Springer Series in Optical Sciences 246

Iwao Matsuda
Ryuichi Arafune Editors

Nonlinear X-Ray
Spectroscopy
for Materials
Science
Springer Series in Optical Sciences

Volume 246
Springer Series in Optical Sciences is led by Editor-in-Chief William T. Rhodes,
Florida Atlantic University, USA, and provides an expanding selection of research
monographs in all major areas of optics:
• lasers and quantum optics
• ultrafast phenomena
• optical spectroscopy techniques
• optoelectronics
• information optics
• applied laser technology
• industrial applications and
• other topics of contemporary interest.

With this broad coverage of topics the series is useful to research scientists and
engineers who need up-to-date reference books.
Iwao Matsuda · Ryuichi Arafune
Editors

Nonlinear X-Ray
Spectroscopy for Materials
Science
Editors
Iwao Matsuda Ryuichi Arafune
The Institute for Solid State Physics Research Center for Materials
The University of Tokyo Nanoarchitectonics (MANA)
Kashiwa, Chiba, Japan National Institute for Materials Science
(NIMS)
Tsukuba, Ibaraki, Japan

ISSN 0342-4111 ISSN 1556-1534 (electronic)

Springer Series in Optical Sciences
ISBN 978-981-99-6713-1 ISBN 978-981-99-6714-8 (eBook)
https://doi.org/10.1007/978-981-99-6714-8

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Singapore Pte Ltd. 2023

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

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore

Paper in this product is recyclable.

Preface

There has been a rise of a new analysis method of nonlinear X-ray spectroscopy for
materials science. This book provides the basic principles for understanding the issue
and reviews the frontier topics, followed by the future prospects. Readers are expected
to have elementary knowledges of electromagnetism and quantum mechanics.
A spectrum is essentially a set of signals taken with parameters. For example, it
is the photon intensity signal at different wavelengths (photon energies). The depen-
dence provides information of a sample that is characteristic to the materials. Inter-
pretations of experimental spectra can naturally be done when amounts of the signal
are proportional to the incident intensity. However, there are varieties of the optical
phenomena that show the nonlinear dependence. Moreover, they are associated with
events, such as second harmonic generation, that are unusual in everyday life but crit-
ical for our scientific development. These special phenomena are typically induced
with the ultrashort pulse light source, i.e., laser, and the recent technical innova-
tions have pushed the controllable wavelength down to the X-ray region. “X-ray”
was named after a ray of “something unknown (X)” by Röntgen, and it has been a
significant experimental probe to investigate structure and electronic states of mate-
rials. Today, there rises a new research field of the nonlinear X-ray experiments that
have led to the significant discoveries. The understandings have also opened a new
analysis method of the nonlinear X-ray spectroscopy that are found to be useful for
materials science. The frontier topics deserve known widely among researchers, but
the experimental principles are based on the unusual phenomena with the mysterious
light. Thus, the issue is likely hard to get started with, and it is desirable to have a
guide for those who are interested in. This situation has motivated us to develop the
book.
This book is composed of six chapters (Fig. 1). Chapter 1 introduces the basic
principles of the nonlinear responses and the related issues. Chapter 2 guides various
linear X-ray spectroscopies for materials science and provides the useful knowledges
for understanding the following chapters. Chapters 3–5 review the frontier topics of
nonlinear spectroscopy at various wavelength (energy) regions of photons. At the
end of the book, future prospects of this new research field are given.
The editing of this book was done at two sites: (i) Research Center for Materials
Nanoarchitectonics (MANA), National Institute for Materials Science, and (ii) Laser
v
vi Preface

Fig. 1 Structure of the book

and Synchrotron Research Center (LASOR), the Institute for Solid State Physics
(ISSP), the University of Tokyo. We would like to express our sincere appreciations
to all the authors and those acknowledged in the individual chapters. Colleagues who
contributed directly in providing the wonderful illustrations and the critical reading
of the entire manuscript are: Nanami Arafune, Yurina Matsuda, Akiko Itakura, Tasha
Welling, Masafumi Horio, Yuki Tsujikawa, and the ISSP Public Relation Office.
We would like to appreciate you for reading this book and wish to enjoy it.

Ryuichi Arafune
Tsukuba, Ibaraki, Japan

Iwao Matsuda
Kashiwa, Chiba, Japan
Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Ryuichi Arafune
1.1 Nonlinearity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Classical Model of Nonlinear Optical Process . . . . . . . . . . . . . . . . . . 2
1.3 Pragmatic Importance of Nonlinear Spectroscopy
for Materials Science . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Synopsys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2 Linear X-Ray Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Iwao Matsuda
2.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.1.1 Optical Responses of Materials in the X-Ray Region . . . . . . 15
2.1.2 X-Ray Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.1.3 Light–Matter Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.2 X-Ray Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.2.1 X-Ray Absorption Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 28
2.2.2 Photoelectron Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2.3 X-Ray Emission Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.3 Time-Resolved X-Ray Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.3.1 Measurement Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.3.2 Examples of Time-Resolved Photoemission
Spectroscopy Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.4 Researcher’s Guide to Material Characterization with X-Rays . . . . . 49
2.4.1 The Guiding Chart for Experiments . . . . . . . . . . . . . . . . . . . . . 49
2.4.2 Application for Beamtime at the X-Ray Facility . . . . . . . . . . 51
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

vii
viii Contents

3 Probing Nonlinear Light–Matter Interaction in Momentum

Space: Coherent Multiphoton Photoemission Spectroscopy . . . . . . . . . 57
Marcel Reutzel, Andi Li, Zehua Wang, and Hrvoje Petek
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.2 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.2.1 Experimental Setup: The Photoelectron Analyzer . . . . . . . . . 60
3.2.2 Experimental Setup: The Light-Source . . . . . . . . . . . . . . . . . . 60
3.3 mPP: Highly Nonlinear Mapping of the Energy-
and Momentum-Dispersive Electronic Band Structure . . . . . . . . . . . 62
3.3.1 Static mPP in Threshold Order of Photoemission . . . . . . . . . 63
3.3.2 Above Threshold Photoemission . . . . . . . . . . . . . . . . . . . . . . . 63
3.3.3 Toward Full Surface Brillouin Zone Mapping
by Coherent mPP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.4 Coherent Two-Dimensional Photoelectron Spectroscopy . . . . . . . . . 66
3.4.1 Coherent 2D FT Photoelectron Spectroscopy—Optical
Bloch Equation Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.4.2 Coherent 2D FT Photoelectron Spectroscopy
of Ag(111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.4.3 Coherent Above Threshold Photoemission . . . . . . . . . . . . . . . 72
3.5 Ultrafast Quasiparticle Dressing by Light . . . . . . . . . . . . . . . . . . . . . . 74
3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4 Nonlinear Soft X-Ray Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Craig P. Schwartz and Walter S. Drisdell
4.1 Nonlinear Spectroscopy—Development with Visible Light . . . . . . . 84
4.2 Ultrafast X-Ray Light Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.3 Family of Soft X-Ray Nonlinear Spectroscopy . . . . . . . . . . . . . . . . . . 92
4.4 Nonlinear Soft X-Ray Optics and Spectroscopies . . . . . . . . . . . . . . . 95
4.4.1 Multiphoton Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4.4.2 Stimulated Emission/Forward Scattering . . . . . . . . . . . . . . . . 97
4.4.3 Stimulated Raman Scattering . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4.4.4 Four Wave Mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
4.4.5 Soft X-Ray Second Harmonic Generation/Sum
Frequency Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.5 Theoretical Calculations for the Spectral Analysis . . . . . . . . . . . . . . . 110
4.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5 Nonlinear X-Ray Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Yuya Kubota and Kenji Tamasaku
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.2 The Basic Theory of Nonlinear Optics in the Hard X-Ray
Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.2.1 Nonlinear Polarizability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.2.2 X-Ray Second Harmonic Generation . . . . . . . . . . . . . . . . . . . 127
Contents ix

5.2.3 Parametric Down-Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . 128

5.2.4 Sum Frequency Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.3 Featuring Examples of Nonlinear X-Ray Spectroscopy . . . . . . . . . . . 132
5.3.1 X-Ray Two-Photon Absorption Spectroscopy . . . . . . . . . . . . 132
5.3.2 Saturable Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
5.3.3 Atomic X-Ray Laser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
5.3.4 Stimulated X-Ray Emission Spectroscopy . . . . . . . . . . . . . . . 139
5.3.5 X-Ray Transient Grating Spectroscopy . . . . . . . . . . . . . . . . . . 139
5.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6 Future Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
Iwao Matsuda, Craig P. Schwartz, Walter S. Drisdell,
and Ryuichi Arafune
6.1 Toward Multi-dimensional Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 147
6.2 Phase Sensitive Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
6.3 Vacuum Nonlinear X-Ray Optics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
6.4 Developments in Experimental Stations for Materials Science . . . . . 154
6.5 Into the Deep: Nonlinear Science . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Contributors

Ryuichi Arafune Research Center for Materials Nanoarchitectonics (MANA),

National Institute for Materials Science (NIMS), Tsukuba, Japan
Walter S. Drisdell Chemical Sciences Division, Lawrence Berkeley National
Laboratory, Berkeley, CA, USA
Yuya Kubota RIKEN SPring-8 Center, Hyogo, Japan
Andi Li Department of Physics and Astronomy and IQ Initiative, University of
Pittsburgh, Pittsburgh, PA, USA
Iwao Matsuda The Institute for Solid State Physics, The University of Tokyo,
Chiba, Japan
Hrvoje Petek Department of Physics and Astronomy and IQ Initiative, University
of Pittsburgh, Pittsburgh, PA, USA
Marcel Reutzel I. Physikalisches Institut, Georg-August-Universität Göttingen,
Göttingen, Germany
Craig P. Schwartz Nevada Extreme Laboratory, University of Nevada, Las Vegas,
USA
Kenji Tamasaku RIKEN SPring-8 Center, Hyogo, Japan
Zehua Wang Department of Physics and Astronomy and IQ Initiative, University
of Pittsburgh, Pittsburgh, PA, USA

xi
Abbreviations

4WM Four-wave mixing

AED Auger electron diffraction
AES Auger electron spectroscopy
ARPES Angle-resolved photoelectron (photoemission) spectroscopy
ASE Amplified spontaneous emission
ATP Above threshold photoemission
CDI Coherent diffraction imaging
CLS Core-level photoelectron (photoemission) spectroscopy
DFG Difference frequency generation
EXAFS Extended X-ray absorption fine structure
FEL Free electron laser
FT Fourier transform
HHG High harmonic generation
IP Image potential
ITR-mPP Interferometrically time-resolved multi-photon photoemission
KKR Korringa–Kohn–Rostoker
LCLS Linac Coherent Light Source
mPP Multi-photon photoemission
NEXAFS Near-edge X-ray absorption fine structure
OBE Optical Bloch equation
OPA One-photon absorption
OPA Optical parametric amplification
OPO Optical parametric oscillation
PDC Parametric down-conversion
PED Photoelectron diffraction
RIXS Resonant inelastic X-ray scattering
RTTDDFT Real-time time-dependent density functional theory
SA Saturable absorption
SACLA SPring-8 Angstrom compact free-electron laser
SARPES Spin- and angle-resolved photoelectron (photoemission)
spectroscopy

xiii
xiv Abbreviations

SASE Self-amplified spontaneous emission

SFG Sum frequency generation
SHG Second harmonic generation
SR Synchrotron radiation
SS Surface state
SXES Stimulated X-ray emission spectroscopy
SXSHG Soft X-ray second harmonic generation
TG Transient grating
THG Third harmonic generation
TPA Two-photon absorption
tr2PP Time-resolved two-photon photoemission
trARPES Time- and angle-resolved photoelectron spectroscopy
UPS Ultraviolet photoelectron (photoemission) spectroscopy
VG-RTTDDFT Velocity gauge real-time time-dependent density functional
theory
VUV Vacuum ultraviolet
XAFS X-ray absorption fine structure
XAS X-ray absorption spectroscopy
XES X-ray emission spectroscopy
XFEL X-ray free electron laser
XMCD X-ray magnetic circular dichroism
XMLD X-ray magnetic linear dichroism
XPS X-ray photoelectron (photoemission) spectroscopy
XRD X-ray diffraction
XTG X-ray transient grating
Chapter 1
Introduction

Ryuichi Arafune

Abstract This book is primarily intended for students and researchers who are
actively engaged in experimental work within the field of materials science. It offers
an in-depth exploration of cutting-edge techniques and tools designed for measure-
ment of electron dynamics through nonlinear spectroscopy. While it serves as an
invaluable resource for thouse already immersed in this area of study, it also aims to
inspire readers of nonlinear optical spectroscopy within the field of material science.
In this intoroductory chapter, we have underscored the significance of nonlinearity in
material science, elucidated the common basic aspects of nonlinear optical response,
and outlined how nonlinear optical spectroscopy contributes to the understanding of
the dynamics of materials.

1.1 Nonlinearity

There are a surprisingly large number of terms and concepts with negative prefix in
all fields of science. Even with in the realm of the physical sciences, we encounter
terms like non-equilibrium system, non-holonomic system, non-local potential, irre-
versible process, anharmonicity, non-Abelian theory, asymmetry, inelastic process,
non-integral system of coordinates, and many more. Initially, all the concepts in
various fields of science without the negative prefix “Non-” such as equilibrium
system, reversible process, Abelian theory, and many more were developed first.
However, as the field expanded, the need for “Non-” fields also became increasingly
important. Due to the rapid pace of development and the vast diversity of the field,
it has continued to progress under the name “Non-(something)” without being given
a new name. It is important to emphasize that the presence of the negative prefix
does not imply that the discipline is dealing with exceptional or uncommon matters.
On the contrary, it is a broader and deeper field of study, offering a treasure trove

R. Arafune (B)
Research Center for Materials Nanoarchitectonics (MANA), National Institute for Materials
Science (NIMS), Tsukuba, Japan
e-mail: ARAFUNE.Ryuichi@nims.go.jp

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 1
I. Matsuda and R. Arafune (eds.), Nonlinear X-Ray Spectroscopy for Materials Science,
Springer Series in Optical Sciences 246, https://doi.org/10.1007/978-981-99-6714-8_1
2 R. Arafune

of interesting research subjects. Nonlinearity, the subject of this book is no excep-

tion; rather, it serves as a prime example and the foundation concept underlying the
various “Non-(something)” fields mentioned above.
The most crucial characteristic of nonlinearity is the absence of the superposi-
tion principle. Linear phenomena have developed a simple, easy-to-understand, and
elegant system of logic based on the validity of the superposition principle, which
means that they can be subdivided into individual elements and analyzed. However,
on the other hand, it is easy to imagine how boring the world would be if all science
could be described in linear terms. Each element would exist independently and
would not be able to cooperate.
The primary goals of materials science are to create and discover novel function-
ality of materials. Furthermore, understanding “What is a novel functionality?” is
also essential for the development of science. The coordination and interaction of
individual elements play a pivotal role in developing the functions of materials,
rendering nonlinearity the essence of material science. In condensed matter physics,
nonlinearity has been a significant focal point, including its engineering application.
Its significance is anticipated to grow even futher in the future.

1.2 Classical Model of Nonlinear Optical Process

X-ray nonlinear spectroscopy, one of the major techniques introduced in this book,
has attracted much attention as a practical spectroscopy method due to the recent
development of the X-ray free electron laser. The nonlinear optical spectroscopy
begun with the infrared/visible laser technology. As the importance of nonlinear
optics has been recognized, efforts have been made to shorten the wavelengths.
The X-ray nonlinear spectroscopy can be regarded as a milestone in this direction.
This great deal of attention is due to the importance and effectiveness of nonlinear
spectroscopy developed with the short-pulse laser technology. On the other hand, it
should be noted that X-ray nonlinear spectroscopy is not a simple short-wavelength
version of ordinary optical nonlinear spectroscopy.
In the context of material science, X-rays are not merely shorter-wavelength elec-
tromagnetic radiation, but that their wavelength (λ) is of the order of magnitude
of the bonding distance of matter, especially for hard X-ray region. An X-ray has
been conventionally classified into two groups, “soft” or “hard”, by the range of
photon energy (ℏω). A soft X-ray is named for the photon energy region below
ℏω = 2 keV (λ = 0.6 nm), while a hard X-ray above ℏω = 2 keV. This feature
makes a decisive difference in their principles and applications. As a result, clas-
sical/macroscopic treatment in the analysis might be difficult in X-ray nonlinear
spectroscopy. The macroscopic nonlinear response, which is based on classical elec-
tromagnetism, can be described because we implicitly assume that the spatial scale
over which the electric field fluctuates is essentially large compared to the size of the
atoms and molecules, which are the fundamental building blocks of matter. It would
be essentially important to understand X-ray nonlinear spectroscopy by describing
1 Introduction 3

the interaction between X-ray and matter from a semiclassical/quantum mechan-

ical viewpoint. However, if there is no knowledge of nonlinear optical responses
described from the classical viewpoint, it may be difficult to understand the theo-
retical framework and its direction. Therefore, we will introduce the fundamental
and essential features of the nonlinear optical response from a classical viewpoint to
establish a basis for understanding nonlinear spectroscopy [1].
The interaction between light and matter, such as reflection, refraction, scattering,
emission, and absorption, can be treated as a linear response between the incident
light field and the material when the incident radiation field is weak. The intensity of
the output signal is linearly proportional to that of the input light. Furthermore, the
energy of the output photon or electron is identical to that of the input photon. From
a quantum mechanical view, it is often conceived as a situation in which a single
photon excites a ground state to another state. Nonlinear optical response (Fig. 1.1) is
used to refer to cases that fall outside this view, including: (1) the response of material
subjected to interactions with two or more independent incident fields and (2) the
situation where linear response theory is inadequate for treating how the material
behaves, as in the case of very intense incident radiation. From a macroscopic point
of view, nonlinear optical response means that D = ε(E)E, i.e., the permittivity is
described as a function of the electric field.
Now, let us describe the optical properties of a material in terms of a series of
electric fields (power series) to the polarization:

.
P = χ (1) · E + χ (2) : EE + χ (3) .. EEE + · · · . (1.1)

The first term represents the normal linear response, while the second term repre-
sents the lowest order of nonlinear susceptibility. In many cases, this optical nonlin-
earity is small (which is why such a series expansion makes sense), and thus, it
requires a short-pulse light that can produce a high electric field. It should be noted
that nonlinear optical responses are also realized that are not well described by
perturbative expansions at present.
Next, let us describe the physical significance of the second-order nonlinear
susceptibility in a simple but realistic model. In this context, we examine how a

Fig. 1.1 Linear and nonlinear optical response

4 R. Arafune

single electron in a plasma is affected by a linearly polarized electric field described

as:
1( )
By = E x = E exp(ikz − i ωt) + E ∗ exp(−ikz + i ωt) , (1.2)
2

where k = ωc . Equations of motion of the electron are:

e m ẋ
m ẍ = eE x − ż B y −
c τ
m ẏ
m ÿ = −
τ
e m ż
m z̈ = ż B y − (1.3)
c τ
Here, the phenomenological collision time τ is introduced, which describes the
damping of the motion. Assuming the velocity of the electron is sufficiently slow
compared with the velocity of the light and that the force by the magnetic field can be
treated as a perturbation, these equations can be solved by successive approximation
in the form of a Fourier series. The first or linear approximation term is a well-known
result:
−eE exp(ikz − iωt)
x(ω) = ( ) (1.4)
m ω2 + i ω/τ

From this, one can obtain the (linear) dipole moment: ex(ω), and this dipole
moment describes the Rayleigh scattering from the free electron gas. By substituting
the linear solution into the equation of motion, one can obtain the lowest order of the
nonlinear approximation of the electron position.

−ie2 E 2 exp(2ikz − 2i ωt)

z(2ω) = ( )( ) (1.5)
m 2 c 4ω + 2iτ ω2 + iω τ

As analogous to the linear term, ez(2ω) describes the scattering of light with a
frequency twice that of the incident light.
The above idea is extended to the plasma with an average density of No . The
linear polarization is:

Px = χ (1) (ω)E x (ω) = No ex(ω) (1.6)

Assuming ωτ « 1, one can deduce the familiar result for the (linear) susceptibility
of a plasma:

ε − 1 = 4π χ (1) = −4π No e2 /mω2 (1.7)

1 Introduction 5

Similarly, the nonlinear polarization at the second harmonic frequency can be

given by

Pz (2ω) = No ez(2ω). (1.8)

By considering the direction of the polarization to the incident electric field, one
would see that no coherent radiation at 2 ω is generated. This is because we have
assumed that the plasma is centrosymmetric.
Next, we go one step further from the free electron plasma model. The frequency
dependence of the nonlinear susceptibility based on the Lorenz model is described.
One would see that the nonlinear susceptibility is sensitive to the anharmonicity. Let
us consider the anharmonic oscillator with damping:
e
ẍ + ⎡ ẋ + ωo2 x + ax 2 + bx 3 = E(t), (1.9)
M
driven by the electric field. Here, ω0 is the resonant frequency, and ⎡ is the damping
constant. And we assume that the electric field is the summation of the fields with
frequencies ω1 and ω2 :

E(t) = Re{(E 1 exp(ik1 z − i ω1 t) + E 2 exp(ik2 z − iω2 t)}. (1.10)

To proceed with the analysis, one expands the displacement x by the power series:

x(t) = x (1) (t) + x (2) (t) + x (3) (t) + · · · (1.11)

Thus, the linear term can be described by

e
ẍ (1) + ⎡ ẋ (1) + ωo2 x (1) = E(t), (1.12)
M
and its solution can be written as:
[
1 e 1 1
x (1) (t) = E 1 exp(ik1 z − i ω1 t) +
2 M D(ω1 ) D(ω2 )
]
E 2 exp(ik2 z − i ω2 t) + c.c., (1.13)

where the notation

D(ω) = ω0 − ω − i⎡ω = D ∗ (−ω) (1.14)

is introduced. Note that the x (1) (t) term is proportional to the electric field.
The lowest nonlinear-order term can be described by:
( )2
ẍ (2) + ⎡ ẋ (2) + ωo2 x (2) + a x (1) = 0 (1.15)
6 R. Arafune

By substituting x (1) into the above equation, one can obtain x (2) (t).
x (2) (t) consists of many terms, and each term can be represented by the following:

ae2 E 12
x (2) (t) = − exp(2ik1 z − 2i ω1 t) + c.c. (1.16)
4M 2 D 2 (ω1 )D(2ω1 )

ae2 E 22
x (2) (t) = − exp(2ik2 z − 2i ω2 t) + c.c. (1.17)
4M D (ω2 )D(2ω2 )
2 2

ae2 E1 E2
x (2) (t) = − exp(i(k1 + k2 )z − i(ω1 + ω2 )t) + c.c.
2M D(ω1 + ω2 )D(ω1 )D(ω2 )
2

(1.18)
ae2 E1 E2
x (2) (t) = − exp(i(k1 − k2 )z − i(ω1 − ω2 )t) + c.c.
2M D(ω1 − ω2 )D(ω1 )D(ω2 )
2

(1.19)

(and there is the DC component of the displacement which is independent of time).

One would understand the meanings of each term intuitively. In this model, the dipole
moment is represented by ex(t). Thus, each term corresponds to the second-order
nonlinear optical response. More specifically, each term corresponds:
• The second harmonic generation (SHG) of the light with frequency ω1 .
• The second harmonic generation of the light with frequency ω2 .
• The sum frequency generation (SFG) of the light with ω1 and ω2 .
• The difference frequency generation (DFG) of the light with ω1 and ω2 .

The respective excitation processes are also shown in Fig. 1.2.

Fig. 1.2 Energy level scheme for the second nonlinear optical response as described in
(1.16)−(1.19)
1 Introduction 7

One application of the nonlinear optical process is the energy conversion of light.
For example, SHG can convert low-energy infrared light into visible light while
keeping the coherent characteristics as well as the original incident light. This means
that the nonlinear processes can extend the wavelength range of coherent light. From
a viewpoint of material science, a basic question arises here “What are the bene-
fits of probing nonlinear optical responses from a materials science perspective?”
Starting from (1.1), let us see how nonlinear optical responses relate to other “non-
(something)” characteristics. Generally, experiments that use second-order nonlinear
optical effects are conducted using crystalline media and surface interfaces. In other
words, isotropic gases and liquids are unlikely to be used, although the liquid/gas
interface is an important target for nonlinear optics. Let us confirm simply that the
second-order nonlinear response does not occur in symmetric media. In a system with
inversion symmetry, the generated dipole must be − P(r) = P(− r), regardless of its
order. On the other hand, a nonlinear dipole of the second order must satisfy both
−P(r) = −χ (2) : E(r)E(r) and P(−r) = χ (2) : E(−r)E(−r) = χ (2) : E(r)E(r).
Thus, χ (2) must be zero. In other words, the second-order nonlinear optical response
occurs at the point where the symmetry is broken. From this brief discussion, one
would see that nonlinear spectroscopy is a powerful tool for probing symmetry.
The third-order term in (1.1) can be treated in a similar manner to the second-order
term. One can obtain the third-order term of the displacement x (3) . x (3) contains more
terms than x (2) (t). Here, let us introduce an interesting term of x (3) . The frequency
of the term is 2ω1 − ω2 :

e2 1
x (3) (t) = −
4M 2 D(2ω1 − ω2 )D(ω1 )2 D(ω2 )∗
[ ]
3 2a 2 1 a2
× b− −
2 D(ω1 − ω2 ) D(2ω1 )
× E 12 E 2∗ exp(i(2k1 − k2 )z − i(2ω1 − ω2 )t) + c.c. (1.20)

This term describes the non-degenerate four-wave mixing. In this process, there
are two interactions: one is between waves with frequency ω1 and another with
frequency ω2 . As a result, the wave with frequency 2ω1 − ω2 is generated. Note
that for a = 0 and b = 0, the above terms are zero. The anharmonicity is the key to
generating the nonlinear optical response.
Note that the dispersion of the nonlinear susceptibility is described by a set of more
than one frequency. The dispersive property is characterized by the denominators in
(1.16)−(1.19). The profile of the 1/D(ω) is shown in Fig. 1.3a. Here, let us examine
the dispersive feature of the sum frequency generation, which is a resultant of the
two input waves with frequency ω1 and ω2 . By examining (1.18), one would see
that nonlinear susceptibility is resonantly enhanced when ω1 or ω2 is equal or nearly
equal as well as when the ω1 +ω2 is matched with the resonant frequency (Fig. 1.3b).
To make this chapter a more comprehensive introduction to nonlinear optical
response, it is important to highlight that current laser technology can generate strong
electric fields for which the perturbation theory no longer applies. Nonlinear optical
8 R. Arafune

Fig. 1.3 a Profile of 1/D(ω) (1.14) which represents the dispersion characteristics of the nonlinear
susceptibility. b Schematic diagram of energy level for the possible resonant processes (resonant
frequency: ω0 ) involved in SFG. The solid horizontal line and dotted line represent the stationary
state and the virtual state, respectively

responses originating from these strong electric fields are also important in science.
As noted above, the polarization created by the incident light field is sufficiently
described with the linear response alone when the electric field is very low. The
perturbation expansion expressed in (1.1) is valid when the electric field is not suffi-
ciently weak for a linear response, but we can still consider that the electric field
is sufficiently weak in the case of a nonlinear response. As a criterion for judging
whether the perturbation expansion is reasonable or not, the electric field felt by an
electron in an atom, a molecule, or a solid would be appropriate, which is approximate
109 V/m. When considering the interaction of matter with the light field whose inten-
sity is comparable to it, the perturbation expansion become inapplicable. Instead, as
an effective approach to describe the nonlinear responses for an atom under such a
high electric field, a model in which the laser electric field deforms the Coulomb
potential of an atom, and the atom is ionized by the electrons tunneling through the
deformed potential (tunnel ionization) is known [2]. While the atom is ionized, the
electron driven by the laser field can remain in the vicinity of the ion, and then the
electron can recollide with the ionized atom. As the result, the electron loses the
energy and is emitted as radiation. Harmonic generation originating from the tunnel
1 Introduction 9

ionization is referred to as high harmonic generation (HHG) and can create high-
energy photons (> 100 eV) and light pulses with the ultrashort temporal duration
(< 100 as). Thus, the HHG-related technique potentially provides the tabletop light
source that generates extreme ultraviolet light and leads to the technology for inves-
tigating material properties within an extremely short time scale. Indeed, there is an
active research field in which HHG is used. It is called attoscience [3]. At this time
scale, nuclear motion is almost negligible (recall that, within classical approximation,
the period of the electron motion in the ground state of the hydrogen atom occurs
over a time of 150 as). With few exceptions, there is currently vigorous research on
the direct observation and control of electron wave packets using isolated atoms and
molecules in a vacuum as the sample (this topic will not be covered in this book).

1.3 Pragmatic Importance of Nonlinear Spectroscopy

for Materials Science

Through the nonlinear optical response, one can reveal significant deviations from
the symmetry and/or harmonicity of the system. Here, let us discuss another impor-
tant aspect of nonlinear spectroscopy. It is difficult to achieve the high electric field
that leads to nonlinear optical responses by irradiating the conventional continuous
laser on the sample. It is well known that the pulsed light generated through tech-
niques such as mode-locking and Q-switching is suitable to achieve a higher electric
field. At present, pulsed light sources are essential equipment for nonlinear spec-
troscopy. Indeed, nonlinear optical spectroscopy had its origins with the invention
of the pulsed laser. The utilization of pulsed light which is an inevitable feature of
nonlinear optical responses caused an essential revolution in spectroscopy adding
the temporal resolution with very high resolution. And that revolution continues to
develop to the present day.
One of the interesting aspects of solid-state properties is their dynamical nature,
which unfolds on exceedingly brief time scales. Time, a fundamental variable in
physics and chemistry, plays a pivotal role. The temporal evolution of a particle
or wave packet is the center of the fundamental equations which describe the
phenomena, whether in classical or quantum mechanics. Comprehending these prop-
erties is not only crucial for fundamental scientific understanding but also holds
significant importance for engineering applications. For example, gaining essential
knowledge about the high-speed response of electrons, which carries information
bits in current information-processing devices, is important for the development of
new processing devices.
The short time scales discussed in this book are mainly those with a time resolution
of tens of femtoseconds to tens of picoseconds. What would a time scale of this
order (10 fs–10 ps) correspond to in materials science? One of the most typical
examples is nuclear motion. Consider a molecule as a simple system. Apart from
translational motion, molecules have several periodic degrees of freedom of motion,
10 R. Arafune

vibration, and rotation. Depending on the potential between atoms and the mass of
the atoms, the periodicity is of the order of picoseconds to femtoseconds. Similar
characteristics can be observed in solids, where the atoms remain in relatively well-
defined positions in a lattice. Their motion is limited as a phonon, which appears
with time on the order of subpicoseconds. “How does a change in the position of the
nucleus affect the electronic state?” is a naturally exciting and interesting question.
Furthermore, questions such as how long does the photoexcited electron relax its
energy on a given time scale? How does the electron move in real and reciprocal
space during the relaxation process? What symmetry does the electronic state of the
photoexcited system have? How is its time evolution described? These questions are
also fundamental and important.
Pulsed light is an ideal tool for studying ultrafast processes of various origins. It
provides a “snapshot” measurement of the time evolution of how a system triggered
by pulsed light relaxes. This technique, known as the pump-probe experiment, is
one of the most fundamental and widely used techniques for measuring physical
properties using pulsed light. In the pump-probe experiment, a pulsed light labeled
“pump pulse” or “excitation pulse” interacts with the sample and excites it to a new
non-equilibrium state. The sample then relaxes to an equilibrium state, which may
or may not be the same as before the pulsed light excitation. The relaxation process
to equilibrium can be mapped using a second pulse of light, often referred to as the
“probe pulse” or “test pulse”. In the typical pump-probe experiment, the probe pulse
is intentionally delayed relative to the pump pulse, and the signal from the probe
pulse is measured as a function of the delay time betwen two pulses. The temporal
limit of the experiment is determined by the duration of both the pump and probe
pulses. If the system relaxes to the equilibrium state before the pulse light is applied,
the repetition time of the pulse light can be adjusted to be longer than the time for
the system to relax to the equilibrium state, allowing repeated measurements and
obtaining spectra with a high signal-to-noise ratio.
Here, let us take transient absorption spectroscopy as an example, which was
developed at an early stage and is probably the most widely used time-resolved
optical measurement using the pump-probe technique. While it is classified under
the third-order nonlinear optical spectroscopy, the use of the formalism of nonlinear
spectroscopy is not required to interpret the experiment in most cases, because of its
brevity. To illustrate this technique as simple as possible, let us set up an ensemble
of two-level systems as our sample (Fig. 1.4). Here, we assume that all electrons
occupy the ground state at thermal equilibrium. Two pulses separated by a delay τ
are focused on the sample. One would see that the sample act as a saturable absorber.
In this experiment, the pump pulse will saturate the absorber (if the intensity of the
pump pulse is sufficiently high, the sample will be transparent). The delayed weak
probe pulse light will provide the absorption coefficient α in the current example:

α = σ ΔN , (1.21)

where σ is the absorption cross section and ΔN is the population difference between
the upper and lower level over the transition. After pump-pulse excitation, the sample
1 Introduction 11

Fig. 1.4 Schematic diagram of transient absorption spectroscopy

then relaxes to an equilibrium state. By considering a rate equation, the absorption

coefficient follows:
( )
τ
α(τ ) = α0 + Δα exp − , (1.22)
T1

where Δα describes the change in the absorption produced by the pump-pulse

light, T1 is the energy relaxation time of the electron in the transition state. Thus,
the dynamics of the relaxation to equilibrium can be mapped by measuring the
absorbance as the function of the delay.
Based on the above discussion, it can be understood that transient absorption
spectroscopy measures the light-triggered dynamics of the bulk properties of a
sample. However, with advancements in materials science and solid-state physics,
the non-bulk properties of materials, i.e., surfaces and interfaces, have become more
important. The second-order nonlinear optical spectroscopy is ideal for character-
izing surface and interface properties because they are inherently broken in their
central inversion symmetry. When the bulk properties of the sample have central
inversion symmetry, measuring the SHG and SFG signals allows for the selective
investigation of surface properties [4]. These nonlinear optical spectroscopies can
be readily integrated with the pump-probe technique. This allows for the examina-
tion of surface dynamics, such as symmetry change of the surface structure and the
diffusion of adsorbed molecules on the surface, with high temporal resolution. The
findings gained from these techniques are important for understanding the electron/
phonon dynamics at surfaces of solids, which are related to chemical reactions at
surfaces including catalytic reactions.
The pump-probe technique can be universally applied to investigate photo-
induced dynamics and has been combined with many spectroscopic methods. A
noteworthy combination is that with electron spectroscopy [5]. In this book, the
combination of angle-resolved photoemission spectroscopy (ARPES) [6] and the
pump-probe technique will be discussed, especially in terms of electron dynamics
such as the relaxation process of hot electrons near solid surfaces and coherence in
12 R. Arafune

multiphoton excitation processes. It should be noted that this “time-resolved photoe-

mission spectroscopy” often requires a compromise between energy and time reso-
lution. The energy resolution of current ARPES has reached the sub-100 meV order
[7]. On the other hand, as noted above, current techniques can provide light pulses
with sub-100 attosecond durations. Certainly, combining these two techniques would
be a very interesting subject. However, it should be noted that the resulting data from
this combination does not simply mean both ultra-high temporal resolution and high
energy resolution.
Time-resolved spectroscopy often provides beautiful (both scientifically and
graphically) data, which can significantly enhance our understanding of the decay
dynamics and coherence of excited electrons. The reader will recognize that many of
the interesting findings revealed in the following chapters are attributed to the prop-
erties of the excitation light source. In this context, time-resolved nonlinear X-ray
spectroscopy promises to offer distinctive insights into material dynamics. Realizing
the full potential of this technique hinges on further advancements in the light sources.
Hence the continued development of light sources is of paramount importance for
its successful implementation and continued progress.

1.4 Synopsys

This book is composed of six chapters. In this introductory chapter, the importance
of “non-something “ research field, the classical description of the nonlinear optical
response, and the significance of the nonlinear spectroscopy for material sciences
have been introduced, which are assumed to be prescribed. After this chapter, Chap. 2
guides various analyses of linear X-ray spectroscopy. These measurement methods
have been useful in materials science, and they have played a role as the foundation
of nonlinear X-ray spectroscopy. The journey into state-of-the-art in nonlinear spec-
troscopy commences in Chap. 3, and the photon energy used increases as the chapter
progresses. Chap. 3 provides the first example of the nonlinear spectroscopy, coherent
multiphoton photoemission spectroscopy. Photoemission spectroscopy has been one
of the standard methods to probe electronic states in matters, and the chapter reviews
its evolution toward the nonlinear regime. Integrating extremely highly controlled
laser technology with photoemission spectroscopy techniques has demonstrated the
ability to quantify the coherent response of solid-state materials. In Chaps. 4 and 5, the
frontier research of nonlinear spectroscopy with X-ray will be reviewed. Chapter 4
showcases the results of various experiments on nonlinear optical response using soft
X-rays. It is of note that the nonlinear response with a soft X-ray is essentially under-
stood by simply extending knowledge of those with an infrared ~ visible ray. On the
other hand, the macroscopic description for the nonlinear optical response for hard
X-ray is not fully useful, as noted above. Therefore, Chap. 5 that focuses on nonlinear
hard X-ray spectroscopy starts from the detailed theoretical description. After that
the important features discovered only through hard X-ray will be reviewed. Finally,
1 Introduction 13

in Chap. 6, we look ahead and provide insights into the future prospects of this
emerging field of research.
Before closing this chapter, it should also be essential to highlight that nonlinear
spectroscopy serves as a valuable domain for exploring nonlinear physics. Nonlinear
optics is strongly related to the important topics of nonlinear physics, such as Chaos
and Solitons. Many of these phenomena cannot be fully explained by solving linear
equations or by using perturbation theory. The powerful point of nonlinear optics is
to experimentally “visualize” the fundamental properties of a phenomenon that are
difficult to interpret analytically. Since material science itself is a complex field, the
integration of Chaos phenomena and material science remains a relatively unexplored
area and is still limited. We believe that this will become an increasingly important
research theme in the future.

Bibliography

1. Y.R. Shen, The Principles of Nonlinear Optics (Wiley, London, 1984).

2. T. Brabec, F. Krausz, Intense few-cycle laser fields: frontiers of nonlinear optics. Rev. Modern
Phys. 72 (2), 545 (2000); J. Eden, High-order harmonic generation and other intense optical
field-matter interactions: review of recent experimental and theoretical advances. Prog. Quantum
Electron. 28 (3–4), 197 (2004); H.C. Kapteyn, M.M. Murnane, I.P. Christov, Extreme nonlinear
optics: coherent X rays from lasers, Phys. Today 58(3), 39 (2005)
3. F. Krausz, M. Ivanov, Attosecond physics. Rev. Mod. Phys. 81(1), 163 (2009)
4. H.-L. Dai, W. Ho, Laser Spectroscopy and Photochemistry on Metal Surfaces (World Scientific
Publishing Company, Singapore, 1995)
5. U. Bovensiepen, H. Petek, M. Wolf, Dynamics at Solid State Surfaces and Interfaces (WILEY-
VCH Verlag GmbH & Co. KGaA, Weinheim, 2010)
6. S. Hüfner, Photoelectron Spectroscopy (Springer, Berlin, 2003)
7. K. Okazaki, Y. Ota, Y. Kotani, W. Malaeb, Y. Ishida, T. Shimojima, T. Kiss, S. Watanabe, C.-T.
Chen, K. Kihou, C.H. Lee, A. Iyo, H. Eisaki, T. Saito, H. Fukazawa, Y. Kohori, K. Hashimoto,
T. Shibauchi, Y. Matsuda, H. Ikeda, H. Miyahara, R. Arita, A. Chainani, S. Shin, Octet-line node
structure of superconducting order parameter in KFe2As2. Science 337(6100), 1314 (2012)
Chapter 2
Linear X-Ray Spectroscopy

Iwao Matsuda

Abstract This chapter starts with the basics of light–matter interactions, followed by
explanations of the linear X-ray spectroscopy, such as X-ray absorption spectroscopy,
photoelectron spectroscopy, and X-ray emission spectroscopy, with the experimental
examples. The chapter also focuses on the time-resolved measurement of X-ray
spectroscopy as one of the operando experiments. In this chapter, various analyses
with X-ray are summarized in the guide chart. Nowadays, experiments of X-ray
spectroscopy can be made at the X-ray facilities, such as synchrotron radiation and
X-ray free electron laser. At the end of the chapter, a typical procedure for applying
beamtime is introduced as a flow diagram.

2.1 Basics

2.1.1 Optical Responses of Materials in the X-Ray Region

We see things to understand them, as known from the proverb, “to see is to believe
(seeing is believing)”. When we study a material, we actually make a little change
(perturbation) to the target and observe the result. Even a direct visualization requires
illumination of light to an object (Fig. 2.1).
In a general experiment of materials science, the situation is described as [1–3]

P = ε0 χ E, (2.1)

where a physical quantity, P, is linearly generated by a gentle perturbation of E

to the sample (the linear response) and ε0 is the vacuum dielectric constant. The
response function, χ, contains material-specific information, and, thus, it leads to
an understanding of the object itself. For example, if E is the electric field, the
polarization P can be measured to derive the susceptibility, χ. When the system is

I. Matsuda (B)
The University of Tokyo, Tokyo, Japan
e-mail: imatsuda@issp.u-tokyo.ac.jp

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 15
I. Matsuda and R. Arafune (eds.), Nonlinear X-Ray Spectroscopy for Materials Science,
Springer Series in Optical Sciences 246, https://doi.org/10.1007/978-981-99-6714-8_2
16 I. Matsuda

Fig. 2.1 When there is no light, there is no sight. a Light-OFF. b Light-ON

homogeneous both in time and space, the macroscopic polarization can be written
as
∑∫ ( ) ( )
Pα (r, t) = ε0 dt ' d3 r ' χαβ r − r ' ; t − t ' E β r ' , t ' . (2.2)
β

With the vector and tensor representations, it can be written as

∫
( ) ( )
P(r, t) = ε0 dt ' d3 r ' χ r − r ' ; t − t ' · E r ' , t ' . (2.3)

( )
Due to the causality, χ r − r ' ; t − t ' = 0 when t – t ' < 0. If the spatial
change (temporal (variation ) of the envelope function) of E is smaller (slower)
than that of χ, E r ' , t ' can be replaced with E(r, t). By the definition, χ ≡
( ) ( )
∫ dt ' d3 r , χ r − r , ; t − t ' = ∫ dt '' d3 r ,, χ r ,, ; t '' , thus, the equation becomes simple
by the constant susceptibility tensor χ:

P(r, t) = ε0 χ · E(r, t). (2.4)

By the Fourier conversion, it is:

P(k, ω) = ε0 χ (k, ω) · E(k, ω). (2.5)

Focusing on the time and frequency dependence, χ (ω) can be given by the Fourier
transform of χ (t) as

∫∞
χ (ω) = dtχ (t)eiωt . (2.6)
−∞
2 Linear X-Ray Spectroscopy 17

The time dependence, t, is adopted for the causality, χ (t) = 0 when t < 0. In
general, χ (ω) is a complex number, χ (ω) = χ 1 (ω) + i χ 2 (ω), and it has a relation,
χ (−ω) = χ ∗ (ω). The real numbers, E(t) and P(t), are expressed as:

E(r, t) = E ω (r)e−i ωt + E ∗ω (r)eiωt . (2.7)

P(r, t) = ε0 χ (ω)E ω (r)e−iωt + ε0 χ ∗ (ω) E ∗ω (r)eiωt . (2.8)

Since χ (t) is a real number, χ 1 (ω) and χ 2 (ω) are even and odd functions of ω,
respectively.
The appearance is a consequence of light–matter interactions that provide infor-
mation on the target material. Let us consider an energy transfer between P of a mate-
rial and E of a light through the interaction. Temporal variation of the energy U p per
∂U
unit volume stored during the polarization of the medium is given by: ∂t p = ∂∂tP E.
Taking the average over a time sufficiently longer than the time period of the
electromagnetic wave, it becomes

∂U p
= 2ε0 ωχ 2 (ω)|E ω (r)|2 . (2.9)
∂t
∂U
By the light absorption in a material, ∂t p becomes positive and, thus, χ 2 (ω) > 0.
Since the incident photon intensity, I 0 , can be given by the cycle-averaged Poynting
∂U
vector magnitude, I0 = 21 ε0 c|E ω (r)|2 , the absorption intensity, I (∝ ∂t p ), has the
following relationship (c is the speed of light):

I ∝ I0 . (2.10)

When the optical properties of a material do not change with the incident light,
the response is linear and is described in term of the linear response function.
However, when the incident light intensity is extremely large, the polarization P
additionally contains the nonlinear terms, P (NL) , with nonlinear susceptibilities [4–6],
χ (2) , χ (3) , . . .:
∫
P (NL) (r, t) = ε0 dt1 d3 r 1 dt2 d3 r 2

χ (2) (r − r 1 , t − t1 ; r − r 2 , t − t2 ) : E(r 1 , t1 )E(r 2 , t2 )

∫
+ ε0 dt1 d3 r 1 dt2 d3 r 2 dt3 d3 r 3

χ (3) (r − r 1 , t − t1 ; r − r 2 , t − t2 ;
.
r − r 3 , t − t3 )..E(r 1 , t1 ) E(r 2 , t2 )E(r 3 , t3 ) + · · · (2.11)
18 I. Matsuda

.
It is of note that the expressions, “:” and “..” in (2.11) are the vector and tensor
representations, as shown in “·” in (2.3).∑For example, the term with the 2nd-order
susceptibility, χ (2) , is generally given as χi(2) (2)
jk E j E k . The quantity, χ , is the third-
(2)
order tensor or the 3 × 3 × 3 matrix, χi jk . The symbol “:” represents the interaction
with the two E vectors (E j and E k ), and it is adopted to abbreviate the summation
and the subscript expressions. The third-order susceptibility, χ (3) , is the fourth-order
tensor that reacts with three E vectors and, thus, the polarization term is express with
.
“..”. In a simple form, the total polarization, P (total) , can be given by a summation of
the linear, P, and the nonlinear, P (NL) , terms as

.
P (total) = P + P (NL) = ε0 χ (1) · E + ε0 χ (2) : E E + ε0 χ (3) ..E E E + · · · . (2.12)

Then, consequently, the optical characteristics depend on the intensity, and the
response becomes nonlinear. In this chapter, linear optical responses, such as light
absorption or light scattering, are described. Nonlinear optical phenomena, such
as nonlinear light absorption and stimulated light scattering, are described in the
subsequent chapters.
Interactions between light (electromagnetic wave) and matter include light absorp-
tion, light emission, and scattering processes. What we visualize with our eyes is a
consequence of visible-light optical events that cover the wavelength range between
380 nm (violet) and 770 nm (red). However, the visible-light region is very narrow
in nature, as shown in Fig. 2.2. Light that is invisible to our eyes, such as infrared,
ultraviolet, and X-rays, has a much wider range. The “invisible light” also exhibits
absorption, emission, and scattering processes as a result of light–matter interactions,
but the phenomena appear differently from those with visible light. For example,
when a material is irradiated with ultraviolet or X-rays, electrons can be excited out
of the sample (i.e., the photoelectric effect). It has been also well-known in radiog-
raphy that X-rays, transmitted through a human body, can unveil the bone structure.
These unique characteristics of X-ray have led to developments of various analytical
methods in materials science. For example, X-rays can be used to probe electronic
states and atomic structures of a material because the photon energy and wavelength
match the atomic scale as well as electron-binding energies, as shown in Fig. 2.2.
X-ray beams have been categorized into two regions: X-rays or hard X-rays
(0.1 nm or shorter) and soft X-rays (0.1 nm or longer). Conventionally, X-rays have
been used for structure analyses, while SX have been used to probe electronic states
of materials [7–19]. The nomenclature of the detailed wavelength range has been not
clear for the invisible regions. Conventionally, the wavelength range over 10–100 nm
has been called extreme ultraviolet (XUV, EUV), or vacuum ultraviolet (VUV), while
that over 0.1–10 nm is SX. To make it simple for the public, “soft X-ray (SX)” has
been used to cover both two ranges and this chapter follows the trend.
As shown in Fig. 2.2, X-ray photon energies can exceed the binding energies of
core levels, such as the 1s and 2p orbitals of atoms [7]. Because energies of the
inner-shell levels depend on the element and its chemical state, X-ray spectroscopy
2 Linear X-Ray Spectroscopy 19

Fig. 2.2 Relationship between energy states of light and matter, using molecules as an example.
The interaction depends on the energy level of each state

has been typically used to probe these levels. Because an X-ray beam exhibits light
absorption, emission, and scattering with materials, there are many experimental
techniques, such as X-ray absorption spectroscopy, X-ray emission spectroscopy,
and X-ray scattering [7–19]. At X-ray facilities over the world, such as synchrotron
radiation (SR), these methods are used for elemental/chemical analysis, structure
determination, electronic structure analysis, and spin/magnetic analysis.
For a comprehensive understanding in this chapter, I would like to classify these
analysis methods based on the semi-classical quantum theory. Focusing on spec-
troscopy, some of the methods are described in detail with the experimental spectra.
A summary chart of the methods is provided so that readers can seek for the appro-
priate one for their research. A flow for applying the user experiment at the X-ray
facility or the “beamtime” application is also introduced. This chapter has subsec-
tions, starting with explanations of X-ray sources (Sect. 2.1.2). When there is no light,
a matter is invisible (Fig. 2.1). We have neither light–matter interactions nor spec-
troscopy. Typical descriptions of light–matter interaction in experiments of materials
science are based on semi-classical quantum mechanics (Sect. 2.1.3). Spectroscopy
measurements have negligible effects on the target sample; thus, the interactions
can be quantitatively described by perturbation theory that supports linearity of the
optical response of the sample. Section 2.2 deals with the principles of absorption
20 I. Matsuda

and emission, with examples of X-ray absorption spectroscopy (Sect. 2.2.1), photo-
electron spectroscopy (Sect. 2.2.2), and X-ray emission spectroscopy (Sect. 2.2.3).
Section 2.3 presents a topic of the time-resolved experiment as an example of the
advanced spectroscopy that is related to the issue of the book. In the last section
(Sect. 2.4), materials analyses with X-ray are summarized as a guide chart based on
the fundamental light–matter interactions. I believe the chart should be useful for
readers to seek for the appropriate experimental approach.

2.1.2 X-Ray Sources

There are many (soft) X-ray sources today, such as synchrotron radiation (SR), X-ray
free electron lasers, and high harmonic generation in lasers (Fig. 2.3). These sources
have been fully incorporated in routine methods essential for materials science [20–
26].
A source has optical characteristics, such as monochromaticity, wavelength
tunability, coherence, intensity, brightness, degree of polarization, and pulse width,
that should be selected appropriately based on purpose of the experiment. In mate-
rials science, most samples have mm-scale sizes; however, only a μm ~ nm region
on the surface are typically irradiated with light during an experiment. Thus, we need
a bright beam source that irradiates this small area, rather than a point source that
emits light radially (such as the sun). It is thus useful to adopt the quantity brilliance
to compare the performance of light sources:

Fig. 2.3 Photographs of light sources and facilities: Laser at LASOR (Laser and Synchrotron
Research Center, The Institute for Solid State Physics, the University of Tokyo) and undulator at
SPring-8 BL07LSU. Synchrotron radiation (SPring-8) and X-ray free electron laser (SACLA). By
courtesy of Takeshi Suzuki, Kozo Okazaki, and Jiro Itatani. By permission of RIKEN
2 Linear X-Ray Spectroscopy 21
[ ]
a number of photons
[brilliance] = [ ]2 [ ]2 [ ]
[time] length angle divergence energy resolution

In addition to the photon flux ([a number of photons][time]−1 [length]−2 ), the

two-dimensional angular divergence of the beam and the energy resolution (ΔE/
E) are also included in the definition. The number of particles for “high brilliance”
corresponds to the number traveling with high directionality under conditions of
constant time, area, and energy resolution. For most experiments, it is desirable that
the light source has high brilliance. At the SR beamline, the brilliance is generally
expressed as [photons/s/mm2 /mrad2 with a 0.01% band width] or [photons/s/mm2 /
mrad2 /0.01% BW]. The beamline brilliance at the SPring-8 synchrotron radiation
facility is a factor of 1010 larger than that of sunlight on the earth’s surface, although
the emitting power of the sun is much enormous. Recently, it has become necessary
to define peak brilliance that represents brilliance per light pulse to characterize ultra-
short pulse lasers, such as X-ray free electron lasers. Experiments with such a light
source are associated with data-acquisition using a single shot, where information in
individual shots is critical. Conventional brilliance can be termed the average bril-
liance to distinguish it from the peak brilliance. Because of the steady development
of light sources, average or peak brilliance is updated over time; some of the data
are found in references or webpages of the SR facilities.
A laser light source is based on light amplification by stimulated emission of
radiation (LASER). The optical characteristics are (1) monochromaticity, (2) spatial
and temporal coherence, (3) femtosecond pulse widths in the case of mode-locked
oscillations, (4) high intensity, and, recently, (5) soft X-ray wavelengths from higher-
order harmonic generation.
Synchrotron radiation (SR) is light (orbital radiation) generated with charged
particles (electrons) at nearly the speed of light that change the motion (acceleration)
by bending electromagnets or periodic magnet arrays (undulators or wigglers). As
the light source, (1) it can continuously emit from infrared ray to X-ray; (2) it can
control linear and circular polarizations of the light; (3) it provides highly oriented
beam; (4) it has pulse widths of several tens of picoseconds; and (5) it can generate
the spatially quasi-coherent beam.
In a free electron laser (FEL), light generation and amplification are sequentially
made by interactions between electrons and light (SR) meandering in a periodic
magnet array. The FEL sources can be broadly divided into two types: a multi-pass
type that combines a laser resonator and an undulator, and a single-pass type that
is based on self-amplified spontaneous emission (SASE) in a long undulator. The
former FEL requires an appropriate optical mirror for the resonator, and the photon
energy is typically used up to the level of visible light. In contrast, the latter FEL
requires no mirror and can thus generate photon energies up to the X-ray region.
Existing SASE-FEL facilities provide the light sources of (1) wavelengths up to X-
ray, (2) monochromaticity, (3) femtosecond pulse duration, (4) ultrahigh intensity,
and (5) full spatial coherence. It is worth mentioning that the performance has been
22 I. Matsuda

Fig. 2.4 Typical

specifications of the light
sources. Pulse width/
wavelength characteristics of
lasers, synchrotron radiation,
and free electron lasers.
Example of the SPring-8 soft
X-ray beamline is indicated
in the diagram. SXFEL and
XFEL are the abbreviations
for soft X-ray free electron
laser and X-ray free electron
laser, respectively, while
HHG for high harmonic
generation

improved further by development of the FEL source that is seeded by a laser (seeded
FEL).
Figure 2.4 summarizes these light sources with respect to wavelength and pulse
width. The sources are for soft X-ray experiments, but the actual ranges are quite
distinctive. Lasers generate ultrashort optical pulses and extend photon energies to the
soft X-ray region by higher-order harmonic generation. Synchrotron radiation (SR)
produces ultrashort wavelengths, but the pulse width is in the 10 ps scale (typically
50 ps) due to the limitation of the electron bunch in the storage ring. Soft X-ray FEL
(SXFEL) and X-ray FEL (XFEL) cover both the ultrashort pulse and ultrashort wave-
length regions. As shown in Fig. 2.4, characters of the light sources are compensating
each other and also share some regions. Experimentalists select the most appropriate
X-ray source for their purposes. Nowadays, the cutting-edge researchers conduct
advanced measurement by a combination of the different light sources. For example,
one can make a series of time-resolved X-ray experiments using both XFEL and SR
to trace dynamical events from femtoseconds (fs) to microseconds (μs), capturing a
whole picture of the phenomenon. Knowledges of light sources have been significant
in optical science.

2.1.3 Light–Matter Interactions

Here, we focus on unique soft X-ray and X-ray regions. As noted above in Sect. 2.1.1,
soft X-rays include the vacuum ultraviolet (VUV) region.
2 Linear X-Ray Spectroscopy 23

When materials are irradiated with soft X-rays, particles (electrons, ions) with
various wavelengths and energies are generated, as shown in Fig. 2.5. These
phenomena are caused essentially by light absorption and scattering. A magnitude
of the light–matter interaction is expressed by a “cross section” that means the event
probability.
Figure 2.6 shows the dependence of photo-induced events in the X-ray energy
range. Absorption dominates in the soft X-ray (VUV) region, but the absorption
cross sections decrease with increasing photon energy. However, at a certain energy,
there is an increase (a jump) in the cross-sectional area. This is the absorption edge
of a core–shell, such as a K-shell, L-shell, or M-shell, and energy position of the
edge-jump is determined by each element. In the soft X-ray region, the contribution
of scattering, such as Rayleigh scattering, appears. Furthermore, in the X-ray region,
Compton scattering becomes dominant, and the scattering cross section becomes
larger than the absorption cross section.
Absorption and scattering events can be distinguished in terms of processes that
are associated with the vector potential A of the electromagnetic wave (light). A is a
vector quantity introduced to quantitatively describe the electromagnetic field and is
adopted as a parameter in the Hamiltonian of the light–electron interaction, as given
below. In the electromagnetic theory, the electric field E and magnetic field B have
the following relationships:

∂A
E = −∇φ − , (2.13)
∂t

B = ∇ × A, (2.14)

where φ is the scalar potential. Because physical properties of materials are mainly
determined by the electrons, the interaction between light and matter essentially
corresponds to the interaction between electromagnetic waves and electrons. The

Fig. 2.5 Interactions between matter and light (vacuum ultraviolet rays, soft X-rays, and X-rays):
The events can be classified as the one or two A process, where A is a vector potential of the light
24 I. Matsuda

Fig. 2.6 Schematic drawing

of the total optical cross
section of carbon atom at
various photon energy with
the contribution processes of
absorption (photoemission),
coherent scattering
(Rayleigh scattering), and
incoherent scattering
(Compton scattering) [7]

essence can be captured by the semiclassical theory that treats electronic states in a
material quantum-mechanically and electromagnetic fields classically [27–31].
The Schrödinger equation of the interaction between the electromagnetic field
and the electron is expressed in terms of A and φ:
{ }
1 ∂Ψ
HΨ = (p + eA)2 − eφ + V Ψ = i h , (2.15)
2m ∂t

where V is the potential in the material that electrons sense other than the electromag-
netic wave. By rearranging the terms in the equation, p = −ih∇, the Hamiltonian
H transforms as:

p2 e e2
H= +V + (p · A + A · p) − eφ + A·A (2.16)
2m 2m 2m

e e2
= H0 + (2A · p − i h∇ · A) − eφ + A · A, (2.17)
2m 2m
where H 0 is the Hamiltonian of an electron in a material that does not contain an
electromagnetic field. If we choose the radial gauge condition eφ = 0 and ∇ · A = 0
from the gauge invariance, we can make the Hamiltonian simpler and treat that
the electromagnetic wave is spatially uniform. This situation is realized in soft X-
ray experiments because the wavelengths are a factor of 10–100 greater than the
spread of electrons around atoms. Thus, A can be regarded as a constant in actual
e2
experiments, and the ∇ · A term is negligible. The 2m A · A term makes the dominant
2 Linear X-Ray Spectroscopy 25

contribution in the light scattering events in X-ray region (Fig. 2.6) [7]. To keep our
argument simple, we neglect the term at a moment. Then, the Hamiltonian becomes:
e
H = H0 + (A · p) = H0 + Hop . (2.18)
m
When the magnitude of the Hamiltonian H op is sufficiently small relative to H 0 ,
perturbation theory can be applied to deal with H op . This treatment corresponds to
the fact that light–matter interactions during measurements affect the sample trivially
and guarantee linearity in the optical response.
In quantum mechanics, the occurrence of phenomena is expressed as a probability.
Application of the perturbation theory provides a transition probability P from an
initial state |i⟩ to a final state |f⟩ by light–matter interactions with different orders.
First-Order Perturbation
|/ | e | \|2
| | | |
P1 ∝ | f | (A · p)|i | . (2.19)
m

Second-Order Perturbation
| (⟨ | )|
|∑ f | e (A · p)||n ⟩⟨n || e (A · p)||i ⟩ ⟨ f || e (A · p)||n ⟩⟨n || e (A · p)||i ⟩ |2
| |
P2 ∝ | m m
+ m m
| .
| n E i − E n + hν E f − E n − hν |
(2.20)

In these equations, E i , E f , and E n are the energies of the initial (i), final ( f ), and
intermediate (n) ˝states, respectively. ⟨f |H op |i⟩ is a matrix element that corresponds
to the integral Ψ ∗f Hop Ψi dxdydz, using the expression of the wave functions Ψ
(Ψ f ) for the initial (final) state. P1 in (2.19) corresponds to Fermi’s Golden Rule,
|⟨ | | ⟩|
P1 = 2π | f | Hop |i |2 ρ, where ρ is density of states at the energy of the final state. The
h
transition probability P1 is described by the square of the matrix element. P2 in (2.20)
is characterized by the appearance of an intermediate state |n⟩, and it corresponds
to the Kramers–Heisenberg–Dirac dispersion equation. It is of note that hν in the
equation is the incident photon energy.
Using quantum field theory, A is linear with respect to photon creation and annihi-
lation operators [27–31]. That is, the transition probability of a first-order perturbation
with one A (2.19) corresponds to the absorption or emission (e.g., fluorescence) of
light. Whereas the transition probability of the second-order perturbation contains
two A (2.20), and it corresponds to the scattering process. The terms in the formula
contain the energy difference in the denominator, and thus, the processes describe
the resonant scattering and resonant inelastic X-ray scattering (RIXS). It is of note
e2
that application of the first-order perturbation of the 2m A · A term, neglected earlier
from (2.17), results in giving the probability of the non-resonant scattering, such as
Thomson scattering. In a scattering process in a crystal, the term contributes to the
Bragg peaks in diffraction. It is also responsible for non-resonant inelastic scattering.
26 I. Matsuda

Fig. 2.7 Feynman diagrams of interactions between a photon and an electron. Time passes from
bottom to top, a and b correspond to light absorption and emission in the one A process, c–e are
the examples of the two A process corresponding to the light scattering. (c) is scattering in the first
perturbation (Thomson scattering), and d, e are scattering process by the second perturbation of
p·A via a virtual intermediate state n

The effect appears in high-energy X-rays, as shown in Fig. 2.6. Figure 2.7 summa-
rizes these processes with Feynman diagrams. Figure 2.7a, b correspond to (2.19),
while Fig. 2.7c represents Thomson scattering. Figure 2.7d, e describe the first and
second terms in (2.20), respectively.
Figure 2.8a–e illustrates events associated with the absorption process. When
the optical transition energy is below the vacuum level (a), electrons remain in the
transition to the unoccupied level, but when it exceeds the vacuum level (b), they are
emitted into the vacuum (photoelectric effect). Holes are generated at the core level
in processes (a, b), and both are accompanied by the secondary processes shown in
(c, d). Fluorescence emission (c) occurs when light is generated by the transition of
electrons in the valence band to the core level, and (d) occurs when electrons are
emitted into the vacuum by the non-radiative Auger process. Process (e) is called the
resonant photoelectric effect, which involves (a), (b), and (d). Thus, when the incident
light matches the energy difference between the core and the unoccupied levels, there
are electrons emitted in vacuum by the photoelectric effect and the Auger process
that have the same kinetic energy and share the same final states. In this case, the
final-state wavefunctions interfere constructively, resulting in enhancement of the
photoelectron intensity. Therefore, one can use the resonant photoelectric effect to
extract information on an element or an orbital of the probing electronic states.
Figures 2.8f–h are events of the scattering process. In Rayleigh scattering (elastic
X-ray scattering) in (f), electrons at the core level eventually make transitions to the
2 Linear X-Ray Spectroscopy 27

Fig. 2.8 Absorption/scattering processes, where A and k mean a vector potential and a wavenumber
of the light wave, respectively. a Absorption (electronic transition to an unoccupied state), b absorp-
tion (photoelectric effect), c fluorescence (secondary process of absorption), d Auger process
(secondary process of absorption), e resonant photoelectric effect [(a), (b), and (d) mixed process],
f Rayleigh scattering (elastic X-ray scattering), g resonant elastic X-ray scattering, h resonant
Raman scattering (resonant inelastic X-ray scattering)

original level through intermediate states. As a result, photon energy (wavenumber)

of the incident light and the emitted light are the same. When the intermediate state
coincides with the unoccupied state, as shown in (g), the resonance term in (2.20)
increases, resulting in significant enhancement of the scattering process. This process
is the resonance (resonant) scattering or resonant elastic X-ray scattering. There is
a case in the resonant scattering process, shown in (h), that the scattered light has
28 I. Matsuda

smaller photon energy (lower wavenumber) than the incident light. This event is the
resonance Raman scattering or resonant inelastic X-ray scattering. The energy loss
(Raman shift) carries information on a material, such as band gap in the electronic
states or the collective modes of phonons.
The interaction of light and electrons also yields information on the spin state of
the electrons. In that case, we consider the interaction Hamiltonian of the electron
and the electromagnetic field, including the spin σ, as follows [16]. These derivations
incorporate relativistic effects.
( )
' e e2 eh e2 h ∂A
Hop = (A · p) + A·A+ σ · (∇ × A) − σ· × A . (2.21)
m 2m 2m (2mc)2 ∂t

The transition probability of absorption and scattering process, for example, can be
obtained by assembling the perturbation terms of the one and two A process, respec-
tively. The principle is simple, but the calculation may be complicated. It is worth
mentioning that, in X-ray scattering, there is a magnetic scattering process because
of the magnetic interaction between the electromagnetic field and the electron spin
or the electron orbital angular momentum, as described in (2.21). The cross section
is reduced by a factor of hω/mc2 or (hω/mc2 )2 with respect to the photon energy hω,
relative to that for scattering attributed to charge. Note that mc2 = 511 keV, where m
is the electron rest mass. Therefore, the scattering intensity itself is extremely small.
Recently, however, high-intensity X-rays have been used to analyze the magnetic
structure of crystals. In the soft X-ray region, resonance magnetic scattering occurs
with X-rays having energies matching the absorption edge and the large scattering
signal can be obtained.
A simple description of semi-classical quantum mechanics can classify varieties of
X-ray analysis methods, based on absorption/emission (one A-process) and scattering
(two A-process). The relationships are summarized in Sect. 2.4 as a chart in an easy-
to-understand manner that will help one find the most appropriate approach. A deeper
understanding and details can be found in the references of the chapter. Some of the
X-rays spectroscopy methods are discussed in the following sections.

2.2 X-Ray Spectroscopy

2.2.1 X-Ray Absorption Spectroscopy

In an absorption experiment, the incident light intensity I 0 on the sample and the
intensity I of the light transmitted through the sample are measured. Based on the
Lambert–Beer law, absorption satisfies the relation:

I = I0 e−μρd . (2.22)
2 Linear X-Ray Spectroscopy 29

The exponent μρd corresponds to absorption coefficient μ (m2 /kg), density ρ

(kg/m3 ), and thickness d (m) of a sample.
One obtains:

I ∝ I0 , (2.23)

which confirms the linearity of the optical response. Because ρ and d can be known in
advance, μ can be obtained experimentally. The value can be expressed as the amount
per atom μ = NAwA σabs , where N A is the Avogadro coefficient, Aw is the atomic weight,
and σabs is the absorption scattering cross section per atom, introduced in Fig. 2.6
[7].
The penetration of soft X-rays through materials is very small; thus, it is difficult to
measure the intensity of transmitted X-rays in normal measurements. However, when
a material absorbs X-rays and emits mainly electrons, the emission is proportional
to the amount of absorption. Therefore, by measuring with the electron yield, it is
possible to obtain a soft X-ray absorption spectrum of a sample even in a case of no
transmission (Fig. 2.9). The total electron yield is used in many soft X-ray absorption
measurements because it is relatively easy to obtain spectra by measuring the current
flowing through the sample (specimen current or drain current). The emitted electrons
can include Auger electrons and secondary electrons in addition to photoelectrons.
The partial electron yield method selects these electrons at the specific energy range.
The fluorescence yield captures fluorescent X-rays emitted from the sample. It is
worth mentioning that yields of the electron and the photon provide different probing
depths of the signal. The measurement can be surface-sensitive (electron-probe) or
the bulk-sensitive (photon-probe). Thus, one can choose one of the yield methods to
select the probing sample region (surface or bulk) or one can combine the different
yield methods to capture a whole picture of the sample (surface and bulk).
X-ray absorption fine structure is a technique that can directly investigate unoc-
cupied states of a material [9–11]. Absorption increases when a certain energy is
reached, as seen for the absorption edge in Fig. 2.6. This corresponds to the photon

Fig. 2.9 Measurement

set-ups for X-ray absorption
spectroscopy
30 I. Matsuda

energy becoming large enough to excite an electron in a certain core level, and the
energy position depends on the element. In addition, a fine structure appears near the
absorption edge, as shown schematically in Fig. 2.10. X-ray absorption spectroscopy
can be used to analyze spectral features such as near-edge X-ray absorption fine struc-
ture (NEXAFS) or the X-ray absorption near-edge structure (XANES). When the
investigation ranges up to several hundred eV, it is called extended X-ray absorption
fine structure (EXAFS).
In NEXAFS, the transition matrix has the core level as the initial state |i⟩. The final
states |f ⟩ are (1) the unoccupied levels, (2) the Rydberg levels, or (3) the continuum
level, exhibiting the characteristic spectral shapes in each energy range (Fig. 2.10).

Fig. 2.10 Spectral appearance of the photoabsorption near the absorption edge for a binuclear
molecule in near-edge X-ray absorption fine structure (NEXAFS) and X-ray absorption fine structure
(EXAFS). E vac represents the vacuum level
2 Linear X-Ray Spectroscopy 31

Compared to Fig. 2.8a, b, the discussion has now extended to states above the vacuum
level. In region (1), the final state can be an antibonding orbital (σ *, π *) of a molecule
or a conduction band of a crystal. The Rydberg state in region (2) appears uniquely
near the vacuum level. These electrons far away from the nucleus are mainly affected
by interactions with the potential of the ion core, as shielded by the other electrons;
thus, they have orbits with high principal quantum numbers. In region (3), the final
state corresponds to an electron with a continuum energy that is partially trapped
by a potential of the material (molecule) or an electron wave that is scattered by
surrounding atoms. In solids, regions (1) and (3) are focused for analyzing a sample.
In the case of the region (1), Fig. 2.11 shows examples of NEXAFS spectra of
fullerene (C 60 ), highly oriented pyrolytic graphite (HOPG), and diamond. All the
materials show absorption edges at the carbon K-shell, and the spectral structures
appear differently. The features depend on the material and have reproducibility.
Because of this unique character, NEXAFS spectra have been used as fingerprints to
identify the material. Machine learning or big data analysis has been performed for
accurate material identifications [32, 33].
Figure 2.12a shows incidence-angle-dependent NEXAFS results for (a) HOPG
and (b) hexagonal boron nitride (h-BN), which are highly oriented because of the
stacking two-dimensional structures [34–37]. The peak structure near the absorption
edge corresponds to the unoccupied states (π *, σ *) of the material. By changing the
incident angle of the SR beam on the sample surface, one can find that the π * peak
at 285 eV and the σ * peak at 291 eV for HOPG changes significantly with respect to
the incident angle, and shows the opposite behavior. This is because the π *-orbital
spreads in a perpendicular direction to the honeycomb plane of HOPG, while the

Fig. 2.11 For various

carbon materials, near-edge
X-ray absorption fine
structure (NEXAFS) spectra
of fullerene (C 60 ), graphite,
and diamond, taken at the
K-shell absorption edge and
at normal incidence. By
courtesy of Masato Niibe
32 I. Matsuda

Fig. 2.12 Incidence-angle dependence of X-ray absorption spectra of a carbon K-shell absorption
edge in highly oriented pyrolytic graphite (HOPG) and b nitrogen-K-shell absorption edge in
hexagonal boron nitride (h-BN). The angle is measured from the sample surface normal with 0°
corresponding to normal incidence and 60° corresponding to oblique incidence. By courtesy of
Masato Niibe

σ *-orbital spreads in the plane. In another example, Fig. 2.12b shows the incident-
angle dependence of the K-shell soft X-ray absorption spectrum of nitrogen (N) in
h-BN. The π * and σ * peaks correspondingly exhibit the same behavior as seen for
graphite. In this way, the orientation of the molecular orbitals (or the direction of
chemical bonding) can be analyzed from the incident-angle dependence of the soft
X-ray absorption spectrum.
The electron, excited to the continuum level of (3), behaves as a photoelectron
wave and scatters between the X-ray absorbing atom (photoelectron emitter) and
surrounding atoms. This causes the electron interference in the final state, and the
EXAFS spectrum shows the intensity modulation with photon energy (Fig. 2.10). By
converting from a spectrum of the photon energy to that of the electron wavenumber
(k), its modulation structure ξ is generally expressed as:

∑ { }
1 2r j ( )
ξ (k) ∝ Nj exp − sin 2kr j + δ . (2.24)
j
(kr j )2 λ(k)

Here, the λ(k) is the decay length. The phase, kr, of the photoelectron wave changes
between atoms at distance r, and the absorption is modulated spectroscopically. This
is the EXAFS method, and, by analyzing this modulated structure, it is possible to
obtain information the local atomic structure, such as the coordination number N
and the bond distance r around the absorbing atom. The phase shift, δ, is determined
from the results of a standard sample using the same element as the scatterer. Because
ξ (k) is a sine function, the Fourier transform of EXAFS can also be used to obtain
a radial distribution function for structural analysis.
2 Linear X-Ray Spectroscopy 33

2.2.2 Photoelectron Spectroscopy

Photoelectron spectroscopy directly probes occupied valence states and core-level

states of a material [12, 13]. It is a spectroscopic application of the photoelectric
effect, in which a material is irradiated with light that has energy higher than the
work function W, and the emitted electrons or photoelectrons are analyzed (Fig. 2.13).
From the conservation of energy, kinetic energy (E k ) of the electrons in vacuum given
as a function of the photon energy hυ:

E k = hν − E B − W, (2.25)

where E B is the binding energy of electrons in the material based on the Fermi
level E F , and W generally has a value of 4–5 eV. The higher E B (deeper energy level)
electron requires the higher photon energy hυ. Ultraviolet (UV) to vacuum ultraviolet
(VUV) ray can emit valence electrons, and soft X-ray to X-ray can emit core-level

Fig. 2.13 Principle of photoelectron spectroscopy

34 I. Matsuda

electrons around the nucleus (Fig. 2.13). Electrons in the valence bands of solids
or in the highest-occupied molecular orbitals (HOMO) of molecules are critical for
processes such as electrical conduction and chemical reactions, and photoelectron
spectroscopy can investigate these states. The E B values of the core level depend on
the element, while the additional E B shifts (chemical shifts) depend on the valence
levels and the chemical environment. Therefore, chemical composition can also be
determined quantitatively from the core level E B and its photoelectron intensity.
Based on the information obtained from a material by photon energy, the former
is so-called ultraviolet photoelectron (photoemission) spectroscopy (UPS), and the
latter is X-ray photoelectron (photoemission) spectroscopy (XPS) historically or
core-level photoelectron spectroscopy (CLS).
While an energy level of an electronic state ψ(E) can be probed by photoemis-
sion spectroscopy, other parameters, such as momentum k or spin vector σ, can also
be directly evaluated by extensions of the methodology. Angle-resolved photoemis-
sion spectroscopy (ARPES) maps out band-dispersion curves of the electronic state
ψ(E, k), based on the momentum conservation rule. By installing a spin polarimeter,
spin-resolved experiments can be performed to determine the spin-polarized band
structure of the electronic state ψ(E, k, σ) of a sample. Recently, measurements
can also be conducted with spatial and temporal resolutions. Spatial mappings have
been performed by scanning focused X-ray beams or by imaging with photoelec-
tron microscopy to determine the electronic states ψ(E, k, σ, r) of a non-uniform
sample. The spatial resolution nowadays reaches in the nanoscale. Time-resolved
measurements of photoemission spectroscopy have recently been performed to trace
temporal evolutions of electronic states ψ(E, k, σ, r, t) during dynamic events in real
time. This topic is focused in Sect. 2.3.1 as one of the operando experiments in X-
ray spectroscopy. Details of the individual measurement methods of photoemission
spectroscopy can be found in [12, 13].
The photoelectron intensity I PES ([a number of photoelectrons][time]−1 ) for an
electronic state that is characterized by E and k can be simply expressed by taking a
product of transition probability and distribution of electrons over the possible range.
( ) ¨ |/ | | \|2
2π | |e | | ( )
IPES = | f | (A · p)|i | ρ(E) f FD (E, T )δ E f (k) − E i (k) − hν dkdE.
h m
(2.26)

IPES is described by a matrix transition probability (2.19), the Fermi–Dirac distri-

bution function f FD (E, T ), and density of states ρ(E). In the dipole approximation,
(2.26) can transformed to:
¨
( )2
IPES ∝ A0 ω f i |⟨ f |(er A )|i⟩|2
( )
× ρ(E) f FD (E, T )δ E f (k) − E i (k) − hν dkdE, (2.27)
2 Linear X-Ray Spectroscopy 35
( )
where ω f i = E f − E i /h. When the vector potential A = A0 exp{i (kr − ω f i t},
the electric field E is given by (2.13) and the amplitude E 0 = A0 ω f i . Because the
incident photon intensity I 0 is given by I0 = 21 ε0 cE 02 , one obtains:

IPES ∝ I0 , (2.28)

which confirms the linearity of the optical response.

Figure 2.14 collects spectra of atomic sheets, as examples of XPS [38–40]. The
sample spectra in Fig. 2.14a, b are obtained from a graphene layer on the hydrogen-
terminated SiC substrate. A survey XPS spectrum in Fig. 2.14a shows core-level
peaks of the carbon (C), silicon (Si), and oxygen (O) atoms. The result quantitatively
unveils not only the composing element but also the impurity. Intensity of each
peak corresponds to amount of each element in the sample. Figure 2.14b focuses
on the C 1s core-level region. The peak is composed of two components that are
assigned to carbon atoms of graphene and silicon carbide. The spectrum can be
decomposed by curve-fitting with Voigt functions. The function is convoluted by
a Lorentzian function that characterizes the material and a Gaussian function that
represents the experimental resolution. One can find that two chemical species appear
separately at different energy positions and thus XPS is significant for analyzing a
material. Figure 2.14c presents an XPS spectrum of a new material, the atomic sheet
of hydrogen boride (HB). Photoemission spectroscopy has also played a significant
role for material synthesis.

Fig. 2.14 a Survey spectrum of X-ray photoelectron spectroscopy of a graphene layer on the
hydrogen-terminated SiC substrate, taken at hυ = 650 eV (synchrotron radiation). b Core-level
spectrum at the C 1s peak in (a). Two carbon peaks are assigned to the graphene layer and the SiC
substrate. A spectrum is curve-fitted by the Voigt functions and the background. c B 1s core-level
photoemission spectrum of the hydrogen boride (HB) sheet, measured with a Mg X-ray tube. By
courtesy of Fumio Komori and Takahiro Kondo
36 I. Matsuda

2.2.3 X-Ray Emission Spectroscopy

In X-ray emission spectroscopy [9, 10, 14–16], X-rays are incident on a sample using
a monochromatic source such as a SR beamline, and the emitted X-rays from the
sample are dispersed by a diffraction grating to acquire the spectrum (Fig. 2.15). The
emission spectrum, which differs from the incident energy, contains fluorescence
and inelastic scattering components. X-ray fluorescence is a secondary process that
occurs after absorption, as it is generated by electronic transitions to form holes in
the inner shell (Fig. 2.8c). Because it is determined by the position of the core level,
the fluorescence energy is unique to the element (characteristic X-ray) and is useful
for elemental component analysis.
Carbon K-edge X-ray emission spectra of carbon allotropes excited by SR are
shown in Fig. 2.16. The excitation energy was 310 eV, a bit larger than the C K-edge.
This is the energy distribution of soft X-rays emitted when the 2p-orbital electron
transitions to the 1s vacancy according to the selection rule and basically reflects the
electron state density of the 2p orbital. Therefore, various electronic states for each
compound appear clearly as changes in the shape of the spectrum. From such soft
X-ray emission spectra, it is possible to analyze the electronic and chemical states
of occupied orbitals, which can be used for fingerprint analyses.
Figure 2.17 shows the takeoff-angle dependence of soft X-ray emissions at (a) the
C K-shell absorption edge of HOPG and (b) the N K-shell absorption edge of h-BN
[41]. In these figures, each emission intensities are normalized to keep the intensity of
the tail on the low energy side the same. With such normalization, the intensity of σ
emission on the lower energy side becomes approximately equal value. On the other
hand, it can be seen that the π emission intensity on the higher energy side shows
a large takeoff-angle dependence. This corresponds well with the incident-angle
dependence of NEXAFS seen in Fig. 2.12. The reason for the angular dependence

Fig. 2.15 X-ray emission

spectroscopy arrangement. A
typical emission
spectrometer consists of a
diffraction grating and a
detector
2 Linear X-Ray Spectroscopy 37

Fig. 2.16 A collection of carbon K-edge X-ray emission (fluorescence) spectra of carbonaceous
materials. By courtesy of Masato Niibe

is that the π orbital is perpendicular to the honeycomb plane and the σ orbital is
parallel to the plane, as in the same case of NEXAFS.
In Fig. 2.18, an example of the resonant inelastic X-ray scattering (RIXS) exper-
iment is given for a case of graphite (HOPG) [42]. Since excitation energy range of
RIXS corresponds to that of the absorption edge, the X-ray absorption spectrum of
the sample is collected at the C K-shell, as shown in Fig. 2.18a. The spectral peaks
at 285.4 and 291.8 eV are assigned to the photoexcitations to the π * and σ * states,
respectively, as presented in Figs. 2.11 and 2.12. A measurement of the sample was
made with the surface normal at 55° relative to the incident beam. Figure 2.18b is
the RIXS map that was recorded at the same geometry with the X-ray spectrometer
placed at 90° scattering angle. The prominent features at emission energies between
270 and 280 eV are from emissions (Figs. 2.16 and 2.17) and also from inelastic scat-
tering by the band transitions. In the map, a diagonal line of the elastic peak (zero
energy loss) appears as the white line. Focus on a region of the yellow box, one can
find inelastic peaks of phonon excitations, as shown in Fig. 2.18c. High-resolution
measurements of RIXS nowadays unveil not only electronic states but also collective
modes and the associated couplings in a material.
38 I. Matsuda

Fig. 2.17 a Takeoff-angle dependence of X-ray emission spectra of (a) carbon K-shell absorption
edge in highly oriented pyrolytic graphite (HOPG) and b nitrogen K-shell absorption edge in
hexagonal boron nitride (h-BN). The takeoff angle, α, is measured from the sample surface, with
15° corresponding to very small oblique angle. In this measurement, the X-ray incident and takeoff
directions are fixed at 90°, and the takeoff angle can be changed by rotating the sample around the
vertical axis. By taking the takeoff angle, α, as shown in the inset, the incident angle and the takeoff
angle are always expressed as the same value. By courtesy of Masato Niibe

Fig. 2.18 a An X-ray absorption spectrum at the C K-edge of a HOPG sample. b Results of resonant
inelastic X-ray scattering (RIXS), mapped with excitation and emission energies. c RIXS spectra
around the elastic peaks, observed at excitation energies between 291 and 292 eV (light yellow
box). Reprinted with permission from Feng et al. [42]. Copyright (2020) by the American Physical
Society
2 Linear X-Ray Spectroscopy 39

2.3 Time-Resolved X-Ray Spectroscopy

2.3.1 Measurement Principles

Measurements of X-ray spectroscopy have been used to characterize sample proper-

ties, such as structures or electronic states. Today, the information has been measured
to examine functionalities of a material by in situ or operando experiments under
the working condition. Operando X-ray spectroscopy measurements have been
performed, for example, during electrochemical operations in a battery, redox reac-
tions at the catalytic surface, or photovoltaic events in a solar cell [34]. There have
been significant developments in sample environments in an experimental chamber
of X-ray spectroscopy, such as electrochemical cells and near-ambient-pressure cell,
and also in measurement techniques with temporal and/or spatial resolution(s). In
this section, the time-resolved measurement is focused since the technique is related
to experimental techniques of nonlinear X-ray spectroscopy, which is the issue of
this book. For readers who wish to learn about the operando experiments extensively,
selected issues of the advanced X-ray spectroscopies, such as ambient-pressure or
nanospace spectroscopy, can be found in [34, 43].
A time-resolved measurement is conducted to track temporal evolution of a
dynamic event in real time [38, 43–53]. Interests of chronometric research can be
ephemeral and also eternal. The clock time can be as short as femtoseconds (fs,
10–15 s) and as long as centuries (100 years). In human society, the former is required
for fast communications of information, while the latter is needed for the safe storage,
for example. Such an enormously wide range of the possible target is one of the diffi-
culties in designing an experimental set-up for the time-resolved measurement. In
general, a researcher adopts a system that covers a certain range of time and experi-
mentally investigates a dynamical phenomenon of the sample that has a critical step
in the temporal region.
In materials science, research of the fs-time scale phenomena has been conducted
to investigate elementary process, such as electron–electron interactions, electron–
coherent phonon coupling, and electron–phonon coupling. The temporal region is
also significant to capture transient states or bond-breaking/making moments in
chemical reaction. Experiments of ultrafast time-resolved spectroscopy have been
demanded to directly monitor such ultrafast events. On the other hand, a chem-
ical reaction, in general, is composed of various steps, such as diffusion, formation
of reaction intermediates and desorption. A whole reaction process is complicated
and takes over time scales that are longer than picoseconds (ps). It is of note that
most chemical reactions are governed by kinetics, and it is necessary to reveal the
rate-determining step. Therefore, comprehensive and detailed understandings of a
chemical reaction require the time-resolved experiment for a wide range of time that
could even take over days. It is, thus, necessary to design a time-resolved experiment
that can be completed in reasonable but enough time to achieve the research goal.
Figure 2.19 summarizes experimental information at various time scales. Indi-
vidual time is indicated with distance that light travels during the corresponding
40 I. Matsuda

Fig. 2.19 Chronical structures at various time scales (distance that light travels during a period
of time). Optical specifications of X-ray sources, dynamic events, and measurement methods are
indicted. Abbreviations are as: attosecond, fs: femtosecond, ps: picosecond, ns: nanosecond, μs:
microsecond, ms: millisecond, s: second, ks: kilosecond, and Ms: megasecond. HHG, XFEL, and SR
indicate high harmonic generation, X-ray free electron laser, and synchrotron radiation, respectively

period. One finds that timing of the fs-ps range can reasonably be controlled by a
delay stage. A distance of m ~ km at μs matches with sizes of the SR facilities in the
world, and the period corresponds to a lap time of electrons in the storage ring.
Pulse widths of X-ray lasers, XFEL and HHG, range around fs, and the sources
can be used for time-resolved experiments to trace ultrafast dynamic phenomena.
For the measurement, one adopts the pump-probe method that generates a transient
state of a sample by an optical pump pulse, followed by an optical probe to detect
signals from the sample. To obtain the enough sample information, the measurement
is carried out repeatedly. A period of the cycle is needed to be longer than the recovery
time of an event to keep the same initial condition (before pumping). The repetition
rate typically ranges Hz ~ MHz. It is worth mentioning that there are two types
of the XFEL resources that operate at frequencies of the Hz and (sub)-MHz range,
indicated as the XFEL-pulse intervals #1 and #2 in Fig. 2.19, respectively. The former
is useful for making ultrahigh pulse energy of a pump, while the latter is suitable for
increasing detection cycles of a probe.
In a case of the synchrotron radiation, the pulse width (50–100 ps) is limited by
an operation of the electron storage ring. The SR pulse interval is determined by
distance between the electrons (electron bunches) that emit SR pulses, and it ranges
typically in ns-scale. By making an operation of the several bunch or single bunch
mode, the interval can be tuned up to the lap time (typically μs). During the SR
2 Linear X-Ray Spectroscopy 41

beamtime for a user, typically awarded for hours ~ days, time-resolved experiments
have been conducted to investigate dynamic events in various time scales (sub-ns ~
ks).
In general, designing of a time-resolved experiment starts by choosing time struc-
ture of the pump, probe, and detection system. Figure 2.20 shows examples of the
combinations. A pump is needed as a trigger to initiate a dynamic event. As shown
in Fig. 2.20a, one can trace a sample before pumping (Pump OFF), during pumping
(Pump ON), and after pumping (Pump OFF) with a continuous probe and a detection
system that operates fast enough to complete the measurement at the various stages
of pumping. In this scheme, the temporal resolution is limited by the detection time.
The measurement design has been adopted to trace irreversible phenomena. When
a pump is a short pulse, one can consider making a single detection after a delay
time, as presented in Fig. 2.20b. The scheme is useful when temporal width of the
pump pulse is much shorter than the detection time window. In a case of examining
reversible phenomena, the measurement can be repeated in cycles to accumulate
enough signals in the detector. Inversely, one can set a pulsed probe, instead of
continuous one, and keep the detection continuous. This scheme, Fig. 2.20c, is the
pump-probe method. Since time resolution of the method is determined by temporal
width of the pump and the probe, it is useful to capture ultrafast phenomena when a
pump and a probe are ultrashort optical pulses. In a typical time-resolved experiment
of such an ultrafast dynamic event, time range of the measurement is set as an order
of the optical pulse width. It is thus clever to limit the detection time, Fig. 2.20d,
to eliminate noises that are detected during the interval. Comparing the schemes in
Fig. 2.20 with the temporal structure of the light sources, shown in Fig. 2.19, one can
find that X-ray lasers of XFEL and HHG are suitable for the pump-probe methods,
Fig. 2.20c, d, while SR can be used for all the methods.
Figure 2.21a shows an experimental system using a laser with the high harmonic
generation (HHG). A timing jitter between pulsed beams of a pump and a probe can
be suppressed by sharing the same laser source. A soft X-ray probe is generated by
injecting the laser pulse in to the HHG system (see Fig. 2.22 for detail). To improve
efficiency of HHG, photon energy (wavelength) of the original laser pulse is often
doubled (halved) by a β-BaB2 O4 or BBO crystal before the injection, as shown in
Fig. 2.22. The delay time is controlled by changing the optical path by a movable
stage.
Figure 2.21b shows a schematic drawing of an electron storage ring in a SR
facility. The storage ring is fundamentally composed of (i) bending magnets that
ensure revolutions or cycles of electron bunches and simultaneously induces radiation
of continuous spectrum, (ii) an undulator that radiates quasi-monochromatic light,
and (iii) a radio frequency (RF) cavity that accelerates the electrons (replenishes
energy to the electrons). Since SR originates from individual electron bunches, the
emitted SR light has an inherent nature of pulsed light; the temporal width of SR light
corresponds to the size of an electron bunch and the repetition rate is determined by
integer division of the frequency at the RF cavity.
Figure 2.21c presents the basic principle of a free electron laser (FEL) of the
oscillator-type or the multi-pass type. Electrons in a bunch gain or lose their energy
42 I. Matsuda

Fig. 2.20 Examples of measurement schemes for time-resolved experiments operated by a combi-
nation of the pump, probe, and detection system. a A pulsed pump, continuous probe, and cycles
of detection. Operations of the elements are described by “ON” and “OFF.” b A short pump pulse,
continuous probe and a single-time detection. The delay time, Δt, is defined between pump and
detection time. c Short pulses of a pump and a probe with continuous detection. The delay time, Δt,
is defined between the pump and the probe. The scheme corresponds to the pump-probe method.
d The pump-probe method with limited time for detection of (c) that has been adopted to reduce
the background noise at a detector

through interaction with electromagnetic field, as in the case at the RF cavity. In

an undulator, when the oscillation period of electrons matches with a frequency of
light (electromagnetic wave) injected through the undulator, the energy is transferred
resonantly between electron and light. Choosing a proper phase shift between two
oscillations, the intensity of SR light increases when it passes through the undulator
again. The SR beam, reflected by appropriate mirrors, is amplified by repeating
the cycle and it is, eventually, transmitted to a sample. It is a sort of a resonant
cavity, within which the electrons exchange energy with a trapped radiation field.
It is of note that a generation scheme of FEL is fundamentally distinct from that of
conventional quantum lasers which utilize the population inversion in a laser media.
An advantage of FEL is continuous tunability of photon energy by changing the
undulator parameters, which is in contrast to the discrete choice of photon energy
for quantum lasers. Due to the technical limitation of mirrors, a wavelength of the
2 Linear X-Ray Spectroscopy 43

Fig. 2.21 A collection of pulse light sources for X-ray. a A laser system with high harmonic
generation (HHG), b undulator in an electron storage ring of synchrotron radiation (SR), c an
oscillator-type free electron laser (FEL), d a combination of laser and SR, e the laser-slicing source,
and f self-amplified spontaneous emission (SASE)-type X-ray FEL [43]
44 I. Matsuda

Fig. 2.22 Example of a set-up for time-resolved photoelectron spectroscopy experiment with high
harmonic generation of laser. Soft X-ray HHG pulses are generated by irradiating the laser beam at
noble gases, followed by selections of photons at specific energy by multi-layered mirrors. Pump
pulses are guided by another path with delay stage to make the delay time [43, 45, 48]. By courtesy
of Takeshi Suzuki

oscillator-type FEL typically ranges from the millimeter wave to the visible light and
a pulse width is restricted by the electron bunch.
As summarized in Fig. 2.19, there are characteristic chronical parameters for
the individual electron storage rings at various SR facilities. The time structure is
composed of three key “clocks”: a lap time, an interval, and a duration of the electron
bunch. One revolution (a lap time) of an electron bunch in the storage ring, running at
the velocity of about light speed, takes an order of microseconds (μs). Time interval
between neighboring bunches depends on the electron filling pattern in the ring. The
minimum time interval is determined by the frequency of the RF cavity, typically
about nanoseconds (ns). The pulse width is determined by the bunch length, and
it is in an order of 10 picoseconds (ps). One can recognize that the time structure
of electron storage rings is hierarchic and the time layer is almost three orders of
magnitude different from the neighbors. Using the SR pulses as a probe, a pump-
probe method has been performed over the SR facilities with a femtosecond laser
pulse as a pump. Figure 2.21d illustrates the beam paths from the two light sources
toward a sample. It is of note that the timing jitter affect the temporal resolution when
two sources are different, especially in fast time scale. Thus, the most critical factor
in the pump–probe experiment is the timing control between pump laser and probe
SR. Through recent technical developments at the SR facilities, the timing is now
adjusted better than 10 ps and the time-resolution is now limited by the SR pulse
width. Details of the timing control between SR and laser pulses are described in
[43, 46].
2 Linear X-Ray Spectroscopy 45

To achieve time resolution better than picoseconds, there exist several schemes
which generate femtosecond (fs) ultrashort pulses in the vacuum ultraviolet (VUV),
soft X-ray, and X-ray region. As an example of fs x-ray pulse generation, the “laser-
slicing” scheme is shown in Fig. 2.21e. In the modulator section, electron bunch
is shot in the magnetic field, such as in a wiggler, by a high-intensity ultrashort
laser pulse to induce a very short electron modulation in an electron bunch, forming
the non-uniform electron distribution. In the subsequent passage through magnet
field in an undulator or a bending magnet (the radiator section), the spiked fraction
radiates ultrashort SR pulse, while the remnant part generates the (coherent) SR.
This “laser-slicing” technique is currently under operation at various SR facilities in
the world [47]. Although the laser-slicing technique stably generates fs light pulses,
the actual measurement using the beam is quite challenging due to very low photon
flux. The situation is easily understood by a simple estimation. Repetition frequency
of the high-intense laser is typically kHz, which is 1/100,000 of the SR frequency
(typically on the order of 100 MHz) and a ratio of the spiked region to the total
electron bunch is 100 fs/ 50 ps = 1/500. Thus, photon flux of the laser-slicing fs light
pulses is about 10–8 of that of the usual SR pulses.
Demands of the intense fs pulses of VUV, soft X-ray, and X-ray have been
increased recently. It has been discussed that an interaction between light and electron
is the key phenomenon to overcome the technical difficulties. As described above,
the oscillator-type FEL has not been extrapolated to X-ray wavelengths yet due to
the limited efficiencies and power-handling capacities of X-ray mirrors. Instead, a
scheme to generate X-ray laser without mirrors has been identified (a single-pass
FEL) [23, 24]. Traveling through a sufficiently long undulator, electrons in a bunch
are subjected to the combined force of undulator magnets and their own SR beam.
Then, they begin to form microbunches, spaced by one wavelength of the dominant
SR. Constructive interference of the radiation from these microbunches produces an
X-ray beam with full transverse coherence. The instantaneous or peak brilliance is
ten billion-fold greater and the pulse width (< 10 to 100 fs) is thousand-fold shorter
than SR. Such FEL amplification is traditionally called “self-amplified spontaneous
emission” (SASE). Nowadays, the technical developments have been carried out and
we have varieties of advanced XFEL sources today.

2.3.2 Examples of Time-Resolved Photoemission

Spectroscopy Measurement

Figure 2.22 shows an example (Fig. 2.21a) of the time-resolved photoelectron spec-
troscopy experimental system using a laser with HHG [43, 45, 48]. With the laser that
has the pulse width of femtoseconds, the time-resolved photoemission spectroscopy
measurement system can capture temporal changes in electronic states during ultra-
fast phenomena. Titanium sapphire (Ti:S) laser, which is a general ultrashort pulse
laser, has a wavelength of 800 nm (hυ = 1.55 eV) and the energy is doubled by BBO
46 I. Matsuda

(hυ = 3.10 eV). The laser is injected into a jet of noble gas in the HHG chamber,
followed by monitoring by a spectrometer and by monochromatizing by multi-layer
mirrors. In this time-resolved photoemission spectroscopy system, soft X-rays with
light energies of hυ = 20–60 eV are used for experiments and electronic states of a
sample have been investigated near the Fermi level.
Time-resolved photoemission spectroscopy allows us to directly observe changes
in the actual carrier dynamics as the occupancy in electronic states or bands. These
data can be band images obtained by wavenumber-resolved (angle-resolved) photoe-
mission spectroscopy, measured at various delay time (Δt) between the pump light
and the probe light. Figure 2.23 is an example of such temporal evolutions of the band
diagrams that are observed near the K point of a graphene layer on a SiC substrate
[45].
In general, an energy dispersion curve of the (nearly) free electron is quadratic
but in a case of graphene, the curve is linear, as called the Dirac cones, and the
crossing point is called the Dirac point. Therefore, electrons in graphene behave as if
massless. Such a Dirac band can be directly observed by photoelectron spectroscopy
as shown in Fig. 2.23, and it can be confirmed that the Dirac point exists at binding
energy = 0.0 eV (Fermi level).
Before the pump light irradiation, Fig. 2.23a, only the lower Dirac cone is
observed, but after the pump light irradiation, (b–f), the upper Dirac cone can also be
seen due to the electron excitation to the unoccupied states. Furthermore, by exam-
ining the change in photoelectron intensity for each Δt, the relaxation process of
non-equilibrium carriers in the Dirac band can be tracked quantitatively. As shown
in Fig. 2.23, the electrons in the upper Dirac cone relax in the femtosecond time scale

Fig. 2.23 Results of femto- to picosecond time-resolved photoemission spectroscopy of graphene

layers on a C-face SiC substrate. The measurement was performed by the pump-probe method,
where a pump is a laser beam of hυ = 1.55 eV and a probe is the HHG beam of hυ = 27.9 eV.
a–f Ultrafast temporal changes of (a–f) the photoemission band diagrams and g–k the difference,
taken at various delay time, Δt, between a pump and a probe [45]
2 Linear X-Ray Spectroscopy 47

Fig. 2.24 Example of a set-up for time-resolved photoelectron spectroscopy experiment at the SR
beamline, SPring-8 BL07LSU [43, 46]

and also accumulate near the Dirac point, where the relaxation slows down. Quan-
titative understanding of specific relaxation times is important for the development
of graphene devices, and this research provides material dynamics information from
theory to industrial applications [45].
Figure 2.24 illustrates an example (Fig. 2.21d) of the time-resolved photoelec-
tron spectroscopy system at the SR beamline, SPring-8 BL07LSU [38, 43, 46]. By
installing the femtosecond laser system at the beamline, one can conduct the pump-
probe experiment with pump (laser) and probe (SR). Timing controls of the pump and
probe are made electrically based on the master oscillator that controls the motion of
the electron bunches in the storage ring. Using the soft X-ray beam, one can detect
not only valence bands but also core levels in a material. As a result, even in non-
uniform samples, it has the characteristic of being able to trace time changes in an
element-selective manner.
Figure 2.25 shows a series of (sub)nanosecond time-resolved core-level photo-
electron spectra of graphene layers on the C-face SiC substrate [38]. By using a
photon beam at hυ = 740 eV as a probe, one can measure photoelectron spectra
of both the C 1s core level at binding energy of ~ 285 eV and Si 2p with binding
energy of ~ 100 eV. Two peaks in the C 1s spectra are assigned to carbon atoms
in the graphene (G) layer and the SiC substrate, respectively. In this way, the infor-
mation of the graphene layer and the substrate can be measured simultaneously. By
optical pumping with the laser, there is an energy shifts of all the core-level peaks
due to generation of the surface photovoltage effect and to the following relaxation.
The measurement unveils temporal variations of the energy shift at components and
elements in the functional interface that directly links to understanding of carrier
dynamics [38, 43, 49, 50].
This chapter section gave examples of the time-resolved experiment of photoe-
mission spectroscopy with synchrotron radiation (SR) and the laboratory-based laser
(HHG). It should be mentioned that such measurements of the time-resolved photoe-
mission spectroscopy can also be conducted at the XFEL facility, FLASH, today [51].
48 I. Matsuda

Fig. 2.25 Results of nanosecond time-resolved photoemission spectroscopy of graphene layers (G)
on a C-face SiC substrate. The measurement was performed by the pump-probe method, where a
pump is a laser beam of hυ = 3.1 eV and a probe is the SR beam of hυ = 740 eV. a and b are
temporal changes of the C 1s and Si 2p core-level spectra with respect to delay time (Δt) [38]

The experiment had been considered difficult due to the space charge effect [52]. The
effect is associated with the ultrahigh and ultrashort XFEL pulse, resulting in high
densities of photoelectrons. Due to the significant Coulomb interaction (repulsion)
between electrons, the photoemission spectra are significantly disturbed. However,
with the special mode of the XFEL operation with the MHz-frequency, time-resolved
measurement of angle-resolved photoemission and core-level photoemission spec-
troscopies have been made successfully [51]. Similar to the photoemission spec-
troscopy stations at beamlines of the SR facilities, the XFEL facility is becoming to
open the beamline station for the time-resolved photoemission spectroscopy [53].
2 Linear X-Ray Spectroscopy 49

2.4 Researcher’s Guide to Material Characterization

with X-Rays

2.4.1 The Guiding Chart for Experiments

There are many measurement methods using light and particle signals generated
by X-ray irradiation [7–19]. They have often confused researchers who seek for
the appropriate methods for their experiments. In Figs. 2.26 and 2.27, methods are
summarized with fundamental sample analyses, such as elemental/chemical analysis,
structure determination, electronic state analyses, and spin/magnetism analyses. The
classifications are absorption (one A-process) and scattering (two A-process), as
described in Sect. 2.1.
In the optical experiments, such as elemental analysis and electronic-state
measurements, it seems that the same information can be obtained by the different
methods. However, in practice, we analyze electrons emitted by excitation light and
use light scattered by electrons in materials. Because electrons and photons have
different probe lengths, the former is more useful for analyzing material surfaces,
and the latter is more useful for analyzing material interiors. The latter is also suitable

Fig. 2.26 Interactions between light and electrons (one A-process) and the relationship with each
measurement method
50 I. Matsuda

Fig. 2.27 Interactions between light and electrons (two A-process) and the relationship with each
measurement method

for measurements under external fields, such as magnetic fields. In this way, spec-
troscopy and diffraction methods can be used selectively according to the research
purpose and, by assembling all the dataset on the same sample, one can perform a
diversified analysis. Usage of X-ray beams benefit by adopting electrons as the mate-
rial signals, as shown in Fig. 2.5. The detected electrons carry various information,
like receiving an e-mail from the sample (Fig. 2.28).

Fig. 2.28 X-ray

spectroscopy is like getting
an e-mail from a sample
2 Linear X-Ray Spectroscopy 51

Fig. 2.29 A user flow diagram for the beamtime experiment at the X-ray facilities

2.4.2 Application for Beamtime at the X-Ray Facility

There are many facilities for X-ray experiments, such as synchrotron radiation and
X-ray free electron laser, all over the world. Some are in Japan, as shown in Fig. 2.3.
One can use these valuable and cutting-edge experimental resources. Therefore, I
would like to introduce the actual “user flow” below and in Fig. 2.29.

(1) Consider the measurements and experiments.

(2) At the facilities, the incident beam source is a “beamline” that is designed and
developed to match with the purpose of the experimental station. The beamline
information can be found, for example, in the website of the facilities. Hence,
consult with the person in charge of the beamline.
(3) Submit a proposal of the research project by the deadline.
(4) Project proposals will be reviewed by a committee of the experts in the research
fields. Because the operational time (the user beamtime) in the dedicated facility
is limited, the proposal can be rejected. In that case, the committee provides the
useful feedback comments. Then, restart from (2). The staff in charge of the
beamline will be happy to consult with the researcher.
(5) When the project proposal is awarded, the researcher will receive a notifica-
tion of acceptance, and, in general, will be assigned a project number with the
beamtime schedule.
(6) In a dedicated facility, procedures such as chemical management and radiation
safety training are performed so that experiments can be conducted safely and
smoothly. After successfully completing these steps, the researcher will go to
the beamline where the experimental station is located.
(7) The experiment will then be conducted during the designated period (beamtime).
(8) Analyze data obtained during the beamtime, and, if necessary, perform theo-
retical simulations of the data. Then, discuss and summarize the results with
co-researchers.
(9) Present the research results at academic conferences and publish them as
research papers to share the valuable information. Academic communications
via presentations and publications provide positive feedbacks in the research,
deepen the understanding, and stimulate the subsequent research. In addition,
the researcher is requested to submit a report of the beamtime to the facility
52 I. Matsuda

where the experiments were conducted. On the facility side, such information
is compiled and used to improve the future usage.
At the beginning of this chapter, the well-known proverb, “To see is to believe
(seeing is believing),” was mentioned. While the proverb is in English, it is common
to the whole world and there is a Japanese version. In Japan, the proverb was extended
by people to provide more lessons and can be expressed as “One-eye witness is better
than many hearsays.” One may extend it for a researcher as:
“One-eye observation is better than many hearsays.”

“One consideration is better than many observations.”

“One action is better than many considerations.”

“One publication is better than many actions.”

It is more important “to observe by yourself” than “to just listen to others a lot,”
and “to consider on your own” is more important than “to just observe many things.“
Then, it is more important “to take real actions” than “to only consider in your
brain,” and it is worth having publications to share with others than “to just keep it
in oneself.” I hope readers will perform more and more research to solve individual
problems or mysteries, and then share the findings or knowledges in the public.

2.5 Summary

In the present chapter, I introduced varieties of linear X-ray spectroscopy with the
basic principles that have been used to analyze electronic (chemical) states and struc-
ture of materials. Then, the analyses are summarized in a chart diagram so that a reader
can seek for the appropriate ones for the experiment. It is important to mention again
that some of the analyses, such as absorption spectroscopy, can be made with different
types of the signal detections, such as electrons and photons. Since these signals have
the unique probing lengths, one can made the surface-sensitive (electron-probe) and
the bulk-sensitive (photon-probe) measurements to capture a whole picture of a
sample. A choice of the probe is an important and interesting feature of the X-ray
experiments. It is also significant to note that a combination of the different anal-
yses and methods provides the multi-dimensional data that unveils a sample from
different aspects. These concepts have led to developments of operando experiments
that investigates electronic (chemical) states, for example, of functional materials
during the working condition. X-ray spectroscopy has become the useful analysis
method in both academic and industrial fields. The experiment can be made as the
user beamtime at the X-ray facilities in the world. I wish readers to utilize the chapter
and to conduct their research of materials science.

Acknowledgements I appreciate Masahito Niibe, Masafumi Horio, Masashige Miyamoto,

Junichiro Mizuki, Yoshihito Tanaka, Yasuhisa Tezuka, Fumio Komori, Takeshi Suzuki, and Takahiro
2 Linear X-Ray Spectroscopy 53

Kondo for valuable discussion, careful readings, and data usages for this chapter. This work was
supported by JSPS KAKENHI grants (Grants No. JP21H05012 and No. JP18H03874) and by JST,
CREST Grant No. JPMJCR21O4, Japan.

References

1. J. D. Jackson, Classical Electrodynamics (Wiley, London, 1998)

2. N. Mermin, N. Ashcroft, Solid State Physics (Brooks/Cole, Pacific Grove, 1976)
3. C. Kittel, Introduction to Solid State Physics (Wiley, London, 2004)
4. R.W. Boyd, Nonlinear Optics (Academic, Cambridge, 2020)
5. G. New, Introduction to Nonlinear Optics (Cambridge University Press, Cambridge, 2011)
6. V.V. Shtykov, S.M. Smolskiy, Introduction to Quantum Electronics and Nonlinear Optics
(Springer, Berlin, 2020)
7. X-RAY DATA BOOKLET Center for X-ray Optics and Advanced Light Source}, Lawrence
Berkeley National Laboratory. http://xdb.lbl.gov/
8. J.A. Samson, D.L. Ederer, Vacuum Ultraviolet Spectroscopy (Academic, Cambridge, 2000)
9. F. de Groot, A. Kotani, Core-Level Spectroscopy of Solids (CRC Press, Boca Raton, 2008)
10. F. de Groot, High-resolution X-ray emission and X-ray absorption spectroscopy. Chem. Rev.
101, 1779 (2001)
11. J. Stohr, NEXAFS Spectroscopy (Springer, Berlin, 2003).
12. S.Hufner, Photoelectron Spectroscopy: Principles and Applications (Springer, Berlin, 2003)
13. S. Suga, A. Sekiyama, Photoelectron Spectroscopy: Bulk and Surface Electronic Structures
(Springer, Berlin, 2014)
14. F. Gel’mukhanov, H. Agren, Resonant X-ray Raman Scattering (Elsevier, Amsterdam, 1999)
15. A. Kotani, S. Shin, Resonant inelastic X-ray scattering spectra for electrons in solids. Rev.
Mod. Phys. 73, 203 (2001)
16. L.J.P. Ament, M. van Veenendaal, T.P. Devereaux, J.P. Hill, J. van den Brink, Resonant inelastic
X-ray scattering studies of elementary excitations. Rev. Mod. Phys. 83, 705 (2011)
17. E. Zolotoyabko, Basic Concepts of X-Ray Diffraction (Wiley, London, 2014)
18. D. McMorrow, J. Als-Nielsen, Elements of Modern X-ray Physics (Wiley, London, 2011)
19. J. Miao, T. Ishikawa, I.K. Robinson, M.M. Murnane, Beyond crystallography: Diffractive
imaging using coherent X-ray light sources. Science 348, 530 (2015)
20. D. Attwood, Soft X-Rays and Extreme Ultraviolet Radiation (Cambridge University Press,
Cambridge, 1999)
21. H. Wiedemann, Synchrotron Radiation (Springer, Berlin, 2002)
22. S. Mobilio, F. Boscherini, C. Meneghini, Synchrotron Radiation: Basics, Methods and
Applications (Springer, Berlin, 2014)
23. C. Pellegrini, A. Marinelli, S. Reiche, The physics of X-ray free-electron lasers. Rev. Mod.
Phys. 88, 015006 (2016)
24. K.-J. Kim, Z. Huang, R. Lindberg, Synchrotron Radiation and Free-Electron Lasers: Principles
of Coherent X-Ray Generation (Cambridge University Press, Cambridge, 2017)
25. P. Jaegle, Coherent Sources of XUV Radiation: Soft X-Ray Lasers and High-Order Harmonic
Generation (Springer, Berlin, 2005)
26. D. Popmintchev, B.R. Galloway, M.-C. Chen, F. Dollar, C.A. Mancuso, A. Hankla, L. Miaja-
Avila, G. O’Neil, J.M. Shaw, G. Fan, S. Ališauskas, G. Andriukaitis, T. Balčiunas, O.D. Mücke,
A. Pugzlys, A. Baltuška, H.C. Kapteyn, T. Popmintchev, M.M. Murnane, Near- and extended-
edge X-ray-absorption fine-structure spectroscopy using ultrafast coherent high-order harmonic
supercontinua. Phys. Rev. Lett. 120, 093002 (2018)
27. J.J. Sakurai, Modern Quantum Mechanics (Benjamin-Cummings Publishing Co., Subs. of
Addison Wesley Longman, San Francisco, 1985)
54 I. Matsuda

28. L. Schiff, Quantum Mechanics (McGraw-Hill, New York, 1969)

29. H. Rubin, Landau (Wiley, Quantum Mechanics II, 1995)
30. A. Corney, Atomic and Laser Spectroscopy (Clarendon Press, Oxford, 1977)
31. B. Henderson, G.F. Imbusch, Optical Spectroscopy of Inorganic Solids (Oxford, 2006)
32. T. Mizoguchi, S. Kiyohara, Machine learning approaches for ELNES/XANES. Microscopy 92
(2020)
33. Y. Suzuki, H. Hino, M. Kotsugi, K. Ono, Automated estimation of materials parameter for X-
ray absorption and electron energy-loss spectra with similarity measures. NPJ Comput. Mater.
5, 39 (2019)
34. I. Matsuda (ed.), Monatomic Two-Dimensional Layers: Modern Experimental Approaches for
Structure, Properties, and Industrial Use (Elsevier, Amsterdam, 2019)
35. I. Matsuda, K. Wu ed., 2D Boron: Boraphene, Borophene, Boronene (Springer, Berlin, 2021)
36. X. Zhang, Y. Tsujikawa, I. Tateishi, M. Niibe, T. Wada, M. Horio, M. Hikichi, Y. Ando, K.
Yubuta, T. Kondo, I. Matsuda, Electronic topological transition of 2D boron by the ion exchange
reaction. J. Phys. Chem. C 126, 12802 (2022)
37. M. Niibe, M. Cameau, N. T. Cuong, O. Ilemona Sunday, X. Zhang, Y. Tsujikawa, S. Okada, K.
Yubuta, T. Kondo, I. Matsuda, Electronic structure of a borophene layer in rare-earth aluminum/
chromium boride and its hydrogenated derivative, borophane. Phys. Rev. Materials 5, 084007
(2021)
38. T. Someya, H. Fukidome, N. Endo, K. Takahashi, S. Yamamoto, I. Matsuda, Interfacial carrier
dynamics of graphene on SiC, traced by the full-range time-resolved core-level photoemission
spectroscopy. Appl. Phys. Lett. 113, 051601 (2018)
39. X. Zhang, M. Hikichi, T. Iimori, Y. Tsujikawa, M. Horio, K. Yubuta, F. Komori, M. Miyauchi,
T. Kondo, I. Matsuda, Accelerating synthesis of topological borophane by the HCl assisted ion
exchange reaction. Molecules 28, 2985 (2023)
40. H. Nishino, T. Fujita, N.T. Cuong, S. Tominaka, M. Miyauchi, S. Iimura, A. Hirata, N.
Umezawa, S. Okada, E. Nishibori, A. Fujino, T. Fujimori, S. Ito, J. Nakamura, H. Hosono,
T. Kondo, Formation and characterization of hydrogen boride sheets derived from MgB2 by
cation exchange. J. Am. Chem. Soc. 139, 13761 (2017)
41. M. Niibe, N. Takehira, T. Tokushima, Electronic structure of boron doped HOPG: selec-
tive observation of carbon and trace dope boron by means of X-ray emission and absorption
spectroscopy, e-J. Surf. Sci. Nanotech. 16, 122 (2018)
42. X. Feng, S. Sallis, Y.-C. Shao, R. Qiao, Y.-S. Liu, L.C. Kao, A.S. Tremsin, Z. Hussain, W.
Yang, J. Guo, Y.-D. Chuang, Disparate exciton-phonon couplings for zone-center and boundary
phonons in solid-state graphite. Phys. Rev. Lett. 125, 116401 (2020)
43. S. Yamamoto, I. Matsuda, Time-resolved photoemission spectroscopy with synchrotron
radiation: Past, Present, and Future. J. Phys. Soc. Jpn. 82, 021003 (2013)
44. Y. Hayashi, Y. Tanaka, T. Kirimura, N. Tsukuda, E. Kuramoto, T. Ishikawa, Acoustic pulse
echoes probed with time-resolved X-ray triple-crystal diffractometry. Phys. Rev. Lett. 96,
115505 (2006)
45. T. Someya, H. Fukidome, H. Watanabe, T. Yamamoto, M. Okada, H. Suzuki, Y. Ogawa, T.
Iimori, N. Ishii, T. Kanai, K. Tashima, B. Feng, S. Yamamoto, J. Itatani, F. Komori, K. Okazaki,
S. Shin, I. Matsuda, Suppression of supercollision carrier cooling in high mobility graphene
on SiC(0001). Phys. Rev. B 95, 165303 (2017)
46. M. Ogawa, S. Yamamoto, Y. Kousa, F. Nakamura, R. Yukawa, A. Fukushima, A. Harasawa, H.
Kondo, Y. Tanaka, A. Kakizaki, I. Matsuda, Development of soft X-ray time-resolved photoe-
mission spectroscopy system with a two-dimensional angle-resolved time-of-flight analyzer at
SPring-8 BL07LSU. Rev. Sci. Instrum. 83, 023109 (2012)
47. S. Khan, K. Holldack, T. Kachel, R. Mitzner, T. Quast, Femtosecond undulator radiation from
sliced electron bunches. Phys. Rev. Lett. 97, 074801 (2006)
48. T. Suzuki, S. Shin, K. Okazaki, HHG-laser-based time- and angle-resolved photoemission
spectroscopy of quantum materials. J. Electron Spectrosc. Relat. Phenom. 251, 147105 (2021)
49. K. Ozawa, M. Emori, S. Yamamoto, R. Yukawa, Sh. Yamamoto, R. Hobara, K. Fujikawa,
H. Sakama, I. Matsuda, Electron−hole recombination time at TiO2 single-crystal surfaces:
influence of surface band bending. J. Phys. Chem. Lett. 5, 1953 (2014)
2 Linear X-Ray Spectroscopy 55

50. K. Ozawa, S. Yamamoto, R. Yukawa, R.-Y. Liu, M. Emori, K. Inoue, T. Higuchi, H. Sakama,
K. Mase, I. Matsuda, What determines the lifetime of photoexcited carriers on TiO2 surfaces?
J. Phys. Chem. C 120, 29283–29289 (2016)
51. D. Kutnyakhov et al., Time- and momentum-resolved photoemission studies using time-of-
flight momentum microscopy at a free-electron laser. Rev. Sci. Instrum. 91, 013109 (2020)
52. A. Pietzsch et al., Towards time resolved core level photoelectron spectroscopy with
femtosecond X-ray free-electron lasers. New J. Phys. 10, 033004 (2008)
53. I. Matsuda, Y. Kubota, Recent progress in spectroscopies using soft X-ray free-electron Lasers.
Chem. Lett. 50, 1336 (2021)
Chapter 3
Probing Nonlinear Light–Matter
Interaction in Momentum Space:
Coherent Multiphoton Photoemission
Spectroscopy

Marcel Reutzel, Andi Li, Zehua Wang, and Hrvoje Petek

Abstract Optical fields interacting with metal surfaces can drive collective free
electron plasma currents and single-particle dipole excitations. The outcome of these
coherent interactions is immediately evident in nearly perfect images that appear in
metal mirrors. Yet, performing quantum state-resolved measurements and electronic
charge/spin actuations of metals is extremely challenging because of interaction-
driven decoherence of the polarization of the electronic system. In this chapter,
we describe interferometrically time-resolved multiphoton photoemission spec-
troscopy as a quantum state-resolved method for investigating coherent responses
of solids. We perform 3D (energy, momentum, and time)-resolved nonlinear photo-
electron spectroscopy to record the coherent response of single crystal surfaces with
attosecond accuracy. The measurements resolve the coherent polarization oscillations
in response to the driving light field, the resonant frequencies of the metal sample, and
the optical field-induced dressing of the electronic band structure. As a model system,
we focus on the coherent responses of single crystal silver and copper surfaces to
intense ~20 fs laser pulses in the IR-UV spectral range. Noble metals provide well-
known band structures, deeply studied attosecond collective responses, and simple
surface preparation methods that guide the development of experimental and theoret-
ical extension of coherent nonlinear photoemission spectroscopy to more complex
materials. To photoexcite electrons above the vacuum level in a nonlinear manner,
the excitation creates Floquet ladders of quasi-energy states up to fifth order in the
driving light field, including signatures of above threshold photoemission. Finally,
we show that at high driving field amplitudes, electrons follow both the space and
optical field-dependent periodic potentials causing non-perturbative modifications

M. Reutzel (B)
I. Physikalisches Institut, Georg-August-Universität Göttingen, Göttingen, Germany
e-mail: marcel.reutzel@phys.uni-goettingen.de
A. Li · Z. Wang · H. Petek (B)
Department of Physics and Astronomy and IQ Initiative, University of Pittsburgh, Pittsburgh, PA,
USA
e-mail: petek@pitt.edu

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 57
I. Matsuda and R. Arafune (eds.), Nonlinear X-Ray Spectroscopy for Materials Science,
Springer Series in Optical Sciences 246, https://doi.org/10.1007/978-981-99-6714-8_3
58 M. Reutzel et al.

of metal crystal electronic structures. Hence, nonlinear interferometric photoelec-

tron spectroscopy enables to quantify, as well as to transiently modify, the coherent
responses of solid-state materials.

3.1 Introduction

The microscopic understanding of light–matter interactions in the condensed phase

is a key ingredient for the implementation of the next-generation opto-electronic
devices. In this context, recent efforts strive for the understanding of coherent light–
matter interaction and the creation new and emergent material phases under the
irradiation with femtosecond light-pulses [1, 2]. Under intense illumination, light
pulses do not only induce dipole transitions and drive intraband currents, but can
also significantly perturb the materials electronic band structure, create new photon-
dressed electronic band dispersions, and change the correlated many-body interac-
tions between the electron–phonon-spin subsystems [3–16]. Experimentally, there
are multiple approaches to probe such light-dressed material phases. On the one
side, all-optical techniques have identified the dominant timescales of ultrafast mate-
rial control [17]. Complementary, distinct changes in the electronic, phononic, and
spin subsystems have been identified using time- and angle-resolved photoelectron
spectroscopy (trARPES) [18] and ultrafast electron scattering techniques [19].
In this chapter, we focus on the application of highly nonlinear spectroscopies
to address such questions. High-harmonic generation (HHG) has been shown to be
extremely powerful to characterize the nonlinear material response [20–25]. In these
experiments, intense light pulses in the mid-infrared to terahertz regime impinge
onto the sample surface, and the emitted HHG light is detected and analyzed.
While the experiment provides a manifold insight onto microscopic processes, as
any photon-in-photon-out technique, its information is integrated over the energy–
momentum coordinates of the crystalline solid. Hence, it lacks specificity to the
energy–momentum-dispersive electronic band structure, or, more generally, the spec-
tral function, in which all the information of the single-particle band structure and
many-body interactions is imprinted in the photoemission experiment [26].
In this chapter, we review the complementary approach of multiphoton photoemis-
sion spectroscopy (mPP) [27–36]. In contrast to HHG, where the nonlinear response
of the condensed matter material is probed by the detection of high-frequency
photons, mPP is a photon-in-electron-out technique. As such, mPP measures the
system’s response by detecting photoelectrons with energies much larger than the
driving light-field frequency. By varying the driving light frequency, the number
of photons m necessary to overcome the work function of the material of interest
changes systematically (Fig. 3.1). Additional to this mth-order outcome, in a process
termed above threshold photoemission (ATP), (m + n)th-order replicas of the main
photoemission peak can be detected. Often, experiments detecting highly nonlinear
mPP have been performed on sharp metal tips or nanostructured surfaces where the
nonlinearity can be increased by plasmonic fields or lightning-rod enhancements
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 59

[36–45] with the goal to analyze non-perturbative light–matter interactions. Signifi-

cantly less effort has been reported to employ mPP to well-defined crystalline solids,
where the parallel momentum information is preserved [27–35]. Importantly, as the
highly nonlinear variant of trARPES, mPP is directly sensitive to the energy and the
in-plane momentum of the photoelectrons. As such, it carries distinct information on
the periodically driven energy–momentum-dispersive band structure. In this chapter,
we discuss the experimental basics of interferometrically time-resolved mPP (ITR-
mPP) and explore its application in the perturbative regime of coherent light–matter
interaction where it probes optical dipole transitions. Finally, we discuss the applica-
tion of ITR-mPP to study the onset of non-perturbative light–matter interaction and
provide an outlook toward Floquet engineered material phases.

ATP
6
two-color

Ev
Usp
4
IP1
E-EF (eV)

2PP
2
1PP 3PP
4PP

SS
0 EF
Lsp
M Γ K
-0.2 -0.1 0.0 0.1 0.2
k|| (Å −1)

Fig. 3.1 Excitation diagram of mPP and ATP based on the example of the Ag(111) surface. The
surface-projected band gap (white area) energetically separates the upper and lower sp-bands (L sp
and U sp ) and hosts the Shockley surface state (SS) and the first image potential state (IP1). By
tuning the photon energy (of the vertical arrows), the lowest order of photoemission m to overcome
the vacuum level increases from 1 to 4 from left to right. Higher-order replicas of the photoemission
spectral features can be detected in ATP, i.e., in nonlinear order of m + n (shown for n = 1 with
dashed red and orange arrows). In addition, a two-color mPP (m = 2) experiment is indicated by
the red and the blue arrow. Adopted from Ref. [27]
60 M. Reutzel et al.

3.2 Experimental Setup

Angle-resolved photoemission spectroscopy (ARPES) has a long tradition as a probe

of the spectral function of a manifold of material classes [18, 46, 47]. While the basics
of ARPES are detailed in many monographs, such as the book by Hüfner [48], recent
experimental developments and case studies are summarized in the review on the
beginners [47] and the expert [18] level. In the following two sections, we will briefly
review the necessary experimental requirements from the extension of ARPES to a
highly nonlinear driving scheme.

3.2.1 Experimental Setup: The Photoelectron Analyzer

Modern photoelectron analyzers are designed to provide direct access to the

energy and at least one in-plane momentum coordinate of the photoemitted single-
particle electrons. Here, we will briefly highlight hemispherical energy analyzer and
momentum microscope methods and their demonstrated applications in mPP exper-
iments. In mPP, and, more generally, trARPES, the conventional approach employs
hemispherical electron energy–momentum analyzers that are combined with a two-
dimensional photoelectron detector to provide direct access to the emission angle
Θ (i.e., the in-plane momentum, k || ) and the energy E of the photoelectrons; exem-
plarily mPP (m = 2) data obtained on pristine Ag(111) with such an setup are shown
in Fig. 3.2a [27, 49]. Complementary, time-of-flight momentum microscopes can be
used to simultaneously collect multidimensional data sets that contain information on
both in-plane momentum coordinates of the photoelectrons and their energy [50, 51];
exemplary mPP (m = 2) data measured on the pristine Ag(110) surface are shown in
Fig. 3.1b [52, 53]. Because of the higher-dimension photoelectron detection scheme,
the momentum microscope has some advantages over the hemispherical electron
analyzer. The momentum microscope approach, however, is severely constrained in
regard to space charge electron repulsion effects, which limit its dynamic detection
range [54–56]. In highly nonlinear mPP experiments that involve high photoelectron
currents, hemispherical electron analyzer detection may be preferable.

3.2.2 Experimental Setup: The Light-Source

On the side of the light source, naturally, there are many parameters that can be
optimized for the desired experiment. High-resolution ARPES experiments are opti-
mized for a detailed band mapping of the occupied band structure and the extraction
of many-body band renormalizations. These experiments are typically performed at
large-scale synchrotron facilities that offer high-photon flux in the VUV-soft X-ray
range, tunable photon energy, and small bandwidth [18, 26, 57].
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 61

(a) Ag(111) k|| = 0 (b)

x1000
IP
7.0 100
Final State Energy (eV)

x1 100
k|| = 0.18 —1
SS

6.0

1
sp—
band

5.0
0
—0.2 —0.1 0.0 0.1 0.2 Intensity
k|| ( —1)

Fig. 3.2 Exemplary mPP (m = 2) data measured with an (a) hemispherical electron analyzer (b) and
a time-of-flight momentum microscope. a In 2PP with 3.32 eV, the hemispherical energy analyzer
provides direct access to the final state energy and the in-plane momentum of the photoemitted elec-
trons collected from the Ag(111) surface. The bulk sp-band transition (U sp ←− L sp ), the occupied
SS state, and a weak signal from the IP1 state are labeled in the energy-distribution-curve taken at
k || =0 Å−1 . b The time-of-flight momentum microscope provides three-dimensional access to the
photoemission data as a function of final state energy and two in-plane momenta. When excited
with 3.72 eV photons, the accessible momentum range is limited by the photoemission horizon. The
spectral signatures of surface states (S1, S2, S3) and bulk transitions (B1, B2, B3, B4) are labeled
in the figure and their assignments are reported in Ref. [52]. a and b reproduced from [27, 52]

In order to study ultrafast and coherent light–matter interaction, photoemission

spectroscopy has to be performed with femtosecond laser pulses in a pump-probe
scheme. Early realizations of such experiments, however, were limited to nanosecond
laser pulses to investigate the occupied and unoccupied electronic band structure of
pristine metal surfaces [58, 59]. With the advent of femtosecond laser technology,
time-resolved two-photon photoemission spectroscopy (tr2PP) has been the experi-
ment of choice to probe the coherent and incoherent electron dynamics in different
material classes [60–65]. In the implementation of a two-color experiment, a pump
laser pulse of one color occupies an initially unoccupied state that is probed by a time-
delayed laser pulse of different color (Fig. 3.1). In this way, the inelastic lifetimes
1/T 1 of, for example, image potential states on pristine and adsorbate covered metal
surfaces have been studied in detail [62, 66–68]. Newer implementations of tr2PP,
nowadays largely termed trARPES, combined optical pump pulses with extreme
ultraviolet probe pulses that facilitate the access to large in-plane momenta and thus
to the full momentum range of the surface Brillouin zone [4, 5, 15, 69–73].
In order to access the coherent response of a material to a light stimulus, and,
specifically, to experimentally extract the decoherence time 1/T 2ij of a coherent
polarization field built up between two states i and j, the mPP experiment has to
62 M. Reutzel et al.

ITR

Final State Energy (eV)

6.0
1

—1
k|| = 0
5.0 0

-0.1 0 30 45 60
0.1 -30 -15 0 15
k|| ( —1
) Delay (fs)

Fig. 3.3 Artistic illustration of the 3D data set that is collected in ITR-mPP (m = 4) on pristine
Ag(111) (1.40 eV photon excitation). The ITR-mPP experiments collect snapshots of the photoe-
mission intensity as a function of the photoelectron energy, the in-plane momentum, and the pump-
probe delay. As the delay is varied in 50 as steps, constructive and destructive interferences lead
to the modulation of the photoemission intensity on the time scale of the optical cycle (back panel
displayed for k || =0 Å−1 ). At a detuned three-photon resonance between the IP1 and the SS states,
both states are separately resolved on the energy axis, whereas the in-plane momentum width of
the SS state is limited by its occupation range (upper feature) and the IP1 state data is limited by
the acceptance angle of the photoelectron detector

be performed with an interferometric pump-probe scheme, i.e., so-called interfero-

metrically time-resolved mPP (ITR-mPP) [27, 60, 74–80]. In common implementa-
tions of a single-color experiment, a phase-locked pump-probe pulse pair is created
by passing an ultrashort laser pulse through a Mach–Zehnder interferometer. By
varying the delay of one interferometer arm with interferometric precision, sub-50
as resolution movies of the energy- and momentum-dependent coherent electron
dynamics can be resolved (Fig. 3.3). As such, this experiment has been applied
to study surface plasmon polariton propagation [42, 81, 82], coherent light–matter
interaction on metals [49, 75, 80, 83, 84], adsorbate covered surfaces [85–87], and
in coherent photoemission electron microscopy [88–95]. In an alternative approach,
recent efforts used birefringent wedges [96] to measure ITR-mPP data, promising a
more straightforward alignment of the optical setup, and an increased stability of the
interferometric measurement.

3.3 mPP: Highly Nonlinear Mapping of the Energy-

and Momentum-Dispersive Electronic Band Structure

Static mPP can be used to map the occupied and unoccupied energy–momentum-
dispersive band structure of crystalline solids. In this section, we show the potential
that it provides on the example of the Shockley surface state and the Rydberg series
of image potential states on the pristine Cu(111) and Ag(110) surfaces. First, we
describe the mPP process with the threshold nonlinear order of m that is necessary to
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 63

overcome the material’s work function. Second, we systematically characterize the

contribution of above threshold photoemission (ATP) to the mPP spectra. Finally, we
motivate the application of highly nonlinear mPP to measure coherent light–matter
interaction at large in-plane momenta, i.e., throughout the full surface Brillouin zone.

3.3.1 Static mPP in Threshold Order of Photoemission

In the (111) projection of the electronic band structure of Cu, the Shockley surface
state (SS) and the n = 1 image potential state (IP1) reside in the L-projected band gap
with energy minima at the ⎡-point of the surface Brillouin zone; higher-order states
of the Rydberg series of image potential states such as the n = 2 image potential state
(IP2) lie resonant with the surface project upper sp-band (band diagram in Fig. 3.4b),
and thus are not strictly localized at the surface. mPP is capable to directly map the
energy–momentum dispersion of these surface states: Using 1.53 eV photons, the
lowest order of photoemission to overcome the work function of Cu(111) is m =
4 (Φ ≈ 4.9 eV). For the chosen photon energy, the IP1 state is excited resonantly
in a 3-photon process from the occupied SS state, leading to a strong enhancement
of photoemission yield at the ⎡-point. Consequently, the SS and the IP1 states are
detected at a final state energy of 5.6 eV above the Fermi level. Because of the
different effective masses of the SS and the IP1 state, however, the excitation becomes
detuned for larger in-plane momenta k || and the parabolic dispersion of both states
under near-resonant excitation conditions is observed separately. Notably, in the in-
plane momentum region where the resonance condition is detuned, the photoemission
yield is significantly reduced. If the driving photon energy is increased to 1.57 eV,
the excitation becomes detuned over the full accessible in-plane momentum range
and, concomitant, the photoemission intensity becomes more smoothly distributed
in momentum space. A further increase of the photon energy to 1.67 eV leads to a
pronounced enhancement of photoemission yield at large in-plane momenta, which
can be attributed to onset of a resonant 3-photon transition from the SS state into
the IP2 state. In short, mPP is fully capable to map the electronic band structure
in the full accessible momentum range, and its efficiency is strongly dependent on
the nonlinear resonance conditions between the occupied and unoccupied electronic
bands, which are the SS and IP1 or IP2 states in the example of Cu(111).

3.3.2 Above Threshold Photoemission

At sufficiently high driving field strength, higher-order replicas of the lowest-order

photoemission signature become accessible. While the photoemission intensity in
Fig. 3.4a is dominated by signal in nonlinear order m = 4, higher-order replicas in
the order of m + n = 5 and m + n = 6 are clearly detected. At the excitation conditions
used for the data shown in Fig. 3.4, each higher order of photoemission is roughly a
64 M. Reutzel et al.

(a) ω = 1.53 eV (b) ω = 1.57 eV (c) ω = 1.67 eV (d)

10.0 8 Usp
(6)
SS
SS(6) SS(6)
9.0 6
Final State Energy (eV)

x240 x217 x540

x15 x22 x30 IP2 E
SS(5) vac

E-EF (eV)
8.0 SS(5) SS(5)
4 IP1
(5)
IP1 (5)
IP1 (5) IP1
7.0
x1 (4) x1 x1 2
1 SS SS(4)
SS(4)
6.0 1 1
IP1(4) IP1(4) SS
IP1(4) 0 0 EF
5.0 Cu(111) 0 0 Lsp
0.0 0.1 0.2 0.0 0.1 0.2 0.0 0.1 0.2 0.0 0.1 0.2
k|| ( —1) k|| ( —1) k|| ( —1) k|| ( —1)

Fig. 3.4 Exemplary mPP and ATP data obtained from Cu(111). a–c From left to right, the energy–
momentum-resolved mPP spectra are obtained with increasing photon energy. The IP1(i) and the
SS(i) state are observed in threshold order of photoemission i = m = 4. In addition, ATP leads to
the formation of higher-order replicas in i = m + n = 4 + 1 and 4 + 2. b Excitation diagram for
1.67 eV photons. Reproduced from [28]

factor 10 less intense. This is a first indication that these experiments are performed
in the perturbative regime where still the lowest order of photoemission dominates
the signal [28]. The excitation of multiple orders of nonlinear interaction, and conse-
quently photoemission, can be understood in terms of generation of a Floquet quasi-
energy ladder, where levels excited to energies above the work function can decay into
the vacuum-free electron continuum [27, 28], as will be explained further below. In
the non-perturbative regime, one might expect to observe that higher-order processes
(m + n) become more intense than the signal obtained in threshold photoemission
with nonlinear order m [38]. Nevertheless, energy–momentum-resolved spectra of
well-ordered, single-crystalline surfaces carry rich information on the allowed dipole
transitions under intense optical excitation. Specifically, it is directly apparent that
the parabolic dispersion of the surface bands is unchanged. Moreover, the enhanced
photoemission yield is clearly detected at the in-plane momenta where the SS to
IP1 (IP2) transition occurs in a three-photon resonance, indicating that indeed such
resonant conditions lead to higher-order transitions that are observed as ATP signal.
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 65

3.3.3 Toward Full Surface Brillouin Zone Mapping

by Coherent mPP

So far, we have applied mPP spectroscopy to characterize the energy–momentum-

dispersive electronic band structure of the (111) crystalline orientation of noble metal
surfaces such as silver or copper. On these facets, the surface projected band gap
and the surface bands are found around the center of the surface Brillouin zone,
i.e., at the ⎡-point. However, there are many modern quantum materials such as
graphene, whose important electronic band structure features are found at large in-
plane momenta, i.e., at the corners of the surface Brillouin zone. If one aims to
probe these bands in an ARPES experiment, one has to realize that probing larger in-
plane momenta requires excitation of electrons to higher energies such that they can
perform work against the surface potential [48]. This is because the energy in elec-
tron surface parallel motion is conserved and is not available to overcome the work
function in photoemission [32]. Consequently, the accessible in-plane momentum
range in a photoemission experiment is constrained by the so-called photoemission
horizon that is described by E kin = hk|| /2m e (green parabola in Fig. 3.5a; E kin :
photoelectron kinetic energy; h: Plank constant, me : electron mass). In a nutshell, a
photon energy of approximately 20 eV is necessary to probe the first surface Bril-
louin zone with a typical extension of < 2 Å−1 . Of course, the necessary excitation is
easily accessible in static ARPES experiments at synchrotrons or with conventional
gas discharge lamps in the laboratory. It is more challenging for femtosecond lasers
excitation. Here, the last years have shown a tremendous development in the installa-
tion of table-top high-repetition rate high-harmonic generation beamlines [97–103]
that enable generation of sufficiently high photon energies with a few ten femtosecond
pulse duration. Complementary, highly nonlinear mPP provides a powerful approach
to study the coherent light–matter interaction without the need to first create extreme
ultraviolet light. Instead of creating high energy photons in a nonlinear process, it
is possible to utilize this nonlinear process in order to photoexcite electrons to a
sufficiently high energy in order to overcome the photoemission horizon (Fig. 3.5a).
We illustrate the potential of this approach with the example of the Ag(110) surface
[104]. The (110) surface occurs at a junction of two (111) planes. Consequently, the
band gaps of (110) surface also occur in the ⎡ − L direction, but this is an off-normal
direction for (110). As in the case of the (111) facets, a surface projected band gap
is formed on this crystal orientation. The band gaps of the (110) surface support
Shockley surface bands (S oc labels the occupied SS, and S un labels the unoccupied
SS) at the Y-point, corresponding to in-plane momenta of ≈ 0.7 Å−1 (Fig. 3.5a).
When performing a mPP experiment with 1.73 eV photons, a m = 3 nonlinear
process is sufficiently energetic to overcome the work function of 4.2 eV at the
⎡-point. For example, the Fermi level E F can be probed in 3PP in vicinity of the
⎡-point (Fig. 3.5b). As can be seen in the excitation diagram in Fig. 3.5a, however,
at the Y-point, the energy of the photoelectrons probing the occupied surface band
is not sufficient to overcome the photoemission horizon. Nevertheless, in mPP, the
noted surface bands can be detected in next nonlinear order: For m + n = 4, the
66 M. Reutzel et al.

(a) Γ Y (b) Γ Y
ω= 1.73 eV

Binding Energy (eV)

4PP

Ekin (eV)

Final State Energy (eV)

(4)
6 2 Soc
7.0 EF(4) 0
3PP
Evac
E—EF (eV)

4 0
6.0 1
2 Sun E (3)
Max
F

Soc 5.0
0 EF
Min

0.0 0.5 1.0 0.0 0.3 0.6

ky ( —1) ky ( —1)

Fig. 3.5 Illustration of the opportunity to use high-order mPP to probe electronic bands at large
in-plane momenta. a While 3PP with 1.73 eV photons is sufficient to overcome the vacuum level at
the ⎡-point, the occupied surface state S oc at the Y-point is not accessible because it remains hidden
below the photoemission horizon (green parabola). The S oc state, however, can be probed in the
next higher order of photoemission, i.e., in 4PP. b Corresponding mPP data in which the occupied
surface state is probed in 4PP, but not in 3PP. In contrast, the Fermi level at the ⎡-point is accessible
in 3PP and in 4PP. Reproduced from [104]

energy is sufficient to overcome the photoemission horizon, and, indeed, we detect

the S un ← S oc resonant transition in (1 + 3)PP (Fig. 3.5b). In direct comparison to
data obtained in 1PP [105] and 2PP (Fig. 3.2b) [52], we can thereby confirm that
the energy–momentum dispersion (i.e., the effective mass) of the surface band is
well-reproduced in (1 + 3)PP, exemplifying thus the potential of highly nonlinear
mPP to probe the occupied and unoccupied electronic band structure at large in-plane
momenta. Moreover, the full potential of ITR-mPP, as will be discussed in the next
section, can then be employed to map these bands.

3.4 Coherent Two-Dimensional Photoelectron Spectroscopy

The strength of the time-resolved photoemission technique is the direct access to the
energy–momentum-resolved population dynamics in excited states. In conventional
two-color pump-probe experiments; typically, the phase information and thus the
important insight into coherent light–matter interaction are not accessible. In order
to gain access to the momentum-resolved coherent light–matter interaction, the time-
resolved photoemission experiment has to be extended to a phase-resolved pump-
probe scheme. This facilitates the evaluation of the interferometrically time-resolved
photoemission data akin to a two-dimensional spectroscopy [106, 107], providing
direct access to the momentum-resolved coherent light–matter interaction.
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 67

In this context, so far, we have discussed mPP as a static spectroscopy tool to

map the energy–momentum-dispersive surface projected band structure, as applied
to noble metal surfaces. Its full potential, however, becomes accessible if the exper-
iment is extended into the time-domain in a phase-resolved driving scheme. In the
following, first, we will introduce the concept of coherent two-dimensional Fourier
transform photoelectron spectroscopy (2D FT photoelectron spectroscopy) on the
model example of a three-level system that we simulate within the optical Bloch
equation (OBE) approach. Subsequently, we will compare these simple model calcu-
lations with actual ITR-mPP data collected on the Ag(111) surface; this comparison
of OBE simulations and experimental data on Ag(111) will provide a direct bench-
mark of the potential of 2D FT photoelectron spectroscopy. Using this experimental
approach, subsequently, we will show that the mPP and ATP processes from a discrete
Shockley surface state are indeed a coherent process.

3.4.1 Coherent 2D FT Photoelectron Spectroscopy—Optical

Bloch Equation Modeling

The general motivation of using OBE simulations to model time-resolved mPP and
ARPES experiments is to simplify the complex dynamics in a condensed matter
system to a more atom-like system described only by a few discrete levels. This
is schematically illustrated in the three-level system shown in Fig. 3.6b. All loss
and decoherence process, which in the solid state are given by electron–electron and
electron–phonon scattering processes, are added phenomenologically to the diagonal
and off-diagonal elements of the density matrix, respectively, as 1/T 1i and 1/T 2ij
rates (i, j: levels) [27, 108, 109]. Details on the application of the optical Bloch
equations to the evaluation of time-resolved photoemission data can be found in
many references, e.g., [27, 70, 108–111]; here, we want to focus on the evaluation
of the OBE simulation akin to a two-dimensional spectroscopy.
For the actual modeling of the mPP process with m = 2, we construct a three-level
system assembling an occupied SS state, an unoccupied intermediate state (IP), and
a final state continuum (Fig. 3.6b). Note, that the continuum can only be modeled
by OBE by adding a dense discrete energy level structure [109]. Dipole transitions
between all states are induced by the coherent driving with a 20 fs laser pulse,
which is consistent with the experimental driving field in related experiments.The
photon energy is chosen such that the intermediate IP state can be occupied only in
a largely detuned transition. The photoemission spectrum is modeled by evaluating
the occupation of the discretized final state continuum, such as shown in the energy-
distribution-curve in the top of Fig. 3.6a. Under the detuned excitation condition
between SS and IP, only negligible occupation is found at the probing energy of the
IP state, but SS state occupation is efficiently excited into the final state continuum
in a two-photon process. In order to model the coherent light–matter interaction,
we repeat the OBE calculation with a phase-locked pulse pair and tune the delay
68 M. Reutzel et al.

(a) Final State Energy (eV) (c)

FT A. (a.u.)
Intensity
SS SS 2ω l
6.4 6.6 6.8
IP 1ω l
IP 0ω l
24
Delay (fs)

50 ωl 2ω l
22 25
ωl 6.0

Polarization Energy (eV)

23
x2.2
20
Delay (fs)

4.0
(b)
EFinal 0 1ω l
0.3
Evac 0.7 2.0
IP
–50 x1 0
0 SS IP 0.0 0ω l
6.5 7.0 Intensity (a.u.) 6.5 7.0 FT A. (a.u.)
SS Final State Energy (eV) Final State Energy (eV)

Fig. 3.6 Modeling of the ITR-mPP experimental data with optical Bloch equations, and its appli-
cation to 2D FT photoelectron spectroscopy. a ITR-mPP (m = 2) experiments can be modeled by a
three-level system b built up by an initially occupied SS state, an initially unoccupied intermediate
state IP, and a final state continuum E Final . In the model, the final state occupation is calculated as
a function of pump-probe delay for the SS and the IP state in a detuned transition. The top and the
right-hand side line-profiles show energy-distribution curves at 0 fs and interferometric correlation
traces in the energy regions of interest (color-coded), respectively. The inset highlights the tilting
of the interference fringes in the time-domain data representation. c After Fourier transformation
of the delay time axis, the polarization energy can be correlated with the final state energy. In the
second-order correlation experiment, Fourier amplitude is found at the harmonics of the driving
laser field oscillating at 0ω, 1ω, and 2ω. Reproduced from [27]

between both in pulses in sub-100 as steps; in this way, energy-resolved interfero-

metric two-pulse correlation traces are collected (main panel of Fig. 3.6a). In this
data representation, the occupation of the final state continuum is color-coded and
we can clearly identify how the occupation is modulated as a function of pump-probe
delay on the timescale of the optical cycle of the laser pulses due to constructive and
destructive interferences. Prominently, at a pump-probe delay of > 20 fs, we find a
tilting of the interference fringes with increasing final state energy (inset in Fig. 3.6a).
This time-domain tilting of the interference fringes can be more intuitively under-
stood when evaluating the ITR-mPP simulation akin to a two-dimensional spec-
troscopy. In related work, the fringe tilting was said to have a hyperbolic dependence
on the relative phase [112]. For this, we Fourier transform the delay axis in order
to relate the polarization energy (excitation energy) with respect to the final state
energy of the detected occupation (Fig. 3.6c); the Fourier amplitude is shown on the
color-coded heatmap. As we simulate a correlation experiment of second order, we
find Fourier amplitude at polarization frequencies that correspond to the harmonics
of the driving frequency ω, i.e., at 0ω, 1ω, and 2ω. This observation directly high-
lights that indeed, the mPP experiments measure the rectification of the nonlinear
coherent polarization field to a photoelectron current.
The slowly varying (decaying) contribution to the occupation dynamics at zero
polarization energy (0ω) can be attributed to the phase-averaged intermediate state
population dynamics; this is the common transient that is typically accessed in a
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 69

two-color mPP (m = 2) experiment [66–68]. From this signal Fourier component,

the inelastic decay times 1/T 1i of level i can be extracted [60, 113]. The 1ω and 2ω
components represent the linear and nonlinear coherent polarizations, respectively,
which the driving field induces to excite the nonlinear transitions in the sample.
Notably, from evaluation of the 1ω and 2ω coherences, one can obtain physical
insights onto the linear coherences that are coupled by linear superposition (SS &
IP and IP & final state continuum) and the nonlinear coherences (SS and final state
continuum), respectively, and thus extract their dephasing 1/T 2ij rates.
Hence, in order to explore the coherence of the two-photon transition from the
discrete SS state into the final state continuum in more detail, Fig. 3.7a shows
expanded elements of the 2D FT spectra corresponding to the 0ω, 1ω, and 2ω
components. Strikingly, we find that the Fourier amplitude profiles show very distinct
dependence on the polarization and the final state energy: While the 0ω is not disper-
sive, the 1ω and 2ω components show a distinct linear dispersion, where the slope
increases from the 0ω to the 2ω-component. This distinct shape of the 2D FT spectra
is a direct result of the coherent excitation process and the conservation of energy
in the photoexcitation of a discrete initial state with a spectrally broad laser pulse
(Fig. 3.7b). Consequently, the slope of the 2D FT spectra can be described by the
ratio k/m, where m is the order of the photoexcitation process and k is the nonlinear
polarization order. Specifically, for the case of the three-level model simulation of
occupation transfer from the SS state into the final state continuum in a coherent two-
photon process, we find slopes of 0/2, 1/2, and 2/2 for the 0ω, 1ω, and 2ω-components,
respectively.
Naturally, these fingerprints of coherent light–matter interaction identified in the
2D FT photoelectron spectra derive from the signal evolution in the time domain. In
the expanded part of Fig. 3.6a, one clearly finds the already mentioned hyperbolic
tilting of the interference fringes toward time zero with increasing pump-probe delay.
This tilting results from the coherent excitation process as the optical cycle frequency
follows the spectral component of the broadband laser pulse [114–116].

3.4.2 Coherent 2D FT Photoelectron Spectroscopy

of Ag(111)

The Shockley surface band of Ag(111) provides a simple model system to system-
atically study the impact of the nonlinear order of photoemission (m = 4) onto the
coherence of the photoemission process. In this section, in direct comparison to the
OBE simulations, we will show how it provides a benchmark response for applica-
tion of 2D-FT to more complex systems. For this purpose, it is important to note that
this band probably has the longest dephasing time among occupied states of metal
surfaces [66].
70 M. Reutzel et al.

Fig. 3.7 a Blow ups of the (a) (b)

2D FT photoelectron spectra 2ω l
of Fig. 3.6c. The polarization
and final state energy 6.7
dispersion of the Fourier
amplitude can be described
by the ratio k/m, where k and 6.5

Polarization Energy (eV)

Pol. En.
m are the nonlinear
polarization order and the 2ω l
overall photoemission order, 1ω l
respectively. b Schematic
illustration of the 3.4
conservation of energy in a
coherent photoemission
process: The shape of the 2D 3.2
spectra can be understood
based on the correlation of 1ω l
the excitation frequency
component of the laser pulse 0ωl
with a specific final state 0.1
energy. Reproduced from 1
[27]
—0.1 Fin. State En.
0
6.4 6.6 6.8
Final State Energy (eV)

For this task, we perform a ITR-4PP experiment on the Ag(111) surface and
report images of the photoemission intensity as a function of final state energy and
in-plane momentum as the pump-probe delay is varied with interferometric precision
(Fig. 3.8a). While we observe photoemission intensity from the IP1 state and the
SS state, for the sake of simplicity, we concentrate on the SS feature. Notably, we
find that the full energy–momentum-dispersive SS state spectrum apparently shifts
down in energy on the time scale of a single optical cycle. Indeed, this is direct
verification of the coherence of the 4PP process of the SS state. The energy shift
reflects which frequency component of the excitation pulse comes into in-phase
relation at a particular delay.
While the energy–momentum-resolved data record a striking illustration of coher-
ence, it is more straightforwardly analyzed by plotting the pump-probe delay evolu-
tion of the final state energy for a selected in-plane momentum (k || =0 Å−1 , Fig. 3.8b),
similar to the OBE simulations. First, we observe the high-peak-to-baseline ratio in
the interferometric two-pulse correlation traces that is a consequence of the high
nonlinearity of the photoemission process (right panel in Fig. 3.8b). In addition, we
find a strong periodic enhancement and suppression of photoemission on the delay
axis oscillating with the optical cycle of the 1.40 eV photons used in the experi-
ment. Again, this directly reflects the constructive and destructive interference in the
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 71

(a)
Final State Energy (eV)
SS
5.6

5.2
IP
4.8 28.9 fs 29.5 fs 30.1 fs
-0.1 0.0 0.1 -0.1 0.0 0.1 -0.1 0.0 0.1
k|| (eV) k|| (eV) k|| (eV)

(b) 5.5 (c) 5.7 4ω

Intensity

l 4ωl
SS
IP
40 5.5 4/4

3ωl 3ωl

Polarization Energy (eV)

4.3
50
4.1 3/4 3/4
30
2ωl 2ωl
2.9
Delay (fs)

2.7 2/4 2/4

0 1ωl 1ωl
1.5
1.3 1/4
0.6 1/4

0ωl 0ωl
0.1 1

–50 IP SS -0.1
0 0 0/4 0/4

5.0 5.5 0 20 0 50 5.4 5.6 5.8 5.4 5.6 5.8

Final State Energy (eV) Norm. Intensity Final State Energy (eV)

Fig. 3.8 ITR-4PP experiment on Ag(111). a Selected energy–momentum-resolved snapshots of

the SS and the IP state taken within a single optical cycle of the driving laser pulse (time stamps given
in the figure). Due to the coherence of the 4-photon process, the SS state photoelectrons are detected
at a decreasing final state energy for increasing pump-probe delay. b Time-domain representation
of the ITR-4PP experiment evaluated selectively for k || =0 Å−1 . On the time axis, the photoemis-
sion intensity is strongly modulated on the timescale of the optical cycle due to constructive and
destructive interference. Moreover, at a pump-probe delay of ~ 30 fs, the interference fringes are
tilted. c Fourier analysis of the data shown in (b). Fourier amplitude is observed at the harmonics
of the driving frequency and the dispersion of the 2D FT spectra is characterized by the rate k/
m, as given in the figure. On the right-hand side, an OBE model simulation with a 5-level system
confirms the k/m ratio in the 4PP process. Reproduced and adopted from [27]

nonlinear photoemission process at approximately the optical cycle (≈ 2.7 fs) of the
optical field as the pump-probe delay is advanced in subcycle steps. The interference
fringes record a direct signature of coherence in the 4-photon promotion of SS elec-
trons to above the vacuum level: At a pump-probe delay of around 30 fs, we find the
interference fringes to tilt toward time zero with increasing final state energy, fully
consistent with the OBE model simulations of the ITR-mPP experiment. Actually,
this tilting of the interference fringes then directly reflects the apparent upshift of
72 M. Reutzel et al.

the energy–momentum-dispersive SS state in Fig. 3.8a, i.e., as the delay is advanced

within a single optical cycle.
Having compared the time-domain appearance of the ITR-mPP data between
experiment and OBE simulation, in the next step, we evaluate the coherent response
akin to 2D spectroscopy [106, 107]. We Fourier transform the time-domain inter-
ferometric data to obtain 2D FT photoelectron spectra in the conjugate frequency
domain, as shown in Fig. 3.8c. Again, we find Fourier amplitude at the driving laser
frequency and its harmonics, indicating that coherent polarization fields up to fourth
order are detected in the 4-photon photoexcitation process of the SS state. Signif-
icantly, the 2D FT spectra have distinct shapes: Their slopes take on the ratio k/m
where k is the polarization order and m is the photoemission order (here: m = 4).
In direct comparison to the OBE simulation, this data analysis directly shows that
the 4PP process from the SS state is a coherent process. For further verification of
this statement, we repeat the OBE modeling for a 5-level system in order to describe
the 4-photon process in a ladder of transitions via excitation that is detuned from
the IP state and proceeds through virtual intermediate states up to the photoemission
continuum (Fig. 3.8d). As detailed in Ref. [27], we can reproduce the overall shapes
of the experimental 2D FT spectra. Note that the energy width of the 2D FT spectra
contains information on the decoherence rate of the coherent polarization [117].
Moreover, by systematically varying the photoemission order from m = 4 to m = 3
and 2, the shape of the 2D FT spectra behave as expected [27, 117]. Related inter-
ferometric time-resolved nonlinear photoemission experiments have been applied to
studies of dephasing and coherent control of electronic bands in surface and bulk
metals [11, 27, 75–77, 83], metal surfaces with organic/inorganic adsorbates [85–
87], metal/semiconductor interfaces [113, 118], ultrafast microscopy of trivial [42,
88, 119] and topological plasmonic fields [89, 90, 93, 120–122].

3.4.3 Coherent Above Threshold Photoemission

Above threshold photoemission describes the detection of photoelectrons with ener-

gies that can only be reached if the electron absorbs more photons than necessary to
overcome the material’s work function [123, 124]. In general, this process of ATP
can be envisaged in two ways: On the one-side, it might be possible that a free photo-
electron is created in front of the metal surface, which is subsequently dressed by the
laser field. Otherwise, the driving laser field can directly induce coherent polarization
field in a Floquet ladder oscillating at the frequency where the photoelectrons are
rectified, and, as such coherently lead to photoemission at above threshold energy.
If both processes occur with similar amplitudes, one might expect the two processes
to show evidence of two-path interference. In this section, we employ the ITR-mPP
experiment in the manner of the 2D FT photoelectron spectroscopy in order to show
that we indeed find a direct contribution from higher-order coherent polarization field
at the order of ATP.
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 73

For this, we repeat the ITR-mPP experiment on Ag(111), but now evaluate the
photoemission signal not in lowest order of photoemission (m = 4), but focus on
the m + n = (4 + 1)PP data (Fig. 3.9). As discussed in the 4PP experiment and the
OBE simulations, for ATP from the occupied Shockley surface band, we clearly find
tilted interference fringes in the time domain and the distinct shape of the 2D FT
photoelectron spectra in Fourier space. Specifically for the ATP process that occurs
in the order of m = 5, we find that the slopes of the 2D FT spectra are nicely described
by the ratio k/5.
This experimental observation indeed shows the mPP experiment detects highly
nonlinear polarization fields created in the sample as the system is driven by intense
near-infrared laser pulses. Consequently, under our experimental conditions, we
conclude that ATP is a one-step process where the highly nonlinear coherent polar-
ization field is directly rectified such that photoelectrons are detected in an above
threshold process.

(a) ITR—5PP, ATP (b) 5PP (ATP), 2D—FT

7.0
Intensity

SS 3ωl
IP 40 4.3

4.1
3/5
Polarization Energy (eV)

50 ATP
30
2ωl
2.9
x2.5
Delay (fs)

2.7
0 2/5
1ωl
1 1.5
1

1.3
—50
0 x1 IP SS
0 1/5
7.0 7.5 0 50 0 200 7.0 7.2 7.4
Final State Energy (eV) Norm. Intensity Final State Energy (eV)

Fig. 3.9 Interferometrically time-resolved measurements can be applied in order to characterize

the coherence of the ATP process. a, b The same data as initially shown in Fig. 3.8 is analyzed, but
now explicitly not in threshold order of photoemission m but in ATP, i.e., in m + n = 4 + 1 = 5.
a In the time domain, the interference fringes show clear tilting toward time zero with increasing
pump-probe delay (inset). b In the 2D FT spectra, the shape of the FT amplitude can be described
by the ratio k/5. Reproduced from [27]
74 M. Reutzel et al.

3.5 Ultrafast Quasiparticle Dressing by Light

In the final section of this chapter, we will explore the parameter regime where the
driving electric field becomes sufficiently strong to modify the equilibrium quasi-
particle band structure of the solid-state system. In the context of Floquet theory [2,
125, 126], it is proposed that a time-periodic oscillating field can lead to the coherent
formation of light-dressed band structure whose properties are composed of the
solid itself and those of the light field [126]. In this regime, if a resonance condition
between an occupied and an initially unoccupied state is driven at a sufficiently high
field, band structures with light-induced band gaps can emerge, as initially found
in semiconductors [127, 128], and more recently on metal surfaces [11], topolog-
ical insulators [9], graphene [10], and black phosphorus [12]. Likewise, in atomic
physics, it is well established that a light field can lead to the renormalization of
the atomic eigenstates, requiring the formulation of new light-dressed eigenstates
[129], which have been identified in all-optical spectroscopies of atomic gas species
as the Autler–Townes splitting [130]. So far, we have explored that the coherent
polarization fields imprinted onto the solid-state material by a time-periodic external
perturbation can be directly measured with energy- and momentum resolution with
ITR-mPP. In the following, using ITR-mPP and 2D FT photoelectron spectroscopy,
we will show that it is indeed possible to extract information on the coherent response
leading to light-dressed band structure of Cu(111) surface bands [11].
Figure 3.10a shows an overview of ITR-4PP data collected with 1.54 eV pulses
on pristine Cu(111). In contrast to the last section where we focused on non-resonant
excitation pathway of the SS state, the photon energy is now chosen such that
it drives a three-photon resonant excitation from the SS to the IP1 state at the
⎡-point; with increasing in-plane momentum k || , the resonance condition detunes
(Fig. 3.3a). When considering the IP1 ← SS resonance as a coherently driven two-
level system, Rabi oscillations will be induced, if the driving field is sufficiently
strong and the decoherence time is sufficiently long [129]. The strong driving field
causes the mixing of the coupled states such that this coherently driven system
becomes strongly perturbed, and needs to be described by a new set of light-dressed
eigenstates E ± , corresponding to Autler–Townes doublet [130]. The energetic split-
ting of the transiently dressed states is described by the generalized Rabi splitting
∼ √
hΩ R (τ, k) = h μeff E3 (τ ) + Δ2 (k), (μeff : effective dipole moment of the three-
photon transition; E(τ ): electric field strength; Δ(k): in-plane momentum dependent
detuning). The challenge in the ITR-4PP experiment now is the experimental access
to the energy–momentum dependence of the coherent polarization fields and how
they carry information on the light-induced band renormalizations.
In order to access the energy–momentum-dispersive coherent polarization fields
in different harmonic orders of the driving laser’s harmonics, we Fourier analyze the
ITR-4PP data as detailed in Ref. [11]: In short, in the 2D FT spectra, we band-pass
filter the frequency components oscillating with frequencies close to 2ω and inverse
Fourier transform the data back into the time domain. In the data in Fig. 3.10b,
the amplitude of the coherent polarization field is then plotted on the color scale as
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 75

(a) ITR

Final State Energy (eV)

6.0
1

—1
k|| = 0
5.0 0

-0.1 0 30 45 60
0.1 -30 -15 0 15
k|| ( —1
) Delay (fs)

(b)
Final State Energy (eV)

2ωl
6.0
1

—1
k|| = 0
5.0 0

-0.1 0 30 45 60
0.1 -30 -15 0 15
k|| ( —1
(c) ) Delay (fs)
Final State Energy (eV)

0 fs 7 fs 15 fs 65 fs SS
E+
6.0

IP

E— 2ωl 2ωl 2ωl 2ωl

5.0
0 0.1 0 0.1 0 0.1 0.0 0.1
—1 —1 —1 —1
k|| ( ) k|| ( ) k|| ( ) k|| ( )

Fig. 3.10 If the IP1 ← SS transition is driven resonantly at a sufficiently high electric field
strength, a coherent manipulation of the electronic band structure is observed. a Energy- and
momentum-resolved ITR-4PP data collected from pristine Cu(111). b After band-pass filtering
the 2ω-component and inverse Fourier transformation, the energy–momentum dispersion of the
coherent polarization field oscillating at 2ω can be identified. c Selected energy–momentum images
from the multidimensional data shown in (b). At large delays (65 fs), the energy–momentum disper-
sion of the coherent polarization field can be described by the unperturbed dispersions of the SS
and the IP1 state (solid black lines). In temporal overlap of the pulse pair, the electric field leads
to a coherent renormalization of the electronic band structure, which is directly imprinted onto the
2ω-component of the coherent polarization field. Reproduced from [11]

function of pump-probe time delay, energy, and in-plane momentum. Notably, we

find that the amplitude of the coherent polarization field oscillating at 2ω mimics the
energy–momentum dispersion of the SS and the IP1 state at pump-probe delays of
65 fs, i.e., when the local electric field strength inducing the photoexcitation process
is dominantly given by a single-laser pulse (Fig. 3.10c, right panel). When the pump-
probe laser pulses come into temporal overlap, however, the electric field strength is
significantly increased, leading to a stronger AC Stark splitting of the SS band, i.e.,
a larger generalized Rabi frequency and thus a measurable light-induced band gap
opening between the light-dressed eigenstates E ± . During the time period of overlap
of the pump- and the probe-laser pulses, we then indeed find a systematic opening and
closing of the light-induced band gap (Fig. 3.10c, left panels). Notably, as expected,
the detuning of the three-photon resonance toward larger in-plane momenta leads
76 M. Reutzel et al.

to a weaker influence of the driving electric field strength on the observed band
dispersion (Fig. 3.10c).
We consider this observation of the gap opening in the energy–momentum
dispersion that appears in the 2ω coherent polarization field as a direct evidence
for the coherent engineering of the electronic band structure of the noble metal
surface. Notably, the light-dressed electronic band structure is obscured, i.e., not
yet detectable, in the phase-averaged data shown in Fig. 3.3a, highlighting that the
analysis of the phase-stable mPP experiment has significant advantages to probe
light-induced phenomena as it approaches the non-perturbative regime. As a photoe-
mission experiment, it will be applicable to further material systems, providing, in
addition to two-color trARPES [9, 12], a powerful energy- and momentum-sensitive
technique to probe the coherent control of condensed matter material phases.

3.6 Conclusion

The coherent response of metals to optical fields is familiar in our daily lives as
coherent reflections of optical fields from metallic mirrors. Measuring such coherent
responses to optical fields by optical methods is extremely challenging [131], because
it is defined by the coherent response of free electrons occupying electronic bands
with several eV bandwidth. Consequently, the coherent response leading to the
screening, and therefore, optical reflection at a metal surface occurs on the attosecond
time scale, as defined by the collective plasmon response of metals [132]. The energy
and momentum resolution of photoemission spectroscopy enables measurements
of quasiparticle decoherence from photoemission linewidths, though this strongly
depends on the sample quality [133], and can easily lead to incorrect conclusions
with poorly prepared samples. The nonlinear ITR-mPP spectroscopy enables the
direct measurement of coherence of surface and bulk electronic bands of metals
particularly in the electron [27, 75] and hole [77] excited states, as illustrated in this
chapter. Beyond probing the linear and nonlinear polarization dephasing, the applied
Coulomb field of femtosecond laser excitation can be sufficiently strong to drive the
Rabi frequency near the carrier wave frequency and thereby modify the electronic
band structure of a solid [11], and even exert fields that exceed the binding energy
of electrons in metals. Moreover, the optical fields can impose topology that tran-
siently modifies the time inversion and parity symmetry of coherent responses of
metals [89, 121]. Therefore, the coherent ITR-mPP spectroscopy is a route to probe
the non-perturbative responses of metals and to transiently impress electronic fields
that modify the electronic properties of solids by Floquet and Poincaré engineering
methods [27, 28, 134, 135].

Acknowledgements We acknowledge fruitful discussions with Namitha Ann James. M. R.

acknowledges support through the Alexander von Humboldt Foundation for his Feodor Lynen
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 77

PostDoc fellowship at the University of Pittsburgh in the group of H.P. In addition, M.R. acknowl-
edges funding through the Deutsche Forschungsgemeinschaft (DFG, German Research Founda-
tion)—217133147/SFB 1073, project B10, and inspiring discussions with the newly established
ITR-mPP team in Göttingen: Hannah Strauch, Marco Merboldt, and Stefan Mathias. H.P. thanks
partial financial support from the National Science Foundation Grant No. CHE-2102601, and
promotion of ITR-mPP methods by Shijing Tan.

References

1. D.N. Basov, R.D. Averitt, D. Hsieh, Towards properties on demand in quantum materials.
Nat. Mater. 16, 1077 (2017)
2. A. de la Torre, D.M. Kennes, M. Claassen, S. Gerber, J.W. McIver, M.A. Sentef, Colloquium:
nonthermal pathways to ultrafast control in quantum materials. Rev. Mod. Phys. 93, 041002
(2021)
3. D. Fausti et al., Light-induced superconductivity in a stripe-ordered cuprate. Science 331, 189
(2011)
4. T. Rohwer et al., Collapse of long-range charge order tracked by time-resolved photoemission
at high momenta. Nature 471, 490 (2011)
5. C.W. Nicholson, A. Lücke, W.G. Schmidt, M. Puppin, L. Rettig, R. Ernstorfer, M. Wolf,
Beyond the molecular movie: Dynamics of bands and bonds during a photoinduced phase
transition. Science 362, 821 (2018)
6. F. Schmitt et al., Transient electronic structure and melting of a charge density wave in TbTe3 .
Science 321, 1649 (2008)
7. L. Stojchevska, I. Vaskivskyi, T. Mertelj, P. Kusar, D. Svetin, S. Brazovskii, D. Mihailovic,
Ultrafast switching to a stable hidden quantum state in an electronic crystal. Science 344, 177
(2014)
8. E.J. Sie et al., An ultrafast symmetry switch in a Weyl semimetal. Nature 565, 61 (2019)
9. F. Mahmood, C.-K. Chan, Z. Alpichshev, D. Gardner, Y. Lee, P.A. Lee, N. Gedik, Selective
scattering between Floquet-Bloch and Volkov states in a topological insulator. Nat. Phys. 12,
306 (2016)
10. J.W. McIver, B. Schulte, F.U. Stein, T. Matsuyama, G. Jotzu, G. Meier, A. Cavalleri, Light-
induced anomalous Hall effect in graphene. Nat. Phys. 16, 38 (2020)
11. M. Reutzel, A. Li, Z. Wang, H. Petek, Coherent multidimensional photoelectron spectroscopy
of ultrafast quasiparticle dressing by light. Nat. Commun. 11, 2230 (2020)
12. S. Zhou et al., Pseudospin-selective Floquet band engineering in black phosphorus. Nature
614, 75 (2023)
13. T. Frigge et al., Optically excited structural transition in atomic wires on surfaces at the
quantum limit. Nature 544, 207 (2017)
14. J.G. Horstmann, H. Böckmann, B. Wit, F. Kurtz, G. Storeck, C. Ropers, Coherent control of
a surface structural phase transition. Nature 583, 232 (2020)
15. M. Düvel et al., Far-from-equilibrium electron–phonon interactions in optically excited
graphene. Nano Lett. 22, 4897 (2022)
16. A. Kogar et al., Light-induced charge density wave in LaTe3 . Nat. Phys. 16, 159 (2020)
17. A. Zong, B.R. Nebgen, S.-C. Lin, J.A. Spies, M. Zuerch, Emerging ultrafast techniques for
studying quantum materials. Nat. Rev. Mater. 8, 224 (2023)
18. J.A. Sobota, Y. He, Z.-X. Shen, Angle-resolved photoemission studies of quantum materials.
Rev. Mod. Phys. 93, 025006 (2021)
19. D. Filippetto, P. Musumeci, R.K. Li, B.J. Siwick, M.R. Otto, M. Centurion, J.P.F. Nunes,
Ultrafast electron diffraction: visualizing dynamic states of matter. Rev. Mod. Phys. 94, 045004
(2022)
78 M. Reutzel et al.

20. S. Ghimire, D.A. Reis, High-harmonic generation from solids. Nat. Phys. 15, 10 (2019)
21. S. Ghimire, A.D. DiChiara, E. Sistrunk, P. Agostini, L.F. DiMauro, D.A. Reis, Observation
of high-order harmonic generation in a bulk crystal. Nat. Phys. 7, 138 (2011)
22. F. Langer et al., Lightwave-driven quasiparticle collisions on a subcycle timescale. Nature
533, 225 (2016)
23. C.P. Schmid et al., Tunable non-integer high-harmonic generation in a topological insulator.
Nature 593, 385 (2021)
24. C. Heide et al., Probing topological phase transitions using high-harmonic generation. Nat.
Photon. 16, 620 (2022)
25. J. Freudenstein et al., Attosecond clocking of correlations between Bloch electrons. Nature
610, 290 (2022)
26. A. Damascelli, Probing the electronic structure of complex systems by ARPES. Phys. Scr.
T109, 61 (2004)
27. M. Reutzel, A. Li, H. Petek, Coherent two-dimensional multiphoton photoelectron spec-
troscopy of metal surfaces. Phys. Rev. X 9, 011044 (2019)
28. M. Reutzel, A. Li, H. Petek, Above-threshold multiphoton photoemission from noble metal
surfaces. Phys. Rev. B 101, 075409 (2020)
29. W.S. Fann, R. Storz, J. Bokor, Observation of above-threshold multiphoton photoelectric
emission from image-potential surface states. Phys. Rev. B 44, 10980 (1991)
30. M. Aeschlimann, C.A. Schmuttenmaer, H.E. Elsayed-Ali, R.J.D. Miller, J. Cao, Y. Gao, D.A.
Mantell, Observation of surface enhanced multiphoton photoemission from metal surfaces in
the short pulse limit. J. Chem. Phys. 102, 8606 (1995)
31. F. Banfi, C. Giannetti, G. Ferrini, G. Galimberti, S. Pagliara, D. Fausti, F. Parmigiani, Exper-
imental evidence of above-threshold photoemission in solids. Phys. Rev. Lett. 94, 037601
(2005)
32. F. Bisio, M. Nývlt, J. Franta, H. Petek, J. Kirschner, Mechanisms of high-order perturbative
photoemission from Cu(001). Phys. Rev. Lett. 96, 087601 (2006)
33. G. Saathoff, L. Miaja-Avila, M. Aeschlimann, M.M. Murnane, H.C. Kapteyn, Laser-assisted
photoemission from surfaces. Phys. Rev. A 77, 022903 (2008)
34. F. Bisio, A. Winkelmann, C.-T. Chiang, H. Petek, J. Kirschner, Band structure effects in above
threshold photoemission. J. Phys. Condens. Matter 23, 485002 (2011)
35. F. Sirotti, N. Beaulieu, A. Bendounan, M.G. Silly, C. Chauvet, G. Malinowski, G. Fratesi,
V. Véniard, G. Onida, Multiphoton k-resolved photoemission from gold surface states with
800-nm femtosecond laser pulses. Phys. Rev. B 90, 035401 (2014)
36. P. Dreher, D. Janoschka, A. Neuhaus, B. Frank, H. Giessen, M. Horn-von Hoegen, F.-J. Meyer
zu Heringdorf, Quantitative determination of the electric field strength in a plasmon focus
from ponderomotive energy shifts. Nanophotonics 11, 3687 (2022)
37. R. Bormann, M. Gulde, A. Weismann, S.V. Yalunin, C. Ropers, Tip-enhanced strong-field
photoemission. Phys. Rev. Lett. 105, 147601 (2010)
38. M. Schenk, M. Kruger, P. Hommelhoff, Strong-field above-threshold photoemission from
sharp metal tips. Phys. Rev. Lett. 105, 257601 (2010)
39. P. Dombi et al., Strong-field nano-optics. Rev. Mod. Phys. 92, 025003 (2020)
40. M. Merschdorf, W. Pfeiffer, A. Thon, S. Voll, G. Gerber, Stimulated emission induced by
exciton–exciton scattering in ZnO/ZnMgO multiquantum wells up to room temperature. Appl.
Phys. A 71, 547 (2000)
41. P. Dombi, A. Hörl, P. Rácz, I. Márton, A. Trügler, J.R. Krenn, U. Hohenester, Ultrafast
strong-field photoemission from plasmonic nanoparticles. Nano Lett. 13, 674 (2013)
42. M. D˛abrowski, Y. Dai, H. Petek, Ultrafast microscopy: imaging light with photoelectrons on
the nano-femto scale. J. Phys. Chem. Lett. 8, 4446 (2017)
43. M. Sivis, N. Pazos-Perez, R. Yu, R. Alvarez-Puebla, F.J. García de Abajo, C. Ropers,
Continuous-wave multiphoton photoemission from plasmonic nanostars. Commun. Phys. 1,
13 (2018)
44. A. Li, M. Reutzel, Z. Wang, D. Novko, B. Gumhalter, H. Petek, Plasmonic photoemission
from single-crystalline silver. ACS Photon. 8, 247 (2021)
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 79

45. H. Petek, A. Li, X. Li, S. Tan, M. Reutzel, Plasmonic decay into hot electrons in silver. Prog.
Surf. Sci. 98, 100707 (2023)
46. A. Damascelli, Z. Hussain, Z.-X. Shen, Angle-resolved photoemission studies of the cuprate
superconductors. Rev. Mod. Phys. 75, 473 (2003)
47. H. Zhang, T. Pincelli, C. Jozwiak, T. Kondo, R. Ernstorfer, T. Sato, S. Zhou, Angle-resolved
photoemission spectroscopy. Nat. Rev. Methods Primers 2, 54 (2022)
48. S. Hüfner, Photoelectron Spectroscopy: Principles and Applications (Springer, Berlin, 2003)
49. M. Reutzel, A. Li, B. Gumhalter, H. Petek, Nonlinear plasmonic photoelectron response of
Ag(111). Phys. Rev. Lett. 123, 017404 (2019)
50. B. Krömker, M. Escher, D. Funnemann, D. Hartung, H. Engelhard, J. Kirschner, Development
of a momentum microscope for time resolved band structure imaging. Rev. Sci. Instrum. 79,
053702 (2008)
51. K. Medjanik et al., Direct 3D mapping of the fermi surface and fermi velocity. Nat. Mat. 16,
615 (2017)
52. A. Li et al., Multidimensional multiphoton momentum microscopy of the anisotropic Ag(110)
surface. Phys. Rev. B 105, 075105 (2022)
53. M. Keunecke et al., Electromagnetic dressing of the electron energy spectrum of Au(111) at
high momenta. Phys. Rev. B 102, 161403 (2020)
54. J. Maklar et al., A quantitative comparison of time-of-flight momentum microscopes and hemi-
spherical analyzers for time- and angle-resolved photoemission spectroscopy experiments.
Rev. Sci. Ins. 91, 123112 (2020)
55. B. Schönhense et al., Multidimensional photoemission spectroscopy—the space-charge limit.
New J. Phys. 20, 033004 (2018)
56. G. Schönhense et al., Suppression of the vacuum space-charge effect in fs-photoemission by
a retarding electrostatic front lens. Rev. Sci. Instrum. 92, 053703 (2021)
57. E. Rotenberg, A. Bostwick, microARPES and nanoARPES at diffraction-limited light sources:
opportunities and performance gains. J. Synch. Radiati. 21, 1048 (2014)
58. K. Giesen, F. Hage, F.J. Himpsel, H.J. Riess, W. Steinmann, Two-photon photoemission via
image-potential states. Phys. Rev. Lett. 55, 300 (1985)
59. W. Steinmann, Magneto-optical Kerr effect and perpendicular magnetic anisotropy of
evaporated and sputtered Co/Pt multilayer structures. Appl. Phys. A 49, 365 (1989)
60. H. Petek, S. Ogawa, Femtosecond time-resolved two-photon photoemission studies of electron
dynamics in metals. Prog. Surf. Sci. 56, 239 (1997)
61. M. Wolf, Femtosecond dynamics of electronic excitations at metal surfaces. Surf. Sci. 377–
379, 343 (1997)
62. M. Bauer, A. Marienfeld, M. Aeschlimann, Hot electron lifetimes in metals probed by time-
resolved two-photon photoemission. Prog. Surf. Sci. 90, 319 (2015)
63. U. Höfer, I.L. Shumay, C. Reuss, U. Thomann, W. Wallauer, T. Fauster, Time-resolved
coherent photoelectron spectroscopy of quantized electronic states on metal surfaces. Science
277, 1480 (1997)
64. U. Bovensiepen, H. Petek, M. Wolf, Dynamics at Solid State Surfaces and Interfaces, Vol. 1:
Current Developments (Wiley-VCH Verlag Gmbh & Co., Weinheim, 2010), Vol. 1: Current
developments
65. U. Bovensiepen, H. Petek, M. Wolf, Dynamics at Solid State Surfaces and Interfaces, Vol. 2:
Fundamentals (Wiley-VCH Verlag Gmbh & Co., Weinheim, 2012), Vol. 2: Fundamentals
66. P.M. Echenique, R. Berndt, E.V. Chulkov, T. Fauster, A. Goldmann, U. Höfer, Decay of
electronic excitations at metal surfaces. Surf. Sci. Rep. 52, 219 (2004)
67. T. Fauster, M. Weinelt, U. Höfer, Quasi-elastic scattering of electrons in image-potential states.
Prog. Surf. Sci. 82, 224 (2007)
68. M. Weinelt, Time-resolved two-photon photoemission from metal surfaces. J. Phys. Condens.
Matter 14, R1099 (2002)
69. D. Schmitt et al., Formation of moiré interlayer excitons in space and time. Nature 608, 499
(2022)
70. R. Wallauer et al., Tracing orbital images on ultrafast time scales. Science 371, 1056 (2021)
80 M. Reutzel et al.

71. J. Madéo et al., Directly visualizing the momentum-forbidden dark excitons and their
dynamics in atomically thin semiconductors. Science 370, 1199 (2020)
72. J.C. Johannsen et al., Direct view of hot carrier dynamics in graphene. Phys. Rev. Lett. 111,
027403 (2013)
73. I. Gierz et al., Snapshots of non-equilibrium Dirac carrier distributions in graphene. Nat.
Mater. 12, 1119 (2013)
74. J. Lehmann, M. Merschdorf, W. Pfeiffer, A. Thon, S. Voll, G. Gerber, Surface plasmon
dynamics in silver nanoparticles studied by femtosecond time-resolved photoemission. Phys.
Rev. Lett. 85, 2921 (2000)
75. H. Petek, A.P. Heberle, W. Nessler, H. Nagano, S. Kubota, S. Matsunami, N. Moriya, S.
Ogawa, Optical phase control of coherent electron dynamics in metals. Phys. Rev. Lett. 79,
4649 (1997)
76. X. Cui, C. Wang, A. Argondizzo, S. Garrett-Roe, B. Gumhalter, H. Petek, Transient excitons
at metal surfaces. Nat. Phys. 10, 505 (2014)
77. H. Petek, H. Nagano, S. Ogawa, Hole decoherence of d bands in copper. Phys. Rev. Lett. 83,
832 (1999)
78. J. Güdde, M. Rohleder, T. Meier, S.W. Koch, U. Höfer, Time-resolved investigation of
coherently controlled electric currents at a metal surface. Science 318, 1287 (2007)
79. T. Eul, E. Prinz, M. Hartelt, B. Frisch, M. Aeschlimann, B. Stadtmüller, Coherent response of
the electronic system driven by non-interfering laser pulses. Nat. Commun. 13, 3324 (2022)
80. M. Bauer, S. Pawlik, M. Aeschlimann, Decay dynamics of photoexcited alkali chemisorbates:
real-time investigations in the femtosecond regime. Phys. Rev. B 60, 5016 (1999)
81. A. Kubo, N. Pontius, H. Petek, Femtosecond microscopy of surface plasmon polariton wave
packet evolution at the silver/vacuum interface. Nano Lett. 7, 470 (2007)
82. M. Aeschlimann, M. Bauer, D. Bayer, T. Brixner, F.J. Garcia de Abajo, W. Pfeiffer, M. Rohmer,
C. Spindler, F. Steeb, Adaptive subwavelength control of nano-optical fields. Nature 446, 301
(2007)
83. S. Ogawa, H. Nagano, H. Petek, A.P. Heberle, Optical dephasing in Cu(111) measured by
interferometric two-photon time-resolved photoemission. Phys. Rev. Lett. 78, 1339 (1997)
84. M. Merschdorf, C. Kennerknecht, W. Pfeiffer, Collective and single-particle dynamics in
time-resolved two-photon photoemission. Phys. Rev. B 70, 193401 (2004)
85. H. Petek, M.J. Weida, H. Nagano, S. Ogawa, Real-time observation of adsorbate atom motion
above a metal surface. Science 288, 1402 (2000)
86. S. Ogawa, H. Nagano, H. Petek, Phase and energy relaxation in an antibonding surface state:
Cs/Cu(111). Phys. Rev. Lett. 82, 1931 (1999)
87. M. Aeschlimann et al., Observation of optical coherence in a disordered metal-molecule
interface by coherent optical two-dimensional photoelectron spectroscopy. Phys. Rev. B 105,
205415 (2022)
88. M. Dabrowski, Y. Dai, H. Petek, Ultrafast photoemission electron microscopy: imaging
plasmons in space and time. Chem. Rev. 120, 6247 (2020)
89. Y. Dai, Z. Zhou, A. Ghosh, R.S.K. Mong, A. Kubo, C.-B. Huang, H. Petek, Plasmonic
topological quasiparticle on the nanometre and femtosecond scales. Nature 588, 616 (2020)
90. Y. Dai, Z. Zhou, A. Ghosh, S. Yang, C.-B. Huang, H. Petek, Ultrafast nanofemto photoemission
electron microscopy of vectorial plasmonic fields. MRS Bull. 46, 738 (2021)
91. E. Prinz, M. Hartelt, G. Spektor, M. Orenstein, M. Aeschlimann, Orbital angular momentum
in nanoplasmonic vortices. ACS Photon. 10, 340 (2023)
92. P. Dreher, D. Janoschka, B. Frank, H. Giessen, F.-J. Meyer zu Heringdorf, Focused
surface plasmon polaritons coherently couple to electronic states in above-threshold electron
emission. Commun. Phys. 6, 15 (2023)
93. G. Spektor et al., Revealing the subfemtosecond dynamics of orbital angular momentum in
nanoplasmonic vortices. Science 355, 1187 (2017)
94. G. Spektor, D. Kilbane, A.K. Mahro, M. Hartelt, E. Prinz, M. Aeschlimann, M. Orenstein,
Mixing the light spin with plasmon orbit by nonlinear light-matter interaction in gold. Phys.
Rev. X 9, 021031 (2019)
3 Probing Nonlinear Light–Matter Interaction in Momentum Space … 81

95. B. Huber et al., Space- and time-resolved UV-to-NIR surface spectroscopy and 2D nanoscopy
at 1 MHz repetition rate. Rev. Sci. Ins. 90, 113103 (2019)
96. J. Réhault, M. Maiuri, A. Oriana, G. Cerullo, Two-dimensional electronic spectroscopy with
birefringent wedges. Rev. Sci. Ins. 85, 123107 (2014)
97. C.M. Heyl, J. Güdde, A. L’Huillier, U. Höfer, High-order harmonic generation with μJ laser
pulses at high repetition rates. J. Phys. B 45, 074020 (2012)
98. C.-T. Chiang, A. Blättermann, M. Huth, J. Kirschner, W. Widdra, High-order harmonic gener-
ation at 4 MHz as a light source for time-of-flight photoemission spectroscopy. Appl. Phys.
Lett. 101, 071116 (2012)
99. S. Hädrich, A. Klenke, J. Rothhardt, M. Krebs, A. Hoffmann, O. Pronin, V. Pervak, J. Limpert,
A. Tünnermann, High photon flux table-top coherent extreme-ultraviolet source. Nat. Photon.
8, 779 (2014)
100. M. Puppin et al., Time- and angle-resolved photoemission spectroscopy of solids in the
extreme ultraviolet at 500 kHz repetition rate. Rev. Sci. Instrum. 90, 023104 (2019)
101. M. Keunecke et al., Time-resolved momentum microscopy with a 1 MHz high-harmonic
extreme ultraviolet beamline. Rev. Sci. Instrum. 91, 063905 (2020)
102. A. Kunin et al., Momentum-resolved exciton coupling and valley polarization dynamics in
monolayer WS2 . Phys. Rev. Lett. 130, 046202 (2023)
103. X. Li, M.A.R. Reber, C. Corder, Y. Chen, P. Zhao, T.K. Allison, High-power ultrafast Yb:fiber
laser frequency combs using commercially available components and basic fiber tools. Rev.
Sci. Instrum. 87, 093114 (2016)
104. A. Li, N.A. James, T. Wang, Z. Wang, H. Petek, M. Reutzel, Towards full surface Brillouin
zone mapping by coherent multi-photon photoemission. New J. Phys. 22, 073035 (2020)
105. A. Gerlach, G. Meister, R. Matzdorf, A. Goldmann, High-resolution photoemission study of
the Ȳ surface state on Ag(110). Surf. Sci. 443, 221 (1999)
106. S.T. Cundiff, S. Mukamel, Optical multidimensional coherent spectroscopy. Phys. Today 66,
44 (2013)
107. S. Mukamel, Multidimensional femtosecond correlation spectroscopies of electronic and
vibrational excitations. Annu. Rev. Phys. Chem. 51, 691 (2000)
108. T. Klamroth, P. Saalfrank, U. Höfer, Open-system density-matrix approach to image-potential
dynamics of electrons at Cu(100): energy- and time-resolved two-photon photoemission
spectra. Phys. Rev. B 64, 035420 (2001)
109. M.J. Weida, S. Ogawa, H. Nagano, H. Petek, Ultrafast interferometric pump–probe correlation
measurements in systems with broadened bands or continua. J. Opt. Soc. Am. B 17, 1443
(2000)
110. S. Dong et al., Direct measurement of key exciton properties: energy, dynamics, and spatial
distribution of the wave function. Nat. Sci. 1, e10010 (2021)
111. N. Pontius, V. Sametoglu, H. Petek, Simulation of two-photon photoemission from the bulk
sp-bands of Ag(111). Phys. Rev. B. 72, 115105 (2005). http://link.aps.org/abstract/PRB/v72/
e115105
112. J. Mauritsson et al., Attosecond electron spectroscopy using a novel interferometric pump-
probe technique. Phys. Rev. Lett. 105, 053001 (2010)
113. S. Tan, Y. Dai, S. Zhang, L. Liu, J. Zhao, H. Petek, Coherent electron transfer at the Ag/
graphite heterojunction interface. Phys. Rev. Lett. 120, 126801 (2018)
114. R. Trebino, Frequency-Resolved Optical Gating: The Measurement of Ultrashort Laser Pulses
(Springer US, New York, 2000)
115. A. Anderson, K.S. Deryckx, X.G. Xu, G. Steinmeyer, M.B. Raschke, Few-femtosecond
plasmon dephasing of a single metallic nanostructure from optical response function
reconstruction by interferometric frequency resolved optical gating. Nano Lett. 10, 2519
(2010)
116. M. Mascheck, S. Schmidt, M. Silies, T. Yatsui, K. Kitamura, M. Ohtsu, D. Leipold, E. Runge,
C. Lienau, Observing the localization of light in space and time by ultrafast second-harmonic
microscopy. Nat. Photon. 6, 293 (2012)
117. H. Strauch, Georg-August-Universität Göttingen (2022)
82 M. Reutzel et al.

118. S. Tan, A. Argondizzo, J. Ren, L. Liu, J. Zhao, H. Petek, Plasmonic coupling at a metal/
semiconductor interface. Nat. Photon. 11, 806 (2017)
119. D. Podbiel, P. Kahl, F.-J. Meyer zu Heringdorf, Analysis of the contrast in normal-incidence
surface plasmon photoemission microscopy in a pump–probe experiment with adjustable
polarization. Appl. Phys. B 122, 90 (2016)
120. A. Ghosh, S. Yang, Y. Dai, Z. Zhou, T. Wang, C.-B. Huang, H. Petek, A topological lattice
of plasmonic merons. Appl. Phys. Rev. 8, 041413 (2021)
121. Y. Dai, Z. Zhou, A. Ghosh, K. Kapoor, M. D˛abrowski, A. Kubo, C.-B. Huang, H. Petek,
Ultrafast microscopy of a twisted plasmonic spin skyrmion. Appl. Phys. Rev. 9, 011420
(2022)
122. T. J. Davis, D. Janoschka, P. Dreher, B. Frank, F.-J. Meyer zu Heringdorf, H. Giessen, Ultrafast
vector imaging of plasmonic skyrmion dynamics with deep subwavelength resolution. Science
368, eaba6415 (2020)
123. R.R. Freeman, P.H. Bucksbaum, H. Milchberg, S. Darack, D. Schumacher, M.E. Geusic,
Above-threshold ionization with subpicosecond laser pulses. Phys. Rev. Lett. 59, 1092 (1987)
124. W. Becker, F. Grasbon, R. Kopold, D.B. Milošević, G.G. Paulus, H. Walther, in Advances in
Atomic, Molecular, and Optical Physics. ed. by B. Bederson, H. Walther (Academic Press,
New York, 2002), p.35
125. J.H. Shirley, Solution of the Schrödinger equation with a Hamiltonian periodic in time. Phys.
Rev. 138, B979 (1965)
126. T. Oka, S. Kitamura, Floquet engineering of quantum materials. Ann. Rev. Condens. Matter
Phys. 10, 387 (2019)
127. O.D. Mücke, T. Tritschler, M. Wegener, U. Morgner, F.X. Kärtner, Signatures of carrier-wave
Rabi flopping in GaAs. Phys. Rev. Lett. 87, 057401 (2001)
128. Q.T. Vu, H. Haug, O.A. Mücke, T. Tritschler, M. Wegener, G. Khitrova, H.M. Gibbs, Light-
induced gaps in semiconductor band-to-band transitions. Phys. Rev. Lett. 92, 217403 (2004)
129. M. Fleischhauer, A. Imamoglu, J.P. Marangos, Electromagnetically induced transparency:
optics in coherent media. Rev. Mod. Phys. 77, 633 (2005)
130. S.H. Autler, C.H. Townes, Stark effect in rapidly varying fields. Phys. Rev. 100, 703 (1955)
131. W.H. Knox, Dispersion measurements for femtosecond-pulse generation and applications.
Appl. Phys. B 58, 225 (1994)
132. V.M. Silkin, P. Lazić, N. Došlić, H. Petek, B. Gumhalter, Ultrafast electronic response of
Ag(111) and Cu(111) surfaces: from early excitonic transients to saturated image potential.
Phys. Rev. B 92, 155405 (2015)
133. F. Reinert, G. Nicolay, S. Schmidt, D. Ehm, S. Hüfner, Direct measurements of the L-gap
surface states on the (111) face of noble metals by photoelectron spectroscopy. Phys. Rev. B
63, 115415 (2001)
134. B. Gumhalter, D. Novko, H. Petek, Electron emission from plasmonically induced Floquet
bands at metal surfaces. Phys. Rev. B 106, 035422 (2022)
135. Y. Dai, A. Ghosh, S. Yang, Z. Zhou, C.-B. Huang, H. Petek, Poincaré engineering of surface
plasmon polaritons. Nat. Rev. Phys. 4, 562 (2022)
Chapter 4
Nonlinear Soft X-Ray Spectroscopy

Craig P. Schwartz and Walter S. Drisdell

Abstract The history and current state of the field of soft X-ray nonlinear optics is
discussed. The development of soft X-ray nonlinear optics has been largely based
upon the earlier work done with visible and infrared spectroscopy. While nonlinear
optics in those frequency regimes are relatively mature today, soft X-ray nonlinear
optics is a young field. This delayed development was due to the lack of intense ultra-
fast X-ray light sources, but with the recent rise of X-ray free electron lasers and other
intense X-ray sources, there has been an explosion in the field. The specific soft X-ray
nonlinear optics techniques that have been employed so far are discussed in detail
and the underlying physical principles are discussed. These include non-sequential
multiple-photon absorption, stimulated emission to drive lasing, and various forms
of stimulated Raman scattering. These techniques have enabled new measurements
that were previously infeasible, providing access to dipole forbidden states, driving
anisotropic X-ray emission, and allowing for the understanding of excited state
surfaces in an unprecedented way. Additionally, four wave mixing has been demon-
strated and a dedicated four wave beamline built for the study of complex phenomena
such as diffusional dynamics. By performing four wave mixing in the soft X-ray
region, one accesses particularly large momentum transfer and thus spatially small
volumes. Second harmonic generation has also been demonstrated in the soft X-ray
region, allowing for the study of symmetry breaking and surfaces in materials ranging
from batteries to graphite. This technique has recently been extended to a variation
of sum frequency generation, where the soft X-ray beam is mixed with a visible laser
pulse and has been shown to be a sensitive probe of charge localization. Finally, the
available computational methods which can be used to model and understand these
phenomena are discussed.

C. P. Schwartz (B)
Nevada Extreme Laboratory, University of Nevada, Las Vegas, Las Vegas, NV, USA
e-mail: craig.schwartz@unlv.edu
W. S. Drisdell (B)
Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA
e-mail: wsdrisdell@lbl.gov

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 83
I. Matsuda and R. Arafune (eds.), Nonlinear X-Ray Spectroscopy for Materials Science,
Springer Series in Optical Sciences 246, https://doi.org/10.1007/978-981-99-6714-8_4
84 C. P. Schwartz and W. S. Drisdell

4.1 Nonlinear Spectroscopy—Development with Visible

Light

Over the course of an average day, a person may see all sorts of wondrous things,
but they will all be in the regime of linear optics. That is to say, almost all everyday
visual occurrences happen in a regime where the electric field associated with light
can be considered weak. This means that if one makes the light brighter, a material
illuminated by that light will appear brighter, but otherwise its appearance will not
fundamentally change [1]. This was the state of the observable world for most of
human existence, and only the relatively recent development of bright light sources
(notably lasers) has allowed for the development of nonlinear optics [2, 3].
While it may seem like magic, strong fields enable phenomena to be seen which
otherwise are not possible as shown in Fig. 4.1. Specifically, the advent of a strong
electric field can lead to all sorts of phenomena which otherwise would not be observ-
able. Strong laser fields lead to a strong polarization density in a material which then
reacts nonlinearly with the electric field. One of the first applications of lasers was to
demonstrate second harmonic generation (SHG) on a crystal, a phenomenon which
combines two photons of the same energy to generate a single photon of twice the
energy [3]. While the energy is linear (two photons combine to give identical energy
as from the single generated photon), SHG is a nonlinear effect as the signal scales
quadratically with respect to input power. Unfortunately, even with the advent of
lasers, the signal was so weak compared to the much stronger input energy (funda-
mental) that the small signal shown in the first manuscript figure was believed by the
journal editors to be a speck of dust and was removed. The electric fields required
are typically quite large (> 108 V/m) but can be readily achieved with modern lasers
[4, 5].
Given the small signal, it is reasonable to ask why there has been so much work
on nonlinear optics. This is because, in a variety of cases, the only way to determine
certain properties of materials is by using nonlinear optics. Many types of information
cannot be obtained by other methods, such as surface specific phenomena, access
to diffusion rates for phenomena such as heat transfer, self-focusing, generation of
different frequencies of light, and spectroscopy with different selection rules; these
have provided insight in fields ranging from biology to electrical engineering. Most
of these phenomena and the related information simply cannot be obtained in any
other manner besides using nonlinear optics. The exploitation of nonlinear optics is
therefore essential in many fields. For instance, to study buried interfaces, the only
widely accessible measurements use nonlinear optics.
An optical event is when as an incoming light wave induces electric polarization
in a material that generates the outgoing light wave. Nonlinear optics are defined by
being sensitive to the electric field. In general, the instantaneous polarization can be
represented as
( )
P(t) = ε0 χ 1 E 1 (t) + χ 2 E 2 (t) + χ 3 E 3 (t) + · · · , (4.1)
4 Nonlinear Soft X-Ray Spectroscopy 85

Fig. 4.1 Second harmonic generation (SHG), can seem like magic, combining two photons to
generate a single photon at twice the energy. As the signal (2ω) varies as the square of the input
field (ω), a strong field is required. This particular technique is often used to probe surfaces and
symmetry breaking

where P(t) is the polarization, ε0 is the dielectric of free space, χ n is the nth order
susceptibility in terms of the electric field E(t). χ n is usually a tensor of rank n +
1 that is sensitive to the polarization and symmetry of the system. A second order
process such as second harmonic generation would have a χ 2 consisting of a rank
3 tensor, and the higher order terms are not necessary in that case. There are many
varieties of nonlinear optical effects that have been observed experimentally. Some
of them are listed in Table 4.1 with characteristic information. It is worth mentioning
that some nonlinear optical phenomena, such as multiple ionization or saturable
absorption, have also been observed with a linear response of the polarization with
a linear susceptibility, χ 1 . Such optical events themselves are “nonlinear” but the
86 C. P. Schwartz and W. S. Drisdell

equation defines them as “linear”. These examples are given in Table 4.2 and the
description is given in Sect. 4.3.
The case of wave mixing (where two or more photons are combined to generate
a single photon) has shown use in both the generation of tailored light and the study
of materials, as it enables generation of light at different wavelengths [25, 26]. A
whole host of related techniques can be used to mix multiple photons. The mixing of
two degenerate photons is SHG, as mentioned above, and mixing three degenerate
photons to generate a 4th is called third harmonic generation (THG). Mixing two
photons of different energy, where the resultant photon is the sum of the energies of
the input photons, is called sum frequency generation (SFG), and mixing two photons
where the resultant photon energy is the difference between the input photon energies
is called difference frequency generation (DFG). The mixing of an even number of
photons (generally two) has the special property of requiring symmetry breaking in a
material for signal to occur, which requires either a noncentrosymmetric material, or
an interface. This is not true with an odd number of photons which is bulk sensitive,
as will be the case in four wave mixing, which provides similar information to
third harmonic generation—both probe χ 3 , described below—but have different
phase matching conditions. Combining larger numbers of photons is just generally
termed high harmonic generation (HHG) [5, 27]. These techniques have shown a
host of applications in the study of materials and in preparing tailored laser light to
probe other materials through the generation of specific frequencies. This idea of
generating different frequencies to probe matter is frequently used with other wave
mixing techniques such as optical parametric amplification (OPA), optical parametric
oscillation (OPO), and optical parametric generation (OPG) [28–30]. Indeed, much
of modern laser spectroscopy would not be possible without these techniques. The
details of performing these techniques have been reviewed extensively before and
the reader is encouraged to seek out those manuscripts if more detailed knowledge
is necessary [4, 26].
Nonlinear optics, however, is not limited to wave mixing and includes a host
of other processes that are only possible as the field becomes strong enough to
lead to nonlinear effects. These include but are not limited to: transient grating
(i.e., 4-wave mixing), where diffusional processes can be mapped spectroscopically;
Raman amplification which enhances the Raman signal under large fields; the optical
Kerr effect which leads to a change in the refractive index with field; and multi-
photon absorption (i.e., two-photon absorption), wherein multiple (two) photons are
absorbed simultaneously, among many other techniques [2, 31–34]. Many of these
processes provide ways of probing matter in a manner for which there are either
no available alternatives or no options with all the benefits of these methods. It is
important to note that just because an effect can require a strong field to occur, does
not mean the process is nonlinear—the signal must change nonlinearly with respect
to the electric field. For instance, pump-probe spectroscopy where an initial pump
prepares an excited state which is then probed by a second pulse is not necessarily a
nonlinear phenomenon even though typically somewhat large fields are required to
drive a noticeable fraction of the system into an excited state [27].
4 Nonlinear Soft X-Ray Spectroscopy 87

Table 4.1 A list of nonlinear optical phenomena with nonlinear susceptibility, optical properties,
and representative references
Optical phenomena Nonlinear Notes on the Selected references for soft
susceptibility phenomena X-ray experiment and section
in this chapter
Second harmonic χ (2) Sensitive to symmetry [6, 7]
generation (SHG) breaking as an even Section 4.4.5
2ω1 → ω2 order process.
Generates a photon at
twice the input photon
energy
Third harmonic χ (3) Generates a photon Not yet demonstrated in the
generation (THG) with three times the X-ray regime. See [8] for more
3ω1 → ω3 input energy. Has discussion from the optical
been used in the regime
optical regime as a
form of microscopy
Sum frequency χ (2) Similar selection rules A variation with 3 input
generation (SFG) to SHG, but with photons is shown in [9]
ω1 + ω2 → ω3 input photons of Section 4.4.5
different energy
Difference frequency χ (2) Similar selection rules Has been discussed
generation (DFG) to SHG and SFG theoretically in the hard X-ray
ω1 − ω2 → ω3 regime, but not yet
demonstrated with two photons
in the soft X-ray regime [10]
High harmonic Varies Used in the optical [11, 12]
generation (HHG) regime to generate Section 4.2
X-ray pulses
Optical parametric χ (2) Used in the optical Similar technique used in the
amplification (OPA) regime to tune laser hard X-ray regime as X-ray
frequencies parametric down-conversion
[13, 14]
Optical parametric χ (2) Can be thought of as a Not demonstrated in the X-ray
oscillation (OPO) way to improve the regime
focus and power of
light generated with
an OPA
4-wave mixing χ (3) A series of related [15, 16]
techniques where 3 Section 4.4.4
waves interact to
generate a 4th wave.
Amongst the most
common is transient
grating
(continued)
88 C. P. Schwartz and W. S. Drisdell

Table 4.1 (continued)

Optical phenomena Nonlinear Notes on the Selected references for soft
susceptibility phenomena X-ray experiment and section
in this chapter
Raman amplification Can vary Method of amplifying Not demonstrated in the X-ray
a Raman signal regime
Optical Kerr effect Can vary A variety of related Not demonstrated in the X-ray
techniques where the regime
refractive index of a
material varies with
the input electric field
Two-photon χ (3) Two (or in certain [17, 18]
absorption (TPA) cases more) photons Section 4.4.1
are absorbed,
allowing the probing
of different selection
rules
Stimulated emission N/A Drives a population [19, 20]
inversion which then
undergoes a
stimulated emission
process, like lasing
Superfluorescence N/A Ensemble of emitters [21, 22]
coupling to emit an Section 4.4.2
intense burst of light
Stimulated Raman χ (3) A process where a [23, 24]
species is driven into
a higher excited state
before being driven
back down into a
lower state and
emitting a photon

Table 4.2 A list of nonlinear optical phenomena with linear susceptibility, optical properties and
representative references
Technique name Nonlinear Notes on the technique Selected references
susceptibility and section in this
chapter
Multiple ionization χ (1) Multiple electrons are [17]
removed from a bound Section 4.3
system simultaneously or
nearly simultaneously
Saturable absorption χ (1) At high fluxes, absorption [19, 20]
saturates Section 4.3
4 Nonlinear Soft X-Ray Spectroscopy 89

To use nonlinear optics experimentally, a variety of complications must be care-

fully understood and potentially optimized to maximize signal. One example is phase
matching. The phase of light must be considered because the index of refraction of
a material generally changes as a function of wavelength, meaning the angle of the
crystal being studied where the incoming and outgoing waves interfere will have a
large effect on the signal. For a second order process such as SFG, at position x in a
crystal system, the second order polarization is

P 2 (x, t) ∝ E 1n1 E 2n2 ei[(k1 +k2 )x−ω3 t] + (complex conjugate), (4.2)

where E 1 and E 2 are two components of the electric field at frequencies ωa and
indices na , with the wave vector ka = na (ωa ) ωa /c where c is the speed of light, and
na (ωa ) is the index of refraction at frequency ωa . For SFG, definitionally ω1 + ω2 =
ω3 . Therefore, phase matching, which requires constructive interference, can only
be true if the output wave vector (k→3 ) is equal to the input wave vectors (k→1 , k→2 ).

k→3 = k→1 + k→2 . (4.3)

This statement can be generalized for a system of n − 1 input vectors such that

k→n = k→1 + · · · + k→n−1 . (4.4)

This is known as the phase matching conditions, albeit in certain cases the vectors
may subtract instead of adding, as is the case for difference frequency generation. For
certain systems, for instance those with geometric constraints, such matching is not
always possible, making measurements challenging [5]. Perhaps the largest challenge
in general, is that the demands on the light sources are quite high, requiring an intense
light source of monochromatic and coherent light. Even today, these sources simply
do not exist for every frequency range, and even when they do exist they can be quite
expensive to use and maintain. It is almost a certainty that the quality and availability
sources will continue to improve in the future, as they have for the past 50 years.
The key to development of nonlinear optics has been the development of powerful
light sources [5]. The development of bright X-ray light sources has lagged signif-
icantly behind those in the visible, leading to a similar lag in the development of
X-ray nonlinear optics. The current state of the art is significantly less advanced than
that in the visible. In addition, X-rays are generally harder to work with than visible
light, as they are harder to focus and reflect, and require specialized and expensive
optics for these purposes [35]. Only the recent rise of X-ray free electron lasers, with
their high peak photon flux, has enabled significant work and development in the
field of X-ray nonlinear optics [36]. These light sources are discussed in more detail
in the next section.
Visible nonlinear optics and X-ray nonlinear optics in the so-called hard X-ray
regime (photon energies greater than 3000 eV) are fundamentally different. In the
visible the spectroscopies are dipole selected, but in the hard X-ray regime the dipole
approximation breaks down leading to different effects, such as SHG being bulk
90 C. P. Schwartz and W. S. Drisdell

sensitive in centrosymmetric media, and more accessibility of the nonlinear Compton

effect [37–39]. It appears, however, that in the soft X-ray regime (photon energy less
than 1000 eV) nonlinear optics can be described well by the dipole approximation and
the previous equations can be readily applied, and as such visible nonlinear optics
can be used to guide the development of soft X-ray nonlinear optics. The region
below approximately 150 eV is sometimes referred to as extreme ultraviolet (XUV),
although the cutoff between XUV and soft X-ray is not well defined. We wish to
note the description of the development of visible nonlinear optics and its detailed
uses are covered only in a cursory fashion here. For a more thorough treatment, the
reader is encouraged to read the several books that have been published on the topic
[5, 27].

4.2 Ultrafast X-Ray Light Sources

In almost all cases, to observe nonlinear effects one needs a strong field. (The case of
X-ray parametric down-conversion is skipped here, but it is a hard X-ray phenomenon
and not relevant at the energies discussed in this chapter [13, 14].) The original X-ray
sources, dating back to their invention by Roentgen in 1895, were based on X-ray
tubes. While these were improved over the course of decades by developing rotating
anodes which allowed for increased load, the continuous nature of the X-ray emission
and the relatively low flux precludes their use in generating the large fields necessary
to perform nonlinear optics. Additionally, these sources are not generally and readily
tunable in terms of the strongest energy available [40].
The next step in the development of X-ray sources was the development of
synchrotron light sources. These machines were discovered by Veksler in 1944 [41],
built by McMillan in 1945 [42] and the observation of synchrotron radiation was made
in 1947 by General Electric [43]. These machines move electrons at relativistic speeds
around a ring, and as the electrons are accelerated, they emit radiation, a phenomenon
which was not initially expected. Following this discovery, synchrotron light sources
have undergone continuous improvements, from their early use being run parasit-
ically from early synchrotron particle accelerators, to their current designs which
approach the diffraction limit. Synchrotron light sources may now be close to their
ultimate performance limits [44], having become the workhorses of X-ray science.
They deliver stable X-ray light with a high average flux, but their design is not opti-
mized for high field science. This is because while the machines generally run at a
high enough pulse rate to approach a continuous source (MHz), their peak power
(or equivalently their power per shot) is quite low, with pulse lengths typically in
the hundreds of picoseconds and the number of monochromatic photons per pulse
often approaching 106 [45]. While this is many orders of magnitude higher than an
X-ray tube (depending on how this is counted, 15 orders of magnitude may be a
conservative estimate for the increase in monochromatic flux), the field strength is
still much too low to perform most nonlinear optical experiments.
4 Nonlinear Soft X-Ray Spectroscopy 91

The advent of X-ray free electron lasers (FELs), however, brought even more
intense X-ray sources with the potential to drive nonlinear optical phenomena. These
lasers use unbound (‘free’) electrons as the lasing medium [46]. Because the X-
ray radiation they generate does not reflect well, a reflection-based lasing cavity
has not yet been built and the radiation must be generated on a single pass of the
electron beam, leading these machines to be large (European XFEL is 3.4 km long),
and therefore quite expensive. Despite the cost of these machines, their tremendous
power (up to mJ per pulse), short pulse length (as short as hundreds of attoseconds),
and relatively high repetition rate (now approaching MHz), makes these machines
ideal for performing nonlinear optics. Indeed, the vast majority of nonlinear X-
ray experiments that have ever been performed have been performed with these
machines [47–49]. They are not without their drawbacks, however, as the typical
lasing method, self-amplified spontaneous emission (SASE), in essence amplifies
noise and leads to a broad and non-Gaussian spectral and temporal profile [50].
There has been work to improve this by performing self-seeding, which has shown
improved spectral quality and has seen rapid improvements recently [48, 49]. Perhaps
the optimal solution would be to use external seeding, where an initial laser pulse
is amplified, as is done at FERMI in Italy [51]. While this provides a very coherent
and monochromatic source, it does so at the cost of laser power, which is lower
than SASE FELs. Additionally, as amplifying an external laser source amplifies
the noise, using this technique to develop pulses into the hard X-ray regime and
maintaining decent quality pulses has proved challenging so far and will require
the developments of several new technologies [52]. The details of how X-rays from
X-ray tubes, synchrotrons and FELs are generated has been handled elsewhere [44].
The overall progress can be seen in a graph plotting photon energy versus peak
brilliance—which is roughly proportional to monochromatic photon flux [45]. As
shown in Fig. 4.3, conventional lasers can be quite intense in terms of photon power,
but struggle to produce high energy photons. This is contrasted with synchrotrons
which are quite good at generating high energy photons but struggle to produce high
peak flux. With synchrotrons near their performance limits, free electron lasers are
opening entirely new fields of study by enabling high flux and high photon energy.
An example of a modern X-ray free electron laser is shown in Fig. 4.2, taken at the
soft X-ray FEL at SACLA in Japan.
Finally, it is worth acknowledging recent advances in lab-based X-ray sources.
Beyond X-ray tubes, X-rays can be generated in a variety of ways, such as by
impacting a metal with an intense laser pulse, but this provides inadequate photon flux
for nonlinear experiments [53]. Perhaps more likely to be of use for X-ray nonlinear
optics in the future is HHG-based sources, which upconvert a laser pulse into the soft
X-ray region [11, 12]. While at present these sources provide inadequate peak power
for nonlinear optics, they are very attractive in that they maintain excellent pulse
quality and are significantly smaller financial investments than FELs. Current work
has largely focused on increasing the average power of the machine by increasing
the repetition rate, increasing the photon energy to hundreds of eV and decreasing
the pulse length of these sources, but perhaps if more work is performed to optimize
92 C. P. Schwartz and W. S. Drisdell

Fig. 4.2 A photo of soft XFEL at SACLA. By permission of RIKEN

peak flux performing nonlinear optical experiments at soft X-ray energies will be
possible with HHG sources [54].

4.3 Family of Soft X-Ray Nonlinear Spectroscopy

Well-Established Linear Spectroscopies

Before describing nonlinear X-ray spectroscopies already developed, it is perhaps
beneficial to briefly describe the uses of linear X-ray spectroscopy. X-ray spec-
troscopy uses a high energy photon to excite core or semi-core electrons in an atom,
molecule, or material system. These core or semi-core electrons are generally well
separated in energy for each element, allowing X-ray spectroscopy to probe different
elements in a system individually [55]. This contrasts with visible spectroscopy,
which typically probes a delocalized state which can extend over an entire system
[56]. Also, as the excitation is high energy, the spectroscopy is often sensitive to
neighboring molecules as the excited state is often quite diffuse spatially. It therefore
can be a very sensitive probe; for instance the X-ray spectrum of water is significantly
more feature rich than the infrared (IR) spectrum of water [57].
X-ray spectroscopy generally takes one of several forms. In X-ray absorption
spectroscopy (XAS), the input energy is scanned and the amount of photons absorbed
by the sample is measured [55]. The absorption transition can be described (as was
detailed in Chap. 2) by Fermi’s golden rule as
4 Nonlinear Soft X-Ray Spectroscopy 93

Fig. 4.3 Plot of photon energy versus peak brilliance, roughly proportional to peak photon flux.
Conventional lasers are high flux but low photon energy whereas synchrotrons are high energy
but low flux. The X-ray free electron laser is uniquely high flux and high energy. X-ray tubes can
produce high energy X-rays, but their flux is so low it is off the scale of the plot. HHG sources
are at present above synchrotron peak brightness (up to ~ 1025 photons/s/mrad2 /mm2 /0.1%/-BW at
100 eV). The pace of their improvement is significantly faster than that of synchrotrons, however.
Taken from Boutet and Yabashi [45]. Reprinted with permission from Boutet and Yabashi [45].
Copyright by Springer Nature

2π ||⟨ ⟩|2
⎡i→ f = f |H ' |i | ρ(E f ), (4.5)
h
where the transition probability from the initial state (i) to the final state ( f ) is given
by ⎡i→ f . H ' is the Hamiltonian describing the transition from initial to final, ρ(E f )
is the density of states, and h is Planck’s constant divided by 2π. In the case of
XAS, the transition is from a relaxed state (i) to a core excited state ( f ). As this is a
single photon process, there must be a single unit of angular momentum difference
between i and f , often meaning a transition from an s to a p-type orbital. As f is
unoccupied, this provides a sensitive probe of the unoccupied projected density of
states (for instance K-shell excitations are from the 1s shell into unoccupied p-type
states). In X-ray emission spectroscopy (XES), the input energy is well above the
core or semi-core electron binding energy, and the emitted photons are measured as
a function of energy. Many of these are elastic, emitted when the excited electron
relaxes directly back to its original core or semi-core state. But there is also inelastic
94 C. P. Schwartz and W. S. Drisdell

emission, from electrons transitioning from valence states into the empty core hole
left by the X-ray excitation. This provides a measure of the projected density of
states of the occupied density of states [58, 59]. In terms of Fermi’s golden rule, i is
a core hole state and f is the relaxed state it transitions into. As the process is single
photon, the same selection rules a XAS apply. X-ray emission spectra can also be
measured when the input X-rays are resonant with an excitation to a specific final
state, known as resonant inelastic X-ray scattering (RIXS). This in essence combines
X-ray absorption and X-ray emission spectroscopy [60].
It is also possible to examine the electrons emitted from a system. X-ray photo-
electron spectroscopy probes the energies of the initial occupied electronic states by
measuring the energy of the photoelectrons excited from these states by an incident
X-ray photon. Introducing incident electrons of constant energy and measuring the
emitted photon energy is called inverse photoemission and is sensitive to the unoccu-
pied states [61]. All of these techniques would be classified as linear, but all nonlinear
soft X-ray spectroscopy are built upon linear techniques in some way. Linear X-ray
spectroscopy is described in detail in Chap. 2.
High Field Linear Soft X-Ray Spectroscopies
Beyond measurements that can be made at synchrotrons, there are other linear effects
that can be observed at FELs. The most common of these are discussed briefly here.
Among the first papers to be produced from an X-ray free electron laser (XFEL)
was a demonstration of the multiple ionization of neon [62]. In terms of Fermi’s
golden rule, this means it goes through a series of excitations from the ground state,
where after the first excitation the ground state has become some ionized state. It was
shown that the highly intense X-ray pulses would drive the formation of core holes
by removing the inner core electrons through direct excitation; the valence electrons
would then fall into the core within the duration of the X-ray pulse, which then
removed these formerly valence electrons as well. This meant the way to maximize
the ionization—as was demonstrated for both an atom (Ne) and a molecule (N2 )—
was to increase the length of the laser pulse, not the field strength [62, 63]. So while
this is an interesting high field phenomenon, it is not an example of nonlinear optics.
Other related high field phenomena in the soft X-ray (and demonstrated as well
in the hard X-ray regime on Fe) include the phenomena of saturable absorption (SA)
and X-ray induced transparency [64, 65]. In SA, the amount of absorption is so high
that the final states are filled and absorption is saturated, manifesting as a decreasing
amount of the rate of absorption with increasing laser power. When the ground state
is fully depopulated by a phenomenon such as SA, this will give rise to the related
phenomenon of X-ray induced transparency, wherein X-rays which would normally
be absorbed by the material are instead fully transmitted, because all the core electrons
have been depopulated. This may have use in futuristic advanced materials where
X-ray transparency can be turned on and off. This has been demonstrated at the L-
edge of solid Al [66], and the N-edge of tin [67]. While both these phenomenon are
dependent on the pulse power per area, the effect is not nonlinear with respect to the
electric field, so here we will not consider it a nonlinear optical effect, although such
a term is used elsewhere.
4 Nonlinear Soft X-Ray Spectroscopy 95

4.4 Nonlinear Soft X-Ray Optics and Spectroscopies

This section will deal primarily with various soft X-ray nonlinear optical techniques.
These include multiphoton absorption, stimulated phenomena, stimulated Raman,
four wave mixing, and second harmonic generation. Some of the key results are
summarized in Table 4.1, listing nonlinear susceptibility and a few representative
papers.

4.4.1 Multiphoton Absorption

One of the most used nonlinear optical techniques in the visible is two-photon absorp-
tion (TPA) [68]. With this technique, rather than a single photon being absorbed,
followed by another photon, two photons (generally of the same energy for practical
reasons) are absorbed simultaneously. This has shown use in the visible as a method
of photolithography for generating narrow etched lines, as only the most intense
part of the laser pulse will have the requisite intensity for two-photon absorption,
ultimately leading to higher resolution linewidths [69, 70].
TPA also has different selection rules, and can therefore access states that are not
normally allowed [71]. Fermi’s golden rule to the first order (as shown previously in
Eq. (4.5)) is inadequate for this two-photon process. This is because the simultaneous
absorption of two photons allows for transitions to states that are dipole forbidden
with a single photon, or that require a change of two units of angular momentum.
In the X-ray region, TPA forbids s → p transitions but allows (for instance) s → d
transitions, or s → s transitions (allowed angular momentum changes for two photons
are 0, ± 2) generating an entirely different spectrum than the low field single photon
absorption spectrum [72]. This allows for matter to be characterized in a way that is
not possible at lower fields; these otherwise inaccessible states can have interesting
properties such as different symmetry and different lifetimes. TPA also has a different
absorption rate compared to linear absorption. Linear absorption allows transmission
proportional to Beer–Lambert law, corresponding to an exponential decay in intensity
with regards to sample thickness

I (x) = I0 e−αx , (4.6)

where I(x) is the intensity, I 0 is the input intensity, α is the single photon absorption
coefficient, and x is the sample thickness. By contrast two-photon absorption is
described by

I0
I (x) = , (4.7)
1 + βx I0
96 C. P. Schwartz and W. S. Drisdell

with β being the two-photon absorption coefficient and the other terms the same as
in the single photon absorption case. The linear intensity is dominant at low field
strength, but this is not always true. In the limit of incredibly high fields, two-photon
absorption and other multiple-photon absorption processes can become dominant.
Indeed, this is how the evidence of TPA in the soft X-ray regime was noticed. The
production of Ne9+ was observed at lower photon energies than what was required
for Ne8+ [17]. Following careful theoretical analysis, Doumy et al. assigned it to
two-photon absorption; they cleverly changed the input energy for ionization and
showed that the absorption rate was different as the energy was moved from 1110
to 1225 eV, going from a quadratic scaling factor with respect to pulse power to one
that was roughly linear (exponent of 2.0 vs. 1.1, respectively). 1110 eV should not
be enough energy to ionize Ne8+ into Ne9+ , allowing for TPA without the possibility
of single photon absorption. The reason the scaling factor is not 1.0 for the case of
1225 eV is probably related to other processes beyond simple absorption occurring,
including TPA amongst many other possible effects, as shown in Fig. 4.4. The cross
section of TPA in the case of Ne9+ was many orders of magnitude higher than what
had been expected, which is likely why it was observable.
Two-photon absorption had previously been seen for helium at 41.5 eV [73, 74]. In
this case, light generated from an HHG source was incident upon helium, generating
a variety of ions [18]. Because the spectra of helium are so well understood and
the quality of the data was so high, the research team was able to determine the
cross section and compare it to theory, with reasonable agreement, although there is
some error related to unknown amounts of 83.0 eV contamination from the source.
While the importance of this specific result is perhaps limited, the ability to do these
measurements in a lab setting is a significant achievement and opens the possibility
of future nonlinear measurements in lab settings.

Fig. 4.4 Normalized ratio of Ne9+ to Ne8+ for excitation with 1110 eV (filled circles) and 1225 eV
(open squares). The fit for the 1110 eV showed a quadratic dependence with flux (E 2.0 ) whereas the
fit for 1225 eV showed a roughly linear dependence (E 1.1 ) [17]. Reprinted with permission from
Doumy et al. [17]. Copyright (2011) by the American Physical Society
4 Nonlinear Soft X-Ray Spectroscopy 97

Finally, there have been two sets of measurements on graphite. The first measure-
ment [75] showed two-photon absorption near the carbon K-edge. However, this
view has been revised with more recent data and is now ascribed to shifting in the
edge energy over the finite length of the pulse [65]. In the newer work, if is found
that TPA is fundamentally competing with SA, described in Sect. 4.3. What is seen
is that at relatively low fluences SA occurs, and at higher fluences the importance
of TPA is observable. In solids the processes occur at different rates than in the
case of gases, due to the very different local environment when comparing a gas to
condensed matter.
This all indicates that spectroscopy at very high fluences will need to be carefully
considered going forward, as multiple effects can occur. This complexity inherently
grows as the systems grow more complicated. However, it ultimately provides unique
insight into materials and offers unique abilities to control light and matter in ways
that are otherwise not possible.

4.4.2 Stimulated Emission/Forward Scattering

While the generation of laser light from a material via stimulated (rather than spon-
taneous) emission is generally not considered a nonlinear optical technique as the
high fields do not drive the process, the process requires intense X-ray sources and
leads to a significant nonlinear enhancement in the number of photons generated as
compared to spontaneous emission. In the case of Ne, an X-ray laser was made by
pumping a gas with very intense X-ray pulses which causes a population inversion of
the neon single excited ion. This transition to stimulated emission from spontaneous
emission for 960 eV input photons occurred at roughly 0.15 mJ power [76]. For an
infinitesimally thin sample with interaction length dx, the emission probabilities P
can be described as

P = σρatom dx, (4.8)

and

P = σρch dx, (4.9)

where σ is the cross section of either absorption or stimulation, ρatom and ρch are
the number density of absorbing atoms and core holes, respectively. As such, more
core holes linearly increase the probability of stimulated emission. In the limit where
stimulated emission is large (P ~ 1), the ratio of densities can be expressed as

ρch λ
= , (4.10)
ρatom dx
98 C. P. Schwartz and W. S. Drisdell

where λ is the absorption length. For high efficiency, the absorption length relative to
the interaction length has to be similar to the ratio of excited atoms. For comparable
interaction and absorption lengths it would require full excitation of the system, that
is to say a population inversion will be required, i.e., lasing will occur. The ability to
drive this type of lasing phenomenon should open the way to future nonlinear X-ray
studies. This was later extended to show that in certain situations superfluorescence,
a quantum effect leading to the collective emission of fluorescent light, similar to
but distinct from amplified stimulated emission, could also occur [21, 22]. There has
been a variety of work expanding upon this in the hard X-ray regime with potential
spectroscopic applications [77–79].
The demonstration of stimulated emission for materials science with soft X-rays
was made in 2013 [19]. This work was motivated by X-ray emission generally being
an isotropic process with the efficiency for low energy excitations often below one
percent [80]. This means that sample damage from nonradiative decay, such as Auger
decay, limits the ability to make measurements on samples that are sensitive to radi-
ation damage, notably including biological and organic molecules. Using stimulated
emission instead allows for a higher conversion efficiency and thus less sample
damage by mitigating processes such as Auger decay.
This is identical to what drove the lasing phenomenon seen in neon. Beye et al.
propose that stimulated emission can be used with nonlinear optical techniques to
enable measurements such as X-ray emission on samples that are highly susceptible
to damage.
This work was expanded upon more recently by studying the emission from Co/Pd
multilayers. Through careful design of the X-ray pulses, Stöhr and coworkers were
able to observe an enhancement of 106 for stimulated versus spontaneous emission
[81]. This large enhancement was found for both inelastic X-ray scattering as well
as elastic X-ray scattering [81]. However, they also found that there were effects
caused by the temporally spiky nature of the SASE radiation used in this study, as
the electrons would diffuse between spikes within a pulse causing unwanted spectral
effects. Thus, they suggest that the ultimate way to minimize the unwanted effect
and enhance the stimulated RIXS spectrum would be for a single short (possibly
hundreds of attosecond) spike. This would enable linear spectroscopy efficiently in
the high photon flux regime often associated with nonlinear optics.
On the same Co/Pd system at the Co L-edge, differences in the absorption and the
magnetic circular dichroism spectra were observed when comparing XFEL measure-
ments to synchrotron measurements. It appears that over the intensity range available
at XFELs near absorption edges, the intense X-ray pulse changes the valence elec-
tronic structure, which can then be observed by differences in absorption [20]. This
paper also saw enhancement in elastic X-ray scattering. Such an enhancement has
been observed previously in a series of manuscripts. It was first proposed that this
would occur in high intensity coherent X-ray pulses leading to a deviation from the
Beer–Lambert law [82]. The results verifying this theory showed up only a year
later, when high fluence led to a loss of diffraction signal in Co/Pd systems [83].
This was definitively shown two years later by a loss of diffraction signal with a
simultaneous rise in forward elastic scattering for the same system [84]. Stöhr and
4 Nonlinear Soft X-Ray Spectroscopy 99

coworkers were able to explain how the ratio of stimulated to spontaneous emission
affected transmission and diffraction contrast as a function of intensity.

4.4.3 Stimulated Raman Scattering

A critical milestone in the development of 2D X-ray spectroscopy is a demonstra-

tion of the ability to control the electronic state of molecules as well as control the
motion of the constituent atoms such that they can be driven in specific relative
motion. This can in principle be used to drive highly specific reactions as well as
understand the motion of atoms. While this was theorized for a long time with X-rays
[85], the cutting-edge experiment to coherently drive matter from an excited state to
another excited state is relatively recent, using stimulated electronic X-ray Raman
scattering [23]. In normal Raman scattering, a single pump photon is input ωPump ,
and some of the energy is lost (or gained) inelastically, corresponding to a difference
of ωSignal . In contrast, in stimulated Raman, two photons are input, the pump photon
ωPump , and the Stokes photon ωStokes . When the difference between the two photons
(ωPump − ωStokes ) corresponds to a Raman transition ωSignal , a large enhancement is
seen due to stimulated Raman scattering. In the frontier X-ray experiment of stim-
ulated Raman scattering, an X-ray pulse pumped Ne gas into an excited state and
then resonantly deexcited the atom into a lower energy state accompanied by photon
emission. This is possible due to the wide energies available in a SASE pulse and its
finite pulse length, ultimately leading to an emission band in this case, slightly lower
then the input photon energy. In this case the input energy was roughly 870 eV and
the emitted energy was roughly 850 eV. This pushing of an atom from a strongly
excited state into a less excited state by resonant deexcitation is a demonstration of
control of an excited state. This type of exquisite experimental control will be neces-
sary for the ultimate realization of 2D X-ray spectroscopy. Unfortunately, despite
substantial effort to extend this technique [86], advances have been slow due to a
range of experimental challenges. One pulse schema suffer from the inherent jitter
and irreproducibility of individual shots, making data analysis and definitive proof
hard at SASE beamlines, and the spectral width of seeded lasers are too low to easily
observe Raman deexcitation in a single pulse. More advanced techniques require
two pulses with well-tuned parameters and high power. This is the kind of problem
which is challenging for current FELs as the FELs with excellent spectral properties
have limited energy tunability for two pulses and lower power, and the FELs with
tunability for a second pulse and high power produce noisy pulses. It does seem
inevitable that newer machines will enable these measurements eventually.
In 2020 while studying nitric oxide, researchers were able to drive NO into a
core hole excited state, before pushing another electron back down into the core hole
leaving the element in a relatively stable excited state [87]. This excited state could
be used to drive chemical reactions and otherwise prepare molecules into specifically
chosen states and extends upon the previous work by demonstration atomic control in
a molecular system. It also provides the opportunity to track the motion of electrons
100 C. P. Schwartz and W. S. Drisdell

within a molecule with higher spatial and temporal resolution than can be obtained
with direct excitation of the same energetic states. This technique requires fast pulses
and high fields, so the increasing availability of higher intensity X-ray pulses should
make such experiments more accessible.
Finally, it was noticed that when exciting neon in a particularly strong resonant
X-ray field, it was possible to drive it from an excited state (hundreds of eV) into
a slightly excited yet stable neutral state (few eV) which does not undergo recoil
after excitation due to no emission of an electron [24]. This is done by exciting the
electron and then driving it back down into a neutral but excited ground state that
is below the binding energy of neon, preventing the possibility of the of emission
an electron. When imaging an atomic beam of xenon, it is quite apparent that these
excited atoms exist, as the excited xenon molecules show up as a solid line on
the 2-dimensional imaging detector, tracking the path of the beam of atoms which
was inserted. Interestingly, this phenomenon is seen only at 855 eV excitation, not
860 eV excitation, indicating it is sensitive to the specific nature of the excited state.
Regardless, such exquisite control is a demonstrative step toward 2D spectroscopy
of neon atoms. This is shown in detail in Fig. 4.5.

4.4.4 Four Wave Mixing

There is only a single beamline in the world dedicated to the study of nonlinear
phenomena: TIMER at FERMI. That beamline is dedicated specifically to the four
wave mixing (4WM) technique, a powerful method demonstrated with optical pulses
before its more recent push into the X-ray regime [88]. In this technique, two (typi-
cally degenerate) pulses are incident upon a surface with some angle relative to each
other. The positive and negative interference between the two waves generate areas
which are excited and other areas which are not excited in the form of a transient
grating. This allows for time scale of properties from attoseconds (if the initial pulses
are fast enough) to dynamics of hundreds of picoseconds as the system returns to
initial conditions. Over time this energy will diffuse, and it can be used to study
various dynamic properties such as thermal transport, coherent phonon diffusion,
and ultrafast magnetism by probing it with a third wave, which scatters off the tran-
sient grating and in the process generates a fourth wave which is then detected [31].
This technique is quite important in the visible and has been used to study a variety
of materials such as radiation materials, quantum materials, and magnetic materials
[89–91]. It is important that phase matching conditions be met such that the total
momentum of the four pulses cancels out. In the case of the geometry used in typical
transient grating experiments

k→FWM = k→1 − k→2 + k→3 , (4.11)

4 Nonlinear Soft X-Ray Spectroscopy 101

Fig. 4.5 a Excitation and decay pathway in neon for X-ray scattering near 855 eV. The initial pulse
excited the laser into a state Ne[1s−1 3p] state, followed by either Raman emission to the Ne[2p−1 3p]
state or auger emission (blue line to gray horizontal line). The Ne[2p−1 3p] state can decay to a long
lived metastable Ne[2p−1 3s] state which can be detected. b Experimental setup for this experiment
including position-sensitive MCP detector to observe excited Ne atoms. c Distribution of excited
neon atoms deflected as a result of spontaneous X-ray scattering leading to momentum transfer with
an incident photon energy of 855 eV. The narrow vertical line is not seen when excited at 860 eV
and is due to the mechanism described in a. Adapted from Eichmann et al. [24]. Reprinted with
permission from Eichmann et al. [24]. Copyright by Science AAAS

where k→FWM is the signal, k→1 and k→2 form the grating, and k→3 is the probe pulse. In the
general case where the grating is made up 1 frequency, k→1 and k→2 will be the same
energy. The signal is generally time dependent due to both a long-term exponential
decay of the signal and the sinusoidal ‘beating’ of the signal. This generally takes
the form of
| |2
Signal ∝ |e−Δt/τi sin 2π vi Δt | , (4.12)

where Signal is the signal, Δt is the time delay, and τi and vi are the frequencies and
damping times of the impulsive stimulated mode. There are a variety of prefactors
102 C. P. Schwartz and W. S. Drisdell

which can be dependent on the specific nature of the measurement and will otherwise
affect the absolute signal but not the general shape.
A higher frequency grating, and thus a higher momentum transfer and energy,
allows one to go from studying molecular excitations to studying electronic excita-
tions, in addition to the thermal and acoustic modes. This means that properties on a
smaller length scale, approaching the size of a unit cell can be studied. If pulses as
high energy as hard X-ray pulses were used as a grating the length scales that could
be studied approach nanometers and would allow for determination of phenomena
like thermal diffusion on an unprecedentedly small length scale. The length scale is
limited by the wavelength of the pulses that make up the grating, so shorter wave-
length allows for more microscopic information. Specifically, the spacing of the
grating is given by

L = λ/2 sin θ, (4.13)

where L is the spatial periodicity, λ is the frequency of light used in the grating, and θ
is the angle between the two pulses that make up the grating. For that reason, there has
been tremendous effort put into the development of soft X-ray 4WM and thus lower
λ, as soft X-rays have significantly shorter wavelengths than the UV wavelengths
used previously, and is a necessary step to shorter frequencies, as well as allowing
for the study of phenomena on the tens of nanometers length scale. The key to these
measurements is that the transient grating must be made from relatively coherent
light, as otherwise the grating will be of insufficiently high quality. Thus a seeded
FEL with a beam splitter was key to making the cutting-edge measurements [15].
The measurements were therefore conducted at FERMI, an externally seeded
FEL, on SiO2 , likely chosen because it is easy to work with and made for a strong
demonstration [15]. Part of the challenge is overlapping two XFEL pulses in space
and time while also meeting the phase matching conditions. In this demonstration, the
probe pulse used was an optical pulse, which allows for high intensity and a frequency
that exploits an optical resonance, allowing for one to examine the destruction of a
specific state due to the existence of the grating. The experimental and conceptual
setup is shown in Fig. 4.6. The phase matching conditions are also shown in Fig. 4.6.
Because the FEL is inherently ultrafast, it is quite simple to demonstrate the time-
dependent nature of the technique. The ultrafast beating as the transient grating
sinusoidally oscillates was readily detected even in these initial measurements.
Using an X-ray probe pulse with an X-ray transient grating, however—a so called
all X-ray transient grating experiment—would allow for one to exploit the element
specific nature of X-rays as a probe, so it was perhaps inevitable that such a measure-
ment would be accomplished [92]. By cleverly exploiting the harmonics generated
at FERMI, the same group demonstrated an all X-ray experiment from both a thin
and a thicker sample which allows for different experimental geometries. They were
able to estimate the effective cross section for 4WM in the all XUV setup of 6 ×
10−24 m2 V−2 . While in this case the probe pulse was not used in a resonant scheme,
this experiment clearly opens the possibility of exploiting that phenomenon.
4 Nonlinear Soft X-Ray Spectroscopy 103

Fig. 4.6 a Experimental schematic of FWM experiment. ɅEUV = 27.6 nm and λopt = 392.8 nm
were used here. The CCD is placed in the expected direction of the FWM signal beam (kFWM ), as
determined by the phase matching conditions. b Phase matching conditions used here where kopt ,
kEUV1, and kEUV2 are the wavevectors of the optical and of the two FEL pulses, respectively [15].
Reprinted with permission from Bencivenga et al. [15]. Copyright by Springer Nature

Silicon nitride has become the most reliable standard for X-ray 4WM experiments
because it’s cheap, readily available, and reasonably robust. In fact, after SiO2 , the
next measurement published was on Si3 N4 [93]. These initial measurements estab-
lished that 4WM is a general technique and could be applied to a wide variety of
systems, and that the systems—not surprisingly—will show different cross sections
and different time constants depending on the specific material being studied. It
was also discovered that these systems can have quite complicated dynamics, with
Si3 N4 displaying both a slow (extending over 100 ps) and rapid component (as fast as
100 fs can be resolved) to its time decay following excitation. The power dependence
was also investigated, and the system behaved as expected indicating that four wave
mixing in the X-ray regime will behave similarly to that of optical 4WM. Perhaps
the most important result of this study was showing that sample damage is going
to be a problem for X-ray 4WM experiments. While certain samples can work, at
present there is always a risk of sample damage before signal can be obtained and
experiments must be carefully planned.
However, silicon nitride is a particularly robust sample and this makes it an excel-
lent tool for the curious scientist. Because it is so robust, Si3 N4 was used as an
example for a resonant transient grating (in this case, resonant with the Si L-edge)
[94]. The probe pulse remained optical, and the work was otherwise similar to the
previous work on Si3 N4 with an optical probe, the large difference being the switched
pump energy. By resonantly exciting the silicon, the method of thermalization was
changed dramatically as the silicon can relax by newly accessible channels involving
core hole decay such as Auger and fluorescence. This led to faster dynamics when
the silicon nitride is resonantly excited.
The most recent and most advanced published FEL work using 4WM concerned
optimization of grating resolution. Using both Si3 N4 and crystalline silicon as
samples, the authors were able to show different dynamics between the two samples
104 C. P. Schwartz and W. S. Drisdell

[16]. Of more significance in the long term, they used grating wavelengths of 13.3 nm
spacing, producing period spacings of 28 nm. This means it is possible to observe
thermal and coherent phonon dynamics on length spacings as small as tens of
nanometers. At this length scale, it becomes possible to study the effects of defects
in materials on phenomena such as heat flow and magnetic dynamics. It may even
be possible to investigate structural effects in liquids, although for simple liquids
such as water the ideal grating spacing will be even smaller, which will require using
higher X-ray energies for the grating.
The most comprehensive X-ray 4WM study compared diamond, glass, and
Bi4 Ge3 O12 [95]. By performing Fourier analysis on the resulting spectrum, the
observed signal could be assigned to Rayleigh surface waves and bulk longitudinal
waves moving throughout the sample, with frequencies ranging from THz to GHz.
Waves that were combinations of these waves were also observed—a combined
Rayleigh and longitudinal wave for instance. Bi4 Ge3 O12 was found to have a coherent
optical phonon mode at approximately 3 THz. The authors point out that this method
can be used to generate waves of a wide variety of acoustic frequencies without the
need of making a transducer. While this is true, a transducer is generally more prac-
tical and cost effective. It will be more immediately useful in explaining particularly
interesting phenomena such as microscopic heat diffusion.
Finally, there has been work reported on the 4WM of silicon carbide [96]. In this
case, a transient grating with a period of 84 nm was used. At these length scales the
Fourier heat equations break down, but the system can be fully described by Lamb
waves, elastic waves that travel along the surface. Additionally, there was a slight
angular difference in the signal corresponding to the signal strength. Although not
definitively shown, the authors speculate it is related to time-dependent non-isotropic
changes in refractive index.
There has also been extensive work in demonstrating four wave mixing with
gaseous samples using lab-based HHG systems. While these sources lose the pulse
power and often the energy range of FELs, they have large advantages in that
the dedicated equipment is comparatively cheap and optimized for the measure-
ments, and signal can be averaged indefinitely without needing to accommodate
other measurements.
Amongst the first to make such measurements were Leone and Neumark, who
led a series of measurements mixing XUV with near IR in a 4WM setup. They first
demonstrated that they could drive neon into an electronically resonant state (the ns/
nd state) which would then transition to the ground state [97]. They also successfully
demonstrated the expected behavior with regards to signal versus input power inten-
sities and the expected sinusoidal decay of signal as is seen in other frequency ranges.
One interesting aspect of this study was that the signal changed monotonically with
increasing gas pressure, suggesting phase matching conditions are important even
in these conditions. A similar result demonstrating electronic sensitivity was shown
by a separate research group demonstrating ultrafast oscillations of neon [98]. The
work on phase matching has been extended to really explore the detailed effects of
phase matching conditions [99].
4 Nonlinear Soft X-Ray Spectroscopy 105

Eventually, a noncollinear setup where the HHG pulse was offset from the two
Near IR (NIR) pulses was demonstrated [100]. In this arrangement, the HHG pulse
intercepts the NIR pulse, combines with two NIR photons, and generates a photon
offset angularly from both the NIR and HHG pulse. This setup is highly desirable
because it removes the background and also allows one to see various four wave
mixing process and their different excitation pathways, and in certain cases weaker 6
wave mixing processes were observed. The research team were able to show coherent
Rydberg wave packets, which could be prepared in different ways depending upon the
experimental details. They point out this is a significant step toward multidimensional
spectroscopy with coherent dynamics. This would be similar to 2D-infrared but with
X-ray pulses, allowing for 2D-X-ray spectroscopy. This work has now been extended
to self-heterodyne, enabling even cleaner spectra, and up to eight distinct states in
the case of helium [101]. The more states that can be seen, the more accurately the
electronic structure of the material can be understood.
It was subsequently demonstrated that dynamics as fast as attoseconds could be
probed. First, krypton was studied and dynamics as fast as 22 fs were demonstrated
[102]. However, because the technique is so sensitive, multiple different decay chan-
nels with different decay times are all simultaneously accessible. Because the signal
is convoluted by all these decay channels determining the ultimate time resolution
is difficult, but probably was noticeably faster than 22 fs. This was later proven
by some of the same authors [103]. Ultimately, by working with attosecond XUV
pulses in helium mixed with few cycle NIR pulses, they demonstrate that the reso-
lution approaches hundreds of attoseconds. It is certainly better than 1.5 fs which
can be determined from the higher order diffraction peaks from the transient grating
and could be assigned based on excellent matching with the Maxwell–Schrodinger
equation.
One clever variation of four wave mixing is to use three nondegenerate pulses.
While this makes the experimental setup more difficult, it enables the effect of each
individual pulse to be separated and studied in a simpler manner. This has been
accomplished and was used to study lifetimes of argon as short as 200 fs [104].
Such a setup may ultimately have limited time resolution as mixing three separate
pulses inherently increases jitter and uncertainty. However, this is not a theoretical
limit, and subsequent work did claim substantially higher time resolution (prob-
ably approaching 10 fs) [105]. The ability to have three separate pulses may prove
important in the demonstration of future multidimensional measurements.
Four wave mixing has already shown uses beyond simple demonstration spec-
troscopy. It has been used to narrow the bandwidth of XUV pulses, and such a
scheme should be extendable to the X-ray regime [106]. However, this is probably
not necessary for seeded FELs which have intrinsically narrow bandwidth. Finally,
we wish to note that 4WM been extended into the hard X-ray regime, albeit using
a very different setup [107]. For matters requiring investigations on the truly atomic
scale, hard X-ray 4WM provides a unique probe for which there are no currently
existing analogous techniques.
106 C. P. Schwartz and W. S. Drisdell

4.4.5 Soft X-Ray Second Harmonic Generation/Sum

Frequency Generation

As discussed above, one of the most common uses of nonlinear optics is to combine
waves. When one combines two waves through either SFG or SHG the signal will
only be generated when there is no center of inversion, i.e., the measurement occurs at
a place that lacks centrosymmetry. This is generally, although not always, surface and
interfaces. This centrosymmetry argument does not hold when one combines three
photons to generate a fourth, as in 4WM. Broken inversion symmetry can be shown
to be essential for even order nonlinear processes by considering the expansion,

∑
n
P→ = ∈0 χ (i ) E→ i , (4.14)
i=1

where P→ is the polarization, ∈0 is the permittivity of free space, χ (i) is the nonlinear
susceptibility of order i, and E→ i is the electric field associated with order i and n is the
arbitrarily large polarization expansion. If the inversion is defined as Iinv P→ = − P→
then it can be shown that
( )
∑ n ∑n
→ →
Iinv P = − P = Iinv ∈0 (i ) → i
χ E = ∈0 (− 1)i χ (i) E→ i , (4.15)
i=1 i=1

and adding the first equation to the second yields

∑
n/2
P→ − P→ = 0 = 2∈0 χ (2i) E→ 2i . (4.16)
i=1

Proving that the signal will be zero when the process is even order and inversion
symmetry is present. This argument does not hold in the hard X-ray regime for both
SFG and SHG [38, 39]. This is because in this regime, as the wavelength is on the order
of the density fluctuations in the electron density of a material, in essence everything
becomes a surface and SFG/SHG signal is generated in the bulk, causing the previous
equation to break down at small wavelengths. At these particularly short wavelengths
the dipole approximation can also become inadequate, as the electric field may not
be constant over the length scale of the excited transition. What would happen in the
soft X-ray regime, with photon wavelengths between visible and hard X-ray, was not
obvious before the soft X-ray SHG measurements were made. Typically, soft X-rays
processes are dipole selected, but this had not been proven for even order nonlinear
processes.
The pioneering demonstration of SHG in the soft X-ray regime used graphite
samples in transmission mode, exploiting the coherence of the FERMI FEL source
[6]. Signal was observed at incident photon energies below the carbon K-edge, reso-
nant with the carbon K-edge and above the carbon K-edge, but generally close to
4 Nonlinear Soft X-Ray Spectroscopy 107

the carbon K-edge. A quadratic power dependence in accordance with theory was
seen. DFT calculations were also performed to try to simulate the spectrum, but with
only three experimental energy points the simulations could not be fully validated
by experiment. More significantly, both the calculations and the experiments showed
that the measurements were surface sensitive. This was done theoretically by deter-
mining the signal as a function of slab thickness and noting that is saturates after
about 4 atomic layers. Experimentally, SHG signal was shown to be constant for
transmission samples of varying thickness, implying that signal was generated only
from the front and back surfaces. This measurement was performed in a single shot
mode, where the graphite films were destroyed with each shot. The experimental
geometry (transmission) is detailed in Fig. 4.7a.
A different research team simultaneously conducted separate SXSHG measure-
ments at SACLA, studying a noncentrosymmetric crystal, GaFeO3 , at the Fe M-edge
[7]. This contrasts with the previous experiments at FERMI, but by using a bulk
crystal these measurements were made in a reflection geometry, and perhaps most
significantly, below the damage threshold, meaning the measurements can be made
on rare and hard to produce samples. In this case, several energy points near the Fe
M-edge in double resonance were measured and with enough energetic data points
to be able to show definitively that SHG occurred. Once again the expected quadratic
signal was observed. The reflection geometry and associated theory is shown in part
b of Fig. 4.7. These two measurements collectively demonstrated second harmonic

Fig. 4.7 a Experimental setup of a transmission experiment on graphite performing SXSHG at

FERMI free electron laser (left). In this experiment, pulses were transmitted through sheets of
graphite at roughly resonant with the carbon K-edge (right) generating signal at roughly 600 eV
and a quadratic dependence with no thickness effect was seen (middle). This suggests SXSHG is
largely surface sensitive [6]. Reprinted with permission from Lam et al. [6]. Copyright (2018) by the
American Physical Society. b Experimental setup of a reflection experiment on GaFeO3 performing
SXSHG at SACLA (left). In this experiment, pulses at half resonance (right) were reflected off the
sample. Once again a quadratic signal dependence on input power was observed [7]. Reprinted with
permission from Yamamoto et al. [7]. Copyright (2018) by the American Physical Society
108 C. P. Schwartz and W. S. Drisdell

generation at input frequencies that were resonant with the one photon transition,
generating a photon at twice the resonant frequency (2ω), as well as input frequencies
that are half the resonant energy of a single electron transition, generation photons
at a frequency of ω.
A study of boron films covered by a thin 100 nm layer of a polymer was conducted
in transmission mode [108]. This boron/polymer interface was chosen as an example
of a buried interface, which are generally difficult to probe with existing experimental
methods. In this case, by tuning the FEL source to the boron K-edge, the X-rays
passed through the polymer which contained no boron and were thus only probing
the boron layer. Distinct spectra were observed for the boron/polymer (B/P) interface
compared to a bare boron control sample, and the researchers were able to deter-
mine the separation between the polymer and the boron layer to within ~ 0.2 Å by
comparing calculations to experiments. In comparison, XAS of boron/vacuum and
boron/polymer were roughly identical as shown in Fig. 4.8. One of the advantages of
this study over the previous carbon work was more spectral points, enabling a better
interpretation of the data. This is a measurement which is only currently possible
with soft X-ray SHG (SXSHG). No other techniques can both penetrate to the buried
interface and offer the high sensitivity demonstrated in the SXSHG data.
Using the SACLA setup, several more measurements have been performed that
take advantage of operating below the damage threshold. Lithium osmate is a so-
called polar metal below a transition temperature [109], meaning it no longer has
centrosymmetry and SHG was easily observable. However, when the temperature
was raised and the crystal undergoes a phase transition into a centrosymmetric state,
no signal was observable. The SHG spectra were used to determine the lithium bond

Fig. 4.8 SXSHG spectra of boron/vacuum (B/V) and boron/polymer (B/P) interfaces. B/V (dark
red triangle) and B/P (dark blue circle) interfaces, shown along with the linear XAS of boron
(light red line) and boron/polymer (light blue line), with arrows showing the appropriate y axis. The
difference in XAS is believed to be due to differences in the background, whereas the SXSHG signal
is significantly different [108]. Reprinted with permission from Schwartz et al. [108]. Copyright
(2021) by the American Physical Society
4 Nonlinear Soft X-Ray Spectroscopy 109

distances in the crystal based upon comparison to calculations. While this informa-
tion can also be obtained from other techniques like neutron scattering, performing
this experiment spectroscopically allows for the possibility of studying the system
following an excitation laser pulse. It should be possible to watch the system ‘recover’
into the polar metal state following excitation, as the detailed way in which this tran-
sition occurs is not presently known. Such a measurement should now be possible
with SXSHG.
Another measurement was made on lithium niobate [110]. This is a standard
noncentrosymmetric crystal used in nonlinear optical measurements, for example as
an optical parametric oscillator [111]. In this case, the lithium and niobium peaks
overlap spectrally, so the resultant SHG spectrum was quite complicated. However,
with the aid of calculations even spectra with contributions coming from more than
one element can be disentangled. The most important result from this study is that, as
is seen in the visible, there is a polarization dependence to the signal at a given energy
[112]. This can be used to understand the symmetry of a crystal in the visible, but at
X-ray energies it reveals the symmetry around a given element. By measuring a wide
variety of different angles, it is possible to obtain detailed polarization analysis. It
additionally confirms that SXSHG is dipole selected. This method of using a polarizer
to detect SHG has potential in that it efficiently suppresses the background, but the
mirrors only function within a limited spectral range. This can be mitigated using
multiple sets of different mirrors. This of course increases the equipment cost, but
given the expense of FEL experiments, the expense could be considered minor.
More recent studies have moved beyond fundamental physics to probe chemistry.
A recent work demonstrated that in a solid-state battery system, a large spectral
shift was seen between the surface SHG measurements and the bulk absorption
measurements [113]. Because of the sensitivity of these measurements, the shift
could be assigned to the lithium having more freedom to move (fewer librational
constraints) at the surface than in the bulk. This is an important structural property for
battery designs and is the cause of the previously unassigned large surface resistance
seen in a variety of solid-state batteries. This implies a means of mitigating this
problem through careful battery design.
There has also been a successful demonstration of two NIR photons and one
XUV photon combining in a variation of SFG on LiF [9]. Working below damage
required using the three photon scheme so the signal would be bulk sensitive, as
the two-photon transition was likely too weak to be measured at the relatively low
fluences used here. The authors demonstrated both SFG (XUV + two NIR photons)
and difference frequency generation (DFG) (XUV − two NIR photons). They also
showed that the overlap between the NIR and XUV photon must be relatively precise,
or else the signal drops to zero as is expected. Interestingly, the observed line width
was sharper than the bandwidth of the FEL pulse, which suggest the spectral limit
is due to the excitonic LiF state probed. This has potentially use in the development
of high-resolution spectroscopy at FELs, although may not be particularly useful at
seeded FELs.
Finally, we should note there has been a SXSHG measurement made without an
FEL [114]. Using a large laser system with a power of ~ 1011 W/cm2 , which is roughly
110 C. P. Schwartz and W. S. Drisdell

5 orders less power density then typical from an FEL, SXSHG was successfully
demonstrated in lab at the titanium M-edge. This opens the possibility of performing
these types of measurements in labs going forward and hopefully one day will negate
the need for a national facility such as an FEL. This particular system, however, is
substantially larger than an HHG source and lacks the tunability of a FEL. However,
there has been large progress in the development of lab-based X-ray sources.
Thus, SXSHG represents a technique with great potential for studying surfaces and
buried interfaces, dynamics of noncentrosymmetric materials, or dynamic symmetry
breaking in centrosymmetric materials. It is maturing as new detection schemes
become available and a variety of methods to understand the resultant spectra have
been developed. New FELs such as LCLS-II and developments in lab-based light
sources may make this technique a standard technique for surface studies in the
future.

4.5 Theoretical Calculations for the Spectral Analysis

It was only shortly after the laser was invented that the first nonlinear optics measure-
ments were made [3]. This was possible because theoretical predictions made well in
advance of the development of the laser had predicted the phenomena, and to some
extent even how to perform such measurements. This has been true to some extent
also in the field of X-ray nonlinear optics, with measurements such as hard X-ray
SHG having been guided by earlier theoretical work [39]. The guiding theory has
taken on a variety of forms, and the most widely used conceptual frameworks are
discussed further here.
Since nonlinear optics are well established in the visible, theoretical frameworks
used in the visible can be applied in the soft X-ray regime in certain cases as both
are dipole selected and have similar selection rules. Perhaps the clearest example of
this was the study of LiF by four wave mixing [9], where the measured single photon
X-ray absorption data was combined with reflectivity and known nonlinear optical
constants to predict the sum and difference frequency generation spectra [5]. This
type of technique, where well-established laws of physics in the visible can be applied
to a problem in the soft X-ray regime, is quite powerful when adequate information
about a system is known. But well-established values and equations must exist, and
this is not always, or even generally, the case.
Because of limited information in most cases in X-ray nonlinear optics, such an
approach is not generally applicable. More general approaches such as those that
propagate motions of atoms, electrons, and fields are available, including techniques
such as velocity gauge real-time time-dependent density functional theory (VG-
RTTDDFT) [115–117]. This method is, like many used for calculations involving
soft X-ray nonlinear optics, incredibly expensive, but it gives accurate results and can
describe a wide variety of phenomena, including second harmonic generation, sum
frequency generation and multiphoton absorption and can potentially be extended
to simulate other measurements. This method works by propagating the fields, the
4 Nonlinear Soft X-Ray Spectroscopy 111

motions of the electrons, and the motions of the atoms in a real-time formalism.
This means in principle it is possible to describe the electronic structure and the
resulting change in electronic structure as a function of time as fields interact with
the system. There is another approach, based on the first principles calculation that
uses the Korringa–Kohn–Rostoker (KKR) method. The calculation can be made by
the software, such as AkaiKKR (Machikaneyama) [118, 119]. In the method, the
second order susceptibility for SHG is expressed with the Green’s functions and the
dipole matrix element that are calculated by the density functional calculation. The
calculation successfully reproduced result of the soft X-ray SHG experiment [7]. The
technique is highly feasible and it can easily be applied to the non-uniform system,
such as impurity or superlattice. Such a technique is inherently powerful, but the
computational expense means it requires large amounts of supercomputing time and
is limited to relatively simple systems.
For this reason, there has been significant effort put into extending cheaper
methods that exist in the visible but did not yet exist in the X-ray region, partic-
ularly using linear response [6, 120, 121]. While linear response techniques are
almost universally cheaper than RTTDDFT, they come with their own costs. First,
the formalism must be built to solve a relatively specific problem or class of prob-
lems. Secondly, the formalism must be tested to ensure that it works and provides
reasonably accurate results. Since nonlinear X-ray optics is a new field, there is often
no data to test against, and the ability to perform tests at high fields is limited by FEL
time which is difficult to obtain [36]. This is particularly true if the experiment is seen
as risky, as may happen if it is based on highly original and possibly controversial
theoretical work. There is at present no easy solution to this other than for more free
electron lasers and more supercomputers to be built, but this is an expensive solution
and one that will likely not be readily championed by governments around the world.
Still, some interesting and powerful methods have been developed. Perhaps some
of the nicest examples of this have been in the variety of methods used to describe
advanced X-ray Raman techniques [23, 85, 86, 122, 123]. These have proven very
useful in the development of novel and X-ray specific forms of Raman spectroscopy,
as the specific measurements appear to have been simulated in detail before they
were initially attempted, leading ultimately to the initial demonstration of stimulated
Raman. However, several of the suggested forms of Raman either have not been
made to work yet, or may prove ultimately impossible to perform due to some yet
unknown issue with the calculations. Still, attempting to develop theories which can
be tested is an important aspect of development of nonlinear X-ray spectroscopy and
should be encouraged.
Going beyond this, several authors, among them Mukamel and Santra, have been
engaged in forward looking theory, supporting experiments that may be feasible in
10, 20, or 50 years but are not feasible currently and do not look to be feasible in
the immediate future [124–128]. This can serve the purpose of inspiring long-term
experiments and goals for large national and international facilities. It also promotes
thinking about dream experiments and the ultimate limit of what could be achieved
with X-ray nonlinear optics.
112 C. P. Schwartz and W. S. Drisdell

However, one opportunity for real advancement in X-ray spectroscopy in general,

and in nonlinear optics in particular is the development of predictive actionable
theories. While there is certainly value in predicting experiments with four rotatable,
phase and mode locked attosecond, infinitely tunable Fourier-limited FEL pulses, a
real opportunity for growth where new theorists could enter is in performing calcu-
lations to guide and inform currently feasible experiments. Even for problems where
there exists adequate theory to calculate spectra, the methods are often too expensive
to be performed in advance of the experiment. For instance, performing VG-RTTDFT
is quite expensive, and therefore impractical for performing calculations of spectra
in advance of an experiment, when it is still not known whether the measurement
will work. In practice, the theorists who have worked on SHG as of the time of
this writing have always only offered to calculate results after the experiment was
completed. Thus, there certainly would be a welcome niche for potentially less accu-
rate, but significantly cheaper and therefore more predictive calculations, because
they can be used to guide experiments. Given the incredible expense of FEL time
(approaching $10,000 an hour [129]), such an investment in theory might also make
sense financially even if it requires a relatively large investment in supercomputing
time.

4.6 Summary

The development of high intensity X-ray sources has opened the door into a whole
new field of X-ray physics. In the soft X-ray region, nonlinear optics has been shown
to be quite similar conceptually to work that is already present in the visible with the
added benefit of shorter wavelengths and atomic specificity. Rapid progress has been
demonstrated on a wide variety of fronts, ranging from multiphoton absorption to
wave mixing. As progress on both sources and experimental setups continue and are
in their early stages, we expect to see rapid progress. Theoretical efforts in certain
cases have guided many of the existing experimental methods and hopefully will be
even more important in the future.
The early work laying the foundations of a variety of soft X-ray nonlinear optics
measurements has now largely been done, and now the community must begin the
process of making more useful and less proof of principle measurements. The ability
to do this will depend on a variety of experimental and theoretical developments. Of
particular benefit to the experimentalists would be phase locked coherent attosecond
lasers with high pulse rates and high repetition rates, beam splitters, flexible mirrors
and generally anything which makes soft X-rays more flexible to work with. This
will enable measurements which are not just analogous to the measurements made
in the visible but can push beyond that, exploiting X-ray photons sensitivity to core
excitations and the ability to control electronic structure in a novel manner.

Acknowledgements We thank Hisazumi Akai for his comment on theoretical calculation of

nonlinear spectral analysis. We appreciate Tomoaki Senoo, XingYu Su, Toshihide Sumi and Iwao
4 Nonlinear Soft X-Ray Spectroscopy 113

Matsuda for editing and revising the chapter. This work was supported by the U.S. Department of
Energy, Office of Science, Office of Basic Energy Sciences under Award Number DE-SC-0023397.

References

1. J.M. Hollas, Modern Spectroscopy, 4th edn. (Wiley, Chichester, Hoboken, NJ, 2004)
2. W. Kaiser, C.G.B. Garrett, Two-photon excitation in CaF2 : Eu2+ . Phys. Rev. Lett. 7, 229
(1961)
3. P.A. Franken, A.E. Hill, C.W. Peters, G. Weinreich, Generation of optical harmonics. Phys.
Rev. Lett. 7, 118 (1961)
4. U. Keller, Ultrafast Lasers: A Comprehensive Introduction to Fundamental Principles with
Practical Applications (Springer, Cham, 2021)
5. R.W. Boyd, Nonlinear Optics, 3rd edn. (Academic Press, Amsterdam, Boston, 2008)
6. R.K. Lam et al., Soft X-ray second harmonic generation as an interfacial probe. Phys. Rev.
Lett. 120, 023901 (2018)
7. S. Yamamoto et al., Element selectivity in second-harmonic generation of GaFeO3 by a
soft-X-ray free-electron laser. Phys. Rev. Lett. 120, 223902 (2018)
8. B. Weigelin, G.-J. Bakker, P. Friedl, Third harmonic generation microscopy of cells and tissue
organization. J. Cell Sci. (2016). https://doi.org/10.1242/jcs.152272
9. H. Rottke et al., Probing electron and hole colocalization by resonant four-wave mixing
spectroscopy in the extreme ultraviolet. Sci. Adv. 8, eabn5127 (2022)
10. E. Shwartz, S. Shwartz, Difference-frequency generation of optical radiation from two-color
X-ray pulses. Opt. Express 23, 7471 (2015)
11. T. Helk, M. Zürch, C. Spielmann, Perspective: towards single shot time-resolved microscopy
using short wavelength table-top light sources. Struct. Dyn. 6, 010902 (2019)
12. Y. Fu, K. Nishimura, R. Shao, A. Suda, K. Midorikawa, P. Lan, E.J. Takahashi, High effi-
ciency ultrafast water-window harmonic generation for single-shot soft X-ray spectroscopy.
Commun. Phys. 3, 1 (2020)
13. A. Schori, C. Bömer, D. Borodin, S.P. Collins, B. Detlefs, M. Moretti Sala, S. Yudovich, S.
Shwartz, Parametric down-conversion of X rays into the optical regime. Phys. Rev. Lett. 119,
253902 (2017)
14. B. Adams, Y. Nishino, D.V. Novikov, G. Materlik, D.M. Mills, Parametric down conversion
of X-rays, recent experiments. Nucl. Instrum. Methods Phys. Res. Sect. A 467–468, 1019
(2001)
15. F. Bencivenga et al., Four-wave mixing experiments with extreme ultraviolet transient
gratings. Nature 520, 7546 (2015)
16. F. Bencivenga et al., Nanoscale transient gratings excited and probed by extreme ultraviolet
femtosecond pulses. Sci. Adv. 5, eaaw5805 (2019)
17. G. Doumy et al., Nonlinear atomic response to intense ultrashort X rays. Phys. Rev. Lett. 106,
083002 (2011)
18. H. Hasegawa, E.J. Takahashi, Y. Nabekawa, K.L. Ishikawa, K. Midorikawa, Multiphoton
ionization of He by using intense high-order harmonics in the soft-X-ray region. Phys. Rev.
A 71, 023407 (2005)
19. M. Beye, S. Schreck, F. Sorgenfrei, C. Trabant, N. Pontius, C. Schüßler-Langeheine, W. Wurth,
A. Föhlisch, Stimulated X-ray emission for materials science. Nature 501, 7466 (2013)
20. D.J. Higley et al., Femtosecond X-ray induced changes of the electronic and magnetic response
of solids from electron redistribution. Nat. Commun. 10, 1 (2019)
21. L. Mercadier et al., Evidence of extreme ultraviolet superfluorescence in xenon. Phys. Rev.
Lett. 123, 023201 (2019)
114 C. P. Schwartz and W. S. Drisdell

22. J.R. Harries et al., Superfluorescence, free-induction decay, and four-wave mixing: propaga-
tion of free-electron laser pulses through a dense sample of helium ions. Phys. Rev. Lett. 121,
263201 (2018)
23. C. Weninger, M. Purvis, D. Ryan, R.A. London, J.D. Bozek, C. Bostedt, A. Graf, G. Brown,
J.J. Rocca, N. Rohringer, Stimulated electronic X-ray Raman scattering. Phys. Rev. Lett. 111,
233902 (2013)
24. U. Eichmann et al., Photon-recoil imaging: expanding the view of nonlinear X-ray physics.
Science 369, 1630 (2020)
25. J. Chen, C.-L. Hu, F. Kong, J.-G. Mao, High-performance second-harmonic-generation (SHG)
materials: new developments and new strategies. Acc. Chem. Res. 54, 2775 (2021)
26. F.J. Duarte (ed.), Tunable Laser Applications, 3rd edn. (CRC Press, Taylor & Francis Group,
Boca Raton, FL, 2016)
27. Y.R. Shen, The Principles of Nonlinear Optics, Wiley classics library edn. (Wiley-Interscience,
Hoboken, NJ, 2003)
28. J.A. Giordmaine, R.C. Miller, Tunable coherent parametric oscillation in LiNbO3 at optical
frequencies. Phys. Rev. Lett. 14, 973 (1965)
29. S.A. Akhmanov, A.I. Kovrigin, A.S. Piskarskas, V.V. Fadeev, R.V. Khokhlov, Observation of
parametric amplification in the optical range. Sov. J. Exp. Theor. Phys. Lett. 2, 191 (1965)
30. L. Carrion, J.-P. Girardeau-Montaut, Development of a simple model for optical parametric
generation. J. Opt. Soc. Am. B JOSAB 17, 78 (2000)
31. F. Hofmann, M.P. Short, C.A. Dennett, Transient grating spectroscopy: an ultrarapid,
nondestructive materials evaluation technique. MRS Bull. 44, 392 (2019)
32. E. Poem, A. Golenchenko, O. Davidson, O. Arenfrid, R. Finkelstein, O. Firstenberg, Pulsed-
pump phosphorus-doped fiber Raman amplifier around 1260 nm for applications in quantum
non-linear optics. Opt. Express OE 28, 32738 (2020)
33. P. Weinberger, John Kerr and his effects found in 1877 and 1878. Philos. Mag. Lett. 88, 897
(2008)
34. M. Protopapas, C.H. Keitel, P.L. Knight, Atomic physics with super-high intensity lasers.
Rep. Prog. Phys. 60, 389 (1997)
35. C.A. Hoffman, R.G. Driggers, Encyclopedia of Optical and Photonic Engineering, 2nd edn.
(CRC Press, Boca Raton, FL, 2016)
36. C. Bostedt, S. Boutet, D.M. Fritz, Z. Huang, H.J. Lee, H.T. Lemke, A. Robert, W.F. Schlotter,
J.J. Turner, G.J. Williams, Linac coherent light source: the first five years. Rev. Mod. Phys.
88, 015007 (2016)
37. M. Fuchs et al., Anomalous nonlinear X-ray Compton scattering. Nat. Phys. 11, 11 (2015)
38. T.E. Glover et al., X-ray and optical wave mixing. Nature 488, 7413 (2012)
39. S. Shwartz et al., X-ray second harmonic generation. Phys. Rev. Lett. 112, 163901 (2014)
40. V. Honkimäki, J. Sleight, P. Suortti, Characteristic X-ray flux from sealed Cr, Cu, Mo, Ag and
W tubes. J. Appl. Cryst. 23, 5 (1990)
41. V. Veksler, Concerning some new methods of acceleration of relativistic particles. Phys. Rev.
69, 244 (1946)
42. E.M. McMillan, The synchrotron—a proposed high energy particle accelerator. Phys. Rev.
68, 143 (1945)
43. F.R. Elder, A.M. Gurewitsch, R.V. Langmuir, H.C. Pollock, Radiation from electrons in a
synchrotron. Phys. Rev. 71, 829 (1947)
44. J. Als-Nielsen, D. McMorrow, Elements of Modern X-Ray Physics, 2nd edn. (Wiley, Hoboken,
2011)
45. S. Boutet, M. Yabashi, X-ray free electron lasers and their applications, in X-Ray Free Electron
Lasers: A Revolution in Structural Biology, ed. by S. Boutet, P. Fromme, M.S. Hunter (Springer
International Publishing, Cham, 2018), pp. 1–21
46. G. Margaritondo, P. Rebernik Ribic, A simplified description of X-ray free-electron lasers. J.
Synchrotron Radiat. 18, 2 (2011)
47. J. Duris et al., Tunable isolated attosecond X-ray pulses with gigawatt peak power from a
free-electron laser. Nat. Photon. 14, 1 (2020)
4 Nonlinear Soft X-Ray Spectroscopy 115

48. E. Allaria et al., Two-stage seeded soft-X-ray free-electron laser. Nat. Photon. 7, 11 (2013)
49. I. Nam et al., High-brightness self-seeded X-ray free-electron laser covering the 3.5 keV to
14.6 keV range. Nat. Photon. 15, 6 (2021)
50. I. Inoue et al., Generation of narrow-band X-ray free-electron laser via reflection self-seeding.
Nat. Photon. 13, 5 (2019)
51. E. Allaria et al., The FERMI free-electron lasers. J. Synchrotron Radiat. 22, 3 (2015)
52. G. Penco et al., Nonlinear harmonics of a seeded free-electron laser as a coherent and ultrafast
probe to investigate matter at the water window and beyond. Phys. Rev. A 105, 053524 (2022)
53. B.W. Adams, D. DeCiccio, M. Michon, P. Chmielniak, T. Parkman, G. Diebold, K. Bisogno,
C. Rose-Petruck, LPXS: a high-performance, recirculating liquid-metal laser-driven plasma
X-ray source, in Advances in X-Ray/EUV Optics and Components XVI, vol. 11837 (SPIE,
2021), pp. 59–69
54. T. Saule et al., High-flux ultrafast extreme-ultraviolet photoemission spectroscopy at 18.4
MHz pulse repetition rate. Nat. Commun. 10, 1 (2019)
55. J. Stöhr, NEXAFS Spectroscopy (Springer, Berlin, London, 2011)
56. G. Gauglitz, D.S. Moore (eds.), Handbook of Spectroscopy, 2nd, completely revised and
enlarged edn. (Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2014)
57. J.D. Smith, C.D. Cappa, K.R. Wilson, B.M. Messer, R.C. Cohen, R.J. Saykally, Energetics
of hydrogen bond network rearrangements in liquid water. Science 306, 851 (2004)
58. U. Bergmann, P. Glatzel, X-ray emission spectroscopy. Photosynth. Res. 102, 255 (2009)
59. G. Smolentsev, A.V. Soldatov, J. Messinger, K. Merz, T. Weyhermüller, U. Bergmann, Y.
Pushkar, J. Yano, V.K. Yachandra, P. Glatzel, X-ray emission spectroscopy to study ligand
valence orbitals in Mn coordination complexes. J. Am. Chem. Soc. 131, 13161 (2009)
60. F. de Groot, A. Kotani, Core Level Spectroscopy of Solids (CRC Press, Boca Raton, 2008)
61. S. Hüfner, Photoelectron Spectroscopy: Principles and Applications (Springer Berlin Heidel-
berg, Berlin, Heidelberg, 2013)
62. L. Young et al., Femtosecond electronic response of atoms to ultra-intense X-rays. Nature
466, 7302 (2010)
63. M. Hoener et al., Ultraintense X-ray induced ionization, dissociation, and frustrated absorption
in molecular nitrogen. Phys. Rev. Lett. 104, 253002 (2010)
64. H. Yoneda, Y. Inubushi, M. Yabashi, T. Katayama, T. Ishikawa, H. Ohashi, H. Yumoto, K.
Yamauchi, H. Mimura, H. Kitamura, Saturable absorption of intense hard X-rays in iron. Nat.
Commun. 5, 1 (2014)
65. L. Hoffmann et al., Saturable absorption of free-electron laser radiation by graphite near the
carbon K-edge, arXiv:2112.12585 [Cond-Mat, Physics:Physics] (2022)
66. B. Nagler et al., Turning solid aluminium transparent by intense soft X-ray photoionization.
Nat. Phys. 5, 9 (2009)
67. H. Yoneda et al., Ultra-fast switching of light by absorption saturation in vacuum ultra-violet
region. Opt. Express OE 17, 23443 (2009)
68. M. Pawlicki, H.A. Collins, R.G. Denning, H.L. Anderson, Two-photon absorption and the
design of two-photon dyes. Angew. Chem. Int. Ed. 48, 3244 (2009)
69. C.N. LaFratta, J.T. Fourkas, T. Baldacchini, R.A. Farrer, Multiphoton fabrication. Angew.
Chem. Int. Ed. 46, 6238 (2007)
70. A. Hayat, A. Nevet, P. Ginzburg, M. Orenstein, Applications of two-photon processes in
semiconductor photonic devices: invited review. Semicond. Sci. Technol. 26, 083001 (2011)
71. M. Rumi, J.W. Perry, Two-photon absorption: an overview of measurements and principles.
Adv. Opt. Photon. 2, 451 (2010)
72. J.H. Eberly, P. Lambropoulos (eds.), Multiphoton Processes: Proceedings of an International
Conference at the University of Rochester, Rochester, N.Y., June 6–9, 1977 (Wiley, New York,
1978)
73. Y. Nabekawa, H. Hasegawa, E.J. Takahashi, K. Midorikawa, Production of doubly charged
helium ions by two-photon absorption of an intense sub-10-fs soft X-ray pulse at 42 eV photon
energy. Phys. Rev. Lett. 94, 043001 (2005)
116 C. P. Schwartz and W. S. Drisdell

74. N. Miyamoto, M. Kamei, D. Yoshitomi, T. Kanai, T. Sekikawa, T. Nakajima, S. Watanabe,

Observation of two-photon above-threshold ionization of rare gases by XUV harmonic
photons. Phys. Rev. Lett. 93, 083903 (2004)
75. R.K. Lam et al., Two-photon absorption of soft X-ray free electron laser radiation by graphite
near the carbon K-absorption edge. Chem. Phys. Lett. 703, 112 (2018)
76. N. Rohringer et al., Atomic inner-shell X-ray laser at 1.46 nanometres pumped by an X-ray
free-electron laser. Nature 481, 488 (2012)
77. T. Kroll et al., Observation of seeded Mn Kβ stimulated X-ray emission using two-color X-ray
free-electron laser pulses. Phys. Rev. Lett. 125, 037404 (2020)
78. T. Kroll et al., Stimulated X-ray emission spectroscopy in transition metal complexes. Phys.
Rev. Lett. 120, 133203 (2018)
79. Y. Zhang et al., Generation of intense phase-stable femtosecond hard X-ray pulse pairs. Proc.
Natl. Acad. Sci. 119, e2119616119 (2022)
80. A.C. Thompson, X-Ray Data Booklet (Lawrence Berkeley National Laboratory, University
of California, 2001)
81. D.J. Higley et al., Stimulated resonant inelastic X-ray scattering in a solid. Commun. Phys.
5, 1 (2022)
82. J. Stöhr, A. Scherz, Creation of X-ray transparency of matter by stimulated elastic forward
scattering. Phys. Rev. Lett. 115, 107402 (2015)
83. B. Wu et al., Elimination of X-ray diffraction through stimulated X-ray transmission. Phys.
Rev. Lett. 117, 027401 (2016)
84. Z. Chen et al., Ultrafast self-induced X-ray transparency and loss of magnetic diffraction.
Phys. Rev. Lett. 121, 137403 (2018)
85. N. Rohringer, X-ray Raman scattering: a building block for nonlinear spectroscopy. Philos.
Trans. R. Soc. A Math. Phys. Eng. Sci. 377, 20170471 (2019)
86. V. Kimberg et al., Stimulated X-ray Raman scattering—a critical assessment of the building
block of nonlinear X-ray spectroscopy. Faraday Discuss. 194, 305 (2016)
87. J.T. O’Neal et al., Electronic population transfer via impulsive stimulated X-ray Raman
scattering with attosecond soft-X-ray pulses. Phys. Rev. Lett. 125, 073203 (2020)
88. C. Masciovecchio et al., EIS: the scattering beamline at FERMI. J. Synchrotron Radiat. 22, 3
(2015)
89. M.P. Short, C.A. Dennett, S.E. Ferry, Y. Yang, V.K. Mishra, J.K. Eliason, A. Vega-Flick, A.A.
Maznev, K.A. Nelson, Applications of transient grating spectroscopy to radiation materials
science. JOM 67, 1840 (2015)
90. J. Janušonis, T. Jansma, C.L. Chang, Q. Liu, A. Gatilova, A.M. Lomonosov, V. Shalagatskyi,
T. Pezeril, V.V. Temnov, R.I. Tobey, Transient grating spectroscopy in magnetic thin films:
simultaneous detection of elastic and magnetic dynamics. Sci. Rep. 6, 1 (2016)
91. A.A. Maznev et al., Propagation of THz acoustic wave packets in GaN at room temperature.
Appl. Phys. Lett. 112, 061903 (2018)
92. L. Foglia et al., First evidence of purely extreme-ultraviolet four-wave mixing. Phys. Rev.
Lett. 120, 263901 (2018)
93. F. Bencivenga et al., Four-wave-mixing experiments with seeded free electron lasers. Faraday
Discuss. 194, 283 (2016)
94. R. Bohinc et al., Nonlinear XUV-optical transient grating spectroscopy at the Si L2,3–edge.
Appl. Phys. Lett. 114, 181101 (2019)
95. A.A. Maznev et al., Generation of coherent phonons by coherent extreme ultraviolet radiation
in a transient grating experiment. Appl. Phys. Lett. 113, 221905 (2018)
96. D. Naumenko et al., Thermoelasticity of nanoscale silicon carbide membranes excited by
extreme ultraviolet transient gratings: implications for mechanical and thermal management.
ACS Appl. Nano Mater. 2, 5132 (2019)
97. W. Cao, E.R. Warrick, A. Fidler, S.R. Leone, D.M. Neumark, Near-resonant four-wave
mixing of attosecond extreme-ultraviolet pulses with near-infrared pulses in neon: detection
of electronic coherences. Phys. Rev. A 94, 021802 (2016)
4 Nonlinear Soft X-Ray Spectroscopy 117

98. T. Ding et al., Time-resolved four-wave-mixing spectroscopy for inner-valence transitions.

Opt. Lett. 41, 709 (2016)
99. K.A. Tran, K.B. Dinh, P. Hannaford, L.V. Dao, Phase-matched nonlinear wave-mixing
processes in XUV region with multicolor lasers. Appl. Opt. 58, 2540 (2019)
100. W. Cao, E.R. Warrick, A. Fidler, D.M. Neumark, S.R. Leone, Noncollinear wave mixing of
attosecond XUV and few-cycle optical laser pulses in gas-phase atoms: toward multidimen-
sional spectroscopy involving XUV excitations. Phys. Rev. A 94, 053846 (2016)
101. A.P. Fidler, E.R. Warrick, H.J.B. Marroux, E. Bloch, D.M. Neumark, S.R. Leone, Self-
heterodyned detection of dressed state coherences in helium by noncollinear extreme
ultraviolet wave mixing with attosecond pulses. J. Phys. Photon. 2, 034003 (2020)
102. A.P. Fidler, H.J.B. Marroux, E.R. Warrick, E. Bloch, W. Cao, S.R. Leone, D.M. Neumark,
Autoionization dynamics of (2P1/2)ns/d states in krypton probed by noncollinear wave mixing
with attosecond extreme ultraviolet and few-cycle near infrared pulses. J. Chem. Phys. 151,
114305 (2019)
103. A.P. Fidler, S.J. Camp, E.R. Warrick, E. Bloch, H.J.B. Marroux, D.M. Neumark, K.J. Schafer,
M.B. Gaarde, S.R. Leone, Nonlinear XUV signal generation probed by transient grating
spectroscopy with attosecond pulses. Nat. Commun. 10, 1384 (2019)
104. L. Van Dao, K.A. Tran, P. Hannaford, Cascaded four-wave mixing in the XUV region. Opt.
Lett. 43, 134 (2018)
105. K.B. Dinh, K.A. Tran, P. Hannaford, L.V. Dao, Four-wave mixing of extreme ultraviolet pulses
and infrared pulses for studies of atomic dynamics. J. Opt. Soc. Am. B 36, 3046 (2019)
106. L. Drescher, O. Kornilov, T. Witting, V. Shokeen, M.J.J. Vrakking, B. Schütte, Extreme-
ultraviolet spectral compression by four-wave mixing. Nat. Photon. 15, 263 (2021)
107. J.R. Rouxel et al., Hard X-ray transient grating spectroscopy on bismuth germanate. Nat.
Photon. 15, 7 (2021)
108. C.P. Schwartz et al., Angstrom-resolved interfacial structure in buried organic-inorganic
junctions. Phys. Rev. Lett. 127, 096801 (2021)
109. E. Berger et al., Extreme ultraviolet second harmonic generation spectroscopy in a polar metal.
Nano Lett. 21, 6095 (2021)
110. C.B. Uzundal et al., Polarization-resolved extreme-ultraviolet second-harmonic generation
from LiNbO3 . Phys. Rev. Lett. 127, 237402 (2021)
111. J. Lu, A.A. Sayem, Z. Gong, J.B. Surya, C.-L. Zou, H.X. Tang, Ultralow-threshold thin-film
lithium niobate optical parametric oscillator. Optica 8, 539 (2021)
112. M. Sliwa, K. Nakatani, T. Asahi, P.G. Lacroix, R.B. Pansu, H. Masuhara, Polarization and
wavelength dependent nonlinear optical properties of a photo-switchable organic crystal.
Chem. Phys. Lett. 437, 212 (2007)
113. C. Woodahl et al., Dynamics of interfacial lithium ions revealed by second harmonic extreme
ultraviolet spectroscopy. Nature (2022). Under review
114. T. Helk et al., Table-top extreme ultraviolet second harmonic generation. Sci. Adv. 7, eabe2265
(n.d.)
115. C.D. Pemmaraju, Valence and core excitons in solids from velocity-gauge real-time TDDFT
with range-separated hybrid functionals: an LCAO approach. Comput. Condens. Matter 18,
e00348 (2019)
116. C.D. Pemmaraju, Simulation of attosecond transient soft X-ray absorption in solids using
generalized Kohn–Sham real-time time-dependent density functional theory. New J. Phys.
22, 083063 (2020)
117. M. Noda et al., SALMON: scalable ab-initio light-matter simulator for optics and nanoscience.
Comput. Phys. Commun. 235, 356 (2019)
118. H. Akai, Electronic structure Ni–Pd alloys calculated by the self-consistent KKR-CPA
method. J. Phys. Soc. Jpn. 51, 468 (1982)
119. H. Akai, Fast Korringa–Kohn–Rostoker coherent potential approximation and its application
to FCC Ni-Fe systems. J. Phys. Condens. Matter 1, 8045 (1989)
120. S. Sharma, C. Ambrosch-Draxl, Second-harmonic optical response from first principles. Phys.
Scr. 2004, 128 (2004)
118 C. P. Schwartz and W. S. Drisdell

121. K.E. Dorfman, F. Schlawin, S. Mukamel, Nonlinear optical signals and spectroscopy with
quantum light. Rev. Mod. Phys. 88, 045008 (2016)
122. J.D. Biggs, Y. Zhang, D. Healion, S. Mukamel, Watching energy transfer in metalloporphyrin
heterodimers using stimulated X-ray Raman spectroscopy. Proc. Natl. Acad. Sci. 110, 15597
(2013)
123. S.M. Cavaletto, D. Keefer, S. Mukamel, High temporal and spectral resolution of stimulated
X-ray Raman signals with stochastic free-electron-laser pulses. Phys. Rev. X 11, 011029
(2021)
124. A. Karamatskou, R.E. Goetz, C.P. Koch, R. Santra, Suppression of hole decoherence in
ultrafast photoionization. Phys. Rev. A 101, 043405 (2020)
125. J.M. Schäfer, L. Inhester, S.-K. Son, R.F. Fink, R. Santra, Electron and fluorescence spectra
of a water molecule irradiated by an X-ray free-electron laser pulse. Phys. Rev. A 97, 053415
(2018)
126. J.R. Rouxel, M. Kowalewski, S. Mukamel, Diffraction-detected sum frequency generation:
novel ultrafast X-ray probe of molecular dynamics. J. Phys. Chem. Lett. 9, 3392 (2018)
127. G. Batignani, C. Sansone, C. Ferrante, G. Fumero, S. Mukamel, T. Scopigno, Excited-state
energy surfaces in molecules revealed by impulsive stimulated Raman excitation profiles. J.
Phys. Chem. Lett. 12, 9239 (2021)
128. S.M. Cavaletto, S. Mukamel, Probing delocalized current densities in selenophene by resonant
X-ray sum-frequency generation. J. Chem. Theory Comput. 17, 367 (2021)
129. Facts and Figures, https://www.xfel.eu/facility/overview/facts_amp_figures/index_eng.html
Chapter 5
Nonlinear X-Ray Spectroscopy

Yuya Kubota and Kenji Tamasaku

Abstract This chapter focuses on nonlinear spectroscopies in the hard X-ray region
(wavelength 0.1–1 nm). X-ray nonlinear phenomena differ from those in longer wave-
lengths because of unique X-ray interactions with matter. Theoretical aspect of X-
ray-matter interactions is briefly reviewed for calculation of nonlinear polarizability
together with some associated experiments. Then, nonlinear X-ray spectroscopy is
overviewed by focusing on experimental studies.

5.1 Introduction

In this chapter, we discuss nonlinear spectroscopy in the hard X-ray region (wave-
length 0.1–1 nm) and related topics. Studies of X-ray nonlinear optics have been
developed gradually using synchrotron radiation, but the advent of high-brilliance
X-ray free-electron lasers (XFELs) has brought a breakthrough in this field, and been
leading to nonlinear X-ray spectroscopy. This trend has been supported by focus-
ing systems, which can realize unprecedented intensity required for the nonlinear
processes. First of all, we briefly introduce XFEL and the focusing systems.
A free-electron laser (FEL) relies on interaction between electromagnetic wave
and relativistic electron beam in alternating magnetic field created by an undulator.
The interaction imprints a density modulation with a period of the wavelength on
the electron beam, which finally emits electromagnetic wave in phase. Because the
interaction is very weak, the FEL requires a cavity, which considerably elongates the
effective interaction length. It is lack of X-ray cavities that had prevented researchers
from realizing XFEL. This difficulty was cleared by the self-amplified spontaneous
emission (SASE) principle [1, 2], where the interaction length was increased by
using a very long undulator. A drawback of the SASE principle is that the pulse
characteristics, such as pulse energy, wavelength, and spectral shape, fluctuate shot

Y. Kubota (B) · K. Tamasaku (B)

RIKEN SPring-8 Center, Hyogo, Japan
e-mail: kubota@spring8.or.jp
K. Tamasaku
e-mail: tamasaku@spring8.or.jp
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 119
I. Matsuda and R. Arafune (eds.), Nonlinear X-Ray Spectroscopy for Materials Science,
Springer Series in Optical Sciences 246, https://doi.org/10.1007/978-981-99-6714-8_5
120 Y. Kubota and K. Tamasaku

by shot, because the lasing process starts from shot noise. The instabilities of SASE-
XFEL can be improved by seeding [3, 4]. In this scheme, the undulator is split into
two parts. A monochromator set between them selects a single wavelength from the
SASE radiation of the upstream undulator. Then, the monochromatic beam is led
to the downstream undulator, stabilizing the wavelength and the lasing process. The
split undulators can also be used to produce two-color XFEL by operating them at
different wavelengths [5].
There are currently five XFEL facilities in operation worldwide, the Linac Coher-
ent Light Source (LCLS) in the USA [6], the SPring-8 Angstrom Compact free-
electron LAser (SACLA) in Japan [7], the hard X-ray free-electron laser at the
Pohang Accelerator Laboratory (PAL-XFEL) in the Republic of Korea [8], swissFEL
in Switzerland [9], and the European X-Ray Free-Electron Laser Facility (European
XFEL) in Germany [10] in chronological order of the first lasing. Figure 5.1 shows
some pictures of SACLA. The distinctive features of XFEL compared to other X-ray
sources are high brightness, ultrashort pulse duration, and high transverse coher-
ence, which have opened up new frontiers in science. For example, formation and
dissociation of chemical bonds after optical laser excitation were revealed by tak-
ing advantage of the ultrashort pulse duration on a femtosecond timescale [11–13].
In solid-state physics, ultrafast phenomena in photoexcited quantum materials have
also been extensively studied by using the femtosecond pulse duration [14–16].
The highly brilliant XFEL pulses, which enable data acquisition in a single shot,
have brought opportunities to perform experiments under instantaneous extreme
conditions such as strong magnetic fields [17–19] and high pressure created by a
high-power optical laser pulse [20–22]. One of the breakthroughs by XFEL may
be a so-called diffract-before-destroy scheme [23]. X-rays have been a powerful
tool for crystallography, but radicals and reactants produced during X-ray irradiation
inevitably damage radiation-sensitive biological specimens. The intrinsic structures
before damage can be obtained from the diffraction pattern taken by a single expo-
sure with the high-brilliant and ultrashort XFEL pulse. This scheme has contributed,
for example, to investigation of the oxygen-evolving complex of photosystem II [24,
25]. Single-shot coherent diffraction imaging (CDI) experiments can be performed
by combining the high transverse coherence. Similar to the diffract-before-destroy
scheme, single-shot CDI is applicable to damage-free observation of non-periodic
and nano-sized particles, such as living cells [26].
In the study of X-ray nonlinear optics, the peak intensity is of crucial importance.
Benefiting from the high brilliance of XFEL, a number of studies in X-ray nonlinear
optics have been reported. In these studies, Kirkpatrick-Baez (KB) mirrors [27]
have been used to obtain a high peak intensity. The KB mirror systems developed
at SACLA are summarized in Table 5.1. A unique one was the two-stage focusing
system, where the first KB mirror expanded the beam and the second focused it [28].
The elaborate optical design allowed a longer working distance for flexible sample
environment at a cost of complicated alignment procedure.
Hard X-rays have characteristics that the wavelength is close to the interatomic
distance of crystals and the photon energy is close to the binding energy of the inner-
shell electrons of metallic elements. Unique interactions between X-rays and matter
5 Nonlinear X-Ray Spectroscopy 121

Fig. 5.1 Pictures of a aerial view, b accelerator, and c undulator of SACLA. With permission of
RIKEN

Table 5.1 KB mirror systems developed at SACLA

System Beam size Peak intensity Working distance (m)
(.W/cm2 ).b
.1 µm
General-purpose KB .10
18 1.35
mirror [29]
Two-stage focusing 50 nm .10
20 0.35
system.a [28]
100-exa focusing 100–200 nm .10
20 0.115
system [30]
The beam size is in full width at half maximum (FWHM)
.
a Not available
b
. Estimated for a pink beam
122 Y. Kubota and K. Tamasaku

may produce interesting nonlinear phenomena possible only in the X-ray region.
In the following Sect. 5.2, we review a basic theory of nonlinear phenomena in the
hard X-ray region and present some associated studies. Then, in Sect. 5.3, we provide
highlights of notable studies of nonlinear X-ray spectroscopy. The last Sect. 5.4 gives
summary and future prospects.

5.2 The Basic Theory of Nonlinear Optics in the Hard

X-Ray Region

5.2.1 Nonlinear Polarizability

In general, the interaction between electromagnetic waves and materials can be

described using polarizability. We will derive the expression of the polarizability
by considering the current density induced by electric fields using elementary quan-
tum mechanics. This is because we cannot directly calculate the polarizability in the
X-ray region. Note that we use the Gaussian system of units in this chapter.
Electron density in a material at a certain time, .t, can be expressed by
∑
ρ(r, t) =
. δ(r − r l (t)), (5.1)
l

where . r l (t) is the position of the .l-th electron at .t. The current density is written as
∑
.J (r, t) = −e δ(r − r l (t))vl (t), (5.2)
l

where .vl (t) is the velocity of the .l-th electron at .t, and -e is the charge of the electron.
We calculate the Fourier transform of Eq. (5.2), and obtain
∫∫
. J̃ (K , ω) = J (r, t)ei(ωt−K ·r) drdt
∑
≃ e−iK ·r l j̃ l (ω). (5.3)
l

Here, . j̃ l (ω) is the Fourier transform of . j l (t) = −evl (t). We approximated . r l (t) by
r , because the time scale of electron motion is much longer than the oscillation
. l
period of X-rays, i.e., . r l (t) does not change while responding to X-rays.
The Hamiltonian for the electron system interacting with electromagnetic fields
is given by

∑[ 1 { e }2 ]
.H= pi + A(r i , t) + V (r i ) , (5.4)
i
2m c
5 Nonlinear X-Ray Spectroscopy 123

where . pi and . r i are the momentum and position operators for the .i-th electron,
respectively, m is the mass of the electron, c is the speed of light in vacuum, and .V
is the Coulomb potential of the nucleus. We ignored spin and interactions among
electrons. Below, we employ the Coulomb gauge, where .∇ · A = 0, and write the
vector potential:

1{ } c { }
A(r, t) =
. A(r, t) + A∗ (r, t) = ∈ E 0 ei(K ·r−ωt) − ∈ ∗ E 0∗ e−i(K ·r−ωt) ,
2 2iω
(5.5)
where .∈ is the polarization vector and . E 0 is the amplitude of the electric field.
We consider the perturbed system by . A, and divide Eq. (5.4) into the unperturbed
Hamiltonian and the perturbing one:

∑ { p2 }
.H0 = + V (r i ) ,
i
(5.6)
i
2m
∑{ e e2
}
'
.H = p · A(r i , t) + A (r i , t) .
2
(5.7)
i
mc i 2mc2

We used an identity, . p · A = A · p, in the Coulomb gauge.

Now, we calculate the current density of the perturbed system using the standard
perturbation theory. In this section, we will incorporate up to second-order terms. We
consider two X-ray fields with the wave vectors and the angular frequencies, . K 1 , .ω1 ,
and . K 2 , .ω2 , respectively, and calculate the current density, .J 3 , with . K 3 , .ω3 induced
by the nonlinear interaction between these waves and electrons. The perturbation
term Eq. (5.7) becomes

e e2
H' =
. p · {A1 (r, t) + A2 (r, t)} + {A1 (r, t) + A2 (r, t)}2 . (5.8)
mc 2mc2
We calculate the Fourier transform of the expectation value of the current density,
J = −(e/m) ( p + eA/c), to obtain the nonlinear polarizability:
.

∫ ( e)
J̃ 3 (K 3 , ω3 ) =
. ⟨Ψ(t)|e−iK 3 ·r̂ −
m
[ e{ }]
× p̂ + A1 (r̂, t) + A2 (r̂, t) |Ψ(t)⟩eiω3 t dt. (5.9)
c

Up to the second order, .|Ψ(t)⟩ is written as

∑{ }
δng + an(1, pA) (t) + an(1,A ) (t) + an(2, pA pA) (t) |n⟩e−iωn t ,
2
|Ψ(t)⟩ =
. (5.10)
n
124 Y. Kubota and K. Tamasaku

(1, pA)
and .an(1,A ) are coefficients for the first-order perturbed state by the
2
where .an
(2, pA pA)
. p · A and .A terms in Eq. (5.8), respectively, and .an
2
is for the second-order
perturbed state by the . p · A term.
In the rest of this section, we focus on a specific nonlinear process with .ω3 =
ω1 + ω2 , namely, sum-frequency generation (SFG). Below, we pick up typical matrix
elements in Eq. (5.9), which satisfy .ω3 = ω1 + ω2 .
( e)e
(1, p A−
.⟨Ψ (0) (t)|e−iK 3 ·r̂ − A−
2 (r̂, t)|Ψ
1 ) (t)⟩, (5.11)
( e)m c
− −
(0)
.⟨Ψ (t)|e−iK 3 ·r̂ − p̂|Ψ (1,A1 A2 ) (t)⟩, (5.12)
( m e) − −
(0) −iK 3 ·r̂
.⟨Ψ (t)|e − p̂|Ψ (2, p A1 p A2 ) (t)⟩, (5.13)
m
( e)
(1, p A+
1 ) (t)|e−iK 3 ·r̂
−
.⟨Ψ − p̂|Ψ (1, p A2 ) (t)⟩. (5.14)
m
Here the .± symbols on . A represent the sign of the angular frequency of the field,
e.g., . A−
2 for .−ω2 .
The relevant coefficients to Eqs. (5.11)–(5.14) are given by

(1, p A−
1)
ie ⟨n|∈ 1 · p̂eiK 1 ·r̂ |g⟩E 1 ei(ωng −ω1 )t
a
. n (t) = (5.15)
2mℏω1 ωng − ω1
(1,A− −
1 A2 )
e2 ∈ 1 · ∈ 2 ⟨n|ei(K 1 +K 2 )·r̂ |g⟩E 1 E 2 ei(ωng −ω1 −ω2 )t
a
. n (t) = , (5.16)
4mℏω1 ω2 ωng − ω1 − ω2
(2, p A− −
1 p A2 )
e2 E 1 E 2
a
. n (t) = −
4m 2 ℏ2 ω1 ω2
∑ ⟨n|∈ 2 · p̂eiK 2 ·r̂ |l⟩⟨l|∈ 1 · p̂eiK 1 ·r̂ |g⟩ ei(ωng −ω1 −ω2 )t
× . (5.17)
l
ωlg − ω1 ωng − ω1 − ω2

Here, .ℏ is the Planck constant divided by 2.π .

Finally, we obtain the nonlinear current density for .ω3 = ω1 + ω2 by calculating
Eq. (5.9) using Eqs. (5.10) and (5.15)–(5.17):

. J̃ 3 (K 3 , ω3 ) = −iπ ω3 β̃(K 3 , −K 1 , −K 2 , ω1 , ω2 ) : ∈ 1 ∈ 2 E 1 E 2 δ(ω3 − ω1 − ω2 ),

ie3
β̃(K 3 , −K 1 , −K 2 , ω1 , ω2 ) : ∈ 1 ∈ 2 = (U + B), (5.18)
m 3 ℏ2 ω1 ω2 ω3

where .β̃ is the Fourier transform of the second-order nonlinear polarizability.

) the contribution into two parts, .U and .B, where .U originates from
We( divide
.( p A) A -type perturbations and .B from .( p A) ( p A) ( p A)-type.
2
5 Nonlinear X-Ray Spectroscopy 125
{
∑ ⟨g|∈ 2 ei(K 2 −K 3 )·r̂ |n⟩⟨n|eiK 1 ·r̂ ∈ 1 · p̂|g⟩
.U = −mℏ
n
ωng − ω1
⟨g|eiK 1 ·r̂ ∈ 1 · p̂|n⟩⟨n|∈ 2 ei(K 2 −K 3 )·r̂ |g⟩
+
ωng + ω1
⟨g|∈ 1 ei(K 1 −K 3 )·r̂ |n⟩⟨n|eiK 2 ·r̂ ∈ 2 · p̂|g⟩
+
ωng − ω2
⟨g|eiK 2 ·r̂ ∈ 2 · p̂|n⟩⟨n|∈ 1 ei(K 1 −K 3 )·r̂ |g⟩
+
ωng + ω2
∈ 1 · ∈ 2 ⟨g|e−iK 3 ·r̂ p̂|n⟩⟨n|ei(K 1 +K 2 )·r̂ |g⟩
+
ωng − ω3
}
∈ 1 · ∈ 2 ⟨g|ei(K 1 +K 2 )·r̂ |n⟩⟨n|e−iK 3 ·r̂ p̂|g⟩
+ , (5.19)
ωng + ω3
{
∑ ⟨g|e−iK 3 ·r̂ p̂|n⟩⟨n|eiK 1 ·r̂ ∈ 1 · p̂|l⟩⟨l|eiK 2 ·r̂ ∈ 2 · p̂|g⟩
.B =
n,l
(ωng − ω3 )(ωlg − ω2 )
⟨g|e−iK 3 ·r̂ p̂|n⟩⟨n|eiK 2 ·r̂ ∈ 2 · p̂|l⟩⟨l|eiK 1 ·r̂ ∈ 1 · p̂|g⟩
+
(ωng − ω3 )(ωlg − ω1 )
⟨g|eiK 2 ·r̂ ∈ 2 · p̂|l⟩⟨l|eiK 1 ·r̂ ∈ 1 · p̂|n⟩⟨n|e−iK 3 ·r̂ p̂|g⟩
+
(ωng + ω3 )(ωlg + ω2 )
⟨g|eiK 1 ·r̂ ∈ 1 · p̂|l⟩⟨l|eiK 2 ·r̂ ∈ 2 · p̂|n⟩⟨n|e−iK 3 ·r̂ p̂|g⟩
+
(ωng + ω3 )(ωlg + ω1 )
⟨g|eiK 2 ·r̂ ∈ 2 · p̂|l⟩⟨l|e−iK 3 ·r̂ p̂|n⟩⟨n|eiK 1 ·r̂ ∈ 1 · p̂|g⟩
+
(ωng − ω1 )(ωlg + ω2 )
}
⟨g|eiK 1 ·r̂ ∈ 1 · p̂|n⟩⟨n|e−iK 3 ·r̂ p̂|l⟩⟨l|eiK 2 ·r̂ ∈ 2 · p̂|g⟩
+ . (5.20)
(ωng + ω1 )(ωlg − ω2 )

The second-order X-ray nonlinear process involves three waves. It is one of unique
features of X-ray nonlinear optics that one wave can have longer wavelength than
X-rays. However, we limit our discussion here to a special case, where all three waves
are X-rays. Other situation, where one wave is in the optical region, will be discussed
in Sect. 5.2.4. The three-X-rays case was first calculated by Freund and Levine [31],
and then classically by Eisenberger and McCall [32].
We evaluate the first term of .U [Eq. (5.19)],

∑ ⟨g|ei(K 2 −K 3 )·r̂ ∈ 2 |n⟩⟨n|eiK 1 ·r̂ ∈ 1 · p̂|g⟩

.U1 = −mℏ . (5.21)
n
ωng − ω1

We assume that the nonlinear crystal is composed of light elements, such as beryllium
(Be) and carbon. This assumption is not only for convenience in the calculation but
also valid for experiments, because the light elements are advantageous for absorp-
126 Y. Kubota and K. Tamasaku

tion. Considering .ωng « ω1 for the light elements, and ignoring the continuum state
in the vacuum, we obtain

mℏ ∑
U1 ≃
. ⟨g|ei(K 2 −K 3 )·r̂ ∈ 2 |n⟩⟨n|eiK 1 ·r̂ ∈ 1 · p̂|g⟩
ω1 n
mℏ
= ⟨g|e−iS·r̂ ∈ 2 (∈ 1 · p̂)|g⟩, (5.22)
ω1

where we defined . S = K 3 − K 1 − K 2 . We used the completeness of .|n⟩, i.e.,

∑
. n |n⟩⟨n| = 1. The second term of Eq. (5.19) can be calculated in the same manner,

∑ ⟨g|eiK 1 ·r̂ ∈ 1 · p̂|n⟩⟨n|∈ 2 ei(K 2 −K 3 )·r̂ |g⟩

U2 = − mℏ
.
n
ωng + ω1
mℏ
≃− ⟨g|∈ 2 (∈ 1 · p̂)e−iS·r̂ |g⟩. (5.23)
ω1

We can further calculate .U1 + U2 using .∈ 1 · K 1 = 0 and . p = −iℏ∇, and find that
we can relate it to a physical quantity:

mℏ
U1 + U2 =
. ∈ 2 ⟨g|[e−iS·r̂ , ∈ 1 · p̂]|g⟩
ω1
mℏ2
= ∈ 2 (∈ 1 · S)ρ̃(S). (5.24)
ω1

Here, .ρ̃ represents the Fourier transform of .ρ(r). Repeating similar calculations for
the remaining terms in Eq. (5.19), we finally obtain
{ }
1 1 1
U = −mℏ2 ρ̃(S)
. (∈ 1 · ∈ 2 )S − ∈ 2 (∈ 1 · S) − ∈ 1 (∈ 2 · S) . (5.25)
ω3 ω1 ω2

We find that .U is determined by the Fourier transform of .ρ when all three waves are
X-rays. We note that the X-ray linear polarizability is also determined by .ρ̃.
We discuss which mechanism, .U or .B, dominates .β̃. Because .U can be small
when . S is small, the contribution of .U is smaller than .B in the visible region.
However, it can be shown that the contribution of .B is negligible in the X-ray region.
We evaluate the first term of Eq. (5.20), .B1 . In the case of light elements, it can be
approximated as .eiK ·r̂ = 1 because .⟨K · r⟩ « 1. Then, .B1 can be approximated by

∑ ⟨g|r̂|n⟩⟨n|∈ 1 · r̂|l⟩⟨l|∈ 2 · r̂|g⟩

B1 = −
. im 3 ωgn ωnl ωlg . (5.26)
(ωng − ω3 )(ωlg − ω2 )

Regarding the size of dipoles as the Bohr radius, .a0 , we evaluate .B1 :

∑ a03
B1 ∼ −
. im 3 ωgn ωnl ωlg . (5.27)
ω3 ω2
5 Nonlinear X-Ray Spectroscopy 127

Major resonant frequencies are in the visible and ultraviolet range, which we set to
Ωuv . We also set the X-ray frequencies, .ω2 and .ω3 , to .ωx . Note that .Ωuv /ωx ∼ 10−4 .
.
The magnitude of .B1 is approximately given by

ℏ6 Ω3uv
.|B1 | ∼ . (5.28)
e6 ωx2

On the other hand, since .|S| is comparable to the wave vector of X-rays, .|U| can
be approximated by

. |U| ∼ mℏ2 ρ̃/c. (5.29)

We assume .ρ̃ ∼ 10 in crystals, and obtain,

| |
| B1 |
| ∼ 1 (ℏΩuv ) 1 ∼ 10−8 ,
3
| (5.30)
.
| U | α 3 mc2 (ℏω )2 ρ̃
x

where .α represents the fine structure constant. The above evaluation can be adapted
to other terms of .B. Finally, we can conclude that .U dominates .β̃ in the X-ray
region.
Since the . r-operator appears three times in Eq. (5.26), the sign of .B1 changes
under space inversion by . r ' = −r. The same applies to the other terms of .B. There-
fore, when a crystal has inversion symmetry, .B is zero within the dipole approx-
imation. On the other hand, .U has no such restriction [Eq. (5.25)]. Because the .U
contribution dominates .β̃ in the X-ray region, second-order nonlinear-optical pro-
cesses are possible even in nonlinear crystals with inversion symmetry.

5.2.2 X-Ray Second Harmonic Generation

We consider X-ray second harmonic generation (SHG), which is a special case of

SFG, i.e., .ω1 = ω2 . Since the linear and nonlinear polarizabilities are very small
even in crystals, we can ignore local-field correction in the X-ray region. Then,
the nonlinear polarizability can be regarded as a nonlinear susceptibility. Here, we
assume that the crystal is infinitely large. Applying Eqs. (5.18) and (5.25) to the case
of the crystal and replacing .ρ̃ with . FH0 /vc , the nonlinear susceptibility is expressed
as
∑ (2)
(2)
.χ (r) = χ H eiH·r , (5.31)
H
(2) 1 ∗
.χ H = ∈ · β̃ (H, ω1 , ω1 ) : ∈ 1 ∈ 1 (5.32)
2vc 2 cell
ie3 FH0 { ∗ }
. = −
4 v
∈ 2 · H − 4(∈ ∗2 · ∈ 1 )(∈ 1 · H) , (5.33)
8m ω1 c
2
128 Y. Kubota and K. Tamasaku

where . H, .β̃ cell , and .vc represent the reciprocal lattice vector, the nonlinear polariz-
ability of the unit cell, and the volume of the unit cell, respectively. . FH0 is the crystal
structure factor excluding the dispersion corrections and expressed as
∫
. FH0 = ρcell (r)e−iH·r dr, (5.34)

where .ρcell is the electron density of the unit cell.

Solving wave equations including the nonlinear terms, we obtain the intensity of
SHG: ( )
32π 3 2 2 (2) 2 2 Δk H l
. I2 (l) = K 2 l |χ H | I1 sinc2 , (5.35)
c 2

where .l is the crystal length, . I1 is the intensity of the fundamental wave, and we
defined
.Δk H = 2k 1 − k 2 + H. (5.36)

The efficiency of SHG is maximum for .Δk H = 0, which is called a phase-matching

condition. In the visible region, birefringence of the nonlinear crystal enables to
achieve the phase-matching condition at . H = 0 (collinear geometry, i.e., .2k1 = k2 ).
On the other hand, X-ray SHG is observed as nonlinear diffraction associated with
. H, because it is difficult to use birefringence in the X-ray region.
X-ray SHG was first observed by Shwartz et al. at SACLA [33]. The 7.3-keV
monochromatic XFEL beam was focused down to 1.5-.µm size by the general-
purpose KB mirror (Table 5.1), producing a peak intensity of .1016 W/cm2 . A 0.48-
mm-thick (111) diamond was used for the nonlinear crystal, and aligned to satisfy
the phase-matching condition with . H = (0, 2, −2) in the transmission geometry
(Fig. 5.2a). Because the .011̄ reflection is forbidden for the fundamental wave, elastic
scattering, which can produce strong noise, is suppressed to a very low level. The
14.6-keV SHG signal grows quadratically as the incident pulse energy (Fig. 5.2b) as
expected from Eq. (5.35).

5.2.3 Parametric Down-Conversion

Parametric down-conversion (PDC) is one of the second-order nonlinear-optical pro-

cesses. In this process, an incident photon nonlinearly interacts with materials, and
spontaneously splits into two photons. The incident photon is called “pump”, while
the generated two photons are “signal” and “idler”. PDC can be regarded as an inverse
process of SFG and SHG.
X-ray PDC was first observed by Eisenberger et al. [32] with a Be crystal in 1971.
The synchrotron PDC experiment was performed by Yoda et al. at the Photon Factory,
KEK [34]. Figure 5.3 shows the experimental setup and the coincidence events as a
function of the angle measured from the Bragg condition. They excited a 0.6 mm-
5 Nonlinear X-Ray Spectroscopy 129

Fig. 5.2 Second harmonic generation. a Schematic experimental setup. A diamond (111) thin plate
placed on the focus was set to satisfy the phase-matching condition (bottom). b Second harmonic
signal rate as a function of the incident pulse energy. The inset shows a histogram of the pulse
energy. Reprinted with permission from Shwartz et al. [33]. Copyright (2014) by the American
Physical Society

Fig. 5.3 X-ray parametric down-conversion. a Schematic experimental setup. b Rocking curves
for the diamond 111 nonlinear diffraction (closed circles) and the linear (Bragg) diffraction (open
circles). Reprinted with permission from Yoda et al. [34]. Copyright by IUCr Journals

thick diamond by 19-keV X-rays and detected 9.5-keV signal and idler photons.
Because the signal and idler photons are generated at the same time, one can recog-
nize the PDC event, when one simultaneously detects two photons. Pulsed nature of
synchrotron radiation and fast avalanche photodiode (APD) detectors enabled simul-
taneous detection at a very low background level (Fig. 5.3b). Since 2000, X-ray PDC
experiments have been performed by, for example, Adams et al. [35] and Shwartz
et al. [36]. Unfortunately, no new information on the nonlinear medium is available
with X-ray PDC, since it depends on . FH0 for the three X-ray waves case [Eq. (5.33)].
However, quantum-optical studies with X-ray PDC have reported successful quan-
tum ghost imaging using entanglement between the signal and idler [37, 38] and
generation of heralded X-ray photons [39].
130 Y. Kubota and K. Tamasaku

5.2.4 Sum Frequency Generation

This section concludes by introducing an X-ray SFG experiment with infrared light.
For this purpose, we first consider the X-ray nonlinear polarizability including one
long-wavelength light [40]. We assume .ω1 « ω2,3 and .ωng « ω2,3 in Eq. (5.19),
which can be applied to light elements. The first two terms in Eq. (5.19),
{
∑ ⟨g|∈ 2 ei(K 2 −K 3 )·r̂ |n⟩⟨n|eiK 1 ·r̂ ∈ 1 · p̂|g⟩
.U = − mℏ
n
ωng − ω1
}
⟨g|eiK 1 ·r̂ ∈ 1 · p̂|n⟩⟨n|∈ 2 ei(K 2 −K 3 )·r̂ |g⟩
+ , (5.37)
ωng + ω1

are dominant. Below, we take the dipole approximation and assume a phase-matching
condition of
. K 3 = K 1 + K 2 + H. (5.38)

We calculate Eq. (5.37) and obtain

∫
m 2 ℏ2 ω1
.U=− ∈2 H · {α(r, ω1 ) · ∈ 1 } e−iH·r dr, (5.39)
e2

where .α(r, ω) is the linear polarizability for the long-wavelength light. When the
response of the material is isotropic (.α = α I), .U can be expressed by
∫
m 2 ℏ2 ω1
U=−
. (H · ∈ 1 ) ∈ 2 α(r, ω1 )e−iH·r dr. (5.40)
e2

From Eq. (5.18), the nonlinear polarizability is written as

ie
.β̃ cell (H, ω1 , ω2 ) : ∈ 1 ∈ 2 = − (H · ∈ 1 ) ∈ 2 α cell
H (ω1 ), (5.41)
mω2 ω3

where .α cell
H is defined with the polarizability of the unit cell, .αcell (r), as
∫
.α cell
H = αcell (r)e−iH·r dr. (5.42)

The SFG with infrared and X-ray photons was observed by Glover et al. at LCLS
[41]. The infrared (1.55-eV) and X-ray (8-keV) pulses were simultaneously illumi-
nated onto a diamond crystal (Fig. 5.4a). The SFG signal was observed at a photon
energy of 8 keV + 1.55 eV, as shown in Fig. 5.4b. Figure 5.4c shows the SFG count
depending on the polarization direction of the infrared light. Note that both the X-ray
and infrared photons are linearly polarized. The signal is maximum when the angle
5 Nonlinear X-Ray Spectroscopy 131

Fig. 5.4 Sum frequency generation. a Schematic experimental setup. b SFG signal as a function
of the energy analyzer angle. The inset shows the energy spectrum of the SFG signal. c Optical
polarization dependence of the SFG signal. Reprinted with permission from Glover et al. [41].
Copyright by Springer Nature

of optical polarization vector is in the diffraction plane (.0◦ ) and is zero when it is
normal to the diffraction plane (.±90◦ ), as expected from the polarization factor of
. H · ∈ 1 in Eq. (5.41).
132 Y. Kubota and K. Tamasaku

5.3 Featuring Examples of Nonlinear X-Ray Spectroscopy

5.3.1 X-Ray Two-Photon Absorption Spectroscopy

We begin this section by considering nonlinear absorption, i.e., X-ray multi-photon

absorption. Young et al. first observed the X-ray multi-photon absorption process
using soft X-ray FEL at LCLS [42]. Starting with this research, many experimental
and theoretical studies on X-ray multi-photon absorption have been reported in the
soft X-ray region [43]. The multi-photon absorption in the hard X-ray region was
first reported for xenon (Xe) by Fukuzawa et al., where they observed .Xe24+ caused
by absorbing five 5.5-keV photons [44]. Rudenko et al. observed .I48+ , which was
accounted for by twenty-photon absorption in .CH3 I at 8.3 keV [45]. One important
aim for investigation of hard X-ray multi-photon absorption is its close connection to
single-molecule structure analysis with ultra-intense X-rays [23]. However, for the
sake of simplicity, we deal with X-ray two-photon absorption (TPA) in this section.
There are two processes in X-ray TPA; one is sequential (sequential TPA) where
an atom absorbs two photons step by step, and the other is direct (direct TPA) where
there is no real intermediate state. First, we discuss the difference between the two
processes by comparing the TPA cross-section.
The cross-section can be written as [46]

.σ (2) = σe(1) τe σg(1) , (5.43)

(1)
where .σe,g represents the one-photon absorption (OPA) cross-section in the interme-
diate and ground states, respectively, and .τe is the lifetime of the intermediate state.
The OPA cross-section is given by Fermi’s golden rule

4π 2 αω2f i | |
σ (1) (ω) =
. |⟨k f |∈ · r̂|1s⟩|2 δ(ω f i − ω), (5.44)
ω
where .α represents the fine structure constant. We are considering the process where
a 1s electron is excited to the continuum state in the vacuum, .|k f ⟩, with an energy
(1)
difference of .ℏω f i . The values of .σe,g are comparable for both the sequential and
direct absorption processes. The intermediate state in the sequential process may
have a lifetime of .τe = 0.1−1 fs, while .τe for the direct process is determined by
the uncertainty principle. For example, when the X-ray photon energy is 5 keV, we
estimate .τe ∼ 1/ω = 1.3 × 10−19 s. It results in that .σ (2) of the direct TPA is about
four orders of magnitude smaller than that of the sequential one.
Figure 5.5 shows the sequential TPA creating a double core-hole state in krypton
(Kr) atom by Tamasaku et al. at SACLA [47]. In this process, the first photon excites
a 1s electron to the continuum, and then the second photon does the other 1s electron
before the single core-hole state decays. Two weak peaks denoted by . K h α1,2 on the
higher energy tail of . K α are hypersatellites, the X-ray fluorescence from the . K -shell
double core-hole state. An intensity ratio of . K α to . K h α indicates that about 0.1%
5 Nonlinear X-Ray Spectroscopy 133

Fig. 5.5 Double core-hole creation. a X-ray fluorescence spectra of Kr measured with high (circles)
and low intensities (triangles). b Pulse-energy dependence of the . K h α2 intensity. Reprinted with
permission from Tamasaku et al. [47]. Copyright (2013) by the American Physical Society

of the excited states is in the double core-hole state. Figure 5.5b shows the . K h α2
intensity as a function of the incident X-ray pulse energy, which shows a quadratic
dependence expected for the two-photon process. This result indicates that the Kr
atoms in the . K -shell single core-hole state interact with X-rays within the lifetime
as short as 0.24 fs [48].
Next, we consider direct TPA. For simplicity, we assume excitation by a single
beam with . K and .ω. From Eq. (5.7), there can be two contributions to .σ (2) ; one
originates from the second-order perturbation of . p · A, another from the first-order
perturbation of .A2 . Using Fermi’s golden rule, we obtain
| |
|∑ ω ω ⟨k |∈ · r̂|n⟩⟨n|∈ · r̂|1s⟩ |2
(2, pA) | f n ni f |
.σ (ω) = 8π α | | δ(ω f i − 2ω), (5.45)
3 2
| n ω ωni − ω |
| |
(2,A )
(ω) = 32π 3re2 |⟨k f |v · r̂|1s⟩| δ(ω f i − 2ω),
2 2
.σ (5.46)

where .re is the classical electron radius, and .v is a unit vector in the . K direction
written as c
.v = K. (5.47)
2ω

From Eqs. (5.44) and (5.46), .σ (2,A ) may be expressed by

2

πre2 (1)
σ (2,A ) (ω, ω) =
2
. σ (2ω). (5.48)
αω

We consider which of . p · A and .A2 is dominant in the direct TPA. From

Eqs. (5.43) and (5.48), this question can be solved by comparing .re2 /α to .σe(1) . In
134 Y. Kubota and K. Tamasaku

Fig. 5.6 Direct two-photon absorption. a Schematic experimental setup. b Pulse-energy depen-
dence of Ge TPA fluorescence (circles). Dashed line indicates a quadratic function. Reprinted with
permission from Tamasaku et al. [50]. Copyright by Springer Nature

the case of the direct TPA near absorption edges, .σe(1) may be evaluated to be
−20
.∼ 10 cm.2 . Because this value is about three orders of magnitude larger than
−23
.r e /α = 1.1 × 10 cm2 , . p · A is dominant in the direct TPA near absorption edges.
2

On the other hand, Varma et al. performed numerical calculations indicating that the
contribution from .A2 to the direct TPA dominates above a photon energy of 6.8 keV
in atomic hydrogen [49]. Since the absorption edge of hydrogen is 13.6 eV, this
process is above-threshold ionization.
An experiment of the direct TPA for germanium (Ge) was performed by Tamasaku
et al. at SACLA [50]. The XFEL beam was focused to .110 × 140 nm2 by the
two-stage focusing system (Table 5.1 and Fig. 5.6a), yielding a peak intensity of
.∼ 10
19
W/cm2 . The photon energy was set to 5.6 keV half of 11.2 keV, just above
the Ge . K -shell binding energy of 11.1 keV. The . K α1 fluorescence was detected by a
spectrometer consisting of an analyzer crystal and a charge-coupled device (CCD).
Figure 5.6b shows the pulse-energy dependence of two-photon excited . K α1 fluores-
cence counts. The signal is approximately proportional to the square of the pulse
energy, which indicates that the . K α1 fluorescence is due to the direct TPA. Ghimire
et al. also observed the direct TPA in zirconium (Zr) [51]. The experimental values
of .σ (2) for Ge and Zr were consistent with the theoretically predicted . Z −6 scaling
low for hydrogenic ions. Since TPA requires two photons at the same time, it leads to
a unique application of X-ray intensity correlator. Osaka et al. constructed an X-ray
intensity correlator using a split-delay optics, successfully measured an X-ray auto-
correlation function at 9 keV, and determined the pulse duration of only .7.6 ± 0.8 fs
[52].
One can see deviation from the quadratic dependence in the high pulse-energy
region. When the peak intensity is very high, many Ge atoms could be excited by the
OPA process. A Ge atom with a core hole in the . L shell has a higher . K -shell binding
5 Nonlinear X-Ray Spectroscopy 135

Fig. 5.7 X-ray two-photon

absorption spectroscopy.
TPA (squares) and OPA
(solid line) spectra of Cu
metal. Reprinted with
permission from Tamasaku
et al. [53]. Copyright (2018)
by the American Physical
Society

energy due to reduced screening of the nucleus potential by the. L shell. Since the total
energy transfer by TPA is just above the . K edge in the ground state, the L-shell core-
hole states cannot absorb two photons, suppressing TPA. Such a saturable trend will
be discussed in Sect. 5.3.2. Although application of TPA to nonlinear spectroscopy
may look straightforward, it requires an advanced experimental technique to measure
the TPA signal at much lower peak intensities in order to study the undamaged
electronic structure.
The first TPA spectroscopy was performed for copper (Cu) metal by Tamasaku et
al. at SACLA [53]. The peak intensity of .5 × 1015 W/cm2 was carefully determined
by checking the threshold where the OPA spectrum started to change. The weak TPA
signal due to the lower peak intensity was measured by considerable suppression of
higher harmonics of XFEL radiation and a more efficient detecting system using the
photon-energy resolution of CCD and omitting an analyzer crystal.
The TPA spectrum is quite different from OPA as shown in Fig. 5.7. Not only the
spectral shape but the absorption edge differed, which indicates that each spectrum
represents different electronic states. The physical background can be readily under-
stood by comparing Eqs. (5.44) and (5.45). The OPA process includes one dipole
transition, while TPA does two. Because the dipole transition is allowed between
states with different parities, OPA in Cu metal is mainly due to a transition from
1s to 4p (1s.→4p). In the case of TPA, the dominant transition is considered to be
1s.→3d. So, the spectral shape, which relates to a partial density of the final state,
differs between the two spectra. The 3d orbital is closer to the nucleus than 4p, and
therefore is more tightly bound when a core hole is created in the 1s orbital. This
corresponds to the difference in the absorption edge. The sensitivity of the TPA spec-
troscopy to the 3d orbital is expected to serve for investigation of 3d transition metal
compounds, which include various functional materials.
136 Y. Kubota and K. Tamasaku

5.3.2 Saturable Absorption

When a sample is irradiated with high-intensity X-rays, the population of atoms in the
ground state decreases. Above a certain threshold intensity, few atoms can absorb
X-rays in the photon-energy region above the intrinsic absorption edges, and the
sample becomes transparent. This phenomenon is called saturable absorption (SA).
In the hard X-ray region, Yoneda et al. reported the first observation of SA for iron
(Fe) using high-brilliant XFEL pulses with a pulse duration of .∼ 7 fs at SACLA [54].
Figure 5.8a shows the schematic experimental setup. The required X-ray intensity
for SA, . ISA , can be roughly estimated from . ISA ∼ n atom ℏω/(μτc ), where .n atom , .ℏω,
.μ, and .τc represent the number of atoms in a unit volume, the photon energy, the

X-ray linear absorption coefficient, and the core-hole lifetime, respectively. They
estimated . ISA to be .2 × 1019 W/cm2 . Such high-intensity XFEL pulse was obtained
by the two-stage focusing system (Table 5.1). Figures 5.8b and c show the intensity
dependence of absorption coefficient and transmission, respectively, around the Fe
. K edge. When the XFEL pulses with .10
20
W/cm2 were injected, the transmission at
7.13 keV (above the Fe . K edge) increased by a factor of 10 compared to the normal
absorption condition (called cold state).
Then, SA at the aluminum . K edge has been also observed by Rackstraw et al.
at LCLS [55]. In contrast to the above study at SACLA, they discussed relatively
slow phenomena in .∼ 100-fs pulse duration of LCLS, such as change in the . L-shell
population and collisional ionization in addition to the mechanism discussed above.
The time during which SA occurs is determined by .τc . Thus, X-ray pulse dura-
tion can be shortened by using SA as an ultrafast shutter, which was demonstrated
by Inoue et al. at SACLA [56]. The 9-keV XFEL beam was focused down to
.130 × 165 nm size by the 100-exa focusing system (Table 5.1), producing a peak
2

intensity of.10 W/cm2 . The pulse duration was evaluated by an intensity correlation
19

technique [57]. Figure 5.9a shows the result of SA in Cu. The transmittance at 9 keV,
above the Cu . K edge (8.979 keV), increases as the incident X-ray fluence. The pulse
duration after the saturable absorber decreases by .∼ 35% (Fig. 5.9b).

5.3.3 Atomic X-Ray Laser

Most atoms are in the core-hole state under the SA condition. At this instant, there are
no atoms in the relaxed states, which will be reached by emitting X-ray fluorescence.
Thus, high-intensity X-rays can create population inversion, which is prerequisite for
lasing by the conventional scheme. Unlike in the visible region, where a specific three-
or four-level system is required to create the population inversion, in the X-ray region,
any element can act as the lasing medium. This is because X-ray photon energy can
excite inner-shell electrons to the continuum and the absorption photon energy is
different from the emission. In addition, the relayed states reached by emission are
not the ground state. This atomic X-ray laser was realized in the soft X-ray region
5 Nonlinear X-Ray Spectroscopy 137

Fig. 5.8 Saturable absorption. a Schematic experimental setup. b Absorption coefficients for the
Fe foil measured in the vicinity of the Fe . K edge (7113 eV) at different X-ray intensities. c Incident
intensity dependence of the transmission .T normalized to the cold condition .Tcold at 7130 eV
(squares). Reprinted with permission from Yoneda et al. [54]. Copyright by Springer Nature

Fig. 5.9 Pulse shortening by saturable absorber. a Transmittance of 10-.µm-thick Cu foil at 9 keV.
Dotted line shows the transmittance in the cold state. b Pulse duration of the incidence and transmis-
sion. Reprinted with permission from Inoue et al. [56]. Copyright (2021) by the American Physical
Society
138 Y. Kubota and K. Tamasaku

Fig. 5.10 X-ray atomic laser. a, b Schematic experimental setup. XFEL pulses are generated
with undulators in single-color mode (the ASE experiment) (a) or two-color mode (the seeding
experiment) (b). c, d Output energy of. K α emission as a function of the pump pulse-energy density.
c The output energies (blue squares) in the ASE experiment. d The output energies in the seeding
experiment (red squares) and those in the non-seeding condition (gray squares). Green squares
represent typical energies of the input seeding pulse. e, f Typical spectra of amplified . K α emission
in the ASE experiment (e) and the seeding experiment (f). Reprinted with permission from Yoneda
et al. [59]. Copyright by Springer Nature

using neon gas for the first time by Rohringer et al. [58]. In the hard X-ray region,
Yoneda et al. reported the atomic X-ray laser with Cu at SACLA [59]. Figure 5.10
shows their experimental setup and results. The two-stage focusing system (Table 5.1)
generated a 120-nm spot. XFEL pulses with an intensity of about .1019 W/cm2 were
tuned to the Cu . K edge and produced sufficient population inversion to generate
strong amplified spontaneous emission (ASE) at the Cu . K α lines.
In addition, the two-color mode [5] was used to stabilize the lasing process, where
one color tuned to the Cu . K edge (pumping) and the other for the seed (starting) light
tuned to the Cu . K α. The output power of the seeded atomic laser is larger than that
of ASE, as shown in Fig. 5.10d. Furthermore, the bandwidth of the seeded atomic
laser (1.7 eV at minimum) is narrower than that of ASE, and its photon energy can
be switched between . K α1 and . K α2 by tuning the photon energy of seed light.
5 Nonlinear X-Ray Spectroscopy 139

Other unconventional X-ray radiation process is superfluorescence, where coher-

ent radiation is emitted from atoms excited by shorter pulses than the lifetime of
the excited state. This quantum optical emission was reported in the vacuum ultra-
violet (VUV) region by Harries et al. [60] and for the .57 Fe Mössbauer transition by
Chumakov et al. [61].

5.3.4 Stimulated X-Ray Emission Spectroscopy

The ASE process found an interesting spectroscopic application, namely, stimulated

X-ray emission spectroscopy (SXES). The first SXES in short wavelengths was
performed in the VUV region for silicon [62]. In the hard X-ray region, Kroll et al.
observed the stimulated . K α emission from solutions of manganese (Mn) complexes
at LCLS [63]. They have subsequently succeeded in observing the seeded stimulated
emission for the Mn . Kβ [64]. In general, the . Kβ emission is chemically more
sensitive but an order of magnitude weaker than the. K α emission. They made use of a
two-color XFEL to detect the weak signal of the stimulated . Kβ emission (Figs. 5.11a
and b). They used NaMnO.4 solution to perform the Mn. Kβ SXES. The photon energy
of the pump pulse was set above the Mn . K edge to 6.6 keV, and the seed pulse were
tuned to the . Kβ line at 6.49 keV. The size and peak intensity of pump pulse were
.∼ 150 nm and.10 W/cm2 , respectively. Figure 5.11c shows spectra detected around
20

the Mn . Kβ with sample (red) and without (blue). The no-sample curve reflects the
averaged SASE spectrum, while the spectrum with the sample indicates the seeded
. Kβ SXES signal with a single sharp peak. Comparing the observed seeded stimulated
. Kβ emission signal with the spontaneous . Kβ emission into the same solid angle,
they estimated a signal enhancement to be more than .105 . Since . Kβ X-ray emission
spectroscopy is sensitive to the spin state, oxidation, and ligand environment of 3d
transition metal, this demonstration of the . Kβ SXES opens up rich avenues for
investigation in chemistry and materials science.

5.3.5 X-Ray Transient Grating Spectroscopy

This section concludes with an introduction to X-ray transient grating (XTG) spec-
troscopy. Transient grating (TG) is one of four-wave-mixing processes where two
simultaneous excitation waves generate an interference pattern within a sample. The
pattern works as a grating of spatially modulated excitation, which can diffract a
controlled time-delayed probe wave. Rouxel et al. demonstrated a TG spectroscopy
with XFEL pulses [65]. Short wavelength of X-rays can generate gratings with peri-
ods down to nanometer scales, which allows observation of the behavior of excited
bulk materials with large momentum transfer, high spatial resolution, and element
selectivity.
140 Y. Kubota and K. Tamasaku

Fig. 5.11 Stimulated X-ray emission spectroscopy. a Schematic experimental setup using a par-
tially transparent . K α analyzer and . Kβ analyzer for simultaneous detection of . K α ASE and seeded
stimulated . Kβ emission signals. b Schematics of the stimulated emission processes. c The seeded
stimulated . Kβ emission spectrum (red), and averaged seed spectrum (blue). Reprinted with per-
mission from Kroll et al. [64]. Copyright (2020) by the American Physical Society

Fig. 5.12 X-ray transient grating spectroscopy. a Schematic experimental setup. b Delay time (.t)
dependence of the XTG signal (red circles). Reprinted with permission from Rouxel et al. [65].
Copyright by Springer Nature

Figure 5.12a shows the setup of XTG spectroscopy at SwissFEL [65]. A diamond
phase grating with a period of 960 nm diffracted a 7.1-keV XFEL beam to generate
an interference pattern with a period of 770 nm in .Bi4 Ge3 O12 (BGO). The size and
peak intensity of XFEL beam at the sample were .250 × 150 µm2 and .1011 W/cm2 ,
respectively. They successfully detected 400-nm probe laser pulses diffracted by
TG. Figure 5.12b shows a delay time (.t) dependence of the intensity of diffracted
optical wave (XTG signal). The XTG signal rises at the temporal overlap between the
excitation and probe waves, i.e.,.t = 0, followed by decay and oscillations. This result
indicates generation of a photoexcited A.1 coherent phonon with a frequency of .2.6 ±
0.1 THz in BGO. This research demonstrates usefulness of the XTG spectroscopy
for probing properties of bulk materials. The XTG spectroscopy would be a powerful
method for investigation of materials science, such as ultrafast dynamics of lattice,
charge, and spin states on the nanoscale.
5 Nonlinear X-Ray Spectroscopy 141

5.4 Summary

This chapter begins with a theoretical description of nonlinear optics in the hard X-
ray region and provides several noteworthy studies. As mentioned at the beginning
of this chapter, we hope the readers have understood that there are many nonlin-
ear phenomena unique to the hard X-ray region. The fact that the .A2 term in the
nonrelativistic Hamiltonian plays important role in the linear X-ray scattering, i.e.,
the Thomson scattering, also has a close relation to the nonlinear scattering. The
non-resonant nature of the .A2 interaction makes X-rays see the electron itself, and
leads to structural analysis. The dominant terms in the second-order nonlinear sus-
ceptibility include the .A2 term, which allows PDC and SFG to unveil the “structure”
of the electronic distribution responding to the long-wavelength light. On the other
hand, the imaginary part of the linear and the third-order nonlinear susceptibility
is determined by the . p · A contribution. Therefore, the linear and the two-photon
absorption has the same mechanism as that in the optical region. Only deep in the
above-threshold ionization region, the .A2 term may exceed the . p · A absorption
mechanism. It is still not clear which term decides the third-order nonlinear suscep-
tibility. The third-order nonlinear process includes four photons, which are connected
by various combination of the . p · A and .A2 interactions. Futhermore, one or two of
the four photons can be in the longer wavelength region from infrared to soft X-rays,
making this problem very complicated.
Many basic nonlinear processes have been observed as we saw in this chapter.
One of the most important nonlinear processes to be observed might be coherent
Raman scattering, which includes stimulated Raman scattering and coherent anti-
stokes Raman scattering. Both are the third-order nonlinear process, however, the
former relies on the imaginary part of the third-order nonlinear susceptibility, while
the latter on the magnitude. The coherent Raman scattering can be used for nonlin-
ear spectroscopy, and could enhance the sensitivity of the spontaneous Raman spec-
troscopy. Experimentally, the coherent Raman scattering requires two monochro-
matic and intense X-ray beams with controllable photon-energy separation around
an electronic excitation energy of interest. Since the third-order nonlinear process is
observable only in high-intensity region, it may compete with the sample damage.
This can be relaxed by using the Fourier-limited pulse, which has the highest inten-
sity at given pulse energy and bandwidth. Advance of XFELs would enable X-ray
nonlinear spectroscopy to become a versatile tool to resolve physical, chemical, and
biological problems.

Acknowledgements We thank Prof. Iwao Matsuda for the valuable discussion.

References

1. A.M. Kondratenko, E.L. Saldin, Generation of coherent radiation by a relativistic electron

beam in an ondulator. Part. Accel. 10, 207 (1980)
142 Y. Kubota and K. Tamasaku

2. R. Bonifacio, C. Pellegrini, L.M. Narducci, Collective instabilities and high-gain regime in a

free electron laser. Opt. Commun. 50, 373 (1984)
3. J. Amann, W. Berg, V. Blank, F.-J. Decker, Y. Ding, P. Emma, Y. Feng, J. Frisch, D. Fritz, J.
Hastings et al., Demonstration of self-seeding in a hard-X-ray free-electron laser. Nat. Photonics
6, 693 (2012)
4. I. Inoue, T. Osaka, T. Hara, T. Tanaka, T. Inagaki, T. Fukui, S. Goto, Y. Inubushi, H. Kimura,
R. Kinjo et al., Generation of narrow-band X-ray free-electron laser via reflection self-seeding.
Nat. Photonics 13, 319 (2019)
5. T. Hara, Y. Inubushi, T. Katayama, T. Sato, H. Tanaka, T. Tanaka, T. Togashi, K. Togawa, K.
Tono, M. Yabashi et al., Two-colour hard X-ray free-electron laser with wide tunability. Nat.
Commun. 4, 2919 (2013)
6. P. Emma, R. Akre, J. Arthur, R. Bionta, C. Bostedt, J. Bozek, A. Brachmann, P. Bucksbaum, R.
Coffee, F.-J. Decker et al., First lasing and operation of an ångstrom-wavelength free-electron
laser. Nat. Photonics 4, 641 (2010)
7. T. Ishikawa, H. Aoyagi, T. Asaka, Y. Asano, N. Azumi, T. Bizen, H. Ego, K. Fukami, T. Fukui,
Y. Furukawa et al., A compact X-ray free-electron laser emitting in the sub-ångström region.
Nat. Photonics 6, 540 (2012)
8. H.-S. Kang, C.-K. Min, H. Heo, C. Kim, H. Yang, G. Kim, I. Nam, S.Y. Baek, H.-J. Choi, G.
Mun et al., Hard X-ray free-electron laser with femtosecond-scale timing jitter. Nat. Photonics
11, 708 (2017)
9. E. Prat, R. Abela, M. Aiba, A. Alarcon, J. Alex, Y. Arbelo, C. Arrell, V. Arsov, C. Bacellar,
C. Beard et al., A compact and cost-effective hard X-ray free-electron laser driven by a high-
brightness and low-energy electron beam. Nat. Photonics 14, 748 (2020)
10. W. Decking, S. Abeghyan, P. Abramian, A. Abramsky, A. Aguirre, C. Albrecht, P. Alou, M.
Altarelli, P. Altmann, K. Amyan et al., A MHz-repetition-rate hard X-ray free-electron laser
driven by a superconducting linear accelerator. Nat. Photonics 14, 391 (2020)
11. B. Erk, R. Boll, S. Trippel, D. Anielski, L. Foucar, B. Rudek, S.W. Epp, R. Coffee, S. Carron,
S. Schorb et al., Imaging charge transfer in iodomethane upon x-ray photoabsorption. Science
345, 288 (2014)
12. K.H. Kim, J.G. Kim, S. Nozawa, T. Sato, K.Y. Oang, T.W. Kim, H. Ki, J. Jo, S. Park, C. Song
et al., Direct observation of bond formation in solution with femtosecond X-ray scattering.
Nature 518, 385 (2015)
13. J.G. Kim, S. Nozawa, H. Kim, E.H. Choi, T. Sato, T.W. Kim, K.H. Kim, H. Ki, J. Kim, M.
Choi et al., Mapping the emergence of molecular vibrations mediating bond formation. Nature
582, 520 (2020)
14. M. Buzzi, M. Först, R. Mankowsky, A. Cavalleri, Probing dynamics in quantum materials with
femtosecond X-rays. Nat. Rev. Mater. 3, 299 (2018)
15. T. Suzuki, Y. Kubota, A. Nakamura, T. Shimojima, K. Takubo, S. Ito, K. Yamamoto, S. Michi-
mae, H. Sato, H. Hiramatsu et al., Ultrafast optical stress on BaFe.2 As.2 . Phys. Rev. Res. 3,
033222 (2021)
16. Y. Kubota, Y. Tanaka, T. Togashi, T. Ebisu, K. Tamasaku, H. Osawa, T. Wada, O. Sugino,
I. Matsuda, M. Yabashi, Suppression of atomic displacive excitation in photo-induced A.1g
phonon mode of bismuth unveiled by low-temperature time-resolved x-ray diffraction. Appl.
Phys. Lett. 122, 092201 (2023)
17. S. Gerber, H. Jang, H. Nojiri, S. Matsuzawa, H. Yasumura, D.A. Bonn, R. Liang, W.N. Hardy,
Z. Islam, A. Mehta et al., Three-dimensional charge density wave order in YBa.2 Cu.3 O.6.67 at
high magnetic fields. Science 350, 949 (2015)
18. H. Jang, S. Song, T. Kihara, Y. Liu, S.-J. Lee, S.-Y. Park, M. Kim, H.-D. Kim, G. Coslovich,
S. Nakata et al., Characterization of photoinduced normal state through charge density wave
in superconducting YBa.2 Cu.3 O.6.67 . Science Advances 8, eabk0832 (2022)
19. A. Ikeda, Y.H. Matsuda, X. Zhou, S. Peng, Y. Ishii, T. Yajima, Y. Kubota, I. Inoue, Y. Inubushi,
K. Tono et al., Generating 77 T using a portable pulse magnet for single-shot quantum beam
experiments. Appl. Phys. Lett. 120, 142403 (2022)
5 Nonlinear X-Ray Spectroscopy 143

20. D. Kraus, J. Vorberger, A. Pak, N.J. Hartley, L.B. Fletcher, S. Frydrych, E. Galtier, E.J. Gamboa,
D.O. Gericke, S.H. Glenzer et al., Formation of diamonds in laser-compressed hydrocarbons
at planetary interior conditions. Nat. Astron. 1, 606 (2017)
21. T. Okuchi, Y. Seto, N. Tomioka, T. Matsuoka, B. Albertazzi, N.J. Hartley, Y. Inubushi, K.
Katagiri, R. Kodama, T.A. Pikuz et al., Ultrafast olivine-ringwoodite transformation during
shock compression. Nat. Commun. 12, 4305 (2021)
22. Z. He, M. Rödel, J. Lütgert, A. Bergermann, M. Bethkenhagen, D. Chekrygina, T.E. Cowan,
A. Descamps, M. French, E. Galtier et al., Diamond formation kinetics in shock-compressed
C—H—O samples recorded by small-angle x-ray scattering and x-ray diffraction. Science
Advances 8, eabo0617 (2022)
23. R. Neutze, R. Wouts, D. van der Spoel, E. Weckert, J. Hajdu, Potential for biomolecular imaging
with femtosecond X-ray pulses. Nature 406, 752 (2000)
24. M. Suga, F. Akita, K. Hirata, G. Ueno, H. Murakami, Y. Nakajima, T. Shimizu, K. Yamashita,
M. Yamamoto, H. Ago et al., Native structure of photosystem II at 1.95 Å resolution viewed
by femtosecond X-ray pulses. Nature 517, 99 (2015)
25. M. Suga, F. Akita, M. Sugahara, M. Kubo, Y. Nakajima, T. Nakane, K. Yamashita, Y. Umena,
M. Nakabayashi, T. Yamane et al., Light-induced structural changes and the site of O=O bond
formation in PSII caught by XFEL. Nature 543, 131 (2017)
26. T. Kimura, Y. Joti, A. Shibuya, C. Song, S. Kim, K. Tono, M. Yabashi, M. Tamakoshi, T.
Moriya, T. Oshima et al., Imaging live cell in micro-liquid enclosure by X-ray laser diffraction.
Nat. Commun. 5, 3052 (2014)
27. P. Kirkpatrick, A.V. Baez, Formation of optical images by X-rays. J. Opt. Soc. Am. 38, 766
(1948)
28. H. Mimura, H. Yumoto, S. Matsuyama, T. Koyama, K. Tono, Y. Inubushi, T. Togashi, T. Sato,
J. Kim, R. Fukui et al., Generation of 10.20 W cm.−2 hard X-ray laser pulses with two-stage
reflective focusing system. Nat. Commun. 5, 3539 (2014)
29. K. Tono, T. Togashi, Y. Inubushi, T. Sato, T. Katayama, K. Ogawa, H. Ohashi, H. Kimura, S.
Takahashi, K. Takeshita et al., Beamline, experimental stations and photon beam diagnostics
for the hard x-ray free electron laser of SACLA. New J. Phys. 15, 083035 (2013)
30. H. Yumoto, Y. Inubushi, T. Osaka, I. Inoue, T. Koyama, K. Tono, M. Yabashi, H. Ohashi,
Nanofocusing optics for an X-ray free-electron laser generating an extreme intensity of 100
EW/cm.2 using total reflection mirrors. Appl. Sci. 10, 2611 (2020)
31. I. Freund, B.F. Levine, Parametric conversion of X rays. Phys. Rev. Lett. 23, 854 (1969)
32. P. Eisenberger, S.L. McCall, X-ray parametric conversion. Phys. Rev. Lett. 26, 684 (1971)
33. S. Shwartz, M. Fuchs, J.B. Hastings, Y. Inubushi, T. Ishikawa, T. Katayama, D.A. Reis, T. Sato,
K. Tono, M. Yabashi et al., X-ray second harmonic generation. Phys. Rev. Lett. 112, 163901
(2014)
34. Y. Yoda, T. Suzuki, X.-W. Zhang, K. Hirano, S. Kikuta, X-ray parametric scattering by a
diamond crystal. J. Synchrotron Radiat. 5, 980 (1998)
35. B. Adams, P. Fernandez, W.-K. Lee, G. Materlik, D.M. Mills, D.V. Novikov, Parametric down
conversion of X-ray photons. J. Synchrotron Radiat. 7, 81 (2000)
36. S. Shwartz, R.N. Coffee, J.M. Feldkamp, Y. Feng, J.B. Hastings, G.Y. Yin, S.E. Harris, X-ray
parametric down-conversion in the Langevin regime. Phys. Rev. Lett. 109, 013602 (2012)
37. A. Schori, D. Borodin, K. Tamasaku, S. Shwartz, Ghost imaging with paired x-ray photons.
Phys. Rev. A 97, 063804 (2018)
38. S. Sofer, E. Strizhevsky, A. Schori, K. Tamasaku, S. Shwartz, Quantum enhanced X-ray detec-
tion. Phys. Rev. X 9, 031033 (2019)
39. E. Strizhevsky, D. Borodin, A. Schori, S. Francoual, R. Röhlsberger, S. Shwartz, Efficient
interaction of heralded X-ray photons with a beam splitter. Phys. Rev. Lett. 127, 013603 (2021)
40. K. Tamasaku, K. Sawada, E. Nishibori, T. Ishikawa, Visualizing the local optical response to
extreme-ultraviolet radiation with a resolution of .λ/380. Nat. Phys. 7, 705 (2011)
41. T.E. Glover, D.M. Fritz, M. Cammarata, T.K. Allison, S. Coh, J.M. Feldkamp, H. Lemke, D.
Zhu, Y. Feng, R.N. Coffee et al., X-ray and optical wave mixing. Nature 488, 603 (2012)
144 Y. Kubota and K. Tamasaku

42. L. Young, E.P. Kanter, B. Krässig, Y. Li, A.M. March, S.T. Pratt, R. Santra, S.H. Southworth,
N. Rohringer, L.F. DiMauro et al., Femtosecond electronic response of atoms to ultra-intense
X-rays. Nature 466, 56 (2010)
43. C. Bostedt, J.D. Bozek, P.H. Bucksbaum, R.N. Coffee, J.B. Hastings, Z. Huang, R.W. Lee, S.
Schorb, J.N. Corlett, P. Denes et al., Ultra-fast and ultra-intense x-ray sciences: first results
from the Linac Coherent Light Source free-electron laser. J. Phys. B At. Mol. Opt. Phys. 46,
164003 (2013)
44. H. Fukuzawa, S.-K. Son, K. Motomura, S. Mondal, K. Nagaya, S. Wada, X.-J. Liu, R. Feifel, T.
Tachibana, Y. Ito et al., Deep inner-shell multiphoton ionization by intense X-ray free-electron
laser pulses. Phys. Rev. Lett. 110, 173005 (2013)
45. A. Rudenko, L. Inhester, K. Hanasaki, X. Li, S.J. Robatjazi, B. Erk, R. Boll, K. Toyota, Y.
Hao, O. Vendrell et al., Femtosecond response of polyatomic molecules to ultra-intense hard
X-rays. Nature 546, 129 (2017)
46. P. Lambropoulos, X. Tang, Multiple excitation and ionization of atoms by strong lasers. J. Opt.
Soc. Am. B 4, 821 (1987)
47. K. Tamasaku, M. Nagasono, H. Iwayama, E. Shigemasa, Y. Inubushi, T. Tanaka, K. Tono,
T. Togashi, T. Sato, T. Katayama et al., Double core-hole creation by sequential attosecond
photoionization. Phys. Rev. Lett. 111, 043001 (2013)
48. M.O. Krause, J.H. Oliver, Natural widths of atomic. K and. L levels,. K α X-ray lines and several
. K L L Auger lines. J. Phys. Chem. Ref. Data 8, 329 (1979)
49. H.R. Varma, M.F. Ciappina, N. Rohringer, R. Santra, Above-threshold ionization in the x-ray
regime. Phys. Rev. A 80, 053424 (2009)
50. K. Tamasaku, E. Shigemasa, Y. Inubushi, T. Katayama, K. Sawada, H. Yumoto, H. Ohashi,
H. Mimura, M. Yabashi, K. Yamauchi et al., X-ray two-photon absorption competing against
single and sequential multiphoton processes. Nat. Photonics 8, 313 (2014)
51. S. Ghimire, M. Fuchs, J. Hastings, S.C. Herrmann, Y. Inubushi, J. Pines, S. Shwartz, M. Yabashi,
D.A. Reis, Nonsequential two-photon absorption from the . K shell in solid zirconium. Phys.
Rev. A 94, 043418 (2016)
52. T. Osaka, I. Inoue, J. Yamada, Y. Inubushi, S. Matsumura, Y. Sano, K. Tono, K. Yamauchi, K.
Tamasaku, M. Yabashi, Hard x-ray intensity autocorrelation using direct two-photon absorp-
tion. Phys. Rev. Res. 4, L012035 (2022)
53. K. Tamasaku, E. Shigemasa, Y. Inubushi, I. Inoue, T. Osaka, T. Katayama, M. Yabashi, A.
Koide, T. Yokoyama, T. Ishikawa, Nonlinear spectroscopy with X-ray two-photon absorption
in metallic copper. Phys. Rev. Lett. 121, 083901 (2018)
54. H. Yoneda, Y. Inubushi, M. Yabashi, T. Katayama, T. Ishikawa, H. Ohashi, H. Yumoto, K.
Yamauchi, H. Mimura, H. Kitamura, Saturable absorption of intense hard X-rays in iron. Nat.
Commun. 5, 5080 (2014)
55. D.S. Rackstraw, O. Ciricosta, S.M. Vinko, B. Barbrel, T. Burian, J. Chalupský, B.I. Cho, H.-K.
Chung, G.L. Dakovski, K. Engelhorn et al., Saturable absorption of an X-ray free-electron-laser
heated solid-density aluminum plasma. Phys. Rev. Lett. 114, 015003 (2015)
56. I. Inoue, Y. Inubushi, T. Osaka, J. Yamada, K. Tamasaku, H. Yoneda, M. Yabashi, Shortening
X-ray pulse duration via saturable absorption. Phys. Rev. Lett. 127, 163903 (2021)
57. I. Inoue, K. Tamasaku, T. Osaka, Y. Inubushi, M. Yabashi, Determination of X-ray pulse
duration via intensity correlation measurements of X-ray fluorescence. J. Synchrotron Radiat.
26, 2050 (2019)
58. N. Rohringer, D. Ryan, R.A. London, M. Purvis, F. Albert, J. Dunn, J.D. Bozek, C. Bostedt,
A. Graf, R. Hill et al., Atomic inner-shell X-ray laser at 1.46 nanometres pumped by an X-ray
free-electron laser. Nature 481, 488 (2012)
59. H. Yoneda, Y. Inubushi, K. Nagamine, Y. Michine, H. Ohashi, H. Yumoto, K. Yamauchi, H.
Mimura, H. Kitamura, T. Katayama et al., Atomic inner-shell laser at 1.5-ångström wavelength
pumped by an X-ray free-electron laser. Nature 524, 446 (2015)
60. J.R. Harries, H. Iwayama, S. Kuma, M. Iizawa, N. Suzuki, Y. Azuma, I. Inoue, S. Owada,
T. Togashi, K. Tono et al., Superfluorescence, free-induction decay, and four-wave mixing:
propagation of free-electron laser pulses through a dense sample of helium ions. Phys. Rev.
Lett. 121, 263201 (2018)
5 Nonlinear X-Ray Spectroscopy 145

61. A.I. Chumakov, A.Q.R. Baron, I. Sergueev, C. Strohm, O. Leupold, Y. Shvyd’ko, G.V. Smirnov,
R. Rüffer, Y. Inubushi, M. Yabashi et al., Superradiance of an ensemble of nuclei excited by a
free electron laser. Nat. Phys. 14, 261 (2018)
62. M. Beye, S. Schreck, F. Sorgenfrei, C. Trabant, N. Pontius, C. Schüßler-Langeheine, W. Wurth,
A. Föhlisch, Stimulated X-ray emission for materials science. Nature 501, 191 (2013)
63. T. Kroll, C. Weninger, R. Alonso-Mori, D. Sokaras, D. Zhu, L. Mercadier, V.P. Majety, A.
Marinelli, A. Lutman, M.W. Guetg et al., Stimulated X-ray emission spectroscopy in transition
metal complexes. Phys. Rev. Lett. 120, 133203 (2018)
64. T. Kroll, C. Weninger, F.D. Fuller, M.W. Guetg, A. Benediktovitch, Y. Zhang, A. Marinelli,
R. Alonso-Mori, A. Aquila, M. Liang et al., Observation of seeded Mn K.β stimulated X-ray
emission using two-color X-ray free-electron laser pulses. Phys. Rev. Lett. 125, 037404 (2020)
65. J.R. Rouxel, D. Fainozzi, R. Mankowsky, B. Rösner, G. Seniutinas, R. Mincigrucci, S. Catalini,
L. Foglia, R. Cucini, F. Döring et al., Hard X-ray transient grating spectroscopy on bismuth
germanate. Nat. Photonics 15, 499 (2021)
Chapter 6
Future Prospects

Iwao Matsuda, Craig P. Schwartz, Walter S. Drisdell, and Ryuichi Arafune

Abstract This chapter describes the future prospects of research with nonlinear
X-ray spectroscopy, based on recent achievements at the XFEL facilities. The
measurement techniques have been challenging the cutting edge and, at the same
time, the experimental environment has become user-friendly for academic and
industrial research. Nonlinear X-ray spectroscopy will soon become one of the
standard analysis methods for materials science.

6.1 Toward Multi-dimensional Spectroscopy

Some of the most powerful techniques in all of spectroscopy, not merely nonlinear
spectroscopy, are the so-called 2-dimensional techniques, including 2D-IR, 2D-
NMR, and 2D-visible [1]. In these techniques, a material is pumped by two or more
separate pulses and the energy transfer between the associated states is observed
as a function of time between the pulses. This allows for detailed understanding
of ultrafast energy flow and has been used to provide insight into phenomena such
as photosynthesis, and in the case of 2D-NMR is even used to solve protein struc-
tures [2]. In general the  spectrum can be thought of as a 2D plot of a function,
2Dspectrum ω1 , tdelay , ω2 , where ω1 is the first frequency used, ω2 is the second

I. Matsuda
The Institute for Solid State Physics, The University of Tokyo, Chiba, Japan
e-mail: imatsuda@issp.u-tokyo.ac.jp
C. P. Schwartz
Nevada Extreme Laboratory, University of Nevada, Las Vegas, USA
e-mail: craig.schwartz@unlv.edu
W. S. Drisdell (B)
Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA
e-mail: wsdrisdell@lbl.gov
R. Arafune
Research Center for Materials Nanoarchitectonics (MANA), National Institute for Materials
Science (NIMS), Tsukuba, Japan
e-mail: ARAFUNE.Ryuichi@nims.go.jp

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 147
I. Matsuda and R. Arafune (eds.), Nonlinear X-Ray Spectroscopy for Materials Science,
Springer Series in Optical Sciences 246, https://doi.org/10.1007/978-981-99-6714-8_6
148 I. Matsuda et al.

frequency used, and t delay is the time delay between the two pulses. In this way,
at a time delay of tdelay the signal that is generated when pumped with frequencies
ω1 on the x-axis and ω2 on the y-axis can be plotted, giving a 2D plot with signal
intensity the third dimension. This generally enables one to study the energy transfer
that occurs between the two pulses in time, for instance in 2D-IR one might watch
the energy transfer between two different modes of the water molecule and use this
to understand how energy transferred between molecules. In the case of 2D-visible
one can watch how the energy redistributes between electronic states. This approach
would be particularly powerful in the case of X-ray spectroscopy because it will be
possible to watch energy transfer from one particular atom into another by exploiting
the atomic sensitivity of X-rays, enabling one to track intramolecular energy flow
with unprecedented resolution and understanding.
If this could be extended to the X-ray regime, it would provide unprecedented
insight into the flow of energy on an atom specific basis and at time scales potentially
as short as attoseconds. Such measurements are therefore a motivating factor for the
design of future X-ray light sources, and some thought has been put into how to ulti-
mately achieve these measurements. Unfortunately, extending these measurements
from the IR to the X-ray will require overcoming numerous logistical difficulties.
In an ideal case, as these are coherent spectroscopies, one would like to have a
source that is spatially and temporally coherent, which may be true of externally
seeded sources such as FERMI but is not true of SASE sources. This benefits all
measurements by providing cleaner temporal and spectral resolution but is particu-
larly important for types of spectroscopy where coherence is important, including
phenomena such as four wave mixing. Perhaps this is most important for 2D spec-
troscopy, where it is difficult to achieve such measurements otherwise. There has
been progress improving coherence at seeded FELs such as FERMI [3].
Besides coherence, phase stability is also desired. One would like multiple pulses
to be phase locked to each other, to eliminate problems with relative phase shifts
between pulses as time is scanned. This means the phase of each pulse’s arrival can be
controlled besides just the delay time between the pulses. Phase instability would lead
to another variable and could make techniques such as 2D spectroscopy significantly
more challenging to understand if present. This was demonstrated at a seeded FEL
by frequency interferometry [4]. Phase stability has also been accomplished both by
using high spatial coherent beams that are then split, [5] or by using clever seeding
schemes [6].
As X-ray pulses are inherently ionizing, one must also worry about the core hole
lifetime. These can range from tens of femtoseconds to attoseconds depending on
the energy of the core hole being probed, with higher energy core holes being more
unstable and decaying more rapidly [7]. As the Auger decay process will inherently
cause large changes in the system, having a probe pulse occur before Auger emission
can occur is a desirable property for 2D-spectroscopy. This would require ultrafast
pulses on the time scales of attoseconds, with tunable delays ideally ranging from
attoseconds to femtoseconds. Such pulses have been demonstrated previously at
LCLS as well as various HHG sources [8–10].
6 Future Prospects 149

Since 2D-spectroscopy involves comparing one energy to a second potentially

different energy, it requires the ability to separately tune the energy of different
pulses. Again, there has been some progress on this at LCLS, demonstrating energy
tunability of more than 20% for pulses with separations as short as hundreds of
attoseconds [11]. Additionally, one would like to be able to control the angles of
incidence of the separate pulses and generally have full control over the experi-
mental geometry. Because X-ray pulses are generally harder to manipulate spatially
than infrared and visible pulses, steerable only with limited options such as grazing
incidence reflections, Bragg diffractions, and energy specific optics such as zone
plates, this is a significant challenge. However, if one is willing to work in restricted
geometric conditions that would be necessary only for a single given experiment,
such an experiment can often be designed.
Finally, all of these measurements are complicated and would benefit from excel-
lent statistics which can be obtained by having a high repetition rate and a high
pulse power. There has been substantial progress on this, with new FEL facilities
consistently pushing forward with more pulses and often more power. The statistical
portion of the problem is perhaps the portion of the problem the global research
community is most prepared to solve [12, 13].
The individual pieces which would make such a 2D-X-ray measurement possible
exist but putting all these desirable properties together at the same time is a challenge
which does not appear likely to be solved soon. However, there has been substantial
progress toward making these types of measurements, and it does appear inevitable
that such measurements will eventually be made. The potential importance of such
measurements to enable us to understand matter on its natural length and time scale
is enormous.

6.2 Phase Sensitive Spectroscopy

Coherent control is a necessary step in the development of a variety of advanced spec-

troscopy measurements wherein one can tune the atomic state of an atom. However,
coherent control requires exquisite control over light so as to “dress” an atom—in
this context dressing means to put the atom into a specific quantum state. By using
highly tuned waveforms and pulse shapers, a molecule can be “pushed” and “pulled”
through specific transitions to get to a desired end point. These pulses often end up
with such complicated forms that there isn’t a simple analytical way to describe
the pulse shape. The ability to control matter with light, however, is highly desir-
able not just as a requirement for spectroscopy, but also because coherent control
can potentially be used to drive specific reactions with incredible specificity, such as
choosing reaction products of, for example, the dissociation of C2 H2 . Demonstrating
the phenomena in the X-ray regime is difficult because the requirements on the laser
are sizeable.
Coherent control of a seeded beamline was demonstrated on neon ions using
two pulses from a coherent FEL, specifically FERMI [14]. When using two different
150 I. Matsuda et al.

pulses, this requires high longitudinal coherence, something which SASE FELs typi-
cally struggle with. By using two colors at 63.0 and 31.5 nm, the asymmetry of the
angular distribution of photoelectrons could be controlled with a temporal resolution
as fine as 3 attoseconds. In this experiment the two pulses were incident upon neon
gas and by controlling the relative phase between the two pulses, the asymmetry of
emission could be steered. By choosing the properties of the laser light carefully, the
excited atom can be controlled and the decay mechanism for photoelectron emission
can be tuned to favor the p channels versus s and d channels. Such careful control on
a 3 as time scale is unprecedented and could only have been achieved by operating
in the X-ray regime.
One definitive way of showing coherent control is by demonstrating Rabi oscil-
lations. These occur when an electromagnetic field can drive an atom between two
different states as a function of time. While these have been observed for decades,
only recently was this demonstrated with X-ray light, exploiting the coherence of
the FERMI source [15]. The authors demonstrated an Autler-Townes doublet with
the expected spacing from Rabi oscillations at the resonant energy, but when moving
slightly above or slightly below the energy, a singlet was restored, as shown in
Fig. 6.1. Based on detailed theory they were also able to demonstrate an avoided
crossing between the two excited states. This is a significant step toward ultrafast
manipulation of matter with atomic specificity.
Perhaps even more remarkably, attosecond interferometry was demonstrated from
a SASE source. In this case, unlike in the previous example, the beam energy is broad
and has very limited longitudinal coherence. By using cleverly designed optics, phase
control was demonstrated with time resolution as short as 129 attoseconds [16].
Specifically, with a single pulse and a beam splitter it was not necessary to phase
lock two different pulses as a single pulse is intrinsically phase locked with itself.
While this is almost two orders of magnitude worse than the resolution demonstrated
at FERMI, the X-ray source was significantly less coherent and therefore harder to
utilize. In this experiment, by using a split and delay line that simultaneously acts as

Fig. 6.1 Experimental

kinetic energy of exciting
helium with 23.740 eV (red),
23.753 eV (black), and
23.766 eV (blue). The
characteristic doublet is only
seen at when on resonance at
23.753 eV and is not seen
when detuned. This is
indicative of a Rabi
oscillation and is a powerful
demonstration of coherent
control. Reprinted from [15]
by CC BY 4.0
6 Future Prospects 151

a grating due to a cleverly designed optic, the authors were able to choose a specific
wavelength, and then delay one pulse relative to the other (of the same energy)
generating an interference pattern when incident upon xenon atoms. This shows up
as a difference in the ion positions on an imaging detector. Thus, even with relatively
impure light some control over matter at ultrafast timescales can be demonstrated,
although likely never with as fine control as when the FEL is seeded.
This same beam splitter which allows for phase stability between pulses has been
used to demonstrate nonlinear coherence effects. Working with neon atoms at the L-
edge, Ding et al. [5] demonstrated these effects with pump-probe transient absorption
in which both the pump and probe pulses came from the FEL. They demonstrated
transient effects on a very fast timescale (2.4 fs) as well as spectral shifts of 10 meV
due to the Stark effect. While the resolution of these measurements was limited by
the spectrometer employed, in principle there is effectively no limit on the spectral
resolution. The temporal resolution is limited by the coherence of the FLASH laser,
but also can be improved going forward. These results represented a substantial
step forward toward performing 2D (or even higher dimensional) spectroscopy with
X-rays.
Ding et al. later improved upon their original study, demonstrating a result which
is relatively standard for nonlinear optics in the visible, but had eluded scientists in
the X-ray regime [17]. Once again working at the L-edge of Ne, they measured the
chirp in frequency space of a FEL pulse by using transient absorption spectroscopy.
While this technique is well established and relatively standard in the infrared and the
visible, making these measurements at ~ 50 eV required the development of a novel
model for time-dependent absorption. Using this model, they were able to extract
and quantify the chirp. This measurement can in principle be used to characterize and
optimize the FEL properties, and in this case they showed that there was a systematic
shift in the chirp with increasing pulse energy. As the measurement occurs without
external fields the authors claim it can be used in a variety of measurements as it
offers a convenient way to characterize the FEL and represents another step on the
path toward 2D spectroscopy.
Most recently, self-phase modulation has been demonstrated, where the refractive
index of a thin film of magnesium changes within a duration of a single incident X-
ray pulse, leading to a phase shift and a slight change in the spectrum [18]. For
soft X-rays, the phase modulation changed as a function of moving over the edge,
leading from a nonlinear spectral blue shift broadening below the edge and to a red
shift broadening above the edge. This has shown applications in other frequency
ranges for pulse compression both spatially and temporally as well as in spectral
broadening and the generation of supercontinuum light [19, 20]. These properties
will also likely prove useful in the ultimate development of 2D-X-ray spectroscopy.
152 I. Matsuda et al.

6.3 Vacuum Nonlinear X-Ray Optics

Through the book, it is shown that interactions between light and matter have led to
developments of linear and nonlinear X-ray spectroscopy for materials science. In
a case of the measurement with soft X-ray, experiments have been made under the
vacuum condition. A vacuum is a space devoid of matter but the vacuum itself can
be a target of nonlinear science.
One of the characteristic quantities for vacuum science is Schwinger limit.
According to quantum electrodynamics (QED), an electron–positron pair is gener-
ated in vacuum when the electric field exceeds the Schwinger limit [21],

m 2 c3
E QED = = 1.3 × 1018 V/m
e
Thus, neglecting the effect of the magnetic field, no electromagnetic wave exists
under the condition. An ultimate goal for X-ray laser development is to push this
limit and to examine the nonlinear optical phenomena. So far, the current XFEL
source does not have enough optical performance to reach the limit. However, one
may observe the nonlinear optical effect in vacuum in the future.
As detailed in Chap. 2, a Thomson scattering event is described as the first order
process of a Hamiltonian with the A · A or A2 term and it can be recalled in the
Feynman diagram in Fig. 6.2a. The diagram depicts a scattering event at a certain
time with an electron from p to p’ and a photon from K to K’. This description is an
approximate form in the nonrelativistic regime. In relativistic quantum mechanics,
the diagram is essentially given as Fig. 6.2b. The intermediate state (p”) is associ-
ated with the antiparticle of an electron, a positron, and the electron–positron pair
display a polarization in vacuum. Practically, a difference on the order of the photon
associations in Fig. 6.2b can be neglected at an X-ray photon energy of 10 keV since
the total system lacks the energy of 1 MeV necessary for 2 mc2 . Therefore, Thomson
scattering can be described by a single crossing point, as shown in Fig. 6.2a. On the
other hand, the existence of virtual polarization indicates there are also nonlinear
optical phenomena in vacuum. Such an investigation of vacuum nonlinear X-ray
optics directly links to investigation of “vacuum” itself.
In QED, the principle of superposition of electromagnetic waves is no longer
valid and photons scatter each other, associated with the electron–positron pair. The
differential cross section of scattering between the linearly polarized photons are
given by [22, 23]:

dσγ γ →γ γ α 4 ωCMS
6  
= 260 cos4 CMS + 328 cos2 CMS + 580
d (180π ) m2 8

when a photon energy is much smaller than mc2 . ωCMS and CMS are photon energy
and scattering angle in the centroid system. In the formula, one can find that the
differential cross section increases by the 6th power with ωCMS and thus detection may
6 Future Prospects 153

Fig. 6.2 Thomson scattering event in the a nonrelativistic and b relativistic regimes. Momentum
of an electron is given by p, while wavenumber of a photon is by K

be possible with high photon energies such as X-rays. On the other hand, interestingly,
quantum mechanics allows for the possibility that the photon-photon scattering event
is associated with an undiscovered elementary particle beyond the standard model
(Fig. 6.3). A proposed candidate is Axion, which has also been invoked as dark matter.
It should be noted that initial experiments have been conducted at SACLA [24, 25],
but further technical improvements, such as installation of the higher sensitivity
detector, are required to definitively settle the issue. Vacuum nonlinear X-ray optics
or spectroscopy will become the intriguing and significant topic in fundamental
physics.

Fig. 6.3 Feynman diagrams of photon-photon scattering from (K 1 , K 2 ) to (K 3 , K 4 ) a in the regime

of quantum electrodynamics and b in a possible regime with the undiscovered elementary particle
154 I. Matsuda et al.

6.4 Developments in Experimental Stations for Materials

Science

As described in Chap. 2 X-ray spectroscopy is nowadays used to trace evolutions of

electronic states or atomic structures of functional materials under working condi-
tions, such as batteries and catalysts. Such so-called “operando” spectroscopy exper-
iments can also be carried out with nonlinear X-ray spectroscopy, to provide unique
information about a sample that cannot be obtained with conventional linear X-ray
spectroscopies. Measurements of the nonlinear X-ray optical phenomena, shown
in Chaps 4 and 5, are made with a photon-in and photon-out scheme. Therefore,
signal can be detected under a variety of external environments for a sample, such
as electric current, gate voltage, and magnetic field. At an XFEL facility, such as
SACLA, an experimental station for nonlinear X-ray spectroscopy has opened, for
both academic and industrial users. The experimental chamber is equipped with
sample transfer, temperature control, external field application, ellipsometer, spec-
trometer, and nanobeam focusing systems. Moreover, recent developments of nano-
focusing techniques for the X-ray beams have reached spot sizes of < 100 nm for soft
X-ray [26] and < 10 nm for hard X-ray [27]. Technical innovations of nano-focusing
mirrors have generated the high-efficient, achromatic and brilliant beam that has led
us to probe nano-sized samples for nanotechnology and also to enhance the nonlinear
X-ray signals [28–30]. The user-friendly station (Fig. 6.4) has already been used in
the industrial area, such as investigations of lithium batteries and solar cells [31–
33]. The recent observation of magnetization-induced soft X-ray second harmonic
generation has found to be useful for research of magnetics and spintronics [34].
Nonlinear X-ray spectroscopy will soon be one of the standard analysis methods to
characterize a sample and to trace its functionalities.
At X-ray beamlines, guiding and shaping of the X-ray beam has been achieved
using geometric optics or wave optics based on linear responses of light with optical
elements, such as mirrors. It has been the quality of the optical components, such
slope error or roughness of a mirror, that has determined the optical specification, such
as beam-size and pulse-width. Nowadays, the fabrication is becoming so established
that performance is reaching the limit that is expected purely by the optical principles.
A rise of nonlinear X-ray optics and spectroscopy opens a new research field to
develop the new fundamental principles for beam technology that use the nonlinear
optical effect to control the X-ray beam. One example is shortening of the X-ray
pulse duration by the saturable absorption that is associated with the inner-shell
photoionization [35]. Such a technical development has a high potential to break
through the conventional technical limits at the beamline.
X-ray free electron laser light sources have been developed significantly with
high power, ultrashort pulses, and high repetition rate [36]. The brilliant optical
performance also paves new and various ways for nonlinear X-ray spectroscopy.
We note that the combination of nonlinear X-ray optical phenomena with coherent
imaging methods is also being explored. It is certain that the research field will expand
6 Future Prospects 155

Fig. 6.4 Multi-purpose experimental station for nonlinear X-ray spectroscopy at the soft X-ray
beamline BL1 at SACLA. By permission of RIKEN

and various mysteries of materials science will be unveiled by the new analysis
methods.

6.5 Into the Deep: Nonlinear Science

In the previous chapter, we stated that it is important to find new functionalities of

materials as well as to consider “what is a new function?” for material science. The
target “new function” may differ from researcher to researcher, but we would like to
present one of our ideas.
156 I. Matsuda et al.

The difference between “right” and “left” is one of the fundamental concepts that
pervade all of science, from particle physics to biological activity. The violation of
the law of conservation of parity in particle physics shows that there is an essential
left–right distinction in our world. In life science as well, amino acids in DNA have a
chiral structure, but only molecules of one chirality constitute life. The synthesis of
molecules with only homochiral structures is one of the important topics in chemistry,
as exemplified by the study of asymmetric reactions by Noyori [37].
The challenge of creating and controlling the difference between “right and left” in
dynamic processes, rather than the static “right and left” in material structures, is now
at the forefront of material science. The differences between “rightward” and “left-
ward” properties are called nonreciprocal phenomena. We believe that the pursuit
of nonreciprocal phenomena for quantum (electrons and phonons) and degrees of
freedom (spin and valley) will lead to the development of new control methods for
materials, information, and energy transfer methods. Since diodes that realize elec-
tron rectification are widely used, nonreciprocal phenomena have not been recog-
nized as an important issue in material science. Recently, however, several examples
of dynamic control of nonrelativistic performance by new quantum mechanisms have
been demonstrated, and this is beginning to be recognized as an interesting issue.
This is because, compared to the static “right–left” difference in structure, nonre-
ciprocal phenomena involve not only space-reversal symmetry breaking but also
time-reversal symmetry breaking, and thus have richer physics inherent in them.
Considering the benefits that diodes have brought to modern society, the impact of
facilitating the control of nonrelativistic phenomena in other quantum and degrees
of freedom would be extremely large. The readers would have recognized that the
mechanisms of photosynthesis and the transport of proteins, information, and energy
in the cell are important questions in the life sciences as well. The fundamental ques-
tion of “What determines the direction in which matter, information, and energy
move?” should be closely related to the nonreciprocal phenomena.
Readers of this book may think that nonreciprocity is a topic that has come out
of nowhere. However, we believe that the experimental techniques discussed in this
book can make a significant contribution to the development and advancement of
nonreciprocal functions of matter in the future and are therefore highly relevant to this
book. Controlling the coherent response of matter with high temporal precision and
detecting the asymmetry inherent in matter with high sensitivity, which are discussed
throughout this book, are essential for developing the nonreciprocal response of
quantum and degrees of freedom in materials.

References

1. P. Hamm, M.T. Zanni, Concepts and Methods of 2D Infrared Spectroscopy (Cambridge

University Press, Cambridge, 2011)
2. J. Keeler, Understanding NMR Spectroscopy, 2nd edn. (John Wiley and Sons, Chichester, U.K.,
2010)
6 Future Prospects 157

3. A. Fabris et al., Fermi upgrade plans, in Proceedings of the 7th Int. Particle Accelerator
Conference, vol. IPAC2016, (2016), p. 4
4. D. Gauthier, P.R. Ribič, G. De Ninno, E. Allaria, P. Cinquegrana, M.B. Danailov, A. Demi-
dovich, E. Ferrari, L. Giannessi, Generation of phase-locked pulses from a seeded free-electron
laser. Phys. Rev. Lett. 116, 024801 (2016)
5. T. Ding et al., Nonlinear coherence effects in transient-absorption ion spectroscopy with
stochastic extreme-ultraviolet free-electron laser pulses. Phys. Rev. Lett. 123, 103001 (2019)
6. P.K. Maroju et al., Attosecond pulse shaping using a seeded free-electron laser. Nature 578,
7795 (2020)
7. C. Nicolas, C. Miron, Lifetime broadening of core-excited and -ionized states. J. Electron
Spectrosc. Relat. Phenom. 185, 267 (2012)
8. P.K. Maroju et al., Complex attosecond waveform synthesis at FEL FERMI. Appl. Sci. 11, 21
(2021)
9. J.T. O’Neal et al., Electronic population transfer via impulsive stimulated X-ray raman
scattering with attosecond soft-X-Ray pulses. Phys. Rev. Lett. 125, 073203 (2020)
10. J. Li, J. Lu, A. Chew, S. Han, J. Li, Y. Wu, H. Wang, S. Ghimire, Z. Chang, Attosecond science
based on high harmonic generation from gases and solids. Nat. Commun. 11, 1 (2020)
11. Y. Sun, F.-J. Decker, J. Turner, S. Song, A. Robert, D. Zhu, Pulse intensity characterization of
the LCLS nanosecond double-bunch mode of operation. J. Synchrotron. Rad. 25, 3 (2018)
12. J. Stohr, Linac Coherent Light Source II (LCLS-II) Conceptual Design Report, No. SLAC-R-
978, 1029479, 2011
13. I. Georgescu, The first decade of XFELs. Nat. Rev. Phys. 2, 7 (2020)
14. K.C. Prince et al., Coherent control with a short-wavelength free-electron laser. Nature Photon.
10, 3 (2016)
15. S. Nandi et al., Observation of Rabi dynamics with a short-wavelength free-electron laser.
Nature 608, 7923 (2022)
16. S. Usenko, A. Przystawik, M.A. Jakob, L.L. Lazzarino, G. Brenner, S. Toleikis, C. Haunhorst,
D. Kip, T. Laarmann, Attosecond interferometry with self-amplified spontaneous emission of
a free-electron laser. Nat. Commun. 8, 1 (2017)
17. T. Ding et al., Measuring the frequency chirp of extreme-ultraviolet free-electron laser pulses
by transient absorption spectroscopy. Nat. Commun. 12, 1 (2021)
18. C. Ferrante et al., Non-linear self-driven spectral tuning of extreme ultraviolet femtosecond
pulses in monoatomic materials. Light Sci. Appl. 10, 1 (2021)
19. Y. Yamaguchi, R. Hida, T. Suzuki, F. Kannari, Supercontinuum generation by self-phase modu-
lation and induced phase modulation at fused silica thin plate array, in 2017 Conference on
Lasers and Electro-Optics Pacific Rim, (Optica Publishing Group, 2017), p. s1402
20. A.M. Weiner, J.P. Heritage, R.H. Stolen, Self-phase modulation and optical pulse compression
influenced by stimulated raman scattering in fibers. J. Opt. Soc. Am. B, JOSAB 5, 364 (1988)
21. J. Schwinger, On Gauge Invariance and Vacuum Polarization. Phys. Rev. 82, 664 (1951)
22. B. Tollis, Dispersive approach to photon-photon scattering. Nuovo Cimento 32, 757 (1964)
23. B. Tollis, The scattering of photons by photons. Nuovo Cimento 35, 1182 (1965)
24. T. Inada, T. Yamaji, S. Adachi, T. Namba, S. Asai, T. Kobayashi, K. Tamasaku, Y. Tanaka, Y.
Inubushi, K. Sawada, M. Yabashi, T. Ishikawa, Search for photon–photon elastic scattering in
the X-ray region. Phys. Lett. B 732, 356 (2014)
25. Y. Seino, Search for the vacuum polarization using a high-field laser and an XFEL, doctoral
thesis, the University of Tokyo, 2021
26. K. Horiba, Y. Nakamura, N. Nagamura, S. Toyoda, H. Kumigashira, M. Oshima, K. Amemiya,
Y. Senba, H. Ohashi, Scanning photoelectron microscope for nanoscale three-dimensional
spatial-resolved electron spectroscopy for chemical analysis. Rev. Sci. Instrum. 82, 113701
(2011)
27. H. Mimura, S. Handa, T. Kimura, H. Yumoto, D. Yamakawa, H. Yokoyama, S. Matsuyama,
K. Inagaki, K. Yamamura, Y. Sano, K. Tamasaku, Y. Nishino, M. Yabashi, T. Ishikawa, K.
Yamauchi, Breaking the 10 nm barrier in hard-X-ray focusing. Nat. Phys. 6, 122 (2010)
158 I. Matsuda et al.

28. H. Motoyama, S. Owada, G. Yamaguchi, T. Kume, S. Egawa, K. Tono, Y. Inubushi, T. Koyama,

M. Yabashi, H. Ohashi, H. Mimura, Intense sub-micrometre focusing of soft X-ray free-electron
laser beyond 1016 W cm−2 with an ellipsoidal mirror. J. Synchrotron Rad. 26, 1406 (2019)
29. Y. Takeo, H. Motoyama, T. Shimamura, T. Kimura, T. Kume, Y. Matsuzawa, T. Saito, Y.
Imamura, H. Miyashita, K. Hiragur, H. Hashizume, Y. Senba, H. Kishimoto, H. Ohashi, H.
Mimura, A highly efficient nanofocusing system for soft x rays. Appl. Phys. Lett. 117, 151104
(2020)
30. T. Kume, K. Hiraguri, Y. Imamura, Y. Matsuzawa, T. Saito, H. Miyashita, Y. Takeo, Y. Senba,
H. Kishimoto, H. Ohashi, H. Hashizume, H. Mimura, Fabrication of soft x-ray monolithic
Wolter mirror based on surface scanning measurement using touch probe. Rev. Sci. Instrum.
93, 063101 (2022)
31. A. Zong, R.R. Nebgen, S.-C. Lin, J.A. Spies, M. Zuerch, Emerging ultrafast techniques for
studying quantum materials. Nat. Rev. Mater. 8, 224 (2023)
32. C. Woodahl, S. Jamnuch, A. Amado et al., Probing lithium mobility at a solid electrolyte
surface. Nat. Mater. (2023). https://doi.org/10.1038/s41563-023-01535-y
33. M. Horio, T. Sumi, J. Bullock, Y. Hirata, M. Miyamoto, B. Nebgen, T. Wada, T. Senoo, Y.
Tsujikawa, Y. Kubota, S. Owada, K. Tono, M. Yabashi, T. Iimori, M.W. Zuerch, I. Matsuda,
C.P. Schwartz, W.S. Drisdell, Detecting driving potentials at the buried SiO2 nanolayers in
solar cells by chemical-selective nonlinear X-ray spectroscopy. Appl. Phys. Lett. 123, 031602
(2023)
34. T. Sumi, M. Horio, T. Senoo, Y. Kubota, G. Yamaguchi, T. Wada, M. Miyamoto, K. Yamaguchi,
Y. Tsujikawa, Y. Sato, M. Niibe, Y. Hirata, Y. Miyauchi, D. Oshima, T. Kato, S. Owada, K.
Tono, M. Yabashi, I. Matsuda, Observing soft x-ray magnetization-induced second harmonic
generation at a heterojunction interface. Appl. Phys. Lett. 122, 171601 (2023)
35. I. Inoue, Y. Inubushi, T. Osaka, J. Yamada, K. Tamasaku, H. Yoneda, M. Yabashi, Shortening
X-ray pulse duration via saturable absorption. Phys. Rev. Lett. 127, 163903 (2021)
36. N. Huang, H. Deng, B. Liu, D. Wang, Z. Zhao, Features and futures of X-ray free-electron
lasers. The Innovation 2, 100097 (2021)
37. R. Noyori, Monograph: asymmetric catalysis in organic synthesis (John Wiley & Sons, New
York, 1994)
Index

A N
Above threshold photoemission, 57, 58, 63, Nonlinear optics, 2, 7, 13, 83, 84, 86,
72 89–91, 94, 98, 106, 110–112, 119,
Absorption spectroscopy, 10, 11, 15, 19, 120, 122, 125, 141, 151
20, 28–30, 52, 92, 132, 135, 151 Nonlinear optical susceptibility, 3, 87, 88
Anharmonicity, 1, 5, 7
Atomic X-ray laser, 136, 138
P
Parametric down-conversion, 87, 90, 128,
129, 141
C Phase sensitive spectroscopy, 149
Coherent multi-photon photoemission, 12 Photoelectron spectroscopy, 15, 20, 33–35,
44–47, 57, 58, 66–69, 72, 74, 94
Photoemission, 11, 12, 24, 34, 35, 45–48,
57–67, 69–73, 76, 94
E Pump-probe, 10, 11, 40–42, 46–48, 61, 62,
Emission spectroscopy, 15, 19, 20, 36, 93, 66, 68–71, 73, 75, 86, 151
94, 139

S
Saturable absorption, 85, 88, 94, 136, 137,
F 154
Floquet engineering, 76 Second harmonic generation, 6, 83–85, 87,
Four wave mixing, 86, 95, 100, 103–105, 95, 106, 108, 110, 127, 129, 154
110, 148 Stimulated Raman scattering, 83, 99, 141
Stimulated X-ray emission spectroscopy,
139, 140
Sum frequency generation, 6, 7, 83, 86, 87,
M 106, 110, 130, 131
Multi-dimensional spectroscopy, 105, 147 Susceptibility, 3–5, 7, 8, 15–18, 85, 87, 88,
Multiphoton absorption, 86, 95, 110, 112 95, 106, 111, 127, 141

© The Editor(s) (if applicable) and The Author(s), under exclusive license 159
to Springer Nature Singapore Pte Ltd. 2023
I. Matsuda and R. Arafune (eds.), Nonlinear X-Ray Spectroscopy for Materials Science,
Springer Series in Optical Sciences 246, https://doi.org/10.1007/978-981-99-6714-8
160 Index

Synchrotron radiation, 15, 19–22, 35, 40, V

43, 47, 51, 90, 119, 129 Vacuum nonlinear, 152, 153

T X
Time-resolved, 10, 12, 15, 20, 22, 34, X-ray free electron laser, 2, 15, 20–22, 40,
39–42, 44–48, 57, 59, 61, 62, 66, 67, 51, 83, 89, 91, 93, 94, 154
72, 73 X-ray spectroscopy, 12, 15, 18, 28, 39, 50,
Two-photon absorption, 86, 88, 95–97, 132, 52, 92, 94, 99, 111, 112, 119, 122,
134, 135, 141 132, 147, 148, 151, 152, 154, 155