system and bath degrees of freedom, the following identity holds.

1   
PL1,I (t)P(Ô) =
. ρ̂b T rb Ĥ1,I (t), ρ̂b T rb Ô
h̄
1    
= ρ̂b T rb Ĥ1,I (t)ρ̂b T rb Ô
h̄
   
−T rb ρ̂b T rb Ô Ĥ1,I (t)

1     
= ρ̂b T rb Ĥ1,I (t)ρ̂b T rb Ô
h̄
   
−T rb Ô T rb ρ̂b Ĥ1,I (t)

1    
= ρ̂b T rb Ĥ1,I (t)ρ̂b , T rb Ô = 0. (12.55)
h̄

In the
 second
 equality of the above equations, the fact that .ρ̂b commutes with
.T rb Ô , which becomes independent of the bath after trace over the bath has
   
been used. However, note that .T rb Ô and .T rb Ĥ1,I (t)ρ̂b , which still remain
operators in the system space, do not commute in general. Thus, for the identity to
hold for any kind of .Ô, the latter has to become zero as follows:
   
T rb Ĥ1,I (t)ρ̂b = ei Ĥs t/h̄ T rb ei Ĥb t/h̄ Ĥsb e−i Ĥb t/h̄ ρ̂b e−i Ĥs t/h̄ = 0.
. (12.56)
 
For .ρ̂b = e−β Ĥb /T rb e−β Ĥb or any other form of .ρ̂b that commutes with .Ĥb , the
 
above condition implies .T rb Ĥsb ρ̂b = 0, which means that the average of system-
bath interaction over the bath density operator is zero. In case .Ĥsb does not satisfy
this 
condition,
 it is always possible to redefine
 the
 system Hamiltonian by adding
.T rb Ĥsb ρ̂b and then define .Ĥsb − T rb Ĥsb ρ̂b as a new system-bath interaction

term. Thus, in order to guarantee the condition of Eq. (12.30), here we assume that
 
. Ĥb , ρ̂b = 0, . (12.57)
 
T rb Ĥsb ρ̂b = 0. (12.58)
406 12 Special Topics

12.2.2.1 Formally Exact QMEs

For the present case, the time-nonlocal time evolution equation for the projected
part of the density operator, Eq. (12.42), is expressed as

d
 t   t 
. Pρ̂I (t) = − dτ PLsb,I (t) exp(+) −i dτ ′ QLsb,I (τ ′ ) QLsb,I (τ )Pρ̂I (τ )
dt 0 τ
  t 
−iPLsb,I (t) exp(+) −i dτ QLsb,I (τ ) Qρ̂I (0), (12.59)
t0

where .Lsb,I (t)(·) ≡ [Ĥsb,I (t), (·)]/h̄. The above equation can be used to obtain the
exact time-nonlocal QME for the reduced system density operator in the interaction
picture, as defined by Eq. (12.49).

Exact Time-Nonlocal QME for Reduced System Density Operator

Taking trace of Eq. (12.59) over the bath degrees of freedom, we obtain the
following formally exact time-nonlocal QME for reduced system density
operator in the interaction picture, .ρ̂s,I (t) :

d
 t
. ρ̂s,I (t) = − dτ Ksb (t, τ )ρ̂s,I (τ ) + Ic,sb (t), (12.60)
dt 0
where
t
−i τ dτ ′ QLsb,I (τ ′ )
. Ksb (t, τ ) = T rb Lsb,I (t)e(+) QLsb,I (τ )ρ̂b , (12.61)
.

t
−i dτ QLsb,I (τ )
Ic,sb (t) = −iT rb Lsb,I (t)e(+) 0 Qρ̂I (0) . (12.62)

In the Schrödinger picture, the corresponding time-nonlocal QME

can be obtained by taking derivative of .ρ̂s (t) = e−i Ls t ρ̂s,I (t) =
e−i Ĥs t/h̄ ρ̂s,I (t)ei Ĥs t/h̄ and employing Eq. (12.60). The resulting expression
is as follows:
 t
d −i Ls t
 
. ρ̂s (t) = −iLs ρ̂s (t) − e dτ Ksb (t, τ ) ei Ĥs τ/h̄ ρ̂s (τ )e−i Ĥs τ )/h̄
dt 0

+e−i Ls t Ic,sb (t). (12.63)

On the other hand, under the same assumptions of Eqs. (12.57) and (12.58), the
time-local QME for .ρ̂s,I (t), Eq. (12.48), is expressed as
12.2 Quantum Master Equation for Open System Quantum Dynamics 407

d
. Pρ̂I (t) = −PLsb,I (t)(1 + iŴsb,I (t))−1
dt
 t t t
−i dτ ′ QLsb,I (τ ′ ) i dτ ′ Lsb,I (τ ′ )
× dτ e(+) τ QLsb,I (τ )Pe(−)τ PρI (t)
0
t ˆ
−1 −i 0 dτ QLsb,I (τ )
−iPLsb,I (t)(1 + iŴsb,I (t)) e(+) Qρ̂(0).
(12.64)

where .Ŵsb,I (t) is defined by Eq. (12.47) with .L1,I (t) replaced by .Lsb,I (t). The
exact time-local QME for .ρ̂s,I (t) thus can be obtained easily from the above
equation as summarized below.

Exact Time-Local QME for Reduced System Density Operator

Taking trace of Eq. (12.64) over the bath degrees of freedom, we obtain the
following exact time-local QME for the reduced system density operator in
the interaction picture:

d
. ρ̂s,I (t) = −Rsb (t)ρ̂s,I (t) + Il,sb (t), (12.65)
dt
where
 t
. Rsb (t) = dτ T rb Lsb,I (t)(1 + iŴsb,I (t))−1
0
t t
−i dτ ′ QLsb,I (τ ′ ) i dτ ′ Lsb,I (τ ′ )
×e(+) τ QLsb,I (τ )Pe(−)τ ρ̂b ,

. (12.66)
t
−i dτ QLsb,I (τ )
Il,sb (t) = −iT rb Lsb,I (t)(1 + iŴsb,I (t, t0 ))−1 e(+) 0 Qρ̂(0) .

(12.67)

In the Schrödinger picture, the corresponding time-local QME can

be obtained by taking the derivative of .ρ̂s (t) = e−i Lt ρ̂s,I (t) =
e−i Ĥs t/h̄ ρ̂s,I (t)ei Ĥs t/h̄ and employing Eq. (12.65). The resulting expression
is
d  
. ρ̂s (t) = −iLs ρ̂s (t) − e−i Ls t Rsb (t) ei Ĥs t/h̄ ρ̂s (t)e−i Ĥs t/h̄
dt
+e−i Ls t I (t). l,sb (12.68)
408 12 Special Topics

12.2.2.2 Second Order QMEs

The kernel in time-nonlocal QME, the relaxation super-operator in time-local

QME, and the inhomogeneous terms in both equations can be expanded in terms
of .Ĥsb,I (t). The lowest order approximations with respect to .Ĥsb,I (t) for both
equations in the interaction picture are of second order.
Let us first consider the time-nonlocal QME, Eq. (12.60). Keeping only terms
up to the second order of .Lsb,I (t) in Eqs. (12.61) and (12.62), and employing the
identity of Eq. (12.30), we obtain the following second order approximations for the
terms involved in the time-nonlocal QME,

(2)
Ksb (t, τ ) = T rb Lsb,I (t)Lsb,I (τ )ρ̂b , .
. (12.69)

I(2)
c,sb (t) = −iT rb L̂sb,I (t)ρ̂I (0)
 t
− dτ T rb Lsb,I (t)Lsb,I (τ )Qρ̂I (0) . (12.70)
0

For the time-local QME, Eq. (12.65), similar 2nd order approximations can be
made in Eqs. (12.66) and (12.67) as follows:
 t  t
(2) (2)
. Rsb (t) = dτ T rb Lsb,I (t)Lsb,I (τ )ρ̂b = dτ Ksb (t, τ ), (12.71)
.
0 0

I(2) (1)
l,sb (t) = −iT rb Lsb,I (t)(1 − iŴsb,I (t))Qρ̂I (0)
 t
− dτ T rb Lsb,I (t)QLsb,I (τ )Qρ̂I (0) . (12.72)
0

(1) t (1)
In the above expression, .Ŵsb,I (t) = t0 dτ QLsb,I (τ )P. Clearly, .Ŵsb,I (t)Qρ̂I (0) = 0
because .PQ = 0. Therefore,
 t
(2)
. Il,sb (t) = −iT rb L̂sb,I (t)ρ̂I (0) − dτ T rb Lsb,I (t)Lsb,I (τ )Qρ̂I (t0 )
t0
(2)
= Ic,sb (t). (12.73)

Thus, the relaxation super-operator for the 2nd order time-local QME is equal
to the integration of the kernel of the 2nd order time-nonlocal QME, while the
inhomogeneous terms for both equations are the same.
In the Schrödinger picture, the corresponding second order equations can
be obtained by using the above second order approximations in Eqs. (12.63)
12.2 Quantum Master Equation for Open System Quantum Dynamics 409

and (12.68). Before presenting these equations, it is instructive to consider a generic

form of the system-bath interaction Hamiltonian given by

.Ĥsb = Ŝn B̂n , (12.74)

n

where .Ŝn and .B̂n are not necessarily Hermitian. Since .Ĥsb is Hermitian, this means
that the sum should also includes those for Hermitian conjugates of .Ŝn and .B̂n in
such cases. With this convention, .Ĥsb given by Eq. (12.74) can now represent a wide
range of system-bath interactions.
In the interaction picture,

Ĥsb,I (t) =
. Ŝn,I (t)B̂n,I (t). (12.75)
n

where

. Ŝn,I (t) = ei Ĥs t/h̄ Ŝn e−i Ĥs t/h̄ , . (12.76)

B̂n,I (t) = ei Ĥb t/h̄ B̂n e−i Ĥb t/h̄ . (12.77)

Inserting Eq. (12.75) into Eq. (12.69), we find that

 
(2)
.Ksb (t, τ )ρ̂s,I (τ ) = T rb [Ŝn,I (t)B̂n,I (t), [Ŝn′ ,I (τ )B̂n′ ,I (τ ), ρ̂b ρ̂s,I (τ )]]
n n′

(2)
= Cnn ′ (t, τ )[Ŝn,I (t), Ŝn′ ,I (τ )ρ̂s,I (τ )]
n n′

+Cn(2)
′ n (τ, t)[ρ̂s,I (τ )Ŝn′ ,I (τ ), Ŝn,I (t)] , (12.78)

with

(2) 1
Cnn
. ′ (t, τ ) = T rb {B̂n,I (t)B̂n′ ,I (τ )ρ̂b }. (12.79)
h̄2

In the second equality of Eq. (12.78), the fact that .Ŝn,I (t), .Ŝn′ ,I (τ ), and .ρ̂s,I (τ ) all
commute with .B̂n,I (t), .B̂n′ ,I (τ ), and .ρ̂b and the cyclic invariance in the trace of bath
operators have been used. Also, note that the first and second sums in the second
(2)
equality of Eq. (12.78) are Hermitian conjugates of each other since .Cnn′ (t, τ ) and
(2)
.C ′ (τ, t) are complex conjugates of each other. Combining Eq. (12.78) with the
nn
integrand in the second term on the righthand side of Eq. (12.63), we find that
410 12 Special Topics

 
e−i Ls t Ksb (t, τ ) ei Ĥs τ/h̄ ρ̂s (τ )e−i Ĥs τ/h̄
(2)
.


(2)
= Cnn′ (t, τ )e−i Ĥs t/h̄ [Ŝn,I (t), Ŝn′ ,I (τ )ei Ĥs τ/h̄ ρ̂s (τ )e−i Ĥs τ )/h̄ ]ei Ĥs t/h̄
n n′

(2)
+Cn′ n (τ, t)e−i Ĥs t/h̄ [ei Ĥs τ/h̄ ρ̂s (τ )e−i Ĥs τ/h̄ Ŝn′ ,I (τ ), Ŝn,I (t)]ei Ĥs t/h̄

(2)
= Cnn′ (t, τ )[Ŝn , e−i Ĥs (t−τ )/h̄ Ŝn′ ρ̂s (τ )ei Ĥs (t−τ )/h̄ ]
n n′

(2)
+Cn′ n (τ, t)[e−i Ĥs (t−τ )/h̄ ρ̂s (τ )Ŝn′ ei Ĥs (t−τ )/h̄ , Ŝn ] . (12.80)

Similarly for the second order inhomogeneous term Eq. (12.70), we find that

(1)
 
(2)
.Ic,sb (t) = −i Ŝn,I (t), F̂n (t)
n
 t  (2)

− dτ Ŝn,I (t), Ŝn′ ,I (τ )F̂nn′ (t, τ )
n n′ 0

(2)
 
+ F̂n′ n (τ, t)Ŝn′ ,I (τ ), Ŝn,I (t) , (12.81)

where
(1) 1
. F̂n (t) = T rb {B̂n,I (t)ρ̂I (0)}, . (12.82)
h̄
(2) 1
F̂nn′ (t, τ ) = 2 T rb {B̂n (t)B̂n′ (τ )Qρ̂I (0)}. (12.83)
h̄

Note that the above two terms are still operators in the system space. Therefore, the
inhomogeneous term in Eq. (12.63) within the second order approximation can be
expressed as

(1)
 
e−i Ls t Ic,sb (t) = −i
(2)
. Ŝn , e−i Ĥs t/h̄ F̂n (t)ei Ĥs t/h̄
n
 t  (2)

− dτ Ŝn , e−i Ĥs (t−τ )/h̄ Ŝn′ F̂nn′ (t, τ )ei Ĥs (t−τ )/h̄
n n′ 0

(2)
 
+ e−i Ĥs (t−τ )/h̄ F̂n′ n (τ, t)Ŝn′ ei Ĥs (t−τ )/h̄ , Ŝn

≡ I(2) (t). (12.84)

Now the general second order time-nonlocal QME for .Ĥsb given by Eq. (12.74) can
be obtained by combining expressions obtained above.
12.2 Quantum Master Equation for Open System Quantum Dynamics 411

Second Order Time-Nonlocal QME in the Schrödinger Picture

For the type of the system-bath interaction Hamiltonian given by Eq. (12.74),
employing Eqs. (12.80) and (12.84) in Eq. (12.63), we obtain the following
second order time-nonlocal QME in the Schrödinger picture:

d
. ρ̂s (t) = −iLs ρ̂s (t)
dt
 t 
(2)
− dτ Cnn′ (t, τ )[Ŝn , e−i Ĥs (t−τ )/h̄ Ŝn′ ρ̂s (τ )ei Ĥs (t−τ )/h̄ ]
n n′ 0

(2)
+Cn′ n (τ, t)[e−i Ĥs (t−τ )/h̄ ρ̂s (τ )Ŝn′ ei Ĥs (t−τ )/h̄ , Ŝn ]

+I(2) (t). (12.85)

For the derivation of the second order time-local equation in the Schrödinger
picture, first note that the relaxation super-operator .R(2)
sb (t) given by Eq. (12.71) is
(2)
simply the integration of .K̂sb (t, τ ) over .τ . Thus, for .Ĥsb given by Eq. (12.74), we
find that
 t 
(2) (2)
.R
sb (t) ρ̂s,I (t) = dτ Cnn ′ (t, τ )[Ŝn,I (t), Ŝn′ ,I (τ )ρ̂s,I (t)]
n n′ 0

+Cn(2)
′ n (τ, t)[ρ̂s,I (t)Ŝn′ ,I (τ ), Ŝn,I (t)] . (12.86)

Therefore, the second term on the righthand side of Eq. (12.68), employing the
second order approximation given above, can be expressed as
 
e−i Ls t Rsb (t) ei Ĥs t/h̄ ρ̂s (t)e−i Ĥs t/h̄
(2)
.

 t 
(2)
= dτ Cnn′ (t, τ )[Ŝn , e−i Ĥs (t−τ )/h̄ Ŝn′ ei Ĥs (t−τ )/h̄ ρ̂s (t)]
n n′ 0

(2)
+Cn′ n (τ, t)[ρ̂s (t)e−i Ĥs (t−τ )/h̄ Ŝn′ ei Ĥs (t−τ )/h̄ , Ŝn ] . (12.87)

The above expression in combination with the inhomogeneous term leads to a

general second order time-local QME as summarized below.
412 12 Special Topics

Second Order Time-Local QME in the Schrödinger Picture

For the type of the system-bath interaction Hamiltonian given by Eq. (12.74),
employing Eqs. (12.87) and (12.84) in Eq. (12.68), we obtain the following
second order time-nonlocal QME in the Schrödinger picture:

d
. ρ̂s (t) = −iLs ρ̂s (t)
dt
 t 
(2)
− dτ Cnn′ (t, τ )[Ŝn , e−i Ĥs (t−τ )/h̄ Ŝn′ ei Ĥs (t−τ )/h̄ ρ̂s (t)]
n n′ 0

(2)
+Cn′ n (τ, t)[ρ̂s (t)e−i Ĥs (t−τ )/h̄ Ŝn′ ei Ĥs (t−τ )/h̄ , Ŝn ]

+I(2) (t). (12.88)

The approach of deriving second order approximations described above can be

extended further to derive higher order perturbative approximations in both time-
nonlocal and time-local forms [37, 69, 70]. However, such perturbative expansions
result in much more complicated expressions in general while they may not lead to
significant improvements. Rather, approaches to account for terms up to the infinite
order (in principle) through hierarchy of equations of motion approach [71–74] or an
approach to solve unprojected part through numerical solution of integral equation,
which is called generalized QME [75, 76], have been more successful. Alternatively,
it is possible to account for some higher order terms through unitary transformation
that effectively reduces the system-bath interaction, while still using the second
order approximation [77, 78].

12.3 Green’s Function Approach

Mathematically, Green’s function is an inverse of a differential operator and

enables finding the solution of an inhomogeneous differential equation employing
those for the homogeneous equation [30, 79, 80]. Let us consider the following
inhomogeneous differential equation,

L̂(ξ )f (ξ ) = S(ξ ),
. (12.89)

where .L̂(ξ ) is a differential operator with .ξ as a variable and .S(ξ ) is the source
term. Although one dimensional notation is used here, .ξ can be multidimensional in
general. For example, .ξ can represent either the time or spatial coordinates, or both
12.3 Green’s Function Approach 413

of them. Green function for the above differential equation is defined as the solution
of the following equation with a delta function source term.

L̂(ξ )G(ξ, ξ ′ ) = δ(ξ − ξ ′ ).

. (12.90)

Then, the general solution of Eq. (12.89) can be expressed as follows:


f (ξ ) = fh (ξ ) +
. dξ ′ G(ξ, ξ ′ )S(ξ ′ ), (12.91)

where .fh (ξ ) is the homogeneous solution of the differential equation such that
L̂(ξ )fh (ξ ) = 0. The fact that .f (ξ ) given above indeed is a solution of Eq. (12.89)
.

is shown below.
   
.L̂(ξ ) fh (ξ ) + ds G(ξ, ξ )S(ξ ) = dξ ′ L̂ξ G(ξ, ξ ′ )S(ξ ′ )
′ ′ ′


= dξ ′ δ(ξ − ξ ′ )S(ξ ′ ) = S(ξ ).

(12.92)

Employing Green’s function is convenient because it can incorporate boundary

conditions within its form. In addition, solutions involving Green’s functions are
easier to develop systematic approximations.
Green’s function can also be used for obtaining (or approximating) the homoge-
nous solution of a complicated differential operator in terms of those for simpler
ones. For example, assume that the differential operator .L̂(ξ ) can be divided into
two terms,

L̂(ξ ) = L̂0 (ξ ) + L̂1 (ξ ),

. (12.93)

and that the Green function for .L̂0 (ξ ), which is denoted as .G0 (ξ, ξ ′ ), can be
obtained easily. Then, the differential equation for the homogeneous solution for
.L̂(ξ ) can be expressed as

L̂0 (ξ )f (ξ ) = −L̂1 (ξ )f (ξ ).
. (12.94)

Treating the righthand side of the above equation as a source term, we can express
the solution .f (ξ ) in term of .G0 (ξ, ξ ′ ) as

f (ξ ) = f0 (ξ ) −
. ds ′ G0 (ξ, ξ ′ )L̂1 (ξ ′ )f (ξ ′ ). (12.95)

Although the above equation still involves unknown .f (ξ ′ ) on the righthand side, a
formal solution can be obtained by iterating the righthand side as follows:
414 12 Special Topics


f (ξ ) = f0 (ξ ) −
. dξ1 G0 (ξ, ξ1 )L̂1 (ξ1 )f0 (ξ1 )
 
+ dξ1 dξ2 G0 (ξ, ξ1 )L̂1 (ξ1 )G0 (ξ1 , ξ2 )L̂1 (ξ2 )f0 (ξ2 )

+···
 
+(−1)n dξ1 · · · dξn G0 (ξ, ξ1 )L̂1 (ξ1 ) · · · G0 (ξn−1 , ξn )L̂1 (ξn )f0 (ξn )

+··· . (12.96)

Given that the above series converge quickly, one can truncate the above expansion
at finite order, obtaining a perturbative approximation. More advanced approaches
are to employ integral transformation and/or diagrammatic techniques [18] to
account for major terms up to an infinite order.
The standard Green’s function in quantum mechanics refers to that for the
Schrödinger equation and has played an important role in scattering and dynamics
problems [3, 17, 81]. In fact, the real and imaginary time propagators defined in the
path integral representation can also be viewed as examples of Green’s function.
However, more generally, Green’s function approach in quantum mechanics
often refers to a general class of powerful technique to solve quantum many
body problems in combination with quantum field theory [18, 81]. Operational
definitions of Green’s functions in this approach are correlation functions of field
operators, which retain all the major spatiotemporal dynamics of quantum particles
in interacting environments and can be used for the calculation of all the relevant
physical properties. The remainder of this section will provide a very brief account
of this approach, starting from the definition of field operators. For comprehensive
accounts of theoretical basis and important technical details of the Green’s function
approach, readers are strongly recommended to refer to more advanced references
[18, 82].

12.3.1 Second Quantization and Field Operators

Let us consider the lowering and raising operators .b̂ and .b̂† for a harmonic
oscillator as defined by Eqs. (3.51) and (3.52) in Chap. 3. The two operators satisfy
Eqs. (3.53)–(3.55) and can be used to express the Hamiltonian as Eq. (3.57). For a
given eigenstate .|v of the harmonic oscillator, application of .b̂† produces .|v + 1
according to Eq. (3.67) and application of .b̂ to .|v produces .|v − 1 according to
Eq. (3.70).
In the second quantization formulation of quantum mechanics, each eigenstate
.|v of the harmonic oscillator Hamiltonian can be viewed as a state with v number

of phonons. Then, the ground vibrational state .|0 can be defined as the “vacuum"
state of phonons and .b̂† and .b̂ now become creation and annihilation operators
of phonons. This view of quantum harmonic oscillator is equivalent to the one
12.3 Green’s Function Approach 415

described in Chap. 3 and allows expressing all physical observables and states only
in terms of products of .b̂’s and .b̂† ’s applied to the vacuum state.
The second quantization formulation is particularly useful when there are many
(possibly infinite number of) harmonic oscillator normal modes because there is
no need to introduce eigenstates with many (possibly infinite number of) indices.
Instead, for a system of .Ns independent harmonic oscillator normal modes, we can
define .b̂j† and .b̂j , with .j = 1, · · · , Ns , as the creation and annihilation operators of
a phonon for each mode.
Phonons for each mode are identical particles satisfying boson statistics. In fact,
the definition of creation and annihilation operators for phonons can be extended to
other types of bosons as well. For this, it is first necessary to define the occupation
number space of bosons. Let us introduce a state in the occupation number space,

|n1 , · · · , nk , · · · , nNs  : A quantum state where there are n1 particles

.

in state 1, · · · , nk particles in state k, · · ·

and nNs particles in state Ns (12.97)

Then, the creation and annihilation operators for bosons in the kth state, which are
in general denoted as .b̂k† and .b̂k , can be defined as described below.

Creation and Annihilation Operators for Bosons

For the state in the occupation number space defined by Eq. (12.97), creation
and annihilation operators for bosons in the kth state are defined by

b̂k† |n1 , · · · , nk , · · · , nNs  =


. nk + 1|n1 , · · · , nk + 1, · · · , nNs , (12.98).
√
b̂k |n1 , · · · , nk , · · · , nNs  = nk |n1 , · · · , nk − 1, · · · , nNs , (12.99)

and satisfy the following commutator identities:

. [b̂j , b̂k† ] = δj k , . (12.100)

[b̂j† , b̂k† ] = [b̂j , b̂k ] = 0. (12.101)

Then, the state in the occupation number space, Eq. (12.97), can be expressed
as
Ns
 1  †  nk
.|n1 , · · · , nk , · · · , nNs  = √ b̂ |vac, (12.102)
k=1
nk ! k

where .|vac is the vacuum state without any boson particles occupying the
states. Note that the above definition is independent of the order of creation
operators for different indices because they commute with each other.
416 12 Special Topics

The creation and annihilation operators introduced above define a number

operator .n̂k = âk† âk for bosons in the state k as follows:

b̂k† b̂k |n1 , · · · , nk , · · · , nNs  = nk |n1 , · · · , nk , · · · , nNs ,

. (12.103)

which can be shown directly by consecutive applications of Eqs. (12.98)

and (12.99).
The creation and annihilation operators defined above can also be used for
photons, which are bosons, for each frequency and polarization direction. If photons
are confined in a small volume, only certain frequencies satisfying the boundary
conditions are allowed. Thus, the states of photons in such case are discrete.
However, as the volume becomes infinite, the spectrum of photon frequencies
becomes continuous and the corresponding Hamiltonian of light involves integration
over the entire frequency range.
For fermions such as electrons, the occupation number state such as Eq. (12.97)
can still be used. However, the number of particles in each state can be only either
0 or 1. In addition, exchange of the states of two particles should result in change
of the sign of the state. All of these properties can be naturally encoded by defining
creation and annihilation operators satisfying different identities as described below.

Creation and Annihilation Operators for Fermions

Creation and annihilation operators for states r and s of fermions, .âr† and .âs ,
satisfy the following anti-commutation relationship:

.{âr† , âs } ≡ âr† âs + âs âr† = δrs . (12.104)

On the other hand, between annihilation or creation operators, the following

anti-commutation identities hold.

{âr , âs } = {âr† , âs† } = 0.

. (12.105)

Then, the state, Eq. (12.97), can be expressed as

nN
. |n1 , · · · , nr , · · · , nNs  = (â1† )n1 (â2† )n2 · · · (âs† )ns · · · (âN†s ) s |vac,
(12.106)

where .|vac is the vacuum state without any fermion particles occupying the
states and each .nr = 0 or 1.
12.3 Green’s Function Approach 417

Equation (12.105) for .r = s implies that .âs2 = âs†2 = 0 for each s. On the other
hand, Eq. (12.104) for .r = s implies that .âs† âs = 1 − âs âs† . Thus, combining the two
cases, we find that

(âs† âs )2 = âs† âs − âs† âs2 âs† = âs† âs .

. (12.107)

The above identity is true only if the eigenvalue of .âs† âs is either 0 or 1. Thus, the
number operator for each state s in this case is given by .n̂s = âs† âs as well. On the
other hand, the number operators for different states (.r = s) commute with each
other as shown below.

[n̂r , n̂s ] = âr† âr âs† âs − âs† âs âr† âr = −âr† âs† âr âs − âs† âs âr† âr
.

= âr† âs† âs âr − âs† âs âr† âr = −âs† âr† âs âr − âs† âs âr† âr
= âs† âs âr† âr − âs† âs âr† âr = 0. (12.108)

This means that simultaneous diagonalization of any set of the .n̂r is possible, which
validates the occupation number state defined by Eq. (12.106).
In general, creation of a single boson or fermion particle does not have to be
limited to a specific state but can be any linear superposition of those for different
states. In addition, it is often useful to have a definite spatial information on
the created particle because interactions between particles are local in space, in
particular, for actual particles such as electrons. For this reason, it is useful to
consider a general linear combination of creation or annihilation operators projected
onto the position space. These are called field operators and are defined below.

Field Operators
Field operators are linear combinations of creation or annihilation operators
defined as

. ψ̂(r) = ψk (r)ĉk , . (12.109)

k

ψ̂ † (r) = ψk† (r)ĉk† , (12.110)

k

where .ĉk and .ĉk† are annihilation and creation operators for either bosons
or fermions, and .ψk (r) is the wavefunction (or vector of wavefunctions for
particles with nonzero spins) for the state k. it is assumed that the summation
index k runs over a complete set of single-particle quantum states.
418 12 Special Topics

For the case of bosons, the field operators defined above satisfy the following
commutator identity:

.[ψ̂(r), ψ̂ † (r′ )] = ψk (r)ψk∗′ (r′ )[b̂k , b̂k†′ ] = ψk (r)ψk∗ (r′ )

k k′ k

= r|ψk ψk |r′  = δ(r − r′ ). (12.111)

k

On the other hand, for fermions with spin half such as electrons, the field operator
defined above satisfy the following anti-commutator identity:
   
. ψ̂γ (r), ψ̂γ† ′ (r′ ) = ∗
ψk,γ (r)ψk,γ ′ †
′ (r ) âk , âk ′ = ∗
ψk,γ (r)ψk,γ ′
′ (r )
k k′ k

= δγ γ ′ r|ψk ψk |r′  = δγ γ ′ δ(r − r′ ), (12.112)

k

where .γ , .γ ′ represent either spin up or down.

The field operators defined above can be used to represent physical observables.
For example, the  operator for any one-body function given by a sum of all the
particles, .A = j A(rj ), can be expressed as

. = dr ψ̂ † (r)A(r)ψ̂(r). (12.113)

It is also possible to prove that the Hamiltonian operator for identical particles of
mass m subject to external potential .vext (r) and with pair-wise interaction potential
′
.Vint (r, r ) can be expressed as

 
h̄2 2
  
′
.Ĥ = dr dr ψ̂ (r) − † ′
∇r δ(r − r ) ψ̂(r′ ) + dr ψ̂ † (r)vext (r)ψ̂(r)
2m
  
1
+ dr dr′ ψ̂ † (r)ψ̂ † (r′ )Vint (r, r′ )ψ̂(r′ )ψ̂(r). (12.114)
2

12.3.2 Ground State (Zero Temperature) Green’s Functions

For the total Hamiltonian .Ĥ of an interacting many particle system, let us define
time dependent field operators in the Heisenberg picture,

. ψ̂H γ (rt) = ei Ĥ t/h̄ ψ̂γ (r)e−i Ĥ t/h̄ , . (12.115)

†
ψ̂H γ (rt) = e ψ̂γ (r)e−i Ĥ t/h̄ .
i Ĥ t/h̄ †
(12.116)
12.3 Green’s Function Approach 419

Then, the single particle Green’s function with respect to the normalized ground
state .| 0  is defined as
 
′ ′ † ′ ′
.Gγ γ ′ (rt, r t ) ≡ −i 0 |T ψ̂H γ (r, t)ψ̂
H γ ′ (r , t ) | 0 , (12.117)

where .T is the chronological time ordering super-operator. Thus, the above Green’s
functions can be expressed as
 †
−i 0 |ψ̂H γ (rt)ψ̂H ′ ′ ′
′ ′ γ ′ (r , t )| 0  , t > t
.Gγ γ ′ (rt, r t ) = † (12.118)
∓i 0 |ψ̂H ′ ′ ′
γ (r , t )ψ̂H γ ′ (rt)| 0  , t > t

In the second row on the righthand side of the above equation, the .− sign is for
bosons and the .+ sign is for fermions. This sign convention makes the above
definition of Green’s function consistent with the commutator and anti-commutator
identities of bosons and fermions, respectively. Green’s functions for two particle
field operators can be defined in a similar manner.
The Green’s function defined above provides direct information on the excitation
properties. For example, in the Fourier transforms of this Green’s function with
respect to time and position, the location and strength of poles respectively give
direct information on excitation energies and strengths. Thus, in combination with
powerful diagrammatic techniques [18] and more recently with direct numerical
iteration techniques, which allow calculation of Green’s functions for interacting
many-particle systems from those for simple reference systems, one can calculate
the key properties related to ionizations or excitations without going through
calculation of many-body wavefunctions.

12.3.3 Nonequilibrium Green’s Functions

Nonequilibrium Green’s functions serve as important tools for calculating transport

properties [83, 84] in open environments including the case where there is particle
exchange with reservoir. Within the Keldysh formalism [85], this can be done by
defining contour ordered correlation functions of field operators. For this, let us
omit writing explicitly the spatial coordinates and denote each operator in terms of
time. Let us also drop the subscript H denoting the Heisenberg picture.
For any nonequilibrium process that can be defined to have evolved from an
equilibrium grand canonical distribution in the infinite past, one can define the
following contour ordered Green’s function :
 
′ † ′
.G(τ, τ ) = −iTc ψ̂(τ )ψ̂ (τ ) , (12.119)

where .· · ·  means trace over the grand canonical density operator at infinite past
(.t = −∞) and .Tc is the normal time ordering operator along the closed time contour
that starts from .−∞ to .∞ and then comes back to .−∞. In case .τ comes later than
′ ′
.τ along the contour, .G(τ, τ ) becomes the following greater Green’s function:
420 12 Special Topics

G> (τ, τ ′ ) = −iψ̂(τ )ψ̂ † (τ ′ ).

. (12.120)

On the other hand, in case .τ ′ comes later than .τ along the contour, .G(τ, τ ′ ) becomes
the following lesser Green’s function:

G< (τ, τ ′ ) = ∓iψ̂ † (τ ′ )ψ̂(τ ).

. (12.121)

where the .− sign is for bosons and .+ sign is for fermions. Note that these greater
and lesser Green’s functions are well defined even when the averaging is over the
ground state. In fact, the two cases of Eq. (12.118) correspond to these greater and
lesser Green’s functions.
It is also useful to define retarded and advanced Green’s functions as follows:

Gr (τ, τ ′ ) = −i(τ − τ ′ )ψ̂(τ )ψ̂ † (τ ′ ) + ψ̂ † (τ ′ )ψ̂(τ )

.

= (τ − τ ′ ) G> (τ, τ ′ ) − G< (τ, τ ′ ) , .


(12.122)
Ga (τ, τ ′ ) = i(τ ′ − τ )ψ̂(τ )ψ̂ † (τ ′ ) + ψ̂ † (τ ′ )ψ̂(τ )
= (τ ′ − τ ) G< (τ, τ ′ ) − G> (τ, τ ′ ) .

(12.123)

Finally, let us also define two regular time ordered Green functions defined along
a straight real time line as follows:
 
G(+) (τ, τ ′ ) = −iT ψ̂(τ )ψ̂ † (τ ′ ) , .
. (12.124)
 
G(−) (τ, τ ′ ) = −iT̃ ψ̂(τ )ψ̂ † (τ ′ ) , (12.125)

where .T is the chronological time ordering super-operator and .T̃ is the anti-
chronological time ordering super-operator. Then, we can show that

G(+) (τ, τ ′ ) = (τ − τ ′ )G> (τ, τ ′ ) + (τ ′ − τ )G< (τ, τ ′ ),

.

. (12.126)
(−) ′ ′ > ′ ′ < ′
G (τ, τ ) = (τ − τ )G (τ, τ ) + (τ − τ )G (τ, τ ).
(12.127)

Combining Eqs. (12.122), (12.123), (12.126), and (12.127), we can also establish
the following identities:

G< (τ, τ ′ ) = G(+) (τ, τ ′ ) − Gr (τ, τ ′ )

.

= G(−) (τ, τ ′ ) + Ga (τ, τ ′ ), . (12.128)

G> (τ, τ ′ ) = G(+) (τ, τ ′ ) − Ga (τ, τ ′ )
= G(−) (τ, τ ′ ) + Gr (τ, τ ′ ). (12.129)
12.3 Green’s Function Approach 421

A great deal of dynamical properties in nonequilibrium processes can be

expressed in terms of the following convolution of contour-order Green’s functions:

′
.A(τ, τ ) = dτ1 B(τ, τ1 )C(τ1 , τ ′ ). (12.130)
c

Langreth [86] showed that the above convolution can be related to those for standard
types of Green functions employing a two-looped contour which passes through .∞
twice with t and .t ′ on each loop. Important examples of these are
 ∞
A< (t, t ′ ) = dt1 B r (t, t1 )C < (t1 , t ′ ) + B < (t, t1 )C a (t1 , t ′ ) ,
! "
. (12.131)
.
−∞
 ∞
A> (t, t ′ ) = dt1 B > (t, t1 )C a (t1 , t ′ ) + B r (t, t1 )C > (t1 , t ′ ) ,
! "
(12.132)
.
−∞
 ∞
Ar (t, t ′ ) = dt1 B r (t, t1 )C r (t1 , t ′ ), . (12.133)
−∞
 ∞
Aa (t, t ′ ) = dt1 B a (t, t1 )C a (t1 , t ′ ). (12.134)
−∞

Application of the above identities makes it possible to reduce expressions for

nonequilibrium transport properties to those involving standard Green’s functions
that have clear analogues in the equilibrium case and thus can be calculated
extending well established techniques for equilibrium Green’s functions.
References

1. D.A. McQuarrie, Quantum Chemistry, 2nd edn. (University Science Books, Mill Valley, 2008)
2. I.N. Levine, Quantum Chemistry, 7th edn. (Pearson Education, Upper Saddle River, 2014)
3. G.C. Schatz, M.A. Ratner, Quantum Mechanics in Chemistry (Dover Publications, Mineola,
2002)
4. S. Gasiorowicz, Quantum Physics (John Wiley & Sons, New York, 1974)
5. P.A.M. Dirac, The Principles of Quantum Mechanics (Oxford University Press, Oxford, 1967)
6. J.J. Sakurai, Modern Quantum Mechanics (The Benjamin/Cummings Publishing Company,
Menlo Park, 1985)
7. L.D. Landau, E.M. Lifshitz, Quantum Mechanics (Non. Relativistic Theory), 3rd edn. (Perga-
mon Press, Oxford, 1977)
8. C. Cohen-Tannoudji, B. Diu, F. Laloe, Quantum Mechanics (John Wiley & Sons, New York,
1970)
9. F.A. Cotton, Chemical Applications of Group Theory, 3rd edn. (John Wiley & Sons, New York,
1990)
10. M. Tinkham, Group Theory and Quantum Mechanics (Dover Publications, Mineola, 2003)
11. P. Atkins, J. dePaula, Physical Chemistry, Tenth Edition (Oxford University Press, Oxford,
2014)
12. R.J. Silbey, R.A. Alberty, M.G. Bawendi, Physical Chemistry, 4th edn. (John Wiley & Sons,
Hoboken, 2005)
13. A. Cooksy, Physical Chemistry: Quantum Chemistry and Molecular Interactions (Pearson,
London, 2013)
14. A. Szabo, N.S. Ostlund, Modern Quantum Chemistry (McGraw-Hill New York, 1989)
15. C.J. Cramer, Essentials of Computational Chemistry, 2nd edn. (John Wiley & Sons, Chichester,
2004)
16. R.P. Feynman, A.R. Hibbs, Quantum Mechanics and Path Integrals (McGraw-Hill Book
Company, New York, 1965)
17. H. Kleinert, Path Integrals in Quantum Mechanics, Statistics, Polymer Physics, and Financial
Markets (World Scientific, Singapore, 2009)
18. A.L. Fetter, J.D. Walecka, Quantum Theory of Many-Particle Systems (McGraw-Hill Book
Company, New York, 1971)
19. H. Goldstein, Classical Mechanics (Addison-Wesley Publishing Company, New York, 1980)
20. R. Loudon, The Quantum Theory of Light, 3rd edn. (Oxford University Press, Oxford, 2000)
21. C.J. Davisson, L.H. Germer, Reflection of electrons by a crystal of nickel. Proc. Natl. Acad.
Sci. U.S.A. 14, 317 (1928)
22. M. Jammer, The Philosophy of Quantum Mechanics (John Wiley & Sons, New York, 1974)

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 423
S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3
424 References

23. G.B. Arfken, H.J. Weber, F.E. Harris, Mathematical Methods for Physicists, 7th edn. (Aca-
demic Press, Waltham, 2013)
24. J. von Neumann, Mathematical Foundations of Quantum Mechanics, translated by R. T. Beyer
(Princeton University Press, Princeton, 1955)
25. A. Peres, Quantum Theory: Concept and Methods (Kluwer Academic Publishers, Dordrecht,
1995)
26. R.B. Griffiths, What quantum measurements measure. Phys. Rev. A 96, 032110 (2017)
27. R. Omnès, Consistent interpretations of quantum mechanics. Rev. Mod. Phys. 64, 339 (1992)
28. P.C. Hohenberg, Colloquium: an introduction to consistent quantum theory. Rev. Mod. Phys.
82, 2835 (2010)
29. J.E.B.Wilson, J.C. Decius, P.C. Cross, Molecular Vibrations: The Theory of Infrared and
Raman Vibrational Spectra (Dover Publications, New York, 1955)
30. P.M. Morse, H. Feshbach, Methods of Theoretical Physics, Part 1 (McGraw-Hill Book
Company, New York, 1953)
31. J.M. Hollas, Modern Spectroscopy, 4th edn. (John Wiley & Sons, Chichester, 2004)
32. M. Whippman, Orbital angular momentum in quantum mechanics. Am. J. Phys. 34, 656
(1966)
33. R.A. Marcus, On the theory of oxidation-reduction reactions involving electron transfer. I. J.
Chem. Phys. 24, 966 (1956)
34. R.A. Marcus, N. Sutin, Electron transfers in chemistry and biology. Biochim. Biophys. Acta
811, 265 (1985)
35. T. Förster, Intermolecular energy migration and fluorescence. Ann. Phys. (Berlin) 437, 55
(1948)
36. S. Jan, M.D. Newton, Closed form expressions of quantum electron transfer rate based on the
stationary phase approximation. J. Phys. Chem. B 110, 18996 (2006)
37. S.J. Jang, Dynamics of Molecular Excitons (Nanophotonics Series) (Elsevier, Amsterdam,
2020)
38. J.D. Jackson, Classical Electrodynamics (John Wiley and Sons, New York, 1975)
39. S. Mukamel, Principles of Nonlinear Spectroscopy (Oxford University Press, New York, 1995)
40. J.A. Cina, Getting Started on Time-Resolved Molecular Specrtoscopy (Oxford University
Press, Oxford, 2022)
41. J.C. Garrison, R.Y. Chiao, Quantum Optics (Oxford University Press, Oxford, 2014)
42. C.C.J. Roothaan, New developments in molecular orbital theory. Rev. Mod. Phys. 23, 69 (1951)
43. R. Weinstock, Calculus of Variations (McGraw-Hill Book Company New York, 1952)
44. T.H. Dunning, Jr., P.J. Hay, Gaussian basis sets for molecular calculations, in Methods of
Electronic Structure Theory: Modern Theoretical Chemistry, vol. 3, ed. by H.F. Schaefer
(Springer, Boston, 1977)
45. J.A. Pople, A priori geometry predictions, in Methods of Electronic Structure Theory: Modern
Theoretical Chemistry, vol. 4, ed. by H.F. Schaefer (Springer, Boston, 1977)
46. S. Wilson, Basis sets, in Advances in Chemical Physics: Ab initio Methods in Quantum
Chemistry Part I, vol. 67, ed. by K.P. Lawley (John Wiley & Sons, New York, 1987)
47. I. Shavitt, The method of configuration interaction, in Methods of Electronic Structure Theory:
Modern Theoretical Chemistry, vol. 3, ed. by H.F. Schaefer (Springer, Boston, 1977)
48. P.D. Jorgensen, J. Simons, Second Quantization-Based Methods in Quantum Chemistry
(Academic Press, New York, 1981)
49. P. Hohenberg, W. Kohn, Inhomogeneous electron gas. Phys. Rev. 136, B864 (1964)
50. R.G. Parr, W. Yang, Density-Functional Theory of Atoms and Molecules (Oxford Science
Publications, Oxford, 1989)
51. M. Levy, Universal variational functionals of electron densities, first-order density matrices,
and natural spin-orbitals and solution of the v-representability problem. Proc. Natl. Acad. Sci.
U.S.A. 76, 6062 (1979)
52. W. Kohn, L.J. Sham, Self consistent equations including exchange and correlation effects.
Phys. Rev. 140, A1133 (1965)
References 425

53. W. Koch, M.C. Holthausen, A Chemist’s Guide to Density Functional Theory, 2nd edn. (Wiley-
VCH, Weinheim, 2002)
54. A.M. Teale et al., DFT exchange: sharing perspectives on the workhorse of quantum chemistry
and materials science. Phys. Chem. Chem. Phys. 24, 28700 (2022)
55. E. Runge, E.K.U. Gross, Density-functional Theory for time dependent systems. Phys. Rev.
Lett. 52, 997 (1984)
56. R.P. Feynman, Statistical Mechanics (Addison-Wesley Publishing Company, New York, 1972)
57. L.S. Schulman, Techniques and Applications of Path Integration (Wiley-Interscience, New
York, 1981)
58. E. Trotter, On the products of semi-groups of operators. Proc. Am. Math. Soc. 10, 545 (1958)
59. U. Weiss, Series in Modern Condensed Matter Physics. Quantum Dissipative Systems, vol. 2
(World Scientific, Singapore, 1993)
60. N. Makri, Time-dependent quantum methods for large systems. Ann. Rev. Phys. Chem. 50,
167 (1999)
61. I. Oppenheim, K.E. Shuler, G.H. Weiss (eds.), Stochastic Processes in Chemical Physics: The
Master Equation (The MIT Press, Cambridge, 1977)
62. R. Zwanzig, Time correlation functions and transport coefficients in statistical mechanics. Ann.
Rev. Phys. Chem. 16, 67 (1965)
63. N.G. van Kampen, I. Oppenheim, Langevin and master equation in quantum mechanics. J.
Stat. Phys. 87, 1325 (1997)
64. A. Nitzan, Chemical Dynamics in Condensed Phases (Oxford University Press, Oxford, 2006)
65. F. Shibata, T. Arimitsu, Expansion formulas in nonequilibrium statistical mechanics. J. Phys.
Soc. Jpn. 49, 891 (1980)
66. R. Kubo, Generalized cumulant expansion method. J. Phy. Soc. Jpn. 17, 1100 (1962)
67. B. Yoon, J.M. Deutch, J.H. Freed, A comparison of generalized cumulant and projection
operator methods in spin-relaxation theory. J. Chem. Phys. 62, 4687 (1975)
68. S. Mukamel, I. Oppenheim, J. Ross, Statistical reduction for strongly driven simple quantum
systems. Phys. Rev. A 17, 1988 (1978)
69. S. Jang, J. Cao, R.J. Silbey, Fourth order quantum master equation and its Markovian bath
limit. J. Chem. Phys. 116, 2705 (2002)
70. H.P. Breuer, B. Kappler, F. Petruccione, The time-convolution less projection operator
technique in the quantum theory of dissipation and decoherence. Ann. Phys. 291, 36 (2001)
71. Y. Tanimura, Stochastic Liouville, Langevin, Fokker-Planck, and master equation approaches
to quantum dissipative systems. J. Phys. Soc. Jpn. 75, 082001 (2006)
72. Y. Tanimura, Real-time and imaginary-time quantum hierarchical Fokker-Planck equation. J.
Chem. Phys. 142, 144110 (2015)
73. R.X. Xu, Y.J. Yan, Dynamics of quantum dissipation systems interacting with bosonic
canonical bath: hierarchical equations of motion approach. Phys. Rev. E 75, 031107 (2007)
74. M. Schröter, S.D. Ivanov, J. Schulze, S.P. Polyutov, Y. Yan, T. Pullerits, O. Kühn, Exciton-
vibrational coupling in the dynamics and spectroscopy of Frenkel excitons in molecular
aggregates. Phys. Rep. 567, 1 (2015)
75. M.-L. Zhang, B.J. Ka, E. Geva, Nonequilibrium quantum dynamics in the condensed phase via
the generalized quantum master equation. J. Chem.Phys. 125, 0044106 (2006)
76. Q. Shi, E. Geva, A semiclassical generalized quantum master equation for an arbitrary system-
bath coupling. J. Chem. Phys. 120, 10647 (2004)
77. S. Jang, T. Berkelbach, D.R. Reichman, Coherent quantum dynamics in donor-bridge-acceptor
nonadiabatic processes: beyond the hopping and super-exchange mechanisms. New J. Phys.
15, 105020 (2013)
78. S.J. Jang, Partially polaron-transformed quantum master equation for exciton and charge
transport dynamics. J. Chem. Phys 157, 104107 (2022)
79. R. Courant, D. Hilbert, Methods of Mathematical Physics, vol. I (Interscience Publishers, New
York, 1953)
80. J. Mathews, R.L. Walker, Mathematical Methods of Physics (W. A. Benjamin, Menlo Park,
1969)
426 References

81. E. Merzbacher, Quantum Mechanics, 2nd edn. (John Wiley & Sons, Amstredam, 1970)
82. J. Linderberg, Y. Öhrn, Propagators in Quantum Chemistry (Academic Press, London/New
York, 1973)
83. J. Ramer, H. Smith, Quantum field-theoretical methods in transport theory of metals. Rev. Mod.
Phys. 58, 323 (1986)
84. A.-P. Jauho, N.S. Wingreen, Y. Meir, Time-dependent transport in interacting and noninteract-
ing resonant-tunneling systems. Phys. Rev. B 50, 5528 (1994)
85. L.V. Keldysh, Diagram technique for nonequilibrium processes. Sov. Phys. JETP 20, 1018
(1965)
86. D. C. Langreth, Linear and Nonlinear Electron Transport in Solids. Nato Advanced Study
Institute, Series B: Physics, vol. 17, ed. by J.T. Devreese, V.E. Van Doren (Plenum, New York,
1976)
Index

A C
Angular momentum degeneracy, 152 Cauchy-Schwarz inequality, 59, 60
Angular momentum quantum number, 182 Causality
total, 248 principle of, 10
total orbital, 247 relationship of, 26
Anharmonicity, 88, 89 Center of mass coordinates, 113
Anti-Stokes vibrational transition, 88 Centrifugal distortion, 155
Atomic term symbol, 189 constant, 157
Atomic units, 260 energy, 157
Aufbau (building-up) principle, 244 Class, 300
Classical mechanics, 1
Classical point particles, 1
B Commutator, 47
Balmer series, 174 Commutator identities, 48
Basis Configuration interaction, 380
complete, 49 Coulomb gauge, 357, 359
complete and orthonormal, 50 Coulomb integral, 265
orthonormal, 49 Coulomb interaction, 369
Blackbody, 6 Coulomb potential, 171, 176
Blackbody radiation, 6 Coupled cluster (CC) method, 380
Bohr radius, 173 Cylindrical symmetry, 292
Bohr’s model, 172, 173
Bond order, 284
Born-Oppenheimer approximation, 276, 311 D
Born-Oppenheimer potential energy, 279 Davisson-Germer’s experiment, 9
Bose-Einstein condensation, 185 De Broglie, 172
Boson, 185, 415 De Broglie wavelength, 190
Boundary condition, 21 Degeneracy, 110
Box, one-dimensional, 20 Degenerate states, 110
Bra, 36 Degrees of freedom
position, 42 rotational, 116
space, 38 vibrational, 116

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 427
S. J. Jang, Quantum Mechanics for Chemistry,
https://doi.org/10.1007/978-3-031-30218-3
428 Index

Delocalization energy, .π -bonding, 297 Fluorescence, 315

Density functional theory (DFT), 384 Fock operator, 370, 376
Kohn-Sham, 384 Force operator, time dependent, 334
time dependent (TD), 385 Förster rate, 354
Density matrix, electron, 382 Franck-Condon approximation, 315
Density, one-electron, 382 Franck-Condon factor, 315
Density operator, 341, 342 Frequency, 4
canonical equilibrium, 347, 351, 395
one electron, 382
G
reduced system, 404, 406
Gauge transformation, 357
time dependent, 349
Gaussian basis function, 379
two electron, 382
Gaussian basis function, contracted, 379
unnormalized canonical equilibrium, 397
Gaussian integration, 393
Deuteron, 185
Gaussian type basis function, 379
Diatomic molecule, 83
Geometry optimization, 379
Dielectric constant, 357
Gerade, 283
Differential equation
Great orthogonality theorem, 322
homogeneous, 412
Green’s function, 412
inhomogeneous, 412
advanced, 420
Dipole approximation, 339, 360
contour ordered, 419, 421
Dirac-delta function, 43
greater, 420
Dirac-delta function, three dimensional, 102
ground state, 419
Direct product, 102
lesser, 420
Dissociation energy, 89
nonequilibrium, 419
retarded, 420
Group
E
Abelian, 300
Effective single-electron potential, 242
definition of, 300
Eigenfunction, 15
equivalent matrix representations of, 301
Eigenvalue, 15
matrix representation of, 301
Einstein, A., 7
multiplication, 300
Electromagnetic field, 3
multiplication table, 300
Electromagnetic wave, 4
order of, 300
Electromagnetic wave, energy density, 5
representation of, 301
Electron, 185
theory, 120, 299
Electron-electron repulsion, 242, 257
Group elements, 300
Electronic configuration, 284
Group elements, conjugate, 300
excited state, 245
ground state, 245
Electronic states H
adiabatic, 277 Hamiltonian, 2
ground, 383 harmonic oscillator, 71
Equation of motion, Newtonian, 1 rotational, 132
Evolution equation, time-nonlocal, 402 Hamiltonian operator
Exchange integral, 265 for atoms, 241
Exchange interaction, 369 Hamiltonian operator, 15
Expectation value, 18, 19 effective one-electron, 262
effective single electron, 268
polyatomic molecule, 275
F Hamilton’s equations, 1
Fermion, 185, 242, 244, 416 Harmonic oscillator, 69
Fermi’s golden rule (FGR), 339 classical, 69
FGR rate, 340, 354 quantum, 80
Field operator, 417 three dimensional, 111
Index 429

Hartree approximation, 260, 261, 263 Laguerre polynomial, 178

Hartree-Fock (HF) Laguerre polynomial, associated, 179
approximation, 263, 268 Lande’s interval rule, 252
energy, 268 Laplacian, 175
equation, canonical form, 372 Laporte’s rule, 253
equation, restricted, 373 LCAO-MO
method, 378 approximation, 289, 292, 379
Heisenberg picture, 333 symmetry adapted, 302, 309
Heisenberg’s uncertainty principle, 9 Legendre equation, associated, 139
Helium atom, 254 Linear combination, 19
Hermite equation, 73 Lowest unoccupied molecular orbital (LUMO),
Hermite polynomial, 75, 154 310
Hermite polynomial, recursion relationship, 77 LS-coupling, 248
Hermitian conjugate, 40 Lyman series, 174
Hessian matrix, 117, 379
HF-Roothaan equation, 376, 377
Highest occupied molecular orbital (HOMO), M
310 Magnetic permeability, 357
Hückel approximation, 292 Magnetic quantum number, 182
Hund’s rules, 251, 291 Many-electron atoms, 243
Hydrogen atom, 173 Marcus rate, 354
Hydrogen-like system, 187 Material wave, 9
Matrix representation, 303
I Matrix representation, character of, 303
Identity resolution, 392 Maxwell’s equations, 3, 7, 357
Improper rotation, 299 Measurement, 12, 50
Independent electron model, 242, 255 positive operator valued measure (POVM),
Influence functional, 398 51
Interaction picture, 335, 399 projective, 51
Internal conversion, 316 Minimum uncertainty state, 78
Intersystem crossing, 316 Molecular orbital, linear combination of
Inversion symmetry, 288, 292 atomic orbitals as (LCAO-MO), 280
IR absorption, 87 Moment of inertia, 135, 138, 150, 153
IR spectroscopy, selection rules, 86 Momentum operator, three dimensional, 104
Morse potential, 88
Multidimensional space, 101
J Multi-reference, 380
Jacobi identity, 49
jj -coupling, 249
N
Neutron, 185
K
Non-radiative transition, 316
Keldysh formalism, 419
Norm, 39
Ket, 36
Normalization condition, 22
position, 42
Normal mode, 118
space, 37
Normal mode frequency, 118
Kinetic energy, rotational, 149
Koopmans’ theorem, 287, 288
Kronecker-delta symbol, 25
O
Observables
L compatible, 49
Lagrange multiplier, 367 maximally compatible, 49
Laguerre equation, 179 Occupation number space, 415
430 Index

Operator, 40 spin of, 253

annihilation, 414, 416 Physical observable, 42
canonical momentum, 113 Planck constant, 7
compatible, 49 Planck, M., 6
creation, 414, 416 Plane wave, 358
field, 414 Point particle, 3
Hermitian, 17, 41, 54 Point symmetry operation, 298
identity, 41 Polarizability tensor, 120
linear, 16 Position basis, 392
lowering, 79 Position bra, three dimensional, 102
number, 416 Position ket, three dimensional, 102
raising, 79 Position representation, 112
time evolution, 54, 332, 392 Position representation, multidimensional,
time-ordered exponential, 348 112
unitary, 52 Potential well-depth, 89
Orbital, 243 Principal quantum number, 181, 245
anti-bonding, 284 Product
bonding, 284 associative rule, 39
degenerate, 243 direct, 56
Orthogonality, 18 distributive rule, 40
Orthogonality theorem, 307, 317 inner, 38
Overlap matrix element, 206 outer, 39
Projection super-operator, 399, 404
Propagator
P imaginary time, 396
Paschen series, 174 real time, 392
Path integral, 391 Proper rotation, 299
Pauli exclusion principle, 242 Proton, 185
Pauli principle, 290
P-branch, 155, 159
Perturbation approximation, first order time Q
dependent, 338 Q-branch, 155
Perturbation energy Quantum Liouville equation, 342
degenerate first order, 219 Quantum Liouville equation, interaction
degenerate second order, 221 picture, 399
first order, 214 Quantum Liouvillian, 343
second order, 215 Quantum master equation (QME), 399
Perturbation state second order time-local, 408, 411
degenerate first order, 222 second order time-nonlocal, 408
degenerate zeroth order, 219 time-local, 406, 407
first order, 214 time-nonlocal, 406
second order, 216 Quantum mechanics, 11
Perturbation theory Quantum number, 22
complete active space (CAS), 380 Quantum state
degenerate, 217 mixed, 341, 342
non-degenerate, 212 pure, 332
time dependent, 335, 348 Quantum Trouvaille equation, 398
time independent, 210
Phonon, 414
Phosphorescence, 316 R
Photoelectric effect, 7 Radial probability density, 179
Photon, 8, 185, 416 R-branch, 155, 159
energy of, 8 Rearrangement theorem, 318
speed of, 8 Reduced mass, 115
Index 431

Reflection operator, 299 photoelectron, 288

Reflection symmetry, 292 Raman, 120, 311
Refractive index, 358 rotational Raman, 150, 152
Relativistic effect, 269 selection rule, 311
Representation vibrational, 86
irreducible, 304 vibrational Raman, 87
reducible, 304 Spherical harmonics, 141, 175
R-head reversal, 159 Spin eigenstate, 185
Ritz combination principle, 174 Spin multiplicity, 290, 316
Rotating wave approximation, 339 Spin operator, 185
Rotational Raman Spin-orbit coupling, 187, 316
anti-Stokes transition, 152 Spin-orbit interaction, 269
selection rule, 152 Spin quantum number, total, 245, 247
Stokes transition, 152 Spin states
Rotational wavefunction, 136 antisymmetric, 246
Ro-vibrational transition, 150, 153 symmetric, 246
Russell-Saunders coupling, 248 Stationary state, 14
Rydberg constant, 173 Stokes vibrational transition, 88
Super-operator
negatively time ordered exponential, 400
S positively time ordered exponential, 400
Schönflies notation, 301 Symmetry, 297
Schrödinger equation, 11 Symmetry element, 298
Dirac notation, 106 Symmetry operation, 298
time dependent, 24, 55, 332 Symmetry operation, point, 298, 301
time independent, 15
time independent and three dimensional,
105 T
Schrödinger picture, 332 Term symbol, molecular, 289
Schrödinger’s theory, 9 Time local equation, 403
Schur’s Lemma, 320 Time ordering, chronological, 348
Second quantization, 414 Total spin quantum number, 185
Secular equation, 208 Trace, 343
Selection rules, 291 Transition dipole, 339
general many-electron atoms, 253 Trial state, 206
orbital angular momentum, 188 Triplet, 246
spin, 188 Triplet-triplet transition, 254
Self consistent field (SCF) Trotter factorization, 392
approximation, 254, 258, 260 Two-electron spin-orbit state, 246
complete active space (CAS), 380 Two-electron orbital state, 246
Self consistent field (SCF) method, Two-electron spin state, 246
multi-configuration (MC), 380
Separation of variables, 109
Separation of variables method, 102, 111 U
Shell, 243 Ultraviolet divergence, 7
Singlet-singlet transition, 254 Uncertainty principle, Heisenberg’s, 61
Slater determinant, 263, 364 Uncertainty relationship, general, 59, 61
Slater-type basis function, 379 Ungerade, 283
Sodium D lines, 254 Unitary transformation, 53, 344
Spectroscopy
atomic, 252
electronic absorption, 315 V
IR, 86, 120, 311 Vacuum state, 414
microwave, 150 Variational principle, 204, 280
432 Index

Velocity, phase, 4 W
Vibrational modes Wave, 3
IR active, 120 Wavefunction, 11
Raman active, 120 Wavelength, 4
silent, 120 Wavelength, de Brogile’s, 9
Vibrational Raman Wave-particle duality, 10
anti-Stokes transition, 88, 120 Wavevector, 4
selection rules, 88
Stokes transition, 88, 120
Z
Vibrational term, 87
Zeeman effect, 182