NanoScience and Technology

Hilmi Ünlü
Norman J.M. Horing
Jaroslaw Dabowski Editors

Low-Dimensional
and Nanostructured
Materials and
Devices
Properties, Synthesis, Characterization,
Modelling and Applications
NanoScience and Technology

Series editors
Phaedon Avouris, Yorktown Heights, USA
Bharat Bhushan, Columbus, USA
Dieter Bimberg, Berlin, Germany
Klaus von Klitzing, Stuttgart, Germany
Cun-Zheng Ning, Tempe, USA
Roland Wiesendanger, Hamburg, Germany
The series NanoScience and Technology is focused on the fascinating nano-world,
mesoscopic physics, analysis with atomic resolution, nano and quantum-effect
devices, nanomechanics and atomic-scale processes. All the basic aspects and
technology-oriented developments in this emerging discipline are covered by
comprehensive and timely books. The series constitutes a survey of the relevant
special topics, which are presented by leading experts in the field. These books will
appeal to researchers, engineers, and advanced students.

More information about this series at http://www.springer.com/series/3705

Hilmi Ünlü Norman J.M. Horing
•

Jaroslaw Dabowski
Editors

Low-Dimensional
and Nanostructured
Materials and Devices
Properties, Synthesis, Characterization,
Modelling and Applications

123
Editors
Hilmi Ünlü Jaroslaw Dabowski
Department of Physics Engineering, Faculty Innovations for High Performance
of Science and Letters Microelectronics (IHP) GmbH
Istanbul Technical University Frankfurt
Istanbul Germany
Turkey

Norman J.M. Horing

Department of Physics and Engineering
Physics
Stevens Institute of Technology
Hoboken, NJ
USA

ISSN 1434-4904 ISSN 2197-7127 (electronic)

NanoScience and Technology
ISBN 978-3-319-25338-1 ISBN 978-3-319-25340-4 (eBook)
DOI 10.1007/978-3-319-25340-4

Library of Congress Control Number: 2015953262

Springer Cham Heidelberg New York Dordrecht London

© Springer International Publishing Switzerland 2016
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made.

Printed on acid-free paper

Springer International Publishing AG Switzerland is part of Springer Science+Business Media

(www.springer.com)
My work in this volume is dedicated to my
wife, Yeliz and my children, Melek,
Merve and Miray, in deep appreciation for
their support and encouragements.

—Hilmi Ünlü

My work in this volume is dedicated to my

wife, Jerilyn, in deep appreciation
for her enduring forbearance and love.

—Norman J.M. Horing

Preface

Since the invention of the first transistor in 1949, there has been a continuous
avalanche of growth and development in the science and technology of semicon-
ductor materials and device applications. The thrust of the advances has involved
the steady lowering of the dimensions in which confined charge carriers can move,
ultimately reaching the nano-world. The term nano, a prefix derived from the Greek
word for dwarf (νᾶνος), is used to indicate sizes in the range of about one to a
hundred nanometers, or 10–9 to 10–7 m. By 1970, systems thin enough to be
regarded as two dimensional (for example, semiconductor inversion layers, etc.)
were being analyzed, and these were rapidly followed by one-dimensional quantum
wires and zero-dimensional quantum dots. In conjunction with technology, “nano”
was used for the first time by Norio Taniguchi in 1974. As he predicted,
nanoscience and nanotechnology emerged strongly as research fields in the 1980s:
the scanning tunneling microscope was invented by Gerd Binnig and Heinrich
Rohrer in 1981 and the first carbon nanomaterial, C60 buckyballs, was discovered
by Richard Smalley, Robert Curl, Harold Kroto, James Heath, and Sean O'Brien in
1985, followed by carbon nanotubes in 1991 found by Sumio Iijima. With the
discovery of graphene in 2004 by Andre Geim and Kostya Novoselov, the world
was presented with a two-dimensional device-friendly carbon material just a single
atom thick, and this was followed by similar materials including silicene, ger-
manene, stanene, metal dichalcogenides, and topological insulators.
The requisite support basis of nanotechnology, laden with promising develop-
ments tantamount to a new industrial revolution, has been a qualitatively reliable
understanding of the underlying physics of semiconducting materials accompanied
by quantitatively precise predictions of device performance. This has led to new
concepts and techniques of semiconductor growth that have facilitated the emer-
gence of a generation of advanced devices with more complex functionality and
much higher densities for electronic, computational, and optical applications.
Advances in the growth of semiconductor thin films of differing structural, elec-
tronic, and optical properties, and the diminution of layer thickness approaching
atomic dimensions, have provided new opportunities for fundamental scientific

vii
viii Preface

studies and technological applications of semiconductors in new devices.

Moreover, contemporary fabrication technologies have made it possible to reduce
device dimensions to the point where size effects must be properly described
quantum mechanically in order to reliably predict the potential and performance of
low-dimensional semiconductor systems for electronic and optical applications.
Advances in single and multilayer thin film growth technologies (e.g., molecular
beam epitaxy (MBE), and metal organic chemical vapor deposition (MOCVD)),
have made it possible to produce semiconductor heterostructure thin films based on
group IV–IV, III–V and II–VI semiconductors in binary/binary and alloy/binary
forms, having engineered electronic and optical properties that are not available in
nature. One can now control the alloy composition and doping in ternary, quater-
nary and pentanary IV–IV, III–V and II–VI semiconductors over atomic distances.
These growth techniques, taken jointly with advanced characterization and fabri-
cation techniques have facilitated the development of a number of high perfor-
mance devices for fast signal processing, as well as some novel structures that are of
fundamental interest to solid state scientists and device engineers. The epitaxial
layers are so thin that quantum mechanical effects govern the operation of such
heterostructure devices. These materials, coupled with device design, have enabled
the emergence of novel devices in which signals propagate faster, promising to
replace silicon with compound semiconductors having much higher electron
mobility and velocity. Low-dimensional bipolar and unipolar compound semi-
conductor devices (e.g., GaAs based heterojunction bipolar transistors (HBTs),
modulation doped field effect transistors (MODFETs), light-emitting diodes, lasers,
etc.) operate much faster than conventional homostructure silicon devices (e.g.,
bipolar junction transistors (BJTs), metal oxide field effect transistors (MOSFETs),
etc.), leading to a many-thousand-fold increase in speed, which is of crucial
importance to the electronic and optical communication and computer industries.
This book describes recent scientific and technological developments of
low-dimensional nanomaterials in over 20 review and research articles. It begins
with metrology and methodology, with simple carbon nanomaterials (nanotubes
and graphene), and with metal oxide thin films, nanowires, and quantum dots. More
complex issues associated with the environment and with energy production and
storage follow. Furthermore, important achievements in materials pertinent to the
fields of biology and medicine are also reviewed, exhibiting an outstanding con-
fluence of basic physical science and vital human endeavor. Finally, after a brief
excursion into device physics, new and interesting developments relating to
quantum computing are addressed.
In recognition of the vastly important role that low-dimensional semiconductor
systems are now playing in all fields of science and technology and the even greater
role that they are poised to play in high-functionality devices for electronic, com-
putational, optical, biological, and medical systems that will support much greater
economic development and human well-being, much of this book has been pre-
pared in a way that permits access to the various subjects by readers who are not
Preface ix

experts at the outset. This is to say that this book will be helpful to graduate
students and young scientists who want to develop an understanding of the subject;
and it will also be informative to seasoned scientists and engineers who are
knowledgeable in some areas and wish to broaden their perspective in others in this
multidisciplinary field.

Istanbul Hilmi Ünlü

Hoboken Norman J.M. Horing
Frankfurt Jaroslaw Dabowski
Contents

1 Modelling of Heterostructures for Low Dimensional Devices. . ... 1

H. Hakan Gürel, Özden Akıncı and Hilmi Ünlü
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 2
1.2 Issues in Modelling of Electronic Structure
in Heterostructures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2.1 Interface Strain Effects in Heterostructures . . . . . . . . . 6
1.2.2 Composition Effects in Heterostructures. . . . . . . . . . . 10
1.3 Semiempirical Tight Binding Modeling of Heterostructures . . . 12
1.3.1 Semiempirical Sp3 Tight Binding Modeling . . . . . . . . 13
1.3.2 Semiempirical Sp3s* Tight Binding Modeling . . . . . . 21
1.3.3 Semiempirical Sp3d5s* Tight Binding Modeling . . . . . 27
1.3.4 Semiempirical Sp3d5 Tight Binding Modeling . . . . . . 30
1.4 Density Functional Theory Modelling of Heterostructures . . . . 32
1.5 Modeling of Band Offsets in Heterostructures . . . . . . . . . . . . 40
1.6 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

2 Aspects of the Modeling of Low Dimensional

Quantum Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 49
Norman J.M. Horing
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 49
2.2 3D Harmonic Oscillator with a Superposed
Quantum Dot (Fig. 2.1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.3 Double Quantum Dot Systems in Low Dimensions . . . . . . . . . 53
2.4 Quantum Wire on a 2D Sheet in a Uniform Magnetic Field . . . 56
2.5 Inverse Dielectric Function of a Periodic Superlattice . . . . . . . 59
2.6 An Impenetrable Barrier: Quantum Particle Dynamics
in a Highly Singular 1D-Potential . . . . . . . . . . . . . . . . . . . .. 64
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 70

xi
xii Contents

3 Wave Propagation and Diffraction Through a Subwavelength

Nano-Hole in a 2D Plasmonic Screen . . . . . . . . . . . . . . . . . . . ... 73
Norman J.M. Horing and Désiré Miessein
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 73
3.2 Scalar Green’s Function for a 2D Plasmonic Layer
with a Nano-Hole Aperture Embedded in a 3D Bulk
Host Medium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 74
3.3 Scalar Field Response of a Perforated 2D Plasmonic
Layer to an Incident Wave U0 ðr; xÞ: Transmission . . . . . . ... 79
3.4 Numerical Analysis of Helmholtz Scalar Wave
Propagation in the Vicinity of a Perforated Plasmonic
Layer with a Nano-Hole . . . . . . . . . . . . . . . . . . . . . . . . ... 86
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 103
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 104

4 The Challenge to Develop Metrology at the Nanoscale . . . . ..... 105

R. Ince
4.1 Metrology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.2 Nanotechnology Generations, Definitions, and Visions . . . . . . 108
4.3 Nanotechnology Research Drives . . . . . . . . . . . . . . . . . . . . . 109
4.3.1 The Semiconductor Industry. . . . . . . . . . . . . . . . . . . 109
4.3.2 The Healthcare Industry. . . . . . . . . . . . . . . . . . . . . . 110
4.4 Nanotechnology Artefacts . . . . . . . . . . . . . . . . . . . . . . . . . . 111
4.5 Nanomaterials Measurement Challenges. . . . . . . . . . . . . . . . . 112
4.6 Nanomaterials Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . 113
4.6.1 Spectroscopic Ellipsometry (SE) . . . . . . . . . . . . . . . . 114
4.6.2 Metrological AFM . . . . . . . . . . . . . . . . . . . . . . . . . 115
4.6.3 Scanning Electron Microscopy (SEM) . . . . . . . . . . . . 117
4.7 Other Methods for Dimensional Nanometrology . . . . . . . . . . . 118
4.7.1 X-ray Interferometry (XRI) . . . . . . . . . . . . . . . . . . . 119
4.7.2 Small Angle X-ray Scattering
Diffractometer (SAXS) . . . . . . . . . . . . . . . . . . . . . . 120
4.7.3 Electron/X-ray Diffraction . . . . . . . . . . . . . . . . . . . . 121
4.7.4 Raman Spectroscopy as a Nanometrology Tool . . . . . 121
4.8 The Challenge for Nanometrology . . . . . . . . . . . . . . . . . . . . 122
4.9 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

5 Terahertz Devices and Systems for the Spectroscopic Analysis

of Biomolecules—“Complexity Great and Small” . . . . . . . . . . ... 131
Anthony John Vickers and David Crompton
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 131
5.2 THz Time Domain Spectroscopy . . . . . . . . . . . . . . . . . . ... 132
5.3 Small Complexity—Simple Polar Liquids . . . . . . . . . . . . ... 134
Contents xiii

5.4 Great Complexity—Proteins . . . . . . . . . . . . . . . . . . . . . . . . . 138

5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

6 Recent Progress in XAFS Study for Semiconducting

Thin Films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 149
Takafumi Miyanaga and Takashi Azuhata
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
6.2 XAFS Spectroscopy: Basic Theory and Measurements . . . . . . 150
6.2.1 Basic Theory of XAFS . . . . . . . . . . . . . . . . . . . . . . 150
6.2.2 Measurements of XAFS. . . . . . . . . . . . . . . . . . . . . . 154
6.3 Application to Semiconducting Thin Films. . . . . . . . . . . . . . . 155
6.3.1 c-Plane InGaN SQWs . . . . . . . . . . . . . . . . . . . . . . . 155
6.3.2 c-Plane InGaN MQW . . . . . . . . . . . . . . . . . . . . . . . 158
6.3.3 m-Plane InGaN Thin Film . . . . . . . . . . . . . . . . . . . . 159
6.3.4 m-Plane AlGaN Thin Films . . . . . . . . . . . . . . . . . . . 161
6.3.5 c-Plane MgZnO Thin Film. . . . . . . . . . . . . . . . . . . . 164
6.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168

7 Pulsed-Laser Generation of Nanostructures . . . . . . . . . . . . . . . .. 171

Selcuk Akturk
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 171
7.2 Self-formation of Nanostructures Through
Pulsed-Laser Ablation . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 172
7.3 Formation of Nanoscale Patterns by Femtosecond
Laser Ablation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 173
7.3.1 General Principles . . . . . . . . . . . . . . . . . . . . . . . . .. 173
7.3.2 Femtosecond Laser Nanomachining on Thin Films
with Bessel Beams . . . . . . . . . . . . . . . . . . . . . . . . . 175
7.3.3 Experimental Results. . . . . . . . . . . . . . . . . . . . . . . . 177
7.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179

8 Graphene for Silicon Microelectronics: Ab Initio Modeling

of Graphene Nucleation and Growth . . . . . . . . . . . . . . . . . ..... 181
Jarek Dabrowski, Gunther Lippert and Grzegorz Lupina
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
8.2 Approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
8.3 Carbon on Graphene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
8.4 Silicon on Graphene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
8.5 Carbon on h-BN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
8.6 Carbon on Mica . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
8.7 Graphene Base Transistor . . . . . . . . . . . . . . . . . . . . . . . . . . 194
xiv Contents

8.8 Carbon on Germanium . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196

8.9 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . 201
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202

9 Recent Progress on Nonlocal Graphene/Surface Plasmons . . . . . . 205

Norman J.M. Horing, A. Iurov, G. Gumbs, A. Politano
and G. Chiarello
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
9.2 Nonlocal Dielectric Response of a Slab Plasma . . . . . . . . . . . 206
9.3 2D Plasma Coulomb-Coupled with a Slab Plasma. . . . . . . . . . 214
9.4 Numerical Results for Plasmon Dispersion:
Graphene Layers Interacting with Semi-infinite Conductor . . . . 217
9.5 Experimental Studies on Epitaxial Graphene. . . . . . . . . . . . . . 221
9.6 Influence of Adsorbed and Intercalated Atoms . . . . . . . . . . . . 227
9.7 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Appendix 1: Dynamic Nonlocal Polarization Function
for Free-Standing Graphene with no Bandgap;
Brief Summary of the Results Derived in [19, 20] . . . . . . . . . 229
Appendix 2: Dynamic Nonlocal Polarization Function for Graphene
with a Finite Energy Bandgap; Brief Summary
of the Results Derived in [18] . . . . . . . . . . . . . . . . . . . . . . . 231
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234

10 Semiconducting Carbon Nanotubes: Properties,

Characterization and Selected Applications . . . . . . . . . . . . ..... 239
Chiara Pintossi and Luigi Sangaletti
10.1 Introduction to Carbon Nanotubes. . . . . . . . . . . . . . . . . . . . . 239
10.2 CNTs Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
10.3 Carbon Nanotubes Applications . . . . . . . . . . . . . . . . . . . . . . 246
10.3.1 Selected Applications of Semiconducting CNTs . . . . . 246
10.3.2 CNTs for Photovoltaic Applications . . . . . . . . . . . . . 247
10.3.3 CNT Interaction with Gases: From Surface
Chemistry to Devices . . . . . . . . . . . . . . . . . . ..... 250
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 257

11 Effects of Charging and Perpendicular Electric Field

on Graphene Oxide. . . . . . . . . . . . . . . . . . . . . . . . . . . ........ 261
H. Hakan Gürel, M. Topsakal and S. Ciraci
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
11.2 Methodolgy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
11.3 Theoretical Investigations of Charged Nanosystems. . . . . . . . . 263
11.4 Interaction of H2O, OH, O and H with Graphene . . . . . . . . . . 267
11.4.1 Binding of H2O to Graphene . . . . . . . . . . . . . . . . . . 267
11.4.2 Binding of OH to Graphene . . . . . . . . . . . . . . . . . . . 267
Contents xv

11.4.3 Binding of O to Graphene . . . . . . . . . . . . ........ 269

11.4.4 Binding of H and H2 to Graphene. . . . . . . ........ 269
11.5 Effects of an Electric Field and Charging. . . . . . . . ........ 270
11.5.1 Effects of an Electric Field and Charging
on Adsorbed O. . . . . . . . . . . . . . . . . . . . ........ 270
11.5.2 Effects of an Electric Field and Charging
on Adsorbed OH . . . . . . . . . . . . . . . . . . . . . . . . . . 275
11.6 Desorption of Oxygen from GOX . . . . . . . . . . . . . . . . . . . . . 278
11.6.1 Formation of Oxygen Molecule . . . . . . . . . . . . . . . . 278
11.6.2 Interaction Between Adsorbed H and O. . . . . . . . . . . 280
11.6.3 Interaction Between Adsorbed H and OH . . . . . . . . . 282
11.6.4 Interaction Between Two OH Co-Adsorbed
in Close Proximity . . . . . . . . . . . . . . . . . ........ 284
11.7 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........ 286
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........ 288

12 Structural and Optical Properties of Tungsten Oxide

Based Thin Films and Nanofibers . . . . . . . . . . . . . . . . . . . . . ... 291
E.O. Zayim and A. Tabatabaei Mohseni
12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 291
12.1.1 Amorphous and Crystalline Tungsten
Oxide Based Nanomaterials . . . . . . . . . . . . . . . . ... 292
12.2 Tungsten Oxide Based Nanomaterials . . . . . . . . . . . . . . . ... 295
12.2.1 Tungsten Oxide Based Thin Films
and Mesoporous Thin Films . . . . . . . . . . . . . . . . ... 296
12.2.2 Tungsten Oxide Based Nanofibers and Nanowires ... 298
12.3 Chromogenic Properties and Applications
of Tungsten Oxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 303
12.3.1 Electrochromic Properties of Tungsten
Oxide Films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
12.3.2 Coloration Phenomena in WO3 . . . . . . . . . . . . . . . . . 305
12.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305

13 Electron Accumulation in InN Thin Films and Nanowires ...... 309

L. Colakerol Arslan and K.E. Smith
13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
13.2 Fermi Level Pinning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
13.3 Surface Electron Accumulation in InN Thin Films . . . . . . . . . 312
13.4 Surface Charge Accumulation on InN NWs . . . . . . . . . . . . . . 317
13.5 Control of Surface Electron Accumulation
in InN Nanowires . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 320
13.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 323
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 324
xvi Contents

14 Optical and Structural Properties of Quantum Dots . . . . . . . . ... 327

M.H. Yükselici, A. Aşıkoğlu Bozkurt, Ç. Allahverdi, Z. Nassar,
D. Bulut, B. Can Ömür, M.K. Torun and A.T. İnce
14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
14.2 CdSexS1−x Nanocrystals. . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
14.3 Investigation of Raman Spectroscopy for CdTe Thin Film . . . . 329
14.3.1 Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . 329
14.3.2 Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
14.3.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
14.3.4 Importance of the Subject . . . . . . . . . . . . . . . . . . . . 331
14.3.5 Section Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 331
14.4 Steady State Photoluminescence Spectroscopy . . . . . . . . . . . . 332
14.5 The Progression of Strain and Micro-electric Field
Dependent Urbach Energy with Deposition Time
in Chemical Bath Deposited CdS Thin Films . . . . . . . . . . . . . 335
14.5.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335
14.5.2 Modeling of the Urbach Tail . . . . . . . . . . . . . . . . . . 341
14.5.3 The Progression of Strain with Deposition Time . . . . . 343
14.5.4 Section Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . 345
14.6 In Situ Low Temperature Optical Absorption Spectroscopy . . . 346
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347

15 One-Dimensional Nano-structured Solar Cells . . . . . . . . . . . .... 351

H. Karaağaç, E. Peksu, E.U. Arici and M. Saif Islam
15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 352
15.2 One-Dimensional Nanostructures Based Solar
Cell Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 352
15.3 Synthesis of One-Dimensional Nanostructures . . . . . . . . .... 355
15.3.1 Solution-Based Synthesis of 1-D Nanostructures . .... 356
15.3.2 Vapor Phase-Based Synthesis
of 1-D Nanostructures . . . . . . . . . . . . . . . . . . . . . . . 359
15.4 Common Materials for 1-D Nano-Structured Solar Cells . . . . . 361
15.4.1 Silicon. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
15.4.2 Zinc-Oxide. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
15.4.3 Titanium-Dioxide . . . . . . . . . . . . . . . . . . . . . . . . . . 376
15.4.4 Carbon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380
15.5 Summary and Future Outlook. . . . . . . . . . . . . . . . . . . . . . . . 389
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391

16 Computational Studies of Bismuth-Doped Zinc

Oxide Nanowires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
Çetin Kılıç, Mehmet Aras and Sümeyra Güler-Kılıç
16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
Contents xvii

16.2 Computational Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . 403

16.2.1 Supercell Approach . . . . . . . . . . . . . . . . . . . . . . . . . 403
16.2.2 Defect Calculations . . . . . . . . . . . . . . . . . . . . . . . . . 404
16.2.3 Density- and Hybrid-Functional+U Calculations . . . . . 408
16.2.4 Computational Settings . . . . . . . . . . . . . . . . . . . . . . 411
16.3 Structure and Energetics of ZnO Nanowires . . . . . . . . . . . . . . 412
16.4 Defect Energetics and Transition Levels
in ZnO:Bi Nanowire . . . . . . . . . . . . . . . . . . . . . . . . . . .... 414
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 419

17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts .... 423

Juying Lei, Hong Li, Jinlong Zhang and Masakazu Anpo
17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 424
17.2 Phases of TiO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 425
17.2.1 Structure Properties of Rutile, Anatase
and Brookite . . . . . . . . . . . . . . . . . . . . . . . . . .... 425
17.2.2 Stability and Phase Transformation . . . . . . . . . . .... 426
17.2.3 Photocatalytic Activity of Rutile, Anatase
and Brookite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427
17.3 Synthesis of Mixed-Phase TiO2 Photocatalysts . . . . . . . . . . . . 429
17.3.1 Hydrothermal Method and Solvothermal Method . . . . 429
17.3.2 Microemulsion-mediated Solvothermal Method. . . . . . 434
17.3.3 Sol-Gel Method . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
17.3.4 Solvent Mixing and Calcination Method . . . . . . . . . . 437
17.3.5 High-Temperature Calcination Method . . . . . . . . . . . 438
17.4 Applications of Mixed-Phase TiO2 in Photocatalysis . . . . . . . . 441
17.4.1 Photocatalytic Hydrogen Production . . . . . . . . . . . . . 441
17.4.2 Photocatalytic Reduction of CO2 with Water
on Mixed-Phase TiO2 . . . . . . . . . . . . . . . . . . . .... 444
17.4.3 Photocatalytic Degradation of Organic Pollutants
on Mixed-Phase TiO2 . . . . . . . . . . . . . . . . . . . .... 446
17.5 Mechanism of the Enhanced Photocatalytic Activities
by the Mixed-Phase TiO2 Photocatalysis . . . . . . . . . . . . .... 451
17.6 Conclusion and Outlook . . . . . . . . . . . . . . . . . . . . . . . .... 456
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 457

18 Electrochemical Impedance Study on

Poly(Alkylenedioxy)Thiophene Nanostructures:
Solvent and Potential Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
A. Sezai Sarac and Aslı Gencturk
18.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
xviii Contents

18.2 Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 464

18.2.1 Chemicals . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 464
18.2.2 Preparation of Carbon Fiber
Microelectrode (CFME) . . . . . . . . . . . . . . . . . . . . . . 464
18.2.3 Apparatus and Procedure . . . . . . . . . . . . . . . . . . . . . 464
18.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
18.3.1 Electropolymerization of ProDOT-Me2 on CFME . . . . 465
18.3.2 FTIR-ATR Characterisation of PProDOT-Me2
Film on CFME. . . . . . . . . . . . . . . . . . . . . . . . . ... 466
18.3.3 Electrochemical Impedance Spectroscopy. . . . . . . ... 467
18.3.4 Potential Effect on EIS of PProDOT-Me2/CFME
Coated Electrode . . . . . . . . . . . . . . . . . . . . . . . ... 468
18.3.5 Electrolyte and Solvent Effects on ProDOT-Me2
Coated CFMEs. . . . . . . . . . . . . . . . . . . . . . . . . ... 469
18.3.6 Electrical Equivalent Circuit Modeling . . . . . . . . ... 472
18.3.7 Morphology of Nanoporous and Compact
Conductive Polymer Coatings on Carbon Fiber. . . ... 474
18.4 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 474
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 474

19 Application of Nanoporous Zeolites for the Removal

of Ammonium from Wastewaters: A Review . . . . . . . . . . . . .... 477
Mustafa Turan
19.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
19.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480
19.2.1 Batch Adsorption Studies. . . . . . . . . . . . . . . . . . . . . 480
19.2.2 Column Adsorption Studies . . . . . . . . . . . . . . . . . . . 482
19.2.3 Modification of Zeolite . . . . . . . . . . . . . . . . . . . . . . 484
19.2.4 Synthesized Zeolites . . . . . . . . . . . . . . . . . . . . . . . . 485
19.2.5 Regeneration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
19.3 Nanoporous Zeolites for Ammonium Removal
from Wastewaters . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 488
19.3.1 Ammonium Adsorption Capacities
and Other Parameters for Zeolites
in the Batch Systems. . . . . . . . . . . . . . . . . . . . .... 488
19.3.2 Ammonium Removal from Wastewaters
in the Batch and Column Systems. . . . . . . . . . . .... 495
19.3.3 Ammonium Removal from Wastewaters
in the Fixed Bed Systems . . . . . . . . . . . . . . . . .... 497
19.3.4 Ammonium Removal for Wastewater Treatment
Systems Combined with Zeolites . . . . . . . . . . . .... 498
19.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 501
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 501
Contents xix

20 Synthesis and Biological Applications of Quantum Dots . . . . .... 505

Manolya Kukut Hatipoglu, Seda Kelestemur and Mustafa Culha
20.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 505
20.2 Synthesis of CdSe and CdSe/ZnS QDs . . . . . . . . . . . . . .... 506
20.3 Design of CdSe and CdSe/ZnS QDs for Biological
Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 508
20.3.1 Ligand Exchange Process . . . . . . . . . . . . . . . . . . . . 509
20.3.2 Silanization Process. . . . . . . . . . . . . . . . . . . . . . . . . 509
20.3.3 Amphiphilic Molecules . . . . . . . . . . . . . . . . . . . . . . 510
20.3.4 PEG and Phospholipid Micelle . . . . . . . . . . . . . . . . . 511
20.3.5 Biomacromolecules . . . . . . . . . . . . . . . . . . . . . . . . . 512
20.4 Biological Applications of QDs . . . . . . . . . . . . . . . . . . . . . . 514
20.4.1 In Vitro Targeting with Antibody Conjugation . . . . . . 515
20.4.2 In Vitro Targeting with Peptide Conjugation . . . . . . . 517
20.4.3 In Vitro Targeting with Small
Molecule Conjugation . . . . . . . . . . . . . . . . . . . .... 518
20.4.4 In Vivo Applications of QDs . . . . . . . . . . . . . . .... 520
20.4.5 In Vivo Vascular Imaging . . . . . . . . . . . . . . . . .... 522
20.4.6 Förster Resonance Energy Transfer (FRET)
Based Applications . . . . . . . . . . . . . . . . . . . . . .... 523
20.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 525
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 526

21 Bionanotechnology: Lessons from Nature for Better

Material Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 535
F.N. Kök
21.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535
21.2 Biomineralization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 536
21.3 Biomimetic Proteins: Receptors, Catalysts, Channels . . . . . . . . 538
21.4 Optics/Biophotonics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 539
21.5 Natural Adhesives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 540
21.6 Biointerfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 541
21.6.1 Self-cleaning Surfaces . . . . . . . . . . . . . . . . . . . . . . . 542
21.6.2 Bioinspired Interfaces for Better Biocompatibility . . . . 543
21.7 Biomimetic Membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544
21.7.1 Cell Membrane Mimics . . . . . . . . . . . . . . . . . . . . . . 544
21.7.2 Membranes for Water Treatment. . . . . . . . . . . . . . . . 546
21.8 Hints from Nature for Endurance . . . . . . . . . . . . . . . . . . . . . 546
21.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
xx Contents

22 Quantum Dots in Bionanotechnology and Medical Sciences:

Power of the Small . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 555
İ. Ergal and A.T. Akarsubasi
22.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
22.2 Types and Characteristics of QDs . . . . . . . . . . . . . . . . . . . . . 556
22.3 Advantages and Disadvantages of QDs . . . . . . . . . . . . . . . . . 557
22.4 Synthesis of QDs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 558
22.5 Toxicity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
22.6 Surface Modification and Functionalization . . . . . . . . . . . . . . 560
22.6.1 Surface Coatings to Minimize Hydrodynamic Size . . . 561
22.7 Biocompatibility in QDs (Bioconjugation) . . . . . . . . . . . . . . . 562
22.8 Next Generation QDs (Silica, Carbon, Metal Nanocluster) . . . . 564
22.8.1 Metal Nanoclusters . . . . . . . . . . . . . . . . . . . . . . . . . 564
22.8.2 Carbon Dots (C-Dots/GQDs) . . . . . . . . . . . . . . . . . . 564
22.8.3 Silicon Dots (Si QDs) . . . . . . . . . . . . . . . . . . . . . . . 565
22.9 Applications of QDs in Bionanotechnology
and Nanomedicine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
22.9.1 Cell Labeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
22.9.2 In Vitro Imaging. . . . . . . . . . . . . . . . . . . . . . . . . . . 566
22.9.3 In Vivo Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . 568
22.10 Dual-Modality Imaging with QDs . . . . . . . . . . . . . . . . . . . . . 571
22.10.1 Fluorescence/MRI . . . . . . . . . . . . . . . . . . . . . . . . . . 571
22.10.2 Fluorescence/CT . . . . . . . . . . . . . . . . . . . . . . . . . . . 571
22.10.3 Fluorescence/PET . . . . . . . . . . . . . . . . . . . . . . . . . . 572
22.11 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 572
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 573

23 Nanomedicine . . . . . . . . . . . . . . . . . . . . . . . . . .............. 579

Eda Tahir Turanlı and Elif Everest
23.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 579
23.2 Nanopharmaceuticals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 581
23.3 Diagnostic and Theranostic Nanomedicine . . . . . . . . . . . . . . . 583
23.4 Ethics and Regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584
23.5 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 585
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 586

24 Microfluidics and Its Applications in Bionanotechnology ....... 589

Z.P. Çakar and B. Sönmez
24.1 Introduction to Bionanotechnology and Microfluidics . . . . . . . 589
24.2 Microfluidic PCR Applications . . . . . . . . . . . . . . . . . . . . . . . 592
24.3 Microfluidic DNA Microarray Systems . . . . . . . . . . . . . . . . . 594
24.4 Microfluidic Applications in Electrophoresis. . . . . . . . . . . . . . 595
24.5 Microfluidic Bioreactors . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
24.6 Monitoring Microbial Behaviour by Microfluidics. . . . . . . . . . 599
Contents xxi

24.7 The Use and Potential of Microfluidics in Microbial

Strain Development. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 600
24.8 Microfluidic Applications in Single Cell Studies . . . . . . . . . . . 602
24.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 602
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603

25 Non-Markovian Dynamics of Qubit Systems: Quantum-State

Diffusion Equations Versus Master Equations . . . . . . . . . . . .... 609
Yusui Chen and Ting Yu
25.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609
25.2 Non-Markovian Quantum-State Diffusion Approach . . . . . . . . 610
25.3 Non-Markovian Master Equation Approach . . . . . . . . . . . . . . 615
25.4 Multiple-Qubit Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617
25.4.1 Two-Qubit Systems. . . . . . . . . . . . . . . . . . . . . . . . . 619
25.4.2 Three-Qubit Systems . . . . . . . . . . . . . . . . . . . . . . . . 623
25.4.3 A Note on General N-Qubit Systems. . . . . . . . . . . . . 627
25.5 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 628
Appendix 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 628
Appendix 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 630
Appendix 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633

26 Computing with Emerging Nanotechnologies . . . . . . .......... 635

M. Altun
26.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
26.2 Computing with Nano-crossbar Arrays . . . . . . . . . . . . . . . . . 636
26.2.1 Implementing Boolean Logic Functions . . . . . . . . . . . 638
26.2.2 Defect Tolerance. . . . . . . . . . . . . . . . . . . . . . . . . . . 643
26.2.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . 649
26.3 Stochastic Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 652
26.3.1 Reducing Error Rates . . . . . . . . . . . . . . . . . . . . . . . 654
26.3.2 Error Free Stochastic Computing . . . . . . . . . . . . . . . 657
26.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 661
Contributors

A.T. Akarsubasi Molecular Biology and Genetics Department, Istanbul Technical

University, Maslak, Istanbul, Turkey
Özden Akıncı Computational Science and Engineering Programme , Informatics
Institute, Istanbul Technical University, Maslak, Istanbul, Turkey
Selcuk Akturk Department of Physics, Istanbul Technical University, Maslak,
Istanbul, Turkey
Ç. Allahverdi Faculty of Engineering, Electrical-Electronics Engineering
Department, Toros University, Mersin, Turkey
M. Altun Electronics and Communication Engineering Department, Istanbul
Technical University, Maslak, Istanbul, Turkey
Masakazu Anpo Department of Applied Chemistry, Graduate School of
Engineering, Osaka Prefecture University, Sakai, Osaka, Japan
Mehmet Aras Department of Physics, Gebze Technical University, Gebze,
Kocaeli, Turkey
E.U. Arici Energy Institute, Istanbul Technical University, Maslak, Istanbul,
Turkey
A. Aşıkoğlu Bozkurt Faculty of Science and Letters, Department of Physics,
Yıldız Technical University, Istanbul, Turkey
Takashi Azuhata Department of Advanced Physics, Hirosaki University,
Hirosaki, Aomori, Japan
D. Bulut Department of Physics, İnönü Mah. Kayışdağı Cad., Yeditepe
University, Ataşehir, İstanbul, Turkey

xxiii
xxiv Contributors

Z.P. Çakar Faculty of Science and Letters, Department of Molecular Biology and
Genetics, Istanbul Technical University, Maslak, Istanbul, Turkey; Dr. Orhan
Öcalgiray Molecular Biology, Biotechnology and Genetics Research Center
(ITU-MOBGAM), Istanbul Technical University, Maslak, Istanbul, Turkey
B. Can Ömür Faculty of Science and Letters, Department of Physics, Yıldız
Technical University, Istanbul, Turkey
Yusui Chen Department of Physics and Engineering Physics, Stevens Institute of
Technology, Hoboken, NJ, USA
G. Chiarello Dipartimento di Fisica, Universitá degli Studi della Calabria, Rende,
Italy
S. Ciraci Department of Physics, Bilkent University, Ankara, Turkey
L. Colakerol Arslan Department of Physics, Gebze Technical University,
Kocaeli, Turkey
David Crompton School of Computer Science and Electronic Engineering,
University of Essex, Colchester, Essex, UK
Mustafa Culha Genetics and Bioengineering Department, Yeditepe University,
Faculty of Engineering and Architecture, Istanbul, Turkey
Jarek Dabrowski IHP, Frankfurt(Oder), Germany
İ. Ergal Molecular Biology and Genetics Department, İstanbul Technical
University, Maslak, İstanbul, Turkey
Elif Everest Graduate School of Science Engineering and Technology, Molecular
Biology-Genetics and Biotechnology Programme, Dr. Orhan Öcalgiray Molecular
Biology and Genetics Research Centre, Istanbul Technical University, Istanbul,
Turkey
Aslı Gencturk Department of Chemistry, Polymer Science and Technology,
Istanbul Technical University, Istanbul, Turkey
Sümeyra Güler-Kılıç Department of Physics, Gebze Technical University,
Gebze, Kocaeli, Turkey
G. Gumbs Department of Physics, Hunter College, Cuny, NY, USA; Donostia
International Physics Center (DIPC), San Sebastian, Basque Country, Spain
H. Hakan Gürel Technology Faculty, Information Systems Engineering
Department, Computational Science and Engineering Programme, Kocaeli
University, Kocaeli, Turkey
Manolya Kukut Hatipoglu Faculty of Engineering and Architecture, Genetics
and Bioengineering Department, Yeditepe University, Istanbul, Turkey
Contributors xxv

Norman J.M. Horing Department of Physics and Engineering Physics, Stevens

Institute of Technology, Hoboken, NJ, USA
R. Ince Department of Physics, Yeditepe University, Istanbul, Turkey
A.T. İnce Department of Physics, İnönüMah. Kayışdağı Cad., Yeditepe
University, Ataşehir, Istanbul, Turkey
A. Iurov Center for High Technology Materials, University of New Mexico,
Albuquerque, NM, USA
H. Karaağaç Department of Physics, Istanbul Technical University, Maslak,
Istanbul, Turkey
Seda Kelestemur Faculty of Engineering and Architecture, Genetics and
Bioengineering Department, Yeditepe University, Istanbul, Turkey; Istanbul
Medipol University Regenerative and Restorative Medicine Research Center,
Istanbul, Turkey
Çetin Kılıç Department of Physics, Gebze Technical University, Gebze, Kocaeli,
Turkey
F.N. Kök Molecular Biology and Genetics Department, Istanbul Technical
University, Maslak, Istanbul, Turkey
Juying Lei State Environmental Protection Key Laboratory of Environmental Risk
Assessment and Control on Chemical Process, East China University of Science
and Technology, Shanghai, People’s Republic of China
Hong Li Key Lab for Advanced Materials and Institute of Fine Chemicals, East
China University of Science and Technology, Shanghai, People’s Republic of
China
Gunther Lippert IHP, Frankfurt(Oder), Germany
Grzegorz Lupina IHP, Frankfurt(Oder), Germany
Désiré Miessein Department of Physics and Engineering Physics, Stevens Institute
of Technology, Hoboken, NJ, USA
Takafumi Miyanaga Department of Advanced Physics, Hirosaki University,
Hirosaki, Aomori, Japan
Z. Nassar Faculty of Science and Letters, Department of Physics, Yıldız Technical
University, Istanbul, Turkey
E. Peksu Department of Physics, Istanbul Technical University, Maslak, Istanbul,
Turkey
Chiara Pintossi I-LAMP, Interdisciplinary Laboratory for Advanced Materials
Physics, Università Cattolica del Sacro Cuore, Brescia, Italy
xxvi Contributors

A. Politano Dipartimento di Fisica, Universitá degli Studi della Calabria, Rende,

Italy
M. Saif Islam Department of Electrical and Computer Engineering, University of
California at Davis, Davis, CA, USA
Luigi Sangaletti I-LAMP, Interdisciplinary Laboratory for Advanced Materials
Physics, Università Cattolica del Sacro Cuore, Brescia, Italy
A. Sezai Sarac Department of Chemistry, Polymer Science and Technology,
Istanbul Technical University, Istanbul, Turkey
K.E. Smith Department of Physics, Boston University, Boston, MA, USA; School
of Chemical Sciences, The MacDiarmid Institute for Advanced Materials and
Nanotechnology, The University of Auckland, Auckland, New Zealand
B. Sönmez Faculty of Science and Letters, Department of Molecular Biology and
Genetics, Istanbul Technical University, Maslak, Istanbul, Turkey; Dr. Orhan
Öcalgiray Molecular Biology, Biotechnology and Genetics Research Center
(ITU-MOBGAM), Istanbul Technical University, Maslak, Istanbul, Turkey
A. Tabatabaei Mohseni Nanoscience and Nanotechnology Program, Istanbul
Technical University, Istanbul, Turkey
M. Topsakal UNAM-National Nanotechnology Research Center, Bilkent
University, Ankara, Turkey
M.K. Torun Department of Physics, İnönüMah. Kayışdağı Cad., Yeditepe
University, Ataşehir, İstanbul, Turkey
Mustafa Turan Department of Environmental Engineering, İstanbul Technical
University, İstanbul, Turkey
Eda Tahir Turanlı Molecular Biology and Genetics Department, Istanbul
Technical University, Istanbul, Turkey; Graduate School of Science Engineering
and Technology, Molecular Biology-Genetics and Biotechnology Programme,
Dr. Orhan Öcalgiray Molecular Biology and Genetics Research Centre, Istanbul
Technical University, Istanbul, Turkey
Hilmi Ünlü Computational Science and Engineering Programme , Informatics
Institute, İstanbul Technical University, Maslak, Istanbul, Turkey; Faculty of
Science and Letters, Department of Physics, İstanbul Technical University, Maslak,
Istanbul, Turkey
Anthony John Vickers School of Computer Science and Electronic Engineering,
University of Essex, Colchester, Essex, UK
Ting Yu Department of Physics and Engineering Physics, Stevens Institute of
Technology, Hoboken, NJ, USA
Contributors xxvii

M.H. Yükselici Faculty of Science and Letters, Department of Physics, Yıldız

Technical University, Istanbul, Turkey
E.O. Zayim Physics Department, Istanbul Technical University, Istanbul, Turkey
Jinlong Zhang Key Lab for Advanced Materials and Institute of Fine Chemicals,
East China University of Science and Technology, Shanghai, People’s Republic of
China
Chapter 1
Modelling of Heterostructures for Low
Dimensional Devices

H. Hakan Gürel, Özden Akıncı and Hilmi Ünlü

Abstract Advancement in the theoretical understanding and experimental devel-

opment of the science and technology of low dimensional electronic and optical
devices requires qualitatively reliable and quantitatively precise theoretical mod-
elling of the structural, electronic and optical properties of semiconducting mate-
rials and their heterostructures to predict their potential profiles. In this chapter, we
review the calculation techniques of electronic band structures of III–V and II–VI
compounds and their heterostructures. We focus on the semiempirical tight binding
theory (with sp3, sp3s*, sp3d5s* and sp3d5 orbital sets) and density functional theory
(DFT), which, in turn, employs the modified Becke-Johnson exchange-correlation
potential with a local density approximation (DFT-MBJLDA). We conclude that
the density functional theory and semiempirical tight binding theory can easily be
employed in relation to charge transport in heterostructure devices as well as in the
accurate design and simulation of low dimensional semiconductor electronic and
optical devices.

H. Hakan Gürel
Computational Science and Engineering Programme, Information Systems Engineering
Department, Technology Faculty, Kocaeli University, 41380 Kocaeli, Turkey
e-mail: [email protected]
Ö. Akıncı
H. Ünlü (&)
Computational Science and Engineering Programme , Informatics Institute, İstanbul
Technical University, 34469 Maslak, Istanbul, Turkey
e-mail: [email protected]
Ö. Akıncı
e-mail: [email protected]
H. Ünlü
Department of Physics, Faculty of Science and Letters, İstanbul Technical University, 34469
Maslak, Istanbul, Turkey

© Springer International Publishing Switzerland 2016 1

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_1
2 H. Hakan Gürel et al.

1.1 Introduction

Progress in the theoretical understanding and experimental development of the sci-

ence and technology of semiconductor devices has grown rapidly ever since the
invention of first transistor in 1949 [1–6]. Advances in growing semiconductor thin
films having differing structural, electronic and optical properties with varying
composition and layer thickness (even approaching atomic dimensions) have pro-
vided new opportunities and challenges in basic scientific studies and in their device
applications in the electronics industry. The combination of advanced growth, char-
acterization and fabrication technologies has led to the production of an impressive
number of high performance semiconductor devices for fast signal processing and the
discovery of some novel structures that are of special interest to solid state scientists
and device engineers. One can now control the alloy composition and doping in
ternary and quaternary alloys based on group IV–IV, III–V and II–VI semiconductor
compounds over atomic distances, as low as tens of angstroms. These man-made
heterostructures, are easily grown as lattice mismatched substrates (e.g. growth of
GaAs on Si substrates [5, 6]) by modern crystal growth techniques including
molecular beam epitaxy (MBE) and metal organic chemical vapor deposition
(MOCVD). The epitaxial layers are so thin that quantum mechanical effects are
prominently realized. The resulting contemporary low dimensional semiconductor
devices (e.g., heterostructure bipolar transistors (HBTs) and modulation doped field
effect transistors (MODFETs), nanowires, quantum dots, etc.) are known to operate
much faster than conventional silicon devices (e.g., silicon bipolar junction transistors
(BJTs) and metal oxide semiconductor field effect transistors (MOSFETs)); this is
crucial for the electronic and optical communication and computer industries.
When the semiconductor composition changes abruptly across the interface
between two constituents, the difference in their energy bandgaps is accommodated
by the discontinuities in the conduction and valence bands [7, 8]. There are three
types of interface formation between a wide bandgap semiconductor and a narrow
bandgap semiconductor shown in Figs. 1.1, 1.2 and 1.3;
(i) Type I heterostructure: The bandgap of the barrier semiconductor overlaps
(straddling lineup) that of the well and equilibrium Fermi level is near the
middle of the bandgap on both sides, as shown in Fig. 1.1. Both an electron
and a hole tend to localize in the narrow bandgap quantum well. Such
semiconductor heterostructure is useful in optoelectronic applications such as
lasers, with both electrons and holes participate to device operation.
(ii) Type II heterostructure: The bandgap of the barrier semiconductor partially
overlaps (staggered lineup) that of the well and the equilibrium Fermi level is
close to the conduction band (or valence band) on one side while it is near the
middle of the bandgap on other side, as shown in Fig. 1.2. The potential
energy gradient tends to spatially separate electron and hole on different sides
of heterointerface. Large potential barrier in conduction (or valence) band at
heterointerface leads to a better electron (hole) confinement in field effect
transistors with higher electron (hole) concentration. Position of equilibrium
1 Modelling of Heterostructures for Low Dimensional Devices 3

Fig. 1.1 The schematic energy band diagram of Type I heterostructure: electrons and holes
localize in the narrow gap quantum well

Fig. 1.2 Schematic energy band diagram of Type II heterostructure: the potential energy gradient
tends to spatially separate electrons and holes on different sides of heterointerface

Fermi level determines the density of carrier confinement on either side of the
heterointerface. When equilibrium Fermi level is close to conduction band of
wide bandgap constituent (left) the density of electrons in its conduction band
is greater but when it is closer to the valence band of narrow bandgap con-
stituent (right) the number of holes is greater there.
(iii) Type III heterostructure: This is the extreme case of Type II band alignment.
The bandgap of the barrier does not overlap at all (broken gap lineup), as
shown in Fig. 1.3. The equilibrium Fermi level is well above the conduction
band minimum (or well below the valence band maximum) on one side while
it is near the middle of the bandgap on the other side. Position of equilibrium
Fermi level determines the carrier confinement on either side of the
heterointerface. When equilibrium Fermi level is close to conduction band of
wide bandgap constituent (left) the number of electrons in its conduction band
is greater but when it is closer to the valence band of narrow bandgap con-
stituent (right) the number of holes is greater there.
4 H. Hakan Gürel et al.

Fig. 1.3 Schematic energy band diagram of Type II heterostructure: the potential energy gradient
tends to spatially separate electrons and holes on different sides of heterointerface

Fig. 1.4 The schematic energy band diagram of a forward biased N-AlGaAs/p‐GaAs emitter/base
part a of Npn heterojunction bipolar transistor (HBT) in forward active mode

The impact of heterostructures on semiconductor device physics and technology

is twofold [1, 2]. The first impact of heterostructures is a high charge carrier injection
efficiency that can be obtained in an anisotype heterojuction bipolar transistor
(HBT), in which the charge carriers are flowing from a wide bandgap emitter (e.g.,
AlGaAs) to narrow bandgap base (e.g., GaAs) as shown in Fig. 1.4. When a Np
hetero-emitter is forward biased, the potential barrier blocking electron emission
from the emitter to the base is lowered and electrons injected from the emitter diffuse
across the base and are collected in the collector. Furthermore, holes injected from
the base into the emitter are blocked by the valence band offset as ΔEv/kT.
The second impact of heterostructures involves the confinement of charge car-
riers in a narrow bandgap two dimensional electron gas (2DEG) quantum well
structure to reduce their scattering by parent impurities in a doped wide bandgap
1 Modelling of Heterostructures for Low Dimensional Devices 5

Fig. 1.5 Schematic energy

band diagram of an
AlGaAs/GaAs 2DEG. The
triangular well in a narrow
bandgap GaAs well leads to
quantization of the allowed
energy levels E0 and E1 near
the heterointerface

barrier layer, as shown in Fig. 1.5. The charge carrier scattering that limits the high
speed character of MOSFETs can be minimized by replacing an
oxide/semiconductor junction with a widegap/narrowgap semiconductor hetero-
junction to separate the free charge carriers in a nominally undoped narrow bandgap
quantum well (e.g., i-GaAs) from their parent impurities in the doped wide bandgap
(e.g., N-AlGaAs) barrier layer, leading to a much higher mobility and a high current
from the drain to the source of a MODFET [2, 4].
Conduction and valence band offsets at the AlGaAs/GaAs interface lead to a
depletion of electrons from the wide-bandgap AlGaAs barrier and an accumulation of
electrons in the narrow-bandgap GaAs channel where they are quantized due to the
potential ‘notch’ at the interface. Quantized energy levels are formed in this well, with
the lowest level filled and the second level partially or fully filled by electrons,
depending on the electron density; the Fermi level lies between the first and the second
sub-bands. Free electrons confined in the channel conduct current parallel to the
hetero-interface between the source and the drain of a FET when an electric field is
applied parallel to the hetero-interface. As the wide bandgap and narrow bandgap
semiconductors reach thermal equilibrium, their Fermi levels align at the heteroint-
erface and the resultant energy band structure is as shown in Fig. 1.4 for a Np
heterostructure bipolar device and in Fig. 1.5 for a unipolar heterostructure device,
respectively.

1.2 Issues in Modelling of Electronic Structure

in Heterostructures

Ever since the invention of the first transistor in 1949, device scientists and engi-
neers have witnessed an amazing growth in semiconductor science and technology.
Advances in the growth of semiconductor thin films of different structural,
6 H. Hakan Gürel et al.

electronic and optical properties and with layer thickness approaching atomic
dimensions has provided new opportunities in fundamental science and technology
of semiconductors for device applications. Furthermore, contemporary fabrication
technologies have further made it possible to reduce the device dimensions to the
point where quantum size effects must be described in order to realistically describe
the operation and reliably predict the potential and performance of low dimensional
semiconductor devices for electronic and optical applications. Such advancement
could not have been possible without a qualitatively reliable understanding of the
basic physics of semiconducting materials and quantitatively precise potential
predictions and performance of devices, leading to new concepts in the semicon-
ductor growth that allowed previously many unknown devices with more complex
functionality and much higher densities for electronic and optical applications.
In order to emphasize the importance of heterostructure in low dimensional
semiconductor systems it is necessary to understand the interface formation and
modelling of the electronic structure of heterostructure constituents. The composi-
tion variation in alloy constituent and lattice mismatch between two semiconductors
and thermal expansion of lattice constant with the growth temperature will cause
strain across the interface that will modify the electronic properties of both materials
and consequently, the energy of the moving charge carriers across the interface that
influence the device performance [7, 8]. Qualitatively reliable and quantitatively
precise modelling and simulation of electronic properties of semiconductor con-
stituents is important for a better prediction of their potential in making low
dimensional electronic and optical devices. In the following we will give a brief
discussion about the critical structural properties and how they can be implemented
in theoretical models for the calculations of electronic properties of heterostructures.

1.2.1 Interface Strain Effects in Heterostructures

When two semiconductors with different lattice constants are grown upon each
other, strains will develop across the heterointerface increasing with layer thickness.
As long as the thickness of epilayer is kept under a critical thickness, the lattice
mismatch will be accommodated by uniform elastic strain, as shown in Fig. 1.6.
The interface strain will modify the structural and electronic properties of the
constituent semiconductors in directions parallel and perpendicular to the growth
direction. Therefore, the lattice matching is known to be important to the electronic
properties of semiconductor layers in heteroepitaxy. Until the early 1980s, the
studies of heterostructures had focused on semiconducting materials with close
lattice matching with a substrate, such as an AlGaAs/GaAs heterostructure quantum
well. The constraint on lattice matching in crystal growth is essential for relatively
thick epitaxial layers since a large lattice mismatch can cause the occurrence of
misfit dislocations with large densities. Such lattice-matched growth will prevent
the generation of misfit dislocations that would degrade charge transport and lower
device performance.
1 Modelling of Heterostructures for Low Dimensional Devices 7

Fig. 1.6 Strained

heteroepitaxy is a result of
growth of materials with
dissimilar lattice constants

The lattice constant of strained epilayer along the growth direction is equal to
that of the substrate and is expanded or compressed in the direction parallel to the
interface, shown in Fig. 1.6. Considering the (001) heterepitaxial growth, elastic
theory allows one to decompose the biaxial strain tensor into the sum of hydrostatic
and uniaxial strains along the growth direction.




afjj af? C12
exx ¼ eyy ¼ efjj ¼  1 ; ezz ¼ ef? ¼ 1 ¼2 efjj ð1:1Þ
af0 af0 C11 f

where ef ? and ef jj are the strain components perpendicular and parallel to growth
direction. The epilayer lattice constant will be equal to that of the substrate along
the growth direction: af jj ¼ as and is expanded by the bulk value of its Poisson ratio
perpendicular to the growth direction:
"

#
C12 af k
a f ? ¼ af 0 12 1 ; ð1:2Þ
C11 f af 0

where, afo is the bulk lattice constant and C11 and C12 are the bulk elastic stiffness
constants of the strained layer. as is the bulk lattice constant of the substrate.
Equations (1.1) and (1.2), suggest that when a thin layer is deposited on a buffer (or
substrate) with large mismatch, the epilayer will be under strain. The lattice con-
stant of the epilayer along and perpendicular to the interface will change to mini-
mize its elastic energy. The interface strains due to lattice mismatch and thermal
expansion gradients over the crystal growth temperature can co-exist in heterolayers
in low dimensional heterostructures; it is difficult to assign the observed stress to
8 H. Hakan Gürel et al.

Fig. 1.7 The schematic view of energy band diagram of a semiconductor under tensile strain
bandgap decreases (left) and compressive strain bandgap increases (right). Heavy hole (HH), light
hole (LH) and split-off (SO) bands are also shifted with respect to top of the valence band energy

either one or the other. The thermal expansion of lattice constants of epilayer and
the substrate with the growth temperature cause the lattice constant parallel and
perpendicular to the hetero-interface to change with temperature. High resolution
x-ray scattering measurements of GaAs grown on a Si substrate [9] show that the
thermal expansion of GaAs perpendicular to the growth direction follows the
thermal expansion of a Si substrate af k ðTÞ ¼ as ðTÞ and is therefore smaller than
that of bulk GaAs. However, the thermal expansion along the growth direction
exceeds the bulk GaAs value by the Poisson ratio as a result of the in-plane
constraint;

C12f
af? ðT) ¼ af ðT) þ 2 ½af ðT)  as ðT); ð1:3Þ
C11f

where as ðTÞ and af ðT) are the bulk linear thermal expansion coefficients of Si and
GaAs. Strain across the interface that will modify the electronic properties of both
materials, including the band offsets. Compressive (or tensile) strain in epilayer
results in an increase (or decrease) in conduction and valence band energy levels
[2], as shown in Fig. 1.7.
The uniaxial component of the biaxial strain tensor splits the heavy-hole,
light-hole and split-off valence band edges relative to the average valence band
edge. The heavy-hole, light‐hole and split‐off band energies relative to the average
valence band edge Ev are
1 Modelling of Heterostructures for Low Dimensional Devices 9

1 1
D  dE;
Evh ðeÞ ¼ Ev ðeÞ þ ð1:4aÞ
3 2
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1 1 9
Evl ðeÞ ¼ Ev ðeÞ  D þ dE þ D2 þ DdE þ dE 2 ; ð1:4bÞ
6 4 2 4
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1 1 9
Evs ðeÞ ¼ Ev ðeÞ  D þ dE  D2 þ DdE þ dE 2 ; ð1:4cÞ
6 4 2 4

where dE ¼ 2bðezz  exx Þ ¼ 2bðe?  ejj Þ and b is the shear deformation potential
which describes splitting in the valence band energy due to the [001] uniaxial strain.
Ev ðeÞ is the average valence band maximum under hydrostatic strain. The hydro-
static component of biaxial strain tensor corresponds to the relative volume change
of the strained epilayer; TrðeÞ ¼ 2e? þ e== ¼ DV=V0 ¼ ðV  V0 Þ=V0 , which leads
to shifts of the conduction band minimums relative to the average valence band
maximum at the Γ point.
According to a statistical thermodynamic model in which the conduction elec-
trons and valence holes are treated as charged chemical particles, the conduction
and valence band energy levels energy are expressed as a function of pressure at
any temperature as [10, 11]
 
ai P2 ð1 þ B0 Þ 3
Ei ðT; PÞ ¼ Ei ð0; P0 Þ þ CiP
0
Tð1  ln TÞ  P  P ; ð1:5Þ
B 2B 6B2

where i represents the conduction (c) or valence (v) band energy levels, P is the
applied pressure, T is the temperature, Ei(0, P0) are the conduction or valence band
edges at Γ, L and X high symmetry points. Eci and Ev are the conduction and
valence band edges at C, L and X points with deformation potentials aci ¼
Bð@Ecl =@PÞ and av ¼ Bð@Ev =@PÞ. B is the bulk modulus; B0 ¼ @B=@P is its
0
derivative. CiP is the standard heat capacity of conduction electrons and valence
holes. CcP ¼ CnP
0 0
 C0P
0
¼ CpP
0
þ DCP0 and CvP 0
¼ CpP0
are the standard heat
capacities of conduction electrons and valence holes; CnP ¼ CpP
0 0
¼ ð5=2Þk, where k
is the Boltzmann’s constant. DCP ¼ CnP þ CpP  C0P is the heat capacity of
0 0 0 0

reaction of free electron and hole formation obtained by fitting (1.6) to the exper-
imental data [12, 13] and empirical pseudopotential bandgap energy [14] and at
high symmetry points in the first Brillouin zone of semiconductors:
 
agl P2 ð1 þ B0 ÞP3
Egl ðT; PÞ ¼ Egi þ DCP0 Tð1  ln TÞ  P  ; ð1:6Þ
B 2B 6B2

where P ¼ 2Bf Cf ef == and P ¼ 3Bs esjj for the epilayer and the substrate,
respectively. In the case of (001) pseudomorphic growth, e ¼ ef jj ¼ ðaf jj  af Þ=af is
the strain in the epilayer along the growth direction and Cf ¼ ðC11  C12 Þ=C11 .
10 H. Hakan Gürel et al.

Any interface strain will modify the energy band structure and charge transport in
heterobipolar and unipolar devices. It should be noted that as the device dimension
is reduced to nanoscale of the order of Bohr radius of charged carriers, the use of
continuum elastic theory becomes questionable [15]. In such a case one should use
the atomic elasticity theory known as valence force field (VFF) approach [16],
especially as the lattice mismatch at heterointerface becomes large.

1.2.2 Composition Effects in Heterostructures

Semiconductor alloys based on IV–VI, III–V and II–VI compounds are important
in fabricating low dimensional bipolar and unipolar heterostructure devices since
their structural and electronic properties (e.g., lattice constants and bandgaps) can
be tailored independently. Therefore, reliable and accurate determination of com-
position variation of lattice constant and bandgap energies are very important. In the
theoretical determination of composition effects on the structural properties such as
lattice constants, a virtual crystal approximation (VCA) is often used [17, 18] in
which the compositional disorder effect is neglected. Since in VCA the alloy
potential is taken as the concentration weighted average of the constituent potentials
the bandgap energy is linear function of alloy composition. However, many
experimental studies report that the bandgap energies of semiconductor alloys is
nonlinear function of composition. Furthermore, it is believed that the composi-
tional disorder, which is related to the differences of electronegativity of atoms
forming a ternary semiconductor, plays a major role in determining its bowing of
the bandgap energy when lattice mismatch induced strain plays a crucial role in
heterostructure electronic properties.
In our recent studies [19–28] the effects of composition and strain on the elec-
tronic properties of semiconductor alloy constituents in heterostructures are
implemented in the semiempirical tight binding models and first principles DFT
with MBJLDA functional in terms of host bond length and distorted bond length by
the substitutional impurity without any adjustable parameter. In determining the
composition effects on band structures, we do not follow the common practice
which employs the so called virtual crystal approximation (VCA) to the TB
Hamiltonian matrix elements and bond length, with or without the compositional
disorder of the semiconductor alloys, since the detailed treatment requires inclusion
of the compositional disorder effect on the electronic properties. Instead, we employ
the method of the modified virtual crystal approximation (MVCA) in which one
formulates the composition dependence of the bond length of ternary material [19–
28]. The MVCA allows one to accurately take into account the effect of
disorder-induced nonlinear variation of the lattice constant on the TB parameters
used in calculating the band structure properties. The composition dependence of
bond length (or lattice constant) of the ABC ternary is written as the sum of
1 Modelling of Heterostructures for Low Dimensional Devices 11

undistorted bond length (dVCA ¼ ð1  xÞdAC 0

þ xdBC
0
) due to the virtual crystal
approximation (VCA) and the distorted bond length
(drelax ¼ xð1  xÞdc ðdBC  dAC Þ) due to cation-anion relaxation of binary in ternary
0 0

[19–28]:

d(x) ¼ ð1  xÞdAC ðxÞ þ xdBC ðxÞ ¼ dVCA ðxÞ þ xð1  xÞdc ðd0AC  d0BC Þ; ð1:7aÞ

dBC ðx) ¼ d0BC  ð1  xÞnAC:B ðd0BC  d0AC Þ; ð1:7bÞ

dAC ðx) ¼ d0AC  xnBC:A ðd0AC  d0BC Þ; ð1:7cÞ

where dAC ðxÞ and dBC ðxÞ are the bond lengths of AC and BC binaries in an ABC
0 0
ternary, dAC and dBC are the undistorted bond lengths of the host materials AC and
BC and nAC:B and nBC:A are two dimensionless relaxation parameters [29],

1 1
nAC:B ¼ ; nBC:A ¼ ; ð1:7dÞ
1þ aAC
6aBC ð1 þ 10 baAC
AC
Þ 1þ aBC
6eAC ð1 þ 10 baBC
BC
Þ

dc is the difference between dimensionless relaxation parameters:

dc ¼ nAC:B  nBC:A . As an example, Fig. 1.8 shows the composition variation of the
lattice constants and interface strain in AlGaN and InGaN ternaries in AlGaN/GaN
and InGaN/GaN heterostructures, respectively. Compressive strain at the
InGaN/GaN heterointerface decreases the lattice constant of the InGaN ternary
constituent as composition increases. Whereas tensile strain at the AlGaN/GaN
heterointerface increases the lattice constant of the AlGaN ternary constituent as
composition increases, such an increase or decrease in lattice constant due to
interface strain will change the electronic properties of the heterostructure ternary
constituents. Therefore, it is essential that we include the compositional disorder
effect on the structural and electronic properties of constituents in lattice mis-
matched heterostructures. Equation (1.7a–1.7d) can then be used to take into
account the composition effects on the band structures of the ternary constituent of
heterostructures in the theoretical models such as first principles WIEN2K simu-
lations package with the MBJLDA functional embedded in DFT and semiempirical
sp3, sp3s*, sp3d5s* and sp3d5 tight binding models.
The diagonal or off-diagonal matrix elements in the NN sp3d5 and 2NN sp3s* TB
Hamiltonian matrix elements for an ABC ternary semiconductor are expressed as
[19–28]

Ea=b ðxÞ ¼ ð1  xÞEa=b ðACÞ þ xEa=b ðBCÞ þ xð1  xÞ½Ea=b ðACÞ  Ea=b ðBCÞ;
ð1:8Þ

where Ea=b ðACÞ and Ea=b ðBCÞ represent the fitted energies of the s, p and d states
of anion and cation atoms forming the AC and BC binary compounds. This allows
12 H. Hakan Gürel et al.

Fig. 1.8 Lattice constant versus alloy composition (left) and interface strain versus alloy
composition (right) for AlGaN/GaN and InGaN/GaN GaN heterostructures

one to take into account the composition variations of the diagonal and off-diagonal
elements of the TB Hamiltonian matrix for a ternary ABC semiconductor can be
taken as a nonlinear function of alloy composition.

1.3 Semiempirical Tight Binding Modeling

of Heterostructures

Advances in growing multilayered structures with layer thicknesses approaching

atomic dimensions have provided new opportunities and new challenges for the
theoretical modelling of electronic structures of low dimensional heterostructures
based on the universally accepted fundamental principles of solid state physics and
quantum mechanics. The electronic properties of low dimensional bipolar and
unipolar semiconductor device structures are often calculated based on the fol-
lowing theoretical models: (i) First principle ab initio methods, such as density
functional theory (DFT) [29, 30], (ii) methods, such as local/empirical pseudopo-
tential method [14], or (iii) tight binding method (also known as the linear
1 Modelling of Heterostructures for Low Dimensional Devices 13

combination of atomic orbitals (LCAO) method) [19–28, 31–38], and k.p method
[39, 40]. The first principles ab initio methods are computationally expensive and
require heavy parallel computations. Since the ab initio models are based on the
calculations of ground state properties, they can only give limited physical insight
about the energy band structure at high temperatures and pressures.
On the other hand, the semiempirical tight binding models are less expensive
than the ab initio methods and can easily be implemented in calculating electronic
properties of low dimensional heterostructures such as nanowires or quantum dots.
The tight binding model has recently become popular, because of its simplicity and
ability to realistically describe the structural and dielectric properties of semicon-
ductors in terms of chemical bonds. The tight binding model is an atomistic
approach and is well suited for calculating the electronic band structure of semi-
conductor heterostructures such as two dimensional quantum wells and zero
dimensional quantum dots. In the following Sects. (1.3.1–1.3.4), we will give a
detailed discussion of the semiempirical tight binding theory with first and second
nearest neighbor (NN and 2NN) sp3 and sp3s* atomic orbitals sets and first nearest
neighbor (NN) sp3d5 orbitals set with spin-orbit coupling of cation (Al, Ga; In, Cd,
Zn) and anion (P, As, Sb, S, Se, Te) atoms for calculating the electronic structure of
III–V and II–VI compounds and their heterostructures. We shall later on (Sect. 1.4)
compare the semiempirical tight binding models with the density functional theory
(DFT) that uses the modified Becke-Johnson exchange-correlation potential with
the local density approximation (LDA), called MBJLDA functional, for calculating
the band structure of group III–V and II–VI compounds and their alloys.

1.3.1 Semiempirical Sp3 Tight Binding Modeling

The elementary tight-binding method, also known as the linear combination of

atomic-orbitals (LCAO), provides a physically insightful formulation of energy
band theory in a crystalline solid, which permits tracing trends from system to
system entirely in terms of the fundamental physical features [31–34]. In the
semiempirical tight binding theory of semiconductors one assumes that the valence
electrons are tightly bound to their nuclei as in the free atom. Anion and cation
atoms are brought together until their separations become comparable to the lattice
constant of semiconductors, at which point their wave functions will overlap. In the
Slater-Koster formalism, the crystal potential of a binary semiconductor is defined
as a sum of the symmetrical spherical potentials around each atom. This allows the
electronic wave function wak to be written as linear combination of atomic orbitals
ua ðkÞ(called the Löwdin orbitals):

X X ei:k:ri
jwak i ¼ ua jua ðkÞi ¼ pffiffiffiffi jua ðr  ri Þi; ð1:9Þ
a i N
14 H. Hakan Gürel et al.

where r is the position of electron with respect to origin of coordinate system is real
space, k is the wave vector, N is the number of atoms, and ri is the crystal lattice
site. The energy state is given by the eigenvalue of the linear equation written in
matrix form [31–34];
X 
Hab ðkÞ  Sab ðkÞE ub ¼ 0; ð1:10Þ
b

where E is the eigenvalue and Hab ¼ hua ðkÞjH ub ðkÞ and Sab ¼ hua ðkÞjub ðkÞ
are the Hamiltonian and overlap matrices, respectively.
   
Hcc Hca 1 Sca
Hab ¼ ; Sab ¼ ; ð1:11Þ
Hac Haa Sac 1

Here α correspond to a cation s (p) orbital and β corresponds to an anion s

(p) orbital and ub is the wave function coefficient. In the orthogonal sp3 tight
binding formalism in which only the first nearest neighbor interactions are taken
into account, the Bloch functions in (1.8) are taken to be orthogonal so that the
overlap matrix Sab ðkÞ is an identity matrix. There are nine independent matrix
elements, namely the four on-site atomic energies, Esa ; Esc ; Epa and Epc (with
a = anion and c = cation) and five hopping terms, Ess ; Exx ; Esa pc ; Esc pa and Exy . The
diagonal elements (Esa ; Esc ; Epa and Epc ,) are expressed as the (4 × 4) matrix
2 3 2 c 3
Esa 0 0 0 Es 0 0 0
6 0 Epa 0 0 7 6 0 Epc 0 0 7
Haa ¼6
4 0
7; H ¼ 6 7; ð1:12aÞ
0 Epa 0 5 cc 4 0 0 Epc 0 5
0 0 0 Epa 0 0 0 Epc

The off-diagonal elements (Ess ; Exx ; Esa pc ; Esc pa and Exy ) are expressed as the
(4 × 4) matrix
2 3
B0 Ess B1 Esp B2 Ess B3 Esp
6 B E B2 Exy 7
6 0 sp B0 Exx B3 Exy 7
Hac ¼ 6 7;
4 B2 Esp B3 Exy B0 Exx B1 Exy 5
B3 Esp B2 Exy B1 Exy B0 Exx
2  3 ð1:12bÞ
B0 Ess B1 Esp B2 Ess B3 Esp
6 B E B0 Exx B3 Exy B2 Exy 7
6 sp 7
Hca ¼ 6 1 7;
4 B2 Esp B3 Exy B0 Exx B1 Exy 5
B3 Esp B2 Exy B1 Exy B0 Exx

where Bi is the complex conjugate of matrix element Bi , defined in (1.15a). The
diagonal matrix elements, representing the interactions between the same p-orbitals
of cation and anion atoms yield symmetric sums; off-diagonal matrix elements
1 Modelling of Heterostructures for Low Dimensional Devices 15

representing the interactions between different p-orbitals of cation and anion atoms
yield asymmetric sums, given as:
 c a
D c a E  c a

px H px ¼ py H py ¼ pz H pz ¼ Exx B0 ðkÞ; ð1:13aÞ

 c a E D c a
D
 E

px H py ¼ py H px ¼ Exy B3 ðkÞ; pcy H paz ¼ pcz H pay ¼ Exy B1 ðkÞ;
ð1:13bÞ
D
 E  c a
 c a

pcy H paz ¼ pcz H pay ¼ Exy B1 ðkÞ; px H pz ¼ pz H px ¼ Exy B2 ðkÞ;
ð1:13cÞ

Interactions between an s orbital of atom A and the px ; py ; pz orbitals of

neighboring atoms B are

X
4
X
4
hsc jH jsa i ¼ Ess eik:rn ¼ Ess B0 ðkÞ; hsc jH pax ¼ Esp eik:rn ¼ Esp B1 ðkÞ;
n¼1 n¼1
ð1:14aÞ

E X
4
X
4

hsc jH pay ¼ Esp eik:rn ¼ Esp B2 ðkÞ; hsc jH paz ¼ Esp eik:rn ¼ Esp B3 ðkÞ;
n¼1 n¼1
ð1:14bÞ







kx a ky a kz a kx a ky a kz a
B0 ðk) ¼ 4Cos Cos Cos  4iSin Sin Sin ;
2 2 2 2 2 2
ð1:15aÞ







kx a ky a kz a kx a ky a kz a
B1 ðk) ¼ 4Cos Sin Sin þ 4iSin Cos Cos ;
2 2 2 2 2 2
ð1:15bÞ







kx a ky a kz a kx a ky a kz a
B2 ðk) ¼ 4Sin Cos Sin þ 4iSin Sin Cos ;
2 2 2 2 2 2
ð1:15cÞ







kx a ky a kz a kx a ky a kz a
B3 ðk) ¼ 4Sin Sin Cos  4iCos Cos Sin ;
2 2 2 2 2 2
ð1:15dÞ
16 H. Hakan Gürel et al.

Fig. 1.9 Crystal unit cell

group III–V and II–VI
compounds with zinc-blende
structure

pffiffiffiffiffiffiffi
where i ¼ 1 and r1 = (a/2)(1, 1, 1), r2 = (a/2)(1, −1, −1), r3 = (a/2)(−1, 1, −1)
and r4 = (a/2)(−1, −1, 1) are the displacement vectors of nearest neighbors.
Figure 1.9 shows the schematic view of unit cells for tetrahedral semiconductors.
In the first nearest neighbor (NN) sp3 tight binding theory, the acceptable fea-
tures of electronic band structure of semiconductors are dominated by the diagonal
terms (on-site atomic energies:Esa ; Esc ; Epa and Epc ) and off-diagonal terms
(Ess ; Exx ; Esa pc ; Esc pa and Exy ) representing the nearest neighbor interactions. The
first two off-diagonal elements Ess and Exx can be obtained from the bandgap at the
C point ðkx ¼ ky ¼ kz ¼ 0Þ for which B1 ; B2 and B3 are all zero, but B0 is nonzero.
The solution of the matrix (1.8) at the C point (k = 0) then yields
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
EðC1c=v Þ ¼ ðEsa þ Esc Þ  ½ðEsa  Esc Þ=22 þ 16Ess 2; ð1:16Þ
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
EðC15c=v Þ ¼ ðEpa þ Epc Þ  ½ðEpa  Epc Þ=22 þ 16Exx 2 ; ð1:17Þ
2

where EðC15v Þ are EðC1c Þ the top of the triply degenerate valence band and bottom
of the conduction band, respectively. The two unknown parameters Ess and Exx can
be determined by inverting (1.7) provided we have accurate s and p energies Es and
Ep for the anion and cation atoms, the bandgap at C point and the width of the
valence band. The next three off-diagonal matrix elements Esa pc ; Esc pa and Exy are
obtained from the bandgap data at the X and L high symmetry points. At the X high
symmetry point ðkx ¼ 1; ky ¼ kz ¼ 0Þ the Block sums are B0 ¼ 0; B1 ¼ 4i; B2 ¼ 0
and B3 ¼ 0, so that conduction and valence band energies are
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
EðX3c=v Þ ¼ ðEsa þ Epc Þ  ½ðEsa  Epc Þ=22 þ 16Es2a pc ; ð1:18aÞ
2
1 Modelling of Heterostructures for Low Dimensional Devices 17

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
EðX1c=v Þ ¼ ðEsa þ Epc Þ  ½ðEsa  Epc Þ=22 þ 16Es2c pa ; ð1:18bÞ
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
EðX5c=v Þ ¼ ðEpa þ Epc Þ  ½ðEpa  Epc Þ=22 þ 16Exy 2 ; ð1:18cÞ
2

Equations (1.18a) can be inverted for Esa pc ; Esc pa and Exy in terms of the
bandgap data at the X symmetry point with appropriate s and p energies. Finally at
the L symmetry point ðkx ¼ ky ¼ 1; kz ¼ 0Þ the conduction and valence band
energies are
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
EðL3c=v Þ ¼ ðEpa þ Epc Þ  ½ðEpa  Epc Þ=22 þ 4ðExx þ Exy Þ2 ; ð1:19Þ
2

Likewise L energy levels can be improved. It should be noted that fitting of

(1.16)–(1.19) to the bandgap data [12, 13] and empirical pseudopotential bandgaps
[14] at high symmetry points is reliable only for the valence band but not to the
conduction bands; this is vital in studying the electronic structures of low dimen-
sional heterostructures. In this respect we find it useful to use the semiempirical
second nearest neighbor (2NN) sp3 tight binding models [35, 36]. Talwar and Ting
[35] incorporate the second nearest neighbor (2NN) interactions of cation and anion
atoms in the semiempirical sp3 tight binding formalism by constructing a nonzero
symmetrized (8 × 8) Hamiltonian matrix with 23 two-center 2NN integrals. The
(2NN) sp3 TB Hamiltonian matrix elements are given as

X
Hab ¼ hua ðkÞjH ub ðkÞ ¼ Eab þ 00
Iab ð0; iÞeik:ri þ Hab þ Hso ; ð1:20Þ
i6¼0

where Eab represents the intra-atomic integrals, coupling atomic orbitals located in
the same cell. The second term Iαβ(0, i) is the integral that represents the NN
interactions, coupling atomic orbitals located in different cells. Third term repre-
sents 2NN interaction integrals. Finally, the last term represents the spin-orbit
interactions. In this 2NN sp3 TB approach there are 23 TB parameters to be
determined from the fitting of the TB model energy bands to experimental data [12,
13] and empirical pseudopotential bands at symmetry points [14].
The modified virtual crystal approximation (MVCA) can be implemented in the
semiempirical tight binding model [19–28] by using (1.7a–1.7d) and (1.8) for the
composition variation of bond length and diagonal and off-diagonal elements in the
TB Hamiltonian matrix for ternary constituent of heterostructures. This allows one to
take into account the composition variations of the off-diagonal terms in the TB
Hamiltonian matrix elements. Likewise, the TB parameters representing the diag-
onal terms in the Hamiltonian matrix for ABC ternary semiconductor can be taken
nonlinear function of composition using (1.8). We can then use the semiempirical
NN or 2NN sp3 TB model calculate the composition effects on electronic properties
of ternary/binary heterostructures by taking account of lattice mismatch induced
18 H. Hakan Gürel et al.

Fig. 1.10 Band structure of GaAs obtained using 2NN sp3 TB model of Talwar and Ting [35]
(dashed lines) and 2NN sp3 TB model of Loehr and Talwar [36] (solid lines)

interface strain that causes modification of the energy levels of the heterostructure
constituents.
Figure 1.10 shows the band diagram of GaAs obtained using the NN and 2NN
sp3 TB model due to the parametrization of Talwar and Ting [35] and Loehr and
Talwar [36]. As shown in Fig. 1.10, improvements made in the band structure
calculations by the 2NN sp3 TB model of Loehr and Talwar is visible when it is
compared with that of Talwar and Ting at varying values of wave vector over the
entire first Brillouin zone, although both produce the experimental bandgap data
and empirical pseudopotential bands at symmetry points. However, the low
dimensional device performance modeling requires accurate numerical values for
the electron and hole masses since they determine the charge transport and device
performance. In aiming for this goal, Loehr and Talwar inverted the expressions for
energy levels at the C, L and X high symmetry point, fitting the electron and hole
effective masses, to reduce the number of free tight binding parameters from 23 to
8. The 2NN sp3 TB parametrization of Loehr and Talwar yields much better
conduction band dispersion curves than that of Talwar and Ting. The optimized
2NN sp3 TB parameters of Loehr and Talwar reproduce the electron mass at the
lowest conduction valley and heavy hole effective mass of the valence band at C
point, in addition to its ability to reproduce the correct values of the critical point
energies at the C, L and X high symmetry points.
1 Modelling of Heterostructures for Low Dimensional Devices 19

Fig. 1.11 Band structure of AlGaAs with x = 0.20 (left) and x = 0.30 (right) calculated with the
2NN sp3 TB parameters of Talwar and Ting [23] (dashed lines) and Loehr and Talwar [24] (solid
lines)

Using the tight binding parameters of Talwar and Ting [35] and Loehr and
Talwar [36] we calculated the composition effects on band structure of AlGaAs
ternary semiconductor and the results are displayed in Fig. 1.11 for x = 20 and 30 %
of aluminum fraction. Furthermore, the composition variation of band gaps of
AlGaAs ternary in lattice matched AlGaAs/GaAs heterostructure is shown in
Fig. 1.12 (left) and of InGaAs ternary in pseudomorphic InGaAs/GaAs
heterostructures is shown in Fig. 1.12 (right), for which band gaps at 0 K are
computed using the 2NN sp3 TB model with the TB parameters of Talwar and Ting
and Loehr and Talwar compared with experimental data and empirical pseudopo-
tential model results at symmetry points. As can be seen from Fig. 1.12 (left), there
is an excellent agreement between theory and experiment for lattice matched
AlGaAs/GaAs heterostructures. However, it should be noted that there is a con-
siderable lattice mismatch across many of the III–V and II–VI ternary/binary
heterointerfaces (e.g., InGaAs/GaAs). Interface strain due to lattice mismatch
causes a shift in the lattice constant of an epilayer: a ¼ ð1 þ eÞa0 , where e is the
symmetric strain tensor. Therefore, the bond lengths, defined in (1.7a–1.7d), and
TB matrix elements, defined in (1.8), will be modified with strain. The effect of
20 H. Hakan Gürel et al.

Fig. 1.12 Predicted strain effects on bandgaps of AlGaAs in lattice matched AlGaAs/GaAs (left)
and of InGaAs in pseudomorphic InGaAs/GaAs (right) heterostructures

interface strain is evident in the case of InGaAs/GaAs heterostructure, as seen in

Fig. 1.12 (right). There is a large strain shift in the band gaps of the InGaAs ternary
constituent in an InGaAs/GaAs heterostructure at symmetry points due to large
lattice mismatch that increases with indium mole fraction. There is good agreement
between theory and experiment for the InGaAs ternary as bulk at around 0 K.
Qualitatively reliable and quantitatively accurate determination of the strain
effects on the energy levels by fitting the off-site tight binding matrix elements to a
set of observables is difficult. The off-site TB matrix elements representing nearest
neighbor interactions, known as the hopping strength, will be modified with respect
to their unstrained values and are often determined by assuming that they obey the
Harrison Scaling Law [34]: Vll0 m ðeÞ ¼ Vll0 m ða=ao Þgllm , where Vll0 m ðeÞ is the strained
and Vll0 m the unstrained value of interaction potential for anion and cation atoms.
The exponents gllm are determined to reproduce the strain variations of the band
structure of relevant semiconductors under hydrostatic pressure, namely the volume
deformation potential agl ¼ Bð@Egl =@PÞ for the corresponding band gap energies
EgC ; EgL and EgX at symmetry points. Since these bandgaps depend on experimental
data, in order to obtain reliable and accurate TB parameters, one must go through a
fitting process that depends on the mapping of a large number of orbital coupling
1 Modelling of Heterostructures for Low Dimensional Devices 21

parameters on the set of observables; in many cases there are not many analytical
expressions available. One can overcome this difficulty by using a so-called sta-
tistical thermodynamic model [10, 11], which considers the conduction electrons
and valence holes as distinct electrically chemical species, to study the interface
strain effects on the electronic structure of heterostructures at symmetry points,
defined by (1.5) and (1.6). In this model, one first expresses the shifts in the
conduction and valence band edges at C, L and X high symmetry points obtained
by using the 2NN sp3 TB orbitals basis sets as a function of pressure at any lattice
temperature and then obtains the refined tight binding matrix elements to find the
band structure.
Although the Slater-Koster type semiempirical 2NN sp3s tight binding treatment
of electronic structures yields a good description of valence band dispersion curves,
the conduction band dispersion curves are inaccurately given, especially the indirect
band gap at the X symmetry point is not well reproduced. Vogl et al. [34] intro-
duced a nearest neighbour sp3s* tight binding model in order to include the influ-
ence of excited d-states, which will be discussed next.

1.3.2 Semiempirical Sp3s* Tight Binding Modeling

In the semiempirical nearest neighbor sp3s* tight binding model of Vogl et al. [41],
each atom is described by its outer valence s orbital, three p orbitals and fictitious
excited s* orbital is added to mimic the effects of higher lying d-states. This results
in a semiempirical NN sp3s* tight binding model with a total of 13 parameters,
which are determined by comparing the predicted bandgaps with those produced by
empirical pseudopotential model [14] at high symmetry points. Adding the excited
s* state and spin-orbit coupling to sp3 orbital basis set makes it possible to accu-
rately calculate band dispersion curves at the X high symmetry point. Furthermore,
the inclusion of 2NN interactions of cation and anion atoms in the sp3s* TB model
yields better fit of conduction band dispersion curve at the L symmetry point.

The sub-matrices (diagonal (Hcc and Haa ) and off-diagonal (Hca and Hac ¼ Hca )
3 *
elements of the 2NN sp s Hamiltonian matrix Hab are written as [41]
2 3
Esc eðs; px ÞB6 eðs; px ÞB5 eðs; px ÞB4 0
6 eðs; px ÞB6 Epc eðpx ; py ÞB4 eðpx ; py ÞB5 0 7
6 7
Hcc ¼ 6
6 eðs; px ÞB5 eðpx ; py ÞB4 Epc eðpx ; py ÞB6 0 7 7;
4 eðs; px ÞB4 eðpx ; py ÞB5 eðpx ; py ÞB6 Epc 0 5
0 0 0 0 Esc
ð1:21aÞ
22 H. Hakan Gürel et al.
2 3
Esa eðs; px ÞB6 eðs; px ÞB5 eðs; px ÞB4 0
6 eðs; px ÞB6 Epa eðpx ; py ÞB4 eðpx ; py ÞB5 0 7
6 7
Haa ¼ 6
6 eðs; px ÞB5 eðpx ; py ÞB4 Epa eðpx ; py ÞB6 0 7 7; ð1:21bÞ
4 eðs; px ÞB4 eðpx ; py ÞB5 eðpx ; py ÞB6 Epa 0 5
0 0 0 0 Esa
2 3
B0 Ess B1 Esp B2 Esp B3 Esp 0
6 B1 Esp B0 Exx B3 Exy B2 Exy B1 Eps 7
6 7
Hca ¼ 6
6 B2 Esp B3 Exy B0 Exx B1 Exy B2 Eps 7 
7; Hac ¼ ðHca Þ ; ð1:21cÞ
4 B3 Esp B2 Exy B1 Exy B0 Exx B3 Eps 5
0 B1 Es p B2 Es p B3 Es p 0

where Esa ; Esc ; Epa ; Epc ; Esa , and Es c are diagonal elements of Hab matrix and repre-
sent the on-site atomic energies of cation and anion atoms.
Ess ; Exx ; Esa pc ; Esc pa ; Exy ; Es p and Eps are the off-diagonal elements of Hab matrix
and represent the hopping terms (transfer matrix elements). Finally, esx ¼
eðscðaÞ; px cðaÞÞ and exy ¼ eðpx cðaÞ; py cðaÞÞ) are the two 2NN transfer matrix ele-
ments for the cation and anion atoms. Here s and p refer to the basis states and a and
c refer to anion (e.g., As, Sb, N) and cation (e.g., Al, Ga and In) atoms, respectively.
In the 2NN sp3s* TB Hamiltonian matrix, Bi is the complex conjugate of the matrix
element Bi which gives the k wave vector dependence and the first four elements
are the same as in (1.15a) and other three are


B4 ðk) ¼ 4Sinðkx aÞSin ky a ; ð1:22aÞ

B5 ðk) ¼ 4Sinðkx aÞSinðkz aÞ; ð1:22bÞ


B6 ðkÞ ¼ 4Sin ky a Sinðkz aÞ; ð1:22cÞ
pffiffiffiffiffiffiffi
where i ¼ 1 and displacement vectors of nearest neighbors. Including the
spin-orbit coupling the size of the (10 × 10) 2NN sp3s* TB Hamiltonian matrix
increases to (20 × 20) matrix which is diagonalized for each k vector to obtain the
band structure [42]. The spin-orbit effects are included with the 2NN sp3s* orbitals
set by coupling different spin states of different on-site p orbitals through the
spin-orbit interaction. Tight binding parameters of 2NN sp3s* orbitals set is
obtained by fitting the obtained band gaps given in Table 1.1 to those produced by
empirical pseudopotential theory [14]. Having reliable diagonal matrix elements
one can make realistic tight binding parametrization of the off-diagonal matrix
elements representing the first nearest neighbor (NN) and/or 2NN interactions,
which are obtained by focusing on the reproduction of the band gap energies at
symmetry points in the energy dispersion curve. As examples of 2NN sp3s* tight
binding parametrization, the bandgaps given in Table 1.1 [21, 23] for several group
III–V compounds may be used to calculate electronic band structure of their
heterostructures.
1 Modelling of Heterostructures for Low Dimensional Devices 23

Table 1.1 Bandgaps at symmetry points of AlAs, GaAs, InAs, GaP, AlN, GaN and InN
compounds obtained using the 2NN sp3s* TB model [21, 23] and experimental data [12] are given
in parenthesis
(eV) GaAs InAs GaP AlN GaN InN
EgC 1.52 0.43 2.88 5.99 3.30 0.90
(1.52) (0.42) (2.88) (6.00) (3.27) (0.90)
EgX 1.98 2.28 2.33 4.90 4.70 2.83
(2.03) (2.50) (2.16) (4.80) (4.70) (3.00)
EgL 1.82 1.61 2.56 8.63 6.10 3.81
(1.85) (1.43) (2.79)

Table 1.2 2NN sp3s* TB parameters for GaAs, InAs, GaP, AlN, GaN and InN obtained using
Table 1.1
(eV) GaAs InAs GaP AlN GaN InN
Es,a −8.4399 −9.5381 −8.1124 −11.505 −12.915 −12.860
Ep,a 0.9252 0.7733 1.0952 4.3815 3.1697 1.9800
Es,c −2.6569 −2.7219 −2.1976 0.5047 −1.5844 −0.3994
Ep,c 3.5523 3.5834 4.0851 10.2184 9.0302 8.0200
Es*,a 6.6235 7.2730 8.4796 12.0400 12.2000 10.6300
Es*,c 7.4249 6.6095 7.1563 13.7400 12.2000 13.0000
4Vs,s −6.4210 −5.6052 −7.4909 −9.8077 −8.8996 −4.2285
4Vx,x 1.9850 1.8398 2.1516 6.6900 5.3500 3.9800
4Vx,y 4.9100 4.3977 5.1213 8.9400 8.6200 7.4100
4Vsa,pc 4.2390 3.0205 4.2724 7.8500 6.4000 3.8100
4Vpa,sc 5.15358 5.3894 6.3075 7.6800 7.2400 6.1900
4Vs*a,pc 3.80624 3.2191 4.8184 8.0300 7.0600 6.8800
4Vpa,s*c 4.7009 3.7234 50534 2.4700 1.8200 3.3600
esx 0.2459 0.1441 0.2325 −1.4000 0.9500 0.6150
exy −0.1050 0.0249 −0.2200 6.9000 1.0100 0.7100
ka 0.0553 0.1385 0.0578 0.0035 0.0035 0.0035
kc 0.1338 0.1290 0.0222 0.0070 0.0410 0.1100

As an example of 2NN sp3s* TB parametrization, the diagonal and off-diagonal

tight binding parameters for AlAs, GaAs, InAs, GaP, AlN, GaN and InN are given
in Table 1.2 [21, 23].
Using the 2NN sp3s* TB model the electronic band structures of GaAs, GaN,
InN and AlN are calculated [21, 23] and are shown in Fig. 1.13, which reproduce
the conduction and valence band structures, including the heavy hole and light hole
bands as well as spin-orbit splitting bands. Tight binding interaction parameters
4Vs*,p and 4Vp,s* were adjusted to fit to the X bands and 2NN interaction param-
eters esx and exy to get a good fit to the L bands in reproducing the empirical
pseudopotential energy bands [14]. As shown in Fig. 1.13, adding the excited s*
state to the sp3 orbitals set on the cation and anion atoms with 2NN interactions and
24 H. Hakan Gürel et al.

Fig. 1.13 Band structures of AlN, GaN, InN and GaAs obtained using the 2NN sp3s* TB
parameters given in Table 1.2

spin-orbit coupling of p-states, improves the simulation of the conduction band

structure of III–V compounds, especially at the X symmetry point, reproducing the
empirical pseudopotential bands at symmetry points of energy dispersion curve,
which cannot be done with conventional NN and 2NN sp3 TB models.
As an example of implementing the modified virtual crystal approximation
(MVCA) for the implementation of the composition effects in semiempirical NN or
2NN sp3s* tight binding model, the tight binding parameters given in Tables 1.1
and 1.2 are used in the calculations of the electronic structures of AlGaN, InGaN
and GaAsN nitride ternaries in k-space for various alloy compositions, as displayed
in Fig. 1.14 showing the expected trend in band structures. The main features of the
composition effects on the energy band properties of AlGaN/GaN, InGaN/GaN and
GaAsN/GaAs ternary/binary group III-nitride heterostructures are summarized in
Fig. 1.15, in which the principal band gap energies of AlGaN, InGaN and GaAsN
ternaries are plotted as functions of interface strain for the entire composition range.
As seen from Fig. 1.15, lattice mismatch interface strain effects on the fundamental
band gaps at Γ, L and X symmetry points in AlGaN/GaN (left), InGaN/GaN
(middle) and GaAsN/GaAs (right) heterostructures can be quite large when the
band gap (or conduction band) deformation potential is large.
1 Modelling of Heterostructures for Low Dimensional Devices 25

Fig. 1.14 Band structures of AlGaN, InGaN and GaAsN obtained using the 2NN sp3s* TB
orbitals set with optimized energies given in Table 1.2

Predicted fundamental bandgaps of AlGaN, InGaN and GaAsN ternaries at the

Γ, L and X points, especially at Γ, are in excellent agreement with experiment [12].
The principal band gaps of AlGaN and InGaN conventional nitrides increase with
composition. However, the band gap of GaAsN diluted nitride at the Γ symmetry
point decreases with alloy composition for x < 0.25, and then increases for
0.26 < x < 1, in agreement with experiment. It is gratifying to note that using the
optimized 2NN sp3s* tight binding model parameters for GaAs, GaN,AlN and InN
compounds given in Tables 1.1 and 1.2, we are able to obtain the nonlinear
composition dependence of the principal band gaps of AlGaN, InGaN and GaAsN
ternary semiconductors without any empirical fitting or any adjustable parameter.
This observation suggests that by using the optimized 2NN sp3s* TB parameters for
bulk GaAs, GaN, AlN and InN binary compounds, given in Table 1.2, the 2NN
sp3s* TB model allows one to determine the nonlinear composition dependence of
the principal band gaps of nitride-based ternary semiconductors without any
empirical fitting nor any adjustable parameter.
Cadmium and zinc based group II–VI compounds (CdX and ZnX (X = S, Se,
Te)) and their ternary/binary heterostructures are known to have considerable
potential for making optical and photovoltaic devices (e.g., light emitting diodes,
laser diodes, infrared detectors, photovoltaic devices, and quantum dots [1, 2]).
Therefore, similar calculations were carried out for the electronic band structure of
26 H. Hakan Gürel et al.

Fig. 1.15 Predicted interface strain effects on the bandgaps of ternaries in AlGaN/GaN (left),
InGaN/GaN (middle) and GaAsN/GaAs (right) heterosructures compared with experiment [12]

cadmium and zinc based group II–VI compounds (CdX and ZnX (X = S, Se, Te))
and their ZnCdS/CdS (left) and ZnTeS/ZnS ternary/binary heterostructures by using
the 2NN sp3s* TB parametrization, which yield the same accuracy for the con-
duction band structure of CdS, CdSe, CdTe, ZnS, ZnSe and ZnTe. Figure 1.16
shows the composition and interface strain effects on band gaps of ternaries in
ZnCdS/CdS (left) and ZnTeS/ZnS (right) heterostructures compared with experi-
ment [12]. It should be noted that there is a considerable lattice mismatch across
many of the ternary/binary heterointerfaces. Therefore, the composition dependent
bond lengths, defined in (1.7a–1.7d), and tight binding matrix elements in the
semiempirical NN sp3s* orbitals set, defined in (1.24), and hence the electronic
properties of heterostructure alloy constituents will be modified with respect to their
unstrained values. Furthermore, accurate description of the second conduction band
and the transverse effective masses at the X- and L-symmetry points are found to be
in good agreement with experiment, leading to a reliable and accurate TB model
calculation of electronic properties of semiconductors. We should note that the
accuracy of 2NN sp3s* TB parametrization depends on the accurate description of
band structures of III–V and II–VI compounds by the empirical pseudopotential
theory and bandgap data.
1 Modelling of Heterostructures for Low Dimensional Devices 27

Fig. 1.16 Predicted interface strain effects on band gaps of ternaries in ZnCdS/CdS (left) and
ZnTeS/ZnS (right) heterosructures compared with experiment [12]

Although the inclusion of s* excited state in the 2NN sp3 TB model yields
accurate calculations of the conduction band structure features at the X high
symmetry point, it does so by modelling the average of p-d interactions and is
insufficient to determine energy levels above 6.0 eV. Jancu et al. [42, 43] suggested
that from a solid state physics point of view, the actual behaviour of excited d-states
is not reliably addressed in the sp3s* tight binding m of valence band and con-
duction band dispersion curves involving the C, X and L high symmetry points,
which will be discussed next.

1.3.3 Semiempirical Sp3d5s* Tight Binding Modeling

In the semiempirical nearest neighbor sp3d5s* tight binding model proposed by

Jancu et al. [42, 43], the excited d-states contribute critically to both the valence
band maximum at the Γ symmetry point and to the conduction band dispersion
curves at the X and L symmetry points and is explicitly included in NN sp3d5s*
orbitals set. The band gaps at high symmetry points in the Brillouin zone and
28 H. Hakan Gürel et al.

Fig. 1.17 Band structure and density of states of a GaAs compound calculated using the 2NN
sp3s* (dashed) and NN sp3d5s* (solid) tight binding models

optimized tight binding parameters are given in Tables 1.1 and 1.2 for the 2NN
sp3s* 2NN TB model and those of the NN sp3d5s* TB model were taken from
Jancu et al. [42]. Results of 2NN sp3s* TB and NN sp3d5s* TB models are com-
pared in Fig. 1.17 for electronic band structure and density of states of GaAs. As
can be seen from Fig. 1.17, both the valence band and conduction band dispersion
curves of GaAs obtained by using the NN sp3d5s* TB model overcomes most of the
limitations of the earlier 2NN sp3 and sp3s*TB models. As Fig. 1.17 (left) for GaAs
show, the NN sp3d5s* TB model overcomes most of the limitations of the earlier of
2NN sp3 and sp3s* 2NN TB models for both valence and conduction band dis-
persion curves. Furthermore, accurate description of the second conduction band
and the transverse effective masses at the X- and L-symmetry points that are in
good agreement with experiment suggests that the NN sp3d5s* TB parametrization
makes it possible to accurately calculate the electronic and optical properties
involving symmetry points at the edge of the Brillouin zone of tetrahedral semi-
conductors. It is noted here that the NN sp3d5s* TB model is reliable for the
calculation of dispersion curves for energies up to 6 eV above the valence band
maximum, correctly reproducing the orbital character of band edges, and their
1 Modelling of Heterostructures for Low Dimensional Devices 29

Fig. 1.18 Electronic band structure of AlAs and GaAs obtained using NN sp3d5s* TB model

behavior under strain, at the C, X and L symmetry points of the first Brillouin zone
of compound semiconductors.
In a search for physically realistic and numerically accurate calculations of band
structures of group III–V and II–VI compound semiconductors, Sapra et al. [44]
have argued that the adding of an s* state to an sp3 orbitals set does not represent the
true contribution of d-states to both valence band and conduction band dispersion
curves and have shown that the NN sp3d5 tight binding model is sufficient to
calculate the electronic structure of group III–V and II–VI compounds and suggests
that the use of any fictitious s* state is not needed in tight binding modeling to
calculate the energy band structures. However, as can be seen from Fig. 1.18, the
use of the more complete nearest neighbor sp3d5s* orbitals set imposes drastically
heavier computational requirement on the calculation of electronic band structures
of III–V and II–VI compound semiconductors, without affording realistic physical
insight to the solid state interpretation of its contribution to the problem.
In the following section we use a semiempirical NN sp3 tight binding
parametrization of the problems of nearest neighbor interactions and spin-orbit
30 H. Hakan Gürel et al.

Table 1.3 Bandgaps of CdS, CdSe and ZnS obtained from fitting the NN sp3d5 and 2NN sp3s*
TB models results [24] in the empirical pseudopotential bands [14] and measured band gaps [12,
13] at high symmetry points
Bandgap (eV) CdS ZnS CdSe
sp3s* sp3d5 sp3s* sp3d5 sp3s* sp3d5
EgC 2.476 2.555 3.680 3.702 1.887 1.937
EgX 4.341 4.696 5.103 5.190 3.784 3.779
EgL 3.983 5.193 4.810 4.641 3.097 3.027

coupling of p-states taken into account in the calculations of the band structures of
III–V and II–VI compound semiconductors.

1.3.4 Semiempirical Sp3d5 Tight Binding Modeling

In this section we discuss the semiempirical sp3d5 tight binding parametrization

with the first nearest neighbor interaction and spin-orbit coupling of p-states taken
into account in the calculation of the electronic structure of constituents of semi-
conductor heterostructures. We write the semiempirical NN sp3d5 TB Hamiltonian
matrix as [24]

X
Hab ¼ hua ðkÞjH ub ðkÞ ¼ Eab þ Iab ð0; iÞeik:ri þ Hso ; ð1:23Þ
i6¼0

where Eab is the on-site energy for the b orbital (s, p, d) at the atomic site a (cation
and anion); and it represents the intra-atomic integrals, which couple atomic orbitals
located in the same cell, and Iab ð0; iÞ represents the first nearest neighbor
(NN) interaction integrals, known as the hopping term, which couples atomic
orbitals located in different cells.
The nine state atomic-like (s; x; y; z; xy; yz; zx; x2  y2 ; 3z2  r 2 ) basis set
describes each atom of the semiconductor. Hab has 19 independent matrix elements:
9 on-site and 10 off-site elements. The inclusion of spin-orbit coupling to the sp3d5
basis set, in which the spin-orbit interaction is given by two parameters; ka ¼
\xa " jHso jza # [ for anion and kc ¼ \xc " jHso jzc # [ for cation atom, adds
two extra tight binding parameters. Therefore, Hab has total of 21 independent
matrix elements, including the nearest neighbor interaction and spin orbit coupling
of p-states. The diagonal and off-diagonal matrix elements are determined by fitting
the empirical pseudopotential energy band dispersions to calculated energy band
dispersions using the semiempirical NN sp3d5 TB parametrization for a given
semiconductor. One starts with calculating the electronic band structure by esti-
mating the values of on-site matrix elements and off-site matrix elements and then
carries out a least-squares error minimization fitting procedure at a number of high
1 Modelling of Heterostructures for Low Dimensional Devices 31

Table 1.4 Optimized s, p (eV) CdS ZnS CdSe

and s* TB parameters for
CdS, CdSe and ZnS obtained Es,a −11.5300 −11.6100 −9.63
from fitting the 2NN sp3s* TB Ep,a 0.5300 1.4800 1.326
model [24] to empirical Es,c 1.8300 1.1100 0.03
pseudopotential band [14] at Ep,c 5.8700 6.5200 4.73
symmetry points
Es*,a 7.1300 8.0800 7.53
Es*,c 6.8700 8.0200 5.72
4Vs,s −3.0700 −6.3000 −4.64
4Vx,x 1.7600 3.1100 2.64
4Vx,y 4.2300 5.0000 5.36
4Vs,p 2.1700 5.1600 4.57
4Vp,s 5.4800 5.1700 5.54
4Vs*,p 1.9900 2.8900 3.05
4Vp,s* 3.0600 1.7500 2.49
esx 0.1000 0.2000 0.0
exy −0.0100 −0.1500 0.0
ka 0.0250 [6] 0.0250 [6] 0.1434
kc 0.0130 [14] 0.0270 [14] 0.0591

symmetry points in the valence and conduction band dispersion curves to fit band
gap energies obtained from the empirical pseudopotential method [14] (Table 1.3).
We now compare the predictions of semiempirical NN sp3d5 and 2NN sp3s* TB
parametrizations, both of which include spin-orbit coupling of p-states [28].
Tables 1.4 and 1.5 give the number of optimized TB parameters for the NN sp3d5
and 2NN sp3s* TB models. Figure 1.19 compare the electronic band structure of
CdS, ZnS and CdSe calculated using NN sp3d5 and 2NN sp3s* TB models, both
reproducing the band gaps as well as the valence band and conduction band dis-
persion curves at Γ, X and L symmetry points, which cannot be done with con-
ventional NN sp3 and 2NN sp3 TB models.
As can be seen from Fig. 1.19 the tight binding model with an sp3d5 orbital
basis, including the nearest neighbor interactions of cation and anion atoms and
spin-orbit coupling of p-states, is adequate to accurately reproduce the band gaps at
Γ, X and L symmetry points and both the valence band and conduction band
dispersion curves. When compared with the 2NN sp3s* TB model, the NN sp3d5 TB
model better simulates both the valence band and conduction band dispersion
curves. This is due to the fact that the inclusion of s*-excited state is included by
modeling the average of p–d interactions, and it does not permit the inclusion of
excited d-orbitals to a sp3 orbitals set. Therefore, the actual behavior and contri-
bution of the excited d-states in the band structure calculations is not reliably
reproduced, suggesting that the sp3s* TB model is of limited value for determi-
nation of optical properties of semiconductors involving high symmetry points. We
should point out that both TB models with NN sp3d5 and 2NN sp3s* orbitals sets are
32 H. Hakan Gürel et al.

Table 1.5 Optimized s, p (eV) CdS ZnS CdSe

and d-TB parameters for CdS,
CdSe and ZnS compounds sc 1.8300 1.3600 0.0300
obtained from fitting the NN pc 5.8700 6.5200 4.7300
sp3d5 TB model [24] to dc ðt2 Þ −6.8300 −5.8200 −7.3100
empirical pseudopotential dc ðeÞ −7.4400 −6.2100 −7.8100
bands at symmetry points [14]
sa −11.5300 −14.6100 −9.6300
pa 0.9300 1.7800 1.3260
da ðt2 Þ 14.4300 15.5400 15.2600
da ðeÞ 13.1500 13.6000 14.1000
ssr −0.8440 −1.0050 −1.2600
spr 2.2729 2.5387 2.3989
ppr 1.68963 1.3343 1.9789
ppp 2.5500 4.3775 3.3400
psr −0.10750 −0.3725 −0.6800
dsr 0.0000 −2.5900 −1.2200
dpr −1.1400 −0.0500 −0.0100
dpp −1.2900 −2.7800 −1.0900
sdr 1.7500 1.3700 1.5200
pdr 1.5580 2.1100 1.7800
pdp −0.3500 −0.4500 −0.3200
ka 0.0250 0.0250 0.1434
kc 0.07600 0.0270 0.0591

semiempirical and their basis lies in a good description of the electronic band
structures of tetrahedral semiconductors.

1.4 Density Functional Theory Modelling

of Heterostructures

The conventional density functional theory (DFT) enables us to have a

parameter-free description of structural and electronic properties of semiconductors
and has its foundations in the work of Kohn and his collaborators [30, 45]. The
variation method is used to calculate the ground state properties of a many-body
system with the charge density, expressed in terms of single-particle electronic
orbitals

X
N X 2
nð~
rÞ ¼ f i;k ui;k ðr) [ ; ð1:24Þ
i¼1 k
1 Modelling of Heterostructures for Low Dimensional Devices 33

Fig. 1.19 Band structure of CdS, CdSe and ZnS compounds obtained using the 2NN sp3s* and
NN sp3d5 TB models with the tight binding parameters given in Tables 1.4 and 1.5

which plays the central role in the calculation of the structural and electronic
properties of semiconductors. The major consequence of the
Kohn-Hohenberg-Sham studies [30, 45] is that the single particle states ui;k ðr) are
determined by the solution of a set of equations similar to the Schrodinger equation:


h2 2
Hui;k ¼  r þ Vext þ Veff ui;k ¼ Ei ui;k ; ð1:25Þ
2m i

where ui;k ðr) are the one-electron wave functions, Vext is the external potential of
the nuclei, and Veff ðr) the effective potential defined as

dExc ðn(r)
Veff ðr) ¼ VH ðr) þ Vxc ðr) ¼ VH ðr) þ ; ð1:26Þ
dn(r)

where VH ðr) is the Hartree Coulomb term and Exc ðn(r) is the exchange-correlation
functional. Since the electron density n(r) is involved in the definition of effective
potential, (1.24) must be solved consistently with (1.25) and (1.26). Although the
set of (1.24)–(1.26) yields, in principle, an exact solution to the
quantum-mechanical problem, there are approximations involved in the
34 H. Hakan Gürel et al.

exchange-correlation potential Vxc ðr). In this respect, Local Density Approximation

(LDA) and Generalized Gradient Approximations (GGA) have proven to be
effective for a large number of semiconductors.
The conventional DFT calculations based on (1.24)–(1.26) provide satisfactory
results for ground state properties (e.g., total energies, lattice constant, bulk mod-
ulus) but give unsatisfactory results for the electronic properties (e.g., band gaps
and effective masses) [46]. For example, DFT predicted bandgaps are too small
compared to experimental data [12] and predicted by empirical pseudopotential
theory [14]. This difficulty can be avoided using various approaches (e.g., GW
approximation or hybrid functional) that have been implemented in DFT model
calculations. Even with GW and hybrid functionals, conventional DFT results in
bandgap error on the order of 10–20 % as compared with experimental data [12].
As a remedy, the exchange-correlation potential contribution to the DFT band gap
is shifted by using a so-called “scissor operator”, in accordance with the suggestion
of Fiorentini and Balderschi [46]. The difference between the LDA and experi-
mental band gaps (DE) scales with the optical dielectric constant (DE ffi 9:1=e1 )
[26, 27].
Recently, Tran and Blaha [47] proposed a new exchange-correlation potential,
called the modified Becke-Johnson density functional (MBJLDA), which combines
the Becke-Johnson exchange potential and the local density approximation (LDA)
correlation potential in DFT band structure calculations. The MBJLDA functional
is an exchange-correlation (XC) potential that is obtained as the functional
derivative of the XC-energy functional EXC with respect to the electron density n(r)
MBJ
(Vxc ¼ dEXC ½nðrÞ=dnðrÞ) taken from LDA. The computational cost of DFT with
the MBJLDA functional is comparable with DFT-LDA and DFT-GGA. It yields
bandgaps which are in good agreement with experimental data [12, 13] and
empirical pseudopotential theory [14]. We calculated the electronic properties of
CdX and ZnX (X = S, Se, Te) II–VI compounds and their ternary/binary
heterostructures using the WIEN2K [48] simulation package that uses the density
functional theory (DFT) with MBJLDA functional in the band structure
calculations.
Figure 1.20 exhibits the crystal structure of CdSe and ZnS compounds. CdX and
ZnX (X = S, Se, Te) have a zinc-blende crystal structure (space group F-43 m No:
216) in which the Cd/Zn atoms are located at (0, 0, 0) and S (or Se) at (0.25, 0.25,
0.25). Converged results are obtained using 10,000 k points in the first Brillouin
zone with RMTKmax = 8.50, where RMT represents the smallest muffin-thin radius
and Kmax is the maximum size of the reciprocal lattice vectors. LSDA, GGA and
MBJLDA, as implemented in the WIEN2K simulations package, are used for
exchange and correlation potentials in the calculations. SCF iterations are repeated
until the total energy converges to a point less than 10−4 Ryd. The WIEN2K
simulations with the MBJLDA functional embedded in DFT yields a remarkable
improvement over LDA and GGA in the calculation of the structural properties
(e.g., lattice constants and bulk modulus) and electronic structure properties (e.g.,
bandgaps) of the compounds (see Tables 1.6 and 1.7). However, we should point
1 Modelling of Heterostructures for Low Dimensional Devices 35

Fig. 1.20 Crystal structure of ZnS and CdSe compounds

Table 1.6 Lattice constant (a0), bulk modulus (B0) and its pressure derivative (B00 ) for CdX
(X = S, Se, Te) and ZnX (X = S, Se, Te) calculated using WIEN2K, with DFT-MBJLDA
Property CdS CdSe CdTe ZnS ZnSe ZnTe
a0(A0) 5.779a 5.994 6.425 5.470 5.540
5.985
5.839b 6.088 6.500 5.460 5.690
6.039
5.810c 6.084 6.480 5.320 5.669
6.030
(5.830)d 6.089
B0 68.765 53.799 47.778 71.474 82.379 54.038
66.01 56.79 46.76 91.71 73.91 55.56
64.30 55.53 44.50 77.3 62.50 50.90
– – – – –
B′ 4.375 2.364 5.200 3.550 3.646 4.381
4.20 4.470 – 5.10
4.31 5.00 5.00 –
– –
a DFT-MBJLDA, b 2NN sp3s* TBM, c NN sp3d5 TBM, (…)d: Measured lattice constant

out that the MBJLDA only underestimates the band gaps between −2.50 and −7 %
for CdTe, ZnS, ZnSe, ZnTe and overestimates up to *4 % for CdS and CdSe. In
most cases (except ZnS), MBJLDA is in better agreement with experiment than
GW and G’W’. Bandgap energies at L, C and X symmetry points, calculated by
using WIEN2K simulations package, with the MBJLDA functional embedded in
36 H. Hakan Gürel et al.

Table 1.7 Comparison of calculated bandgap energies of CdX (X = S, Se, Te) and ZnX (X = S,
Se, Te)
(eV) CdS CdSe CdTe ZnS ZnSe ZnTe
EgC 2.660a 1.89 1.56 3.66 2.67 2.22
2.476b 1.887 1.59 3.680 2.83 2.39
c
2.555 1.937 0.76 3.702 1.91 1.53
(2.55) (1.82) (1.60) (3.82) (2.82) (2.39)
EgX 4.28 3.40 2.36 4.48 3.61 2.45
3.983 3.097 2.46 4.810 3.99 2.80
5.193 3.027 1.83 4.641 2.14 1.62
EgL 4.71 4.04 3.03 4.42 3.78 2.71
4.341 3.784 3.57 5.103 4.54 3.43
4.696 3.779 4.76 5.190 5.04 3.34
a DFT-MBJLDA, b 2NN sp3s* TBM, NN sp3d5 TBM, (…): Measured fundamental bandgap

DFT, and using NN sp3d5 and 2NN sp3s* TB models, are given in Tables 1.6 and
1.7 for CdS, CdSe, CdTe, ZnS, ZnSe and ZnTe II–VI compounds.
As shown in Table 1.6, MBJLDA functional based DFT calculations of struc-
tural properties of compound semiconductors (such as lattice constant, bulk mod-
ulus and its pressure derivative) are in good agreement with the experimental data
[12]. Furthermore, as shown in Table 1.7, WIEN2K simulations with MBJLDA
embedded in DFT are in good agreement with the experimental data in terms of
band gaps. Figures 1.21 and 1.22 demonstrate use of MBJLDA functional in
WIEN2K simulations package in calculating the electronic band structure of
compound semiconductors yields much better results than those of LSDA for CdX
(X = S, Se, Te) and GGA band structures for ZnX (X = S, Se, Te). Accurate
modeling of band structure properties have a key role in the design of cadmium and
zinc chalcogenides based device applications. According to current calculations
MBJLDA performed better than other conventional DFT functionals and GW
approximations to calculate band structure. For this reason, MBJLDA is a very
effective method and it can be used for a wide range of semiconductors. Accurate
modelling of band structure has a key role in the design of cadmium and zinc
chalcogenides-based device applications.
The comparison of the band structure dispersion curves obtained by using the
WIEN2K simulation package with MBJLDA functional embedded in DFT and the
semiempirical tight binding theories with NN sp3d5 and 2NN sp3s* orbitals sets is
shown in Fig. 1.23 for CdS and ZnS and in Fig. 1.24 for CdSe, CdTe and ZnSe
compound semiconductors, respectively. As displayed in Figs. 1.23 and 1.24, the
MBJLDA functional based DFT and semiempirical NN sp3d5 and 2NN sp3s*
orbitals sets TB models accurately reproduce the band gaps and band dispersion
curves at high symmetry points when they are compared with empirical pseu-
dopotential calculations for bulk II–VI compounds and their heterostructures. The
three different approaches almost equally well reproduce the band gaps at Γ, X and
L high symmetry points and valence band and conduction band dispersion curves in
1 Modelling of Heterostructures for Low Dimensional Devices 37

Fig. 1.21 Band structure of CdS, CdSe and ZnS obtained using WIEN2K with DFT-MBJLDA
(solid lines) and DFT-LSDA (dashed lines)

the entire Brillouin zone of bulk CdX and ZnX (X = S, Se, Te) II–VI compounds
and their heterostructures CdZnS/CdS, CdSTe/CdTe and ZnSSe/ZnSe.
We now discuss the use of DFT with MBJLDA functional and of the
semiempirical NN sp3d5 and 2NN sp3s* TB models to calculate the electronic band
structures of ZnSSe/ZnSe and CdSTe/CdTe heterostructures. The predicted com-
position and interface strain effects on the fundamental band gaps of ZnSSe and
CdSTe in ZnSSe/ZnSe and CdSTe/CdTe heterostructures are shown in Fig. 1.25
(left, right), respectively. The calculations are carried out using the WIEN2K
simulations based on DFT with the MBJLDA functional (dark-solid line) and the
semiempirical NN sp3d5 TB model (red-solid lines) and 2NN sp3s* TB models
(dashed-blue lines) with tight binding parameters given in Tables 1.3, 1.4 and 1.5,
integrated with the statistical thermodynamic model. (Since the measured band gaps
are near 0 K, we ignored the logarithmic term, which accounts for the
electron-phonon interactions for temperature dependence.) We can state that strain
effect on the fundamental bandgaps can be quite large when the interface strain
increases for large deformation potential and high alloy composition. The predicted
band gaps, especially at the C point, are in excellent agreement with experiment
[12].
38 H. Hakan Gürel et al.

Fig. 1.22 Band structure of ZnS, ZnSe and ZnTe obtained using WIEN2K with DFT-MBJLDA
functional (solid lines) and DFT-GGA (dashed lines)

In conclusion, as demonstrated in Figs. 1.23 and 1.24 for CdX and ZnX (X=S,
Se, Te) binaries and in 1.25 for ZnSSe and CdSTe ternaries of ZnSSe/ZnSe and
CdSTe/CdTe heterostructures, the WIEN2K simulations with DFT-MBJLDA
functional and semiempirical NN sp3d5 and 2NN sp3s* TB methods described in
this work accurately reproduce the band gaps and band dispersion curves at high
symmetry points of Brillouin zone when they are compared with empirical pseu-
dopotential calculations for these binary and ternary semiconductors. The three
different approaches almost identically reproduce the band gaps at the C, X and L
high symmetry points, as well as the valence band and conduction band dispersion
curves in the entire Brillouin zones of the bulk CdX and ZnX (X = S, Se, Te)
binaries and ZnSSe and CdSTe ternaries of ZnSSe/ZnSe and CdSTe/CdTe
heterostructures, respectively.
As can be seen from Fig. 1.25, the tight binding model with an sp3d5 orbital
basis, including the NN interactions of cation and anion atoms and spin-orbit
coupling of p-states, is adequate to accurately reproduce the band gaps at the C, X
and L symmetry points and both the valence band and conduction band dispersion
curves, obviating the need for any fictitious excited s* state in the TB model
calculations. When the two semiempirical NN sp3d5 and 2NN sp3s* tight binding
parameterizations are compared with each other, they both give reasonable
1 Modelling of Heterostructures for Low Dimensional Devices 39

Fig. 1.23 Band structures of CdS and ZnS obtained using WIEN2K with DFT-MBJLDA
(dark-solid lines), NN sp3d5 TB (red-solid line) and 2NN sp3s* TB (dashed-blue line) models

descriptions of the energy levels in the vicinity of the bottom of the conduction
band and the top of the valence band of both binary and ternary II–VI compounds.
However, the NN sp3d5 TB model does better than the 2NN sp3s* TB model in
accurately reproducing band gaps as well as valence band and conduction band
dispersion curves fitted to the empirical pseudopotential calculations at the L high
symmetry point of these compounds. Furthermore, since the inclusion of the fic-
titious s* excited state is done by modelling the average of p-d interactions, the
2NN sp3s* TB parametrization does not permit the inclusion of excited d-orbitals to
an sp3 basis set. Consequently, the actual behavior and contribution of excited
d-states in the band structure calculations is reliably and accurately reflected in the
NN sp3d5 TB model. We should point out that both the NN sp3d5 and 2NN sp3s*
TB models are semiempirical and they rely on a good description of the band
structures that are produced exactly by WIEN2K simulations package with
MBLDA functional embedded in DFT [47, 48] and empirical pseudopotential
theory [14].
40 H. Hakan Gürel et al.

Fig. 1.24 Band structures of CdSe, CdTe and ZnSe obtained using WIEN2K with DFT-MBJLDA
(dark-solid lines), NN sp3d5 TB (red-solid line) and 2NN sp3s* TB (dashed-blue line) models

1.5 Modeling of Band Offsets in Heterostructures

One of the key issues in low dimensional heterostructure semiconductor device

technology is the understanding the formation and determining the magnitude of
conduction and valence band structure across heterointerfaces (e.g., band offsets),
shown in Fig. 1.26.
The potential barriers dominate various device properties such as injection
efficiency in heterobipolar devices (e.g., HBTs) and carrier confinement in unipolar
devices (e.g., MODFETs), and have received considerable attention among device
scientists and engineers over the years. When the semiconductor composition
changes abruptly at interface between constituents, the difference in their energy
bands is accommodated by the discontinuities in the conduction and valence bands
across the interface. In analyzing charge transport in bipolar and unipolar
heterostructure devices, the crucial problem is to determine the formation and
magnitude of potential barriers relating to conduction bands and valence bands
across the heterointerface. In the context of the semiempirical tight binding model
1 Modelling of Heterostructures for Low Dimensional Devices 41

Fig. 1.25 Strain effects on the bandgaps of ZnSSe and CdSTe in ZnSSe/ZnSe and CdSTe/CdTe
heterostructures as a function of composition, obtained using the WIEN2K package with
DFT-MBJLDA (dark-solid line), and the NN sp3d5 TB (red-solid line) and 2NN sp3s* TB
(dashed-blue line) models for the TB parameters given in Tables 1.3, 1.4 and 1.5

Fig. 1.26 Schematic view of band diagram of Type I (left) and Type II (right) heterostructures

discussed in Sect. 1.2, the valence band offset across an ABC/AC ternary/binary
heterostructure can be obtained by taking the difference between the valence band
energies of the constituents that are screened with the optical dielectric constants of
the constituents [11]:
42 H. Hakan Gürel et al.



Ev Ev
DEv ¼  ; ð1:27Þ
e1 BC e 1 ABC

where Ev ¼ Ev ðC15 Þ is top of the valence band at the Γ high symmetry point and
e1 ðABCÞ and e1 ðBCÞ are the optical dielectric constants of the ABC ternary and
BC binary constituents of heterostructure, respectively:

e1 ðABCÞ ¼ xe1 ðACÞ þ ð1  xÞe1 ðBCÞ þ  xð1  xÞdc ðe1 ðBCÞ  e1 ðACÞÞ;
ð1:28Þ

The conduction band offset for a given Γ, L and X high symmetry point of the
Brillouin zone is given as the difference between the respective band gap differ-
ences, written as [11]:

DEci ¼ EciABC  EciBC ¼ DEgi  DEv ; ð1:29Þ

where Eci ¼ EC6c ; EL6c and EX6c are the bottom of the conduction bands at the Γ, L
and X high symmetry points, determined using the semiempirical tight binding
theory with the sp3, sp3s*, sp3d5s* or sp3d5orbitals sets. Here DEgi ¼ Egi ðABCÞ 
Egi ðBCÞ is the difference between the band gaps of ABC ternary and BC binary
compounds with band gaps Egi ðABCÞ and Egi ðBCÞ, where Egi ¼ EgC ; EgL and EgX
are the principal bandgaps at the Γ, L and X symmetry points.
Once the valence band energies and band gaps at high symmetry points are
known, it is easy to determine the conduction band offsets of a given heterostructure
from (1.29).
The potential barrier in the conduction band and the potential step in the valence
band at the interface significantly influence carrier transport and device performance
[2], making reliable and accurate modeling of interface energy band structure
crucial to designing the functionality of group III–V and II–VI heterostructures and
their optimization in low dimensional semiconductor electronic and optical devices.
Here, we focus on modeling the effects of composition and interface strain on
electronic band structure, band gaps and band offsets of ternary/binary group III–V
and II–VI compounds based heterostructures for which experimental data are
available for comparison. In this, we compare the predictions of the nearest
neighbor (NN) sp3d5 tight binding model, with spin-orbit coupling of p-states, to
those of the widely used second nearest neighbor (2NN) sp3s* model, with
spin-orbit coupling of p-states. Tables 1.3, 1.4 and 1.5 give the number of tight
binding parameters for the 2NN sp3s* TB and NN sp3d5 tight TB models used in the
calculations. Figure 1.27 shows the composition and strain variations of the band
gap difference and valence band offsets at lattice matched AlGaAs/GaAs and
pseudomorphic InGaAs/GaAs heterostructures.
1 Modelling of Heterostructures for Low Dimensional Devices 43

Fig. 1.27 Composition effects on bandgap differences and valence band offsets of lattice matched
AlGaAs/GaAs (left) and strained InGaAs/GaAs (right) heterostructures

The band offsets in III–V nitrides based ternary/binary heterostructures are

shown in Fig. 1.28 for AlGaN/GaN, InGaN/GaN and GaAsN/GaAs as functions of
interface strain for the entire composition range (0  x  1). As shown in Fig. 1.28,
the interface strain effect on the valence band offsets in AlGaN/GaN (left), InGaN/
GaN (middle) and GaAsN/GaAs (right) dilute nitrides based heterostructures is
rather small because of the smaller valence band deformation potentials. However,
the interface strain effects on conduction band offsets at the Γ, L and X high
symmetry points can be considerably larger because the conduction band defor-
mation potentials are large.
The predicted valence band offsets are in good agreement with experimental
findings [12, 13]. As shown in Figs. 1.27 and 1.28, strain effects on the band gap
differences and on the conduction band offsets can be quite large since the lattice
mismatch increases with an increase in the ternary mole fraction. As we pointed out
earlier, there is a considerable lattice mismatch across many of the ternary/binary
heterointerfaces. Interface strain due to lattice mismatch causes a shift in the lattice
44 H. Hakan Gürel et al.

Fig. 1.28 Predicted interface strain effects on the valence band offsets of AlGaN/GaN (left),
InGaN/GaN (middle) and GaAsN/GaAs (right) heterostructures

constant of the epilayer: a ¼ ð1 þ eÞa0 , where e is the symmetric strain tensor.

Therefore, the bond lengths and on-site and off-site TB matrix elements will be
modified with strain. Consequently, the electronic properties of heterostructures
such as conduction and valence band offsets will be modified with respect to their
unstrained values. As an example, Fig. 1.29 shows the composition and strain
effects on conduction and valence band offsets of CdSTe/CdTe and ZnSSe/ZnSe
heterostructures calculated by using the WIEN2K simulation package based on the
DFT with MBJLDA functional and semiempirical NN sp3d5 and 2NN sp3s* TB
models for optimized tight binding parameters given in Tables 1.3, 1.4 and 1.5 As
can be seen in this comparison, the interface strain effects on the conduction band
offsets of ZnSSe/ZnSe and CdSTe/CdTe heterostructures can be quite large when
the interface strain increases for large the deformation potential and high alloy
composition.
1 Modelling of Heterostructures for Low Dimensional Devices 45

Fig. 1.29 Predicted composition and interface strain effects on the band offsets of ZnSSe/ZnSe
and CdSTe/CdTe heterostructures, obtained using the WIEN2K with DFT-MBJLDA (dark-solid
line) compared with the NN sp3d5 (red-solid line) and 2NN sp3s*(dashed-blue line) TB models for
parameters given in Tables 1.1 and 1.2 of [30]

1.6 Conclusion

Ever since the invention of the first transistor in 1949, device scientists and engi-
neers have witnessed amazing achievements in the semiconductor science and
technology. Advances in the growth of semiconductor thin films with layer
thickness approaching atomic dimensions and fabrication of low dimensional
electronic and optical devices has provided new opportunities in fundamental sci-
ence and technology of semiconductors and semiconductor devices. Such
advancement could not have been possible without a qualitatively reliable under-
standing of the basic physics of semiconducting materials and quantitatively precise
potential predictions and performance of devices, leading to new concepts in the
semiconductor growth that allowed previously many unknown devices with more
complex functionality and much higher densities for electronic and optical
applications.
46 H. Hakan Gürel et al.

This review has presented comparative predictions of electronic band structures

of group III–V and II–VI compounds and their heterostructures obtained using the
semiempirical tight binding theory based on the 2NN sp3, 2NN sp3s*, NN sp3d5s*
and NN sp3d5 orbitals sets in comparison with the WIEN2K simulations package
based on the density functional theory (DFT) with the modified modified
Becke-Johnson exchange-correlation potential with the local density approximation
(LDA), called MBJLDA functional. We conclude that DFT with MBJLDA func-
tional and the semiempirical NN sp3d5 TB model are in good agreement with
experimental data for band gaps and band offsets and have considerable potential in
the design and optimization of group III–V and II–VI compounds and their ternaries
used for the fabrication of electronic and optical devices. When compared to the
2NN sp3s* TB model, NN sp3d5 TB parametrization better simulates both the
valence band and conduction band dispersion curves. The actual contribution of the
excited d-states in the band structure calculations is reliably reproduced in the case
of NN sp3d5 orbitals set tight binding parameterization. We conclude that DFT with
MBJLDA and semiempirical NN sp3d5 TB model show good agreement with
experiment for band gaps and band offsets and has a considerable potential in the
design and optimization of group III–V and II–VI compounds and their ternaries
used for the fabrication of low dimensional electronic and optical devices.

References

1. H. Ünlü, N.J.M. Horing (eds.), Low Dimensional Semiconductor Structure: Characterization

and Applications (Springer, Heidelberg, 2013)
2. H. Morkoç, H. Ünlü, G. Ji, Principles and technology of MODFETs. Wiley 1, 2 (1991)
3. H. Ünlü, H. Morkoç, Solid State Technol. 31, 83 (1988)
4. H. Morkoç, H. Ünlü, in Semiconductor and Semimetals, vol. 24 ed. by R. Dingle (1987),
p. 135
5. H. Ünlü, H. Morkoç, and S. Iyer, in Gallium Arsenide Technology, vol. 2 ed. by D.K. Ferry
(1990), p. 231
6. H. Morkoç, H. Ünlü, H. Zabel, N. Otsuka, Solid State Technol. 31, 71 (1988)
7. H. Kroemer, Proc. IRE 45, 1535 (1957)
8. H. Kroemer, Rev. Mod. Phys. 73, 783 (2000)
9. N. Lucas, H. Zabel, H. Morkoç, H. Ünlü, Appl. Phys. Lett. 52, 2117 (1988)
10. H. Ünlü, Solid State Electron. 35, 1343 (1992)
11. H. Ünlü, Phys. Status Solidi B 216, 107 (1999); B 223, 195 (2001); B 229 581 (2002); B 235,
248 (2003)
12. O. Madelung (ed), Numerical Data and Functional Relationships in Science and Technology,
17a, Springer, Berlin (1982); ibid 17d, Springer, Berlin (1984)
13. Vurgaftman, I., Meyer, J.R. ve Ram-Mohan, L.R. (2001), J. Appl. Phys. 89, 5815 (2001)
14. M.L. Cohen, J.R. Chelikowsky, Electronic Structure and Optical Properties of
Semiconductors (2nd edn), Springer, Berlin (1989)
15. C. Pryor, J. Kim, L.M. Wang, A.J. Willamson, A. Zunger, J. Appl. Phys. 83, 2548 (1998)
16. P. Keating, Phys. Rev. 145, 637 (1966)
17. K. Shim, H. Rabitz, Phys. Rev. B 57, 12874 (1998)
18. M. Rabah, B. Sahraui, B. Bouhafs, B. Abbar, H. Abid, Phys. Stat. Solidi B 238, 156 (2003)
19. H.H. Gürel, Ö. Akıncı, H. Ünlü, Comput. Mater. Sci. 33, 269 (2005)
1 Modelling of Heterostructures for Low Dimensional Devices 47

20. Ö. Akıncı, H.H. Gürel, H. Ünlü, Thin Solid Films 511–512, 684 (2006)
21. H.H. Gürel, Ö. Akıncı, H. Ünlü, Superlattices and Microstructures, 40 (4–6), 588 (2006)
22. H.H. Gürel, Ö. Akıncı, H. Ünlü, Physica Status Solidi (c) 4(2), 316 (2007)
23. Ö. Akıncı, H.H. Gürel, H. Ünlü, Physica Status Solidi (c) 5(2), 478 (2008)
24. H.H. Gürel, Ö. Akıncı, H. Ünlü, Thin Solid Films 516, 7098 (2008)
25. Ö. Akıncı, H.H. Gürel, H. Ünlü, J. Nanoscience Nanotechnology 8, 540 (2008)
26. Ö. Akıncı, Ö., H.H. Gürel, H.H. and H. Ünlü, Thin Solid Films, 517, 2431 (2009)
27. H.H. Gürel, O. Akıncı, H. Ünlü, Superlattices Microstruct. 51, 725 (2012)
28. H.H. Gürel, H. Ünlü, Mater. Sci. Semicond. Process. 16, 1619 (2013)
29. J.L. Martins, A. Zunger, Physical Review B 30, 6217 (1984)
30. P. Hohenberg and W. Kohn, Phys. Rev. 136 B, 864 (1964)
31. J.C. Slater and G. F. Koster (1954), Physical Review, 94, 1498 (1954)
32. D.J. Chadi, M.L. Cohen, Physica Status Solidi 68, 405 (1975)
33. D.J. Chadi, Physical Review B 16, 790 (1977)
34. W.A. Harrison, Electronic Structure and the Properties of Solids, Freeman (1980)
35. D.N. Talwar, C.S. Ting, Physical Review B 25, 2660 (1982)
36. J.P. Loehr, Talwar, Phys. Rev. B 55, 4353 (1997)
37. T.B. Boykin, L.J. Gamble, G. Klimeck, R.C. Bowen, Phys. Rev. B 59, 7301 (1999)
38. Y. Fu, K.A. Chao, Phys. Rev. B 43, 4119 (1991)
39. A.D. Carlo, Semiconductor Sci. and Tech. 18, R1 (2001)
40. E.P. O′Reilly, A. Lindsay, A., S. Tomic, M.K. Saadi, Semicond. Sci. Technol. 17, 870 (2002)
41. P. Vogl, H.P. Hjalmarson, J.D. Dow, J. Phys. Chem. Solids 44, 365 (1983)
42. M. Jancu, F. R.Scholz, F. Beltram, F. Bassani, Phys. Rev. B 57, 6493 (1998)
43. J.M. Jancu, F. Bassani, F. Della Sala, R. Schols, Appl. Phys. Lett. 81, 4838 (2002)
44. S. Sapra, N. Shanthi, D.D. Sarma, Phys. Rev. B 66, 205202 (2002)
45. W. Kohn, J.L. Sham, Phys. Rev. 140A, 1133 (1965)
46. V. Fiorentini, A. Baldereschi, Phys. Rev. B 51, 17196 (1995)
47. F. Tran, P. Blaha, Phys. Rev. Lett. 102, 226401 (2009)
48. P. Blaha, K. Schwarz, G.K.H. Madsen, D. Kvasnicka, J. Luitz, WIEN2k, An Augmented Plane
Wave Plus Local Orbitals Program For Calculating Crystal Properties, Vienna University of
Technology, Vienna, Austria, 2001
Chapter 2
Aspects of the Modeling of Low
Dimensional Quantum Systems

Norman J.M. Horing

Abstract The mathematical modeling of low dimensional quantum systems is

discussed in this chapter. In particular, the use of generalized functions in such
modeling is illustrated in some detail, including applications of the Dirac delta
function and its derivative (“delta-prime”) in determining quantum mechanical
Schrödinger Green’s functions describing the dynamics of various low dimensional
systems. The illustrations include quantum dots, wires and wells (and a superlattice)
in various dimensions. Also, the one-dimensional “delta-prime” potential is shown
to provide an impenetrable barrier.

2.1 Introduction

The advent and rapid development of the fabrication of low dimensional semi-
conductor materials, replete with the promise of nanostructures upon which a whole
new generation of quantum electronic, optical and computational devices can be
based, has stimulated an enormous effort to explore the physical properties of such
materials, and how they can be manipulated to greatest advantage. Practically all the
fields of science and engineering are involved in this massive effort throughout the
world. Mathematical modeling has an important role in this matter, enabling highly
tractable analyses that provide insight into the quantum dynamical behavior of
nanostructures and their possible optimization. In this chapter on low dimensional
quantum systems, in which the direction/range of particle motion is sharply limited
by deep (strong) wells (barriers) in its potential profile, the role of generalized
functions, in particular the Dirac delta function, will be discussed in model cal-
culations of the properties of such structures in three, two, one and zero dimensional
systems. The generalized functions are useful in representing potentials that confine

N.J.M. Horing (&)

Department of Physics and Engineering Physics, Stevens Institute of Technology,
Hoboken, New Jersey 07030, USA
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 49

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_2
50 N.J.M. Horing

electrons into spatially small subsystems, such as quantum dots, wires, wells and
lattices thereof, lodged in a host medium that may itself be confined in lower
dimensions; for example, quantum dots can occur in bulk three-dimensional
structures, but can also occur on a thin planar two-dimensional layer and on a
one-dimensional quantum wire. Such electron confinement by a potential usually
has some wave function extension outside the spatial boundaries of the potential,
associated with tunneling phenomenology. However, if the confining potential is so
strong (high) that such external wavefunction extension is negligible, then an
object, such as a thin quantum well, may be viewed as zero dimensional in the
confinement direction(s), and may be reasonably represented by a generalized
function in regard to that spatial direction(s). Both of these important uses of
generalized function models, in particular Dirac delta functions, in representing
(i) confining potentials, and, also, (ii) spatial extension of a low dimensional narrow
quantum well, will be discussed here.
The illustrative model calculations to be discussed here include the use of Dirac
delta function confining potentials in the determination of Schrödinger Green’s
functions for:
(i) a zero dimensional quantum dot in a 3D bulk medium permeated by an
isotropic simple harmonic oscillator potential,
(ii) double quantum dot systems lodged in a thin 2D layer, also in a thin 1D wire,
(iii) a one dimensional quantum wire on a thin 2D layer in a normal magnetic field.
Also, we present the example of (iv) the space-time inversion of the dynamic,
nonlocal dielectric function of a type 1 superlattice. Finally, the physical features of
the “delta-prime” derivative of a Dirac-delta function potential will be treated as
well.
The calculational procedure generally followed here involves rewriting the
differential Schrödinger Green’s function equation for the system at hand as an
integral equation in terms of the decoupled Green’s function without special low
dimensional features (such as quantum dots, wires, etc.). The resulting Dyson-type
integral equation involves the decoupled Green’s function as its driving term and
also as part of its kernel. The kernel also involves the special low dimensional
features described in terms of generalized functions (such as the Dirac delta
function and its derivative), which considerably simplify the equation, so that it
may be solved algebraically or by a spatial Fourier transform in directions of
translational invariance. The associated energy spectrum is discussed in terms of the
frequency poles of the resulting Green’s function. In regard to the inversion of the
dynamic, nonlocal dielectric function of a superlattice, it is facilitated by an
“ansatz” expressing its spatial structure in terms of Dirac delta functions. The
resulting superlattice structure factor describes a continuous band of superlattice
plasmons.
2 Aspects of the Modeling of Low Dimensional Quantum Systems 51

2.2 3D Harmonic Oscillator with a Superposed Quantum

Dot (Fig. 2.1)

To start, we consider a three dimensional isotropic Schrödinger oscillator in a bulk

medium, which, by itself, has been viewed as a quantum dot capable of having
many electrons in the plethora of its many states. The associated retarded Green’s
function GHO h ! 1)
3D satisfies (



@ 1 2 mx20 2 HO
i þ r  r G3D ðr; r0 ; t; t0 Þ ¼ dð3Þ ðr  r0 Þdðt  t0 Þ;
@t 2m 2

and is given in position-time representation as [1]

0
3D ðr; r ; TÞ
GHO
 3=2  
mx0 imx0
¼ g þ ðTÞ exp ½ðr  r þ r0  r0 Þ cos x0 T  2r  r0  ;
2pi sin x0 T 2 sin x0 T
ð2:1Þ

where r ¼ ðx1 ; x2 ; x3 Þ, T ¼ t  t0 , and x0 is the natural frequency of the oscillator

of mass m (g þ ðTÞ ¼ g þ ðt  t0 Þ is the Heaviside unit step function). The role of a
superposed quantum dot (at r0 ) in the electron energy spectrum and propagation is
represented by the addition of a potential term of the form

UðrÞ ¼ adð3Þ ðr  r0 Þ; ð2:2Þ

where dð3Þ ðr  r0 Þ is a three dimensional Dirac delta function, and a ¼ U0 V\0
describes the product of the quantum well depth U0 times its volume V. The
retarded Green’s function Gðr; r0 ; xÞ for electron propagation in this bulk medium
satisfies the differential equation (frequency/energy representation).


1 2 mx20 2
xþ r  r  UðrÞ Gðr; r0 ; xÞ ¼ dð3Þ ðr  r0 Þ: ð2:3Þ
2m 2

Rewriting this as an integral equation in three dimensions, we have

Z
0 0
Gðr; r ; xÞ ¼ G3D ðr; r ; xÞ þ
HO
d 3 r00 GHO 00 00 00 0
3D ðr; r ; xÞUðr ÞGðr ; r ; xÞ; ð2:4Þ

00
where GHO3D is given by (2.1). Introducing (2.2) on the right of (2.4), the r -inte-
gration leads to
52 N.J.M. Horing

Fig. 2.1 Schematic of a 3D

harmonic oscillator potential
(red) with a Dirac delta
function potential well
(green) at r0

Gðr; r0 ; xÞ ¼ GHO 0 0
3D ðr; r ; xÞ þ aG3D ðr; r0 ; xÞGðr0 ; r ; xÞ:
HO
ð2:5Þ

To solve, set r ¼ r0 in (2.5), with the result

1
Gðr0 ; r0 ; xÞ ¼ GHO ðr0 ; r0 ; xÞ; ð2:6Þ
1 aG3D ðr0 ; r0 ; xÞ 3D
HO

whence the full Green’s function for the “delta-dot”superposed on a 3D oscillator is

given by

1
Gðr; r0 ; xÞ ¼ GHO 0
3D ðr; r ; xÞ þ aG3D ðr; r0 ; xÞ
HO 0
3D ðr0 ; r ; xÞ:
GHO
1  aGHO
3D ðr0 ; r0 ; xÞ
ð2:7Þ

The new set of coupled electron energy states occurring as frequency poles of
GðxÞ is introduced by the vanishing of the denominator of the last term on the right
of (2.7),

1 ¼ aGHO
3D ðr0 ; r0 ; xÞ: ð2:8Þ

Fourier transorming (2.1) from time to frequency/energy representation, we have

Z1  3=2  
mx0 imx0 r02
1¼a dTeixT exp ½cos x0 T  1 : ð2:9Þ
2pi sin x0 T sin x0 T
0
2 Aspects of the Modeling of Low Dimensional Quantum Systems 53

The right side of (2.9) is readily expanded in the eigenfunctions /n ðrÞ of the 3D
harmonic oscillator hamiltonian alone as (x ! x þ i0 þ ; En are the associated
decoupled oscillator eigenenergies)

Z1 X X j/ ðrÞj2
1¼a dTeixT ðiÞj/n ðrÞj2 eiEn T ¼ a n
: ð2:10Þ
n n
x  En
0

(An alternative evaluation could be developed from the generating function of

Laguerre functions [2].) As the 3D oscillator energy spectrum En is discrete (and
uniformly spaced), the structure of the coupled “delta-dot” eigenenergy dispersion
relation of (2.10) makes it clear that the new coupled set of energy roots will occur
shifted from each value of En, with n ¼ ðn1 ; n2 ; n3 Þ referring to the three sets of
integers corresponding to the three spatial directions of the bulk oscillator medium.

2.3 Double Quantum Dot Systems in Low Dimensions

Considering a GaAs semiconductor inversion layer or other 2D system (e.g. gra-

phene) in which electron motion is confined to the x–y plane at z ¼ 0, we examine
the role of a double quantum dot induced by a 2D double-well potential (a \0)
constituted of Dirac delta functions:
X
UðrÞ ¼ a þ dð2Þ ðr  r þ Þ þ a dð2Þ ðr  r Þ ¼ a dð2Þ ðr  r Þ; ð2:11Þ


where r ¼ x^i þ y^j and the two potential wells defining the double dot are located at
r ¼ ^id=2 with strengths a ¼ U0 A \0 representing the product of the well
depths U0 at r times their common area A  a2 . The retarded Green’s function
for electron propagation on the sheet, Gðr1 ; r2 ; xÞ, including the dot regions, obeys
the differential Schrödinger equation (frequency representation)

  
1 @2 @2
xþ þ  Uðr1 Þ Gðr1 ; r2 ; xÞ ¼ dð2Þ ðr1  r2 Þ: ð2:12Þ
2m @x2 @y2

Alternatively, this may be rewritten as a two dimensional integral equation as

Gðr1 ; r2 ; xÞ ¼ G02D ðr1 ; r2 ; xÞ

Z
ð2:13Þ
þ d 2 r3 G02D ðr1 ; r3 ; xÞUðr3 ÞGðr3 ; r2 ; xÞ;

where G02D is the infinite sheet two dimensional Schrödinger Green’s function for
electron propagation on the sheet in the absence of the double quantum dot
potential wells. Employing (2.11) and integrating with respect to r3 , we have
54 N.J.M. Horing

Gðr1 ; r2 ; xÞ ¼ G02D ðr1 ; r2 ; xÞ

X
þ a G02D ðr1 ; r ; xÞGðr ; r2 ; xÞ: ð2:14Þ


Setting r1 ¼ r here, Gðr ; r2 ; xÞ is readily obtained algebraically, with the

result

Gðr ; r2 ; xÞ
D; G02D ðr ; r2 ; xÞ þ D; G02D ðr ; r2 ; xÞ ð2:15Þ
¼ ;
D; D;  D; D;

where

D; ¼ d;  a G02D ðr ; r ; xÞ ð2:16Þ

and

D; ¼ a G02D ðr ; r ; xÞ: ð2:17Þ

The final double-dot Green’s function is given by (2.14)–(2.17) as

X
Gðr1 ; r2 ; xÞ ¼ G02D ðr1 ; r2 ; xÞ þ a G02D ðr1 ; r ; xÞ

ð2:18Þ
D; G02D ðr ; r2 ; xÞ þ D; G02D ðr ; r2 ; xÞ
 :
D; D;  D; D;

The coupled subband energy eigenstates associated with the double quantum dot
arise from the frequency poles of its Green’s function, as given by the dispersion
relation:

detðD; D;  D; D; Þ ¼ 0: ð2:19Þ

The retarded infinite sheet Schrödinger Green’s function for a spinless 2D

electron (absent dots) is well known as (
h ! 1) [3]

imðX 2 þ Y 2 Þ
m exp½ 2T 
GB2D ðr1 ; r2 ; TÞ ¼ g þ ðTÞ ; ð2:20Þ
2p T

in position-time representation, with T ¼ t1  t2 , X ¼ x1  x2 , Y ¼ y1  y2 and

g þ ðTÞ is the Heaviside unit step function. Fourier transforming from time to fre-
quency representation, and setting r1 ¼ r2 ¼ r as needed in (2.16), (X ¼ 0,
Y ¼ 0), we obtain ðx ! x þ i0 þ Þ
2 Aspects of the Modeling of Low Dimensional Quantum Systems 55

Z1
m eixT
G02D ð0; 0; xÞ ¼ dT : ð2:21Þ
2p T
0

This T-integral diverges as T ! 0 for X ¼ Y ¼ 0. The divergence is an artifact

of assuming a dð2Þ ðrÞ potential which confines the dot to a single point. It may be
removed by noting that the original integral equation involves integration over a
small, but finite radius “a”, so we should install a “smear” radius “a” putting
X 2 þ Y 2 ! a2 (instead of “0”) in (2.20), leading to the replacement
Z1
m dT ixT 
G02D ð0; 0; xÞ ! G02D ða; xÞ  e exp ima2 =2T ; ð2:22Þ
2p T
0

which is convergent. It is given by a Hankel function of the first kind, order zero
ð1Þ
(H0 ), yielding

im ð1Þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi im h pffiffiffiffiffiffiffiffiffiffiffiffiffiffi  pffiffiffiffiffiffiffiffiffiffiffiffiffiffii

G02D ð0; 0; xÞ !  H0 2ma2 x ¼  J0 2ma2 x þ iY0 2ma2 x :
2 2
ð2:23Þ

For a small (nano) quantum dot, 2ma2 x 1; this may be approximated as [4]
m pffiffiffiffiffiffiffiffiffiffiffiffiffiffi  m pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
G02D ð0; 0; xÞ
Y0 2ma2 x
ln ma2 x=2 ; ð2:24Þ
2 p

since Y0 then becomes dominant.

The energy eigenvalue dispersion relation, (2.19), for the 2D Schrödinger cou-
pled quantum double dot system is given by (note that jr  r j ¼ d;
jr  r j ! a; suppress x):
  2
1  a þ G02D ðaÞ 1  a G02D ðaÞ  a þ a G02D ðdÞ ¼ 0: ð2:25Þ

If the dots are widely separated, d  a, then G02D ðdÞ ! 0 due to the extremely
rapid variation of the time-dependent exponential on the right of (2.20). The result
is two completely independent decoupled modes given by
 
1  a þ G02D ðaÞ ¼ 1  a G02D ðaÞ ¼ 0: ð2:26Þ

However, if the separation of the dots shrinks dramatically, d ! a, then (2.25)

becomes
1  ða þ þ a ÞG02D ðaÞ ¼ 0; ð2:27Þ
56 N.J.M. Horing

characterizing a single mode dispersion relation for one quantum dot of combined
strength a þ þ a . Of course, the fully coupled mode solutions of (2.25) are con-
siderably more complicated because of the transcendental character of that dis-
persion relation.
Another interesting case is that of a double quantum dot superposed on electrons
confined to move along a straight line in just one dimension, x. The associated
Schrödinger Green’s function for free 1D electrons in the absence of the double-dot
is [3]

im  pffiffiffiffiffiffiffiffiffiffi 
G01D ðx1 ; x2 ; xÞ ¼ pffiffiffiffiffiffiffiffiffiffi exp i 2mxjx1  x2 j : ð2:28Þ
2mx

The “double-dot” Green’s function in one-dimension is given by (2.18) with the

replacement G02D ! G01D (note that G01D of (2.28) is finite when x1 ¼ x2 ). Similar
remarks apply to the dispersion relations of (2.25)–(2.27). For the “single-mode”
type limits of (2.26), (2.27), the corresponding 1D bound state is given by

iam a2 m
1 þ pffiffiffiffiffiffiffiffiffiffi ¼ 0; or x¼ ; ð2:29Þ
2mx 2

as found by Glasser, et al. (Phys. Rev. B 42, 7630 (1990)).

2.4 Quantum Wire on a 2D Sheet in a Uniform Magnetic

Field

To examine the effects of a normal magnetic field on the dynamics of a quantum

wire along the y-axis at x ¼ 0 (on the two dimensional x–y plane at z ¼ 0), the
confinement of electrons to the wire may be represented by a one-dimensional
Dirac-delta function potential as (r ¼ ðx; yÞ on the 2D sheet)

UðrÞ ¼ adðxÞ; ð2:30Þ

where a ¼ U0 d\0, with U0 as the potential well depth and d as its width. The
governing Schrödinger Green’s function equation is similar to (2.12), but with the
kinetic Hamiltonian part adjusted to include the magnetic field; and U(r) is given by
(2.30) for the quantum wire. As usual, the partial differential Schrödinger equation
can be reformulated as an integral equation given in frequency/energy representa-
tion by
Z
Gðr1 ; r2 ; xÞ ¼ GB2D ðr1 ; r2 ; xÞ þ d ð2Þ r3 GB2D ðr1 ; r3 ; xÞUðr3 ÞGðr3 ; r2 ; xÞ; ð2:31Þ
2 Aspects of the Modeling of Low Dimensional Quantum Systems 57

where GB2D ðr1 ; r2 ; xÞ is the two-dimensional magnetic field Green’s function on the
2D planar sheet, with no wire present. It involves gauge dependence and lacks
spatial translational invariance through the Peierls phase factor Cðr; r0 Þ: (AðrÞ is the
vector potential of the uniform magnetic field)
0 1
Zr
i:e
Cðr; r0 Þ ¼ exp@ dr00  Aðr00 ÞA; ð2:32Þ

hc
r0

such that GB2D has the form

GB2D ðr; r0 ; xÞ ¼ Cðr; r0 ÞG02D ðr; r0 ; xÞ; ð2:33Þ

where G02D ðr; r0 ; xÞ ¼ G02D ðr  r0 ; xÞ is effectively translationally invariant on the

2D sheet. Employing (2.30) in (2.31) yields (suppress x)
Z
Gðr1 ; r2 Þ ¼ GB2D ðr1 ; r2 Þ þ a d ð2Þ r3 GB2D ðr1 ; r3 Þdðx3 ÞGðr3 ; r2 Þ; ð2:34Þ

and executing the x3 -integration,

Gðx1 ; y1 ; x2 ; y2 Þ ¼ GB2D ðx1 ; y1 ; x2 ; y2 Þ

Z
ð2:35Þ
þ a dy3 GB2D ðx1 ; y1 ; x3 ¼ 0; y3 ÞGðx3 ¼ 0; y3 ; x2 ; y2 Þ:

Focusing attention on electron propagation along the wire with x1 ¼ x2 ¼ 0, we

Rr
note that the Peierls phase integral, r0 dr00  Aðr00 Þ, which occurs in G as well as GB2D
(and is generally dependent on the path from r0 to r in the x–y plane) in conse-
quence of gauge covariance, is now confined to a straight line path along the y-axis.
Accordingly, its role in (2.35) is eliminated and with

Gðr1 ; r2 ; xÞ ¼ Cðr1 ; r2 ÞG0 ðr1 ; r2 ; xÞ; ð2:36Þ

we have the Green’s function “on-axis” determined by

G0 ðx1 ¼ 0; y1 ; x2 ¼ 0; y2 Þ ¼ G02D ðx1 ¼ 0; y1 ; x2 ¼ 0; y2 Þ

Z
þ a dy3 G02D ðx1 ¼ 0; y1 ; x3 ¼ 0; y3 ÞG0 ðx3 ¼ 0; y3 ; x2 ¼ 0; y2 Þ:

ð2:37Þ

Since G02D is translationally invariant, this integral equation can be solved by a

single momentum transform in the y-direction (y1  y2 ! py ), yielding
58 N.J.M. Horing

1
G0 ðx1 ¼ 0; x2 ¼ 0; py Þ ¼ G0 ðx1 ¼ 0; x2 ¼ 0; py Þ:
1  aG02D ðx1 ¼ 0; x2 ¼ 0; py Þ 2D
ð2:38Þ

Finally, the Green’s function for electron propagation along the quantum wire is
determined by substituting (2.38) as the last G0 -factor on the right of (2.37)
transformed to py -representation, with the result

G0 ðx1 ¼ 0; x2 ¼ 0; py ; xÞ ¼ G02D ðx1 ¼ 0; x2 ¼ 0; py ; xÞ þ aG02D ðx1 ¼ 0; x2 ¼ 0; py ; xÞ

1
 1  aG02D ðx1 ¼ 0; x2 ¼ 0; py ; xÞ G02D ðx1 ¼ 0; x2 ¼ 0; py ; xÞ
ð2:39Þ

as a function of frequency/energy (and py ). (Off-axis propagation is more compli-

cated.) It is immediately apparent that the quantum wire energy spectrum is
determined by the zeros of the denominator of (2.39), (restore x)

1  aG02D ðx1 ¼ 0; x2 ¼ 0; py ; xÞ ¼ 0: ð2:40Þ

Note that the energy spectrum of (2.40) supplants the original spectrum asso-
ciated with the frequency poles of G02D ðx1 ¼ 0; x2 ¼ 0; py ; xÞ, since Gðx1 ¼ 0; x2 ¼
0; py ; xÞ vanishes identically when G02D ðx1 ¼ 0; x2 ¼ 0; py ; xÞ ! 1 at its pole
positions, as can be seen from (2.39).
The 2D magnetic field Landau-quantized retarded Schrödinger Green’s function
is given in position-time representation as [3, 5]


mxc imxc ðX 2 þ Y 2 Þ
G02D ðr1 ; r2 ; TÞ ¼ g þ ðTÞ exp = sinðxc T=2Þ; ð2:41Þ
4p 4 tanðxc T=2Þ

where X ¼ x1  x2 ; Y ¼ y1  y2 ; T ¼ t1  t2 ; xc is the cyclotron frequency, m is

the effective mass and g þ ðTÞ is the Heaviside unit step function. Fourier trans-
forming Y ! py and T ! x, we have

mxc X
1
emxc X =4
2

G02D ðX; py ; xÞ ¼
p n¼0 x  ½n þ 1=2xc
Z1
 dY cosðpy YÞexpðmxc Y 2 =4ÞLn ðmxc ½X 2 þ Y 2 =2Þ;
0
ð2:42Þ

where the T-integral was carried out using the integrand as the generator of a series
of Laguerre polynomials Ln ðxÞ [2]. The Y-integration is facilitated by expanding
2 Aspects of the Modeling of Low Dimensional Quantum Systems 59

L ðmxc ½X 2 þ Y 2 =2Þ
n in products
 of Hermite polynomials [6],
n
is the binomial coefficient
k

n   rffiffiffiffiffiffiffiffiffi 
mxc X1
ð1Þn emxc X =4 X
2
n mxc
G02D ðX; py ; xÞ ¼ H2n2k X
p n¼0 n! x  ½n þ 1=2xc k¼0 k 2
Z 1 rffiffiffiffiffiffiffiffiffi 
mxc
 dY cosðpy YÞexpðmxc Y 2 =4ÞH2k Y :
0 2
ð2:43Þ

The Y-integral can now be evaluated in terms of a variant of Hermite polyno-

 pffiffiffi
mials defined as Hen ðxÞ ¼ 2n=2 Hn x= 2 [7]:
rffiffiffiffiffiffiffiffiffi
mxc mxc X 2 =4 p2y =mxc X 1
ð1Þn
G02D ðX; py ; xÞ ¼ e e
p n¼0
n!ðx  ½n þ 1=2xc Þ
    rffiffiffiffiffiffiffiffiffi    ð2:44Þ
X 1
n k
n mxc 2py
 H2n2k X He2k pffiffiffiffiffiffiffiffiffi :
k¼0
2 k 2 mxc

In regard to the eigenenergy spectrum, the usual 2D Landau quantized energy

levels in the absence of a quantum wire (a ¼ 0) appear as the frequency poles of
GB2D in the first term on the right of (2.35). The coupled eigenenergies associated
with the quantum wire in a magnetic field perpendicular to the 2D sheet arise from
(2.40). For electron propagation along the wire, x1 ¼ x2 ¼ 0, the analysis above
yields the coupled (a 6¼ 0) dispersion relation as
rffiffiffiffiffiffiffiffiffi 1
 rffiffiffiffiffiffiffiffiffi2
a mxc 
hp2y =mxc X
h 2n =n!
1¼ e H n py : ð2:45Þ


h 3=2 p n¼0
mxc x  ½n þ 1=2xc

For a very narrow nanowire, a2 m=
h3 xc ¼ d 2 mU02 =
h3 xc 1, it is clear that

there is an energy root xn near each multiple of the cyclotron frequency, nxc , close
enough so that the nearest pole alone determines it as
rffiffiffiffiffiffiffiffiffi n
 rffiffiffiffiffiffiffiffiffi2
a
mxc 2 
hp2y =mxc
h
xn ¼ ½n þ 1=2xc þ 3=2 e H n py : ð2:46Þ

h p n! mxc

2.5 Inverse Dielectric Function of a Periodic Superlattice

The screening function Kð1; 2Þ [1 ¼ r1 ; t1 ¼ ðx1 ; y1 ; z1 ; t1 Þ] of a periodic superlat-

tice (Fig. 2.2) provides an interesting and useful example of the use of Dirac delta
60 N.J.M. Horing

functions in representing nano-thin systems, in particular 2D dynamic, nonlocal

plasmonic sheets. The frequency poles of Kð1; 2Þ are the superlattice plasmons and
its zero frequency limit describes static shielding. Its role is subsumed in the linear
relations
Z
dVð1Þ
Kð1; 2Þ ¼ or Vð1Þ ¼ d 4 2Kð1; 2ÞUð2Þ; ð2:47Þ
dUð2Þ

which indicate that the action of Kð1; 2Þ on an impressed potential Uð2Þ at

space-time point 2 yields the resulting effective potential Vð1Þ at 1. Its relation to
the dielectric function eð1; 2Þ,

eð1; 2Þ ¼ dð1  2Þ þ 4pa0 ð1; 2Þ; ð2:48Þ

where 4pa0 ð1; 2Þ is the polarizability, defines Kð1; 2Þ as the space-time matrix
inverse of eð1; 2Þ:
Z
d 4 3 Kð1; 3Þeð3; 2Þ ¼ d4 ð1; 2Þ ! 4D space-time unit matrix: ð2:49Þ

Alternatively expressed, Kð1; 2Þ obeys the integral equation [3]

Z Z
Kð1; 2Þ ¼ dð1  2Þ þ d 4 3 d 4 4 vð1  3ÞRð3; 4ÞKð4; 2Þ; ð2:50Þ

where Rð1; 2Þ is the density pertubation response function of the system, mð1  2Þ is
the Coulomb potential and

dqð1Þ
Rð1; 2Þ ¼ ¼ iG1 ð1; 2ÞG1 ð2; 1 þ Þ; ð2:51Þ
dVð2Þ

with G1 as the single particle noninteracting Green’s function in the random phase
approximation. It is readily seen that this corresponds to the polarizability as
Z
4pa0 ð1; 2Þ ¼  d 4 3 vð1  3ÞRð3; 2Þ: ð2:52Þ

Furthermore,
Z
eð1; 2Þ ¼ dð1  2Þ  d 4 3 vð1  3ÞRð3; 2Þ: ð2:53Þ

As an illustration of the usefulness of Dirac delta functions in representing the

spatial extension of a low dimensional system, we consider an infinite superlattice
of identical 2D planar sheets of electrons parallel to the x-y plane having period
a along the z-axis. Each sheet is characterized by a 2D density perturbation
2 Aspects of the Modeling of Low Dimensional Quantum Systems 61

Fig. 2.2 Schematic of a type

1 superlattice

response function of (2.51): Assuming that only the lowest electron energy subband
/0 ðzÞ for motion in the z-direction across the sheet n is energetically accessible, an
associated eigenfunction expansion of Rð1; 2Þ has the form
2 2
Rð1; 2Þ ! Rðx1  x2 ; y1  y2 ; t1  t2 Þ ! j/0 ðz1  naÞj j/0 ðz2  naÞj . As the
width of the sheet shrinks to zero, so does /0 ðz1  naÞ, except at z ¼ 0 where it
grows to accommodate the normalization to unity,

i:e:; j/0 ðz1  naÞj2 ! dðz1  naÞ:

Furthermore, since there is translational invariance in the x-y plane of the sheet,
Rð1; 2; xÞ can be Fourier transformed in terms of a single 2D wave vector Q in the
x-y plane, such that

X
1
RðQ; z1 ; z2 ; xÞ ¼ dðz1  naÞdðz2  naÞR2D ðQ; xÞ; ð2:54Þ
n¼1

since the densities of the various sheets are additive. The corresponding 2D wave
vector transform of the Coulomb potential coupling is (charge strength k)
62 N.J.M. Horing

vð1  3Þ ! ðk=QÞexpðQjz1  z3 jÞ; ð2:55Þ

so (2.53) in Q-representation is given by

X
1
eðQ; z1 ; z2 ; xÞ ¼ dðz1  z2 Þ  4pa2D
0 ðQ; xÞ eQjz1 naj dðz2  naÞ; ð2:56Þ
n¼1

where the z3 -integral has been executed and we have defined the corresponding
polarizablility of a single 2D sheet as

0 ðQ; xÞ ¼ kR ðQ; xÞ=Q:

4pa2D ð2:57Þ
2D

To address the problem of inverting eðQ; z1 ; z2 ; xÞ as a matrix in z1 ; z2 , we make

the ansatz for KðQ; z2 ; z3 ; xÞ:

X
1
KðQ; z2 ; z3 ; xÞ ¼ dðz2  z3 Þ þ Cn0 ðQ; z2 ; z3 ; xÞdðz3  n0 aÞ; ð2:58Þ
n0 ¼1

with the coefficients Cn0 to be determined. The inversion condition, written in the
form
Z
dz2 eðQ; z1 ; z2 ; xÞKðQ; z2 ; z3 ; xÞ ¼ dðz1  z3 Þ ð2:59Þ

with (2.56), (2.58) leads to (suppress Q and x)

1 h
X i
Qjz1 n0 aj
Cn0 ðz1 ; n0 aÞ  4pa2D
0 e dðz3  n0 aÞ
n0 ¼1
ð2:60Þ
X
1 X
1
¼ 4pa2D
0 eQjz1 naj Cn0 ðna; n0 aÞdðz3  n0 aÞ;
n¼1 n0 ¼1

where the z2 -integration has been carried out. Equating the coefficients of like delta
functions, we have

X
1
0
Cn0 ðz1 ; n0 aÞ  4pa2D
0 eQjz1 naj Cn0 ðna; n0 aÞ ¼ 4pa2D
0 e
Qjz1 n aj
: ð2:61Þ
n¼1

To solve for Cn0 ðz1 ; naÞ, it is necessary to determine Cn0 ðna; n0 aÞ. Because of
periodicity, it is to be expected that

Cn0 ðma; n0 aÞ ¼ Gðm  n0 Þ; ð2:62Þ

2 Aspects of the Modeling of Low Dimensional Quantum Systems 63

and setting z1 ¼ ma in (2.61) yields

X
1
0
Gðm  n0 Þ  4pa2D
0 eQajmnj Gðn  n0 Þ ¼ 4pa2D
0 e
Qajmn j
: ð2:63Þ
n¼1

Introducing a Fourier (series) analysis with

Zp=a
a0 0
Gðm  n Þ ¼ dqeiqaðmn Þ gðqÞ; ð2:64Þ
2p
p=a

leads to the result [8]

 1
4pa2D
0 sinhðQaÞ sinhðQaÞ
gðqÞ ¼ 1  4pa2D ; ð2:65Þ
coshðQaÞ  cosðqaÞ 0
coshðQaÞ  cosðqaÞ

from which Gðn  n0 Þ (2.64) and Cn0 ðna; n0 aÞ (2.62) may be determined: In turn,
this provides the coefficients Cn0 ðz1 ; n0 aÞ through (2.61). This suffices to yield the
full solution for the inverse dielectric screening function KðQ; z1 ; z2 ; xÞ (since
Cn0 ðQ; z2 ; z3 ; xÞ in (2.58) involves z3 only at z3 ¼ n0 a). Introducing the definition of
the structure factor SðQ; qÞ,

sinhðQaÞ
SðQ; qÞ ¼ ; ð2:66Þ
coshðQaÞ  cosðqaÞ

the final result for the superlattice screening function is given by

X
1
KðQ; z1 ; z2 ; xÞ ¼ dðz1  z2 Þ þ dðz2  n0 aÞCn0 ðQ; z1 ; n0 a; xÞ
n0 ¼1
X
1
¼ dðz1  z2 Þ þ 4pa2D
0 ðQ; xÞ dðz2  n0 aÞ
n0 ¼1
0 1
X
1 Zp
1 0 ðQ; xÞSðQ; x=aÞ A
4pa2D
 @eQjz1 n aj þ
0
Qjz1 naj ixðnn0 Þ
e dxe :
n¼1
2p 0 ðQ; xÞSðQ; x=aÞ
1  4pa2D
p

ð2:67Þ

The spectrum of superlattice plasmons determined by the dispersion relation

arising from the vanishing of the denominator of the integrand on the right of
(2.67),

1  4pa2D
0 ðQ; xÞSðQ; qÞ ¼ 0; ð2:68Þ
64 N.J.M. Horing

involves a continuous band of mode frequencies [9–24] associated with the integral
limits p=a  q  p=a. The use of Dirac delta functions here has facilitated the
explicit determination of the inverse dielectric screening function as well as the
dynamic, nonlocal plasmon dispersion relation.

2.6 An Impenetrable Barrier: Quantum Particle

Dynamics in a Highly Singular 1D-Potential

The use of the one dimensional Dirac delta function ðdðxÞÞ itself as a potential has
proven to be quite straightforward; however, the inclusion of its derivative ðd0 ðxÞÞ
has brought forth substantial controversy [25–29]. As far back as 1986, Seba [28]
found that electron transmission through that highly singular potential could not
take place. However, other researchers have appended boundary conditions to d0 ðxÞ
and claimed that electron transmission can occur [30–39].
We recently studied this problem by constructing the full, exact Green’s function
for the Dirac-delta-function-derivative model, defining it solely in terms of the usual
derivative property ½d0 ðxÞ  dðdðxÞÞ=dx under integration by parts, with no appeal to
additional boundary conditions [40]. Aside from such an integration by parts, we
employ only the conventional, well established properties R 1of the Dirac delta function,
dðxÞ, in terms of integration as a generalized function, 1 dx dðx  aÞf ðxÞ ¼ f ðaÞ,

and differentiation, d g þ ðx  aÞ =dx ¼ dðx  aÞ, where g þ ðxÞ is the Heaviside unit
step function (which is understood to have the value g þ ð0Þ ¼ 1=2, to which a Fourier
series representation converges at the position of the step). Our avoidance of
boundary conditions is due to the fact that they are not at all necessary under the usual
0
conventions and that they may distort the meaning of d ðxÞ beyond recognition. The
Green’s function we obtain analytically in closed form confirms Seba’s finding that
there can be no electron transmission through the d0 ðxÞ-singular potential if the
potential profile is otherwise well-behaved, and that an incident Schrödinger
wave-packet must be totally reflected.
Here, we broaden these one-dimensional considerations to take into account a
spatially variable mass, mðxÞ, and a reasonably well behaved spatially variable
potential, VðxÞ, and a dðxÞ-potential, all in addition to the d0 ðxÞ-potential previously
examined. It is in this very general situation that we construct the full, exact
Schrödinger Green’s function in closed form, and show that no electron wave
packet transmission through d0 ðxÞ can occur. Our analysis of this problem is in
agreement with aspects of the formulation by Park [38], but the latter is quite
limited in application to one specific case.
Allowing for a variable mass, mðxÞ, the one-dimensional Schrödinger Green’s
function equation for a system with time-translational invariance has the (hermitian)
Sturm-Liouville form in frequency representation t  t0 ! x (we suppress the
2 Aspects of the Modeling of Low Dimensional Quantum Systems 65

explicit appearance of x, so Gðx; x0 ; t  t0 Þ ! Gðx; x0 ; xÞ  Gðx; x0 Þ), as given by

[41]

  
@ 1 @
þ VðxÞ þ UðxÞ Gðx; x0 Þ ¼ dðx  x0 Þ: ð2:69Þ
@x mðxÞ @x

Here, VðxÞ is understood to be a relatively well behaved potential (which can

accommodate finite discontinuities using the well known Green’s function joining
technique employed in the theory of surface/interface states [42–46]; it also
includes an x-term from the Fourier time-transform), and the highly singular
delta-function-type generalized function potentials are relegated to UðxÞ as

UðxÞ ¼ adðxÞ þ bd0 ðxÞ ð2:70Þ

(a, b are constants and d0 ðxÞ  d ðdðxÞÞ=dx).

To start, we define an auxiliary Green’s function, G0 ðx; x0 Þ, as the inverse of the
Sturm-Liouville operator excluding UðxÞ:

  
@ 1 @
þ VðxÞ G0 ðx; x0 Þ ¼ dðx  x0 Þ: ð2:71Þ
@x mðxÞ @x

Taking y1 ðxÞ and y2 ðxÞ as two linearly independent solutions of the homoge-
neous counterpart of (2.71), with y1 chosen to satisfy the boundary condition at the
lower limit and y2 doing so at the upper limit, the solution of (2.71) is known to be
[41]
 
mðx0 Þ y1 ðxÞy2 ðx0 Þ for x\x0
G0 ðx; x0 Þ ¼ ; ð2:72Þ
Dðy1 ; y2 Þ y2 ðxÞy1 ðx0 Þ for x [ x0

where Dðy1 ; y2 Þ is the Wronskian of the two solutions, y1 ðx0 Þ; y2 ðx0 Þ, evaluated at x0 :
 
y y01 
Dðy1 ; y2 Þ ¼ det 1 ; ð2:73Þ
y2 y02 

(y0  dyðx0 Þ=dx0 ).

Considering y1 , y2 to be known, and hence G0 is known, (2.69) may be rewritten
as

Z1
0 0
Gðx; x Þ ¼ G0 ðx; x Þ þ dx00 G0 ðx; x00 ÞUðx00 ÞGðx00 ; x0 Þ; ð2:74Þ
1
66 N.J.M. Horing

or

Gðx; x0 Þ ¼ G0 ðx; x0 Þ  aG0 ðx; 0ÞGð0; x0 Þ

Z1
ð2:75Þ
þb dx00 d0 ðx00 ÞG0 ðx; x00 ÞGðx00 ; x0 Þ:
1

Integrating the last term on the right by parts, this becomes

Z1
0 0 0 @
Gðx; x Þ ¼ G0 ðx; x Þ  aG0 ðx; 0ÞGð0; x Þ  b dx00 dðx00 Þ ½G0 ðx; x00 ÞGðx00 ; x0 Þ:
@x00
1
ð2:76Þ

Introducing “left” and “right” derivative notation (L and R, respectively),

@ 
00
Gðx00 ; x0 Þ  @ðLÞ Gðx00 ; x0 Þ ;
@x
@ 
00
Gðx0 ; x00 Þ  @ðRÞ Gðx0 ; x00 Þ ; ð2:77Þ
@x
@ @ h i
0 00 0 00
Gðx ; x Þ  @ 2
ðL;RÞ Gðx ; x Þ ;
@x0 @x00

Equation (2.76) may be written as


Gðx; x0 Þ ¼ G0 ðx; x0 Þ  aG0 ðx; 0ÞGð0; x0 Þ  b @ðRÞ G0 ðx; 0Þ Gð0; x0 Þ
 ð2:78Þ
 bG0 ðx; 0Þ @ðLÞ Gð0; x0 Þ :

To solve, we need to determine Gð0; x0 Þ and @ðLÞ Gð0; x0 Þ : Setting x ! 0, we
obtain (2.78) as

Gð0; x0 Þ ¼ G0 ð0; x0 Þ  aG0 ð0; 0ÞGð0; x0 Þ  b @ðRÞ G0 ð0; 0Þ Gð0; x0 Þ
 ð2:79Þ
 bG0 ð0; 0Þ @ðLÞ Gð0; x0 Þ :

Forming the left derivative of (2.78), we have

  
@ðLÞ Gðx; x0 Þ ¼ @ðLÞ G0 ðx; x0 Þ  a @ðLÞ G0 ðx; 0Þ Gð0; x0 Þ
h i  
 b @ðL;RÞ
2
G0 ðx; 0Þ Gð0; x0 Þ  b @ðLÞ G0 ðx; 0Þ @ðLÞ Gð0; x0 Þ ;
ð2:80Þ
2 Aspects of the Modeling of Low Dimensional Quantum Systems 67

and putting x ! 0 in (2.80) yields

  
@ðLÞ Gð0; x0 Þ ¼ @ðLÞ G0 ð0; x0 Þ  a @ðLÞ G0 ð0; 0Þ Gð0; x0 Þ
h i  
 b @ðL;RÞ
2
G0 ð0; 0Þ Gð0; x0 Þ  b @ðLÞ G0 ð0; 0Þ @ðLÞ Gð0; x0 Þ ;
ð2:81Þ

which expresses @ðLÞ Gð0; x0 Þ in terms of Gð0; x0 Þ as
    
1 þ b @ðLÞ G0 ð0; 0Þ @ðLÞ Gð0; x0 Þ ¼ @ðLÞ G0 ð0; x0 Þ  a @ðLÞ G0 ð0; 0Þ Gð0; x0 Þ
h i
 b @ðL;RÞ
2
G0 ð0; 0Þ Gð0; x0 Þ;
ð2:82Þ

or
   1
@ðLÞ Gð0; x0 Þ ¼ 1 þ b @ðLÞ G0 ð0; 0Þ
   h i 
 @ðLÞ G0 ð0; x0 Þ  a @ðLÞ G0 ð0; 0Þ Gð0; x0 Þ  b @ðL;RÞ
2
G0 ð0; 0Þ Gð0; x0 Þ :

ð2:83Þ

Employing this result in (2.79), we have


Gð0; x0 Þ ¼ G0 ð0; x0 Þ  aG0 ð0; 0ÞGð0; x0 Þ  b @ðRÞ G0 ð0; 0Þ Gð0; x0 Þ
  1   
 bG0 ð0; 0Þ 1 þ b @ðLÞ G0 ð0; 0Þ @ðLÞ G0 ð0; x0 Þ  a @ðLÞ G0 ð0; 0Þ Gð0; x0 Þ
h i o
b @ðL;RÞ
2
G0 ð0; 0Þ Gð0; x0 Þ ;

ð2:84Þ

which yields Gð0; x0 Þ as

n    1
Gð0; x0 Þ ¼ 1 þ aG0 ð0; 0Þ þ b @ðRÞ G0 ð0; 0Þ  bG0 ð0; 0Þ 1 þ b @ðLÞ G0 ð0; 0Þ
  h io1
 a @ðLÞ G0 ð0; 0Þ þ b @ðL;RÞ
2
G0 ð0; 0Þ
h   1 i
 G0 ð0; x0 Þ  bG0 ð0; 0Þ 1 þ b @ðLÞ G0 ð0; 0Þ @ðLÞ G0 ð0; x0 Þ :
ð2:85Þ

Equation (2.85) may now be substituted into the right side of (2.83) to obtain
½@ðLÞ Gð0; x0 Þ in terms of G0 and its derivatives alone. Finally, the substitution of
these results for Gð0; x0 Þ and ½@ðLÞ Gð0; x0 Þ as indicated above into the right side of
(2.78) yields the full, exact Green’s function for the highly singular 1-D potential
UðxÞ of (2.70) joined onto any relatively well behaved 1-D potential VðxÞ, such as a
harmonic oscillator and/or a constant electric field potential independent of time.
68 N.J.M. Horing

For the special case of b ¼ 0, we obtain the corresponding result for the dðxÞ-
potential with no d0 ðxÞ-potential.

aG0 ðx; 0ÞG0 ð0; x0 Þ

Gðx; x0 Þ ¼ G0 ðx; x0 Þ  : ð2:86Þ
1 þ aG0 ð0; 0Þ

To examine the role of the highly singular potentials in the 1-D Green’s function,
we rewrite (2.72) in the form
 
G0 ðx; x0 Þ ¼ Cðx0 Þ g þ ðx0  xÞy1 ðxÞy2 ðx0 Þ þ g þ ðx  x0 Þy2 ðxÞy1 ðx0 Þ ð2:87Þ

where g þ ðxÞ is the Heaviside unit step function and Cðx0 Þ  mðx0 Þ=Dðy1 ; y2 Þ. Since
g þ ð0Þ ¼ 1=2,
 
1 1
G0 ð0; 0Þ ¼ Cð0Þ y1 ð0Þy2 ð0Þ þ y2 ð0Þy1 ð0Þ ¼ Cð0Þy1 ð0Þy2 ð0Þ: ð2:88Þ
2 2

Differentiating (2.87) to form @ðLÞ G0 ðx; x0 Þ , there is a cancellation of terms
involving Dirac delta functions arising from @ðLÞ g þ ðx  x0 Þ, etc., and we obtain
  
@ðLÞ G0 ðx; x0 Þ ¼ Cðx0 Þ g þ ðx0  xÞy01 ðxÞy2 ðx0 Þ þ g þ ðx  x0 Þy02 ðxÞy1 ðx0 Þ : ð2:89Þ
0 0 0 0
Differentiating again to form ½@ðL;RÞ
2
Gðx; x0 Þ, we have (C ðx Þ  dðCðx ÞÞ=dx ,
etc.)
 
½@ðL;RÞ
2
G0 ðx; x0 Þ ¼ C 0 ðx0 Þ g þ ðx0  xÞy01 ðxÞy2 ðx0 Þ þ g þ ðx  x0 Þy02 ðxÞy1 ðx0 Þ
 
0
dðx0  xÞy01 ðxÞy2 ðx0 Þ þ g þ ðx0  xÞy01 ðxÞy02 ðx0 Þ
þ Cðx Þ ;
dðx  x0 Þy02 ðxÞy1 ðx0 Þ þ g þ ðx  x0 Þy02 ðxÞy01 ðx0 Þ
ð2:90Þ

which may be written more compactly using the definition of the Wronskian,
Dx0 ðy1 ; y2 Þ, as (subscript “x0 ” denotes evaluation at x0 )
h i
@ðL;RÞ
2
G0 ðx; x0 Þ ¼  Cðx0 Þdðx0  xÞDx0 ðy1 ; y2 Þ
þ Cðx0 Þ½g þ ðx0  xÞy01 ðxÞy02 ðx0 Þ þ g þ ðx  x0 Þy02 ðxÞy01 ðx0 Þ
þ C0 ðx0 Þfg þ ðx0  xÞy01 ðxÞy2 ðx0 Þ þ g þ ðx  x0 Þy02 ðxÞy1 ðx0 Þg:
ð2:91Þ

Clearly, the first term on the right of (2.91) shows that

2 Aspects of the Modeling of Low Dimensional Quantum Systems 69

h i
 2 
 @ðL;RÞ G0 ð0; 0Þ  ! 1: ð2:92Þ
h i
 2 
In view of the huge value of  @ðL;RÞ G0 ð0; 0Þ , we may write (2.85) for b 6¼ 0 as

 ( )
0
1 þ b @ ðLÞ G ð0; 0Þ bG ð0; 0Þ @ðLÞ G ð0; x Þ
Gð0; x0 Þ ¼  i G0 ð0; x0 Þ 
0 0 0
h  ;
b2 G0 ð0; 0Þ @ðL;RÞ
2 G0 ð0; 0Þ 1 þ b @ðLÞ G0 ð0; 0Þ

ð2:93Þ

and considering particle transmission through the highly singular potential at the
origin (with x\0 and x0 [ 0), this means that (2.92), (2.93)

Gð0; x0 Þ ¼ 0: ðb 6¼ 0Þ ð2:94Þ

On the same basis, (2.83) for @ðLÞ Gð0; x0 Þ may be written for b 6¼ 0 as

½@ðLÞ Gð0; x0 Þ ¼ D1 @ðLÞ G0 ð0; x0 Þ
1  
þ G0 ð0; x0 Þ  bD1 G0 ð0; 0Þ @ðLÞ G0 ð0; x0 Þ ;
bG0 ð0; 0Þ
ð2:95Þ

where we have defined D as

D ¼ 1 þ b½@L G0 ð0; 0Þ: ð2:96Þ

Here, we see that the role of DU ¼ adðxÞ as part of the generalized potential
U is negligible in Gðx; x0 Þ for well behaved potentials VðxÞ so long as b 6¼ 0.
Employing (2.93) and (2.95), we find Gðx; x0 Þ of (2.78) for b 6¼ 0 as

Gðx; x0 Þ ¼ G0 ðx; x0 Þ  bG0 ðx; 0ÞðD1 @ðLÞ G0 ð0; x0 Þ
1  
þ G0 ð0; x0 Þ  bD1 G0 ð0; 0Þ @ðLÞ G0 ð0; x0 Þ ;
bG0 ð0; 0Þ

or

G0 ðx; 0ÞG0 ð0; x0 Þ

Gðx; x0 Þ ¼ G0 ðx; x0 Þ  : ð2:97Þ
G0 ð0; 0Þ

Considering that the Fourier transform of Gðx; x0 ; xÞ to direct time representation

represents the quantum mechanical amplitude for a Schrödinger particle (wave
packet) to be transmitted from position x0 at time t0 to x at a later time t, such that
70 N.J.M. Horing

Z Z1
dx ixðtt0 Þ
Wout ðx; tÞ ¼ e dx0 Gðx; x0 ; xÞWin ðx0 ; t0 Þ; ð2:98Þ
2p
1

it is clear that an electron wave packet, Win ðx0 \0; t0 Þ in the region of incidence,
x0 \0, cannot be transmitted to Wout ðx [ 0; tÞ in the outgoing wave region, x [ 0,
on the other side of the highly singular potential bd0 ðxÞ because (2.97) and (2.72)
jointly yield a vanishing result


y2 ðxÞy1 ð0Þy2 ð0Þy1 ðx0 Þ
Gðx [ 0; x0 \0; xÞ ¼ Cðx0 Þ y2 ðxÞy1 ðx0 Þ  ¼ 0: ð2:99Þ
y1 ð0Þy2 ð0Þ

This very general result means that there is no possibility of particle transmission
from x0 \0 through the bd0 ðxÞ-potential, even in the presence of well behaved
potentials VðxÞ as well as the presence of DU ¼ adðxÞ, provided b 6¼ 0.

References

1. V. Fessatidis, N.J.M. Horing, K. Sabeeh, Philos. Mag. B79, 77 (1999)

2. A. Erdelyi, Higher Transcendental Functions. vol. II, Eq. (17), p. 189. (McGraw-Hill, New
York, 1953)
3. N.J.M. Horing, Quantum Statistical Field Theory: Schwinger’s Variational Method. (Oxford
University Press, to be published)
4. A. Erdelyi, Higher Transcendental Functions, vol. II, Eq. (33), p. 8. (McGraw-Hill, New
York, 1953)
5. N. Horing, M. Yildiz, Ann. Phys. (NY) 97, 216–241 (1976)
6. A. Erdelyi, Higher Transcendental Functions, vol. II, Eq. (32), p. 195. (McGraw-Hill, New
York, 1953)
7. A. Erdelyi, Tables of Integral Transforms, vol. I. (McGraw-Hill, New York, 1954)
8. I.S. Gradshteyn, I.M. Ryzhik, Tables of Integrals, Series and Products (Academic, New York,
1965)
9. P.B. Visscher, L.M. Falicov, Phys. Rev. B 3, 2541 (1971)
10. D. Grecu, Phys. Rev. B 8, 1958 (1972)
11. A.L. Fetter, Ann. Phys. (N. Y.) 88, 1 (1974)
12. M. Apostol, Z. Phys, B 22, 13 (1975)
13. M. Kobayashi et al., J. Phys. Soc. Jpn. 39, 18 (1975)
14. J. Mizuno et al., J. Phys. Soc. Jpn. 39, 983 (1975)
15. A. Caille et al., Solid State Commun. 24, 805 (1977)
16. L. Bloss and E. M., Brody, Solid State Commun. 43, 523 (1982)
17. W.L. Bloss, Solid State Commun. 44, 363 (1982)
18. D. Olego et al., Phys. Rev. B 25, 7867 (1982)
19. P. Ruden, G.H. Dohler, Phys. Rev. B 27, 3538 (1983)
20. P. Ruden, G.H. Dohler, Phys. Rev. B 27, 3547 (1983)
21. S. Das Sarma, A. Madhukar, Phys. Rev. B 23, 805 (1981); S. Das Sarma, J.J. Quinn, ibid. 25,
7603 (1982); 27, 6516 (1983)
22. S. Das Sarma, Phys. Rev. B 28, 2240 (1983); Appl. Surf. Sci. 11/12, 535 (1982)
23. G.G. de la Cruz et al., J. Phys. Chem. Solids 44, 807 (1983)
2 Aspects of the Modeling of Low Dimensional Quantum Systems 71

24. A.C. Tselis, J.J. Quinn, Phys. Rev. B 29, 2021 (1984); 29, 3318 (1984)
25. P.L. Christiansen, H.C. Arnbak, A.V. Zolotaryuk, V.N. Ermakov, Y.B. Gaididei, J. Phys. A:
Math. Gen. 36, 7589 (2003)
26. F.A.B. Coutinho, Y. Nogami, J. Fernando Perez; J Phys. A. Math. Gen. 30, 3937 (1997)
27. Y. Golovaty; arXiv:1201.2610v1 [math.SP] 12 Jan 2012
28. P. Šeba, Rep. Math. Phys. 24, 111 (1986)
29. F.M. Toyama, Y. Nogami, J. Phys. A: Math. Theor. 40, F685 (2007)
30. M. Gadella, M.L. Glasser, L.M. Nieto, Int. J. Theor. Phys. 50, 2144 (2011)
31. M. Gadella, J. Negro, L.M. Nieto, Phys. Lett. A 373, 1310 (2009)
32. M. Gadella, M.L. Glasser, L.M. Nieto, Int. J. Theor. Phys. 50, 2191 (2011)
33. J.J. Alvarez, M. Gadella, M.L. Glasser, L.P. Lara, L.M. Nieto, J. Phys: Conf. Ser. 284, 012009
(2011)
34. A.V. Zolotaryuk, Y. Zolotaryuk, arXiv:1202.1117v1 [math-ph] 6 Feb 2012
35. J.F. Brasche, L. Nizhnik, arXiv:1112.2545v1 [math.FA] 12 Dec 2011
36. P. Kurasov, Math. Anal. Appl. 201, 297 (1996)
37. S. Kocinak, V. Milanovic, Mod. Phys. Lett. B 26, 1250092 (2012)
38. D.K. Park, J. Phys. A Math.: Gen. 29, 6407 (1996)
39. Haydar Uncu, Devrim Tarhan, Ersan Demiralp, Özgür E. Müstecaphoğlu, Phys. Rev. A 76,
013618 (2007)
40. N.J.M. Horing, unpublished
41. P.M. Morse, H. Feschbach, Methods of Theoretical Physics, vol. 1. (McGraw-Hill, New York,
832, 1953)
42. F. Garcia-Moliner, F. Flores, Introduction to the Theory of Solid Surfaces. (Cambridge
University Press, Cambridge, 1979)
43. F. Garcia-Moliner, V.R. Velasco, Theory of Single and Multiple Interfaces. (World Scientific,
Singapore, 1992)
44. S.G. Davison, J.D. Levine, Solid State Physics, vol. 25, eds. by H. Ehrenreich, F. Seitz, D.
Turnbull (Academic Press, New York, 1970)
45. S.G. Davison, M. Steslicka, Basic Theory of Surface States (Clarendon Press, Oxford, 1992)
46. J. Inglesfield, J. Phys. C; Solid State Phys. 4, L 14, (1971)
Chapter 3
Wave Propagation and Diffraction
Through a Subwavelength Nano-Hole
in a 2D Plasmonic Screen

Norman J.M. Horing and Désiré Miessein

Abstract This article addresses the subject of subwavelength transmission and

diffraction by a two dimensional plasmonic screen/layer with an embedded
nano-hole aperture. The dynamical development of a prototypical scalar wave
(incident at an arbitrary angle) is traced in terms of the associated Helmholtz
Green’s function. The latter is derived through the exact analytic solution of a
succession of integral equations which obviate the need to specify boundary con-
ditions separately. The resulting wave transmission/diffraction is evaluated
numerically and exhibited graphically: It includes the finite plasmonic response of
the 2D layer jointly with that of the subwavelength nano-hole aperture, for an
arbitrary angle of incidence.

3.1 Introduction

Wave transmission and diffraction in the presence of a subwavelength nano-hole in

a screen (radius R
k) has been the subject of extensive investigation in recent
years due to a variety of interesting optical phenomena [1–12]. The properties of the
screen are often represented by metallic boundary conditions, which, however, may
not accurately describe its plasmonic polarization response. In this article we dis-
cuss the transmission/diffraction of a prototypical scalar field incident on a thin two
dimensional plasmonic layer, having well defined finite polarizability, containing a
subwavelength aperture. The two dimensional layer is taken to be at the plane z = 0,
ð3DÞ
embedded in a 3D bulk host medium having background dielectric constant eb
with the aperture at its center, x = y = 0. Employing a succession of integral

N.J.M. Horing (&)  D. Miessein

Department of Physics and Engineering Physics, Stevens Institute of Technology, Hoboken,
NJ 07030, USA
e-mail: [email protected]
D. Miessein
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 73

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_3
74 N.J.M. Horing and D. Miessein

equations for the scalar Green’s function of the associated Helmholtz problem
describing the evolution of the wave in the transmission/diffraction process, the
need for boundary conditions is obviated, and explicit analytic solutions are
obtained that accurately account for both the presence of the subwavelength
nano-hole and the finite plasmonic response of the layer.
In Sect. 3.2, the scalar Helmholtz Green’s function of the two dimensional
plasmonic layer containing a nano-hole embedded in a 3D bulk host medium is
obtained through the analytic solutions of a succession of integral equations. The
trasmission/diffraction of a plane scalar wave at an arbitrary angle of incidence on
the perforated 2D layer is discussed in Sect. 3.3. The numerical analysis of
Helmholtz wave propagation in the vicinity of the plasmonic layer containing a
nano-hole, based on the analytic results obtained in Sect. 3.3, is presented in detail
in the figures of Sect. 3.4. A summary of conclusions is contained in Sect. 3.5.

3.2 Scalar Green’s Function for a 2D Plasmonic Layer

with a Nano-Hole Aperture Embedded in a 3D Bulk
Host Medium

Considering a two dimensional (2D) plasmonic layer on the plane z = 0, embedded

ð3DÞ
in a three dimensional bulk host medium with background dielectric constant eb ,
the scalar Helmholtz Green’s function of the system Gfs satisfies the equation (ω—
frequency representation) [13–15]

 Z
x2 ð3DÞ 0 x2 ð2DÞ
r2  e G fs ðr; r ; xÞ  d 3 r0 afs ðr; r00 ; xÞGfs ðr00 ; r0 ; xÞ
c2 b c2
¼ dð3Þ ðr  r0 Þ; ð3:1Þ

ð2DÞ
where the polarizability afs of the 2D plasmonic layer has the form

ð2DÞ ð2DÞ
afs ðr; r0 ; xÞ ¼ afs ðrk ; r0k ; xÞ d dðzÞdðz0 Þ: ð3:2Þ

ð2DÞ
Here, d is the thickness of the plasmonic sheet, r ¼ ðrk ; zÞ and afs ðrk ; r0k ; xÞ is
the 2D plasmonic polarizability on the 2D sheet and δ(z) is the Dirac delta function
needed to lock the polarizability onto the plane of the 2D layer at z = 0. To solve for
Gfs ðr; r0 ; xÞ, we employ the bulk Helmholtz Green’s function, G3D ðr; r0 ; xÞ, given
by


x2 ð3DÞ
r  2 eb
2
G3D ðr; r0 ; xÞ ¼ d3D ðr  r0 Þ; ð3:3Þ
c
3 Wave Propagation and Diffraction Through a Subwavelength … 75

in lateral wavevector representation ðrk  r0k 7! kk Þ,

0
eik? jzz j
G3D ðkk ; z; z0 ; xÞ ¼  ð3:4Þ
2ik?
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffi
ð3DÞ
where k? ¼  q2x  kk2 and qx ¼ ðx=cÞ eb .
The Helmholtz Green’s function including the plasmonic sheet, Gfs ðr; r0 ; xÞ,
satisfies the integral equation

Gfs ðr; r0 ; xÞ ¼ G3D ðr; r0 ; xÞ

Z Z
x2 3 00
þ 2 d r d 3 r000 G3D ðr; r00 ; xÞa2D 00 000 000 0
fs ðr ; r ; xÞGfs ðr ; r ; xÞ:
c
ð3:5Þ

Fourier transforming this in the lateral plane of translational invariance

(rk  r0k ! kk ) and employing (3.2), G3D ðkk ; 0; z0 ; xÞ is readily obtained alge-
braically and leads to Gfs as

Gfs ðkk ; z; z0 ; xÞ ¼ G3D ðkk ; z; z0 ; xÞ

ð2DÞ 0
x2 G3D ðkk ; z; 0; xÞafs ðkk ; xÞG3D ðkk ; 0; z ; xÞ ð3:6Þ
þ 2d ð2DÞ
:
c 1  x2 da ðkk ; xÞG3D ðkk ; 0; 0; xÞ
2
c fs

Using (3.4), this yields

0 0
0 eik? jzz j ceik? ðjzj þ jz jÞ
Gfs ðkk ; z; z ; xÞ ¼  þ ; ð3:7Þ
2ik? 2ik? ð2ik? þ cÞ

ð2DÞ ð2DÞ
where c ¼ xc2 d afs ðkk ; xÞ. For metallic densities, afs ðkk ; xÞ tends to negative
2

infinity and γ also approaches negative infinity, so (3.7) reduces to

ik? jzz0 j 0 
0 e  eik? ðjzj þ jz jÞ
Gfs ðkk ; z; z ; xÞ ¼  ; ð3:8Þ
2ik?

which indicates perfect reflection and no transmission of an impinging scalar wave.

It is useful to represent Gfs ðrk ; 0; z; 0; xÞ in lateral position representation: with
z0 ¼ 0, (3.7) yields

eik? jzj
Gfs ðkk ; z; 0; xÞ ¼  : ð3:9Þ
ð2ik? þ cÞ
76 N.J.M. Horing and D. Miessein

Bearing in mind that

x2 d ð2DÞ
c¼ a ðkk ; xÞ ð3:10Þ
c2 fs

and


ð2DÞ 2pie2 n2D qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
afs ðkk ; xÞ ¼ q 2  k2 ;
x k ð3:11Þ
m x2

is the polarizability of the 2D plasmonic layer (with n2D as the 2D equilibrium

density and m* is effective mass), we have

1 eik? jzj
Gfs ðkk ; z; 0; xÞ ¼ 
1 þ l 2ik?
ð3:12Þ
1
¼ G3D ðkk ; z; 0; xÞ;
1þl

or, in position representation,

pffiffiffiffiffiffiffiffiffiffiffi ffi
iq r2 þ jzj2
1 e x k
Gfs ðrk ; 0; z; 0; xÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; ð3:13Þ
4p½1 þ l r 2 þ jzj2
k

where l ¼ pe2 d n2D =m c2 .

Addressing the problem of a two dimensional plasmonic layer S1 perforated
by a nano-scale aperture S2 of radius R at r ¼ 0 in the (x − y) plane, the full
sheet polarizability is modified by subtracting the part associated with the hole
(Fig. 3.1)

ð2DÞ ð2DÞ
að2DÞ ðr; r0 ; xÞ ¼ afs ðr; r0 ; xÞ  ah ðr; r0 ; xÞ ð3:14Þ

ð2DÞ
where ah ðr; r0 ; xÞ is the part of the layer polarizability excised from the
nano-hole. The scalar Green’s function for this system, including the aperture,
satisfies the integral equation given by:

Gðr; r0 ; xÞ ¼ Gfs ðr; r0 ; xÞ

Z Z
x2 3 00
 2 d r d 3 r000 Gfs ðr; r00 ; xÞa2D 00 000 000 0
h ðr ; r ; xÞGðr ; r ; xÞ:
c
ð3:15Þ
3 Wave Propagation and Diffraction Through a Subwavelength … 77

Fig. 3.1 Two dimensional plasmonic layer (thickness d, embedded at z = 0 in a three dimensional
bulk host medium) with a nano-hole of radius R at the origin of the (x − y) plane

Here, the polarizability part excised from the nano-hole is defined as

ð2DÞ ð2DÞ
ah ðr; r0 ; xÞ ¼ afs ðx; y; x0 ; y0 ; xÞd dðzÞdðz0 Þ
a 
b 
 g þ  jxj g þ  jyj ð3:16Þ
2 2
a 

0 b 0
 g þ  jx j g þ  jy j ;
2 2

where g þ ða2  jxjÞ is the Heaviside unit step function representing a cut-off function
imposed to confine the integration range on the 2D sheet to the nano-hole
dimension ðR aÞ and δ(z) is the Dirac delta function needed to lock the
78 N.J.M. Horing and D. Miessein

polarizability onto the plane of the 2D plasmonic layer. A simple approximation of

(3.16) for very small radius leads to g þ ða2  jxjÞ 7! a dðxÞ and

ð2DÞ ð2DÞ
ah ðr; r0 ; xÞ
A2 afs ðrk ; r0k ; xÞd2 ðrk Þd2 ðr0k Þd dðzÞdðz0 Þ; ð3:17Þ

(A is nano-hole area) and with this, (3.15) yields

Gðrk ; r0k ; z; z0 ; xÞ ¼ Gfs ðrk ; r0k ; z; z0 ; xÞ  bGfs ðrk ; 0; z; 0; xÞGð0; r0k ; 0; z0 ; xÞ;
ð3:18Þ

where b ¼ cA2 with c ¼ xc2d a2D 0

2
fs ð½rk ¼ 0; ½rk ¼ 0; xÞ. Equation (3.18) may be
solved algebraically by setting rk ¼ 0 and z = 0, with the closed form analytic result

Gðrk ; r0k ; z; z0 ; xÞ ¼ Gfs ðrk ; r0k ; z; z0 ; xÞ

bGfs ðrk ; 0; z; 0; xÞGfs ð0; r0k ; 0; z0 ; xÞ ð3:19Þ
 :
1 þ bGfs ð½rk ¼ 0  ½r0k ¼ 0; z ¼ 0; z0 ¼ 0; xÞ

Anticipating that the transmitted scalar wave is controlled by Gðrk ; 0; z; 0; xÞ,

setting r0k ¼ 0 and z0 ¼ 0 yields

Gfs ðrk ; 0; z; 0; xÞ
Gðrk ; 0; z; 0; xÞ ¼ : ð3:20Þ
1 þ bGfs ð½rk ¼ 0  ½r0k ¼ 0; z ¼ 0; z0 ¼ 0; xÞ

which involves a divergent integral in the denominator when all its positional
arguments vanish. Specifically, rewriting (3.11) in lateral position representation
with rk ¼ 0, r0k ¼ 0, and z = 0, z0 ¼ 0, the resulting kk -integral is divergent at the
upper limit,

 Z1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð2DÞ 1 2pie2 n2D
afs ð0; 0; xÞ ¼ dkk kk q2x  kk2 ; ð3:21Þ
2p m x
 2
0

and similar remarks apply to Gfs ð0; 0; 0; 0; xÞ:

8 9
>
<Zqx Z1 >
i dkk kk dkk kk =
Gfs ð0; 0; 0; 0; xÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  i qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð3:22Þ
4p½1 þ l >
: q2x  kk2 kk2  q2x> ;
0 q x

These divergencies are artifacts of confining the aperture to zero radial size in
(3.17), and recognizing that the integral equation, (3.15), is actually extended over a
finite range R, it is necessary to adjust the result to accommodate a “ smear”
average in which jrk j, jr0k j 7! 0 is replaced by 0 7! R; alternatively, in Fourier lateral
3 Wave Propagation and Diffraction Through a Subwavelength … 79

kk -integrations, the range is restricted to a large but finite range of kk \ R1 , with

frequencies in the range qx R\1 for nano-holes of subwavelength dimensions,
R\k q1x . The resulting integrations are elementary, yielding

 h i3=2 
ð2DÞ 1 2pie2 n2D
afs ð½rk ¼ 0; ½r0k ¼ 0; xÞ
3
ðqx RÞ þ i 1  ðqx RÞ 2
6pR3 m x2
ð3:23Þ

and
 h i3=2 
l 3 2
c¼ i ð q x RÞ  1  ðq x RÞ : ð3:24Þ
3pR3

Similarly,
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
Gfs ð½rk ¼ 0  ½r0k ¼ 0; z ¼ 0; z0 ¼ 0; xÞ ¼ iqx R þ 1  ðqx RÞ2 :
4pð1 þ lÞR
ð3:25Þ

3.3 Scalar Field Response of a Perforated 2D Plasmonic

Layer to an Incident Wave U 0 ðr; xÞ: Transmission

The model scalar field system employed here, which is governed by Helmholtz
dynamics, is driven by a source density qðr; tÞ which is the sum of the induced
density, qind ðr; tÞ, and an external density, qext ðr; tÞ, in accordance with the inho-
mogeneous equation (Fig. 3.2)

 Z
x2 ð3DÞ x2
r  2 eb
2
Uðr; xÞ  2 d 3 r0 að2DÞ ðr; r0 ; xÞUðr0 ; xÞ ¼ 4pqext ðr; xÞ:
c c
ð3:26Þ

Here, að2DÞ ðr; r0 ; xÞ is the polarizability induced in the perforated 2D layer,

ð2DÞ ð2DÞ
a ðr; r0 ; xÞ ¼ afs ðr; r0 ; xÞ  ah ðr; r0 ; xÞ (3.14), (3.17). Comparison of
ð2DÞ

(3.26) with (3.1) immediately shows that

Z
Uðr; xÞ ¼ 4p d 3 r0 Gðr; r0 ; xÞqext ðr0 ; xÞ; ð3:27Þ
80 N.J.M. Horing and D. Miessein

Fig. 3.2 Two dimensional plasmonic layer (thickness d, embedded at z = 0 in a three dimensional
bulk host medium) with a nano-hole of radius R at the origin of the (x − y) plane, shown with
incident, reflected and transmitted wave vectors (k0 ; kr ; kt ) for waves (U0(x, y, z; t), Ur(x, y, z; t),
Ut(x, y, z; t))

which is typical for linear systems. To connect this with the incident field U0 ðr; xÞ
generated by a distant source qext ðr; xÞ, we follow the techniques of [13–15] noting
that
Z
U0 ðr; xÞ ¼ 4p d 3 r0 G3D ðr; r0 ; xÞqext ðr0 ; xÞ; ð3:28Þ

or, in positional matrix notation

U0 ¼ 4pG3D qext ; ð3:29Þ

3 Wave Propagation and Diffraction Through a Subwavelength … 81

so that

U ¼ GG1
3D U0 : ð3:30Þ

By the same token [13–15], (3.15) may be written as the positional integral
(matrix) equation

x2 ð2DÞ
G ¼ Gfs  Gfs ah G; ð3:31Þ
c2

or, multiplying by G−1 from the right and G1

fs from the left,

x2 ð2DÞ
G1 ¼ G1
fs þ a : ð3:32Þ
c2 h

Alternatively, analogous treatment of (3.5) yields

x2 ð2DÞ
G1 1
fs ¼ G3D  a ; ð3:33Þ
c2 fs

so that

x2 h ð2DÞ ð2DÞ
i
G1 ¼ G1
3D  a fs  a h
c2 ð3:34Þ
x2 ð2DÞ
¼ G1
fs þ 2 ah :
c

Consequently, U may be written as

x2 ð2DÞ x2 ð2DÞ
U ¼ GG1
3D U0 ¼ U0 þ G a fs U 0  G ah U 0 : ð3:35Þ
c2 c2

In explicit position/frequency representation, this result reads as

Z Z Z Z
x2 2 0 2 00 0
Uðrk ; z; xÞ ¼ U0 ðrk ; z; xÞ þ 2 d rk d rk dz dz00 Gðrk ; r0k ; z; z0 ; xÞ
c
0 00 0 00 00 00
fs ðrk ; rk ; z ; z ; xÞ U0 ðrk ; z ; xÞ
 a2D
Z Z Z Z
x2
 2 d 2 r0k d 2 r00k dz0 dz00 Gðrk ; r0k ; z; z0 ; xÞ
c
0 00 0 00 00 00
h ðrk ; rk ; z ; z ; xÞ U0 ðrk ; z ; xÞ:
 a2D
ð3:36Þ
82 N.J.M. Horing and D. Miessein

0 00
Bearing in mind that a2D
fs involves dðz Þdðz Þ—functions (3.2) and ah involves
2D

d2 ðrk0 Þd2 ðrk00 Þdðz0 Þdðz00 Þ—functions (3.17) facilitating integrations on the right of
(3.36), we obtain

Uðrk ; z; xÞ ¼ U0 ðrk ; z; xÞ þ I1 ðrk ; z; xÞ  I2 ðrk ; z; xÞ; ð3:37Þ

where (3.19) has been employed in evaluating the integrals I1 ðrk ; z; xÞ and
I2 ðrk ; z; xÞ as
Z ( Z )
x2 d 1 ikk :r00k
I1 ðrk ; z; xÞ ¼ 2 d 2 r00k 2
d kk e ikk :rk
Gfs ðkk ; z; 0; xÞe a2D
fs ðkk ; xÞ
c ð2pÞ2
 U0 ðr00k ; 0; xÞ
 
x2 d bGfs ðrk ; 0; z; 0; xÞ
 2
c 1 þ b Gfs ð0; 0; 0; 0; xÞ
Z ( Z )
1 ik :r 00
 d 2 r00k d 2 kk Gfs ðkk ; 0; 0; xÞa2D
fs ðkk ; xÞ e
k k

ð2pÞ2
 U0 ðr00k ; 0; xÞ
ð3:38Þ

and

I2 ðrk ; z; xÞ ¼ bGðrk ; 0; z; 0; xÞU0 ð0; 0; xÞ

 
Gfs ðrk ; 0; z; 0; xÞ ð3:39Þ
¼b U0 ð0; 0; xÞ;
1 þ bGfs ð0; 0; 0; 0; xÞ

with Gfs expressed in lateral position representation. It should be noted that in the
absence of a nano-hole, b ¼ cA2 ¼ 0 and I2 ðrk ; z; xÞ ¼ 0. In this case, the field
Uðrk ; z; xÞ 7! Ufs ðrk ; z; xÞ is that of the full sheet alone, so that

Ufs ðrk ; z; xÞ ¼ U0 ðrk ; z; xÞ þ I1 ðrk ; z; xÞ

¼ U0 ðrk ; z; xÞ
Z Z
Z 
d x2 2 0 2 00 1 ikk :ðrk r0k Þ
þ 2 d rk d rk 2
d kk e Gfs ðkk ; z; 0; xÞ
c ð2pÞ2
Z !
1 ik0 :ðr0 r00 Þ
 2
d 2 k0k e k k k a2D 0 00
fs ðkk ; xÞ U0 ðr ; 0; xÞ:
ð2pÞ
ð3:40Þ
3 Wave Propagation and Diffraction Through a Subwavelength … 83

Using (3.9), this yields

Ufs ðrk ; z; xÞ ¼ U0 ðrk ; z; xÞ

Z Z
l 1 2 00 ik jzj þ ikk :ðrk r00k Þ
 d r k d 2 kk e ? U0 ðr00 ; 0; xÞ:
½1 þ l ð2pÞ 2

ð3:41Þ

Considering an infinite incident plane wave-train of the form

U0 ðr; tÞ ¼ U0 ei½k0 :r  x0 t
i ½k0k :rk þ k0? z  x0 t
ð3:42Þ
¼ U0 e ;

where U0 is the amplitude, its frequency representation is given by

i ½k0k :rk þ k0? z

U0 ðrk ; z; xÞ ¼ 2pU0 e dðx  x0 Þ; ð3:43Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffi
ð3DÞ
where k0? ¼ q2x0  k02k , qx0 ¼ xc0 eb is the incident wavenumber, and k0k and
k0? are the parallel and perpendicular components of the incident wave vector,
respectively.
The scalar field in the vicinity of the 2D plasmonic screen in the absence of the
nano-hole is determined by (3.41) jointly with (3.43); in time representation (ex-
ecuting the r00k - and kk -integrations) it is given by (k0? [ 0)

i ½k0k :rk þ k0? jzj  x0 t

Ufs ðrk ; z; tÞ ¼ U0 ðrk ; z; tÞ  b U0 e ð3:44Þ

where b  1 þl l. The incident and reflected waves occurring in the region z < 0 are
described jointly as

Ufs ðrk ; z; tÞ ¼ U0 ðrk ; z; tÞ 1  b e2 i k0? z ð3:45Þ

and its square modulus relative to that of the incident wave is given by
  
Ufs ðrk ; z; tÞ2
  ¼ 1 þ b2  2 b cos½2 k0 z: ð3:46Þ
 U  ?
0

The total transmitted wave occurring in the region z > 0 is given by

h i
i k0k :rk þ k0? z  x0 t
Ufþ
s
ðrk ; z; tÞ ¼ U0 ðrk ; z; tÞ  b U0 e ; ð3:47Þ
84 N.J.M. Horing and D. Miessein

and its square modulus (relative to the incident wave) is given by

 
U þ ðr ; z; tÞ2
 fs k  2
  ¼ ð1  bÞ : ð3:48Þ
 U0 

It is readily seen that this result correctly reproduces the complete absence of a
transmitted wave in the limit in which the plasmonic layer becomes a perfectly
ð2DÞ
conducting screen. In this case, afs tends to negative infinity and b approaches 1.
Alternatively, for z > 0, (3.47) takes the form
h i
i k0k :rk þ k0? z  x0 t
Ufþ
s
ðrk ; z; tÞ ¼ U0 ðrk ; z; tÞ  U0 e ; ð3:49Þ

or

Ufþ
s
ðrk ; z; tÞ ¼ 0; ð3:50Þ

which also shows that there is no transmission. On the other hand, for z < 0, (3.45)
becomes

i ½k0k :rk  k0? z  x0 t

Ufs ðrk ; z; tÞ ¼ U0 ðrk ; z; tÞ  U0 e ð3:51Þ

which indicates total reflection.

Addressing scalar field propagation in the vicinity of a perforated plasmonic
layer, (3.37), (3.38), (3.39) and (3.43) yield a delta function, dð2Þ ðk0k  kk Þ after r00k -
integration, which facilitates the kk -integration, with result for I1 ðrk ; z; xÞ as

x2 d ik0k :rk
I1 ðrk ; z; xÞ ¼ Gfs ðk0k ; z; 0; xÞ a2D
fs ðk0k ; xÞ2 pdðx  x0 ÞU0 e
c2
 
x2 d b Gfs ðrk ; 0; z; 0; xÞ ð3:52Þ
 2
c 1 þ b Gfs ð0; 0; 0; 0; xÞ
 Gfs ðk0k ; 0; 0; xÞa2D
fs ðk0k ; xÞ2 p dðx  x0 ÞU0 ;

and using (3.20), we have

x2 d ik0k :rk
I1 ðrk ; z; xÞ ¼ Gfs ðk0k ; z; 0; xÞ a2D
fs ðk0k ; xÞ2 pdðx  x0 ÞU0 e
c2
x2 d ð3:53Þ
 2 b Gðrk ; 0; z; 0; xÞGfs ðk0k ; 0; 0; xÞ
c
 a2Dfs ðk0k ; xÞ2 p dðx  x0 ÞU0 :
3 Wave Propagation and Diffraction Through a Subwavelength … 85

Bearing in mind (3.37) for Uðrk ; z; xÞ and using (3.39) for I2 ðrk ; z; xÞ with (3.53)
for I1 ðrk ; z; xÞ, and transforming from frequency to time representation, we obtain
h i
i k0k rk þ k0? jzjx0 t
Uðrk ; z; tÞ ¼ U0 ðrk ; z; tÞ  bU0 e ð3:54Þ
 b0 ½1  b U0 Gðrk ; 0; z; 0; x0 Þei x0 t ;

which can also be written in terms of Ufs as

Uðrk ; z; tÞ ¼ Ufs ðrk ; z; tÞ  b0 ½1  b U0 Gðrk ; 0; z; 0; x0 Þei x0 t ; ð3:55Þ

where
 h i3=2 
l 3 2
b0 ¼ c0 A2 ; c0 ¼ i ð q x0 RÞ  1  ðq x0 RÞ ð3:56Þ
3pR3

and Ufs ðrk ; z; tÞ is given by (3.44).

Equation (3.55) describes the total scalar field generated by the perforated 2D
plasmonic layer in the presence of the nano-hole. This result can be analyzed in terms
of the square modulus of Uðrk ; z; tÞ for z < 0 and for z > 0, both of which are given by
   
Uðrk ; z; tÞ2 Ufs ðrk ; z; tÞ2
  ¼  þ b0 b ½1  b2 Gðrk ; 0; z; 0; x0 Þ G ðrk ; 0; z; 0; x0 Þ
 U   U  0
0 0
" h i !#
i ½k0k :rk þ k0? z i k0k :rk þ k0? jzj
 ½1  b b0 G ðrk ; 0; z; 0; x0 Þ e  be
" h i h i !#
i k0k :rk þ k0? z i k0k :rk þ k0? jzj
 ½1  b b0 Gðrk ; 0; z; 0; x0 Þ e  be :

ð3:57Þ

This equation, which represents the total field distribution generated by the 2D
plasmonic layer in the presence of the nano-hole, is examined numerically
below using the Green’s function of (3.20) jointly with the results above for Gfs
(3.13, 3.25).
86 N.J.M. Horing and D. Miessein

3.4 Numerical Analysis of Helmholtz Scalar Wave

Propagation in the Vicinity of a Perforated Plasmonic
Layer with a Nano-Hole
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Considering a subwavelength nano-hole, ðqx0 RÞ\1 and 1  ðqx0 RÞ2 is real, the
associated frequency range for the incident scalar wave is given by
c
x0 \ qffiffiffiffiffiffiffiffiffi  x
e 0; ð3:58Þ
ð3DÞ
R eb

with xe 0 as the cut-off angular frequency. Based on the above results, the wave
amplitude field distributions for z < 0 and z > 0 are evaluated using (3.57),
exhibiting dependence on the radius R of the nano-hole.
For the transmitted scalar wave in the region z > 0, jzj ¼ z, the field distribution
of the wave is given by
 þ   2
U ðrk ; z; tÞ2 Ufsþ ðrk ; z; tÞ
  ¼  þ b0 b0 ½1  b Gðrk ; 0; z; 0; x0 Þ G ðrk ; 0; z; 0; x0 Þ
2
 U0   U0 
" h i h i !#
i k0k :rk þ k0? z i k0k :rk þ k0? z
 ½1  b b0 G ðrk ; 0; z; 0; x0 Þ e  be
" h i h i !#
i k0k :rk þ k0? z i k0k :rk þ k0? z
 ½1  b b0 Gðrk ; 0; z; 0; x0 Þ e  be :

ð3:59Þ

Considering the scalar wave to be incident upon the perforated 2D plasmonic

layer (with a nano-hole) at an angle θ0 with respect to the z-axis in the (x − z) plane

Fig. 3.3 The incident wave vector k0 in (x − z) plane impinging on the 2D plasmonic layer at
angle θ0 with respect to the normal z-axis, with frequency ω0
3 Wave Propagation and Diffraction Through a Subwavelength … 87

of incidence (Fig. 3.3), the components of the incident wave vector k0 are defined
as follows (k0  qx0 ):

k0k ¼ k0x ¼ qx0 sinðh0 Þ; k0y  0; k0? ¼ qx0 cosðh0 Þ: ð3:60Þ

Substitution of this into (3.59) yields

 þ 
U ðx; y; z; tÞ2
  ¼ ½1  b2
 U0 
" #
R2 b2 ½1  3ðqx0 RÞ2 þ 3ðqx0 RÞ4 
 1þ 2 2 4
½ð12  bÞ þ 3ð16  bÞðqx0 RÞ  3ð16  bÞbðqx0 RÞ ðrk2 þ z2 Þ
þ ½1  b2
2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffi 3
2bR 1  ðqx0 RÞ2 ½ð12  bÞ  3ð4  bÞðqx0 RÞ2  3bðqx0 RÞ4  cos½k0x x þ k0? z  qx0 rk2 þ z2 
6 7
4 5
½ð12  bÞ2 þ 3ð16  bÞðqx0 RÞ2  3ð16  bÞbðqx0 RÞ4 ðrk2 þ z2 Þ1=2

þ ½1  b2
2 qffiffiffiffiffiffiffiffiffiffiffiffiffi 3
2bRðqx0 RÞ½b  3ð4  bÞðqx0 RÞ2 þ 3bðqx0 RÞ4  sin½k0x x þ k0? z  qx0 rk2 þ z2 
 4 5;
½ð12  bÞ2 þ 3ð16  bÞðqx0 RÞ2  3ð16  bÞbðqx0 RÞ4 ðrk2 þ z2 Þ1=2

ð3:61Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where rk ¼ x2 þ y2 for qx0 R\1.

Fig. 3.4 θ0 = 0°—Transmitted field square modulii as functions of x produced by a perforated 2D

plasmonic layer of GaAs of thickness d ¼ 10 nm, effective mass m ¼ 0:067 m0 (where m0 is the
ð3DÞ
free-electron mass, n2D ¼ 4  1015 =cm2 with dielectric constant of the host medium eb ¼ 1 in
the presence of a nano-hole of radius R ¼ 5 nm) for angle of incidence h0 ¼ 0 , with incident
wavelength k0 ¼ 1000 nm and frequency f0 ¼ 300 THz for z ¼ 50 R [near-field (Blue)], z ¼ 250 R
[middle-field (Red)] and z ¼ 1000 R [far-field (Black)] regimes
88 N.J.M. Horing and D. Miessein

For the reflected and incident waves in the region z < 0, jzj ¼ z, the field
distribution is given by
    
U ðrk ; z; tÞ2 Ufs ðrk ; z; tÞ2
  ¼  þ b0 b ½1  b2 Gðrk ; 0; z; 0; x0 Þ G ðrk ; 0; z; 0; x0 Þ
 U0   U0  0
" h i h i !#
i k0k :rk þ k0? z i k0k :rk  k0? z
 ½1  b b0 G ðrk ; 0; z; 0; x0 Þ e  be
" h i h i !#
i k0k :rk þ k0? z i k0k :rk  k0? z
 ½1  b b0 Gðrk ; 0; z; 0; x0 Þ e  be :

ð3:62Þ

The numerical analysis is focused on the transmitted field, with z > 0 for
R ¼ 5 nm. The transmitted field distributions are exhibited as functions of x for
z > 0 with the parameters chosen as f0 ¼ x2p0 ¼ 300 THz and k0 ¼ 1000 nm. These
field distributions show the nature of the transmitted scalar radiation fields in the
subwavelength regime (R
k) in terms of near-field, middle-field and far-field
zones of diffraction for various values of z at y = 0.
Figures 3.4, 3.5, 3.6 and 3.7, exhibit the square modulii of the transmitted fields
jU þ ðx; y; z; tÞ=U0 j2 as functions of x at y = 0 for θ0 = 0°, θ0 = 30°, θ0 = 60°, and

Fig. 3.5 θ0 = 30°—Transmitted field square modulii as functions of x produced by a perforated

2D plasmonic layer of GaAs of thickness d ¼ 10 nm, effective mass m ¼ 0:067 m0 (where m0 is
ð3DÞ
the free-electron mass, n2D ¼ 4  1015 =cm2 with dielectric constant of the host medium eb ¼ 1

in the presence of a nano-hole of radius R ¼ 5 nm) for angle of incidence h0 ¼ 30 , with incident
wavelength k0 ¼ 1000 nm and frequency f0 ¼ 300 THz for z ¼ 50 R [near-field (Blue)], z ¼ 250 R
[middle-field (Red)] and z ¼ 1000 R [far-field (Black)] regimes
3 Wave Propagation and Diffraction Through a Subwavelength … 89

Fig. 3.6 h0 ¼ 60 —Transmitted field square modulii as functions of x produced by a perforated

2D plasmonic layer of GaAs of thickness d ¼ 10 nm, effective mass m ¼ 0:067 m0 (where m0 is
ð3DÞ
the free-electron mass, n2D ¼ 4  1015 =cm2 with dielectric constant of the host medium eb ¼ 1

in the presence of a nano-hole of radius R ¼ 5 nm) for angle of incidence h0 ¼ 60 , with incident
wavelength k0 ¼ 1000 nm and frequency f0 ¼ 300 THz for z ¼ 50 R [near-field (Blue)], z ¼ 250 R
[middle-field (Red)] and z ¼ 1000 R [far-field (Black)] regimes

θ0 = 80°, respectively, with several values of z = 50 R, z = 250 R and z = 1000 R in

each figure (taking R ¼ 5 nm and incident wave frequency f0 ¼ x2p0 ¼ 300 THz).
In addition to the transmitted field intensities illustrated above as functions of
x for fixed y = 0, we also exhibit the field amplitude distributions
jU þ ðx; y; z; tÞ=U0 j2 of the transmitted field intensities for the entire x − y plane in
terms of 3D and density plots, which characterize the spread of the transmitted
fields as they propagate in the positive z direction. The evaluations exhibited below
in Figs. 3.8, 3.9, 3.10 and 3.11 include angles of incidence θ0 = 0°, 30°, 60° and
80°, respectively; and each figure has constituent pairs of 3D and density plots for
each of the diffraction zones [near (z ¼ 50 R), middle (z ¼ 250 R) and far
(z ¼ 1000 R) zones].
 þ 2
 
Furthermore, the field distributions U ðx;y;z;tÞ
U0  of the transmitted waves have been
evaluated as functions of distance from the 2D perforated screen (z). It is readily seen
that for normal incidence (θ0 = 0°), the amplitude of the transmitted field distribution
decreases as distance from the aperture in the (x − y) plane, rk , increases, and it also
decreases as a function of increasing z. This distribution flattens uniformly for large
z. These features are illustrated in the 3D and density plots of the field as functions of
x and z for y = 0 in Fig. 3.12 below for θ0 = 0°, revealing the sharp decay of the field as
a function of z while spreading in the x-direction.
90 N.J.M. Horing and D. Miessein

Fig. 3.7 θ0 = 80°—Transmitted field square modulii as functions of x produced by a perforated

2D plasmonic layer of GaAs of thickness d ¼ 10 nm, effective mass m ¼ 0:067 m0 (where m0 is
ð3DÞ
the free-electron mass, n2D ¼ 4  1015 =cm2 with dielectric constant of the host medium eb ¼ 1

in the presence of a nano-hole of radius R ¼ 5 nm) for angle of incidence h0 ¼ 80 , with incident
wavelength k0 ¼ 1000 nm and frequency f0 ¼ 300 THz for z ¼ 50 R [near-field (Blue)], z ¼ 250 R
[middle-field (Red)] and z ¼ 1000 R [far-field (Black)] regimes
 þ 2
 
Moreover, we have also calculated the field distributions U ðx;y;z;tÞ
U0  of the
transmitted wave as functions of the incident frequency f = f0 (THz) at a fixed finite
x-value (x ¼ 200 R) with y = 0 for incident angles h0 ¼ 0 , h0 ¼ 30 , h0 ¼ 60 , and
h0 ¼ 80 , for the near-field (z ¼ 50 R), middle-field (z ¼ 250 R) and far-field
(z ¼ 1000 R) zones of diffraction. These results are shown in Figs. 3.13, 3.14, 3.15
and 3.16:
Furthermore, the 3D and density plot patterns of the transmitted field distribu-
 þ 2
 
tions U ðx;y;z;tÞ
U0  are exhibited in Figs. 3.17, 3.18, 3.19 and 3.20 as functions of
position x and incoming frequency f ¼ f0 for incident angles h0 ¼ 0 ; 30 ; 60 ; 80
with values of z ¼ 50 R; 250 R; and 1000 R for near, middle and far zones,
respectively, (y ≡ 0). At normal incidence, symmetry is observed in the intensity
distributions in the 3D plots in position space, while the distributions decay as
functions of frequency.
In addition to the field distributions of transmitted waves as functions of position
 þ 2
 
and frequency f (THz), we have also determined the field distributions U ðx;y;z;tÞ
U0  of
transmitted waves as functions of incident angle θ0 for near-field, middle-field and
far-field zones of diffraction for z ¼ 50 R; 250 R and 1000 R, respectively, and x ¼
500 R with y = 0. These results are presented in Fig. 3.21 ðR ¼ 5 nm and x ¼ 500 RÞ.
3 Wave Propagation and Diffraction Through a Subwavelength … 91

Fig. 3.8 θ0 = 0°—Transmitted field distribution of a perforated GaAs layer in terms of 3D

(a, c, e) and density (b, d, f) plots
92 N.J.M. Horing and D. Miessein

(a) (b)

(c) (d)

(e) (f)

Fig. 3.9 θ0 = 30°—Transmitted field distribution of a perforated GaAs layer in terms of 3D

(a, c, e) and density (b, d, f) plots
3 Wave Propagation and Diffraction Through a Subwavelength … 93

(a) (b)

(c) (d)

(e) (f)

Fig. 3.10 θ0 = 60°—Transmitted field distribution of a perforated GaAs layer in terms of 3D

(a, c, e) and density (b, d, f) plots
94 N.J.M. Horing and D. Miessein

(a) (b)

(c) (d)

(e) (f)

Fig. 3.11 θ0 = 80°—Transmitted field distribution of a perforated GaAs layer in terms of 3D

(a, c, e) and density (b, d, f) plots
3 Wave Propagation and Diffraction Through a Subwavelength … 95

Fig. 3.12 Transmitted field distribution as a function of x and z for y ≡ 0 (a 3D and b density
plots) of the 2D plasmonic layer of GaAs of thickness d ¼ 10 nm in the presence of a nanohole of
R ¼ 5 nm at angle h0 ¼ 0 , with incident wavelength k0 ¼ 1000 nm and frequency
f ¼ f0 ¼ 300 THz
96 N.J.M. Horing and D. Miessein

Fig. 3.13 h0 ¼ 0 and x ¼ 200 R (y  0): Transmitted field distributions as functions of frequency
f produced by a perforated 2D plasmonic layer of GaAs of thickness d ¼ 10 nm, effective mass
m ¼ 0:067 m0 (where m0 is the free-electron mass, n2D ¼ 4  1015 =cm2 with dielectric constant
ð3DÞ
of the host medium eb ¼ 1 in the presence of the nano-hole of R ¼ 5 nm); for z ¼ 50 R (Blue),
z ¼ 250 R (Red) and z ¼ 1000 R (Black)

Fig. 3.14 h0 ¼ 30 and x ¼ 200 R (y  0): Transmitted field distributions as functions of

frequency f produced by a perforated 2D plasmonic layer of GaAs of thickness d ¼ 10 nm,
effective mass m ¼ 0:067 m0 (where m0 is the free-electron mass, n2D ¼ 4  1015 =cm2 with
ð3DÞ
dielectric constant of the host medium eb ¼ 1 in the presence of the nano-hole of R ¼ 5 nm); for
z ¼ 50 R (Blue), z ¼ 250 R (Red) and z ¼ 1000 R (Black)
3 Wave Propagation and Diffraction Through a Subwavelength … 97

Fig. 3.15 h0 ¼ 60 and x ¼ 200 R (y  0): Transmitted field distributions as functions of

frequency f produced by a perforated 2D plasmonic layer of GaAs of thickness d ¼ 10 nm,
effective mass m ¼ 0:067 m0 (where m0 is the free-electron mass, n2D ¼ 4  1015 =cm2 with
ð3DÞ
dielectric constant of the host medium eb ¼ 1 in the presence of the nano-hole of R ¼ 5 nm); for
z ¼ 50 R (Blue), z ¼ 250 R (Red) and z ¼ 1000 R (Black)

Fig. 3.16 h0 ¼ 80 and x ¼ 200 R (y  0): Transmitted field distributions as functions of

frequency f produced by a perforated 2D plasmonic layer of GaAs of thickness d ¼ 10 nm,
effective mass m ¼ 0:067 m0 (where m0 is the free-electron mass, n2D = 4 × 1015/cm2 with
ð3DÞ
dielectric constant of the host medium eb ¼ 1 in the presence of the nano-hole of R ¼ 5 nm); for
z ¼ 50 R (Blue), z ¼ 250 R (Red) and z ¼ 1000 R (Black)
98 N.J.M. Horing and D. Miessein

Fig. 3.17 Field distributions as functions of x and f = f0 in terms of 3D (a, c, e) and density
(b, d, f) plots
3 Wave Propagation and Diffraction Through a Subwavelength … 99

Fig. 3.18 Field distributions as functions of x and f = f0 in terms of 3D (a, c, e) and density
(b, d, f) plots
100 N.J.M. Horing and D. Miessein

Fig. 3.19 Field distributions as functions of x and f = f0 in terms of 3D (a, c, e) and density
(b, d, f) plots

 þ 2
 
It is also of interest to analyze the field distributions U ðx;y;z;tÞ
U0  of the trans-
mitted waves as functions of the nano-hole radius R for a fixed value of z with
x = y = 0. In this case, the acceptable radius range for a subwavelength nano-hole is
given by R \ 1=qx0 as
3 Wave Propagation and Diffraction Through a Subwavelength … 101

Fig. 3.20 Field distributions as functions of x and f = f0 in terms of 3D (a, c, e) and density
(b, d, f) plots
102 N.J.M. Horing and D. Miessein

Fig. 3.21 Transmitted field distributions produced by the 2D plasmonic layer of GaAs of
thickness d ¼ 10 nm, effective mass m ¼ 0:067 m0 where m0 is the free-electron mass, n2D ¼
ð3DÞ
4  1015 =cm2 with dielectric constant of the host medium eb ¼ 1 in the presence of the
nano-hole of R ¼ 5 nm, with incident wavelength k0 ¼ 1000 nm and linear frequency f0 ¼
300 THz as functions of θ0 for z ¼ 50 R (near-field) (Blue), z ¼ 250 R (middle-field) (Red) and
z ¼ 1000 R (far-field) (Black)

Fig. 3.22 Transmitted field distributions produced by a perforated 2D plasmonic layer (GaAs of
thickness d ¼ 10 nm) as functions of the hole radius R ðnmÞ with incident wavelength and
frequency: k0 ¼ 1000 nm and f0 ¼ 300 THz (Blue), k0 ¼ 500 nm and f0 ¼ 600 THz (Red), and
k0 ¼ 250 nm and f0 ¼ 1200 THz (Black), for z ¼ 100 nm with x = y = 0 and normal incidence
3 Wave Propagation and Diffraction Through a Subwavelength … 103

c e
R qffiffiffiffiffiffiffiffiffi  R: ð3:63Þ
ð3DÞ
x0 eb

The fixed point for which the transmitted field distributions are exhibited as
functions of R is z ¼ 100 nm; x ¼ y ¼ 0. This is illustrated in Fig. 3.22 for normal
incidence, h0 ¼ 0 , for f0 ¼ 300; 600; 1200 THz.

3.5 Summary

The subject of wave propagation, transmission and diffraction through a sub-

wavelength nanohole aperture embedded in a thin plasmonic layer/screen has been
reviewed in this article in terms of a prototypical scalar wave. This analytic
approach involving closed form solutions of successive integral equations has
facilitated relatively simple numerical computations exhibited in Sect. 3.4, and is
not in any way restricted to a metallic screen. Moreover, this formulation, which is
based on the use of successive integral equations for the associated Helmholtz
Green’s function involved, automatically incorporates the boundary conditions,
which would otherwise need to be addressed explicitly. It also incorporates the role
of the thin plasmonic layer, which is smeared by its lateral wavenumber
dependence.
In the case of normal incidence, the calculated results shown in the figures of
Sect. 3.4 exhibit a positional broadening of the central transmission maximum as
z increases from the near to the far zones. This is to be expected as a simple
geometric spreading of the radiation emanating from the nano-hole itself, but it is
also interesting to note that the transmitted radiation is well supported at large
lateral distances from the nano-hole by transmission directly through the plasmonic
sheet. Moreover, the figures also exhibit interference fringes as a function of lateral
distance from the nano-hole in presence of transmission through the sheet as well as
through the hole; and as the lateral distance from the hole increases, this interfer-
ence diminishes and the transmission tends to flatten to that emerging through the
sheet alone, in the absence of the distant nano-hole. As the angle of incidence
increases, the figures of Sect. 3.4 show that the axis of the central maximum follows
it, accompanied by a spatial compression of interference fringe maxima forward of
the axis and a spatial thinning of the fringe maxima behind it. Furthermore, the
figures also exhibit the dependence of the transmitted/diffracted field intensity on
incident frequency f ≡ f0 and nano-hole size (Fig. 3.22).
Finally, the interested reader will find an account of the corresponding electro-
magnetic vector field propagation/transmission/diffraction through a subwavelength
nano-hole aperture in a thin plasmonic screen in research articles by Horing et al.
[15], and Miessein et al. [16].

Acknowledgments We gratefully acknowledge support from the NSF-AGEP program.

104 N.J.M. Horing and D. Miessein

References

1. H. Cao, A. Nahata, Influence of aperture shape on the transmission properties of a periodic

array of subwavelength apertures. Opt. Express 12, 16 (2004)
2. L. Chen, J.T. Robinson, M. Lipson, Role of radiation and surface plasmon polaritons in the
optical interactions between a nano-slit and a nano-groove on a metal surface. Opt. Express 14,
26 (2006)
3. C. Genet, T. Ebbesen, Light in tiny holes. Nature 445, 39–46 (2007)
4. R. Gordon, L.K.S. Kumar, A.G. Brolo, Resonant light transmission through a nanohole in a
metal film. IEEE Trans. Nanotechnol. 5(3), 291–294 (2006)
5. A. Hajiaboli, M. Kahrizi, V. Truong, Optical behaviour of thick gold and silver films with
periodic circular nanohole arrays. J. Phys. D: Appl. Phys. 45, 485105 (2012)
6. S.V. Kukhlevsky, M. Mechler, O. Samek, K. Janssens, Analytical model of the enhanced light
transmission through subwavelength metal slits: green’s function formalism versus
Rayleigh’sexpansion. Appl. Phys. B: Lasers Optics 84, 19–24 (2006)
7. H. Leong, J. Guo, A surface plasmon resonance spectrometer using a super-period nanohole
array. Opt. Express 20(19), 21318–21323 (2012)
8. P.N. Melentiev, A.E. Afansiev, A.A. Kuzin, A.V. Zablotskiy, A.S. Baturin, V.I. Balykin,
Single nanohole and photonic crystal: wavelength selective enhanced transmission of light.
Opt. Express 19, 23 (2011)
9. O.M. Mendez, M. Cadilhac, R. Petit, Diffraction of a two-dimensional electromagnetic beam
wave by a thick slit pierced in a perfectly conducting screen. J. Opt. Soc. Am. 73, 3 (1983)
10. J.M. McMahon, J. Henzie, T.W. Odom, G.C. Schatz, S.K. Gray, Tailoring the sensing
capabilities of nanohole arrays in gold films with Rayleigh anomaly-surface plasmon
polaritons. Opt. Express 15(26), 18119–18129 (2007)
11. V.A.G. Rivera, in Plasmonics—Principles and Applications, chapter 7, ed. by Kim, Kiyoung
(InTech 2012), p.157
12. Y. Wang, Y. Qin, Z. Zhang, IEEE Photon. J. 6(4) (2014)
13. N.J.M. Horing, T.Y. Bagaeva, V.V. Popov, Excitation of radiative polaritons in a
two-dimensional excitonic layer by a light pulse. J. Opt. Soc. Am. B 24, 2428–2435 (2007)
14. N.J.M. Horing, Radiative plasmonic-polariton dispersion relation for a thin metallic foil with
interband damping transitions. IEEE Sens. J. 8, 771–774 (2008)
15. N.J.M. Horing, D. Miessein, G. Gumbs, Electromagnetic wave transmission through a
subwavelength nano-hole in a two-dimensional plasmonic layer. J. Opt. Soc. Am. A 32(6),
1184–1198 (2015)
16. D. Miessein, N.J.M. Horing, H. Lenzing, Incident-angle dependence of electromagnetic wave
transmission through a nano-hole in a thin plasmonic semiconductor layer, in preparation for
publication
Chapter 4
The Challenge to Develop Metrology
at the Nanoscale

R. Ince

Abstract Since nanotechnology goods are manufactured and utilised by the

community, legal metrology, human safety, and the environment demand traceable
measurement techniques. This is the business of the international network of
measurements called metrology. After 2017 the realisation of the international
metre will be through the lattice parameter of silicon or another suitable crystalline
material. Many NMI’s have developed traceable instrumentation systems primarily
for AFM, but only partly for SEM and optical instrumentation. None of the existing
techniques is able to meet the present requirements for reliable metrology of
nanomaterials. However suitable reference materials are being developed alongside
standardised sample preparation methods. Present second generation nanostructures
are complex requiring multiparameter and ensemble measurements that AFM or
SEM cannot offer. Ensemble techniques reveal the sub-nanometre detail required in
the healthcare and electronics industry. The next generation of manufacturing in
these industries will be three-dimensional complex sub-nm architectures, and
nanometrology is currently being driven there.

4.1 Metrology

Metrology is the science of measurement; its mission and implementation is

international and ensures that each of the seven international SI units [1] is reliable
at all the levels shown in Fig. 4.1. This is achieved via traceability to the repre-
sentation of each unit held in national laboratories through a chain of unbroken
documented calibrations. Representation of each unit is the experiment that realises
the unit in national standards laboratories. The national representations must
therefore be traceable to the international values (some still held at the international

R. Ince (&)
Department of Physics, Yeditepe University,
Kayisdagi Caddessi, Kadikoy, 34755 Istanbul, Turkey
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 105

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_4
106 R. Ince

Decreasing
measurement
accuracy
BIPM

National standards

Primary standards

Working standards

Research, industrial and community instruments

Fig. 4.1 Traceability pyramid showing laboratory hierarchy. Absolute accuracy refers to how
close a measurement is to a traceable standard held in the national standards laboratory

bureau of weights and measures, BIPM in Paris) by international comparisons

which is described by an uncertainty (a value expressing how near to the real value
the measurement is). This creates a highly correlated set of international base units.
Scientists themselves are often unaware that such a vast network of perpetual
calibrations exists at so many levels (traceability pyramid) and underscores all
global measurements for science, trade, and research. Indeed the SI is just as
important a part of the national infrastructure as roads, buildings and bridges; the
traceability pyramids in Fig. 4.1 illustrate the quintessence of the organization.
Thus, metrological validation is achieved through comparisons and peer review
resulting in a number that expresses the uncertainty called the combined uncer-
tainty, Uc together with the level of confidence (usually 95 %). The value Uc
includes in it all the uncertainties that contribute to the measurements and all those
which may affect the measurement as the root sum of the squares of each:
Pythagorus’ theory taken to the nth dimension.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Uc ¼ u21 þ u22 þ    u2n ð4:1Þ

At the beginning of the SI system in 1869 the experiments involved transfer

artefacts; macro sized objects such as a platinum-iridium cylinder (1 kg) for mass or
a bar (1 m) for length. These have very well characterised values. These artefacts
are calibrated against others via metrological instruments.
However, even in 1870 James Clerk Maxwell pointed out that artefacts were
subject to local conditions, thus presented a barrier to imperishable standards. He
predicted their replacement with the parameters of microscopic, universal standards.
Over the twentieth century this artefact barrier has been removed and most
4 The Challenge to Develop Metrology at the Nanoscale 107

Fig. 4.2 The Silicon unit

cell, a cubic packing
arrangement of 8 atoms. Its
volume is determined from
the lattice parameter a, the
length of one its sides

standards have long been replaced by fundamental constants of nature. The last
artefact the unit of mass is due to be replaced in 2017 with a definition from
Planck’s constant, h [2].
The method by which this has been achieved in the international Avogadro
project (2003) [3] is testament to the singular importance of optical and spectro-
scopic methods in determining dimensions of nanostructures. The lattice constant a,
volume, and density ρ of a 1 kg single-crystal sphere, highly enriched with the 28Si
were all determined with unprecedented accuracy via optical and spectroscopic
methods. The structure of the silicon unit cell is shown in Fig. 4.2.
This enabled determination of the Avogadro number NA by the equation

NA ¼ nM qa3 ð4:2Þ

where n is the number of atoms per unit cell and M is molar mass.
The Avogadro number is directly linked to Planck’s constant via three other SI
microscopic representations as:

ca2 Ar M
h¼ ð4:3Þ
2R1 NA

where α is the hyperfine splitting constant, R∞ is the Rydberg gas constant, Ar is

relative atomic mass.
The SI definition of length is based on the speed of light, c, in a vacuum. This is
realised by measurement of stabilised laser wavelengths using highly sophisticated
interferometry of fixed length metal block artefacts (gauge blocks) to accuracies of
about 3 × 10−8 under highly controlled conditions; here the wavelength itself must
be sub-divided to the nanometre and sub-nm scales required. However at primary
level it is achieved by comparator calibration against calibration artefacts called
gauge blocks (0.5–100 mm) to an accuracy of 2 × 10−6 under highly controlled
conditions [4]. These macroscale artefacts can be picked up and placed in position
108 R. Ince

for comparison; imagine the challenge of extending the traceability to nanoscale

artefacts that cannot even be seen by the naked eye. What does a straight line or a
flat surface mean on the nanoscale?

4.2 Nanotechnology Generations, Definitions, and Visions

Technology has innovated away from the 1st generation of passive nanostructures
which performed single tasks to the 2nd generation of nanotechnology ‘Active
nanostructures’ which multitask e.g. targeted delivery of drugs, the third generation
will be nanosystems with thousands of interacting components and several years
into the future further innovation to integrated nanosystems will form the 4th
generation of systems within systems e.g. molecular manufacturing of genes inside
the DNA of targeted cells, and nanosurgery for healing wounds on the cellular
level. The aim is to achieve nanoscale control of matter by processes that construct
precise three-dimensional structures through molecular manufacturing. This is
challenging because quantum physics becomes more important as structures shrink
to nanoscale. However, although we have an incomplete knowledge of quantum
physics, nature has not; it has built these machines for billions of years as
self-assembled viruses and motors, springs, tracks and actuators within cells [5].
In 2011 the European the Commission adopted the definition of nanotechnology
as the study of phenomena and fine-tuning of materials at atomic, molecular and
macromolecular scales, where properties differ significantly to those at a larger
scale. The US National Nanotechnology Initiative defines nanotechnology as the
understanding and control of matter engineering of nanosystems and manufacturing
them. The measurement of length or size, force, mass, electrical and other prop-
erties is included in nanometrology.
Nanotechnology has two primary research thrusts
• Nanoscale electronics (nanowires, connections, memory systems, displays,
nanosensors, NEMS, and the protein computer)
• Medicine (drug delivery via functionalization, improved imaging, materials to
repair blood vessels and cartilage, nanorobots [6])
Nanotechnology also has two secondary research thrusts
• Nanoenergy (Longer range electric cars with super batteries, strong lightweight
materials via nanocomposites, cheaper energy)
• Environment (purifying sea water, cleaning water supplies, removing
contaminants)
Nanotechnology also has three ivory tower projects under consideration:
• The Space elevator [7–10],
• The protein-based computer [11],
4 The Challenge to Develop Metrology at the Nanoscale 109

• Molecular fabricators [12–14],

• Nanomachines for cellular repair [12, 15–18].

4.3 Nanotechnology Research Drives

From ancient times, metrology has always been driven by trade and industry; today
half of all R&D spending is in healthcare and semiconductors, nanometrology is
driven by these industries. Healthcare depends on molecular metrology (the
bottom-up approach), whilst the semiconductor industry depends on engineering
metrology (top-down approach).
Although research in molecular nanotechnology has only been boosted in the
past 10 years, a research rollercoaster began in 1948 in the semiconductor industry,
within a few years the world’s first transistor computer was built at the University
of Manchester in 1953 and commercialized within 3 years [19]. Burgeoning
demand for integration into more and more products, and for ever denser chips
(currently 995 million transistors per chip at Intel) with better performance and
speeds, innovated dimensional metrology.
Measurements of nanosystems are similar to those of the macroscopic world:
distance, width, height, form, texture/roughness, volume and physical and chemical
characteristics. However, unlike macroscale metrology, the dimensions of nanos-
tructures play a major role in almost all their physical and chemical properties
(functionality). For example, when crystal size is smaller than the mean free path of
an electron, the conductivity of the crystal is modified; the colour of nanogold goes
from the near infra-red through the full visible spectrum to blue, dependent on both
size 2–100 nm and shape; seemingly inert substances become catalysts at nan-
odimensions, stresses in nanosystems become discretized. The latter (as mentioned
in the previous section) directly affect the semiconductor and healthcare industries.
In anticipation of this profound link, the most common way to classify nano
structures is via their dimensions. Thus, dimensional metrology is the natural basis
for all nanometrology; but the link between nanostructure dimension (nanodi-
mensional metrology) and functionality must be strengthened.

4.3.1 The Semiconductor Industry

Metrology reduces the cost of manufacturing by bringing in more robust processes,

preventing scrap, and helping in ramping and maintaining yield [20].
Metrology tools are required to measure and monitor film thickness, refractive
index, resistivity, and stress; metrology systems for defect inspection are essential.
For high performance the manufacturing parameters must be controlled: IC line
widths and edge profiles must fall within tight tolerances. The minimum feature size
110 R. Ince

Table 4.1 Metrology systems utilised for measurements in the semiconductor industry
Process step Measurement Metrology system
Silicon Resistivity (dependent on thickness), Four-point probe,
manufacturing/epitaxy thickness FTIR
Wafer inspection Defects Defect inspection
system
Dielectric deposition Thickness, refractive index, stress, dielectric Ellipsometer,
constant stress gauge
Etching Pattern defects, uniformity, etch profile and AFM, SEM,
selectivity reflectometer
Lithography Critical dimension, pattern defects SEM
Yield monitoring Correlation of metrology and inspection Data management
results to yield system

in microchips today is 32 nm wide, yet research aims to decrease this to 20 nm

using extreme UV nanolithography. Film thickness must be accurate and uniform
and the devices must be free of defects that affect yield: defects as small as a third of
the linewidth cause electrical failure. Another important factor is in monitoring of
Nano dimensions, since semiconductor wafers tend to change thickness over time
due to environmental factors.
As film thickness decreases, it approaches that of the interlayer itself, making it
extremely difficult to manufacture or measure with assured accuracy [21]. This is
due to limits on instrumentation, data analysis procedures, and the fact that for
process control, measurements need to be made to a level below atomic dimensions
(Table 4.1).

4.3.2 The Healthcare Industry

Molecular nanometrology aims to advance the little explored and difficult field of
measuring nanometre distances/structures in situ, which lies at the heart of
molecular biology and medicine. This will lead to improved targeting of drugs to
diseased organs and cells reduce side effects and improve efficacy of drugs,
improved medical imaging technologies (due to utilisation of magnetic nanoparti-
cles iron oxide) will diagnose diseases earlier on, enabling precise and effective
intervention resulting in lower costs.
Important biomedical application areas involve trapping viruses, [22] bacteria,
living cells, repairing bones, implant of soft tissues, functionalised nanoparticles for
in vitro diagnostics. Here greater control on the nanoscale is required.
Many new drugs have poor solubility or poor permeability, however, through
nanotechnology scientists have realised that size distributions less than 100 nm
increase surface area, solubility and bioavailability. Thus, ‘active pharmaceutical
ingredients’ is a hot topic [23, 24].
4 The Challenge to Develop Metrology at the Nanoscale 111

Table 4.2 Metrology systems utilised for measurements in the healthcare industry
Process Measurement Metrology system
Investigate nanostructures Dimension Transmission electron
internally microscopy (TEM)
Trap: Force, dimension,
Dielectric spheres, viruses, motion tracking
bacteria, living cells, organelles,
small metal particles, DNA
strands.
Cell sorting tracking bacteria
Imaging Difficult in biology due AFM
to low conductivity
Imaging Surface topography, Optical Tweezers creates an
correlation with ‘optical trap’ to hold
various physical particles < 1 nm at its centre
properties
Process Measurement Scanning tunnelling
microscope (STM)

Instruments utilised in the healthcare sciences for nanometrology are Electron

Beam Techniques, Scanning Probe Techniques, Optical Tweezers Table 4.2.

4.4 Nanotechnology Artefacts

First we will discuss nanosystems as possible nanometrology artefacts for future

nanometrology based on nanosystems presently available due to research, and then
we will discuss present nanometrology instruments available for their calibration.
There are 13 main elements and compounds in the literature from which
nanostructures are produced. The references in the Table 4.3 show their major area
of application in the past 3 years only.
Nanoparticles of the materials in Table 4.3 are mainly produced when their solid
particles are vaporised whilst moving on a carrier gas and then allowed to cool in a
chamber. They may be vaporised by plasma, laser ablation, or high voltage radio
frequency pulses across electrodes.
Presently, there are five overarching nanostructures broadly implemented in the
above research thrusts in order: films, composites, nanowires, crystals, rods,
fullerenes, tubes, and quantum dots. There is also a menagerie of flourishing sec-
ondary nanostructures that demonstrate potential for innovation of any of the major
research thrusts. The references attached in the Table 4.4 are noteworthy as the most
cited research papers on nanostructures (not including review papers) in each of the
past 3 years (Table 4.5):
112 R. Ince

Table 4.3 Most widely utilised nano materials

Material Application References Times cited
Carbon
Nanoparticles EST [60] 111
Buckyballs [61] Nanobot components [62] 25
Nanotubes [63] E [64] 111
Graphene [65] E [66] 204
Diamondoid [6, 67, 68] Mechanosynthesis [69] 11
Platinum Enhanced catalysis [70] 182
Gold Chemo-sensor [71] 120
Silica TDD [72] 102
Boron nitride electronic components [73] 89
Boron TDD [74] 15
Silicon EST, nanoantennae/metamaterials [75, 76] 86, 85
Silver Light amplification/photonics [77] 68
Iron oxide TDD [78] 63
Iron Enhanced catalysis [79, 80] 28, 28
Titanium dioxide Enhanced electric biosensor [81] 59
Palladium Enhanced catalysis [82] 40
Neodymium Sub tissue optical probes [83] 37
ECT/EST energy conversion/storage technology, E electronic devices and components, TDD
targeted drug delivery

4.5 Nanomaterials Measurement Challenges

Physical dimensions of nanosystems must be measured with traceability and novel

chemical and physical phenomena accounted for in order to engineer and manu-
facture nanosystems for precise applications. The definition of a nanomaterial
includes measurements of the size and size distributions of Nano-objects embedded
in another material [25]. During fabrication these systems require measurement
down to accuracies of 0.1 nm. However, robust, easy to use measurement methods
for their characterisation are not present. This means a revolutionary ‘holistic’
approach is required for nanoscale metrology. This must involve morphing some of
the techniques below or development of Innovative techniques such as X-ray
interference, X-ray diffraction, small angle X-ray spectroscopy (SAXS) and
non-linear optical microscopy and spectroscopy (e.g. Raman).
4 The Challenge to Develop Metrology at the Nanoscale 113

Table 4.4 Overarching nanostructures

Overarching Topic Times References
nanostructures cited
Films 2013 SC 901 [84]
2014 Chemo sensor 8 [85]
2015 Dental adhesives 1 [86]
Composites 2013 SC 373 [87]
2014 EST 80 [88]
2015 Bio sensor, TDD, tissue 6, 5, 5 [89, 90]
engineering
Wires 2013 SC 244 [91]
2014 EST 60 [92]
2015 PC 5 [93]
Crystals 2013 EST 135 [94]
2014 PC 37 [95]
2015 SC 9 [96]
Rods 2013 SC 134 [97]
2014 PC 36 [98]
2015 EST/ECT, artificial retinas 3,1 [99, 100]
Fullerenes 2013 SC 130 [101]
2014 SC 30 [102]
2015 SC 5 [103]
Tubes 2013 Carbon computer 111 [64]
2014 EST 60 [92]
2015 Bio sensor 6 [104]
Quantum dots 2013 Optical imaging agents 109 [105]
2014 SC 58 [106]
2015 PC 4 [107]
SC solar cells, PC Photocatalysis for transportation fuels

4.6 Nanomaterials Instruments

Spectroscopic ellipsometers, scanning probe microscopes and AFM and SEM are
widely used in nanotechnology, however, legal metrology, human safety and the
environment demand traceable measurement techniques—nanometrology. This
requires experience in the determination of uncertainties and metrological valida-
tion through comparisons and peer review. The metrological problems associated
with these devices are discussed below.
114 R. Ince

Table 4.5 Secondary nanostructures

Secondary Topic Times References
nanostructures cited
Fibers 2013 Functional materials/E 901 [108]
2014 EST 8 [109]
2015 PC/ECT 1 [110]
Nanopores 2013 EST 373 [111]
2014 Molecular sieve 80 [112]
2015 E 6,5,5 [113]
Cages 2013 EST 244 [114]
2014 TDD 60 [115]
2015 Catalytic enhancement 5 [116]
Pillars 2013 EST 135 [117]
2014 Enhanced ferromagnetism 37 [118]
2015 Reversible adhesives, chemo sensor 9 [119, 120]
Needles 2013 EST 134 [121]
2014 ECT 36 [122]
2015 EST 3,1 [123]
Meshes 2013 Catalytic enhancement 130 [124]
2014 Foldable photoelectronics/ 30 [125]
muscle-like transducers
2015 Energy conservation 5 [126]
Foams 2013 Efficient reaction medium 111 [127]
2014 Nanoscaffolding/architecture, 60 [128, 129]
efficient reaction medium
2015 Super insulation, radiation protection [130, 131]

4.6.1 Spectroscopic Ellipsometry (SE)

This is an indirect, non-imaging method for dimensional measurement of thin film

thickness but also extricates refractive index. They are a non-contact/non-
destructive method for solid/air or liquid/air interfaces and are used at every
stage of semiconductor processing. Fresnel’s equations are utilised to calculate
theoretical spectra to be compared with the measured spectra. Regression analysis is
then performed by varying the parameters of interest until a best fit is obtained [26].
However, some materials have a refractive index which varies as a function of
position and more elaborate models are required to analyse them; this directly
affects accuracy (Table 4.6).
The preferred approach for establishing and maintaining accuracy of measure-
ment tool performance is through use of transfer artefacts materials [27–29].
4 The Challenge to Develop Metrology at the Nanoscale 115

Table 4.6 Typical commercial SE

Spectral Resolution/nm Measurement Accuracy (NMI)/ Repeatability Possible
range/nm time/s nm (NMI)/nm traceability
440– 2 Typical 5 d = 100 nm ± 0.04 d = 15 nm ± 0.02 Wafer
1000 transfer
artefacts
from NMIa
a
Only for very well-characterised materials systems (e.g. Silicon oxide on silicon)

4.6.1.1 Main Measurement Uncertainty Contributions

• Choice of starting assumptions during data analysis, e.g., different dielectric

functions available in the literature
• Correlation of variables when extracting multiple parameters in the thin film
regime increases the uncertainty for evaluated parameters.
These latter two issues may not be of concern in a particular manufacturing plant
because a fixed choice for the silicon dielectric function can be locally standardized
on, and the main concern will be detecting small changes from baseline values.
However, these issues strongly impact the certification of reference materials. It is
not yet possible with SE measurements alone to extract values of all the parameters
for a thin film with its interface layer and to provide certified values with 1 %
uncertainty at a 95 % confidence level.

4.6.2 Metrological AFM

This is a nanotechnology analogy of the coordinate measuring machine (CMM), the

present system for calibrating gauge blocks. It realises real three-dimensional
nanometrology, and like the CMM the scanning method over small areas is time
consuming. However, a need exists to modify existing, commercial nanoscale
measurement systems into routinely available instruments which provide traceable
measurements for nanometre scale calibration [24, 30].
In order to ensure traceability at a higher level, several NMIs have developed
metrological AFMs. Presently about 17 NMIs worldwide either have metrological
AFM in use or in development [31–34]. The largest NMIs have already developed
second or third generation metrological AFMs [30, 35–39]. Several international
comparisons have been organized in nanometrology [40–44]. One of the main
standards is step height, h (0.3–2000 nm).
A step height standard in amorphous silicon dioxide is defined in analogy to the
international standard organisation’s (ISO) 5436 definition [43] (Fig. 4.3).
116 R. Ince

Fig. 4.3 A step-height W

standard [43]
1/2W
1/2W 1/2W

A C B

2.5W

Table 4.7 Metrological AFM

Uncertainty/nm Measurement Expanded Repeatability Traceability
time/s Uncertainty (NMI)/nm
Uc/nm
0.1 [132] 15 μm × 12 μm. SHN 43.00 −0.8 NMI
>5 min 1.06 [133] via
0.1 SHN = 1012.50 −0.26 laser
1.14 [133] standards
0.18 SHN = 7.0 0.1
0.43 [134]
0.18 SHN 1000 nm 0.4
0.89 [134]
SHN Step height nominal value, Uc at 95 % confidence

A coordinated and optimised approach to the development of traceability at the

nanometre scale is pursued in Europe by the European Association of National
Metrology Institutes (EURAMET) projects to ensure (Tables 4.7):
• Development of ready-to-use methodologies for calibration of

– step height
– line width
– grating pitch
– surface roughness parameters
• Development of a methodology for general shape measurements of the various
nanostructures
• Study of the effects of data post-processing
• Elaboration of good practice guides to minimise operator influence on the results
• Preparation and realisation of an inter-laboratory comparison with all interested
European NMIs on

– step height
– line width (Fig. 4.4)
4 The Challenge to Develop Metrology at the Nanoscale 117

Fig. 4.4 Transfer standard image in AFM [58]

4.6.2.1 Main Measurement Uncertainty Contributions

Laser calibration, step height roughness, sample temperature, imperfect shape of

tip.

4.6.3 Scanning Electron Microscopy (SEM)

Projects a two-dimensional image of a three-dimensional object, so is not appro-

priate for measurements of three-dimensional profiles. The resultant image is pro-
cessed to realise traceable measurements.
The interaction of secondary electrons with nanostructures being measured
changes the dimension. Modelling is required to de-convolute this from the image
[45].
SEM can inspect, characterise, and measure huge quantities of nanomaterials
(such as carbon nanotubes or cellulose nanocrystals) and over a much greater area
of the sample and at higher speed than AFM. However, SEM has poorer resolution
than AFM.
Measurements impact the healthcare (nanobio) and semiconductor industries.
A nanostructure can be calibrated by a metrological AFM and can then be used
as a 3D calibration artefact for other types of SPMs or SEMs. Gold nanoparticle
reference materials of size 10 nm and over are provided by some NMI’s [46]
(Table 4.8).

Table 4.8 Metrological SEM

Resolution/nm Measurement Expanded Repeatability/nm Traceability
[46] time/s uncertainty [136] [137]
Uc/nm [135]
Voltage 60 mm × 100–1500 0.3 Metrological
30 keV 0.4 130 mm < 60 1% AFM
Voltage 1 keV
1.6
Uc at 95 % confidence
118 R. Ince

Fig. 4.5 The same transfer

standard from Fig. 4.4
projected as a
two-dimensional SEM image
at 20.0 kV 10 μm × 10 μm
[58]

4.6.3.1 Main Measurement Uncertainty Contributions

Sample vibration, Proper instrument calibration, modelling of SEM image, choice

of operating parameters/understanding trade-offs, Beam-induced specimen con-
tamination [47] (Fig. 4.5).

4.7 Other Methods for Dimensional Nanometrology

The metrological systems described above must be further refined since they cannot
be used in deep holes or for sub-nm precision. For example, the feasibility of
utilizing CNT as probe tips is under investigation, since their high aspect ratio will
reproduce deep trenches, whilst maintaining high precision due to their nanoscale
diameters. Diamondoid tips are necessary both for sub-nm precision and for
avoiding wear which occurs on silicon tips [24]. Furthermore, there is the unique
challenge of gaining structural information on large-scale nanosystems with random
density inhomogeneities such as nanoparticles in suspension and dilute biological
samples; ensemble methods are required for this.
Thus the nanostructures in Tables 4.2 and 4.3 can be utilized either as parts of
nano-characterisation instruments or as reference materials to validate their per-
formances. Additionally the repeatability and uncertainties of the instruments must
be decreased. The following are alternate novel techniques that are currently under
investigation for future nanometrology:
4 The Challenge to Develop Metrology at the Nanoscale 119

4.7.1 X-ray Interferometry (XRI)

This is currently being developed in the European Metrology Research Programme

(EMRP) as a technique for disseminating traceable measurements of sub-nanometre
step heights. Silicon reference samples have mono-atomic step terraces arranged in
an amphitheatre-like structure.
X-ray and visible light interferometry can be combined for better precision. In
visible light subdivision of an optical fringe may give rise to errors that are sig-
nificant at the sub-nanometre level. The X-ray interferometer is used to subdivide
the optical fringes. Traceability to the meter is achieved via traceable calibrations of
the lattice parameter of silicon and of the laser frequency. Each X-ray fringe cor-
responds to a displacement equal to the lattice parameter of silicon (*0.19 nm) for
the (220) lattice planes.
The X-ray interferometer’s translation stage must be moved in steps of discrete
numbers of X-ray fringes, so servo positioning capability at the picometer level has
been developed and includes quadrature counting of X-ray fringes with a resolution
of 24 pm [48] (Fig. 4.6).
X-rays from a Mo Kα source are split by the first crystal then recombined. When
the analyser is moved in the direction orthogonal to the diffracting planes, the
periodic variation in the transmitted and diffracted X-ray intensities (caused by the
diffracting-plane spacing) is observed. The analyser displacement δx is measured by

δx X-ray fringes
X-ray source
192 pm
C1 C2 C3
Mo Kα

λ/4
Optical fringes
M
316 nm
PBS
P

PM

He-Ne laser

Fig. 4.6 C1, C2, C3 Si crystals, M = mirror, PBS polarising beam splitter, PM phase modulator,
P = polariser [59]
120 R. Ince

the optical interferometer. Polarization encoding and phase modulation enable

picometre and nanoradian resolutions to be achieved. Uncertainties at 95 % con-
fidence level are ±30 pm [49].

4.7.2 Small Angle X-ray Scattering Diffractometer (SAXS)

This is deflection of collimated radiation away from a straight trajectory after

interacting with structures much larger than the wavelength of the radiation; these
deflections are small (0.1–10°). It has unique potential for analysing inner structures
of disordered systems and gains structural information on large-scales systems
(1 nm–hundreds of nm) with random density inhomogeneities. For nanoparticles in
suspension and dilute biological samples, this ensemble method is essential for
dimensional metrology.
It is an indirect method of obtaining dimensional information of sample structure
and requiring intense, monochromatic X-rays of low divergence. This is because
biological molecules and nanostructures normally scatter weakly and the measured
solutions are dilute. Such directional, high intensity light can only be acquired from
synchrotron radiation beam lines of typical size (100 × 100 μm) to provide detailed
data for 10–500 nm structures placed with high uniformity in pitch.
By selection of suitable monochromator, X-ray wavelengths from several
nanometres down to 0.1 nm at very low angles (typically 0.1–10°) can be utilised to
determine dimensional measurements of samples from a plot of the scattered
electron ıntensity (I) against momentum transfer(s) at very low angles. The particle
diameter and the size distribution are obtained from the scattering pattern as well as
structural information: shape and size of molecules, characteristic distances of
partially ordered materials, pore sizes, and other data [50, 51].
The solvent intensity is subtracted to obtain a scattering curve representing
infinite dilution. Thus, the isotropic intensity distribution remaining is due to ran-
dom positions and orientations of particles which, for monodisperse non-interacting
particles depend on the scattering from a single particle averaged over all orien-
tations (Fig. 4.7).
As for SE, analysis is based on complex fitting to multi-parameter geometric
models; for example, a diffraction grating is modelled as a stack of trapezoids, so
uniqueness of fit is not guaranteed since it cannot represent the complexity of the
sample. A new paradigm is required to elevate scattering and diffraction mea-
surements to quantitative dimensional metrology; this can be achieved by devel-
oping a materials model. The uniqueness of the sample structure and shape would
then depend on physical parameters (thermodynamics or statistical mechanics) of
the model, not the fitting procedure [52].
4 The Challenge to Develop Metrology at the Nanoscale 121

Fig. 4.7 Schematic of a small angle X-ray scattering diffractometer for analysis of dilute,
biological and nanostructured samples

4.7.3 Electron/X-ray Diffraction

Suitable for transparent or opaque structures and offers non-destructive, fast col-
lection of data. High-resolution diffraction can provide dimensional information
with a relatively simple scan; however, the data have inherent complex error
function [53]. Parallel beam diffractometers (traceable to XRI) have self-calibrating
goniometers (stable, high-resolution encoders that have been calibrated using the
circle closure method) incorporated within them.
To ensure traceability to System Internationale (SI) based units, certification
measurements must be performed on this type of goniometer. This can be achieved
via Standard Reference Materials (SRM’s).
It has been developed to an optical ruler imaging system based on
cross-correlation of a feature-rich diffraction pattern (a robust ruler) and silicon
sample wafers. The diffraction pattern is projected onto the sample. This produces
metrology with a spatial precision of 17.2 nm across silicon wafers [54].

4.7.4 Raman Spectroscopy as a Nanometrology Tool

This is a non-linear optical spectroscopy technique that is capable of providing

dimensional information at the nanoscale.
Two-dimensional band features of graphene are highly reproducible and, toge-
ther with the G-peak position, can be used to count the number of graphene layers.
Raman features remain the same when you place graphene on arbitrary substrate
[55, 56] (Fig. 4.8).
122 R. Ince

Fig. 4.8 Raman spectra of (a) 15000

graphene at 300 K, excitation Experimental result
wavelength = 488 nm Fitted result
a shows broadening and blue Lorentzian peaks
shift b shows highly 5 layers
reproducible 2D-band

Intensity (arbitrary units)

features, and together with the 10000
G peak position can be used
4 layers
to count the number of
graphene layers
3 layers

5000

2 layers

1 layer
0
2600 2650 2700 2750 2800
Raman shift [cm-1)

(b) 24000

20000
5 layers
Intensity (arbitrary units)

16000
4 layers
12000
3 layers
8000

2 layers
4000
1 layer
0
2300 2400 2500 2600 2700 2800 2900 3000
Raman shift (cm-1)

4.8 The Challenge for Nanometrology

The challenge is to develop globally accepted standards and measurement tech-

niques which are reliable, stable, precise and accurate. This will involve the
development of innovative systems and standards to meet the needs of the
next-generation of manufacturing. These new systems will involve imaging,
modelling and manipulating matter at the nanometre length scale.
Various techniques have been discussed above; recently a few standards have
been published for terminology, characterization of nanomaterials and nanoparticles
4 The Challenge to Develop Metrology at the Nanoscale 123

based on some physical phenomena observed on particle interactions or forces at

nanoscale using SPM measurement tools (AFM, SEM, Interferometers) [57].
To meet the requirements of future manufacturing and healthcare industries
suitable reference materials must be developed and standardised sample preparation
methods must be developed. Dimensional metrology must be correlated with
functionality by the creation of an international database. Monte Carlo modelling
software can be utilised to generate a library of waveforms (for microscopies and
spectroscopies) for three-dimensional structures with shape parameters (e.g. widths,
angles, radii of curvature) spanning a range of values around that expected together
with functional data. The task is then to identify three-dimensional shapes with
modelled images that fit the one acquired alongside expected functional behaviour.
As described in Tables 4.4 and 4.5 present second generation nanostructures are
more complex than the first generation passive nanotechnology products. They are
now functional structures requiring more complex measurement tasks to be per-
formed: shape, size distribution, characteristic distances of partially ordered mate-
rials, pore sizes, and other data. The need for more accurate, precise and fast
ensemble techniques will only grow as nanotechnology stretches into the third and
fourth generation: integrated nanosystems.
Traceability is one of the fundamental requirements for manufacturing and
assembly of products, but it is not possible to ensure traceability at nanoscale yet.
There are suitable calibration artefacts for achieving traceability, but the measure-
ment equipment required to ensure traceability has not been developed. National
reference standards are not always available and uncertainty estimation models
have yet to be developed.

4.9 Conclusion

Sub-division of optical laser wavelengths is naturally limited, so research must

pursue the realisation of the metre through the lattice parameter of silicon or another
suitable crystalline material. Reference samples will be arranged as mono-atomic
steps on silicon crystal surfaces or two-dimensional and three-dimensional pyra-
mids and cubes to calibrate nanotechnology instruments. XRI must be developed to
replace visible light interferometry. This will involve improving its hardware, optics
and data processing and provide a route to traceability to the lattice constant ‘a’ as
the metre realisation.
High resolution microscopes and integral non-imaging methods (ellipsometry,
scatterometry) are widely used for R&D purposes. Many NMI’s have developed
traceable instrumentation systems primarily for AFM, but only partly for SEM and
optical instrumentation due to lack of understanding of the probe-sample interac-
tion. This can be improved by advanced modelling techniques, such as Molecular
Dynamic Modelling (MDC), Monte-Carlo-Simulations.
Nanomaterials can be utilised for the validation of the measurement performance
of Nano-characterisation instruments. However, suitable reference materials and
124 R. Ince

methods must be developed for this purpose. Bridging the gap between well
characterised reference nanomaterials and manufactured nanomaterials embedded
in complex matrix materials is a challenge. There is a great variety of different
nanomaterials, so a suitable categorisation approach is required to limit the char-
acterisation efforts to a meaningful manageable amount. The Tables 4.1, 4.2 and 4.3
describe thirteen different main materials and fifteen nano architectures in common
use, these are a platform for such reference materials.
The success of nanoelectronics is a precondition for any technological innova-
tion; expected future key challenges for the metrology and standards of to be
developed are well described in the ITRS roadmap document. Next generation
nanoelectronics technology will develop into three-dimensional complex sub-nm
architectures such as FinFET’s, and nanometrology is currently being driven there.
None of the existing techniques is able to meet the requirements for reliable
metrology of nanomaterials. To meet these requirements suitable reference mate-
rials must be developed and standardised sample preparation methods must be
developed. Dimensional metrology must be correlated with analytical metrology
and functionality, possibly by the creation of an international database.
Innovative techniques such as non-linear optical microscopy and spectroscopy,
X-ray scattering and FRET have to be developed further and intercompared with
presently employed nanometrology methods such as AFM and SEM to validate
accuracy. These innovative techniques have greater capability than present systems
for sensing the more complex structures and samples that are of keen interest to the
healthcare industry. For nanometrology to move into the biological environment
these techniques must be developed.
In addition they have great potential of realising the correlation gap between
analytical and dimensional metrology. These are important goals and incorporation
of these innovative techniques into nanometrology per se will provide the health-
care and semiconductor industries with remarkable tools.

Acknowledgments I would like to thank Seda Aytekin for her assistance in finding the relevant
references.

References

1. SI Brochure www.bipm.org
2. B. Andreas et al., Determination of the Avogadro constant by counting the atoms in a 28Si
crystal. Phys. Rev. Lett. 106, 030801 (2011)
3. http://www.bipm.org/en/bipm/mass/avogadro
4. Yeditepe University dimensional calibration services (YUKAL) handbook (Istanbul, Turkey,
2010)
5. E.L. Wolf, Nanophysics and nanotechnology (Wiley-VCH Weinheim, Germany, 2005),
pp. 28–37
6. Freitas R.A. Jr, Meeting the challenge of building diamondoid medical nanorobots. Int.
J. Robot. Res. 28, 548–557 (2009)
7. A.C. Clarke, The Fountains of Paradise Ballantine (Harcourt, New York, 1978)
4 The Challenge to Develop Metrology at the Nanoscale 125

8. S. Knudsen, L. Golubovic, Rotating space elevators: physics of celestial scale spinning

strings. Eur. Phys. J. Plus 129, 1–24 (2014)
9. N.M. Pugno, Towards the Artsutanov s dream of the space elevator: The ultimate design of a
35 GPa strong tether thanks to graphene. Acta Astronaut. 82, 221–224 (2013)
10. P.K. Aravind, The physics of the space elevator. Am. J. Phys. 75, 125–130 (2007)
11. R.R. Birge, Protein-based computers. Sci. Am. 272, 90–95 (1995)
12. K.E. Drexler, Engines of creation: the coming era of nanotechnology (Anchor books edition,
1986), pp. 7–12
13. K.E. Drexler, Nanosystems: Molecular Machinery Manufacturing and Computation
(Wiley-interscience, 1992), pp. 191–248
14. K.E. Drexler, Engines of Creation 2.0: The Coming Era of Nanotechnology, 20th
anniversary e-book edition, pp. 517–535 (2007)
15. R.D. Kamm, R. Bashir, Creating living cellular machines. Ann. Biomed. Eng. 42, 445–459
(2014)
16. P. Ball, Natural strategies for the molecular engineer. Nanotechnology 13, R15–R28 (2002)
17. A.A. Barhorst, O.P. Harrison, G.D. Bachand, Modelling elasto-mechanical phenomena
involved in the motor-driven assembly of nanomaterials, in ASME 19th International
Conference on Design Theory and Methodology/1st International Conference on Micro &
Nano Systems, vol. 3, pp. 669–677 (2008)
18. G.A. Ozin, I. Manners, S. Fournier-Bidoz et al., Dream nanomachines. Adv Mat 17, 3011
(2005)
19. D.P. Anderson, T. Kilburn, A pioneer of computer design. IEEE Ann. Hist. Comp. 31, 84
(2009)
20. A.C. Diebold, K. Monahan, Next-generation metrology must meet challenges. Solid State
Technol. 41, 50 (1998)
21. K. Seshan (ed.), The Handbook of Thin Film Deposition: Processes and Technologies, 6, 241
(2002)
22. M.A. Jamakhani, M.R. Jadhav, G.S. Kamble, V.R. Gambhire, Nanometrology in biological
and medical sciences. Int. J. Adv. Biotechnol. Res. 2, 213–223 (2011)
23. L. Peltonen, J. Hirvonen, Pharmaceutical nanocrystals by nanomilling: critical process
parameters particle fracturing and stabilization method. Eur. J. Pharm. Biopharm. 81, 214–
222 (2012)
24. Euromet (2014) The strategic research agenda of EURAMET
25. JRC Reference Report: Requirements on measurements for the implementation of the
European Commission definition of the term nanomaterial
26. J.A. Woollam, B. Johs, C.M. Herzinger et al., Overview of variable angle spectroscopic
ellipsometry, in Critical reviews of optical science and technology: Optical Metrology CR72,
ed. by A-J. Ghanim (1999), pp. 3–28
27. R.K. Leach, C. Giusca, Results from a comparison of optical thin film thickness
measurement. NPL Report UK (2009)
28. M.P. Seah, S.J. Spencer, F. Bensebaa et al., Critical review of the current status of thickness
measurements for ultra-thin SiO2 on Si: Part V results of a CCQM pilot study. Surf.
Interfaces Anal. 36, 1269–1303 (2004)
29. B. Belzer, K. Eberhardt, D. Chandler-Horowitz et al., Thin film reference materials
development final report for CRADA CN-1364. NIST Gaithersburg 2012 (1998)
30. Euromet EMPIR Call, Traceable AFM measurement capability. Industry and Research
Potential Selected Research Topic: SRT-r 05, 1–3 (2014)
31. V. Korpelainen, J. Seppä, A. Lassila, Design and characterization of MIKES metrological
atomic force microscope. Prec. Eng. 34, 735–744 (2010)
32. S. Ducourtieux, B. Poyet, J. David, Development of a metrological atomic force microscope
with minimized Abbe error and differential interferometer based real-time position control.
Measur. Sci. Technol. 22, 094010 (2011)
33. B. Babic, C.H. Freund, J. Herrmann et al., Metrological scanning probe microscope based on
a quartz tuning fork detector. J. Micro/Nanolith MEMS MOEMS 11, 011003 (2012)
126 R. Ince

34. P. Klapetek, M. Valtr, M. Matula, A long-range scanning probe microscope for automotive
reflector optical quality inspection. Measur. Sci. Technol. 22, 094011 (2011)
35. G. Dai, F. Pohlenz, X. Min et al., Metrological large range scanning probe microscope. Rev.
Sci. Instrum. 75, 962–969 (2004)
36. I. Misumi, S. Gonda, Q. Huang et al., Sub-hundred nanometre pitch measurement using an
AFM with differential laser interferometers for designing usable lateral scales. Measur. Sci.
Technol. 16, 2080–2090 (2005)
37. J.A. Kramar, Nanometre resolution metrology with the molecular measuring machine.
Measur. Sci. Technol. 16, 2121–2128 (2005)
38. J. Haycocks, K. Jackson, Traceable calibration of transfer standards for scanning probe
microscopy. Prec. Eng. 29, 168–175 (2005)
39. V. Korpelainen, A. Lassila, Calibration of a commercial AFM: traceability for a coordinate
system. Measur. Sci. Technol. 18, 395–403 (2007)
40. F. Meli, NANO4-1D Gratings. Final report (2000)
41. J. Garnaes, K. Dirscherl, NANO5—2D Grating—Final report CCL-S4. Metrologia. Tech.
Suppl. 45, 04003 (2008)
42. L. Koenders, R. Bergmans, J. Garnaes et al., Comparison on nanometrology: nano 2—step
height. Metrologia 40, 04001 (2003)
43. L. Klapetek, P. Koenders, F. Meli, G.B. Picotto, Comparison on step height measurements in
the nano and micrometre range by scanning force microscopes. Metrologia 43, 04001 (2006)
44. G. Hamann, Positioning in sub-nanometer resolution High-Accuracy and Traceable AFM
Measurements on Semiconductors and Step Standards. Whitepaper Physik Instrumente
GmbH (2013)
45. M.T. Postek, A.E. Vladar, Modelling for accurate dimensional scanning electron microscope
metrology: then and now. Scanning Special Iss. 33, 111–125 (2011)
46. M.T. Postek, A.E. Vladar, J.S. Villarrubiar, Nanometrology solutions using an ultra-high
resolution in-lens SEM (NIST White Paper, 2011)
47. J. Cazaux, Errors in nanometrology by SEM. Nanotechnol 15, 1195–1199 (2004)
48. A. Yacoot, U. Kuetgens, Sub-atomic dimensional metrology: developments in the control of
x-ray interferometers. Measur. Sci. Technol. 23, 074003 (2012)
49. G. Basile, P. Becker, A. Bergamin et al., Combined optical and X-ray interferometry for
high-precision dimensional metrology. Proc. Royal Soc. A-Math. Phys. Eng. Sci. 456, 701–
729 (2000)
50. L. Koenders, Defining a European Strategy for Nanometrology- Dimensional and
Mechanical Aspects (Co-nanomet Physikalisch-Technische Bundesanstalt, Budapest, 2011)
51. M. Krumrey, G. Gleber, F. Scholze, J. Wernecke, Synchrotron radiation-based x-ray
reflection and scattering techniques for dimensional nanometrology. Measur. Sci. Technol.
22, 094032 (2011)
52. NIST, Dimensional metrology for nanofabrication. Materials Science and Engineering
Division White Paper (2009)
53. J.P. Cline, D.A. Windover, Materials structure characterization. Diffraction Metrology and
Standards NIST White Paper (2014)
54. N. Yoshimizu, A. Lal, C.R. Pollock, Nanometrology optical ruler imaging system using
diffraction from a quasiperiodic structure. Opt. Express 18, 20827–20838 (2010)
55. I. Calizo, S. Ghosh, D. Teweldebrhan et al., Raman nanometrology of graphene on arbitrary
substrates and at variable temperature. Proc. SPIE 7037, 70371B (2008)
56. A.A. Balandin, Thermal transport in graphene and graphene multi-layers. Seminar Report
Nano-Device Laboratory, Univ California-Riverside (2008)
57. A. Vladár, PML Scientists Develop 3D SEM Metrology for 10 nm Structures.
Semiconductor and dimensional division NIST White Paper (2014)
58. P. Todua, Nanometrology nanoscale reference materials for high-resolution microscopy.
national scientific metrology center/center for surface and vacuum research (CSVR) Moscow
Institute of Physics and Technology (MIPT). Vienna, Austria (2011)
4 The Challenge to Develop Metrology at the Nanoscale 127

59. P. Becker, M. Jentschel, G. Mana, G. Zosi, Towards an atomic realisation of the kilogram:
the measurements of NA and NAh, in Metrology and Fundamental Constants, ed. by T.
Haensch, S. Leschiutta, A. Wallard (IOS Press, Italy, 2007), pp. 519–530
60. C. He, S. Wu, N. Zhao et al., Carbon-encapsulated Fe3O4 nanoparticles as a high-rate lithium
ion battery anode material. ACS Nano 7, 4459–4469 (2013)
61. H.W. Kroto, J.R. Heath, S.C. OBrien, R.F. Curl, R.E. Smalley, C60: Buckminsterfullerene.
Nature 318, 162–163 (1985)
62. H. Isobe, S. Hitosugi, T. Yamasaki et al., Molecular bearings of finite carbon nanotubes and
fullerenes in ensemble rolling motion. Chem. Sci. 4, 1293–1297 (2013)
63. S. Iijima, T. Ichihashi, Single-shell carbon nanotubes of 1-nm diameter. Nature 363, 603–605
(1993)
64. M.M. Shulaker, G. Hills, N. Patil et al., Carbon nanotube computer. Nature 501, 526 (2013)
65. A.K. Geim, K.S. Novoselov, The rise of graphene. Nat. Mater. 6, 183–191 (2007)
66. T. Georgiou, R. Jalil, B.D. Belle et al., Vertical field-effect transistor based on graphene-WS2
heterostructures for flexible and transparent electronics. Nat. Nanotechnol. 8, 100–103
(2013)
67. R.C. Merkle, R.A. Jr Freitas, Theoretical analysis of a carbon-carbon dimer placement tool
for diamond mechanosynthesis. J. Nanosci. Nanotechnol. 3, 319–324 (2003)
68. D.G. Allis, K.E. Drexler, Design and analysis of a molecular tool for carbon transfer in
mechanosynthesis. J. Comp. Theor. Nanosci. 2, 45–55 (2005)
69. A. Yamamoto, T. Hamada, I. Hisaki et al., Dynamically deformable cube-like
hydrogen-bonding networks in water-responsive diamondoid porous organic salts. Angew.
Chem. Int. Ed. 52, 1709–1712 (2013)
70. D. Wang, H.L. Xin, R. Hovden et al., Structurally ordered intermetallic platinum-cobalt
core-shell nanoparticles with enhanced activity and stability as oxygen reduction
electrocatalysts. Nat. Mater. 12, 81–87 (2013)
71. H. Li, D.X. Chen, Y.L. Sun et al., Viologen-mediated assembly of and sensing with
carboxylatopillar[5]arene-modified gold nanoparticles. J. Am. Chem. Soc. 135, 1570–1576
(2013)
72. J. Liu, W. Bu, L. Pan et al., NIR-triggered anticancer drug delivery by upconverting
nanoparticles with integrated azobenzene-modified mesoporous silica. Angew. Chem. Int.
Ed. 52, 4375–4379 (2013)
73. L. Wang, I. Meric, P.I. Huang et al., One-dimensional electrical contact to a two-dimensional
material. Science 342, 614–617 (2013)
74. P.J. Kueffer, C.A. Maitz, A.A. Khan et al., Boron neutron capture therapy demonstrated in
mice bearing EMT6 tumors following selective delivery of boron by rationally designed
liposomes. Proc. Nat. Acad. Sci. USA 110, 6512–6517 (2013)
75. H. Wu, G. Yu, K. Pan et al., Stable Li-ion battery anodes by in-situ polymerization of
conducting hydrogel to conformally coat silicon nanoparticles. Nat. Commun. 4, 1943 (2013)
76. Y.H. Fu, A.I. Kuznetsov, A.E. Miroshnichenko et al., Directional visible light scattering by
silicon nanoparticles. Nat. Commun. 4, 1527 (2013)
77. H. Choi, S.J. Ko, Y. Choi et al., Versatile surface plasmon resonance of
carbon-dot-supported silver nanoparticles in polymer optoelectronic devices. Nat.
Photonics 7, 732–738 (2013)
78. C. Wang, H. Xu, C. Liang et al., Iron oxide @ polypyrrole nanoparticles as a multifunctional
drug carrier for remotely controlled cancer therapy with synergistic antitumor effect. ACS
Nano 7, 6782–6795 (2013)
79. G. Chen, Y. Zhao, G. Fu et al., Interfacial effects in iron-nickel hydroxide-platinum
nanoparticles enhance catalytic oxidation. Science 344, 495–499 (2014)
80. Y. Hu, J.O. Jensen, W. Zhang et al., Hollow spheres of iron carbide nanoparticles encased in
graphitic layers as oxygen reduction catalysts. Angew. Chem. Int. 53, 3675–3679 (2014)
81. B.J. Sanghavi, A.K. Srivastava, Adsorptive stripping voltammetric determination of
imipramine trimipramine and desipramine employing titanium dioxide nanoparticles and
128 R. Ince

an Amberlite XAD-2 modified glassy carbon paste electrode. Analyst 138, 1395–1404
(2013)
82. S. Sarina, H. Zhu, E. Jaatinen et al., Enhancing catalytic performance of palladium in gold
and palladium alloy nanoparticles for organic synthesis reactions through visible light
irradiation at ambient temperatures. J. Am. Chem. Soc. 135, 5793–5801 (2013)
83. U. Rocha, C. Jacinto, S.W. Ferreira et al., Subtissue thermal sensing based on
neodymium-doped LaF3 nanoparticles. ACS Nano 7, 1188–1199 (2013)
84. J. Burschka, N. Pellet, S.J. Moon et al., Sequential deposition as a route to high-performance
perovskite-sensitized solar cells. Nature 499, 316 (2013)
85. X. Zhu, X. Niu, H. Zhao et al., Doping ionic liquid into Prussian blue-multiwalled carbon
nanotubes modified screen-printed electrode to enhance the nonenzymatic H2O2 sensing
performance. Sens. Act. B 195, 274–280 (2014)
86. C.C. Druck, J.L. Pozzobon, G.L. Callegari et al., Adhesion to Y-TZP ceramic: study of silica
nanofilm coating on the surface of Y-TZP. J. Biomed. Mat. Res. B-App. Biomater. 103, 143–
150 (2015)
87. J.H. Heo, S.H. Im, J.H. Noh et al., Efficient inorganic-organic hybrid heterojunction solar
cells containing perovskite compound and polymeric hole conductors. Nat Photon 7, 487–
492 (2013)
88. T. Stephenson, Z. Li, B. Olsen et al., Lithium ion battery applications of molybdenum
disulfide (MoS2) nanocomposites. Energy Environ. Sci. 7, 209–231 (2014)
89. T. Choi, S.H. Kim, C.W. Lee et al., Synthesis of carbon nanotube-nickel nanocomposites
using atomic layer deposition for high-performance non-enzymatic glucose sensing. Biosens.
Bioelectron. 63, 325–330 (2015)
90. F. Song, X. Li, Q. Wang et al., Nanocomposite hydrogels and their applications in drug
delivery and tissue engineering. J. Biomed. Nanotechnol. 11, 40–52 (2015)
91. J. Wallentin, N. Anttu, D. Asoli et al., InP nanowire array solar cells achieving 13.8 %
efficiency by exceeding the ray optics limit. Science 339, 1057–1060 (2013)
92. P. Yang, Y. Ding, Z. Lin et al., Low-cost high-performance solid-state asymmetric
supercapacitors based on MnO2 nanowires and Fe2O3 nanotubes. Nano Lett. 14, 731–736
(2014)
93. L. Fu, W. Cai, A. Wang et al., Photocatalytic hydrogenation of nitrobenzene to aniline over
tungsten oxide-silver nanowires. Mat. Lett. 142, 201–203 (2015)
94. X. Zhou, L.J. Wan, Y.G. Guo, Binding SnO2 nanocrystals in nitrogen-doped graphene sheets
as anode materials for lithium-ion batteries. Adv. Mater. 25, 2152–2157 (2013)
95. Z. Bian, Zhang T.P. Tachikawa et al., Au/TiO2 superstructure-based plasmonic
photocatalysts exhibiting efficient charge separation and unprecedented activity. J. Am.
Chem. Soc. 136, 458–465 (2014)
96. C.J. Hages, S. Levcenco, C.K. Miskin et al., Improved performance of Ge-alloyed
CZTGeSSe thin-film solar cells through control of elemental losses. Prog. Photovolt. 23,
376–384 (2015)
97. H.S. Kim, J.W. Lee, N. Yantara et al., High efficiency solid-state sensitized solar cell-based
on submicrometer rutile TiO2 nanorod and CH3NH3PbI3 perovskite sensitizer. Nano Lett. 13,
2412–2417 (2013)
98. J. Yu, J. Jin, B. Cheng et al., A noble metal-free reduced graphene oxide-CdS nanorod
composite for the enhanced visible-light photocatalytic reduction of CO2 to solar fuel. J. Mat.
Chem. A 2, 3407–3416 (2014)
99. S. Danwittayakul, M. Jaisai, J. Dutta, Efficient solar photocatalytic degradation of textile
wastewater using ZnO/ZTO composites. App. Cat. B-Environ. 163, 1–8 (2015)
100. L. Bareket, N. Waiskopf, D. Rand, G. Lubin, M. David-Pur et al., Semiconductor
nanorod-carbon nanotube biomimetic films for wire-free photostimulation of blind retinas.
Nano Lett. 14, 6685–6692 (2014)
101. J.Y. Jeng, Y.F. Chiang, M.H. Lee et al., CH3NH3PbI3 perovskite/fullerene
planar-heterojunction hybrid solar cells. Adv. Mater. 25, 3727–3732 (2013)
4 The Challenge to Develop Metrology at the Nanoscale 129

102. Y. Liu, J. Zhao, Z. Li et al., Aggregation and morphology control enables multiple cases of
high-efficiency polymer solar cells. Nat. Commun. 5, 5293 (2014)
103. Y. Lin, J. Wang, Z.G. Zhang et al., An electron acceptor challenging fullerenes for efficient
polymer solar cells. Adv. Mater. 27, 1170–1174 (2015)
104. T. Choi, S.H. Kim, C.W. Lee et al., Synthesis of carbon nanotube-nickel nanocomposites
using atomic layer deposition for high-performance non-enzymatic glucose sensing. Biosens.
Bioelectron. 63, 325–330 (2015)
105. O. Chen, J. Zhao, V.P. Chauhan et al., Compact high-quality CdSe-CdS core-shell
nanocrystals with narrow emission linewidths and suppressed blinking. Nat. Mater. 12, 445–
451 (2013)
106. C. Clavero, Plasmon-induced hot-electron generation at nanoparticle/metal-oxide interfaces
for photovoltaic and photocatalytic devices. Nat. Photon. 8, 95–103 (2014)
107. S.G. Kumar, K.S.R. Koteswara, Zinc oxide based photocatalysis: tailoring surface-bulk
structure and related interfacial charge carrier dynamics for better environmental
applications. RSC Adv. 5, 3306–3351 (2015)
108. S. Agarwal, A. Greiner, J.H. Wendorff, Functional materials by electrospinning of polymers.
Prog. Polym. Sci. 38, 963–991 (2013)
109. C. Zhu, X. Mu, P.A. van Aken et al., Single-layered ultrasmall nanoplates of MoS2
embedded in carbon nanofibers with excellent electrochemical performance for lithium and
sodium storage. Angew. Chem. Int. Ed. 53, 2152–2156 (2014)
110. C. Liu, L. Wang, Y. Tang et al., Vertical single or few-layer MoS2 nanosheets rooting into
TiO2 nanofibers for highly efficient photocatalytic hydrogen evolution. App. Cat. B-Environ.
164, 1–9 (2015)
111. S.I. Kim, J.S. Lee, H.J. Ahn et al., Facile route to an efficient NiO supercapacitor with a
three-dimensional nanonetwork morphology. ACS Appl. Mater. Interfaces 5, 1596–1603
(2013)
112. R.K. Joshi, P. Carbone, F.C. Wang et al., Precise and ultrafast molecular sieving through
graphene oxide membranes. Science 343, 752–754 (2014)
113. H. Zhang, X. Hou, Z. Yang et al., Bio-inspired smart single asymmetric hourglass
nanochannels for continuous shape and ion transport control. Small 11, 786–791 (2015)
114. M.H. Oh, T. Yu, S.H. Yu et al., Galvanic replacement reactions in metal oxide nanocrystals.
Science 340, 964–968 (2013)
115. S. Dasgupta, T. Auth, G. Gompper, Shape and orientation matter for the cellular uptake of
nonspherical particles. Nano Lett. 14, 687–693 (2014)
116. J. Min, F. Wang, Y. Cai et al., Azeotropic distillation assisted fabrication of silver nanocages
and their catalytic property for reduction of 4-nitrophenol. Chem. Commun. 51, 761–764
(2015)
117. Z. Yu, B. Duong, D. Abbitt et al., Highly ordered MnO2 nanopillars for enhanced
supercapacitor performance. Adv. Mater. 25, 3302–3306 (2013)
118. D.D. Wang, G.Z. Xing, F. Yan et al., Ferromagnetic (Mn N)-codoped ZnO nanopillars array:
experimental and computational insights. Appl. Phys. Lett. 104, 022412 (2014)
119. D.M. Drotlef, E. Appel, H. Peisker et al., Morphological studies of the toe pads of the rock
frog Staurois parvus (family: Ranidae) and their relevance to the development of new
biomimetically inspired reversible adhesives. Int. Focus 5, 0036 (2015)
120. K. Wu, T. Rindzevicius, M.S. Schmidt et al., Wafer-scale leaning silver nanopillars for
molecular detection at ultra-low concentrations. J. Phys. Chem. C 119, 2053–2062 (2015)
121. W. Liu, Y.Q. Feng, X.B. Yan et al., Superior micro-supercapacitors based on graphene
quantum dots. Adv. Funct. Mater. 23, 4111–4122 (2013)
122. X. Liu, Z. Chang, L. Luo et al., Hierarchical ZnxCo3-xO4 Nanoarrays with High Activity for
Electrocatalytic Oxygen Evolution. Chem of Mat 26, 1889–1895 (2014)
123. M. Yu, J. Chen, J. Liu et al., Mesoporous NiCo2O4 nanoneedles grown on 3D
graphene-nickel foam for supercapacitor and methanol electro-oxidation. Electrochim.
Acta 151, 99–108 (2015)
130 R. Ince

124. X. Wang, L. Jiao, K. Sheng et al., Solution-processable graphene nanomeshes with

controlled pore structures. Sci. Rep. 3, 1996 (2013)
125. C.F. Guo, T. Sun, Q. Liu et al., Highly stretchable and transparent nanomesh electrodes made
by grain boundary lithography. Nat. Commun. 5, 3121 (2014)
126. B. Piccione, D.S. Gianola, Tunable thermoelectric transport in nanomeshes via elastic strain
engineering. Appl. Phys. Lett. 106, 113101 (2015)
127. S. Costeux, L. Zhu, Low density thermoplastic nanofoams nucleated by nanoparticles.
Polymer 54, 2785–2795 (2013)
128. A. Wittstock, M. Baeumer, Catalysis by unsupported skeletal gold catalysts. Acc. Chem.
Res. 47, 731–739 (2014)
129. S. Sen, D. Liu, G.T.R. Palmore, Electrochemical reduction of CO2 at copper nanofoams.
ACS Catalysis 4, 3091–3095 (2014)
130. S. Costeux, I. Khan, S.P. Bunker et al., Experimental study and modelling of nanofoams
formation from single phase acrylic copolymers. J. Cell. Plastic. 51, 197–221 (2015)
131. C. Anders, E.M. Bringa, H.M. Urbassek, Sputtering of a metal nanofoam by Au ions. Nucl.
Instrum. Methods. Phys. Res. Sect. B-Beam Interact. Mater. Atom 342, 234–239 (2015)
132. R. Köning, J. Flügge, D. Hüser et al., Dimensional micro- and nanometrology at PTB, in
2011 Proceedings of 8th International Conference on Measurement, Smolenice, Slovakia
95:106 (2011)
133. L. Koenders, EUROMET project 707 step height standards. PTB Final Report (2005)
134. V. Korpelainen, J. Seppä, A. Lassila, Measurement strategies and uncertainty estimations for
pitch and step height calibrations by metrological AFM. Proc. SPIE 8036, 80360Q1 (2011)
135. J.B. Warren, B.J. Panessa-Warren, A comparison of nanometrology using TEM and SEM.
Microsc. Microanal. 11(2), 1932–1933 (2005)
136. T. Ikegami, A. Yamaguchi, M. Tanaka et al., Evolution and future of critical dimension
measurement system for semiconductor processes. Hitachi high Technologies Inc White
paper 203–209 (2011)
137. Nanosensors, Transfer Standards for Scanning Probe Microscopy (Nano World AG Data
Sheet, 2013)
Chapter 5
Terahertz Devices and Systems
for the Spectroscopic Analysis
of Biomolecules—“Complexity Great
and Small”

Anthony John Vickers and David Crompton

Abstract The world has an ever-growing need to understand the complexity of

biomolecules from fields as far ranging as drug design Hajduk and Geer, Nat Rev 6
(3):211, 2007, [1] and Congreve et al., Prog Med Chem 53(1), 2014, [2], crop
characterization Ge et al., 15(6):12560, 2015, [3] and organic electronics Gao et al.,
Phys Rev Lett 114(12):128701, 2015, [4]. There are many scientific techniques for
studying bio-molecules namely mass spectroscopy (MS), nuclear magnetic reso-
nance (NMR), droplet single biomolecule studies, spectroscopy, biocalorimetry,
chromatography, crystallography, electrophoresis, and bioinformatics. In this
chapter the focus will be on the study of biomolecules using the relatively new
technique of terahertz spectroscopy. This chapter will draw on the THz spec-
troscopy work of the authors and others to demonstrate the need to understand both
the biomolecule and its water environment, which provide both great (biomolecule)
and small (water) complexity.

5.1 Introduction

Terahertz spectroscopy has developed over the past two decades and has started to
make an impact on the understanding of biomolecules. One of the key issues
regarding the study of biomolecules is to study them in their natural aqueous
environment as this offers the greatest possibility of understanding their natural
behavior. This provides both a challenge and an opportunity in THz spectroscopy
as water strongly absorbs in the THz frequency range [5]. This strong water THz
absorption could be seen as a disadvantage making observation of the biomolecule
absorption difficult if not impossible. However, as we shall discuss in this chapter, it
can be argued that the modeling and study of biomolecules cannot be undertaken in

A.J. Vickers (&)
D. Crompton

School of Computer Science and Electronic Engineering, University of Essex,
Wivenhoe Park, Colchester, Essex CO4 3SQ, UK
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 131

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_5
132 A.J. Vickers and D. Crompton

isolation from its aqueous environment. Taking the proposition to another level it is
perhaps possible to argue that biomolecules have some water molecules so strongly
associated with them that they are truly a part of the protein. This is the basis of a
new drug design software package, WaterMap [6].

5.2 THz Time Domain Spectroscopy

Terahertz time domain spectroscopy (THz-TDS) describes a technique that has

become widely adopted in the field of THz spectroscopy. In THz-TDS, picosecond
THz pulses are generated from a photoconductive antenna pumped by a femtosec-
ond laser. High frequency sampling of the generated THz pulse is achieved by
delaying a probe laser, in time, to sample the THz EM field. By changing the time
interval of the probe laser, the THz EM field is sampled at each time interval until a
complete THz pulse has been sampled. The resulting THz pulse contains all the
frequency information of the THz beam in the time domain, which can be extracted
by a Fourier transform algorithm. A typical frequency range for photoconductive
antennas is 0.1–4 THz. Detection of the THz pulse is achieved using techniques such
as photodetectors or electro-optic sampling. In the case of a THz photodetector the
THz pulse creates an electric field across a pair of electrodes deposited on the
photodetector material, for example GaAs. The THz pulse biases the detector, a
signal being generated by the delayed optical femtosecond pulse. The signal, for a
given time delay, is then averaged using a lock-in amplifier. Modulation of the
lock-in amplifier is either achieved through the modulation of the THz emitter bias or
through the optical modulation of the femtosecond laser pulse train. THz photode-
tectors have characteristic resonant frequencies that effect the efficiency as a function
of frequency. Electro-optic (EO) sampling is a technique often used in place of the
photoconductive detector in THz-TDS spectroscopy. Unlike the photoconductive
antennas, EO sampling does not have resonant frequencies, so has the same effi-
ciency in THz detection across the frequency range. The principle of the EO tech-
nique is based on an EO medium, often constructed from a ZnTe crystal, which
introduces birefringence when under the influence of an external THz EM field.
Typically for a THz-TDS system, the EO medium induces an elliptical birefringence
to the probe laser beam, dependent on the strength of the instantaneous THz EM
field. A Wollaston prism splits the now elliptically polarized probe laser into
component p and s polarizations, which are subsequently detected by two balanced
photodiodes. The difference in p and s magnitude is then directly proportional to the
strength of the THz EM field. As with the photoconductive antenna, the probe laser
is delayed in time to detect the entire THz pulse. Without the THz EM field, the
probe laser will still undergo some equilibrium birefringence resulting in a non-zero
detection. To compensate for this effect, a quarter-wave plate is placed between the
Wollaston prism and the EO medium. The quarter-wave plate is then rotated till the
closest value of 0 V is achieved on the balanced photo detector. The p and s polar-
izations should now have equal magnitude at the detector for the null result.
5 Terahertz Devices and Systems … 133

7
6
Terahertz Electric Field 5
4
(Arb. Units)
3
2
1
0
0 10 20 30 40 50 60 70
-1
Time (ps)
-2
-3

Fig. 5.1 A typical THz pulse recorded using a time domain system

The overall system consisting of a THz emitter, pumped by a femtosecond laser,

which also provides the variable time delayed probe pulse for the THz detector, is
referred to as a THz Time Domain Spectrometer (TDS) and provides a sampling
rate beyond current sampling oscilloscope technology [7]. A THz pulse sampled
using the above method is displayed in Fig. 5.1.
The choice of THz emitter and detector for a time domain system influences the
overall design of the spectrometer. For detectors and emitters without any associ-
ated focusing element, a four parabolic mirror system is usually adopted. The first
parabolic mirror gathers the THz emission from the emitter and generates a parallel
beam. The second parabolic mirror focuses this parallel beam onto the sample cell.
A third parabolic mirror gathers the THz beam as it expands after the sample cell,
again producing a parallel THz beam. The fourth parabolic mirror then focuses the
THz beam onto the detector. An alternative design, in the case where both THz
emitter and detector have associated focusing elements, requires only two parabolic
mirrors [8]. This design is sketched in Fig. 5.2. In this case a photoconductive
emitter and detector are used. Both emitter and detector have Si lenses positioned at
the point of emission and detection. The Si lenses can be manipulated in x–y to
center the lens on the emitter and detector to achieve optimum signal generation and
detection.
In the spectrometer shown in Fig. 5.2 the sample cell is an Attentuated Total
Reflection (ATR) cell, although in this design the ATR sample cell can be replaced
by a transmission cell. Attenuated total reflectance cells work on the principal of the
evanescent wave effect observed at the point of total internal reflection. This
evanescent wave extends out in an exponentially decaying wave that interacts with
a sample.
This method then avoids direct transmission through the sample. To achieve
total internal reflection of the THz beam, a general design utilises a dove prism
placed within a THz beam path, shown in Fig. 5.2. In biological systems, this is a
key advantage over transmission systems, which require short sample path lengths.
The evanescent tail is also highly sensitive to changes in sample concentrations
making ATR ideal for complex biological molecules that are difficult to purify in
134 A.J. Vickers and D. Crompton

Lock in
Amplifier

Fig. 5.2 A typical THz time domain spectrometer showing a photoconductive emitter and
detector, and an Attenuated Total Reflection (ATR) sample stage [8]

high concentrations. Techniques like high throughput protein-drug candidate

screening require accurate results with minimum protein sample usage. The con-
centrations required for protein-drug binding in THz transmittance systems are too
high to make this a viable option. This is one example where the sensitivity of the
ATR unit could be a dominating feature in its adoption in the THz frequency range.

5.3 Small Complexity—Simple Polar Liquids

One of the key areas of interest for THz spectroscopy is its ability to probe the low
frequency dynamics of water and other polar liquids. These dynamics are linked to
the continuous breaking and reformation of the hydrogen bonds that universally
bind polar liquids together across a hydrogen bond network. As water is a solvent
present in almost all biological processes, this network will almost certainly interact
with these processes through its dynamics. In the following section we explore the
composition of the hydrogen bond dynamics and how these dynamics underpin the
low frequencies in the THz regime.
ATR THz studies of simple polar liquids have been undertaken by a number of
groups including ourselves [9, 10]. We prepared an aqueous solution series (0, 10,
20 …,100 %) of propan-2-ol. In turn, a drop of each solution placed on the ATR
sample cell, creating a sample with a depth of approximately 0.5 mm. The average
5 Terahertz Devices and Systems … 135

2000
Blank

Transmission Spectrum
1800
1600 IPA 100%
1400
IPA 90%
1200
1000 IPA 80%
800
IPA 70%
600
400 IPA 60%
200
IPA 50%
0
0.05 0.55 1.05 IPA 40%
Frequency (THz)
IPA 30%

Fig. 5.3 The frequency spectrum of varying IPA concentrations was calculated from the FFT of
THz pulse response. A high detection indicates less THz radiation has been absorbed by the
sample as more radiation has reached the detector. IPA has significantly less THz absorbance
compared to water with the max peak showing a linear response to the addition of IPA. The
standard error is approximately 8 units

spectra of each solution was calculated from three consecutive scans with 10
averages per time step. The peltier cooler was set to maintain the sample cell
temperature at 21 °C, and the spectrometer was dry nitrogen purged to prevent
water absorption by the THz beam on its passage through the spectrometer.
Figure 5.3 shows the THz frequency spectrum for the propan-2-ol concentration
series. The frequency spectrum, at each concentration, is defined by singular large,
broad peak. Broad peaks are a ubiquitous feature in THz spectroscopy and are seen
in THz studies using aqueous solutions [11]. The broad peaks are caused by
molecular vibrations associated with THz absorption not occupying discrete energy
levels and instead harmonically fluctuate at frequencies across the THz regime. It is
therefore difficult to assign specific vibrational modes to frequencies as many
vibrational modes can overlap in the same frequency space.
The large disparity between water and propan-2-ol THz absorption is best
explained by the physical properties of the molecules that cause THz absorbance.
Molecules have two key characteristics that absorb THz radiation. The first is the
collective motions of large molecules, typically manifested in the torsional motion
of hundreds of atoms along a molecule’s backbone. Propan-2-ol is a relatively small
molecule and so these collective motions are unlikely to contribute to
propan-2-ol THz absorption. The second characteristic that leads to THz absorption
is the existence of an extended dynamical hydrogen bond network. Propan-2-ol has
a functional hydroxyl group located at the center of the carbon chain that is able to
form hydrogen bonds. So intuitively propan-2-ol must be able to form a hydrogen
bond network that absorbs THz radiation. The lack of THz absorbance then indi-
cates these hydrogen bonds are significantly different to those found in water.
The geometry of water allows the acceptance and donation of up to four
hydrogen bonds freely arrange from any angle on the molecule, whereas
propan-2-ol has a carbon chain that is likely to sterically hinder the formation of
136 A.J. Vickers and D. Crompton

Table 5.1 Recalculation of atomic partial charges

Propan-2-ol molecule constituents Atomic point charges
Jorgensen charges New propan-2-ol charges
q(O) −0.683 −0.620
q(HO) 0.418 0.388
q(C) −0.18 −0.456
q(C) −0.12 −0.444
q(CO) 0.145 0.148
q(H) 0.06 0.141

hydrogen bonds. Propan-2-ol is also a moderately larger molecule compared to

water, creating a lower hydrogen bond density per volume of solution. It can be
concluded the observed propan-2-ol absorbed THz spectrum is caused by the lack
of hydrogen bond density and the relative small size of the molecule.
To help clarify the underlying hydrogen bond dynamics of the polar liquid
concentration series, molecular dynamics are used to view the hydrogen bond
formation events associated with the THz frequency range. This is followed by an
analysis of the hydrogen bond network to determine the dynamical effects that are
not seen solely in the THz-TDS spectrum.
For this study, the OPLS-All Atom force field [12] was selected for its accuracy
in liquid simulation dynamics. Water was modeled by the TIP4P water model as it
works well with OPLS-All Atom and accurately models the tetrahedral shape of the
water molecule. The propan-2-ol topology parameters were based on the topology
of propan-1-ol, also built by Jorgensen et al. [13], using GAUSSIAN [14] to
calculate the charge distribution for propan-2-ol (Table 5.1).
The initial system for the simulation consisted of 1000 alcohol molecules with a
XYZ box size of 3.8 nm for ethanol and 4.3 nm for propan-2-ol. To generate the
different solution concentrations, a percentage volume of alcohol molecules were
removed at random from the system and then solvated with the TIP4P water. The
appropriate number of waters to add was selected by the GROMACS solvation
algorithm, with a van der Waals tolerance of 0.105 and a density of 1. The number
of water molecules added to each alcohol concentration is shown in Table 5.2. The
true percentage of alcohol to water molecules after solvation is also shown in the
table, which differs slightly from the alcohol volume to water ratio.
Each concentration solution was simulated for 1 ns to allow adjustment of the
total size of its system box to equilibrate the pressure. This had the effect of slightly
increasing the box size at higher concentrations of alcohol due to their lower density
than water. From the production runs the hydrogen bond auto correlation, the total
number of hydrogen bonds and hydrogen bond length were taken from twenty
averages of 10 ps simulations. The temperature was maintained at 300 K for these
simulations with an NPT ensemble and a 2 fs step between each time frame output.
From the simulation, the number of water molecules added to the system does
not increase linearly with decreasing alcohol concentration. Instead a smooth
gradual increase in the number of water molecules and hydrogen bonds is seen as
5 Terahertz Devices and Systems … 137

Table 5.2 Polar liquid system parameters

Alcohol Alcohol Water molecules in Water molecules in 4.3 nm
volume % molecules in 3.8 nm ethanol box propan-2-ol box
box
0 0 1728 2617
5 50 1555 2397
10 100 1389 2170
30 300 821 1395
50 500 400 845
70 700 165 412
80 800 74 200
90 900 30 64
95 950 10 39
100 1000 0 0

the alcohol concentration changes. The THz absorption aqueous solution series
therefore cannot be explained by a simple linear model of increasing water
hydrogen bonds.
The non linear increase in water molecules can be explained by the efficiencies
in packing alcohol-alcohol, alcohol-water and water-water bonds in the system.
Water, with its optimised geometry can efficiently place hydrogen bonds, and in the
simulation, pure water forms 3.34 bonds per water molecule. As alcohol is added to
the system, alcohols are sterically hindered from forming hydrogen bonds by their
carbon chains, causing inefficient packing of water molecules and a decrease in the
relative density of the system.
The reason why we observe a linear increase THz absorption in the experimental
results, but not equally linear increasing number of hydrogen bonds can be
explained in the terms of the dynamics of the hydrogen bonds. Strong and
long-lived hydrogen bonds have been previously correlated with higher THz
absorption [15]. In two systems with equal hydrogen bond density, the stronger and
longer-lived hydrogen bonds will absorb more THz radiation than those with
weaker and shorter lifetimes.
The strength of hydrogen bonds can be assumed to directly relate to the lifetime
of the hydrogen bond. The strength-lifetime relationship is calculated from the
molecular dynamic simulations using the Hydrogen Bond Auto Correlation func-
tion [16]. The existence of hydrogen bonds is calculated for any bond forming pair
within 35 nm. The HBAC function used in this experiment allows for the
re-formation of the same hydrogen bond, that is a donor and acceptor may move
away from the hydrogen bond geometry and subsequently move back. This is then
counted as one continuous hydrogen bond.
HBAC is calculated for 1, 2.5 and 5 picoseconds for the propan-2-ol concen-
tration series. A high HBAC can indicate strong and long-lived hydrogen bonds
relative to those with a lower HBAC [15]. Below an HBAC value of 0.5, half of the
original hydrogen bonds have been broken. At high propan-2-ol concentrations,
138 A.J. Vickers and D. Crompton

extreme hydrogen bond lifetimes are seen, suggesting these hydrogen bonds are
locked in position by steric constraints of the alcohol.
The high HBAC values in the alcohol systems then seem to suggest alcohol forms
strong hydrogen bonds, despite the perceived inefficient packing of the water mole-
cules. The HBAC results are then counter intuitive, as alcohols are expected to weaken
hydrogen bonds through a weak hydroxyl group and carbon steric effects. As a further
test of hydrogen bond strength the average distance distribution of the hydrogen bonds
was calculated for the propan-2-ol concentration series. Hydrogen bond length can be
directly correlated to the strength of hydrogen bonds, with shorter bonds having a
greater strength than the equivalent longer bonds [17]. The result of the distance
distribution indicates the bonds in high alcohol concentrations are stretched relative to
those found in water by approximately 0.01 Å between pure propan-2-ol and water.
These results then disagree with the assumption that HBAC predicts the hydrogen
bonds in the alcohol solutions are stronger. From the distance distribution of hydrogen
bonds and the HBAC it can be concluded the hydrogen bonds in the alcohol solutions
are both sterically hindered and relatively weak in comparison to bulk water.
The contradiction in results between HBAC and hydrogen bond distance dis-
tribution suggests the THz absorption of hydrogen bonds cannot be simply
explained in terms of the hydrogen bond strength. Instead it appears it is the
hydrogen bond dynamics, controlled by both the strength of the bond and steric
hindrance from the alcohol, that dominate the THz absorbance at the low hydrogen
bond density seen in the high propan-2-ol concentrations.
This leads to the conclusion that the alcohol solutions have weaker and sterically
hindered hydrogen bonds, which have increased THz absorption properties com-
pared to non-sterically hindered and strong hydrogen bonds of bulk water. After
50 % water concentration the weaker, but more abundant water-water hydrogen
bonds begin to dominate the system and increase in THz absorption through their
efficient packing of hydrogen bonds at high water concentrations. It is then the
combine effects of sterically hindered hydrogen bonds at high propan-2-ol con-
centrations and the sheer number of packed hydrogen bonds in pure water that cause
the linear increase in THz absorption observed in the experimental results. A similar
result has been observed when water is exposed to a hydrophobic region of a protein
[18]. The hydrophobic regions of the protein act in the same manner as the alcohol
solutions, actively hindering the dynamics of hydrogen bonding, by repelling water
away from the protein surface. This led to an observed increase in THz absorption,
which this report attributes to the increase of sterically hindered waters.

5.4 Great Complexity—Proteins

Low frequency dynamics are not exclusive to hydrogen bond networks, but also
exist as large collective motions across proteins. These collective motions change as
the protein undergoes biological processes and often underpin the activation or
deactivation of the protein. THz radiation is able to probe these collective motions
5 Terahertz Devices and Systems … 139

and in this section we use THz radiation to monitor the activation of G-Protein
coupled receptors.
G-Protein coupled receptors (GPCRs) are one of the largest and most diverse
group of membrane proteins found within the animal kingdom [19]. GPCRs are
characterized by a common structure of seven transmembrane helices (TM1,
TM2,..,TM7) that allow them to form cytoplasmic and extracellular domains across
the cellular lipid bilayer [20]. Their principal role is to act as a transmembrane
messenger and regulator of many biological functions, which has made them a
highly prized drug target, with 40–60 % of all modern medicines believed to target
a GPCR [21, 22]. Due to their significance in drug candidate research, GPCRs are
also one of the most widely studied classes of proteins.
The typical modes of action for GPCR drugs or ligands can be broadly classed in
two groups of agonists and antagonists, where the drug promotes or inhibits the
GPCR activation [23]. By binding to the GPCR, the drug or ligand can alter the
flow of activation energy across the protein by altering the molecular constraints of
the protein. This flow of activation energy is strongly linked to the low frequency
normal modes of the molecule, which sample the local spatial geometry [24, 25].
The sampling of the local geometry can lead to conformational changes, which
ultimately activate or deactivate the protein. The activated and deactivated low
frequency modes appear to be conserved between GPCRs [26] as the low frequency
modes seem to be an inherent property of the secondary structure of the protein
[27]. The THz frequency regime is able to directly probe these low frequency
modes and therefore sample the activation and conformational state of the protein.
In this section, the low frequency dynamics of GPCR Bovine Rhodopsin are
studied in an aqueous environment in both a membrane and non-membrane solu-
tion. Bovine rhodopsin is a key protein in monochromatic vision in cattle. Within
the GPCR super family rhodopsin is a type A, the largest class of GPCRs. The
proteins in this family are called opsins (visual purple), light sensitive proteins,
which are non-homologous, but have evolved to perform a similar function and
share a similar structure. Opsins are further sub divided into two groups, type I
Prokaryotic [28] and type II Animal [29].
Rhodopsin being a photoreactive protein makes it an ideal GPCR to study, as it
can be simply activated via light illumination without affecting chemical compo-
sition of the solution. ATR THz-TDS spectroscopy is then used to observe any
change in the low frequency dynamics of the protein in both the inactive dark and
active light state. Any shift in low frequency dynamics should indicate a confor-
mational change upon protein activation. By testing the rhodopsin in membrane and
non-membrane solution, the impact of the membrane on the low frequency
dynamics can also be established.
Further to ATR THz-TDS spectroscopy of rhodopsin, elastic networks are
computationally calculated from crystallised rhodopsin proteins in both activation
states. From these elastic models, the normal modes of the protein are calculated,
which are synonymous with the low frequency modes. The normal modes of the
elastic models are then used to help interpret data in the experimentally observed
THz spectrum.
140 A.J. Vickers and D. Crompton

Bovine rhodopsin and all opsin type proteins are activated through photoiso-
merisation of the chromophore, where specific wavelengths of the electromagnetic
spectrum act as the primary stimulus [30]. The rhodopsin chromophore retinal is
linked to the protein via a protonated Schiff base located at Lys296 on the TM7
helix [20]. This location places the positively charged Schiff base into a
hydrophobic area neutralized by a neighboring Glu113 residue. This structure helps
stabilize the inactive rhodopsin molecule until it is activated [31]. The retinal, in the
inactive state, is initially in the cis conformation. When illuminated, photoiso-
merisation of the retinal occurs, resulting in a cis to all-trans formation between the
carbon-11 and carbon-12 double bonds [20]. The initial cis to all-trans takes just a
few femtoseconds once photo excited [32]. This extension of the chromophore
leads to the series of conformational changes that ultimately lead to the activation of
the rhodopsin molecule within a few milliseconds [33].
Inside the core of rhodopsin exists what is believed to be a highly conserved
water network that interacts with the protein through hydrogen bonding [34]. The
hydrogen bond network is thought to help meditate conformational changes across
the rhodopsin protein and assist the color tuning of the rhodopsin chromophore that
helps stabilize the ground state of the GPCR [35]. The sensitivity of THz radiation
to the hydrogen bond dynamics and the importance of hydrogen bonds to protein
structure potentially makes THz radiation highly sensitive to the activation state of
the protein.
Activation-critical hydrogen bonding is also observed between residues within the
hydrophobic core of the protein. During activation the β-ionone ring moves toward
TM5, this movement rearranges the hydrogen bond network leading to a weakening
of the hydrogen bond to the Glu122 residue. A change in Glu122 absorption occurs
between the Meta IIa and the Meta IIb states suggesting that in the Meta IIa state the
hydrogen bond network is still in the inactive conformation [36]. The disruption of the
Schiff base upon activation also causes a rearrangement of the core hydrogen bond
network around TM1, TM2, and TM7 with the help of Asp83 [37].
Previous THz-TDS studies of rhodopsin have used thin film dried samples [26]
and small aqueous sample transmission path lengths (15 µm) [38] to prevent high
water absorptions. For the experimental method in this chapter, THz-TDS is
combined with an ATR sample unit to achieve high sensitivity despite the aqueous
solvent. The theoretical method uses NMA to calculate the low frequency modes of
the protein by using an elastic model based on the atomic coordinates from x-ray
crystallography of protein structures. From each low frequency mode, the fluctu-
ations of the protein residues can be seen.
For the THz-TDS experiment, rhodopsin is activated by white light illumination,
so any change in the THz spectrum between activation states should be due to the
low frequency dynamics and hydrogen bonding within the protein. Rhodopsin is
tested both in a lipid bilayer membrane and non-membrane solution to observe if
the membrane has any noticeable impact on the protein’s low frequency dynamics.
Any change in the low frequency dynamics between inactive, active, membrane and
non-membrane results should be revealed by a change in the THz spectrum and the
level of residue fluctuations in the NMA.
5 Terahertz Devices and Systems … 141

Bovine rhodopsin rod outer segments were prepared from bovine retina using
sucrose gradient centrifugation. The samples were urea washed and suspended in an
aqueous buffer (20 mM Tris-HCl [pH 7.2], 2 mM EDTA, 1 mM 2-mercaptoethanol,
0.1 mM PMSF). The resulting concentration was approximately 2 mg/ml [38]. The
rhodopsin samples were prepared in both membranes (BRM) and non-membrane
(BR). A dim red light is used for lighting during preparation to prevent accidental
activation of the protein.
The THz-TDS with ATR unit, described in Sect. 5.2, was used to resolve the
THz spectrum of each bovine rhodopsin sample. The experiments were carried out
in a controlled environment with a temperature of 20 °C and a dry nitrogen purged
atmosphere. The samples were kept in the dark to avoid accidental activation and
were pipetted in turn onto the sample facet of the dove prism. A total of four scans
for each sample was taken, with an average of 30 scans of each step of the
THz-TDS translation stage. This resulted in four sets of data, the inactive bovine
rhodopsin membrane dark state (BRMDark) and bovine rhodopsin non-membrane
dark state (BRDark). Followed by after 1 min illumination with a white LED, the
active bovine rhodopsin membrane light state (BRMLight) and bovine rhodopsin
non-membrane light state (BRLight).
Figure 5.4 shows, the peak terahertz electric field response of BRMDark,
BRMLight, BRDark and BRLight samples. The total amplitude of the THz electric
field appears to be unaffected by the protein’s activation state or its membrane
environment. The phase of the THz pulse, however, varies between each of the
samples. For the membrane and non-membrane solutions BRMDark-BRDark and
BRMLight-BRLight respective samples, the phase change appears as a +0.1 ps time
shift. This is not unexpected as the membranes in the bovine rhodopsin solution will
displace some of the water, which has a high refractive index in the THz region that
will affect the phase of the THz pulse [15].

BRMDark
Terahertz Electric Field (Arb. Units)

500
450 BRMLight

400 BRDark
BRLight
350
300
400
Terahertz Electric

250
FIeld (Arb. Units)

200 200
150
0
100 0 20 40 60
50 -200 Time (ps)
0
4.4 4.5 4.6 4.7 4.8 4.9 5
Time (ps)

Fig. 5.4 THz-TDS of bovine rhodopsin: key differences between the samples THz pulses can be
seen at the 4.7 ps peak. Error bars show the standard deviation between data averages. Inset The
entire THz pulse recorded for each sample
142 A.J. Vickers and D. Crompton

Upon protein activation, a phase change is also observed between the inactive
dark state and active light state. For membrane and non-membrane solutions, this
phase change from dark to light state is +0.63 and +0.6 ps respectively. As the
chemical composition of the sample from dark to light is unchanged, this suggests
the rhodopsin protein has undergone a conformational change. The more pro-
nounced phase change in the membrane solution may indicate the membrane has a
non-trivial influence on the conformational change of the protein.
From the THz electric field response, a fast Fourier transform algorithm is used
to generate the THz transmission frequency spectrum of the sample. Background
correction is applied to the sample spectrum by division by the reference trans-
mission spectrum. In this instance, the reference spectrum is the ATR prism with no
sample. The result of this calculation is shown in Fig. 5.5.
There are three key features of the rhodopsin absorption spectrum that can be
observed in these results. First it appears rhodopsin in both aqueous and membrane
systems absorb more radiation at lower THz frequencies. The low frequency modes
of the protein are expected to significantly contribute the absorptions in this region
at these frequencies. Secondly, the rhodopsin in membrane absorbs less THz
radiation than the rhodopsin in aqueous solution. The membranes displacement of
highly absorbing waters may account for the decrease in THz absorption between
the rhodopsin samples, as previous reports have shown membranes absorb signif-
icantly less THz radiation than water [39]. Although the membranes do not sig-
nificantly absorb THz radiation, they are likely to affect the THz absorbing low
frequency dynamics of the protein through steric or electrostatic interaction due to
their close proximity. The potential of this interaction is further investigate in the
difference spectrum below. The third feature is the light activated state is observed
to absorb more THz radiation than the dark inactive state for both aqueous and
membrane solutions. This indicates the unlocking of low frequency modes upon
protein activation, with the increased low frequency modes absorbing more THz
radiation.

0.6
Transmission Spectrum

0.55

0.5
BRDark
0.45
BRLight
0.4
BRMDark
0.35
BRMLight
0.3
0.2 0.3 0.4 0.5 0.6 0.7 0.8
Frequency (THz)

Fig. 5.5 THz frequency spectrum of bovine rhodopsin: the results have been corrected for
spectrometer absorptions and the standard error is calculated from the average of four sample scans
5 Terahertz Devices and Systems … 143

0.03
BR

Difference Absorbance
0.025

0.02 BRM

0.015

0.01

0.005

0
0.2 0.4 0.6 0.8 1 1.2 1.4
-0.005 Frequency (THz)

Fig. 5.6 Difference absorbance of rhodopsin in aqueous and membrane solutions

To help elucidate the THz characteristics of the activation states of rhodopsin,

the difference spectrum between BRMDark-BRMLight and BRDark-BRLight is
calculated. This effectively removes the THz characteristics of the water and
membrane molecules, leaving the contributions of the low frequency modes from
the protein inactive-active conformational change. The results of the difference
spectrum are shown in Fig. 5.6.
There appears to be significant differences between the inactive state and active
state between 0.2 and 1.4 THz for both membrane and non-membrane systems.
This difference is more pronounced in the membrane solution with an average
difference of 0.005 (Arb.Units) higher than the rhodopsin in aqueous solution. The
membrane appears to have a significant effect on the low frequency dynamics of
rhodopsin, making a more defined response between the two activation states. This
suggests the membrane may stabilise the low frequency dynamics between inactive
and active rhodopsin, where the non-membrane protein low frequency modes are
less restricted. After 1.4 THz, the difference spectrum remains at approximately
zero (Arb. Units), indicating the majority of the activation low frequency modes lie
between 0 and 1.4 THz.
For the purpose of molecular dynamic simulations, there are effectively four
samples in the rhodopsin THz-TDS experiments, inactive and active rhodopsin both
in- and not-in membrane solution. Inactive rhodopsin was provided by the protein
data bank (PDB) file 1U19 [40], where the retinal remains chemically bound to the
protein. Active rhodopsin was modelled by the PDB file 2 × 72 [41], which has
been crystallised after the dissociation of the retinal ligand. Topologies for the
protein was generated using GROMACS. A pre-equilibrated 1-palmitoyl-2-oleoyl-
sn-glycero-3-phosphocholine (POPC) membrane lipid bilayer, by Andreas Kukol,
was used to imbed the membrane proteins [42].
To generate the non-membrane samples, each of the inactive and active state
proteins was solvated with SPC water with a 20 Å buffer away from the protein
surface. The membranes systems were made by superimposing the membrane onto
the protein in VMD, which is then followed by a 20 Å SPC buffer solvation.
GROMACS is then used to “grow” the protein in the membrane removing any
144 A.J. Vickers and D. Crompton

steric clashes [43]. This results in four model systems; inactive no-membrane,
inactive membrane, active no-membrane and active membrane. Each system is then
minimised to 1 kJ mol−1, followed by 200 ps of NVT ensemble and 1 ns of NPT
ensemble. Standard Berendsen temperature and pressure coupling were used to
maintain the system at 1 ATM and 300 K, with periodic boundary conditions.
The normal mode analysis for each model rhodopsin system was calculated
using an anisotropic network model (ANM). The calculation was performed on all
backbone carbons excluding the initial 33 residues and all residues up to residue
334. These excluded residues are the C and N terminus of the protein, which are
loose chains that may mask the low frequencies of the protein. The first six modes
of the calculation are trivial, representing the rotational movement of the protein, so
are also ignored.
The results of this calculation, for the first three non-trivial modes, are shown in
Fig. 5.7, as the mean square fluctuation of each residue. From Fig. 5.7, there can be
seen significant differences between each active state, typically by a change in peak
amplitudes at residues 100, 240 and 370 in non-membrane and residues 145 and
240 in membrane solution. The residues that form these peaks are prominently
found in the loop regions of the protein, so a relatively high mobility is expected. In
membrane there are significantly fewer peaks in the square fluctuation across the
protein’s residues, suggesting the membrane is suppressing some of the mean
square fluctuations. A change is also observed in the frequency of the normal modes
between each of the active states. In non-membrane solution the frequency appears
to undergo a red shift, for the first and third modes, when the protein is activated. In
membrane, only a small red shift is observed for the first mode and large blue shifts
for the second and third mode.
Figure 5.7c, d shows in more detail the square fluctuations of α-helical structure
in the membrane solution, which are highlighted in red. A clear increase in square
fluctuation can be seen in modes 1 and 3, with mode 2 remaining relatively the
same except at loop residue 150. This same distinct increase in α-helical structure
square fluctuations, between activation states, is ill-defined in the non-membrane
systems.
In conclusion, in membrane, there is a greater observed difference spectrum
between the low frequency modes of the inactive and active states compared to the
non-membrane solutions, but membranes themselves have minimal THz absorption
[39]. NMA of the protein shows the membrane suppresses the low frequency modes
in the α-helical and loop structures of the protein. Despite the suppression, the
membrane still allows for an increase in the low frequency modes upon protein
activation.
Without membrane, the low frequency modes of the α-helices and loops appear
less defined, combined with significantly larger square fluctuations across the
protein for both activation states. Suggesting the membranes may act as a stabilising
agent, helping the protein maintain the inactive and active conformation.
It seems likely that the protein has evolved to predominantly exist in either
inactive or active state when in its natural membrane environment. Taken out of this
environment has allowed the protein to sample more conformational states, shown
5 Terahertz Devices and Systems … 145

Fig. 5.7 a Mean square (a)

Mean square fluctuiations ((Å)

0.7
fluctuation as a function of Mode 1
0.6 Mode 2
residue number for the
0.5 Mode 3
non-membrane inactive
bovine rhodopsin (1U19). 0.4
b Mean square fluctuation as 0.3
a function of residue number 0.2
for the non-membrane active 0.1
bovine rhodopsin (2 × 72).
0
c Mean square fluctuation as a 0 50 100 150 200 250 300
function of residue number Residue number
for the membrane inactive
bovine rhodopsin (2 × 72). (b)

Mean square fluctuiations ((Å)

0.7
Mode 1
d Mean square fluctuation as
0.6 Mode 2
a function of residue number
0.5 Mode 3
for the membrane active
bovine rhodopsin (2 × 72) 0.4

0.3

0.2

0.1

0
0 50 100 150 200 250 300 350
Residue number

(c) 0.04
Mean square fluctuiations (Å)

0.035
TM1 TM2 TM3 TM4 TM5 TM6 TM7
0.03
0.025
0.02
0.015
0.01
0.005
0
0 50 100 150 200 250 300
Residue number

(d) 0.04
Mean square fluctuation ((Å)

0.035
0.03
0.025
0.02
0.015
0.01
0.005
0
0 50 100 150 200 250 300
Residue number
146 A.J. Vickers and D. Crompton

by the increased NMA square fluctuations in the non-membrane solutions. Despite

being able to sample more conformations, many of these may be inactive or par-
tially active. The membrane, however, stabilises the inactive state and may force the
protein towards a full activation by restricting these square fluctuations in the
protein. The increased difference, seen in the THz difference spectrum, between the
membrane activation states may then be explained by the membrane interaction
with the low frequency modes.
In the discussion above, the changes in the THz spectrum of the protein were
linked to the low frequency dynamics of the protein’s α-helical structure. Also the
water and hydrogen bonding plays an active role in the function and structure of
rhodopsin. Other reports have shown that the dynamics of the hydrogen bonding
are directly influenced by the solute and the low frequency dynamics of biomole-
cules can be coupled to the solvating water [40, 42]. This suggests the change in the
THz between activation states may be partly influenced by the waters coupled to the
low frequency dynamics of the protein.
GLU122 and surrounding residues have been identified as key residues on
forming protein activation dependent hydrogen bonds [36]. To assess the hydrogen
bonds around this residue in the model systems, HBAC analysis and the total
number of the hydrogen bonds are calculated 1 nm around GLU122. The HBAC is
shown to be higher in the inactive form than the active form, indicating sterically
hindered bonds prior to protein activation. This is in agreement with the observed
weakening of the hydrogen bonds during activation of the protein [36].
Normally an increased restriction and density of hydrogen bonds would indicate
stronger THz absorbing properties of the system, which has been observed in the
previous section on polar liquids. Ignoring the low frequency mode contributions,
this would suggest the inactive state around GLU122 absorbs more THz radiation
than the active state of the protein. But the THz absorption spectrum of the acti-
vation states reveals the reverse is true. This indicates the low frequency dynamics
of the protein are then the most significant contributor to the THz characteristics of
the system. The low frequency modes, however, may actively contribute to the
weakening of the hydrogen bond network, allowing the ligand to fully dissociate
from the system.
Future molecular dynamics simulations may reveal if the low frequency
dynamics are directly coupled to the weakening of the hydrogen bonds at the ligand
binding site. Further THz-TDS experiments may also explore the intermediary
states of rhodopsin activation, where the research in this chapter is solely focused
on the inactive and active states of rhodopsin. Of particular interest is the change
from the Meta I state to the Meta II state, where the TM6 helix shifts releasing
activation energy. The sensitivity of the ATR THz-TDS should be able to detect the
TM6 shift as an unlocking of low frequency modes. It may then be possible to link
the shift as a critical step to protein activation, by measuring its contribution to the
active state low frequency modes.
5 Terahertz Devices and Systems … 147

5.5 Conclusions

THz TD spectroscopy has been demonstrated as tool to study biomolecules in an

aqueous environment. The use of an attenuated total reflection sample cell enables
rapid sample change, avoids the need for standardized volume cells, and does not
require the spectrometer to be repeatedly nitrogen purged when changing samples.
We have show through the experimental results and the simulations that water plays
a very important role in the overall dynamics of biomolecules. This work and the
work of others has provided some initial steps towards a long term goal of applying
biomelecular engineering to the rapid design of drugs, an aim that would provide
significant benefit to world health.

References

1. P.J. Hajduk, J. Geer, Nat. Rev. Drug Discov. 6(3), 211 (2007)
2. M. Congreve, J.M. Dias, F.H. Marshall, Prog. Med. Chem. 53, 1 (2014)
3. H. Ge, Y. Jiang, F. Lian, Y. Zhang, S. Xia, Sensors 15(6), 12560 (2015)
4. F. Gao, W. Tress, J. Wang, O. Inganas, Phys. Rev. Lett. 114(12), 128701 (2015)
5. J. Xu, W. Plaxco, S.J. Allen, Protein Sci. 15, 1175 (2006)
6. Nimbus Theraputics, Our technologies, nimbustx.com. http://nimbustx.com/our-approach/our-
technologies. Accessed 15 July 2015
7. P. Thoma, V. Judin, Optik & Photonik 8, 58 (2013)
8. D. Crompton, A.J. Vickers, in 13th Biennial Baltic Semiconductor Conference, vol. 13 (2012)
9. D. Crompton A.J. Vickers, in Computer Science and Electronic Engineering Conference, vol.
47 (2012)
10. H. Naito, Y. Ogawa, K. Shiraga, N. Kondo, I. Osaka, A. Kubota, in International Symposium
on Systems Integration (SII), vol. 192 (2011)
11. D.H. Choi, H. Son, S. Jung, J. Park, W.Y. Park, O.S. Kwon, G.S. Park, in Infrared. Millimeter,
and THz Waves, 1 (2012)
12. W.L. Jorgensen, in Encyclopedia of Computational Chemistry (2002)
13. W.L. Jorgensen, D.S. Maxwell, J. Tirado-Rives, J. Am. Chem. Soc. 118, 11225 (1996)
14. M. Frisch, G. Trucks, H.B. Schlegel, G. Scuseria, M. Robb, J. Cheeseman, G. Scalmani, V.
Barone, B. Mennucci, G. Petersson, Inc., Wallingford, CT, 270, 271 (2009)
15. M. Heyden, Fakultat fur Chemie und Biochemie (Ruhr-Universitat Bochum, Bochum, 2010)
16. D. van der Spoel, E. Lindahl, B. Hess, A. Van Buuren, E. Apol, P. Meulenhoff, D. Tieleman,
A. Sijbers, K. Feenstra, R. van Drunen, Gromacs User manual version 3.3 (2008)
17. S. Goryainov, Physica B 407, 4233 (2012)
18. T.Q. Luong, P.K. Verma, R.K. Mitra, M. Havenith, Biophys. J. 101, 925 (2011)
19. K. Palczewski, T. Kumasaka, T. Hori, C.A. Behnke, H. Motoshima, B.A. Fox, I.L. Trong, D.
C. Teller, T. Okada, R.E. Stenkamp, Science Signaling 289, 739 (2000)
20. D.M. Rosenbaum, S.G. Rasmussen, B.K. Kobilka, Nature 459, 356 (2009)
21. J.P. Overington, B. Al-Lazikani, A.L. Hopkins, Nat. Rev. Drug Discovery 5, 993 (2006)
22. Y. Arinaminpathy, E. Khurana, D.M. Engelman, M.B. Gerstein, Drug Discovery Today 14,
1130 (2009)
23. B. Bosier, E. Hermans, Trends Pharmacol. Sci. 28, 438 (2007)
24. K.C. Chou, Biophys. J. 48, 289 (1985)
25. K.C. Chou, Biophys. Chem. 30, 3 (1988)
148 A.J. Vickers and D. Crompton

26. R. Balu, H. Zhang, E. Zukowski, J.-Y. Chen, A. Markelz, S. Gregurick, Biophys. J. 94, 3217
(2008)
27. K. Yamamoto, K. Tominaga, H. Sasakawa, A. Tamura, H. Murakami, H. Ohtake, N.
Sarukura, Biophys. J. 89, L22 (2005)
28. M.X. Ruiz-Gonzalez, I. Marín, J. Mol. Evol. 58, 348 (2004)
29. Y. Shichida, Philos. Trans. R. Soc. B: Biol. Sci. 364, 2881 (2009)
30. J. Buczyłko, J.C. Saari, R.K. Crouch, K. Palczewski, J. Biol. Chem. 271, 20621 (1996)
31. G.B. Cohen, D.D. Oprian, P.R. Robinson, Biochemistry 31, 12592 (1992)
32. R.S. Liu, G.S. Hammond, Photochem. Photobiol. Sci. 2, 835 (2003)
33. J.-P. Vilardaga, M. Bünemann, C. Krasel, M. Castro, M.J. Lohse, Nat. Biotechnol. 21, 807
(2003)
34. R. Nygaard, T.M. Frimurer, B. Holst, M.M. Rosenkilde, T.W. Schwartz, Trends Pharmacol.
Sci. 30, 249 (2009)
35. T. Okada, Y. Fujiyoshi, M. Silow, J. Navarro, E.M. Landau, Y. Shichida, Proc. Natl. Acad.
Sci. 99, 5982 (2002)
36. H.-W. Choe, Y.J. Kim, J.H. Park, T. Morizumi, E.F. Pai, N. Krauß, K.P. Hofmann,
P. Scheerer, O.P. Ernst, Nature 471, 651 (2011)
37. P.T. Ho, J.M. Moran, K.Y. Lo, Astrophys. J. Lett. 616, L1 (2004)
38. A. Vickers, R. Dudley, P. Reeves, C. Reynolds, S. Gadde, T. Nithyanand, Y. Ma, in Infrared
and Millimeter Waves, vol. 395 (2007)
39. K. Tielrooij, D. Paparo, L. Piatkowski, H. Bakker, M. Bonn, Biophys. J. 97, 2484 (2009)
40. F. Despa, A. Fernández, R.S. Berry, Phys. Rev. Lett. 93, 228104 (2004)
41. J. Standfuss, P.C. Edwards, A. D’Antona, M. Fransen, G. Xie, D.D. Oprian, G.F. Schertler,
Nature 471, 656 (2011)
42. P. Fenimore, H. Frauenfelder, B. McMahon, F. Parak, Proc. Natl. Acad. Sci. 99, 16047 (2002)
43. M.G. Wolf, M. Hoefling, C. Aponte-Santamaria, H. Grubmüller, G. Groenhof, J. Comput.
Chem. 31, 2169 (2010)
Chapter 6
Recent Progress in XAFS Study
for Semiconducting Thin Films

Takafumi Miyanaga and Takashi Azuhata

Abstract X-ray absorption fine structure (XAFS) is a powerful structural analysis

tool for studying the local structure in thin films. In the recent developments of this
technology, high-brightness light emitting diodes and laser diodes are applied to
semiconducting InGaN nanolayers. In this chapter, we provide a brief review on
basic theory and measurements of XAFS, and then present some recent results of
XAFS application to c-plane (0001) InGaN, m-plane (1
100) InGaN, m-plane
AlGaN, and c-plane MgZnO to reveal the atomic distribution and the interatomic
distances in the films.

6.1 Introduction

X-ray absorption fine structure (XAFS) is a well-established structural analysis

technique used for the characterization of materials in a variety of forms [1–3]. In
particular, XAFS is a powerful tool in investigating local structures of thin layers
composed of two or more elements. The characteristics of XAFS are
element-specificity and electron-shell-specificity, which make XAFS a useful probe
complementary to other techniques, to obtain structural information and electronic
structure. XAFS is considered to be established in the 1970s with pioneering work
by Lytle, Sayers, and Stern [4] after struggling for many years [5]. XAFS has
progressed tremendously in the development of experimental techniques and the-
oretical understanding. Modern synchrotron light sources have played an important
role and have provided a large variety of new applications, such as structural
characterization of nanostructures and time-resolved studies for dynamic processes.
Due to the progress in theoretical understanding, more exact XAFS functions can
be calculated with computer code. XAFS is presently used over many fields;
physics, chemistry, material sciences, geology, biology, and environmental science.

T. Miyanaga (&)
T. Azuhata

Department of Advanced Physics, Hirosaki University, Hirosaki, Aomori 036-8561, Japan
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 149

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_6
150 T. Miyanaga and T. Azuhata

Semiconductors are basic materials used in a variety of electronic and photonic

devices and are indispensable in today’s society. Presently, a new world of semi-
conductor nanomaterials is being extensively explored. Semiconducting materials
also play an important role in sustainable energy supply, which is one of the great
global challenges for the 21st century to reduce greenhouse gases such as CO2, and
to provide clean energy especially after the Fukushima nuclear accident. By directly
converting sunlight to electricity, photovoltaic devices such as solar cells can
provide a versatile and renewable energy source.
The most effective utilization of these semiconductor materials needs knowledge
of their structural properties which connect with electrical, optical, and magnetic
properties and so can enhance device performance. XAFS has provided unique and
valuable structural and electronic information in a large number of semiconductor
systems.
GaN is one of the direct wide band gap semiconductors and has potential for
high-power devices and radio-frequency applications. Due to the direct band gap it
is also attractive for high performance light emitting devices, such as light emitting
diodes (LEDs) and laser diodes (LDs). The first high efficient blue LED was
realized using GaN and its mixtures with AlN and InN. The advantage of shorter
wavelength in optical storage devices is to provide higher storage density. And the
combination of the devices that emit red, green, and blue light is utilized in panel
displays and white-light emission. In 2014, the Nobel Prize in physics was awarded
in this field. On the other hand, ZnO-based semiconductors are also of current
research interest due to the direct wide gap and piezo- and pyroelectric properties as
materials competitive to GaN-based ones.
Such devices consist of single-quantum-well (SQW) or multi-quantum-well
(MQW) structures to control electronic and, consequently, optical properties. The
electronic structures in SQWs and MQWs are different substantially from those of
bulk materials, because electrons are confined in one space direction. In quantum
wells (QWs), the electron-hole recombination rate is improved and the lumines-
cence efficiency is enhanced. Naturally, the structure and atomic distribution in
QWs is different from bulk, so it is important to study them by XAFS. The latter
can be considered a hybrid technique because the information for both local
structure and electronic structure is included, which makes XAFS a unique ana-
lytical technique to study semiconducting thin films.

6.2 XAFS Spectroscopy: Basic Theory and Measurements

6.2.1 Basic Theory of XAFS

XAFS is the fine structure on X-ray absorption coefficient spectra around an

absorption edge of a particular element. When monochromatic X-rays are irradiated
on a sample, the intensity of the transmitted X-ray, It, depends on the photon energy
6 Recent Progress in XAFS Study for Semiconducting Thin Films 151

sample
d
incident X ray, I 0 transmitted X ray, I t

fluorescence X ray, I f

Fig. 6.1 Schematic representation of transmission and fluorescence mode

E and sample thickness d (Fig. 6.1). According to the Beer-Lambert law, It is

expressed as

It ¼ I0 elðEÞd : ð6:1Þ

Here, I0 is the intensity of the incident X-ray and μ(E) is the energy-dependent
X-ray absorption coefficient. Since μ(E) * a Z4/m E3 where a, Z, and m are
respectively the density of the sample, atomic number, and electron mass, the
magnitude of μ(E) monotonically decreases as the photon energy is increased.
However, if the photon energy reaches the binding energy of core electrons, the
energy of a photon is consumed to excite a core electron and the photon is anni-
hilated to create a photoelectron and a core hole. As the result, a new absorption
channel is opened and μ(E) shows steep increase around the absorption edge. The
core hole recombines with an electron at the higher energy state with X-ray
fluorescence or the creation of an Auger electron.
According to the quantum mechanical perturbation theory, transition rate is
proportional to the square of the corresponding matrix element M [6]. Thus, μ(E) is

l/jM j2 ¼ jhf jHI jiij2 ¼ jhf0 þ df jHI jiij2 ffi jhf0 jHI jiij2 þ 2Rehf0 jHI jii hdf jHI jii;
ð6:2Þ

where jii and j f i are the initial and final states, respectively, and HI is the inter-
action Hamiltonian between the X-ray photon and the electron. In the case of
K-edge absorption, jii is the 1s electron state. jf0 i is the final state for a hypothetical
isolated atom. jf0 þ df i is the final state with interference between outgoing and
incoming electron waves after the electron is backscattered by surrounding atoms,
which contains the essential information on XAFS. In this equation, the second
order term of jdf i is neglected. The first term jhf0 jHjiij2 on the right side corre-
sponds to μ0 and XAFS function χ(k) is

vðkÞ ¼ ðl  l0 Þ=l0 /hdf jHI jii; ð6:3Þ

152 T. Miyanaga and T. Azuhata

where k is the photoelectron wavenumber converted from E using

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
k ¼ 2m ðE  E0 Þ=
h2 .
For higher energy region of about 60–1000 eV from the absorption edge, the
expression can be simplified by some approximations: Muffin-tin approximation for
the atomic potential, one-electron approximation in the absorption process, dipole
approximation for the photon-electron interaction, plane-wave approximation for
the photoelectron, and single scattering approximation. The fine structure in this
energy region is called Extended XAFS (EXAFS) and the expression can be
simplified as
X N  S2 ðkÞ   
vðkÞ ¼ j 0 fj ðk; pÞe2ðr2j k2 þ rj =kðkÞÞ sin 2krj þ 2dðkÞ þ /j ðkÞ ; ð6:4Þ
kr 2j
j

where rj, N*j , and σj are the interatomic distance, the effective coordination number,
and the root-mean-square relative displacement (or Debye-Waller factor), respec-
tively, for atomic pair j. These parameters can be determined by the non-linear least
square fitting (so-called curve fitting) method by, e.g., XANADU [7] and Athena
and Artemis code [8]. The phase shift φj(k), the backscattering amplitude fj(k), and
the electron mean free path λ(k) are generally obtained from theoretical calculation,
for example, FEFF code [9].
In principle, the fitting of the theoretical curve to experimental data can be
performed over the entire EXAFS spectrum. More frequently, however, the fit is
carried out only on a filtered portion of the data which corresponds to the first
several coordination shells. Comparison of the theory with experimental data can be
performed in the r-space after Fourier transform or in the k-space after
back-transform. The maximum number of independent parameters that can be
reasonably obtained from one EXAFS spectrum is proposed as the following
equation [3]:

Nind  2DkDr=p; ð6:5Þ

where Δk(=kmax – kmin) and Δr(=rmax – rmin) are intervals in each space of the
spectrum.
For crystalline and oriented samples, the effective coordination number N*j is
expressed as
X
Nj ¼ 3 cos2 hi ; ð6:6Þ
i

where θi is the angle between the direction of ith photoelectron scattering atom from
the X-ray absorbing atom and the polarization direction of the incident X-ray. The
Debye-Waller factor can be separated into a thermal vibration term and statistical
disorder term as
6 Recent Progress in XAFS Study for Semiconducting Thin Films 153

r2 ¼ r2T þ r2S : ð6:7Þ

The thermal vibration process (*10−13 s) is much slower than the X-ray
absorption process (*10−16 s) and the effect reduces the amplitude of the XAFS
signal. This process is well understood by the Einstein model within a harmonic
approximation and σ2T can be expressed as
 

h h
xE
r2T ¼ coth ; ð6:8Þ
2lxE 2kB T

where ωE is Einstein frequency and μ is reduced mass of the atomic pair, respec-
tively. This equation is quite useful for the harmonic or weak anharmonic systems.
For strong anharmonic systems, the cumulant expansion is valuable. The temper-
ature dependent factor in the EXAFS function with relative displacement of Δα can
be expressed as the following cumulant expansion [10]:
" #
 X
1
ð2ik Þn 
e 2ikDa
¼ exp Dna c : ð6:9Þ
n¼1
n!

The odd terms appear in the phase part and the even terms appear in the
amplitude part of the EXAFS function. The higher order terms include the anhar-
monic contribution in the atomic potential.
On the other hand, the absorption around the absorption edge is mainly attrib-
uted to the transition of core electrons to unoccupied bound states. This feature of
the absorption coefficient spectrum is referred to as X-ray absorption near edge
structure (XANES) or near edge XAFS (NEXAFS). XANES reflects the electronic
structure of the X-ray absorbing atoms and the local structure around them.
Theoretical calculations in this region are quite complicated and interpretation of
the spectrum is still an active research topic.
The intensity of pre-edge peak depends on the symmetry character of the X-ray
absorbing atom. For example, in the K-edge of Ti, V, Cr, Mn, a 1s → 3d peak is
dipole-forbidden for Oh symmetry but dipole-allowed for Td or D4h symmetry. One
can easily check the symmetry by observing the intensity of the pre-edge peak. The
energy position of the threshold of each edge in the absorption spectrum is quite
sensitive to the valence state of the X-ray absorbing atom and approximately lin-
early depends on it. This characteristic is very useful to the mixed valence state and
one can follow the change of the valence in battery materials being charged or
discharged by in situ XAFS measurements.
154 T. Miyanaga and T. Azuhata

6.2.2 Measurements of XAFS

Photon source: Most X-ray absorption spectra are measured using synchrotron
radiation sources because an X-ray beam with high intensity and a continuous
energy spectrum is required. Electrons are forced to go around a storage ring with
the aid of a series of bending magnets and emit polarized pulsed X-rays of high
intensity and a continuous energy spectrum over a wide energy range. In addition to
bending magnets, insertion devices such as a wiggler or undulator are also installed
in modern synchrotron facilities. To collimate and focus the X-ray beam mirrors are
used, followed by slits to adjust the beam size. To select the required X-ray energy,
a double-crystal monochromator is utilized based on the Bragg diffraction condition

nk ¼ 2d sinh; ð6:10Þ

where n, λ, d, and θ are an integer, the X-ray wavelength, the lattice spacing of the
diffracting crystal, and the angle between the beam and the diffracting crystal,
respectively. X-rays of higher harmonics ðn  2Þ included in the incident beam can
be removed by another mirror. The X-ray absorption coefficient can be determined
from the relative intensities of the incident and transmitted X-ray beams, X-ray
fluorescence, or Auger electrons (corresponding to transmission mode), fluores-
cence mode, and electron yield mode, respectively.
Transmission mode: Intensity of incident and transmitted X-ray beams is mea-
sured using ionization chambers filled with inert gas and the absorption coefficient μ
can be obtained from (6.1). For transmission measurements, appropriately con-
centrated samples are needed. Furthermore, the samples are required to be highly
homogeneous and to have constant thickness without pinholes. Thus, if necessary,
samples should be prepared by crushing an appropriate amount of materials and
mixing it with a suitable binder such as BN, polyethylene, or cellulose which are
transparent to X-rays.
Fluorescence mode: Intensity of the characteristic fluorescent X-ray is usually
measured using a simple ionization chamber (Lytle detector) or energy-dispersive
detectors such as a silicon drift detector and a germanium solid state detector. The
intensity of X-ray fluorescence (If in Fig. 6.1) is represented as

I0 li ðEÞef ðX=4pÞ  
 
If ¼    1  exp  lT ðE Þ þ lT Ef d ð6:11Þ
lT ðE Þ þ lT Ef

 lT ðE Þ is the total absorption coefficient of the sample for incident X-ray,
where
lT Ef is the absorption coefficient for the fluorescent X-ray, li ðEÞ is the
absorption coefficient for the element i of interest (essence of XAFS signal), Ω is
the solid angle of the detector, ef is fluorescence efficiency, and d is the sample
thickness. It should be noted that the self-absorption effect must be taken into
account for thick or concentrated samples. The intensity of the fluorescence line
normalized by that of incident X-rays (If/I0) is proportional to li ðEÞ under the
6 Recent Progress in XAFS Study for Semiconducting Thin Films 155

conditions: dilute sample as ca. 0.01 mol dm−1 or thin layer under ca. 100 nm. For
multi-element samples, the characteristic X-ray must be separated from other X-ray
such as lines due to elastic scattering and Compton scattering and other fluores-
cence lines. Si and Ge detectors are capable of extracting the intensity of the
fluorescence X-ray of interest. To improve S/N ratio of the signal, a multi-element
detector is necessary. Fluorescence mode has a great advantage for highly dilute
samples such as thin films deposited on thick substrates.
Electron yield mode: In this mode, the amount of electrons emitted from the
sample surface by X-ray irradiation, such as photoelectrons, secondary electrons,
and Auger electrons, is measured. Relatively short mean free path of the electrons
makes this technique surface sensitive (5–10 nm from the surface) and free from the
self-absorption effect. This mode is useful for XAFS in the low energy (soft X-ray)
region.
Another unique method is XEOL (X-ray Excited Optical Luminescence), in
which the intensity of photoluminescence emitted from the sample provides the
XAFS function. This method is capable of distinguishing the electronic structure or
the local structure around particular chemical species from other species by the
difference in their photon energies (see Chap. 11 in [2]).

6.3 Application to Semiconducting Thin Films

6.3.1 c-Plane InGaN SQWs

InxGa1−xN is a key material in high-brightness blue and green LEDs and

purplish-blue LDs [11]. Although such devices have very high densities of
threading dislocations, they show high quantum efficiency in contrast to conven-
tional III–V and II–VI semiconductor-based devices [12]. The fluctuation of In
atom concentration in InGaN active layers has been proposed as its origin [13]. In
this model, injected carriers are localized into fluctuation-induced potential minima
and can efficiently contribute to radiative recombination before capture by
dislocation-activated non-radiative recombination centers, which is considered to
lead to the high quantum efficiency. Therefore, in order to make the emission
mechanism clear in InGaN-based light-emitting devices, it is important to clarify
the local structure around In atoms in active layers. In this section, we present the
results of EXAFS measurements of In K edge for high quality c-plane (0001)
InxGa1−xN (x = 0.145, 0.20, 0.275) SQW LED wafers by fluorescence detection
using a linearly polarized intense synchrotron X-ray beam and a high performance
detector. Polarization direction of X-ray electric field was set to be both horizontal
and vertical to SQWs. The anisotropy of the local structure around In atoms is
discussed [14].
The samples used in this study are InxGa1−xN (x = 0.145, 0.20, 0.275) SQWs of
3 nm thickness, which were grown by metal organic chemical vapor deposition
156 T. Miyanaga and T. Azuhata

(MOCVD) on sapphire (0001) substrates. Each sample consists of a GaN buffer

layer, an n-GaN:Si barrier layer (4000 nm), an InxGa1−xN SQW (3 nm), a p-
Al0.2Ga0.8N:Mg barrier layer (100 nm), and a p-GaN layer (500 nm). The values of
x were determined using the relation between photoluminescence (PL) emission
energy and x for 3 nm thick InxGa1−xN QWs, the uncertainty of which is estimated
to be ±0.025 [15].
Figure 6.2 shows In K-edge X-ray absorption spectra of the In0.20Ga0.80N SQW
for (a) horizontal (electric field of the X-ray, E, is perpendicular to the c-axis of the
sample) and (b) vertical direction. The spectra were collected on BL10XU,
BL01B1, and BL07B1 at SPring-8, Japan.
The background was subtracted and EXAFS oscillating function χ(k) was
extracted as shown in Fig. 6.3a. In general, χ(k) is multiplied by k or k2, k in this
case. X-ray photon energy was converted into wavenumber of photoelectron as
discussed previously. Figure 6.3b shows radial distribution functions obtained by
Fourier transform of the kχ(k) functions. The peaks around 1.6 Å and 2.8 Å cor-
respond to In–N and In–In/In–Ga atomic pairs, respectively. After the EXAFS

(a) (b)
0.7 0.08

0.6 0.07
t

0.06
t

0.5

0.4 0.05
Horizontal 0.04 Vertical
0.3
27.6 27.8 28 28.2 28.4 28.6 27.6 27.8 28 28.2 28.4 28.6
E / keV E / keV

Fig. 6.2 In K-edge X-ray absorption spectra of the In0.20Ga0.80N SQW for a horizontal (E is
perpendicular to the c-axis) and b vertical direction [14]

(a) Horizontal (b) Horizontal

0.2 Vertical Vertical
0.1
|FT(r)| / arb.unit
-1

0.1
(k) /

0 0.05
k

-0.1

-0.2 0
0 2 4 6 8
5 10 r/
-1
k/

Fig. 6.3 a EXAFS kχ(k) functions and b their Fourier transforms for the In0.20Ga0.80N SQW [14]
6 Recent Progress in XAFS Study for Semiconducting Thin Films 157

functions are fitted to the theoretical curves as (6.4), the structure parameters are
obtained.
As the results, it was found that both rIn–Ga and rIn–In for six out-of-plane atoms
are larger than those for six in-plane atoms, suggesting that the InGaN SQWs are
biaxially compressed in the ab plane by the effect of adjacent layers. Because rIn–N
is almost constant for both directions, the angle of In–N–Ga/In in the horizontal
direction is found to be smaller than that in the vertical direction. It should be noted
that the values of rIn–In in our particular samples are almost independent of x and
close to that of metallic indium 3.2–3.3 Å which is predicted in zinc-blende InGaN
and is also expected in wurtzite InGaN to cause strong coupling of valence states by
Kent and Zunger [16].
The local InN molar fraction around In atoms, y, can be expressed using effective
coordination numbers as

  

y ¼ NInIn = NInIn þ NInGa : ð6:12Þ

This is essentially different from the average InN molar fraction x obtained from
PL emission energy [15], since EXAFS is sensitive only to the local structure
around specific atoms, In in this case. It should be noted that the value of the
denominator is 9.0 (12.0) for in-plane (out-of-plane) molar fraction.
In Fig. 6.4, the values of y/x are plotted as a function of x for the three InxGa1−xN
SQW samples. The In local structure can be classified into three categories: (1) If
y = x, In atoms are randomly distributed; (2) if y ≤ x, In atoms are isolated and
superlattice structures exist; (3) if y > x, In atoms are aggregated with each other and
separated from Ga atoms. As can be seen in Fig. 6.4, it is found that the In atom
distribution is statistically random in the horizontal direction in all the samples.
Note that the accidental aggregation of In atoms is possible even in statistically
random alloys which are not necessarily homogeneous. On the other hand, In atoms
in the vertical direction for x = 0.20 are found to be aggregated with each other and
are located at the top and bottom, that is, In atoms tend to be just above and/or just

1.8

1.6
y/x

1.4

1.2

0.8
0.1 0.15 0.2 0.25 0.3
x

Fig. 6.4 The plot of y/x with x for the three InGaN SQWs. The schematic model for In atom
distribution is also shown (see text)
158 T. Miyanaga and T. Azuhata

below In atoms. Although higher experimental accuracy is necessary to make a

decision, such vertical aggregation is likely to exist also in the In0.145Ga0.855N SQW
because the plot is above the solid line corresponding to y = x. For x = 0.275, y in
the vertical direction is almost equal to x, indicating random distribution. The
average InN molar fractions x = 0.145, 0.20, and 0.275 correspond to blue, green,
and amber LEDs, respectively. According to the emission energy dependence of the
external quantum efficiency for InGaN LEDs by Mukai et al. [17], InGaN-based
blue to green LEDs show higher external quantum efficiency. Thus, the present
EXAFS results suggest that the aggregation of In atoms in the vertical direction in
active layers correlates with the higher quantum efficiency of blue to green InGaN
LEDs.

6.3.2 c-Plane InGaN MQW

The local structure around In atoms in the InxGa1−xN (x = 0.145, 0.20, 0.275)
SQW LED wafers were discussed in the previous section: In atoms are aggregated
and located at the top and bottom in the vertical direction of the SQW planes. The
In–In in-plane distance is smaller than the out-of-plane distance. This means that
the SQWs of InGaN are biaxially compressed in the ab plane by the effect of
adjacent layers.
On the other hand, MQWs are used in LDs to improve device performance such
as threshold current density and optical gain. In this section, we discuss the local
structure around In atoms in a c-plane InxGa1−xN MQW, where the sample surface
is set to be horizontal and vertical to the X-ray polarization electric vector [18]. The
results are discussed and compared with those for the previous In0.2Ga0.8N SQW
study [14].
The sample was an MQW structure of 10 periods of In0.2Ga0.8N (2.5 nm) and
In0.05Ga0.95N (7.5 nm), grown by MOCVD on a sapphire (0001) substrate [19].
X-ray absorption measurements were carried out at BL10XU of SPring-8.
Figure 6.5 shows (a) the In K-edge EXAFS kχ(k) functions and (b) Fourier
transforms of the InxGa1−xN MQW for horizontal (solid line) and vertical (dashed
line) polarization directions. The spectral difference between these two directions is
relatively small. The EXAFS amplitude for the MQW is larger than that for the
SQW. The range for the Fourier transform is 2.7–13.3 Å−1. The 1st peak at 1.5 Å in
Fig. 6.5b corresponds to the In–N atomic pair, while the 2nd peak at 2.8 A cor-
responds to the In–In and In–Ga atomic pairs. The Debye-Waller factors for the In–
In (around 0.04 Å) and In–Ga (around 0.06 Å) atomic pairs in the MQW are smaller
than those in the SQW (0.08 Å and 0.07 Å, respectively) [14]. In particular, the
difference for the In–In pairs is prominent. This result corresponds to the difference
in the intensities of the 2nd peaks in Fig. 6.5b.
In the case of the SQW, the In–In proportion (or coordination number, N) is
higher for the vertical direction (N = 4.0) than for horizontal direction (N = 2.1).
This indicates that the In atom is aggregated in the vertical direction [2]. On the
6 Recent Progress in XAFS Study for Semiconducting Thin Films 159

Fig. 6.5 a In K-edge kχ(k) and b Fourier transform of the InxGa1−xN MQW (xav = 0.088) for
horizontal (solid line) and vertical (dashed line) directions [18]

other hand, in the present InGaN MQW, N for In–In is the same for the horizontal
and vertical direction (N = 1.2). This result indicates that the nearest-neighbor In
atom is isotropically distributed around the absorber In atoms in the InGaN MQW.
The average In concentration is 0.088 in this InGaN MQW. Therefore, N is
expected to be 1.1 if In atoms are randomly distributed and the value of N (=1.2) is
quite close to this value. The interatomic distances (In–In 3.29 Å and In–Ga 3.27 Å)
in the vertical direction is longer than that in the horizontal direction (In–In 3.23 Å
and In–Ga 3.21 Å) for the SQW, where the InGaN layer is biaxially compressed in
the ab plane by the effect of the adjacent layers. On the other hand, those in both
directions are the same for the InGaN MQW (In–In 3.25 Å and In–Ga 3.24 Å).
Judging from these results, it is concluded that the strain in the InxGa1−xN MQW is
reduced compared to that in the SQW [14].

6.3.3 m-Plane InGaN Thin Film

As discussed in the previous section, in the c-plane InxGa1−xN QWs, the localized
exciton emission exhibits high internal quantum efficiency despite the presence of
high-density threading dislocations. The internal quantum efficiency, however,
decreases abruptly for the emission wavelength longer than around 520 nm (InN
molar fraction x < 0.2) due to the quantum-confined Stark effects (QCSEs). The
effects are caused by the electrostatic field perpendicular to QWs due to the
spontaneous and piezoelectric polarization [13, 15, 20]. In contrast to polar c-plane
QWs, nonpolar m-plane ð1
100Þ and a-plane ð11
20Þ QWs can avoid the QCSEs
because the polar c-axis is on the QW plane. Thus, nonpolar InGaN epilayers are
attracting much attention to achieve higher internal quantum efficiency. The
structural study for nonpolar InGaN films is important to reveal the emission
mechanism. However, there is little research in this area presently. Recently,
Chichibu et al. reported m-plane InxGa1−xN (x < 0.14) thin films grown on
160 T. Miyanaga and T. Azuhata

Fig. 6.6 Atomic structure and crystal axes for the model of the m-plane In0.06Ga0.94N. Black
circles represent Ga or In atoms and gray ones N atoms. The atom labelled “X” is the X-ray
absorbing In atom and the numbers of “1”, “2”, and “3” show three types of atomic positions

freestanding m-plane GaN substrates. The films were confirmed to be coherently

grown by an X-ray reciprocal space mapping (XRSM) method [21], but the local
structures around In atoms are still unknown. In this section, we present the results
of In K-edge EXAFS analyses for one of the m-plane InGaN films, an In0.06Ga0.94N
film. The following three configurations were used: the electric field vector of the
incident X-ray is parallel to the c-axis [0001], the m-axis ½1
100 , and the a-axis
½11
20 . A step-by-step procedure is introduced here to analyze the EXAFS data for
the three different directions individually [22].
The In0.06Ga0.94N film was deposited by MOVPE on a 1.5-μm-thick GaN buffer
layer on a 325-μm-thick GaN substrate prepared by a halide vapor phase exitaxy
method [23]. X-ray absorption measurements were made at a beam line of NW10A
of Photon Factory Advanced Ring, KEK. XAFS data were collected with a
double-crystal monochromator of Si (311) crystals. Indium Kα-fluorescence emis-
sion was measured using a 19-element Ge solid-state detector.
Figure 6.6 shows the atomic structure and the crystal axes of the wurtzite
structure. The numbers of “1”, “2”, and “3” show three types of atomic positions.
For the analyses of the local structure for the second nearest neighbor (2NN) In–
Ga and In–In, the following step-by-step procedure was used. Firstly, the E ‖ c data
were analyzed (E is the electric vector of the X-ray). In this configuration, only the
type “1” atoms are detected by XAFS. Their effective coordination number N*1 was
set to be 12. The type “2” and “3” atoms are not detected because the plane
including these types of atoms and the X-ray absorbing In atom (labelled “X” in
Fig. 6.6) is perpendicular to E. Thus, N* for these types of atoms are N*2 = N*3 = 0.
Strictly speaking, these six atoms labelled “1” are not equivalent for the strained m-
plane wurtzite structure. However, it is quite difficult to analyze them separately
because of too many fitting parameters. Thus, XAFS spectra were analyzed by
assuming that the type “1” six Ga/In are equivalent.
6 Recent Progress in XAFS Study for Semiconducting Thin Films 161

Secondly, the E || m data, to which only the type “1” and “2” atoms contribute,
were analyzed. Although σ may depend on the polarization direction of the incident
X-ray, it was assumed to be isotropic in this study. For r, σ, and the ratio of N*In–Ga/
N*In–In for type “1” atoms, the values determined from the above E || c data fitting
were used. N*1and N*2 were fixed to be 3 and 9, respectively. At this step, r, σ, and
N*In–Ga/NIn–In
*
for type “2” atoms were determined.
Finally, the E || a data were analyzed. In this configuration, all types of atoms
contribute to the XAFS signal. The values of r, σ, and N*In–Ga/NIn–In
*
for type “1” and
“2” atoms, which were obtained from the above two steps, were used. N*1and N*2,
and N*3 were set to be 3, 3, and 6, respectively.
We describe only preliminary results here. From the coordination numbers, In
atoms were shown to be localized in all directions. The In–Ga distance in the
c-direction was found to be longer than that in strain-free virtual crystals in spite of
the compressive strain along the a- and c-axes, which can be explained by the
partial strain-cancellation mechanism due to the anisotropic atomic structure on the
m-plane. It was also found that the local strain in coherently grown m-plane InGaN
is more relaxed than that in coherently grown c-plane InGaN.

6.3.4 m-Plane AlGaN Thin Films

Group III-nitride semiconductors have attracted much attention as materials for blue
and ultraviolet light emitters [24]. The most studied member of this group, wurtzite
GaN, has a direct band gap of 3.42 eV, while pseudo-binary alloys, AlxGa1−xN,
provide a direct gap up to 6.0 eV. It is attractive that it covers the wide range of
ultraviolet rays. There are several XAFS studies of Ga K-edge for AlxGa1−xN
(x = 0–0.4) [25] and (x = 0–0.6) [26] films, where the interatomic distances and the
coordination numbers of Ga–Al and Ga–Ga pairs are discussed. On the other hand,
quite a few Al K-edge XAFS studies have been reported for AlGaN films. Recently,
it was pointed out that the problem is the reduction in oscillator strength of
electron-hole pairs in quantum wells due to the QCSEs caused by the spontaneous
and piezoelectric polarization discontinuity at c-plane heterointerfaces. To avoid
these effects in AlGaN heterostructures, epitaxial growth in nonpolar orientation
such as m- or a-plane has been attracting attention as shown in the previous section.
However, the local structure and the basic physical properties of nonpolar AlGaN
films are not fully understood.
In this section, the polarized XAFS studies for Al K-edge of m-plane AlxGa1−xN
thin films are presented. In particular, the interatomic distance of 2NN Al–Al and
Al–Ga atomic pairs and the localization of Al atoms in m-plane AlxGa1−xN are
discussed [27].
140-nm-thick m-plane AlxGa1−xN films (x = 0.32 and 0.58) were grown by the
NH3 source molecular beam epitaxy method on m-plane GaN substrates. It is
reported that the Al0.32Ga0.68N film is pseudomorphic but the Al0.58Ga0.42N film is
162 T. Miyanaga and T. Azuhata

a-Al0.32 Ga0.68 N c-Al0.32 Ga0.68 N a-Al0.58 Ga0.42 N c-Al0.58 Ga0.42 N

m-Al0.32 Ga0.68 N m-Al0.58 Ga0.42 N
(a) 2 (b) 2

1 1

(k)
(k)

0 0

k2
k2

-1 -1

-2 -2
2 4 6 8 2 4 6 8
-1 -1
k/ k/

Fig. 6.7 EXAFS k2χ(k) functions for Al K-edge of a the m-plane Al0.32Ga0.68N film and b the
Al0.58Ga0.42N film [27]

partially or nearly fully relaxed [28]. Thus, it is important to study the local
structure around Al atoms in the films by EXAFS.
Al K-edge (1560 eV) XAFS was measured on BL11A at Photon Factory, KEK,
in fluorescence mode using a silicon drift detector. To measure the polarization
dependent XAFS, we set the sample in three configurations with the electric field
vector of X-ray parallel to the (1) a-axis ½11
20 , (2) c-axis [0001], and (3) m-axis
½1
100 .
Figure 6.7 shows EXAFS k2χ(k) spectra of Al K-edge for the m-plane
(a) Al0.32Ga0.68N and (b) Al0.58Ga0.42N films, where “a-”, “c-”, and “m-” denote
that the polarization direction of the incident X-ray is parallel to the a-, c-, and
m-axes, respectively. The data quality is acceptable.
Figures 6.8a, b show Fourier transforms for Al K-edge of the m-plane
Al0.32Ga0.68N and Al0.58Ga0.42N films, respectively. The k-range for the Fourier
transforms is 2.1–7.75 Å−1. The authors analyzed polarization dependent XAFS
data individually.
Although in general it is not so easy to analyze Al K-edge EXAFS because of the
low energy of the K-edge and the narrow energy range of EXAFS region, the
EXAFS curve fitting analysis was performed. The step-by-step procedure has been
proposed for more precise polarization analysis in the previous section, however
only the simple two-shell fitting are applied for Al–Al and Al–Ga pairs in each
direction for this systems. The structural parameters obtained from the curve-fitting
analysis are listed in Table 6.1.
For both the films, the interatomic distances of Al–Al and Al–Ga for the
a-axis-polarized spectra are about 3.17 ± 0.02 Å, which is close to Ga–Ga distance
in GaN. On the other hand, those for the m-axis spectra are about 3.10 ± 0.02 Å,
which is close to Al–Al distance in AlN. It is suggested that local structures along
the a-axis of the films are strongly affected by GaN substrates. This result is similar
to that for m-plane In0.06Ga0.94N XAFS, but in that case the 2NN interatomic
6 Recent Progress in XAFS Study for Semiconducting Thin Films 163

1
(a) 1 a-Al 0.32 Ga0.68 N (b) a-Al 0.58 Ga0.42 N
c-Al 0.32 Ga0.68 N c-Al 0.58 Ga0.42 N
0.8 0.8 m-Al 0.58 Ga0.42 N
m-Al 0.32 Ga0.68 N

0.6 0.6
|FT|

|FT|
0.4 0.4

0.2 0.2

0 0
0 2 4 6 8 0 2 4 6 8
r/ r/

Fig. 6.8 Fourier transforms for Al K-edge of a the m-plane Al0.32Ga0.68N film and b the
Al0.58Ga0.42N film [27]

Table 6.1 Structural rAl–Al/Å rAl–Ga/Å NAl/(NAl + NGa)

parameters for the AlxGa1−xN
films obtained from EXAFS a-Al0.32Ga0.68N 3.17 3.17 0.63
curve fitting [27] c-Al0.32Ga0.68N 3.11 3.18 0.67
m-Al0.32Ga0.68N 3.09 3.10 0.67
a-Al0.58Ga0.42N 3.17 3.18 0.58
c-Al0.58Ga0.42N 3.10 3.11 0.67
m-Al0.58Ga0.42N 3.11 3.12 0.59

distance (In–Ga) is shortened by the effect of GaN substrates. The interatomic

distances of Al–Ga for the c-axis spectra are 3.18 ± 0.02 Å for Al0.32Ga0.68N and
3.11 ± 0.02 Å for Al0.58Ga0.42N, respectively. This difference may be caused by the
fact that the Al0.32Ga0.68N film is pseudomorphic but that the Al0.58Ga0.42N film is
partially or nearly fully relaxed. Yu et al. [26] studied Ga K-edge XAFS for AlxGa1
−xN (x = 0–0.6) films and reported that the interatomic distance of Ga–Al is almost
constant to be 3.10 Å and that of Ga–Ga changes linearly from 3.18 Å for GaN to
3.14 Å for Al0.6Ga0.4N. The present polarization dependent analysis shows that the
interatomic distances of Al–Al and Al–Ga for the m-axis spectra are close to their
Ga–Al distance (3.10 Å) and those for the a-axis spectra are close to their Ga–Ga
distance for GaN (3.18 Å). The reason of these discrepancies may be that the
present samples were grown on m-plane GaN substrates, while their ones on
c-plane sapphire substrates.
The proportion of the 2NN Al atoms around Al atoms in the films is evaluated as
NAl/(NAl + NGa). Those for the Al0.32Ga0.68N and Al0.58Ga0.42N films are around
0.63 ± 0.05 for each direction, suggesting that Al atoms are almost randomly
distributed in the Al0.58Ga0.42N film but are localized in the Al0.32Ga0.68N film
within the simple model. Polarization dependence of the localization of Al atoms
was not observed. Yu et al. [26] reported that Ga and Al atoms are randomly
distributed in the concentration range of x = 0–0.6. It is suggested that Al atoms are
164 T. Miyanaga and T. Azuhata

localized in the Al0.32Ga0.68N film. Hazu et al. have studied spatio-time-resolved

cathode luminescence for an m-plane Al0.25Ga0.75N film and suggested the
localization of Al atoms [29].

6.3.5 c-Plane MgZnO Thin Film

MgxZn1−xO has attracted a lot of attention as a useful semiconducting material,

because the band gap can be controlled by Mg composition [30, 31]. This is
expected to be a new LED and/or LD material in place of InGaN for marketing
reasons. It is known that, after annealing, MgxZn1−xO films with x > 0.18
decompose into a rock salt and a wurtzite phase, while the sample with x = 0.12
remains as a single phase of the wurtzite, in which the c-parameter in the wurtzite
phase approximately follows Vegard’s law [32]. In such a mixed crystal semi-
conductor as MgxZn1−xO, the local structure around each atom is important for the
study of the light emitting mechanism.The local structure around the low concen-
tration Mg atom is actually more attractive, since the interatomic distance does not
simply correspond to Vegard’s law. In this section, the Zn K-edge and Mg K-edge
XAFS results for c-plane Mg0.06Zn0.94O film are discussed [33].
The sample of Mg0.06Zn0.94O film fabricated on ð11
20Þ Al2O3 substrate by
Helicon Wave excited Plasma Sputtering Epitaxy (HWPSE) method. For this
technique, the plasma chamber is separated from the sputtering chamber in order to
avoid the re-sputtering and surface damage caused by the high energy sputtering
particles. The thickness of the Mg0.06Zn0.94O layer is 780 nm on the buffer layer of
AlN and Al2O3 substrate. The detailed sample preparation is described elsewhere
[30]. XAFS measurements were performed at BL7C for Zn K-edge and BL11A for
Mg K-edge in Photon Factory (KEK, Tsukuba) [33]. The fluorescence XAFS
spectra were measured using a Lytle detector for Zn K-edge and a silicon drift
detector for Mg K-edge. For the wurtzite anisotropy of this sample, the measure-
ments were performed by polarized XAFS: (1) the electric field of the incident
X-ray is horizontal and (2) the electric field is vertical to the sample plane of the
Mg0.06Zn0.94O film.
A. Results of Zn K-edge analyses
First, we describe the analytical procedure of Zn K-edge EXAFS for the
Mg0.06Zn0.94O film, which is not too thin for the condition that the fluorescence
limit is applicable. Figure 6.9 shows the schematic measurement set up for the
sample direction and X-ray beam. To give accurate XAFS spectra by the fluorescent
method, two conditions are generally required; (1) thick sample and dilute target
element or (2) thin sample and dense target element as shown in the previous
section. In this sample, the condition for the measurement of Zn K-edge corre-
sponds to the latter (2). However, unfortunately, the effective thickness of the
vertical measurement is about 15 μm and out of the condition. Hence the vertical
6 Recent Progress in XAFS Study for Semiconducting Thin Films 165

Fig. 6.9 The schematic measurement set up for the sample direction and X-ray beam. a The
horizontal direction between the electric field of X-ray and the sample plane, b the vertical
direction between them, and c 45° between them [33]

(a) (b)
10 horizontal 10
corrected
vertical
|FT|

|FT|

5 5

0 0
0 2 4 6 8 10 0 2 4 6 8 10
r( ) r( )

Fig. 6.10 Fourier transform of Zn K-edge for a the horizontal and b the vertical direction of the
sample plane of Mg0.06Zn0.94O. In b, the corrected vertical spectrum is also shown [33]

measurement does not give a correct spectrum and the amplitude of the signal is
reduced. The spectrum obtained with 45° between the sample plane and the X-ray
beam (Fig. 6.9c) contains the equal amount of information of the horizontal and
vertical. So the horizontal signal was subtracted from 45° signal to obtain the
corrected vertical spectrum. Figure 6.10 shows the example of the Fourier trans-
forms for the comparison with the horizontal and 45° spectra.
Next, we discuss the results of Zn K-edge analyses for the Mg0.06Zn0.94O film. In
Fig. 6.10b, the intensity of the Fourier transform for the corrected vertical data seems
to be improved compared with the horizontal data. The first nearest peak corresponds
to a Zn–O atomic pair and the second one to Zn–Zn and Zn–Mg atomic pairs. The
contributions from Zn–Zn and Zn–Mg cannot be separated by the Fourier transform
166 T. Miyanaga and T. Azuhata

method. The non-linear-least-square-fitting method was applied to these two nearest

peaks. In the analyses for the second nearest peak, the contribution from Mg atoms
cannot be recognized, due to the small amount of Mg atoms in Mg0.06Zn0.94O. The
coordination numbers for the corrected vertical data are well improved to be close to
the wurtzite structure. From these structural analyses for Zn K-edge EXAFS, the 1st
nearest Zn–O distance is 1.963 Å, that of 2nd nearest Zn–Zn is 3.249 Å, and 4th
nearest Zn–O is 3.861 Å. The local structure of the present Mg0.06Zn0.94O layer is
quite close to that of the bulk ZnO crystal [34].
B. Results of Mg K-edge analyses
In this subsection we discuss the result of the Mg K-edge XAFS analyses for the
Mg0.06Zn0.94O layer. Figure 6.11a shows the X-ray absorption spectra of
Mg K-edge for Mg0.06Zn0.94O for the horizontal direction (in plane) and the vertical
direction (the sample is inclined about 6° to the c-axis). High quality data are
obtained up to 1600 eV. Figures 6.11b, c show the EXAFS kχ(k) oscillation
functions and their Fourier transforms for the horizontal (in plane) and the vertical
(c-axis) direction, respectively. A clear polarization dependence is observed also in
the Mg K-edge XAFS.

(a) (b) 2
in plane
in plane c-axis
2 c-axis
1
k (k)
µ(E)

1
0

0
1300 1400 1500 2 4 6 8
-1
E/eV k/
(c) 0.3
in plane
c-axis
Mg-O M
Mg-Zn/ Mg
0.2
|FT|

0.1

0
0 2 4 6 8 10
r/

Fig. 6.11 a X-ray absorption spectra, b kχ(k) and c Fourier transforms of Mg K-edge for
Mg0.06Zn0.94O for the horizontal direction (in plane, red) and the vertical direction (the sample is
inclined at about 6° to the c-axis, green) [33]
6 Recent Progress in XAFS Study for Semiconducting Thin Films 167

(a) (b)
2 2

µ(E)
µ(E)

1 1

0 0
1310 1315 1320 1325 1330 1310 1315 1320 1325 1330
E /eV E /eV

Fig. 6.12 Calculated XANES spectra with measured one (EXP) for a horizontal and b vertical
directions. Mg0, Mg1, … show the number of Mg atoms around an X-ray absorption Mg atom [33]

The data range of kχ(k) is limited to 8.2 Å−1 (in Fig. 6.11b), but the first and
second peaks are well distinguished in the Fourier transform (in (c)). The first
nearest peak corresponds to the Mg–O atomic pair and the second one does to the
Mg–Zn and Mg–Mg pairs. Each contribution of Mg–Zn and Mg–Mg cannot be
separated in FT as is the case of previous Zn K-edge EXAFS.
Although rather good quality XAFS data were obtained, the parameter fitting by
non-linear-least-square method is not so easy, so we just compare the simulation
spectra calculated by FEFF code with the experimental data. Figures 6.12 shows
measured and calculated XANES for (a) horizontal and (b) vertical direction,
respectively. For the model calculation, the atomic positions are fixed at the
coordinates considered from the EXAFS result previously discussed and only the
number of Mg atoms is changed as a parameter (the Zn atom in the 2NN position is
replaced by Mg atom.) In Fig. 6.12a, the peaks at 1295, 1313, and 1323 eV are well
reproduced by FEFF calculation. The peaks at 1313 and 1323 eV are sensitive to
the number of surrounding Mg atoms; these intensities change and the positions are
shifted to higher energy with increasing the number of Mg atoms. The energy shift
is pronounced in the case that the number of Mg atoms is larger than 4. Similar
features are observed in Fig. 6.12b: the peak at 1317 eV is sensitive to increasing
Mg atoms and slightly shifts to higher energy, whereas peaks at 1312 and 1323 eV
are decreasing. Judging from these simulation results, the number of surrounding
Mg atoms in the second nearest shell from the X-ray absorbing Mg atom may be
0–3. Actually, for the limited non-linear-least-square-fitting in which the coordi-
nation number of Mg is fixed at the values of 0, 1, 2, … 6, better fitting results were
obtained for the model where less Mg atoms located in the 2NN positions.
It was concluded that Mg atoms are distributed randomly in the present
Mg0.06Zn0.94O layer. This result contrasts with the case of 3-nm-thick InxGa1−xN
(x = 0.145, 0.20, 0.275) SQWs as previously discussed [4]. In the case of InGaN,
the polarization dependence is remarkable and In atoms are aggregated in the
vertical direction of In0.2Ga0.8N but randomly distributed in the horizontal
168 T. Miyanaga and T. Azuhata

direction. Recently, Shibata et al. studied the photoluminescence characterization of

MgxZn1−xO (x = 0.05, 0.11, 0.15) epitaxial films [31] and presented that the degree
of the localization or the compositional fluctuation of Mg atoms increases with
increasing the concentration of Mg.
According to the structural parameters obtained from the
non-linear-least-square-fitting under the limited condition that no Mg atom exists in
the 2NN from X-ray absorbing Mg atoms (not shown here), for the 1NN distance of
Mg–O, Mg–O (horizontal) is 1.957 Å and Mg–O (vertical) is 1.983 Å. Thus, the
local structure in the Mg0.06Zn0.94O layer is almost the same as the bulk ZnO
structure. For the 2NN, on the other hand, the atomic distances of Mg–Zn (3.283 Å
for horizontal and 3.262 Å for vertical) are slightly longer than those of Zn–Zn
(3.249 Å for horizontal and 3.220 Å for vertical from the present EXAFS study
[33]) in the Mg0.06Zn0.94O layer and those in bulk ZnO (3.239 Å for horizontal and
3.208 Å for vertical from [34]).

6.4 Summary

In this chapter we have provided several examples of XAFS application to the local
structure analysis of semiconducting thin films: InGaN, AlGaN, and MgZnO. For
such thin films, it is not so easy to study the structure of active layers by con-
ventional methods. XAFS can provide unique information on atomic distribution
and electronic structure of the small area in the active layer of thin films. In general,
XAFS is considered as a method to give isotropic structural information, but it is
capable of analyzing anisotropic structure in epitaxial films by use of
linearly-polarized X-ray from synchrotron radiation sources. While the XAFS
analysis in low and medium energy region (soft and tender X-ray) is not so accurate
in comparison with high energy region (hard X-ray), recent development of the
high performance 3rd and higher generation synchrotron radiation source provides
an advantage in lower energy XAFS studies.

Acknowledgments The authors gratefully acknowledge Prof. S.F. Chichibu and Dr. K. Nitta for
significant contributions and fruitful collaborations. The authors thank the Photon Factory in KEK,
and JASRI in SPring-8 for financial and experimental support.

References

1. G. Bunker, Introduction to XAFS (Cambridge University Press, Cambridge, 2010)

2. C.S. Schnohr, M.C. Ridgway, X-ray Absorption Spectroscopy of Semiconductors (Springer,
Heidelberg, 2015)
3. P. Fornasini, Synchrotron Radiation: Basics, Methods and Applications, eds. by S. Mobilio,
F. Boscherini, C. Meneghini, (Springer, Heidelberg, 2015) Chap. 6
4. D. Sayers, E.A. Stern, F. Lytle, Phys. Rev. Lett. 27, 1204 (1971)
6 Recent Progress in XAFS Study for Semiconducting Thin Films 169

5. http://www.exafsco.com/techpapers/index.html
6. J. Als-Nielsen, D. McMorrow, Elements of Modern X-ray Physics, (Wiley, Chichester, 2011)
7. H. Sakane, T. Miyanaga, I. Watanabe, N. Matsubayashi, S. Ikeda, Y. Yokoyama, Jpn. J. Appl.
Phys. 32, 4641 (1993)
8. http://cars.uchicago.edu/ifeffit/Downloads
9. A.L. Ankudinov, J.J. Rehr, Phys. Rev. B 62, 2437 (2000)
10. T. Miyanaga, T. Fujikawa, J. Spectrosc. Dyn. 1, 4 (2011)
11. S. Nakamura, G. Fasol, The Blue Laser Diode (Springer, Heidelberg, 1997)
12. S.F. Chichibu, Y. Kawakami, T. Sota, in Introduction to Nitride Semiconductor Blue Lasers
and Light Emitting Diodes, eds. by S. Nakamura, S.F. Chichibu (Taylor & Francis, London,
2000), Chap. 5
13. S. Chichibu, T. Azuhata, T. Sota, S. Nakamura, Appl. Phys. Lett. 69, 4188 (1996)
14. T. Miyanaga, T. Azuhata, S. Matsuda, Y. Ishikawa, S. Sasaki, T. Uruga, H. Tanida, S.F.
Chichibu, T. Sota, Phys. Rev. B 76, 035314 (2007)
15. S.F. Chichibu, T. Sota, K. Wada, O. Brandt, K.H. Ploog, S.P. DenBaars, S. Nakamura, Phys.
Status Solidi A 183, 91 (2001)
16. P.R.C. Kent, A. Zunger, Appl. Phys. Lett. 79, 1977 (2001)
17. T. Mukai, M. Yamada, S. Nakamura, Jpn. J. Appl. Phys., Part 1 38, 3976 (1999)
18. S. Sasaki, T. Miyanaga, T. Azuhata, T. Uruga, H. Tanida, S.F. Chichibu, T. Sota, A.I.P. Conf,
Proc. 882, 499 (2008)
19. S.F. Chichibu, A.C. Abare, M.P. Mack, M.S. Minsky, T. Deguchi, D. Cohen, P. Kozoboy,
S.B. Fleischer, S. Keller, J.S. Speck, J.E. Bowers, E. Hu, U.K. Mishra, L.A. Coldren,
S.P. DenBaars, K. Wada, T. Sota, S. Nakamura, Mater. Sci. Eng., B 59, 298 (1999)
20. T. Takeuchi, S. Sota, M. Katsuragawa, M. Komori, H. Takeuchi, H. Amano, I. Akasaki, Jpn.
J. Appl. Phys. 36, L382 (1997)
21. S.F. Chichibu, H. Yamaguchi, L. Zhao, M. Kubota, T. Onuma, K. Okamoto, H. Ohta, Appl.
Phys. Lett. 93, 151908 (2008)
22. T. Miyanaga, T. Azuhata, K. Nitta, S.F. Chichibu, unpublished
23. T. Koyama, T. Onuma, H. Masui, A. Chakraborty, B.A. Haskell, S. Keller, U.K. Mishra, J.S.
Speck, S. Nakamura, S.P. DenBaars, T. Sota, S.F. Chichibu, Appl. Phys. Lett. 89, 091906 (2006)
24. F.H. Morkoc, S.N. Mohammad, Science 267, 51 (1995)
25. K.E. Miyano, J.C. Woicik, L.H. Robins, C.E. Bouldin, D.K. Wickenden, Appl. Phys. Lett. 70,
2108 (1997)
26. K.M. Yu, C.J. Glover, M.C. Ridgway, W.S. Wong, W. Yang, Appl. Phys. Lett. 75, 4097 (1999)
27. T. Miyanaga, T. Azuhata, K. Nakajima, H. Nagoya, K. Hazu, S.F. Chichibu, J. Phys: Conf.
Ser. 502, 012031 (2014)
28. K. Hazu, T. Hoshi, M. Kagaya, T. Onuma, S.F. Chichibu, J. Appl. Phys. 107, 033701 (2010)
29. K. Hazu, M. Kagaya, T. Hoshi, T. Onuma, S.F. Chichibu, J. Vac. Sci. Technol., B 29, 21208
(2011)
30. T. Koyama, T. Ohmori, N. Shibata, T. Onuma, S.F. Chichibu, J. Vac. Sci. Technol., B 22,
2220 (2002)
31. H. Shibata, H. Tampo, K. Matsubara, A. Yamada, K. Sakurai, S. Ishizuka, S. Niki, Appl.
Phys. Lett. 90, 124104 (2007)
32. H. Ryoken, N. Ohashi, I. Sakaguchi, Y. Adachi, S. Hishita, H. Haneda, J. Cryst. Growth 287,
134 (2006)
33. T. Yamada, T. Miyanaga, T. Azuhata, T. Koyama, S.F. Chichibu, Y. Kitajima, e-J. Surf. Sci.
Nanotech. 7, 596 (2009)
34. H. Karzel, W. Potzel, M. Köfferlein, W. Schiessl, M. Steiner, U. Hiller, G.M. Kalvius, D.W.
Mitchell, T.P. Das, P. Blaha, K. Schwarz, M.P. Pasternak, Phys. Rev. B 53, 1425 (1996)
Chapter 7
Pulsed-Laser Generation
of Nanostructures

Selcuk Akturk

Abstract We discussed two main approaches for generation of nanometer-scale

structures using pulsed lasers. In the first, a bulk material is ablated in vacuum, gas
or liquid and consequent re-solidification of the evaporated material in clusters
generate the desired nanoparticles. While one can play with various experimental
parameters (laser powers, type of liquids, background gas etc.) to control particle
sizes, there is always a relatively broad distribution of particle sizes. Nevertheless,
particles with sizes down to few nanometers are achievable. In the second approach,
fs lasers are used to pattern material surfaces. This process is more controlled in the
sense that the desired geometries are generated by scanning of the sample. On the
other hand, best achievable resolutions are worse as compared to the former
approach. Resolutions better than 100 nm over large surfaces is very difficult.
Finally, special beam shaping methods forms an interesting frontier in pushing
forward the fs laser nanomachining methods towards broader practical uses.

7.1 Introduction

Ablation of materials by pulsed lasers constitute a versatile approach to generation

of nanometer-scale structures. In this chapter, we will discuss two distinct
approaches for pulsed-laser generation of nanostructures. In the first case, laser
pulses are sent to the surface of the target material in bulk form. The evaporated
material then solidifies in the form of clusters of various sizes, usually in
sub-micron range. The cluster formation can take place in liquid or gas environ-
ment. In the second approach that we will discuss, the process is more “controlled”.
Here, ultrashort laser pulses (with femtosecond pulse durations) are focused to
small spots and by working near the ablation-threshold, volumes with sub-micron
dimensions are removed. Nanostructures of desired forms and geometries are

S. Akturk (&)
Department of Physics, Istanbul Technical University, 34469 Maslak, Istanbul, Turkey
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 171

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_7
172 S. Akturk

generated by scanning the sample under laser illumination. We will describe the
first approach (which is a relatively more mature field, and many good reviews
already exist) in a brief overview manner, while we will elaborate more on the
second, more recently emerging case.

7.2 Self-formation of Nanostructures Through

Pulsed-Laser Ablation

Laser-assisted formation of nanostructures has a history of nearly three decades. In

one of the earlier works, continuous-wave laser are used to manipulate the size
distribution of metal clusters [1]. The physical mechanism of size change is basi-
cally absorption of laser energy by the particles, consequent heating and division.
The method is thus suitable for processing of nanoparticles, already prepared by
other means. In 1997, Morales and Lieber demonstrated the use of pulsed laser
ablation in a gas environment to generate semiconductor nanowires [2]. Cluster
formation and vapor-liquid-solid (VLS) growth yields the wire-type structures.
Following years has seen an expanding use of laser ablation to generate nanopar-
ticles of various materials and geometries.
Pulsed-laser self-formation of nanoparticles is mainly performed in gas/vacuum
or liquid environments. Experimental principles of both cases is outlined in Fig. 7.1.
In the first case, the target material is ablated with laser pulses and the evaporated
material forms clusters on a stopping target. Physical mechanisms of particles
formation depends strongly on the nature and the conditions of the environment [3].

Fig. 7.1 Schematic representation of laser-assisted nanoparticle generation in gas/vacuum (left)

and liquid (right) environments
7 Pulsed-Laser Generation of Nanostructures 173

Nanoparticles may form in the gas phase or from nucleation on the target substrate.
An extensive review of methods and applications for the gas-phase case is collected
by Kruis et al. [4]. Particles sizes are typically below 100 nm. The necessity of
plasma formation on the ablation target material usually calls for
nanosecond-pulsed lasers. It is also possible to use picosecond and femtosecond
pulsed lasers for the same purpose, even though the particle formation mechanisms
are quite different [5]. Processes performed in vacuum allow a more detailed
investigation of events taking place from the laser absorption up to the final particle
solidification [6].
In the second and more commonly used scheme of self-formation of nanopar-
ticles, the ablation process takes place in a liquid environment (Fig. 7.1). Desired
material of the bulk form is immersed in a background liquid, and then ablated with
a pulsed laser, forming an emulsion of nanoparticles [7]. Particles sizes, again
typically below 100 nm are obtained and controlled via laser fluence (pulse energy
per unit area) and background liquid. Most commonly used materials are metals
(Au, Ag, Ti etc.) [8] and semiconductors [9], and most common background liquids
are water and ethanol. Different background liquids are also investigated for size
stabilization [10]. As with the gas/vacuum case, ultrashort pulse durations are
studied for the same purpose, both from a fundamental point of view (to investigate
nucleation dynamics), as well as to optimize particle size distributions [11].
Lastly, nanoparticles formed by the laser processes briefly outlined here also find
range of applications, especially in biology, biomedicine and chemistry. We refer
the reader to excellent reviews already published in the literature [12, 13].

7.3 Formation of Nanoscale Patterns by Femtosecond

Laser Ablation

7.3.1 General Principles

As mentioned above, self-formation of nanostructures through laser ablation in

vacuum, gas or liquid environments has a relatively long history, and there are
many good reviews in the literature. Hence, we presented a brief flavor from a quite
broad range of methods, materials and applications. In this section we will describe,
in a deeper detail, a relatively younger field of using femtosecond (fs) laser pulses
for generation of nanometer-scale structures. The two processes described in the
previous and present sections are fundamentally different. In the former, nanopar-
ticles are self-formed into clusters from a gas and/or plasma phase medium (formed
from bulk, using a laser). In the latter case, the process is more “controlled” in the
sense that the laser beam has a well-defined ablation volume, and the process is
performed by scanning the sample over the desired pattern. One can also make a
crude categorization of the two cases as bottom-up and top-down, even though the
nomenclature is not as distinct as the case of chemical processes, for example.
174 S. Akturk

Before delving into the experimental details, it would be useful to briefly discuss
the spirit of material processing with fs laser pulses. In general, laser ablation
processes are based on absorption of light energy (through whatever physical
process) by the electrons of the medium and consequent heating, melting and
vaporization [14]. Fs laser pulses has several aspects that separate them from
longer-pulse lasers. First of all, the pulse duration is typically orders of magnitude
shorter that heat diffusion times, as a result, there is essentially no thermalization
during the pulse period [15–17]. For transparent materials, long-pulse ablation is
initiated by noise electrons, through avalanche ionization, while fs laser generate
plasma via tunnel and multiphoton ionizations [18, 19]. As the latter processes are
deterministic, while the former being more dependent on the presence of impurities,
fs laser ablation yield much more precise, repeatable and well-defined structures
(free of heat affected zones, burrs etc.) [20].
While the field of laser material processing dates back to essentially a couple
years after the inversion of the laser, nanometer-scale resolutions (here we define
resolution as smallest ablation dimensions) have only been possible after the
emergence of fs laser material processing. In a typical laser material processing
experiment, the resolution is defined mainly by the overall heat affected zone and
also by the focused laser beam spot size. Furthermore, the minimum focusable spot
sizes are determined by wave nature of light and limited to about the wavelength of
the laser (called diffraction limit). As a result of combination of both effects, even
when UV lasers (e.g. excimers) are used, best achievable resolutions for CW and
long-pulsed lasers are limited to well above micrometers.
The situation changes dramatically for fs laser pulses, due to the deterministic
nature and minimal heat diffusion effects mentioned above. First of all, transverse
spatial modes of stable laser cavities are Gaussian beams and in the most common
lowest order (the so-called TEM00 mode) have a single-peak bell-like shapes [21].
The Gaussian shape is also preserved through focusing, with transformation to a
different spot size. When a fs laser beam is focused, to a near-diffraction-limited
spot, of diameter say *1 µm (for a visible or near-infrared laser), the intensity
distribution would look schematically like the cross section illustrated in Fig. 7.2.
As a result, by adjusting the laser intensity to proper levels, such that only the
central region of the focused beam exceeds the ablation threshold, resolutions well

Fig. 7.2 Generation of

sub-diffraction-limit ablation
sizes by working near critical
intensity
7 Pulsed-Laser Generation of Nanostructures 175

below the diffraction limit, and in the nanometer range can be achieved [22, 23]. In
one of the earliest proofs of this principle of “optics at critical intensity”, resolutions
of *30 nm are obtained with a laser wavelength at 1053 nm [22]. The ultimate
resolution of this approach is limited by the pulse-to-pulse stability of the laser
system and mechanical stability of the overall setup. The method is suitable for a
broad range of materials including metals, dielectrics and polymers [24].

7.3.2 Femtosecond Laser Nanomachining on Thin Films

with Bessel Beams

Nanometer-scale ablation by fs laser pulses explained above by itself is not suffi-

cient to generate functional nanostructures. The process simply yields a sensitive
“tool” to process larger-scale materials. One method for generating functional
nanostructures with this tool is to start with “thin films” (films with sub-micron
thicknesses) of desired materials, deposited on a suitable substrate, and generate a
two dimensional ablation pattern, via scanning of the fs laser beam [25, 26].
This nanofabrication method, although limited to materials of which thin films
can be formed, yields a direct, versatile and relatively simple approach. A typical
schematic of such experiments is shown in Fig. 7.3. With a fs laser of high repe-
tition rate, millimeter-scale regions can be processed within minutes. As a result, the
approach is particularly suitable for prototype investigations (not so much for serial
production). Most processes are performed in ambient air.
In the experimental realization of fs laser nanomachining, particularly on large
(tens of microns to millimiter-scale) surfaces, optical alignments and sample
positioning are quite critical and requires extreme care. High resolutions require
high-numerical aperture focusing, and the resulting tightly-focused beams diverge
very rapidly. Roughly speaking, beams focused to transverse sizes of * wave-
length stay intense only within a longitudinal depth of * one wavelength. Since the
essence of fs nanomachining is working sensitively close to the ablation threshold,
minute changes in the intensity due to beam divergence, sample positioning and
sample surface roughness can yield deleterious effects.

Fig. 7.3 Typical layout of a fs laser material processing experiment

176 S. Akturk

These critical constraints resulting from rapid diffraction of focused Gaussian

beams can be remedied by exploiting the so-called “non-diffracting beams” [27].
These beams have transverse field envelope profiles in the form of Bessel function
of the first kind and have the property of retaining the same intensity and beam
profile regardless of free-space propagation [28, 29]. Gaussian beam outputs of
typical laser cavities can be turned into quasi-Bessel beams by using conical lenses,
or axicons [30, 31]. Some example profiles of theoretical and experimental Bessel
beams is shown in Fig. 7.4. Defining the width of a Bessel beam as the breadth of
the central peak, the focal depths (length of Bessel zones) of Bessel beams can be
one or two orders of magnitude longer as compared to Gaussian beams of the same
width.
The potential advantages of Bessel beams in fs nanomachining should be clear.
Since the beam size does not change and the intensity changes very slowly within
the Bessel zone, the constraints related to positioning and alignments mentioned
above are significantly relaxed. Furthermore, it is much easier to use Bessel beams
in the ultraviolet wavelengths (by using a suitable axicon), as compared to Gaussian

Fig. 7.4 Theoretical transverse profile (a) and longitudinal intensity variation (b) of a Bessel
beam. Experimental Bessel beams generated by an axicon has a very similar transverse profile
(c) closely following the Bessel function (d)
7 Pulsed-Laser Generation of Nanostructures 177

beams, which require highly costly aberration-corrected objectives [32]. Note that,
when working close to ablation threshold, multiple-rings surrounding the main peak
do not contribute to the ablation pattern.

7.3.3 Experimental Results

The virtues of Bessel beams are exploited in many recent fs laser ablation exper-
iments. In a transparent sample, Bessel beams can form high-aspect-ratio channels
without mechanical scanning [33–35]. Similarly, it is also possible to ablate sur-
faces (in this case, one does not exploit long-focal-line nature of Bessel beams, but
more the non-diffracting nature) [36]. In this section, we will present some
experimental results on using Bessel beam fs lasers to generate nanostructures on
thin film surfaces.
Figure 7.5 shows nanometer scale channels and nanoparticle arrays formed on
Au thin films (*20 nm thickness) deposited on glass [37]. The laser used in this
process is a Yb:Glass based chirped-pulse amplification (CPA) system, delivering
550 fs pulses at 1 kHz. Second-harmonic of the output (515 nm) is generated to
enhance resolution. A glass axicon with 40° base angle is used to generate Bessel
beams. The diameter of Bessel beams decreases with increasing axicon base angle
[38], and this axicon is very close to the maximum, allowed for avoiding total
internal reflection inside the glass.
Metallic nanostructures of types shown in Fig. 7.5 can be readily used for
applications. Metal nanostructure arrays are particularly suitable for plasmonics
applications [39]. Plasmonic behaviors are observed in structures generated with
this method [40].

Fig. 7.5 Nanometer scale channels (left) and nanoparticle arrays (right) formed on Au thin films
178 S. Akturk

Fig. 7.6 SEM (left) and optical (right) images of graphene-on-SiO2 samples ablated with fs laser
Bessel beams

The second example that we present is on the emerging field of two-dimensional

(single-atomic layer) materials such as graphene [41]. These materials have exotic
physical behaviors and potentials for application in the future of electronics. Yet,
high resolution patterning of graphene and similar structures is still a challenge. In
this regard, the methods summarized in this section provides important opportu-
nities. In recent proof-of-principle experiments, nanometer-scale patterning of
graphene surface is demonstrated [42]. An example of graphene surface, patterned
with fs laser Bessel beams (with experimental parameters similar to above) is
shown in Fig. 7.6.

7.4 Conclusions

In this chapter, we discussed two main approaches for generation of

nanometer-scale structures using pulsed lasers. In the first, a bulk material is ablated
in vacuum, gas or liquid and consequent re-solidification of the evaporated material
in clusters generate the desired nanoparticles. While one can play with various
experimental parameters (laser powers, type of liquids, background gas etc.) to
control particle sizes, there is always a relatively broad distribution of particle sizes.
Nevertheless, particles with sizes down to few nanometers are achievable. In the
second approach, fs lasers are used to pattern material surfaces. This process is
more controlled in the sense that the desired geometries are generated by scanning
of the sample. On the other hand, best achievable resolutions are worse as compared
to the former approach. Resolutions better than 100 nm over large surfaces is very
difficult. Finally, special beam shaping methods forms an interesting frontier in
pushing forward the fs laser nanomachining methods towards broader practical
uses.
7 Pulsed-Laser Generation of Nanostructures 179

References

1. M. Vollmer, R. Weidenauer, W. Hoheisel, U. Schulte, F. Träger, Size manipulation of metal

particles with laser light. Phys. Rev. B. 40(18), 12509–12512 (1989)
2. A.M. Morales, C.M. Lieber, A laser ablation method for the synthesis of crystalline
semiconductor nanowires. Science 279(5348), 208–211 (1998)
3. D.B. Geohegan, A.A. Puretzky, G. Duscher, S.J. Pennycook, Time-resolved imaging of gas
phase nanoparticle synthesis by laser ablation. Appl. Phys. Lett. 72(23), 2987–2989 (1998)
4. F.E. Kruis, H. Fissan, A. Peled, Synthesis of nanoparticles in the gas phase for electronic,
optical and magnetic applications—a review. J. Aerosol Sci. 29(5–6), 511–535 (1998)
5. S. Eliezer, N. Eliaz, E. Grossman, D. Fisher, I. Gouzman, Z. Henis et al., Synthesis of
nanoparticles with femtosecond laser pulses. Phys. Rev. B. 69(14), 144119 (2004)
6. S. Amoruso, G. Ausanio, R. Bruzzese, M. Vitiello, X. Wang, Femtosecond laser pulse
irradiation of solid targets as a general route to nanoparticle formation in a vacuum. Phys. Rev.
B. 71(3), 033406 (2005)
7. S.I. Dolgaev, A.V. Simakin, V.V. Voronov, G.A. Shafeev, F. Bozon-Verduraz, Nanoparticles
produced by laser ablation of solids in liquid environment. Appl. Surf. Sci. 186(1–4), 546–551
(2002)
8. A.V. Simakin, V.V. Voronov, N.A. Kirichenko, G.A. Shafeev, Nanoparticles produced by
laser ablation of solids in liquid environment. Appl. Phys. A 79(4–6), 1127–1132 (2004)
9. K.V. Anikin, N.N. Melnik, A.V. Simakin, G.A. Shafeev, V.V. Voronov, A.G. Vitukhnovsky,
Formation of ZnSe and CdS quantum dots via laser ablation in liquids. Chem. Phys. Lett. 366
(3–4), 357–360 (2002)
10. J.-P. Sylvestre, A.V. Kabashin, E. Sacher, M. Meunier, J.H.T. Luong, Stabilization and size
control of gold nanoparticles during laser ablation in aqueous cyclodextrins. J. Am. Chem.
Soc. 126(23), 7176–7177 (2004)
11. J.-P. Sylvestre, A.V. Kabashin, E. Sacher, M. Meunier, Femtosecond laser ablation of gold in
water: influence of the laser-produced plasma on the nanoparticle size distribution. Appl. Phys.
A 80(4), 753–758 (2004)
12. I.H. El-Sayed, X. Huang, M.A. El-Sayed, Surface plasmon resonance scattering and
absorption of anti-EGFR antibody conjugated gold nanoparticles in cancer diagnostics:
applications in oral cancer. Nano Lett. 5(5), 829–834 (2005)
13. M.-C. Daniel, D. Astruc, Gold nanoparticles: assembly, supramolecular chemistry,
quantum-size-related properties, and applications toward biology, catalysis, and
nanotechnology. Chem. Rev. 104(1), 293–346 (2004)
14. W.M. Steen, Laser Material Processing, 3rd edn. (Springer, London, 2008)
15. B.C. Stuart, M.D. Feit, S. Herman, A.M. Rubenchik, B.W. Shore, M.D. Perry,
Nanosecond-to-femtosecond laser-induced breakdown in dielectrics. Phys. Rev. B. 53(4),
1749 (1996)
16. B.C. Stuart, M.D. Feit, A.M. Rubenchik, B.W. Shore, M.D. Perry, Laser-induced damage in
dielectrics with nanosecond to subpicosecond pulses. Phys. Rev. Lett. 74(12), 2248 (1995)
17. S. Nolte, C. Momma, H. Jacobs, A. Tünnermann, B.N. Chichkov, B. Wellegehausen et al.,
Ablation of metals by ultrashort laser pulses. J. Opt. Soc. Am. B. 14(10), 2716–2722 (1997)
18. X. Liu, D. Du, G. Mourou, Laser ablation and micromachining with ultrashort laser pulses.
IEEE J. Quantum Electron. 33(10), 1706–1716 (1997)
19. D. Du, X. Liu, G. Korn, J. Squier, G. Mourou, Laser-induced breakdown by impact ionization
in SiO[sub 2] with pulse widths from 7 ns to 150 fs. Appl. Phys. Lett. 64(23), 3071–3073
(1994)
20. R.R. Gattass, E. Mazur, Femtosecond laser micromachining in transparent materials. Nat.
Photon. 2(4), 219–225 (2008)
21. A.E. Siegman, Lasers (Univ Science Books, 1986)
22. A.P. Joglekar, H. Liu, E. Meyhöfer, G. Mourou, A.J. Hunt, Optics at critical intensity:
applications to nanomorphing. Proc. Natl. Acad. Sci. USA 101(16), 5856–5861 (2004)
180 S. Akturk

23. P.P. Pronko, S.K. Dutta, J. Squier, J.V. Rudd, D. Du, G. Mourou, Machining of sub-micron
holes using a femtosecond laser at 800 nm. Opt. Commun. 114(1–2), 106–110 (1995)
24. F. Korte, J. Serbin, J. Koch, A. Egbert, C. Fallnich, A. Ostendorf et al., Towards
nanostructuring with femtosecond laser pulses. Appl. Phys. A 77(2), 229–235 (2003)
25. J. Kim, S. Na, Metal thin film ablation with femtosecond pulsed laser. Opt. Laser Technol. 39
(7), 1443–1448 (2007)
26. V. Mizeikis, S. Juodkazis, K. Sun, H. Misawa, Fabrication of micro- and nanostructures in thin
metallic films by femtosecond laser ablation. [cited 2015 Jun 11], pp. 75910 J–75910 J–11
(2010). http://dx.doi.org/10.1117/12.843074
27. K. Dholakia, Optics: Against the spread of the light. Nature 451(7177), 413 (2008)
28. J. Durnin, J.J. Miceli, J.H. Eberly, Diffraction-free beams. Phys. Rev. Lett. 58(15), 1499–1501
(1987)
29. J. Durnin, Exact solutions for nondiffracting beams. I. The scalar theory. J. Opt. Soc. Am. A 4
(4), 651–654 (1987)
30. J.H. McLeod, Axicons and their uses. J Opt Soc Am. 50(2), 166–169 (1960)
31. J. Arlt, K. Dholakia, Generation of high-order Bessel beams by use of an axicon. Opt.
Commun. 177(1–6), 297–301 (2000)
32. R. Sahin, T. Ersoy, S. Akturk, Ablation of metal thin films using femtosecond laser Bessel
vortex beams. Appl. Phys. A 9, 1–5 (2014)
33. F. Courvoisier, J. Zhang, M.K. Bhuyan, M. Jacquot, J.M. Dudley, Applications of
femtosecond Bessel beams to laser ablation. Appl. Phys. A. 1–6
34. M.K. Bhuyan, F. Courvoisier, P.A. Lacourt, M. Jacquot, R. Salut, L. Furfaro et al., High
aspect ratio nanochannel machining using single shot femtosecond Bessel beams. Appl. Phys.
Lett. 97(8), 081102 (2010)
35. M.K. Bhuyan, F. Courvoisier, P.-A. Lacourt, M. Jacquot, L. Furfaro, M.J. Withford et al.,
High aspect ratio taper-free microchannel fabrication using femtosecond Bessel beams. Opt.
Express 18(2), 566–574 (2010)
36. F. Courvoisier, P.-A. Lacourt, M. Jacquot, M.K. Bhuyan, L. Furfaro, J.M. Dudley, Surface
nanoprocessing with nondiffracting femtosecond Bessel beams. Opt. Lett. 34(20), 3163–3165
(2009)
37. B. Yalizay, T. Ersoy, B. Soylu, S. Akturk, Fabrication of nanometer-size structures in metal
thin films using femtosecond laser Bessel beams. Appl. Phys. Lett. 100(3), 031104–031104–3
(2012)
38. S. Akturk, B. Zhou, B. Pasquiou, M. Franco, A. Mysyrowicz, Intensity distribution around the
focal regions of real axicons. Opt. Commun. 281(17), 4240–4244 (2008)
39. S.A. Maier, Plasmonics: Fundamentals and Applications (Springer, 2007), 234 pp
40. R. Sahin, Y. Morova, E. Simsek, S. Akturk, Bessel-beam-written nanoslit arrays and
characterization of their optical response. Appl. Phys. Lett. 102(19), 193106 (2013)
41. A.K. Geim, K.S. Novoselov, The rise of graphene. Nat. Mater. 6(3), 183–191 (2007)
42. R. Sahin, E. Simsek, S. Akturk, Nanoscale patterning of graphene through femtosecond laser
ablation. Appl. Phys. Lett. 104(5), 053118 (2014)
Chapter 8
Graphene for Silicon Microelectronics:
Ab Initio Modeling of Graphene
Nucleation and Growth

Jarek Dabrowski, Gunther Lippert and Grzegorz Lupina

Abstract Graphene electronics is expected to complement the conventional Si

technologies. Graphene processing should thus be compatible with the mainstream
Si technology: CMOS. Ideally, it should be possible to grow graphene directly on a
Si wafer, but this does not work. Large area graphene can be grown on Cu or on Ni,
its transfer to silicon must then follow, which is problematic. Researchers try
therefore to grow graphene on CMOS compatible substrates, such as on Ge/Si(001)
wafers. Ab initio modeling, particularly when used in combination with experi-
mental data, can elucidate the mechanisms that govern the process of nucleation and
growth of graphene, and thus provide assistance in the design of experiments and
production processes. We overview our results in this context, startig from atomic C
deposited on (chemically inert) graphene, through the similar cases of Si deposited
on graphene and C deposited on hexagonal boron nitride, and the case of carbon on
a non-inert insulator (SiO2-like surface of mica), up to C atoms and hydrocarbon
molecules building graphene on Ge(001) surfaces.

8.1 Introduction

Graphene, a two-dimensional sheet of sp2-bonded carbon atoms, is a semi-metal

with linear energy-momentum relation around the Fermi energy. This photon-like
law has its origin in the symmetry of the atomic lattice and makes the charge
carriers massless, just like photons are. One of the consequences is that the sheet
conductivity of even a relatively highly defected graphene sheet may be comparable

J. Dabrowski (&)
G. Lippert
G. Lupina

IHP, Im Technologiepark 25, 15236 Frankfurt(Oder), Germany
e-mail: [email protected]
G. Lippert
e-mail: [email protected]
G. Lupina
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 181

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_8
182 J. Dabrowski et al.

to what would be needed for a heterojunction bipolar transistor (HBT) designed to

operate at terahertz frequencies. And perfect enough graphene used as the channel
material may enable a field effect transistor (FET) to work in the THz range.
Graphene is thus supposed to give the traditional microelectronics based on Si FETs
or HBTs the chance to cross the terahertz boundary [1–3].
Graphene can be formed from SiC on Si [4, 5] or grown on a metal substrate
(usually, on Cu) and then transferred to Si [6]. However, practical constraints make
both these approaches difficult to integrate with the chip production. For example,
transfer from copper leads to contamination with Cu atoms [7]. Graphene can also
be grown directly on Si or on SiO2 by plasma-enhanced chemical vapor deposition
(PE-CVD) [8]. Yet such films have too many graphene layers for FET-like appli-
cations. Moreover, chemical interaction of C with Si results in the formation of
carbide, excluding the usage of this material as the base electrode in HBT-like
devices. On the other hand, graphene can be grown on Ge by CVD [9] or even by
molecular beam epitaxy (MBE), i.e., from atomic source [10]. This is good news,
because Ge is a material that is fully compatible with the standard silicon
technology.
In this chapter we work out some theoretical aspects of the nucleation and
growth of graphene on semi-metallic and non-metallic substrates. We put most
weight on the chemically simplest case, which is deposition of carbon atoms in a
vacuum MBE chamber. We begin with the single-species system: carbon on gra-
phene. Comparison to the growth of Si films on graphene is then done [11]. We use
these two cases together with that of carbon on hexagonal boron nitride (hBN) to
illustrate the behavior of atoms on a substrate with which they interact only weakly.
We then turn attention to the substrate that is insulating and in most cases chem-
ically inert, but yet highly reactive to carbon: to hexagonal SiO2 surfaces and to
similar materials, exemplified by mica. Finally, we analyze the process of nucle-
ation and growth of graphene on the Ge(001) surface, for which we make also a
brief excursion into the CVD process. The latter section is preceded by remarks on
the graphene base transistor (GBT) [12–14], a device in which graphene grown
directly on Ge may be a good combination of materials.
In the latter context, we performed calculations for electron tunneling through
graphene sheet [13]. It turns out that tunneling probability may have to be
accounted for in the simulations of GBT electrical performance, particularly when
the lateral momentum is conserved in the tunneling process, as it may be the case if
hexagonal BN (hBN) is used to separate the bottom electrode from the base.
Hexagonal BN is generally interesting as a substrate for graphene because its
interaction with carbon is weak enough for high carrier mobility to be retained in
graphene but seems to be strong enough to allow for some epitaxial relationship
between the substrate and the film. It may be viewed as a potential substrate also for
CMOS-GFET integration. We learned that graphene nucleates on h-BN at
pre-existing defect sites, including C and B2O3 impurities, and estimated the
sticking coefficient of C as a function of average distance between the adsorbing
sites [15].
8 Graphene for Silicon Microelectronics … 183

One of the first successful experimental growth studies of graphene on an

insulating surface have however been performed by MBE deposition of C on mica,
i.e., on a surface closely resembling SiO2 but having few surface defects. From
theoretical analysis we learned that due to strong interaction of C with O and Si,
only deposition on defect-free, insulating oxide surfaces may lead to graphene films
of acceptable quality [16–18]. While this can be achieved on mica, the density of
nucleation sites on SiO2 is too high: the mica process cannot be transferred to SiO2.
The final substrate considered in this Chapter is germanium. In contrast to the
case of Si, deposition of C onto Ge does not deteriorate the substrate by carbide
formation. However, graphene grown by MBE on Ge(001) is separated from the
substrate by a poorly conducting film that is probably built of nanocrystalline
graphene interconnected by Ge atoms [10] and which may be better to be elimi-
nated if an application in a THz device is targeted. An ab initio study of the
interaction of carbon and hydrocarbons with Ge(001) provided insight into the
nucleation of this film.

8.2 Approach

Electronic structures, total energies, forces, and atomic positions were calculated in
the framework of the Density Functional theory (DFT) as implemented in the
Quantum Espresso package designed for massively parallel computations [19]. The
exchange and correlation energy of electrons was approximated by the gradient
corrected functional of Perdew, Burke and Ernzerhof [20]. Spin polarization was
accounted for. The atoms were described by the pseudopotential method using
custom ultrasoft potentials and a plane wave basis set with the energy cut-off of 30
Ry for the wave functions and 240 eV for the electron density. The calculations for
Ge(001) substrates (rectangular surface cell) were done with two special k-points
from the Brillouin zone (or with a sampling derived from this k-point set), while the
calculations for substrates with hexagonal surface cells (hexagonal boron nitride,
model mica, graphene) were done with the C sampling (or with a C-derived
sampling).1 Structures were assumed to be converged when the maximum force
dropped below 0.01 eV/Å and the expected change of the total energy by further
relaxation was insignificant for the final conclusion. Energy barriers on migration
and reaction paths and the corresponding atomic configurations have been obtained
using the Nudged Elastic Band/Climbing Image (NEB/CI) algorithm [21, 22],

1
Derived means here “using the same set but taken from Brillouin zone of a larger surface cell”.
This is done with caution, as it costs computing time. For example, calculation using the C point
from the 24 × 24 cell of graphene (this is a fully converged k-point set) requires 10 times more
wall time (or 10 times more cores if the same wall time is requested) than the standard calcu-
lation done with the C point from the 6 × 6 cell (this is a poor set for adsorption energies, but a
reasonable one for diffusion barriers). Therefore, the structures are computed using the funda-
mental k-point set and only in the most interesting cases refined using the fully converged set.
184 J. Dabrowski et al.

which searches for saddle points on the energy surface. Where applicable, the
energies and barriers are compatible with those obtained in previous DFT studies
[23–25].

8.3 Carbon on Graphene

When carbon atoms are deposited on any material and graphene then nucleates and
grows, sooner or later more and more of the incoming carbon atoms do not hit the
original substrate but they land on the newly grown graphene film. Carbon on
graphene, while being the chemically simplest system among those treated in this
Chapter, is therefore also the one of the fundamental importance.
A carbon atom approaches graphene from vacuum without any barrier. When
adsorbed, its most stable position is a bridge site: it stands above two neighboring C
atoms of graphene, making two bonds to their protruding orbitals. These were
originally pz orbitals, but the interaction with the additional atom adds an s com-
ponent, transforming them towards a hybrid with sp3 character. This creates tension
in the substrate, because in response to the changed hybridization, the preferred
angles to the neighboring atoms change as well, from sp2 towards tetrahedral sp3
angles. A tiny elevation is thus produced, with the adsorbed C on its top (Fig. 8.1).
The adsorption energy, equal to the desorption barrier, is small: 1.3 eV, only by
0.7 eV higher than the barrier for diffusion. The consequence is that when the
temperature of the substrate is as high as during typical deposition (above 800 °C),
most of the adsorbed C desorbs after diffusing for only short time. The probability
that a deposited C atom will combine with another one into a C2 dimer can be
roughly estimated by the product of the deposition rate R and the lifetime tlife of the
ad-atom on the surface (Fig. 8.2). When p ¼ tlife R approaches unity, the adsorbed C
may be hit by another atom arriving from vacuum.2 But at typical deposition
temperatures of around 850 °C (the dashed line in Fig. 8.2) and realistic deposition
rates R the life time is way too short for spontaneous formation of carbon clusters
that stick to graphene.
The probability of C2 formation in fact higher. This is also because a C atom
sitting on the surface can kick its C neighbor out from the sheet and take its place in
the crystal. The result of equivalent to the C ad-atom punching through the gra-
phene, from its top to its bottom surface. Once on the other side, the atom may bind
itself to the underlaying material, so that the life time of atomic C in the adsorbed
state increases. Nevertheless, it turns out that trimers (C3) are unbound on gra-
phene: they desorb immediately after being formed. This means that preexisting
nucleation sites are needed for another graphene layer to form on top of the
complete one (Fig. 8.3).

This ignores the probability that two ad-atoms find one another; for full treatment, see Fig. 8.5a.
2
8 Graphene for Silicon Microelectronics … 185

Fig. 8.1 Adsorption, diffusion and desorption of C atoms on graphene. The gray trails connect the
initial and final atomic configurations on the NEB/CI reaction path

Fig. 8.2 The C2 dimer formation probability p on graphene, estimated as tlife R, where tlife is the
lifetime of C ad-atom on the surface and R is the deposition rate. A typical deposition rate R = 10
monolayers per minute is assumed. The lifetime is obtained from tlife ¼ fdes expðEdes =kTÞ, where
Edes is the desorption barrier of 1.3 eV (Fig. 8.1) and the attempt frequency fdes is taken to be
fdes ¼ 1013 s−1

Fig. 8.3 Collection zone around an adsorbing defect on graphene (e.g., step edge). To be
adsorbed, the C atom must fall within the distance w from the defect. The atoms that land further
away, desorb to vacuum without contributing to the graphene growth

Because the time span tlife between the adsorption and desorption events is
limited, so is the distance walked on the surface by the adsorbed atom. To the
growth of graphene can therefore contribute only those of the impinging atoms,
which hit the surface close enough to the nucleation center. Using the random walk
theory one can easily show [15] that the collection zone width w (defined as such
186 J. Dabrowski et al.

distance from a linear adsorbing boundary within which ad-atoms can be treated as
contributing to growth and beyond which they can be treated as non-contributing)
equals to
sffiffiffiffiffiffiffi  
fmig Edes  Emig
w¼s exp ; ð8:1Þ
fdes 2kT

where s is a distance of the order of the jump length, fmig and fdes are attempt
frequencies for diffusion and desorption, Emig and Edes are the corresponding energy
barriers, and kT is the thermal energy. Please note that 1=w is thermally activated
with only half of the difference between the desorption and diffusion barriers. The
factor 12 in the exponential term stems from the random character of the walk.
Indeed, it is the well-known result of the random walk theory that the distance by
pffiffiffiffi
which a random walker departs from the initial point after N jumps scales as N .
At 850 °C, the estimated zone width is w ¼ 10 nm, or several lattice constants.
The MBE growth temperature cannot be much smaller than that; the graphene
quality would otherwise deteriorate. The sticking coefficient of C on good graphene
is thus tiny: very many monolayers must be deposited to produce a single ad-layer.

8.4 Silicon on Graphene

In order to build a useful graphene transistor, one must deposit some material on the
graphene sheet. In the case of a GBT, this can be a silicon film, for example.
Figure 8.4 illustrates a somewhat intriguing effect observed [11] after MBE
deposition of Si atoms onto graphene (transferred from Cu onto oxidized Si).
Though the film is in general rough and grainy, one notes that in the case of sample
1 there are regions where a relative smooth Si layer was produced. A closer
examination reveals that where the film is smooth, the graphene was somewhat

Fig. 8.4 Scanning electron microscope (SEM) images of two graphene samples after smiliar MBE
deposition process of Si at 550 °C [11]. The Si overlayer in sample 1 is closed but inhomogeneous,
apart from quite a flat region that grew on an island of a thicker (probably bilyer) graphene.
A similar though larger island is visible on sample 2, but in this case it is not overgrown
8 Graphene for Silicon Microelectronics … 187

Fig. 8.5 a Spontaneous nucleation probabilities and b collection zone widths w, obtained for
MBE deposition of Si atoms [11] and C atoms [15] on graphene from ab initio results by applying
the random walk theory [15]

thicker, probably consisting of two carbon layers (such islands appear often on
single-layer graphene sheets). But in contrast to that, the same kind of islands on
sample 2 were not overgrown. Graphene has been transfered to both samples in the
same process; the difference is that graphene on sample 1 stems from different
provider than on sample 2.
The solution to this puzzle is suggested by results of ab initio calculations [11].
First, one finds that Si atoms stick to graphene even worse than C atoms do
(Fig. 8.5). The adsorption energy (and the desorption barrier) of Si on undoped
graphene is only 0.58 eV, less than half of that for C. On p-type graphene it may
increase noticeably due to electron transfer from Si (Fig. 8.5). The self-nucleation
probability3 is low at 550 °C and above, but many dimers form below about 200 °C
for undoped or about 450 °C for p-doped graphene because two ad-atoms spent
enough time on the surface and visit enough sites there to find one another
(Fig. 8.5a); yet this contributes only to Si loss, because in contrast to C dimers, Si
dimers do not stick to graphene at all, apart from being attracted to it by weak van
der Waals forces. But when nucleation sites are present on graphene, Si atoms may
stick to them, and the collection zone increases strongly when the graphene is
p-doped (Fig. 8.5b). At 550 °C (the dashed line in Fig. 8.5b), the collection zone
width on p-type graphene is estimated to be of the order of 100 nm, which com-
pares well to the distance between the grains in Fig. 8.4. Overgrowth with Si is
therefore possible even at 550 °C if graphene is p-doped and if there is enough
nucleation sites available. Indeed, the nucleation was found to correlate with the
surface concentration of PMMA rests.4

Figure 8.5a includes the probability that two ad-atoms find one another on the surface.
3

4
PMMA is the polymer used during graphene transfer from Cu to the target wafer.
188 J. Dabrowski et al.

8.5 Carbon on h-BN

Hexagonal boron nitride (hBN) is a layered compound with a graphene-like atomic

structure. The graphene lattice sites are now occupied by alternating B and N atoms,
and the lattice constant differs by only 4 % from the lattice constant of graphene.
Consequently, some epitaxial relationship may be achieved in the hBN/graphene
system, whereby the mismatch of the lattice constant leads to the appearance of
Moiré patterns in AFM and STM images and to electronic super-structure effects.
When C atoms are deposited on hBN, graphene can be formed, but the mor-
phology of the resulting film varies. We have therefore analyzed by ab initio cal-
culations the influence of typical impurities on the nucleation of graphene on hBN
and compared the predictions to the experimental data obtained on hBN flakes with
non-negligible concentration of carbon, oxygen, and boron oxide (Fig. 8.6).
Figure 8.7 compares the adsorption data for C ad-atoms and ad-dimers on perfect
hBN and perfect graphene. It is clear that sticking of carbon to both substrates is
comparable and weak. It is thus no surprise that also on hBN the self-nucleation
path of carbon is broken by desorption of carbon trimers, C3.
It may be interesting to verify whether the formation of C tetramers, C4, from
two dimers may provide a nucleation path bypassing the trimer desorption. The
answer is: yes, it can, but this mechanism has low efficiency. During the whole
deposition process, only about 200 tetramers were created in this way in the area
imaged in Fig. 8.6. This makes a negligible fraction of the number of carbon atoms
that are incorporated into graphene during the same time.
But C atoms, dimers, and also bigger C clusters including C3 are trapped by C–C
donor-acceptor pairs: two substitutional C atoms occupying two neighboring sites
in the hBN lattice (Fig. 8.8). This explains the appearance of isolated graphene
islands.

Fig. 8.6 AFM image of hBN after deposition of C atoms by MBE at 850 °C [15]. Flat areas,
isolated islands, lines of islands, island-free steps, and areas covered by coalesced islands are
recognizable
8 Graphene for Silicon Microelectronics … 189

Fig. 8.7 Comparison of the atomic structures and energies on perfect hBN and on perfect
graphene for a C ad-atoms and b C2 ad-dimers [15]. Carbon is black, boron is dark green, nitrogen
is light green, DE ¼ Edes  Emig . Life times between adsorption and desorption events are given
for Tsubstrate = 850 °C, assuming the attempt frequencies fdes and fmig to be equal to 1013 s−1

Fig. 8.8 Trapping of C ad-atoms, ad-dimers and ad-trimers by C–C donor-acceptor pairs

From the ab initio data and from random walk theory it follows that [15]:
1. MBE carbon desorbs from perfect hBN and from perfect graphene, at least at
substrate temperatures as high as needed to obtain reasonably crystalline films.
2. The observed island heights are compatible with the total exposure to carbon
and with theoretical sticking coefficients obtained for the density of adsorbing
boundaries suggested by the analysis of the Raman spectra.
3. The sticking coefficient computed for average the monatomic step density
observed on the fresh substrate is high enough to justify the assumption that the
flat areas visible in AFM images are covered by monolayer graphene.
4. The islands in the densely covered areas may have nucleated on B2O3 inclusions
and on oxygen-carbon donor-acceptor pairs. Contribution from carbon-carbon
donor-acceptor pairs and from larger C [?] precipitates (graphene-like inclusions)
190 J. Dabrowski et al.

is possible as well, provided that the size distribution of carbon clusters in the
virgin hBN substrate is far from thermal equilibrium.
5. The lines of islands are likely to be associated with nucleation on
boron-terminated substrate steps that have been oxidized by ambient oxygen.

8.6 Carbon on Mica

The case of atomic carbon deposited on the surface of mica [16–18] is interesting
for several reasons. First, the surface of mica is built mostly of SiO2 that, due to the
hexagonal arrangement of surface atoms, is free of dangling bonds. SiO2 is the
natural insulator used in Si microelectronics, although the technological SiO2 is
amorphous and its surface has a certain density of defects with Si bonds that are not
saturated by oxygen nor by another Si atom. Second, the defect density on mica
surfaces is usually low; this includes low concentration of steps, as mica is a layered
material. Third, atomic C combines easily with oxygen (burning to CO) and with Si
(forming SiC). Given this third property, it is intriguing that high quality graphene
flakes can be grown on mica, and that large areas of mica may remain unaffected
even after prolonged exposure to high flux of carbon atoms (Fig. 8.9).
Three different hypotheses can be formulated [26] to explain this effect
(Fig. 8.10):
A. The surface of mica reflects carbon atoms and graphene grows only at defects.
B. Carbon ad-atoms diffuse on the surface of mica to nucleation centers (defects)
faster than they bind the surface oxygen atoms into volatile CO molecules;
C. Carbon atoms do reduce the surface (CO is emitted) and then intervene into
oxygen vacancies to make bonds with Si, but the substitutional CO produced in
this way diffuses to nucleation centers faster than it can escape as CO after
reacting with a surface oxygen atom, while the oxygen removed from the sur-
face is efficiently restored by O diffusion from subsurface and/or bulk regions of
mica.

Fig. 8.9 MBE graphene on mica. Left Optical image. Right AFM overlayed with Raman data
8 Graphene for Silicon Microelectronics … 191

Fig. 8.10 Schematic illustration of hypotheses explaining the formation of graphene on mica

Fig. 8.11 Atomic structure of C on mica, as acquired after ten of ten molecular dynamic runs. C is
black, O is yellow, Si is red, Al is brown, H is blue. At the substrate temperatures used in the
experiment, 10 ms suffice for this atom to escape as CO

Hypothesis A can be tested by ab initio molecular dynamics: a carbon atom with

velocity similar to that of atoms in the MBE beam is placed above the surface of
mica or of hexagonal SiO2. Its interaction with the surface is then monitored. After
a few runs with various initial parameters, one can obtain a reasonable picture of
what can and what cannot happen. And it appears that hypothesis A fails
(Fig. 8.11).
192 J. Dabrowski et al.

Fig. 8.12 a C atom (black) diffusing under the surface of model muscovite mica. b Second
ground state of adsorbed C, with the energy the same as that in Fig. 8.11 but with only 10 ns of
conversion time to CO. O is yellow, Si is red, Al is brown, H is blue

To test hypothesis B, one should therefore search for such diffusion paths, which
avoid a close contact between C and twofold-coordinated oxygen. Such a path may
indeed exist under the surface (Fig. 8.12a). But the barrier for diffusion turns out to
be clearly higher than that for a detour that the diffusing atom can take to the surface
and then towards its escape as CO. Moreover, the C atom can reach the alternative
ground state (Fig. 8.12b), from which it escapes as CO even faster than from the
ground state shown in Fig. 8.11. Thus, hypothesis B fails as well.
Somewhat disappointingly, the remaining hypothesis C also fails. On the one
hand, oxygen vacancy diffusion is possible. The vacancy must be positively
charged, or the barrier is high. The same is known in SiO2, and the barriers are
similar. But although carbon can diffuse on the surface by a vacancy-mediated
mechanism, the diffusion barrier is higher than the barrier for CO emission.
One should therefore re-examine the hypothesis A, keeping in mind that the
reflection of the incoming C atoms does not have to be complete, because oxygen
vacancy diffusion may help to restore the surface stoichiometry to one that contains
enough oxygen to suppress the formation of carbide that would survive the contact
with ambient oxygen when the sample is removed from the reaction chamber.
The reason why C is reflected from the surface may be associated with the
internal excitation of the C electrons caused by the interaction between the carbon
and oxygen orbitals before the bond is formed. Figure 8.13 illustrates this effect. It
does not lead to internal excitation during unconstrained DFT molecular dynamics,
because such calculations force the system to the ground state. But the energy
transfer from the C 2p electrons to mica does require certain time. If this time is
shorter than the time after which the internally excited atom would be reflected to
vacuum (i.e., shorter than about 60 fs), the atom eventually is indeed reflected back.
This is confirmed by DFT calculations with constrained orbital occupation.
Energy transfer from the atom to the substrate is expected to be more efficient
when there are partially occupied states that are sufficiently close in energy and in
the distance to the excited C 2p orbitals. This means that surface regions which are
defected are more sticky to C; graphene and then graphite can grow there. All in
8 Graphene for Silicon Microelectronics … 193

Fig. 8.13 Pauli repulsion effect. Blue C 2p2 orbital, initially degenerate and with spin 1. Violet and
red valence and conduction states of mica (O orbitals). C px and py try to make bonds to O, 2pz and
2s2 are repelled from VB top. When pz remains occupied, the C atom turns back. This is confirmed
by DFT calculations with constrained orbital occupation: all 10 molecular dynamics runs resulted
with reflection back to vacuum

Fig. 8.14 Pingpong ball analogy of the Pauli repulsion effect. The ball bounces from the table
because it becomes internally excited. If the excitation energy is efficiently transferred to the target,
as to the beach sand, the ball stops

one, the behavior of carbon on the mica surface seems to resemble the behavior of a
pingpong ball (Fig. 8.14), which bounces from a table (perfect areas of the surface)
but sticks to sand (defected areas of the surface).
Figure 8.15 summarizes the proposed growth mechanism. It takes into account
the diffusivity of O vacancies. The seed may be either a dust particle, or damaged
surface, or the area where the vacancies cannot become positively charged and for
that reason the interaction with C gradually converts the oxide there into carbide.
For a number of reasons, mica cannot be used in CMOS technology. The
technological consequence of the above mechanism is thus that the growth of
graphene on oxides with twofold-coordinated O, like on SiO2, is hard to achieve:
the surface density of defects is too high there. Such a high defect density combined
with low surface diffusivity of C results in a nanographene-like film that is then
difficult to overgrow with graphene crystals of at least lm-size, as needed for the
transistors.
194 J. Dabrowski et al.

Fig. 8.15 The proposed mechanism of graphene and graphite growth from C atomic beam on
mica

8.7 Graphene Base Transistor

In a Field Effect Transistor (FET), the current flowing through the channel formed
below the gate oxide is controlled by the potential of the gate electrode on the
opposite side of the gate oxide: the channel, the gate oxide, and the gate electrode
form a capacitor. In a graphene FET (GFET), the channel is made of graphene, so
that the current that is being controlled flows in the plane of the graphene sheet.
The Graphene Base Transistor (GBT) is based on a completely different device
concept, akin to that of a bipolar transistor or simply to that of a vacuum tube: the
charge carrier flow is controlled by a base electrode placed between the cathode
(emitter) and the anode (collector). The carriers flow in the direction normal to the
plane of the base electrode, which is graphene in the GBT (Fig. 8.16a).
One of the practical problems of a GFET is that when a semi-metal is used as the
channel material, then the transistor cannot be switched off. The gate voltage
modulation can only modulate the channel resistance, but the channel remains
always conductive. In other words, a GFET is a poor switch, and as such it is not
suited for logics: in a microprocessor, transistors in the OFF state should conduct no
current, otherwise unacceptable power losses occur and the circuit is not only too
expensive in use, but it simply heats up so much that it cannot work at all. The
situation in a GBT is totally different: when a suitable emitter-collector bias is
applied, the emitter-base bias modulation can switch the state of the transistor
between non-conductive OFF and conductive ON (Fig. 8.16a).
Figure 8.16b demonstrates the influence of the graphene sheet on the trans-
mission probability. Transmission probabilities were computed from self-consistent
band structures of two unbiased Co electrodes placed at various distances from one
another; for the separation of 1.9 nm, one calculation was done with a graphene
sheet placed midway between the electrodes. It is clear that insertion of graphene
increases the transmission, but the effect is not dramatic: the graphene sheet
8 Graphene for Silicon Microelectronics … 195

Fig. 8.16 a The GBT concept with graded collector insulation [13]; blue line is the unbiased
device, red line is the device in the ON state, ECI and BCI are the base-emitter and base-collector
insulators. b Energy distribution of the relative transmission probability for a graphene sheet
placed centrally between two Co electrodes separated by 1.9 nm. Various lines correspond to
various reference configurations. C edge is the computed position of graphene conduction band
bottom at C

behaves approximately as a tunneling barrier. The reason is that the electrons

traveling in the direction normal to the sheet see the forbidden gap that graphene
has at C.
This result relies however on the conservation of the lateral momentum of
electrons and is therefore strictly valid only if graphene is not a strong perturbation
to the symmetry of the system. In most of the practically useful cases, graphene
would destroy the lateral symmetry completely, so that efficient channels away
from C may appear even for electrons traveling along the normal. The only
exception may be graphene placed on hBN. Therefore, for most of the layers that
separate graphene from the emitter and collector electrodes (Fig. 8.16a) it may be
still legitimate to treat graphene as fully (or nearly fully) permeable to
emitter-collector electrons.
The calculations summarized in Fig. 8.17 have been done for a device in which
the graphene base electrode is placed on a Ge(001) film, whereby graphene was
assumed to be a moderately strong barrier for electrons traveling across the sheet.
The results indicate that such a device may be capable of operating in the THz
range. GBTs built with graphene placed between a Ge(001) wafer and overgrown
with a germanium film have been analysed by other researches [27], with similar
conclusions: that such devices may be able to operate with terahertz frequencies.
196 J. Dabrowski et al.

Fig. 8.17 High-frequency performance (cutoff frequency) of the GBT, simulated for a GBT with
Ge(001) “insulation” between an erbium germanide emitter and graphene, assuming that graphene
is a moderate tunneling barrier for electrons. The red solid and the blue dashed line correspond to
the simulation done with and without taking the quantum capacitance of graphene into account. It
follows that the Ge/graphene device should be capable of operating in the THz range. In addition,
the device is unique in that it can produce high output power: the design of the base-collector
insulator allows for high voltage to be applied between the emitter and the collector

8.8 Carbon on Germanium

As noted in Sect. 8.7, from the design point of view it would be good to have Ge
(001) as the substrate on which graphene is grown in the active device region. For
this purpose, relatively small graphene flakes suffice: the width of the device may be
smaller than a micron, and the whole area may be of the order of a lm2. On the
other hand, if a good graphene could be grown on germanium in the Si wafer-size
scale (that is, with the diameter of 300 mm or more), then some of the problems
associated with the graphene transfer to target Si wafer would be solved: in contrast
to atoms from the metal substrates used nowadays for graphene growth, Ge atoms
does not contaminate the production tools. This would open other technological
options for production of CMOS-compatible graphene-based devices.
Graphene can indeed be grown on Ge(001) substrates, including Ge(001) films
deposited by a CMOS-compatible technique on Si(001) wafers. Figure 8.18 illus-
trates this by some experimental data obtained for MBE graphene on samples cut
from patterned Ge(001)/Si(001) wafers. Germanium pillars have been embedded
into a SiO2 matrix and then exposed to atomic carbon in an MBE chamber. Raman
spectroscopy analysis shows that there is a dramatic difference in the quality of the
film obtained on the Ge pillars and on the surrounding SiO2. This is exemplified in
the upper panel of Fig. 8.18 by the 2D/G Raman mode ratio. The higher the ratio is,
the more does the film resemble a single-layer graphene; in principle, the ratio
8 Graphene for Silicon Microelectronics … 197

Fig. 8.18 MBE graphene on germanium. From top Map of 2D/G Raman mode intensity ratio, and
the optical microscope image of this area (Ge pillar surrounded by SiO2); high ratio (red) is seen
on Ge and indicates the presence of graphene, while low ratio (blue) is seen on SiO2 and indicates
a predominantly sp2-bonded carbon film. Atomic force microscope (AFM) image taken from the
region covered by graphene; the growth process resulted in roughening of the surface. High
resolution transmission electron microscopy (HR TEM) cross-section of a MBE graphene/Ge(001)
sample with a Ge overlayer deposited on top of the graphene after the growth of graphene was
complete; several graphene layers on top of crystalline Ge(001) can be recognized

should exceed 1 if the Raman signal originates in such a graphene.5 The film
obtained on germanium may be described as a single-layer graphene or a few-layer
graphene consisting of layers decoupled from one another [10], and the domain size
estimated from the Raman spectrum is some tens of nanometers. Scanning electron
microscopy images indicate the presence of much larger flakes of micrometer size
(Fig. 8.19); in addition, the straight edges of the graphene flakes seen in Fig. 8.19
suggest that the flakes consist of at most few domains of monocrystalline graphene.
The difference in the estimated domain size may be associated with the few-layer
structure of the graphene, as discussed further on: the flakes have grown on top of
the interface layer built of smaller graphene crystallites that is still not fully covered
by the flakes of better graphene and remains to be seen as the gray areas in
Fig. 8.19.
Here, we are most interested in the reason for the significant difference between
the growth of graphene on Ge(001) pillars and on the surrounding SiO2 planes.

5
The 2D/G ratio is not the only criterion here, and it is not the absolute one. For more discussion,
see [10] (the case of MBE graphene on Ge) and references therein.
198 J. Dabrowski et al.

Fig. 8.19 Scanning electron microscope (SEM) image of graphene on a Ge pillar such as that in
Fig. 8.18. Black areas are more conducting than the gray areas and are interpreted as graphene
flakes. The gray areas may be associated with a more polycrystalline film, possibly with Ge
intervening at grain boundaries

Fig. 8.20 Molecular dynamics simulation of the reaction of C with Ge(001). a The incoming
atom. b A Ge dimer breaks, the C atom lands under the surface. c The C atom may substitute a Ge
atom from the broken dimer; the other Ge can now diffuse on top of the surface. d Alternatively,
the Ge dimer survives and the C atom diffuses under the surface

Figure 8.20 uses the atomic structures obtained during ab initio molecular
dynamics simulation of the interaction of an incoming C atom with the Ge(001)
surface to illustrate two fundamental processes invoked by this interaction. When
the C atom hits the surface, about 4.4 eV is released. Part of this energy is received
by two Ge dimer atoms, and part is retained by the C atom. The Ge dimer initially
breaks, and the C atom finds itself under the surface (Fig. 8.20b). The C atom may
now substitute one of the atoms of the broken dimer, transferring the other one into
8 Graphene for Silicon Microelectronics … 199

a Ge ad-atom that diffuses on the surface (Fig. 8.20c). Alternatively, the Ge dimer
may survive; in this case, the C atom becomes an Interstitialcy that diffuses under
the surface. The probability that it as next substitutes a fourfold-coordinated Ge
atom is low (because the barrier for this process is high). The Interstitialcy avoids
also going deeper into the bulk of Ge; this is because lot of stress would be
generated. Since the GeC phase (in contrast to the SiC phase) is unstable and the C
Interstitialcy is mobile, longer deposition times results in the formation of graphene
flakes.
Besides the instability of Ge carbide, the key reason for the radically different
behavior of C on SiO2 (and also on Si) than on Ge is therefore good (sub-)surface
diffusivity of the C Interstitialcy under (and of small C clusters—on) the Ge(001)
surface. This is probably similar to the behavior of the C Interstitialcy under the Si
(001) surface; however, in the later case the SiC phase is stable and its formation
(presumably happening by the surface Si ejection process similar to the Ge ejection
from the Ge–Ge surface dimer) efficiently competes with the formation of graphene.
The ejection of surface Ge by C atoms leads to problems: the ejected Ge atoms
attach themselves preferably to the edges of the growing graphene. They may thus
remain in the grain boundaries, preventing the formation of larger graphene
crystals.
On the other hand. when a small graphene molecule is built by coalescence of
interstitialcies arriving from under the surface, it is likely to make as many possible
bonds to the substrate (Fig. 8.21a). This parallel configuration is the ground state of
the molecule. Tiny C clusters (dimers, trimers, and even hexamer strings) diffuse
easily as well—but on top of, not under the surface. Coalescence of C strings may
end up in a metastable molecule that is raised (Fig. 8.21b). Such cases are unfa-
vorable for a good graphene, but here the ejected Ge have a positive contribution:
they help to “glue” the molecule back to the substrate (Fig. 8.22).
Figure 8.23 contains a summary of the basic processes leading to nucleation and
growth of graphene on Ge(001) from atomic carbon. The kick-in process is the
reverse process to Ge ejection (or kick-out) by carbon and it produces a complete
Ge dimer and a subsurface C Interstitialcy. Note that kick-out may be initiated not
only by the incoming C atom, but also by a diffusing C Interstitialcy. For simplicity,
the contribution of carbon strings to the growing graphene is not shown in
Fig. 8.23.

Fig. 8.21 A small graphene molecule a lying and b standing on Ge(001)

200 J. Dabrowski et al.

Fig. 8.22 a A small graphene molecule, half-standing on and partially glued to Ge(001) by an
ejected Ge atom (red). b A piece of graphene with Ge atoms attached to its edge

Fig. 8.23 Schematic presentation of the graphene formation on Ge(001) from atomic carbon

The first layer of graphene on Ge(001) is therefore quite poor, presumably due to
its nanocrystalline nature stabilized by Ge atoms preventing good coalescence of
graphene islands. Polycrystallinity and Ge ejection lead also to the roughening of
the surface. Nevertheless, the density of such defects as standing graphene mole-
cules or highly disordered regions of sp2 carbon can be low enough to allow for this
interfacial layer to be overgrown by a much better graphene sheet [10].
Ejection of surface Ge appears to be largely an effect associated with the pres-
ence of atomic carbon. When hydrocarbon molecules are supplied instead, the
major diffusing species are—according to our ab initio calculations—such their
fragments as CH2 or C2H3 (depending on the precursor used). Consequently, there
seems to be the possibility to suppress the formation of the interface layer in a CVD
process; yet up to now, the interface layer might be there (Fig. 8.24, cf. the gray
areas).
The problem that is common to the MBE and the CVD growth of graphene on
germanium is the high substrate temperature needed to produce good graphene.
Significantly better graphene is grown at temperatures close to the melting point of
Ge, but at such conditions there is additional, strong surface roughening effect. The
major challenge is therefore to reduce the growth temperature down into the range
at which the Ge/Si(001) is stable against roughening induced by surface melting.
8 Graphene for Silicon Microelectronics … 201

Fig. 8.24 Scanning electron microscope (SEM) image of graphene grown in a cold wall CVD
chamber from C2H4. Black areas are graphene flakes. This is confirmed by the observation of the
Dirac cone in angle-resolved ultraviolet photoelectron spectroscopy (ARUPS) [28]

8.9 Summary and Conclusions

Physical mechanisms involved in the process of graphene growth on several sub-

strates are analyzed on the basis of theoretical (ab initio DFT) and experimental
data. The substrates considered here are: graphene, hexagonal boron nitride (hBN),
hexagonal SiO2 and mica, and Ge(001). The focus is on the deposition of atomic
carbon, apart from some remarks on CVD from hydrocarbons on Ge(001). A link to
the device design and fabrication is provided by sections on the deposition of Si on
graphene and on the performance analysis of a graphene base transistor (GBT).
C ad-atoms diffuse fast on graphene and on hBN. But this does not lead to
efficient self-nucleation, because the binding of trimers (C3) to such substrates is
practically null. In order to overgrow these materials with graphene, one must
provide high enough density of nucleation centers. Ab initio data can be used in a
simple formula derived by the random walk theory to estimate the collection zone
width w along a linear defect to which carbon sticks; from that one can obtain the
sticking coefficient r of C atoms as r ¼ w=L, where L is the average distance
between such absorbing defects. Sticking coefficient on high-quality substrates
turns out to be low, because w is small, due to desorption of ad-atoms at temper-
atures of interest.
Chemical interaction of C atoms with perfect graphene or perfect hBN is weak.
The opposite is true for SiO2-like substrates, including mica, where it leads to CO
and SiC formation. In spite of that, atomic C leaves large areas of mica unreacted,
and graphene is grown. This is attributable to (a) reflection of many C atoms back to
vacuum by a mechanism associated with internal excitation of C 2p electrons before
a chemical bond to the perfect substrate is formed, and (b) to surface re-oxidation
by O vacancy diffusion. This works on mica and should work on a perfect
hexagonal SiO2. But the growth on SiO2 and on similar substrates fails, probably
202 J. Dabrowski et al.

due to too high density of energy-dissipative centers, to which the carbon atoms
stick.
Atomic carbon interacts strongly with Ge(001) as well. However, there are two
key differences between Ge and SiO2. First, C atoms and small C clusters diffuse
efficiently on Ge(001), so that the crystallites that are already formed can collect the
carbon that fell down on the surface far away from them. Second, no stable carbide
phase is built by Ge and C; this is in contrast to stability of SiC. The end effect is
that Ge can be overgrown with good MBE graphene. The process is not unprob-
lematic, though. The interaction of C atoms with Ge(001) produces mobile Ge
atoms, which decorate edges of the growing graphene and may hinder the formation
of a wide-area graphene of sufficiently high sheet conductivity. As hydrocarbons do
not eject Ge ad-atoms, this may possibly be improved when CVD is used instead of
MBE.

Acknowledgments The authors wish to dedicate this work to the memory of Prof. Wolfgang
Mehr, the inventor and the pioneer of research on graphene base transistors. We thank our IHP
colleagues: Yuji Yamamoto for CVD growth of Ge(001) films on Si wafers, Oksana Fursenko for
taking AFM images, Markus Schubert for TEM and EDX characterization, Julia Kitzmann for
graphene transfer, Thomas Schroeder for discussions, ideas, and encouragement, and Andre Wolff
for technology support. Financial support from the European Commission through the GRADE
project (No. 317839) and computing time support from the Jülich Supercomputing Center of the
John von Neumann Institute for Computing (project hfo06) are gratefully acknowledged.

References

1. A.K. Geim, Graphene: Status and Prospects. Science 324, 1530–1534 (2009)
2. Y.M. Lin, C. Dimitrakopoulos, K.A. Jenkins, D.B. Farmer, H.Y. Chiu, A. Grill, P. Avouris,
100-GHz transistors from wafer-scale epitaxial graphene. Science 327, 662–662 (2010)
3. F. Schwierz, Graphene transistors. Nat. Nanotechnol. 5, 487–496 (2010)
4. H.C. Kang, H. Karasawa, Y. Miyamoto, H. Handa, T. Suemitsu, M. Suemitsu, T. Otsuji,
Epitaxial graphene field-effect transistors on silicon substrates. Solid State Electron. 54, 1010–
1014 (2010)
5. J. Kedzierski, P.L. Hsu, P. Healey, P.W. Wyatt, C.L. Keast, M. Sprinkle, C. Berger, W.A. de
Heer, Epitaxial graphene transistors on SiC substrates. IEEE Trans. Electron. Dev. 55, 2078–
2085 (2008)
6. K.S. Kim, Y. Zhao, H. Jang, S.Y. Lee, J.M. Kim, K.S. Kim, J.H. Ahn, P. Kim, J.Y. Choi, B.H.
Hong, Large-scale pattern growth of graphene films for stretchable transparent electrodes.
Nature 457, 706–710 (2009)
7. G. Lupina, J. Kitzmann, I. Costina, M. Lukosius, C. Wenger, A. Wolff, S. Vaziri, M. Östling,
I. Pasternak, A. Krajewska, W. Strupinski, S. Kataria, A. Gahoi, M.C. Lemme, G. Ruhl, G.
Zoth, O. Luxenhofer, W. Mehr, Residual metallic contamination of transferred chemical vapor
deposited graphene. ACS Nano 9, 4776–4785 (2015)
8. A. Malesevic, R. Vitchev, K. Schouteden, A. Volodin, L. Zhang, G. Van Tendeloo, A.
Vanhulsel, C. Van Haesendonck, Synthesis of few-layer graphene via microwave PECVD.
Nanotechnology 19, 305604 (2008)
9. G. Wang, M. Zhang, Y. Zhu, G. Ding, D. Jiang, Q. Guo, S. Liu, X. Xie, P.K. Chu, Z. Di, X.
Wang, Direct growth of graphene film on germanium substrate. Sci. Rep. 3, 2465 (2013)
8 Graphene for Silicon Microelectronics … 203

10. G. Lippert, J. Dabrowski, T. Schroeder, M.A. Schubert, Y. Yamamoto, F. Herziger,

J. Maultzsch, J. Baringhaus, C. Tegenkamp, M.C. Asensio, J. Avila, G. Lupina, Graphene
grown on Ge(001) from atomic source. Carbon 75, 104–112 (2014)
11. G. Lupina, J. Kitzmann, M. Lukosius, J. Dabrowski, A. Wolff, W. Mehr, Deposition of thin
silicon layers on transferred large area graphene. Appl. Phys. Lett. 103, 263101 (2013)
12. F. Driussi, P. Palestri, L. Selmi, Modeling, simulation and design of the vertical graphene base
transistor. Microelectron. Eng. 109:338–341; S. Venica, F. Driussi, P. Palestri, D. Esseni, S.
Vaziri, L. Selmi (2014) Simulation of DC and RF performance of the graphene base transistor.
IEEE Trans. Electron Devices 61, 2570–2576 (2013)
13. W. Mehr, J. Dabrowski, J.C. Scheytt, G. Lippert, Y.H. Xie, M.C. Lemme, M. Östling, G.
Lupina, Vertical graphene base transistor. IEEE Electron Device Lett. 33, 691–693 (2012)
14. S. Vaziri, G. Lupina, C. Henkel, A.D. Smith, M. Östling, J. Dabrowski, G. Lippert, W. Mehr,
M.C. Lemme, A Graphene-Based Hot Electron Transistor. ACS Nano Lett. 13:1435–1439;
Vaziri S, Lupina G, Paussa A, Smith AD, Henkel C, Lippert G, Dabrowski J, Mehr W, Östling
M, and Lemme MC (2013) A manufacturable process integration approach for graphene
devices. Solid-State Electronics 84, 185–190 (2013)
15. J. Dabrowski, G. Lippert, T. Schroeder, G. Lupina, Role of defects in the process of graphene
growth on hexagonal boron nitride from atomic carbon. Appl. Phys. Lett. 105, 191610 (2014)
16. G. Lippert, J. Dabrowski, M.C. Lemme, C. Marcus, O. Seifarth, G. Lupina, Direct graphene
growth on insulator. Phys. Status Solidi B 248, 2619–2622 (2011)
17. G. Lippert, J. Dabrowski, Y. Yamamoto, F. Herziger, J. Maultzsch, J. Baringhaus, C.
Tegenkamp, M.C. Lemme, W. Mehr, G. Lupina, Molecular beam epitaxy of graphene on
mica. Phys. Status Solidi B 249, 2305–2312 (2012)
18. G. Lippert, J. Dabrowski, Y. Yamamoto, F. Herziger, J. Maultzsch, M.C. Lemme, W. Mehr,
G. Lupina, Molecular beam growth of micrometer-size graphene on mica. Carbon 52, 40–48
(2013)
19. P. Giannozzi et al., QUANTUM ESPRESSO: a modular and open-source software project for
quantum simulations of materials. J. Phys. Cond. Mat. 21, 395502 (2009)
20. J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made simple. Phys.
Rev. Lett. 77, 3865–3868 (1996)
21. G. Henkelman, B.P. Uberuaga, Hannes Jónsson, A climbing image nudged elastic band
method for finding saddle points and minimum energy paths. J. Chem. Phys. 113, 9901–9904
(2000)
22. H. Jónsson, G. Mills, K.W. Jacobsen, Nudged Elastic Band Method for Finding Minimum
Energy Paths of Transitions. in Classical and Quantum Dynamics in Condensed Phase
Simulations eds. by B.J. Berne, G. Ciccotti, D.F. Coker (World Scientific, Singapore, 1998),
p. 385
23. C. Ataca, S. Ciraci, Functionalization of BN honeycomb structure by adsorption and
substitution of foreign atoms. Phys. Rev. B 82, 165402 (2010)
24. N. Berseneva, A. Gulans, A.V. Krasheninnikov, R.M. Nieminen, Electronic structure of boron
nitride sheets doped with carbon from first-principles calculations. Phys. Rev. B 87, 035404
(2013)
25. A. van de Walle, M. Asta, P.W. Voorhees, First-principles calculation of the effect of strain on
the diffusion of Ge adatoms on Si and Ge(001) surfaces. Phys. Rev. B 67, 041308 (2013)
26. J. Dabrowski, A. Fleszar, G. Lippert, G. Lupina, Ab initio Modeling of Growth of Grapene for
Silicon-Compatible Microelectronics, in Proceedings of NIC Symposium, 12–13 February
2014, ed. by K. Binder, G. Mnster, M. Kremer (Jülich, Germany, Forschungszentrum Jülich,
Jülich, 2014), p. 207
27. G. Lippert, J. Dabrowski, W. Mehr, G. Baccarani, A. Gnudi, V. Di Lecce (2015) Low band
gap graphene transistor. European patent application 14194150.0
28. M.C. Asensio, J. Avila, J. Baringhaus, I. Colambo, J. Dabrowski, G. Lippert, G. Lupina (2015)
unpublished
Chapter 9
Recent Progress on Nonlocal
Graphene/Surface Plasmons

Norman J.M. Horing, A. Iurov, G. Gumbs, A. Politano

and G. Chiarello

Abstract We review recent experimental and theoretical studies of the non-local

plasmon dispersion relations of both single and double layers of graphene which are
Coulomb-coupled to a thick conducting medium. High-resolution electron energy
loss spectroscopy (HREELS) was employed in the investigations. A mean-field
theory (R.P.A.) formulation was used to simulate and explain the experimental
results, with the undamped plasmon excitation spectrum calculated for arbitrary
wave number. Our numerical calculations show that when the separation a between
a graphene layer and the surface is less than a critical value ac = 0.4k−1
F , the lower
acoustic plasmon is overdamped. This result seems to explain the experimentally
observed behavior for the plasmon mode intensity as a function of wave vector. The
damping, as well as the critical distance, changes in the presence of an energy
bandgap for graphene. We also report similar damping features of the plasmon
modes for a pair of graphene layers. However, the main difference arising in the
case when there are two layers is that if the separation between the layer nearest the
surface and the surface is less than ac, then both the symmetric and antisymmetric
modes become damped, in different ranges of wave vector.

N.J.M. Horing (&)

Department of Physics, Stevens Institute of Technology, Hoboken, NJ 07030, USA
e-mail: [email protected]
A. Iurov
Center for High Technology Materials, University of New Mexico,
Albuquerque, NM 87106, USA
G. Gumbs
Department of Physics, Hunter College, Cuny, NY 01065, USA
G. Gumbs
Donostia International Physics Center (DIPC), P de Manuel Lardizabal, 4,
20018 San Sebastian, Basque Country, Spain
A. Politano
G. Chiarello
Dipartimento di Fisica, Universitá degli Studi della Calabria, 87036 Rende (Cs), Italy

© Springer International Publishing Switzerland 2016 205

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_9
206 N.J.M. Horing et al.

9.1 Introduction

This review article is concerned with plasma phenomenology in graphene [1–8];

both nonlocal as well as local features are discussed, including wavevector shifting
of plasmon modes and Landau damping. It is hard to overstate the importance of
graphene (and similar materials, such as silicene [9]). As an ultrathin layer of
carbon/graphite just a single atom thick in a planar hexagonal honeycombed array,
graphene has an energy spectrum proportional to linear momentum that is char-
acteristic of relativistic Dirac electrons with no gap between the conduction and
valence bands and, correspondingly, no mass in its native state. It has been the
subject of truly extraordinary research efforts because of its great potential to bring
about a massive change in the electronic device and computer industries based on
its exceptional “device-friendly” electrical conductivity and sensing properties. The
electrical mobility in graphene reaches up to 200,000 cm2 Vs−1, over two orders of
magnitude greater than that of present silicon-based materials, over 20 times that of
gallium arsenide and over twice that of indium antimonide. It has a long
mean-free-path (l  400 nm at room temperature) and it is stable to 3000 K, as well
as having great strength. Its planar form is consonant with well-developed present
device fabrication techniques. Graphene-based chemical sensors can detect one part
per billion of various active gases with high sensitivity based on adsorbed mole-
cules acting as donors/acceptors, jointly with graphene’s high conductivity and low
noise. These, and other features of graphene, have commanded the attention of the
scientific/engineering world, generating a massive flood of research activity.
Graphene’s electromagnetic/optical response features are also important and its
dielectric and plasmonic properties (local and nonlocal) play a major role in this. As
indicated above, this subject will be discussed in detail here in two important
experimental configurations: (1) free-standing graphene, and (2) graphene
Coulomb-coupled with a slab of substrate plasma. The pertinent equation for the
fundamental inverse dielectric “screening” function of the latter involves the cor-
responding function for the slab of substrate plasma alone: we start the discussion
with the theoretical background for that since it may be expected to play an
important role in further future studies. The slab plasma discussion is followed by a
review of the present graphene plasma literature and recent theoretical results.
Finally, and most importantly, there is a thorough discussion of the subject from an
experimental point of view.

9.2 Nonlocal Dielectric Response of a Slab Plasma

Plasma nonlocality has been a subject of great interest in both classical and quantum
physics. It is important in dynamic plasma screening phenomenology, including the
plasmon spectrum, intensity and natural damping, as well as in the static shielding of
charge interactions. In low dimensional systems, the boundaries involved can signif-
icantly affect such features of plasma dynamics, including all aspects of nonlocal
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 207

behavior. Moreover, the associated geometric constraints have become ever more
important as semiconductor device sizes have diminished, now at the nanometer level
and even entering the regime of a “single atom thick” layer, such as graphene and
silicene. Of course, “thick” and semi-infinite plasma layers also involve nonlocal
plasmas, and their dielectric response properties will be discussed here on both a
stand-alone basis, as well as in the role of a substrate interacting with a thin
two-dimensional sheet of graphene. This review will address both the theoretical and
experimental aspects of such a nonlocal system starting with the analysis of the
dynamic, nonlocal inverse dielectric response function K(1, 2) of a finite slab of plasma
in the random phase approximation (RPA) ð1 ¼ r1 ; t1 ; 2 ¼ r2 ; t2 Þ. This function is the
space-time matrix inverse of the direct dielectric function ɛ(1, 2), such that the effective
potential V(1) generated by an applied potential U(2) is given by [10–14].
Z
Vð1Þ ¼ d2Kð1; 2ÞUð2Þ or Kð1; 2Þ ¼ dVð1Þ=dUð2Þ; ð9:1Þ

for a linear medium. Since V(1) is developed by the density perturbation ρ(1) due to
polarization of the plasma (with interparticle coulomb potential v) by U(2), it is
given by
Z
Vð1Þ ¼ Uð1Þ þ d3 vð1  3Þqð3Þ; ð9:2Þ

leading to the integral equation (δ(1 − 2) is a space-time Dirac delta function):

Z
dVð1Þ dqð3Þ
Kð1; 2Þ ¼ ¼ dð1  2Þ þ d3 vð1  3Þ
dUð2Þ dUð2Þ
Z Z ð9:3Þ
dqð3Þ dVð4Þ
¼ dð1  2Þ þ d4 d3 vð1  3Þ ;
dVð4Þ dUð2Þ

or
Z
Kð1; 2Þ ¼ dð1  2Þ  d4 að1; 4ÞKð4; 2Þ ð9:4Þ

where the polarizability α(1, 4) is defined as

Z
dqð3Þ
að1; 4Þ ¼  d3 vð1  3Þ ð9:5Þ
dVð4Þ

and

dqð3Þ
Rð3; 4Þ ¼ ¼ iG1 ð1; 2ÞG1 ð2; 1 þ Þ ð9:6Þ
dVð4Þ
208 N.J.M. Horing et al.

is the “ring diagram” of the density perturbation response function R, expressed in

terms of noninteracting one-electron thermodynamic Green’s functions G1 within
the random phase approximation (RPA) [15]. Such RPA results for the density
perturbation response function and the effective potential, as well as the inverse
dielectric function, for three-, two- and one-dimensional plasmas based on the use
of noninteracting one-electron Green’s functions in (9.6) are very well known. Our
focus here will be on a thick plasma slab (including the semi-infinite plasma limit)
in which it is embedded in a substrate in the vicinity of a thin two-dimensional
plasma such as graphene or silicene [9].
Of course, plasma boundaries have been the subject of intensive investigation in
the past. Older reviews, articles and books by Ritchie [16], Raether [17, 18],
Feibelman [19], Liebsch [20], Kushwaha [21] and others summarize a great deal of
work on the role of boundaries and nonlocality in thick bounded plasmas.
A particularly informative study was carried out by Newns [22], which was further
elaborated by Horing [23] in the determination of antisymmetric (A) and symmetric
(S) effective potentials V(A,S) generated by corresponding applied potentials U(A,S)
associated with sources ∇22U(2) = 4πS(2) [note that the convention ∇22U(2) → 4πS(2)
employed here differs from the usual one in the sign of S(2)]. In this analysis for
infinite-barrier-model boundaries, the effective potentials V(A,S) (as well as S(A,S)) of a
slab of plasma in the region 0 ≤ z ≤ d/2 are continued antisymmetrically (A) or
symmetrically (S) into the adjoining region d/2 < z < d across the midpoint z = d/2;
and it yields results in a mixed ðz; QÞ and ω representation given as follows for the
regions z ≤ 0 and 0 ≤ z ≤ d [note that translational invariance in the lateral (x, y) plane
permits representation in terms of a single conjugate lateral wavevector
R ¼ ðx1  x2 ; y1  y2 Þ ! Q ¼ ðQx ; Qy Þ]:
n h i o
ðA;SÞ ðA;SÞ IðA;SÞ IðA;SÞ
VQ ðz; xÞ ¼ g þ ðzÞ eQz VQ ð0; xÞ  UQ ð0; xÞ þ UQ ðz; xÞ
2
 X X   h
4 4 ðA;SÞ
þ g þ ðzÞg þ ðd  zÞ  cos qðA;SÞ z EQqq
1
0 2pSQq0 ðxÞ
d ðA;SÞ q0 ðA;SÞ
q
ii
ðA;SÞ IðA;SÞ
þ QVQ ð0; xÞ  2QUQ ð0; xÞ ;
ð9:7Þ

where η+(z) is the Heaviside unit step function and

Z0
2p 0 ðA;SÞ
UQ ðz; z0 ; xÞ dz0 eQjzz j SQ ðz0 ; xÞ
IðA;SÞ
¼ for z\0; ð9:8Þ
Q
1

represents the potential in the region where z < 0 due to that part of the source
Q (z′; ω) lying within z < 0 and
charge S(A,S)
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 209

0 1
 1 8p X X
ðA;SÞ ðA;SÞ @2U IðA;SÞ ð0; xÞ  eðA;SÞ 1 ðA;SÞ A
VQ ð0; xÞ ¼ 1 þ eQ Q EQqq0 SQq0 ðxÞ
d Q 0
ðA;SÞ q ðA;SÞ q

ð9:9Þ

and
0 11
ðA;SÞ 4Q X X
 eQ ðxÞ ¼ @
ðA;SÞ
eQ E 1 0 A : ð9:10Þ
d ðA;SÞ q0 ðA;SÞ Qqq
q

In these equations above we have

 
EQqq0  EQqq0 ðxÞ ¼ Q2 þ q2 dqq0 =gq þ 4pRQqq0 ð9:11Þ
 
¼ 4p DQq dqq0 =gq  AQqq0 ; ð9:12Þ

with

RQqq0 ¼ DQq dqq0 =gq  AQqq0 ð9:13Þ

as the density-perturbation response function R(1, 2) = δρ(1)/δV(2) in (Q, q, q′)-

representation having a diagonal part D and a non-diagonal part −A. In this,
ΔQq = (4π)−1(Q2 + q2) + DQq. The notations (A, S) refer to antisymmetric or
symmetric potentials and sources, and for the antisymmetric case, q(A), q
′(A) = (2n + 1)π/d, n = 0, 1, 2, …, ∞ and ηq ≡ 1, whereas for the symmetric case,
q(S), q′(S) = 2nπ/d, n = 0, 1, 2, 3…, ∞ and ηq = 1 for q > 0, but gq ¼ 12 for q = 0. Both
of the Fourier series sums over qðAÞ ¼ ð2n þ 1Þ pd and over qðSÞ ¼ 2np d represent the
same function in the half interval ½0  z \ d2, but in the other half interval
½d2 \ z  d, they represent antisymmetric or symmetric continuations, respec-
tively, across z ¼ d2. The Fourier series transforms employed here are defined as

r ¼ ðR; zÞ and q ¼ ðQ; qÞ :
Z
ðA;SÞ 2X d2 Q ðA;SÞ
f ðR; zÞ ¼ g eiQ
R cosðqðA;SÞ zÞfQq ;
d ðA;SÞ q ð2pÞ 2
q

Zd Z ð9:14Þ
ðA;SÞ ðA;SÞ ðA;SÞ
and fQq ¼ dz d Re2 iQ
R
cosðq zÞfQq ðR; zÞ:
0

The diagonal part ðDQq dqq0 =gq Þ of the density perturbation response function
(R) corresponds to the familiar noninteracting Fermi gas result subject to periodic
boundary conditions, whereas the non-diagonal part (AQq) arises from boundary
210 N.J.M. Horing et al.

Fig. 9.1 Slab plasma in region II where z = [0, d/2) with potential V continued across z = d/2 into
region z = [d/2, d]: a V(S) symmetrically continued potential; b V(A) antisymmetrically continued
potential; c V average of the symmetric potential V(S) and antisymmetric potential V(A) with equal
weight as d → ∞(d/2 → ∞)

conditions forcing the electron Green’s functions G1 to vanish at the boundaries

(having effects localized within small spatial regions limited to an inverse Fermi
wavenumber near the boundaries), for the infinite barrier model considered here.
While each of the two Fourier series for V(A) Q (z, ω) and VQ (z, ω) individually
(S)

properly represent the effective potential in the slab region 0 ≤ z ≤ d/2, as generated
by corresponding applied potentials U(A) and U(S) and associated sources, it is
necessary to average them to eliminate the appearance of spurious sources in the
continuation region d/2 < z < d, which is distant for large d (Fig. 9.1). Thus, we
have

1 h ðAÞ ðSÞ
i 1X
ðA;SÞ
VQ ðz; xÞ ¼ VQ ðz; xÞ þ VQ ðz; xÞ ¼ V ðz; xÞ for large d ð9:15Þ
2 2 AþS Q

and

1 X ðA;SÞ
UQ ðz; xÞ ¼ U ðz; xÞ: ð9:16Þ
2 AþS Q

It is at once apparent that (9.1) yields K(1, 1′) directly upon choosing the Dirac
delta function U(2) = δ(2 − 1′) as a model potential,
Z Z
V10 ð1Þ ¼ dð2ÞKð1; 2ÞUð2Þ ¼ dð2ÞKð1; 2Þdð2  10 Þ ¼ Kð1; 10 Þ: ð9:17Þ

Using this, one can describe K(1, 1′) for a thick slab of plasma with planar
boundaries by using (9.7) for V(A,S)(1) jointly with (9.15), generated by the math-
ematical model potential U(2) = δ(2 − 1′). These considerations yield VQ(z, ω) for a
~ Qqq0 ¼ gq EQqq0 )
thick slab as (define E
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 211

" ! #
1 X ðA;SÞ IðA;SÞ IðA;SÞ
VQ ðz; xÞ ¼ g þ ðzÞ e Qz
VQ ð0; xÞ  2UQ ð0; xÞ þ 2UQ ðz; xÞ
2 AþS
2
1 X 4 X X   h
 g þ ðzÞ 4 cos qðA;SÞ z E~ Qqq
1 ðA;SÞ
0 2pSQq0 ðxÞ
2 AþS
d
qðA;SÞ q0 ðA;SÞ
ii
ðA;SÞ IðA;SÞ
þ QVQ ð0; xÞ  2QUQ ð0; xÞ ;
ð9:18Þ
1
where terms of order d are neglected (but could be important in a calculation of
surface energy), and
0 1
  8p ðA;SÞ X X ~ 1 ðA;SÞ
ðA;SÞ ðA;SÞ 1 @
EQqq0 SQq0 ðxÞA
IðA;SÞ
VQ ð0; xÞ ¼ 1 þ eQ 2UQ ð0; xÞ  eQ
d ðA;SÞ 0ðA;SÞ
q q

ð9:19Þ

and
 1 4Q X X ~ 1
ðA;SÞ
eQ ¼ E 0: ð9:20Þ
d ðA;SÞ q0 ðA;SÞ Qqq
q

The resulting solution for the inverse dielectric function Kð1; 10 Þ ¼ V10 ð1Þ of a
thick slab, neglecting terms of order O(1/d), may be written in a mixed represen-
tation as
" !
1 X ðA;SÞ
0 1 0 0 0
KðQ; z; z ; xÞ ¼ g þ ðzÞ e Qz
VQ ð0; xÞ  Q d ðz Þ  dðz Þ
2 AþS
i
þ Q1 d0 ðz0 ÞeQjzj þ dðz0 ÞHðzÞeQjzj þ 2g þ ðz0 Þdðz  z0 Þ
1 X 4 X X   h
ðA;SÞ
 g þ ðzÞ cos qðA;SÞ z E~ Qqq
1
0 QVQ ð0; xÞ
2 AþS
d
qðA;SÞ q0 ðA;SÞ

 ðq02 þ Q2 Þ cosðq0 ðA; SÞz0 Þg þ ðz0 Þ  Qdðz0 Þ
ð9:21Þ
212 N.J.M. Horing et al.

where Θ(z) ≡ η+(z) – η + (−z) and δ′(z) ≡ d(δ(z))/dz, with

8
< 8p ðA;SÞ X X ~ 1
ðA;SÞ ðA;SÞ
VQ ð0; xÞ ¼ ð1 þ eQ Þ1 Q1 d0 ðz0 Þ þ dðz0 Þ  eQ EQqq0
: d
qðA;SÞ q0ðA;SÞ ð9:22Þ

0 0  02  
d ðz Þ q þ Q2
  cosðq0ðA;SÞ z0 Þg þ ðz0 Þ :
2p 2p

Equations (9.21, 9.22, 9.17) above include the role of non-diagonal density
response matrix elements, AQqq0 as well as the diagonal ones, DQqq0 dqq0 =gq .
Considering the semi-infinite limit d → ∞, it is reasonable to neglect the
nondiagonal elements AQqq0 as well as terms of order O(1/d), and passing to the
P R
d 1
continuum limit qðA;SÞ ! 2p 0 dq (also ∑A+S → 2), we have

  eQ Qz 0
K Q; z; z0 ; x ¼ g þ ðzÞ dðz  z0 Þ  e dðz Þ
1 þ eQ

2eQ  
þ 3D
K1 Q; z0 ; x eQz g þ ðz0 Þ
1 þ eQ

 
  QeQ 0 2QeQ 3D  0
 0
þ g þ ðzÞ v3D
1 Q; z; x dðz Þ  K Q; z ; x g þ ðz Þ
1 þ eQ 1 þ eQ 1
 3D    
þ K1 Q; z þ z0 ; x þ K1 3D
Q; z  z0 ; x g þ ðz0 Þ ;
ð9:23Þ

where

Z1
ðA;SÞ 2Q  1
ðeQ Þ1 ¼ e1 ¼ dq ðQ2 þ q2 Þe3D
1 ðq; xÞ ; ð9:24Þ
Q
p
0

and e3D
1 ðq; xÞ (3D denotes three dimensional) is the dynamic, nonlocal bulk dielectric
function of the thick-slab material. Furthermore, (9.23) involves the definitions

Z1
 0
 1
3D
K1 Q; z ; x ¼ p dq cosðqz0 Þ=e3D
1 ðq; xÞ ð9:25Þ
0

and

Z1
  
v3D
1 Q; z ; x ¼ 2p1
0
dq cosðqz0 Þ= ðq2 þ Q2 Þe3D
1 ðq; xÞ : ð9:26Þ
0

A fuller account of the derivation of these results is provided in [21].

9 Recent Progress on Nonlocal Graphene/Surface Plasmons 213

If the bulk plasma within the semi-infinite slab is fully local in the sense that
1 ðq; xÞ ! e1 ðxÞ  eðxÞ, then the corresponding local inverse dielectric func-
e3D 3D

tion is given by

 
0 0 0 1  eðxÞ
KðQ; z; z ; xÞ ¼ g þ ðzÞ dðz  z Þ þ dðz Þe Qz
1 þ eðxÞ

0
  ð9:27Þ
dðz  z Þ 0 Qz 1 eðxÞ  1
þ g þ ðzÞ þ dðz Þe :
eðxÞ eðxÞ eðxÞ þ 1

While (9.27) is useful because of its relative simplicity, it is important to

understand that its validity is restricted because the q-integrations in (9.24–9.26)
extend over an infinite integration range. This admixes the effects of the
boundary/image length scale with that of the q-nonlocality dependence, eliminating
the possibility of an unqualified limit q → 0 and modifying the plasmon depen-
dence on Q: Furthermore, the imaginary part of e3D 1 ðq; xÞ participates in these q-
integrations, imparting a damping contribution to the surface plasmon roots even in
the low-Q limit (which is not represented in (9.27)). In fact, the “nonlocal” linear Q-
correction to the surface plasmon, and its imaginary part involving damping,
depend sensitively on the nature of the bounding surface. Nevertheless, there is a
regime in which (9.27) is meaningful. This can be seen by examining the parameter
measuring the importance of nonlocality in the bulk dielectric function e3D 1 ðq; xÞ,
 1=2
namely pM c  mxp =Ethermal
2
, where the characteristic thermal energy
Ethermal = μ is the Fermi energy in the degenerate case at hand. For q ≪ pM c it is
reasonable to neglect nonlocality, at least in the surface plasmon frequency xs ¼
pffiffiffi
xp = 2 as well as in comparison with other more pertinent sources of nonlocal
behavior (but it cannot be neglected in the damping of the surface plasmon). For
Cu, xp  1:66  1016 s1 , we have the critical nonlocality wavenumber parameter
as pM c  10
10
m−1. By the way of comparison, the corresponding critical
wavenumber for massless graphene, measured by pG
pffiffi c  pF EF =hxp with
2 1
xp ¼ ð3:22  1019 ÞpG c
1=2
, is given by p G
c  4  10 m : Thus, while nonlo-
cality is important in graphene for wavenumbers in the vicinity of (and above) pG c , it
10 1
is relatively unimportant in surface plasmon phenomenology up to pM c  10 m .
The high wavenumber range between pG M
c and pc will be discussed in detail here
taking account of graphene nonlocality, but not that of the metal. Initially, we
review the fully local range for wavenumbers well below both pG M
c and pc (which, of
course, excludes intrinsically nonlocal damping).
Note that henceforth the symbols Q and q| for lateral wavevector are used
interchangeably, Q ≡ q|.
214 N.J.M. Horing et al.

9.3 2D Plasma Coulomb-Coupled with a Slab Plasma

To address the problem of a two dimensional sheet of plasma interacting with a slab
of substrate plasma, it is reasonable to expect that the perturbed densities of the two
are mutually independent and additive, ρ = ρSl + ρ2D, to the extent that tunneling
between the two plasmonic systems is negligible. Accordingly, the polarizabilities
are additive and the inverse dielectric function K for the composite system is given
by (subscripts “Sl” and “2D” refer to the slab plasma and 2D sheet, respectively)

K 1 ¼ e ¼ 1 þ aSl þ a2D  eSl þ a2D ¼ KSl1 þ a2D ð9:28Þ

or

K ¼ KSl  KSl
a2D
K; ð9:29Þ

where KSl refers to the inverse dielectric function of the slab plasma alone, as
discussed above. This integral equation is readily solved in position representation
(bear in mind that the symbol for lateral wavenumber, formerly Q above, will now
be q|| [Q ≡ q||], which we supress along with frequency ω):

Z1 Z1
0
Kðz1 ; z2 Þ ¼ KSl ðz1 ; z2 Þ  dz dz00 KSl ðz1 ; z0 Þa2D ðz0 ; z00 ÞKðz00 ; z2 Þ: ð9:30Þ
1 1

Simplification occurs because localization of the 2D sheet to the plane at

z = a means that

Z1
0 00
a2D ðz ; z Þ ¼ dz000 vðz0 ; z000 ÞDðz000 ; z00 Þ ð9:31Þ
1

where (v is the Coulomb potential)

ð0Þ ð0Þ
Dðz000 ; z00 Þ ¼ P2D ðqjj ; xÞdðz000  aÞdðz00  aÞ  R2D ðqjj ; xÞdðz000  aÞdðz00  aÞ;
ð9:32Þ

with Π(0)
2D(q||, ω) ≡ − R2D(q||, ω) as the 2D ring diagram of the RPA for the 2D
(0)

plasma sheet. This yields

Z1
ð0Þ
Kðz1 ; z2 Þ ¼ KSl ðz1 ; z2 Þ  P2D ðqjj ; xÞ dz0 KSl ðz1 ; z0 Þvðz0  aÞKða; z2 Þ ð9:33Þ
1

and the solution for K(a, z2) may be obtained algebraically as

9 Recent Progress on Nonlocal Graphene/Surface Plasmons 215

KSl ða; z2 Þ
Kða; z2 Þ ¼ ; ð9:34Þ
SC ðqjj ; xÞ

where the “inverse dispersion factor” SC(q||, ω), whose zeros determine the coupled
plasmon frequencies (and amplitudes) of the composite system is given by
8 1 9
<Z =
ð0Þ
SC ðqjj ; xÞ  1 þ P2D ðqjj ; xÞ dz0 KSl ða; z0 Þvðz0  aÞ : ð9:35Þ
: ;
1

The resulting inverse dielectric function for the composite system is given in
mixed z-q||-ω representation by [12, 14, 24]
8 1 9
K ða; z Þ <Z =
ð0Þ
dz0 KSl ðz1 ; z0 Þvðz0  aÞ :
Sl 2
Kðz1 ; z2 Þ ¼ KSl ðz1 ; z2 Þ  P2D ðqjj ; xÞ
SC ðqjj ; xÞ : ;
1
ð9:36Þ

The fully local composite system result for K(z1, z2) of a semi-infinite substrate
plasma coupled with a 2D sheet plasma is obtained by employing KSl from (9.25) in
(9.35) and (9.36). Correspondingly, the interacting plasmon roots arise from the
vanishing of SC(q||, ω) of (9.34) using the local limit of Π(0)
2D(q||, ω) given as [25]

 
ð0Þ 2l D2
P2D ðqjj ; xÞ Cq2jj =x2 with C¼ 1  ; ð9:37Þ
ph2 l2

which accomodates both gapless as well as gapped graphene, with Δ as the gap
between the valence and conduction bands. The fully local plasmon modes follow
as (ɛs is the background dielectric constant)
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

  2
pe2 C x 2 pe2 Cx2p 2aq xp 2 Ce2 p
p
x ¼
2
qjj þ
e jj q þ
jj  q jj ; ð9:38Þ
es 2 es 2 es

and for large separation aq|| → ∞, the plasmons disengage (ωs is the decoupled
surface plasmon and ω22D is the decoupled 2D plasmon)

x2p 2pe2 C
x2þ ¼  x2s ; x2 ¼ qjj  x22D ; ð9:39Þ
2 es

whereas small separation yields the fully coupled hybridized modes as

216 N.J.M. Horing et al.

x22D x2s x22D x2s

x2þ ¼ x2s þ x22D  2aqjj ; x2 ¼ 2aqjj ; ð9:40Þ
x2s þ x22D x2s þ x22D

with plasmon linearity exhibited at low wavenumber (q||a ≪ 1) as

rffiffiffiffiffiffiffiffiffi
pCe2 paC
xþ xs þ qjj ; x 2e qjj : ð9:41Þ
es x s es
1
Both of these modes are linear in lateral wavenumber, differing from the q2jj -
dependence for free-standing graphene or a 2DEG [25].
Generalizing our considerations to include an arbitrary number N of 2D layers
which are Coulomb-coupled to a semi-infinite substrate as well as to each other, we
have
8 9
N <
X Z1 =
ð0Þ
djj0 þ P2D;j ðqk ; xÞ dz0 KSI ðaj0 ; z0 Þvðz0  aj Þ Kðaj ; z2 Þ ¼ KSI ðaj0 ; z2 Þ:
j¼1
: ;
1
ð9:42Þ

Employing the local approximation for KSI, i.e.




0 0 1  eB ðxÞ
0 qk z
KSI ðz; z ; qk ; xÞ ¼ hðzÞ dðz  z Þ þ dðz Þe
1 þ eB ðxÞ
 0


dðz  z Þ 0 qk z 1 eB ðxÞ  1
þ hðzÞ þ dðz Þe ;
eB ðxÞ eB ðxÞ eB ðxÞ þ 1

leads to the matrix form

0 1 0 1
Kða1 ; z2 Þ KSI ða1 ; z2 Þ
B Kða2 ; z2 Þ C 1 $ ðNÞ B KSI ða2 ; z2 Þ C
B C¼ N ðqjj ; xÞB C ð9:43Þ
@


A ðNÞ @


A
Sc ðqjj ; xÞ
Kða2 ; N2 Þ KSI ðaN ; z2 Þ

ðNÞ $ ðNÞ
where Sc ðqk ; xÞ ¼ Det M ðqk ; xÞ with matrix elements given by
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 217

Z1
ðNÞ ð0Þ
Mjj0 ðqk ; xÞ ¼ djj0 þ P2D;j ðqk ; xÞ dz0 KSI ðaj0 ; z0 Þvðz0  aj Þ
1
2pe2 ð0Þ ð9:44Þ
¼ djj0 þ P ðqk ; xÞ
es qjj 2D;j
" #
x2p
 eqk jaj aj0 j þ eqk ðaj þ aj0 Þ :
2x2  x2p

$ ðNÞ $ ðNÞ
and N ðqk ; xÞ is the transpose cofactor matrix of M ðqk ; xÞ.
To address the role of graphene nonlocality in the coupled plasmon spectrum in
the wavenumber region between pG M
c and pc , the nonlocal graphene polarizability
ring diagram may be employed in (9.36) and (9.35), with KSl again taken to be local
as KSl of (9.25). The nonlocal gapless graphene polarizability was derived by
Wünsch et al. [26] and by Hwang and Das Sarma [27] in the degenerate limit; for
convenience it is exhibited in Appendix 1. The corresponding result for gapped
graphene was derived by Pyatkovskij [25] and is exhibited in Appendix 2.

9.4 Numerical Results for Plasmon Dispersion: Graphene

Layers Interacting with Semi-infinite Conductor

Calculated numerical results for the plasmon dispersion of a system consisting of a

semi-infinite conducting medium which is Coulomb-coupled to a single layer of
ungapped graphene are shown in Fig. 9.2, and for two layers they are exhibited in
Fig. 9.3. We note that the plasmon spectra and Landau damping both depend
crucially on the separation between the constituents. Landau damping is determined
by the imaginary part of the polarization function. For a single layer, Fig. 9.2
illustrates that when the plasmon mode enters a region with ℑmΠ(0)(q, ω) ≠ 0, the
mode is damped. Our calculations also show that when the distance a is less than a
critical value ac = 0.4k−1
F , the lower acoustic plasmon is overdamped. On the other
hand, our calculation shows that when the conducting surface-MLG separation
exceeds the critical distance ac, the intensity of the low-frequency plasmon in the
long wavelength regime is high, and may be detected up to some cut-off wave
vector qc. For qk [ qc , the intensity of ωc is very weak until the plasmon wave
vector exceeds some value q0c . However, as indicated above, for gapless graphene,
when the surface-MLG separation is less than ac, the surface plasmon intensity in
the long wavelength regime is weak and the mode only appears at shorter wave-
length when qk [ q0c . This behavior seems consistent with the reported experi-
mental data: It is obviously the case if the plasmon branch goes below the main
diagonal ω = vFq. Such damping, as well as the critical distance, changes in the
presence of an energy bandgap for graphene.
218 N.J.M. Horing et al.

(a) (b)

(c) (d)

(e) (f)

Fig. 9.2 Plasmon modes for a semi-infinite conductor interacting with a monolayer of ungapped
graphene with various surface-to-layer separations. The left panels a, c and e present density plots
of the inverse dispersion function S1 þ
C ðqk ; x þ i0 Þ with peaks corresponding to the plasmon
modes. The right panels b, d and f show exact numerical solutions for the plasmon branches, both
Landau damped and undamped. The distances chosen are a kF ¼ 0:2, 1.0 and 0.5, correspond-
ingly. All the plots are provided for extrinsic graphene (doped) with zero energy bandgap
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 219

(a) (b)

(c) (d)

(e) (f)

Fig. 9.3 Plasmon excitations for a semi-infinite conducting medium interacting with two
ungapped graphene layers located at chosen distances from the surface. The left panels a, c and
e give density plots of the inverse dispersion function 1=ScN¼2 ðqk ; x þ i0 þ Þ with peaks
corresponding to the plasmon modes. The right panels b, d and f show the exact numerical
solutions for the plasmon branches, both Landau damped and undamped. The plots show results
for various distances between the surface and the layers: a1 kF ¼ 2:0, 0.2 and 0.5, respectively. The
second layer is placed at a distance a2, which is twice as large as a1. All the plots are provided for
extrinsic graphene (doped) with zero energy bandgap
220 N.J.M. Horing et al.

(a) (b)

(c) (d)

Fig. 9.4 Plasmon excitation spectra for a semi-infinite conductor Coulomb coupled with a single
sheet of monolayer gapped graphene located at a chosen distance from the surface. The left-hand
panels exhibit density plots of the inverse dispersion function 1=ScN¼1 ðqk ; x þ i0 þ Þ with peaks
corresponding to the plasmon modes. The right panels show the numerical solutions for the
Landau damped (dashed blue lines) and undamped (red curves) plasmon branches

Similar calculated results for two ungapped graphene layers and a surface are
presented in Fig. 9.3. The principal difference arising in the case when there are two
Coulomb-coupled layers is that if the separation between the layer nearest the
surface and the surface is less than ac, then both the symmetric and antisymmetric
modes become damped in different ranges of wave vector. It is important to
mention that the upper plasmon branch (symmetric mode) remains almost
unchanged for all cases, with either one or two graphene layers.
The role played by the energy band gap is an important part of our considera-
tions. For monolayer graphene, the existence of an energy gap leads to an extended
region of undamped plasmons [25]. We have focused on the regions outside of the
ð0Þ
single-particle excitation continuum with Im P2D ðqk ; xÞ ¼ 0, since the plasmons
in these regions are not Landau-damped. In Fig. 9.4, we have plotted the plasmon
dispersion relation for the single graphene layer with an energy bandgap for two
chosen distances between the graphene layer and the conducting surface. These
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 221

(a) (b) (c)

Fig. 9.5 Schematic illustrations of the (ω-q||)-space regions having differing analytical expres-
sions for the non-interacting polarization function given in Appendix 2 as taken from [25]. Both
real and imaginary parts each have distinct analytical expressions, as described in [25]. The regions
with ω > ℏvFq|| above the main diagonal are denoted by Ω1–Ω5, while the regions below this
diagonal are denoted by Q1–Q4. Regions with zero ℑmΠ(0) 2D(q, ω), namely Q4, Ω1 and Ω5
correspond to regions where undamped plasmons exist. Each schematic plot illustrates the case
with a different energy bandgap: a Δ = 0.1 μ; b Δ = 0.5 μ; and c Δ = 0.8 μ

results show that for a conducting substrate surface plasmon frequency denoted by
pffiffiffi
xs ¼ xp = 2, there is a coupled/hybrid plasmon frequency ωc of the composite
system which lies below ωs. Furthermore, the intensity of this mode depends on the
distance of the graphene layer from the surface of the conductor as well as on the
energy band gap between the valence and conduction bands of the graphene
monolayer (MLG).
In Fig. 9.5, we illustrate the regions in which both the real and imaginary parts of
the polarization function have distinct analytical expressions for several bandgaps.
As mentioned above, we have given particular attention to these regions with
ImΠ(0)(q, ω) = 0 outside of the single-particle excitation continuum, since the
plasmons in these regions are not Landau-damped. The detailed analytic expres-
sions for the polarization function in each region are given in Appendix 2.

9.5 Experimental Studies on Epitaxial Graphene

Electron energy loss spectroscopy (EELS) is the main experimental technique

employed in the investigation of collective electronic excitations (see [28–30] for a
review of this technique). In particular, while photons can only access the
long-wavelength region, EELS is able to scan an extended range in reciprocal space
(up to 0.4–0.5 Å−1) [29], encompassing both the long- and the short-wavelength
regions. Figure 9.6 shows a typical loss spectrum of plasmon modes for graphene.
While a square-root-like dispersion of the intraband plasmon has been reported
for free-standing graphene and graphene/SiC, experiments performed on monolayer
222 N.J.M. Horing et al.

Fig. 9.6 Intraband and interband plasmons in epitaxial graphene on Pt(111)

Fig. 9.7 Dispersion relation

of the ASP in MLG/Pt(111)
(data taken from [32]) and
MLG/Ir(111) (data taken from
[34])

graphene grown on metals exhibit a linear dispersion for the intraband plasmon. In
particular, experiments for graphene on Pt(111) [31–33] and on Ir(111) [34] have
shown a nearly identical dispersion relation (Fig. 9.7).
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 223

Fig. 9.8 Dependence of the

HREELS spectrum for
monolayer graphene (MLG)
grown on Pt(111) on the
primary electron beam
energy. Spectra have been
recorded in the same
scattering geometry
(incidence angle θi = 55° and
scattering angle θs = 51°,
respectively, measured with
respect to the sample normal).
In the top axis, the energy
scale is reported in THz

The experimental study of plasmon modes of graphene/metals merits special

attention. Graphene-metal contacts [35–44] are omnipresent and inevitable com-
ponents of graphene-based devices and, thus, the investigation of the nature and
dispersion of plasmon modes at graphene/metal interfaces is a key step toward
engineering plasmonic applications of graphene.
The linear dispersion of 2D plasmons at graphene/metal interfaces is a conse-
quence of the coexistence of the bounded metal electron gas with the π-charge
density of graphene in close proximity. It resembles the acoustic surface plasmon
(ASP) at metal surfaces with a partially occupied surface state band within a wide
bulk energy gap [45, 46]. The non-local character of the dielectric function [11, 14,
24] and of the screening processes in graphene [47] prevents the sheet plasmon
from being screened out by the 3D bulk states of the metal substrate.
The group velocity of intraband plasmons at graphene/metal interfaces
( 106 m/s), extracted from the slope of the dispersion relation, is similar to those
calculated for ASP, thus indicating a similar nature of the two collective excitations.
The group velocity of the sheet plasmons at graphene/metal interfaces is about
two orders of magnitude lower than the speed of light, and thus its direct excitation
by light is not possible. However, nanometer-size objects at such surfaces,
including atomic steps or molecular structures, can facilitate coupling between the
sheet plasmons and light. The linear behavior of its dispersion implies that both
phase and group velocities of the collective excitation are the same, so signals can
be transmitted undistorted along the surface with potential applications in
graphene-based nano-optical devices.
224 N.J.M. Horing et al.

Fig. 9.9 HREELS spectra for

MLG/Pt(111), recorded as a
function of the scattering
angle. The primary energy is
7 eV, while the incidence
angle is fixed at 55° with
respect to the sample normal

In addition to the ordinary intraband plasmon, a few recent theoretical works

show the existence of a nonlinear plasmon in free-standing doped graphene [48,
49]. Although a similar excitation has also been observed experimentally [33, 50]
(see the nonlinear mode in Figs. 9.8, 9.9 and 9.10), its origin is still unclear and
currently under investigation.
The electronic response of self-doped graphene supported on metal substrates
shows a remarkable dependence on the primary electron beam energy. Loss spectra,
obtained by using impinging electrons of low kinetic energy 7–8 eV, show both
acoustic and nonlinear plasmons.
For primary energies below 6 eV, only a peak around 0:5 eV is observed. At
E p ¼ 8 eV, two clearly distinct plasmon modes may be resolved spectroscopically.
The second plasmon mode around 1:65 eV increases its spectral weight and it is
predominant in the spectrum recorded for E p ¼ 16 eV. In addition to these peaks,
graphene optical phonons are also clearly observed at 0.1 and 0:2 eV [51, 52]. This
striking dependence on the primary energy explains why previous experimental
studies were not able to resolve the two plasmon modes [32, 34].
The HREELS spectra for E p ¼ 7 eV, recorded as a function of the scattering
angle, are reported in Fig. 9.9. The two plasmon modes develop with variation of
the scattering angle. The higher-energy mode emerges in off-specular geometries,
for values of θs lower than 50° with respect to the sample normal. This mode is
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 225

Fig. 9.10 Plasmon dispersion in MLG/Pt(111). The size of the markers reflects the intensity of the
plasmon modes, normalized to that of the elastic peak. Markers are colored following a color scale
reported on the right. The light gray area indicates the continuum of interband single-particle
excitations (SPE), while the dark gray represents the area of the continuum of intraband SPEs. The
plasmon mode enters the Landau damping regime of interband electron-hole excitations when its
dispersion relation intercepts the boundary for the continuum of interband SPEs. The dotted line
represents an interpolation of the nonlinear mode for small momenta, for which the intensity
vanishes

characterized by a linear dispersion (acoustic plasmon). By contrast, the low-energy

mode has a nonlinear dispersion relation (Fig. 9.10). Its intensity reaches a maxi-
mum and then decreases for higher momenta.
The dispersion curve of the nonlinear plasmon is well fitted by a power law:

hxnonlinear ¼ Bqnk ð9:45Þ

n
with B ¼ ð0:66
0:02Þ eV Å and n = (0.12 ± 0.01). In Fig. 9.10, we also show the
electron-hole continuum or single-particle excitation (SPE) region.
The acoustic plasmon lies above the upper edge vF qk of the intraband excitations
and it can be damped by interband transitions. In contrast to this, the nonlinear
plasmon is located below vF qk and thus it is damped by intraband transitions.
The nonlinear plasmon was first predicted by Gao et al. [48] for doped
free-standing graphene. It should be noted that theoretical work based on an
oversimplified isotropic graphene band structure in the vicinity of the K point
cannot explain it [53]. Its nonlinear dispersion, considered jointly with its high
spectral weight and its lower energy, makes it a natural candidate for applications in
graphene-based nonlinear plasmonics [54, 55].
226 N.J.M. Horing et al.

Fig. 9.11 Behavior of the nonlinear mode as a function of parallel momentum transfer qk for
monolayer graphene (MLG) on Ru(0001). The intensity of the plasmon modes, normalized to that
of the elastic peak, is indicated by the size of the markers, which are colored following a color
scale reported on the right. The solid line represents the best fit. In the inset, a selected loss
spectrum for MLG/Ru(0001) has been reported. Two distinct loss features are present

It should be borne in mind that in undoped graphene, both the acoustic plasmon
and the nonlinear mode do not occur [48, 56]. In the case of MLG/Pt(111), the
p-type doping of the graphene sheet at this interface is responsible for the existence
of the above-mentioned two plasmon modes.
We also note that when the acoustic plasmon coexists in the EELS spectrum
with the nonlinear mode, its intensity in the range 0.6–1.0 eV is particularly weak
(as highlighted by the size of markers in Fig. 9.10) in comparison with the case of
impinging energies above 12 eV, for which only the acoustic plasmon is recorded.
Thus, the coexistence of the two modes induces a gap in the dispersion relation
when considering the intensity of the plasmon modes.
Finally, in order to investigate whether the nonlinear mode depends on the
graphene/metal interface, we have also studied its dispersion for MLG/Ru(0001)
(Fig. 9.11). We find that two plasmon modes also exist in this case, even though the
dispersion of the nonlinear mode is slightly different due to the dissimilar properties
of this graphene/metal interface from those of MLG/Pt(111) (see [57–59] for more
details).
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 227

Fig. 9.12 Dispersion relation

of the intraband plasmon for
several different values of
electron density n. The
intraband plasmon excitation
arising from intrinsic electron
density n ¼ 1:2  1013 cm−2
changes when more electrons
are added by K doping.
Adapted from [62]

9.6 Influence of Adsorbed and Intercalated Atoms

The tunability of graphene plasmons is a promising opportunity for the emerging

field of graphene-based plasmonics. Both the conductivity of graphene and the
dispersion of plasmons in graphene are related to the Fermi energy (or chemical
potential at room temperature). The Fermi energy
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
EF l ph2 v2F n ð9:46Þ

can be easily controlled (with a concomitant tuning of plasmon modes) by changing

the charge concentration, which, in turn, can be realized by varying a gate potential
or chemical doping [60]. In consideration of possible applications, it should be
mentioned that distributing carriers into multiple graphene/insulator stacks effi-
ciently enhances the plasmonic resonance frequency and its magnitude [61].
Highly electropositive or electronegative chemical species can easily dope
graphene due to charge transfer. In Fig. 9.12, results for K-doped graphene/SiC [62]
are reported. With the addition of more electrons in the π* band, the Fermi level
shifts upward away from the Dirac point ED, causing a continuous blue-shift of the
plasmon energy with increasing potassium coverage.
Similarly, a red-shift has been reported in water-exposed MLG/Pt(111) [31].
However, in the latter study, charge doping affects the frequency of the plasmon
228 N.J.M. Horing et al.

Fig. 9.13 Plasmon dispersion

for Cs-intercalated
graphene/Ni(111). The linear
dispersion of the π plasmon in
pristine, free-standing
graphene is recovered [63,
64]. Adapted from [65]

modes but not the acoustic behavior of the dispersion relation. Nevertheless, a
reduction of the group velocity by about 11 % has been observed [31].
In contrast, experiments performed for Na-doped MLG/Ir (111) [34] and
F4-TCNQ/MLG/SiC [65] by Pfnür’s group show negligible influence of doping on
the dispersion curve while noticeable changes in the damping mechanisms of the
plasmon modes are reported.
Interesting results have been reported for plasmon dispersion in Cs-intercalated
graphene/Ni(111) [66]. Without intercalation, the close surface proximity of gra-
phene on Ni(111) exhibits a strong hybridization between Ni d bands and π states of
graphene [67, 68]. The concomitant change of the energy spectrum results in the
loss of the Dirac cones at the graphene K-points, and, with this, the linear plasmon
spectrum is lost. However, intercalated chemical species may decouple such orbital
interactions between the graphene sheet and the substrate [69, 70], restoring the
integrity of graphene’s Dirac cones. Correspondingly, the influence of intercalated
atoms on the electronic structure of MLG/Ni(111) can manifest itself in the col-
lective electronic properties of the intercalated system involving just the Coulomb
coupling of the graphene sheet with a substrate Ni plasma (devoid of orbital
hybridization effects with Ni). Thus, the dispersion curve of the π plasmon, in
comparison with the analogous dispersion curve obtained in the absence of inter-
calated Cs, shows that alkali metal atoms make graphene effectively
quasi-freestanding in this sense, with a much lowered charge transfer, and with the
recovery of the linear plasmon dispersion typical of free-standing graphene whose
interaction with the substrate Ni is solely Coulombic (Fig. 9.13).
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 229

9.7 Concluding Remarks

In summary, we have reviewed nonlocal plasmon dispersion in the Random Phase

Approximation for monolayer graphene (as well as for a pair of layers) interacting
with a conducting substrate: this was done for arbitrary wavelength, also examining
dependence on layer-surface separation. We have emphasized some interesting
effects of Landau damping of the plasmon modes which depend on the layer
separation distance. Specifically, our numerical simulations exhibited in Fig. 9.2
indicate that, for a single layer whose distance is less than a critical value ac, the
lower acoustic plasmon mode is severely Landau damped; this behavior is con-
sistent with reported experimental data. This may be seen in our plots when the
plasmon energy lies below the main diagonal ω = vFq. The damping, as well as the
critical distance, changes in the presence of an energy bandgap for graphene.
In regard to plasmon damping, we also discuss results for two graphene layers
and a conductor, which are presented in Fig. 9.3. But, a major difference between the
case in which there are two graphene layers rather than one is that, for separations
between the closest layer and the conducting surface less than the critical value ac,
both the symmetric and antisymmetric modes become damped in different ranges of
wave vector. It should be noted that the upper plasmon branch (symmetric mode)
remains almost unaffected by Landau damping for all cases, involving either one or
two graphene layers. Furthermore, the role of an energy gap in the graphene plasmon
spectrum and damping has been addressed here in connection with Figs. 9.4 and 9.5.
Moreover, considerable attention to the experimental studies of graphene plasmons
is provided in Figs. 9.6–9.13. The reader should be aware that other materials having
the Dirac-like energy spectrum of graphene are currently under investigation, in
particular silicene and, most recently, topological insulators [71].
A fuller exposition of the theory and its bearing on experimental data is provided
by Gumbs et al. [24]. The experimental data clearly confirm linear plasmon disper-
sion in the long wavelength limit. The simulations reviewed here go beyond this
regime to a fully general numerical description of nonlocal effects in monolayer and
bilayer graphene, including dependencies on spatial separation a and gap strength.
Further developments of this work to include the roles of nonlocality of the substrate
plasma and of its finite width may be facilitated by the results provided in Sect. 9.2.

Acknowledgments This work was supported in part by contract # FA 9453-13-1-0291 of AFRL.

Appendix 1: Dynamic Nonlocal Polarization Function

for Free-Standing Graphene with no Bandgap; Brief
Summary of the Results Derived in [19, 20]

The 2D RPA density perturbation response function,

ð0Þ e
dq=dV ¼ R2D ðp; xÞ  D0 Rðx; vÞ, for free-standing gapless Graphene in the T = 0
230 N.J.M. Horing et al.

degenerate limit is given in terms of dimensionless frequency and wavenumber

variables defined by ν = ω/EF = ω/μ and x = p/pF, respectively, as (note that
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
D0  c1 gs gv q2D =p; gs and gv are spin and valley degeneracies, gs = gv = 2;
ℏ → 1):

e mÞ ¼ R
Rðx; e þ ðx; mÞ þ R
e  ðx; mÞ; ð9:47Þ

with (θ(z) ≡ η+(z) = Heaviside unit step function)

e þ ðx; mÞ ¼ R
R e 1þ ðx; mÞhðm  xÞ þ R
e 2þ ðx; mÞhðx  mÞ ð9:48Þ

e   R)
where (define P e

1
e þ ðx; mÞ ¼ 1  pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e 1þ ðx; mÞ ¼ Re P
Re R 1 ff1 ðx; mÞhðj2 þ mj  xÞ
8 m  x2
2
ð9:49Þ
þ sgnðm  2 þ xÞf1 ðx; mÞhðj2  mj  xÞ
þ f2 ðx; mÞ½hðx þ 2  mÞ þ hð2  x  mÞg;

1
e þ ðx; mÞ ¼ 1  pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e 2þ ðx; mÞ ¼ Re P
Re R 2 ff3 ðx; mÞhðx  jm þ 2jÞ
8 x2  m2
þ f3 ðx; mÞhðx  jm  2jÞ ð9:50Þ

px2
þ ½hðjm þ 2j  xÞ þ hðjm  2j  xÞ ;
2

1
e þ ðx; mÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e 1þ ðx; mÞ ¼ Im P
Im R 1 ff3 ðx; mÞhðx  jm  2jÞ
8 m2  x2
 ð9:51Þ
px2
þ ½hðx þ 2  mÞ þ hð2  x  mÞ ;
2

e þ ðx; mÞ ¼ hðp
e 2þ ðx; mÞ ¼ Im P
Im R
m  x þ 2Þ
ffiffiffiffiffiffiffiffiffiffiffiffiffiffi ½f4 ðx; mÞ  f4 ðx; mÞhð2  x  mÞ
2
8 x2  m2
ð9:52Þ

and

e  ðx; mÞ ¼ px2 hðx  mÞ px2 hðm  xÞ

R pffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ i pffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð9:53Þ
8 x2  m2 8 m2  x2

The quantities f1(x, ν), f2(x, ν), f3(x, ν), f4(x, ν) are defined as
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 231

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2 þ mÞ2 x2 þ ð2 þ mÞ
f1 ðx; mÞ ¼ ð2 þ mÞ ð2 þ mÞ2 x2  x2 ln pffiffiffiffiffiffiffiffiffiffiffiffiffiffi  ; ð9:54Þ
 2 
 m  x2 þ m
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
m m2  x2
f2 ðx; mÞ ¼ x ln 2
; ð9:55Þ
x
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2þm
f3 ðx; mÞ ¼ ð2 þ mÞ x2  ð2 þ mÞ2 þ x2 sin1 ; ð9:56Þ
x
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2 þ mÞ2 x2 þ ð2 þ mÞ
f4 ðx; mÞ ¼ ð2 þ mÞ ð2 þ mÞ2 x2  x2 ln : ð9:57Þ
x

Appendix 2: Dynamic Nonlocal Polarization Function

for Graphene with a Finite Energy Bandgap; Brief
Summary of the Results Derived in [18]

The 2D RPA ring diagram polarization function for graphene with a gap Δ may be
expressed as

ð0Þ g
Z X k
ðk þ qÞ þ D2

0
P2D ðq; xÞ ¼ 2 2
d k 1 þ ss
4p s;s0 ¼

e k e j k þ qj
ð9:58Þ
f ðsek Þ  f ðs0 ek þ q Þ

sek  s0 ek þ q  hx  ihd

Since we limit our considerations to zero temperature, T = 0, the Fermi-Dirac

distribution function is reduced to the Heaviside step function f(ɛ, μ;
T → 0) = η+(μ − ɛ), so (9.58) is simplified to

ð0Þ l
P2D ðq; xÞ ¼ v1 l
 ðq; xÞ þ v þ ðq; xÞ þ v ðq; xÞ ; ð9:59Þ

where
Z X 2
g 0 k
ðk þ qÞ þ D
va
¼ 2 d kg þ ða  D  q Þ
2 2 2 2
1
ss
4p s;s0 ¼

ek ejk þ qj
  ð9:60Þ
1 1
 
hx þ ek  ejk þ qj þ ihd hx  ek
ejk þ qj þ ihd

and
232 N.J.M. Horing et al.

ð0Þ
P2D ðq; xÞ ¼ Pð0Þ ðq; xÞ þ g þ ðl  DÞPð1Þ ðq; xÞ; ð9:61Þ

where

Pð0Þ ðq; xÞ ¼ v1

 ðq; xÞ
ð9:62Þ
Pð1Þ ðq; xÞ ¼ vlþ ðq; xÞ þ vl ðq; xÞ:

The following notations are employed to specify the expressions involved in the
polarization function:

g hv2F q2
F ðq; xÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð9:63Þ
16p
jx2  ðvF qÞ2 j

and
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
u 4D2
X 0 ¼ t1   ; ð9:64Þ
h2 x2  ðvF qÞ2

along with definitions of the following functions:

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
G\ ðxÞ ¼ x X 20  x2  ð2  X 0 Þ arccosðx=X 0 Þ; ð9:65Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
G [ ðxÞ ¼ x x2  X 20  ð2  X 0 Þ arccoshðx=X 0 Þ; ð9:66Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffi
G [ ðxÞ ¼ x x2  X 20  ð2  X 0 Þarcsinhðx= X 20 Þ: ð9:67Þ

Finally, the polarization function is given in the following form:

Real part
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 233

gl
RePð0Þ ðq; xÞ ¼  þ F ðq; xÞ
h2 v2F
8
> 0 ) Q1
>
>  
>
>
> G\ hM
> ) Q2
>
>  
vF q
 
>
>
>
> G P
þ G M
)
>
> \ hvF q \ hvF q Q3
>
>    
>
>
> G [ P  G\ P
> ) Q4
>
> 
h v q hvF q
>
>  
F
  ð9:68Þ
< P M
 G [ hvF q  G\ hvF q ) X1
>
>  
>
>
>
> G [ hvPF q ) X2
>
>    
>
>
>
> G [ hvP q  G [ hM ) X3
>
>
>
>  
F
 
vF q
>
>
>
> M
G [ hvF q þ G [ hvF q P
) X4
>
>
>
>    
>
:G P  G M
0 hvF q 0 hvF q ) X1

Imaginary part

ImPð0Þ ðq; xÞ ¼ F ðq; xÞ

8    
>
> G [ hvPF q  G [ hM ) Q1
>
> vF q
>
>  
>
>
>
> G [ hvPF q ) Q2
>
>
>
> )
>
> 0 Q3
>
>
>
<0 ) Q4 ð9:69Þ
 0 ) X1
>
>  
>
>
>
> G M
) X2
>
>
\ hvF q
>
> 
>
> p 2  X 20 ) X3
>
>
> 
>
>
> p 2  X0 ) X4
2
>
:
0 ) X5
where

P ¼ 2l þ hx and M ¼ 2l  hx: ð9:70Þ

The analytic expressions provided in the left columns above for the real and
imaginary parts of Π0 pertain to the frequency wavenumber regions marked by the
Q’s and Ω’s in the corresponding right column as indicated in Fig. 9.4. Specifically,
these ω-q regions are defined as follows:
234 N.J.M. Horing et al.

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Q1 ) hx\l  ðhvF Þ2 ðq  kFl Þ2 þ D2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Q2 )
l  ðhvF Þ2 ðq  kFl Þ2 þ D2 \hx\
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
l þ ðhvF Þ2 ðq þ kFl Þ2 þ D2 ð9:71Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Q3 ) hx\  l þ ðhvF Þ2 ðq þ kFl Þ2 þ D2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Q4 ) l þ ðhvF Þ2 ðq þ kFl Þ2 þ D2 \hx\hvF q;

and

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X1 ) q\2kF & ðhvF Þ2 q2 þ 4D2 \hx\
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
\  l þ ðhvF Þ2 ðq  kFl Þ2 þ D2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X2 ) l þ ðhvF Þ2 ðq  kFl Þ2 þ D2 \
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
l þ ðhvF Þ2 ðq  kFl Þ2 þ D2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð9:72Þ
X3 ) hx [ l þ ðhvF Þ2 ðq þ kFl Þ2 þ D2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X4 ) q [ 2kF & ðhvF Þ2 q2 þ 4D2 \hx\
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
\  l þ ðhvF Þ2 ðq  kFl Þ2 þ D2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X5 ) hvF q\hx\ ðhvF Þ2 q2 þ 4D2 :

References

1. D.S.L. Abergel, V. Apalkov, J. Berashevich, K. Ziegler, T. Chakraborty, Properties of

graphene: a theoretical perspective. Adv. Phys. 59(4), 261482 (2010)
2. M.I. Katsnelson, Graphene: Carbon in Two Dimensions (Cambridge University Press, 2012)
3. H.S. Aoki, M. Dresselhaus, Physics of Graphene (Springer, 2014)
4. E.L. Wolf, Graphene: A New Paradigm in Condensed Matter and Device Physics (Oxford
University Press, 2013)
5. G. Gumbs, D. Huang, O. Roslyak, Electronic and photonic properties of graphene layers and
carbon nanoribbons. Philos. Trans. R. Soc. A 368, 5351–5556 (2010)
6. A.H.C. Neto, F. Guinea, N.M.R. Peres, K.S. Novoselov, A.K. Geim, The electronic properties
of graphene. Rev. Mod. Phys. 81(1), 109 (2009)
7. S. Das Sarma, S. Adam, E.H. Hwang, E. Rossi, Electronic transport in two-dimensional
graphene. Rev. Mod. Phys. 83(2), 407 (2011)
8. L.E.F.F. Torres, S. Roche, J. Charlier, Introduction to Graphene-Based Nanomaterials From
Electronic Structure to Quantum Transport (Cambridge University Press, 2014).
ISBN:9781107030831
9. C.J. Tabert, E.J. Nicol, Dynamical polarization function, plasmons, and screening in silicene
and other buckled honeycomb lattices. Phy. Rev. B 89(19), 195410 (2014)
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 235

10. N.J. Horing, Quantum theory of electron gas plasma oscillations in a magnetic field. Ann.
Phys. (NY) 31, 1–63 (1965)
11. N.J.M. Horing, Aspects of the theory of graphene. Philos. Trans. R. Soc. A 368(1932), 5525–
5556 (2010)
12. N.J.M. Horing, Inhomogeneous structure of the nonlocal dynamic dielectric response of a
quantum-well bound state and a three dimensional band of extended states. Phys. Rev. B 59
(8), 5648 (1999)
13. N.J.M. Horing, G. Gumbs, T. Park, Coupling of 2D plasmons to nonlocal bulk plasmons.
Phys. B 299(1), 165172 (2001)
14. N.J.M. Horing, Coupling of graphene and surface plasmons. Phys. Rev. B 80(19), 193401
(2009)
15. P.C. Martin, J. Schwinger, Theory of many-particle systems. I Phys. Rev. 115(6), 1342 (1959)
16. R.H. Ritchie, Surface plasmons in solids. Surf. Sci. 34(1), 119 (1973)
17. H. Raether, Surface plasmons on smooth and rough surfaces and on gratings. Springer Tracts
Mod. Phys. 111, (1988)
18. H. Raether, Excitations of plasmons and interband transitions by electrons. Springer Tracts
Mod. Phys. 88, (1980)
19. P.J. Feibelman, Surface electromagnetic fields. Prog. Surf. Sci. 12(4), 287407 (1982)
20. A. Liebsch, Electronic Excitations at Metal Surfaces (Plenum Pub Corp, New York, 1997).
ISBN 10:0306455455
21. M.S. Kushwaha, Plasmons and magnetoplasmons in semiconductor heterostructures. Surf. Sci.
Rep. 41(1), 1416 (2001)
22. D.M. Newns, Dielectric response of a semi-infinite degenerate electron gas. Phys. Rev. B 1(8),
3304 (1970)
23. N.J.M. Horing, E. Kamen, H. Cui, Inverse dielectric function of a bounded solid-state plasma.
Phys. Rev. B 32(4), 2184 (1985)
24. G. Gumbs, A. Iurov, N.J.M. Horing, Non-local plasma spectrum of graphene interacting with
a thick conductor. Phys. Rev. B 91(23), (2015)
25. P.K. Pyatkovskiy, Dynamical polarization, screening, and plasmons in gapped graphene.
J. Phys.: Condens. Matter 21(2), 025506 (2009)
26. B. Wünsch, T. Stauber, F. Sols, F. Guinea, Dynamical polarization of graphene at finite
doping. New J. Phys. 8(12), 318 (2006)
27. E.H. Hwang, S. Das Sarma, Dielectric function, screening, and plasmons in two-dimensional
graphene. Phys. Rev. B 75(20), 205418 (2007)
28. M. Rocca, Low-energy investigation of surface electronic excitations on metals. Surf. Sci.
Rep. 22(12), 171 (1995)
29. A. Politano, G. Chiarello, G. Benedek, E.V. Chulkov, P.M. Echenique, Vibrational
spectroscopy and theory of alkali metal adsorption and co-adsorption on single-crystal
surfaces. Surf. Sci. Rep. 68(3–4), 305–389 (2013)
30. A. Politano, G. Chiarello, The influence of electron confinement, quantum size effects, and
film morphology on the dispersion and the damping of plasmonic modes in Ag and Au thin
films. Prog. Surf. Sci. 90(2), 144–193 (2015)
31. A. Politano, A.R. Marino, G. Chiarello, Effects of a humid environment on the sheet plasmon
resonance in epitaxial graphene. Phys. Rev. B 86, 085420 (2012)
32. A. Politano, A.R. Marino, V. Formoso, D. Farías, R. Miranda, G. Chiarello, Evidence for
acoustic-like plasmons on epitaxial graphene on Pt(111). Phys. Rev. B 84, 033401 (2011)
33. A. Politano, V. Formoso, G. Chiarello, Evidence of composite plasmonphonon modes in the
electronic response of epitaxial graphene. J. Phys.: Condens. Matter 25(34), 345303 (2013)
34. T. Langer, D.F. Förster, C. Busse, T. Michely, H. Pfnür, C. Tegenkamp, Sheet plasmons in
modulated graphene on Ir(111). New J. Phys. 13(5), 053006 (2011)
35. G. Giovannetti, P.A. Khomyakov, G. Brocks, V.M. Karpan, J. van den Brink, P.J. Kelly,
Doping graphene with metal contacts. Phys. Rev. Lett. 101, 026803 (2008)
236 N.J.M. Horing et al.

36. C. Gong, D. Hinojos, W. Wang, N. Nijem, B. Shan, R.M. Wallace, K. Cho, Y.J. Chabal,
MetalGrapheneMetal sandwich contacts for enhanced interface bonding and work function
control. ACS Nano 6(6), 5381–5387 (2012)
37. K.L. Grosse, B. Myung-Ho, L.E.P. Feifei, W.P. King, Nanoscale Joule heating, Peltier cooling
and current crowding at graphene-metal contacts. ACS Nano 6(6), 5381–5387 (2012)
38. C. Gong, S. McDonnell, X. Qin, A. Azcatl, H. Dong, Y.J. Chabal, K. Cho, R.M. Wallace,
Realistic metalgraphene contact structures. ACS Nano 8(1), 642649 (2013)
39. A. Politano, G. Chiarello, Quenching of plasmons modes in air-exposed graphene-Ru contacts
for plasmonic devices. Appl. Phys. Lett. 102(20), 201608 (2013)
40. J.T. Smith, A.D. Franklin, D.B. Farmer, C.D. Dimitrakopoulos, Reducing contact resistance in
graphene devices through contact area patterning. ACS Nano 7(4), 3661–3667 (2013)
41. C. Archambault, A. Rochefort, States Modulation in Graphene Nanoribbons through Metal
Contacts. ACS Nano 7(6), 5414–5420 (2013)
42. P. Janthon, F. Viñes, S.M. Kozlov, J. Limtrakul, F. Illas, Theoretical assessment of
graphene-metal contacts. J. Chem. Phys. 138(24), 244701 (2013)
43. F. Menges, H. Riel, A. Stemmer, C. Dimitrakopoulos, B. Gotsmann, Thermal transport into
graphene through nanoscopic contacts. Phys. Rev. Lett. 111(20), 205901 (2013)
44. A. Politano, G. Chiarello, Unravelling suitable graphene-metal contacts for graphene-based
plasmonic devices. Nanoscale 5(17), 8215–8220 (2013)
45. V.M. Silkin, A. García-Lekue, J.M. Pitarke, E.V. Chulkov, E. Zaremba, P.M. Echenique,
Novel low-energy collective excitation at metal surfaces. EPL (Europhys. Lett.) 66(2), 260
(2004)
46. V.M. Silkin, J.M. Pitarke, E.V. Chulkov, P.M. Echenique, Acoustic surface plasmons in the
noble metals Cu, Ag, and Au. Phys. Rev. B 72(11), 115435 (2005)
47. M. van Schilfgaarde, M.I. Katsnelson, First-principles theory of nonlocal screening in
graphene. Phys. Rev. B 83(8), 081409 (2011)
48. Y. Gao, Z. Yuan, Anisotropic low-energy plasmon excitations in doped graphene: An ab initio
study. Solid State Commun. 151(14), 1009–1013 (2011)
49. F.M.D. Pellegrino, G.G.N. Angilella, R. Pucci, Dynamical polarization of graphene under
strain. Phys. Rev. B 82(11), 115434 (2010)
50. A. Politano, G. Chiarello, Emergence of a nonlinear plasmon in the electronic response of
doped graphene. Carbon 71, 176–180 (2014)
51. A. Politano, A.R. Marino, D. Campi, D. Farías, R. Miranda, G. Chiarello, Elastic properties of
a macro- scopic graphene sample from phonon dispersion measurements. Carbon 50(13),
4903–4910 (2012)
52. A. Politano, A.R. Marino, G. Chiarello, Phonon dispersion of quasi-freestanding graphene on
Pt (111). J. Phys.: Condens. Matter 24(10), 104025 (2012)
53. E.H. Hwang, S. Das Sarma, Dielectric function, screening, and plasmons in two-dimensional
graphene. Phys. Rev. B 75(20), 205418 (2007)
54. A.V. Gorbach, Nonlinear graphene plasmonics: amplitude equation for surface plasmons.
Phys. Rev. A 87(1), 013830 (2013)
55. M. Kauranen, A.V. Zayats, Nonlinear plasmonics. Nat. Photonics 6(11), 737748 (2012)
56. V. Despoja, D. Novko, K. Dekanić, M. Šunjić, L. Marušić, Two-dimensional and p plasmon
spectra in pristine and doped graphene. Phys. Rev. B 87(7), 075447 (2013)
57. B. Borca, S. Barja, M. Garnica, M. Minniti, A. Politano, J. M. Rodriguez-García,
J. J. Hinarejos, D. Farías, A. L. Vásquez de Parga, R. Miranda, Electronic and geometric
corrugation of periodically rippled, self-nanostructured graphene epitaxially grown on Ru
(0001). New J. Phys. 12(9), 093018 (2010)
58. M. Batzill, The surface science of graphene: Metal interfaces, CVD synthesis, nanoribbons,
chemical modifications and, defects. Surf. Sci. Rep. 67(3), 83115 (2012)
59. J. Wintterlin, M.L. Bocquet, Graphene on metal surfaces. Surf. Sci. 603(10), 1841–1852
(2009)
60. Z. Fang, Y. Wang, Z. Liu, A. Schlather, P.M. Ajayan, F.H.L. Koppens, P. Nordlander, N.
J. Halas, Plasmon-induced doping of graphene. ACS Nano 6(11), 10222–10228 (2012)
9 Recent Progress on Nonlocal Graphene/Surface Plasmons 237

61. H. Yan, X. Li, B. Chandra, G. Tulevski, Y. Wu, M. Freitag, W. Zhu, P. Avouris, F. Xia,
Tunable infrared plasmonic devices using graphene/insulator stacks. Nat. Nanotechnol. 7(5),
330–334 (2012)
62. S.Y. Shin, N.D. Kim, J.G. Kim, K.S. Kim, D.Y. Noh, K.S. Kim, J.W. Chung, Control of the p
plasmon in a single layer graphene by charge doping. Appl. Phys. Lett. 99(8), 082110 (2011)
63. M.K. Kinyanjui, C. Kramberger, T. Pichler, J.C. Meyer, P. Wachsmuth, G. Benner, U. Kaiser,
Direct probe of linearly dispersing 2D interband plasmons in a free-standing graphene
monolayer. EPL (Europhys. Lett.) 97(5), 57005 (2012)
64. C. Kramberger, R. Hambach, C. Giorgetti, M.H. Rümmeli, M. Knupfer, J. Fink, B. Büchner,
Reining Lucia, E. Einarsson, S. Maruyama et al., Linear plasmon dispersion in single-wall
carbon nanotubes and the collective excitation spectrum of graphene. Phys. Rev. Lett. 100(19),
196803 (2008)
65. C. Tegenkamp, H. Pfnür, T. Langer, J. Baringhaus, H.W. Schumacher, Plasmon electronhole
resonance in epitaxial graphene. J. Phys.: Condens. Matter 23(1), 012001 (2011)
66. A. Cupolillo, N. Ligato, L.S. Caputi, Plasmon dispersion in quasifreestanding graphene on Ni
(111). Appl. Phys. Lett. 102(11), 111609 (2013)
67. G. Bertoni, L. Calmels, A. Altibelli, V. Serin, First-principles calculation of the electronic
structure and EELS spectra at the graphene/Ni(111) interface. Phys. Rev. B 71(7), 075402
(2005)
68. M. Hasegawa, K. Nishidate, T. Hosokai, N. Yoshimoto, Electronic-structure modification of
graphene on Ni(111) surface by the intercalation of a noble metal. Phys. Rev. B 78, 085439
(2013)
69. D. Farias, A.M. Shikin, K.H. Rieder, YuS Dedkov, Synthesis of a weakly bonded graphite
monolayer on Ni(111) by intercalation of silver. J. Phys.: Condens. Matter 11(43), 8453
(1999)
70. A.M. Shikin, D. Farías, V.K. Adamchuk, K. Rieder, Surface phonon dispersion of a graphite
monolayer adsorbed on Ni(111) and its modification caused by intercalation of Yb La and Cu
layers. Surf. Sci. 424(1), 155–167 (1999)
71. A. Politano, V.M. Silkin, I.A. Nechaev, M.S. Vitiello, L. Viti, Z.S. Aliev, M.B. Babanly, G.
Chiarello, P.M. Echenique, E.V. Chulkov, Interplay of surface and Dirac plasmons in
topological insulators: the case of Bi2Se3. Phys. Rev. Lett. (2015)
Chapter 10
Semiconducting Carbon Nanotubes:
Properties, Characterization
and Selected Applications

Chiara Pintossi and Luigi Sangaletti

Abstract Carbon nanotubes are challenging materials from the point of view of
nanotechnology, because of their peculiar electrical, mechanical and optical prop-
erties arising from their monodimensional geometry. Here, the properties of carbon
nanotubes are discussed, starting from their crystalline structure, in order to
understand their optical, electrical and vibrational behavior. In the second section,
the most popular CNT synthesis mechanism are presented, while the last section is
devoted to the CNTs applications, focusing on photovoltaic and gas sensor devices.

10.1 Introduction to Carbon Nanotubes

Even though carbon filaments with a diameter of less than 10 nm were produced in
70s and 80s [1–5], it was only after the first observation with high resolution
transmission electron microscopy (HRTEM) of carbon nanotubes (CNTs) by Iijima
in 1991 [6] that CNT field boosted up significantly, attracting more and more
scientist all over the world.
The simplest and fundamental type of CNTs is represented by single walled
carbon nanotubes (SWCNT) which can be schematized as a single graphene sheet,
rolled up to form a cylinder.
Because the tube properties [7] are dramatically influenced by the way in which
the graphene sheet is rolled, it’s worth deeply understanding the geometry of
nanotubes and their relationship with graphene crystalline structure.
Starting from the basis vectors of the graphene hexagonal honeycomb Bravais
lattice, a1 and a2 represented in Fig. 10.1, it is possible to express the circumference

C. Pintossi
L. Sangaletti (&)

I-LAMP, Interdisciplinary Laboratory for Advanced Materials Physics,
Università Cattolica del Sacro Cuore, Brescia, Italy
e-mail: [email protected]
C. Pintossi
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 239

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_10
240 C. Pintossi and L. Sangaletti

Fig. 10.1 a Schematic representation of the graphene Bravais lattice in which the vectors a1, a2,
Ch and T are highlighted for a (5, 2) SWCNT. b A SWCNT obtained by rolling a
graphene monolayer

of any carbon nanotube in term of the chiral vector Ch, which is defined as the
linear combination of the basis vectors: Ch = na1 + ma2. The pair of integer (n, m)
uniquely defines the chiral vector and is often called simply nanotube chirality.
Nanotubes with m = 0 are called zigzag, if m = n, the tube is denoted as armchair,
while all the other CNTs are simply called chiral.
The other unit lattice vector T, perpendicular to the chiral vector can be also
expressed in terms of n, m, a1 and a2 as:

ð2m þ nÞ ðm þ 2nÞ
T¼ a1  a2
bR bR

where bR is equal to 3b, if m − n is a multiple of 3b, or bR = b, when m − n is not a

multiple of 3b, with b being the greatest common divisor between m and n.
The tube diameter d is given by:
pffiffiffi
Ch 3aCC pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
d¼ ¼ m2 þ mn þ n2
p p

where aC–C is the lenght of the carbon–carbon bond length (1.42 Å), i.e. the
hexagon side in Fig. 10.1.
The chiral angle θ, defined as the angle between Ch and a1, is:
 pffiffiffi 
3n
h ¼ tan1
2m þ n

and varies from θ = 0 for zigzag nanotubes to θ = 30° for the armchair ones.
The number of hexagons in each unit cell is given by:
10 Semiconducting Carbon Nanotubes … 241

Fig. 10.2 a Tight-binding band structure of graphene. The wave vector k for a 1D CNT is shown
in the 2D Brillouin zone of graphite, in red, for b metallic and c semiconducting tubes

2  2 
N¼ m þ mn þ n2 ;
bR

it follows that the CNT unit cell is N time larger than that of graphene.
Theoretical studies predicted CNT electronic properties to be very sensitive to
their geometrical structure. In fact, although graphene is a zero-gap semiconductor,
single walled carbon nanotube can be either metallic or semiconducting, depending
only on their chiral index (n, m).
The electronic structure around the Fermi level of an isolated graphene sheet is
given by an occupied π-band and an unoccupied π*-band, characterized by having
both linear dispersion and by crossing the Fermi level at the K point of the Brillouin
zone (Fig. 10.2a). When a graphene sheet is rolled to form a tube, only some
k states are allowed, due to the periodicity boundary conditions imposed in the
circumferential direction. Whenever the set of allowed k states includes the K point
(Fig. 10.2b), the system has a non-zero density of states (DOS) at the Fermi level
and is then a metal with two linear dispersing bands; otherwise, when the K point is
located at one-third of the distance between two adjacent allowed k (Fig. 10.2c), the
CNT is a semiconductor with a variable energy band gap. Nanotubes with
n − m = 3b, where b is an integer, are metallic, all the other are semiconductors.
Each parallel k-line describes one pair of conduction and valence sub-bands in the
nanotube. Transitions at the bandgaps of such sub-band pairs lead to strong optical
resonances, and these transitions are traditionally labelled Sii for semiconducting
and Mii for metallic nanotubes, where i is the sub-band index.
Multi-walled nanotubes (MWNTs) consist of multiple rolled layers (concentric
tubes) of graphene in a Russian doll model. Its individual shells can be described as
SWNTs, which can be metallic or semiconducting. Because of statistical probability
and restrictions on the relative diameters of the individual tubes, a MWNT is
usually a zero-gap metal.
The 1D density of state of CNT, derived straightforward from the graphite one
[8], becomes large when the energy dispersion relation becomes flat as a function of
242 C. Pintossi and L. Sangaletti

Fig. 10.3 The density of state

of graphene (a) is compared
to the CNTs one (b). The 1D
confinement in CNTs gives
rise to the characteristic van
Hove singularities

k, giving rise to one-dimensional van Hove singularities (vHs) in the DOS, as

shown in Fig. 10.3. VHs determine many solid state properties of carbon nanotubes,
which affect the spectra observed by scanning tunneling spectroscopy (STS),
optical absorption, and resonant Raman spectroscopy.
By tight binding calculation a simple relation between nanotube diameter d and
the transition energies Esc,m, the energy distance between the van Hove singulari-
ties, valid both for semiconducting (sc) and metallic (m) SWCNT can be defined as:
c0 aCC
Esc;m ¼ 2i
d

with i = 1 for S11, i = 2 for S22 and i = 3 for M11; γ0 = 2.9 eV is the nearest
neighbour carbon–carbon interaction energy and aC–C = 0.144 nm is the nearest
neighbour carbon–carbon distance. In UV/Vis/NIR absorption spectroscopy [9], the
optical transitions, arising from vHs, are probed. By absorption of light, the elec-
trons in the van Hove singularities of the valence band jumps to the corresponding
energy levels in the conduction band (the S11 and S22 in semiconducting tubes). In
principle, the same process happens in the case of metallic SWCNTs. However, the
spacing between the van Hove singularities is larger so that only M11 transitions can
be probed in the UV/Vis region (Fig. 10.4).
Also in the phonon DOS, carbon nanotubes exhibit other unusual aspects
regarding their phonon dispersion relations [10], such as four acoustic branches. In
addition to the longitudinal acoustic and transverse acoustic modes, there are two
acoustic twist modes for rigid rotation around the tube axis, which are important for
heat transport and charge carrier scattering. In SWNT Raman spectroscopy, the
G-band spectra, present also in graphene, which is split into many features around
1580 cm−1, and the lower frequency radial breathing mode (RBM) are usually the
strongest features, and are both first-order Raman modes. The RBM is a unique
phonon mode, appearing only in carbon nanotubes and its observation in the Raman
10 Semiconducting Carbon Nanotubes … 243

Fig. 10.4 The density of state of semiconducting (a) and metallc (b) are sketched, highlighting the
S11, S22 an M11 transitions. These transitions give rise to the three peaks in the optical absorption
spectrum (c)

Fig. 10.5 Raman spectrum of

a SWCNT. The vibrational
modes associated to RBMs
and the G-band are
represented

spectrum provides direct evidence of the presence of SWNTs. The RBM is a

bond-stretching out-of-plane phonon mode for which all the carbon atoms move
coherently in the radial direction, and whose frequency is about 100–500 cm−1.
The RBM frequency is inversely proportional to the tube diameter.
Carbon materials presents also two relatively weak Raman signals whose pho-
non frequencies change with different laser excitation energy, which is called
“dispersive” behavior. In graphite as well as in SWNTs, the D-band at 1350 cm−1
and the G’-band at 2700 cm−1, excited by a laser photon energy of 2.41 eV, are,
respectively, due to one- and two-phonon, second-order Raman scattering pro-
cesses. In particular, because the D-band was firstly observed in the presence of
defects in an otherwise perfect infinite graphite structure, it has been associated to
structural defects in the sample (Fig. 10.5).
244 C. Pintossi and L. Sangaletti

10.2 CNTs Synthesis

Since their discover, a lot of efforts have been made in the optimization of CNTs
synthesis techniques, in order to produce large quantities of pure, diameter sorted
tube, with low fabrication costs [11]. Even though most of these processes usually
take place in vacuum, new techniques have recently been developed for synthetizing
CNTs at atmospheric pressure in order to make nanotubes more commercially viable.
In this paragraph, arc discharge, laser ablation, plasma torch and chemical vapor
deposition will be presented and discussed with more details.
• Arc discharge [12] has been the first CNT production method used since 1991.
In fact, nanotubes were first observed in the carbon soot of graphite electrodes
during an arc discharge that was aimed to produce fullerenes.
In this method two carbon rods are placed end to end, separated by approxi-
mately 1 mm, in an enclosure that is usually filled with helium at relatively low
pressure (between 50 and 700 mbar). A direct current of 50–100 A driven by
approximately 20 V creates a high temperature (*4000 K) discharge between
the two electrodes. The discharge vaporizes the anode carbon rod and forms a
small rod shaped deposit on the cathode. The yield for this method is up to 30 %
by weight and it can be used to produce both single- and multi-walled nanotubes
with lengths of up to 50 µm with a low number of structural defects. For
producing single-walled CNTs, metal catalysts, such as Fe, Co and Ni, are
needed. The SWCNTs diameter distribution depends on the composition of the
catalyst and on the growth conditions. If no catalysts are used, mostly MWCNTs
are produced with a typical outer diameter of approximately 10 nm (Fig. 10.6)
• Laser ablation [13]: a pulsed laser vaporizes a graphite target, placed in a
furnace at roughly 1200 °C, containing small amounts of a metal catalyst.
During laser ablation, a flow of inert gas flows through the growth chamber to
carry the grown nanotubes to be collected on a cooler surface. The produced
nanotubes are mostly in the form of ropes (bundles) consisting of tens of

Fig. 10.6 Schematic drawing

of an arc discharge apparatus
for CNTs production
10 Semiconducting Carbon Nanotubes … 245

Fig. 10.7 Representation of a

laser ablation system

individual nanotubes close-packed together via van der Waals forces. The yield
of nanotube synthesis by this process is roughly 70 % (Fig. 10.7).
• Plasma torch [14] the plasma torch operation conditions are similar to those
used in the arc discharge and laser ablation approaches, but, instead of graphite
vapors, a carbon-containing gas is used to supply the carbon necessary for the
production of SWNT. In this way, the growth of SWNT is more efficient be-
cause decomposing a carbon containing gas can be 10 times less
energy-consuming than graphite vaporization. To produce a continuous process,
a gas mixture composed of argon, ethylene and ferrocene is introduced into a
plasma torch, where it is atomized by the atmospheric pressure plasma, which
has the form of an intense flame. The fumes created by the flame are found to
contain SWNT, amorphous carbon as well as metallic and carbon nanoparticles
(Fig. 10.8).
• Chemical vapor deposition (CVD) [15] is nowadays the most popular method
of producing CNTs. In this process, thermal decomposition of a hydrocarbon
vapor is achieved in the presence of a metal catalyst. The process involves the
passage of an hydrocarbon vapor through a tubular reactor in which a catalyst
material is present at a temperature sufficiently high (600–1200 °C) to decom-
pose the hydrocarbon. CNTs grow on the catalyst in the reactor and are collected
upon cooling the system to room temperature. In the case of a liquid hydrocarbon
(benzene, alcohol, etc.), the liquid is heated in a flask and an inert gas is purged
through it, which in turn carries the hydrocarbon vapor into the reaction zone. If a
solid hydrocarbon is used, it can be directly kept in the low-temperature zone of
the reaction tube. Volatile materials (camphor, naphthalene, ferrocence etc.)
directly turn from solid to vapor, and perform CVD while passing over the
catalyst kept in the high-temperature zone. Like the CNT precursors, also the
catalyst precursors in CVD may be used in any form: solid, liquid or gas, which
may be suitably placed inside the reactor or fed from outside. Pyrolysis of the
catalyst vapor at a suitable temperature liberates metal nanoparticles in situ (such
a process is known as floating catalyst method). Alternatively, catalyst-coated
substrates can be placed in the hot zone of the furnace to catalyze the CNT
growth.
246 C. Pintossi and L. Sangaletti

Fig. 10.8 Schematic drawing

of a plasma torch system

Compared to arc-discharge and laser-ablation methods, CVD is a simple and

economic technique for synthesizing CNTs at relatively low temperature and
ambient pressure. In the case of MWCNTs, arc- and laser-grown tube possess better
crystallinity compared to the CVD-grown ones, but the crystallinity of SWCNTs
grown by CVD is close to that grown by arc or laser methods. CVD result to
overcome arc and laser methods in yield and purity, reaching the best performances
in terms of structure control and CNT architecture. CVD is versatile because it
works with various type hydrocarbons in any state (solid, liquid or gas), enables the
use of various substrates, and allows CNT growth in a variety of forms, such as
powder, thin or thick films, aligned or entangled, straight or coiled nanotubes with
the possibility of controlling the growth parameters (Fig. 10.9).

10.3 Carbon Nanotubes Applications

10.3.1 Selected Applications of Semiconducting CNTs

So far CNTs have been tested for a considerable number of applications spanning
quite different areas from nanoelectronics to biosensing [16–24]. In particular, they
have been considered in micro- and nanoelectronics, photonics, in the field of solar
cells, for hydrogen and energy storage, as chemiresistors in gas sensing and
10 Semiconducting Carbon Nanotubes … 247

Fig. 10.9 Schematic representation of a chemical vapor deposition (CVD) chamber

environmental monitoring, and as structural materials in composites. Among these

applications, nanoelectronics is most probably the one that merges issues related to
basic physics (e.g. transport in 1D systems) to the capability to design, engineer,
produce, and test novel architectures for electronic components at the nanoscale.
A lot of effort was focused on the capability to produce single-wire FET devices. It
is interesting to note that FET-based devices have also been proposed in the field of
gas sensing [25], showing that cross-fertilization among different research areas is a
process that CNT can enable due to their versatility as functional materials.
In the following, two applications are discussed, namely the use of SWCNT in
hybrid solar cells, and the development of gas sensors based on SWCNT layers.
These application share as a common feature the relatively easy preparation route of
the active SWCNT layer and the possibility to achieve quite good results without
resorting to control the layer at the nanoscale. Indeed even with CNT bundles,
devices can be brought to operation, enabling the test of novel ideas and device
architectures at relatively low cost.

10.3.2 CNTs for Photovoltaic Applications

In recent years, CNTs have played different roles in photovoltaic devices.

Nanotubes are in fact used as a part of the photoactive layer, as a transparent
electrode or in dye-sensitized solar cells, but among the most promising alternatives
248 C. Pintossi and L. Sangaletti

Fig. 10.10 A sketch of a hybrid CNT/Si solar cell (a) and its section view (b), with the CNT film
on top (in gray), two Au contact (in yellow), n-type silicon in silver and silicon oxide in purple.
J-V curves (c) and Si2p XPS peaks (d) are presented for different oxidation time: pristine sample
(pink circles), without oxides (gray diamonds) and at the optimal oxidation state (blue triangles)

to silicon photovoltaics (PV) there are hybrid devices based on heterojunctions in

which the p-type silicon is completely replaced by SWCNT, which are naturally
p-doped in air by molecular oxygen (Fig. 10.10a, b).
The remarkable power conversion efficiencies (PCE) so far displayed by devices
based on these junctions [26, 27] are strictly related to the peculiar physical
properties of CNTs. On one side, single-SWCNT are excellent candidates for
participating in photogeneration processes, because of their capability of absorbing
light over a wide range of the solar spectrum, thanks to an energy bandgap variable
from 0 to 3 eV, depending on the tube diameter and to the presence in their DOS of
the vHs. On the other side, it has been recently proved [28] that SWCNTs do not
merely act as a light absorber layer but, due to their extraordinary electrical con-
ductivity (*106 S/cm) [29], they are also able to provide efficient transport paths
for carriers, which have a very high probability to reach electrodes and generate
photocurrent in devices based on CNTs photoexcitation processes [30].
A considerable effort has been recently made to increase the efficiency of
CNT-Si heterojunctions, until reaching a record PCE of about 15 % [31]. However,
much has still to be done in order to identify the mechanisms behind the device
operation and determine the parameters that mostly affect the cell behavior. In
10 Semiconducting Carbon Nanotubes … 249

addition to a number of technical issues related to the cell geometry, to contacting

pads, and to the choice of the CNT layer, several studies have pointed out [32, 33]
that a silicon oxide layer grows at the interface between monocrystalline silicon and
CNTs not only during the manufacturing process but also once the junction is built
and that the properties of this layer affect the overall cell performances.
The role played by the oxide layer is of fundamental importance for under-
standing the working mechanism behind these devices that could behave either as
p-n junctions [34] or as a metal-insulator-semiconductor (MIS) devices [35]. In
particular, in a p-n junction model, the best efficiency should be reached once all the
insulating silicon oxides are removed and perfect junctions between CNTs and Si
can be built. On the other hand, in MIS junctions a relatively low potential barrier
between CNTs and silicon yields a high number of majority carriers recombina-
tions, resulting in a saturation current larger than that of a p-n junction diode and, as
a consequence, in a low open circuit voltage. In these devices a thin layer of
insulating oxide is beneficial because it helps to confine electrons in silicon,
avoiding them to fast recombine in CNTs [36] (Fig. 10.10c). Moreover, a thin
insulating layer may also have the advantage to passivate interface states, hindering
charge trapping, lowering recombinations and increasing the device open circuit
voltage.
The choice between the two junction schemes (i.e. p-n vs. MIS) requires a
specific investigation on the oxide role through direct evidence of the Si oxidation
states during manufacturing and ageing processes to tailor the oxide layer
properties.
Oxide removal from silicon upon HF exposure is expected, as it is a standard
treatment to etch silicon wafers and, on this basis, similar results have been claimed
when CNT/Si heteojunctions were HF etched [37]. The possibility to access the
buried interface layer by angle resolved X-ray photoemission spectroscopy
(AR-XPS) has recently been demonstrated (Fig. 10.10d) and the existence of an
oxide layer with optimal thickness was inferred by several groups [32, 33, 38] to
rationalize the behavior of the cells in the presence or absence of oxides. Therefore,
in spite of a seemingly straightforward preparation route, the CNT/SiOx/Si interface
represents a quite complex system from the point of view of materials, as it matches
a relatively low-density CNT bundle layer with a nanostructured and chemically
inhomogeneous SiOx layer grown on a n-doped silicon wafer.
In particular, in devices based on bundles of randomly aligned SWCNT the
presence of a thin layer of oxides has always lead to better cell performances [32,
33, 38].
As shown by Jia et al. [32], the electrical parameters of a CNT/Si solar cell can
be modulated by changing the thickness of the SiOx layer, either by removing
oxides with HF or by letting these oxides regrow, through either oxidation in
ambient air or exposure to nitric acid.
For understanding the operation mechanism, a combination of suitable spec-
troscopic tools has been used recently [39]: AR-XPS for a direct probe of the buried
oxide interface, Raman spectroscopy to obtain information on CNT properties, and
250 C. Pintossi and L. Sangaletti

the acquisition of current density-voltage (J-V) characteristics for monitoring the

behavior of the photovoltaic device after exposure to acid.
Based on this experimental approach, It has been demonstrated that a relatively
mild HF etching does not change the CNT film properties, acting only as silicon
oxides etcher. This finding addresses a long-standing question on the unaligned
CNT-Si hybrid junctions, probing the role of the complex silicon oxide interface to
tailor the PV cell performances. The build-up of an interface Si oxide layer is
required to increase the efficiency. This is assumed as a strong indication that the
junction can be described with a MIS model rather than a p-n junction. These
results indicate that the SiOx buried interface can be regarded as a passivating and
inversion layer, which improves the cell performances by hindering the e-h pair
recombination.

10.3.3 CNT Interaction with Gases: From Surface

Chemistry to Devices

Up to now CNT have been tested against many polluting gases, as they are known
to interact with many gas molecules that, depending on their reducing or oxidizing
properties, may inject or extract electrons from the CNTs, resulting in a detectable
electrical signal. Indeed, a large body of studies reports the response of CNTs to
many gases, in particular to ammonia and nitrogen dioxide, as prototypes of
reducing and oxidizing gas, respectively. Many CNT-based gas sensors devices
have been tested so far, showing that CNTs are a promising class of materials for
the development of gas sensors [40–53].
The gas sensing capability of CNT is based on the possible interactions of
molecules with the CNT walls. The basic interaction mechanism can be clearly
evidenced in semiconducting CNT. It has been shown by Collins et al. [54] that the
presence of oxygen induces significant effects on the conductivity and the DOS of
CNTs. Indeed, CNTs readily become p-doped when exposed to the oxygen
molecules present in air and therefore the interaction with molecules capable to
transfer charge to and from CNTs can increase or decrease the number of carries
(holes), depending on the reducing or oxidizing properties of the target gas mole-
cules. These interactions have been explored in experimental [41, 54–57] and
theoretical works using ab initio calculations and density functional theory
(DFT) [58–60]. The distribution of electrons around the CNT make it electro-
chemically active. Therefore when oxidizing or reducing gas molecules, e.g. NO2
or NH3 respectively, are adsorbed onto carbon nanotubes surface, the charge
transfer affects the electronic properties in terms of band structure, DOS, and
electrical conductivity of CNTs. Many studies have been inspired by the experi-
mental evidences presented in [55], where the variation of the electrical properties
of CNTs upon exposure to NH3 and NO2 was clearly evidenced. On the theoretical
side, one of the first works based on DFT was performed by Zhao et al. [58] who
10 Semiconducting Carbon Nanotubes … 251

Fig. 10.11 Electronic DOS of (10, 0) carbon nanotubes after adsorption of NH3 (top panel), and
NO2 (bottom panel) compared with the DOS of (10, 0) pristine CNTs. Dashed lines denote the
Fermi level. b Contour plot of the electron charge density (slice passing through SWCNT and NO2
molecule) [58]

studied the interaction of a large set of molecules (NO2, O2, H2O, NH3, CH4, CO2,
H2, N2, Ar) and SWCNTs with different chirality, clarifying the mechanisms of
charge transfer, the adsorption sites, and the CNT-molecule minimum distance. The
gas molecule adsorption on the SWCNT surface may induce a significant variation
of the electronic properties due to the charge transfer. In Fig. 10.11, the density of
electronic states of pure (10, 0) carbon nanotubes and upon absorption of NH3 and
NO2 (respectively in middle, top, and bottom panel) is shown. The NH3 effect is
predicted to be weaker with respect to NO2 interaction which shifts the Fermi level
into the valence band. Nevertheless, detection of NH3 is still a benchmarking
measure to test the capability of CNT to interact with small reducing molecules.
In turn, the strong interaction of NO2 with CNTs may result in poorly reversible
interactions, leaving the CNT layer poisoned by NO2 and therefore, to some extent,
less capable to restore the pristine electrical parameters and to proceed with the
detection of further exposure to this target molecule.
According to Zhao et al. [58] for molecules like H2O, NH3, CO2, which are
donors, i.e. they inject electrons on the nanotube surface, the charge transfer
towards the tube is low (0.01–0.035 electron per molecule) and the adsorption
energy defined as Ea(d) = Etot(tube + molecule) − Etot(tube) − Etot(molecule) is
weak (0.2 eV), d being the equilibrium distance between the tube and the molecule,
while for the NO2 acceptor which induces high charge transfer from the tube (0.06–
0.14 electron per molecule), the adsorption energies are also greater (0.4 eV).
252 C. Pintossi and L. Sangaletti

In order to exemplify the interaction between gas molecules and bundle of

SWCNTs, in Fig. 10.12 we report the case of an H2 molecule adsorbed on (10, 10)
SWCNTs bundle considered by Zhao et al. [58]. As one can observe from the table
on the right side of Fig. 10.12, the adsorption energy and charge transfer in the
interstitial and groove sites are greater with respect to that on the surface sites. This
can be explained by considering the greater number of carbon nanotubes that
interact with the molecule. It is important to highlight that the possibility to be
adsorbed in interstitial sites is reduced for large molecules.
From the experimental point of view, reference measurements have been pre-
sented by Goldoni et al. [41], based on X-ray photoemission spectroscopy (XPS). In
particular, the interaction between SWCNTs and gas molecules like SO2, NH3, O2
or NO2 has been studied. The authors show that the reaction with oxygen is not due
to carbon nanotube interaction, but to the presence of residual contaminants of the
purification process (especially the species containing Na). Once these contami-
nants are completely removed, SWCNTs appear not very sensitive to O2, CO, H2O
e N2, while a strong sensitivity to NO2, SO2, and NH3 is still observed. In
Fig. 10.13 XPS C 1s spectra of the SWCNT clean sample, i.e. without

Fig. 10.12 Illustration of possible adsorption sites for the H2 molecule in the (10, 10) SWCNTs
bundle. a Surface; b pore; c groove; d interstitial. In the table on the right, the equilibrium
tube-molecule distance (d), adsorption energy (Ea) and charge transfer (Q) on the different
adsorption sites are reported [58]

Fig. 10.13 C 1s spectra of the clean SWCNT bucky-paper measured while exposing the sample at
150 K to SO2 (a), NH3 (b), and NO2 (c) at partial pressures of about 10−8 mbar [41]
10 Semiconducting Carbon Nanotubes … 253

contaminants, during gas exposure of SO2, NH3 and NO2 are shown. The shift of C
1s peak during gas exposure is toward lower binding energies in the case of SO2
and NO2 (Fig. 10.13a, c), while during NH3 exposure (Fig. 10.13b) the shift is
smaller and toward higher binding energies. If these binding energy shifts are
ascribed to shifts of the nanotube chemical potential, the NO2 and SO2 molecules
bound to SWCNTs act as charge acceptors, while the NH3 molecules act as charge
donors, in full agreement with the experimental observations reported by Kong
et al. [55]. and theoretical results [58]. The results of [41], clearly address one
typical problem of CNT synthesis, i.e. the capability to control contaminations
related to the CNT growth and sample preparation, mostly in terms of residual
impurities from metal catalysts, as well as surfactants used to disperse CNTs and
prepare thin layers.
In addition to the above mentioned capability to interact with target gas mole-
cules, SWCNTs soon resulted to be appealing systems to be used as gas sensors for
three other reasons at least: the huge surface to volume ratio, the remarkable
transport properties that favor charge drift towards the electrodes, and their
one-dimensional nature that makes them very sensitive to very small external
perturbations [51]. Furthermore, gas sensors based on SWCNTs are able to operate
at room temperature, coping with constraints on power consumption in portable
devices. In fact, the use of nanostructured metal oxide layers in gas sensing (so far
the most diffuse materials in chemiresistors) requires to operate at relatively high
temperatures (few hundreds of °C) by coupling the sensing layer with suitable
resistive heaters.
As discussed above, gas molecules adsorbed on the carbon nanotube surface
could act as dopants (donors or acceptors), they could shift the Fermi level or
change the band structure of SWCNT. Such a change may be easily detected, once
the nanotubes are inserted within an electronic device. The different types of
electrochemical sensors based on SWCNTs electrical properties (e.g. resistance [44,
55, 61–63], capacitance [64] or frequency [65] variation, ionization chamber with
CNTs [66]) can be grouped in: sensors based on impedance, piezoelectric
transducer-based sensors and solid state miniaturized sensors [67]. The chemire-
sistor and the chemFET (chemical Field Effect Transistor) belong to the solid state
miniaturized sensors category. In Fig. 10.14, the CNT-based devices and the
electronic circuit schemes in the chemiresistor and chem-FET configuration are
shown.
In the chemiresistor configuration, the CNTs create a bridge between two
electrodes (to which a certain potential is applied), allowing the current flow. When
a reactive gas molecule is adsorbed on the nanotube surface, the charge transfer
alters the electrical resistance of the CNTs.
The chemFET configuration is similar to the chemiresistor configuration, except
for the fact that the current flowing through the CNTs between source (S) and drain
(D) electrodes is modulated by a third electrode, the gate (G), separated from the
other two by a thin layer of insulating material (typically SiO2). This metal/CNT
semiconductor/metal system has the characteristics of a p-type transistor [55]. An
254 C. Pintossi and L. Sangaletti

Fig. 10.14 CNT device and electronic circuit scheme in the chemiresistor (a) and chem-FET
(b) configuration

Fig. 10.15 Electrical response of a semiconducting SWCNT to gas molecules in a chem-FET

configuration, but at fixed gate voltage (Figure adapted from [55]). a Conductance (under
Vgate = +4 V, in an initial insulating state) versus time in a 200 ppm NO2 flow. b Conductance
(Vg = 0 V, in an initial conducting state) versus time recorded with the same SWCNT sample in a
flow of Ar containing 1 % NH3. These conductance variations to oxidizing (i.e. NO2) or reducing
(i.e. NH3) molecules indicate the p-type character of the SWCNT response

example of response to NO2 and NH3 exposure of a CNT-based chemFET is shown

in Fig. 10.15.
In Table 10.1 a summary of pristine CNT chemiresistors and ChemFETs sensing
performance is shown [50]. The chemFET configuration tends to be more sensitive
to gas because of the possibility to modulate the conductance of CNTs via the gate
voltage. However it requires more sophisticated electronics and much higher costs
[44]. Chemiresistors are often chosen for the gas sensing measurements because
they are simpler to produce and to manage. For this type of sensors interdigitated
electrodes (IDE) are often used to improve charge collection from the CNT layer.
The main techniques to deposit CNTs are: direct CVD on the substrate; deposition
of a solution in which the CNTs are dispersed, e.g. by drop-casting method or
ink-jet printing. Finally, in order to align CNTs between the electrodes, a method
called dielectrophoresis, which consists on applying an AC electric field between
the electrodes during the deposition process, has been used by Suheiro et al. [63] to
build a MWCNT-based ammonia gas sensor. Usually the signal is processed as
10 Semiconducting Carbon Nanotubes … 255

Table 10.1 Summary of selected sensing performance of pristine CNT chemiresistors and
chemFETs (N/S = Not-stated) [50]
CNT type Sensor Targeted analytes Detection Response Reversibility
configuration limit time (s)
Single SWNT ChemFET NO2, NH3 2 ppm (NO2) <600 Irreversible
0.1 % (NH3)
SWNTs ChemFET Alcoholic vapors N/S 5–150 Reversible
(methanol, ethanol, (−20 V gate
1-propanol, bias
2-propanol and potential)
tert-butanol.)
SWNTs Chemiresistor O2 N/S N/S Reversible
SWNTs ChemFET DMMP <1 ppb 1000 Reversible
(3 V gate
bias)
MWNTs Chemiresistor NO2 l0 ppb N/S Reversible
(165 °C)
SWNTs Chemiresistor NO2, Nitrotoluene 44 ppb (NO2) 600 Reversible
262 ppb (UV)
(Nitrotoluene)
MWNTs Chemiresistor NH3 10 ppm *100 Reversible
SWNTs Chemiresistor SOCl2, DMMP 100 ppm 10 Irreversible
SWNTs Chemiresistor O3 6 pph <600 Reversible
MWNTs Chemiresistor NO2 5–10 ppb −600 Reversible
(165 °C) (165 °C)
SWNTs Chemiresistor Methanol, acetone N/S *100 N/S
SWNTs Chemiresistor H2O N/S 10–100 Reversible
Carboxylated Chemiresistor CO 1 ppm *100 Reversible
SWNT

ΔR/R0, (where ΔR = R − R0, R being the sensor resistance measured during

exposure to gas and R0 the baseline resistance before gas exposure) and the
capability of the sensor to track variations in the target gas concentration is
described by the calibration curve, where the ΔR/R0 values are plotted vs the target
gas concentration. A schematic drawing of these measurements is shown in
Fig. 10.16. The figure has been drawn for the case of a reducing molecule (i.e.
ammonia) and the resistance increase is determined by the injection of electrons in
the p-type CNTs.
Novel perspectives in the development of chemiresistor gas sensors (CGS) can
be determined by the capability to meet the requirements of environmental moni-
toring, in particular the sensitivity in the low ppb range.
Recently, low-cost CNT-based layers prepared on plastic substrates have been
produced to monitor the presence of ammonia in the environment, where the NH3
concentrations in the low-ppb range are expected [68]. A systematic monitoring of
ammonia concentrations [69] is mandatory to reduce the hazard for human health
256 C. Pintossi and L. Sangaletti

Fig. 10.16 a Sketch of sensor response to gas exposure and b response curve

and vegetation determined by the widespread use of ammonia and its derivatives as
agricultural nitrogen fertilizers, which is known to determine a severe environ-
mental problems, such as the eutrophication of semi-natural ecosystems, acidifi-
cation of soils, formation of fine particulate matter in the atmosphere, and alteration
of the global greenhouse balance. In spite of this urgency, the detection of ammonia
atmospheric concentrations in urban areas has been so far widely overlooked, since
its average levels are usually low, i.e. in the 20–30 ppb range [70], while novel
ammonia CGS are not usually tested at these low levels.
Ammonia is one of the molecules most frequently used to test the capability of
these devices to operate as CGS. However, little attention has been paid to the
capability of these sensors to measure sub-ppm [NH3] in air. In fact, ppt sensitivity
of pristine CNTs to ammonia has been demonstrated, in inert Ar atmosphere under
UV irradiation [71], a detection limit (DL) of 50 ppb has been achieved for
ammonia diluted in Ar using CNTs functionalized with polianyline [72], and finally
a detection limit of 30 ppb has been demonstrated for CNT layers on plastic
substrates operating in ambient air [68].
Another emerging issue in the development of high sensitivity CNT-based
devices is represented by hybrid systems where the CNTs are coupled with oxide
nanoparticles. When decorated with oxide nanoparticles, the CNT layers can be
regarded as hybrid systems, displaying properties that can be either ascribed to the
superposition of the parent materials, or to novel properties emerging from the
junction between the CNT wall and the nanoparticle. In the field of gas sensing,
hybrid systems with CNT have been mostly created through decoration with SnO2,
WO3 or TiO2 nanoparticles [73, 74]. These oxides alone, in the form of thin or thick
films, are known as the most diffused materials for gas sensing, with remarkable
sensitivity towards hazardous such gas as NO2, NOx, etc. It is therefore a straight
choice to create hybrids were CNT can provide superior charge transport and
eventually gas sensing capability to be combined to that of oxide. Here, the p-n
heterojunction formed between, e.g., an n-type metal oxide and a p-type carbon
support is supposed to play an important role in the sensing mechanism of MOs
coated CNT heterostructures [75–78].
10 Semiconducting Carbon Nanotubes … 257

References

1. M. Endo. Mecanisme de croissance en phase vapeur de bres de carbone (The growth

mechanism of vapor-grown carbon fibers). Ph.D. thesis, University of Orleans, Orleans,
France, 1975. (in French)
2. M. Endo. Ph.D. thesis, Nagoya University, Japan, 1978. (in Japanese)
3. S. Iijima, J. Cryst. Growth 55, 675–683 (1980)
4. A. Oberlin, M. Endo, T. Koyama, Carbon 14, 133 (1976)
5. A. Oberlin, M. Endo, T. Koyama, J. Cryst. Growth 32, 335–349 (1976)
6. S. Iijima, Nature 354, 56–58 (1991)
7. M.S. Dresselhaus, G. Dresselhaus, Ph Avouris, Carbon Nanotubes. Synthesis, Structure
Properties, and Applications (Springer, 2001)
8. P. Castrucci, C. Scilletta, S. Del Gobbo, M. Scarselli, L. Camilli, M. Simeoni, B. Delley, A.
Continenza, M. De Crescenzi, Nanotechnology 22, 115701 (2011)
9. I. Stemmler, C. Backes, Absorption Spectroscopy as a Powerful Technique for the
Characterization of Single-Walled Carbon Nanotubes (White paper, PerkinElmer, 2013)
10. M.S. Dresselhaus, G. Dresselhaus, R. Saito, A. Jorio, Raman spectroscopy of carbon
nanotubes. Phys. Rep. 409, 47–99 (2005)
11. J. Prasek, J. Drbohlavova, J. Chomoucka, J. Hubalek, O. Jasek, V. Adam, R. Kizek, Methods
for carbon nanotubes synthesis—review. J. Mater. Chem. 21, 15872 (2011)
12. N. Arora, N.N. Sharma, Arc discharge synthesis of carbon nanotubes: Comprehensive review.
Diam. Relat. Mater. 50, 135–150 (2014)
13. T. Guo, P. Nikolaev, A. Thess, D.T. Colbert, R.E. Smalley, Catalytic growth of single-walled
nanotubes by laser vaporization. Chem. Phys. Lett. 243, 49–54 (1995)
14. M. Pacheco, J. Pacheco, M. Valdivia, L. Bernal, R. Valdivia, A. Huczko, H. Lange, A. Cruz,
R. Lopez-Callejas, Synthesis of Carbon Nanostructures by Using Thermal Plasma Torch.
Braz. J. Phys. 34, 4B (2004)
15. J. Kong, A.M. Cassell, Hongjie Dai, Chemical vapor deposition of methane for single-walled
carbon nanotubes. Chem. Phys. Lett. 292, 567–574 (1998)
16. M.F.L. De Volder, S.H. Tawfick, R.H. Baughman, A.J. Hart, Carbon Nanotubes: Present and
Future Commercial Applications. Science 339(6119), 535–539 (2013)
17. M. Meyyappan (ed.), Carbon Nanotubes: Science and Applications (CRC Press, 2005)
18. A. Javey, J. Kong (eds.), Carbon Nanotube Electronics (Springer Science + Business Media,
2009)
19. Dirk M. Guldi and Nazario Martín (eds.), Carbon Nanotubes and Related Structures,
Synthesis, Characterization, Functionalization, and Applications (Wiley-VHC, 2010)
20. S. Yamashita, Y. Saito, J.H. Choi (eds.) Carbon Nanotubes and graphene for photonics
applications (Woodhead Publishing, 2013)
21. C. Hierold (ed.), Advanced Micro & Nanosystems, vol. 8, Carbon Nanotube Devices:
Properties, Modeling, Integration and Applications (Wiley-VHC, Weinheim, 2008)
22. Rüdiger Klingeler Robert, B. Sim (eds.), Carbon Nanotubes for Biomedical Applications
(Springer, Berlin, Heidelberg, 2011)
23. Tony McNally, Petra Potschke (eds.), Polymer Carbon Nanotube Composites: Preparation:
Properties and Applications (Woodhead Publishing, London, 2011)
24. Amin Salehi-Khojin, Fatemeh Khalili-Araghi, Marcelo A. Kuroda, Kevin Y. Lin, Jean-Pierre
Leburton, Richard I. Masel, On the sensing mechanism in carbon nanotube chemiresistors.
ACS Nano 5(1), 153–158 (2011)
25. J. Kong, N.R. Franklin, C. Zhou, M.G. Chapline, S. Peng, K. Cho, H. Dai. Science 287, 622
(2000)
26. D.D. Tune, B.S. Flavel, R. Krupke, J.G. Shapter, Carbon nanotube-silicon solar cells. Adv.
Energy Mater. 2, 1043–1055 (2012)
27. P. Castrucci, Carbon nanotube/silicon hybrid heterojunctions for photovoltaic devices. Adv.
Nano Res. 2, 23–56 (2014)
258 C. Pintossi and L. Sangaletti

28. D.D. Tune, F. Hennrich, S. Dehm, M.F.G. Klein, K. Glaser, A. Colsmann, J.G. Shapter, U.
Lemmer, M.M. Kappes, R. Krupke, B.S. Flavel, The role of nanotubes in carbon
nanotube-silicon solar cells. Adv. Energy Mater. 3, 1091–1097 (2013)
29. P.L. Mc Euen, J.Y. Park, Electron transport in single-walled carbon nanotubes. MRS Bull. 29,
272–275 (2004)
30. S. Ponzoni, G. Galimberti, L. Sangaletti, P. Castrucci, S. Del Gobbo, M. Morbidoni, M.
Scarselli, S. Pagliara, Selective optical switching of interface-coupled relaxation dynamics in
carbon nanotube–Si heterojunctions. J. Phys. Chem. C 118, 24110–24116 (2014)
31. E. Shi, L. Zhang, Z. Li, P. Li, Y. Shang, Y. Jia, J. Wei, K. Wang, H. Zhu, D. Wu, S. Zhang, A.
Cao, TiO2-coated carbon nanotube-silicon solar cells with efficiency of 15 %. Sci. Rep. 2
(884), 1–5 (2012)
32. Y. Jia, A. Cao, F. Kang, P. Li, X. Gui, L. Zhang, E. Shi, J. Wei, K. Wang, H. Zhu, D. Wu,
Strong and reversible modulation of carbon nanotube-silicon heterojunctions solar cells by an
interfacial oxide layer. Phys. Chem. Chem. Phys. 14, 8391–8396 (2012)
33. D.D. Tune, A.J. Blanch, R. Krupke, B.S. Flavel, J.G. Shapter, Nanotube film metallicity and
its effect on the performance of carbon nanotube-silicon solar cells. Phys. Status Solidi A 211,
1479–1487 (2014)
34. Y. Jung, X. Li, N.K. Rajan, A.D. Taylor, M.A. Reed, Record high efficiency single-walled
carbon nanotube/silicon p–n junction solar cells. Nano Lett. 13, 96–99 (2013)
35. P. Wadhwa, B. Liu, M.A. McCarthy, Z. Wu, A.G. Rinzler, Electronic junction control in a
nanotube-semiconducting schottky junction solar cell. Nano Lett. 10, 5001–5005 (2010)
36. H.C. Card, Photovoltaic properties of MIS-schottky barriers. Solid State Electron. 20, 971–976
(1977)
37. X. Li, J.S. Huang, S. Nejati, L. McMillon, S. Huang, C.O. Osuji, N. Hazari, A.D. Taylor, Role
of HF in oxygen removal from carbon nanotubes: implications for high performance carbon
electronics. Nano Lett. 14, 6179–6184 (2014)
38. C. Pintossi, G. Salvinelli, G. Drera, S. Pagliara, L. Sangaletti, S. Del Gobbo, M. Morbidoni,
M. Scarselli, M. De Crescenzi, P. Castrucci, Direct evidence of chemically inhomogeneous,
nanostructured, Si–O buried interfaces and their effect on the efficiency of carbon nanotube/Si
photovoltaic heterojunctions. J. Phys. Chem. C 117, 18688–18696 (2013)
39. C. Pintossi, S. Pagliara, G. Drera, F. De Nicola, P. Castrucci, M. De Crescenzi, M. Crivellari,
M. Boscardin, L. Sangaletti, Steering the efficiency of carbon nanotube-silicon photovoltaic
cells by acid vapor exposure: a real-time spectroscopic tracking. ACS Appl. Mater. Interfaces
7(18), 9436–9444 (2015)
40. K.A. Mirica, J.G. Weis, J.M. Schnorr, B. Esser, T.M. Swager, Mechanical drawing of gas
sensors on paper. Angew. Chem. Int. Ed. 51, 10740–10745 (2012)
41. A. Goldoni, L. Petaccia, L. Gregoratti, B. Kaulich, A. Barinov, S. Lizzit, A. Laurita, L.
Sangaletti, R. Larciprete, Carbon 42, 2099 (2004)
42. S. Chopra, K. McGuire, N. Gothard, A.M. Rao, A. Pham, Appl. Phys. Lett. 83, 2280 (2003)
43. T. Someya, J. Small, P. Kim, C. Nuckolls, J.T. Yardley, Nano Lett. 3, 877 (2003)
44. J. Li, Y. Lu, Q. Ye, M. Cinke, J. Han, M. Meyyappan, Nano Lett. 3, 929 (2003)
45. J.P. Novak, E.S. Snow, E.J. Houser, D. Park, J.L. Stepnowski, R.A. McGill, Appl. Phys. Lett.
83, 4026 (2003)
46. F. Picaud, C. Girardet, A.M. Rao, J. Appl. Phys. 105, 014315 (2009)
47. P. Vichchulada, P.Q. Zhang, M.D. Lay, Analyst 132, 719 (2007)
48. D.R. Kauffman, A. Star, Angew. Chem. 47, 6550 (2008)
49. M. Penza, R. Rossi, M. Alvisi, M.A. Signore, E. Serra, J. Phys. D Appl. Phys. 42, 072002
(2009)
50. T. Zhang, S. Mubeen, N.V. Myung, M.A. Deshusses, Nanotechnology 19, 332001 (2008)
51. A. Goldoni, L. Petaccia, S. Lizzit, R. Larciprete, J. Phys.: Condens. Matter 22, 013001 (2010)
52. M. Penza, R. Rossi, M. Alvisi, E. Serra, Nanotechnology 21, 105501 (2010)
53. M. Penza, P.J. Martin, J.T.W. Yeow, Carbon Nanotube Gas sensors, in Gas Sensing
Fundamentals. Springer Series on Chemical Sensors and Biosensors, ed. by C.-D. Kohl, T.
Wagner (Springer, Berlin, Heidelberg, 2014)
10 Semiconducting Carbon Nanotubes … 259

54. P.G. Collins, K. Bradley, M. Ishigami, A. Zettl, Science 287, 1801 (2000)
55. J. Kong, N.R. Franklin, C. Zhou, M.G. Chapline, S. Peng, K. Cho, H. Dai, Science 287, 622
(2000)
56. G.U. Sumanasekera, C.K.W. Adu, S. Fang, P.C. Eklund, Phys. Rev. Lett. 85, 1096 (2000)
57. X.P. Tang, A. Kleinhammes, H. Shimoda, L. Fleming, K.Y. Bennoune, S. Sinha, C. Bower,
O. Zhou, Y. Wu, Science 288, 492 (2000)
58. J. Zhao, A. Buldum, J. Han, J.P. Lu, Gas molecule adsorption in carbon nanotubes and
nanotube bundles. Nanotechnology 13, 195–200 (2002)
59. H. Chang, J.D. Lee, S.M. Lee, Y.H. Lee, Appl. Phys. Lett. 79, 3863 (2001)
60. B.B. Shirvani, J. Beheshtian, G. Parsafar, N.L. Hadipour, Comput. Mater. Sci. 48, 655 (2010)
61. L. Valentini, C. Cantalini, I. Armentano, J.M. Kenny, L. Lozzi, S. Santucci. 21, 1996 (2003)
62. C.Y. Lee, S. Baik, J.Q. Zhang, R.I. Masel, M.S. Strano, J. Phys. Chem. B 110, 11055 (2006)
63. J. Suehiro, G.B. Zhou, M. Hara, J. Phys. D Appl. Phys. 36, 109 (2003)
64. E.S. Snow, F.K. Perkins, E.J. Houser, S.C. Badescu, T.L. Reinecke, Science 307, 1942 (2005)
65. S. Chopra, A. Pham, J. Gaillard, A. Parker, A.M. Rao, Appl. Phys. Lett. 80, 4632 (2002)
66. A. Modi, N. Koratkar, E. Lass, B.Q. Wei, P.M. Ajayan, Nature 424, 171 (2003)
67. C. Brett, Other types of sensors: Impedance-based sensors, fet sensors, acoustic sensors, in
Environmental Analysis by Electrochemical Sensors and Biosensors, Nanostructure Science
and Technology, eds. by L.M. Moretto, K. Kalcher (Springer, 2014), pp. 351–370
68. F. Rigoni, S. Tognolini, P. Borghetti, G. Drera, S. Pagliara, A. Goldoni, L. Sangaletti, Analyst
138, 7392–7399 (2013)
69. B. Timmer, W. Olthuis, A. van den Berg, Sens. Actuators B 107, 666 (2005)
70. M. Chiesa, F. Rigoni, M. Paderno, P. Borghetti, G. Gagliotti, M. Bertoni, A. Ballarin, Denti,
L. Schiavina, A. Goldoni and L. Sangaletti. J. Environ. Monit. 14, 1565 (2012)
71. G. Chen, T.M. Paronyan, E.M. Pigas, A.R. Harutyunyan, Sci. Rep. 2, 343 (2012)
72. T. Zhang, M.B. Nix, B.-Y. Yoo, M.A. Deshusses, N.V. Myung, Electroanalysis 18, 1153
(2006)
73. T. Kerdcharoen, C. Wongchoosuk, Carbon nanotube and metal oxide hybrid materials for gas
sensing. in Semiconductor Gas Sensors, Woodhead Publishing Series in Electronic and
Optical Materials, eds. by R. Jaaniso, O.K. Tan, pp. 386–407 (2013)
74. D. Eder, Carbon nanotube-inorganic hybrids. Chem. Rev. 110, 1348–1385 (2010)
75. C. Marichy, P.A. Russo, M. Latino, J.P. Tessonnier, M.G. Willinger, N. Donato, G. Neri, N.
Pinna, Tin dioxide–carbon heterostructures applied to gas sensing: structure-dependent
properties and general sensing mechanism. J. Phys. Chem. C 117, 19729–19739 (2013)
76. C. Marichy, N. Donato, M.-G. Willinger, M. Latino, D. Karpinsky, S.-H. Yu, G. Neri, N.
Pinna, Tin dioxide sensing layer grown on tubular nanostructures by a non-aqueous atomic
layer deposition process. Adv. Funct. Mater. 21, 658–666 (2011)
77. E. Llobet, R. Ionescu, E.H. Espinoza, R. Leghrib, A. Felten, R. Erni, Sens. Actuators B 131,
174 (2008)
78. B.-Y. Wei, M.-C. Hsu, P.-G. Su, H.-M. Lin, R.-J. Wu, H.-J. Lai, Sens. Actuators B Chem.
101, 81–89 (2004)
Chapter 11
Effects of Charging and Perpendicular
Electric Field on Graphene Oxide

H. Hakan Gürel, M. Topsakal and S. Ciraci

Abstract We present a first-principles study of the effects of charging, electric field

on the oxidation/deoxidation of graphene oxide consisting of epoxy and hydroxly
groups. We first determined the proper basis set which prevents the spurious
spilling of electrons of graphene oxide, when negatively charged or exerted by
perpendicular electric field and treated with periodic boundary conditions. Applied
perpendicular electric field to graphene surface is provided side specific function-
alization. We showed that the bonds between oxygen and graphene are weakened
under applied electric field. For specific values of excess charge or perpendicular
electric field, oxygen atom moves to the top site from bridge site which is normally
absorbed in equilibrium. Individually adsorbed oxygen atoms cannot form oxygen
molecules due to an energy barrier. This energy barrier is dramatically weakened,
when negatively charged or exerted by an electric field. Beside the epoxy groups,
hydroxyl groups have an important role of oxidation/deoxidation of graphene
oxide. Charging and perpendicular electric field mediates the reduction of graphene
oxide through the formation of H2O and H2O2. Our results explain the role of
external effects to the reduction of graphene oxide.

H. Hakan Gürel
Technology Faculty, Information Systems Engineering Department,
Kocaeli University, 41380 Kocaeli, Turkey
e-mail: [email protected]
M. Topsakal
UNAM-National Nanotechnology Research Center, Bilkent University,
06800 Ankara, Turkey
e-mail: [email protected]
S. Ciraci (&)
Department of Physics, Bilkent University, 06800 Ankara, Turkey
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 261

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_11
262 H. Hakan Gürel et al.

11.1 Introduction

Graphene oxide (GOX) has been an important material because it allows the pro-
duction of large scale graphene sheets through the reduction of oxidized multilayer
graphene [1, 2]. Despite the great amount of experimental and theoretical research
carried out recently, [1–16] a thorough understanding of the interaction of epoxy
and hydroxyl groups with graphene resulting in oxidation/deoxidation process are
not fully understood yet. Active research to date has concluded that the oxidation
and reduction of graphene are, in fact, rather complex and comprise the interplay of
various molecules and atoms, such as O, O2, CO2, OH, H, H2O, as well as external
agents. GOX has been also a subject of interest because the electronic properties of
graphene, in particular its linear p and p
bands, [17] which cross at the Fermi level
undergo dramatic changes upon oxidation [18–21]. Introducing a band gap, which
varies with oxygen coverage and hence changes semimetallic graphene into a
semiconductor, has been an active field of study in graphene based nanoelectronics.
Furthermore, it has been reported that the oxidized surface of graphene can be
changed reversibly between dark and light spots with the applied lateral electric
field [21]. To tune the band gap, the extent of reduction can also be controlled by
applied perpendicular electric field, whereby deoxidized spots are attained under-
neath an STM tip under a specific bias [21]. These results show that the properties
of graphene can be modified by the controlled and reversible reduction/oxidation of
GOX, which are realized by charging it or by applying perpendicular electric field.

11.2 Methodolgy

We carried out spin-polarized and spin-unpolarized calculations within density

functional theory (DFT) using projector-augmented wave potentials [22]. The
exchange-correlation potential is approximated by generalized gradient approxi-
mation using Perdew, Burke and Ernzerhof (PBE) functional [23]. Initially, we used
two different basis sets, namely plane waves (PW) and local basis (AO) sets for
reasons explained below, which are applied by using first-principles simulation
packages VASP [24] and SIESTA, [25] respectively. Based on the results derived
from the test calculations, all of our conclusions concerning the effect of the
charging and electric field are obtained from AO calculations using SIESTA [25].
A plane-wave basis set with kinetic energy cutoff of 500 eV is used in PW
calculations [24]. Atomic positions and lattice constants are optimized by using the
conjugate gradient method, where the total energy and atomic forces are minimized.
The convergence for energy is chosen as 10−5 eV between two steps.
Oxygen-adatom adsorbed to one side of graphene breaks inversion symmetry and a
net electric-dipole moment is generated perpendicular to the graphene surface.
Dipole corrections [26] are applied in order to remove spurious dipole interactions
between periodic images for the neutral calculations.
11 Effects of Charging and Perpendicular Electric Field … 263

In AO calculations [25], we use double f polarized basis set and the eigenstates
of the Kohn-Sham Hamiltonian are expressed as linear combinations of numerical
atomic orbitals. A 200 Ryd mesh cut-off is chosen and the self-consistent calcu-
lations are performed with a mixing rate of 0.1. The Brillouin zone is sampled with
a Monkhorst-Pack mesh [27] with (5 × 5 × 1) k-points, whereas we use
(45 × 45 × 1) k-points in specific systems. Core electrons are replaced by
norm-conserving, nonlocal Truoiller-Martins pseudopotentials [28]. The vacuum
spacing between graphene layers in adjacent supercells is taken 15 Å. The con-
vergence criterion for the density matrix is taken as 10−4.

11.3 Theoretical Investigations of Charged Nanosystems

The theoretical investigations of charged nanosystems have been an active field of

study. Developing appropriate formalisms to provide reliable predictions on the
effects of charging and applied electric field have been the subject matter of several
studies [26, 29–43].
The aim of this study is to investigate the effects of electric field and charging on
OH, H and O which are adsorbed to graphene. Here, we performed the calculations
of charged GOX by adding desired amount of excess electrons for the case of
negative charging or by removing electrons for the case of positive charging, where
both cases are treated using periodic boundary conditions (PBC). Throughout the
study, Q [ 0 (or surface charge density r
¼ Q=A in Coulomb/m2 , A being the area
of the cell) indicates the positive charging, namely number of depleted electrons per
cell; Q \ 0 indicates the negative charging, namely the number of excess electrons
per cell, and Q ¼ 0 is the neutral cell. The bare (super)cell is made up of ðn  nÞ
primitive unit cells of graphene; each supercell comprises 2n2 carbon atoms and
8 n2 valence electrons. We assume that graphene planes, which are repeating
periodically along the z-axis are parallel to the (x, y)-plane. The electric field E,
which is applied perpendicularly to the graphene plane, is specified as positive i.e.
Efield [ 0, if it is along z-direction (or it is pointing towards adsorbates, i.e. H, OH,
O). This electric field induces electronic charge transfer from the adsorbate to
graphene and the case is vice versa if the direction of E is reversed, i.e. Efield \ 0.
Electric field induced charge transfer modifies the charge distribution and hence
affects the physical and chemical properties.
We first examine the limitations of PBC method, where two-dimensional gra-
phene layers separated by large spacing s are repeated periodically along z-direc-
tion. We carried out first-principles PW calculations as well as AO calculations
using linear combinations of numerical atomic orbitals and investigated the effect of
charging on the graphene layer.
In Fig. 11.1a–c we present electronic band structures of graphene calculated by
PW and AO for Q = +0.2 (
r = +0.61 C/m2 ), 0, −0.2 electrons/cell (

r = −0.61 C/m2 )
using the primitive cell (n = 1). Further details about the changes in the band
264 H. Hakan Gürel et al.

PW (plane-wave) AO (atomic-orbital)
+0.2 e/cell
(a)
5 5
0 0
EF

-5 -5

neutral
(b)
Energy (eV)

5 5
0 0
EF

-5 -5

-0.2 e/cell
(c)
5 5
0 0
EF

-5 -5

Γ Μ Κ Γ Γ Μ Κ Γ
(d) e-
ΦΒ Vacuum
EF

w
V(z)

|
s ∞
Graphene
s
x 10-3
(e) 5
0.6
0.5
λ (e/A)

4 o 0.4
0.3
λ (e/A)

o 3 | 0.2
0.1
0 2 4 6 8 10 12 14 16 18 20
| 2 o
z (A)
1
8.09 e 0.11 e
0 2 4 6 8 10 12 14 16 18 20
o
z (A)
 11 Effects of Charging and Perpendicular Electric Field … 265

b Fig. 11.1 a–c Energy band structures of graphene calculated for Q = +0.2 e/per primitive unit cell
r ¼ þ 0:61 C/m2 ), 0 and −0.2 e/per primitive unit cell (

(
r ¼ 0:61 C/m2 ). Left (right) panels are
calculated using PW basis set using VASP code (AO basis set using SIESTA code). d Schematic
description of the plane-averaged potential of negatively charged graphene, VðzÞ
dipping at the
center of vacuum spacing and forming a quantum well like structure. The vacuum spacing between
two adjacent graphenes is denoted by s. The triangular potential barrier and its width are UB and
w, respectively. The quantum well confining electrons, which are spilled from charged graphene of
PW calculations is illustrated by blue lines; the well without bound electrons shown by red lines
corresponds to AO calculations. For one single graphene layer corresponding to the “actual case”,

! 1Þ ! 1 for s ! 1. e Plane-averaged (linear) charge density,

Vðz kðzÞ obtained from
PW calculations for Q = −0.2 e/cell (
r ¼ 0:61 C/m2 ) and s = 20 Å.
kðzÞ in full y-axis range is
shown by inset. Here “cell” indicates the primitive unit cell, and s = 20 Å. This figure is taken from
[29]

structure of graphene upon charging have been discussed in our recent studies
[42, 43]. In the case of Q = +0.2 (
r = +0.61 C/m2 ), the Fermi level shifts down or
the Dirac point raises for both PW and AO calculations. Accordingly, semimetallic
graphene turns into a hole-doped metallic system.
For neutral case (Q = 0), band structures calculated using PW and AO are
similar. The work function is the difference of the reference vacuum energy and the
Fermi level. The value of work function extracted from Fig. 11.1b is 4.26 eV and
this is in good agreement with earlier study [44]. Briefly, the results calculated
using PW and AO are similar. Minute differences originate from different pseu-
dopotentials used in calculations. Moreover, the perfect convergences of different
basis sets can be achieved only by using very large cutoff values.
As for the case of negative charging with Q = −0.2 e/cell (
r ¼ 0:61 C/m2 ), the
Fermi level shifts up significantly in AO results. In the case of PW results the Fermi
level shifts up only slightly and then is pinned by the free electron-like parabolic
bands touching to the Fermi level; thereafter the position of the Fermi level is
practically unaltered with increasing negative charging as shown in Fig. 11.1c. The
dramatic difference for the case of Q \ 0 between PW and AO results can be
understood through the plane-averaged electronic potential along the z-direction,


i.e. VðzÞ

as described schematically in Fig. 11.1d. In the actual case, VðzÞ passes
through a maximum near the surface and goes to 1 as z ! 1 at both sides of
one single graphene layer, which is charged with Q \ 0. Then the electrons in the
graphene can spill to vacuum only if they tunnel through tunneling across a wide
triangular barrier. The width of the barrier, w decreases with increasing negative
charging. Accordingly, the tunneling current would be negligibly low and excess
electrons would practically be trapped in graphene if the value of jQj is not very
high. However, when treated within PBC, VðzÞ
of periodically repeating graphene
layers with a vacuum spacing s between them can make a dip reminiscent of a
quantum well at the center of the spacing as shown in Fig. 11.1d. Under these
circumstances, Kohn-Sham Hamiltonian using PW can acquire solutions in this
quantum well, which are localized along the z-direction, but free-electron like in the
(x, y)-plane parallel to graphene. These states form the parabolic bands (kx ; ky ) plane
266 H. Hakan Gürel et al.

as shown in Fig. 11.1c. If the quantum well dips below the Fermi level with
increasing s or Q, excess electrons in graphene start to be accommodated in these
2D free-electron like bands. This ends up with the spilling of electrons into the
vacuum region.
In Fig. 11.1e we present the plane-averaged charge density,
kðzÞ calculated
within PW. By integrating
kðzÞ along the z-direction, it is concluded that graphene
can keep only −0.09 e/cell (
r ¼ 0:28 C/m2 ) out of −0.2 e/cell (
r ¼ 0:61 C/m2 )
excess electrons. The remaining −0.11 e/cell (
r ¼ 0:33 C/m ) are spilled out into
2

the vacuum, namely to the region between two adjacent graphene planes. Clearly,
the spilling of excess electron, which can be normally trapped in graphene, is an
artifact of PBC. The amount of excess electrons spilled to the vacuum spacing
increases with increasing s and increasing negative charging. While wide s is not
convenient because of excessive charge spilling, small s may prevent excess
electrons from spilling into the vacuum at very low negative charging and can yield
reasonable solutions as long as the coupling of the adjacent layers is negligible.
However, the charge spilling can set in when the negative charging exceeds a
threshold value corresponding to the given s. The dependency of electronic
structure and hence the band gap opening on s in PW calculations of silicene
(graphene like Si) under perpendicular electric field has been also reported recently
[45]. The perpendicular electric field breaks the mirror symmetry of VðzÞ
and gives
rise to a sawtooth like variation. At high E and large s, electrons of silicene spilling
to low potential site result in a situation discussed above and in [30] above for
Q \ 0.
The following conclusions have been arrived regarding the features and limi-
tations of first-principles methods in treating these external effects using plane wave
(PW) and local basis (AO) sets: (i) When negatively charged, the electronic
potential between periodically repeating graphene layers makes a dip at the center
of the vacuum spacing s. This dip forms a quantum well like structure and its depth
from the Fermi level increases with s and with excess charge Q \ 0. (ii) PW cal-
culations result in charge spilling to the vacuum region for large s, which is the
consequence of the artifact of PBC. (iii) On the other hand, such a spilling does not
occur in AO calculations even if PBC is used, since local basis orbitals at carbon
sites fail to represent states in the quantum well like potential at the middle of the
vacuum spacing. (iv) This is the artifact of AO calculations, which turned out to be
its advantage, whereby the artifact of PBC is tacitly canceled out. (v) The electronic
potential of one single graphene layer trapping the excess charge in the actual case
is close to that obtained by AO calculations using PBC. Hence the results con-
cerning charging obtained from AO calculations are expected to be very close to the
results of the actual case and are physical. (vi) This analysis can also be extended to
graphene exerted by an perpendicular electric field E. While the excess electrons
can spill to the vacuum region at the lower energy side of the saw tooth like
electronic potential treated by PW and by using PBC, AO calculations provide
predictions close to the actual case of single graphene under the perpendicular E.
11 Effects of Charging and Perpendicular Electric Field … 267

According to the above analysis concerning the artifact of PW basis set using
PBC, our results in this study are obtained by performing first-principles spin
polarized calculations within Density Functional Theory (DFT) [46] using linear
combination of numerical atomic orbitals (LCNAO). Therefore, present results are
expected to be very close to the results of the actual case corresponding to single
graphene layer under E or containing excess Q \ 0.

11.4 Interaction of H2O, OH, O and H with Graphene

The interactions of H2O, OH, O and H with graphene are essential for the reduction
of GOX through desorption of H2O, O2 and H2O2. Here we characterize their
interactions in equilibrium conditions by calculating the optimized binding geom-
etry, the corresponding binding energy, the minimum energy barrier in their dif-
fusion and the path of diffusion with least energy barrier.

11.4.1 Binding of H2O to Graphene

H2O, an indispensable ingredient in the oxidation/reduction process of GOX, has

chemical as well as van der Waals (vdW) interaction with graphene [47]. Since the
generalized gradient approximation (GGA) does not include vdW interaction, it
underestimates the binding energy between H2O and graphene. The binding energy
calculated by using GGA lies in the range of 18–47 meV depending on its orien-
tation and position [48]. On the other hand, the local density approximation
(LDA) overestimates the binding energy to be 151 meV. Therefore, the binding
energy of H2O including vdW interaction is estimated to be between 18 and
151 meV. It is really a weak interaction and a water molecule physisorbed to
graphene can be desorbed near the room temperature.

11.4.2 Binding of OH to Graphene

It is demonstrated experimentally that OH is one of the critical functional groups

existing in GOX [49, 50]. The binding energies and magnetic moments of OH
adsorbed to graphene are calculated by AO for hollow (H), top (T) and bridge
(B) sites on graphene layer. As shown in Fig. 11.2a, the top (T) site is most favorable
energetically with a binding energy Eb ¼ 0:97 eV and the magnetic moment l ¼
0:51 lB . Carbon atoms underlying OH are displaced slightly upwards from the plane
of graphene. Earlier calculations found the binding energy of OH in the energy range
between 0.54 and 0.86 eV [51–56]. The energy landscaping shown in Fig. 11.2b is
calculated by relaxing all the position of carbon atoms, as well as the height z of
268 H. Hakan Gürel et al.

(a)

0 0
2.94 A 1.49 A 2.49 A0
E T = 0.49 eV ET = 0 E T = 0.51 eV
μ = 0.55 μ B μ = 0.51 μ B μ = 0.60 μ B

(b) (c)
H B 1 H
0.5
B
Energy (eV)

0.51 eV
0.49 eV

0.96 eV

0.74 eV
0.25 0.5
B H

0.5 0.4 0.3 0.2 0.1 0.0

0 0
T T T T

Fig. 11.2 Interaction and binding between OH/H and graphene surface. a Top and side views of
atomic configurations, total energy ET and magnetic moments l in Bohr magneton lB , of OH
adsorbed to hollow (H), top (T) and bridge (B) sites on graphene. The zero of energy is set to ET at
the top site. Large brown, large red and small yellow balls represent carbon, oxygen and hydrogen
atoms, respectively. b The energy landscaping of OH adsorbed to different sites in the honeycomb
structure and the variation of energy of OH migrating along symmetry sites on a hexagon, i.e. T →
H → B → T. The minimum energy barrier between T and H-sites is EB ¼ 0:49 eV. A possible
path of minimum energy barrier for the diffusion of adsorbed OH is shown by stars. c The
variation of energy of H adatom migrating along the symmetry sites on a hexagon, i.e. T → H →
B → T. The minimum energy barrier occurs between T and B-sites is EB ¼ 0:74 eV. Accordingly,
H adatom migrates above the C–C bonds. Equilibrium binding energies of OH and H occur at the
top site as 0.97 and 0.76 eV, respectively. Calculations are performed using a system, where a
single OH or H is adsorbed to each (4 × 4) supercell of graphene. This figure is taken from [29]

single adsorbed OH, while its x and y coordinates are fixed in the (4 × 4) supercell of
graphene. These calculations are repeated for 36 × 36 (x, y)-grid points on a hexagon.
One can deduce the minimum energy barrier to the diffusion to be EB ¼ 0:49 eV
from the variation of the calculated total energies along the symmetry directions,
T → H → B → T of the hexagon presented in Fig. 11.2b. This comparatively low
energy barrier allows easy migration of OH on graphene at elevated temperatures.
We expect that EB calculated at low coverage is modified at very high coverage due
to increased hydroxyl-hydroxyl interaction [15, 57].
11 Effects of Charging and Perpendicular Electric Field … 269

11.4.3 Binding of O to Graphene

When neutral, single oxygen is adsorbed at the bridge site of graphene, namely
above the center of the C–C bonds of honeycomb structure. Calculations yield that
the bridge (B-)site is energetically the most favorable adsorption site; the top (T-)
sites (on top of carbon atoms) are not favorable. PW calculations [15] predicted the
binding energies of a single oxygen at the B-site as 2.35, 2.40, 2.43 and 2.43 eV
using (2 × 2), (3 × 3), (4 × 4) and (5 × 5) supercells of graphene, respectively. The
binding energy is practically unaltered for supercells larger than (5 × 5) [15].
Using AO we found that the binding energy of oxygen on (5 × 5) graphene
supercell is 2.34 eV.
The formation energy related with the oxidation of graphene is negative and
hence graphene cannot be oxidized through oxygen molecules [15, 58]. Actually,
O2 is physisorbed to bare graphene surface with a very weak binding energy of
58 meV calculated without van der Waals correction; but it raises to 115 meV when
the van der Waals correction [59] is included. Only at defect sites like holes and
vacancies, O2 can dissociate [60] and its constituent oxygen atoms become
adsorbed to carbon atoms having a lower coordination numbers [58]. In contrast to
O2, the bonding of free oxygen to graphene is rather strong and changes between
2.43 and 3.20 eV depending on the coverage [15]. The crucial question to be
addressed is how oxygen atoms can remain strongly bound to graphene despite
their negative formation energy and why bound oxygen atoms are prevented from
desorption through the formation of oxygen molecule. Recent studies have pro-
vided the energy barrier for the diffusion of adsorbed O on graphene and for the
desorption of O through the formation of O2 and CO2 [15, 16].

11.4.4 Binding of H and H2 to Graphene

The binding of atomic hydrogen H on graphene has been treated in earlier studies
[61–63]. Binding energies reported by different authors vary in a wide range of
energy [64]. They lie between 0.47 and 1.44 eV with a majority of data being
between 0.6 and 0.85 eV. This variability can be ascribed to differences in struc-
tures, optimization procedure, and computational methodology used to calculate the
chemisorption of H on graphite [64]. Here, for the sake of completeness, we cal-
culate the binding energy using the same calculation parameters and local basis set
used throughout the present work. In agreement with previous studies, we found
that the strongest binding of H atom occurs at the top site with a binding energy of
Eb ¼ 0:76 eV. The variation of the total energies of H adatom moving along
specific directions of the honeycomb structure are also calculated and the minimum
energy barrier to the diffusion is found to be EB ¼ 0:74 eV as shown in Fig. 11.2c.
Similar to water molecule, the binding of hydrogen molecule H2, to graphene is
weak. Calculations with the LDA VWN [65] functional result in a binding energy
270 H. Hakan Gürel et al.

of 93 meV for molecular hydrogen. Due to the lack of vdW interaction, GGA
calculations using PW91 and PBE functional predict relatively lower binding
energies of 23 and 13 meV, respectively [66].

11.5 Effects of an Electric Field and Charging

11.5.1 Effects of an Electric Field and Charging

on Adsorbed O

The effects of applied electric field and charging are interrelated. Upon the
adsorption of oxygen atom, 0.78 electrons are transferred from carbon atoms to
oxygen in neutral case. The electric field applied perpendicularly to the graphene E,
which is specified as positive if it is along z-direction (or it is pointing towards
oxygen adatom). This electric field induces electronic charge transfer from the
adsorbed oxygen to graphene or vice versa if its direction is reversed. Electric field
induced charge transfer modifies the charge distribution and hence affects the
physical and chemical properties. Results obtained from this section will enlighten
recent experimental studies performed for similar systems [20, 21].
Interesting effects of electric field on oxygen adsorbed graphene are summarized in
Fig. 11.3. The binding energy of the oxygen adatom increases with applied negative
electric field, which is perpendicular to the surface and pointing the direction opposite
to oxygen. The height h of oxygen adatom from the graphene plane also increases,
even if one expects the opposite trend. This paradoxical situation originates from the
definition of the binding energy, Eb , of oxygen adsorbed to graphene, which is given
by Eb ¼ ET ½O þ ET ½graphene; E  ET ½O þ graphene; E, in terms of the total
energies of single neutral oxygen atom ET[O], of bare graphene under E,
ET ½graphene; E and single oxygen adsorbed to graphene under applied E,
ET ½O þ graphene; E all calculated using (5 × 5) supercell. According to this defini-
tion, increasing Eb does not mean that the bond between oxygen and graphene has
become stronger. In fact, that h increases with increasing negative E implies the
opposite situation.
Before we explain this paradoxical situation, we first consider charge rear-
rangements caused by E. Induced charge transfer depends on the direction of E. In
Fig. 11.4 we present the effects of applied perpendicular electric field on the charge
distribution and potential energy VðzÞ.
An electric field E = −1.0 V/Å transfers
electrons from the bottom side to the upper side of the graphene and also causes to
further transfer of carbon electrons to oxygen adatom as shown in Fig. 11.4b. At the
end, the total energy of whole system increases. This effect creates an interesting
situation also for the bare graphene by breaking the projection symmetry between
its two sides. The direction of electron transfer is reversed when the direction of the
applied electric field is reversed. Accordingly, one can monitor various properties of
bare graphene, in particular its chemical activity by applying perpendicular electric
11 Effects of Charging and Perpendicular Electric Field … 271

z h
y
x
1.75
2.6
1.74

Oxygen height, h (A)

2.5 1.73
Eb (eV)

2.4 1.72

1.71
2.3
1.70
2.2
1.69
−1 −0.5 0 0.5 1 1.5 2
E−field (V/A)

Fig. 11.3 Variation of the binding energy Eb , of oxygen adatom, and height h, from the graphene
plane with the electric field E, applied perpendicular to the graphene plane. E is taken positive, if it
is along z-direction (or it is pointing towards oxygen adatom) and vise versa. (Results are obtained
from AO calculations using PBC.) This figure is taken from [29]

field and hence by modifying electron concentration at both sides as seen in

Fig. 11.4c.
The self-consistent potential energy averaged on the planes perpendicular to


z-axis, VðzÞ is presented in Fig. 11.4d. Within PBC VðzÞ
displays a saw teeth
behavior. The form of the potential suggests interesting situations for both sides of
graphene. For example, one can monitor the work function on both sides. Side
specific effects induced by perpendicular electric field are pronouced in single layer
honeycomb structures consisting of two or three atomic planes, such as silicene
[45, 67, 68], single layer transition metal dichalcogenides [69] and graphene
bilayer.
The effect of applied perpendicular electric field on the calculated total energies
of oxygen adsorbed at different sites on the NEB path between two adjacent B-sites
through the T-site is presented in Fig. 11.5. The difference of total energy DE,
between oxygen adsorbed at the T-site and B-site is as large as *1.2 eV for E =
+2.0 V/Å. However, when the direction of E is reversed, DE decreases and
becomes even negative for E = −2.0 V/Å. This is a dramatic effect and is expected
to bear on the reduction/oxidation of graphene surfaces. Since the energy barrier
between the B-site and T-site can be modified by the applied electric field, the
monitoring of oxygen diffusion at the graphene surface will be possible through
applied electric field. We note that the magnitudes of the electric fields of 1–2 V/Å
are high but they are in the range, which can be applied at least for short times [70].
We now resolve the above paradoxical situation, namely that the equilibrium
oxygen graphene distance h increases with Eb increases as shown in Fig. 11.3 and
272 H. Hakan Gürel et al.

(a)

z
y x

(b)

-ρ

+ρ Ε −field
(c)
Ε −field

(d)
10

0
V(z) (eV)

| -10

E = -1.0 V/A
-20 E = 0.0 V/A
E = +1.0 V/A

10 20 30
o
z (A)

Fig. 11.4 a Atomic configuration showing a single oxygen atom adsorbed at the B-site of a
(5 × 5) graphene supercell with red and brown balls indicating oxygen adatom and carbon atoms
of graphene, respectively. The (xy)-plane coincides with the atomic plane of bare graphene. z-
direction is perpendicular to graphene. b and c The difference charge density Dq for the electric
field antiparallele (E\0) and parallel (E [ 0) to z-direction, respectively. Yellow (blue)
isosurfaces indicate electron accumulation (depletion) induced by E applied in different directions.
d Self-consistent field potential energy averaged on the planes perpendicular to z-axis throughout


the supercell, VðzÞ, for the cases E ¼ 0, E\0, and E [ 0. (Results are obtained from AO
calculations using PBC.) This figure is taken from [29]

develop a criterion for the strength of bond between graphene and oxygen adatom.
To this end, we consider the system consisting of a (5 × 5) graphene supercells with
a single oxygen adatom, which are either charged by Q or under a perpendicular
electric field, E as shown in Fig. 11.6a. We calculated the optimized total energies
of this system while oxygen adatom is pulled along z-direction perpendicular to the
11 Effects of Charging and Perpendicular Electric Field … 273

Ε −field
1.2 +2.0
+1.5
+1.0
1.0 +0.5
0.0
−0.5
−1.0
0.8 −1.5
−2.0
−2.5
0.6
ET (eV)

0.4

0.2

-0.2

-0.4
(bridge) (top) (bridge)
Position

Fig. 11.5 Variation of the total energy ET , of single oxygen atom adsorbed to each (5 × 5)
graphene supercell under applied perpendicular electric fields E on a NEB path between two
adjacent B-sites through a T-site. The total energy at the B-site is set to zero for all cases. The unit
of E is V/Å. (Results are obtained from AO calculations using PBC.) This figure is taken from [29]

plane of graphene. For each fixed value Dd of oxygen adatom from its equilibrium
height h, carbon atoms around oxygen adatom are relaxed, while the rest of the
atoms are fixed to prevent graphene from displacement. These analyses are con-
tinued by varying Dd for different values of Q [e per (5 × 5) cell] and for different
values of E perpendicular to the graphene plane.
The variation of the total energies with pulling Dd are plotted in Fig. 11.6b, c for
different values of Q and E, respectively. For the sake of comparison, we also
included the pulling curves for Q = 0 and E = 0. These figures convey interesting
features regarding the effects of either charging or applied perpendicular electric field
on the strength of the bond between graphene and oxygen adatom: Normally, the
energy associated with pulling Ep ¼ ET ½O þ graphene; Q; E; Dd 
ET ½O þ graphene; Q; E; Dd ¼ 0 increases with increasing Dd, since the system is
strained and pulled upwards. Eventually it passes through a maximum value denoted
by Ep
and drops suddenly at about 0:8\Dd\0:9 Å. Our analysis suggests that Ep

can be taken as a measure for the strength of the bond between oxygen adatom and
graphene. We note that Ep
, namely the energy barrier to pull out the adsorbed
oxygen adatom from graphene surface is *2.60 eV for both Q = 0 and E = 0. This
energy consistent with the binding energy calculated for the bridge site. An inter-
esting feature of the present analysis is that Ep
is strongly dependent on charging and
electric field. While Ep
increase with Q [ 0, it decreases dramatically for Q\0. For
example, Ep
¼ 1:46 eV for Q = −1.0 e/cell (or −0.02 e per carbon atom or r
¼
274 H. Hakan Gürel et al.

(a)

pulling of oxygen

(b) 3
Q = 1e removed
2.5 Q = neutral
2 Q = 1e added
Ep (eV)

Q = 2e added
1.5
1
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Δd (A)
(c) 3 E = −2.0 (V/A)
E = −1.0 (V/A)
Ep (eV)

2.5
E = 0.0 (V/A)
2 E = +1.0 (V/A)
Ep (eV)

E = +3.0 (V/A)
1.5 Δd (A)
1
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Δd (A)
(d) -5
-4.5
-4
E−field (V/A)

-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
0 -0.5 -1 -1.5 -2 -2.5 -3 -3.5 -4
Charging , Q (e/cell)
 11 Effects of Charging and Perpendicular Electric Field … 275

b Fig. 11.6 Variation of the pulling energy Ep , of a single oxygen adatom adsorbed to each (5 × 5)
supercell of graphene charged by Q or exerted by E with pulling distance along the z-axis, Dd from
its equilibrium height h. a Atomic configuration. Red (large) and brown (small) balls indicate
oxygen and carbon atoms, respectively. b Variation of Ep with Dd, along z-axis for different charging
values Q. c Variation of Ep with Dd, along z-axis for different applied electric field E. All cases have
their own equilibrium heights h. The maximum value of the pulling energy for each case is specified
by Ep
. As shown by inset for the case of Q = 0 and E = 0, a bistability occurs when oxygen atom
approaches to graphene from a long distance. d Positions of adsorbed oxygen atom under different
values of electric filed E and/or charging Q. Green, red and blue dots indicate respectively, oxygen
atom adsorbed to the bridge-site, top-site and desorbed by moving away from graphene plane.
(Results are obtained from AO calculations using PBC.) This figure is taken from [29]

0:12 C/m2 ). Similarly, Ep
can be as low as *1.14 eV under the perpendicular
electric field E = −2 V/Å. As shown by inset in Fig. 11.6c, a bistability occurs if
oxygen atom moves in the reverse direction and hence approaches to the graphene
from Dd [ 1:0 Å. Notably, for Q \ 0 Ep curve doesn’t experience a sharp fall
passing its maximum value. This is related with the excess charge on oxygen atom.
These are important results and demonstrate that it is easier to desorb oxygen ada-
toms from graphene by negatively charging or by applying perpendicular electric
field. Under high local charging and local electric field, which can be attained by the
sharp tip of a Scanning Tunneling Microscope or by a gate voltage the reduction of
GOX through the desorption of O adatoms can be achieved.
Since the negatively charging or negative perpendicular electric field both
weaken the bond between oxygen and graphene and hence lower Ep
, we next
explore their effects on the bond between graphene and oxygen adatom when they
both coexist. Thus the negative E of different magnitudes exerts on a negatively
charged system consisting of single oxygen adatom adsorbed to each (5 × 5)
supercell of graphene. We found that the effects same as in Fig. 11.6b, c are attained
by applying relatively lower electric field when the system is negatively charged. In
particular, adsorbed oxygen moves high above graphene and becomes weakly
bound under the electric field E = −2.5 V/Å, if the whole system is charged by Q =

¼ 0:12 C/m2 ). We note that this value is much smaller than the
−1.0 e/cell (or r
electric field required for desorption of oxygen from a neutral system. In Fig. 11.6d
we schematically show different positions of oxygen adatom (the bridge-, top-site
and moved away from graphene for desorption) occurring under different values of
negative E and negative Q.

11.5.2 Effects of an Electric Field and Charging

on Adsorbed OH

The effects of perpendicular electric field E and charging Q on OH adsorbed to

graphene occur through the modification of the equilibrium charge distribution. In
Fig. 11.7 we present atomic geometry of adsorption, the isosurfaces of difference
276 H. Hakan Gürel et al.

(a)

(b)
-Δρ

+Δρ Ε field < 0 Ε field > 0

(c)
Added 1e/cell Removed 1e/cell

Fig. 11.7 a Equilibrium atomic configuration of one OH adsorbed to the top site of the (4 × 4)
supercell of graphene. b Isosurfaces of the difference charge density Dq under perpendicular
electric field Efield ¼ 1:0 V/Å. Negative and positive Dq are shown by yellow and turquoise
isosurfaces, respectively. c Isosurfaces of the difference charge density for Q ¼  1.0 e/supercell.
This figure is taken from [31]

charge density, Dq ¼ q½EorQ  q½E ¼ 0; Q ¼ 0. For Efield \ 0, electronic charge

is transferred from the bottom site of graphene towards OH and upper site leading to
accumulation of more charge on OH. The charge transfer is reversed when the
direction of the perpendicular electric field is reversed i.e. Efield [ 0. The negative
charging Q \ 0 realized by adding electrons in the system gives rise to electron
accumulation at OH. This situation is, however, reversed for positive charging, Q [ 0
realized by removing electrons from the OH+graphene system. Clearly, the interac-
tion of OH with graphene, hence the strength of the bond is modified depending on the
magnitude and direction of E, as well as the sign and magnitude of charging.
Here we further investigate the strength of the bond by calculating the optimized
total energy of OH+graphene system as OH is pulled in a perpendicular direction
(z-direction) gradually for different values of Q [e per (4 × 4) cell] and for different
values of Efield . For each fixed value Dz of OH from its equilibrium height, all
carbon atoms in the supercell are relaxed, while carbon atoms at the corners of
supercell are fixed to prevent graphene from displacement. This analysis is con-
tinued by varying Dz. The variation of the total energies with pulling Dz are plotted
in Fig. 11.8b, c for different values of Q and Efield , respectively. For the sake of
comparison, we also included the pulling curves for Q = 0 and E = 0. These figures
11 Effects of Charging and Perpendicular Electric Field … 277

(a)
z
pulling of OH

(b) 2
Q = neutral
1.6 Q = 1e added
Q = 1e removed
Ep (eV)

1.2

0.8

0.4

0
0 0.2 0.4 0.6 0.8 1 1.2
Δz (A0 )
(c) 2
E=0
0
E = -1 V/A
0
1.5 E = 1 V/A
0
E = -0.5 V/A
0
E = 0.5 V/A

1
Ep (eV)

0.5

-0.5
0 0.2 0.4 0.6 0.8 1 1.2
Δz (A ) 0

Fig. 11.8 a Atomic configuration consisting of one OH adsorbed to each (4 × 4) supercell of

graphene, where OH is pulled in the perpendicular direction by the distance Dz from its
equilibrium height on graphene surface. Large brown, large red and small yellow balls stand for
C, O and H atoms, respectively. b The total energy Ep , versus the distance Dz for different values
of Q. c Ep , versus the pulling distance Dz for different values of Efield . The zero of energy is set to
the total energies corresponding to Dz ¼ 0. This figure is taken from [31]
278 H. Hakan Gürel et al.

clarify the effects of either Q or E on the strength of the OH-graphene bond as

discussed in the previous sections: Normally, the energy associated with pulling
Ep ¼ ET ½OH þ graphene; Q; E; Dz  ET ½OH þ graphene; Q; E; Dz ¼ 0 increases
with increasing Dz, since the system is strained and pulled upwards. Eventually it
passes through a maximum value denoted by Ep
and drops suddenly at about
0:5\Dz\1:0 Å. Our analysis suggests that Ep
can be taken as a measure for the
strength of the bond between OH and graphene. We note that Ep
, namely the energy
barrier to pull out the adsorbed OH from graphene surface is *1.3 eV for both Q =
0 and E = 0. This energy is 0.33 eV larger than the equilibrium binding energy Eb
of OH, since it corresponds to the strained configuration of underlying graphene,
where carbon atoms at the corners of the supercell are fixed. An interesting feature
of the present analysis is that Ep
is strongly dependent on charging and electric
field. While Ep
increases with Q [ 0, it decreases dramatically for Q \ 0. For
example, Ep
 1:8 eV for Q = +1.0 e/cell, but it decreases to Ep
 0:9 eV for Q =
−1.0 e/cell. Notably, for Q \ 0 Ep curve doesn’t experience a sharp fall passing its
maximum value. This is related with the excess charge on OH. Similarly, Ep
can be
as low as *0.35 eV under the perpendicular electric field Efield ¼ 1:0 V/Å. We
note that a bistability [29] may occur if OH moves in the reverse direction and
hence approaches to the graphene from Dz [ 1:2 Å. These results also confirm that
it is easier to desorb OH and to achieve the reduction of GOX by negatively
charging or by applying Efield \ 0. Under high local charging and local electric
field, which can be attained by the sharp tip of a Scanning Tunneling Microscope or
by a gate voltage the reduction of GOX can be achieved easily.

11.6 Desorption of Oxygen from GOX

In the above sections we discussed interaction of single O, H, H2, H2O and OH with
graphene surface and also revealed how the binding and related properties of O and
OH are affected with applied E and Q. The interaction of oxygen atom with
graphene surface was investigated thoroughly in earlier studies [15, 29, 58]. In this
section we investigate the binary interactions among H, O and OH. Our objective is
to reveal whether oxygen atoms can desorb from GOX via hydroxyl groups and
how the reduction process is affected by Q and E.

11.6.1 Formation of Oxygen Molecule

Having examined the binding energy, binding site and diffusion of single oxygen
adatom and their variation with charging and applied electric field, we next consider
the formation of O2, which can be essential in deoxidation. Since single oxygen
11 Effects of Charging and Perpendicular Electric Field … 279

adatom adsorbed to graphene has negative formation energy, one cannot expect that
O2 molecule dissociates into two atomic oxygens, both adsorbed to graphene.
However, O2 molecules can dissociate at the edges of vacancy defects or holes and
subsequently the constituent atomic oxygens are adsorbed to carbon atoms with
lower coordination [58]. In view of the negative formation energy, one normally
expects that two oxygen atoms adsorbed individually to graphene can readily form
O2 in an exothermic process when two atoms are at close proximity. Therefore, the
interaction of two oxygen adatom on graphene is essential for the formation of O2.
In Fig. 11.9a we examined O–O interaction for Q \ 0 and E = 0 while one O is at
the position identified as “p” and marked by the arrow as shown by inset, the other
one is diffusing from a distant bridge-site identified as “a”. Initially, the interaction
between them is weak, but develops as one O atom is migrating from a towards
p through b, c and d positions. Here a and c correspond to the bridge- and b and d to
the top-sites. Here we consider three situation, which are illustrated by three curves
denoted by I, II and III. For the curve I, the (x, y)-coordinates of all carbon atoms of
graphene are kept fixed, but their z-coordinates are relaxed as the first O adatom
migrates from a to the second O at p. While the first O is forced to migrate through
the path of the minimum energy barrier by optimizing its height from graphene, the
second O is fully relaxed. Overcoming the barrier of *3.3 eV corresponding to
configuration K, O2 is desorbed at G. Curve I is similar to the curve in Fig. 11.5c of
[15] calculated using PW method. For the curve II, only one carbon atom out of 72
in the supercell is fixed to prevent underlying graphene from displacing in the
course of forced migration of the first O adatom. A relatively smaller barrier of
*2.3 eV develops between d and H, and eventually one O adatom is desorbed once
H is overcame at L. Apparently curve I and curve II are associated with high energy
barrier for the desorption of O2 or single oxygen atom from GOX. A different
migration path is followed for the curve III: Once the first O adatom arrived at d, it
is fixed there while the second O adatom at p is forced to the top site at e. To attain
the final configuration J, an energy barrier of 1.3 eV from c has to be overcame.
After J, two O adatoms form O2 and desorb from graphene. Clearly, this path
described by curve III has much lower barrier than those F and H. The energy
barriers of curve I–III are lowered with negative charging.
Larciprete et al. [16] pointed out a dual path described by inset A in
Fig. 11.9b, which allows two O adatoms at the bridge sites to move along C–C
bonds towards to adjacent top sites. They calculated the energy barrier to be
1.13 eV using PW method and found that it is in good agreement with tem-
perature programmed desorption (TPD) data. They observed significant O2 des-
orption at *500 K during the thermal annealing of low oxygen density GOX.
Here, we first examine the low-energy-barrier dual path proposed by Larciprete
et al. [16] and calculate the energy barrier for Q \ 0 and E = 0 using AO method
by relaxing all atoms in the (5 × 5) supercell except one carbon atom to hinder
the displacement of graphene layer. In Fig. 11.9b we present the energy variation
of the dual path corresponding to curve I through A, B, C and D configurations
shown by inset. The energy barrier at B configuration is rather low as compared
to F and H configurations in Fig. 11.9a. Overcoming an energy barrier of 0.8 eV,
280 H. Hakan Gürel et al.

two O adatoms become attached to two adjacent top sites of graphene described
as the configuration C. The energy barrier calculated by AO is *0.3 eV lower
than predicted by Larciprete et al. [16] perhaps due to the different parameters of
calculations and different basis set. If one of these O atoms is forced towards the
other one an energy barrier of 0.4 eV develops between C and D. The formation
of O2 and desorption is achieved at E when D is overcame. A different path
specified as curve II in Fig. 11.9b goes to F from C, where one of O adatoms of
the configuration C is raised. This way, the formation of O2 attained without an
energy barrier after the configuration C. The atomic configurations corresponding
to various stages are described by insets.
Having discussed two paths with low energy barrier leading to the formation of
O2 from GOX, we next examine how these barriers are affected by external agents
such as negative charging and applied electric field. When charged by Q = −4 e per
r ¼ 0:332 C/m2 ), the energy variation of curve I and curve II is mod-
supercell (

ified as presented by curve III in Fig. 11.9b. Two oxygen adatoms migrating from the
configuration A towards the configuration C encounter a small energy barrier of
*0.1 eV. This is really small barrier, which can be further lowered or completely
suppressed with increasing excess electrons. The application of the electric field of
E = −2 V/Å perpendicular to the plane of graphene induces a similar effect shown by
curve IV: While the barrier of 0.8 eV at B is suppressed, a small barrier of *0.2 eV
near C appears. The energy barrier in curve III and curve IV are rather small and
explain why GOX can easily be deoxidized under external effects [20, 21].

11.6.2 Interaction Between Adsorbed H and O

Formation of OH from adsorbed O and H is another process contrubiting to deoxi-

dation of GOX. Here we examine the interaction between coadsorbed H and O atoms
to see how OH can form. We consider three different paths for H atom approaching the
adsorbed O atom. (i) Both H and O are initially coadsorbed; O is adsorbed at the bridge
site and initially H is adsorbed at the top site at the close proximity of O adatom as
shown in Fig. 11.10a. Here we move adsorbed H atom on its migration path with
minimum energy barrier on graphene by fixing its x- and y-coordinates, but fully
relaxing its z-coordinate, as well as all the coordinates of adsorbed O atom and of all
carbon atoms of graphene. As the coadsorbed H approaches the adsorbed O, the
energy falls suddenly by *1.4 eV when OH forms. This exothermic process occurs
without any barrier. At this moment, owing to the constraints in the approach of H
adatom, O adatom desorbs to form strong O–H bond. Eventually, OH becomes
detached from graphene. Apparently, the bond energy of OH compensate the binding
energies of O and H atoms with graphene, as well as the energy lowering of *1.4 eV
of whole system. However, as shown in Fig. 11.10a, an energy barrier of 1.3 eV can
develop as the adsorbed H approaches the adsorbed O, if the graphene is fixed from
corner atoms, rather than it is fully relaxed. (ii) Along the second path shown in
11 Effects of Charging and Perpendicular Electric Field … 281

(a)
K
3
p
b H
d
2 a c e I
L
II
1
b

d
III
0 c 1.3 eV J
a

−1 p p p
K
H d
d d d G
−2
ENERGY (eV)

L J
G
−3
(b) 1 B
0.8 eV

0.5 I D
IV 0.4 eV
0
A C
III II F
−0.5 C
A B

−1

−1.5
C D E

−2
E
C F E
−2.5
Atomic Configurations
 282 H. Hakan Gürel et al.

b Fig. 11.9 Interaction between two oxygen adatoms (epoxy groups) at close proximity. a Variation
of the interaction energy between two oxygen adatom are presented for three different case
represented by curves I–III associated with single path. Calculations are carried out for a
rectangular supercell consisting of 72 carbon atoms and two oxygen adatoms. Details are given in
the text. The top inset shows the isosurface charge densities of C–C bonds and positions of two
oxygen adatoms in the course of migration. Bottom insets describes how O2 and O are desorbed
from various configurations. b Variation of the interaction energy between two oxygen adatoms
involving dual paths are presented by curve I and II. Relevant atomic configurations A-F
associated with dual paths and ending with O2 desorption are shown by insets. The modification of

¼ 0:332 C/m2 ) and
curve I and II under excess electronic charge (Q = −4 e per supercell or r
applied electric field (E = −2 V/Å) (Results are obtained from AO calculations using PBC.) This
figure is taken from [29]

Fig. 11.10b, where O is adsorbed at the bridge site, free H is approaching it from the
top vertically. At a specific distance overcoming an energy barrier of 0.1 eV, OH is
formed in an exothermic process by lowering the energy of the system by 2.6 eV.
Because of the strategy of approach, O adatom, which is detached from graphene
forms weakly bound OH. The gain of energy through the formation of free OH
compensate the desorption of O adatom. Upon the adsorption of OH the total energy
can be further lowered to *3.3 eV. (iii) Along the third path shown in Fig. 11.10c,
whereby O is adsorbed, free H is approaching it horizontally to form OH. The for-
mation of OH occurs without any barrier, and the total energy is first lowered 2.6 eV
once weakly bound OH is formed. Thereafter, the energy is further lowered to
*3.3 eV upon the adsorption of OH. It appears that the formation of OH through the
interaction between O adatom and H atom, which is either free in the environment or
coadsorbed to graphene can occur easily by the release of significant energy.
Formation of OH adsorbed on graphene is a crucial step towards the formation of H2O
from hydroxyl groups.

11.6.3 Interaction Between Adsorbed H and OH

Here we consider also three different cases to investigate the interaction between H
and adsorbed OH. (i) As described in Fig. 11.11a, initially one H and OH are coad-
sorbed at close proximity and occupy the top sites of two outer carbon atoms of two
adjacent C–C bonds. When a weakly bound H2O is formed, the energy gained from
this process compensates the sum of the binding energies of H and OH and further
lowers the total energy by 2.65 eV. The cases shown in Fig. 11.11b, c are similar to the
above analysis and involve the interaction between free H atom and adsorbed OH. In
the case described in Fig. 11.11b, the barrier is only 30 meV between adsorbed OH
and free H atom. Once this small barrier is overcame, a weakly bound H2O forms. In
this exothermic process, an energy of 4.5 eV is released. This is a significant energy
which can trigger other reactions. The reaction described in Fig. 11.11c takes place in
two stages: First, free H atom is bonded to graphene temporarily, eventually an energy
of 4.5 eV is released, when a weakly bound H2O is formed.
11 Effects of Charging and Perpendicular Electric Field … 283

Fig. 11.10 Variation of the (a) 1.5

total energy ET as H atom is
approaching the adsorbed 1
oxygen from different
1.34 eV
directions by a displacement
0.5
Ds. a H adatom on graphene

ET (eV)
is approaching oxygen atom
adsorbed at the bridge site. 0
Energy variation presented by
black (red) dashed lines −0.5
corresponds to graphene
substrate, which is fully
−1
relaxed (fixed at the corner
atoms). Ds follows the energy
−1.5
path with minimum barrier 0 0.5 1 1.5
described in the text. b Free H
atom is approaching the (b) 0.5
adsorbed O atom vertically
0 0.1 eV
from the top and is forming
weakly bound OH. Here
−0.5
Ds ¼ Dz. c H atom is 2.9 A0
ET (eV)

approaching the adsorbed O

−1
atom horizontally. Weakly
bound OH corresponds to an
−1.5
intermediate state, which is
exited by *0.9 eV relative to
−2
adsorbed OH. Calculations
are performed using (4 × 4)
−2.5
supercell of graphene.
Large-brown, large-red and
−3
small-yellow atoms are 0 0.5 1 1.5 2 2.5
carbon, oxygen and hydrogen
atoms, respectively. This
(c) 0.5
figure is taken from [31] 0
2.8 A0
−0.5

−1
ET (eV)

−1.5

−2

−2.5

−3

−3.5
0 0.5 1 1.5 2 2.5
Δs (A0)
284 H. Hakan Gürel et al.

Fig. 11.11 a Initially, H and (a)

OH are coadsorbed and are at
the close proximity occupying
outer top sites of two adjacent
C–C bonds. When H2O is H
formed the total energy is
lowered by −2.65 eV. E T= -2.65 eV
b Variation of the total energy
ET as a free H atom (b) 0.5
approaches to OH from the 0 30 meV
top by Ds ¼ Dz. Formation −0.5 2.9 A0
of H2O occurs upon
−1
overcoming an energy barrier
of 30 meV. c As a free H −1.5
atom approaching −2
ET (eV)

horizontally first it is adsorbed −2.5

to graphene and then form
−3
H2O. Whole process is
exothermic and release −3.5
*4.5 eV. H2O by itself is −4
weakly bound to graphene −4.5
and can desorb easily.
Calculations are performed −5
0 0.5 1 1.5 2 2.5 3
using (4 × 4) supercell of
graphene. This figure is taken (c) 0.5
2.8 A0
from [31] 0
−0.5
−1
−1.5
ET (eV)

−2
−2.5
−3
−3.5
−4
−4.5
0 0.5 1 1.5 2 2.5 3
Δs (A0)

11.6.4 Interaction Between Two OH Co-Adsorbed in Close

Proximity

Finally, we examine the interaction and chemical processes, when two OH coad-
sorbed at the close proximity approach each other. The interaction between two
coadsorbed OH’s is relevant for the deoxidation of GOX for the reasons pointed out
at the beginning. Here, we move one OH along the path of minimum energy barrier
towards the other OH. While moving one OH, its x- and y-coordinates are fixed
11 Effects of Charging and Perpendicular Electric Field … 285

(a) 0.4 Neutral

2e added B
4e added
0.2
A
0
C I
−0.2

−0.4

−0.6 D II

−0.8

D III
−1
ENERGY (eV)

−1.2

(b) 1 Neutral
−0.5 V/A0
−1 V/A0
0.5 V/A0
0.5 B

A
0 C a

E b
D c
−0.5

d
−1
D

−1.5
Atomic Configurations

A B C D E

Fig. 11.12 Variation of the total energy ET as two OH atoms adsorbed at close proximity are
approaching each other. a The system is charged by Q = −2 and Q = −4 electrons per supercell.
b The system is under an electric field Efield ¼ þ 0:5, −0.5 and −1.0 V/Å. The neutral system (i.e.
Q = 0 and Efield ¼ 0) is also shown by green dashed lines for the sake of comparison. Various
atomic configurations starting from A going through B and ending in C or D or E are described at
the bottom of the figure. Large brown, large red and small yellow balls represent C, O and H
atoms, respectively. The zero of energy is set to the total energies of the initial configuration-A i.e.
two adsorbed OH’s are widely separated. Calculations are performed using (6 × 6) supercell of
graphene. This figure is taken from [31]
286 H. Hakan Gürel et al.

along the migration path, but its z-coordinate, all the coordinates of the second OH,
as well as all the coordinates of graphene atoms are fully relaxed. In this case, an
energy barrier of *0.4 eV prevents these two OH from engaging in a chemical
reaction. Note that due to OH–OH interaction this barrier is smaller than EB in
Fig. 11.2. When the energy barrier is overcame, the chemical reaction sets in to
form one weakly bound H2O molecule and one O atom bound to the bridge site.
The former is prone to desorb easily and hence to remove one adsorbed O atom
from GOX. The relevant processes, A → B → C, and corresponding energy
variation are shown in Fig. 11.12.
Since a significant barrier is involved in OH–OH interaction, here we examine
how the energy barrier at B is modified externally by charging or by applying
perpendicular electric field. First, we consider the case, where the system is neg-
atively charged by implementing two excess electrons, i.e. Q = −2 electrons/cell.
Under these circumstances, the energy barrier is dramatically lowered to *0.1 eV.
Overcoming this small barrier, the configuration-A proceeds to the configuration-D,
whereby two coadsorbed OH form one H2O molecule and one O atom adsorbed to
the top site. Interestingly, for Q = −4 electrons/cell, the energy barrier completely
disappears and the system directly passed from the configuration-A to the
configuration-D forming again weakly bound H2O and O atom adsorbed at the
bridge site.
As shown in Fig. 11.12b, the OH–OH interaction under applied perpendicular
E is reminiscent of the above charged cases. While the energy barrier of neutral
system under Efield ¼ 0 corresponds to the configuration-B, it increases to 0.95 eV
under Efield ¼ þ 0:5 V/Å. However, it decreases to 0.35 and to 0.05 eV under
Efield ¼ 0:5 V/Å and Efield ¼ 1:0 V/Å, respectively. Under Efield ¼ þ 0:5 V/Å,
the system transforms to the configuration-E, whereby H2O2 is released.
The variations of interaction energies of two coadsorbed OH under excess
charge and/or perpendicular electric field demonstrate the critical role played by
hydroxyl groups in the reduction process of GOX. Since epoxy and hydroxyl
groups coexist in GOX, desorption of oxygen adatom can take place through H2O
or H2O2 almost spontaneously under appropriate Q or E.

11.7 Conclusion

In order to understand recent experimental works reporting reduction of GOX by

charging, by heating the system, or by applying electric field, we investigated the
effects of these external agents on graphene. First, we discussed the limitations and
artifacts of methods using plane wave (PW), and local orbital (AO), basis sets
within periodic boundary conditions in treating the systems which are either
charged or exerted by perpendicular electric field. While both methods have been
successful in treating the atomic and electronic structure of various neutral or
positively charged adatom-surface complexes, AO calculations hindering the
11 Effects of Charging and Perpendicular Electric Field … 287

spurious vacuum charging within PBC are found to be appropriate to treat the
systems, which are either negatively charged or are exerted by a perpendicular
electric field. It appears that an artifact of AO calculations cancels out another
artifact of PBC occuring in PW calculations. Our related analyses are believed to be
very valuable for future theoretical studies including excess charge and electric
field.
Determining that AO calculations can provide the proper treatment of one single
graphene layer, which is charged negatively or exerted by a perpendicular electric
field, we carried out structure optimized, self-consistent field calculations under
specific constraints and NEB calculations between well defined initial and final
configurations. We found that the strength of the bond between graphene and
oxygen adatom is weakened with negative charging and/or with perpendicular
electric field applied in the direction opposite to oxygen. Under negative perpen-
dicular electric field and negative charging, both coexisting in the system the
desorption and hence reduction of GOX can take place much easier. Despite the
negative formation energy, the formation of oxygen molecule from individual
oxygen adatoms adsorbed to graphene are hindered by the energy barriers of 0.8–
1.3 eV. The calculated energy barriers are significantly lower than the sum of the
binding energies of two single oxygen adatom due to the concerted action. In this
respect, we note the possibility of another reaction path even with lower energy
barrier.
We also considered interactions among coadsorbed H, O, and OH. Adsorbed or
free hydrogen atoms can easily interact with oxygen adatom to form OH. Moreover,
free or adsorbed hydrogen atoms can also interact with adsorbed OH to from H2O.
Adsorbed OH by themselves can diffuse relatively easy and interact with each
other. However, an energy barrier of 0.4 eV hinders them to engage in a chemical
process. We showed that by negatively charging the system or by applying per-
pendicular field one can suppress this energy barrier and promote the chemical
reaction to form H2O. H2O by itself is very weakly bound to graphene and can
desorb in ambient temperature. Each desorbed H2O removes one oxygen atom from
graphene oxide.
The calculated energy barriers on various reaction paths are shown to be lowered
by charging the system with excess electrons or by applying a perpendicular electric
field. This explains why the reduction of GOX through desorption of oxygen or O2
is facilitated as reported by earlier experimental studies. Notably, perpendicular
electric field can be used also for side specific functionalization of single and multi
layered nanostructures. Finally, we note that the levels of charging or applied
electric field comparable to those used in this study can be achieved locally in
experiments through the tip of STM or femtosecond laser systems.

Acknowledgments The Authors acknowledge useful and stimulating discussions with

Professor A. Dana and Mr O. Ekiz. Dr H.H. Gürel acknowledges the support of
TUBITAK-BIDEB. Part of the computational resources has been provided by
TUBITAK ULAKBIM, High Performance and Grid Computing Center (TR-Grid e-Infrastructure)
and UYBHM at Istanbul Technical University through Grant No. 2-024-2007. The content of this
review has been published in [29, 31, 32, 42, 43]
288 H. Hakan Gürel et al.

References

1. H.C. Sniepp, J.-L. Li, M.C. McAllister, H. Sai, M. Herrera-Alonso, D.H. Adamson, R.K.
Prud’homme, R. Car, D.A. Saville, I.A. Aksay, Functionalized single graphene sheets derived
from splitting graphite oxide. J. Phys. Chem. B 110, 8535–8539 (2006)
2. J.-L. Li, K.N. Kudin, M.J. McAllister, R.K. Prudhomme, I.A. Aksay, R. Car, Oxygen-driven
unzipping of graphitic materials. Phys. Rev. Lett. 96, 176101–176105 (2006)
3. D.W. Boukhvalov, M.I. Katsnelson, Modeling of graphite oxide. J. Am. Chem. Soc. 130,
10697–10701 (2008)
4. J. Yan, L. Xian, M.Y. Chou, Structural and electronic properties of oxidized graphene. Phys.
Rev. Lett. 103, 086802–086806 (2009)
5. J.-A. Yan, M.Y. Chou, Oxidation functional groups on graphene: structural and electronic
properties. Phys. Rev. B 82, 125403–125413 (2010)
6. L. Wang, Y.Y. Sun, K. Lee, D. West, Z.F. Chen, J.J. Zhao, S. Zhang, Stability of graphene
oxide phases from first-principles calculations. Phys. Rev. B 82, 161406–161409 (2010)
7. A. Bagri, C. Mattevi, M. Acik, Y.J. Chabal, M. Chhowalla, V.B. Shenoy, Structural evolution
during the reduction of chemically derived graphene oxide. Nat. Chem. 2, 581–587 (2010)
8. H.J. Xiang, S.-H. Wei, X.G. Gong, Structural motifs in oxidized graphene: a genetic algorithm
study based on density functional theory. Phys. Rev. B 82, 035416–5 (2010)
9. E.C. Mattson, H. Pu, S. Cui, M.A. Schofield, S. Rhim, G. Lu, M.J. Nasse, R.S. Ruoff, M.
Weinert, M. Gajdardziska-Josifovska, J. Chen, C.J. Hirschmugl, Evidence of nanocrystalline
semiconducting graphene monoxide during thermal reduction of graphene oxide in vacuum.
ACS Nano 5, 9710–9717 (2011)
10. S. Pei, H. Cheng, The reduction of graphene oxide. Carbon 50, 3210–3228 (2012)
11. S. Mao, H. Pu, J. Chen, Graphene oxide and its reduction: modeling and experimental
progress. RSC Adv. 2, 2643–2662 (2012)
12. H. Huang, Z. Li, J. She, W. Wang, Oxygen density dependent band gap of reduced graphene
oxide. J. Appl. Phys. 111, 054317–4 (2012)
13. S. Kim, S. Zhou, Y. Hu, M. Acik, Y.J. Chabal, C. Berger, W. de Heer, A. Bongiorno, E.
Riedo, Room-temperature metastability of multilayer graphene oxide films. Nat. Mater. 11,
544–549 (2012)
14. T. Sun, S. Fabris, Mechanisms for oxidative unzipping and cutting of graphene. Nano Lett. 12,
17–21 (2012)
15. M. Topsakal, S. Ciraci, Domain formation on oxidized graphene. Phys. Rev. B 86, 205402–
205413 (2012)
16. R. Larciprete, S. Fabris, T. Sun, P. Lacovig, A. Baraldi, S. Lizzit, Dual path mechanism in the
thermal reduction of graphene oxide. J. Am. Chem. Soc. 133, 17315–17321 (2011)
17. K. Novoselov, A. Geim, S. Morozov, D. Jiang, Y. Zhang, S. Dubonos, I. Grigorieva, A.
Firsov, Electric field effect in atomically thin carbon films. Science 306, 666–669 (2004)
18. D.A. Dikin, S. Stankovich, E.J. Zimney, R. Piner, G.H.B. Dommett, G. Evmenenko, S.T.
Nguyen, R.S. Ruoff, Preparation and characterization of graphene oxide paper. Nature 448,
457–460 (2007)
19. P.P. Yao, P.L. Chen, L. Jiang, H.P. Zhao, H.F. Zhu, D. Zhou, W.P. Hu, B.H. Han, M.H. Liu,
Electric current induced reduction of graphene oxide and its application as gap electrodes in
organic photoswitching devices. Adv. Mater. 22, 5008 (2010)
20. Z. Wei, D. Wang, S. Kim, S.Y. Kim, Y. Hu, M.K. Yakes, A.R. Laracuente, Z. Dai, S.R.
Marder, C. Berger, W.P. King, W.A. de Heer, P.E. Sheehan, E. Riedo, Nanoscale tunable
reduction of graphene oxide for graphene electronics. Science 328, 1373–137 (2010)
21. O.O. Ekiz, M. Urel, H. Guner, A.K. Mizrak, A. Dana, Reversible electrical reduction and
oxidation of graphene oxide. ACS Nano 5, 2475–2482 (2011)
22. P. Blochl, Projector augmented-wave method. Phys. Rev. B 50, 17953–17979 (1994)
23. J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made simple. Phys.
Rev. Lett. 77, 3865–3868 (1996)
11 Effects of Charging and Perpendicular Electric Field … 289

24. G. Kresse, J. Furthmuller, Efficient iterative schemes for ab initio total-energy calculations
using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 (1996)
25. J.M. Soler, E. Artacho, J.D. Gale, A. Garcia, J. Junquera, P. Ordejon, D. Sanchez-Portal,
The SIESTA method for Ab initio order-N materials simulation. J. Phys.: Conden. Matter 14
(11), 2745–2779 (2002)
26. G. Makov, M.C. Payne, Periodic boundary conditions in ab initio calculations. Phys. Rev.
B 51, 4014–4022 (1995)
27. H.J. Monkhorst, J.D. Pack, Special points for Brillouin-zone integrations. Phys. Rev. B 13,
5188 (1976)
28. N. Troullier, J.L. Martins, Efficient pseudopotentials for plane-wave calculations. Phys. Rev.
B 1991, 43 (1993)
29. M. Topsakal, H.H. Gurel, S. Ciraci, Effects of charging and electric field on graphene oxide.
J. Phys. Chem. C 117, 5943 (2013)
30. H.H. Gurel, V.O. Ozcelik, S. Ciraci, Effects of charging and perpendicular electric field on the
properties of silicene and germanene. J. Phys.: Condens. Matter 25, 305007 (2014)
31. H.H. Gurel, S. Ciraci, Enhanced reduction of graphene oxide by means of charging and
electric fields applied to hydroxyl groups. J. Phys.: Condens. Matter 25, 435304 (2014)
32. H.H. Gurel, S. Ciraci, Effects of charging and electric field on graphene functionalized with
titanium. J. Phys.: Condens. Matter 25, 275302 (2014)
33. R. Poloni, A.S. Miguel, M.V. Fernandez-Serra, A first principles study of the effect of charge
doping on the 1D polymerization of C60. J. Phys.: Condens. Matter 24, 095501–095507
(2012)
34. C. Attaccalite, L. Wirtz, M. Lazzeri, F. Mauri, A. Rubio, Doped graphene as tunable
electron-phonon coupling material. Nano Lett. 10, 1172–1176 (2010)
35. B. Aradi, B. Hourahine, T. Frauenheim, DFTB+, a sparse matrix-based implementation of the
DFTB method. J. Phys. Chem. A 111, 5678–5684 (2007)
36. A.S. Bernard, I.K. Snook, Transformation of graphene into graphane in the absence of
hydrogen. Carbon 48, 981–987 (2010)
37. B. Delley, An all-electron numerical method for solving the local density functional for
polyatomic molecules. J. Chem. Phys. 92, 508–517 (1990)
38. Z.M. Ao, A.D. Hernandez-Nieves, F.M. Peeters, S. Lia, The electric field as a novel switch for
uptake/release of hydrogen for storage in nitrogen doped graphene. Phys. Chem. Chem. Phys.
14, 1463–1467 (2012)
39. K.T. Chan, H. Lee, M.L. Cohen, Gated adatoms on graphene studied with first-principles
calculations. Phys. Rev. B 83, 035405–035410 (2011)
40. V.W. Brar, R. Decker, H.M. Solowan, Y. Wang, L. Maserati, K.T. Chan, H. Lee, C.O. Girit,
A. Zettl, S.G. Louie, M.L. Cohen, M.F. Crommie, Gate-controlled ionization and screening of
cobalt adatoms on a graphene surface. Nat. Phys. 7, 43 (2011)
41. A.M. Suarez, L.R. Radovic, E. Bar-Ziv, J.O. Sofo, Gate voltage control of oxygen diffusion on
graphene. Phys. Rev. Lett. 106, 146802–146804 (2011)
42. M. Topsakal, S. Ciraci, Static charging of graphene and graphite slabs. Appl. Phys. Lett. 98,
131908–131911 (2011)
43. M. Topsakal, S. Ciraci, Effects of static charging and exfoliation of layered crystals. Phys.
Rev. B 85, 045121–045127 (2012)
44. K.T. Chan, J.B. Neaton, M.L. Cohen, First-principles study of metal adatom adsorption on
graphene. Phys. Rev. B 77, 235430–12 (2008)
45. N.D. Drummond, V. Zolyomi, V.I. Falko, Electrically tunable band gap in silicene. Phys. Rev.
B 85, 075423-7 (2012)
46. P. Hohenberg, W. Kohn, Inhomogeneous Electron Gas Phys. Rev. 136, B864 (1964); W.
Kohn, L.J. Sham, Self-consistent equations including exchange and correlation effects. ibid.
140, A1133 (1965)
47. X. Lin, J. Ni, C. Fang, Adsorbtion capacity of H2O, NH3, CO, NO2 on pristine graphene.
J. Appl. Phys. 113, 034306 (2013)
290 H. Hakan Gürel et al.

48. O. Leenaerts, B. Partoens, F.M. Peeters, Adsorption of H2O, NH3, CO, NO2, and NO on
graphene: a first-principles study. Phys. Rev. B 77, 125416 (2007)
49. W.W. Cai, R.D. Piner, F.J. Stadermann, S. Park, M.A. Shaibat, Y. Ishii, D.X. Yang, A.
Velamakanni, S.J. An, M. Stoller, Synthesis and solid-state NMR structural characterization of
C-13-labeled graphite oxide. Science 321, 1815–1817 (2008)
50. T. Szabo, O. Berkesi, P. Forgo, K. Josepovits, Y. Sanakis, D. Petridis, I. Dekany, Evolution of
surface functional groups in a series of progressively oxidized graphite oxides. Chem. Mater.
18, 2740–2749 (2006)
51. A. Jelea, F. Marinelli, Y. Ferro, A. Allouche, C. Brosset, Quantum study of hydrogen oxygen
graphite interactions. Carbon 42, 3189–3198 (2004)
52. R.J.W.E. Lahaye, H.K. Jeong, C.Y. Park, Y.H. Lee, Density functional theory study of
graphite oxide for different oxidation levels. Phys. Rev. B 79, 125435 (2009)
53. M.C. Kim, G.S. Hwang, R.S.J. Ruoff, Epoxide reduction with hydrazine on graphene: a first
principles study. Chem. Phys. 131, 064704 (2009)
54. S.C. Xu, S. Irle, D.G. Musaev, M.C.J. Lin, Quantum chemical study of the dissociative
adsorption of OH and H2O on pristine and defective graphite (0001) surfaces: reaction
mechanism and kinetics. Phys. Chem. C 111, 1355–1365 (2007)
55. P.A.J. Denis, Density functional investigation of thioepoxidated and thiolated graphene. Phys.
Chem. C 113, 5612–5619 (2009)
56. Q. Zhu, Y.H. Lu, J.Z. Jiang, Stability and properties of two-dimensional graphene hydroxide.
J. Phys. Chem. Lett. 2, 1310–1314 (2011)
57. H. Khantha, N.A. Cordero, L.M. Molina, D.A. Alonso, L.A. Grifalco, Interaction of lithium
with graphene: an ab initio study. Phys. Rev B 70, 125422 (2004)
58. M. Topsakal, H. Sahin, C. Ciraci, Graphene coatings: an efficient protection from oxidation.
Phys. Rev. B 85, 155445–155447 (2012)
59. S. Grimme, Semiempirical GGA-type density functional constructed with a long-range
dispersion correction. J. Comput. Chem. 27, 1787 (2006)
60. H.H. Gurel, V.O. Ozcelik, S. Ciraci, Dissociative adsorption of molecules on graphene and
silicene. J. Phys. Chem. C 118(47) (2014)
61. L. Jeloaica, V. Sidis, DFT investigation of the adsorption of atomic hydrogen on a
cluster-model graphite surface. Chem. Phys. Lett. 300, 157 (1999)
62. X. Sha, B. Jackson, First-principles study of the structural and energetic properties of H atoms
on a graphite (0001) surface. Surf. Sci. 496, 318 (2002)
63. S. Casolo, O.M. Lvvik, R. Martinazzo, G.F. Tantardini, Understanding adsorption of
hydrogen atoms on graphene. J. Chem. Phys. 30, 054704 (2009)
64. V.V. Ivanovskaya, A. Zobelli, D. Teillet-Billy, N. Rougeau, V. Sidis, P.R. Briddon, Hydrogen
adsorption on graphene: a first principles study. Eur. Phys. J. B 76, 481 (2010)
65. S.H. Vosko, L. Wilk, M. Nusair, Accurate spin-dependent electron liquid correlation energies
for local spin density calculations: a critical analysis. Can. J. Phys. 58, 1200 (1980)
66. D. Henwood, J.D. Carey, Ab initio investigation of molecular hydrogen physisorbtion on
graphene and carbon nanotubes. Phys. Rev. B 75, 245413 (2007)
67. S. Cahangirov, M. Topsakal, E. Akturk, H. Sahin, S. Ciraci, Two- and one-dimensional
honeycomb structures of silicon and germanium. Phys. Rev. Lett. 102, 236804–236811 (2009)
68. H. Sahin, S. Cahangirov, M. Topsakal, Bekaroglu, E,; Akturk, E.; Senger R.T.; Ciraci,
S. Monolayer honeycomb structures of group-IV elements and III-V binary compounds:
first-principles calculations. Phys. Rev. B 80, 155453 (2009)
69. C. Ataca, H. Sahin, S. Ciraci, Stable, single-layer MX2 transition-metal oxides and
dichalcogenides in a honeycomb-like structure. J. Phys. Chem. C 116, 8983–8999 (2012)
70. P.P. Yao, P.L. Chen, L. Jiang, H.P. Zhao, H.F. Zhu, D. Zhou, W.P. Hu, B.H. Han, M.H. Liu,
Electric current induced reduction of graphene oxide and its application as gap electrodes in
organic photoswitching devices. Adv. Mater. 22, 5008 (2010)
Chapter 12
Structural and Optical Properties
of Tungsten Oxide Based Thin Films
and Nanofibers

E.O. Zayim and A. Tabatabaei Mohseni

Abstract Tungsten oxide nanomaterials confined to one and two dimensions can
be prepared with tungsten metal powder, tungsten chloride, peroxotungstic acid and
acetylated peroxotungstic acid precursors by sol-gel, evaporation and electrode-
position techniques. Nanofibers and nanowires of tungsten oxide are synthesized by
organic/inorganic blend of tungsten hexachloride, tungsten metal powder and
polyvinylpyrrolidone with electrospinning technique. Standard and mesoporous
tungsten oxide thin films are prepared from an ethanolic solution of tungsten
hexachloride. Several polymers were employed as a template to generate the
mesoporous structure. Additionally a detailed systematic study of the evaporated
tungsten oxide thin films has been carried out at progressively increasing temper-
atures. Overall, the optical, electrochemical and structural properties of the
deposited films were examined in both liquid and solid electrolytes. All solid
electrochromic devices were fabricated using tungsten oxide active electrochromic
layers. The fabrication and evaluation of a prototype solid-state electrochromic
device are also described.

12.1 Introduction

Tungsten oxide (WO3), also known as tungsten trioxide or tungstic anhydride, is an

n-type semiconductor with a wide band gap of 2.5–3.2 eV [1]. Tungsten trioxide is
a thermally stable and water insoluble. Tungsten oxides are unique materials that
have been extensively studied because of their unique physical and chemical
properties and widespread applications. Industrial applications of tungsten oxide

E.O. Zayim (&)

Physics Department, Istanbul Technical University, Maslak, Istanbul, Turkey
e-mail: [email protected]
A. Tabatabaei Mohseni
Nanoscience & Nanotechnology Program, Istanbul Technical University, Maslak,
Istanbul, Turkey

© Springer International Publishing Switzerland 2016 291

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_12
292 E.O. Zayim and A. Tabatabaei Mohseni

coatings have encouraged many research groups to investigate this specific material
from different perspectives.
Thin films and nanomaterials of tungsten trioxide are rigorously investigated for
chromogenic devices, photo- and electrochromic “smart” windows, erasable
optical-storage devices, photocatalysts, biological elements [2], electrodes for solar
cells, lithium-ion batteries [3, 4], supercapacitors [5], x-ray screen phosphors,
fireproofing fabrics, humidity and temperature sensors and in gas-sensing elements
[6, 7]. They exhibit further important properties and functionalities. Recent works
present a general review of nanostructured WOx, their properties, methods of
synthesis, and a description of how they can be used in unique ways for different
applications. This oxide film shows both reversible wettability conversion between
superhydrophobic and super-hydrophilic states.

12.1.1 Amorphous and Crystalline Tungsten Oxide Based

Nanomaterials

Tungsten oxide based nanomaterials prepared by various techniques, such as

thermal evaporation, chemical vapor deposition, sputtering, and sol–gel methods,
possess different micro- and nanostructural properties. As in every chemical and
physical deposition method, heat treatment plays an important role, because
structural, optical and electrical properties will change by annealing. Therefore, it is
important to characterize WO3 films as a function of both the fabrication and the
annealing conditions. WO3 film grown at substrate temperature which is less than
300 °C is amorphous and transparent. Crystallization of the films can be observed at
a temperature of 350 °C or higher [8].
Presently, scientists and engineers are paying much attention to nanostructured
oxides with zero-dimensional quantum dots, one-dimensional nanowires and
nanorods, and two-dimensional nanosheets and nanodisks, due to their novel
physical and chemical properties which are different from their corresponding bulks
and amorphous tungsten oxide films [9].
Tungsten trioxide (WO3) has a perovskite structure (with no A-site cations) [10]
which is referred to any material with the same type of crystal structure with general
chemical formula of ABX3, where ‘A’ and ‘B’ are two cations of very different sizes,
and X is an anion that bonds to both. As shown in Fig. 12.1 the tungsten ions occupy
the corners of a primitive unit cell and oxygen ions bisect the unit cell edges [11].
Investigations by Ramana et al. on thin film of WO3 deposited by pulsed-laser
deposition (PLD) technique, represent the following crystal structure sequences
depending on their temperature increment, which is also true for bulk tungsten
oxides: amorphous (<300 °C) → monoclinic (350 °C) → orthorhombic
(500 °C) → hexagonal (740 °C) → tetragonal [8].
12 Structural and Optical Properties of Tungsten Oxide … 293

Fig. 12.1 Unit cell for the

perovskite lattice and the
octahedral symmetry

Fig. 12.2 X-ray diffraction

patterns of the PLD WO3–x
thin films: a ITO; b Room
Temperature; c annealed at
100 °C; d annealed at 200 °C;
e annealed at 300 °C;
f annealed at 400 °C;
g annealed at 500 °C

The physical properties of slightly oxygen-deficient analogues tungsten trioxide,

WO3–x, possess several crystallographic phases [12]. WO3–x films have been
extensively studied for their chromism, which exhibit different chromic properties
for different levels of oxygen deficiency: x > 0.5 films are metallic and conductive,
0.3 < x < 0.5 films are blue and conductive, and x < 0.3 films are transparent and
resistive; these findings are independent of the film preparation method [13].
Figure 12.2 demonstrates that amorphous WO3–x thin films deposited by PLD
remain in an amorphous phase up to 300 °C. The thin films form a crystalline
structure by post annealing at 400 and 500 °C [14].
Reactive RF magnetron sputtering can be applied for depositing a slightly
reduced amorphous WO3–x comes up with the same result for crystallization of the
294 E.O. Zayim and A. Tabatabaei Mohseni

Fig. 12.3 XRD patterns for as-deposited and annealed WO3–x films coated by reactive RF
magnetron sputtering, a Room Temperature; b annealed at 400 °C

Fig. 12.4 XRD patterns of

thermal evaporated tungsten
oxide thin films annealed at
different temperatures. (+)
indicates the peaks from the
substrate

film by annealing at 400 °C in vacuum [15]. Figure 12.3 shows the XRD patterns
for the as-prepared and annealed tungsten oxide films. No XRD peaks were
observed for the as-prepared films. Annealed films have a hexagonal single phase.
Thermally evaporated tungsten oxide films were obtained from WO3 powder.
As-deposited amorphous WO3–x films were subjected to different annealing tem-
peratures to obtain various crystalline phases. They were annealed at progressively
increasing temperatures ranging from 350 to 450 °C in oxygen ambient. Figure 12.4
demonstrates that amorphous WO3–x films remain in an amorphous phase up to
385 °C and begin to crystallize at 390 °C and then are completely crystallized at
450 °C. The peaks in the XRD pattern can be attributed to a monoclinic crystalline
WO3 structure.
12 Structural and Optical Properties of Tungsten Oxide … 295

Fig. 12.5 Raman spectra of

thermal evaporated tungsten
oxides thin films annealed at
different temperatures

Raman spectra of the thermal evaporated WO3–x films are shown in Fig. 12.5.
The spectrum of an as-deposited film (a) shows a broad peak at 770 cm−1, due to
the W6+–O bonds. There is also a relatively sharp peak at 950 cm−1, which has been
assigned to the W6+=O stretching mode of terminal oxygen atoms. In addition,
there is a weaker, broad peak at 220 cm−1, which has been appeared from the
presence of W4+ states. The peak at 770 cm−1 gets sharper and one new peak at
719 cm−1 develops at 385 °C. At 390 °C new peaks at 719 and 807 cm−1
develop. At 400 °C annealing temperature, these peaks get sharper and the
950 cm−1 (C) component decreases. The two strongest peaks appear at 719 (A) and
807 (B) cm−1 in the Raman spectrum of crystalline WO3. Also two new peaks at
127 and 330 cm−1 develop in crystalline films. Overall, the amorphous WO3 film
begins to crystallize at 390 °C and is completely crystallized at 450 °C. A peak at
950 cm−1, which is a key characteristic of the amorphous phase, remains up to
400 °C indicating the coexistence of both crystalline and amorphous phases [16].

12.2 Tungsten Oxide Based Nanomaterials

Nanoscaled materials produced based on tungsten oxide have various physical and
chemical properties according to their crystallography, confinement in
one-dimension or two-dimensions, oxygen deficiency and consequently the band
gap. In the following we will take a look at properties of tungsten oxide in thin film
and nanofiber forms.
296 E.O. Zayim and A. Tabatabaei Mohseni

12.2.1 Tungsten Oxide Based Thin Films

and Mesoporous Thin Films

Surfactants are able to modify and control the properties of electrode surfaces. The
influence of surface active agents on the kinetics of electron transfer reactions at
electrodes have been investigated in the past few decades [17]. Since the discovery
of surfactant templated mesoporous materials, the templating method has been
extended to a range of transition metal oxides [16]. Mesoporous tungsten oxide thin
films prepared by nonionic surfactants as templates, show high-rate ion-insertion
performance when used as electrochromic layers.
Different types of polymers were employed as a template to generate the
mesoporous structure. These polymers are poly(ethylene glycol) (PEG), poly-
ethylene glycol-ran-propylene glycol and polystyrene-co-allyl-alcohol. Here the
application of PEG is given as an example. To avoid the varying degrees of
crystallinity as a result of thermal treatment, a UV illumination method has been
employed to remove the polymer surfactant. This room temperature approach uses
ozone, generated during UV illumination in air to oxidize the organic compounds.
The electrochromic and optical properties of the mesoporous films are described
and compared to standard sol-gel WO3 films. Results are also presented on the
samples prepared by thermal treatment. It has been demonstrated that the UV
illumination treatment is a superior method to remove templates, leading to more
detailed investigation of the effect of mesoporosity on the electrokinetics of ion
insertion into WO3 films. These mesoporous materials exhibit superior high rate
ion-insertion performance when used as electrochromic layers, which is attributed
to the high surface area of mesoporous WO3.
The introduction of mesoporosity into WO3 layers has been predicted to improve
their functional film properties due to the porous nature of these materials and better
penetration of electrolytes into the inorganic framework. The improved elec-
trochromic (EC) properties of mesoporous tungsten oxide films produced via these
templates are described and compared to standard sol-gel films of tungsten oxides.
SEM was used to study the surface structure of all deposited layers shown in
Figs. 12.6 and 12.7. The images indicate that polymer generates a remarkable

(a) (b)

Fig. 12.6 SEM image of a as-deposited tungsten oxide film; b calcined tungsten oxide film at
400 °C
12 Structural and Optical Properties of Tungsten Oxide … 297

Fig. 12.7 SEM image of a as-deposited polymer-added tungsten oxide film; b calcined poly-
mer-added tungsten oxide film at 400 °C

Fig. 12.8 XRD patterns of WO3–x thin films: a As-deposited tungsten oxide film; b Tungsten
oxide film treated by UV illumination; c annealed tungsten oxide film at 400 °C; d As-deposited
polymer-added tungsten oxide film; e Polymer-added tungsten oxide film treated by UV
illumination; f annealed polymer-added tungsten oxide film at 400 °C

change on the structural properties of tungsten oxide films. As shown in SEM

results, while as-deposited tungsten oxide film has a porous structure, the surface of
the as-deposited polymer-added tungsten oxide films is uniform and flat. Heat
treatment causes a remarkable difference in the films’ images.
Characterizations of tungsten oxide films were carried out with XRD and FTIR
measurements. Standard and mesoporous tungsten oxide films remain almost
amorphous up to 400 °C and crystallize at 400 °C. As-deposited and UV illumi-
nated films are amorphous. Polymer-added tungsten oxide films have changed the
crystal phases depending on the polymer matrix. Using polymer as a surfactant in
the tungsten oxide films may retard crystallization and created different crystal
phases. Calcined standard tungsten oxide films have triclinic crystalline structure.
Following spectral deconvolution analysis of both films, their different crystal
structures are presented in Fig. 12.8. PEG polymers are in linear and rigid rod form;
298 E.O. Zayim and A. Tabatabaei Mohseni

Fig. 12.9 Cyclic

Voltammetry (CV) of
tungsten oxide film treated by
UV illumination (WO3 under
UV) and polymer-added
tungsten oxide film treated by
UV illumination (PEG/WO3
under UV)

this may positively affect the speed and the rate of crystallization of the structures in
which PEG polymers are added.
All as-deposited films were then exposed to electrochemical and electrochromic
investigations using LiClO4 in propylene carbonate as an electrolyte. The CV of
tungsten oxide films are given in Fig. 12.9. The results demonstrated that the
polymer causes modifications of the electrochemical and structural properties of the
films.
It is observed that better CV results (charge insertion/extraction) were obtained
for mesoporous films as shown in the Fig. 12.9. Supposably using PEG polymer as
a surfactant increases the active surface area. These mesoporous films exhibit
superior high-rate ion-insertion performance in lithium liquid electrolyte as
expected.

12.2.2 Tungsten Oxide Based Nanofibers and Nanowires

One-dimensional (1D) nanostructures of tungsten trioxide include nanowires [18],

nanotubes [19], nanofibers [7], nanorods [20] and nanoribbons [21] which offer the
best morphology for most of the electro-functional devices [22].
Tungsten oxide nanomaterials can be produced via various techniques including
electrochemical etching [23], chemical vapor deposition (CVD) [24], hydrothermal
reaction [25] and etc.
Among these methods and techniques, electrospinning is one of the most widely
used processes for the production of nanofibers known since the 1930s. This
technique employs electrostatic forces for stretching the viscoelastic fluid [26].
Electrospinning offers a relatively simple and versatile method for generating
fibular mesostructures. This process involves the production of continuous
one-dimensional nanofibers under electrostatic force of the charges on the surface
of a liquid droplet in an electric field as strong as several kV. cm−1.
12 Structural and Optical Properties of Tungsten Oxide … 299

Here, method of producing nanofibers and nanowires of tungsten oxide via

electrospinning has been applied using two different precursors.
WO3 fibers can be produced by electrospinning technique using tungsten
hexachloride (WCl6) and tungsten (W) metallic powder as precursors. Surface and
structural characterizations were performed by SEM, XRD and FTIR to demon-
strate the crystallinity and structure of the fibers.
The electrospinning technique requires a viscose polymeric solution. This
organic/inorganic blend contains precursors and polymer ingredients and their
solvents. The sol-gel W metal powder and WCl6 precursor mixed with a 10 %
polyvinylpyrrolidone (PVP) solution in ethanol were loaded into a plastic syringe
after stirring for minimum 5 hours. PVP has been prefered because it is a suitable
capping reagent for various metal nanoparticles with low environmental and health
risks.

12.2.2.1 Tungsten Oxide Fibers with Metallic Tungsten Precursor

To prepare the organic/inorganic blend for the electrospinning process, 10 ml of

cold 30 % H2O2 solution was added to 0.7 g metallic tungsten powder. After
stirring for about 4 hours, the powder dissolved completely and a faint yellowish
solution (tungstatic acid) was obtained. Then 10 ml ethanol and 1.0 g PVP were
added into the above solution, followed by magnetic stirring for 2 hours. The
solution was then loaded into a plastic syringe connected to a 1 cm short needle
made of copper by a plastic pipe. The needle was connected to a high-voltages
supply that is capable of generating DC voltages up to 35 kV. In this experiment a
voltage of 15 kV was applied for electrospinning conducted in the air at room
temperature. Annealing electrospun fibers in air at 500 °C for 2 h completely
evaporates the solvents and polymer to attain calcined WO3 nanofibers.
Microstructural properties of tungsten oxide fibers were investigated by
FE-SEM. Homogenous nanofibers of tungsten oxide were prepared using metallic
tungsten precursor as seen in the SEM images presented in Fig. 12.10. The diameter
of the nanofibers are in the range of 112–184 nm. SEM images of calcined
nanofibers of tungsten oxide are shown in Fig. 12.11. The thickness of the fibers are
observed to be 35 % reduced that is between 42 and 152 nm.
FTIR analysis of as-prepared electrospun tungsten oxide nanofibers by a metallic
tungsten precursor is given in Fig. 12.12. Inorganic compounds have vibrational
bands mainly below 1200 cm−1. The 816 cm−1 band is assigned to the out of
stretching vibrations of W–O–W mode, when hydrogen is located at a coplanar
square of oxygen atoms. A relatively weak band at 1073 cm−1, which is assigned to
the plane deformational (bending) W–OH mode, was found. The peak at 1424 cm−1
is assigned to ν(OH) and δ(OH) in OH, W–O group. In the frequency range of 400–
1100 cm−1, the peak in 977 cm−1 (W=O terminal modes of surface grains) and W–
O–W bridging mode 902 cm−1 exist exhibiting the stretching vibrations of W–O
bounds [27]. In addition, the sharp peak in 1290 cm−1 is assigned to C–N stretches
of the aromatic amines group. Organic compounds usually have vibrational bands
300 E.O. Zayim and A. Tabatabaei Mohseni

Fig. 12.10 SEM images of electrospun tungsten oxide nanofibers by W/PVP, a 5 µm scale,
b 1 µm scale

Fig. 12.11 SEM images of calcined tungsten oxide nanofibers by W/PVP, a 5 µm scale, b 0.5 µm
scale

Fig. 12.12 FTIR of 0.07

1652
W/PVP blend
electrospun tungsten oxide Electrospun Nanofibers
nanofibers of W/PVP blend 0.06

0.05
Absorbance

90
34
12
14

0.04
816
0.03
3340 2950
977
0.02

0.01

0.00
4500 4000 3500 3000 2500 2000 1500 1000 500
-1
Wavenumber (cm )

mainly after 1500 cm−1. The sharp peak around 1652 cm−1 is assigned to C=O
stretch of α, β–unsaturated aldehydes the ketone functional group. The medium
peak at 2950 cm−1 reveals the C–H stretch from an alkane group and finally the
sharp and broad shoulder around 3340 cm−1 is assigned to O–H stretch, from H–
bonded alcohol the phenol functional group.
12 Structural and Optical Properties of Tungsten Oxide … 301

12.2.2.2 Tungsten Oxide Fibers with Tungsten Hexachloride

Precursor

Tungsten hexachloride (WCl6 ≥ 99.9 % trace metals basis) powder was dissolved in
ethanol (CH3CH2OH), then mixed with polyvinylpyrrolidone (PVP) where N,
N-dimethylformamide (DMF) was used as its solvent. The electrospinning process
was held at room temperature using 20 kV DC voltage while the distance between
the spinning nozzle and the collection plate was 15 cm, feeding rate was 6 μl/min
and substrate was a glass on aluminum foil.
SEM images of resultant electrospun nanofibers of the blend are presented in
Figs. 12.13 and 12.14. The diameters of the composite nanofibers are distributed in
the range of 190 and 350 nm with an average diameter of 270 nm.
To eliminate polymeric compounds from the film structure and obtain an inor-
ganic surface the nanofiber mat was calcined. After calcination at 500 °C for 1 h,
homogenous thin fibers in the range of 57 and 110 nm were obtained with an
average diameter of 85 nm as shown in Fig. 12.14. After the calcination process, the
average diameter of the nanofibers was reduced by nearly 70 %, proving that
organic compounds have been removed from the film structure and tungsten oxide
is the dominant remaining material in the surface. After calcination at 500 °C for
1 h, we obtained the thinner fibers as shown in SEM images.
FTIR analysis of as-prepared electrospun nanofibers of tungsten oxide using
WCl6/PVP blend is given in Fig. 12.15. The 679 cm−1 band is assigned to the out of
plane deformation W–O–W mode, when hydrogen is located at a coplanar square of
oxygen atoms, the 816 cm−1 peak is assigned to the out of plane stretching
vibrations of the W–O–W mode. A relatively weak band at 1062 cm−1, which is
assigned to the plane deformational (bending) W–OH mode, was found. This is due
to the creation of weakly bonded W–OH groups is formed in the as-deposited film.
In the frequency range from 400 to 1100 cm−1, a shoulder is observed around
979 cm−1 (W=O terminal modes of surface grains) and a W–O–W bridging mode at
902 cm−1 [27]. Organic compounds mostly have vibrational bands after 1500 cm−1.
The sharp peak around 1690 cm−1 is assigned to C=O stretch of α, β–unsaturated
aldehydes, the ketone functional group. The medium peak at 2960 cm−1 reveals the
C–H stretch from the alkane group and finally the sharp and broad shoulder around

Fig. 12.13 SEM images of electrospun nanofibers by WCl6/PVP, a 10 µm scale, b 1 µm scale

302 E.O. Zayim and A. Tabatabaei Mohseni

Fig. 12.14 SEM images of calcined nanofibers of tungsten oxide by WCl6/PVP blend

Fig. 12.15 FTIR of 1690

0.07 WCl 6 /PVP blend
electrospun tungsten oxide
Electrospun nanofibers
nanofibers by WCl6/PVP 0.06
blend
Absorbance (%)

0.05

0.04 1423
3330 816
0.03 2960

9
97
678
0.02

2
90
62
0.01

10
4500 4000 3500 3000 2500 2000 1500 1000 500
Wavenumber (cm-1)

3330 cm−1 is assigned to O–H stretch, from H–bonded alcohol, the phenol func-
tional group. In addition, the medium peak in 1423 is related to the C–H bend of
one of the alkane groups.
To eliminate polymeric compounds from the film structure, and obtain an
inorganic surface the nanofiber mat was calcined. After calcination at 500 °C for
1 h, homogenous thin fibers in the range of 57 and 110 nm were obtained with an
average diameter of 85 nm. After the calcination process, the average diameter of
the nanofibers was reduced by nearly 70 %, proving that organic compounds have
been removed from the film structure and tungsten oxide is the dominant remaining
material in the surface.
Table 12.1 compares the diameter variations of WO3 fibers obtained by two
different precursors explained in the latter section. The average values indicate that
as-prepared nanofibers of tungsten oxide polymeric hybrid are thinner for metallic
tungsten precursor (148 vs. 270 nm), but for calcined tungsten oxide nanofibers,
tungsten hexachloride precursor yields in thinner fibers and nanowires (83 vs.
100 nm).
12 Structural and Optical Properties of Tungsten Oxide … 303

Table 12.1 Diameter range of the nanofibers in different steps

Diameter As-prepared Calcined As-prepared Calcined
nanofibers by nanofibers by nanofibers by W nanofibers by W
WCl6 precursor WCl6 precursor metallic precursor metallic
(nm) (nm) (nm) precursor (nm)
Min 189 57 112 42
Max 350 110 184 158
Average 270 83 148 100

12.3 Chromogenic Properties and Applications

of Tungsten Oxide

Optically active thin film coatings can change their optical properties as a function
of external stimuli. These substances, recently named ‘chromogenics’, include both
inorganic and organic materials. Among various chromogenic property of films,
well known ones are electrochromic (EC), thermochromic, photochromic and
gaschromic. Materials with controllable light absorbance, transmittance or reflec-
tance possess great technical relevance, because of their high potential applications.
In the near future, optically active films may be utilized to regulate the throughput
radiation energy for windows in the buildings and cars, to maintain comfortable
lighting and temperature, in sunglasses, as an optically active filter, or in systems
with variable reflectance, as automobile rear-view mirrors, in sensors, in detectors,
in displays, as road signs, and so forth.
Tungsten oxide is the most investigated chromogenic material, especially used
for the preparation of EC devices due to its high stability and fully reversible
coloration ability.

12.3.1 Electrochromic Properties of Tungsten Oxide Films

The optical, electrochemical and structural properties of the deposited films were
examined in both liquid and solid nafion electrolytes. Herein, electrochromic
devices were fabricated by tungsten oxide active electrochromic layer prepared by
tungsten chloride precursor and nafion electrolytes. The use of a polymer elec-
trolyte, in contrast to commonly used liquid electrolytes, allows integration of the
electrode into fully solid state devices with the following configuration: ITO/WO3/
Nafion/ITO.
The colored and bleached state in nafion solution was represented in Fig. 12.16.
The potentiostatic measurements of produced films, were performed with the
generation of an I-t curve. The film was colored at −1 V and bleached at +1 V, with
each step being 60 s.
304 E.O. Zayim and A. Tabatabaei Mohseni

Fig. 12.16 Tungsten oxide electrodeposited film in nafion solution electrolytes a bleached state
and b colored state

The in situ transmittance versus wavelength curves for tungsten oxide elec-
trodeposited film by the WCl6 precursor in nafion solution is shown in Fig. 12.17.
The coloration and bleaching currents decay quickly for tungsten oxide elec-
trodeposited film using the WCl6 precursor. Additionality, in situ optical trans-
mittance spectra of tungsten oxide electrodeposited film by the WCl6 precursor was
taken, the transmittance spectra for tungsten oxide electrodeposited film by the

Fig. 12.17 In-situ optical

transmittance spectra of
tungsten oxide
electrodeposited film by the
WCl6 precursor in nafion
liquid electrolyte
12 Structural and Optical Properties of Tungsten Oxide … 305

WCl6 precursor exhibit high optical modulation. Transmittance values of the col-
ored tungsten oxide film is 5.9 but bleached tungsten oxide has a transmittance of
74.0 at 550 nm. Transmittance values of the colored and bleached tungsten oxide
films were obtained as 2.70 and 80.0 respectively at 680 nm.

12.3.2 Coloration Phenomena in WO3

A variety of excitation processes such as thermal, optical, electrical, ionizing

radiation, heating in vacuum, inert and reducing atmosphere, can be used for col-
oration of a material such as tungsten oxide. The spectrum of coloration is essen-
tially similar in all cases. Coloration is structure-sensitive, while the absorption
peak for amorphous WO3 films is around 1.2 eV, for crystalline WO3 films the
absorption peak is around 0.7 eV.
Coloration occurs most efficiently in highly disordered amorphous WO3–x films.
Only sub-stoichiometric oxidized tungsten oxide crystalline films (WO3) show
coloration (Film formation occurs at high temperature under a reducing atmosphere
such as vacuum, Ar, H2, etc.). The photochromic effect (PC) occurs when films are
irradiated with photon energies within the band gap (max at 3.6 eV) [28]. Ambient
conditions can also effect optical and electrical coloration. Optical coloration effi-
ciency increases in the presence of H2O. Although there is no electrical coloration
in vacuum, thermal, optical, electron and ion beam induced coloration is observed.

12.4 Conclusions

After four decades of extensive R&D on electrochromic (EC) properties of metal

oxides, WO3 still remains the leading EC-material. Major advances have been made
in the fabrication and characterization (structural, optical, and electrical) of
EC-materials and devices based on metal oxides. In this review, well-known
preparation techniques of nanostructured tungsten oxide are briefly introduced, their
similarities and differences according to the nanostructure of tungsten oxide pro-
duced are discussed. Two important factors, crystallinity of tungsten oxide film and
the surface morphology play an important role in chromogenic properties of
tungsten oxide nanomaterials.

References

1. X. Chi, Tungsten trioxide nanotubes with high sensitive and selective properties to acetone.
Sensors and Actuators B: Chemical 194, 33–37 (2014)
2. S. Novak, Effect of ingested tungsten oxide (WOx) nanofibers on digestive gland tissue of
porcellio scaber. Am. Chem. Soc. 47, 11284–11292 (2013)
306 E.O. Zayim and A. Tabatabaei Mohseni

3. J. Lee, Simple fabrication of flexible electrodes with high metal-oxide content electrospun
reduced tungsten oxidecarbon nanofibers for lithium ion battery applications. Nanoscale 6,
10147–10155 (2014)
4. W. Li, Nanostructured WO3 thin film as a new anode material for lithium-ion batteries. Appl.
Surf. Sci. 256, 2447–2452 (2010)
5. S. Cong, Y. Tian, Q. Li, Z. Zhao, F. Geng, Single-crystalline tungsten oxide quantum dots for
fast pseudocapacitor and electrochromic applications. Adv. Mater. 26, 4260–4267 (2014)
6. Y. Qin, Nanowire (nanorod) arrays-constructed tungsten oxide hierarchical structure and its
unique NO2-sensing performances. J. Alloys Compd. 615, 616–623 (2014)
7. T.A. Nguyen, S. Park, J.B. Kim, T.K. Kim, G.H. Seong, J. Choo, Y.S. Kim, Polycrystalline
tungsten oxide nanofibers for gas-sensing applications. Sens. Actuators B: Chem. 160(1), 549–
554 (2011)
8. C.V. Ramana, S. Utsunomiya, R.C. Ewing, C.M. Julien, U. Becker, Structural stability and
phase transitions in WO3 thin films. J Phys. Chem. B 110, 10430–10435 (2006)
9. X. Chang, S. Shibin, Z. Li, X. Xu, Y. Qiu, Assembly of tungsten oxide nanobundles and their
electrochromic properties. Elsevier, Appl. Surf. Sci. 257, 5726–5730 (2011)
10. J. Sungpanich, Fabrication of WO3 nanofibers by high voltage electrospinning. Mater. Lett.
65, 3000–3004 (2011)
11. k Bange, Colouration of tungsten oxide films: a model for optically active coatings. Elsevier,
Solar Energy Mater. Solar Cells 58, 1–131 (1999)
12. E.K.H. Salje, S. Rehmann, F. Pobell, D. Morris, K.S. Knight, T. Herrmannsdorfer, M.T. Dove,
Crystal structure and paramagnetic behaviour of e-WO3–x. J. Phys.: Condens. Matter 9, 6563–
6577 (1997)
13. R. Chatten, A.V. Chadwick, A. Rougier, P.J.D. Lindan, The Oxygen Vacancy in Crystal
Phases of WO3. J Phys. Chem. B 109, 3146–3156 (2005)
14. M.H. Kim, H.W. Choi, K.H. Kim, Properties of WO3–x electrochromic thin film prepared by
reactive sputtering with various post annealing temperatures. IOP Jpn. Soc. Appl. Phys. 52, 11
(2013)
15. H. Miyazaki, Electrical and optical properties of amorphous and crystalline nondoped WO3–x
films prepared by reactive RF sputtering. J. Appl. Phys. 47(9), 7295–7297 (2008)
16. E. Ozkan, S.H. Lee, P. Liu, C.E. Tracy, F.Z. Tepehan, J.R. Pitts, S.K. Deb, Electrochromic
and optical properties of mesoporous tungsten oxide films. J. Solid State Ionics 149, 139–146
(2002)
17. R. Vittal, H. Gomathi, K.J. Kim, Beneficial role of surfactants in electrochemistry and in the
modification of electrodes. Adv. Colloid Interface Sci. 119(1), 55–68 (2006)
18. W. Cheng, Y. Ju, P. Payamyar, D. Primc, J. Rao, C. Willa, M. Niederberger, Large-area
alignment of tungsten oxide nanowires over flat and patterned substrates for room-temperature
gas sensing. Angew. Chem. Int. Ed. 54(1), 340–344 (2015)
19. X. Chi, C. Liu, L. Liu, Y. Li, Z. Wang, X. Bo, C. Su, Tungsten trioxide nanotubes with high
sensitive and selective properties to acetone. Sens. Actuators B: Chem. 194, 33–37 (2014)
20. P.V. Tong, N.D. Hoa, N.V. Duy, N.V. Hieu, Micro-wheels composed of self-assembled
tungsten oxide nanorods for highly sensitive detection of low level toxic chlorine gas. RSC
Advances 5(32), 25204–25207 (2015)
21. M. Ahmadi, S. Sahoo, R. Younesi, A.P. Gaur, R.S. Katiyar, M.J. Guinel, WO3 nano-ribbons:
their phase transformation from tungstite (WO3·H2O) to tungsten oxide (WO3). J. Mater. Sci.
49(17), 5899–5909 (2014)
22. S. Salmaoui, F. Sediri, N. Gharbi, Characterization of h-WO3 nanorods synthesized by
hydrothermal process. Polyhedron 29(7), 1771–1775 (2010)
23. G. Gu, B. Zheng, W.Q. Han, S. Roth, J. Liu, Tungsten oxide nanowires on tungsten substrates.
Nano Lett. 2(8), 849–851 (2002)
24. T.D. Manning, I.P. Parkin, M.E. Pemble, D. Sheel, D. Vernardou, Intelligent window
coatings: atmospheric pressure chemical vapor deposition of tungsten-doped vanadium
dioxide. Chem. Mater. 16(4), 744–749 (2004)
12 Structural and Optical Properties of Tungsten Oxide … 307

25. H. Hassani, E. Marzbanrad, C. Zamani, B. Raissi, Effect of hydrothermal duration on synthesis

of WO3 nanorods. J. Mater. Sci.: Mater. Electron. 22(9), 1264–1268 (2011)
26. R. Sahay, V. Thavasi, S. Ramakrishna, Design modifications in electrospinning setup for
advanced applications. J. Nanomater. 17 (2011)
27. J. Díaz-Reyes, V. Dorantes-García, A. Pérez-Benítez, J.A. Balderas-López, Obtaining of films
of tungsten trioxide (WO3) by resistive heating of a tungsten filament. Superficies Vacio 21(2),
12–17 (2008)
28. S.K. Deb, Opportunities and challenges in science and technology of WO3 for electrochromic
and related applications. Sol. Energy Mater. Sol. Cells 92(2), 245–258 (2008)
Chapter 13
Electron Accumulation in InN Thin Films
and Nanowires

L. Colakerol Arslan and K.E. Smith

Abstract An overview on the electron accumulation layer on InN thin film and
nanowire surfaces is provided. The interactions between the valence and conduction
bands due to the narrow band gap and high electron density at the surface of these
materials have a big influence on the electronic structure and the device performance
of these materials. We first review the current understanding on the electron accu-
mulation on InN thin films, pointing out the role of defects and dislocations on the
unintentional n-type conductivity. Then we carry out detailed investigation on tuning
the surface charge properties of InN nanowires depending on the growth process.

13.1 Introduction

The narrow band gap semiconductors such as InAs, InSb, and InN are subject of
intense study because of their combined electrical and optical functionalities. Their
distinctive material characteristics such as high electron mobility, small effective
mass and strong spin-orbit coupling provide excellent charge transport properties
for applications in ultrahigh-speed nanophotonic devices. Among these materials,
InN has attracted huge attention recently because of its near-surface properties. The
low dissociation temperature of InN and the extremely high equilibrium vapor
pressure of nitrogen prevented the preparation of high quality single crystals,
leaving the optical, electrical and physical properties of this material difficult to

L. Colakerol Arslan (&)

Department of Physics, Gebze Technical University, 41400 Kocaeli, Turkey
e-mail: [email protected]
K.E. Smith
Department of Physics, Boston University, 02215 Boston, MA, USA
e-mail: [email protected]
K.E. Smith
School of Chemical Sciences, The MacDiarmid Institute for Advanced Materials
and Nanotechnology, The University of Auckland, 1142 Auckland, New Zealand

© Springer International Publishing Switzerland 2016 309

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_13
310 L. Colakerol Arslan and K.E. Smith

determine. Early InN samples, prepared by reactive RF sputtering of indium targets

with nitrogen [1], have a band gap value of 1.9 eV due to the presence of oxygen in
the films during growth [2]. Recently, the energy of the fundamental band gap has
undergone a remarkable revision from the previously accepted value of approxi-
mately 1.9 eV to the much lower value of approximately 0.65 eV as a consequence
of the development of epitaxial growth methods for high quality crystalline thin
films of InN [3–5]. The band gap of the InxGa1−xN ternary alloy system now spans
the near infrared (0.65 eV, x = 1) to the UV (3.4 eV, x = 0), enabling the entire
optical window to be encompassed by a single material system [6]. This makes
InxGa1−xN potentially suitable for applications such as high-brightness white light
emitting diodes and high-efficiency solar cells. In addition, mobilities over
3000 cm2/Vs have been measured at room temperature [7], leading to high per-
formance transistors operating at high frequencies up to the several THz range. The
use of InN based optoelectronic devices offers the potential of an
environmental-friendly red emitter with no toxic element, which may replace GaAs
based devices. In order to fully realize the potential of InN, a broader investigation
of the surface and interface properties was performed, since these will have a major
influence on the design of low dimensional devices.
InN is unusual in that there is strong experimental evidence for the existence of
an intrinsic electron accumulation layer near the surface of the thin films, where the
density of electrons is over an order of magnitude larger than in the bulk. The
existence of a large electron accumulation layer at all surface planes and the pos-
sibility of spin-splitting may also play an important role in electron spin transport
and spintronics and extends InN’s possible applications [8]. This electron accu-
mulation could also potentially be used for gas sensor applications and for the next
generation of THz radiation generators and sensors [9]. On the other hand, the high
electron density makes it difficult to achieve p-type doping in InN. Therefore,
understanding of the nature of this electron accumulation and surface electronic
properties is of a great importance in the design and the analysis of high perfor-
mance nitride-based electronic devices.
Adding up these entire qualities one can conclude that InN represents one of the
most versatile semiconductor material systems. However, due to the lack of
homoepitaxial substrates for nitride epitaxy the crystalline quality of planar films is
not perfect. Despite high defect densities, nitride devices are extraordinarily per-
forming. Recently, InN nanostructures in the forms of nanowires (NWs) have also
received great attention due to the high surface-to-volume ratio and enhanced
electron accumulation at the surface. NWs, which can be grown as single
nano-crystals, show fewer structural defects than planar films and therefore, are
expected to further improve the device quality. Especially, when the high surface
electron density is combined with high mobility, InN NWs become attractive
materials for the production of high electron mobility transistors operating at high
frequencies. In addition, InN NWs can also be grown on several different substrates,
including silicon, therefore the possibility of combining these highly lattice mis-
matched materials provides easy integration of these nanowires with the established
Si microelectronics technology.
13 Electron Accumulation in InN Thin Films and Nanowires 311

In this chapter, we have focused on the study of near surface electron accu-
mulation in InN thin films and NWs. We begin by exploring the various mecha-
nisms behind the origin of electron accumulation at InN surfaces. The accumulation
layer is discussed in terms of the bulk Fermi level (EF) lying below the pinned
surface EF, with a confining potential formed normal to surface due to the down-
ward band bending facilitated by donor type surface states or impurities. Then the
effect of these charge accumulation on the electronic properties of these materials
was investigated. The effect of growth properties and reduced dimensions on the EF
pinning mechanism on InN nanowire surfaces is also discussed. Mg-doped InN
NWs and the possibility to achieve p-type conductivity in InN NWs are presented.

13.2 Fermi Level Pinning

The pinning of the Fermi level (EF) in semiconductors is observed for a wide class
of physical phenomena, including interface formation, surface processes, chemical
impurity doping, and high-energy radiation effects. For free surfaces of semicon-
ductors, EF pinning is driven by sufficiently high density of surface states of both
donor and acceptor character [10, 11]. These surface states arise from the termi-
nation (or reconstruction) of the bulk crystal at the surface. They can also be caused
by impurities or defects. If there surface states exist in the gap, they pin the Fermi
level upward or downward depending on their position relative to charge neutrality
level (CNL) [12]. The charge neutrality at the surface is satisfied when the total
charge due to surface states, Qss, is compensated by an equal but opposite charge
near the surface region called space charge, Qsc. In the presence of ionized surface
states at semiconductor free surfaces, charge neutrality is ensured by the bending of
electronic bands due to the Fermi level being pinned at the surface. If the Fermi
level is located above the CNL, the surface states are negatively charged and a
depletion layer is created on the surface. If the electrons are filled to a point below
the CNL, the surface states are positively charged and this results in an accumu-
lation layer near the surface. The CNL allows a unique correspondence to be made
between the band bending at the semiconductor surface and the population of
surface states. For an intrinsic semiconductor, also called an undoped semicon-
ductor, with homogeneous surfaces, the Fermi level should coincide with the CNL
of the surface states.
Narrow band gap III–V semiconductors usually show an accumulation layer due
to pinning of the Fermi level above the conduction band edge. This electron
accumulation in these materials can be explained by the extremely low conduction
band minimum at the Γ point compared with at other points in k-space. Therefore,
the CNL is located above the CBM, leading to positive surface charge, Qss, as seen
in Fig. 13.1. To compensate this surface charge, a negative space charge builds up
on the surface and Fermi level is pinned below the CNL, resulting in an electron
accumulation layer. The observed electron accumulation at the surface of these
materials is due to the presence of positively charged donor-type surface states.
312 L. Colakerol Arslan and K.E. Smith

Fig. 13.1 Illustration of a E

surface accumulati on layer. ECNL
EF is located below the charge QSS EC
neutrality level, the surface is EF
positively charged forming an
QSC
accumulation layer

EV

N(E)

Similar change of surface potential and the corresponding charge accumulation

layer formation region occurs near surfaces of InN NWs. In this case, the nanowire
can be modeled by two distinct regions along the lateral dimension, including the
space charge region near the surface and the neutral region in the bulk.

13.3 Surface Electron Accumulation in InN Thin Films

There have been several reports regarding surface electron accumulation on InN
thin films. Evidence for the accumulation layer on InN thin films comes from high
resolution electron energy loss spectroscopy (HREELS) studies of clean InN sur-
faces [13, 14], which revealed a conduction band plasmon whose energy increased
as the incident electron energy was reduced, indicating a higher electron density
within 80 Å of the surface than in the bulk. Further evidence comes from sheet
carrier density measurements as a function of film thickness, capacitance-voltage
measurements, a photoemission study of defective InN coated with Ti [15, 16].
Recent tunneling spectroscopy experiments add further credence to the existence of
this layer, with features associated with electronic sub-bands being observed,
indicating that the electrons are in quantized states [17].
The most direct evidence of electron accumulation layer on InN is obtained from
the high resolution angle resolved photoelectron spectroscopy (ARPES) spectra
showing the energy region between EF and the top of the valence band, as shown in
Fig. 13.2. ARPES is one of the most direct methods to determine the energy dis-
persion of electronic states in solids. This technique measures the energy and the
angular distribution of electrons emitted from the surface of a material when it is
illuminated with monochromatic photons of sufficient energy. ARPES reveals two
discrete states, centered around the Г-point. The states were characterized by varying
the photon energy, and thus probing the band dispersion along kz (surface normal)
13 Electron Accumulation in InN Thin Films and Nanowires 313

Fig. 13.2 ARPES

photocurrent intensity map of EF
states within 3 eV of EF. 0,0
hυ = 69 eV and sample
temperature was 177 K. QW # 2
Sample was annealed to
300 °C in UHV for 30 min.
The false color intensity -0,5
reflects the photocurrent, with
lighter intensity indicating QW # 1

Binding energy (eV) relative to E F

higher current
-1,0 -0,2 0,0 0,2

VBM

-1,5

-2,0

-2,5

-0,2 -0,1 0,0 0,1 0,2

-1
wavevector, k || (Å )

and angle of emission. The negligible dispersion along kz and narrow range of
emission angles over which these features are observed, indicating that the emitting
states are highly localized in k space. The measured separation between the VBM
and EF is 1.2 eV for this surface, which is higher than the band gap of InN, con-
firming that EF is above the conduction band minimum (CBM). The energy dif-
ference between EF and the CBM for this surface is 0.65 eV. Therefore, these states
are the quantum well states in the accumulation layer in InN due to the quantization
of electrons trapped in the conduction band minimum. These states, although
derived from the conduction band, arise from the existence of a potential well
perpendicular to the film surface. Downward band bending forms a one-dimensional
(1D) potential well and the resulting two dimensional (2D) electron gas is quantized
along the surface normal. This band bending is the result of donor-type surface states
or N vacancies pinning EF above the CBM [15, 18]. Similar behavior is also
observed from other In containing III–V compounds such as InAs (EF = 0.36 eV)
[19], InSb (EF = 0.18 eV) [20], and InP (EF = 1.35 eV) [21].
314 L. Colakerol Arslan and K.E. Smith

The existence of 2D electron accumulation layer due to downward band bending

and the small value of the energy gap in InN lead to strong interaction of quantized
electron states and valence band states. Thus, the dispersion of the top of the
valence band in InN within the quantum well potential associated with the electron
accumulation layer is affected. Although InN is well known to be a direct gap
semiconductor in the bulk, intrinsic quantum well potential produces complicated
hole dispersion curves and a minimum in the dispersion of the top of the valence
band is measured at the Γ-point, as can be seen in Fig. 13.2. Although two dis-
persing heavy and light hole valence bands locates around the top of the valence
band, the intensity of the heavy hole valence band vanishes near zone center,
resulting in a camel-back dispersion of the top of the valence band [22].
There is a significant body of experimental evidence that InN is a direct gap
semiconductor, including a recent study of the bulk electronic structure using reso-
nant x-ray emission spectroscopy and state-of-the art hybrid density functional theory
[23]. Clear intermixing between the heavy-hole and light-hole valence bands in the
intrinsic quantum well potential is observed associated with the near-surface electron
accumulation layer; results in an inverted band structure, with the valence band
maximum lying away from the Brillouin zone center. Such structures have been
previously been deduced from earlier transport and tunneling experiments from
engineered single HgTe/Hg1−xCdxTe [24] and GaAs/AlAs quantum wells [25].
Theoretical modeling of electronic band structure of InN demonstrates that InN films
under biaxial strain presents similar inverted valence band structure [26]. Tensile
strain occurs during the early stages of growth on the InN films grown on GaN (0001)
buffer layer due to elastic deformation of the crystal to fill the gaps between InN
islands during coalescence [27, 28]. Chen et al. showed that InN films grown on GaN
buffer layer form InN islands although InN films grown InN buffer layer after nitri-
fication does not form island structures [29]. Since the sample showing an inverted
band structure was grown on InN buffer layer, don’t have such strain to cause valence
band mixing. Therefore intermixing between the heavy-hole and light-hole valence
bands in the intrinsic quantum well potential is observed associated with the
near-surface electron accumulation layer; results in an inverted band structure, with
the valence band maximum lying away from the Brillouin zone center.
The origin of the surface states responsible for the Fermi level pinning mech-
anism at InN surfaces is associated with donor-like impurities or crystal defects.
Due to nitrogen vacancies, dislocations, and incorporation of oxygen or hydrogen
in the bulk during crystal growth; even high quality InN layers grown by
plasma-assisted molecular beam epitaxy still show bulk free electron density of
1017 cm−3 [30–32]. According to models derived on the basis of single field
electrical Hall effect measurements, the positively charged N vacancies associated
with dislocations are considered to be the origin of unintentional n-type conduc-
tivity [33]. Moreover, surface reconstruction, surface adsorbates and formation of In
overlayers or droplets are also identified as factors that cause donor type surface
states [34, 35]. Models derived on the basis of single field electrical Hall effect
measurements favor positively charged N vacancies (V+N) as the major origin of
n-type conductivity.
13 Electron Accumulation in InN Thin Films and Nanowires 315

In fact, ARPES spectra obtained from InN that is treated with two cycles of 500 eV
Ar+ ion bombardment and annealing in UHV to 300 °C show that modifying the InN
surface leads to an increase in charge density in the conduction band [36]. As can be
seen in Fig. 13.3a, the subband energy levels are localized at 0.511 and 0.797 eV
below EF after sputtering and annealing, while they are located at 0.233 and
0.662 below EF for clean InN surface. The increase in the energy levels is directly
related to increase in the charge density at the bottom of the conduction band. It is
likely that the bombardment of InN surface with inert gas ion leads to N vacancies
since this method may leave binary or ternary nitrides in a non-stoichiometric state
[37, 38].
A series of momentum distribution curves (MDCs) extracted from Fig. 13.3a is
presented in Fig. 13.3b to emphasize the dispersion of the quantized states in the
conduction band. Dispersion of the bands is fitted assuming both parabolic (dashed
gray) and non-parabolic bands (solid black). The non-parabolic curve is fitted to the
outer band which is more resolved and has larger non-parabolicity effect because of
its higher energy and larger occupation probability. It can be seen clearly that the
dispersion of the states appreciably deviate from the parabolic dispersions. Rather,
it is linear away from the subband minima and the non-parabolic dispersion follows
the relation given by Kane’s two band k.p model for narrow band gap semicon-
ductors [39]. Note that non-parabolic single-band tight binding approach has also
been applied to model surface electron accumulation at InN surfaces [40]. In narrow

(a) (b)

0,00
EF

-0,25
Binding Energy (eV)

Intensity (arbt. units)

-0,50

-0,75

-1,00

-0,3 -0,2 -0,1 0,0 0,1 0,2 0,3 -0,3 -0,2 -0,1 0,0 0,1 0,2 0,3
-1
-1 k(Å )
Momentum (Å )

Fig. 13.3 a ARPES photocurrent intensity map of states within 1.2 eV of EF. hυ = 70 eV, and
sample temperature was 60 K. Sample was prepared by two cycles of 500 eV Ar+ ion
bombardment and annealing in UHV to 300 °C. The momentum direction is along ΓΣΜ in the
surface plane. b Momentum distribution curve derived dispersion of the conduction band subbands
extracted from the spectra on the left. The dispersion of the states are fitted by parabolic (dashed
gray) and non-parabolic 2D (solid black) based on Kane’s model [39] with effective masses
respectively 0.16m0 and 0.1m0
316 L. Colakerol Arslan and K.E. Smith

Fig. 13.4 The derived free electron concentration n0 as a function of the nanowire radius r in a
logarithmic scale. The line is a linear fit that gives a slope k = −2.3 ± 2. The inset shows the plot in
a linear scale, and the curve is given by n0 ∝ 1 = r2.3. Reprinted with permission from [50].
Copyright 2013, AIP Publishing LLC

band gap semiconductors, strong k.p interaction between the conduction band and
the valence bands results in strong non-parabolicity in the conduction band [41].
Similar non-parabolic dispersion of InN conduction band is also observed by other
groups by measuring the relation between the effective mass and carrier concen-
tration [42].
Much debate is ongoing about the origin of the Fermi level pinning whether it is
an intrinsic property or it occurs under certain reconstruction conditions. There are
recent reports on the absence of surface Fermi level pinning on non-polar surfaces,
which were cleaved under ultra-high vacuum conditions [43]. Theoretical calcula-
tions also suggest that surface charge accumulation would not be expected on
non-polar surfaces [44]. However, the universal nature of the electron accumulation
is found at both polar and nonpolar InN surfaces under ambient conditions due to
contamination or adatom adsorption. Various growth conditions or surface treatment
methods may affect the density and the microscopic nature of charge accumulation.
However, all InN surfaces exhibit a tendency toward electron accumulation, because
the bulk Fermi level is located below CNL level. This is confirmed by King et al.
who showed that neither the film polarity nor the growth parameters have any effect
on the Fermi level pinning mechanism [45]. This is in contrast to the interpretation of
recent Hall effect measurements, which suggested different surface sheet densities
for In- and N-polarity samples and the samples grown at N-rich and In-rich condi-
tions. However, theoretical and experimental investigations show that the variation
in the sheet density of the surface electron accumulation with buffer layer, polarity,
or growth conditions is related with the dislocation densities at the interface [46–48].
13 Electron Accumulation in InN Thin Films and Nanowires 317

Although this intrinsic band bending and electron accumulation may affect the
electronic properties near the surface region, they have relatively low influence on
the bulk electronic properties in semiconductor thin films. In semiconductor
nanowires, on the other hand, the dimensions of the NW are often on the order of
the Debye screening length, so charge accumulation layer has a larger impact on the
electronic properties due to high surface-to-volume ratio.

13.4 Surface Charge Accumulation on InN NWs

InN nanowires are highly interesting as they can exhibit considerable confinement
effects which provide them significantly different properties than InN thin films.
Although it is quite challenging to grow InN thin films with high crystalline quality
due to the lack of hemoepitaxial substrates, NWs can be form on various substrates
with lower defect densities than planar films. Numerous studies are focused on the
investigation of InN NW structure devices with improved quality and the possibility
of integration with silicon technology [49–51]. Moreover, surface space charge
layer on polar and non-polar surfaces of InN NWs renders them ideal candidates for
use in nanoscale sensor devices and infrared light-emitting diodes and lasers [52].
The thermodynamically stable phase of InN NWs is hexagonal wurtzite structure
with two possible orientations depending on the growth method. The MBE grown
nanowires are oriented along the [0001] polar direction, with their sidewalls being
nonpolar m-planes [49]; while the ones deposited by chemical vapor deposition
using the vapor-liquid-solid (VLS) growth mechanism are oriented along the
[11–20] direction, and the sidewalls are formed by the polar {0001} facets [53].
The strong ionicity of the metal-nitrogen bond and absence of inversion symmetry
leads to strong macroscopic polarization along the [0001] direction in the
III-nitrides at zero strain [54, 55]. If an external electric field is applied to the III–V
nitride lattice, crystal lattice will induce piezoelectric polarization to accommodate
the stress. The overall polarization strength of crystal affects the free electron
motion inside the crystal. Due to extraordinary surface-to-volume ratio and absence
of interface affects, polarity of the surface has a dramatic effect on surface Fermi
level pinning and therefore optical and electrical properties of InN nanowires,
although the surface Fermi level pinning was shown to be the same for a-plane and
for both polarities of c-plane InN.
The lateral surfaces of most of the InN NWs exhibit high density of accumulated
electrons due to the presence of positively charged donor-type defects. An accu-
mulation layer is formed due to Fermi level pinning on the sidewalls of NW
surface, leaving a region with free-electron concentration in the middle of the wire.
Extend of this region depends on the thickness of the wire. Several efforts have
been made to characterize and to control EF pinning and surface charge properties
of InN nanowires. To prove the presence of a surface accumulation layer, electrical
measurements were performed on InN nanowires and the resistance was displayed
as a function of wire radius. It is expected that the conductance should scale with
318 L. Colakerol Arslan and K.E. Smith

the cross sectional area of the wire (πr2nw) as the current flows through the bulk of a
wire, while the conductance should scale with πrnw as the current flows through the
surface of a wire. The data presented by Richter et al. showed an approximately
linear relation between the conductance and diameter for wires with a diameter
below 100 nm, suggesting that major contribution to conduction through InN
nanowires is coming from the surface accumulation layer [56]. A decreasing
resistivity with decreasing wire diameter is also obtained by others [57]. Another
supporting evidence for the existence of surface electron accumulation on InN NW
surface comes from the analysis of the slope of the conductance as a function of
temperature [56]. The resistance of InN NW decreases linearly with decreasing
temperature. This metallic like behavior indicates that the main conductance con-
tribution arises from the surface electron accumulation layer. The pinning of the
Fermi level 0.6–0.7 eV above the conduction band minimum at the surface of the
nanowires is determined by polymer electrolyte gating and three-dimensional
electrostatic modeling of charge distribution [58].
The existence of an electron accumulation layer on InN surface is also proven by
photoluminescence (PL) spectroscopy which has been a key technique to study
the optical processes in semiconductors with particular sensitivity to point defects.
The shape and peak position of the PL spectra of InN NWs grown under In rich
conditions with electron densities in the range of 1 × 1018 to 6 × 1018 cm−3 exhibit a
weak dependence on the laser intensity which can be characterized by the presence
a weak band filling effect due to photogenerated charge carriers [59]. The reason for
the weak excitation power dependence is a large broadening of the PL peak.
Moreover, a high energy tail exists in the PL spectrum and it extends to higher
energies with increasing temperature. The authors have correlated the observed
features in the PL spectra with fluctuations of the Fermi level along the wire,
relative to the conduction band edge and modelled based on an accumulation layer
at the nanowire surface.
Raman spectroscopy is also used as an effective method to investigate surface
properties of semiconductors as it is a nondestructive and no contact characteri-
zation tool. In most cases, Raman spectra taken perpendicular to c-axis of InN wires
display a strong uncoupled E1(LO) mode near the frequency of the A1(LO) phonon
in addition to bulk phonon modes of EH 2 (LO) and A1(LO). The physical origin of
this LO mode is thought to be the scattering of photons associated with the charge
density fluctuations. The change in the E1(LO) frequency with excitation wave-
length indicates the changes in the free carrier concentration through the wire
thickness due to the different probing depths. Furthermore, enhancement in the
relative intensity of E1(LO) mode is observed for the samples with morphologies
benefiting the scattering through the lateral walls [60]. Therefore this mode is
attributed to free electron accumulation on NW surfaces. A similar LO mode is also
observed in Raman spectrum of a single InN NW independently of the scattering
geometry, indicating the presence of high electron concentration at the wire sur-
faces [61].
Detailed investigation on the origin of an electron accumulation layer at InN NW
surface is performed by using an electrical nanoprobing technique in a scanning
13 Electron Accumulation in InN Thin Films and Nanowires 319

electron microscope environment [50]. It is observed that the derived free electron
density increases as n * r−2.3 shown in Fig. 13.4. The existence of high carrier
concentration in small-radii NWs is explained by the presence of surface states.
Considering the fact that the diameter of nanowires grown at low temperatures is
smaller than that of the ones grown at higher temperatures, this is an indication of
that nanowires with smaller radii have relatively higher carrier concentration due to
surface defects. Recent study on the surface charge properties of InN nanowires has
also revealed that the surface defects are mainly responsible for large carrier con-
centration and corresponding electron accumulation on InN surfaces [62]. In fact,
the surface charge concentration can be tuned by varying the nanowire morphology
through In seeding mediated growth approach [62].
InN NWs grown with the use of an in situ deposited In seeding layer under
nitrogen rich conditions have carrier concentrations of ∼1 × 1015−16 cm−3.
Moreover, the optical excitation power and temperature has a strong influence on
their PL [63, 64]. Such behavior is a clear indication of low carrier density and
absence or a negligible level of surface electron accumulation. The authors recorded
an emission peak at 0.64 eV (at 300 K) due to band-to-band recombination.
According to ARPES results, the Fermi level position of undoped InN NWs is
located 0.5 eV above the VBM, which would imply that the near-surface EF for InN
NWs lies well below the CBM, as shown in Fig. 13.5a. In contrast, the PL spectra

Fig. 13.5 a X-ray photoelectron spectroscopy of intrinsic and moderately Si-doped InN
nanowires. b and c show the carrier distributions and recombination processes in intrinsic and InN:
Si nanowires, respectively. The blue circles represent residual free charge carriers, and the dotted
red circles represent photogenerated charge carriers. Reprinted with permission from [63].
Copyright (2012) by the American Physical Society
320 L. Colakerol Arslan and K.E. Smith

of Si doped InN NWs show two major peaks at 0.65 and *0.75 eV. While the low
energy emission is associated with band-to-band carrier recombination in the rel-
atively low-doped bulk region, the high energy peak is associated with the emission
of Mahan exciton in the high-doped nanowire near-surface region. Such behavior is
confirmed by the pinning of the EF * 0.35 eV above the CBM. A schematic band
diagrams to illustrate the carrier recombination processes are shown in Fig. 13.5b, c
for intrinsic and Si-doped InN NWs, respectively. Photogenerated electron-hole
pairs are distributed nearly uniformly across the cross section of undoped InN NWs.
For InN:Si nanowires, however, a surface charge density results in a downward
band bending. Then photoexcited electrons will migrate to the surface along the
potential slope produced by band-bending, leading PL emissions from bulk states
and from surface trap states.
Werner et al. Showed that the surfaces of InN NWs are covered by a thin layer of
In2O3 according to the results of XPS. The shell conductivity was ascribed to the
accumulation layer located at the interface between InN and In2O3 layers formed
during exposure to air [57].

13.5 Control of Surface Electron Accumulation in InN

Nanowires

The band bending can be detrimental for technological applications because it

precludes external control of the electronic properties and the realization of p-type
conductivity in InN although it is essential for the fabrication of p-n junctions and
building the basis of all electronic devices. Growth process and the quality of NWs
have a major influence on the band bending mechanism.
The 2D electron accumulation level on the surfaces of intrinsically doped InN
nanowires can be controlled by varying the epitaxial growth process. In this section,
we will discuss the effects of growth parameters on the 2D electron accumulation
layer on InN NW surfaces. Similar to InN thin films, early InN NWs prepared by
solvo-thermal method, catalytic chemical vapor deposition, and vapor-liquid-solid
were poorly crystalline and the band gap value was about 1.9 eV [65–67]. Recent
improvements in its molecular beam epitaxial growth process have resulted in
superior quality InN nanowire structures with a band gap of *0.7 eV on various
substrates becoming available [62, 68, 69]. However, most of the MBE grown InN
NWs exhibit uncontrolled tapered morphologies and poor crystal quality because of
the low decomposition temperature of InN, high vapor pressure of nitrogen over
InN and the lack of lattice matched substrates. Mi and coworkers synthesized InN
NWs directly on silicon substrate with the utilization of in-situ deposited In seeding
layer [69]. The use of In seeding layer promotes the nucleation of InN and provides
the growth of non-tapered InN NWs with a well-defined hexagonal structure along
[0001] direction. Figure 13.6 shows the SEM images of InN nanowires grown at
different temperatures. Optimum substrate temperature of 480 °C results in NWs
13 Electron Accumulation in InN Thin Films and Nanowires 321

Fig. 13.6 The role of surface defects on InN nanowire quality and the near-surface Fermi-level.
a and b SEM images of nontapered and tapered InN nanowires, respectively. c The XPS spectra
for nontapered and tapered InN nanowires measured from the sidewalls. Reprinted with
permission from [70]. Copyright (2014) by John Wiley and Sons, Inc.

with a uniform diameter as well as smooth hexagonal facets. Lower growth tem-
peratures result in a strong tapering of the NW. The InN NWs synthesized at
optimal substrate temperature have a rather low carrier concentration of low-to-mid
1013 cm−3 and high electron mobility of 12,000 cm2/Vs [50]. The valence band
spectrum taken from the tapered NWs sidewalls shows that EF is located about
*0.15 eV below the CBM (see Fig. 13.6c) The EF of non-tapered InN nanowires,
on the other hand, is positioned *0.5 eV above the VBM, which implies that the
surface Fermi level is not pinned above the conduction band. This was a major step
toward realization of p-type doping.
Another promising evidence for p-type conductivity in InN nanowires comes
from transport measurements. Due to the large surface to volume ratio in NWs, the
surface has an even higher impact on electrical transport properties. The electrical
properties (I-V) of as fabricated single InN nanowire is characterized a power-law
dependence of the type I ∝ Vβ, with good ohmic behavior (β * 1) at lower biases,
higher order response at larger biases, β > 2, associated with the trap-filled space
charge limited conductivity due to the presence of charge traps at high bias. It is
well known that when a dc voltage is applied to insulators or nearly insulating
materials, space charge is formed. Therefore, such non-linear behavior implies that
InN NWs have extremely low background concentration, i.e., the EF of those
non-tapered InN nanowires is located below the conduction band edge.
Following the observation of the lack of surface electron accumulation on
cleaved InN surfaces and theoretical calculations on the surface states of InN which
322 L. Colakerol Arslan and K.E. Smith

Fig. 13.7 The PL spectra of Mg-doped InN nanowires measured at 7 K under a 9 mW optical
excitation. The PL spectrum of nondoped InN nanowires measured at 20 K under the same power
is also shown for a comparison. The spectra were normalized by the PL peak intensity of the
nondoped InN nanowires and were shifted for display purpose. The dotted green curve shows the
PL spectrum measured under an excitation of 200 μW from the Mg-doped sample with a Mg cell
temperature of 190 °C; and the inset shows the PL peak intensity ratio of PLPL over PH PL as a
function of the excitation power. Reprinted with permission from [78]. Copyright (2013) American
Chemical Society

indicate that the Fermi-level can be unpinned on nonpolar planes, there has been
intense study in the production of p-type doping in InN [70–76]. Since Mg has been
proven to be an effective p-type dopant in GaN [77], Mg is also considered to be the
most suitable candidate for obtaining p-type InN. However most studies did not
confirm the presence of hole carriers in Mg doped InN because the surface potential
originated by the depletion layer prevents the electric field from penetrating into the
bulk of the film transport measurements are mostly dominated by surface charge
layer and could not be used to determine bulk carrier density. Nevertheless, some
evidence of the p-type conductivity in Mg doped InN NWs have been provided by
PL and ARPES measurements [78]. Figure 13.7 shows the PL spectra of Mg-doped
InN nanowires exhibiting non-tapered surface morphology measured at 7 K along
with the PL spectrum of nondoped InN nanowire measured at 20 K. While the PL
spectrum of the nondoped InN NWs is dominated by a single peak at 0.69 eV, PL
spectrum of Mg doped InN NWs contains two emission peaks at energies
ELPL * 0.61 eV and EH PL * 0.67 eV. With the increase of Mg doping, only PL
L

peak can be observed and it is redshifted due to band gap renormalization.

Enhanced defect incorporation leads to disappearance of high energy PLH peak. To
further investigate the origin of these two emission peaks, excitation power
dependence of PL spectra was examined, as shown in the inset of Fig. 13.7. At low
excitation power, low energy peak is dominant. Since holes at the Mg-acceptor
13 Electron Accumulation in InN Thin Films and Nanowires 323

energy level and electrons in conduction band level participate in radiation

recombination, this lower energy peak is ascribed to Mg-dopant associated acceptor
energy level. The intensity of high energy peak, on the other hand, increases with
the excitation power indicating that it is originating from valence to conduction
band transitions in InN NWs.
ARPES measurements were performed at a grazing incidence on the samples
with various Mg doping level to determine the position of the Fermi level at the
surface as a function of bulk carrier concentration [78]. No states associated with
the surface electron accumulation layer observed around the leading edges of the
valence band photoemission spectra for high Mg-doped InN NWs. For moderate
Mg concentrations (TMg < 220°), the Fermi level is located *0.4 eV above the
VBM independent of the doping level. The surface Fermi level positions for high
Mg concentrations (TMG > 230°) monotonically decrease with increasing Mg
density and reaches to the level of 0.1 eV above the VBM at highest concentration,
indicating the p-type doping of InN NWs. Because NW growth temperature is
much higher than the Mg effusion cell temperature, it is quite challenging to
incorporate Mg atoms around the surface region. However, at elevated Mg doping
levels, enhanced acceptor incorporation in the near-surface region can balance to a
certain extent the large surface desorption of Mg atoms during the epitaxial growth
process.

13.6 Conclusions

This chapter is a review of experimental progress that has been achieved in

understanding and controlling the surface electron accumulation in InN thin films
and NWs. The existence of the intrinsic electron accumulation layer is explained in
terms of a particularly low Γ-point conduction band minimum in wurtzite InN. As a
result, the surface Fermi level is pinned high in the conduction band due to excess
charge at the surface created by impurities, dislocations or donor-type surface
states. Although nearly all InN thin films exhibit a strong charge accumulation, the
degree of charge accumulation in InN NWs depends on the growth mechanism and
specific surface conditions. In fact, the absence of an accumulation layer on highly
crystalline InN NWs raises the question if the Fermi level pinning is an intrinsic
property of InN or if it only occurs due to growth related defects or surface
reconstruction.

Acknowledgments Thank Asst. Prof. L.F.J. Piper and Prof. T.D. Moustakas for their contribu-
tions to this work.
324 L. Colakerol Arslan and K.E. Smith

References

1. A.G. Bhuiyan, A. Hashimoto, A. Yamamoto, J. Appl. Phys. 94, 2779–2808 (2003)

2. T.L. Tansley, C.P. Foley, J. Appl. Phys. 59, 3241–3244 (1986)
3. V.Y. Davydov, A.A. Klochikhin, R.P. Seisyan, V.V. Emtsev, S.V. Ivanov, F. Bechstedt,
J. Furthmuller, H. Harima, V. Mudryi, J. Aderhold, O. Semchinova, J. Graul, Phys. Status
Solidi B 229, R1–R3 (2002)
4. J. Wu, W. Walukiewicz, K.M. Yu, J.W. Ager, E.E. Haller, H. Lu, W.J. Schaff, Y. Saito,
Y. Nanishi, Appl. Phys. Lett. 80, 3967–3969 (2002)
5. T. Matsuoka, H. Okamoto, M. Nakao, H. Harima, E. Kurimoto, Appl. Phys. Lett. 81,
1246–1248 (2002)
6. J. Wu, W. Walukiewicz, K.M. Yu, J.W. Ager, E.E. Haller, H. Lu, W.J. Schaff, Appl. Phys.
Lett. 80, 4741–4743 (2002)
7. X.Q. Wang, S.T. Liu, N. Ma, L. Feng, G. Chen, F.J. Xu, N. Tang, S. Huang, K.J. Chen,
S.Q. Zhou, B. Shen, Appl. Phys. Express 5, 015502 (2012)
8. C.M. Yin, H.T. Yuan, X.Q. Wang, S.T. Liu, S. Zhang, N. Tang, F.J. Xu, Z.Y. Chen,
H. Shimotani, Y. Iwasa, Y.H. Chen, W.K. Ge, B. Shen, Nano Lett. 13, 2024–2029 (2013)
9. H. Ahn, Y.J. Yeh, Y.L. Hong, S. Gwo, Appl. Phys. Lett. 95, 232104 (2009)
10. H. Lüth, Surfaces and Interfaces of Solids (Springer, Berlin, 1993)
11. P.Y. Yu, M. Cardona, Fundamentals of Semiconductors (Springer-Verlag, Berlin, 1996)
12. R.T. Tung, Mat Sci Eng R 35, 1–138 (2001)
13. I. Mahboob, T.D. Veal, C.F. McConville, H. Lu, W.J. Schaff, Phys. Rev. Lett. 92, 036804
(2004)
14. I. Mahboob, T.D. Veal, L.F.J. Piper, C.F. McConville, H. Lu, W.J. Schaff, J. Furthmuller,
F. Bechstedt, Phys. Rev. B 69, 201307(R) (2004)
15. H. Lu, W.J. Schaff, L.F. Eastman, C.E. Stutz, Appl. Phys. Lett. 82, 1736–1738 (2003)
16. K.A. Rickert, A.B. Ellis, F.J. Himpsel, H. Lu, W. Schaff, J.M. Redwing, F. Dwikusuma,
T.F. Kuech, Appl. Phys. Lett. 82, 3254–3256 (2003)
17. T.D. Veal, L.F.J. Piper, M.R. Phillips, M.H. Zareie, H. Lu, W.J. Schaff, C.F. McConville,
Phys. Status Solidi A 203, 85–92 (2006)
18. C. Stampfl, C.G. Van de Walle, D. Vogel, P. Krüger, J Pollmann Phys. Rev. B 61, 7846–7849
(2000)
19. L.Ö. Olsson, C.B.M. Andersson, M.C. Håkansson, J. Kanski, L. Ilver, U.O. Karlsson, Phys.
Rev. Lett. 76, 3626–3629 (1996)
20. V.Y. Aristov, V.M. Zhilin, C. Grupp, A. Taleb-Ibrahimi, H.J. Kim, P.S. Mangat,
P. Soukiassian, G. Le Lay, Appl. Surf. Sci. 166, 263–267 (2000)
21. G. Abstreiter, R. Huber, G. Trankle, B. Vinter, Solid State Commun. 47, 651–654 (1983)
22. L. Colakerol, L.F.J. Piper, A. Fedorov, T.C. Chen, T.D. Moustakas, K.E. Smith, EPL
(Europhysics Letters) 83, 47003 (2008)
23. L.F.J. Piper, L. Colakerol, T. Learmonth, P.A. Glans, K.E. Smith, F. Fuchs, J. Furthmueller,
F. Bechstedt, T.C. Chen, T.D. Moustakas, J.H. Guo, Phys. Rev. B 76, 245204 (2007)
24. K. Ortner, X.C. Zhang, A. Pfeuffer-Jeschke, C.R. Becker, G. Landwehr, L.W. Molenkamp,
Phys. Rev. B 66, 075322 (2002)
25. R.K. Hayden, D.K. Maude, L. Eaves, E.C. Valadares, M. Henini, F.W. Sheard, O.H. Hughes,
J.C. Portal, L. Cury, Phys. Rev. Lett. 66, 1749–1752 (1991)
26. J. Bhattacharyya, S. Ghosh, Physica status solidia (a) 204, 439–446 (2007)
27. A. Delimitis, P. Komninou, G.P. Dimitrakopulos, T. Kehagias, J. Kioseoglou, T. Karakostas,
G. Nouet, Appl. Phys. Lett. 90, 061920 (2007)
28. T. Bottcher, S. Einfeldt, S. Figge, R. Chierchia, H. Heinke, D. Hommel, J.S. Speck, Appl.
Phys. Lett. 78, 1976–1978 (2001)
29. T.C.P. Chen, C. Thomidis, J. Abell, W. Li, T.D. Moustakas, J. Cryst. Growth 288, 254–260
(2006)
13 Electron Accumulation in InN Thin Films and Nanowires 325

30. V. Darakchieva, K. Lorenz, N.P. Barradas, E. Alves, B. Monemar, M. Schubert, N. Franco,

C.L. Hsiao, L.C. Chen, W.J. Schaff, L.W. Tu, T. Yamaguchi, Y. Nanishi, Appl. Phys. Lett. 96,
081907 (2010)
31. E. Kalesaki, J. Kioseoglou, L. Lymperakis, P. Komninou, T. Karakostas, Appl. Phys. Lett. 98,
072103 (2011)
32. N. Miller, E.E. Haller, G. Koblmuller, C. Gallinat, J.S. Speck, W.J. Schaff, M.E. Hawkridge,
K.M. Yu, J.W. Ager, Phys. Rev. B 84, 075315 (2011)
33. V. Cimalla, V. Lebedev, F.M. Morales, R. Goldhahn, O. Ambacher, Appl. Phys. Lett. 89,
172109 (2006)
34. T. Nagata, G. Koblmuller, O. Bierwagen, C.S. Gallinat, J.S. Speck, Appl. Phys. Lett. 95,
132104 (2009)
35. D. Segev, C.G. Van de Walle, Europhys. Lett. 76, 305–311 (2006)
36. L. Colakerol, T.D. Veal, H.K. Jeong, L. Plucinski, A. DeMasi, T. Learmonth, P.A. Glans,
S.C. Wang, Y.F. Zhang, L.F.J. Piper, P.H. Jefferson, A. Fedorov, T.C. Chen, T.D. Moustakas,
C.F. McConville, K.E. Smith, Phys. Rev. Lett. 97, 237601 (2006)
37. Y.H. Lai, C.T. Yeh, J.M. Hwang, H.L. Hwang, C.T. Chen, W.H. Hung, J. Phys. Chem. B 105,
10029–10036 (2001)
38. S. Krischok, V. Yanev, O. Balykov, M. Himmerlich, J.A. Schaefer, R. Kosiba, G. Ecke, I.
Cimalla, V. Cimalla, O. Ambacher, H. Lu, W.J. Schaff, L.F. Eastman, Surf. Sci. 566, 849–855
(2004)
39. E.O. Kane, J. Phys. Chem. Solids 1, 249–261 (1957)
40. K.H. Bevan, M.S. Hossain, Semicond. Sci. Tech. 27, 105029 (2012)
41. B.R. Nag, Springer, Berlin, (1980)
42. W. Walukiewicz, J.W. Ager, K.M. Yu, Z. Liliental-Weber, J. Wu, S.X. Li, R.E. Jones,
J.D. Denlinger, J. Phys. D Appl. Phys. 39, R83–R99 (2006)
43. C.L. Wu, H.M. Lee, C.T. Kuo, C.H. Chen, S. Gwo, Phys. Rev. Lett. 101, 106803 (2008)
44. C.G. Van de Walle, D. Segev, J. Appl. Phys. 101, 081704 (2007)
45. P.D.C. King, T.D. Veal, C.S. Gallinat, G. Koblmuller, L.R. Bailey, J.S. Speck,
C.F. McConville, J. Appl. Phys. 104, 103703 (2008)
46. Y. Ishitani, M. Fujiwara, X. Wang, S.B. Che, A. Yoshikawa, Appl Phys Lett, 92 (2008)
47. G.F. Brown, J.W. Ager, W. Walukiewicz, W.J. Schaff, J. Wu, Appl. Phys. Lett. 93, 262105
(2008)
48. V. Lebedev, V. Cimalla, T. Baumann, O. Ambacher, F.M. Morales, J.G. Lozano, D. Gonzalez,
J. Appl. Phys. 100, 094903 (2006)
49. T. Stoica, R. Meijers, R. Calarco, T. Richter, H. Luth, J. Cryst. Growth 290, 241–247 (2006)
50. S. Zhao, O. Salehzadeh, S. Alagha, K.L. Kavanagh, S.P. Watkins, Z. Mi, Appl. Phys. Lett.
102, 073102 (2013)
51. B.H. Le, S.R. Zhao, N.H. Tran, Z.T. Mi, Appl. Phys. Lett. 105, 231124 (2014)
52. E. Calleja, J. Grandal, M.A. Sanchez-Garcia, M. Niebelschutz, V. Cimalla, O. Ambacher,
Appl. Phys. Lett. 90, 262110 (2007)
53. R.S. Chen, T.H. Yang, H.Y. Chen, L.C. Chen, K.H. Chen, Y.J. Yang, C.H. Su, C.R. Lin,
Nanotechnology 22, 425702 (2011)
54. O. Ambacher, J. Phys. D Appl. Phys. 31, 2653–2710 (1998)
55. F. Bernardini, V. Fiorentini, D. Vanderbilt, Phys. Rev. B 56, 10024–10027 (1997)
56. T. Richter, H. Luth, T. Schapers, R. Meijers, K. Jeganathan, S.E. Hernandez, R. Calarco,
M. Marso, Nanotechnology 20, 405206 (2009)
57. F. Werner, F. Limbach, M. Carsten, C. Denker, J. Malindretos, A. Rizzi, Nano Lett. 9,
1567–1571 (2009)
58. D.R. Khanal, W. Walukiewicz, J. Grandal, E. Calleja, J. Wu, Appl. Phys. Lett. 95, 173114
(2009)
59. T. Stoica, R.J. Meijers, R. Calarco, T. Richter, E. Sutter, H. Luth, Nano Lett. 6, 1541–1547
(2006)
60. S. Lazic, E. Gallardo, J.M. Calleja, F. Agullo-Rueda, J. Grandal, M.A. Sanchez-Garcia,
E. Calleja, E. Luna, A. Trampert, Phys. Rev. B 76, 205319 (2007)
326 L. Colakerol Arslan and K.E. Smith

61. E.O. Schafer-Nolte, T. Stoica, T. Gotschke, F.A. Limbach, E. Sutter, P. Sutter,

D. Grutzmacher, R. Calarco, Nanotechnology 21 (2010)
62. S. Zhao, S. Fathololoumi, K.H. Bevan, D.P. Liu, M.G. Kibria, Q. Li, G.T. Wang, H. Guo,
Z. Mi, Nano Lett. 12, 2877–2882 (2012)
63. S. Zhao, Z. Mi, M.G. Kibria, Q. Li, G.T. Wang, Phys. Rev. B 85, 245313 (2012)
64. S.R. Zhao, Z.T. Mi, B.H. Le, Gallium Nitride Materials and Devices IX 8986, 898617 (2014)
65. Y.J. Bai, Z.G. Liu, X.G. Xu, D.L. Cui, X.P. Hao, X. Feng, Q.L. Wang, J. Cryst. Growth 241,
189–192 (2002)
66. Z.H. Lan, W.M. Wang, C.L. Sun, S.C. Shi, C.W. Hsu, T.T. Chen, K.H. Chen, C.C. Chen,
Y.F. Chen, L.C. Chen, J. Cryst. Growth 269, 87–94 (2004)
67. J. Zhang, L. Zhang, X.S. Peng, X.F. Wang, J. Mater. Chem. 12, 802–804 (2002)
68. X.Q. Wang, S.B. Che, Y. Ishitani, A. Yoshikawa, Physica Status Solidi C—Curr. Topics Solid
State Phys. 3(6), 1561–1565 (2006)
69. Y.L. Chang, F. Li, A. Fatehi, Z.T. Mi, Nanotechnology 20, 345203 (2009)
70. J.H. Chai, T.H. Myers, Y.W. Song, R.J. Reeves, W.M. Linhart, R.J.H. Morris, T.D. Veal,
M.G. Dowsett, C.F. McConville, S.M. Durbin, J. Vac. Sci. Technol. B 30, 02b124 (2012)
71. J.H. Chai, Y.W. Song, R.J. Reeves, S.M. Durbin, Phys. Status Solidi A 209, 95–99 (2012)
72. M.A. Mayer, S. Choi, O. Bierwagen, H.M. Smith, E.E. Haller, J.S. Speck, W. Walukiewicz,
J. Appl. Phys. 110, 123707 (2011)
73. I. Wilke, Ultrafast Phenom. Semicond. Nanostruct. Mater. XIV, 76001 (2010)
74. J.W. Ager, N. Miller, R.E. Jones, K.M. Yu, J. Wu, W.J. Schaff, W. Walukiewicz, Physica
Status Solidi B-Basic Solid State Phys. 245, 873–877 (2008)
75. P.D.C. King, T.D. Veal, P.H. Jefferson, C.F. McConville, H. Lu, W.J. Schaff, Phys. Rev. B 75,
115312 (2007)
76. J.W. Ager, K.M. Yu, R.E. Jones, D.M. Yamaguchi, S.X. Li, W. Walukiewicz, E.E. Haller,
H. Lu, W.J. Schaff, AIP Conf. Proc. 893, 343–344 (2007)
77. K.M. Tracy, W.J. Mecouch, R.F. Davis, R.J. Nemanich, J. Appl. Phys. 94, 3163–3172 (2003)
78. S. Zhao, B.H. Le, D.P. Liu, X.D. Liu, M.G. Kibria, T. Szkopek, H. Guo, Z. Mi, Nano Lett. 13,
5509–5513 (2013)
Chapter 14
Optical and Structural Properties
of Quantum Dots

M.H. Yükselici, A. Aşıkoğlu Bozkurt, Ç. Allahverdi, Z. Nassar,

D. Bulut, B. Can Ömür, M.K. Torun and A.T. İnce

Abstract We report (i) thickness dependent evolution of structural disorder, strain

on crystalline planes and grain size in chemical bath deposited (CBD) CdS thin
films studied through a combinative evaluation of the results of optical absorption,
Raman spectroscopies, X-Ray diffraction (XRD) and Scanning Electron
Microscopy (SEM). (ii) refer briefly to CdSexS1-x nanocrystals in liquid and
(iii) address quantum size effect in CdSexS1-x quantum dots embedded in glass
studied through steady state photoluminescence spectroscopy. The asymptotic
optical absorption edge is red shifted while the long wavelength tail narrows with
increasing thickness which is proportional to deposition time. We employ effective
mass theory under quantum size effect to estimate average grain size from the
energetic position of asymptotic optical absorption edge. The long wavelength tail
optical absorption is due presumably to the micro-electric field induced by crys-
talline defects. The transmission probability through the potential energy barrier
created by micro-electric field is calculated with the help of WKB (Wentzel,
Kramers, Brillouin) approximation. We conclude that as the deposition time
increases from 10 to 150 min, the average grain radius changes by 2 nm, Urbach
energy and the electric micro-field decrease from 600 to 400 meV and 2240–
820 kV/mm respectively. The shift in XRD pattern shows that the compressive
strain decreases with growth. The Raman LO1 vibrational mode display an increase
up to 22 min of deposition time and then a decrease.

M.H. Yükselici (&)
A. Aşıkoğlu Bozkurt
Z. Nassar
B. Can Ömür

Faculty of Science and Letters, Department of Physics, Yıldız Technical University,
Davutpaşa Campus, Esenler, Davutpaşa Mah., Davutpaşa Caddesi, 34220 Istanbul, Turkey
e-mail: [email protected]
Ç. Allahverdi
Electrical-Electronics Engineering Department, Faculty of Engineering,
Toros University, 45 Evler Campus/Bahçelievler Mahallesi,
16. Street, No: 1/7 Yenisehir, 33140 Mersin, Turkey
D. Bulut
M.K. Torun
A.T. İnce
Department of Physics, İnönüMah. Kayışdağı Cad., Yeditepe University, 26
AğustosYerleşimi, 34755 Ataşehir, İstanbul, Turkey

© Springer International Publishing Switzerland 2016 327

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_14
328 M.H. Yükselici et al.

14.1 Introduction

Cd chalcogenide compounds of II–VI semiconductor nanocrystals (quantum dots)

are of special interest due to their technological applications especially in solar
cells, opto-electronics, biotechnology, medicine and light emitting diodes. When
the size of bulk semiconductor is reduced to nanometer scales, the crystal gains new
optical and electronic properties. Energy levels are quantized as in atoms and the
energy difference between levels widens with the inverse of the size squared. The
efficiency for the luminescence emission of the levels increases. The techniques for
obtaining II–VI nanocrystals with good luminescent characteristics and uniform
size distribution include colloidal synthesis of nanocrystals from precursors at the
presence of capping agents or stabilizers in order to achieve the desired
size-dependent optical properties and solid phase precipitation of the crystallites in
glass matrix through nucleation followed by diffusion limited growth and then
Oswalt ripening among others. II–VI thin films deposited by various techniques
such as solution growth and thermal evaporation possess optical and structural
properties which might be explained with the help of quantum size effect. In this
chapter we firstly review our research on (i) CdSexS1−x nanocrystals which have
been extensively studied model system and (ii) the most proper nanocrystal of
group II–VI semiconductors which take an advantage of strong quantum size effect
that is, CdTe of band gap energy of 1.5 eV (850 nm) and thirdly report
(iii) thickness dependent evolution of structural disorder, strain on crystalline planes
and grain size in chemical bath deposited CdS thin films. The asymptotic optical
absorption edge is red shifted while the long wavelength tail narrows with
increasing thickness which is proportional to deposition time. We employ effective
mass theory under quantum size effect to estimate average grain size from the
energetic position of asymptotic optical absorption edge. The long wavelength tail
optical absorption is due presumably to the micro-electric field induced by crys-
talline defects. The transmission probability through the potential energy barrier
created by micro-electric field is calculated with the help of WKB approximation.
We conclude that as the deposition time increases from 10 to 150 min, the average
grain radius changes by 2 nm, Urbach energy and the electric micro-field decrease
from 600 to 400 meV and 2240 to 820 kV/mm respectively.

14.2 CdSexS1−x Nanocrystals

Synthesis of binary and ternary II–VI group semiconductor nanocrystal systems

which have convenient size, size distribution, shape and composition in liquid media
is highly important with respect to recent technological applications. For instance,
the II–VI group semiconductor nanocrystals which have sufficient quantum effi-
ciency are used in active layers of light emission diodes (LEDs) in order to improve
color purity and color variety of the diodes [1]. Semiconductor nanocrystals can be
14 Optical and Structural Properties of Quantum Dots 329

(a) (b) (c) (d)

Photoluminescence Intensity (a.u.)

500

400

300

200

100

0
475 500 525 550 575 600 625 650 675 700
Wavelength (nm)

Fig. 14.1 CdSe nanocrystals dispersed in toluene. Their appearance under daylight (a) and under
*366 nm UV light (b). HRTEM image (c) and photoluminescence spectrum (d) of these
nanocrystals. TEM scale of 5 nm is given at the bottom left corner in (c)

formed and grown in aqueous and non-aqueous liquids with methods of

microemulsion [2], chemical reduction [3], cold and hot injection [4] etc. As shown
in Fig. 14.1a, CdSe nanocrystals growth in non-coordinating octadecene with hot
injection method and then dispersed into toluene are shown. These nanocrystals
whose surfaces are modified with stearic acid (C17H35COOH) molecules can be kept
in toluene longer than one year. Their appearance under *366 nm UV light is
shown in Fig. 14.1b. It can be deduced that these nanocrystals are nearly spherical
and have diameter of *3 nm from their high resolution transmission electron
(HRTEM) image given at Fig. 14.1c. Measured photoluminescence spectrum of
these synthesized nanocrystals is shown in Fig. 14.1d.

14.3 Investigation of Raman Spectroscopy for CdTe

Thin Film

Four CdTe thin film samples are produced by evaporating CdTe compound onto
glass substrates by using physical vapor deposition (PVD) technique. The samples
are investigated structurally and optically by Raman scattering. Relative to the bulk
phonon frequency, longitudinal acoustics (LA) phonon frequency (≈100 cm−1) is
redshifted for all samples. This shift may be produced as a result of the effect of
finite particle size and/or strain in the thin film.

14.3.1 Experimental Details

The data for Raman scattering were measured at room temperature by using a
spectrometer employing Rayleigh line rejection filter. The samples were excited by
3 mW laser having a wave length of 532 nm and a spot size of 1 µm.
330 M.H. Yükselici et al.

14.3.2 Modelling

First order Raman spectrum can be fitted to the confinement model discussed in [5]:
2p
Z
q¼ a
jC ð0; qÞj2 d 3 q
I ðxÞ ¼ ð3:1Þ
½x  xðqÞ2 þ ðC0 =2Þ2
q¼0

 
jCð0; qÞj2 ¼ exp q2 d 2 =2a ð3:2Þ

where d is the diameter of the grain, a is the lattice parameter, q is the phonon wave
vector, a is a value determines how rapidly the wave function of the phonon decays
as the boundary is approached, C0 is the natural line width of the optical phonon in
bulk, xðqÞ is the dispersion curve:

xðqÞ ¼ x0  Dx sin2 ðqa=4Þ ð3:3Þ

x0 is the phonon frequency and Dx is the width of the dispersion curve.

14.3.3 Discussion

First order Raman spectra for the CdTe thin film samples are plotted in Fig. 14.2
according to the model in (3.1). Figure 14.2 shows lines centered at phonon fre-
quencies closed to that for the L-critical-point LA phonon which has a frequency of
3:25 THz ¼ 108:333 cm1 [6]. In this work, we observe that phonon frequency red
shifts as grain size decreases. This decrease in phonon frequency can be attributed to
the effect of compressive strain and/or phonon confinement due to finite grain size.

Fig. 14.2 Intensity against Intensity (a.u.)

Raman shift for CdTe thin 1.0
films. Line position is
50 nm
*100 cm−1 which is close to
L-critical-point LA phonon 0.8

0.6 200nm

0.4
350nm

500nm
0.2
Bulk

80 90 100 110 120

Raman Shift (cm-1)
14 Optical and Structural Properties of Quantum Dots 331

Table 14.1 Line position, line width and grain size calculated by Raman spectroscopy for CdTe
thin film
Thickness (nm) Line position (cm−1) Line width (cm−1) Rave (nm)
50 98.9 18.4 1.380
200 99.6 16.4 1.602
350 100.6 16.4 1.750
500 99.4 18.0 1.984
Bulk Measured 100.2 16.8
[2] 108.333

Referring to Table 14.1, the grain size calculated by Raman spectroscopy

decreases to a value less than Bohr radius of CdTe bulk (7.3 nm [7]) as the film
thickness decreases and hence the phonon confinement becomes stronger with
decreasing the thickness causing red shift in phonon frequency.

14.3.4 Importance of the Subject

In general, the compounds of group II–VI can be considered as materials of

valuable practical importance, the fact that they exhibit many considerable
solid-state phenomena. As a special case of these compounds, CdTe is a material
which can be considered as a completely representative of the group II–VI for its
distinctive physical properties. Moreover, CdTe is a promising compound since it is
used in electronic and optical manufactures. Studies about CdTe are scarce espe-
cially those related to Raman spectroscopy. Some articles have investigated the
structural and optical properties of CdTe using XRD and/or optical transmission
spectroscopy but they have not used Raman spectroscopy, the references [8–15] are
given as examples for some of these articles. On the other hand, there are lim-
ited number of articles which are related to Raman spectroscopy analysis for CdTe
[16–18]. These articles didn’t use the confinement model described in the present
work, which has the advantage of that it can compute the fitting parameters like
grain size and line width, for analysing the spectrum; in [19] the data was fitted to
Lorentzian profile and the spectrum in [20] was studied using a different model
described in [21], while [16–18] did not use the fitting at all and hence they could
not estimate the grain size and line width.

14.3.5 Section Summary

PVD technique was used for depositing CdTe on glass plates. Four samples of
CdTe thin films of different thicknesses were obtained. The samples are investi-
gated by Raman spectroscopy to study the effect of grain size on phonon
332 M.H. Yükselici et al.

confinement. The confinement model evaluated grain sizes ranging from about 1.4
to 2.0 nm. The importance of this work lies in the fact that there are few articles that
concern to analysis of Raman spectra for CdTe using the above model.

14.4 Steady State Photoluminescence Spectroscopy

The experimental data, results and discussions presented in this section are mainly
cited from and/or based on [22, 23]. In previous studies of CdSxSe1−x which is a
member of group II–IV quantum dots, it is revealed that photoluminescence (PL)
spectra generally has a three peak structure; one of them is close to the absorption
(ABS) edge due to direct electron hole recombination, the other one is attributed to
recombination from another energy levels within the band gap and the last low
energy peak is related to deep trap levels. The first peak mentioned above is
assigned to volume contribution and the other peaks are presumed to be surface
states [24, 25]. Hache et al. studied with emphasis on evolution of intensity of the
excitation beam in PL spectra. It is employed that at low intensity there is only one
peak which is related to deep trap levels. As the intensity increases, higher energy
peak (L1) due to direct electron hole pair recombination remains stationary, whereas
lower energy peak (L2) that leads to recombination through the new unidentified
states becomes visible. At higher intensities, both of the peaks show up and begin to
get sharpener with respect to decrease in deep trap level luminescence. In addition
to this, they also figured out that recombination from another states has more
efficiency than direct electron hole pair recombination. G. Mei et al., who studied
on variously sized quantum dots embedded in glass matrix, pointed out that the PL
peak near the absorption edge is related to bound-exciton recombination and
indicates the impurity or surface states of CdSxSe1−x quantum dots. As the particle
size gets smaller, more surface states occur in PL spectrum. In other words, since
the surface to volume ratio of quantum dot spheres increases, the PL spectrum is
dominated by surface states of decreasing particle sizes [26, 22]. Rodden et al.
scrutinized the effect of annealing at different temperatures and doping concentra-
tions for CdSxSe1−x samples [27]. In PL spectra a shift is observed through the
lower energies by increasing temperatures or doping concentrations. Inreasing
temperature results in an increase of crystallite size. As the crystallite size becomes
larger, the effective energy gap of the CdSxSe1−x sample shifts towards lower
energies, indicating an existence of three-dimensional quantum confinement of
electrons and holes. A reduce in doping concentrations causes a growth of smaller
particles, hence a shift towards the higher energy is observed in PL spectra. In more
recent works, nanocrystals in a range of 2.3–4.2 nm radii are analyzed with a
quantized state effective mass theory by Allahverdi et al. [23]. They figured out the
PL spectra with the help of the model Ravindran et al. and also concentrate on deep
trap peak and amount of its shift from the first exciton band. The peaks which are
located at *1.56 and *1.6 eV remains stationary as the heat treatment progresses.
However, deep trap peak in PL and the first exciton peak in ABS shift towards
14 Optical and Structural Properties of Quantum Dots 333

lower energies (38 meV for deep trap peak) due to quantum confinement effect. The
shift can also arise from the variation in S content that may increase/decrease with
the heat treatment [26]. Since the shift due to S content is about *20 meV, it is
presumed that the redshift of 38 meV in deep trap peak represents the quantum
confinement effect. With the help of the model Ravindran et al. presented,
Allahverdi et al. observed that a decrease in energy occurs between the deep trap
and first exciton peak with increasing average particle radius. It can be inferred
from the ABS and PL spectra, the narrower shift, the larger are the particles. This
indicates that deep trap levels have less sensitivity to variations in average quantum
dot radius than that of direct recombination levels in Cd0.08Se0.92 quantum dots.
We have focussed on trap sites of variously sized samples of CdSxSe1−x QDs in
glass. As-received RG695 Schott borosilicate glass doped with CdS0.08Se0.92 was
melted at 1000 °C for 15 min in order to dissolve the particles and after quenching
to room temperature, the glass was annealed at 450 °C for 5 h to decrease the stress
in it. Finally the samples were exposed to heat treatment gradually from 600 to
675 °C for different periods of times as listed in Table 14.2 [22].
In order to identify the trap sites of gradually annealed CdSxSe1−x samples, we
present the steady state PL spectra (light colored curve) with the Gaussian model
lineshape (dark colored curve) on it in Fig. 14.3 [22].
The modelling function was employed below, was also desribed in [28]. Hence,
the calculated PL intesity has the following form
" 2 # ( " # " #)
A hm  Etrap X 1 ð R  R0 Þ 2 B ðhm  ER Þ2
IPL ðhmÞ ¼ pffiffiffiffiffiffi exp þ p ffiffiffiffiffi
ffi exp  p ffiffiffiffiffiffiffiffi
ffi exp 
2prR 2C2trap R 2prR 2r2R 2pC 2C2

ð4:1Þ

Table 14.2 Heat treated CdSxSe1−x samples for different temperatures and times
Sample Annealing temperature (°C) Time (h) R0* (nm) R0** (nm) σ* (nm)
number
1 600 1/2 – – –
2 625 1 1.7 ± 0.38 2.1 1.2 ± 0.3
3 625 2 1.8 ± 0.36 2.2 1.3 ± 0.3
4 625 4 2.9 ± 0.10 2.5 0.32 ± 0.01
5 625 8 3.3 ± 0.73 2.7 0.22 ± 0.05
6 650 2 3.4 ± 0.17 3.1 0.55 ± 0.03
7 650 4 4.3 ± 0.87 3.4 0.70 ± 0.14
8 650 8 4.8 ± 0.17 3.5 0.70 ± 0.03
9 675 2 7.2 ± 0.52 4.5 1.3 ± 0.1
10 675 4 8.5 ± 1.3 4.7 1.3 ± 0.2
Average radius (R0*) and size dispersion (σ*) calculated by the employed model of PL spectra.
Average radius (R0**) obtained from the energetic position of the first exciton peak is pointed out
for comparison
334 M.H. Yükselici et al.

where A and B = the constants, R0 = radius of the spherical quantum dot, rR = s-

tandard deviation of the radius, C = 2 % width of homogenious broadening,
Ctrap = 10–20 % width of trap band broadening.
We consider that C ¼ ðDE=E Þ  100 where DE indicates the standard deviation
homogenious at an energy of E. The size dependent first exciton transition energy is
characterized by following relation [29].

0:14 0:356
ER ðeVÞ ¼ Eg ðeVÞ  þ ð4:2Þ
RðnmÞ l½RðnmÞ2

The first term leads to the bulk band gap of CdS0.08Se0.92 of Eg ¼ 1:727 eV
while the second term is responsible for the Coulomb potential energy between an
electron and a hole. l represents 0:106m0 that m0 indicates electron rest mass.
As seen in Fig. 14.3, in the first periods of times and temperatures the dark
colored curves 2 and 3 exhibits one peak at *1.8 eV. We presume that this peak is
attributed to volume traps and the model figures out that it is almost stationary. When
the heat treatment progresses, another peak begins to resolve at *2.1 eV and shifts

Fig. 14.3 Light colored lines

indicate the steady state PL
spectra obtained with 488 nm
cw Ar+ laser of the intesity of
*70 nW/mm2. Dark colored
lines refer to simulated PL
spectra. Average nanocrystal
radius ðR0 Þ and size
dispersion ðr0 Þ are the best fit
parameters which are
presented beside each pair of
curve. Heat treatment
temperatures and times
numbered as (1) 600 ½, (2)
625, 1; (3) 625, 2; (4) 625, 4;
(5) 625, 8; (6) 650, 2; (7) 650,
4; (8) 650, 8; (9) 675, 2; (10)
675, 4, respectively. See
Table 14.2 for details
14 Optical and Structural Properties of Quantum Dots 335

through the lower energies for samples 4 and 5. During the gradual increase in heat
treatment, this peak overlaps with the stationary peak which leads to the volume trap
band from samples 5 to 10. According to the modelling, this peak is related to a size
dependent band and identified as a surface assisted electron hole recombination
band. It is redshifted from the first exciton peak by amount of 120 meV and the
length of the shift decreases with respect to increasing radius of the QDs.

14.5 The Progression of Strain and Micro-electric Field

Dependent Urbach Energy with Deposition Time
in Chemical Bath Deposited CdS Thin Films

Thin film CdS semiconductors are of special interest due to their technological
applications in solar cells. CdS can be used as a window material in
hetero-junctions due to its wide band gap energy of *2.5 eV [30–34]. Chemical
bath deposition (CBD) is a low cost method to grow thin films over wide surfaces
[35–40]. Most of studies on CBD CdS have gone into the optimization of exper-
imental parameters such as bath temperature, pH value, concentration of reactants,
deposition time and post-deposition heat treatment. On the other hand, recent
Raman studies show that the interfacial strain and phonon confinement modify the
phonon vibrational modes. A 0.35 % change in lattice constant for CdSe nanotubes
produces a blue shift of 2.4 cm−1 in LO1 mode in Raman spectrum [41]. CdSe
nanocrystals of an average radius of 3 nm and a size distribution of 10 % or FWHM
of 0.3 nm result in a red shift of 1. 82 cm−1 in LO1 mode relative to bulk structure
of the same composition [42]. In this section we employ optical absorption spec-
troscopy (ABS), Scanning Electron Microscopy (SEM) and X-Ray Diffraction
(XRD) spectroscopy for initial characterization of thin film CBD CdS structures.
We analyze the optical absorption (ABS) data with the help of Wentzel, Kramers,
Brillouin (WKB) approach to relate the long wavelength tail below the fundamental
absorption edge to micro-electric fields induced by crystalline disorder. Finally we
employ resonant Raman spectroscopy to assess the relative weight of results
obtained by XRD data with the help of Grüneisen model and phonon confinement.

14.5.1 Experimental

14.5.1.1 Sample Preparation

CdS thin films were deposited on glass substrates by chemical bath deposition
(CBD). Glass substrates were cleaned and dipped into a 60 °C reaction bath of
pH = 11.6 prepared from 0.1 M of cadmium sulfate, 1 M of thiourea each separately
336 M.H. Yükselici et al.

Fig. 14.4 (OD × hν)2 plotted 30 (5)

against photon energy (hν) for
chemical bath-deposited CdS
thin films. Deposition times 25 (4)
are numbered as follows: (3)
1 → 10, 2 → 16, 3 → 22, (2)
4 → 40, 5 → 150 min 20

2
(1)

(OD x E)
15

10

0
1.8 2.2 2.6 3 3.4 3.8 4.2
Energy (eV)

dissolved in 30 ml of distilled water and 5 % ammonium hydroxide according to the

procedure reported in detail elsewhere [43]. A set of 5 glass slides deposited for 10,
16, 22, 40 and 150 min under stirring at 250 rpm is scrutinized in this work.

14.5.1.2 Optical Absorption

Optical density (OD) for each sample is calculated by optical transmission mea-
surements. As shown in Fig. 14.4, the straight-line portions of the (OD × hν)2 − hν
curves are extrapolated to obtain the band gap energies (Eg) as intercept with the
x axis according to the relation (OD × hν)2 = (A*d)2(hν − Eg), where A* is a
constant and d film thickness [44]. As the deposition time increases, the asymptotic
absorption edge with a tail structure shifts to lower energies. A solution to
Schrödinger Equation for a particle in a spherical box under effective mass model
yields a relation between the energetic position of first exciton peak and average
particle’s radius [45]. The blue shift is related to grain size by
1
ER ðeVÞ ¼ Eg ðeVÞ  0:1655=R ðnmÞ þ 0:376=l½R ðnmÞ2 where the first term is
bulk band gap of CdS crystal of Eg = 2.402 eV, the second term the attractive
Coulomb potential energy and the third term the kinetic energy for a particle in a
spherical box (μ = 0.1674m0, m0 is the electron rest mass). The average grain radii
Rave (nm) for each film are calculated as 1.21, 1.71, 1.91, 2.60 and 3.59 nm for
samples 1, 2, 3, 4 and 5 respectively.

1
Derived by using [16].
14 Optical and Structural Properties of Quantum Dots 337

Fig. 14.5 Surface SEM

micrographs for sample 2
(top) and sample 4 (bottom)

14.5.1.3 SEM and EDS

JEOL/35M-6335F scanning electron microscope (SEM) was employed for the

surface morphological analysis. SEM surface micrographs are shown in Fig. 14.5
and cross-sectional micrographs in Fig. 14.6 for samples 2 and 4 respectively. SEM
micrographs have an irregular and anfractuous surface like a web pattern. We
assume that powdery and nonsticky film deposition give rise to an extremely rough
surface. Figure 14.7 shows Energy Dispersive X-ray spectrum (EDS) of CdS film
deposited for 40 min (sample 4). The most intense peak is due to Si of atomic % of
22.47 which is presumably from glass substrate. The second and third intense peaks
might be due to Cd of atomic % of 11.89 and S of 11.55.
338 M.H. Yükselici et al.

Fig. 14.6 Cross sectional SEM micrographs for sample 2 (top) and sample 4 (bottom)

Element Weight% Atomic%

OK 20.24 45.23
Na K 2.15 3.35
Si K 17.64 22.47
SK 10.36 11.55
Ca K 2.81 2.51
Cd L 46.8 14.89

Fig. 14.7 Energy dispersive X-ray spectrum (EDS) of CdS thin film deposited for 40 min (sample
4)
14 Optical and Structural Properties of Quantum Dots 339

14.5.1.4 XRD

XRD patterns recorded on Panalytical MPD diffractometer are given in Fig. 14.8.
We fit each XRD spectrum with a Lorentzian line profile for each
peak/shoulder/hump and a cubic background function and then find best fitting
parameters for peak positions and full width at half maximum (FWHM). We pre-
sume that the peaks at 2θ = 26.5°, 43.9°, 51.9° and 54.7° correspond to H[002]/C
[111], H[110]/C[220], H[112]/C[311] and H[004]/C[222] planes of hexagonal
(H) (ICDD PDF No: 03-065-8873) and/or cubic (C) (ICDD PDF No: 03-065-2887)
planes of CdS crystal. The peaks at 18.3o and 31.6o are assigned to cubic planes of
CdSO4 structure (ICDD PDF No: 00-003-0453). The reported studies show that
CBD CdS thin films grow on glass substrates in cubic zincblende, hexagonal
wurtzite or mixed H and C structures depending on bath temperature and
post-deposition heat-treatment. The broad structure at 65.6° is assigned to [322]
cubic plane. A shoulder at 25.3° and a hump at 28.2° identified at the earlier stages
of film growth are assigned to [110] and [101] hexagonal planes of CdS crystal
respectively. We observe that the peak at 26.6° grows at the expense of the clear
structures at between 25° and 35° for thin film deposited for 150 min. The peak
structures assigned to H[100] and H[101] disappear for later stages of thin film
growth. However the persistence of C[322] suggests a crystal growth which favors
cubic phase with the peak at 26.5° assigned to C[111].

Fig. 14.8 XRD pattern for

CBD CdS thin films of
different thicknesses
5
3

4
Intensity (a. u.)

2 3

10 20 30 40 50 60 70
2 (deg)
340 M.H. Yükselici et al.

14.5.1.5 Raman Spectroscopy

Resonant Raman scattering measurements were carried out at room temperature on

Renishaw 250 mm focal length in Via Reflex Spectrometer system using Rayleigh
line rejection filter allowing Raman spectrum to 100 cm−1 from a laser line. An
accuracy of ±0.1 cm−1 was estimated for the spectral region studied. The Raman
spectra are shown in Fig. 14.9. The Raman spectra are excited by a cw laser line at
532 nm and a total power of 3 mW within a spot size of 1 μm on the sample. Because
the x-ray pattern in Fig. 14.8 displays quite amorphous background, the Raman
spectra were collected from crystalline regions in the substrate. The spectra display a
sharp peak at between 298–305 cm−1 attributed to zone center longitudinal optical
(LO) mode of cubic phonons and hexagonal A1(LO) + E1(LO) phonons and over-
tones at 592–600 cm−1 [46]. We assume that the broad structure at the low frequency
side of the most intense mode for the film deposited for 10 min is possibly originated
from a cubic phonon mode based on a previous study by O. Z-Angel et al. [46] who
reported that E1(TO) mode of cubic structure at 240 cm−1 is shifted to 277 cm−1 as a
result of cubic to hexagonal phase transformation. A Lorentzian fit which is centered
at 278 cm−1 is consistent with their proposal. It is also apparent that the most intense
mode is asymmetrically broadened toward the lower frequency which can be fitted
by a Lorentzian profile centered at *297 cm−1 which might be due to disordered
activated zone edge optical phonons as reported by previous researchers [47].
A broad peak at around 400 cm−1 which appears for the thinnest film is due pre-
sumably to Si–O vibrational modes from glass substrate (Fig. 14.9).
The shift in LO1 peak position is related to grain radius by digitizing the cal-
culated curve between LO1 mode and grain radius at 295 °C in [48]. We find a
relation for the radius dependent vibrational frequency: xðRÞ ðcm1 Þ ¼ 303:44 
20:57= R2 ðnm2 Þ.

Fig. 14.9 Raman spectra for

CBD CdS thin films of 140000
different thicknesses
120000 5

100000
Intensity (a. u.)

4
80000
3
60000
2

40000
1
20000

0
100 200 300 400 500 600 700 800 900
-1
Raman Shift (cm )
14 Optical and Structural Properties of Quantum Dots 341

14.5.2 Modeling of the Urbach Tail

The defect levels below the band gap energy broaden due mainly to thermal
vibrations and phonon-electron interactions. The long wavelength absorption tail is
explained by the excitation of a low energy electron to conduction band through
tunneling the potential energy incline created by electric micro fields [49]. The
schematic energy band diagram is shown in Fig. 14.10 without field (a) and with
field (b). The potential well of depth Uo represents a trap level over which an
excited charge from the valence band tunnels through the classically forbidden
region II as shown in (b) under electric field. The exponential tail below the fun-
damental absorption edge (Eg) related to the degree of structural disorder is
determined by Urbach rule. We follow the procedure as described by Liboff [50]
and derive transmission probability for an electron to tunnel through a potential
barrier under electric field. Micro-field (F) of defects employing basic quantum
mechanics [50]. In the WKB approximation the wave function for the charged
particles is approximated by modified plane wave form:

uð xÞ ¼ AeiSðxÞ ð5:1Þ

where Sð xÞ ¼ S0 ð xÞ þ hS1 ð xÞ þ 1=2h2 S2 ð xÞ þ


is the WKB expansion. After

substituting (5.1) in Schrödinger equation and keeping only the first term in the
expansion, we obtain the WKB solution in region III [50],
2 x 3
Z
A p
uIII ð xÞ ¼ pffiffiffi exp4i kdx  5 ð5:2Þ
k 4
x2

(b) +E
(a) conduction band
F≠ 0
Eg
U0
E g′
Eg E
I II III

valance band +x
x2 = ( Eg − E ) / eF
x1 = Eg − Eg′ / eF

Fig. 14.10 The schematic energy band diagram with no field (a) with field (b) applied
342 M.H. Yükselici et al.

which is related to uI over uII by connecting formulas,

" #
C Zx
0 p
uI ð xÞ ¼ pffiffiffi rexp i k dx  ð5:3Þ
k x1 4
" #
Rx2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where r ¼ exp Kdx and k or k0 ¼ 2mðE  UÞh2 and K ¼ 2mðU  EÞh2
x1
are the magnitudes of wave vectors in regions I ðU ¼ eFx þ Eg  U0 \E) or III
ðU ¼ eFx þ Eg \EÞ and in region II ðU ¼ eFx þ Eg [ EÞ respectively. The
transmission probability (T) which is the ratio of transmitted (Jtrans) to incident
(Jinc) current densities in regions III and I respectively is calculated as T = r−2 where
h
  du 
J ¼ 2mi u dx  u dudx [50]. A photon of energy E = hυ below the shifted band gap
Eg0 will be absorbed depending on the transmission probability and therefore we
propose that α ≈ T based on the earlier work by Dow and Redfield [51] and
calculate the absorption coefficient by integrating the argument of r from x1 ¼
ðEg0  EÞ=eF to x2 ¼ ðEg  EÞ=eF,
" #
ðEg0  EÞ3=2
aðE; F Þ ¼ C1 exp C2 ð5:4Þ
F
qffiffiffiffiffiffiffiffiffiffiffiffi
where C1 is a constant, C2 ¼ 4e3 2l=h2 and Eu is Urbach energy. In Fig. 14.11, the
calculated optical tail densities are plotted against photon energies for different
thickness films along with experimental tail optical densities. As the electric
micro-field increases, the absorption tail width related to Urbach energy increases.
We model each optical absorption tail for different thickness films and find best
average electric micro-field values. In Fig. 14.12, we plot electric micro-field as a
function of deposition time. We see that as the deposition time increases, electric
micro-field strength decreases. As thickness of the film of *2 μm at 16 min of
deposition time increases to *2.5 μm at the end of 40 min, the average electric
micro-field decreases. Average electric micro-field can be described by a power law
decay function given by the expression: F(kV/mm) = 3.4 × [t(min)]−0.32.
We relate electric micro-field to Urbach energy to explore the relation between
disorder and internal electric field induced by crystalline disorder. The optical
density is given by,
 
ht  Eg
OD ¼ OD0 exp ð5:5Þ
Eu

here OD0 is a constant. The Urbach tail width represented by the Urbach energy
(Eu) is calculated from the slope of least-squares linear fit to the straight portion of
lnOD against the hν curve [52]. The Urbach energy is plotted against the average
14 Optical and Structural Properties of Quantum Dots 343

Fig. 14.11 Experimental and

calculated (dashed curve) (5) (4) (3) (2) (1)
1.0
optical absorption tail spectra
below band gap energy for
different thickness thin films
0.8

0.6

OD 0.4

0.2

0.0
1.5 2.0 2.5 3.0 3.5
Energy(eV)

electric micro-field in Fig. 14.13. The right vertical axis is the unitless structural
disorder parameter (X) calculated by the equation [53],
  
Ep Ep
Eu ¼ X þ coth ð5:6Þ
2r0 2kBT

where Ep of 25 meV is the phonon energy, σ0 of 1.19 is a material-dependent

parameter and X is a measure of structural disorder. The straight line through the
data points is a linear fit given by y = 0.16x + 254 where y = Eu(meV) or X(%) and
x = F(kV/mm).

14.5.3 The Progression of Strain with Deposition Time

The XRD diffraction lines broaden due to finite crystalline size, strain and instru-
mental error. The size broadening is given by Debye-Scherrer relation

Kk
Dð2hhkl Þsize ¼ ð5:7Þ
2Rave cos hhkl

where Rave is the average radius of grains. The strain broadening and the strain
coefficient are given by,
344 M.H. Yükselici et al.

Fig. 14.12 Electric

micro-field plotted against
deposition time. The curve is
a power law fit the data 2163

F (kV/mm)
1663

1163

663
0 50 100 150
Deposition Time (min)

Fig. 14.13 Average electric 61

micro-field against Urbach 650
energy and disorder. The
56
straight line is a linear fit to 600
the data with slope of 0.16
51
550
E u (meV)

X
500 46

450 41

400 36

350 31
500 1000 1500 2000 2500
F (kV/mm)

Dð2hhkl Þstrain ¼ 2 2 tan hhkl ð5:8aÞ

dhkl  dhkl
bulk
2¼  bulk
ð5:8bÞ
dhkl

We approximate the angular dependent instrumental error function through a

scan on the Si powder as Δ(2θhkl)inst(mrad) = −9 × tan θhkl + 4.5/cos θhkl which
estimates that the maximum instrumental broadening could be only 20 % of the
least FWHM.
14 Optical and Structural Properties of Quantum Dots 345

Table 14.3 The summary of the results for different thickness CdS thin films
Deposition Urbach Grain radius F Strain × 10−3 Raman Eg Thickness
time (min) energy (nm) (kV/mm) shift (eV) (μm) SEM
(meV) ABS Raman XRD Raman (cm−1)
10 605 1.21 1.93 2239 9.55 2.24 297.90 3.72 –
16 445 1.71 3.47 1244 7.73 0.19 301.74 3.08 *2.0
22 417 1.91 3.88 1077 4.80 0.09 302.08 2.93 –
40 377 2.60 3.24 848 3.52 0.33 301.49 2.67 *2.5
150 396 3.59 1.96 823 3.48 3.13 298.11 2.53 –

We related peak position for C[111] for each sample to the interplanar distance
d111 by employing Bragg Law for cubic structure with a bulk lattice constant of
a = 0.582 nm. The progression of strain values calculated by employing XRD data
and (5.8b) are listed in Table 14.3.
Raman spectroscopy was employed to assess the relative weigh of the result for
the strain obtained by XRD. Vibrational modes in Raman spectra are fitted by a
Lorentzian line profile for each vibrational mode and a cubic background function.
The best fitting parameters for the LO1 peak position of Pi are recorded to calculate
the strain coefficient which is related to the vibrational frequency by [54],

1
Pi  Dxci c
Da LOBi
1
¼ ð5:9Þ
a 3

where LOBi is the bulk frequency of LO phonon, Dxci is the red-shift due to phonon
confinement and γ is the Grüneisen parameter of 1.51 for bulk CdS [55]. Previous
studies show that the shift in Raman frequency due to phonon confinement is
negligible and can be ignored compared to the shift due to strain. The strain values
calculated by using LO1 peak positions in Fig. 14.9 are tabulated in Table 14.3.

14.5.4 Section Conclusion

Different thickness CdS thin films were grown on glass by CBD. The red shift
observed in asymptotic optical absorption edge is used to calculate the average
grain radii from 1.2 to 3.6 nm with the help of effective mass model under quantum
confinement of charged carriers. The optical absorption below the band gap energy
is simulated by a simple quantum mechanical model which presumes the WKB
solution to Schrödinger equation under a disorder induced micro-electric field. The
simulation shows that the micro-electric field decreases with grain size. The
exponential tail width proportional to Urbach energy varies nearly linearly with
micro-field. The results extracted directly or through calculation from the optical
absorption, XRD, SEM and Raman data are summarized in Table 14.3.
346 M.H. Yükselici et al.

The red shift in Raman LO1 vibrational modes and the shift in C(111) diffraction
peak position in XRD spectra yield a compressive strain on the order of power of 3.
We conclude that as the deposition time increases from 10 to 150 min, the average
grain radius changes by 2 nm, Urbach energy and the electric micro-field pre-
sumably induced by crystalline disorder decrease approximately from 600 to
400 meV and 2240–820 kV/mm respectively. The shift in XRD pattern shows that
the compressive strain decreases with growth from 9.55 × 10−3 to 3.48 × 10−3. The
Raman LO1 vibrational mode displays an increase up to 22 min of deposition time
and then a decrease. There might be two contributions to observed shift in LO
phonon mode: strain and phonon confinement. Phonon confinement leads to a
decrease in frequency because the zone center LO phonon frequency of
303.44 cm−1 for bulk CdS is at a maximum. The shift due to strain might be
positive or negative. At the earlier stages of grain growth when the size is 1.2 nm in
radius, we observe a decrease of about 5 cm−1 which is too high to be due to
phonon confinement since Raman frequency is red shifted by only 1 cm−1 as the
nanocrystalline radius change between 1 and 3 nm as reported in [56]. As shown in
Table 14.3 the strain calculated by Grüneisen model decreases with deposition time
up to 22 min and increases later on which might be due to a phase transition to
cubic structure from a mixed phase of hexagonal and cubic structures at the earlier
stages of film growth as is evident from XRD pattern.

14.6 In Situ Low Temperature Optical Absorption

Spectroscopy

The low temperature optical transmission measurements were conducted on one

CdSe doped glass sample by employing a custom-made liquid nitrogen optical
chamber. A diagram of the sample geometry is shown in Fig. 14.14. The temper-
ature was monitored with a thermo-couple positioned within 2 cm of the center of
the crystal fixed on a sample holder in the chamber. The light from a 50 W tungsten
lamp, which was powered by a constant current source, was collected by a lens,
passed through the quartz window of the chamber and then the sample. The
transmitted light was collected by another lens behind the quartz window at the

Fig. 14.14 The experimental setup for in situ low temperature optical absorption spectroscopy
14 Optical and Structural Properties of Quantum Dots 347

Fig. 14.15 The energetic 2,35

position of asymptotic optical
absorption edge plotted

Band gap energy (eV)

against temperature. The line
through the data is a linear fit
with a slope of −4 × 10−4
eV/K 2,30

2,25
100 150 200 250 300
Temperature (K)

opposite side of the chamber and focused on the entrance slit of a ¼ m spectro-
graph. The transmitted spectra were measured as a function of temperature at
between 104 and 253 K. The band gap energy is plotted as a function of temper-
ature in Fig. 14.15. The line is a least-squares straight line fit whose slope is
4
dT ¼ 4  10
dE
eV=K:

Acknowledgements This work has been supported by Yıldız Technical University Scientific
Research Project Coordination under project nos, 2012-01-01-KAP09 and 2012-01-01-KAP03
2011-01-01-DOP01. XRD and SEM were conducted at TÜBİTAK MRS Industrial services. We
thank Dr. Barış Yağcı at Surface Science and Technology Center (KUYTAM), Koç University for
conducting Raman measurement.

References

1. V. Wood, V. Bulović, Colloidal quantum dot light-emitting devices. Nano Rev. 1(5202), 1–7
(2010)
2. L. Liu, Q. Peng, Y. Li, An effective oxidation route to blue emission CdSe quantum dots.
Inorg. Chem. 47(8), 3182–3187 (2008)
3. J. Yoon, W. Chae, S. Im, Y. Kim, Mild synthesis of ultra-small CdSe quantum dots in
ethylenediamine solution. Mater. Lett. 59, 1430–1433 (2005)
4. C. Mell Donegá, P. Liljeroth, D. Vanmaekelbergh, Physicochemical evaluation of the
hot-injection method, a synthesis route for monodisperse nanocrystals. Small 12, 1152–1116
(2005)
5. A.K. Arora, M. Rajalakshmi, T.R. Ravindran, Phonon confinement in nanostructured
materials. Encycl. Nanosci. Nanotechnol. 8, 499–512 (2004)
6. H. Matsuo, T. Soma, Lattice dynamics of tetrahedral compounds with the local Heine‐
Abarenkov model potential. Phys. Stat. Sol. (b) 124, 37–44 (1984)
7. F. Wu, J.W. Lewis, D.S. Kliger, J.Z. Zhang, Unusual excitation intensity dependence of
fluorescence of CdTe nanoparticles. J. Chem. Phys. 118, 12 (2003)
8. M.H. Yükselici, Ç. Allahverdi, Size-dependent photo-induced shift of the first exciton band in
CdTe quantum dots in glass prepared by a two-stage heat-treatment process. J. Lumin. 128(3),
537 (2008)
348 M.H. Yükselici et al.

9. V.M. Nikale, S.S. Shinde, C.H. Bhosale, K.Y. Rajpure, Physical properties of spray deposited
CdTe thin films: PEC performance. J. Semicond. 32, 033001–1 (2011)
10. H.R. Moutinho, F.S. Hasoon, F. Abulfotuh, L.L. Kazmerski, Investigation of polycrystalline
CdTe thin films deposited by physical vapor deposition, close-spaced sublimation, and
sputtering. J. Vac. Sci. Technol., A 13, 2877 (1995)
11. Ç. Allahverdi, M.H. Yükselici, Temperature dependence of absorption band edge of CdTe
nanocrystals in glass. New J. Phys. 10, 103029 (2008)
12. G. Gordillo, J.M. Florez, L.C. Hernandez, Preparation and characterization of CdTe thin films
deposited by CSS. Sol. Energy Mater. Sol. Cells 37, 273–281 (1995)
13. M. H. Yükselici, Ç. Allahverdi, Solid-phase precipitation of CdTe nanoparticles in glass, Phys.
Stat. Sol. (b) 236(3), 649 (2003)
14. S. Lalitha, R. Sathyamoorthy, S. Senthilarasu, A. Subbarayan, K. Natarajan, Characterization
of CdTe thin film—dependence of structural and optical properties on temperature and
thickness. Sol. Energy Mater. Sol. Cells 82, 187–199 (2004)
15. S. Wageh, A.A. Higazy, M.A. Algradee, optical properties and activation energy of a novel
system of CdTe nanoparticles embedded in phosphate glass matrix. J. Modern Phys. 913–921
(2011)
16. P.T.C. Freire, M.A. Araujo Silva, V.C.S. Reynoso, A.R. Vaz, V. Lemos, Pressure Raman
scattering of CdTe quantum dots. Phys. Rev. B 55(11), 6743 (1997)
17. S.S. Islam, S. Rath, K.P. Jain, S.C. Abbi, Forbidden one-LO-phonon resonant Raman
scattering and multiphonon scattering in pure CdTe crystals. Phys. Rev. B 46(8), 4982 (1997)
18. V.S. Vinogradov, G. Karczewski, I.V. Kucherenko, N.N. Mel’nik, P. Fernandez, Raman
spectra of structures with CdTe-, ZnTe-, and CdSe-based quantum dots and their relation to the
fabrication technology. Phys. Solid State 50(1), 164–167 (2008)
19. V. Dzhagan, I. Lokteva, C. Himcinschi, X. Jin, J. Kolny-Olesiak, D.R.T. Zahn, Phonon
Raman spectra of colloidal CdTe nanocrystals: effect of size, non-stoichiometry and ligand
exchange. Nanoscale Res. Lett. 6, 79 (2011)
20. A.G. Rolo, M.I. Vasilevskiy, N.P. Gaponik, A.L. Rogach, M.J.M. Gomes, Confined optical
vibrations in CdTe quantum dots and clusters. Phys. Stat. Sol. (b) 229(1), 433–437 (2002)
21. M.I. Vasilevskiy, A.G. Rolo, M.J.M. Gomes, O.V. Vikhrova, C. Ricolleau, Impact of disorder
on optical phonons confined in CdS nano-crystallites embedded in a SiO2 matrix. J. Phys.:
Condens. Matter 13, 3491–3509 (2001)
22. B.C. Ömür, A. Aşıkoğlu, Ç. Allahverdi, M.H. Yükselici, CdSxSe1-x quantum dots studied
through optical absorption, steady state photoluminescence and resonant Raman spectroscopy.
J. Mater. Sci. 45, 112–117 (2010)
23. Ç. Allahverdi, M.H. Yükselici, R. Turan, A. Seyhan, Photoluminescence spectroscopy in the
study of growth of CdSexS1−x nanocrystals in glass. Semicond. Sci. Technol. 19, 1005 (2004)
24. N.F. Borrelli, D.W. Hall, H.J. Holland, D.W. Smith, Quantum confinement effects of
semiconducting microcrystallites in glass. J. Appl. Phys. 61, 5399 (1987)
25. F. Hache, M.C. Kelin, D. Ricard, C. Flytzanis, Photoluminescence study of Schott commercial
and experimental CdSSe-doped glasses: observation of surface states. J. Opt. Soc. Am. B. 8,
1802 (1991)
26. G. Mei, A photoluminescence study of CdSexS1-x semiconductor quantum dots. J. Phys.:
Condens. Matter 4, 7521 (1992)
27. W.S.O. Rodden, C.N. Ironside, C.M.S. Torres, A study of the growth of CdSexS1-x crystallites
within a glass matrix. Semicond. Sci. Technol. 9, 1839 (1994)
28. T.R. Ravindran, A.K. Arora, B. Balamurugan, B.R. Mehta, Inhomogeneous broadening in the
photoluminescence spectrum of CdS nanoparticles. Nanostruct. Mater. 11, 603–609 (1999)
29. F. Henneberger, J. Puls, C. Spiegelberg, A. Schülzgen, H. Rossman, V. Jungnickel, A.I.
Ekimov, Optical and electro-optical properties of II-VI quantum dots. Semicond. Sci. Technol.
16, A41 (1991)
30. A.D. Compaan, The status of and challenges in CdTe thin-film solar-cell technology. Mater.
Res. Soc. Symp. Proc. 808, 545–555 (2004)
14 Optical and Structural Properties of Quantum Dots 349

31. J. Britt, C. Ferekides, Thin-film CdS/CdTe solar cell with 15.8 % efficiency. Appl. Phys. Lett.
62, 2851–2852 (1993)
32. K.L. Chopra, P.D. Paulson, V. Dutta, Thin-film solar cells: an overview. Prog. Photovolt. Res.
Appl. 12, 69–92 (2004)
33. S. Chun, Y. Jung, J. Kim, D. Kim, The analysis of CdS thin film at the processes of
manufacturing CdS/CdTe solar cells. J. Cryst. Growth 326, 152–156 (2011)
34. C. Gretener, J. Perrenoud, L. Kranz, L. Kneer, R. Schmitt, S. Buecheler, A.N. Tiwari,
CdTe/CdS thin film solar cells grown in substrate configuration. Prog. Photovolt. Res. Appl.
(2012). doi:10.1002/pip.2233
35. C. Sahu, S.N. Sahu, Preparation of CdS semiconductor thin films by a solution growth
technique. Thin Solid Films 235, 17–19 (1993)
36. R. Ortega-Borges, D. Lincot, Mechanism of chemical bath deposition of cadmium sulfide thin
films in the ammonia-thiourea system in situ kinetic study and modelization. J. Electrochem.
Soc. 140, 3464–3473 (1993)
37. J.M. Doña, J. Herrero, Chemical bath deposition of CdS thin films: an approach to the
chemical mechanism through study of the film microstructure. J. Electrochem. Soc. 144,
4081–4091 (1997)
38. C. Guill’en, M. A. Mart′ınez, J. Herrero, Accurate control of thin film CdS growth process by
adjusting the chemical bath deposition parameters. Thin Solid Films 335, 37–42 (1998)
39. R.S Mane, C.D. Lokhande, Chemical deposition method for metal chalcogenide thin films
Mater. Chem. Phys. 65, 1–31 (2000)
40. J. Herrero, M.T. Gutiérrez, J.M. Doña, M.A. Martínez, A.M. Chaparro, R. Bayon,
Photovoltaic windows by chemical bath deposition. Thin Solid Films 361–362, 28–33 (2000)
41. R. Venugopal, P.-I. Lin, C.-C. Liu, Y.-T. Chen, Surface-enhanced Raman scattering and
polarized photoluminescence from catalytically grown CdSe nanobelts and sheets. J. Am.
Chem. Soc. 127, 11262–11268 (2005)
42. E.S.F. Neto, N.O. Dantas, S.W. da Silva, P.C. Morais, M.A. Pereira da Silva, Confirming the
lattice contraction in CdSe nanocrystals grown in a glass matrix by Raman scattering.
J. Raman Spectrosc. 41, 1302–1305 (2010)
43. A. Aşıkoğlu, M.H. Yükselici, Evolution of the energy band structure in
chemical-bath-deposited CdS thin films studied by optical absorption spectroscopy.
Semicond. Sci. Technol. 26, 055012 (2011)
44. J.I. Pankove, Optical Processes in Semiconductors (Dover, New York, 1971)
45. F. Henneberger, J. Puls, A. Ch Spiegelberg, H.Rossman Schulzgen, V. Jungnickel, A.I.
Ekimov, Optical and electro-optical properties of II-VI quantum dots. Semicond. Sci. Technol.
6, A41–A50 (1991)
46. O. Zelya-Angel, Raman studies in CdS thin films in the evolution from cubic to hexagonal
phase. Solid State Commun. 104, 161–166 (1997)
47. S. Mishra, A. Ingale, U.N. Roy, A. Gupta, Study of annealing-induced changes in CdS thin
films using X-ray diffraction and Raman spectroscopy. Thin Solid Films 516, 91–98 (2007)
48. E.S. Freitas Neto, N.O. Dantas, S.W. da Silva, P.C. Morais, M.A. Pereira-da-Silva, A.J.D.
Moreno, V. López-Richard, G.E. Marques, C. Trallero-Giner, Temperature-dependent Raman
study of thermal parameters in CdS quantum dots, Nanotechnology 23, 1–8 (2012)
49. H. Hartmut, S.W. Koch, Quantum Theory of the Optical and Electronic Properties of
Semiconductors, 4th edn. (World Scientific, Singapure, 2004)
50. R.L. Liboff, Introductory Quantum Mechanics, 2nd edn. (Addison-Wesley, San Francisco,
1980)
51. J.D. Dow, D. Redfield, toward a unified theory of Urbach’s rule and exponential absorption
edges. Phys. Rev. B 5, 594–610 (1972)
52. M.V. Kurik, Urbach Rule. Phys. Stat. Sol. (a) 8, 9–45 (1971)
53. G.D. Cody, T. Tiedje, B. Abeles, B. Brooks, Y. Goldstein, Disorder and the optical absorption
edge of hydrogenated amorphoussilicon. J. Phys. Colloques C4, 301–304 (1981)
350 M.H. Yükselici et al.

54. V. Spagnolo, G. Scamarcio, M. Lugara, G.C. Righini, Raman scattering in CdTe1-xSex and
CdS1-xSex nanocrystals embedded in glass. Superlattices Microstruct. 16, 51–54 (1994)
55. Landolt-Borstein, Numerical Data and Functional Relationships in Science and Technology,
vol. III/41b. (Springer, Berlin, 1999)
56. X.S. Zhao, J. Shroeder, P. Persans, T. Bilodeau, Resonant-Raman-scattering and
photoluminescence studies in glass-composite and colloidal CdS. Phys. Rev. B 43, 12580–
12589 (1991)
Chapter 15
One-Dimensional Nano-structured
Solar Cells

H. Karaağaç, E. Peksu, E.U. Arici and M. Saif Islam

Abstract The solar light harvesting has long been regarded as promising way to
meet the increasing world’s annual energy consumption as well as the solution to
prevent the detrimental long-term effect of carbon-monoxide emission released by
fossil fuel sources. Due to the high cost of today’s conventional PV technology,
however, it is not possible to compete with the energy supplied from fossil fuel
sources. The use of one-dimensional nanostructures, including nanowires (NWs),
nanorods (NRs), nanopillars (NPs) and nanotubes (NTs) in solar cells with different
device architectures (e.g. axial, radial, and nanorod/nanowire array embedded in a
thin film) provides peculiar and fascinating advantages over single-crystalline and
thin film based solar cells in terms of power conversion efficiency and manufacturing
cost due to their large surface/interface area, the ability to grow single-crystalline
nanowires on inexpensive substrates without resorting to complex epitaxial routes,
single-crystalline structure and light trapping function. In this chapter, we review the
recent studies conducted on nanowire/nanorod arrays based solar cells with different
device architectures for the realization of high-efficiency solar cells at an econom-
ically viable cost.

H. Karaağaç (&)
E. Peksu

Department of Physics, Istanbul Technical University, 34469 Maslak, Istanbul, Turkey
e-mail: [email protected]
E.U. Arici
Energy Institute, Istanbul Technical University, 34469 Maslak, Istanbul, Turkey
e-mail: [email protected]
M. Saif Islam
Department of Electrical and Computer Engineering, University of California at Davis, Davis,
CA 95616, USA
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 351

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_15
352 H. Karaağaç et al.

15.1 Introduction

Today’s world annual energy consumption has been estimated to be around

15 TWYs and it has been predicted to be around 30 TWYs by 2050 [1]. To meet
this huge energy requirement, harnessing energy from the sun has long been
regarded as a promising solution due to its abundance and free from contaminations
and carbon emissions [2]. The direct conversion of solar energy is realized through
solar cells, which is based on photovoltaic (PV) effect [3]. Due to the high cost of
today’s conventional PV technology it is not possible to compete with the energy
supplied from fossil fuel sources. However, in recent years, there has been an
increasing interest in reducing the price, or cost per watt, of solar cells. In particular,
the use of nanostructures, defined as structures having at least one dimension in the
range of 1–100 nm, including nanowires (NWs), quantum wells (QWs), and
quantum dots (QDs), in solar cells is identified as promising approach to generate
inexpensive photovoltaic electricity [4]. The use of nanostructures in solar cells
offers many advantages over their conventional planer counterparts owing to their
unique properties such as implanting new physical mechanisms, tailoring the
material properties and low fabrication cost. In particular, one-dimensional (1D)
nanostructures including nanowires, nanorods (NRs), nanopillars (NPs) and nan-
otubes (NTs) provide peculiar and fascinating advantages over thin films in various
opto-electronic device applications such as solar cells and sensors, arising from
their large surface/interface area, single-crystalline structure and light trapping
function [5].
The objective of this chapter is to review recent research activities on 1-D
nanostructures based solar cells. First section of this chapter deals with the most
widely preferred 1D nanostructures based solar cell architectures/models. Next
section focuses on the most commonly employed growth techniques for the syn-
thesis of 1D nanostructures. Third section includes well-known materials (Silicon,
Zinc-Oxide, Titanium-dioxide, and carbon) based 1D nanostructures based solar
cells. The final section concludes with prospective on 1D nanostructures and their
solar cell application.

15.2 One-Dimensional Nanostructures Based Solar Cell

Architectures

In the past years, a variety of concepts have been proposed for the realization of
high-efficiency solar cells at an economically viable cost [6, 7]. Utilization of
one-dimensional (1D) nanowire/nanorods array for the construction of three-
dimensional (3D) structures is among the widely preferred approach to fabricate
high-efficiency solar cells at lower cost. In recent decades, various photovoltaic
(PV) device architectures based on nanowires/nanorods/nanopillars have been
developed to construct 3D solar cells. In particular, the device configuration based
15 One-Dimensional Nano-structured Solar Cells 353

on nanowire/nanorods array with radial, axial, and nanowire/nanorods array

embedded in thin film junctions are the most widely preferred types of PV devices
[5, 8, 9], as depicted in Fig. 15.1. Each of these device architectures has its own
advantages and drawbacks [5]. All types of these structures offer many potential
benefits over the traditional planer thin-film or wafer-based solar cells in terms of
cost, effective charge separation, strain relaxation and some electrical and optical
properties [10]. Of the aforementioned solar cell architectures, core-shell (radial)
configuration takes advantages of all these effects, such as high surface/volume ratio,
reduced surface reflection and enhanced carrier collection efficiency. The core-shell
configuration provides not only tuning the optical properties of the structure by
adjusting the size and pitch of nanowire array but also the possibility to improve the
carrier collection efficiency by varying the core-shell radii [5]. However, in this type
device architecture a conformal coating requirement still remains a problem [11]. For
the nanowire/nanorods array embedded in a thin film configuration, although there is
an enhanced carrier collection efficiency, its optical properties are almost similar to
thin film or wafer-based solar cell architecture owing to its planer top surface [11].
Of the 1D nanostructures based solar cell models, PV devices based on axial junction
configuration have attracted less attention due to the limitation of the geometry in
effective charge collections and the challenges in material synthesis [12]. It is,
therefore, critical to fabricate sharp axial p-n junctions and well controlled doping
level to achieve effective charge separation and charge transportation to counter
electrodes with such a device configuration [12].
The past decade has seen increasingly rapid advances in the use of
nanowires/nanorods for the realization of third-generation solar cells with either
all-inorganic or inorganic-organic hybrid material combinations based on one of the
aforementioned solar cell device architectures [13, 14]. In 2005, the first Si NWs
based core-shell solar cell with a 14.5 % efficiency (theoretically) was proposed
[15]. On the other hand, the efficiency of first Si NW solar cell having core-shell
architecture with thin amorphous n-Si on p-Si NWs was only *0.1 % [16]. The
low efficiency of the constructed device was attributed to junction quality and
impurities, which are assumed to be key parameters having direct effect on deter-
mining the efficiency of this type device structure. Another study based on Si NW
core-shell solar cell was reported by Garnett and Yang using two-step procedures
[9]. N-type Si NW array was derived from Si wafer using chemical etching, which

Fig. 15.1 One-dimensional nanostructures based solar cell architectures: a core-shell, b axial,
c nanowire/nanorod array embedded in thin film junctions [5]
354 H. Karaağaç et al.

was sequentially decorated with a p-type Si layer using chemical vapor deposition.
The solar cell parameters of the device exhibited a Jsc (short-current density) of
4.28 mA/cm2, Voc (open-circuit voltage) of 0.29 V and FF (fill-factor) of
0.33. Based on these parameters, the power conversion efficiency (PCE) was cal-
culated to be 0.46 %. The measured low PCE with this device geometry was
attributed to the existence of severe interfacial recombinations and high series
resistance, suggested to be overcome by surface passivation of Si NWs and
reducing the bulk resistivity of p-Si cap layer.
There are also a number of studies in literature on nanowire array embedded in
thin film type solar cells. For instance, Yang et al. demonstrated an all-oxide solar
cell, in which n-type ZnO NW array covered by a p-type Cu2O nanoparticles was
utilized to realize environmentally benign stable solar cells [17]. It was revealed that
the morphology, the thickness and the phase of Cu2O nanoparticles were the key
factors determining the performance of solar cell. It was also realized that when an
immediate blocking layer introduced between the electrodes (having direct contact
with NWs) and the absorber layer, there was a dramatic increase in the overall
efficiency of the solar cell. For instance, a 10 nm TiO2 blocking layer between the
transparent electrode and Cu2O resulted in 50 times higher efficiency (0. 055 %)
with respect to the device configuration without the blocking layer. In another
study, Fan et al. [18] demonstrated a solar cell consisting of n-CdS nanopillar array
grown by CVD in a porous anodic alumina membrane (AAM) embedded in p-CdTe
polycrystalline thin film by CVD so as to enhance light absorption and carrier
collection efficiency. The performance of solar cell was tested under different
illumination conditions, ranging from 17 to 100 mWcm−2. For instance, under
A.M.1.5 G illumination condition, the n-CdS/p-CdTe nanopillars solar cell exhib-
ited a power conversion efficiency of *6 %, which is known to be higher than
values that have been reported so far for the solar cell devices utilizing the same
concepts such as the orthogonalizing the direction of light absorption and
photo-generated carrier collection. Results of this investigation suggested that the
power conversion efficiency achieved, *6 %, could be improved by applying
several device-fabrication, contact issues, and nanopillars-geometry related opti-
mization processes including the adjustment of the inter-pillar spacing of CdS pillar
array, the transparency and resistances of contacts, coating nanopillars with a
antireflection layer, and tuning the penetration depth of CdS nanopillars in the
polycrystalline CdTe thin film layer. Additionally, in order to realize a flexible solar
cell, CdS/CdTe nanopillars solar cell was constructed in a plastic layer
(Polydimethylsiloxane, PDMS), which exhibited almost the same power conversion
efficiency that obtained for the solar cells constructed on the rigid substrates.
Furthermore, under different bending conditions device showed a little change in
the cell performance. In a different study, single nanowire solar cells consisting of
GaAs nanowires in radial p-n configuration were reported [8], which exhibited a
0.8 % power conversion efficiency with FF * 0.26. The same group also inves-
tigated numerical simulations of current-voltage characteristics for III–V nanowire
core-shell architecture in order to find the optimum nanowire geometry for the
realization of high-efficiency solar cells [19]. Their analysis revealed that a *30 %
15 One-Dimensional Nano-structured Solar Cells 355

power conversion efficiency could be achieved using the core-shell device archi-
tecture by implanting effective surface passivation as well as designing the p-n
junction with top contact geometry and minimal tip height. In another major study,
Colombo et al. reported a 4.5 % power conversion efficiency for a radial p-i-n diode
nanowire solar cell based on GaAs [20]. In addition to CdS, Si, ZnO and GaAs,
core-shell solar cells based on GaN nanopillars were also reported as heterostruc-
tured solar cell with power conversion efficiency of *2.7 % under AM 1.5 G
illumination [21].
There have been relatively few recent studies on solar cells based on axial device
architecture due to the challenges in synthesis of materials and the limitations in the
charge collection. Sivakov et al. constructed an axial architecture based solar cell
composed of multicrystalline p+-n-n+-doped Si layer using electroless etching
technique [22]. To form axial geometry, first p+, n, and n+ amorphous Si films were
deposited alternately by electron-beam evaporation technique, which was followed
by crystallization using a laser beam, and then chemical etching to synthesize Si
NWs. Results demonstrated that Si NWs exhibited a significant reduction in
reflectance in the wavelength range of 300–1000 nm, when compared with its
planer counterpart. Power conversion efficiency for the device was calculated to be
1.4 % under standard measurement conditions (A.M. 1.5 illumination). In another
major study, GaAs nanowire array based solar cells with p-i-n axial junctions were
reported by Yao et al. [23]. The results of investigation revealed that GaAs
nanowire array was highly absorptive when synthesized with a low filling ratio. In
addition to this, it was found that thicker nanowire arrays without surface passi-
vation treatment were favorable owing to the high surface recombination velocity.
A power conversion efficiency of 7.58 % was achieved for the solar cell with p-i-n
axial junction consisting of a junction depth of 100 nm and nanowire diameter of
320 nm, indicating that GaAs nanowire array with axial geometry is among the
promising candidates for the realization of light-efficiency and low-cost solar cells,
which are regarded as next generation photovoltaics.

15.3 Synthesis of One-Dimensional Nanostructures

In this section, the most widely employed fabrication methods for the synthesis of
nanostructures will be reviewed. In particular, the synthesis of one-dimensional
nanostructures based on top-down and bottom-up methods will be introduced.
In general, top-down and bottom-up methods are regarded as two different
approaches that embody all the fabrication methods for the synthesis of 1-D
nanostructures. Fabrication methods based on top-down approach includes the
reduction of the size of bulk material to the nanoscale by using combined litho-
graphic steps and etching process. It is well known fact that this approach was first
employed in the microelectronic industry. However, following the advances in thin
film deposition and lithography techniques this approach has been used to syn-
thesize one-dimensional nanostructures as well. Although a great success has been
356 H. Karaağaç et al.

achieved in the fabrication of 1-D microstructures by using top-down approach, the

limit in the size of features (width, diameter) that can be formed imposed by
lithographic patterning (e.g., electron beam) has constrained its wide use in the last
half century [24]. However, this goal can be realized by bottom-up approach,
regarded as a more promising way for the synthesis of 1-D nanostructures due to its
precisely controlled growth dynamics. In other words, this approach mainly based
on the synthesis of 1-D nanostructures from some basic blocks such as atoms,
molecules and clusters in a precisely controlled fashion.
Generally, the methods for the fabrication of 1-D nanostructures are presented
under two different routes. These are solution- and vapor-based techniques. Here,
several commonly employed fabrication techniques for the synthesis of 1-D
nanostructures will be presented.

15.3.1 Solution-Based Synthesis of 1-D Nanostructures

The fabrication of one-dimensional (1-D) nanostructures along one specific orien-

tation requires the break of growth symmetry that can be accomplished in a wide
range of solution-based routes.
Hydrothermal technique is one of the most commonly used solution-based
techniques. Among several other solution-based techniques, hydrothermal is of a
special interest due to its many advantages such as its low growth temperature, a
variety of control parameters for tuning properties, and allowing the input of various
elements into the crystal at relatively low temperatures [25]. The synthesis process
using hydrothermal technique involves the use of a mixture of precursor and reagent
to either regulate or template crystal growth like amine, which is followed by a series
of chemical reactions under different pressures and temperatures. So far, a wide
variety of 1-D nanostructures, such as ZnO NWs, TiO2 NRs and MnO2 NWs, have
been synthesized using this technique [26–28]. Hydrothermal technique is particu-
larly preferred to synthesize vertically well-aligned ZnO NWs that can be employed
for a wide variety of opto-electronic devices [29–36]. For instance, recently, we
demonstrated a 3D photodetectors based on ZnO NWs grown by hydrothermal
technique (Fig. 15.2) [37].
Some of solution-based techniques are particularly convenient for certain types
of materials. For instance, metal assisted chemical etching (MACE) is widely used
to derive Si NWs from single-crystalline silicon (Si) wafers [38]. The fabrication of
Si NWs can also be accomplished by using a number of techniques, such as
vapor-liquid-solid (VLS) growth, electrochemical, or lithography-related etching
routes [39]. However, most of these techniques are requiring complicated expensive
equipments with demand of high vacuum and temperature. On the other hand,
MACE is a simple and a low-cost technique that can be employed for the synthesis
of vertically well-oriented Si NWs. It offers several advantages over the afore-
mentioned techniques including its ability to tune the physical parameters (e.g.,
diameter, length, and density) by adjusting growth parameters (e.g., growth
15 One-Dimensional Nano-structured Solar Cells 357

Fig. 15.2 SEM images of Si micro pillars/walls decorated with ZnO NWs grown by hydrothermal
technique: a Si microwalls and b Si micropillars decorated with AZO seed layer (100 nm), ZnO
NWs grown on c Si microwalls and d Si micropillars

temperature, precursor concentration and growth time), no requirement of expen-

sive equipment and allowing mass producible fabrication of Si NWs. In a typical
MACE process, a Si wafer first covered by a metal, which is subsequently
immersed in a solution consisted of hydrofluoric (HF) acid and an oxidative agent
[40]. Since the metal coated parts of Si wafer has a faster etching rate compared
with uncoated parts, it leads to a selective etching as metal nanoparticles propagates
through the Si wafer. Based on transferred metal (e.g., silver, gold, copper) pattern
and the nature of used solution (AgNO3/HF or H2O2/HF), Si nanostructures with
various morphologies (e.g., cylinder, pillar, pyramid, wire, holes) can be realized
[40]. In MACE process, a noble metal, such as silver (Ag), behaves as local cathode
and injects holes into Si substrate in contact with metal layer, resulting in formation
of SiO2. Since HF attacks SiO2, it can be easily etched away. In this way, Si
nanostructures with a specific pattern and morphology can be synthesized, as shown
in Fig. 15.3 [41].
One-dimensional nanostructures such as nanowires can also be synthesized via
the sonochemical technique. Similar to hydrothermal technique, this route is based
on a seed formation in a specific bulk solution. The energy required for the seed
formation is supplied from ultrasonic waves as opposed to the case in hydrothermal
approach, which is based on a temperature gradient driven process [42]. In a typical
sonochemical process, seeds are formed by a phenomenon called cavitation.
Regions where the pressure of liquid is below the vapor pressure are acting as the
source of these cavitations, which are generated by ultrasonic waves (15–400 kHz).
In other words, ultrasonic waves triggers first the bubble formation, then col-
laption and finally energy release that can be used for the seeding. The localized
358 H. Karaağaç et al.

Fig. 15.3 One and two dimensional Si nano/micro structures synthesized using MACE process:
a Si-pillars, b Si-microwalls, c Si-microholes with specific pattern, c, d different Si-micro
geometrical features and f Si-nanowires

heat release during the collapse of bubbles is around 5000 K. It is this high local
heat release that triggers the formation of seeds at extreme conditions. In recent
years, there has been an increasing amount of literature on the growth of 1-D
nanostructures by using sonochemical approach. For instance, in 2002, Gates and
co-workers reported the synthesis of selenium (Se) nanowires using this technique
[43]. To grow Se NWs first an aqueous solution composed of selenious acid and
hydrazine was used to form Se colloids. After filtering these colloids through a
membrane, they were dried and then dispersed in an alcoholic solution. Following
these stages, solution was sonicated for about 30 s (42 kHz, 0.15 W/m2). The
applied ultrasonic waves resulted in the formation of clusters of the Se colloids with
irregular shapes, acting as seeds responsible for the growth of Se NWs based on
solid-solution-solid mechanism. In another study, Kumar et al. reported the syn-
thesis of magnetic nanorods using sonochemical process [44]. The sonochemical
method was also employed for the growth of ZnO nanowires. We recently reported
the synthesis of ZnO NWs in a single step, which took less than an hour growth
cycle [45]. This technique allowed us to grow ZnO NWs on virtually any substrates
(Si, Cr, polymer, cotton and human hair) as long as they withstand both water and
15 One-Dimensional Nano-structured Solar Cells 359

Fig. 15.4 SEM images of ZnO nanowires grown on a1–a3 human hair, b cotton c Silicon (Si),
d chromium (Cr) and e1–e3 organic polymer (PMMA) substrates using sonochemical technique

alcohol, as shown in Fig. 15.4. Although sonochemical technique is relatively rapid

and non-hazardous, it is very challenging to control the shape and size of nanowires
(e.g., length and orientation).

15.3.2 Vapor Phase-Based Synthesis of 1-D Nanostructures

Vapor-phase based synthesis of one-dimensional (1-D) nanostructures are partic-

ularly preferred for a certain material systems [42]. The growth kinetics behind the
vapor-phase based methods are generally catalogued into vapor-liquid-solid (VLS)
and vapor-solid (VS) mechanisms.

15.3.2.1 Vapor-Liquid-Solid (VLS) Mechanism

The VLS growth mechanism involves selectively reaction of vapor-phase precur-

sors on a substrate pre-coated with patterned catalyst layer. A selective growth is
achieved through the catalyst particles, which form a eutectic liquid solution with
nanowire material.
Following the dissolution of precursors from gas phase into the liquid drops, the
supersaturation of solution results in precipitation of nuclei, which act as seed
points for the growth of 1-D nanostructures [46]. During the growth process, both
the diameter and location of 1-D nanostructures can be precisely controlled via the
position and droplet size of catalysts; hence 1-D nanostructures with a specific
360 H. Karaağaç et al.

pattern and density can be produced [47]. A schematic representation for the growth
of Si NWs by VLS mechanism is presented in Fig. 15.5.
The use of 1-D nanostructures by VLS mechanism was first reported by Wagner
and co-workers to grow the unidirectional Si crystal growth [49]. There is a large
volume of published studies on VLS mechanism based synthesis of 1-D nanos-
tructures of a wide range of materials including elemental semiconductors (Ge and
Si), III–V semiconductors (GaAs, GaN), II–VI semiconductors (ZnS, CdS and
CdSe), and oxides (ZnO and SiO2) [47, 50–54]. For instance, recently, we syn-
thesized Si NWs via a VLS process by using Au thin film catalyst deposited on Si
substrate for transfer printing from Si mother substrate to any carrier substrate, as
shown Fig. 15.6 [55].

Fig. 15.5 VLS growth mechanism for Si nanowires using Au catalyst [48]

Fig. 15.6 SEM images of Si nanowires (NWs) grown by VLS process. a Si NWs grown on Si
(111) wafer using Au metal catalyst, b–d high magnification SEM images of transferred VLS
grown Si NWs from Si wafer and PMMA polymer (embedded in polymer)
15 One-Dimensional Nano-structured Solar Cells 361

Fig. 15.7 A typical VS growth mechanism [57]

15.3.2.2 Vapor-Solid (VS) Mechanism

In contrast to VLS route, a catalyst is not employed in a typical vapor-solid

(VS) route to synthesize one-dimensional nanostructures. In this process, vapor
phase of 1-D nanostructure material is generated either by decomposition of pre-
cursors or thermal evaporation, which is subsequently transferred by carrier gases
(e.g., N2 or Ar) and condensed on a substrate kept a lower temperature [56], as
shown in Fig. 15.7.
As long as one can control some growth parameters, particularly nucleation, it is
possible to synthesize 1-D nanostructures by VS mechanism. During the growth
cycle of 1-D nanostructures, it is generally believed that nucleation mechanism is
driven by either certain defects (e.g., screw dislocation) or anisotropic growth.
Although the growth mechanism is not well-understood, it is suggested that at the
initial stage of nuclei formation there is a competition between certain crystallo-
graphic directions having different growth rates. In the end, one of these directions
wins this race and hence promotes the formation of 1-D nanostructure in this
favored direction [58]. By now, a considerable amount of literature has been
published on VS route based synthesis of one-dimensional nanostructures including
nanowires (Si3N4, SiC), metal-oxides (oxides of Zn, In, Al, and Sn), nanobelts
(ZnO, In2O3, and SnO) [56, 59, 60] and carbon nanotubes [61–63].

15.4 Common Materials for 1-D Nano-Structured Solar

Cells

In this section, we mainly introduce several commonly used materials such as

silicon, zinc-oxide and titanium-dioxide that we have employed in our recent
studies for the fabrication of one-dimensional nanostructures based solar cells.
362 H. Karaağaç et al.

In addition, we review the studies on the utilization of carbon nanotubes (CNTs) in

dye sensitized solar cells as well as strategies devoted to the development of CNTs
based more efficient photoanodes.

15.4.1 Silicon

The vast majority of today’s solar cells is based on silicon (Si) material due to its
various advantages including its abundance, stability, high reliability, and well
established processing techniques. However, the requirement of high quality single
crystalline Si for the power conversion efficiencies over 20 % increases the cost
tremendously. To reduce the cost, recent research efforts are mainly focused on
structural changes in devices to reach highly efficient, low cost solar cells. In this
regard, Si nanowires (NWs) have received significant interest because of the intro-
duction of new physical mechanisms such as quantum confinement, allowing for the
tuning of material properties and overcoming the theoretical efficiency limits for
conventional planar p-n junction solar cells. To date a considerable amount of liter-
ature has been published on Si nanowires/nanorods based solar cells. However, this
subsection will mainly focus on our studies based on synthesis of Si NWs by elec-
troless etching technique and solar cells based on their integration into
p-AgGa0.5In0.5Se2 (AGIS) chalcopyrite thin films. In addition, our study on the
fabrication of a hybrid solar cell based on Si-NWs/PCBM will be introduced in detail.

15.4.1.1 n-Si-NWs/p-AGIS Nanowires Embedded in a Thin Film Type

Solar Cell

Over the last two decades, a large a growing body of literature have been reported
on developing high efficient low cost inorganic solar cells. A wide variety of
materials are employed for achieving this goal, including poly-crystalline Si,
amorphous Si, polymers, II–VI and I–III–VI2 chalcopyrite (I–III–VI2) compounds.
In particular, chalcopyrite materials are strong candidates for the realization of
highly-efficient solar cells based on the construction of p-n heterojunctions utilizing
II–VI semiconductors due to their similar crystal structures, which is the key factor
to reduce lattice mismatch in the device structure.
Copper (Cu–III–VI2) and silver (Ag–III–VI2) based chalcopyrites are well
known members for their high absorption coefficient (105 cm−1) and an optimum
bandgap energy for the solar spectrum [64, 65]. So far, a number of studies have
been reported on Ag based chalcopyrite thin film solar cells including n-Ag(In1
−xGax)Se2/CdS:ZnO(Al) [66], p-AgInSe2/n-CdS [67], p-AgGaSe2/n-CdS [68] and
p-AgGa0.25In0.75Se2/n-Zn0.35Cd0.65S [69] heterojunctions with power conversion
efficiencies ranging from 4.5 to 7.8 %. In addition to incorporating new materials
into solar cells, structural changes in devices may also serve the goal to reach high
efficient solar cells at lower cost. Although there are many studies reported on one
15 One-Dimensional Nano-structured Solar Cells 363

of these two major research areas individually, so far, however, too little attention
has been paid to combine these studies. In this sense, introducing Si NWs into
chalcopyrite systems is very promising for fabricating high efficient inorganic solar
cells.
In our recent study we reported the synthesis of Si NWs and solar cells based on
their integration into p-AgGa0.5In0.5Se2 (AGIS) chalcopyrite thin films. It was the
first study reporting fabrication and characterization of Si NWs/p-AGIS hetero-
junction based thin film solar cells.
Si NWs were synthesized using metal assisted chemical etching (MACE)
technique, details of which is given in earlier section. In order to synthesize Si
NWs, an n-type Si wafer was immersed in a solution consisting of 4.6 M hydro-
fluoric acid (HF) and 0.002 M silver nitride (AgNO3) [70]. The fabrication of Si
NWs was accomplished based on the oxidation-reduction reactions took place in
solution. After the synthesis of n-Si NWs, they were decorated with AGIS thin films
(800 nm) to form p-n heterojunction based solar cells (see Fig. 15.8) [64].
Following this, indium (In) metallic top contacts with dot pattern were deposited by
thermal evaporation. For the back contacts of the solar cell, Si wafer was coated
with silver (Ag) by direct current (DC) sputtering.
For the synthesis of Si NWs, we have preferred the MACE technique applied to
n-type (100) Si wafers (10 Ω cm resistivity). From the previous studies it is known
that MACE route weakly dependent upon the orientation and doping type of the
silicon wafer [71]. The growth parameters such as solution temperature and growth
time were 60 °C and 5 min, respectively. The cross sectional SEM images of
synthesized Si NWs are given in Fig. 15.9. From these images, it is apparent that that
there are vertically well oriented Si NWs on the Si substrate with an etching depth of
*700 nm corresponding to the average NW length. Using MACE technique it is
possible to tune the lengths and radii of NWs by adjusting growth parameters such as
growth time, etchant concentration and the solution temperature [70].
Once the Si NWs were synthesized, AGIS thin films were deposited by
electron-beam technique to realize Si NWs embedded in a thin film type solar cell.
To determine the structural, chemical and optical properties of the AGIS thin films
prior to the fabrication of p-n junction, AGIS films were deposited onto glass
substrates as well. In addition, in order to understand the effect of post-annealing on

Fig. 15.8 Three-dimensional schematic views of the fabricated n-Si NWs/p-AGIS heterojunction
solar cell
364 H. Karaağaç et al.

Fig. 15.9 SEM images of synthesized Si NWs using electroless etching techniques

Fig. 15.10 Transmission 0.8

2e+10
spectra recorded for as-grown as-grown
and AGIS films annealed at o o
300 C
2
2e+10 400 C
(αhν) (cm eV)
o
different temperatures. Inset 0.6
-1
400 C
figure presents (αhν)2 versus 1e+10 Eg=1.56 eV
Transmittance

(hν) relation for AGIS film

2

5e+9
annealed at 400 °C
0.4
1.4 1.6 1.8 2.0
hν (eV)

0.2

0.0
300 400 500 600 700 800 900
Wavelength (nm)

the properties of the films, AGIS thin films were subjected to annealing process in
the temperature range of 300–400 °C for 30 min in N2 gas atmosphere.
Transmission measurements were carried out in the wavelength range of
325–900 nm to determine the band gap of the deposited AGIS thin films. The
spectrum obtained for both as-grown and annealed AGIS thin films are shown in
Fig. 15.10. As can be seen from the transmission spectra, the average transmittance
is around 40 % at 800 nm for all samples. Absorption coefficient (α) was evaluated
via the transmission spectra. Through the relation between photon energy (hν) and
absorption coefficient, (αhν) = A(hν − Eg)n, the nature and energy value of the band
gap of AGIS films were determined from the plot of (αhν)2 versus (hν), shown as an
inset in Fig. 15.10 for an AGIS film annealed at 400 °C. The results obtained from
the analysis show that there is a direct nature (n = 1/2) of the optical transition and
the optical band gap value is around 1.56 eV, which is exactly matching the most
abundant part of the solar spectrum (1.3–1.5 eV).
A nanowires embedded in a thin film type solar cell was constructed by
embedding n-Si NWs in a p-AGIS poly-crystalline thin film. The cross-sectional
view of the fabricated solar cell with AgGa0.5In0.5Se2/n-Si (NWs)/n-Si structure is
presented in Fig. 15.11. The dark current (I)–voltage (V) characteristic of
In/p-AGIS/n-Si NWs/n-Si/Ag solar cell was tested at room temperature to determine
the formation of p-n junction between n-Si NWs and p-AGIS thin film. As shown in
15 One-Dimensional Nano-structured Solar Cells 365

Fig. 15.11 SEM image of the fabricated n-Si NWs/p-AGIS structured solar cell (cross sectional
view)

Fig. 15.12 Current (I)- 10 -2

voltage (V) characteristic of
n-Si NWs/p-AGIS 10 -3
heterojunction measured in 10 -4
dark condition
Current (A)

10 -5

10 -6

10 -7

10 -8

10 -9

10 -10
-3.2 -2.4 -1.6 -0.8 0.0 0.8 1.6 2.4 3.2
Voltage (Volt)

Fig. 15.12, the device shows the typical p-n junction characteristics with rectification
ratio (Iforward/Ireverse) of 100 at 1 V, which verifies the formation of p-n
heterojunction.
The performance of fabricated solar cell was tested by a solar simulator under
AM 1.5 illumination (100 mW/cm2). The I–V characteristic recorded under the
replicated solar spectrum is presented in Fig. 15.13. The power conversion effi-
ciency (η) of a typical solar cell can be calculated from

Voc Isc FF
g¼ ð15:1Þ
Pin

where Voc, Isc, FF and Pin are open circuit voltage, short circuit current, fill factor
and incident power, respectively.
The Voc, Jsc, and FF of the solar cell were found to be 0.34 V, 25.38 mA/cm2 and
63 %, respectively, corresponding to a power conversion efficiency of 5.50 %.
Besides these, series (Rs) and shunt (Rsh) resistances, showing the quality of solar
cell, were also calculated through the slope of the I–V curve near the Jsc and Voc,
which were calculated to be 32 kΩ and 15 Ω, respectively.
366 H. Karaağaç et al.

Fig. 15.13 I–V characteristic

of n-Si NWs/p-AGIS
heterojunction solar cell
recorded under the replicated
solar spectrum

In reviewing the literature, no studies were found on a solar cell based on n-Si
NWs embedded in p-AGIS thin films; therefore, it is not possible to compare our
results with literature. Although constructed with different materials combination, a
solar cell with a similar architecture, based on n-CdS nanopillar arrays embedded in
a p-type polycrystalline CdTe thin film, was reported in literature [18]. A 6 %
power conversion efficiency was achieved by embedding CdS nanopillars into
CdTe thin film.
To date, a numerous studies have been reported in the literature on fabrication of
Ag-based chalcopyrite thin film heterojunction planer solar cells constructed with a
wide variety of materials including CdS: ZnO (Al), n-CdS and n-Zn0.35Cd0.65S thin
films with power conversion efficiencies ranging from 4.5 to 7.8 % [66–69]. In our
study, the efficiency achieved with Si NWs based solar cell lies in range of reported
values. For instance, when compared with n-Ag(In1−xGax)Se2/CdS:ZnO(Al) planar
thin film heterojunction solar cells, we have measured an almost two times larger
short circuit current density with the n-Si NWs/p-AGIS solar cell structure. The
enhancement in current density is expected outcome for a hybrid nanowire-thin film
heterojunction solar cells when compared with its planar counterpart since it
provides both light trapping and efficient charge collection to the electrodes for
generated free carriers. In addition, in a typical planer thin film based solar cell,
there is high probability of formation of recombination and trap centers, reducing
the collection efficiency of generated free carriers, due to the polycrystalline
structure of films (e.g., grain boundaries, point defects). That’s why, although the
two architectures have different n-type materials, the observed high short current
density (25.38 mA/cm2) for n-Si NWs/p-AGIS relative to planar n-Ag(In1−xGax)
Se2/CdS:ZnO(Al) (14.50 mA/cm2) is quite reasonable, which highlights the unique
advantages of our constructed solar cell based on using Si nanowires.
Although the energy conversion efficiency obtained from n-Si NWs/p-AGIS solar
cell is lower than that reported for its planer counterparts constructed with silicon and
some Ag-chalcopyrite thin films, it is possible to realize more efficient NW based
15 One-Dimensional Nano-structured Solar Cells 367

solar cells by optimization of some fabrication steps including NW synthesis, thin

film layer deposition, contact issues, and surface passivation processes.
Among these optimization processes, for instance, surface passivation alone would
increase the energy conversion efficiency of devices to over 5.50 %. Up to now, many
research groups have focused on the effect of passivation on efficiency of solar cells
based on Si NWs. Gunavan and Guha reported that with incorporation of 10 nm thick
Al2O3 conformal passivation layer onto Si NWs’ surface the solar efficiency increased
from 1 to 1.8 % by increment of 47 mV in Voc and 5 mA/cm2 in Jsc [72]. In another major
study, Kim et al. reported the introducing a-SiN:H for the passivation of Si micro-wire
based solar cell [73]. It was observed that by introducing the passivation the power
conversion efficiency was improved from 8.9 to 10.6 %.
In our study, although the AGIS has been employed as absorber thin film layer
in which Si NWs were embedded, the same device architecture can be used with a
wide variety of other PV materials including CdTe, CdS, and CuInSe2 to improve
the energy conversion efficiency beyond 5.50 % achieved in this work. In addition,
as we demonstrated in our previous studies, Si NWs, micro-walls and micro-pillars
transfer printed from the single crystal wafer to any arbitrary substrate [55, 74, 75]
via bottom-up methods may considerably reduce the cost of such solar cells, as
shown in Fig. 15.14.

Fig. 15.14 a Schematic representation of vertical transfer printing process for 1-D and 2-D Si
micro/nano structures, b the vertical embossing tool used during transfer printing, c Si micro
pillars (MPLs) produced by deep reactive ion etching (DRIE), d and e SEM images showing root
location of the MPLs on the mother substrate before and after transferring process, respectively,
f–i transferred MPLs embedded in polymer, j Si microwalls (MWAs) produced by DRIE, k roots
of MWAs following the transfer printing, m and n low and high magnification SEM images of
MWAs transferred into polymer, respectively, p Si NWs on Si wafer synthesized by using
electroless etching technique, r and s show the high magnification SEM images of the
electroless-etched Si NWs transferred from mother substrate
368 H. Karaağaç et al.

15.4.1.2 Si-NWs/PCBM Structured Hybrid Solar Cell

In the past few years, an intense interest has emerged in combining an organic
material with NWs for the fabrication of hybrid solar cells due to their significant
advantages over the conventional organic solar cells [76–80].
The requirement of charge separation at an interface and low diffusion lengths of
excitons in a typical organic material limits the efficiency of organic solar cells.
A reasonable approach to tackle these problems could be to integrate nanowires
(NWs) into a photoactive organic material. In recent years, there has been an
increasing amount of literature on incorporation of NWs, such as Si, CdS, ZnO, and
TiO2, into an organic film [81]. For instance, recently, we have reported a hybrid
solar cell based on integration of Si NWs into PCBM thin film layer [82]. The aim
of this study was to increase diffusion length of excitons in the organic layer and
enhancing the interface area that can be realized through the integration of NWs due
to their large surface/volume ratio and ability to provide single crystalline channels
for photo-carrier charge collection. It is the first study reporting the fabrication of a
hybrid solar cell based on PCBM coated vertically well-aligned p-Si NW arrays.
P-type Si nanowires (NWs) were synthesized using a p-type Si wafer (with
(100) orientation and (1–10 Ω cm) resistivity) by electroless etching technique. On
completion of Si NWs synthesis, p-type PCBM (120 nm) polymer solution was
spin coated onto the substrate. Following this, the device annealed at 120 °C for
30 min under vacuum. For the top contact electrodes dot patterned (1.6 mm2) ITO
(200 nm thick) were deposited using RF sputtering, while for the back contacts a
250 nm thick Al was deposited onto the back side of the Si wafer. The schematic
representation of the device structure and its energy level diagram are shown in
Fig. 15.15 [83].
The morphology of synthesized Si nanowires (NWs), and the structure of the
fabricated hybrid solar cell were revealed via the SEM micrographs, shown in
Fig. 15.16. It can be seen from the SEM images that average length and diameter of

Fig. 15.15 a Schematic representation of p-Si NWs/PCBM/ITO structured hybrid solar cell and
b its energy level diagram
15 One-Dimensional Nano-structured Solar Cells 369

Fig. 15.16 a Tilted and b cross sectional view of synthesized Si NWs by electroless etching
technique. Cross sectional SEM images of Si NWs first coated with c PCBM organic thin film and
then with d transparent conductive oxide (ITO) top contacts

Fig. 15.17 The I–V 16

characteristic of the Al/p-Si
NWs/PCBM/ITO structured -8
hybrid solar cell measured
Ln(J)(A/cm )

12
2

under dark condition -10

-5
Current (A)x10

8 -12

0.0 0.1 0.2 0.3

4 V (V)
Dark
0
-2 -1 0 1 2
Voltage (V)

Si NWs are around 3 µm and *70 nm, respectively. It is also apparent from these
images that the deposited PCBM by spin coating resulted a conformal thin film
layer over the Si NWs matrix. The component layers of p-Si NWs/PCBM/ITO
structured hybrid solar cell can be easily distinguished from the cross-sectional
view shown in Fig. 15.16d.
Figure 15.17 presents the I–V characteristic of the Al/p-Si NWs/PCBM/ITO
structured solar cell under dark condition. It is apparent from the graph that the
device exhibits a standard rectifying diode characteristic with a current rectification
ratio of 78 at 0.5 V, verifying the heterojunction between Si NWs and PCBM
organic layer.
The I–V characteristics of the solar cell under visible (VIS, 1050 nm) and
ultra-violet (UV, 350 nm) light illuminations are presented in Fig. 15.18a, b,
370 H. Karaağaç et al.

Fig. 15.18 The I–V (a)

characteristics of the Al/p-Si
NWs/PCBM/ITO structured 0
Dark
hybrid solar cell measured
under different a visible and

-5
Current (A)x10
b UV light illumination -2
intensities VIS1

-4
VIS2

-6 Visible Light
VIS3

-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2

Voltage (V)

(b) 0.1
Dark
0.0

UV1
-0.1
-5
Current (A)x10

-0.2
UV2

-0.3

-0.4

-0.5 UV3 UV Light

-0.6
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2
Voltage (V)

respectively. From these figures we can see that the measured current gradually
increases when the intensity of both UV (UV3 > UV2 > UV1) and VIS
(VIS3 > VIS2 > VIS3) lights is increased, a clear verification of that both p-Si NW
and PCBM components in the device structure contribute the photocurrent. In other
words, it is reasonable to expect a contribution to photocurrent in UV and VIS part
of spectrum from Si NW and PCBM since the respective band gaps of these
materials are 1.1 and 2 eV (energy difference between the PCBM-LUMO and
PCBM-HOMO) [83]. These results suggest that this type of a hybrid heterostructure
of p-Si NW/PCBM can be utilized to fabricate portable, low cost and high efficient
UV or VIS sensitive photodiodes.
In order to determine the photovoltaic behavior of the p-Si NW/PCBM hybrid
solar cell, the J–V (current density vs. voltage) characteristic was tested under AM
1.5G illumination, as presented in Fig. 15.19.
The Voc, Isc, FF and η of the fabricated hybrid solar cell were found as 0.19 V,
8.1 A/cm2, 27 and 0.41 %, respectively. To our best knowledge, this is the first
study reported on integration of Si NWs into the PCBM organic thin film. When the
15 One-Dimensional Nano-structured Solar Cells 371

Fig. 15.19 The photovoltaic 60

behavior of the p-Si 8

J(mA/cm )
Light

2
NWs/PCBM/ITO structured
hybrid solar cell tested under 40
4
AM 1.56 illumination

J (mA/cm )
2
20 0
0.0 0.1 0.2
V (V)
0

-20
Dark
Light

-40
-1.5 -1.0 -0.5 0.0 0.5 1.0
V (V)

performance of this solar cell is compared with its planar p-Si/PCBM hybrid solar
cell (tested under 6 mW/cm2), with Voc = 0.12 V and Isc = 0.1 µA reported in
literature [76, 83], there is significant improvement in all parameters following the
integration of Si NWs into PCBM layer. This improvement can be attributed to high
efficiency of carrier collection, light trapping and formation of high interface
between Si NW and PCBM provided by unique feature of NWs. The power con-
version efficiency of 0.41 % recorded with Al/p-Sİ NW/PCBM/ITO structured
hybrid solar cell can be further improved by several optimization process including
the geometry and size of nanowires, passivation process and some contact issues.

15.4.2 Zinc-Oxide

In this part of chapter, we will mainly focus on our recent study based on synthesis
of ZnO NWs by hydrothermal technique and solar cells based on their integration
into a chalcopyrite thin film (p-AgGa0.5In0.5Se2 (AGIS)) and decoration around
silicon-micro pillars/walls. These are the first results on fabrication and character-
ization of n-ZnO NWs/p-AGIS heterojunction based thin film solar cells. In addi-
tion, we will briefly review the research conducted on ZnO one-dimensional
nanostructures based solar cells with different architectures.

15.4.2.1 n-ZnO NWs/p-AGIS Heterojunction Based Thin Film Solar

Cells

ZnO is an n-type wide band gap (3.37 eV) semiconductor oxide with a large exciton
binding energy (60 meV). It can be employed for the fabrication of a wide range
of high-performance opto-electronic devices because of its unique properties.
372 H. Karaağaç et al.

In particular, zinc-oxide (ZnO) nanowires (NWs) have been extensively studied for
a large number of applications including field ionization gas sensors, light emitting
diodes, solar cells, photodetectors, chemical sensors and field effect transistors due
to their unique physical properties as well as their excellent chemical and thermal
properties [84–90].
So far, many works have been reported on one-dimensional ZnO nanostructures
based solar cells [91–98]. We reported a nanowires embedded in a thin film type
solar cell consisted of ZnO NWs and a chalcopyrite thin film [33]. Using
hydrothermal technique, vertically well-oriented ZnO NWs were grown onto both a
glass substrate pre-coated with indium thin oxide (ITO) and n-Si wafer substrate. In
order to synthesize ZnO NWs, the required ZnO seed layers with a wide variety of
thicknesses (10–750 nm) were deposited onto ITO/glass and n-Si wafer substrates
using radio-frequency (RF) magnetron sputtering. Following this, seed layer
pre-coated substrates were immersed in a solution consisting of equimolar 25 mM
(Zn(NO3)2) 6H2O (Zinc nitrate hexahydrate, Sigma Aldrich) and HMTA
(Hexamethylenetetramine, Sigma Aldrich) as precursors in deionized (DI) water.
Growth parameters such as solution temperature and the growth time were selected
to be 90 °C and 2–3 h, respectively. For the purpose of construction of a p-n
heterojunction based solar cell, ZnO-NWs/ITO/glass-substrates (with 750 nm ZnO
seed layer) were then coated with a 800 nm thick AgGa0.5In0.5Se2 (AGIS) thin
films, details of deposition and structural and optical properties of which was
discussed in previous section and elsewhere [64].
To investigate the morphology and some physical parameters of grown ZnO
NWs such as length, diameter and density, scanning electron microscopy
(SEM) measurements were conducted. The effect of substrate on the quality of
grown ZnO NWs was investigated by growing NWs on both glass and Silicon
substrates, pre-coated with 10 nm ZnO seed layers. Figure 15.20 shows SEM
images recorded for ZnO NWs synthesized on these substrates. It can be seen from
the images that highly dense and vertically oriented ZnO NWs with a length of
about 0.6–1.0 µm and diameters in the range of 70–125 nm were grown on both
substrates. It can also be seen that the preferential growth direction of ZnO is along
the c-axis of the substrate.
To reveal the effect of the seed layer thickness on the quality of grown ZnO
NWs, ZnO seed layers with various thicknesses (10–750 nm) were prepared on
glass substrates, SEM images of which are presented in Fig. 15.21.
It was observed that when the thickness of the seed layer was increased, there
was significant improvement in quality of NWs in terms of uniformity and orien-
tation in the vertical direction. Moreover, it was deduced that when thickness was
increased, NWs with larger diameters were grown, which was attributed to the
coalescence of vertically-oriented NWs grown on the thicker seed layer. These
findings were in a good agreement with results reported in some previous studies
[99]. In general, the low-quality ZnO NWs are expected when they are grown on
thin ZnO seed layer due to the poor crystal characteristic of seed layer. It can thus
be suggested that the density and diameter of ZnO NWs can be tuned by adjusting
the thickness of ZnO seed layer.
15 One-Dimensional Nano-structured Solar Cells 373

Fig. 15.20 SEM images recorded for ZnO NWs grown on a soda-lime glass and b silicon
substrates

Fig. 15.21 ZnO NWs grown

on soda-lime glass substrates
pre-coated with a 10 nm,
b 500 nm, and c 750 nm thick
ZnO seed layers deposited by
RF sputtering technique
374 H. Karaağaç et al.

The effect of the growth time on the length of synthesized ZnO NWs by hy-
drothermal technique was also studied by keeping all growth parameters constant
except the growth time. The comparison NWs synthesized for growth times of
2 and 3 h is shown in Fig. 15.22. Results demonstrated that once the growth time
exceeded 2 h, the ZnO thin film grew instead of nanowires due to the coalescence
of longer NWs grown at the end of longer time growth period.
ZnO NWs based solar cell was constructed by integration of n-ZnO NWs into a
p-AGIS poly-crystalline thin film. The device then annealed at 400 °C in N2 gas
atmosphere for 30 min to enhance the crystallinity of absorber layer (AGIS), which
is a heat-energy driven process. The structure of constructed solar cell,
AgGa0.5In0.5Se2/n-ZnO (NWs)/ZnO (seed layer)/ITO/glass, was viewed through
the recorded SEM images, presented in Fig. 15.23. The top and bottom contacts of
the device for current collection were thermal evaporated indium (with 1 mm
diameter dots) and RF sputtered ITO layer, respectively. The dark I–V character-
istic of the device at room temperature is shown in Fig. 15.24a. It can be seen the
figure, the device exhibits a typical diode behaviour (p-n junction characteristics)
with rectification ratio (Iforward/Ireverse) of 10 at 1 V, verifying the formation of p-n
junction between ZnO NWs and AGIS thin film.

Fig. 15.22 The comparison

of ZnO NWs synthesized for
growth times of a 2 h and
b3h
15 One-Dimensional Nano-structured Solar Cells 375

Fig. 15.23 The cross sectional SEM image of Glass/ITO/ZnO (seed layer)/n-ZnO NWs/p-AGIS
structured heterojunction solar cell

Fig. 15.24 The I–V (a) 10-3

characteristics of the
fabricated solar cell under
a dark and b light 10 -4
illumination (AM 1.56)
Current (A)

10 -5

10 -6

10 -7

10 -8
-3 -2 -1 0 1 2 3
Voltage (V)

(b)
3
J (mA/cm )

Voc=0.26 V
2

2 2
Jsc=3.18 mA/cm

Rsh=20 kΩ
Rs=5 kΩ
FF=35 %
1 η=0.37 %

0
0.00 0.05 0.10 0.15 0.20 0.25
Voltage (V)
376 H. Karaağaç et al.

The photovoltaic performance of the solar cell was determined under standard
test conditions using a solar simulator (under AM 1.5 illumination (100 mW/cm2)),
J–V characteristics of which is shown in Fig. 15.24b.
The power conversion efficiency was found to be 0.37 %. There are a wide
variety of potential factors that can determine the efficiency of a solar cell. In our
case, the most probable factors behind the observed low efficiency of the ZnO NW
cell are very low shunt resistance (Rsh = 20 kΩ), high series resistance (Rs = 5 kΩ)
and some fabrication related shortcomings.
These have a direct effect on obtaining a low short circuit current density
(Jsc = 3.18 mA/cm2) and open circuit voltage (Voc = 0.26 V). In addition, the
geometry of the NWs (length, diameter and vertically orientation), the embedded
length of NWs in the thin film, the NW-thin film interface and the crystal quality of
the deposited AGIS thin film can be regarded as the potential factors responsible for
the high series and low shunt resistances. In particular, the co-existence of two
separate phases (AGIS and Ag) in thin film structure of absorber layer deduced
from XRD study may result in the formation of disorder related local defects which
promote the recombination of photogenerated free carriers.

15.4.3 Titanium-Dioxide

In recent years, TiO2 (Titanium Dioxide), one of the most important wide band gap
materials (*3.0 eV for rutile TiO2 structure), has received a considerable attention
for their potential applications in various areas including solar cells [100], water
photoelectrolysis [101], field emission devices [102], and gas sensors [103].
In particular, intense efforts have been devoted to TiO2 nanorods (NRs) based
solar cells to realize high performance solar cells at lower cost.
Among the aforementioned one-dimensional nanostructures based solar cell
architectures (axial, core-shell (or radial), and NRs embedded in thin films), the
core-shell is of particular interest to improve the power conversion efficiency over
the traditional planer thin film based solar cell due to due to its enhanced interaction
junction area, improved carrier transport to electrodes, and ability to ortogonalize
the light absorption and charge collection.
In other words, with this device architecture not only enhanced light absorption
but also efficient charge collection to electrodes are accomplished on condition that
core radius and shell thickness are optimized properly. To date a considerable
amount of literature has been published on TiO2 NRs based dye-sensitized solar
cells [104–115]. Few studies, however, have focused on TiO2 NRs based
semiconductor-sensitized solar cells [116]. Due to superior properties of TiO2 NRs
and excellent properties of CdTe as an absorber layer with an ideal band gap
matching the most abundant part of solar spectrum (1.45 eV), TiO2/CdTe is a one
of the most promising material combinations in the fabrication of high-efficient
solar cells at lower cost [117]. In this section of chapter, we introduce our recent
15 One-Dimensional Nano-structured Solar Cells 377

study based on an n-T TiO2/p-CdTe core-shell structured solar cell constructed by a

two-step route combining hydrothermal method and sputtering technique.

15.4.3.1 n-TiO2/p-CdTe Core-Shell Structured Solar Cells

Vertically-well oriented TiO2 nanorods (NRs) were grown on a fluorine-doped thin

oxide (FTO)-pre-coated glass substrate using hydrothermal growth technique, detail
of which was given elsewhere [118, 119]. After synthesis of T TiO2 NRs, in order
to form an n-TiO2/p-CdTe core-shell type solar cell, NRs were decorated with a
*250-nm-thick CdTe thin film deposited by sputtering technique. Following this,
10 nm Cu and 40 nm Au metallic layers of top contacts with dot pattern were
deposited by thermal evaporation. The device was then annealed at 120 °C for
30 min to form ohmic type contacts.
The SEM images of synthesized TiO2 nanorods (NRs) grown on a FTO glass
substrate recorded with different magnifications are presented in Fig. 15.25. As can
be seen from Fig. 15.25a, uniformly distributed TiO2 NRs were grown successfully
on FTO substrate. It can also be seen from Fig. 15.25b that the dimension of NRs is
around 1.8 µm for the length and 100 nm for the average diameter.
Figure 15.25c, d show cross-sectional and tilted view of SEM images of fabri-
cated TiO2/CdTe structured core-shell solar cell. It is worth to remark that the
diameter of core-shell TiO2 increases from bottom to top, resulting in formation of a

Fig. 15.25 a Low and b high magnification SEM images of synthesized TiO2 NRs on FTO
pre-coated glass substrates, c cross sectional and d tilted view of the completed solar cell device
structure
378 H. Karaağaç et al.

baseball bat-like core shell solar cell following the deposition of CdTe thin film
layer.
TEM images of TiO2 NRs coated with CdTe thin film are shown in Fig. 15.26.
The baseball bat-like solar cell structure is deduced from these images as well,
mainly attributable to the deposition parameters of CdTe. In particular, deposition
rate is a key parameter determining the form of core-shell due to the necessity of a
balanced crystal growth rate and deposition rate for a conformal coating of 1-D
nanostructures. It is usually expected that in the case of a high crystal growth rate
compared to deposition rate by sputtering process there is a sufficient time required
for the effective pre-cursor nucleation on NRs surface, which then results in this
type formations [116].
In order to analyze the interface region formed between TiO2 and CdTe, a high
resolution TEM image was recorded. As can be seen from Fig. 15.26c, the
boundary between TiO2 and CdTe phases are distinguishable as well as crystal
planes associated with each phases.
To investigate the optical properties of the constructed device and TiO2 NRs the
ultraviolet-visible-near infrared (UV-VIS-NIR) absorption spectra were recorded in
the 350–1000 nm wavelength range, as shown in Fig. 15.27. It is apparent from the
absorption spectrum that there are two onsets, which are at 885 (1.4 eV) and
410 nm (3.02 eV) associated with band gaps of CdTe and TiO2, respectively. This
result can be taken as a strong evidence for the contribution of both materials to the
absorption process, resulting in the observed broadening of absorption spectrum for
a wide wavelength range including UV, VIS, and NIR. The high sensitivity in these
wavelength range under solar spectrum is a crucial necessity for the fabrication of
high efficient solar cells since the most intense solar spectrum reaches on the surface
of the earth is in the 1.4–1.6 eV photon energy range.

Fig. 15.26 TEM images recorded for TiO2 NRs decorated with sputtered CdTe thin film layer:
a low magnification image showing the formation of a core-shell type solar cell, b image revealing
the systematic variation in thickness of deposited CdTe layer on TiO2 NRs from bottom to top,
c high magnification image revealing the atomic planes of CdTe and TiO2 phases
15 One-Dimensional Nano-structured Solar Cells 379

Fig. 15.27 Absorption

spectra recorded for TiO2
NRs and TiO2/CdTe

Absorbance (in arb. u.)

core-shell structure
TiO2/CdTe

TiO2

400 500 600 700 800 900 1000

Wavelength (nm)

Fig. 15.28 a Responsivity (a)

spectrum obtained for TiO2/ 2.0
Photoresponsivity (A/W)

CdTe core-shell solar cell, 0.5 V

b current density (J)–voltage 1.5
(V) characteristics of the solar
cell measured under dark and
light illumination (AM 1.56) 1.0

0.5

0.0
300 450 600 750 900
Wavelength (nm)

(b) 4
J(mA/cm2)

9
2
6 0
J (mA/cm2)

-2
3 0.0 0.2 0.4 0.6
dark Voltage (V)
0

light
-3

-6
-1.0 -0.5 0.0 0.5
Voltage (V)

Figure 15.28a, b show the responsivity as a function of wavelength (300–

900 nm) at 0.5 V and J–V characteristic of n-TiO2/p-CdTe structured core-shell
solar cell, respectively. The variation in phoresponsivity observed in Fig. 15.20a is
consistent with the results extracted from absorption spectrum. In other words, there
are three humps that can be resolved from the photoresponsivity located at 600 nm
380 H. Karaağaç et al.

(2.06 eV), 775 nm (1.6 eV), and 854 nm (1.41 eV). Of these peaks, the one
observed at 854 nm (1.41 eV) is associated with the optic band gap of CdTe. As for
the ones observed at 600 and 775 nm, they are likely to be stemming from the
defect states formed either in TiO2 crystal or at interface formed between TiO2 and
CdTe in core-shell structure.
The J–V characteristic of n-TiO2/p-CdTe core-shell solar cell under standard test
conditions (A.M. 1.5G) is shown in Fig. 15.28b. Dot patterned Cu/Au and FTO
substrates were preferred for top and bottom electrodes of solar cells, respectively
during the measurement. It is apparent from the graph that there is a rectification of
current, verifying the formation of a p-n heterojunction between CdTe and TiO2. It
is also clear from the figure that device shows the photovoltaic behavior under
illumination. The solar cell parameters, Voc, Jsc, FF and power conversion efficiency
(η) were found to be as 0.34 V, 1.27 mA/cm2, 28 and 0.12 %, respectively.
As it is the first study reporting the solar cell parameters for a TiO2-CdTe
core-shell solar cell, it is not possible to compare with any study in literature.
Recently, however, a core-shell solar cell based on CdTe decorating TiO2
nanotube arrays was reported by Zhang et al. [117]. The solar parameters such as
the Voc and Jsc for this device structure were 0.23 V and 1.23 mA/cm2, respec-
tively and with a corresponding power conversion efficiency of 0.10 %, slightly
lower than that we obtained with TiO2-CdTe core-shell solar cell.
The low power conversion efficiency achieved with n-TiO2(NRs)/p-CdTe(thin
film) core-shell solar cell, 0.12 %, could be improved by applying several opti-
mization process such as improving the quality of CdTe thin film (e.g., grain size,
conductivity, carrier concentration, and mobility), adjusting the morphology (e.g.,
length, radius, density and orientation) of synthesized TiO2 NRs properly, electrical
contacts to CdTe coated TiO2 NRs. In addition to these, the post-production
optimization process such as CdCl2 treatment to CdTe thin film may have a pro-
nounceable effect on improving the power conversion efficiency of solar cell over
0.12 % due to its effect on increasing the grain sizes in polycrystalline structure of
CdTe.
In addition to our studies, the solar cell device performances of nanowires
derived from a wide variety of materials reported in literature are summarized in
Table 15.1.

15.4.4 Carbon

Carbon nanotubes (CNTs) present new opportunities to design third-generation

photovoltaic devices by enhancing the efficiency and further reducing the pro-
duction costs to make them competitive in the market. CNTs are investigated in
bulk heterojunction [133] as well as in inorganic thin film [134] solar cells. In this
section, we review the studies about the utilization of CNTs in dye sensitized solar
cells. We describe briefly the material properties of CNTs first. Strategies devoted
15 One-Dimensional Nano-structured Solar Cells 381

Table 15.1 Solar cell device performances of nanowires derived from a wide variety of materials
a b c d
Solar cell Material (Jsc) (Voc) (FF) (η) References
type (mA/cm2) (V) %
Core-shell Si NWs 4.28 0.29 0.33 0.46 [9]
CdS-Cu2S – 0.61 0.80 5.40 [120]
ZnO/CdS 7.23 −1.55 – 3.53 [121]
ZnO/TiO2/P3HT 0.77 0.51 0.54 0.23 [122]
p/i/n Si NWs 23.9 0.26 0.55 3.40 [123]
TCO/a-Si/Si NWs 27 0.476 0.56 7.29 [124]
amorph-Si NWs 13.9 0.830 0.524 6.00 [125]
CuO/ZnO 0.89 0.445 – 0.17 [126]
Si 4.28 0.29 0.33 0.46 [127]
GaAs 0.20 0.20 0.27 0.83 [128]
InP 13.72 0.43 0.57 3.37 [8]
ZnO/ZnS – 0.09 0.28 – [129]
Axial GaAs 21.08 0.565 0.6365 7.58 [23]
p+nn+-Si 13 0.450 0.30 4.40 [130]
Embedded PEDOT:PSS-Si 9.38 0.43 0.45 1.82 [131]
NWs
ZnO/AgNW/ZnO 31 0.458 0.45 6.37 [132]
ZnO/Cu2O2 1.43 0.15 0.25 0.05 [22]
a
Short current density
b
Open circuit voltage
c
Fill factor
d
Efficiency

to the development of more efficient photoanode materials and to replace the

counter electrode (Pt) by carbon nanotubes will be described.

15.4.4.1 Carbon Nanotubes

Carbon nanotubes (CNTs) are cylinder-shaped macromolecules with large aspect

ratios. The length axes can grow in a range of 100 nm up to 20 cm depending on the
synthesis conditions. In the most general case, a CNT is composed of a concentric
arrangement of many cylinders. Such multi-walled nanotubes (MWNTs) can reach
diameters of up to 100 nm. Single-walled nanotubes (SWNTs) possess the simplest
geometry and have been observed with diameters ranging from 0.4 to 3 nm. The
formation of a SWNT can be visualized through the rolling of a graphene sheet.
Based on the orientation of the tube axis with respect to the hexagonal lattice, the
structure of a nanotube can be completely specified through its chiral vector.
A SWNT behaves either as a metal or as a semiconductor, depending on its chiral
vector [135].
382 H. Karaağaç et al.

A critical factor for the application of CNTs in solar cells is their purity. To
remove the by-products and catalysts, a detailed purification strategy is a necessity
in most cases. Multi-step purification procedures using surfactants etc. are described
in detail in [136, 137].
SWNTs as well as MWNTs are investigated as electrode material. Both of them
can be an ideal channel for collecting and transporting charges across
light-harvesting assemblies depending on the priorities by device construction.
The work function of CNTs with intrinsic values of 4.5–5.1 eV are in the same
range to the work function of indium tin oxide (4.4–4.9 eV), the mostly used as a
transparent electrode in organic photovoltaics [138]. Recent reports have also
shown that CNT films can be produced with sheet resistance values as low as
160 V/m−2 along with a transmittance of 87 % at 550 nm [139]. These values are
still higher than those of ITO films on glass or plastic with sheet resistances as low
as 4 and 10 V/m−2, respectively, but still in the necessary conductivity range for
photovoltaic applications.
Additionally, the random network properties of CNT films facilitate maximum
mechanical flexibility when compared to ITO films deposited on polymer substrates
as demonstrated by Parekh et al. [140].
If the homogeneity and transparency of the carbon nanotube thin films is
important, SWNTs are better suited than MWNTs. On the other hand, MWNTs are
purely metallic, whereas SWNT synthesis always contain metallic and p-type
semiconductive SWNTs as product. To obtain much greater precision over their
electronic properties semiconducting and metallic SWNTs should be either chem-
ically separated by using selective surfactants [141] or chemically doped. Heavily
doping of SWNTs can be established using strong acids (HNO3 and/or SOCl2)
[142] or inorganic solvent mixtures (hydrofluoric acid, nitric acid and gold
(III) chloride hydrate) [143] by the metallization of p-type SWNTs through effec-
tive pinning of the Fermi level inside the valence band. It should be also noted that
the metallization of SWNTs is a quasi reversible process [144]. Extended testing
carried out in air and at room temperature using a protecting layer of poly
(3,4-ethylenedioxythiophene)–poly(styrenesulfonate) aqueous solution (PEDOT:
PSS, Baytron) on the top of the SWCNT layer show stable sheet resistances for a
aging period of 1400 h at 80 °C [145]. In 2012, sheet resistances change less than
10 % over 20 days in ambient and less than 2 % with overnight heating to 300 °C in
air has been achieved by doping SWNTs using MoOx [146]. Because of the well
defined metallic behaviour, MWNTs are mostly investigated in DSCs.
There are several wet-chemistry methods for fabricating isotropic nanotube
films, including drop casting from solvents, spin coating, air brushing, vacuum
filtering and quasi Langmuir-Blodgett deposition. Electrospinning and gas flow
growth of vertically aligned CNTs on suitable substrates are the mostly used
methods for oriented CNT films.
The problems to overcome for isotropic CNT films are the poor uniformity,
controllable film thickness and the fluctuations due to the van der Waals interactions
15 One-Dimensional Nano-structured Solar Cells 383

between the nanotubes. The formation of bundled nanotubes could adversely affect
the transport and recombination in solar cells. To overcome the carbon nanotube
clustering formation surfactants such as sodium dodecylbenzenesulfonate, sodium
dodecyl sulphate and sodium carboxymethyl cellulose are used [147, 148]. After
washing out the excess amount of surfactants, using sodium dodecyl sulphate for
example, Kim et al. obtained a sheet resistance of 57 V/sq at 65 % transmittance and
a conductivity in excess of 6700 s cm−1—values that are comparable with those of
surfactant-free CNT films [148], process of which is given in Fig. 15.29.
Additionally, the strong interaction between surfactant and CNTs allows dispersions
to be prepared without excessive sonication produces avoiding surface defects and
shortening of the tube length [149].
Although the wet-chemistry methods work well with randomly dispersed CNTs,
they are not suitable for vertically aligned CNTs. The aligned structure will be
damaged in the wet processing because the upper ends of the neighboring nan-
otubes have been seen to bundle together and cause some nanotubes to lay down.
Usually, the thin film fabrication process of the as-grown aligned CNTs requires
a mechanical densification after monolithically removement of CNTs from the
substrate. By controlling the extent of compression the CNT density can be tailored.
An UV curable epoxy-precursor contacted to the CNT forest diffuses via capillary
action and is then cured under well defined preparation parameter [150].

Fig. 15.29 General preparation scheme of CNT thin films as electrodes and its morphology [148]
384 H. Karaağaç et al.

CNTs as Photoanode Material in DSCs

In DSCs, a wide band gap n-type semiconductor material, anatase TiO2 nanopar-
ticles (generally around 20 nm in diameter) are deposited and interconnected on
fluorine-doped tin dioxide FTO (F:SnO2) coated conducting glass as photo anode,
which is attached by dye molecules (sensitizers). The design using mesoporous
TiO2 completely derives from the understanding suggesting that a large interfacial
surface area must be averted in solar cell devices in order to enhance the number of
dye molecules on the TiO2 electrode for light harvesting. The electronic transition
occurs upon light absorption in the sensitizers, which leads to the formation of an
excited state of the dye (exciton). Charge separation occurs at the interface by
excited dye injecting photo-induced electrons into the conduction band (CB) of
mesoporous TiO2 working electrode. The injected electrons are afterwards trans-
ported through the interconnected mesoporous TiO2 network towards the FTO
substrate of photo anode. After electron injection, the dye molecule in its oxidized
form is subsequently regenerated by electron donation from the liquid electrolyte
(−/I3− redox couple) or solid state hole conductor (such as 2,2′,7,7′-tetrakis(N,N-
dimethoxyphenylamine)-9,9′-spiro bifluorene, termed as spiro-OMeTAD [14],
P3HT [17] etc.). Meanwhile the holes are transferred via the redox couple towards
the counter-electrode (CE) to complete the circuit [151].
The output voltage of a DSC device is determined by the energy difference
between the redox potential of the electrolyte and the electron quasi-Fermi level
(EF) in the TiO2 under working conditions (Fig. 15.30). The photocurrent output
depends on the absorption spectrum of the dye and the absorbed photon to current
conversion efficiency, also known as the internal quantum efficiency (IQE).
The intrinsic properties of p-type semiconductor, such as low hole mobility, fast
interfacial recombination rate and insufficient infiltration into the mesoporous photo
anode are the main limiting factors of the solar cell efficiency.

Fig. 15.30 Schematic

operating mechanism of
liquid-electrolyte
dye-sensitize solar cell based
on mesoporous TiO2
photoanode and I−/I3− redox
couple [152]
15 One-Dimensional Nano-structured Solar Cells 385

Using nanoparous materials, a number of the charges are usually trapped by

isolated nanoparticles, surface states, or defect states. The so-called trap effects slow
down the charge transport through diffusion and limit the device efficiency, which
has been also proved by time-resolved photocurrent and photovoltage measure-
ments [153, 154] and modelling studies [155, 156].
An idea to improve the charge collection was realized by integration of metallic
carbon nanotubes into the described systems.
Two different strategies were developed:
(a) Incorporating CNTs into a titanium photoanode
(b) Core/shell structures incorporating CNT inside TiO2 nanofiber or decorating
CNT walls by TiO2 nanoparticles.
In Table 15.2, the solar cell parameter of the ‘best’ performing devices together
with the references for detailed information are summarized.
A general optimization step for the applications a–b is the necessity to optimize
the volume fraction of CNTs in photo anodes to find the optimal balance between
improving the electron transport properties by adding CNTs and the dye loading
capacity of the titania. It is well recognized that the optimal solar cell efficiencies
can be obtained when the dye is chemisorbed in a closely packed monolayer on
titania, whereas the chemisorption of multilayers decreases the overall performance.
Hereby, the interaction between the hydroxyl group of the dye and titania has a
significant importance [155]. Following this concept, MWNTs were oxidized in a
hydrogen peroxide (H2O2) solution under ultrasonication condition for 24 h at the
temperature 50 °C to produce finely dispersed MWNTs terminated with carboxylic
acid groups to enhance the adhesion of the titania nanoparticles on the MWNT
surface [157]. An overall efficiency of 4.97 % has been achieved by this approach.
The efficiency enhancement in comparison to MWNT free reference cell was ca.
49 % (Table 15.2, a1). In another study, using acid treated MWNTs (without OH
groups on the side walls) by simply adding into a titania paste and annealing at
150 °C, the efficiency enhancement was only 20 % relative to its MWNT free
reference (Table 15.2, a2).
Using the same concept, the efficiency of DSCs was as high as 8.1 % in 2013
(Table 15.2, a3) [159]. Optimizing the photoanode morphology by reducing the
grain boundaries using a post-treatment step at 500 °C, hence improving the charge
collection and increasing the electron diffusion coefficient was resulted to an
improvement of 25 % cell efficiency relative to the efficiency of MWNT free
photoanodes.
But the photocurrent or energy conversion efficiency was not monotonously
increased with the CNT concentration. An argument is related to the intrinsic feature
of the carbon nanostructures, which should be taken into account: As it is well
known that carbon species, including carbon black, activated carbon, graphene, as
well as CNTs, possesses catalytic activities for conversion of iodine species. For
photoelectrode, conversion of triiodide to iodide ions means loss (recombination) of
the photogenerated electrons, which is unfavorable to charge collection and should
be suppressed.
386 H. Karaağaç et al.

Table 15.2 The use of carbon nanotubes as photoanode

Photo anode Device I–V characteristics Efficiency References
0.1 wt% MWNT H2O2 TiO2/MWNT/N719/Pt 0.63 V/13.5 mA/cm2/ 4.97 [157]
treated (a1) (liquid electrolyte) FF: 59.00 %
– TiO2/N719/Pt (liquid 0.60 V/8.5 mA/cm2/ 3.32
electrolyte) FF: 65.00 %
0.1 wt% MWNT Low TiO2/MWNT/N3/Pt 0.78 V/9.08 mA/cm2/ 5.02 [158]
temperature (liquid electrolyte) FF: 70.80 %
processing (150 °C)
(a2)
– TiO2/N3/Pt (liquid 0.76 V/7.97 mA/cm2/ 4.15
electrolyte) FF: 68.50 %
0.01 wt% MWNT in TiO2/MWNT/N719/Pt 0.72 V/15.6 mA/cm2/ 8.10 [159]
TiO2 paste/500 °C (a3) (liquid electrolyte) FF: 71.00 %
(a3) including light TiO2/MWNT/N719/Pt 0.76 V/16.0 mA/cm2/ 9.00
scattering TiO2 (liquid electrolyte) FF: 74.00 %
nanoparticles
0.07 wt% MWNT in TiO2/MWNT/N719/Pt 0.70 V/11.6 mA/cm2/ 5.90
TiO2 paste/500 °C (liquid electrolyte) FF: 73.00 %
– TiO2/N719/Pt (liquid 0.75 V/12.7 mA/cm2/ 6.50
electrolyte) FF: 71.0 %
TiO2/0.2 MWNT TiO2/MWNT/N3/Pt 0.82 V/11.98 mA/cm2/ 6.12 [160]
electrospinning rice (liquid electrolyte) FF: 62.0 %
GRAIN form (b1)
– 0.81 V/9.40 mA/cm2/ 4.61
FF: 60.0 %
TiO2/0.1 wt% MWNT TiO2/MWNT/N7/Pt 0.75 V/18.53 mA/cm2/ 10.24 [161]
electrospinning (liquid electrolyte) FF: 73.6 %
elongated nanowires
15 μm layer thickness
(b2)
TiO2/0.1 wt% MWNT TiO2/MWNT/N7/Pt 0.72 V/11.10 mA/cm2/ 5.84
electrospinning 6 μm (liquid electrolyte) FF: 73.0 %
layer thickness
TiO2/0.15 wt% TiO2/MWNT/N7/Pt 0.78 V/12.11 mA/cm2/ 7.13
MWNT (liquid electrolyte) FF: 75.6 %
electrospinning 13 μm
layer thickness
– TiO2/N7/Pt (liquid 0.73 V/15.4 mA/cm2/ 5.63
electrolyte) FF: 50.0 %

Additionally, an important parameter was the loss of transparency of the pho-

toanode. For higher CNT concentrations (above 0.01 wt%), the light absorption of
the CNTs play a role, resulting in the loss of solar radiation available for pho-
tocurrent generation. By using additional titania to enhance the light scattering in
the photo active area, the efficiency was as high as 9.0 % (Table 15.2, a3).
Given the MWCNT is incorporated fully inside the titania according to the
concept (b), this can rapidly capture and transport the photogenerated electrons and
concurrently reduce the undesirable recombination and back-reaction.
15 One-Dimensional Nano-structured Solar Cells 387

Rice-grain shaped TiO2/MWNT nanocomposite photoanodes were prepared by

electrospun with efficiencies of 6.1 % (Table 15.2, b1).
In a similar approach, TiO2/PVP/MWCNT composite nanofibers were first
electrospun at 70 kV on fluorine-doped tin oxide (FTO) glass from a precursor
solution that contained titanium isopropoxide, polyvinylpyrrolidone, acetic acid and
ethanol. A layer of TiO2/MWCNT nanorod were peeled off from the original FTO
glass after calcination and then transferred to another FTO glass with an ultra-thin
adhesive TiO2 paste. The TiO2/MWCNT nanorod photoanode was obtained after
calcinating at 450 °C for 2 h [161].
Upon optimization of the parameters for the photoanode, an efficiency as high as
10.24 % has been attained for a photoanode with CNT nanorods that incorporates
0.1 % MWCNTs in the precursor solution. It is worthy to note that the improved
efficiency for the photoanode incorporating MWCNTs is largely due to the
improved FF, an indication that shows the improvement of charge transport
properties.
But also using this concept, the porosity of the titania layer should be high enough
enabling sufficient light harvesting by dye molecules. The best solar cell parameters
were reported for a mesoporous photo anode with a thickness of ca. 15 µm.
Decreasing the layer thickness to 6 µm under the same preparation conditions, the
solar cell efficiency was only 5.8 % by constant fill factor. However, upon further
increase the amount of MWNT in the precursor solution to a value of 0.15 %, Jsc
decreased in both sets of photoanode devices, demonstrating the fact that loading of
the dye molecules was not sufficient enough in these devices. Despite this reverse
trend, the FF kept increasing with increasing the amount of MWNT, an indication
suggests that the recombination is also suppressed well in the later case.

CNTs as Counter Electrode Material in DSCs

In DSCs, the redox reaction occur at the interface of electrolyte/counter electrode

and electrolyte/dye molecule. The reaction on the counter electrode is the reduction
of tri-iodide to iodine. Because of the catalytic activity of CNTs for the reduction of
DSC electrolyte tri-iodide [162], there is great effort to replace the costy counter
electrode Au/Pt by solution processable cheaper carbon based alternatives [163].
The carbon nanotubes that show results that are closest to Pt are usually
single-walled carbon nanotubes, with the MWNTs showing lower overall power
conversion efficiency when compared to SWNTs. While there is a clear advantage
to using carbon SWNTs in the counter electrodes of DSSCs, there is greater
development in the method of mass production of carbon MWNTs, which makes it
valuable to investigate methods to improve the reductive properties of MWNTs in
the tri-iodide/iodide redox couple in the DSSCs.
The effect of different CNT blends as cathode on the DSC performance is listed
in Table 15.3.
388 H. Karaağaç et al.

Table 15.3 The use of carbon nanotubes as counter electrode materials in dye-sensitized solar
cells
Cathode I–V characteristics Efficiency References
(%)
SWNTs (c1) 0.68 V/9.51 mA/cm2/FF: 62.0 % 4.03 [164]
MWNTs (c2) 0.69 V/10.55 mA/cm2/FF: 60.0 % 4.36
– 0.63 V/8.11 mA/cm2/FF: 63.0 % 3.22
MWNT-blended 0.80 V/11.99 mA/cm2/FF: 74.0 % 7.13 [165]
incarboxymethylcellulose (c3)
– 0.80 V/11.99 mA/cm2/FF: 68.0 % 6.52
MWNT (c4) 0.74 V/15.54 mA/cm2/FF: 57.6 % 6.62 [166]
MWNT/Graphene (c4) 0.74 V/16.04 mA/cm2/FF: 62.7 % 7.55
Graphene (c4) 0.75 V/13.09 mA/cm2/FF: 47.3 % 4.64
Pt as reference 0.79 V/16.27 mA/cm2/FF: 68.8 % 8.80
MWNT (c5) 0.70 V/9.55 mA/cm2/FF: 52.0 % 3.50 [166]
MWNT-TiN 0.75 V/12.74 mA/cm2/FF: 57.0 % 5.40
TiN nanoparticle 0.66 V/9.28 mA/cm2/FF: 35.0 % 2.10
Pt as reference 0.73 V/12.83 mA/cm2/FF: 60 % 5.68
Vertically aligned CNTs (c6) 0.66 V/13.9 mA/cm2/FF: 58.0 % 5.20 [167]
Isotropic CNTs 0.61 V/11.8 mA/cm2/FF: 600 % 3.70
Pt as reference 0.61 V/17.2 mA/cm2/FF: 64.0 % 6.60

In 2010, SWNTs as well as MWNTs were investigated as counter electrode

materials with slightly better performance (Table 15.3-c1) for MWNTs in com-
parison to PtCl4 electrode [164].
Highly disordered multiwalled CNTs prepared from solution using doctor blade
technique were successfully used as counter electrode for DSCs through a simple
low-temperature fabrication process. The CNT layer acts hereby as a better catalyst,
enhancing the fill factor of the devices (Table 15.3, c3). Variations in electrolyte
solutions (an electrolyte containing 1-butyl-3-methylimidazolium instead of Li ion
in acetonitrile) demonstrated that electrolytes containing larger molecules perform
better using CNT than Pt electrode [165].
The highly catalytic graphene is functionalized onto the carbon MWNTs to
support the reduction of the tri-iodide in the redox couple that mediates the charge
between the two electrodes of the DSSC [166]. The study of these cells supports the
idea that for efficient performance of a carbon-based counter electrode, there needs
to be many sharp atomic edges exposed to the ions in the electrolyte for increased
catalytic effect. Graphene devices only shows a very poor performance because of
forming smooth layers on top of one another, limiting the total surface area for the
electrolyte interaction. MWNTs alone, are better suited than graphene, but the best
morphology is given by blending the materials together leading to a nanostructured
surface area with lowest sheet resistance of the cathode (Table 15.3, c4).
15 One-Dimensional Nano-structured Solar Cells 389

Similarly, titanium nitrides (TiN) with intrinsic electro catalytic activity for the
triiodide reduction is blended with MWNTs to form DSC cathodes [168]. A DSC
composed of the highly ordered TiN nanotube arrays shows comparable perfor-
mance with typical Pt counter electrode [169]. However, TiN nanoparticle film
electrode alone has lower fill factor (FF) owing to the poor electron transport effi-
ciency across nanoparticles. Furthermore, the introduction of conducting paths of
CNTs into TiN can improve electrical conductivity and capacitance of the cathode.
The surface of CNTs was functionalized by TiN—by thermal hydrolysis of TiOSO4
on CNTs. As shown in Table 15.2, c4, the synergic effect of CNTs and TiN is
verified. The performance of the TiN-CNT counter electrode is comparable with that
of the typical Pt counter electrode. In the TiN anchored CNTs, the superior elec-
trocatalytic activity of TiN nanoparticles is undoubtedly the driving force for the
high photovoltaic performance, demonstrated by cyclic voltammetric measurements
in the reference paper. The introduction of TiN nanoparticles on CNTs not only
decreases the charge-transfer resistance in the electrolyte/electrode interface, but also
reduces the diffusion impedance of triiodide ions, leading to the high Jsc and FF for
DSSCs. Meanwhile, the electron-transport network formed by CNTs also plays an
indispensable role in the high photovoltaic performance (Table 15.3, c5).
CNTs have been also blended in conducting polymers such as poly
3,4-ethylenedioxythiophene:polystyrenesulfonate, but the effect of MWNTs was
rather insignificant in comparison to PEDOT:PSS only devices [170].
Also vertically aligned carbon nanotubes are investigated as counter electrodes
[167]. Ultrathin-alumina coated stainless steel substrates decorated with Fe
nanoparticles were used as substrates for the CNT growth. DSCs using vertically
aligned CNTs had a better performance than DSCs using randomly dispersed CNTs
(without any surface modification). But the DSCs using vertically aligned CNT
electrodes do exhibit an increased series resistance compared to traditional Pt cells
(C6). The source of this increased resistance has been demonstrated to be the charge
transfer resistance at the CNT/electrolyte interface.
Functionalization of the CNT surfaces as well as post-growth surface treatments
using inorganic nanoparticles as given in examples above might reduce the acti-
vation barrier for the triiodide reduction and thereby reduce the charge transfer
resistance.

15.5 Summary and Future Outlook

The huge demand for clean energy can be supplied from the sun due to its abun-
dance and free from contaminations and carbon emissions. However, owing to the
high cost, imposed by quantity and quality of potential materials, and inadequate
power conversion efficiencies of existing conventional solar cells it is not possible
to compete with the energy supplied from fossil fuel sources. To meet this energy
requirement, there has been an increasing interest in recent years in both reducing
the price and increasing the efficiency of solar cells by use of one-dimensional (1-D)
390 H. Karaağaç et al.

nanostructures, including nanowires (NWs), quantum wells (QWs), and quantum

dots (QDs), in solar cells. 1-D nanostructured solar cells, particularly nanowires and
nanopillars based ones, address the main requirement of high-efficient solar cells at
lower cost in several ways including high carrier collection efficiency by decou-
pling the light absorption and carrier collection directions, the ability to grow
single-crystalline nanowires on inexpensive substrates without resorting to complex
epitaxial routes, and minimization of optical losses (e.g., reducing reflection and
enhancing absorption) by light trapping. In the past years, several solar cell device
architectures based on nanowires/nanorods/nanopillars have been developed to
achieve this goal. In particular, device architectures of nanowire/nanorods array
with radial, axial, and nanowire/nanorods array embedded in thin film geometries
are the most widely studied ones, all of which offer significant advantages over the
traditional planer thin-film or wafer-based solar cells in terms of cost, effective
charge separation, strain relaxation and some electrical and optical properties.
Over the last two decades, a considerable amount of literature has been published
on one-dimensional nanostructure based solar cells by the use of a wide variety of
materials (e.g., poly-crystalline Si, amorphous Si, polymers, II–VI and I–III–VI2
chalcopyrite (I–III–VI2) compounds. In this chapter, we mainly focused on several
commonly used materials such as silicon, zinc-oxide and titanium-dioxide that we
have employed in our recent studies for the fabrication of one-dimensional nanos-
tructures based solar cells as well as the studies reported in literature on the uti-
lization of carbon nanotubes (CNTs) in dye sensitized solar cells. In one of these
studies, for instance, we demonstrated a solar cell based on ZnO and Si Nanowires
and their integration into a chalcopyrite thin film (AGIS). Photovoltaic properties
were determined under AM 1.5G illumination. For the ZnO NWs based solar cell,
0.37 % efficiency was measured, while it was 5.5 % for the Si NWs based solar cell.
While at present, the energy conversion efficiencies of n-ZnO-NWs/p-AGIS and
n-Si NWs/p-AGIS solar cells are lower than the conventional planar silicon and
some Ag-chalcopyrite solar cells, to further improve the efficiency of the device
many optimization processes are required such as nanowire growth, thin film layer
deposition and contact formation, and surface passivation processes to inhibit sur-
face recombination. It is reasonable to expect that surface passivation alone would
increase the energy conversion efficiency of devices to over 5.50 % that we have
obtained for n-Si NWs/p-AGIS solar cell. In addition, although the AGIS has been
used as the thin film in which Si and ZnO NWs were embedded, the same device
architecture can be employed for other potential PV materials (e.g., CdTe, CdS, and
CuInSe2) to enhance the energy conversion efficiency. Moreover, Si NWs transfer
printed from the single crystal wafer to any arbitrary substrate through several
bottom-up processes may drastically lower the cost of such solar cells.
Carbon nanotube/titania blends function well as transparent, working electrodes
and as such could offer improved performance of DSSCs through the reduction of
recombination processes by more effective transport of electrons. The most critical
factor hereby is to find the balance between the concentration of light harvesting
dyes attached onto the titania and the number of carbon nanotube charge transport
channels on porous titania electrode. Performance gains up to 9–10 % for DSSCs
15 One-Dimensional Nano-structured Solar Cells 391

employing liquid electrode has been reported using CNT blends in titania. Also as
counter electrode, the replacement of Pt by ‘potentially’ low-cost carbon nanotubes
as alternative materials, is progressing well. However, the mass-production tech-
nologies have to be further improved to obtain low-cost carbon nanotube materials
for large area solar cell applications. The polychirality of nanotubes together with
purity requirements reported so far certainly confounds this issue and thus new
ideas for the development of highly uniform carbon nanotube mass production has
to be realized.
In summary, the use of one-dimensional nanostructures in several type of solar
cell device architecture is a potential approach for the realization of cost effective
high efficiency solar cells. However, to fully exploit the unique properties of these
structures much work is needed. In other words, several fabrication and opti-
mization stages, such as surface passivation to eliminate surface/interface recom-
bination centers arising from large surface/volume ratio, the uniform doping
concentration for homo-junction solar cells, appropriate material systems to meet
lattice-matching for hetero-junction solar cells, chemical and mechanical stability,
and some practical issues (e.g., rapid scaling, device packaging, and integration into
device modules) are required to realize one-dimensional nanostructures based solar
cells with efficiencies comparable to those achieved with their planer counterparts at
much lower production cost.

References

1. Energy BSRoW, Technical Report, BP (2010)

2. P. Würfel, Frontmatter. Physics of solar cells (Wiley-VCH Verlag GmbH, 2007), pp. I–XII.
doi:10.1002/9783527618545.fmatter
3. J.S. Li, H.Y. Yu, Y.L. Li, Aligned Si nanowire-based solar cells. Nanoscale 3(12),
4888–4900 (2011)
4. Y.N. Xia, P.D. Yang, Y.G. Sun, Y.Y. Wu, B. Mayers, B. Gates, Y.D. Yin, F. Kim, Y.Q.
Yan, One-dimensional nanostructures: synthesis, characterization, and applications. Adv.
Mater. 15(5), 353–389 (2003)
5. Z.Y. Fan, D.J. Ruebusch, A.A. Rathore, R. Kapadia, O. Ergen, P.W. Leu, A. Javey,
Challenges and prospects of nanopillar-based solar cells. Nano Res 2(11), 829–843 (2009)
6. P.V. Kamat, Meeting the clean energy demand: nanostructure architectures for solar energy
conversion. J. Phys. Chem. C 111(7), 2834–2860 (2007)
7. M. Law, L.E. Greene, J.C. Johnson, R. Saykally, P.D. Yang, Nanowire dye-sensitized solar
cells. Nat. Mater. 4(6), 455–459 (2005)
8. J.A. Czaban, D.A. Thompson, R.R. LaPierre, GaAs core-shell nanowires for photovoltaic
applications. Nano Lett. 9(1), 148–154 (2009)
9. E.C. Garnett, P.D. Yang, Silicon nanowire radial p-n junction solar cells. J. Am. Chem. Soc.
130(29), 9224–9225 (2008)
10. E.C. Garnett, M.L. Brongersma, Y. Cui, M.D. McGehee, Nanowire solar cells. Annu. Rev.
Mater. Res. 41, 269–295 (2011)
11. R. Kapadia, Z.Y. Fan, K. Takei, A. Javey, Nanopillar photovoltaics: materials, processes, and
devices. Nano Energy 1(1), 132–144 (2012)
12. Three-Dimensional Nano Architectures: Designing Next Generation Devices (Springer, New
York, London, 2011). doi:10.1007/978-1-4419-9822-4
392 H. Karaağaç et al.

13. B.Z. Tian, X.L. Zheng, T.J. Kempa, Y. Fang, N.F. Yu, G.H. Yu, J.L. Huang, C.M. Lieber,
Coaxial silicon nanowires as solar cells and nanoelectronic power sources. Nature 449(7164),
885–890 (2007)
14. Y. Zhang, L.W. Wang, A. Mascarenhas, Quantum coaxial cables for solar energy harvesting.
Nano Lett. 7(5), 1264–1269 (2007)
15. B.M. Kayes, H.A. Atwater, N.S. Lewis, Comparison of the device physics principles of
planar and radial p-n junction nanorod solar cells. J. Appl. Phys. 97(11), 114302–114313
(2005)
16. L. Tsakalakos, J. Balch, J. Fronheiser, B.A. Korevaar, O. Sulima, J. Rand, Silicon nanowire
solar cells. Appl Phys Lett 91(23) (2007)
17. B.D. Yuhas, P.D. Yang, Nanowire-based all-oxide solar cells. J. Am. Chem. Soc. 131(10),
3756–3761 (2009)
18. Z.Y. Fan, H. Razavi, J.W. Do, A. Moriwaki, O. Ergen, Y.L. Chueh, P.W. Leu, J.C. Ho, T.
Takahashi, L.A. Reichertz, S. Neale, K. Yu, M. Wu, J.W. Ager, A. Javey, Three-dimensional
nanopillar-array photovoltaics on low-cost and flexible substrates. Nat. Mater. 8(8), 648–653
(2009)
19. R.R. LaPierre, Numerical model of current-voltage characteristics and efficiency of GaAs
nanowire solar cells. J. Appl. Phys. 109(3), 034311–034316 (2011)
20. C. Colombo, M. Heiss, M. Gratzel, A.F.I. Morral, Gallium arsenide p-i-n radial structures for
photovoltaic applications. Appl. Phys. Lett. 94(17), 173108 (2009)
21. Y.B. Tang, Z.H. Chen, H.S. Song, C.S. Lee, H.T. Cong, H.M. Cheng, W.J. Zhang, I. Bello,
S.T. Lee, Vertically aligned p-type single-crystalline gan nanorod arrays on n-type si for
heterojunction photovoltaic cells. Nano Lett. 8(12), 4191–4195 (2008)
22. V. Sivakov, G. Andra, A. Gawlik, A. Berger, J. Plentz, F. Falk, S.H. Christiansen, Silicon
nanowire-based solar cells on glass: synthesis, optical properties, and cell parameters. Nano
Lett. 9(4), 1549–1554 (2009)
23. M.Q. Yao, N.F. Huang, S. Cong, C.Y. Chi, M.A. Seyedi, Y.T. Lin, Y. Cao, M.L. Povinelli,
P.D. Dapkus, C.W. Zhou, GaAs nanowire array solar cells with axial p-i-n junctions. Nano
Lett. 14(6), 3293–3303 (2014)
24. A. Mews, Nanomaterials handbook. edited by Yury Gogotsi. Angew. Chem. Int. Ed. 46(13),
2143 (2007). doi:10.1002/anie.200685445
25. L.N. Dem’yanets, L.E. Li, T.G. Uvarova, Zinc oxide: hydrothermal growth of nano- and bulk
crystals and their luminescent properties. J. Mater. Sci. 41(5), 1439–1444 (2006). doi:10.
1007/s10853-006-7457-z
26. H.-E. Wang, Z. Chen, Y.H. Leung, C. Luan, C. Liu, Y. Tang, C. Yan, W. Zhang, J.A.
Zapien, I. Bello, S.-T. Lee, Hydrothermal synthesis of ordered single-crystalline rutile TiO2
nanorod arrays on different substrates. Appl. Phys. Lett. 96(26), 263104 (2010). doi:10.1063/
1.3442913
27. X. Wang, Y. Li, Selected-control hydrothermal synthesis of α- and β-MnO2 single crystal
nanowires. J. Am. Chem. Soc. 124(12), 2880–2881 (2002). doi:10.1021/ja0177105
28. H. Zhitao, L. Sisi, C. Jinkui, C. Yong, Controlled growth of well-aligned ZnO nanowire
arrays using the improved hydrothermal method. J. Semiconduct. 34(6), 063002 (2013)
29. M. Ahmad, M.A. Iqbal, J. Kiely, R. Luxton, M. Jabeen, Low temperature hydrothermal
synthesis of ZnO nanowires for nanogenerator: effect of gold electrode on the output voltage
of nanogenerator. Indian J. Eng. Mater. S 21(6), 672–676 (2014)
30. S.N. Bai, S.C. Wu, Synthesis of ZnO nanowires by the hydrothermal method, using sol-gel
prepared ZnO seed films. J. Mater. Sci-Mater El 22(4), 339–344 (2011)
31. R. Hao, X. Deng, Y.B. Yang, D.Y. Chen, Research progress in preparation and applications
of ZnO nanowire/rod arrays by hydrothermal method. Acta Chim. Sinica 72(12), 1199–1208
(2014)
32. H.S. Jang, B. Son, H. Song, G.Y. Jung, H.C. Ko, Controlled hydrothermal growth of
multi-length-scale ZnO nanowires using liquid masking layers. J. Mater. Sci. 49(23), 8000–
8009 (2014)
15 One-Dimensional Nano-structured Solar Cells 393

33. H. Karaagac, M. Parlak, E. Yengel, M.S. Islam, Heterojunction solar cells with integrated Si
and ZnO nanowires and a chalcopyrite thin film. Mater. Chem. Phys. 140(1), 382–390 (2013)
34. D.P. Neveling, T.S. van den Heever, R. Bucher, W.J. Perold, L.M.T. Dicks, Effect of seed
layer deposition, au film layer thickness and crystal orientation on the synthesis of
hydrothermally grown ZnO nanowires. Curr. Nanosci. 10(6), 827–836 (2014)
35. I.J. No, S. Lee, S.H. Kim, J.W. Cho, P.K. Shin, Morphology control of ZnO nanowires
grown by hydrothermal methods using Au nanodots on Al doped ZnO seed layer. Jpn.
J. Appl. Phys. 52(2), 025003 (2013)
36. Y.K. Tseng, M.C. Hung, S.L. Su, S.K. Li, Using the hydrothermal method to grow p-type
ZnO nanowires on Al-doped ZnO thin film to fabricate a homojunction diode. J. Nanosci.
Nanotechnol. 14(10), 7907–7910 (2014)
37. H. Karaagac, V.J. Logeeswaran, M.S. Islam, Fabrication of 3D-silicon micropillars/walls
decorated with aluminum-ZnO/ZnO nanowires for optoelectric devices. Phys. Status Solidi A
210(7), 1377–1380 (2013)
38. K. Peng, Y. Xu, Y. Wu, Y. Yan, S.-T. Lee, J. Zhu, Aligned single-crystalline Si nanowire
arrays for photovoltaic applications. Small 1(11), 1062–1067 (2005). doi:10.1002/smll.
200500137
39. K.J. Morton, G. Nieberg, S.F. Bai, S.Y. Chou, Wafer-scale patterning of sub-40 nm diameter
and high aspect ratio (>50:1) silicon pillar arrays by nanoimprint and etching.
Nanotechnology 19(34), 345301 (2008)
40. Z. Huang, N. Geyer, P. Werner, J. de Boor, U. Gösele, Metal-assisted chemical etching of
silicon: a review. Adv. Mater. 23(2), 285–308 (2011). doi:10.1002/adma.201001784
41. H. Karaagac, M.S. Islam, Enhanced field ionization enabled by metal induced surface states
on semiconductor nanotips. Adv. Funct. Mater. 24(15), 2224–2232 (2014)
42. M. Meyyappan MS, Inorganic Nanowires: Applications, Properties and Characterization.
(CRC Press, 2010)
43. B. Gates, B. Mayers, A. Grossman, Y. Xia, A sonochemical approach to the synthesis of
crystalline selenium nanowires in solutions and on solid supports. Adv. Mater. 14(23), 1749–
1752 (2002)
44. R.V. Kumar, Y. Koltypin, X.N. Xu, Y. Yeshurun, A. Gedanken, I. Felner, Fabrication of
magnetite nanorods by ultrasound irradiation. J. Appl. Phys. 89(11), 6324–6328 (2001)
45. A.P. Nayak, A.M. Katzenmeyer, J.-Y. Kim, M.K. Kwon, Y. Gosho, M. Saif Islam, Purely
sonochemical route for oriented zinc oxide nanowire growth on arbitrary substrate. Proc.
SPIE 7683, 738312. doi:10.1117/12.851755
46. Y.Y. Wu, P.D. Yang, Direct observation of vapor-liquid-solid nanowire growth. J. Am.
Chem. Soc. 123(13), 3165–3166 (2001)
47. M.H. Huang, S. Mao, H. Feick, H.Q. Yan, Y.Y. Wu, H. Kind, E. Weber, R. Russo, P.D.
Yang, Room-temperature ultraviolet nanowire nanolasers. Science 292(5523), 1897–1899
(2001)
48. W. Lu, C.M. Lieber, Semiconductor nanowires. J. Phys. D Appl. Phys. 39(21), R387–R406
(2006)
49. S. Han, R.S. Wagner, Grain boundary effects on carrier transport in undoped polycrystalline
chemical-vapor-deposited diamond. Appl. Phys. Lett. 68(21), 3016–3018 (1996)
50. C.C. Chen, C.C. Yeh, Large-scale catalytic synthesis of crystalline gallium nitride nanowires.
Adv. Mater. 12(10), 738 (2000)
51. Y.J. Chen, J.B. Li, Y.S. Han, X.Z. Yang, J.H. Dai, The effect of Mg vapor source on the
formation of MgO whiskers and sheets. J. Cryst. Growth 245(1–2), 163–170 (2002)
52. X.F. Duan, C.M. Lieber, General synthesis of compound semiconductor nanowires. Adv.
Mater. 12(4), 298–302 (2000)
53. Y.W. Wang, L.D. Zhang, C.H. Liang, G.Z. Wang, X.S. Peng, Catalytic growth and
photoluminescence properties of semiconductor single-crystal ZnS nanowires. Chem. Phys.
Lett. 357(3–4), 314–318 (2002)
394 H. Karaağaç et al.

54. J. Zhang, X.S. Peng, X.F. Wang, Y.W. Wang, L.D. Zhang, Micro-Raman investigation
of GaN nanowires prepared by direct reaction Ga with NH3. Chem. Phys. Lett. 345(5–6),
372–376 (2001). doi:10.1016/S0009-2614(01)00905-8
55. M. Triplett, H. Nishimura, M. Ombaba, V.J. Logeeswarren, M. Yee, K.G. Polat, J.Y. Oh, T.
Fuyuki, F. Leonard, M.S. Islam, High-precision transfer-printing and integration of vertically
oriented semiconductor arrays for flexible device fabrication. Nano Res. 7(7), 998–1006
(2014)
56. C.N.R. Rao, F.L. Deepak, G. Gundiah, A. Govindaraj, Inorganic nanowires. Prog. Solid
State Ch 31(1–2), 5–147 (2003)
57. P.L. Dong, X.D. Wang, M. Zhang, M. Guo, S. Seetharaman, The preparation and
characterization of beta-SiAlON nanostructure whiskers. J. Nanomater. 2008, 282187–
282192 (2008)
58. Y.J. Hsu, S.Y. Lu, Vapor-solid growth of Sn nanowires: growth mechanism and
superconductivity. J. Phys. Chem. B 109(10), 4398–4403 (2005)
59. Z.W. Pan, Z.R. Dai, Z.L. Wang, Nanobelts of semiconducting oxides. Science 291(5510),
1947–1949 (2001)
60. Y.J. Zhang, N.L. Wang, S.P. Gao, R.R. He, S. Miao, J. Liu, J. Zhu, X. Zhang, A simple
method to synthesize nanowires. Chem. Mater. 14(8), 3564–3568 (2002)
61. S.S. Fan, M.G. Chapline, N.R. Franklin, T.W. Tombler, A.M. Cassell, H.J. Dai, Self-oriented
regular arrays of carbon nanotubes and their field emission properties. Science 283(5401),
512–514 (1999)
62. S. Sakurai, M. Inaguma, D.N. Futaba, M. Yumura, K. Hata, A fundamental limitation of
small diameter single-walled carbon nanotube synthesis-a scaling rule of the carbon nanotube
yield with catalyst volume. Materials 6(7), 2633–2641 (2013)
63. M. Xu, D.N. Futaba, M. Yumura, K. Hata, Alignment control of carbon nanotube forest from
random to nearly perfectly aligned by utilizing the crowding effect. ACS Nano 6(7),
5837–5844 (2012)
64. H. Karaagac, M. Kaleli, M. Parlak, Characterization of AgGa0.5In0.5Se2 thin films
deposited by electron-beam technique. J. Phys. D-Appl. Phys. 42(16) (2009). doi:Artn
165413; doi:10.1088/0022-3727/42/16/165413
65. I. Repins, M.A. Contreras, B. Egaas, C. DeHart, J. Scharf, C.L. Perkins, B. To, R. Noufi,
19.9 %-efficient ZnO/CdS/CuInGaSe2 solar cell with 81.2 % fill factor. Prog. Photovolt. 16
(3), 235–239 (2008). doi:10.1002/Pip.822
66. K. Yamada, N. Hoshino, T. Nakada, Crystallographic and electrical properties of wide gap
Ag(In1-x, Ga-x)Se-2 thin films and solar cells. Sci. Technol. Adv. Mat. 7(1), 42–45 (2006).
doi:10.1016/j.stam.2005.11.016
67. P.P. Ramesh, O.M. Hussain, S. Uthanna, B.S. Naidu, P.J. Reddy, Photovoltaic performance
of p-AgInSe2/n-CdS thin film heterojunctions. Mater. Lett. 34(3–6), 217–221 (1998)
68. Y.S. Murthy, O.M. Hussain, B.S. Naidu, P.J. Reddy, Characterization of P-Aggase2/N-Cds
thin-film heterojunction. Mater. Lett. 10(11–12), 504–508 (1991)
69. G.H. Chandra, O.M. Hussain, S. Uthanna, B.S. Naidu, Characterization of
p-AgGa0.25In0.75Se2/n-Zn0.35Cd0.65S polycrystalline thin film heterojunctions. Mat.
Sci. Eng. B-Solid 86(1), 60–63 (2001)
70. B. Ozdemir, M. Kulakci, R. Turan, H.E. Unalan, Effect of electroless etching parameters on
the growth and reflection properties of silicon nanowires. Nanotechnology 22(15), 155606
(2011). doi:Artn 155606; doi:10.1088/0957-4484/22/15/155606
71. K.Q. Peng, J.J. Hu, Y.J. Yan, Y. Wu, H. Fang, Y. Xu, S.T. Lee, J. Zhu, Fabrication of
single-crystalline silicon nanowires by scratching a silicon surface with catalytic metal
particles. Adv. Funct. Mater. 16(3), 387–394 (2006). doi:10.1002/adfm.200500392
72. O. Gunawan, S. Guha, Characteristics of vapor-liquid-solid grown silicon nanowire solar
cells. Sol. Energy Mat. Sol. C 93(8), 1388–1393 (2009). doi:10.1016/j.solmat.2009.02.024
73. D.R. Kim, C.H. Lee, P.M. Rao, I.S. Cho, X.L. Zheng, Hybrid Si microwire and planar solar
cells: passivation and characterization. Nano Lett. 11(7), 2704–2708 (2011). doi:10.1021/
Nl2009636
15 One-Dimensional Nano-structured Solar Cells 395

74. V.J. Logeeswaran, A.M. Katzenmeyer, M.S. Islam, Harvesting and transferring vertical pillar
arrays of single-crystal semiconductor devices to arbitrary substrates. IEEE T Electron. Dev.
57(8), 1856–1864 (2010). doi:10.1109/Ted.2010.2051195
75. M.M. Ombaba, L.V. Jayaraman, M.S. Islam, Precision stress localization during mechanical
harvesting of vertically oriented semiconductor micro- and nanostructure arrays. Appl. Phys.
Lett. 104(24), 243109 (2014)
76. H. Bi, R.R. LaPierre, A GaAs nanowire/P3HT hybrid photovoltaic device. Nanotechnology
20(46), 465205 (2009). doi:Artn 465205; doi:10.1088/0957-4484/20/46/465205
77. J. Davenas, S.B. Dkhil, D. Cornu, A. Rybak, Silicon nanowire/P3HT hybrid solar cells: effect
of the electron localization at wire nanodiameters. Energy Proc. 31, 136–143 (2012). doi:10.
1016/j.egypro.2012.11.175
78. B. Eisenhawer, S. Sensfuss, V. Sivakov, M. Pietsch, G. Andra, F. Falk, Increasing the
efficiency of polymer solar cells by silicon nanowires. Nanotechnology 22(31), 315401
(2011). doi:Artn 315401; doi:10.1088/0957-4484/22/31/315401
79. L.N. He, C.Y. Jiang, H. Wang, D. Lai, Rusli, Si nanowires organic semiconductor hybrid
heterojunction solar cells toward 10 % efficiency. ACS Appl. Mater. Inter. 4(3), 1704–1708
(2012). doi:10.1021/Am201838y
80. Y. Kang, D. Kim, Well-aligned CdS nanorod conjugated polymer solar cells. Sol. Energy
Mat. Sol. C 90(2), 166–174 (2006). doi:10.1016/j.solmat.2005.03.001
81. J.S. Huang, C.Y. Hsiao, S.J. Syu, J.J. Chao, C.F. Lin, Well-aligned single-crystalline silicon
nanowire hybrid solar cells on glass. Sol. Energy Mat. Sol. C 93(5), 621–624 (2009). doi:10.
1016/j.solmat.2008.12.016
82. H. Karaagac, A hybrid solar cell based on silicon nanowire and organic thin film. Physica
Status solidi (a) 211(11), 2503–2508 (2014). doi:10.1002/pssa.201431320
83. G.J. Matt, T. Fromherz, M. Bednorz, S. Zamiri, G. Goncalves, C. Lungenschmied, D.
Meissner, H. Sitter, N.S. Sariciftci, C.J. Brabec, G. Bauer, Fullerene sensitized silicon for
near-to mid-infrared light detection. Adv. Mater. 22(5), 647 (2010). doi:10.1002/adma.
200901383
84. J. Bae et al., Si nanowire metal–insulator–semiconductor photodetectors as efficient light
harvesters. Nanotechnology 21(9), 095502 (2010)
85. Y. Cui, Q. Wei, H. Park, C.M. Lieber, Nanowire nanosensors for highly sensitive and
selective detection of biological and chemical species. Science 293(5533), 1289–1292
(2001). doi:10.1126/science.1062711
86. T.C.H. Yamada, S. Ishidac, Y. Arakawac, Si-nanowire optical waveguide devices for optical
communications. Proc. SPIE 6019, 60192X (2005)
87. L. Hu, G. Chen, Analysis of optical absorption in silicon nanowire arrays for photovoltaic
applications. Nano Lett. 7(11), 3249–3252 (2007). doi:10.1021/nl071018b
88. L. Liao, H.B. Lu, M. Shuai, J.C. Li, Y.L. Liu, C. Liu, Z.X. Shen, T. Yu, A novel gas sensor
based on field ionization from ZnO nanowires: moderate working voltage and high stability.
Nanotechnology 19(17), 175501 (2008). doi:Artn 175501; doi:10.1088/0957-4484/19/17/
175501
89. F. Qian, S. Gradečak, Y. Li, C.-Y. Wen, C.M. Lieber, Core/multishell nanowire
heterostructures as multicolor, high-efficiency light-emitting diodes. Nano Lett. 5(11),
2287–2291 (2005). doi:10.1021/nl051689e
90. V. Schmidt, H. Riel, S. Senz, S. Karg, W. Riess, U. Gösele, Realization of a silicon nanowire
vertical surround-gate field-effect transistor. Small 2(1), 85–88 (2006). doi:10.1002/smll.
200500181
91. P.D. Li, C.M. Sun, T.G. Jiu, G.J. Wang, J. Li, X.F. Li, J.F. Fangt, High-performance inverted
solar cells based on blend films of ZnO naoparticles and TiO2 nanorods as a cathode buffer
layer. ACS Appl. Mater. Inter. 6(6), 4074–4080 (2014)
92. J.P. Liu, S.S. Wang, Z.Q. Bian, M. Shan, C.H. Huang, Organic/inorganic hybrid solar cells
with vertically oriented ZnO nanowires. Appl. Phys. Lett. 94(17), 173107 (2009)
396 H. Karaağaç et al.

93. O. Lupan, V.M. Guerin, I.M. Tiginyanu, V.V. Ursaki, L. Chow, H. Heinrich, T. Pauporte,
Well-aligned arrays of vertically oriented ZnO nanowires electrodeposited on ITO-coated
glass and their integration in dye sensitized solar cells. J. Photoch. Photobio A 211(1), 65–73
(2010)
94. V. Strano, E. Smecca, V. Depauw, C. Trompoukis, A. Alberti, R. Reitano, I. Crupi, I.
Gordon, S. Mirabella, Low-cost high-haze films based on ZnO nanorods for light scattering
in thin c-Si solar cells. Appl. Phys. Lett. 106(1), 013901 (2015)
95. D.I. Suh, S.Y. Lee, T.H. Kim, J.M. Chun, E.K. Suh, O.B. Yang, S.K. Lee, The fabrication
and characterization of dye-sensitized solar cells with a branched structure of ZnO nanowires.
Chem. Phys. Lett. 442(4–6), 348–353 (2007)
96. M.T. Tsai, Z.P. Yang, T.S. Jing, H.H. Hsieh, Y.C. Yao, T.Y. Lin, Y.F. Chen, Y.J. Lee,
Achieving graded refractive index by use of ZnO nanorods/TiO2 layer to enhance
omnidirectional photovoltaic performances of InGaP/GaAs/Ge triple-junction solar cells.
Sol. Energy Mat. Sol C 136, 17–24 (2015)
97. J. Zhang, W.X. Que, P. Zhong, G.Q. Zhu, p-Cu2O/n-ZnO nanowires on ITO glass for solar
cells. J. Nanosci. Nanotechnol. 10(11), 7473–7476 (2010)
98. Y.F. Zhu, W.Z. Shen, Synthesis of ZnO nanoplates decorated rhombus-shaped ZnO
nanorods and their application in solar cells. Physica E 59, 110–116 (2014)
99. L.W. Ji, S.M. Peng, J.S. Wu, W.S. Shih, C.Z. Wu, I.T. Tang, Effect of seed layer on the
growth of well-aligned ZnO nanowires. J. Phys. Chem. Solids 70(10), 1359–1362 (2009).
doi:10.1016/j.jpcs.2009.07.029
100. L. De Marco, M. Manca, R. Giannuzzi, F. Malara, G. Melcarne, G. Ciccarella, I. Zama, R.
Cingolani, G. Gigli, Novel preparation method of TiO2-nanorod-based photoelectrodes for
dye-sensitized solar cells with improved light-harvesting efficiency. J. Phys. Chem. C 114(9),
4228–4236 (2010). doi:10.1021/Jp910346d
101. I.S. Cho, Z.B. Chen, A.J. Forman, D.R. Kim, P.M. Rao, T.F. Jaramillo, X.L. Zheng,
Branched TiO2 nanorods for photoelectrochemical hydrogen production. Nano Lett. 11(11),
4978–4984 (2011). doi:10.1021/Nl2029392
102. J.B. Chen, C.W. Wang, Y.M. Kang, D.S. Li, W.D. Zhu, F. Zhou, Investigation of
temperature-dependent field emission from single crystal TiO2 nanorods. Appl. Surf. Sci. 258
(20), 8279–8282 (2012). doi:10.1016/j.apsusc.2012.05.037
103. H.W. Lin, Y.H. Chang, C. Chen, Facile fabrication of TiO2 nanorod arrays for gas sensing
using double-layered anodic oxidation method. J. Electrochem. Soc. 159(1), K5–K9 (2012).
doi:10.1149/2.013201jes
104. J.K. Chen, W.Y. Fu, G.Y. Yuan, A. Runa, H. Bala, X.D. Wang, G. Sun, J.L. Cao, Z.Y.
Zhang, Fabrication of TiO2 nanocrystals/nanorods composites thin film electrode: Enhanced
performance of dye-sensitized solar cells. Mater. Lett. 135, 229–232 (2014)
105. K. Fan, W. Zhang, T.Y. Peng, J.N. Chen, F. Yang, Application of TiO2 fusiform nanorods
for dye-sensitized solar cells with significantly improved efficiency. J. Phys. Chem. C 115
(34), 17213–17219 (2011)
106. Z.M. He, J. Liu, J.W. Miao, B. Liu, T.T.Y. Tan, A one-pot solvothermal synthesis of
hierarchical microspheres with radially assembled single-crystalline TiO2-nanorods for high
performance dye-sensitized solar cells. J. Mater. Chem. C 2(8), 1381–1385 (2014)
107. Y.H. Jung, K.H. Park, J.S. Oh, D.H. Kim, C.K. Hong, Effect of TiO2 rutile nanorods on the
photoelectrodes of dye-sensitized solar cells. Nanoscale Res. Lett. 8, 37 (2013)
108. S.H. Kang, Thickness effect of single crystalline TiO2 nanorods for dye-sensitized solar cells.
J. Nanosci. Nanotechnol. 14(8), 6318–6321 (2014)
109. S. Kathirvel, C.C. Su, H.C. Lin, B.R. Chen, W.R. Li, Facile non-hydrolytic solvothermal
synthesis of one dimensional TiO2 nanorods for efficient dye-sensitized solar cells. Mater.
Lett. 129, 149–152 (2014)
110. P.L. Kuo, T.S. Jan, C.H. Liao, C.C. Chen, K.M. Lee, Syntheses of size-varied nanorods TiO2
and blending effects on efficiency for dye-sensitized solar cells. J. Power Sources 235,
297–302 (2013)
15 One-Dimensional Nano-structured Solar Cells 397

111. J. Liu, J. Luo, W.G. Yang, Y.L. Wang, L.Y. Zhu, Y.Y. Xu, Y. Tang, Y.J. Hu, C. Wang, Y.G.
Chen, W.M. Shi, Synthesis of single-crystalline anatase TiO2 nanorods with
high-performance dye-sensitized solar cells. J. Mater. Sci. Technol. 31(1), 106–109 (2015)
112. Y.D. Park, K. Anabuki, S. Kim, K.W. Park, D.H. Lee, S.H. Um, J. Kim, J.H. Cho,
Fabrication of stable electrospun TiO2 nanorods for high-performance dye-sensitized solar
cells. Macromol. Res. 21(6), 636–640 (2013)
113. M.K. Wang, J. Bai, F. Le Formal, S.J. Moon, L. Cevey-Ha, R. Humphry-Baker, C. Gratzel, S.
M. Zakeeruddin, M. Gratzel, Solid-state dye-sensitized solar cells using ordered TiO2 nanorods
on transparent conductive oxide as photoanodes. J. Phys. Chem. C 116(5), 3266–3273 (2012)
114. Y.L. Xie, P.C. Lin, S.Q. Hu, Y.C. Lu, L. Li, H. Wang, Growth of ZnO nanorods on TiO2
nanoparticles films and their application to the electrode of dye-sensitized solar cells.
J. Mater. Sci-Mater El 25(6), 2665–2670 (2014)
115. W.J. Zhang, Y. Xie, D.H. Xiong, X.W. Zeng, Z.H. Li, M.K. Wang, Y.B. Cheng, W. Chen,
K.Y. Yan, S.H. Yang, TiO2 nanorods: a facile size- and shape-tunable synthesis and effective
improvement of charge collection kinetics for dye-sensitized solar cells. ACS Appl. Mater.
Inter. 6(12), 9698–9704 (2014)
116. B.W. Luo, Y. Deng, Y. Wang, Z.W. Zhang, M. Tan, Heterogeneous flammulina
velutipes-like CdTe/TiO2 nanorod array: A promising composite nanostructure for solar
cell application. J. Alloy. Compd. 517, 192–197 (2012). doi:10.1016/j.jallcom.2011.12.090
117. H.B. Zhang, M.J. Zhang, C.B. Tian, N. Li, P. Lin, Z.H. Li, S.W. Du, An effective method for
the synthesis of 3D inorganic Ln(III)-K(I) sulfate open frameworks with unusually high
thermal stability: in situ generation of sulfate anions. J. Mater. Chem. 22(14), 6831–6837
(2012). doi:10.1039/C2jm16779d
118. M. Ghaffari, M.B. Cosar, H.I. Yavuz, M. Ozenbas, A.K. Okyay, Effect of Au nano-particles
on TiO2 nanorod electrode in dye-sensitized solar cells. Electrochim. Acta 76, 446–452
(2012). doi:10.1016/j.electacta.2012.05.058
119. H. Karaagac, L.E. Aygun, M. Parlak, M. Ghaffari, N. Biyikli, A.K. Okyay, Au/TiO2
nanorod-based schottky-type UV photodetectors. Phys. Status Solidi-R 6(11), 442–444
(2012). doi:10.1002/pssr.201206379
120. J. Tang, Z. Huo, S. Brittman, H. Gao, P. Yang, Solution-processed core-shell nanowires for
efficient photovoltaic cells. Nat. Nano 6(9), 568–572 (2011)
121. Y. Tak, S.J. Hong, J.S. Lee, K. Yong, Fabrication of ZnO/CdS core/shell nanowire arrays for
efficient solar energy conversion. J. Mater. Chem. 19(33), 5945–5951 (2009). doi:10.1039/
B904993B
122. L.E. Greene, M. Law, B.D. Yuhas, P. Yang, ZnO−TiO2 core−shell nanorod/P3HT solar
cells. J. Phys. Chem. C 111(50), 18451–18456 (2007). doi:10.1021/jp077593l
123. B. Tian, X. Zheng, T.J. Kempa, Y. Fang, N. Yu, G. Yu, J. Huang, C.M. Lieber, Coaxial
silicon nanowires as solar cells and nanoelectronic power sources. Nature 449
(7164):885–889 (2007)
124. G. Jia, M. Steglich, I. Sill, F. Falk, Core–shell heterojunction solar cells on silicon nanowire
arrays. Sol Energy Mat. Sol. C 96, 226–230 (2012). doi:10.1016/j.solmat.2011.09.062
125. M.M. Adachi, M.P. Anantram, K.S. Karim, Core-shell silicon nanowire solar cells. Sci.
Rep. 3 (2013). doi:http://www.nature.com/srep/2013/130326/srep01546/abs/srep01546.
html#supplementary-information
126. X. Zhao, P. Wang, Y. Gao, X. Xu, Z. Yan, N. Ren, CuO/ZnO core/shell nanowire arrays and
their photovoltaics application. Mater. Lett. 132, 409–412 (2014). doi:10.1016/j.matlet.2014.
06.124
127. B.M. Kayes, H.A. Atwater, N.S. Lewis, Comparison of the device physics principles of
planar and radial p-n junction nanorod solar cells. J. Appl. Phys. 97(11), 114302 (2005).
doi:10.1063/1.1901835
128. S.E. Han, G. Chen, Optical absorption enhancement in silicon nanohole arrays for solar
photovoltaics. Nano Lett. 10(3), 1012–1015 (2010). doi:10.1021/nl904187m
398 H. Karaağaç et al.

129. L. Aé, D. Kieven, J. Chen, R. Klenk, T. Rissom, Y. Tang, M.C. Lux-Steiner, ZnO nanorod
arrays as an antireflective coating for Cu(In, Ga)Se2 thin film solar cells. Prog. Photovoltaics
Res. Appl. 18(3), 209–213 (2010). doi:10.1002/pip.946
130. T. Stelzner, M. Pietsch, G. Andra, F. Falk, E. Ose, S. Christiansen, Silicon nanowire-based
solar cells. Nanotechnology 19(29), 295203 (2008)
131. S.A. Moiz, A.M. Nahhas, H.D. Um, S.W. Jee, H.K. Cho, S.W. Kim, J.H. Lee, A stamped
PEDOT:PSS-silicon nanowire hybrid solar cell. Nanotechnology 23(14), 145401 (2012)
132. A. Kim, Y. Won, K. Woo, C.-H. Kim, J. Moon, Highly transparent low resistance ZnO/Ag
nanowire/ZnO composite electrode for thin film solar cells. ACS Nano 7(2), 1081–1091
(2013). doi:10.1021/nn305491x
133. S. Cataldo, P. Salice, E. Menna, B. Pignataro, Carbon nanotubes and organic solar cells.
Energy Environ. Sci. 5(3), 5919–5940 (2012)
134. P.-L. Ong, W.B. Euler, I.A. Levitsky, Hybrid solar cells based on single-walled carbon
nanotubes/Si heterojunctions. Nanotechnology 21(10), 105203 (2010)
135. M.S. Dresselhaus GD, P. Avouris, Carbon Nanotubes: Synthesis, Properties and
Applications, vol. 80 (Springer, Berlin, 2001). doi:10.1007/3-540-39947-X
136. P.X. Hou, C. Liu, H.M. Cheng, Purification of carbon nanotubes. Carbon 46(15), 2003–2025
(2008)
137. L.A. Montoro, J.M. Rosolen, A multi-step treatment to effective purification of single-walled
carbon nanotubes. Carbon 44(15), 3293–3301 (2006)
138. C.M. Aguirre, S. Auvray, S. Pigeon, R. Izquierdo, P. Desjardins, R. Martel, Carbon nanotube
sheets as electrodes in organic light-emitting diodes. Appl. Phys. Lett. 88(18) (2006)
139. D.H. Zhang, K. Ryu, X.L. Liu, E. Polikarpov, J. Ly, M.E. Tompson, C.W. Zhou,
Transparent, conductive, and flexible carbon nanotube films and their application in organic
light-emitting diodes. Nano Lett. 6(9), 1880–1886 (2006)
140. B.B. Parekh, G. Fanchini, G. Eda, M. Chhowalla, Improved conductivity of transparent
single-wall carbon nanotube thin films via stable postdeposition functionalization. Appl. Phys.
Lett. 90(12), 121913 (2007)
141. T. Tanaka, Y. Urabe, D. Nishide, H. Kataura, Discovery of surfactants for
metal/semiconductor separation of single-wall carbon nanotubes via high-throughput
screening. J. Am. Chem. Soc. 133(44), 17610–17613 (2011)
142. P. Havu, M.J. Hashemi, M. Kaukonen, E.T. Seppälä, R.M. Nieminen, Effect of gating and
pressure on the electronic transport properties of crossed nanotube junctions: formation of a
Schottky barrier. J. Phys.: Condens. Matter 23(11), 112203 (2011)
143. Y. Jung, X.K. Li, N.K. Rajan, A.D. Tayor, M.A. Reed, Record high efficiency single-walled
carbon nanotube/silicon p-n junction solar cells. Nano Lett. 13(1), 95–99 (2013)
144. F. Hennrich, R. Wellmann, S. Malik, S. Lebedkin, M.M. Kappes, Reversible modification of
the absorption properties of single-walled carbon nanotube thin films via nitric acid exposure.
Phys. Chem. Chem. Phys. 5(1), 178–183 (2003)
145. R. Jackson, B. Domercq, R. Jain, B. Kippelen, S. Graham, Stability of doped transparent
carbon nanotube electrodes. Adv. Funct. Mater. 18(17), 2548–2554 (2008)
146. S.L. Hellstrom, M. Vosgueritchian, R.M. Stoltenberg, I. Irfan, M. Hammock, Y.B. Wang, C.
C. Jia, X.F. Guo, Y.L. Gao, Z.N. Bao, Strong and stable doping of carbon nanotubes and
graphene by MoOx for transparent electrodes. Nano Lett. 12(7), 3574–3580 (2012)
147. H.-Z. Geng, K.K. Kim, K.P. So, Y.S. Lee, Y. Chang, Y.H. Lee, Effect of acid treatment on
carbon nanotube-based flexible transparent conducting films. J. Am. Chem. Soc. 129(25),
7758–7759 (2007). doi:10.1021/ja0722224
148. S. Kim, J. Yim, X. Wang, D.D.C. Bradley, S. Lee, J.C. Demello, Spin- and spray-deposited
single-walled carbon-nanotube electrodes for organic solar cells. Adv. Funct. Mater. 20(14),
2310–2316 (2010)
15 One-Dimensional Nano-structured Solar Cells 399

149. M.S. Strano, V.C. Moore, M.K. Miller, M.J. Allen, E.H. Haroz, C. Kittrell, R.H. Hauge, R.E.
Smalley, The role of surfactant adsorption during ultrasonication in the dispersion of
single-walled carbon nanotubes. J. Nanosci. Nanotechnol. 3(1–2), 81–86 (2003)
150. H. Cebeci, R.G. de Villoria, A.J. Hart, B.L. Wardle, Multifunctional properties of high
volume fraction aligned carbon nanotube polymer composites with controlled morphology.
Compos. Sci. Technol. 69(15–16), 2649–2656 (2009)
151. A. Hagfeldt, M. Gratzel, Light-induced redox reactions in nanocrystalline systems. Chem.
Rev. 95(1), 49–68 (1995)
152. CCISolar, Dye sensitized solar cell—DSCC (2015). http://www.ccisolar.caltech.edu/index.
php?module=webpage&id=113&page=3
153. A.C. Fisher, L.M. Peter, E.A. Ponomarev, A.B. Walker, K.G.U. Wijayantha, Intensity
dependence of the back reaction and transport of electrons in dye-sensitized nanacrystalline
TiO2 solar cells. J. Phys. Chem. B 104(5), 949–958 (2000)
154. T. Oekermann, D. Zhang, T. Yoshida, H. Minoura, Electron transport and back reaction in
nanocrystalline TiO2 films prepared by hydrothermal crystallization. J. Phys. Chem. B 108
(7), 2227–2235 (2004)
155. R.C. Nelson, Energy transfers between sensitizer and substrate. III. sensitization by thick dye
films. J. Opt. Soc. Am. 51(11), 1182–1186 (1961). doi:10.1364/JOSA.51.001182
156. J. van de Lagemaat, A.J. Frank, Nonthermalized electron transport in dye-sensitized
nanocrystalline TiO2 films: transient photocurrent and random-walk modeling studies.
J. Phys. Chem. B 105(45), 11194–11205 (2001). doi:10.1021/jp0118468
157. T.Y. Lee, P.S. Alegaonkar, J.B. Yoo, Fabrication of dye sensitized solar cell using TiO2
coated carbon nanotubes. Thin Solid Films 515(12), 5131–5135 (2007)
158. K.-M. Lee, C.-W. Hu, H.-W. Chen, K.-C. Ho, Incorporating carbon nanotube in a
low-temperature fabrication process for dye-sensitized TiO2 solar cells. Sol. Energy Mat. Sol.
C 92(12), 1628–1633 (2008). doi:10.1016/j.solmat.2008.07.012
159. K.T. Dembele, G.S. Selopal, C. Soldano, R. Nechache, J.C. Rimada, I. Concina, G.
Sberveglieri, F. Rosei, A. Vomiero, Hybrid carbon nanotubes-TiO2 photoanodes for high
efficiency dye-sensitized solar cells. J. Phys. Chem. C 117(28), 14510–14517 (2013)
160. Z. Peining, A.S. Nair, Y. Shengyuan, P. Shengjie, N.K. Elumalai, S. Ramakrishna, Rice
grain-shaped TiO2–CNT composite—a functional material with a novel morphology for
dye-sensitized solar cells. J. Photochem. Photobiol. A 231(1), 9–18 (2012). doi:10.1016/j.
jphotochem.2012.01.002
161. L.J. Yang, W.W.F. Leung, Electrospun TiO2 nanorods with carbon nanotubes for efficient
electron collection in dye-sensitized solar cells. Adv. Mater. 25(12), 1792–1795 (2013)
162. H.S. Wroblowa, A. Saunders, Flow-through electrodes: II. The I3/−/I−redox couple.
J. Electroanal. Chem. Interfacial Electrochem. 42(3), 329–346 (1973). doi:10.1016/S0022-
0728(73)80323-7
163. K. Suzuki, M. Yamaguchi, M. Kumagai, S. Yanagida, Application of carbon nanotubes to
counter electrodes of dye-sensitized solar cells. Chem. Lett. 32(1), 28–29 (2003). doi:10.
1246/cl.2003.28
164. S. Uk Lee, W. Seok Choi, B. Hong, A comparative study of dye-sensitized solar cells added
carbon nanotubes to electrolyte and counter electrodes. Sol. Energy Mat. Sol. C 94(4), 680–
685 (2010). doi:10.1016/j.solmat.2009.11.030
165. S.H. Seo, S.Y. Kim, B.K. Koo, S.I. Cha, D.Y. Lee, Influence of electrolyte composition on
the photovoltaic performance and stability of dye-sensitized solar cells with multiwalled
carbon nanotube catalysts. Langmuir 26(12), 10341–10346 (2010)
166. J. Velten, A.J. Mozer, D. Li, D. Officer, G. Wallace, R. Baughman, A. Zakhidov, Carbon
nanotube/graphene nanocomposite as efficient counter electrodes in dye-sensitized solar
cells. Nanotechnology 23(8), 085201 (2012)
167. H. Anwar, A.E. George, I.G. Hill, Vertically-aligned carbon nanotube counter electrodes for
dye-sensitized solar cells. Sol. Energy 88, 129–136 (2013)
400 H. Karaağaç et al.

168. G.-R. Li, F. Wang, Q.-W. Jiang, X.-P. Gao, P.-W. Shen, Carbon nanotubes with titanium
nitride as a low-cost counter-electrode material for dye-sensitized solar cells. Angew. Chem.
Int. Ed. 49(21), 3653–3656 (2010). doi:10.1002/anie.201000659
169. Q.W. Jiang, G.R. Li, X.P. Gao, Highly ordered TiN nanotube arrays as counter electrodes for
dye-sensitized solar cells. Chem. Commun. 44, 6720–6722 (2009)
170. H.-J. Shin, S.S. Jeon, S.S. Im, CNT/PEDOT core/shell nanostructures as a counter electrode
for dye-sensitized solar cells. Synthetic Met. 161(13–14), 1284–1288 (2011). doi:10.1016/j.
synthmet.2011.04.024
Chapter 16
Computational Studies of Bismuth-Doped
Zinc Oxide Nanowires

Çetin Kılıç, Mehmet Aras and Sümeyra Güler-Kılıç

Abstract A computational modeling methodology is described, which is based on

the total-energy and defect calculations using the density and hybrid functionals in
combination with the supercell approach. This methodology is employed to study
the structure and energetics of zinc oxide (ZnO) nanowires, and the electrical
properties of a bismuth-doped zinc oxide (ZnO:Bi) nanowire. A simple model is
developed, where the nanowire formation energy is expressed as a function of the
nanowire diameter and cross-sectional morphology (hexagonal or triangular).
Defect calculations are employed to characterize the ZnO:Bi nanowire, in regard to
the location and charge-state of the dopant. The latter indicates that the dopant
atoms are mostly substituted into the Zn sites on the nanowire surface, and therefore
Bi in ZnO electrically acts as a donor.

16.1 Introduction

The doping of one-dimensional ZnO nanostructures has been attracting a growing

interest in recent decades [1], which is driven by not only scientific curiosity but
also potential device applications [2, 3]. In accord with this research trend, an
increasing number of computational studies, e.g., [4–10], has been devoted to the
modeling of doped ZnO nanowires in order to elucidate their behavior in regard to a
variety of phenomena such as p-type electrical conduction [11–13], which were
mostly performed within the framework of the density functional theory (DFT). In
this chapter, we utilize a modeling methodology for studying the doping of ZnO

Ç. Kılıç (&)
M. Aras
S. Güler-Kılıç

Department of Physics, Gebze Technical University, 41400 Gebze, Kocaeli, Turkey
e-mail: [email protected]
M. Aras
e-mail: [email protected]
S. Güler-Kılıç
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 401

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_16
402 Ç. Kılıç et al.

nanowires via computational methods, which is developed in Sect. 16.2. This

methodology is exemplified by application to (i) undoped ZnO nanowires in
Sect. 16.3, and (ii) a bismuth-doped zinc oxide (ZnO:Bi) nanowire in Sect. 16.4
where we focus on the electrical properties of this nanowire.
Although bismuth is known to act as an acceptor in ZnO thin films [14, 15], it
has a low solubility in zinc oxide [16] so that Bi-doped ZnO films usually exhibit n-
type electrical conduction. This is so because undoped ZnO is naturally n-type due
to the existence of intrinsic donors such as zinc interstitials and/or unintentional
dopants such as hydrogen, and therefore the holes produced by the acceptors are
compensated by the native defects of ZnO, hindering p-type doping [17]. On the
other hand, a recent experimental study [15] suggests that stable and reproducible p-
type conductivity could be achieved by doping ZnO thin films by Bi provided that
the bismuth atoms are incorporated substitutionally into the oxygen sites, forming
acceptor-like BiO defects. Needless to say, the latter is usually difficult due to the
segregation behavior of Bi in ZnO, which is a well-known phenomenon owing to
its occurrence in zinc oxide varistors[16, 18]. In the latter systems, Bi segregates to
the grain boundaries where it leads to the formation of acceptor-like defect com-
plexes [19]. As long as the p-doped grain boundaries could act as electron traps, the
electron-rich (i.e., n-type) interior regions are surrounded by an electrostatic barrier
created at the Bi–decorated grain boundaries. Hence, the varistor has a high
resistance at low voltages. The nonlinear current-voltage characteristic of the
varistor is produced by the release of the electrons at the breakdown voltage.
Accordingly, the varistor has a low resistance at high voltages. The electrical
properties of undoped and Bi-doped ZnO thin films are compared in Table 16.1. It
is seen that doping by Bi results in (i) a significant decrease in the electron con-
centration of the sample of [14], and (ii) a substantial increase in the hole con-
centration of the sample of [15]. Both of the latter are indicators for the acceptor
behavior originated from doping of ZnO by Bi.
The electrical behavior of ZnO:Bi nanowires should be contrasted with that of
thin films since bismuth acts as a donor in ZnO nanowires [20]. A comparison of
the electrical properties of undoped and Bi-doped ZnO nanowires is given in
Table 16.2. It is seen that both undoped and Bi-doped ZnO nanowires have n-type
conduction, and the electron concentration is increased by one order of magnitude
as a result of doping by Bi. Consequently, it is likely that Bi substitutes Zn site to

Table 16.1 Electrical properties of undoped and Bi-doped ZnO thin films
Conduction Carrier Mobility Resistivity Thickness References
type concentration (cm2/Vs) (Ω cm) (nm)
(cm−3)
ZnO n −2.7 × 1018 7.4 0.3 660 [14]
ZnO: n −9.1 × 1016 15.9 4.3 [14]
Bi
ZnO n −6.6 × 1017 122.1 0.077 200 [15]
ZnO: p 4.8 × 1019 5.9 0.022 [15]
Bi
16 Computational Studies of Bismuth-Doped Zinc Oxide Nanowires 403

Table 16.2 Electrical properties of undoped and Bi-doped ZnO nanowires

Conduction Carrier Mobility Resistivity Radius References
type concentration (cm2/Vs) (Ω cm) (nm)
(cm−1)
ZnO n −4.0 × 107 17.2 0.2 60 [21]
ZnO: n −3.5 × 108 1.5 0.1 20 [20]
Bi

form a donor-like defect BiZn in a positive charge state, releasing electrons [20].
Interestingly, the charge state of Bi in ZnO thin films is also positive [14], which
imply that BiZn forms in the sample of [14] too. On the contrary, the acceptor
behavior of bismuth in the sample of [15] is attributed to Bio, as mentioned above.
Hence, comparative analysis of experimental findings results in some degree of
controversy in regard to the questions: (i) What is the charge state of Bi in ZnO?
(ii) Does Bi substitute Zn or O sites? (iii) Does Bi act as a donor or acceptor in
ZnO? It is thus fortunate that the state and location of Bi in ZnO could also be
characterized by using the aforementioned modeling methodology, notwithstanding
the experimental controversies. It is the latter task that is accomplished in this
chapter, with focus on the nanowires.
The rest of this chapter is organized as follows. Section 16.2 is devoted to the
elements of modeling methodology employed in the following sections, where the
computational methods of calculation are described in detail. We present the results
of calculations performed for the structure and energetics of the undoped ZnO
nanowires in Sect. 16.3. The electrical properties of ZnO:Bi nanowire are studied in
Sect. 16.4 with the aid of defect calculations.

16.2 Computational Modeling

In the modeling of nanowires, we employ the total-energy and electronic-structure

calculations based on the (hybrid) density functional theory+U approach in combi-
nation with a variety of periodic supercells, where U is the Hubbard on-site repulsion.
The supercell approach is also adopted in our defect calculations that are utilized to
investigate doping of a zinc oxide nanowire with bismuth. Hence, in this section, we
first describe the supercell approach, which is followed by a discussion on the defect
calculations. Next, we expound on the density- and hybrid-functional+U calculations,
and lastly, a summary of the computational details are given.

16.2.1 Supercell Approach

The nanowires under consideration here are cut from bulk ZnO in wurtzite structure
in such a way that the wire axis coincide with the [0001] direction of wurtzite,
404 Ç. Kılıç et al.

Fig. 16.1 The tetragonal supercell of dimension 5c × 5c × c used to study undoped ZnO
nanowires. The [0001] direction of wurtzite is indicated by the vertical arrow

which have either a hexagonal or triangular cross section. The supercells containing
these nanowires are devised to include vacuum regions in order to make the
nanowire extend to infinity only along the [0001] direction of wurtzite although the
supercells themselves are subject to the Bloch periodicity condition in all directions.
Figure 16.1 displays a tetragonal supercell of dimension 5c × 5c × c that is
employed for undoped ZnO nanowires, where c denotes the length of periodicity
along the wire axis. A cubic supercell of side length L = 5c is used for doped ZnO
nanowires, which is displayed in Fig. 16.2. Generating a supercell configuration for
substitutional doping of ZnO nanowire is also schematically illustrated in the latter
figure. The relevant supercell configurations (i.e., atomistic structures contained in
the aforementioned supercells) are all optimized via minimization of the total
energy that is computed as described below in Sect. 16.2.3 (with the computational
settings given in Sect. 16.2.4). Prior to the latter, the equilibrium value of c is
determined for each nanowire, which corresponds to the energy minimum.

16.2.2 Defect Calculations

It has become customary to perform defect calculations using the supercell

approach in combination with the total-energy calculations (cf. Sect. 16.2.3) within
16 Computational Studies of Bismuth-Doped Zinc Oxide Nanowires 405

Fig. 16.2 The cubic supercell of length L = 5c employed for doped ZnO nanowires. The vertical
arrow indicates the wire axis that coincides with the [0001] direction of wurtzite. Generating a
supercell configuration for studying a Bi-doped ZnO nanowire is schematically illustrated, where
an extrinsic defect BiZn is created

the framework of a simple thermodynamic model where a host material is con-

sidered in equilibrium with the reservoirs of atoms and the Fermi sea of electrons.
The intrinsic and/or extrinsic defects are created by exchanging atoms and/or
electrons between the host and the reservoirs. This is illustrated schematically in
Fig. 16.2 for the substitutional doping of ZnO nanowire with Bi, where an extrinsic
defect is created by transferring a Zn (Bi) atom to (from) the zinc (bismuth)
reservoir from (to) the host, which will be denoted as BiZn. Note that a charged
defect BiqZn would be created if q electrons were also transferred to the Fermi
reservoir. It is also notable that transferring q electrons from the host to the Fermi
reservoir (from the Fermi reservoir to the host) implies using a positively-charged
(negatively-charged) supercell, which enables modeling of donor-like (acceptor-
like) defects. For BiqZn , the energetics of defect formation could be described as
follows: Assume that the ZnO nanowire is in equilibrium with the zinc, oxygen,
bismuth, and Fermi reservoirs consisting of mZn ; mO , mBi ; and me particles with
chemical potentials lZn ; lO , lBi ; and le , respectively. The total energy of a com-
bined system made of a defect-free nanowire and all reservoirs is
E0 þ mZn lZn þ mO lO þ mBi lBi þ me le , where E0 denote the total energy of the
supercell containing the defect-free nanowire. Once Zn and Bi atoms and q elec-
trons are exchanged between the nanowire and the reservoirs to form BiqZn in the
nanowire, the total energy becomes EðBiqZn Þ þ ðmZn þ 1ÞlZn þ mO lO þ ðmBi 
1ÞlBi þ ðme þ qÞle where EðBiqZn Þ denote the total energy of the supercell con-
taining the nanowire with defect BiqZn . Thus, the defect formation energy DHf ,
406 Ç. Kılıç et al.

which is defined as the change in energy associated with the formation of the
defect, is given by

DHf ðBiqZn Þ ¼ ½EðBiqZn Þ  E0  þ lZn  lBi þ qle ; ð16:1Þ

for BiqZn , which is clearly a function of the chemical potentials lZn , lBi , and le . It is
customary to set the zero of the atomic chemical potential la (α = Zn, O, or Bi) to
the energy per atom E(α) of an elemental form of the respective atomic species. On
the other hand, the electrochemical potential of electrons le , viz. the Fermi level EF ,
is conventionally referenced to the valence band maximum (VBM) energy Ev of the
host. Consequently, (16.1) is modified as

DHf ðBiqZn Þ ¼ ½EðBiqZn Þ  E0 þ EðZnÞ  EðBiÞ þ qEv  þ lZn  lBi þ qEF : ð16:2Þ

A similar derivation for the defect BiqO that is formed by substituting the dopant
on an O site (to replace a constituent O atom) results in

DHf ðBiqO Þ ¼ ½EðBiqO Þ  E0 þ EðOÞ  EðBiÞ þ qEv  þ lO  lBi þ qEF ; ð16:3Þ

which is a function of the chemical potentials lO , and lBi , and EF . If the dopant Bi
is incorporated into an interstitial site to form Biqi , then the defect formation energy
is given by

DHf ðBiqi Þ ¼ ½EðBiqi Þ  E0  EðBiÞ þ qEv   lBi þ qEF : ð16:4Þ

It should be emphasized that the defect formation energy DHf is an indicator for
the abundance of the defect under given thermodynamic conditions since it is a
significant portion of the Gibbs energy of formation that determines the equilibrium
defect concentration. DHf is therefore usually computed for a relevant range of the
atomic chemical potentials la as well as the Fermi level EF . It is further to be
reminded that this simple thermodynamic model has long been used to study the
defect structure of solids [21–23] forasmuch as the derivations given above are
generally valid for any material system including solids, which was also employed
in former studies [24–27] of one of the authors.
It is clear in the foregoing formulation that the defect formation energy DHf is
linearly proportional (with a negative slope) to the Fermi level EF for q > 0 (q < 0)
while DHf is independent of EF for q = 0, cf. (16.2)–(16.4). Thus DHf is usually
studied as a function of EF in order to determine the charge state of the defect (or
dopant) under consideration. Note that the value of EF eventually depends on the
free-carrier concentration. The defect transition energy eðq1 =q2 Þ of defect D is
defined as the value of EF for which DHf ðDq1 Þ ¼ DHf ðDq2 Þ, which is obtained via

EðDq2 Þ  EðDq1 Þ
eðq1 =q2 Þ ¼ : ð16:5Þ
q1  q2
16 Computational Studies of Bismuth-Doped Zinc Oxide Nanowires 407

Note that eðq1 =q2 Þ is independent of the atomic chemical potentials la . If EF is

smaller (larger) than eðq1 =q2 Þ, then the charge state of D is q1 ðq2 Þ. Thus, eðq1 =q2 Þ
gives the value of the Fermi level at which the transition from charge state q1 to q2
takes place. The defect transition energies correspond to thermal ionization ener-
gies, and accordingly they are used to determine the position of electrical (donor or
acceptor) levels with respect to the band edges in order to reveal the behavior of
defects (or dopants). For example, considering a defect with possible charge states
q = −, 0, and +, if the difference Ev  eð0=Þ ½eð þ =0Þ  Ec  is positive (or neg-
ative with a magnitude within a few kB T), then the defect yields a shallow acceptor
[donor] level. Here Ec denotes the conduction band minimum (CBM) energy.
Otherwise, the electrical level is deep since it would require an amount of energy
appreciably greater than the thermal energy kB T to ionize this defect.
As for the atomic chemical potentials fla g appearing in (16.2)–(16.4), it should
be reminded that the actual value of la depends on the kinetic or thermodynamic
conditions during growth or in the course of equilibration processes. It is thus
conventional to employ a lower and an upper bound for each la , which would
enable one to study typical representative cases. For the constituent atoms Zn and
O, these lower and upper bounds are determined as follows: It is required to have
lZn ; lO  0 because the host system (i.e., ZnO nanowire) would otherwise
decompose into its elemental constituents. Moreover,

lZn þ lO ¼ DHN ; ð16:6Þ

assuming that chemical equilibrium is maintained between the host system and the
atomic reservoirs, where DHN denotes the heat of formation (per formula unit) of
the nanowire made of N Zn–O pairs. The latter implies DHN  la ða ¼ Zn or OÞ so
that the lower bound is equal to DHN . Hence

DHN  la  0 ða ¼ Zn or OÞ: ð16:7Þ

la is equal to upper (lower) bound if the host system is in an environment that is

extremely rich (poor) in element α. For example, O–poor condition implies
lO ¼ DHN , whereas O–rich condition means lO ¼ 0. Accordingly, lZn ¼ 0 and
lZn ¼ DHN under O–poor (i.e., Zn–rich) and O–rich (i.e., Zn–poor) conditions,
respectively.
The value of the Bi chemical potential is set to the adsorption energy of a Bi atom
on the outer surface of the ZnO nanowire. The lowest-energy adatom configuration
is shown in Fig. 16.3, for which the adsorption energy Ead ¼ EðBi=nanowireÞ 
EðnanowireÞ  EðBiÞ is −0.577 eV. Note that Ead \0 for the adatom configuration
displayed in Fig. 16.3, indicating the attractive interaction between the adatom and
the nanowire. As for the results presented in Sect. 16.4, we set lBi ¼ Ead . The latter
means that the zero of lBi is taken as the energy of the Bi atom EðBiÞ. On the other
hand, the ground-state phase is used as a reference for the constituent elements of the
nanowire, viz. Zn and O, i.e., the zero of lZn and lO are set to the energy per atom
EðZnÞ and EðOÞ of bulk Zn metal and O2 molecule, respectively.
408 Ç. Kılıç et al.

Fig. 16.3 The lowest-energy configuration employed to calculate the adsorption energy Bi on the
ZnO nanowire surface

16.2.3 Density- and Hybrid-Functional+U Calculations

First-principles calculations based on the density functional theory [28, 29]

(DFT) have been the de facto method of choice in ab initio materials modeling for
computational efficiency [30]. It is nevertheless notable that the success (or failure)
of DFT in realistic modeling depends inherently on the quality of approximation for
the exchange-correlation energy Exc functional. The standard DFT calculations
performed within the local density approximation [29] (LDA) or the generalized
gradient approximation using the Perdew-Burke-Ernzerhof (PBE) functional [31]
are not entirely satisfactory for modeling materials containing localized electrons
such as some transition-metal or rare-earth compounds. Thus, the local/semilocal
functionals are often either combined with Hubbard U, as in the LDA+U method
[32], or mixed with a fraction of Hartree-Fock exchange energy [33], as in the hybrid
functional scheme [34]. In a recent study [35] devoted to a set of II–VI semicon-
ductors with localized d states, we found that employing the hybrid functional
scheme and the DFT+U method as complementary approaches results in an
improved description not only of the electronic structure but also of the crystal
structure and energetics. In contrast, the improvements due to employing a hybrid
functional or Hubbard U alone are not sufficient to make the predictions agree with
the experimental data, especially for wide band gap semiconductors such as ZnO that
has localized d electrons. The latter is particularly valid in predicting the band gap Eg
as long as the band gap is underestimated in proportionality with 1=e1 , cf. the
Appendix of [35], where e1 denotes the high-frequency dielectric constant. It is
demonstrated in [35] that this well-known band-gap problem could be cured by
adopting a hybrid-functional+U approach where the range-separated
Heyd-Scuseria-Ernzerhof (HSE) hybrid functional [36] is combined with the
Hubbard U. This can be briefly explained as follows: the band-gap correction due to
the Hubbard term turns out to be proportional to U=e1 , cf. Fig. 4 of [35], which
complements the HSE-calculated band gap (being underestimated in proportionality
with 1=e1 ) provided that U is set to have a suitable value. Consequently, the suitable
U value is determined by simply matching the experimental band gap, which is
denoted as U*. This is illustrated in Fig. 16.4 where the calculated band gap is
plotted as a function of U. The variation of the band gap with U is seemingly linear.
16 Computational Studies of Bismuth-Doped Zinc Oxide Nanowires 409

Fig. 16.4 Setting the value of U* by matching the experimental band gap. For wurtzite ZnO bulk
crystal, the band gaps (solid circles) obtained in our HSE+U calculations is plotted with respect to
the effective Hubbard parameter U. The red line represents a linear fit to the calculated points. The
vertical line marks U*

Table 16.3 The calculated (PBE, PBE+U, HSE, HSE+U*) and experimental values of the
wurtzite lattice parameters a and c, the wurtzite internal parameter u, the formation energy DHf ,
the d band position ed relative to the valence band maximum, and the band gap Eg for wurtzite
ZnO bulk crystal
Method a(Å) c(Å) c/a u DHf (eV) ed (eV) Eg (eV)
PBE 3.2872 5.3059 1.6141 0.3790 −2.8858 −4.66 0.72
PBE+U 3.1801 5.1208 1.6102 0.3798 −3.6770 −6.25 1.91
HSE 3.2606 5.2332 1.6023 0.3811 −3.1196 −5.71 2.45
HSE+U* 3.1850 5.1070 1.6036 0.3812 −3.7409 −7.03 3.37
Experiment 3.2495 5.2069 1.6024 0.3819 −3.6326 −7.50 3.37

Hence, the U value that corresponds to the experimental band gap is obtained via a
linear fit, yielding U  ¼ 5:95 eV for ZnO.
It should be emphasized that the results obtained for a set of II–VI semicon-
ductors with d electrons [35] prove that an improved description of the electronic
and crystal structure and energetics is obtained via HSE+U* calculations, compared
to the respective HSE or PBE+U calculations. Comparative inspection of rows of
Table 16.3 reveals that the latter is particularly true for wurtzite ZnO bulk crystal,
where the calculated and experimental values are given for the following quantities:
the wurtzite lattice parameters a and c, the wurtzite internal parameter u, the for-
mation energy DHf , the d band position ed relative to the valence band maximum,
and the band gap Eg .
Within the HSE+U approach, the exchange-correlation energy functional is
effectively generalized to
410 Ç. Kılıç et al.

HSE þ U
Exc ¼ aExHF;SR ðxÞ þ ð1  aÞExPBE;SR ðxÞ þ ExPBE;LR ðxÞ þ EcPBE
!
UX X r X r r ð16:8Þ
þ qii  qij qji ;
2 r i i;j

where the first line gives the range-separated HSE functional, and the simplified
(rotationally invariant) formulation [37] of the DFT+U method is adopted in the
second line. Note that the effective Hubbard parameter U ¼ U  J is introduced as
the difference between the on-site Coulomb U and exchange J parameters in the
latter formulation [37] where qrij denotes the density matrix of localized electrons
with spin projection σ. In (16.8), ExHF;SR and ExPBE;SR are the short-range contri-
bution from the HF and PBE exchange functionals, respectively, while ExPBE;LR is
the long-range contribution from the PBE exchange functional, and EcPBE denotes
the correlation energy functional within the generalized gradient approximation.
Note that range separation is accomplished by splitting the Coulomb operator into
two components with the aid of the screening parameter ω, i.e.,
1=r ¼ ½1  erfðxrÞ=r þ erfðxrÞ=r, and the screened Coulomb potential is applied
only to the exchange interaction [36]. The exchange mixing coefficient α is then
used only for the short-range components of the HF and PBE exchange functionals.
HSE þ U
Clearly, Exc reduces to the HSE functional for U ¼ 0. On the other hand, the
PBE functional is recovered in (16.8) for U ¼ 0; a ¼ 0, and x ! 1. Hence, (16.8)
yields the PBE+U functional for U 6¼ 0; a ¼ 0, and x ! 1.
It should be pointed out from a practical point of view that employing
hybrid-functionals for the defect calculations (cf. Sect. 16.2.2) becomes an
exhaustive task once one needs to use large supercells, e.g., Fig. 16.2, especially if
the computational resources are moderate. For this reason and encouraged by the
results of a number of test calculations, we devised a computational strategy where
the PBE+U and HSE+U* calculations are employed successively: The former is
used to perform structure optimizations, which is followed by computing the
electronic structure via the latter. We found that an adequate U value for PBE
+U could be determined so that structure optimizations via the PBE+U and HSE
+U* calculations yield very similar equilibrium geometries. In practice, the U value
for PBE+U could be set by matching some structural parameter predicted via the
PBE+U and HSE+U* calculations. This is demonstrated in Fig. 16.5 for the unit
cell volume. It is seen in Fig. 16.5 that the PBE+U and HSE+U* values for the
equilibrium unit cell volume turn out to be identical for U ¼ 7:7 eV, given that
U  ¼ 5:95 eV. Test calculations performed for a thin ZnO nanowire with the
foregoing values of U and U* indicate that employing the PBE+U and HSE+U*
calculations successively introduces an uncertainty on the order of 10−2 Å in the
equilibrium positions of the nanowire atoms.
16 Computational Studies of Bismuth-Doped Zinc Oxide Nanowires 411

Fig. 16.5 Setting the value of U for PBE+U by matching the HSE+U*-optimized unit cell
volume. For wurtzite ZnO bulk crystal, the PBE+U-optimized volume (empty circles) is plotted
with respect to the effective Hubbard parameter U. The red curve represents a polynomial fit to the
calculated points. The vertical line marks the value of U for PBE+U

16.2.4 Computational Settings

The density-functional (PBE+U) as well as hybrid-functional (HSE+U*) calcula-

tions were carried out using the projector augmented-wave (PAW) method [38], as
implemented [39] in the Vienna ab initio simulation package [40] (VASP). Plane
wave basis sets were used to represent the electronic states, which were determined
by imposing a kinetic energy cutoff of 400 eV. The 2s and 2p, 3d and 4s, and 6s and
6p states were treated as valence states for oxygen, zinc, and bismuth, respectively.
The Hubbard U was applied only to the Zn 3d states. The Brillouin zone of the
tetragonal supercells used for undoped nanowires, cf. Fig. 16.1, was sampled by
1 × 1 × 7 k-point mesh, which were generated according to the Monkhorst-Pack
scheme [41]. The Γ point was used for sampling the Brillouin zone of the cubic
supercells employed for doped nanowires, cf. Fig. 16.2. Convergence criterion for
the electronic self-consistency was set up to 10−6 eV. The charged supercell cal-
culations were performed with the aid of a neutralizing jellium background. By
studying the supercell energy as a function of supercell size L, we estimated that the
finite-size error is on the order of 0.2 eV. In the optimization of the atomistic
structures, ionic relaxations were performed for each supercell configuration to
minimize the total energy, until the maximum value of residual forces on atoms was
reduced to be smaller than 10−2 eV/Å.
412 Ç. Kılıç et al.

16.3 Structure and Energetics of ZnO Nanowires

The optimized structures of the ZnO nanowires studied here are displayed in
Fig. 16.6, which have either hexagonal (upper panels) or triangular (lower panels)
cross sections. The chemical formula (ZnO)N is used throughout to indicate these
structures. Note that N = 6, 24, or 54 (13, 22, or 33) for the hexagonal (triangular)
nanowires. The inradius r and the circumradius R are introduced to differentiate the
interior (i.e., bulk-like) and surface regions, which are indicated by the green and blue
circles, respectively, in Fig. 16.6. The diameter D of the nanowire is given by
pffiffiffiffiffi pffiffiffiffiffi
D ¼ 2 rR, which is set to be equal to twice the midradius rR. D is plotted with
respect to the inverse number of Zn–O pairs (1/N) in the upper panel of Fig. 16.7. The
middle and lower panels in Fig. 16.7 display the percent changes ΔD and Δc in D and
c, respectively, as a function of 1/N, relative to the unrelaxed nanowire structure. It is
seen that the former takes only negative values, i.e., DD\0, while the latter takes
only positive values, i.e., 0\Dc, for all nanowires, upon structural relaxation. This
means that the equilibrium structures of the nanowires are elongated along the wire
axis and at the same time radially shrinked, compared to the unrelaxed structures.
Moreover, the variation of the curves in the middle and lower panels of Fig. 16.7
follows a similar trend, indicating that the radial shrinkage is roughly proportional to
the axial elongation. The equilibrium structures displayed in Fig. 16.6 are further
characterized by measuring the O–Zn interatomic distances that are given in
Table 16.4. The optimized (PBE+U) values for the latter in wurtzite ZnO bulk crystal
are 1.94 and 1.95 Å for O(n)–Zn(n) and O(n)–Zn(n′) with n; n0 ! 1, respectively. It
is thus clear that the O–Zn interatomic distances are equal to or slightly larger than
the respective bulk values in the interior region of the nanowires, which indicates a
bulk-like coordination. On the contrary, the O–Zn interatomic distances are signifi-
cantly smaller than the respective bulk values for the surface region of the nanowires,
which implies the occurrence of substantial surface reconstruction. Accordingly, the
relaxed nanowires possess a heterogeneous structure.
Figure 16.8 shows the variation of the negative of the nanowire formation energy
DHN with the ratio Ns =N where Ns denotes the number of Zn–O pairs residing on
the outer surface of the nanowire. The solid symbols in the upper and lower panels of
Fig. 16.8 represent the calculated (PBE+U and HSE+U*, respectively) values. In
each graph, the solid diamond (square) marks the experimental (calculated) value for
DHf of wurtzite ZnO bulk crystal. The calculation results (i.e., the solid symbols)
are extrapolated to the bulk limit with a model equation given by

N  Ns Ns  Nl Nl
DHN ¼ Cb þ Cs þ Cl ; ð16:9Þ
N N N

which is drawn by the solid line in Fig. 16.8. Here Nl denotes the number of Zn–O
pairs located at the edges of the nanowire, Cb , Cs and Cl are the fitting coefficients,
whose values are given in Table 16.5. Note that (i) Nl =N, (ii) ðNs  Nl Þ=N, and
(iii) ðN  Ns Þ=N give the fraction of Zn–O pairs located (i) at the edges, (ii) on the
16 Computational Studies of Bismuth-Doped Zinc Oxide Nanowires 413

Fig. 16.6 Optimized atomistic structures of ZnO nanowires with hexagonal (upper panels) or
triangular (lower panels) cross sections. The surface region of a nanowire consists of atoms
confined between the green and blue circles of radius r and R, respectively, while atoms confined
within the green circle, with a bulk-like environment, reside in the interior region of the nanowire

faces, and (iii) in the interior region of the nanowire. In Fig. 16.8, the empty
symbols represent the values obtained via (16.9). Hence, the formation energy is
obtained for any size, i.e., 6  N  1, which would enable one to study the size-
and shape-dependent energetics of ZnO nanowires. It has recently been
414 Ç. Kılıç et al.

Fig. 16.7 The diameter D (upper panel), the percent change in D (middle panel) and c (lower
panel) as a function of the inverse number of Zn–O pairs

demonstrated that the size- and shape-dependent energetics of a variety of

nanocrystals [42–44] could be studied in a similar fashion. It is to be noted that this
way of studying the nanocrystal energetics makes it possible to extend the com-
putational findings obtained for small systems to large systems as long as (16.9) is
valid not only ultrathin wires such as (ZnO)6 but also ZnO bulk crystal.

16.4 Defect Energetics and Transition Levels in ZnO:Bi

Nanowire

In the section, we first employ defect calculations to obtain the formation energy for
the Bi dopants incorporated into the (ZnO)24 nanowire on either the substitutional
or the interstitial sites. We then utilize the latter to determine the charge-state
16 Computational Studies of Bismuth-Doped Zinc Oxide Nanowires 415

Table 16.4 The O–Zn interatomic distances (in Å) in (ZnO)N nanowires

Region ðZnOÞ6 ðZnOÞ13 ðZnOÞ22 ðZnOÞ24 ðZnOÞ33 ðZnOÞ54
Surface O(1)–Zn(1) 1.88 1.85 1.85 1.83 1.85 1.82
O(1)–Zn(3) 1.89 1.90
O(2)–Zn(1) 1.92 1.92 1.91 1.92 1.91
O(2)–Zn(2) 1.88 1.88 1.87
O(3)–Zn(3) 1.86
O(1)–Zn(1´) 1.83 1.83 1.83 1.81 1.83 1.81
O(2)–Zn(2´) 1.99 1.97 1.96 1.99 1.95
O(3)–Zn(3´) 1.80
Interior O(4)–Zn(4) 1.95 1.94 1.94 1.92,1.94
O(5)–Zn(5) 1.95
O(5)–Zn(4) 1.96
O(4)–Zn(4´) 2.00 1.97 1.99 1.95,1.97
O(5)–Zn(5´) 1.96
The respective values in wurtzite ZnO bulk crystal are 1.94 and 1.95 Å for O(n)–Zn(n) and O(n)–
Zn(n′) with n; n0 ! 1, respectively

Fig. 16.8 The formation energy DHN of (ZnO)N nanowires obtained via the PBE+U (upper
panel) and HSE+U* (lower panel) calculations versus the ratio of the number of Zn–O pairs
located on the outer surface of the nanowire Ns to N
416 Ç. Kılıç et al.

Table 16.5 The values of Cb , Cross section Cb Cs Cl

Cs and Cl (in eV per formula
unit) introduced in (16.9) Hexagonal PBE+U 3.684 3.264 2.731
HSE+U* 3.750 3.277 2.707
Triangular PBE+U 3.685 3.201 2.340
HSE+U* 3.752 3.213 2.286

Fig. 16.9 The cross marks

represent possible doping
sites for doping (ZnO)24
nanowire with Bi

transition energies in order to reveal the electrical behavior of the Bi dopant. For the
location of the dopant in the host nanowire, possible doping sites are displayed in
Fig. 16.9, which are represented by the cross marks. The latter consist of a number
of interior and surface sites where Bi incorporation would cause the following
extrinsic defects:
• BiZn ðbÞ, i.e., the dopant replacing a Zn atom in the bulk-like region
• BiO ðbÞ, i.e., the dopant replacing an O atom in the bulk-like region
• Bii ðbÞ, i.e., the dopant placed on the interstitial site at the center of the nanowire
• BiZn ðssÞ, i.e., the dopant replacing a Zn atom in the subsurface region
• BiO ðssÞ, i.e., the dopant replacing an O atom in the subsurface region
• BiZn ðsÞ, i.e., the dopant replacing a Zn atom on the surface
• BiO ðsÞ, i.e., the dopant replacing an O atom on the surface
Note that only inequivalent sites are considered. Furthermore, the sites where Bi
incorporation is found to lead to an unstable configuration are not represented.
Figure 16.10 displays the plots of calculated formation energy as a function of
the Fermi level EF for the foregoing surface (lower panels) and interior (upper
panels) defects, cf. Fig. 16.9, under O–poor (left panels) and O–rich (right panels)
conditions. The valence band maximum (VBM) and conduction band minimum
(CBM) of the nanowire are marked by the vertical lines. The defect (charge-state)
16 Computational Studies of Bismuth-Doped Zinc Oxide Nanowires 417

Fig. 16.10 The plots of formation energy as a function of the Fermi level EF for the surface (lower
panels) and interior (upper panels) defects, cf. Fig. 16.9, under O–poor (left panels) and O–rich
(right panels) conditions. The valence band maximum (VBM) and conduction band minimum
(CBM) are marked by the vertical lines. The defect (charge-state) transition energies are indicated
by the solid circles

transition energies are indicated by the solid circles. Figure 16.11 is plotted in order
to represent the transition energies (horizontal line segments) relative to the band
edge energies. Table 16.6 lists the energy differences Ec  eðq1 =q2 Þ (for donor-like
defects) and eðq1 =q2 Þ  Ev (for acceptor-like defects), where Ec and Ev denote the
CBM and VBM energies, respectively.
Analyzing the results presented in Fig. 16.10, the formation energy is negative
for the various foregoing defects for some range of the Fermi level. Thus, it is likely
that many of these defects will form spontaneously under suitable conditions. Note
that a defect forms spontaneously if its formation energy is negative, i.e., DHf  0.
However, one should also notice that the formation energy of BiZn ðsÞ is almost
always lower then that of the other defects regardless of the value of the Fermi level
as well as the atomic chemical potentials. The only exception to this is that BiO ðsÞ
has a lower formation energy for a narrow range of EF around the CBM energy (as
in n-type ZnO). Accordingly, it is most likely that BiZn ðsÞ dominates in the defect
418 Ç. Kılıç et al.

Table 16.6 The calculated (PBE+U) values for Ec  eðq1 =q2 Þ and eðq1 =q2 Þ  Ev
Defect q1 q2 Ec  eðq1 =q2 Þ (eV) eðq1 =q2 Þ  Ev (eV)
Donor-like Bii ðbÞ +++ ++ 0.98
++ + 0.58
BiZn ðbÞ + 0 0.12
BiZn ðsÞ + 0 0.14
(0.10)
BiZn ðssÞ + 0 0.15
Acceptor-like BiO ðbÞ 0 – 1.85
BiO ðsÞ 0 – 1.83
(2.61)
BiO ðssÞ 0 – 1.68
The values given in parentheses are obtained via the HSE+U* calculations

Fig. 16.11 The positions of electrical levels, i.e., the defect transition energies eðq1 =q2 Þ, are
indicated by the horizontal line segments

structure of ZnO:Bi nanowire. If we restrict our attention to the defects forming in

the bulk-like region and compare the interior defects among themselves, cf. the
upper panels in Fig. 16.11, we see that BiZn ðbÞ has the lowest formation energy
relative to other defects forming in the bulk-like region. Thus, one could conclude
that the incorporation of Bi as a substitutional dopant on the Zn sites is more likely,
compared to the rest of the alternatives. This conclusion is clear for O–rich con-
dition. However, the situation is more subtle for n-type nanowire under O–poor
conditions. In the latter case, both BiZn ðsÞ and BiO ðsÞ have negative formation
energies so that the Bi atoms substitute into not only Zn but also O sites. It is also
interesting to note that a segregation behavior could be attributed to Bi in ZnO
nanowire because the surface defects have lower formation energies compared to
the respective bulk-like defects.
The defect transition energies derived from the formation energies, i.e., the
positions of electrical levels, are presented in Fig. 16.11 which is accompanied by
16 Computational Studies of Bismuth-Doped Zinc Oxide Nanowires 419

Table 16.6 where the numerical values are given for Ec  eðq1 =q2 Þ and
eðq1 =q2 Þ  Ev . It is clear that the donor-like defects BiZn ðbÞ; BiZn ðssÞ, and BiZn ðsÞ
all form shallow electrical levels (since Ec  eðq1 =q2 Þ  0:1 eV) whereas the
acceptor-like defects BiO ðbÞ; BiO ðssÞ, and BiO ðsÞ all form deep electrical levels
(since eðq1 =q2 Þ  Ev [ 1 eV). Note that the charge state of Bii ðbÞ is + in n-type
ZnO whose Fermi level is near to the CBM, indicating that Bii ðbÞ would act as a
shallow donor too. Thus, Bi forms either a shallow donor or a deep acceptor in the
ZnO nanowire regardless of the location of doping site. This indicates that the
incorporation of Bi in ZnO nanowire would not produce p-type doping even if Bi
substitutes into the O site as an acceptor. This analysis makes it clear that Bi in ZnO
electrically acts as a donor.
Since the band gap is not well-described in semilocal density approximations
[35], which is known to cause problems in the defect calculations performed for
bulk ZnO, e.g., [45], we compute the transition energies for BiZn ðsÞ and BiO ðsÞ
within HSE+U* description. Here it is of interest to see if the energy differences
given in Table 16.6 vary or not while the band gap increases from 2.35 eV (PBE
+U) to 3.75 eV (HSE+U*). As seen in Fig. 16.11, the position of eð0=Þ of BiO ðsÞ
remains relatively unchanged on the absolute scale whereas eð þ =0Þ of BiZn ðsÞ
roughly follows the upward shift of CBM. Thus, the difference Ec  eðq1 =q2 Þ
remains relatively unchanged in Table 16.6 whereas there is a significant increase in
the value of eðq1 =q2 Þ  Ev . The latter indicates that the band edge correction
obtained via the HSE+U* description makes the electrical level of BiZn ðsÞ and
BiO ðsÞ become even more shallow and deeper, respectively. Hence, the conclusion
of the preceding paragraph (i.e., that Bi in ZnO electrically acts as a donor) is not
susceptible to the band gap error.

Acknowledgments The authors acknowledge financial support from TUBITAK through Grant
114F155.

References

1. C.-L. Hsu, S.-J. Chang, Doped ZnO 1D nanostructures: synthesis, properties, and
photodetector application. Small 10, 1613–6829 (2014)
2. L. Schmidt-Mende, J.L. MacManus-Driscoll, ZnO—nanostructures, defects, and devices.
Mater. Today 10, 40–48 (2007)
3. S. Xu, Z.L. Wang, One-dimensional ZnO nanostructures: solution growth and functional
properties. Nano Research 4, 1013–1098 (2011)
4. Q. Wang, Q. Sun, G. Chen, Y. Kawazoe, P. Jena, Vacancy-induced magnetism in ZnO thin
films and nanowires. Phys. Rev. B 77, 205411 (2008)
5. R. Qin, J. Zheng, J. Lu, L. Wang, L. Lai, G. Luo, J. Zhou, H. Li, Z. Gao, G. Li, W.N. Mei,
Origin of p-type doping in zinc oxide nanowires induced by phosphorus doping: A first
principles study. J. Phys. Chem. C 113, 9541–9545 (2009)
6. G. Jingyun, Z. Xinzheng, S. Yanghui, Z. Qing, Y. Dapeng, Compensation mechanism in
N-doped ZnO nanowires. Nanotechnology 21, 245703 (2010)
7. Y. Li, X. Zhao, W. Fan, Structural, electronic, and optical properties of Ag-doped ZnO
nanowires: first principles study. J. Phys. Chem. C 115, 3552–3557 (2011)
420 Ç. Kılıç et al.

8. M.A. Thomas, W.W. Sun, J.B. Cui, Mechanism of Ag doping in ZnO nanowires by
electrodeposition: experimental and theoretical insights. J. Phys. Chem. C 116, 6383–6391
(2012)
9. D. Wang, Q. Chen, G. Xing, J. Yi, B.S. Rahman, J. Ding, J. Wang, T. Wu, Robust
room-temperature ferromagnetism with giant anisotropy in Nd-doped ZnO nanowire arrays.
Nano Lett. 12, 3994–4000 (2012)
10. D.D. Wang, G.Z. Xing, F. Yan, Y.S. Yan, S. Li, Ferromagnetic (Mn, N)-codoped ZnO
nanopillars array: Experimental and computational insights. Appl. Phys. Lett. 104, 022412
(2014)
11. Y.-W. Song, K. Kim, J.P. Ahn, G.-E. Jang, S.Y. Lee, Physically processed Ag-doped ZnO
nanowires for all-ZnO p–n diodes. Nanotechnology 20, 275606 (2009)
12. J.S. Lee, S.N. Cha, J.M. Kim, H.W. Nam, S.H. Lee, W.B. Ko, K.L. Wang, J.G. Park, J.
P. Hong, p-type conduction characteristics of lithium-doped ZnO nanowires. Adv. Mater. 23,
4183–4187 (2011)
13. S.K.S. Parashar, B.S. Murty, S. Repp, S. Weber, E. Erdem, Investigation of intrinsic defects in
core-shell structured ZnO nanocrystals. J. Appl. Phys. 111, 113712 (2012)
14. F.X. Xiu, L.J. Mandalapu, Z. Yang, J.L. Liu, G.F. Liu, J.A. Yarmoff, Bi-induced acceptor
states in ZnO by molecular-beam epitaxy. Appl. Phys. Lett. 85, 052103 (2006)
15. J.W. Lee, N.G. Subramaniam, J.C. Lee, S. Kumar, T.W. Kang, Study of stable p-type
conductivity in bismuth-doped ZnO films grown by pulsed-laser deposition. EPL 95, 47002
(2011)
16. A. Smith, J.F. Baurnard, P. Abelard, M.F. Denanot, ac impedance measurements and V-I
characteristics for Co-, Mn-, or Bi-doped ZnO. J. Appl. Phys. 65, 5119–5125 (1989)
17. Ü. Özgür, I. Alivov Ya, C. Liu, A. Teke, M.A. Reshchikov, S. Doğan, V. Avrutin, S.-J. Cho,
H. Morkoç, A comprehensive review of ZnO materials, devices. J. Appl. Phys. 98, 041301
(2005)
18. K.-I. Kobayashi, O. Wada, M. Kobayashi, Y. Takada, Continuous existence of bismuth at
grain boundaries of zinc oxide varistor without intergranular phase. J. Am. Ceram. Soc. 81,
2071–2076 (1998)
19. H.S. Domingos, J.M. Carlsson, P.D. Bristowe, B. Hellsing, The formation of defect complexes
in a ZnO grain boundary. Interface Sci. 12, 227–234 (2004)
20. C. Xu, J. Chun, E. Kim Dong, J.-J. Kim, C B, T. Joo, Electrical properties, near band edge
emission of Bi-doped ZnO nanowires. Appl. Phys. Lett. 90, 083113 (2007)
21. G.A. Baraff, M. Schlüter, Electronic structure, total energies, and abundances of the
elementary point defects in GaAs. Phys. Rev. Lett. 55, 1327–1330 (1985)
22. S.B. Zhang, J.E. Northrup, Chemical potential dependence of defect formation energies in
GaAs: application to Ga self-diffusion. Phys. Rev. Lett. 67, 2339–2342 (1991)
23. C. Freysoldt, B. Grabowski, T. Hickel, J. Neugebauer, G. Kresse, A. Janotti, C.G. Van de
Walle, First-principles calculations for point defects in solids. Rev. Mod. Phys. 86, 253–305
(2014)
24. Ç. Kılıç, A. Zunger, Origins of coexistence of conductivity and transparency in SnO2. Phys.
Rev. Lett. 88, 095501 (2002)
25. Ç. Kılıç, A. Zunger, n-type doping of oxides by hydrogen. Appl. Phys. Lett. 81, 73–75 (2002)
26. Ç. Kılıç, A. Zunger, Doping of chalcopyrites by hydrogen. Appl. Phys. Lett. 81, 2007–2009
(2003)
27. Ç. Kılıç, A. Zunger, n-type doping and passivation of CuInSe2 and CuGaSe2 by hydrogen.
Phys. Rev. B 68, 075201 (2003)
28. P. Hohenberg, W. Kohn, Inhomogeneous electron gas. Phys. Rev. 136, B864–B871 (1964)
29. W. Kohn, L.J. Sham, Self-consistent equations including exchange and correlation effects.
Phys. Rev. 140, A1133–A1138 (1965)
30. W. Kohn, Nobel lecture: electronic structure of matter—wave functions and density
functionals. http://www.nobelprize.org/nobel_prizes/chemistry/laureates/1998/kohn-lecture.
html. Cited 28 Jan 1999
16 Computational Studies of Bismuth-Doped Zinc Oxide Nanowires 421

31. J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made simple. Phys.
Rev. Lett. 77, 3865–3868 (1996)
32. V.I. Anisimov, F. Aryasetiawan, A.I. Lichtenstein, First-principles calculations of the
electronic structure and spectra of strongly correlated systems: the LDA+U method. J. Phys.:
Condens. Matter 9, 767–808 (1997)
33. A.D. Becke, A new mixing of Hartree-Fock and local density-functional theories. J. Chem.
Phys. 98, 1372–1377 (1993)
34. A. Seidl, A. Görling, P. Vogl, J.A. Majewski, M. Levy, Generalized Kohn-Sham schemes and
the band-gap problem. Phys. Rev. B 53, 3764–3774 (1996)
35. M. Aras, Ç. Kılıç, Combined hybrid functional and DFT+U calculations for metal
chalcogenides. J. Chem. Phys. 141, 044106 (2014)
36. J. Heyd, G.E. Scuseria, M. Ernzerhof, Hybrid functionals based on a screened Coulomb
potential. J. Chem. Phys. 118, 8207–8215 (2003)
37. S.L. Dudarev, G.A. Botton, S.Y. Savrasov, C.J. Humphreys, A.P. Sutton, Electron-energy-loss
spectra and the structural stability of nickel oxide: an LSDA+U study. Phys. Rev. B 57, 1505–
1509 (1998)
38. P.E. Blöchl, Projector augmented-wave method. Phys. Rev. B 50, 17953–17979 (1994)
39. G. Kresse, D. Joubert, From ultrasoft pseudopotentials to the projector augmented-wave
method. Phys. Rev. B 59, 1758–1775 (1999)
40. G. Kresse, F. Furthmüller, Efficient iterative schemes for ab initio total-energy calculations
using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 (1996)
41. H.J. Monkhorst, J.D. Pack, Special points for brillouin-zone integrations. Phys. Rev. B 13,
5188–5192 (1976)
42. Ç. Kılıç, Size- and shape-dependent energetics of transition-metal nanocrystals. Solid State
Commun. 150, 2333–2336 (2010)
43. Ç. Kılıç, Energy-distance relation for fcc transition metal nanocrystals. EPL 93, 26004 (2011)
44. M.İ.T. Balta, Ç. Kılıç, Nanocrystal energetics via quantum similarity measures. Modelling
Simul. Mater. Sci. Eng. 22, 025009 (2014)
45. A. Alkauskas, A. Pasquarello, Band-edge problem in the theoretical determination of defect
energy levels: The O vacancy in ZnO as a benchmark case. Phys. Rev. B 84, 125206 (2011)
46. Z. Fan, D. Wang, P.-C. Chang, W.-Y. Tseng, J.G. Lu, ZnO nanowire field-effect transistor,
oxygen sensing property. Appl. Phys. Lett. 85, 5923–5925 (2004)
Chapter 17
Mixed-Phase TiO2 Nanomaterials
as Efficient Photocatalysts

Juying Lei, Hong Li, Jinlong Zhang and Masakazu Anpo

Abstract TiO2, as one of the most promising photocatalysts, exists different phases
such as anatase, rutile and brookite. These phases exhibit different properties and
consequently different photocatalytic performances. In addition, mixed-phase TiO2
have been demonstrated to have enhanced photocatalytic activity relative to
pure-phase TiO2. In the past two decades, many research works have been done on
the synthesis of different kinds of mixed-phase TiO2 and their applications to
photocatalysis. In this review, we firstly give an introduction of three main types of
TiO2 phases as mentioned above, including their structural properties, stability,
phase transformation and photocatalytic activity. And then we pay more attention
on the synthesis of the mixed-phase TiO2. Six preparation methods are introduced
in details, which are hydrothermal method, solvothermal method, microemulsion-
mediated solvothermal method, sol-gel method, solvent mixing and calcination
method and high-temperature calcination method. Following this, three kinds of
applications of the mixed-phase TiO2 in the photocatalysis field are comprehen-
sively highlighted, including photocatalytic production of hydrogen, reduction of
CO2 and degradation of organic pollutants. As the photocatalytic activity of the
mixed-phase TiO2 is usually higher than the single phase TiO2, the mechanism for
the enhancing effects of the mixed phases are discussed. Finally, the existing
problems of mixed-phase TiO2 are summarized and the application prospects of this
kind of nanomaterials are outlooked.

H. Li
J. Zhang
Key Lab for Advanced Materials and Institute of Fine Chemicals,
East China University of Science and Technology, 130 Meilong Road,
Shanghai 200237, People’s Republic of China
e-mail: [email protected]
J. Lei
State Environmental Protection Key Laboratory of Environmental Risk Assessment
and Control on Chemical Process, East China University of Science and Technology,
130 Meilong Road, Shanghai 200237, People’s Republic of China
M. Anpo (&)
Department of Applied Chemistry, Graduate School of Engineering,
Osaka Prefecture University, Naka Ku, 1-1 Gakuencho, Sakai 599-8531, Osaka, Japan
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 423

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_17
424 J. Lei et al.

17.1 Introduction

In recent years, there has been great concern over the many serious environmental
problems and lack of natural clean energy resources that we face on a global scale.
The increase in world population and industrial growth have all led to accelerated
energy consumption and the unabated release of toxic agents and industrial wastes
into the air and waterways, leading to pollution-related diseases, global warming
and abnormal climatic changes. Thus, environmentally harmonious, clean and safe
scientific technologies to address energy as well as pollution and climatic change
are major challenges facing most scientists nowadays. Our scientists hope to con-
tribute to the development ecologically clean, safe, and sustainable chemical
technologies, materials and processes.
Although there are many approaches to challenge these issues, ever since Honda
and Fujishima [1] found the UV light-induced cleavage of water into H2 and O2
using a TiO2 photo-electrode (photosensitizing effect of TiO2 electrode), there has
been enormous interest in the use of TiO2 and other extended oxide and chalco-
genide semiconductors. The purpose is environmental remediation, where organic
toxic materials at low concentrations are photocatalytically converted to harmless
oxidation products [2–29]. TiO2 as a photocatalyst has many attractive features; a
convenient band gap between its valence band and conduction band of 3.2 eV (ca.
400 nm), high stability, low cost, non-toxicity and good performance in the
oxidation of organic pollutants to CO2 and H2O. For these reasons, TiO2-based
photocatalysts have received much attention for various applications in the fields of
the generation of clean energy and protection of our environment.
As mentioned above, with its various advantages, TiO2 has attracted many
researchers to perform systematic in-depth studies of it, promoting the application
process in various areas related to energy and environment. The related research
include studies on photocatalytic mechanism [3–5], regulation of crystal structure
and morphology [6–15], improvement of photocatalysis and photoelectric con-
version efficiency [16–24], expansion of the optical response range [25–29].
However, traditional TiO2 has defects in the catalytic performance, which are
mainly caused by the following two aspects: on one hand, the large band gap limits
its adsorption in the UV region (λ ≤ 387 nm), and the absorption of visible light is
almost zero, so that the utilization of the sunlight is low efficiency [3–5]; on the
other hand, the recombination of photo-generated electrons and holes is facile and
serious, which greatly limits the photocatalytic performance of TiO2 [30, 31].
Researches have demonstrated that, modification of TiO2 by organic dye
photosensitization [32–36], noble metal deposition [37–41] or doping [42–45]
and semiconductor compounding [45–49] can effectively improve the photon uti-
lization and reduce the recombination rate of photo-generated electrons with holes,
thereby improving its photocatalytic efficiency. However, the above methods have
disadvantages such as poor reaction stability and controllability, high cost, com-
plicated operation process amongst others. In these respects, mixed-phase TiO2
with relatively simple preparation process, low cost and adjustable crystal type has
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 425

attracted much attention. In the mixed-phase photocatalyst, the photo-generated

electrons and holes can effectively separate in the two-phase interface, thus reducing
their recombination rate. Moreover, the rutile in the mixed phases have a narrower
band gap, which can enhance the utilization of visible light to some extent. These
advantages can improve the comprehensive properties of TiO2 photocatalyst.
Research results have shown that, adding rutile TiO2 into pure anatase TiO2 to form
mixed-phase TiO2 of rutile and anatase, can significantly improve the photocatalytic
activity of TiO2 [50–53]. After these studies, other kinds of mixed-phase TiO2 have
also been studied in depth, such as anatase/brookite [54, 55], anatase/brookite/rutile
[56] and brookite/TiO2(B) [57]. The photocatalytic experimental results of these
studies showed that, the photocatalytic activity of mixed-phase TiO2 is higher than
that of single-phase TiO2. This is because different phases of TiO2 have different
band positions; their combination can effectively promote the separation of
photo-generated electron-hole pairs. In addition, catalytic “hot spots” are supposed
to be created at the interface of different phases, therefore photocatalysis efficiency
can be improved [51, 56].
In this review, first, we would like to briefly introduce three main kinds of TiO2
phases, and then we will focus on recent advances in the development of mixed-
phase TiO2 photocatalysts, including the synthetic methods for the mixed-phase
catalysts, the applications of the catalysts to various photocatalytic reaction systems
such as photocatalytic hydrogen production, photoreduction of CO2 and photo-
catalytic degradation of organic pollutants. Following this the mechanism of
enhanced photocatalytic activity of the mixed-phase TiO2 will be presented.
Finally, we will summarize the existing problems and overview the application
prospects of the mixed-phase TiO2.

17.2 Phases of TiO2

TiO2 has eight types of crystal phases, which are rutile, anatase, brookite, TiO2-B,
TiO2-R, TiO2-H, TiO2-II, and TiO2-III [58]. Among them, rutile, anatase and
brookite are naturally occurring oxides of titanium at atmospheric pressure and have
been researched mostly for applications [59, 60]. The other five phases have also
been investigated [61–64], however, because they are high-pressure phases, they
are of minor significance for research and development applications. Therefore, we
focus on only the phases of rutile, anatase and brookite in this review.

17.2.1 Structure Properties of Rutile, Anatase and Brookite

The three kinds of TiO2 phases have different crystallographic properties, which are
summarized in Table 17.1 [65].
426 J. Lei et al.

Table 17.1 Crystallographic property of anatase, rutile, and brookite TiO2

Crystal Density System Space group Cell parameters (nm)
structure (g/cm3) a b c
Rutile 4.240 Tetragonal D14
4b-P42/mnm 0.4584 – 0.2953
Anatase 3.830 Tetragonal D19
4a -I41/amd 0.3758 – 0.9514
Brookite 4.170 Rhombohedral D15
2h-Pbca 0.9166 0.5436 0.5135

Fig. 17.1 Crystalline structure of a anatase, b brookite and c rutile [68]

In their structures, the basic building block consists of a titanium atom sur-
rounded by six oxygen atoms forming a TiO6 octahedron in a more or less distorted
configuration. In all the three TiO2 phase structures, the stacking of the TiO6
octahedra results in threefold coordinated oxygen atoms. The fundamental struc-
tural units in these three TiO2 crystals form from TiO6 octahedron units and have
different lattice configurations as presented in Fig. 17.1. In rutile, TiO6 octahedra
were linked by sharing an edge along the c-axis to form chains. These chains are
then interlinked by sharing corner oxygen atoms to form a three-dimensional lattice.
Conversely in anatase, the three-dimensional lattice is formed only by edge shared
bonding among TiO6 octahedra. It means that the octahedra in anatase share four
edges and are arranged in zigzag chains. In brookite, the octahedra share both edges
and corners, forming an orthorhombic structure [66, 67].
In determining these crystal structures and estimating the crystal grain size of
anatase, rutile and brookite, the X-ray diffraction (XRD) experimental method is
usually used. Anatase peaks in X-ray diffraction occur at θ = 12.65°, 18.9°, and
24.054°, the rutile peaks are found at θ = 13.75°, 18.1°, and 27.2° while brookite
peaks are encountered at θ = 12.65°, 12.85°, 15.4°, and 18.1°; θ represents the
X-ray diffraction angle [69, 70].

17.2.2 Stability and Phase Transformation

Among the three types of phases, rutile is the most stable phase, which won’t
decompose or undergo a phase transformation even at very high temperatures.
However, anatase and brookite are metastable; they can be transformed into
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 427

thermodynamically stable rutile when calcined to a certain temperature. There has

been extensive research into the phase transfer mechanism of TiO2 during heating.
Shannon [71] described the anatase to rutile transformation from crystallography to
be a nucleation and growth process, in which rutile phase firstly nucleates on the
surface of anatase and then expands to the bulk. Due to the great differences
between anatase and rutile, the transformation is reconstructive, which means that
the transformation process involves the breaking and reforming of bonds [72]. In
the course of the transformation to rutile, the {112} planes in anatase are retained as
the {100} planes in the rutile product. In these planes, Ti and O atoms synergis-
tically rearrange. Most of the Ti atoms move to a new location to form rutile by the
breaking of two Ti-O bonds in the TiO6 octahedron. Therefore, the formation of
oxygen vacancies can accelerate the phase transformation and the formation of Ti
interstitials inhibits the transformation. This anatase to rutile transformation is a
non-equilibrium phase transformation, which occurs in a certain range of temper-
ature (400–1000 °C); thus temperature is closely related to impurities, particle size,
surface area etc. Impurities and processing atmosphere strongly influence the phase
transition temperature and rate because they can result in different defect structure.
In general, impurities such as the oxides of Ce, Li, K, Na, Fe, and Mn which can
increase the oxygen vacancies usually promote the phase transformation, while
impurities like S, P, and W usually inhibit the phase transformation. A reductive
atmosphere such as H2 and Cl2 accelerate the transformation and gases which are
conducive to the formation of Ti interstitials slow the phase transformation.
Artificial preparation favors the synthesis of anatase nanoparticles, especially the
preparation of TiO2 in aqueous solution [73]; this is because the energy of the three
types of phases are quite close. When the nanoparticles are small enough (<13 nm),
the small surface free energy will become the determinant for the phase transfor-
mation [74]. For TiO2 nanocrystals smaller than 11 nm, anatase is the most stable
phase. When the nanocrystals are bigger than 35 nm, the rutile phase reflects the
thermodynamic stability. And the stability of brookite lies between the anatase and
rutile. In fact, brookite as a metastable phase, most of its physical parameters are
between those of anatase and rutile, such as the band width of anatase, brookite and
rutile are 3.19, 3.11 and 3.0 eV respectively [75].

17.2.3 Photocatalytic Activity of Rutile, Anatase

and Brookite

Generally, anatase exhibits the highest photocatalytic activity among the three types
of phases, which are due to the following reasons:
(1) The band gap of rutile, anatase and brookite are 3.19, 3.11 and 3.0 eV
respectively [75], from which we can see that anatase has the widest band gap.
Therefore, the electron-hole pair has more positive or more negative potential,
leading to higher oxidation ability [76];
428 J. Lei et al.

(2) The surface of anatase has a strong adsorption ability for H2O, O2 and OH,
which is favorable for high photocatalytic activity because the adsorption
capacity of the surface has a great influence on the catalytic activity in the
photocatalytic reaction and strong adsorption capacity is beneficial to high
activity;
(3) Compared to rutile and brookite, anatase usually has smaller grain size and
larger specific surface area in the process of crystallization, which can improve
the photocatalytic activity.
However, because the photocatalytic activity is also greatly influenced by the
crystallization process, the above rules are not suitable for all situations. Under the
same conditions, when the amorphous TiO2 crystallize, rutile usually forms large
grains with poor adsorption ability and thus has low catalytic activity. But if the
rutile can have the same grain size and surface properties as anatase, it can also have
high photocatalytic activity. For example, Lee et al. [77] found that anatase TiO2
can transfer to rutile TiO2 by laser exposure. During the transformation, the specific
surface area and grain size remained unchanged. This resultant rutile photocatalyst
exhibited considerable high catalytic activity. Tsai et al. [78] prepared anatase and
rutile TiO2 by different methods and investigated their photocatalytic activity
toward degradation of phenol. They found the catalytic activity was related to the
preparation methods and calcinations temperature. Under certain conditions, rutile
TiO2 exhibited very high catalytic activity. Therefore, whether anatase or rutile,
they are both likely to have high photocatalytic activity which greatly depends on
the surface properties and the grain size. In addition, Ohno et al. [79] reported that
the photocatalytic activity of different phases of TiO2 is related to the electron
acceptors in the system. When the electron acceptor is O2, the photocatalytic
activity of anatase is high while that of rutile is relatively low. And when Fe3+ is the
electron acceptor, rutile showed higher catalytic activity. This is because O2 as an
electron acceptor is quite sensitive to properties of the catalyst materials in the
photocatalytic reactions. The surface structure of rutile and/or its low energy band
may make it have low transfer efficiency of electrons to O2 and thus rutile has low
catalytic activity when O2 is the electron acceptor. And because most researchers
use O2 for the studies of photocatalytic reactions, rutile usually exhibits low cat-
alytic activity.
The photocatlytic activity of brookite is quite controversial; the main reasons
may be that there have been few studies and experiments on brookite and it is
difficult to prepare pure brookite TiO2. The general products are mixed-phase TiO2,
including anatase/brookite [80], brookite/rutile [81] or anatase/brookite/rutile [44].
Recent studies have demonstrated that, the mixed phase composited TiO2
crystals in an appropriate composition ratio have higher photocatalytic activity than
that of single phase TiO2 crystal. Bacsa et al. [82] found that 100 % anatase or
100 % rutile has relatively lower photocatalytic activity, but a mixture of anatase
and rutile in various ratios achieved better catalytic activity than pure anatase or
pure rutile; the mixture with 30 % rutile and 70 % anatase showed the best catalytic
activity. Thus the two kinds of crystals have a certain synergistic effect for
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 429

enhanced photocatalysis. And the commercial photocatlyst P25, which has con-
siderable high photocatalytic activity, is also a kind of mix-phase TiO2 rather than
pure anatase. Therefore, great efforts have been devoted to investigating the syn-
thesis and photocatalysis applications of mixed-phase TiO2.

17.3 Synthesis of Mixed-Phase TiO2 Photocatalysts

It can be observed from the structure and properties of the three types of crystals,
that adjusting the experimental parameters of the synthesis can cause conversion
among different phase types allowing different mixed-phase TiO2 photocatalysts to
be prepared. At present, there is a variety of methods to prepare mixed-phase TiO2,
such as pulsed laser deposition (PLD), hydrolysis method, hydrothermal method,
solvothermal method, micromulsion-mediated solvothermal method, solvent mix-
ing and calcination method (SMC method), high-temperature calcination method,
and high temperature gas phase decomposition method as well as others [83]. In the
preparation process of mixed-phase TiO2, with different synthesis methods, the
important influence factors for the ratio of mixed phase crystals are different. In all
kinds of synthesis methods, the mixed phase ratio, morphology and surface prop-
erties can be controlled by optimizing the experimental conditions, such as tem-
perature, pressure, concentration, and types of reagents. The ratio of different
crystals in a perfect mixed-phase photocatalyst should be able to be tuned according
to the actual needs, and the product should have controllable shape, uniform dis-
persion, and should be uneasy to agglomerate. The preparation conditions and
influence factors will be discussed in the following sub-section, combining concrete
preparation methods.

17.3.1 Hydrothermal Method and Solvothermal Method

The hydrothermal method is used most commonly in the laboratory preparation of

TiO2 nanomaterials [84–91]. It refers to a method of synthesizing materials in a
sealed reaction vessel (such as an autoclave), using aqueous solution as the reaction
system, by heating or other means, to create an environment of high temperature
and high pressure. In this condition, the vapor pressure of water becomes higher
while the density and surface tension decrease and the ion product increases. This
leads to dissolution and recrystallization of the substances which are insoluble in
water at room temperature. This method belongs to the liquid phase chemical
synthesis method. The main factors influencing the properties of the product by the
hydrothermal method include vessel volume, reaction temperature, reaction time,
and heating rate.
430 J. Lei et al.

Chen and Sum et al. [92] synthesized highly crystalline pure brookite and
two-phase anatase/brookite TiO2 nanostructures via a simple hydrothermal method
with TiS2 as the precursors in NaOH solutions. The control of the phase compo-
sition has been achieved by varying the solution concentration and reaction time. In
the reaction system, the TiS2 underwent a hydrolysis reaction as follows:

TiS2 þ 4 NaOH ! Ti4 þ þ 4 OH þ 2 Na2 S ð17:1Þ

Ti4 þ þ xðOHÞ ! TiðOHÞ4x

x ð17:2Þ

At the beginning of the reaction, the hydrolysis of TiS2 formed Ti4+ (17.1). Then
in the hydrothermal process, Ti4+ reacted with OH− in the solution (17.2). When the
concentration of OH− is low, the hydrolysis rate of Ti4+ is slow and the Ti4+
connect with OH− to form anatase TiO2; when the OH− concentration is high, the
hydrothesis rate of Ti4+ is fast leading to the formation of mixed-phase TiO2 of
brookite and anatase. The mass fraction of brookite (B) and anatase (A) can be
obtained respectively by calculating with the following formula [93]:

WB ¼ KB IB =ðKA IA þ KB IB Þ ð17:3Þ

WA ¼ KA IA =ðKA IA þ KB IB Þ ð17:4Þ

In the formula, IA and IB represent the integral intensity of the strongest

diffraction peaks in anatase and brookite respectively, and KA = 0.886, KB = 2.721.
Figure 17.2a, b show the influences of different NaOH concentrations and different
reaction times on TiO2 crystal. From the graphs, it is observed that with increase of
NaOH concentration, the content of brookite increases, and when the reaction time
is more than 6 h, the crystalline form of the product did not change significantly. In
addition, brookite can be obtained by conversion of sodium titanate. Since Na+ is
able to stabilize the layered structure, when there is no Na+ in the reaction system, a
layered structure is not stable, and the hydrothermal process causes the structure to
collapse, forming anatase TiO2; when Na+ is excessive, the layered structure is very
stable and does not easily collapse, ultimately forming titanate; when Na+ is
moderate, the layered structure partially collapses, and part of the layered structure
remains unstable, forming brookite (Fig. 17.2c). The above results are consistent
with the literature [88].
Zhang et al. [94] introduced NaCl and NH4OH in the same time into the
hydrothermal reaction system, using TiCl3 as a titanium source. The results showed
that the phase composition of the products can be controlled by changing the
concentration of NaCl and the volume ratio of NH4OH to H2O in the reaction
system (Fig. 17.3a). They also believe that Na+ is able to stabilize the layered
titanate structure, which was formed by the hydrolysis of TiCl3 in aqueous
ammonia in their reaction system. In the absence of NaCl, NH4+ ions exist in the
interlayer of titanate to balance the negative charges of the layered titanate.
However, under hydrothermal treatment at high temperature up to 200 °C, NH4+ is
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 431

Fig. 17.2 The composition of anatase and brookite in the mixed phase TiO2 synthesized with
a various NaOH concentrations and b reaction times under hydrothermal reaction [92]; The
scheme of formation of anatase and brookite (c) [92]

hydrolyzed to form NH4OH and is released from the interlayer. The H+ ions pro-
duced from the hydrolysis of NH4+ ions then induce a hydroxyl condensation
reaction to form anatase TiO2. In the presence of NaCl, Na+ can help to stabilize the
layered structure with NH4+. Na+ and NH4+ exist simultaneously in the interlayer of
titanate to balance the negative charges of titanate. When the layered titanate was
treated hydrothermally continuously at a high temperature for several hours, the
increased hydrolysis of NH4+ led to the collapse of layered structures. However,
some bonds in the structure are maintained by Na+ ions. Therefore, the structural
transformation is delayed which results in the formation of a brookite-like structure
[95–97]. The as-formed brookite-like structure expands to form a brookite lattice.
Increasing the concentration of NaCl will enhance the powder of the brookite lattice
to compete with the anatase lattice. Therefore, the brookite content in the
mixed-phase TiO2 increases with an increase in concentration of NaCl. In addition
432 J. Lei et al.

Fig. 17.3 a Relationship between the contents of brookite in products and the applied
concentration of NaCl [94]; b Relationship between the contents of brookite and the applied molar
ratio of tartaric acid to TiCl3 [98]

to this research, Zhang et al. [98] developed another facial hydrothermal method for
the preparation of anatase/brookite TiO2 using tartaric acid (C4H6O6) as the phase
content regulator. Different amounts of C4H6O6 were added to the hydrothermal
reaction system using TiCl3 as the titanium source and NaOH for adjusting the pH
of the reaction solution. The contents of anatase and brookite in the TiO2 particles
were successfully controlled by simply adjusting the molar ratio of C4H6O6 to TiCl3
(Fig. 17.3b). The mechanism of the phase evolution between brookite and anatase
was explained according to the ligand field theory [99]. They suggested that
C4H6O6 can chelate with Ti to form a stable titanium complex. When the C4H6O6/
TiCl3 molar ratio is less than 0.75 in the reaction system, there are two forms of Ti
species. One is the insoluble Ti-contained species Ti(OH)4; the other is the soluble
Ti-contained complex [Ti(OH)x(C4H6O6)y]z-. The amorphous Ti(OH)4 could first
transform to layered titanate under hydrothermal treatment, which would be
transformed to brookite under a given concentration of Na+ and OH− [95]. Due to
the large steric hindrance of carboxylic acidic ligands, the [Ti(OH)x(C4H6O6)y]z-
complexes should combine together by sharing equatorial or apical edges and being
arranged in zigzag chains, which benefited the formation of anatase crystallites, thus
resulting in an anatase/brookite mixed-phase TiO2. They also pointed out that the
brookite as the pure phase or the main phase was not obtained when the pH value of
the system was less than 9 or the hydrothermal temperature was lower than 180 °C.
That is to say, it is difficult to control the contents of anatase and brookite in
samples by changing the C4H6O6/TiCl3 molar ratio while pH < 9 or temperature
<180 °C in the hydrothermal reaction system.
A variation of the hydrothermal method is the solvothermal method wherein
organic solvents such as ethanol, glycol and toluene are used instead of an aqueous
environment. Similar to the hydrothermal method, in the solvothermal method the
crystal types and morphology of the TiO2 nanomaterials can both be controlled by
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 433

regulating parameters such as temperature, pressure inside the system, the reaction
time, and the titanium source.
Li et al. [100] synthesized anatase/rutile mixed-phase TiO2 with different rutile
content using the solvothermal method. They obtained mixed-phase TiO2 crystals
by hydrolysis of tetraisopropyltitanate in an acid alcoholic solution and studied the
effect of hydrochloric acid on the rutile content in the mixed-phase crystal. The
results showed that low H2O/Ti mole ratio favors the formation of anatase/rutile
mixed-phase TiO2, otherwise it will form brookite/anatase mixed-phase TiO2. Lei
et al. [101] prepared anatase/rutile mixed-phase TiO2 crystal in a low temperature
(80 °C) solvothermal reaction by pre-oxidizing TiCl3 into Ti4+ with HNO3 followed
by diluting with urea, water and ethanol. The anatase content in the mixed-phase
crystal increases with the increase of ethanol content in the solution (Fig. 17.4). The
Scherrer formula is used to calculate the average particle size of anatase and rutile
in the mixed-phase crystal; they were both below 10 nm.
The mixed-phase TiO2 nanomaterials prepared by hydrothermal and solvother-
mal methods are usually well crystallized and don’t need to be calcinated under
high temperature. The size and phase type of the product can be regulated by simply
changing the experimental parameters such as the type of acid or base in the
reaction system, the reaction temperature, the reaction time, the autoclave pressure
and so on. Besides these advantages, they also have some drawbacks including the
requirement for the equipment to withstand high temperature and high pressure, and
experimental safety concerns. Moreover, because the hydrothermal and solvother-
mal process can’t be monitored in real time and usually proceeds without stirring,
sometimes the reaction is incomplete and it is difficult to obtain uniform product.
Since a variety of organic solvents can be used in the solvothermal method, it is
easier to control the morphology and crystalline of the products compared with the
hydrothermal method. Thus, the solvothermal method is considered to have better
application prospects.

Fig. 17.4 Relationship

between the contents of
anatase and the volume ratio
of ethanol in a solvothermal
reaction [101]
434 J. Lei et al.

17.3.2 Microemulsion-mediated Solvothermal Method

The method of microemulsion refers to the process that, firstly, two immiscible
solvents form emulsion in the role of surfactants, and then nanomaterials are
obtained in the emulsion through nucleation, coalescence, agglomeration, and heat
treatment in the oil-in-water or water-in-oil micro-bubbles. In this process, the two
immiscible solvents are divided into many micro-reactors (micro-bubbles) by sur-
factant, in which the reactants are well dispersed and a uniform nucleation occurs. In
addition, the growth of the nucleus can be limited in the micro-reactors. Therefore,
the particle size and stability of the nanomaterials can be controlled precisely.
Compared with other traditional preparation methods, the microemulsion method
has been found to have obvious advantages in terms of the synthesis of nanoma-
terials with superior monodispersity and interfacial properties. Moreover, this
method is a very versatile technique which allows the preparation of a great variety
of nanomaterials in combination with other techniques. Recently, the combination of
the microemulsion method with solvothermal method, namely the microemulsion-
mediated solvothermal method, has been used to prepare the mixed-phase TiO2
crystals [66, 102, 103].
Yan et al. [66] prepared anatase/rutile mixed-phase TiO2 crystal through the
combination of microemulsion and solvothermal method. In the reaction system,
tert-octylphenoxypolyethoxyethanol (Triton X-100) as the surfactant and n-hexanol
as the co-surfactant were mixed with cyclohexane to serve as the continuous oil
phase. Tetrabutyl titanate and (NH4)2SO4 were dissolved in the hydrochloric acid as
the aqueous phase. The aqueous phase was added dropwise to the oil phase, forming
a clear microemulsion, which was then solvothermally treated below 120 °C for 13 h
giving the mixed-phase TiO2 product. The resultant product has surface area of
about 86–169 m2/g and grain sizes of anatase and rutile of about 15 nm and 10 nm
respectively. Their results showed that the change of SO42− concentration impacts
the content of anatase in the mixed-phase crystal (Fig. 17.5). With the increase of
SO42− concentration, the anatase content in the mixed-phase TiO2 crystal increased.
Different polymorphs can be obtained by affecting the polycondensation of TiO6
octahedra with SO42− (Fig. 17.6a). The SO42− interacts with the hydroxy groups on
the surface of TiO6 octahedra to form hydrogen bonds during the hydrothermal
reaction process (Fig. 17.6b). Because of the steric effect of SO42−, the octahedron
with SO42− and another octahedron would polycondense along the converse
direction in order to decrease the repulsion (Fig. 17.6c), and the orientation of the
third octahedron is more conducive to the formation of an anatase nucleus
(Fig. 17.6d). When the concentration of SO42− stays at a lower level, a rutile
structure can be easily formed in order to ensure the lowest free energy in the system.
Therefore, controlled adjustment of anatase content in mixed TiO2 crystal can be
achieved by adjusting the concentration of SO42−. In the same oil phase system,
Zhang et al. [103] prepared mixed-phase TiO2 crystal with different anatase and
rutile content by the microemulsion-mediated solvothermal method, using a mixture
of TiCl3, urea, and hydrochloric acid solution as the aqueous phase. The results
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 435

Fig. 17.5 Relationship

between the contents of
anatase and the mole
percentage of SO42− under the
micromulsion-mediated
hydrothermal method [66]

Fig. 17.6 Proposed

mechanism: a the orientation
of the third octahedron
determines whether a rutile or
an anatase nucleus is formed;
b interaction between SO42−
and TiO62− octahedral
hydroxyls; c two TiO62−
octahedra share edge in the
presence of SO42−;
d formation of anatase in the
presence of SO42− [66]

showed that with the increase of urea concentration, anatase content in the
mixed-phase TiO2 increased. The hydrochloric acid in the reaction solution sup-
pressed the generation of anatase and promoted the formation of rutile phase.
Different from SO42−, the Cl− has small radius and weak steric hindrance. TiO6
octahedra polycondensed in chain to form a rutile structure according to the
436 J. Lei et al.

minimum energy principle. Therefore, high concentration of Cl− favors the forma-
tion of rutile TiO2 and a high concentration of SO42− favors the formation of anatase
TiO2.
Although the experiment process is relatively complicated in the preparation of
mixed-phase TiO2 by the microemulsion-mediated solvothermal method, the grain
size of the product is small and the phases can be controlled, which gives this
method good prospects for development.

17.3.3 Sol-Gel Method

Sol-gel method is one of the commonly used methods for preparation of TiO2
nanomaterials. The sol-gel method has advantages such as processing at
low-temperature and the ability to prepare materials in various shapes, thus, it is one
of the most promising techniques to prepare amorphous and crystalline materials.
Usually the titanium precursor used in the sol-gel reaction system is titanium
alkoxides or titanium halides. Firstly, a sol is prepared by hydrolysis of titanium
precursor. After aging the sol for a certain time, a three-dimensional cross-linked
gel is obtained; amorphous white powder is obtained after grinding the gel. Finally,
crystallized TiO2 is produced by high temperature calcination.
Scotti et al. [52] synthesized anatase/rutile mixed-phase TiO2 crystal by a sol-gel
method and investigated the influence of HCl/Ti and H2O/Ti molar ratio on the phase
type of the product. The experimental process was as follows: TiCl4 and tri-block
copolymer were dissolved in ethanol then water and HCl were added to adjust the pH
of the solution, obtaining a sol. A gel is formed after aging the sol for 3–13 days. After
drying and calcination, TiO2 crystals with different anatase and rutile content were
obtained. In the products, pure rutile phase (Fig. 17.7a) showed chestnut burr
aggregates of radially growing elongated nanocrystals with average sizes of 10–20 nm
in width and 100–200 nm in length; pure anatase phase (Fig. 17.7b) showed aggre-
gates of almost square-ended nanoparticles with average sizes of 5–15 nm. The
mixed-phase sample (Fig. 17.7c) was observed to have two phases with the small
anatase particles surrounding the chestnut burr aggregates of rutile. The phase content
of the products can be controlled by systematically changing the H2O:Ti (rw) and
HCl:Ti (ra) molar ratios in the reaction system. Results have shown that, in the acidic
titanium halide or titanium alkoxide precursor solution, with the increase of H2O
content, concentration of the Ti4+ concentration decreased, causing the condensation
reaction rate of TiO6 octahedra to decrease easily forming rutile TiO2 which has better
thermodynamically stability (Fig. 17.7d, e).
Mixed-phase TiO2 crystals obtained by the sol-gel method are small particles of
high purity, but this usually requires long time aging and the particles are easy to
agglomerate after high temperature calcination, which may have a significant
impact on the photocatalytic activity of TiO2.
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 437

Fig. 17.7 TEM images of a pure rutile, b pure anatase, and c mixed anatase 48 wt% and rutile
52 wt% sample; Plots of rutile content (wt%) in TiO2 samples: d vs. H2O:Ti molar ratio, e vs.
titanium concentration [52]

17.3.4 Solvent Mixing and Calcination Method

The solvent mixing and calcination method (SMC method) refers to mixing TiO2
with different phases in some solvent and then evaporating the solvent completely
followed by calcinations at high temperature to make a close contact among dif-
ferent crystal phases. Zachariah et al. [104] synthesized anatase/rutile mixed-phase
TiO2 by the SMC method using anhydrous isopropanol as the solvent. The crys-
talline size of pure anatase is 10 nm as calculated from the Scherrer equation, and
the grain size of pure rutile is 40 nm. In the resultant mixed-phase TiO2 containing
40 % rutile, the size of anatase is 29 nm, and size of rutile is 47 nm. From
Fig. 17.8a, b, pure anatase can be observed consisting of small particles, and the
overall particle size is 100–150 nm. The particle size of pure rutile is 300–500 nm
(Fig. 17.8c, d). The morphology of the mixed-phase TiO2 is a mixture of anatase
and rutile particles (Fig. 17.8c, f). The phase interface can be seen from the high
resolution transmission electron microscopy (HRTEM), indicating there is a close
interaction between the two phases. Bojinova et al. [105] prepared an anatase/rutile
mixed-phase crystal TiO2 by the SMC method using ethanol as the solvent. In the
mixed-phase TiO2, the grain size of anatase is 42 nm, and that of rutile is 56 nm, as
calculated from the Scherrer formula,. Liu et al. [106] made the anatase small
particles adsorbed onto rutile nanorods through a layer-by-layer self-assembly
technique, using polystyrene sulfonate (PSS) as a medium and finally the PSS was
438 J. Lei et al.

Fig. 17.8 SEM and TEM images of a, b anatse, c, d rutile, and e, f mixed-phase TiO2 with an
optimum rutile content of 40 wt% produced by the SMC method [104]

removed by calcination. The anatase content in the mixed-phase TiO2 can be

controlled by changing the number of loading cycles.
By using the SMC method for the synthesis of mixed-phase TiO2, the phase
content of the product can be regulated easily. However, because of the final
calcination process at high temperature, the TiO2 is easy to agglomerate. In addi-
tion, it’s usually difficult to achieve uniform mixing of anatase and rutile, leading to
products with many pure anatase aggregates as well as pure rutile aggregates.
Therefore, the photocatalytic performance of the mixed-phase TiO2 prepared by this
method is subject to limitations.

17.3.5 High-Temperature Calcination Method

The high-temperature calcination method is one of the earliest methods used to

synthesize mixed-phase TiO2, which is mostly used to research the phase transition
of TiO2 [84].
Nair et al. [107] obtained mixed-phase TiO2 crystal with different anatase and
rutile contents by the high-temperature calcination method. They firstly prepared
small anatase particles by a sol-gel method and then calcinated the anatase particles
under elevated temperatures. It was found that the samples calcined at low tem-
peratures (T < 600 °C) remained pure anatase phase. The phase transformation
began at 650 °C. When the temperature was raised to 850 °C, the product becomes
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 439

pure rutile. In the mixed phase products, the grain size of anatase was 33.66–
51.48 nm, and that of rutile was 45.2–60.6 nm, as calculated by the Scherrer
formula. As shown in Fig. 17.9, the particle size of anatase was about 100 nm and
rutile was about 200 nm; the mixed-phase particle size was between the above two
and increased with calcination temperature. The phase transformation is a process
of nucleation and growth. However, there is some dispute about the concrete
process. Gouma et al. [108] thought that the rutile firstly formed nuclear on the
surface of anatase, then extended to the bulk. Zhang et al. [109] verified the phase
transformation of anatase-rutile occurred in the bulk firstly, and then spread to the
surface with the increase of calcination temperature by UV Raman spectroscopy.
The phase’s transformation diagram is shown in Fig. 17.10.
The high-temperature calcination method is very simple and can be used to
prepare mixed-phase TiO2 with perfect polymorphs and tunable phase content.
However, the hard agglomeration of TiO2 is serious, and the particles size is large,
resulting in limitations of the photocatalytic activity.
Overall, both in the SMC method and the high-temperature calcination method,
a high temperature is required to obtain the mixed-phase TiO2 or close contact
between different crystalline forms. As the electron micrographs and XRD results
showed, the grain size of crystal after calcination is significantly larger than that of
mixed-phase TiO2 crystals prepared by the in situ method, and the agglomeration is

Fig. 17.9 SEM and EDAX images of titania catalysts calcined at different temperature: a 600 °C,
b 700 °C, c 750 °C, d 800 °C [107]
440 J. Lei et al.

Fig. 17.10 Proposed schemes for the phase transformation of TiO2 with increasing calcination
temperature [109]

Table 17.2 Rate constants Synthesis method Rate References

for the photocatalytic constants/min−1
degradation experiment with
mixed-phase TiO2 Hydrothermal method 0.019 [53]
synthesized by different Sol-gel method 2.400 [51]
methods Microemulsion-mediated 0.030 [75]
Solvothermal method
SMC method 0.023 [77]
Calcination method 0.003 [79]

also serious, all of which can easily cause decrease of photocatalytic activity.
Therefore, in recent years, these methods are used relatively less.
The photoctalytic activity rate constants of mixed-phase TiO2 prepared by dif-
ferent methods are summarized in Table 17.2. From this table, we can see that the
reaction rate of TiO2 synthesized by in situ methods, such as the sol-gel method and
microemulsion-mediated hydrothermal method is generally higher than the high-
temperature or SMC method. This is because the latter requires high-temperature
calcination, resulting in larger particle size and serious agglomeration that reduce
the contact area with the degradation target, and impact the photocatalytic activity.
However, the reaction rate of mixed-phase TiO2 prepared by Zheng et al. [54] was
small, which may be because high-temperature calcination leads to a hollow
structure, which at the same time, decrease the specific surface area and lower the
activity. Photocatalytic activities of mixed-phase TiO2 prepared under different
conditions are also influenced by many other factors: product morphology, particle
size, crystal proportion.
In addition to the commonly used approaches mentioned above, there are some
new methods to prepare mixed-phase TiO2, for example the microwave heating
method [110]. The particle size of TiO2 synthesized by this method is smaller than
that prepared by heating with an oil bath. With the development of new tech-
nologies, it is believed that there will be many new methods for synthesizing
mixed-phase TiO2.
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 441

17.4 Applications of Mixed-Phase TiO2 in Photocatalysis

17.4.1 Photocatalytic Hydrogen Production

Hydrogen has been widely recognized as an ideal energy source that could contribute
to the replacement of a significant fraction of fossil fuels. However, nowadays the
large amount of H2 demanded is mostly produced from fossil feedstock, mainly by
steam reforming of methane. It is therefore mandatory to develop sustainable
methods for hydrogen production. Among them, one of the most promising tech-
nologies is the photocatalytic hydrogen production, which combines various positive
aspects, such as sustainability of the primary energy source (the solar light), the
renewability of the starting feedstock and the possible production of by-products with
a high added value [111]. And TiO2 is the most investigated photocatalyst for H2
production due to its non-toxicity, relatively cheap cost, and excellent stability under
the reaction conditions. However, presently, the solar-to-hydrogen energy conver-
sion efficiency of TiO2 is too low for the technology to be economically sound. The
main barriers are the rapid recombination of photo-generated electron-hole pairs as
well as backward reaction and the poor activation of TiO2 by visible light. In response
to these deficiencies, many investigators have been conducting research with an
emphasis on effective remediation methods like modification of TiO2 by means of
metal loading, metal ion doping, dye sensitization, composite semiconductor, anion
doping, and metal ion implantation. In addition to these efforts, mixed-phase TiO2
have also been explored for hydrogen production [92, 111–115], since multiphasic
TiO2 materials demonstrated higher performances with respect to the correspondent
monophasic systems in many photocatalytic processes.
Chen et al. [92] successfully synthesized pure anatase nanoparticles, pure
brookite nanoplates, and two-phase anatase/brookite composite by a simple
hydrothermal method. Photocatalytic activity of the as-synthesized samples for
hydrogen production was investigated in methanol solution. Results have shown
that the photocatalytic activity is higher for the two-phase anatase/brookite TiO2
compared to pure brookite nanoplates, pure anatase nanoparticles and their physi-
cally mixed sample, as shown in Fig. 17.11. Moreover, in comparison with the
highly active two-phase commercial P25, the synthesized two-phase anatase/
brookite TiO2 is 220 % more active when measured by the H2 yield per unit area of
the photocatalyst surface. Similar results were obtained by Montini et al. [112].
They also prepared TiO2 nanocomposite with anatase/brookite composition by
hydrothermal treatments and the as-prepared anatase/brookite nanocomposites
showed higher H2 production compared to a reference TiO2 prepared by conven-
tional sol-gel synthesis. Furthermore, they found that the anatase/brookite ratio in
the nanocomposite greatly affects the photocatalytic activity in H2 production.
In addition to the research applying mixed phase anatase/brookite TiO2 as the
photocatalyst for hydrogen production, a great deal of studies have also been done
using the mixed phase anatase/rutile TiO2. Li et al. [113] developed photocatalysts
with a tuned anatase/rutile structure by calcination of P25 at different temperatures and
442 J. Lei et al.

Fig. 17.11 Hydrogen evolved per gram of catalyst per hour under UV−vis irradiation in aqueous
methanol solution over 0.3 wt% Pt loaded photocatalysts. A, B and R denote anatase, brookite and
rutile respectively [92]

investigated their catalytic activity for hydrogen production by photocatalytic bio-

mass reforming. Surprisingly, they found that the photocatalytic activity of the
thermal-treated P25 for hydrogen production can be greatly increased, compared with
P25 without any treatment. The overall photocatalytic activity for hydrogen pro-
duction on thermal-treated P25 can be enhanced up to 3–5 times. It is proposed that the
anatase/rutile junction structure is mainly responsible for the improved photocatalytic
performance. Their work implies that the photocatalytic performance of TiO2 could be
further improved by elaborately designing the anatase/rutile structure. Similar
conclusions were obtained by Amal et al. [114]. They did a systematical study on
photocatalytic H2 evolution over mixed-phase TiO2 as a function of anatase and rutile
phase compositions with methanol as hole scavengers. The TiO2 nanoparticles they
synthesized contain 4–95 mol% anatase, with the remaining being rutile. Synergistic
effects in terms of H2 evolution were observed for a wide range of anatase contents,
from 13 to 79 mol%. No synergistic effect was observed for the physically mixed
anatase and rutile particles due to insufficient physical contact. Besides the above
research on particulate TiO2 (Fig. 17.12a, b), recently Yu et al. [115] synthesized a
kind of anatase/rutile TiO2 nanofiber photocatalyst (Fig. 17.12b, c). The enhanced H2
production performance was also observed in the obtained anatase/rutile composite
nanofibers. The nanofibers with 45 wt% rutile phase and 55 wt% anatase phase,
exhibiting the highest photocatalytic H2 production rate of 324 mmol h−1 and apparent
QE of 20.9 % at 365 nm.
Although the research on the mixed-phase TiO2 of anatase/rutile all demon-
strated that the combination of the two phases can greatly enhance the photocat-
alytic hydrogen production efficiency, the low utilization rate of visible light is still
a big problem which limits its practical application. The relatively narrow band gap
of rutile allows some visible light absorption; however this is not viable for practical
applications. Further efforts have been devoted to expand the visible light
absorption of the mixed-phase TiO2. Keller et al. [116] reported Au modified TiO2
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 443

Fig. 17.12 TEM images of particulate anatase and rutile TiO2; c, d TEM and HRTEM images of
fibrous anatase and rutile TiO2 [113]

anatase/rutile catalyst (Fig. 17.13) for hydrogen production. They investigated the
light absorption of the catalysts with and without Au deposition by UV–Vis light
absorption spectra. Before the Au deposition, the samples with rutile phase whose
content was higher than that of P25 showed light absorption extended up to
550 nm, but with a relatively low intensity. After Au deposition, the catalysts
showed an obvious modification of the absorption properties. This yielded an
additional absorption band around 550 nm, which is attributed to a plasmon res-
onance phenomenon [117] due to collective oscillations of the conduction electrons
located on the 6 s orbital of gold and induced by the incident electromagnetic wave.
In addition to this plasmon absorption, deposition of gold also shifted the
absorption spectrum deeper into the visible light range, in the 380–450 nm regions.
This better light absorption in the visible region was demonstrated to contribute to
better results for the hydrogen production under simulated solar light. Furthermore,
they pointed out that, there are various factors which were crucial to enhance H2
evolution efficiency: (i) the surface, crystallographic, and porosity properties of the
TiO2 anatase/rutile catalyst, (ii) the anatase/rutile ratio, (iii) the nature and content
of the metallic co-catalyst, (iv) the metal-support interactions, and (v) the relative
amount of sacrificial reagent. The influence of these different factors was studied in
detail by them. In optimized conditions, important H2 production efficiency
(120 μmol min−1) was obtained over days without deactivation and with very low
amounts of sacrificial reagent.
444 J. Lei et al.

Fig. 17.13 a, b TEM images of mixed-phase TiO2 photocatalysts with Au deposition; c, d Light
absorption properties of mixed-phase TiO2 photocatalysts without and with Au deposition [116]

17.4.2 Photocatalytic Reduction of CO2 with Water

on Mixed-Phase TiO2

CO2 is well known to be one of the main causes for the greenhouse effect which
leads to global warming. At the same time CO2 is also a promising carbon resource
because it can be converted into various useful chemical compounds and fuels such
as CH4, CH3OH, HCOOH. Therefore, in order to reduce the emissions of CO2 and
to achieve a sustainable energy future, novel materials and new technologies have
been developed to convert CO2. Besides the methods of solar thermo-chemical
conversion and electrochemical reduction of CO2, solar-activated photocatalytic
reduction of CO2 with water by TiO2 at room temperature and atmospheric pressure
is attractive due to its “green chemistry” and relatively low cost.
During the process of CO2 photoreduction with H2O, photo-illumination of the
catalyst surface induces the generation of electron-hole pairs in TiO2. The excited
electrons in the conduction band (CB) of TiO2 could migrate to the surface and
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 445

reduce CO2. Meanwhile, the holes left in the valence band (VB) of TiO2 could
oxidize H2O into O2.
The above photocatalytic process has been demonstrated to be closely related to
the phase type of TiO2. Li et al. [118] investigated the CO2 photoreduction with
water vapor on three TiO2 nanocrystal polymorphs (anatase, rutile, and brookite).
Their experimental results showed that, the photocatalytic reduction activity follows
the order: anatase > brookite > rutile. The rutile was the least active, mainly due to
the fast e− and h+ recombination in rutile. They also investigated the photoreduction
activity of the TiO2 catalysts with a helium treatment. The photoreduction data
indicated that helium treated catalysts were more active than non-pre-treated ones.
And the catalytic activity for production of CO and CH4 from photoreduction of
CO2 has the order: brookite > anatase > rutile (Fig. 17.14), where brookite exhibited
the highest photocatalytic activity. Thus their study also implies that the brookite
phase is a promising material for CO2 reduction, they performed further research on
this topic, including the investigation of brookite-containing mixed phases. They
prepared bicrystalline anatase/brookite TiO2 through a hydrothermal method [119].
The as-prepared bicrystalline TiO2 were also applied for CO2 photoreduction in the
presence of water vapor for production of CO and CH4. The photocatalytic activ-
ities were compared with those of pure anatase, pure brookite, and a commercial
anatase/rutile TiO2 (P25). The results in Fig. 17.15 showed that bicrystalline ana-
tase/brookite was generally more active than pure anatase, brookite, and P25. The
bicrystalline mixture with a composition of 75 % anatase and 25 % brookite
showed the highest photocatalytic activity, which was nearly twice as high as that
of pure anatase (A100) and three times as high as that of single-phase brookite
(B100). The higher activity of bicrystalline anatase/brookite is speculated to be
ascribed to the interactions between the anatase and brookite nanocrystals.

Fig. 17.14 The top figures

are TEM images of different
TiO2 polymorphs: anatase,
brookite and rutile (from left
to right). The bottom figure is
the production of CO and
CH4 on the three different
TiO2 polymorphs [118]
446 J. Lei et al.

Fig. 17.15 The top figures are TEM and HRTEM images of mixed-phase anatase/brookite TiO2
(A75B25). The bottom figure is production of CO on various TiO2 catalysts (A = anatase,
B = brookite, the subscript number are the phase fraction of anatase and brookite) [119]

In addition, the anatase-rich bicrystalline anatase/brookite mixtures are superior to

anatase/rutile mixtures P25, indicating the interaction between anatase and brookite
seems to be more effective than that between anatase and rutile (as in P25) in the
photoreduction of CO2.

17.4.3 Photocatalytic Degradation of Organic Pollutants

on Mixed-Phase TiO2

In 1977, Frank and Bard reported for the first time the photocatalytic decomposition
of cyanide by TiO2 in aqueous medium under sunlight [120]. Since then, the
application of TiO2 for photocatalytic degradation of organic pollutants has gained
wide attention due to its effectiveness in degrading and mineralizing the recalcitrant
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 447

organic compounds as well as the possibility of utilizing the solar UV and

visible-light spectrum. Mixed-phase TiO2 photocatalysts, which usually have better
photocatalytic activity than single-phase TiO2, have also been widely applied for
the degradation of organic pollutants.
Many studies have been carried out to examine the photocatalytic degradation of
organic dyes in the presence of mixed-phase TiO2 as photocatalyst. For instance,
methyl blue (MB) were degraded in the aqueous solution by anatase/rutile TiO2
heterojunction nanoflowers under simulated solar light irradiation [121]. It was
found that 72 % of MB could be degraded in 120 min with the mixed-phase catalyst
prepared in optimized experimental conditions while only 30 % of MB was
degraded in the solution with pure anantase. In addition, the TiO2 nanoflowers
showed excellent stability after 9 cycles under the same conditions. These results
suggested that the mixed phase anatase/rutile TiO2 heterojunction nanoflowers have
great potential for the future photodegradation of real dye waste water. Since high
surface area is conducive to high catalytic activity, mesoporous mixed-phase
anatase/rutile TiO2 were also prepared [122] (Fig. 17.16a) and applied for the

Fig. 17.16 a TEM image of mesoporous mixed-phase anatase/rutile TiO2 (10 %V-Ti-700); b N2
sorption isotherms of 10 %V-Ti-700. The inset figure is the BJH desorption pore-size distributions;
c DR UV−vis spectra for various vanadium doped mixed phase samples; d photocatalytic
decomposition of 100 mL, 10−4 MB dye under visible light in 2 h by using 100 mg of 5 %
V-Ti-700, 10 %V-Ti-700, 5 %V-Ti-800, 10 %V-Ti-800, and P25 samples, respectively [122]
448 J. Lei et al.

photocatalytic degradation of MB. The optimum mesoporous catalyst obtained

(10 %V-Ti-700) possess surface area up to 94 m2 g−1 and an average pore size of
5.6 nm (Fig. 17.16b). In addition, the catalyst was in situ doped with vanadium
which helped to extend the absorption of TiO2 to the visible light region (>400 nm)
(Fig. 17.16c). As a result, the high surface area, the mixed-phase effect and the
vanadium doping qualified the mesoporous mixed-phase anatase/rutile TiO2 high
photocatalytic activity under visible light which are superior to the commercial P25
(Fig. 17.16d).
In addition to the anatase/rutile TiO2, brookite/rutile TiO2 were also explored as
photocatalysts for the degradation of dyes by Zhang et al. [81]. They synthesize
mixed-phase TiO2 nanocrystals with a tunable brookite-to-rutile ratio by a facile
controllable one pot hydrothermal method. The photocatalytic activities of the
resulting TiO2 nanocrystals were examined in the degradation of Rhodamine B
under artificial solar light. The TiO2 nanocrystals with 38 % brookite and 62 %
rutile exhibited the highest photocatalytic activity among the as-prepared samples,
which was about 6 times that of the commercial P25. Recently,Chen et al. grew
anatase grains on the surface of brookite petals resulting in a brookite/anatase TiO2
hybrid [55], which appeared to have superior photocatalytic activity for the single
phase during the degradation of methyl orange (MO) and 2,4-dichlorophenol
(2,4-DCP). The hybrid contains 60 % brookite and 40 % anatase and exhibited the
highest activity; the degradation rate constants of which are 2.27 and 1.80 times
higher than that of the corresponding physically mixed sample for the degradation
of MO and 2,4-DCP, respectively. Following this further Ag modification of the
brookite/anatase composite [123] was carried out. Ag0 clusters with an average
diameter of ca. 1.5 nm formed on the surface of the mixed-phase composite
(Fig. 17.17a, b). The photocatalytic performance of the as-prepared catalysts was
evaluated in terms of the degradation of methyl orange (MO). As shown in
Fig. 17.17d, the Ag0-loaded brookite/anatase composite had a higher degradation
reaction rate constant than that of the pure anatase and brookite, indicating a better
photocatalytic activity. Furthermore, by comparing the results in Fig. 17.17c, d, it
can be found that the Ag0-loaded mixed-phase sample had better catalytic perfor-
mance than that of Ag free sample. The enhanced photocatalytic reactivity is
attributed to the significant improvement in separation of the photo-generated
electrons and holes, which can be ascribed to two aspects: (1) synergistic effect of
brookite and anatase in the composite, and (2) the schottky barrier at the interface of
Ag0 and TiO2.
While most research focused on photocatalytic degradation properties of
biphasic TiO2 catalysts, Que et al. explored brookite/anatase/rutile coexisting TiO2
composite as the photocatalyst [124]. Their results indicated that in the
brookite/anatase/rutile coexisting nanopowders, the brookite and anatase phases
were crystallized into irregular nanoparticles <20 nm in diameter, whereas the rutile
phase was crystallized into single-crystalline nanorods *20 nm in diameter and
100–500 nm in length, as shown in Fig. 17.18a. This triphasic TiO2 catalyst was
demonstrated to have better photocatalytic activity during degradation of methyl
orange than the biphasic commercial P25 under irradiation of UV light. The sample
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 449

Fig. 17.17 TEM images of 2.0 % Ag0-TiO2 (brookite/anatase) (a, b); photocatalytic degradation
of MO with Ag-free (c) and 2.0 % Ag0-loaded (d) TiO2 with different crystal phase composition
under UV irradiation [55]

with 29.9 % anatase, 27.9 % brookite, 42.2 % rutile (referred as T2) was shown to
have the highest photocatalytic activity, yielding over 90 % bleaching of methyl
orange solution in 20 min (Fig. 17.18b). The degradation rate constant k of this
sample was 0.10180 min−1, almost twice as high as that of P25 (k = 0.05397 min−1).
In addition, recycle experiments of the photocatalyst were also carried out by using
the best photocatalyst sample T2 as shown in Fig. 17.18c. It can be seen that after
five recycles, the photocatalytic activity of the sample shows only a slight loss,
indicating a good stability of the triphasic TiO2 catalyst in the photocatalytic
degradation process. Similarly, Shao et al. also developed brookite/anatase/rutile
triphasic TiO2 and the catalyst was demonstrated to have excellent photocatalytic
performance in the degradation of methylene blue solutions.
Apart from dyes, some other organic compounds have also been used as the
model pollutants in the mixed-phase TiO2 photocatalysis system. For example,
phenol and its derivatives have been widely studied as photodegradation target
molecules, since they are considered to be one type of primary pollutants due to
their danger to organisms at low concentrations [125]. Lu et al. [126] prepared a
series of TiO2 samples with different anatase-to-rutile ratios, and studied the role of
crystal phase in photocatalytic oxidation of phenol in water. They found that
samples with higher anatase-to-rutile ratios have higher activities for phenol
450 J. Lei et al.

Fig. 17.18 a Scheme and TEM image of the brookite/anatase/rutile nanocomposites;

b Photocatalytic degradation of the MO aqueous solution by using the brookite/anatase/rutile
nanocomposites synthesized with various reagent ingredients, and the inset shows the removal of
MO monitored by TOC with the sample T2 as photocatalysts; c Cycling degradation curves of the
brookite/anatase/rutile coexisting nanocomposites (sample T2) [124]

degradation. Tian et al. [127] reported Cr-doped TiO2 (Cr-TiO2) nanoparticles with
anatase and rutile bicrystalline phases for the photocatalytic degradation of
2,4-dichlorophenol (2,4-DCP). It was revealed that Cr3+ doping can not only
effectively extend the visible light response of TiO2 (Fig. 17.19a) but also promote
the anatase-to-rutile transformation. The photocatalytic activities of different
Cr-TiO2 photocatalytsts were evaluated in terms of the photocatalytic degradation
of 2, 4-dichlorophenol (2, 4-DCP) under visible light irradiation (Fig. 17.19b). They
found appropriate Cr3+ doping can significantly enhance the visible-light photo-
catalytic activity of TiO2, which is attributed to the improvement of visible light
response as well as suitable anatase-to-rutile ratio. In addition, excess Cr3+ doping
is detrimental to the improvement of visible light photocatalytic activity, due to the
formation of Cr2O3 clusters as well as too high a rutile content.
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 451

Fig. 17.19 a UV-vis DRS and b degradation curves of 2, 4-DCP over undoped TiO2 and different
Cr-TiO2 samples under visible light irradiation: a undoped TiO2, b 0.5 % Cr-TiO2, c 1 % Cr-TiO2,
d 2 % Cr-TiO2, and e 5 % Cr-TiO2. The inset is the absorption spectra of 2, 4-DCP over different
samples after visible light irradiation for 8 h [127]

17.5 Mechanism of the Enhanced Photocatalytic Activities

by the Mixed-Phase TiO2 Photocatalysis

Generally, most of the photo-generated electrons and holes of the single-phase TiO2
will recombine in the bulk, and only a small amount of them can migrate to the
surface to take part in the oxidation-reduction reactions with absorbed molecules,
leading to relatively low photocatalytic efficiency. In 1991, Bickley et al. [128] first
proposed the mechanism of enhanced photocatalytic activity of P25 which is
composed of 80 % anatase and 20 % rutile. They proved that the mixed-phase
structure of P25 is an anatase structure coated with a layer of rutile film by TEM.
By evaluating the photocatalytic activities, they found that the photocatalytic
activity of the mixed-phase TiO2 was better than the activity of any other
single-phase TiO2 photocatalyst. Because the anatase and rutile have different band
gaps, a bending band is generated on the interface of the two phases, as shown in
Fig. 17.20. Under light irradiation, the photo-generated electrons migrate from
rutile phase to anatase phase, while the holes migrate from anatase phase to rutile
phase, so that the electrons and holes can be separated effectively, causing the
mixed-phase P25 to have a high photcatalytic activity. Meanwhile, they speculated
that the mechanism of the enhanced photocatalytic activity of mixed-phase TiO2 is
actually more complex and requires further study. Subsequently, researchers have
studied the mobility direction of the photo-generated carriers, the molecular
dynamics characterizations of the mixed-phase interface [129], the band structure of
mixed-phase TiO2 [130, 131], the suitable phase proportion in the mixed-phase
TiO2 and so on.
452 J. Lei et al.

Fig. 17.20 Schematic band

structure diagrams of
P25 [128]

In 1995, Datye et al. [132]. refuted the view of Bickley et al. [128] on the
mixed-phase crystal structure in P25 by X-ray diffraction (XRD) and high-
resolution transmission electron microscopy (HRTEM) techniques. Their charac-
terization results showed that the mixed-phase P25 consists of individual single
crystal particles of the anatase and the rutile phases of titania rather than particles of
anatase covered by a layer of rutile. This view was confirmed by Zhang et al. [133]
and Ohno et al. [134]. But they did not pursue further research on the mechanism of
the enhanced photocatalytic performance resulted from the mixing of two phases.
In 2002, Kawahara et al. [135] designed a film model to study the migration
direction of the photo-generated electrons and holes in anatase/rutile mixed-phase
crystal under light irradiation. In their experiments, the mixed-phase crystal TiO2
film was immersed into an AgNO3 solution, and it was exposed to light over a
period of time under an argon atmosphere. Experimental results showed that there
were a large number of Ag particles on the surface of the rutile phase. It is possible
that Ag+ caused a reduction reaction on the rutile surface, indicating that the
photo-generated electrons migrated from the conduction band of anatase to rutile
(Fig. 17.21).
In 2003, Sun et al. [137, 138] deposited Pt on P25 for the photocatalytic
degradation of phenol. However, they found that loading P25 with Pt didn’t help to
increase phenol decomposition and total carbon removal rates. In addition, they
found that Pt appeared on the surface of anatase. Therefore, they proposed a charge
separation mechanism for the mixed-phase P25 as “contact of the two phases
causing the bending of the conduction band resulting in the holes in anatase
transferring to rutile, while the electrons cannot transfer from anatase to rutile”, as
shown in Fig. 17.22. As a result, holes were concentrated in rutile and electrons
were left in the anatase particles, which indicated that oxidation happens mostly on
rutile and reduction mostly on anatase, thus the Pt deposition on anatase cannot
increase its photo efficiency for phenol oxidation in water.
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 453

Fig. 17.21 A proposed schematic illustrations showing the migration of the electron transfer from
anatase to rutile [135, 136]

Fig. 17.22 a Band structure before anatase and rutile contact each other; b Band structure and the
separation of photo-generated charge carriers after anatase and rutile contact each other [137]

In the same year, Hurum et al. [51] characterized the charge separation process
of P25 by electron paramagnetic resonance (EPR). When P25 mixed-phase crystal
was excited by visible light, due to the wide band gap of anatase, no electrons and
holes were generated. However, in electron paramagnetic resonance spectroscopy,
the electrons trapping sites were found on anatase surface, and it could be con-
cluded that the electrons in rutile generated by visible light excitation transferred to
the lower energy anatase lattice trapping sites, leading to a more stable charge
separation and enhancing the photocatalytic activity of P25 (Fig. 17.23). Thereafter,
this mechanism was confirmed by Liu et al. [139] through the photocatalytic
activity experiments of anatase/rutile mixed-phase TiO2 nanotubes. EPR spec-
troscopy is a new characterization method developed in recent years, which is a
spectroscopic method to detect the unpaired electrons in the sample. Since then,
researchers have began to characterize the migration of photo-generated electrons
and holes in the mixed-phase TiO2 using the EPR technique [140].
In 2006, Wang et al. [141] proposed an “antenna mechanism” to explain the
mechanism of enhanced photocatalytic activity in mixed crystal phases. The
specific mechanism is as follows: in the photocatalytic degradation process, TiO2
absorbs light to produce photo-generated electrons and holes, which then take part
454 J. Lei et al.

Fig. 17.23 A proposed schematic illustrations showing the charge separation using EPR [51]

in the oxidation-reduction reactions with the target molecules adsorbing on the

catalyst’s surface; However, the depth in liquid which light can reach is limited, so
that the deeper the liquid molecules, the less light they receive. Therefore, the
particles irradiated by light can absorb photo energy, this energy can then be
transferred from to particles deeper in the liquid, finally oxidation-reduction reac-
tion takes place on the particles which light cannot reach, enhancing the photo-
catalytic activity. In the process of photo energy being transferred to the
photocatalyst deeper in the liquid, the long-chain particles act as an antenna system
transferring the photon energy from the location of light absorption to the location
of reaction, thus this is called an “antenna mechanism” (Fig. 17.24).
However, the research objects of the above-mentioned theoretical models are all
P25 which has a fixed phase ratio of anatase and rutile. Because mixed-phase TiO2
with different phase ratios have different photocatalytic activities, none of these
existing models can adequately explain the physical phenomenon responsible for
the existence of an optimum phase content in mixed-phase nanocrystalline TiO2
corresponding to the maximum photocatalytic activity. Therefore, further research
is needed to investigate the photocatalytic mechanism of mixed-phase TiO2 with
different phase ratios.
In 2008, Zachariah et al. [104] proposed a new mechanism model, which takes
the factors of crystallite size distribution and phase composition into account. This
mechanism overcomes the limitations of the existing models proposed earlier in the
literature in terms of explaining the phenomenon of an existing optimal phase ratio
in the mixed-phase TiO2 for the best photocatalytic activity. The mechanism is
based on the charge separation mechanism proposed by Sun et al. [137]. In the
mechanism, when the size of nano TiO2 is identical (Fig. 17.25a), the band gap of

Fig. 17.24 Antenna effect by network structure leading to enhanced photocatalytic activity [141]
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 455

Fig. 17.25 The newly

proposed mechanism model
based on the band-gap
variation in the connected
nanocrystallites as a function
of the size distribution and the
phase involved [104].
A, R, CB, and VB represent
anatase-TiO2, rutile-TiO2, the
conduction band, and the
valence band, respectively
[104]

adjacent anatase crystallites is the same, so there is no driving force for the
migration of photo-generated holes, resulting in high recombination rate of
photo-generated charges and low photocatalytic activity. If the sizes of anatase
crystallite are different and are below the critical size (Fig. 17.25b, c), the band gaps
of the connected crystallites would depend on their size. Thus, photo-generated
holes in one crystallite can easily escape into the other leading to effective charge
separation. When a small amount of rutile was mixed with anatase (Fig. 17.25d),
due to the different band gap, photo-generated holes will transfer from anatase to
rutile, so that the photocatalytic activity is further enhanced. However, if the rutile
content exceeds a certain amount, the migration of photo-generated holes in
mixed-phase crystal TiO2 is limited (Fig. 17.25e), and the photocatalytic activity
begins to decline. Given that the crystallite sizes of pure rutile are above the critical
size (Fig. 17.25f), their band gap values are the same, then the migration of pho-
togenerated holes does not occur, and the photocatalytic activity is minimum. The
key point of this mechanism is that whether the TiO2 is mixed-phase crystalline or
456 J. Lei et al.

single crystalline, as long as the band gap values are different, the photo-generated
electron-hole pairs will be separated, thereby improving the photocatalytic activity.
In summary, the above mechanisms for the enhanced photocatalytic perfor-
mance of mixed-phase TiO2 all involve the migration behavior of photo-generated
electrons and holes. In these mechanisms, the mixed-phase crystal structure enables
the effective separation of photo-generated electrons and holes, while the specific
paths of migration are still controversial. The determining factors for photocatalytic
activity of mixed-phase crystal TiO2 include grain size, structural forms (core-shell
structure, cladding structure, random composite, etc.), phase composition. These
factors lead to the emergence of bending band, trapping sites and other phenomena
in the mixed-phase crystal. Therefore, under different experimental conditions, the
obtained mixed-phase TiO2 differ in photocatalytic mechanism, so further research
is still required.

17.6 Conclusion and Outlook

Research in the synthetic methods and applications in photocatalysis of

mixed-phase TiO2 have made great progress since the discovery of photocatalytic
water splitting and degradation of organic pollutants by TiO2 nanomatrials in the
1970s. The superior photocatalytic performance of mixed-phase TiO2 to
single-phase TiO2 has been well received and the mechanism of the enhanced
photocatalytic activities by the mixed-phase TiO2 photocatalysis has also been
widely studied. However, there are still many problems remaining to be resolved.
For example, (1) Particulate mixed-phase TiO2 nanomaterials are prone to
agglomeration, which will greatly decrease the photocatalytic activity and hinder
the further development of the materials; (2) Although the relatively narrow band
gap of rutile extends the absorption of the mixed-phase TiO2 to some of the visible
light range, it is still insufficient to utilize all visible light from sun light, limiting its
application in photocatalysis; (3) Although researchers all agree that the
mixed-phase crystal structure enables the effective separation of photogenerated
electrons and holes, thereby enhancing the photocatalytic activity, there is much
controversy about the mobility direction of photo-generated electrons and holes.
Further research is essential to confirm the mechanism of the enhanced photocat-
alytic activity of mixed-phase TiO2. Therefore, developing mixed-phase TiO2
nanomaterials and combining other advantageous structures (such as hierarchical
structure), enhancing the visible light utilization, and exploring the mechanism by
newly developed characterization techniques will continue to be the challenges and
hot research topics in this area.
Although mixed-phase crystal TiO2 has been studied for decades, it is still a
popular topic due to its excellent photocatalytic activity. Nowadays, with increas-
ingly serious energy and environmental problems, the applications of mixed-phase
TiO2 is anticipated to attract considerable attention.
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 457

References

1. A. Fujishima, K. Honda, Nature 238, 37–38 (1972)

2. J.H. Carey, J. Lawrence, T. Hm, Bull. Environ. Contam. Toxicol. 16, 697–701 (1977)
3. S.N. Habisreutinger, S.L. Mende, J.K. Stolarczyk, Angewandte Chemie 52, 7372–7408
(2013)
4. H. Ke, K.C.D. Robson, P.G. Johansson, C.P. Berlinguette, G.J. Meyer, J. Am. Chem. Soc.
134, 8352–8355 (2012)
5. Q. Guo, C. Xu, Z. Ren, W. Yang, Z. Ma, D. Dai, H. Fan, T.K. Minton, X. Yang, J. Am.
Chem. Soc. 134, 2827–2830 (2012)
6. B. Tian, F. Chen, J. Zhang, M. Anpo, J. Colloid Interface Sci. 303, 142–148 (2006)
7. M. Tomkiewicz, G. Dagan, Z. Zhu, Res. Chem. Intermed. 20, 701–710 (1994)
8. S. Zhu, G. Xie, X. Yang, Z. Cui, Mater. Res. Bull. 48, 1961–1966 (2013)
9. T. Beuvier, M. Richard-Plouet, G.M. Mancini-Le, T. Brousse, O. Crosnier, L. Brohan, Inorg.
Chem. 49, 8457–8464 (2010)
10. X. Xin, M. Scheiner, M. Ye, Z. Lin, Langmuir 27, 14594–14598 (2011)
11. E. Hosono, S. Fujihara, H. Imai, I. Honma, I. Masaki, H. Zhou, ACS Nano 1, 273–278
(2007)
12. A.K. Sinha, S. Jana, S. Pande, S. Saha, A. Pal, T. Pal, S. Sarkar, M. Pradhan, M. Basu, Cryst.
Eng. Comm. 7, 1210–1212 (2009)
13. Q.Q. Cheng, Y. Cao, L. Yang, P.P. Zhang, K. Wang, H.J. Wang, Mater. Res. Bull. 46,
372–377 (2011)
14. Y. Wang, L. Zhang, K. Deng, X. Chen, Z. Zou, J. Phys. Chem. C 111, 2709–2714 (2007)
15. J. Wei, J. Yao, X. Zhang, W. Zhu, H. Wang, M.J. Rhodes, Mater. Lett. 61, 4610–4613
(2007)
16. Y. Hu, Angew. Chem. Int. Ed. 51, 12410–12412 (2012)
17. J.K. Oh, J.K. Lee, H.S. Kim, S.B. Han, K.W. Park, Chem. Mater. 22, 1114–1118 (2009)
18. M. Ye, H.Y. Liu, C. Lin, Z. Lin, Small 9, 312–321 (2013)
19. S. Peng, S. Ding, J. Yuan, W.L. Xiong, D.H. Kim, ACS Nano 5, 7617–7626 (2011)
20. E. Lioz, G. Peng, X. Zhaosheng, P. Qin, C. Aravind Kumar, L. Bin, M.K. Nazeeruddin, G.T.
Michael, J. Am. Chem. Soc. (2012)
21. W. Guo, C. Xu, X. Wang, S. Wang, C. Pan, C. Lin, Z. Wang, J. Am. Chem. Soc. 134,
4437–4441 (2012)
22. Y.Q. Wang, G. Lin, Y.G. Guo, L. Hong, X.Q. He, S. Tsukimoto, Y. Ikuhara, L.J. Wan,
J. Am. Chem. Soc 134, 7874–7879 (2012)
23. S.S.K. Lee, P. Schmuki, J. Am. Chem. Soc. 134, 11316–11318 (2012)
24. W.N. Wang, W.J. An, B. Ramalingam, S. Mukherjee, D.M. Niedzwiedzki, S.
Gangopadhyay, P. Biswas, J. Am. Chem. Soc. 134, 11276–11281 (2012)
25. S.A.K. Leghari, S. Sajjad, F. Chen, J. Zhang, Chem. Eng. J. 166, 906–915 (2011)
26. F. Chen, W. Zou, W. Qu, J. Zhang, Catal. Commun. 10, 1510–1513 (2009)
27. M. Xing, J. Zhang, F. Chen, Appl. Catal. B 89, 563–569 (2009)
28. T.H. Kim, C. Gómez-Solís, E. Moctezuma, S.W. Lee, Res. Chem. Intermed. 40, 1595–1605
(2014)
29. F. Zuo, L. Wang, T. Wu, Z. Zhang, D. Borchardt, P. Feng, J. Am. Chem. Soc. 132,
11856–11857 (2010)
30. J.K. Yong, H.L. Mi, H.J. Kim, G. Lim, Y.S. Choi, N.G. Park, K. Kim, I.L. Wan, Adv. Mater.
21, 3668–3673 (2009)
31. S. Liu, Q. Li, C. Hou, X. Feng, Z. Guan, J. Alloy. Compd. 575, 128–136 (2013)
32. M. Cai, X. Pan, W. Liu, J. Sheng, X. Fang, C. Zhang, Z. Huo, H. Tian, S. Xiao, S. Dai,
J. Mater. Chem. A 15, 4885–4892 (2013)
33. M.S. Liang, C.C. Khaw, C.C. Liu, S.P. Chin, J. Wang, H. Li, Ceram. Int. 39, 1519–1523
(2013)
458 J. Lei et al.

34. H. Kim, H.H. Yun, G. Cho, D. Kim, N. Lim, M. Pyo, Electrochim. Acta 56, 9476–9481
(2011)
35. C. Zhang, Y. Huang, S. Chen, H. Tian, L.E. Mo, L. Hu, Z. Huo, F. Kong, Y. Ma, S. Dai,
J. Phys. Chem. C 116, 19807–19813 (2012)
36. E.G. Bae, H. Kim, Y.H. Hwang, K.S. Sohn, M. Pyo, J. Mater. Chem 22, 551–556 (2011)
37. A.A. Ashkarran, M. Ghavamipour, H. Hamidinezhad, H. Haddadi, Res. Chem. Intermed.
(2014)
38. M. Haruta, B.S. Uphade, S. Tsubota, A. Miyamoto, Res. Chem. Intermed. 24, 329–336
(1998)
39. B. Tian, J. Zhang, T. Tong, F. Chen, Appl. Catal. B 79, 394–401 (2008)
40. Y. Wu, H. Liu, J. Zhang, F. Chen, J. Phys. Chem. C 113, 14689–14695 (2009)
41. W. Wang, J. Zhang, F. Chen, M. Anpo, D. He, Res. Chem. Intermed. 36, 163–172 (2010)
42. W. Yan, C. Feng, Z. Min, J. Yang, Z. Zhang, Appl. Catal. B 100, 84–90 (2010)
43. C. Feng, Y. Wang, J. Zhang, L. Yu, D. Li, J. Yang, Z. Zhang, Appl. Catal. B Environ.
113–114, 61–71 2012
44. A. Charanpahari, S.S. Umare, S.P. Gokhale, V. Sudarsan, B. Sreedhar, R. Sasikala, Appl.
Catal. A 443, 96–102 (2012)
45. B. Tian, C. Li, F. Gu, H. Jiang, Catal. Commun. 10, 925–929 (2009)
46. P. Zhang, C. Shao, X. Li, M. Zhang, X. Zhang, Y. Sun, Y. Liu, J. Hazard. Mater. 237–238,
331–338 (2012)
47. Z.H. Zhang, Y. Yuan, L.H. Liang, Y.X. Cheng, G.Y. Shi, L.T. Jin, J. Hazard. Mater. 158,
517–522 (2008)
48. G. Chattopadhyaya, D.G. Macdonald, N.N. Bakhshi, J.S.S. Mohammadzadeh, A.K. Dalai,
Catal. Lett. 108, 1–5 (2006)
49. M. Yang, Y. Men, S. Li, G. Chen, Appl. Catal. A 433, 26–34 (2012)
50. S. Ren, R. Bechstein, L. S03, R.T. Vang, M. Sillassen, B. Esbjörnsson, A. Palmqvist, F.
Besenbacher, J. Phys. Chem. C 115, 24287–24292 (2011)
51. D.C. Hurum, A.G. Agrios, K.A. Gray, T. Rajh, M.C. Thurnauer, J. Phys. Chem. B 107,
4545–4549 (2003)
52. R. Scotti, I.R. Bellobono, C. Canevali, C. Cannas, M. Catti, M. D’Arienzo, A. Musinu, S.
Polizzi, M. Sommariva, A. Testino, Chem. Mater. 20, 4051–4061 (2008)
53. V. Puddu, H. Choi, D.D. Dionysiou, G.L. Puma, Appl. Catal. B 94, 211–218 (2010)
54. R. Zheng, X. Meng, F. Tang, Appl. Surf. Sci. 255, 5989–5994 (2009)
55. Y. Jiao, F. Chen, B. Zhao, H. Yang, J. Zhang, Colloids Surf., A 402, 66–71 (2012)
56. A.D. Paola, G. Cufalo, M. Addamo, M. Bellardita, R. Campostrini, M. Ischia, R. Ceccato,
L. Palmisano, Colloids Surf., A 317, 366–376 (2008)
57. W. Li, C. Liu, Y. Zhou, Y. Bai, X. Feng, Z. Yang, L. Lu, X. Lu, K.Y. Chan, J. Phys. Chem.
C 112, 20539–20545 (2008)
58. D. Hanaor, C. Sorrell, J. Mater. Sci. 46, 855–874 (2011)
59. J.A. Gamboa, D.M. Pasquevich, J. Am. Ceram. Soc. 75, 2934–2938 (1992)
60. X.Z. Ding, Y.Z. He, J. Mater. Sci. Lett. 15, 320–322 (1996)
61. J. Muscat, V. Swamy, N.M. Harrison, Phys. Rev. B 65, 392–397 (2002)
62. T. Arlt, M. Bermejo, M.A. Blanco, L. Gerward, J.Z. Jiang, J. Staun Olsen, J.M. Recio, Phys.
Rev. B 61, 14414–14419 (2000)
63. R. Ren, Z. Yang, L.L. Shaw, J. Mater. Sci. 35, 6015–6026 (6012) (2000)
64. N.A. Dubrovinskaia, L.S. Dubrovinsky, R. Ahuja, V.B. Prokopenko, V. Dmitriev,
H.P. Weber, J.M. Osorio-Guillen, B. Johansson, Phys. Rev. Lett. 87, 455–475 (2001)
65. U. Diebold, Surf. Sci. Rep. 48, 53–229 (2003)
66. M. Yan, F. Chen, J. Zhang, M. Anpo, J. Phys. Chem. B, 109 (2005)
67. Y. Li, T.J. White, S.H. Lim, J. Solid State Chem. 177, 1372–1381 (2004)
68. A.R. Khataee, M.B. Kasiri, J. Mol. Catal. A: Chem. 328, 8–26 (2010)
69. N. Daneshvar, M.H. Rasoulifard, A.R. Khataee, F. Hosseinzadeh, J. Hazard. Mater. 143,
95–101 (2007)
70. M.P. Moret, R. Zallen, D.P. Vijay, S.B. Desu, Thin Solid Films 366, 8–10(13) (2000)
17 Mixed-Phase TiO2 Nanomaterials as Efficient Photocatalysts 459

71. R.D. Shannon, J.A. Pask, J. Am. Ceram. Soc. 48, 391–398 (1965)
72. M. Batzill, E.H. Morales, U. Diebold, Phys. Rev. Lett. 96 (2006)
73. G. Oskam, A. Nellore, R.L. Penn, P.C. Searson, J. Phys. Chem. B 107, 1734–1738 (2003)
74. H. Zhang, J.F. Banfield, J. Mater. Chem. 9, 2073–2076 (1998)
75. J.G. Li, T. Ishigaki, X. Sun, J. Phys. Chem. C 111, 4969–4976 (2007)
76. R.I. Bickley, T. Gonzalez-Carreno, J.S. Lees, L. Palmisano, R.J.D. Tilley, J. Solid State
Chem. 92, 178–190 (1991)
77. S.-K. Lee, P.K.J. Robertson, A. Mills, D. McStay, J. Photochem. Photobiol. A 122, 69–71
(1999)
78. S.J. Tsai, S. Cheng, Catal. Today 33, 227–237 (1997)
79. T. Ohno, K. Sarukawa, M. Matsumura, J. Phys. Chem. B 105, 2417–2420 (2001)
80. T. Ozawa, M. Iwasaki, H. Tada, T. Akita, K. Tanaka, S. Ito, J. Colloid Interface Sci. 281,
510–513 (2005)
81. H. Xu, L. Zhang, J. Phys. Chem. C 113, 1785–1790 (2009)
82. R.R. Bacsa, J. Kiwi, Appl. Catal. B 16, 19–29 (1998)
83. K.Y. Jung, S.B. Park, H.D. Jang, Catal. Commun. 5, 491–497 (2004)
84. S. Leghari, S. Sajjad, F. Chen, J. Zhang, Chem. Eng. J. 3, 906–915 (2011)
85. J. Zhu, W. Zheng, B. He, J. Zhang, M. Anpo, J. Mol. Catal. A: Chem. 216, 35–43 (2004)
86. Y. Wu, M. Xing, J. Zhang, J. Hazard. Mater. 192, 368–373 (2011)
87. J. Ng, X. Wang, D.D. Sun, Appl. Catal. B 110, 260–272 (2011)
88. J. Ovenstone, K. Yanagisawa, Chem. Mater. 11, 2770–2774 (1999)
89. G. Li, S. Ciston, Z.V. Saponjic, L. Chen, N.M. Dimitrijevic, T. Rajh, K.A. Gray, J. Catal.
253, 105–110 (2008)
90. M. Fehse, F. Fischer, C. Tessier, L. Stievano, L. Monconduit, J. Power Sources 231, 23–28
(2013)
91. Y. Zhang, J. Chen, X. Li, Catal. Lett. 139, 129–133(125) 2010
92. Q. Tay, X. Liu, Y. Tang, Z. Jiang, T.C. Sum, Z. Chen, J. Phys. Chem. C 117, 14973–14982
(2013)
93. H. Zhang, J.F. Banfield, J. Phys. Chem. B 104, 3481–3487 (2000)
94. X. Shen, B. Tian, J. Zhang, Catal. Today 201, 151–158 (2013)
95. B. Zhao, F. Chen, Q. Huang, J. Zhang, Chem. Commun. 34, 5115–5117 (2009)
96. L.M. Zhao, Z.J. Zhang, S.Y. Zhang, P. Cui, W. Shi, B. Zhao, P. Cheng, D.Z. Liao, S.P. Yan,
Cryst. Eng. Comm. 3, 907–913 (2011)
97. B. Zhao, F. Chen, Y. Jiao, J. Zhang, J. Mater. Chem. 20, 7990–7997 (2010)
98. X. Shen, J. Zhang, B. Tian, M. Anpo, J. Mater. Sci. 47, 5743–5751 (2012)
99. H. Cheng, J. Ma, Z. Zhao, L. Qi, Chem. Mater. 7, 663–671 (1995)
100. G. Li, K.A. Gray, Chem. Mater. 19, 1143–1146 (2007)
101. S. Lei, W. Duan, J. Environ. Sci. 20, 1263–1267 (2008)
102. M. Wu, J. Long, A. Huang, Y. Luo, S. Feng, R. Xu, Langmuir 15, 8822–8825 (1999)
103. X. Shen, J. Zhang, B. Tian, J. Hazard. Mater. 192, 651–657 (2011)
104. A. Zachariah, K.V. Baiju, S. Shukla, K.S. Deepa, J. James, K.G.K. Warrier, J. Phys. Chem.
C 112, 11345–11356 (2008)
105. A. Bojinova, R. Kralchevska, I. Poulios, C. Dushkin, Mater. Chem. Phys. 106, 187–192
(2007)
106. Z. Liu, X. Zhang, S. Nishimoto, J. Ming, D.A. Tryk, T. Murakami, A. Fujishima, Langmuir
23, 10916–10919 (2007)
107. R.G. Nair, S. Paul, S.K. Samdarshi, Sol. Energy Mater. Sol. Cells 95, 1901–1907 (2011)
108. P.I. Gouma, M.J. Mills, J. Am. Ceram. Soc. 84, 619–622 (2001)
109. Z. Jing, M. Li, Z. Feng, J. Chen, C. Li, J. Phys. Chem. B 110, 927–935 (2006)
110. Y.C. Hsu, H.C. Lin, C.H. Chen, Y.T. Liao, C.M. Yang, J. Solid State Chem. 183, 1917–1924
(2010)
111. M. Ni, M.K.H. Leung, D.Y.C. Leung, K. Sumathy, Renew. Sustain. Energy Rev. 11,
401–425 (2007)
460 J. Lei et al.

112. I.R. Oca09a, A. Beltram, J.J.D. Jaén, G. Adami, T. Montini, P. Fornasiero, Inorganica
Chimica Acta (2015)
113. Q. Xu, Y. Ma, J. Zhang, X. Wang, Z. Feng, C. Li, J. Catal. 278, 329–335 (2011)
114. Y.K. Kho, A. Iwase, W.Y. Teoh, L. M01dler, A. Kudo, R. Amal, J. Phys. Chem. C, 114,
2821–2829 (2010)
115. F. Xu, W. Xiao, B. Cheng, J. Yu, Int. J. Hydrogen Energy 39, 15394–15402 (2014)
116. O. Rosseler, M.V. Shankar, M.K.-L. Du, L. Schmidlin, N. Keller, V. Keller, J. Catal. 269,
179–190 (2010)
117. G. Marcı̀, M. Addamo, V. Augugliaro, S. Coluccia, E. Garcı́a-López, V. Loddo, G. Martra,
L. Palmisano, M. Schiavello, J. Photochem. Photobiology A: Chem. 160, 105–114 (2003)
118. L. Liu, H. Zhao, J.M. Andino, Y. Li, ACS Catal. 2, 1817–1828 (2012)
119. H. Zhao, L. Liu, J.M. Andino, Y. Li, J. Mater. Chem. A 1, 8209–8216 (2013)
120. S.N. Frank, A.J. Bard, J. Am. Chem. Soc. 99, 303–304 (2002)
121. H. Xu, G. Li, G. Zhu, K. Zhu, S. Jin, Catal. Commun. 62, 52–56 (2015)
122. Z. Luo, A. S. Poyraz, C. H. Kuo, R. Miao, Y. Meng, S. Y. Chen, T. Jiang, C. Wenos,
S. L. Suib, Z. Luo, Chem. Mater. 27 (2014)
123. B. Zhao, Y. Jiao, F. Chen, H. Yang, J. Zhang, J. Mol. Catal. A: Chem. 348, 114–119 (2011)
124. Y. Liao, W. Que, Q. Jia, Y. He, J. Zhang, P. Zhong, J. Mater. Chem. 22, 7937–7944 (2012)
125. E. Grabowska, J. Reszczyńska, A. Zaleska, Water Res. 46, 5453–5471 (2012)
126. Z. Ding, G. Lu, P. Greenfield, J. Phys. Chem. B 104, 4815–4820 (2000)
127. B. Tian, C. Li, J. Zhang, Chem. Eng. J. 191, 402–409 (2012)
128. R.I. Bickley, T. Gonzalez-Carreno, J.S. Lees, L. Palmisano, R.J.D. Tilley, J. Solid State
Chem. (United States), 92:1, 178–190 (1991)
129. N.A. Deskins, S. Kerisit, K.M. Rosso, M. Dupuis, J. Phys. Chem. C 111, 9290–9298 (2007)
130. P. Deák, B. Aradi, T. Frauenheim, J. Phys. Chem. C 115, 3443–3446 (2011)
131. D.O. Scanlon, C.W. Dunnill, J. Buckeridge, S.A. Shevlin, A.J. Logsdail, S.M. Woodley,
C.R.A. Catlow, M.J. Powell, R.G. Palgrave, I.P. Parkin, Nat. Mater. 12, 798–801 (2013)
132. A.K. Datye, G. Riegel, J.R. Bolton, M. Huang, M.R. Prairie, J. Solid State Chem. 115,
236–239 (1995)
133. Z. Zhang, C.C. Wang, R. Zakaria, J.Y. Ying, J. Phys. Chem. B 102, 10871–10878 (1998)
134. T. Ohno, K. Sarukawa, K. Tokieda, M. Matsumura, J. Catal. 203, 82–86 (2001)
135. T. Kawahara, Y. Konishi, H. Tada, N. Tohge, J. Nishii, S. Ito, Angew. Chem. Int. Ed. Engl.
41, 2811–2813 (2002)
136. G. Li, K.A. Gray, Chem. Phys. 339, 173–187 (2007)
137. B. Sun, A.V. Vorontsov, P.G. Smirniotis, Langmuir 19, 3151–3156 (2003)
138. S. Bo, P.G. Smirniotis, Catal. Today 88, 49–59 (2003)
139. B. Liu, L. Peng, J. Alloy. Compd. 571, 145–152 (2013)
140. G. Li, L. Chen, M.E. Graham, K.A. Gray, J. Mol. Catal. A: Chem. 275, 30–35 (2007)
141. C.-Y. Wang, R. Pagel, J.K. Dohrmann, D.W. Bahnemann, C. R. Chim. 9, 761–773 (2006)
Chapter 18
Electrochemical Impedance Study
on Poly(Alkylenedioxy)Thiophene
Nanostructures: Solvent
and Potential Effect

A. Sezai Sarac and Aslı Gencturk

Abstract Conducting polymers can be doped and dedoped rapidly to a high charge
density, hence, they can be applied as active materials for supercapacitors. Higher
energy densities can be achieved because charging occurs through very thin
thicknesses from the nano to microscale range. Taking into account the costs and
compatibility of the materials, the modification of carbon fiber by electrocoating of
poly(3,4-alkylenedioxythiophene)s for microsupercapacitor applications seems to
be a very attractive method. [3,4-(2,2-dimethylpropylenedioxy)thiophene] was
electrodeposited cyclovoltametrically onto the carbon fiber micro electrode. An
electrochemical impedance spectroscopic study was performed at applied potential,
in different electrolytes and solvents and evaluated with our previous findings by
reviewing.

18.1 Introduction

Owing to its good electrochemical properties and its low oxidation potential, its high
conductivity [1], unusual thermal stability in the oxidized state, poly(3,4 ethylene-
dioxythiophene) (PEDOT) has been widely investigated [1, 2]. The electron
donating oxygen atoms in 3- and 4- position not only reduce the oxidation potential
of the aromatic ring but also prevent α,β-coupling during the polymerization. The
ethylene bridges minimize steric distortion effects resulting in a high stereoregularity
of the polymer chain so that good-conjugation is guaranteed. Although in special
fields of application PEDOT-type materials with improved properties were needed.

A. Sezai Sarac (&)
A. Gencturk

Department of Chemistry, Polymer Science and Technology,
Istanbul Technical University, Maslak, 34469 Istanbul, Turkey
e-mail: [email protected]
URL: http://web.itu.edu.tr/*sarac/

© Springer International Publishing Switzerland 2016 461

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_18
462 A. Sezai Sarac and A. Gencturk

The conductivity and stability of PEDOT makes it useful as a conducting electrode

in electrochemical supercapacitors [3].
As a consequence different types of poly(3,4-alkylenedioxythiophene)s struc-
turally related to PEDOT have been developed [4–6]. In comparison with PEDOT
the structure of 3,4-(2,2-dimethylpropylenedioxy)thiophene (PProDOT-Me2) con-
tains propylene bridges with quaternary central carbon and tetrahedral symmetry
which should allow to stabilize [7] high levels of electrical charge. Poly
[3,4-(2,2-dimethylpropylenedioxy)thiophene] (PProDOT-Me2) is a very promising
candidate for use in electrochromic devices switching between slightly different
colors than PEDOT [8] with higher coloration efficiency [9], higher electrochromic
contrast ratio [10] and very fast switching times. Using PProDOT-Me2 several
prototypes of devices [11–17] have been fabricated.
The tetrahedral substitution pattern of PProDOT-Me2 causes the alkyl groups to
be positioned above and below the plane of the Π conjugated chain, can allow high
doping levels and inhibit π stacking [7], and the quaternary central carbon of the
propylene bridge of 3,4-(2,2-dimethylpropylenedioxy)thiophene should cause
structural conditions that can favour high charge capacitance in well ordered
polymer films. Electrochromic properties of PProDOT-Me2 film was studied on the
potentiostatically deposited ITO coated glass slides previously. PProDOT-Me2
shows good contrast in the visible region. This higher contrast is due to its higher
percentage of transmittance in the oxidised form. At an applied potential of
−600 mV, the polymer is in its fully neutral form; the band gap of the polymer Eg
was calculated as 1.85 eV from onset of π to π* transition [18].
Carbon based materials are well known for the electrochemical applications. As
an electrode material, carbon has several advantages compared to other electrode
materials, i.e. good polarisation of the electrode, high surface area, processability,
low cost, accessibility, thermal and chemical stability. Porous carbon, offers large
surface area and good polarisation due to its porosity, which makes it one of the
advantages electrode materials for supercapacitors applications [19–22].
Carbon fibers have unique properties such as high strength, high modulus and
low density. Recently, the surface properties of carbon fibers (CF) and modification
of carbon fibre surfaces by electropolymerising with thiophene derivatives and their
copolymers onto CF have been investigated in detail [23–25]. Ethylcarbazole-bis-
EDOT-ethylcarbazole comonomer was electrochemically coated onto carbon fibre
microelectrodes (CFME) and stable films were obtained at high potentials [26].
Micron size carbon fiber electrodes (CFMEs) show superior performance in
potentiodynamic studies due to their size and cylindrical shape. They can be used
once (disposable) a new surface area will be present at each measurement compared
to conventional electrodes.
Generally nano- and micro-scale electrocoatings of conjugated polymers on
carbon fiber easily allow the characterization of the deposited films by electro-
chemical, spectroscopic, and morphological techniques [27–31]. The comparison of
18 Electrochemical Impedance Study … 463

different polymers coated under identical electrochemical conditions gives reliable

information [32–35], for the evaluation of the materials in terms of application. The
high stability and electroactivity was obtained for PEDOT on CFME which is
important for the view of practical applications.
For high capacitance the carbon electrodes should display a high specific surface
area that can easily be accessed by the electrolyte; carbon porosity and surface
chemistry are thus of primary importance. The carbons should display a high pore
volume and pores at least wider than the size of the counterions. Although pore
walls should be thicker than the screening length of the electric field in the solid part
of the double layer without limiting the charge accommodation on the two sides of
the carbon walls [36, 37]. Functional groups are also important because they affect
the electronic properties of the carbon itself and its surface polarizability and can
attract–repel the ions with different effects on carbon wettability.
The electrochemical impedance spectroscopy (EIS) is one of the most effective
tools to obtain information about double layer capacitance, resistance, determina-
tion of the rate of the charge transfer and charge transport processes, solution
resistance, etc. [19, 38]. It was used to explain behaviour of polymer coated
electrodes by established theories using two models which are known as uniform
and porous medium.
Electrochemical capacitors (supercapacitors, ultracapacitors) have been inten-
sively investigated since they are expected to be useful for high-power, in partic-
ular, pulse-power applications. Several materials for this application have been
reviewed [39, 40]. Supercapacitors [41] can be classified into two types: double
layer supercapacitors, which store the energy within the electrochemical double
layer at the electrode/electrolyte interface and redox supercapacitors. For redox
supercapacitors, among several types of electrode materials, the use of conducting
polymers has received much attention due to their use in both aqueous and organic
electrolytes and their wide potential working range [42].
Supercapacitors are becoming increasingly important as an intermittent power
supply as well as for load levelling and energy storage in fields ranging from
distributed energy grids to domestic and transport applications [43–45].
Modified carbon fiber electrodes can be good candidates for storing the charges
by two mechanisms: double layer effect, and pseudocapacitance, which can store
more energy than double layer effects.
The paper presents an electrochemical deposition, characterization, impedance
and morphologic study of PProDOT-Me2 which is electrocoated on carbon fiber
microelectrode. The electrochemical characteristics of PProDOT-Me2 on carbon
fiber microelectrode (CFME) in an electrolyte solution of tetrabutylammonium
hexafluorophosphate (Bu4NPF6) in acetonitrile (ACN) shows that it can be used as
an active electrode material for electrochemical supercapacitor applications.
464 A. Sezai Sarac and A. Gencturk

18.2 Experimental Details

18.2.1 Chemicals

Acetonitrile (ACN), dimethylformamide (DMF) and propylenecarbonate (PC) were

used as received from Riedel-de Haen Chemical without further purification.
Tetrabutylammonium hexafluorophosphate (Bu4NPF6), tetrabutylammonium
tetrafloroborate (Bu4NBF4), tetraethylammonium hexafluorophosphate (Et4NPF6),
tetraethylammonium tetrafloroborate (Et4NBF4) were used from Fluka (%98)
Chemicals. SGL SIGRAFIL C 320 B (A high strength and high modulus of
elasticity coupled with high electrical conductivity carbon fibers) (SGL Carbon
Group) containing single filaments in a roving were used as working electrodes.
ProDOT-Me2 monomer was prepared in three synthetic steps from commercially
available starting materials using Mitsunobu conditions [46] followed by alkaline
hydrolysis [47], and decarboxylation [48] of the resulting precursor.

18.2.2 Preparation of Carbon Fiber

Microelectrode (CFME)

A carbon fiber was used as working electrode (3 cm of the CFME with an average
diameter of *7 µm) attached to a copper wire. About 1 cm of the CFMEs were
dipped into the solution to keep the electrode area constant and the rest of the
electrode was covered with the Teflon type. The CFMEs were cleaned with acetone
and then dried up with an air-dryer before the experiments.

18.2.3 Apparatus and Procedure

Cyclic voltammograms (CV) of the polymer coated CFMEs and electrochemical

impedance spectroscopy (with supplied Power Sine software package) were per-
formed on a Princeton Applied Research (PAR) Parstat 2263 potentiostat, which is
a self-contained unit that combines potentiostatic circuitry with phase-sensitive
detection. The potentiostat was connected to a Faraday cage (BAS Cell Stand C3).
Electropolymerisation was performed using a three electrode system with a
CFME as a working electrode, a platinum (Pt) wire as a counter and a silver
(Ag) wire as a pseudoreference electrode. The cell has a volume of 5 mL solution,
arranged in a way at a distance of, 1 cm from each other. The pseudoreference
electrode was calibrated externally using a 5 mM solution of ferrocene/ferrocenium
(Fc/Fc+) redox couple in the electrolyte and potentials were reported against the Ag
electrode.
18 Electrochemical Impedance Study … 465

After the electropolymerization process, coated CFME was washed with ACN
and EIS measurements were performed in monomer free electrolyte solution using a
three electrode system. Electrochemical polymerizations (electrocoatings) were
performed in acetonitrile (ACN) containing 0.1 M Bu4NPF6 at 100 mV/s and
electrochemical impedance measurements were performed in DMF, ACN, and PC
containing 0.1 M electrolyte at applied potential. An electrochemical impedance
software (PowerSine) was used to carry out impedance measurements by scanning
in the 100 kHz–10 mHz frequency range with an applied AC signal amplitude of
10 mV and DC potentials were applied before EIS measurements. The impedance
spectra were analysed using an AC impedance data analysis software (Z SimpWin
V3.10).
Thin films of homopolymers, electrocoated onto carbon fiber surface were
analyzed by FT-IR reflectance spectrophotometry (Perkin-Elmer, Spectrum One,
with Universal ATR attachment with a diamond and ZnSe crystal C70951).
The morphological features of the coated carbon fibre electrodes were observed
using a Hitachi S-2700 scanning electron microscopy (Nissei Sangyo GmbH,
Rathingen, Germany), which was connected to an energy dispersive X-ray
(EDX) microanalyser (Kevex type delta V, Foster City, CA, USA). The excitation
energy was 10 keV at a beam current of 0.5 nA.

18.3 Results and Discussion

18.3.1 Electropolymerization of ProDOT-Me2 on CFME

Oxidative electropolymerization of 5 mM ProDOT-Me2 was performed potentio-

dynamically between 0.0 and 1.6 V versus Ag in ACN with 0.1 M Bu4NPF6 as the
electrolyte. Polymer film was electrodeposited by CV on a CFME (area 0.02 cm2)
at 100 mV s−1, it was scanned in monomer free solution between 20 and
400 mV s−1 in 0.1 M Bu4NPF6/ACN. The typical CV of the polymeric film exhibits
a well defined and reversible redox process as shown in Fig. 18.1.
The processes involved thus give rise in CV to almost mirror image voltam-
mograms that are the charateristics of capacitor discharge and recharge behavior,
and also to response currents that increase with scan rate. The current of anodic
peak potential, ipa depends linearly on the scan rate (up to 40th cycle). The seg-
ments of the film consisting of the longer (and porous in some cases) polymer
chains (filled more with dopant) are more compact than the short chain segments.
The incorporation of ions into the compact parts of the film during oxidation is slow
therefore the process can be controlled by diffusion. Multisweep cyclic voltam-
mogram of 5 mM ProDOT-Me2 in 0.1 M Bu4NPF6/ACN on CFME show an
increasing current density with each cycle (Epa1 = 0.373 V, Epa2 = 0.524 V,
Epc1 = 0.291 V, Epc2 = 0.484 V at 20th cycle), resulting in the formation of thin film
of conducting polymer on CFME at 100 mV/s and 20 cycle. The cyclic
466 A. Sezai Sarac and A. Gencturk

200
2000
h

2
Current density / mA/cm
1000 a
2
Current density / mA/cm 150

100 -1000
a

-2000 h

50 -0,2 0,0 0,2 0,4 0,6 0,8 1,0 1,2

Potential / V

0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6

Potential / V

Fig. 18.1 Cyclic voltammetry of electrodeposition of monomer at 100 mV s−1, 15 cycle, inset
electrocoated CFME scanned in monomer free electrolyte at 20–400 mV s−1

voltammogram of a polymer coated CFME in monomer free solution with different

scan rates (0.1 M Bu4NPF6/ACN) shows two oxidation peaks (Epa1 = 0.360 V,
Epa2 = 0.483 V) and corresponding reduction peaks (Epc1 = 0.184 V, Epc2 = 0.389 V
(weak)) (inset Fig. 18.1). The current response of the film was directly proportional
to the scan rate indicating a diffusion limited process, showing that the polymer film
is electroactive from 0.0 to 1.6 V and sufficiently ensuring rapid film charging and
discharging.
The capacitive behaviour of the CFME/film was first noticed from CV experi-
ments in a monomer free solution (the inset in Fig. 18.1). The variation of specific
capacitance CCV calculated from CV with respect to potential and the scan rate of
the CFME/ProDOT-Me2/electrolyte system. Capacitance values from CV were
calculated by applying the equation CCV = i/ν ≈ 2 mF (where i: current density
(A) v: scan rate (V/s)) and the current densities obtained from the oxidation and
reduction peaks.

18.3.2 FTIR-ATR Characterisation of PProDOT-Me2

Film on CFME

The electropolymerized thin film structure and doping with the respective anion of
the supporting electrolyte was followed by FTIR-ATR. When the cycle number
increased, the appearance of characteristic peaks of PProDOT-Me2 was observed.
18 Electrochemical Impedance Study … 467

The peak intensities increased with increase in charge which confirms the film
formation of PProDOT-Me2 on CFME [49].
A pattern of five main bands (1470, 1287, 1155, doublet at 1017 and 826 cm−1)
were detected for a PProDOT(Me)2 film coated on CFME during application of
charge in the electrogrowth process (with 100 mV/s−1 scan rate and 5–30 cycles).
The band at 1470 cm−1 (aromatic stretching of C=C bond) and a peak at 1287 cm−1
(in plane deformation of C–H bond at β position) are known to be characteristic
vibrational peaks of polythiophene. Vibrations at 1469–1470, 1295–1286 and
1160–1155 cm−1 originate from the stretching of C-C and C=C bonds in the
thiophene ring.
Further vibrations from the C–S bond in the thiophene ring can be seen at
833–825 cm−1 assigned to C–S stretching. Vibrations at 1017–1048 cm−1 are
assigned to stretching in the alkylenedioxy group [50].

18.3.3 Electrochemical Impedance Spectroscopy

Conducting polymer systems offer opportunies for storing electrochemical energy

largely through their redox capacitance. The essential features of the electro-
chemical behavior of the conducting polymers that make them suitable as elec-
trochemical capacitor materials, the continuous range of states of oxidation that
arise with increasing electrode potential and the reversibility of the faradaic pro-
cesses corresponding to charge withdrawal and reinjection.
The impedance is a complex quantity defined as Z(ω) = v(ω)/I(ω), where v(ω) is
the transform of an alternating potential signal applied to the system and I(ω) is the
transform of the alternating current signal that results. The impedance is a function
of the frequency of the signal. It is most generally represented as Z(ω) = |Z|eθi. Its
magnitude or modulus is |Z|, and its phase angle is θ. Both |Z| and θ may also be
functions of the frequency. As the imaginary contribution increases, the phase angle
increases. For a capacitor, the impedance is purely imaginary, the phase angle is 90o
and the current is out of phase with the voltage by 90°.
However, in a resistor, the impedance is real, the current and the voltage are in
phase, and the phase angle is zero. A large phase angle indicates that the impedance
is predominantly capacitive, while for small angles, it is resistive.
A working electrode/solution interface responds to various perturbations (i.e.
potential steps and potential scans) which are usually of large amplitude, and they
generally drive the working electrode far from equilibrium. Another approach is to
perturb the cell by an alternating (usually sinusoidal) signal of small amplitude
(nominally a few milivolts peak to peak, in our case 10 mV was applied) and
observe the manner in which the system follows the perturbation at steady state.
An ideal double layer capacitance behavior of an electrode material in electro-
chemical supercapacitors is expressed in the form of a rectangular shape of the
voltammetry characteristics, and Bode phase angles close to 90°. Polymer coated
electrodes used in this study show an ideal capacitive line by the application of
468 A. Sezai Sarac and A. Gencturk

electrochemical impedance spectroscopic measurements, indicating fast charge

transfer at the carbon fiber/polymer and polymer/solution interfaces.
The low frequency capacitance values from impedance spectroscopy were
obtained from the slope of a plot of the imaginary component ZIM of the impedance
at low frequencies versus the inverse of the reciprocal frequency f using the fol-
lowing equation [51]

CLF ¼ ð2pfZIM Þ1 ð18:1Þ

18.3.4 Potential Effect on EIS of PProDOT-Me2/CFME

Coated Electrode

Electrochemical impedance measurements have been performed at different scan

rates in the range of 0–400 mV/s and different scan number in the range of 5–40
cycle to find the optimum conditions of coating and also different applied potentials
in ACN solution in the range of −0.1–1.3 V with a potential step of 0.1 V to cover
polymerization range of the ProDOT-Me2 monomer. The specific capacitance
values of PProDOT-Me2 film which were calculated from (Equation 18.1) using
impedance data for frequencies of 10 and 16 mHz with applied DC potentials
between 0.2 and 1.3 V in monomer free electrolyte solutions [18].
Low frequency capacitance increases with increase of diameter of polymer films
(avarage values are calculated from the SEM pictures). Bode phase angle versus
frequency data for PProDOT-Me2/CFME prepared with two different charge den-
sities showed high capacitance. Low frequency capacitance values of PProDOT-
Me2/CFME (CLF) (for the scan rate of 100 mV/s during electrogrowth) were
calculated for the scan numbers of 5, 20 and 40 cycles are 0.29, 1.39 and 12.05
mF/cm2, respectively [18].
Although they are not exactly corresponding to CLF, these results are comparable
with the literature values obtained for PEDOT/Pt (in 0.1 M LiClO4/PC) as 5 mF
(surface area: 0.03 cm2) [52], and for PEDOT/Pt (in 0.1 M KCI/H2O) as 2.11 mF
(surface area: 0.07 cm2) [53].
In this study, electropolymerization was performed at 100 mV/s and 20 cycle
Bu4NPF6/ACN and EIS measurements were performed at Bu4NPF6/PC, and
Bu4NPF6/DMF electrolyte solution and also in Bu4NPF6, Bu4NBF4, NaClO4,
LiClO4/ACN. Capacitive behaviour of PProDOT-Me2 film in PC and DMF solu-
tions were exhibited similar trend with ACN in electrochemical impedance
measurement.
Figure 18.2 shows Nyquist plots, Magnitude of Z, and Bode Phase plots of
coated CFME. Bode phase angle in which the frequency dependence of the system
is clearer compared to Nyquist plots. We can separate the plot into three frequency
regions namely low, medium and high frequency regions. Low frequency region
starts with 10 mHz and finishes with 1 Hz, in this region (except 1.3 V) polymer
18 Electrochemical Impedance Study … 469

100
-0.1 Vmsd
-0.1 Vcal
0.2 Vmsd
80 0.2 V cal
0.4 Vmsd
1000

Phase of angle / Degree

0.4 V cal
1.0 Vmsd
60
1.0 cal
800 1.3 Vmsd
1.3 Vcal

Magnitude of Z / kOhm
40
100

600 20
ZIM / kOhm

0
10
400
0,01 0,1 1 10 100 1000 10000
Frequency / Hz

1
200

0,01 0,1 1 10 100 1000 10000 100000

Frequency / Hz
0

0 200 400 600 800

ZRE / kOhm

Fig. 18.2 Nyquist plots, (middle inset Magnitude of Z, right inset Bode Phase,) EIS
measurements were performed in Bu4NPF6/DMF at applied potentials (ProDOT-Me2 film
deposited at 100 mV/s, 20 cycle in Bu4NPF6/ACN)

films show a maximum peak at 100 mHz with a phase angle of (−87°). Between
1 and 100 Hz, phase angles sharply decrease. Two peaks were observed at applied
potential at 1.3 V from the bode phase plot in Bu4NPF6/DMF, one of them at
100 mHz (−55o) and the other at 1000 Hz (−35°), on the other hand, in our previous
paper only one peak was observed in Bu4NPF6/ACN. From the magnitude of Z
plot, magnitude of impedance exhibits a large drop in the range of 0.01–10 Hz.
From these observations we can conclude that the most capacitive behavior was
obtained for the potential of 0.4 V.

18.3.5 Electrolyte and Solvent Effects on ProDOT-Me2

Coated CFMEs

Figure 18.3 shows variation of low frequency capacitance values of PProDOT-Me2

film in different electrolytes in ACN. CFME/PProDOT-Me2 film has capacitance
values higher in Bu4NPF6 compare to other electrolyte solutions. Cyclic voltammetric
values have exhibited a peak at *0.4 V which is the oxidation peak, similar trend was
also observed for the case of capacitances obtained from EIS measurements.
Figure 18.4 shows variation of low frequency capacitance values of
PProDOT-Me2 film on CFME in different solutions with different polarity.
470 A. Sezai Sarac and A. Gencturk

Fig. 18.3 Variation of low f=0.01 Hz Bu4NPF6 /ACN

frequency capacitance (at 300

Low Frequency Capacitance / mF

Et4NPF6 /ACN
10 mHz) of the Et4NBF4 /ACN
PProDOT-Me2 film which 250 Bu4NBF4 /ACN
was electrodeposited in 5 mM
ProDOT-Me2 at 100 mV/s, 20 200
cycle in 0.1 M Bu4NPF6/ACN
(EIS was performed in
150
Bu4NPF6, Et4NPF6,
Bu4NBF4, Et4NBF4/ACN)
100

50

0
0,0 0,2 0,4 0,6 0,8 1,0 1,2
Potential / V

Fig. 18.4 Variation of low Bu4NPF6/ACN

frequency capacitance (at 300
Low Frequency Capacitance / µ F

Bu4NPF6/PC
10 mHz) of the Bu4NPF6/DMF
PProDOT-Me2 film which 250 f=10mHz
was electrodeposited in 5 mM
ProDOT-Me2 at 100 mV/s, 20
200
cycle in 0.1 M Bu4NPF6/ACN
(EIS was performed in
Bu4NPF6/ACN, Bu4NPF6/ 150
PC, Bu4NPF6/DMF)
100

50

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Potential / V

CFME/PProDOT-Me2 film has capacitance values higher in ACN compared to PC

and DMF. Shape of the capacitance curves at various potentials in ACN, PC and
DMF exhibits similar fashion. The peak which is observed at around 0.4 V in ACN,
PC and DMF solvents exhibits higher specific capacitances by corresponding the
oxidation peak of the polymer (Fig. 18.1).
Figure 18.5 shows the dependence of polarity index and double layer capaci-
tance of the modified electrode. Double layer capacitance (CDL) and low frequency
capacitance (CLF) of the modified electrodes decrease, with the increasing of the
polarity index (P′) (ACN: 5.8, PC: 6.1, DMF: 6.4) [54]. High polarity result a
negative effects on double layer capacitance.
Figure 18.6 shows the dependence of polarity index and double layer capaci-
tance of the electrode and polarization resistance. Double layer capacitance (CDL) of
18 Electrochemical Impedance Study … 471

ACN
300

Low Frequency Capacitance / mF

Double Layer Capacitance/ µF
100

250

80 PC
200

60 DMF
150

5,8 6,0 6,2 6,4

Polarity Index / P'

Fig. 18.5 Dependence of polarity index and double layer capacitance of the electrode and low
frequency capacitance (Filled circle Low frequency capacitance)

110
1400
DMF

Polarization Resistance / kOhm

D ouble Layer Capacitance /µF

100 1300

PC 1200
90
1100

80 1000

900
70
800
ACN
60 700

600
50
5,8 5,9 6,0 6,1 6,2 6,3 6,4
Polarity Index / P'

Fig. 18.6 Dependence of polarity index and double layer capacitance of the electrode and
polarization resistance

the modified electrodes decrease and polarization resistance increases, with the
increasing of the polarity index (P′).
Potential and electrolyte effect on capacitance for other derivatives of Poly
(alkylenedioxy)thiophenes like 3,4-(2,2-dibenzylpropylenedioxy)thiophene and
3,4-(2,2-dibutylpropylenedioxy)thiophene was described in detail in our previous
papers [55, 56]. Bu4NPF6 caused the highest capacitance from the conductive salts
investigation.
472 A. Sezai Sarac and A. Gencturk

18.3.6 Electrical Equivalent Circuit Modeling

The electrochemical parameters of the CFME/PProDOT-Me2/electroyte system

were evaluated using the ZSimpWin (version 3.10) software from Princeton
Applied Research.
Two electrical equivalent circuits were proposed in the simulation of the
impedance behaviour of the film from the experimentally obtained impedance data
in previous paper for the condition of applied of the potential effect [18].
The model (Fig. 18.7) was built using series components; the first one is the bulk
solution resistance of the polymer and the electrolyte RS, the second one the parallel
combination of the double layer capacitance Cdl and R1 is the polarization resis-
tance of the electrolyte. A series connection to R1 was made up using Constant
Phase Element (CPE) in parallel with R2 and W (R2 is the charge transfer and W is
the Warburg impedance of the polymer). χ2 is the function defined as the sum of the
squares of the residuals. The chi squared (χ2) was minimised below 10−4.
Proposed equivalent circuit model was successfully applied to the experimental
data which was performed in monomer free solution in different electrolytes and
solvents to explain the electrical interfacial properties between the carbon fiber
microelectrode, the polymer film and electrolyte in this potential region. Simulation
results are given in Tables 18.1, 18.2 and 18.3 shows the fitting of the results of the

Fig. 18.7 SEM of CFME and ProDOT-(Me)2/CFME a uncoated b coated at 100 mV/s and
equivalent electrical circuit used in simulation
18 Electrochemical Impedance Study … 473

Table 18.1 Dependence of the parameters calculated for PProDOT-Me2 film, at different applied
potential
E = 0.2 V E = 0.4 V E = 0.7 V E = 1.0 V E = 1.3 V
Rs/Ω 685.8 679.8 681.3 683.2 718.9
Cdl/µF 26.2 57.7 49.9 37.3 0.33
R1/Ω 87.8 89.6 894.1 73.7 2157.0
Q/Yo/µS s−n 26.3 69.3 58.8 49.5 46.9
R2/kΩ 627.1 1300.0 904.3 524.5 49.8
W/Yo/µS s−n 0.87 6.03 5.42 119.2 7.051 × 106
Chi squared 8.91 × 10−5 3.40 × 10−5 4.46 × 10−5 6.78 × 10−5 1.15 × 10−4
[(5 mM ProDOT-Me2 was deposited at 100 mV/s, 20 cycle in 0.1 M Bu4NPF6/ACN), (n * 0.96–
0.84)]
EIS measurements were performed in 0.1 M Bu4 NPF6/DMF

Table 18.2 Dependence of the parameters calculated from the model (EIS performed in different
electrolyte solutions)
Bu4NPF6 Et4NPF6 Et4NBF4 Bu4NBF4
Rs/Ω 1852.0 856.3 858.0 350.6
Cdl/µF 102.6 55.9 29.62 120.8
R1/kΩ 47.2 0.71 0.11 0.16
−n
Q/Yo/µS s 48.24 77.83 31.68 70.32
R2/kΩ 631.8 1473.0 6238.0 432.4
W/Yo/µS s−n 5.43 3.18 1.58 368.0
Chi squared 4.14 × 10−5 9.86 × 10−5 4.98 × 10−5 8.47 × 10−5
[(5 mM ProDOT-Me2 deposited electrochemically at 100 mV/s, 20 cycle in 0.1 M Bu4NPF6/
ACN), (n * 0.96–0.93)]

Table 18.3 Dependence of ACN PC DMF

the parameters calculated
from the Model (EIS Rs/Ω 1852 1211 680
performed in different Cdl/µF 102.6 61.05 57.72
electrolyte solutions) R1/kΩ 47.1 0.072 0.089
Q/Yo/µS s−n 48.2 127.6 69.34
R2/kΩ 632 1111 1344
W/Yo/µS s−n 5.43 6.28 6.04
Chi squared 4.14 × 10−5 6.13 × 10−5 3.40 × 10−5
[(5 mM ProDOT-Me2 deposited electrochemically at 100 mV/s,
20 cycle in 0.1 M Bu4NPF6/ACN) (n * 0.96)]

equivalent circuit from 0.2 to 1.3 V for the proposed Model, calculated chi squared
values show that it is smaller than 1 × 10−4, indicating very good agreement
between experimental data and simulation results in the 0.2–1.3 V range
(Table 18.1). Rs was found to be almost constant and it is independent of Edc
474 A. Sezai Sarac and A. Gencturk

(Rs * 680 Ω, for Edc = 0.2–1.3 V). As it was indicated in literature [19] the solution
resistance Rs is not very much dependent on film thickness and Edc. These
experimental results allow describing Rs as the solution resistance.

18.3.7 Morphology of Nanoporous and Compact

Conductive Polymer Coatings on Carbon Fiber

SEM pictures of uncoated carbon fiber and PProDOT-Me2 coated carbon fiber
morphology is given in Fig. 18.7. The uncoated CFME surface was smooth with its
striations, and coated CFME surface exhibits highly porous structure at 100 mV/s.
A large amount of electrolyte solution can gain access to the interior of the film
through these pores. This open structure will allow fast transport of ions in the
supporting electrolyte solution, and as a result, the effective diffusion length of the
ions in the conducting polymer is much shorter than the thickness of the film.
Consequently, this process created the porous morphology, the pores were filled
with the material being deposited (including dopant anion) decreasing the rough-
ness, leading to an increase in impedance.

18.4 Conclusion

Simulation results of CFME/PProDOT-Me2 system showed that electrical circuit

model was successfully applied to the experimental data. From the effect of
potentials on behaviour of capacitance in Bu4NPF6/DMF, polymer film shows
higher capacitance values at 0.4 V compared to other applied potentials. Shape of
the capacitance curves at various potentials in ACN, PC and DMF exhibits similar
fashion but, PProDOT-Me2 film has highest capacitance values for Bu4NPF6 in
ACN compared to different electrolytes and highest capacitance was obtained in
acetonitrile (ACN) compared to PC and DMF. Double layer capacitance (CDL) and
low frequency capacitance (CLF) of the modified electrodes decrease while polar-
ization resistance increases, with the increasing of the polarity index (P′).

References

1. G. Heywang, F. Jonas, Adv. Mater. 4, 116 (1992)

2. F. Jonas, L. Schrader, Synth. Met. 41, 831–836 (1991)
3. K.S. Ryu, Y.G. Lee, Y.S. Hong, Y.J. Park, X. Wu, K.M. Kim, M.G. Kang, N.G. Park, S.H.
Chang, Electrochim. Acta 50, 843–847 (2004)
4. D.M. Welsh, L.J. Kloeppner, L. Madrigal, M.R. Pinto, B.C. Thompson, K.S. Schanze, K.A.
Abboud, D. Powell, J.R. Reynolds, Macromolecules 35, 6517 (2002)
18 Electrochemical Impedance Study … 475

5. A. Cirpan, A.A. Argun, C.R.G. Grenier, B.D. Reeves, J.R. Reynolds, J. Mater. Chem. 13,
2422 (2003)
6. S.P. Mishra, R. Sahoo, A.V. Ambade, A.Q. Contractor, A. Kumar, J. Mater. Chem. 14, 1896
(2004)
7. C.L. Gaupp, D.M. Welsh, J.R. Reynolds, Macromol. Rapid Commun. 23, 885 (2002)
8. B.C. Thompson, P. Schottland, K. Zong, J.R. Reynolds, Chem. Mater. 12, 1563 (2000)
9. C.L. Gaupp, D.M. Welsh, R.D. Rauh, J.R. Reynolds, Chem. Mater. 14, 3964 (2002)
10. D.M. Welsh, A. Kumar, E.W. Meijer, J.R. Reynolds, Adv. Mater. 11, 1379 (1999)
11. I. Schwendeman, J. Hwang, D.M. Welsh, D.B. Tanner, J.R. Reynolds, Adv. Mater. 13, 634
(2001)
12. C. Xu, L. Liu, S. Legenski, M. Le Guilly, M. Taya, A. Weidner, Proc. SPIE-Int. Soc. Opt.
Eng. 4692, 442 (2002)
13. C. Xu, L. Liu, S. Legenski, M. Le Guilly, M. Taya, A. Weidner, Proc. SPIE-Int. Soc. Opt.
Eng. 5051, 404 (2003)
14. C. Xu, L. Liu, S.E. Legenski, D. Ning, M. Taya, J. Mater. Res. 19, 2072 (2004)
15. P.-H. Aubert, A.A. Argun, A. Cirpan, D.B. Tanner, J.R. Reynolds, Chem. Mater. 16, 2386
(2004)
16. A.A. Argun, M. Berard, P.-H. Aubert, J.R. Reynolds, Adv. Mater. 17, 422 (2005)
17. A.A. Argun, J.R. Reynolds, J. Mater. Chem. 15, 1793 (2005)
18. A.S. Sarac, B. Schulz, A. Gencturk, H.D. Gilsing, Surf. Eng. 24, 5 (2008)
19. E. Barsoukov, J.R. Macdonald, Impedance spectroscopy: theory, experiment, and
applications, 2nd edn. (Wiley Interscience Publications, NJ, 2005)
20. W.C. Chen, T.C. Wen, H.S. Teng, Electrochim. Acta 48, 641–649 (2003)
21. T.C. Weng, H.S. Teng, J. Electrochem. Soc. 148, A368–A373 (2001)
22. A. Lewandowski, M. Zajder, E. Frackowiak, F. Beguin, Electrochim. Acta 46, 2777–2780
(2001)
23. A.S. Sarac, J. Springer, Surf. Coat. Technol. 160, 227–238 (2002)
24. A.S. Sarac, U. Evans, M. Serantoni, J. Clohessy, V.J. Cunnane, Surf. Coat. Technol. 182,
7–13 (2004)
25. A.S. Sarac, G. Sonmez, F.C. Cebeci, J. Appl. Electrochem. 33, 295–301 (2003)
26. E. Sezer, A.H.S. Sarac, E.A. Parlak, J. Appl. Electrochem. 33, 1233–1237 (2003)
27. A.S. Sarac, J. Springer, Surface Coating Tech. 160, 227 (2002)
28. A.S. Sarac, U. Ev ans, M. Serantoni, V.J. Cunnane, Carbon 41, 2725 (2003)
29. A.S. Sarac, U. Ev ans, M. Serantoni, J. Clohessy, V.J. Cunnane, Surf. Coat. Tech. 182, 7
(2004)
30. A.S. Sarac, S.A.M.T. ofail, M. Serantoni, J. Henry, V.J. Cunnane, J.B. McMonagle, Appl.
Surf. Sci. 222, 148 (2004)
31. M. Serantoni, A.S. Sarac, D. Sutton, Surf. Coat. Tech. 194, 36 (2005)
32. M. Jamal, A.S. Sarac, E. Magner, Sens. Acts. B 97, 59 (2004)
33. A.S. Sarac, M. Serantoni, S.A.M.T. Ofail, V.J. Cunnane, J. Nanosci. Nanotechnol. 5, 1677
(2005)
34. A.S. Sarac, A. Bismarck, J. Springer, J. Mater. Sci. 41, 389 (2006)
35. A.S. Sarac, Microelec. Eng. 83, 1534 (2006)
36. C.O. Ania, J. Pernak, F. Stefaniak, E. Raymundo-Pinero, F. Beguin, Carbon 44, 3126 (2006)
37. O. Barbieri, M. Hahn, A. Herzog, R. Kötz, Carbon 43, 1303 (2005)
38. A.J. Bard, L.R. Faulkner, Electrochemical methods: fundamentals and applications, 2nd edn.
(John Wiley, New York, 2001)
39. S. Sarangapani, B.V. Tilak, C.P. Chen, J. Electrochem. Soc. 143, 3791 (1996)
40. A.F. Burke, T.C. Murphy, In: D.H. Doughty, B. Vyas, T. Takamura, J.R. Huff (eds.),
Materials for Electrochemical Energy Storage and Conversion: Batteries, Capacitors and Fuel
Cells, Materials Research Society, Pittsburgh, 1995, p. 375
41. E. Frackowiak, F. Beguin, Carbon 39, 937–950 (2001)
42. J.P. Ferraris, M.M. Eissa, I.D. Brotherston, D.C. Loveday, Chem. Mater. 10, 3528–3535
(1998)
476 A. Sezai Sarac and A. Gencturk

43. A.S. Arico`, P. Bruce, B. Scrosati, J.-M. Tarascon, W. Van Schalkwijk, Nature Mater. 4, 366
(2005)
44. R. Kötz, M. Carlen, Electrochim. Acta 45, 2483 (2000)
45. A.G. Pandolfo, A.F. Hollenkamp, J. Power Sources 157, 11 (2006)
46. K. Zong, L. Madrigal, L. Groenendaal, J.R. Reynolds, Chem. Commun. 2498 (2002)
47. B. Sankaran, J.R. Reynolds, Macromolecules 30, 2582 (1997)
48. G. Rauchschwalbe, F. Jonas, EP 1142888, (10.10.2001), CA 135 (2001) 289187e
49. A.S. Sarac, H.D. Gilsing, A. Gencturk, B. Schulz, Progr. Organic Coating 60, 281–236 (2007)
50. S. Trasatti, P. Kurzweil, Platinum Met. Rev. 38, 46 (1994)
51. P. Fiordiponti, G. Pistoia, Electrochim. Acta 34, 215–221 (1989)
52. V. Noel, H. Randriamahazaka, C. Chevrot, J. Electroanal. Chem. 558, 41 (2003)
53. J. Bobacka, A. Lewenstam, A. Ivaska, J. Electroanal. Chem. 489, 17 (2000)
54. R.C. Weast, CRC Handbook of Chemistry and Physics, 1st edn. (1988)
55. A.S. Sarac, S.E. Ozgul, A. Gencturk, B. Schulz, H.-D. Gilsing, H. Faltz, Prog. Organic
Coatings 69, 527–533 (2010)
56. M.C. Turhan, A.S. Sarac, A. Gencturk, H.-D. Gilsing, H. Faltz, B. Schulz, Synthetic Metals
162, 511–515 (2012)
Chapter 19
Application of Nanoporous Zeolites
for the Removal of Ammonium
from Wastewaters: A Review

Mustafa Turan

Abstract This contribution deals with low-cost nanoporous zeolite for the removal
of ammonium from wastewaters. Adsorption processes are widely used to remove
certain classes of pollutants from wastewaters, especially those that are not easily
biodegradable. Natural and synthetic zeolites are widely used as adsorbents for gas
or liquid purification. Natural zeolite is abundant in the world and is usually
regarded as low-cost material. In this review, an extensive list of nanoporous
zeolites used for ammonium removal in the literature has been compiled. The
review evaluates (i) ammonium adsorption capacities and other parameters in batch
systems; (ii) ammonium removal efficiency from wastewaters in fixed-bed systems;
and (iii) a combination of biological treatment and adsorption on zeolite for the
removal of ammonium from wastewaters. It is important to note that the adsorption
capacities of the zeolites presented in this paper vary depending on the character-
istics of zeolite, the extent of chemical modifications and the concentration of
adsorbate.

19.1 Introduction

The ammonium sources, which are municipal, agricultural and industrial, contribute
to the accelerated eutrophication of lakes and rivers, dissolved oxygen depletion
and fish toxicity in receiving water. Complete removal of ammonium from process
or waste effluents is required due to its extreme toxicity to most fish species.
A variety of biological and physicochemical methods and technologies have been
proposed for the removal of ammonium from the environment and industrial water
systems [1–3].

M. Turan (&)
Department of Environmental Engineering, İstanbul Technical University,
34469 İstanbul, Turkey
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 477

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_19
478 M. Turan

Commonly, nitrogen removal is achieved by a biological nitrification–denitri-

fication process, where ammonium (NH4+) is transformed first to nitrite, then to
nitrate, and finally to nitrogen gas [4]. However, since biological methods do not
respond well to shock loads of ammonia, unacceptable peaks over the discharging
levels may frequently appear in the effluent NH4+ concentrations [5]. Besides, the
treatment of ammonium nitrogen from wastewaters of low organic content by a
biological process usually needs to be supplemented with a carbon source, which
may add to the treatment cost. In such a situation, ion-exchange offers an alternative
method for the removal of NH4+.
Sanitary landfilling is one of the main disposal options for municipal waste
throughout the world. Leachate generated from landfill sites is characterized with
high concentrations of organic matter, inorganic salts, heavy metals and ammo-
nium–nitrogen. The discharging of leachate may pose a great threat to the envi-
ronment. Among the leachate constituents, ammonium is the most significant
pollutant in the long term, and its concentration can be found up to several thousand
mg/L [6–8]. The release of high level of ammonium without proper treatment
decreases the performance of conventional biological treatment systems and
accelerates eutrophication, dissolved oxygen depletion and toxicity of living
organisms in surface waters [9]. Therefore, the removal of high ammonium con-
centration from leachate has become a critical issue in recent years. Various
treatment technologies have been applied to remove ammonium from leachate, such
as biological processes [10, 11], electro-oxidation [12, 13], air stripping [8, 9],
chemical precipitation [14, 15], adsorption and ion exchange [6, 16]. Among these
treatment methods, ion exchange using zeolites is very attractive due to its low cost
and relatively simple application.
The adsorption process provides an attractive alternative for the removal of
ammonium from wastewaters while low-cost and readily available adsorbents have
been used for various applications. In the literature, aluminosilicates such as clays,
zeolites, bentonite and sepiolite are the most commonly used low-cost adsorbents
for environmental applications [5, 6, 16]. Zeolites are microporous (pore size
<2 nm) crystalline aluminosilicates that are widely used in separation, filtration,
adsorption and catalysis [17]. The catalytic performance of zeolites is generally
attributed to the existence of a network of micropores with uniform size and shape.
However, micropores can also be detrimental to catalytic reactions by limiting the
diffusion of reagents and/or products throughout the crystals. One possibility to
minimize diffusion limitations is to use zeolite nanocrystals, with a size typically
smaller than 0.5 μm [18, 19]. ZSM-5 with very regular hollow architectures can be
easily obtained from calcined silicalite-1 nanocrystals [20].
Nanoporous zeolites have negative charges that arise due to isomorphous sub-
stitution of Al3+ for Si4+ and its typical unit cell formula is given either as
Na6[(AlO2)6 (SiO2)30] · 24H2O or (Na2, K2, Ca, Mg)3[(AlO2)6(SiO2)30] · 24H2O
[21]. The framework structure may contain linked cages, cavities or channels which
are of the right size to allow small molecules to enter (Fig. 19.1). Three dimensional
crystal structure of zeolite contains two dimensional channels which embody some
ion exchangeable cations such as Na, K, Ca and Mg. These exchangeable cations
19 Application of Nanoporous Zeolites for the Removal … 479

Fig. 19.1 Zeolite structure

Table 19.1 Chemical composition of natural zeolites for comparison

Constituent Turkish Australian Chinese Croatian Turkish Serbia
(%) zeolite zeolite zeolite zeolite zeolite zeolite
[5] [41] [42] [43] [28] [33]
SiO2 74.4 65.02 65.72 64.93 70.27 66.36
Al2O3 11.5 12.76 13.5 13.39 12.90 13.81
CaO 2.00 2.88 3.10 2.00 2.05 3.65
K2O 5.00 1.85 3.14 1.30 1.49 0.99
Fe2O3 1.10 1.82 1.30 2.07 1.38 1.69
MgO 0.50 0.95 0.63 1.08 1.71 1.03
Na2O 0.60 0.50 1.16 2.40 3.21 1.03
TiO2 0.10 0.34 0.212 – 0.19 –
P2O5 0.02 0.10 0.040 – 0.04 –
MnO2 0.001 0.04 0.050 – 0.02 –
Loss on 5.85 11.61 11.12 9.63 6.57 13.40
ignition

give rise to the ion-exchange properties of the material [22]. Such sorptive prop-
erties have been utilized for a variety of purposes such as removal of ammonium [5,
23–28], heavy metal [29–33] and dye [34–39].
Various natural zeolites around the world have shown varying ion-exchange
capacities for cations such as ammonium and heavy metal ions. Some zeolites also
show adsorption of anions and organics from aqueous solution. Modification of
natural zeolites can be done by several methods such as acid treatment, ion
exchange and surfactant functionalisation, making the modified zeolites achieving
higher adsorption capacity for organics and anions [40]. Table 19.1 presents the
chemical analysis of natural zeolites for comparison.
This review (i) presents batch adsorption studies, where adsorption equilibrium,
kinetic models and thermodynamic parameters are considered; (ii) evaluates
480 M. Turan

column adsorption studies with adsorption column design parameters and bed depth
service time (BDST) model; (iii) examines modification of natural zeolites and
synthesized zeolites as well as regeneration process; (iv) describes ammonium
removal efficiency and other parameters for zeolites in the batch and fixed-bed
column systems; (v) investigates ammonium removal efficiency for wastewater
treatment systems combined with zeolites.

19.2 Methodology

19.2.1 Batch Adsorption Studies

19.2.1.1 Equilibrium Isotherms

Sorption studies were performed by batch technique to obtain rate and equilibrium
data. The removal efficiency (%) of adsorbent (zeolite) and the adsorption capacity
of solution (ammonium) (qe, mg/g) was calculated, respectively, as follows;

Removal efficiency ð%Þ ¼ ðCi  Ce Þ=Ci  100 ð19:1Þ

qe ¼ ðCi  Ce ÞV=m ð19:2Þ

where Ci and Ce are the initial and equilibrium solution concentrations (mg/L),
respectively, V the volume of the solutions (L) and m the weight of adsorbent used
(g). The widely used Langmuir isotherm [44] has found successful application in
many real sorption processes and the linear form of Langmuir model is expressed
as;

Ce =qe ¼ 1=kb þ Ce =b ð19:3Þ

where Ce is the concentration of adsorbate in equilibrium solution (mg/L) and the

value of b (mg/g) and k (L/mg) are maximum adsorption capacity of adsorbent and
the adsorption energy coefficient, respectively. The well-known Freundlich iso-
therm [45] used for isothermal adsorption is a special case for heterogeneous sur-
face energy, in which the energy term of the Langmuir equation varies as a function
of surface coverage strictly due to the variation of the sorption. The linear form of
Freundlich model is given as;

log qe ¼ log Kf þ ð1=nÞ log Ce ð19:4Þ

where Kf is roughly an indicator of the adsorption capacity and n is the adsorption

intensity. The magnitude of the exponent 1/n gives an indication of the favorability
of adsorption. Values, n > 1 represent favorable adsorption condition.
19 Application of Nanoporous Zeolites for the Removal … 481

19.2.1.2 Kinetics and Thermodynamics

In order to investigate the mechanism of adsorption kinetics, the pseudo-first-order

adsorption, the pseudo-second-order adsorption and the intraparticle diffusion
model were used to test dynamical experimental data. The first-order rate expres-
sion of Lagergren [46] is expressed as follows;

logðqe qÞ ¼ log qe k1 t=2:303 ð19:5Þ

where qe and q are the amounts of dye adsorbed on adsorbent (mg/g) at equilibrium
and at time t, respectively and k1 is the rate constant of first-order adsorption
(1/min). In many cases the first-order equation of Lagergren does not fit well to the
whole range of contact time and is generally applicable over the initial stage of the
adsorption processes [47]. The second-order kinetic model [48] is defined as;


t=q ¼ 1= k2 q2e t=qe ð19:6Þ

and

h ¼ k2 q2e ð19:7Þ

where k2 (g/mg min) is the rate constant of second-order adsorption and h is the
initial adsorption rate (mg/g min). The intraparticle diffusion equation can be
described as;

q ¼ ki t0:5 ð19:8Þ

where ki is intraparticle diffusion rate constant (mg/g min−0.5). The

pseudo-second-order rate constant of solute adsorption is expressed as a function of
temperature by the Arrhenius type relationship;

ln k ¼ ln A  Ea =RT ð19:9Þ

where Ea is the Arrhenius activation energy of sorption, representing the minimum

energy that reactants must have for the reaction to proceed, A is the Arrhenius
factor, R is the gas constant and is equal to 8.314 J/(mol K) and T is the solution
temperature.
The thermodynamic parameters such as change in free energy (ΔGo), enthalpy
(ΔHo) and entropy (ΔSo) were determined using the following equations [49, 50];

KC ¼ Ce =CAe ð19:10Þ

DGo ¼ RT lnKC ð19:11Þ

482 M. Turan

logKC ¼ DSo =ð2:303RÞ  DHo =ð2:303RTÞ ð19:12Þ

where KC is the equilibrium constant, CAe is the amount of adsorbate (g) adsorbed
on the adsorbent per dm3 of the solution at equilibrium, Ce is the equilibrium
concentration (g/dm3) of the dye in the solution, T is the solution temperature
(K) and R is the gas constant. Generally, the absolute magnitude of the change in
free energy for physisorption is between −20 and 0 kJ/mol; chemisorption has a
range of −80 to −400 kJ/mol [51]. The negative values of the standard enthalpy
change (ΔH°) indicate that the interaction of the adsorbate with adsorbent is
exothermic in nature.

19.2.2 Column Adsorption Studies

19.2.2.1 Column Design Parameters

Adsorption performance of the fixed-bed can be evaluated while bed volumes

(BV) at breakthrough point (C/C0 = 0.1) is maximum. The breakthrough curves
were constructed by plotting the normalized effluent concentration (C/C0) versus
time (t) and/or bed volumes (BV). The bed volumes, BV, and the empty bed contact
time, EBCT, at the fixed-bed column are defined as follows;

BV ¼ VF =VR ¼ Qt =VR ð19:13Þ

EBCT ¼ VR =Q ð19:14Þ

where VF is the total volume of wastewater treated during the adsorption process at
time t (L), VR is the fixed-bed volume of zeolite (L), C0 is the influent concentration
(mg/L), C is the effluent concentration at time t (mg/L), Q is the feed flow rate in the
fixed-bed (L/min) and t is the service time (min). The formation and the movement
of the adsorption zone can be evaluated numerically [52, 53]. The time required for
the adsorption zone to become established and move completely out of the bed at
exhaust time is;
tExh ¼ VExh =Q ¼ BVExh EBCT ð19:15Þ

The rate at which the adsorption zone (Uz) is moving up or down through the bed is;

Uz ¼ hz =tz ¼ h=ðtExh  tf Þ ð19:16Þ

From (19.16), the height of the adsorption zone (hz) is obtained;

hz ¼ htz =ðtExh  tf Þ ð19:17Þ

19 Application of Nanoporous Zeolites for the Removal … 483

where VExh is the total volume of wastewater treated in the zeolite column at
exhaust time (L), hz is the height of adsorption zone (cm), h is the total bed height
(cm) and tf is the time required for the adsorption zone to initially form (min). The tf
value can be found as follows;

tf ¼ ð1  FÞtz ð19:18Þ

At breakthrough, the fraction (F) of adsorbent present in the adsorption zone still
possessing the ability to remove solute is;
Z VExh
F ¼ Sz =Smax ¼ ðC0  CÞdV=C0 ðVExh  Vb Þ ð19:19Þ
Vb

where Vb is the total volume of the wastewater treated to the breakthrough point
(L), Sz is the amount of solute that has been removed by the adsorption zone from
breakthrough to exhaustion; and Smax is the amount of solute removed by the
adsorption zone when completely exhausted. The percentage of the total column
saturated at breakthrough point is

% saturation ¼ ½1 þ ðF  1Þhz =h  100 ð19:20Þ

19.2.2.2 Bed Depth Service Time (BDST) Model

The objective of fixed-bed operations is to reduce the concentration in the effluent so

that it does not exceed a specific breakthrough concentration (Cb). The original work
on the BDST model was carried out by Bohart and Adams, who proposed a rela-
tionship between bed depth, Z, and the time taken for breakthrough to occur [54].
The service time, t, is related to the process conditions and operating parameters;
 
lnðC0 =Cb  1Þ ¼ ln ekadN0Z=V  1  kad C0 t ð19:21Þ

Hutchins [55] proposed a linear relationship between the bed depth and service
time;

t ¼ N0 Z=ðC0 VÞ  ð1=kad C0 ÞlnðC0 =Cb  1Þ ð19:22Þ

where Cb is the solute concentration at breakthrough point (mg/L), kad is the

adsorption rate constant (L/(mg h)), N0 is the adsorption capacity (mg of solute/L of
adsorbent), Z is the bed depth of column (cm), V is the linear flow velocity of feed
to bed (cm/h) and t is the service time of column (h). The critical bed depth, Z0, is
the minimum column height necessary to produce an effluent concentration Cb. By
letting t = 0, Z0 is obtained from (19.22)
484 M. Turan

Z0 ¼ ðV=kad N0 ÞlnðC0 =Cb  1Þ ð19:23Þ

Equation (19.22) enables the service time, t, of an adsorption bed to be deter-

mined by a specified bed depth, Z, of adsorbent. t and Z are correlated with the
process parameters and the influent concentration of solute, solution flow-rate and
the adsorption capacity [56] (19.22) can be expressed in the following form:

t ¼ mZ þ n ð19:24aÞ

where

m ¼ slope ¼ N0 =C0 V ð19:24bÞ

n ¼ intercept ¼ ð1=kad C0 ÞlnðC0 =Cb  1Þ ð19:24cÞ

19.2.3 Modification of Zeolite

Several modification methods have been tried to enhance the ammonium adsorption
capacity of zeolite, including hydrothermal pretreatment, NaOH, HCl or NaCl
solution treatment, integrated calcination, and others.
Sodium-natural zeolite (SNZ): The optimum modification conditions for Yemeni
natural zeolite on the effectiveness of ammonium ion (NH4+) removal was inves-
tigated [57]. The natural zeolite (NZ) powder (200–230 mesh) was washed with
distilled water to remove undesirable materials and dried at a temperature of 100 °C
for 24 h and then modified with sodium chloride. The suspension was stirred in
500 ml conical flasks using a magnetic stirrer water bath at a rate of 120 rpm and
90 °C for 2 h. Subsequently, the suspension was filtered and washed with distilled
water. The wet modified material was dried at 100 °C in an oven for 24 h. In a
similar study [58], natural zeolite was also modified with NaCl to investigate the
mechanisms for ammonium adsorption onto zeolite.
Silicate-Carbon Modified Zeolite (SCMZ): The Silicate-Carbon Modified Zeolite
(SCMZ) was used for removing ammonium from drinking water [59]. The modi-
fication process applied in the preparation of SCMZ consisted of three steps:
(1) The clinoptilolite powder (74 μm) was repeatedly washed with tap water and
then dried at 100 °C for 4 h. (2) The zeolite was dried and 1000 g of the dry zeolite
was dispersed into 1 L sodium chloride solution. The sample was repeatedly
washed with tap water and then dried at 100 °C for 4 h. Finally, the dried sample
was ground and sieved to 74 μm again and designated as sample A. (3) Sample A,
Na2SiO3 and powdered activated carbon were mixed at weight/weight ratio of
100:9:2, and mixed evenly; then, 10 % tap water was added and stirred again. The
mixture was shaped into a cylinder by an extrusion method and then dried at 100 °C
for 2 h and calcined in a muffled furnace at 500 °C for 2 h and the SCMZ filter was
obtained.
19 Application of Nanoporous Zeolites for the Removal … 485

Calcium formed zeolite (CaY): A modified clinoptilolite zeolite-Ca2+-formed

(CaY) was prepared for the removal of ammonium ions from aqueous solutions
[60]. The chosen clinoptilolite was ground and sieved into 0.425–0.970 mm particle
size and then washed twice with distilled water (volume ratio of liquid/solid 3:1).
NaY was gained by treating the samples three times with saturated NaCl solution at
boiling point for 2 h and then changed into NH4+-formed clinoptilolite (NH4Y) by
washing it with 1 mol/l NH4Cl solution for 20 min at room temperature. Finally,
CaY was prepared through washing the NH4Y with Ca(OH)2 solution at boiling
point for 2–3 h.
Integrated modification process: a natural calcium-rich zeolite modified by an
integrated process was used for ammonium ion uptake (AIU) from drinking water
[42]. This process consisted of pretreatment (grinding and sieving), sodium salt
modification and calcination. The introduction of sodium ion enhanced the heat
resistance of the raw material from 150 to 400 °C and Na+ ion treatment followed
by calcination effectively improved pore and surface properties of zeolite.

19.2.4 Synthesized Zeolites

Due to its uniform micropore structure and high surface area, synthetic zeolite
usually has a higher adsorption capability compared with natural zeolite. That’s
why many researchers have investigated the synthesis of zeolite from low-cost raw
materials like coal fly ash, red mud, fly ash, halloysite etc. in order to obtain
low-cost and effective synthetic zeolites.
Zeolite synthesized from halloysite mineral: The adsorption behavior of
ammonium ions (NH4+) from aqueous solution onto synthesized NaA zeolite was
investigated [61]. NaA zeolite was prepared using alkaline fusion followed by
hydrothermal treatment method. 2 g of halloysite powder was placed in a Ni
crucible and fused with 2.6 g NaOH powder at 673 K for 2 h to promote their
reactivity. The fused mass obtained was cooled and mixed with deionized water
(38 mL) by magnetic stirring for 30 min until the reaction gel was homogenized.
The aluminosilicate gel was aged for 2 h at 313 K in sealed Teflon reactor to
rearrange the reactant for forming nuclei. After that, the mixture was crystallized at
363 K for 5 h under static condition. Solid powder was filtered-off, washed and
dried at 383 K for 12 h and the product was obtained. In a similar study [62], the
removal of ammonium from wastewater by zeolite synthesized from halloysite was
also investigated.
Zeolite synthesized from fly ash by a fusion method: A fusion method, involving
alkaline fusion followed by hydrothermal treatment, was adopted for the synthesis
of zeolite [63, 64]. 9 g of fly ash were mixed with NaOH powder to obtain a
homogeneous mixture. The ratio of fly ash to NaOH was 1:1.3 (w/w). The mixture
was then heated in a nickel crucible in air at 600 °C for 90 min. The fusion products
were ground and poured into a plastic bottle, followed by the addition of distilled
water to form a mixture that contains 17.25 % of the fusion product (w/w). The
486 M. Turan

mixture was stirred intensely at 80 °C for 1 h to form an aluminosilicate gel and

then poured into a stainless alloy autoclave and kept in oven at 100 °C for 9 h. After
hydrothermal treatment, the precipitated samples were extracted and washed with
distilled water until the pH of the samples was about 9. The samples were dried at
100 °C for 12 h and sieved through a 100 mesh (<150 μm). In another study [65],
the simultaneous removal of ammonium and phosphate by zeolite synthesized from
fly ash (ZFA) and saturated with different cations was investigated.
Integrated synthesis of zeolites 4A and Na–P1 using coal fly ash: Fly ash from
Brazilian coal combustion was used for green integrated synthesis of zeolite 4A and

Closed
Reactor
NaOH (3.0 mol/L) Coal fly ash UB
24 h, 100 oC

Filtred with extracted

silicon
Filtering and
Washing

Drying 2 h, 100 oC Closed Metallic aluminum

Reactor (NaOH 3.0 mol/L)

4 h,
90-95 oC

Zeolite Na-P1 Effluent Synthesis

110 mL

Filtering

Cleaning up pH 10

Drying 2 h, 100 oC

Zeolite 4 A

Fig. 19.2 Flow diagram of the zeolites synthesis 4A and Na–P1 by integrated process 1 [66]
19 Application of Nanoporous Zeolites for the Removal … 487

Na–P1 [66]. Integrated synthesis of zeolites 4A and Na–P1 was carried out using
two different pathways, denominated processes 1 and 2. In process 1 (Fig. 19.2) ash
was activated by NaOH solution (3.0 mol/L and L/S ratio of 6 L/kg) using the
conventional hydrothermal process in a closed borosilicate reactor at 100 °C for
24 h. The resulting product was separated by filtration and washed several times
with deionized water to remove excess sodium and lower the alkaline content, then
dried (100 °C for 2 h) and stored in a desiccator. The molar ratio was adjusted to 1.0
(for zeolite 4A formation) by adding an alkaline solution (3.0 mol/L) containing Al,
previously dissolved from aluminum metal powder. The resulting solution was
transferred to a borosilicate glass reactor, stirred manually (2 min at 25 °C) and
immediately heated at 90 °C for 1.5 h, followed by another 2.5 h at 95 °C.
Zeolite synthesized from red mud: The removal of ammonium from aqueous
using zeolite synthesized from red mud was studied [67]. Red mud was initially
sieved and the particles smaller than 100 mesh were kept for zeolite synthesis. The
hydrothermal method was adopted for the synthesis of zeolite. Extra silicon
(Na2SiO3 · 9H2O, Nona hydrate of sodium silicate) was added to make the total
mole ratio Si/Al = 4.5, and a NaOH solution was added to make ratio Na/Si = 1.3.
The mixture was then heated in a homogeneous reactor for an 8 h aging time, and a
16 h crystallization time at a temperature of 120 °C. After applying the above
treatment, the precipitated samples were extracted and washed with distilled water
until the pH of the samples was about 9. Finally, the samples were dried at 100 °C
for 12 h for further studies. In order to reduce the negative influence of high Fe2O3
content (40.96 %) of red mud to zeolite synthesis, high alkalinity conditions were
used for zeolite synthesis in this study.

19.2.5 Regeneration

The ammonium-removal efficiency of zeolite decreases after operating for a long

duration. Thus, zeolite should be regenerated before using it again. There are
various regeneration methods such as heating regeneration, acid regeneration, NaCl
regeneration and Ca(OH)2 regeneration (Table 19.2). The breakthrough zeolite

Table 19.2 Methods of zeolite regeneration [59]

Regenerate Step of regeneration
methods
Heating The breakthrough zeolite was calcined in a muffle furnace at 200 °C for
regeneration 24 h
Acid The breakthrough zeolite was treated with 1.0 mol/L HCl solution for 24 h,
regeneration repeatedly washed with tap water, and dried at 45 °C for 4 h
NaCl The breakthrough zeolite was treated with 2.0 mol/L NaCl solution for
regeneration 24 h, repeatedly washed with tap water and dried at 100 °C for 4 h
Ca(OH)2 Clinoptilolite saturated with ammonium ions was regenerated by 0.2 mol/l
regeneration Ca(OH)2 emulsion at 100–110 °C
488 M. Turan

regenerated using NaCl maintained its initial adsorption capacity. The following
reaction occurred in the process of NaCl regeneration:

NaCl þ NH4  Zeolite , NH4 Cl þ Na4  Zeolite ð19:25Þ

In the process of regeneration, Na+ exchanged the NH4+ that was adsorbed on
zeolite, and the ammonium-removing property of zeolite was restored [59].
A consideration of practical interest with regard to the application of a new
exchanger is its desorption and regeneration ability. The Ca(OH)2 regeneration
happened as follows:
þ
NH4ðZÞ þ CaðOHÞ2 ! Ca2ðZÞþ þ NH4 OH ð19:26aÞ

NH4 OH ! NH3 " þ H2 O ð19:26bÞ

This operation went on with generating free ammonia, so some ammoniated

by-products (such as NH4Cl) can be gained. The mass of the zeolite during the
regeneration and the adsorption capacity of the regeneration modified zeolite for
ammonium ions after regeneration were measured, respectively. The column was
regenerated with Ca(OH)2 emulsion at pH 11–12 and at flow velocity of 5 ml/min [60].

19.3 Nanoporous Zeolites for Ammonium Removal

from Wastewaters

19.3.1 Ammonium Adsorption Capacities and Other

Parameters for Zeolites in the Batch Systems

An overview of nanoporous zeolites for the removal of ammonium is presented in

this article along with their adsorption capacities and other parameters and the
experimental conditions used in batch systems (Table 19.3). In 2004, the ion uptake
of ammonium ion onto New Zealand clinoptilolite and mordenite are compared
[27]. The results showed that the mordenite exhibited higher overall uptake con-
centrations at equilibrium compared to clinoptilolite having solution concentrations
greater than 80 mg/L. The ion exchange capacity of New Zealand clinoptilolite for
NH4+ ion was found as 6.59 mg/g.
In 2006, the ion exchange kinetics and equilibrium isotherms of ammonium ion
on the natural Turkish zeolite (clinoptilolite) was studied [5]. Ammonium
adsorption capacity increased from 1 to 4.75 mg/g with the increase of initial NH4+
concentration (Fig. 19.3). For lower initial concentration of NH4+, equilibrium time
was lower than that of the higher concentration due to the increased competition for
the active sites with the increase of NH4+ concentration. This is consistent with the
ion-exchange surface becoming increasingly saturated with ammonium ion [68].
Table 19.3 Ammonium adsorption capacities and other parameters for natural and modified zeolites
19

Adsorbent Wastewater dP (mm)/Sp(m2/g) qm (mg/g) Ci (mg/L) t (min) pH References

Zeolite (Iranian) Contaminated surfacewater 49.7 10–100 5–90 2–10 [75]
Synthesized zeo. NaA-H Synthetic wastewater 0.0005–0.001 44.3 10–300 120 7 [61]
Acid treat. zeo. (Azerbaijan) Synthetic wastewater 43.47 10–200 2–120 4–9 [84]
Synthesized zeo.Y-H Synthetic wastewater 0.001–2/79.62 38.64 10–300 60 6–7 [62]
Synthesized zeolite FA Domestic wastewater <0.15 37.45 10–300 40 8 [63]
Zeolite (Jordanian) Synthetic wastewater 0.5–1 37.175 15–50 2880 5.5–9 [90]
Zeolite (Zeolith, Deutch.e.K.) Synthetic wastewater 0.013 36 10–200 60 6–8.5 [76]
Synthesized zeolite (ZFA) Simulated wastewater 80 mesh 35.31 2.5–1200 1440 4.98–6 [65]
Zeolite (USA, TR, EU) Synthetic wastewater 0.355–2.50 22.8–32.4 5–500 250 7.5 [89]
Synthesized zeo. FA Na–PI Swine wastewater 0.003–0.005 31–4 1205–125 30 7.11;10.7 [66]
Zeolite (New Zealand) Simulated wastewater 29 25–200 5760 [69]
Zeolite NZ, BZ, BAZ (Chinese) Synthetic wastewater 1–15/6.64 18.3–28.9 16.5 5400 [79]
Synthe. zeo. FA LC-Z, HC-Z Synthetic wastewater <0.15 28.65–6.12 10–300 0–360 8 [64]
Modified zeolite MWZ (Chinese) Synthetic wastewater 0.083 23.83 0–200 360 6.5 [85]
Acid teated zeolite (Turkey) Municipal wastewater 1–2 23.70 8.8–40.3 5–120 4–7 [28]
Application of Nanoporous Zeolites for the Removal …

Zeolite (Croatian) Synthetic wastewater 0.1–0.2 21.6 25.7–700 250 3–8 [70]
Modified zeolite (Japan) Synthetic wastewater 0.001–25/20–115 12.5–19.5 400 1440 [82]
Zeolite/modified zeo. (Bulgarian) Synthetic wastewater 0.2–1/26 7.85–18.4 50–250 60 2.5–9 [83]
Zeolite (Iranian) Synthetic wastewater 0.5–2.3/22 17.8 10–220 180 5.5–7.6 [88]
Synthesized zeolite RM Synthetic wastewater 0.007; 0.01 17.5 5–500 30 2–11 [67]
Zeolite/modif. Zeo. (Chinese) Synthetic wastewater 1.43;0.80/14; 60.8 11.0–17.3 10–4000 15–1440 3.4–11.1 [58]
Zeolite/modified zeo. (Chinese) Synthetic wastewater 2;1/10.53–30.1 4.04–17.24 25 120 7.1 [42]
Zeolite (Turkey) Synthetic wastewater 0.3–0.6 15.4 5–150 720 6.5 [72]
Zeolite (Iranian) Synthetic wastewater 0.6–0.8 11.31–13.27 90–3610 10–300 [78]
(continued)
489
Table 19.3 (continued)
490

Adsorbent Wastewater dP (mm)/Sp(m2/g) qm (mg/g) Ci (mg/L) t (min) pH References

Zeolite (Romania) Synthetic wastewater 0.2–0.4/12–35.7 13.15–11.85 70 1440 7 [91]
Zeolite (Turkey) Synthetic wastewater 0.3–0.6 11.25 20–100 5–50 4–10 [71]
Zeolite (Iranian) Synthetic wastewater 0.6–0.8 11.31–13.27 90–3610 10–300 [78]
Zeolite (Romania) Synthetic wastewater 0.2–0.4/12–35.7 13.15–11.85 70 1440 7 [91]
Zeolite (Turkey) Synthetic wastewater 0.3–0.6 11.25 20–100 5–50 4–10 [71]
Zeolite (Turkey) Municipal wastewater 0.5–1 10.4 20 1440 7.3 [94]
Zeolite (Turkey) Synthetic wastewater 0.075 9.64 20–400 1440 8 [39]
Synthesized zeo.-Na–PI Acid mine drainage 0.038–0.25 9.6 30 3.3 [86]
Zeolite (Chinese) Synthetic wastewater 0.074–0.3 9.41 50–300 300 4–9 [81]
Modified zeo.-CaY (Chinese) Simulated wastewater 0.425–0.970 3.03–7.67 300 120 4–9 [60]
Zeolite (Turkey) Contaminat. Surf. water 1–2 and 2–2.8 7.4 1–1400 60 6.5 [26]
Zeo. (New Zealand) Simulated wastewater 0.5–0.71 6.59 10–200 4320 7.5 [27]
Zeolite (Australian) Greywater 6.30 5–50 480 2–10 [41]
Zeolite (Chinese) Landfill leachate 1–2 5.87–6.02 2000–2300 480 3–9.9 [92]
Zeolite (Turkey) Synthetic wastewater 1–1.4/11.8 5.95 25–150 60 7 [5]
Zeolite (New Zealand) Wastewater 0.25;2.83/30–40 5.74–5.72 15–750 1020 5.4–8.2 [87]
Zeolite (USA) Synthetic wastewater 2–4/521 3.18–4.5 0–500 1440 [93]
Zeolite (Croatian) Leakage waters WD 0.5–2.0 4.2 800 15–360 7.3 [43]
Zeolite (Hungarian) Synthetic wastewater 0–0.125/1500 3.79 50–500 30–180 6–7 [77]
Zeolite/modif. zeolite (Yemeni) Synthetic wastewater 0.008;64/36; 41 2.83; 3.68 10–250 20–120 7 [57]
H halloysite; FA fly ash; RM red mud; Na-ZFA zeolite synthesized from fly ash; LC-Z low-calcium zeolite; HC-Z high calcium zeolite; Na-PI zeolite
synthesized from fly ash; Zeolite Y zeolite synthesized from halloysite; NZ natural fresh zeolite; BZ bio-zeolite made from natural fresh zeolite; BAZ bio-zeolite
made from ammonium-bearing zeolite; MWZ microwave-treated zeolite; WD waste dump
M. Turan
19 Application of Nanoporous Zeolites for the Removal … 491

Fig. 19.3 Effect of contact time and initial concentration on the uptake of NH4+ by natural zeolite
(T = 40 °C, solid concentration = 10 g/L, agitation rate, 200 rpm) [5]

NH4+–N removal increased in the range of 8.8–40.3 mg NH4+–N/L and also

decreased after 40.3 mg NH4+–N/L. This is attributed to the weak binding between
ion exchanger material (zeolite) and NH4+ ion and can be expected from zeolite
having micropores and macropores [69]. The maximum amount of exchanged
NH4+ using natural Turkish clinoptilolite was 5.95 mg/g at 25 °C.
The ability of a Croatian clinoptilolite- and montmorillonite-rich tuffs to take up
ammonium from aqueous solutions was investigated [70]. The clinoptilolite effi-
ciency for ammonium removal was far superior (by 45–50 %) to the removal effi-
ciency of montmorillonite. Maximum loadings in ammonium exchange in
clinoptilolite tuff, montmorillonite tuff, alkaline modified montmorillonite tuff and
acid modified montmorillonite tuff, as calculated by the Langmuir equation, are 1.20,
0.54, 0.64, and 0.21 mmol/g NH4+, respectively. Doğan et al. [71] investigated the
parameters affecting on the equilibrium of ammonium exchange by zeolite and the
applicability of linear and nonlinear regression for the estimation of kinetics and
isotherm models. In a similar study, Gunay [72] evaluated the experimental data of
ammonium exchange by natural Bigadic clinoptilolite using nonlinear regression
analysis. Thermodynamic investigation indicated that ammonium removal by
clinoptilolite was favorable at lower temperatures and exothermic in nature.
The effect of shaking time on the removal of NH4+ ion using natural zeolite was
investigated [39]. Removal efficiency of NH4+ ion increased with increasing
shaking time and 75 % of NH4+ ion removal was completed within 15 min. The
removal efficiency increased by raising the pH from 4 to 8 and reached a maximum
value at pH 8 and then decreased. The decrease in the NH4+ removal efficiency after
pH 8 may be due to two factors: (i) partial dissolution of the natural zeolite [73];
(ii) the conversion probability of NH4+ into NH3 [74]. Moussavi et al. [75]
examined the applicability of natural zeolite for simultaneous removal of ammonia
and humic acid, two of the most encountered contaminants in the surface waters.
492 M. Turan

Fig. 19.4 The effect of

temperature on NH4+ ion
removal using the natural
zeolite [39]

The maximum experimental adsorption capacities of ammonia and humic acid as

binary components were 49.7 and 10.5 mg/g, respectively.
The effect of temperature on ammonium adsorption was studied at 25, 40 and 55 °
C [75] and at 21, 30, 40, and 50 °C [39]. The results indicate that the amount of NH4+
exchanged onto zeolite increases with a decrease in temperature as seen in Fig. 19.4,
which may be due to a tendency for the ammonium molecules to escape from the solid
phase to the bulk phase with an increase in temperature of the solution. This result
also indicated that the adsorption process of NH4+ onto natural zeolite was
exothermic in nature. In 2009, removal of both ammonium and phosphorus nutrients
by natural zeolite has been studied in lab scale by using a mechanically stirred batch
system [76]. Results indicated that the removal efficiency and the uptake of ammo-
nium by zeolite are both dependent on the initial ammonium and zeolite concentra-
tions, indicating rise in uptake of ammonium by zeolite for higher ratios. Widiastuti
et al. [41] also investigated the effectiveness of removing ammonium ion and the
theoretical aspects of adsorption as well as desorption–regeneration studies. Natural
zeolite shows good performance with up to 97 % of ammonium removal. Gibbs free
energy change (ΔG°) at 25 °C is negative (−19.52 kJ/mol) indicating the spontaneous
nature of the adsorption process. The enthalpy change (ΔH°) is positive
(30.96 kJ/mol) indicating endothermic adsorption process. The entropy change (ΔS°)
at 25 °C is also positive (0.169 kJ/mol K) indicating increasing randomness at the
solid-solution interface during adsorption.
In another study, the adsorption capacity of clinoptilolite zeolite towards
ammonium ions was evaluated [77]. The highest adsorption capacity of clinop-
tilolite towards ammonium ions was found 3.79 mg/g for the initial concentration of
300 mg/L. The highest removal level reached by clinoptilolite for ammonium was
99.74 % for the initial concentration of ammonia of 50 mg/L. Malekian et al. [78]
studied to remove ammonium (NH4+) from aqueous solutions with different Na+
19 Application of Nanoporous Zeolites for the Removal … 493

concentrations by natural Iranian zeolite. The maximum NH4+ exchange capacity

by the zeolite at equilibrium was measured at 0.03 M Na+ concentration, which was
11.31 and 13.27 mgNH4+/g for millimeter- and nanometer-sized zeolites, respec-
tively. Wen et al. [79] reported the sorption kinetics of ammonium onto three types
of zeolite, where he found that ion exchange increased with decreasing particle size
of the zeolite. The capacity sequence was found to be NZ > BZ > BAZ while larger
particle size (8–15 mm) was used (Table 19.3). Compared to the virgin zeolite, the
ion exchange rate in the bio-zeolite was reduced about 25–30 % and the
rate-controlling step for ion exchange shifted from pore diffusion in the virgin
zeolite to film diffusion in the bio-zeolite [80].
In 2010, the removal of ammonium ion from aqueous solutions using natural
Chinese (Chende) zeolite was studied [81]. The effect of other cations on the
removal of ammonium followed the order of preference Na+ > K+ > Ca2+ > Mg2+ at
identical mass concentrations and the effect of the presence of individual anions
followed the order of preference carbonate > chloride > sulfate > phosphate at
identical mass concentrations of ammonium ions. The maximum amount of NH4+
adsorbed by the zeolite at equilibrium (M) was 9.41 mg/g in the individual presence
of phosphate and 5.05 mg/g in the individual presence of carbonate.
Chemical modifications of zeolite were also conducted to improve its removal
performance and adsorption capacity for ammonium ions. In 2009, it was studied to
modify the natural clinoptilolite zeolite available locally (Akita Prefecture, Japan)
for its ammonium ions retention capacity [82]. The natural clinoptilolite was mod-
ified chemically and mechanically with changing time duration of sodium hydroxide
treatment and ball to powder mass ratio in wet ball milling, respectively. The
maximum NH4+ retention capacities of natural clinoptilolite (NZeo), clinoptilolite
treated with NaOH solution for 72 h (Zeo-72) and wet-milled clinoptilolite (WM-50)
according to Langmuir model were 0.89, 1.15 and 1.39 mmol/g, respectively.
Vassileva and Voikova [83] investigated ammonium ions removal from aqueous
solutions by natural and NaCl pretreated Bulgarian clinoptilolite. The maximum
adsorption capacities for ammonium ions shown by natural and pretreated clinop-
tilolites (CL Na and CL Na(t)) were 7.85, 12.29 and 18.40 mg/g, respectively.
In a another study, Lin et al. [58] investigated the mechanisms of ammonium
adsorption onto natural zeolite and NaCl-modified counterpart under different
ammonium levels (10–4000 mg-N/L) and initial pH 3.4–11.1. Alshameri et al. [57]
investigated the potential use of Yemeni (Al-Ahyuq) natural zeolite on the effec-
tiveness of ammonium ion removal in a batch system. The highest ion exchange
capacity was obtained at 35 °C and pH of 8. In 2014, Khosravi et al. [84] studied
the removal of ammonium ion from aqueous solutions using the natural Western
Azerbaijan zeolite. NaCl treatment of the adsorbent showed an improvement of
ammonium ion removal from aqueous solutions over other treatments. Maximum
uptake for ammonium ion was 43.47 mg/g.
The development of physical and chemical properties of a natural calcium-rich
zeolite modified by an integrated process, as well as the relation between the
development and ammonium ion uptake (AIU), was studied [42]. By the optimum
modification process, NH4+ ion exchange capacity was largely enhanced and the
494 M. Turan

maximum AIU (17.24 mg/g) of the modified zeolite was 4.3 times that of the raw
zeolite (4.04 mg/g). In 2008, natural Chinese zeolite pretreated by sodium chloride
solution under microwave irradiation was used to remove ammonium from aqueous
solution [85]. The microwave-treated zeolite had the highest ammonium adsorption
capacity, followed by the thermal-treated zeolite and the fresh zeolite.
Many researchers have reported zeolite synthesis using halloysite mineral, fly ash
and red mud etc. for ammonium ion removal. In 2006, Wu et al. [65] studied Na-ZFA
(zeolite synthesized from fly ash) converted into Ca-, Mg-, Al-, and Fe-ZFA by salt
treatment and the simultaneous removal of ammonium and phosphate by ZFA sat-
urated with different cations. It was shown that Al3+-ZFA had the highest removal
efficiencies (80–98 %) for ammonium. It was concluded that ZFA could be used in
simultaneous removal of NH4+ and phosphate at low concentrations with pre satu-
ration by an appropriate cation such as Al3+ through salt treatment. Zhao et al. [62]
investigated ammonium adsorption properties of the synthesized zeolite Y using
batch experiments. Due to its low cost, high adsorption capacity and fast adsorption
rate, the zeolite Y synthesized from halloysite has the potential to be utilized for the
cost-effective removal of ammonium from wastewater. The maximum adsorption
capacity is 38.64 mg/g at 25 °C. In another similar research, adsorption behavior of
ammonium ions (NH4+) from aqueous solution onto NaA zeolite was investigated
[61]. A maximum adsorption capacity of 44.3 mg/g of NH4+ was achieved.
Zhang et al. [63] investigated the removal of ammonium from aqueous solutions
using zeolite synthesized from fly ash by a fusion method, which combines alkaline
fusion followed by hydrothermal treatment. The maximum ammonium uptake
capacity obtained is 24.3 mg/g. The regenerated zeolite has almost the same
ammonium uptake capacity as the original zeolite. In another similar study [64],
zeolites were synthesized from low-calcium (LC-Z) and high-calcium (HC-Z) fly
ashes, respectively. The obtained maximum ammonium uptake capacities are
23.8 mg/g for LC-Z and 3.17 mg/g for HC-Z in the synthetic solution. The Ca2+
leaching and the lower zeolite content in HC-Z account for its lower uptake
capacity. In 2015 [66], Brazilian coal fly ashes were used for integrated synthesis of
zeolites 4A and Na–P1 by two different routes and under moderate operating
conditions (temperature and pressure). Treatment of swine wastewater with zeolite
Na–P1 resulted in a high removal capacity for total ammoniacal nitrogen (31 mg/g).
In another study [86], the zeolite synthesized in the optimized process (77 % Na–
P1) was used to remove contaminants from acid mine drainage (AMD). Treatment of
this wastewater with 10 g/L of synthetic zeolite for 30 min presented high removal of
arsenic, nickel, calcium, copper, iron and manganese ions and the partial yet signifi-
cant removal of ammonium, magnesium, potassium and zinc ions. Zhao et al. [67] also
studied the removal of ammonium from aqueous solutions using zeolite synthesized
from red mud. The CEC increased from 81.9 to 111 mmol/100 g during the synthesis
process. The maximum ammonium adsorption capacity obtained is 17.5 mg/g.
19 Application of Nanoporous Zeolites for the Removal … 495

19.3.2 Ammonium Removal from Wastewaters in the Batch

and Column Systems

The removal of ammonium ions from wastewater using natural zeolite was carried
out in both batch and fixed bed (column) systems in the literature. In 2005, Sarioğlu
[28] studied the ammonium ion adsorption onto Doğantepe (Turkey) zeolite.
Increasing initial ammonium nitrogen concentration from 5.0 to 12.0 mg/L,
increased the exchange capacity from 0.70 to 1.08 mg NH4+–N/g (Table 19.3). The
flowrate and pH values, at which the highest adsorption capacities were obtained,
were found to be 0.5 ml/min (0.87 mg NH4+–N/g zeolite) and 4 ml/min (mg NH4+–
N/L), respectively (Table 19.4). The cation exchange capacity of Dogantepe zeolite
was found to be 164.62 meq/100 g.
In 2003, the ammonia removal from contaminated drinking water using
clinoptilolite zeolite was studied in bottle adsorption tests and optimum adsorption
conditions were established [26]. In the fixed bed studies, the breakthrough curves
showed that increasing the influent concentration (10–20 mg/L) increased the
effluent concentration. Twice-regenerated clinoptilolite showed the best perfor-
mance compared with the others. The regeneration of clinoptilolite with 30 g/L
NaCl solution at pH 11.5 was found to have activated the clinoptilolite and
increased the ion exchange performance.
Nguyen and Tanner [87] evaluated the capacity of two natural New Zealand
zeolites (clinoptilolite and mordenite) to remove NH4+ from a range of wastewaters
under both batch and flow through conditions. The maximum NH4+ removal
capacity of the New Zealand zeolites studied (5.7–8.3 gNH4–N/kg) is within the
range (5–15 g NH4–N/kg) reported for overseas zeolites. Ashrafizadeh et al. [88]
studied the capability of Iranian natural clinoptilolite for ammonia removal from
aqueous solutions. In the batch system, an adsorption capacity about 17.8 mgNH4+/
g at initial ammonia concentration of 50 mg/L was obtained The adsorption
capacity of clinoptilolite in the continuous mode was about 15.16 and 15.36 mg
NH4+/g for the original and regenerated types of clinoptilolites, respectively.
In 2008, removal of low-concentration ammonium by eight commercial natural
zeolites was studied [89]. In the batch tests, the ammonium adsorption capacity (up
to 32.4 mg NH4þ =g) decreased with increasing particle size (355–2500 µm). In the
column tests, the adsorption capacity of chabazite was 48.3 mg NH4+/g. The
zeolites were rapidly regenerated by an alkaline sodium chloride solution fed to the
columns. Mashal et al. [90] investigated the performance of natural Jordanian
zeolite tuff to remove ammonia from aqueous solutions. Breakthrough time
increased by increasing the bed depth as well as decreasing zeolite particle size,
solution flow-rate, initial NH4+ concentration and pH. In 2011, zeolitic volcanic
tuffs (ZVT) originated from Transylvania, Romania were used to remove ammo-
nium ions from model solution [91]. The ionic exchange process was realized in
batch and fixed bed column conditions. It was found that ionic exchange efficiency
varies according to their zeolite content.
Table 19.4 Ammonium removal efficiency and other parameters for zeolites in fixed-bed systems
496

Adsorbent Wastewater dp (mm) Ci Q (mL/min) Removal BVb(BV)/tb (h) References

(mg/L) (%)
Zeolite (Turkey) Contaminated 1–2 and 2–2.8 10–20 70.2 90 275 and 140 [26]
drinkingwater
Zeolite (Turkey) Municipal wastewater 1–2 5–12 4, 2, 1 and 0.5 80 /4, 11, 25 and 56 [28]
Zeolite (Croatian) Leakage water 0.5–2 820 25 100 4.5 [43]
Zeolite (Turkey) Landfill leachate 0.6–1.2 100–400 6.8–21 90 29.11–22.47 [16]
Zeolite (Turkey) Landfill leachate 0.6–1.2 200 21, 11 and 6.8 90 /7, 30 and 65 [6]
Zeolite (Chinese) Landfill leachate 1–2 2057 4–9 69.4–31.6 /3 [92]
Zeolite (Chinese) Synthetic wastewater 0.25–0.70 50 2 100 /22 [95]
Zeo. (New Zealand) Synthetic wastewater 0.25–2.83 5–15 0.47; 15.9 100 60 [87]
Modif. zeo. CaY Simulated wastewater 0.425–0.970 250 10–70 100 78 [60]
(Chinese)
Zeolite (Iranian) Synthetic wastewater 0.25–2 49 54 BV/h 100 200 [88]
Zeo. (USA, TR, EU) Synthetic wastewater 0.5–0.7 4.5 5 100 1400 [89]
SCMZ (Chinese) Contaminated 0.074 2.5–10 8–12 m/h 90 /47 [59]
drinkingwater
Zeolite (USA) Synthetic wastewater 0.6–2.8 300 2–3 90 /100 [93]
Zeo. (Ukraine)/NaA Simulated wastewater 0.71–1 27–580 28.8–29 and 31.7–31 89.8–97.9 25–217 and 35– [96]
BV/h 272
Zeo. (Jordanian) Synthetic wastewater 0.5–0.71 15–50 100–250 90 /1.5;0.4 [90]
Zeolite (Turkey) Municipal wastewater 0.5–1; 1-2-2-4 20 0.3 m/h 100 /54, 66, and 78 [94]
Zeolite (Ukraine) Contaminated 0.3–0.6;0.6– 2 5 m/h 70–89; 54– [97]
drinkingwater 1.5 94
Zeo. (Azarbaijan) Synthetic wastewater 0.5–2.0 1–50 1–3 99.43 [99]
Zeolite (Romania) Synthetic wastewater 0.2–0.4 70 3.3 97–100 /0.5–3.7 [91]
SCMZ silicate-carbon modified zeolite (SCMZ); NaA zeolite synthesized from fly ash; tb breakthrough time; BVb bed volumes (BV) at breakthrough point
M. Turan
19 Application of Nanoporous Zeolites for the Removal … 497

Various researches have been conducted to remove ammonium from landfill

leachates. In 2008, Karadağ et al. [6] studied the removal of ammonium ion (NH4+)
from sanitary landfill leachate (Istanbul, Turkey) using zeolite (clinoptilolite) in
both batch and column studies. The competitive effect was observed to be more
effective in lower ammonium concentration and higher total cation concentrations.
Conditioning and regeneration using NaCl improved removal efficiency of
ammonium from leachate. In a similar study, Ye et al. [92] investigated the removal
of ammonium from landfill leachate using natural zeolite. In the batch studies, the
ammonium removal efficiency increased with an increase in zeolite dosage from 25
to 150 g/L and an increase in stirring speed from 200 to 250 r/min. The optimal pH
for the removal of ammonium was found as 7.1. In the column studies, the total
ammonium removal percentage during 180 min operation time decreased with the
flow rate though the ion-exchange capacity varied to a very small extent with the
flow rate ranging from 4 to 9 mL/min.
Ji et al. [60] investigated the removal of ammonium ions from aqueous solution
using a modified clinoptilolite-Ca2+-formed clinoptilolite (CaY) prepared from
natural clinoptilolite. The batch results show that the CaY has a high selectivity for
NH4+ and the exchange decreases with increasing temperature. The column results
indicated that the effluent of simulated wastewater treated with CaY could meet the
integrated wastewater discharge standard of China. Cyrus and Reddy [93] inves-
tigated the efficiency of zeolite in the removal of NH4+–N from swine wastewater
and desorption of the sorbed NH4+–N from the zeolite saturated with N. Column
studies indicated that, columns with flow rates of 2 and 3 mL/min, reached the
breakthrough point within about 100 h. In 2013, a three-column filtration system
was designed and operated for the investigation of ammonium and suspended solids
removal from wastewater by using natural zeolite [94]. The optimum results were
obtained as 100 and 38 % ammonium and suspended solids removals, respectively
for 0.5 mm diameter during an operation period of 78 h.

19.3.3 Ammonium Removal from Wastewaters in the Fixed

Bed Systems

Various researches have been done on zeolite fixed bed column in recent years.
Wang et al. [95] studied the removal of ammonia nitrogen with a low concentration
(50 mg/L) from aqueous solution in a zeolite column. They found that the
wastewater with 50 mg/L ammonia nitrogen took 7 h to flow through a modified
zeolite (10 g) column having diameters of 24–64 meshes at a flow rate of 2 mL/min.
The saturation extent of adsorption was up to 7.95 mg/g and the saturation
adsorption time was 22 h (Table 19.4). Malovanyy et al. [96] studied the con-
centration of ammonium from municipal wastewater using ion exchange process. In
total 23 runs of saturation and regeneration, synthetic wastewaters of different kinds
and pretreated municipal wastewater were used.
498 M. Turan

Experimental investigation of ammonium ion removal from drinking water was

carried out using natural zeolite (clinoptilolite) [97]. The efficiency fluctuated from
89 to 70 % (0.3–0.6 mm particle size) and from 94 to 54 % (0.6–1.5 mm particle
size). In 2011, clinoptilolite zeolite was processed, first, by sodium chloride solu-
tion and then mixed with Na2SiO3 and powdered activated carbon. The mixture
went through extrusion forming and calcination to prepare the Silicate-Carbon
Modified Zeolite (SCMZ) that was used for removing ammonium from drinking
water using column system [98]. The adsorption capacity of SCMZ at a solution pH
of 7, filtration rate of 10 m/h and the initial ammonium concentration of 5 mg/L was
0.115 mg NH4+–N/g, which was much higher than that of natural zeolite. SCMZ
filtering media would exhibit high ammonium-removal efficiency when regenerated
by three cycles with NaCl solution.
Farkas et al. [43] investigated the treatment of leakage waters from waste dumps
using activated carbon and natural zeolite clinoptilolite. Leakage waters were
characterized by a high content of ammonium (820 mg/L) and organic pollutants
(1033 mg/L C). Addition of activated carbon, regardless of its mass, increased the
exchange of ammonium ions to only 5.7 mg NH4+/g zeolite. In the model solution
having the same ammonium concentration as the real solution, 17.70 mg NH4+/g
zeolite was exchanged. The zeolite fixed bed was used to remove ammonium from
landfill leachate [16]. Higher effluent volumes and removal rates were produced by
lower ammonium concentrations and increased expansion ratios in the
fluidized-bed column reduced the treatment efficiency. Combining ammonia
stripping and an upflow fixed-bed system is feasible in reducing higher ammonium
concentrations to satisfactory levels. In 2012, Khosravi et al. [99] studied the
removal of ammonium ion from wastewater using natural Western Azerbaijan
zeolite. The maximum removal was achieved at initial ammonium concentration of
30 mg/L, flow rate of 1 mL/min and pH 6.

19.3.4 Ammonium Removal for Wastewater Treatment

Systems Combined with Zeolites

The influence of combined systems with zeolites for wastewater treatment was
investigated by a number of researchers. Turan et al. [100] studied a combined
anaerobic fluidized bed and zeolite fixed bed system in sanitary landfill leachate
treatment. While excellent ammonia removal (>90 %) was obtained with the
untreated zeolite, the regenerated zeolites showed higher performance. This com-
bined anaerobic and adsorption system is an effective tool to remove high COD and
high ammonia in landfill leachate (Table 19.5).
Multifunctional permeable reactive barrier (multibarrier), combining microbial
degradation and abiotic ion exchange processes for removal of ammonium from
landfill leachate was studied [101]. Feeding the columns with leachate deprived
from ammonium resulted in a complete release of the previously sorbed ammonium
 19

Table 19.5 Ammonium removal efficiency for wastewater treatment systems combined with zeolites
Adsorbent Wastewater dp (mm) Ci (mg/L) Q (mL/min) Removal (%) BVb(BV)/tb (h) pH References
AFB/Zeo. (Turkey) Landfill leach./AFB 0.5 410 9.03 90 /480 [100]
PRB/Nitrif/Zeolite Landfill leachate 1.0–2.0 244–260 0.07 ± 0.02 97.5–96.8 18; 13.5 and 41 7.6 [101]
PRB/Zeo. (Chinese) Contaminated Groundwater 1–2 0.3–36.6 24 m/h 60 /7200 6.4– [102]
8
BAF/Zeo. (Chinese) Textile wastewater 2.4–4 8 4 35.1 [103]
C/F-F/Powder zeo.4A Landfill leachate 0.003 1644 3.94 91 /17 7.74 [104]
BA/Silver-zeolite Synthetic wastewater 0.002–0.003 5.04 20–37 0.25–2 [105]
NBC/Zeo. (Chilean) Synthetic wastewater 1 707.1 70 /240–288 7.2 [106]
AnMBRs/Zeo.USA Wastewater/AnMBR permeate 8 × 40 mesh 31 4–8 BV/h 95 /48–24 7.2 [107]
PRB/Zeo. (Chinese) Contaminated Groundwater 3–5 1488.7 0.5 70.8 /200 7.9 [108]
Application of Nanoporous Zeolites for the Removal …

CW/Zeo. (Chinese) Contaminated surfacewater 1.21– 0.4 m3/h 83–94 /35.4 8.4 [109]
3.70
CW/Zeo. (Chinese) Synthetic wastewater 0.27–0.45 100– 97.1–72.0 /1–72 6 [110]
1500
AFB anaerobic fluidized bed; PRB permeable reactive barrier; BAF biological aerated filter; C/F-F coagulation/flocculation-Fenton; AnMBRs anaerobic
membrane bioreactors; CW constructed wetland; BA bactericidal activity; NBC nitrifying biomass carrier
499
500 M. Turan

from the clinoptilolite due to exchange with metal cations present in the leachate. Li
et al. [102] studied to remove NH4+–N from contaminated groundwater in a
pilot-scale zeolite permeable reactive barrier (PRB). After some 300 days of
operation, about 40 % of the influent NH4+–N was converted to nitrite and nitrate
through nitrification; where zeolite contributed to ∼60 % of the observed removal of
NH4+–N through adsorption.
Ammonium nitrogen removal characteristics of zeolite media in Biological
Aerated Filter (BAF) process for treatment of textile wastewater was studied [103].
Several biofilters were conducted to compare the performance of natural zeolite
with sand and granular activated carbon (GAC) as packing media. The removal of
NH4+–N by ion exchange, nitrification and biological synthesis was estimated as
35.1, 40.2 and 22.3 %, respectively. Turki et al. [104] examined the application of
an integrated technique that consisted of coagulation/flocculation-Fenton-powder
zeolite adsorption methods for the treatment of raw leachates from a landfill site of
Bizerte (Northern Tunisia). The removal efficiency of ammonia was 91 %. In 2012,
the antimicrobial activity of silver-zeolite against Escherichia coli, Vibrio harveyi,
Vibrio cholerae and Vibrio parahaemolyticus was examined in liquid medium
[105]. Silver-zeolite removed 20–37 % ammonia from aqueous solutions.
A batch nitrification process was studied using synthetic wastewater as substrate
and Chilean natural zeolite as biomass carrier at ambient temperatures (20 °C)
[106]. Ammonium N removals higher than 70 % were obtained in experimental sets
II and III when zeolite was used. The third experimental set (III) revealed that the
highest N consumption took place between days 11 and 16 of digestion. Deng et al.
[107] found that ammonium exchange of natural zeolite could be an economical
method of nitrogen removal from the permeates of anaerobic membrane bioreactors
(AnMBRs). A significant bottleneck of zeolite processes would be the requirement
of substituting exhausted zeolite with virgin one due to the reductions in ammo-
nium exchange capacity and regeneration efficiency (RE).
The optimum composition of permeable reactive barrier (PRB) materials for
remediating groundwater heavily contaminated by landfill leachate was investigated
[108]. In column tests, various mixtures of zero-valent iron (ZVI), zeolite (Zeo) and
activated carbon (AC) with 0.01–0.25, 3.0–5.0 and 0.7–1.0 mm grain sizes,
respectively were used. The optimum weight ratio of the three PRB materials for
removing the contaminants and maintaining adequate hydraulic conductivity was
found to be 5:1:4.
A strip-like wetland was constructed in the riparian zone for investigation of
ammonium nitrogen (NH3–N) removal in the Peach River [109]. An
oxygen-deficient condition with dissolved oxygen of 0.87–1.60 mg/L, was
observed in the zeolite layer, which benefits anaerobic ammonium oxidation
(anammox) bacteria survival. The intensified test driven by artificial pumping
shows that average removal rates of NH3–N and TN are 41.6 mg/m3d and
63.2 mg/m3d, respectively. In a similar study [110], the kinetic adsorption of
ammonium nitrogen (NH4+–N) by CW substrate materials such as blast furnace
19 Application of Nanoporous Zeolites for the Removal … 501

slag (BFS), zeolite, ceramsite, vermiculite, gravel, paddy soil, red soil and turf, was
investigated using batch experiments and kinetic adsorption isotherms. The maxi-
mum adsorption capacity of NH4+–N, estimated as 33.333 g/kg for zeolite.

19.4 Conclusions

In this review, the recent development of nanoporous zeolites in water and

wastewater treatment has been studied worldwide for ammonium removal. The
properties and modification of natural zeolite are discussed. Modification of natural
zeolites can be done in several methods such as hydrothermal pretreatment, NaOH,
HCl or NaCl solution treatment, making the modified zeolites achieve higher
adsorption capacity for the removal of ammonium ions. Due to uniform micropore
structure and high surface area, synthetic zeolite usually has a higher adsorption
capability compared with natural zeolite. Therefore, in order to obtain low-cost and
effective synthetic zeolite, many researchers have investigated the synthesis of
zeolite from coal fly ash, red mud, fly ash, halloysite etc. An overview of nano-
porous zeolites for the removal of ammonium is presented in this article along with
their adsorption capacities and other parameters and the experimental conditions
used in batch systems. Various researches on the zeolite fixed bed columns have
been done in recent years. The influence of combined systems with zeolites for
wastewater treatment was investigated by a number of researchers. Thus, the use of
zeolite as a low-cost adsorbent may contribute to the sustainability of the envi-
ronment. Undoubtedly nanoporous zeolites offer a lot of promising benefits for
commercial purposes in the future.

References

1. M. Rozic, S. Cerjan-Stefanovic, S. Kurajica, V. Vancina, E. Hodzic, Water Res. 34, 3675

(2000)
2. M.S. Celik, B. Ozdemir, M. Turan, I. Koyuncu, G. Atesok, H.Z. Sarikaya, Water Sci.
Technol. Wat. Supply 1, 81 (2001)
3. L. Lin, G.Y.S. Chan, B.L. Jiang, C.Y. Lan, Waste Manage. 27, 1376 (2007)
4. A. Thornton, P. Pearce, S.A. Parsons, Water Res. 41, 433 (2007)
5. D. Karadag, Y. Koc, M. Turan, B. Armagan, J. Hazard. Mater. 136, 604 (2006)
6. D. Karadag, S. Tok, E. Akgul, M. Turan, M. Ozturk, A. Demir, J. Hazard. Mater. 153, 60
(2008)
7. H. Zhang, X.G. Wu, X.W. Li, Chem. Eng. J. 210, 188 (2012)
8. F.M. Ferraz, J. Povinelli, E.M. Vieira, Environ. Technol. 34, 2317 (2013)
9. J.C. Campos, D. Moura, A.P. Costa, L. Yokoyama, F.V.D. Araujo, M.C. Cammarota, L.
Cardillo, J. Environ. Sci. Health A 48, 1105 (2013)
10. H. Alvarez-Vazquez, B. Jefferson, S.J. Judd, J. Chem. Technol. Biotechnol. 79, 1043 (2004)
11. J.L. Gao, M. Chys, W. Audenaert, Y.L. He, S.W.H. Van Hulle, Environ. Technol. 35, 1226
(2014)
502 M. Turan

12. A. Cabeza, A. Urtiaga, M.J. Rivero, I. Ortiz, J. Hazard. Mater. 144, 715 (2007)
13. H. Zhang, Y.L. Li, X.G. Wu, Y.J. Zhang, D.B. Zhang, Waste Manage. 30, 2096 (2010)
14. D. Kim, H.D. Ryu, M.S. Kim, J. Kim, S.I. Lee, J. Hazard. Mater. 146, 81 (2007)
15. T. Zhang, L.L. Ding, H.Q. Ren, J. Hazard. Mater. 166, 911 (2009)
16. D. Karadag, E. Akkaya, A. Demir, A. Saral, M. Turan, M. Ozturk, Ind. Eng. Chem. Res. 47,
9552 (2008)
17. A. Corma, Chem. Rev. 95, 559 (1995)
18. C.J.H. Jacobsen, C. Madsen, T.V.W. Janssens, H.J. Jakobsen, J. Skibsted, Micropor.
Mesopor. Mater. 39, 393 (2000)
19. L. Tosheva, V.P. Valtchev, Chem. Mater. 17, 2494 (2005)
20. Y. Wang, A. Tuel, Micropor. Mesopor. Mater. 113, 286 (2008)
21. G.M. Haggerty, R.S. Bowman, Environ. Sci. Technol. 28, 452 (1994)
22. L.L. Ames, Am. Mineral. 45, 689 (1960)
23. S.E. Jorgensen, Water Res. 10, 213 (1975)
24. M. Gaspard, A. Neveu, G. Matin, Wate Res. 17, 279 (1983)
25. I. Rodriguez-Iznaga, A. Gomez, G. Rodriguez-Fuentes, A. Benitez-Aguilar,
J. Serrano-Ballan, Micropor. Mesopor. Mater. 53, 71 (2002)
26. M. Turan, M.S. Celik, J. Water Supply Res. Technol. Aqua 52, 159 (2003)
27. L.R. Weatherley, N.D. Miladinovic, Water Res. 38, 4305 (2004)
28. M. Sarioglu, Separ. Purif. Technol. 41, 1 (2005)
29. V.J. Inglezakis, M.D. Loizidou, H.P. Grigoropoulou, Water Res. 36, 2784 (2002)
30. E. Erdem, N. Karapinar, R. Donat, J. Colloid Interf. Sci. 280, 309 (2004)
31. M. Turan, U. Mart, B. Yüksel, M.S. Çelik, Chemosphere 60, 1487 (2005)
32. M.K. Doula, Water Res. 40, 3167 (2006)
33. N.V. Medvidovic, J. Peric, M. Trgo et al., Micropor. Mesopor. Mater. 105, 298 (2007)
34. B. Armagan, O. Ozdemir, M.S. Celik, M. Turan, J. Chem. Technol. Biotechnol. 78, 725
(2003)
35. O. Ozdemir, B. Armagan, M. Turan, M.S. Celik, Dyes Pigments 62, 49 (2004)
36. D. Karadag, M. Turan, E. Akgul, S. Tok, A. Faki, J. Chem. Eng. Data 52, 1615 (2007)
37. D. Karadag, E. Akgul, S. Tok, F. Erturk, M.A. Kaya, M. Turan, J. Chem. Eng. Data 52, 2436
(2007)
38. J. Torres-Perez, M. Solache-Rios, M.T. Olguin, Separ. Sci. Technol. 42, 299 (2007)
39. K. Saltalı, A. Sarı, M. Aydın, J. Hazard. Mater. 141, 258 (2007)
40. S. Wang, Y. Peng, Chem. Eng. J. 156, 11 (2010)
41. N. Widiastuti, H.W. Wu, H.M. Ang, D.K. Zhang, Desalination 277, 15 (2011)
42. Z. Liang, J. Ni, J. Hazard. Mater. 166, 52 (2009)
43. A. Farkas, M. Rozic, Z. Barbaric-Mikocevic, J. Hazard. Mater. 117, 25 (2005)
44. I. Langmuir, J. Am. Chem. Soc. 40, 1361 (1918)
45. H. Freundlich, Z. Phys, Chemei 57, 384 (1906)
46. S. Lagergren, Kungliga Svenska Vetenskapsaka-demiens. Handlingar 24, 1 (1898)
47. G. McKay, Y.S. Ho, Water Res. 33, 578 (1999)
48. G. McKay, Y.S. Ho, Process Biochem. 34, 451 (1999)
49. G.C. Catena, F.V. Bright, Anal. Chem. 61, 905 (1989)
50. C. Namasivayam, R.T. Yamuna, Environ. Pollut. 89, 1 (1995)
51. H. Nollet, M. Roels, P. Lutgen, P. Van der Meeren, W. Verstraete, Chemosphere 53, 655
(2003)
52. L.D. Benefield, J.F. Judkins Jr, B.L. Weand, Process Chemistry for Water and Wastewater
Treatment (Prentice Hall, Englewood Cliffs, 1982)
53. S. Kundu, A.K. Gupta, J. Colloid Interface Sci. 290, 52 (2005)
54. G.S. Bohart, E.Q. Adams, J. Am. Chem. Soc. 42, 523 (1920)
55. R.A. Hutchins, Chem. Eng. 80, 133 (1973)
56. G.M. Walker, L.R. Weatherley, Water Res. 31, 2093 (1997)
57. A. Alshameri, A. Ibrahim, A.M. Assabri, X. Lei, H. Wang, C. Yan, Powder Technol. 258, 20
(2014)
19 Application of Nanoporous Zeolites for the Removal … 503

58. L. Lin, Z. Lei, L. Wang, X. Liu, Y. Zhang, C. Wan, D.-J. Lee, J.H. Tay, Separ. Purif.
Technol. 103, 15 (2013)
59. M. Li, X. Zhu, F. Zhu, G. Ren, G. Cao, L. Song, Desalination 271, 295 (2011)
60. Z.-Y. Ji, J.-S. Yuan, X.-G. Li, J. Hazard. Mater. 141, 483 (2007)
61. Y. Zhao, B. Zhang, X. Zhang, J. Wang, J. Liu, R. Chen, J. Hazard. Mater. 178, 658 (2010)
62. Y. Zhao, B. Zhang, Y. Zhang, J. Wang, J. Liu, R. Chen, Separ. Sci. Technol. 45, 1066 (2010)
63. M. Zhang, H. Zhang, D. Xu, L. Han, D. Niu, B. Tian, J. Zhang, L. Zhang, W. Wu,
Desalination 271, 111 (2011)
64. M. Zhang, H. Zhang, D. Xu, L. Han, D. Niu, L. Zhang, W. Wu, B. Tian, Desalination 277,
46 (2011)
65. D. Wu, B. Zhang, C. Li, Z. Zhang, H. Kong, J. Colloid Interf. Sci. 304, 300 (2006)
66. A.M. Cardoso, M.B. Horn, L.S. Ferret, C.M.N. Azevedo, M. Pires, J. Hazard. Mater. 287, 69
(2015)
67. Y. Zhao, Y. Niu, X. Hu, B. Xi, X. Peng, W. Liu, W. Guan, L. Wang, Desalin. Water Treat. 1,
1 (2015)
68. N.P. Hankins, S. Pliankarom, N. Hilal, Separ. Sci. Technol. 39, 3639 (2004)
69. T.C. Jorgensen, L.R. Weatherley, Water Res. 37, 1723 (2003)
70. M. Rozic, S. Cerjan-Stefanovic, L. Curkovic, Croat. Chem. Acta 75, 255 (2002)
71. D. Karadag, Y. Koc, M. Turan, M. Ozturk, J. Hazard. Mater. 144, 432 (2007)
72. A. Gunay, J. Hazard. Mater. 148, 708 (2007)
73. J.B. Dixon, S.B. Weed (eds.), Mineral Soil Environment (SSSA Pub. Inc., Madison, WI,
1989), pp. 874–911
74. E. Maranon, M. Ulmanu, Y. Fernandez, I. Anger, L. Castrillon, J. Hazard. Mater. 137, 1402
(2006)
75. G. Moussavi, S. Talebi, M. Farrokhi, R.M. Sabouti, Chem. Eng. J. 171, 1159 (2011)
76. N. Karapınar, J. Hazard. Mater. 170, 1186 (2009)
77. M. Zabochnicka-ŚWiątek, K. Mali Ńska, Global NEST J. 12, 256 (2010)
78. R. Malekian, J. Abedi-Koupai, S.S. Eslamian, S.F. Mousavi, K.C. Abbaspour, M. Afyuni,
Appl. Clay Sci. 51, 323 (2011)
79. D. Wen, Y.-S. Ho, X. Tang, J. Hazard. Mater. 133, 252 (2006)
80. O. Lahav, M. Green, Water SA 26, 51 (2000)
81. H.M. Huang, X.M. Xiao, B. Yan, L.P. Yang, J. Hazard. Mater. 175, 247 (2010)
82. V.K. Jha, S. Hayashi, J. Hazard. Mater. 169, 29 (2009)
83. P. Vassileva, D. Voikova, J. Hazard. Mater. 170, 948 (2009)
84. A. Khosravi, M. Esmhosseini, S. Khezri, Res. Chem. Intermed. 40, 2905 (2014)
85. L. Lei, X. Li, X. Zhang, Separ. Purif. Technol. 58, 359 (2008)
86. A.M. Cardoso, Fuel 139, 59 (2015)
87. M.L. Nguyen, C.C. Tanner, New Zealand J. Agricul. Res. 41, 427 (1998)
88. S.N. Ashrafizadeh, Z. Khorasani, M. Gorjiara, Separ. Sci. Technol. 43, 960 (2008)
89. J. Langwaldt, Separ. Sci. Technol. 43, 2166 (2008)
90. A. Mashal, J. Abu-Dahrieh, A.A. Ahmed, L. Oyedele, N. Haimour, A. Al-Haj-Ali, D.
Rooney, World J. Sci. Technol. Sustain. Develop. 11, 144 (2014)
91. A. Maicaneanu, H. Bedelean, S. Burca, M. Stanca, Separ. Sci. Technol. 46, 1621 (2011)
92. Z. Ye, J. Wang, L. Sun, D. Zhang, H. Zhang, Environ. Technol. 1, 1 (2014)
93. J.S. Cyrus, G.B. Reddy, J. Environ. Sci. Health A 46, 408 (2011)
94. D. Alkas, B. Beler, Baykal, C. Kinaci. Desalin. Water Treat. 51, 3047 (2013)
95. H. Wang, H. Gui, W. Yang, D. Li, W. Tan, M. Yang, C.J. Barrow, Desalin. Water Treat. 52,
753 (2014)
96. A. Malovanyy, H. Sakalova, Y. Yatchyshyn, E. Plaza, M. Malovanyy, Desalination 329, 93
(2013)
97. A. Mažeikiene, M. Valentukevičiene, J. Jankauskas, J. Environ. Eng. Landscape Manage. 18,
54 (2010)
98. M. Li, X. Zhu, F. Zhu, G. Ren, G. Cao, L. Song, Desalination 271, 295 (2011)
504 M. Turan

99. A. Khosravi, M. Esmhosseini, J. Jalili, S. Khezri, J. Incl, Phenom. Macrocycl. Chem. 74, 383
(2012)
100. M. Turan, H. Gulsen, M.S. Çelik, J. Environ. Eng. 131, 815 (2005)
101. T. Van Nooten, L. Diels, L. Bastiaens, Sci. Technol. 44, 3486 (2010)
102. S. Li, G. Huang, X. Kong, Y. Yang, F. Liu, G. Hou, H. Chen, Water Sci. Technol. 70, 1540
(2014)
103. W.-S. Hang, H.-T. Tran, D.-H. Park, R.-H. Zhang, D.-H. Ahn, J. Ind. Eng. Chem. 15, 524
(2009)
104. N. Turki, K. Elghniji, D. Belhaj, J. Bouzid, Desalin. Water Treat. 1, 1 (2014)
105. K.K. Krishnani, Y. Zhang, L. Xiong, Y. Yan, R. Boopathy, A. Mulchandani, Bioresour.
Technol. 117, 86 (2012)
106. S.J. Montalvo, L.E. Guerrero, R. Borja, Int. J. Environ. Sci. Technol. 11, 43 (2014)
107. Q. Deng, B.R. Dhar, E. Elbeshbishy, H.-S. Lee, Environ. Technol. 35, 2008 (2014)
108. D. Zhou, Y. Li, Y. Zhang, C. Zhang, X. Li, Z. Chen, J. Huang, X. Li, G. Flores, M. Kamon,
J. Contamin. Hydrol. 168, 1 (2014)
109. Y. Pei, J. Wang, Z. Wang, B. Tian, J. Environ. Sci. Health A 46, 715 (2011)
110. W.L. Zhu, L.H. Cui, Y. Ouyang, C.F. Long, X.D. Tang, Pedosphere 21, 454 (2011)
Chapter 20
Synthesis and Biological Applications
of Quantum Dots

Manolya Kukut Hatipoglu, Seda Kelestemur and Mustafa Culha

Abstract One of the fastest moving fields in nanotechnology is the application of

quantum dots (QDs) in biology. These light-emitting materials are a new class of
fluorescent labels in biomedical field. The superior properties such as resistance to
photobleaching, a broad range of excitation and size-tunable light emission facili-
tate the applications of QDs in nanodiagnostics, imaging and targeted delivery. In
addition, new synthesis approaches and modification strategies with several
biodegradable polymers and biomolecules bring in water solubility features to QDs,
which lead the way to the applications including cellular labelling, deep-tissue
imaging and assay labelling. Here we discuss the synthesis, surface modification
strategies and in vivo /in vitro biological applications of QDs.

20.1 Introduction

The development of biocompatible nanomaterials is one of the fastest moving fields

of nanotechnology due to promising application potentials in nanomedicine.
Quantum dots (QDs) are drawing attention among other nanomaterials such as
fullerenes, carbon nanotubes, graphenes, noble metal nanoparticles, and metal
oxide nanoparticles. The QDs are the fluorophores offering a unmatched promise in
the nanomedicine toolbox [1]. The strong light absorbance, bright fluorescence,
narrow and symmetric emission bands, broad range of excitation and high photo-
stability are among unique optical properties of QDs [2]. These tiny light-emitting
particles have several potential applications including biosensing, bioimaging, bio

M.K. Hatipoglu
S. Kelestemur
M. Culha (&)

Genetics and Bioengineering Department, Yeditepe University,
Faculty of Engineering and Architecture, Kayısdagi, Istanbul, Turkey
e-mail: [email protected]; [email protected]
S. Kelestemur
Istanbul Medipol University Regenerative and Restorative Medicine Research Center,
Kavacık, Istanbul, Turkey

© Springer International Publishing Switzerland 2016 505

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_20
506 M.K. Hatipoglu et al.

assay, targeted drug delivery and new therapeutic agent development [3]. The
diversity of the applications of QDs in nanomedicine field is attributed to their
exceptional photophysical properties [4]. QDs consist of a semiconductor core with
an outer shell of another semiconductor material with a larger spectral band gap [1].
They can be synthesized with diameters from a few nanometers to a few
micrometers with controlled size distribution [4]. Synthesis method can be tailored
to specific requirements with core, shell and coating characteristics that all affect the
optical properties of QDs [5]. The size and composition of QDs with numerous
surface functionalization possibilities make them amenable materials for medical
applications [1]. The utilization of QDs in vitro and in vivo bioimaging for real time
and long-term applications are offering exciting opportunities. The conjunction of
QDs with biorecognition molecules makes them perfect fluorescent probes for
detection, imaging and labelling in vivo and vitro applications. This chapter pro-
vides a short review of synthesis, surface modification strategies and biological
imaging applications of QDs in vitro and in vivo.

20.2 Synthesis of CdSe and CdSe/ZnS QDs

The synthesis methods for CdSe quantum dots (QDs) have been improved over the
past years. Although it is very challenging to control the size and shape of the QDs,
and it is now possible with the developed synthesis methods. Nucleation of
organometallic precursors in solvent is an efficient approach for synthesizing QDs.
The inorganic precipitation was carried out at high temperature in glasses or by
using chemical approaches at room temperature [6]. Micellular methods were
performed to obtain monodisperse QDs through precipitation of Cd and Se pre-
cursors in micelle solutions [7, 8]. The quantum effect of nanocrystals depends on
their size distributions and surface reconstructions. The micellular approaches are
insufficient in elimination of the intrinsic traps during the synthesis and needed
remarkable improvement to obtain narrow sized QDs [6].
Finally, the CdSe QDs with high optical quality and controlled size were synthesized
by Bawendi and co-workers using tri-n-octylphosphine (TOP) and tri-n-octylphosphine
oxide (TOPO) as stabilizing agents [9]. The organometallic reagents, dimethylcadmium,
bis(trimethylsily) selenium and bis(tert-dimeyhlybutylsilyl) tellerium, were injected
rapidly into the reaction solvent to control the nucleation step. The size selective pre-
cipitation allowed obtaining monodisperse QDs in different size ranges. Alivisatos and
co-workers have developed a synthesis method [10] by increasing the injection tem-
perature during the nucleation step and decreasing the temperature in the growth phase,
which affected the size, shape and optical properties of QDs. Also, tributylphosphine
was used instead of TOP, which the excess amount can be easily removed at high
temperatures. The most important improvement on this method was obtaining nearly
monodisperse nanocrystals without a need of a size selective precipitation [11]. In 1998,
Peng, Wickham and Alivisatos focused on the study of size distribution of the QDs.
Ostwald ripening is the theory that explains the diffusion of dissolved small crystals on
20 Synthesis and Biological Applications of Quantum Dots 507

the surface of the large crystals, which are thermodynamically favored according to
larger crystals due to their lower surface energy. Alivisatos and co-workers demon-
strated the effect of addition monomer reagent in the reaction solvent to obtain narrow
size distributed nanocrystals by using Gibbs-Thompson equation [12].

Sr ¼ Sb expð2rVm =rRT Þ ð20:1Þ

Sr is the dispersibility of the nanocrystal and Sb is corresponding to the bulk

solid; σ is specific surface energy; r is the radius of the nanocrystal; Vm is the molar
volume of the materials; R is the gas constant; and T is the temperature.

f2rVm =rRT  1; ð20:2Þ

The diffusion controlled growth rate of the particle with size r is

dr=dt ¼ K ð1=r þ 1=dÞð1=r   1=r Þ ð20:3Þ

K is a constant proportional to the diffusion constant of the monomer. δ is the

thickness of the diffusion layer. At a fixed concentration, r* is the particle radius,
which the nanocrystal is stable in suspension with proper monomer concentration.
The small nanocrystals can grow faster than the larger particles when the
nanocrystals size is larger than the r*. In the study of Alivisatos et al. it was
mentioned that the narrow sized nanocrystals can be obtained through adjusting the
concentration of the monomer [11]. Although, nearly monodispersed QDs were
synthesized but their quantum yield was low due to the organic compounds on their
surfaces, which were used in the synthesis procedure. It was demonstrated that
wrapping the surface with an inorganic compound is promising to increase the
quantum yield [8]. Hines and Guyot-Sionnest first began working on obtaining
monodisperse and highly luminescence quantum yielded QDs through coating the
CdSe surface with ZnS, which have a large band gap area increasing the fluorescence
properties [13]. The CdSe QDs were synthesized by using the method developed by
Bawendi and co-workers. The Zn/S/TOP solution was injected in an optimized
period of time at 300 °C. The CdSe/ZnS core shell structures were synthesized at
different size ranges by Bawendi et al. [14]. The ZnS capping layer increased the
quantum yield up to 50 % was a promising result paved the way for future studies.
The main difficulty of the core shell synthesis procedure is achieving a perfect ZnS
coating on the surface of CdSe. During the injection, Zn and S atoms tend to grown
on ZnS layer composed on CdSe. Injection Cd atoms during the growth phase
provide a perfect coverage of the QDs. It was demonstrated that the complete shell
coating increase the quantum efficiency of nanocrystals effectively [15].
Important progress has been made in recent years for the synthesis of QDs with
high optical properties. Green synthesis methods were developed to obtain
non-toxic QDs with a simple method. The synthesis of QDs in aqueous media is
more environmentally friendly approach according to organic phase synthesis. Liu
and coworkers synthesized QDs at lower injection temperature by providing long
508 M.K. Hatipoglu et al.

Fig. 20.1 Fluprescence of CdSe QDs synthesized at room temperature with different emission.
Reprinted with permission from [14] Copyright (2008) American Chemical Society

reaction times for the nucleation and growth phase of nanocrystals as seen in
Fig. 20.1 [16]. The thiol bounded QDs were synthesized with 22 % photolumi-
nescence quantum yield used as an alternative method to organic phase synthesis
[17]. Siy and co-workers obtained a narrowly distributed in size and high quality
QDs synthesized low reaction temperature compared to the hot injection approaches
[18]. Eventually Park and co-workers developed a new approach by using amino
acid cysteine and derivatives to synthesize very small sized hydrophilic QDs by
controlling size growth. A sol-gel method is effective for the synthesis of
monodispersed and properly shaped nanocrystals [19]. The great interest for the
green synthesis approaches [20] for QDs favored the development of safety sol-gel
synthesis methods by optimizing the reaction parameters [21].

20.3 Design of CdSe and CdSe/ZnS QDs for Biological

Applications

The CdSe and CdSe/ZnS QDs were synthesized in organic solvents as described
above. The core nanocrystals possess high fluorescence efficiency and stability in
organic media and coating the surface with ZnS layer increases the quantum yield
and quality. However, the organic ligand capped nanocrystals cannot be used in
biological applications due to their hydrophobic behavior. The QDs should be
soluble in biological buffers while maintaining their physicochemical properties and
size distributions. A surface modification is needed to provide hydrophilic behavior
to the QDs. Also, the surface modification prevents the release of Cd+2 ions, which
have a very toxic effect and limits the biological applications of the QDs [22]. There
are several approaches for modification of CdSe or CdSe/ZnS QDs to be used in
biological applications. There are several routes to design biocompatible
nanocrystals including, ligand exchange, silica coating and conjugation with
20 Synthesis and Biological Applications of Quantum Dots 509

amphiphilic molecules, poly ethylene glycol (PEG), phospholipid micelles and

biomacromolecules.

20.3.1 Ligand Exchange Process

Ligand exchange process is used to exchange the hydrophobic ligands on the

surface of QDs with thiol-ended ligands, which bound on the surface through
chemisorption. Mercaptoacetic acid [23], mercaptopropanoic acid [24], cystamine
[25], cysteine [26] are used with carboxyl(−COOH), amine(−NH2) and hydroxyl
(−OH) groups on their free edges allowed water solubility and further modifica-
tions. Although it was reported the ligand exchange process between the organic
ligands on the surface of QDs and thiolalkyl molecules do not increase the size of
the QDs, it reduces the quantum yield and the stability [27]. The decrease in the
quantum efficiency depends on the redox energy levels of thiol molecules. When
the valence band of the nanocrsytal is lower than the redox energy of the thiol
molecule, the holetrapping is thermodynamically favorable through the thiol
molecule. The hole trapping inhibits the radiative recombination of exciton resulted
with the decrease of the quantum yield [28].
In biological applications, the CdSe/ZnS QDs are routinely used, which the thiol
groups bound onto the ZnS layer through weak disulfide bonds [29]. The thiol
molecules bounded on the surface of QDs also form disulfide linkages and causes
oxidation and aggregation. To increase the long-term stability, dihydrolipoic acid
(DHLA) was used as a ligand exchange molecule. DHLA forms chelate with the
metal atoms on the surface of QDs and provide stability only in the basic conditions
[30]. Using DHLA and PEG together is another approach to provide usage of the
QDs also in acidic conditions [31]. Dithiol-containing molecules such as tridentate
[32] and multidentate ligands [33], and also multivalent thiol polymer ligands [34]
are used to provide long-term stability [35]. In a recent study, poly acrylamide acid
was used as a modifier in ligand exchange approach and provided solubility to the
QDs however it quenched the fluorescence [36].

20.3.2 Silanization Process

Coating the surface of the CdSe and CdSe/ZnS QDs with a silica layer is another
preferred approach. A silane-coupling agent can be used to generate a silica shell
coating on the surface, completely. The silica coating provides stability to the QDs
at neutral pH and under high salt conditions. Further, the modifications can also be
pursued with the selected ligands and polymers. In an aqueous solution, water
interacts with the silica layer through hydrogen bonding and this interaction pro-
vides a steric repulsion between the nanocrystals, which reduces agglomeration. As
an advantage of the silanization, the fluorescence properties of nanocrystals are not
510 M.K. Hatipoglu et al.

Fig. 20.2 Structure of silane coupling agent

quenched after silica coating process. The first route for silanization process is
direct ligand exchange process where generally 3-mercaptopropyl-trimethoxysilane
is used [37] to exchange with TOPO or the other hydrophobic ligands on the
surface of nanocrystals [38]. The general formula for a typical silane coupling agent
bears two classes of functionality as seen in Fig. 20.2. X is a hydrolyzable group,
which is generally an alkoxy, acyloxy, halogen or amine. First, the hydrolysis takes
place followed by the formation of silanol group which can condense with the other
silanol groups to from siloxane linkages [23, 39].
The second route is Stober approach, a sol-gel process, used to coat the surface
with a silica layer by using tetraethyl ortosilicate (TEOS) [40]. Stober method is
easy to apply and high monodisperse QDs can be obtained [41]. The optimization
of the silane coupling agent concentration is important to prevent the aggregation of
QDs and if it cannot be achieved, the aggregates may be encapsulated in a silica
shell, which increase the size of the structure.

20.3.3 Amphiphilic Molecules

The main purpose of coating the surface of QDs with an amphiphilic molecule is
the interaction of the hydrophobic residue of the molecule with the hydrophobic
ligands on the surface while the hydrophilic residue provides solubility in aqueous
environment. The synthesis procedure is started in an organic solvent and followed
by adding aqueous solution after evaporating the organic solvent [42].
Poly(acrylicacid) (PAA), a polymer with carboxylic acid groups, can be used as
an amphiphilic molecule for the attachment of an aliphatic amine molecule such as
N-octylamine (OA) and 5-amino-1-pentanol [43]. It was also shown that using
3-(dimethylaminoporpyl) carbodimide as a cross-linker for lysine modification
on the PAA-OA increase the stability of functionalized CdSe QDs in aqueous
environment [44].
Poly(maleic anhydride) copolymers are generally synthesized through copoly-
merization between maleic anhydride and olefins [45]. Designed maleic anhydride
based copolymers such as poly(maleic anhydride-alt-1-tetradecene) [46],
poly(maleicanhydride altoctadecene) [47], poly(maleic anhydride-alt-1-decene) [48],
were used for the surface functionalization with the aim of employing in biological
applications [49].
Block copolymers are used as surface modifiers through forming micellar
structure around the QDs. The amphiphilic block copolymers assembled in proper
20 Synthesis and Biological Applications of Quantum Dots 511

solvents and formed the micellar nanocrystal structures, which could be used in
several biological applications [50]. Using block copolymers as surface modifiers
have advantages; first the fluorescence properties of QDs are not quenched because
the polymer is not directly interacting with the nanocrystal’s core atoms. Second,
they are stable in aqueous solutions due to the hydrophobic domains around the
nanocrystals [51].

20.3.4 PEG and Phospholipid Micelle

PEG is a polymer used as a surface modifier for QDs, which is employed in several
biological applications such as drug delivery [52], imaging [53] and in vivo studies
[54]. Non-specific protein binding is prevented by encapsulation the nanocrystal
with PEG and also the modified nanocrystals cannot recognized by the reticuloen-
dothelial system (RES), which decrease their circulation time in the organism [55].
PEG polymer can be used by modification with thiol, amine and carboxyl active
groups. As the thiol groups bound on the nanocrystal’s surface and PEG chains are
responsible to provide solubility in all the acidic, basic and salty conditions of aqueous
media, where active amine and carboxyl residues allow further modifications [56].
PEG is also a preferred material for designing amphiphilic block copolymers,
where in a recent study, poly(ethylene glycol)-graft-chitosan was synthesized and
used to modify CdSe QDs. The reaction scheme was shown in Fig. 20.3. The results

Fig. 20.3 Reaction scheme of modification of CdSe QDs with poly(ethylene glycol)-
graft-chitosan [48]. Reprinted from Colloids and Surfaces B-Biointerfaces, Vol. 115, Jiang Z,
Zhao C, Liu X, Synthesis of poly(ethylene glycol)-graft-chitosan and using as ligand for
fabrication of water-soluble quantum dots, 260–266, Copyright (2014), with permission from
Elsevier
512 M.K. Hatipoglu et al.

were promising that the modification reduces the cellular toxicity of QDs and also
allowed using in biological applications through providing water solubility [57].
CdSe and CdSe/ZnS QDs can be encapsulated in phospholipid micelles to use
both in vivo and in vitro imaging [58]. The encapsulation approach does not alter
the fluorescence and surface properties of the QDs and can be used in drug delivery.
The water insoluble anticancer drugs can be delivered by this safe delivery system
[59]. When the nanocrystals encapsulated with poly(ethylene glycol)-phospholipid
micelle, PEG improve the solubility and circulation time of the QDs and also it can
be modified with antibodies to use in targeted delivery [60].

20.3.5 Biomacromolecules

CdSe and CdSe/ZnS QDs are modified with biomacromolecules such as peptides,
proteins, nucleic acids and polysaccharides to use in biological applications. First,
the surface of the QDs is modified with several ligands, which allows for further
modification with biomacromolecules through non-covalent interaction and cova-
lent binding.
Non-covalent interactions of QDs with biomacromolecules occur through elec-
trostatic interaction, hydrogen bonding and through metal-coordination. The
water-soluble QDs can conjunct with the proteins through electrostatic interactions.
The positively charged domains of the proteins can self-assemble on to the nega-
tively charged surface modified QDs [61]. It was demonstrated by Lee et al. that the
PEG modified water soluble QDs formed a complex with negatively charged
deoxyribose nucleic acid (DNA) with an electrostatic interaction [62]. The strength
of hydrogen bonding interaction between QDs and biomolecules depends on the pH
and ionic strength of the aqueous media [63]. The association of proteins on the
QDs was explained in a study that the hydrogen bonding between carboxyl groups
of proteins and amine terminated silica coated QDs is thermodynamically favorable
process [64].
The peptide and QDs conjugates are generally used for cellular delivery appli-
cations [65]. As another approach, the peptides self-assemble on the QDs’ surface
through expressed polyhistidine sequences via metal affinity coordination with zinc
ions on the surface of CdSe/ZnS QDs [66]. A mixture of different peptides can
cover the surface with the self-assembly approach, which have unique functional-
ities in a single peptide-QDs conjugate system [67]. Although non-covalent inter-
action is an easy process and not need a cross-linker agent for conjugation but it is
difficult to control the binding percentages of modifier on the surface of the
nanocrystals. The strength of the interaction is also not very stable in cellular
environment, where the free biomolecules have affinity to bind on the surface of the
QDs through ligand exchange process. The water soluble QDs containing func-
tional groups (−NH2, −COOH, −OH and −SH) can be modified with biomolecules
through covalent attachment to provide stability in aqueous media. In general,
surface modified QDs with free carboxyl active groups can bind with active amine
20 Synthesis and Biological Applications of Quantum Dots 513

groups of proteins, peptides and nucleic acid using a cross-linker 1-ethyl-3-

(dimethylaminopropyl) carbodiimide hydrochloride (EDC) [68]. N-hydroxy suc-
cinimide (NHS), which has to be used for formation of a stable intermediate product
to increase the yield of the reaction, while EDC cannot react with the amine groups
rapidly. Biomacromolecules are covalently attached to the quantum dots through
amide bond formation using EDC and NHS in two-step process [63, 69]. The
reaction scheme is shown in Fig. 20.4.
Thiol groups have affinity for chemisorption onto the CdSe/ZnS QDs, which can
allow the direct attachment of biomacromolecules without using a chemical agent.
Through this approach the hydrophobic layer on the CdSe/ZnS QDs replace with
the thiol containing biomacromolecules, which is called ligand exchange process.
Thiol containing peptides, proteins and oligonucleotides are used to provide
water-solubility and stability to the QDs [70]. In an aqueous synthesis approach, the
QDs are synthesized in water including thiol containing biomacromolecules to
obtain non-toxic and practicable QDs in bioimaging and biolabelling [71].
Bioortogonal click chemistry has pronounced interest in recent years, which the
functional groups of biomacromolecules are not affected from the reaction process
[72]. The reaction is called as “click”, which is the of Huisgen 1,3-dipolar cycload-
dition reaction between azides and terminal alkynes and Cu(I) are used as catalyst
[73]. The reaction scheme is shown in Fig. 20.5. Although this approach used as a
surface modification strategy for QDs, it quenches the fluorescence property [74].
The development of copper-free click chemistry provides opportunity for conjuga-
tion biomacromolecules on the surface of QDs without altering their properties [75].
Carbohydrates play a very important role in biological systems and can cova-
lently attached to a terminal polymer chain of a diblock amphiphilic polymer
through Huisgen-type click chemistry as demonstrated in a recent study [76]. The
QDs were encapsulated into the glycoconjugated polymer system, which are used
for their binding affinities to Concanavalin A with nontoxic behaviors as seen in
Fig. 20.6 [76].

Fig. 20.4 Schematic diagram of the covalent attachment of biomolecule with QDs

Fig. 20.5 Cu-catalyzed azide-alkyne cycloaddition

514 M.K. Hatipoglu et al.

Fig. 20.6 Schematic diagram

of poly(isoprene)-b-poly
(ethylene glycol) diblock
copolymer modification of
QDs with unmodified (a) and
glycoconjugated polymer
(b) [64]

20.4 Biological Applications of QDs

The aim of using CdSe and CdSe/ZnS QDs in biological applications is to develop
probes with high selectivity, versatility, capacity to penetrate cells and organelles
[2, 77]. However, there are certain biological challenges to face such as biocom-
pability, non-specific binding and toxicity. The reproducible fabrication is another
difficulty to meet. A large amount of work on applications of QDs is devoted for
bioconjunction and biological labelling for tracking and imaging [78]. The first
application of QDs as label in biology was reported in 1998 [79]. QDs are amaz-
ingly appropriate for immunolabeling, cell motility assays, in situ hybridization and
live cell markers [80, 81]. The engineered QDs (water-soluble) can be cross-linked
to biomolecules such as proteins, peptides, oligonucleotides, and antibodies to
target specific biomolecules [82]. The fact that the applications of QDs are limited
depending on the administration routes and dispersion within the organism. They
tend to aggregate in cytoplasms of living cells and commonly trapped in endosomes
and lysosomes. There are varieties of tracking and detection methods of QDs
including confocal microscopy, total internal reflection microscopy, wide-field
epifluorescence microscopy and fluorometry [83, 84].
QDs have a diverse array of applications in imaging biological samples by a
unique property that several QDs can be excited at the same wavelength of light [1,
85, 86]. This superiority acquires a reputation to QDs as optical bar codes of
biological screening [87]. These labels can be specific to a particular analyte or a
cell type since one wavelength could be directed [88]. QDs can be employed to
obtain information location, saturation and distribution of marked cells or proteins
[85]. This multiplexing potential is practicable on cancer diagnostics and research
20 Synthesis and Biological Applications of Quantum Dots 515

to detect cancer biomarkers [85]. Detection of several proteins on Western blots

with QDs opens up a new era in proteomics field [81]. In a study of Makrides et al.
showed that streptavidin coated QDs would be used in detection of two proteins in a
Western blot assay [89]. Additionally, their stability and resistance to metabolic
pathways in live cells allow the usage of QDs in long term studies [90].

20.4.1 In Vitro Targeting with Antibody Conjugation

High photostability, resistance to metabolic degradation and lack of cytotoxicity

facilitates the utilization of QDs for imaging thick tissues for a long period of time
without any damage to the pattern [80].
QDs bioconjugates have been found as a powerful tool to recognize and track
plasma membrane agents on living cells [82]. The conjunction of QDs with cancer
specific agents such as peptides, ligands and antibodies makes them propitious for
cancer detection and imaging. Figure 20.7 shows some of the QD-conjugate
structures used in biological applications.
One of the reasons of the myriad of applications of QDs conjugates is the
antibody-QD conjugate, which is the simplest way to label if an antibody exists for
the target. It is the fastest and easiest route to follow. For instance, hairy-related 2
protein (Her2) receptor, which is a well-known cancer marker over-expressed in
many breast cancers, was detected by Wu et al. by the development of
streptavidin-coated QDs [90]. The assay employed humanized anti-Her2 antibody
and a biotinylated goat anti-human IgG secondary antibody on SK-BR-3 breast
cancer cells.
Another similar approach was used to detect intracellular targets including
microtubules in mouse 3T3 fibroblasts, nuclear antigens, and F-actin filaments via

Fig. 20.7 QD-conjugate structures used in biological applications. Reprinted from Advanced
Drug Delivery Reviews, Vol. 60, Andrew M. Smith, Hongwei Duan, Aaron M. Mohs, Shuming
Nie, Bioconjugated quantum dots for in vivo molecular and cellular imaging, 1226–1240,
Copyright (2008), with permission from Elsevier
516 M.K. Hatipoglu et al.

streptavidin-coated QDs. A single wavelength was used for the detection on

SK-BR-3 cancer cells [84]. MCF-7 and BT-474 breast cancer cells were selectively
labelled as an extension of this work by Yezhelyev et al. [91]. They used conjugates
of QDs with antibodies for Her2, epidermal growth factor receptor (EGFR),
estrogen receptor (ER), progesterone receptor (PR) and mammalian target of
rapamycin (m-TOR). The labelled cells were detected by visible and near infrared
(NIR). Also, Kawashima et al. employed the same system, the conjugates of QDs
with EGF and anti EGFR antibody, for targeting EGFR in human ovarian epi-
dermoid carcinoma cells (A431) [91, 92]. Tada et al. reported that, KPL-4 breast
cancer cells were selectively labeled with conjugate of QDs with Herceptin, similar
to the previous studies [91, 93]. Also, Weng et al. implemented the labeling system
to a multimodal approach by drug delivery of an antibody via immunoliposomes
and imaging cells with QDs [94]. They used carbodiimide chemistry between
carboxylic acid functionalized QDs and a primary amine group of liposome for
coupling. Selective labeling of cancer cells, high-contrast fluorescence imaging and
carrying anti-cancer drugs are just a few advantages of multimodal immunolipo-
somes. In a previous study of Zhang et al. mentioned that, conjugate of QDs with
anti-type 1 insulin-like growth factor receptor (IGF1R) was used on up-regulated
IGF1R in MCF-7 breast cancer cells for detection and imaging. The results claimed
that the QD-anti-IGFR1 conjugate is bearing promising candidates for imaging of
breast cancer cells [95]. Figure 20.8 shows the fluorescence images of human
pancreatic cancer cells (MiaPaCa) treated with InP/ZnS QD-anti-Claudin-4 anti-
body conjugates, which is a study of Yong et al. [96]. They claimed that conjugates
of QDs with anti-Claudin-4 antibody and anti-prostate stem cell antigen
(anti-PSCA) selectively detected by human pancreatic cells. The proteins Claudin-4
and PSCA are generally over-expressed in both primary and metastatic pancreatic

Fig. 20.8 Fluorescence images of human pancreatic cancer cells incubated with a InP
QD-anti-Claudin-4 antibody conjugate and b InP QD without antibody. Reprinted with permission
from a [94] and b [96]. Copyright (2008, 2009) American Chemical Society
20 Synthesis and Biological Applications of Quantum Dots 517

cancer cells. Lidke et al. showed that, QD conjugates could be used to distinguish
cultured human cancer cells [97]. Coupling CdSe-ZnS QDs with EGF, which is a
very small protein specific to erbB/HER membrane receptor, enables the detection
of this protein at the single-molecule level with a continuous observation of protein
diffusion on the cellular membrane. Gao et al. used prostate specific membrane
antigen (PSMA) coupled with QDs by the injection to the tail vines of mice for the
detection of prostate cancer xenografts [98]. Detection of human prostate cancer
cells (C4-2) proceeded through labelling an antibody for PSMA with QDs.
A conjugate of QDs was developed with three different peptides for molecular
imaging of brain and lung endothelial cells and breast carcinoma cells. The studies
were carried both in vivo and in vitro by Akerman et al. [99]. The cellular orga-
nelles including nucleus and mitochondria can also be monitored with QDs-peptide
conjugates [84]. In another study, Larson et al. used water-soluble QDs in an
imaging application that indicated the stability of QDs was over 9 months by
fluorescence correlation spectroscopy [100]. In the latest study Gao et al. used
QD-antibody probes by detecting the staining intensity [101, 102]. They used QDs
based multicolor multicycle molecular profiling technology to target LaminA,
Cox-4, Ki67 and HSP90 in single cells, which is reported by Gao et al. in 2013
[101, 102].

20.4.2 In Vitro Targeting with Peptide Conjugation

A single QD can be conjugated with several proteins and peptides simultaneously

depending on its size (see Fig. 20.9) [103, 104]. Medintz et al. showed that
approximately 15–20 maltose binding proteins (MBP, 44 kDa) can be conjugated to

Fig. 20.9 Comparison of QD size to a MBP molecule. 555-nm-emitting CdSe/ZnS core/shell QD,
diameter *60 Å, surface-functionalized with dihydrolipoic acid (red shell *9–11 Å) has a
diameter *78–82 Å. The diagram depicts the homogeneous orientation MBP assumes relative to
the QD MBP a midsize protein (Mr * 44 kDa) has dimensions of 30 × 40 × 65 Å [105].
Copyright (2004) National Academy of Sciences, U.S.A
518 M.K. Hatipoglu et al.

a 6 nm sized QD [103]. Conjunction of proteins and peptides to QDs can be divided

into three categories as carbodiimide chemistry, direct binding of thiolated proteins
to the QDs and adsorption or non-covalent interaction of engineered proteins.
QDs-streptavidin conjugates can be counted as an example of carbodiimide
chemistry with high quantum yield, which is also commercially available.
The direct attachment of proteins to the QDs occurs between cysteine residues
and the surface of QDs [99, 106]. Non-specific adsorption of proteins to the QDs
has handicaps including lack of homogeneity of proteins on the surface of QDs and
inability to control protein/QD ratio. Homogeneity is an important issue that,
heterogeneous orientation of proteins may result in protein-peptide/QD conjugates
without functionality. Peptides are efficient carriers of QDs inside the living cells.
There are typical examples of Tat peptides [107], polyarginine [108], cholera toxin
B [109] and allatostatin [110] as QD carriers. Curtis and co-workers demonstrated
that the conjugates of Tat peptides can be used as vehicles to carry QDs inside
human fibroblast cells and nucleus of COS 1 cells [67]. An intercellular delivery of
QDs was shown by Waggoner et al. by simply conjugating QDs with poly
(L-arginine), where QDs was internalized in HeLa MG63 and 3T3 cells [108].
Another example of internalization is cholera toxin B coated QDs that the conju-
gates well dispersed in cytoplasm of various cell lines [109]. Additionally, a neu-
ropeptide from Drosophila melanogaster, which is called as allatostatin, found as a
potential carrier of QDs [110]. Utilization of peptide modified QDs is generally
used to target cellular proteins including growth factor receptors, G protein-coupled
receptors, integrins and ion channels [111]. Also, there are many examples of QD
conjunctions with HIV TAT, PEP-1, polyarginine and SV40 T antigen providing
intercellular delivery [98, 112]. Conjunction of arginine-glycine-aspartic acid
(RGD, ̴ 30–50) peptides with QDs used to observe avb3 integrins in mouse tumor
via NIR [113]. In a recent study, anti-hemagglutinin (HA) conjugated QDs was
used to investigate the kinetics of G protein coupled receptor (GPCRs) and endo-
somal trafficking pathways, where QDs were used as fluorescent probes [114].

20.4.3 In Vitro Targeting with Small Molecule Conjugation

High affinity and selectivity of ligands as small molecules to the cell surface
receptors make them propitious candidates as targeting agents [111]. Some
examples of small molecules are shown in Table 20.1. The ligands with high
affinity lead the way to the protein of interest for QDs. Labeling the binding site of
the protein by QDs bring many opportunities as monitoring the protein and helping
to clarify the regulation of protein expression and cell surface mobility [111].
Ligand-conjugated QDs may also be utilized in drug discovery that each ligand can
be conjugated to different sized QDs, which means different color for a multiplexed
fluorescent assay [111]. The major difficulty in ligand-QDs conjugates is tailoring
ligand according to the needs such as a linker arm to optimize binding, which
requires a good command of organic chemistry.
20 Synthesis and Biological Applications of Quantum Dots 519

Table 20.1 Small molecule ligands that have been conjugated to quantum dots
PEGylated serotonin derivative [116]

Muscimol derivative [117]

Dopamine transporter antagonist [118]

Seratonin transporter antagonist [119]

GPI a tumor targeting ligand [120]

A derivative of the NSAI D naproxen [121]

Dopamine [122]

Glutathione [123]

According to a study in 2007, aptamer labeled QDs could selectively be

delivered to prostate cancer cells [115]. The A10 RNA was used as an aptamer to
label PSMA positive LNCaP prostate cells. The labeling system was found as
successful to distinguish the difference between positive and negative LNCaP
prostate cells. In a study of Kawashima et al. they used the ligand-QDs conjugates
as a biological probe and found that CHO and A431 cells were efficiently labeled
with EGF as a result of specific binding of EGF to EGFR [92]. The results of this
study unleash the route for detection of various cancer cells since EGRF is an
over-expressed protein in many cancer types.
520 M.K. Hatipoglu et al.

20.4.4 In Vivo Applications of QDs

Many optical techniques have been employed for biomedical imaging using animal
models but the poor transmission of visible light through biological tissue prevents
its utilizations. Weissleder claimed that, NIR might be applicable for deep tissue
imaging [124]. A few of organic dyes can be used in this prospect but photo-
bleaching limits their use for the purpose [125]. Unlike the organic dyes, QDs are
ideal materials for imaging in living tissue since their emission wavelengths can be
tuned into the desired excitation region of the spectrum. The emission wavelength
of the QDs can be adjusted by altering their composition and size for their exci-
tation in the NIR region as photostable fluorophores [126]. The general adminis-
tration of QDs or any other nanoparticles is systemic intravenous delivery to the
blood stream. Depending on to administration route; the interaction of QDs with
plasma components, biodistribution in organs and non-specific interaction with
cells are important issues to deal with. The biodistribution mechanism starts with
the administration of QDs through adsorption by opsonins, which mark them for
phagocytosis. The pathways of QDs after administration are shown in Fig. 20.10.
There are several reports indicating the nonspecific uptake of QDs by RES, liver
and spleen, which do not reflect the actual behavior of QDs [127–129]. The first
in vivo application of QDs was performed by Akermen et al. in 2002 [130]. They
administrated the QDs-peptide conjugates through tail vein in mouse and demon-
strated the distribution of the QDs in the lung blood vessels. Based on Ballou et al.
findings, it can be said that QDs remain in the organs of RES and can be detectable
for at least 4 months from initial administration [127–129]. QDs can last for long
periods of time without degradation into toxic elements.
Fisher et al. demonstrated that surface modification of QD is holding key for
recognition of the particles. According to their study, albumin coated QDs were
removed from the circulation and held in the liver after a tail vein injection, which
was much faster than unmodified QDs [128]. From a clinical point of view; it is
assumed that the accumulation of the QDs can be counted as negligible, as long as
they are within the size of renal excretion. Recent publications about in vivo
applications of QDs have based on this assumption. Another important issue for
targeted imaging is the half-life of contrast agents in blood stream, which is related
to biodistribution of the particles. It was found that the surface properties of QDs
bearing a great importance in biodistribution as demonstrated in Ballou et al. study
on mice [127].
Gao et al. used an antibody conjugate of QDs encapsulated in PEG to track PSMA
[98]. The conjugate bound to human prostate xenografts, selectively. They claimed
that the signals from QD conjugate were brighter than the green fluorescent protein
(GFP), which was used for a similar manner, previously. Dubertret and co-workers
followed a parallel approach of the employment of QDs for imaging [131]. They
coated QDs with PEG-phosphatidylethanolamine and PEG-phosphatidylcholine,
which are micelle-forming block copolymers [131]. In an another example, a triblock
copolymer consist of poly(butyl acrylate), poly(ethyl acrylate) and poly
20 Synthesis and Biological Applications of Quantum Dots 521

Fig. 20.10 Schematic diagram showing QD interactions with blood immune cells and plasma
proteins. The probable modes of interactions including QD opsonization and phagocytosis by
leucocytes (e.g., monocytes), nonspecific QD–cell membrane interactions (electrostatic or
hydrophobic), and fluid-phase pinocytosis. Reprinted from Advanced Drug Delivery Reviews,
Vol. 60, Andrew M. Smith, Hongwei Duan, Aaron M. Mohs, Shuming Nie, Bioconjugated
quantum dots for in vivo molecular and cellular imaging, 1226–1240, Copyright (2008), with
permission from Elsevier

(methacrylate) was used to modify the surface of QDs. The modified QDs were
conjugated with PSMA antibody and they were able to observe prostate cancer
successfully in nude mice [98]. The triblock copolymer acted as reactive surface for
conjunction of PSMA antibody. Figure 20.11 shows the localized QD—antibody
conjugates on tumor sites of a mouse.
Another developing area of in vivo applications of QDs is imaging of inflam-
mation related to variety of diseases such as diabetes, atherosclerosis, colitis and
asthma [132]. Many different cell types and surface proteins take place in inflam-
mation process. A study on investigation of inflammation in retinal endothelium of
streptozotocin-treated diabetic rats showed that, different color-coded QDs could be
detected and identification of increased retinal vasculature was possible on diabetic
animals [133]. The conjugates of various QDs with monoclonal antibodies specific
to leukocytes and neutrophils were used for the imaging.
522 M.K. Hatipoglu et al.

Fig. 20.11 In vivo

fluorescence image of tumor
bearing mice using QD
probes with PEG-PSMA
antibody conjugates
Reprinted by permission from
Mcmillan Publishers Ltd:
[Nature Biotechnology], [98],
Copyright (2004)

20.4.5 In Vivo Vascular Imaging

The initial successful in vivo applications of QDs were accomplished on the car-
diovascular and lymphatic systems of mammals that it was a non-specific imaging
of tumor vasculature. It was shown that the intravenously injected QDs might be
employed as marker for fluorescent detection in capillaries of adipose tissue and
skin of living mouse [100]. This study triggered the future investigations on
non-specific in vivo imaging of tumor vasculatures, lymph nodes and lymphatic
drainage by bioconjugated QDs. For instance, PEG-phosphatidylethanolamine
modified QDs were used to target and imaging of tumor vasculature related with
MCaIV isogenic mouse adenocarcinoma tumor implants in C3H mice by Stroh
et al. [134]. In general, the NIR was used to excite the QDs for deep tissue
penetration, which cannot be done with organic dyes [135]. The excitation with
NIR was used in many studies for imaging including the coronary vasculature of a
rat heart and the blood vessels of chicken embryos [136].
The size of the imaging probe is more important in lymphatic system. The first
attempt to use QDs for fluorescence imaging of lymph nodes was reported by
Frangioni et al. that they used oligomeric phosphine coated CdTe-CdSe QDs on mice
and pig [126]. They were able to detect the signals of QDs from 1 cm beneath the skin
pre-surgically and even under bloody surgical conditions. As the hydrodynamic
volume of probes gets smaller, the penetration of imaging agents into lymphatic
system increases. If the QDs has overall hydrodynamic volume around 9 nm, they
might able to label five nodes [137]. This technique would help the surgeons to
identify lymph nodes by fluorescence and excise nodes draining from metastatic
tumors since the lymph nodes are one of the most important prognostic signs of
20 Synthesis and Biological Applications of Quantum Dots 523

cancers. Additionally, there are examples of the utilization of the QD-assisted

mapping on identification of lymph nodes downstream from the lungs [129, 138],
esophagus [139] and from subcutaneous tumors [139]. QDs could be fluorescently
observed up to five different nodes by the multiplexing capability [140].
In a recent study, anti-hemagglutinin (HA) conjugated QDs was used to
investigate the kinetics of G protein coupled receptor (GPCRs) endosomal traf-
ficking pathways, where QDs were used as fluorescent probes [114]. The latest
study is reported by Gao et al. in 2013 about QDs based multicolor multicycle
molecular profiling technology to target LaminA, Cox-4, Ki67 and HSP90 in single
cells [102]. They used QD-antibody probes by detecting the staining intensity. The
latest study on QDs was on modifying QDs with RGD peptides as NIR emitting
probes for specific tumor targeting by Wang et al. [141]. The tumor was clearly
detected after 30 min. of administration of QDs. A distinct red fluorescence was
observed after 2.5 h of injection with a maximum tumor to background ratio [142].

20.4.6 Förster Resonance Energy Transfer (FRET) Based

Applications

Förster resonance energy transfer (FRET) theory was first put forward by Theodor
Förster in 1948 [143]. FRET is allowed an energy transfer from a donor to an
acceptor through a dipol-dipol interaction, which is used to identify a protein-protein
interaction, conformational changes in biological molecules and ligand receptor
binding in the range of 1–20 nm [144]. In general, fluorophores such as fluorescent
dyes, proteins and polymers are used as energy donors in FRET technique. The QDs
are unique fluorophores with their broad absorption and narrow emission profiles,
high excitation coefficients and brightness and also large surface area allows using as
a biosensor through conjugation with proper biomolecules [145]. Also, QDs can be
synthesized at different sizes, which can prevent the spectral overlapping between
donor and acceptor. All these advantages increase the efficiency of FRET technique
and make QDs preferable fluorophores for imaging [103].
Despite the superior optical properties, QDs are not appropriate materials as
acceptors depending on the large extinction coefficients and log excited-state life-
times [146]. In previous studies, it was shown that QDs are strong FRET donors to
organic dye acceptors in multiple applications [146, 147]. Algar and Krull showed
that, the non-radiative energy transfer between QD donors and organics dyes or
fluorescent proteins (FP) are in parallel with the dipole-dipole interaction of Forster
theory [66, 146].
The most important point to consider before choosing a QD-dye or QD-FP pair
is the spatial proximity between two fluorophores that has a great impact on energy
transfer efficiency. The distance between donor and the acceptor called as Förster
distance (R0), which can be determined by the photophysical properties of the
donor and the acceptor molecules (J.R. Lakowicz, Principles of flourescence
spectroscopy, 3rd ed. Springer, New York, USA (1999)). The quantum yield of the
524 M.K. Hatipoglu et al.

QD will be towering in the absence of acceptor, for a FRET pair, where the QD
used as a donor. The QDs were found as strong donors when paired with the
acceptor molecules with a strong spectral overlap [148]. Another issue while
choosing the QDs to employ in FRET is the surface coating of the QDs. A study in
2008 showed that the modifiers on the surface of the QDs affect the energy transfer
to FPs. The small ligand coatings, which allow shorter distances between QDs and
FPS, result in higher energy transfer in vitro assays [148, 149]. In general conju-
gates of QDs with biological molecules are employed in FRET for immunoassay
studies. For instance, McKinney et al. used the QDs conjugates to monitor the
protein interactions in Holiday Junction [150]. DNA undergoes a conformational
change in the presence of Mg2+ ions during recombination phase [151]. They
monitored the change in the emission by the movement of arm due to Mg2+ ions
where QDs are used as donor and Cy5 used as the acceptor on the two arms of
DNA [151]. Another study on enzyme assays pointed out the importance of QDs
for FRET experiments [103].
The QDs can be used to develop as biosensors in different applications using
FRET technique. Escherichia coli maltose-binding protein conjugated QDs were
used as a carbohydrate biosensor by Medintz and co-workers [103]. Detection of
nucleic acids through QDs FRET approach is also an interested area to
develop. A streptavidin conjugated CdSe/ZnS QDs was used as an energy donor for
sensitive detection of Cy5 labelled DNA by Zang et al. in 2005. Detection was
occurred with hybridization of biotinylated oligonucleotide with Cy5 labelled DNA
as seen schematically in Fig. 20.12 [152]. The FRET efficiency found maximum of

Fig. 20.12 Design of QD based nanosensor (a) and fluorescence emission from Cy5 using QD as
an energy donor (b). Reprinted with permits from [6] Copyright (2005) Nature Publishing Group
20 Synthesis and Biological Applications of Quantum Dots 525

54-fold greater than when single molecules are bound [153]. The main limitation of
using QDs in FRET applications is the increased size of the QDs due to the surface
modification process to improve water-solubility. Using thiolated molecules is a
good strategy to prevent the increasing of the size of the QDs. The conjugate of
QDs with thiolated fluorophore labelled DNA was used as an energy donor with
high efficiency [63]. In a study of Zhou and co-workers, tri(ethylene glycol) was
used as a linker for coating the QDs surface with DNA, which prevent the non-
specific interactions between QDs and DNA [68]. The differentiation of the FRET
interactions becomes difficult in case of non-specific bindings [154].
There are various examples of QDs in FRET applications on gene technology.
A high number of studies demonstrated that the conjugates of QDs with oligonu-
cleotide sequences bound through surface carboxylic groups may be utilized to bind
DNA or mRNA [155]. For instance, Patolsky et al. demonstrated that, QDs could
be used to determine the kinetics of telomerisation and DNA replication via FRET
[156]. The strength of the quenching is affected by the interparticle distance, which
was shown by the gold conjugated DNA with QDs [157]. The percentages of
quenching were found as 79 [157] and 85 [158] depending on the short interparticle
distances addition to non-radiative interactions. Gueroi et al. were also claimed that
the interaction of QD-DNA complexes with clean oligonucleotides, which were not
bearing any Au, affect the emission yield, significantly [157].
Besides the DNA technology, QDs has noticeable applications in RNA tech-
nology. The QDs are generally employed in the detection of mRNA through in situ
hybridization (ISH) and in RNA interference applications with siRNA [103].
The ISH techniques are used in the detection of expression of specific mRNA
transcripts in mouse mid brain, where the success of QDs have already proven
[159]. The results of QDs are much better than the most sensitive organic fluor-
ophore on labelling of four different mRNA transcripts in neurons. Combination of
immunohistochemistry with ISH techniques stand to visualize the localization of
growth hormone and prolactin proteins related to their mRNA [103].
The QDs are also utilized in RNA interference applications to determine the
extent of gene knockdown in a cell through the difference in brightness, which
became an important tool to monitor the gene functions [160].

20.5 Conclusions

In this chapter, some of the recent advancements in the synthesis methods, bio-
conjunction strategies and in vivo/in vitro biological applications of QDs are
summarized. The synthesis of QDs and understanding their optical properties open
a new era in biomedical detection and imaging. The development of new synthesis
and modification approaches increased the fluorescence properties of QDs and their
compatibility with biological systems.
QDs are already proved itself with their size-tunable and stable photolumines-
cence, semiconductivity, biocompability and availability in multicolor features. The
526 M.K. Hatipoglu et al.

multiplex detection of biomolecules to multicolor whole animal imaging drew

much attention to QDs. They are brighter than the dye molecules and the large
surface area provides the opportunity to attach multiple cancer markers, which
make them promising materials for multimodal imaging in vivo and in vitro.
Surface chemistry, polymer chemistry and bioconjugate chemistry were success-
fully employed for the modification of QDs up to date. The utilization of QDs to
in vivo and in vitro targeting of prostate, breast, pancreatic and liver cancers was
demonstrated through QD-antibody, QD-peptide and QD-small molecules bio-
conjugates. The deep-tissue imaging applications of QDs with NIR are another
aspect to be considered as improvement in biomedical field. In vitro/in vivo
detection and imaging studies with cancer markers and other reporter molecules are
already established in animal models.
The spectral characteristics of QDs provide a unique opportunity to utilize them
in FRET applications as donor fluorophores. QDs have gained their place on gene
technology, cell tracking, pathogens and toxin detections as FRET donors. The
future of the QDs as FRET donors depend on the QD-FRET innovative approaches.
It is expected that, the surface modifications and the developments in synthesis will
reduce the negative effects of QDs in spite of their complex nature.
The recent advancements on bioconjugated QDs clearly revealed that the QDs
are powerful tools on cancer diagnosis and prognosis. However, there is still need
for improvement on issues of QDs probes such as specificity of targeting, signaling
intensity, multimodality and therapeutic potentials. The next challenge of utilization
of QDs into biomedical field is the commercialization of the products. Although
there are some commercially available products of QDs for research purposes, they
have a long way to go before they can hit the market.

Acknowledgments The authors acknowledge Yeditepe University and TUBİTAK for their
financial support.

References

1. H.M.E. Azzazy, M.M.H. Mansour, S.C. Kazmierczak, From diagnostics to therapy:

prospects of quantum dots. Clin. Biochem. 40, 917–927 (2007)
2. P. Juzenas, W. Chen, Y.-P. Sun, MaN Coelho, R. Generalov, N. Generalova, I.L.
Christensen, Quantum dots and nanoparticles for photodynamic and radiation therapies of
cancer. Adv. Drug Deliv. Rev. 60, 1600–1614 (2008)
3. Q. Huo, A perspective on bioconjugated nanoparticles and quantum dots. Colloids Surf., B
59, 1–10 (2007)
4. T. Jamieson, R. Bakhshi, D. Petrova, R. Pocock, M. Imani, A.M. Seifalian, Biological
applications of quantum dots. Biomaterials 28, 4717–4732 (2007)
5. S. Santra, K. Wang, R. Tapec, W. Tan, Development of novel dye-doped silica nanoparticles
for biomarker application. J. Biomed. Opt. 6, 160 (2001)
6. L. Brus, Quantum crystallites and nonlinear optics. Appl. Phys. Mater. Sci. Process. 53,
465–474 (1991)
20 Synthesis and Biological Applications of Quantum Dots 527

7. M.G. Bawendi, A.R. Kortan, M.L. Steigerwald, L.E. Brus, X-ray structural characterization
of larger cdse semiconductor clusters. J. Chem. Phys. 91, 7282–7290 (1989)
8. A.R. Kortan, R. Hull, R.L. Opila, M.G. Bawendi, M.L. Steigerwald, P.J. Carroll, L.E. Brus,
Nucleation and growth of CdSe on ZnS quantum crystallite seeds, and vice versa, in inverse
micelle media. J. Am. Chem. Soc. 112, 1327–1332 (1990)
9. C.B. Murray, D.J. Norris, M.G. Bawendi, Synthesis and characterization of nearly
monodisperse cde (e = s, se, te) semiconductor nanocrystallites. J. Am. Chem. Soc. 115,
8706–8715 (1993)
10. J.E.B. Katari, V.L. Colvin, A.P. Alivisatos, X-ray photoelectron-spectroscopy of cdse
nanocrystals with applications to studies of the nanocrystal surface. J. Phys. Chem. 98, 4109–
4117 (1994)
11. X.G. Peng, J. Wickham, A.P. Alivisatos, Kinetics of ii-vi and iii-v colloidal semiconductor
nanocrystal growth: “Focusing” of size distributions. J. Am. Chem. Soc. 120, 5343–5344
(1998)
12. T. Sugimoto, Preparation of monodispersed colloidal particles. Adv. Colloid Interface Sci.
28, 65–108 (1987)
13. M.A. Hines, P. Guyot-Sionnest, Synthesis and characterization of strongly luminescing
ZnS-capped cdse nanocrystals. J. Phys. Chem. 100, 468–471 (1996)
14. B.O. Dabbousi, J. Rodriguezviejo, F.V. Mikulec, J.R. Heine, H. Mattoussi, R. Ober, K.F.
Jensen, M.G. Bawendi, (CdSe)ZnS core-shell quantum dots: synthesis and characterization
of a size series of highly luminescent nanocrystallites. J. Phys. Chem. B 101, 9463–9475
(1997)
15. J. Mcbride, J. Treadway, L.C. Feldman, S.J. Pennycook, S.J. Rosenthal, Structural basis for
near unity quantum yield core/shell nanostructures. Nano Lett. 6, 1496–1501 (2006)
16. L. Liu, Q. Peng, Y. Li, Preparation of cdse quantum dots with full color emission based on a
room temperature injection technique. Inorg. Chem. 47, 5022–5028 (2008)
17. X. Chen, J.L. Hutchison, P.J. Dobson, G. Wakefield, Highly luminescent monodisperse cdse
nanoparticles synthesized in aqueous solution. J. Mater. Sci. 44, 285–292 (2009)
18. J.T. Siy, E.M. Brauser, M.H. Bartl, Low-temperature synthesis of cdse nanocrystal quantum
dots. Chem. Commun. 47, 364–366 (2011)
19. I.U. Arachchige, S.L. Brock, Sol-gel assembly of cdse nanoparticles to form porous aerogel
networks. J. Am. Chem. Soc. 128, 7964–7971 (2006)
20. M.R. Gaeeni, M. Tohidian, M. Mapes-Ara, Green synthesis of cdse colloidal nanocrystals
with strong green emission by the sol-gel method. Ind. Eng. Chem. Res. 53, 7598–7603
(2014)
21. I.U. Arachchige, S.L. Brock, Sol-gel methods for the assembly of metal chalcogenide
quantum dots. Acc. Chem. Res. 40, 801–809 (2007)
22. A.M. Derfus, W.C.W. Chan, S.N. Bhatia, Probing the cytotoxicity of semiconductor
quantum dots. Nano Lett. 4, 11–18 (2004)
23. W.C.W. Chan, S.M. Nie, Quantum dot bioconjugates for ultrasensitive nonisotopic
detection. Science 281, 2016–2018 (1998)
24. C. Kirchner, T. Liedl, S. Kudera, T. Pellegrino, A.M. Javier, H.E. Gaub, S. Stolzle, N. Fertig,
W.J. Parak, Cytotoxicity of colloidal CdSe and CdSe/ZnS nanoparticles. Nano Lett. 5, 331–
338 (2005)
25. U. Resch-Genger, M. Grabolle, S. Cavaliere-Jaricot, R. Nitschke, T. Nann, Quantum dots
versus organic dyes as fluorescent labels. Nat. Methods 5, 763–775 (2008)
26. W. Liu, H.S. Choi, J.P. Zimmer, E. Tanaka, J.V. Frangioni, M. Bawendi, Compact
cysteine-coated cdse(zncds) quantum dots for in vivo applications. J. Am. Chem. Soc. 129,
14530–14543 (2007)
27. A.S. Karakoti, R. Shukla, R. Shanker, S. Singh, Surface functionalization of quantum dots
for biological applications. Adv. Colloid Interface Sci. 215, 28–45 (2015)
28. S.F. Wuister, C.D. Donega, A. Meijerink, Influence of thiol capping on the exciton
luminescence and decay kinetics of CdTe and CdSe quantum. J. Phys. Chem. B 108, 17393–
17397 (2004)
528 M.K. Hatipoglu et al.

29. R. Pejchal, M.L. Ludwig, Cobalamin-independent methionine synthase (mete): a face-to-face

double barrel that evolved by gene duplication. PLoS Biol. 3, 254–265 (2005)
30. Y. Zhang, A. Clapp, Overview of stabilizing ligands for biocompatible quantum dot
nanocrystals. Sensors 11, 11036–11055 (2011)
31. M.H. Stewart, K. Susumu, B.C. Mei, I.L. Medintz, J.B. Delehanty, J.B. Blanco-Canosa, P.E.
Dawson, H. Mattoussi, Multidentate poly(ethylene glycol) ligands provide colloidal stability
to semiconductor and metallic nanocrystals in extreme conditions. J. Am. Chem. Soc. 132,
9804–9813 (2010)
32. A.T. Nagaraja, A. Sooresh, K.E. Meissner, M.J. Mcshane, Processing and characterization of
stable, ph-sensitive layer-by-layer modified colloidal quantum dots. ACS Nano 7, 6194–6202
(2013)
33. E. Gravel, C. Tanguy, E. Cassette, T. Pons, F. Knittel, N. Bernards, A. Garofalakis, F.
Duconge, B. Dubertret, E. Doris, Compact tridentate ligands for enhanced aqueous stability
of quantum dots and in vivo imaging. Chem. Sci. 4, 411–417 (2013)
34. L. Liu, X. Guo, Y. Li, X. Zhong, Bifunctional multidentate ligand modified highly stable
water-soluble quantum dots. Inorg. Chem. 49, 3768–3775 (2010)
35. H. Duan, M. Kuang, Y.A. Wangi, Quantum dots with multivalent and compact polymer
coatings for efficient fluorescence resonance energy transfer and self-assembled biotagging.
Chem. Mater. 22, 4372–4378 (2010)
36. G.R. Bardajee, R. Soleyman, Surface passivation of cdse-topo quantum dots by poly(acrylic
acid): solvent sensitivity and photo-induced emission in water. Iran. Polym. J. 22, 885–890
(2013)
37. M.A. Correa-Duarte, M. Giersig, L.M. Liz-Marzan, Stabilization of cds semiconductor
nanoparticles against photodegradation by a silica coating procedure. Chem. Phys. Lett. 286,
497–501 (1998)
38. M.A. Correa-Duarte, Y. Kobayashi, R.A. Caruso, L.M. Liz-Marzan, Photodegradation of
sio2-coated cds nanoparticles within silica gels. J. Nanosci. Nanotechnol. 1, 95–99 (2001)
39. A.G. Kanaras, F.S. Kamounah, K. Schaumburg, C.J. Kiely, M. Brust, Thioalkylated
tetraethylene glycol: a new ligand for water soluble monolayer protected gold clusters.
Chem. Commun. 2294–2295 (2002)
40. W. Stober, A. Fink, E. Bohn, Controlled growth of monodisperse silica spheres in micron
size range. J. Colloid Interface Sci. 26, 62 (1968)
41. M. Darbandi, R. Thomann, T. Nann, Single quantum dots in silica spheres by microemulsion
synthesis. Chem. Mater. 17, 5720–5725 (2005)
42. R.E. Anderson, W.C.W. Chan, Systematic investigation of preparing biocompatible, single,
and small ZnS-capped CdSe quantum dots with amphiphilic polymers. ACS Nano 2, 1341–
1352 (2008)
43. R.C. Somers, P.T. Snee, M.G. Bawendi, D.G. Nocera, Energy transfer of CdSe/ZnS
nanocrystals encapsulated with rhodamine-dye functionalized poly(acrylic acid).
J. Photochem. Photobiol. Chem. 248, 24–29 (2012)
44. C. Luccardini, C. Tribet, F. Vial, V. Marchi-Artzner, M. Dahan, Size, charge, and
interactions with giant lipid vesicles of quantum dots coated with an amphiphilic
macromolecule. Langmuir 22, 2304–2310 (2006)
45. W. Regel, C. Schneider, Poly(maleic anhydride) - synthesis and proof of structure.
Macromol. Chem. Phys. Makromol. Chem. 182, 237–242 (1981)
46. T. Pellegrino, L. Manna, S. Kudera, T. Liedl, D. Koktysh, A.L. Rogach, S. Keller, J. Radler,
G. Natile, W.J. Parak, Hydrophobic nanocrystals coated with an amphiphilic polymer shell: a
general route to water soluble nanocrystals. Nano Lett. 4, 703–707 (2004)
47. R. Di Corato, A. Quarta, P. Piacenza, A. Ragusa, A. Figuerola, R. Buonsanti, R. Cingolani,
L. Manna, T. Pellegrino, Water solubilization of hydrophobic nanocrystals by means of poly
(maleic anhydride-alt-1-octadecene). J. Mater. Chem. 18, 1991–1996 (2008)
48. L. Qi, X. Gao, Quantum dot-amphipol nanocomplex for intracellular delivery and real-time
imaging of sirna. ACS Nano 2, 1403–1410 (2008)
20 Synthesis and Biological Applications of Quantum Dots 529

49. N. Tomczak, R. Liu, J.G. Vancso, Polymer-coated quantum dots. Nanoscale 5, 12018–12032
(2013)
50. X. Nie, J. Xu, J. Cui, B. Yang, W. Jiang, Encapsulation of semiconductor quantum dots into
the central cores of block copolymer cylindrical and toroidal micelles. RSC Adv. 3, 24625–
24633 (2013)
51. Y.-K. Lee, S.M. Hong, J.S. Kim, J.H. Im, H.S. Min, E. Subramanyam, K.M. Huh,
Encapsulation of CdSe/ZnS quantum dots in poly(ethylene glycol)-poly(d, l-lactide) micelle
for biomedical imaging and detection. Macromol. Res. 15, 330–336 (2007)
52. S. Ghaderi, B. Ramesh, A.M. Seifalian, Fluorescence nanoparticles “quantum dots” as drug
delivery system and their toxicity: a review. J. Drug Target. 19, 475–486 (2011)
53. J. Dimitrijevic, L. Krapf, C. Wolter, C. Schmidtke, J.-P. Merkl, T. Jochum, A. Kornowski, A.
Schueth, A. Gebert, G. Huettmann, T. Vossmeyer, H. Weller, Cdse/cds-quantum rods:
fluorescent probes for in vivo two-photon laser scanning microscopy. Nanoscale 6, 10413–
10422 (2014)
54. V. Kulvietis, E. Zurauskas, R. Rotomskis, Distribution of polyethylene glycol coated
quantum dots in mice skin. Exp. Dermatol. 22, 157–159 (2013)
55. E.L. Bentzen, I.D. Tomlinson, J. Mason, P. Gresch, M.R. Warnement, D. Wright, E.
Sanders-Bush, R. Blakely, S.J. Rosenthal, Surface modification to reduce nonspecific
binding of quantum dots in live cell assays. Bioconjug. Chem. 16, 1488–1494 (2005)
56. I. Yildiz, E. Deniz, B. Mccaughan, S.F. Cruickshank, J.F. Callan, F.M. Raymo, Hydrophilic
CdSe-ZnS core-shell quantum dots with reactive functional groups on their surface.
Langmuir 26, 11503–11511 (2010)
57. Z. Jiang, C. Zhao, X. Liu, Synthesis of poly(ethylene glycol)-graft-chitosan and using as
ligand for fabrication of water-soluble quantum dots. Colloids Surf., B 115, 260–266 (2014)
58. A. Papagiannaros, T. Levchenko, W. Hartner, D. Mongayt, V. Torchilin, Quantum dots
encapsulated in phospholipid micelles for imaging and quantification of tumors in the
near-infrared region. Nanomed. Nanotechnol. Biol. Med. 5, 216–224 (2009)
59. V.P. Torchilin, Micellar nanocarriers: pharmaceutical perspectives. Pharm. Res. 24, 1–16
(2007)
60. F. Erogbogbo, K.-T. Yong, R. Hu, W.-C. Law, H. Ding, C.-W. Chang, P.N. Prasad, M.T.
Swihart, Biocompatible magnetofluorescent probes: luminescent silicon quantum dots
coupled with superparamagnetic iron(iii) oxide. ACS Nano 4, 5131–5138 (2010)
61. J. Pathak, K. Rawat, S. Sanwlani, H.B. Bohidar, Interaction of globular plasma proteins with
water-soluble cdse quantum dots. ChemPhysChem 16, 1777–1786 (2015)
62. J. Lee, Y. Choi, J. Kim, E. Park, R. Song, Positively charged compact quantum dot-DNA
complexes for detection of nucleic acids. ChemPhysChem 10, 806–811 (2009)
63. W.R. Algar, U.J. Krull, Adsorption and hybridization of oligonucleotides on mercaptoacetic
acid-capped CdSe/ZnS quantum dots and quantum dot-oligonucleotide conjugates. Langmuir
22, 11346–11352 (2006)
64. P. Liu, N. Na, L. Huang, D. He, C. Huang, J. Ouyang, The application of amine-terminated
silicon quantum dots on the imaging of human serum proteins after polyacrylamide gel
electrophoresis (page). Chem. Eur. J. 18, 1438–1443 (2012)
65. S.M. Rozenzhak, M.P. Kadakia, T.M. Caserta, T.R. Westbrook, M.O. Stone, R.R. Naik,
Cellular internalization and targeting of semiconductor quantum dots. Chem. Commun.
2217–2219 (2005)
66. I.L. Medintz, S.A. Trammell, H. Mattoussi, J.M. Mauro, Reversible modulation of quantum
dot photoluminescence using a protein-bound photochromic fluorescence resonance energy
transfer acceptor. J. Am. Chem. Soc. 126, 30–31 (2004)
67. J.B. Delehanty, I.L. Medintz, T. Pons, F.M. Brunel, P.E. Dawson, H. Mattoussi,
Self-assembled quantum dot-peptide bioconjugates for selective intracellular delivery.
Bioconjug. Chem. 17, 920–927 (2006)
68. D. Zhou, L. Ying, X. Hong, E.A. Hall, C. Abell, D. Klenerman, A compact functional
quantum dot-DNA conjugate: preparation, hybridization, and specific label-free DNA
detection. Langmuir 24, 1659–1664 (2008)
530 M.K. Hatipoglu et al.

69. G.H.T. Au, L. Mejias, V.K. Swami, A.D. Brooks, W.Y. Shih, W.-H. Shih, Quantitative
assessment of tn antigen in breast tissue micro-arrays using cdse aqueous quantum dots.
Biomaterials 35, 2971–2980 (2014)
70. W. Cai, X. Chen, Preparation of peptide-conjugated quantum dots for tumor
vasculature-targeted imaging. Nat. Protoc. 3, 89–96 (2008)
71. A. Samanta, Z. Deng, Y. Liu, Aqueous synthesis of glutathione-capped CdTe/CdS/ZnS and
CdTe/CdSe/ZnS core/shell/shell nanocrystal heterostructures. Langmuir 28, 8205–8215
(2012)
72. R. Bilan, F. Fleury, I. Nabiey, A. Sukhanova, Quantum dot surface chemistry and
functionalization for cell targeting and imaging. Bioconjug. Chem. 26, 609–624 (2015)
73. A. Bernardin, A. Cazet, L. Guyon, P. Delannoy, F. Vinet, D. Bonnaffe, I. Texier, Copper-free
click chemistry for highly luminescent quantum dot conjugates: application to in vivo
metabolic imaging. Bioconjug. Chem. 21, 583–588 (2010)
74. W.H. Binder, R. Sachsenhofer, C.J. Straif, R. Zirbs, Surface-modified nanoparticles via
thermal and cu(i)-mediated “click” chemistry: generation of luminescent CdSe nanoparticles
with polar ligands guiding supramolecular recognition. J. Mater. Chem. 17, 2125–2132
(2007)
75. M.B. Meza, Bead-based HTS applications in drug discovery. Drug Discov. Today 38–41
(2000)
76. C. Schmidtke, A.-M. Kreuziger, D. Alpers, A. Jacobsen, Y. Leshch, R. Eggers, H. Kloust, T.
Huong, J. Ostermann, T. Schotten, J. Thiem, J. Thimm, H. Weller, Glycoconjugated
amphiphilic polymers via click-chemistry for the encapsulation of quantum dots. Langmuir
29, 12593–12600 (2013)
77. M. J t, Quantum dots finally come of age. Nat. Biotechnol. 21, 32–33 (2003)
78. A.M. Smith, X.H. Gao, S.M. Nie, Quantum dot nanocrystals for in vivo molecular and
cellular imaging. Photochem. Photobiol. 80, 377–385 (2004)
79. P. Alivisatos, The use of nanocrystals in biological detection. Nat. Biotechnol. 22, 47–52
(2003)
80. A.P. Alivisatos, W.W. Gu, C. Larabell, Quantum dots as cellular probes, in Quantum Dots as
Cellular Probes (Annual Reviews, Palo Alto. 2005), pp. 55–76
81. M.P. Bruchez, Turning all the lights on: quantum dots in cellular assays. Curr. Opin. Chem.
Biol. 9, 533–537 (2005)
82. A. Smith, H. Duan, A. Mohs, S. Nie, Bioconjugated quantum dots for in vivo molecular and
cellular imaging. Adv. Drug Deliv. Rev. 60, 1226–1240 (2008)
83. K.K. Jain, Nanotechnology in clinical laboratory diagnostics. Clin. Chim. Acta 358, 37–54
(2005)
84. X. Michalet, F.F. Pinaud, L.A. Bentolila, J.M. Tsay, S. Doose, J.J. Li, G. Sundaresan, A.M.
Wu, S.S. Gambhir, S. Weiss, Quantum dots for live cells, in vivo imaging, and diagnostics.
Science 307, 538–544 (2005)
85. A.M. Smith, S. Dave, S.M. Nie, L. True, X.H. Gao, Multicolor quantum dots for molecular
diagnostics of cancer. Expert Rev. Mol. Diagn. 6, 231–244 (2006)
86. W.W. Yu, E. Chang, R. Drezek, V.L. Colvin, Water-soluble quantum dots for biomedical
applications. Biochem. Biophys. Res. Commun. 348, 781–786 (2006)
87. K.K. Jain, Nanodiagnostics: application of nanotechnology in molecular diagnostics. Expert
Rev. Mol. Diagn. 3, 153–161 (2003)
88. M. Bruchez, M. Moronne, P. Gin, S. Weiss, A.P. Alivisatos, Semiconductor nanocrystals as
fluorescent biological labels. Science 281, 2013–2016 (1998)
89. S.C. Makrides, C. Gasbarro, J.M. Bello, Bioconjugation of quantum dot luminescent probes
for western blot analysis. Biotechniques 39, 501–506 (2005)
90. J.K. Jaiswal, S.M. Simon, Potentials and pitfalls of fluorescent quantum dots for biological
imaging. Trends Cell Biol. 14, 497–504 (2004)
91. V. Biju, S. Mundayoor, R.V. Omkumar, A. Anas, M. Ishikawa, Bioconjugated quantum dots
for cancer research: present status, prospects and remaining issues. Biotechnol. Adv. 28,
199–213 (2010)
20 Synthesis and Biological Applications of Quantum Dots 531

92. N. Kawashima, K. Nakayama, K. Itoh, T. Itoh, M. Ishikawa, V. Biju, Reversible

dimerization of egfr revealed by single-molecule fluorescence imaging using quantum
dots. Chem. Eur. J. 16, 1186–1192 (2010)
93. H. Tada, H. Higuchi, T.M. Wanatabe, N. Ohuchi, In vivo real-time tracking of single
quantum dots conjugated with monoclonal anti-her2 antibody in tumors of mice. Cancer Res.
67, 1138–1144 (2007)
94. K.C. Weng, C.O. Noble, B. Papahadjopoulos-Sternberg, F.F. Chen, D.C. Drummond, D.B.
Kirpotin, D. Wang, Y.K. Hom, B. Hann, J.W. Park, Targeted tumor cell internalization and
imaging of multifunctional quantum dot-conjugated immunoliposomes in vitro and in vivo.
Nano Lett. 8, 2851–2857 (2008)
95. H. Zhang, D. Sachdev, C. Wang, A. Hubel, M. Gaillard-Kelly, D. Yee, Detection and
downregulation of type i igf receptor expression by antibody-conjugated quantum dots in
breast cancer cells. Breast Cancer Res. Treat. 114, 277–285 (2009)
96. K.-T. Yong, H. Ding, I. Roy, W.-C. Law, E.J. Bergey, A. Maitra, P.N. Prasad, Imaging
pancreatic cancer using bioconjugated inp quantum dots. ACS Nano 3, 502–510 (2009)
97. D.S. Lidke, P. Nagy, R. Heintzmann, D.J. Arndt-Jovin, J.N. Post, H.E. Grecco, E.A.
Jares-Erijman, T.M. Jovin, Quantum dot ligands provide new insights into erbb/her
receptor-mediated signal transduction. Nat. Biotechnol. 22, 198–203 (2004)
98. X.H. Gao, Y.Y. Cui, R.M. Levenson, L.W.K. Chung, S.M. Nie, In vivo cancer targeting and
imaging with semiconductor quantum dots. Nat. Biotechnol. 22, 969–976 (2004)
99. M.E. Akerman, W.C.W. Chan, P. Laakkonen, S.N. Bhatia, E. Ruoslahti, Nanocrystal
targeting in vivo. Proc. Nat. Acad. Sci. U.S.A. 99, 12617–12621 (2002)
100. D.R. Larson, W.R. Zipfel, R.M. Williams, S.W. Clark, M.P. Bruchez, F.W. Wise, W.W.
Webb, Water-soluble quantum dots for multiphoton fluorescence imaging in vivo. Science
300, 1434–1436 (2003)
101. X. Ji, F. Peng, Y. Zhong, Y. Su, Y. He, Fluorescent quantum dots: synthesis, biomedical
optical imaging, and biosafety assessment. Colloids Surf., B 124, 132–139 (2014)
102. P. Zrazhevskiy, X.H. Gao, Quantum dot imaging platform for single-cell molecular profiling.
Nat. Commun. 4, 12 (2013)
103. I.L. Medintz, A.R. Clapp, H. Mattoussi, E.R. Goldman, B. Fisher, J.M. Mauro,
Self-assembled nanoscale biosensors based on quantum dot fret donors. Nat. Mater. 2,
630–638 (2003)
104. L. Igor, H.T.U. Medintz, Ellen R. Goldman, Hedi Mattoussi, Quantum dot bioconjugates for
imaging, labelling and sensing. Nat. Mater. 4, 435–446 (2005)
105. I.L. Medintz, J.H. Konnert, A.R. Clapp, I. Stanish, M.E. Twigg, H. Mattoussi, J.M. Mauro, J.
R. Deschamps, A fluorescence resonance energy transfer-derived structure of a quantum
dot-protein bioconjugate nanoassembly. Proc. Nat. Acad. Sci. U.S.A. 101, 9612–9617 (2004)
106. F. Pinaud, D. King, H.P. Moore, S. Weiss, Bioactivation and cell targeting of semiconductor
CdSe/ZnS nanocrystals with phytochelatin-related peptides. J. Am. Chem. Soc. 126, 6115–
6123 (2004)
107. J.M. De La Fuente, M. Fandel, C.C. Berry, M. Riehle, L. Cronin, G. Aitchison, A.S.G.
Curtis, Quantum dots protected with tiopronin: a new fluorescence system for cell-biology
studies. Chembiochem 6, 989–991 (2005)
108. B.C. Lagerholm, M.M. Wang, L.A. Ernst, D.H. Ly, H.J. Liu, M.P. Bruchez, A.S. Waggoner,
Multicolor coding of cells with cationic peptide coated quantum dots. Nano Lett. 4, 2019–
2022 (2004)
109. S.K. Chakraborty, JaJ Fitzpatrick, J.A. Phillippi, S. Andreko, A.S. Waggoner, M.P. Bruchez,
B. Ballou, Cholera toxin b conjugated quantum dots for live cell labeling. Nano Lett. 7,
2618–2626 (2007)
110. V. Biju, D. Muraleedharan, K. Nakayama, Y. Shinohara, T. Itoh, Y. Baba, M. Ishikawa,
Quantum dot-insect neuropeptide conjugates for fluorescence imaging, transfection, and
nucleus targeting of living cells. Langmuir 23, 10254–10261 (2007)
111. S.J. Rosenthal, J.C. Chang, O. Kovtun, J.R. Mcbride, I.D. Tomlinson, Biocompatible
quantum dots for biological applications. Chem. Biol. 18, 10–24 (2011)
532 M.K. Hatipoglu et al.

112. G. Ruan, A. Agrawal, A.I. Marcus, S. Nie, Imaging and tracking of tat peptide-conjugated
quantum dots in living cells: new insights into nanoparticle uptake, intracellular transport,
and vesicle shedding. J. Am. Chem. Soc. 129, 14759–14766 (2007)
113. B.R. Smith, Z. Cheng, A. De, A.L. Koh, R. Sinclair, S.S. Gambhir, Real-time intravital
imaging of rgd-quantum dot binding to luminal endothelium in mouse tumor neovasculature.
Nano Lett. 8, 2599–2606 (2008)
114. K.M. Fichter, M. Flajolet, P. Greengard, T.Q. Vu, Kinetics of g-protein-coupled receptor
endosomal trafficking pathways revealed by single quantum dots. Proc. Nat. Acad. Sci. U.S.
A. 107, 18658–18663 (2010)
115. V. Bagalkot, L. Zhang, E. Levy-Nissenbaum, S. Jon, P.W. Kantoff, R. Langer, O.C.
Farokhzad, Quantum dot—aptamer conjugates for synchronous cancer imaging, therapy, and
sensing of drug delivery based on bi-fluorescence resonance energy transfer. Nano Lett. 7,
3065–3070 (2007)
116. S.J. Rosenthal, A. Tomlinson, E.M. Adkins, S. Schroeter, S. Adams, L. Swafford, J. Mcbride,
Y.Q. Wang, L.J. Defelice, R.D. Blakely, Targeting cell surface receptors with
ligand-conjugated nanocrystals. J. Am. Chem. Soc. 124, 4586–4594 (2002)
117. H.A. Gussin, I.D. Tomlinson, D.M. Little, M.R. Warnement, H. Qian, S.J. Rosenthal, D.R.
Pepperberg, Binding of muscimol-conjugated quantum dots to gaba(c) receptors. J. Am.
Chem. Soc. 128, 15701–15713 (2006)
118. I.D. Tomlinson, J.N. Mason, R.D. Blakely, S.J. Rosenthal, High affinity inhibitors of the
dopamine transporter (dat): novel biotinylated ligands for conjugation to quantum dots.
Bioorg. Med. Chem. Lett. 16, 4664–4667 (2006)
119. I.D. Tomlinson, M.R. Warnerment, J.N. Mason, M.J. Vergne, D.M. Hercules, R.D. Blakely,
S.J. Rosenthal, Synthesis and characterization of a pegylated derivative of 3-
(1,2,3,6-tetrahydro-pyridin-4yl)-1 h-indole (idt199): a high affinity sert ligand for
conjugation to quantum dots. Bioorg. Med. Chem. Lett. 17, 5656–5660 (2007)
120. H.S. Choi, W. Liu, F. Liu, K. Nasr, P. Misra, M.G. Bawendi, J.V. Frangioni, Design
considerations for tumour-targeted nanoparticles. Nat. Nanotechnol. 5, 42–47 (2010)
121. S.J. Byrne, B. Le Bon, S.A. Corr, M. Stefanko, C. O’connor, Y.K. Gun’ko, Y.P. Rakovich, J.
F. Donegan, Y. Williams, Y. Volkov, P. Evans, Synthesis, characterisation, and biological
studies of CdTe quantum dot-naproxen conjugates. Chemmedchem 2, 183–190 (2007)
122. S.J. Clarke, C.A. Hollmann, Z.J. Zhang, D. Suffern, S.E. Bradforth, N.M. Dimitrijevic, W.G.
Minarik, J.L. Nadeau, Photophysics of dopamine-modified quantumdots and effects on
biological systems. Nat. Mater. 5, 409–417 (2006)
123. C. Tortiglione, A. Quarta, A. Tino, L. Manna, R. Cingolani, T. Pellegrino, Synthesis and
biological assay of gsh functionalized fluorescent quantum dots for staining hydra vulgaris.
Bioconjug. Chem. 18, 829–835 (2007)
124. R. Weissleder, A clearer vision for in vivo imaging. Nat. Biotechnol. 19, 316–317 (2001)
125. J.V. Frangioni, In vivo near-infrared fluorescence imaging. Curr. Opin. Chem. Biol. 7, 626–
634 (2003)
126. S. Kim, Y.T. Lim, E.G. Soltesz, A.M. De Grand, J. Lee, A. Nakayama, J.A. Parker, T.
Mihaljevic, R.G. Laurence, D.M. Dor, L.H. Cohn, M.G. Bawendi, J.V. Frangioni,
Near-infrared fluorescent type ii quantum dots for sentinel lymph node mapping. Nat.
Biotechnol. 22, 93–97 (2004)
127. B. Ballou, B.C. Lagerholm, L.A. Ernst, M.P. Bruchez, A.S. Waggoner, Noninvasive imaging
of quantum dots in mice. Bioconjug. Chem. 15, 79–86 (2004)
128. H.C. Fischer, L. Liu, K.S. Pang, W.C.W. Chan, Pharmacokinetics of nanoscale quantum
dots: in vivo distribution, sequestration, and clearance in the rat. Adv. Funct. Mater. 16,
1299–1305 (2006)
129. E.G. Soltesz, S. Kim, R.G. Laurence, A.M. Degrand, C.P. Parungo, D.M. Dor, L.H. Cohn,
M.G. Bawendi, J.V. Frangioni, T. Mihaljevic, Intraoperative sentinel lymph node mapping of
the lung using near-infrared fluorescent quantum dots. Ann. Thorac. Surg. 79, 269–277
(2005)
20 Synthesis and Biological Applications of Quantum Dots 533

130. M.E. Akerman, W.C.W. Chan, P. Laakkonen, S.N. Bhatia, E. Ruoslahti, Nanocrystal
targeting in vivo. Proc. Nat. Acad. Sci. 99, 12617–12621 (2002)
131. B. Dubertret, P. Skourides, D.J. Norris, V. Noireaux, A.H. Brivanlou, A. Libchaber, In vivo
imaging of quantum dots encapsulated in phospholipid micelles. Science 298, 1759–1762
(2002)
132. A. Karwa, E. Papazoglou, K. Pourrezaei, S. Tyagi, S. Murthy, Imaging biomarkers of
inflammation in situ with fun ctionalized quantum dots in the dextran sodium sulfate
(dss) model of mouse colitis. Inflamm. Res. 56, 502–510 (2007)
133. A. Jayagopal, P.K. Russ, F.R. Haselton, Surface engineering of quantum dots for in vivo
vascular imaging. Bioconjug. Chem. 18, 1424–1433 (2007)
134. M. Stroh, J.P. Zimmer, D.G. Duda, T.S. Levchenko, K.S. Cohen, E.B. Brown, D.T. Scadden,
V.P. Torchilin, M.G. Bawendi, D. Fukumura, R.K. Jain, Quantum dots spectrally distinguish
multiple species within the tumor milieu in vivo. Nat. Med. 11, 678–682 (2005)
135. M. Rubart, Two-photon microscopy of cells and tissue. Circ. Res. 95, 1154–1166 (2004)
136. J.D. Smith, G.W. Fisher, A.S. Waggoner, P.G. Campbell, The use of quantum dots for
analysis of chick cam vasculature. Microvasc. Res. 73, 75–83 (2007)
137. X.G. Peng, M.C. Schlamp, A.V. Kadavanich, A.P. Alivisatos, Epitaxial growth of highly
luminescent cdse/cds core/shell nanocrystals with photostability and electronic accessibility.
J. Am. Chem. Soc. 119, 7019–7029 (1997)
138. C.P. Parungo, Y.L. Colson, S.W. Kim, S. Kim, L.H. Cohn, M.G. Bawendi, J.V. Frangioni,
Sentinel lymph node mapping of the pleural space. Chest 127, 1799–1804 (2005)
139. B. Ballou, L.A. Ernst, S. Andreko, T. Harper, JaJ Fitzpatrick, A.S. Waggoner, M.P. Bruchez,
Sentinel lymph node imaging using quantum dots in mouse tumor models. Bioconjug. Chem.
18, 389–396 (2007)
140. H. Kobayashi, Y. Hama, Y. Koyama, T. Barrett, C.a S. Regino, Y. Urano, P.L. Choyke,
Simultaneous multicolor imaging of five different lymphatic basins using quantum dots.
Nano Lett. 7, 1711–1716 (2007)
141. X. Ji, F. Peng, Y. Zhong, Y. Su, Y. He, Fluorescent quantum dots: synthesis, biomedical
optical imaging, and biosafety assessment. Colloids Surf., B 124, 132–139 (2014)
142. J. Wang, Y. Lu, F. Peng, Y. Zhong, Y. Zhou, X. Jiang, Y. Su, Y. He, Photostable
water-dispersible nir-emitting CdTe/CdS/ZnS core-shell-shell quantum dots for
high-resolution tumor targeting. Biomaterials 34, 9509–9518 (2013)
143. T. Forster, Zwischenmolekulare energiewanderung und fluoreszenz. Annalen Der Physik 2,
55–75 (1948)
144. A. Miyawaki, Visualization of the spatial and temporal dynamics of intracellular signaling.
Dev. Cell 4, 295–305 (2003)
145. E.A. Jares-Erijman, T.M. Jovin, Fret imaging. Nat. Biotechnol. 21, 1387–1395 (2003)
146. W.R. Algar, U.J. Krull, Quantum dots as donors in fluorescence resonance energy transfer for
the bioanalysis of nucleic acids, proteins, and other biological molecules. Anal. Bioanal.
Chem. 391, 1609–1618 (2008)
147. N. Mcgrath, M. Barroso, Quantum dots as fluorescence resonance energy transfer donors in
cells. J. Biomed. Opt. 13–21 (2008)
148. A.M. Dennis, D.C. Sotto, B.C. Mei, I.L. Medintz, H. Mattoussi, G. Bao, Surface ligand
effects on metal-affinity coordination to quantum dots: implications for nanoprobe
self-assembly. Bioconjug. Chem. 21, 1160–1170 (2010)
149. A.M. Dennis, G. Bao, Quantum dot-fluorescent protein pairs as novel fluorescence resonance
energy transfer probes. Nano Lett. 8, 1439–1445 (2008)
150. S.A. Mckinney, A.C. Declais, D.M.J. Lilley, T. Ha, Structural dynamics of individual
holliday junctions. Nat. Struct. Biol. 10, 93–97 (2003)
151. S. Hohng, T. Ha, Single-molecule quantum-dot fluorescence resonance energy transfer.
Chemphyschem 6, 956–960 (2005)
152. C.Y. Zhang, H.C. Yeh, M.T. Kuroki, T.H. Wang, Single-quantum-dot-based DNA
nanosensor. Nat. Mater. 4, 826–831 (2005)
534 M.K. Hatipoglu et al.

153. A.R. Clapp, I.L. Medintz, J.M. Mauro, B.R. Fisher, M.G. Bawendi, H. Mattoussi,
Fluorescence resonance energy transfer between quantum dot donors and dye-labeled protein
acceptors. J. Am. Chem. Soc. 126, 301–310 (2004)
154. I.L. Medintz, H. Mattoussi, Quantum dot-based resonance energy transfer and its growing
application in biology. Phys. Chem. Chem. Phys. 11, 17–45 (2009)
155. S. Pathak, S.K. Choi, N. Arnheim, M.E. Thompson, Hydroxylated quantum dots as
luminescent probes for in situ hybridization. J. Am. Chem. Soc. 123, 4103–4104 (2001)
156. F. Patolsky, R. Gill, Y. Weizmann, T. Mokari, U. Banin, I. Willner, Lighting-up the
dynamics of telomerization and DNA replication by CdSe-ZnS quantum dots. J. Am. Chem.
Soc. 125, 13918–13919 (2003)
157. Z. Gueroui, A. Libchaber, Single-molecule measurements of gold-quenched quantum dots.
Phys. Rev. Lett. 93 (2004)
158. L. Dyadyusha, H. Yin, S. Jaiswal, T. Brown, J.J. Baumberg, F.P. Booy, T. Melvin,
Quenching of CdSe quantum dot emission, a new approach for biosensing. Chem. Commun.
3201–3203 (2005)
159. P.M. Chan, T. Yuen, F. Ruf, J. Gonzalez-Maeso, S.C. Sealfon, Method for multiplex cellular
detection of mRNAs using quantum dot fluorescent in situ hybridization. Nucleic Acids Res.
33 (2005)
160. A.A. Chen, A.M. Derfus, S.R. Khetani, S.N. Bhatia, Quantum dots to monitor RNAi delivery
and improve gene silencing. Nucleic Acids Res. 33 (2005)
Chapter 21
Bionanotechnology: Lessons from Nature
for Better Material Properties

F.N. Kök

Abstract For millions of years, nature has built hierarchically organized intricate
systems with interesting material properties that synthetic materials often fail to
replicate. With the advances in instrumentation for both characterization and
manipulation in nano-scale, it has now become possible to comprehend the
molecular mechanisms and structures behind that success and mimic them.
Biomimicry should not be understood as a superficial imitation of the biological
systems. It should rather be interpreted as the inspiration from the structure-function
relationships observed in biological systems to construct new hierarchical structures
with improved properties.

21.1 Introduction

For centuries, humankind has been both astonished by and inspired from the nature
and tried to understand and mimic the natural systems to establish new materials
and machines. Long before the invention of sophisticated instruments to observe
and manipulate in micro or nano-scale, bat wings had been an inspiration for
Leonardo da Vinci to design a flying machine or imitation of human arm move-
ments had been used to design robotic arms [1]. In today’s world, however, the
concept of hierarchical design could be understood starting from molecular level
and this gives us the opportunity to design smart, highly sophisticated materials
starting from molecules or functional modules like lipid membranes, ion channels,
motor proteins, etc.
Nature not only gives inspiration for better materials or systems but also for the
establishment of less power hungry production technologies with minimum waste
generation. The synthesis of sophisticated materials was done under mild reaction

F.N. Kök (&)

Molecular Biology and Genetics Department, Istanbul Technical University,
34469 Maslak, Istanbul, Turkey
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 535

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_21
536 F.N. Kök

conditions from aqueous precursors as opposed to the conventional technologies

which require the usage of organic solvents and extreme conditions (high tem-
perature, pressure, acidity, etc.). This makes the bioinspired technologies more
environmental friendly and the materials produced tend to lack the toxic residues;
all this makes them intrinsically biocompatible [2]. The applications of these
materials are not only limited to medical areas. A diverse industrial field ranging
from photovoltaics to catalysis and pollution control could take advantage of these
attractive production technologies.
The process of self-assembly, which could be defined as the autonomous
organization of components into patterns or structures without human intervention
is one of the most important mechanisms behind the biological structures and
function [3]. Self-assembly processes depend on non-covalent bonds (e.g. hydrogen
bonds, ionic bonds, and hydrophobic interactions) which are individually weak but
extremely effective working in harmony for the construction of strong structures.
Folded polypeptides, assembly of pore forming peptides, amyloid forming peptides,
cell bilayer membrane, prokaryotic S-layers and lipid tubules, viral particles are
some of the examples of the diverse processes and structures which are organized
by self-assembly.
Self-assembly is also an attractive bottom-up strategy for the fabrication of
nanostructures owing to highly ordered nature of the process and considerable
degree of flexibility and diversity provided by biological molecules [4, 5].
Understanding these mechanisms helps scientists to design novel systems
exhibiting unusual properties, even in nano-scale. These natural processes could be
combined by inorganic processes and hybrid materials could also be designed.
Supramolecular self-assembly and inorganic sol–gel polymerization, for example,
could be combined to form higher self-organized hybrid materials with complex
design and intricate properties [6]. A single type of molecule is rarely used in nature
and materials are mostly arranged in composite systems. By this mean, a single
building block, like collagen, is used as a major component in different tissues that
require different properties, by applying different chemistries and assemblies [7].
Implementation of self-organization to obtain self-organized hybrid materials was
explained in different studies and reviews [6, 8]. In this chapter, an overview of
diverse applications will be given.

21.2 Biomineralization

Biomineralization results in materials with different size and shapes that take part in
many interesting activities in living organisms. Although mechanical works like
support or chewing and ion storage are the most prominent functions of these
biominerals, they could also be used as navigation and optical tools. These minerals
may be extremely ordered and are referred to as crystalline minerals (or biogenic
crystals), or they may be relatively disordered and are referred to as amorphous
minerals [9]. Biomineralization is an organic matrix-mediated process rather than a
21 Bionanotechnology: Lessons from Nature … 537

simple thermodynamic and kinetic process of crystallization of inorganic mineral

ions [10]. These biological determinants, which self-assemble and used as templates
to control the nucleation, growth or morphology, have not been completely
resolved yet, even in the systems where the molecules which are crucial for the
event are known [11].
Mineralized tissue composed of calcium phosphate (e.g. bone and teeth) and
calcium carbonate (e.g. seashells) is one of the most common examples for
biomineralization. In the form of hydroxyapatite (HA), calcium phosphate is the
most common mineral in vertebrate tissues providing mechanical durability to
bones and teeth. In addition, being a stable, biocompatible, non-toxic and cheap
inorganic material, it is a material of choice for the coating of hard tissue implants
as well as carriers or ingredients in cosmetics and pharmaceutical applications [12,
13]. Calcium phosphate mineralization is therefore an important inspiration for
scientist to construct highly organized, bioactive materials at near-ambient condi-
tions avoiding high temperatures or organic solvents which are required for con-
ventional production [14]. This process is generally done by bioinspired
precipitation of inorganic minerals from aqueous solution using peptides, proteins,
protein mimics, synthetic peptides, carbohydrates and DNA [15].
Although they are less abundant than calcium phosphate/carbonate minerals,
silica from diatoms [16] and sponges [17] and magnetite from magnetic bacteria
[11] are other interesting materials synthesized in nature. Magnetite, for example, is
an iron oxide mineral that is found in various organisms ranging from unicellular
bacteria to fish and it serves as a geonavigational or mechanical tool. In magne-
totactic bacteria, in which its synthesis has been studied in detail, these monodis-
perse magnetic nanoparticles form chains which termed as magnetosomes [11].
Their potential in medical applications as MRI contrast agents makes in vitro
synthesis of these materials attractive. For this, a typical strategy is to first identify
the biologically determinant molecules and their characterization. These molecules
are generally used for template-based synthesis and this method to syntheses
inorganic materials was well-established [11, 15, 18, 19]. Consequently the efforts
have been focused on the production—by similar approach—of materials, which
are not biomineralized by living organisms (e.g. gold, silver, platinum) [14].
The discovery of new template molecules for both known and new biominerals
could be done by two different approaches. In rational design, in silico modelling
techniques based on existing knowledge was used to design new sequences [20].
The other method, on the other hand, requires no prior knowledge on existing
molecules and basically depends on an evolutionary approach involving the
selection of peptides from random peptide libraries constructed with surface display
methods [11, 21, 22]. By successive amplification-selection cycles, high affinity
peptides could be enriched. However, one should keep in mind that strong binding
affinity towards the minerals does not always translate into better biomineralization
directing ability and peptides with no significant binding affinity towards hydrox-
yapatite, for example, resulted in faster kinetics in biomineralization [19]. This
could be explained by the interaction of the peptide with the soluble precursor ions
rather than the mineral surface that the selection was actually done. Apart from
538 F.N. Kök

peptides, various polymers were also tried for their potential to assist bioinspired
calcium phosphate mineralization from aqueous solution and gave successful
results [15, 18].

21.3 Biomimetic Proteins: Receptors, Catalysts, Channels

Although proteins provide a selective recognition for many different molecules,

lack of stability and high production cost often limit their usage in high-throughput
systems. Strategies of supramolecular chemistry, on the other hand, offer an
alternative to design new biomimetic proteins inspired from the biological ana-
logues but providing better stability and applicability in production and operation.
Most widely employed strategy involves the usage of template-directed syntheses
and molecular imprinting technologies to produce molecularly imprinted polymers
(MIPs) [23]. MIPs are generally produced to mimic natural receptors and therefore
called as synthetic receptors. They are mostly synthesized by copolymerizing
functional monomers in the presence of a particular template molecule (Fig. 21.1)
[24]. When the target is removed, cavities are obtained with a high affinity and
selectivity for the corresponding template molecule. MIPs that could selectively
recognize small molecules like histamine, serotonin and L-nicotine can be produced
and used in the biosensor construction [25].

Fig. 21.1 Schematic representation of the molecular imprinting process. Reversible interactions
between the template and the polymerizable functional monomer can involve different
mechanisms explained in figure. After polymerization, a porous matrix in which the template
sites are located is formed. Removal of the template leaves an empty area defined by the shape of
the template and the specific interactions that it forms with functional monomers. The target
analyte can selectively bind to this imprint. Reprinted with permission from Alexander et al. [24].
Copyright (2006) John Wiley and Sons
21 Bionanotechnology: Lessons from Nature … 539

MIPs could also be used to produce enzyme mimics with high stability in
organic solvents, extreme pH and temperature conditions, which are needed for
most of the industrial processes. Although robust enzymes with tailored selectivities
can be obtained, the catalytic efficiency of these systems is, in general, still well
below that of the natural enzymes [26].
Apart from MIPs, hybrid approaches can also be used, especially in the fabri-
cation of nanoscale pores with high selectivity based on specific biochemical
properties [27]. Solid-state fabrication of pores in nanoscale has some challenges
such a difficulties in adjustment of the pore size and specificity, non-specific
interactions and clogging of the pores. Another alternative is the usage of wild-type
and genetically modified biological pores in the construction of selective and
sensitive,biosensors [28, 29] but instability of proteins and difficulties in their
purification may limit their large scale production. These limitations and challenges
can be addressed by biomimetic approaches. Olfactory systems of insects, for
example, have nanochannels coated with a lipid layer which helps to preconcentrate
odorant molecules before transporting. A similar system was adapted to a synthetic
pore and the resultant lipid-coated nanopores had better selectivity and caused less
non-specific adsorption [30]. In addition, translocation speed of proteins can be
tuned by regulating the lipid-bilayer viscosity.

21.4 Optics/Biophotonics

One of the exiting areas of bioinspired materials is the development of components

for devices, such as optics, photonics, photovoltaics, circuits and data storage [14].
Human eye uses gradient refractive index lenses with better focusing power than
traditional glass ones [31]. Although this type of lenses can be fabricated using
various methods [32], their usage as ocular implants is limited due to their poor
performance and lack of biocompatibility. Both limitations had been addressed by
several different bioinspired approaches. Template-mediated biomineralization
using an amphiphilic biopolymer-like surfactant, for example, was shown to be
useful to mediate the self-assembly of calcium carbonate from solution to construct
uniform hemispherical microlenses [33]. These lenses require no extreme condi-
tions for fabrication and are biocompatible. Insect eyes, on the other hand, contain a
microstructured part which significantly reduces the reflectance of the lens surface
[34]. Apart from insect eyes, distinctively organized biosilica fibers which form
spicules of the glass sponge Euplectella aspergillum of deep seas possesses inter-
esting optical properties and could be an inspiration to construct optical materials
with better properties [17]. Similar microstructured materials have been shown to be
effective not only to produce optical lenses but also anti-reflective coatings [35] and
even solar cells [36]. High cost and low efficiency of the existing solar cells are the
main challenges in this area and anti-reflection coatings inspired from the corneal
nipple arrays found in moths’ eyes had been found to significantly increase the
conversion efficiency [36]. Another strategy relies on natural photosynthesis
540 F.N. Kök

mechanism in which the conversion of sunlight into usable energy is efficiently

realized. Artificial photosynthetic systems can be constructed by implementing
biological elements like enzyme with nanoparticles to increase the efficiency and
robustness of these photocatalytic hybrid systems [37].
The shift from natural pigments to synthetic dyes made the production of more
economical textiles with diverse colors but their heavy usage led to environmental
problems due to their toxic and even carcinogenic nature. Biomimetic approach
may offer a completely different way of producing bright colors, using photonic
structures rather than pigments [38–40]. Butterflies and beetles display striking
colors originated from a wide range of structural mechanisms such as multilayer
films, diffraction gratings and quasi-ordered scattering [41]. Morphotex, for
example, is an unstained, structurally colored fiber, which is based on the biomi-
metic conception for the microscopic structure of Morpho butterfly’s wings. By
precisely controlling the layer thickness of the structure according to visible
wavelength, four types of basic colors such as red, green, blue and violet can be
developed [42]. In addition, some beetle species demonstrate reversible color
changes, based on the hydration of a porous ordered three-dimensional matrix [43].
These structures could be used in the development of optic displays and biomimetic
sensors.
Beetles and butterflies are not the only options for scientists to design
color-shifting systems. A phage-bundle nanostructure inspired from wattle structure
of turkey, for example, was used to design a tunable, colorimetric sensor for
trinitrotoluene (TNT) detection [44]. Wattle is composed of bundles of collagen
enriched with lots of blood vessels and in relaxed turkey the bundles are small and
transparent, exposing the blood vessels through the skin, creating a red color. When
a turkey gets excited, however, the bundles expand, become opaque and the color
shifts from red to white, or even blue. Similar structure was mimicked using
self-assembled genetically engineered phage displaying TNT-binding peptide
motifs. On exposure to various volatile organic chemicals, the structures rapidly
swell and undergo distinct color changes.

21.5 Natural Adhesives

Natural adhesives derived from marine organisms, especially marine mussels (M.
edulis), have been extensively studied for the past three decades and the role of
proteins in the success of these adhesives is well-established [45]. These materials
makes the adhesion to metal and mineral surfaces possible even in wet saline
environments at ambient temperatures, which could not be achieved by man-made
ones. Although their potency and non-toxicity is an advantage for both medical and
industrial applications, difficulties in extracting and purifying sufficient quantities
from the producing organism or low product yield have limited their commer-
cialization. Moreover, recombinant production of mussel proteins often yields
products with reduced functionalities compared to native proteins. Biomimetic
21 Bionanotechnology: Lessons from Nature … 541

approach, on the other hand, could make the design of practical analogs of these
adhesive materials possible; indeed improved functionalities can be achieved at
lower cost [46].
The knowledge on the mechanism behind this performance was piled up over
several decades using various approaches of molecular biology, chemistry and
physic fields [47–51]. Mussel adhesive proteins (MAPs), i.e. polyphenolic proteins,
collagen, and polyphenol oxidase, are important components of byssus responsible
for adhesion. When needed, these proteins were secreted into the seawater where
the curing process occurs. Quick diffusion of these proteins to the sea environment
was prevented by complex coacervation, a phenomenon occurred in the presence of
two oppositely charged polyelectrolyte solutions at isoelectric pH [52]. Tyrosine
residues in polyphenolic proteins are converted to 3,4-dihydroxy-phenylalanine
(DOPA) by polyphenol oxidase and high DOPA content was associated with good
adhesion [45]. Although the exact mechanism was not known, various processes
are suggested to work concurrently [45] and complex formation of metal ions, metal
oxides and semimetals with DOPA-containing proteins to form crosslinked struc-
tures is one of them [51]. Collagen proteins, on the other hand, mainly provides the
mechanical features (e.g. elasticity, extensibility, strength) needed for shock
absorbing ability and endurance [53].
Although some pieces of the puzzle are still missing, the existing knowledge
helped researchers to design biomimetic strategies for the production of non-toxic
and more effective adhesives. One way of producing biomimetic adhesives is to
synthesize simplified MAP mimics composed of shorter peptides using functional
groups such as DOPA and hydroxyproline rather than developing whole protein
[54]. The other way is the usage of knowledge on adhesion mechanism to develop
completely new molecules. In one of these strategies, L-poly(lactic acid) com-
posites reinforced with vegetable fibers together with dopamine had been shown to
exhibit improved mechanical properties compared with conventional ones [55].

21.6 Biointerfaces

As already mentioned, nature organizes functional units by weak interactions and

chirality. Various functional units including proteins, DNA double helix, phos-
pholipid membranes, photosystems, and biominerals are self-assembled or directed
by these interactions [56]. Another important issue in nature is the interaction of
nature’s elements with its surrounding environment via dynamic biointerfaces,
which is also choreographed by highly selective weak interactions responsive to
different external stimuli (light, pH, temperature, adsorption, etc.). With the help of
the conformational transitions induced by external stimuli, surface characteristics
could be changed which subsequently leading to macroscopic functions. Interaction
with weak forces also brings reversibility to the mechanism and various biological
events could not take place without reversibility. Stimuli-responsive smart polymers
like thermo-responsive poly(N-isopropylacrylamide) (PNIPAAm) can be used for
542 F.N. Kök

this purpose but still have limited applicability. For this reason various different
strategies from the design of three-component molecular systems including
“recognition-mediating-function” (RMF) to mimic complex interactions in vivo
[56] to the arrangement of surface topography based on different structural designs
[4, 57, 58] were used. In this chapter, two different motivations for interface design,
production of self-cleaning surfaces and biocompatible surfaces, are covered.

21.6.1 Self-cleaning Surfaces

One of the well-known biomimetic approaches in surface design was based on the
discovery of so called “lotus effect” based on the self-cleaning ability of lotus
flower petals [59]. Apart from plants, a wide range of living organisms (e.g. insect
wings, water strider legs, shark skin, gecko feet, spider silks, bird feathers) had been
investigated for their unique properties resulting different wettability profiles and
self-cleaning features and Gecko (Gekko gecko) feet is one of the attractive
examples since these animals can climb vertical surfaces without the need of
cleaning their feet [60]. This behavior was explained by the presence of adhesive
foot hairs (or setae), a uniform array on overlapping lamellar pads, which increases
the intermolecular interactions with the profile of the substrate and thus enhance the
adhesion [60, 61].
Self-cleaning surfaces exhibit low adhesion for a broad range of foulants. This is
achieved with the help of a complex interplay between chemistry and surface
morphology and can be classified in two classes: hydrophobic surfaces or hydro-
philic (underwater oleophobic) surfaces [57]. Superhydrophobic non-wetting
capability is a fundamental property of typical self-cleaning biological surfaces
and very different surface structures can lead to similar outcome. The underwater
oleophobic self-cleaning surfaces, on the other hand, are inspired from the hydrated
skin of marine organisms and the cooperation of physical heterogeneity and high
hydration energy moieties work synergistically to prevent oil-fouling [57, 62].
Studies were then focused on the design of artificial self-cleaning adhesive
surfaces using different materials and production methods [58, 61, 63] and bio-
mimetic surfaces with similar wetting behavior of that of the natural ones had been
obtained (Fig. 21.2) [64]. Wu et al. [64] used an improved laser interference
lithography to produce a surface topography resembling to an array of a string of
pearls. By tailoring their height along x and y directions, and designing their shape
and size, a controlled anisotropic biosurfaces were prepared. Prepared surface
showed a similar wetting behavior of that of rice leaf which was known as one of
the most typical examples for anisotropic wetting caused by those directional
microstructures (Fig. 21.2). These petal inspired surfaces could also be produced by
flexible replication methods like microinjection compression molding needed for
the mass-production of different hierarchical topographies (e.g. biomimics of rice or
lotus leaf) using designed templates and possess good self-cleaning performance in
submicron contaminants and good mechanical robustness [65]. Production of
21 Bionanotechnology: Lessons from Nature … 543

Fig. 21.2 Comparison of a biomimetic anisotropic surfaces and c natural rice leaf. Water droplets
and contact angles associated with b biomimetic and d natural surface. Reprinted with permission
from Wu et al. [64]. Copyright (2010) American Chemical Society

micropatterned vertically aligned carbon nanotubes on a silicon substrate is another

approach to mimic setae in gecko feet to obtain sticky tapes with higher shear
resistance than the natural gecko feet without compromising its remarkable
self-cleaning abilities [63].

21.6.2 Bioinspired Interfaces for Better Biocompatibility

One of the most crucial properties of biomaterials that we use to restore or replace
damaged or lost tissue functions is biocompatibility. The term “biocompatibility”
has evolved over years and it is now considered as a characteristic of a material-
biological host system and not a property of a material [66]. For the success of a
biomaterial, relationships between materials and host tissue should be understood
so that materials could be designed to elicit the most appropriate response.
Designing the host-material interface has therefore a great importance on the
integration of the biomaterial to the body by triggering an appropriate tissue
response and directing cells to attach, proliferate or even differentiate to desired
lineages. Organs and tissues have a definite micro-architecture and use specific
recognition motifs so when these elements are implemented in the design, they
could increase the success of implants in the body.
544 F.N. Kök

Designing implants in nano/micro level is therefore upmost importance to

achieve a long-term success and various biomimetic approaches have been used to
enhance the performance of implants and tissue engineering scaffolds [67].
Modification of implant surfaces by self-organized peptides to produce a
bio-inspired interface is one of the approaches to address this problem [68, 69]. In
one of these studies, a chimeric peptide composed of a peptide specifically binding
to titanium and another one inducing a regulatory pathway activating osteogenic
differentiation of mesenchymal stem cells for bone formation was used [69].
Attachment of peptide by molecular recognition and self-assembly have been found
to prevent the problems arising from chemical coupling and provided a mild and
effective way to stimulate osteogenesis in mesenchymal stem cells in vitro.
Self-assembled double layer formed by collagen I and IV have been proposed as
another approach to provide microenvironment mimicking the structure and orga-
nization of the native tissue [70]. The results showed that self-assembled collagen
structures supported the cell maturation and functionality compared to dip-coated
collagen surface. This also shows that not only the molecule itself but how it is
organized on the surface affects the success of the system and biomimetic design
strategies lead to more superior structures.

21.7 Biomimetic Membranes

Cell membrane has a complex structure composed of various elements and is

therefore hard to mimic. Artificial bilayer membranes can be designed but incor-
poration of membrane proteins into these structures is still an important challenge
due to the problems associated with their production, purification and instability in
aqueous solutions. Membrane biomimics are generally used in two main areas, as
described in the following sections.

21.7.1 Cell Membrane Mimics

Cell membrane mimics constituted in vitro are interesting and valuable systems in
basic membrane research. Cell membranes mimics are generally composed of
phospholipids that are already found in cell membrane. They could be formed either
in freely floating form like liposomes and nanodiscs [71–73] or on a solid surface as
solid-supported bilayer lipid membranes (sBLMs), tethered bilayer membranes
(tBLMs) [74–77], etc. The properties of these systems can be tailored both in terms of
phospholipid type and composition and membrane protein repertoire. This gives the
opportunity to study membrane protein–ligand and membrane protein-membrane
interactions in vitro conditions which cannot be done easily in vivo conditions due to
many external and internal factors involved in cellular organization. These cell
membrane mimics with or without proteins are not only promising for basic research
21 Bionanotechnology: Lessons from Nature … 545

Fig. 21.3 Construction of tBLM on gold-coated surface. a Activation of the surface.

b Attachment of tethering molecule (PEG2000 modified lipid). c Construction of protein-free
tBLM. d Construction of protein-incorporated tBLM (molecules were not presented in their actual
sizes). Reprinted from Inci et al. [77]. Copyright (2015), with permission from Elsevier

but also for different applications especially in medical fields to construct sensors and
drug screening systems.
Tethered bilayer membranes are especially attractive due to their ability to
incorporate membrane proteins of different sizes by preserving their biological
activities. Different molecules (e.g. peptides, polymers, etc.) and approaches can be
used to tether the lipid bilayer to the solid surface [78]. For a membrane protein
with small extra-membrane part, a short tethering molecule is enough for successful
incorporation [76]. When, however, a membrane protein with large extra-
membrane domain has to be incorporated into tBLM, more space is needed in
between the solid surface and the bilayer. This restriction can be overcome by the
usage of long tethering molecules. To construct a tBLM consisting p-glycoprotein,
for example, a modified polymer spacer (polyethylene glycol-2000-modified lipid
molecule) can provide this necessary space in order not to disrupt the active con-
formation of the protein [77]. Construction of this system needs extensive opti-
mizations, and generally done in several steps such as: (i) activation of the surface
(Fig. 21.3a), (ii) attachment of tethering layer (Fig. 21.3b), and (iii) construction of
whole bilayer without (Fig. 21.3c) or with (Fig. 21.3d) membrane protein via
liposome spreading. This protein incorporated tBLMs can be used to investigate
drug (e.g. statin-based cholesterol lowering drug)-membrane protein
(p-glycoprotein) interactions in an isolated environment [77] and this work can also
be extended to the study of binding kinetics of drugs to different membrane related
components.
546 F.N. Kök

21.7.2 Membranes for Water Treatment

Increasing amount of wastewater and decreasing access to clean water supplies

makes water purification and wastewater treatment inevitable to obtain usable water
from new sources. Membrane technology for the production of high quality water
has therefore gained importance in recent years and non-traditional water sources
originated from agricultural/industrial wastewater, seawater, etc. have been studied
for this purpose. Due to the diverse nature of these sources, different strategies have
to be applied to improve the separation performance and biomimetic approaches
have provided new insights to the challenges associated with that.
One of the important problems in the conventional oil removing membranes is
the easy fouling by oils. Self-cleaning ability of various living organisms discussed
in Sect. 21.6.1 could be adopted to design membranes with low oil fouling property
[57]. Hierarchical structures of fish scale had inspired researchers to design poly-
acrylamide hydrogel-coated mesh membranes that have been successfully used in
gravity-driven oil-water separation process [62].
In many water treatment applications, not only large contaminants like bacteria
but also small ones like metal ions are needed to be removed and this is mainly
done by sieving mechanism based on the membrane pore size. To obtain nano-
porous membranes with good resolution and high throughput, a well-known natural
counterpart, cell membrane and components associated with that for selective
transport, can be used to develop new strategies. One of the most widely studied
examples is the usage of aquaporins (AQPs) in desalination process [79, 80]. Cell
membrane separates the intracellular functions from the effect of surrounding
environment and only selectively allows the passage of different molecules when
needed. Even water could not pass from this membrane and need specialized
membrane proteins, i.e. aquaporins, to enter. AQPs let only water pass, rejecting all
other solutes [81], and therefore they have been intensively studied as candidates to
construct desalination membranes [80]. When the protein-to-polymer ratio was
optimized, the permeability values 3000 times higher than that of the pure polymer
and AQP based composite membranes with competitive water permeability and ion
rejection capacity compared with the existing systems could be obtained [82–84].

21.8 Hints from Nature for Endurance

Hostile and changing environmental conditions induce severe changes in the

properties and durability of the exposed materials. To deal with various stresses,
living organisms synthesized a wide range of composite materials with desired
properties [85]. Protein fiber-reinforced composites constituting byssus of mussels
21 Bionanotechnology: Lessons from Nature … 547

were an inspiration for the production of synthetic materials with gradual

mechanical properties which can withstand various environmental conditions [86].
Another example is the unusual mechanical tensile strength of egg capsules syn-
thesized by marine snails from the family Melongenidae [87]. This behavior was
found to be originated from conformational changes of the protein backbone under
mechanical stress. Similar studies revealed the involvement of protein fibrils to
obtain hybrid materials with better mechanical properties and this led to design of
high-performance materials with locally defined properties (like mechanics,
porosity). This type of gradual composites has gained importance in different fields
like automotive and aircraft technology as well as the biomedical field [86].
Materials are generally limited by their resistance to fracture rather than strength
and therefore most critical structures, such as bridges, ships, nuclear pressure
vessels, are manufactured from materials that are comparatively low in strength but
high in toughness [88]. Hierarchical structural architectures found in nature could
be an answer to produce advanced hybrid materials with better strength and
toughness properties. Although it sounds very promising, biomimetic approach has
yielded few technological advances, simply because of the lack of processing
techniques in practical dimensions needed for construction industry [88]. In general,
direct mimicry of the biological systems would not be enough to find engineering
solutions to improve mechanical properties. Living organisms use the materials that
are found in their environment and except from the extremophiles, they survive in
environments that have restricted mechanical loads, fairly narrow temperature
regimes, and so on, so their structures are not necessarily resistant to every con-
dition they may face during an industrial application [89]. It is therefore important
to be “inspired” from the nature to build new structure and if needed using different
materials but not directly try to mimic it.
From nature’s various strategies to construct armored protection systems, scaled
skins, in particular, are attractive due to their remarkable mechanical properties
[90]. This ultrathin structure is not only compliant, resistant to penetration and
lightweight but also its discrete, spatially overlapped scales attached to a flexible
dermis layer allow flexibility needed for the animals’ motion [91]. For example, the
arrangement of the scales of teleost fish is such that the soft dermis is covered by at
least three layers of scales [92]. Inspired from this elegant multilayer structure, a
synthetic fish skin material was designed by Funk et al. [90] (Fig. 21.4) to replicate
the structural, mechanical, and functional aspects of natural teleost fish skin. Key
characteristics of this complex structure was defined as low-modulus elastic mesh
or “dermis” layer that holds relatively rigid plastic scales and constructed system
was tested for its mechanical properties. This design resulted to an attractive
low-weight, low-profile protective coating for soft materials.
Another strategy that nature uses to increase the competitiveness of living sys-
tems in changing environmental conditions is the use of protective covers such as
prokaryotic S-layers. These S-layer proteins can self-assemble in solution and at
548 F.N. Kök

Fig. 21.4 Photographs of the skin of a the striped red mullet (Mullus surmuletus) and b the
synthetic material when undeformed (left) and under a bending deformation (right). Adapted with
permission from Funk et al. [90]. Copyright (2015) American Chemical Society

interfaces and can be used as protective coats for mechanically unstable structures
like liposomes and planar lipid membranes [93]. Instability of these artificial lipid
systems often hinders their commercial usage and this approach might help to
improve their longevity and, in particular, offer a solution to the problems emerging
upon their drying [94].

21.9 Conclusions

Nature provides an immense source of materials and systems, and mimicking the
composition, structure and formation principles of these systems help researchers to
design materials with improved properties. A tremendous amount of knowledge has
been accumulated for the last couple of decades about the molecular mechanisms
and structures behind natural materials and systems. Understanding the mechanisms
behind the success of the living systems and exploit them for practical purposes
requires interdisciplinary research with the cooperative effort of biologists, physi-
cists, chemists, material scientists and engineers. Biomimicry does not mean to
produce the exact copy of the biological systems by bare imitation. Once the
molecular mechanisms and structure-function relationships have been understood,
they could be adopted to construct completely new hierarchical structures with
properties tailored for specific applications through an interdisciplinary effort.
21 Bionanotechnology: Lessons from Nature … 549

References

1. M. Billard, J. Matari, Learning human arm movements by imitation: evaluation of a

biologically inspired connectionist architecture. Robot Auton. Syst. 941, 1–16 (2001)
2. X. Li, H. Xu, Z.S. Chen, G. Chen, Biosynthesis of nanoparticles by microorganisms and their
applications. J. Nanomater. 2011, 270974 (16 pp) (2011)
3. G.M. Whitesides, B. Grzybowski, Self-assembly at all scales. Science 295, 2418–2421 (2002)
4. J.H. Zhang, Y.R. Zhao, S.Y. Han, C.X. Chen, H. Xu, Self-assembly of surfactant-like peptides
and their applications. Sci. China Chem. 57, 1634–1645 (2014)
5. A. Rawat, R. Nagaraj, Covalently attached fatty acyl chains alter the aggregation behavior of
an amyloidogenic peptide derived from human ß(2)-microglobulin. J. Pept. Sci. 12, 770–783
(2013)
6. M. Barboiu, Constitutional hybrid materials—Toward selection of functions. Eur. J. Inorg.
Chem. 2015, 1112–1125 (2015)
7. S. Lapidot, S. Meirovitch, S. Sharon, A. Heyman, D.L. Kaplan, O. Shoseyov, Clues for
biomimetics from natural composite materials. Nanomedicine 7, 1409–1423 (2012)
8. M. Hnilova, B. Taktak Karaca, J. Park, C. Jia, B.R. Wilson, M. Sarikaya, C. Tamerler,
Fabrication of hierarchical hybrid structures using bio-enabled layer-by-layer self-assembly.
Biotechnol. Bioeng. 109, 1120–1130 (2012)
9. L. Addadi, S. Weiner, Biomineralization: mineral formation by organisms. Phys. Scr. 89,
098003 (13 pp) (2014)
10. T.-Y. Ning, X.-H. Xu, L.-F. Zhu, X.-P. Zhu, C.H. Chu, L.-K. Liu, Q.-L. Li, Biomimetic
mineralization of dentin induced by agarose gel loaded with calcium phosphate. J. Biomed.
Mater. Res., Part B 100B, 138–144 (2012)
11. J. Baumgartner, M.A. Carillo, K.M. Eckes, P. Werner, D. Faivre, Biomimetic magnetite
formation: from biocombinatorial approaches to mineralization effects. Langmuir 30, 2129–
2136 (2014)
12. A. Sinha, T. Mishra, N. Ravishankar, Polymer assisted hydroxyapatite microspheres suitable
for biomedical application. J. Mater. Sci. Mater. Med. 19, 2009–2013 (2008)
13. S. Onder, A.C. Calikoglu-Koyuncu, K. Kazmanli, M. Urgen, G.T. Kose, F.N. Kok, Behavior
of mammalian cells on magnesium substituted bare and hydroxyapatite deposited (Ti, Mg)N
coatings. New Biotechnol. (2015). doi:10.1016/j.nbt.2014.11.006
14. J.M. Galloway, J.P. Bramble, S.S. Staniland, Biomimetic synthesis of materials for
technology. Chem. Eur. J. 19, 8710–8725 (2013)
15. S. Schweizer, A. Taubert, Polymer-controlled, bio-inspired calcium phosphate mineralization
from aqueous solution. Macromol. Biosci. 7, 1085–1099 (2007)
16. M. Hildebrand, Diatoms, biomineralization processes, and genomics. Chem. Rev. 108, 4855–
4874 (2008)
17. J. Aizenberg, V.C. Sundar, A.D. Yablon, J.C. Weaver, G. Chen, Biological glass fibers:
correlation between optical and structural properties. Proc. Natl. Acad. Sci. USA 101, 3358–
3363 (2004)
18. K. Bleek, A. Taubert, New developments in polymer-controlled, bioinspired calcium
phosphate mineralization from aqueous solution. Acta Biomater. 9, 6283–6321 (2013)
19. M. Gungormus, E.E. Oren, J.A. Horst, H. Fong, M. Hnilova, M.J. Somerman, M.L. Snead, R.
Samudrala, C. Tamerler, M. Sarikaya, Cementomimetics—constructing a cementum-like
biomineralized microlayer via amelogenin-derived peptides. Int. J. Oral Sci. 4, 69–77 (2012)
20. E.E. Oren, C. Tamerler, D. Sahin, M. Hnilova, U.O.S. Seker, M. Sarikaya, R. Samudrala, A
novel knowledge-based approach to design inorganic-binding peptides. Bioinformatics 23,
2816–2822 (2007)
21. J. Kim, Y. Rheem, B. Yoo, Y. Chong, K.N. Bozhilov, D. Kim, M.J. Sadowsky, H.-G. Hur, N.
V. Myung, Peptide-mediated shape- and size-tunable synthesis of gold nanostructures. Acta
Biomater. 6, 2681–2689 (2010)
550 F.N. Kök

22. M. Sarikaya, C. Tamerler, A.K.Y. Jen, K. Schulten, F. Baneyx, Molecular biomimetics:

nanotechnology through biology. Nat. Mater. 2, 577–585 (2003)
23. F.L. Dickert, Biomimetic receptors and sensors. Sensors 14, 22525–22531 (2014)
24. C. Alexander, H.S. Andersson, L.I. Andersson, R.J. Ansell, N. Kirsch, I.A. Nicholls,
J. O’Mahony, M.J. Whitcombe, Molecular imprinting science and technology: a survey of the
literature for the years up to and including 2003. J. Mol. Recog. 19, 106–180 (2006)
25. G. Wackers, T. Vandenryt, P. Cornelis, E. Kellens, R. Thoelen, W. de Ceuninck,
P. Losada-Pérez, B. van Grinsven, M. Peeters, P. Wagner, Array formatting of the
heat-transfer method (HTM) for the detection of small organic molecules by molecularly
imprinted polymers. Sensors 14, 11016–11030 (2014)
26. M. Resmini, Molecularly imprinted polymers as biomimetic catalysts. Anal. Bioanal. Chem.
402, 3021–3026 (2012)
27. S.W. Kowalczyk, T.R. Blosser, C. Dekker, Biomimetic nanopores: learning from and about
nature. Trends Biotechnol. 29, 607–614 (2011)
28. S. Howorka, Z. Siwy, Nanopore analytics: sensing of single molecules. Chem. Soc. Rev. 38,
2360–2384 (2009)
29. T. Nowotny, M. de Bruyne, A.Z. Berna, C.G. Warr, S.C. Trowell, Drsophila olfactory
receptors as classifiers for volatiles from disparate real world applications. Bioinspir. Biomim.
9, 046007 (13 pp) (2014)
30. E.C. Yusko, J.M. Johnson, S. Majd, P. Prangkio, R.C. Rollings, J. Li, J. Yang, M. Mayer,
Controlling protein translocation through nanopores with bio-inspired fluid walls. Nat.
Nanotechnol. 6, 253–260 (2011)
31. Y. Jin, H. Tai, A. Hiltner, E. Baer, J.S. Shirk, New class of bioinspired lenses with a gradient
refractive index. J. Appl. Polym. Sci. 103, 1834–1841 (2007)
32. H. Ren, S.-T. Wu, Adaptive liquid crystal lens with large focal length tenability. Opt. Express
14, 11292–11298 (2006)
33. K. Lee, W. Wagermaier, A. Masic, K.P. Kommareddy, M. Bennet, I. Manjubala, S.-W. Lee, S.
B. Park, H. Colfen, P. Fratzl, Self-assembly of amorphous calcium carbonate microlens arrays.
Nat. Commun. 3, 725 (2012)
34. D.G. Stavenga, S. Foletti, G. Palasantzas, K. Arikawa, Light on the moth-eye corneal nipple
array of butterflies. Proc. R. Soc. B 273, 661–667 (2006)
35. C. Morhard, C. Pacholski, D. Lehr, R. Brunner, M. Helgert, M. Sundermann, J.P. Spatz,
Tailored antireflective biomimetic nanostructures for UV applications. Nanotechnology 21,
425301 (2010)
36. R. Dewan, S. Fischer, W.B. Meyer-Rochow, Y. Ozdemir, S. Hamraz, D. Knipp, Studying
nanostructured nipple arrays of moth eye facets helps to design better thin film solar cells.
Bioinspiration Biomimetics 7, 016003 (2012)
37. P.W. King, Designing interfaces of hydrogenase–nanomaterial hybrids for efficient solar
conversion. Bioch. Biophys. Acta 1827, 949–957 (2013)
38. K. Watanabe, T. Hoshino, K. Kanda, Y. Haruyama, S. Matsui, Brilliant blue observation from
a morpho-butterfly-scale quasi-structure. Jap. J. Appl. Phys. 44, L48–L50 (2005)
39. J. Xu, Z. Guo, Biomimetic photonic materials with tunable structural colors. J. Colloid Interf
Sci. 406, 1–17 (2013)
40. K. Yu, T. Fan, S. Lou, D. Zhang, Biomimetic optical materials: Integration of nature’s design
for manipulation of light. Prog. Mater Sci. 58, 825–873 (2013)
41 A.E. Seago, P. Brady, J.-P. Vigneron, T.D. Schultz, Gold bugs and beyond: a review of
iridescence and structural colour mechanisms in beetles (Coleoptera). J. R. Soc. Interface 6,
S165–S184 (2009)
42. http://www.nanotechproject.org/cpi/products/morphotex-r-fiber/
43. M. Rassart, J.-F. Colomer, T. Tabarrant, J.P. Vigneron, Diffractive hygrochromic effect in the
cuticle of the hercules beetle Dynastes hercules. New J. Phys. 10, 033014 (14 pp) (2008)
44. J.-W. Oh, W.-J. Chung, K. Heo, H.-E. Jin, B.Y. Lee, E. Wang, C. Zueger, W. Wong, J. Meyer,
C. Kim, S.-Y. Lee, W.-G. Kim, M. Zemla, M. Auer, A. Hexemer, S.-W. Lee, Biomimetic
virus-based colourimetric sensors. Nat. Commun. 5, 3043 (2014)
21 Bionanotechnology: Lessons from Nature … 551

45. N. Bandara, H. Zeng, J. Wu, Marine mussel adhesion: biochemistry, mechanisms, and
biomimetics. J. Adhesion Sci. Technol. 27, 2139–2162 (2013)
46. H.J. Cha, D.S. Hwang, S. Lim, Development of bioadhesives from marine mussels. Biotech.
J. 3, 631–638 (2008)
47. J.H. Waite, Adhesion a la moule. Integr. Comp. Biol. 42, 1172–1180 (2002)
48. Q. Lin, D. Gourdon, C. Sun, N. Holten-Andersen, T.H. Anderson, J.H. Waite, J.N.
Israelachvili, Adhesion mechanisms of the mussel foot proteins mfp-1 and mfp-3. PNAS USA
104, 3782–3786 (2007)
49. S. Haemers, G.J.M. Koper, G. Frens, Effect of oxidation rate on cross-linking of mussel
adhesive proteins. Biomacromolecules 4, 632–640 (2003)
50. D.S. Hwang, Y. Gim, H.J. Yoo, H.J. Cha, Practical recombinant hybrid mussel bioadhesive
fp-151. Biomaterials 28, 3560–3568 (2007)
51 D.S. Hwang, H. Zeng, A. Masic, M.J. Harrington, J.N. Israelachvili, J.H. Waite, Protein and
metal dependent interactions of a prominent protein in mussel adhesive plaques. J. Biol.
Chem. 285, 25850–25858 (2010)
52. D. Hwang, H. Zeng, A. Srivastava, D.V. Krogstad, M. Tirrell, J.N. Israelachvili, J.H. Waite,
Viscosity and interfacial properties in a mussel-inspired adhesive coacervate. Soft Matter 6,
3232–3236 (2010)
53. H.G. Silverman, F.F. Roberto, Understanding marine mussel adhesion. Marine Biotechnol. 9,
661–681 (2007)
54. M. Yu, T. DeSimone, T.J. Deming, Synthetic analogs of marine mussel cement protein.
Polym. Prepar. 38, 101–102 (1997)
55. A. Bourmaud, J. Riviere, D. Le Antoine, G. Raj, C. Baley, Investigations of the use of a
musselinspired compatibilizer to improve the matrix-fiber adhesion of a biocomposites.
Polym. Testing 28, 668–672 (2009)
56. B. Chang, M. Zhang, G. Qing, T. Sun, Dynamic biointerfaces: from recognition to function.
Small 11, 1097–1112 (2015)
57. J. Zhao, X. Zhao, Z. Jianga, Z. Li, X. Fana, J. Zhu, H. Wu, Y. Su, D. Yang, F. Pan, J. Shi,
Biomimetic and bioinspired membranes: preparation and application. Prog. Polym. Sci. 39,
1668–1720 (2014)
58. B. Bhushan, R.A. Sayer, Surface characterization and friction of a bio-inspired reversible
adhesive tape. Microsyst. Technol. 13, 71–78 (2007)
59. C. Neinhuis, W. Barthlott, Characterization and distribution of water-repellent, self-cleaning
plant surfaces. Ann. Bot. 79, 667–677 (1997)
60. W.R. Hansen, K. Autumn, Evidence for self-cleaning in gecko setae. PNAS 102, 385–389
(2005)
61. T.S. Kustandi, V.D. Samper, W.S. Ng, A.S. Chong, H. Ga, Fabrication of a gecko-like
hierarchical fibril array using a bonded porous alumina template. J. Micromech. Microeng. 17,
75–81 (2007)
62. Z. Xue, S. Wang, L. Lin, L. Chen, M. Liu, L. Feng, L. Jiang, A novel superhydrophilic and
underwater superoleophobic hydrogel-coated mesh for oil/water separation. Adv. Mater. 23,
4270–7273 (2011)
63. S. Sethi, L. Ge, L. Ci, P.M. Ajayan, A. Dhinojwala, Gecko-inspired carbon nanotube-based
self-cleaning adhesives. Nano Lett. 8, 822–825 (2008)
64. S.-Z. Wu, D. Wu, J. Yao, Q.-D. Chen, J.-N. Wang, L.-G. Niu, H.-H. Fang, H.-B. Sun,
One-step preparation of regular micropearl arrays for two-direction controllable anisotropic
wetting. Langmuir 26, 12012–12016 (2010)
65. W.-S. Guan, H.-X. Huang, A.-F. Chen, Tuning 3D topography on biomimetic surface for
efficient self-cleaning and microfluidic manipulation. J. Micromech. Microeng. 25, 035001
(9 pp) (2015)
66. D.F. Williams, There is no such thing as a biocompatible material. Biomaterials 35, 10009–
10014 (2014)
67. D. Singh, D. Singh, S. Zo, S.S. Han, Nano-biomimetics for nano/micro tissue regeneration.
J. Biomed. Nanotechnol. 10, 3141–3161 (2014)
552 F.N. Kök

68. H. Yazici, H. Fong, B. Wilson, E.E. Oren, F.A. Amos, H. Zhang, J.S. Evans, M.L. Snead, M.
Sarikaya, C. Tamerler, Biological response on a titanium implant-grade surface functionalized
with modular peptides. Acta Biomater. 9, 5341–5352 (2013)
69. Y. Zhou, M.L. Snead, C. Tamerler, Bio-inspired hard-to-soft interface for implant integration
to bone. Nanomed. Nanotech. Bio Med. 11, 431–434 (2015)
70. A.E. Sorkio, E.P. Vuorimaa-Laukkanen, H.M. Hakola, H. Liang, T.A. Ujula, J.
J. Valle-Delgado, M. Osterberg, M.L. Yliperttula, H. Skottman, Biomimetic collagen I and
IV double layer Langmuir-Schaefer films as microenvironment for human pluripotent stem
cell derived retinal pigment epithelial cells. Biomaterials 51, 257–269 (2015)
71. Y. Dayani, N. Malmstadt, Liposomes with double-stranded DNA anchoring the bilayer to a
hydrogel Core. Biomacromolecules 14, 3380–3385 (2013)
72. I.G. Denisov, Y.V. Grinkova, A.A. Lazarides, S.G. Sligar, Directed self-assembly of
monodisperse phospholipid bilayer nanodiscs with controlled size. J. Am. Chem. Soc. 126,
3477–3487 (2004)
73. F. Hagn, G. Wagner, Structure refinement and membrane positioning of selectively labeled
OmpX in phospholipid nanodiscs. J. Biomol. NMR 61, 249–260 (2015)
74. R.P. Richter, J.L.K. Him, A. Brisson, Supported lipid membranes. Mater Today 6, 32–37
(2003)
75. J.A. Jackman, W. Knoll, N.-J. Cho, Biotechnology applications of tethered lipid bilayer
membranes. Materials 5, 2637–2657 (2012)
76. E.-K. Sinner, U. Reuning, F.N. Kok, B. Saccà, L. Moroder, W. Knoll, D. Oesterhelt,
Incorporation of integrins into artificial planar lipid membranes: characterization by
plasmon-enhanced fluorescence spectroscopy. Anal. Biochem. 333, 216–224 (2004)
77. F. Inci, U. Celik, B. Turken, H.O. Özer, F.N. Kok, Construction of p-glycoprotein
incorporated tethered lipid bilayer membranes. Biochem. Biophys. Rep. (2015). doi:10.1016/j.
bbrep.2015.05.012
78. S. Rebaud, O. Maniti, A.P. Girard-Egrot, Tethered bilayer lipid membranes (tBLMs): interest
and applications for biological membrane investigations. Biochimie 107, 135–142 (2014)
79. M. Kumar, M. Grzelakowski, J. Zilles, M. Clark, W. Meier, Highly permeable polymeric
membranes based on the incorporation of the functional water channel protein. Aquaporin Z.
Proc. Natl. Acad. Sci. USA 104, 20719–20724 (2007)
80. M. Grzelakowski, M.F. Cherenet, Y.-X. Shen, M. Kumar, A framework for accurate
evaluation of the promise of aquaporin based biomimetic membranes. J. Membr. Sci. 479,
223–231 (2015)
81. P. Agre, Aquaporin water channels (Nobel lecture). Angew. Chem. Int. Ed. 43, 4278–4290
(2004)
82. P.H.H. Duong, T.S. Chung, K. Jeyaseelan, A. Armugam, Z. Chen, Hong M. YangJ, Planar
biomimetic aquaporin-incorporated triblock copolymer membranes on porous alumina
supports for nanofiltration. J. Membr. Sci. 409–410, 34–43 (2012)
83. P.S. Zhong, T.S. Chung, K. Jeyaseelan, A. Armugam, Aquaporin-embedded biomimetic
membranes for nanofiltration. J. Membr. Sci. 407–408, 27–33 (2012)
84. X. Li, R. Wang, F. Wicaksana, C. Tang, J. Torres, A.G. Fane, Preparation of high performance
nanofiltration (NF) membranes incorporated with aquaporin Z. J. Membr. Sci. 450, 181–188
(2014)
85. E. Munch, M.E. Launey, D.H. Alsem, E. Saiz, A.P. Tomsia, R.O. Ritchie, Tough, bio-inspired
hybrid materials. Science 322, 1516–1520 (2008)
86 A. Hagenau, M.H. Suhrea, T.R. Scheibel, Nature as a blueprint for polymer material concepts:
Protein fiber-reinforced composites as holdfasts of mussels. Prog. Polym. Sci. 39, 1564–1583
(2014)
87. A. Miserez, J.C. Weaver, O. Chaudhurid, Biological materials and molecular biomimetics –
filling up the empty soft materials space for tissue engineering applications. J. Mater. Chem.
B 3, 13–24 (2015)
88. M.E. Launey, R.O. Ritchie, On the fracture toughness of advanced materials. Adv. Mater. 21,
2103–2110 (2009)
21 Bionanotechnology: Lessons from Nature … 553

89. G. Mayer, Rigid biological systems as models for synthetic composites. Science 310, 1144–
1147 (2005)
90. N. Funk, M. Vera, L.J. Szewciw, F. Barthelat, M.P. Stoykovich, F.J. Vernerey, Bioinspired
fabrication and characterization of a synthetic fish skin for the protection of soft materials.
ACS Appl. Mater. Interfaces 7, 5972–5983 (2015)
91. D.G. Elliott, In the laboratory fish. In: G.K. Ostrander (ed.) Academic Press, San Diego,
Chap. 5, p. 99 (2000)
92. D. Zhu, L. Szewciw, F. Vernerey, F. Barthelat, Puncture resistance of the scaled skin from
striped bass: collective mechanisms and inspiration for new flexible armor designs. J. Mech.
Behav. Biomed. Mater. 24, 30–40 (2013)
93. U.B. Sleytr, B. Schuster, E.-M. Egelseer, D. Pum, S-layers: principles and applications. FEMS
Microbiol. Rev. 38, 823–864 (2014)
94. B. Schuster, U.B. Sleytr, Biomimetic interfaces based on S-layer proteins, lipid membranes
and functional biomolecules. J. R. Soc. Interface 11, 20140232 (2014)
Chapter 22
Quantum Dots in Bionanotechnology
and Medical Sciences: Power of the Small

İ. Ergal and A.T. Akarsubasi

Abstract Among the other bionanotechnology and nanomedicine tools Quantum

Dots (QDs) have pioneered numerous advances in both fundamental and applied
sciences; due to their unique optical, electronic and physical properties namely
photostability, single source excitation, narrow emission, multiplexing capabilities
and high quantum yield as compared to traditional fluorophore materials. New
generation QDs with better biocompatibility allows studying intracellular processes
in human beings at the single molecule level, to use in vitro imaging for bionan-
otechnological purposes, long term in vivo observation of cellular events, tumor
targeting and even in diagnostics. However, further developments are necessary for
robust and reproducible methods to conjugate QDs with many different biomole-
cules under stringent biological requirements to improve its commercial availability
and further use in clinical diagnostics.

22.1 Introduction

Quantum Dots (QDs) have distinct advantages over traditional fluorescent organic
dyes and green fluorescent proteins on account of their longer photobleaching
capacity, spectral overlapping and low signal intensity [1–3].
Principally for early diagnostics, the requirement for highly sensitive and
specific biological information has pushed the need for such nano crystal structures
like QDs to be used in clinical applications. Because of their unique properties, QDs

İ. Ergal
A.T. Akarsubasi (&)

Molecular Biology and Genetics Department,
İstanbul Technical University, Maslak, İstanbul, Turkey
e-mail: [email protected]
İ. Ergal
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 555

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_22
556 İ. Ergal and A.T. Akarsubasi

have become one of the most investigated subject in biology and medicine. QDs
have already been used in many areas of biotechnology particularly, in vitro and
in vivo imaging of sentinel lymph nodes [4, 5], tumor-specific receptors [6, 7],
malignant tumor detectors [8], tumor immune responses [9].

22.2 Types and Characteristics of QDs

QDs are semi conductor nanocrystals which shows particular optical properties.
Unique capabilities of QDs are mainly coming from five distinct properties. First,
nano-scale size of these crystals ranging from 1 to 20 nm make them efficient
probes used in biological and medical applications. Second, their size-tunable
fluorescence defined as ‘quantum confinement effect gives them a wide range of
absorption and emission wavelengths from visible to the near infrared (NIR).
Thirdly, combination of tunable-size for variety of fluorescence, with large Stokes
shifts and narrow emission spectrum make simultaneous excitation of multiple
colors possible. Fourth, QDs have high signal intensity due to huge absorption
extinction coefficients and high quantum yield. Last but not least, since they are
inorganic, they exhibit more resistance for photobleaching called long-term pho-
tostability [10–13]. Discovery of QDs begins with first observation of them in glass
crystals and pioneered by Efros and Ekimov in early 1980s [14].
The typical QDs consist of a semiconductor core composed of II–IV group of
elements where CdX is the most investigated QDs due to their excellent optical and
electrochemical properties.
Advancement of the science and technology has led to many research projects on
QDs synthesis and applications which was driven after mid-eighties. However, it
took nearly two decades of promotion in QD research until the successful synthesis
of colloidal CdX (X = S, Se, Te) QDs [15].
One of the major drawbacks which restricts the application of QDs for in vivo
bionanotechnological research is the toxicity associated with heavy metals; par-
ticularly the cadmium, lead, or arsenic containing QDs.
If the surfaces (shells) are not properly covered or protected by ligands, the toxic
heavy metal ions can easily be permeated into biological systems.
Evolving alternative designs of fluorescent nanocrystals with high biocompati-
bility has become crucial because of the limitation of heavy metal containing QDs.
Basically, a few methods are commonly used, one of them is to generate highly
fluorescent nanoparticles with nontoxic elements or to explore luminescent con-
jugated polymer dots.
Recently, scientists also focused on fluorescent polymer encapsulation and
functionalization to yield conjugated polymer dots (CP dots) for bioimaging [16,
17]. CP dots which have appropriate absorption range (350–550 nm) for fluores-
cence microscopy and laser excitation, can exhibit higher brightness than any other
nanoparticles under the same conditions [18]. Another type of QDs is the silicon
nanoparticles (<10 nm) which has unique properties such as resistance to
22 Quantum Dots in Bionanotechnology and Medical Sciences … 557

photobleaching and wide emission range with relatively high quantum yields.
Silicon based QDs (Si QDs) have advantages over traditional fluorescent organic
dyes when they are used as fluorescent probes for bioimaging due to their nontoxic
structure [19]. It has been reported that Si QDs are 10 times safer than
Cd-containing QDs [20].

22.3 Advantages and Disadvantages of QDs

Quantum dots are nanoscale light-emitting crystals that are emerging as a new class
of biological probes (specifically in the imaging field). They could replace organic
dyes and fluorescent proteins. The fundamental advantages of using quantum dots
are their high quantum yield and strong emission intensity, in addition to the broad
emission spectrum as a function of tunable particle size [21].
The QDs can be excited by UV light and emit light, changing from visible to
infrared wavelengths. Their broad excitation spectrum and narrow emission spec-
trum with almost no spectral overlap makes them attractive for high resolution
imaging of multiple species at the same time without the need for complex optic
detection systems.
QDs have higher signal-to-noise ratios compared to traditional fluorochromes.
The high sensitivity of the QDs allows precise detection even in the presence of
strong autofluorescent signals especially during in vivo imaging [22, 23]. During
the live cell imaging long term monitoring of the biological sample has been
enabled by QDs because of their resistance to photobleaching. Additionally, QDs
can be used in multicolor optical coding for high throughput screening and also
used in microarrays, bioimaging, FACS (fluorescence-activated cell sorting) anal-
ysis, and monitoring pharmacokinetics as contrast agents [24].
One of the major limitations of quantum dots can emerge from the stability of the
core shell structure. Most commonly used materials in commercially available QDs
are a core of CdSe and a shell of ZnS. In order to provide hydrophilic character for
better biocompatibility for this inorganic structure, amphiphilic polymers are used to
cap the shell layer and provide reactive sites for further linking to proteins. Therefore
the stability of the shell layer provides better control on solubility and possible
release of core materials (Cd ions) to the surrounding which may result in toxicity
[25]. That is why QDs have limited immediate clinical use, but rather has focused
applications on animal testing and in vitro assay developments. However, before we
move forward toward widespread use of quantum dots in biological systems, further
research is needed [22], since it’s toxicity to humans is still being debated.
Despite, size dependent tuning methodology mentioned as an advantage in terms
of covering multi-color in the system; due to the production strategy, different
hydrodynamic sizes makes such multiplexing procedures difficult in biological
systems.
Another disadvantage is the blinking property of the QDs. They tend to blink at
the single dot level that exhibit some limitation in absolute fluorescence
558 İ. Ergal and A.T. Akarsubasi

quantification. However, when used for imaging of biomarkers blinking doesn’t

affect much as there are very many of quantum dots in a sample to allow proper
averaging.

22.4 Synthesis of QDs

Over three decades ago, the first QD synthesis has been reported. Since then variety
of methods have been developed for synthesis of QDs. QDs can be manufactured
with different sizes starting from few nanometers up to few micrometers and size
distribution can be controlled using specific growth techniques under high tem-
peratures. QDs can be designed in a way that, product can be in bare core;
core/shell or core/shell/shell architecture. One of the main parameters over here
besides the size distribution is the choice of shell and coatings. This has particular
importance since the shell stabilizes the nanocyrstal core and gives specific pho-
tophysical properties.
These methods can be classified broadly into two types, based on the solvent
used in production processes: the organometallic method [26–29] and the aqueous
method [30–32].
In organometallic synthesis, high boiling point and high coordinating capability
media such as trioctylphosphine oxide (TOPO) is used with both metal and
chalcogen elements (Oxygen family from group 6 in the periodic table). QDs
synthesized with this methodology have high fluorescence quantum yields due to
perfect crystal structures. Another advantage of this method is to provide narrow
size distribution. Yet, in typical organometallic method, QDs are synthesized with
the hydrophobic surface which is the major drawback for biological applications.
Changing the surface properties of QDs with coating with a water-soluble shell and
ligand-exchange is resulted with significant loss in fluorescence signal.
One of the most important disadvantages of the organometallic synthesis is the
cost and troublesome process; however, still it is considered as one of the main
synthesis methodology due to high quantum yield and better means of controlled
size distribution.
On the other hand, the aqueous synthesis has some advantages over
organometallic synthesis such as being environment-friendly and has relatively low
cost production procedure resulting from improved simplicity, reproducibility and
less toxicity. Therefore the aqueous synthesis has gained better reputation as pre-
ferred method for synthesis of QDs, in spite of the lower quantum yields and
broader size distribution [33, 34].
Especially, with the properties such as higher stability and better compatibility in
biological environment, hydrophilic thiol-capped QDs are become much popular
compared to those prepared in organic solvent. Various methods have been reported
to synthesize the thiol-capped QDs, including hydrothermal synthesis and
microwave-assisted irradiation [35–38].
22 Quantum Dots in Bionanotechnology and Medical Sciences … 559

In the hydrothermal synthesis method, an airtight container which contains all

reaction reagents is heated to supercritical temperature under the high pressure.
Thus, the reaction time and surface defects of QDs can be efficiently reduced in
these conditions [39]. Since, the reaction temperature is high; the processes of
nucleus formation and crystal growth can be separated. As a result QDs with
narrow size distribution produced and the existing problem in the traditional
aqueous synthesis can be overcome.
Another attractive method for QDs synthesis is the microwave-assisted irradia-
tion. Almost a decade ago, CdTe and CdSe-CdS have been simply synthesized by
QDs by microwave-assisted irradiation method [40]. Microwave irradiation method
is based on a heating source which is used to optimize synthesis conditions. Water
is the most suitable solvent in this system. The quantum yields of QDs are able to
reach as high as 17 % with the quick heating the whole solution (above 100 °C).
Production of QDs with aqueous methods enables the use of QDs in biological
researches without additional procedures such as ligand exchange.

22.5 Toxicity

For the application of QDs in biological studies, safety, and biocompatibility issues
must be addressed. Potential adverse human health effects have been suggested
depending on toxicological studies, but the fundamental cause-effect relationships
are still not clear [41].
QDs are toxic mainly coming from in vitro studies where cadmium-containing
particles killed cells in culture. Thus, many researchers are believed that QDs will
never be used for diagnostics and treating patients.
It has been assumed that since QDs are toxic to cells, they must also be toxic to
humans despite there is not sufficient scientific evidence. Actually, cell-based
toxicity studies are rarely transferable to more complex biological systems to be
able to conclude with such extrapolation. Besides, yet they have not been tested in
animal models at physiologically appropriate doses.
Therefore, it is crucial to detect their potential toxicity in cells and on organisms
and to define the interaction of QDs with biological systems.
Several studies have shown that the toxicity of QDs depends on many factors
which can be categorized as inherent physicochemical properties and environmental
conditions [41–43]. QDs should be characterized individually by their types since
QDs are heterogeneous. They vary in their core and shell composition, sizes,
charges, concentrations, in their surface chemistry and functional groups, oxidation,
and mechanical stability and all these variations have been involved as contributing
factors to its toxicity [42]. QDs toxicity primarily depends on heavy metal based
toxicity and also induce cytotoxicity generated by reactive oxygen species (ROS).
Also, the quantum size effect which resulted from nanoparticle toxicity is their size
dependent intracellular routing where the nanoscale particles are able to reach
organelles that are inaccessible to metal ions.
560 İ. Ergal and A.T. Akarsubasi

Very few studies mentioned, in vivo QDs toxicity validation tests. The one which
has been applied on rodents via bioimaging studies, helped to identify body areas at
risk for toxicity. The organs of the reticuloendothelial system (RES) took up and
retained the majority of given QDs [44, 45]. RES is the part of body’s defense
mechanism which consist of phagocytic cells located in the liver, spleen and lymph
nodes. Due to such data, it is proposed that QDs mainly degraded in these organs.
Despite accumulation of QDs in the RES organs has been reported; little focus
has been placed on the QDs clearance from the body. There are two excretion
pathways available for metal containing particles: following renal clearance
(through kidneys to urine) or through the liver’s biliary system into feces. These
results lead to a question: Why are QDs apparently toxic in vitro but seems safer
in vivo? Actually rather than mentioning about toxicity of the QDs; the question
should be “What is the lethal dose?” Recent, studies have demonstrated that QD
physicochemical properties affects the cell interaction and the potential for toxicity.
Despite all the evolution of QDs over the past decade, we are still far from
carrying QD-based technology from lab scale to a patient’s bedside. Moreover, new
strategies to engineer safer QDs, combined with advanced methodology are
important steps forward in closing the gap between in vitro and in vivo testing and,
at the end, try to find out the answer to QD toxicity.

22.6 Surface Modification and Functionalization

In order to apply QDs to biological studies; biocompatibility and toxicity are the
main issues that must be solved. Very hydrophobic nanocrystals are generated as a
result of high quantum yield QDs and these nanocrystals are only soluble in
non-polar solvents such as toluene or chloroform.
Because of the need for the aqueous solutions and buffers in biological appli-
cations, their water solubility must be increased while optical properties conserved.
This purpose can be achieved by modification of the surface nature while keeping
the core intact. Usually, carboxylate, amino or hydroxyl like functional groups has
been used to modify surfaces. Hydrophilic ligand exchange or encapsulation using
hydrophobic interactions are two main approaches for the functionalization of the
QD surfaces (Fig. 22.1).
Surface modification of QDs synthesized via the organometallic route can
improve the aqueous solubility for further biological applications while protect
them from degradation and photobleaching. Polyethyleneglycol (PEG) can provide
functional groups for further bioconjugation that also helps to reduce non-specific
uptake in the non-targeted organs. Many literature reports have shown that PEG
coating can change the power of biodistribution of QDs in animals [46, 47].
Combination of PEG and other polymers, poly(lactic-co-glycolic acid) (PLGA)
and liposomes has also been investigated for surface modification of QDs [26, 27].
Encapsulating QDs with highly biocompatible silica can render a hydrophilic
surface and also facilitate the incorporation of various functional groups such as
22 Quantum Dots in Bionanotechnology and Medical Sciences … 561

Fig. 22.1 Aqueous solubility modifications of QDs: fabrication of functionalized QDs for
biological applications. Copyright (2011) Journal of Nanomaterials

carboxyl, amine, and thiol groups for further bioconjugation [48, 49]. Moreover, the
release of toxic QD components into the biological environment can be prevented
with silica shell [28, 29].
QD surfaces can have hydrophobic ligands that can be surface-modified to
render them water-soluble and biocompatible. The release of free cadmium and
reactive oxygen species (ROS) production are often discussed in isolation, it is
more likely that these mechanisms act in concert to produce QD toxicity.

22.6.1 Surface Coatings to Minimize Hydrodynamic Size

Recently, a multifunctional, multidentate polymer ligand for generating ultra-small,

highly compact QDs which still preserve the excellent optical properties of the QDs
have developed by Smith and Nie [50, 51].
These multidentate polymers can displace the existing ligands on the QDs and
tightly bind to the nanocrystal surface in a closed “loops-and-trains” conformation.
This unique design eliminates the hydrophobic barrier layer and causes the polymer
ligand to tightly conform the nanocrystal surface, resulting in an exceptionally thin
polymer shell and small overall particle size. In contrast to water-soluble QDs with
small-molecule ligands, the multidentate binding of the polymer provides excellent
colloidal stability, resistance to photobleaching, and high quantum yield. Using this
strategy, investigators have prepared high-quality QDs with a hydrodynamic size of
4–6 nm [51]. Importantly, Frangioni and colleagues [52] have demonstrated that
QDs with a hydrodynamic diameter of <5.5 nm undergo rapid renal clearance,
making these size-minimized nanoparticles ideal for in vivo imaging, wherein the
potential toxicity of the heavy metal containing QD has been a significant
impediment to clinical adoption. In addition to size, another problem is that current
QDs are often “sticky” because they have a tendency to bind nonspecifically to
proteins, cellular membranes, or extracellular matrixes [30, 53].
562 İ. Ergal and A.T. Akarsubasi

Nonspecific binding reduces the signal-to-noise ratio and limits detection sen-
sitivity and immunostaining specificity. Nonspecific binding can also lead to
false-positive staining for biomarkers in fluids, cells, and tissues. In particular, QDs
with highly negative or positive surface charges, such as surface coatings con-
taining carboxylic acids or amines, can exhibit strong nonspecific binding to cells
and tissues [30, 53, 54] as well as to proteins in serum and blood. Since most
biomolecules are charged or have charged domains [55], QDs could interact
electrostatically with many soluble proteins in solution or with biomolecules on the
cell surface and in the cytoplasm, resulting in the nonspecific binding.
To reduce nonspecific binding, PEGs are often attached to the organic coating
layer of QDs [30, 53–56]. PEGylated QDs have a nearly neutral surface charge and
can maintain colloidal stability through steric repulsion between the PEG chains.

22.7 Biocompatibility in QDs (Bioconjugation)

Preparation of the carefully selected functional surface coatings is one of the most
important stages to be able to use the advantage of the unique properties of QDs in
biological applications. Depending on the application environment, and desired
application mechanisms such as in vivo/in vitro imaging, diagnostics, biosensors
etc. Numerous QD bioconjugation strategies exist (Fig. 22.2) such as covalent
linkages, polyhistidine-metal-affinity coordination, electrostatic adhesion, and
chemoselective ligation [57].
The most important aspects that should be taken into consideration for bio-
conjugation are: (i) overall dimensions of bioconjugate (a polymer coated QD with
a long linker attached to a large protein might be spatially disadvantageous for
energy transfer applications or intracellular measurements); (ii) functionality
(conserving photophysical properties and binding properties of the biomolecule);
(iii) reproducibility; and (iv) multiplexing capacity (number of biomolecules per
QD or vice versa).
A pathway is provided by the proper application of these varying conjugation
schemes for controlling critical criteria of the biomolecules attachment including
valency, affinity, biomolecular orientation, QD-biomolecular spacing, and orienta-
tion of the target binding spot.
Nanoscale sizes of QDs allow attachment of multiple biomolecules to one QD, as
well as the attachment of multiple QDs in larger biomolecules such as viruses. The first
step for biocompatibility of QDs as mentioned above functionalization of the surface
for water-solubility. Transferring the inorganic QDs into water-soluble and bio-
compatible fluorescent nanoparticles offers variety of possibilities. Once the func-
tionalized QDs are stable in aqueous solution, many different classical as well as novel
bioconjugation methods are available, ranging from random conjugation (mostly
targeting primary amines or thiols with NHS-ester or maleimide-functionalized QDs)
to controlled bio-orthogonal conjugation. Another technique for attaching proteins,
peptides, DNA and similar biomolecules to QDs. It is also a relatively easy and stable
22 Quantum Dots in Bionanotechnology and Medical Sciences … 563

Fig. 22.2 QDs toxicity depend on physicochemical properties. Shape, core composition, size, and
shell composition can be manipulated during QD synthesis. Also surface ligands can be added to
increase solubilization. An additional coating can further protect the QD core from oxidation.
Copyright (2012) American Chemical Society

(kd_10–9 M) conjugation method. As a result of the convenient commercial avail-

ability of readily prepared QD bioconjugates and their stable binding (kd_10–14 M),
probably the most often applied conjugation is based on biotin–streptavidin recog-
nition. Many other technologies such as HaloTag (Promega), SNAP-tag and
copper-free click chemistry can also be used for QD-biolabeling.
564 İ. Ergal and A.T. Akarsubasi

22.8 Next Generation QDs

(Silica, Carbon, Metal Nanocluster)

New challenge in the QD technology is improving the optimized brightness,

minimized hydrodynamic sizes, resistance to nonspecific interactions, and
site-specific ligand conjugation. In the case of hydrodynamic size of QD,
QD-tagged proteins will be used substituted for GFP-tagged proteins inside living
cells [58–60] when it is reduced to level of green fluorescent proteins (GFPs).
Because of the inefficacy of nanoscale dots for optical absorbance and coating with
a thin polymer layer; the challenging task is improving some protein-sized dots.
New opportunities in next generation QDs for molecular medicine and
bio-diagnostic applications have being performed with novel studies and related
approaches.

22.8.1 Metal Nanoclusters

New class of fluorescent labels for biological applications which are called
fluorescent metal nanoclusters, known with their ultra-small sizes, good biocom-
patibility, and excellent photostability, have become a good alternative to QDs.
Especially, Au and Ag nanoclusters are favorable among them. Due to the extre-
mely low quantum yield (QY), photoluminescence from the noble metals, started
with little attention and much more interests have been attracted by researchers until
the much enhanced QY reached [61]. Till today, a lot of Au and Ag NCs Metal
nanocyrstals with different scaffolds (protein, peptide, and oligonucleotide) have
been developed and applied for the detection of thiol compounds [62], metal ions
[56, 63], protein [64, 65], DNA [66], RNA [67] and in vivo bioimaging.
DNA stabilized silver nanoclusters have advantages in design and applications,
due to easy assembly of DNA sequences; such as, Werner and colleagues designed
a nanocluster beacon to detect the human Braf oncogene DNA sequence. Thus,
newly developed fluorescent metal nanoclusters leads to promising opportunities
for in vivo bioimaging [68].

22.8.2 Carbon Dots (C-Dots/GQDs)

C-dots which were first obtained in 2004 during purification of single-walled car-
bon nanotubes through preparative electrophoresis have excellent optical properties
[69]. Thus, C-dots displayed great potentials in biological applications. Because of
the high optical property of C-dots; high performance complex catalyst systems
with C-dots are constructed, in order to provide the usage of full spectrum of
sunlight effectively [70–73]. Due to low autofluorescence and high tissue
22 Quantum Dots in Bionanotechnology and Medical Sciences … 565

transparency in the near infrared (NIR) region, C-dots emission in the NIR under
the excitation, is particularly significant for in vivo bionanotechnology [74, 75]. As
a type of C-dots, recently the graphene QDs (GQDs) have also gained a lot of
interest from researchers because of their excellent optical and electronic properties
[76, 77]. Despite belonging to same family, there are some discrepancies between
GQDs and C-dots [78]. The C-dots are either amorphous or crystalline, while
GQDs are graphene lattices inside the dots [79]. GQDs have been employed in the
production of numerous sensors and bioimaging [78–82]. Because of the high
biocompatibility and optical properties, C-dots and GQDs have enhanced potentials
in imaging applications of nanobiotechnology such as InP [83], InP/ZnS [84],
CuInS2/ZnS [85, 86], Ag2Se [87], Ag2S [88].

22.8.3 Silicon Dots (Si QDs)

Silicon has been known with poor optical properties for a long time. There are three
distinct photoluminescence bands for Si QDs, sequentially in the infrared, in the
red, and one in the blue light range. The most important advantage of Si QDs is
their good biocompatibility which were claimed to be much safer than Cd-based
QDs under UV irradiation (at least 10 times) [25], and even nanoscale Si has been
proposed as a food additive [89]. Recently, several methods have been reported to
produce optically stable, hydrophilic Si QDs [48]. However, one of the main dis-
advantages for their applications in bioimaging results from their oxidative
degradation in vivo. Thus, surface modifications such as PEGylated micelle
encapsulation and bioconjugation are necessary [22]. These Si QDs could be used
in multiple cancer-related in vivo applications (sentinel lymph node mapping,
tumor vasculature targeting and multicolor NIR imaging in live mice) and also
showed great potentials as biocompatible fluorescent probes [49].

22.9 Applications of QDs in Bionanotechnology

and Nanomedicine

22.9.1 Cell Labeling

One of the most common uses of QDs in bionanotechnology is labeling cells and
cellular structures as fluorescent probes. Thus, earliest usage of quantum dots as
more stable, brighter and multiplexing nanostructures is to label many of the intra
and inter-cellular structures such as cytoskeletal proteins, organelles and elucidation
of intracellular processes instead of fluorescent dyes. Organic dyes have been used
for cell labelling for decades; however, multicolor labelling has been a bottleneck
for such procedures untill the potential use of Quantum dots. Because of their broad
566 İ. Ergal and A.T. Akarsubasi

absorption and narrow excitation spectra combined with retarded photobleaching

makes those molecules appropriate for quantitative long term cellular imaging
which reduces signal crosstalk and increases the number of labels in a single
system. Long term photostability gives inarguably a significant advantage to QDs
over fluorescent dyes during time scale procedures. One of the most important
drawbacks in this application is QDs toxicity which has been overcome by new
generation of biocompatible water soluble QDs. There are mainly two types of
imaging: in vitro and in vivo.

22.9.2 In Vitro Imaging

QDs are used to create specific signals for biosensing depending on their unique
electrochemical properties. QDs are robust fluorescent nano-sized structures that are
compatible with conventional biosensing techniques used in the research and
diagnostics. Using different resonance energy transfer (RET) processes means a
nonradiative energy transferred without overlapping as in the excitation and
emission of fluorophores. Electron exchange with overlapping orbitals depends on
distance that usually exponentially decreases. Fluorescence RET (FRET), chemi-
luminescent RET (CRET) and bioluminescence RET (BRET) efficiencies show an
inversely related to sixth power of radius distance dependence, nanometal RET
(NSET) an inversely related to fourth power of radius distance dependence, and
dipole to metal particle energy transfer (DMPET) mainly follows the FRET
mechanism for small distances. They are therefore limited to shorter distances.
Following methods demonstrate the versatility and functionality of these nanop-
robes used as biosensors in nanomedicine and research. There are five main transfer
mechanisms applied with QDs (Fig. 22.3).

22.9.2.1 FRET, QDs Used as Energy Donors

In other words an organic dye fluorophore can be excited efficiently at an absorption

minimum of the acceptor then transferred to QDs which minimizes direct acceptor
excitation. The QD is quenched in that terms, The FRET efficiency increases with
the help of multiplexing QDs.

22.9.2.2 FRET QDs Used as Acceptors

The broad absorption spectrum of the QD becomes an advantage since QD can be

excited at almost any wavelength; however this may lead to inefficient FRET due to
small ratio of excited molecules. The problem has been solved using long
excited-state lifetimes such as up to several milliseconds and time interval
excitation.
22 Quantum Dots in Bionanotechnology and Medical Sciences … 567

Fig. 22.3 Quantum dot assays: Illustration of the energy transfer schemes utilized in various types
of QD assay paradigms. Each assay requires an energy input (electromagnetic, chemical,
biochemical or electrical) that is transmitted to the QDs either directly or through an input
intermediate. This energy modifies the electron behaviour of the QDs, which then release the
energy, either directly or through an output intermediate, in the form of emitted light or charge
transfer. Copyright (2014) Royal Society of Chemistry

22.9.2.3 CRET or BRET Using QDs as Acceptors

It is not possible to create luminescence in deep tissues with external light exci-
tation. However, a chemical reaction like the oxidation of luminol in CRET or
luciferin in BRET can be triggered at the desired position using bioluminescent
donors. Thus, the excitation of the chemiluminescent or bioluminescent donor is
quenched and the QD is stimulated for FRET effect that creates a smart probe.

22.9.2.4 NSET Using QDs as Donors

The advantage of energy transfer between QDs and Au–NPs mechanisms where the
QD donor is quenched and the Au–NP acts as dark quencher is mainly allowing
energy transfer over larger distances compared to FRET (nearly 20 nm).

22.9.2.5 CT Using QDs as Donors or Acceptors

In Charge transfer (CT), different QDs can be quenched by a single CT quencher,

which gives an advantage for color multiplexing. It’s mechanism basically, trans-
ferring electrons from or to proximal molecules with matching oxidation levels to or
from mainly the QD surface states which leads to QD photoluminescence quenching.
568 İ. Ergal and A.T. Akarsubasi

For the improvement of bionanotechnological applications, smart probes

development offers a promising future for QD-based energy and CT systems
regarding bionanotechnology applications, especially for in vitro diagnostics
(IVD) and cellular imaging [90–92].

22.9.3 In Vivo Imaging

There are two types of approaches here as passive targeting which depends on
permeability and accumulation on the target cells or tissue. On the other hand active
targeting has been applied to recognition of specific cellular components like
organelles or surface proteins through ligands targetted with QDs. There are several
characteristics which makes a fluorescence probe ideal for in vivo targeting such as;
photostability, high biocompatibility, low toxicity and minimal rethiculoendothelial
(RES) uptake. There are many reports in the literature mentioning about importance
of surface coating and particle sizes effect in vivo behavior of QDs.

22.9.3.1 Peptide-Conjugated QDs

Peptide-quantum dot nanoconjugate for labeling are preferred targeting ligands

because of their small sizes. Also many peptides can be simultaneously attached to
a single QDs to enhance its specificity. Furthermore, ease of synthesis and low
immunogenicity are some of the other advantages of peptides. The strength of
peptide-target interaction is highly variable as in antibody recognition. Moreover,
covalent conjugation of multiple peptides to one quantum dot which is achieved
through water soluble crosslinking reagents (such as EDAC, SIA and Traut’s
reagent) may considerably increase the binding capacity of the resulting complexes
to the target via multivalent interactions. Streptavidin-biotin noncovalent
self-assembly is another route for the utilization of peptide targeted QDs. After the
first in vivo peptideconjugated QDs targeted imaging was reported in 2006, the
arginine-glycine-aspartate (RGD) peptide has also been widely used in nanome-
dicine [53]. Since Cd-free RGD peptides have better biocompatibility [83, 93, 94],
they may find broader bionanotechnological future applications.
Peptide conjugated quantum dots have also been successfully used for targeting
of cellular proteins such as growth factor receptors, G protein-coupled receptors,
integrins, and ion channels [35, 37, 95]. In particular, quantum dots targeted to
protein transduction domains namely HIV TAT, Pep-1, poly-arginine, and SV40 T
antigen have been used in the research of intracellular delivery [35, 36, 96]. As a
result, peptide conjugated quantum dots offer exclusive advantages over
antibody-mediated targeting, and their potential as bionanotechnological probes is
being actively explored.
22 Quantum Dots in Bionanotechnology and Medical Sciences … 569

22.9.3.2 Antibody Conjugated QDs

Antibodies are widely used as a class of diverse and specific ligands for targeted
imaging and therapy. One of the simplest and quickest labeling route is to use an
antibody-quantum dot conjugate, such as using antibodies for F and G proteins to
detect the presence and follow the progression of respiratory syncytial viral
infection [97]. Another example is using antibody-conjugated quantum dots for
tumors in mice to target a prostate-specific membrane antigen in vivo [31]. In
particular, anti-HER2 quantum dot conjugates has been used imaging breast cancer
cells both in vivo [33] and in vitro [32]. Biotinylated antibodies can also be used
with streptavidin-coated quantum dots or they can be directly conjugated to the
quantum dot. Even a simple commercial kit is available for linking antibodies to
AMP quantum dots requiring no prior quantum dot experience (Invitrogen).
Besides, a wide range of antibody-quantum dot conjugates is also commercially
available.
Despite many monoclonal antibodies have been used for tumor targeting and
imaging, disadvantages to this approach include the relatively large size of anti-
bodies which limits the number of ligands attached to the surface of each QD. This
size affects the penetration into solid tumors. Alternatively, antibody heavy- and
light- chain variable domains connected by a flexible peptide linker, is a much
smaller targeting ligand than intact antibodies (25 vs. 150 kDa), which can maintain
high binding affinity and specificity to the antigen. Nonetheless, antibody-quantum
dot conjugates are one of the most common methodology of the choice for the most
of the quantum dots in bionanotechnology literature.

22.9.3.3 Ligand Conjugated QDs

Targeting the QDs directly to the binding site of the protein of interest via ligands (also
called as small molecules) can initiate a dynamic process and enable highly selective
faster monitoring [98]. Using ligand-conjugated QD helps to study factors regulating
protein expression and cell surface mobility, besides allosteric modulation of the
binding site which allows a multiplexing assay for drug discovery. Variety of
ligand-conjugated to a different size (color) QDs can be used in such assays. Also,
displacement assays are possible in which the binding site targeted test compound will
displace the probe, enabling a fluorescence-based assay for drug discovery.
Major disadvantage of the methods with ligand-conjugated QD is the synthesis
of an optimized ligand that requires fairly sophisticated organic chemistry to design
(Fig. 22.4).

22.9.3.4 High-Affinity Fusion Tag Targeting Approaches

Recently new approaches based on high-affinity fusion tags labelled QDs for cel-
lular targets have been reported. A genetically engineered high-affinity fusion tag
570 İ. Ergal and A.T. Akarsubasi

Fig. 22.4 Bioconjugation schemes for QD functionalization including (a) thiol coordination;
(b) polyhistidine groups; (c) amine/carboxyl moieties; (d) PEG; (e) hydrophobic interactions;
(f) electrostatic adhesion; (g) nickel mediated polyhistidine; (h) maleimide covalent linkage;
(i) active esters; and (j) biotin-streptavidin complexes. Figure used with permissions of publisher,
Copyright (2014) Sensors

that is usually a short peptide sequence with a recognition site, is used to express a
cellular target for further labelling with QDs [99, 100]. For instance, Dahan and
colleagues used trisnitriloacetic acid quantum dot conjugate which has a
sub-nanomolar affinity for linear polyhistidine motifs to target a decahistidine tag
fused to the N-terminus of type-1 interferon receptor subunit [101]. There are other
studies that have used fusion tags in conjunction with quantum dots include CrAsH
[102] and Halotag [103].

22.9.3.5 Nonspecific Binding

One of the main drawback of quantum dots is nonspecific cellular binding in a range
of in vivo nanobiotechnology applications, such as live cell imaging and tissue
staining.
Quantum dots may show nonspecific cellular binding that has been observed in
several systems and primarily attributed to electrostatic interactions in between
22 Quantum Dots in Bionanotechnology and Medical Sciences … 571

probes surface (e.g., carboxyl) and target cell surface [30, 104]. Therefore,
researchers must rigorously examine selected QDs for nonspecific interactions with
the surface functional groups and should choose a modification approach that re-
sults in minimal nonspecific binding. Moreover, the issue of nonspecific binding
should not be taken into consideration in the presentation of research results.

22.10 Dual-Modality Imaging with QDs

Imaging techniques and particularly preclinical imaging systems has been improved
massively over the last decade. However, yet no single imaging technique is perfect
to obtain all the necessary information for a given study. For instance, one of the
main challenges faced in fluorescence imaging is the quantification and deep tissue
penetration. Magnetic resonance imaging (MRI) has outstanding soft tissue contrast
and good resolution, but rather poor sensitivity. On the other hand; Positron
emission tomography (PET) is superior in sensitivity, quantitation, and tissue
penetration, nonetheless has relatively low resolution. Thus, in many clinical
applications to overcome these disadvantages single probe using multiple imaging
techniques has been applied to provide more detailed information.
QD-based nanoprobes will be better tools for these applications rather than
traditional fluorophores.

22.10.1 Fluorescence/MRI

Using the RGD peptides as ligands for tumor targeting with Paramagnetic
gadolinium coated QDs and PEGylated lipids to develop a dual-modality probe for
MRI has been reported by Mulder et al. in 2006 [105]. The high relativity, and the
specificity of this nano-probe makes it an excellent contrast agent for
nanobiomedical imaging purposes Although it has been confirmed by several
studies that the safety and tumor targeting capability of these nanocarriers, more
studies required to fully realize its potential for dual-modality imaging. Recently, a
core/shell nanoprobe was investigated for dual-modality imaging of breastcancer,
which was composed of an IO NP core [106] used with MRI to get detailed
anatomical structure of the tumor.

22.10.2 Fluorescence/CT

One of the most reliable and fast diagnostic tool is X-ray computed tomography
(CT) in many of the clinical applications which also has high spatial resolution.
Iodinated molecules and compounds used as contrast agents for CT; however, these
572 İ. Ergal and A.T. Akarsubasi

contrast agents have short lifetime in the blood or lymphatic vessels, which is a
major disadvantage. Recently, QDs that have been combined with iodinated
molecules can potentially overcome such disadvantages and leads to better future
promising results [107]. For instance, diagnostic tools in cardiovascular diseases, as
one of the leading cause of death worldwide, is very important specifically in
clinical settings for improving patient’s onset [108, 109]. QDs embedded inside
hydrophobic iodinated oil core was shown to be a good alternative for targeting
macrophages in atherosclerotic plaques both in vitro and in vivo.

22.10.3 Fluorescence/PET

PET is widely used imaging technique in cancer diagnosis, evaluation and during
the treatment follow up; due to its high sensitivity, good quantitation capability, and
superb tissue penetration. The first targeted dualmodality fluorescence/PET probe
based on QDs was reported in 2007 that was used RGD peptide as the targeting
ligand for the DOTA (1,4,7,10-tetraazacyclododecane-N,N′,N′′,N′′′-tetraacetic acid)
chelator (DOTA-QD-RGD). The results between the ones measured by in vivo PET
imaging and those measured by ex vivo NIRF imaging or tissue homogenate
fluorescence were in good correlation. Thereafter, pathology examination has
shown that DOTA-QD-RGD targets primarily the tumor vasculature with little
extravasation similar as the previous report using RGD conjugated QDs without
64Cu-labeling [95].

22.11 Conclusions

In this review, selected properties, applications and performance of QDs have been
discussed specifically in the context of bionanotechnology and nanomedicine.
Among the other bionanotechnological tools these nanocyrstal structures have led
the way for numerous advances in both fundamental and applied sciences.
Main reason for these advancements is that; QDs exhibit unique optical, elec-
tronic and physical properties as compared to traditional fluorophore materials
namely photostability, single source excitation, narrow emission, multiplexing
capabilities and high quantum yield. Recently, significant progress has been made in
the synthesis of QDs and their biocompatibilities in terms of both theoretical models
and experimental results.
Quantum dots, specifically those which are called new generation, offer more
future promising application potential than both traditional fluorescent probes and
also CdSe QDs to study intracellular processes in human beings at the single
molecule level. Since they have better biocompatibility, these QDs can be used for
in vitro imaging in bionanotechnological applications, long term in vivo
22 Quantum Dots in Bionanotechnology and Medical Sciences … 573

observation of cellular events, tumor targeting and even in diagnostics. Yet, in the
areas like labeling of proteins, tagging nucleic acids and in DNA moments an
extensive research is going on.
Biomedical applications of QDs can serve in an extensive area of research and
applications, such as microscopy, multiplexed histology, flow cytometry, drug
delivery, in vivo long term whole organism and clinical imaging, photodynamic
therapy, tissue mapping and demarcation, real time detection of intra and inter-
cellular events, signaling biosensing, environment and biodefense etc.
What can we expect for the future clinical applications of quantum dots?
The requirement for collecting more biological information from extremely low
concentrations of samples from any organisms (human being till microbiological
specimens), principally for early diagnostics or during the treatment processes, has
pushed the need for such highly sensitive and capability of multiplexing molecules
like quantum dots.
However, the lack of robust and reproducible methods to conjugate many dif-
ferent biomolecules such as antibodies, protein markers, DNA and RNA to QDs in
a systematic manner with control over their ratio, orientation, and its noncovalent
interactions continues to retard their further use in clinical diagnostics. It may seem
like QDs are slowly improving, however, those unresolved problems leads to
extensive new research areas in science. As a result of such efforts, we may expect
more commercial products incorporating QDs for clinical diagnostic, and research
purposes, such as particles that can be used for dual-modality imaging.
We believe that the future of quantum dots in life sciences is as bright as the
quantum dots themselves.

Acknowledgments I would like to give my sincere gratitudes to Prof. Dr. Hilmi Ünlü for his
critical reviews. I would like to express my sincere thanks to all my colleagues especially, Mss.
Busra Ahata (BSc), Mr. Mert Kumru (MSc) and Mr. Halil Kurt (PhD) for their valuable contri-
butions throughout the preparation of this chapter.

References

1. W.C.W. Chan, S.M. Nie, Quantum dot bioconjugates for ultrasensitive nonisotopic
detection. Science 281(5385), 2016–2018 (1998)
2. B. Dubertret, P. Skourides, D.J. Norris, V. Noireaux, A.H. Brivanlou, A. Libchaber, In vivo
imaging of quantum dots encapsulated in phospholipid micelles. Science 298(5599), 1759–
1762 (2002)
3. A. Konkar, S.Y. Lu, A. Madhukar, S.M. Hughes, A.P. Alivisatos, Semiconductor
nanocrystal quantum dots on single crystal semiconductor substrates: high resolution
transmission electron microscopy. Nano Lett. 5(5), 969–973 (2005)
4. A.M. Smith, S. Dave, S.M. Nie, L. True, X. Gao, Multicolor quantum dots for molecular
diagnostics of cancer. Expert Rev. Mol. Diagn. 6(2), 231–244 (2006)
5. B. Ballou, L.A. Ernst, S. Andreko, Sentinel lymph node imaging using quantum dots in
mouse tumor models. Bioconjug. Chem. 18(2), 389–396 (2007)
6. P. Diagaradjane, J.M. Orenstein-Cardona, N.E. Colon-Casasnovas, Imaging epidermal
growth factor receptor expression in vivo: pharmacokinetic and biodistribution
574 İ. Ergal and A.T. Akarsubasi

characterization of a bioconjugated quantum dot nanoprobe. Clin. Cancer Res. 14(3), 731–
774 (2008)
7. D.S. Lidke, P. Nagy, R. Heintzmann, D.J. Arndt-Jovin, J.N. Post, H.E. Grecco, E.A.
Jares-Erijman, T.M. Jovin, Quantum dot ligands provide new insights into erbB/HER
receptor-mediated signal transduction. Nat. Biotech. 22, 198–203 (2004)
8. L.F. Qi, X.H. Gao, Quantum dot—amphipol nanocomplex for intracellular delivery and
real-time imaging of siRNA. ACS Nano 2(7), 1403–1410 (2008)
9. D. Sen, T.J. Deerinck, M.H. Ellisman, I. Parker, M.D. Cahalan, Quantum dots for tracking
dendritic cells and priming an immune response in vitro and in vivo. PLoS One 3(9) (2008)
10. P. Alivisatos, The use of nanocrystals in biological detection. Nat. Biotechnol. 22(1), 47–52
(2004)
11. W.R. Algar, A.J. Tavares, U.J. Krull, Beyond labels: a review of the application of quantum
dots as integrated components of assays, bioprobes, and biosensors utilizing optical
transduction. Anal. Chim. Acta 673(1), 1–25 (2010)
12. S.J. Rosenthal, J.C. Chang, O. Kovtun, J.R. McBride, I.D. Tomlinson, Biocompatible
quantum dots for biological applications. Chem. Biol. 18(1), 10–24 (2011)
13. A.M. Smith, H.W. Duan, A.M. Mohs, S.M. Nie, Bioconjugated quantum dots for in vivo
molecular and cellular imaging. Adv. Drug Deliv. Rev. 60(11), 1226–1240 (2008)
14. A.I. Ekimoy, A.A. Onushchenko, Quantum size effect in three-dimensional microscopic
semiconductor crystals. JETP Lett. 34(6), 345–349 (1981)
15. C.B. Murray, D.J. Norris, M.G. Bawendi, Synthesis and characterization of nearly
monodisperse CdE (E = sulfur, selenium, tellurium) semiconductor nanocrystallites.
J. Am. Chem. Soc. 115(19), 8706–8715 (1993)
16. C.F. Wu, C. Szymanski, J. McNeill, Preparation and encapsulation of highly fluorescent
conjugated polymer nanoparticles. Langmuir 22(7), 2956–2960 (2006)
17. C.F. Wu, C. Szymanski, Z. Cain, J. McNeill, Conjugated polymer dots for multiphoton
fluorescence imaging. J. Am. Chem. Soc. 129(43), 12904–12905 (2007)
18. C.F. Wu, B. Bull, C. Szymanski, K. Christensen, J. McNeill, Multicolor conjugated polymer
dots for biological fluorescence imaging. ACS Nano 2(11), 2415–2423 (2008)
19. S.T. Selvan, Silica-coated quantum dots and magnetic nanoparticles for bioimaging
applications. Biointerphases 5(3), FA110–FA115 (2010)
20. X.H. Gao, S.M. Nie, Doping mesoporous materials with multicolor quantum dots. J. Phys.
Chem. B 107(42), 11575–11578 (2003)
21. I.L. Medintz, A.R. Clapp, H. Mattoussi, E.R. Goldman, B. Fisher, J.M. Mauro, Nat. Mater. 2,
630–638 (2003)
22. F. Erogbogbo, K.T. Yong, I. Roy, In vivo targeted cancer imaging, sentinel lymph node
mapping and multichannel imaging with biocompatible silicon nanocrystals. Nano Lett. 5(1),
413–423 (2011)
23. B. Bruhn, Fabrication and characterization of single luminescing quantum dots from 1D
silicon nanostructures. Doctoral Thesis (2012)
24. J.H. Park, L. Gu, G. VonMaltzahn, E. Ruoslahti, S.N. Bhatia, M.J. Sailor, Biodegradable
luminescent porous silicon nanoparticles for in vivo applications. Nat. Mater. 8(4), 331–336
(2009)
25. J.H. Shen, Y.H. Zhu, X.L. Yang, C.Z. Li, Graphene quantum dots: emergent nanolights for
bioimaging, sensors, catalysis and photovoltaic devices. Chem. Commun. 48, 3686–3699
(2012)
26. P. Pierobon, S. Achouri, S. Courty, A.R. Dunn, J.A. Spudich, M. Dahan, G. Cappello,
Velocity, processively, and individual steps of single myosin V molecules in live cells.
Biophys. J. 96, 4268–4275 (2009)
27. Y. Ishihama, T. Funatsu, Single molecule tracking of quantum dot labeled mRNAs in a cell
nucleus. Biochem. Biophys. Res. Commun. 381, 33–38 (2009)
28. K.M. Tsoi, Q. Dai, B.A. Alman, W.C. Chan, Are quantum dots toxic? Exploring the
discrepancy between cell culture and animal studies. Acc. Chem. Res. 46(3), 662–671 (2012)
22 Quantum Dots in Bionanotechnology and Medical Sciences … 575

29. G. Ruan, A. Agrawal, A.I. Marcus, S. Nie, Imaging and tracking of Tat peptide-conjugated
quantum dots in living cells: new insights into nanoparticle uptake, intracellular transport,
and vesicle shedding. J. Am. Chem. Soc. 129, 14759–14766 (2007)
30. E.L. Bentzen, I.D. Tomlinson, J. Mason, P. Gresch, M.R. Warnement, D. Wright, E.
Sanders-Bush, R. Blakely, S.J. Rosenthal, Surface modification to reduce nonspecific
binding of quantum dots in live cell assays. Bioconjug. Chem. 16, 1488–1494 (2005)
31. X. Gao, Y. Cui, R.M. Levenson, L.W.K. Chung, S. Nie, In vivo cancer targeting and imaging
with semiconductor quantum dots. Nat. Biotech. 22, 969–976 (2004)
32. X. Wu, H. Liu, J. Liu, K.N. Haley, J.A. Treadway, J.P. Larson, N. Ge, F. Peale, M.
P. Bruchez, Immunofluorescent labeling of cancer marker Her2 and other cellular targets
with semiconductor quantum dots. Nat. Biotech. 21, 41–46 (2002)
33. H. Tada, H. Higuchi, T.M. Wanatabe, N. Ohuchi, In vivo real-time tracking of single
quantum dots conjugated with monoclonal anti-HER2 antibody in tumors of mice. Cancer
Res. 67, 1138–1144 (2007)
34. S. Pathak, M.C. Davidson, G.A. Silva, Characterization of the functional binding properties
of antibody conjugated quantum dots. Nano Lett. 7, 1839–1845 (2007)
35. R.L. Orndorff, S.J. Rosenthal, Neurotoxin quantum dot conjugates detect endogenous targets
expressed in live cancer cells. Nano Lett. 9, 2589–2599 (2009)
36. R.L. Orndorff, M.R. Warnement, J.N. Mason, R.D. Blakely, S.J. Rosenthal, Quantum dot
ex vivo labeling of neuromuscular synapses. Nano Lett. 8, 780–785 (2008)
37. T.Q. Vu, R. Maddipati, T.A. Blute, B.J. Nehilla, L. Nusblat, T.A. Desai, Peptide-conjugated
quantum dots activate neuronal receptors and initiate downstream signaling of neurite
growth. Nano Lett. 5, 603–607 (2005)
38. I.D. Tomlinson, J.N. Mason, R.D. Blakely, S.J. Rosenthal, Peptide-conjugated quantum dots.
Methods Mol. Biol. 303, 51–60 (2005)
39. D. Sparks, T. Hubbard, Micromachined needles and lancets with design adjustable bevel
angles. J. Micromech. Microeng. 14(8), 1230 (2004)
40. M.J. Cloninger, Biological applications of dendrimers. Curr. Opin. Chem. Biol. 6(6), 742–
748 (2002)
41. A.M. Derfus, W.C.W. Chan, S.N. Bhatia, Probing the cytotoxicity of semiconductor
quantum dots. Nano Lett. 4(1), 11–18 (2004)
42. R. Hardman, Toxicological review of quantum dots: toxicity depends on physicochemical
and environmental factors. Environ. Health Perspect. 114(2), 165–172 (2006)
43. Y. Pan, S. Neuss, A. Leifert, Size-dependent cytotoxicity of gold nanoparticles. Small 3(11),
941–1949 (2007)
44. X. Michalet, F.F. Pinaud, L.A. Bentolila, Quantum dots for live cells, in vivo imaging, and
diagnostics. Science 307(5709), 538–544 (2005)
45. P.N. Prasad, Biophotonics (Wiley-Interscience, Hoboken, NJ, USA, 2003)
46. S.M. Smith, R. Renden, H. von Gersdorff, Synaptic vesicle endocytosis: fast and slow modes
of membrane retrieval. Trends Neurosci. 31, 559–568 (2008)
47. S. An, D. Zenisek, Regulation of exocytosis in neurons and neuroendocrine cells. Curr. Opin.
Neurobiol. 14, 522–530 (2004)
48. H.T. Li, Z.H. Kang, Y. Liu, S.T. Lee, Carbon nanodots: synthesis, properties and
applications. J. Mater. Chem. 22, 24230–24253 (2012)
49. H. Ming, Z. Ma, Y. Liu, K.M. Pan, H. Yu, F. Wang, Z.H. Kang, Large scale electrochemical
synthesis of high quality carbon nanodots and their photocatalytic property. Dalton Trans. 41,
9526–9531 (2012)
50. A.M. Smith, S. Nie, Next-generation quantum dots. Nat. Biotechnol. 27, 732–733 (2009)
51. A.M. Smith, S. Nie, Minimizing the hydrodynamic size of quantum dots with multifunctional
multidentate polymer ligands. J. Am. Chem. Soc. 130, 11278–11279 (2008)
52. H.S. Choi, W. Liu, P. Misra, E. Tanaka, J.P. Zimmer, Renal clearance of quantum dots. Nat.
Biotechnol. 25, 1165–1170 (2007)
576 İ. Ergal and A.T. Akarsubasi

53. B.A. Kairdolf, M.C. Mancini, A.M. Smith, S. Nie, Minimizing nonspecific cellular binding
of quantum dots with hydroxyl-derivatized surface coatings. Anal. Chem. 80, 3029–3034
(2008)
54. S. Pathak, S.K. Choi, N. Arnheim, M.E. Thompson, Hydroxylated quantum dots as
luminescent probes for in situ hybridization. J. Am. Chem. Soc. 123, 4103–4104 (2001)
55. F.B. Sheinerman, R. Norel, B. Honig, Electrostatic aspects of protein–protein interactions.
Curr. Opin. Struct. Biol. 10, 153–159 (2000)
56. X. Wu, H. Liu, J. Liu, K.N. Haley, J.A. Treadway, Immunofluorescent labeling of cancer
marker Her2 and other cellular targets with semiconductor quantum dots. Nat. Biotechnol.
21, 41–46 (2003)
57. S.F. Lim, R. Riehn, W.S. Ryu, N. Khanarian, C.K. Tung, D. Tank, R.H. Austin, In vivo and
scanning electron microscopy imaging of upconverting nanophosphors in caenorhabditis
elegans. Nano Lett. 6(2), 169–174 (2006)
58. L. Li, T.J. Daou, I. Texier, Highly luminescent CuInS2/ZnS core/shell nanocrystals:
cadmium-free quantum dots for in vivo imaging. Chem. Mater. 21, 2422–2429 (2009)
59. J. Park, J. Nam, N. Won, Compact and stable quantum dots with positive, negative, or
zwitterionic surface: specific cell interactions and non-specific adsorptions by the surface
charges. Adv. Funct. Mater. 21, 1558–1566 (2011)
60. E. Muro, A. Fragola, T. Pons, Comparing intracellular stability and targeting of sulfobetaine
quantum dots with other surface chemistries in live cells. Small 8, 1029–1037 (2012)
61. M. Howarth, W.H. Liu, S. Puthenveetil, Monovalent, reduced-size quantum dots for imaging
receptors on living cells. Nat. Methods 5, 397–399 (2008)
62. X.F. Hu, T.C. Liu, Y.C. Cao, Characterization of the coupling of quantum dots and
immunoglobulin antibodies. Anal. Bioanal. Chem. 386, 1665–1671 (2006)
63. T. Maldiney, M.U. Kaikkonen, J. Seguin, In vitro targeting of avidin-expressing glioma cells
with biotinylated persistent luminescence nanoparticles. Bioconjug. Chem. 23, 472–478
(2012)
64. L. Chen, X.W. Zhang, G.H. Zhou, Simultaneous determination of human enterovirus 71 and
coxsackievirus B3 by dual-color quantum dots and homogeneous immunoassay. Anal.
Chem. 84, 3200–3207 (2012)
65. P.M. Allen, W.H. Liu, V.P. Chauhan, InAs(ZnCdS) quantum dots optimized for biological
imaging in the near-infrared. J. Am. Chem. Soc. 132, 470–471 (2010)
66. P.K. Bae, K.N. Kim, S.J. Lee, The modification of quantum dot probes used for the targeted
imaging of his-tagged fusion proteins. Biomaterials 30, 836–842 (2009)
67. H.Y. Park, K. Kim, S. Hong, Compact and versatile nickel-nitrilotriacetate-modified
quantum dots for protein imaging and forster resonance energy transfer based assay.
Langmuir 26, 7327–7333 (2010)
68. B.C. Lagerholm, M. Wang, L.A. Ernst, D.H. Ly, H. Liu, M.P. Bruchez, A.S. Waggoner,
Multicolor coding of cells with cationic peptide coated quantum dots. Nano Lett. 4(10),
2019–2022 (2004)
69. K. Susumu, I.L. Medintz, J.B. Delehanty, K. Boeneman, H. Mattoussi, Modification of poly
(ethylene glycol)-capped quantum dots with nickel nitrilotriacetic acid and self-assembly
with histidine-tagged proteins. J. Phys. Chem. C 114, 13526–13531 (2010)
70. S.H. Bhang, N. Won, T.J. Lee, Hyaluronic acid-quantum dot conjugates for in vivo
lymphatic vessel imaging. ACS Nano 3, 1389–1398 (2009)
71. H.W. Yu, I.S. Kim, R. Niessner, D. Knopp, Multiplex competitive microbead-based flow
cytometric immunoassay using quantum dot fluorescent labels. Anal. Chim. Acta 750, 191–
198 (2012)
72. J.S. Liou, B.R. Liu, A.L. Martin, Protein transduction in human cells is enhanced by
cell-penetrating peptides fused with an endosomolytic HA2 sequence. Peptides 37, 273–284
(2012)
73. Y. Zhang, M.K. So, A.M. Loening, HaloTag protein-mediated site-specific conjugation of
bioluminescent proteins to quantum dots. Angew. Chem. Int. Ed. Engl. 45, 4936–4940
(2006)
22 Quantum Dots in Bionanotechnology and Medical Sciences … 577

74. A. Charalambous, I. Antoniades, N. Christodoulou, P.A. Skourides, Split-inteins for

simultaneous, site-specific conjugation of quantum dots to multiple protein targets in vivo.
J. Nanobiotechnol. 9, 37 (2011)
75. A.M. Smith, H. Duan, A.M. Mohs, S. Nie, Bioconjugated quantum dots for in vivo
molecular and cellular imaging. Adv. Drug Deliv. Rev. 60, 1226–1240 (2008)
76. M. Baker, Nanotechnology imaging probes: smaller and more stable. Nat. Methods 7, 957–
962 (2010)
77. A.M. Smith, M.M. Wen, S. Nie, Imaging dynamic cellular events with quantum dots—the
bright future. Biochemist 32, 12–17 (2010)
78. A. Mooradian, Photoluminescence of metals. Phys. Rev. Lett. 22(5), 185–187 (1969)
79. Z.Z. Huang, F. Pu, Y.H. Lin, J.S. Ren, X.G. Qu, Modulating DNA-templated silver
nanoclusters for fluorescence turn-on detection of thiol compounds. Chem. Commun. 47,
3487–3489 (2011)
80. Y.T. Su, G.Y. Lan, W.Y. Chen, H.T. Chang, Detection of copper ions through recovery of
the fluorescence of DNA templated copper/silver nanoclusters in the presence of
mercaptopropionic acid. Anal. Chem. 82, 8566–8572 (2010)
81. G.Y. Lan, C.C. Huang, H.T. Chang, Silver nanoclusters as fluorescent probes for selective
and sensitive detection of copper ions. Chem. Commun. 46, 1257–1259 (2010)
82. J. Sharma, H.C. Yeh, H. Yoo, J.H. Werner, J.S. Martinez, Silver nanocluster aptamers: in situ
generation of intrinsically fluorescent recognition ligands for protein detection. Chem.
Commun. 47, 2294–2296 (2011)
83. J.J. Li, X.Q. Zhong, H.Q. Zhang, X.C. Le, J.J. Zhu, Binding-induced fluorescence turn on
assay using aptamer-functionalized silver nanocluster DNA probes. Anal. Chem. 84, 5170–
5174 (2012)
84. H.C. Yeh, J. Sharma, I.M. Shih, D.M. Vu, J.S. Martinez, J.H. Werner, A fluorescence light
up Ag nanocluster probe that discriminates single-nucleotide variants by emission color.
J. Am. Chem. Soc. 134, 11550–11558 (2012)
85. H.F. Dong, S. Jin, H.X. Ju, K.H. Hao, L.P. Xu, H.T. Lu, X.J. Zhang, Trace and label-free
microRNA detection using oligonucleotide encapsulated silver nanoclusters as probes. Anal.
Chem. 84, 8670–8674 (2012)
86. J.L. Wang, G. Zhang, Q.W. Li, H. Jiang, C.Y. Liu, C. Amatore, X.M. Wang, In vivo self bio-
imaging of tumors through in situ biosynthesized fluorescent gold nanoclusters. Sci Rep 3,
1157–1162 (2013)
87. X.Y. Xu, R. Ray, Y.L. Gu, H.J. Ploehn, L. Gearheart, K. Raker, W.A. Scrivens,
Electrophoretic analysis and purification of fluorescent single-walled carbon nanotube
fragments. J. Am. Chem. Soc. 126(40), 12736–12737 (2004)
88. L. Cao, X. Wang, M.J. Meziani, F. Lu, H. Wang, P.G. Luo, Y. Lin, B.A. Harruff, L.M. Veca,
D. Murray, S.Y. Xie, Y.P. Sun, Carbon dots for multiphoton bioimaging. J. Am. Chem. Soc.
129(37), 11318–11319 (2007)
89. S.Y. Ding, M. Jones, M.P. Tucker, J.M. Nedeljkovic, J. Wall, M.N. Simon, G. Rumbles, M.
E. Himmel, Nano Lett. 3, 1581 (2003)
90. B. Hötzer et al., Fluorescence in nanobiotechnology: sophisticated fluorophores for novel
applications. SMALL (2012)
91. K.E. Sapsford et al., Materials for fluorescence resonance energy transfer analysis: beyond
traditional donor–acceptor combinations. Angew. Chem. Int. Ed. 45, 4562–4588 (2006)
92. I.L. Medintz et al., Interactions between redox complexes and semiconductor quantum dots
coupled via a peptide bridge. J. Am. Chem. Soc. 130, 16745–16756 (2008)
93. T. Pons, H. Mattoussi, Investigating biological processes at the single molecule level using
luminescent quantum dots. Ann. Biomed. Eng. 37, 1934–1959 (2009)
94. J. Schlessinger, Cell signaling by receptor tyrosine kinases. Cell 103, 211–225 (2000)
95. W. Cai, D.-W. Shin, K. Chen, O. Gheysens, Q. Cao, S.X. Wang, S.S. Gambhir, X. Chen,
Peptide-labeled near-infrared quantum dots for imaging tumor vasculature in living subjects.
Nano Lett. 6, 669–676 (2006)
578 İ. Ergal and A.T. Akarsubasi

96. B.R. Smith, Z. Cheng, A. De, A.L. Koh, R. Sinclair, S.S. Gambhir, Real-time intravital
imaging of RGD: quantum dot binding to luminal endothelium in mouse tumor
neovasculature. Nano Lett. 8, 2599–2606 (2008)
97. E.L. Bentzen, F. House, T.J. Utley, J.E. Crowe, D.W. Wright, Progression of respiratory
syncytial virus infection monitored by fluorescent quantum dot probes. Nano Lett. 5, 591–
595 (2005)
98. S.J. Rosenthal, I. Tomlinson, E.M. Adkins, S. Schroeter, S. Adams, L. Swafford, J. McBride,
Y. Wang, L.J. DeFelice, R.D. Blakely, Targeting cell surface receptors with
ligand-conjugated nanocrystals. J. Am. Chem. Soc. 124, 4586–4594 (2002)
99. M. Howarth, K. Takao, Y. Hayashi, A.Y. Ting, Targeting quantum dots to surface proteins in
living cells with biotin ligase. Proc. Natl. Acad. Sci. U.S.A. 102, 7583–7588 (2005)
100. M. Howarth, A.Y. Ting, Imaging proteins in live mammalian cells with biotin ligase and
monovalent streptavidin. Nat. Protoc. 3, 534–545 (2008)
101. V. Roullier, S. Clarke, C. You, F. Pinaud, G.R. Gouzer, D. Schaible, V.R. Marchi-Artzner,
J. Piehler, M. Dahan, High-affinity labeling and tracking of individual histidine-tagged
proteins in live cells using Ni2 + Tris-nitrilotriacetic acid quantum dot conjugates. Nano Lett.
9, 1228–1234 (2009)
102. E. Genin, O. Carion, B. Mahler, B. Dubertret, N. Arhel, P. Charneau, E. Doris, C.
Mioskowski, CrAsH_quantum dot nanohybrids for smart targeting of proteins. J. Am. Chem.
Soc. 130, 8596–8597 (2008)
103. M.-K. So, H. Yao, J. Rao, HaloTag protein-mediated specific labeling of living cells with
quantum dots. Biochem. Biophys. Res. Comm. 374, 419–423 (2008)
104. D. Gerion, W.J. Parak, S.C. Williams, D. Zanchet, C.M. Micheel, A.P. Alivisatos, Sorting
fluorescent nanocrystals with DNA. J. Am. Chem. Soc. 124, 7070–7074 (2002)
105. W.J.M. Mulder, R. Koole, R.J. Brandwijk et al., Quantum dots with a paramagnetic coating
as a bimodal molecular imaging probe. Nano Lett. 6, 1–6 (2006)
106. Q. Ma, Y. Nakane, Y. Mori et al., Multilayered, core/shell nanoprobes based on magnetic
ferric oxide particles and quantum dots for multimodality imaging of breast cancer tumors.
Biomaterials 33:8486–8494 (2012)
107. J. Ding, Y. Wang, M. Ma et al., CT/fluorescence dual-modal nanoemulsion platform for
investigating atherosclerotic plaques. Biomaterials 34, 209–216 (2013)
108. J.H. Rudd, F. Hyafil, Z.A. Fayad, Inflammation imaging in atherosclerosis. Arterioscler.
Thromb. Vasc. Biol. 29, 1009–1016 (2009)
109. T. Temma, H. Saji, Radiolabelled probes for imaging of atherosclerotic plaques. Am. J. Nucl.
Med. Mol. Imaging 2, 432–447 (2012)
Chapter 23
Nanomedicine

Eda Tahir Turanlı and Elif Everest

Abstract Nanomedicine is application of nanotechnology to diagnosis, treatment

and prevention of diseases, in which multidisciplinary approaches are used com-
bining chemistry, physics, biology, genetics, and medicine. Currently, there are a
number of approved nanomedicines some of which have been developed to treat
diseases using two main different approaches; passive or active targeting. Others
include in vitro and in vivo diagnostic nanoparticles that have been shown to have
many advantages compared to the traditional methods. Although there are some
safety issues regarding development and use of nanoparticles in medicine and
controversies surrounding the need for nanomedicine-specific regulations; many
nanomedicines have been approved to date, and further improvements are being
pursued with promising success. In this chapter, we summarise the use, advantages,
current and future statuses of pharmaceutical and diagnostic nanomedicines,
including ethical issues and regulations.

23.1 Introduction

Nanomedicine is a multidisciplinary area, which combines chemistry, physics,

biology, genetics, medicine and develops nano-sized tools to diagnose, treat, and
prevent diseases (Fig. 23.1). There are several definitions of nanomedicine that
slightly differ from each other. However according to the European Science
Foundation’s (ESF) Forward Look Nanomedicine [1], the definition is as follows:

E.T. Turanlı (&)

Molecular Biology and Genetics Department, Istanbul Technical University, Maslak, 34469
Istanbul, Turkey
e-mail: [email protected]
E.T. Turanlı
E. Everest
Graduate School of Science Engineering and Technology, Molecular Biology-Genetics and
Biotechnology Programme, Dr. Orhan Öcalgiray Molecular Biology and Genetics Research
Centre, Istanbul Technical University, Maslak, 34469 Istanbul, Turkey

© Springer International Publishing Switzerland 2016 579

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_23
580 E.T. Turanlı and E. Everest

Medicine

Materials
Chemistry
science

Physico-
Biology chemical
characterisation

Genetics Nanomedicine Epigenetics

Fig. 23.1 Interdiciplinary nature of nanomedicine

Nanomedicine uses nano-sized tools for the diagnosis, prevention and treatment of disease
and to gain increased understanding of the complex underlying patho-physiology of dis-
ease. The ultimate goal is improved quality-of-life.

The definition encompasses diagnostic nanoparticles such as in vitro biosensors

and surgical tools, and in vivo imaging agents and monitoring technologies, as well
as therapeutic biomaterials that are used for drug delivery. According to the
European Commission’s (EC) Joint Research Centre report [2]; nanomedicines are
defined as particles with 1–1000 nm size, and biochips with nanoscale components
and polymer therapeutics are also classified as nanomedicine. The significant
development of nanomedicine has been achieved within the last couple of decades,
though the first results were identified in the 1970s [3]. The early pioneer studies
have shown that liposomal formulations of existing drugs have the potential of
increasing therapeutic efficiency and reducing toxicity in non-clinical models,
though these effects could not be well-demonstrated in patients [4]. Therefore,
along with the difficulties in the commercial-scale production of nanoparticles, only
limited investment and research have been made in nanoparticles. In 2003, the ESF
conducted a study involving more than 100 experts which examined the status of
nanotechnology in medical applications, and published a policy summarising their
recommendations for further improvement of the field in 2005 [5]. In addition;
other reports [6–8] and establishments including European Technology Platform
Nanomedicine, European Foundation for Clinical Nanomedicine, Nanomedicine
Development Centres of National Institutes of Health (NIH) were developed for
further improvement of the field.
23 Nanomedicine 581

All noncommunicable diseases (NCD), also known as chronic diseases, such as

cancer, diabetes, and cardiovascular diseases cause damage of cells, tissues, organs,
and ultimately related systems in the body. NCDs represent 63 % of annual deaths
[9], constituting a large proportion of death causes in the world. Collins, director of
NIH, states that among approximately 4,000 diseases whose biological mechanisms
have been elucidated, only 250 of them have available treatments currently [10].
Unfortunately, most of the available medical approaches cannot be classified as
‘’satisfactory treatment’’, enabling only symptomatic recovery of the patients,
rather than being fully curative [11]. In addition to the evident need for new
effective treatment strategies, early diagnosis that can increase survival rate is yet
another important issue for patients who have diseases like cancer, which needs to
be further improved. Similarly, in many neurodegenerative diseases such as mul-
tiple sclerosis (MS), patients can manifest clinical symptoms after the initiation of
biological disease activity [12]. Therefore, it is very important to identify indi-
viduals at risk for these diseases, in which treatment at earlier stages usually defines
milder disease course. Nanomedicine is a promising avenue for at least partly filling
these huge gaps in diagnostic and therapeutic approaches in the field of medicine.
To date, many successful nanomedicines have been developed using multidisci-
plinary approaches, and further improvements are being pursued. In this chapter, we
review the current status of pharmaceutical and diagnostic nanomedicines, as well
as the ethical issues and regulations regarding the field of nanomedicine.

23.2 Nanopharmaceuticals

Application of nanopharmaceuticals aims to specifically direct diseased tissues or

cells, thereby preventing side effects caused by damages to healthy tissue in con-
ventional treatments, and improving the therapy efficacy. These nanoparticles used
as drug carriers have many advantages, especially over the standard cancer thera-
pies. Conventional chemotherapies cause death of healthy cells along with cancer
cells, leading to a load of serious adverse side effects. By delivering nanoparticles
directly to the tumour tissue, prevention of the drugs interacting with healthy tissue
before they reach the target tissue are achieved. Moreover, efficacy of the treatments
is enhanced, since drug absorption proportion solely into cancer cells is increased.
There have been studies on three generation vectorisation strategies for
nanomedicines, which are nanospheres and nanocapsules, hydrophilic polymer like
polyethylene glycol (PEG)-coated nanoparticles, and vectors constituting of
biodegradable cores and polymer envelopes with membrane recognition ligands.
Using the vectorisation systems, many nanopharmaceutical products have already
been marketed (Table 23.1). Nanoparticlate drug delivery has been achieved by a
number of different approaches, of which passive targeting relies on “enhanced
permeability and retention (EPR)’’ principle [13]. In EPR effect, particles smaller
than 100–200 nm are taken up to tumour cells more easily due to their leaky
blood vessels with greater porosity compared to the normal vasculature.
582 E.T. Turanlı and E. Everest

Table 23.1 Examples of nanomedicines

Product Indication Status Nanotechnology Company
name
Doxil Cancer (various) Approved PEGylated Johnson and
liposomal Johnson/Schering-Plough
doxorubicin
Myocet Breast cancer Approved Liposomal Cephalon
doxorubicin
Pegasys Hepatitis C Approved PEGylated interferon Roche
alpha 2a
Abraxane Breast cancer Approved Albumin-paclitaxel Abraxis (Celgene)
nanoparticles
Depodur Pain relief Approved Liposomal morphine EKR Therapeutics Inc
Venofer Anemia Approved Iron oxide, sucrose Fresenius
Rapamune Immunosuppressive Approved NanoCrystal Elan Pfizer/Wyeth
ThermoDox Hepatocellular Phase 3 trial Temperature Celsion
carcinoma sensitive liposomal
doxorubicine+RFA
CPX-351 Acute myeloid Phase 3 trial Liposomal Celator Pharmaceuticals
leukemia cytarabine and
daunorubicine
combination
BIND-014 Prostate cancer, Phase 2 trial Targeted BIND Biosciences
lung cancer nanoparticle
MM-398 Pancreatic cancer Phase 3 trial Liposome Merrimack
Pharmaceuticals
NKTR-105 Cancer (various) Phase 1 trial PEGylated docetaxel Nektar

When nanoparticles are taken up to the tumour vessels, they have a long retention
time due to poor lymphatic drainage in tumours. To increase the EPR effect, lipid
composition of nanoparticles have been modified, allowing prolonged circulation in
blood. Doxil (doxorubicin HCl liposome injection) for instance, developed to be
used in cancer treatment, was designed based on this principle [14]. Both in xeno-
graft models, and in the first clinical trial, enhanced accumulation of Doxil in tumour
tissues was shown compared to the non-liposomal parent, free doxorubicine [15, 16].
In addition to the passive targeting, other promising approaches have been
investigated mostly focusing on active targeting, in which nanoparticles with
molecules that bind receptors specifically expressed on diseased cells are generated,
allowing more accurate targeting. In this context, N-(2-Hydroxypropyl)
methacrylamide (HPMA) copolymer-doxorubicine conjugates including galac-
tosamine for targeting asialoglycoprotein receptor on hepatocytes and hepatoma
was generated, revealing that after 24 h of dose application, 15–20 % liver targeting
is achieved, although the higher level of accumulation was observed in normal liver
(16.9 %) than in hepatoma (3.2 %) [17]. Still, the accumulation level in hepatoma
was 12–50 times increased, compared to administration of free doxorubicine.
Moreover, nanoshells that absorb light at different wavelengths generating heat
23 Nanomedicine 583

have been designed for an application of light with specific wavelength inducing
hyperthermia in cancer cells once they are delivered the nanoshells by active tar-
geting. An example for this system is ThermoDox, which is a radio frequency
ablation (RFA)-sensitive liposomal formulation of doxorubicin [18]. When induced
by RFA, the active drug is locally released into tumour cells, which is suggested to
improve the outcome of hepatocellular carcinoma patients who receive RFA.
Currently, there are 93 clinical trials with the “nano’’ prefix on NIH (as of
16.06.2015), which passed the pre-clinical stages [19]. However, it is known that
only about 5 % of drugs that enter clinical trials for all types of cancer therapies are
later approved to be marketed [20]. This emphasises the importance of biomarkers
for predicting parameters of developing nanomedicines, since the results of
potential nanomedicines in vitro culture/in vivo animal model studies and human
subjects can greatly vary. Potential nanomedicine biomarkers include;
• Markers showing healthy tissue toxicity
• Evidence of diseased tissue targeting
• Presence of receptors for active targeting
• Presence of activating enzymes and conditions
• Absence of resistance markers
• Early evidence of response
These potential nanomedicine-specific biomarkers mostly predicting targeting
and toxicity should be identified and tested with respect to each product for
enhanced efficacy and safety, which will improve patient selection for the therapies.
However, studies on nanomedicine biomarkers are at early stages and required to be
carefully developed with respect to each condition.

23.3 Diagnostic and Theranostic Nanomedicine

Use of nanoparticles in medical diagnosis have many advantages compared to the

traditional methods. Such new generation in vitro and in vivo diagnostic devices are
both smaller, faster, more specific and sensitive. In terms of in vitro diagnosis, these
devices also require smaller samples from the individuals. An in vitro nano-sized
diagnostic device, a bionsensor, is composed of a biological material that can
recognise the presence, activity, or level of a biological molecule, which is then
converted to a quantifiable signal by a transducer. Biosensors can be used both in
diagnosis of a disease and monitoring a therapy progress. In vivo imaging diag-
nostic devices comprise identifying tracers or contrast agents that mark the related
sites in the body. Commonly used agents include fluorescent dyes, radionuclides,
superparamagnetic iron oxides, and heavy elements. Targeted molecular imaging
can provide detection of localisation of a tumour, as well as disease states, sites of
inflammation, visualisation of vascular structures etc., allowing detection of dis-
eases as early as possible. In vivo diagnostics also include implantable devices such
as swallowable imaging pills, that enable monitoring of circulating molecules.
584 E.T. Turanlı and E. Everest

Although nano-harvesting of biomarkers have the potential to significantly improve

early disease detection, there is a limitation that albumin and immunoglobulin
constitute about 90 % of all molecules in blood, thereby preventing high signals
from the trace substances. However, developing polymer-based nanoparticles seem
to overcome this problem.
Nano-biopsy, in which a small amount of cells and molecules are obtained from
the target area, is specifically advantageous in the detection of brain tumours, since
its diagnosis is generally the most difficult one to make among all cancer types, and
removing brain tissue by conventional biopsy is more dangerous due to the delicate
nature of the brain. In a study of Stott et al. [21] for instance, a microchip named
HB-Chip was developed to capture tumour cells from circulating blood (CTCs),
which may later be used in the assessment of therapy effectiveness. The system is
based on antibody-coated walls of the chip that specifically bind to cancer cells. The
unique property of the chip is that it does not just detect cancer cells, but it also
traps them, allowing further analyses on these cells, which may eliminate the need
for conventional biopsy in certain cancer types in the future.
Combination of the diagnostic capability of a device with therapeutic property is
referred by the term ‘nanotheranostics’. In the study of Muthu et al. [22] for example,
MCF-7 breast cancer cell line was targeted by D-alpha tocopheryl polyethylene glycol
1000 succinate (TPGS)-coated liposomes with folic acid specific to folate receptor on
the cells. The nanoparticle included docetaxel for therapy and quantum dots for
imaging, leading to enhanced cellular uptake and toxicity for cancer cells [22].
Examples for nano-imaging products are shown in Table 23.2.

23.4 Ethics and Regulation

Nanotechnology in medical applications raises issues such as safety, sensitivity of

genetic information and health care resources, as well as the interface between human
and ‘machine’ like implants and chips. Application of nanomedicines to patients may

Table 23.2 Examples of nano-imaging agents

Product Indication Status Nanotechnology Company
name
Endorem Liver imaging Approved SPION dextran coating AMAG
Pharmaceuticals/Guerbet
SA
Gastromark Gastrointestinal Approved SPION silicone coating AMAG Pharmaceuticals
imaging
Lumirem Gastrointestinal Approved SPION silicone coating AMAG Pharmaceuticals
imaging
Resovist Small liver lesion Approved SPION Bayer Healthcare
imaging Pharmaceuticals
KER-0001 Angiogenesis Phase 1 Targeted Kereos
imaging trial nanoparticle/fluorine-19
23 Nanomedicine 585

cause serious adverse effects by rapid inflammatory-like hypersensitivity reactions or

slower specific responses against the nanoparticles. Indeed, possible toxic effects of
specific nanomedicines have been implicated in a number of studies [23, 24]. Also,
long-term consequences of nanomedicines with no observed toxic effects in
pre-clinical and clinical studies have not been well-known yet. Germline therapy with
nanomedicines or possible effects of any medication with nanoparticles on germline
cells raise yet another ethical issue, since they can cause serious damages to be later
passed down to next generations. Regarding the safety issues, people who work in this
field also stand at risk due to the possibility of being exposed to the nanoparticles by
inhalation, swallowing, or interaction with skin [25, 26]. In addition, the clinical trial
literature of first generation nanomedicines is rarely cited in new studies, even though
the there is a huge clinical data from these first generation products. Thus, conducting
new clinical trials without considering earlier experiences is another ethical issue in
clinical research of nanomedicines. Therefore, there should be a comprehensive
ethical analyses regarding nanomedicines based on bioethic principles, in order to
eliminate their possible adverse impacts on public health and balance the improve-
ment of the nanomedicine field and restrictions for sustaining ethical, social, and legal
boundaries.
Regulatory framework for current nanomedicines are based on benefit/risk
assessments, even though there are no specific procedures. The first generation
nanomedicines have been progressed from laboratory to clinical use by considering
existing regulatory procedures. However, some criticisms concerning clinical safety
of these products have emerged, suggesting that toxicological evaluation of the
products should be done on a case by case basis. According to the US Food and
Drug Administration (FDA) though, there is no need for a regulation specific to
nanomedicines, since they have no significant differences from any other new
technology. Still, there is an ongoing debate regarding nanomedicine-specific
regulations, which will shape the procedures to be followed later on [27–29].

23.5 Conclusion

Considering there are many approved nanomedicines, and many others under
clinical trials, a significant number of therapeutic and diagnostic nanoparticles will
be further developed. Although different carrier designs and targeting strategies for
nanomedicines with promising success have been developed, toxicity issues have
been mostly disregarded. In case of overcoming the safety, as well as other ethical
issues, the field of nanomedicine seems to make a significant contribution to
healthcare in the future.
586 E.T. Turanlı and E. Everest

References

1. European Science Foundation’s Forward Look Nanomedicine: An EMRC Consensus Opinion

(2005). http://www.esf.org
2. V. Wagner, B. Hüsing, S. Gaisser, Nanomedicine: Drivers for development and possible
impacts. European Commission’s (EC) Joint Research Centre (JRC) Report No. 46744, 1–116
(2008)
3. G. Gregoriadis, B.E. Ryman, Liposomes as carriers of enzymes or drugs: a new approach to
the treatment of storage diseases. Biochem. J. 124, 58P (1971)
4. P.D. Brown, P.R. Patel, Nanomedicine: a pharma perspective. Wiley Interdiscip Rev
Nanomed Nanobiotechnol 7, 125–130 (2015)
5. The European Science Foundation, Nanomedicine: An ESF—European Medical Research
Councils (EMRC) Forward Look report (2005). http://www.esf.org
6. Commission of the European Communities Communication, Towards a European Strategy for
Nanotechnology. EU, DG Research, Brussels, (2004). www.cordis.lu/nanotechnology
7. National Institutes of Health, NIH roadmap: nanomedicine. NIH, USA, (2004). http://
nihroadmap.nih.gov
8. National Institutes of Health—National Cancer Institute, Cancer nanotechnology plan: a
strategic initiative to transform clinical oncology and basic research through the
directed application of nanotechnology, NCI, NIH, USA, (2004). http://nano.cancer.gov/
alliance_cancer_nanotechnology_plan.pdf
9. World Health Organization, 10 facts on noncommunicable diseases, (2013). http://www.who.int
10. F.S. Collins, We need better drugs—now. TEDMed Conference (Long Beach, CA, USA,
2012)
11. T. Bartfai, G.V. Lees, The Future of Drug Discovery: Who Decides which Diseases to Treat?
(Academic Press, 2013)
12. D.T. Okuda, A. Siva, O. Kantarci, M. Inglese, I. Katz, M. Tutuncu, B.M. Keegan, S. Donlon,
H. Hua le, A. Vidal-Jordana, X. Montalban, A. Rovira, M. Tintoré, M.P. Amato, B. Brochet,
J. de Seze, D. Brassat, P. Vermersch, N. De Stefano, M.P. Sormani, D. Pelletier, C. Lebrun,
Radiologically Isolated Syndrome Consortium (RISC), Club Francophone de la Sclérose en
Plaques (CFSEP) Radiologically isolated syndrome: 5-year risk for an initial clinical event.
PLoS One 9, e90509 (2014)
13. Y. Matsumura, H. Maeda, A new concept for macromolecular therapies in cancer
chemotherapy: mechanism of tumouritropic accumulation of proteins and the antitumour
agent smancs. Cancer Res. 6, 6387–6392 (1986)
14. Y. Barenholz, Doxil®—the first FDA-approved nano-drug: lessons learned. J. Control Releas.
160, 117–134 (2012)
15. J. Vaage, E. Barberá-Guillem, R. Abra, A. Huang, P. Working, Tissue distribution and
therapeutic effect of intravenous free or encapsulated liposomal doxorubicin on human
prostate carcinoma xenografts. Cancer 73, 1478–1484 (1994)
16. A. Gabizon, R. Catane, B. Uziely, B. Kaufman, T. Safra, R. Cohen, F. Martin, A. Huang, Y.
Barenholz, Prolonged circulation time and enhanced accumulation in malignant exudates of
doxorubicin encapsulated in polyethylene-glycol coated liposomes. Cancer Res. 54, 987–992
(1994)
17. L.W. Seymour, D.R. Ferry, D. Anderson, S. Hesslewood, P.J. Julyan, R. Poyner, J. Doran, A.
M. Young, S. Burtles, D.J. Kerr, Cancer Research Campaign Phase I/II Clinical Trials
committee, Hepatic drug targeting: phase I evaluation of polymer bound doxorubicin. J. Clin.
Oncol. 20, 1668–1676 (2002)
18. B. Wood, R.T.P. Poon, Z. Neeman, M. Eugeni, J. Locklin, S. Dromi, S. Kachala, R. Probhakar,
W. Hahne, S.K. Libutti, Phase I study of thermally sensitive liposomes containing doxorubicin
(ThermoDox [TD]) given during radiofrequency ablation (RFA) in patients with unresectable
hepatic malignancies. ASCO Gastrointestinal Cancers Symposium (2007). abstr. 188
19. ClinicalTrials.gov, a service of the U.S National Institutes of Health. http://clinicaltrials.gov
23 Nanomedicine 587

20. D. Sarker, P. Workman (2007) Pharmacodynamic biomarkers for molecular cancer

therapeutics. Adv. Cancer Res. 96, 213–268
21. S.L. Stott, C.H. Hsu, D.I. Tsukrov, M. Yu, D.T. Miyamoto, B.A. Waltman, S.M. Rothenberg,
A.M. Shah, M.E. Smas, G.K. Korir, F.P. Jr Floyd, A.J. Gilman, J.B. Lord, D. Winokur, S.
Springer, D. Irimia, S. Nagrath, L.V. Sequist, R.J. Lee, K.J. Isselbacher, S. Maheswaran, D.A.
Haber, M. Toner, Isolation of circulating tumor cells using a microvortex-generating
herringbone-chip. Proc. Natl. Acad. Sci. 107, 18392–18397 (2010)
22. M.S. Muthu, S.A. Kulkarni, A. Raju, S.S. Feng, Theranostic liposomes of TPGS coating for
targeted co-delivery of docetaxel and quantum dots. Biomaterials 33, 3494–3501 (2012)
23. G. Oberdörster, Pulmonary effects of inhaled ultrafine particles. Int. Arch. Occup. Environ.
Health 74, 1–8 (2001)
24. P. Jani, G.W. Halbert, J. Langridge, A.T. Florence, Nanoparticle uptake by the rat
gastrointestinal mucosa: quantitation and particle size dependency. J. Pharm. Pharmacol. 42,
821–826 (1990)
25. J. Lademann, H. Weigmann, C. Rickmeyer, H. Barthelmes, H. Schaefer, G. Mueller, W.
Sterry, Penetration of titanium dioxide microparticles in a sunscreen formulation into the
horny layer and the follicular orifice. Skin Pharmacol. Appl. Skin Physiol. 12, 247–256 (1999)
26. G. Oberdörster, Z. Sharp, V. Atudorei, A. Elder, R. Gelein, A. Lunts, W. Kreyling, C. Cox,
Extrapulmonary translocation of ultrafine carbon particles following whole-body inhalation
exposure of rats. J. Toxicol. Environ. Health A 65, 1531–1543 (2002)
27. R. Gaspar, R. Duncan, Polymeric carriers: preclinical safety and the regulatory implications for
design and development of polymer therapeutics. Adv. Drug Deliv. Rev. 61, 1220–1231 (2009)
28. M. Eaton, Nanomedicine: industry-wise research. Nat. Mater. 6, 251–253 (2007)
29. R. Nijhara, K. Balakrishnan, Bringing nanomedicines to market: regulatory challenges,
opportunities, and uncertainties. Nanomed 2, 127–136 (2006)
Chapter 24
Microfluidics and Its Applications
in Bionanotechnology

Z.P. Çakar and B. Sönmez

Abstract Several advantages of microfluidic systems such as rapid control,

reduced size, low costs, large-scale integration and in vivo simulation of cellular
microenvironments make them suitable for diverse bionanotechnological applica-
tions. This chapter reviews major current and potential microfluidic applications in
bionanotechnology, including miniaturized devices related to common molecular
biological techniques such as polymerase chain reaction (PCR), DNA microarray
and electrophoresis; microfluidic bioreactors; and monitoring microbial behaviour
by microfluidics. Additionally, present applications and future potential of micro-
fluidics in microbial strain development and single cell analysis/characterization are
discussed.

24.1 Introduction to Bionanotechnology and Microfluidics

Biological systems consist of complex molecular machineries, as shown by modern

molecular biology research. Individual molecules are the functional constituents of
these complex machineries. There are two major reasons that make living systems
crucial for nanoscience and nanotechnology. One reason is the strong evidence
provided by biological systems for the functionality of nanostructured systems. This

Z.P. Çakar (&)
B. Sönmez

Department of Molecular Biology & Genetics, Faculty of Science and Letters,
Istanbul Technical University, 34469 Maslak, Istanbul, Turkey
e-mail: [email protected]
B. Sönmez
e-mail: [email protected]
Z.P. Çakar
B. Sönmez
Dr. Orhan Öcalgiray Molecular Biology, Biotechnology and Genetics Research Center
(ITU-MOBGAM), Istanbul Technical University, Maslak, Istanbul 34469, Turkey

© Springer International Publishing Switzerland 2016 589

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_24
590 Z.P. Çakar and B. Sönmez

is also known as biomimetics, a technology development approach inspired by

biology. The second reason is that understanding life at nanoscale has a wide range
of important applications in medicine, agriculture, engineering and sustainability of
life. Using biomimetic approaches, self-assembly of biological molecules in vitro,
namely outside a living organism, has been used to reconstitute biological function,
which is the basis of bionanotechnology [1].
The invention of Polymerase Chain Reaction (PCR) method is a breakthrough in
modern molecular biology, and also has significant impact on bionanotechnology.
This method enables in vitro DNA replication. In this method, there is no need for
the whole molecular machinery to denature DNA and to polymerize it, but finding a
thermostable DNA polymerase, such as the one isolated from the thermophilic
bacterium Thermus aquaticus, is necessary. At the RNA level, DNA microarrays
constitute a powerful method by reading the RNA present in a biological sample
and determining which genes are expressed at a particular time or under particular
conditions. Thus, they provide valuable information about the state of a biological
system [1].
The microelectronics industry initiated the progress in miniaturization, where
fabrication sizes became comparable to those of biological molecules. Additionally,
progress in chemistry allows the synthesis of an increasing number of constituents
with a precision at atomic level. Thus, nanostructured materials can be fabricated
and used for a wide range of nanotechnology applications, including diagnostics
and detection [1].
Microfluidics can be defined as manipulation and precise control of fluids in
channels that have cross-sectional dimensions in 10–100 µm range [2]. This central
technology in many miniaturized systems has found a wide range of chemical,
biological and medical applications. Early microfluidics systems were fabricated
using microelectronics-derived technologies, such as photolithography and etching
in silicon and glass. However, silicon is an expensive material and is not envi-
ronmentally friendly. Additionally, it is an opaque material in the visible/UV range
of spectrum which prevents its use for optical detection. Glass, on the other hand, is
transparent, but it is amorphous, making vertical side walls difficult to etch. Most
importantly, sealing of silicon and glass required clean room facilities, and high
voltages or temperatures [2]. These difficulties were overcome by the use of new
materials such as inexpensive polymers. In the late 90s, soft-lithography using
polymer molding was introduced to fabricate cheap microfluidic devices.
Particularly, soft-lithography of poly-di-methyl-siloxane (PDMS) became the basis
of microfluidic systems for cell biology applications [3].
There are many advantages of microfluidics systems for biological applications:
these systems are fast-responding microsystems, where changes in media, envi-
ronmental changes and temperature control are fast. As the reduced size also
reduces the effects of inertial forces compared to frictional ones, laminar flow is
observed as the flow regime in microfluidic channels. Laminar flow allows static
and dynamic gradient formation at subcellular resolution. Other practical
24 Microfluidics and Its Applications in Bionanotechnology 591

advantages of microfluidic systems are low reagent consumption, simultaneous and

independent manipulation of large number of cells, automatic generation of various
unique conditions, and large-scale integration. Additionally, the versatility of
microfluidic devices allows simulation of in vivo cellular microenvironment,
including vascularization and nutrient stress conditions [3].
The major present challenge of microfluidic systems is the relatively slow dif-
fusive mixing resulting from the laminar flow regime, which limits applications that
require a rapidly homogenizing flow. To overcome this limitation, the use of
integrated mixers has been suggested [3, 4]. Other challenges of microfluidic
systems include difficulties in adapting biological protocols to fit experiments in
microsystems, such as media and cell concentrations; the bias resulting from bio-
molecule absorption or adsorption from the medium due to the affinity of PDMS for
small hydrophobic molecules; and osmolarity changes resulting from media drying
because of the permeability of PDMS to water vapor [3].
Technological advances in bionanotechnology include a variety of microfluidic
applications, also for systems biomedicine [5]. As systems biology approach views
and models a biological system as a whole, it also requires large amounts of data
and information for analysis. Microfluidic system examples that help fulfill this
requirement are an integrated microfluidic device for high-throughput measurement
of molecular affinities between DNA and transcription factors to build inter-relation
networks [6]; and integrated use of digital PCR with microfluidic manipulation and
microarrays to obtain data from a single bacterium [7]. The latter is crucial for
single cell analysis of heterogeneous microbial populations. Technological advan-
ces in bionanotechnology that involve microfluidics systems include nano- and
microfluidic systems for handling fluids in high-throughput biological experiments
with low sample volume [8]; real time quantitative reverse transcription polymerase
chain reaction (qRT-PCR) applications with integrated microfluidic chips for single
cell quantification of gene expression levels [9]; microfluidic devices with extra-
cellular matrix-mimicking hydrogel applications for in vitro high-throughput drug
screening, such as organ-on-a-chip platforms for systematic investigations of
pharmacokinetics [10]; high-throughput detection of target molecules by combining
conventional label-free detection methods with micro- or nanofluidic platforms
[11]; and next generation DNA sequencing which is a breakthrough in genome and
transcriptome analysis [5].
The present chapter reviews a wide range of current and potential microfluidics
applications in bionanotechnology (Fig. 24.1). Firstly, it focuses on microfluidics
applications in commonly used biological techniques such as PCR, DNA
microarray systems, and electrophoresis. It then discusses the development of
microfluidic bioreactors and their applications. Application of microfluidics tech-
nologies for monitoring microbial behavior and for microbial strain development is
also reviewed, with an emphasis on evolutionary engineering approach to select for
improved microbial strains. Finally, microfluidics applications in single cell studies
are discussed.
592 Z.P. Çakar and B. Sönmez

Fig. 24.1 Bionanotechnological microfluidics application areas reviewed in the present chapter

24.2 Microfluidic PCR Applications

Micro-electro-mechanical-systems (MEMS) technology enabled miniaturization of

analytical devices used in biology and chemistry which had a strong influence on a
variety of fields such as medical diagnostics, microbial detection, etc. Among
miniaturized analytical devices, those related to PCR are studied extensively [12].
A review article published in 2005 summarizes properties of an automated
chip-based PCR device for simple and rapid amplification of nucleic acids. The
rapid temperature ramping ability and microliter-range sample volumes are the
advantages of the described system [13]. Developments in microfluidics technology
also include real-time PCR applications: in 2006, it was reported that a real-time
PCR analyser was developed that is able to analyse 2 µl reactions in glass capil-
laries [14].
Application of microfluidic technologies to PCR has also been useful for the
detection of low-abundant DNA point mutations. For this purpose, flow-through
biochips have been microfabricated using two different materials: a polycarbonate
chip for performing the PCR reaction and the allele-specific ligation detection
reaction in a continuous-flow format, and a poly(methyl methacrylate) (PMMA)
chip for the detection of the allele-specific ligation detection reaction products [15].
Forensic science is a field that makes routine use of PCR-based analytical
techniques. Thus, development of microfluidic PCR devices also influenced
forensic medicine. The use of microfluidic devices for forensic DNA analysis has
been reviewed in 2007, where advances in microfluidic separation of
short-tandem-repeat (STR) fragments, and the analytical processes related to
sample preparation have been discussed, along with the general advantages and
challenges of microfluidic techniques [16]. In 2010, the development of an
24 Microfluidics and Its Applications in Bionanotechnology 593

Table 24.1 Selected reports on improvement of miniaturized PCR devices

Improvement area References
Droplet technologies [19–22]
Transportation mechanism [23]
Extrusion-driving method based on microfabricated PDMS [24]
pneumatic valves
Vapor-diffusion barrier for pressure reduction [25]
Integrated design and optimization of a cyberphysical digital-microfluidic [26]
biochip

automated instrument for human STR identification has been reported: the instru-
ment involved a lab-on-a-chip automation system to enable DNA fingerprinting that
is compatible with the CODIS standards used by forensic scientists. The minia-
turized instrument aimed to improve costs, duration and simplicity of the analysis
and to exclude human intervention [17]. A very recent forensic science article
reported an improvement of the miniaturized PCR-STR typing method by using
digital agarose droplet microfluidics: the improvement is based on minimizing
inhibition by PCR inhibitors. The inhibitors were efficiently removed from the
porous agarose microgel droplets that carry the DNA template through washing and
significant dilution. Thus, the improved method could allow analysis of
low-abundance DNA targets in the presence of inhibitors [18].
Regarding improvement of miniaturized PCR devices, there are various reports
in the literature (Table 24.1). Some studies focused on droplets: the first
lab-on-a-chip system for picoliter droplet generation and PCR amplification
including real-time fluorescence detection enabled PCR in isolated droplets of
picoliter range [19]. A droplet microfluidic chip allowing amplification-free
detection of single pathogenic cells was also developed: in that chip, a peptide
nucleic acid fluorescence resonance energy transfer probe was used to detect 16S
rRNA in pathogenic cells [20]. The advantages of droplet microfluidics in the
high-throughput analysis of single cells and molecules in cancer research, diag-
nostics and therapy with exceptionally high sensitivity have also been reviewed
recently [21]. Similarly, novel droplet platforms for the detection of disease
biomarkers have also been reviewed recently, by emphasizing the importance of
microfluidic droplet technologies for personalized medicine applications [22].
Other studies on the improvement of miniaturized PCR systems include
improvement of the transport of DNA samples and PCR reagents by a novel liquid
transportation mechanism based on a suction-type membrane and three microvalves
[23], introduction of a novel extrusion driving method based on microfabricated
PDMS pneumatic valves [24], reduction of the pressure increase in fully closed
microfluidic PCR systems due to high operating temperatures [25], and an inte-
grated design and optimization of a cyberphysical digital-microfluidic biochip for
the entire miniaturized PCR procedure [26].
594 Z.P. Çakar and B. Sönmez

24.3 Microfluidic DNA Microarray Systems

DNA microarray technology is a powerful tool for medicine, biomedical and basic
research, as DNA microarrays can be used to search for nucleic acids and molecular
signatures related to a disease or to monitor a physiological/environmental condi-
tion. Microfluidics has many advantages such as small sample volumes, minimized
reagent consumption and costs, automated sample preparation and reduced sample
processing time. Thus, the use of microfluidics in DNA microarray technology also
reduces reagent costs and hybridization assay times, provides high-throughput
sample processing, and integration and automation abilities of the front-end sample
processing steps. Challenges of microfluidic DNA microarray technology include
the need for low-cost manufacturing methods for the fluidic chips, providing good
interfaces to the macro-world, minimizing high surface-to-volume-ratio-related
problems such as non-specific analyte-wall interactions, the need for materials for
the optical readout phases, and complete integration of the optical and electrical
components to the microfluidics [27]. As flow is challenging in microfluidic sys-
tems, various methods developed to generate nanoliter flows in the microchannels
of the microfluidic DNA microarray systems have been discussed in a more recent
review. These methods include electrokinetic control, vacuum suction and syringe
pumping. Of these, vacuum suction and syringe pumping are pressure-driven, and
do not require consideration of the physicochemical properties of solutions, which
allows flexibility. Centrifugal force has also been employed to drive liquid move-
ment in microchannels, by using the body force from the liquid. Besides, no
additional solution interface contacts from electrodes, syringes and tubing exist.
When flow is provided by centrifugal force, parallel hybridizations are also made
easily [28].
DNA microarrays are usually processed under one condition, which requires that
all probes of the array have to function optimally under that particular assay con-
dition. However, cross-hybridizition is a common microarray problem, which
results from a poor combination of the probes chosen and the assay conditions.
Microfluidics-based technologies provide new array processors which are powerful
tools for constructing highly specific DNA microarrays according to the needs of
samples, including special samples that have DNA regions with highly varying G
+C content. Thus, high sample throughput can be combined with optimal assay
conditions [29].
Semiconductor nanoparticles known as quantum dots have unique optical fea-
tures such as high brightness, narrow emission band, etc. which make them
advantageous over traditional organic fluorophores. Thus, they have been replacing
traditional organic fluorophores to be used as simple fluorescent reporters in a
variety of biological techniques such as immunoassays, microarrays, fluorescent
imaging, etc. Combination of advanced quantum dot probes with microfluidics and
confocal fluorescence spectroscopy further enhances the detection limit to single
molecule level [30].
24 Microfluidics and Its Applications in Bionanotechnology 595

There are various reports in the literature regarding present and future medical
applications of microfluidic DNA microarray systems. An important example is the
use of a microfabricated DNA microarray system for the pharmacogenomic profiling
of cytochrome P450-mediated drug metabolism which can also be viewed as a work
related to personalized medicine [31]. A more recent review discussed developments
in enzymatically assisted detection of nucleic acids on microarrays and the potential
integration of these new techniques into personalized medicine [32].
Applications of microfluidic DNA microarray systems are not limited to medi-
cine. Basic and environmental microbiology research also makes use of micro-
fluidic DNA microarray systems. Investigation of microbial diversity and detection
of specific microorganisms (e.g. pathogens) in environmental samples such as
wastewater are allowed without cultivation, by advances in molecular biology
methods including microarray assays [33]. The use of chip-based microfluidic
systems such as microarrays and microfluidic biochips for rapid pathogen sensing
has also been reviewed previously [34].

24.4 Microfluidic Applications in Electrophoresis

Analysis of biological systems or microbial populations at the single cell level requires
techniques that can rapidly detect and quantify a variety of analytes from extremely
small sample volumes with very high sensitivity and selectivity. For this purpose,
microelectrophoretic separations which use both traditional capillary electrophoresis
and modern microfluidic methods are employed. Microelectrophoretic separation
techniques allow quantification of diverse analytes in cells, such as amino acids,
neurotransmitters, large proteins and subcellular organelles [35]. In a recent review on
microchip electrophoresis, it has been reported that electrophoresis has been the most
common separation method used in microfluidic applications for more than two
decades [36].
Capillary electrophoresis is a widely used technique for water quality analysis,
drug development and quality control, proteomics and DNA analysis,
counter-terrorism and monitoring of corrosion. For common use of this technique in
the field, portable and in situ capillary electrophoresis systems are necessary [37].
Thus, microchip capillary electrophoresis technology has been developed.
Similar to other microfluidic systems, issues related to flow have also been
studied in microchip capillary electrophoresis systems. Electroosmotic flow is an
important property of fluids in microchannels, and methods to measure electroos-
motic flow have been reviewed for both capillary electrophoresis and microchip
capillary electrophoresis [38]. Other methods discussed in the literature to improve
microchip electrophoresis systems include methods to construct contactless con-
ductivity detectors [39], pressure-driven and electrokinetic methods for reliable and
596 Z.P. Çakar and B. Sönmez

reproducible injection of representative samples in microchip electrophoresis sys-

tems [40], and methods for online sample preconcentration of all analytes in the
sample reservoir before they reach the separation channel of the microfluidic
capillary electrophoresis system [41].
Today, microchip capillary electrophoresis systems have several applications.
The importance and clinical application potential of microchip capillary elec-
trophoresis systems has already been foreseen and reviewed about two decades ago
[42]. Rapid, sensitive and reliable detection and analysis of nerve agents by cap-
illary electrophoresis and lab-on-a-chip technology has also been discussed [43]. As
nerve agents are generally used in local conflicts or by local terroristic groups, it is
important to develop such rapid and easy detection techniques as counter-terrorism
applications. Considering the innovative microfluidic technologies that are applied
to capillary electrophoresis, capillary electrophoresis is becoming increasingly
important for green analytical separation and Green Chemistry applications
including food analysis [44], where the use of solvents and reagents can be reduced,
energy consumption is minimized by rapid analysis, and providing miniaturized,
portable equipment [45]. The field of applied microbiology also benefits from
microfluidic capillary electrophoresis applications: adaptation of capillary elec-
trophoresis to microchip format and coupling it with mass spectrometry has been
reported as a powerful approach for rapid and reliable analysis of bacterial outer
membrane proteins. As these proteins are important for bacterial antibiotic sensi-
tivity and pathogenicity, their detection is critical [46].
Other microfluidic developments in the field of electrophoresis are summarized
in Table 24.2. These include two-dimensional (2-D) separation of proteins in
microfluidic devices by a miniaturized 2-D gel electrophoresis system [47];
implementation of isoelectric focusing in microfluidic chips [48]; miniaturization of
free-flow electrophoresis which is a method performing electrophoretic separation
on a continuous analyte stream, as it is flowing through a planar flow channel where
the electric field is applied perpendicularly to the flow [49]; microfluidic elec-
trophoretic mobility shift assays [50]; and microdialysis coupled to microchip
electrophoresis for in vivo and in vitro monitoring of biological events in a
‘separation-based sensor’ mode [51].

Table 24.2 Some examples of microfluidic developments in the field of electrophoresis

Name of the developed system References
Miniaturized 2-D gel electrophoresis system [47]
Isoelectric focusing system in microfluidic chips [48]
Miniaturized free-flow electrophoresis system [49]
Microfluidic electrophoretic mobility shift assay (EMSA) [50]
Microdialysis coupled to microchip electrophoresis [51]
24 Microfluidics and Its Applications in Bionanotechnology 597

24.5 Microfluidic Bioreactors

As a result of the miniaturization of biomedical and biochemical devices for

micro-electromechanical (MEMS) systems, a variety of products have been
developed, including biochips/biosensors, drug delivery systems, tissue scaffolds
and bioreactors. For biochemical and biomedical applications, polymeric materials
have been developed to replace the use of silicon, glass or quartz in MEMS devices,
as these polymers are cheaper, easy to process and biocompatible [52]. A typical
example is a microfluidic polydimethysiloxane (PDMS) bioreactor that was used
for large-scale cultivation of hepatocytes, where ten PDMS layers were stacked
together as four single culture chambers, with high gas permeability of PDMS
allowing sufficient oxygen supply to the microfluidic channels. With these modi-
fications, a culture with higher cell density and sufficient oxygen and nutrient
supply could be obtained, which could make the microfluidic bioreactor suitable for
large-scale cultures [53]. Another example of microfluidic bioreactor systems is
related to regenerative medicine process development: a novel, autoclavable,
microfabricated scale-down device has been developed and inactivated mouse
embryonic fibroblasts along with human embryonic stem cell colonies have been
seeded to grow human embryonic stem cells [54]. In a more recent study, adherent
cell cultures were grown and characterized in an online and automated fashion by
using a microfabricated bioreactor. The developed bioreactor system allowed the
automation of flow control, environmental control, and provided time-course data
of the cell cultures for continuous monitoring of their characteristics [55]. Other
microfluidic bioreactor applications include optimization of production and evalu-
ation of retroviruses and other viral vectors for potential gene therapy applications
[56], and high-throughput cytotoxicity assays allowed by long-term cultivation in
microfluidic bioreactors with continuous perfusion [57].
Long-term cultivation in microfluidic bioreactors is particularly important for
drug discovery and development to test the effects of drugs or toxin exposure. An
early report is on the development of a ‘NanoLiterBioReactor’ (NBR) for long-term
mammalian cell cultures. NBR prototypes have been fabricated with PDMS-based
microfluidic components, and three different cell lines including fibroblasts,
Chinese Hamster Ovary cells and hepatocytes were tested for cultivation, and high
viabilities were achieved [58]. Periodic ‘flow-stop’ perfusion microchannel biore-
actors have also been developed for long-term cultures of mammalian and human
embryonic stem cells. These bioreactors enabled co-culturing of undifferentiated
colonies of human embryonic stem cells on human foreskin fibroblast feeder cells
in microchannels. It was suggested that this has been possible due to two properties
of the novel method: short temporal exposure to flow followed by long static
incubation [59]. Another study is on the development of a microfluidic traps system
that allows prolonged culture of human embryonic stem cell aggregates. The for-
mation of these aggregates is necessary to induce embryonic stem cell differenti-
ation. Thus, the designed traps system facilitates cell capture and aggregation and
598 Z.P. Çakar and B. Sönmez

allows efficient gas/nutrient exchange, suggesting its future use in embryogenesis

research [60].
The use of bioreactors to design three dimensional (3D) artificial cellular envi-
ronments is another major field of research and development. Designing 3D bio-
logical fluidic environments for advanced cell cultures requires the use of
microfluidic devices to tightly regulate fluid direction and behaviour in a geomet-
rically limited environment. These requirements are met by the modern cultivation
devices like bioreactors and multicompartment systems [61]. The major advantage
of 3D culture systems is the fact that they reveal the in vivo conditions much better
than the 2D cultivation systems. Thus, technical systems are used to generate 3D,
organ-like structures [62]. The synthesis of monodisperse water-in-oil microemul-
sions in microfluidic devices is important for precise bead control and high surface
area, and these emulsions are used for the synthesis of highly porous polymer beads
which have a high potential to be used in 3D cell culture applications [63].
A variety of studies and reviews exist in the literature on the improvement of
microbioreactors, their monitoring, and control. An expert panel review article on
monitoring and control of microbioreactors was published by the Working Group
of the European Section of Biochemical Engineering Science (ESBES) in the
European Federation of Biotechnology (EFB), by emphasizing the importance of
developments in microbioreactor monitoring and control on future bioprocess
development and in vitro toxicity testing [64]. Other studies on improvement of
microbioreactors are listed in Table 24.3. These include development of a
fluorescence lifetime-based optical imaging technique in microfluidic bioreactors
that allows oxygen monitoring in continuous cell culture systems [65], integration
of analyte-specific holographic sensors into microfluidic devices for the real-time
pH monitoring in microbioreactors [66], development of a microfluidic bioreactor
for continuous amperometric measurement of glucose [67], development of
energy-efficient nanobubbles flexible in injection and production [68], nanoporous
membrane-sealed microfluidic devices to improve cell viability and avoid perfusion
requirement and related complications [69], and in vitro biosynthesis of metal
nanoparticles in microdroplets [70].

Table 24.3 Selected reports on the improvement of microbioreactors

Improvement area References
Oxygen monitoring [65]
Real-time pH-monitoring [66]
Continuous amperometric glucose measurement [67]
Energy-efficient nanobubble development (flexibility in injection [68]
and production)
Nanoporous membrane-sealing of microbioreactors to improve cell viability and [69]
avoid perfusion and its complications
In vitro biosynthesis of metal nanoparticles in microdroplets [70]
24 Microfluidics and Its Applications in Bionanotechnology 599

It is important to note that, apart from medicine, basic and applied microbio-
logical research also benefits significantly from the developments in microfluidic
bioreactor technology. An early example is a report on the development of a
portable aerobic microbioreactor which allowed direct microscopic monitoring of
the time-integrated growth response of a methanogenic bacterium, Methanosaeta
concilii, and determination of the optimum growth conditions for the tested
microorganism [71]. An integrated microbioreactor with the ability to cultivate
microorganisms in distinct droplets without the need of valves, mixers or pumps
has also been developed using digital microfluidics, for automated cultivation and
analysis of a variety of microorganisms including bacteria, algae and yeast [72].

24.6 Monitoring Microbial Behaviour by Microfluidics

An important contribution of microfluidics to the fields of microbiology and

microbial ecology is the acquired ability to observe microbial behaviour in highly
controlled microenvironments, ranging from single cells to mixed microbial com-
munities. Thus, natural microbial habitats could be mimicked at the micrometer
scale. It has been suggested that this enabling technology could be exploited to
control important microbial processes including antibiotic resistance and biofilm
formation [73]. Some microfluidics application examples to monitor microbial
behaviour are summarized in Table 24.4. Design of a microfluidic biochip by
combining contactless dielectric microsensors with microfluidics for online moni-
toring of fungal biofilm dynamics has already been reported [74]. Additionally,
secondary flow structures resulting due to semi-confined features in a microfluidic
device have also been studied to investigate their effects on bacterial biofilm for-
mation and dynamics [75].
Another important microbial behaviour is ‘chemotaxis’, the ability of some
microorganisms to move towards an attractant and to move away from a repellent
compound. Microfluidic technologies have also been used to study chemotaxis, as
they allow observations at high spatial and temporal resolution in controlled
microenvironments. Chemotaxis of free-swimming bacteria is challenging to
observe by microfluidics. However, recent development of advanced microdevices
generated flow-free and steady gradients. Particularly, the microfluidic gradient

Table 24.4 Microfluidics Type of microbial behaviour Reference number

application examples to monitored (s)
monitor microbial behaviour
Biofilm formation and dynamics [74, 75]
Chemotaxis [76, 77, 79]
Response to drugs/antibiotics [78–80]
Quorum sensing [81]
600 Z.P. Çakar and B. Sönmez

generators are promising to investigate the basic mechanisms of bacterial chemo-

taxis, and the dynamics of chemotaxis in realistic environments [76]. In a previous
study, microfluidics and image analysis of swimming behaviour were combined to
study chemoattraction to dimethylsulfoniopropionate, a compound produced by
phytoplanktons, and the influences on marine planktonic food webs and global
climate change were discussed [77].
Microfluidic technology has also been used to study microbial behaviour under
the influence of drugs. The use of a microchip-based system to study the effects of
the antibiotic, ampicillin, on the bacterium Escherichia coli has been reported [78].
Similarly, a microfluidic device was used to expose the pathogenic bacterium
Vibrio cholerae to an antibiotic and to investigate its behavioural response [79]. In a
more recent study, a 3D microfluidic culture device was used to study pharmaco-
dynamic effects of antibiotics on bacterial cells. For this purpose, bacterial cells
were encapsulated in an agarose-based gel in a commercially available micro-
fluidics chamber, and time-lapse photography was applied to observe bacterial
responses to a linear concentration gradient of antibiotics [80].
The ability of bacteria to sense other bacteria or ‘the crowd’ in their surroundings
is an important phenomenon called ‘quorum sensing’ which influences many
physiological characteristics such as bacterial growth rate and metabolism, antibi-
otic resistance and biofilm formation. In a recent study; a lithographic technique
called micro-3D printing was used to organize bacteria in complex aggregates, to
study the influence of spatial and environmental parameters on social behaviours of
bacteria, such as quorum sensing [81].

24.7 The Use and Potential of Microfluidics

in Microbial Strain Development

The aim of process development in microbial fermentations is to obtain a high

production yield. Development of microbial strains used in fermentations is a
common strategy to obtain high yields and more efficient bioprocesses. Microbial
strain development usually requires application of high-throughput screening and
optimization procedures [82].
Directed evolution is a high-throughput strategy used for microbial strain
development. It is an experimental system mimicking natural evolution in a test
tube at the molecular level, in a directed and rapid fashion. Thus, it is also called
‘laboratory evolution’. Directed evolution has been viewed as an aspect of
nanobiotechnology, which ‘has evolved from protein engineering, via chemical
biology or combinatorial chemistry, to nanobiotechnology’ [83]. Although the term
‘directed evolution’ is more commonly used to describe methods that improve
protein/enzyme properties, it also describes methods to improve properties of whole
24 Microfluidics and Its Applications in Bionanotechnology 601

microbial cells. Those methods are more commonly referred to as ‘evolutionary

engineering’, as discussed previously [84]. The basis of directed evolution methods
is random mutagenesis and high-throughput screening, either at whole cell or at
protein level.
The main focus of our research group is to improve stress resistance of the
industrially important yeast, Saccharomyces cerevisiae. Our ultimate aim is to
obtain robust yeast strains for more efficient yeast bioprocesses, and to understand
the complex molecular mechanisms of resistance to a variety of stress types by
using genomic, transcriptomic and/or proteomic approaches. We have been using
evolutionary engineering strategy to obtain stress-resistant S. cerevisiae strains, by
applying random mutagenesis with the use of a mutagenic agent called ethyl
methane sulfonate, followed by selection under stress conditions. The selection
procedure is applied either as successive batch cultivations in culture tubes or shake
flasks, or as prolonged chemostat cultivations in a lab-scale bioreactor [85–88].
Using evolutionary engineering, we have successfully obtained S. cerevisiae strains
with multistress resistance [86], cobalt resistance [87, 89], and nickel resistance
[88]. Thus, evolutionary engineering is a powerful strategy to obtain yeast strains
with desired properties, as reviewed previously [90]. However, prolonged batch
cultivations in culture tubes or shake flasks cannot provide a good control over
bioprocess conditions and the data are limited to end-point measurements.
Alternatively, the use of long-term chemostat cultivation for selection allows better
bioprocess control, but is more expensive and more labor intensive regarding
sterilization, assembly, sensor calibration, operation, and cleaning of the bioreactor.
Thus, the use of microfluidic bioreactors with online control over cultivation data
and automation, and reduced cost and time would be beneficial for microbial
bioprocess development applications, including strain development by evolutionary
engineering [82, 90].
During our evolutionary engineering studies, we discovered that the final
stress-resistant populations obtained from selection experiments were highly
heterogeneous regarding their resistance levels against the particular stress type
applied during selection. This phenomenon of ‘population heterogeneity’ is
important for evolutionary engineering studies, indicating the multigenic and
complex nature of stress resistance, as discussed previously [90], and it emphasizes
the importance of carefully analysing many individual resistant mutants, to find
those with the highest resistance levels. In fact, cell-to-cell heterogeneity in
microbial bioprocesses can result from a variety of biological and environmental
factors, and is known to have a severe impact on productivity. Thus, the design of
innovative microfluidic systems that can analyse single cells in a microbial culture
also allows detailed analysis of population heterogeneity and bioreactor inhomo-
geneity. Fluorescence-based imaging technologies integrated in microfluidic sys-
tems are generally used to analyse population heterogeneity. It is expected that
microfluidics will become a major single-cell analytical technique in biotechno-
logical processes and strain characterization [91].
602 Z.P. Çakar and B. Sönmez

24.8 Microfluidic Applications in Single Cell Studies

Microfluidics is suitable for single-cell characterization and analysis, because it

allows precise microenvironment control, high-throughput manipulation, and
multi-parameter label-free detection of individual cells [92]. Developments in
microfluidic technology for chemical-content analysis of single cells have been
reviewed previously, by also discussing applications at DNA, RNA and protein
levels [93].
Microfluidics applications for single cell studies are found in diverse research
fields. One of them is the field of drug discovery, development and delivery [94,
95]. Microfluidic drug delivery devices that make use of a concentration gradient
generator for drug delivery at individual cell level are of particular interest [96].
Stem cell research also benefits from microfluidic single-cell applications: human
stem cell differentiation can be regulated by using microfluidic single-cell tech-
nologies [97]. In the field of applied and environmental microbiology, single cell
genomics of bacteria and archaea has been enabled by microfluidic devices [98]. In
a recent review, the use of nanofabrication and microfluidic techniques to study
individual bacteria and multispecies communities was discussed. It was suggested
that the use of these techniques can help gain insight into complex bacterial pro-
cesses like aging, electron transport, quorum sensing, and investigate microbial
communities by single cell genomics [99].

24.9 Conclusions

Microfluidic systems have several advantages that make them suitable for bio-
nanotechnological applications: rapid control, reduced size, low reagent con-
sumption, low costs, simultaneous and independent manipulation of a large number
of cells, large-scale integration and in vivo simulation of cellular microenviron-
ments, etc. The present challenges of microfluidic systems such as slow diffusive
mixing due to laminar flow, difficulties in adapting biological protocols to fit
microfluidic experiments and other issues are being studied to improve microfluidic
systems.
As presented in this chapter, there are diverse microfluidics applications in
bionanotechnology, and microfluidics seems to have a great potential particularly
for future applications related to single cell analysis and characterization. Today,
DNA sequencing is more common and feasible, compared to sequencing of other
molecules, because of the powerful PCR method for easy amplification of the DNA
sample to be sequenced, and the existing high-throughput next generation DNA
sequencing technologies. Development of new and miniaturized technologies for
sequencing and analyzing proteins, carbohydrates and other biologically important
polymers is an important challenge for microfluidics research related to
bionanotechnology.
24 Microfluidics and Its Applications in Bionanotechnology 603

Acknowledgments Our research presented in this chapter was supported by the COST Action
CM0902, Turkish Scientific and Technological Research Council (TÜBİTAK) (project no:
105T314, 107T284, 109T638, PI: ZPÇ), Istanbul Technical University (ITU) Research Funds
(project no: 30108, 34200, PI: ZPÇ). BS is financially supported by the Faculty Member Training
Programme (ÖYP) of the Council of Higher Education (YÖK) in Turkey.

References

1. NanoInk Inc. (2010) Introduction to nanoscale science and technology, Version 1, Chapter 5,
Introduction to Nanobiology (Illinois, USA), pp. 162–209
2. J. Cooper McDonald, D.C. Duffy, J.R. Anderson, D.T. Chiu, H. Wu, O.J.A. Schueller, G.M.
Whitesides, Fabrication of microfluidic systems in poly(dimethylsiloxane). Electrophoresis 21,
27–40 (2000)
3. G. Velve-Casquillas, M. Le Berre, M. Piel, P.T. Tran, Microfluidic tools for cell biological
research. Nano Today 5, 28–47 (2010)
4. C.J. Campbell, B.A. Grzybowski, Microfluidic mixers: from microfabricated to
self-assembling devices. Philos. Trans. A Math. Phys. Eng. Sci. 362, 1069–1086 (2004)
5. E.-S. Kim, E.H. Ahn, E. Chung, D.-H. Kim, Recent advances in nanobiotechnology and
high-throughput molecular techniques for systems biomedicine. Mol. Cells 36, 477–484
(2013)
6. S.J. Maerkl, S.R. Quake, A systems approach to measuring the binding energy landscapes of
transcription factors. Science 315, 233–237 (2007)
7. E.A. Ottesen, J.W. Hong, S.R. Quake, J.R. Leadbetter, Microfluidic digital PCR enables
multigene analysis of individual environmental bacteria. Science 314, 1464–1467 (2006)
8. J. Kim, M. Junkin, D.H. Kim, S. Kwon, Y.S. Shin, P.K. Wong, B.K. Gale, Applications,
techniques, and microfluidic interfacing for nanoscale biosensing. Microfluid. Nanofluid. 7,
149–167 (2009)
9. V. Sanchez-Freire, A.D. Ebert, T. Kalisky, S.R. Quake, J.C. Wu, Microfluidic single-cell
real-time PCR for comparative analysis of gene expression patterns. Nat. Protoc. 7, 829–838
(2012)
10. J.H. Tsui, W. Lee, S.H. Pun, J. Kim, D.H. Kim, Microfluidics-assisted in vitro drug screening
and carrier production. Adv. Drug Deliv. Rev. 65, 1575–1588 (2013)
11. Y. Zhang, Y. Tang, Y.H. Hsieh, C.Y. Hsu, J. Xi, K.J. Lin, X. Jiang, Towards a
high-throughput label-free detection system combining localized-surface plasmon resonance
and microfluidics. Lab Chip 12, 3012–3015 (2012)
12. C.S. Zhang, J.L. Xu, W.L. Ma, W.L. Zheng, PCR microfluidic devices for DNA amplification.
Biotechnol. Adv. 24, 243–284 (2006)
13. G. Munchow, D. Dadic, F. Daffing, S. Hardt, K.S. Drese, Automated chip-based device for
simple and fast nucleic acid amplification. Expert Rev. Mol. Diagn. 5, 613–620 (2005)
14. P.N. Ngatchou, M.R. Holl, C.H. Fisher, M.S. Saini, J.C. Dong, T.T.H. Ren, W.H. Pence, D.L.
Cunningham, S.E. Moody, D.A. Donaldson, D.R. Meldrum, A real-time PCR analyzer
compatible with high-throughput automated processing of 2 µL reactions in glass capillaries.
IEEE Trans. Autom. Sci. Eng. 3, 141–151 (2006)
15. M. Hashimoto, F. Barany, S.A. Soper, Polymerase chain reaction/ligase detection
reaction/hybridization assays using flow-through microfluidic devices for the detection of
low-abundant DNA point mutations. Biosens. Bioelectron. 21, 1915–1923 (2006)
16. K.M. Horsman, J.M. Bienvenue, K.R. Blasier, J.P. Landers, Forensic DNA analysis on
microfluidic devices: a review. J. Forensic Sci. 52, 784–799 (2007)
17. C. Hurth, S.D. Smith, A.R. Nordquist, R. Lenigk, B. Duane, D. Nguyen, A. Surve, A.
J. Hopwood, M.D. Estes, J.N. Yang, Z. Cai, X.J. Chen, J.G. Lee-Edghill, N. Moran, K. Elliott,
604 Z.P. Çakar and B. Sönmez

G. Tully, F. Zenhausern, An automated instrument for human STR identification: design,

characterization, and experimental validation. Electrophoresis 31, 3510–3517 (2010)
18. T. Geng, R.A. Mathies, Minimizing inhibition of PCR-STR typing using digital agarose
droplet microfluidics. Forensic Sci. Int-Genetics 14, 203–209 (2015)
19. N.R. Beer, B.Y. Hindson, E.K. Wheeler, S.B. Hall, K.A. Rose, I.M. Kennedy, B.W. Colston,
On-chip, real-time, single-copy polymerase chain reaction in picoliter droplets. Anal. Chem.
79, 8471–8475 (2007)
20. T.D. Rane, H.C. Zec, C. Puleo, A.P. Lee, T.H. Wang, Droplet microfluidics for
amplification-free genetic detection of single cells. Lab Chip 12, 3341–3347 (2012)
21. D.K. Kang, M.M. Ali, K.X. Zhang, E.J. Pone, W.A. Zhao, Droplet microfluidics for
single-molecule and single-cell analysis in cancer research, diagnosis and therapy. Trac-Trend
Anal. Chem. 58, 145–153 (2014)
22. H. Zec, D.J. Shin, T.H. Wang, Novel droplet platforms for the detection of disease biomarkers.
Expert Rev. Mol. Diagn. 14, 787–801 (2014)
23. L.J. Chien, J.H. Wang, T.M. Hsieh, P.H. Chen, P.J. Chen, D.S. Lee, C.H. Luo, G.B. Lee, A
micro circulating PCR chip using a suction-type membrane for fluidic transport. Med.
Microdevices 11, 359–367 (2009)
24. Z.R. Xu, X. Wang, X.F. Fan, J.H. Wang, An extrusion fluidic driving method for
continuous-flow polymerase chain reaction on a microfluidic chip. Microchim. Acta 168, 71–
78 (2010)
25. G. Czilwik, I. Schwarz, M. Keller, S. Wadle, S. Zehnle, F. von Stetten, D. Mark, R. Zengerle,
N. Paust, Microfluidic vapor-diffusion barrier for pressure reduction in fully closed PCR
modules. Lab Chip 15, 1084–1091 (2015)
26. Y. Luo, B.B. Bhattacharya, T.Y. Ho, K. Chakrabarty, Design and optimization of a
cyberphysical digital-microfluidic biochip for the polymerase chain reaction. IEEE Trans.
Comput. Aid. D 34, 29–42 (2015)
27. C. Situma, M. Hashimoto, S.A. Soper, Merging microfluidics with microarray-based
bioassays. Biomol. Eng. 23, 213–231 (2006)
28. L. Wang, P.C.H. Li, Microfluidic DNA microarray analysis: a review. Anal. Chim. Acta 687,
12–27 (2011)
29. M. Dufva, J. Petersen, L. Poulsen, Increasing the specificity and function of DNA microarrays
by processing arrays at different stringencies. Anal. Bioanal. Chem. 395, 669–677 (2009)
30. Y. Zhang, T.H. Wang, Quantum dot enabled molecular sensing and diagnostics. Theranostics
2, 631–654 (2012)
31. L. Bissonnette, M.G. Bergeron, Next revolution in the molecular theranostics of infectious
diseases: microfabricated systems for personalized medicine. Expert Rev. Mol. Diagn. 6, 433–
450 (2006)
32. D.A. Khodakov, A.V. Ellis, Recent developments in nucleic acid identification using
solid-phase enzymatic assays. Microchim. Acta 181, 1633–1646 (2014)
33. K.A. Gilbride, D.Y. Lee, L.A. Beaudette, Molecular techniques in wastewater: understanding
microbial communities, detecting pathogens, and real-time process control. J. Microbiol.
Meth. 66, 1–20 (2006)
34. J. Mairhofer, K. Roppert, P. Ertl, Microfluidic systems for pathogen sensing: a review. Sensors
9, 4804–4823 (2009)
35. L.M. Borland, S. Kottegoda, K.S. Phillips, N.L. Allbritton, Chemical analysis of single cells.
Annu. Rev. Anal. Chem. 1, 191–227 (2008)
36. E.R. Castro, A. Manz, Present state of microchip electrophoresis: state of the art and routine
applications. J. Chromatogr. A 1382, 66–85 (2015)
37. A.P. Lewis, A. Cranny, N.R. Harris, N.G. Green, J.A. Wharton, R.J.K. Wood, K.R. Stokes,
Review on the development of truly portable and in-situ capillary electrophoresis systems.
Meas. Sci. Technol. 24, 042001 (2013)
38. W. Wang, F. Zhou, L. Zhao, J.R. Zhang, J.J. Zhu, Measurement of electroosmotic flow in
capillary and microchip electrophoresis. J. Chromatogr. A 1170, 1–8 (2007)
24 Microfluidics and Its Applications in Bionanotechnology 605

39. M. Pumera, Contactless conductivity detection for microfluidics: designs and applications.
Talanta 74, 358–364 (2007)
40. J.M. Karlinsey, Sample introduction techniques for microchip electrophoresis: a review. Anal.
Chim. Acta 725, 1–13 (2012)
41. S. Yamamoto, In situ photopolymerization of polyacrylamide-based preconcentrator on a
microfluidic chip for capillary electrophoresis. Yakugaku Zasshi 132, 1031–1035 (2012)
42. C.L. Colyer, T. Tang, N. Chiem, D.J. Harrison, Clinical potential of microchip capillary
electrophoresis systems. Electrophoresis 18, 1733–1741 (1997)
43. M. Pumera, Analysis of nerve agents using capillary electrophoresis and laboratory-on-a-chip
technology. J. Chromatogr. A 1113, 5–13 (2006)
44. M. Koel, M. Borissova, M. Vaher, M. Kaljurand, Developments in the application of Green
Chemistry principles to food analysis: capillary electrophoresis for the analysis of ingredients
in food products. Agro Food Ind Hi Tec 22, 27–29 (2011)
45. M. Kaljurand, M. Koel, Recent advancements on greening analytical separation. Crit. Rev.
Anal. Chem. 41, 2–20 (2011)
46. I. Kustos, B. Kocsis, F. Kilar, Bacterial outer membrane protein analysis by electrophoresis
and microchip technology. Expert Rev. Proteomic 4, 91–106 (2007)
47. H. Chen, Z.H. Fan, Two-dimensional protein separation in microfluidic devices.
Electrophoresis 30, 758–765 (2009)
48. G.J. Sommer, A.V. Hatch, IEF in microfluidic devices. Electrophoresis 30, 742–757 (2009)
49. R.T. Turgeon, M.T. Bowser, Micro free-flow electrophoresis: theory and applications. Anal.
Bioanal. Chem. 394, 187–198 (2009)
50. Y.C. Pan, K. Karns, A.E. Herr, Microfluidic electrophoretic mobility shift assays for
quantitative biochemical analysis. Electrophoresis 35, 2078–2090 (2014)
51. R.A. Saylor, S.M. Lunte, A review of microdialysis coupled to microchip electrophoresis for
monitoring biological events. J. Chromatogr. A 1382, 48–64 (2015)
52. L.J. Lee, BioMEMS and micro-/nano-processing of polymers—an overview. J. Chin. Inst.
Chem. Eng, 34, 25–46 (2003)
53. E. Leclerc, Y. Sakai, T. Fujii, Microfluidic PDMS (polydimethylsiloxane) bioreactor for
large-scale culture of hepatocytes. Biotechnol Progr 20, 750–755 (2004)
54. M. Reichen, R.J. Macown, N. Jaccard, A. Super, L. Ruban, L.D. Griffin, F.S. Veraitch,
N. Szita, Microfabricated modular scale-down device for regenerative medicine process
development. PLoS ONE 7, e52246 (2012)
55. N. Jaccard, R.J. Macown, A. Super, L.D. Griffin, F.S. Veraitch, N. Szita, Automated and
online characterization of adherent cell culture growth in a microfabricated bioreactor. JALA
19, 437–443 (2014)
56. H.L.N. Vu, Y.W. Li, M. Casali, D. Irimia, Z. Megeed, M.L. Yarmush, A microfluidic
bioreactor for increased active retrovirus output. Lab Chip 8, 75–80 (2008)
57. S.T. Yang, X.D. Zhang, Y. Wen, Microbioreactors for high-throughput cytotoxicity assays.
Curr. Opin. Drug Disc. 11, 111–127 (2008)
58. A. Prokop, Z. Prokop, D. Schaffer, E. Kozlov, J. Wikswo, D. Cliffel, F. Baudenbacher,
NanoLiterBioReactor: long-term mammalian cell culture at nanofabricated scale. Biomed.
Microdevices 6, 325–339 (2004)
59. N. Korin, A. Bransky, U. Dinnar, S. Levenberg, Periodic ‘flow-stop’ perfusion microchannel
bioreactors for mammalian and human embryonic stem cell long-term culture. Biomed.
Microdevices 11, 87–94 (2009)
60. M. Khoury, A. Bransky, N. Korin, L.C. Konak, G. Enikolopov, I. Tzchori, S. Levenberg, A
microfluidic traps system supporting prolonged culture of human embryonic stem cells
aggregates. Biomed. Microdevices 12, 1001–1008 (2010)
61. A. Schober, U. Fernekorn, S. Singh, G. Schlingloff, M. Gebinoga, J. Hampl, A. Williamson,
Mimicking the biological world: methods for the 3D structuring of artificial cellular
environments. Eng. Life Sci. 13, 352–367 (2013)
62. A. Schober, U. Fernekorn, B. Lubbers, J. Hampl, F. Weise, G. Schlingloff, M. Gebinoga, M.
Worgull, M. Schneider, C. Augspurger, C. Hildmann, M. Kittler, M. Donahue, Applied nano
606 Z.P. Çakar and B. Sönmez

bio systems with microfluidics and biosensors for three-dimensional cell culture. Materialwiss
Werkst 42, 139–146 (2011)
63. F. Lapierre, N.R. Cameron, J. Oakeshott, T. Peat, Y.G. Zhu, How to fabricate robust
microfluidic systems for a dollar. Micro/Nano Materials, Devices, and Systems, Book Series:
Proceedings of SPIE 3, 89232Y (2013)
64. K.V. Gernaey, F. Baganz, E. Franco-Lara, F. Kensy, U. Kruhne, M. Luebberstedt, U. Marx, E.
Palmqvist, A. Schmid, F. Schubert, C.F. Mandenius, Biotechnol. J. 7, 1308–1314 (2012)
65. D. Sud, G. Mehta, K. Mehta, J. Linderman, S. Takayama, M.A. Mycek, Optical imaging in
microfluidic bioreactors enables oxygen monitoring for continuous cell culture. J. Biomed.
Opt. 11, 050504 (2006)
66. L.L. Bell, A.A. Seshia, C.A.B. Davidson, C.R. Lowe, Integration of holographic sensors into
microfluidics for the real-time pH sensing of L. casei. Eurosensors XXIV Conference, Book
series: Procedia Engineering 5, 1352–1355 (2010)
67. S. Talaei, O. Frey, S. Psoma, P.D. van der Wal, N.F. de Rooij, Smart SU-8 pillars
implemented in a microfluidic bioreactor for continuous measurement of glucose.
Eurosensors XXIV Conference, Book series: Procedia Engineering 5, 448–451 (2010)
68. W.B. Zimmerman, V. Tesar, H.C.H. Bandulasena, Towards energy efficient nanobubble
generation with fluidic oscillation. Curr. Opin. Colloid Interface 16, 350–356 (2011)
69. S.N. Masand, L. Mignone, J.D. Zahn, D.I. Shreiber, Nanoporous membrane-sealed
microfluidic devices for improved cell viability. Biomed. Microdevices 13, 955–961 (2011)
70. K.G. Lee, J. Hong, K.W. Wang, N.S. Heo, D.H. Kim, S.Y. Lee, S.J. Lee, T.J. Park, In vitro
biosynthesis of metal nanoparticles in microdroplets. ACS Nano 6, 6998–7008 (2012)
71. B. Steinhaus, M.L. Garcia, A.Q. Shen, L.T. Angenent, A portable anaerobic microbioreactor
reveals optimum growth conditions for the methanogen Methanosaeta concilii. Appl. Environ.
Microbiol. 73, 1653–1658 (2007)
72. S.H. Au, S.C.C. Shih, A.R. Wheeler, Integrated microbioreactor for culture and analysis of
bacteria, algae and yeast. Biomed. Microdevices 13, 41–50 (2011)
73. R. Rusconi, M. Garren, R. Stocker, Microfluidics expanding the frontiers of microbial
ecology. Ann. Rev. Biophys. 43, 65–91 (2014)
74. L. Richter, C. Stepper, A. Mak, A. Reinthaler, R. Heer, M. Kast, H. Bruckl, P. Ertl,
Development of a microfluidic biochip for online monitoring of fungal biofilm dynamics. Lab
Chip 7, 1723–1731 (2007)
75. A. Kumar, D. Karig, R. Acharya, S. Neethirajan, P.P. Mukherjee, S. Retterer, M.J. Doktycz,
Microscale confinement features can affect biofilm formation. Microfluid. Nanofluid. 14, 895–
902 (2013)
76. T. Ahmed, T.S. Shimizu, R. Stocker, Microfluidics for bacterial chemotaxis. Integr. Biol. 2,
604–629 (2010)
77. J.R. Seymour, R. Simo, T. Ahmed, R. Stocker, Chemoattraction to
dimethylsulfoniopropionate throughout the marine microbial food web. Science 329, 342–
345 (2010)
78. S. Arora, C.S. Lim, J.Y. Foo, M.K. Sakharkar, P. Dixit, A.Q. Liu, J.M. Miao, Microchip
system for monitoring microbial physiological behaviour under drug influences. J. Eng. Med.
223, 777–786 (2009)
79. J.R. Graff, S.R. Forschner-Dancause, S. Menden-Deuer, R.A. Long, D.C. Rowley, Vibrio
cholerae exploits sub-lethal concentrations of a competitor-produced antibiotic to avoid toxic
interactions. Front Microbiol. 4, 8 (2013)
80. Z.N. Hou, Y. An, K. Hjort, K. Hjort, L. Sandegren, Z.G. Wu, Time lapse investigation of
antibiotic susceptibility using a microfluidic linear gradient 3D culture device. Lab Chip 14,
3409–3418 (2014)
81. J.L. Connell, J. Kim, J.B. Shear, A.J. Bard, M. Whiteley, Real-time monitoring of quorum
sensing in 3D-printed bacterial aggregates using scanning electrochemical microscopy. Proc.
Natl. Acad. Sci. USA 111, 18255–18260 (2014)
24 Microfluidics and Its Applications in Bionanotechnology 607

82. H.M. Hegab, A. ElMekawy, T. Stakenborg, Review of microfluidic microbioreactor

technology for high-throughput submerged microbiological cultivation. Biomicrofluidics 7,
021502 (2013)
83. E.S. Kim, Directed evolution: a historical explanation into an evolutionary experimental
system of nanobiotechnology, 1965-2006. Minerva 46, 463–484 (2008)
84. Z.P. Çakar, Metabolic and evolutionary engineering research in Turkey and beyond.
Biotechnol. J. 4, 992–1002 (2009)
85. C. Alkım, B. Turanlı-Yıldız, Z.P. Çakar, Evolutionary engineering of yeast, in Yeast metabolic
engineering: methods and protocols, Method Mol Biol: 1152, ed. by V. Mapelli (Humana
Press, Dordrecht, 2014), pp. 169–183
86. Z.P. Çakar, U.O.S. Seker, C. Tamerler, M. Sonderegger, U. Sauer, Evolutionary engineering
of multiple-stress resistant Saccharomyces cerevisiae. FEMS Yeast Res. 5, 569–578 (2005)
87. Z.P. Çakar, C. Alkim, B. Turanli, N. Tokman, S. Akman, M. Sarikaya, C. Tamerler, L.
Benbadis, J.M. Francois, Isolation of cobalt hyper-resistant mutants of Saccharomyces
cerevisiae by in vivo evolutionary engineering approach. J. Biotechnol. 143, 130–138 (2009)
88. G. Küçükgöze, C. Alkım, Ü. Yılmaz, H.I. Kısakesen, S. Gündüz, S. Akman, Z.P. Çakar,
Evolutionary engineering and transcriptomic analysis of nickel-resistant Saccharomyces
cerevisiae. FEMS Yeast Res. 13, 731–746 (2013)
89. C. Alkim, L. Benbadis, U. Yilmaz, Z.P. Cakar, J.M. François, Mechanisms other than
activation of the iron regulon account for the hyper-resistance to cobalt of a Saccharomyces
cerevisiae strain obtained by evolutionary engineering. Metallomics 5, 1043–1060 (2013)
90. Z.P. Çakar, B. Turanlı-Yıldız, C. Alkım, Ü. Yılmaz, Evolutionary engineering of
Saccharomyces cerevisiae for improved industrially important properties. FEMS Yeast Res.
12, 171–182 (2012)
91. A. Grunberger, W. Wiechert, D. Kohlheyer, Single-cell microfluidics: opportunity for
bioprocess development. Curr. Opin. Biotechnol. 29, 15–23 (2014)
92. W.L. Tang, N. Xiang, D. Huang, X.J. Zhang, X.Z. Gu, Z.H. Ni, Micro-fluidics-based
single-cell biophysical characterization. Prog. Chem. 26, 1050–1064 (2014)
93. R.N. Zare, S. Kim, Microfluidic platforms for single-cell analysis. Annu. Rev. Biomed. Eng.
12, 187–201 (2010)
94. L.F. Kang, B.G. Chung, R. Langer, A. Khademhosseini, Microfluidics for drug discovery and
development: from target selection to product lifecycle management. Drug Discov. Today 13,
1–13 (2008)
95. D. Lombardi, P.S. Dittrich, Advances in microfluidics for drug discovery. Expert Opin. Drug
Discov. 5, 1081–1094 (2010)
96. N.T. Nguyen, S.A.M. Shaegh, N. Kashaninejad, D.T. Phan, Design, fabrication and
characterization of drug delivery systems based on lab-on-a-chip technology. Adv. Drug
Deliver Rev. 65, 1403–1419 (2013)
97. E. Cimetta, G. Vunjak-Novakovic, Microscale technologies for regulating human stem cell
differentiation. Exp. Biol. Med. 239, 1255–1263 (2014)
98. P.C. Blainey, The future is now: single-cell genomics of bacteria and archaea. FEMS
Microbiol. Rev. 37, 407–427 (2013)
99. F.J.H. Hol, C. Dekker, Zooming into see the bigger picture: microfluidic and nanofabrication
tools to study bacteria. Science 346, 1251821 (2014)
Chapter 25
Non-Markovian Dynamics of Qubit
Systems: Quantum-State Diffusion
Equations Versus Master Equations

Yusui Chen and Ting Yu

Abstract In this review we discuss recent progress in the theory of open quantum
systems based on non-Markovian quantum state diffusion and master equations. In
particular, we show that an exact master equation for an open quantum system
consisting of a few qubits can be explicitly constructed by using the corresponding
non-Markovian quantum state diffusion equation. The exact master equation arises
naturally from the quantum decoherence dynamics of qubit systems collectively
interacting with a colored noise. We illustrate our general theoretical formalism by
the explicit construction of a three-qubit system coupled to a non-Markovian
bosonic environment. This exact qubit master equation accurately characterizes the
time evolution of the qubit system in various parameter domains, and paves the way
for investigation of the memory effect of an open quantum system in a
non-Markovian regime without any approximation.

25.1 Introduction

Recent advances in open quantum systems, quantum dissipative dynamics and

quantum information science have attracted enormous interest in examining the
quantum dynamics of open systems in various time domains and coupling strength
ranges. Although the Lindblad master equation is a powerful theoretical tool to
study an open quantum system under the Born-Markov approximation, such a
Markov method will not be valid when the system is strongly coupled to an
environment or the surrounding environment has a structured spectrum. In this case,
it is inevitable to employ a non-Markovian quantum approach. However, unlike in

Y. Chen (&)
T. Yu
Department of Physics and Engineering Physics, Stevens Institute of Technology,
Castle Point on Hudson, Hoboken, NJ 07030, USA
e-mail: [email protected]
T. Yu
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 609

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_25
610 Y. Chen and T. Yu

the case of the standard Markov regimes, deriving the evolution equation that
governs the density operator for a non-Markovian open quantum system is a long
outstanding open problem. The recently developed non-Markovian quantum state
diffusion (QSD) approach [1] offers an alternative way of solving the
non-Markovian open quantum systems. However, from a more fundamental point
of view, particularly in conjunction with the investigation of quantum decoherence
and non-equilibrium quantum transport, a non-Markovian master equation
approach that can be applied to both strong coupling regimes and structured
medium is also highly desirable.
In this paper, we report a systematic theoretical approach that can be imple-
mented easily for realistic quantum systems such as multiple-qubit systems [2]. The
paper is organized as follows. In Sect. 25.2, we describe the principle ideas of
establishing stochastic Schrödinger equations for a generic open quantum system
coupled to a bosonic bath. We further present our recent work on developing a
systematic non-Markovian master equation based on the stochastic non-Markovian
QSD approach in Sect. 25.3. In Sect. 25.4, as examples, we study both two-qubit
and three-qubit systems analytically with our new master equation approach. Some
technical details are left to appendices.

25.2 Non-Markovian Quantum-State Diffusion Approach

The model under consideration is a generic open quantum system linearly coupled
to a zero-temperature bosonic environment. The total Hamiltonian may be written
as (setting

h ¼ 1) [3–5]:

Htot ¼ Hsys þ Hint þ Henv

X y
 X
y ð25:1Þ
¼ Hsys þ gk Lbk þ gk Ly bk þ xk bk bk ;
k k

where Hsys is the Hamiltonian of an arbitrary quantum system of interest, such as

spins, atoms, quantum harmonic oscillators, cavities etc. The operator L is an
arbitrary system operator, describing the coupling between the system of interest
y
and its environment. bk ðbk Þ is the bosonic annihilation (creation) operator of kth
mode in the environment, satisfying the usual commutation relations for bosons,
y y y
½bk ; bk0  ¼ dk;k0 and ½bk ; bk0  ¼ ½bk ; bk0  ¼ 0.
In the interaction picture with respect to the environment, the total Hamiltonian
can be rewritten as (the rest of this paper is discussed in the interaction picture),
X y

Htot ¼ Hsys þ gk Lbk eixk t þ gk Ly bk eixk t : ð25:2Þ
k
25 Non-Markovian Dynamics of Qubit Systems … 611

With the total Hamiltonian, the evolution for the state of the total system jwtot ðtÞi
is governed by the standard Schrödinger equation,
" #
X y ixk t

 y ixk t
@t jwtot ðtÞi ¼ i Hsys þ gk Lbk e þ gk L bk e jwtot ðtÞi: ð25:3Þ
k

For a real-world problem, solving above Schrödinger equation in a

non-Markovian regime is by no means an easy task due to the complexity arising
from a large number of environmental variables and strong system-environment
coupling. Therefore, it is desirable to develop a dynamical approach for dealing
with a reduced density operator describing open quantum systems only. The
quantum-state diffusion approach was developed based on a special choice of
environmental basis consisting of a set of Bargmann coherent states
jzi ¼ jz1 i  jz2 i 


 jzk i 


. For each mode, the Bargmann state is defined
as

y
jzk i ¼ ezk bk j0i;

satisfying the following properties,

bk jzk i ¼ zk jzk i;
y @
bk jzk i ¼ jzk i:
@zk

It should be noted that the Bargmann states completeness identity is given by,
Z
d 2 z jzj2
I¼ e jzihzj;
p

where d 2 z ¼ d 2 z1 d 2 z2


. Then the state jwtot ðtÞi for the combined total system can
be expanded as,
Z
d 2 z jzj2
jwtot ðtÞi ¼ e jzihzjwtot ðtÞi
p
Z 2
d z jzj2
¼ e jwt ðz Þi  jzi; ð25:4Þ
p

where

jwt ðz Þi ¼ hzjwtot i:

Note that jwt ðz Þi is a pure state in the system’s Hilbert space, containing the
complex variables z that will be interpreted as complex Gaussian random vari-
ables. For reasons to be explained later, jwt ðz Þi is called a quantum trajectory
612 Y. Chen and T. Yu

[1, 6]. Remarkably, the reduced density operator qt at time point t for the system of
interest can be recovered by the quantum pure state as shown below. By definition,
the reduced density operator qt may be obtained from qtot by performing the partial
trace over the environmental variables. For this purpose, we choose Bargmann
coherent states as our basis,

qt ¼ trenv ðqtot Þ
Z 2
d z jzj2
¼ e hzjwtot ihwtot jzi
p
Z 2
d z jzj2
¼ e jwt ðz Þihwt ðzÞj
p
¼ Mðjwt ðz Þihwt ðzÞjÞ; ð25:5Þ

where the symbol

Z
d 2 z jzj2
Mð
Þ ¼ e ð
Þ ð25:6Þ
p

stands for the statistical average over the random variables z [1, 6, 7].
From (25.3), one can derive a stochastic differential equation for a quantum
trajectory when the environmental bath is in a vacuum state [1],
" #
X y ixk t


@t jwt ðz Þi ¼ ihzj Hsys þ gk Lbk e þ h:c: jwtot ðtÞi
k
" #
X @ ixk t
¼ iHsys þ Lzt  iLy gk e jwt ðz Þi; ð25:7Þ
k
@zk

where
X
zt ¼ i gk zk eixk t ð25:8Þ
k

is a complex Gaussian process.

In a more general situation where the environment is in a thermal equilibrium
state
P y
eb k
xk bk bk
qenv ð0Þ ¼ ;
Z
P y
where b ¼ kB1T and Z is the partition function Z ¼ trðeb k
x k bk bk
Þ, the bath cor-
relation function can be written in the following form
25 Non-Markovian Dynamics of Qubit Systems … 613

X  
2 xk
aðt; sÞ ¼ jgk j coth cos xk ðt  sÞ  i sin xk ðt  sÞ :
k
2kB T

For the zero temperature case, the correlation function reduces to

X
aðt; sÞjT¼0 ¼ jgk j2 eixk ðtsÞ : ð25:9Þ
k

It is interesting to note that the stochastic process defined in (25.8) satisfies,

Mðzt Þ ¼ 0;
Mðzt zs Þ ¼ 0; ð25:10Þ
Mðzt zs Þ ¼ aðt; sÞ:

Equation (25.10) shows that zt typically represents a non-Markovian Gaussian

process characterised by the correlation aðt; sÞ. Taking the Lorenz spectrum as an
example,

C 1
JðxÞ ¼ ;
2p ðx  xs þ Xc Þ2 þ c2

we can explicitly show that the correlation function takes a very simple form,

Cc ðc þ iXc Þjtsj

aðt; sÞ ¼ e ; ð25:11Þ
2

which is commonly called the Ornstein-Uhlenbeck type correlation function. Xc

represents the central frequency of the environment and 1c is the correlation-time of
the environment. When the parameter c ! 1, the Ornstein-Uhlenbeck correlation
function recovers the well-known Markov approximation described by a Dirac delta
function,

aðt; sÞ  Cdðt; sÞ:

In (25.7), the term @z@ jwt ðz Þi can be cast as a functional derivative by using the
k
chain rule,

X X Zt
@ @zs d
i gk Ly eixk t  jwt ðz Þi ¼ i gk Ly eixk t ds jw ðz Þi
k
@zk k
@zk dzs t
0
Zt
d
¼ Ly dsaðt; sÞ jw ðz Þi:
dzs t
0
614 Y. Chen and T. Yu

By defining the O operator,

d
Oðt; s; z Þjwt ðz Þi ¼ jw ðz Þi; ð25:12Þ
dzs t

the non-Markovian quantum-state diffusion (QSD) equation driven by the complex

Gaussian process zt is written as,
 
@t jwt ðz Þi ¼ iHsys þ Lzt  Ly Oðt;

z Þ jwt ðz Þi; ð25:13Þ

R

z Þ ¼ t dsaðt; sÞOðt; s; z Þ.
where Oðt; 0
The exact non-Markovian QSD equations are generic for open quantum system
models represented by (25.1). Note that these non-Markovian stochastic equations
are derived from the generic microscopic Hamiltonian (25.1) or (25.2) without any
approximation. For practical numerical simulations, it is useful to recast the QSD
equation into a time convolutionless form by introducing a time-local operator O.
The dynamical equation of the O operator can be determined by its consistency
condition,

@ d d @

jwt ðz Þi   jwt ðz Þi:
@t dzs dzs @t

Putting the definition of O operator (25.12) and the QSD (25.13) into above
equation, the dynamical equation of O operator is given by,

z Þ; Oðt; s; z Þ  Ly dO :
@t Oðt; s; z Þ ¼ ½iHsys þ Lzt  Ly Oðt; ð25:14Þ
dzs

with the initial condition

Oðt; s ¼ t; z Þ ¼ L: ð25:15Þ

For many interesting models, such as dephasing models [8], multiple-qubit

dissipative systems [9–12], and quantum Brownian motion [13], the exact
non-Markovian QSD equations have been established [5, 14–17]. Consequently,
one can study the non-Markovian behaviors of quantum decoherence and quantum
entanglement, based on the numerically recovered reduced density operator qt .
However, from a more fundamental point of view, it is known that the corre-
sponding non-Markovian master equations are very useful in describing quantum
dissipative dynamics, quantum transport processes, and quantum decoherence.
Therefore, it is of great interest to establish a generic relation between the stochastic
QSD equations and their master equation counterparts.
25 Non-Markovian Dynamics of Qubit Systems … 615

25.3 Non-Markovian Master Equation Approach

After discussing the non-Markovian QSD approach, we will study the relationship
between the non-Markovian QSD and master equation approaches in this section.
As a fundamental tool, the master equation governs the evolution of the reduced
density operator for an open quantum system. However, deriving a systematic
non-Markovian master equation for a generic open quantum system is a rather
difficult problem. Up to now, exact master equations are available only for some
specific models, such as the dephasing model, qubit dissipative model, and
Brownian motion model [5, 7, 13–26]. Traditionally in quantum optics, in the case
of weak coupling and broadband approximation, one can adequately describe the
dynamics of atoms coupled to a quantized radiation field by a Lindblad master
equation [27],

C y
@t qt ¼ ½iHsys ; qt   ðL Lqt þ qt Ly L  2Lqt Ly Þ; ð25:16Þ
2

where qt is the reduce density operator of the system of interest, L is the Lindblad
operator and C represents a decay rate. However, when the Born-Markov
approximation ceases to be valid as shown in many cases involving strong cou-
plings and structured spectrum distributions, non-Markovian dynamics has to be
invoked. It is shown that the non-Markovian dynamics can bring new interesting
physical phenomena, such as a regeneration of quantum entanglement, slow
quantum coherence decay and so on. In this section, we show a systematic way of
deriving the non-Markovian master equations from stochastic QSD equations.
As shown in (25.5), the reduced density matrix qt can be formally recovered by
taking the statistical average over all the quantum trajectories,

qt ¼ M½jwt ðz Þihwt ðzÞj:

From this starting point, we can write down the formal master equation as,

@t qt ¼ ½iHsys ; qt  þ LM½zt Pt   Ly M½Oðt;
y ðt; zÞL;

z ÞPt  þ M½zt Pt Ly  M½Pt O
ð25:17Þ

where Pt is the stochastic projection operator Pt ðz; z Þ ¼ jwt ðz Þihwt ðzÞj.
By applying the Novikov’s theorem [8],

Zt
dPt
M½zt Pt  ¼ dsM½zt zs M½ ;
dzs
0
616 Y. Chen and T. Yu

it is easy to obtain the following results,

y ;
M½zt Pt  ¼ M½Pt O
ð25:18Þ

t :
M½zt Pt  ¼ M½OP

The detailed proof of the above results can be found in the Appendix 1.
Therefore, the formal master equations can be written as

@t qt ¼ ½iHsys ; qt  þ ½L; RðtÞ  ½Ly ; Ry ðtÞ: ð25:19Þ

where,

y Þ:
RðtÞ ¼ MðPt O

As a note, we point out that non-Markovian master equations may provide a

possibility to find an exact analytical solution. Even in numerical simulations, in
some cases, such as small quantum systems, a master equation can significantly
reduce computational complexity. Generally, the O operator contains noise terms,

t Þ is still hard to derive analytically.
therefore, the term MðOP
Example Here we consider the one qubit dissipative model as an example to show
how to use Novikov’s theorem to derive an exact master equation. The total
Hamiltonian in this case is given by [1, 8],

x X y X
Htot ¼ rz þ r  gk bk eixk t þ r þ gk bk eixk t :
2 k k

Then, the non-Markovian QSD (25.13) can be explicitly written as,

x
t ðz Þi:
@t jwt ðz Þi ¼ ði rz þ r zt  r þ OÞjw ð25:20Þ
2

And the O operator takes the form of

Oðt; sÞ ¼ f ðt; sÞr ; ð25:21Þ

where the coefficient function f ðt; sÞ satisfies the initial condition f ðt; tÞ ¼ 1 and it
obeys the equation of motion,

@t f ðt; sÞ ¼ ixf þ Ff ;
Zt
FðtÞ ¼ dsaðt; sÞf ðt; sÞ:
0
25 Non-Markovian Dynamics of Qubit Systems … 617

Here we choose the Ornstein-Uhlenbeck type correlation function (25.11) as an

example, such that the coefficient function FðtÞ satisfies

d Cc
FðtÞ ¼  cF þ ixF þ F 2 ;
dt 2
Fð0Þ ¼ 0:

Using Novikov’s theorem (25.18), we have

t Þ ¼ FðtÞr qt ;
MðOP

y Þ ¼ F  ðtÞqt r þ :
MðPt O

Then the exact master equation can be shown explicitly as

x
@t qt ¼ ½i rz ; qt   ðFr þ r qt þ F  qt r þ r  ðF þ F  Þr qt r þ Þ: ð25:22Þ
2

Next, we check its Markov limit: writing the correlation function in the form

aðt; sÞ ¼ Cdðt; sÞ; ð25:23Þ

then FðtÞ can be calculated as

Zt
C
FðtÞ ¼ dsCdðt; sÞf ðt; sÞ ¼ :
2
0

The master equation in the Markov limit is easily obtained from (25.22),
h x i C
@t qt ¼ i rz ; qt  ðr þ r qt þ qt r þ r  2r qt r þ Þ; ð25:24Þ
2 2

which clearly takes the standard Lindblad form.

25.4 Multiple-Qubit Systems

In this section, we discuss a multiple-qubit system coupled to a common bosonic

environment. The multiple-qubit model is of interest in quantum information as it
represents a quantum memory realised by two-level systems such as spins or atoms
[28–34]. Studies of dissipation and decoherence for multiple qubit systems are
useful to understand quantum decoherence control and quantum disentanglement
processes. Such studies can help us to develop new theoretical and experimental
618 Y. Chen and T. Yu

strategies to control quantum decoherence [35–37]. Here, we consider a generic

N-qubit model,
X y X
Htot ¼ Hsys þ L gk bk eixk t þ Ly gk bk eixk t ;
k k
X xj X 
Hsys ¼ rzj þ Jxy rxj rxj þ 1 þ ryj ryj þ 1 ;
j
2 j

P
where L ¼ j jj rj is the dissipative coupling operator of the system, jj is the
coupling constant for jth qubit. The non-Markovian QSD equation is written as
 
@t jwt ðz Þi ¼ iHsys þ Lzt  Ly O

jwt ðz Þi; ð25:25Þ

where O operator is determined by the following equation,

dOðtÞ
@t Oðt; s; z Þ ¼ ½iHsys þ Lzt  Ly OðtÞ;

Oðt; sÞ  Ly ; ð25:26Þ
dzs
P
together with the initial condition Oðt; s ¼ tÞ ¼ j jj rj .
Differing from the previous simple example, O operator is no longer free of
noise when the size of the system increases. In general, the O operator is typically
involved with noise z . Note that O operator can be formally written in the func-
tional expansion of noise [8],

Zt Zt Zt
Oðt; s; z Þ ¼ O0 ðt; sÞ þ ds1 zs1 O1 ðt; s; s1 Þ þ ds1 ds2 zs1 zs2 O2 ðt; s; s1 ; s2 Þ þ


;
0 0 0
ð25:27Þ

where O0 is the zeroth order, which does not contain noise z ; also, operators On by
definition do not contain noise. For a simple example, the one qubit case, O ¼
f ðt; sÞr is a special case in which O operator only contains the O0 term. The initial
conditions for each term of the O operator are [13],

O0 ðt; s ¼ tÞ ¼ L;
On ðt; s ¼ tÞ ¼ 0:

Substituting (25.27) into (25.26), we have a set of coupled differential equations

for each term On in the O operator (Appendix 2),
25 Non-Markovian Dynamics of Qubit Systems … 619

@t O0 ðt; sÞ ¼ ½iHsys  Ly O
0 ðtÞ; O0 ðt; sÞ  Ly O
1 ðt; sÞ;

@t O1 ðt; s; s1 Þ ¼ ½iHsys  Ly O
0 ðtÞ; O1 ðt; s; s1 Þ  ½Ly O
1 ðt; s1 Þ; O0 ðt; sÞ

ð25:28Þ
 Ly ðO

ðt; s ; sÞ þ O
2 1

ðt; s; s ÞÞ;
2 1
etc:;

together with the boundary conditions

O1 ðt; s; tÞ ¼ ½L; O0 ðt; sÞ;

O2 ðt; s; t; s1 Þ þ O2 ðt; s; s1 ; tÞ ¼ ½L; O1 ðt; s; s1 Þ;
etc:

As we have shown in (25.19), explicitly finding RðtÞ is the key to determine the
exact master equation. In the next section, we will exhibit the detail of deriving RðtÞ
for some important qubit systems.

25.4.1 Two-Qubit Systems

For simplicity, we take the two-qubit system as our first example to show the details
of our analytical derivation. The two-qubit system has generated enormous interest
due to its relevance in quantum computing and quantum information. For example,
the entanglement measure for a qubit system takes a particular simple form for the
two-qubit system known as concurrence [38]. The Hamiltonian for the two-qubit
model is given by,
X y X
Htot ¼ Hsys þ L gk bk eixk t þ Ly gk bk eixk t ;
k k
x1 1 x2 2  
Hsys ¼ rz þ rz þ Jxy r1x r2x þ r1y r2y ;
2 2
L ¼ j1 r1 þ j2 r2 :

As discussed above, the non-Markovian QSD equation is given by,

 
@t jwt ðz Þi ¼ iHsys þ Lzt  Ly O

jwt ðz Þi; ð25:29Þ

where the O operator can be written as

Zt
Oðt; s; z Þ ¼ O0 ðt; sÞ þ ds1 zs1 O1 ðt; s; s1 Þ; ð25:30Þ
0
620 Y. Chen and T. Yu

where,

O0 ðt; sÞ ¼ f1 ðt; sÞr1 þ f2 ðt; sÞr2 þ f3 ðt; sÞr1z r2 þ f4 ðt; sÞr1 r2z ; ð25:31Þ

O1 ðt; s; s1 Þ ¼ f5 ðt; s; s1 Þr1 r2 : ð25:32Þ

Inserting the explicit form of O operator into the equation of motion (25.26),

Zt
@t O0 ¼ ½iHsys  Ly O

0 ; O0   Ly dsaðt; sÞf5 ðt; s; sÞr1 r2 ; ð25:33Þ
0

@t O1 ¼ ½iHsys ; O1   ½Ly O

0 ; O1   ½Ly O

1 ; O0 ; ð25:34Þ

we have the evolution equations for the coefficient functions as

@t f1 ¼ ix1 f1  2iJxy f3 þ ðj1 F1 þ j2 F3 Þf1 þ j2 ðF4  F1 Þf3

j2 ð25:35Þ
þ ðj1 F4 þ j2 F3 Þf4  F5 ;
2

@t f2 ¼ ix2 f2  2iJxy f4 þ ðj1 F4 þ j2 F2 Þf2 þ ðj1 F4 þ j2 F3 Þf3

j1 ð25:36Þ
þ j1 ðF3  F2 Þf4  F5 ;
2

@t f3 ¼ ix2 f3  2iJxy f1 þ j1 ðF3  F2 Þf1 þ ðj1 F4 þ j2 F3 Þf2

j1 ð25:37Þ
þ ðj1 F4 þ j2 F2 Þf3  F5 ;
2

@t f4 ¼ ix1 f4  2iJxy f2 þ ðj1 F4 þ j2 F3 Þf1 þ j2 ðF4  F1 Þf2

j2 ð25:38Þ
þ ðj1 F1 þ j2 F3 Þf4  F5 ;
2

@t f5 ¼ iðx1 þ x2 Þf5 þ ðj1 F1 þ j1 F4 þ j2 F2 þ j2 F3 Þf5

ð25:39Þ
þ ðj1 f1  j1 f4 þ j2 f2  j2 f3 ÞF5 ;
Rt Rt
where Fj ðtÞ ¼ 0 dsaðt; sÞfj ðt; sÞ (j ¼ 1; 2; 3; 4) and F5 ðt; sÞ ¼ 0 dsaðt; sÞf5 ðt; s; sÞ.
Based on the previous discussion, we have the initial conditions as

f1 ðt; tÞ ¼ j1 ; f2 ðt; tÞ ¼ j2 ; ð25:40Þ

f3 ðt; tÞ ¼ 0; f4 ðt; tÞ ¼ 0; ð25:41Þ

f5 ðt; t; s1 Þ ¼ 0; ð25:42Þ
25 Non-Markovian Dynamics of Qubit Systems … 621

and the boundary condition,

f5 ðt; s; tÞ ¼ 2ðj1 f3 ðt; sÞ þ j2 f4 ðt; sÞÞ: ð25:43Þ

y Þ can be evaluated explicitly. By the

For this two-qubit model, RðtÞ ¼ MðPt O
ansatz of O operator,

yÞ
RðtÞ ¼ MðPt O
Zt

y þ Pt
¼ MðPt O
y ðt; s1 ÞÞ:
ds1 zs1 O
0 1
0

Since both O0 and O1 are free of noise, therefore, we have,

Zt

y þ
RðtÞ ¼ qt O
y ðt; s1 Þ:
ds1 Mðzs1 Pt ÞO ð25:44Þ
0 1
0

Applying Novikov’s theorem (25.18), we obtain,

Zt
Mðzs1 Pt Þ ¼ ds2 aðs1 ; s2 ÞMðOðt; s2 ÞPt Þ
0
2 3
Zt Zt
¼ ds2 aðs1 ; s2 Þ4O0 ðt; s2 Þqt þ ds3 O1 ðt; s2 ; s3 ÞMðzs3 Pt Þ5:
0 0

Repeatedly applying Novikov’s theorem, we get,

Zt
Mðzs3 Pt Þ ¼ ds4 aðs3 ; s4 ÞMðPt Oy ðt; s4 ÞÞ
0
2 3
Zt Zt
y y
¼ ds4 aðs3 ; s4 Þ4MðPt O0 ðt; s4 ÞÞ þ ds5 MðPt zs5 ÞO1 ðt; s4 ; s5 Þ5:
0 0

In general, repeating the Novikov theorem may generate an infinite number of

terms. However, as shown below, for our two-qubit model, we can get a closed
equation in a finite number of steps. Note that, if we put all the results back into
RðtÞ, we have
622 Y. Chen and T. Yu

Zt Zt

y þ
RðtÞ ¼ qt O ds1
y ðt; s1 Þ
ds2 aðs1 ; s2 ÞO0 ðt; s2 Þqt O
0 1
0 0
Zt Zt Zt
þ ds1 ds2 aðs1 ; s2 Þ
y ðt; s1 Þ:
ds3 O1 ðt; s2 ; s3 ÞMðzs3 Pt ÞO ð25:45Þ
1
0 0 0

It is easy to check that,

y
Mðzs3 Pt ÞO1 ðt; s; s1 Þ ¼ 0;

since

y y
O0 O1 ¼ 0;
y y
O1 O1 ¼ 0:

The two identities are called “forbidden conditions” [2], which result in a closed
noise-free RðtÞ operator,

Zt Zt

y þ
RðtÞ ¼ qt O ds1
y ðt; s1 Þ:
ds2 aðs1 ; s2 ÞO0 ðt; s2 Þqt O ð25:46Þ
0 1
0 0

Finally, we determine the exact non-Markovian master equation for the

two-qubit system in a bosonic environment. Here, we explicitly exhibit RðtÞ with
coefficient functions:

Ry ðtÞ ¼ ðF1 r1 þ F2 r2 þ F3 r1z r2 þ F4 r1 r2z Þqt

  ð25:47Þ
þ r1 r2 qt r1 ðtÞr1þ þ r2 ðtÞr2þ þ r3 ðtÞr1z r2þ þ r4 ðtÞr1þ r2z ;
Rt Rt
where rj ðtÞ ¼ 0 ds1 0 ds2 aðs1 ; s2 Þfj ðt; s2 ÞF5 ðt; s1 Þ, ðj ¼ 1; 2; 3; 4Þ.
In Fig. 25.1, we show the dynamics of quantum entanglement in the two-qubit
system. For calculational simplicity, we choose the Ornstein-Uhlenbeck type of
correlation function (25.11) in our numerical simulation. Figure 25.1a shows a few
single-trajectory paths, numerically simulated by the non-Markovian QSD equa-
tion. In Fig. 25.1b, we use 100-trajectory averaged (dash-dotted curve) and
1000-trajectory (dashed curve) averaged reduced density operators qt to simulate
the entanglement dynamics. Also we show the result simulated by using the
non-Markovian master equation (solid line). The non-Markovian dynamics for
1000 quantum trajectories shows a high degree of agreement with the master
equation approach.
25 Non-Markovian Dynamics of Qubit Systems … 623

(a) 1 (b) 1

Concurrence

Concurrence
0.6 0.6

100 Trajectory
1000 Trajectory
Master Equation

0.2 0.2
0 2 4 6 8 10 0 2 4 6 8 10
ωt ωt

Fig. 25.1 Quantum entanglement in two-qubit system, initially prepared in a Bell state
p1ffiffi ðj10i þ j01iÞ. We show the results: a a set of single-trajectory evolution (dashed), and b 100
2
trajectories averaged (dash-dotted), 1000 trajectories averaged (dashed) and master equation
(solid). The parameters are set as: x1 ¼ x2 ¼ x, j1 ¼ j2 ¼ 1, Jxy ¼ 0 and c ¼ 0:1

25.4.2 Three-Qubit Systems

As another interesting example, in this section, we extend our derivation for the
two-qubit system to the case of a three-qubit model. With the derived
non-Markovian master equation, we study quantum decoherence and quantum
disentanglement in a multiple-qubit system. Although there is no convenient
computable measure of entanglement for multipartite systems, we can still inves-
tigate the entanglement transfer between two qubits in a multiple-qubit system. The
total Hamiltonian for the three-qubit system (shown in Fig. 25.2) is,
X y X
Htot ¼ Hsys þ L gk bk eixk t þ Ly gk bk eixk t ;
k k
X
3 2 
X 
xj j
Hsys ¼ r þ Jxy rxj rxj þ 1 þ ryj ryj þ 1 ;
j¼1
2 z j¼1

P
where L ¼ 3j¼1 jj rj is the Lindblad operator coupling the system to the envi-
ronment. The non-Markovian QSD equation in this case is given by,
 
@t jwt ðz Þi ¼ iHsys þ Lzt  Ly O

jwt ðz Þi: ð25:48Þ

For the three-qubit dissipative model, the O operator contains up to the

second-order of noise and can be written in a functional expansion as
624 Y. Chen and T. Yu

Fig. 25.2 Schematic of the

3-qubit system coupled to a
common environment

Zt Zt Zt

Oðt; s; z Þ ¼ O0 ðt; sÞ þ ds1 zs1 O1 ðt; s; s1 Þ þ ds1 ds2 zs1 zs2 O2 ðt; s; s1 ; s2 Þ:
0 0 0

Here, we do not explicitly show the form of O operators. However, we still have
the boundary conditions and initial conditions from the O operator evolution
equation. The evolution equations for O0 , O1 and O2 are

@t O0 ðt; sÞ ¼ ½iHsys  Ly O

0 ; O0   Ly O

1 ðt; sÞ; ð25:49Þ

@t O1 ðt; s; s1 Þ ¼ ½iHsys ; O1   ½Ly O
0 ; O1   ½Ly O
1 ; O0 

ð25:50Þ
 L y ðO

2 ðt; s; s1 Þ þ O

2 ðt; s1 ; sÞÞ;

@t O2 ðt; s; s1 ; s2 Þ ¼ ½iHsys ; O2   ½Ly O
0 ; O2   ½Ly O

2 ; O0 

 ½Ly O

1 ðt; s1 Þ; O1 ðt; s; s2 Þ  ½Ly O

1 ðt; s2 Þ; O1 ðt; s; s1 Þ:
ð25:51Þ

The boundary conditions are

O1 ðt; s; tÞ ¼ ½L; O0 ðt; sÞ;

O2 ðt; s; t; s1 Þ þ O2 ðt; s; s1 ; tÞ ¼ ½L; O1 ðt; s; s1 Þ:

The initial conditions are

O0 ðt; s ¼ tÞ ¼ L;
O1 ðt; s ¼ t; s1 Þ ¼ 0;
O2 ðt; s ¼ t; s1 ; s2 Þ ¼ 0:
25 Non-Markovian Dynamics of Qubit Systems … 625

And the “forbidden conditions” are

O0 O2 ¼ 0;
O1 O2 ¼ 0;
O2 O2 ¼ 0: ð25:52Þ

Equations (25.49–25.51), together with their initial conditions fully determine

the O operator. In order to derive the exact master equation, the last step is to

y  in the form
evaluate RðtÞ ¼ M½Pt O

Zt Zt Zt

y þ
RðtÞ ¼ qt O
y ðt; s1 Þ þ
ds1 Mðzs1 Pt ÞO ds1
y ðt; s1 ; s3 Þ:
ds3 Mðzs1 zs3 Pt ÞO
0 1 2
0 0 0
ð25:53Þ

Similar to the two-qubit case, employment of the Novikov theorem (25.18) and
the forbidden conditions (25.52) leads to RðtÞ of (25.53) in the form

Zt Zt

y þ
R ¼ qt O ds1
y ðt; s1 Þ
ds2 aðs1 ; s2 ÞO0 ðt; s2 Þqt O
0 1
0 0
Zt Zt Zt Zt
y
y ðt; s1 Þ
þ ds1 ds2 ds3 ds4 aðs1 ; s2 Þaðs3 ; s4 ÞO1 ðt; s2 ; s3 Þqt O0 ðt; s4 ÞO 1
0 0 0 0
Zt Zt Zt Zt
þ ds1 ds2 ds3
y ðt; s1 ; s2 Þ
ds4 aðs1 ; s3 Þaðs2 ; s4 ÞO0 ðt; s3 ÞO0 ðt; s4 Þqt O 2
0 0 0 0
Zt Zt Zt Zt
þ ds1 ds2 ds3
y ðt; s1 ; s2 Þ:
ds4 aðs1 ; s3 Þaðs2 ; s4 ÞO1 ðt; s3 ; s4 Þqt O 2
0 0 0 0
ð25:54Þ

The detailed derivation of this can be found in Appendix 3. With the exact form
of RðtÞ, the exact non-Markovian master equation may be explicitly obtained,

@t qt ¼ ½iHsys ; qt  þ ½L; R  ½Ly ; Ry : ð25:55Þ

It should be noted that in the above derivation, the correlation function aðt; sÞ
can have an arbitrary form. Therefore our derivation of the exact master equation is
completely general.
In Fig. 25.3, we plot the entanglement dynamics of a pair of qubits in the 3-qubit
model with four different initial states, including a separate state and three
626 Y. Chen and T. Yu

(a) (b) 0.06

0.4
γ =0.4
Concurrence

Concurrence
γ =0.9
γ =1.5 0.04
0.2
0.02

0 0
0 10 20 30 0 10 20 30
ωt ωt
(c) (d)
1 1
Concurrence

Concurrence
0.5 0.5

0 0
0 10 20 30 0 10 20 30
ωt ωt

Fig. 25.3 The dynamics of entanglement between qubit 1 and qubit 2 (see Fig. 25.2) with
pffiffiffi pffiffiffi
different initial states. a j111i, b ðj111i þ j000iÞ= 2, c ðj100i þ j010i þ j001iÞ= 3,
pffiffiffi
d ðj110i þ j101i þ j011iÞ= 3

maximally entangled states (GHZ state and W state). Without loss of generality, the
concurrence between qubit 1 and qubit 2 is studied. In Fig. 25.3a, b, since the initial
three-qubit states are j111i and p1ffiffi2 ðj111i þ j000iÞ respectively, there is no entan-
glement between the qubit-pair considered. When we choose different memory
times 1=c (taking Ornstein-Uhlenbeck noise as an example again (25.11)), the
degrees of the generated quantum entanglement are different. When c ¼ 0:4, a
typical non-Markovian regime, the maximally generated entanglement is much
higher than that in the case with c ¼ 1:5 representing the Markov limit. In
Fig. 25.3c, d, the initial GHZ state of the three-qubit system is maximally entan-
gled, and the reduced density matrices for qubits 1 and 2 are also entangled. When
c ¼ 0:4, the early revival of entanglement in both cases is a typical non-Markovian
feature.
Furthermore, we consider the entanglement transfer between two pairs of qubits.
In Fig. 25.4, we prepare a Bell state for the qubit-pair 1 and 2. The idea is to observe
the way entanglement transfers from qubits 1 and 2 to qubits 2 and 3. Because of
the symmetry of the model, the behaviors of quantum entanglements C13 and C23
are identical. In Fig. 25.4a, c, the correlation parameter c ¼ 0:4 is fixed, therefore
these two graphs show the short-time behavior of non-Markovian entanglement
evolution. For different initial states, the speed of generating quantum entanglement
is also different. In Fig. 25.4b, d, with the environment close to the Markov limit
with c ¼ 1:5, we see that the entanglement drops to its final steady state quickly, as
expected. It is interesting to note that the quantum entanglement between a pair of
qubits does not actually vanish for a long time. Contrary to the two-qubit system
25 Non-Markovian Dynamics of Qubit Systems … 627

(a) 1 (b) 1
C12

Concurrence
C13

0.5 0.5

0 0
0 10 20 30 0 10 20 30
ωt ωt

(c) 1 (d) 1

Concurrence
0.5 0.5

0 0
0 10 20 30 0 10 20 30
ωt ωt

Fig. 25.4 The dynamics of quantum entanglement between three qubit-pairs C12 (qubit 1 and 2,
solid) and C13 (qubit 1 and 3, dashed). Left column shows a non-Markovian regime with c ¼ 0:4.
Right column shows a regime close to Markov limit (we choose c ¼ 1:5). a and b use the same
pffiffiffi pffiffiffi
initial state ðj11i þ j00iÞ  j0i= 2; while c and d use the initial state ðj10i þ j01iÞ  j0i= 2

dissipatively coupled to a bosonic environment, most two-qubit entangled states

pffiffiffi
will be disentangled eventually, except for the Bell state ðj10i  j01iÞ= 2, which
preserves the quantum information due to the decoherence-free subspace. However,
as shown in Fig. 25.4b, d, quantum entanglement can be stored in a pair of qubits
robustly. This result can be naturally extended to N-qubit systems; the capacity of
storing quantum information will increase as the size of quantum system is
enlarged.

25.4.3 A Note on General N-Qubit Systems

We remark that the previous derivations for the two-qubit and the three-qubit sys-
tems can be extended to the more general case of N-qubit systems, with the Lindblad
P
operator L ¼ j jj rj . The general procedure for generalizing our results to N-qubit
systems is highlighted as follows. First, we need to determine the maximum order of
noise in the O operator. It is easy to prove that LN þ 1 ¼ 0 for a N-qubit system, and
the last term of the O operator, ON1 , must be in the form of LN . And the highest
order of noise in the O operator is N  1 [11]. For example, the O operator contains
the first-order noise in the two-qubit model, and up to the second order of noise in the
three-qubit models. Similarly, there is at most N  1 order of noise for the
N-qubit models. Second, we need to determine the “forbidden conditions”.
628 Y. Chen and T. Yu

The close condition for qubit is ðrj Þ2 ¼ 0. One can see that if two O operator
components Oj and Ok satisfy this condition j þ k [ N  2, then Oj Ok ¼ 0.
Therefore, generally, one can obtain the explicit form of RðtÞ ¼ MðP O
y Þ, by t
calculating

Zt Zt ! " #
Y Y d
Mðzs1


zs2j1 Pt Þ ¼


ds2


ds2j aðs2j1 ; s2j Þ M ð ÞPt :
j j
dzs2j
0 0
ð25:56Þ

Once the closed form of the RðtÞ operator is obtained, the exact master equation
is determined.

25.5 Conclusion

In this paper, based on the non-Markovian QSD approach, we analytically and

numerically investigate multiple-qubit systems dissipatively coupled to a
non-Markovian zero-temperature bosonic environment. We have explicitly
demonstrated how to establish an exact non-Markovian master equation from the
corresponding quantum state diffusion equation. Our approach is very flexible in
the sense that it can be readily modified to solve many other types of models such as
hybrid systems consisting of qubits, qutrits, continuous variable systems and
multiple-environment systems, to name a few [39]. The time-local exact master
equation approach studied in this paper represents a new advance in our investi-
gations of non-Markovian quantum dynamics and non-equilibrium quantum
dynamics. We expect that our newly developed theoretical approach will be useful
in attacking many real-world problems.

Appendix 1

Here we supply a proof of the Novikov theorem. To make the proof more generic,
we calculate the term Mðzs Pt Þ, where s and t are two independent time indexes. In
this, Mðzt Pt Þ is the limit case in which s ¼ t. By the definition of ensemble average
in (25.6), we have [8]
Z
d 2 z jzj2
Mðzs Pt Þ ¼ e zs Pt : ð25:57Þ
p
25 Non-Markovian Dynamics of Qubit Systems … 629

P
where jzj2 ¼ k jzk j2 and d 2 z ¼ d 2 z1 d 2 z2


. With the definition of
P  ixk s
z s ¼ i k gk z k e , we have
Z !
d 2 z1 d 2 z2 Y X
jzn j2  ixk s
Mðzs Pt Þ ¼


e i gk zk e Pt :
p p n k

Since all zk are independent to each other, the above integration can be simplified
as
!Z
X Y Z d 2 zn 2 dzk dzk jzk j2
Mðzs Pt Þ ¼ i gk eixk s ejzn j e zk Pt :
k n6¼k
p p

Integrating by parts, then we have,

Z
dzk dzk jzk j2
e zk P t
p
Z

dzk dzk @ jzk j2

¼  e Pt
p @zk
Z 

dzk dzk @ jzk j2 jzk j2 @
¼  e Pt þ e Pt
p @zk @zk
Z
dzk dzk jzk j2 @
¼ e Pt :
p @zk

Then

X Z
d 2 z jzj2 @
Mðzs Pt Þ ¼ i gk eixk s e Pt :
k
p @zk

Using the functional derivative chain rule,

X Z Zt
d 2 z jzj2 @zs d
Mðzs Pt Þ ¼ i gk eixk s e ds Pt
k
p @zk dzs
0
Z Zt
d 2 z jzj2
¼ e dsaðs; sÞOðt; s; z ÞPt ;
p
0
Zt
Mðzs Pt Þ ¼ dsaðs; sÞM½Oðt; s; z ÞPt :
0
630 Y. Chen and T. Yu

Now we have the Novikov theorem,

Zt
Mðzs Pt Þ ¼ dsMðzs zs ÞM½Oðt; s; z ÞPt ;
0
Zt
Mðzs Pt Þ ¼ dsMðzs zs ÞM½Pt Oy ðt; s; zÞ:
0

In the limit s ¼ t, we obtain

z ÞPt Þ;
Mðzt Pt Þ ¼ MðOðt;
ð25:58Þ
Mðz P Þ ¼ MðP O
y ðt; zÞÞ:
t t
t

Appendix 2

Inserting the expansion series of O operator (25.27) into the O operator evolution
(25.26), we have

Zt
@t Oðt; sÞ ¼ @t O0 ðt; sÞ þ zt O1 ðt; s; tÞ þ ds1 zs1 @t O1 ðt; s;1 Þ þ


; ð25:59Þ
0

for the left hand side. Furthermore, the right hand side of (25.26) can be expanded
as

dO
½iHsys þ Lzt  Ly O;

O  Ly
dzs
Zt Zt
¼ ½iHsys þ Lzt
0 ; O0   Ly d
 Ly O dsaðt; sÞ ds1 zs1 O1 ðt; s; s1 Þ
dzs
0 0
Zt Zt Zt
þ ½iHsys þ Lzt ; ds1 zs1 O1   ½Ly O

0; ds1 zs1 O1   ½Ly
1; O
ds1 zs1 O
0
0 0 0
Zt Zt Zt
d
 Ly dsaðt; sÞ ds1 ds2 zs1 zs2 O2 ðt; s; s1 ; s2 Þ
dzs
0 0 0
þ


: ð25:60Þ
25 Non-Markovian Dynamics of Qubit Systems … 631

Rt

¼
By the definition O dsaðt; sÞOðt; s; z Þ, we can calculate the terms
0

Zt Zt
d
dsaðt; sÞ ds1 zs1 O1 ðt; s; s1 Þ
dzs
0 0
Zt Zt
¼ dsaðt; sÞ ds1 dðs; s1 ÞO1 ðt; s; s1 Þ
0 0
Zt
¼
1 ðt; sÞ;
dsaðt; sÞO1 ðt; s; sÞ ¼ O
0

and

Zt Zt Zt
d
dsaðt; sÞ ds1 ds2 zs1 zs2 O2 ðt; s; s1 ; s2 Þ
dzs
0 0 0
Zt Zt Zt
¼ dsaðt; sÞ ds1 ds2 zs1 dðs; s2 ÞO2 ðt; s; s1 ; s2 Þ
0 0 0
Zt Zt Zt
þ dsaðt; sÞ ds1 ds2 zs2 dðs; s1 ÞO2 ðt; s; s1 ; s2 Þ
0 0 0
Zt Zt Zt Zt
¼ dsaðt; sÞ ds1 zs1 O2 ðt; s; s1 ; sÞ þ dsaðt; sÞ ds2 zs2 O2 ðt; s; s; s2 Þ
0 0 0 0
Zt Zt
¼ dsaðt; sÞ ds1 zs1 ðO2 ðt; s; s1 ; sÞ þ O2 ðt; s; s; s1 ÞÞ
0 0
Zt
¼
2 ðt; s1 ; sÞ þ O
ds1 zs1 ðO
2 ðt; s; s1 ÞÞ:
0

Equating the two sides for each order of noise z , we obtain a set of dynamical
equations for the On ðn ¼ 1; 2; . . .Þ. For the non-noise term, we have

@t O0 ¼ ½iHsys  Ly O

0 ; O0   Ly O

1:
632 Y. Chen and T. Yu

For the first-order noise terms, we have

Zt
ds1 zs1 @t O1
0
Zt n o
¼ ds1 zs1 ½iHsys Ly O

0 ; O1   ½Ly O

1 ; O0   Ly ðO

2 ðt; s1 ; sÞ þ O

2 ðt; s; s1 ÞÞ ;
0

and the evolution equation for O1 is obtained as

@t O1 ¼ ½iHsys  Ly O

0 ; O1   ½Ly O

1 ; O0   Ly ðO

2 ðt; s1 ; sÞ þ O

2 ðt; s; s1 ÞÞ:

Similarly, the set of coupled dynamical equations for all On can be determined
sequentially. For the terms containing zt , the boundary conditions can be obtained as
O1 ðt; s; tÞ ¼ ½L; O0 ðt; sÞ;
O2 ðt; s; s1 ; tÞ þ O2 ðt; s; t; s1 Þ ¼ ½L; O1 ðt; s; s1 Þ;
etc:

Appendix 3

In order to explicitly derive the RðtÞ for the three-qubit system model, we need to
calculate two terms Mfzs1 Pt g and Mfzs1 zs3 Pt g. Since the term Mfzs1 zs3 Pt g contains
second order of noise, it can be evaluated by using Novikov’s theorem twice (25.18).
Zt
Mfzs1 Pt g ¼ ds2 aðs1 ; s2 ÞMfOðt; s2 ÞPt g
0
2 3
Zt Zt
¼ ds2 aðs1 ; s2 Þ4O0 ðt; s2 Þqt þ ds3 O1 ðt; s2 ; s3 ÞMfzs3 Pt g5
0 0
Zt Zt Zt
þ ds2 aðs1 ; s2 Þ ds3 ds5 O2 ðt; s2 ; s3 ; s5 ÞMfzs3 zs5 Pt g;
0 0 0
Zt
Mfzs1 zs3 Pt g ¼ ds2 aðs1 ; s2 ÞMfzs3 Oðt; s2 ÞPt g
0
Zt Zt ( )
dOðt; s2 Þ
¼ ds2 ds4 aðs1 ; s2 Þaðs3 ; s4 ÞM Pt
dzs4
0 0
Zt Zt
þ ds2 ds4 aðs1 ; s2 Þaðs3 ; s4 ÞMfOðt; s2 ÞOðt; s4 ÞPt g:
0 0
25 Non-Markovian Dynamics of Qubit Systems … 633

After eliminating the zero terms by the “forbidden conditions”, RðtÞ can be
explicitly shown as (25.54).

References

1. L. Diósi, N. Gisin, W.T. Strunz, Non-Markovian quantum state diffusion. Phys. Rev. A 58,
1699–1712 (1998)
2. Y. Chen, J.Q. You, T. Yu, Exact non-Markovian master equations for multiple qubit systems:
quantum-trajectory approach. Phys. Rev. A 90, 052104 (2014)
3. C. Gardiner, P. Zoller, Quantum Noise (Springer-Verlag, Berlin Heidelberg, 2004)
4. Caldeira Ao, A.J. Leggett, Path integral approach to quantum Brownian motion. Phys. A 121,
587–616 (1983)
5. B.L. Hu, J.P. Paz, Y. Zhang, Quantum Brownian motion in a general environment: exact
master equation with nonlocal dissipation and colored noise. Phys. Rev. D 45, 2843–2861
(1992)
6. W.T. Strunz, L. Diósi, N. Gisin, Open system dynamics with non-Markovian quantum
trajectories. Phys. Rev. Lett. 82, 1801–1805 (1999)
7. W.T. Strunz, L. Diósi, N. Gisin, T. Yu, Quantum trajectories for Brownian motion. Phys. Rev.
Lett. 83, 4909–4913 (1999)
8. T. Yu, L. Diósi, N. Gisin, W.T. Strunz, Non-Markovian quantum-state diffusion: perturbation
approach. Phys. Rev. A 60, 91–103 (1999)
9. J. Jing, T. Yu, Non-Markovian relaxation of a three-level system: quantum trajectory
approach. Phys. Rev. Lett. 105, 240403 (2010)
10. X. Zhao, J. Jing, B. Corn, T. Yu, Dynamics of interacting qubits coupled to a common bath:
non-Markovian quantum-state-diffusion approach. Phys. Rev. A 84, 032101 (2011)
11. J. Jing, X. Zhao, J.Q. You, T. Yu, Time-local quantum-state-diffusion equation for multilevel
quantum systems. Phys. Rev. A 85, 042106 (2012)
12. J. Jing, X. Zhao, J.Q. You, W.T. Strunz, T. Yu, Many-body quantum trajectories of
non-Markovian open systems. Phys. Rev. A 88, 052122 (2013)
13. T. Yu, Non-Markovian quantum trajectories versus master equations: finite-temperature heat
bath. Phys. Rev. A 69, 062107 (2004)
14. C. Anastopoulos, B.L. Hu, Two-level atom-field interaction: exact master equations for
non-Markovian dynamics, decoherence, and relaxation. Phys. Rev. A 62, 033821 (2000)
15. C. Chou, T. Yu, B.L. Hu, Exact master equation and quantum decoherence of two coupled
harmonic oscillators in a general environment. Phys. Rev. E 77011112 (2008)
16. C. Anastopoulos, S. Shresta, B.L. Hu, Non-Markovian entanglement dynamics of two qubits
interacting with a common electromagnetic field. Quant. Inf. Process. 8, 549–563 (2009)
17. C.H. Fleming, A. Roura, B.L. Hu, Exact analytical solutions to the master equation of
quantum Brownian motion for a general environment. Ann. Phys. 326, 1207–1258 (2011)
18. H.M. Wiseman, Stochastic quantum dynamics of a continuously monitored laser. Phys. Rev.
A 47, 5180–5192 (1993)
19. H.P. Breuer, W. Huber, F. Petruccione, Fast Monte Carlo algorithm for nonequilibrium
systems. Phys. Rev. E 53, 4232–4235 (1996)
20. N. Gisin, I.C. Percival, The quantum-state diffusion model applied to open systems. J. Phys.
A: Math. Gen. 25, 5677–5691 (1992)
21. B.L. Hu, J.P. Paz, Y. Zhang, Quantum Brownian motion in a general environment II:
nonlinear coupling and perturbative approach. Phys. Rev. D 47, 1576–1594 (1993)
22. S. Lin, C. Chou, B.L. Hu, Disentanglement of two harmonic oscillators in relativistic motion.
Phys. Rev. D 78, 125025 (2008)
634 Y. Chen and T. Yu

23. C.H. Fleming, B.L. Hu, Non-Markovian dynamics of open quantum systems: stochastic
equations and their perturbative solutions. Ann. Phys. 327, 1238–1276 (2012)
24. J.P. Paz, A.J. Roncaglia, Dynamics of the entanglement between two oscillators in the same
environment. Phys. Rev. Lett. 100, 220401 (2008)
25. W. Zhang, P. Lo, H. Xiong, M. Tu, F. Nori, General non-Markovian dynamics of open
quantum systems. Phys. Rev. Lett. 109, 170402 (2012)
26. W.T. Strunz, T. Yu, Convolutionless non-Markovian master equations and quantum
trajectories: Brownian motion. Phys. Rev. A 69, 052115 (2004)
27. G. Lindblad, Entropy, information and quantum measurements. Commun. Math. Phys. 33,
305–322 (1973)
28. J. Gea-Banacloche, Qubit-qubit interaction in quantum computers. Phys. Rev. A 57, R1
(1998)
29. F. Setiawan, H. Hui, J.P. Kestner, X. Wang, S.D. Sarma, Robust two-qubit gates for
exchange-coupled qubits. Phys. Rev. B 89, 085314 (2014)
30. A.C. Doherty, M.P. Wardrop, Two-qubit gates for resonant exchange qubits. Phys. Rev. Lett.
111, 050503 (2013)
31. L.A. Wu, D.A. Lidar, Dressed qubits. Phys. Rev. Lett. 91, 097904 (2003)
32. A.M. Childs, W. van Dam, Quantum algorithms for algebraic problems. Rev. Mod. Phys. 82,
1–52 (2010)
33. P. Kok, W.J. Munro, K. Nemoto, T.C. Ralph, J.P. Dowling, G.J. Milburn, Linear optical
quantum computing with photonic qubits. Rev. Mod. Phys. 79, 135–174 (2007)
34. P. van Loock, S.L. Braunstein, Multipartite entanglement for continuous variables: a quantum
teleportation network. Phys. Rev. Lett. 84, 3482–3485 (2000)
35. F. Xue, S.X. Yu, C.P. Sun, Quantum control limited by quantum decoherence. Phys. Rev.
A 73, 013403 (2006)
36. A.M. Brańczyk, P.E.M.F. Mendonça, A. Gilchrist, A.C. Doherty, S.D. Bartlett, Quantum
control of a single qubit. Phys. Rev. A 75, 012329 (2007)
37. F. Delgado, Quantum control on entangled bipartite qubits. Phys. Rev. A 81, 042317 (2010)
38. W.K. Wootters, Entanglement of formation of an arbitrary state of two qubits. Phys. Rev. Lett.
80, 2245–2248 (1998)
39. Y. Chen, J.Q. You, T. Yu, Generic non-Markovian master equations for multilevel systems.
Submitted to Phys. Rev. A (2015)
Chapter 26
Computing with Emerging
Nanotechnologies

M. Altun

Abstract As current CMOS based technologies are approaching their anticipated

limits, emerging nanotechnologies start to replace their role in electronic circuits.
New computing models have been proposed. This chapter overviews both deter-
ministic and stochastic computing models targeting nano-crossbar switching arrays
and emerging low-density circuits. These models are demonstrated with imple-
mentations using Boolean and arithmetic logic. Performance parameters of the
models such as area, speed, and accuracy, are also evaluated in comparison with
those of conventional circuits.

26.1 Introduction

In 1965, Gordon Moore made an influential prediction about CMOS size shrinking,
formulated as the Moore Law stating that the number of transistors on a chip
doubles every 18–24 months [1]. His prediction has kept its validity for decades.
Nowadays this trend has reached a critical point and it is widely accepted that the
trend will end in the next decade. Even Gordon accepted that his prediction will
lose it validity in near future [2]. At this point, research is shifting to novel forms of
nanoarchitectures including nano-crossbar arrays and probabilistic/stochastic cir-
cuits and systems [3–5]. Such technologies have apparent advantages over con-
ventional CMOS technologies, such as high performance capacity and easy
manufacturability.
Nano-crossbar arrays are regular and dense structures that are generally fabri-
cated by self-assembly as opposed to lithography based conventional and relatively
costly CMOS fabrication techniques [6]. Conventional lithographic techniques face
severe challenges for emerging nanotechnologies due to their need for directed

M. Altun (&)
Electronics and Communication Engineering Department, Istanbul Technical University,
Ayazağa Campus, 34469 Maslak, Istanbul, Turkey
e-mail: [email protected]

© Springer International Publishing Switzerland 2016 635

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4_26
636 M. Altun

manipulation of molecules which is quite costly in nanoscale. Along with these

advantages in terms of circuit size and fabrication, nano-crossbar arrays have
drawbacks including reliability issues and CMOS integration problems, standing
against commercial production. While reliability of nano-crossbars have been sat-
isfactorily improved using reconfigurable architectures and new defect tolerance
techniques [5, 7, 8], discussed later in this chapter, CMOS integration is still a
major problem; CMOL is the strongest candidate for this problem [9]. Indeed, if all
parts of a computing system can be successfully realized with nano-crossbar arrays
then there will be no need for integration, but current state-of-the art has not reached
this point, hopefully in the next decade.
The concept of using stochastic computing models is not new; it dates back to a
seminal paper by John von Neumann in 1956 [10]. With the advent of a variety of
types of emerging nanoscale technologies, the model has found renewed interest
[3, 11]. Unlike conventional CMOS that is solely based on deterministic operations,
stochastic circuits use probabilities as inputs and outputs. This feature is invaluable
to cope with uncertainties seen in emerging nanotechnologies in the form of
variability, reliability, and noise problems [4]. Stochastic computing also offers
smaller circuit implementations for arithmetic functions using much fewer tran-
sistors compared to conventional CMOS circuits. This feature attracts low density
obligated technologies such as printed/flexible electronics [12]. Here, the main
drawback is high error rates seen in stochastic computing by nature. Methods, as
discussed later in this chapter, have been proposed to improve it.
In this chapter, we focus on computing models for nano-crossbar arrays and
stochastic circuits. These models are demonstrated with implementations using
Boolean and arithmetic logic. Performance parameters of the models such as area,
speed, and accuracy, are also evaluated in comparison with those of conventional
circuits. This chapter is organized as follows. In Sect. 26.2, we investigate
nano-crossbar array based computing models. We present Boolean function
implementation and defect tolerance techniques in Sects. 26.2.1 and 26.2.2,
respectively. We evaluate the techniques on standard benchmark circuits in
Sect. 26.2.3. In Sect. 26.3, we introduce stochastic computing models and their
application areas. We present techniques to reduce error rates and to achieve error
free stochastic computing in Sects. 26.3.1 and 26.3.2, respectively. In Sect. 26.4, we
present conclusions.

26.2 Computing with Nano-crossbar Arrays

Unlike conventional CMOS that can be patterned in complex ways with lithogra-
phy, self-assembled nanoscale systems generally consist of regular structures.
Logical functions and memory elements are achieved with arrays of crossbar-type
switches. In this study, we target this type of switching arrays where each cross-
point behaves as a switch, either two-terminal or four-terminal. This is illustrated in
Fig. 26.1. Depending on the used technology, a two-terminal switch based
26 Computing with Emerging Nanotechnologies 637

Two-terminal switch
Switching nano array
Closed Open

Four-terminal switch
Closed Open

Fig. 26.1 A switching crossbar nanoarray modeled with two-terminal and four-terminal switches

Fig. 26.2 a Nano-crossbar [14] based on b diode [14], c FET [15], d four-terminal switch [23]
crosspoints

crosspoint can be modeled as a diode [13, 14] or a FET [15, 16]. This is illustrated
in Fig. 26.2. Note that both diode and FET based crosspoints conduct current in one
direction. However, four-terminal switches conduct current in multiple directions.
We implement Boolean functions by considering array sizes. Table 26.1 com-
pares different implementation methodologies for few XOR functions (Parity
functions) regarding array sizes. The columns “diode based” and “FET based”
represent two-terminal switch based implementation methodologies. These
methodologies have been proposed to implement simple logic functions [17, 18].
638 M. Altun

Table 26.1 Array sizes for nanoarray computing models; XOR2 = x1 ⊕ x2,
XOR3 = x1 ⊕ x2 ⊕ x3, and XOR4 = x1 ⊕ x2 ⊕ x3 ⊕ x4
Two-terminal switch based Four-terminal switch based nanoarray
nanoarray models models
Diode based FET based Four-terminal Four-terminal based
(Optimal) [13] (Optimal) [16] based [23] (Optimal) [36]
XOR2 2 × 5 array 4 × 4 array 2 × 2 array 2 × 2 array 4 switches
10 switches 16 switches 4 switches
XOR3 4 × 7 array 6 × 8 array 4 × 4 array 3 × 3 array 9 switches
28 switches 48 switches 16 switches
XOR4 8 × 9 array 8 × 16 array 8 × 8 array 3 × 5 array 15 switches
72 switches 128 switches 64 switches

In this study, we generalize them to be applicable for any given Boolean function
with offering optimal array size formulations. The last two columns represent
four-terminal switch based implementation methodologies that offer favorably
better results.
Defect rates are much higher for nano-crossbars compared to conventional
CMOS circuits [19]. Therefore developing new defect tolerance techniques for
nano-crossbars is a must, especially for high defect rates up to 20 %. Tolerating
such high defect rates necessitates using reconfigurable crossbar architectures and
redundancy [5, 20]. A predetermined design with static crossbars is not capable for
defect tolerance because it is not possible to create alternative routes for defective
regions. On the contrary, reconfigurable designs can be manipulated for defect
tolerance. In this study, we assess and compare defect tolerance performances of the
three different reconfigurable nano-crossbar architectures/technologies, represented
in Fig. 26.2, that is conducted through finding a valid mapping in accordance with
the proposed algorithm and defect maps in case of randomly distributed defects. We
consider randomly occurred stuck-open and stuck-closed crosspoint defects causing
permanently open and closed defective crosspoint devices or switches, respectively.
This study is at the technology-independent level. The presented synthesis and
optimization methods are applicable to variety of nanoarray based emerging tech-
nologies including nanowire and nanotube crossbar arrays [13, 15–18], magnetic
switch-based structures [8], arrays of single-electron transistors [21], and memris-
tive arrays [22].

26.2.1 Implementing Boolean Logic Functions

We investigate three major implementation methodologies developed for switching

nanoarrays. We classify them as two-terminal or four-terminal switch based.
26 Computing with Emerging Nanotechnologies 639

26.2.1.1 Two-Terminal Switch Based Methodologies

These methodologies consider each crosspoint of an array as a two-terminal switch

that behaves like a diode or a FET. Since diodes and FETs conduct current through
their two terminals that are anode and cathode for diodes and source and drain for
FETs, they are fundamentally two-terminal switches.
Boolean functions are implemented by using conventional techniques from
diode-resistor logic and CMOS logic with an important constraint regarding
nanoarray structures. Boolean functions should be implemented in their
sum-of-products (SOP) forms; other forms such as factored or BDD (Binary
Decision Diagram) cannot be used since these forms require manipulation/wiring of
switches that is not applicable for self-assembled nanoarrays. Figure 26.3 shows
implementations of a Boolean function XOR2 with diode and with FET based
nanoarrays.
Array size formulations: Given a target Boolean function f, we derive formulas
of the array sizes required to implement f. This is shown in Table 26.2. For diode
based implementations, each product of f requires a row (horizontal line), and each
literal of f requires a column (vertical line) in an array. Additionally, one extra
column is needed to obtain the output. For FET based implementations, each
product of f and its dual, f D, requires a column, and each literal of f requires a row in

Fig. 26.3 a Diode and b FET (a)

based nanoarrays R2
implementing
XOR2 = x1 ⊕ x2 with 2 × 5
and 4 × 4 arrays, respectively R1

f
R3 x1 x2 x1 x2

(b)

x1

x1

x2

x2

f
640 M. Altun

an array. As an example shown in Fig. 26.3: f = XOR2 = x1 x2 + x1 x2 has 4 literals

and 2 products; f D = x1 x2 + x1 x2 has 2 products. This results in array sizes of 2 × 5
and 4 × 4 for diode and FET based implementations, respectively. Note that both
formulas, for diode and FET, always result in optimal array sizes; no further
reduction is possible.

26.2.1.2 Four-Terminal Switch Based Methodology

This methodology considers each crosspoint of an array as a four-terminal switch.

An example is shown in Fig. 26.4. The four terminals of the switch are all either
mutually connected (ON) or disconnected (OFF). Boolean functions are imple-
mented with top-to-bottom paths in an array by taking the sum (OR) of the product
(AND) of literals along each path. This makes Boolean functions implemented in
their sum-of-products (SOP) forms. Figure 26.5a, b show the implementations of a
Boolean function XOR2 in an array and lattice representations, respectively.
Figure 26.5c shows a lattice of four-terminal switches implementing a Boolean
function x1x2x3 + x1x2x5x6 + x2x3x4x5 + x4x5x6. The function is computed by taking
the sum of the products of the literals along each path. These products are x1x2x3,
x1x2x5x6, x2x3x4x5, and x4x5x6.
Array size formulation: Given a target Boolean function f, the array size for-
mula was proposed by Altun and Riedel [23] that is shown in Table 26.3. In their
implementation, each product of f and its dual, f D, require a column and a row,
respectively, in an array. As an example shown in Fig. 26.5a, f = XOR2 = x1 x2 +
x1 x2 and f D = x1 x2 + x1 x2 have both 2 products. This results in an array size of
2 × 2.
Examining the array size formulas in Tables 26.2 and 26.3, we see that while the
formulas in Table 26.2 always result in optimal sizes, the sizes derived from the
formula in Table 26.3, that is, for four-terminal switch based arrays, are not

Table 26.2 Array size formulas for diode and FET based implementations
Type Array size formulas (Optimal)
Diode (Number of products in f) x (“number of literals in f” + 1)
FET (Number of literals in f) x (“number of products in f” + “number of products in f D”)

ON OFF

Four-terminal
switch

Fig. 26.4 A four-terminal switch with its two states

26 Computing with Emerging Nanotechnologies 641

(a) (b) (c)

TOP
TOP

x1 x2 x1 x4
x1 x2

x2 x5
x2 x1
x2 x1
BOTTOM
x3 x6
f R
R
BOTTOM

Fig. 26.5 a Four-terminal switch based nanoarray and b its lattice representation implementing
XOR2 = x1 ⊕ x2 with a size of 2 × 2. c Four-terminal switch based lattice implementing
x1x2x3 + x1x2x5x6 + x2x3x4x5 + x4x5x6

Table 26.3 Array size Type Array size formula (Non-optimal)

formula for four-terminal
switch based implementation Four-terminal (Number of products in f)
x (number of products in f D)

necessarily optimal. In the following part we present an algorithm that finds an

optimal size implementation of any given target Boolean function.
Finding whether a certain array with assigned literals to its switches implements
a target function is the main problem in finding optimal sizes. This problem requires
to check if each assignment of 0’s and 1’s to the switches, corresponding to a row
of the target function’s truth table, results in logic 1 (a top-to-bottom path of 1’s
exists). To check this we have to enumerate all top-to-bottom paths; the size of this
task grows exponentially with the array size. This is a general statement that holds
also for our algorithm described below.
Our algorithm finds optimal array sizes to implement given target Boolean
functions with arrays of four-terminal switches in four steps:
(1) Obtain irredundant sum-of-products (ISOP) expressions of a given-target
function fT and its dual fD
T . Determine the upper bound on the array size using
the formula in Table 26.3.
Upper Bound (UB) = (number of products in fT) × (number of products in fD T ).
The implementable lower bound (LB) values are taken from the lower bound
table proposed in [23].
(2) List the array shapes (RxC) (which are in between LB and UB) into the ‘List
of Implementable Nanoarray Shapes’ and sort them regarding of array sizes, in
ascending order. While ordering, first take the array shape which has lower
number of rows (e.g. if the kth shape is “3 × 4”, then the (k + 1)th shape can be
“4 × 3”.). Suppose that there are total of N different shapes in the list. For step
3, start with n = 1 (1 ≤ n ≤ N).
642 M. Altun

(3) Compute the value of the following statement for the nth shape.
The Statement: An array which has the shape in the nth line of the list is
implementable for fT.
If the statement is TRUE
Change UB to the R×C (save the design);
Go to the step 4;
If the statement is FALSE
Increase the number “n” by 1 (n = n+1);
Repeat step 3.
(4) Declare that UB is optimal size for given-target function fT can be realized in.
Our algorithm is mainly based on finding a design in a certain sized array such
that the design implements fT. Our algorithm does not check every possible design.
If it did then it would be intractable even for small sized arrays. For example, if a
target function fT having 6 variables, 8 literals, is tested on a 3 × 4 array then there
are 1210 possible designs and 26 truth table rows. Note that for each of the 12
switches in the array there are 10 different options; it might be one of the 8 literals,
0, or 1. In this scenario, the algorithm would have to check 1210 × 26 truth table
rows. To overcome this problem, we discard a significant portion of designs to be
checked. For this purpose, we offer 3 major improvements:
I. We create a library of reduced number of R × 2 sized sub-designs. We use them
to achieve R × C sized designs. While creating sub-designs we exploit the
following simple lemmas. First lemma allows us to discard designs imple-
menting a product (s) that does not imply fT. The second lemma allows us to
discard designs with “0” assignments to the switches if fT has a product having a
single literal.

Lemma 1 If a design has a path realizing a product p for which fT ≠ fT + p, then

the design can not implement fT.
Proof Since p is not an implicant of fT, then a design including p implements a
different function.
Lemma 2 If a function fT has a single variable product term p = x then the
algorithm does not need to assign “0” to the switches.
Proof All the “0” assignments can be replaced with x’s without a loss of generality.
II. If there is a product of fT such that the number of literals of the product equals
to the number of switches in the longest top-to-bottom path in the array, then
we settle that particular product onto that particular path.
III. We discard designs having fewer number of total literals than the total number
literals of fT.
These improvements make our algorithm much faster. As an example, suppose
that XOR3 is given as a target function for which the improved algorithm runs
roughly 400 times faster. For 3 × 2 sized sub-designs, there are 86 = 262,144 designs.
26 Computing with Emerging Nanotechnologies 643

With applying the proposed improvements, this number is reduced to 12,114,

roughly 20 times smaller than the unimproved one. Since we use two sub-arrays for
XOR3, for the optimal array size of 3 × 4, the improved algorithm works 400 times
faster.

26.2.2 Defect Tolerance

In this section, defect tolerance performance of switching nano-crossbar arrays is

extensively studied. Three types of nanoarrays where each crosspoint behaves as a
diode, FET, and four-terminal switch, are considered. For each crosspoint, both
stuck-open and stuck-closed defect probabilities are independently taken into
consideration. A fast heuristic algorithm using indexing and mapping techniques is
presented [20]. The algorithm measures defect tolerance performances of the
crossbar arrays that are expected to implement a certain given function. The
algorithm’s effectiveness is demonstrated on standard benchmark circuits that
shows 99 % accuracy compared with an exhaustive search. The benchmark results,
presented in the next section, also show that not only the used technology, the
nanoarray type, but more significantly the specifics of given functions affect defect
tolerance performances.
Mapping a target logic function on a defective crossbar is an NP-complete
problem [24]. In the worst-case scenario, an N × M crossbar has N!M! permutations
that is intractable for a reasonable computing time. Different algorithms and
heuristics are presented to tackle this issue. Graph based models are proposed in
[25, 26] that use a fan-out embedding heuristic and a maximum flow algorithm,
respectively. In addition to graph based approaches, “Integer Linear Programming”
is used in [27] that employs a pruning centered approach with certain constraints. It
is shown in [27] that defect tolerance results might dramatically depend on the
chosen algorithm correlated to the algorithm’s accuracy. We test and compare our
algorithm with an exhaustive search to establish an accuracy of 99 %. However, it
should be noted that large crossbars are computationally intractable to be included
in exhaustive search so we only consider crossbars up to 7 × 7 size for comparison.
In addition, the approaches mentioned so far are using pre-determined crossbar
sizes to find a mapping for a chosen logic function. We use an optimal crossbar to
realize a logic function which means that both the function and the crossbar
matrices have the same size. We test logic functions in Irredundant
Sum-of-Products (ISOP) form that is consistent with using optimal crossbar sizes.
Also we include three different logic families for comparison which departs from
the mentioned studies in the literature.
We present a heuristic algorithm that creates and compares index representations
of a given function and a defective certain sized crossbar to be used to implement
the function. We show that if the index representations are not matched then the
defective crossbar cannot be used to implement the function; otherwise, it can be
used. We prove that this is a necessary and sufficient condition. Our algorithm
644 M. Altun

eliminates considerable amount of crossbar mapping permutations that is the main

headache for the mentioned studies in the literature. Furthermore, our viewpoint is
comprehensive, that is, concerned both with the types of the crossbar technologies
and the characteristics of given functions. The presented indexing based algorithm
is direct used for diode and CMOS based logic of nano-crossbars. For four-terminal
switch based logic, the presented algorithm is partially used; a conventional
matrix-based matching is mainly performed. Also we include three different logic
families for comparison which departs from stated studies. It is important to
determine features of different families due to post-production selection according
to inherent defect types.
As follows, we first explain the algorithm used for diode and CMOS based logic
of nano-crossbars [20]. Then, we briefly explain the defect tolerance technique used
for four-terminal switch based logic. In the next section, we present experimental
results and elaborate on them.

26.2.2.1 The Algorithm for Diode and CMOS Based Logic

The outline of our four-step algorithm is shown below. We will then explain each
step in details. The algorithm will be demonstrated with an example in Fig. 26.6. It
should be noted that the example and the following explanations are for stuck-open
defects, nevertheless they can be applied easily for stuck-closed defects by con-
sidering defects as 1s (as opposed to 0s) to be matched with 1s (as opposed to 0s) in
the function matrix.

Input: Function matrix and crossbar (defective) matrix

Output: If there is a matching, “YES”; otherwise “NO”

Step 1: If the number of defective switches is greater than the corre-

sponding elements in the function matrix, then return “NO”.
Step 2: Sort matrices according to the row and column index, if the
crossbar matrix has at least one row or column index greater than a
row or column index of the function matrix, return “NO”.
Step 3: If the number of defects is equal to or smaller than the worst-case
limit WC, return “YES”.
Step 4: Find the reduced matrix and find set of double indices. Start
subarray search. If a subarray is found with the equal set of double
indices as the reduced matrix with 20,000 trials, return “YES”;
otherwise return “NO”.
26 Computing with Emerging Nanotechnologies 645

The explanations of the algorithm steps for stuck-open defects:

Step 1: If the number of defective switches is greater than the corresponding
elements in the function matrix, then return “NO”.
We consider stuck-open defects so algorithm checks 0s in the crossbar
matrix to match 0s in the function matrix. If 0s in the crossbar matrix is
greater than 0s in the function matrix, then it is not possible to find
mapping.
Step 2: Sort matrices according to the row and column index, if the crossbar
matrix has at least one row or column index greater than a row or
column index of the function matrix, return “NO”.
We sort matrices according to the row and column indices. For example, in
Fig. 26.6 index sets of the sorted function and crossbar matrices for 0 are:

IR; F ¼ f2; 2; 2; 1g IC; F ¼ f2; 2; 2; 1g

IR; C ¼ f2; 1; 0; 0g IC; C ¼ f2; 1; 0; 0g:

There is a perfect matching between sets. However, if a member of a

crossbar set would be greater than corresponding member in a function
set, there would be no matching. This would mean there are excessive

Fig. 26.6 Fourth step of the

presented algorithm: sorting
the function matrix, crossbar
reducing, and subarray search
646 M. Altun

defective elements for matching. For the example in Fig. 26.6, this is not
an issue so we proceed to the step 3.
Step 3: If the number of defects is equal to or smaller than the worst-case limit
WC, return “YES”.
The worst-case limit of a function matrix is the maximum number of
tolerable defects in any defect distribution related to the row and column
index. We find WC with using sets of row and column indices. First we
choose minimum members in sets of row and column indices, sepa-
rately. After that we choose the minimum between two members
obtained in the first step. WC gives us minimum row and column index.
If crossbar matrix has defective elements less than or equal to Wc, then
defects can be matched with any row and column in a function matrix.
For illustration, let us now follow the process of finding the Wc of the
function matrix shown in Fig. 26.6.

IR; F ¼ f2; 2; 2; 1gIC; F ¼ f2; 2; 2; 1g

min{IR, F} = 1 and min{IC, F} = 1 so WC = 1

In Fig. 26.7, the crossbar matrix has three defective elements, so we cannot
conclude if there is a matching without performing the fourth step.
Step 4: Find the reduced matrix and find set of double indices. Start subarray
search. If a subarray is found with the equal set of double indices as the
reduced matrix with 20,000 trials, return “YES”; otherwise return
“NO”.
In a crossbar matrix, Xs corresponding to functional switches do not change the
double index of a matrix element. For this reason, we erase the columns and rows
consisting of only such elements (Xs) for compactness. The acquired matrix keeps
the same set of double indices. Figure 26.7 shows an example for this.
In the next step, we use a subarray search to find a matching between a reduced
matrix and a subarray. Then, the function and crossbar matrices are sorted
according to the sets of indices. We use this to increase the chance of finding
matrices with the same set of double indices. Since matching elements are collected
to the one side, the search progresses diagonally. It can be seen from Fig. 26.6 that
searching a subarray checks only the set of double indices of a chosen subarray, due
to the Double Index Theorem, presented below. As long as they have the same set
of double indices, permutation of matrices is not necessary. Once two matrices are
found with the same set, it means there is a mapping, so the algorithm returns
“YES”.

Fig. 26.7 Reduced matrix for subarray search

26 Computing with Emerging Nanotechnologies 647

Double Index Theorem There is a one-to-one matching between two matrices if

and only if their set of double indices are equivavelent.
Lemma 1 Row and column permutations do not alter the double index of a matrix
element.
Numbers of the double index are defined with the row and column indices in
which the element is found. Therefore even though permutations changes the
position of a row or a column, element is still in the same row and column with the
same row and column indices which defines the double index.
Lemma 2 Set of double indices is unique for a given matrix.
Proof The proof is by contradiction. Let’s assume such a matrix that has two
different set of double indices. Therefore sets should have different double indices
for one element or more. Double indices in a matrix are determined according to a
row and column index. Since matrix is not altered and has same number of elements
in rows and columns for a chosen value of 0 or 1, it is not possible to have different
row and column indices that comprises double indices. This contradicts with the
assumption of having different double index for an element or more.
Lemma 3 Set of double indices for a matrix does not change with row and column
permutations.
Proof Rows and columns of a matrix is consisted of matrix elements which have
the same double index after permutations according to Lemma 1. Therefore set
stays the same since its members are not changed.
Proof of the Double Index Theorem Sufficiency If there is a one-to-one matching
between two matrices, then they are identical by definition. Therefore their set of
double indices is equivavelent according to Lemma 2.
Necessity Lemma 2 states that set of indices is unique for a certain matrix and
Lemma 3 states that set of double indices stays the same after row and column
permutation. Therefore, if two matrices have the same set of double indices, then
they are either identical or permutation of one another; in both case they can be
matched one-to-one.
Subarray search has a trial limit of 20,000. We choose this value because when
compared with exhaustive search, it gives 99 % accuracy. When we run the
algorithm by removing this limitation, we see that there is almost no change in the
values. This validates the heuristic algorithm presented above interpreted as a
negative result meaning that there is no mapping at all. In subarray search, the
double index theorem is only valid for matrices with the same number of elements
to be matched. In case of unequal number of elements, new defective elements
should be introduced to the reduced matrix to equalize the number of elements. If
there are N missing elements for matching, 2N possibilities are to be considered.
Instead of doing this, functional switches denoted with x are marked with 0 and
defective switches with 1. Same is applied to the subarray in the function matrix.
Next, element by element multiplication of matrices is executed. Since functional
648 M. Altun

switches can be matched with 0s and 1s in the function matrix, the resulting matrix
can be compared with the reduced matrix. If they are equal, there is a matching.
In the CMOS based design two matrices are used to model a logic function. All
principles used for diode based design are valid, with the exception of input per-
mutations. In the diode based design both inputs (columns) and products (rows) can
be permutated with respect to the crossbar which are checked for a mapping. In the
CMOS based design, the first matrix is for the function itself and the second matrix
for its complement. Important distinction is that inputs are in the same order for
both matrices. For this reason while a mapping is searched for a crossbar, rows of
matrices can be permutated independently; however inputs must be in the same
order for both matrices.

26.2.2.2 Defect Tolerance for Four-Terminal Switch Based Logic

For four-terminal switch based design, we partially use the above presented algo-
rithm. We use matrix based defect maps similar to those used there. Our defect
tolerance technique mainly depends on permutation trials. Therefore, we mention it
only briefly.
Defect Map Due to its layout method, the four-terminal switch based design [17]
uses every switch on the crossbar. Therefore there is no unused switch like those in
diode and CMOS based designs. However, certain functions yield redundant paths
or extra connections between top and bottom plates. Figure 26.8 shows occurrence
of extra connections. If a defect existing in a crossbar appears only on one of the
connections, then it can be compensated with the other connection and the correct
result can still be achieved. A defect map of a function implemented with
four-terminal switch based design displays these type of connections.

Fig. 26.8 a Four-terminal implementation of f with two connections between top and bottom
plates; b in case of a stuck-open defect, first connection is broken down, however since there is a
second connection f evaluates correctly
26 Computing with Emerging Nanotechnologies 649

26.2.3 Simulation Results

In Table 26.4, we report synthesis results for standard benchmark circuits [28]. We
treat each output of a benchmark circuit as a separate target function. The number of
products for each target function f T and its dual fTD are obtained through
sum-of-products minimization using the program Espresso [29]. The array size
values for “Diode”, “CMOS”, and “4-terminal” are calculated by using the for-
mulas in Tables 26.2 and 26.3. The array size values for “Optimal 4-terminal” are
obtained using the presented optimization algorithm in Sect. 26.2.1: Implementing
Boolean logic functions.
Examining the numbers in Table 26.4, we always see the same sequence from
the worst to the best result as “CMOS”, “Diode”, “4-terminal”, and “Optimal
4-terminal”. This proves that models based on four-terminal switches overwhelm
those based on two-terminal switches regarding the array size. Further, the numbers
obtained by our optimal synthesis method compares very favorably to the numbers
obtained by previous methods.
In Table 26.5, we use standard benchmark circuits to measure defect tolerance
performances of different nano-crossbar technologies. We consider stuck-open and
stuck-closed defect probabilities/rates of 10 and 20 % for each crosspoint inde-
pendently. Simulations are conducted in Matlab. Crossbars with random defects are
produced with Matlab’s predetermined function generator. To obtain defect toler-
ance values, a sample size of around 600 is used. At this level the defect tolerance
fluctuation stabilizes. All experiments run on a 1.70-GHz Intel Core i5 CPU (only
single core used) with 4.00 GB memory. It takes 0.2 s for each sample in average to
check a valid mapping that satisfies an accuracy of 99 % compared with an
exhaustive search.
Table 26.5 shows the results of benchmark functions with respect to defect rates
and defect types as well as the crossbar technologies. Considering the technologies
and the related logic synthesis methodologies, the diode based logic always has a
better defect tolerance performance compared with the CMOS based one. The
reason behind this is directly connected to the number of matchings necessitated for
valid mapping. Since the CMOS based logic uses two different planes for function
realization, it needs to satisfy two matchings instead of one. Another important
conclusion is that the four-terminal switch based design yields better results for
stuck-closed defects than for stuck-open ones since the design generally requires to
assign the same literals to multiple switches on the same conduction path.
Characteristics of the functions also play an important role in the defect tolerance.
Since stuck-open defects are tolerated with zeros in the functions’ matrices, func-
tions with relatively higher number of products compared to their number of literals
have a better chance for tolerating these defects. On the contrary, functions with
relatively higher number of literals compared to their number of products have a
better chance for tolerating stuck-closed defects. For example, Ex33 in Table 26.1
has a 14 % defect tolerance for stuck-closed and 99 % for stuck-open type defects.
650 M. Altun

Table 26.4 Array sizes of Benchmark CMOS Diode 4-terminal Optimal

three different nano-crossbar 4-terminal
based logic families
Alu 0 30 18 6 6
Alu 1 30 18 6 6
Alu 2 30 18 6 6
Alu J 30 18 6 6
B12 0 80 32 24 12
B12 1 120 70 35 16
B12 3 30 20 8 8
B12 4 42 28 8 3
B12 6 132 77 35 18
B12 7 110 66 24 18
B12 8 90 70 14 14
C17 0 36 18 9 6
C17 1 30 20 8 8
Clpl 0 64 32 16 12
Clpl 1 36 18 9 9
Clpl 2 16 8 4 4
Clpl 3 144 72 36 18
Clpl 4 100 50 25 15
Del 1 25 10 6 6
Del 2 72 36 16 12
Del 5 35 15 12 6
Del 6 36 18 9 6
Ex5 31 156 104 32 24
Ex5 33 110 77 21 21
Ex5 46 81 54 18 18
Ex5 49 72 54 12 12
Ex5 50 81 63 14 14
Ex5 61 64 48 12 12
Ex5 62 49 35 10 10
Misexl 1 48 16 8 8
Misexl 2 132 55 35 15
Misexl 3 156 60 40 24
MLsexl 4 121 44 28 16
Misexl 5 90 45 25 15
Misexl 6 143 66 42 18
Misexl 7 81 36 20 15
Mp2d 4 345 75 90 24
Newrag 108 72 32 18
26 Computing with Emerging Nanotechnologies 651

Table 26.5 Defect tolerance performances of three different nano-crossbar based logic families
Circuit Diode CMOS Four-terminal
name Stuck-closed Stuck-open Stuck-closed Stuck-open Stuck-closed Stuck-open
10 % 20 % 10 % 20 % 10 % 20 % 10 % 20 % 10 % 20 % 10 % 20 %
Alu 0 85 54 99 96 63 50 93 72 86 64 52 26
Alu 1 85 54 99 96 83 53 97 73 86 64 52 26
Alu 2 85 54 99 96 83 53 97 73 86 64 52 26
Alu 3 85 54 99 96 83 53 94 73 86 64 52 26
B12 0 98 80 98 74 46 5 95 59 28 7 28 7
B12 1 92 33 99 75 58 1 99 91 36 9 58 19
B12 3 96 77 96 78 90 58 93 68 42 17 43 16
B12 4 84 40 99 96 79 25 93 74 42 17 43 16
B12 6 68 2 95 47 14 1 99 85 27 7 21 11
B12 7 44 2 99 95 34 1 82 75 42 11 22 3
B12 8 32 1 99 97 24 1 99 99 82 40 22 4
C17 0 95 78 99 94 92 70 98 87 53 26 53 26
CI7 1 96 77 96 78 91 64 92 69 43 16 43 16
ClplO 97 69 99 98 78 23 99 92 62 29 53 20
Clpl 1 98 84 99 95 98 83 98 82 39 14 48 20
Clpl 2 97 82 99 94 93 79 98 92 67 40 65 42
Clpl 3 87 53 99 81 49 1 50 21 18 3 41 10
Clpl 4 91 41 99 97 74 6 63 50 18 3 41 12
Del 1 99 97 95 75 84 52 93 73 52 25 52 25
Del 2 93 55 99 96 68 9 99 96 28 6 28 6
Del 5 99 95 97 85 96 84 84 53 65 38 53 26
Del 6 95 79 99 88 94 70 98 86 53 25 53 25
Ex5 31 56 5 99 95 30 1 83 64 35 7 22 2
Ex5 33 14 1 99 98 9 1 60 43 66 28 25 4
Ex5 46 45 5 99 99 38 1 84 65 17 1 28 6
Ex5 49 3 0 99 99 1 1 84 65 90 32 28 6
ExS 50 23 1 99 99 22 1 87 45 93 75 22 4
Ex5 61 29 2 99 99 25 1 98 78 90 32 43 16
Ex5 62 28 1 98 85 23 1 96 74 95 74 37 12
Misexl 1 99 96 92 66 65 17 52 9 44 16 43 16
Mlsexl 2 78 18 99 92 30 1 98 87 29 6 36 10
Misexl 3 94 38 99 86 10 1 96 67 8 1 38 11
Misexl 4 93 44 99 94 8 1 99 89 27 5 38 10
Misexl 5 86 45 97 80 63 3 95 64 26 4 42 14
Misexl 6 89 28 99 86 26 1 93 73 12 2 29 5
Misexl 7 95 57 99 92 50 1 99 93 21 3 49 15
Newtag 59 4 99 98 52 1 96 52 62 22 30 7
652 M. Altun

Significant difference between values occurs because of the mentioned literal

related properties. Using Ex33 is more favorable for a crossbar with higher prob-
ability of stuck-open type defects.

26.3 Stochastic Computing

Stochastic computing (SC), represents values in time domain by random bit streams
[29] and was first presented in a paper written by John von Neumann in 1956 [10].
These values are interpreted as probabilities. Therefore, the available range is [0,1]
interval. For instance, if a 16 bit length stream contains 12 1s, it represents p = 0.75
independently from the position of 1s in the stream. For example, both
(0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1) and (0,1,0,1,1,1,1,1,1,0,1,1,1,1,1,0) represent the
same probability p = 0.75. The same probability values can be obtained with
different bit stream lengths. Complicated stream operations can be performed with
simple circuits in SC. Multiplication and scaled addition are basic arithmetic
operations of SC.
Stochastic multiplication can be realized with one AND gate independently of
the input bit stream length. In contrary to that, conventional (digital) CMOS
multipliers need about 1000 gates for 16 bit multipliers and this number is expo-
nentially increasing with the number of bits. Due to the interval [0,1], addition
cannot be performed in SC. Instead, scaled addition that guarantees to remain in the
interval is described. MUX is used for scaled addition. Figure 26.9 shows stochastic
multiplication and scaled addition operations.
Error rates
Different combination of streams represent same values and this yields error rates in
SC. These error rates are extremely high for short bit streams. To reduce error rates,
longer bit streams should be used. Figure 26.10 illustrates this; in order have error
rates below 1 %, streams having more than 1000 bits are needed. These extremely
long streams extend the operation time. Hence, classical SC can not be effectively
used in mathematical operations.

Fig. 26.9 Stochastic implementation of arithmetic operations: a Multiplication. b Scaled addition

26 Computing with Emerging Nanotechnologies 653

Fig. 26.10 Error rates versus number of bits in the stream for AND gate while p1 = 1/2 and
p2 = 1/2

Fig. 26.11 Stochastic multiplication with and without an error

We calculate this error as

 
z  zexpected 
ð26:1Þ
zexpected

where z is the probability value of the output. And expected value varies for
different logic gates.
In order to get “1” at the output of an AND gate, both of the inputs have to be
“1”. Assume that the probability of being “1” is p1 for the 1st input and p2 for the
2nd input. Hence expected value of multiplication/AND gate is z = p1 p2. Following
the same logic, expected value at the output of a MUX is z = p1 ps + p2 (1 – ps).
Let two input probability values be p1 and p2 respectively depending on the
number of 1s in streams. The expected value at the output is p1 × p2. However, the
same probability obtained with different permutations of input may not always yield
the expected results. At this point error rates should be considered. This is illus-
trated in Fig. 26.11. The expected value can be obtained from the AND gate
symbolized with a. However, output value with an error rate is obtained from the
AND gate symbolized with b in Fig. 26.11.
Effect of dependency in stochastic computing
In conventional SC, bit streams which have Bernoulli distribution are independent
from each other. If input streams become dependent, which is not desire, expected
result for classical SC will change. If the inverse of the first input is applied as the
second input to an AND gate, “0” will be obtained at the output. It is expected that
z = p (1 − p), but due to dependency this expectation fails. Moreover, if the same
inputs are applied, the output will be same as the input. It acts like a buffer. It is
654 M. Altun

Fig. 26.12 Effect of dependency in SC

expected that z = p p, but due to dependency this expectation fails. These examples
are shown in Fig. 26.12.
Application areas
SC needs less area than conventional hardware implementations [4]. On the con-
trary, error rates are worse. Hence, the commercial application fields are limited
(image processing, LDPC codes) [30]. Recently, printed/flexible electronics
becomes a major candidate for stochastic computing. Printed electronics strictly
requires low density circuits that is a perfect fit with stochastic computing if error
rates can be reduced to proper values [31]. This is our motivation. In this study, we
offer a method based on improving the generation of bit streams. In Sect. 26.3.1, a
detailed information about our method and the conventional one is given.
Comparative analysis made between two methods demonstrate our methods suc-
cess. We also propose a method to achieve error free stochastic computing in
Sect. 26.3.2

26.3.1 Reducing Error Rates

We analyze and make comparison between two SC methods, namely random bit
assigning and shuffling methods.

26.3.1.1 Random Bit Assigning Method

In order to generate bit streams in SC, conventional random bit generators are used
[30]. We name this technique as random bit assigning method (RBAM) [32]. Let
desired input probability value for RBAM be p1 and let random number generator
generate values between [0–1] with binomial distribution. If the generated value is
smaller than p1, “1” is added to the stream. If the generated value is bigger than p1
“0” is added to the stream. Because this is a stochastic procedure, there exists
difference between the desired and the obtained input values. There are three 1s and
five 0s in the generated 8 bit stream in Table 26.6. For this reason, generated input bit
stream is physically pe = 0.375. Consequently, the result at the output will be faulty.
26 Computing with Emerging Nanotechnologies 655

Table 26.6 8 bit length Generated number from generator Bit stream
stream generated with RBAM
for p1 = 0.5 0.89967 0
0.45847 1
0.81537 0
0.20818 1
0.71289 0
0.43615 1
0.64214 0

26.3.1.2 Random Bit Shuffling Method

Let the input desired value for random bit shuffling method (RBSM) be p1. In this
method, depending on the desired probability value and the length of the bit stream,
a bit stream containing necessary and sufficient amounts of 1 is generated.
Fisher-Yates shuffling algorithm [33], adapted to computers by Durstenfeld [34], is
applied. In this way, it is guaranteed that the desired and the generated input values
are probabilistically identical. With this approach output error rates will be
decreased. We use basic logic gates to compare these two methods. The results are
obtained by fixing the two inputs, changing the inputs between [0, 1] interval and
using different bit lengths.

RBSM–RBAM Comparison for p1 = 1/2, p2 = 1/2 Bit Streams

As shown in Table 26.7, shuffling method gives better results for AND and OR
gates. The longer bit length reduces the error rates as expected and the difference of
error rates between two methods also is reduced. The expectation is that error rates
of two methods overlap at infinity. Two methods give similar results for XOR gate
between 4 and 128 bits, however for longer bit streams shuffling method gives
better results.

RBSM–RBAM Comparison While 64-Bit Inputs p1 and p2 Changing

The graphics in Fig. 26.13 are symmetrical with respect to the diagonal line. Two
inputs for the ideal operating range of AND gate should be 0.5 and/or greater.
Average error rates for RBAM and RBSM are 30.11 and 19.31 %, respectively. The
results are better for RBSM.
The graphics in Fig. 26.14 are symmetrical with respect to the diagonal line. The
error rates of an OR gate are better than those for an AND gate. Two inputs for the
ideal operating range of an OR gate should be 0.3 and/or greater. Average error
rates for RBAM and RBSM are 6.07 and 2.24 %, respectively. The results are better
for RBSM.
656 M. Altun

Table 26.7 Obtained error Number of RBAM RBSM

rates with different bit lengths bit AND OR XOR AND OR XOR
and three logic gates (%)
4 64.3 20.82 37.55 32.55 11.2 33.65
8 46.58 15.38 27.1 26.44 8.77 25.76
12 39.92 12.98 22.52 21.22 7.24 21.85
16 33.38 11.05 19.4 19.04 6.42 19.09
20 30.28 10.22 17.68 17.58 5.85 17.24
24 27.75 9.38 16.35 16.01 5.2 15.66
32 24.28 8.2 14.02 13.82 4.67 13.18
64 17.1 5.68 9.65 8.32 3.05 10.24
100 13.62 4.58 7.88 7.93 2.58 7.82
128 12.12 4.05 7.08 6.21 2.23 7.1
256 8.52 2.8 4.98 3.98 1.6 3.57
512 6.22 2 3.55 2.6 1.34 2.03
1024 4.42 1.4 2.45 1.95 0.35 1.56

Fig. 26.13 p1-p2-error graph for a RBAM and b RBSM AND gate

The graphics in Fig. 26.15 are symmetrical with respect to both diagonal lines.
The error rates of an XOR gate are better than those for an AND gate but worse
than those for an OR gate. Two inputs for the ideal operating range of XOR gate
should be 0.3 and/or greater. However, the two inputs should not take place on the
diagonal line at the same time. Average error rates for RBAM and RBSM are 10.94
and 6.58 %, respectively. The results are better for RBSM.
26 Computing with Emerging Nanotechnologies 657

Fig. 26.14 p1-p2-error graph for a RBAM and b RBSM OR gate

Fig. 26.15 p1-p2-error graph for a RBAM and b RBSM XOR gate

26.3.2 Error Free Stochastic Computing

Conventional wisdom in stochastic computing is that input bit streams should be

independent; we previously discussed potential accuracy problems related to
dependencies. In this work, however, we use dependent inputs to improve accuracy;
we achieve error free outputs.

26.3.2.1 Realizing Error Free Multiplication with 0.5

We suggest an error free block for multiplication by 0.5 shown in Fig. 26.16. With
this block, the input stream is multiplied by 0.5 and zero error at the output stream
658 M. Altun

is achieved. However if the length of the input stream is n, the length of the output
stream will be 2n because of the merging operation. The merge block can be
realized with a delay switch. Thus, the stream obtained from the second AND gate
waits the stream from the first AND gate to be finished.

26.3.2.2 Generating Any Probability Value Without an Error

We adapt the block in Fig. 26.16 to switch circuits [35] in order to obtain any
probability value. It is illustrated in Fig. 26.17. Inputs are always 0.5 (0.12 = 1/2).
In order to shift the binary value one digit to the right and add “0” to the most
significant bit, the circuit block based on AND gate is used (0.012 = 1/4). In order

Fig. 26.16 Error free multiplication for 0.5 p1

Fig. 26.17 Probability generation tree

26 Computing with Emerging Nanotechnologies 659

to shift the binary value one digit to right and add “1” to the most significant bit, the
circuit block based on OR gate is used (0.112 = 3/4). By cascading the blocks, any
desired probability value can be accurately obtained without an error.

26.4 Conclusions

This chapter overviews both deterministic and stochastic computing models tar-
geting nano-crossbar switching arrays and emerging low-density circuits. These
models are demonstrated with implementations using Boolean and arithmetic logic.
Performance parameters of the models such as area, reliability, and accuracy, are
also evaluated.

Acknowledgments This work is supported by TUBITAK (The Scientific and Technological

Council of Turkey) Career Program #113E760 and EU (European Union) H2020-MCSA Research
and Innovation Staff Exchange Program #691178.

References

1. L. Wilson, International technology roadmap for semiconductors (ITRS). Semiconductor

Industry Association (2013)
2. M. Dubash, Moore’s Law is dead, says Gordon Moore. Techworld (April 2005)
3. M. Haselman, S. Hauck, The future of integrated circuits: a survey of nanoelectronics. Proc.
IEEE 98(1), 11–38 (2010)
4. W. Qian, X. Li, M.D. Riedel, K. Bazargan, D.J. Lilja, An architecture for fault-tolerant
computation with stochastic logic. IEEE Trans. Comput. 60(1), 93–105 (2011)
5. D.B. Strukov, K.K. Likharev, Reconfigurable nano-crossbar architectures. Nanoelectronics
Inf. Technol. 543–562 (2012)
6. G.M. Whitesides, B. Grzybowski, Self-assembly at all scales. Science 295(5564), 2418–2421
(2002)
7. M. Altun, M.D. Riedel, Robust computation through percolation: synthesizing logic with
percolation in nanoscale lattices. Int. J. Nanotechnol. Mol. Comput. (IJNMC) 3(2), 12–30
(2011)
8. A. Khitun, M. Bao, K.L. Wang, Spin wave magnetic nanofabric: a new approach to spin-based
logic circuitry. IEEE Trans. Magn. 44(9), 2141–2152 (2008)
9. Z. Abid, M. Liu, W. Wang, 3D integration of CMOL structures for FPGA applications. IEEE
Trans. Comput. 60(4), 463–471 (2011)
10. J. Von Neumann, Probabilistic logics and the synthesis of reliable organisms from unreliable
components. Automata Stud. 34, 43–98 (1956)
11. S. Gaba, P. Knag, Z. Zhang, W. Lu, Circuits and Systems (ISCAS), in 2014 IEEE
International Symposium on Memristive devices for stochastic computing. IEEE (2014, June),
pp. 2592–2595
12. A.C. Arias, J.D. MacKenzie, I. McCulloch, J. Rivnay, A. Salleo, Materials and applications
for large area electronics: solution-based ap-proaches. Chem. Rev. 110(1), 3–24 (2010)
13. Y. Huang, X. Duan, Y. Cui, L.J. Lauhon, K.H. Kim, C.M. Lieber, Logic gates and
computation from assembled nanowire building blocks. Science 294(5545), 1313–1317
(2001)
660 M. Altun

14. W. Lu, C.M. Lieber, Nanoelectronics from the bottom up. Nat. Mater. 6(11), 841–850 (2007)
15. P. Avouris, Molecular electronics with carbon nanotubes. Acc. Chem. Res. 35(12), 1026–1034
(2002)
16. G. Snider, U.S. Patent No. 6,919,740. Washington, DC: U.S. Patent and Trademark Office
(2005)
17. Z. Chen, et al., An integrated logic circuit assembled on a single carbon nanotube. Science
311.5768, 1735–1735 (2006)
18. H. Yan, H.S. Choe, S. Nam, Y. Hu, S. Das, J.F. Klemic, C.M. Lieber, Programmable nanowire
circuits for nanoprocessors. Nature 470(7333), 240–244 (2011)
19. C.P. Collier, E.W. Wong, M. Belohradský, F.M. Raymo, J.F. Stoddart, P.J. Kuekes, J.R. Heath,
Electronically configurable molecular-based logic gates. Science 285(5426), 391–394 (1999)
20. O. Tunali, M. Altun, Defect tolerance in diode, FET, and four-terminal switch based
nano-crossbar arrays, in IEEE International Symposium on Nanoscale Architectures, 2015.
NANOARCH 2015. pp. 82–87. IEEE (2015, July)
21. Y.C. Chen, S. Eachempati, C.Y. Wang, S. Datta, Y. Xie, V. Narayanan, Automated mapping
for reconfigurable single-electron transistor arrays, in Proceedings of the 48th Design
Automation Conference, pp. 878–883. ACM (2011, June)
22. Y. Levy, J. Bruck, Y. Cassuto, E.G. Friedman, A. Kolodny, E. Yaakobi, S. Kvatinsky, Logic
operations in memory using a memristive Akers array. Microelectron. J. 45(11), 1429–1437 (2014)
23. M. Altun, M.D. Riedel, Logic synthesis for switching lattices. IEEE Trans. Comput. 61(11),
1588–1600 (2012)
24. A.M.S. Shrestha, S. Tayu, S. Ueno, Orthogonal ray graphs and nano-PLA design. In ISCAS
(2009, May), pp. 2930–2933
25. W. Rao, A. Orailoglu, R. Karri, Topology aware mapping of logic functions onto
nanowire-based crossbar architectures, in Proceedings of the 43rd Annual Design
Automation Conference. ACM (2006, July), pp. 723–726
26. J. Huang, M.B. Tahoori, F. Lombardi, On the defect tolerance of nano-scale two-dimensional
crossbars, in Proceedings of 19th IEEE International Symposium on Defect and Fault
Tolerance in VLSI Systems, 2004. DFT 2004. IEEE (2004, October), pp. 96–104
27. M. Zamani, H. Mirzaei, M.B. Tahoori, ILP formulations for variation/defect-tolerant logic
mapping on crossbar nano-architectures. ACM J. Emerg. Technol. Comput. Syst. (JETC) 9(3),
21 (2013)
28. K. McElvain, in Distributed as part of the MCNC International Workshop on Logic Synthesis
IWLS’93 benchmark set: Version 4.0 (vol. 93) (1993, May)
29. B.R. Gaines (1967, April). Stochastic computing. In Proceedings of the April 18–20, 1967,
Spring Joint Computer Conference. ACM, pp. 149–156
30. A. Alaghi, J.P. Hayes, Survey of stochastic computing. ACM Trans. Embed. Comput. Syst.
(TECS) 12(2s), 92 (2013)
31. W.S. Wong, A. Salleo (eds.), Flexible electronics: materials and applications (vol. 11).
Springer Science & Business Media (2009)
32. S. Yavuz, M. Altun, Stokastik Hesaplamada Hata Oranlarını Azaltmak için Rastgele Bit
Karıştırma Yöntemi—Random Bit Shuffling Method for Reducing Error Rates in Stochastic
Computing, in Elektrik—Elektronik—Bilgisayar ve Biyomedikal Mühendisliği Sempozyumu,
pp. 728–732. ELECO (2014)
33. R.A. Fisher, F. Yates, Statistical tables for biological, agricultural and medical research, 3rd
edn. (1949)
34. R. Durstenfeld, Algorithm 235: random permutation. Commun. ACM 7(7), 420 (1964)
35. H. Zhou, P.L. Loh, J. Bruck, The synthesis and analysis of stochastic switching circuits.
(2012). arXiv preprint arXiv:1209.0715
36. C. Morgul, M. Altun, Synthesis and optimization of switching nanoarrays, in 2015 IEEE 18th
International Symposium on Design and Diagnostics of Electronic Circuits & Systems
(DDECS). pp. 161–164. IEEE (2015, April)
Index

A Amorphous silicon dioxide, 115

Ab initio, 201 Amphiphilic molecules, 510
Ab initio modeling, 181 Anatase, 426, 430, 431, 436, 437
Ablation, 174 Anatase and rutile, 434
Ablation threshold, 174 Anatase/brookite, 432, 441, 445, 446
Absorbing light, 248 Anatase/rutile, 433, 434, 436, 439, 442, 447
Absorption (ABS), 332 Anatase TiO2, 384
Absorption coefficient, 364 Angle of incidence, 103
Absorption edge, 341 Angle resolved X-ray photoemission
Absorption properties, 443 spectroscopy, 249
Absorption spectrum, 378 Anharmonic contribution, 153
Acceptor, 189, 253, 566 Anisotropic growth, 361
Accuracy, 647 Antenna mechanism, 453, 454
ACN, 470 Antibody conjugate, 520
Acoustic plasmon, 205, 225 Antibody conjugated QDs, 215, 569
Acoustic plasmon (lower), 217 AO, 262
Acoustic surface plasmon (ASP), 222, 223 Aperture, 73, 74, 76, 78
Activation states, 143 Applications, 246, 441
Active nanostructures, 108 Applications of QDs, 506
Active pharmaceutical ingredients, 110 bionanotechnology and nanomedicine, 565
Ad-atom, 184–190, 199, 201, 275 Applied potential, 208
Adsorption, 183–185, 187–189, 478, 481, 483, Aptamer, 519
492, 500 Aquaporins, 546
Adsorption ability, 428 Aqueous synthesis, 558
Adsorption capacity, 495 AR-XPS, 249
Advantages, 424, 433, 436 Arc discharge, 244
AFM, 113 Armored protection systems, 547
Ag modification, 448 ARPES, 312, 322
AgGa0.5In0.5Se2, 372 Array size formulas, 640
Agglomeration, 439 Arrays of single-electron transistors, 638
Aggregates, 438 Artificial preparation, 427
AlGaAs/GaAs heterostructure, 19 Artificial retinas, 113
AlGaN/GaN, 24, 43 Asialoglycoprotein receptor, 582
AlGaN, 161 Asymptotic absorption edge, 336
Algorithm, 641 ATR prism, 142
Alkalinity, 487 Attentuated total reflection, 133
Ammonium, 477, 478, 487, 492, 495, 498 Au modified, 442
Amorphous, 292, 294 Autofluorescent, 557
Amorphous phase, 295 Average error rates, 656

© Springer International Publishing Switzerland 2016 661

H. Ünlü et al. (eds.), Low-Dimensional and Nanostructured
Materials and Devices, NanoScience and Technology,
DOI 10.1007/978-3-319-25340-4
662 Index

Axial, 353 Biphasic TiO2, 448

Axicons, 176 Bismuth-doped zinc oxide, 401, 402
Axis of central maximum, 103 Bit lengths, 656
Bit streams, 652, 655
B Blend, 301
Bacterial biofilm, 599 Bloch functions, 14
π band, 227 Block copolymers, 510
d band, 228 Blue shift, 227, 336
Band bending, 311 Bode phase, 468
Band of frequencies, 64 Boron, 190
Band offsets, 40 Boron oxide, 182, 188, 189
Band structure, 453 Bottom-up approach, 356
Bandgap, 205, 221, 229, 364, 427, 455 Boundary condition of O operator, 619, 624
Bands (conduction, valence), 206 Bound state-one dimensional, 56
Bargmann coherent state, 611 Bovine Rhodopsin, 139
Barrier layer, 5 Breakthrough, 482, 483, 487
Batch, 495 Brookite, 426, 428, 430, 431, 445
Beer-Lambert law, 151 Brookite/anatase, 448
Benchmark circuits, 649 Bulk dielectric function, 212
Benchmark functions, 649 Bulk heterojunction, 380
Bernoulli distribution, 653 Bulk plasma, 213
Bessel beam, 175, 176 Bulks, 292
Bi-doped ZnO, 402, 403, 405
Bi-doped ZnO films, 402 C
Bilayer, 229 C4H6O6/TiCl3molar ratio, 432
Bilayer graphene, 217, 219, 229 Calcination, 301, 387
Binding, 269 Calcium phosphate, 537
Biocompatability, 543, 557, 559 3D calibration artefact, 117
in QDs, 562 Capacitance, 467
Biocompatible, 536 Capacitive, 466
Biocompatible silica, 560 Capacitor, 467
Bioconjugation, 514, 562 Capacity of storing quantum information, 627
Bioconjunction, 514 Carbodiimide chemistry, 518
Biodistribution, 520 Carbohydrates, 513
Bioimaging, 557 Carbon, 181–184, 187–193, 197, 199–202
Biointerfaces, 541 Carbon cluster, 188, 190, 199
Biological, 500 Carbon computer, 113
Biological labelling, 514 Carbon dimer, 184, 185
Biomacromolecules, 512 Carbon dots, 564
Biomarkers, 583 Carbon fibers, 462
Biomaterials, 543 Carbon monoxide, 190–192, 201
Biomedical applications of QDs, 573 Carbon nanotubes, 117, 239, 380, 387, 388
Biomelecular engineering, 147 Carbon trimer, 184, 188, 201
Biomimetic, 540 Cd-free RGD peptides, 568
Biomimicry, 548 Cd0.08Se0.92 quantum dots, 333
Biomineralization, 536 CdCl2treatment, 380
Biomolecule, 131 CdS.08Se.92, 334
Bionanotechnology, 565, 589, 591, 602 CdSe nanocrystals, 329
Bionsensor, 583 CdSe QDs, 572
Biophotonics, 539 CdSexS1−x Nanocrystals, 328
Biosensing, 246 CdSe/ZnS core shell, 507
Biosensors, 539 CdSO4 structure, 339
Biotechnology, 556 CdSTe/CdTe, 44
Biotin–streptavidin recognition, 563 CdS thin film, 328, 336, 338–340
Index 663

CdSxSe1−x, 332, 333 Contaminants, 252

CdTe/CdS, 328, 367, 378, 380, 390 Contamination, 182, 196
CdTe thin film, 329–331 Control, surface electron accumulation
CdX and ZnX (X = S, Se, Te) binaries, 38 in InN nanowires, 314, 320, 321, 323
3D cell culture, 598 Copolymer, 520
Cell labeling, 565 Copper, 182
Cell membrane mimics, 544 Copper-free click chemistry, 513
Certification of reference materials, 115 Core-shell, 353, 379
CFME, 462 Core-shell solar cell, 353
Chalcopyrites, 362 Correlation function, 613
Challenges, 424, 456 Coulomb coupled, 214, 216, 220
Change the band structure, 253 Coulomb potential, 214
I–V characteristic, 369 Counter electrode, 384, 387, 388
Charge collection, 376 Coupled plasmons, 215
Charge neutrality level, 311 Coupled plasmon spectrum, 217
Charge separation, 453 Covalent attachment, 513
Charge separation mechanism, 452, 454 Cr-TiO2, 451
Charge transfer (CT), 252, 571 Cr3+doping, 450
Charging, 263 CRET/BRET, 567
ChemFETs, 254 Critical distance, 205, 229
Chemical bath deposition (CBD), 335 Critical wavenumber, 213
Chemical vapor deposition, 182, 200, 202, 245 Cross-correlation, 121
Chemisorption, 509 Crossbar-type switches, 636
Chemotaxis, 599 3D culture systems, 598
Chiral vector, 240 Crystal growth, 360
Chromism, 293 Crystal phases, 425
Chronic diseases, 581 Crystalline, 295
Circuit, 472 Crystallite sizes, 455
Click chemistry, 513 Crystallization, 428, 487
Clinical trials, 583 Crystallographic phases, 293
Clinoptilolite, 488, 491, 492, 497 CT using QDs
Closed noise-free operator, 622 as donors/acceptors, 567
Cluster, 172 CuInSe2, 367, 390
CMOS, 181, 182, 193, 196 Cumulant expansion, 153
CNT chemiresistors, 254 Current density-voltage (J-V) characteristics,
CO2 photoreduction, 445 250
Coadsorbed, 284 Cut-off frequency, 86, 196
Colored and bleached tungsten oxide films, 305 Cyclic voltammetric measurements, 389
Combined uncertainty, 106 Cyclic voltammograms, 464
Compact QDs, 561
Complex Gaussian random variable, 611 D
Composite system, 215 Damping, 213, 228
Composition dependent bond lengths, 26 Data, 472
IV–VI, III–V and II–VI compounds, 10 D-band, 243
Concentration of SO42−, 434 Debye-Waller factor, 152
Conclusion, 456 Decoherence-free space, 627
Conducting polymers, 389 Defect calculations, 401, 403, 404, 410, 414,
Conducting screen, 84 419
Conducting substrate, 229 Defect distribution, 646
Conduction band offset, 42 Defect formation energy, 405, 406
Conductivity, 380 Defect maps, 648
Conductor, 220 Defect probabilities/rates, 643, 649
Conformational changes, 140 Defect tolerance, 636, 649
Consistency condition of O operator, 614 Defect transition energy, 406, 407, 418
664 Index

Degenerate case, 213 Drug delivery, 516, 581

Degradation of organic pollutants, 446 Drugs, 580
Delta-dot, 52 D-states, 29
Delta-prime, 49, 50 Dual, 640
Density functional theory (DFT), 1, 183, 184, Dual-modality imaging, 571
192, 193 Dye sensitized solar cells, 380, 388
Density functional theory modelling, 32 Dynamic, 212
Density of states, 241 Dynamics of quantum entanglement, 622
Density perturbation, 208 Dyson integral equation, 50
Density perturbation response, 61
Density perturbation response function, 209 E
Density plot, 89, 91–95 EDS, 337
Deoxidized, 280 Effective potential, 60, 208
Deposition rate, 184, 185 Efficiency, 480
Desorption, 184–189, 201, 492 Eigenenergy, 53
Deterministic, 659 Eigenenergy spectrum, 59
DFT+U method, 408, 410 Einstein frequency, 153
DFT with MBJLDA functional, 37, 46 EIS, 468
Diamondoid, 118 Elastic stiffness constants, 7
2,4-dichlorophenol, 450 Electric field, 263
Dielectric, 206 Electric micro-field, 342, 344
Diffraction, 73, 103 Electrical, 305
Diffraction zones, 88–90 Electrical conductivity, 248
Diffusion, 183–186, 190, 192, 193, 199, 201, Electrical measurements, 318
481 Electrochemical, 298, 356, 464
Dimensional metrology, 109 Electrochromic, 298
Dimensional nanometrology, 118 Electrodeposited, 465
3,4-(2,2-dimethylpropylenedioxy)thiophene, Electrolytes, 461
462 Electron-beam evaporation technique, 355
Diode, 638 Electron-hole continuum, 225
Diode behaviour, 374 Electron-hole excitations, 225
Dirac cones, 228 Electron acceptors, 428
Dirac delta function, 49, 74, 77, 207, 210 Electron accumulation, 311
Dirac-delta-function-derivative, 64 Electron beam energy, 224
Dirac-like energy spectrum, 229 Electron density, 319
Dirac point, 227 Electron energy loss spectroscopy (EELS),
Direct attachment, 518 221, 226
Directed evolution, 600 Electron propagation along wire, 58
Dispersion, 205 Electron transmission, 64
Dispersion relation, 54 Electron yield mode, 155
Distorted bond length, 11 Electronic properties, 309, 314
Divergent integral, 78 Electrophoresis, 595, 596
DMF, 470 Electropolymerisation, 464
DNA microarrays, 594 Electrospinning, 298
DNA sequencing, 602 Electrospun, 301
Doctor blade technique, 388 Electrospun tungsten oxide nanofibers, 299
Donor, 189, 253, 567 Emerging low-density circuits, 659
Doped, 218, 219 Empirical pseudopotential bandgaps, 17
Doping, 227, 228, 319, 321 Empirical pseudopotential bands, 31
Double indices, 646 Empirical pseudopotential calculations, 36
Double layer, 463 Encapsulation, 512
Double quantum dot, 50, 53 Energy, 345
Drawbacks, 433 Energy barrier, 184, 185, 187, 192, 195, 199
Drinking water, 495 Energy donors, 566
Index 665

Energy loss spectrum, 221 Fluidized bed, 498

Energy resources, 424 Fluorescence imaging, 516
Engineering metrology, 109 Fluorescence mode, 154
Enhanced ferromagnetism, 114 Fluorine-doped thin oxide (FTO), 377
Enhanced photocatalytic activity, 451 Fluorochromes, 557
Ensemble method, 120 Fly ash, 494
Entanglement transfer, 626 Foldable photoelectronics, 114
Environmental problems, 424 Forbidden condition, 622, 625
Epilayer, 8 Formal master equation, 615
Epitaxial Graphene, 221 Formation, 282
EPR, 453 Four-terminal, 638
Equation of motion of O operator, 620 Free-standing, 206
Equilibrium, 480 Free-standing graphene, 221, 228, 229
Equivalent, 472 Frequency dependence, 96, 97
Error free multiplication, 657 FRET, 523, 566
Error free stochastic computing, 657 applications, 525, 526
Error rates, 653, 655 donors, 523
Etchant concentration, 363 pair, 523
Evanescent wave, 133 Freundlich, 480
Evolution equation of O operator, 624, 632 FTIR-ATR, 466
Evolutionary engineering, 601 Fullerenes, 113
Exact master equation, 616, 628 Functional expansion of O operator, 618
Exchange-correlation functional, 33 Functional materials, 114
Excited state, 27 Functional switches, 646
Exciton, 384 Fundamental constants of nature, 107
Exciton binding energy, 371 Fusion tags, 570
Exhaustion, 483
Exhaustive search, 647 G
Expanded, 116 GaAsN/GaAs, 24, 43, 95–97, 102, 132, 355
Experimental, 473 Gallium arsenide, 206
Extended XAFS (EXAFS), 152 Gap, 206
External quantum efficiency, 158 Gapless graphene, 217, 229
Extreme UV nanolithography, 110 Gapless graphene polarizability, 217
Extrinsic graphene, 218, 219 Gapped graphene, 217, 220
Gas sensing, 250
F Gate potential, 227
FDA, 585 Gauge dependence, 57
Feature size in microchips, 110 GAUSSIAN, 136
Femtosecond, 173 Gaussian, 174
Femtosecond laser, 132 G-band, 242, 243
Fermi energy, 213, 227 Gene technology, 525
Fermi level, 5, 227, 382 Generalized functions, 49
Fermi level pinning, 311, 314, 323 Germanium, 182, 183, 195–197, 200
Fictitious excited s* orbital, 21 Germanium ad-atom, 202
Field amplitude distributions, 89 Germanium dimer, 198, 199
Field distribution, 85, 86, 88 Germline cells, 585
Field effect transistor (FET), 182, 194, 638 GGA, 270
Field of full sheet, 82 Glass substrates, 372
Fill factor, 365, 381, 387 Globally accepted standards, 122
Filtration, 498 GLU122, 146
First Brillouin zone, 18 Glycoconjugated polymer system, 513
First-principles, 266 G-Protein coupled receptors, 139
First principle ab initio methods, 12 Grüneisen parameter, 345
Fixed-bed, 482, 495 Grain radius, 345
666 Index

Graphene, 121, 178, 181, 182, 184, 186, 187, High-resolution electron energy loss
194, 196, 205–208, 213, 221, 228, 229, spectroscopy (HREELS), 205
239, 262, 385 High temperature, 439
Graphene band structure, 225 High-temperature calcination method, 438
Graphene bandgap, 217 H2 molecule adsorbed, 252
Graphene base transistor, 182, 186, 194–196 H2O, 267
Graphene/metal interface, 223, 226 H2O/Ti mole ratio, 433
Graphene nonlocality, 217 Homogenous thin fibers, 302
Graphene oxide (GOX), 262 Hopping terms, 22
Green’s function, 49, 51, 53, 56, 64, 67, Hot injection, 508
73, 85 H2 production, 442
Green synthesis methods, 507 HREELS, 223, 224
GROMACS, 136 HRTEM, 452
Group II–VI compounds, 25 HSE+U* calculations, 409, 410, 418
Group velocity, 223, 228 HSE+U approach, 409
Growth process, 319 Hubbard U, 408, 411
Growth time, 363, 374 Huisgen-type click chemistry, 513
Hybrid-functional+U approach, 408
H Hybrid heterostructure, 370
Halloysite, 494 Hybrid solar cell, 368, 369
Hamiltonian, 14 Hybridization, 228
Hamiltonian matrix, 22 Hydrazine, 358
Harmonic oscillator, 51 Hydrogen and energy storage, 246
Harrison Scaling Law, 20 Hydrogen bond dynamics, 134
Hartree Coulomb term, 33 Hydrogen bonds, 138
Hazard, 255 Hydrogen production, 441
HBAC, 137 Hydrophilic QDs, 508
Healthcare, 585 Hydrophobic nanocrystals, 560
Healthcare Industry, 110 Hydrostatic strain, 9
Heavy-hole, 8 Hydrothermal, 429, 432, 487
α-helices, 144 Hydrothermal synthesis, 558
Helicon Wave excited Plasma Sputtering Hydrothermal technique, 357, 371, 374
Epitaxy (HWPSE), 164 Hydroxyl groups, 278
Helium treatment, 445 Hyperfine splitting constant, 107
Helmholtz, 73, 74, 79
Helmholtz Green’s function, 103 I
Helmholtz scalar wave, 86 Image processing, 654
Heterepitaxial growth, 7 Imaginary part, 213
Heterointerface, 2, 161 Imaging, 514
Heterojuction bipolar transistor (HBT), 4 Immunostaining specificity, 562
Heterostructures, 1, 161 Impedance, 461
Type I, 2 Impenetrable barrier, 49
Type II, 2 In-Situ Low Temperature Optical Absorption
Type III, 3 Spectroscopy, 346
II–VI heterostructures, 42 In ternary and quaternary alloys, 2
III–V heterostructures, 42 In vitro imaging, 566
H2evolution efficiency, 443 In vitro targeting, 515, 518, 526
Hexagonal boron nitride, 182, 188, 188–190, In vivo, 526
195, 201 in vivo applications, 520
Hexamethylenetetramine, 372 In vivo imaging, 568
HF exposure, 249 in vivo vascular imaging, 522
Hierarchical structural architectures, 547 Incident field, 80
High-Affinity Fusion Tag Targeting Incident wave, 79, 88
Approaches, 569 Incident wave vector, 83, 87
Index 667

Infinite barrier model, 210 Lattice matched AlGaAs/GaAs, 42

Inflammation, 521 Lattice mismatch, 6, 19, 362
Influence factors, 429 Layer-surface separation, 229
InGaN/GaN, 24, 43 Layered titanate, 430
Initial condition of O operator, 618, 624 Layers, single and double, 205
InN, 309, 310 LCNAO, 267
InN thin films, 312, 314 LDPC codes, 654
Integral equation, 73–76, 103 Leachate, 478, 497, 498
Integrated nanosystems, 108 Ligand conjugated QDs, 569
Inter-laboratory comparison, 116 Ligand exchange, 508, 510
Interaction picture, 610 Light absorption, 354, 444
Interband, 222 Light amplification/photonics, 112
Interband plasmons, 222 Light emitting diodes (LEDs), 150
Interband transitions, 225 Light‐hole, 8
Intercalated, 228 Light trapping, 366
Intercalated graphene, 228 Lindblad master equation, 615
Interface strain, 6 Linear dispersion, 222
Interference fringes, 103 Linear medium, 207
Internal quantum efficiency, 159 Linear plasmon, 228
International Avogadro project, 107 Lipid bilayer, 140
International comparisons, 115 Liposomes, 584
Interstitial atom, 199 Local, 206, 213
Intraband excitations, 225 Local defects, 376
Intraband plasmons, 223 Local density approximation (LDA), 34, 269
Intrinsic band bending, 317 Long and short wavelengths, 221
Introduction, 239 Long wavelength limit, 229
Inverse dielectric function, 50, 59, 206, 208, Lotus effect, 542
211, 213, 214 Low dimensional, 49
Inverse dispersion factor, 215 Löwdin orbitals, 13
Inverse dispersion function, 218–220 Low frequency dynamics, 139
Inverted band structure, 314 Lower bound, 641
InxGa1−xN, 155
Ion-insertion, 298 M
ITRS roadmap, 124 M11, 242
Magnetic field, 56
K Magnetic nanoparticles, 110
Kinetic, 481 Magnetic switch-based structures, 638
Kohn-Hohenberg-Sham studies, 33 Magnetite, 537
Kohn-Sham, 263 Magnitude of Z, 468
k.p method, 13 Markers, 583
k-points, 225, 228 Markov limit, 617
Massless graphene, 213
L Master equation, 609
Lab-on-a-chip, 593 Matching, 646
Landau damped, 218–220 Maximally entangled state, 626
Landau damping, 217, 225, 229 MBJLDA functional, 11, 36
Landau quantization, 58 Mechanism, 451
Langmuir, 480 Mechanism model, 454
Langmuir-Blodgett deposition, 382 Mechanosynthesis, 112
Laser-assisted, 172 Memory time, 626
Laser ablation, 244 Memristive arrays, 638
Laser calibration, 117 Mesoporous, 296, 297
Laser diodes (LDs), 150 Metal-insulator-semiconductor (MIS) devices,
Lateral wave vector, 75 249
668 Index

Metal, 213 Molecular beam epitaxy, 2, 182, 183, 186–191,

Metal assisted chemical etching, 356 197, 200, 202
Metal based toxicity, 559 Molecular dynamic modelling, 123
Metal nanoclusters, 564 Molecular imprinting, 538
Metal nanoparticles, 357 Molecular metrology, 109
Metal organic chemical vapor deposition Monoclinic crystalline, 294
(MOCVD), 2 Monolayer, 220, 229
Metallic, 241 Monolayer graphene (MLG), 218, 220, 222,
Metallic tungsten precursor, 299 223, 226, 229
Methods, 429 Monte-Carlo-Simulations, 123
Methyl orange, 448 Monte Carlo modelling, 123
Metrological AFMs, 115 Montmorillonite, 491
Metrology, 105 Morphology, 474
MgZnO, 164 MOVPE, 160
Mica, 181–183, 190, 192–194, 201 MQW, 158
Microbial behaviour, 599 mRNA, 525
Microbial strain development, 600 Muffin-tin approximation, 152
Microbioreactors, 598 Multi-walled nanotubes, 241
Microchannels, 597 Multimodal approach, 516
Microchip capillary electrophoresis, 596 Multiphoton, 174
Microelectrophoretic separation, 595 Multiple-qubit system, 617, 628
Microemulsion-mediated solvothermal method, Multiplexed fluorescent assay, 518
434 Multiplexing, 523, 557
Microfluidic bioreactors, 597, 598 Multiwalled CNTs, 388
Microfluidic channels, 597
Microfluidic DNA microarray, 595 N
Microfluidic DNA microarray technology, 594 Na+, 430
Microfluidic PCR systems, 593 NaCl, 430
Microfluidic polydimethysiloxane (PDMS) Nano-biopsy, 584
bioreactor, 597 Nano-crossbar arrays, 635
Microfluidics, 589, 591, 592, 594, 599, 600, Nano-crossbar switching arrays, 659
602 Nano-hole, 73, 74, 80, 82, 83, 85, 86, 95–97,
Micropore, 485, 491, 501 102, 103
Microwave-assisted irradiation, 558 Nano-imaging, 584
Migration behavior, 456 Nano-scale, 76
Migration direction, 452 Nano-scale aperture, 76
Miniaturized 2-D gel electrophoresis, 596 Nanoantennae/metamaterials, 112
Minimization, 649 Nanobiomedical, 571
Mixed-phase TiO2, 423, 424, 440, Nanobot components, 112
441, 444 Nanocarriers, 571
MLG, 221, 224–226, 228 Nanochannels, 539
Mobility, 206 Nanocrystal radius, 334
Mobility direction, 451 Nanoelectronics, 246, 262
Model, 472 Nanofiber, 295, 299, 301
Model potential, 210 Nanogold, 109
Modelling, 49, 333 Nanomachines, 109
Modelling of electronic structure, 5 Nanomachining, 175, 176
Modification, 484, 493 Nanomaterials, 292
Modification of TiO2, 424 Nanomaterials embedded in complex matrix
Modified, 474 materials, 124
Modified Becke-Johnson density functional Nanomedicine, 572, 579, 585
(MBJLDA), 34 Nanometrology, 111
Modified virtual crystal approximation Nanoparticlate, 581
(MVCA), 10, 17, 24 Nanopharmaceuticals, 581
Index 669

Nanopillar arrays, 366 Nonlinear plasmon, 224, 225

Nanopillars, 351 Nonlocal, 206, 208, 212, 229
Nanoporous, 474, 477, 488, 501, 546 Nonlocal plasmon, 229
Nanoshells, 583 Nonlocality, 213
Nanostructures, 49, 172, 352 Nonparabolic dispersion, InN conduction band,
Nanosystems, 108 315
Nanotechnology, 580 Nonpolar m-plane, 159
Nanotheranostics, 584 Nonspecific binding, 570
Nanotube, 380 Nonspecific uptake, 520
Nanowire and nanotube crossbar arrays, 638 Novikov’s theorem, 615, 616, 621, 628
Nanowire growth, 317 NSET, 567
Nanowire growth mode, 320 Nucleation, 173, 181–185, 187, 188, 190, 199,
Nanowire structure, 320 201, 361
Nanowires, 2, 299, 390 Nudged elastic band, 183
Narrow band gap, 310 Number of 2D layers, 216
Natural adhesives, 540 Number of literals, 640
Nearest neighbor (NN) sp3 tight binding Number of products, 640
theory, 16 Numerical analysis, 86, 88
Nearest neighbor interaction, 30 N vacancy, 315, 316
Neurodegenerative diseases, 581 Nyquist plots, 468
Next generation QDs, 564
NH4OH, 430 O
N-(2-Hydroxypropyl) methacrylamide Obtain any probability, 658
(HPMA), 582 Off-diagonal matrix elements, 14
NIR, 522 OH/H, 268
Nitrification, 500 Ohmic type contacts, 377
NMA, 144 On-site atomic energies, 16
2NN and TB models, 28 One-dimensional nanostructures, 355
2NN sp3s* orbitals, 22 One-dimensional ZnO nanostructures, 401
2NN sp3s* orbitals sets, 36 One-to-one matching, 647
2NN sp3s* TB model, 23, 31 One qubit dissipative model, 616
2NN sp3s* TB parametrization, 26 Open circuit voltage, 365, 376, 381
2NN sp3s* tight binding parametrization, 22 O operator, 614, 619
2NN sp3 TB Hamiltonian matrix elements, 17 O operator coefficient function, 620
NN sp3d5 s* orbitals set, 27 O operator initial condition, 614
NN sp3d5 TB Hamiltonian matrix, 30 OPLS-All Atom, 136
NN sp3d5, 36 Opsins, 139
NN sp3d5 TB model, 31 Opsonins, 520
NN sp3d5 and 2NN sp3s* TB models, 37 Optical, 291, 305
NN sp3d5 tight binding model, 29 Optical absorption, 327, 336
NN sp3 tight binding parametrization, 29 Optical absorption tail, 343
NO2, O2, H2O, NH3, CH4, CO2, H2, N2, Ar, Optical coloration efficiency, 305
251 Optical dielectric constants, 42
NO2 or NH3, 250 Optical imaging agents, 113
Non-covalent interactions, 512 Optical phonons, 224
Non-diffracting, 177 Optical ruler imaging system, 121
Non-Markovian, 609 Optical Tweezers, 111
Non-Markovian master equation, 622 Optimal phase ratio, 454
Non-Markovian QSD equation, 614, 623 Optimal size, 642
Non-optimal, 641 Optimum phase content, 454
Non-parabolic dispersion, InN conduction Opto-electronic devices, 371
band, 316 Organic dyes, 447
Nonlinear dispersion, 225 Organic solar cells, 368
Nonlinear mode, 225 Ornstein-Uhlenbeck, 622
670 Index

Ornstein-Uhlenbeck correlation function, 613 Photogenerated free carriers, 376

Oscillator eigenfunctions, 53 Photoisomerisation, 140
Ostwald ripening, 506 Photoluminescence (PL), 168, 332, 333
Outlook, 456 Photoluminescence spectroscopy, 318, 320,
Overdamped, 217 322
Overlap matrix, 14 Photon Factory Advanced Ring, KEK, 160
Oxide nanoparticles, 256 Photovoltaic, 248, 352
Oxygen, 188–190, 192, 269 Photovoltaic behavior, 370
Oxygen deficiency, 293 PL spectra, 334
Planar boundaries, 210
P Plane wave-train, 83
P25, 441, 446 Plasma, 173
Pair of layers, 229 2D, 214
Passive nanostructures, 108 layer, 207
Pauli repulsion, 193 slab, 208
PBC method, 263 torch, 245
PC, 470 π plasmon, 228
PCBM, 368 Plasmon, 205, 206, 213, 220
p-d interactions, 27, 39 Plasmon dispersion, 64, 217, 225
p-doped, 250 Plasmon modes, 218, 219
PEDOT:PSS, 389 Plasmonic layer, 74, 76, 80, 95, 103
PEG, 511 Plasmonic response, 74
PEG polymer, 298 Plasmonics, 177
PEGylated QDs, 562 Plasmonic screen, 73
Peierls phase integral, 57 Plastic substrates, 255
Peptide conjugation, 517 3D plots, 89, 91–95
Peptide libraries, 537 p-n heterojunction, 365
Peptideconjugated QDs, 568 p-n junction, 249, 365
Perforated 2D layer, 79 Poisson ratio, 7
Perforated GaAs layer, 91, 92–94 Polarizability, 60, 73, 74, 76, 79, 207, 214
Perforated plasmonic layer, 74, 79, 85, 86, 96, Polarization, 221
97, 102 Polarization of gapped graphene, 231
Perforated plasmonic wave, 87 Poly(L-arginine), 518
Permeable, 498 Polymer, 296
Permutations, 647 Polymerase Chain Reaction (PCR), 591–593
Perovskite structure, 292 Polyvinylpyrrolidone (PVP), 299, 301
PET, 572 POPC, 143
Phase content, 436 Popular topic, 456
Phase transformation, 426, 438, 440 Population heterogeneity, 601
Phase transition, 427 Porous, 474
Phenol, 449 Porous anodic alumina, 354
Phonon confinement, 346 Positional broadening, 103
Phospholipid micelles, 512 Post-annealing, 363
Photo anode, 384, 387 Potassium coverage, 227
Photocatalysis, 423 Potential effect, 468
Photocatalytic, 441 Potentiodynamically, 465
Photocatalytic activity, 425, 427, 428 Power conversion efficiency, 248, 354
Photocatalytic decomposition, 447 20–30 ppb range, 256
Photocatalytic degradation, 448, 450 PProDOT-Me2, 462
Photocatalytic reduction activity, 445 Pretreatment, 485
Photocurrent, 370, 384 Printed/flexible electronics, 654
Photodetectors, 132 Probabilities, 652
Photo-generated carriers, 451 Probability generation tree, 658
Photo-generated electrons, 452 Probe-sample interaction, 123
Index 671

Problems, 456 RBSM, 655

Production of CO and CH4, 445 Reaction time, 430
Propan-2-ol, 135 Real time, 591
Proposed mechanism, 435 Real-time PCR, 592
Prostate cancer, 519 Realisation of the metre, 123
Protein-based computer, 108 Recombination, 424
Protein activation, 139 Reconfigurable crossbar architectures, 638
Proteins, 138 Rectification ratio, 365, 369
Pseudomorphic InGaAs/GaAs heterostructures, Red-shift, 227, 330
19, 42 Red shifted, 346
p-type doping, 226, 323 Reduced density matrix, 615
PVD, 329 Reduced matrix, 646
PW, 262 Reduction of CO2 with water, 444
Reflected wave, 88
Q Regeneration, 487
QD—antibody conjugates, 521 Relationship, 435
QD toxicity, 560 Relativistic Dirac electrons, 206
QDs bioconjugates, 515 Removal, 480
QDs with antibodies, 516 Repeatability, 116
Quantitative reverse transcription polymerase Representation, 105
chain reaction (qRT-PCR), 591 Resistance, 474
Quantized energy levels, 5 Resonant Raman scattering, 340
Quantum-confined Stark effects (QCSEs), 159 Responsivity, 379
Quantum confinement effect, 333 Reversible coloration ability, 303
Quantum dot, 2, 50, 390, 505 Review, 423, 425
three dimensional, 51 Ring diagram, 208, 231
wires wells, 50 2D, 214
Quantum effect, 506 RNA technology, 525
Quantum mechanical amplitude, 69 Rolled, 239
Quantum mechanical perturbation theory, 151 Room temperature, 253
Quantum state diffusion equation, 609 RPA, 207, 214
Quantum systems, 49 2D RPA, 231
Quantum trajectory, 611 2D RPA density perturbation, 229
Quantum well, 5, 390 ΔR/R0, 255
Quantum wire, 50, 56 Rutile, 426, 436, 437
Quantum yield, 507
Quasi-Fermi level, 384 S
Quasi-freestanding, 228 S11, 242
S22, 242
R Saccharomyces cerevisiae, 601
Radial, 353 Scalar field, 79, 85
Radial breathing mode, 242 Scalar Green’s function, 74, 76
Radial distribution functions, 156 Scalar wave, 73, 74, 86, 103
Raman, 242, 318 Scanning tunneling microscope, 275
Raman shift, 345 Schrödinger Green’s function, 50
Raman spectra, 295 Schrodinger equation, 33
Raman spectroscopy, 249, 331, 340, 345 Screening, 223
Random bit assigning, 654 Screening function, 59, 206
Random bit shuffling, 655 Seed layer, 372
Random phase approximation (RPA), 60, 208, Self-assemble, 547
229 Self-assembled nanoscale systems, 636
Random walk theory, 185–187, 189, 201 Self-assembly, 512, 536, 544
Rate constants, 440 Self-cleaning, 542, 546
RBAM, 655 Self-doped graphene, 224
672 Index

Self-formation, 172, 173 Size distribution, 455

SEM, 113, 337, 347, 377, 474 Slab plasma, 206, 214
SEM and TEM images, 438 Slater-Koster formalism, 13
II–VI semiconductors, 360 S-layer proteins, 547
III–V semiconductors, 360 Small angle X-ray scattering diffractometer,
Semi-infinite conductor, 217–219 120
Semi-infinite limit, 212 SnO2, 256
Semi-infinite slab, 213 SO42−concentration, 434
Semi-infinite substrate, 216 Sol-gel method, 436, 508
Semiconductor, 49, 150, 207 Solar cell, 351, 369, 539
Semi conductor nanocrystals, 556 Solar energy, 352
Semiconducting, 241 Solar simulator, 376
Semiconductor devices, 2 Solar spectrum, 362, 365, 378
Semiconductor Industry, 109 Solid-solution-solid mechanism, 358
Semiconductor nanocrystals, 328 Solution-based techniques, 356
Semiempirical NN sp3d5 TB model, 46 Solution concentration, 430
Semiempirical Sp3 tight binding modeling, 13 Solution temperature, 363
Semiempirical Sp3d5 tight binding modeling, Solvent mixing and calcination method, 437
30 Solvents, 461
Semiempirical Sp3s* tight binding modeling, Solvothermal, 432
21 Sonication, 383
Semiempirical tight binding theory, 1 Sonochemical technique, 357
Sensitivity, 255 Sort matrices, 645
Series resistance, 376 Source, 80
S*-excited state, 31 Sp3d5 s* orbitals set, 29
Sheet resistances, 382 Sp3d5 orbital basis, 31
Shift the binary value, 658 Space elevator, 108
Shift the Fermi level, 253 Spatial compression, 103
Short circuit current, 365 SPE interband, 225
Short circuit current density, 366, 376 SPE intraband, 225
Short current density, 381 Spectroscopic ellipsometers, 113
Shunt resistance, 376 Spectroscopy, 467
SI microscopic representations, 107 Speed of generating quantum entanglement,
Si nanowires, 368 626
Sidewall electron accumulation, 317 Spin coating, 369
SIESTA, 262 Spin orbit coupling of p-states, 30
Silanization, 509 Split-off valence band, 8
Silicene, 206–208, 229 SPring-8, 156
Silicon, 181, 182, 186, 187, 206 Sputtering technique, 377
Silicon carbide (SiC), 182, 190, 192, 193, 199, Square-root-like dispersion, 221
201 SQW, 155
Silicon dioxide, 181–183, 190–193, 196, 197, Stability, 426, 449
199, 201 Standard, 297
Silicon dots, 565 Standard Berendsen, 144
Silicon oxide layer, 249 Standard heat capacity, 9
Silicon, zinc-oxide, 361 Standard Reference Materials (SRM’s), 121
Simulation, 473, 649 S* state, 29
Single-cell characterization and analysis, 602 π states, 228
Single-particle electronic orbitals, 32 Static shielding, 60
Single-particle excitation (SPE), 225 Statistical average, 612
Single-walled carbon nanotubes, 387 Statistical thermodynamic model, 9, 21
Single cell genomics, 602 Step height standard, 115
Single walled carbon nanotube, 241 Sterically hindered, 138
Size- and shape-dependent energetics, 413 Sticking coefficient, 182, 186, 187, 189, 201
Index 673

Stober approach, 510 Synthesis results, 649

Stochastic computing, 636, 652 Synthetic, 485
Stochastic computing models, 659
Stochastic multiplication, 653 T
Strain, 343 Target function, 642
Strain effects, 20 Targeting of drugs, 110
Strong coupling, 610 Tartaric acid, 432
Structural, 291, 298, 305 Telomerisation, 525
Structural disorder, 343 TEM, 378
Structural materials, 247 TEM images, 445
Structural properties, 36 Terahertz spectroscopy, 131
Structure factor, 63 Ternary/binary heterointerfaces, 43
Structure properties, 425 Ternary semiconductor, 11
Structured medium, 610 Tethered bilayer membranes, 545
Stuck-closed, 643, 649 Tetrabutylammonium hexafluorophosphate,
Stuck-open, 643, 649 464
Sturm-Liouville, 64 Tetrabutylammonium tetrafloroborate, 464
Sub-division of optical laser wavelengths, 123 Tetraethylammonium hexafluorophosphate,
Sub-nanometre step heights, 119 464
Sub-nm precision, 118 Tetrahedral semiconductors, 16
Sub-nm scales, 107 Theranostic, 583
Subarray, 646 Thermal equilibrium state, 612
Substitutional atom, 188, 190 Thermal evaporation, 361
Substrate, 208 Thermal expansion, 6
Substrate plasma, 206, 214 Thermodynamic Green’s function, 208
Subwavelength, 73, 74, 79, 86, 103 Thick slab, 210, 212
Sum-of-products (SOP), 640 Thickness of the film, 342
Super insulation, 114 Thin film, 175, 177, 292, 295, 327, 351, 376
Supercapacitors, 463 Thiolated molecules, 525
Supercell, 263 Thiolated proteins, 518
Supercell approach, 401, 403, 404 Third-generation solar cells, 353
Superhydrophobic, 542 Three-qubit dissipative model, 623
Superlattice, 49, 59, 157 THz emitter, 133
Supramolecular chemistry, 538 Tight binding, 242
Surface, 205 Tight binding method, 12
Surface charge accumulation on InN Tight binding model with an sp3d5 orbital
Nanowires, 317–319 basis, 38
Surface electron accumulation, 309, 310, 312, Time-local, 628
314–316 TiO2, 256, 378, 424
Surface modification, 508, 560 TiO2nanofiber, 385
Surface morphology, 188–190, 197, 200, 202 TiO2nanorods, 376
Surface passivation, 367 Tissue engineering, 113
Surface passivation treatment, 355 Titania nanoparticles, 385
Surface plasmon, 213 Titanium-dioxide, 361, 376
Surface steps, 185, 188–190 Titanium nitrides, 389
Surface-to-layer separation, 218, 219 TM6 helix, 146
Surfactant, 298, 382 Top-down approach, 356
Switching crossbar nanoarray, 637 Topological insulators, 229
Symmetric/antisymmetric modes, 205 Total reflection, 84
Synchrotron light sources, 149 Toxicity, 585
Synchrotron radiation beam, 120 Toxicological studies, 559
Synergistic effects, 442 Traceability, 105
Synthesis, 429, 506 Traceability to the lattice constant, 123
Synthesis methods, 506 Traceable measurement techniques, 113
674 Index

Transfer artefacts, 106 van der Waals (vdW), 267

Transmembrane, 139 van Hove singularities (vHs), 242, 248
Transmission, 73, 74, 79, 103, 364 Vanadium doping, 448
Transmission maximum, 103 Vapor-liquid-solid (VLS), 359
Transmission mode, 154 Vapor-phase based synthesis, 359
Transmitted field, 88 Vapor-solid (VS), 359
Transmitted field distribution, 91–102 Variable mass, 64
transmitted field intensities, 89 VASP, 262
Transmitted scalar wave, 86 vdW forces, 187
Transmitted wave, 83 Virtual crystal approximation (VCA), 10
Transparent conductive oxide, 369 Visible, 369
Trapping viruses, 110
Treatment, 581 W
Triphasic TiO2, 448, 449 W–O–W mode, 301
Truoiller-Martins, 263 Wafer inspection, 110
Truth table, 642 Wafer transfer artefacts, 115
Ttransport measurements, 322 Wastewater, 497
Tungstatic acid, 299 Water, 135
Tungsten hexachloride, 301 Water solubility, 512
Tungsten oxide, 291, 295, 297 Wave packet transmission, 69
Tungsten trioxide, 292 Wave propagation, 73, 74, 103
Tunneling of electrons, 194 WCl6/PVP blend, 301
Two dimensional (2D), 74 Wet-chemistry, 382
Two dimensional electron gas (2DEG), 4 Wetland, 500
Two dimensional plasmon, 103 WIEN2K simulations package, 11, 34, 46
Two-qubit system, 619 WKB approximation, 341
Two-step procedures, 353 WO3, 256
Type 1 superlattice, 50 Wurtzite, 157
Types and characteristics of QD, 556
X
U Γ, X and L symmetry points, 38
Ultrasonic waves, 358 X-ray absorption fine structure (XAFS), 149
Ultra-violet, 369 X-ray absorption near edge structure
Uncertainty, 106, 116 (XANES), 153
Undamped, 218, 219 X-ray diffraction (XRD), 339, 347, 426
Undamped plasmons, 220 X-ray fringes, 119
Undamped plasmon spectrum, 205 X-ray interferometry, 119
Undistorted bond length, 11 X-ray reciprocal space mapping (XRSM), 160
Upper bound, 641 XANADU, 152
Upper plasmon branch, 220 XPS, 252
Urbach, 345 XRD pattern, 294
Urbach energy, 335, 342
Urbach rule, 341 Z
Urbach tail, 341, 342 Zeolite, 477, 480, 484, 485, 493, 501
Urea concentration, 435 Zero bandgap, 218, 219
Use of quantum dots in biological systems, 557 Zero radial size, 78
UV/Vis/NIR absorption spectroscopy, 242 Zero temperature, 231
UV illumination, 296 Zinc-blende crystal structure, 34
Zinc nitrate hexahydrate, 372
V Zinc oxide varistors, 402
Vacancy defects, 279 Zinc oxide (ZnO), 371, 372
Valence and conduction bands, 215 ZnO nanowires, 401–405, 412, 413
Valence band edge, 321 ZnO thin film, 374
Valence band offset, 4, 41 ZnSSe/ZnSe, 44