Springer Tracts in Modern Physics 282

Kazutaka Nakamura

Quantum
Phononics
Introduction to Ultrafast Dynamics of
Optical Phonons
Springer Tracts in Modern Physics

Volume 282

Series editors
Yan Chen, Department of Physics, Fudan University, Shanghai, China
Atsushi Fujimori, Department of Physics, University of Tokyo, Tokyo, Japan
Thomas Müller, Institut für Experimentelle Kernphysik, Universität Karlsruhe,
Karlsruhe, Germany
William C. Stwalley, Department of Physics, University of Connecticut, Storrs,
USA
Jianke Yang, Department of Mathematics and Statistics, University of Vermont,
Burlington, VT, USA
Springer Tracts in Modern Physics provides comprehensive and critical reviews of
topics of current interest in physics. The following fields are emphasized:
– Elementary Particle Physics
– Condensed Matter Physics
– Light Matter Interaction
– Atomic and Molecular Physics
– Complex Systems
– Fundamental Astrophysics
Suitable reviews of other fields can also be accepted. The Editors encourage
prospective authors to correspond with them in advance of submitting a manuscript.
For reviews of topics belonging to the above mentioned fields, they should address
the responsible Editor as listed in “Contact the Editors”.

More information about this series at http://www.springer.com/series/426

Kazutaka Nakamura

Quantum Phononics
Introduction to Ultrafast Dynamics of Optical
Phonons

123
Kazutaka Nakamura
Laboratory for Materials and Structures,
Institute of Innovative Research
Tokyo Institute of Technology
Yokohama, Japan

ISSN 0081-3869 ISSN 1615-0430 (electronic)

Springer Tracts in Modern Physics
ISBN 978-3-030-11923-2 ISBN 978-3-030-11924-9 (eBook)
https://doi.org/10.1007/978-3-030-11924-9

Library of Congress Control Number: 2019930564

© Springer Nature Switzerland AG 2019

This work is subject to copyright. All rights are reserved 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, express 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
Preface

Phonon are quantums of lattice vibrations and are bosons as well as photon. The
phonons are nowadays coherently generated and controlled by using an ultrashort
optical pulse. Quantum phononics is a new field in condensed-matter physics and
materials science, which includes the generation, detection, and the engineering and
coherent control of quantum states of lattice vibrations. This book describes the
fundamentals of generation, detection, and coherent control of optical phonons
using ultrashort laser pulses. While they are important for heat transfer and
capacity, acoustic phonons are not included in this book. The required basic
knowledge of quantum mechanics, quantum optics, solid state physics, and non-
linear spectroscopy is presented in Chaps. 1–4, which were developed during a
one-quarter course for graduate students in the department of materials science at
Tokyo Institute of Technology. Chapters 5–7 are based on the author’s recent
journal papers on coherent phonons and their coherent control. I have not tried to be
encyclopedic in the treatment, either in terms of complete references or a discussion
of every type of experiment and theory. Reference are included for more detailed
background or historical interest. Instead, I tried to explain detailed derivations for
important equations, which are often difficult for experimental physicists and
materials scientists.
Chapter 1 reviews the basics of quantum mechanics, state vectors, time evolu-
tion of a quantum state, and perturbation expansion, which are used throughout the
text. In Chap. 2, the density operator and a double-sided Feynman diagram are
introduced, which are used to describe nonlinear optical processes. Chapter 3
explains the quantum mechanics of the harmonic oscillator, which is of essential
importance in field quantization and phonon dynamics. Coherent and squeezed
states, which are important in quantum optics, are introduced. Chapter 4 reviews
lattice vibrations and field quantization and introduces the “phonon”. In Chaps. 5
and 6, we discuss coherent phonons using ultrafast optical measurements and the
quantum mechanical description of the generation and detection mechanism.
Section 5.3 shows squeezed phonons. In Chap. 7, we discuss the coherent control of
optical phonons using an ultrashort pulse train and its selected applications.

v
vi Preface

I especially thank Prof. Yosuke Kayanuma, my co-worker for several years, for
his guidance and advice on theory. I thank my colleagues and students participated
in my research projects on coherent phonons, especially Prof. Masahiro Kitajima,
Prof. Oleg V. Misochko, Prof. Fujio Minami, Prof. Yutaka Shikano, Dr. Yasuaki
Okano, Dr. Hiroshi Takahashi, Dr. Jianbo Hu, and Dr. Katsura Norimatsu. I also
thank Gordon Han Ying Li for checking English and equations in this book.
Immeasurable thanks are due to my wife Mitsuko, without her constant compan-
ionship and support this book would never have been possible. My thanks also go to
Dr. Claus Ascheron, Physics Editor of Springer-Verlag at Heidelberg (now retired),
for his kind support.

Yokohama, Japan Kazutaka Nakamura

Contents

1 Time Evolution of Quantum State . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Description of Quantum State . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 Hermitian Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.2 Projection Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Measurements, Observable, and Expectation Value . . . . . . . . . . . . 3
1.4 Time Evolution of a Quantum System . . . . . . . . . . . . . . . . . . . . . 4
1.4.1 Schrödinger Picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4.2 Heisenberg Picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4.3 Interaction Picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2 Density Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1 Pure State and Mixed State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Density Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Definition of Density Operator . . . . . . . . . . . . . . . . . . . . . 11
2.2.2 Time Evolution of the Density Operator . . . . . . . . . . . . . . 13
2.2.3 Perturbative Expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3 Density Operator for a Two-Level System . . . . . . . . . . . . . . . . . . 18
2.3.1 Interaction-Free Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.2 Time-Dependent Interaction Case . . . . . . . . . . . . . . . . . . . 19
2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3 Harmonic Oscillator and Coherent and Squeezed States . . . . . . . . . 25
3.1 Hamiltonian and Energy Eigenstate . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.1 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.2 Annihilation and Creation Operators . . . . . . . . . . . . . . . . . 29
3.1.3 Wave Function in a Position Space . . . . . . . . . . . . . . . . . . 30

vii
viii Contents

3.2 Time Evolution of a State in the Harmonic Oscillator System . . . . 32

3.3 Coherent States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3.1 Definition of the Coherent State . . . . . . . . . . . . . . . . . . . . 34
3.3.2 Uncertainty Relation of the Coherent State . . . . . . . . . . . . 36
3.3.3 Displaced Vacuum State . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.4 Time Evolution of the Coherent State . . . . . . . . . . . . . . . . 38
3.4 Squeezed States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.4.1 Squeezed Vacuum States . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.4.2 More General Squeezed States . . . . . . . . . . . . . . . . . . . . . 48
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4 Lattice Vibration and Phonon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.1 Linear Atomic Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.1.1 Classical Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.1.2 Quantum Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2 Linear Diatomic Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.2.1 Classical Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.2.2 Quantum Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5 Coherent Phonons: Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.1 A Brief History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.1.1 Experiments of Coherent Optical Phonons . . . . . . . . . . . . 67
5.1.2 Generation Mechanism of Optical Coherent Phonons . . . . . 68
5.2 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.2.1 Optical Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.2.2 X-Ray Diffraction Measurement . . . . . . . . . . . . . . . . . . . . 75
5.3 Squeezed Phonons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6 Coherent Phonons: Quantum Theory . . . . . . . . . . . . . . . . . . . . . . . . 81
6.1 Generation Mechanism with Displaced Harmonic Oscillator . . . . . 81
6.1.1 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.1.2 Transition Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.1.3 Second-Order Perturbation . . . . . . . . . . . . . . . . . . . . . . . . 85
6.1.4 Time Evolution of Phonon Amplitude . . . . . . . . . . . . . . . . 91
6.1.5 Gaussian Pulse Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
6.2 Four-Level Model and Double-Sided Feynman Diagrams . . . . . . . 103
6.2.1 Impulsive Absorption Process . . . . . . . . . . . . . . . . . . . . . . 103
6.2.2 Impulsive Stimulated Raman Scattering Process . . . . . . . . 106
Contents ix

6.3 Extension to Band Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

6.4 Optical Detection Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
7 Coherent Control of Optical Phonons . . . . . . . . . . . . . . . . . . . . . . . . 115
7.1 Coherent Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
7.2 Experiment of Coherent Control of Optical Phonons . . . . . . . . . . 115
7.2.1 Coherent Control of Optical Phonons in Diamond . . . . . . . 116
7.3 Coherent Control Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
7.3.1 Transparent Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
7.3.2 Opaque Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
7.4 Selected Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
7.4.1 Mode-Selective Excitation . . . . . . . . . . . . . . . . . . . . . . . . 121
7.4.2 Mode Coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
7.4.3 Controlling Phase Transition . . . . . . . . . . . . . . . . . . . . . . 122
7.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Appendix A: Linear Diatomic Chain: Normal Coordinate
and Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Appendix B: Ultrashort Laser Technology . . . . . . . . . . . . . . . . . . . . . . . . 131
Appendix C: Mathematical Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Chapter 1
Time Evolution of Quantum State

In this chapter, we summarize basic knowledge of quantum mechanics required for

representing a phonon and its time evolution. Ket and bra vectors are used to represent
a quantum state. Time evolution of the quantum state is expressed by using three
ways: Schrödinger picture, Heisenberg picture, and interaction picture. Perturbation
expansion of a time-dependent ket vector is explained for the interaction picture.

1.1 Description of Quantum State

In quantum mechanics, a physical state is represented by a state vector in a complex

vector space [1]. Using the Dirac notation [2], a quantum state is represented by a ket
vector denoted by |ϕ, which can be represented by a column vector with complex
components. We now introduce the bra vector, which is denoted by ϕ|, dual to the
ket vector.1 In a two-dimensional vector space, |ϕ is represented by
 
C1
|ϕ = , (1.1)
C2

where C1 and C2 are complex numbers. When we set another ket vector |ϕ  
 
 C1
|ϕ  = , (1.2)
C2

1 We consider a dual vector space using ket and bra vector spaces.
© Springer Nature Switzerland AG 2019 1
K. Nakamura, Quantum Phononics, Springer Tracts in Modern Physics 282,
https://doi.org/10.1007/978-3-030-11924-9_1
2 1 Time Evolution of Quantum State

the vector inner product is defined by

 

  C1
ϕ |ϕ = C1∗ C2∗ = C1∗ C1 + C2∗ C2 , (1.3)
C2

where C1∗ and C2∗ are complex conjugates of C1 and C2 , respectively. The inner
product of their own is

ϕ|ϕ = C1∗ C1 + C2∗ C2 = |C1 |2 + |C2 |2 . (1.4)

√
ϕ|ϕ is known as the norm and corresponds to the length of the ket vector |ϕ.
When two vectors |ϕ and |ϕ   are orthogonal each other, their inner product is zero:
ϕ  |ϕ = 0.

1.2 Operator

An operator  acts on a ket vector from the left side Â|ϕ and the resulting product
is another ket vector. When the ket vector is represented by a N -dimensional vector,
the operator  corresponds to the N × N matrix. When a vector |ϕ satisfies the
following equation
Â|ϕ = a|ϕ, (1.5)

where a is the constant, |ϕ and a are called eigenvector and eigenvalue for Â.

1.2.1 Hermitian Operator

The Hermitian conjugate operator for a linear operator  is expressed by † . By

setting the matrix elements Ai, j , those of the Hermitian conjugate operators are
defined by ( † )i, j = (A j,i )∗ . When the Hermitian conjugate operator † is the same
as the original operator Â, this operator is called Hermitian operator. The Hermitian
operator has the following properties:
• Eigenvalues are real.
• Eigenvectors, belonging to the different eigenvalues, are orthogonal each other.
Thus, the operator for the observable physical quantity is an Hermitian operator.
Furthermore, eigenvectors of an Hermitian operator are complete system, and the
normalized eigenvectors are used as the basis set of an orthogonal system. Any sate
|ϕ can be expanded using all of eigenvectors |φi  of the Hermitian operator

|ϕ = Ci |φi , (1.6)
i

where Ci is a complex number.

1.2 Operator 3

1.2.2 Projection Operator

Setting an operator P̂(a) = |aa|, in which |a is normalized, the operator selects
the component of a ket vector (|ϕ) parallel to |a:

P̂(a)|ϕ = |aa|ϕ = a|ϕ|a, (1.7)

where a|ϕ is a classical number. The operator is known as the projection operator
along the base ket vector |a [1]. Using the normalized orthogonal system {|ai }, the
identity operator Iˆ is expressed using the projection operator
 
Iˆ = P̂(ai ) = |ai ai |. (1.8)
i i

When we operate the identity operator on any operator B̂, the operator does not
change:  
B̂ = Iˆ B̂ Iˆ = |ai ai | B̂|a j a j | = ai | B̂|a j |ai a j |. (1.9)
i, j i, j

Thus, any operator can be represented as linear combinations of |ai a j |. In partic-
ular, the Hermitian operator  can be expressed by a linear combination of its own
eigenvectors {|αi }, where Â|αi  = αi |αi :
 
 = Ci |αi |ai αi | = Ci P̂(αi ). (1.10)
i i

This expression is called spectral decomposition of the Hermitian operator.

1.3 Measurements, Observable, and Expectation Value

A dynamical variable such as position and momentum is represented by an operator

that can act on a ket vector. A measurement of the dynamical variable gives one of its
eigenvalues.2 When the system |ϕ is represented  by a linear combination of eigen-
states |ai  of the dynamical variable Â, |ϕ = i Ci |ai , before the measurement,
one of eigenstates |ai  is selected and its eigenvalue ai is obtained by the measure-
ment. We do not know which eigenstate is selected, but the probability Prob(ai ) that
the |ai  is selected is obtained by

Prob(ai ) = |ai |ϕ|2 = |Ci |2 . (1.11)

2 Dirac wrote in his book “In this way we see that a measurement always causes the system to jump

onto an eigenstate of the dynamical variable that is being measured, the eigenvalue this eigenstate
belongs to being equal to the result of the measurement” [2]. Dirac’s words are also referred to in [1].
4 1 Time Evolution of Quantum State

The expectation value of  taken with respect to state |ϕ is defined by

 Â ≡ ϕ| Â|ϕ

 
= ϕ|a j a j | Â|ai ai |ϕ = C ∗j a j |ai |ai Ci
i, j i, j
 
= ai |Ci | =
2
Prob(ai )ai . (1.12)
i i

When eigenstates of a real dynamical variable form a complete set, it is called an

observable. An observable is represented by a Hermitian operator because its eigen-
values are real.

1.4 Time Evolution of a Quantum System

There are three types of representation for time evolution of a quantum system (equa-
tion of motion in quantum mechanics): Schrödinger, Heisenberg, and Interaction
pictures [3].

1.4.1 Schrödinger Picture

In the Schrödinger picture, time evolution of a quantum system is represented by

time evolution of state vectors. When we write the state vector at time t as |ϕ(t) for
a closed quantum system, time evolution of this state is defined by the Schrödinger
equation:
d
i |ϕ(t) = Ĥ |ϕ(t), (1.13)
dt

where Ĥ is the Hermitian operator representing the total energy of the system. Ĥ is
called the Hamiltonian and is the time-independent operator for the closed system.
When there is no degeneracy, the energy eigenstate |n for the eigenvalue E n satisfies

Ĥ |n = E n |n, (1.14)

which is called time-independent Schrödinger equation. When the state |ϕ(t) starts
from the energy eigenstate |n, |ϕ(0) = |n, the state at time t is expressed as

|ϕ(t) = e−i En t/ |n. (1.15)

1.4 Time Evolution of a Quantum System 5

Only the phase factor is different between |ϕ(t) and |ϕ(0). The phase factor rotates
with an angular frequency ωn = E n /. In the closed system, the probability of detect-
ing the energy eigenstate does not change and the energy eigenstate is a steady state.
Since an arbitrary state in the closed system is expanded
 using energy eigenstates,
any initial state |ψ(0) is represented by |ψ(0) = n Cn |n. Then the state |ψ(t)
at time t is expressed by
 
|ψ(t) = Cn e−i En t/ |n = Cn e−iωn t |n. (1.16)
n n

1.4.1.1 Time-Evolution Operator

The norm of the state vector |ψ(t) is expressed by


ψ(t)|ψ(t) = Cn∗ Cn  eiωn t e−iωn t n|n = ψ(0)|ψ(0), (1.17)
n,n 

and is constant throughout the time evolution. Thus, the transformation from |ψ(0)
to |ψ(t) is a unitary transformation. We consider the unitary operator Û (t) for the
time evolution,
|ψ(t) = Û (t)|ψ(0). (1.18)

This unitary operator is called the time-evolution operator. Using the Schrödinger
equation, the time-evolution operator satisfies

d
i Û (t) = Ĥ Û (t), (1.19)
dt

with an initial condition Û (0) = Iˆ. If the Hamiltonian is time independent, the time-
evolution operator is expressed by using an exponential form3

−i Ĥ t
Û (t) = exp . (1.20)


If the Hamiltonian is time dependent, the equation for Û (t) is modified as

d
i Û (t) = Ĥ (t)Û (t). (1.21)
dt

3 The exponential of a linear operator  is defined by as a power series expansion: exp  ≡ Iˆ +

 + (1/2!) Â2 + (1/3!) Â3 · · · .
6 1 Time Evolution of Quantum State

Integrating this equation, we get formally

t t
d
i Û (t1 )dt1 = Ĥ (t1 )Û (t1 )dt1
0 dt1 0
t
iÛ (t) = iÛ (0) + Ĥ (t1 )Û (t1 )dt1 . (1.22)
0

If there is no change in |ψ(t) from |ψ(0) at t = 0, then Û (0) = 1 and we get

t
1
Û (t) = 1 + Ĥ (t1 )Û (t1 )dt1 . (1.23)
i 0

Inserting Û (t) defined by this equation into the integration, we get

 
1 t 1 t1
Û (t) = 1 + Ĥ (t1 ) 1 + Ĥ (t2 )Û (t2 )dt2 dt1
i 0 i 0
 2 t  t1 
1 t 1
= 1+ Ĥ (t1 )dt1 + Ĥ (t1 ) Ĥ (t2 )Û (t2 )dt2 dt1
i 0 i 0 0
t  2 t t2
1 1
= 1+ dt1 Ĥ (t1 ) + dt2 dt1 Ĥ (t2 ) Ĥ (t1 )Û (t1 ), (1.24)
i 0 i 0 0

where t1 and t2 replaced each other in the third term.

By repeating again this procedure, we get
 2 t
1 t 1 t2
Û (t) = 1 + dt1 Ĥ (t1 ) + dt2 dt1 Ĥ (t2 ) Ĥ (t1 )
i 0 i 0 0
 3 t t3 t2
1
+ dt3 dt2 dt1 Ĥ (t3 ) Ĥ (t2 ) Ĥ (t1 )Û (t1 ). (1.25)
i 0 0 0

By repeating this procedure indefinitely, we finally get

∞  
1 n t tn t2
Û (t) = dtn dtn−1 · · · dt1 Ĥ (tn ) Ĥ (tn−1 ) · · · Ĥ (t1 )
n=0
i 0 0 0
 
1 t  
= exp+ Ĥ (t )dt , (1.26)
i 0

where exp+ (x) is called the time-ordered exponential.

Consider a small time period Δt = t/N , the time-ordered integral is expressed by
 t  N −1  
1   1
exp+ Ĥ (t )dt = lim exp+ Ĥ (nΔt) . (1.27)
i 0 N →∞ i
n=0

This can be understood as the Hamiltonian is constant during the small period Δt.
1.4 Time Evolution of a Quantum System 7

1.4.2 Heisenberg Picture

The linear operator of a physical quantity is time dependent in the Heisenberg picture,
while in the Schrödinger picture, the operator is time independent and the state vector
is time dependent. The expectation value of physical observable  at time t is obtained
in the Schrödinger representation

At = ϕ(t)| Â|ϕ(t) = ϕ(0)|Û † (t) ÂÛ (t)|ϕ(0). (1.28)

If we set the time-dependent operator ÂH (t) and the time-independent state vector
|ϕH as

ÂH (t) = Û † (t) ÂÛ (t) (1.29)

|ϕH = |ϕ(0), (1.30)

where the subscript H denotes the Heisenberg representation, we get

At = ϕH | ÂH |ϕH . (1.31)

Therefore, the expectation value at time t can be expressed by using the time-
dependent linear operator and the time-independent state vector. In the Heisenberg
picture, the time evolution of ÂH (t) is expressed by4

d
i ÂH (t) = ÂH (t), Ĥ . (1.32)
dt

1.4.3 Interaction Picture

The interaction picture has characteristics somewhat intermediate representation

between the Schrödinger and Heisenberg representation. In this representation, we
separate the Hamiltonian into two parts Ĥ = Ĥ0 + ĤI , where Ĥ0 is the main Hamil-
tonian for the system and ĤI is for an interaction between the system and its environ-
ment. In addition, Ĥ0 is time independent and ĤI is time dependent.5 This is the case
for a system that interacts with light, which is an oscillating electromagnetic field. A
physical observable operator in the interaction picture ÂI (t) and a state vector |ϕI (t)
are defined by

4 The commutator between  and B̂ is defined as Â, B̂ =  B̂ − B̂ Â.

5 The interaction picture is useful to investigate light–matter interaction and nonlinear spectroscopy,

because the system interacts with an oscillating electromagnetic filed of light.

8 1 Time Evolution of Quantum State
   
i −i
ÂI (t) = exp Ĥ0 t  exp Ĥ0 t (1.33)
 
 
i
|ϕI (t) = exp Ĥ0 t |ϕ(t). (1.34)


Using the Schrödinger equation, we get

d
i |ϕI (t) = ĤI (t)|ϕI (t). (1.35)
dt
The solution is formally
   
i −i
|ϕI (t) = exp Ĥ0 t exp Ĥ t |ϕ(0). (1.36)
 

In addition, ÂI (t) obeys the equation

d i
ÂI (t) = Ĥ0 , ÂI (t) . (1.37)
dt 

The interaction picture can be understood as Ĥ0 being treated in the Heisenberg
picture and the interaction part ĤI in the Schrödinger picture.

1.4.3.1 Perturbative Expansion

When the Hamiltonian has a time-dependent perturbative interaction Ĥ (t) = Ĥ0 +

Ĥ  (t), time evolution of the ket vector is expressed by (1.34) in the interaction
representation. The solution is obtained formally as
∞ 
 
−i n t tn t2
|ϕI (t) = |ϕI (t0 ) + dtn dtn−1 · · · dt1
n=1
 t0 t0 t0

ĤI (tn ) ĤI (tn−1 ) · · · ĤI (t1 )|ϕI (t0 ), (1.38)

where the initial state is |ϕI (t0 ). The time-evolution operator from t0 to t is defined
as Û (t, t0 ) ≡ exp(−i Ĥ0 (t − t0 )/). The ket vector in the Schrödinger representation
is expressed as |ϕ(t) = Û (t, t0 )|ϕI (t) and |ϕ(t0 ) = |ϕI (t0 ). Using these relations,
we get
∞ 
 
−i n t tn t2
Û (t, t0 )|ϕ(t) = |ϕ(t0 ) +
†
dtn dtn−1 · · · dt1
n=1
 t0 t0 t0

ĤI (tn ) ĤI (tn−1 ) · · · ĤI (t1 )|ϕI (t0 ), (1.39)
1.4 Time Evolution of a Quantum System 9

and then
∞ 
 
−i n t tn t2
|ϕ(t) = Û (t, t0 )|ϕ(t0 ) + dtn dtn−1 · · · dt1
n=1
 t0 t0 t0

Û (t, t0 ) ĤI (tn ) ĤI (tn−1 ) · · · ĤI (t1 )|ϕI (t0 )

 ∞  
−i n t tn t2
= Û (t, t0 )|ϕ(t0 ) + dtn dtn−1 · · · dt1
n=1
 t0 t0 t0

Û (t, t0 )Û † (tn , t0 ) Ĥ  (tn )Û (t, tn )Û † (tn−1 , t0 ) Ĥ  (tn−1 )

· · · Ĥ  (t2 )Û (t2 , t0 )Û † (t1 , t0 ) Ĥ  (t1 )Û (t1 , t0 )|ϕ(t0 ). (1.40)

By using the relation: Û (tn , tn−1 ) = Û (tn , t0 )Û (t0 , tn−1 ) = Û (tn , t0 )Û † (tn−1 , t0 ), the
above equation can be expressed as
∞ 
 
−i n t tn t2
|ϕ(t) = Û (t, t0 )|ϕ(t0 ) + dtn dtn−1 · · · dt1
n=1
 t0 t0 t0

Û (t, tn ) Ĥ  (tn )Û (tn , tn−1 ) Ĥ  (tn−1 )

· · · Ĥ  (t2 )Û (t2 , t1 ) Ĥ  (t1 )Û (t1 , t0 )|ϕ(t0 ). (1.41)

This indicates that the system propagates with Û (t1 , t0 ) from t0 under the time-
independent Hamiltonian Ĥ0 until time t1 , when the system interacts with the inter-
action Hamiltonian (perturbation) Ĥ  (t1 ). After the interaction, the system propagates
with Û (t2 , t1 ) until time t2 and this process repeats until time t. Û (t, t0 )|ϕ(t0 ) is the
zeroth-order perturbation ket vector, and each term in the summation is the nth-order
one. For example, the second-order perturbation is
 2
−i t t2
|ϕ (2) (t) = dt2 dt1
 t0 t0

Û (t, t2 ) Ĥ  (t2 )Û (t2 , t1 ) Ĥ  (t1 )Û (t1 , t0 )|ϕ(t0 ). (1.42)

This is shown graphically in a single-sided Feynman diagram (Fig. 1.1) [4], where
time flows from the left side to the right side.

Fig. 1.1 Single-sided

Feynman diagram for the
second-order perturbation
10 1 Time Evolution of Quantum State

1.5 Summary

The quantum state is expressed using the ket and bra vectors. Dynamical variables
are represented by linear operators that act on ket vectors. Time evolution of the
quantum state is expressed by using three ways: the Schrödinger, Heisenberg, and
interaction pictures. The time-evolution operator Û (t), which is a unitary operator, is
introduced, and the ket at time t is expressed by |Ψ (t) = Û (t)|Ψ (0). The perturba-
tive expansion is used to express the time evolution of the ket vector in the interaction
picture and is shown graphically using the single-sided Feynman diagram [5].

References

1. Sakurai, J.J.: Modern Quantum Mechanics, Revised edn. Addison-Wesley Publishing Company
Inc., Reading (1994)
2. Dirac, P.A.M.: The Principles of Quantum Mechanics, 4th edn. Oxford University Press, Oxford
(1958)
3. Tannor, D.J.: Introduction to Quantum Mechanics, A Time-Dependent Perspective. University
Science Books, California (2007)
4. Hamm, P.: Principles of Nonlinear Optical Spectroscopy: A Practical Approach. http://www.
mitr.p.lodz.pl/evu/lectures/Hamm.pdf
5. I wrote this chapter referencing Refs. [1–4] and following books: Takahiro Sunagawa and
Masahiro Ueda, Ryoshi Sokutei to Ryoshi Seigyo (in Japanese, Quantum Measurement and
Quantum Control), Science Sya (2016); Yoshio Kuramoto and Junichi Ezawa, Ryoushi Rikigaku
(in Japanese, Quantum Mechanics), Asakura Shyoten (2008); Keiji Igi and Hikari Kawai, Kiso
Ryousi Rikigaku (in Japanese, Fundamental Quantum Mechanics), Kodansya (2007); Masahito
Ueda, Gendai Ryoushi Buturigaku (in Japanese, Modern Quantum Physics), Baifukan (2004);
Akira Shimizu, Shin-han Rryoushi Ron no Kiso (in Japanese, New edition Fundamental of
Quantum Physics), Science Shya (2004)
Chapter 2
Density Operator

This chapter describes the density operator, which is used to calculate light-matter
interaction and the generation of coherent phonons in Chap. 6. The density operator
can represent both a pure quantum state and a mixed state. The von Neumann equation
for the time evolution of the density operator is derived. The perturbative expansion of
its time evolution is explained and represented with double-sided Feynman diagrams.
Time evolution in a two-level system is shown as an example.

2.1 Pure State and Mixed State

A pure state is a state in which a system is defined by a state vector |ϕ with a unit
probability. On the other hand, when the system includes several quantum states |ϕi 
with a ratio of pi , thestate is called a mixed state. This state is quite different from a
√
superposition state ( i pi |ϕi ) and cannot be expressed by a ket vector. A mixed
state is expressed by a density operator. The mixed state is important when we treat
a quantum system consisting of many particles. In this system, some particles are
often in different quantum states and the system is expressed with a density operator.
On the other hand, when all particles are in a single state, the system is in a pure state.

2.2 Density Operator

2.2.1 Definition of Density Operator

When a system consists of state vectors |ϕi  with probabilities of pi , the density
operator ρ̂ of this system is defined by

© Springer Nature Switzerland AG 2019 11

K. Nakamura, Quantum Phononics, Springer Tracts in Modern Physics 282,
https://doi.org/10.1007/978-3-030-11924-9_2
12 2 Density Operator

ρ̂ ≡ pi |ϕi ϕi |, (2.1)
i


where i pi = 1 [1, 2]. The density operator is expressed by an outer product of ket
and bra vectors, which corresponds to a matrix. The probability that we detect the
system in the state |ϕi  is expressed by ϕi |ρ̂|ϕi  using the density operator. When a
system is in a pure state and has only one state |ψ, the density operator is

ρ̂ = |ψψ|. (2.2)

The density operator is a Hermitian operator because the probability pi is real and
can be expanded with a orthonormal basis set {|ϕi }. Using an orthonormal basis set,
the trace of an arbitrary operator  is defined by

T r [ Â] ≡ ϕi | Â|ϕi . (2.3)
i


Let us consider a superposition state |φ = i ci |ϕi , where |φ is normalized
and {|ϕi } is a orthonormal basis set. The density operator of this superposition state
is calculated as
 
ρ̂ = |φφ| = ci c∗j |ϕi ϕ j | = ρ̂  + ci c∗j |ϕi ϕ j |, (2.4)
ij i= j

where ρ̂  is the density operator of the mixed state composed of |ϕi . Then the density
operator of the superposition state has additional off-diagonal terms of ci c∗j |ϕi ϕ j |
compared to that of the mixed state. These off-diagonal terms are due to interfer-
ence between the quantum states. A state, which has such a interference term, has
“coherence”.
Here, we consider, for example, a system including N -particles, in which each
particle has only two levels (a ground state |1 and an excited state |2), and N1 and
N2 particles are in the |1 and |2 states, respectively. Figure 2.1 shows its typical
example. When there is no interaction between each particle, the density operator is
expressed by
N1 N2
ρ̂ = |11| + |22|, (2.5)
N N
where N = N1 + N2 . This is an example of the mixed state. On the other √ hand, in
a√pure state, all particles are expressed by the single ket vector |ψ = N1 /N |1 +
N2 /N |2 and the density operator is expressed by

N1 N2 N1 N2
ρ̂ = |ψψ| = |11| + |22| + (|12| + |12|) . (2.6)
N N N2

The last term shows an interference between two states and the coherence.
2.2 Density Operator 13

Fig. 2.1 A schematic of

pure and mixed states. The
particle has only two states
(|1 and |2). In the mixed
state, particles are in the
ground state (|1 represented
by black circles) or the
excited state (|2 represented
by white circles). In the pure
state, all particles are in the
√ state |ψ√=
same
N1 /N |1 + N2 /N |2
represented by gray circles

2.2.2 Time Evolution of the Density Operator

We consider a density operator (ρ̂(t)) at time t as


ρ̂(t) = pi |ϕi (t)ϕi (t)|. (2.7)
i

Since |ϕi (t) obeys the Schrödinger equation, the time evolution of the density oper-
ator is calculated by

d  d
ρ̂(t) = pi |ϕi (t)ϕi (t)|
dt i
dt
  d|ϕi (t) dϕi (t)|

= pi ϕi (t)| + |ϕi (t)
i
dt dt
  
i i
= pi − Ĥ |ϕi (t)ϕi (t)| + |ϕi (t)ϕi (t)| Ĥ
i
 
1  
= Ĥ , ρ̂(t) , (2.8)
i

which is called the von Neumann equation.1

1 This equation corresponds to the Liouville equation for classical mechanics. Then, this is also
called the quantum Liouville equation or Liouville–von Neumann equation [2].
14 2 Density Operator

When the Hamiltonian is separated into two parts as Ĥ = Ĥ0 + Ĥ  (t), where Ĥ0
is time independent, the density operator can be treated with the interaction picture.
The density operator ρ̂I (t) and the interaction ĤI (t) in the interaction picture are
given by

i Ĥ0 t −i Ĥ0 t
ρ̂I (t) = exp ρ̂(t) exp
 
i Ĥ0 t −i Ĥ0 t
ĤI (t) = exp Ĥ  (t) exp , (2.9)
 

and time evolution of the density operator is given by

d 1   
ρ̂I (t) = ĤI , ρ̂I (t) . (2.10)
dt i
The von Neumann equation in the interaction picture is formally the same as that in
the Schrödinger picture.

2.2.3 Perturbative Expansion

The solution of the von Neumann equation (4.11) is obtained formally as

1 t  
ρ̂(t) = ρ̂(0) + Ĥ (t), ρ̂(0) dt. (2.11)
i 0

We calculate this integral by using perturbative approximation. At the zeroth

order, ρ̂ (0) (t) = ρ̂(0). Inserting this into the integration, we get the first-order
approximation

1 t  
ρ̂ (1) (t) = Ĥ (t1 ), ρ̂(0) dt1 . (2.12)
i 0

This corresponds to that the interaction occurs once at time t1 . Inserting ρ̂ (1) (t) into
the integral, we get the second-order approximation
 2 t t2   
1
ρ̂ (2) (t) = dt2 dt1 Ĥ (t2 ), Ĥ (t1 ), ρ̂(0) . (2.13)
i 0 0

Repeating this procedure one after another, we get the nth-order approximation
2.2 Density Operator 15
 n t tn t2
1
ρ̂ (n) (t) = dtn dtn−1 dt1 · · ·
i 0 0 0
    
Ĥ (tn ), Ĥ (tn−1 ), . . . Ĥ (t1 ), ρ̂(0) . . . . (2.14)

Finally, we get
∞

ρ̂(t) = ρ̂ (n) (t). (2.15)
n=0

The first-order approximation includes a term

 
Ĥ (t1 ), ρ̂(0) = Ĥ (t1 )ρ̂(0) − ρ̂(0) Ĥ (t1 )

= Ĥ (t1 )|ϕ(0)ϕ(0)| − |ϕ(0)ϕ(0)| Ĥ (t1 ), (2.16)

which indicates that |ϕ(0) and ϕ(0)| get interaction once at time t1 . In a similar
way, the second-order approximation includes
  
Ĥ (t2 ), Ĥ (t1 ), ρ̂(0) = Ĥ (t2 ) Ĥ (t1 )|ϕ(0)ϕ(0)| − Ĥ (t2 )|ϕ(0)ϕ(0)| Ĥ (t1 )

− Ĥ (t1 )|ϕ(0)ϕ(0)| Ĥ (t2 ) − |ϕ(0)ϕ(0)| Ĥ (t1 ) Ĥ (t2 ), (2.17)

which consists of the ket having interaction at t1 and t2 , the ket having interaction at
t2 with the bra having interaction at t1 , the ket having interaction at t2 with the bra
having interaction at t1 , and the bra having interaction at t1 and t2 .

2.2.3.1 Perturbative Interaction

The von Neumann equation in the interaction picture is formally equivalent to that in
the Schrödinger picture. Then, the density operator in the interaction picture is also
expressed as
∞ 
 
−i n t tn t2
ρ̂I (t) = ρ̂I (t0 ) + dtn dtn−1 · · · dt1
n=1
 t0 t0 t0
    
ĤI (tn ), ĤI (tn−1 ), . . . , ĤI (t1 ), ρ̂I (t0 ) . . . . (2.18)

From this equation, the density operator in the Schrödinger picture is expressed as
∞ 
 
−i n t tn t2
ρ̂(t) = Û (t, t0 )ρ̂(t0 )Û (t, t0 ) +
†
dtn dtn−1 · · · dt1
n=1
 t0 t0 t0
    
Û (t, t0 ) ĤI (tn ), ĤI (tn−1 ), . . . , ĤI (t1 ), ρ̂(t0 ) . . . Û † (t, t0 ), (2.19)
16 2 Density Operator

Fig. 2.2 Double-sided

Feynman diagrams for the
first-order perturbation. a for
the first term, b for the
second term in (2.22)

where we used ρ̂I (t0 ) = ρ̂(t0 ). The first term is the zeroth-order perturbation and the
time-evolution operator Û (t, t0 ) and its Hermitian conjugate operator act on ket and
bra vectors, respectively. The first-order perturbation is obtained as
   
−i t
ρ̂ (1) (t) = dt1 Û (t, t0 ) ĤI (t1 ), ρ̂(t0 ) Û † (t, t0 ). (2.20)
 t0

The term in the integral is calculated by

Û (t, t0 ) ĤI (t1 )ρ̂(t0 ) − ρ̂(t0 ) ĤI (t1 ) Û † (t, t0 )

= Û (t, t0 )Û † (t1 , t0 ) Ĥ  (t1 )Û (t1 , t0 )|ϕ(t0 )ϕ(t0 )|Û † (t, t0 )
−Û (t, t0 )|ϕ(t0 )ϕ(t0 )|Û † (t1 , t0 ) Ĥ  (t1 )Û (t1 , t0 )Û † (t, t0 )
= Û (t, t1 ) Ĥ  (t1 )Û (t1 , t0 )|ϕ(t0 )ϕ(t0 )|Û † (t, t0 )
−Û (t, t0 )|ϕ(t0 )ϕ(t0 )|Û † (t1 , t0 ) Ĥ  (t1 )Û † (t, t1 ). (2.21)

The first term shows that the ket propagates under Ĥ0 until time t1 , interacts with
perturbation Ĥ  (t1 ) at time t1 , and propagates again under Ĥ0 and the bra propagates
under Ĥ0 without perturbation. In the second term, the perturbative interaction occurs
for the bra. This interpretation is expressed by using double-sided Feynman diagrams
(Fig. 2.2). The upper and lower lines show propagation of the ket and bra vectors,
respectively. Time is running from the left to the right.
Similar calculation gives the second-order perturbation as
 2   
−i t t2
ρ̂ (2) (t) = dt2 dt1 Û (t, t0 ) ĤI (t2 ), ĤI (t1 ), ρ̂(t0 ) Û † (t, t0 )
 t0 t0
2.2 Density Operator 17

Fig. 2.3 Double-sided

Feynman diagrams for the
first-order perturbation

 2
−i t t2
= dt2 dt1
 t0 t0

(Û (t, t2 ) Ĥ  (t2 )Û (t2 , t1 ) Ĥ  (t1 )Û (t1 , t0 )|ϕ(t0 )ϕ(t0 )|Û † (t, t0 )
− Û (t, t2 ) Ĥ  (t2 )Û (t2 , t0 )|ϕ(t0 )ϕ(t0 )|Û † (t1 , t0 ) Ĥ  (t1 )Û † (t, t1 )
− Û (t, t1 ) Ĥ  (t1 )Û (t1 , t0 )|ϕ(t0 )ϕ(t0 )|Û † (t2 , t0 ) Ĥ  (t2 )Û † (t, t2 )
+ Û (t, t0 )|ϕ(t0 )ϕ(t0 )|Û † (t1 , t0 ) Ĥ  (t1 )Û † (t2 , t1 ) Ĥ  (t2 )Û † (t, t2 )).
(2.22)

The perturbative interaction occurs two times for the ket or the bra in the first or
the fourth term, respectively, and occurs one time for both the ket and bra in the
second and third terms. The double-sided Feynman diagrams for the second-order
perturbation is shown in Fig. 2.3, in which (a)–(d) correspond each term in the (2.22).
18 2 Density Operator

2.2.3.2 Double-Sided Feynman Diagram Rule

The perturbative expansion of the time evolution of the density operator is expressed
using double-sided Feynman diagrams. Here, we list several rules for the double-
sided Feynman diagram [3].
• Time is running from the left to the right.
• The double-sided Feynman diagram consists of two lines which expressed time
evolution of ket and bra vectors.
• Interaction with the interaction Hamiltonian Ĥ  (t) is represented by arrows.
• Each diagram has a sign (−1)n , where n is the number of interaction to the bra
vector (arrows to the bottom line).

2.3 Density Operator for a Two-Level System

As an example of the density operator, here we consider a two-level system which

interacts with optical pulses. The system consists of the basis {|1, |2}: |1 and |2
represent the ground and excited states, respectively.

2.3.1 Interaction-Free Case

We consider, at first, an interaction-free case. The Hamiltonian is defined by

Ĥ = ε1 |11| + ε2 |22|, (2.23)

where ε1 and ε2 are the energy and ε1 < ε2 . The Hamiltonian is also expressed using
the matrix:
 
ε1 0
Ĥ = . (2.24)
0 ε2

The time evolution of the density operator, which is shown in the matrix form, is
obtained from the von Neumann equation
   
d ρ11 ρ12 −i 0 (ε1 − ε2 )ρ12
= , (2.25)
dt ρ21 ρ22  (ε2 − ε1 )ρ21 0

for the interaction-free case. The diagonal components, which correspond to popu-
lations, are constant (ρ11 (t) = ρ11 (0) and ρ22 (t) = ρ22 (0)) in time. The off-diagonal
components, which correspond to polarizations, oscillate
2.3 Density Operator for a Two-Level System 19

ρ12 (t) = eiω21 t ρ12 (0)

ρ21 (t) = e−iω21 t ρ21 (0), (2.26)

where ω21 = (ε2 − ε1 )/.

Using the density operator, phase relaxation can be phenomenologically intro-
duced as

ρ˙12 (t) = iω21 ρ12 (t) − Γ ρ12 (t)

ρ˙21 (t) = −iω21 ρ21 (t) − Γ ρ21 (t), (2.27)

where Γ is a relaxation rate, and we get

ρ12 (t) = eiω21 t e−Γ t ρ12 (0)

ρ21 (t) = e−iω21 t e−Γ t ρ21 (0). (2.28)

2.3.2 Time-Dependent Interaction Case

When the oscillating perturbation V̂ (t)2

V̂ (t) = γ e−iωt |21| + γ eiωt |12| (2.29)

is applied to the two-level system, the total Hamiltonian ( Ĥ (t) = Ĥ0 + V̂ (t)) is
expressed by
 
ε1 γ eiωt
Ĥ (t) = . (2.30)
γ e−iωt ε2

The von Neumann equation is

   
d ρ11 ρ12 −i γ eiωt ρ21 − γ e−iωt ρ12 (ε1 − ε2 )ρ12 + γ eiωt (ρ22 − ρ11 )
= .
dt ρ21 ρ22  (ε2 − ε1 )ρ21 + γ e−iωt (ρ11 − ρ22 ) γ e−iωt ρ12 − γ eiωt ρ21
(2.31)

2 This interaction corresponds to an dipole interaction between the two-level system and light with
an angular frequency of ω. When we use μ as a dipole moment and E 0 as an amplitude of elec-
tric field, γ = μE 0 . γ e−iωt |21| and γ eiωt |12| correspond to the excitation process with light
absorption and the de-excitation process with emission. In addition, the rotating wave approximation
is assumed.
20 2 Density Operator

By using the frequency Ω ≡ γ /, we get

ρ̇11 = −iΩ(eiωt ρ21 − e−iωt ρ12 )

ρ̇12 = iω21 ρ12 − iΩeiωt (ρ22 − ρ11 )
ρ̇21 = −iω21 ρ12 − iΩe−iωt (ρ11 − ρ22 )
ρ̇22 = −iΩ(e−iωt ρ12 − eiωt ρ21 ). (2.32)

The sum of the population in |1 and |2 is time independent, because ρ˙11 + ρ˙22 = 0.
From the results of the interaction-free case, the off-diagonal terms have e−iω21 t or
eiω21 t. Then, we use a rotating frame defined as ρ̃12 = e−iω21 t ρ12 , ρ̃21 = eiω21 t ρ21 ,
ρ̃11 = ρ11 , and ρ̃22 = ρ22 . In the rotating frame, the density matrix elements are

ρ̃˙12 = −iΩei(ω−ω21 )t (ρ̃22 − ρ̃11 )

ρ̃˙21 = iΩe−i(ω−ω21 )t (ρ̃22 − ρ̃11 )
ρ̃˙11 = −iΩei(ω−ω21 )t ρ̃21 + iΩe−i(ω−ω21 )t ρ̃12
ρ̃˙22 = iΩei(ω−ω21 )t ρ̃21 − iΩe−i(ω−ω21 )t ρ̃12 . (2.33)

For the resonant condition (ω − ω21 = 0), the equations are greatly simplified, and
we get

ρ̃˙12 − ρ̃˙21 = 2iΩ(ρ̃11 − ρ̃22 )

ρ̃˙11 − ρ̃˙22 = 2iΩ(ρ̃12 − ρ̃21 ), (2.34)

and

ρ̃¨11 − ρ̃¨22 = 2iΩ(ρ̃˙12 − ρ̃˙21 ) = −4Ω 2 (ρ̃11 − ρ̃22 ). (2.35)

If we put ρ̃11 − ρ̃22 = Ceat , we get a = ±i2Ω. For the initial condition of ρ11 (0) = 1
and ρ22 (0) = 0, the coefficient C is obtained to be 1/2. Then, we get

ei2Ωt + e−i2Ωt
ρ11 − ρ22 = = cos 2Ωt. (2.36)
2
By using ρ11 + ρ22 = 1, we finally get

1 − cos 2Ωt
ρ11 = = cos2 Ωt. (2.37)
2

The result shows that the system absorbs energy from the interaction potential V̂ (t)
in the time range between 0 and Ω/2 and emits in the time range between Ω/2 and
Ω (as shown in Fig. 2.4). This absorption and emission process periodically repeat
with a frequency of Ω, which is called Rabi frequency.
2.3 Density Operator for a Two-Level System 21

Fig. 2.4 Time evolution of 1.0

the population (ρ11 , and ρ22 )
of the ground and excited
states. The solid and dotted 0.8
curves show ρ11 and ρ22 ,
respectively

Population
0.6

0.4

0.2

0.0
0 1 2 3 4
time (π/Ω)

When we introduce the phase relaxation (Γ ) in this two-level system, ρ˙˜12 and ρ˙˜21
are obtained by

ρ̃˙12 = −Γ ρ˜12 − iΩei(ω−ω21 )t (ρ̃22 − ρ̃11 )

ρ̃˙21 = −Γ ρ˜21 iΩe−i(ω−ω21 )t (ρ̃22 − ρ̃11 ). (2.38)

At resonance, a similar calculation gives

 
1 e−Γ t √Γ 2 −4Ω 2 t √
− Γ 2 −4Ω 2 t
ρ11 = 1+ e +e . (2.39)
2 2

If we represent the relaxation rate in units of the Rabi oscillation (Γ = nΩ), the
population ρ11 is a damped oscillation

1 
ρ11 = 1 + e−nΩt cos( 4 − n 2 Ωt) , (2.40)
2
if 0 ≤ n ≤ 2 and

1 
ρ11 = 1 + e−nΩt cosh( n 2 − 4Ωt) , (2.41)
2
if n > 2.
The time evolution of the population (ρ11 and ρ22 ) is shown in Figs. 2.5 and
2.6 for Γ = 0.5 and Γ = 2.5 , respectively. Both ρ11 and ρ22 approach to the
steady-state value (1/2).
22 2 Density Operator

Fig. 2.5 Time evolution of 1.0

the population (ρ11 , and ρ22 )
of the ground and excited
states with relaxation 0.8
(Γ = 0.5 ). The solid and

Population
dotted curves show ρ11 and 0.6
ρ22 , respectively

0.4

0.2

0.0
0 1 2 3 4
time (π/Ω)

Fig. 2.6 Time evolution of 1.0

the population (ρ11 , and ρ22 )
of the ground and excited
states with relaxation 0.8
(Γ = 2.5 ). The solid and
Population

dotted curves show ρ11 and 0.6

ρ22 , respectively
0.4

0.2

0.0
0 1 2 3 4
time (π/Ω)

2.4 Summary

The density operator, which can describe a pure state and mixed state, is described.
The time evolution of the density matrix is obtained by the von Neumann equation.
The perturbative expansion of the time evolution of the density operator is explained
and represented by using a double-sided Feynman diagram. The time evolution of
the populations in the excited and ground states in the two-level model is calculated
[4].

References

1. Schatz, G.C., Ratner, M.A.: Quantum Mechanics in Chemistry. Dover Publication Inc., Mineola
(2002)
2. Tannor, D.J.: Introduction to Quantum Mechanics, A Time-Dependent Perspective. University
Science Books, California (2007)
References 23

3. Mukamel, S.: Principles of Nonlinear Optical Spectroscopy. Oxford University Press, New York
(1995)
4. I wrote this chapter referencing Refs. [1–3] and following books: Yoshio Kuramoto and Junichi
Ezawa, Ryoushi Rikigaku (in Japanese, Quantum Mechanics), Asakura Shyoten (2008); Keiji
Igi and Hikari Kawai, Kiso Ryousi Rikigaku (in Japanese, Fundamental Quantum Mechanics),
Kodansya (2007); Takahiro Sunagawa and Masahiro Ueda, Ryoshi Sokutei to Ryoshi Seigyo (in
Japanese, Quantum Measurement and Quantum Control), Science Sya (2016), Masahito Ueda
Gendai Ryoushi Buturigaku (in Japanese, Modern Quantum Physics), Baifukan (2004); Akira
Shimizu, Shin-han Rryoushi Ron no Kiso (in Japanese, New edition Fundamental of Quantum
Physics), Science Shya (2004); Kyo Inoue, Kogaku Kei no Tameno Ryoushi Kogaku (in Japanese,
Quantum Optics for Engineer), Morikita Shyoten (2015); Masahiro Matsuoka, Ryoushi Kogaku
(in Japanese, Quantum Optics), Shokabou (2000)
Chapter 3
Harmonic Oscillator and Coherent
and Squeezed States

This chapter describes the quantum mechanics of a harmonic oscillator, which is of

essential importance in treating a phonon, using creation and annihilation operators.
A number state, a coherent state, and a squeezed state are introduced. The number
state is an eigenstate of the energy of the harmonic oscillator. The coherent state is a
minimum uncertainty state. In the squeezed state, fluctuation of one of the conjugated
variables is reduced and lowered than that of the vacuum state. Coherent and squeezed
states are important in quantum optics and phononics. Time evolution of the mean
value of position and variance for the harmonic oscillator is discussed.

3.1 Hamiltonian and Energy Eigenstate

3.1.1 Hamiltonian

The classical Hamiltonian H of a one-dimensional harmonic oscillator with a mass

m and an angular frequency ω is

p2 mω2 x 2
H= + , (3.1)
2m 2
where x and p are position and momentum, respectively (Fig. 3.1).
 Replacing
 x and p to linear operators x̂ and p̂, which satisfy commutation relation
x̂, p̂ = i, we get the quantum mechanical Hamiltonian Ĥ by

p̂ 2 mω2 x̂ 2
Ĥ = + . (3.2)
2m 2

© Springer Nature Switzerland AG 2019 25

K. Nakamura, Quantum Phononics, Springer Tracts in Modern Physics 282,
https://doi.org/10.1007/978-3-030-11924-9_3
26 3 Harmonic Oscillator and Coherent and Squeezed States

Fig. 3.1 Potential energy of V(x)

the harmonic oscillator

Here, we introduce new operators â and â † [1], which are defined using x̂ and p̂ by

1  
â ≡ √ mω x̂ + i p̂ ,
2mω
1  
â † ≡ √ mω x̂ − i p̂ . (3.3)
2mω

The position and momentum operators (x̂ and p̂) are expressed by

  
x̂ = â + â † ,
2mω

mω  
p̂ = i −â + â † , (3.4)
2

using
 the new operators. Inserting this relationship into the commutation relation
x̂, p̂ , we get
 
x̂, p̂ = x̂ p̂ − p̂ x̂
−i  
= (â + â † )(â − â † ) − (â − â † )(â + â † )
2  
= −i â † â − â â †
 
= i â, â † . (3.5)

Then, the commutation relation between â and â † is

 
â, â † = 1, (3.6)
 
because x̂, p̂ = i. The operator â is not a Hermitian operator. The Hamiltonian
is expressed by using â and â † by
3.1 Hamiltonian and Energy Eigenstate 27
 2  2
1 mω   mω2   
Ĥ = i −â + â † + â + â †
2m 2 2 2mω
ω  † 
= â â + â † â
2
1
= ω â † â + . (3.7)
2

If we introduce a new Hermitian operator n̂ 1 by

n̂ ≡ â † â, (3.8)

the Hamiltonian is represented by a very simple form

1
Ĥ = ω n̂ + , (3.9)
2

using a single linear operator n̂.

3.1.1.1 Eigenenergy and Energy Eigenstates

Because the Hamiltonian Ĥ of the harmonic oscillator is represented by the single

linear operator n̂, it is enough to solve an eigenvalue equation for n̂ to obtain an
energy eigenvalue and eigenstate. Let us consider that the eigenvalue equation holds
for n̂ as

n̂|n = n|n, (3.10)

where n and |n are eigenvalue and eigenstate, respectively. Operating â on (3.10)
from the left side, the left-hand side of the equation is
   
â n̂|n = â â † â|n = â † â + 1 |n = n̂ + 1 |n, (3.11)

and the right-hand side of the equation is

ân|n = n â|n. (3.12)

Then, we get
 
n̂ + 1 |n = n â|n, (3.13)

1 n̂ † = (â † â)† = (â)† (â † )† = â † â = n̂, because (  B̂)† = B̂ † † .

28 3 Harmonic Oscillator and Coherent and Squeezed States

and

n̂ â|n = (n − 1)â|n. (3.14)

This equation means that â|n is also an eigenstate of the operator n̂ and its eigenvalue
is n − 1. We named an eigenstate belonging to the eigenvalue n |n, then an eigenstate
belonging to the eigenvalue n − 1 is |n − 1. The state â|n is expressed by

â|n = Cn |n − 1, (3.15)

where Cn is a constant. When we calculate the inner product with itself, the left-hand
side is

n|â † â|n = n|n̂|n = n, (3.16)

and the right-hand side is

n − 1|Cn∗ Cn |n − 1 = |Cn |2 ≥ 0. (3.17)

√
Then, we find that n ≥ 0 and Cn = ± n. We choose a positive number, because the
global phase can be set arbitrarily. Equation (3.15) becomes
√
â|n = n|n − 1. (3.18)

n should be a nonnegative integer: n = 0, 1, 2, 3, . . .. We explain this below [2]. The

operator â changes an eigenstate to another eigenstate belonging to the eigenvalue
reduced by one. Operating this operator m times to the eigenstate |n, we get the
eigenstate |n − m as
 m
â |n = n(n − 1) · · · (n − m + 1)|n − m. (3.19)

The eigenvalue is nonnegative and n ≥ m. Suppose an integer m 0 which satisfies

n ≥ m 0 > n − 1, the eigenstates should be
 m 0 )
â |n = 0, (3.20)

and
 m 0 +1)
â |n = 0. (3.21)

If (3.21) does not hold, we get a negative eigenvalue (n − (m 0 + 1)). By operating

â † to (3.21) from the left side, we get
 m +1    m  m
â † â 0 |n = â † â â 0 |n = n̂ â 0 |n = 0. (3.22)
3.1 Hamiltonian and Energy Eigenstate 29

Using (3.19), we get

 m
n̂ â 0 |n = (n − m 0 ) n(n − 1) · · · (n − m + 1)|n − m = 0. (3.23)

Then, we get n = m 0 . Since m 0 is a nonnegative integer, the eigenvalue n is a non-

negative integer.2 In conclusion, |n is an eigenstate of both the operator n̂ and the
Hamiltonian Ĥ , then we get the eigenvalue equation

1 1
Ĥ |n = ω n̂ + |n = ω n + |n, (3.24)
2 2

where n is a nonnegative integer (n = 0, 1, 2, . . .). Eigenenergies E n for the eigen-

state |n is
1
E n = ω n + . (3.25)
2

The harmonic oscillator has energy of ω/2 even at the lowest state |0, which is
called the zero-point energy, and a constant energy spacing of ω. The eigenstate
|n is called an occupation number state,3 because well-defined numbers of quanta
(ω) are included in the state. |0 is the lowest energy state and called the ground
state or the vacuum state.

3.1.2 Annihilation and Creation Operators

The operator â shifts a number state |n to the lower state |n − 1, and is called the
annihilation operator [1]. On the other hand, its conjugated operator â † is the creation
operator, which shifts the state to the upper state |n + 1. Operating â † to (3.10) from
the left side, the left-hand side is
   
â † n̂|n = â † â † â|n = â † â â † − 1 |n = n̂ − 1 â † |n (3.26)

and the right-hand side is

â † n|n = n â † |n. (3.27)

Then, we get

n̂ â † |n = (n + 1)â † |n, (3.28)

which means that â † |n is also an eigenvector with the eigenvalue of n + 1. So we

can write â † |n as

2m is, of course, a nonnegative integer, because it is number of times for the operation.
3 The occupation number state is also called the Fock state for photons in quantum optics.
30 3 Harmonic Oscillator and Coherent and Squeezed States

â † |n = Cn |n + 1, (3.29)

where Cn is a normalized constant. When we calculate the inner product with itself,
the left-hand side is

n|â â † |n = n|(n̂ + 1)|n = n + 1, (3.30)

and the right-hand side is

n + 1|Cn∗ Cn |n + 1 = |Cn |2 ≥ 0. (3.31)

Then, we get
√
â † |n = n + 1|n + 1. (3.32)

The creation or annihilation operator shifts an eigenstate upward or downward,

respectively. The lowest eigenstate is |0 and obeys

â|0 = 0. (3.33)

Operating â † n times to the lowest state |0, we get upper number states one by one
as [1]
|1 = â † |0
1 1
|2 = √ â † |1 = √ (â † )2 |0
2 2·1
1 1
|3 = √ â † |2 = √ (â † )3 |0
3 3·2·1
..
.
1 1
|n = √ â † |n − 1 = √ (â † )n |0. (3.34)
n n!

The annihilation
 and creation
 operators
 obey commutation relations with the number
operator as â, n̂ = â and â † , n̂ = â † .

3.1.3 Wave Function in a Position Space

The wave function ϕ0 (x) of the lowest state |0 is obtained in a position space by
solving (3.33).4 In the one-dimensional position space, the position and momentum
operators are expressed as

4 The wave function ϕ( x) is obtained by projection of the ket vector |ϕ into position space and has
the relationship ϕ( x) = x|ϕ.
3.1 Hamiltonian and Energy Eigenstate 31

x̂ = x
d
p̂ = −i , (3.35)
dx
and the annihilation operator is represented by

1 d
â = √ mωx +  . (3.36)
2mω dx

Then, the differential equation for the wave function ϕ0 (x) is

d
mωx +  ϕ0 (x) = 0, (3.37)
dx

and the solution is

−mωx 2
ϕ0 (x) = C exp . (3.38)
2

The normalization coefficient C is calculated by

∞
|ϕ0 (x)|2 dx = 1, (3.39)
−∞

and obtained as

mω
C= . (3.40)
π

Then, the wave function ϕ0 (x) is a Gaussian function centered at x = 0. The wave
functions for the upper number states are obtained by operating the creation operator

Fig. 3.2 The wave function

V(x)
for n = 0, n = 1, and n = 2
states. E 0 , E 1 , and E 2 are
ω/2, 3ω/2, and 5ω/2,
respectively n=2

n=1

n=0

x
32 3 Harmonic Oscillator and Coherent and Squeezed States

â † one by one to ϕ0 (x). For example, the wave function ϕ1 (x) for |1 is

1 d mω −mωx 2
ϕ1 (x) = â ϕ0 (x) = √
†
mωx −  exp
2mω dx π 2
2m −mωx 2
= √ x exp . (3.41)
 2π 2

The wave function ϕ1 (x) is a odd function and has a node at x = 0. The ϕ2 (x) is
also calculated by â † ϕ1 (x). The wave functions ϕ0 (x), ϕ1 (x), and ϕ1 (x) are shown
in Fig. 3.2.

3.2 Time Evolution of a State in the Harmonic

Oscillator System

The number state |n(t) at time t is defined as

1
|n(t) = exp −i n + ωt |n, (3.42)
2

because its energy is ω(n + 1/2). Of course, the probability distribution function
of the number state does not change in time n(t)|n(t) = n|n, because it is a
stationary state. The expectation value of position for the number state is calculated
to be zero:
 

x̂ = n(t)|x̂|n(t) = n| (â + â ) |n = 0.
†
(3.43)
2mω

The standard deviation σ is obtained by

σ = x̂ 2  − x̂2 , (3.44)

where x̂ 2  is calculated by

 2

x̂  = n|
2
(â + â † ) |n
2mω
  
= n| â 2 + â â † + â † â + (â † )2 |n
2mω
   (2n + 1)
= n| 2n̂ + 1 |n = . (3.45)
2mω 2mω
3.2 Time Evolution of a State in the Harmonic Oscillator System 33

Then, we get the standard deviation as


(2n + 1)
σ = . (3.46)
2mω

When the position of the harmonic oscillator is measured, the observed value is scat-
tered at every observation and its standard deviation is σ , with an average value of
zero. This feature is different from the characteristics of classical harmonic oscil-
lators, in which the position changes in time with its angular frequency ω. Such
oscillation of position can be realized in a superposition state of adjoining number
states. Suppose the superposition state of |0 and |1 as

1   ω
|ϕ(t) = √ |0 + e−iωt |1 e−i 2 t . (3.47)
2

The expectation value of position in this state is


1     
x = 0| + eiωt 1| â + â † |0 + e−iωt |1
2 2mω


= cos(ωt). (3.48)
2mω

The expectation value of x 2  is obtained by

1     
x 2  = 0| + eiωt 1| 2n̂ + 1 |0 + e−iωt |1
2 2mω

= , (3.49)
mω
and the standard deviation is

 
σ = − cos2 (ωt)
mω 2mω


= (1 + sin2 (ωt)). (3.50)
2mω

Then, the standard deviation for position also oscillates, which means the shape of
the probability distribution function changes in time.
On the other hand, a superposition state composed of number states, which differ
more than two, shows no oscillation in expectation value of position, because the
position operator includes only â and â † . Suppose the superposition state of |0
and |2 are

1   ω
|ϕ(t) = √ |0 + e−i2ωt |2 e−i 2 t . (3.51)
2
34 3 Harmonic Oscillator and Coherent and Squeezed States

The expectation value of position in the state is


1     
x = 0| + ei2ωt 2| â + â † |0 + e−2ωt |2 = 0. (3.52)
2 2mω

The expectation value of x 2  is obtained by

1    2  
x 2  = 0| + ei2ωt 2| â + (â † )2 + 2n̂ + 1 |0 + e−i2ωt |2
2 2mω
 √ 
= 3 + 2 cos(2ωt) , (3.53)
2mω
and the standard deviation is the same value
 √ 
σ = 3 + 2 cos(2ωt) . (3.54)
2mω
Therefore, in the superposition state, the mean value does not change and only the
standard deviation oscillates with 2ω.

3.3 Coherent States

3.3.1 Definition of the Coherent State

A coherent state5 is one of the most important states, which is used in quantum
optics and quantum information technologies [4]. The coherent state is defined by
an eigenstate of the annihilation operator â. Consider an eigenvalue and eigenstate
of â as α and |α

â|α = α|α. (3.55)

The eigenvalue α is, in general, complex, because â is not Hermitian. When we

expand this coherent state using number states |n, we get

|α = Cn |n, (3.56)
n

where Cn = n|α. Specific representation of Cn is given as follows. Operating n|

on (3.55) from the left side gives

5 The concept of the coherent state is first proposed by Schrödinger [3]. Much work on the coherent

state in quantum optics was performed by R. J. Glauber. The coherent state is also called the Glauber
state in quantum optics.
3.3 Coherent States 35

n|â|α = αn|α. (3.57)

On the other hand, from (3.32), n|â is

√
n|â = n + 1n + 1|, (3.58)

then, we get
√
n|â|α = n + 1n + 1|α. (3.59)

Comparing (3.57) and (3.59), we get

α
n + 1|α = √ n|α. (3.60)
n+1

Applying this relationship recursively, Cn is obtained as

α
Cn = n|α = √ n − 1|α
n
α 2
αn
= √ n − 2|α · · · = √ 0|α. (3.61)
n(n − 1) n!

Then, we get
 αn
|α = 0|α √ |n, (3.62)
n n!

because 0|α is a constant. Normalization of |α

 |α|2n
= |0|α|2 e|α| = 1
2
|α = |0|α|2 (3.63)
n
n!

requires

|0|α|2 = e−|α| .
2
(3.64)

Finally, we get the expansion form of |α as

 αn
|α = e−|α| /2
2
√ |n, (3.65)
n n!

where the global phase is set to be 1. The coherent state is the superposition state of
an infinite number of the number states. When we measure the harmonic oscillator
in the coherent state, the probability to find it in the eigenstate |n is
36 3 Harmonic Oscillator and Coherent and Squeezed States

Fig. 3.3 A typical example (a) 1.0

of the population of number
states |n for the coherent
state with α = 0.5 (a) and 0.8
α = 2 (b)

Population Pn
0.6

0.4

0.2

0.0
0 4 8 12
n
(b) 1.0

0.8
Population Pn

0.6

0.4

0.2

0.0
0 4 8 12
n

|α|2n
P(n) = |n|α|2 = e−|α|
2
, (3.66)
n!

which is a Poisson distribution.6 Then, the the coherent state is the superposition state
of energy eigenstates of the harmonic oscillator with a Poisson distribution. Typical
examples of the population of the number states in the coherent state is shown in
Fig. 3.3.

3.3.2 Uncertainty Relation of the Coherent State

Here, we describe the uncertainty relation between position and momentum in the
coherent state of the harmonic oscillator. The expectation values for position and

6 ThePoisson distribution P(n) = e−λ λk /k! represents the probability of events occurring k times
per unit time, when the events occur randomly λ per unit time.
3.3 Coherent States 37

momentum are
 
 
x = α|x̂|α = α|(â + â )|α =
†
(α + α ∗ )
2mω 2mω
 
mω mω
 p = α| p̂|α = −i α|(â − â )|α = −i
†
(α − α ∗ ), (3.67)
2 2

and x 2  and  p 2  are


x 2  = α|x̂ 2 |α = α|(â + â † )(â + â † )|α
2mω
   
= α|(â 2 + (â † )2 + 2â † â + 1)|α = (α + α ∗ )2 + 1
2mω 2mω
mω
 p 2  = α| p̂ 2 |α = − α|(â − â † )(â − â † )|α
2
mω  
=− (α − α ∗ )2 − 1 . (3.68)
2
Then, we get the variance


( x)2 =
2mω
mω
( p) =
2
, (3.69)
2
and the uncertainty relation is given by


( x)2 ( p)2 = . (3.70)
2
It shows the coherent state is a minimum uncertainty state for position and
momentum.

3.3.3 Displaced Vacuum State

Here, we describe that the coherent state is also derived using a displaced harmonic
oscillator. Consider the initial state as the ground state |0 of the harmonic oscillator,
in which the potential minimum is at x = 0 position (schematically shown in Fig. 3.4).
When this harmonic potential is suddenly shifted α along the x-coordinate, the initial
state transfers to a new state. This operation is described by a displacement operator
D̂(α):

D̂(α) = exp(α â † − α ∗ â). (3.71)

38 3 Harmonic Oscillator and Coherent and Squeezed States

Fig. 3.4 Schematic of the

coherent state expressed as V(x)
the shifted vacuum state. The
dotted curve shows a shifted
harmonic potential. The
dashed Gaussian is the
vacuum state

This operator is an unitary operator and satisfies D̂ † (α) = D̂(−α). The displacement
operator shifts the annihilation operator by α:

D̂ † (α)â D̂(α) = â + α. (3.72)

Operating D̂(α) on |0 gives

D̂(α)|0 = exp(α â † − α ∗ â)|0

1 † 
= exp(α â † ) exp(−α ∗ â) exp − α â , −α ∗ â |0
2
|α|2
= exp − exp(α â † ) exp(−α ∗ â)|0
2
|α|2
= exp − exp(α â † )|0
2
|α|2  αn
= exp − √ |n. (3.73)
2 n n!

Then, the suddenly displaced vacuum state is the coherent state.

3.3.4 Time Evolution of the Coherent State

The time-dependent coherent state is expressed by

∞
αn
|α(t) = e−i Ĥ t/ |α = 0|α √ e−i Ĥ t/ |n
n=0 n!
3.3 Coherent States 39

∞
αn
= 0|α √ e−i(n+1/2)ωt/ |n
n=0 n!
∞
 (αe−iωt â † )n
= 0|αe−iωt/2 |0
n=0
n!
 
= e−|α| /2 −iωt/2
exp αe−iωt â † |0.
2
e (3.74)

Comparing this equation with the time-independent coherent state expanded in the
number states
 
|α = e−|α| /2
2
exp α â † |0, (3.75)

we know that α is replaced by αe−iωt and the phase factor e−iωt/2 is multiplied in
the time-dependent coherent state. Then, we can rewrite as

|α(t) = e−iωt/2 |αe−iωt . (3.76)

The expectation values of position and momentum are obtained by replacing α to

αe−iωt in (3.67):

2
x(t) = α cos(ωt)
mω
√
 p(t) = = 2mωα sin(ωt), (3.77)

for a real α value. Then, the expectation value of position oscillates with the angular
frequency ω. As shown in (3.69), the variances of both position and momentum do
not depend on α and are time independent. This implies that the shape of the wave
packet of the coherent state does not change in time. Then, the coherent state is
similar to a classical harmonic oscillator.

3.4 Squeezed States

3.4.1 Squeezed Vacuum States

The coherent state is a minimum uncertainty state for position and momentum
(ΔxΔp = /2). In a squeezed state, one of these two variances is reduced while
maintaining the minimum uncertainty state [4]. We define new operators (b̂ and
b̂† ) using Bogoliubov transformation to the annihilation and creation operators (â
and â † ) as

b̂ = μâ + ν â † = (cosh γ )â + (sinh γ )â †

b̂† = μâ † + ν â = (cosh γ )â † + (sinh γ )â, (3.78)
40 3 Harmonic Oscillator and Coherent and Squeezed States

where μ and ν are hyperbolic functions of a constant γ : μ = cosh γ , ν = sinh γ .

Because of μ2 − ν 2 = 1, the commutation between b̂ and b̂† is given by

[b̂, b̂† ] = μ2 − ν 2 = 1. (3.79)

Using the new operators, x̂ and p̂ are expressed as

  
  †  −γ 
x̂ = â + â = e b̂† + b̂
2mω 2mω
  
mω  †  mω †
p̂ = i â − â = ieγ b̂ − b̂ . (3.80)
2 2

Using eγ defined by

γ mω
e ≡ , (3.81)
mω

x̂ and p̂ are represented by

 

x̂ = b̂† + b̂
2m ω
 
m ω
p̂ = i b̂† − b̂ . (3.82)
2

Therefore, b̂ and b̂† are considered to be annihilation and creation operators for the
harmonic oscillator ( Ĥb = ω (b̂† b̂ + 1/2)) defined by the mass m and the angular
frequency of ω [5]. A sharpened potential for the harmonic oscillator is schematically
shown in Fig. 3.5. We set the ground state (vacuum state) of this Hamiltonian as |0b ,
which satisfies b̂|0b = 0. At this state, Δx and Δp are obtained as
 
  −γ
Δx = = e
2m ω 2mω
 
m ω mω γ
Δp = = e . (3.83)
2 2

This state is the squeezed vacuum state, which is also a minimum uncertainty state
(ΔxΔp = /2) and Δx is reduced compared to that of the ground state in the original
harmonic oscillator with (m, ω) for a positive γ value.
|0b is expanded with number states of Ĥ :
∞

|0b = Cn |n. (3.84)
n=0
3.4 Squeezed States 41

Fig. 3.5 Schematic of the V(x)

squeezed vacuum state. The
dotted curve shows a
sharpened harmonic
potential. The dashed
Gaussian is the vacuum state

Inserting this to b̂|0b = 0 gives

∞
   ∞
 √ √ 
Cn μâ + ν â † |n = Cn μ n|n − 1 + ν n + 1|n + 1
n=0 n=0
∞
 √ √ 
= Cn+1 μ n + 1 + Cn−1 ν n |n = 0. (3.85)
n=0

Then, the coefficient of |n should be zero and we get the recursion relation of

ν n
Cn+1 =− Cn−1 . (3.86)
μ n+1

The |0b contains only even or odd number states. We choose the even solution
because the ground state |0 is included. When we start from C0 , C2m is obtained by

−ν 1
C2 = C0 ,
μ 2
 2

−ν 3 −ν 3·1
C4 = C2 = C0
μ 4 μ 4·2
 3

−ν 5 −ν 5·3·1
C6 = C4 = C0
μ 6 μ 6·4·2
..
.
 m

−ν 2m − 1 −ν (2m − 1)!!
C2m = C2(m−1) = C0
μ 2m μ 2m!!

(2m − 1)!!
= (−1)m (tanh γ )2m C0 , (3.87)
(2m)!!
42 3 Harmonic Oscillator and Coherent and Squeezed States

where (2m)!! is the double factorial function ((2m)!! = 2m · (2m − 1) · (2m −

2) · · · 0) and m is an integer. C0 is obtained by normalizing the |0b as
∞

||0b |2 = |C2m |2 = 1, (3.88)
m=0

which leads to
 ∞

 (tanh γ )2m (2m − 1)!!
|C0 | 2
1+ = 1. (3.89)
m=1
(2m)!!

If we use the Taylor expansion of the function

∞
 (2n − 1)!!
f (z) = (1 − z)− 2 = 1 +
1
zn , (3.90)
n=1
2n n!

we get

∞
(2m − 1)!!  − 1
1+ (tanh2 γ )m = 1 − tanh2 γ 2 , (3.91)
m=1
(2m)!!

where (2m)!! = 2m m!. The coefficient |C0 | is obtained by


 1 1
|C0 | = 1 − tanh2 γ 2 = √ . (3.92)
cosh γ

Finally, |0b is expressed by


∞
1 (2m − 1)!!
|0b = √ (−1)m (tanh γ )m |2m
cosh γ m=1 (2m)!!
∞ √
1  (2m)!
= √ (−1) m
(tanh γ )m |2m, (3.93)
cosh γ m=1 2m m!

where (2m − 1)!! = (2m)!/(2m m!). The population of the number state is

(2m)! (tanh γ )2m

P2m = |2m|0b |2 =
22m (2m!)2 cosh γ
P2m+1 = |2m + 1|0b | = 0.
2
(3.94)
3.4 Squeezed States 43

Fig. 3.6 Typical examples

(a) 0.25
of the population of number
states |m for the squeezed
vacuum state with γ = 2 (a) 0.20
and γ = 6 (b)

Population Pn
0.15

0.10

0.05

0.00
0 5 10 15
n
(b)
-3
4x10
Population Pn

0
0 5 10 15
n

The squeezed vacuum state |0b consists of only even number states. A typical
example of the population of the number states in the squeezed states is shown
in Fig. 3.6.
Next, we consider time evolution of the squeezed vacuum state. Since the time
evolution of the even number states are expressed by exp(−iω(2m + 1/2)t)|2m,
the time-dependent squeezed vacuum state |0(t)b is
∞ √
1  (2m)!
|0(t)b = √ (−1) m
(tanh γ e−i2ωt )m |2m, (3.95)
cosh γ m=1 2m m!

where the phase exp(−iωt/2) is omitted. This means that tanh γ should be replaced
by tanh γ e−i2ωt . Then, we need to set sinh γ → sinh γ e−i2ωt , which corresponds to
ν = e−i2ωt sinh γ . Here, we introduce a squeeze operator Ŝ(ξ ) as

1 ∗ 2 
Ŝ(ξ ) ≡ exp ξ â − ξ â †2 , (3.96)
2
44 3 Harmonic Oscillator and Coherent and Squeezed States

where ξ = γ eiθ , and set the squeezed vacuum state |ξ  as

|ξ  = Ŝ(ξ )|0. (3.97)

The Ŝ(ξ ) is a unitary operator and Ŝ † (ξ ) = Ŝ −1 (ξ ) = Ŝ(−ξ ). By using the Baker–

Hausdrof lemma

(iλ)2
eiλ Â B̂e−iλ Â = B̂ + iλ[ Â, B̂] + [ Â, [ Â, B̂]] + · · · , (3.98)
2!

we calculate Ŝ † (ξ )â Ŝ(ξ ). When we consider  and B̂ as

i 1 ∗ 2 
 = ξ â − ξ â †2
λ2
B̂ = â, (3.99)

we get

i
[ Â, B̂] = [(ξ ∗ â 2 − ξ â †2 ), â]
2λ
i i
= (−ξ )[â †2 , â] = (2ξ )â † . (3.100)
2λ 2λ

Inserting this result, we get

2
i
[ Â, [ Â, B̂]] = (2ξ )[(ξ ∗ â 2 − ξ â †2 ), â † ]
2λ
2 2
i i
= (2ξ )ξ ∗ [â 2 , â † ] = (2ξ )(2)ξ ∗ â. (3.101)
2λ 2λ

The operators â and â † appear alternately in commutators and we get

2
i (iλ)2 i
Ŝ † (ξ )â Ŝ(ξ ) = â + (iλ) (2ξ )â † + (2|ξ |)2 â + · · ·
2λ 2! 2λ
|ξ | 2
ξ |ξ | †
2
= â − ξ â † + â − â · · ·
2! 3!
= â cosh γ − â † eiθ sinh γ = âμ − â † ν, (3.102)

which is the same of the Bogoliubov transformation (3.78) by replacing γ by −γ .

This confirms that Ŝ(ξ ) is the operator which transfers the vacuum state |0 to the
squeezed vacuum state.
Next, we calculate expectation values of creation and annihilation operators for
the squeezed vacuum state in order to evaluate the expectation value and variance of
3.4 Squeezed States 45

position (x̂) and momentum ( p̂). At first we calculate the expectation values for the
creation and annihilation operators.

ξ |â|ξ  = 0| Ŝ † (ξ )â Ŝ(ξ )|0

= 0|â cosh γ − â † eiθ sinh γ |0 = 0, (3.103)
ξ |â † |ξ  = 0| Ŝ † (ξ )â † Ŝ(ξ )|0
= 0|â † cosh γ − âe−iθ sinh γ |0 = 0, (3.104)
ξ |â â|ξ  = 0| Ŝ (ξ )â Ŝ(ξ ) Ŝ (ξ )â Ŝ(ξ )|0
† † † †

= 0|(â † cosh γ − âe−iθ sinh γ )(â cosh γ − â † eiθ sinh γ )|0

= sinh2 γ , (3.105)
ξ |â â |ξ  = 0| Ŝ † (ξ )â Ŝ(ξ ) Ŝ † (ξ )â † Ŝ(ξ )|0
†

= 0|(âr cosh γ − â † eiθ sinh γ )(â † cosh γ − âe−iθ sinh γ )|0

= cosh2 γ , (3.106)
ξ |â â|ξ  = 0| Ŝ † (ξ )â Ŝ(ξ ) Ŝ † (ξ )â Ŝ(ξ )|0
= 0|(âr cosh γ − â † eiθ sinh γ )(âr cosh γ − â † eiθ sinh γ )|0
= −eiθ sinh γ cosh2 γ , (3.107)
ξ |â â |ξ  = 0| Ŝ (ξ )â Ŝ(ξ ) Ŝ (ξ )â Ŝ(ξ )|0
† † † † † †

= 0|(â † cosh γ − âe−iθ sinh γ )(â † cosh γ − âe−iθ sinh γ )|0

= −e−iθ sinh γ cosh2 γ . (3.108)

Using above equations, we calculate ξ |x̂|ξ , ξ | p̂|ξ , ξ |x̂ 2 |ξ , and ξ | p̂ 2 |ξ :



ξ |x̂|ξ  = ξ |(â + â † )|ξ  = 0, (3.109)
2mω

mω
ξ | p̂|ξ  = i ξ |(−â + â † )|ξ  = 0, (3.110)
2

ξ |x̂ 2 |ξ  = ξ |(â 2 + â â † + â † â + â †2 )|ξ 
2mω
  
= cosh2 γ + sinh2 γ − 2 cos θ cosh γ sinh γ , (3.111)
2mω
mω
ξ | p̂ |ξ  = −
2
ξ |(â 2 − â â † − â † â + â †2 )|ξ 
2
mω  
= ξ | cosh2 γ + sinh2 γ + 2 cos θ cosh γ sinh γ . (3.112)
2
Then, the expectation values of position and momentum are zero, which are the same
as the vacuum state. The variance of position (x̂) is the same value of the expectation
value of x̂ 2 . For θ = 0, we get
46 3 Harmonic Oscillator and Coherent and Squeezed States

Fig. 3.7 Schematic drawing (a)

of the error ellipse of the
squeezed vacuum state with
γ
√ = 0.6. x is in a unit of
/(2mω)
√ and y is in a unit
of mω/2. a θ = 0 and b
θ = π . The center circle
represents the error circle of
the vacuum state

(b)

  −2γ
(Δx)2 = (cosh γ − sinh γ )2 = e
2mω 2mω
mω mω 2γ
(Δp)2 = (cosh γ + sinh γ )2 = e , (3.113)
2 2

and squeezing exists in the x̂ quadrature for γ > 0. On the other hand, for θ =
π , squeezing exists in the p̂ quadrature. Figure 3.7 shows the error ellipse for the
squeezed states.
For an arbitrary angle θ , we use a rotational transformation to operators (x̂ and
p̂) and new operators (x̂2 and p̂2 ):

x̂2 cos θ2 sin θ2 x̂

= , (3.114)
p̂2 − sin θ2 cos θ2 p̂
3.4 Squeezed States 47

Fig. 3.8 Schematic drawing

of the error ellipse of the
squeezed vacuum state with
γ = 0.6 and
√ θ = π/4. x is in
a unit of /(2mω)
√ and y is
in a unit of mω/2. The
center circle represents the
error circle of the vacuum
state

which corresponds to7

 
x̂2 + i p̂2 = x̂ + i p̂ e−iθ/2 . (3.115)

Then, we get x̂2 = e−iθ/2 x̂ and p̂2 = e−iθ/2 p̂, and the creation and annihilation
operators for the new coordinates (â2 and â2† ) are defined by â2 = e−iθ/2 â and
â2† = e−iθ/2 â † .
The squeeze operator is expressed using â2 and â2† as

1  −iθ 2 
Ŝ(ξ ) = Ŝ(r eiθ ) = exp r e â − r eiθ â †2
2
1  −iθ/2 2 
= exp r (e â) − r (eiθ â † )2
2
1 
= exp r â22 − r â2†2 . (3.116)
2

This means that the squeeze operator acts with θ = 0 to x̂2 and p̂2 . Then, the variance
is obtained as
 −2γ
(Δx2 )2 = e
2mω
mω 2γ
(Δp2 )2 = e , (3.117)
2

7 This is because (x̂ + i p̂) exp(−iθ/2) = (x̂ + i p̂)(cos(θ/2) − i sin(θ/2)) = (cos(θ/2)x̂ +

sin(θ/2) p̂) + i(− sin(θ/2)x̂ + cos(θ/2) p̂).
48 3 Harmonic Oscillator and Coherent and Squeezed States

Fig. 3.9 Time dependence

of variance ((Δx)2 ) in the
vacuum squeezed state with
γ = 0.6

2
Δx)
π/ω)

and the squeeze occurs in the x2 direction. Figure 3.8 shows the error ellipse at
θ = π/4.
Time evolution of variance is calculated using (3.113) and replacing θ by 2ωt:

  
(Δx)2 = ξ |x̂ 2 |ξ  = cosh2 γ + sinh2 γ − 2 cos(2ωt) cosh γ sinh γ
2mω
mω  
(Δp) = ξ | p̂ |ξ  =
2 2
cosh2 γ + sinh2 γ + 2 cos(2ωt) cosh γ sinh γ .
2
(3.118)

Then, the variance in the squeezed vacuum state oscillates with the angular frequency
of 2ω as shown in Fig. 3.9, though the variance is constant for the coherent state.

3.4.2 More General Squeezed States

More general squeezed states are obtained by

|α, ξ  = D̂(α)|ξ  = D̂(α) Ŝ(ξ )|0. (3.119)

The expectation value of the annihilation operator for the state is calculated as

α, ξ |â|α, ξ  = ξ | D̂ † (α)â D̂(α)|ξ  = ξ |(â + α)|ξ 

= α, (3.120)

where we used (3.72) and (3.103). A similar calculation gives

α, ξ |â † |α, ξ  = α ∗

α, ξ |â 2 |α, ξ  = α 2 − eiθ sinh γ cosh2 γ = α 2 + ξ |â 2 |ξ 
α, ξ |â †2 |α, ξ  = α ∗2 − e−iθ sinh γ cosh2 γ = α ∗2 + ξ |â †2 |ξ 
3.4 Squeezed States 49

α, ξ |â † â|α, ξ  = |α|2 + sinh2 = |α|2 + ξ |â † â|ξ 

α, ξ |â â † |α, ξ  = |α|2 + cosh2 γ = |α|2 + ξ |â â † |ξ . (3.121)

Using these equations, we get the expectation value of the position and the momen-
tum as


x̂ = α, ξ |x̂|α, ξ  = α, ξ |(â + â † )|α, ξ 
2mω


= (α + α ∗ )
2mω

mω
 p̂ = α, ξ | p̂|α, ξ  = −i α, ξ |(â − â † )|α, ξ 
2

mω
= −i (α − α ∗ ), (3.122)
2

which are the same as those of the coherent states (3.67). Similarly, we get


x̂ 2  = α, ξ |(â â + â † â + â â † + â † â † )|α, ξ 
2mω

= ξ |x̂ 2 |ξ  + (α 2 + α ∗2 + 2|α|2 )
2mω
mω
 p̂ 2  = − α, ξ |(â â + â † â − â â † − â † â † )|α, ξ 
2
mω 2
= ξ | p̂ 2 |ξ  − (α + α ∗2 − 2|α|2 ). (3.123)
2
The variances are obtained as

(Δx)2 = x̂ 2  − (x̂)2 = ξ |x̂ 2 |ξ 

(Δp)2 =  p̂ 2  − ( p̂)2 = ξ | p̂ 2 |ξ , (3.124)

and the same as those of the squeezed vacuum state (3.113). Therefore, in the general
squeezed state |α, ξ , the mean value of the position is the same as that of the coherent
state |α and the variance is the same as that of the squeezed vacuum state |ξ . The
time evolution of the position x̂ and the variance (Δx)2 oscillates with the frequency
of ω and 2ω, respectively.

3.5 Summary

The one-dimensional harmonic oscillator was investigated quantum mechanically.

The Hamiltonian of the harmonic oscillator was expressed using the creation â and
annihilation â † operators. The energy eigenstate is expressed using the number state
50 3 Harmonic Oscillator and Coherent and Squeezed States

|n. The coherent state, which is a minimum uncertainty state, is introduced. The
squeezed state, in which the variance of the position and the momentum is reduced
while maintaining the minimum uncertainty state, is also introduced. In the coherent
state, the mean value of the position oscillates with the angular frequency ω with
keeping its variance constant. In the vacuum squeezed state, the mean value of the
position is constant and its variance oscillates with 2ω [6].

References

1. Sakurai, J.J.: Modern Quantum Mechanics, Revised edn. Addison-Wesley Publishing Company
Inc., Reading (1994)
2. Takahashi, Y.: Ba no Ryoshiron I (in Japanese: Quantum Field Theory I). Baifukan, Tokyo
(1974)
3. Schrödinger, E.: Der steige Ubergang von der Mikro- zur Makromechanik. Die Nautureis-
senschaften 14, 664–666 (1926)
4. Gerry, C.C., Knight, P.L.: Introductory Quantum Optics. Cambridge University Press, New York
(2008)
5. Zwiebach, B.: Quantum dynamics (2013). https://ocw.mit.edu/courses/physics/8-05-quantum-
physics-ii-fall-2013/lecture-notes/MIT8_05F13_Chap_06.pdf
6. I wrote this chapter referencing Refs. [1–5] and following books: Yoshio Kuramoto and Junichi
Ezawa, Ryoushi Rikigaku (in Japanese, Quantum Mechanics), Asakura Shyoten (2008); Keiji
Igi and Hikari Kawai, Kiso Ryousi Rikigaku (in Japanese, Fundamental Quantum Mechanics),
Kodansya (2007); Masahito Ueda Gendai Ryoushi Buturigaku (in Japanese, Modern Quantum
Physics), Baifukan (2004); Akira Shimizu, Shin-han Rryoushi Ron no Kiso (in Japanese, New
edition Fundamental of Quantum Physics), Science Shya (2004); Kyo Inoue, Kogaku Kei no
Tameno Ryoushi Kogaku (in Japanese, Quantum Optics for Engineer), Morikita Shyoten (2015);
Masahiro Matsuoka, Ryoushi Kogaku (in Japanese, Quantum Optics), Shokabou (2000)
Chapter 4
Lattice Vibration and Phonon

A phonon is a quantum description of lattice vibrations in solids. This chapter sum-

maries basics of lattice vibration and phonons using a linear atomic chain. At first,
we calculate dynamics of atomic motions with classical mechanics and introduce a
plane wave expansion to express collective atomic motions. A phonon is introduced
by using field quantization. Optical and acoustic phonons are explained by using a
linear diatomic chain.

4.1 Linear Atomic Chain

We consider a simple linear atomic chain as a model of one-dimensional lattice

vibration. At first we treat it in classical mechanics and then in quantum mechanics.

4.1.1 Classical Treatment

Suppose atoms with mass of m stay in line with spacing a.1 The displacement of the
lth atom from its equilibrium position is set to be q l as shown in Fig. 4.1. The spring
constant between atoms is f . The equation of motion of this atom is

m q̈ l = f (q l+1 + q l−1 − 2q l ). (4.1)

We adapt a periodic boundary condition (q l = q l+N ) for a total of N atoms. Using a

plane wave expansion, the displacement is expressed by

1 Here,we referred to the method described in [1] for classical and quantum mechanics of a linear
atomic chain.
© Springer Nature Switzerland AG 2019 51
K. Nakamura, Quantum Phononics, Springer Tracts in Modern Physics 282,
https://doi.org/10.1007/978-3-030-11924-9_4
52 4 Lattice Vibration and Phonon

Fig. 4.1 Schematic of the linear atomic chain. The unit cell with the lattice constant a contains one
atom with mass of m. The force constant is f , and q l is atomic displacement in the lth unit cell

1
q l (t) = √ eikla Q k (t), (4.2)
N

where k is the wave vector which satisfies

2nπ
k= . (4.3)
Na
√
n is an integer in a range of (−N /2, N /2). 1/ N is the normalization factor, which
satisfies

N    
1 ikla ∗ 1 ik  la
√ e √ e = δk,k  . (4.4)
l=1
N N

Using this plane wave expansion, we get

 
Q̈ k (t) = D eika + e−ika − 2 Q k (t), (4.5)

where D = f /m. This is the equation of simple harmonic oscillation for Q k , and
then we express Q k as

Q k (t) = e−iωk t Ak , (4.6)

and get the equation

 
− ωk2 e−iωk t Ak = D eika + e−ika − 2 e−iωk t Ak . (4.7)

Then we get a dispersion relation between the frequency (ω) and the wave vector (k)
as
√
ωk = 2 D| sin(ka/2)|. (4.8)

The dispersion curve is shown in Fig. 4.2. The frequency ωk is equal to ω−k .
4.1 Linear Atomic Chain 53

Fig. 4.2 Phonon dispersion 2.5

Frequency (in unit of sqrt(D))

curve of the linear atomic
chain 2.0

1.5

1.0

0.5

0.0
-1.0 -0.5 0.0 0.5 1.0
Wave number (π/a)

The displacement of the lth atom is expressed as

 1 1

q l (t) = √ eikla e−iωk t Ak + √ e−ikla eiωk t A∗k
k
N N
  
1 1
= √ eikla Q k (t) + √ e−ikla Q ∗k (t) . (4.9)
k
N N

The Hamiltonian H of the lattice vibration is obtained by

 pl (t)2  2
f  l
H= + q (t) − q l+1 (t) , (4.10)
l
2m 2 l

where pl (t) is the momentum

 1 
pl (t) = m q̇ l (t) = m √ eikla Q̇ k (t) + c.c.
k
N
  
1
= −iωk m √ eikla Q k (t) + c.c. . (4.11)
k
N

Here c.c. means the complex conjugate. The kinetic energy T is obtained by
 
pl (t)2 1   1 ikla
T = = −iωk m √ e Q k (t) + c.c.
2m 2m l k
N
 
1 ik  la
× −iωk  m √ e Q k  (t) + c.c. . (4.12)
k
N
54 4 Lattice Vibration and Phonon

Using the orthogonal condition

1  ikla+ik  la
e = δk,−k , (4.13)
N l

we rewrite the kinetic energy as


m
T = − ωk ω−k Q k (t)Q −k (t) − ωk ω−k Q ∗k (t)Q ∗−k (t)
2 k

∗ 2 ∗
+ωk Q k (t)Q k (t) + ωk Q k (t)Q k (t)
2


m 2
= ω − Q k (t)Q −k (t) − Q ∗k (t)Q ∗−k (t)
2 k k

∗ ∗
+Q k (t)Q k (t) + Q k (t)Q k (t) . (4.14)

The potential energy V is obtained by

 
f 
V = (q l (t))2 + (q l+1 (t))2 − 2q l (t)q l+1 (t)
2 l
 
f 
= 2(q l (t))2 − q l−1 (t)q l (t) − q l+1 (t)q l (t) , (4.15)
2 l

by using different pairs of displacements. Each term of the right-hand side in the
equation is expressed as V1 , V2 , and V3 . V1 is obtained by

 1 1 −ikla ∗

V1 = f √ e Q k (t) + √ e
ikla
Q k (t)
l k
N N
 1 1 −ik  la ∗

ik  la
× √ e Q k  (t) + √ e Q k  (t)
k
N N

= f Q k (t)Q −k (t) + Q k (t)Q ∗k (t)
k

+Q ∗k (t)Q k (t) + Q ∗k (t)Q ∗−k (t) . (4.16)

A similar calculation gives V2 and V3 as


− f  −ila
V2 = e Q k (t)Q −k (t) + e−ika Q k (t)Q ∗k (t)
2 k
4.1 Linear Atomic Chain 55

+eika Q ∗k (t)Q k (t) + eika Q ∗k (t)Q ∗−k (t)

− f  ila
V3 = e Q k (t)Q −k (t) + eika Q k (t)Q ∗k (t)
2 k

+e−ika Q ∗k (t)Q k (t) + e−ika Q ∗k (t)Q ∗−k (t) . (4.17)

Using the dispersion relation

f −ika f mωk2
f − e − eika = , (4.18)
2 2 2
we get

mωk2  
V = Q k (t)Q −k (t) + Q k (t)Q ∗k (t) + Q ∗k (t)Q k (t) + Q ∗k (t)Q ∗−k (t) .
2
(4.19)

Then we get the Hamiltonian as


H =T +V = mωk2 (Q ∗k (t)Q k (t) + Q k (t)Q ∗k (t)). (4.20)
k

By introducing the dimensionless parameter bk , we can simplify the formula as

 ωk
H= (bk∗ bk + bk bk∗ ), (4.21)
k
2

where the relation between bk and Q k is


2mωk
bk = Q k (t). (4.22)


The lattice vibration is expressed as the summation of the harmonic oscillators with
their coordinates Q k (t), which are not the same as the atomic displacements q l (t).2

4.1.2 Quantum Treatment

For the canonical quantization, the coordinate ql and momentum pl are replaced by
linear operators q̂l and p̂l , which satisfy the commutation relations

√
2 At the Γ -point (k = 0), the atomic displacement in the lth unit cell is the same of Q k (t)/ N .
56 4 Lattice Vibration and Phonon
 l n
q̂ , p̂ = iδl,n ,
 l n 
q̂ , q̂ = p̂l , p̂ n = 0. (4.23)

In the Schrödinger picture, a dynamical variable is expressed by a time-independent

operator, where we omit time dependence in variables and operators. The amplitudes
of lattice vibration Q k and Q ∗k are also replaced by Q̂ k and Q̂ †k and the q̂ l are expressed
by

 1 1 −ikla †

q̂ =
l
√ e Q̂ k + √ e
ikla
Q̂ k
k
N N
 1
= √ eikla Q̂ k + Q̂ †−k , (4.24)
k
N

because the summations of k and −k are the same for the periodic boundary
conditions. The Fourier transformation of q̂ l is expressed by

1  −ikla l
N
√ e q̂ = Q̂ k + Q̂ †−k . (4.25)
N l=1

Here we introduce a new operator α̂k as

α̂k ≡ Q̂ k + Q̂ †−k . (4.26)

Applying the same procedure to the momentum operator p̂l gives

 1 ikla 1 −ikla †

p̂ =
l
−iωk m √ e Q̂ k + iωk m √ e Q̂ k ,
k
N N
m  ikla
= √ e −iωk Q̂ k + iω−k Q̂ †−k . (4.27)
N k

Its Fourier transform is

1  −ikla l
N
√ e p̂ = −iωk m Q̂ k + iω−k m Q̂ †−k
N l=1
= β̂k mωk , (4.28)

where we used ωk = ω−k and introduced a new operator β̂

β̂k ≡ −i Q̂ k + i Q̂ †−k . (4.29)

4.1 Linear Atomic Chain 57

Using (4.26) and (4.29), we get

1
Q̂ k = α̂k + i β̂k ,
2
1
Q̂ †−k = α̂k − i β̂k . (4.30)
2

The commutation relation between Q̂ k and Q̂ †k  at the same time is calculated by

[ Q̂ k , Q̂ †k  ] = Q̂ k Q̂ †k  − Q̂ †k  Q̂ k
 
1
= ˆ
(α̂k + i βk )(α̂−k − i β̂−k ) − (α̂−k − i β̂−k )(α̂k + i βk )
    ˆ
4
 
1
= [α̂k , α̂−k ] + [β̂k , β̂−k ] + i[β̂k , α̂−k ] − i[α̂k , β̂−k ] . (4.31)
   
4

The operators α̂k and α̂k† and β̂k and β̂k† are commutative,  because α̂k and β̂k include

only q̂ l or p̂l , respectively: α̂k , α̂−k  = 0, β̂k , β̂−k  = 0. Then we get
 
1
[ Q̂ k , Q̂ †k  ] = i[β̂−k  , α̂k ] + i[β̂k , α̂−k  ] . (4.32)
4

We rewrite this commutation relation using q̂ l and p̂l and get


1  ik  l  a  −ikla l  l
N N
i
[ Q̂ k , Q̂ †k  ] = e e [ p̂ , q̂ ]
4N m ωk  l  =1 l=1

1  −ikl  a  ik  la l  l
N N
+ e e [ p̂ , q̂ ] . (4.33)
ωk l  =1 l=1


Using the commutation relation [ p̂l , q̂ l ] = −iδl,l  we get
 N   
  1 1 
ia(k  −k)
[ Q̂ k , Q̂ †k  ] = + e = δk,k  . (4.34)
4N m l=1
ωk  ωk  2mωk

We introduce a dimensionless operator b̂k as


2mωk
b̂k ≡ Q̂ k , (4.35)


where this operator satisfies the commutation relation

[b̂k , b̂k† ] = δk,k  . (4.36)

58 4 Lattice Vibration and Phonon

Finally, the Hamiltonian is obtained by

  
† 1
Ĥ = ωk b̂k b̂k + . (4.37)
k
2

This Hamiltonian has the same form as the harmonic oscillator, and the operators
b̂k† and b̂k are creation and annihilation operators, respectively. This quantum of the
lattice vibration is called a phonon.3 The phonon is a boson like the photon. In the
stationary condition, the lattice wave with the wave number k is explained to be
occupied by n k phonons.

4.2 Linear Diatomic Chain

4.2.1 Classical Treatment

In the linear atomic chain, there is only one phonon mode (acoustic phonon). Next,
we consider a linear diatomic chain, which includes two atoms (atom 1 and atom 2)
in a unit cell.4 The atom 1 and atom 2 have masses of m 1 and m 2 , respectively. a is
a length of the unit cell. In the linear diatomic chain, another phonon mode (optical
phonon) exists in addition to the acoustic phonon. The atomic displacement of each
atom in the lth unit cell is expressed by qκl for atom κ, κ = 1, 2. By adapting the
periodic boundary condition for N unit cells, qκl+N = qκl (Fig. 4.3).
The kinetic energy T and the potential energy V are obtained by

 ( p l )2
κ
T =
l,κ
2m κ
1  2  2
V = f q2l − q1l + g q1l+1 − q2l , (4.38)
2 l

where pκl is a momentum of the atom κ and defined by

d l
pκl = m κ q̇κl = m κ q . (4.39)
dt κ

3A phonon is a collective oscillation, and is sometimes referred to as a quasi particle. The quasi
particle is a fictitious body consisting of the original real individual particle pulse train of disturbed
neighbors [2].
4 We referred to the method described in [3] for classical and quantum mechanics of the linear

diatomic chain.
4.2 Linear Diatomic Chain 59

Fig. 4.3 Schematic of the linear diatomic chain. The unit cell contains two atoms (atom 1 and
atom 2 represented by black and white balls) with masses of m 1 and m 2 , respectively. The force
constants between two atom 1 and atom 2 are f and g. x(l) represents the position of the lth unit
cell. x1 and x2 represent the equilibrium position of the atom 1 and atom 2 from the edge of the
unit cell, respectively

For the atom 1 in the lth unit cell, the equation of motion is

d2 l dV
m1 2
q1 = − l
dt dq1
1 d  2  2
=− l
f q2l − q1l + g q1l+1 − q2l
2 dq
 l 1 l  
= f q2 − q1 + g q2l−1 − q1l , (4.40)

where −dV /dq1l is the force acting on the atom 1. The equation for atom 2 is calcu-
lated in a similar way:

d2 l dV    
m2 2
q1 = − l = f q1l − q2l + g q1l+1 − q2l . (4.41)
dt dq2

Using the plane wave expansion, displacements are expressed by

1 1
A0 (k)aκ (k)eikxκ −iω(k)t + c.c.,
l
qκl = √ (4.42)
2 N mκ

where k is the wave number, ω(k) is the vibration frequency, and xκl is the equilibrium
position of the atom κ in the lth unit cell. Inserting (4.42) into (4.40), we get
 
√ l 1 l 1 l
− (ω(k))2 m 1 a1 (k)eikx1 = f √ a2 (k)eikx2 − √ a1 (k)eikx1
m2 m1
 
1 l 1 l
+ g √ a2 (k)eikx2 − √ a1 (k)eikx1 , (4.43)
m2 m1

where we neglected the complex conjugate term. The equilibrium position of each
atom has relationships as follows: x1l = x(l) + x1, x2l = x(l) + x2, and x2l−1 =
60 4 Lattice Vibration and Phonon

x(l) + x2 − a. Using these relations, removing the factor eikx(l) gives

 
2√ 1 1
− (ω(k)) m 1 a1 (k)e ikx1
= f √ a2 (k)e ikx2
− √ a1 (k)e ikx1
m2 m1
 
1 −ika ikx2 1
+ g √ a2 (k)e e − √ a1 (k)e ikx1
. (4.44)
m2 m1

Rearranging the equation gives

f +g f + ge−ika ik(x2−x1)
(ω(k))2 a1 (k) = a1 (k) − √ e a2 (k). (4.45)
m1 m1m2

A similar calculation for a2 (k) gives

f +g f + geika −ik(x2−x1)
(ω(k))2 a2 (k) = a2 (k) − √ e a1 (k). (4.46)
m2 m1m2

These equations are expressed by using a matrix

  
D11 − ω(k)2 D12 a1 (k)
= 0, (4.47)
D21 D22 − ω(k)2 a2 (k)

where
f +g
D11 =
m1
f + ge−ika ik(x2−x1)
D12 = √ e
m1m2
f +g
D22 =
m2
f + geika −ik(x2−x1)
D21 = √ e . (4.48)
m1m2

a1 (k) and a2 (k) have nonzero solutions when ω(k) satisfies

 
 D11 − ω(k)2 D12 
  = 0, (4.49)
 D21 D22 − ω(k)2 

then

(D11 − ω(k)2 )(D22 − ω(k)2 ) − D12 D21 = 0,

ω(k)4 − (D11 + D22 )ω(k)2 + D11 D22 − D12 D21 = 0. (4.50)
4.2 Linear Diatomic Chain 61

Here we calculate
f +g f +g m1 + m2
D11 + D22 = + = ( f + g)
m1 m2 m1m2
f +g
= ,
μ
 
2 fg 4 fg ka
D11 D22 − D12 D21 = (1 − cos(ka)) = sin2 , (4.51)
m1m2 m1m2 2

where μ is a reduced mass μ ≡ (m 1 m 2 )/(m 1 + m 2 ). Then we get the quadratic

equation of ω2 (k) as
 
 2 2 f +g 2 4 fg ka
ω (k) − ω (k) + sin2 . (4.52)
μ m1m2 2

The solution of ω2 (k) is

   
f +g 16μ2 fg ka
ω2 (k) = 1± 1− sin2 . (4.53)
2μ m 1 m 2 ( f + g)2 2

Since frequencies are positive, we get

  

 f +g  
 16μ2 fg 2 ka
ω j (k) = 1± 1− sin , (4.54)
2μ m 1 m 2 ( f + g)2 2

where j = 1, 2 corresponding to + term and − term, respectively. The correspond-

ing eigenvector is defined by using aκ, j (k). The frequency ω j is called a normal
frequency. The lattice vibration with the normal frequency is also called a normal
mode. In the normal mode, all atoms oscillate with the same frequency ω j in phase.
In general, the lattice vibration consists of many normal mode oscillations, and then
the oscillation of each atom is not in phase.

4.2.1.1 Example of Linear Diatomic Chain

Here, we consider a special case for the linear diatomic chain: f = g for the force
constant, m 2 > m 1 for the atomic mass and x1 = 0, x2 = a/2 for the equilibrium
atomic position. From (4.54), the eigenfrequency is
  

 3f  
 8 2 ka
ω j (k) = 1 ± 1 − sin . (4.55)
2m 1 9 2
62 4 Lattice Vibration and Phonon

Fig. 4.4 Phonon dispersion 2.0

Frequency (in unit of sqrt(f/m1))

curves of the diatomic linear Optical
chain with the force constant
( f = g), the atomic mass 1.5
(m 2 = 2m 1 ), and lattice
constant of a in the first
Brillouin zone. The higher 1.0
frequency mode is an optical
mode and the lower one is an
acoustic mode 0.5
Acoustic

0.0
-1.0 -0.5 0.0 0.5 1.0
Wave number (π /a)

The frequency is plotted as a function of the wave number in Fig. 4.4. There is
a gap, in which the frequency does not exist, between the high- and low-frequency
branches.
At ka  0 condition (long-wavelength
√ approximation), the eigenfrequencies were
obtained to be ω1 = 0 and ω2 = 3 f /m 1 , where we set the low and high frequencies
ω1 and ω2 , respectively. The atomic displacements are calculated by using (4.42) as

q1l
=1 (4.56)
q2l

for the j = 1 branch and

q1l m2
=− = −2 (4.57)
q2l m1

for the j = 2 branch. All atoms move in the same direction with the same displace-
ment, and the whole chain moves in the j = 1 branch, which is called an acoustic
mode. In the j = 2 branch, two neighboring atoms move in opposite directions to
each other and the center of mass does not move. This oscillation is called an optical
mode, because it can induce electronic polarization and interacts with the electronic
√ light for ions. At√around the ka = π condition, the eigen frequencies are
field of
ω1 = f /m 1 and ω2 = 2 f /m 1 for the acoustic and optical modes, respectively.
The optical phonons with a wave number of k ∼ 0 are quite important because they
are excited by light, for example, in Raman spectroscopy (Fig. 4.5).

4.2.1.2 Normal Coordinates and Hamiltonian

The displacement of atom κ with the wavenumber k is expressed by using normal

coordinates Q j (k) as
4.2 Linear Diatomic Chain 63

Fig. 4.5 Directions of atomic displacement for the acoustic mode (a) and the optical mode (b) at
ka  0

⎛ ⎞
1  
qκl = √ ⎝ l
aκ, j (k)eikxκ Q j (k) + aκ, j (−k)e−ikxκ Q j (−k)⎠ , (4.58)
l

N m κ j,k≥0 j,k>0

where
1 
Q j (k) = A0, j (k)e−iω j (k)t + A0, j (k)eiω j (−k)t
2
1 
= A0, j (k)e−iω j (k)t + A∗0, j (−k)eiω j (k)t . (4.59)
2
The most important merit to use normal coordinates is that the Hamiltonian is diag-
onalized:
1  
H =T +V = P j (k)P j∗ (k) + Q j (k)Q ∗j (k)ω2j . (4.60)
2 k, j

Derivation of this relation is shown in Appendix A. Equations of motion in a Hamil-

tonian form are
∂H
Q̇ j (k) = = P j (k)
∂ P j∗ (k)
−∂ H
Ṗ j (k) = , (4.61)
∂ Q ∗j (k)

then we get

Q̈ j (k) = Ṗ j (k) = −ω2j Q j (k), (4.62)

which is an equation of motion for uncoupled 2N harmonic oscillators. In a normal

coordinate which belongs to an eigenfrequency defined by the mode and the wave
number, all atoms oscillate with the same phase.
64 4 Lattice Vibration and Phonon

4.2.2 Quantum Treatment

Using the same method as the linear atomic chain, we replace Q j (k) and P j (k) by
Q̂ j (k) and P̂ j (k), which satisfy the commutation relation:
 
Q̂ ∗j (k), P̂ j  (k  ) = Q̂ j (k), P̂ j∗ (k  ) = iΔ(k − k  )δ j, j 
 
Q̂ ∗j (k), Q̂ j (k  ) = P̂ j∗ (k), P̂ j (k  ) = 0, (4.63)

where Δ(k − k  ) = 1 for k = k  ± 2π n/a and Δ(k − k  ) = 0 for otherwise. In addi-

tion, we introduce creation and annihilation operators (b̂†j (k) and b̂ j (k)) as
 
ω j (k) 1
b̂ j (k) = Q̂ j (k) + i P̂ j (k)
2 2ω j (k)
 
ω j (k) 1
b̂†j (k) = Q̂ j (k) − i P̂ j (k). (4.64)
2 2ω j (k)

The Hamiltonian is expressed by

1 ∗
H = P̂ j (k) P̂ j (k) + ω2j Q̂ ∗j (k) Q̂ j (k)
2 k, j
 
1 1
= ω j b̂†j (k)b̂ j (k) + . (4.65)
2 k, j 2

Therefore, the system is calculated as uncoupled 2N harmonic oscillators.

4.3 Summary

A phonon is a quantum description of lattice vibrations in solids. At first, we calculate

dynamics of atomic motions with classical mechanics and introduce a plane wave
expansion to express collective atomic motions. A phonon is introduced by using field
quantization. The phonon in a normal mode is expressed by the Hamiltonian which
has the same form used for the harmonic oscillator using the creation and annihilation
operators for the normal mode. Optical and acoustic phonons are explained by using
a linear diatomic chain.
References 65

References

1. Haken, H.: Quantum Filed Theory of Solids. North-Holland Publishing Company, Amsterdam
(1976)
2. Bruesch, P.: Phonons: Theory and Experiments I. Springer Series in Solid-State Sciences 34.
Springer, NewYork (1982)
3. Mattuck, R.D.: A Guide to Feynman Diagrams in the Many-Body Problems, 2nd edn. Dover
Publications Inc., New York (1992)
Chapter 5
Coherent Phonons: Experiment

Coherent phonons are coherently excited by an ultrashort optical pulse and used to
study dynamics of optical phonons. In this chapter, we present a brief history of
experiments and phenomenological theories of coherent phonons. A typical exam-
ple of the experimental technique and results of coherent phonons using a pump
and probe protocol is explained. Several selected examples of coherent phonons in
semiconductors and semimetals are presented. In addition, the squeezed phonons,
which are essential nonclassical states, are described.

5.1 A Brief History

5.1.1 Experiments of Coherent Optical Phonons

When an ultrashort optical pulse with duration much shorter than an oscillation
period of optical phonons is irradiated on materials, the optical phonons are excited
impulsively and coherently. Oscillation of the induced optical phonons is detected via
a transient change in reflectivity or transmissivity. Such impulsively excited phonons
are called coherent phonons.1 Experimental studies of coherent optical phonons
started in the mid-1980s. Keith Nelson and his coworkers demonstrated femtosecond
time-resolved measurements of optical phonons in the organic molecular crystal
(α-perylene) [1]. They used 70 fs optical pulses with wavelength of 620 nm and
performed pump–probe experiments. The α-perylene crystal has two pairs of four
planar molecules in a unit cell. They observed coherent oscillations due to vibrational
(33 and 80 cm−1 ) and translational (104 cm−1 ) modes.

1 It should be mentioned that the coherent phonons are not necessarily a coherent state of the phonon.

© Springer Nature Switzerland AG 2019 67

K. Nakamura, Quantum Phononics, Springer Tracts in Modern Physics 282,
https://doi.org/10.1007/978-3-030-11924-9_5
68 5 Coherent Phonons: Experiment

Cho et al. [2] studied transient reflectivity of a semiconductor crystal (GaAs)

and found 8.8-THz coherent phonon oscillations due to the longitudinal optical
(LO) phonon in GaAs. Coherent phonons in semimetals (bismuth and antimony)
were observed through reflectivity modulation by Cheng et al. [3]. The observed
phonon mode was an isotropic optical (A1g ) mode with 2.9 and 4.5-THz frequencies
for the Bi and Sb samples, respectively. Chwalek et al. [4] demonstrated the time-
resolved observation of A1g -mode phonon in the semiconducting cuprate compound
YBa2 Cu3 O6+x (x < 0.4) using 100 fs pulses.
Coherent phonons have now been extensively studied in various materials, for
example, in semimetals [5–7], semiconductors [8–13], superconductors [14–17],
oxides [18–20], topological insulators [21–26], and carbon materials [27–30], using
pump–probe optical measurements.

5.1.2 Generation Mechanism of Optical Coherent Phonons

Generation mechanisms of the optical coherent phonons are often classified into
three mechanisms: the impulsive stimulated Raman scattering (ISRS) [31, 32] for
transparent conditions, the displacive excitation of coherent phonons (DECP) [33]
for opaque conditions, and the screening of the surface-charge field for a polar semi-
conductor such as gallium arsenide [2, 34]. The equation of motion for an optical
phonon is expressed in a normal coordinate Q using a classical harmonic oscillator
model:

d2 Q dQ
2
+ 2γ + ω02 Q = F(t), (5.1)
dt dt
where γ and ω0 are the vibrational damping constant and frequency, respectively,
F(t) is the driving force.
The ISRS mechanism was first proposed by Yan et al. [31]. When an ultrashort
optical pulse passes through a Raman-active material, phonons, which have vibra-
tional periods longer than the pulse width, are excited by the ISRS process, because
the Stokes frequency is contained within the bandwidth of the pulse. The driving
force is given by
 
1 ∂α
F(t) = N E 2, (5.2)
2 ∂Q

where N is the number density of oscillators, α is the differential polarizability tensor,

and E is the electric field of the pulse. The obtained oscillation of Q is a sine-like
oscillation (Q 0 sin(ω0 t)), which starts from the equilibrium position.
The DECP mechanism was proposed by Zeiger et al. [33] for coherent optical
phonons with A1 symmetry in semimetal materials and semiconductors. In this mech-
anism, the origin of the oscillation is a change in the quasi-equilibrium A1 nuclear
coordinate Q 0 (t) from the equilibrium position Q 0 = 0 before laser irradiation. This
5.1 A Brief History 69

shift is due to photoexcited electrons (n(t): electron density per unit volume) in the
excited state. They assumed the linear dependence of the shift on the electron density:
Q 0 (t) = κn(t). The driving force is expressed as

F(t) = ω0 2Q 0 (t) = ω02 κn(t)

dn(t)
= ρ E p g(t) − βn(t), (5.3)
dt

where E p and g(t) are the energy and a normalized pulse-shape function of the
optical pulse, respectively, ρ is a constant of proportionality for carrier generation,
and β is a rate constant for relaxation. The oscillation shows a cos(ω0 t) dependence
in DECP mechanism. The DECP mechanism is analogous to the mechanism using
a displaced potential for the electronic excited state, which is proposed for coherent
oscillations in large organic molecules.
For polar semiconductors, another generation mechanism due to a change in a
surface-charge field is proposed [2, 34]. The driving force of this mechanism is the
sudden depolarization of the crystal lattice due to the ultrafast change of intrinsic
surface-charge field in its depletion layer which is given by
 t
F(t) = J j (t  )dt  , (5.4)
∞

where J j (t) is a current associated with drift of photoexcited carriers in the surface-
charge field.
Kuznetsov and Stanton [35] proposed a microscopic theory in which the dynam-
ics of electrons and phonons are described by kinetic equations using quantum-
mechanical operators. They used the Hamiltonian:
 †

Ĥel = εαk ĉαk ĉαk + ωq b̂q† b̂q
k,α q
  
†
+ Mkq b̂q + b̂q† ĉαk ĉαk+q , (5.5)
α,k,q

where ĉ† , ĉ are the electron creation and annihilation operators in k space, respec-
tively, εαk is the energy dispersion in band α = {c, v} (conduction or valence band).
ωq , b̂† , and b̂ are the phonon angular frequency, the phonon creation, and annihilation
operators. The last term is an interaction term describing electron–phonon coupling,
and Mkq is the coupling constant related to a deformation potential. They defined
a coherent amplitude Dq ≡ b̂ + b̂†  of the qth phonon mode using the statistical
average b̂ and b̂†  and obtained the equation of motion using Dq :

∂2 
α α
Dq + ω 2
Dq = −2ωq Mkq n k,k+q , (5.6)
∂t 2 q
α,k
70 5 Coherent Phonons: Experiment

where n αk,k+q is the electronic density matrix. The derived equation is similar to the
phenomenological equation for classical oscillators.
Numerical simulations for generating coherent phonons have been studied using
the time-dependent density functional theory (TDDFT) by Yabana and his coworkers
[36, 37]. They used a classical electronic field for the optical pulse, the first principle
theory for electronic states using DFT, and classical dynamics for the atomic motion.
The ISRS process is confirmed to be a mechanism for generating the driving force
in a diamond sample.

5.2 Experiments

The coherent phonons are usually excited by irradiation of an ultrashort optical pulse
and detected by using transient optical transmission (or reflection) measurements
using a pump–probe technique. Recently, using ultrashort X-ray pulses, the coherent
phonon oscillation is also detected via a time-resolved X-ray diffraction technique.

5.2.1 Optical Measurement

The coherent optical phonons are excited and measured via pump–probe transmission
(or reflection) measurements using femtosecond laser pulses. The pump pulse with
duration shorter than a vibration period of a phonon induces the phonons with finite
timing. Induced phonons oscillate in phase and cause a modulation in macroscopic
electric susceptibility, which can be detected via transmissivity (or reflectivity) of
the probe pulse. Figure 5.1 shows a typical example of the pump–probe experiments.
Femtosecond optical pulses with a central wavelength of 800 nm are generated
by a Ti:sapphire oscillator, which is excited by 532 nm light from a continuous wave
laser. The femtosecond pulse is separated into two pulses (pump and probe pulses)
by a beam splitter. The pump and probe pulses pass through different optical paths,
which have their lengths controlled. Since the light travels with the constant light
speed (2.99 × 108 m/s), the arrival time at the sample position is different for the
pump and probe pulses. The difference of 299 nm in the optical paths corresponds
to 1 fs. In this setup, path length for the probe pulse is controlled by using a shaker
(Scan delay unit), which oscillates at 20 Hz. Intensity (signal intensity: Is ) of the
probe pulse transmitted through a sample is detected by a photodiode (PD2), which
does not need a high time resolution. The probe pulse intensity (reference intensity:
Ir ) is monitored by a photodiode (PD1). The differential signal (Is − Ir ) is introduced
to a current amplifier and detected by an oscilloscope. The pump and probe pulses
are focused on the sample by using a lens or a parabolic mirror.2 The output from the

2A parabolic mirror is often used to focus an ultrashort optical pulse shorter than approximately
30 fs in order to avoid strong chirp effects by a lens.
5.2 Experiments 71

Fig. 5.1 Schematic of the experimental setup for coherent phonon measurements using a pump and
probe technique. This is a setup for the transient transmission measurement. BS: beam splitter, CM:
chirp mirror, PD: photodiode, Amp: current amplifier, PC: personal computer. Solid and dashed
lines represent the optical path and the electric connection, respectively

Fig. 5.2 Transient

-6

5
Transmittance change (ΔT/T0) 10

transmittance from a single

crystal of diamond measured
by using a pump and probe
0
technique. Inset is an
enlarged signal in the delay
range between 500 and
-5
600 fs. The oscillation period
is 25.1 ± 0.03 fs
-10
600 650 700 750 800

-500 0 500 1000 1500 2000

Pump probe delay (fs)

laser oscillator is reflected by a pair of chirp mirrors in order to compensate for the
group velocity dispersion of optics and to become the shortest pulse at the sample
position.
Figure 5.2 shows a typical example of the femtosecond time-resolved transmit-
tance measurement of diamond using the pump and probe technique and sub-10 fs
laser pulses at room temperatures.3 There is a strong peak at zero delay due to the non-

3 The used sample is a single crystal of diamond with a (100) face. The sample was fabricated by
chemical vapor deposition and obtained from EPD corporation and has intermediate type between
Ib and IIa with a size of 5 mm square and 0.7 mm thickness [38]. The optical pulse is characterized
immediately behind the output port of the laser oscillator using a spectrometer and a frequency-
resolved autocorrelation (FRAC) technique. The spectrum showed a maximum-intensity wavelength
72 5 Coherent Phonons: Experiment

linear response of diamond by overlapping the pump and probe pulses in the sample.
After the strong peak, there is a modulation caused by the coherent optical phonons
in diamond. Inset in Fig. 5.2 is the enlarged signal in delay between 600 and 800 fs.
The oscillation period is 25.1 ± 0.03 fs (frequency of 39.9 ± 0.05 THz), which is
the same value of the optical phonon of diamond at Γ -point (k = 0) obtained by
Raman spectroscopy.4 The used laser pulse is near infrared, and its energy (1.5 eV)
is well below the band gap (7.3 eV) of diamond. Then, the generation mechanism in
this experiment is ISRS. The oscillation is well reproduced by a damped harmonic
oscillation
ΔT (t)
= C exp(−t/τ ) sin(ωt + θ ), (5.7)
T0

where C is an oscillation amplitude, τ is a coherent time, ω is an angular frequency,

and θ is an initial phase. The coherent time is obtained to be approximately 6 ps.

5.2.1.1 Semimetals

Coherent phonons in semimetals were first reported by Cheng et al. [3]. They mea-
sured femtosecond time-resolved reflection from bismuth and antimony samples
using 70 fs pulses of laser light at 1.98 eV (625.8 nm) and the pump and probe tech-
nique. Bismuth and antimony crystallize in the A7 structure, which is a trigonally
distorted cubic structure with two atoms per unit cell. There are two optical phonon
modes: isotropic A1g and anisotropic E g modes. The transient reflection intensity
showed coherent oscillations with frequencies of 2.9 and 4.5 THz, which correspond
to the A1g optical phonon frequencies at Γ -point (k = 0) for bismuth and antimony,
respectively. The anisotropic E g mode phonon oscillation was not observed in their
time-resolved measurement.
By using an electro-optical (EO) sampling technique [34], the E g mode phonon
oscillation was observed in bismuth [39]. In this technique, the polarization of the
probe pulse is set perpendicular to that of the pump pulse. The reflected probe pulse
is directed into a polarizing beamsplitter, and the parallel and perpendicular compo-
nents are detected by using balanced photodetectors. Thus, the difference between
two orthogonal components Rx − R y is detected. A schematic of the EO sampling
is shown in Fig. 5.3. Using this technique, the isotropic oscillation ( A1g ) signal is
suppressed and the weak anisotropic oscillation (E g ) signal can be observed [39].
In addition to the A1g coherent phonons, the E g coherent phonons oscillating with
oscillation period of 475 fs (2.1 THz) were observed. The phonon decay time was

of 792 nm with a bandwidth of approximately 200 nm, and the estimated pulsed width was 8.2 fs at
full width at half maximum from the FRAC measurement.
4 Coherent phonons in diamond were first reported by Ishioka et al. [28] by using sub-10 fs, 395 nm

pulses via transient reflection measurements.

5.2 Experiments 73

Fig. 5.3 Schematic of reflective electro-optic (EO) sampling technique. The reflected probe pulse
is directed to a polarizing beam splitter (PBS), and p- and ps-polarized components are separated
and detected by photodiodes (PD1 and PD2). PD1 and PD2 have reverse-bias voltages. An arrow
shows a polarization direction of pulses

obtained to be 3.71 and 2.21 ps for A1g and E g , respectively, by using curve fitting
with two damping oscillations.

5.2.1.2 Semiconductors

Transient pump–probe reflectivity measurements have been performed to observe

coherent oscillation of longitudinal optical (LO) phonons in GaAs by Cho et al.
[2]. The sample used was (100)-oriented intrinsic GaAs (i-GaAs). They used 50-fs
pulses at 2 eV and observed a periodic oscillation with an oscillation frequency of
8.8 ± 0.15 THz (oscillation period of ∼114 fs), which corresponds to the LO phonon
in i-GaAs. The dephasing time was obtained to be 2.0 ps at excitation density close
to 1017 cm−3 and 0.7 ps at density of 1018 cm−3 at room temperature. Coherent LO
phonons were also observed in n- and p-doped GaAs (N D = 2 × 1018 cm−3 ) using the
reflective EO sampling [34]. They measured cosine-like oscillation with a frequency
of 8.75 THz. The phase of the coherent LO-phonon oscillation in n-doped GaAs
was shifted by π from that in p-doped GaAs. The initial phase is determined by
the sign of the driving force, which is due to the screening of the surface-space-
charge field. In a polar semiconductor, the valence and conduction bands are bent
in a depletion layer near the surface due to pining of the Fermi level at the charged
surface state. Thus, a macroscopic surface-space-charge field is created perpendicular
to the sample surface. When electron–hole pairs or electronic polarization are created
by pumping in the depletion layer, the space-charge-field is screened and a sudden
change of constraining force causes driving force of the lattice vibration. The bands
bend upward and downward at the surface for n- and p-doped GaAs, respectively.
Then the driving force in the n- and p-doped GaAs has different directions and
the initial phase was shifted by π . In a complementary experiment, Pfeifer et al.
[34] applied an external electric field to the sample and found an enhancement of
the phonon amplitude. Typical examples of the pump-probe transient reflectivity
measurements of GaAs are shown in Fig. 5.4.
74 5 Coherent Phonons: Experiment

Fig. 5.4 Reflectivity change (a)

in the photoexcited intrinsic 4
GaAs (a), p-doped GaAs (b) 2
and n-doped GaAs (c) 0
measured using pump–probe
-2
reflective electro-optical

-6
Refelectivity change ( ΔR/R) 10
-4
sampling. The coherent
oscillation due to LO (b)
phonons is shown in the 4
intrinsic GaAs. In the 2
n-doped GaAs, there is 0
beating due to two coherent -2
oscillations (LO phonon and -4
LOPC). In the p-doped
GaAs, very small oscillation 4
(c)
of LOPC overlaps with that 2
of LO phonons 0
-2
-4
-1 0 1 2 3 4
Pump probe delay (ps)

In addition to LO phonons, the LO phonon–plasmon coupled (LOPC) oscillation

is excited in n-doped GaAs [40]. The√frequency of the plasmon, which is proportional
to square root of electron density ( Ne ), approaches to the LO-phonon frequency
at Ne ∼ 1018 cm−3 and the plasmon oscillation couples to LO-phonon oscillation.
Hu et al. [41] irradiated a 45-fs pulse at 1.55 eV on the n-doped GaAs (100) with
Ne = 1018 cm−3 sample and excited the total density (Ne = 4.2 ± 0.5 × 1018 cm−3 )
of photoexcited and intrinsic carriers. The coherent oscillations of the LO phonon
and LOPC are 8.8 and 7.7 THz, respectively, at room temperature. The dephasing
time of LO phonon (1.5 ps) is longer than that of LOPC (0.9 ps) and increases as
temperature decreases. The temperature dependence of the decay rate 1/τ was well
fitted by the Klemens channel
 
1 1
= Γ0 1 + 2 , (5.8)
τ exp(ω L A /(2k B T )) − 1

where Γ0 is a fitting parameter (fitted value is 0.26 ps−1 ), k B is the Boltzmann

constant, and T is the temperature. In the Klemens channel, an optical phonon decays
into two acoustic phonons with a frequency of ω L A = ω L O /2 and equal but opposite
wave vectors. On the other hand, the dephasing time of LOPC was less temperature
dependent.
Many experimental examples of the coherent phonons in bulk and low-dimensional
semiconductors are reviewed in the literature [9].
5.2 Experiments 75

5.2.2 X-Ray Diffraction Measurement

Coherent optical phonons have also been detected by using ultrafast time-resolved
X-ray diffraction, where coherent phonons are excited by a femtosecond optical pulse
and detected by using an ultrashort X-ray pulse [42–46]. The first clear experiments
were demonstrated by Sokolowski-Tinten et al. [42] on a bismuth film sample. They
performed femtosecond time-resolved X-ray diffraction using a laser pump and X-ray
probe technique with 120-fs laser pulses and laser-plasma X-rays (LPX).5 The X-ray
diffraction intensity of the (222) reflection showed oscillation just after the pump-
laser pulse irradiation with an oscillation period of 467 fs (frequency of 2.12 THz).
This oscillation is assigned to the A1g optical phonon. The observed frequency was
downshifted from that of the pristine sample (2.92 THz) because the sample was
highly excited by the laser irradiation. The diffraction intensity is dependent on the
square modulus of the structure factor F(h, k, l, t), where h, k, and l are Miller
indices and t is time. The time dependence of the structure factor is expressed by


N
 
F(h, k, l, t) = f i exp −iG j · r j + δ j , (5.9)
j=1

where f j , r j , and δ j are the atomic scattering factor, the crystallographic atomic
position, and the atomic-deviation vector of the jth atom in a unit cell. For coher-
ently excited phonons, the atomic-deviation vector is δ j = u j sin(ωt + θ ), where
u j and θ are the maximum deviation vector and initial phase. Thus, the diffraction
intensity oscillates according to the optical phonon oscillation. Unlike the optical
reflection or transmission measurements, the value of the atomic displacement is
directly determined using the X-ray diffraction experiment. Sokolowski-Tinten et al.
[42] reported that the maximum atomic displacement is approximately 5–8% of the
nearest-neighbor distance (0.31 nm) between the bismuth atoms in the A7 structure
at pump-laser fluence of 6 mJ/cm2 .

5.3 Squeezed Phonons

As discussed in Chap. 3, a squeezed state is an essential nonclassical state. In the

squeezed state, it is possible to reduce the fluctuations of one variable to a level
lower than that of the vacuum state at the expense of the fluctuations of its conjugate
variable. Because of this possibility to overcome the limitation of the uncertainty

5 When the metal target is irradiated by an intense ultrashort laser pulse, with intensity much larger
than 1016 W/cm2 , the rising edge of the pulse has enough energy to produce plasma. Electrons in the
plasma are accelerated more than keV by the main part of the pulse. The high-energy electrons collide
with target atom, excite inner shell electrons, and produce characteristic X-rays. For example, short-
pulsed K α and K β emissions are reported for a copper target for 42-fs laser-irradiation experiment
with a power density between 3 × 1016 and 2 × 1017 W/cm2 [47].
76 5 Coherent Phonons: Experiment

principle and to reduce fluctuations or quantum noise, squeezed states have attracted
much attention, especially in quantum optics. Squeezed states in condensed matters
including squeezed phonons are also studied and have promising applications such
as gravitation-wave detection and quantum communication.
The squeezed phonons were first reported using transient transmission experiment
with femtosecond laser pulses for a KTaO3 crystal by Garrett et al. [48]. The measured
transmittance coherently oscillated with a frequency close to twice the frequency of
the transverse acoustic (2TA) phonon near the zone boundary. These oscillations are
due to two phonon-squeezed states excited by the second-order Raman scattering.
They estimated that the squeeze factor is approximately 3 × 10−6 of an integrated
pulse intensity of I0 = 19 µJ/cm2 .
Generation of phonon-squeezed states by the second-order Raman scattering is
also theoretically proposed by Hu and Nori [49]. They considered both the sponta-
neous Raman and impulsive Raman processes. In the impulsive Raman process, they
used a δ function for the optical pulse and the Hamiltonian defined by

Ĥq = Ĥq − λq δ(t) Q̂ q Q̂ −q , (5.10)

where Ĥq is the Hamiltonian of the phonon system, and λq is the product of the
second-order polarizability tensor and electric fields of the optical pulses. By solving
the Schrödinger equation with separating the free oscillator terms and the two-phonon
creation and annihilation terms, the time-dependent wave function was obtained as

Ĥq t iλq Ĥq t

|ψq (t) = exp exp
i 2 ωq

 
× exp ζq∗ b̂q b̂−q − ζq b̂q† b̂−q
†
|ψ(0− ), (5.11)

where ζq = −iλq exp(−iλq /). Then, the second-order Raman process causes the
two-mode quadrature-squeezed operator on the vacuum state.
In the general squeezed states, the mean value and variance of position oscillate
with ω and 2ω, where ω is the phonon frequency, respectively. This feature has also
been reported in transient reflection measurements of Bi and GaAs using femtosec-
ond laser pulses with a pump–probe protocol [50]. They measured the amplitude
(m) of the oscillatory waveform in transient reflectivity. They repeated the measure-
ment 100 times and calculated the variance, σ 2 = (m − m)2 , at each delay. The
transient reflectivity in Bi shows 2.93-THz oscillation, corresponding to the A1g
phonon frequency (ω), in amplitude and 2ω frequency in variance. Similar features
are observed for the longitudinal optical phonons (frequency of 8.54 THz) in GaAs.
They suggested that the observed phase-dependent noise indicates the elliptical shape
for the uncertainty contour and the squeezing of phonons.
There are several reports on the squeezed phonons, and further discussions are
still continuing [51–53].
5.4 Summary 77

5.4 Summary

The coherent phonons are phonons excited by an ultrashort optical pulse and used
to study dynamics of optical phonons. A brief history of the experiments and phe-
nomenological theories of the coherent optical phonons is described. The experi-
mental setup and measurement technique for the coherent phonons in diamond are
presented as examples. Several selected examples of coherent phonons in semicon-
ductors and semimetals are presented. In addition, the squeezed phonons, which are
essential nonclassical states, are described.

References

1. De Silvestri, D., Fujimoto, J.G., Ippen, E.P., Gamble, E.B., Williams, Jr. L.R., Nelson, K.A.:
Femtosecond time-resolved measurements of optic phonon dephasing by impulsive stimulated
Raman scattering in α-perylene crystal from 20 to 300 K. Chem. Phys. Lett. 116, 146–152
(1985)
2. Cho, G.C., Kütt, W., Kurz, H.: Subpicosecond time-resolved coherent-phonon oscillation in
GaAs. Phys. Rev. Lett. 65, 764–766 (1990)
3. Cheng, T.K., Brorson, S.D., Kazeroonian, A.S., Moodera, J.S., Dresselhaus, G., Dresselhaus,
M.S., Ippen, E.P.: Impulsive excitation of coherent phonons observed in reflection in bismuth
and antimony. Appl. Phys. Lett. 57, 1004–1006 (1990)
4. Chwalek, J.M., Uher, C., Whitaker, J.F., Mourou, G.A., Agostinelli, J.A.: Subpicosecond time-
resolved studies of coherent phonon oscillations in thin-film YBa2 Cu3 O6+x (x<0.4). Appl.
Phys. Lett. 58, 980–982 (1991)
5. Misochko, O.V., Hase, M., Ishioka, K., Kitajima, M.: Observation of an amplitude collapse
and revival of chirped coherent phonons in bismuth. Phys. Rev. Lett. 92, 19740 (2004)
6. Katsuki, H., Delagnes, J.D.C., Hosaka, K., Ishioka, K., Chiba, H., Zijlestra, E.S., Garcia, M.E.,
Takahashi, H., Watanabe, K., Kitajima, M., Matsumoto, Y., Nakamura, K.G., Ohmori, K.: All-
optical control and visualization of ultrafast two-dimensional atomic motions in a single crystal
of bismuth. Nat. Commun. 4, 2801 (2013)
7. Cheng, Y.-H., Gao, F.Y., Teitelbaum, S.W., Nelson, K.A.: Coherent control of optical phonons
in bismuth. Phys. Rev. B 96, 134302 (2017)
8. Dekorsy, T., Chi, G.C., Kurz, H.: Coherent phonons in condensed media. Light. Scatt. Solids
VIII 76, 169–209 (2000). and references therein
9. Först, M., Dekorsy, T.: Coherent phonons in bulk and low-dimensional semiconductors. In: De
Silvestri, S., Cerullo, G., Lanzani, G. (eds.) Coherent Vibration Dynamics. CRC Press, New
York (2008). and references therein
10. Yee, K.J., Lee, K.G., Oh, E., Kim, D.S., Lim, Y.S.: Coherent optical phonons oscillations in
bulk GaN excited by far below band gap photons. Phys. Rev. Lett. 88, 105501 (2002)
11. Hase, M., Kitajima, M., Constantinecu, A.M., Petek, H.: The birth of a quasiparticle in silicon
observed in time-frequency space. Nature 426, 51–54 (2003)
12. Hase, M., Katsuragara, M., Constantinecu, A.M., Petek, H.: Frequency comb generation at
terahertz frequencies by coherent phonon excitation in silicon. Nat. Photonics 6, 243–247
(2012)
13. Yoshino, S., Oohata, G., Mizoguchi, K.: Dynamical Fano-like interference between Rabi oscil-
lations and phonons in a semiconductor microcavity system. Phys. Rev. Lett. 115, 157402
(2015)
14. Misochko, O.V., Georgiev, N., Dekorsy, T., Helm, M.: Two crossovers in the pseudogap regime
of YBa2 Cu3 O7−δ supreconductors observed by ultrafast spectroscopy. Phys. Rev. Lett. 89,
067002 (2002)
78 5 Coherent Phonons: Experiment

15. Mansart, B., Boschetto, D., Savoia, A., Rullier-Albenque, F., Forget, A., Colson, D., Rousse,
A., Marsi, M.: Observation of a coherent optical phonon in the iron pnictide superconductor
Ba(Fe1−x Cox )2 As2 (x = 0.06 and 0.08). Phys. Rev. B 80, 172504 (2009)
16. Takahashi, H., Kamihara, Y., Koguchi, H., Atou, T., Hosono, H., Katayama, I., Takeda, J.,
Kitajima, M., Nakamura, K.G.: Coherent optical phonons in the iron oxypnictide
SmFeAsO1−x Fx (x = 0.075). J. Phys. Soc. Jpn. 80, 013707 (2011)
17. Okano, Y., Katsuki, H., Nakagawa, Y., Takahashi, H., Nakamura, K.G., Ohmori, K.: Opti-
cal manipulation of coherent phonons in superconducting YBa2 Cu4 O7−δ thin films. Faraday
Discuss. 153, 375–382 (2011)
18. Lee, I.H., Yee, K.J., Lee, K.G., Oh, E., Kim, D.S.: Coherent optical phonon mode oscillations
in wurtzite ZnO excited by femtosecond pulses. J. Appl. Phys. 93, 4939–4941 (2003)
19. Hu, J., Misochko, O.V., Takahashi, H., Koguchi, H., Eda, T., Nakamura, K.G.: Ultrafast zone-
center coherent lattice dynamics in ferroelectric lithium tantalate. Sci. Technol. Adv. Mater.
12, 034409 (2011)
20. Wall, S., Wegkamp, D., Foglia, L., Appavoo, K., Nag, J., Haglund, R.F., Jr. Stähler, Wolf,
J.: Ultrafast changes in lattice symmetry probed by coherent phonons. Nat. Commun 3, 721
(2012)
21. Wang, Y., Xu, X., Venkatasubramanian, R.: Reduction in coherent phonon lifetime in
Bi2 Te3 /Sb2 Te3 superlattices. Appl. Phys. Lett. 93, 113114 (2008)
22. Kamarajyu, N., Kumar, S., Sood, A.K.: Temperature-dependent chirped coherent phonon
dynamics in Bi2 Te3 using high-intensity femtosecond laser pulses. EPL 92, 47007 (2010)
23. Norimatsu, K., Hu, J., Goto, A., Igarashi, K., Sasagawa, T., Nakamura, K.G.: Coherent optical
phonons in a Bi2 Se3 single crystal measured via transient anisotropic reflectivity. Solid State
Commun. 157, 58–61 (2013)
24. Flock, J., Dekorsy, T., Misochko, O.V.: Coherent lattice dynamics of the topological insulator
Bi2 Te3 probed by ultrafast spectroscopy. Appl. Phys. Lett. 105, 011902 (2014)
25. Norimatsu, K., Hada, M., Yamamoto, S., Sasagawa, T., Kitajima, M., Kayanuma, Y., Nakamura,
K.G.: Dynamics of all Raman active coherent phonons in Sb2 Te3 . J. Appl. Phys. 117, 143102
(2015)
26. Hu, J., Igarashi, K., Sasagawa, T., Nakamura, K.G., Misochko, O.V.: Femtosecond study of
A1g phonons in the strong 3D topological insulators: From pump-probe to coherent control.
Appl. Phys. Lett. 112, 031901 (2018)
27. Mishina, T., Nitta, K., Matsumoto, Y.: Coherent lattice vibration of interlayer shearing mode
of graphite. Phys. Rev. B 62, 2908–2911 (2000)
28. Ishioka, K., Hase, M., Kitajima, M., Petek, H.: Coherent optical phonons in diamond. Appl.
Phys. Lett. 89, 231916 (2006)
29. Gambetta, A., Manzoni, C., Menna, E., Meneghetti, M., Cerullo, G., Lanzani, G., Tretiak,
S., Piryatinski, A., Saxena, A., Matin, R.L., Bishop, A.R.: Real-time observation of nonlinear
coherent phonon dynamics in singe-walled carbon nanotubes. Nat. Phys. 2, 515–520 (2006)
30. Kim, J.-H., Nugraha, A.R.T., Booshehri, L.G., Hároz, E.H., Satom, K., Sanders, G.D., Yee,
K.-J., Lim, Y.S., Stanton, C.J., Saito, R.: Coherent phonons in carbon nanotubes and graphene.
Chem. Phys. 413, 55–80 (2013)
31. Yan, Y.-X., Gamble Jr. E.B., Nelson, K.A.: Impulsive stimulated scattering: General importance
in femtosecond laser pulse interactions with matter, and spectroscopic applications. J. Chem.
Phys. 83, 5391 (1985)
32. Merlin, R.: Generating coherent THz phonons with light pulses. Solid State Commun. 102,
107–220 (1997)
33. Zeiger, H.J., Vidal, J., Cheng, T.K., Ippen, E.P., Dresselhaus, G., Dresselhaus, M.S.: Theory
for displacive excitation of coherent phonons. Phys. Rev. B 45, 768–778 (1992)
34. Pfeifer, T., Dekorsy, T., Kütt, W., Kurz, H.: Generation mechanism for coherent LO phonons
in surface-space-charge fields of III–V-compounds. Appl. Phys. A 55, 482–488 (1992)
35. Kuznetsov, A.V., Stanton, C.J.: Theory of coherent phonon oscillations in semiconductors.
Phys. Rev. Lett. 43, 3243–3246 (1994)
References 79

36. Shinihara, Y., Kawashita, Y., Iwata, J.-I., Yabana, K., Otobe, T., Bertsch, G.F.: First-principles
description for coherent phonon generation in diamond. J. Phys.: Condens. Matter 22, 38412
(2010)
37. Shinihara, Y., Yabana, K., Kawashita, Y., Iwata, J.-I., Otobe, T., Bertsch, G.F.: Coherent phonon
generation in time-dependent density functional theory. Phys. Rev. B 82, 155110 (2010)
38. Sasaki, H., Tanaka, R., Okano, Y., Minami, F., Kayanuma, Y., Shikano, Y., Nakamura, K.G.:
Coherent control theory and experiment of optical phonons in diamond. Sci. Rep. 8, 96909
(2018). https://doi.org/10.1038/s41598-018-27734-1.
39. Hase, M., Mizoguchi, K., Harima, H., Nakashima, S., Tani, M., Sakai, K., Hangyo, M.: Optical
control of optical phonons in bismuth filma. Appl. Phys. Lett. 69, 2474–2476 (1996)
40. Cho, G.C., Dekorsy, T., Bakker, H.J., Hövel, R., Kurz, H.: Generation and relaxation of coherent
majority phonons. Phys. Rev. Lett. 77, 4062–4065 (1996)
41. Hu, J., Zhang, H., Sun, Y., Misochko, O.V., Nakamura, K.G.: Temperature effect on the coupling
between coherent longitudinal phonons and plasmons in n-type and p-type GaAs. Phys. Rev.
B 97, 165307 (2018)
42. Sokolowski-Tinten, K., Blome, C., Blums, J., Cavalleri, A., Dietrich, C., Tarasevitch, A.,
Uschmann, I., Förster, E., Kammler, M., Horn-von-Hoegen, M., von der Linde, D.: Femo-
tosecond X-ray measurement of coherent lattice vibrations near the lindemann stability limit.
Nature 422, 287 (2003)
43. Nakamura, K.G., Ishii, S., Ishitsu, S., Shiokawa, M., Takahashi, H., Dharmalingam, K., Irisawa,
J., Hironaka, Y., Ishioka, K., Kitajima, M.: Femtosecond time-resolved X-ray diffraction from
optical coherent phonons in CdTe(111) crystal. Appl. Phys. Lett. 93, 061905 (2008)
44. Johnson, S.L., Beaud, P., Vorobeva, E., Milne, C.J., Murray, É.D., Fahy, S., Ingland, G.: Directly
observing squeezed phonon states with femtosecond x-ray diffraction. Phy. Rev. Lett. 102,
175503 (2009)
45. Harmand, M., Coffee, R., Bionta, M.R., Chollet, M., French, D., Zhu, D., Fritz, D.M., Lemke,
H.T., Medvedev, N., Ziaja, B., Toleikis, S., Cammarata, M.: Achieving few-femtosecond time-
sorting at har X-ray free-electron lasers. Nat. Photonics 7, 215–218 (2013)
46. Johnson, S.L., Beaud, P., Möhr-Vorobeva, E., Caviezel, A., Ingold, G., Milne, C.J.: Direct
observation of non-fully-symmetric coherent optical phonons by femtosecond x-ray diffraction.
Phys. Rev. B 87, 054301 (2013)
47. Yoshida, M., Fujimoto, Y., Hironaka, Y., Nakamura, K.G., Kondo, K., Ohtani, M., Tsunemi,
H.: Generation of picosecond hard x rays by tera watt laser focusing on a copper target. Appl.
Phys. Lett. 73, 2393–2395 (1998)
48. Garrett, G.A., Rojo, A.R., Sood, A.K., Whitaker, J.F., Merlin, R.: Vacuum squeezing of solids:
macroscopic quantum states driven by light pulses. Science 275, 1638–1640 (1997)
49. Hu, X., Nori, F.: Phonon squeezed states generated by second-order Raman scattering. Phys.
Rev. Lett. 79, 4605–4608 (1997)
50. Misochko, O.V., Sakai, K., Nakashima, S.: Phase-dependent noise in femtosecond pump-probe
experiments on Bi and GaAs. Phys. Rev. B 61, 11225–11228 (2000)
51. Misochko, O.V., Hu, J., Nakamura, K.G.: Controlling phonon squeezing correlation via one-
and two- phonon interference. Phys. Lett. A 375, 4141–4146 (2011)
52. Esposito, M., Titimbo, K., Zimmermann, K., Giusti, F., Randi, F., Boschetoo, D., Floreanini,
R., Benatti, F., Fausti, D.: Photon number statistics uncover the fluctuations in non-equilibrium
lattice dynamics. Nat. Commu. 6, 10249 (2015)
53. Benatti, F., Esposito, M., Fausi, D., Floreanini, R., Titimbo, K., Zimmermann, K.: Generation
and detection of squeezed phonons in lattice dynamics by ultrafast optical excitations. New J.
Phys. 19, 02302 (2017)
Chapter 6
Coherent Phonons: Quantum Theory

In this chapter, we describe a unified quantum mechanical description for the genera-
tion and detection of coherent optical phonons using a simple model. Two electronic
states coupled with displaced harmonic oscillators, which represent phonons, are
used. The dipole interaction between an optical pulse and the system is assumed
and treated as a perturbation interaction. Time evolution of the density operator is
calculated using the second-order perturbation approximation.

6.1 Generation Mechanism with Displaced Harmonic

Oscillator

There are several quantum theories for generating coherent phonons. The different
theoretical treatments are often used to describe DECP and ISRS processes [1–5].
Here, we attempt to describe the generation of coherent phonons using a unified
quantum mechanical theory [6] in both opaque and transparent conditions.

6.1.1 Hamiltonian

Here, we consider a two-level system for the electronic state and a one-dimensional
harmonic oscillator for the optical phonon at the Γ point with a wave vector k = 0.1
Figure 6.1 shows a schematic of the model potential. |g and |e indicate the electronic

1 Quantum mechanical models with the two-level electronic states are often used to study vibrational

dynamics of a molecular system [7–10].

© Springer Nature Switzerland AG 2019 81
K. Nakamura, Quantum Phononics, Springer Tracts in Modern Physics 282,
https://doi.org/10.1007/978-3-030-11924-9_6
82 6 Coherent Phonons: Quantum Theory

Fig. 6.1 Schematic of the V(Q)

model potential consisting of
two electronic states and |e>
displaced harmonic
potentials

|g>

ground and excited states, respectively. The Hamiltonian Ĥ0 of the materials system
is
 
Ĥ0 = Ĥg |gg| + ε + Ĥe |ee| (6.1)

Ĥg = ωb̂† b̂ (6.2)

 
Ĥe = ωb̂ b̂ + αω b̂† + b̂ ,
†
(6.3)

where ω is the angular frequency and b̂† and b̂ are the creation and annihilation oper-
ators of the phonon, respectively [6]. Ĥg and Ĥe are Hamiltonians for the harmonic
oscillator in the electronic ground and excited states. The zero-point energy of the
harmonic oscillator is set to be zero. α represents shift between the harmonic oscilla-
tor of the electronic ground and that in the excited states.2 ε is a energy gap between
the electronic ground and excited states. In general, the creation and annihilation
operators are dependent on the wave vector k and expressed as b̂k† and b̂k . Here, we
use only the b̂† and b̂ which represent the operators at k ∼ 0. The reason is that,
according to the phase matching (or momentum conservation), the wave vectors of
phonons which can be excited by the optical processes are only those lying close
to the Γ -point (k ∼ 0), because the optical wavelength is much larger than the lat-
tice constant. The eigenmodes of optical phonons form a continuum around k = 0.
The electromagnetic field of the optical pulse interacts only with atoms within the
penetration depth (δL). In opaque conditions, the penetration depth is usually much
smaller than the sample thickness. This effect relaxes the condition of phase match-
ing and allows the modes with δk ∼ 1/(δL) around k = 0 to be excited. In addition,
the spot size of the pump pulse is smaller than the surface area of the crystal, and the
phase-matching condition along the lateral direction is also relaxed.
The dipole interaction between the materials system and the incident light is
assumed and the interaction Hamiltonian is represented by

2 α2 is the Huang–Rhys factor, which is considered to be small (α 2 < 1) in a bulk solid.

6.1 Generation Mechanism with Displaced Harmonic Oscillator 83
 
ĤI (t) = μE 0 f (t) e−iΩt |eg| + eiΩt |ge| , (6.4)

where μ is a transition dipole, E 0 and f (t) are an amplitude and an envelope of

electric field of the light pulse. Ω is the angular frequency of the light. The operator
|eg| causes a transition from the electronic ground state |g to the excited state
|e. The light pulse is assumed to be Fourier-transform limited. Here, we use the
rotating wave approximation (RWA). The material system Hamiltonian Ĥ0 is time
independent and the interaction Hamiltonian ĤI (t) is time dependent.
The time evolution of the electron–phonon coupled state is obtained by solving
the time-dependent Schrödinger equation as given below:

d  
i |Ψ (t) = Ĥ0 + ĤI (t) |Ψ (t). (6.5)
dt
The solution is given formally by using a time-ordered exponential:
    t 
1 1
|Ψ (t) = exp Ĥ0 t exp+ Ĥ  I (t  )dt 
i i −∞
×|Ψ (−∞), (6.6)

with
   
i −i
Ĥ  I (t  ) = exp Ĥ0 t ĤI (t  ) exp Ĥ0 t , (6.7)
 

where |Ψ (−∞) is the Ket vector of the initial state at t = −∞. Ĥ  I (t  ) is given by
using (6.1) and (6.3):
   
i
Ĥ  I (t) = exp Ĥg |gg| + (ε + Ĥe |ee|)t

× μE 0 f (t)(e−iΩt |eg| + eiΩt |ge|)
 
−i
× exp ( Ĥg |gg| + (ε + Ĥe )|ee|)t

     
i −i
= μE 0 f (t) exp (ε − Ω + Ĥe )t |eg| exp Ĥg t + H.c. , (6.8)
 

where we used an expansion of exponential.3

3 When we set B̂ = |ee|, the exponential exp( Ĥe B̂t) is expressed by exp( Ĥe B̂t) =
n ( Ât) Since B̂ 2 = (|ee|)(|ee|) = |ee| = B̂, then B̂ n = B̂ for n ≥ 1. Finally, we get
n B̂ n .

exp( Ĥe |ee|t)|eg| = exp( Ĥe t)|eg|.

84 6 Coherent Phonons: Quantum Theory

Fig. 6.2 Schematics of

Feynman diagrams for the
impulsive absorption (a) and
ISRS (b and c). The single
and double lines represent
the electronic ground and
excited states, respectively.
Time passes from the left
side to the right side

6.1.2 Transition Processes

In the abovementioned system, we consider two transition processes: an impulsive

absorption process and an impulsive stimulated Raman scattering (ISRS) process.
Figure 6.2 shows double-sided Feynman diagrams for the absorption (a) and ISRS
(b and c) processes. Time passes from the left side to the right side.
The initial state is set to be the zero-phonon state |0 in the electronic ground
state |g, which is expressed by |g, 0. Then, the density matrix of the initial state
is |g, 0g, 0|. Here, we consider the second-order perturbation, in which the inter-
action between the materials system and the light pulse occurs two times. In the
absorption process, |g and g| change to |e and e| by the first and second inter-
actions, respectively, and the final state becomes |ee| for the electronic state. The
phonon is generated in the electronic excited state. On the other hand, in the ISRS
process, |g changes to |e by the first interaction and changes to |g by the second
interaction. The final state is in the electronic ground state (|gg|). The phonon is
induced in the electronic ground state. It is worth noting that both the absorption and
ISRS processes coexist in the quantum mechanics. These processes are expressed
with the second-order perturbation calculation.
The IA and ISRS processes refer to the phonon-generation processes via the
population and polarization in the electronic excited states, respectively. The DECP
mechanism for an opaque condition is included in the IA process. The ISRS process
6.1 Generation Mechanism with Displaced Harmonic Oscillator 85

includes not only the ISRS mechanism, which is previously proposed (Chap. 5)
for a transparent condition, but also the phonon-generation mechanism in an opaque
condition via the photo-induced electronic polarization. The screening of the surface-
space-charge field is also included in either the IA or ISRS processes depending on
the excitation condition, because both the photoexcited electron–hole pair and the
polarization can change the surface-space-charge field, which causes deformation of
the potential.4 Even at the resonance condition, the ISRS process is different from
the absorption and emission process, because the population in the electronic state
is not induced by the ISRS process.

6.1.3 Second-Order Perturbation

Here, we consider a density operator ρ(t) ≡ |ψ(t)ψ(t)|. The density operator at

the initial state is ρ(0) = |g, 0g, 0|. Then, |ψ(t) at time t is obtained by
    t 
i 1
|ψ(t) = exp − Ĥ0 t exp+  Ĥ  I (t  )dt  |g, 0. (6.9)
 i −∞

We define two operators as

 
i
† (t) = exp (ε − Ω + Ĥe )t (6.10)

 
−i
B̂(t) = exp Ĥg t , (6.11)


where these operators act on the phonon states. Using these operators, Ĥ  I (t) is
expressed by
 
Ĥ  I (t) = μE 0 f (t) † (t)|eg| B̂(t) + B̂ † (t)|ge| Â(t) . (6.12)

Next, we expand the time-ordered exponential in the second-order perturbation and

get

4 The harmonic potential can be deformed under a long-range external field such as the surface-
charge field. Supposing a uniform electric field (F(x) = −dx) acting along the phonon coordinate
(x), the effective charge field can be treated as uniform in such a short scale. When the external
field potential is applied to a harmonic potential U (x) = kx 2 /2, the potential changes to U  (x) =
U (x) + F(x) = k(x − d/k)2 /2 − d2 /(2k). Then, the potential minimum position and energy shift
are d/k and −d2 /(2k), respectively. The slope d of the external fields in the electronic excited state
is lower than that in the ground state, because the surface screening is suppressed by a electron–hole
pair or electronic polarization. Then, the effective harmonic potential in the excited state is displaced
from that in the ground state.
86 6 Coherent Phonons: Quantum Theory
  t 
1
exp+ Ĥ  I (t  )dt 
i −∞
  2  t  t 
1 t   1
=1+ 
Ĥ I (t )dt + Ĥ  I (t  ) Ĥ  I (t  )dt  dt 
i −∞ i −∞ −∞
  
μE 0 t
=1+ f (t  ) † (t  )|eg| B̂(t  ) + B̂ † (t  )|ge| Â(t  ) dt 
i −∞
    t  
μE 0 2 t
+ f (t  ) † (t  )|eg| B̂(t  ) + B̂ † (t  )|ge| Â(t  )
i −∞ −∞
 
× f (t ) Â (t )|eg| B̂(t  ) + B̂ † (t  )|ge| Â(t  ) dt  dt  .
 † 
(6.13)

The third term in the right-hand side is obtained as

 2   t 
μE 0 t
f (t  ) f (t  ) † (t  ) B̂ † (t  )|ee| B̂(t  ) Â(t  )
i −∞ −∞

+ B̂ † (t  ) † (t  )|gg| Â(t  ) B̂(t  ) dt  dt  (6.14)

by using e|g = 0. We define new two operators as

 t  
F̂(t) = f (t  ) † (t  )|eg| B̂(t  ) + B̂ † (t  )|ge| Â(t  ) dt 
−∞
 t  t
Ĝ(t) = f (t  ) f (t  ) † (t  ) B̂ † (t  )|ee| B̂(t  ) Â(t  )
−∞ −∞

+ B̂ † (t  ) † (t  )|gg| Â(t  ) B̂(t  ) dt  dt  , (6.15)

and simplify the ket vector as

   2
−i μE 0 μE 0
|ψ(t) = exp Ĥ0 t 1+ F̂(t) + Ĝ(t) |g, 0. (6.16)
 i i

When the operators ( F̂(t) and Ĝ(t)) act on the |g, 0 state, we can neglect the terms
with |ge| and |ee| and the effective operators are
 t  
F̂(t) = f (t  ) † (t  )|eg| B̂(t  ) dt 
−∞
 t  t
Ĝ(t) = f (t  ) f (t  ) B̂ † (t  ) † (t  )|gg| Â(t  ) B̂(t  ) dt  dt  . (6.17)
−∞ −∞
6.1 Generation Mechanism with Displaced Harmonic Oscillator 87

The density operator is obtained by

ρ̂(t) = |ψ(t)ψ(t)|
     
−i μE 0 μE 0 2
= exp Ĥ0 t 1+ F̂(t) + Ĝ(t) |g, 0
 i i
     
μE 0 μE 0 2 † i
× g, 0| 1 + F̂ † (t) + Ĝ (t) exp Ĥ0 t . (6.18)
−i −i 

Here, we focus our attention on the terms with μ2 , which are the second-order
interaction with light and denote this as ρ̂ (2) (t):
      
−i μE 0 μE 0 i
ρ̂ (2) (t) = exp Ĥ0 t F̂ † (t) exp
F̂(t)|g, 0g, 0| Ĥ0 t
 i −i 
  2  
−i μE 0 i
+ exp Ĥ0 t Ĝ(t)|g, 0g, 0| exp Ĥ0 t
 i 
     
−i μE 0 2 † i
+ exp Ĥ0 t |g, 0g, 0| Ĝ (t) exp Ĥ0 t , (6.19)
 −i 

where we denote the terms in the right-hand side as ρ̂1(2) (t), ρ̂2(2) (t), and ρ̂3(2) (t). ρ̂1(2) (t)
corresponds to the impulsive absorption process ρ̂2(2) (t), and ρ̂3(2) (t) corresponds to
the impulsive stimulated Raman scattering process.

6.1.3.1 Impulsive Absorption

Let us calculate the ρ̂1(2) (t) term as given below:

 2    
μE 0 −i i
ρ̂1(2) (t) = exp Ĥ0 t F̂(t)|g, 0g, 0| F̂ (t) exp
†
Ĥ0 t
  
 2   t  
μE 0 −i
= exp Ĥ0 t dt  f (t  ) † (t  )|eg| B̂(t  ) |g, 0
  −∞
 t    
i
× g, 0| dt  f (t  ) B̂ † (t  )|eg| Â(t  ) exp Ĥ0 t . (6.20)
−∞ 

The former term in the right-hand side becomes

  t  
−i
exp Ĥ0 t dt  f (t  ) † (t  )|eg| B̂(t  ) |g, 0
 −∞
 t   
  −i  
= dt f (t ) exp Ĥg |gg| + ε + Ĥe |ee| t
−∞ 
88 6 Coherent Phonons: Quantum Theory
     
i −i
× exp ε − Ω + Ĥe t  exp Ĥg t  |e, 0
 
 t      
−i −i
= dt  f (t  ) exp ε − Ω + Ĥe (t − t  ) exp Ωt |e, 0,
−∞  
(6.21)

where we used
     1 2 
−i
exp Ĥg t  |0 = 1 + ωb̂† b̂ + ωb̂† b̂ + · · · |0 = |0, (6.22)
 2

because b̂† b̂|0 = 0. A similar calculation gives the latter half in the right-hand side
in (6.20) as
 t      
 i   i
e, 0| dt f (t ) exp ε − Ω + Ĥe (t − t ) exp Ωt . (6.23)
−∞  

Then, we get the density operator as

 2   
μE 0 t
−i  
ρ̂1(2) (t) = 
dt f (t ) exp  
ε − Ω + Ĥe (t − t ) |e, 0
 −∞ 
 t    
  i 
e, 0| dt f (t ) exp ε − Ω + Ĥe (t − t ) ,
−∞ 
(6.24)

which shows the system is in the electronic excited state at time t. The expectation
value for the electronic excitation state is
    
μE 0 2 t −i  
e|ρ̂1(2) (t)|e = dt  f (t  ) exp ε − Ω + Ĥe (t − t  ) |0
 −∞ 
 t    
i
0| dt  f (t  ) exp ε − Ω + Ĥe (t − t  )
−∞ 
 2
μE 0
= |ψ p (t)ψ p (t)|, (6.25)


where |ψ p (t) is the phonon ket vector:

  
t
−i   


|ψ p (t) = dt f (t ) exp ε − Ω + Ĥe (t − t ) |0. (6.26)
−∞ 

The final electronic state is the excited state and this process corresponds to the
absorption process.
6.1 Generation Mechanism with Displaced Harmonic Oscillator 89

6.1.3.2 Impulsive Stimulated Raman Scattering Process

Next, we calculate the ρ̂2(2) (t) term in a similar way as

 2 
  
μE 0 −i i
ρ̂2(2) (t) = exp
Ĥ0 t Ĝ(t)|g, 0g, 0| exp Ĥ0 t
  
 2    t  t
μE 0 −i
= exp Ĥ0 t f (t  ) f (t  ) B̂ † (t  ) † (t  )
  −∞ −∞
 
 
   i
× |gg| Â(t ) B̂(t ) dt dt |g, 0g, 0| exp Ĥ0 t . (6.27)


Here
     
 i
  i
B̂ (t ) Â (t )|g = exp
† †
Ĥg t exp ε − Ω + Ĥe t  |g, (6.28)
 

and
 
  
  −i   −i 
g| Â(t ) B̂(t ) = g| exp ε − Ω + Ĥe t exp Ĥg t , (6.29)
 

then, we get
 
i
B̂ † (t  ) † (t  )|gg| Â(t  ) B̂(t  ) = exp Ĥg t 

     
−i −i
× exp ε − Ω + Ĥe (t  − t  ) exp Ĥg t  |gg|. (6.30)
 

The integral is then expressed by

     
t t
  i  −i  
 
f (t ) f (t ) exp Ĥg t exp ε − Ω + Ĥe (t − t )
−∞ −∞  
 
−i
× exp Ĥg t  dt  dt  |g, 0

 t  t  
  i 
= f (t ) f (t ) exp Ĥg t
−∞ −∞ 
 
−i  
× exp ε − Ω + Ĥe (t − t ) dt  dt  |g, 0
 
(6.31)

90 6 Coherent Phonons: Quantum Theory

In addition,
   
i i
g, 0| exp Ĥ0 t = |g, 0 exp Ĥg |gg|t = g, 0|, (6.32)
 

then, we get
     t  t  
μE 0 2 −i i
ρ̂2(2) (t) = exp Ĥ0 t  
f (t ) f (t ) exp Ĥg t 
i  −∞ −∞ 
 
−i  
exp ε − Ω + Ĥe (t  − t  ) dt  dt  |g, 0g, 0|

    t  
μE 0 2 t   −i 
= f (t ) f (t ) exp Ĥg (t − t )
i −∞ −∞ 
 
−i  
exp ε − Ω + Ĥe (t  − t  ) dt  dt  |g, 0g, 0|. (6.33)


Therefore, the expectation value for the electronic states, which corresponds to the
trace for the electronic states, is
     t  t  
μE 0 2 −i i
g|ρ̂2(2) (t)|g = g| exp Ĥ0 t  
f (t ) f (t ) exp Ĥg t 
i  −∞ −∞ 
 
−i  
exp ε − Ω + Ĥe (t  − t  ) dt  dt  |g, 0g, 0||g

    t  
μE 0 2 t −i
= f (t  ) f (t  ) exp Ĥg (t − t  )
i −∞ −∞ 
   
−i
exp ε − Ω + Ĥe (t  − t  ) dt  dt  |00|

= |ψ p (t)0|, (6.34)

where |ψ p (t) is the ket for a phonon state

  t  
t
−i
|ψ p (t) = f (t  ) f (t  ) exp
Ĥg (t − t  )
−∞ −∞ 
   
−i
exp ε − Ω + Ĥe (t  − t  ) dt  dt  |0. (6.35)


ρ̂3(2) (t) is also calculated in a similar way to give

ρ̂3(2) (t) = |0ψ p (t)|, (6.36)

6.1 Generation Mechanism with Displaced Harmonic Oscillator 91

which is the Hermitian conjugate of ρ̂2(2) (t). In the paths represented by ρ̂2(2) (t)
and ρ̂3(2) (t), the initial and final states are in the electronic ground state and the light
interaction occurs for the bra or ket state. These paths correspond to the ISRS process.

6.1.4 Time Evolution of Phonon Amplitude

The expectation value of the phonon amplitude is calculated by Q(t) as given

below:

Q(t) = T r ( Q̂ ρ̂ (2) (t))

= T r ( Q̂ ρ̂1(2) (t)) + T r ( Q̂ ρ̂2(2) (t)) + T r ( Q̂ ρ̂3(2) (t)). (6.37)

6.1.4.1 Impulsive Absorption Process

The phonon ket in the electronic excited state which is expressed by (6.26) as
 t    
−i −i
|ψ p (t) = f (t  ) exp (ε − Ω)(t − t  ) exp Ĥe (t − t  ) dt  |0. (6.38)
−∞  

The Hamiltonian for the harmonic oscillator in the |e is

Ĥe = ωb̂† b̂ + αω(b̂† + b̂)

= D̂ † (α) Ĥg D̂(α) − ωα 2 , (6.39)

where
 D̂(α)  is the displacement operator and expressed by D̂(α) ≡ exp
α b̂† − α ∗ b̂ . This relation is shown by
 
D̂ † (α) Ĥg D̂(α) = ω D̂ † (α)b̂† b̂ D̂(α)
 
= ω D̂ † (α)b̂† D̂ D̂ † b̂ D̂(α)
  
= ω b̂† + α b̂ + α

= Ĥe + ωα 2 . (6.40)

Using the displacement operator, the coherent state |α is expressed by |α =
D̂(α)|0. Using these relations, we get
92 6 Coherent Phonons: Quantum Theory
   
−i −i  
exp Ĥe t = D̂ † (α) exp D̂(α) exp iωα 2 t
Ĥg t
 
   
= exp iωα 2 t D̂ † (α) exp −iωb̂† b̂ D̂(α), (6.41)

and |ψ p (t) is obtained as

  
t
i  
|ψ p (t) = f (t ) exp − (ε − Ω)(t − t ) exp iωα 2 (t − t  )
 
−∞ 
 
× D̂ (α) exp −iωb̂ b̂(t − t  ) D̂(α)dt  |0
† †

 t  
i
= f (t  ) exp − (ε − Ω − α 2 ω)(t − t  )
−∞ 

× D̂ † (α)|αe−iω(t−t ) dt  , (6.42)

where |αe−iω(t−t )  is the time-dependent coherent state. By setting a new parameter
B ≡ −(ε − Ω − α 2 ω)/(ω), the expectation value of the phonon amplitude in
the electronic excited state is obtained as

  † 
Q̄ = ψ p (t)| b̂ + b̂ |ψ p (t)
2ω
  t  t
  
= f (t  ) f (t  )e−iωB(t−t ) eiωB(t−t )
2ω −∞ −∞

  
× αe−iω(t−t ) | D̂(α) b̂† + b̂ D̂ † (α)|αe−iω(t−t ) dt  dt 
  t  t
  
= f (t  ) f (t  )e−iωB(t −t )
2ω −∞ −∞

  
× αe−iω(t−t ) | b̂† − α + b̂ − α |αe−iω(t−t ) dt  dt 
  t  t
  
= f (t  ) f (t  )e−iωB(t −t )
2ω −∞ −∞
  
  
× α eiω(t−t ) + e−iω(t−t ) − 2 αe−iω(t−t ) |αe−iω(t−t ) dt  dt  . (6.43)

The time-dependent coherent state is expressed by

∞
 αn 
|αe−iω(t−t )  = e−|α| /2
√ e−iω(t−t )n |n,
2
(6.44)
n=0 n!
6.1 Generation Mechanism with Displaced Harmonic Oscillator 93

and
∞

 αm 
αe−iω(t−t ) | = e−|α| /2
m| √ eiω(t−t )m .
2
(6.45)
m=0 m!

Using these relations, we get

∞

  α 2n 
−t  )n
αe−iω(t−t ) |αe−iω(t−t )  = e−|α| e−iω(t
2

n=0
n!
    

= exp −|α|2 exp |α|2 e−iω(t −t )
   

= exp |α|2 eiω(t −t ) − 1 . (6.46)

Then, Q̄ is obtained as
  t  t  
    
Q̄ = f (t  ) f (t  )e−iωB(t −t ) × α eiω(t−t ) + e−iω(t−t ) − 2
2ω −∞ −∞
   

× exp |α|2 eiω(t −t ) − 1 dt  dt  . (6.47)

Finally, we obtain the expectation value of the phonon amplitude by inserting the
form of B as
 
μE 0 2
Q A (t) = T r [ Q̂ ρ̂2(2) ] = Q̄

    t  t  
μE 0 2   
=α f (t  ) f (t  ) eiω(t−t ) + e−iω(t−t ) − 2
 2ω −∞ −∞
   
  
× exp −α 1 + iω(t  − t  ) − eiω(t −t ) ei(ε−Ω)(t −t )/ dt  dt  .
2

(6.48)

If the displacement α between the excited-state and ground-state potentials is small

(α 1), we regard α 2 ≈ 0 and get a simple form as
 2     
μE 0  t t
 
Q A (t) = α f (t  ) f (t  ) eiω(t−t ) + e−iω(t−t ) − 2
 2ω −∞ −∞
 
× ei(ε−Ω)(t −t )/
dt  dt  . (6.49)

6.1.4.2 Impulsive Stimulated Raman Scattering Process

In the ISRS process, the final state is the electronic ground state, which is given by
94 6 Coherent Phonons: Quantum Theory
 2
μE 0  
g|ρ̂2(2) (t) + ρ̂3(2) (t)|g =− |ψ p (t)0| + |0ψ p (t)| . (6.50)


The |ψ p (t) is expressed using the initial state |0 at t = −∞ as

  t  
t
−i
|ψ p (t) = f (t  ) f (t  ) exp
Ĥg (t − t  )
−∞ −∞ 
   
−i −i
× exp (ε − Ω + Ĥe )(t  − t  )) exp Ĥg t  dt  dt  |0. (6.51)
 

The expectation value of the phonon amplitude induced by the ISRS is

 2 
μE 0   
Q R (t) = − 0|b̂|ψ p (t) + ψ p (t)|b̂† |0 . (6.52)
 2ω

The first term of this equation is calculated by

 t  t  
−i
0|b̂|ψ p (t) = f (t  ) f (t  )0|b̂ exp Ĥg (t − t  )
−∞ −∞ 
   
−i −i
× exp (ε − Ω)(t  − t  ) exp Ĥe (t  − t  ) dt  dt  |0.
 
(6.53)

Using the relationship

 
−i 2    
Ĥe (t − t ) = eiωα (t −t ) D̂ † (α)e−iωb̂ b̂(t −t ) D̂(α),
  †
exp (6.54)


we get
 
−i 2    
exp Ĥe (t − t ) |0 = eiωα (t −t ) D̂ † (α)|αe−iω(t −t ) .
 
(6.55)


On the other hand, there is a relationship

eiωb̂ b̂t b̂e−iωb̂ = e−iωt b̂

† †
b̂t
(6.56)

and then
b̂e−iωb̂ = e−iωt e−iωb̂ b̂t b̂.
† †
b̂t
(6.57)

Using this relation, we get

   
−i  −iω(t−t  ) −i 
b̂ exp Ĥg (t − t ) = e exp Ĥg (t − t ) b̂ (6.58)
 
6.1 Generation Mechanism with Displaced Harmonic Oscillator 95

and
   
−i −i
0|b̂ exp Ĥg (t − t  ) exp Ĥe (t  − t  ) |0
 

(t  −t  ))  
= e−iω(t−t ) e−iωα 0|b̂ D̂ † (α)|αe−iω(t −t ) .
2
(6.59)

This braket is simplified as

   
0|b̂ D̂ † (α)|αe−iω(t −t )  = 0| D̂ † (α) D̂(α)b̂ D̂ † (α)|αe−iω(t −t ) 
 
= 0| D̂ † (α)(b̂ − α)|αe−iω(t −t ) 
 
= α|(b̂ − α)|αe−iω(t −t ) 
  
  
= α e−iω(t −t ) − 1 α|αe−iω(t −t ) . (6.60)

Using these equations, we get

 t  t
  
dt dt  f (t  ) f (t  )e−iω(t−t ) e−iωα (t −t )
2
0|b̂|ψ p (t) =
−∞ −∞
  
   
  
× α e−iω(t −t ) − 1 exp α 2 (e−iω(t −t ) − 1) e−i(ε−Ω)(t −t )/
 t  t  
 

=α dt dt  f (t  ) f (t  ) e−iω(t−t ) − e−iω(t−t )
−∞ −∞
   
  
× exp −α 1 − iω(t  − t  ) + e−iω(t −t ) e−i(ε−Ω)(t −t )/ .
2

(6.61)

Finally, we get the expectation value of the phonon amplitude as

    t  t  
μE 0 2   
Q R (t) = α dt  dt  f (t  ) f (t  ) e−iω(t−t ) − e−iω(t−t )
 2ω −∞ −∞
   
  
× exp −α 1 − iω(t − t ) + e−iω(t −t ) e−i(ε−Ω)(t −t )/ + c.c.
2  

(6.62)

For a small α case, the equation is simplified as

 2    t
μE 0  t

Q R (t) = α dt dt  f (t  ) f (t  )
 −∞ −∞ 2ω
  
  
e−iω(t−t ) − e−iω(t−t ) e−i(ε−Ω)(t −t )/ + c.c. (6.63)
96 6 Coherent Phonons: Quantum Theory

6.1.5 Gaussian Pulse Case

The time evolution of the phonon amplitude is obtained by numerical calculation of

(6.48) and (6.63). For some simple cases, for example, the Gaussian function of the
envelope of an optical pulse, we obtain analytical solutions. Here, we use an optical
pulse E(t) with a Gaussian envelope function:

E(t) = E 0 f (t) (exp(iΩt) + exp(−iΩt)) (6.64)

 2
1 −t
f (t) = √ exp , (6.65)
πσ σ2

intensity of the electric field, Ω is the optical frequency, σ is the

where E 0 is the peak√
pulse width, and 1/( π σ ) is a normalization factor. We hereafter set (μE 0 /)2 = 1
and α 1 for simplifying calculations.

6.1.5.1 Resonance Condition

At the resonance condition of the excitation optical pulse (ε − Ω = 0), the ampli-
tude of the phonons excited via the absorption process is expressed
 t  t
 
Q A (t) = α dt dt  f (t  ) f (t  )
2ω −∞ −∞
  

× e−iω(t−t ) + eiω(t−t ) − 2 . (6.66)


Here, f (t) is a real function and e−iω(t−t ) = cos ω(t − t  ) − i sin ω(t − t  ) and

eiω(t−t ) = cos ω(t − t  ) − i sin ω(t − t  ). Because t  and t  are independent and
have the same integration range, the sine-function terms become zero. Then, we get
  t  t

Q A (t) = α dt  dt  f (t  ) f (t  )
2ω −∞ −∞
 
× cos ω(t − t  ) + cos ω(t − t  ) − 2 . (6.67)

The similar calculation gives the expected amplitude of the phonons excited by the
ISRS process which is given by
  t  t
 
Q R (t) = α dt dt  f (t  ) f (t  )
2ω −∞ −∞
  

× e−iω(t−t ) − e−iω(t−t ) + c.c.
  t  t
 
= 2α dt dt  f (t  ) f (t  )
2ω −∞ −∞
 
× cos ω(t − t  ) − cos ω(t − t  ) . (6.68)
6.1 Generation Mechanism with Displaced Harmonic Oscillator 97

Impulsive absorption process:

For a long time delay condition (t σ ), the upper limit of the integral can be replaced
by ∞. We integrate separately for t  and t  , because they are independent:
 ∞  ∞
 1 −iωt /σ 2 iωt 
dt  f (t  )e−iω(t−t ) = √ e−t dt  .
2
e e (6.69)
−∞ πσ −∞

By using the Gaussian integral, we can calculate as

 ∞  ∞   
/σ 2 iωt  (t  − iσ ω/2)2 −σ 2 ω2
e−t dt  = 
2
e exp − dt exp
−∞ −∞ σ2 4
 
√ −σ ω 2 2
= π σ exp . (6.70)
4

Then, the phonon amplitude is obtained as

      
 −σ 2 ω2 −iωt −σ 2 ω2 iωt
Q A (t) = α exp e + exp e −2
2ω 4 4
    
 −σ 2 ω2
= 2α exp cos(ωt) − 1
2ω 4
  2 2 
−σ ω
= Q 0 exp cos(ωt) − 1 , (6.71)
4
√
where Q 0 ≡ 2α /2ω. The result means that oscillation of the phonon amplitude
is cosine-like. If we take a short pulse limit (σ ω → 0), the phonon amplitude is
expressed by

Q A (t) = Q 0 (cos ωt − 1) . (6.72)

On the other hand, the phonon amplitude decreases along with increasing pulse
width.
ISRS process:
We also consider the t σ condition and set the upper limit of the integration ∞:
 ∞  ∞   

−iωt 
Q R (t) = Q 0 e dt dt  f (t  ) f (t  ) eiωt − eiωt + c.c. (6.73)
−∞ −∞

Here, we set new parameters (s and u) by

 
s ≡ t  + t  /2 (6.74)
 
u ≡ t −t , (6.75)
98 6 Coherent Phonons: Quantum Theory

which have the integral range [−∞, ∞] for s and [0, ∞] for u, respectively. t 
and t  are expressed by t  = s + u/2 and t  = s − u/2, respectively. Using these
parameters, we get
   
  iωt  iωt  1 −(t 2 + t 2 )  iωt  iωt 

f (t ) f (t ) e −e = exp e − e , (6.76)
πσ2 σ2

and

t 2 + t 2 = (s + u/2)2 + (s − u/2)2 = 2s 2 + u 2 /2 (6.77)

 
eiωt − eiωt = eiω(s+u/2) − eiω(s−u/2)
= eiωs 2i sin(ωu/2). (6.78)

Using these equations, we get

  ∞
e−iωt ∞
Q R (t) = Q 0 ds du
π σ 2 −∞ 0
   ωu 
2s 2 u2
× exp − 2 − e iωs
2i sin + c.c. (6.79)
σ 2σ 2 2

Here, we use
 ∞   √  
2s 2 π −σ 2 ω2
ds exp − 2 + iωs = √ exp (6.80)
−∞ σ 2σ 8

and
 ∞    ωu  √    σ ω2√2
u2 −σ 2 ω2
du exp − 2 sin = 2σ exp t 2 dt. (6.81)
0 2σ 2 8 0

This integral is defined by the pulse width and the phonon frequency and constant
for a given condition. When we set this constant value as
   σ ω2√2
4 −σ 2 ω2
A ≡ Q 0 √ exp t 2 dt (6.82)
π 4 0

the phonon amplitude is given by

Q R (t) = A sin ωt. (6.83)

The amplitude of phonons induced by ISRS oscillates sine-like, though that of the
impulsive absorption oscillates cosine-like. It is worth noting that the optical phonon
amplitude goes to zero for the short pulse limit (σ ω → 0), because the integral A
goes to zero.
6.1 Generation Mechanism with Displaced Harmonic Oscillator 99

0.4 (a)
0.2

0.0
Mean value of phonon coordinate (Q0)

-0.2

-0.4
0.5 (b)
0.0
-0.5
-1.0
-1.5
-2.0
0.5 (c)
0.0
-0.5
-1.0
-1.5
-2.0
-0.5 0.0 0.5 1.0 1.5 2.0 2.5
Delay (T )

Fig. 6.3 Time evolution of the mean value of the phonon amplitude with pulse width (FWHM) of
0.1T (solid curves), 0.55T (dotted curves), and 0.9T (dashed curves) for the vibrational period T
of the optical phonon. The bottom axis represents the pump–probe delay in unit of T . a represents
Q R  excited by ISRS process, b represents Q A  excited by IA process. The total amplitude
Q = Q R  + Q A is shown in c. This figure is obtained by modifying Fig. 3 of the paper, K.G.
Nakamura et al., Physical Review B 92, 144304 (2015)

6.1.5.2 Numerical Calculation: Pulse Width Dependence

Figure 6.3 shows the numerical results of the time evolution of the mean value of
the phonon amplitude and its pulse width dependence. We examined three pulse
width conditions (0.1T , 0.55T , and 0.9T ), where T is the vibrational period of the
optical phonon. For Q A (t), the approximate formula (6.71) agrees well with the
numerical results in the region of time after the passage of the pulse (delay larger than
T ). Violent oscillations of the coherent phonons change to a gradual adaption of a
new equilibrium as the pulse width becomes large. The phonon amplitude decreases
as the pulse width increases. On the other hand, for Q R (t), the phonon amplitude
at the pulse width of 0.55T is larger than that at 0.1T and 0.9T .
In order to discriminate experimentally between Q A (t) and Q R (t), it is nec-
essary to use the electronic-state-selective measurement of the coherent phonons.
100 6 Coherent Phonons: Quantum Theory

Fig. 6.4 The dependence of

1.0

Amplitude of oscillation (Q0)

the pulse width on the
phonon amplitudes Q A (t)
(solid curve) and Q R (t) 0.8
(dotted curve). This figure is
obtained by modifying Fig. 4 0.6
of the paper, K.G. Nakamura
et al., Physical Review B 92, 0.4
144304 (2015)
0.2

0.0
0.0 0.4 0.8 1.2 1.6
Pulse width (T )

However, in the commonly used transient reflectivity or transmissivity measure-

ments, the electronic state is not identified. We get information only on the total
value Q(t) = Q A (t) + Q R (t), which is shown in Fig. 6.3c. Q(t) shows an
oscillation with the frequency ω, which is almost cosine-like at short pulse width
such as 0.1T . As the pulse width increases, the contribution from the ISRS process
increases, and the initial phase of the oscillation changes.
In Fig. 6.4, the phonon amplitudes are plotted for IA (solid curve) and ISRS
(dotted curve) as a function of the pulse width (FWHM). The phonon amplitude by
ISRS process increases with an increase of the pulse width, and reaches maximum
at 0.55T and decreases at a larger pulse width. The dominant process changes from
IA to ISRS around a pulse width of 0.55T .

6.1.5.3 Nonresonant Condition: Detuning Effect

We consider detuning effects on the coherent phonon generation, when a frequency

of the optical pulse does not match the energy gap between electronic two levels.
We restrict small α value (α 1) and neglect the perturbation terms higher than the
second order. The detuning is defined by ξ ≡ (ε − Ω)/.
For the impulsive absorption process, the phonon amplitude is obtained analyti-
cally as a similar way for the resonant case:
    
−σ 2 ξ 2 −σ 2 (ω2 + 2ξ ω)
Q A (t) = Q 0 exp exp cos(ωt) − 1 . (6.84)
2 4

This shows that the phonon amplitude decreases as the detuning increases.
For the ISRS process, the calculation is a little bit complicated. The phonon
amplitude is expressed by
6.1 Generation Mechanism with Displaced Harmonic Oscillator 101
 ∞
1
Q R (t) = √ e−σ ω /8 e−iωt due−u /2σ
2 2 2 2

2π 0
 
× e−iωu/2 − e−iωu/2 e−iξ u + c.c. (6.85)

The integral term gives a complex number, then an initial phase of the phonon-
amplitude oscillation is dependent on the detuning. Here, we evaluate the large
detuning case (ξ σ 1) compared to the pulse width. In this case, the real part
of the integral term is negligibly small. For the positive value of η, the integral is
calculated by
 ∞     
u2 π σ 2 η2
du exp − 2 − iηu = σ exp −
0 2σ 2 2
   σ η/√2
√ σ η2
2
2
− i 2σ exp − et dt. (6.86)
2 0

The real part is smaller than the imaginary part value for the ησ 1 case. Then, we
can get

Q R (t) = B sin(ωt), (6.87)

where
 
√ −σ 2 ω2   ω  ω 
B ≡ Q 0 2 exp D ξ+ −D ξ− . (6.88)
8 2 2

D(x) is the Dawson function and expressed by

  √
σ x/ 2
−σ 2 x 2 2
D(x) ≡ exp et dt. (6.89)
2 0

For the large x case, D(x) is expanded with x −1 as D(x) ∼ a0 + a1 x −1 + a2 x −2 +

· · · , and dD(x)/dx is expressed by

d σ
D(x) = −σ 2 x D(x) + √ . (6.90)
dx 2

These equations give

1
D(x)  √ . (6.91)
2σ x
102 6 Coherent Phonons: Quantum Theory

Finally, we get the approximate form of B:

  
2 −σ 2 ω2 ω
B  −Q 0 exp . (6.92)
π 8 σξ2

6.1.5.4 Numerical Calculation: Detuning Dependence

Figure 6.5 shows the numerical results of the time evolution of phonon amplitude
Q A (t), Q R (t), and Q(t) with a pulse width of 0.1T with detuning of ΔE −
0, 3ω, 5ω, and 10ω. The phonon amplitude Q A (t) approaches to zero as the
detuning increases. Q R (t) starts to move in the same direction as Q A (t) at large
detuning.

0.2 (a)
0.1
0.0
Mean value of phonon coordinate (Q0 )

-0.1
-0.2
(b)
0.0
-0.5
-1.0
-1.5
-2.0
(c)
0.0
-0.5
-1.0
-1.5
-2.0
0.0 0.5 1.0 1.5 2.0 2.5
Delay (T )

Fig. 6.5 Time evolution of the phonon amplitude with detuning ΔE of 0ω (solid curves: no detun-
ing), 3ω (dotted curves), and 5ω (dashed curves). a represents Q R (t), b represents Q A (t),
and c represents Q(t). T is the vibrational period of the phonon. The pulse width was set to 0.1T .
This figure is obtained by modifying Fig. 5 of the paper, K.G. Nakamura et al., Physical Review B
92, 144304 (2015)
6.1 Generation Mechanism with Displaced Harmonic Oscillator 103

Fig. 6.6 The detuning 1

dependence of the phonon

Amplitude of oscillation (Q0 )

4
amplitude for Q A (t) (solid
curve) and Q R (t) (dotted 2

curve). The pulse width was 0.1

0.1T . This figure is obtained 4
by modifying Fig. 6 of the
2
paper, K.G. Nakamura et al.,
Physical Review B 92, 0.01
144304 (2015) 4
2

0.001
0 2 4 6 8 10
Detuning ( )

Figure 6.6 shows the phonon amplitude as a function of the detuning. Both the
phonon amplitudes Q R (t) and Q A (t) decrease as the detuning increases. For
large detuning, the phonon amplitude Q R (t) becomes larger than Q A (t) since no
light absorption occurs, and the dominant generation process of the coherent phonons
is subject to the ISRS process.

6.2 Four-Level Model and Double-Sided Feynman

Diagrams

In a weak coupling (α < 1) case, the generation of coherent phonons can be cal-
culated using a four-level system consisting of two electronic states (|g and |e)
and zero-phonon and one-phonon states (|0 and |1). There are four levels: |g|0,
|g|1,|e|0, and |e|1. The four-level model is easily used to calculate dynamics
compared to that of the displaced harmonic oscillators (Sect. 6.1), while the both
show the same results in a weak coupling case.

6.2.1 Impulsive Absorption Process

Transition processes are represented using double-sided Feynman diagrams. Figure

6.7 shows an impulsive absorption process with the |g, 0 → |e, 1 and g, 0| →
e, 0| transitions. The final state (|e, 1e, 0|) is in the electronic excited state with
vibrational polarization. We set the initial state |g, 0g, 0| at t = −∞. At time
t  , |g, 0 transitions to |e, 1 via the dipole interaction with the incident pulse and
104 6 Coherent Phonons: Quantum Theory

Fig. 6.7 Diagram of the

impulsive absorption
process. Time flows from the
left to the right. The upper
part of the figure shows an
envelope of the optical pulse.
Interaction between the
optical pulse and the system
occurs at time t  and t 

the electron–phonon coupling (α b̂† ). The total interaction causes the multiplicative
coefficient
   
μE(t  ) μE 0 
α =α f (t  )e−iΩt . (6.93)
i i

At time between t  and t  , the state |e, 1g, 0| has a time evolution factor of
 
e−i(ε/+ω)(t −t ) . (6.94)

At time t  , g, 0| transfers to e, 0| via the dipole interaction with the incident pulse.
This interaction causes the multiplicative coefficient
   
μE ∗ (t  ) μE 0 
= f (t  )eiΩt . (6.95)
−i −i

After time t  , the state |e, 1e, 0| propagates with a factor of


e−iω(t−t ) . (6.96)

Through these processes, we get the time evolution of the density matrix by
 
  μE 0 −iΩt   
ρ1 (t : t , t ) = α e f (t  )e−i(ε/+ω)(t −t )
i
 
μE 0  
× f (t  )eiΩt e−iω(t−t ) |e, 1e, 0|
−i
 
μE 0 2   
=α f (t  ) f (t  )e−iΔ(t −t ) e−iω(t−t ) |e, 1e, 0|, (6.97)

6.2 Four-Level Model and Double-Sided Feynman Diagrams 105

where Δ = ε/ − Ω. By integrating ρ(t : t  , t  ) over t  and t  , we get

    t
μE 0 2 t   
ρ1 (t) = α dt  dt  f (t  ) f (t  )e−iΔ(t −t ) e−iω(t−t )
 −∞ −∞
× |e, 1e, 0|. (6.98)

Another path is g, 0| → e, 0| at time t  and |g, 0 → |e, 1 at time t  . Its density
matrix is given by
    t
μE 0 2 t   
ρ2 (t) = α dt  dt  f (t  ) f (t  )eiΔ(t −t ) e−iω(t−t )
 −∞ −∞
× |e, 1e, 0|. (6.99)

In addition to these two paths, their Hermitian conjugate paths are possible. The total
density matrix ρ(t) is obtained by ρ(t) = ρ1 (t) + ρ2 (t) + H.c.
At the resonance condition (Δ = 0), the density matrix is expressed as
    t
μE 0 2 t 
ρ(t) = α dt dt  f (t  ) f (t  )
 −∞ −∞
  

× e−iω(t−t ) + e−iω(t−t ) |e, 1e, 0|
  
 
+ eiω(t−t ) + eiω(t−t ) |e, 0e, 1| . (6.100)

In the absorption processes, phonons are excited in the electronic excited state, in
which the potential energy is expressed by the displaced harmonic oscillator, and the
annihilation and creation operators are expressed by b̂ − α and b̂† − α, respectively.
The phonon coordinate operator Q̂ A is expressed by

  
Q̂ A = b̂ + b̂† − 2α . (6.101)
2ω

Therefore, the |e, 0e, 0| state also contributes to the shifted position, and the density
operator ρ  (t) for the |g, 0g, 0| → |e, 0e, 0| transition is
 2    
 μE 0 t

t
   
ρ (t) = dt dt  f (t  ) f (t  ) e−iΔ(t −t ) + e−iΔ(t −t ) |e, 0e, 0|
 −∞ −∞
 2  
μE 0 t t
= dt  dt  f (t  ) f (t  )|e, 0e, 0|. (6.102)
 −∞ −∞
106 6 Coherent Phonons: Quantum Theory

The expectation value of the phonon amplitude  Q̂ A  is calculated as

 Q̂ A (t) = T r ((ρ(t) + ρ  (t)) Q̂ A )

    t
μE 0 2 t 
=α dt dt  f (t  ) f (t  )
 −∞ −∞
 
× cos(ω(t − t  )) + cos(ω(t − t  )) − 2 . (6.103)

This is consistent with the result (6.49) with ε = ω.

6.2.2 Impulsive Stimulated Raman Scattering Process

Next, we consider the ISRS process for exciting the coherent phonons in the electronic
ground state.
Figure 6.8a and b show the double-sided Feynman diagrams of the impul-
sive stimulated Raman scattering process with the |g, 0 → |e, 1 → |g, 1 and
|g, 0 → |e, 0 → |g, 1 transitions, respectively. The final state (|g, 1g, 0|) is in
the electronic ground state with vibrational polarization. We set the initial state
|g, 0g, 0| at t = −∞. In the path shown by Fig. 6.8a, at time t  , |g, 0 changes
to |e, 1 via the dipole interaction with the incident pulse and the electron–phonon
coupling (α b̂† ). The total interaction causes the multiplicative coefficient
   
μE(t  ) μE 0 
α =α f (t  )e−iΩt , (6.104)
i i

Fig. 6.8 Diagram of the

impulsive stimulated Raman
scattering process. a passes
through |e, 1 state and b
passes through the |e, 0
state. Time flows from the
left to the right. The upper
part of the figure shows an
envelope of the optical pulse.
Interaction between the
optical pulse and the system
occurs at time t  and t 
6.2 Four-Level Model and Double-Sided Feynman Diagrams 107

which is the same as that in the IA process (Fig. 6.7). At time between t  and t  , the
state |e, 1g, 0| has a time evolution factor of
 
e−i(ε/+ω)(t −t ) . (6.105)

At time t  , |e, 1 changes to |g, 1 via the dipole interaction with the incident pulse.
This interaction causes the multiplicative coefficient
   
μE ∗ (t  ) μE 0 
= f (t  )eiΩt , (6.106)
i i

because the light goes out from the material system. After time t  , the state |g, 1g, 0|
evolves with a factor of

e−iω(t−t ) . (6.107)

Through these processes, we get the time evolution of the density matrix as

μE 0 −iΩt   
ρ1 (t : t  , t  ) = α e f (t  )e−i(ε/+ω)(t −t )
i
 
μE 0  
× f (t  )eiΩt e−iω(t−t ) |g, 1g, 0|
i
 
μE 0 2   
= −α f (t  ) f (t  )e−iΔ(t −t ) e−iω(t−t ) |g, 1g, 0|.

(6.108)

By integrating ρ(t : t  , t  ) over t  and t  , we get

    t
μE 0 2 t    
ρ1 (t) = −α dt dt  f (t  ) f (t  )e−iΔ(t −t ) e−iω(t−t )
 −∞ −∞
× |g, 1g, 0|. (6.109)

Another path (Fig. 6.8b) is |g, 0 → |e, 0 at time t  and |e, 0 → |g, 1 at time t  .
At time t  , |g, 0 changes to |e, 0 via the dipole interaction with the incident pulse.
The interaction causes the multiply coefficient
   
μE(t  ) μE 0 
= f (t  )e−iΩt . (6.110)
i i

At time between t  and t  , the state |e, 0g, 0| has a time evolution factor of
 
e−i(ε/)(t −t ) . (6.111)
108 6 Coherent Phonons: Quantum Theory

At time t  , |e, 0 changes to |g, 1 via the dipole interaction with the incident pulse and
the electron–phonon coupling. This interaction causes the multiplicative coefficient
   
μE ∗ (t  ) μE 0 
−α = −α f (t  )eiΩt . (6.112)
i i

It is worth noting that the electron–phonon coupling interaction is −α b̂ for the tran-
sition from the electronic excited to the ground state, because the ground-state har-
monic potential shifts in the −α direction. After time t  , the state |g, 1g, 0| evolves
with a factor of

e−iω(t−t ) . (6.113)

Through these processes, we get the time evolution of the density matrix as
 
  μE 0 −iΩt   
ρ2 (t : t , t ) = e f (t  )e−i(ε/)(t −t ) (−α)
i
 
μE 0  
× f (t  )eiΩt e−iω(t−t ) |g, 1g, 0|
i
 
μE 0 2   
=α f (t  ) f (t  )e−iΔ(t −t ) e−iω(t−t ) |g, 1g, 0|. (6.114)


By integrating ρ2 (t : t  , t  ) over t  and t  , we get

    t
μE 0 2 t    
ρ2 (t) = α dt dt  f (t  ) f (t  )e−iΔ(t −t ) e−iω(t−t )
 −∞ −∞
× |g, 1g, 0|. (6.115)

In addition to these two paths, their Hermitian conjugate paths are possible. The total
density matrix ρ(t) is obtained by ρ(t) = ρ1 (t) + ρ2 (t) + H.c. At the resonance
condition (Δ = 0), the density matrix ρ(t) is obtained to be
 2   t
μE 0 t

ρ(t) = α dt dt  f (t  ) f (t  )
 −∞ −∞
 
−iω(t−t  ) −iω(t−t  )
× e −e |g, 1g, 0| + H.c. (6.116)

In the stimulated Raman scattering process, phonons are excited in the electronic
ground state, and the phonon coordinate operator Q̂ R is expressed by

  
Q̂ R = b̂ + b̂† . (6.117)
2ω
6.2 Four-Level Model and Double-Sided Feynman Diagrams 109

The expectation value of the phonon amplitude  Q̂ R  is calculated as

 Q̂ R (t) =T r (ρ(t) Q̂ R )
     t
 μE 0 2 t  
=2α dt  dt  f (t  ) f (t  ) cos(ω(t − t  )) − cos(ω(t − t  )) .
2ω  −∞ −∞
(6.118)

This is the same as (6.63).

6.3 Extension to Band Model

The two-level system for the electronic state is extended to a two band model, in
which electronic excited states with a different wave vector k have different energy
levels [11]. The one-dimensional harmonic potentials are also assumed for the optical
phonons. The Hamiltonian is expressed as

Ĥ0 = Ĥg |gg| + Ĥk |kk|
k

Ĥg = ωb̂† b̂
Ĥk = εk + ωb̂† b̂ + αω(b̂ + b̂† ), (6.119)

where |k is the state for which an electron with wave vector k is excited from |g to
the conduction band with the excitation energy of εk . The creation and annihilation
operators of the LO phonon at the Γ -point with energy ω are denoted b̂† and b̂,
respectively. The deformation potential interaction having dimensionless coupling
constant α, which is independent of the k vector, is assumed. Using the rotating wave
approximation, the interaction Hamiltonian between pump pulse and the electronic
state is given by
  
ĤI (t) = μk E 0 f (t) e−iΩt |kg| + eiΩt |gk| , (6.120)
k

in which μk is the transition dipole moment from |g to |k and f (t) is the envelope
of the pulse. A similar calculation used for the two-level model is used to solve the
Schrödinger equation using Hamiltonians for the two-band model (Fig. 6.9).

6.4 Optical Detection Mechanism

The pump–probe transmission (or reflection measurement) can be treated as a third-

order nonlinear optical response with a heterodyne detection. Here, we use the four-
level model consisting of two electronic states (|g and |e) and two phonon states (|0
110 6 Coherent Phonons: Quantum Theory

Fig. 6.9 A schematic of the V(Q)

potential for the two-band
model with displaced
harmonic oscillators

band

and |1) under transparent conditions. The pump pulse generates phonon polarization
in the electronic ground state by ISRS and the second-order density operator is given
by
 2   t
(2) μE 0 t

ρ (t) = α dt dt  f (t  ) f (t  )
 −∞ −∞
 
−iω(t−t  ) −iω(t−t  )  
× e −e e−iΔ(t −t ) |g, 1g, 0| + H.c., (6.121)

where Δ ≡ ε − Ω is the detuning. By using a Gaussian pulse envelope with pulse

width σ ,
 2
1 t
f (t) = √ exp − 2 , (6.122)
πσC σ
√
where C is a dimensionless normalization factor ( π σ C = 1). Following the cal-
culation of Sect. 6.1.5, we get

|μ|2 |E 0 |2 −iωt ω
ρ (2) (t) = iα e−σ ω /8 |g, 1g, 0|,
2 2
e √ (6.123)
 2
2σ C Δ
2 2

and its Hermite conjugate. The expectation value of the excited coherent optical
phonons Q can be calculated as
 
(2) 
Q(t) = T r [Qρ (t)] = T r (b̂ + b̂† )ρ (2) (t)
2ω


= A|E 0 |2 sin(ωt), (6.124)
2ω
6.4 Optical Detection Mechanism 111

Fig. 6.10 Schematic of the pump–probe heterodyne detection of transient transmission. E pr (t)
is the electronic field of the probe pulse and Ps (t) is the induced polarization. The dotted arrow
indicates dipole radiation from the induced polarization. PD is a photo detector

where, we set
μ2 ω
e−σ ω /8 .
2 2
A=α √ (6.125)
 2
2σ C Δ
2 2

The phonon amplitude is proportional to the intensity of the pump pulse. The dis-
placement of atoms defined by the phonon amplitude can not be directly detected
by the optical response, while it can be detected by using the X-ray diffraction. In
the pump and probe protocol, the probe pulse induces electronic polarization Ps (t),
which causes dipole radiation. When a heterodyne detection of the transmitted probe
pulse is investigated, the detection intensity Ih (t) is

Ih (t) = Ωl × Im(E 3∗ (t)Ps (t)), (6.126)

where l is the thickness of the sample [10], E 3∗ (t) is the strength of the electronic
field of the probe pulse, and Im(A) is to get a imaginary part of A. Figure 6.10 shows
a schematic drawing of the experimental configuration.
The probe pulse irradiates the sample at time t p , which is a delay between the
pump and probe pulses. There are eight pathways (Fig. 6.11) as given below:

path1 : |g, 1g, 0| → |e, 1g, 0| → |g, 0g, 0|

path2 : |g, 1g, 0| → |e, 0g, 0| → |g, 0g, 0|
path3 : |g, 1g, 0| → |g, 1e, 1| → |g, 1g, 1|
path4 : |g, 1g, 0| → |g, 1e, 0| → |g, 1g, 1| (6.127)

and their Hermite conjugates. The third-order density operator for the path 1 is
obtained by

iμ
ρ1(3) (t  ) = i A|E 0 |2 E 3 e−iωt p

 t
   
× dt  f 3 (t  )e−iωt e−iΩt e−i(ε+ω)(t −t )/ |e, 1g, 0|, (6.128)
−∞
112 6 Coherent Phonons: Quantum Theory

Fig. 6.11 The double-sided

Feynman diagrams for a
detection scheme in the
pump and probe
measurement

where f 3 (t  ) is the Gaussian pulse envelope of the probe pulse and t p is the pump–
probe delay. The polarization (P(t)) induced by the probe pulse is obtained using
the polarization operator P̂ op ≡ μ|ge| + μ∗ |eg| by P(t) = T r (ρ (3) (t) P̂ op ). The
polarization for the path 1 is obtained as

μ2
P1 (t  ) = α A|E 0 |2 E 3 e−iωt p

 t
   
× dt  f 3 (t  )e−iωt e−iΩt e−i(ε+ω)(t −t )/ |e, 1g, 0|, (6.129)
−∞

and the time-integrated intensity (I1 (t p )) of the product between the probe light and
polarization is as
 ∞
I1 (t p ) = E 3 f 3∗ (t  )P1 (t  )dt 
−∞
μ2
= α A|E 0 |2 |E 3 |2 e−iωt p

 ∞  t
 
× dt  dt  f 3 (t  )eiΩt f 3 (t  )e−iωt
−∞ −∞
−iΩt  −i(ε+ω)(t  −t  )/
×e e . (6.130)
6.4 Optical Detection Mechanism 113

For the Gaussian pulse, we get

μ2
I1 (t p ) = α A|E 0 |2 |E 3 |2 e−iωt p
 ∞ π σ 2C 2
 ∞
e−2s /σ e−iωs due−u /(2σ ) ei(Δ−ω/2)u
2 2 2 2
×
−∞ 0
√
μ |E 0 | |E 3 | −iωt p π −σ 2 ω2 /8 iα A
2 2 2
≈ e √ e , (6.131)
π σ 2 C 2 2σ Δ − ω/2

where s = (t  + t  )/2 and u = t  − t  . This is simply expressed by

1
I1 (t p ) = i B|E 0 |2 |E 3 |2 eiωt p , (6.132)
Δ − ω/2

where
√
π ωα 2 μ2 −ω2 σ 2 /4
B= e . (6.133)
2σ 2 C 4 Δ2 3

A similar calculation shows that I4 (t p ) = I1 (t p ) and

1
I2 (t p ) = I3 (t p ) = i B|E 0 |2 |E 3 |2 e−iωt p . (6.134)
Δ + ω/2

Thus, the total intensity is

 2i Bω
I (t p ) = Ii (t p ) + H.c. = |E 0 |2 |E 3 |2 e−iωtt p + H.c.
i
Δ2
4Bω
= |E 0 |2 |E 3 |2 sin(ωt p ). (6.135)
Δ2
The results show that the change in transmission is proportional to the pump and
probe intensity and proportional to the phonon amplitude Q(t  ) by comparing
(6.124).

6.5 Summary

The generation processes of the coherent optical phonons were described quan-
tum mechanically by using the two electronic levels and the displaced harmonic
oscillators. Two generation processes (IA and ISRS) were investigated. Using the
second-order perturbation, the mean value of the phonon amplitude was calculated
for both the IA and ISRS processes. Detailed calculation was done with the Gaussian
114 6 Coherent Phonons: Quantum Theory

pulse for the Fourier-transform-limited case. The phonon oscillation is cosine-like

and sine-like for the IA and ISRS processes, respectively. Both the IA and ISRS pro-
cesses coexist in quantum mechanics. The ratio between the IA and ISRS processes is
dependent on the pulse width and detuning. For the small deformation case (α 1),
both the IA and ISRS processes are approximately expressed using the four-level
model. The detection process of the EO sampling was also calculated using the same
model.

References

1. Scholz, R., Pfeifer, T., Kurz, H.: Density-matrix theory of coherent phonon oscillations in
germanium. Phys. Rev. B 24, 16229–16236 (1993)
2. Kuznetsov, A.V., Stanton, C.J.: Theory of coherent phonon oscillations in semiconductors.
Phys. Rev. Lett. 43, 3243–3246 (1994)
3. Hu, X., Nori, F.: Quantum phonon optics: coherent and squeezed atomic displacement. Phys.
Rev. B 53, 2419–2424 (1996)
4. Merlin, R.: Generating coherent THz phonons with light pulses. Solid State Commun. 102,
107–220 (1997)
5. Stevens, T.E., Kuhl, J., Merlin, R.: Coherent phonon generation and two stimulated Raman
tensors. Phys. Rev. B 65, 144304 (2002)
6. Nakamura, K.G., Shikano, Y., Kayanuma, Y.: Influence of pulse width and detuning on coherent
phonon generation. Phys. Rev. B92, 144304 (2015)
7. Pollard, W.T., Fragnito, H.L., Bigot, J.-Y., Shank, C.V., Mathies, R.A.: Quantum-mechanical
theory for 6 fs dynamic absorption spectroscopy. Chem. Phys. Lett. 168, 239–245 (1990)
8. Banin, U., Bartana, A., Ruthman, S., Kosloff, R.: Impulsive excitation of coherent vibrational
motion ground surface dynamics induced by intense short pulse. J. Chem. Phys. 101, 8461–
8481 (1994)
9. Cerullo, G., Manzoni, C.: Time domain vibrational spectroscopy: principle and experimental
tools. In: De Silvestri, S., Cerullo, G., Lanzani, G. (eds.) Coherent Vibration Dynamics. CRC
Press, Boca Raton (2008)
10. Mukamel, S.: Principles of Nonlinear Optical Spectroscopy. Oxford University Press, New
York (1995)
11. Nakamura, K.G., Ohya, K., Takahashi, H., Tsuruta, T., Sasaki, H., Uozumi, S., Norimatsu, K.,
Kitajima, M., Shikano, Y., Kayanuma, Y.: Spectrally resolved detection in transient-reflectivity
measurements of coherent optical phonons. Phys. Rev. B94, 024303 (2016)
Chapter 7
Coherent Control of Optical Phonons

This chapter describes the coherent control of optical phonons. We investigate, as

an example, experiments and theoretical treatment of the coherent control of optical
phonons in diamond, because it is one of the simplest systems. A four-level model
consisting of two electronic states with two phonon states is used to calculate the
coherent control of the optical phonons. Several selected examples of coherent control
experiments of optical phonons are also presented.

7.1 Coherent Control

Coherent control is a technique to control quantum states using laser light and was
originally developed for controlling chemical reactions using coherent two-photon
process to assist chemical reactions via electronic excited states [1–3]. It has been
performed for other physical properties, for example, electronic, vibrational, and
rotational states of atoms and molecules, and electrons, excitons, spins, and phonons
in condensed matter [4–8]. Here, we describe the coherent control of optical phonons
using multiple ultrashort pulses.

7.2 Experiment of Coherent Control of Optical Phonons

The coherent control of optical phonons was first demonstrated on α-perylene molec-
ular crystals at 5 K using multiple femtosecond pulses by Weiner et al. [9]. They
irradiated a femtosecond pulse train consisting of more than 10 pulses, which was
generated through a pulse-shaping technique, on the sample for exciting the opti-
cal phonons via the ISRS process in a transparent condition. They controlled the
oscillation amplitude of the specific phonon mode to be approximately 5 THz owing
to the frequency of the pulse train. A few years later, the coherent control of lon-
gitudinal optical (LO) phonon dynamics was demonstrated in a single crystal of a
© Springer Nature Switzerland AG 2019 115
K. Nakamura, Quantum Phononics, Springer Tracts in Modern Physics 282,
https://doi.org/10.1007/978-3-030-11924-9_7
116 7 Coherent Control of Optical Phonons

semiconductor (gallium arsenide) in an opaque condition using a pair of femtosec-

ond pulses at room temperature [10]. The first pump pulse generated 8.75 THz LO
phonons by an ultrafast screening of the surface-space-charge field. The amplitude of
the phonons was enhanced or reduced by the second pulse when the delay between
the two pump pulses matched an integer or a half-integer multiple of the oscilla-
tion period through constructive or destructive interference, respectively. Similar
coherent control experiments with double femtosecond pulses were performed on
semimetal (bismuth) films [11] by a displacive excitation mechanism. The amplitude
dependence on the delay between the two pump pulses was well explained using a
combination of two damped oscillations. In addition to the amplitude control, the
phase change in the controlled phonons was reported and its behavior deviated from
a linear dependence, which was predicted by theoretical calculations. The coherent
control of optical phonons is currently being used for selective excitation [9, 12, 13]
and control of a structural transformation [14].

7.2.1 Coherent Control of Optical Phonons in Diamond

The coherent control of optical phonons in diamond is given as an example of coherent

control using a pair of ultrashort optical pulses. The experimental setup used for this
experiment is the same described in Fig. 5.1. The sub-10-fs optical pulse is separated
into two pulses by using a beam splitter. One is used for the pump pulse and the other
for the probe pulse. The pump pulse is directed to a Michelson-type interferometer
(as shown in Fig. 7.1), and a pair of pump pulses (pump 1 and pump 2) is generated.
Relative delay between pump 1 and pump 2 is precisely controlled by controlling
difference between their optical path lengths.

Fig. 7.1 Schematic of the

Michelson-type
interferometer, which is used
to generate a pair of
femtosecond pulses. BS is a
beam splitter with 1:1 ratio
7.2 Experiment of Coherent Control of Optical Phonons 117

Fig. 7.2 Transient

transmittance measurement
in femtosecond laser pumped (a) 238.2 fs
diamond using a pair of
10
pump pulses for the
(b) 242.5 fs
pump–pump delay of 283.2

-6
fs (a), 242.5 fs (b), 250.9 fs

Transmission change (ΔT/T0) 10

(c), 258.7 fs (d), and 263.7 fs (c) 250.9 fs
(e). This figure is obtained
by modifying Fig. 3 of the 0
paper, H. Sasaki et al., (d) 258.7 fs
Scientific Reports 8, 9609
(2018)
(e) 263.7 fs
-10

-20

-30
-400 -200 0 200 400 600 800
Delay between pump and probe pulses (fs)

Figure 7.2 shows transient transmittance pumped by a pair of femtosecond optical

pulses. The coherent oscillation due to optical phonons induced by the pump 1 is
enhanced or suppressed by the pump 2 at delay 250.9 fs or 263.7 fs, respectively.
The oscillation amplitude and its initial phase after the pump 2 irradiation are
shown in Fig. 7.3. The phonon amplitude is enhanced and suppressed at integer
multiples and half-integer multiples of the vibrational period (25 fs).

7.3 Coherent Control Theory

7.3.1 Transparent Condition

At first, we consider the coherent control of optical phonons in diamond, which is

described above. Phonon excitation processes by two pump pulses are expressed
by the following double-sided Feynman diagrams (Fig. 7.4) for the second-order
perturbation. The first diagram (a) indicates the phonons that are generated by the
pump 1. The electronic excitation and relaxation occur with the pulse. In a similar
way, the phonons are generated by the pump 2 in the second diagram (b). The third
diagram indicates that the excitation occurs in the pump 1 and the relaxation occurs
118 7 Coherent Control of Optical Phonons

Fig. 7.3 The amplitude (a) (a)

2.0

Amplitude (normalized)
and phase (b) of the
controlled oscillation after
the pump 2 against the 1.5
pump–pump delay τ . The
amplitude is normalized 1.0
using that obtained after
excitation by the pump 1. 0.5
The solid circles are the
experimental data, and the
0.0
solid curves are obtained by
the theoretical calculation.
This figure is obtained by
0.2 (b)
modifying Fig. 4 of the paper,
0.0
Initial phase (π)

H. Sasaki et al., Scientific

Reports 8, 9609 (2018) -0.2

-0.4

-0.6

-0.8
230 240 250 260
Delay between pump pulses (fs)

in the pump 2. The third process is important only when the two pump pulses overlap
for the nonresonant condition (transparent condition). All processes are within the
second-order perturbation. The density operator for the second diagram (ρ2(2) ) is the
same as the first one (ρ1(2) ), which is shown in Chap. 6, except for the time delay:

ρ (2) (t) = ρ1(2) + ρ2(2)

 
= i A |E 1 |2 e−iωt + |E 2 |2 e−iω(t−τ ) |g, 1g, 0| + H.c., (7.1)

where E 1 and E 2 are the electric field amplitude of the pump 1 and pump 2, τ is the
delay between pump 1 and pump 2, and

μ2 ω
e−σ ω /8 .
2 2
A=α √ (7.2)
2 2C 2 Δ2

The mean value of the phonon amplitude Q(t) is calculated as


  
Q(t) = A |E 1 |2 sin(ωt) + |E 2 |2 sin(ω(t − τ )) . (7.3)
2ω

The result shows that the phonon amplitude after the second pump pulse irradiation
is a sum of two sinusoidal functions induced by each pulse. The phonon amplitude
and initial phase at the timing of irradiation of pump 2 are calculated for diamond
and shown in Fig. 7.3 as solid curves.
7.3 Coherent Control Theory 119

Fig. 7.4 Diagram of the impulsive stimulated Raman scattering paths. a passes through |e, 1 state
and b passes through the |e, 0 state. Time flows from the felt to the right. The upper part of the
figure shows an envelope of the optical pulse. Interaction between the optical pulse and the system
occurs at time t and t

It is worth noting that the above coherent control is due to constructive or destruc-
tive interference between phonon states excited by the pump 1 and pump 2. This is
because we cannot distinguish which pulse excites the phonons while keeping the
phonon coherence. The phonons are excited by the second-order perturbation, which
interacts with the light twice in one pulse. This is different from the classical control
of pendulum oscillations with two impacts by two pulses.

7.3.1.1 Short-Delay Region

When the delay between the pump 1 and pump 2 is very short, the third path repre-
sented as (c) in Fig. 7.4 is important. The density operator is calculated as

 μ 2  t  t
ρ3(2) (t, τ ) = −α e−iΩτ e−iωt dt  dt  f 1 (t  ) f 2 ( f  − τ )
 −∞ −∞
   
ε − Ω  
× exp −i (t − t  ) eiωt − eiωt |g, 1g, 0| + H.c. (7.4)


When the two pulses overlap, there is another transition path, in which the excitation
and relaxation are induced by the pump 2 and pump 1, respectively. The density
operator of this path is
120 7 Coherent Control of Optical Phonons

 μ 2  t  t
ρ4(2) (t, τ ) = −α e iΩτ −iωt
e dt  dt  f 1 (t  ) f 2 ( f  − τ )
 −∞ −∞
   
ε − Ω  
× exp −i (t − t  ) eiωt − eiωt |g, 1g, 0| + H.c. (7.5)


With ρ3(2) (t, τ ) and ρ4(2) (t, τ ) included, the expectation value of the phonon amplitude
should be modified with the optical frequency Ω. This oscillation can be observed
when the delay between two pump pulses is controlled with a step shorter than the
optical cycle.

7.3.2 Opaque Condition

In the opaque conditions, the impulsive absorption paths are possible in addition to
the ISRS paths described above. The final state should be |e, 1e, 0| or |e, 0e, 1|.
The paths (a) and (b) are the same as the single pulse excitation described in
Chap. 6. The density operators for the path (c) and (d) in Fig. 7.5 are

Fig. 7.5 Diagrams of the coherent control of optical phonons with the impulsive absorption paths.
Transition occurs by only the pump 1 (a) or the pump 2 (b). (c) and (d) represent the transition
occurring by both the pump 1 and pump 2. Time flows from the felt to the right. The upper part
of the figure shows an envelope of the optical pulse. Interaction between the optical pulse and the
system occurs at time t and t
7.3 Coherent Control Theory 121

 μ 2  t  t
ρ3(2) (t, τ ) = α e−iΩτ dt  dt  f 1 (t  ) f 2 ( f  − τ )
 −∞ −∞

ε − Ω 
× exp −i (t − t  )

  

× e−iω(t−t ) |e, 1e, 0| + eiω(t−t ) |e, 0e, 1| , (7.6)

and
 μ 2  t  t
ρ4(2) (t, τ ) = α eiΩτ dt  dt  f 1 (t  ) f 2 ( f  − τ )
 −∞ −∞

ε − Ω 
× exp −i (t − t  )

  

× e−iω(t−t ) |e, 1e, 0| + eiω(t−t ) |e, 0e, 1| . (7.7)

This shows that the phonon amplitude is also modulated by the optical frequency.

7.4 Selected Examples

There are many applications of the coherent control of optical phonons such as the
mode-selective excitation and phase-transition control. We describe several examples
of the coherent control experiments of the coherent phonons.

7.4.1 Mode-Selective Excitation

YBa2 Cu3 O7−δ is a typical example of a high-temperature transition superconductor

and has two optical phonons, Ba–O and Cu–O modes, at frequencies of 3.4 and 4.3
THz, respectively. Both phonon modes are excited by an ultrashort pulse. By using a
double-pulse excitation, we can selectively excite either the Ba–O mode phonons or
the Cu–O mode phonons, because their vibrational periods are different: 294.1 and
232.6 fs for the Ba–O and Cu–O modes, respectively. Takahashi et al. [12] selectively
excited the phonon modes using a pair of femtosecond pulses and showed that the
Ba–O and Cu–O modes are strongly suppressed at the double-pulse separation time
of 135.0 and 108.5 fs, which are close to a half-integer multiple of the phonon periods.
At the separation time of 135.0 fs, only the Cu–O mode oscillation was observed.
The phonon amplitude of the Ba–O and Cu–O modes are enhanced at separation
time of 270 and 217 fs, respectively.
Katsuki et al. [13] demonstrated all-optical control and visualization of ultra-
fast two-dimensional atomic motions in a single crystal of bismuth using a pair of
chirped femtosecond pulses. Bismuth has two optical phonons A1g and E g modes
122 7 Coherent Control of Optical Phonons

with different frequencies of 3 and 2 THz, respectively. Both phonon modes are
excited by an ultrashort optical pulse and detected via the EO sampling technique.
In the A1g and E g modes, the Bi atoms oscillate in the longitudinal and lateral direc-
tions, when the (0001) surface is used. The relative intensity ratio between the A1g
and E g modes, which corresponds to the displacement direction of atoms in the
two-dimensional space, was controlled by controlling the delay between two pump
pulses. The reflectivity change shows beat structures, which are produced by overlap-
ping two oscillations, whose temporal evolution changes drastically as they change
the delay between two pump pulses on the attosecond timescale. They mapped these
beat structures into a two-dimensional space atomic displacement with the density
functional theory calculation of the reflectivity change.

7.4.2 Mode Coupling

Not only the LO phonons (ω L /2π = 8.7 THz) but also the optical phonon–plasmon
coupled (LOPC) mode oscillations (ω− /2π = 7.7 THz) are coherently controlled
by using a pair of pump pulses at delays between 315 and 345 fs [15]. In n-GaAs,
the oscillation amplitudes of the LO phonons and LOPC mode were harmonically
modulated according to their frequencies by the pump–probe delay. On the other
hand, in p-GaAs, the oscillation amplitude of the LOPC mode was modulated in
phase with that of the LO phonons at the period of LO phonons. They analyzed
the results using a phenomenological model by assuming that the LOPC formation
is delayed from the LO-phonon excitation. The lifetime of the LOPC plays a key
role in understanding different results in two samples. The lifetime of the LOPC
is ∼200 fs in p-GaAs, and ∼800 fs in n-GaAs. Therefore, when the second pump
pulse arrives at the sample, the LOPC excited by the first pump pulse has already
decayed significantly in p-GaAs. Thus, the LOPC excited by the second pump pulse
has nearly no interference with that excited by the first one, and the dynamics are
determined mainly by the interference of LO phonons. The excited LO phonons
cause the phonon–plasmon coupling. The delay time of the LOPC formation was
estimated to be 100 fs for n-GaAs and 120 fs for p-GaAs. In addition to the control
of phonon amplitudes, they reported that the lifetime of the LOPC was coherently
controlled, while that of LO phonons was independent of the pump–pulse delay.

7.4.3 Controlling Phase Transition

Ge2 Sb2 Te5 (GST) is a chalcogenide phase-change material, in which phase transfor-
mation between the amorphous and crystal phase is optically induced. Makino et al.
[16] demonstrated that the phase change from amorphous into crystalline states is
manipulated by controlling atomic motions through selectively exciting vibrational
modes that involve Ge atom using a pair of femtosecond pulses. The GST superlattice
7.4 Selected Examples 123

sample was irradiated by the two pump pulses with delay of 276 fs, which is close
to the time period of the local A1 mode of the octahedral GeTe6 (in crystalline
phase). The pump pulses coherently enhanced the vibrational amplitude toward the
crystalline phase.

7.5 Summary

We described the experiments and theoretical treatments of the coherent control of

optical phonons in diamond, because it is one of the simplest systems. A four-level
model consisting of two electronic states with two phonon states is used to calculate
the coherent control of the phonon amplitude. Several selected examples of coherent
control experiments of optical phonons are also presented.

References

1. Tannor, D.J., Rice, S.A.: Control of selectivity of chemical reaction via control of wave packet
evolution. J. Chem. Phys. 83, 5013 (1985)
2. Brumer, P., Shapiro, M.: Control of unimolecular reactions using coherent light. Chem. Phys.
Lett. 126, 541 (1986)
3. Tannor, D.J., Kosloff, R., Rice, S.A.: Coherent pulse sequence induce control of selectivity of
reactions: exact quantum mechanical calculations. J. Chem. Phys. 85, 5805 (1986)
4. Ramsay, A.J.: A review of the coherent optical control of the exciton and spin states of semi-
conductor quantum dots. Semicond. Sci. Techol. 25, 103001 (2010). and references therein
5. Katsuki, H., Takei, N., Sommer, C., Ohmori, K.: Ultrafast coherent control of condensed matter
with attosecond precison. Acc. Chem. Res. 51, 1174–1184 (2018)
6. Okano, Y., Katsuki, H., Nakagawa, Y., Takahashi, H., Nakamura, K.G., Ohmori, K.: Opti-
cal manipulation of coherent phonons in superconducting YBa2 Cu4 O7−δ thin films. Faraday
Discuss. 153, 375–382 (2011)
7. Cheng, Y.-H., Gao, F.Y., Teitelbaum, S.W., Nelson, K.A.: Coherent control of optical phonons
in bismuth. Phys. Rev. B 96, 134302 (2017)
8. Hu, J., Igarashi, K., Sasagawa, T., Nakamura, K.G., Misochko, O.V.: Femtosecond study of
A1g phonons in the strong 3D topological insulators: From pump-probe to coherent control.
Appl. Phys. Lett. 112, 031901 (2018)
9. Weiner, A.M., Leaird, D.E., Wiederrecht, G.P., Nelson, K.A.: Femtosecond multiple-pulse
impulsive stimulated Raman scattering spectroscopy. J. Opt. Soc. Am. B 8, 1264 (1991)
10. Dekorsky, T., Kütt, W., Pfeifer, T., Kurz, H.: Coherent Control of LO-Phonon Dynamics in
Opaque Semiconductors by Femtosecond Laser Pulses. Europhys. Lett. 23, 223 (1993)
11. Hase, M., Mizoguchi, K., Harima, H., Nakashima, S., Tani, M., Sakai, K., Hangyo, M.: Optical
control of coherent optical phonons in bismuth films. Appl. Phys. Lett. 69, 2474 (1996)
12. Takahashi, H., Kato, K., Nakano, H., Kitajima, M., Ohmori, K., Nakamura, K.G.: Optical con-
trol and mode selective excitation of coherent phonons in YBa2 Cu3 O7−δ . Solid State Commun.
149, 1955 (2009)
13. Katsuki, H., Delagnes, J.C., Hosaka, K., Ishioka, K., Chiba, H., Zijlstra, E.S., Garcia, M.E.,
Takahashi, H., Watanabe, K., Kitajima, M., Matsumoto, Y., Nakamura, K.G., Ohmori, K.: All-
optical control and visualization of ultrafast two-dimensional atomic motions in a single crystal
of bismuth. Nat. Commun. 4, 2801 (2013)
124 7 Coherent Control of Optical Phonons

14. Hase, M., Fons, P., Mitrofanov, K., Kolobov, A.V., Tominaga, J.: Femtosecond structural trans-
formation of phase-change materials far from equilibrium monitored by coherent phonons.
Nat. Commun. 6, 8367 (2015)
15. Hu, J., Misochko, O.V., Goto, A., Nakamura, K.G.: Delayed formation of coherent LO phonon-
plasmon coupled modes in n- and p-type GaAs measured using a femtoseocnd coherent control
technique. Phys. Rev. B 86, 235145 (2012)
16. Makino, K., Tominaga, J., Hase, M.: Ultrafast optical manipulation of atomic arrangements in
chalcogenide alloy memory materials. Opt. Express 19, 138475 (2011)
Appendix A
Linear Diatomic Chain: Normal Coordinate
and Hamiltonian

Here, we describe the normal coordinates of the linear diatomic chain and diagonalize
its Hamiltonian.1

A.1 Normal Coordinates

The displacement of atom κ with the wavenumber k is expressed by

1  
ikxκl −iω j (k)t
l
qκ, j (k) = √ A 0, j (k)a κ, j (k)e + c.c. . (A.1)
2 N mκ

When we set q > 0, the displacement of the atomκ at the lth cell is expressed by

1  
A0, j (k)aκ (k)eikxκ −iω j (k)t + c.c.
l
qκl = √
2 N m κ j,k≥0
1  
A0, j (−k)aκ (−k)e−ikxκ −iω j (−k)t + c.c. .
l
+ √ (A.2)
2 N m κ j,k>0

This solution is more generally expressed in terms of N complex normal coordinate

Q kj as
⎛ ⎞
1  
qκl =√ ⎝ l
aκ, j (k)e κ Q j +
ikx k
aκ, j (−k)e −ikx l
κ Q −k
j
⎠, (A.3)
N m κ j,k≥0 j,k>0

where

1 We use the method described in [1].

© Springer Nature Switzerland AG 2019 125
K. Nakamura, Quantum Phononics, Springer Tracts in Modern Physics 282,
https://doi.org/10.1007/978-3-030-11924-9
126 Appendix A: Linear Diatomic Chain: Normal Coordinate and Hamiltonian

1
Q kj = A0, j (k)e−iω j (k)t + A0, j (k)eiω j (−k)t
2
1
= A0, j (k)e−iω j (k)t + A∗0, j (−k)eiω j (k)t . (A.4)
2
∗
j (−k) = aκ, j (k), we calculate qκ as
l∗
Using aκ,
⎛ ⎞
1  
qκl∗ = √ ⎝ a ∗ (k)e−ikxκ Q k∗
l
j +
∗
aκ, j (−k)e
ikxκ −k∗ ⎠
l
Qj
N m κ j,k≥0 κ, j j,k>0
⎛ ⎞
1  
= √ ⎝ a ∗ (k)e−ikxκ Q k∗
l
j +
∗
aκ, j (−k)e
ikxκ −k∗ ⎠
l
Qj . (A.5)
N m κ j,k≥0 κ, j j,k>0

−k
Since the displacement qκl is real, qκl∗ = qκl . Then, we get Q k∗
j = Q j . One of the
most important merits of using the normal coordinates is that the Hamiltonian is
diagonalized by using the normal coordinate (as shown in the next subsection).

A.2 Diagonalizing Hamiltonian

Hamiltonian H is a sum of kinetic energy T and potential energy V : H = T + V .

The potential energy is

 pκl 2
T = , (A.6)
l,κ
2m κ

where pκl is a momentum and defined by

d l
pκl = m κ q = m κ q̇κl
dt κ
mκ 
= aκ, j (k) exp(ikxκl ) Q̇ j (k). (A.7)
N k, j

Using the normal coordinates, T is

⎛ ⎞
1  
T = ⎝ aκ, j (k) exp(ikxκl ) Q̇ j (k)⎠
2N l,κ k, j
⎛ ⎞

×⎝ aκ, j  (k  ) exp(ik  xκl ) Q̇ j  (k  )⎠
k, j 
Appendix A: Linear Diatomic Chain: Normal Coordinate and Hamiltonian 127

1  
= mκ aκ, j (k)aκ, j  (k  ) exp(i(k + k  )la)
2N l,κ k,k  , j, j 

× exp(i(k + k  )xκ ) Q̇ j (k) Q̇ j  (k  ). (A.8)

This is not zero for k  = −k condition because l exp(i(k − k  )la) = N Δ(k − k  ).

Then, we get

1
T = aκ, j (k)aκ, j  (−k) Q̇ j (k) Q̇ j  (−k)
2 κ k, j, j 
1 ∗ ∗
= aκ, j (k)aκ, j  (k) Q̇ j (k) Q̇ j  (k)
2 κ k, j, j 
1
= Q̇ j (k) Q̇ ∗j  (k), (A.9)
2 k, j

∗
where we use κ aκ, j (k)aκ, j  (k) = δ j, j  .
Next we calculate the potential energy V using (4.38):

1
V = f S + gS  , (A.10)
2
where
 2
S= q2l − q1l
l
 2

S = q1l+1 − q2l . (A.11)
l

Using the displacements described with normal coordinates

 1
q2l = √ a2, j (k) exp(ik(la + x2 ))Q j (k)
q, j
N m2
 1
q1l = √ a1, j (k) exp(ik(la + x1 ))Q j (k)
q, j
N m2
 1
q1l+1 = √ a1, j (k) exp(ik(la + x1 + a))Q j (k), (A.12)
q, j
N m2
128 Appendix A: Linear Diatomic Chain: Normal Coordinate and Hamiltonian

we get
⎛ ⎞
   1
q2l
2
= ⎝ √ a2, j (k) exp(ik(la + x2 ))Q j (k)⎠
l l k, j
N m2
⎛ ⎞
 1
×⎝ √ a2, j  (k  ) exp(ik  (la + x2 ))Q j  (k  )⎠
k, j 
N m 2

1  
= a2, j (k)a2, j  (k  ) exp(i(k + k  )la)
N m 2 l k,k  , j, j 
× exp(i(k + k  )x2 )Q j (k)Q j  (k  )
1 
= a2, j (k)a2, j  (−k)Q j (k)Q j  (−k)
m 2 k, j, j 
1  ∗ ∗
= a2, j (k)a2, j  (k)Q j (k)Q j  (k). (A.13)
m 2 k, j, j 

Similar calculation gives

 2 1  ∗ ∗
q1l = a1, j (k)a1, j  (k)Q j (k)Q j  (k), (A.14)
l
m 1 k, j, j 

⎛ ⎞
   1
q1l q2l = ⎝ √ a1, j (k) exp(ik(la + x1 ))Q j (k)⎠
l l k, j
N m 1
⎛ ⎞
 1
×⎝ √ a2, j  (k  ) exp(ik  (la + x2 ))Q j  (k  )⎠

k ,j  N m 2

1 
∗ ∗
= √ a1, j (k)a2, j  (k)Q j (k)Q j  (k) exp(ik(x 1 − x 2 ))
m 1 m 2 k, j, j 
1 
∗ ∗
= √ a1, j (k)a2, j  (k)Q j (k)Q j  (k)ψ12 , (A.15)
m 1 m 2 k, j, j 

where ψ12 ≡ exp(ik(x1 − x2 )). Similarly, we get

⎛ ⎞
   1
q2l q1l = ⎝ √ a1, j (k) exp(ik(la + x2 ))Q j (k)⎠
l l k, j
N m2
Appendix A: Linear Diatomic Chain: Normal Coordinate and Hamiltonian 129
⎛ ⎞
 1
×⎝ √ a1, j  (k  ) exp(ik  (la + x1 ))Q j  (k  )⎠

k ,j  N m 1

1 
= √ a ∗ (k)a2, j  (k)Q j (k)Q ∗j  (k)ψ12
∗
. (A.16)
m 1 m 2 k, j, j  1, j

Using these equations, we get

 2 2

S= q2l + q2l q1l + q1l q2l + q1l
l
 1 1
= Q j (k)Q ∗j  (k) ∗
a1, j (k)a1, j  (k) +
∗
a2, j (k)a2, j  (k)
k, j, j 
m1 m2
1 
∗ ∗ ∗
− √ a1, j (k)a2, j  (k)ψ 12 + a 1, j  (k)a 2, j (k)ψ 12 . (A.17)
m1m2

The S  term is also calculated in the similar way:

 1 1
S = Q j (k)Q ∗j  (k) ∗
a1, j (k)a1, j  (k) +
∗
a2, j (k)a2, j  (k)
k, j, j 
m1 m2
1 
∗ ∗ ∗
− √ a1, j (k)a2, j  (k)Ψ12 + a1, j  (k)a2, j (k)Ψ12 , (A.18)
m1m2

where ψ12 in S is replaced by Ψ12 , which is defined by Ψ12 ≡ exp(ik(x1 − x2 + a)).

The potential energy V is obtained as

1  f +g f +g
V = Q j (k)Q ∗j  (k) ∗
a1, j (k)a1, j  (k) +
∗
a2, j (k)a2, j  (k)
2 k, j, j  m1 m2
1  ∗ ∗ ∗ ∗

− √ a1, j (k)a2, j  (k) ( f ψ12 + gΨ12 ) + a1, j  (k)a2, j (k) f ψ12 + gΨ12 .
m1m2
(A.19)

Using (4.48), the terms are replaced as

1 ∗
−√ ( f ψ12 + gΨ12 ) = D12 (k) = D21 (k)
m1m2
1 ∗ ∗ ∗
−√ f ψ12 + gΨ12 = D12 (k) = D21 (k), (A.20)
m1m2

and we get
130 Appendix A: Linear Diatomic Chain: Normal Coordinate and Hamiltonian

1 
V = Q j (k)Q ∗j  (k) D11 (k)a1, j (k)a1, ∗ ∗
j  (k) + D22 (k)a2, j (k)a2, j  (k)
2 k, j, j 

∗ ∗
+ D21 (k)a1, j (k)a2, j  (k) + D12 (k)a1, j  (k)a2, j (k)

1 
= Q j (k)Q ∗j  (k) D11 (k)a1, j (k) + D12 (k)a2, j (k) a1, ∗
j  (k)
2 k, j, j 

∗ ∗
+ D22 (k)a2, j (k) + D21 (k)a1, j (k) a 2, j  (k) . (A.21)

Using (4.50), the equation is simplified as

1  
V = Q j (k)Q ∗j  (k) ω2j a1, j (k)a1,
∗ ∗
j  (k) + ω j a2, j (k)a2, j  (k)
2
2 k, j, j 
1
= Q j (k)Q ∗j (k)ω2j . (A.22)
2 k, j

The Lagrangian is expressed using (A.8) and (A.22)

1 
L =T −V = Q̇ j (k) Q̇ ∗j (k) − Q j (k)Q ∗j (k)ω2j , (A.23)
2 k, j

and the generalized momentum P j (k) conjugate to the normal coordinate Q j (k) is
defined from the Lagrangian:

∂L
P j (k) = = Q̇ j (k). (A.24)
∂ Q̇ j (k)

The Hamiltonian expressed using the normal coordinates and the conjugated mo-
mentum is
1 
H =T +V = P j (k)P j∗ (k) + Q j (k)Q ∗j (k)ω2j . (A.25)
2 k, j
Appendix B
Ultrashort Laser Technology

An ultrashort optical pulse is required to excite coherent optical phonons, because

its pulse width should be shorter than vibrational period (lower than several hundred
femtoseconds). Then, we show a brief summary of ultrashort laser technology.

B.1 Pulse and Spectrum Width

An ultrashort pulse consists of many continuous waves with different frequencies.

Let us consider a simple Gaussian light pulse, which does not have time-dependent
phase,

E(t) = E 0 e−t /σ 2 iω0 t

2
e , (B.1)

where E 0 is the amplitude of the electric field, σ is the pulse width, and ω0 is the
angular frequency of the light. The spectrum of the pulse is calculated by the Fourier
transformation of the pulse as
 ∞
1 σ
E(t)e−iωt dt = √ E 0 e−(ω−ω0 ) σ /4 .
2 2
E(ω) = √ (B.2)
2π −∞ 2

The spectrum has a Gaussian form at the center angular frequency of ω0 . The inten-
sities of the pulse and the spectrum are defined by I (t) = E(t)E ∗ (t) and I (ω) =
E(ω)E ∗ (ω), respectively. The widths (full√ width of half maximum) of the pulse
and the spectrum are obtained as Δt = σ 2 ln 2 and Δω = 2π × Δν = 4 ln 2/σ ,
respectively [2]. The relation between the pulse width and the frequency is

2 ln 2
ΔtΔν = ≈ 0.441. (B.3)
π

© Springer Nature Switzerland AG 2019 131

K. Nakamura, Quantum Phononics, Springer Tracts in Modern Physics 282,
https://doi.org/10.1007/978-3-030-11924-9
132 Appendix B: Ultrashort Laser Technology

In more general case, the pulse includes a time-dependent phase term (eiθ(t) and
e−iθ(ω) ) and ΔtΔν is bigger than 0.441. The pulse satisfying (B.3) is called the
Fourier-transform-limited pulse.
When the Fourier-transform-limited pulse passes through optical materials such
as a lens and a beamsplitter, the pulse gets a frequency-dependent phase change [3].
The pulse after passing the optics is expressed as
 ∞
1
E  (t) = √ E(ω)eiθ(ω) eiωt dω. (B.4)
2π ∞

The phase change is expressed by the Taylor expansion around ω0

 
dθ  1 d2 θ 
θ (ω) = θ0 + (ω − ω ) + (ω − ω0 )2
dω ω0 2 dω2 ω0
0


1 d3 θ 
+ (ω − ω0 )3 + · · · , (B.5)
6 dω3  ω0

where θ0 is a constant, the second term is a group velocity delay, and the third term
is a group velocity dispersion. If we consider only the third term using

1 d2 θ 
δ≡ , (B.6)
2 dω2 ω0

the electric field amplitude is given by

 ∞
1
E  (t) = √ E(ω)eiθ(ω) eiωt dω
2π ∞
 ∞  
E0 σ (ω − ω0 )2 σ 2
= √ exp − exp(iδ(ω − ω0 )2 ) exp(iωt)dω
2 π ∞ 4
 ∞   2  
E0 σ σ
= √ eiω0 t exp − − iδ u 2 + itu du, (B.7)
2 π ∞ 4

where we set u = ω − ω0 . Using the Gaussian integral formula, we get

 
E0 σ 4π −(ω − ω0 )2
E  (t) = √ eiω0 t exp
2 π σ 2 − i4δ σ 2 − i4δ
 
E 0 σ iω0 t 4π −i4δt 2
= √ e exp
2 π σ 2 − i4δ σ 4 + 16δ 2
 2 
t
× exp − 2 . (B.8)
σ + 16δ 2 /σ 2
Appendix B: Ultrashort Laser Technology 133

The result indicates that the frequency is dependent on time and the pulse width is
larger than that of the original pulse. This effect is called “chirp”.

B.2 Optical Interference

The pulse width of femtosecond optical pulses cannot be measured directly by an

electronic devise, because it has not enough time resolution. The pulse shape is mon-
itored via optical interference for a Fourier transfer limited pulse [4]. A femtosecond
pulse is divided into two pulses (pulse 1 and pulse 2) by a beam splitter. The pump
1 and pump 2 are mixed after passing the different optical paths and detected by a
slow photodetector. The electric fields of the pump 1 and pump 2 are
 
t2
E 1 (t) = E 0 exp − 2 e−iωt
σ
 
(t − τ )2 −iω(t−τ )
E 2 (t) = E 0 exp − e , (B.9)
σ2

where τ is delay between the pump 1 and pump 2. The intensity I (τ ) detected by
the detector is
 ∞  ∞
I (τ ) = |E 1 (t) + E 2 (t)| dt =
2
(E 1 (t) + E 2 (t))(E 1∗ (t) + E 2∗ (t))dt
−∞ −∞
 ∞    
2t 2 2(t − τ )2
= |E 0 |2 exp − 2 + exp −
−∞ σ σ2
 2 
t + (t − τ )2
+ 2 exp − cos(ωτ )dt
σ2
   
√ τ2
= |E 0 |2 2π σ 1 + exp − 2 cos(ωτ ) . (B.10)
2σ
Appendix C
Mathematical Formula

C.1 Gaussian Integral

The Gaussian integral is

 ∞
π
e−ax dx =
2
I (a) = , (C.1)
−∞ a

where Re(a) > 0.

The square of I is calculated as
 ∞ 2  ∞   ∞ 
e−ax dx e−ax dx e−ay dy
2 2 2
I (a)2 = =
−∞ −∞ −∞
 ∞  ∞
−a(x 2 +y 2 )
= e dxdy. (C.2)
−∞ −∞

Using polar coordinates (x = r cos θ and y = r sin θ ), the integral is

 ∞  2π  ∞
e−ar dθr dr = 2π e−ar r dr
2 2
I (a)2 =
0 0 0
 ∞
e−ar
2
π
= 2π = . (C.3)
−2a a
0
√
Finally, we get I (a) = π/a.

C.1.1 Two Complex Parameters’ Case

Let us consider the Gaussian integral with two complex parameters (α and β)
© Springer Nature Switzerland AG 2019 135
K. Nakamura, Quantum Phononics, Springer Tracts in Modern Physics 282,
https://doi.org/10.1007/978-3-030-11924-9
136 Appendix C: Mathematical Formula
 ∞
e−α(x+β) dx,
2
I (α, β) = (C.4)
−∞

where Re(α) > 0.2 The integrand function goes to zero at x → ±∞ for the Re(α)>0
condition, and the integral I (α, β) converges. The partial derivative of I (α, β) by
β is
 ∞  ∞
∂ I (α, β) ∂ −α(x+β)2
−2α(x + β)e−α(x+β) dx
2
= e dx =
∂β −∞ ∂β −∞
 
2 ∞
= e−α(x+β) = 0. (C.5)
−∞

Then I (α, β) is independent on β. The partial derivative of I (α, β) by α is

 ∞
∂ I (α, β)
−(x + β)2 e−α(x+β) dx
2
=
∂α −∞

x +β 
∞ 
−2α(x + β)e−α(x+β) dx
2
=
−∞ 2α
   ∞
x + β −α(x+β)2 ∞ 1 −α(x+β)2
= e − e dx
2α −∞ −∞ 2α
1
= − I (α, β). (C.6)
2α
Therefore, we get
   
1 1
I (α, β) = C exp − ln α = C exp − ln(|α| + i arg(α))
2 2
 
C i
= √ exp − arg(α) , (C.7)
|α| 2

where C is a constant, and arg(α) is the argument of α. If we set α = a > 0, arg(a) = 0

and
C
I (a, β) = √
a
π
= I (a) = . (C.8)
a
√
Then we get C = π . Finally, we get
 ∞  
−α(x+β)2 π i
I (α, β) = e dx = exp − arg(α) . (C.9)
−∞ |α| 2

2 We use the method described in [5].

Appendix C: Mathematical Formula 137

Using this formula, we get

 ∞
π
e−a(x+ib) dx =
2
, (C.10)
−∞ a

where a and b are real.

C.1.2 Fourier Transformation

Fourier transformation of the Gaussian function is also important and often used in
the text.
 ∞  ∞
−ax 2 ikx
e−a(x −ikx/a) dx
2
e e dx =
−∞ −∞
 ∞
e−a(x−ik/(2a)) e−k /(4a) dx
2 2
=
−∞
π −k 2 /(4a)
= I (a, −ik/(2a))e−k /(4a)
2
= e . (C.11)
a

I (a, −ik/(2a)) is obtained by using (C.9) and arg(a) = 0.

References

1. Bruesch, P.: Phonons: Theory and Experiments I. Springer Series in Solid-State Sciences, vol.
34. Springer, Berlin (1982)
2. Rulliere, C. (ed.): Femtosecond Laser Pulses, Principles and Experiments, 2nd edn. Springer
Science+Business Media Inc, Berlin (2005)
3. Watanabe, S.: Cyotanparusu hassei gijyutu (in Japanese: Generation of ultrashort pulse). Kogaku
24, 378–383 (1995)
4. Trebino, R.: Frequency-Resolved Optical Grating: The measurement of ultrashort laser pulses.
Kluwer Academic Publishers, Norwell (2000)
5. Nomoto: Gauss sekibun no ippannka (in Japanese: General Solution of Gaussian Integral). http://
www.eng.niigata-u.ac.jp/~nomoto/15.html
Index

A Cosine-like oscillation, 73
Acoustic mode, 62 Coupling constant, 69
Acoustic phonon, 58 Creation operator, 29, 64, 105
A1g -mode phonon, 68 Current, 69
Annihilation operator, 29, 34, 64, 105
Atomic deviation, 75
D
Damped harmonic oscillation, 72
B Damping constant, 68
Baker–Hausdrof lemma, 44 Dawson function, 101
Band model, 109 Decay time, 72
Bandwidth, 68 Deformation potential, 109
Bogoliubov transformation, 39, 44 Density functional theory, 122
Boson, 58 Density matrix, 20
Bra vector, 1 Density operator, 11, 85
Dephasing, 74
Depolarization, 69
C Destructive interference, 116, 119
Canonical quantization, 55 Detuning, 100
Chemical reaction, 115 Diagonal component, 18
Chirp mirror, 71 Diamond, 116
Classical number, 3 Dimensionless operator, 57
Closed system, 4 Dimensionless parameter, 55
Coherence, 12 Dipole interaction, 82, 107
Coherent control, 115 Dispersion curve, 52
Coherent phonons, 67 Dispersion relation, 52, 55
Coherent state, 34, 39 Displaced harmonic oscillator, 37, 105
Collective oscillations, 58 Displaced potential, 69
Commulator, 44 Displaced vacuum state, 38
Commutation relation, 25, 26, 30, 55, 57, 64 Displacement, 51
Commutator, 7 Displacement operator, 37, 91
Complete set, 4 Displacive excitation of coherent phonons,
Complex vector space, 1 68
Conduction bad, 109 Double factorial function, 42
Conduction band, 73 Double femtosecond pulses, 116
Constructive interference, 116, 119 Double-sided Feynman diagram, 17, 84,
Cosine-like, 97 103, 117
© Springer Nature Switzerland AG 2019 139
K. Nakamura, Quantum Phononics, Springer Tracts in Modern Physics 282,
https://doi.org/10.1007/978-3-030-11924-9
140 Index

Driving force, 68 Group velocity dispersion, 71

Dynamical variable, 3, 56

H
E Hamiltonian, 4
Eigenstate, 27 Harmonic oscillator, 25, 35
Eigenvalue, 2, 27 Haung–Rhys factor, 82
Eigenvalue equation, 27 Heisenberg picture, 7
Eigenvector, 2 Hermitial operator, 27
Electric susceptibility, 70 Hermitian conjugate, 2, 105
Electromagnetic field, 7, 82 Hermitian operator, 2
Electron annihilation operator, 69 Heterodyne detection, 111
Electron creation operator, 69 High-temperature transition superconduc-
Electron density, 69 tor, 121
Electron density matrix, 70
Electronic excited state, 69
Electronic ground state, 84, 108 I
Electronic state, 81, 88 Identity operator, 3
Electronic-state-selective measurement, 99 Imaginary part, 111
Electron-hole pair, 73 Impulsive absorption process, 87
Electron-phonon coupling, 104, 106 Impulsive Raman, 76
Electro-optical sampling, 72 Impulsive stimulated Raman scattering, 68
Energy gap, 82 Impulsive stimulated Raman scattering pro-
Envelope, 96 cess, 87
Equation of motion, 4 Initial state, 85
Equilibrium position, 51, 59 Inner product, 2
Error ellipse, 46 Interaction-free, 18
Even number state, 43 Interaction Hamiltonian, 9, 82
Excitation density, 73 Interaction picture, 7
Excitation energy, 109
Excited state, 12
Expectation value, 4, 7 K
Ket vector, 1
Kinetic energy, 54
F Klemens channel, 74
Femtosecond laser pulse, 70
Fermi level, 73
Fluctuation, 76 L
Fock state, 29 Laser-plasma X-rays, 75
Force, 59 Lattice vibration, 53
Fourier transformation, 56 Lifetime, 122
Fourier-transform limited, 83 Linear atomic chain, 51
Four-level system, 103 Linear diatomic chain, 58
Liouville-von Neumann equation, 13
Longitudinal optical phonon, 68
G Long-wavelength approximation, 62
Γ -point, 72 Lowest state, 29
Gaussian function, 31, 96
Gaussian integral, 97
Gaussian pulse, 110 M
General squeezed state, 48 Matrix, 12
Glauber state, 34 Matrix form, 18
Gravitational-wave detection, 76 Measurement, 3
Ground state, 12, 29 Michelson-type interferometer, 116
Index 141

Minimum uncertainty state, 37, 39 Photoexcited electron, 69

Mixed state, 11 Photon, 58
Plane wave expansion, 51, 59
Plasmon, 74
N Poisson distribution, 36
Nearest-neighbor distance, 75 Polarizability tensor, 76
Node, 32 Polarization operator, 112
Nonclassical state, 75 Polarizing beamsplitter, 72
Nonlinear spectroscopy, 7 Position operator, 33
Norm, 2 Position space, 30
Normal coordinate, 62, 68 Potential energy, 54
Normal frequency, 61 Potential minimum, 37
Normalization factor, 96 Probability, 3
Normal mode, 61 Probability distribution function, 32
Number state, 32 Projection operator, 3
Pulse width, 68, 98
Pulse-shaping, 115
O Pump-probe technique, 70
Observable, 4 Pure state, 11
Observed value, 33
Occupation number state, 29
Odd function, 32 Q
Off-diagonal, 12 Quantum communication, 76
Opaque condition, 68, 116 Quantum Liouville equation, 13
Operator, 2 Quantum noise, 76
Optical frequency, 120 Quantum optics, 34
Optical mode, 62 Quantum state, 115
Optical phonon, 58 Quasi-particle, 58
Optical pulse, 96
Orthogonal condition, 54
Orthonormal basis set, 12 R
Oscillating perturbation, 19 Rabi frequency, 20
Oscillation period, 67 Rabi oscillation, 21
Outer product, 12 Raman-active, 68
Raman spectroscopy, 62, 72
Recursion relation, 41
P Relaxation rate, 19
Penetration depth, 82 Resonant condition, 20
Periodic boundary condition, 51, 56, 58 Rotating frame, 20
Perturbative interaction, 8 Rotating wave approximation, 83, 109
Phase-change material, 122
Phase-dependent noise, 76
Phase factor, 5, 39 S
Phase matching, 82 Schrödinger equation, 4
Phase relaxation, 19 Schrödinger picture, 4, 56
Phenomenological equation, 70 Second-order perturbation, 84, 117
Phonon, 58 Second-order Raman scattering, 76
Phonon amplitude, 91, 94, 97 Sharpened potential, 40
Phonon coherence, 119 Short pulse limit, 97
Phonon coordinate operator, 105 Sine-like oscillation, 68
Phonon polarization, 110 Single-sided Feynman diagram, 9
Phonon state, 90 Spectral decomposition, 3
Phonon-plasmon coupled oscillation, 74 Spontaneous Raman, 76
Photodiode, 70 Spring constant, 51
142 Index

Squeezed operator, 43 Two-dimensional atomic motion, 121

Squeezed state, 39 Two-level system, 18, 81
Squeezed vacuum state, 40 Two-photon process, 115
Squeeze factor, 76
Squeeze operator, 76
Standard deviation, 32 U
State vector, 1 Ultrashort optical pulse, 67
Statistical average, 69 Ultrashort X-ray pulse, 75
Steady state, 5 Uncertainty principle, 76
Structure factor, 75 Uncertainty relation, 36
Superposition state, 11, 33 Unitary operator, 5, 38, 44
Surface-charge field, 68 Unit cell, 58
Surface-space-charge field, 73, 116

V
T
Vacuum state, 29, 75
Taylor expansion, 42
Valence band, 73
Time-dependent coherent state, 39
Variance, 37, 76
Time-dependent density functional theory,
70 Vibrational period, 99
Time-dependent squeezed state, 43 Von Neumann equation, 13
Time-evolution operator, 5, 8, 16
Time-ordered exponential, 6, 83
Ti:sapphire oscillator, 70 W
Trace, 12 Wave function, 30
Transient optical transmission, 70 Wave packet, 39
Transient reflectivity, 68 Wave vector, 52
Transient transmittance, 117
Transition dipole, 83
Transition process, 84, 103 Z
Transparent condition, 68, 110, 115 Zero-phonon state, 84
Transverse acoustic phonon, 76 Zero-point energy, 29, 82