Ferroic Materials For Smart Systems From Fundamentals To Device Applications PDF

Ferroic Materials for Smart Systems
Ferroic Materials for Smart Systems
From Fundamentals to Device Applications
Jiyan Dai
Author All books published by Wiley-VCH
are carefully produced. Nevertheless,
Prof. Jiyan Dai authors, editors, and publisher do not
The Hong Kong Polytechnic University warrant the information contained in
Department of Applied Physics these books, including this book, to
Hung Hom be free of errors. Readers are advised
Kowloon to keep in mind that statements, data,
Hong Kong illustrations, procedural details or other
items may inadvertently be inaccurate.
Cover
© antoniokhr/Getty Images; Library of Congress Card No.:
(graph) Courtesy of Professor Jiyan Dai, applied for
The Hong Kong Polytechnic Univeristy
British Library Cataloguing-in-Publication
Data
A catalogue record for this book is
available from the British Library.
Bibliographic information published by

the Deutsche Nationalbibliothek
The Deutsche Nationalbibliothek lists
this publication in the Deutsche
Nationalbibliografie; detailed
bibliographic data are available on the
Internet at <http://dnb.d-nb.de>.
© 2020 Wiley-VCH Verlag GmbH &

Co. KGaA, Boschstr. 12, 69469
Weinheim, Germany
All rights reserved (including those of

translation into other languages). No
part of this book may be reproduced in
any form – by photoprinting,
microfilm, or any other means – nor
transmitted or translated into a
machine language without written
permission from the publishers.
Registered names, trademarks, etc. used
in this book, even when not specifically
marked as such, are not to be
considered unprotected by law.
Print ISBN: 978-3-527-34476-5

ePDF ISBN: 978-3-527-81534-0
ePub ISBN: 978-3-527-81537-1
oBook ISBN: 978-3-527-81538-8
Typesetting SPi Global, Chennai, India

Printing and Binding
Printed on acid-free paper
10 9 8 7 6 5 4 3 2 1
v
Contents
1 General Introduction: Smart Materials, Sensors, and

Actuators 1
1.1 Smart System 2
1.2 Device Application of Ferroelectric Materials 5
1.2.1 Piezoelectric Device Applications 6
1.2.2 Infrared Sensor 7
1.2.3 Ferroelectric RAM (FeRAM) 8
1.3 Device Application of Ferromagnetic Materials 9
1.3.1 Spin-Transfer Torque Memory 9
1.3.2 Magnetic Field Sensor Based on Multiferroic Device 9
1.4 Ferroelastic Material and Device Application 10
1.5 Scope of This Book 12
References 13
2 Introduction to Ferroelectrics 15
2.1 What Is Ferroelectrics? 15
2.1.1 P–E Loop 15
2.1.2 Relationships Between Dielectric, Piezoelectric, Pyroelectric, and
Ferroelectric 16
2.1.2.1 Ferroelectric–Dielectric 16
2.1.2.2 Ferroelectric–Piezoelectric 17
2.1.2.3 Ferroelectric–Pyroelectric 18
2.2 Origin of Ferroelectrics 18
2.2.1 Structure-Induced Phase Change from Paraelectric to
Ferroelectric 18
2.2.2 Soft Phonon Mode 19
2.3 Theory of Ferroelectric Phase Transition 21
2.3.1 Landau Free Energy and Curie–Weiss Law 21
2.3.2 Landau Theory of First-Order Phase Transition 23
2.3.3 Landau Theory of a Second-Order Phase Transition 26
2.4 Ferroelectric Domains and Domain Switching 28
2.4.1 Domain Structure 28
2.4.2 Ferroelectric Switching 28
2.5 Ferroelectric Materials 29
vi Contents
2.5.1 From BaTiO3 to SrTiO3 29

2.5.2 From PbTiO3 to PbZrO3 32
2.5.3 Antiferroelectric PbZrO3 33
2.5.3.1 Pb(Zrx Ti1−x )O3 (PZT) 35
2.5.4 Relaxor Ferroelectrics 36
2.5.4.1 Relaxor Ferroelectrics: PMN-xPT Single Crystal 37
2.5.4.2 Polar Nano Regions 38
2.5.4.3 Morphotropic Phase Boundary (MPB) of PMN-PT Crystal 40
2.6 Ferroelectric Domain and Phase Field Calculation 41
References 42
3 Device Applications of Ferroelectrics 47

3.1 Ferroelectric Random-Access Memory 47
3.2 Ferroelectric Tunneling Non-volatile Memory 50
3.2.1 Tunneling Models 51
3.2.2 Metal–Ferroelectric–Semiconductor Tunnel Junction 55
3.2.3 Ferroelectric Tunneling Memristor 56
3.2.4 Strain Modulation to Ferroelectric Memory 57
3.3 Pyroelectric Effect and Infrared Sensor Application 58
3.3.1 Pyroelectric Coefficient 58
3.3.2 Pyroelectric Infrared Sensor 59
3.3.3 Pyroelectric Figures of Merit 60
3.4 Application in Microwave Device 63
3.5 Ferroelectric Photovoltaics 65
3.6 Electrocaloric Effect 67
References 68
4 Ferroelectric Characterizations 73
4.1 P–E Loop Measurement 73
4.2 Temperature-Dependent Dielectric Permittivity Measurement 76
4.3 Piezoresponse Force Microscopy (PFM) 77
4.3.1 Imaging Mechanism of PFM 77
4.3.2 Out-of-plane Polarization (OPP) and In-plane Polarization (IPP)
PFM 80
4.3.2.1 Electrostatic Force in PFM 83
4.3.2.2 Perspectives of PFM Technique 84
4.4 Structural Characterization 86
4.5 Domain Imaging and Polarization Mapping by Transmission Electron
Microscopy 87
4.5.1 Selected Area Electron Diffraction (SAED) 88
4.5.2 Convergent Beam Electron Diffraction (CBED) for Tetragonality
Measurement 91
References 92
5 Recent Advances in Ferroelectric Research 95

5.1 Size Limit of Ferroelectricity 95
5.2 Ferroelectricity in Emerging 2D Materials 96
Contents vii
5.3 Ferroelectric Vortex 99

5.4 Molecular Ferroelectrics 104
5.5 Ferroelectricity in HfO2 and ZrO2 Fluorite Oxide Thin Films 106
5.6 Ferroic Properties in Hybrid Perovskites 114
References 117
6 Piezoelectric Effect: Basic Theory 123

6.1 General Introduction to Piezoelectric Effect 123
6.2 Piezoelectric Constant Measurement 124
6.2.1 Piezoelectric Charge Constant 125
6.2.2 Piezoelectric Voltage Constant 126
6.2.3 Dielectric Permittivity 127
6.2.4 Young’s Modulus (Elastic Stiffness) 127
6.2.5 Elastic Compliance 127
6.2.6 Electromechanical Coupling Factor 128
6.2.6.1 How to Measure Electromechanical Coupling Factor? 129
6.3 Equivalent Circuit 132
6.4 Characterization of Piezoelectric Resonator Based on a Resonance
Technique 135
6.4.1 Length Extensional Mode of a Rod 135
6.4.2 Extensional Vibration Mode of a Long Plate 138
6.4.3 Thickness Shear Mode of a Thin Plate 139
6.4.4 Thickness Mode of a Thin Disc/Plate 140
6.4.5 Radial Mode in a Thin Disc 141
6.4.6 Mechanical Quality Factor 141
References 141
7 Piezoelectric Devices 143

7.1 Piezoelectric Ultrasonic Transducers 143
7.1.1 Structure of Ultrasonic Transducers 143
7.1.2 Theoretical Models of Ultrasonic Transducer (KLM Model) 145
7.1.3 Characterization of Ultrasonic Transducers 147
7.1.3.1 Bandwidth (BW) 147
7.1.3.2 Sensitivity of the Transducer 148
7.1.3.3 Resolution 148
7.1.4 Types of Ultrasonic Transducers 149
7.1.4.1 Medical Application 149
7.1.5 Piezoelectric Film Application in Ultrasound Transducers 149
7.1.6 Challenges and Trend of Developing New Advanced Ultrasound
Transducers 150
7.2 Ultrasonic Motor 150
7.2.1 Terminologies 151
7.2.2 Design of USM 153
7.3 Surface Acoustics Wave Devices 154
7.3.1 Interdigital Electrode in SAW Device 155
7.3.2 Acoustic Wave 155
7.3.3 Piezoelectric Property Considerations for SAW Devices 157
viii Contents
7.3.4 Characterization of SAW Devices 159

7.3.5 Lead-Free Piezoelectric Materials 161
References 163
8 Ferromagnetics: From Material to Device 165

8.1 General Introduction to Ferromagnetics 165
8.2 Ferromagnetic Phase Transition: Landau Free-Energy Theory 168
8.3 Domain and Domain Wall 169
8.4 Magnetoresistance Effect and Device 171
8.4.1 Anisotropic Magnetoresistance (AMR) 171
8.4.2 Giant Magnetoresistance (GMR) 172
8.4.3 Colossal Magnetoresistance (CMR) 175
8.4.4 Tunneling Magnetoresistance (TMR) 176
8.4.4.1 Spin-Transfer Torque Random-Access Memory (STT-RAM) 177
8.5 Magnetostrictive Effect and Device Applications 178
8.5.1 Magnetostrictive Properties of Terfenol-D 180
8.5.2 Magnetostrictive Ultrasonic Transducer 183
8.5.3 Magnetoelastic Effect 184
8.5.3.1 Magnetomechanical Strain Gauge 185
8.6 Characterizations of Ferromagnetism 186
8.6.1 Vibrating Sample Magnetometer (VSM) 186
8.6.2 Superconducting Quantum Interference Device (SQUID) 187
8.6.3 Magnetic Force Microscopy (MFM) 188
8.6.4 Magneto-Optical Kerr Effect (MOKE) 189
8.7 Hall Effect 191
8.7.1 Ordinary Hall Effect 191
8.7.2 Anomalous Hall Effect 191
8.7.3 Spin Hall Effect 192
References 193
9 Multiferroics: Single Phase and Composites 197

9.1 Introduction on Multiferroic 197
9.2 Magnetoelectric Effect 199
9.3 Why Are There so Few Magnetic Ferroelectrics? 199
9.4 Single Phase Multiferroic Materials 200
9.4.1 Switching Mechanism in BFO Films 204
9.5 ME Composite Materials 205
9.6 Modeling the Interfacial Coupling in Multilayered
ME Thin Film 207
9.6.1 PZT/CFO Multilayered Heterostructures 207
9.6.2 Ferroelectric Properties of PZT/CFO Multilayers 209
References 212
10 217
Device Application of Multiferroics
10.1 ME Composite Devices 217
10.1.1 Effect of Preload Stress 221
10.2 Memory Devices Based on Multiferroic Thin Films 223
Contents ix
10.3 Memory Devices Based on Multiferroic Tunneling 224

References 229
11 Ferroelasticity and Shape Memory Alloy 231

11.1 Shape Memory Alloy 231
11.1.1 SMA Phase Change Mechanism 232
11.1.2 Nonlinearity in SMA 233
11.1.3 One-Way and Two-Way Shape Memory Effect 235
11.1.4 Superelastic Effect (SE) 235
11.1.5 Application Examples of SMAs 236
11.2 Ferromagnetic Shape Memory Alloys 237
11.2.1 Formation of Twin Variants 238
11.2.2 Challenges for Ni–Mn–Ga SMA 242
11.2.3 Device Application of MSMA 243
References 244
Index 247
1
General Introduction: Smart Materials, Sensors, and

Actuators
The early twenty-first century has foreseen acceleration of innovations in

robotics and automations as well as artificial intelligence (AI), where sensors/
transducers and smart materials play very important roles. The concept of AI
has been around since the late 1950s; however, it’s only since the first decade of
the twenty-first century that excitement about it has really begun to grow due
to the ability of fast computation and abundant size of memory devices. A very
successful demonstration of AI is Google’s AlphaGo, which is the first computer
program to defeat a professional human Go player (see Figure 1.1). Another
successful application of AI is the unmanned vehicles and aircrafts where large
number of sensors and actuators are used.
One may ask, what is the relation between the key word “ferroic materials” of
this book and the mentioned robotics, automations, and AI? The answer is that
these smart and intelligence systems rely on large amount of data from sensors
and memories for machine learning and actuators for close-looped feedback con-
trol systems; and among these sensors, actuators, and memories, ferroic materials
play very important roles.
For example, the piezoelectric property of a ferroelectric material (one of the
most typical ferroic materials) can be used for ultrasound sensors to detect dis-
tance of your car from a wall for auto parking system. A ferroelectric material can
be used as the functional element for many kinds of sensors from pressure sensor
to acceleration sensor, infrared sensor, etc. Beyond that, a ferroelectric polariza-
tion and its switching can also be used in memory devices such as ferroelectric
random-access memory (FeRAM) where the ferroelectric layer acts as gate insu-
lator in a field-effect transistor (FET) structure. Ferroelectric tunneling-based
resistive random-access memory (RRAM) has also been demonstrated, and such
ferroelectric-based memory has been shown to be able to perform as an artificial
synapse. More interestingly, artificial neural networks (ANNs) based on these
ferroic synapses can realize brain-like computing and AI functions such as image
recognition. As shown in Figure 1.2, synapses with BiFeO3 (BFO) ferroelectric
layer has been successfully demonstrated.
This book will tell you fundamentals and characterization methods of ferroic
materials, physics, and technologies behind ferroic device design and applica-
tions as well as their recent advances.
Ferroic Materials for Smart Systems: From Fundamentals to Device Applications,

First Edition. Jiyan Dai.
© 2020 Wiley-VCH Verlag GmbH & Co. KGaA. Published 2020 by Wiley-VCH Verlag GmbH & Co. KGaA.
2 1 General Introduction: Smart Materials, Sensors, and Actuators
Figure 1.1 AI beats human chess player.
Synapse
Nucleus
Axon
Axon
Memristor Input
Co
BFO Synapses
CCMO To pre-neuron
To post-neuron Output
Figure 1.2 A cross-bar structure of synapse and artificial neuron networks based on a
cross-bar structure. Ferroelectric thin films such as BiFeO3 can be used as the junction material.
Source: Adapted from Boyn et al. (2017).
1.1 Smart System

A smart system, such as a self-driving car or remote-control aircraft, is a sys-
tem that relies on sensors and actuators to realize instant feedback of controlled
variables (CVs) such as speed, height, etc. The basic component of a smart sys-
tem usually contains sensors, actuators, and control system. An intelligent smart
system needs large amount of data processing and memories, while ferromag-
netic and ferroelectric materials have been implemented in realizing non-volatile
memristors. Beyond that, memories based on ferroelectric thin films may also
find application in electronic synapses as building blocks toward building ANN.
1.1 Smart System 3
Proportional
+ Integral Work
Sensor
Derivative
Actuator
(a) (b)
Figure 1.3 (a) Photo of a remote control copter and (b) diagram of a PID feedback control
system where sensors and actuators are implemented.
As an example, a smart system of remote control copter relying on

proportional–integral–derivative (PID) feedback control system is shown
in Figure 1.3. PID is a three-order feedback control system that has been widely
used in auto driving vehicles and auto-pilot airplanes to make the dynamic
system operate smoothly or being stable during video imaging. Equation (1.1)
illustrates the mechanism of PID control where three terms of proportional gain
(P), integral gain (I), and derivative feedback (D) can provide instant response to
cure the error (E) between the set point (speed of vehicle or height of the copter)
and controlled variable (CV):
t
dCV (t)
MV (t) = PE(t) + I E(t ′ )dt ′ − D (1.1)
∫0 dt
In this feedback control system, we can find applications of ferroic materials, for
example, piezoelectric-based gyroscope, surface acoustic wave device for wireless
communication, and ferroelectric-based infrared detector. In gyroscope, rotation
and acceleration can be sensed by measuring induced voltage generated by piezo-
electric effect; a surface acoustic wave device based on piezoelectric effect is used
for communication band selection; and a ferroelectric-based infrared detector
can be used as an intruder sensor making the copter able to find people for res-
cue mission. Figure 1.4 shows the finite element modeling (FEM) simulation of
three resonant motions in a Pb(Zrx, Ti1−x )O3 (PZT)-based gyroscope and photo
of a fabricated gyroscope (Chang and Chen 2017).
Sensors: Devices that can “sense” a change in some physical characteristics and
perform an electrical input function are commonly called sensors. For example,
a strain sensor converts mechanical strain into electrical signal.
Actuators: Devices that perform an output function are generally called actua-
tors. An actuator can be utilized to control some external moduli or output
mechanical movement such as ultrasonic wave. For example, an atomic force
microscope (AFM) uses piezoelectric actuators to realize scanning along three
directions.
Transducers: Both sensors and actuators are collectively known as transducers
because they are used to convert energy of one kind into energy of another
kind. Transducers can be used to sense a wide range of different energy forms
Y
MEMS scanner
X Horizontal scan
Z Vertical
scan
PZT on stainless steel
1 mm
Gyro sensors 4 mm
Resonant in Resonant in Resonant in
Y-direction Y-direction X-direction
Figure 1.4 FEM simulations of three resonant motions in a PZT-based gyroscope and photo of
a fabricated gyroscope. Source: Adapted from Chang and Chen (2017).
such as movement, electrical signals, thermal or magnetic energy, etc. The type
of input or output transducers being used really depends on the type of signal
or process being “sensed” or “controlled,” but we can define a transducer as a
device that converts one physical quantity into another.
A smart system needs sensors and actuators to realize the sensing functions
such as distance, movement, and acceleration as well as actions. These sensors
and actuators use smart materials to realize the conversion between different
energies and moduli to electrical signals such as voltage, current, and capaci-
tance. Of course, many sensor devices are made of semiconductors such as the
FET, but this is not the focus of this book.
“Smart material” is a very large concept, in fact, there is no stupid material (a
joke), i.e. all materials are smart in some way since they all have their own prop-
erties and response to external stimuli. But in this book, we restrict the “smart
materials” to those materials with “ferroic” characteristics. We focus on basic
physics, materials science, structures, devices, and applications of ferroic mate-
rials for smart systems. The ferroic materials are usually classified as possessing
one of the followings based on coupling of stimuli:
(i) Ferroelectric, which is also piezoelectric when electromechanically coupled
and pyroelectric when thermoelectrically coupled.
(ii) Ferromagnetic, which is also magnetostrictive when magnetomechanically
coupled.
(iii) Ferroelastic, which also includes shape memory when thermomechanically
coupled.
Among these ferroics, we can see that strain, electric polarization and magne-
tization, and their interplay or coupling are involved. We call a material as ferroic
material if it possesses at least one of the properties of ferroelectric, ferromag-
netic and ferroelastic.
If we look at the diagram shown in Figure 1.5, we can see that the coupling
and interplay between electricity, mechanics, magnetism, heat, and optics result
in many smart functions, such as ferroelectric, piezoelectric, pyroelectric, ferro-
magnetic, electromechanical, etc. One book cannot cover all of them, but those
belong to ferroic materials and devices especially in the form of thin films will be
Electricity
E
ici
ty Ma lectr
Elec
tr
ic
gn om
lec ion
alor
eto ag
oe ict ele ne
tro-o
ez str ctr tism
troc
Pi ctr
o
ici
e
ptic
El ty
Elec
Mechanics El Magnetism
as Magnetostriction
to-
op ric
tic
s c alo
to
optic
The
Elas oelastic
ne
ag
rm
M
toca
eto-
elec city
Opto tovoltai
tricit
Pho
i
n
The roelectr
loric
ism
Mag
elec
net
ag
rmo
tricit
om
Py
erm
y
cs
Th
Thermo-optics
Heat Optics
Figure 1.5 Diagram showing coupling between different moduli and the clarification of smart
materials.
extensively introduced in this book. Before going into details, some application
examples of ferroic materials in smart systems are given in this chapter.
1.2 Device Application of Ferroelectric Materials

When people talk about applications of ferroelectric materials, the first thing
jumps out is most possibly the PZT (lead–zirconium–titanate with chemical
formula Pb(Zrx Ti1−x )O3 ), which is known as an excellent piezoelectric material.
As the most popular ferroelectric, PZT is also the most important piezoelectric
material in commercial applications. Piezoelectric materials have very broad
applications in many fields, from medical ultrasound imaging to ultrasonic
wire bonding machine in semiconductor industry, from pressure sensors to
accelerometer, etc. The market size of piezoelectric materials is more than US
$1 billion now and is expected to be US $1.68 billion by 2025 (GRAND VIEW
RESEARCH).
Another field of application of ferroelectric materials is the infrared sensors
based on their pyroelectric property, which is also one of the most important
properties of a ferroelectric material. Beyond these well-known applications,
another important application based on the switching of ferroelectric polariza-
tion is the non-volatile memory device such as FeRAM. Examples are given in
the following and details will be introduced in the following chapters.
1.2.1 Piezoelectric Device Applications

An example of smart system using piezoelectric material is the distance radar
system in a car or a sonar system in submarines as shown in Figure 1.6, where the
key sensing element is based on piezoelectric material to realize the conversion
between electrical energy and acoustic energy for sending and receiving sound
waves. Other application examples of piezoelectric devices include active damp-
ing system, micro-scanning system in scanning probe imaging instrument (such
as AFM), force sensor, accelerometer, energy harvesting, etc.
Medical ultrasound imaging system with piezoelectric material as the trans-
ducer to convert electrical and acoustic energies is another very good example
of device application where the piezoelectric material plays the roles of sens-
ing and actuating functions. Figure 1.7 shows photos of ultrasound transducers
developed in our group. Knowledge in ultrasound transducer fabrication, char-
acterization, and applications will be intensively introduced in Chapter 6.
A very new application example is piezoelectric-based fingerprint ID system in
mobile phone. The currently used finger identification system is based on capac-
itance measurement to obtain two-dimensional (2D) information of fingerprint,
but it faces the problem of difficulty to identify the fingerprint when the finger is
dirty or wet. Ultrasound fingerprint identification system based on piezoelectric
ultrasonic transducer and imaging system can obtain a three-dimensional image
(a) (b)
Figure 1.6 Piezoelectric materials-based sonar system for car (a) and submarine (b).
(a) (b)
Figure 1.7 (a) Transducers and (b) B-mode image of a wire phantom acquired with
PolyU-made array ultrasound transducer.
PMUT unit
Capping layer
Coupling materials + MEMS
Piezoelectric layer
Cavity
CMOS wafer
Bottom electrode Top electrode
Figure 1.8 Illustration of concept of a ultrasonic transducer-based fingerprint ID system based

on complementary metal-oxide-semiconductor micro-electro-mechanical systems
(CMOS-MEMS) technology.
of fingerprint with a certain depth. This can overcome the problems of the cur-
rent fingerprint identification system in most mobile phones. InvenSense, Inc. is
one of the main suppliers of this solution, and Figure 1.8 is an illustration of the
ultrasonic fingerprint system.
1.2.2 Infrared Sensor

An infrared sensor is usually made of a ferroelectric material, which is also
pyroelectric that generates surface electric charges when exposed to heat
in the form of infrared radiation. A pyroelectric-based infrared sensor can detect
the temperature change but produce no response for a steady temperature since
the pyroelectric sensing element can only produce polarization change-induced
electric charge when the sensor is subject to temperature change. Figure 1.9a
shows a photo of a real infrared detector with its internal device structure
illustrated in Figure 1.9b, where the active element is made of pyroelectric
materials such as LiTaO3 . Those pyroelectric materials with their polarization
able to be switched are ferroelectrics. Therefore, pyroelectric sensors that are
widely used as infrared detectors are important device applications for ferroic
materials in a smart system.
Window
Absorbing electrode layer

Bottom electrode Pyroelectric plate
Circuit board
(a) (b)
Figure 1.9 A photo of an infrared detector (a) and illustration of its internal structure (b).
1.2.3 Ferroelectric RAM (FeRAM)

Ferroelectric RAM (FeRAM, F-RAM, or FRAM) is a random-access memory that
is similar to Dynamic Random Access Memory (DRAM) in structure but uses a
ferroelectric layer instead of a dielectric layer to achieve non-volatility. FeRAM
is one of a growing member of alternative non-volatile random-access memory
technologies that offers the same functionality as flash memory.
Advantages of FeRAM over flash memory include lower power usage, faster
write performance, and much greater maximum read/write endurance (about
1010 –1014 cycles). FeRAMs have data retention of more than 10 years at +85 ∘ C
(up to many decades at lower temperatures). Market disadvantages of FeRAM
are much lower storage densities than flash devices and higher cost.
A ferroelectric material has a nonlinear relationship between the applied elec-
tric field and the stored charge. Specifically, the ferroelectric characteristic has
the form of a hysteresis loop, which is very similar in shape to the hysteresis
loop of ferromagnetic materials. The dielectric constant of a ferroelectric is typi-
cally much higher than that of a linear dielectric because of the effects of electric
dipoles formed in the crystal structure of the ferroelectric material. When an
external electric field is applied across a dielectric, the dipoles tend to align them-
selves with the field direction. This alignment process is produced by small shifts
in the positions of ions and shifts in the distributions of electric charges in the
crystal structure. After the charges are removed, the dipoles retain their polariza-
tion state. Binary “0”s and “1”s are stored as one of the two possible electric polar-
izations in each data storage cell. For example, in Figure 1.10, a “1” is encoded
using the negative remnant polarization “−Pr ,” and a “0” is encoded using the pos-
itive remnant polarization “+Pr ”. In this FET structure with a ferroelectric layer as
gate dielectric, the two polarization states correspond to different V th , resulting
in a memory window within which the ON and OFF states of the FET can be read.
Spontaneous
polarization of
ferroelectric layer
Gate Memory
window
Source Drain – – – –
– – – – – – – –– –
N N
–––––––––
IDS P
–––––––––
– – – –
VG
Figure 1.10 Schematic diagram of field-effect transistor (FET) and the current–voltage (I–V)
characteristics induced by two different polarization state.
1.3 Device Application of Ferromagnetic Materials 9
FeRAM remains a relatively small part of the overall semiconductor market.

In 2016, worldwide semiconductor sales were US $338.93 billion (according to
WSTS, SIA), with the flash memory market accounting for US $59.2 billion
(according to IC insights (Cho 2018)). The 2017 annual revenue growth of
Cypress semiconductor, perhaps the major FeRAM vendor, were reported to
be US $2.33 billion. The much larger sales of flash memory compared with the
alternative FeRAMs support a much larger research and development effort.
Flash memory is produced using semiconductor linewidths of 15 nm at Renesas
Electronics Corporation (2017). Flash memory can store multiple bits per cell
(currently Samsung has announced the 64-layer 512-Gb in the NAND flash
devices). As a result of innovations in flash cell design, the number of bits per
flash cell is projected to increase to double or even to triple. As a consequence,
the areal bit densities of flash memory are much higher than those of FeRAM,
and thus the cost per bit of flash memory is orders of magnitude lower than that
of FeRAM.
1.3 Device Application of Ferromagnetic Materials

Among many successful applications of ferromagnetic-based devices, memory
device based on ferromagnetic material is one of the most successful examples,
especially in the thin film form. This is manifested by the very large market of
magnetic hard disc in computing systems. But in most recent years, solid state
memory (mainly flash memory) is superseding the magnetic hard disc. Never-
theless, ferromagnetic material also finds its application in non-volatile memories
such as spin-transfer torque memory.
1.3.1 Spin-Transfer Torque Memory

Spin-transfer torque can be used to flip the active elements in magnetic
random-access memory. Spin-transfer torque magnetic random-access memory
(STT-RAM or STT-MRAM) has the advantages of lower power consumption
and better scalability over conventional magnetoresistive random-access mem-
ory (MRAM), which uses magnetic field to flip the active elements. Spin-transfer
torque technology has the potential to make possible MRAM devices combining
low current requirements and reduced cost; however, the amount of current
needed to reorient the magnetization at present is too high for most commer-
cial applications, and the reduction of this current density alone is the basis
for present academic research in spin electronics. Figure 1.11 is a schematic
diagram of spin valve structure, while arrows indicate the magnetization
direction.
1.3.2 Magnetic Field Sensor Based on Multiferroic Device

If a material possesses more than one of the ferroic properties of ferroelectric,
ferromagnetic and ferroelastic, it is called multiferroics. Unfortunately, such
Free layer
Pin layer
Low resistance state High resistance state
Figure 1.11 Schematic diagram of spin valve structure where arrows indicate the
magnetization directions.
materials are rare and are usually strong in one property but very weak in
another, such as BiFeO3 , which is very strong in ferroelectrics but very weak in
ferromagnetic (it is antiferromagnetic in fact). This makes multiferroic materials
hard to be practically applied in devices. However, people have been trying to
make composite materials such as piezoelectric with magnetostrictive mate-
rials, where the mechanical coupling between them makes the “multiferroic”
meaningful for device application, for example, making very sensitive magnetic
field sensor.
The magnetoelectric (ME) effect is the phenomenon of inducing magnetiza-
tion by an applied electric field (E) or polarization by magnetic field (H). Many
efforts have been devoted to improve the limit of detection of the ME compos-
ite at low frequency range, and values of ∼10−7 Oe at 1 Hz has been reported
(Wang et al. 2011). Based on the magnetic–strain–electric coupling, scientists
have demonstrated dc magnetic field sensor with a detection limit of 2 × 10−5 Oe
to dc magnetic field with a nonlinear ME magnetic effect (Li et al. 2017). Ferro-
electric material PZT and magnetostrictive material Metglas have been imple-
mented in the composite device (see Figure 1.12).
1.4 Ferroelastic Material and Device Application

Shape memory alloys (SMAs) are the most typical ferroelastic material, which
is an important member of ferroics. The shape memory characteristics origi-
nates from the phase transition between high temperature austenite phase and
low temperature martensite phase, where the shape at cubic-austenite phase can
be resumed from low temperature martensite phase whose lattice can be largely
twisted (see Figure 1.13).
The SMAs have been widely used in devices from brace of orthodontia
and other medical applications, air jet, satellite antenna, etc. Research was
carried out in developing systems that would optimize the chevron “immersion”
1.4 Ferroelastic Material and Device Application 11
ID electrodes
Metglas
Piezofiber
Epoxy Kapton
VirginiaTech
Figure 1.12 Outline of ME device from Virginia Tech and schematic of the cross-section of the
ME composite. Source: Wang et al. (2011). Adapted with permission of John Wiley and Sons.
He
ati
ng
Stress/load
Loading
Cooling
Heating
Temperature
Figure 1.13 Phase transition between high temperature austenite phase and low
temperature martensite phase, where the shape at cubic-austenite phase can be resumed
from low temperature martensite phase whose lattice can be largely twisted.
into the jet flow based on the flight condition. As shown in Figure 1.14a,
SMAs activated by heat were developed that would allow for full chevron
immersion in jet flow during high thrust requirements (e.g. during take-off )
and not immersing it during cruise where the thrust efficiency is of greater
importance (Anon n.d.).
(a) (b)
Figure 1.14 (a) Brace of orthodontia using shape memory alloys and (b) arthrodesis device
developed by Karnes et al.
For broken bone rehabilitation, a SMA plate with a memory transfer tempera-
ture close to body temperature can be attached to both ends of the broken bone as
shown in Figure 1.14b. From body heat, the plate will contract and retain its origi-
nal shape, therefore exerting a compression force on the broken bone at the place
of fracture. After the bone has healed, the plate continues exerting the compres-
sive force and aids in strengthening during rehabilitation (Garlock et al. 2017).
1.5 Scope of This Book

In Chapters 2–5, fundamentals of ferroelectrics, applications of ferroelectric
materials, recent advances, and advanced measurement and testing techniques
in ferroelectrics will be introduced. In particular, device applications of fer-
roelectric materials in thin film form will be introduced including FeRAM,
ferroelectric tunneling-based resistive switching, etc. The recent advances
include ferroelectricity in emerging materials such as 2D materials and high-k
gate dielectric material HfO2 , while the advanced characterization technologies
include the piezoresponse force microscopy (by imaging and switching ferro-
electric domains) and Cs-corrected transmission electron microscopy (TEM)
where atomic level ionic displacement can be identified.
As the most important property application of ferroelectric materials,
fundamentals of piezoelectric physics and engineering considerations for device
design and fabrication are introduced in Chapters 6 and 7.
In Chapter 8, starting with a brief introduction on origin of ferromagnetism and
its analogy to ferroelectrics, device applications, particularly for magnetostrictive
devices, are introduced.
Chapters 9 and 10 will introduce the multiferroics of materials possessing both
ferromagnetic and ferroelectric orders including single phase and composite
materials. In particular, devices based on the integration of ferroelectric and
ferromagnetic materials such as multiferroic memory device and ME coupling
device for sensor applications will be introduced.
In Chapter 11, ferroelastic materials represented by SMA and magnetic SMAs
as well as their device applications will be introduced.
References 13
References
Boyn, S., Grollier, J., Lecerf, G. et al. (2017). Learning through ferroelectric domain
dynamics in solid-state synapses. Nature Communications 8: 1–7.
Chang, C.-Y. and Chen, T.-L. (2017). Design, fabrication, and modeling of a novel
dual-axis control input PZT gyroscope. Sensors 17 (11): 2505.
Cho, J. (2018). Amid contradictory forecast: IC insights: ‘Memory chips will grow at
annual rate of 5% only on average by 2022’. Seoul, Korea: BusinessKorea.
Garlock, A., Karnes, W.M., Fonte, M. et al. (2017). Arthrodesis devices for
generating and applying compression within joints. US 2017/0296241 A1,
Available at: https://patents.google.com/patent/US20170296241A1/en.
Li, M., Dong, C., Zhou, H. et al. (2017). Highly sensitive DC magnetic field sensor
based on nonlinear ME effect. IEEE Sensors Letters 1 (6): 1–4.
Renesas Electronics Corporation (2017). Renesas electronics achieves large-scale
memory operation in fin-shaped MONOS flash memory for industry’s first
high-performance, highly reliable MCUs in 16/14nm process nodes and beyond.
Wang, Y., Gray, D., Berry, D. et al. (2011). An extremely low equivalent magnetic
noise magnetoelectric sensor. Advanced Materials 23 (35): 4111–4114. Available
at: https://doi.org/10.1002/adma.201100773.
15
Introduction to Ferroelectrics
2.1 What Is Ferroelectrics?

Ferroelectric can be defined as switchable spontaneous polarization at
temperatures below the Curie point (T c ). The spontaneous polarization and
its switching make the ferroelectric material exhibit polarization–electric field
hysteresis loop (called P–E loop) as shown in Figure 2.1, where the Pr is called
remnant polarization, Ps is the saturated polarization, and Ec is the coercive field,
a field at which the polarization direction can be switched. A very important
characteristic of ferroelectric materials is the presence of domain structure
where each domain has its own polarization direction. The formation of domain
structure is to align dipoles in different directions in order to minimize elec-
trostatic and elastic energy, and it is the domain switching and domain wall
movement that produce hysteresis and saturation nonlinearities. As shown in
the inset of the figure, the polarization switching is accompanied by domain wall
movement and new domain formation.
It is worthy to note that, except with presence of charged defects, the polariza-
tions are usually aligned head-to-tail. However, in some nanostructured ferro-
electrics, or under a special strained situation, charged domain walls, topological
domains such as vertex type of domains can be formed. Details will be introduced
in Chapter 5.
2.1.1 P–E Loop

People usually use the term of P–E loop to prove the ferroelectricity. But, strictly
speaking, it should be called D–E loop since the so-called polarization is actually
obtained by measuring the electric displacement D through capacitance mea-
surement. Nevertheless, it is approximately correct, since for ferroelectric mate-
rial, the electric polarization is nearly equal to the electric displacement.
For a dielectric material, the presence of an electric field E makes the elec-
tron cloud slightly shifted inducing a local electric dipole moment that is called
electric displacement, while for a ferroelectric material, its polarization density P
(spontaneous polarization or induced by electric field) will dominate the electric
displacement D that can be defined as
D = 𝜖0 E + P (2.1)

16 2 Introduction to Ferroelectrics
P Figure 2.1 Typical P–E loop from a

ferroelectric crystal. Insets illustrate
domain structures for the
Ps corresponding polarizations. As the
field is increased, the polarization of
Pr domains with an unfavorable direction
of polarization will start to switch in
the direction of the field.
E
–Ec +Ec
–Pr
where 𝜖 0 is the vacuum permittivity or called permittivity of free space and

P depends linearly on the electric field in a homogeneous and isotropic
material, i.e.
P = 𝜖0 𝜒 e E (2.2)
where 𝜒 e is electric susceptibility of the material. Therefore, we have
D = 𝜖0 (1 + 𝜒e )E = 𝜖E (2.3)
where the permittivity
𝜖 = 𝜖0 𝜖r (2.4)
and relative permittivity
𝜖r = 1 + 𝜒e (2.5)
For a ferroelectric material, its electric displacement is contributed mainly by
ionic displacement, which is much larger than the contribution from vacuum
dielectric; therefore, the polarization P is approximately equal to D. Practically,
P–E loop is measured by a Sawyer–Tower method that will be introduced in
Chapter 3.
2.1.2 Relationships Between Dielectric, Piezoelectric, Pyroelectric,

and Ferroelectric
2.1.2.1 Ferroelectric–Dielectric
Why does a ferroelectric material have a very large dielectric constant?
In a large category, a ferroelectric belongs to dielectrics. There are three pri-
mary contributions to electric polarization: electronic, ionic, and dipole reori-
entation; and the polarization is expressed quantitatively as the sum of electric
dipoles per unit volume (C/m2 ).
2.1 What Is Ferroelectrics? 17
Dipole
Electronic Ionic
reorientation
E=0
E>0
Center of negative charge
Electric field
Figure 2.2 Schematic diagram showing the origin of the electric polarization.
For all dielectric materials, the electron clouds deform under electric field,
forming electric dipoles. Electric polarization from this electron clouds defor-
mation is usually much smaller compared with ionic displacement. In ionic
crystals, when an electric field is applied, cations and anions are attracted to the
cathode and anode, respectively. While for a ferroelectric, electric polarization
reorientation (rotation or reversing) will result in a significant change of electric
displacement (see Figure 2.2).
This diagram also illustrates that a ferroelectric falls into a larger category called
dielectrics. Let’s look at their difference with an example by comparing refrac-
tive index of a typical dielectric material SiO2 and a typical ferroelectric mate-
rial BaTiO
√ 3 . By following relationship of refractive index with dielectric constant
n = 𝜖r , with dielectric constant of about 3.6 for SiO2 , its refractive index is
about 1.4–1.5. However, for a ferroelectric material, its dielectric constant is usu-
ally very large due to the reorientation of electric polarization, a few hundred for
BaTiO3 , for example, but we cannot
√ say that its refractive index is also very large
based on the relation of n = 𝜖r . In fact, the refractive index of a material is an
optical constant that only works in optical frequency regime in which electron
cloud can respond to the optical frequency of electric field (say 1014 Hz), but the
ferroelectric polarization
√ cannot follow the optical frequency to rotate or switch,
i.e. this equation n = 𝜖r does not work for ferroelectrics in optical frequency.
2.1.2.2 Ferroelectric–Piezoelectric
Many materials such as AlN, GaN, and ZnO exhibit electric polarization orig-
inated from their crystal structure symmetry breaking, but their polarizations
cannot be switched by external electric field. These electrically polarized mate-
rials are piezoelectric but not ferroelectrics. Figure 2.3 shows the relationships
between the piezoelectric and ferroelectric materials as well as pyroelectric and
dielectric materials. A ferroelectric is also piezoelectric since the electric polar-
ization change induced by mechanical strain will generate electric charges and
voltage at the surface of a ferroelectric plate, i.e. piezoelectricity.
Figure 2.3 Relationships of the

piezoelectric, ferroelectric,
pyroelectric, and dielectric.
Ferroelectric
Pyroelectric
Piezoelectric
Dielectric
2.1.2.3 Ferroelectric–Pyroelectric
As illustrated in Figure 2.3, a ferroelectric is also a pyroelectric, but a pyroelectric
material may not be a ferroelectric. Pyroelectricity is a phenomenon of tempera-
ture dependence of the spontaneous polarization. As the temperature of a crys-
tal is changed, electric charges corresponding to the change of the spontaneous
polarization appear on the surface of the sample. Among many pyroelectric mate-
rials, only those whose spontaneous polarization can be reversed by an electric
field belong to ferroelectrics.
2.2 Origin of Ferroelectrics

2.2.1 Structure-Induced Phase Change from Paraelectric
to Ferroelectric
The origin of ferroelectric comes from space inversion symmetry breaking, i.e.
the structure symmetry breaking results in spontaneous polarization of a crystal
and generates the ferroelectricity. A simple understanding is that there must be a
relative shift of cations and anions, i.e. non-coincidence of positive and negative
charge centers, in the crystal unit cell. This will not happen in a crystal of cubic
structure with central symmetry. To generate electric polarization in a perovskite
structure, lattice distortion along either ⟨100⟩, ⟨110⟩, or ⟨111⟩ is a must, plus rel-
ative shift of cations and anions. This only happens when the crystal is below its
Curie temperature. The following describes the formation of ferroelectricity from
the crystal symmetry point of view:
• Crystals can be classified into 32 point groups according to their crystallo-

graphic symmetry, and these point groups can be divided into two classes, one
with a center of symmetry and the other without, as indicated in Table 2.1.
• Among these 32 point groups, 21 of them do not have a center of sym-
metry. Except the point group 432, in crystals belonging to 20 of these
non-centrosymmetric point groups, positive and negative charges are gen-
erated on the crystal surfaces when a stress is applied; so these crystals are
known as piezoelectrics.
2.2 Origin of Ferroelectrics 19
Table 2.1 Crystallographic classification according to crystal centrosymmetry.
Crystal system
Symmetry Polarity
Cubic Hexagonal Tetragonal Rhombohedral Orthorhombic Monclinic Triclinic
Centro (11) m3m m3 6/mmm 6/m 4/mmm 4/m 3m 3 mmm 2/m
32 point group Non-polar
432 622 422
of (22) 23 6 4 32 222
crystallographic Non-centro 43m 6m2 42m
symmetry (22) 2
Polar (10) 6mm 6 4mm 4 3m 3 mm2 1
m
Inside are piezoelectrics (20)
Inside are pyroelectrics (10)
Source: Adapted from Uchino (2009).
Ti
Ti O Ti
Ps
Ps
Ba
T > Tc T < Tc
Figure 2.4 Illustration of crystal structure symmetry breaking-induced ferroelectricity in

BaTiO3 .
• Among these 20 point groups, 10 of them have a unique polar direction, and
these 10 polar classes are pyroelectrics. Among these pyroelectrics, only those
whose polar can be reversed by external electric field are ferroelectrics.
BaTiO3 is an example of a typical ferroelectric crystal with perovskite struc-
ture as illustrated in Figure 2.4. Above the Curie temperature T c , the paraelectric
cubic structure is centrosymmetric, whereas in the tetragonal phase below T c , it
is energetically favorable for the O2− ions to be shifted slightly below face centers,
and Ti4+ ions are shifted upward from the unit cell center. The relative change in
positions of the Ti4+ and O2− ions produces a spontaneous polarization Ps as well
as the non-centrosymmetric structure. More ferroelectric phases of BaTiO3 will
be introduced later.
2.2.2 Soft Phonon Mode

Ionic displacement can be expected through lattice vibrations at a finite temper-
ature. Figure 2.5 shows some of the possible lattice vibrations in a perovskite-like
crystal, where
Panel (a) shows an initial cubic (symmetrical) structure.
Panel (b) is a symmetrically elongated one.
Panel (c) has coherently shifted center cations exhibiting ferroelectricity. If the
shift is anti-polarized, it results in antiferroelectricity (AFE) that will be intro-
duced later.
Elongation
direction
(a) (b)
(c)
Figure 2.5 Starting from the original cubic structure (a), if (b) is stabilized, only oxygen
octahedra are distorted without generating dipole moments (acoustic mode). When (c) is
stabilized, dipole moments are generated (optical mode). The final stabilized state (c)
corresponds to ferroelectric state.
Since only oxygen octahedra are distorted in Figure 2.5b without generating
electric polarization, its vibration is a kind of phonon mode. However, if the par-
ticular mode Figure 2.5c becomes stabilized, with decreasing temperature, the
vibration frequency decreases (soft phonon mode), and finally at a certain phase
transition temperature (Curie temperature), this frequency becomes zero, i.e. the
electric polarization state is stabilized.
Since the soft modes in ferroelectrics lead to electric polarization, they are opti-
cally active and can be detected by means of optical spectroscopy in the spectra of
dielectric permittivity (real and imaginary parts). Therefore, the vibration shown
in Figure 2.5c is also called optical mode.
Spectroscopic studies of the soft phonon modes provide a very powerful tool
for investigating the ferroelectric transitions. Figure 2.6 shows Raman spectra of
BaTiO3 phase transition, where the 305 cm−1 peak reduces its sharpness as the
in situ temperature increases and becomes indistinct at 150 ∘ C, indicating that
tetragonal phase is transformed to cubic phase. This trend is consistent with the
results that the peak around 310 cm−1 appeared below Curie temperature (T c )
and vanished above T c (Hayashi et al. 2013).
The dielectric permittivity in the soft mode is governed by three laws (Cochran
1960; Shirane et al. 1970; Luspin et al. 1980):
(a) Static dielectric permittivity 𝜖 r (0) produced by the soft mode obeys the
Curie–Weiss law:
𝜖r (0) ∼ (T − Tc )−1 (2.6)
(b) The eigen frequency 𝜔(T) of the soft mode follows the Cochran behavior:
1
𝜔(T) ∼ (T − T0 ) 2 (2.7)
where T 0 is the soft mode condensation temperature.

Intensity (a.u.)
200 °C
150 °C
100 °C
50 °C
25 °C
200 300 400 500 600 700 800

Raman shift (cm–1)
Figure 2.6 In situ Raman spectra of BaTiO3 particles measured at different temperatures: 25,
50, 100, 150, and 200 ∘ C. Source: Hayashi et al. (2013). Adapted with permission of Elsevier.
(c) The static dielectric constant and the soft mode frequency are connected via
the Lyddane–Sachs–Teller (LST) relation:
𝜖∞ 𝜔2 (T)
(T) = T 2 (2.8)
𝜖r (0) 𝜔L
where 𝜖 ∞ is the high-frequency dielectric constant and 𝜔T and 𝜔L are the
transverse and longitudinal frequencies of the corresponding vibrations,
respectively.
2.3 Theory of Ferroelectric Phase Transition

2.3.1 Landau Free Energy and Curie–Weiss Law
Ferroelectric phase transition can be well described by thermodynamic theory.
Landau addressed this problem by considering free energy in the form of expan-
sion of ferroelectric polarization P, from which fundamental properties of the fer-
roelectric material such as polarization and dielectric permittivity can be derived.
For simplicity, we only consider one-dimensional case, so the expression of Lan-
dau free energy as a function of polarization and temperature T is
F = F0 (T) + F(P, T) (2.9)
1 2 1 4 1 6
F(P, T) = 𝛼P + 𝛽P + 𝛾P + … (2.10)
2 4 6
where F 0 does not depend on the order parameter P and it describes the tem-
perature dependence of free energy of high temperature phase near the phase
transition (Graz University of Technology n.d.; Uchino 2009). The coefficients
𝛼, 𝛽, and 𝛾 are temperature dependent in general, while their signs are different
for the first- and second-order phase transitions. The reason to have only even
number of powers is that when the polarization changes signs, the free energy of
the crystal will not change. Since this phenomenological formula describes the
free energy across the paraelectric to ferroelectric phase transition, it should be
applicable to all temperature ranges.
In our calculation, however, the coefficient 𝛼 is the only coefficient that is
assumed to be temperature dependent. The fraction numbers are for conve-
nience when we differentiate F in respect of P. To define the relation of 𝛼 and T,
the free energy for certain polarization should be considered. As in paraelectric
state (i.e. P = 0), the free energy should be zero (F(P, T) = 0) at any temperature
above its Curie point. In ferroelectric state, there are two situations for free
energy, either F(P, T) > 0 or F(P, T) < 0. For free energy greater than zero, the
paraelectric state should be realized because it always tends to the minimum
energy state. Therefore, the free energy for a certain polarization must be lower
than zero in order to stabilize the ferroelectric state. Thus, the coefficient 𝛼 of
the P2 term must be negative in ferroelectric state, while it is positive passing
through zero at temperature T 0 . According to this concept, a relation of 𝛼 and T
is formed as the following:
𝛼 = 𝛼0 (T − T0 ) (2.11)
where 𝛼 0 is a temperature-independent constant and T 0 the Curie–Weiss tem-
perature, which is equal to or lower than the actual transition temperature T c
(Curie temperature).
By differentiating the free energy, electric field E related to the equilibrium
polarization can be expressed as
𝜕F
E= = 𝛼P + 𝛽P3 + 𝛾P5 (2.12)
𝜕P
By considering the existence of polarization without external electric field, i.e.
E = 0, we can get
P(𝛼 + 𝛽P2 + 𝛾P4 ) = 0 (2.13)
This equation has two possible solutions, among them the trivial solution P = 0
that corresponds to a paraelectric
√
state is not our concern for ferroelectric state.
−𝛽± 𝛽 2 −4𝛼𝛾
Another finite solution P2 = 2𝛾
corresponds to a ferroelectric state.
From Landau free energy, the temperature-dependent dielectric constant
(relative permittivity) of a ferroelectric material at ferroelectric and para-
electric phases (below and above Curie temperature) can be derived. For
the first order of phase transition from paraelectric phase to ferroelectric
phase, the temperature-dependent Landau free energy, dielectric constant, and
spontaneous polarization are illustrated in Figure 2.7. One can see that the
characteristic of the first-order ferroelectric phase transition is that the order
parameter Ps drops to zero abruptly and the dielectric constant reaches a finite
T > T1
T c < T < T1
1/ϵr
Ps
T = Tc
T0 < T < Tc ϵr
T < T0
(a) (b) T0 Tc T1
Figure 2.7 (a) Free energy, (b) dielectric constant and spontaneous polarization, of first-order
phase transition.
peak value continuously; while 1/𝜖 r is linear at temperatures lower or higher

than Curie point. The meaning of T 0 and T 1 will be shown in the following
derivations.
2.3.2 Landau Theory of First-Order Phase Transition

For second-order phase transition, the order parameter grows continuously from
zero at the phase transition so the first two terms of the power series will dom-
inate. If the free energy is expanded to the sixth order in the order parameter,
the system will undergo a first-order phase transition with 𝛼 > 0, 𝛽 < 0, and 𝛾 > 0
(Wang 2010).
1 2 1 4 1 6
F(P, T) = 𝛼P + 𝛽P + 𝛾P (2.14)
2 4 6
When we try to find the existent polarization P without the external electric field
E, the free energy is minimized.
𝜕F
E= = 0 = 𝛼P + 𝛽P3 + 𝛾P5
𝜕P
P=0 or 𝛼 + 𝛽P2 + 𝛾P4 = 0
So, nonzero solution of P is
√
2
−𝛽 ± 𝛽 2 − 4𝛼𝛾
P =
2𝛾
Here the “−” sign is meaningless since it will result in P2 < 0. So,
√
2
−𝛽 + 𝛽 2 − 4𝛼𝛾
P =
2𝛾
Since 𝛼 = 𝛼 0 (T − T 0 ), the solutions of P are
√ √
−𝛽 + 𝛽 2 − 4𝛼0 (T − T0 )𝛾
P=± for T < T0
2𝛾
√ √
−𝛽 + 𝛽 2 − 4𝛼0 (T − T0 )𝛾
P = 0, ± for T0 < T < T1
2𝛾
P=0 for T > T1
The temperature T 1 is where the two nonzero solutions become unstable. For
T > T 1 , there is only one free energy minimum. For T 1 > T > T 0 , there are three
√ minima. P = 0 is the lowest free energy solution when T 1 > T > T c , while
potential
√
−𝛽+ 𝛽 2 −4𝛼0 (T−Tc )𝛾
P=± 2𝛾
exhibits the lowest free energy when T c > T > T 0 . For
T < T 0 , there are double potential minima of the free energy that correspond to
stable spontaneous polarization.
The difference between the Curie–Weiss temperature T 0 , the Curie tempera-
ture T c , and the ferroelectric limit temperature T 1 can be verified based on the
potential minima obtained in the first-order phase transition.
As the potential minima are obtained from
𝜕F
= E = 𝛼P + 𝛽P3 + 𝛾P5 = 0
𝜕P
which is valid for any temperature below or above Curie temperature, there are
three possible minima including P = 0 (i.e. F = 0). At T = T c , the free energy at
nonzero polarization must be equal to that of the paraelectric state, i.e.
F(P, T)|T=Tc = F(P, T)|P=0 = 0
So,
1 2 1 4 1 6
F(P, T) = 𝛼P + 𝛽P + 𝛾P = 0
2 4 6
then we get
𝛼 + 𝛽P2 + 𝛾P4 = 0
1 1
𝛼 + 𝛽P2 + 𝛾P4 = 0
2 3
By eliminating the P4 term from these two equations,
1 2
2𝛼 + 𝛽P = 0
2
4𝛼
P2 = −
𝛽
Putting P2 = − 4𝛼
𝛽
into the equations,
( ) ( )2
4𝛼 4𝛼
𝛼+𝛽 − +𝛾 − =0
𝛽 𝛽
or
( )2
1 4𝛼
𝛼− 𝛾 − =0
3 𝛽
Taking 𝛼 = 𝛼 0 (T − T 0 ) = 𝛼 0 (T c − T 0 ),
3 𝛽2
𝛼 = 𝛼0 (Tc − T0 ) =
16 𝛾
the Curie temperature is thus calculated as
3 𝛽2
Tc = T0 +
16 𝛼0 𝛾
This indicates that T c is little higher than T 0 . Meanwhile, when T = T 1 , the free
energy has only one solution at P = 0, i.e.
𝛽 2 − 4𝛼0 (T1 − T0 )𝛾 = 0
𝛽2
T1 = T0 +
4𝛼0 𝛾
For this equation, we can identify that T 1 is higher than T c .
As a result, we can conclude that
T0 < Tc < T1
To identify the dielectric constant, we first have to calculate the relative permit-
tivity 𝜖 r which is the response of the system to an electric field.
dP
𝜖r =
dE
In free energy equation, the field has been included to calculate 𝜖 r
𝜕F
=E
𝜕P
E = 𝛼0 (T − Tc )P + 𝛽P3 + 𝛾P5
dE = [𝛼0 (T − Tc ) + 3𝛽P2 + 5𝛾P4 ]dP

√
At the critical temperatures P = 0, −𝛽
2𝛾
, the susceptibility in the vicinity of the
phase transition can be obtained.
dP || 1
at Tc , 𝜖r = =
dE ||P=0 𝛼0 (T − Tc )
dP || 1
at T1 , 𝜖r = | √ −𝛽 =
dE |P= , 𝛼0 (T − T1 )
2𝛾
Based on the equations earlier, we can verify that in case of 𝛽 < 0, the permittivity
shows a maximum and a discontinuity of the spontaneous polarization appears
at T c .
2.3.3 Landau Theory of a Second-Order Phase Transition

At a second-order phase transition, the order parameter increases continuously
from zero starting at the critical temperature of the phase transition (i.e. T c = T 0 ).
It is assumed that 𝛽 > 0, so that the free energy has a minimum for finite values
of the order parameter. When 𝛼 > 0, there is only one minimum at P = 0. When
𝛼 < 0, there are two minima with P ≠ 0 (Figure 2.8).
For the ferroelectric materials following the second-order phase transition, it
can be derived as
1 1
F(P, T) = 𝛼P2 + 𝛽P4 where 𝛼0 > 0, 𝛽 > 0
2 4
It is assumed that 𝛽 > 0 so that the free energy has a minimum for finite values of
the order parameter.
Similar to the previous case, polarization can be found when E = 0.
𝜕F
= 0 = 𝛼P + 𝛽P3 = 𝛼0 (T − Tc )P + 𝛽P3
𝜕P
So,
P=0 or 𝛼0 (T − Tc ) + 𝛽P2 = 0
When T > T c , there is only one minimum at P = 0. When T < T c there are two
minima with P ≠ 0, i.e.
P=0 for T > Tc
√
𝛼0 (Tc − T)
P=± for T < Tc
𝛽
For dielectric constant,
𝜕F
=E
𝜕P
dE = [𝛼0 (T − Tc ) + 3𝛽P2 ]dP
T > Tc
T = Tc
Ps
1/ϵr
T < Tc ϵr
(a) (b) Tc
Figure 2.8 (a) Free energy, (b) dielectric constant and spontaneous polarization, of
second-order phase transition.
dP || 1 C
𝜖r = = = for T > Tc
dE ||P=0 𝛼0 (T − Tc ) (T − Tc )
dP || 1 C
𝜖r = = = for T < Tc
dE ||P2 = 𝛼0 (Tc − T) 2𝛼0 (Tc − T) 2(Tc − T)
𝛽
where C = 𝛼1 is Curie constant.
0
Now we can see that the dielectric permittivity can be derived from
second-order differentiation of Landau free energy with boundary conditions of
E = 0 (without electric field). The temperature-dependent dielectric permittivity
for ferroelectric materials following the first-order phase transition can thus be
derived. In both cases, 1∕𝜖r linearly depends on temperature.
The classification of order of phase transition comes from thermodynamic free
energy according to Ehrenfest (Jaeger 1998; Blundell and Blundell 2009), where
it can be labeled as the lowest derivation (order parameter) of the free energy
that shows discontinuity as a function of other thermodynamic variables during
phase transition. For instance, if a discontinuity is exhibited in the first derivative
of Gibbs free energy, it is classified as first-order phase transition. If the discon-
tinues order parameter is the second order of differentiation of Gibbs free energy,
the corresponding phase transition is classified as second-order phase transi-
tion. Although Ehrenfest classification is clear to understand, it is not a complete
method as it is not suitable if the derivative of free energy diverges. Therefore,
a modern classification is made. Similar to the Ehrenfest classes, two broad cat-
egories are divided by taking the latent heat into account (Maris and Kadanoff
1978).
For example, various solid/liquid/gas transitions usually involve a discontinu-
ous change in density, which is the first derivative of free energy with respect to
pressure. Therefore, they are classified as first-order phase transition based on
Ehrenfest classification. In the view point of modern classification, during the
water to ice phase transition, latent heat is added while the temperature of the
system remains unchanged. The specific discontinuity with respect to energy and
temperature of this phase transition thus assorts to the first-order differentiation
of Gibbs free energy, and it belongs to first-order phase transition associated with
latent heat. While for β-brass (an alloy of copper and zinc, in equal amounts)
phase transition, there is no latent heat. However, when the temperature is lower
than transition temperature, the probability for each copper atom to have more
zinc nearest neighbors increases, which would result in a completely random
arrangement of copper and zinc atoms. To say, the material undergoes a contin-
uous phase transition in which the specific heat has a singularity and it belongs
to second-order phase transition.
For the ferroelectric phase transition, for example, BaTiO3 , there is latent
heat during paraelectric to ferroelectric phase transition, and the spontaneous
polarization abruptly (discontinuously) changes to zero, which is the first-order
differentiation of Landau free energy. Therefore, this phase transition belongs
to first-order phase transition. While for the second-order phase transition, the
Ps continuously changes to zero, but the dielectric permittivity, which is the
second-order of differentiation of Landau free energy, is discontinuous at T c .
Most ferroelectric phase transitions are first order, very few (such as triglycine
sulfate [TGS]) are second-order transition.
2.4 Ferroelectric Domains and Domain Switching

2.4.1 Domain Structure
In order to reduce electrostatic and elastic energy, a ferroelectric material usually
presents a domain structure where every single domain possesses a polarization
direction. Depending on the structure, the domain polarizations are formed into
different angles. For example, in a perovskite ferroelectric material BaTiO3 with
polarizations along ⟨100⟩ and ⟨110⟩ for a pseudocubic unite cell, the domains
are aligned with 180∘ or 90∘ as shown in Figure 2.9a. While if the polarizations
are along ⟨111⟩ directions, 71∘ and 109∘ domains will be formed (angles between
the ⟨111⟩ directions). However, for a hexagonal structure such as LiNbO3 with
polarizations along ⟨0001⟩, only 180∘ domains can be formed. As shown in the
figure, the polarizations in adjacent domains are usually arranged head-to-tail in
order not to form charged domain walls. But some charged domain walls can also
be formed in some crystals if defects exist.
There are also vortex types of domains formed in some ferroelectric materials,
for example, YMnO3 , or in nanostructures and thin films. Figure 2.9b is a vor-
tex domain structure in YMnO3 . These vortex domains have been theoretically
predicted and experimentally observed and will be introduced in Chapter 5.
2.4.2 Ferroelectric Switching

In a ferroelectric material, polarization switching can be realized by applying
electric field and/or mechanical stress. It should be noted that an electric field can
induce both 180∘ and 90∘ polarization switching, whereas a mechanical stress can
only induce 90∘ switching of polarization. This can be understood by considering
an ellipsoid representing a polarized unit cell as illustrated in Figure 2.10. One
can see that an electric field can force the ellipsoid from out-of-plane to in-plane
90° boundary 180° boundary
β+
γ– α–
α+ γ+
(a) (b) β–
Figure 2.9 (a) Ferroelectric domain structure in BaTiO3 and (b) cloverleaf domain patterns and
the enlarged single vortex domain in the ferroelectric YMnO3 . Source: Adapted from Choi et al.
(2010) and Reprinted with permission of Springer Nature: Zhang et al. (2013).
E-field
direction E-field
direction
Adding electric field or
90° switching 180° switching
Stress
Adding mechanical stress or
90° switching
Figure 2.10 Illustration of polarization switching by electric field and mechanical stress.
as well as up-to-down, i.e. 90∘ and 180∘ polarization rotations, respectively. How-
ever, a mechanical stress along the long axis can only press the ellipsoid into an
in-plane ellipsoid.
In a ferroelectric crystal, its polarization switching is accompanied by domain
switching as well as domain wall movement. This process will induce strain in the
crystal, where the non-180∘ switching results in large piezoelectric strain.
2.5 Ferroelectric Materials

The first ever ferroelectric material is Rochelle salt discovered in 1675 by an
apothecary, Pierre Seignette, of La Rochelle in France. But this Rochelle salt is
not suitable for engineering application due to its poor property and stability.
The most typical and applicable ferroelectric material is BaTiO3 , which has
been widely used in electronic components as capacitors due to its very large
dielectric constant and very low dielectric loss. Nevertheless, the most popular
and useful ferroelectric material is Pb(Zrx Ti1−x )O3 (PZT) due to its very large
piezoelectric constant and remnant polarization. The applications of PZT in
sensors and actuators are the implementation of its excellent piezoelectric
property and applications in infrared sensors utilizing its pyroelectric prop-
erty; while some small-scale application of resistive random access memory
(ReRAM) implement its ferroelectric property. Relaxor ferroelectrics such
as Pb(Mg1/3 Nb2/3 )O3 –xPbTiO3 (PMN-xPT) emerged in recent decades have
attracted great attention due to their outstanding ferroelectric and piezoelectric
properties.
2.5.1 From BaTiO3 to SrTiO3

BaTiO3 is a model ferroelectric material, when temperature decreases, it goes
through a series of phase transitions from cubic to tetragonal, orthorhombic, and
then rhombohedral. The spontaneous polarization directions corresponding to
these non-centrosymmetric structures are ⟨100⟩, ⟨110⟩, and ⟨111⟩, respectively,
indexed by pseudocubic structure. The so-called pseudocubic means we still use
a′
Ti
b′ 130 °C c
a′ –90 °C 0 °C
a
b
Ba
Rhombohedral Orthorhombic Tetragonal Cubic
(a = b = c; (a = b = c; (a = b ≠ c; (a = b = c;
α = β = γ ≠ 90°) α ≠ 90°, β = γ = 90°) α = β = γ = 90°) α = β = γ = 90°)
Figure 2.11 Illustration of BaTiO3 structures at different temperatures showing different

polarization directions at different phases.
cubic structure to index the atomic planes and directions for convenience, but in
fact, the crystal structure is not cubic any more due to its distortion.
You may be wondering, the so-called orthorhombic structure in Figure 2.11
should be a monoclinic structure when the cubic unit cell is stretched along
[110] direction. It seems to be right, but since there is relative shift of cations
and anions, the real Bravais unit cell is actually a larger orthorhombic unit cell
with the shadow plane as the a′ b′ plane with c′ perpendicular
√ to a′ b′ plane.
′ ′ ′
The orthorhombic unit cell has a = a, b = c = 2a, i.e. the volume of the
orthorhombic unit cell is doubled, and a′ , b′ , and c′ are not equal, where a is for
pseudocubic unite cell in the distorted structure.
Ceramics based on BaTiO3 are very important in electronic component of
capacitors due to its high dielectric constant and low dielectric loss. BaTiO3 thin
film is also one of the most attractive ferroelectric thin films for memory device
application. This will be introduced in Chapter 4.
Pure SrTiO3 is a quantum paraelectric material where quantum fluctuations of
atomic positions suppress a ferroelectric transition, leading to a so-called incipi-
ent ferroelectric (Müller and Burkard 1979; Zhong and Vanderbilt 1996; Barrett
1952; Hemberger et al. 1995). It means that SrTiO3 is supposed to be ferroelectric
but its Curie temperature is absolute 0 K, which can never be reached. There-
fore temperature-dependent dielectric constant of SrTiO3 increases continuously
when temperature decreases, but it can never have a peak like a normal ferroelec-
tric material (see Figure 2.12). However, one can see that when a bias voltage is
applied on the SrTiO3 crystal, a relaxor-like paraelectric–ferroelectric transition
happens above 0 K. This is because of the existence of polar nano regions (PNRs)
in SrTiO3 at low temperatures. These PNRs can form ferroelectric nano-domains
under electric field, giving rise to relaxor-like ferroelectricity that will be intro-
duced later in this section.
SrTiO3 single crystal has been used as many oxide thin films substrate material
due to its high chemical stability and good lattice matching with many perovskite
oxide materials. When SrTiO3 is doped with BaTiO3 , it becomes (Bax Sr1−x )TiO3
(called BST) and its Curie temperature can be tuned between 0 K to T c of BaTiO3
depending on how much BaTiO3 is added. The study of BST, especially BST thin
films, has been a very hot topic due to its large dielectric tunability which is very
useful in tunable microwave device.
BST is attractive for microwave device application mainly because of two rea-
sons. The first is that the dielectric permittivity of BST strongly depends on the
25
1
1- E = 0 kV/cm
20 2- E = 2 kV/cm
3- E = 5 kV/cm
4- E = 15 kV/cm
ϵ(T ) × 103
15
10
2
5 3
4
0
0 100 200 300
Temperature (K)
Figure 2.12 Dielectric constant of SrTiO3 single crystal as a function of temperature and
biasing field. Source: Vendik et al. (1999). Reprinted with permission of Springer Nature.
bias electric field; this property is termed as dielectric tunability. Another reason
is that BST possesses very low dielectric loss (tan 𝛿) at microwave frequency. In
the following paragraphs, properties and applications of BST are introduced.
In Figure 2.13, an experimental result shows the change in lattice parameters
when different compositions of Ba and Sr are added into the compound. When
the composition of Ba increases, it can be seen that the structure of BST varies,
changing from cubic to tetragonal, proofing the structural transition from SrTiO3
(STO) to BaTiO3 (BTO). Therefore, it can be concluded that the tetragonality
increases when the composition of Ba increases.
0.405
c
Lattice parameters (nm)
0.400
a T
0.395
C
0.390
0.0 0.2 0.4 0.6 0.8 1.0

SrTiO3 Composition BaTiO3
Figure 2.13 Lattice constant of (Bax Sr1−x )TiO3 (both bulk and films) as a function of
composition x. Source: Adapted from Abe and Komatsu (1995).
25 000
x=0
20 000 x = 0.34
x = 0.27
x = 0.47 x = 0.65
x = 0.87
15 000
ϵr(T )
10 000 x = 1.0
5000
0
0 100 200 300 400
Temperature (K)
Figure 2.14 The phase transition behavior (the dielectric constant as a function of
temperature) of (Bax Sr1−x )TiO3 single crystal for various content of Ba–x. Source: Vendik et al.
(1999). Reprinted with permission of Springer Nature.
Figure 2.14 shows a theoretical result that the composition of Ba in BST varies
the Curie temperature as well as the dielectric properties of the compound.
Owing to the increase of tetragonality, when more Ba is added, the Curie
temperature increases. One can also notice that the dielectric constant near T c
is a few orders of magnitude greater than the rest, indicating a paraelectric to
ferroelectric phase transition.
Figure 2.15 shows the BST dielectric constant as a function of bias electric field.
The dielectric constant has a huge difference when E-field is applied. In the case
of x = 0.24, the dielectric constant is greatly reduced from 𝜖 r > 1400 to 𝜖 r < 500
when a field of E = 10 MV/m is applied. The electric field-dependent dielectric
permittivity is commonly described as the dielectric tunability n, which is defined
as the ratio of the dielectric permittivity at zero electric field bias to the permit-
tivity under electric field bias E, i.e.
𝜖r (0) − 𝜖r (E)
n=
𝜖r (0)
Question to students: Why does dielectric constant decrease when an E-field is
applied?
2.5.2 From PbTiO3 to PbZrO3

PbTiO3 (PT) is another model ferroelectric material. As illustrated in Figure 2.16,
above its Curie temperature, the paraelectric cubic structure is centrosymmetric;
but below Curie temperature, it is energetically favorable for the O2− ions to shift
slightly below face centers and Ti4+ ions to shift upward from the unit cell center
forming tetragonal phase. The relative shift in positions of the Ti4+ and O2− ions
produces a spontaneous polarization Ps .
Figure 2.15 Bias field 1600

dependence of (Bax Sr1−x )TiO3
dielectric constant under the
condition of f = 100 kHz at room 1400
temperature. Source: Adapted
from Abe and Komatsu (1995). 1200
Dielectric constant ϵ
1000
800
0.24
600
0.44
400
0
0.68
200
1.0
0
–50 –40 –30 –20 –10 0 10 20 30 40 50
Bias field (MV/m)
Pb
Ti
Ti O
Ps
(a) T > Tc (b) T < Tc
Figure 2.16 (a) Perovskite structure of PbTiO3 in the cubic form above T c and (b) tetragonal
structure of PbTiO3 for T < T c presenting spontaneous polarization.
2.5.3 Antiferroelectric PbZrO3

The notion of antiferroelectric (AFE) dates back to the early 1950s. It describes
a state where chains of ions in the crystal are spontaneously polarized, but
with neighboring chains polarized in antiparallel directions. As a result, the
crystal does not display spontaneous macroscopic polarization or the piezo-
electric effect. PbZrO3 was the first compound identified as an antiferroelectric
crystal (Sawaguchi et al. 1951; Shirane et al. 1951). At that time, Kittel (1951)
proposed a theory of antiparallel ionic displacements in dielectrics using the
antiferromagnetism scheme. In parallel, Sawaguchi et al. assigned the perovskite
lead zirconate, PbZrO3 , as antiferroelectric because of its dielectric behavior.
Thereafter, materials that exhibit a structural phase transition between two
nonpolar phases with a strong dielectric anomaly at the high temperature side of
the transition are named antiferroelectrics, and PbZrO3 remains the prototype
for this large group of materials.
When an antiferroelectric crystal is subject to electric fields, the antiparal-
lel dipoles can be flipped and forced to be parallel, corresponding to an elec-
tric field-induced antiferroelectric-to-ferroelectric phase transition (Cross 1967).
This phase transition manifests itself by the signature of double-hysteresis loops
as the electric polarization P is monitored as a function of applied field E (see
Figure 2.17a). In addition to the abrupt change in polarization, significant changes
in other quantities, such as linear dimensions (Berlincourt et al. 1964; Pan et al.
1989a; Brodeur et al. 1994) and optical properties (Chandani et al. 1989), also
accompany the phase transition. These changes at the phase transition provide
the physical foundation for important engineering applications of antiferroelec-
tric materials in digital displacement transducers (Berlincourt 1966; Uchino 1985;
Pan et al. 1989b), energy storage capacitors, electrocaloric cooling devices (Tut-
tle and Payne 1981; Mischenko et al. 2006; Scott 2011), and flat panel displays
(Fukuda et al. 1994; Takezoe et al. 2010). For most device applications, antiferro-
electric materials in the thin film form are required. As a result, films with various
antiferroelectric compositions have been deposited and characterized (Li et al.
1994; Qu and Tan 2006).
As illustrated in Figure 2.17b, with respect to the pseudocubic perovskite
primitive unit cell that contains
√ one PbZrO
√ 3 molecule, the dimensions of the
orthorhombic cell are a = 2ac , b = 2 2ac , and c = 2ac , where the parameter
ac is the lattice constant for pseudocubic unit cell.
The primary structural feature that accounts for the AFE is the antiparallel
displacements of Pb2+ along the pseudocubic ⟨110⟩ directions, as schematically
shown in Figure 2.17b. At room temperature, the average displacement of Pb2+
is ∼0.025 nm.
Energy storage is an important application of antiferroelectric material. As
illustrated in Figure 2.18a, the energy density W of dielectric capacitors is
4
E b
4
a
c
(a) (b)
Figure 2.17 (a) Typical P–E loop of antiferroelectric and (b) structure models for PbZrO3 -based
antiferroelectric. The orthorhombic unit cell is indicated by the box, while arrows represent the
direction of shifts of Pb ions along the c-axis. Source: Tan et al. (2011). Reprinted with
permission of John Wiley and Sons.
D D D
E E E
(a) (b) (c)
Figure 2.18 The typical electric displacement verses electric field (D–E) hysteresis loops and
energy storage characteristics of the three classes of solid dielectric materials (a) linear,
(b) ferroelectric, and (c) antiferroelectric.
represented by the shadow area that can be calculated by the following formula:
Pmax
U= E ⋅ dD = E dD (upon discharging)
∮ ∫Pr
where Pr is remnant polarization and Pmax is the maximum polarization under
bias electric field E whose change generates electric displacement dD. We can see
that the linear dielectric has relatively low energy density (∼0.01 J/cm3 ) because of
the low dielectric constant. The ferroelectric material owns large electric polar-
ization and hence has very high dielectric permittivity. However, when electric
fields excess the coercive field of the material, the permittivity decreases very
rapidly with the electric fields. Moreover, because of the hysteresis characteristic
of ferroelectric materials, the energy stored in ferroelectric during the charging
process cannot be fully released (see Figure 2.18b). Instead, some of the energy is
converted into heat due to the internal friction of dipole moment reorientation.
The shape of the double-hysteresis loop of antiferroelectric, as shown in
Figure 2.18c, can benefit to the energy storage in two aspects. First, because
of the AFE–FE phase transition, AFE can sustain a high dielectric constant
in relatively high electric field and hence absorbing a large amount of energy.
On the other hand, ferroelectric structures can turn back to AFE phase at low
electric field and release nearly all the energy.
2.5.3.1 Pb(Zrx Ti1−x )O3 (PZT)

When PbZrO3 is doped with PbTiO3 forming Pb(Zrx Ti1−x )O3 (PZT), its fer-
roelectric and piezoelectric properties reach maximum at their morphotropic
phase boundary (MPB); these outstanding properties made PZT dominant in the
commercial application in piezoelectric devices until today. What makes PZT so
significant in piezoelectric properties? The answer may be found from its phase
diagram as shown in Figure 2.19, which was established by Jaffe et al. (1971) and
Noheda et al. (2000).
One can see from PZT phase diagram, as the temperature T decreases, PZT
undergoes a paraelectric-to-ferroelectric phase transition, and the cubic unit cell
is distorted depending on the mole fraction of PbTiO3 . In the Zr-rich region, the
paraelectric phase changes to orthorhombic phase, and an intermediate mon-
oclinic phase exists between the Zr-rich rhombohedral and Ti-rich tetragonal
800
700 Cubic
600 20
–31
–32
Temperature (K)
500 –25
Rhombohedral(HT) –24
–25 19
MB 13
400 Tetragonal
MA
30
300
–25 –26 –22 –12 10 15 26 31
200
Rhombohedral(LT) 23
100 R3c/R3m+Cm
Orthorombic 16 19 22 30 R3m+P4mm+Cm
35
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Ti concentration
Figure 2.19 PZT phase diagram. The crossover between MB and MA structures is marked by a
dashed line. Tetragonal and rhombohedral phase regions are ferroelectric phases. Source:
Adapted from Zhang et al. (2014).
phases. That’s the point! One should notice that the best piezoelectric property
for PZT actually appears at the composition close to the MPB; one of the main
reasons for PZT to possess excellent piezoelectric property is the existence of
the MPB at which the monoclinic phase is the result of structural uncertainty
between the left side rhombohedra and the right side tetragonal structures. It
is because of this “uncertainty,” under a small applied electric field, the structure
distortion can easily happen between the tetragonal and rhombohedra phases, i.e.
non-180∘ polarization rotation, which is the key factor to induce very large strain
in the crystal. Of course, the large strain also comes from domain wall movement
that will be further discussed later.
2.5.4 Relaxor Ferroelectrics

Relaxor ferroelectrics were discovered about 50 years ago. The so-called relaxor
means a delayed dielectric response to high-frequency electric field, and as illus-
trated in Figure 2.20, a relaxor ferroelectric has three characteristics:
1. Phase transition is in a wide temperature range with its diffused dielectric
constant maximum at temperature T m , which represents the transition tem-
perature from ferroelectric phase to paraelectric phase (Figure 2.20a).
2. The ferroelectric relaxation also exhibits a frequency-dependent relative
permittivity, whereas its magnitude decreases with increasing frequency
(Figure 2.20a).
Figure 2.20 (a) 25 000

Frequency-dependent Tm
relative permittivity and (b)
P–E loops as a function of 20 000
temperature of relaxor
Dielectric constant
ferroelectrics.
15 000
1000 f
5000 100 kHz

10 kHz
1 kHz
0.1 kHz
0
–60 –40 –20 0 20 40 60 80 100
(a) Temperature (°C)
30
Electric displacement (μC/cm2)
20
10
0
24 °C
–10 46 °C
72 °C
96 °C
–20
124 °C
152 °C
–30 165 °C
–10 –8 –6 –4 –2 0 2 4 6 8 10
(b) Electric field (kV/cm)
3. Ferroelectricity Pr decreases gradually as temperature increases. Figure 2.20b

illustrates gradually decreased Pr when temperature increases. In fact, even
when temperature is higher than T m , there is still ferroelectricity because of
the existence of so-called nano-polar region (PNR).
2.5.4.1 Relaxor Ferroelectrics: PMN-xPT Single Crystal

Among many relaxor ferroelectrics, Pb(Mg1/3 Nb2/3 )O3 –xPbTiO3 (PMN-xPT
or PMN-PT for simplicity) single crystal is the most attractive one for its
outstanding ferroelectric/piezoelectric properties and engineering applications.
The outstanding properties of PMN-PT single crystal include its high piezo-
electric coefficient (d33 ∼ 2000 pC/N), large electromechanical coupling factor
(k 33 ∼ 92%), and large potential in application of novel devices including elec-
tromechanical sensors, high-performance ultrasound transducers and actuators,
and even infrared sensors. The most important advantage of PMN-PT is that it
can be grown into large size single crystal (at least 3 in.), making them perform
much better than PZT, which can only be used in a polycrystalline ceramic form.
Lead magnesium niobate, Pb(Mg1/3 Nb2/3 )O3 (PMN), is one of the most
interesting relaxor dielectric materials due to its quenched order–disorder
complex structure. The ordered and disordered distribution of Mg2+ and Nb5+
cations on the equivalent lattice B sites in PMN is the essential structural
characteristic of relaxor dielectrics. While lead titanate, PbTiO3 (PT), is a normal
ferroelectric material with long-range order domain structure and has a sharp
phase transition taking place at the Curie temperature T c = 490 ∘ C. PMN-PT
is the solid solution of Pb(Mg1/3 Nb2/3 )O3 and PbTiO3 , in which the substitution
of Ti4+ cations in B sites tends to reduce the random field and enhances the
size and formation of long-range ordered ferroelectric domains. PMN-PT has a
complex perovskite structure with an ABO3 -type unit cell and exhibits a relaxor
ferroelectric property.
2.5.4.2 Polar Nano Regions

One of the unique features of a relaxor ferroelectric is the presence of PNRs in the
crystal below the Burns temperature T B , which is a few hundred degrees above
the T m . Figure 2.21 illustrates the model of PNRs proposed by Cross (1987). The
existence of PNRs below T B ∼ 620 K in PMN is confirmed by inelastic neutron
scattering and X-ray diffraction around the reciprocal lattice points (Vakhrushev
et al. 1996; Naberezhnov et al. 1999; Hirota et al. 2002). The number of PNRs
increases and their dynamics slow down enormously during the sample cooling,
but at the freezing temperature T f , which is typically below the T m , the PNRs
will be frozen into a polarized state. At this polarized state, the average size of
PNRs increases, and a sharp decrease of the number of PNRs occurs due to the
merging of smaller PNRs into the larger ones. The static polarization can be
achieved at any temperature upon further cooling below T f .
It is generally accepted that the dynamic properties of relaxor ferroelectrics are
closely related to the degree of order/disorder and the unique nature of PNRs in
Ferroelectric Superparaelectric Paraelectric
Micro-polar clusters interact Micro-polar develop but Regions are nonpolar and
and material exhibits a have random orientation due material exhibits no
remanent polarization to thermal excitation macroscopic polarization
Increasing temperature
Figure 2.21 Polar mechanisms in relaxor ferroelectric materials as postulated by Cross (1987).
In the superparaelectric phase, the materials exhibit a nonzero root mean square (RMS) dipole
moment but have zero macroscopic polarization. Source: Cross (1987). Reproduced with
permission of Taylor & Francis.
the crystal. Although the information about PNRs can be indirectly measured by
neutron scattering, X-ray diffraction, high-resolution electron microscopy, etc.,
their evolution under different conditions cannot be directly observed and is still
unclear.
The PNRs with sizes around 1.5 and 18 nm in the Pb(Mg1/3 Nb2/3 )O3 and
Pb(Zn1/3 Nb2/3 )O3 crystals, respectively, at room temperature have been
reported (Gehring et al. 2001; Stock et al. 2004; Xu et al. 2004). Compared with
the size of PNRs in PMN determined by means of high-resolution electron
microscopy (Yoshida et al. 1998), PNRs size determined from elastic neutron
scattering is an order of magnitude smaller. Figure 2.22 shows piezoresponse
force (PFM) images of various compositions (001)-oriented PMN-xPT crystals
with different PT content x. It is apparent that the domain size reduces when the
content of PT is reduced, and when x is equal to 20%, the nanosized domains
dominant the domain structure.
The domain structure can also be observed under scanning electron micro-
scope (SEM) if the sample is treated by acid. Figure 2.23 is the laminate domain
structures of PMN-0.35PT crystal, where one can see that when the content of PT
PMN-0.20PT PMN-0.35PT PMN-0.40PT
2 μm
Figure 2.22 Piezoresponse force images of various compositions of (001)-oriented PMN-xPT

crystals. Source: Adapted from Bai et al. (2004).
Figure 2.23 SEM image of

laminate structure of PMN-0.35PT
crystal (courtesy of R.Y. Wang and
L.L. He).
5 μm
increases to more than 30% (x > 0.3), laminate micro-sized domains are formed
to make the crystal more like a normal ferroelectric.
2.5.4.3 Morphotropic Phase Boundary (MPB) of PMN-PT Crystal

PMN-PT single crystals have a complex phase diagram with rhombohedral
(R), tetragonal (T), orthorhombic (O), and monoclinic (M) phases as shown in
Figure 2.24, in which the shadow area with x = 0.28–0.36 of PT is termed as
MPB. When x < 0.28, a stable rhombohedral ferroelectric phase dominates in
the PMN-PT single crystal below the T m . On the other hand, when x > 0.36,
PMN-PT is tetragonal phase and behaves as normal ferroelectrics. By means of
synchrotron X-ray diffraction and neutron diffraction studies, temperature and
electric field-induced phase transitions from rhombohedral phase to monoclinic
phase have been found in the lower boundary of MPB. Fu and Cohen (2000) have
reported their result of first principle calculations, revealing that the very large
piezoelectric response is due to external electric field-induced phase transition
by polarization rotation via monoclinic phase.
After the discovery of a monoclinic phase near the MPB in PZT (Noheda et al.
1999), a large number of works of X-ray diffraction and neutron scattering stud-
ies on PMN-PT and Pb(Zn1/3 Nb2/3 )O3 -PbTiO3 (PZN-PT) have been reported.
Noheda et al. have reported the presence of ferroelectric monoclinic phase in
the MPB of PMN-PT system (Ye et al. 2001; Noheda et al. 2002). Three types
of monoclinic phases, MA , MB , and MC , have been identified in the MPB of elec-
tric field-induced phase diagram. The notations are adopted following Vanderbilt
and Cohen (2001). Figure 2.25 shows the polarization vectors in a pseudocubic
perovskite unit cell.
Pb(Mg1/3Nb2/3)TixO3
500
Cubic
400 Tm
Tetragonal
T (K)
R-O
300
M
Rhombohedral
200
0.1 0.2 0.3 0.4 0.5
X
Figure 2.24 Phase diagram of PMN-xPT. Source: Guo et al. (2002). Adapted with permission of
AIP Publishing.
2.6 Ferroelectric Domain and Phase Field Calculation 41
Figure 2.25 The polarization vectors in MA R

pseudocubic perovskite unit cell. The
T
polarization of MA phase in a plane
between rhombohedral (R) and tetragonal
(T) phases, the polarization of MB phase in Mc
a plane between R and orthorhombic (O) MB
phases, the polarization of MC phase in a
plane between O and T phases.
2.6 Ferroelectric Domain and Phase Field Calculation

When we investigate the ferroelectric domain structure and domain evolution
under electric field or temperature change, a simulation method called phase
field calculation is worth to be mentioned. Phase field calculation is a method to
simulate phase boundaries of different structures by systematically incorporating
effects of coherency strain induced by lattice mismatch and applied stress as well
as external electric and magnetic fields. This method has been applied to many
material processes including solidification, solid-state phase transition, and var-
ious types of complex microstructure changes. Due to the nature of ferroelectric
domains with different polarizations, the phase field calculation has also been
well used to simulate domain structures and evolutions in ferroelectrics where
different oriented ferroelectric domains can be treated as different phases.
Reversible ferroelectric domain switching controlled by thermal treatment has
been recently discovered by Ren (Ren 2004; Liu et al. 2006) in BaTiO3 -based fer-
roelectric materials. Figure 2.26 shows the schematic illustration explaining the
mechanism of the reversible ferroelectric domain switching, which has been pro-
posed by Ren (details of this mechanism is explained in their report).
In ferroelectric phase field calculation, the evolutions of the polarization
P (which is a function of time and space), i.e. the structural phase transition
Electric field
Electric field
Figure 2.26 Schematic illustration of the mechanism of the reversible domain switching. The
light and dark gray regions represent the ferroelectric domains having tetragonal crystal
structure with deferent orientations, respectively. The black arrows indicate the direction of
the c-axis of the tetragonal phase. The small white rectangles represent the spatial symmetry
of the point defects (dopant/impurity ions and oxygen-ion vacancies) in ferroelectrics. E appel is
an electric potential energy induced from the external electric field. Adapted from Koyama
and Onodera (2009).
Cubic Structure Tetragonal Structure
Electric field
50 μm 50 μm
Figure 2.27 Two-dimensional simulation of the structural transformation from cubic to

tetragonal phase in BaTiO3 at 298 K, and polarization domain microstructure change induced
by the external electric field applied along horizontal direction. Adapted from Koyama and
Onodera (2009).
and ferroelectric domain evolution, are described by the time-dependent

Ginzburg–Landau (TDGL) equation:
𝜕P 𝜕Gsys
= −LP
𝜕t 𝜕Pi
where LP is the relaxation kinetic coefficient and Gsys is the total free energy of the
𝜕G
system. The quantity 𝜕Psys is the thermodynamic driving force for the spatial and
i
time-evolution of polarization. The total free energy Gsys is expressed as (Koyama
and Onodera 2009)
[ ]
1 ∑
3
Gsys = Gc (Pi , T) + 𝜅P |∇Pi | dr + Estr + Edipole + Eappel
2
∫r 2 i=1
where Gc (Pi , T) is a bulk free energy density, Estr an elastic strain energy,
Edipole a dipole–dipole interaction energy among polarization moments in the
microstructure, Eappel an electric potential energy induced from the external
electric field, and the second term in the integrant is a ferroelectric domain wall
energy, which is expressed by the same form as the gradient energy in phase-field
method. 𝜅 P is a gradient energy coefficient. It is assumed that the strain field and
electric field are always at equilibrium for a given polarization field distribution.
T. Koyama and H. Onodera (2009) used this method to simulate BaTiO3
domain formation and evolution as ferroelectric phase at room temperature that
is well below Curie point. The results shown in Figure 2.27 illustrate the domain
formation developed into 180∘ and head to tail 90∘ domains. Under electric field,
180∘ domain switching and 90∘ domain rotation are realized by domain wall
movement.
References
Abe, K. and Komatsu, S. (1995). Ferroelectric properties in epitaxially grown
Bax Sr1−x TiO3 thin films. Journal of Applied Physics 77 (1995): 6461.
Bai, F., Li, J., and Viehland, D. (2004). Domain hierarchy in annealed (001)-oriented
Pb(Mg1/3 Nb2/3 )O3−x %PbTiO3 single crystals. Applied Physics Letters 85 (12):
2313–2315. Available at: https://doi.org/10.1063/1.1793353.
References 43
Barrett, J.H. (1952). Dielectric constant in perovskite type crystals. Physical Review
86 (1): 118–120.
Berlincourt, D. (1966). Transducers using forced transitions between ferroelectric
and antiferroelectric states. IEEE Transactions on Sonics and Ultrasonics 13 (4):
116–124.
Berlincourt, D., Krueger, H.H.A., and Jaffe, B. (1964). Stability of phases in modified
lead zirconate with variation in pressure, electric field, temperature and
composition. Journal of Physics and Chemistry of Solids 25 (7): 659–674. Available
at: http://www.sciencedirect.com/science/article/pii/0022369764901751.
Blundell, S.J. and Blundell, K.M. (2009). Concepts in Thermal Physics, 2e. Oxford:
OUP.
Brodeur, R.P., Gachigi, K., Pruna, P.M., and Shrout, T.R. (1994). Ultra-high strain
ceramics with multiple field-induced phase transitions. Journal of the American
Ceramic Society 77 (11): 3042–3044. Available at: https://doi.org/10.1111/j.1151-
2916.1994.tb04546.x.
Chandani, A.D.L., Gorecka, E., Ouchi, Y. et al. (1989). Antiferroelectric chiral
smectic phases responsible for the tristable switching in MHPOBC. Japanese
Journal of Applied Physics 28 (7): L1265–L1268.
Choi, T., Horibe, Y., Cheong, S.W. et al. (2010). Insulating interlocked ferroelectric
and structural antiphase domain walls in multiferroic YMnO3 . Nature Materials
9: 253. Available at: http://dx.doi.org/10.1038/nmat2632.
Cochran, W. (1960). Crystal stability and the theory of ferroelectricity. Advances in
Physics 9 (36): 387–423. Available at: https://doi.org/10.1080/
00018736000101229.
Cross, L.E. (1967). Antiferroelectric-ferroelectric switching in a simple ‘Kittel’
antiferroelectric. Journal of the Physical Society of Japan 23 (1): 77–82.
Cross, L.E. (1987). Relaxor ferroelectrics. Ferroelectrics 76 (1): 241–267. Available at:
https://doi.org/10.1080/00150198708016945.
Fu, H. and Cohen, R.E. (2000). Polarization rotation mechanism for ultrahigh
electromechanical response in single-crystal piezoelectrics. Nature 403: 281–283.
Available at: http://dx.doi.org/10.1038/35002022.
Fukuda, A., Takanishi, Y., Isozaki, T. et al. (1994). Antiferroelectric chiral smectic
liquid crystals. Journal of Materials Chemistry 4 (7): 997–1016. Available at:
http://dx.doi.org/10.1039/JM9940400997.
Gehring, P.M., Wakimoto, S., Ye, Z.G., and Shirane, G. (2001). Soft mode dynamics
above and below the Burns temperature in the relaxor Pb(Mg1/3 Nb2/3 )O3 .
Physical Review Letters 87 (27): 277601.
Graz University of Technology (n.d.). Landau theory of a first order phase transition.
Available at: http://lampx.tugraz.at/~hadley/ss2/landau/first_order.php (accessed
14 August 2019).
Guo, Y., Luo, H., Chen, K. et al. (2002). Effect of composition and poling field on the
properties and ferroelectric phase-stability of Pb(Mg1/3 Nb2/3 )O3 –PbTiO3
crystals. Journal of Applied Physics 92 (10): 6134–6138.
Hayashi, H., Nakamura, T., and Ebina, T. (2013). In-situ Raman spectroscopy of
BaTiO3 particles for tetragonal–cubic transformation. Journal of Physics and
Chemistry of Solids 74 (7): 957–962. Available at: http://www.sciencedirect.com/
science/article/pii/S0022369713000668.
Hemberger, J., Lunkenheimer, P., Böhmer, R. et al. (1995). Electric-field-dependent

dielectric constant and nonlinear susceptibility in SrTiO3 . Physical Review B 52
(18): 13159–13162. Available at: http://journals.aps.org/prb/abstract/10.1103/
PhysRevB.52.13159.
Hirota, K., Ye, Z.G., Wakimoto, S. et al. (2002). Neutron diffuse scattering from polar
nanoregions in the relaxor Pb(Mg1/3 Nb2/3 )O3 . Physical Review B 65 (104105): 1–7.
Jaeger, G. (1998). The Ehrenfest classification of phase transitions: introduction and
evolution. Archive for History of Exact Sciences 53 (1): 51–81. Available at: https://
doi.org/10.1007/s004070050021.
Jaffe, B., Cook, W.R. Jr., and Jaffe, H. (1971). Chapter 7. Solid solutions of Pb(Ti, Zr,
Sn, Hf )O3 . In: Piezoelectric Ceramics (eds. J.P. Roberts and P. Popper), 135–184.
London/New York: Academic Press.
Kittel, C. (1951). Theory of antiferroelectric crystals. Physical Review 82 (5):
729–732.
Koyama, T. and Onodera, H. (2009). Phase-field simulation of ferroelectric domain
microstructure changes in BaTiO3 . Materials Transactions 50 (5): 970–976.
Li, J., Viehland, D.D., Tani, T. et al. (1994). Piezoelectric properties of
sol–gel-derived ferroelectric and antiferroelectric thin layers. Journal of Applied
Physics 75 (1): 442–448. Available at: https://doi.org/10.1063/1.355872.
Liu, W., Chen, W., Ren, X. et al. (2006). Ferroelectric aging effect in hybrid-doped
BaTiO3 ceramics and the associated large recoverable electrostrain. Applied
Physics Letters 89 (17): 172908. Available at: https://doi.org/10.1063/1.2360933.
Luspin, Y., Servoin, J.L., and Gervais, F. (1980). Soft mode spectroscopy in barium
titanate. Journal of Physics C: Solid State Physics 13 (19): 3761. Available at: http://
stacks.iop.org/0022-3719/13/i=19/a=018.
Maris, H.J. and Kadanoff, L.P. (1978). Teaching the renormalization group.
American Journal of Physics 46 (6): 652–657. Available at: http://aapt.scitation
.org/doi/10.1119/1.11224.
Mischenko, A.S., Zhang, Q., Scott, J.F. et al. (2006). Giant electrocaloric effect in
thin-film PbZr0.95 Ti0.05 O3 . Science 311 (5765): 1270–1271. Available at: http://
science.sciencemag.org/content/311/5765/1270.
Müller, K.A. and Burkard, H. (1979). SrTiO3 : an intrinsic quantum paraelectric
below 4 K. Physical Review B 19 (7): 3593–3602.
Naberezhnov, A., Vakhrushev, S., Dorner, B. et al. (1999). Inelastic neutron
scattering study of the relaxor ferroelectric PbMg1/3 Nb2/3 O3 at high temperatures.
European Physical Journal B: Condensed Matter and Complex Systems 11 (1):
13–20. Available at: http://link.springer.com/10.1007/s100510050912.
Noheda, B., Cox, D.E., Shirane, G. et al. (1999). A monoclinic ferroelectric phase in
the Pb(Zr1−x Tix )O3 solid solution. Applied Physics Letters 74 (14): 2059–2061.
Available at: https://doi.org/10.1063/1.123756.
Noheda, B., Gonzalo, J.A., Caballero, A.C. et al. (2000). New features of the
morphotropic phase boundary in the Pb(Zr1−x Tix )O3 system. Ferroelectrics 237
(1): 237–244. Available at: https://doi.org/10.1080/00150190008216254.
Noheda, B., Cox, D.E., Shirane, G. et al. (2002). Phase diagram of the ferroelectric
relaxor (1−x)PbMg1/3 Nb2/3 O3 −xPbTiO3 . Physical Review B 66 (5): 054104.
Available at: https://link.aps.org/doi/10.1103/PhysRevB.66.054104.
References 45
Pan, W., Zhang, Q., Bhalla, A., and Cross, L.E. (1989a). Field-forced
antiferroelectric-to-ferroelectric switching in modified lead zirconate titanate
stannate ceramics. Journal of the American Ceramic Society 72 (4): 571–578.
Available at: https://doi.org/10.1111/j.1151-2916.1989.tb06177.x.
Pan, W.Y., Dam, C.Q., Zhang, Q.M., and Cross, L.E. (1989b). Large displacement
transducers based on electric field forced phase transitions in the tetragonal
(Pb0.97 La0.02 ) (Ti,Zr,Sn)O3 family of ceramics. Journal of Applied Physics 66 (12):
Qu, W. and Tan, X. (2006). Texture control and ferroelectric properties of
Pb(Nb,Zr,Sn,Ti)O3 thin films prepared by chemical solution method. Thin Solid
Films 496 (2): 383–388. Available at: http://www.sciencedirect.com/science/
article/pii/S0040609005017608.
Ren, X. (2004). Large electric-field-induced strain in ferroelectric crystals by
point-defect-mediated reversible domain switching. Nature Materials 3 (2):
91–94.
Sawaguchi, E., Maniwa, H., and Hoshino, S. (1951). Antiferroelectric structure of
lead zirconate. Physical Review 83 (5): 1078.
Scott, J.F. (2011). Electrocaloric materials. Annual Review of Materials Research 41
(1): 229–240. Available at: https://doi.org/10.1146/annurev-matsci-062910-
100341.
Shirane, G., Sawaguchi, E., and Takagi, Y. (1951). Dielectric properties of lead
zirconate. Physical Review 84 (3): 476–481.
Shirane, G., Axe, J.D., Harada, J., and Remeika, J.P. (1970). Soft ferroelectric modes
in lead titanate. Physical Review B 2 (1): 155–159.
Stock, C., Birgeneau, R.J., Wakimoto, S. et al. (2004). Universal static and dynamic
properties of the structural transition in Pb(Zn1/3 Nb2/3 )O3 . Physical Review B 69
(9): 094104.
Takezoe, H., Gorecka, E., and Čepič, M. (2010). Antiferroelectric liquid crystals:
interplay of simplicity and complexity. Reviews of Modern Physics 82 (1): 897–937.
Tan, X., Ma, C., Frederick, J. et al. (2011). The antiferroelectric ↔ ferroelectric phase
transition in lead-containing and lead-free perovskite ceramics. Journal of the
American Ceramic Society 94 (12): 4091–4107.
Tuttle, B.A. and Payne, D.A. (1981). The effects of microstructure on the
electrocaloric properties of Pb(Zr,Sn,Ti)O3 ceramics. Ferroelectrics 37 (1):
603–606. Available at: https://doi.org/10.1080/00150198108223496.
Uchino, K. (1985). Digital displacement transducer using antiferroelectrics.
Japanese Journal of Applied Physics 24: 460–462.
Uchino, K. (2009). Ferroelectric Devices, 2e. CRC Press.
Vakhrushev, S., Nabereznov, A., Sinha, S.K. et al. (1996). Synchrotron X-ray
scattering study of lead magnoniobate relaxor ferroelectric crystals. Journal of
Physics and Chemistry of Solids 57 (10): 1517–1523. Available at: http://www
.sciencedirect.com/science/article/pii/0022369796000224.
Vanderbilt, D. and Cohen, M.H. (2001). Monoclinic and triclinic phases in
higher-order Devonshire theory. Physical Review B 63 (9): 094108.
Vendik, O.G., Hollmann, E.K., Kozyrev, A.B., and Prudan, A.M. (1999). Ferroelectric
tuning of planar and bulk microwave devices. Journal of Superconductivity 12 (2):
325–338. Available at: https://doi.org/10.1023/A:1007797131173.
Wang, C. (2010). Theories and methods of first order ferroelectric phase transitions.
In: Ferroelectrics (ed. I. Coondoo). InTech. Available at: https://www.intechopen
.com/books/ferroelectrics/theories-and-methods-of-first-order-ferroelectric-
phase-transitions.
Xu, G., Shirane, G., Copley, J.R.D., and Gehring, P.M. (2004). Neutron elastic diffuse
scattering study of Pb(Mg1/3 Nb2/3 )O3 . Physical Review B 69: 064112.
Ye, Z.G., Noheda, B., and Dong, M. (2001). Monoclinic phase in the relaxor-based
piezoelectric/ferroelectric Pb(Mg1/3 Nb2/3 )O3 –PbTiO3 system. Physical Review B
64 (18): 184114.
Yoshida, M., Mori, S., and Yamamoto, N. (1998). Transmission electron microscope
observation of relaxor ferroelectric Pb(Mg1/3 Nb2/3 )O3 . Journal of the Korean
Physical Society 32: S993–S995.
Zhang, Q., Tan, G., Yu, R. et al. (2013). Direct observation of multiferroic vortex
domains in YMnO3 . Scientific Reports 3: 2741. Available at: http://dx.doi.org/10
.1038/srep02741.
Zhang, N., Yokota, H., Glazer, A.M. et al. (2014). The missing boundary in the phase
diagram of PbZr1−x Tix O3 . Nature Communications 5 (May): 5231.
Zhong, W. and Vanderbilt, D. (1996). Effect of quantum fluctuations on structural
phase transitions in SrTiO3 and BaTiO3 . Physical Review B 53 (9): 5047–5050.
47
Device Applications of Ferroelectrics
Some smart systems are based on ferroelectric materials; among their many
applications are actually the implementation of their piezoelectric properties
since most excellent piezoelectric materials are also ferroelectrics. The device
applications of piezoelectric property in smart systems will be introduced
in Chapter 7. In this chapter, device applications based on the change of
ferroelectric polarization will be introduced.
3.1 Ferroelectric Random-Access Memory

For memory application, owing to its nonlinear relationship between the applied
electric field and the stored charge, a ferroelectric layer sandwiched by two metal-
lic electrodes with a capacitor-like structure can be used as a non-volatile mem-
ory. When an external electric field is applied to it, the spontaneous polarization
can be reversed by an appropriate electric field applied in the opposite direction
of polarization. Therefore, binary data “0” and “1” can be written and stored by
its polarization directions.
Ferroelectric random-access memory (FeRAM), or called FRAM, is a
non-volatile memory where the non-volatility is achieved by the ferroelectric
polarization in the ferroelectric layer acting as the dielectric layer in either a one
transistor–one capacitor (1T1C) structure (see Figure 3.1a) or a one transistor
(1T) structure (see Figure 3.1b). It is a growing memory technology alternative
to flash memory with advantages such as lower power consumption, faster write
speed, and higher durability. With almost 30 years of development of FeRAM, it
started in mass production commercially. Fujitsu began its volume production
of FeRAM in 1999 and more companies are making multi-megabit density of
FeRAM for niche application where its performance is better than flash memory,
which is very popularly used in USB and SSD.
However, the readout operation in a conventional FeRAM is destructive as the
read method involves writing a bit to each cell. Thus, there are still limitations
in read/write cycle and access time. In addition, the depolarization field that is
opposite to that of polarization within the ferroelectric layer reduces the data
retention time significantly. The depolarization field is an electric field established

48 3 Device Applications of Ferroelectrics
Ferroelectric film
– – – – – – – – – –
Bit line
Word line
Word line
Bit line
N N
P
Plate line
(a)
1T
VR
1V FBB/
RBB
ID ≈ 100
μ A
Low-VT state
(b)
Figure 3.1 (a) FeRAM device and structural diagram (1T1C) made by Fujitsu, where the
ferroelectric layer’s polarization switching induces large difference in capacitance and
therefore the source–drain current representing “0” and “1” states. Source: Redrawn from
Fujitsu (n.d.). (b) FeRAM (or FeFET) from the Ferroelectric Memory Company and structural
diagram showing 1T structure. Source: Adapted from Ferroelectric Memory Company (2018)
and Arimoto and Ishiwara (2004).
inside the polarized layer by the generation of surface charge. This field’s direction
is opposite to the field that induces polarization and tends to depole the ferroelec-
tric layer; therefore, it is called depolarization field as illustrated in Figure 3.2.
Figure 3.3 shows 1T FeRAM diagram, where the gate dielectric layer is fer-
roelectric with polarization up and down representing the “0” and “1” states.
Figure 3.3a illustrates that the surface charge induced by ferroelectric polariza-
tion results in the shift of ON/OFF threshold voltage of field effect transistor
(FET) exhibiting a memory window.
The earlier studies of FeRAM were focused on PZT, since it has much larger
remnant polarization compared with other ferroelectric materials such as
BaTiO3 and PbTiO3 . However, PZT film faces serious fatigue problem, i.e. after
108 times of polarization switching, its ferroelectricity decays significantly. This
is not acceptable for non-volatile memory application (Al-Shareef et al. 1994).
This problem has hindered scientists quite a long time but attracted a lot of
interest of study. Later on, an alternative ferroelectric material SrBi2 Ta2 O9 (SBT)
has been found to be fatigue-free up to 1011 cycles of polarization switching.
3.1 Ferroelectric Random-Access Memory 49
Figure 3.2 Illustration of surface

charges generated in a
ferroelectric layer by polarization
P0 and depolarization field E d
induced by the generated
surface charges.
P0 Poling field
Ed
Gate
– – – –
Spontaneous polarization
of ferroelectric layer Memory
–––––––––
window
Source Drain
– – – – – – – – – – IDS
N N –––––––––
P
– – – –
VG
(a) (b)
Figure 3.3 (a) Diagram of the 1T FeRAM based on a field-effect transistor with ferroelectric
gate dielectric. (b) The ferroelectric polarization-induced interfacial charge alters the on–off
voltage of the FET resulting “0” and “1” states.
SBT also has a relatively smaller coercive field making it able to be operated with
lower voltage.
BiFeO3 (BFO) came into the focus of many scientists in the ferroelectric field
due to the discovery of its very large remnant polarization in thin film form,
which makes it very promising for FeRAM. However, it is very difficult to grow
high-quality BFO thin films with low leakage current and low dielectric loss. This
hinders the mass production of BFO-based FeRAM (Ielmini and Waser 2016).
An emerging ferroelectric material for FeRAM application is HfO2 , which has
been found to be ferroelectric when doped with ZrO2 or other oxides. Even the
ferroelectricity of HfO2 is usually not as good as PZT and BiFeO3 , but since it has
been implemented in CMOS technology as gate insulator, FeRAM with HfO2
gate oxide should be very compatible to current COMS process and therefore is
very attractive to semiconductor industry. However, the origin of ferroelectricity
is still not very clear, even it has been practically demonstrated to be ferroelectric.
Details will be introduced in Chapter 5.
3.2 Ferroelectric Tunneling Non-volatile Memory

With the development of FeRAM, in the meantime, Esaki introduced a concept
of ferroelectric tunnel junction (FTJ) in 1971. A FTJ is composed of two dissimi-
lar metal electrodes with a ferroelectric ultrathin film in-between as a barrier for
tunneling. According to quantum mechanical theory, quantum tunneling occurs
when the dielectric thickness in the FTJ is thin enough. It describes the situation
that there is a finite probability that the wave function may extend beyond the
barrier if the barrier is thin enough, i.e. the carrier can tunnel through the junc-
tion even its energy is below the barrier height. Unlike the classical physics law,
quantum tunneling depends on the potential barrier height. Carriers may tunnel
through the barrier proportional to the tunneling probability, which is a function
of barrier height. In particular, the thickness, bandgap, and polarization of the
ferroelectric film determine the height of potential barrier in the FTJ.
Such a structure takes advantage of ferroelectric polarization that can be
switched by an electric field, which also adds a new function to the tunnel junc-
tion. However, this novel idea requires a few nanometer-thick ferroelectric films
with strong polarization, which is very challenging. Since most people believed
that the critical thickness for ferroelectricity to exist was above 10–15 nm, no
further development or investigation had been done at that time. However,
followed by the rapid improvement of nanotechnology in the 2000s, nanoscale
characterization techniques, and the high-quality crystallinity in the growth of
ferroelectric thin film, this idea was raised again. Since then, there has been a
great attraction to scientists to study ferroelectric tunneling-induced resistive
switching in relation to ferroelectric thin films.
Apart from differentiating the charge polarity in FeRAM, the binary data of
new emerging ferroelectric-based memory depends on the resistance state, i.e.
electroresistance (ER). In a FTJ, its ER also depends on ferroelectric polariza-
tion. The origin of electroresistance effect is introduced by Zhuravlev et al. (2005)
with a simple model of ferroelectric film separating two different metal electrodes
(M1 and M2). To explain this mechanism, let us look at the energy band dia-
gram of a FTJ as shown in Figure 3.4. In this junction structure, the ferroelectric
film should be sufficiently thin but still maintains its ferroelectric polarization,
and the charges induced by polarization at the surface of polarization will repel
or attract electrons next to ferroelectric layer. These interfacial charges build an
electric field giving rise to an additional electrostatic potential at the interface and
therefore induce an asymmetric modulation of the potential barrier profile.
By solving the time-independent Schrödinger equation (3.1) for this junction,
ℏ2 d 2
− 𝛹 (x) + V (x)𝛹 (x) = E𝛹 (x) (3.1)
2m dx2
where ℏ is the reduced Planck constant, m the particle mass, x the distance in the
direction of motion of the particle, V the potential energy of barrier height, and
E the energy of the particle, we can obtain the wave function 𝛹 :
𝛹 (x) = Ae−ikx + Beikx (3.2)
√
with k = 2m(V ℏ2
−E)
δ1 Φ– Φ+
EF EF
δ2
M1 FE M2 M1 FE M2
P P
– + + –
Low barrier height (Φ–) Low barrier height (Φ+)
Figure 3.4 Energy band diagram of a ferroelectric tunnel junction. Polarizations in two
opposite directions cause two average barrier heights resulting in two resistance states. A
rectangular barrier, denoted by dashed lines is assumed when ferroelectric is nonpolarized.
Source: Garcia and Bibes (2014). Reprinted with permission of Springer Nature.
As the transmission of wave function is exponentially proportional to the

square root of barrier height (see Eq. (3.2)), for lower barrier height of potential
barrier, more carriers can pass through when an external voltage is applied;
therefore, the resistance of the junction is relatively lower and is called “ON”
state. On the other hand, when polarization changes to the opposite site, the
barrier becomes higher than normal (shown by dot line); as a result, fewer
carriers can pass through the barrier, and resistance of this system is relatively
higher (i.e. “OFF” state). Hence, the information can be stored because of this
barrier height difference. The ratio of electroresistance (ER) of OFF state to ON
state defines the efficiency of electrical resistance effect by Eq. (3.3):
I(high conduction) − I(low conduction) ROFF − RON
ER = = (3.3)
I(low conduction state) RON
3.2.1 Tunneling Models

In a simple metal–ferroelectric–metal (MFM) structure, electron transport
strongly depends on ferroelectric barrier properties, polarization orientation,
and metal properties. Pantel and Alexe (2010) calculated the influence of
possible factors on MFM current flowing density across the ferroelectric thin
film with three possible transport mechanisms including direct tunneling (DT),
Fowler–Nordheim tunneling (FNT), and thermionic emission (TE). Figure 3.5a
illustrates the transport mechanisms through an ultrathin ferroelectric layer.
ER
5.0 1.0
TE 105 TE TE
102 FNT 4.0 0.5
j (A/cm2)
d (nm)
10–1
FNT
FNT
FNT 3.0 0.0
10–4 DT
DT
10–7 2.0 –0.5
DT TE
–1.5 –1.0 –0.5 0.0 0.5 1.0 1.5 –1.0
.5
.0
.5
0
5
1.0
1.5
0.
0.
V (V)
–1
–1
–0
(a) (b) (c) V (V)
Figure 3.5 (a) Illustration and (b) voltage dependence of the current density contributions for
3.2 nm-thick BaTiO3 of three possible transport mechanisms through ultrathin ferroelectrics:
direct tunneling (DT), Fowler–Nordheim tunneling (FNT), and thermionic emission (TE). Solid
and dashed lines stand for two polarization orientations. (c) Corresponding electroresistance
(ER) as a function of voltage and ferroelectric thickness d. The dashed lines illustrate the
boundaries where each DT, FNT, and TI mechanism dominates. Source: Pantel and Alexe
(2010). Reprinted with permission of American Physical Society.
The direct tunneling (DT) is a quantum-mechanism phenomenon, whereas

electron flow can be estimated in a trapezoidal potential barrier using
Wentzel–Kramers–Brillouin (WKB) approximation. Current characteris-
tics at low voltage in different temperatures can be fitted into Brinkman’s model
and Gruverman’s model. The fitting determines the presence of direct tunneling
through the barrier with polarization-induced interfacial effects. Also, mean
potential barrier height is calculated using the following equation:
{ [( )3 ( ) 3 ]}
exp 𝛼 𝜑2 − eV2 − 𝜑1 + eV2
2 2
IDT = SC [( )1 ( )1 ]
𝛼2 𝜑2 − eV2 − 𝜑1 + eV2
2 2
{ [( )1 ( ) 1 ]}
3eV eV 2 eV 2
× sinh 𝛼 𝜑2 − − 𝜑1 + (3.4)
4 2 2
with
32𝜋em∗
C=− (3.5)
9h3
and
1
8𝜋d(2m∗ ) 2
𝛼= (3.6)
3h(𝜑1 + eV − 𝜑2 )
where I DT is the direct tunneling current given by Gruverman et al. (2009), S the
area of the interface, d the film thickness, h the Planck constant, m* the effective
carrier mass, and Φ1 and Φ2 are potential steps at the two interfaces.
Similar to DT, FNT shares the same physical phenomenon but in different
voltage regime. It is the process whereby electrons tunnel through a barrier in
the presence of a high electric field. Across a triangular-shaped potential barrier,
FNT is present. The tunnel probability can be derived from time-independent
Schrödinger equation, and the calculation of current yields the following
relationship between the current density J FN and the electric field in the barrier E:
( √ 3 )
2m ∗ (e𝜑 ) 2
4 B
JFN = CFN E2 exp − (3.7)
3 eℏ E
where 𝜑B is the barrier height at the conductor/barrier interface.
TE describes the electron flow as the charge carriers that overcome the poten-
tial barrier by thermal energy. Usually, the I–V curves can be fitted by TE model.
The forward bias currents are described by the theory of TE as the following
equation:
( e𝜑 ) [ ( ) ]
eV
Iforward = SA∗ T 2 exp − B exp −1 (3.8)
kT nkB T
where S is the junction area, A* the Richardson constant, T the absolute temper-
ature, e the electron charge, k the Boltzmann constant, n the ideal factor, and 𝜑B
the Schottky barrier height. From the linear relationship in TE modeling, we can
identify the barrier height in OFF state using Eq. (3.8). The corresponding deple-
tion region width, where electrons are depleted in interface, can be estimated by
Eq. (3.9):
[ ]1
2𝜖s 2
WD = (Vbi − V ) (3.9)
eND
where W D is the depletion region width, 𝜖 s the dielectric constant of semicon-
ductor, N D the donor concentration, V bi = 𝜑B + 𝜑n , and 𝜑n = EF − Ec .
In Figure 3.5b, the current density as a function of voltage across 3.2-nm thick
ferroelectric thin film is shown. For the two opposite polarizations (P > 0 and
P < 0 with solid and dashed lines, respectively), the current density behaves
differently in three mechanisms. DT is prominent in low voltage regime, while
FNT is in high voltage regime. With a small range of voltage, DT gives a parallel
current branch for both polarization directions showing the independence of
applied voltage on electroresistance. On the other hand, FNT is barely observed
in low voltage. Because of insufficient band tilting, there is virtually no current
recorded in low voltage regime. In the case of TE, electroresistance is obvious
as its value is given by the change in barrier height on polarization reversal.
However, the total current density decreases compared with the other two.
Therefore, it is usually suitable for thicker films that tunneling cannot happen
(see Figure 3.5c).
Among these three mechanisms, we can understand the possibility in elec-
tronic transportation across ferroelectric thin film. Certainly, parameters related
to those mechanisms, such as effective mass of oxide film in direct tunneling and
FNT as well as temperature in TE, can shift between mechanisms. I–V mea-
surement is the first step to see a ferroelectric junction’s memory characteristics.
However, I–V curves alone are not sufficient to clarify the underlying switching
mechanism that should be attributed to polarization switching.
By performing sequential ferroelectric poling, piezoresponse force microscopy
(PFM) imaging, and conductive-atomic force microscope (C-AFM) imaging
of the poled domains, the resistance difference measured with the C-AFM or
other electrical methods between positively and negatively poled regions would
therefore correspond to the ferroelectric-induced ER effect (Gruverman et al.
2009). Garcia et al. (2009) investigated the local tunneling electroresistance
based on this method and demonstrated resistive readout of the polarization
state through its influence on the tunnel current using PFM and C-AFM at room
temperature even the film is a 1 nm-highly strained BaTiO3 film. The resulting
electroresistance effect scales exponentially with ferroelectric film thickness
and it can reach 75 000% at 3 nm thick. In fact, only by tunneling at such thin
thickness is able to read the polarization state without destroying it, otherwise
undesirable leakage current will dominate. Apart from using conductive tips as
top electrodes, Chanthbouala et al. (2012b) reported tunnel electroresistance on
Au/Co/BaTiO3 (2 nm)/LSMO(30 nm)/NdGaO3 heterostructure with resistance
switching occurring at the coercive voltage of ferroelectric switching. They
show an alternative way to ensure that the electroresistance is originated from
ferroelectric response (Figure 3.6).
Figure 3.6 Out-of-plane PFM

90 (a) phase and (b) amplitude
measurements on
45 Au/Co/BaTiO3 /LSMO/NdGaO3
Phase (°)
heterostructure showing
0 ferroelectric switching versus
resistive switching. (c)
–45 Resistance versus writing
voltage measured in
–90 remanence after applying
(a) successive voltage pulses. The
open and filled circles
represent two different scans
Amplitude (a.u.)
to show reproducibility.
10
Source: Adapted from
Chanthbouala et al. (2012b).
1
(b)
107
Resistance (Ω)
106
105
(c)
–4 –3 –2 –1 0 1 2 3 4
Vwrite (V)
3.2.2 Metal–Ferroelectric–Semiconductor Tunnel Junction

Wen et al. (2013) proposed to replace one of the metal electrodes with a heavily
doped semiconductor to form a metal–ferroelectric–semiconductor (MFS)
heterostructure. Since the transmittance of a tunnel junction depends on both
barrier height and width exponentially, the tunneling electroresistance can be
enhanced by tuning the depletion region in semiconductor surface by the fer-
roelectric field effect. Figure 3.7a shows the schematic drawing of MFS structure
with corresponding potential barrier. At ON state, the polarization is pointing
toward the semiconductor electrode, and the positive bound charges in the ferro-
electric thin layer near the ferroelectric/semiconductor interface drive the n-type
semiconductor surface into accumulation. In this situation, the semiconductor
layer can be treated as a metal electrode and the junction is similar to a MFM FTJ.
With the development of depolarization field, the barrier height is lower than that
in metal/ferroelectric interface, and thus more electrons can tunnel through the
thin barrier layer. While at the OFF state (polarization pointing to the metal elec-
trode), the negative bound charges in ferroelectric have to be screened by immo-
bile ionized donor within the depletion region in the surface of semiconductor. In
contrary to the ON state, depolarization field increases the barrier height in fer-
roelectric/semiconductor interface. Also, due to the presence of depletion region,
electrons have to overcome both increased barrier height and barrier width
induced by ferroelectric field effect, and therefore, tunneling current is greatly
reduced. Figure 3.7b shows the resistance changes in two states based on this phe-
nomenon. With an increased thickness of depletion layer in the semiconductor
electrode, a giant electroresistance can be obtained even at room temperature.
M F S 107
P
wr
w
106 OFF
EF
Resistance (Ω)
EC
EV 105
ON OFF
M F S 104
ON
r w
103
EF P
EC w
EV
–5 –4 –3 –2 –1 0 1 2 3 4 5
(a) (b) Voltage (V)
Figure 3.7 (a) Schematic drawings of the metal/ferroelectric/semiconductor structures and

corresponding potential energy profiles for the low (top) and the high (bottom) resistance
states. (b) Resistance hysteresis loops at room temperature. The loop composed of dark
squares and circles is measured using the pulse train shown schematically in the bottom-left
(top-right) inset. The device is preset to the ON (OFF) state by a positive (negative) pulse. The
corresponding domain evolution is shown schematically in the bottom-right, top-left, and
middle-left insets for the ON, OFF, and intermediate (ON → OFF) states, respectively. Source:
Wen et al. (2013). Reprinted with permission of Springer Nature.
3.2.3 Ferroelectric Tunneling Memristor

Chanthbouala et al. (2012a) introduced a memristive behavior that the ratio of
two resistance changes can be controlled by the nucleation and growth of ferro-
electric domains. They mentioned an important degree of freedom that has not
yet been exploited in FTJs, i.e. the domain structure of the ferroelectric tunnel
barrier. As we know that the domain size in ferroelectrics scales down with film
thickness, so that nanometer-sized domains are expected for ferroelectric tunnel
barriers.
By applying different positive writing voltage pulses on the virtually homo-
geneous up-polarized NiFe/BaTiO3 (5 nm)/LSMO heterostructure, the voltage
starts nucleating the down-polarized domain. From our research results as
shown in Figure 3.8a, one can see the multiple logic states of resistance when
the amplitude of writing voltage pulse increases. This result suggests that
the ferroelectric tunneling-based memory behaves like a memristor whose
1.0 V
1.5 V
2.0 V
1000 2.5 V
OFF
Resistance (kΩ)
100
ON
10
–4 –3 –2 –1 0 1 2 3 4
(a) Voltage (V)
2.5 V 0V 0V
–2.5 V 1.0 V 2.0 V
(b)
Figure 3.8 (a) Tunneling resistance with varying maximum (positive or negative) writing
voltages and (b) configuration extracted from the PFM phase images showing the relative
fraction of ferroelectric domains pointing downward with increased voltage. Source:
Reprinted with permission from Yau et al. (2015).
resistance states can be continuously tuned by writing voltage. The correspond-

ing ferroelectric domain orientation at different levels of resistance is shown in
Figure 3.8b, where one can see that the degree of polarization switching depends
on writing voltage applied. These results suggest that the multiple resistance
states can be modified by controlling the nucleation and growth of ferroelectric
domains, i.e. memristive characteristics. A memristor is considered to be the
fourth electrical component whose resistance has multiple values that can be
altered by writing voltage with different magnitudes or different time sequences.
The so-called artificial neural network is built with these memristors that are
also called artificial synapses since their electrical characteristics are similar to
biological synapses in our neuron system. Therefore, in recent years, scientists
are trying to implement these artificial synapses into artificial neural network to
realize artificial intelligence (AI) function.
3.2.4 Strain Modulation to Ferroelectric Memory

Recently, we have demonstrated the strain engineering in the Pt/BaTiO3 /
Nb:SrTiO3 metal/ferroelectric/semiconductor heterostructured FTJ. As shown
in Figure 3.9, with an extra compressive strain induced by mechanical stress
that is applied beyond the lattice mismatch, the ON/OFF ratio of FTJ can be
increased significantly up to an extremely high value of 107 ; while a mechanical
erasing effect can be attained when a tensile stress is applied. This pure strain
modulation to ferroelectricity and therefore the memory state within a simple
structure avoids other effects, and therefore, verifies the role of ferroelectricity
in FTJ as a non-volatile resistive memory. This external strain engineering also
provides a way to tune resistive switching including ON/OFF ratio and other
performance for future development in non-volatile memory application.
107
Pt 106
BTO
105
NSTO
ON/OFF ratio
104
In-plane
strain ε
103
4.30%
2.18% 102
101
0.18%
c 100
Ba 2+ –4.0 –3.0 –2.0 –1.0 0.0
Ti 4+
a O 2– Strain ε (%)
(a) (b)
Figure 3.9 (a) Illustration of in-plane strain effect and (b) the corresponding ON/OFF ratio of
Pt/BaTiO3 /Nb:SrTiO3 metal/ferroelectric/semiconductor heterostructured ferroelectric tunnel
junction. Source: Adapted from Yau et al. (2017).
3.3 Pyroelectric Effect and Infrared Sensor Application

3.3.1 Pyroelectric Coefficient
Another important application of a ferroelectric material is the implementation
of its pyroelectric property in infrared (IR) sensors. Compared with semiconduc-
tor infrared sensors, advantages of pyroelectric sensors are their wide wavelength
response, uncooled operation, high sensitivity, compact structure, and low cost.
These advantages enable pyroelectric-based infrared sensors a variety of appli-
cations such as temperature sensing, flame and fire detection, gas monitoring,
energy harvesting, night vision, and thermal imaging (Rogalski 2003; Whatmore
1986; Yang et al. 2012; Tien et al. 2009).
Infrared (IR) radiation exists in the electromagnetic spectrum at a wavelength
longer than visible light. To detect infrared radiation, a transducer is required
to convert the infrared radiation to electrically detectable signal with electrical
circuits. Traditional ferroelectric materials, including triglycine sulfate (TGS),
lithium tantalite (LiTaO3 ), and barium strontium titanate (BST) have been
utilized for infrared detector applications (Watton 2011; Buser et al. 1997).
Recently, the superior pyroelectric performance of PMN-PT single crystal
makes it a promising candidate for high-performance uncooled infrared detec-
tors and thermal imagers. The pyroelectric properties of PMN-PT can be nearly
independent of incident beam frequency in the range of 50 Hz to 10 kHz (Yang
et al. 2012).
The pyroelectric effect can be quantified by the pyroelectric coefficient p, which
is defined as
dPs
p= (3.10)
dT
With a pyroelectric coefficient p, a small and gradual temperature change ΔT in
a pyroelectric crystal will cause a change in the spontaneous polarization vector
ΔPS given by
ΔPS = pΔT (3.11)
Given the pyroelectric coefficient, the pyroelectric current induced by tempera-
ture change can be derived by
( )
dPs dPs dT dT
j=− =− × =p (3.12)
dt dT dt dt
Ferroelectric materials have negative pyroelectric coefficients, i.e. an increase in
temperature leads to a decrease in the spontaneous polarization. The polarization
will suddenly fall to zero if the crystal is heated above its Curie temperature. The
temperature-dependent spontaneous polarization Ps of a ferroelectric material
makes the pyroelectric sensing element generate surface electric charges when
exposed to heat in the form of infrared radiation. However, one should know
that a pyroelectric infrared sensor can only detect temperature change but has
no response to a steady temperature since a pyroelectric sensing element can
only sense polarization change when the sensor is subject to temperature change.
However, too fast change of radiation makes the pyroelectric effect reduced due
Thermometer
Electrometer
Thermocouple
Sample
PC Driver Peltier element
Lock-in amplifier
Signal-out Reference Signal-in
Figure 3.10 Schematic setup of dynamic pyroelectric coefficient measurement system.
to slow change of temperature restricted by heat capacity and heat conductiv-

ity. So, there is a frequency limit of temperature change (usually up to 1 kHz,
depending on the size of sensing element and electrode material, thickness, etc.).
The pyroelectric coefficient is usually determined from a dynamic measure-
ment, and its schematic setup is shown in Figure 3.10. This dynamic measurement
method is based on the measurement of pyroelectric current induced by a sinu-
soidal variation of temperature of a heater using a thermoelectric Peltier device.
As the pyroelectric current is proportional to the partial derivative of the sample’s
temperature, the pyroelectric coefficient p can be obtained:
( ) ( )
𝜕P 1 ΔQ 1 dQ∕dt I
p=− =− =− = (3.13)
𝜕T A ΔT A dT∕dt A(dT∕dt)
The pyroelectric current is recorded by an electrometer that acts as a current
to voltage converter, and the amplitude of the voltage is measured by a lock-in
amplifier. The sample’s temperature is sinusoidal modulated around 30 ∘ C with a
frequency in the range of a few mHz. After the signal becomes stable, the reading
of the pyroelectric coefficient can be recorded. The pyroelectric coefficients of
some ferroelectric materials are presented in Table 3.1.
3.3.2 Pyroelectric Infrared Sensor

A pyroelectric sensor consists of a slice of pyroelectric material with metal
electrodes on opposite faces. A ferroelectric material should be poled so that
the polarization direction is perpendicular to the electrode faces and the
material forms a parallel-plate capacitor. The electrode configuration is shown
in Figure 3.11a, and the electric polarization change under thermal radiation is
illustrated in Figure 3.11b,c.
Table 3.1 Pyroelectric coefficient of some typical pyroelectric materials.
Pyroelectric coefficient
Material (𝛍C/(m2 K)) References
PMN-0.26PT (111) 1530 Tang and Luo (2009)

PMN-0.29PT (111) 1280 Tang et al. (2005)
Mn-PMN-0.26PT 1720 Liu et al. (2009)
Mn-PMN-0.29PT 1620 Tang et al. (2006)
LiTaO3 230 Kulwicki et al. (1992)
PZT 330 Kulwicki et al. (1992)
– – – – – – – – – – – – – – – – – –
+ + + + + + + + + + + + + + + + + +
+ + + – + – + + + + – +
+ + +
P – – – P – –– P – – –
– – – – – – – – – – – – – – – – – –
+ + + + + + + + + + + + + + + + + +
IP IP
dT dT dT
0 0 0
dt dt dt
(a) (b) (c)
Figure 3.11 (a) Diagram of pyroelectric infrared sensor, (b) decreased polarization and change
of surface charges when thermal radiation is on, and (c) increased polarization and change of
surface charges when thermal radiation is off. The ellipsoids are only guidance for electric
polarization change.
Since the change of polarization in a solid is accompanied by the change in

number of surface charges, it can be detected by measuring induced current in an
external circuit. If the pyroelectric material is perfectly electrical insulated from
its surroundings, the surface charges are eventually neutralized by charge flow in
the circuit.
Infrared radiation is radiated at the front surface that is usually “blackened” by
black gold (a gold film grown at very poor vacuum is black like a carbon film) or
graphite so that the efficiency of heat absorption can be maximized. The heat dif-
fuses through the graphite layer or gold electrode into the pyroelectric material
and causes finite temperature change. When the amount of radiation striking the
crystal changes, the number of charges associated with the spontaneous polariza-
tion also changes and a small pyroelectric current is thus induced. The current
can then be amplified and measured with a sensitive FET device built into the
sensor. TO5 package is a typical kind of electronic packages that is commonly
applicable to pyroelectric sensors. Poled pyroelectric materials are protected and
isolated from the surrounding by TO5 package, which can significantly reduce the
noise. Figure 3.12 shows the device structure and circuit.
3.3.3 Pyroelectric Figures of Merit

Figure of merit (FOM) is an important parameter to assess the performance of a
pyroelectric sensor. In order to select a suitable pyroelectric material as a sensing
+5 V
Amplifier
Pyroelectric
elements R1 R2
(a) (b)
Figure 3.12 (a) Basic structure and (b) circuit of a pyroelectric detector.
element to fabricate an infrared (IR) sensor, many factors such as the operat-
ing temperature, frequency, and pyroelectric response should be considered. It
is possible to formulate a group of FOMs to assess the performance of IR device.
The current FOM, F i , which describes the influence of the pyroelectric material
on the current responsibility of a pyroelectric sensor, is defined as
p
Fi = (3.14)
𝜌Cp
where p is the pyroelectric coefficient, 𝜌 is the density of the sample, and C p is
the heat capacity.
The voltage FOM, F v , which evaluates the influence of the pyroelectric-material
on the voltage responsibility of a pyroelectric sensor, is expressed as
p
Fv = (3.15)
𝜌Cp 𝜖0 𝜖r
where 𝜖 0 and 𝜖 r are the permittivity of free space and relative permittivity of the
sample, respectively.
The detectivity FOM, F d assesses the influence of the pyroelectric material on
the detection sensitivity of a pyroelectric sensor and is given by
p
Fd = √ (3.16)
𝜌Cp 𝜖0 𝜖r tan 𝛿
where tan 𝛿 is the dielectric loss tangent.
This equation tells us that to make a pyroelectric sensor with high performance,
the pyroelectric coefficient should be as large as possible; other parameters such
as heat capacity, dielectric permittivity and dielectric loss, and specific weight
should be the smaller the better.
PMN-PT single crystal has superior pyroelectric performances to make it
a promising candidate for high-performance uncooled infrared detectors and
thermal imaging. From Table 3.1, the pyroelectric coefficient of PMN-PT single
crystal is proven to be larger than some classical ferroelectric materials such as
PZT ceramic. However, the measured pyroelectric coefficients of PMN-PT are
Table 3.2 Pyroelectric figures of merit of some typical pyroelectric materials.
Material F i (pm/V) F v (m2 /C) F d (𝛍Pa−1/2 ) References
PMN-0.26PT 610 0.11 153 Tang and Luo (2009)

PMN-0.29PT 525 0.11 98 Tang et al. (2005)
Mn-PMN-0.26PT 688 0.12 402 Liu et al. (2009)
Mn-PMN-0.29PT 648 0.11 371 Tang et al. (2006)
LiTaO3 72 0.17 157 Kulwicki et al. (1992)
PZT 122 0.02 12 Kulwicki et al. (1992)
far away from another Mn-doped PMN-PT single crystal. Hence, the pyroelec-
tric coefficient is varied from composition and crystal orientation. (Students:
Think how to choose the crystal orientation for largest pyroelectric response.)
Table 3.2 summarizes PMN-PT materials’ group of FOMs with comparison of
the other ferroelectric materials. We can see that PMN-PT relaxor ferroelectric
crystal has superior FOM compared with the ever best pyroelectric LiTaO3 (Li
et al. 2014; Tang and Luo 2009; Tang et al. 2005, 2006; Liu et al. 2009; Kulwicki
et al. 1992).
Another method to characterize the pyroelectric effect is to use infrared laser
with a chopper and lock-in amplifier. As illustrated in Figure 3.13, an infrared
sensor based on pyroelectric effect can be characterized with infrared radiation
from an incident of infrared laser light, and current or charges can be measured
Pyroelectric materials
Absorption layer
IR laser
A
Chopper
Lock-in
amplifier
IR laser Pyroelectric materials

Absorption layer
IR pulse
Signal generator
Lock-in
amplifier
Figure 3.13 Experimental setup for infrared sensor characterization.

3.4 Application in Microwave Device 63
MoS2
Au e– e– e– e– e– e– Au
MoS2
PMN-PT
S D – – – – – –
C/Au PMN-PT C/Au
MoS2
ITO Au e– e– e– Au
PMN-PT
IR – – –
C/Au
(a) (b) IR
0.018
6 mW/mm2 IR off VD = 0.5 V
0.016
Current (μA)
0.014
0.012
0.010
IR on
0.008
0 200 400 600 800
(c) Time (s)
Figure 3.14 (a) A schematic view of the ferroelectric gate FET, (b) the working mechanism of
the ferroelectric field-effect transistor modulated with IR illumination, and (c) the
time-resolved photocurrent in response to IR on/off at an irradiance of 6 mW/mm2 with
1064 nm laser. Source: Reprinted with permission from Fang et al. (2015). Copyright 2015, The
Optical Society.
by lock-in technique. In the setup, the alternative laser incident can be realized
by a laser chopper or by a laser beam excited by sinusoidal or square wave of
electrical input that is also used as the reference input to the lock-in amplifier.
Figure 3.14 shows our result of an example to integrate pyroelectric effect
with a semiconducting 2D material to form an infrared sensor. This device uses
pyroelectric-induced field effect to modulate the surface conductivity of 2D
MoS2 to enable it to detect infrared radiation, or otherwise it is only sensitive to
visible or UV light.
3.4 Application in Microwave Device

Compared with BaTiO3 , ferroelectric (Bax Sr1−x )TiO3 (BST) exhibits much larger
dielectric constant change with applied dc electric field. This unique property
of BST is in favor to the application in the miniaturization of tunable microwave
devices. The development of microwave devices in voltage-controlled oscillators,
tunable filters, and phase shifters are currently based on this property. Recently,
BST thin films have been employed in many applications, for example, the BST
tunable capacitor used for matching networks in antenna and amplifier. By chang-
ing its capacitance as bias voltage changes, it can tune antennas, improve the
effectiveness of the phone’s antenna, and help the phone to avoid dead spots in
order to improve performance especially in regions with low signal. The elec-
tronically steerable system saves space and weight in satellite communication and
radar system. Not limited to communication technology, BST thin films also have
the potential in integration with semiconductor microelectronic circuits, includ-
ing microwave tunable switches and resonators (Figure 3.15) (Terrence 2013).
Although BST is currently competing with several materials used in the
market, such as semiconductor GaAs and micro-electrical–mechanical sys-
tems (MEMS), the outstanding properties of BST still shows its value in the
development. Table 3.3 shows the comparison of the properties of GaAs, MEMS,
BST Gap
Patch
Substrate
Ground plane
0
Reflection magnitude (dB)
–5
–10
BL- Corner
–15
BL- Emin
BL- Middle
–20 No BL
–25
9 9.5 10 10.5 11 11.5 12
Frequency (GHz)
(a) (b)
Figure 3.15 (a) Antenna devices with tunable BST layer acting as a distributed capacitance
and corresponding reflection (Karnati et al. 2015). (b) The wafer-level antenna devices with BST
thin films. Source: Adapted from Terrence (2013).
Table 3.3 Comparison of the properties of GaAs, MEMS, and BST thin film.
GaAs MEMS BST
Tunability n ∼2–6: 1 ∼1.5–3: 1 ∼2–4: 1

Loss ∼20–50 at 10 GHz Very high ∼20–100 at 10 GHz
Control voltage <15 V <50 V <15 V
Tuning frequency ∼1 MHz >100 kHz ∼1 GHz
Reliability Good Poor Good
Cost High High Low
Power consumption Poor Good Good
Source: Bao et al. (2008). Reproduced with permission of IOP.

3.5 Ferroelectric Photovoltaics 65
and BST. We can see that GaAs and BST show the highest tunability and lowest
dielectric loss even though MEMS is at a similar level of development maturity
as BST. However, the linear decrease of the quality factor with frequency has
limited the applications of GaAs in tunable filters for f > 1 GHz. Additionally,
they are more expensive and consume more power than ferroelectric thin films
in fabrication and operation.
3.5 Ferroelectric Photovoltaics

In 1956, BaTiO3 was found to have weak current under photo illumination. This
is the first time that oxide perovskite was recorded as a kind of photosensitive
material. In 1976, Chen reported his observation of an optically induced refrac-
tive index change in poled ferroelectric crystals of LiNbO3 and LiTaO3 (Chen
1969), making oxide perovskites a possible alternate candidate in photovoltaic
(PV) applications (Choi et al. 2009; Yang et al. 2010; Cao et al. 2012). As shown
in Figure 3.16, the built-in electric field generated from the uniformly polarized
ferroelectric oxide run through the thin film device and would enhance the exci-
ton separation, charge diffusion, and the open circuit voltage (Qin et al. 2008).
However, efficiency obtained from the photovoltaic effect is very low, like 10−7 to
10−5 , until an efficiency of 0.57% was obtained from the judiciously engineered
ferroelectric LiNbO3 crystals (Cao et al. 2012).
The idea of using ferroelectric field to enhance PV effect has been realized in
the P(VDF-TrFE)/polymer organic solar cells as shown in Figure 3.17, where the
poled P(VDF-TrFE) copolymer provides a local electric field to enhance the sep-
aration of electrons and holes in the organic semiconductor layer (Yuan et al.
2011). With such approach, the solar cell has shown an increase of power con-
version efficiency from 1–2% to 4–5%, which is more effective than methods like
morphology and electrode work-function optimization.
Let us look at an example of photovoltaic effect of ferroelectric BNBT:Nb
ceramics PV performance (see Figure 3.18) (Chen et al. 2014). It is found that
Nb doping can significantly enhance the PV effect in BNBT ceramics but at the
expense of deteriorating the ferroelectric and piezoelectric properties.
Light
Top electrode
Ferroelectric
– – – – –
P E thin film
+ + + + +
Bottom electrode
Hole
Electron
Figure 3.16 Schematic illustration of the physical mechanism of the photovoltaic effect in a
ferroelectric.
ITO
Al
Al
Polymer: PCBM
P(VDF-TrFE)
Figure 3.17 The structure of polymer photovoltaic devices with FE interfacial layers and a
schematic diagram of electric field induced by the polarized FE layer and the field-assisted
charge extraction. The right-hand panel illustrates the electric-field distribution and electron
conduction through the P(VDF-TrFE) on the Al side. Source: Yuan et al. (2011). Reprinted with
permission of Springer Nature.
200
Dark
Light
150 A Polarization up
ITO Polarization down
Current density (nA/cm2)
100 v
BNBT
50 ITO
–50
–100
–150
–200
–100 –75 –50 –25 0 25 50 75 100
Voltage (V)
Figure 3.18 Current density–voltage (J–V) curve of BNBT:Nb ceramics under dark and UV
illumination of 365 nm. Inset is the configuration for photovoltaic measurement. Source: Chen
et al. (2014). Reprinted with permission of Elsevier.
Figure 3.19 is the diagram proposed by Huang to illustrate the working princi-
ple of the photovoltaic effect compared with a conventional p–n junction solar
cell (Huang 2010). We can see that the band bending at the domain wall is favor-
able for charge traps and therefore enhance the charge separation. However, we
have to face the truth: a good ferroelectric requires very good insulating property,
3.6 Electrocaloric Effect 67
Domain DW Domain DW Domain

CB
CB
eVoc Eg Recombination
hv VB hv
VB
P P P
n-Type Depletion layer p-Type
(a) (b)
Figure 3.19 Principles of operation for a conventional solar cell and the ferroelectric
alternative. (a) Band diagram of a conventional semiconductor p–n junction, and (b) in the
ferroelectric domain, band bending occurs across the domain walls (DWs). The directions of
electric polarization (P) are shown by arrows. Source: Adapted from Huang (2010).
while the photovoltaic needs a semiconducting property by presenting enough

charge carries. These conflict requirements make a ferroelectric-based solar cell
very hard to reach acceptable energy conversion efficiency for real application.
3.6 Electrocaloric Effect

Ferroelectric materials have very broad applications in energy harvesting based
on the charge generation from piezoelectric effect and solar energy conversion
based on photovoltaic effect where the ferroelectricity plays a very important
role in charge separation. Another important energy conversion for a ferroelec-
tric material is to utilize its entropy changes from poled state to unpoled state to
absorb or release heat.
Electrocaloric (EC) effect utilizing the well-known ferroelectric material
P(VDF-TrFE) copolymer has been reported, and temperature change of 12 ∘ C
has been realized by the EC effect (Neese et al. 2008). Chip level electric cooling
device based on the EC effect has been demonstrated by Zhang’s group (Gu et al.
2014).
Figure 3.20a illustrates the basic mechanism of EC effect based on the entropy
change induced by polarization state change; while Figure 3.20b is a multilayered
structure to increase the heat pump efficiency. This EC effect can be used for
cooling without involving any mechanical movement and cooling gas, making it
very different from the way of current technology for refrigerators.
Reversible ΔS and ΔT for a polymer film with area A as the electric field
changes from E1 to E2 can be found from the pyroelectric coefficient as a
function of the electric field:
E2 ( )
𝜕D
ΔS = − dE
∫E1 𝜕T E
E2 ( )
1 T 𝜕D
ΔT = − dE
𝜌 ∫E1 C 𝜕T E
Electrodes Ferroelectric material
E=0
Heat release Heat absorb
Heat irradiation
– – – – – –
E>0
Figure 3.20 (a) An illustration of the basic principle of the electrocaloric effect in a polar
material and (b) a multilayered electrocaloric device where the electrocaloric films are
ferroelectric. Source: Adapted from Epstein and Malloy (2010).
where 𝜌 is the density

( and
) C is( the)specific heat. These equations are based on the
𝜕D 𝜕S
Maxwell equation 𝜕T = 𝜕E , which establishes the relationship between
E T
the change of polarization and the change of entropy under changing electric
field. By sweeping the electric field from E1 to E2 , heat can be pumped out from
the device and refrigeration can be realized. The recent development in this field
is to find new materials and new device design with higher efficiency toward com-
mercialization.
References
Al-Shareef, H.N., Bellur, K.R., Auciello, O., and Kingon, A.I. (1994). Fatigue and
retention of Pb(Zr0.53 Ti0.47 )O3 thin film capacitors with Pt and RuO2 electrodes.
Integrated Ferroelectrics 5 (3): 185–196. Available at: https://doi.org/10.1080/
10584589408017011.
Arimoto, Y. and Ishiwara, H. (2004). Current status of ferroelectric random-access
memory. MRS Bulletin 29 (11): 823–828.
Bao, P., Jackson, T.J., Wang, X., and Lancaster, M.J. (2008). Barium strontium
titanate thin film varactors for room-temperature microwave device applications.
Journal of Physics D: Applied Physics 41 (063001): 21.
Buser, R.G., Tompsett, M.F., Kruse, P.W. et al. (1997). Ch.4 Hybrid
Phyroelectric-Ferroelectric Bolometer Arrays. In: Uncooled Infrared Imaging
Arrays and Systems, 1e (ed. C.M. Hanson), 154–168. Academic Press.
Cao, D., Zheng, F., Shen, M. et al. (2012). High-efficiency ferroelectric-film solar
cells with an n-type Cu2 O cathode buffer layer. Nano Letters 12 (6): 2803–2809.
Chanthbouala, A., Garcia, V., Barthélémy, A. et al. (2012a). A ferroelectric
memristor. Nature Materials 11: 860–864. Available at: http://www.ncbi.nlm.nih
.gov/pubmed/22983431.
Chanthbouala, A., Crassous, A., Garcia, V. et al. (2012b). Solid-state memories
based on ferroelectric tunnel junctions. Nature Nanotechnology 7 (2): 101–104.
Available at: http://www.ncbi.nlm.nih.gov/pubmed/22138863.
References 69
Chen, F.S. (1969). Optically induced change of refractive indices in LiNbO3 and
LiTaO3 . Journal of Applied Physics 40 (8): 3389–3396. Available at: https://doi
.org/10.1063/1.1658195.
Chen, L., Luo, B.C., Wang, D.Y. et al. (2014). Enhancement of photovoltaic
properties with Nb modified (Bi,Na)TiO3 –BaTiO3 ferroelectric ceramics. Journal
of Alloys and Compounds 587: 339–343. Available at: http://www.sciencedirect
.com/science/article/pii/S0925838813026339.
Choi, T., Lee, S., Cheong, S.W. et al. (2009). Switchable ferroelectric diode and
photovoltaic effect in BiFeO3 . Science 324 (5923): 63–66. Available at: http://
science.sciencemag.org/content/324/5923/63.
Epstein, R.I. and Malloy, K.J. (2010). Electrocaloric refrigerator and multilayer
pyroelectric energy generator, Pub. No.: US 2010/0037624 A1. pp. 1–10. Available
at: https://patents.google.com/patent/US20100037624.
Fang, H., Yan, Q., Dai, J.Y. et al. (2015). Infrared light gated MoS2 field effect
transistor. Optics Express 23 (25): 31908–31914. Available at: http://www
.opticsexpress.org/abstract.cfm?URI=oe-23-25-31908.
Ferroelectric Memory Company (2018). One-transistor FeFET memory. Available
at: https://ferroelectric-memory.com/technology/one-transistor-fefet-memory/
(accessed 19 August 2019).
Fujitsu (n.d.). FRAM structure. Available at: http://www.fujitsu.com/global/
products/devices/semiconductor/memory/fram/overview/structure/ (Accessed
14 March 2018).
Garcia, V. and Bibes, M. (2014). Ferroelectric tunnel junctions for information
storage and processing. Nature Communications 5: 4289. Available at: http://
www.ncbi.nlm.nih.gov/pubmed/25056141.
Garcia, V., Fusil, S., Bibes, M. et al. (2009). Giant tunnel electroresistance for
non-destructive readout of ferroelectric states. Nature 460 (7251): 81–84.
Gruverman, A., Wu, D., Lu, H. et al. (2009). Tunneling electroresistance effect in
ferroelectric tunnel junctions at the nanoscale. Nano Letters 9 (10): 3539–3543.
Gu, H., Qian, X.S., Ye, H.J., and Zhang, Q.M. (2014). An electrocaloric refrigerator
without external regenerator. Applied Physics Letters 105 (16): 162905. Available
at: https://doi.org/10.1063/1.4898812.
Huang, H. (2010). Solar energy: ferroelectric photovoltaics. Nature Photonics 4:
134–135.
Ielmini, D. and Waser, R. (eds.) (2016). Resistive Switching: From Fundamentals of
Nanoionic Redox Processes to Memristive Device Applications. Weinheim,
Germany: Wiley-VCH Verlag GmbH & Co.
Karnati, K.K., Shen, Y., Trampler, M.E. et al. (2015). A BST-integrated capacitively
loaded patch for Ka- and X-band beamsteerable reflectarray antennas in satellite
communications. IEEE Transactions on Antennas and Propagation 63 (4):
1324–1333.
Kulwicki, B.M., Amin, A., Beratan, H.R., and Hanson, C.M. (1992). Pyroelectric
imaging. In: ISAF’92: Proceedings of the Eighth IEEE International Symposium on
Applications of Ferroelectrics, 1. IEEE https://doi.org/10.1109/ISAF.1992.300607.
Li, L., Zhao, X., Luo, H. et al. (2014). Scale effects of low-dimensional relaxor
ferroelectric single crystals and their application in novel pyroelectric infrared
detectors. Advanced Materials 26 (16): 2580–2585. Available at: https://doi.org/

10.1002/adma.201304546.
Liu, L., Li, X., Wu, X. et al. (2009). Dielectric, ferroelectric, and pyroelectric
characterization of Mn-doped 0.74Pb(Mg1/3 Nb2/3 )O3 –0.26PbTiO3 crystals for
infrared detection applications. Applied Physics Letters 95 (19): 192903. Available
at: https://doi.org/10.1063/1.3263139.
Neese, B., Chu, B., Zhang, Q.M. et al. (2008). Large electrocaloric effect in
ferroelectric polymers near room temperature. Science 321 (5890): 821–823.
Available at: http://www.sciencemag.org/cgi/doi/10.1126/science.1159655.
Pantel, D. and Alexe, M. (2010). Electroresistance effects in ferroelectric tunnel
barriers. Physical Review B 82 (13): 134105.
Qin, M., Yao, K., and Liang, Y.C. (2008). High efficient photovoltaics in nanoscaled
ferroelectric thin films. Applied Physics Letters 93 (12): 122904. Available at:
https://doi.org/10.1063/1.2990754.
Rogalski, A. (2003). Infrared detectors: status and trends. Progress in Quantum
Electronics 27 (2): 59–210. Available at: http://www.sciencedirect.com/science/
article/pii/S0079672702000241.
Tang, Y. and Luo, H. (2009). Investigation of the electrical properties of
(1 − x)Pb(Mg1/3 Nb2/3 )O3 –xPbTiO3 single crystals with special reference to
pyroelectric detection. Journal of Physics D: Applied Physics 42 (7): 75406.
Available at: http://stacks.iop.org/0022-3727/42/i=7/a=075406.
Tang, Y., Zhao, X., Feng, X. et al. (2005). Pyroelectric properties of [111]-oriented
Pb(Mg1/3 Nb2/3 )O3 –PbTiO3 crystals. Applied Physics Letters 86 (8): 82901.
Tang, Y., Luo, L., Jia, Y. et al. (2006). Mn-doped 0.71Pb(Mg1/3 Nb2/3 )O3 –0.29PbTiO3
pyroelectric crystals for uncooled infrared focal plane arrays applications. Applied
Physics Letters 89 (16): 162906. Available at: https://doi.org/10.1063/1.2363149.
Terrence, T. (2013). Paratek antenna technology will make an appearance in
BlackBerry Z30! BerryReview. Available at: http://www.berryreview.com/2013/
09/18/paratek-antenna-technology-will-make-appearance-blackberry-z30/
(accessed 12 March 2018).
Tien, N.T., Seol, Y.G., Lee, N. et al. (2009). Utilizing highly crystalline pyroelectric
material as functional gate dielectric in organic thin-film transistors. Advanced
Materials 21 (8): 910–915. Available at: https://doi.org/10.1002/adma.200801831.
Watton, R. (2011). Ferroelectric IR bolometers – from ceramic hybrid arrays to
direct thin film integration. Ferroelectrics 184 (1): 141–150.
Wen, Z., Li, C., Wu, D. et al. (2013). Ferroelectric-field-effect-enhanced
electroresistance in metal/ferroelectric/semiconductor tunnel junctions. Nature
Materials 12 (7): 617–621. Available at: http://www.ncbi.nlm.nih.gov/pubmed/
23685861.
Whatmore, R.W. (1986). Pyroelectric devices and materials. Reports on Progress in
Physics 49 (12): 1335–1386.
Yang, S.Y., Seidel, J., Byrnes, S.J. et al. (2010). Above-bandgap voltages from
ferroelectric photovoltaic devices. Nature Nanotechnology 5 (2): 143–147.
Yang, Y. et al. (2012). Flexible pyroelectric nanogenerators using a composite
structure of lead-free KNbO3 nanowires. Advanced Materials 24 (39):
5357–5362. Available at: https://doi.org/10.1002/adma.201201414.
References 71
Yau, H.M., Yan, Z.B., Dai, J.Y. et al. (2015). Low-field switching four-state nonvolatile
memory based on multiferroic tunnel junctions. Scientific Reports 5: 12826.
Available at: http://www.nature.com/doifinder/10.1038/srep12826.
Yau, H.M., Xi, Z., Dai, J.Y. et al. (2017). Dynamic strain-induced giant
electroresistance and erasing effect in ultrathin ferroelectric tunnel-junction
memory. Physical Review B 95: 214304.
Yuan, Y., Reece, T.J., Huang, J. et al. (2011). Efficiency enhancement in organic solar
cells with ferroelectric polymers. Nature Materials 10: 296–302.
Zhuravlev, M., Sabirianov, R., Jaswal, S., and Tsymbal, E. (2005). Giant
electroresistance in ferroelectric tunnel junctions. Physical Review Letters 94 (24):
246802. Available at: http://link.aps.org/doi/10.1103/PhysRevLett.94.246802.
73
Ferroelectric Characterizations
In recent years, many new emerging materials are being claimed to be ferroelec-
tric, such as two-dimensional (2D) materials MoS2 , SnSe, and topological insula-
tor material CsPbI3 (Liu et al. 2016). Even the solar cell material hybrid perovskite
CH3 NH3 PbO3 has been claimed to be ferroelectric, which is also attributed to be
one of the main reasons to facilitate the photovoltaic efficiency, but the result is
still under debate. To clarify that a material is ferroelectric needs critical exper-
imental evidence such as P–E loop and switchable electric polarization. In this
chapter, basic techniques and recent advances of ferroelectric characterization
methods are introduced.
4.1 P–E Loop Measurement

P–E hysteresis loop is the most direct evidence to prove a material to be ferroelec-
tric, from which the saturation polarization, remnant polarization, and coercive
field can be determined. However, the P–E loop cannot be measured directly; it
can only be obtained by measuring electric displacement D forming D–E hys-
teresis loop, providing that 𝜖 0 E is relatively much smaller compared with the
polarization P. In fact, to a ferroelectric material, this is true, i.e.
D = 𝜖0 E + P ≈ P (4.1)
The measurement of P–E hysteresis loop of a ferroelectric sample is usually
by means of modified Sawyer–Tower circuit that contains the ferroelectric
sample as a capacitor (C x ) in series with a linear reference capacitor (C ref )
(see Figure 4.1). There are two prerequisites for this measurement method: (i)
assuming the charge density (electric displacement D) of the ferroelectric sample
is the same as the amount of charges per unit area on the reference capacitor, and
(ii) the capacitance of the reference capacitor is constant and independent of the
applied voltage and it should be much larger than the capacitance of the sample
(around 1000 times so that the voltage drop across the reference capacitor can
be neglected).
From the Sawyer–Tower measurement,
Vref Cx
= (4.2)
Vx Cref

74 4 Ferroelectric Characterizations
Sample R
Cx Rx
Cref
Figure 4.1 A modified Sawyer–Tower circuit for ferroelectric P–E loop measurement.
where V ref is the voltage for the reference capacitor and V x is the voltage for the
sample. Given the dimension of the sample area S, thickness d, and dielectric
constant 𝜖 r , under electric field E, the sample’s ferroelectric polarization can be
found by
Cx 𝜖𝜖 SV 𝜖𝜖S S
Vref = V = r 0 x = r 0 E= P (4.3)
Cref x Cref d Cref Cref
Then, we have
Vref Cref
P= (4.4)
S
During the hysteresis loop measurement, an ac signal generated by a function
generator and amplified by a voltage amplifier is applied to the sample placed
in a silicone oil bath. The input and output signals of Sawyer–Tower circuit are
observed by a digital oscilloscope, and the data of hysteresis loop at different tem-
peratures are recorded in a computer.
As an example, Figure 4.2 shows our experimental result of P–E loop measure-
ment, from which we can determine that the remnant polarization Pr = 0.25 C/m2
or 25 μC/cm2 and saturation polarization Ps = 0.29 C/m2 or 29 μC/cm2 . The coer-
cive field Ec = 1.80 × 105 V/m.
To obtain a P–E loop of a typical ferroelectric material is not difficult; however,
for many ferroelectric materials with large leakage current (especially thin films),
it is not an easy task. Figure 4.3a sketch an ideal linear response of a capacitor
in the P–E loop measurement, and Figure 4.3b shows an ideal resistor response.
For a leaky ferroelectric thin film, the measured hysteresis loops sometimes only
show ellipsoid feature without saturation as illustrated in Figure 4.3c. This kind
of loop cannot be called P–E loop since it only reflects a leaking dielectric char-
acteristic, and therefore, this kind of result cannot support the conclusion of
ferroelectric claim of the corresponding materials (Scott 2008). A successful P–E
loop should present at least saturation in the loop as shown in Figure 4.3d.
4.1 P–E Loop Measurement 75
0.3 Ps = 0.29 C/m2
0.2
PR = 0.25 C/m2
Polarization (C/m2)
0.1
0.0
Ec = 1.80 × 105 V/m
–0.1
–0.2
–0.3
–1.5 × 106 –1.0 × 106 –5.0 × 105 0.0 5.0 × 105 1.0 × 106 1.5 × 106
Electric field (V/m)
Figure 4.2 Typical P–E loop from a PMN-PT relaxor ferroelectric single crystal.
P P
E E
(a) (b)
P P
E E
(c) (d)
Figure 4.3 (a) An ideal linear response of capacitor, (b) an ideal resistor response, (c) an
unsuccessful “P–E loop” without a saturation, which is a typical dielectric loss induced loop,
and (d) a typical P–E loop showing ferroelectricity.
4.2 Temperature-Dependent Dielectric Permittivity

Measurement
A very important characteristic for ferroelectrics is the presence of a peak in
dielectric permittivity–temperature curve as introduced in the Curie–Weiss law.
Therefore, another practical way to illustrate ferroelectricity of a material is to
measure its temperature-dependent dielectric permittivity 𝜖 r –T, where the peak
indicates ferroelectric to paraelectric phase transition and the corresponding
temperature is the Curie temperature. As shown in Figure 4.4a, the 𝜖 r –T
ϵr (T)
25 000
x=0
x = 0.34
20 000 x = 0.27
x = 0.47 x = 0.65
x = 0.87
15 000
10 000 x = 1.0
5000
0
0 100 200 300 400
(a) Temperature (K)
PMN-PT
2
104ϵ
Frequency
increases
0
–100 –50 0 50 100
(b) Temperature (°C)
Figure 4.4 Dielectric–temperature curve of (a) (Bax Sr1−x )TiO3 where x varies from 0 to 1, i.e.
from SrTiO3 to BaTiO3 and (b) PMN-xPT. Source: Adapted from Vendik et al. (1999) and Cheng
et al. (1998).
curves of (Bax Sr1−x )TiO3 (BST) show different Curie temperatures for different
components of x. The 𝜖 r –T curves at different frequencies also present the
frequency dispersion if a ferroelectric is relaxor type such as PMN-PT as shown
in Figure 4.4b.
4.3 Piezoresponse Force Microscopy (PFM)

Piezoresponse force microscopy (PFM) is a special function in an atomic force
microscope (AFM) that is designed to detect piezoelectric response of a sample
to an applied electric field. This technique has been widely used in ferroelec-
tric domain observation and was systematically introduced by Gruverman in his
book (Gruverman and Kalinin 2007). In this session, details of this technique and
results from our group of research are introduced.
4.3.1 Imaging Mechanism of PFM

In PFM imaging, when an ac electric field is applied between a conductive probe
and bottom electrode of a sample, local oscillatory deformation with the same
frequency as the ac field will be induced on the sample surface due to the converse
piezoelectric effect. Depending on the direction of ferroelectric polarization, this
oscillatory deformation response has 0∘ or 180∘ phase relationship with the added
ac electric field, and this phase angle together with amplitude can be measured by
lock-in technique and imaged to reveal the ferroelectric polarization and domain
structure.
The conductive AFM probe detects the surface oscillation and sends the signal
through the reflected laser beam to photodiode detector and be processed by a
lock-in amplifier. The amplitude measured by the lock-in amplifier represents the
scale of piezoresponse that is proportional to piezoelectric coefficient; and the
phase angle represents the polarization direction, where 0∘ corresponds parallel
and 180∘ the antiparallel alignment of polarization to the electric field added on
the AFM tip. By forming an image with phase angle derived from the lock-in
amplifier, the PFM technique can image ferroelectric domains by presenting the
bright–dark contrast, and the amplitude tells how strong the piezoelectricity is.
In the PFM imaging setup as shown in Figure 4.5, an ac voltage V cos 𝜔t is
applied to the bottom electrode to provide an electric field on the sample while
the conductive probe is electrically grounded, or vice versa. The signal detected
by photodiode from the local polarization, A cos(𝜔t + 𝜑), with the reference
signal V cos 𝜔t will be fed into the lock-in amplifier. By passing through a
low-pass filter, the lock-in amplifier will output a signal with dc component Ã,
which is the time-averaged value of the piezoresponse amplitude and the phase
angle 𝜑 of electromechanical response of the sample to the ac electric field. The
domain images can be obtained by imaging Ã and 𝜑 over an observation area.
Figure 4.6 is a PFM image of a PZT grain in a thin film structure, where one can
see that within one PZT grain there are multiple ferroelectric domains forming
90∘ and 180∘ domain walls. These domain polarizations can be switched by
adding a dc field on the tip.
Output electrical signal
Photodiode detector
Laser source
Low-pass filter
Feedback circuit
Piezo-material
Scanner
Lock-in amplifier
Reference signal
Amplitude, A Phase, φ
Topographic ac modulate signal Domain image
Figure 4.5 Schematic diagram of piezoresponse force microscopy imaging mechanism.
Figure 4.6 (a) Our PFM phase image

120° of a PZT grain in a PZT film showing
(101)
ferroelectric domain structure as
(110) illustrated in (b). Source: Adapted
(011) from Zhao (2005).
[111]
(a) (b)
Not only the domain pattern, when an area is scanned with a conductive APM
tip being added with a dc electric field, but also the polarization of the scanned
area can be switched. Therefore, the PFM technique can also demonstrate polar-
ization switching, which is the key criterion for a ferroelectric. Furthermore, a
so-called butterfly loop (see Figure 4.7) can be achieved by PFM technique when
the piezoresponse amplitude is measured as a function of dc electric field added
on the sample. The butterfly loop illustrates that the increased dc field results
increased polarization leading to larger piezoresponse. It is worthy to note that
the nonzero piezoresponse at zero field is due to the remnant polarization, and
the corresponding phase angle change from 0∘ to 180∘ is due to the polariza-
tion switching. Therefore, this butterfly loop and the corresponding 180∘ phase
change can be used as criteria for clarification of ferroelectricity.
To demonstrate ferroelectric switching is essential to claim ferroelectricity. In
PFM imaging, when a dc voltage is applied on the conductive AFM tip while scan-
ning an area, the subsequent PFM image can reveal the switched polarization as
illustrated in Figure 4.8, where (a) is the amplitude image and (b) the phase image.
Figure 4.7 Our PFM out-of-plane 250

phase hysteresis loop and amplitude
butterfly loop as a function of
applied dc voltage for an epitaxial 200
BaTiO3 thin films.
150
Phase (°)
100
50
0
Amplitude (a.u.)
–6 –4 –2 0 2 4 6
Voltage (V)
a.u. °
1 180
1 μm 1 μm
0 0
(a) (b)
Figure 4.8 Our PFM results from epitaxial BaTiO3 thin films. (a) Amplitude and (b) phase
images recorded after writing an area of 3.0 × 3.0 μm2 with −4.5 V and then the central square
with +4.5 V using a biased conductive tip. Source: Yau et al. (2017). Reproduced with
permission of American Physical Society.
For the switched areas with polarizations facing up and down, the brightness of
amplitude image represents the strength of piezoresponse, while the phase image
represents the polarization direction.
4.3.2 Out-of-plane Polarization (OPP) and In-plane Polarization (IPP)

PFM
The real situation is that in a ferroelectric crystal or film, the polarizations of
domains are not only facing up and down but also in-plane or inclined with
an angle to the surface. Therefore, a PFM should also be able to map domains
with different polarization directions. Fortunately, for the PFM technique,
out-of-plane polarization (OPP) and in-plane polarization (IPP) PFM can be
done simultaneously. In the OPP-PFM, the induced local vertical deformation
is measured by the up-and-down deflection of the cantilever; while the induced
local shear deformation is measured by the torsional twisting of the cantilever
in the IPP-PFM as shown in Figure 4.9. The photo-detector in the AFM can
distinguish the vertical up-and-down movement and torsion (twist) movement
of the tip by sensing the reflected laser beam. There are usually four sectors in
the photo-detector, where (A + B) − (C + D) signal measures the tip’s vertical
movement and (A + C) − (B + D) signal measures the tip’s torsional movement.
The OPP- and IPP-PFM images can be obtained by collecting the piezoelectric
response signals demodulated by two lock-in amplifiers in PFM.
Here we demonstrate the OPP- and IPP-PFM techniques with our results
in PMN-PT single crystal domain imaging. Figure 4.10 shows the OPP- and
IPP-PFM images of the (001)-cut PMN-0.38PT crystal at room temperature,
where stripe-like domains with width of around 700 nm can be observed. In
the PFM images, the dark contrast in the OPP and IPP amplitude images
corresponds to the regions without piezoresponse signal, and the bright regions
correspond to large piezoresponse. While in the OPP phase image, the upward
Torsion Deflection
Out-of-plane
E P E P P
A B
In-plane
a-domain c-domain
C D
X3 X1
PP X2
E E P X2
X1 X3
(a) (b) ε5 = d51E1 ε3 = d33E3

ε1 = d13E3
Figure 4.9 (a) Illustration of four sectors of a photo-detector in AFM. (b) Quantitative
explanation of the contrast formation in IPP- and OPP-PFM. The electric field induced from a
conductive tip causes surface displacement due to the converse piezoelectric effect. Source:
Adapted from Felten et al. (2004).
(a) (b) (c)

OPP
IPP
IPP IPP
IPP
OPP
(d) (e)
Figure 4.10 Piezoresponse (a) OPP phase, (b) OPP amplitude, (c) IPP amplitude images of the
as-grown (001)-cut PMN-0.38PT single crystal with scan size of 5 μm, (d) the schematic
diagram of tetragonal unit cell with six polar directions, and (e) illustration of proposed
domain boundary configuration revealed from the OPP- and IPP-PFM images. Source: Adapted
from Wong et al. (2006).
polarization (pointing out of the paper) appears dark in contrast, and regions
with downward polarization appear bright (please notice that the orientation
depends on the direction of electric field).
One can see that both upward and downward domains appear dark and bright
alternatively in the OPP phase image (Figure 4.10a); while they appear bright in
OPP amplitude image (Figure 4.10b). It is interesting to notice that in the OPP
amplitude image, the dark contrast regions with the width of around 40–90 nm
at the domain boundaries, however, appear bright in the IPP amplitude image
shown in Figure 4.10c. It indicates that the domain boundaries contain large
IPP without OPP. If an IPP phase image is formed, the IPP direction can also
be mapped as bright and dark contrasts. According to a schematic diagram of a
tetragonal unit cell with six polar directions (Figure 4.10d), it can be seen that the
dark and bright stripe-like domains in the OPP phase image represent two oppo-
site OPPs, and four IPPs exist in the domain boundaries. Figure 4.10e shows the
proposed domain boundary configuration revealed from the OPP- and IPP-PFM
images.
In order to understand the nature of PFM imaging, finite element modeling
(FEM) has been carried out to illustrate the surface deformation under a finite
point electrode induced electric field. In the FEM simulation with commercial
software ANSYS, the size of the sample is set as 2 mm × 3 mm × 50 μm, which
is same as the real sample in PFM imaging. A metal layer under the bottom of
sample as bottom electrode is fixed. Because the size of the conductive tip is very
small, the probing electric field is strongly inhomogeneous and decays rapidly
from the tip into the crystal. Therefore, we can substitute the electric field with
the same magnitude as being used in the PFM experiment into the simulation.
The displacements of two samples possessing OPP and IPPs are computed in the
ANSYS simulation.
In the simulation results (Wong 2007), the light blue and purple curves in
Figures 4.11 and 4.12 represent the applied ac electric field across the sample and
the calculated displacement in the x, y, and z directions, respectively. From the
OPP simulation result, it can be seen that along the direction of OPP, denoted
as the z-direction, there is a large displacement as shown in Figure 4.11d, while
there is no obvious displacement along the x and y directions in Figure 4.11b,c.
By contrast, in the IPP simulation result, large displacement along the x direction
can be observed as shown in Figure 4.12b; but there is no obvious displacement
along the y and z directions. Therefore, according to the movement of sample
surface, the IPP and OPP can be detected by PFM tip.
6.25
5
3.75
Out-of-plane ac 2.5
polarization field z 1.25
Value
0
–1.25
x –2.5
–3.75
y
–5
–6.25 (×10−6)
0 –2 –4 –6 –8 1
–1 –3 –5 –7 –9
(a) (b) Time
6.25 6.25
5 5
3.75 3.75
2.5 2.5
1.25 1.25
Value
Value
0 0
–1.25 –1.25
–2.5 –2.5
–3.75 –3.75
–5 –5
–6.25 (×10−6) –6.25 (×10−6)
0 –2 –4 –6 –8 1 0 –2 –4 –6 –8 1
–1 –3 –5 –7 –9 –1 –3 –5 –7 –9
(c) Time (d) Time
Figure 4.11 (a) The sample surface movement of out-of-plane polarization ANSYS simulation.
(b–d) The graph of x, y, and z displacements, respectively. The applied ac electric field is along
the out-of-plane direction. Source: Adapted from Wong (2007).
6.25
ac field 5
3.75
2.5
z 1.25
Value
0
–1.25
In-plane x –2.5
polarization
–3.75
y
–5
–6.25 (×10−6)
0 –2 –4 –6 –8 1
–1 –3 –5 –7 –9
(a) (b) Time
6.25 6.25
5 5
3.75 3.75
2.5 2.5
1.25 1.25
Value
Value
0 0
–1.25 –1.25
–2.5 –2.5
–3.75 –3.75
–5 –5
–6.25 (×10−6) –6.25 (×10−6)
0 –2 –4 –6 –8 1 0 –2 –4 –6 –8 1
–1 –3 –5 –7 –9 –1 –3 –5 –7 –9
(c) Time (d) Time
Figure 4.12 (a) The sample surface movement of in-plane polarization ANSYS simulation.
(b–d) The graph of x, y, and z displacements, respectively. The applied ac electric field is along
the out-of-plane direction. Source: Adapted from Wong (2007).
4.3.2.1 Electrostatic Force in PFM

When an ac electric field is applied to a ferroelectric surface through a conductive
AFM tip, the displacement D of the AFM tip includes three parts as illustrated by
the equation
D = ad33 E + b𝛼E2 + cE2 (4.5)
where the first term is from the piezoresponse of the sample surface with piezo-
electric constant d33 and the second term is contributed by the strain generated
from electrostrictive effect with coefficient 𝛼, which exists in all dielectrics and is
proportional to E2 . The last term is the contribution from the interaction between
the tip cantilever and sample surface due to the electrostatic force (also called
capacitance force since the cantilever and sample surface form a parallel plate
capacitor) that can be repulsive or attractive depending on the positive or nega-
tive charges of the sample surface and the instant voltage of the cantilever. The
constants a, b, and c represent the relative contributions to the tip displacement
due to the imaging condition, such as the cantilever’s force constant, and the
tip/sample surface condition (for example, when the cantilever is on metal elec-
trode or near the edge of the sample, the electrostatic attraction force is much
lower).
PFM amplitude image PFM amplitude image Figure 4.13 Expected

without electrostatic force with electrostatic force piezoresponse amplitude
signals, and accompanying
domain images for (a)
without and (b) with
Amplitude Amplitude electrostatic force.
Ferroelectric polarization Electrostatic force

(a) (b)
During PFM imaging, in order to produce an ac electric field across the sample,
an ac voltage is applied to the bottom electrode; while the conductive tip is
electrically grounded, or vice versa. Since there is an air gap between cantilever
and sample surface, capacitive force between the cantilever and sample surface
needs to be considered in PFM imaging. The capacitive force is an attractive
force between the cantilever and sample surface that alters the PFM image and
shadows out the information of domain boundary structure. Hong et al. (2002)
have investigated the capacitive force effect during PFM imaging and discovered
that, at the center position, the attractive capacitive force leads to a vertical shift
of amplitude signal in the upward and downward domains. Figure 4.13 illustrates
the electrostatic capacitance force effect to the PFM amplitude image, where
panel (a) is what we suppose to see from a ferroelectric polarization induced
contrast and panel (b) is what we see when a strong capacitance force exists.
However, it can be reduced by placing the tip near the edge of sample surface
due to the small capacitive coupling.
In order to avoid the effect coming from attractive force and increase the
signal-to-noise ratio, a relatively large force constant (e.g. 40 N/m) tip should be
used in PFM imaging. With a “shorter” and “stronger” cantilever, the capacitive
force effect can be significantly reduced, and clear domain boundary feature can
be obtained in the PFM images.
4.3.2.2 Perspectives of PFM Technique

PFM is becoming one of the key technologies to investigate ferroelectricity and
domain structures of new emerging materials such as 2D, topological, and hybrid
perovskite materials. For example, the ferroelectricity in halide-perovskite solar
cell material has been considered to be important to its photovoltaic efficiency.
But the ferroelectric domain structure is still arguable. Switchable polarization
has not been demonstrated so far.
It has been said “you can almost obtain PFM image in any material.” Sometimes
this is correct for careless experiments, i.e. non-ferroelectric materials can pro-
duce PFM signals that are similar to those of standard ferroelectrics (Vasudevan
et al. 2017). If you look at Eq. (4.5), the contribution to the PFM image, a domain
contrast stimulated by an external electric field added on a conductive AMF tip
comes from at least three integration force where only the first term is real piezo-
electric response. Electrostrictive and electrostatic responses should also be con-
sidered, especially when the piezoresponse is very small in those non-typical
ferroelectric materials (such as ultrathin films, 2D materials, and other emerg-
ing new materials in which only monolayer or a few layers of some 2D mate-
rials present weak ferroelectricity). To deeply understand the mechanism and
improve the PFM technology (such as resonance tracing; Xie et al. 2012) are very
important in PFM study of very thin films and new materials’ ferroelectricity and
domain structure.
The piezoresponse and PFM sensitivity can usually be enhanced by driving
the ac modulation voltage near the resonance of the cantilever-specimen system
as shown in Figure 4.14(a), and thus magnify the piezoresponse amplitude
A ∼ d33 V 0 Q by orders of magnitude, where Q is the quality factor of the system
Δω
V0eiox
Amplitude
d33V0Q ΔA
d33V0
d33V0ei(ox + φ)
ω0 ω′0 Frequency
7.3 150
6 Fit 100
amplitude
phase
4.7 50
Amplitude (a.u.)
Phase (°)
3.4 0
2.1 –50
0.8 –100
ω1 ω2
–0.5 –150
255 280 305 330 355
Frequency (kHz)
Figure 4.14 Schematics of dual frequency resonance tracking (DFRT) technique: (a)
tip-nanofiber harmonic oscillation system, (b) resonance enhancement using single
frequency, and (c) schematics of DFRT using actual experimental data of piezoresponse versus
ac driving frequency. Source: Xie et al. (2012). Reproduced with permission of The Royal
Society of Chemistry.
typically in the range of 10–100 (Xie et al. 2012). For the single frequency
resonance enhancement technique, the frequency is locked at the resonance in
a particular location. As the resonance shifts during scanning due to the change
of contact stiffness, the lock-in will actually drive the oscillator away from the
resonance at the current location, resulting in loss of enhanced sensitivity,
as schematically shown in Figure 4.14b. The dual frequencies adopted, on
the other hand, allow us to trace resonance when it varies over the sample
surface and minimize the cross-talk with topography, as schematically shown
in Figure 4.14c.
4.4 Structural Characterization

Structural distortion from a centrosymmetric structure is essential for ferro-
electric phase transition, such as from cubic to tetragonal (T), rhombohedral
(R), orthorhombic (O), or monoclinic (M) phases. Using XRD characterization
to determine the structure distortion is a very important step and is also a
convenient method to identify the ferroelectric phase. Figure 4.15a shows that
tetragonal phase results in split of (002) peak where the two equal intensity
peaks are (002) and (020) since c > a for T structure. While for orthorhombic
phase with ⟨110⟩ polarization direction, the (002) peak shows a shoulder peak.
As shown in Figure 4.15b from PMN-0.28PT rhombohedral phase, the (111)
peak splits into two peaks due to the elongation along one particular (111)
direction.
The map of symmetric Bragg peak can also be used to separate tilt (off-set) or
strain in the structure of a material. Figure 4.16a is a reciprocal space map for
pseudocubic {103}c peaks at a representative lower temperature (30 ∘ C), panel
(b) is a reciprocal space map for {113}c peaks at 30 ∘ C, and panel (c) illustrates
the pseudocubic unit cell of the lower temperature phase with a monoclinic dis-
tortion where the c-axis direction tilts up toward [100]c . Figure 4.16(d) is the
reciprocal space mapping for pseudocubic {103}c peaks and (e) {113}c peaks per-
formed at 200 ∘ C, respectively, and (f ) presents the pseudocubic unit cell at the
(002) (020) kα1 [111] poled

PMN–0.28PT
20% LBFO kα2
(002)
Intensity (a.u.)
Intensity (a.u.)
15% LBFO 50 K
(002)
SRO(002)
10% LBFO 150 K
(002)
BFO
300 K
40 42 44 46 48 50 37.5 38.0 38.5 39.0

(a) 2θ (°) (b) 2θ (°)
Figure 4.15 Typical X-ray 𝜃–2𝜃 scans of (a) BiFeO3 and La:BiFeO3 films with different La
concentrations on SrRuO3 /SrTiO3 /Si substrates, and (b) the (111) surface of [111]-poled
PMN-0.28PT single-domain crystal. Source: Adapted from Chu et al. (2008) and Li et al. (2016).
4.5 Domain Imaging and Polarization Mapping by Transmission Electron Microscopy 87
4.15 {103}c 4.20 {113}c ^

cc T = 30 °C (M )
C
(–103)c 4.15
4.10 (–1–13)c
4.10
(–113)c
q⊥ (Å–1)
(0–13)c
4.05 ΔL (013)c 4.05 ΔL
4.00 (1–13)c ^
(113)c bc
4.00 (103)c 3.95
3.90 βL
3.95 30 °C 30 °C ^
ac
1.60 1.65 1.70 1.75
20
25
30
35
40
45
50
2.
2.
2.
2.
2.
2.
2.
(a) (b) (c)
^
cm ^ T = 200 °C (MA)
4.15 {103}c 4.20 {113}c
cc
{113}m {203}m
4.15
4.10 (–103)c (–1–13)c
(0–13)c 4.10
q⊥ (Å–1)
4.05 (–113)c
4.05 2ΔH
(1–13)c
4.00
(103)c ^ ^
4.00 bc am
(013)c 3.95 (113)c
βH
3.90
3.95 200 °C 200 °C βH
^
ac
1.60 1.65 1.70 1.75
20
25
30
35
40
45
50
^
2.
2.
2.
2.
2.
2.
2.
–bm
qII (Å–1) qII (Å–1)
(d) (e) (f)
Figure 4.16 Reciprocal space mapping of the MA and MC phases in PMN-xPT crystal. Source:
Ko et al. (2011). Reprinted with permission of Springer Nature.
higher temperature with the direction of the c-axis tilt toward [110]c. The high
temperature structural phase is similar to the normal R-BiFeO3 film in terms of
the direction of the distortion; however, it is apparently different in terms of the
large c-axis lattice parameter.
4.5 Domain Imaging and Polarization Mapping

by Transmission Electron Microscopy
Transmission electron microscopy (TEM) is an important technology in material
structural characterization including ferroelectric materials. Beyond the general
techniques such as electron diffraction, high-resolution imaging, bright–dark
field imaging, and in situ observation of ferroelectric domain evolution under
electric field, direct mapping of polarization by Cs-corrected high-angle annular
dark-field (HAADF) imaging is emerging as a very powerful technique to directly
mapping ionic displacement and determine the electric polarization. Bright
and dark field images can present domain structures by different contrasts due
to their different scattering of incident electron beam. Figure 4.17(a) shows
existence of 180∘ and 90∘ domain structures in K0.15 Na0.85 NbO3 single crystal,
and the corresponding selected area electron diffraction pattern is shown in
Figure 4.17(b) (Chen et al. 2015).
0.2 μm
(a) (b)
Figure 4.17 (a) Low magnification TEM images of the KNN single crystal along ⟨100⟩ axes
showing typical laminate domain structures and (b) the corresponding selected area electron
diffraction pattern. Source: Adapted from Chen et al. (2015).
4.5.1 Selected Area Electron Diffraction (SAED)

Selected area electron diffraction (SAED), in general, is difficult to distinguish a
ferroelectric phase of either a tetragonal, rhombohedral, or orthorhombic from
a paraelectric cubic phase, since the resolution of electron diffraction pattern is
much lower than XRD. However, if the distortion is relatively large, SAED may
still show split spots.
In the study of antiferroelectrics, SAED can clearly show the structural infor-
mation by presenting superlattice structure in diffraction pattern. As shown in
Figure 4.18, a series of SAED patterns of different zone axes of Bi0.85 Nd0.15 FeO3
ceramics were taken from the regions where the 1/4(00l) superstructure appears.
The fundamental spots in Figure 4.20 come from the pseudocubic perovskite
(a) (b) (c)

– – –
120 004 044 004 204
004 040 200
[210] [100] [010]
(d) (e) – (f)

–
040 200 044 120
040
– –
124 208
–––
[001] [211] [401]
Figure 4.18 SAED patterns of different zone axes of Bi0.85 Nd0.15 FeO3 ceramics taken from the
of superstructure regions. The zone axes corresponding to (a–f ) are [210], [100], [010], [001],
[211] and [401], respectively. Source: Pu et al. (2015). Reprinted with permission of Elsevier.
structure. In comparison with the fundamental spots, the emergence of 1/4{hh0}

and 1/4(00l) diffraction spots indicates √that the√ superstructure resembles the
antiferroelectric Pbam (with unit cell 2a × 2 2a × 4a) structure of PbZrO3
but with quadrupled c-axis instead of the doubled c-axis. It is seen that all the
diffraction
√ spots √ can be well indexed using an orthorhombic unit cell with dimen-
sions 2a × 2 2a × 4a (Pu et al. 2015).
High-resolution TEM image can present clear crystal lattice image of per-
ovskite structure such as BaTiO3 ; however, the imaging mechanism of wave
interference and resolution limitation make it impossible to see the rela-
tive movement of ions from B site to A site. Thanks to the development of
Cs-corrected STEM technique, now people can use it to observe relative shift
of ions from B site to A site by forming real atomic position image. With this
progress, the polarization direction can be directly mapped out from the atomic
images, and therefore the domain structure can be identified. This is a technique
using focused electron beam in the size of ∼0.1 nm to scan sample surface
and using HAADF imaging mode to collect scattered electrons forming the
position image of A and B sites cations in the perovskite crystal structure, where
the heavier atoms appear brighter contrast. A recent research has reported
the observation of polarization vertex (circular polarization direction loop
compared with normally existed 90∘ and 180∘ domains) (Tang et al. 2015).
Wang reported the growth of BiFeO3 (BFO) films on the tetragonal SrRuO3
(SRO)-buffered STO (001) substrate. Using aberration-corrected HAADF
Z-contrast STEM imaging, they visualize, at the atomic scale, the out-of-plane
atomic displacement in the 2D BFO films as shown in Figure 4.19.
BFO is a well-known ionic-displacive ferroelectric material. Both the oxygen
octahedra and the positive (Bi) ions have displacements from the center of the
20 pm
BFO 1
2
SRO
(b)
O
STO DFe
[001]
Fe
c
[100] a Bi
(a) (c)
Figure 4.19 (a) Atomically resolved HAADF-STEM images of 2 unit-cell-thick BiFeO3 . (b) B-site
atomic displacement vector map from the area marked with a white dashed rectangle in (a),
where the yellow arrows superimposed on the atomic image represent the relative
displacement DFe of the Fe atoms as shown in (c). Source: Adapted from Wang et al. (2018).
Fe cation tetragonal cell that give rise to the spontaneous polarization. However,
due to difficulty in imaging the light element (O), the relative displacement
between the Fe cation and the mass center of an elongated rectangle formed by
its four Bi neighbors (denoted as DFe in Figure 4.19) can be used to determine
the polarizations of BFO unit cells.
Another important technique in TEM is the in situ observation with electric
field applied to the sample. Under electric field driven, domain evolution can be
dynamically observed (see Figure 4.20). No domain structure is observed without
electric field, but an applied positive electric field leads to the domain growth
Top electride (Pt/Au)

Pdown
Pup
(a) After contact (0 V) Nb-STO
(b) 3V
(c) 10 V
(d) –4 V
(e) –10 V
Figure 4.20 A series of TEM image of in situ observation of ferroelectric domain evolution
induced by external biases. Source: Han et al. (2014). Reprinted with permission of Springer
Nature.
toward the top electrode. When the negative field is applied, it is noted that the
domain configuration is reversible.
4.5.2 Convergent Beam Electron Diffraction (CBED) for Tetragonality

Measurement
Convergent beam electron diffraction (CBED) is a technique widely used for
determination of structure, symmetry details, and atom positions in a crystal as
small as a few nanometers in size in the field of TEM. Conventionally, until the
discovery of CBED, electron diffraction from a thin crystal in TEM was obtained
using a method called SAED using a parallel electron beam. By replacing the
parallel beam with a convergent beam, the diffraction spots are enlarged into
CBED discs. Compared with the smallest area of SAED (500 nm), the area for
diffraction in CBED is chosen by focusing the incident beam into a very fine spot
(a few nanometers) on the region of interest.
CBED has been widely used for nanostructure analysis. Detailed analyses of
CBED patterns could give good amount of additional and useful information
about a crystal. These include (i) three-dimensional information about the recip-
rocal lattice, (ii) point and space-group symmetry details, (iii) lattice parameters
from regions as fine as 1 nm, (iv) atom positions within a unit cell, and (v) defects
in crystals.
A ferroelectric tetragonal phase can be determined using the CBED method.
Figure 4.21a shows a CBED pattern of the tetragonal phase of PbTiO3 taken
at room temperature with the [100] incidence. The CBED pattern is seen to
have mirror symmetry parallel to the c-axis. Figure 4.21b schematically shows
the corresponding [100]-projected structure of the tetragonal phase, where the
direction of the ferroelectric polarization is indicated by arrow. Figure 4.21c
shows a CBED pattern simulated with the structure in Figure 4.21b using soft-
ware MBFIT for dynamical diffraction calculations. The features of the CBED
pattern of Figure 4.21a are qualitatively reproduced, indicating that the local
structure of the tetragonal phase of PbTiO3 has the symmetry of the tetragonal
structure.
Exp. Sim.
c* Ba c*
001
Ti Ps
000 010 O
c
m b m
(a) (b) (c)
Figure 4.21 (a) CBED pattern of the tetragonal phase of PbTiO3 taken at room temperature
with the [100] incidence, where the mirror expected from the tetragonal structure is indicated
by a line. (b) Corresponding crystal structure of the tetragonal phase with space group P4mm.
(c) CBED pattern simulated with the structure in (b). Source: Tsuda and Tanaka (2013).
Reprinted with permission of IOP.
References
Chen, Y., Wong, C.M., Dai, J.Y. et al. (2015). Ferroelectric domain structures in
⟨001⟩-oriented K0.15 Na0.85 NbO3 lead-free single crystal. AIP Advances 5 (3):
037117. Available at: https://aip.scitation.org/doi/pdf/10.1063/1.4914936?
class=pdf.
Cheng, Z.-Y., Katiyar, R.S., Yao, X., and Bhalla, A.S. (1998). Temperature dependence
of the dielectric constant of relaxor ferroelectrics. Physical Review B 57 (14):
8166–8177. Available at: https://link.aps.org/doi/10.1103/PhysRevB.57.8166.
Chu, Y.H., Zhan, Q., Yang, C.-H. et al. (2008). Low voltage performance of epitaxial
BiFeO3 films on Si substrates through lanthanum substitution. Applied Physics
Letters 92 (10): 102909. Available at: https://doi.org/10.1063/1.2897304.
Felten, F., Schneider, G.A., Saldaña, J.M., and Kalinin, S.V. (2004). Modeling and
measurement of surface displacements in BaTiO3 bulk material in piezoresponse
force microscopy. Journal of Applied Physics 96 (1): 563–568. Available at: https://
doi.org/10.1063/1.1758316.
Gruverman, A. and Kalinin, S.V. (2007). Piezoresponse force microscopy and recent
advances in nanoscale studies of ferroelectrics. In: Frontiers of Ferroelectricity
(eds. S.B. Lang and H.L.W. Chan), 107–116. Boston, MA: Springer US. Available
at: https://doi.org/10.1007/978-0-387-38039-1_10.
Han, M.-G., Marshall, M.S.J., and Zhu, Y. (2014). Interface-induced nonswitchable
domains in ferroelectric thin films. Nature Communications 5: 4693. Available at:
https://doi.org/10.1038/ncomms5693.
Hong, S., Shin, H., Woo, J., and No, K. (2002). Effect of cantilever-sample interaction
on piezoelectric force microscopy. Applied Physics Letters 80 (8): 1453–1455.
Ko, K.-T., Park, J.H., Yang, C.H. et al. (2011). Concurrent transition of ferroelectric
and magnetic ordering near room temperature. Nature Communications 2: 567.
Li, F., Zhang, S., Chen, L.Q. et al. (2016). The origin of ultrahigh piezoelectricity in
relaxor-ferroelectric solid solution crystals. Nature Communications 7: 13807.
Available at: http://dx.doi.org/10.1038/ncomms13807.
Liu, S., Kim, Y., Tan, L.Z., and Rappe, A.M. (2016). Strain-induced ferroelectric
topological insulator. Nano Letters 16 (3): 1663–1668. Available at: https://doi
.org/10.1021/acs.nanolett.5b04545.
Pu, S., Zheng, H., Zou, H. et al. (2015). The microstructure and ferroelectric
property of Nd-doped multiferroic ceramics Bi0.85 Nd0.15 FeO3 . Ceramics
International 41: 5498–5504.
Scott, J.F. (2008). Journal of Physics: Condensed Matter 20: 021001.
Tang, Y.L., Zhu, Y.L., Ma, X.L. et al. (2015). Observation of a periodic array of
flux-closure quadrants in strained ferroelectric PbTiO3 films. Science 348 (6234):
547–551. Available at: http://www.sciencemag.org/cgi/doi/10.1126/science
.1259869.
Tsuda, K. and Tanaka, M. (2013). Convergent-beam electron diffraction study of the
local structure of the tetragonal phase of PbTiO3 . Applied Physics Express 6 (10):
101501. Available at: http://stacks.iop.org/1882-0786/6/i=10/a=101501.
References 93
Vasudevan, R.K., Balke, N., Maksymovych, P. et al. (2017). Ferroelectric or

non-ferroelectric: why so many materials exhibit “ferroelectricity” on the
nanoscale. Applied Physics Reviews 4 (2): 021302.
Vendik, O.G., Hollmann, E.K., Kozyrev, A.B. et al. (1999). Ferroelectric tuning of
planar and bulk microwave devices. Journal of Superconductivity 12 (2): 325–338.
Available at: https://doi.org/10.1023/A:1007797131173.
Wang, H., Tian, H., Chen, J.S. et al. (2018). Direct observation of room-temperature
out-of-plane ferroelectricity and tunneling electroresistance at the
two-dimensional limit. Nature Communications 9 (1): 3319. Available at: https://
doi.org/10.1038/s41467-018-05662-y.
Wong, K., Tian, H., and Chen, J.S. (2007). Study of Domain Structure and Evolution
in PMN-30%PT Single Crystals by Means of Piezoresponse Force Microscopy.
Hong Kong Polytechnic University. Available at: http://theses.lib.polyu.edu.hk/
handle/200/812.
Wong, K.S., Dai, J.Y., Choy, C.L. et al. (2006). Study of domain boundary
polarization in (111)-cut [Pb(Mg1/3 Nb2/3 )O3 ]0.7 (PbTiO3 )0.3 single crystal by
piezoresponse force microscopy. Applied Physics Letters 89 (9): 092906.
Xie, S., Gannepalli, A., Li, J. et al. (2012). High resolution quantitative piezoresponse
force microscopy of BiFeO3 nanofibers with dramatically enhanced sensitivity.
Nanoscale 4 (2): 408–413. Available at: http://xlink.rsc.org/?DOI=C1NR11099C.
Yau, H.M., Xi, Z., Dai, J.Y. et al. (2017). Dynamic strain-induced giant
electroresistance and erasing effect in ultrathin ferroelectric tunnel-junction
memory. Physical Review B 95: 214304.
Zhao, X. (2005). Ferroelectric Domain Study by Piezoresponse Force Microscopy.
Hong Kong Polytechnic University.
95
Recent Advances in Ferroelectric Research
There are very broad applications for ferroelectric materials, especially thin
films, and the research interest in ferroelectrics is showing accelerated increase
in the twenty-first century. One of the main reasons is the successful demon-
stration of ferroelectric non-volatile memory in the field effect transistor (FET)
structure. Other new developments in ferroelectrics are stimulated by nanotech-
nology, where the ferroelectricity (FE) in nanometer scale and the corresponding
advanced characterization techniques for emerging new materials such as
two-dimensional (2D) and topological materials, have been extensively studied.
This chapter introduces the most recent progress in ferroelectric research
in terms of new materials, new phenomena, and advanced characterization
techniques and applications.
5.1 Size Limit of Ferroelectricity

Due to the nature of crystal symmetry breaking-induced ferroelectricity, it is
expected that when a single grain of ferroelectric crystal approaches the regime
of a few nanometers, the distorted crystal lattice, tetragonal, for example, will
relax into cubic structure and crystal symmetry breaking will disappear, mak-
ing the ferroelectricity vanished (Shaw et al. 2000). This is true in a freestand-
ing grain without external stress. Several systematic studies of size effects have
been carried out on powders processed to achieve small particles (Jiang et al.
2000; Ishikawa et al. 1996). These studies have focused on the prototypical ferro-
electric perovskite PbTiO3 , and suppression of ferroelectricity was found for the
small particles with diameters from 10 to 20 nm manifested by the increase of
T c compared with the unstrained and stoichiometric bulk value. These data were
extrapolated to predict a ferroelectric critical size of a few nanometers at room
temperature.
However, in a thin film form of ferroelectric material, due to the substrate
clamping effect and interfacial lattice mismatch, the film could be retained in
a tetragonal structure instead of relaxing to a cubic structure. For example, a
BaTiO3 film as thin as 4 unit cells still behaves very good ferroelectric properties
due to the substrate clamping effect that retains its ferroelectricity. In Garcia’s
work (Garcia et al. 2009), BaTiO3 films were grown with thicknesses ranging

96 5 Recent Advances in Ferroelectric Research
2 180
Height (nm)
Phase (°)
0 0
(a) (b) (c)
Figure 5.1 Demonstration of ferroelectricity for a 1-nm BaTiO3 film grown on top of a 30 nm
LSMO electrode. (a) Topography (636 mm2 ) and (b) PFM out-of-plane (OOP) phase image after
writing eight alternating positive (13.5 V) and negative (23.5 V) 430.5-mm2 voltage stripes. (c)
A 232-mm2 dark square is written on top of the pattern with 13.5 V, and a 131-mm2 white
square (23.5 V) is subsequently written on top of that square. Source: Garcia et al. (2009).
Reprinted with permission of Springer Nature.
between 1 and 50 nm on NdGaO3 substrate with a 30-nm-thick La0.67 Sr0.33 MnO3

(LSMO) buffer. It demonstrated the ferroelectricity of BaTiO3 by piezoresponse
force microscopy (PFM) even in 1-nm thick (see Figure 5.1). The retained ferro-
electricity at a few atomic layer thickness is due to the substrate clamping effect
forcing the BaTiO3 film into a tetragonal structure or otherwise should be a cubic
structure at this thickness.
Fong et al. (2004) have investigated this hypothesis by using X-ray diffraction
to record the structure of PbTiO3 thin films as a function of temperature over
a range of different film thicknesses. Their method allows them to distinguish
between the centrosymmetric structure, in which there is no net polarization, and
the off-centered ferroelectric phase. They observed that the ferroelectric state
persists down to a thickness of 12 Å, which corresponds to just three unit cells
of PbTiO3 . The retaining of ferroelectricity in this size is attributed to the for-
mation of domain structure and interfacial charge to screen the depolarization
field. Otherwise, the thickness limit for ferroelectricity to exist may be doubled
as predicted by first-principle calculation.
Retaining ferroelectricity beyond 5-nm thick is essential to realize ferroelectric
tunneling device application. Therefore, the study of size limitation is a guideline
for ferroelectric application, especially in thin films and nanodevices.
5.2 Ferroelectricity in Emerging 2D Materials

As one of the emerging group of new materials, 2D materials exhibit wide range
of symmetry-breaking quantum phenomena such as crystalline order, supercon-
ductivity, magnetism, and charge density wave, which persist in the limit of a
single-unit cell thickness. However, compared with these well-studied proper-
ties, ferroelectricity in 2D materials is less explored. The reason is that, compared
with those bulk ferroelectric materials, the ferroelectricity in the ultrathin films
is much weaker, and therefore the observation of ferroelectric properties is chal-
lenging. One should know that usually only monolayer or a few layers of the 2D
5.2 Ferroelectricity in Emerging 2D Materials 97
materials are ferroelectric; while their bulk is not, since multilayers average out
each layer’s electric polarization.
Since the first prediction of 2D Van der Waals ferroelectric (Wu et al. 2013),
theoretical studies have proposed that 2D ferroelectrics may appear in distorted
monolayer MoS2 (Shirodkar and Waghmare 2014), buckled honeycomb AB
monolayer (DiSante et al. 2015), phosphorene analogues (GeS, GeSe, SnS,
and SnSe) (Wu and Zeng 2016), hydrogenated carbon nitride (Tu et al. 2017),
and chemically functionalized 2D materials (Wu et al. 2016). For instance,
first-principle calculations predict that the phosphorene analogue (GeS, GeSe,
SnS, and SnSe) monolayers are multiferroic with coupled ferroelectricity and fer-
roelasticity. The structure of SnS and SnSe monolayers are non-centrosymmetric,
and the parallel zigzag chains with dipole moments along the armchair direction
may lead to a spontaneous polarization. If the polarization is switchable, these
materials can be classified as intrinsic ferroelectrics.
Let’s have a look at SnSe or SnS as an example to understand its ferroelectric
nature. Figure 5.2 illustrates ferroelectric switching in SnS/SnSe monolayer. One
can see that when an electric field is applied to the right, the Sn atoms with pos-
itive charge will move toward the right from state I; while the S/Se atoms with
negative charge will move toward the left, then reach an intermediate state II with
a square lattice and finally get to the final state III with the polarization switched.
Therefore, this SnSe (or SnS) is intrinsically ferroelectric. It is interesting to note
that SnS/SnSe adopts a layered orthorhombic structure (space group Pnma) at
room temperature in state I originally. However, when the field is applied, the
structure converts to the nonpolar square structure (Cmcm phase) with the inver-
sion symmetry (intermediate state II) similar to the high temperature condition
(Li et al. 2015; Zhao et al. 2014), then back to the Pnma phase.
I ll Ill
P
Sn
S/Se
P
P
P
y
E-field direction
x
Figure 5.2 Illustrations of (top) top view and (bottom) side view along y direction of the
structure showing the pathway of ferroelectric switching for SnS/SnSe monolayer. Black
arrows denote external electric fields in different directions.
At the same time, the spontaneous in-plane polarization in phosphorene

analogues MXs (GeS, GeSe, SnS, and SnSe) was also identified through
first-principle calculations in another work (Fei et al. 2016). Furthermore,
an effective Hamiltonian and parameters were used to investigate the phase
transition via Monte Carlo simulations. The calculated Curie temperatures of
the monolayer GeS and GeSe are rather high, while the Curie temperature of
SnS and SnSe are low (T c_SnS = 1200 K and T c_SnSe = 326 K).
Beyond monolayers, because of the restored inversion symmetry, the polariza-
tion of even-number-layered MXs is always zero; while the odd-number-layered
MXs show spontaneous polarization. Interestingly, although polarization of the
odd-number-layered MXs decays with the increasing thickness, the Ps is still
comparable to ferroelectric BaTiO3 nanostructures (see Figure 5.3). Thus, it is
possible to observe the ferroelectricity in currently available few-layer chalco-
genides.
However, until 2016, 2D ferroelectricity was finally realized experimentally
in atomic-thick SnTe. The scanning tunneling microscope (STM) method was
employed to probe the ferroelectricity of 1-unit cell SnTe film at the liquid
helium temperature (Chang et al. 2016). Except of the observed domain for-
mation, lattice distortion, and band-bending, the polarization manipulation by
electric field was performed, and the results strongly support the occurrence of
ferroelectricity in the 1-unit cell SnTe film (see Figure 5.4).
2.0 Figure 5.3 (a) Temperature dependence of

Armchair direction polarization obtained from Monte Carlo
Eq. (2) fit simulations of monolayer SnSe and (b)
Polarization (10–10 C/m)
1.5 Zigzag direction polarization at zero temperature versus the

layer number of MXs. Source: Fei et al.
(2016). Reprinted with permission of
American Physical Society.
1.0
0.5
0.0
100 200 300 400 500
(a) Temperature (K)
1.0 GeS
Polarization (C/m2)
0.8 GeSe
SnS
0.6 SnSe
0.4
0.2
0.0
1 3 5 7
(b) MX layers
(a) (b) Domain wall
Figure 5.4 (a) The stripe domain of a 1 unit cell SnTe film and (b) schematic of the lattice
distortion and atom displacement in the ferroelectric phase. Source: Adapted from Chang
et al. (2016).
In bulk form, the highest ferroelectric transition temperature T c of SnTe is only

98 K, owing to the screening effect of charge carriers. However, when the thick-
ness of SnTe film decreases, the T c is strongly enhanced, which is near 270 K
for 1-unit cell SnTe. The strong ferroelectricity enhancement in SnTe films can
be attributed to the lower density of defects and larger energy bandgap, and the
in-plane lattice expansion as well.
5.3 Ferroelectric Vortex

To answer the question whether ferroelectric phase transitions still occur
in low-dimensional structures, Naumov and Fu (Naumov et al. 2004) per-
formed ab initio studies on ferroelectric nanoscale discs and rods of PZT
and demonstrated the existence of previously unknown phase transitions in
zero-dimensional ferroelectric nanoparticles. It is found that phase transitions
do exist in zero-dimensional ferroelectric nanostructures, and the minimum
diameter of the discs that display low-temperature structural bistability is
determined to be 3.2 nm. The phase transitions are very different from those
occurring in bulk material; below the critical temperature, they actually lead to
the formation of spontaneous vortices with nonzero local toroid moment rather
than spontaneous polarization.
Vortex as a topological entity exists in nature, from the shape of a flower to
galaxy, for example, as shown in Figure 5.5. In microscopic world, the vortex
domain structure, characterized by the continuous rotation of ferroelectric polar-
ization or magnetic spin, was firstly predicted more than 70 years ago. Landau
and Lifshitz (1935) and Kittel (1946) showed that formation of circular domains
was likely in ferromagnetic nanodots due to the surface boundary conditions.
In 1979, their 2D structure in terms of winding numbers was investigated and
then extended to three-dimensional vortex domains (Mermin 1979; Williams and
Wright 1998). The toroidal domains can arise for two unrelated physical reasons:
finite size effects and boundary conditions.
(a) (b) (c) (d)
Figure 5.5 Vortex structures in nature from flower (a) to snail (b), tornado (c), and galaxy (d).
The vortex domains are well studied in nanoferromagnets, but rarely reported
in ferroelectrics experimentally since it was extended to ferroelectrics around
1984 (Ginzburg et al. 1984) and predicted by Naumov and Fu (Naumov et al.
2004) from ab initio studies. Compared with magnetic vortex domain, the
ferroelectric vortex is more difficult to be formed due to its metastable state
and is also technologically difficult to be observed. Gruverman et al. (2008)
reported the direct observation of a doughnut-shaped domain pattern appearing
during polarization reversal in micrometer-sized ferroelectric capacitors (see
Figure 5.6). Domain imaging of the PZT disc was performed through the top
electrode of circular 1-μm diameter and square 2.5-μm edge by applying a read-
ing oscillating bias on a 100 ns scale. For the circular capacitors, an unswitched
domain remains in the center forming a vortex domain pattern. However, the
vortex structure remains only for >1 second, after that the unswitched central
domain collapses to leave a uniformly polarized ground state. Though it is
difficult to make unambiguous detection of the in-plane polarization, a 2D map
of the in-plane polarization distribution Pxy shows that the vortex structure is
consistent with the pattern as predicted by simulation, which is supported by
the model of Naumov and Fu.
Recently, Lane Martin and coworkers (Damodaran et al. 2017; Yadav et al.
2016) explored the coexistence of polarization vortices with other forms of polar
1 μs 3 μs Pxy
Figure 5.6 PFM images of instantaneous domain configurations with out-of-plane

polarization component Pz developing at different stages, (a) 1 μs and (b) 3 μs, of polarization
reversal in 1-μm-diameter circular capacitor. (c) two-dimensional maps of polarization
components Pxy corresponding to an instantaneous domain pattern in the 1-μm-diameter
circular capacitor during switching. Source: Adapted from Gruverman et al. (2008).
A single vortex–antivortex pair
SrTiO3
PbTiO3
The curl of the polar displacement
[001]pc
5 nm
[100]pc
Cross-sectional high-resolution STEM image Phase-field simulated polarization vectors
Figure 5.7 Exploring the phase boundary between (SrTiO3 )10 /(PbTiO3 )10 superlattice and
vortex–antivortex phases in each PbTiO3 layer. Source: Yadav et al. (2016). Reprinted with
ordering in finely layered perovskite oxides. They demonstrated how to arrange

different regions in an easily controllable and precise fashion throughout a
system where different orderings are in very close competition with each other
like the PbTiO3 /SrTiO3 superlattice system (see Figure 5.7). Because of the
competition between the tendency of the oxygen octahedral to rotate and the
off-centering of the ions that leads to polarization in a thin layered superlattice
of PbTiO3 and SrTiO3 , a hybrid improper ferroelectricity was produced. The
up and down polarization domains tend to have sharp boundary polarization
transitions, which are described as stripe domains. However, as the DyScO3
substrates have a larger lattice spacing, when this lattice spacing is imposed on
the superlattice, it is easy for the polarization to be in-plane. Under this strain
condition, with tuned layer thickness, highly ordered vortex-like domains form.
When a field is applied in the same direction with both polarization orders, it is
found that it switches the film into a pure vortex state using a conductive atomic
force microscope (AFM) tip, while the opposite direction should maintain the
vortex state. These unique features make the vortex ferroelectricity a potential
for interesting applications in memory technology if one can control these
responses in a local and non-volatile way.
More recently, skyrmion in ferroelectrics or ferroelectric controlled skyrmion
in ferromagnetic systems is becoming another hot topic of research in ferro-
electric community. First proposed by Skyrme in the 1960s (Skyrme 1962), the
localized, particle-like configurations are topologically protected that they can-
not be changed by a continuous deformation of the field configuration unless very
strong perturbations to the system. The configurations in this model are called
skyrmions, and the topologically protected skyrmions can be stabilized in chiral
magnets in the form of stable spin textures, referred to as magnetic skyrmions.
(a) (b)
Figure 5.8 (a) Schematic configurations of a single chiral skyrmion and (b) radial skyrmion
with the corresponding cross-section along the dashed line. The arrows indicate the direction
of the spins, and their colors represent the normal component to the plane, that is, from up
direction (red) to the down direction (blue). Source: Kezsmarki et al. (2015). Reprinted with
y=0 y=π y = –π/2 y = π/2
m=1 ny
nx
m = –1
Figure 5.9 Skyrmion structures with varying vorticity m and helicity 𝛾. The arrows indicate the
direction of the in-plane order parameter component, and the brightness indicates the normal
component to the plane, with white denoting the up direction and black the down direction.
Source: Nagaosa and Tokura (2013). Reprinted with permission of Springer Nature.
The typical spin textures corresponding to a single skyrmion and radial skyrmion
are shown in Figure 5.8a and Figure 5.8b, respectively, each of them shows a vec-
tor field that all the vectors point along a given direction at the external surface
of the skyrmion.
Figure 5.9 shows a few patterns of the different possible tangent vector fields
in 2D xy-plane. Skyrmion structure can be indexed by vorticity m and helicity 𝛾,
where m stands for vorticity and measure how much the pattern rotates and 𝛾
indicates the initial angle that the vector field makes with the x-axis. With this
definition, the skyrmion in Figure 5.8a can be indexed as m = 1 and 𝛾 = 𝜋/2, and
the one shown in Figure 5.8b corresponds to m = 1 and 𝛾 = 𝜋.
By integrating ferroelectricity to magnetic skyrmion system, ferroelectrically
controlled magnetic skyrmions in ultrathin oxide heterostructures may be
formed. As reported by Wang et al. (2018), ferroelectric control of skyrmion
1.0
0.4 ρAHE + ρTHE
0.5
0.2
ρxy – R0H (μΩ cm)
M (μB/Ru)
0.0 0.0
–0.2
–0.5
ρAHE (M)
–0.4 B20S4 ρTHE
–1.0
–4 –2 0 2 4
μ0H (T)
Figure 5.10 Ordinary 𝜌xy , anomalous 𝜌AHE and topological Hall resistivities 𝜌THE versus H for
the BTO/SRO bilayer heterostructures. The 𝜌AHE –H curve is derived from the normalized M–H
curve. The unit of M is μB/Ru. The H-sweeping directions are marked by solid arrows. The inset
is the schematic skyrmion structure. Source: Adapted from Wang et al. (2018).
properties in ultrathin BaTiO3 /SrRuO3 (BTO/SRO) bilayer heterostructures

were first observed via PFM and Hall measurements. They first examined
the basic skyrmion properties by determining transverse Hall resistivity 𝜌xy .
Figure 5.10 shows that the skyrmions are excitations from the ferromagnetic
background, and their emergence is assisted by the ferromagnetic domain
switching. When magnetic field H is aligned parallel with the core spin orien-
tation of the skyrmion and approaches the coercive field (H C ), topological Hall
resistivity (𝜌THE ) increases sharply and peaks at approximately ±1.65 T. Further
sweeping H across H C , 𝜌THE starts to decrease gradually and vanishes at the
critical field (±3.9 T), at which the M–H hysteresis loop closes.
Wang then employed an AFM-based electrical gating approach to realize
the FE-mediated controllability. With different fraction of upward-poling, the
relationship between polarized area and skyrmion phase is demonstrated in
Figure 5.11. By selectively poling the sample at room temperature (Figure 5.11a),
the upward-poled Hall bar exhibits a slight enhancement in 𝜌THE compared with
that in the pristine FE state (Figure 5.11b).
It should be noted that the skyrmions studied in the earlier works can only
be observed at low temperatures in epitaxially grown ultrathin films, and this
type of skyrmions has not been directly relevant for technological application.
However, based on most recent works, the skyrmions can also be found at room
temperatures, which may pave the way for the development of the ultimate small
achievable size for non-volatile memory application.
B20S4 (a) 200

+ –
–8 V
Phase (°)
Vtip = –8 V P +8 V +8 V +8 V
–8 V –8 V
0
Room-T domain writing

(b) 0.4
10 K
ρxy (μΩ cm) 0.2
ρxy
0.0
–0.2 Hsk
ρTHE + ρAHE
–0.4 ρTHE
Low-T hall measurement –4 –2 0 2 4 –4 –2 0 2 4 –4 –2 0 2 4 –4 –2 0 2 4

μ0H (T) μ0H (T) μ0H (T) μ0H (T)
(A) (B)
Figure 5.11 (A) Schematic diagram of the experimental setup for FE domain switching and
Hall measurements. (B) PFM phase images (a), hall resistivity signals (b) of the BTO/SRO bilayer
in different FE poling states. The scale bar in (B) corresponds to 10 μm. The critical H as 𝜌THE
decreases to zero is marked as Hsk . Source: Adapted from Wang et al. (2018).
5.4 Molecular Ferroelectrics

Molecular materials have attracted much attention due to their advantages of
easy modification and fabrication, environmental-friendly and less energy-cost
processing, transparent, light weight, mechanical flexibility, and possible multi-
functionalities, compared with their counterparts of pure inorganic or metallic
ones. Many functionalities and properties including ferroelectricity have been
discovered and realized in molecule-based materials (Zhou et al. 2012; Long and
Yaghi 2009; Cheetham and Rao 2007; Rao et al. 2008). Molecular ferroelectrics
have been proven to be applicable in many device applications such as tempera-
ture sensing, data storage, and energy harvesting.
The first ferroelectrics, Rochelle salt, which was discovered in 1920, is in fact
a molecular material, and later on a few other molecular systems, such as the
well-known potassium dihydrogen phosphate and tri-glycine sulfate, were dis-
covered and developed. However, compared with pure inorganic ferroelectrics
such as BaTiO3 and PZT, molecular ferroelectrics usually have much lower spon-
taneous polarization (Ps ) and dielectric constant (𝜖 ′ ), but relative larger dielec-
tric loss (𝜖 ′′ or tan 𝛿) (Lines and Glass 2001; Jona and Shirane 1962). However,
these shortcomings for molecular ferroelectrics have been gradually overcome
(Horiuchi et al. 2007, 2010, 2012; Horiuchi and Tokura 2008; Fu et al. 2011).
For example, an organic ferroelectric, croconic acid has been reported to pos-
sess a Ps of 21 μC/cm2 at room temperature, comparable to BaTiO3 (Horiuchi
et al. 2010). Very recently, a research team led by Ren-Gen Xiong has reported
(Fu et al. 2013) that a molecular ferroelectric crystal of diisopropylammonium
bromide salt (DIPAB, Figure 5.12) showed a Ps of 23 μC/cm2 , high T c of 426 K,
5.4 Molecular Ferroelectrics 105
b
b α β
c
a c a
Figure 5.12 The structures of the two phases of DIPAB, α phase at 293 K and the β phase at
438 K, and the phase transition routes. Color scheme: C – black, N – cyan, H – white,
Br – golden. Green dashed bonds are N—H· · ·Br hydrogen bonds. Source: Adapted from Fu
et al. (2013).
large 𝜖 ′ up to 103 , and low tan 𝛿 of ∼0.4%, all close to or beyond the pure inorganic
BaTiO3 , representing a breakthrough in the research of molecular ferroelectrics.
Figure 5.12 shows the two phases 𝛼 and 𝛽 of DIPAB crystal. Below the Curie
temperature (T c = 426 K), the 𝛼 phase is in the monoclinic chiral space group P21 ,
which belongs to the polar point group C 2 . When heating the crystals to temper-
atures above T c , it results in the high temperature 𝛽 phase of a centrosymmetric
structure belonging to the monoclinic space group P21 /m in the nonpolar point
group C 2h . The 𝛼 to 𝛽 phase transition is reversible, and accompanied to the tran-
sition, the 𝛼 crystal displays a prominent dielectric anomaly at T c = 426 K, with
the peak 𝜖 ′ as high as 1.6 × 103 at the lowest frequency of 400 Hz (Figure 5.13a).
The polar 𝛼 crystal displayed typical ferroelectric hysteresis loops (Figure 5.13b,
inset). This result opens a new avenue to the applications of molecular ferro-
electrics that are much more environmental friendly, easier to fabricate, lighter
weight, and so on, compared with the pure inorganic oxide ferroelectrics.
1500 400 Hz 20
1 kHz 416 K 418 K
10 10
10 kHz
Ps (μC/cm2)
P (μC/cm2)
1000 100 kHz 0

0
ε′
1 MHz
–10
500 –10
420 K 422 K
–10 0 10
E (kV/cm)
–20
0
380 400 420 440 380 350 400 450
(a) T (K) (b) T (K)
Figure 5.13 (a) The temperature dependence of the dielectric constant of the 𝛼-DIPAB crystal
under several frequencies. (b) The Ps versus T plots measured by a pyroelectric technique, and
inset, the ferroelectric hysteresis loops under several temperatures below T c . Source: Adapted
from Fu et al. (2013).
5.5 Ferroelectricity in HfO2 and ZrO2 Fluorite Oxide

Thin Films
Perovskite-structure ferroelectrics have been intensively studied among the
various ferroelectric materials, and the structural origin of the spontaneous
polarization in perovskite materials is well defined as the oxygen octahedron
distortion, i.e. the relative change in positions of the B-site cation and O2− ions
in a tetragonal ABO3 structure. Recently, ferroelectricity has also been found in
fluorite-structure oxide, mainly HfO2 thin films. Fluorite oxide consists of four
metal and eight oxygen ions in a unit cell. As shown in Figure 5.14, metal ions
are located at eight corners and six face centers, while oxygen ions are at the
eight tetrahedral sites.
Unlike the perovskite-structure, the ferroelectricity of fluorite-structure oxide
has been raised by the distortion of oxygen hexahedron. Figure 5.15 shows
the origins of ferroelectricity in perovskite and fluorite structures. As reported
by Materlik et al. (2015), in fluorite structure, the oxygen hexahedron with
body-centered metal ion can be found in two adjacent unit cells for the cubic
phase, and a nonpolar phase is formed due to the centrosymmetry. When the
cubic unit cell is elongated along a-axis, the orthorhombic polar phase can
be formed, in which four oxygen ions are significantly shifted from the stable
position in a cubic phase, and each of them has two stable positions. As a result,
a ferroelectric phase is formed.
One of the well-known examples of fluorite oxide, hafnium dioxide (HfO2 ),
exhibits three different crystal phases at normal pressure: a monoclinic phase
(P21 /c, m-phase) at room temperature, a tetragonal phase (P42 /nmc, t-phase)
above 2050 K, and finally a cubic phase (Fm3m, c-phase) above 2803 K. It is
known that, same as ZrO2 , HfO2 exhibits a martensitic transformation to the
monoclinic phase when its thin films are cooled down from high temperature
Figure 5.14 Illustration of fluorite

oxide structure, large and small
spheres denote oxygen and metal
ions, respectively.
Nonpolar phase Polar phase
Oxygen octahedron in
perovskite oxide
: Oxygen ion
: Metal ion
Oxygen hexahedron in
fluorite oxide c
a/b
Figure 5.15 A schematic for the structural origin of the ferroelectricity in perovskite and
fluorite ferroelectrics. Source: Adapted from Park et al. (2017).
tetragonal phase. Although t-phase and c-phase can be stabilized at room tem-
perature via doping or nanostructuring, all these structures are centrosymmetric
and reported as non-ferroelectric. However, the recent discovery of the possible
factor such as doping and crystallization with capping layer, HfO2 structure can
be stabilized into non-centrosymmetric orthorhombic phase (Pbc21 ) exhibiting
ferroelectricity (see Figure 5.16).
In 2011, Böscke et al. firstly observed a non-centrosymmetric orthorhombic
phase during the study of silicon doped HfO2 , which exhibited a remnant polar-
ization (Pr ) of ∼10 μC/cm2 at a coercive field of 1 MV/cm (see Figure 5.17). In fact,
ferroelectricity (FE) or antiferroelectric (AFE) can be induced in HfO2 by several
Figure 5.16 The formation of

the orthorhombic phase
proceeds by transformation
from the tetragonal phase. P
The right arrow indicates two
different polarization states of
the ferroelectric phase (large
atoms are O and small atoms
are Hf ).
T phase, P42/nmc
Ferroelectric O phase, Pca21

20
2.6 mol%
15
3.1 mol%
4.3 mol%
10
5.6 mol%
Polarization (μC/cm2)
–5
–10
–15
–20
–2 –1 0 1 2
Field (MV/cm)
Figure 5.17 Polarization measurement of metal–insulator–metal capacitor samples with

different SiO2 admixture. A gradual transition from ferroelectric to antiferroelectric polarization
curves is observed with increasing silicon content. Source: Adapted from Böscke et al. (2011).
Table 5.1 The ferroelectric properties of HfO2 films with various dopants (O: observable, X: not
observable).
Dopant Theoretical Experimental Doping

(valence) Pr (𝛍C/cm2 ) Pr (𝛍C/cm2 ) E c (MV/cm) concentration (%) AFE
Si (4+) 41 24 0.8–1.0 4 O
Zr (4+) 50 34 5 50 X
Y (3+) 40 24 1.2–1.5 5.2 X
Al (3+) 48 16 1.3 4.8 O
Gd (3+) 53 12 1.75 2 X
Sr (2+) 48 23 2.0 9.9 X
La (3+) N/A 45 1.2 N/A N/A
Source: Adapted from Park et al. (2015) and Wei et al. (2018).
dopants (Y, Gd, Al, Sr, La, Mg, Ba, Nd, Sm, Er, Ga, In, Co, N, and Ni), especially
for ZrO2 -doped HfO2 . As shown in Table 5.1, the largest reported value of Pr
is 45 μC/cm2 , observed in La-doped HfO2 film. On the other hand, researchers
have succeeded to suppress the monoclinic phase in sub-10 nm HfO2 by con-
trolling the oxygen fraction in deposition ambient, and a Pr of 13.5 μC/cm2 in a
6.9-nm-thick dopant-free HfO2 film has been observed by Ashish Pal et al. (2017).
HfO2 has a large bandgap (Eg > 5 eV), a larger remnant polarization
(Pr ∼ 45 μC/cm2 ) and coercive field (∼1–2 MV/cm) than conventional ferro-
electric films. The recent discovery of ferroelectricity in HfO2 , which only
appears at the nanoscale and becomes stronger at smaller dimensions, makes
it able to retain ferroelectricity for extremely thin films (<10 nm). Also, the
simple binary oxide structure with high dielectric constant (high-k), silicon
compatibility, and low toxicity makes it very suitable for applications as a gate
dielectric in semiconductor MOS-FET for ferroelectric random-access memory
(FeRAM) application, which has been introduced in Chapter 3. FE/AFE-HfO2
film shows promising applications in ferroelectric random access memory,
pyroelectric energy harvesting and storage.
It is interesting to note that the ferroelectricity of HfO2 only appears at
nanoscale. For the conventional ferroelectrics, the ferroelectric properties in
thin film form can only be retained under the compressive clamping effect with
the substrate. By contrast, remnant polarizations of pure or doped HfO2 is
larger than 15 μC/cm2 when thickness is 10 nm or even thinner (Park et al. 2013,
2014; Wei et al. 2018; Polakowski and Müller 2015). For instance, Polakowski
and Müller (2015) show the P–V and C–V hysteresis loops for a thickness
series of pure HfO2 thin films crystallized in the presence of a PVD-TiN top
electrode (Figure 5.18). It shows the degrading of ferroelectricity when the film
becomes thicker. To explain it, they claimed that the gradual destabilization of
the presumably orthorhombic phase in favor of the monoclinic phase is observed
with increasing thickness.
20
10
20 nm
12 nm
–10 10 nm
8 nm
6 nm
–20
–2.5 –2.0 –1.5 –1.0 –0.5 0.0 0.5 1.0 1.5 2.0 2.5
Field (MV/cm)
Figure 5.18 P–E hysteresis loops for a thickness series of pure HfO2 thin films crystallized in
the presence of a PVD-TiN top electrode. Source: Adapted from Polakowski and Müller (2015).
Several researchers have also studied the phase evolution in HfO2 film with
different thicknesses (Park et al. 2013, 2014, 2017; Wei et al. 2018). The following
evolution trends as a function of thickness are summarized by the experimental
result of grazing incidence X-ray diffraction (GIXRD) and transmission electron
microscopy (TEM) observations. In ultrathin film (thickness < 10 nm), the
amount of o-phase is the highest, and m-phase is absent or very rarely found.
When the thickness increases, the crystal size (grain size) grows further, the
m-phase starts forming with relieving the internal stress. Once the film is
significantly thick (thickness > 20 nm), the m-phase crystal structure appears as
the predominant phase. Thus, the effective electric field applied to ferroelectric
phase is decreased as a consequence of the increase amount of non-ferroelectric
phase. It can therefore explain the generally decreasing polarization with the
increasing thickness.
Since the scaling of the physical thickness of ferroelectric thin film is the
fundamental issue for FeRAM fabrication, the thickness of the conventional
perovskite-structure ferroelectrics is limited to achieve robust ferroelectricity.
The remnant polarization of perovskite oxide is not greater than that of fluorite
oxide, and the difficulty of depositing perovskite-structure ferroelectrics with
two or more cation components in high-aspect-ratio structure via atomic layer
deposition (ALD) is a critical issue. This makes fluorite-structure ferroelectric
able to replace conventional ferroelectrics for memory applications.
However, in 2013, Zhou et al. discovered a critical problem, the so-called
“wake-up effect,” in Si-doped HfO2 ferroelectric film. This effect refers to the
increase in remnant polarization with increasing cycling number of switching
field. Unlike conventional ferroelectrics such as BaTiO3 and PZT, the ferroelec-
tricity of HfO2 relies on large number of switching cycles, making it difficult for
application. Figure 5.19 shows the general switching I–E curves and remnant
polarization change of ferroelectric HfO2 -based thin films with three different
states of field cycling. From the graph, one can see that the three states, starting
Wake-up Fatigue
Pr Wake-up
Pr,rel Fatigue
Polarization (a.u.)
Pr /Pr,relax (a.u.)
Current (a.u.)
Pristine
Electric fied (a.u.)
10–1 100 101 102 103 104 105 106 107 –4 –2 0 2 4

(a) Cycles (b) Electric fied (MV/cm)
Figure 5.19 (a) Evolution of the ferroelectric remnant polarization during bipolar cycling (4 V
at 10 kHz) measured on Sr:HfO2 -based ferroelectric capacitors. (b) Current–voltage
characteristics for three states, pristine, wake-up, and fatigue. Source: Park et al. (2018).
Reprinted with permission of Cambridge University Press.
from pristine, wake-up to fatigue, can be observed during the cycling. The
remnant polarization as a function of cycles shows two distinct trends as
described by Pešić et al. (2016):
i. “Wake-up” – increases with cycling. This increase of the Pr value corresponds
to an opening or de-pinching of the pristine pinches hysteresis loop with field
cycling.
ii. “Fatigue” – the Pr value decreases resulting in a closure of the difference
between positive and negative remnant polarizations after a certain number
of cycles due to aging mechanisms.
As shown in the current–voltage characteristics curve, the switching current
curve has double peaks in both positive and negative bias regions in pristine
state and, consequently, forms a pinched hysteresis loop. It is different from
the ideal switching current curve of a ferroelectric, where only a single peak is
seen at each region. However, the switching current peaks move toward each
other with continuous bipolar cycling and finally merge together in the wake-up
regime. When cycling further increases, the switching polarization decreases,
which corresponds to the typically observed fatigue in many ferroelectrics.
According to the discussion of Pešić et al. (2016) and Schenk et al. (2014),
the behavior of splitting of switching current peaks in the pristine state may
come from different built-in field within the film. The inhomogeneous internal
electric field distribution can originate from polycrystallinity of the film and the
unevenly distributed charged defects, such as oxygen vacancies. A high oxygen
vacancy concentration is expected to be present near the electrodes due to
the reduction of the doped HfO2 layer by the metal nitride electrodes, whose
distribution may be mostly asymmetric. Therefore, the resulting asymmetric
distribution of the oxygen vacancies can be an origin of the internal field in
the pristine material. When it is in cycling process, the oxygen vacancies at
the interface may diffuse into the bulk regions of the ferroelectric-HfO2 -based
films, and their distribution becomes more homogeneous, in which the wake-up
process will occur, accompanied by the merging of the switching current
peaks.
Another mechanism of the wake-up effect is discussed as the field cycling-
induced phase transition. From the observation of a decrease in the dielectric
constant and an increase in remnant polarization with an increasing number
of cycling, a phase transition from a non-ferroelectric phase to a ferroelectric
phase was suggested in 2015 (Lomenzo et al. 2015). Later on, many experimen-
tal results show both direct and indirect evidences suggesting a phase transi-
tion to the ferroelectric o-phase from non-ferroelectric phase, either t-phase or
m-phase. In Figure 5.20, the phase evolution trend shows the schematic tran-
sition to o-phase in pristine, wake-up, and fatigue states, which indicates that
oxygen vacancy distribution as well as the phase transition plays a significant role
in fluorite-structure ferroelectric wake-up process.
Owing to the industrial-compatible fabrication process, most of the
ALD-grown HfO2 films are polycrystalline and contain multiple phases.
That is the reason why a wake-up effect and further treatment have to be done
before obtaining the better ferroelectricity. Recently, highly oriented ferroelectric
TiN
Mix phase
TiOxNy
O-p O- O-
has ph ph
e as as
Vo e e
M- M-
M-p ph ph
has as as
e Sr;HfOx e e
Mixed phase
TiO2
TiN
(a) (b) (c)
Figure 5.20 Phase evolution model of the three different stages of the ferroelectric
capacitance lifetime. (a) Prestine, (b) wake-up, and (c) fatigue. Source: Adapted from Pešić et al.
(2016).
Hf0.5 Zr0.5 O2 thin films are deposited on (001)-oriented LSMO/STO substrate

using pulsed-laser deposition (Wei et al. 2018), and without wake-up cycles,
the film under epitaxial compressive strain and predominantly (111)-oriented
displays large ferroelectric polarization values up to 34 μC/cm2 . Also, it is
curious that Wei et al. discovered a new phase with ferroelectricity besides
conventional phase, i.e. rhombohedral polymorph instead of orthorhombic
phase. Their results proposed that pulsed laser deposition (PLD)-grown thin film
at high temperature enables the in situ crystallization of Hf0.5 Zr0.5 O2 (HZO).
Due to the favorable epitaxial relationships induced by the STO/LSMO stack,
the growing crystallites are subjected to a large epitaxial compressive strain that
elongates the cubic unit cell along the out-of-plane [111] direction, inducing
rhombohedral symmetry with a polar unit cell (see Figure 5.21).
Our group has been working on HfO2 thin films for high-k gate dielectric appli-
cation for many years, and we have first reported the successful growth of epi-
taxial HfO2 on Si and have done pioneer work in doping Al into HfO2 forming
amorphous HfAlOx film (Dai et al. 2003; Lee et al. 2003). Our recent work also
demonstrates the structural characterization and ferroelectric properties of HfO2
film on mica. Mica is emerging as a good candidate for flexible device. Compared
with organic polymers, mica holds the advantages of high stability and sustain-
able for high temperatures, making it very suitable for functional oxide growth
that are usually grown at temperatures higher than 500 ∘ C. With a suitable oxide
electrode epitaxially grown on mica, high-quality HfO2 film (even epitaxy) can
be grown.
Our recent work also demonstrates the structural characterization and
ferroelectric properties of Al-doped HfO2 film on different oriented SrTiO3 . As
a preliminary trial, we have grown HfAlOx film by PLD deposited (La,Sr)MnO3
film on STO substrate. Figure 5.22 shows the HRTEM image of PLD-grown
doped HfO2 /LSMO on STO (111) substrate and its corresponding PFM ampli-
tude and phase hysteresis loops. For the ferroelectric property characterization,
the piezoresponse (the out-of-plane amplitude and phase hysteresis loops) of
35 nm thick Al doped HfO2 on LSMO/STO (111) can be clearly seen. From this
result, coercive voltages of ±4 V can be obtained at the minima of the amplitude
signal.
[111]
[111]
[111] [11-1]
[11-1] [11-1]
LSMO (φ = 15°)
2 nm
(a)
b
[111] a
[-111]
[11-1] ~71° [1-11]
120°
a b
c
(b) (c)
Figure 5.21 (a) Cross-sectional high-angle annular dark field-scanning transmission electron
microscopy (HAADF-STEM) image of a 4-nm-thick film, observed along the zone [111], (b)
sketch of the proposed rhombohedral structure of the HZO film with polarization along the
[111] direction, and (c) two views of the R3m phase obtained for epitaxially compressed HZO
and HfO2 . Green (cyan) spheres represent Hf/Zr (O) atoms. The arrows show the polar
distortion with respect to a reference paraelectric structure with appropriately R3m state.
Source: Wei et al. (2018). Reprinted with permission of Springer Nature.
300 OFF-field
90
250
PFM amplitude (a.u.)
PFM phase (°)
Column-like 200
0
AI:HfO2 150
100
50 –90
LSMO
10 nm [111] 0
–20 –10 0 10 20
Voltage (V)
(a) (b)
Figure 5.22 (a) TEM image of Al:HfO2 /LSMO and (b) PFM amplitude and phase hysteresis
loops from doped HfO2 on STO (111) substrate.
5.6 Ferroic Properties in Hybrid Perovskites

Recent studies have shown that hybrid perovskites have the property of spon-
taneous electric polarization like many inorganic perovskites. For an oxide per-
ovskite ABO3 , the spontaneous electric polarization, i.e. ferroelectricity, comes
from the broken of crystal inversion symmetry, where the B cation shifts slightly
from the geometric center of the BO6 octahedron (Glazer 1972). When the A
point is taken by polar organic cations, the central-symmetry breaking is fur-
ther strengthened because of the organic cations’ intrinsic asymmetric atomic
structure. According to the Berry phase calculation within the modern theory
of polarization (Resta 1994; King-Smith and Vanderbilt 1993), J.M. Frost et al.
(2014a) reported the magnitude of the bulk polarization of four hybrid lead halide
perovskite, which is shown in Table 5.2.
For CH3 NH3 PbI3 , the most widely used material to fabricate hybrid perovskite
solar cells, 38 μC/cm2 of electric polarization has been predicted, which is in
the same level of the ferroelectric oxide perovskite like KNbO3 (30 μC/cm2 )
(Dall’Olio and Dovesi 1997). This group also reported a temperature and electric
field-dependent rich domain structure of twinned molecular dipoles in the
hybrid perovskite by classical Monte Carlo simulation. They propose that the
internal electrical fields associated with microscopic polarization domains
contribute to hysteretic anomalies in the current–voltage response of hybrid
organic–inorganic perovskite solar cells (Frost et al. 2014b).
The reason why the hybrid perovskite solar cells have such unexpected good
performance is not completely understood by scientists today, but a recently
proposed theory claimed that the photovoltaic operation is largely enhanced
by the strong polarization in the following three ways. Firstly, the high local
polarization-induced electric field at the domain walls makes the exciton
separation more efficiently (Yang et al. 2010). Secondly, the long electron and
hole diffusion length results in a low possibility of recombination (Stranks et al.
2013). Thirdly, the open circuit voltage of devices is larger than the bandgap of
the material because of the built-in potential.
As the hybrid perovskite solar cells are being developed rapidly, the idea to
fabricate efficient solar cells based on ferroelectric oxides is also demonstrated
Table 5.2 Calculated properties of four hybrid lead halide perovskites from
density functional theory.
Pseudo-cubic Lattice electronic

Cation lattice constant a (Å) polarization 𝚫P (𝛍C/cm2 )
NH4 6.21 8
CH3 NH3 6.29 38
CF3 NH3 6.35 48
NH2 CHNH2 6.34 63
Source: Frost et al. (2014a). Reproduced with permission of American

Chemical Society.
5.6 Ferroic Properties in Hybrid Perovskites 115
500 200
400
150
Amplitude (a.u.)
50
Phase (°)
300
0 100
200
–50 50
100
200 nm –100
0
0
–4 –3 –2 –1 0 1 2 3 4
(a) (b)
Voltage (V)
Figure 5.23 (a) Surface morphology of CH3 NH3 PbI3 perovskite crystal showing strip-like
domain structure and (b) piezoelectric butterfly loop of CH3 NH3 PbI3 perovskite crystal
(J.Y. Dai 2013 unpublished work).
by Grinberg et al. (2013). Our group has been studying the existence of ferro-
electricity in CH3 NH3 PbI3 single crystal for many years. Figure 5.23 shows the
AFM/PFM results of CH3 NH3 PbI3 single crystal where the surface presents
domain structure and PFM strain-bias butterfly loop. These results suggest that
CH3 NH3 PbI3 single crystal may have ferroelectric property.
Our group has tried to combine ferroelectric material into this perovskite
solar cell structure as shown in Figure 5.24. We proposed structures utilizing fer-
roelectric nanodots embedded in the CH3 NH3 PbI3 perovskite or a ferroelectric
network structure at the bottom and top electrodes. It is expected that polarized
Polarized discrete
ferroelectric layer
Polarized discrete D Polarized discrete
TA
AD
ferroelectric layer
3
ferroelectric layer
bl
3
bl
P Me
eT
3P
3P
Au
Au
M
-O
H
H
-O
3N
iro
3N
iro
Sp
H
P
H
P
Sp
C
C
O
O
Zn
O
Zn
ss
IT
IT
ss
la
G
la
G
(b)
(a)
AD
3
eT
bl
3P
M
Au
-O
H
iro
3N
Sp
H
C
P
Zn
Ferroelectric nanodots
IT
ss
la
G
SEI 5.0 kV ×10 000 1 μm

(c) (d)
Figure 5.24 (a) A discrete ferroelectric layer in between electron-collection layer and ITO
electrode, (b) double ferroelectric layers at cathode and anode, (c) ferroelectric nanodots
embedded 0–3 nanocomposite perovskite solar cell structure, and (d) hexagonal nanocells of
PZT discrete film on ITO/glass made by sol–gel method followed by nano-patterning
technique. The conductivity of the ITO layer underneath has negligible degradation during the
PZT synthesis process (J.Y. Dai and H.J. Fang 2014 unpublished).
ferroelectrics may further prevent the recombination of electron and holes

generated by sunlight in the CH3 NH3 PbI3 perovskite. However, it is difficult to
experimentally increase the efficiency benefited from ferroelectric composite.
Alternating polar and nonpolar structures in ferroic domains resolves a key
puzzle of CH3 NH3 PbI3 that they exhibit strong piezoresponse in polar domains.
Yet, the domains cannot be switched by an electric field applied either locally
through a SPM tip or globally through external electrodes (Strelcov et al. 2017).
However, most recently, Huang’s group believe this is precisely the consequence
of alternating polar and nonpolar domains, since polar domains that are normally
switchable by an electric field are now locked by nonpolar ones that sandwich
them (Huang et al. 2018).
Figure 5.25 shows domains structures of CH3 NH3 PbI3 crystals revealed by
SEM, AFM topography, polarized optical microscopy, and PFM. It is apparent
that they all exhibit characteristic lamellar patterns that have been often seen
in ferroic materials. Striking PFM amplitude mapping acquired through single
frequency scanning is seen in Figure 5.25d, and the alternating domains exhibit
large contrast in piezoresponse up to one order of magnitude difference. This
pm
250
0.0
2 μm 2 μm
–250
(a) (b)
a.u.
545
310
4 μm 2 μm
75.0
(c) (d)
Figure 5.25 Ferroic domain patterns of CH3 NH3 PbI3 crystals revealed by (a) SEM, (b) AFM
topography, (c) polarized optic microscope, and (d) vertical PFM amplitude. Source: Adapted
from Huang et al. (2018).
References 117
suggests that those alternating domains have their different origins. It is found
that polar and nonpolar orders of CH3NH3PbI3 can be distinguished from their
distinct lateral piezoresponse, energy dissipation, first and second harmonic
electromechanical couplings, and temperature variation, even though their dif-
ference in crystalline lattice is very subtle. These findings resolve key questions
regarding polar nature and ferroelectricity of CH3NH3PbI3 and its implication
on photovoltaics.
References
Böscke, T.S., Müller, J., Bräuhaus, D. et al. (2011). Ferroelectricity in hafnium oxide
thin films. Applied Physics Letters 99 (10): 102903. Available at: https://doi.org/10
.1063/1.3634052.
Chang, K., Liu, J., Ji, S.H. et al. (2016). Discovery of robust in-plane ferroelectricity
in atomic-thick SnTe. Science 353 (6296): 274–278. Available at: http://science
.sciencemag.org/content/353/6296/274.
Cheetham, A.K. and Rao, C.N.R. (2007). There’s room in the middle. Science 318
(October): 58–60.
Dai, J.Y., Lee, P.F., Wong, K.H. et al. (2003). Epitaxial growth of yttrium-stabilized
HfO2 high-k gate dielectric thin films on Si. Journal of Applied Physics 94 (2):
Dall’Olio, S. and Dovesi, R. (1997). Spontaneous polarization as a Berry phase of the
Hartree–Fock wave function: the case of KNbO3 . Physical Review B: Condensed
Matter 56 (16): 10105–10114.
Damodaran, A.R., Clarkson, J.D., Martin, L.W. et al. (2017). Phase coexistence and
electric-field control of toroidal order in oxide superlattices. Nature Materials 16:
1003. Available at: http://dx.doi.org/10.1038/nmat4951.
DiSante, D., Stroppa, A., Barone, P. et al. (2015). Emergence of ferroelectricity and
spin-valley properties in two-dimensional honeycomb binary compounds.
Physical Review B: Condensed Matter 91 (161401(R)): 1–6.
Fei, R., Kang, W., and Yang, L. (2016). Ferroelectricity and phase transitions in
monolayer group-IV monochalcogenides. Physical Review Letters 117 (097601):
1–6.
Fong, D.D., Stephenson, G.B., Streiffer, S.K. et al. (2004). Ferroelectricity in ultrathin
perovskite films. Science 304 (5677): 1650–1653. Available at: http://science
.sciencemag.org/content/304/5677/1650.
Frost, J.M., Butler, K.T., Brivio, F. et al. (2014a). Atomistic origins of
high-performance in hybrid halide perovskite solar cells. Nano Letters 14 (5):
2584–2590. Available at: https://doi.org/10.1021/nl500390f.
Frost, J.M., Butler, K.T., and Walsh, A. (2014b). Molecular ferroelectric
contributions to anomalous hysteresis in hybrid perovskite solar cells. APL
Materials 2 (8): 81506. Available at: https://doi.org/10.1063/1.4890246.
Fu, D.-W., Zhang, W., Xiong, R.G. et al. (2011). Diisopropylammonium chloride: a
ferroelectric organic salt with a high phase transition temperature and practical
utilization level of spontaneous polarization. Advanced Materials 23 (47):
5658–5662. Available at: http://dx.doi.org/10.1002/adma.201102938.
Fu, D.-W., Cai, H.L., Liu, Y. et al. (2013). Diisopropylammonium bromide is a

high-temperature molecular ferroelectric crystal. Science 339 (6118): 425–428.
Garcia, V., Fusil, S., Bibes, M. et al. (2009). Giant tunnel electroresistance for
non-destructive readout of ferroelectric states. Nature 460 (7251): 81–84.
Ginzburg, V.L., Gorbatsevich, A.A., Kopayev, Y.V., and Volkov, B.A. (1984). On the
problem of superdiamagnetism. Solid State Communications 50 (4): 339–343.
Available at: http://www.sciencedirect.com/science/article/pii/
0038109884903818.
Glazer, A.M. (1972). The classification of tilted octahedra in perovskites. Acta
Crystallographica B28 (11): 3384–3392. Available at: https://doi.org/10.1107/
S0567740872007976.
Grinberg, I., Vincent West, D., Rappe, A.M. et al. (2013). Perovskite oxides for
visible-light-absorbing ferroelectric and photovoltaic materials. Nature 503:
509–512. Available at: http://dx.doi.org/10.1038/nature12622.
Gruverman, A., Wu, D., Fan, H.J. et al. (2008). Vortex ferroelectric domains. Journal
of Physics: Condensed Matter 20 (34): 342201. Available at: http://stacks.iop.org/
0953-8984/20/i=34/a=342201.
Horiuchi, S. and Tokura, Y. (2008). Organic ferroelectrics. Nature Materials 7:
357–366. Available at: http://dx.doi.org/10.1038/nmat2137.
Horiuchi, S., Kumai, R., and Tokura, Y. (2007). Hydrogen-bonded donor–acceptor
compounds for organic ferroelectric materials. Chemical Communications (23):
2321–2329. Available at: http://dx.doi.org/10.1039/B617881B.
Horiuchi, S., Tokunaga, Y., Tokura, Y. et al. (2010). Above-room-temperature
ferroelectricity in a single-component molecular crystal. Nature 463: 789–792.
Available at: http://dx.doi.org/10.1038/nature08731.
Horiuchi, S., Kagawa, F., Hatahara, K. et al. (2012). Above-room-temperature
ferroelectricity and antiferroelectricity in benzimidazoles. Nature
Communications 3 (1308). Available at: http://dx.doi.org/10.1038/ncomms2322.
Huang, B., Zhao, J., Li, J. et al. (2018). Ferroic domains regulate photocurrent in
single-crystalline CH3 NH3 PbI3 films self-grown on FTO/TiO2 substrate. npj
Quantum Materials 3 (30): 1–8.
Ishikawa, K., Nomura, T., Okada, N., and Takada, K. (1996). Size effect on the phase
transition in PbTiO3 fine particles. Japanese Journal of Applied Physics 35 (9S):
5196. Available at: http://stacks.iop.org/1347-4065/35/i=9S/a=5196.
Jiang, B., Peng, J.L., Bursill, L.A. et al. (2000). Size effects on ferroelectricity of
ultrafine particles of PbTiO3 . Journal of Applied Physics 87 (7): 3462–3467.
Jona, F. and Shirane, G. (1962). Ferroelectric Crystals. Oxford/London/New
York/Paris: Pergamon Press.
Kezsmarki, I., Bordacs, S., Milde, P. et al. (2015). Neel-type skyrmion lattice with
confined orientation in the polar magnetic semiconductor GaV4 S8 . Nature
Materials 14 (11): 1116–1122.
King-Smith, R.D. and Vanderbilt, D. (1993). Theory of polarization of crystalline
solids. Physical Review B: Condensed Matter 47 (3): 1651–1654.
Kittel, C. (1946). Theory of the structure of ferromagnetic domains in films and
small particles. Physical Review 70 (11–12): 965–971.
References 119
Landau, L.D. and Lifshitz, E.M. (1935). On the theory of the dispersion of magnetic
permeability in ferromagnetic bodies. Physikalische Zeitschrift der Sowjetunion 8:
153–164.
Lee, P.F., Dai, J.Y., Wong, K.H. et al. (2003). Growth and characterization of
Hf–aluminate high-k gate dielectric ultrathin films with equivalent oxide
thickness less than 10 Å. Journal of Applied Physics 93 (6): 3665–3667. Available
at: https://doi.org/10.1063/1.1554764.
Li, C.W., Hong, J., Delaire, O. et al. (2015). Orbitally driven giant phonon
anharmonicity in SnSe. Nature Physics 11: 1063. Available at: http://dx.doi.org/10
.1038/nphys3492.
Lines, M.E. and Glass, A.M. (2001). Principles and Applications of Ferroelectrics and
Related Materials. Oxford: Oxford University Press. Available at: http://www
.oxfordscholarship.com/10.1093/acprof:oso/9780198507789.001.0001/acprof-
9780198507789.
Lomenzo, P.D., Takmeel, Q., Nishida, T. et al. (2015). TaN interface properties and
electric field cycling effects on ferroelectric Si-doped HfO2 thin films. Journal of
Applied Physics 117 (13): 134105.
Long, J.R. and Yaghi, O.M. (2009). The pervasive chemistry of metal-organic
frameworks. Chemical Society Reviews 38 (5): 1213–1214. Available at: http://dx
.doi.org/10.1039/B903811F.
Materlik, R., Kunneth, C., and Kersch, A. (2015). The origin of ferroelectricity in
Hf1−x Zrx O2 : a computational investigation and a surface energy model. Journal of
Applied Physics 117 (13): 134109.
Mermin, N.D. (1979). The topological theory of defects in ordered media. Reviews of
Modern Physics 51 (3): 591–648.
Nagaosa, N. and Tokura, Y. (2013). Topological properties and dynamics of
magnetic skyrmions. Nature Nanotechnology 8 (12): 899–911. Available at: http://
www.ncbi.nlm.nih.gov/pubmed/24302027.
Naumov, I.I., Bellaiche, L., and Fu, H. (2004). Unusual phase transitions in
ferroelectric nanodisks and nanorods. Nature 432: 737. Available at: http://dx.doi
.org/10.1038/nature03107.
Pal, A., Narasimhan, V.K., Weeks, S. et al. (2017). Enhancing ferroelectricity in
dopant-free hafnium oxide. Applied Physics Letters 110 (2): 22903. Available at:
https://doi.org/10.1063/1.4973928.
Park, M.H., Kim, H.J., Hwang, C.S. et al. (2013). Evolution of phases and
ferroelectric properties of thin Hf0.5 Zr0.5 O2 films according to the thickness and
annealing temperature. Applied Physics Letters 102 (24): 242905.
Park, M.H., Kim, H.J., Hwang, C.S. et al. (2014). The effects of crystallographic
orientation and strain of thin Hf0.5 Zr0.5 O2 film on its ferroelectricity. Applied
Physics Letters 104 (7): 072901.
Park, M.H., Lee, Y.H., Hwang, C.S. et al. (2015). Ferroelectricity and
antiferroelectricity of doped thin HfO2 -based films. Advanced Materials 27 (11):
1811–1831. Available at: http://dx.doi.org/10.1002/adma.201404531.
Park, M.H., Schenk, T., Schroeder, U. et al. (2017). A comprehensive study on the
structural evolution of HfO2 thin films doped with various dopants. Journal of
Materials Chemistry C 5: 4677–4690.
Park, M.H., Schroeder, U., Hwang, C.S. et al. (2018). Review and perspective on
ferroelectric HfO2 -based thin films for memory applications. MRS
Communications 8 (3): 795–808.
Pešić, M., Fengler, F.P.G., Schroeder, U. et al. (2016). Physical mechanisms behind
the field-cycling behavior of HfO2 -based ferroelectric capacitors. Advanced
Functional Materials 26 (25): 4601–4612.
Polakowski, P. and Müller, J. (2015). Ferroelectricity in undoped hafnium oxide.
Applied Physics Letters 106 (23): 232905.
Rao, C.N.R., Cheetham, A.K., and Thirumurugan, A. (2008). Hybrid
inorganic–organic materials: a new family in condensed matter physics. Journal
of Physics: Condensed Matter 20 (8): 083202. Available at: http://stacks.iop.org/
0953-8984/20/i=8/a=083202.
Resta, R. (1994). Macroscopic polarization in crystalline dielectrics: the geometric
phase approach. Reviews of Modern Physics 66 (3): 899–915.
Schenk, T., Yurchuk, E., Mueller, S. et al. (2014). About the deformation of
ferroelectric hystereses. Applied Physics Reviews 1 (4): 41103. Available at: https://
doi.org/10.1063/1.4902396.
Shaw, T.M., Trolier-McKinstry, S., and McIntyre, P.C. (2000). The properties of
ferroelectric films at small dimensions. Annual Review of Materials Science 30:
263.
Shirodkar, S.N. and Waghmare, U.V. (2014). Emergence of ferroelectricity at a
metal-semiconductor transition in a 1T monolayer of MoS2 . Physical Review
Letters 112 (157601): 1–5.
Skyrme, T.H.R. (1962). A unified field theory of mesons and baryons. Nuclear
Physics 31: 556–569. Available at: http://www.sciencedirect.com/science/article/
pii/0029558262907757.
Stranks, S.D., Eperon, G.E., Snaith, H.J. et al. (2013). Electron-hole diffusion lengths
exceeding 1 micrometer in an organometal trihalide perovskite absorber. Science
342 (6156): 341–344. Available at: http://science.sciencemag.org/content/342/
6156/341.
Strelcov, E., Dong, Q., Li, T. et al. (2017). CH3 NH3 PbI3 perovskites: ferroelasticity
revealed. Science Advances 3 (4): e1602165. Available at: http://advances
.sciencemag.org/content/3/4/e1602165.
Tu, Z., Wu, M., and Zeng, X.C. (2017). Two-dimensional metal-free organic
multiferroic material for design of multifunctional integrated circuits. Journal of
Physical Chemistry Letters 8 (9): 1973–1978. Available at: https://doi.org/10.1021/
acs.jpclett.7b00636.
Wang, L., Lu, Q., Noh, T.W. et al. (2018). Ferroelectrically tunable magnetic
skyrmions in ultrathin oxide heterostructures. Nature Materials 17 (December):
1087–1094. Available at: http://dx.doi.org/10.1038/s41563-018-0204-4.
Wei, Y., Nukala, P., Noheda, B. et al. (2018). A rhombohedral ferroelectric phase in
epitaxially strained Hf0.5 Zr0.5 O2 thin films. Nature Materials 17 (December):
1095–1101. Available at: http://dx.doi.org/10.1038/s41563-018-0196-0.
Williams, W. and Wright, T.M. (1998). High-resolution micromagnetic models of
fine grains of magnetite. Journal of Geophysical Research, B: Solid Earth 103
(B12): 30537–30550. Available at: http://dx.doi.org/10.1029/98JB01120.
References 121
Wu, M. and Zeng, X.C. (2016). Intrinsic ferroelasticity and/or multiferroicity in

two-dimensional phosphorene and phosphorene analogues. Nano Letters 16 (5):
3236–3241. Available at: https://doi.org/10.1021/acs.nanolett.6b00726.
Wu, M., Zeng, X.C., Jena, P. et al. (2013). Hydroxyl-decorated graphene systems as
candidates for organic metal-free ferroelectrics, multiferroics, and
high-performance proton battery cathode materials. Physical Review B:
Condensed Matter 87 (081406(R)): 1–5.
Wu, M., Liu, J., Zeng, X.C. et al. (2016). Ferroelectricity in covalently functionalized
two-dimensional materials: integration of high-mobility semiconductors and
nonvolatile memory. Nano Letters 16 (11): 7309–7315. Available at: https://doi
.org/10.1021/acs.nanolett.6b04309.
Yadav, A.K., Nelson, C.T., Ramesh, R. et al. (2016). Observation of polar vortices in
oxide superlattices. Nature 530: 198. Available at: http://dx.doi.org/10.1038/
nature16463.
Yang, S.Y., Seidel, J., Byrnes, S.J. et al. (2010). Above-bandgap voltages from
ferroelectric photovoltaic devices. Nature Nanotechnology 5 (2): 143–147.
Zhao, L.-D., Lo, S.H., Kanatzidis, M.G. et al. (2014). Ultralow thermal conductivity
and high thermoelectric figure of merit in SnSe crystals. Nature 508 (7496): 373.
Zhou, H.-C., Long, J.R., and Yaghi, O.M. (2012). Introduction to metal–organic
frameworks. Chemical Reviews 112 (2): 673–674. Available at: https://doi.org/10
.1021/cr300014x.
Zhou, D., Xu, J., Li, Q. et al. (2013). Wake-up effects in Si-doped hafnium oxide
ferroelectric thin films. Applied Physics Letters 103 (19): 192904.
123
Piezoelectric Effect: Basic Theory
6.1 General Introduction to Piezoelectric Effect

Piezoelectric effect refers to the generation of electric charges or voltage in
response to a mechanical stress or the generation of strain under applied
electric field. It has been explained in Chapter 2 that all ferroelectrics are
also piezoelectric since the electric field-induced polarization switching and
domain wall movement will generate strain. However, one should note that even
without polarization switching (which is a must to be ferroelectric), changes of
polarization strength under electric field can also induce strain. That is why some
piezoelectric materials such as ZnO and AlN, which possess electric polarization
due to their non-centrosymmetric crystal structure, are not ferroelectric since
their polarizations are not switchable.
Piezoelectric effect can be classified into direct and converse piezoelectric
effects as illustrated in the diagram (Figure 6.1). Direct piezoelectric effect refers
to the generation of charges under stress, while the converse piezoelectric effect
refers to the generation of strain under electric field.
To design a piezoelectric device, the following parameters should be con-
sidered when selecting a suitable piezoelectric material along with its shape,
size, orientation, etc. Among many parameters, piezoelectric constant (charge
constant and voltage constant), dielectric constant, electromechanical coupling
factor, and mechanical quality factor are the most important. Table 6.1 lists
the piezoelectric properties of some important piezoelectric materials. Before
systematic introduction of all parameters, let us have some general view on how
to select a material for device applications.
In general, for high-power applications, for example, ultrasonic motor, hard
PZT is a good choice: here “hard” means high mechanical quality factor. On the
other hand, to pursue a large piezoelectric response with large mechanical strain
output, soft PZT is a good choice: here, “soft” means large strain that can be eas-
ily induced with relatively lower electric field but the output force is relatively
small (its mechanical quality factor is smaller compared with hard piezoelectric
materials). While for ultrasound sensor application, detectors made of PVDF
may be more sensitive, even its piezoelectric charge constant is much smaller
compared with PZT. The reason is that the very large piezoelectric voltage con-
stant of PVDF makes it able to output relatively higher voltage under mechanical

124 6 Piezoelectric Effect: Basic Theory
Applied stress Induced strain
+ + + +
Piezoelectric
plate P
– – – –
Electrodes
Direct piezoelectric effects Converse piezoelectric effects
Figure 6.1 Direct and converse piezoelectric effects.
Table 6.1 Piezoelectric properties of representative piezoelectric materials.
PZT 4 PZT 5H PVDF–TrFE

Parameters Quartz BaTiO3 (hard) (soft) copolymer
d33 (pC/N) 2.3 190 189 593 33

−3
g 33 (10 V m/N) 57.8 12.6 26.1 19.7 380
kt 0.09 0.38 0.51 0.5 0.3
kp 0.33 0.58 0.65
𝜖r 5 1700 1300 3400 6
Qm >105 500 65 3–10
T c (∘ C) 120 328 193
Source: From Uchino (2009) and Chen (2013).
stress. The following sections will give a detailed description of all parameters in
piezoelectric materials.
6.2 Piezoelectric Constant Measurement

According to the directions of electric polarization z 3
and electric field as well as the vibration mode, the
piezoelectric constant and electromechanical cou-
6
pling factor are expressed by d33 , d31 , d15 , k 33 , k 31 ,
and k 15 , where the indications 1, 3, and 5 are defined
P 5 2
in Figure 6.2. Designing different devices needs to
consider different parameters such as dielectric con- 4 y
stant, mechanical compliance, etc. To get the whole 1
set of parameters, usually five samples with different
x
vibration modes are needed.
For an anisotropic piezoelectric material, its phys- Figure 6.2 Definition of
ical constants are related to both the direction of the directions for piezoelectric
constants.
applied mechanical or electrical force and the direc-
tions perpendicular to the applied force. Therefore,
each constant should have two subscripts indicating the directions of the two
related quantities, electric field and strain, for piezoelectricity. The direction of
positive polarization is usually made to coincide with the z-axis of a rectangular
system of x-, y-, and z-axes (Figure 6.2). Directions of x, y, and z are represented
by the subscripts 1, 2, and 3, respectively, and shear about one of these axes is
represented by the subscripts 4, 5, and 6, respectively.
Piezoelectric constant is an important indicator of a material’s suitability
for strain-dependent applications. The following give definition of the most
frequently used physical constants and their interrelating for piezoelectric
materials, such as piezoelectric charge constant d, the piezoelectric voltage
constant g, and the dielectric permittivity 𝜖.
6.2.1 Piezoelectric Charge Constant

The piezoelectric charge constant, dij , is the amount of charges generated per unit
area (polarization P generated) of mechanical stress (𝜎) applied to a piezoelectric
sample or, alternatively, is the mechanical strain (𝜀) generated to a piezoelectric
sample per unit of electric field E applied. As illustrated in Eq. (6.1), the first sub-
script i indicates the direction of the applied field strength or, alternatively, is the
direction of induced polarization when the electric field is zero or constant. The
second subscript j is the direction of the induced strain generated or the applied
stress in the sample, respectively:
( ) ( )
𝜕Pi E 𝜕𝜀j 𝜎
dij = = (6.1)
𝜕𝜎j 𝜕Ei
In Eq. (6.1), the first term corresponds to the direct piezoelectric effect and the
second term corresponds to the converse piezoelectric effect. The superscript E
indicates a zero or constant electric field, and the superscript 𝜎 indicates a zero
or constant stress.
Based on the definition of indication and orientation as shown in Figures 6.2
and 6.3, the often used d33 , d31 , and d15 are defined as follows:
d33 is the induced polarization (electric displacement) in direction 3 per unit
stress applied in direction 3 or induced strain in direction 3 per unit electric
Piezoelectric plate
Electrodes
Polarization direction
(a) (b) (c) Stress direction
Figure 6.3 Diagram showing the definitions of (a) d33 , (b) d31 , and (c) d15 .
field applied in direction 3. In this case, the applied electric field or stress should
be parallel to direction in which piezoelectric plate is polarized.
d31 is similar to the definition of d33 , but for d31 , the applied stress or the induce
strain should be along the direction 1, and the measured polarization or the
applied E-field should be in direction 3. In this case, the applied electric field
or stress should also be parallel to the direction in which piezoelectric plate is
polarized.
d15 is the induced shear strain in direction 5 per unit electric field applied in direc-
tion 1 or the sample’s polarization generated along direction 1 by a shear stress
applied in direction 5. In this case, the applied electric field or stress should be
perpendicular to direction in which piezoelectric sample is polarized.
There are two contributions to the generated polarization Pi , i.e. the applied
stress and electric field. With the absence of an applied field, the polarization
produced by a stress 𝜎 jk , which is a second-rank tensor, is
Pi = dijk 𝜎jk (6.2)
where dijk is the third-rank tensor of piezoelectric charge constant.

The second contribution to induced polarization Pi from the applied elec-
tric field is instantaneous dielectric response to the electric field. It can be
quantified by
Pi = 𝜖0 𝜒ij Ej (6.3)
where 𝜒 ij denotes the dielectric susceptibility measured at constant stress.
By combining these two contributions, the generated polarization to a linear
piezoelectric material can be expressed as
Pi = dijk 𝜎jk + 𝜖0 𝜒ij Ej (6.4)
It should be noted that this expression is only right at low field regime where
the linear relationship is hold. One should also note that dijk is a third-rank ten-
sor, but by invoking elastic symmetries along with electric symmetries due to
poling, the tensor can be reduced to second rank, i.e. what has been introduced
as dij , typically d33 , d31 , and d15 . In fact, d31 = d32 ; and d15 = d24 , and all others
are zero.
For the converse piezoelectric effect, the strain is expressed as
𝜀ij = Sijkl
E
𝜎kl + dkij Ek (6.5)
where SE is the fourth-rank compliance tensor measured at constant field.
6.2.2 Piezoelectric Voltage Constant

The piezoelectric voltage constant, g, is the electric field generated by a piezoelec-
tric material per unit of mechanical stress applied or, alternatively, is the mechani-
cal strain experienced by a piezoelectric material per unit of electric displacement
applied. g is important for assessing a material’s suitability for sensing (sensor)
applications, and the very often used g 33 , g 31 , and g 15 are defined in the same way
as Figures 6.2 and 6.3, i.e.
( ) ( )
𝜕Ei E 𝜕𝜀j 𝜎
gij = − = (6.6)
𝜕𝜎j 𝜕Di
where the “−” sign is due to the fact that the field built from electric polarization
reduces under positive compressive stress.
Dielectric permittivity is the linkage between the piezoelectric charge and volt-
age constants. This can be understood by the relation between voltage and charge,
whereas their ratio is capacitance, which is proportional to permittivity. There-
fore, the relationship between d and g is
d
g= (6.7)
𝜖0 𝜖r
From the relationship of d and g, we can understand why PVDF is a good choice
for sensor applications such as hydrophone and pressure sensor. It is the very low
dielectric constant that gives rise to a larger g factor, even its d factor is much
smaller than PZT; while PZT is excellent for transducer application due to its
large d value, making it able to generate large strain under electric field.
6.2.3 Dielectric Permittivity

The dielectric permittivity, 𝜖 = 𝜖 0 𝜖 r (𝜖 r is relative dielectric constant and
𝜖 0 = 8.85 × 10−12 F/m is vacuum dielectric constant), is the dielectric dis-
placement per unit electric field for a piezoelectric material. Usually, 𝜖 𝜎 is the
permittivity at constant stress, and 𝜖 𝜀 is the permittivity at constant strain.
Dielectric permittivity is second-rank tensor that can be expressed as 𝜖 ij , where
the first subscript i to 𝜖 indicates the direction of the dielectric displacement and
the second j is the direction of electric field.
6.2.4 Young’s Modulus (Elastic Stiffness)

Young’s modulus Y is an indicator of the stiffness (elasticity) of a material. Y is
determined from the value of the stress applied along an axis to the material
divided by the value for the resulting strain along that axis in the range of stress
in which Hooke’s law holds:
𝜎ij
Yijkl =
𝜀kl
6.2.5 Elastic Compliance

Elastic compliance Sijkl is the strain produced in a material per unit of stress
applied and is the reciprocal of the modulus of elasticity (Young’s modulus, Y ).
SD is the compliance under a constant electric displacement; SE is the compliance
under a constant electric field. The first two subscripts indicate the direction of
strain, and the last two is the direction of stress 𝜎:
𝜀ij = Sijkl 𝜎kl
The elastic compliance and Young’s modulus follow the equation

1
Sijkl =
Yijkl
For the general anisotropic piezoelectric material, the stress–strain relation for
a linear elastic material in terms of matrices can be written as
⎡ 𝜀11 ⎤ ⎡ S1111 S1122 S1133 2S1123 2S1131 2S1112 ⎤ ⎡ 𝜎11 ⎤
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ 𝜀22 ⎥ ⎢ S2211 S2222 S2233 2S2223 2S2231 2S2212 ⎥ ⎢ 𝜎22 ⎥
⎢ 𝜀 ⎥ ⎢ S S S 2S3323 2S3331 2S3312 ⎥ ⎢ 𝜎33 ⎥
⎢ 33 ⎥ = ⎢ 3311 3322 3333 ⎥ ⎢ ⎥
⎢ 2𝜀23 ⎥ ⎢ 2S2311 2S2322 2S2333 4S2323 4S2331 4S2312 ⎥ ⎢ 𝜎23 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ 2𝜀31 ⎥ ⎢ 2S3111 2S3122 2S3133 4S3123 4S3131 4S3112 ⎥ ⎢ 𝜎31 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣ 2𝜀12 ⎦ ⎣ 2S1211 2S1222 2S1233 4S1223 4S1231 4S1212 ⎦ ⎣ 𝜎12 ⎦
To simplify, the tensor notation can be rearranged based on the indices in
Figure 6.2, i.e. 11 → 1; 22 → 2; 33 → 3; 23 → 4; 31 → 5; 12 → 6. The utilization of
the structural and electric symmetries along with matrix reindexing then yields
the system
ε1 E
S11 E
S12 E
S13 0 0 0 0 0 d31 σ1
ε2 E E E
S12 S22 S13 0 0 0 0 0 d31 σ2
ε3 E
S13 E
S13 E
S33 0 0 0 0 0 d33 σ3
ε4 0 0 0 E
S44 0 0 0 d15 0 σ4
ε5 = 0 0 0 0 E
S44 0 d15 0 0 σ5
ε6 0 0 0 0 0 E
S66 0 0 0 σ6
σ
D1 0 0 0 0 d15 0 ϵ11 0 0 E1
σ
D2 0 0 0 d15 0 0 0 ϵ 11 0 E2
D3 d31 d31 d33 0 0 0 0 0 ϵ σ33 E3
It summarizes the converse and direct piezoelectric effects. In matrix form, the
components can be written as
𝜀 = SE 𝜎 + d∗ E
D = d𝜎 + 𝜖 𝜎 E
where d* denotes the transpose of d.
Note: With all the previously mentioned definitions, strain and dielectric con-
stant share the same letter 𝜀 as in most literatures, i.e. 𝜀 is used to represent strain
and dielectric constant. For friendly reading, in this book, 𝜀 and 𝜖 denote strain
and dielectric constant, respectively.
6.2.6 Electromechanical Coupling Factor

The electromechanical coupling factor k ij is defined as the conversion efficiency
between the electrical energy and mechanical energy. The ratio of the stored or
converted (mechanical or electrical) energy to the input of the second energy
(electrical or mechanical) is defined as the square of the coupling coefficient. The

subscripts denote the relative direction along which the electrical and mechanical
quantities and the kind of motion involved:
√ √
mechanical energy stored electrical energy stored
k= or k =
electrical energy applied mechanical energy applied
Depending on the vibration modes of shapes and the directions of electric field
and polarization, the electromechanical coupling factors can be defined as k 33 ,
k 31 , k 15 , k t , and k p . Detailed information about the requirement of testing samples
to satisfy these vibration modes will be introduced in later sections.
6.2.6.1 How to Measure Electromechanical Coupling Factor?

A high electromechanical coupling factor k is desirable for efficient energy con-
version. In a well-designed system, piezoelectric elements can exhibit k factor
over 90%. As mentioned earlier, vibration mode-dependent k ij factors can be
defined as illustrated in Figure 6.4, where k 33 , k 31 , and k 15 are defined same as the
way of d33 , d31 , and d15 . While k t is for a thin round disc vibrating in thickness
direction with electrodes on two faces, k p is also for a thin round disc but vibrat-
ing along the radial direction (perpendicular to direction in which piezoelectric
material is polarized). The electric field and polarization direction are the same
as the case of k p in direction 3 (parallel to the thickness in which piezoelectric
material is polarized).
There are two different methods to measure the k factors. One is from the mea-
surement of strain under electric field at static or low frequencies where the d
factor, mechanical compliance, and dielectric permittivity are involved. The other
one is the high-frequency impedance measurement at resonance state.
Poling direction
k33
k15
kp, kt
1 k31 Electrodes
Figure 6.4 Diagram showing the definition (the relationship between the poling and electric
field direction, where the electric field direction is parallel to the two electrodes) of k33 , k31 , k15 ,
kt , and kp.
Under Static or Low Frequency For a rod/bar sample with 33 vibration mode,
2
2
d33
k33 = E 𝜎
S33 𝜖33
For a plate sample with 31 vibration mode,
2 D
d31 S11
2
k31 = E 𝜎
=1− .
S11 𝜖33 E
S11
2
2
d15
k15 = E 𝜎
S15 𝜖11
For a disc sample with thickness vibration mode,
2 𝜀
d33 𝜖33 2
d33
kt2 = = E 𝜎
D
C33 S33 𝜖33
For a disc sample with radial vibration mode (Meitzler et al. 1973),
2 2
2(k P )2 2k31 2d31
kp2 = = = 𝜎
,
1 + υp + 2(k P )2 1 − υp 𝜖33 E
(S11 E
+ S12 )
SE
where υp = − S12E as a planar Poisson’s ratio, and a planar radial piezoelectric cou-
11
p 2
(e31 )2 p 𝜎 2d31
pling coefficient k P is given by (k P )2 = p p
𝜖33 C11
,where 𝜖33 = 𝜖33 − E
S11 E
+S12
,
E
p S11
C11 = E 2 E 2
(S11 ) − (S12 )
p d31
e31 = E E
S11 + S12
High-Frequency Resonance Measurement A more popular way to determine k

factor is through measuring frequency-dependent impedance where the reso-
nance f r (or f m ) and anti-resonance f a (or f n ) states can be identified as shown
in Figure 6.5. Here r denotes resonance, and m means minimum; a denotes
anti-resonance, and n means maximum. Why is the impedance minimum at its
resonance state and maximum at anti-resonance state? This will be explained by
equivalent circuit that is introduced in the following section.
It should be noticed that the sample must be poled before resonance
measurement; otherwise, a piezoelectric material with random polarizations
microscopically and zero polarization macroscopically will not generate strain
and will not present any resonance peak in impedance spectrum.
At higher frequencies, the different k factors at different resonance modes are
given:
For a rod sample with 33 vibration mode,
√ [ ]
𝜋 fa 𝜋 (fa − fr )
k33 = tan
2 fr 2 fa

√ [ ]
√ 𝜋f
√ 𝜋 (fa −fr )
√ 2 fr tan 2 fr
a
k31 = √
√ [ ]
1 + 𝜋2 fa tan 𝜋2 a f r
f (f −f )
r r
For a plate sample with shear vibration mode,

√ [ ]
𝜋 fr 𝜋 (fa − fr )
k15 = tan
2 fa 2 fa
For a disc sample with thickness vibration mode,

√ [ ]
𝜋 fr 𝜋 (fa − fr )
kt = tan
2 fa 2 fa
For a disc sample with radial vibration mode,
2(k P )2
kp2 =
1 + υp + 2(k P )2
where
E
S12
𝜐p = − E
S11
is the planar Poisson ratio.
Nevertheless, in an ultrasonic transducer design, very often, the vibration mode

is not any of the previously mentioned situations due to the restriction of dimen-
sion, so the k factor is called k eff that is quantified as
2
fa2 − fr2
keff =
fa2
Figure 6.5 Frequency-dependent

dielectric permittivity showing fa (fn)
resonance and anti-resonance.
Impedance Z
fr (fm)
Frequency f
Figure 6.6 Equivalent electrical circuit of a

piezoelectric resonator.
Cs Rs Ls
Cp
6.3 Equivalent Circuit

One unique characteristic of a piezoelectric material is the presence of electrical
resonances and anti-resonances in frequency-dependent impedance spectrum.
At frequencies below f r and above f a , the piezoelectric resonator behaves
capacitively; however, between these two frequencies, it behaves inductively.
At resonance state, large strain and capacitance changes are induced, and the
current flow into the sample easily. While at the anti-resonance state, the strains
induced in the sample compensate completely and the current cannot easily
flow into the sample. For each mechanical resonance in the piezo-element,
a resonance/anti-resonance pair will exist in the impedance spectrum. This
characteristic piezoelectric impedance’s behavior can be modeled by the follow-
ing equivalent electrical circuit (see Figure 6.6) representing the piezoelectric
sample’s electrical response as an electrical device. This equivalent circuit is
commonly referred to as Van Dyke’s model and is recommended by the IEEE
Standard on Piezoelectricity.
In this circuit, inductance Ls represents the mass; capacitance C s represents
compliance; resistance Rs represents the internal friction (or dielectric loss) of
the piezoelectric element; and C p is used to represent the capacitance that is
formed because of sample’s mechanical molding. This model is only valid near
the resonance.
This circuit diagram consists of series resonance and parallel resonance, i.e.
two resonance frequencies. If the reactance X C produced by capacitance C s is
equal (in amplitude) and opposite (in phase) to the reactance X L produced by
inductance Ls , the series resonance occurs. Thus, the impedance is the lowest and
approximately equal to the resistance Rs during this condition. If the reactance
of the series wing becomes inductive and equal to the reactance caused due to
capacitance C p , parallel resonance occurs. At this condition, the circuit results in
very high impedance. This can be understood by the following equivalent circuit
analyses.
Now we use a real sample’s impedance spectrum to explain the important
meaning of the equivalent circuit. An example impedance-frequency spectrum
from a relaxor ferroelectric crystal of PMN-0.3PT is shown in Figure 6.7, where
one can see the resonance and anti-resonance peaks as well as 180∘ phase
difference between these two states. Let us analyze the equivalent circuits
by classifying the impedance into A–E five regimes with their corresponding
equivalent circuits shown in Figure 6.8.
6.3 Equivalent Circuit 133
90
fa
B
Impedance Z (Ω)
100
Phase ϕ (º)
E
A C
0
10
fr
D
–90
1
3.0 3.5 4.0 4.5 5.0 5.5 6.0
Frequency f(MHz)
Figure 6.7 Impedance-frequency spectrum of a poled PMN–PT single crystal with f r and f a .
A and E C
XL XL
R R
Cs Rs Rs Ls
Xc Xc
B XL D Rp YL
Ls Rs Cs 1
R
Rp
Lp
Xc Yc
Cp
where YL = 1 = 1
XL jωL
and YC = 1 = jωC
XC
Figure 6.8 Simplified equivalent circuit corresponding to the regions A–E in Figure 6.7.
In the equivalent circuit, impedance Z and its phase angle 𝜑 depend upon the
reactance values of the components; and we know that reactance X is zero when
the circuit element is resistive, positive when the circuit element is inductive, and
negative when it is capacitive. Their impedances are quantified:
ZR = R
1
ZC =
j𝜔C
ZL = j𝜔L
where j is the imaginary notation. In the previously mentioned equations, it is
apparent that the impedance of an inductor is directly proportional to the fre-
quency, while that of capacitor is inversely proportional to frequency. As a result,
in the series LCR circuit, X C is larger than X L when it is at lower frequencies.

Thus, the circuit becomes capacitive that the phase angle becomes negative (i.e.
regime A).
However, in regime B, the inductive reactance magnitude X L increases as fre-
quency increases, while capacitive reactance magnitude X C decreases with the
increase in frequency. At a certain frequency, these two reactances are equal but
opposite in sign, so they will cancel each other and only resistance is left. This
frequency is called resonance frequency f r at which the total impedance reaches
its minimum with Ztotal = R and phase angle 𝜑 = 0. When Z is minimum, i.e.
series resonance occurs,
j(XLs + XCs ) = 0
1
2𝜋f Ls − =0
2𝜋f Cs
1
f = fr = √
2𝜋 Ls Cs
When the frequency further increases, the circuit element becomes inductive
that X L is larger than X C . Thus, it becomes inductive and phase angle is positive
(i.e. region C).
In region D, the piezoelectric material behaves like a parallel LC circuit. As the
frequency increases, the parallel circuit will produce a parallel resonance (also
called anti-resonance) circuit when the resultant current through the parallel
combination is in phase with the applied voltage. At anti-resonance, there is a
large circulating current between the inductor and the capacitor due to the energy
of the oscillations, and then parallel circuit produces current resonance. Dissim-
ilar to region B, the parallel LC circuit acts like an open circuit rather than short
circuit, so the current flow into the circuit is zero at anti-resonance.
For a parallel equivalent circuit, admittance Y can better illustrate the situation,
which is given as
1 1
Y = − j𝜔Cp +
RP j𝜔LP
where at anti-resonance state, the admittances of L and C are the same values
but antiphase resulting in a minimum value of total admittance. Thus, the total
impedance of this resonance circuit at anti-resonance is at its maximum value
creating a circuit condition of high resistance and low current. At f = f a , the
impedance of the circuit represents the so-called maximum dynamic impedance,
i.e. Zd = Z max .
When parallel resonance occurs, i.e. Z is maximum, and the imaginary parts of
Y become zero,
Im(Y ) = 0
so
1
− 2𝜋f LP = 0
2𝜋f Cp
6.4 Characterization of Piezoelectric Resonator Based on a Resonance Technique 135
The solution of this equation is

1
f = fa = √
2𝜋 LP CP
At region E, frequency further increases, the circuit becomes capacitive again,
which is similar to that of region A.
6.4 Characterization of Piezoelectric Resonator Based

on a Resonance Technique
Most parameters of the piezoelectric samples can be evaluated using the fre-
quency spectra of impedance and phase, which are measured by an impedance
analyzer (for example, Agilent 4294A). When an ac signal is applied on a poled
piezoelectric sample, different vibrational modes are excited depending on the
shape of the sample. With the measured resonance and anti-resonance frequen-
cies of specified vibrational modes, most properties of the sample can be eval-
uated. Five samples with specified shapes and polarization directions, including
circular disc, long bar, thin rod, thin square plate, and thin shear plate are usually
used in these measurements (Berlincourt et al. 1964). The piezoelectric parame-
ters related to these geometries are listed in Table 6.2.
Specific resonant modes are excited in samples with specific shapes. In order to
avoid mode coupling and to obtain a pure vibration mode in the impedance and
phase-frequency spectra, thickness and lateral dimensions of the samples should
have substantial difference. In order to use the equations in the IEEE and IRE
standards, assumptions are set for the specific sample listed in the Table 6.3.
The following shows the details of the calculation of the material parameters,
including the elastic compliance constant SE and SD , elastic stiffness constant Y E
and Y D , piezoelectric voltage constant g, piezoelectric charge constant d, elec-
tromechanical coupling factor k, and frequency constant N. The parameters with
superscript D and E represent conditions of the constant electric displacement
and constant electric field, respectively. The constant electric displacement con-
dition corresponds to open circuit (zero current across the electrodes), while the
constant electric field corresponds to the short circuit condition (zero voltage
across the electrodes).
6.4.1 Length Extensional Mode of a Rod

As illustrated in Figure 6.9, a bar of piezoelectric PZT ceramic with its
length along direction 3 (the poling direction), and its end faces electrodes
E D
(normal to the direction 3), is used to determine S33 , S33 , and k 33 . From the
impedance-frequency spectrum, the resonance and anti-resonance frequencies
f r and f a can be determined, from which its piezoelectric, mechanical, and
electrical parameters can be calculated using
[ ]
2 𝜋 fr 𝜋 (fa − fr )
k33 = tan
2 fa 2 fa
Table 6.2 Resonance modes of the five geometries and their related piezoelectric parameters.
Piezoelectric modes of vibration and resonance
Shape Dimensions Poling direction Vibration mode Resonant mode Related parameters
l – length, Piezoelectric Mechanical
w – width,
t – thickness,
d – diameter
𝜎
t > 10d Along t Thickness extension Thickness kt , 𝜖33 D
C33 , C33
E
, S33
E
, S33
D
, Qt
mode resonance
𝜀
l > 5(t, w) Along t Transverse length Length extensional k31 , d31 , g31 , 𝜖33 D
S11 , S11
E
, Q31
mode resonance
𝜎 𝜀
l > 3.5(t, w) Along l Thickness shear Thickness shear k15 , d15 , g15 , 𝜖11 , 𝜖11 D
C55 , S55
D
, S55
E
, Q15
mode resonance
𝜎
d > 10t Along t Thickness extension Thickness kt , 𝜖33 D
C33 , C33
E
, S33
E
, Qt
mode resonance
𝜀 𝜎
d > 10t Along t Radial mode Planar resonance kp , 𝜖33 , 𝜖33 𝜎 E , S11
E
, Qp
Table 6.3 The assumption for specific samples (l – length,

w – width, t – thickness, and D – diameter).
Sample Assumption
Circular disc d≫t

Square plate l, w ≫ t
Long bar l ≫ w, t or t ≫ l, w
Shear plate l, w ≫ t
Figure 6.9 A long rod with the electric

3
field parallel to its length.
Electrodes
Poling
t
direction ∼
d 2
D 1
S33 =
𝜌(2tfa )2
D
E
S33 1
S33 = =
2
1 − k33 𝜌(2tfr )2
√
E 𝜎
d33 = k33 S33 𝜖33
d33
g33 = 𝜎
𝜖33
where the dielectric constant 𝜖ij𝜎 can be determined from the capacitance C0𝜎 mea-
sured at 1 kHz, i.e. 𝜖 LF (a frequency substantially lower than the lowest resonance
frequency of the crystal plate), in which case the measurements yield the dielec-
tric permittivity as constant stress. The equation is given by
𝜖ij𝜎 A
C0𝜎 =
t
where A is the area of the electrode. Usually, the constant strain or “clamped”
relative permittivity 𝜖ij𝜀 can be measured at high frequency (frequencies that are
high compared with the principal natural frequencies of the plate but well below
any ionic resonances), i.e. 𝜖HF = 𝜖ij𝜀 .
𝜖ij𝜀 A
C0𝜎 =
t
In case of length extensional mode,
𝜀
C0𝜎 t
𝜖HF = 𝜖33 =
A
𝜀
𝜎
𝜖33
𝜖LF = 𝜖33 = 2
1 − k33
6.4.2 Extensional Vibration Mode of a Long Plate

E
To determine the transverse mode constants S11 and k 31 , a piezoelectric long plate
sample is used (see Figure 6.11). The lateral extensional mode is excited, and sim-
ilar to Figure 6.10, the corresponding resonance and anti-resonance frequencies
f r and f a can be found, from which electromechanical coupling factor k 31 can be
calculated as
[ ]
𝜋 fa 𝜋 (fa −fr )
2 fr
tan 2 fr
2
k31 = [ ]
1 + 2 f tan 𝜋2 a f r
𝜋 fa (f −f )
r r
E
And its dielectric permittivity and elastic compliance S11 can be determined by
𝜎
C0𝜎 t
𝜖33 = 𝜖LF =
wl
𝜎
𝜖33
𝜖HF = 2
1 − k31
E 1
S11 =
𝜌(2lfr )2
100
1000 000 80
60
40
Impedance Z (Ω)
20
Phase ϕ (º)
100 000
0
–20
–40
10 000
–60
–80
–100
1000
100 200
Frequency f(kHz)
Figure 6.10 Electrical impedance and phase angle versus frequency spectra of a PZT
piezoelectric ceramic rod.
Figure 6.11 A length-expander plate

3
with the electric field perpendicular to
its length.
Poling
direction
Electrodes
w t
∼
l
2
1
The piezoelectric charge and voltage constants of 31 mode are

√
E 𝜎
d31 = k31 S11 𝜖33
d31
g31 = 𝜎
𝜖33
6.4.3 Thickness Shear Mode of a Thin Plate

Figure 6.12 illustrates the relationship of the electrode surface and polariza-
tion direction of a thin shear piezoelectric plate with electric field applied
along direction 1. A shear mode resonance can be exited and its resonance and
anti-resonance frequencies f r and f a can be found, from which electromechanical
coupling factor k 15 can be calculated as
[ ]
k15 = tan
2 fa 2 fa
Other elastic, piezoelectric, and dielectric constants at this shear mode are
given by
D
C44 = 𝜌(2tfa )2
Figure 6.12 A shear plate with the electric 1

field parallel to its thickness.
Electrodes
t w ∼
3
2 Poling
direction
E D 2
C44 = C44 (1 − k15 )
E 1
S44 = E
C44
𝜀
C0𝜎 t
𝜖11 = 𝜖HF =
wl
𝜀
𝜖11
𝜖LF = 2
1 − k15
√
E 𝜎
d15 = k15 S44 𝜖11
d15
g15 = 𝜎
𝜖11
6.4.4 Thickness Mode of a Thin Disc/Plate

A schematic diagram of the piezoelectric thin plate is shown in Figure 6.13. A pure
thickness resonance mode will be excited by electric field. By finding f r and f a ,
electromechanical coupling factor k t can be calculated as
[ ]
kt = tan
2 fa 2 fa
E D
Elastic stiffness C33 and C33 can be determined by
D
C33 = 𝜌(2tfa )2
E
C33 = 4𝜌fa2 t 2 (1 − kt2 ) = C33
D
(1 − kt2 )
𝜀
The free relative permittivity 𝜖33 can be calculated from the capacitance C 0 with
the equation given by
𝜀
C0𝜎 t
𝜖33 = 𝜖HF =
A
𝜀
𝜖33
𝜖LF =
1 − kt2
3 Figure 6.13 A thin plate with the

electric field parallel to its thickness.
Poling
direction Electrodes
t ∼
d
2
1
References 141
Figure 6.14 A thin disc with the 3

electric field parallel to its thickness.
Electrodes
Poling r
direction t ∼
6.4.5 Radial Mode in a Thin Disc

By using a thin disc, the planar electromechanical coupling coefficient k p and
planar Poisson’s ratio 𝜐p can be deduced. The major surfaces of the disc are fully
covered by electrodes, and it is required that its radius r should be large compared
with its thickness t (r/t > 20). The disc is poled in the thickness direction as illus-
trated in Figure 6.14. The radial mode is a special case that it requires three critical
frequencies to analyze. By measuring the fundamental resonance frequency f r1 ,
anti-resonance frequencies f a , and the second series resonance frequency f r2 and
of a disc, two constants k p and 𝜐p can be obtained (Meitzler et al. 1973). More
parameters calculation can be found elsewhere (Anon 1988).
6.4.6 Mechanical Quality Factor

Mechanical quality factor QM is the ratio of the reactance to the resistance in the
series equivalent circuit. Therefore, it can be determined by the equivalent circuit
analysis function built-in to the Agilent 4294 impedance/gain phase analyzer. The
mechanical quality factor QM can be calculated by
𝜔Ls 1 fa2
QM = =
Rs 2𝜋fr Rs (Cp + Cs ) fa2 − fr2
Electrical quality factor QE is the inverse of the dissipation factor tan 𝛿, where
tan 𝛿 can be measured by the impedance analyzer at 1 kHz:
1
QE =
tan 𝛿
References
Anon (1988). IEEE Standard on Piezoelectricity. In: ANSI/IEEE Std 176-1987.
New York, NY: IEEE Standards Board.
Berlincourt, D.A., Curran, D.R., and Jaffe, H. (1964). Piezoelectric and
piezomagnetic materials and their function in transducers. In: Physical Acoustics:
Principles and Methods (ed. W.P. Mason), 169–270. New York/London: Academic
Press. Available at: https://doi.org/10.1016/B978-1-4832-2857-0.50001-0.
Chen, Y. (2013). High-Frequency and Endoscopic Ultrasonic Transducers Based on
PMN–PT and PIN–PMN–PT Single Crystals. The Hong Kong Polytechnic
University. Available at: http://theses.lib.polyu.edu.hk/handle/200/6985.
Meitzler, A.H., Obryan, H.M., and Tiersten, H.F. (1973). Definition and
measurement of radial mode-coupling factors in piezoelectric ceramic materials
with large variations in Poisson’s ratio. IEEE Transactions on Sonics and
Ultrasonics 20 (3): 233–239.
143
Piezoelectric Devices
Piezoelectric materials ha.ve very broad applications in smart systems as trans-

ducers, including sensors and actuators. It is very easy to list a few dozens of
electronic devices utilizing piezoelectric properties. For example, piezoelectric
materials are active elements in many acoustic devices such as ultrasonic trans-
ducers for medical imaging and sonar system and hydrophone for underwater
detections. Actuators utilizing piezoelectric materials include micro-positioners,
active damping systems, and piezoelectric transformers etc. With piezoelectric
thin films, surface acoustic wave (SAW) devices and micro-electro-mechanical
systems (MEMS) devices have been widely used in telecommunications. More
recently, piezoelectric nanostructures such as ZnO nanowires also found appli-
cation in nano-generators for energy harvesting.
In this chapter, device applications of piezoelectric materials in smart systems
in terms of device design, fabrication, characterization, and modeling will be
introduced with examples from our own research work.
7.1 Piezoelectric Ultrasonic Transducers

Acoustic transducers are devices to convert electrical energy to acoustic energy
and vice versa, and those transducers with operating frequency within the range
of ultrasound are called ultrasonic or ultrasound transducers. Most of the ultra-
sound transducers are made of active elements by piezoelectric materials, so they
are called piezoelectric ultrasound transducers. Applications of ultrasound trans-
ducers are very broad, from ultrasound cleaner to ultrasound bonding and from
non-destructive test (NDT) to medical ultrasound (including therapeutics and
diagnostics). In this session we use medical imaging ultrasonic transducers as
examples to introduce the transducers design, fabrication, characterization, and
modeling.
7.1.1 Structure of Ultrasonic Transducers

The basic structure of an ultrasonic transducer is shown in Figure 7.1, where
the main components include matching layer, backing layer, and piezoelectric
(or active) element. The matching layer between the piezoelectric element and

144 7 Piezoelectric Devices
Piezoelectric element
Backing layer Matching layer
Ultrasonic beam
Input pulse Focusing lens
Figure 7.1 The basic geometry of a single-element ultrasonic transducer while focusing lens
can be convex or concave lens depending on the acoustic velocity between lens and
front-loading medium.
medium allows maximum transmission of the ultrasonic wave, and the backing
layer is to damp the back-radiated acoustic wave to reduce the ring down.
Piezoelectric materials, such as PZT ceramics or relaxor ferroelectric single
crystals, are usually used as the active element to generate and receive ultrasonic
waves. Other piezoelectric materials such as PVDF, LiNbO3 , and lead-free
piezoelectric materials have also been used as active element of ultrasonic
transducers.
Transducers with broad bandwidth (BW) and high sensitivity are important
for ultrasonic imaging applications. Usually one or two acoustic matching layers
are used to increase sound transmission into tissue to compensate the acous-
tic impedance (Z) mismatch between piezoelectric material and human tissue.
Acoustic impedance Z is a parameter used to evaluate the acoustic energy trans-
fer between two materials. Acoustic impedance of a medium is defined as
pressure
Z= (7.1)
volume velocity
For materials,
Z = 𝜌c (7.2)
where 𝜌 and c are the density and the sound velocity, respectively, of the
material. The unit of the acoustic impedance is kg/(m2 s) or Rayl. Table 7.1
lists the acoustic impedance for some nonbiological and biological materials
(Zagzebski 1996).
After applying electrical pulses on the piezoelectric active element, the pres-
sure waves are generated in both back and front directions. The transmission
coefficient (𝛼 T ) evaluates the pressure transmits into the front-loading medium:
4Zpf
𝛼T = (7.3)
(1 + Zpf )2 cos2 (km ⋅ L) + (Zmf + Zpm )2 sin2 (km ⋅ L)
Zi
where Zij = Zj
, and Zp , Zf , and Zm are acoustic impedances of the piezoelec-
tric element, front-loading medium, and matching layer, respectively (refer to
Table 7.1 Acoustic impedances of materials.
Tissue Impedance (MRayl)
Air 0.0004
Fat 1.34
Water 1.48
Blood 1.65
Muscle 1.71
Skull bone 7.8
PZT ∼30
PMN-PT ∼28
Epoxy ∼2.0
Note: The acoustic impedances of PZT, PMN-PT,

and epoxy depend on the composition. The values
given in the table are only correct for some range
of compositions.
Source: Adapted from Zagzebski (1996).
Figure 7.1). k m denotes the wave numbers in matching layer, and L the thick-
ness of matching layer. The transmission coefficient is a function of the acoustic
impedance, frequency, and the matching layer thickness.
Without the matching layers, the transmission coefficient 𝛼 T is calculated to be
only about 10% for the values Zf ∼ 1.5 MRayl for human tissue and Zp ∼ 30 MRayl
for PZT ceramic element. Therefore, it is necessary to insert a matching layer
between the loading medium and piezoelectric element to improve the trans-
mission coefficient of the ultrasonic wave at the interface (Zhou et al. 2011; Zhu
2008). When the matching thickness is equal to quarter wavelength, i.e. L = 𝜆4 ,
the transmission coefficient of Eq. (7.3) can be simplified to be
4Zpf
𝛼T = (7.4)
(Zmf + Zpm )2
For example, if a matching layer has an acoustic impedance of 15 MRayl with
a quarter wavelength thickness, for the previously mentioned PZT-based trans-
ducer, 𝛼 T can reach more than 55%.
7.1.2 Theoretical Models of Ultrasonic Transducer (KLM Model)

There are many sophisticated one-dimensional equivalent circuit models that
are used to simulate the behavior of ultrasonic transducer, such as the Mason
model, the Redwood model, and the Krimholtz–Leedom–Matthae (KLM)
model (Krimholtz et al. 1970). Among them, the KLM model is the most popular
one. The commercial transducer simulation software PiezoCAD (version 3.03 for
Windows, Sonic Concepts, Woodinville, WA), which has been extensively used
in transducer design, is based on the KLM model. The effects of matching layers
and backing material can be readily included as sections of transmission lines.
L/2 L/2
+ I1 I2 + Front acoustic
Back acoustic
V1 Z1ʹ v Z2ʹ v V2 port
port
– –
Acoustic transmission line
+ I3 C Cʹ I= L
Electrical 0
V3
port
–
1: Φ
+V3–
Figure 7.2 KLM electrical equivalent circuit model for a piezoelectric transducer. Source:
Adapted from Kirk Shung (2015).
This model divides a piezoelectric element into two halves, each is represented
by an acoustic transmission line. The acoustic transmission line serves as a
secondary circuit that is linked with an electrical primary circuit by an ideal
transformer as shown in Figure 7.2 (Kirk Shung 2015).
With the KLM model, the electrical impedance of the transducer is given as
1 1 Z1 Z2
Zin = + + (7.5)
j𝜔C0 j𝜔C ′ 𝛷2 (Z1 + Z2 )
where Z1 and Z2 are the input impedances of the acoustic transmission line look-
ing toward the front acoustic port and back acoustic port, respectively, and C 0
and C′ are the clamped and series capacitance due to the acoustic transmission
line, respectively. 𝛷 is the electromechanical turns ratio.
Desilets et al. also determined the optimum impedance of the matching layer(s)
based on the KLM model (Desilets et al. 1978). For one matching layer, the acous-
tic impedance of the matching material Z m satisfies Eq. (7.6)
Zm = (Zp Zf2 )1∕3 (7.6)
In order to enhance front matching, two or more matching layers with quarter
wavelength are often used for transducer fabrication. The acoustic impedances
of two matching layers should be as follows (Desilets et al. 1978):
Zm1 = (Zp4 Zf3 )1∕7 (7.7)
Zm2 = (Zp Zf6 )1∕7 (7.8)
It should be pointed out that usually two matching layers with quarter wave-
length thickness of each layer are used in commercial product as a tradeoff of
price and performance. More matching layers and even gradient matching with
the impedance ranging from the piezoelectric element to human tissue is more
favorable in performance, but the fabrication process is complicated and cost is
very high.
7.1.3 Characterization of Ultrasonic Transducers

Ultrasonic transducers are usually characterized by pulse-echo method in water
tank, from which the basic parameters of transducers such as bandwidth, inser-
tion loss (IL), and resolution can be determined. The generated ultrasound field
distribution can also be measured by hydrophone or other techniques. Figure 7.3
shows the setup of pulse-echo measurement, where the ultrasound beam is gen-
erated by the piezoelectric element that also acts as a sensor to detect the reflected
acoustic wave from the target.
7.1.3.1 Bandwidth (BW)

From the measured echo signal and its Fourier transform as shown in Figure 7.3,
the frequency range emitted by a transducer is termed the characteristic band-
width. The bandwidth of a transducer refers to the frequency range at its −6 dB
points (at which the magnitude of the amplitude in the spectrum is 50% of the
Ultrasound analyzer
Ultrasound
absorb tank
Transducer
Designed focus length
Planar reflecting target Bubble-free

DI water
(a)
Frequency (MHz)
5.0 M 10.0 M 15.0 M
1.5
–10
1.0
–20
0.5 –30
Voltage (V)
Amplitude (dB)
–40
0.0
–50
–0.5
–60
–1.0
–70
–1.5 –80
13.0 μ 14.0 μ 15.0 μ 16.0 μ 17.0 μ
Time (s)
(b)
Figure 7.3 (a) Setup of pulse-echo measurement and (b) typical measurement results
showing echo signal and its Fourier transform in frequency domain.
maximum) of the spectrum. The center frequency f c can be determined as

fl + fu
fc = (7.9)
2
where the f l and f u are lower and upper −6 dB frequencies and the −6 dB band-
width is
f −f
BW = u l × 100% (7.10)
fc
7.1.3.2 Sensitivity of the Transducer

A standard method to characterize a transducer’s sensitivity is using the IL,
which is the ratio of the input power Pi delivered to the transducer from the
driving source to the transducer output power Po . By assuming that the input
load resistance Ri is equal to the output load resistance Ro , the IL is simplified as
the ratio of the pulse excitation voltage V i to the transducer and the echo output
voltage V o .
( ) ( ) ( )
Pi Vi2 ∕Ri Vi
IL = 10 log = 10 log 2
= 20 log (7.11)
Po Vo ∕Ro Vo
To obtain the IL of a transducer by Eq. (7.11), practically, a standard method is

to generate a tone burst using an input sinusoidal wave of 20 cycles with ampli-
tude V i at the resonance frequency of the transducer and received echo wave with
amplitude V o .
7.1.3.3 Resolution
The axial and lateral resolutions are the distinguishable minimum distance
between two reflectors along the axis and perpendicular to the ultrasound
beam direction, respectively. Axial resolution (in depth) is determined by the
ultrasound pulse duration, while the lateral resolution is related to the beam
width in the lateral direction.
For a circular transducer working at center frequency f c , the axial resolution,
Raxial , and the lateral resolution, Rlateral at the focus point are
c
Raxial = (7.12)
2BW ⋅ fc
𝜆d
Rlateral = (7.13)
a
where c is the sound velocity in the loading medium, BW the bandwidth of the
transducer, 𝜆 the wavelength at the loading medium, d the focal length, and a
is the diameter of the aperture (see Figure 7.4). Therefore, increasing the trans-
ducer’s bandwidth and center frequency will improve the axial and lateral reso-
lutions of the imaging.
Practically, the resolution can be determined by imaging metal wire phantom
and find the resolution as shown in the figure. As shown in Figure 7.4, the resolu-
tion can be determined by measuring the width of the −6 dB points in real space
from the image signal intensity.
Figure 7.4 Illustration of Well resolved Barely resolved Not resolved

resolution definition
ultrasound imaging, where
−6 dB is commonly used as
the cut-off attenuation.
a
Resolution
6 dB
Rlateral
7.1.4 Types of Ultrasonic Transducers

7.1.4.1 Medical Application
Applications of the medical ultrasound transducers include diagnostic and thera-
peutic. Therapeutic ultrasounds usually require focused and high power in order
to generate enough heat to burn cancer cells, for example, or break stones in the
kidney. For diagnostic imaging purpose, there are many different types of trans-
ducers such as phase array, linear array, annular array, and radial array. In the
past 10 years, our group has been developing these transducers with relaxor fer-
roelectric PMN-PT single crystals. Figure 7.5 shows different types of medical
ultrasound transducers for diagnostic imaging.
7.1.5 Piezoelectric Film Application in Ultrasound Transducers

In order to increase the ultrasound imaging resolution, people have been trying to
further increase the transducers frequency beyond 100 MHz and even to GHz by
implementing thick or thin piezoelectric films such as PZT, ZnO, and AlN (Zhu
et al. 2008). These very high frequency (f c > 100 MHz) transducers not only have
potential applications as clinical tools for the examination of the anterior segment
of the eye, skin, and intravascular imaging but may also be used in bimolecular
imaging and research. Sol–gel coating of piezoelectric film with thickness in the
order of a few micrometers and vacuum coating of the piezoelectric films with
thickness of 100-nm scale have gain success in fabrication of very high-frequency
transducers beyond 100 MHz.
One interesting application of such piezoelectric-miniature ultrasonic
transducer (PMUT) is piezoelectric-based fingerprint ID system in mobile
phone as shown in Figure 1.8. Ultrasound fingerprint identification system
based on piezoelectric ultrasonic transducer and imaging system can obtain a
three-dimensional (3D) image of fingerprint with a certain depth. To realize this
3D fingerprint image, piezoelectric thick film forming a 2D array is needed to
(a) (b)
(c) (d) (e)
Figure 7.5 Different types of medical ultrasound transducers developed from our group. (a)
Single-element focused transducer, (b) single-element rotation transducer for intravascular
imaging, (c) 128-element radial array transducer for endoscopic ultrasound imaging, (d)
annular array for dynamic focusing eye imaging, and (e) an ultrasound image of rabbit eye by
our 40-MHz focus transducer.
form a 2D phase-array transducer that can control the ultrasound beam to be

focused and scanned electronically.
7.1.6 Challenges and Trend of Developing New Advanced Ultrasound

Transducers
Driven by the progress in new piezoelectric materials and new applications, there
are a few fields of further development of ultrasound transducers:
1. Relaxor piezoelectric-based transducer for high sensitivity and larger band-
width.
2. Lead-free piezoelectric-based transducer for environmental-friendly require-
ment.
3. 2D array transducers for 3D imaging.
4. High-frequency transducers up to GHz for cell imaging.
7.2 Ultrasonic Motor

Suggested by the name, an ultrasonic motor (USM) is driven by a traveling ultra-
sound wave transmitted inside the medium of an USM. A traveling wave is a
mechanical wave that is created by a vibrating object and subsequently travels
through a medium where its propagation involves particle interaction causing
Figure 7.6 Simulation of motion on

stator surface.
MX
displacements of neighboring particles by sequence. Compared with electromag-

netic motor, a USM possesses the advantages of large torque, low speed, high
angular resolution and self-lock when stopped. USM can find its unique appli-
cations in some special situations where electromagnetic motor cannot work
properly or satisfy the job.
Like an earthquake, a mechanical traveling wave in a solid material can be clas-
sified into two general types – Rayleigh wave and flexural wave. The Rayleigh wave
creates large amplitude at the surface, but the wavelength is too large and not
suitable for use in USM. However, the flexural wave propagates with a snake-like
motion (the elliptical motion as shown in Figure 7.6) provides the drive in travel-
ing wave motors. It is obvious that an elliptical motion is created on the surface
of the stator.
There are two stages for the energy conversion in USM. The first stage is to con-
vert electrical energy into mechanical energy through piezoelectric conversion by
a piezoelectric material in which a transverse mode of vibration induces a flexu-
ral wave. In the second stage of energy conversion, the high-frequency oscillation
of stator drives rotor into motion by frictional coupling between rotor and sta-
tor surfaces. Similar to the mechanism introduced by Sashida and Kenjo (1991),
Figure 7.7 illustrates this basic principle of USM driven by flexural traveling wave
and our fabricated linear ultrasonic motor.
7.2.1 Terminologies
There are several terminologies to examine the performance of an USM. Usually,
a larger diameter and larger number of waves result in a larger torque and better
performance of the motor. As indicated, the crests of the waves “push” the rotor
forward for linear or circular motion. It can be predicted that more crests would
provide larger “pushing” force on the rotor.
For an USM, the most important characteristic is the torque – velocity depen-
dence. This is also the most significant factor that describes the performance of an
electromagnetic motor. For USM, another important factor is the admittance Y .
As mentioned, Y can be expressed as G + jB. The admittance helps to understand
the resonance behavior and to find the mechanical quality factor. By plotting the
imaginary part of Y against the real part, a typical Nyquist diagram as shown in
Figure 7.8 is obtained, from which resonance and anti-resonance frequencies,
Pressure
Direction of motion
Elliptical motion Traveling wave

Stator metal
P P P P Bulk piezoelectric material
Electrode
~ A ~ B
A sin ωt B sin ωt
(a)
(b)
Figure 7.7 (a) Schematic diagram of motion between stator and rotor and (b) our fabricated
linear motor based on PZT ceramic generated flexural traveling wave.
Im(Y)
f
fh
j0Cd fs
f1
fp
fa fr
1/Rd 1/Rd+1/R Re(Y)
Figure 7.8 Nyquist diagram around its fundamental resonance and anti-resonance
frequencies. Source: Adapted from El and Helbo 2000.
blocking capacitance and resistance, and other parameters can be found. The
maximum and the minimum of the motional admittance occur at the mechanical
resonance (series) f s and anti-resonance (parallel) f p , respectively. The resonance
and the anti-resonance of the whole system are given by f r and f a , respectively,
whereas the maximum and the minimum of the total admittance are given by f h
and f l .
The characteristic frequencies that are commonly used in the evaluation of the
equivalent circuit parameters are f s and f p . When the quality factor is sufficiently
high (Qm > 100), the pair (f s , f p ) is obtained from the average of the two other
pairs:
(fh , fl ) + (fr , fa )
(fs , fp ) = (7.14)
2
If the G–B diagram obtained from impedance measurement is larger and more
rounded, the resonant characteristics of the motor is better.
The mechanical quality factor Qm can be determined from either the Nyquist
diagram or the Bode diagram (Figure 7.9). In the Bode diagram, the sharpness of
the admittance around the resonance frequency is used to determine Qm from
the pass band at −3 dB of the frequency response.
fs
Qm = (7.15)
Δf−3 dB
7.2.2 Design of USM

In the ultrasonic motor, the stator–ceramic–rotor assembly is made to resonate
to create the flexural traveling waves in order to drive the rotor. As mentioned
before the output torque and speed are directly related to the dimensions of the
stator and the adhesion between the stator and the rotor. Depending on the appli-
cation of individual ultrasonic motor, for rotors requiring smaller dimensions, the
torque output would be smaller since the torque depends on the radius, and it is
difficult to fabricate ceramics with large number of waves in small radius.
Figure 7.10 shows diagram of a USM stator structure and a piezoelectric
ceramic disc used in ultrasonic motor. Sectors labeled with dots are polarized
up and the rest sectors are poled down. In theory, the 34 𝜆 region should be
connected to ground in order to allow the flexural wave to transmit through.
Admittance Y
0 dB
–3 dB
fr Frequency (Hz)
Figure 7.9 Bode diagram around its fundamental resonance and anti-resonance frequencies.
+ 280
– 265
+246
+ 258
– 296
– 280
+ 293
+ 297
(a) – 293 (b)
End plane Stator
Case
Rotor Assembly
(c) (d)
Figure 7.10 (a) Diagram of stator and (b) the PZT ring. The numbers indicate d33 measurement
results. (c) the parts and (d) the assembly of the motor.
But in some occasions, to make full use of piezoelectric ceramics, the 34 𝜆 region
would be divided into two sectors and also be poled. In this configuration, the
flexural traveling wave will be produced by exciting both the inputs concurrently.
When an electric field is applied, the positively and negatively poled region
will contract and expand correspondingly. With a 90∘ phase difference between
the two waves, and an unpoled 𝜆/4 region, the flexural waves will travel on
the ceramic surface, and make the teeth-shaped stator vibrate in the same way
driving the rotor to rotate through friction force.
7.3 Surface Acoustics Wave Devices

The SAW device is an electronic component widely used as a frequency filter or
resonator in communication systems such as a mobile phone, optical communi-
cation systems, and so on. It is one of the key devices in modern communication
systems due to its compactness, superior stability, and high performance of fre-
quency characteristics. For a conventional SAW device as shown in Figure 7.11,
a pair of interdigital transducers (IDTs) electrodes are placed on top of a
piezoelectric crystal to generate and detect the SAW signal. The piezoelectric
crystal is periodically poled with the same pattern of the IDTs. The input IDT
applies an electric field to the substrate and launches a SAW by the converse
piezoelectric effect, and the receiver reconverts the acoustic wave into an electri-
cal signal by a direct piezoelectric effect. Acoustic waves propagating along the
Rs
RL
s
Vin
w Surface acoustic wave
Input IDTs Output IDTs
Figure 7.11 Schematic diagram of SAW filters.
surface of a piezoelectric material provide a means of implementing a variety of

signal-processing devices at frequencies ranging from several MHz to a few GHz.
7.3.1 Interdigital Electrode in SAW Device

A SAW device consists of two comb-shaped metal electrodes, called IDT elec-
trode, placed on a piezoelectric material. An electric field created by applying a
voltage to the electrodes induces dynamic strains in the piezoelectric material,
thus creates the elastic waves. When an ac voltage is applied to the electrodes,
the stress wave induced by finger pairs travels along the surface of crystal in both
directions (Gardner et al. 2013). To ensure constructive interference and in-phase
strain, the sum of spacing (s) and width (w) between the adjacent fingers should
be equal to the half of the wavelength (𝜆). i.e.
𝜆
s+w= (7.16)
2
The frequency of the signal, which can be effectively converted to SAW and vice
versa, is given by v = 𝜆f (where v is the phase velocity of SAW). Only this signal
with this particular frequency can be generated with this designed SAW device
with the particular spacing s and w.
7.3.2 Acoustic Wave

The type of acoustic wave generated in a piezoelectric material depends mainly
on the substrate material properties, the crystal cut, and the structure of the
electrodes used to transform the electrical energy into mechanical energy. The
acoustic waves are mainly divided into two types: bulk acoustic wave (BAW) and
surface acoustic wave (SAW). The BAW is an elastic wave propagates in the solid,
while the SAW is a transverse wave propagating along the surface. The acoustic
wave devices usually contain the Rayleigh wave (“true” SAW) and pseudo-SAW
wave propagation, such as leaky-surface acoustic wave (LSAW), surface skim-
ming bulk wave (SSBW), and surface transverse waves (STWs) (Campbell 1998a).
In a conventional SAW device, a piezoelectric substrate is used in the
device fabrication, and the elastic Rayleigh wave is usually used as the
electrical–mechanical–electrical energy conversion through piezoelectric effect.
As shown in Figure 7.12, the Rayleigh wave is a transverse wave that travels
Elliptical displacement of atoms
Figure 7.12 Schematic diagram of Rayleigh wave. Source: Adapted from Gardner et al. (2013).
along the surface and a classical example is the ripples created on the surface of
water by a water droplet. The elastic Rayleigh wave has both a surface-normal
component and a surface-parallel component that is parallel to the direction of
propagation. The energy of the SAW is confined to a zone close to the surface
with a few wavelengths thick.
However, due to the relatively low acoustic velocity in the piezoelectric mate-
rial, the operation frequency is limited in a range of a few hundred MHz. This can-
not satisfy the requirement of high frequency and high bit-rate communication
systems. The increasing demand of high-frequency SAW devices accelerates the
development of SAW devices in the GHz range. The high-frequency SAW devices
may be fabricated by either implementing high-resolution photolithography or by
using high acoustic velocity materials as the substrate with a piezoelectric layer
on top (Nakahata et al. 2003), since the operating frequency is proportional to the
acoustic velocity and inversely proportional to the sum of line width and space
of the IDT (see Figure 7.11). However, the use of high-resolution photolithogra-
phy results in dramatical increase in cost and poorer reliability. Therefore, using
a substrate with high acoustic velocity is preferred, and layer-based SAW devices
have been introduced to achieve this goal.
Most substrates possessing high acoustic velocity are not piezoelectric and
thus must be combined with a piezoelectric layer such as zinc oxide (ZnO), alu-
minum nitride (AlN), lead zirconate titanate (PZT), lithium niobate (LiNbO3 ),
and lithium tantalate (LiTaO3 ) for the generation of the SAW. Consequently, in
the fabrication of high-frequency SAW devices, many combinations of materials
are currently under investigation as potential candidates, such as ZnO/diamond,
LiNbO3 /sapphire, LiNbO3 /diamond, and LiTaO3 /sapphire (Nakahata et al. 2003;
Uchino 2009; Gardner et al. 2013; Campbell 1998a).
In the selection of materials, a systematic study of the required proper-
ties of piezoelectric materials and substrates is needed. Some important
parameters for selecting SAW device structures are (i) acoustic velocity,
(ii) electromechanical coupling coefficient, (iii) film morphology, and (iv)
structural and interfacial qualities. However, none of the material combi-
nations can satisfy all these requirements. The lattice mismatch, thermal
expansion coefficient, and resistance to oxidation of the materials would
directly affect the structure of films and finally adversely affect the per-
formance of the devices. A careful investigation and selection of materials
are essential in the device fabrication. Tables 7.2 and 7.3 compare the rel-
evant properties for some piezoelectric materials and substrates achieved
Table 7.2 Comparison of some piezoelectric materials properties.
Shear wave Electromechanical

acoustic velocity coupling Curie
Materials (ms−1 ) coefficient (%) temperature (∘ C)
LiNbO3 4000 5.5 1210

LiTaO3 3300 0.8 603
AlN 6700 0.7 /
ZnO 2600 1.9 /
PZT 2400 2.3 370
Table 7.3 Comparison of shear acoustic

velocity of substrate materials.
Shear wave
acoustic velocity
Materials (ms−1 )
Diamond 12 000
Sapphire 6 000
Quartz 3 500
SiC 6 830
from SAW devices, respectively (Nakahata et al. 2003; Gardner et al. 2013;
Campbell 1998b).
7.3.3 Piezoelectric Property Considerations for SAW Devices

Started in the 1970s, SAW devices have become attractive as components
incorporated in electronic systems (Morgan 1973). At that time the substrate
(bulk) materials used to fabricate SAW components are limited. Due to the
relatively large electromechanical coupling coefficient and high acoustic velocity,
LiNbO3 has been widely investigated for the SAW device fabrication. The y-cut
z-propagating LiNbO3 is one of the most common materials and orientations
used in fabricating SAW transducers (Reilly et al. 1973). This orientation has
advantages of having strong electromechanical coupling, low attenuation,
and low scattering of waves. The disadvantage is that bulk wave can also be
generated, and those bulk waves traveling close to the surface can excite the
output transducer and produce unwanted output signals (Wong 2002). In
general, the choice of cut or orientation of the substrate materials depends on
the application.
LiNbO3 is one of the most important materials for both fundamental
and applied research in optics and materials science, owing to its electri-
cal, electro-optical, nonlinear optical, and crystallographic properties. As a
ferroelectric crystal, LiNbO3 presents a spontaneous polarization at tempera-
tures below the Curie temperature T c (1210 ∘ C). In the microscopic viewpoint,
lithium niobate consists of a grid of oxygen ion (O2− ) layers in which the lithium
ions (Li+ ) and niobium ions (Nb5+ ) are positioned, as shown in Figure 7.13a.
The displacement of the lithium ions and niobium ions relative to the neutral
configuration (paraelectric phase) gives rise to a dipole moment and thus to the
spontaneous polarization along the c-axis. In the ferroelectric phase as shown
in Figure 7.13b, lithium ions and niobium ions are displaced above or below the
oxygen ion layers. The dipole moment of LiNbO3 is pointing upward so that
the direction of spontaneous polarization is upward. Figure 7.13c shows the
schematic diagram of the crystal structure of lithium niobate in paraelectric
phase where the niobium ions are placed at the center of the oxygen ions
octahedron and the lithium ions are distributed within the next two octahedra.
One possible approach to improve the device performance is to widen the
choice of substrate materials by using piezoelectric films with IDTs to generate
and detect surface waves on non-piezoelectric substrates. LiNbO3 has been
deposited on different substrates or combined with another piezoelectric layer to
enhance the performance of SAW devices. Film and substrate combinations can
be chosen to control the coupling efficiency, to achieve low propagation loss, to
regulate temperature coefficient of delay, and to increase fundamental frequency
and reduce spurious signals. Thus, not many combinations are available. LiNbO3
has also been used as a substrate material incorporating with AlN or ZnO
piezoelectric layer in SAW device application (Wu et al. 2002; Kao et al. 2004;
Kao et al. 2003).
For the LiNbO3 film deposition, it is usually done on the sapphire substrate.
Since these materials are in hexagonal structure and their lattice mismatch are not
so large, high quality films can be obtained by deposition. Shibata et al. claimed
that their grown LiNbO3 films have enough piezoelectricity for SAW device fab-
rication, and it was also shown that different directions of propagation affect
the signal transmission and different modes of acoustic wave transmission were
Li+
Nb5+
P P
O2–
a T < TC T > TC
(a) (b) (c)
Figure 7.13 (a) Stereoscopic view of the ideal crystal stacking of LiNbO3 along the
crystallographic c axis, (b) side view of ferroelectric phase LiNbO3 , and (c) side view of
paraelectric phase LiNbO3 . Source: Kang et al. (2016). Adapted with permission of Royal
Society of Chemistry.
obtained (Shibata et al. 1995). The center frequency of these devices is much
higher than that of the conventional type of SAW devices with similar size of IDT.
7.3.4 Characterization of SAW Devices

The network analyzer is usually used to characterize a SAW device, where the
characteristic admittance (Y ) or impedance (Z) and scattering parameters (S11 ,
S21 , S12 , and S22 ) can be measured over a set of discrete frequency data points
(Gardner et al. 2001). The scattering parameters (S-parameters) are the ratio
of reflection and transmission coefficients between the incident and reflected
waves and can be characterized by a magnitude in decibel (dB) and expressed
as Sij , where i, j = 1 or 2. The S-parameters are the important parameters in
the microwave measurement, since they can well describe the performance of
a device over a range of frequency. These parameters can also be expressed in
different formats, such as rectangular or polar coordinate systems and in Smith
chart. Figure 7.14 shows the details of S-parameters. In characterizing a SAW
device, S21 and S11 are plotted as a function of frequency.
As an example, a SAW device based on the LiNbO3 /diamond contains two
aluminum IDTs with 200 couples of fingers for each pair and an aperture length
of 1000 μm. The IDTs are fabricated by depositing 100-nm-thick aluminum
Incident S21 Transmitted

a1 b2
Forward S11
Reflected Z0 load
b1 a2 = 0
S22
Z0 load a1 = 0 Reflected
b2
Reverse
Transmitted S12 Incident
b1 a2
reflected b reflected b
S11 = = 1 S22 = = 2
incident a1 incident a2
a2 = 0 a1 = 0
transmitted b2 transmitted b1
S21 = = S12 = =
incident a1 incident a2
a2 = 0 a1 = 0
b1 = S11 a1 + S12 a2
b2 = S21a1 + S22a2
Figure 7.14 Detail information of S-parameters.

film on the surface of the LiNbO3 film followed by photolithography and wet
etching process. The designed wavelength of the IDTs was 16 μm (both the line
width and space of IDTs are 4 μm). Figure 7.15 shows the schematic diagram of
LiNbO3 /diamond SAW device.
The performance of the devices is shown in Figure 7.16, where one can see that
the IL is 30 dB at a center frequency of 940 MHz. The calculated phase velocity
(v = 𝜆f ) from the center frequency is 15 000 ms−1 , which are well above the shear
wave sound velocity for diamond and lower than that of the longitudinal wave.
AI IDTs
Δ μm
LiNbO3
piezoelectric film
4 μm
Diamond Al2O3 buffer layer
Figure 7.15 Schematic diagram of SAW device structure (LiNbO3 /Al2 O3 /diamond). Inset is the
aluminum IDTs image.
–30
0
–1 –40
S11 (dB)
S21 (dB)
–2
–50
–3
–60
–4
–5 –70
600 800 1000 1200 1400
Frequency (MHz)
Figure 7.16 Scattering parameters S11 response of the SAW device fabricated on LiNbO3 films
deposited diamond substrate. Source: Adapted from Lam et al. (2004).
Beyond communication application, SAW devices can also be used as sensors

by detecting the SAW that affected by physical stimuli such as weight from sens-
ing elements. This sensing mechanism can be used as very sensitive sensors for
biomaterials and chemical reactions where the weight of biomolecules and reac-
tant can be detected by the wavelength shift.
7.3.5 Lead-Free Piezoelectric Materials

It is well known that, as the most important applications of ferroelectric mate-
rials, almost all commercial piezoelectric actuation and sensing applications are
based on lead zirconate titanate ( PZT) (Jo et al. 2012; Rödel et al. 2009). However,
Pb makes both the manufacturing and disposal of PZT hazardous to people and
the environment (Rödel et al. 2015) in the form of Pb vapor during sintering and
the Pb leaching in soil (Kosec et al. 1998). These issues have led many countries
to regulate and strictly limit the use of Pb in electrical and electronic systems and
devices.
Since the early twenty-first century, there has been significant progress in the
development of lead-free ferroelectrics. Up to date, under certain conditions,
the electromechanical properties of some compositions have shown successful
match or even surpassed commercially available PZT over a wide temperature
range, making them potentially attractive for applications. However, there is still
no single system to be comparable to PZT in satisfying all criteria in application.
The large strains and piezoelectric responses from Pb-free ferroelectrics could be
markedly different to classical ferroelectrics such as PbTiO3 and BaTiO3 .
Lead-free ferroelectrics that have potential for device application can be
broadly classified into two primary groups: (i) ferroelectric and (ii) relaxor ferro-
electric. The first group (i) is primarily based on (Kx ,Na1 − x )NbO3 (KNN) (Saito
et al. 2004), and the second group (ii) is primarily based on (Na1/2 ,Bi1/2 )TiO3
(NBT), a pseudo-rhombohedral ferroelectric. Solid solutions containing MPBs
such as KNN-BT and NBT-BT-KNN show promise, particularly when coupled
Dice the plate

NBT-BT single crystal for the first time
Epoxy
Dice direction
Fill the kerfs with epoxy
for the first time
Fill the kerfs for Dice the plate for the
the second time second time
Dice the plate

for the third time
Fill the kerfs for Ground to remove the top
the third time and bottom part
Figure 7.17 Schematic procedures of the modified dice-and-fill method used for fabricating
1–3 composites based on very fragile piezoelectric materials. Zhou et al. (2012). Reprinted with
permission of Elsevier.
with substitutional doping. Ultimately, more complex solid solutions and doping
schemes will have to be explored because these provide an increasing number
of degrees of freedom for identifying extraordinary properties (Aksel and Jones
2010).
Our group has demonstrated successful application of Pb-free ferroelectric
materials in medical ultrasound transducers. Figure 7.17 shows the diagram of
fabricating NBT-BT/epoxy 1–3 composite for transducer fabrication. This 1–3
composite structure can effectively increase the piezoelectric strain due to the 33
vibration mode of each NBT-BT single crystal element and can also reduce the
acoustic impedance due the low acoustic impedance of epoxy matrix. With such
composite, we have fabricated single-element and array ultrasound transducers,
and the transducers show great performance in both sensitivity and bandwidth
as shown in Figures 7.18 and 7.19.
Matching
layer
NBT-BT / epoxy
1–3 composite
Conductive
adhesive
Backing
material
Coaxial
cable
Figure 7.18 Construction of the NBT–BT/epoxy 1–3 composite single-element and linear array
transducers. Source: Zhou et al. (2012). Reprinted with permission of Elsevier.
0.6
0
0.4
Magnitude (dB)
0.2 BW (–6dB) = 104%

Amplitude (V)
–10
0.0
–20
–0.2
–0.4
–30
–0.6
9 10 11 12 0 1 2 3 4 5 6 7
Time (µs) Frequency (MHz)
(a) (b)
Figure 7.19 (a) Pulse-echo waveform and (b) frequency spectrum of a single array element of
the fabricated NBT–BT/epoxy 1–3 composite linear array ultrasonic transducer. Source: Zhou
et al. (2012). Reprinted with permission of Elsevier.
References 163
References
Aksel, E. and Jones, J.L. (2010). Advances in lead-free piezoelectric materials for
sensors and actuators. Sensors 10 (3): 1935–1954.
Campbell, C.K. (1998a). Basics of piezoelectricity and acoustic waves. In: Surface
Acoustic Wave Devices for Mobile and Wireless Communications, 19–62. San
Diego: Academic Press.
Campbell, C.K. (1998b). Principles of linear-phase SAW filter design. In: Surface
Acoustic Wave Devices for Mobile and Wireless Communications, 67–96. San
Diego: Academic Press.
Desilets, C.S., Fraser, J.D., and Kino, G.S. (1978). The design of efficient broad-band
piezoelectric transducers. IEEE Transactions on Sonics and Ultrasonics 25 (3):
115–125.
El, G.N. and Helbo, J. (2000). Equivalent circuit modeling of a rotary piezoelectric
motor. IASTED International Conference on Modelling and Simulation
(MS’2000), Pittsburgh, Pennsylvania, USA (May 15–17, 2000).
Gardner, J.W., Varadan, V.K., and Awadelkarim, O.O. (2001). IDT microsensor
parameter measurement. In: Microsensors, MEMS, and Smart Devices, 337–339.
Available at: https://doi.org/10.1002/9780470846087.ch11.
Gardner, J.W., Varadan, V.K., and Awadelkarim, O.O. (2013). Introduction to SAW
devices. In: Microsensors, MEMS, and Smart Devices, 303–317. Wiley-Blackwell.
Available at: https://onlinelibrary.wiley.com/doi/abs/10.1002/9780470846087
.ch9.
Jo, W., Dittmer, R., Acosta, M. et al. (2012). Giant electric-field-induced strains in
lead-free ceramics for actuator applications – status and perspective. Journal of
Electroceramics 29 (1): 71–93. Available at: https://doi.org/10.1007/s10832-012-
9742-3.
Kang, X., Sang, Y., Liu, H. et al. (2016). Formation mechanism and elimination
methods for anti-site defects in LiNbO3 /LiTaO3 crystals. CrystEngComm 18 (42):
8136–8146. Available at: http://dx.doi.org/10.1039/C6CE01306F.
Kao, K.S., Cheng, C.C., Chen, Y.C., and Lee, Y.H. (2003). The characteristics of
surface acoustic waves on AlN/LiNbO3 substrates. Applied Physics A 76 (7):
1125–1127. Available at: https://doi.org/10.1007/s00339-002-2022-3.
Kao, K.S., Cheng, C.C., Chen, Y.C., and Chen, C.H. (2004). The dispersion
properties of surface acoustic wave devices on AlN/LiNbO3 film/substrate
structure. Applied Surface Science 230 (1–4): 334–339. Available at: http://www
.sciencedirect.com/science/article/pii/S0169433204001850.
Kirk Shung, K. (2015). Diagnostic Ultrasound: Imaging and Blood Flow
Measurements, 2e. Boca Raton, London, New York: CRC Press.
Kosec, M., Malic, B., Wolny, W. et al. (1998). Effect of chemically aggressive
environment on the electromechanical behaviour of modified lead titanate
ceramics. Journal of the Korean Physical Society 32: S1163–S1166.
Krimholtz, R., Leedom, D.A., and Matthaei, G.L. (1970). New equivalent circuits for
elementary piezoelectric transducers. Electronics Letters 6 (13): 398–399.
Lam, H.K., Dai, J.Y., and Chan, H.L.-W. (2004). Highly-oriented LiNbO3 films on
polycrystalline diamond substrate for high frequency surface acoustic wave
devices. Japanese Journal of Applied Physics 43 (6A): L706–L708. Available at:
http://dx.doi.org/10.1143/JJAP.43.L706.
Morgan, D.R. (1973). Surface acoustic wave devices and applications. Ultrasonics 11
(3): 121. Available at: http://www.sciencedirect.com/science/article/pii/
0041624X73906082.
Nakahata, H., Fujii, S., Higaki, K. et al. (2003). Diamond-based surface acoustic wave
devices. Semiconductor Science and Technology 18 (3): S96–S104. Available at:
http://stacks.iop.org/0268-1242/18/i=3/a=314.
Reilly, N.H.C., Milsom, R.F., and Redwood, M. (1973). Generation of Rayleigh and
bulk waves by interdigital transducers on y-cut z-propagating lithium niobate.
Electronics Letters 9 (18): 419–420.
Rödel, J., Jo, W., Seifert, K.T.P. et al. (2009). Perspective on the development of
lead-free piezoceramics. Journal of the American Ceramic Society 92 (6):
1153–1177. Available at: http://dx.doi.org/10.1111/j.1551-2916.2009.03061.x.
Rödel, J., Webber, K.G., Dittmer, R. et al. (2015). Transferring lead-free piezoelectric
ceramics into application. Journal of the European Ceramic Society 35 (6):
1659–1681. Available at: http://www.sciencedirect.com/science/article/pii/
S0955221914006700.
Saito, Y., Takao, H., Tani, T. et al. (2004). Lead-free piezoceramics. Nature 432: 84.
Sashida, T. and Kenjo, T. (1991). An Introduction to Ultrasonic Motors. Clarendon
Press.
Shibata, Y., Kaya, K., Akashi, K. et al. (1995). Epitaxial growth and surface acoustic
wave properties of lithium niobate films grown by pulsed laser deposition.
Journal of Applied Physics 77 (4): 1498–1503. Available at: https://doi.org/10
.1063/1.358900.
Wong, K.-K. (2002). Acoustic wave propagation and properties. In: Properties of
Lithium Niobate. London: INSPEC, The Institution of Electrical Engineers.
Wu, S., Chen, Y.-C., and Chang, Y.S. (2002). Characterization of AlN films on Y-128∘
LiNbO3 by surface acoustic wave measurement. Japanese Journal of Applied
Physics 41 (Part 1, No. 7A): 4605–4608. Available at: http://dx.doi.org/10.1143/
JJAP.41.4605.
Zagzebski, J.A. (1996). Essentials of Ultrasound Physics. USA: Mosby: Maryland
Heights. Available at: https://books.google.com.hk/books?id=XahsQgAACAAJ.
Zhou, Q., Lam, K.H., Dai, J.Y. et al. (2011). Piezoelectric films for high frequency
ultrasonic transducers in biomedical applications. Progress in Materials Science
56 (2): 139–174. Available at: http://www.sciencedirect.com/science/article/pii/
S0079642510000563.
Zhou, D., Lau, S., Wu, D., and Kirk Shung, K. (2012). Lead-free piezoelectric single
crystal based 1–3 composites for ultrasonic transducer applications. Sensors and
Actuators A: Physical 182: 95–100. Available at: http://www.sciencedirect.com/
science/article/pii/S0924424712003299.
Zhu, J. (2008). Optimization of Matching Layer Design for Medical Ultrasonic
Transducer. The Pennsylvania State University.
Zhu, B.P., Wu, D.W., Zhou, Q.F. et al. (2008). Lead zirconate titanate thick film with
enhanced electrical properties for high frequency transducer applications.
Applied Physics Letters 93: 012905. Available at: https://doi.org/10.1063/1
.2956408.
165
Ferromagnetics: From Material to Device
Ferromagnetic materials exhibit spontaneous magnetization (magnetic moment

per unit volume) at temperatures below the Curie point T c . Magnetic moments
are originated from the orbital and spin of electrons, as shown in Figure 8.1a,b.
Same as ferroelectrics, ferromagnetic materials also exhibit hysteresis loop in the
curve of magnetization–magnetic field as shown in Figure 8.1c. A ferromagnetic
material also tends to form domains when the grain size is larger than a critical
value that is normally a few nanometers.
Since the first discovery of Fe3 O4 (the so-called lodestones) more than
3000 years ago, many ferromagnetic materials have been found and applied
in different fields from magnetic compass to electrical power generation and
telecommunications. In recent decades, magnetic materials have been very
successfully used in data storage from magnetic tape to hard disc. In this book,
we focus on device applications of magnetic materials in smart systems such as
transducers utilizing magnetostrictive effect and thin film-based ferromagnetic
sensors and memory devices.
8.1 General Introduction to Ferromagnetics

The origin of ferromagnetism (FM) is time reversal symmetry breaking due to the
existence of unpaired spin of electrons and the associated current. This statement
can be understood by considering that if time is reversed, the spin or current
flow direction will thus be reversed and this results in magnetization direction
reversion, i.e. time reversal symmetry breaking. Figure 8.2 illustrates time reversal
symmetry breaking.
Any atoms or ions with existing unpaired electrons exhibit ferromag-
netism such as Fe, Co, and Ni. Magnetization can exist, even very weak, in
non-ferromagnetic materials such as graphene as long as their defects or edge
structures induce dangling bonds with unpaired electrons (Ma et al. 2012; Liu
et al. 2013). However, those ferromagnetic materials with practical application
potentials are usually compounds containing Fe, Co, or Ni.
In terms of magnetization, materials can be classified into paramagnetic,
diamagnetic, ferrimagnetic, ferromagnetic, antiferromagnetic (AFM), and super-
paramagnetism. A ferromagnetic material usually forms permanent magnet or

166 8 Ferromagnetics: From Material to Device
Magnetic Magnetic Ms
moment m moment m Mr
Electron
–Hc +Hc H
Atomic Direction
nucleus of spin
(a) (b) (c)
Figure 8.1 Diagram showing the magnetic moment associated with (a) orbital motion and (b)
spin motion of an electron; (c) ferromagnetic M–H loop where Ms and Mr are the saturation
and remnant magnetization and Hc is the coercive magnetic field.
Figure 8.2 Illustration of time reversion symmetry

breaking. Source: Adapted from Eerenstein et al.
m (2006).
m m
Time reversal Spatial inversion

m switched m invariant
is attracted to magnets, and it undergoes phase change to paramagnetic above

Curie temperature through a second-order phase transition. Ferromagnetism
(including ferrimagnetism) is the strongest type; it is the only type that creates
forces strong enough to be felt and is responsible for the common phenomena of
magnetism encountered in everyday life. An AFM has magnetization microscop-
ically but has no macroscopic magnetization due to the cancelation of antiparallel
spins of alternative layers. The temperature from an AFM to paramagnetic phase
transition is called Néel temperature. For example, BiFeO3 is an AFM material
with Néel temperature of 635 K (Kiselev et al. 1963). Ferrimagnetic is a relatively
weak ferromagnetic, such as CoFe2 O4 spinel. A diamagnetic generates weak but
negative magnetization when a magnetic field is applied.
At this point, you may have already seen analogies of ferromagnetics to ferro-
electrics by comparing these terms: ferromagnetic to ferroelectric, paramagnetic
to paraelectric, and AFM to antiferroelectric. The fact is that the theory of ferro-
magnetics was established much earlier than that of ferroelectrics, and scientists
in field of ferroelectrics translated the ferromagnetic theory to ferroelectrics. If
you treat a spin electron as a magneton and a polarized unite cell as polariton,
8.1 General Introduction to Ferromagnetics 167
Table 8.1 Fundamental properties showing analogies between ferromagnetism and

ferroelectricity.
Ferromagnetism Ferroelectricity
Magnetic field H Electric field E

Magnetic induction B (or magnetic flux density) Electric displacement D
B = 𝜇0 H + 𝜇0 M D = 𝜖0 E + P
Vacuum permeability 𝜇0 Vacuum permittivity 𝜖 0
Magnetization M Polarization P
M = 𝜒 mH P = 𝜖0 𝜒 e E
Magnetic susceptibility 𝜒 m Electric susceptibility 𝜒 e
B = 𝜇H D = 𝜖 0 (1 + 𝜒 e )E = 𝜖E
Permeability 𝜇 = 𝜇0 𝜇r Permittivity 𝜖 = 𝜖 0 𝜖 r
Relative permeability 𝜇r = 1 + 𝜒 m Relative permittivity 𝜖 r = 1 + 𝜒 e
they similarly have properties of being switchable and vanished above Curie tem-
perature. There are phenomenological similarities between ferromagnetic and
ferroelectric, where ferromagnetic is due to time-reversal symmetry breaking and
ferroelectric originates from spatial inversion symmetry breaking. More analo-
gies can be seen from Table 8.1 comparing their fundamental properties.
Electrons have an intrinsic property called spin that contributes to their mag-
netic moment. The classical model of electron spin is the electron spinning on its
axis, i.e. either spin-up or spin-down. Then the value of magnetic moment can be
calculated with respect to the spin of an electron. The total magnetic moment of
an atom is the vector sum of the orbital and spin magnetic moments. A magnetic
solid, made up of atoms with magnetic moment, has quantum exchange interac-
tions that tend to align the magnetic moments at low temperature. When T < T c ,
the macroscopic magnetization arises and retains even in the absence of a mag-
netic field. The magnetic moments tend to align in the same direction without the
aid of an external magnetic field. This is known as the ferromagnetic phase. In the
ferromagnetic category, materials are divided into strong and weak ferromagnets.
By introducing the total particle number N = N ↑ + N ↓ and the spin polarization
s = N ↑ − N ↓ , a strong ferromagnet has almost s = 100% at the Fermi energy as
shown in Figure 8.3a, while a weak ferromagnet gets a smaller spin polarization
and paramagnet has no net spin polarization (Figure 8.3b,c).
Meanwhile, there is another type of magnets that possess magnetization
microscopically, but have no macroscopic magnetization due to the cancelation
of antiparallel spins of neighboring pairs – this is known as AFM phase. A
magnet that exhibits no macroscopic magnetization at high temperature (when
H = 0) is called as the paramagnetic phase, in which the magnetic moment
induced by the applied H-field is rather weak. Figure 8.4 shows the relationship
between magnetization and temperature of a ferromagnet, where one can see
that the magnetization starts decreasing close to T c , and when T > T c , magnetic
E E E
EF EF
DOS DOS DOS DOS DOS DOS

(a) (b) (c)
Figure 8.3 Schematic densities of states (DOSs) for ferromagnetism: (a) strong ferromagnet,
(b) weak ferromagnet, and (c) paramagnet.
Mr
Tc T
Figure 8.4 Illustration of magnetization versus temperature. Arrows inside the boxes denote
the magnetic moments alignment.
moments align randomly resulting in zero macroscopic magnetization. This

behavior is similar to polarization versus temperature characteristics of a
ferroelectric under second-order phase transition.
8.2 Ferromagnetic Phase Transition: Landau

Free-Energy Theory
Landau has given a general description of ferromagnetic to paramagnetic phase
transitions with free energy approach by writing out a series expansion for the
free energy close to Curie temperature. The magnetic free energy is defined
as a power series in the order of magnetization M. By only considering the
symmetries, the series must only contain terms that respect the symmetry of the
8.3 Domain and Domain Wall 169
T > Tc
T = Tc
Energy
T < Tc
Magnetization
Figure 8.5 Sketch of Landau free energy F(M) = 𝛼M2 + 𝛽M4 for T > T c , T = T c , and T < Tc .
order parameter:
F(M) = F0 + 𝛼M2 + 𝛽M4 + … (8.1)
F is minimized with respect to M:
𝜕F
= 2𝛼M + 4𝛽M3 = 0 (8.2)
𝜕M
𝛼
M = 0 or M2 = − (8.3)
2𝛽
In Figure 8.5, the symmetry changes precisely at T = T c and the transition occurs
when 𝛼 changes its sign. When T < T c , there are two energy minima M = ± Ms
that indicate 𝛼 < 0 and 𝛽 > 0. When T > T c , a single energy minimum M = 0
indicates 𝛼 > 0 and 𝛽 > 0. The temperature dependence of 𝛼 can be identified as
( )
T − Tc
𝛼(T) = 𝛼0
Tc
With the presence of an external magnetic field H, a linear term is added as
F(M) = 𝛼M2 + 𝛽M4 + … − 𝜇0 HM (8.4)
As shown in Figure 8.6, when an external magnetic field breaks the symmetry,
the global minimum of F(M) changes from two minima to a single minimum.
Therefore, we can see that even when T > T c , a paramagnetic phase of material,
which has no macroscopic magnetization, can also be attracted with the presence
of an externally applied magnetic field.
8.3 Domain and Domain Wall

Ferromagnetic materials also exhibit domain structure where each domain has
its own magnetization direction that can be switched by external magnetic field.
T > Tc
T = Tc
Energy
T < Tc
Magnetization
Figure 8.6 Sketch of Landau free energy F(M) = 𝛼M2 + 𝛽M4 − 𝜇0 HM for T > T c , T = T c , and
T < T c under a positive magnetic field.
There are easy axes in ferromagnetic materials that are the directions magnetic
moments should follow. For example, for a cubic structured FM material such as
Fe, the ⟨100⟩ directions are usually the easy axes, while for a hexagonal-structured
FM material such as Co, the ⟨0001⟩ are the easy axes. Due to this difference, in a
cubic structure, both 90∘ and 180∘ magnetic domains can be formed (Figure 8.7a),
while in a hexagonal structured ferromagnetic material, the domains are usually
aligned with an angle of 180∘ (Figure 8.7b).
The easy axes of cubic-structured Ni are ⟨111⟩, so it can form 180∘ , 71∘ , and
109∘ magnetic domains; these are possible angles between all ⟨111⟩ easy axes. The
magnetization switching can be realized by not only external magnetic field but
also mechanical stress. This mechanical stress-induced magnetization switching
is similar to ferroelectric material whose polarization can be switched by both
electric field and mechanical stress.
(a) (b)
Figure 8.7 Domain structures in cubic (a) and hexagonal (b) structured ferromagnetic
materials.
Bloch wall Néel wall Ising wall

(a) (b) (c)
Figure 8.8 Domain wall structures of (a) Bloch-type ferromagnetic domain walls, (b) Néel-type
ferromagnetic domain walls, and (c) Ising-type ferroelectric domain walls.
Ferromagnetic domain wall structure is somewhat different from ferroelectric

domains in terms of domain wall thickness. In fact, a ferromagnetic domain wall
is much thicker than ferroelectric domain wall thickness, which is only one to a
few atomic layers. Figure 8.8a illustrates the 180∘ domain wall structure in fer-
romagnetic materials, where the more common one is the Bloch wall, but in
thinner films a Néel wall is often favored (Figure 8.8b). By contrast, Figure 8.8c
demonstrates a narrow domain wall (Ising type) structure usually as the case of
a ferroelectric domain structure.
8.4 Magnetoresistance Effect and Device

Among many applications of magnetic thin films, magnetoresistance effect has
achieved great success in sensors and memory devices. By changing the value
of its electrical resistance in an externally applied magnetic field, anisotropic
magnetoresistance (AMR), giant magnetoresistance (GMR), tunneling magne-
toresistance (TMR), and colossal magnetoresistance (CMR) can be observed in
magnetic thin films. The discovery of GMR in 1988 provided scientists with a
new perspective for understanding polarized carriers in ferromagnetic metals
and possibilities to apply this new technologies in non-destructive testing
(García-Martín et al. 2011; Poon et al. 2013) and magnetic storage (Daughton
1992).
8.4.1 Anisotropic Magnetoresistance (AMR)

AMR is a property of a ferromagnetic conductor reported in 1857 by British
physicist Lord Kelvin. AMR is a phenomenon in which a dependence of electrical
resistance on the angle between the directions of electrical current and ferro-
magnetic magnetization. In general, the longitudinal resistance reaches maxi-
mum when the current is parallel to magnetization, and the transverse resistance
reaches minimum when the current is perpendicular to magnetization (Nickel
1995). The AMR ratio is defined as
R∥ − R⟂
AMR = (8.5)
R∥
The AMR is originated from the spin–orbit coupling that leads to spin-dependent
scattering of conduction electrons (Inoue 2013). When the magnetization of
H-field direction
H-field
M
direction
M
e–
e–
θ
(a) (b)
Figure 8.9 Schematic diagram of origin of the AMR effect, (a) high and (b) low resistance
states raised by the current parallel and perpendicular to the magnetization, respectively.
Ovals represent the scattering cross-section of the bound electronic orbits.
a material rotates, the electron cloud of each nucleus deforms and it changes
the amount of conduction electron scattering, i.e. the magnetization direction
rotates the closed orbit orientation with respect to the current direction. As
demonstrated in Figure 8.9a, if the field and magnetization are oriented parallel
to the current, the electronic orbits are oriented perpendicular to the current,
so the cross-section for scattering increases, resulting in a high resistance state.
On the contrary, if the field is transverse to the current, the electronic orbits
are in the plane with current; therefore, less scattering occurs due to the small
cross-section, resulting in a low resistance state (Figure 8.9b).
The relation of electrical resistance R on the angle 𝜃 between the directions of
electrical current and magnetization can be obtained by
R = R⟂ + (R∥ − R⟂ )cos2 𝜃 (8.6)
Owing to the angular dependence, AMR effect is popularly used in
magneto-resistor and sensors. For example, electronic compass for the
measurement of Earth’s magnetic field, traffic detection, and linear position
and angle sensing. A typical AMR magnetoresistive sensor is constructed of a
combination of magneto-resistors, which can be used to measure strength and
direction of magnetic field.
Besides sensors, AMR effect was used as the basis of reading in magnetore-
sistive random-access memory (MRAM). However, its amplitude in ferromag-
netic materials is weak (about 2% variations in resistance on changing the rela-
tive orientation of magnetization and current). The application in reading head
using AMR effect was therefore replaced after the GMR was discovered. How-
ever, AMR effect has attracted great interest again because of investigation on
AFM materials (Kriegner et al. 2016; Lu et al. 2018; Hu et al. 2012). It represents
a major step in the field of AFM spintronics in which the antiferromagnet (with
a zero net moment) governs the transport instead of just playing a passive sup-
porting role in the traditional ferromagnetic (FM) spintronics.
8.4.2 Giant Magnetoresistance (GMR)

GMR is a quantum mechanical magnetoresistance effect observed in multilayers
composed of the super lattice of alternating ferromagnetic and non-magnetic
conductive layers (Baibich et al. 1988). Unlike AMR, GMR requires nano-
structured composites and each layer is only a few nanometers thick. The
magnetoresistance was reported as the largest so far observed for magnetic
metal films (about 40% for Fe/Cr multilayers), so it was called as GMR. The 2007
Nobel Prize in Physics was awarded to Albert Fert and Peter Grünberg for the
discovery of GMR.
Let us look at the details of the work being awarded 2007 Nobel Prize in Physics.
As reported in 1987–1988 (Grünberg et al. 1987; Baibich et al. 1988), they grew
Cr/Fe superlattice structure and measured the in-plane resistance in the presence
of magnetic field. Due to the exchange coupling between the FM Fe layer and Cr
layer, this superlattice structure can have parallel and antiparallel magnetization
layers forming a GMR effect as illustrated in Figure 8.10. One can see that the
maximum magnetoresistance reaches 80% from the superlattice structure with
Cr layer thickness of 0.5 nm. This GMR effect has been successfully implemented
in hard disc drive date storage.
Not only ferromagnetic metals, GMR was also found to occur in some
oxide thin films of manganates of the formula La1−x Ax MnO3 (A = alkaline
earth metal). One of the well-known examples is La1−x Srx MnO3 . The parent
manganate LaMnO3 , AFM insulator, is an orthorhombic perovskite with dis-
torted MnO6 octahedra. When La is replaced by a divalent ion, Mn4+ ions are
created, and when Mn4+ is sufficiently large, the material changes from AFM
to ferromagnetic, exhibiting an insulator-to-metal transition because of double
exchange mechanism. The magnetization and electrical resistivity behavior of
R/R(H = 0)
1.0
0
Fe(3 nm)/Cr(1.8 nm) 0.9

.9
0.8
0 .8 ~+80%
8
3 nm)/Cr(1.2 nm)
Fe(3 0.7
0.6
(0.9 nm)
Fe(3 nm)/Cr(0.9
0.5
–40 –20 0 20 40
H (kG)
Figure 8.10 Magnetoresistance of Fe/Cr superlattice showing the change in the resistance of
Fe/Cr superlattice at 4.2 K in external magnetic field H. The arrow indicates maximum
resistance change. Source: Adapted from Baibich et al (1988).
La1−x Ax MnO3 offers the oxide thin films an opportunity for ferromagnetism and
GMR effect applications (Rao and Mahesh 1997).
The nature of the GMR effect is different from that of AMR – that the
resistance in the plane of the film is isotropic for the GMR effect but depends
on current direction for the AMR. The GMR effect shows a magnetization
orientation-dependent component of resistance that varies between magnetiza-
tions in alternate layers. Figure 8.11a shows the schematic superlattice structure
of ferromagnetic Fe layers alternatively with Cr layers reported by Baibich et al.
In this structure, layers of Fe, sandwiching an ultrathin Cr layer, was found to be
ferromagnetically ordered with a magnetic moment oriental antiparallel to that
Fe
Cr
Fe
Cr
Fe
Cr
Fe
GaAs
(a) External H-field direction
When H ≠ 0, When H = 0,
Fe Fe
Cr Cr
Fe Fe
Rparallel Rantiparallel
(b) (c)
Figure 8.11 (a) Diagram showing the superlattice structure of ferromagnetic Fe layers
alternatively with Cr layers, white arrows denote magnetization direction, and (b) and (c) are
the corresponding low and high resistance states for parallel and antiparallel magnetizations
among neighboring Fe layers.
GMR sensor
AC generator
Figure 8.12 Hybrid probe: eddy current testing (ECT) coil with GMR sensor. Source: Adapted
from Čápová et al. (2008).
of the adjacent Fe layer. Without applied magnetic field, the coupling of Fe films
across Cr interlayers with proper thickness is found to be AFM (Saurenbach
et al. 1988; Grünberg et al. 1987), and it brings the magnetization to zero. GMR
effect is thus achieved in antiferromagnetically coupled Fe/Cr superlattice by
aligning the magnetization of adjacent Fe layers with an external field.
By well controlling the AFM coupling, the ferromagnetic layers can form a state
with antiparallel magnetization. Electrons that transport through this structure
will be highly scattered due to the averaged spin-up and spin-down currents,
resulting in high resistance state (Figure 8.11c). When a magnetic field is applied
in parallel to the film, AFM magnetization can be switched to the magnetic field
direction, making the whole film’s magnetization direction into the field direction
as shown in Figure 8.11b; electrons transport through this layer experience weak
interface scattering and a high transmission for the spin-down electrons, result-
ing in a relatively much lower resistance. When the thickness of the Cr layer or the
temperature increases, more defect and steps may be introduced, which results in
weakened AFM coupling within the Cr layer; this makes the GMR effect relatively
smaller. This model explains the GMR and explains the thickness-dependence of
GMR shown in the diagram.
A GMR sensor has large bandwidth making it applicable to different frequen-
cies of magnetic field. GMR sensors also have high sensitivity and low power
consumption. Figure 8.12 shows an eddy current probe and a GMR magnetic
field sensor being used in power line current stability monitoring by eddy current
detection. GMR sensors have also been used in smart parking system where the
approaching and leaving of vehicles can be sensed by measuring the MR due to
the Earth’s magnetic field change induced by the vehicles’ movement.
8.4.3 Colossal Magnetoresistance (CMR)

CMR is a phenomenon of dramatically changed electrical resistance in the
presence of a magnetic field. Most CMR materials are manganese-based
perovskite oxides such as Lal−x Ax MnO3+𝛿 (A = Ca, Sr). The magnetoresistance
of conventional materials enables changes in resistance of up to 5%, but CMR

materials may exhibit resistance changes by orders of magnitude.
The term colossal has arisen from the huge effects observed, in the order of
ΔR/R(H) = 127 000%, as the resistivity dropped from 11.6 Ω/cm at H = 0 to
9.1 mΩ/cm at H = 6 T (Jin et al. 1994). Because of very large MR effect, and to
distinguish it from the GMR effect, it is called CMR. The understanding and
application of CMR offers tremendous opportunities for the development of
new technologies such as read/write heads for high-capacity magnetic storage
and spintronics.
8.4.4 Tunneling Magnetoresistance (TMR)

Electron tunneling is a quantum-mechanical effect, where electrons can traverse
the potential barrier that exceeds their kinetic energy. When an electric field is
applied to the magnetic tunnel junction (MTJ, consisting of two ferromagnetic
electrodes and an ultrathin insulating barrier), there is a probability for electron
tunneling. As the tunneling is spin dependent (a disproportion in the number
of electrons parallel and antiparallel to the magnetization of a ferromagnet, usu-
ally referred to as majority- and minority-spin electrons), the imbalance leads
to the measurable difference in the tunneling current carried by majority- and
minority-spin electrons. This phenomenon is known as TMR.
TMR effect is originated from spin-conserving tunneling processes between
electrodes with spin-polarized density of state (DOS) as illustrated in Figure 8.13.
From the simple heterostructure, we can see that an oxide tunnel barrier is sand-
wiched by two dissimilar ferromagnetic materials (a free layer and a pin layer). It
should be noted that the free layer has a smaller H c than that of pin layer, thus
the free layer acts as a soft magnet that its magnetization direction can be easily
tuned and the pin layer acts as a hard magnet. In spin and energy conservation
during tunneling, spin-up or spin-down electrons can only tunnel from the ini-
tial spin-up or spin-down state to an unoccupied spin-up or spin-down final state,
respectively. In other words, the magnitude of TMR including current and resis-
tance depends on the alignment of magnetization. Therefore, the TMR can be
maximized with highly spin-polarized materials such as half-metallic manganites
and Heusler alloys.
To obtain the relative resistance change, TMR can be expressed by using
Julliere’s model (Julliere 1975):
Rap − Rp 2P1 P2
TMR = = (8.7)
Rp 1 − P1 P2
where Rap is the resistance in antiparallel magnetization state, Rp in parallel
magnetization state, and P1 and P2 spin polarizations of the two electrodes.
The following equation describes the spin polarization P calculated from the
spin-dependent DOS at the Fermi energy, where N ↑ is spin-up DOS and N ↓ is
the spin-down DOS (Mazin 1999):
N↑ − N↓
P1 and P2 = (8.8)
N↑ + N↓
Free layer (ferromagnetic)
Tunnel barrier (oxide)

Pin layer (ferromagnetic)
Iread Iwrite Iread Iwrite
E E E E
eU
eU
DOS of pinned layer DOS of free layer DOS of pinned layer DOS of free layer
Figure 8.13 Diagram of spin-dependent TMR effect in parallel (left) and antiparallel (right)
magnetization of a MTJ. The solid arrow represents larger tunneling probability.
8.4.4.1 Spin-Transfer Torque Random-Access Memory (STT-RAM)

In order to fulfill the demand of data storage, a memory based on MRAM com-
bining with spin-transfer torque technology was developed recently, which is
called STT-RAM as shown in Figure 8.14. Similar to a hard disk drive using
materials with magnetic properties, STT-RAM also controls the orientation of
magnetization except that it is operated with electrical current; the writing pro-
cess involves an electrical current aligning the spin direction of electrons flowing
through a MTJ element. The resultant resistance difference of the MTJ element
is then used for information readout.
The memory mainly consists of three layers same as MTJ, i.e. an ultrathin oxide
barrier is sandwiched by two dissimilar ferromagnetic layers, one of them has
a larger coercive field H c so that its magnetization alignment is difficult to be
reoriented (it is called pin layer or hard layer). On the other hand, the free layer
has a smaller H c , so its magnetization is much easier to be reoriented (it is called
soft layer). Usually, the data bit is stored in this so-called soft layer according to
its magnetization orientation.
Bit line
MTJ Bit line

Write word line Bypass line
Cladding
MTJ
Gate Gate
Source Drain Source Drain
Silicon substrate Silicon substrate
(a) (b)
Figure 8.14 Diagram of (a) a conventional MRAM cell and (b) STT-RAM cell memory structures.
Similar to a ferroelectric tunnel junction (FTJ), wave function tunnels through

this ferromagnetic layers sandwiched barrier and the memory state depends on
the tunneling resistance. However, unlike FTJ, this effect is spin-polarized elec-
tron tunneling. As illustrated in Figure 8.14, in spin and energy conservation
during tunneling, spin-up or spin-down electrons can only tunnel from the ini-
tial spin-up or spin-down state to an unoccupied spin-up or spin-down final state,
respectively. It means that the magnitude of TMR depends on the alignment of
magnetization. For TMR, if both magnetization alignments are parallel to each
other, electrons will be able to tunnel, and the memory cell is in the low resistance
state. However, if they are in antiparallel, the magnetoresistance is high.
Most MRAMs that are now being developed write data by applying the
magnetic field generated by a current running through a wire near a TMR
element to change the magnetization. That enables fast operation switching but
consumes more power. In the case of STT technology, a large and directional
(spin-polarized) current changes the magnetic orientation of the free layer when
it is writing. When it is reading, data can be identified by the resultant resistance
difference of MTJ.
This STT-RAM consumes less power because the spin valve or MTJ can be
modified using spin-polarized current. The smaller memory cell that enhances
scalability is also in favor of commercial applications. Starting from 2005, Renesas
Technology Corp. and Hynix Semiconductor Inc. licensed Grandis’ intellectual
property (IP) and began developing STT-RAM for embedded and standalone
applications. In 2010, Grandis has developed a 90 nm, 265 Kbit MTJ devices,
and they demonstrated STT-RAM write current of less than 200 μA, a write/read
speed of 20 ns, and an endurance of 1013 cycles.
8.5 Magnetostrictive Effect and Device Applications

Magnetostrictive effect refers to mechanical strains induced by magnetic field.
This effect results from magnetization switching under applied magnetic field
and only happens below Curie temperature when a magnetic material is in the
ferromagnetic phase. This effect exists in all magnetic materials either smaller or
larger and can be positive and negative. The mechanical strain induced by mag-
netic field is called direct magnetostrictive effect, while the magnetization change
resulted from applied mechanical strain is called inverse magnetostrictive effect.
You may have experience to hear the low-frequency noise “electric hum”
near transformers and high-power electrical devices. That sound comes from
the magnetostrictive vibration of magnetic core material in the transformer
at a frequency of power line that is usually 50 Hz. For device applications, in
order to have large response to a magnetic field making it applicable for sensors
or actuators, the magnetostrictive effect should be significant. Table 8.2 lists
properties of magnetostrictive effect of some typical magnetic materials.
In this table, 𝜀max = 32 𝜆s is the maximum strain that can be generated under a
prestress condition, where 𝜆s is the saturation magnetostriction that quantifies
strains generated from a relaxed ferromagnetic state to the state with all domains
aligned along the magnetic field direction. Ms is the saturation magnetization in
response to an applied magnetic field, which should be large for strong transduc-
tion; and T c is the Curie temperature at which a ferromagnetic to paramagnetic
phase transition occurs. The magnetomechanical coupling coefficient k quanti-
fies the efficiency to convert magnetic energy to mechanical energy, or vice versa,
and elastic modulus Y is its mechanical properties quantifying the strain that can
be generated under mechanical stress. These parameters are important for mag-
netostrictive material-based transducer design.
This table provides general guidelines to select a suitable magnetostrictive (MS)
material for device application. Fe, Ni, Co, and CoFe2 O4 possess negative and
much smaller value of maximum strain and usually small magnetomechanical
coupling factor k. These parameters make them less attractive for device applica-
tion. Tb and Dy have very large magnetostriction, but their T c is too low to make
them meaningful for device application.
The most significant one is Terfenol-D (terbium: Ter, iron: Fe, Naval Ordnance
Laboratory: NOL, dysprosium: D), which can produce strains up to 1600 μl/l at
Table 8.2 Properties of magnetostrictive effect of some ferromagnetic materials.
Material 𝜺max (×10−6 ) Ms (T) T c (∘ C) Y (GPa) k
Fe −14 2.15 770 285 —

Ni −50 0.61 358 210 0.31
Co −93 1.79 1120 210 —
Tb 3000 — −48 55.7 —
Dy 6000 — −184 61.4 —
Terfenol-D 1620 1.0 380 110 0.77
Metglas 2605SC 60 1.65 370 25–200 0.92
CoFe2 O4 −885 0.53 520 142 —
All measurements are at room temperature except where specified

Source: Tsai et al. (2013). Reproduced with permission of AIP Publishing.
room temperature and therefore can greatly expand the role of magnetostrictive
smart material for transducer applications. Metglas is also practically very popu-
lar since it has the largest magnetomechanical coupling factor k, making it more
efficient and having broader bandwidth of response to magnetic field.
8.5.1 Magnetostrictive Properties of Terfenol-D

Terfenol-D crystals are normally grown along [112] orientation, while their mag-
netization easy axes are along ⟨111⟩ directions. As what has been illustrated in
piezoelectric that non-180∘ polarization rotation is essential to generate large
strain, it is the same for MS. A Terfenol-D crystal bar along [112] at room tem-
perature has randomly distributed magnetization domains along 8 equiv. ⟨111⟩
directions. To realize non-180∘ switching of magnetization moment, the crystal
should be prestressed by compression in the bar direction, i.e. [112] orientation.
This will make those ⟨111⟩ moments that are not perpendicular to the [112] direc-
tion to be switched to the [111] and [111] directions that are perpendicular to
[112]. This can be demonstrated in Figure 8.15. You can imagine that by apply-
ing a magnetic field along [112] direction, these [111] and [111] moments tend
to be switched back to their original ⟨111⟩ directions through non-180∘ rotation,
therefore maximum strain along [112] can be generated.
The general term magnetostriction refers to strains generated during the
paramagnetic–ferromagnetic phase transition or in response to an applied field
that constitutes magnetoelastic coupling. Usually there are three parameters of
interest: the spontaneous magnetostriction 𝜆0 , saturation magnetostriction 𝜆s,
and Joule magnetostriction 𝜆 below saturation. The relationship between these
parameters and the total strain 𝜖 is illustrated in Figure 8.16.
It has been calculated that if the total strain from the prestressed state to the
final state is 𝜀, the prestress makes the crystal shrink 1/3 𝜀, which is the same as
[112]
[111]
[001]
[112]
Prestress
[110]
[111] 2a
a
Figure 8.15 Terfenol-D crystal orientation where [112] is the rod direction and ⟨111⟩ are
magnetization easy axes.
(a) When T > Tc’
(b) When T < Tc’ and field is absent, λ0 = 1 ε

3
λs = 2 ε
3
(c) When T < Tc’ and H-field is applied along ↑ direction

or under stress along →← direction,
(d) When T < Tc’ and H-field is applied along → direction
Figure 8.16 Illustration of the relationship among the spontaneous magnetostriction 𝜆0 ,

saturation magnetostriction 𝜆s , and the total strain 𝜀. Spheres represent the isotropic domains
of paramagnetic phase above Curie temperature (a) and the ellipsoids represent the
elongated domain shape at ferromagnetic phase below Curie temperature (b)–(d). (b) Without
magnetic field, (c) magnetic field or mechanical stress forces domains along perpendicular
direction, and (d) magnetic field along parallel direction.
the spontaneous magnetostriction 𝜆0 when the crystal passes from paramagnetic

phase to ferromagnetic phase. Without prestress, the crystal can generate 2/3 𝜀
under magnetic field that is called saturation magnetostriction 𝜆s . Detailed
calculation can be found at “Theory of the magnetomechanical effect” (Jiles 1995).
As introduced in Jiles’s book (Jiles 2015), the spontaneous magnetostriction
of a magnetic material when being cooled down to T c below, if a single domain
with magnetization generate maximum strain 𝜀, the spontaneous strain along any
direction 𝜆0 , can be expressed as
𝜀(𝜃) = 𝜀 cos2 𝜃
This equation is only suitable for the case when an isotropic paramagnetic (repre-
sented by the round circle in the diagram) changes to isotropic FM (represented
by ellipsoids). Without external magnetic field, these domains, each of them is
represented by one ellipsoid, are randomly orientated, and the spontaneous strain
generated can be calculated by

𝜋∕2
𝜀
𝜆0 = 𝜀 cos2 𝜃 sin 𝜃d𝜃 =
∫ −𝜋∕2 3
where one can see that the spontaneous strain is 1/3 of the total strain. When
these domains align their magnetization along external magnetic field, the extra
strain (called saturated strain 𝜆s ), is then equal to 2/3 of the total strain, i.e.
2
𝜆s = 𝜀 − 𝜆 0 = 𝜀
3
Figure 8.16d illustrates the maximum strain that can be achieved from such
isotropic magnetostrictive material through prestress to force all domains
into alignment to the direction perpendicular to the external magnetic field,
subsequently applied to rotate the magnetization moment by 90∘ .
As illustrated in Figure 8.16, it also suggests the practical way to measure the
total strain by measuring the difference of strains when the magnetic field is,
respectively, perpendicular and parallel to the rod, i.e.
3
𝜆s|| − 𝜆s⟂ = 𝜆 =𝜀
2 s
The state 𝜆s⟂ can also be obtained by applying prestress forcing the domains into
a state with magnetic moment along the perpendicular direction as illustrated by
the ellipsoids in Figure 8.16d. One can imagine that when the parallel magnetic
field and the prestress along the rod direction reach a metastable sate, minimum
perturbation of ac magnetic field will cause the magnetic momentum to rotate
from perpendicular to parallel or from parallel to perpendicular. In fact, this is
the way to achieve maximum strain in real application that will be further demon-
strated in next chapter.
Another important relationship can be derived from
𝜀(θ) = 𝜀 cos2 𝜃
By substituting with
M = Ms cos 𝜃
and then,
3
𝜆(𝜃) = 𝜆 cos2 𝜃
2 s
This results in a quadratic relationship between the magnetostriction and the
𝜆
magnetization, i.e. 𝜆(M) = 32 Ms M2 .
s
One should notice that electrostrictive effect is also the quadratic relation.
Compared with P–E loop and 𝜀 − E butterfly loop of ferroelectric materials,
M–H and 𝜀 – H exhibit similar hysteresis and butterfly loops as shown in the
Figure 8.17.
1.0
0.8
0.6
0.4
0.2
M/Ms
–0.2
–0.4
f = 1 Hz
–0.6 f = 500 Hz
f = 1000 Hz
–0.8 f = 2000 Hz
–1.0
–1.5 –1 –0.5 0 0.5 1 1.5
(a) H (A/m) ×105
1800
f = 1 Hz
1600
f = 500 Hz
f = 1000 Hz
Magnetostrictive strain (10–6)
1400
f = 2000 Hz
1200
1000
800
600
400
200
0
–1.5 –1 –0.5 0 0.5 1 1.5
(b) H (A/m) ×105
Figure 8.17 The magnetic (a) hysteresis loops and (b) magnetostrictive strain hysteresis loops
of Terfenol-D under the sinusoidal driving magnetic field of same amplitude and different
frequencies. Source: Xiao et al. (2017). Reprinted with permission of Elsevier.
8.5.2 Magnetostrictive Ultrasonic Transducer

Magnetostrictive materials have very broad application in magnetic field-driven
transducers, such as micromachining, spray nozzle in gas injection, ultra-
sound wave generation by magnetic field, etc. Figure 8.18 shows a diagram of
ultrasonic transducer that can generate ultrasound wave driven by magnetic
field.
Figure 8.18 Schematic diagram

of a magnetic field-driven
ultrasonic transducer utilizing
magnetostrictive effect.
Solenoid with
electrical ac input
Magnetostrictive
laminations
Magnetization
and strain induced
by magnetic field
A magnetostrictive ultrasonic transducer consists of laminations of a FM

material to be ultrasonically activated when an electrical coil produces an ac
magnetic field. A magnetostrictive element extends when current passes through
the solenoid and shrinks when the current is off; this causes the ferromagnetic
laminates to vibrate at their resonant frequency. Although piezoelectric trans-
ducers dominate in the ultrasonic world today, magnetostrictive technology is
still available from a limited number of ultrasonic manufacturers and, in fact, is
the technology of choice of some users in some low frequency applications up to
30 kHz.
8.5.3 Magnetoelastic Effect

The inverse magnetostrictive effect (also known as magnetoelastic effect or
Villari effect) refers to the change of the magnetic susceptibility of a material
when a mechanical stress is applied on it. The magnetostriction characterizes
the shape change of a FM material during magnetization, whereas the inverse
magnetostrictive effect characterizes the change of sample’s magnetization
(for a given magnetizing field strength) when a mechanical stress is applied to
the sample. This inverse magnetostriction can be understood by considering
the strain-induced crystal symmetry change, which causes the change of easy
axis leading to magnetization switching. Therefore, under a given uniaxial
mechanical stress, the flux density for a given magnetization field may increase
or decrease.
Magnetoelastic effect can be used in development of force sensors applied
in engineering, such as civil engineering and biomedical engineering. The
force can be measured by detecting the sample’s magnetization change under a
given magnetic field. Maximum sensitivity can be realized if an appropriate dc

magnetic field and prestress are applied.
Analogous to their ferroelectric counterpart, in ferromagnetic materials,
magnetic field and stress induce strain and magnetization that can be linearly
modeled under low field drive. In engineering application such as transducer
design, the linear approximation in modeling can be reasonably accurate;
however, one should notice that in the regime of high field, either electric
or magnetic, the piezoelectric and ferromagnetic responses are nonlinear,
presenting hysteresis characteristics. The physics behind this nonlinearity is the
domain evolution including domain wall movement.
Analogous to piezoelectric linear relation as shown in Chapter 6, the direct
and inverse magnetostrictive effect can be summarized by the linear constitutive
matrix relations
𝜀 = SH 𝜎 + d∗ H
M = d𝜎 + 𝜒 𝜎 H
where 𝜀 is strain, SH compliance at constant field H, 𝜎 stress, M the magne-
tization, 𝜒 𝜎 magnetic susceptibility at constant stress, and d* the transpose of
piezomagnetic coefficient d.
8.5.3.1 Magnetomechanical Strain Gauge

Apart from Terfenol-D, Metglas is another very common magnetostrictive com-
posite that is the amorphous alloy Fe81 Si3.5 B13.5 C2 with its trade name Metglas
2605SC. Originated from its amorphous structure, Metglas has no anisotropic
properties and there are no crystalline grain boundaries to prevent motion of
magnetic domain walls, so it shows excellent magnetic properties such as high
permeability and low loss while having a high-saturation magnetic flux density.
Though the maximum strain is much smaller, about 60 microstrains, compared
with Terfenol-D, Metglas 2605SC exhibits both magnetostriction and magnetoe-
lastic effect. The very good mechanical elastic properties, a very large magne-
tomechanical coupling factor, and the energy conversion efficiency make Metglas
attractive for magnetic micro-electro-mechanical systems (MEMS) devices.
A strain gauge using Metglas 2605SC was predicted to have 103 to 105 more
sensitivity than a conventional strain gauge (Savage and Spano 1982). Figure 8.19
Force
Metglas ribbon
V
Detection coil
Excitation coil
Figure 8.19 Schematic diagram of force sensor using Metglas core.

shows a diagram of the force sensors using magnetoelastic effect. When a force
is added, the magnetoelastic material deforms, and as a result, the strain can be
detected by measuring the flux density change.
8.6 Characterizations of Ferromagnetism

Magnetization and the corresponding properties can be measured directly or
indirectly by employing different instruments and techniques that can be cate-
gorized into inductive, force, magneto-optic, and electrical methods. In the fol-
lowing sections, some commonly used instruments and techniques for magnetic
property characterizations are briefly introduced.
8.6.1 Vibrating Sample Magnetometer (VSM)

Measuring M–H loop is the most fundamental characterization technique to
prove a material to be FM, with which the saturation magnetization, remnant
magnetization, and coercive field can be determined. The M–H loop can be
obtained by vibrating sample magnetometer (VSM) that uses a technique of
induction to measure magnetic moment of a sample.
The change of magnetization can be determined by measuring the change of
magnetic flux density inside the sample if an external magnetic field is known.
During measurement, a sample is placed at one end of the rod and starts vibrat-
ing along vertical direction at a fixed frequency as illustrated in Figure 8.20a.
In the meantime, an external magnetic field, generated by a pair of electromag-
nets, magnetizes the sample along the horizontal parallel direction. The sample’s
magnetic moment then alters with the strength of the applied magnetic field and
results in a M–H loop for ferromagnetism. The basic principle of manipulation is
to detect the induced voltage in pick-up coil when a magnetic flux of the sample
is changing, and the induced voltage e.m.f . is obtained synchronously using
( )
𝜕𝛷 𝜕
e.m.f . = − = B ⋅ dS
𝜕t 𝜕t ∫
where 𝛷 is the flux linked by the coils and dS is an element of vector area.
Referencing to a well-defined sample, the absolute values of magnetic moments
of sample can be attained.
VSM is also applicable to low-temperature characterization by combining with
cryostat equipment, for example, magnetic moment can be detected from 77 to
350 K by using liquid nitrogen in a VSM system. For further cooling, 2 K can also
be achieved if liquid helium is used. Figure 8.20b,c shows the magnetic moment
of La0.7 Sr0.3 MnO3 and Ni0.81 Fe0.19 thin films in response to the external magnetic
field. Except for FM classification, VSM can also be used to observe the strength
of magnetization in in-plane and out-of-plane directions. It is apparent that both
thin films have a stronger magnetization signal parallel to H-field than the per-
pendicular one. It describes that the magnetic dipoles are pointing parallel to the
film instead of pointing perpendicular to the film. In the meantime, an extra high
magnitude of field is required to switch the moment within the same sample for
Sample rod with oscillation
Sample
Pick-up coils
Magnetic field Electromagnets

(a)
150
400 LSMO
LSMO
NiFe 100 NiFe
Moment (μemu)
Moment (μemu)
200
50
0 0
–50
–200
–100
–400
–150
–200 0 200 –200 0 200
Magnetic field (Oe) Magnetic field (Oe)
(b) (c)
Figure 8.20 (a) Schematic diagram of VSM system and M–H loops of two magnetic materials
LSMO and NiFe in in-plane (b) and (c) out-of-plane magnetization directions at 78 K.
the out-of-plane measurement. Based on the results, VSM can also be used for
specifying the easy axis of magnetization for materials. Using inductive method,
for instance, VSM, FM properties including the direction of easy axis, satura-
tion magnetization, and coercivity can be obtained by analyzing M–H loops of
materials. Other common inductive methods include extraction magnetometry,
AC susceptometry, and superconducting quantum interference device (SQUID)
magnetometry.
8.6.2 Superconducting Quantum Interference Device (SQUID)

Although SQUID and VSM are both inductive methods, VSM is not suitable
for materials that are magnetically diluted or that possess very small sample
volumes, i.e. nanoscale magnet (Spinu et al. 2013). The sensitivity of VSM may
be insufficient to properly characterize such materials, while SQUID is a more
sensitive magnetometer used to measure extremely subtle magnetic fields,
based on superconducting loops containing Josephson junctions that is a tunnel
junction between two superconducting layers sandwiched with an ultrathin
insulating layer.
Josephson junctions
Ia
Is
I I
Φ
Ib Is
Superconducting loop
Figure 8.21 Diagram of a DC SQUID.
Quantum effect in conjunction with superconducting detection coil circuitry

are used in SQUID-based magnetometers to measure the magnetic properties
of materials. In a DC SQUID, it has two Josephson junctions in parallel in a
superconducting loop as shown in Figure 8.21. An input biasing current I enters
the superconducting loop and splits into path I a and path I b equally. If an external
magnetic field is applied to the loop, a current I s begins circulating within the
loop that generates an opposite magnetic field against the external flux 𝛷. In
the two branches, the total current in each branch is changed because of the
induced current, i.e. one is I/2 + I s and the another one is I/2 − I s . When the total
current in either branch exceeds the critical current I c of the Josephson junction,
a voltage appears across the junction. The voltage is therefore a function of the
applied magnetic field and the period of the magnetic flux quantum 𝛷0 .
R
ΔV = R ⋅ ΔI = ⋅ Δ𝛷
L
Similar to flux to voltage converter, Δ𝛷 can be estimated as the function of ΔV
using this equation, where R is a shunt resistance connected across the junction
to eliminate the hysteresis and L is the self-inductance of the superconducting
ring. Theoretically, SQUIDs are capable of achieving sensitivities of 10−12 emu,
but practically they are limited to 10−8 emu, because SQUIDs also pick up envi-
ronmental noise similar to VSM. SQUIDs may also be used to perform measure-
ments from low to high temperatures; so temperature-dependent magnetization
and other magnetic properties can also be characterized by SQUIDs.
8.6.3 Magnetic Force Microscopy (MFM)

Besides the macroscopic magnetization determination, ferromagnetic domains
can be observed. Again, similar to the manipulation of piezoelectric materials
using piezoresponse force microscopy (PFM), the magnetic structure and kinds
of magnetic interactions of sample surface can be reconstructed by magnetic
force microscopy (MFM).
Tip with ferromagnet coating

30
25
20
Path of cantilever
μm
15
10
0
Magnetic sample with domains 0 5 10 15 20 25 30
(a) (b) μm
Figure 8.22 (a) Schematic diagram of MFM mechanism and (b) MFM image of a magnetic
recording disc displaying a multi-scale domain structure.
As illustrated in Figure 8.22a, In MFM measurement, when a sharp tip coated

with a ferromagnetic thin film scans a magnetic sample, local oscillatory defor-
mation as the tip-sample force will be induced on the sample surface due to the
magnetic interactions. The frequency and phase change of oscillatory deforma-
tion response possess a kind of relationship with the strength of the magnetic
field gradient, and this phase angle together with amplitude can be measured
by lock-in technique and imaged to reveal the ferromagnetic magnetization and
domain structure.
The magnetic AFM probe detects the surface oscillation and sends the signal
through the reflected laser beam to photodiode detector and be processed by a
lock-in amplifier. The amplitude measured by the lock-in amplifier represents the
strength of magnetization (if well calibrated), and the phase angle represents the
magnetic moment direction; under ideal conditions, 0∘ and 180∘ phase angles
represent the up and down magnetization directions, respectively. By forming an
image with phase angle, the MFM technique can image ferromagnetic domains
by presenting the bright-dark contrast. With this method, magnetic characteris-
tics of such domains and magnetic defects can be revealed on a variety of mate-
rials, especially for their surface properties. Figure 8.22b is a MFM image of a
magnetic recording disc showing magnetic domains on magnetic film represent-
ing memory states of 0 and 1.
8.6.4 Magneto-Optical Kerr Effect (MOKE)

Magneto-optical Kerr effect (MOKE) is originated from the optical anisotropy
of the materials, and it describes the change to light reflected by a magnetized
surface. MOKE imaging is simply based on the rotation of the polarization
plane of linearly polarized light when reflected from the surface of a magnetic
material. By integrating domain contrast relating to the magnitude and direction
of the magnetization in the materials, their electronic and magnetic properties
such as the magnetic domain structure, spin DOSs, and even magnetic phase
transition dynamics can be revealed. MOKE technique has been widely used for
Polarized Reflected light

incident light
Magnetic sample
with domains
(c)
(b)
(a)
Figure 8.23 Schematic illustration showing the optical experiment for observing the (a) polar,
(b) longitudinal, and (c) transverse Kerr effects.
recent progress on high-quality nanostructures and 2D materials (e.g. graphene,

topological insulators, and transition metal dichalcogenides) for enhanced
control of light at nanoscale for integrated photonic and spintronic devices.
In MOKE effect, both polarization and reflected intensity of the reflected light
from a magnetized surface can be changed. As shown in Figure 8.23, when a
polarized light incidents to the magnetic sample and reflects, due to the different
magnetization orientations with respect to the reflecting surface and the plane of
incidence, MOKE can be divided into three types. First, in the polar Kerr effect
(Figure 8.23a), the magnetization is perpendicular to the sample plane and is par-
allel to the plane of incidence. In this case, the MOKE detection is out-of-plane
that the Kerr effect generally alters the polarization of the incident light from
plane to elliptically polarized with the major axis rotated. Nearly normal inci-
dence is usually employed when doing experiments in the polar geometry. Sec-
ond, in the longitudinal Kerr effect as shown in Figure 8.23b, incident light with
angle (typically 60∘ (Celotta et al. 2002)) is employed. The in-plane magnetiza-
tion vector is parallel to both the reflection surface and the incident plane, and
the linearly polarized light incidents on the surface is also elliptically polarized
same as the polar Kerr effect. Third, in the transverse Kerr effect, the magnetiza-
tion is also in the plane of the sample but is perpendicular to the incident plane
(Figure 8.23c). In this case, the incident light is not normal to the reflection sur-
face, but there is no change in the polarization of the incident light. Indeed, only
the intensity of light will be changed, and the change in reflectivity is proportional
to the magnetization orientation.
The longitudinal and transverse Kerr effects are generally used to study the
in-plane magnetic anisotropy, whereas the polar effect is used to study thin
films that exhibit perpendicular anisotropy. Typically, magneto-optic signal and
images can be generated by using MOKE magnetometer and MOKE microscopy,
respectively. For microscopy, magneto-optic images can be obtained either by
8.7 Hall Effect 191
using conventional imaging optics or a finely focused laser spot across the
sample surface. Therefore, the technique is moderately surface sensitive and can
be used to image magnetic domains in thin films that are only a few monolayers
thick.
8.7 Hall Effect

8.7.1 Ordinary Hall Effect
As shown in Figure 8.24a, Hall effect describes the production of a potential dif-
ference (the Hall voltage V H ) across a current-carrying conductor in the presence
of magnetic field, which is perpendicular to both current and the magnetic field.
The Lorentz force F on a charge carrier can be expressed as
F = q(E + (v × B))
where E is the Hall electric field, v the drift velocity of the current, q the charge of
a particle, and B the magnetic induction of the conductor. At a steady state with
F = 0, the Hall voltage can be obtained as
Ix Bz
VH =
nte
where I is the current, n the charge carrier density, t the thickness of conductor,
and e the charge of each electron. The Hall coefficient is defined as
Ey VH t 1
RH = = =−
jx B IB ne
where j is the current density of the carrier electrons and Ey is the induced electric
field. As a result, the carrier density or the magnetic field can be measured with
the aim of Hall effect.
8.7.2 Anomalous Hall Effect

In addition to the ordinary Hall effect presented in semiconductors and metals,
an effect called anomalous Hall effect (AHE) has been discovered claiming that
part of Hall voltage is proportional to the magnetization at fields up to magnetic
saturation (Pugh and Rostoker 1953). AHE is usually much larger than the ordi-
nary Hall effect. Therefore, for ferromagnetic elements Fe, Co, and Ni, the Hall
voltage is determined mainly by the magnetization, resulting in an unusual non-
linear relationship between Hall electric field and magnetic induction coming
from AHE. The AHE has been recognized as a useful tool for measuring M–H
hysteresis loops of perpendicular magnetic recording media, spintronic devices,
and diluted magnetic semiconductors.
The relationship between Hall electric field per current density EH , magnetic
field Hz , and magnetization Mz in most ferromagnetic materials is established as
EH = R0 Hz + R1 Mz
B M
–F –
– – F
– –
– – –
– –
E –
t w E
t w
B=0
I I
(a) (b)
Figure 8.24 (a) Ordinary Hall effect and (b) anomalous Hall effect (AHE). B is the magnetic
induction rise by external magnetic field; M is the intrinsic spontaneous magnetization.
where the first term represents the ordinary Hall effect mentioned, while the sec-
ond term represents the AHE contribution due to the spontaneous magnetization
as illustrated in Figure 8.24b. Unlike R0 that only depends on the carrier density,
R1 depends on the effects of the spin–orbit interaction of spin electrons (Karplus
and Luttinger 1954). As a result, specific magnetic parameters of FM materials,
such as magnetization processes and types of materials, can be determined using
AHE technique.
8.7.3 Spin Hall Effect

There is another transport phenomenon in magnetic materials analogous to the
classical Hall effect – the spin Hall effect (SHE), which consists of the appearance
of spin accumulation on the lateral surfaces of a current-carrying sample. From
Figure 8.25a, we can see that SHE causes electrons carrying different signs of spins
to drift in opposite boundaries, thus generating a transverse pure spin current.
When the current direction is reversed, the directions of spin orientation are also
reversed.
Spin Hall voltage is originated from two possible mechanisms, where an elec-
trical current with moving charges transforms into a spin current with moving
spins without charge flow. The first possible mechanism is spin-dependent Mott
scattering, where carriers with opposite spins diffuse in opposite direction when
σ
Spin current Is Electromotive
σ force EISHE
+
–
Spin curr
ent I
s
(b)
(a) Charge current Ic
Figure 8.25 (a) Spin Hall effect and (b) inverse spin Hall effect, 𝝈 denotes the spin polarization
vector of the spin current.
References 193
colliding with the impurities within the material. The second possible mecha-
nism is caused by the intrinsic properties of the material that the trajectories
of carriers are distorted due to spin–orbit interaction as a consequence of the
asymmetries in the material. Owing to the accumulation of spins of both lateral
boundaries, the current and spin Hall voltage can be detected. By the inverse
spin Hall effect (ISHE, Figure 8.25b) in a strong spin–orbit coupling material,
the spin–orbit interaction bends trajectories of these two electrons in the same
direction. The spin current can then be converted into an electric field EISHE with a
charge current again in the transverse direction, resulting in charge accumulation
at the sample edges.
For the SHE, no magnetic field is required as it is a purely spin-based phe-
nomenon dissimilar to the ordinary Hall effect, but it behaves similar to AHE,
known for a long time in ferromagnet, which also originates from spin–orbit
interaction. SHE can be used to manipulate electron spins electrically, and a pure
spin current can even be generated by non-magnetic materials.
References
Baibich, M.N., Broto, J.M., Fert, A. et al. (1988). Giant magnetoresistance of
(001)Fe/(001)Cr magnetic superlattices. Physical Review Letters 61 (21):
2472–2475.
Čápová, K., Čáp, I., Janoušek, L., and Smetana, M. (2008). Recent trends in
electromagnetic non-destructive sensing. Advances in Electrical and Electronic
Engineering 7 (1): 322–325. Available at: http://hdl.handle.net/10084/84200.
Celotta, R.J., Unguris, J., Kelley, M.H., and Pierce, D.T. (2002). Characterization of
Materials. In: Techniques to measure magnetic domain structures (ed. E.N.
Kaufmann), 1. https://doi.org/10.1002/0471266965.com046. Available at: http://
doi.wiley.com/10.1002/0471266965.com046.
Daughton, J.M. (1992). Magnetoresistive memory technology. Thin Solid Films 216:
162–168.
Eerenstein, W., Mathur, N.D., and Scott, J.F. (2006). Multiferroic and
magnetoelectric materials. Nature 442 (7104): 759–765. Available at: http://www
.ncbi.nlm.nih.gov/pubmed/16915279.
García-Martín, J., Gómez-Gil, J., and Vázquez-Sánchez, E. (2011). Non-destructive
techniques based on eddy current testing. Sensors 11 (3): 2525–2565. Available at:
http://www.mdpi.com/1424-8220/11/3/2525.
Grünberg, P., Schreiber, R., Pang, Y. et al. (1987). Layered magnetic structures:
evidence for antiferromagnetic coupling of Fe layers across Cr interlayers. Journal
of Applied Physics 61 (8): 3750–3752.
Hu, J., Li, Z., Chen, L., and Nan, C. (2012). Design of a voltage-controlled magnetic
random access memory based on anisotropic magnetoresistance in a single
magnetic layer. Advanced Materials 24 (21): 2869–2873. Available at: https://
onlinelibrary.wiley.com/doi/abs/10.1002/adma.201201004.
Inoue, J.I. (2013). GMR, TMR, BMR, and related phenomena. In: Nanomagnetism
and Spintronics, 2e, 15–106. Elsevier B.V. Available at: http://dx.doi.org/10.1016/
B978-0-444-63279-1.00002-2.
Jiles, D. (2015). Introduction to Magnetism and Magnetic Materials, 3e. CRC Press.
Jiles, D.C. (1995). Theory of the magnetomechanical effect. Journal of Physics D:
Applied Physics 28: 1537–1546.
Jin, S., Tiefel, T.H., McCormack, M. et al. (1994). Thousandfold change in resistivity
in magnetoresistive La-Ca-Mn-O films. Science 264 (5157): 413–415. Available at:
http://www.sciencemag.org/cgi/doi/10.1126/science.264.5157.413.
Julliere, M. (1975). Tunneling between ferromagnetic films. Physics Letters A 54 (3):
225–226.
Karplus, R. and Luttinger, J.M. (1954). Hall effect in ferromagnetics. Physical Review
95 (5): 1154–1160.
Kiselev, S.V., Ozerov, R.P., and Zhdanov, G.S. (1963). Detection of magnetic order in
ferroelectric BiFeO3 by neutron diffraction. Soviet Physics – Doklady 7 (8):
742–744.
Kriegner, D., Výborný, K., Olejník, K. et al. (2016). Multiple-stable anisotropic
magnetoresistance memory in antiferromagnetic MnTe. Nature Communications
7: 11623. Available at: http://dx.doi.org/10.1038/ncomms11623.
Liu, Y., Tang, N., Wan, X. et al. (2013). Realization of ferromagnetic graphene oxide
with high magnetization by doping graphene oxide with nitrogen. Scientific
Reports 3: 2566. Available at: http://dx.doi.org/10.1038/srep02566.
Lu, C., Gao, B., Dong, S. et al. (2018). Revealing controllable anisotropic
magnetoresistance in spin–orbit coupled antiferromagnet Sr2 IrO4 . Advanced
Functional Materials 28 (17): 1706589. Available at: https://onlinelibrary.wiley
.com/doi/abs/10.1002/adfm.201706589.
Ma, Y.W., Lu, Y.H., Ding, J. et al. (2012). Room temperature ferromagnetism in
Teflon due to carbon dangling bonds. Nature Communications 3: 727. Available
at: http://dx.doi.org/10.1038/ncomms1689.
Mazin, I.I. (1999). How to define and calculate the degree of spin polarization in
ferromagnets. Physical Review Letters 83 (7): 1427–1430.
Nickel, J. (1995). Magnetoresistance Overview: HP Laboratories Technical Report:
HPL-95-60.
Poon, T.Y., Tse, N.C.F., and Lau, R.W.H. (2013). Extending the GMR current
measurement range with a counteracting magnetic field. Sensors 13 (6):
8042–8059. Available at: http://www.mdpi.com/1424-8220/13/6/8042.
Pugh, E.M. and Rostoker, N. (1953). Hall effect in ferromagnetic materials. Reviews
of Modern Physics 25 (1): 151.
Rao, C.N.R. and Mahesh, R. (1997). Giant magnetoresistance in manganese oxides.
Current Opinion in Solid State and Materials Science 2 (1): 32–39. Available at:
http://www.sciencedirect.com/science/article/pii/S1359028697801027.
Saurenbach, F., Walz, U., Hinchey, L. et al. (1988). Static and dynamic magnetic
properties of Fe-Cr-layered structures with antiferromagnetic interlayer
exchange. Journal of Applied Physics 63 (8): 3473–3475.
Savage, H.T. and Spano, M.L. (1982). Theory and application of highly
magnetoelastic Metglas 2605SC. Journal of Applied Physics 53: 8092.
Spinu, L., Dodrill, B.C., and Radu, C. (2013). Magnetometry measurements. In: ,
1–3. Magnetics Technology International.
Tsai, C.Y., Chen, H.R., Hsieh, W.F. et al. (2013). Stress-mediated magnetic
anisotropy and magnetoelastic coupling in epitaxial multiferroic
References 195
PbTiO3 -CoFe2 O4 nanostructures. Applied Physics Letters 102 (13): 132905.

Xiao, Y., Gou, X.-F., and Zhang, D.-G. (2017). A one-dimension nonlinear hysteretic
constitutive model with elasto-thermo-magnetic coupling for giant
magnetostrictive materials. Journal of Magnetism and Magnetic Materials 441:
642–649.
197
Multiferroics: Single Phase and Composites
9.1 Introduction on Multiferroic

Multiferroics is defined as a physical phenomenon of simultaneous exist of
more than one ferroic order parameters in a material. The basic primary ferroic
order parameters are ferroelectricity, ferromagnetism, and ferroelasticity.
Non-primary order parameters such as antiferroelectricity, antiferromagnetism,
and ferrimagnetism can also be included into the definition of multiferroics.
More than a century ago, Pierre Curie predicted that there should be materials
whose magnetism is induced by electric field E and its polarization by mag-
netic field H (Curie 1894). This conjecture can be illustrated by the following
equations:
M𝛼 = G𝛼𝛽 E𝛽 (9.1)
P𝛼 = G𝛼𝛽 H𝛽 (9.2)
where M and P are ferromagnetism and ferroelectric polarization, respectively,
and G is the corresponding permittivity reflecting the coupling of magnetism and
electric polarization. Subscripts 𝛼 and 𝛽 are, respectively, the directions of field
applied and response.
These two equations illustrate the electrical control of magnetism and mag-
netic control of electricity. The merit of multiferroic is not only the existence of
more than two ferroic properties in one material but also the interplay and mutual
control of each other through magnetoelectric (ME) coupling effect. This will
broaden the horizon of design and application of electromagnetic devices, and it
would be amazing if M–E loop and P–H loops can be easily achieved beyond the
well-known P–E and M–H loops as illustrated in Figure 9.1.
However, until 1959, this conjecture, ferroelectric magnetism or magnetic fer-
roelectricity, was first experimentally observed in Cr2 O3 , and in 1966 Schmid
reported another multiferroic Ni3 B7 O13 (Ascher et al. 1966). The reason that this
conjecture has not been observed for half century since predicted is because of
the fact that it is very rare for a single phase material exhibiting detectable level
of multiferroics (Dzyaloshinskii 1960; Astrov 1960). The physics behind is intro-
duced in the following.
Multiferroic properties are closely linked to crystal structural symmetries.
The primary ferroic properties can be characterized by their symmetry breaking

198 9 Multiferroics: Single Phase and Composites
P M
H E
Figure 9.1 Illustration of multiferroics with expected P–H and M–E hysteresis loops.
under space and time inversions. Space inversion will reverse the direction of
polarization P, while leaving the magnetization M invariant. Time reversal,
in turn, will change the sign of M, while the sign of P remains invariant.
Magnetoelectric multiferroics require simultaneous violation of space and
time inversion symmetries. In BiFeO3 , for example, off-centering of ions gives
rise to an electric polarization, while additional magnetic ordering breaks
time-reversal symmetry. Therefore, BiFeO3 has been extensively studied as
a model multiferroic material that will be particularly introduced in the
following text.
Although scientists have already started experimental observation and theoret-
ical prediction of multiferroics in single phase materials many decades ago, the
term multiferroics was firstly used by H. Schmid in 1994 (Schmid 1994). Since
then, the concept of multiferroics has been expanded to materials that perform
any type of long-range magnetic ordering, a spontaneous electric polarization,
and/or ferroelasticity, and some researchers argue that the term multiferroics
can only be used in materials that exhibit coupling of the ferroic order parame-
ters. The overlap of these properties in ferroic materials is illustrated in Figure 9.2
(Eerenstein et al. 2006).
Multiferroic
Ferroelectric Ferromagnetism
Electrically Magnetically
polarizable polarizable
Magnetoelectric
Figure 9.2 Relationship between ferroelectrics, ferromagnetics, multiferroics, and even the
magnetoelectrics. Source: Adapted from (Eerenstein et al. (2006)).
9.3 Why Are There so Few Magnetic Ferroelectrics? 199
9.2 Magnetoelectric Effect

The magnetoelectric effect is defined as the generation of electric polarization in
a material under an applied external magnetic field, or conversely, the production
of an induced magnetization by an external electric field. For device application,
a detectable ME coupling coefficient is essential.
In ME composites, the ME coupling is associated with mechanical interaction
between the magnetostrictive and piezoelectric phases. The induced polarization
P is related to the magnetic field H by the expression
P = 𝛼H (9.3)
where 𝛼 is the second-rank ME-susceptibility tensor. A parameter of importance
is the ME voltage coefficient 𝛼 E that is defined as
𝜕E
𝛼E = (9.4)
𝜕H
The relationship between 𝛼 and 𝛼 E is 𝛼 = 𝜖 0 𝜖 r 𝛼 E and 𝜖 r is the relative permit-
tivity of the dielectric materials. While in single phase materials, the ME effect
is via the coupling between magnetic and ferroelectric orders; this is extensively
described in Landau’s theory by expressing the free energy F of the system in
terms of an applied magnetic field H and an applied electric field E (Landau and
Lifshitz 1960). From the free energy term, the ME coupling coefficient can be
derived by second-order differentiation.
Expansion of the free energy in isotropic system following Siratori’s expression
(Siratori et al. 1992)
F(E, H) = F0 − P 𝟎 ⋅ E − M 𝟎 ⋅ H − ET ⋅ 𝜒e ⋅ E − H T ⋅ 𝜒m ⋅ H − ET ⋅ 𝛼 ⋅ H − …
where F 0 is the free energy independent of E and H, P0 and M 0 denote the spon-
taneous polarization and magnetization, and tensors 𝜒 e and 𝜒 m are the dielectric
and magnetic susceptibilities. The tensor 𝛼 denotes ME coupling coefficients, cor-
responding to induction of polarization by a magnetic field or of magnetization
by an electric field that is designated as the linear ME effect.
Derivative of F in E leads to the electric polarization:
𝜕F
P=− = P 𝟎 + 2𝜒e ⋅ E + 𝛼 ⋅ H + …
𝜕E
Derivative of F in H leads to the magnetization:
𝜕F
M=− = M 𝟎 + 2𝜒m ⋅ H + 𝛼 ⋅ E + …
𝜕H
ME coupling factor can be found by
𝜕2F 𝜕M 𝜕P
𝛼= = =
𝜕H𝜕E 𝜕E 𝜕H
9.3 Why Are There so Few Magnetic Ferroelectrics?

The reason behind the fact that it is rare for a material to be magnetic-
ferroelectrics is the mutual exclusion of magnetism and ferroelectricity. In
fact, a ferroelectric material must be an insulator, while ferromagnets are often

metals. For example, the driving force for ferromagnetism in the elemental
ferromagnets Fe, Co, and Ni and their alloys is the high density of states at the
Fermi level, resulting in metallicity. The magnetism requires partially occupied
d or f orbitals, while ferroelectricity usually requires that d orbitals should be
empty (Matthias rule requires diamagnetic ions, i.e. d0 ), such as Ti4+ and Ta5+ .
Therefore, multiferroism involves multiple competing factors, among which
d-electron occupancy in the transition metal is critical.
To be a multiferroic with partially occupied d orbital and also distortion of
lattice forming ferroelectric polarization, Jahn–Teller effect plays a crucial role.
The Jahn–Teller effect (JT effect or JTE) is an important mechanism of spon-
taneous symmetry breaking in molecular and solid-state systems; this effect is
named after Hermann Arthur Jahn and Edward Teller, who first reported studies
about it in 1937 (Jahn and Teller 1937). The Jahn–Teller effect, sometimes also
known as Jahn–Teller distortion, describes the distortion of molecules and ions
to a lower energy conformation that is associated with certain electron configu-
rations. The Jahn–Teller theorem essentially states that any nonlinear molecule
with an orbital and spin motion degenerate in electronic state will undergo a geo-
metrical distortion to compensate that degeneracy, because the distortion lowers
the overall energy of the species.
Not only rare in nature, ME voltage response in single phase compounds is also
generally weak. However, relatively strong ME voltage output may be achieved by
artificially integrating magnetostrictive and piezoelectric phases in a composite
material in which the mechanical deformation from magnetostriction results in
measurable voltage generated from piezoelectric effect. In 1970s, Van den Boom-
gaard firstly synthesized bulk composites of cobalt ferrite and nickel ferrite with
BaTiO3 by a solid-state reaction technique, yielding a ME voltage coefficient 𝛼 E
of 130 mV/cm⋅Oe (vanSuchtelen 1972).
In the recent decades, new single phase multiferroics and ME compos-
ites in a variety of forms, such as piezoelectric/magnetostrictive stacks and
polymer–ceramic 0–3 composites (particles in matrix), have attracted consider-
able interest due to their large ME coupling effect and potential application in
novel electronic devices. In the following sessions, recent results on single phase
and composite multiferroic materials will be introduced.
9.4 Single Phase Multiferroic Materials

Typical single phase multiferroics belong to the group of perovskite transition
metal oxides. The most typical single phase multiferroics with relatively large
and measurable ME effect is BiFeO3 . Other examples include rare-earth man-
ganites and ferrites such as TbMnO3 , BiMnO3 , HoMn2 O5 , and LuFe2 O4 . In
addition, some non-oxide materials such as BaNiF4 and spinel chalcogenides,
e.g. ZnCr2 Se4 , are also included in the multiferroic family.
The history of ME multiferroics can be traced back to the 1960s. After the
initial burst of interest of research on antiferromagnetic Cr2 O3 , more and more
materials either artificially prepared or existed in nature are synthesized and

discovered in the form of ceramic, single crystal, and even thin film. Although
the progress of research in multiferroic materials remained static until 2000, the
discovery of strong magnetic and electric coupling in orthorhombic TbMnO3
(Kimura et al. 2003) and TbMn2 O5 (Hur et al. 2004) stimulated people’s interest
again in the field of multiferroic complex oxides. A spontaneous electric polar-
ization in MnWO4 below ∼13 K in the magnetically ordered phase (Heyer et al.
2006) reveals that MnWO4 is a novel multiferroic material. Giant magnetoelastic
coupling in multiferroic hexagonal manganites has been observed in single
crystal YMnO3 (Lee et al. 2008). This extremely large magnetoelastic coupling,
larger by 2 orders of magnitude than any other magnetic materials, is the primary
origin of their multiferroic phenomenon.
Most importantly, the discovery of large ferroelectric polarization (Wang et al.
2003) and the realization of electric-field-controlled magnetic state (Zhao et al.
2006) in epitaxial thin films of antiferromagnetic BiFeO3 (BFO) create a new
degree of freedom to manipulate the ME coupling in nanoscale and pave the
way for using multiferroic ME materials to on-chip integration with new gener-
ations of electronic devices. Currently as the most promising multiferroic mate-
rials, BFO has already shown its potential application in spintronics (Béa et al.
2008) and ferroelectric random access memory (FeRAM) (Mathur 2008). In fact,
the theoretical calculation and experimental investigation on BFO started from
1960s (Roginskaya et al. 1966; Krainik et al. 1966; Kiselev et al. 1963), but it did not
show any practical potentials for applications until the observation of large room
temperature ferroelectric polarization combining interesting magnetic proper-
ties in strongly strained epitaxial BFO films (but is always absent in its bulk coun-
terparts). Among the few reported room temperature single phase ferroelectric
antiferromagnetism (Kimura et al. 2005), BFO thin film shows the highest fer-
roelectric polarization of 90 μC/cm2 (Figure 9.3) along its ferroelectric-preferred
pseudocubic ⟨111⟩c direction (Chu et al. 2007).
The structure and properties of the bulk single crystal form BFO have been
extensively studied, and it has been shown to possess a rhombohedrally distorted
perovskite structure at ground state. Unlike other ferroelectric antiferromagnet-
ics, BFO possesses a very high ferroelectric Curie temperature (T c ) of ∼1100 K
and a high antiferromagnetic Néel temperature (T N ) of ∼640 K. The structure of
BFO is illustrated by distorted perovskite unit cells connected along their body
diagonal, denoted by the pseudocubic ⟨111⟩c direction, to form a rhombohedral
unit cell with ac = 3.96 Å, as shown in Figure 9.4.
The two oxygen octahedrals in the connected perovskite units are rotated
clockwise and counterclockwise around this axis by 13.8∘ , with the Fe cation
shifted by 0.135 Å along the same axis away from the center of an oxygen
octahedron (Zavaliche et al. 2006). The ferroelectric state is achieved by a large
displacement of the Bi ions relative to the FeO6 octahedral. The domain structure
in BFO is illustrated in Figure 9.5, where the eight ⟨111⟩c oriented polarizations
forming 71∘ , 109∘ , and 180∘ domain walls are shown.
This BFO domain structure indicates the formation of eight possible polar-
ization variants corresponding to four structural variants lying along the
pseudocubic ⟨111⟩c directions. It’s magnetic ordering also classify BFO as
120
BFO(111)
BFO(110)
80 BFO(100)
40
–40
–80
–120
–10 –5 0 5 10
Applied voltage (V)
Figure 9.3 Ferroelectric hysteresis loops of epitaxial BFO thin films along different crystalline
directions. Source: Chu et al. (2007). Reprinted with permission of Elsevier.
Pi+ Figure 9.4 (a) Crystal structure of

rhombohedral BFO and its
Pi+/– ferroelectric polarized direction.
(b) Projection along ⟨111⟩c
direction. (c) Projection along
⟨110⟩c direction. Source: Zavaliche
et al. (2006). Reprinted with
(c) permission of Taylor & Francis.
Pi+
(b)
Bi
Fe
(a) O
G-type antiferromagnetic as shown in Figure 9.6, in which the Fe magnetic

moments are aligned ferromagnetically within (111)c and antiferromagnetically
between adjacent (111)c . Antiferromagnetic ordering is of three types particu-
larly in perovskite-type oxides. They are the following. A-type: the intra-plane
coupling is ferromagnetic, while inter-plane coupling is antiferromagnetic.
C-type: the intra-plane coupling is antiferromagnetic, while inter-plane coupling
is ferromagnetic. G-type: both intra-plane and inter-plane coupling are antifer-
romagnetic. Figure 9.6a shows that BFO belongs to G-type (antiferromagnetic
(a)
(b)
(c)
Figure 9.5 Domain structures in BFO with (a) 71∘ , (b) 109∘ , and (c) 180∘ domain walls. Source:
Adapted from Baek et al. (2010).
[111]
MFe1
MFe2
(a) (b)
Figure 9.6 (a) G-type antiferromagnetic plane of BFO is perpendicular to the ferroelectric
polarized [111]c direction, where antiferromagnetic exist in both intra-plane and inter-plane.
(b) Magnetic structure including the Dzyaloshinskii–Moriya interaction. The two Fe3+ magnetic
moment (MFe1 and MFe2 ) are canted in the (111)c plane so that there is a resulting macroscopic
remnant magnetization M. Source: Wu et al. (2016). Reprinted with permission of Elsevier.
in both intra-plane and inter-plane). Consequently, the antiferromagnetic and

ferroelectric orders can be coupled together, namely, the switching of ferroelec-
tric dipoles coupled with the switching of antiferromagnetic plane, indicating
great scientific and technical values in the spintronic, multiferroic tunneling
devices, etc.
The perovskite multiferroic BFO simultaneously shows ferroelectric and anti-
ferromagnetic orders (Smolenskii and Chupis 1982), exhibiting weak magnetism
at room temperature due to a canted spin structure. Spin canting is a phe-
nomenon through which spins are tilted by a small angle about their axis rather
than being exactly parallel or anti-parallel. The antisymmetric exchange is a con-

tribution to the total magnetic exchange interaction between two neighboring
magnetic spins; this term is also called the Dzyaloshinskii–Moriya interaction.
A magnetically ordered system favors a spin canting of otherwise (anti)parallel
aligned magnetic moments and, thus, is a source of weak ferromagnetic behavior
in an antiferromagnet.
9.4.1 Switching Mechanism in BFO Films

The coupling between ferroelectric and antiferromagnetic orders of BFO has
been predicted (Ederer and Spaldin 2005) and experimentally observed (Apos-
tolova and Wesselinowa 2008). Therefore, in order to control the antiferromag-
netic order by electric field in a high-degree of freedom, it becomes significant
to investigate the intrinsic ferroelectric domain configuration and the switching
path as well as the physics behind the domain dynamics under a bias field.
The potential application of BFO in ferroelectric non-volatile memory also
depends on deep understanding on its domain structure and switching mech-
anism. Therefore, ferroelectric switching in monodomain BFO epitaxial thin
film will provide a basic insight into the distributions of electric, magnetic, and
stress fields within domain structures, allowing us to manipulate and control the
domains and domain walls (Seidel et al. 2009).
The ferroelectric domain structure of BFO is related to its crystallographic
direction. As indicated in Figure 9.7, starting from Figure. 9.7(a), the <111>c
Bi
Fe
[001] 71
°
0
(a) (b)
0°
18
109
°
(c) (d)
Figure 9.7 (a) original polarization, (b) after 71∘ rotation, (c) after 109∘ rotation, and (c) after
180∘ rotation. Source: Adapted from Zhao et al. (2006).
9.5 ME Composite Materials 205
polarization can be switched by 71∘ to the direction shown in (b) and 109∘ switch-
ing to (c) and 180∘ switching to (d).
9.5 ME Composite Materials

In order to implement multiferroic properties in device application, reasonably
large ME coupling factor is essential. Therefore, people have started to make
composites and heterostructures exhibiting the coexistence of piezoelectric and
magnetostrictive properties. The ME composite can be in the forms of solid-state
reaction-based composite ceramics, piezoelectric and magnetostrictive-layered
stacks, and even in thin film multilayers and novel nanostructures. The ME
coupling effect in composites originates from the mechanical deformation of
magnetostrictive materials under magnetic field, which induces the accumulated
charge in piezoelectric phase due to piezoelectric effect. For example, piezo-
electric oxide and magnetostrictive ferrite oxide BaTiO3 /CoFe2 O4 composite
can be sintered together using their respective oxide powder, where phase
separation can be realized under high-temperature sintering. The maximum ME
coefficient once reached a value of 130 mV/cm⋅Oe (Van den Boomgaard et al.
1974). However, the phase and element diffusion may happen in the composite
during the sintering, leading to enhanced leakage current and failure of electrical
poling. Although a lot of researchers made their effort in reducing the leakage
current and dielectric loss in the solid-state composite by doping elements in the
respective components (Devan and Chougule 2007; Kadam et al. 2005), the ME
coefficient was not enhanced significantly. Based on the previously mentioned
consideration, people started to put their interests on bulk ME materials with
2–2 structure (multilayer structure), combining the piezoelectric and magne-
tostrictive components together by a conductive epoxy. Srinivasan et al. reported
a ME coefficient of 270 mV/cm⋅Oe in PZT (Pb(Zrx, Ti1−x )O3 )/ferrite laminated
composite (Srinivasan et al. 2003). The multilayered composite was also prepared
as a 2–2 structure with an alternating piezoelectric and magnetostrictive phases,
and a giant ME coefficient of 1500 mV/cm⋅Oe was achieved in the NiFe2 O4 /PZT
multilayered laminate (Srinivasan et al. 2001). The ME coefficient could be
further increased by improving the mechanical coupling in the piezoelectric and
magnetostrictive laminates.
In recent years, polymer materials such as PVDF or P(VDF-TrFE) have been
chosen as promising candidates for the piezoelectric phase in the ME composite
because they have relatively good voltage sensitivity, high electromechanical
coupling, low dielectric constant and low dielectric loss. Moreover, polymeric
piezoelectric materials can stand freely to avoid substrate clamping effect, which
is the main issue for other thin film structures, and their shapes and sizes can also
be easily modified by conventional polymeric processing. Nan et al. developed a
new polymer-based ME composite, in which piezoelectric and magnetostrictive
powders were dispersed into the piezoelectric copolymer, forming a double layer,
sandwich, or even a multilayered structure by bonding the two components
together (shown in Figure 9.8) (Lin et al. 2005). The maximum ME sensitivity
of such composites is about 6 V/cm⋅Oe at about 90 kHz. At the same time,
Figure 9.8 (a) Schematic diagram of

Polarization the composite (tp is the thickness of
the piezoelectric phase and L is the
total thickness) and (b) a typical
TFD-PVDF layer
cross-sectional SEM images of the
L fractured surface of the composite.
PZT-PVDF layer tp TFD denotes Terfenol-D. Source:
Adapted from Lin et al. (2005).
TFD-PVDF layer
(a) (b)
theoretical calculation derived by Green’s function technique also supports the

experimental data (Nan et al. 2001). The record of ME coefficient was broken
again after the appearance of Metglas/PVDF laminate, developed by Dong et al.
which has a ∼300 V/cm⋅Oe of ME coefficient at their resonance frequency
(Figure 9.9) (Zhai et al. 2006).
The demand of miniaturization and high-density storage in next-generation
electronic devices drives people’s interest to develop and design thin films and
nanostructured ME nanocomposites and devices. The coexistence of strong
ferromagnetic and ferroelectric behaviors and realization of ME coupling effect
are necessary in the nanocomposites. The observation of ME coupling effect in
the self-assembled multiferroic BaTiO3 –CoFe2 O4 (BTO–CFO) nanostructures
reveals that the interconversion of energies stored in electric and magnetic fields
can be realized in micro- or even nanodevices (Zheng et al. 2004; Zavaliche
et al. 2007). As shown in Figure 9.10, the self-assembled ME nanocomposite,
consisting of perovskite BaTiO3 (BTO) and spinel CoFe2 O4 (CFO), exhibits
a large magnetization change around the ferroelectric Curie temperature of
BTO, indicating a strong ME coupling response happened in the mesoscopic
order. This effect, for comparison, was not observed in the BTO/CFO 2–2
heterostructure, which means that the large substrate clamping was not released
as much as the 1–3 self-assembled nanostructure composite.
The theoretical calculation, shown in Figure 9.11, also predicts the large
magnetic-field-induced electric polarization in nanostructured multiferroic
Figure 9.9 (a) Photograph of a

Metglas/PVDF laminate, (b) the
unimorph, and (c) three-layer
configuration. Source: Adapted
from Zhai et al. (2006).
(a)
M
P
(b) Unimorph
PVDF
Metglas
M
P
M
(c) T–L
9.6 Modeling the Interfacial Coupling in Multilayered ME Thin Film 207
ΔM = 16 emu/cm3
200
Magnetization (emu/cm3)
150 Spinel CoFe2O4

nanopollars
BaTiO3 matrix
100
1–3 nanostructure Substrate
50
300 320 340 360 380 400 420 440
Temperature (K)
Figure 9.10 Temperature-dependent magnetization in self-assembled nanostructure. Inset:

Schematic diagram of phase separation of spinel and perovskite nanocomposite on perovskite
substrate. Source: Adapted from Zheng et al. (2004).
composite films (Nan et al. 2005), which further proves that the substrate
clamping effect will destroy the mechanical coupling between piezoelectric and
magnetostrictive phase in the 2–2 heterostructured thin film.
9.6 Modeling the Interfacial Coupling in Multilayered

ME Thin Film
In this section, epitaxial PZT/CFO multilayers with perovskite/spinel interface
and residual strain are used as an example to demonstrate the modeling of inter-
facial coupling in multilayered ME thin film. The comparison of ferroelectric and
ferromagnetic properties between the multilayered films and the single phase
is introduced. This interfacial engineering in perovskite/spinel heterostructure
provides an approach to control the ferroelectric, magnetic, and ME interfacial
coupling in epitaxial multilayers and superlattice multiferroics.
9.6.1 PZT/CFO Multilayered Heterostructures

PZT and CFO as well as their epitaxial multilayered nanocomposite thin films
were deposited on Nb:STO single crystal substrates by pulsed laser deposition
(PLD). For comparison purpose, the film thickness is maintained as a constant
for single phase and multilayered films with 3, 5, and 11 layers. Figure 9.12a is
500
400
300
ΔP3(μC/m2)
Nano system 1–3

2–2
200
Bulk system 1–3
2–2
100
0
0 100 200 300 400
(a) H3(kA/m)
Composite
film with 1–3
nanostructure
Substrate
Composite film
with 2–2
structure
Substrate
(b)
Figure 9.11 (a) Theoretical ME coupling effect in nano and bulk systems with 1–3 and 2–2
structures, and (b) the corresponding schematic diagrams of 1–3 and 2–2 nanostructures.
Source: Adapted from Nan et al. (2005).
the low-magnification cross-sectional bright-field image of the PZT/CFO mul-

tilayered film. The relatively dark area corresponds to the PZT phase due to its
relatively heavy elements. Selected area electron diffraction pattern (SAED) is
shown in Figure 9.12b. The cubic-on-cubic epitaxial relationship between PZT,
CFO, and STO substrate with fourfold symmetry is confirmed.
STO022
PZT022
CFO022
500 nm
(a) (b)
Figure 9.12 (a) Low-magnification TEM images of 11-layered PZT/CFO nanocomposite films,
(b) the selected area electron diffraction pattern including PZT, CFO, and STO substrate.
Source: Adapted from Zhang et al. (2010).
9.6.2 Ferroelectric Properties of PZT/CFO Multilayers

P–E loop measurement is essential for ferroelectric characterization. With the
increase of number of layers, remnant polarizations of the PZT film and compo-
sition films of 3, 5 and 11 layers are 31.8, 17.9, 15.3, and 8.5 μC/cm2 , respectively.
The coercive field increases from 80 kV/cm for the PZT film to ∼200 kV/cm for
the multilayered films. Since the coercive field of each PZT layer in the composite
film should be a function of external electric field, volume fraction, and dielectric
constant of PZT and CFO, if we neglect the conductivity of the multilayers, the
actual electric field applied on each PZT layer can be described by
Em
EP = 𝜖 (9.5)
𝜙 P + 𝜙C 𝜖 P
C
where Em and Ep are the electric field applied on the multilayers and the actual
field applied on each PZT layer, respectively; 𝜙P , 𝜙C , 𝜖 P , and 𝜖 C are the volume
fractions and dielectric constants of the PZT and CFO layers in the films. There-
fore, we have to increase the external field in the multilayers in order to maintain
the switching electric field in each PZT layer. It is easy to understand the signif-
icant enhancement of the coercive field in the films compared with that in the
single phase PZT film.
To modeling the number of layer-dependent polarization in the multilayered
thin films, the average residual stress suffered by the PZT phase in the epitaxial
multilayers can be estimated based on the XRD result. The in-plane a-axis lattice
parameters of PZT increase from 4.027 to 4.032, 4.035, and 4.041 Å, which
correspond to single PZT, 3-layered, 5-layered, and 11-layered films. The Young’s
modulus of the epitaxial PZT film is 92.6 GPa as reported elsewhere (Kanno
et al. 2004). Therefore the residual in-plane compressive stress suffered by the
corresponding epitaxial PZT films should be −520, −410, −330, and −204 MPa;
while for the cubic PZT bulk materials without residual stress, the lattice con-
stant a = 4.05 Å (Xu 1991). We adopt Landau–Devonshire’s phenomenological
thermodynamic formalism method to calculate the spontaneous polarization
(Hoon Oh and Jang 1998):
ΔG = 𝛼1 (P12 + P22 + P32 ) + 𝛼11 (P14 + P24 + P34 ) + 𝛼12 (P12 P22 + P22 P32 + P32 P12 )
+ 𝛼111 (P16 + P26 + P36 ) + 𝛼112 [P14 (P22 + P32 ) + P24 (P32 + P12 ) + P34 (P12 + P22 )]
1
+ 𝛼123 P12 P22 P32 − S11 (𝜎12 + 𝜎22 + 𝜎32 ) − S12 (𝜎1 𝜎2 + 𝜎2 𝜎3 + 𝜎3 𝜎1 )
2
1
− S44 (𝜎4 + 𝜎5 + 𝜎62 ) − Q11 (𝜎1 P12 + 𝜎2 P22 + 𝜎3 P32 )
2 2
2
− Q12 [𝜎1 (P22 + P32 ) + 𝜎2 (P32 + P12 ) + 𝜎3 (P12 + P22 )]
− Q44 (𝜎4 P2 P3 + 𝜎5 P3 P1 + 𝜎6 P1 P2 ) (9.6)
where Pi is the magnitude of the polarization vector along the direction i. 𝛼 i is

the dielectric stiffness, 𝛼 ij and 𝛼 ijk are the high-order stiffness coefficients at con-
stant stress, S11 , S12 , and S44 are the elastic compliances measured at constant
polarization, and Q11 , Q12 , and Q44 are the electrostrictive coefficients written in
polarization notation. 𝜎 i is tensile stresses along direction i.
We are now considering the elastic Gibbs function in the presence of a
two-dimensional compressive stress in tetragonal phase. The following relations
hold under this condition:
𝜎1 = 𝜎2 = −𝜎
𝜎3 = 𝜎4 = 𝜎5 = 𝜎6 = 0
P1 = P2 = 0 (9.7)
Therefore,
G = (𝛼1 + 2Q12 𝜎)P32 + 𝛼11 P34 + 𝛼111 P36 − (S11 + S12 )𝜎 2 (9.8)
dG
From the previously mentioned derivation and the condition dP3
= 0, we can
obtain
√
2 2
−𝛼11 ± 𝛼11 2 − 3𝛼111 (𝛼1 + 2Q12 𝜎)
PP = P3 = (9.9)
3𝛼111
Since the “-” in the equation is meaningless, it should be deleted.
In addition, according to the discussion of the “dilution” of the effect of ferro-
electric materials in a dielectric matrix in the BTO–STO–CTO superlattice (Lee
et al. 2005), when we consider the influence of the non-ferroelectric layer, the
electric displacement should be expressed by
( )
𝜖P 𝜖C 𝜙C 𝜖P
D= E+ 1− PP
𝜙 P 𝜖 C + 𝜙C 𝜖 P 𝜙P 𝜎C + 𝜙C 𝜎P
where D and P are the electric displacement and polarization of the multilayers.
Then the average polarization in the electrostatic model can be derived
PP
Pm = 𝜙C 𝜖P
(9.10)
1+ 𝜙P
⋅ 𝜖C
Therefore, we obtain the polarization of the multilayers

√ √
−𝛼11 + 𝛼11 2 −3𝛼111 (𝛼1 +2Q12 𝜎)
3𝛼111
Pm = 𝜙C 𝜖P
(9.11)
1+ 𝜙P
⋅ 𝜖C
where Pm is the average polarization of the PZT/CFO multilayered films and PP

indicates the polarization in each PZT layer.
The theoretical prediction describes the polarization decrease with the release
of in-plane compressive strain, thus qualitatively interprets the experimental
data. We consider the influence of the dielectric matrix in the multilayers
according to Eq. (9.11). The addition of non-ferroelectric phase reduces the
average polarization, compared with the pure PZT films suffered from different
in-plane stresses. It is easy to understand the dramatic decrease of the polar-
ization from single PZT film to the multilayered films as what we observed.
In addition, the size effect discussed by other reports in ferroelectrics may
be another factor to decrease polarization in the multilayers (Wong and Shin
2007). The remnant polarization, if we consider the size effect in PZT phase, is
expressed by
√
1 𝛼0 (Tc − T)
Pr = (9.12)
R 𝛽
where 𝛼 0 and 𝛽 are the modified and original Landau parameters, respectively,
3(t + 4𝛿)
R= (9.13)
2(t + 6𝛿)
32𝜅 (t + 3𝛿)
Tc = T0 − (9.14)
3𝛼0 t (t + 4𝛿)2
where t denotes the thickness of PZT layer and 𝛿 is the so-called extrapolation
length of the PZT/CFO interface; T 0 is the paraelectric to ferroelectric phase
transition temperature of bulk PZT, and 𝜅 is the positive constant representing
the contribution from the spatial variation of polarization.
When we take all of the previously mentioned parameters (Wong and
Shin 2007; Zhong et al. 1994; Liu et al. 2005; Kretschmer and Binder 1979)
as shown in Table 9.1, the theoretical fitting considering the size effect is
in agreement with our experimental observation. We may conclude that
the size effect should exists in our experimental observation and theoretical
calculations.
Table 9.1 Parameters used in calculating the polarization of the PZT film and PZT/CFO
multilayered heterostructures.
𝜶 1 (m/F) 𝜶 11 (m5 /C2 F) 𝜶 111 (m9 /C4 F) Q12 (m4 /C2 ) t 𝜹 𝜶0 𝜷 T0 𝜿
−8 × 106 7.12 × 107 2.35 × 106 −1.54 × 10−2 1 4 1 1 1 1

References
Apostolova, I. and Wesselinowa, J.M. (2008). Magnetic control of ferroelectric
properties in multiferroic BiFeO3 nanoparticles. Solid State Communications 147
(3): 94–97. Available at: http://www.sciencedirect.com/science/article/pii/
S0038109808002500.
Ascher, E. et al. (1966). Some properties of ferromagnetoelectric nickel-iodine
boracite, Ni3 B7 O13 I. Journal of Applied Physics 37 (3): 1404–1405. Available at:
https://doi.org/10.1063/1.1708493.
Astrov, D.N. (1960). The magnetoelectric effect in antiferromagnetics. Soviet Physics
Journal of Experimental and Theoretical Physics 11 (3): 708–709.
Baek, S.H., Jang, H.W., Eom, C.B. et al. (2010). Ferroelastic switching for nanoscale
non-volatile magnetoelectric devices. Nature Materials 9: 309. Available at:
http://dx.doi.org/10.1038/nmat2703.
Béa, H., Bibes, M., Barthélémy, A. et al. (2008). Mechanisms of exchange bias with
multiferroic BiFeO3 epitaxial thin films. Physical Review Letters 100 (1): 017204.
Available at: https://link.aps.org/doi/10.1103/PhysRevLett.100.017204.
Chu, Y.-H., Martin, L.W., Holcomb, M.B., and Ramesh, R. (2007). Controlling
magnetism with multiferroics. Materials Today 10 (10): 16–23. Available at:
Curie, P. (1894). Sur la symétrie dans les phénomènes physiques, symétrie d’un
champ électrique et d’un champ magnétique. Journal of Physics: Theories and
Application 3 (1): 393–415. Available at: https://doi.org/10.1051/jphystap:
018940030039300.
Devan, R.S. and Chougule, B.K. (2007). Effect of composition on coupled electric,
magnetic, and dielectric properties of two phase particulate magnetoelectric
composite. Journal of Applied Physics 101 (1): 014109. Available at: https://doi
.org/10.1063/1.2404773.
Dzyaloshinskii, I.E. (1960). On the magneto-electrical effect in antiferromagnets.
Soviet Physics Journal of Experimental and Theoretical Physics 10 (3): 628–629.
Ederer, C. and Spaldin, N.A. (2005). Weak ferromagnetism and magnetoelectric
coupling in bismuth ferrite. Physical Review B 71 (6): 060401. Available at: https://
link.aps.org/doi/10.1103/PhysRevB.71.060401.
.ncbi.nlm.nih.gov/pubmed/16915279 (accessed 30 May 2014).
Heyer, O., Hollmann, N., Khomskii, D. et al. (2006). A new multiferroic material:
MnWO4 . Journal of Physics: Condensed Matter 18 (39): L471. Available at: http://
stacks.iop.org/0953-8984/18/i=39/a=L01.
Hoon Oh, S. and Jang, H.M. (1998). Enhanced thermodynamic stability of
tetragonal-phase field in epitaxial Pb(Zr,Ti)O3 thin films under a two-dimensional
compressive stress. Applied Physics Letters 72 (12): 1457–1459. Available at:
https://doi.org/10.1063/1.120609.
Hur, N., Park, S., Sharma, P.A. et al. (2004). Electric polarization reversal and
memory in a multiferroic material induced by magnetic fields. Nature 429: 392.
References 213
Jahn, H.A. and Teller, E. (1937). Stability of polyatomic molecules in degenerate

electronic states – I-Orbital degeneracy. Proceedings of the Royal Society of
London. Series A – Mathematical and Physical Sciences 161 (905): 220–235.
Available at: http://rspa.royalsocietypublishing.org/content/161/905/220.abstract.
Kadam, S.L., Kanamadi, C.M., Patankar, K.K., and Chougule, B.K. (2005). Dielectric
behaviour and magnetoelectric effect in Ni0.5 Co0.5 Fe2 O4 +Ba0.8 Pb0.2 TiO3 ME
composites. Materials Letters 59 (2): 215–219. Available at: http://www
.sciencedirect.com/science/article/pii/S0167577X04006950.
Kanno, I., Yokoyama, Y., Kotera, H., and Wasa, K. (2004). Thermodynamic study of
c-axis-oriented epitaxial Pb(Zr,Ti)O3 thin films. Physical Review B 69 (6): 064103.
Kimura, T., Goto, T., Shintani, H. et al. (2003). Magnetic control of ferroelectric
polarization. Nature 426: 55. Available at: http://dx.doi.org/10.1038/nature02018.
Kimura, T., Lawes, G., and Ramirez, A.P. (2005). Electric polarization rotation in a
hexaferrite with long-wavelength magnetic structures. Physical Review Letters 94
(13): 137201.
Kiselev, S.V., Ozerov, R.P., and Zhdanov, G.S. (1963). Detection of magnetic order in
ferroelectric BiFeO3 by neutron diffraction. Soviet Physics – Doklady 7 (8):
742–744.
Krainik, N.N., Khuchua, N.P., Zhdanova, V.V., and Evseev, V.A. (1966). Phase
transition in BiFeO3 . Soviet Physics Solid State 8: 654.
Kretschmer, R. and Binder, K. (1979). Surface effects on phase transitions in
ferroelectrics and dipolar magnets. Physical Review B 20 (3): 1065–1076.
Landau, L.D. and Lifshitz, E.M. (1960). Electrodynamics of Continuous Media, 1e.
Pergamon Press.
Lee, H.N., Christen, H.M., Chisholm, M.F. et al. (2005). Strong polarization
enhancement in asymmetric three-component ferroelectric superlattices. Nature
433: 395. Available at: http://dx.doi.org/10.1038/nature03261.
Lee, S., Pirogov, A., Park, J.G. et al. (2008). Giant magneto-elastic coupling in
multiferroic hexagonal manganites. Nature 451: 805. Available at: http://dx.doi
.org/10.1038/nature06507.
Lin, Y., Cai, N., Nan, C.W. et al. (2005). Giant magnetoelectric effect in multiferroic
laminated composites. Physical Review B 72 (1): 012405. Available at: https://link
.aps.org/doi/10.1103/PhysRevB.72.012405.
Liu, G., Nan, C.W., Xu, Z.K., and Chen, H. (2005). Coupling interaction in
multiferroic BaTiO3 -CoFe2 O4 nanostructures. Journal of Physics D: Applied
Physics 38 (14): 2321. Available at: http://stacks.iop.org/0022-3727/38/i=14/
a=005.
Mathur, N. (2008). Materials science: a desirable wind up. Nature 454: 591. Available
at: http://dx.doi.org/10.1038/454591a.
Nan, C.W., Li, M., and Huang, J.H. (2001). Calculations of giant magnetoelectric
effects in ferroic composites of rare-earth–iron alloys and ferroelectric polymers.
Physical Review B 63 (14): 144415. Available at: https://link.aps.org/doi/10.1103/
PhysRevB.63.144415.
Nan, C.-W., Liu, G., Lin, Y., and Chen, H. (2005). Magnetic-field-induced electric
polarization in multiferroic nanostructures. Physical Review Letters 94 (19):
197203. Available at: https://link.aps.org/doi/10.1103/PhysRevLett.94.197203.
Roginskaya, Y.E., Tomashpol’Skiı̌, Y.Y., Venevtsev, Y.N. et al. (1966). The nature of
the dielectric and magnetic properties of BiFeO3 . Soviet Physics Journal of
Experimental and Theoretical Physics 23: 47.
Schmid, H. (1994). Multi-ferroic magnetoelectrics. Ferroelectrics 162 (1): 317–338.
Available at: https://doi.org/10.1080/00150199408245120.
Seidel, J., Martin, L.W., He, Q. et al. (2009). Conduction at domain walls in oxide
multiferroics. Nature Materials 8: 229. Available at: http://dx.doi.org/10.1038/
nmat2373.
Siratori, K., Kohn, K., and Kita, E. (1992). Magnetoelectric effect in magnetic
materials. Acta Physica Polonica A 81 (4–5): 431–466.
Smolenskii, G.A. and Chupis, I.E. (1982). Ferroelectromagnets. Soviet Physics
Uspekhi 25 (7): 475–493. Available at: https://ufn.ru/en/articles/1982/7/b/.
Srinivasan, G., Rasmussen, E.T., Gallegos, J. et al. (2001). Magnetoelectric bilayer
and multilayer structures of magnetostrictive and piezoelectric oxides. Physical
Review B 64 (21): 214408. Available at: https://link.aps.org/doi/10.1103/PhysRevB
.64.214408.
Srinivasan, G., Rasmussen, E.T., and Hayes, R. (2003). Magnetoelectric effects in
ferrite-lead zirconate titanate layered composites: the influence of zinc
substitution in ferrites. Physical Review B 67 (1): 014418. Available at: https://link
.aps.org/doi/10.1103/PhysRevB.67.014418.
Van den Boomgaard, J., Terrell, D.R., Born, R.A.J., and Giller, H.F.J.I. (1974). An in
situ grown eutectic magnetoelectric composite material: Part I Composition and
unidirectional solidification. Journal of Materials Science 9 (10): 1705–1709.
Available at: https://doi.org/10.1007/BF00540770.
vanSuchtelen, J. (1972). Product properties: a new application of composite
materials. Philips Research Reports 27 (1): 28–37.
Wang, J., Neaton, J.B., Ramesh, R. et al. (2003). Epitaxial BiFeO3 multiferroic thin
film heterostructures. Science 299 (5613): 1719–1722.
Wong, C.K. and Shin, F.G. (2007). A simplified treatment of the Landau theory of
phase transitions for thin ferroelectric films. American Journal of Physics 76 (1):
Wu, J., Fan, Z., Wang, J. et al. (2016). Multiferroic bismuth ferrite-based materials
for multifunctional applications: ceramic bulks, thin films and nanostructures.
Progress in Materials Science 84: 335–402. Available at: http://www.sciencedirect
.com/science/article/pii/S0079642516300585.
Xu, Y. (1991). Perovskite-type ferroelectrics: Part I. In: Ferroelectric Materials and
their Applications (ed. Y. Xu) In Y. B. T.-F. M. and their A. Xu, ed., 101–162.
North-Holland, Amsterdam: Elsevier. Available at: http://www.sciencedirect
.com/science/article/pii/B9780444883544500085.
Zavaliche, F., Yang, S.Y., Ramesh, R. et al. (2006). Multiferroic BiFeO3 films: domain
structure and polarization dynamics. Phase Transitions 79 (12): 991–1017.
Available at: https://doi.org/10.1080/01411590601067144.
References 215
Zavaliche, F., Zhao, T., Zheng, H. et al. (2007). Electrically assisted magnetic
recording in multiferroic nanostructures. Nano Letters 7 (6): 1586–1590.
Available at: https://doi.org/10.1021/nl070465o.
Zhai, J., Dong, S., Xing, Z. et al. (2006). Giant magnetoelectric effect in
metglas/polyvinylidene-fluoride laminates. Applied Physics Letters 89 (8): 083507.
Zhang, J.X., Dai, J.Y., and Chan, H.L.W. (2010). Interfacial engineering and coupling
of electric and magnetic properties in Pb(Zr0.53Ti0.47)O3 /CoFe2 O4 multiferroic
epitaxial multilayers. Journal of Applied Physics 107 (10): 104105. Available at:
https://doi.org/10.1063/1.3386510.
Zhao, T., Scholl, A., Ramesh, R. et al. (2006). Electrical control of antiferromagnetic
domains in multiferroic BiFeO3 films at room temperature. Nature Materials 5:
823–829. Available at: http://dx.doi.org/10.1038/nmat1731.
Zheng, H., Wang, J., Lofland, S.E. et al. (2004). Multiferroic BaTiO3 -CoFe2 O4
nanostructures. Science 303 (5658): 661 LP–663. Available at: http://science
.sciencemag.org/content/303/5658/661.abstract.
Zhong, W.L., Wang, Y.G., Zhang, P.L., and Qu, B.D. (1994). Phenomenological study
of the size effect on phase transitions in ferroelectric particles. Physical Review B
50 (2): 698–703. Available at: https://link.aps.org/doi/10.1103/PhysRevB.50.698.
217
10
Device Application of Multiferroics
Multiferroic materials exhibit coupling between electric polarization and

electron spin, making them applicable in devices where magnetic switching is
being controlled by electric field or vice versa. Composite multiferroic materials
based on the mechanical coupling between piezoelectric and magnetostrictive
effects have found engineering application in devices of highly sensitive magnetic
field sensors. Multistate and multifunctional memories that can be magnetically
written and electrically read or vice versa are also very promising in data storage,
computing, and AI technology. These will be particularly introduced in this
chapter.
10.1 ME Composite Devices

Magnetoelectric (ME) effect is the response of polarization P to a magnetic
field H applied across a ME device. Due to the limitation of single phase
ME material and very low ME coupling factor, scientists have spent a lot of
effort trying to realize device-applicable composite materials, utilizing the
mechanical coupling between the magnetostrictive and piezoelectric effects.
The conversion from magnetic field to electrical signal in ME device is realized
by magneto–elasto–electric coupling through stress interaction between the
magnetostrictive phase and the piezoelectric phase. Figure 10.1 illustrates
different modes of ME composite device, where L represents longitudinal
vibration mode, T represents transverse vibration mode, and C represents
circular vibration mode. During designing of the device, the piezoelectric
coefficient, electromechanical coupling factor, and magnetostrictive effect to
the corresponding variation modes should be considered. The integrated device
resonance frequency is related to the size of each component and their coupling
as well as the preload and therefore is a complicated problem that could be
calculated by finite element modeling.
Similar to a piezoelectric resonance whose performance can be described by
electrical equivalent circuit, an ME device can also be electrically described by
equivalent circuit. Figure 10.2 shows equivalent circuit of such ME device under
resonance state. In this circuit, the 𝜑m is the magnetoelastic coupling factor, and
𝜑p is the elastic–electric coupling factor. With the input magnetic field H and the
induced electrical voltage V , the ME coupling factor can be derived.

218 10 Device Application of Multiferroics
Piezoelectric layer Figure 10.1 Various laminates

M M operated in longitudinal vibration
P P mode: L-longitudinal, T-transverse,
M M
and C-circular. Source: Adapted from
L–L T–L Nan et al. (2008).
Magnetostrictive layer
M M
P P
M M
L–T T–T
M
P M P
M
Push–pull L–T hybrid
M P
Multi-L–T C–C
Mechanical current –C0 Electric current

Applied
magnetic field H
C0
φ mH Induced voltage
V
Zm Lm Cm φp : 1
Magnetoelastic Elastic–electric
coupling φm coupling φp
Figure 10.2 Magnetoelastic–electric equivalent circuit at resonance. Source: Adapted from

Nan et al. (2008) and Dong et al. (2006).
We can design a ME device based on the working principle of L–T mode

laminated composite as shown in Figure 10.3. In this device, Terfenol-D fiber
mainly vibrates along the length direction when the magnetic field is applied
longitudinally. The piezoelectric material vibrates along the length direction
due to mechanical coupling, and electric charges are generated across the
thickness direction via the d31 piezoelectric effect. To enhance the ME sen-
sitivity, epoxy-bonded Terfenol-D continuous fiber composite is used as the
magnetostrictive phase instead of using monolithic Terfenol-D. Composite form
possesses several advantages in comparison with monolithic form, for example,
higher operation bandwidth, lower material cost, and less brittle. Superior ME
coupling generates strong electrical response by external alternating magnetic
field especially when the composites are at resonance giving rise to high magnetic
field sensitivity. Besides, the composite structure can greatly reduce the eddy
current loss that generally occurs in monolithic Terfenol-D.
Terfenol-D continuous fibers Epoxy
M
Piezoelectric plate
P
M
Figure 10.3 Schematic diagram of the ME laminate composite device with Terfenol-D
continuous fiber composite and piezoelectric plate. The arrows represent the magnetization
and polarization directions, respectively.
Multi-purpose FFT analyzer

Signal-out Channel 2 Channel 1
Induced ME voltage (VME)
Swept sinusoidal voltage
Measuring Hac
ME laminate
DC-coupled AC amplifier composite
Signal-in Signal-out
Driving signal for Hac

Helmholtz coil
Electromagnets with
driving signal for Hbias
Figure 10.4 Schematic diagram of an automated ME measurement system.
Standard ME properties using an automated measurement system are shown

in Figure 10.4. The ME voltages (V ME ) generated from the composite can be mea-
sured by applying ac magnetic fields (H ac ) on the composite under a bias magnetic
field (H bias ). The sensitivity of the ME laminated composite/magnetoelectric volt-
age coefficient (MEv ) can be measured by applying H ac of 1 Oe on the composite
at frequency f . The MEv at specific H ac and f is usually calculated as follows:
VME
MEv = (10.1)
Hac
where the H ac along the longitudinal direction is provided by a Helmholtz coil.
H ac and H bias were measured by a pick-up coil and a Gauss/Tesla meter.
To work at resonance frequency is important to reach maximum sensitivity.
Figure 10.5a shows the spectrum of Z and 𝜃 for the ME laminated composite
where the resonance frequency (f r ) of the composite device is 89 kHz, which can
5
10 80
40 Hbias = 20 Oe
Impedance Hbias = 100 Oe
Phase angle 40 Hbias = 300 Oe
(ME)v (V/cmOe)
30 Hbias = 500 Oe
4
10
Hbias = 700 Oe
Z (Ω)
θ (°)
20
10
3
−40
10
−80
0
2
10
20.0k 40.0k 60.0k 80.0k 100.0k 120.0k 140.0k 0 20 40 60 80 100
(a) Frequency (Hz) (b) Frequency (Hz)
Figure 10.5 (a) Electrical impedance (Z) and phase angle (𝜃) spectra for a ME laminated
composite. (b) Magnetoelectric voltage coefficient (ME v ) as a function of frequency f for a ME
laminated composite tested under different dc bias magnetic field. Source: Adapted from
Cheung (2009).
2000 Hbias = 0 Oe
Hbias = 200 Oe
Hbias = 500 Oe
1600
Hbias = 800 Oe
Hbias = 1200 Oe
1200
V3 (mV)
Hbias = 1600 Oe
Hbias = 2000 Oe
800
400
0
0 2 4 6 8 10
H3 (Oe)
Figure 10.6 V ME (V 3 is the voltage across the piezoelectric plate) as a function of Hac with
f = 1 kHz for the ME laminated composite tested under different dc bias magnetic fields.
Source: Adapted from Cheung (2009).
also be identified in Figure 10.5b, but showing right shift under increased bias
field.
Figure 10.6 shows the V ME as a function of H ac with f = 1 kHz for the ME
laminated composite tested under different dc bias magnetic fields. It is appar-
ent that V ME shows excellent linear relationship with the measured range of H ac
under H bias . The linear response to H ac makes the ME laminated composite very
suitable for the application in ac magnetic field measurement.
During the operation of ME device, the magnetic bias field plays a very
important role that affects the sensitivity of the device. Figure 10.7 shows the
optimum H bias for the laminated composite device, where one can see that the
MEv increases with increasing H bias , but it finally reaches a maximum value
at H bias = 400 Oe and then starts to decrease from the maximum. This can be
understood by considering the fact that without bias field, the ac magnetic field
3.5
@ 1 kHz
3.0
ME coefficient (αE) (V/cmOe)
2.5
2.0
1.5
1.0
0.5
0.0
0 200 400 600 800 1000 1200
Hbias (Oe)
Figure 10.7 Magnetoelectric voltage coefficient ME V as a function of dc bias magnetic field

(Hbias ) for a ME laminated composite tested under an ac magnetic field (Hac ) of 1 Oe peak with
frequency f of 1 kHz. Source: Wang et al. (2008). Reprinted with permission of AIP Publishing.
is hard to induce magnetization switching, and therefore the strain generated

by magnetostrictive effect is very small. Under optimized bias field, a small ac
magnetic field can easily induce non-180∘ magnetization rotation and therefore
can generate largest strain in the magnetostrictive laminate. However, a too
large bias field makes the magnetization switching induced by ac field hard, and
therefore, the output strain becomes smaller.
10.1.1 Effect of Preload Stress

In Chapter 7, it is explained that non-180∘ domain rotation is essential to gen-
erate large strain in Terfenol-D, and this can be achieved by adding H bias and
prestress to the Terfenol-D rod. Figure 10.8 shows a ME device illustrating how
to make a ME device with presence of dc magnetic field and preloaded mechani-
cal stress resulting in optimized performance. In this device, a dc magnetic field is
applied by a permanent magnet with the field parallel to the [112] direction of the
Terfenol-D rod. A preloaded mechanical stress is applied by a screw to make the
Terfenol-D and Pb(Mg1/3 Nb2/3 )O3 –xPbTiO3 (PMN-PT) crystal being clamped.
The measured ME response to an ac magnetic field as a function of frequency
and under different level of preload mechanical stress is shown in Figure 10.9,
where two general trends can be found. One is that the resonance frequency
increases as the stress increases; this can be understood by considering that the
stress results in an increase of mechanical compliance. The other trend is that the
ME response increases as the stress increases; this is due to the fact that the stress
applied on Terfenol-D makes the magnetization direction perpendicular to the
rod, and non-180∘ magnetization switching is realized and induces the maximum
ME response. However, overstressed rod will result in fully clamped state so that
22 mm
Stainless frame
Permanent PMN-PT PMN-PT

Terfenol-D Bolt
magnet single crystal single crystal
Figure 10.8 Schematic diagram of a stress-biased PMN-PT single crystal/Terfenol-D ME sensor.

Source: Adapted from Lam et al. (2011).
0 20 40 60 80 100
0.25 0.25
No pressure
0.5 MPa
1.0 MPa
0.20 1.5 MPa 0.20
2.0 MPa
2.5 MPa
ME voltage coefficient (V/Oe)
3.0 MPa
0.15 0.15
0.10 0.10
0.05 0.05
0.00 0.00
0 20 40 60 80 100
Frequency (kHz)
Figure 10.9 Frequency dependence of ME v as a function of preloading stress for a PMN-PT

single crystal/Terfenol-D ME sensor. Source: Lam et al. (2011). Reprinted with permission of AIP
Publishing.
10.2 Memory Devices Based on Multiferroic Thin Films 223
no domain switching can happen; therefore the ME factor decreases significantly

after the peak value.
This type of device can find application in magnetic field sensor. Luo’s group
has increased the sensitivity of magnetic field measurement to 10−10 T at room
temperature.
10.2 Memory Devices Based on Multiferroic Thin Films

Multiferroic thin films can also be used for multifunction and/or multistate mem-
ory devices. For example, as shown in Figure 10.10, a memory device can be
formed based on the coupling between the BiFeO3 (BFO)’s ferroelectricity and
anti-ferromagnetism (FE-AFM) and exchange-bias coupling between the BFO
film and magnetic electrode. BFO is one of the most attractive multiferroic mate-
rials with excellent ferroelectric property and also anti-ferromagnetic and weak
ferromagnetic properties. These properties provide BFO thin film great potential
for memory device applications. Beyond the well-known application in ferroelec-
tric random-access memory (FeRAM) as ferroelectric gate dielectric layer in a
field effect transistor (FET) structure, BFO can also be used as magneto-electric
random access memory (MERAM) based on a cross-bar structure. As shown in
Figure 10.10, the coupling between the BFO’s FE-AFM makes BFO magnetization
moment switchable by electric field. Through exchange bias with a thin ferromag-
netic electrode, a spin valve structure with a tunnel barrier layer between the two
top ferromagnetic layers provides the two resistive states. Interestingly, BFO can
Resistance
Rap
V+
P FE – AFM
Electrode
Rp
V–
P FE – AFM
Electrode
Voltage
Figure 10.10 MERAM based on exchange-bias coupling between a multiferroic that is

ferroelectric and antiferromagnetic (FE-AFM) and a thin ferromagnetic electrode. Source: Bibes
and Barthélémy (2008). Reprinted with permission of Springer Nature.
60 nm
1
δ V (μV)
Output signal 0
–1
3
δ H (Oe)
Read head
CFO
STO
BTO
Substrate
FM layer
FE layer
ME sensor 0
–3
δH dM
S dt
N S
N N S
S N
S N
N
S Iw
NS
S N
SN
N bK 0 0 1 1 0 1 1 1 1 1 0 1 1 0 1 0
Recording medium S aK 0 0 1 1 1 0 1 0 0 0 1 0 1 1 0 0
Figure 10.11 Schematic, TEM image (inset) and voltage output of an ME read head sensor
based on a bilayer CFO/BTO heterostructure grown on a single-crystal SrTiO3 (STO) substrate.
Source: Zhang et al. (2008). Reprinted with permission of AIP Publishing.
act not only as the magnetoelectric active layer but also a ferroelectric tunneling
barrier.
Another method, which has already been implemented in recording read
head application, is a junction working with magnetoelectric (ME) coupling. By
using strain as a media, in-plane magnetic field-induced strain in the magnetic
component due to magnetostrictive effect can be mechanically transferred
to the ferroelectric component and finally can induce an electric polarization
change, and thus voltage change, through the piezoelectric effect. Figure 10.11
demonstrates the ME coupling between the magnetostrictive CoFe2 O4 (CFO)
and ferroelectric BaTiO3 (BTO) (Zhang et al. 2008).
10.3 Memory Devices Based on Multiferroic Tunneling

In Chapter 3, ferroelectric tunneling device as a non-volatile memory has been
introduced. Beyond this ferroelectric junction, if ferromagnetic electrodes are
implemented, memory device based on multiferroic tunnel junction with mul-
tistates and multifunctions can be realized (Gajek et al. 2007; Eerenstein et al.
2006).
A multiferroic tunnel junction is made by combining the structure of ultrathin
ferroelectric tunnel layer and magnetic tunnel junction into one junction with
an ultrathin ferroelectric film sandwiched by two ferromagnetic conductive
electrodes. With this artificial multiferroic tunnel junction (MFTJ), four resis-
tance states can be retrieved by the result of tunneling magnetoresistance
(TMR) and tunneling electroresistance (TER). As a result, multifunctional
(magnetically write and electrically read) four-state memory device can be
achieved. Figure 10.12 is a schematic illustration of an artificial multiferroic
tunnel junction and its working principle. One can see that the combinations
of two ferroelectric polarizations and two ferromagnetic states result in four
resistance states.
FM NM FM
I FE FE
FM NM FM
MTJ FTJ MFTJ
R R R
E
H
H-field E-field
Hc Ec Ec
Hc
Figure 10.12 Schematic illustration of artificial multiferroic tunnel junctions and its working
principle.
Figure 10.13 Schematic Ferromagnetic

diagram of low-field MFTJ. alloy NiFe
M
P
Ferromagnetic Ferroelectric
M barrier BTO
oxide LSMO
Substrate
STO
Aiming at constructing a MFTJ that can fulfill the constraint of less power
consumption, low fields in both ferroelectric and ferromagnetic switching is pre-
ferred. As an example of our work, Figure 10.13 shows a ferroelectric BTO tun-
nel barrier sandwiched by two electrodes composed of different ferromagnetic
materials La0.7 Sr0.3 MnO3 (LSMO) and Permalloy Ni0.81 Fe0.19 (NiFe). NiFe is a
well-known alloy that allows strong spin electron injection and has relatively
much lower coercive field, while LSMO has been predicted to have close to 100%
spin polarization. This structure is expected to form a low-switching-field spin
valve memory.
Figure 10.14 shows resistance–voltage (R–V ) characteristics of non-volatile
resistance switching from this MFTJ device. Unlike RRAM-resistive ran-
dom access memory (ReRAM), no electroforming at higher voltage is
needed before the observation of hysteretic resistance changes. During the
measurements, pulse train writing voltages are applied in the sequence of
0 V → 1.5 V → 0 V → −1.5 V → 0 V with a step of 0.2 V, while reading voltage is
remained constant as 0.1 V. With the top electrode grounded, a positive pulse of
80
70
60
50
Resistance (kΩ)
40
30
20
10
–1.5 –1.0 –0.5 0.0 0.5 1.0 1.5

Voltage (V)
Figure 10.14 Electroresistance change of MFTJ at 40 K for 10 cycles. The corresponding

ferroelectric polarization direction are shown in the bottom right and top right insets for the
ON and OFF states, respectively. Source: Adapted from Yau et al. (2015).
voltage (+V ) is applied on the bottom LSMO electrode, which drives the BaTiO3
ferroelectric polarization pointing to the NiFe electrode and sets the junction to
ON state (low resistance state). In contrast to the ON state, a negative voltage
pulse (−V ) drives the polarization to the opposite direction and sets the junction
to OFF state (high resistance state).
Magnetic characteristics of the NiFe and LSMO electrodes are shown in
Figure 10.15, where two distinct M–H hysteresis loops present in both layers.
NiFe has almost the same magnetization as LSMO; however, the coercive field
of LSMO is 20 Oe, while it is only 5.0 Oe for NiFe in the film plane. The different
coercive fields for these two dissimilar electrodes determine that LSMO acts as
a hard magnet and NiFe electrode acts as a soft magnet in the spin valve. This
result suggests that two resistance states can be achieved by interchanging the
magnetization orientations using external magnetic field.
Figure 10.16 demonstrates the coexistence of TMR and TER in an artificial
MFTJ. By placing the sample in the middle of two electromagnets, alternative
magnetic field is applied to the junction parallel to the film plane. Then, the junc-
tion (ferroelectric barrier) was poled by 1.5 V and −1.5 V for poling upward and
downward, respectively. Among the four stages, two resistance states are formed
due to the electric polarization switching and the other two are due to the parallel
and antiparallel magnetization configurations.
By sweeping the external magnetic field from −100 Oe to 100 Oe and returning
to −100 Oe, two distinct resistance states can be clearly observed for the two
ferroelectric polarization states of up and down, i.e. there are four distinct
800
LSMO
600 NiFe
Magnetization (μemu/cm2)
400
200
–200
–400
–600
–800
–100 –50 0 50 100

Magnetic field (Oe)
Figure 10.15 Magnetism of NiFe and LSMO showing the hysteresis loops. Source: Adapted
from Yau et al. (2015).
resistance states. When −1.5 V is added, ferroelectric polarization first orients

toward LSMO so that the OFF state is reached. While when a −100 Oe magnetic
field is applied, the magnetic moments of both NiFe and LSMO are aligned
in parallel, and the resultant resistance is obtained as indicated by Stage A in
Figure 10.16. Resistance keeps constant when magnetic field changes from
−100 Oe to 5.0 Oe because of the unchanged alignment in two ferromagnetic
layers. Until the field is larger than 5.0 Oe, the soft magnet NiFe starts to switch,
thus, resistance changes due to antiparallel alignment of magnetic moment
between NiFe and LSMO (Stage B). When the field is further increased to
a certain value that is large enough to switch the hard magnet LSMO (i.e.
20 Oe), orientations of magnetization for two ferromagnetic layers will be
aligned in parallel again. As a result, resistance returns to its initial value as
Stage C. Stage D is obtained when an opposite sweeping field is applied and
NiFe is oppositely switched. From the flow chart illustrated in Figure 10.16b,
the configurations of magnetization alignment correlated to magnetic field are
explained.
It is interesting to notice that when the electric polarization is facing down, the
negative TMR exists for the antiparallel magnetization alignment. This can be
explained by the fact that the ferroelectric polarization switching may invert the
spin polarization at the interface (DeTeresa et al. 1999). Since LSMO is p-type
semiconducting, the facing up electric polarization inverts the interfacial layer of
LSMO to n-type. This will change the sign of the spin polarization at the interface,
and then the sign of TMR of the junction may be changed.
175.4
D B
175.2
175.0 A C
Resistance (kΩ)
174.8
174.6
174.4
10.54
A′ C′
10.53 D′ B′
10.52
–100 –50 0 50 100
(a) Magnetic field (Oe)
A B C D
A′ B′ C′ D′
(b)
Figure 10.16 (a) Resistance changes with different magnetic fields at 8 K showing four-state
memory tested on the NiFe/BaTiO3 /LSMO structure. (b) The corresponding schematic diagram
for magnetization and ferroelectric polarization orientations. Source: Adapted from Yau et al.
(2015).
To specify, TMR is expressed in Eq. (10.1) by using Julliere’s model (Julliere

1975).
RAP − RP 2PLSMO PNiFe
= (10.2)
RP 1 − PLSMO PNiFe
where RAP is the resistance in antiparallel magnetization state, RP in parallel mag-
netization state, and PLSMO and PNiFe are the spin polarizations of the two elec-
trodes, where the spin polarization P can be calculated from the spin-dependent
References 229
density of states N at the Fermi energy (refer to Figure 8.3):

[N ↑ (EF ) − N ↓ (EF )]
P=
[N ↑ (EF ) + N ↓ (EF )]
LSMO has been tested to be almost 100% positive spin polarization, while the
spin polarizations of most ferromagnetic metals such as Co and Ni are negative.
Therefore, the TMR of the junction built with NiFe and LSMO should be negative,
i.e. the tunneling resistance decreases when the spin polarizations are antiparal-
lel in NiFe and LSMO. Thus, negative TMR can be observed (B′ and D′ in the
Figure 10.16), and it is reversible when the ferroelectric polarization is reversed
(B and D in the Figure 10.16).
References
Bibes, M. and Barthélémy, A. (2008). Multiferroics: towards a magnetoelectric
memory. Nature Materials 7 (6): 425–426. Available at: http://www.ncbi.nlm.nih
.gov/pubmed/18497843%5Cnhttp://www.nature.com/doifinder/10.1038/
nmat2189.
Cheung, K. (2009). Ultrasonic Transducer Equipped with a magnetoelectric sensor
for weld quality monitoring. Hong Kong Polytechnic University. Available at:
https://theses.lib.polyu.edu.hk/handle/200/3843.
DeTeresa, J.M., Barthélémy, A., Fert, A. et al. (1999). Role of metal-oxide interface in
determining the spin polarization of magnetic tunnel junctions. Science 286
(5439): 507–509. Available at: http://www.sciencemag.org/cgi/doi/10.1126/
science.286.5439.507.
Dong, S., Li, J.-F., and Viehland, D. (2006). Magnetoelectric coupling, efficiency, and
voltage gain effect in piezoelectric-piezomagnetic laminate composites. Journal of
Materials Science 41 (1): 97–106. Available at: https://doi.org/10.1007/s10853-
005-5930-8.
.ncbi.nlm.nih.gov/pubmed/16915279.
Gajek, M., Bibes, M., Fusil, S. et al. (2007). Tunnel junctions with multiferroic
barriers. Nature Materials 6 (4): 296–302. Available at: http://www.ncbi.nlm.nih
.gov/pubmed/17351615.
Julliere, M. (1975). Tunneling between ferromagnetic films. Physics Letters A 54 (3):
225–226.
Lam, K.H., Lo, C.Y., Dai, J.Y. et al. (2011). Enhanced magnetoelectric effect in a
stress-biased lead magnesium niobate-lead titanate single crystal/Terfenol-D
alloy magnetoelectric sensor. Journal of Applied Physics 109 (2): 024505. Available
at: https://doi.org/10.1063/1.3536636.
Nan, C.-W., Bichurin, M.I., Dong, S. et al. (2008). Multiferroic magnetoelectric
composites: historical perspective, status, and future directions. Journal of
Applied Physics 103 (3): 031101. Available at: https://doi.org/10.1063/1.2836410.
Wang, Y., Or, S.W., Chan, H.L.W. et al. (2008). Enhanced magnetoelectric effect in
longitudinal-transverse mode Terfenol-D/Pb(Mg1/3 Nb2/3 )O3 –PbTiO3 laminate
composites with optimal crystal cut. Journal of Applied Physics 103 (12): 124511.
Yau, H.M., Yan, Z.B., Dai, J.Y. et al. (2015). Low-field switching four-state nonvolatile
memory based on multiferroic tunnel junctions. Scientific Reports 5: 12826.
Available at: http://www.nature.com/doifinder/10.1038/srep12826.
Zhang, Y., Li, Z., Nan, C.W. et al. (2008). Demonstration of magnetoelectric read
head of multiferroic heterostructures. Applied Physics Letters 92 (15): 152510.
231
11
Ferroelasticity and Shape Memory Alloy
Shape memory alloys (SMAs), or generally being called shape memory materials
(SMMs) since some polymers also have shape memory characteristics, can be
defined as a class of materials having the ability to “memorize” or retain their
previous shape in response to stimulus such as thermal and mechanical or even
magnetic variations. To be a member in the family of “ferroic,” SMA is charac-
terized as ferroelastic exhibiting hysteresis loop in the curves of strain–stress,
strain–temperature, or strain–magnetic field. Analogous to the other two family
members of ferromagnetic and ferroelectric, ferroelastic materials also present
domain structures, but instead of electrical and magnetic orderings, the domains
are segregated regions of martensite variants. Not only SMA and magnetic SMAs
but also many ferroelectric and ferromagnetic materials are ferroelastic; however,
the desired devices as introduced in previous chapters are mainly related to ferro-
electric and ferromagnetic switching. Many metal oxide materials such as SrTiO3
and LaAlO3 also exhibit ferroelasticity, but the range is relative too small for
engineering application. One of the first full ferroelastic hysteresis loops (Figure
11.1) was seen in 1976 for the prototypic material Pb3 (PO4 )2 (Salje 1993; Salje and
Hoppmann 1976) and was introduced in the review of ferroelastics (Salje 2012).
11.1 Shape Memory Alloy

SMAs have drawn significant attention and interest since the end of last
century in a broad range of engineering applications such as automotive,
aerospace, robotic, and biomedical devices as actuators. The shape mem-
ory (SM) process belongs to solid-state displacive phase transition between
austenite and martensite phases and can generate strains up to 10%, mak-
ing it a fascinating candidate for actuators or even sensors in a smart
system. There are a number of materials, including alloys composited of
nickel–titanium (NiTi), copper–aluminum–nickel (CuAlNi), copper–zinc–
aluminum (CuZnAl), and iron–manganese–silicon (FeMnSi), exhibiting shape
memory effect. This chapter will introduce the fundamentals, device applica-
tions, and modeling of SMA as well as summary of recent advances and new
application opportunities.

232 11 Ferroelasticity and Shape Memory Alloy
Pb b
P Strain (10–4)
c
Stress (bars)
–4 –3 –2 –1 1 2 3 4
Figure 11.1 Ferroelastic hysteresis and atomic switching in Pb3 (PO4 )2 . The structural
mechanism for ferroelasticity originates from the shift of the Pb atom away from the center of
its oxygen cage toward one pair of oxygen ions. This leads to a monoclinic distortion of the
crystal structure that can be inverted or rotated (in the direction of the arrows) under external
stress. Source: Reprinted with permission from Salje (2012).
11.1.1 SMA Phase Change Mechanism

The ferroelastic characteristic of a SMA is the consequence of phase transition
from a high-symmetry austenite phase (cubic structure) to a low-symmetry
martensite phase (a distorted cubic structure). This phase transformation can be
induced by heat, stress, or electromagnetic fields (for example magneto SMA).
Nitinol (NiTi alloy) is the most typical and commercially used SMA that has
superior structure and memory capabilities originated from its austenite phase
(Figure 11.2a) and martensite phase (Figure 11.2b) at high and low temperatures,
respectively. It should be noticed that the volume of a primitive cell (unit cell)
of the martensite structure is double of the austenite structural unit cell; to
specify,
√ the lattice constants b and c in the martensite structure are close to
2a, where a is the lattice constant of the pseudocubic unit cell of austenite
structure.
To illustrate the nature of phase transformations, we consider the austenitic
NiTi lattice cell depicted in Figure 11.3. In this cubic unit cell (where Ni ions
occupy the A site and Ti the B site), we can identify six {110} atomic planes, and
in response to an applied stress, each of the six face-diagonal planes can shear or
shift into four variants. This yields 24 possible variants making the low temper-
ature martensite phase easy to change shape under mechanical stress. All these
martensite variants will change back to austenite cubic structure when temper-
ature increases, and as a consequence, the shape in high-temperature phase is
resumed. This is the working mechanism of SMAs.
Ni
Ti
(a) (b)
Figure 11.2 (a) Austenite phase (B2) and (b) martensite lattice structure (B19’) of
nickel–titanium, where the plane outlined by dash line represents the {100} planes in the
primitive cell of the martensite structure.
Figure 11.3 Lattice cell of NiTi and four martensite variants developed from one face-diagonal
plane.
The phase transformation between austenite and martensite phases is a

first-order transition that can be dictated by the differential scanning calorime-
ter (DSC) measurement as shown in Figure 11.4. DSC is a thermoanalytical
technique in which the difference in the amount of heat flow required to increase
the temperature of a sample and a reference is measured as the function of
temperature. Both the sample and reference are maintained at nearly the same
temperature throughout the experiment. This DSC method shows the amount
of heat released or absorbed by a sample of the alloy as it is cooled or heated
through its phase transformations. The extra amount of heat during the phase
transformation is called latent heat, and the phase change involving latent heat
is classified as first-order phase transition.
11.1.2 Nonlinearity in SMA

The shape memory strain response of nitinol is highly nonlinear as illustrated
in Figure 11.5, where the length is the spontaneous length change during phase
transition. The control of mechanical properties is through the manipulation of
phase transition at temperatures of Ms , Mf , As , and Af , where Ms and Mf are the
martensite start and finish temperatures and As and Af are the austenite start and
finish temperatures.
Cooling
Heat flow
Mf Ms
As Af
Heating
Temperature
Figure 11.4 A schematic diagram of a typical differential scanning calorimeter (DSC) curve for
NiTi shape memory alloy showing presence of latent heat during phase transformation.
Mf As
100
Martensite (%)
Length
ΔL
0
Ms Af
Temperature
Figure 11.5 Typical transformation–temperature curve of NiTi SMA under constant stress as it
is cooled and heated showing hysteresis.
The complete shape memory process showing strain responses to tempera-

ture and stress is illustrated in Figure 11.6. The combination of temperature and
stress-induced phase transformations provides SMAs with their memory capa-
bilities. One can see that at low-temperature martensite phase, large strain is
developed under stress and there is a residual strain from detwinned structure
when stress is released. When the detwinned structure is heated above Af , the
residual strain is recovered and the alloy returns to its original shape, i.e. the shape
memory effect.
Stress
Stress-induced strain
Mf
Strain
Cooling
Af
Heating
Temperature
Figure 11.6 Shape memory effect in a uniaxial SMA.
11.1.3 One-Way and Two-Way Shape Memory Effect

The shape memory effect illustrated in Figure 11.6 is actually a one-way shape
memory effect since there are only two different shapes (the one at higher tem-
perature can be memorized) and the shape change is realized only by stress (or
called loading) at lower temperature, while the original shape at higher temper-
ature will be resumed upon heating. This can be demonstrated more clearly in
Figure 11.7a. One can notice that for the one-way shape memory effect, cooling
from high temperatures does not cause a macroscopic shape change, even there
is an austenite to martensite phase transition.
The so-called two-way shape memory effect is the effect that the material
remembers two different shapes: one at low temperatures and the other one
at high temperatures, i.e. without involving any mechanical stress; these two
different shapes can change from one to the other only by heating and cooling.
As illustrated in Figure 11.7b, the reason to cause this two-way shape memory
effect is due to the mechanical stress-induced plastic deformation (crystal defect)
during the training process.
The two-way shape memory effect should be considered during device appli-
cation. A SMA should be trained before usage, during which overloading results
in accumulated plastic deformation in the martensite structure and even in the
parent austenite structure due to the generation of crystal defect such as staking
dislocations. This property gives rise to a spontaneous shape change of SMA in
both heating and cooling cycles. Therefore, one can see that the two-way shape
memory effect refers to the reversible and spontaneous shape change of SMA
during thermal cycling.
11.1.4 Superelastic Effect (SE)

Besides the shape memory effect, the SMAs also have a property known as
superelastic effect (SE), making SMA different from the other metal alloys. The
He
ati
ng
Stress/load
Loading
Cooling
Heating
(a) Temperature
He
ati
ng
Stress/load
Loading
Cooling
Heating
(b) Temperature
Figure 11.7 Illustration of (a) one-way and (b) two-way memory effects.
SE of Ni–Ti alloys is associated with the recovery of the deformation during

unloading. Figure 11.8 illustrates the superelastic behavior during loading and
unloading above Af associated with stress-induced martensitic transformation
and reversal to austenite upon unloading.
11.1.5 Application Examples of SMAs

SMA can be used for braces and dental arch wires, where the SMA maintains its
shape at a constant temperature; and because of the superelasticity of the mem-
ory metal, the wires retain their original shape after stress has been applied and
removed. The superelastic property of SMAs has also been applied in eyeglass
frames that can be bent back and forth with retained shape, demonstrating its
superelasticity as shown in Figure 11.9.
Figure 11.8 Stress–strain

curve illustrating superelastic Stress
behavior of SMA when T > Af .
Loading
Unloading
Strain
Figure 11.9 Glasses made of shape

memory alloy. Source: Adapted
from Burchett (2019).
SMA has also been used in fire security and protection systems, where the SMA
should have an immediate action to shut down or open the valve in the presence
of increased heat from fire. This can greatly decrease leakage-induced problems
in industries that involve petrochemicals, semiconductors, pharmaceuticals, and
large oil and gas boilers.
Many new applications of SMA are being developed, and even polymer-based
SMMs are emerging. These are beyond the scope of this book.
11.2 Ferromagnetic Shape Memory Alloys

Ferromagnetic shape memory alloys (FSMAs) are an attractive new class of
magneto-active materials that combine the properties of ferromagnetism with
shape memory effect. These materials exhibit giant magnetic field-induced strain
(MFIS) in the martensitic phase, i.e. when a magnetic field is applied, the shape
of the material will change significantly. The mechanism of MFIS originates
from the redistribution or reorientation of martensitic twin variants under an

external magnetic field.
The thermally driven SMA has a very slow response of less than 1 Hz. This is
attributed to the slow process of heating and cooling. However, the combination
of ferromagnetic properties and martensitic transformation in FSMA materials
allows realization of magnetically driven shape memory effect in a frequency up
to kHz regime.
Not only different in mechanisms compared with those of magnetostrictive,
piezoelectric, and electrostrictive, the magneto SMA materials also have greater
obtainable strains. This presents FSMAs with great potential in various smart
structural applications, such as sensors and actuators, which require a large strain
and/or fast response.
Ullakko et al. presented the idea of redistribution or reorientation of marten-
sitic twin variants under an external magnetic field and also experimentally con-
firmed the effect of MFIS in 1996 (Ullakko et al. 1996). They reported MFIS of
nearly 0.2% observed on unstressed Ni–Mn–Ga single crystals in ∼640 kA/m
(79.6 A/m = 1 Oe) magnetic field along the [001] direction at 265 K. This dis-
covery was a major step for the development of FSMA materials, especially in
the MFIS field. Since then, interest in this phenomenon as well as these alloys
has gained increased attention. Also, several additional material systems, such as
Ni–Mn–Fe–Ga, Fe–Pt, Fe–Pd, Co–Ni–Ga, and Ni–Mn–Al, have been tested.
After the report of 0.2% MFIS, extensive works on Ni–Mn–Ga intermetallic
compounds have been carried out. James et al. obtained a 0.5% reversible strain
in an off-stoichiometric Ni–Mn–Ga single crystal (James et al. 1998). Tickle
et al. reported separately a reversible strain of 1.3% and maximal extensional
strain of 4.3% in Ni51.3 Mn24.0 Ga24.7 (numbers indicate at.%) single crystals (Tickle
et al. 1999; Tickle and James 1999). In the next few years, there are a lot of
further development and nearly 10% strain has been obtained in modulated
martensitic single crystal (Sozinov et al. 2002). For dynamical actuation in these
alloys, Henry et al. demonstrated ac-MFIS of 2.6% up to 2 Hz under a bias stress
of 1.1–1.5 MPa (Henry et al. 2002). The cyclic MFIS of Ni–Mn–Ga element
under a pulse magnetic field was studied by Marioni et al. and Suorsa et al
(Suorsa et al. 2004).
One key advantage of Ni–Mn–Ga FSMAs is its giant MFIS under an external
magnetic field. Different from traditional magnetostrictive materials (e.g.
Terfenol-D) that generate strain through magnetic domain rotation, the mecha-
nism of MFIS is attributed to the redistribution or reorientation of martensitic
twin variants, as suggested initially by Ullakko in 1996. Thus, the formation
of martensitic twin variants and motion of twin boundaries under an external
magnetic field are significantly important to understand the mechanism of MFIS
in FSMA materials. In this section, the origin and nature of martensitic twin
variants as well as the mechanism of twin variants motion under an external
magnetic field are presented.
11.2.1 Formation of Twin Variants

Ni–Mn–Ga alloys exhibit a structural (martensitic) phase transformation
from a high-symmetry cubic phase to a low-symmetry martensitic phase (e.g.
tetragonal structure). Figure 11.10a shows atomic structure of Ni2 MnGa in
Ni Mn Ga
[110]
[110]
[001]
[001]
– –
[11 0] [11 0]
– –
[11 0] [11 0]
(a) (b) (c)
Figure 11.10 Illustrations of Ni2 MnGa structured models: (a) 3D cubic structure in the
high-temperature austenitic phase with two kinds of stacking (110) plane in cubic phase.
Five-layered martensite resulted from shear on these {110}c planes in the (b) cubic phase and
(c) modulation. Source: Adapted from Ge (2007).
high-temperature austenite phase. Figure 11.10b,c illustrate how a martensitic

phase is developed from a cubic austenite phase.
The martensitic transformation is illustrated in Figure 11.11a in two dimen-
sions. In the cubic structure, denoted as a gray block, the two crystallographic
axes (three axes in three dimensions) are equivalent. However, a tetragonal
martensite has a long axis (a axis) and a short axis (c axis), which are referred
to as the hard axis and easy axis (the magnetization is oriented along the easy
axis), respectively. Thus, the martensitic structure has two (3 × 3 dimensions)
possible tetragonal variants, i.e. two different deformation directions, during the
tetragonal deformation.
In current case, the easy axis is aligned along c crystallographic axis in the
tetragonal unit cell. The arrows “←” and “↑” in the two variant states denote
the orientation of easy axes, which are referred to as variant I and variant II,
respectively.
During martensitic transformation by a shear stress on the closed packed {110}
cubic planes, usually a twinned structure can be formed during the tetragonal
deformation as illustrated in Figure 11.11b. The twined structure is actually a pair
of unit cells in a lattice that the twin variants are oriented as mirror images. The
mirror symmetry line of the two variants is the twin boundary. In this case, twin
variants have two preferred orientations and can change their directions under
an external magnetic field or external mechanical load, and it is the reorientation
of twin variants that causes a macro-level deformed strain in a sample.
From the microstructure viewpoint, the martensitic twin variants and twin
boundaries are correspondence to magnetic domains and domain walls.
a
a
a c
Variant I
c
Variant II
(a) (b)
Figure 11.11 Schematic illustrations of (a) martensitic transformation and (b) twinned
structure in two dimensions. Arrows indicate the orientations of the variants (or easy axis).
Source: Adapted from Enkovaara et al. (2004).
Figure 11.12 A typical twin

variants with the 90∘ and the
180∘ magnetic domain
Variant I structures. Source: Ge et al.
(2004). Reprinted with
permission of AIP Publishing.
Variant II
[100] [001]
20 μm
[001] [100]
A scanning electron microcopy image presenting this microstructural feature

is shown in Figure 11.12. This is a typical pattern of the two variant, where the
magnetic domains are oriented along the easy axes of the two variants forming
a staircase-like pattern. The twin boundaries form 90∘ domain walls, and the
180∘ domain walls are formed in each variant manifested by the black and white
contrast (Ge et al. 2004). (A question, why different variants show different
contrast in scanning electron microscope (SEM) image?)
In a Ni–Mn–Ga single crystal, the motion of martensitic twin boundaries
results in magnetic domains evolution under an external magnetic field. As
shown in Figure 11.13, initially the single crystal is preset to be a single variant
configuration that can be achieved by applying an external magnetic field or
an external compressive mechanical load along the sample length direction.
Figure 11.13a depicts the 180∘ magnetic domains in a scenario that the overall
Micro-scale
Large strain
Single variant I
H=0
(a)
Applying magnetic field
Twin variants
Hc < H < Hs
Increasing magnetic field

(b)
Single variant II
H ≥ Hs
(c)
Figure 11.13 Schematic illustrations of a Ni–Mn–Ga single crystal under increasing applied
magnetic field levels. Schematics (a)–(c) depict the respective microstructural arrangements of
martensitic variants and magnetic domains. Source: Adapted from Kiefer and Lagoudas (2005).
magnetization is zero without external magnetic field applied. When an external

magnetic field is applied vertically to the sample length direction (horizontal
direction), the magnetic domains tend to align in the direction of the applied field.
When a magnetic field is applied but below a critical field H c , the applied
field only induces a rotation of magnetization away from the original direction.
But when the applied field increases beyond H c , i.e. the magneto-crystalline
anisotropic energy (MAE) is higher than the energy required moving a
twin, the twin variants are movable and variant II will grow accompanied by
reducing the other twin variants. This process generates an axial reorientation
strain, and the magnetization in variant II is eventually oriented fully along
the applied field direction. Once beyond the saturation magnetic field H s ,
variant II becomes dominant and completely replaces variant I. Consequently, as
illustrated in Figure 11.13c, the magnetic domains align themselves completely
along the direction of the applied field. The substitution of variant I with variant
II results in an elongation in that direction and contraction in the perpendicular
directions. Essentially, the motion of the martensitic twin variants can be seen as
the change of volume fraction of the two variants. Thus, twin variants reorienta-
tion strain, i.e. MFIS, is quantified by considering the changes in crystallographic
axes (a and c axes).
5
Strain (%)
3 c
c
a
2 a
0
0 100 200 300 400 500 600 700
H (kA/m)
Figure 11.14 MFIS dependence of applied magnetic field illustrates the change of twin
variants. Source: Adapted from Sozinov et al. (2003).
Figure 11.14 shows the relationship between MFIS and the corresponding
change of twin variants; where the coercive field H c and saturation field H s are
about 250 and 580 kA/m, respectively, and the maximum MFIS is about 5.5%. In
addition, the maximum strain is predictable from the theoretical lattice constant
with the equation
𝜀0 = (1 − c∕a)
where c and a are the short (or easy axis) and long (or hard axis) lattice constants,
respectively.
11.2.2 Challenges for Ni–Mn–Ga SMA

Up to date, ferromagnetic shape memory Ni–Mn–Ga single crystals have shown
giant dc-MFIS in the order of 6–10%, making them suitable for some applica-
tions (e.g. magnetic field sensors, actuators, and converters). However, there are
a few challenges that must be addressed before this FSMA can be largely used in
engineering applications.
First, large MFIS is obtainable only in single crystals; however, the growth of
device-level single crystals is of high cost. Besides, the Ni–Mn–Ga single crystals
present poor formability and brittleness, making it difficult to shape the mate-
rial into required form. Alternatively, some low cost magnetic shape memory
alloy (MSMA) such as NiAl etc. may be a good candidate for applications, but
of course, their MFIS is still much lower than Ni–Mn–Ga single crystals.
Second, even a significant irreversible strain is exhibited under quasi-static
actuations, only a very small reversible strain is available. From the practical
applications point of view, large cyclic strain in ac actuation is desired; therefore,
this disadvantage of significant irreversible strain makes Ni–Mn–Ga single
crystals not suitable for application of high cyclic strain.
Third, data presented in MFIS curves indicate that a relatively large critical field
is needed to initiate shape transformation. Consequently, this relatively large crit-
ical field will increase the size of the drive coil/core, which significantly limits the
application in realizing miniature active systems.
Fourth, the output mechanical stress by Ni–Mn–Ga single crystals driven by
MFIS is actually very small (6–10 MPa). This means that the actuation stress in
practical application is small, which is difficult in meeting the requirement for
large actuation force.
11.2.3 Device Application of MSMA

Similar to the composite device of piezoelectric/magnetostrictive, a het-
erostructure can also be formed by one layer of piezoelectric PMN-PT crystal
sandwiched between two layers of ferromagnetic shape memory Ni–Mn–Ga
crystal (Zeng et al. 2010). This device can be a magnetic field sensor utilizing
its magnetoelectric (ME) effect or an electric field sensor utilizing its converse
ME effect. In Zeng’s work, coefficient of electric field-induced magnetization
𝛼 B , called converse magnetoelectric (CME) effect, of the heterostructure
attains its maximum value in the martensitic–austenitic phase transformation
temperature 28–39 ∘ C. Giant resonance 𝛼 B of 18.6 G/V under a very low bias
magnetic field of 150 Oe has been obtained. Figure 11.15 shows the schematic
Search coil (a)

Ni–Mn–Ga
Vac
Vind PMN–PT
0.9
(b)
MP MAPT AP
0.6 Ms
αB (G/V)
Af
0.3
Heating
Mf As Cooling
0.0
20 25 30 35 40 45 50 55 60
T (°C)
Figure 11.15 (a) Schematic diagram of the proposed heterostructure. (b) CME coefficient
(𝛼 B ) dependence on temperature at an applied ac electric voltage V ac of 20 V peak, a
frequency f of 1 kHz, and a magnetic bias field Hbias of 150 Oe upon heating and cooling.
Source: Zeng et al. (2010). Reprinted with permission of AIP Publishing.
diagram of the proposed heterostructure fabricated by bonding one layer of

piezoelectric PMN-PT crystal plate between two layers of ferromagnetic shape
memory Ni–Mn–Ga crystal plates. The two Ni–Mn–Ga plates have a chemical
composition of Ni49.2 Mn29.6 Ga21.2 , dimensions of 11 × 5 × 0.5 mm3 , and the two
major surfaces are parallel to the (011) plane.
References
Burchett, S. (2019). Global Nitinol Shape Memory Alloy Market 2019 Top Company
Profiles, Competition Status, Trends, Industry Growth, And Forecast 2019–2025.
themarketfact.com. Available at: https://themarketfact.com/2019/05/02/global-
nitinol-shape-memory-alloy-market-2019/ (accessed 21 August 2019).
Enkovaara, J., Ayuela, A., Zayak, A.T. et al. (2004). Magnetically driven shape
memory alloys. Materials Science and Engineering: A 378 (1): 52–60. Available at:
Ge, Y. (2007). The Crystal and Magnetic Microstructure of Ni–Mn–Ga Alloys.
Helsinki University of Technology.
Ge, Y., Heczko, O., Söderberg, O., and Lindroos, V.K. (2004). Various magnetic
domain structures in a Ni–Mn–Ga martensite exhibiting magnetic shape
memory effect. Journal of Applied Physics 96 (4): 2159–2163. Available at: https://
doi.org/10.1063/1.1773381.
Henry, C.P., Bono, D., Feuchtwanger, J. et al. (2002). AC field-induced actuation of
single crystal Ni–Mn–Ga. Journal of Applied Physics 91 (10): 7810–7811.
Available at: https://aip.scitation.org/doi/abs/10.1063/1.1449441.
James, R.D., Tickle, R., and Wuttig, M. (1998). Large field-induced strains in
ferromagnetic shape memory materials. Materials Science and Engineering. A,
Structural Materials: Properties, Microstructure and Processing 273–275:
320–325.
Kiefer, B. and Lagoudas, D.C. (2005). Magnetic field-induced martensitic variant
reorientation in magnetic shape memory alloys. Philosophical Magazine 85
(33–35): 4289–4329. Available at: https://doi.org/10.1080/14786430500363858.
Salje, E.K. (1993). Phase Transitions in Ferroelastic and Co-elastic Crystals.
Cambridge, UK: Cambridge University Press.
Salje, E.K.H. (2012). Ferroelastic materials. Annual Review of Materials Research 42
(1): 265–283. Available at: https://doi.org/10.1146/annurev-matsci-070511-
155022.
Salje, E.K. and Hoppmann, G. (1976). Direct observation of ferroelasticity in
Pb3 (PO4 )2 -Pb3 (VO4 )2 . Materials Research Bulletin 11 (12): 1545–1549. Available
at: http://www.sciencedirect.com/science/article/pii/0025540876901070.
Sozinov, A., Likhachev, A.A., Lanska, N. et al. (2002). Giant magnetic-field-induced
strain in NiMnGa seven-layered martensitic phase. Applied Physics Letters 80
(10): 1746–1748. Available at: https://doi.org/10.1063/1.1458075.
Sozinov, A., Likhachev, A.A., Lanska, N. et al. (2003). Effect of crystal structure on
magnetic-field-induced strain in Ni–Mn–Ga. Proc. SPIE 5053, Smart Structures
and Materials 2003: Active Materials: Behavior and Mechanics 5053: 586–594.
References 245
Suorsa, I., Pagounis, E., and Ullakko, K. (2004). Magnetic shape memory actuator
performance. Journal of Magnetism and Magnetic Materials 272–276 (3):
2029–2030. Available at: http://www.sciencedirect.com/science/article/pii/
S0304885303024922.
Tickle, R. and James, R.D. (1999). Magnetic and magnetomechanical properties of
Ni2MnGa. Journal of Magnetism and Magnetic Materials 195 (3): 627–638.
Available at: http://www.sciencedirect.com/science/article/pii/
S0304885399002929.
Tickle, R., James, R.D., Shield, T. et al. (1999). Ferromagnetic shape memory in the
NiMnGa system. IEEE Transactions on Magnetics 35 (5): 4301–4310.
Ullakko, K., Huang, J.K., Kantner, C. et al. (1996). Large magnetic-field-induced
strains in Ni2MnGa single crystals. Applied Physics Letters 69 (13): 1966–1968.
Zeng, M., Or, S.W., and Chan, H.L.W. (2010). Effect of phase transformation on the
converse magnetoelectric properties of a heterostructure of Ni49.2 Mn29.6 Ga21.2
and 0.7PbMg1/3 Nb2/3 O3-0.3 PbTiO3 crystals. Applied Physics Letters 96 (18):
182503. Available at: https://doi.org/10.1063/1.3427388.
247
Index
a converse magnetoelectric (CME) effect

acoustic wave 3, 144, 147, 154, 155, 243
158 Curie–Weiss law 20–23, 76
anisotropic magnetoresistance 171
anomalous Hall effect (AHE) 191–192 d
antiferroelectricity (AFE) 19, 33, 197 depolarization field 47–49, 55, 96
antiferroelectric material 34 device application, of ferroelectrics
antiferromagnetic coupling 175 electrocaloric effect 67–68
antiferromagnetic ordering 202 ferroelectric random access memory
artificial intelligence (AI) 1, 57 (FeRAM) 47–49
artificial neural networks (ANN) 1, 2, ferroelectric tunnel junction (FTJ)
57 50
microwave device 63–65
b photovoltaics 65–67
bandwidth (BW) 144, 147–148, 150, dielectric constant 8, 16, 17, 21–23, 25,
162, 175, 180, 218 26, 29–33, 35, 36, 63, 74
BaTiO3 phase transition 20, 21, 27 dielectric tunability 30–32
BiFeO3 (BFO) 49, 89, 223 differential scanning calorimeter (DSC)
ferroelectric layer 1 233, 234
switching mechanism 204 direct tunneling (DT) 51–53
Bloch type ferromagnetic domain walls domains, and phase field calculation
171 41–42
Bloch wall 171 domain structure 15, 16, 28, 38, 39, 41,
Bode diagram 153 56, 77, 78, 84, 85, 87–89, 96, 99,
Brinkman’s model 52 169–171, 189, 201, 203, 231
bulk acoustic wave (BAW) 155 dual frequency resonance tracking
butterfly loop 78, 79, 115 (DFRT) technique 85
Dzyaloshinskii–Moriya interaction
c 203, 204
colossal magnetoresistance (CMR)
171, 175–176 e
conventional ferroelectrics 109, 110 Ehrenfest classification 27
convergent beam electron diffraction elastic compliance 127–128, 135, 138,
(CBED) 91 210

248 Index
electric polarization 4, 15–18, 20, 59, dielectric constant 16, 17

60, 67, 73, 87, 97, 114, 123, 124, P–E loop 15–16
127, 197–199, 201, 206, 217, 226 piezoelectric 17–18
electrocaloric effect 67–68 pyroelectricity 18
electromechanical coupling factor 37, soft phonon mode 19–21
123, 124, 128–132, 138–140, 217 structure-induced phase change
electromechanical sensors 37 18–19
electroresistance (ER) 50–55 ferroelectric switching 28–29, 54, 78,
electrostatic force, in piezoresponse 97, 204
force microscopy 83–84 ferroelectric tunnel junction (FTJ)
direct tunneling 52
f electroresistance (ER) 50
feedback control system 1, 3 ferroelectric memory 57
ferroelectric and anti-ferromagnetic memristive behavior 56
(FE-AFM) 223 metal-ferroelectric-semiconductor
ferroelectric-based memory 1, 50 (MFS) heterostructure 55
ferroelectricity nanoscale characterization
HfO2 and ZrO2 fluorite-oxide thin techniques 50
films 106–113 polarization switching 53
size limit 95–96 thermionic emission 52
two-dimensional (2D) materials ferroelectric tunnelling-based resistive
96–99 random-access memory 1
ferroelectric materials 1 ferromagnetic materials, domain and
antiferroelectric PbZrO3 33–36 domain wall 169–171
BaTiO3 to SrTiO3 29–32 ferromagnetic shape memory alloys
and device application 10–12 (FSMAs) 237–244
ferroelectric RAM (FeRAM) 8–9 ferromagnetics, paramagnetic phase
infrared sensor 7 transitions 168
magnetoelectric (ME) effect 10, 11 field effect transistor (FET) structure
Pb(Zrx Ti1–x )O3 (PZT) 35–36 1, 49, 50, 95, 223
PbTiO3 to PbZrO3 32–33 figure of merit (FOMs) 60
piezoelectric device applications finite element modeling (FEM) 3, 81,
6–7 217
spin-transfer torque memory 9 Fowler–Nordheim tunneling (FNT)
ferroelectric phase transition 86 51, 52
Landau free energy and Curie–Weiss
law 21–23 g
Landau theory of first order phase giant magnetoelastic coupling 201
transition 23–26 giant magnetoresistance (GMR)
ferroelectric polarization 1, 5, 17, 21, 171–175
47–50, 74, 77, 84, 91, 99, 108, Gibbs free energy 27
112, 197, 200, 201, 224–229 Google’s AlphaGo 1
ferroelectric random-access memory Gruverman’s model 52
(FeRAM) 1, 47–49, 109, 201,
223 h
ferroelectrics hafnium dioxide (HfO2 ) 106
definition 15 half-metallic manganites 176
Index 249
Hall effect magnetic field-induced strain (MFIS)

anomalous Hall effect (AHE) 224, 237
191–192 magnetic field sensor 9–10, 175, 217,
ordinary Hall effect 191 223, 242, 243
spin Hall effect (SHE) 192–193 magnetic force microscopy (MFM)
Heusler alloys 176 188–189
high-angle annular dark-field (HAADF) magnetic shape memory alloy (MSMA),
89 device application 243–244
high-frequency resonance measurement magnetic tunnel junction (MTJ) element
130–132 176, 224
hybrid perovskites 73, 84, 114–117 magneto-crystalline anisotropic energy
(MAE) 241
i magneto-elastic coupling 201
impedance-frequency spectrum 132, magnetoelectric (ME) 10, 11, 199
133, 135 automated measurement system
industrial-compatible fabrication 219
process, 111 composite materials 205–207
infrared radiation 7, 58, 60, 62, 63 L–T mode 218
in-plane polarization (IPP) 80–86, 98, magnetic bias field 220
100 magnetoelastic coupling factor 217
interdigital electrode 155 magneto-elasto-electric coupling
interdigital transducers (IDTs) 154, 219
159 preload stress 221–223
Ising type ferroelectric domain walls PZT/CFO multilayered
171 heterostructures 207–209
Terfenol-D fiber 218
j magneto-optical Kerr effect (MOKE)
Jahn–Teller effect 200 189–191
Julliere’s model 176, 228 magnetoresistance effect
anisotropic magnetoresistance
k 171–172
Krimholtz–Leedom–Matthae (KLM) colossal magnetoresistance (CMR)
model 145 175–176
giant magnetoresistance (GMR)
l 172–175
Landau free energy and Curie–Weiss spin-transfer torque random access
law 21–23 memory (STT-RAM) 177–178
Landau theory 199 tunnel magnetoresistance (TMR)
of first order phase transition 23–25 176–178
of second order phase transition magnetoresistive random-access
26–28 memory (MRAM) 9, 172, 177
lead-free ferroelectrics 161 magnetostrictive effect 184
LiNbO3 film deposition 158 properties of Terfenol-D 180–183
longitudinal vibration mode 217, 218 ultrasonic transducer 183–184
Mason model 145
m mechanical quality factor 123, 141,
magnetic ferroelectrics 197, 199–200 151, 153
250 Index
medical ultrasound imaging system 6 direct and converse piezoelectric

memory device effects 123
multiferroic thin films 223–224 equivalent circuit 132–135
multiferroic tunneling 224–229 extensional vibration mode of long
metal-ferroelectric-semiconductor plate 138–139
(MFS) heterostructure 55 length extensional mode of rod
Metglas 2605SC 179, 185 135–138
microwave device 30, 63–65 mechanical quality factor 141
molecular materials 104 properties of 123, 124
multiferroic thin films 223–224 radial mode in thin disc 141
multiferroic tunneling 224–229 thickness mode of thin disc/plate
140–141
n thin shear piezoelectric plate 139
nanoferromagnets 100 piezoelectric-miniature ultrasonic
Néel temperature 166, 201 transducer (PMUT) 149
Néel type ferromagnetic domain walls piezoresponse force microscopy (PFM)
171 188
Néel wall 171 imaging mechanism 77–80
Ni-Mn-Ga single crystals 238, 240, in-plane polarization (IPP) 80–86
242, 243 out-of-plane polarization (OPP)
non-destructive test (NDT) 143, 171 80–86
Nyquist diagram 151–153 perspectives of techniques 84–86
polar nano regions 30, 38–40
o polymer materials 205
one transistor-one capacitor (1T1C) preload stress 221–223
structure 47 proportional-integrate-derivative (PID)
ordinary Hall effect 191–193 feed-back control system, 3
out-of-plane polarization (OPP) pulsed laser deposition (PLD) 112, 207
80–86 pyroelectric coefficient 58–62, 67
pyroelectric infrared sensor 58–60
p pyroelectricity 5, 18
paramagnetic phase transitions 166, PZT/CFO multilayers, ferroelectric
168, 179 properties, 209–211
P–E hysteresis loop
ideal linear response of capacitor q
74, 75 quantum effect 188
modified Sawyer–Tower circuit 73 quantum tunneling 50
photovoltaics 65–67, 84, 114
piezoelectric effect r
constants measurement Rayleigh wave, 151, 155, 156
charge constant 125 Redwood model 145
dielectric permittivity 127 relaxor ferroelectrics 29, 36–41
elastic compliance 127–128 characteristics 36–37
electromechanical coupling factor morphotropic phase boundary (MPB)
128–132 of PMN-PT crystal 40–41
voltage constant 126–127 PMN-xPT single crystal 37–38
Young’s modulus 127 polar nano regions 38–40
Index 251
resonance enhancement 85, 86 piezoelectric property considerations

Rochelle salt 29, 104 157–159
symmetric Bragg peak 86
s
selected area electron diffraction t
(SAED) 87–91 temperature-dependent dielectric
shape memory alloys (SMA) 10 permittivity measurement
ferromagnetic shape memory alloys 76–77
(FSMAs) 237–244 Terfenol-D 179–183, 185, 206, 218,
nonlinearity 233–235 219, 221, 222
one-way and two-way shape memory thermionic emission (TE) 51, 52
effect 235 tip-nanofiber harmonic oscillation
phase change mechanism 232–233 system 87
superelastic effect 235–236 tunnel magnetoresistance (TMR) 171,
twin variants, formation of 238–242 176–178, 228
single phase multiferroic materials two-dimensional (2D) materials
antiferromagnetic BiFeO3 (BFO) atomic-thick SnTe 98
201 Hamiltonian and parameters 98
antiferromagnetic ordering 202 in-plane polarization 98
Dzyaloshinskii–Moriya interaction quantum phenomena 96
204 SnS/SnSe monolayer 97
epitaxial BFO films 201 two-way shape-memory effect 235
giant magneto-elastic coupling 201
oxygen octahedron 201 u
rare-earth manganites and ferrites ultrasonic transducers
200 bandwidth (BW) 147, 148
spin canting 203 medical application 149
smart system resolution 148, 149
actuators 3 sensitivity of transducer 148
component 2 structure of 143–145
proportional-integrate-derivative theoretical models of 145–146
(PID) feed-back control ultrasound motor (USM)
system 3 design of 153–154
sensors 3 energy conversion 151
transducers 3–4 Rayleigh wave 151
S-parameters 159 terminologies 151–153
spin canting 203, 204
spin Hall effect (SHE) 192–193 v
spin-transfer torque random access Van Dyke’s Model 132
memory 177–178 vibrating sample magnetometer (VSM)
structural distortion 86 186–187
superelastic effect (SE) 235–236 vortex
surface acoustic wave (SAW) device formation of circular domains 99
acoustic wave 155–157 in-plane polarization 100
characterization 159–161 nanoferromagnets 100
interdigital electrode 155 skyrmion in ferroelectrics 101–103
252 Index
toroidal domains 99 x
up and down polarization domains X-ray diffraction 38–40, 96, 110
101
y
w Young’s modulus 127, 128, 209
wake-up effect 110, 111

Ferroic Materials For Smart Systems From Fundamentals To Device Applications PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ferroic Materials For Smart Systems From Fundamentals To Device Applications PDF

Uploaded by

Copyright:

Available Formats

Ferroic Materials for Smart Systems

Ferroic Materials for Smart Systems

From Fundamentals to Device Applications

Bibliographic information published by

© 2020 Wiley-VCH Verlag GmbH &

All rights reserved (including those of

Print ISBN: 978-3-527-34476-5

Typesetting SPi Global, Chennai, India

Printed on acid-free paper

1 General Introduction: Smart Materials, Sensors, and

2.5.1 From BaTiO3 to SrTiO3 29

3 Device Applications of Ferroelectrics 47

5 Recent Advances in Ferroelectric Research 95

5.3 Ferroelectric Vortex 99

6 Piezoelectric Eﬀect: Basic Theory 123

7 Piezoelectric Devices 143

7.3.4 Characterization of SAW Devices 159

8 Ferromagnetics: From Material to Device 165

9 Multiferroics: Single Phase and Composites 197

10.3 Memory Devices Based on Multiferroic Tunneling 224

11 Ferroelasticity and Shape Memory Alloy 231

General Introduction: Smart Materials, Sensors, and

The early twenty-ﬁrst century has foreseen acceleration of innovations in

Ferroic Materials for Smart Systems: From Fundamentals to Device Applications,

Figure 1.1 AI beats human chess player.

1.1 Smart System

As an example, a smart system of remote control copter relying on

1.2 Device Application of Ferroelectric Materials

1.2.1 Piezoelectric Device Applications

Coupling materials + MEMS

Bottom electrode Top electrode

Figure 1.8 Illustration of concept of a ultrasonic transducer-based ﬁngerprint ID system based

1.2.2 Infrared Sensor

Absorbing electrode layer

1.2.3 Ferroelectric RAM (FeRAM)

FeRAM remains a relatively small part of the overall semiconductor market.

1.3 Device Application of Ferromagnetic Materials

1.3.1 Spin-Transfer Torque Memory

1.3.2 Magnetic Field Sensor Based on Multiferroic Device

Low resistance state High resistance state

1.4 Ferroelastic Material and Device Application

1.5 Scope of This Book

2.1 What Is Ferroelectrics?

2.1.1 P–E Loop

Ferroic Materials for Smart Systems: From Fundamentals to Device Applications,

P Figure 2.1 Typical P–E loop from a

where 𝜖 0 is the vacuum permittivity or called permittivity of free space and

2.1.2 Relationships Between Dielectric, Piezoelectric, Pyroelectric,

Center of negative charge

Figure 2.3 Relationships of the

2.2 Origin of Ferroelectrics

• Crystals can be classiﬁed into 32 point groups according to their crystallo-

Table 2.1 Crystallographic classiﬁcation according to crystal centrosymmetry.

Inside are piezoelectrics (20)

Inside are pyroelectrics (10)

Source: Adapted from Uchino (2009).

Figure 2.4 Illustration of crystal structure symmetry breaking-induced ferroelectricity in

2.2.2 Soft Phonon Mode

𝜖r (0) ∼ (T − Tc )−1 (2.6)

where T 0 is the soft mode condensation temperature.

200 300 400 500 600 700 800

2.3 Theory of Ferroelectric Phase Transition

peak value continuously; while 1/𝜖 r is linear at temperatures lower or higher