Surface Screening Mechanisms in Ferroelectric Thin Films and its Effect on Polarization Dynamics and Domain Structures

Abstract

For over 70 years, ferroelectric materials have been remaining one of the central research topics for condensed matter physics and material science, the interest driven both by fundamental science and applications. However, ferroelectric surfaces, the key component of ferroelectric films and nanostructures, still present a significant theoretical and even conceptual challenge. Indeed, stability of ferroelectric phase per se necessitates screening of polarization charge. At surfaces, this can lead to coupling between ferroelectric and semiconducting properties of material, or with surface (electro) chemistry, going well beyond classical models applicable for ferroelectric interfaces. In this review, we summarize recent studies of surface screening phenomena in ferroelectrics. We provide a brief overview of the historical understanding of physics of ferroelectric surfaces, and existing theoretical models that both introduce screening mechanisms and explore the relationship between screening and relevant aspects of ferroelectric functionalities starting from phase stability itself. Given that majority of ferroelectrics exist in multiple-domain states, we focus on local studies of screening phenomena using scanning probe microscopy techniques. We discuss recent studies of static and dynamic phenomena on ferroelectric surfaces, as well as phenomena observed under lateral transport, light, chemical, and pressure stimuli. We also note that the need for ionic screening renders polarization switching a coupled physical-electrochemical process, and discuss the non-trivial phenomena such as chaotic behavior during domain switching that stem from this.

Full text

1
To be submitted to Rev. Prog. Phys.
Surface Screening Mechanisms in Ferroelectric Thin Films and its Effect on Polarization
Dynamics and Domain Structures
Sergei V. Kalinin,1 Yunseok Kim,2 Dillon Fong,3 and Anna Morozovska4
1 The Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN
37922
2 School of Advanced Materials Science and Engineering, Sungkyunkwan University (SKKU),
Suwon 16419, Republic of Korea
3 Materials Science Division, Argonne National Laboratory, 9700 S. Cass Avenue, Argonne, IL
60439
4 V. Lashkaryov Institute of Semiconductor Physics, National Academy of Science of Ukraine,
41, pr. Nauki, 03028 Kiev, Ukraine
Notice: This manuscript has been authored by UT-Battelle, LLC, under Contract No. DE-
AC0500OR22725 with the U.S. Department of Energy. The United States Government retains
and the publisher, by accepting the article for publication, acknowledges that the United States
Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or
reproduce the published form of this manuscript, or allow others to do so, for the United States
Government purposes. The Department of Energy will provide public access to these results of
federally sponsored research in accordance with the DOE Public Access Plan
(http://energy.gov/downloads/doe-public-access-plan).

2
Outline:
I. Introduction
II. Theory and macroscopic studies of polarization screening on surfaces
II.1. Internal and external screening on surfaces
II.2. Theory of surface screening
II.2.1 Surface screening effects on phase stability
II.2.2 Role of screening on the structure of wall-surface junction
II.2.3. Role of surface screening on wall dynamics
II.3. Coupling between physics and electrochemical behaviors
II.4. Experimental macroscopic studies of polarization screening at surfaces
III. Local studies of polarization screening
III.1. SPM studies of domain structures
III.2. Static studies of work function
III.3. Domain dynamics under constant conditions
III.4. Lateral fields
III.4.1. Lateral switching
III.4.2. Lateral charge injection
III.4.3. Charge dynamics on ferroelectric surfaces
III.5. Variable temperature experiments
III.5.1. Surface potential evolution on heating
III.5.2. Temperature induced domain potential inversion
III.5.3. Thermodynamics of screening
III.6. Environmental effects
III.7. Interplay between screening and size effects in ferroelectric nanostructures
IV. Tip induced switching
IV.1. Charge injection on non-ferroelectric surfaces
IV.2. Potential evolution during switching
IV.3. Backswitching
IV.4. Charge collection phenomena

3
IV.5. Pressure effects
V. Non-local interactions driven by screening charges
V.1. Spatiotemporal chaos during ferroelectric domain switching
V.2. Domain shape instabilities during switching
V.3. Modeling of screening effects impact on the tip-induced polarization switching
V.4. Screening effects during in-plane domain switching
VI. Chemical and light effects
VI.1. Chemical switching
VI.2. Light effects
VI.3. Domain dependent photodeposition
VII. Conclusion and outlook

4
Abstract:
For over 70 years, ferroelectric materials have been remaining one of the central research topics
for condensed matter physics and material science, the interest driven both by fundamental
science and applications. However, ferroelectric surfaces, the key component of ferroelectric
films and nanostructures, still present a significant theoretical and even conceptual challenge.
Indeed, stability of ferroelectric phase per se necessitates screening of polarization charge. At
surfaces, this can lead to coupling between ferroelectric and semiconducting properties of
material, or with surface (electro) chemistry, going well beyond classical models applicable for
ferroelectric interfaces. In this review, we summarize recent studies of surface screening
phenomena in ferroelectrics. We provide a brief overview of the historical understanding of
physics of ferroelectric surfaces, and existing theoretical models that both introduce screening
mechanisms and explore the relationship between screening and relevant aspects of ferroelectric
functionalities starting from phase stability itself. Given that majority of ferroelectrics exist in
multiple-domain states, we focus on local studies of screening phenomena using scanning probe
microscopy techniques. We discuss recent studies of static and dynamic phenomena on
ferroelectric surfaces, as well as phenomena observed under lateral transport, light, chemical, and
pressure stimuli. We also note that the need for ionic screening renders polarization switching a
coupled physical-electrochemical process, and discuss the non-trivial phenomena such as chaotic
behavior during domain switching that stem from this.

5
I. Introduction
For over 70 years, ferroelectric materials have been remaining one of the central research topics
for condensed matter physics and material science, the interest driven both by fundamental
science and applications.[1-3] Throughout this time, the focus of research has shifted from bulk
crystals and ceramics to thin films and nanostructures, both as perspective materials for
information technology and data storage applications and the playground of interesting physics.
Many aspects of ferroelectric behavior including soft mode behavior,[4-7] multiferroic
couplings,[8-12] as well as new functionalities emerging at topological defects[13-23] are now
being extensively studied, the advances made possible by advances in instrumental
characterization techniques such as focused X-ray[24-27] and neutron scattering and electron
microscopy,[28-34] and progress in atomistic theory.[35-38] However, ferroelectric surfaces, the
key component of ferroelectric films and nanostructures, still present a significant theoretical and
even conceptual challenge.
Indeed, since the early days of ferroelectricity it has been recognized that the
discontinuities of polarization at surfaces and interfaces create a bound charge, as a consequence
of fundamental Maxwell electrostatics. The latter gives rise to a depolarization field opposite to
the polarization direction. To reduce the energy of the depolarization field and stabilize the
ferroelectric film, either 180˚ domain structures with antiparallel polarization stripes, or the
surface compensating free charges to the polarization bound charges should be created. In the
absence of the charge compensating process, the ferroelectric phase is absolutely unstable.[1]
However, the nature of these screening charges was traditionally excluded from consideration
and analysis – rather, they were assumed to be (a) always present and (b) irrelevant to the
macroscopic physics of these materials. This assumption is well justified for bulk ferroelectrics
close to equilibrium, where the role of surface effects can be expected to be minor.
This postulate, however, is no longer true on the nanoscale, when the free energies of
surface ionic and electronic screening become comparable to the bulk free energy of the
ferroelectric. In fact, both a range of highly unusual phenomena ranging from hot electron[39-
42] and X-ray[43] emission and fusion[44] to ferroelectric states observed on the nanoscale[45]
can be traced to this surface charge dynamics. Furthermore, any processes involving polarization
switching in system with boundary conditions other than ideal metal electrodes (which provide
unlimited source of screening charges spatially coinciding with polarization bound charges) will

6
now be ruled by the interplay between the polarization and screening dynamics, leading to
coupling between ferroelectric and charge (electronic or ionic) transport behaviors, as
necessitated by charge and mass conservation laws. In other words, polarization cannot be
switched unless screening charges redistribute. These considerations in turn necessitate studies of
the thermodynamic and kinetic behavior of the screening charges, and their role on ferroelectric
functionalities in thin films.
In this review, we summarize recent studies of surface screening phenomena in
ferroelectrics. We provide a brief overview of the historical understanding of phenomena on
ferroelectric surfaces, and existing theoretical models that both introduce screening mechanisms
and explore the relationship between screening and relevant aspects of ferroelectric
functionalities starting from phase stability per se. Given that majority of ferroelectrics exist in
multiple-domain states, in this review we focus on local studies of screening phenomena using
scanning probe microscopy techniques. We discuss recent studies of static and dynamic
phenomena on ferroelectric surfaces, as well as phenomena observed under lateral transport,
light, chemical, and pressure stimuli. We also note that the need for ionic screening renders
polarization switching a coupled physical-electrochemical process, and discuss the non-trivial
phenomena such as chaotic behavior during domain switching that stem from this.
In the light of growing interest to nanoscale ferroelectrics and ferroelectric surfaces, as
well as recent advances in theoretical methods that can capture surface behaviors of
ferroelectrics, we believe that this review will be of interest to a broad range of condensed matter
physicists, theorists, and materials scientist. We further believe that many of these mechanisms
will be applicable to other polar oxide surfaces such as celebrated LAO-STO system,[46, 47] and
can be directly coupled to observed electrochemical behaviors[48-52] and polar responses.[52,
53]
II. Theory and macroscopic of polarization screening on surfaces
In this section, we provide general overview of the theoretical aspects of the surface screening on
ferroelectrics under band bending and electrochemical control, and results of the macroscopic
surface science studies of surface chemistry of ferroelectrics.
II.1. Internal and external screening on surfaces

7
The fundamental aspect of ferroelectric materials is the presence of surface and interface bound
charge due to the discontinuity of polarization fields. Simple estimates illustrate that if this
charge is uncompensated, it will provide bulk-like contribution to free energy of materials, i.e.
corresponding surface energy diverges with system size. These electrostatic considerations were
well recognized since the early days of ferroelectricity, and necessitated the search for
mechanisms for lowering of this depolarization energy. One such mechanism is formation of
ferroelectric domains, extensively analyzed in classical textbooks[54] and further expanded
recently.[55, 56] However, even in the presence of domain structures additional lowering of
depolarization energy can be achieved if polarization charge is compensated, or screened, by
external charges, were it mobile carriers in metallic electrodes or charged ionic species adsorbed
from atmosphere. The energy gain in screening process can be roughly estimated based on the
physical separation between polarization and screening charges from the energy of capacitor
with fixed charge density. While this separation is comparable to system size for bulk
uncompensated ferroelectrics, it is comparable to domain periodicity for domain structures. For
screened systems, it is proportional to physical separation between average location of
polarization (within ferroelectrics) and screening (outside ferroelectrics for adsorption and
metallic electrode, potentially inside ferroelectrics for vacancy screening) charges.
Practically established state of the system is determined both by the kinetics and
thermodynamics of available screening mechanisms. Consider the rapid cooling of ferroelectric
materials below Curie temperature in the medium with low concentration of mobile charges (e.g.
in vacuum). In this case, domain formation is the fastest energy lowering process available to the
system, yielding domain structure with periodicities governed by Kittell law. The slower
screening by mobile surface charges will additionally lower the system energy; however,
reconfiguration of domain systems is hindered by high kinetic barriers to nucleate new domains.
In comparison, very slow cooling of ferroelectric material in the presence of abundant electronic
or ionic screening charges can lead to single domain states, or large periodicity domain systems
(since effective polarization charge density and, consequently, depolarization energy are now
much smaller). It is further important to note that relative preference of screening process is
determined by the medium. In the presence of (inert) conductive metallic electrodes, the
screening will be by electrons and will be associated with relatively large physical separation
between screening and polarization charges that can be represented by dielectric (dead) layer

8
effect. Screening by adsorbed ions on free surfaces can be more favorable energetically. The
important aspect of screening process is the fact that crystal lattice of e.g. oxide ferroelectrics can
change composition, e.g. develops oxygen non-stoichiometry. The vacancies can effectively
screen polarization charge.[57, 58] Similarly, oxidation can happen on metal-ferroelectric
interfaces.[59]
For future discussion, we will distinguish the external and internal screening. The former
is defined as charge compensation phenomena outside the physical boundary of ferroelectrics for
a given surface termination, and may include screening by conductive electrons in metallic
electrodes, adsorbed charged species, or mobile ionic and ionic species in second (external)
component. In the ideal case of external screening, the polarization distribution in the
ferroelectrics is that of ideal material, i.e. polarization is uniform and changes abruptly at the
interface. The internal screening can be defined as that by electrons or oxygen or cationic
vacancies within the ferroelectric material, with associated changes in polarization, carrier, and
ionic species concentration profiles in the vicinity of the surface. Practically, both scenarios can
be realized simultaneously, with the relative contributions controlled both by the
thermodynamics and kinetics of respective screening processes.
II.2. Theory of surface screening
Surface screening of spontaneous polarization in ferroelectrics is typically realized by the mobile
charges adsorbed from the ambient in the case of high or normal humidity.[60-64] In a specific
case of the very weak screening, or its artificial absence due to the experimental conditions
(cleaned surface in dry atmosphere, ultra-high vacuum or thick dielectric layer at the surface) the
screening charges can be localized at surface states caused by the strong band-bending by
depolarization field.[65-69] For both aforementioned cases the screening charges are at least
partially free (mobile), in the sense that the spatial distribution of their quasi two-dimensional
density is ruled by the polarization distribution near the surface. When the screening charges
follow the polarization changes almost immediately, the screening density can be calculated
theoretically in the adiabatic approximation.[66, 68] In the opposite case, one should solve
dynamic relaxation type equations for the spatial-temporal distribution of screening charges.[68]
Due the long-range nature of the depolarization effects, the incomplete surface screening of
ferroelectric polarization strongly influences the domain structure and leads to its pronounced

9
changes both near and relatively far from the surface, and affects phenomena such as domain
wall pinning mechanisms at surfaces and interfaces, as well as nucleation dynamics, domain
shape and period control in thin film under the open-circuited conditions,[70, 71] in films placed
between imperfect "real" electrodes with finite Tomas-Fermi screening length[72] or separated
from the electrodes by ultra-thin dead layers[73] or physical gaps,[74] leads to the formation of
closure domains near free surfaces in multiaxial ferroelectrics,[70, 75, 76] domain wall
broadening in both uniaxial and multiaxial ferroelectrics[77, 78] and crossover between different
screening regimes of the moving domain wall - surface junctions.[79, 80]
II.2.1 Surface screening effects on phase stability
Dead (or screening) layers [figure II.1(a, c)] are responsible for the imperfect screening of the
film ferroelectric polarization that in turn leads to the emergence of multidomain ferroelectric
phase. Typically the phase stability region is between the paraelectric and homogeneous
ferroelectric phase [figure II.1(b, d)]. At that the critical thickness
cr
l
of the size-induced phase
transition into a paraelectric phase can vary in a wide range, from nanometers to micrometers,
depending on the geometry (gap or dead layer thickness), ferroelectric material parameters,
temperature and screening charges concentration [compare figure II.1(b) and II.1(d)]. For the
case of dead layers (or gaps) presence [shown in figure II.1(b)] the line marked ld delineates the
paraelectric and ferroelectric domain phases, while the one marked lms indicates the boundary of
the metastability regions of the monodomain state. The unusual “re-entrant” type shape of
monodomain ferroelectric-paraelectric phase transition boundary [shown in figure II.1(c) for the
open-circuited boundary conditions) originated from the temperature dependence of the
screening length. Further the transition to poly-domain state occurs with either temperature or
film thickness decrease. Note that the approximation
pk»
cr
l
(k is the screening length) is rather
rigorous for the open-circuit conditions.

10
(c) Open-circuited film
10
10
2
10
3
10
4
0
300
600
Temperature T (K)
FE film thickness l (nm)
PE
Monodo
main FE
(d)
PZT
Polydomain
FE
lcr
(a)
(b)
P
3
Screening charge layer
Screening charge layer
Figure II.1. (a) Schematic of the ferroelectric film with thickness l sandwiched between metal
electrodes with screening length λ on a misfit substrate. The misfit between the film and
substrate makes the film a uniaxial ferroelectric with a spontaneous polarization along the z axis.
(b) The phase diagram for BaTiO3/SrRuO3/SrTiO3 films in the coordinates "temperature –
thickness". Panels (a) and (b) are reprinted with permission from [Bratkovsky A M and
Levanyuk A P 2011 Phys. Rev. B 84 045401]. Copyright (2011) by The American Physical
Society [73]. (c) Open-circuited film without electrodes and with diffuse screening charge layer
in the vicinity of film surfaces. (d) Phase diagram in coordinates "film thickness – temperature"
calculated for open-circuited PbZr40Ti60O3 (PZT) film (without electrodes). Panels (c) and (d) are
reprinted with permission from [Eliseev E A, Kalinin S V and Morozovska A N 2015 J. Appl.
Phys. 117 034102]. Copyright 2015, AIP Publishing LLC[71].

11
II.2.2 Role of screening on the structure of wall-surface junction
Among the effects considered in Refs.[68, 70-81], the domain wall broadening seems universal
and especially illustrative for the physical understanding of the role of surface screening charges
on ferroelectric structures.[77] The broadening occurs because bound polarization and surface
screening charges form an electric double layer, and the breaking of this layer by the domain
wall induces stray depolarization field, which in turn changes the domain wall structure. Power
law decay of the stray field results in the power law of polarization saturation near the surface, as
compared to exponential saturation in the bulk. The relative wall broadening is the strongest in
weak uniaxial ferroelectrics and the most noticeable near Curie temperature and does not fully
relax under the temperature decrease allowing for wall pinning effects.
Below we analyze the typical model situations of the surface screening charge either
separated from the ferroelectric polarization break by an ultra-thin dead layer or physical gap of
thickness H, as well as the case of strong polarization gradient at the surface [see figure II.2].
x
z
bound charge
bound charge
surface screening charge
a
H
dielectric dead
layer
ferroelectric
L
P3(x)<0
P3(x)>0
bottom electrode
break of double
electric layer
(a)
(b)
free screening charge
electrode
double electric layer, formed by the surface
screening charges and distributed bond
charges
ferroelectric
Figure II.2. 180o-domain wall structure near the film surface. The double electric layer is
formed due to either the physical dead layer (a) or intrinsic surface effect leading to diminished
polarization at interface (b). Discontinuity of the double electric layer of screening and bound
changes results in an additional stray depolarization field. Reprinted from [Eliseev E A,
Morozovska A N, Kalinin S V, Li Y, Shen J, Glinchuk M D, Chen L Q and Gopalan V 2009 J.
Appl. Phys. 106 084102]. Copyright 2009, AIP Publishing LLC [78].

12
Within the framework of LGD theory spatial distribution of the ferroelectric polarization
component
3
P
in the uniaxial ferroelectric obeys nonlinear time-dependent equation:
( )
d
uijij E
x
P
x
P
x
P
PPPuq
T
t
P3
2
1
3
2
2
2
3
2
2
3
3
2
5
3
3
3333
32
)( =
÷
÷
ø
ö
ç
ç
è
æ¶
¶
+
¶
¶
h
-
¶
¶
z-d+b+-
a+
¶
¶
G
. (II.1)
Kinetic coefficient G is defined by phonon relaxation and hence the Landau-Khalatnikov time is
small
3
a
KG=t
far from immediate vicinity of the ferroelectric phase transition (where the
critical slowing down can occur). Coefficient
()
CT
TT
T-
a
=a )
(
explicitly depends on
temperature T, gradient terms coefficients are
0>z
and
0>h
, expansion coefficients
0>d
,
while bu < 0 for the first order phase transitions or bu > 0 for the second order ones. The
spontaneous polarization PS of mechanically free bulk system is determined from the equation
0)( 42 =
d+b+
aS
SPPT
. However, it was shown that inhomogeneous stress always exists in the
vicinity of domain wall due to the electrostriction coupling. Thus
jk
u
are the strain tensor
components,
ijkl
q
and
ijkl
c
are the components of electrostriction and elastic stiffness tensor
correspondingly related with the elastic stress tensor components
ij
s
via equation of state
ijijlkijkl Pquc s=- 2
333
. These equations should be supplemented the conditions of mechanical
equilibrium,[82, 83]
0
=¶
s¶
iij
x
, and compatibility relations for elastic field,
( )
0
2
=¶
¶¶
mklnjmnikl
xxuee
, where eikl is the permutation symbol or anti-symmetric Levi-Civita
tensor. The homogeneous strain causes the renormalization of the expansion coefficient
( )
( ) ( )
( )
1211
2
1211
1211
2
1211 2
32
2
3
4c
cqq
cc q
q
u+
+
+
-
-
+b=
b
. (II.2)
The boundary conditions for polarization in Eq.(1) have the form:[84]
.0
,0
33
3
3
23
0
3
3
13
=
÷
÷
ø
ö
ç
ç
è
æ¶
¶
l+=
÷
÷
ø
ö
ç
ç
è
æ¶
¶
l-
== Lxx
x
P
P
x
P
P
(II.3)
Extrapolation length
2,1
l
may be different for x3=0 and x3=L. Reported experimental values are
nm502 -
=l
.[29, 85] In the right-hand-side of equation (II.1) stands the depolarization field
d
E
3
,
caused by imperfect screening and inhomogeneous polarization distribution. The field is
determined from Maxwell equations.

13
Assuming, that all uncompensated polarization bond charges are localized in thin near-
surface layer (i.e.
0
)0
,(
3»
¶
>¶ zzxP
), and the stray field and polarization distributions have
the form:
( ) ( )
÷
÷
ø
ö
ç
ç
è
æ÷
÷
ø
ö
ç
ç
è
æ+g
-
÷
÷
ø
ö
ç
ç
è
ægpeg+e e
-= Hz x
zxP
zxE
g
bS
2
arctanarctan
2
,
33
0
3
, (II.4a)
()
( )
( )
( )
÷
÷
ø
ö
ç
ç
è
æ+
g+
pg+p p
eg+e e-e
-» Hz
xzx xH
xPz
xP
g
f
bf
S22
24
)sign(,
33
3333
3
. (II.4b)
Here
( )( )
42
03333 53)(1)( SS
bf PPTT d+b+ae+e=e
is the dielectric permittivity of ferroelectric. The
dielectric anisotropy factor
1133
ee=g
b
;
b
33
e
is the dielectric permittivity of the background or
reference state[86] (typically
b
33
e
£10);
0
e
is the dielectric constant,
g
e
is the isotropic dielectric
permittivity of dead layer with thickness H. Ferroelectric film thickness L is much higher than H
[see figure II.2(a)].
From equations (II.4), the emerging depolarization field induces polarization saturating as
slow as ~ 1/x even for abrupt initial domain wall. Since the factor of order ~PS/(e0e33f) is much
higher than thermodynamic coercive field, the stray field (equation (II.4(a))) would influence the
polarization distribution at distances z much higher than dead layer thickness H. In fact, for
z >> H the length H is no longer the characteristic scale of polarization (equation (II.4(b))) and
depolarization field (equation (II.4(a))), but rather determines the amplitude.
The typical view of the domain wall broadening induced by the incomplete surface
screening via dead layer is shown in figure II.3(a). Coordinate x in the figure is measured in the
correlation length units, where the correlation length
^
R
is given by expression:
( )
42
2
543)(
)(
SS
PP
qT
TR d
++b+a h
=
^
,
()
( ) ( )
( )
11
2
12
12
11
2
1211
1211
2
1211
22
3
2
3
2
c
q
cc
qq
c
c
q
q
q-
+
+
+
-
-
º
. (II.5)
Discontinuity of the double electric layer of screening and bound changes results in an
additional stray depolarization field [see figure II.3(b)]. For the case the domain wall broadening
is incorporated in the polarization profile as [75]:

14
( )
()
( ) ( )
( )
( ) ( )
( )
÷
÷
÷
÷
÷
÷
ø
ö
ç
ç
ç
ç
ç
ç
è
æ
÷
÷
ø
ö
ç
ç
è
æ
+ee+- +ee++
e
e
-
÷
÷
ø
ö
ç
ç
è
æ
÷
÷
ø
ö
ç
ç
è
æ
zeel+
zee-
-
»
^
^
^
^
S
f
f
f
S
b
b
P
dzxR
dzxR
R
d
R
x
P
z
zxP
2
3311
2
2
3311
2
33
11
033
033
3
2
2
ln
4
2
tanh
1
exp
1
, (II.6)
where the distance
( )( )
()
()
l+zee+b+az= 033
22
43 b
S
Pqd
corresponds to the apparent
separation between the surface screening charge and bulk polarization value (i.e. "gradient
scale"). Hence, the distance d plays the role of effective thickness of screening double layer. At
distances
^
>> Rx
the second logarithmic term in equation (II.6) behaves as
( )
( )
( )
2
3311
2
2dzxx f+ee+
after corresponding series expansion, defining the distribution of stray
depolarization field far from the break of double electric layer. Thus the second term in equation
(II.6) proves the power law of the of domain wall profile saturation near the surface, which is
much slower compared to exponential saturation of bulk profile
( )
^
RxP
S
2tanh
[see figure
II.3(b)].
The domain wall width calculated analytically as a function of its depth from the surface
of uniaxial ferroelectric LiTaO3 is shown in figure II.3(c) (solid curves) in comparison with
experimental data in 500 nm thick stoichiometric LiTaO3 (squires, SLT)[87] and 50nm thick
congruent LiTaO3 (triangles, CLT)[88]. Note that the domain wall profiles are strongly
asymmetric, namely at the surface z=0 the width is several times bigger than saturated “bulk”
value near the surface z=L. Dotted curves in figure II.3(c) are numerical calculations by phase
field method.[89] Thus the figure II.3(c) demonstrates that the double electric layers formed by
the bound and surface screening charges can be responsible for the domain broadening effect.
Considering the alternative explanation of the domain walls broadening near the surface in
LiTaO3 due to their interaction with defects having larger density near the surface, we note the
following. Daimon and Cho[29, 85] fabricated and examined two types of LiTaO3 films:
congruent (with numerous defects) and stoichiometric (almost without defects). Results for both
types of samples demonstrated domain walls surface broadening, proving that the effect is
defect-independent.

15
-20
0
20
-1
0
1
z=0
z=5R
^
z>>R
^
0
50
100
0
3
6
CLT
SLT
Depth z (nm)
DW width a (nm)
0
10
20
1
10
L=20R^
L=50R^
L³500R^
Coordinate x/R^
DW width a/2R^
Depth z/R^
PZT
(c)
Polarization P3/PS
(a)
(b)
Figure II.3. (a) Normalized domain wall profile
( )
S
PzxP ,
3
calculated at fixed temperature and
extrapolation length l=0.5R^. Different curves correspond to different distances z/R^=0, 5, ¥
from the surface of thick ferroelectric film. (b) The domain wall width a/2R^ calculated at level
0.76 as a function of depth z/R^ from the surface, extrapolation length l=0.5R^. Material
parameters are for PbZr0.5Ti0.5O3 (PZT). (c) Thickness of domain wall at level 0.76 as a function
of its depth from the surface of LiTaO3. Squires are experimental data from Refs.[29, 85] for
500nm thick stoichiometric LiTaO3 (SLT), triangles correspond to 50nm thick congruent LiTaO3
(CLT). Solid curves are analytical calculations for fitting parameters R^=1.3 nm and l1=0.1 nm
for SLT; while R^=0.7 nm and l1=0.1 nm for CLT, l2>>30nm. Corresponding dotted curves are
numerical calculations by phase field modelling for the same fitting parameters. Reprinted from
[Eliseev E A, Morozovska A N, Kalinin S V, Li Y, Shen J, Glinchuk M D, Chen L Q and
Gopalan V 2009 J. Appl. Phys. 106 084102]. Copyright 2009, AIP Publishing LLC [78].

16
II.2.3. Role of surface screening on wall dynamics
We further discuss the influence of surface screening charge relaxation on the dynamics of the
domain walls near the surface. Eliseev et al[77] consider the planar 180o-domain wall uniformly
moving under homogeneous external field in ferroelectric capacitor with a special attention to
the dynamics of depolarization electric field induced by the bound charge at the wall-surface
junction near the dead layer. The wall dynamics is analyzed within the full Ginzburg-Landau-
Devonshire model, thus providing insight into mesoscopic structure of moving domain wall.
The schematic representation of ferroelectric capacitor with natural or artificially
deposited dielectric dead layer and surface screening charge located at the interface between the
ferroelectric layer and the dead layer is shown in figure II.4. The structure consists of conducting
top and bottom electrodes; wide-band-gap semi-conductor ferroelectric film (f) of thickness L
with dielectric permittivity tensor
f
ij
e
[90] and interfacial surface screening charge layer that
density
),( txs
depends on time t and coordinate x; dielectric dead layer of thickness H with
isotropic dielectric permittivity eg. Note, that due to the effects of “reflections” in bottom
electrode this asymmetric system is equivalent to symmetric capacitor with two dead and
screening charge layers and thickness of ferroelectric doubled.[24] The spontaneous polarization
vector
( )
),(P,0,0 00 tx=P
is pointed either along or opposite the polar axis z and depending on
coordinate x and time t allowing for the domain wall motion, at that
0div 0=P
inside a
ferroelectric layer. Quasi-stationary Maxwell equations should be satisfied for each layer. Inside
the dead layer and outside the screening layer electric potential
j
satisfies the Laplace’s
equation. The boundary conditions of fixed top and bottom electrode potentials and continuous
potential and normal component (n) of displacement on the boundaries between dead layer (d)
and ferroelectric (f) are imposed:
( )
0, =-
=j Hzx
,
( ) ( )
0,0, -=j=+=j z
xzx
,
( )
U
Lzx ==
j,
,
),( txDD
dnfn
s=-
(II.7)

17
x
z
bond charge
bond charge
a
H
dead layer
electrode
DW
ferroelectric
L
U
E
0
P0(x)<0
s(x,t)<0
P0(x)>0
s(x,t)>0
excess charge,
caused by delay
electrode
Figure II.4.
),,(
0tzxP
is spontaneous polarization, Ef is electric field inside the ferroelectric,
voltage U is applied between the electrodes. Screening charge layer, originated from bend
bending, is localized inside ferroelectric near the dead layer boundary. Dotted line indicates the
moving boundary of 180o-domain wall (DW). The normal vector n is pointed from media 1 to
media 2. Reprinted with permission from [Eliseev E A, Morozovska A N, Svechnikov G S,
Rumyantsev E L, Shishkin E I, Shur V Y and Kalinin S V 2008 Phys. Rev. B 78 245409].
Copyright (2008) by The American Physical Society [80].
Let us consider the typical case of the uniformly moving 180-degree domain wall with
the spatially invariant shape,
)(),( 00 vtxPtxP -»
. The unknown domain wall velocity, v, should
be found self-consistently. This approximation is justified given that shape fluctuations in the z-
directions are associated with significant depolarization fields.[91]
The relaxation equation for effective surface charge density s is [77, 80]:
),()(
)
,( 0t
xvtxP
t
tx s-
-=
¶
s
¶
t
(II.8)
where the relaxation time
ree
=t
f
110
, and r is the electric conductivity of the ultra-thin
screening layer. Under the condition of full screening
( ) ( )
xPx
0
0, =s
at the initial moment of
time t=0, the solution of equation (II.8) is

18
() ( )()
ò
÷
÷
ø
ö
ç
ç
è
æt
-
-
¶-
¶
+-
=s
t
m
t
t
xtv
x
P
dtv
tv
x
Pt
x
0
0
0
'
exp
'
'
,
. (II.9)
Hereinafter the following approximations for the polarization distribution within the
domains wall will be used
( ) ( )
( )
î
í
ì<+- >--
=.0,exp1
;0,exp1
0
xax
xax
PxP
S
(II.10)
Here a is the effective half-width of the domain wall, e.g. 0.5-2 nm. Here, PS is spontaneous
polarization.
Substitution of equation (II.10) into equation (II.9) admits exact analytical expression for
the surface charge density (see equation 12 in [77, 80]). Under the condition t>>
t
, i.e. in the
stationary regime, the approximate expression for the surface charge density is:
( ) ( )
ï
ï
î
ï
ï
í
ì
<
÷
÷
ø
ö
ç
ç
è
æ÷
÷
ø
ö
ç
ç
è
æt
-
t+ t
-
÷
ø
ö
ç
è
æ-
t-t
>
÷
ø
ö
ç
è
æ-
-
t+t
+
-»
s.,exp
2
exp
;,exp
,0tvx
v
tvx
v
a
v
a
tv
x
va
v
tv
x
a
t
vx
va
v
Pt
v
xPt
xS
(II.11)
Excess charge density, defined as the difference of surface screening and bond charge densities,
( ) ( )
tv
xPt
xtx --s
=ds 0
,
),(
, is caused by the delay in screening. For the very slow moving wall
(t v®0) excess charge is absent.
The distinctive feature is the fact that the maximum of surface screening charge is located
behind the moving wall at finite width, a, and exactly at the wall in the case a®0. Hence, the
surface screening charge effect on domain wall dynamics can be described only in the case of
finite wall width (see figure II.5(a-c)). Per equation II.11, charge density
),( txds
depends only
on x-v t, i.e. describes the charge wave accompanying moving domain wall at times t>>
t
. In the
limiting case of ultra-thin (or rapidly moving) domain wall,
( )
1<<tva
, approximation
(equation II.11) works with high accuracy at distances
atv
x->
(i.e. in front of the domain
wall) even starting from the small times t<<
t
. Reasonable estimations for the domain wall
intrinsic width
5.05-=a
nm, velocity
36
10
10
-
-
-=
v
m/s and relaxation time
110
3
-=t
-
s
lead to the interval
( )
6
101
-
-=tva
, justifying the limit of
( )
1<<tva
. Thus, one can use
approximation (equation II.11) in the region
atvx ->
.

19
The normal component of the electric field inside the ferroelectric layer has the form:
033 ),,(),,( EtzxEtzxE d+=
,
g
f
g
LH
U
Ee+e e
-= 33
0
, (II.12(a))
( )
( )
( ) ( )( )
( ) ( ) ( )
gge+ge g-s-
pe -
-»
ò
+¥
¥-
LkHkLk
zLkHktktkP
kxi
kdtzxE
g
f
d
sinhtanhcosh
coshtanh),(
~
),(
~
2
exp
),,(
33
0
0
3
. (II.12(b))
Here
ff 1133
ee=g
is the dielectric anisotropy factor.
),(
~0tkP
and
),(
~tks
are Fourier images of
),(
0txP
and
( )
t
x,
s
over coordinate x. The term
),,(
3tzxEd
in equation (II.12(a)) is the internal
electric field produced by the wall and partially screened by the free charges on the top electrode.
The screening is realized by the surface screening charges
( )
tx,s
with delay determined by finite
relaxation time, t. The last term in equation (II.12(a)) is the external electric field induced by
applied bias, U. Far from the wall
g
f
S
d
LH
H
P
Ee+
e
e
-
s
®
33
0
3
as anticipated for the stationary
case.
The x-distribution of the depolarization field
),
,(
3t
zxEd
calculated at different depths z is
shown in figure II.5(d). The field is maximal behind the moving wall for finite intrinsic width a,
and achieves maximum exactly at the wall in the limiting case a®0 (corresponding to the jump
at the wall). The field maximum decreases and diffuses with increasing penetration depth z.
The effect of depolarization field on domain velocity, v, can be found in two limiting
cases, when the velocity of the domain wall is determined as the nucleation rate of at the wall –
surface junction (i) or averaged over the wall surface (ii), which in turn is determined by the full
electric field in this point. Using linear approximation the domain wall velocity dependence on
the electric field is
( )( )
î
í
ì>-m <
=n .,
,,0
thth
th
EEEvE
EE
(II.13)
where m is the wall mobility and Eth is the threshold value of electric field, E is the superposition
of external filed E0 and depolarization field value at the domain wall surface,
),()( 3zvtxEzE ddm =º
. In particular
( )
0)( 0dm
EEvE +=
for the case (i), while
( )
zEEvE dm
+= 0
)(
for the case (ii). The dependence of
dm
E
on the wall velocity v allows two-

20
point Pade approximation that quantitatively reproduces the behavior of exact expressions
equation II.12(b) in the entire region of parameters:
( )
( )
( )
1
1
33330
ln
2
1
-
-
÷
÷
ø
ö
ç
ç
è
æt-
÷
÷
ø
ö
ç
ç
è
æ÷
÷
ø
ö
ç
ç
è
æt
ge+epge
+
t+t
ge+ee
»H
va
v
a
va
v
P
E
g
f
g
g
fS
dm
(II.14)
The derived dependence of field on wall velocity is at first glance unphysical, since the length of
“tail” of uncompensated charge moving after the wall increases with velocity increase. However,
with the increase of tail length (about
tv
) the electric field tends to zero at
0
>z
(similarly to
the vanishing of the electric field outside the flat capacitor).[92] Thus, the damping role of the
internal field at the surface z=0 may be negligible for rapidly moving walls. In other words, there
is a crossover between (I) polarization screening by low-mobility charges
( )
tx,s
and (II)
polarization screening by electrode free charges with giant (in theory infinite) mobility, appeared
when the sluggish charges are belated. Below, we demonstrate the “local” effect on the domain
wall velocity v.
Obtained results are presented in figure II.5(e). At small values of external field the slope
of the velocity dependence is from several times to several orders of magnitude smaller than at
high fields (initial or ideal mobility). The dotted parts of curves correspond to the unstable
regime when the velocity decreases with field increase. The result means possible self-
acceleration of the wall near the top surface, while the effect should become negligibly small
with increasing depth, z. Given that the field strength is maximal at the wall-surface junction and
decreases with the depth increase (see curves 1-3 in figure II.5(e)), this depth dependence is
expected to lead to the domain wall bending near the surface. This effect will be compensated by
the depolarization field produced by the charge excess, and the interplay between the two will
yield the equilibrium domain wall geometry.

21
-5
0
5
-1
0
1
-5
0
5
-1
0
1
-5
0
5
-1
0
1
distance x/(v
t
)
t=0.2t
t=t
t=2t
a=v
t
a=v
t
-2
-1
0
1
0
0.5
1.
1
3
2
0
(a)
Field E*d (norm. units)
Distance x/(v
t
)
t = 2t,
a = 10
-2
v
t
0.1
1
10
10
2
10
3
10
�
3
10
�
2
0.1
1
10
10
2
10
3
1
3
2
E
0
-E
th
(kV/mm)
a = 1 nm,
t = 10
-3
s
I
II
Velocity v (mm/s)
distance x/(v
t
)
distance x/(v
t
)
a=v
t
Charge
s
(n.u.)
(d)
(e)
(b)
(c)
Figure II.5. (a-c) Solid curves are exact distribution of the excess charge
),( txds
, dotted curves
are approximate distribution of
),( txds
calculated from Eq.(11), dashed curves are the
polarization distribution
( )
tvxP -
0
at different moments of time t/
t
= 0.2, 1, 2 (from (a) to (c)),
( )
1=tva
. (d) The dependence of the normalized depolarization field at the domain wall
Sdm
f
d
PEE
330
*
ee
=
on the distance
tvx
calculated for parameters g eg/e33f=0.5,
( )
100=tgvL
and
( )
2
10-
=tva
. Curves 0 - 3 corresponds to the different depth
( )
=tgvz
0, 0.25, 0.5, 1. (e)
Dependence of the domain wall velocity v on the applied electric field calculated for mobility
m = 10-6 mm2/(V sec), PS = 0.75 C/m2, e33f=30, g eg/e33f = 0.5, t = 10-3 sec; a=1 nm. Solid curves
1, 2, 3 corresponds to the H = 1, 3, 10 nm, dashed curve for H = 0 corresponds to the case
without delay in screening. Roman numbers I and II designate two different screening regimes
separated by the unstable region (dashed curves). Reprinted with permission from [Eliseev E A,
Morozovska A N, Svechnikov G S, Rumyantsev E L, Shishkin E I, Shur V Y and Kalinin S V
2008 Phys. Rev. B 78 245409]. Copyright (2008) by The American Physical Society [80].

22
Hence the crossover between two screening regimes: the first one corresponds to the low
domain wall velocity, when the wall drags the sluggish surface screening charges, while the
second regime appeared for high domain wall velocity, when the delay of sluggish screening
charges are essential and the wall depolarization field is screened by the instant free charges
located at the electrode. The “local” motion at the wall-surface junction can be unstable for
nonzero dead layer thickness, since the internal (stray) field at the wall-surface junction
decreases with domain wall velocity increase. The instability may lead to the domain wall
surface bending and actual broadening in thick samples.
II.3. Coupling between physics and electrochemical behaviors
Mixed ionic-electronic conductors, such as solids with rechargeable ions or vacancies, which
also can be mobile, free electrons and/or holes, can display a reversible dynamics of the space
charge layers that leads to a pronounced resistive switching between meta-stable states with high
and low resistance[93-98] and unique dynamic properties (including hysteretic) electro-
mechanical response.[99-106] Though ferroelectrics with mixed type conductivity (FeMIECs)
are promising candidates for the nonvolatile and memristive [107-109] memory devices, the
physical principles of the charge transfer phenomena at their surfaces and interfaces are far from
clear.[102]
Space charge dynamics in ferroelectric thin films were studied theoretically primarily in
the framework of linear drift-diffusion Poisson-Planck-Nernst theory and diluted species
approximation.[105, 107, 110, 111] However one can expect a strong correlation between the
electrochemical and electromechanical response in FeMIECs, because the spatial gradient of
mobile species (ions, vacancies and electrons) concentration near the surface caused by
electromigration (electric field-driven for charged species) and diffusion (concentration gradient-
driven for both charged and neutral species) mechanisms can change the lattice molar volume.
The changes in the volume result in local electrochemical stresses, so called "Vegard stress" or
"chemical pressure".[112, 113] The Vegard mechanism plays a decisive role in the origin and
evolution of local strains caused by the point defect kinetics in solids.[114, 115, 116]
To understand the coupling between ferroelectric and electrochemical phenomena, one
can consider the classical tetragonal ferroelectric for which the polarization components depend

23
on the inner field E as
()
1
11
01 1EP f-ee=
and
( )
21102
1EP
f
-ee=
and
( ) ( )
( )
33303333
1,, EEPEP
bf
-ee+= rr
, where a relative background permittivity
b
ij
e
£10 is
introduced.[90] The ferroelectric permittivity
bf 3333 e>>e
. The spatial distribution of the
ferroelectric polarization
( )
zP f
3
is given by the time-dependent relaxation type Landau-
Ginzburg-Devonshire (T-LGD) equation (II.1) with the boundary conditions (II.3). The
electrochemical surface charges that partially screen the film polarization is localized at the
ferroelectric – dead layer interface. The problem geometry is the same as shown in figure II.2(a),
with the only simplification that we consider the situation far from a domain wall [117].
The system of electrostatic equations of each of the medium (dead layer and ferroelectric
film) acquires the form of Laplace equation in a dead layer and anisotropic Poisson equation in a
ferroelectric. Boundary conditions (BCs) to the electrostatic equations are the equivalence of the
electric potential to the tip electrode potential U at the dead layer/electrode interface
Hz -=
and
the equivalence of the normal component of displacements to the chemical surface charge
density
()
js
at z = 0; the continuity of the electric potential and normal component of
displacements
3303 PED +e=
and
303 ED d
ee=
at dead layer / ferroelectric interface
0=
z
; and
zero potential at the bottom conducting electrode
Lz =
.
To quantify the coupling between ferroelectric phenomena and surface chemistry and to
use a relevant model for
( )
js
, one can be based on the approach developed by Highland and
Stephenson (S&H).[118, 119] Thermodynamic theory of S&H is developed for the ferroelectric
phase transition of an ultrathin film in equilibrium with a chemical environment that supplies
ionic species to compensate its polarization bound charge at the surface. Equations of state and
free energy expressions are developed based on Landau-Ginzburg-Devonshire theory, using
electrochemical equilibria to provide ionic compensation boundary conditions. Calculations are
presented for a monodomain ferroelectric film that top surface is exposed to a controlled oxygen
partial pressure in equilibrium with electronically conducting bottom electrode [see schematic in
figure II.6(a)]. The stability and metastability boundaries of phases of different polarization
orientations ("positive" or "negative") are determined as a function of temperature and oxygen
partial pressure in dependence on the film thickness [figure II.6(b)]. At temperatures below the
thickness-dependent critical one (Tcr(h)), high or low oxygen partial pressure stabilizes positive
or negative polarization, respectively ("chemical screening effect"). In equilibrium with an

24
environment the chemical screening causes new peculiarities in the phase diagram of ultrathin
films, namely a stable nonpolar phase can occur between the ferroelectric phases with positive
and negative polarizations at fixed temperature, under the conditions of partial chemical
screening (when charged surface species concentration is not sufficient to stabilize polar phase).
A synchrotron x-ray study of the equilibrium polarization structure of ultrathin PbTiO3
films on SrRuO3 electrodes epitaxially grown on SrTiO3 substrates, as a function of temperature
and the external oxygen partial pressure (pO2) controlling their surface charge compensation
revealed that the Tcr(h) varies with pO2 and has a minimum at the intermediate pO2, where the
polarization below TC changes sign. The experiments are in semi-quantitative agreement with
the S&H model [figure II.6(c)].
(a)
(b)
(c)
Figure II.6. (a) Schematic of polarization, displacement, electric field, and electric potential in
the bulk and at the interfaces of a ferroelectric film of thickness t and polarization P.
Compensating planes of charge density σ can be considered to reside at a separation λ equal to
the effective screening length in the electrodes. Panel (a) is reprinted with permission from

25
[Stephenson G B and Highland M J 2011 Phys Rev. B 84 064107]. Copyright (2011) by The
American Physical Society [119]. (b) Predicted polarization phase diagrams as a function of
temperature and pO2 for 10 nm PbTiO3 film on SrRuO3 coherently strained to SrTiO3 (001), from
theory for ferroelectric films with ionic compensation. Solid curves show equilibrium phase
boundaries, and dashed red, blue, and white curves show metastability limits for the positive,
negative, and paraelectric phases, respectively. (c) Correspondence between regions of
continuous and nucleated switching mechanisms found previously (open symbols) and the
minimum of observed Tcr(h) (solid circles). The dashed curve is a fit. Panels (b) and (c) are
reprinted with permission from [Highland M J, Fister T T, Fong D D, Fuoss P H, Thompson C,
Eastman J A, Streiffer S K and Stephenson G B 2011 Phys. Rev. Lett. 107 187602]. Copyright
(2011) by The American Physical Society [118].
In the theoretical formalism evolved by SH the equilibrium chemical surface charge
density is controlled by the concentration of surface ions
i
q
as
[ ]
( )
åjq
=js ii
ii A
eZ
0
, (II.15(a))
where e is an elementary charge,
i
Z
is the charge of the surface ions/electrons,
i
q
are relative
concentration of surface ions,
i
A
is the saturation densities of the surface ions, at that i ³ 2 to
reach the charge compensation. The surface charge is controlled by virtual potential j acting at
the interface
0=z
:
( ) ( )
( )
ja+ ja
=jq 1
i
,
() ( )
÷
÷
ø
ö
ç
ç
è
æj
-D-
=ja Tk eZG
p
B
i
i
n
Oi
00
1
2exp
. (II.15(b))
Here
2O
p
is the oxygen partial pressure (relative to atmospheric),
i
n
is the number of surface
ions created per oxygen molecule,
00
i
GD
is the standard free energy of the surface ion formation
at
=
2O
p
1 bar and U=0, and e is the magnitude of the electron charge. Equation (II.15(b)) is
analogous to the Langmuir adsorption isotherm used in interfacial electrochemistry for
adsorption of neutral species onto a conducting electrode exposed to ions in a solution.[81]
Equations (II.15) can be incorporated in the numerical or phase-field formalism, providing
chemical BCs to complement classical fixed potential or fixed charge physical BCs.

26
II.4. Experimental macroscopic studies of polarization screening at surfaces
Due to the strong interaction between extrinsic charges and the ferroelectric surface, great care is
necessary to investigate screening mechanisms at clean ferroelectric surfaces. Even after
following cleaning procedures in an ultrahigh vacuum (UHV) environment (which itself may
lead to unintentional effects including surface reduction and cation unmixing), charges from
hydrogen and other foreign atoms are difficult to avoid and cannot be eliminated entirely. For
this reason, it is advantageous to study the ferroelectric surface in situ, after growth and within
the deposition chamber, such that exposure to uncontrolled species is limited. This was the
strategy employed by different groups studying the behavior of PbTiO3 and BaTiO3 (001)
epitaxial thin films.[121-125]. The PbTiO3 films were grown by metalorganic chemical vapor
deposition (MOCVD) in a chamber installed at the Advanced Photon Source for in situ surface
X-ray diffraction (SXRD) measurements. The BaTiO3 films were grown by pulsed laser
deposition (PLD) or oxide molecular beam epitaxy (MBE) and transferred in vacuum to an
adjacent chamber for low-energy electron diffraction (LEED) studies.
Figure II.7. Atomic structure of the equilibrium PbTiO3 surface. View at left is a section parallel
to the surface through the uppermost TiO2 plane, showing the oxygen ions which form the
c(2×2) reconstruction (lighter shade). The outermost unit cell has an antiferrodistortive structure,
obtained by 10° rotations of the oxygen octahedra around the titanium ions. Reprinted with
permission from [Stephenson G B, Fong D D, Ramana Murty M V, Streiffer S K, Eastman J A,
Auciello O, Fuoss P H, Munkholm A, Aanerud M E M and Thompson C 2003 Physica. B. 336
81-9]. Copyright (2003) with permission from Elsevier [126].

27
Regardless of ferroelectric polarization, solid surfaces and interfaces often restructure to
minimize energy and stress.[127, 128] An antiferrodistortive reconstruction occurs at the PbO-
terminated surface of PbTiO3 (001), resulting in a pattern of counter-rotated TiO2 octahedra in
the layer below the topmost PbO (figure II.7). The c(2×2) symmetry of these surface octahedra
gives rise to sharp half-order surface rods in reciprocal space that can be easily detected by
SXRD[129] and are present in both the paraelectric and ferroelectric phases for PbTiO3 films
grown on SrTiO3 (001). Note that similar antiferrodistortive reconstructions may be expected at
the PbTiO3 / SrTiO3 interface,[130] but the c(2×2) reflections rapidly disappear when the PbO
surface termination is lifted, eventually giving way to a 1×6 reconstruction.[126] In this
particular case, the in-plane surface reconstruction shows little interference with the out-of-plane
ferroelectric polarization favored by the compressive misfit strain, allowing ferroelectric order to
persist down to a thickness of three PbTiO3 unit cells.[131]
Figure II.8. In-plane reciprocal space maps around the PbTiO3 304 peak for 2.4- and 12.1-nm-
thick films at temperatures above and below the Fα to Fβ transition. Both films were grown on <
0.1° miscut substrates. Redder hues indicate higher intensity (log scale). Panels (a) and (c) are
reprinted from [Thompson C, Fong D D, Wang R V, Jiang F, Streiffer S K, Latifi K, Eastman J
A, Fuoss P H and Stephenson G B 2008 Appl. Phys. Lett. 93 182901]. Copyright 2008, AIP
Publishing LLC [45]. Panels (b) and (d) are reprinted from [Streiffer S K, Eastman J A, Fong D

28
D, Thompson C, Munkholm A, Ramana Murty M V, Auciello O, Bai G. R and Stephenson G B
2002 Phys. Rev. Lett. 89 067601]. Copyright (2002) by the American Physical Society [27].
For the low vacuum/PbTiO3/SrTiO3 (001) system, 180° stripe domains form as the film is
cooled below TC, with an in-plane period that scales with thickness according to the Kittel
law.[27] In reciprocal space, the ordered stripe domains generate diffuse intensity (satellites)
adjacent to the PbTiO3 Bragg peaks, as shown in figure II.8, which depicts results for a 2.4-nm-
thick film on the left and a 12.1-nm-thick film on the right.[45] The stripe period is inversely
proportional to the radial distance from the central Bragg peak (at ΔH=ΔK=0) to the diffuse
intensity, and the average stripe morphology can be determined by the inverse Fourier transform
of the diffuse intensity pattern, with figure II.8(a) indicating that the domain walls at 324°C are
preferentially aligned along the 100 directions. However, as the 2.4 nm film is cooled to 135°C
(figure II.8b), the satellites move in towards the Bragg peak and form a torus, demonstrating that
the stripe domains have increased their in-plane period and that their walls lose their preferential
orientation. Apart from the change in morphology, the 12.1-nm-thick PbTiO3 behaves similarly,
as shown in figures II.8(b) and (d). In all cases, the increase in stripe period upon cooling
corresponds to a factor of √2, suggesting that charge compensation has occurred at one of the
two PbTiO3 interfaces,[132] most likely at the surface due to presence of various chemical
species in the MOCVD environment. This was called the Fα to Fβ transition in Refs.[27, 45, 121,
131, 132] At even lower temperatures, the interface with SrTiO3 is also screened, leading to a
mostly monodomain state (called Fγ) with upward polarization. Depth resolved atomic positions
measured through the thickness of these Fγ films show no evidence of a dead layer at the PbTiO3
surface.[133] These studies demonstrate that even in the absence of screening charges, nanoscale
stripe domains can stabilize the ferroelectric phase in ultrathin PbTiO3 films, with half-domain
periods similar to that of the film thickness.
The behavior changes dramatically with the PbTiO3/SrRuO3/SrTiO3 (001) system, as the
presence of conducting SrRuO3 at the bottom interface means that the ferroelectric structure
depends almost entirely on the availability of screening charge at the PbTiO3 surface.[24, 118,
134] As noted in section II.3, the oxygen partial pressure can be used to chemically switch the

29
out-of-plane polarization from monodomain up at high pO2 to monodomain down at low pO2.
Scans along the 30L crystal truncation rod as the pO2 is decreased (from right to left) are shown
in figure II.9(a), for a 5-nm-thick PbTiO3 film at 740 K. The diffuse scatter from up/down
domains is seen in figure II.9(b). It is observed to reach a maximum in intensity at ~10-3 mbar,
where the intensity from the 304 Bragg reflection is minimized. This stems from destructive
interference between the up and domain domains at the Bragg reflection, while the diffuse
intensity results from the incoherent addition of intensities from the up and down domains.
Figure II.9. X-ray scattering distribution near the PbTiO3 304 peak of a 5 nm film at 740 K,
during switching by decreasing pO2. (a) Distribution in the L (out-of-plane) direction along the
Bragg rod. (b) Change in diffuse scattering around the Bragg rod in the K (in-plane) direction,
relative to high pO2 value, integrated from L = 3.82 to 3.91. Redder hues indicate higher
intensity (log scale). Panels (a) and (b) are reprinted from [Wang R V, Fong D D, Jiang F,
Highland M J, Fuoss P H, Thompson C, Kolpak A M, Eastman J A, Streiffer S K, Rappe A M
and Stephenson G B 2009 Phys. Rev. Lett. 102 047601]. Copyright (2009) by the American
Physical Society [134]. (c) Temperature vs pO2 phase diagram for 5-nm-thick PbTiO3 on SrRuO3
coherently strained to SrTiO3 (001). Color scale indicates net polarization. Symbols show points
measured and phase observed. Solid lines indicate TC, and dashed lines are boundaries of 180°
stripe domain regions. Panel (c) is reprinted from [Highland M J, Fister T T, Fong D D, Fuoss P

30
H, Thompson C, Eastman J A, Streiffer S K and Stephenson G B 2011 Phys. Rev. Lett. 107
187602]. Copyright (2011) by the American Physical Society [118].
Temperature vs pO2 phase diagrams may be constructed when the paraelectric film is
first equilibrated at elevated temperatures in a fixed pO2 and then cooled through the phase
transition. Figure II.9(c) provides the experimental counterpart to figure II.6(b) but for a 5-nm-
thick film. In this case, it is seen that TC depends strongly on pO2, decreasing by ~200 K
depending on whether or not screening can take place at the PbTiO3 surface.18 The effect is
largest for ultrathin films, as expected and discussed in section II.3. These equilibrium phase
diagrams show that while decreasing pO2 at 740 K will lead to stripe domain formation, as
observed in figure II.9(b), a similar pO2 sweep at 780 K would result in stabilization of the
paraelectric phase at intermediate oxygen partial pressures. Chemical switching can therefore
change from the more typical nucleated behavior to continuous behavior, with a boundary line
that depends on film thickness (figure II.6(c)). For films that undergo continuous switching, the
internal electric field reaches that of the intrinsic coercive field due to the absence of a nucleation
barrier.[135]
Figure II.10. Characteristic fractional-order diffraction peaks along the H 0 0 azimuth from the
4×1 reconstruction that forms under low-pO2 conditions. Inset shows L scan through the 7/4, 0, 0
peak. Reprinted from [Wang R V, Fong D D, Jiang F, Highland M J, Fuoss P H, Thompson C,
Kolpak A M, Eastman J A, Streiffer S K, Rappe A M and Stephenson G B 2009 Phys. Rev. Lett.
102 047601]. Copyright (2009) by the American Physical Society [134].

31
While changes in pO2 clearly switch the ferroelectric polarization, the nature of the
compensating species at the surface is more difficult to ascertain. To maintain the observed
polarization, the surface requires roughly ½ an electronic charge per unit cell area. In Ref.[24],
density functional theory (DFT) calculations were used to investigate the effect of OH-, O2-, and
H+ adsorbates on the PbO-terminated PbTiO3 surface. They found that an overlayer of OH- or
O2- bonded to the surface Pb cations favored upward polarization, while H+ bonded to the surface
oxygens led to downward polarization. (On the other hand, undissociated H2O and CO2 resulted
in a non-polar state.) Due to their ease of formation in reducing environments and elevated
temperatures,[136] oxygen vacancies are more likely candidates for the positive screening
charge. Since they are typically charged 2+ relative to the lattice site, only one vacancy is
necessary per four surface unit cells. Wang et al. in fact discovered the appearance of a 4×1
reconstruction at the PbTiO3 surface in reducing environments (with domains along both [100]
and [010] directions), giving rise to the quarter-integer reflections shown in figure II.10.[134]
The inset of the figure shows an L-scan through one of the reconstruction peaks; its behavior
shows that the oxygen vacancies are arranged in a two-unit-cell-thick surface layer with long-
range lateral order.
In the case of PLD-grown high vacuum/BaTiO3/SrRuO3/SrTiO3(001) heterostructures,
the BaO-terminated surface shows no change in the in-plane symmetry,[123] although out-of-
plane relaxation and rumpling may occur regardless of the ferroelectric state. Quantitative
analysis of the out-of-plane structure at room temperature by LEED was performed by fitting a
set of diffracted intensity vs. voltage (I-V) curves for different in-plane reflections. Although low
energy electrons cannot penetrate very far (roughly the top three unit cells), the BaTiO3
thicknesses were similar to the probe depth (4 and 10 unit cells). Shin et al. found the BaTiO3
films to be up polarized with bulk-like atomic displacements but that the terminating BaO layer
exhibited only slight upward distortions, possibly due to a competition between surface
relaxation (favoring a downward pointing surface dipole) and the polar field.[123]
Upon exposure to H2O (≥ 6×104 L), the situation reverses. As seen in figure II.11(a-c),
there are no substantial changes in the LEED patterns from the BaO-terminated surface prior to
exposure (a), after exposure to 6×104 L (b), and after exposure to 3×105 L, indicating that at most

32
a single (ordered) layer of H2O adsorbs onto the surface. Changes in the I-V curves with 3×105 L
are shown in figure II.11(d). Several different structural models with full oxygen occupancy
were explored; a topview of the models along with the associated Pendry factor (RP) are shown
in figure II.12(a). The relatively large Pendry factors indicate poor agreement, and the best fits
allow for a large concentration of oxygen vacancies in the topmost BaO layer (~ 70% according
to the best fit shown in figure II.12(c), with the out-of-plane displacements shown in figure
II.12(b). The primary difference with that of the clean surface is the appearance of downward
polarization in the TiO2 layer.
To provide greater insight into the adsorbed structure, Shin et al.[124] conducted a set of
DFT calculations, finding that H2O is energetically favored to dissociate on the BaO surface,
forming Ba(OH)2. The LEED results, however, suggest that the hydroxide layer is disordered,
leading to oxygen vacancies in the rumpled BaO plane and downward polarization in the
underlying unit cells. This is not unlike the results for PbTiO3/SrRuO3/SrTiO3(001), where
surface oxygen vacancies promote downward polarization.[134]

33
Figure II.11. LEED patterns for BaTiO3 films (a) before exposure, and after exposures to H2O
of (b) 6 × 104 L, and (c) 3.6 × 105 L. (d) Experimental LEED I-V curves of 10 ML thick BaTiO3
films before and after exposure to 3.6 × 105 L H2O. (e) Comparison between several (of eight)
experimental and theoretical I(V) curves obtained for the best fit model, which has 30% average
oxygen occupancy and an inverted polarization dipole. Reprinted with permission from [Shin J,
Nascimento V B, Geneste G, Rundgren J, Plummer E W, Dkhil B, Kalinin S V and Baddor A P
2009 Nano Lett. 9 3720-5]. Copyright (2009) ACS Publications [124].

34
Figure II.12. (a) Explored structural models for adsorbed H2O on the BaTiO3 (001) surface with
associated Rp factors. (b) Layer spacings in the suggested structural model. Optimized Pendry Rp
factors showing dependence on (c) oxygen occupancy in the top layer for surfaces with and
without exposure to H2O, and (d) barium occupancy in the top layer after water exposure.
Reprinted with permission from [Shin J, Nascimento V B, Geneste G, Rundgren J, Plummer E
W, Dkhil B, Kalinin S V and Baddor A P 2009 Nano Lett. 9 3720-5]. Copyright (2009) ACS
Publications [124].

35
Figure II.13. Schematic sections of the optimized surface structure of BaTiO3 (001) (a) before
and (b) after water adsorption, as determined from minimization of the Rp factor by out-of-plane
relaxation and in (b) OH− on-top Ti coverage. (c) Schematic diagram showing the two adsorption
processes leading to chemisorbed OH− at the surface of TiO2-terminated BaTiO3, either at a
vacant lattice oxygen site or on-top of a surface Ti. The smaller (red) atoms are O, and the larger
(green) atoms are Ti. Reprinted with permission from [Wang J L, Gaillard F, Pancotti A, Gautier
B, Niu G, Vilquin B, Pillard V, Rodrigues G. L. M. P and Barrett N, 2012 J. Phys. Chem. C 116
21802-9]. Copyright (2012)ACS Publications [125].
Similar studies were conducted by Wang et al. for TiO2-terminated BaTiO3 thin films
grown directly on Nb-doped SrTiO3 (001).[125] Here, the films were deposited by molecular
beam epitaxy in a system permitting in-vacuum transfer to another chamber allowing LEED and
XPS. They also found upward polarization for the as-deposited state using I-V fitting, although
the top TiO2 was slightly downward polarized (figure II.13(a)). After 3.6×105 L of H2O was
introduced into the chamber, the BaTiO3 remained up polarized (figure II.13(b)), unlike the case
of the BaO-terminated BaTiO3 film discussed above. However, the surface oxygen vacancy
concentration was here determined to be relatively small (~1%) both for the as-grown film and
after water exposure, although the LEED results suggest only ~20% of dissociated H2O. As may
be expected, oxygen vacancies serve as active sites for water dissociation (as confirmed by a
separately performed temperature programmed desorption study), but the XPS results suggest
that most of the OH- groups are bonded to the surface Ti ions, as depicted in figure II.13(c). This
chemisorption is hypothesized to be responsible for reversing the direction of the dipoles in the
surface TiO2 and underlying BaO layers shown in figure II.13(b). Interestingly, by analysis of
the so-called OII and OIII features in the O 1s spectra at higher binding energies (~530.5 eV and

36
531.5 eV, respectively), Wang et al. found that the as-grown BaTiO3 films exhibited only
slightly less OH- concentrations than H2O-exposed samples, suggesting that hydrogen can easily
be incorporated during the thin film growth of oxides.
Figure II.14. Band diagrams at ideal ferroelectric surfaces: (a) upward polarization; (b)
downward polarization. For PZT, band bending on the order of 1 eV is expected. (c) Expected
effect of polarization on core-level spectra. Panels (a-c) are reprinted with permission from
[Apostol N G, Stoflea L E, Lungu G A, Tache C A, Popescu D G, Pintilie L and Teodorescu C
M 2013 Mater. Sci. Eng. B-Adv. 178 1317-22]. Copyright (2013) with permission from Elsevier
[137]. (d) Charge transport towards surfaces and possible catalytic activities of both surfaces
(reduction for the upward polarization, oxidation for the downward polarization). Panel (d) is
reprinted with permission from [Ştoflea L E, Apostol N G, Trupină L and Teodorescu C M 2014
J. Mater. Chem. A 2 14386-92]. Copyright (2014) by The Royal Society of Chemistry [138].
Even in the absence of defects or absorbates, charge doping of the surfaces may occur in
ferroelectric films, possibly leading to conductive surfaces.[139-141] Up and down polarizations
naturally result in the type of band bending shown in figures II.14(a) and (b), respectively, and
electrons from the bottom of the conduction band (screening upward polarization) or holes from
the top of the valence band (screening downward polarization) can compensate the surface,
thereby making the surfaces active for reduction or oxidation reactions, respectively (figure
II.14(d)). As XPS can be used to measure band bending (figure II.14(c)), Teodorescu et al. used
this technique to study the surfaces of PbO-terminated PbZr0.2Ti0.8O3 (PZT)/SrRuO3/SrTiO3(001)
heterostructures grown by PLD in a separate chamber and brought in air to the XPS system.[137,
138] They compared three types of samples -- 1) “fresh” samples introduced to the XPS system

37
soon after PLD growth, 2) samples brought to the XPS system one week after growth, and 3)
samples annealed at 400°C in UHV -- and attempted to rationalize the resulting XPS
measurements shown in figure. II.15 by varying levels of carbon contamination. While the Ti 2p
and O 1s spectra shown in figures II.15(c) and (d) for the 400°C annealed film appear
comparable to others in the literature, [125] there is still a substantial carbon 1s peak, and
Teodorescu et al. assign upward polarization for this sample. Compared to this, the Ti 2p and O
1s for the fresh sample are shifted to lower binding energies, suggesting downward polarization,
and the results for the 1-week-old sample lie somewhere in between.

38
Figure II.15. XPS data for the three samples: (a) valence band spectra, with an inset showing the
region near the Fermi level inserted; note the rigid shift of the Pb 5d5/2 level with the valence
band onset; (b) C 1s; (c) O 1s; (d) Ti 2p. Reprinted with permission from [Ştoflea L E, Apostol N
G, Trupină L and Teodorescu C M 2014 J. Mater. Chem. A 2 14386-92]. Copyright (2014) by
The Royal Society of Chemistry [138].

39
Indeed, the main complication stems from the C 1s spectra shown in figure II.15(b): only
the 400°C annealed shows the expected single peak at ~284.6 eV coming from C-C bonds. The
extra high binding energy features in the C 1s spectra for the “fresh” and “1-week” samples are
also observed in the O 1s spectra (figure II.15(c)), indicating the existence of C-O bonded
molecules at the surface. As will be discussed below, such molecules are expected to adsorb
mainly to up-polarized domains. If true, attenuation of the photoelectrons by the carbon
contaminants would primarily show the behavior of down-polarized domains; only the C 1s
spectra would be shifted to higher binding energies, as shown in figure II.15(b). To explain the
evolution of the spectra from “fresh” to “1-week”, the carbon contaminants must migrate from
primarily the up-polarized domains to all domains, and the authors suggest that this results in a
loss of out-of-plane polarization.[138] While some details of this study remain unresolved, it
demonstrates many of the challenges associated with understanding ferroelectric surfaces
exposed to ambient conditions. The simple relationship between XPS binding energies and
polarization depicted in figures II.15(c) and (d) may be inverted when screening occurs via ionic
charges rather than electronic carriers.
In fact, more recent work by the same group made this clear when it showed that a PZT
film totally free of surface carbon forms a down-polarized monodomain state (opposite to that
above).[142] They achieved this by performing repeated anneals in 5×10-5 mbar of O2. However,
rather than attributing downward polarization to the presence of surface oxygen vacancies, they
suggest the existence of a significant concentration of cation vacancies that give rise to mobile
holes that screen the PZT surface. This was consistent with their observation of reduced binding
energies for all the core levels (O 1s, Ti 2p, Zr 3d, and Pb 4f) when compared with carbon
contaminated PZT samples. After dosing the clean surface with 6×103 L of CO at room
temperature, they observed sub-monolayer adsorption, with roughly 1 C or 1 CO molecule
adsorbed per 3-4 surface unit cells. Desorption is most likely in the form of CO2, removing
lattice oxygen from the film and changing the polarization state to monodomain up.

40
Figure II.16. (a) Partial pressures of CO2 isotopes versus sample temperature during CO
oxidation on BaTiO3/SrTiO3(100). Initial pressures: 2 Torr CO, 1 Torr 18O2. (b) Scheme showing
the possible elementary steps associated to the CO oxidation reaction on BaTiO3 thin films. (i)
Reaction of molecular CO with lattice oxygen; (ii) filling of oxygen vacancy by adsorbed or
gaseous oxygen; (iii) dissociative adsorption of O2 or CO; (iv) surface reaction be- tween 2O*
and C*; and (v) reaction of a molecular species (O2 or CO) with an adsorbed atom (C* or O*,
respectively). (1) and (2) depict lattice oxygen and surface oxygen vacancy, respectively.
Adsorbates are depicted with *. Reprinted with permission from [Nassreddine S, Morfin F, Niu
G, Vilquin B, Gaillarda F and Piccoloa L 2014 Surf. Interface Anal. 46 721-5]. Copyright 2014
Publisher Wiley-VCH Verlag GmbH & Co. KGaA [143].
The issue of carbon monoxide oxidation on ferroelectric surfaces was also studied by
Nassreddine et al.[143] Here, they grew BaTiO3 films on Nb:SrTiO3(001) by MBE and
transferred them ex situ to a surface analysis system attached to a catalytic reaction chamber
outfitted with mass spectrometry. The ferroelectric structure of the BaTiO3 was not described,
but the surface was TiO2-terminated as deduced from the (2×2) reconstruction observed by
RHEED. If similar to the MBE-grown BaTiO3 described by Wang et al.[125], the films are
primarily up-polarized in the as-grown state. After transfer to the analysis system, they observed
carbon contamination on the surface and trace amounts of sulfur. As expected, the carbon could
not be removed by a simple anneal in UHV (even up to 700°C), but annealing at 500°C for 1
hour at 10-6 Torr O2 left no detectable carbon signal by Auger electron spectroscopy. Figure
II.16(a) shows the partial pressure of CO2 measured by mass spectrometry when flowing 2 Torr
of CO and 1 Torr of 18O2 into the reaction chamber. First, it should be noted that there is very

41
little catalytic activity, with roughly 0.14 CO2 molecules formed per surface Ti atom per second
during the 150°C-300°C ramp. (On the other hand, Pt(111) forms ~ 3 CO2 molecules per Pt per
second). Second, the same catalytic activity was observed for bare SrTiO3(001), indicating that
ferroelectric polarization had no significant effect on reactivity. Third, the use of isotopic 18O2 as
an input gas shows that at lower temperatures (primarily < 180°C), CO2 is largely made from CO
and lattice oxygen, leaving oxygen vacancies at the surface ((i) in figure II.16(b)), which is
consistent with the Mars-van Krevelen mechanism. At higher temperatures, dissociative
adsorption of 18O2 and CO helps to fill the vacant sites (ii) and leave adsorbed O* and C* (iii),
permitting CO2 formation via adsorbate reaction (iv) or through direct combination of incoming
O2 with C* or incoming CO with O* (v) (the Eley-Rideal mechanism).[143] Although BaTiO3
did not demonstrate high activity in this study, recent calculations have predicted better results
for other ferroelectric materials and reactions when following particular polarization/catalytic
cycles.[144, 145]. Furthermore, there have been several theoretical studies on the creation of
tunable catalysts using ferroelectric oxides as supports for metal catalysts[146, 147] but
experiments along these lines have proven to be highly challenging due to the difficulty in
producing a metal film only one monolayer thick.[148, 149] A more achievable route may be
growth of ultrathin transition metal oxides onto ferroelectric surfaces.[150, 151] For additional
information regarding the activity of ferroelectric surfaces, a review by Khan et al. has recently
been published.[152]
As mentioned above, there can be a large effect of ferroelectric polarization on
adsorption, particularly for polar molecules. Yun et al. found that for pre-poled LiNbO3 (0001)
single crystals, the ferroelectric polarization can strengthen or weaken interaction with the
adsorbed molecule.[153, 154] Prior to introduction of the adsorbates, the surfaces were cleaned
of carbon by annealing in an oxygen plasma at 300°C for 1 hour. Temperature programmed
desorption results are shown in figure II.17(a), for polar acetic acid, and in figure II.17(b), for
non-polar dodecane. The different sets of curves for each pre-poled set refer to different
adsorbate coverages. As observed, the peak desorption temperature (TP) of acetic acid is roughly
100 K higher for upward polarization than for downward polarization; for dodecane, the
desorption temperatures are the same (at ~250 K; the lower peak comes from multilayer
desorption). The effect is more visible in the Redhead plot shown in figure II.17(c), where β is
the heating rate (1 K/s). The exact relationship between adsorption and polarization is difficult to

42
determine from such plots, however, as the polarization also varies as a function of temperature.
Furthermore, one must be careful to distinguish electrostatic effects from those resulting from
differences in surface structure, as noted in Ref.[155]
Similar studies by Garra et al. showed that both polarization and adsorbate coverage were
important.[156, 157] Again for polar molecules (here water and methanol) on pre-poled LiNbO3
(0001), TP was higher for the up-polarized state than the down-polarized state, as shown in
figures II.17(d) and (e). Exposures ranged from 0.01 to 1.3 L, which were then converted into
monolayers assuming a sticking coefficient of 1, as shown in these plots. Clearly the peak
desorption temperature decreases with coverage for both water and methanol, regardless of the
ferroelectric polarization direction. Since these molecules most likely adsorb without
dissociation at these temperatures, this coverage dependence probably originates from a
coverage-dependent desorption energy. This can be caused by hydrogen-driven electrostatic
interactions between adsorbates weakening the interaction with the oxide surface (regardless of
whether or not a ferroelectric material).[156]

43
Figure II.17. (a) Comparison of acetic acid TPD curves for positively (top, red) and negatively
(bottom, black) poled LiNbO3 (0001). The sequence of curves illustrates the coverage
dependence on the two surfaces (1.5×103 to 1.6×104 L); the vertical dashed lines highlight the
difference in desorption peak temperatures. (b) Dodecane TPD curves for several different
coverages (10-80 L) for positively (top, red) and negatively (bottom, black) poled LiNbO3. The
low temperature peak is due to multilayer desorption. The vertical dashed line shows that
monolayer desorption has a peak at the same temperature on both surfaces, so that differences in
the multilayer peak temperatures are due to the higher coverages on the negative surface. (c)
Redhead plots for acetic acid, 2-propanol, and dodecane on positively and negatively poled
LiNbO3 (0001) where β is the heating rate and TP the desorption peak temperature. Panels (a-c)
are reprinted with permission from [Kakekhani A, Ismail-Beigi S and Altman E I 2016 Surf. Sci.
650 302-16]. Copyright (2016) with permission from Elsevier [155]. Dependence of TPD peak
temperature on adsorbate coverage for (d) water and (e) methanol. The initial coverage is given
in monolayer units (ML) and was calculated by assuming a sticking coefficient of 1.0 for both

44
molecules on both surfaces. For most coverages, the desorption temperature (TP) for the positive
surface (filled circles) is 10– 20 K higher than for the negative surface (empty circles). Panels (d)
and (e) are reprinted with permission from [Garra J, Vohs J M and Bonnell D A 2009 Surf. Sci.
603 1106-14]. Copyright (2009) with permission from Elsevier [156].
Finally, it should be noted that X-ray and electron probes can themselves affect the
behavior of ferroelectric surfaces. Popescu et al. found that the degree of band bending gradually
disappears during (focused) soft X-ray irradiation due to the generation of electronic
carriers,[158] while Husanu et al. observed increased out-of-plane polarization with time and
gradual Pb metal precipitation on PZT(111).[159] Wang et al. found that the high flux from
(hard) synchrotron X-rays tend to accelerate the kinetics in evolution of the lattice parameter
with changes in pO2.[134] This is not surprising given that X-rays continuously generate
photoelectrons during illumination. Rault et al. even used low energy electrons (< 10 eV) to both
observe and manipulate the in-plane polarization of BaTiO3 (001).[160] In addition, photons and
electrons have been known to stimulate both cation and anion desorption from surfaces.[161,
162]
III. Local studies of polarization screening
The classical surface science methods are best suited to spatially uniform systems, whereas
domain formation necessitates probing surface properties within a single domain. In this section,
we discuss the application of scanning probe microscopy to explore domain-controlled
functionalities of ferroelectric surfaces.
III.1. SPM studies of domain structures
The ubiquitous formation of domains, the role of domain structures and their dynamics in
ferroelectric materials play in the optical, dielectric, and electromechanical functionalities of
these materials which have generated tremendous interest to local exploration of these
phenomena. These studies can be roughly divided into two broad categories, namely those aimed
at visualization of domain structures and study of the dynamics of domains under the application
of electric fields, temperature, and other stimuli, and studying local materials properties within a

45
domain. Below, we summarize some SPM-based approaches and briefly discuss their potential
for imaging and quantitative probing of ferroelectric properties and screening phenomena. The
reader is referred to other sources for discussion of electron beam and other imaging
methods.[163, 164]
The characteristic feature of domains in multi-axial ferroelectrics is the formation of
surface corrugations at the ferroelastic domain walls, directly related to the crystallographic
structure. These topographic structures are directly amenable to contact and intermittent contact
atomic force microscopy, as reported by a number of authors.[165-174] Beyond imaging
domain structures, high-resolution SPM was proposed as a tool to explore the internal structure
of the domain wall from direct comparison of topographic profile and predictions of Ginzburg-
Landau theory.[175]
While simple to implement, the topographic studies do no not allow (in the absence of
build-in electric fields) to distinguish antiparallel domains, since these do not change surface
topography. More significantly, observation of domain-related topography requires cooling of
ideally flat surface through ferroelectric phase transition. Polishing of material below TC or
polishing material of multi-domain state with subsequent cycling through TC gives rise to
complex surface topographies formed by “ghosts” of pre-existing domains and corrugations at
newly formed domains. Similarly, this mode of domain imaging is inapplicable for surfaces with
significant polishing damage. Finally, observation of topographic structure generally does not
provide insight into the screening mechanism. As an interesting exception, the stripe domain
patterns were directly measured by tapping mode AFM and, further, control of stripe domain
patterns was found to be controlled by a vicinal surface through temperature dependent
measurements by synchrotron x-ray scattering.[45] These domain patterns could be correlated
with screening behavior.

46
Figure III.1. Room-temperature tapping mode AFM images of surface steps and 180° stripe
domains for epitaxial PbTiO3 films on SrTiO3 (001). Images are 500 × 500 nm2 with borders
aligned close to <100> directions. [(a) and (b)] 10-nm-thick films, height image (color scale
range: 0.8 nm) cooled from 735 °C at 0.2 °C/ s and from 655 °C at 1 °C/ s, respectively. (c) 5 nm
phase image (3° range). (d) 10 nm film on 0.5° miscut substrate height image (2.0 nm range).
Reprinted with permission from [Thompson C, Fong D D, Wang R V, Jiang F, Streiffer S K,
Latifi K, Eastman J A, Fuoss P H and Stephenson G B 2008 Appl. Phys. Lett. 93 182901].
Copyright 2008, AIP Publishing LLC [45].
As an additional modality, a number of authors reported observation of ferroelectric
domains in the lateral force microscopy modes though the detection of polarization-dependent
friction forces.[173, 176-180] While, in semiconductors, friction forces recently were correlated
to the surface electronic structure using quantitative data obtained using ultra-high vacuum
atomic force microscopy (AFM) on atomically clean surfaces,[181-183] analogous studies in
ferroelectrics were performed almost exclusively in ambient environment, hindering quantitative
and even qualitative interpretation. Exemplarily, most of friction force studies of ferroelectric
surfaces were done in late 90ies, with recent decade seeing rapid growth of piezoresponse force
(PFM) and Kelvin probe force microscopy (KPFM) studies as described below.

47
A number of authors have explored the mechanical properties of ferroelectric surfaces
using techniques such as frequency-modulation atomic force acoustic microscopy (AFAM),[184,
185] and amplitude based AFAM.[186-190] However, the fundamental limitation of these
studies is the extreme sensitivity of these methods to surface topography, owing to the fact the
tip-surface contact stiffness that determines the resonance frequency shift of the cantilever is a
product of contact area and mechanical properties of material. These two factors cannot be
separated, giving rise to strong direct topographic cross-talk.[191] Furthermore, mechanical
properties of ferroelectrics (for ideal surface) are expected to be similar for antiparallel
polarization orientations and only ferroelastic domains expected to offer small changes in elastic
properties due to the anisotropy of corresponding strain tensors and contribution of piezoelectric
and dielectric properties to corresponding indentation moduli.[192, 193] Hence, AFAM and
similar studies in certain cases can provide qualitative data on domain configurations, and
quantitative information on local mechanical properties. However, inherent error in these studies,
relatively weak sensitivity of elastic properties to electrostatic phenomena render these methods
poorly suited for exploration of surface screening mechanisms.
The characteristic feature of ferroelectric materials is the broad range of electric
functionalities, most notably piezoelectric and electrostrictive electromechanical couplings in the
bulk and presence of surface and interface charges associated with polarization discontinuities.
These factors naturally allow for probing ferroelectric materials using voltage modulated SPMs,
generally based on measuring oscillatory cantilever response to bias applied to the probe. Some
of the notable early works include that of Takata, Hong, Franke and others using complex semi-
contact voltage and mechanically modulated signals.[194-197] However, it was rapidly
recognized that the presence of multiple possible signal sources necessitates decoupling of
electromechanical (piezoelectric and electrostrictive) and electrostatic (polarization charges and
work function) interactions. In response to this challenge, two families of SPM techniques have
emerged - the contact mode based PFM and a family of non-contact voltage modulated
techniques including amplitude modulated KPFM (AM-KPFM) (also known as scanning surface
potential microscopy),[198, 199] frequency modulated KPFM (FM-KPFM),[200-204] and
electrostatic force microscopy (EFM).[205] These two families of technique are presently the
mainstay of local ferroelectric characterization.

48
The PFM and related spectroscopic modes are based on the direct measurement of bias-
induced surface deformation (electromechanical strain). Typically, the measurements are
performed in the strong indentation regime using stiff cantilevers, minimizing the contribution of
electrostatic cantilever-surface interactions, as analyzed by several groups.[206-212] Due to its
simplicity and high spatial resolution (on the ~10 nm scale for many materials), PFM has
emerged as a primary method for probing nanoscale phenomena in ferroelectrics in the last
decade. Applications of PFM for imaging and control of ferroelectric domains is summarized in
a number of archival[213-215] and recent reviews.[216-224] The remarkable aspect of PFM is
that it is sensitive to electromechanical response of material, and provides only indirect
information on the surface state and screening mechanisms. The multitude of available studies
now suggest that while direct information of surface screening cannot be obtained from PFM,
these processes nevertheless strongly affect or even control dynamics of ferroelectric domains
induced by the bias applied to PFM tip, and hence kinetic studies of domain growth and PFM
spectroscopy, as will be discussed below.
Complementary to PFM is non-contact voltage modulated techniques sensitive to stray
electrostatic fields (polarization and screening charges, work function) and dielectric properties
of material, but insensitive to the piezoelectric properties. While these techniques typically offer
lower spatial resolution (50-100 nm), they provide direct information on charge dynamics on
surfaces[225, 226] and surface screening mechanism. Further opportunities are offered by
combinations of voltage modulation approach with other modulation modalities, as exemplified
by e.g. pyroelectric charge microscopy developed by Groten et al.[227]
The brief overview of the SPM modes for probing ferroelectric materials would be
incomplete without mentioning the near-field optical[228-230] and near-field Raman
microscopies.[231] These techniques provide high-resolution optical contrast, but generally are
more sensitive to the bulk properties of ferroelectrics. Finally, scanning nonlinear dielectric
microscopy (SNDM)[232-235] provides high-resolution information on non-linear dielectric
properties and has emerged as high resolution domain imaging mode (arguably with below nm
resolution). However, relatively limited availability of this mode greatly reduced full exploration
of its quantitative potential.
III.2. Static studies of work function

49
The physics of screening phenomena on ferroelectric surfaces and interfaces is directly linked to
the stray electrostatic fields above the surface, rendering these phenomena directly amenable to
KPFM[198, 199] and related non-contact voltage modulated techniques. In AM-KPFM, the
electric potential on a conductive SPM probe is modulated as
()
t
VVV
acdc
tip
w
cos
+=
, (III.1)
where
dc
V
is the static (or slowly changing) potential offset,
ac
V
is the driving voltage, and the
driving frequency
w
is typically chosen close to the free cantilever resonance. The tip bias
results in the capacitive tip-surface force,
( )
2
surftipzel
VVCF -
¢
=
, (III.2)
where
z
C¢
is the (unknown) tip-surface capacitance gradient,
CPD
VV ssurf D
+=
is
electrochemical potential,
s
V
is electrostatic surface potential, and
CPDD
is the contact potential
(work function) difference between the tip and surface. Depending on the experimental
configuration, the voltage modulation can be applied either during the interleave scan (i.e. when
tip retraces a pre-determined surface topography while maintaining constant separation), or
during the acquisition of topographic information (at a different frequency from that used in the
topographic feedback loop).
For a periodically modulated tip bias, the first harmonic components of electrostatic force
between the tip and the surface is
( )
( )
surfdcaczel
VVVCF -
¢
=
w
1
, (III.3)
This periodic force acting on the tip results in cantilever oscillations with magnitude
proportional to the force. In principle, the displacement of the cantilever can be measured
directly (open loop KPFM), but the signal variation in this case is affected both by surface
potential and surface topography (through capacitance gradient term). Alternatively, the signal
can be measured at each point as a function of Vdc, and surface potential can be found as an apex
of parabola equation (III.1) (if full force is measured), or intercept of line in equation (III.3) (if
the first harmonic of the force is measured). However, the elegant way for probing surface
potential obviating the contribution of (unknown) surface topography is through the use of the
feedback loop to keep the oscillation amplitude equal to zero. From equation (III.3), this
condition is achieved for Vdc = Vsurf. Hence, recording microscope-control dc potential offset on
the tip, Vdc, provides information on the (unknown) surface potential, Vsurf.

50
The interpretation of KPFM data strongly depends on materials systems. For metallic or
highly conductive surfaces, measured surface potential is a sum of work function (modified by
adsorption and screening)[236, 237] and local electrostatic potential. For semi-conductive and
dielectric surfaces, the image formation mechanism can be considerably more complex.[238,
239]
Similar principle is employed in FM-KPFM.[200-204] FM-KPFM utilizes a phase-locked
loop (PLL) to determine the bias-induced frequency shift, directly related to tip-surface force
gradient. The bias-dependence of frequency shift is nullified using a second feedback loop,
allowing direct detection of surface potential. The use of two sequential feedback loops results in
relatively slow imaging rates. However, higher localization of electrostatic force gradients
controlling frequency shift allows for very high spatial resolution compared to force-based
signal, and in several cases atomic and molecular level resolution has been reported.[199]
Similar to other SPM techniques, KPFM imaging is non-ideal technique[240] and
experimental measurements contain both systematic (feedback related) errors and topographic
cross-talk. The resolution in AM-KPFM is controlled by electrostatic interactions between the tip
and the surface containing local (tip) and non-local (cantilever components),[205, 241-243]
giving rise to a logarithmic dependence of KPFM contrast with tip-surface separation.[244] The
non-ideality of the feedback gives rise to 1/Vac dependence of the measured signal[245, 246] and
strong topographic cross-talk due to variations of tip-surface capacitance on rough surfaces.[247,
248] More subtle cross-talk (indirect cross-talk[249]) will include e.g. bias and position
dependent variations of the cantilever resonant frequency in AM-KPFM.
Despite these limitations, AM-KPFM and FM-KPFM have emerged as powerful
techniques for probing static and dynamic electric and electrochemical phenomena on the
nanometer and in some cases to molecular and atomic levels.[199, 238, 250-253] Multiple
applications of KPFM to energy conversion and storage materials,[254-257] organic
materials,[237, 258-260] low-dimensional systems,[261-265] electroceramic oxides and
semiconductors[266-271] have been reported. Below, we discuss in detail application of KPFM
for probing screening physics of ferroelectric surfaces.[272-277]
The detailed study of ferroelectric domains on (001) surfaces of prototypical tetragonal
BaTiO3 single crystals polished above TC by combination of topographic AFM and KPFM was
reported by Kalinin and Bonnell.[244] The domain structure of this material is formed by out of

51
plane c domains with two possible c+ and c- orientation, and in-plane a domains with 4 possible
orientations along the (100), (-100), (010), and (0-10) axes. These domains are separated by
ferroelectric 180 walls (e.g. between c+ and c- domains) and ferroelastic 90 domain walls. For the
latter, the tetragonal symmetry of BaTiO3 unit cell results in characteristic surface corrugations
as illustrated on figure III.2. The corrugation angle is directly related to the tetragonality of the
unit call as
q
= p/2-2arctan(a/c), where a and c are the lattice parameters. Given high precision of
conventional metrological AFM platforms, the deviation of the angle from ideal (for
independently determined lattice parameters) can be used as a measure of local strains or
clamping in multi-domain structures.
As discussed in Section II.2.2., the topographic imaging distinguishes ferroelastic walls
only. At the same time, the difference in electric properties of the surface measured as effective
surface potential by KPFM can distinguish c-domains of opposite polarity and c vs. a domains,
as illustrated in figure III.2.[278] Note that potential and topographic information is insufficient
to distinguish antiparallel a domains (invisible in optical microscope) and a1-a2 domain walls
(visible in optical microscope). However, in many cases, the domain structures can be (partially)
reconstructed from the topographic and KPFM data based on the knowledge of crystallographic
compatibility and charge neutrality of the walls.
(f)
(b)
(d)
(c)
(a)
10 mm
10 mm
(e)

52
Figure III.2. (a,b) Surface topography, (c,d) surface potential and (e,f) schematics of domain
structures in (a,c,e) a-domain region with c-domain wedges and in (b,d,f) c-domain region with
a-domain wedges. Reprinted with permission from [Kalinin S V and Bonnell D A 2001 Phys.
Rev. B 63 125411]. Copyright (2001) by The American Physical Society [278].
The directly measurable parameter in KPFM is the surface potential of the surface,
directly related to work function and hence structure of surface dipole layer. For BaTiO3 (BTO)
(100) surface, KPFM imaging yields potential difference between c+ and c- domains as ~150
mV, whereas potential imaging between a and c domains is ~75 mV. Notably, surface potential
is virtually uniform within the domains with rapid changes in the vicinity of ferroelectric and
ferroelastic walls. The width of the transition is ~100-200 nm depending on imaging conditions
and represents the measure of the intrinsic resolution of KPFM, rather than intrinsic wall
width.[279], [280] These observations offer two immediate questions regarding the relationship
between measured potential and domain structure. The first question is the relationship of
measured potential, the polarization charge and screening mechanisms of the BTO surface. The
less obvious question is what is the qualitative relationship between domain potential and the
polarization direction is, i.e. whether domains positive on potential image are c+ or c-.
To address these questions, we consider semi-quantitative model for screening on
ferroelectric surface valid for domain size significantly larger then screening layer width. We
assume that the surface is characterized by the presence of polarization charge
nP ×=
pol
s
and
screening charge equivalent to surface charge density,
s
s
, of the opposite polarity. The
following cases can be distinguished:
1. Completely unscreened surface,
0=
s
s
,
2. Partially screened surface,
spol
ss
->
,
3. Completely screened surface,
spol
ss
-=
,
4. Over-screened surface,
spol
ss
-<
.
As discussed above, completely unscreened surface is extremely unfavorable from an
energetic point of view due to the large depolarization energy. An over-screened surface is likely
to occur during bias-induced domain switching, i.e. under conditions when charge injection on

53
the surface from biased SPM probe is possible or during changes in temperature or chemical
properties of the environment. Partially or completely screened surfaces are likely to be the usual
state of ferroelectric surfaces in air.
Assuming uniform charge densities, the charge distribution on a ferroelectric surface can
be represented as a superposition of a double layer of width, h, dipole moment density
[ ]
spol ,minh
ss
×
and an uncompensated charge component,
spol
ssds
-=
. For future
discussion, it is convenient to introduce degree of screening
pols
ss
a
-=
. It is further assumed
that the screening is symmetric, i.e. the degree of screening for c+ and c- domains is the same.
This assumption is supported by the experimentally observed near-equality of potential
differences between a-c+ and a-c- domains. Depending on the relative spatial localization of the
polarization and screening charges (e.g. on the polarity of dipole layer), surface potential in the
completely screened case can have the same sign as
pol
s
, or be of the opposite sign.

54
Figure III.3. (a,d) Simplified surface charge distribution, (b,e) potential and (c,f) the field in the
vicinity of ferroelectric surface for (a,b,c) unscreened and (d,e,f) completely screened cases.
Surface charge density is 0.25 C/m2, domain size 10 mm, width of the double layer 2 nm.
Reprinted with permission from [Kalinin S V and Bonnell D A 2001 Phys, Rev. B 63 125411].
Copyright (2001) by The American Physical Society [278].
To analyze the origins of image contrast in EFM and KPFM, it is instructive to calculate
the potential and field distributions above ferroelectric surface in the completely screened and
completely unscreened cases. Usually non-contact measurements are performed at tip-surface
0 5 10 15 20
-40
-20
0
20
40 10 nm
100 nm
1 mm
Potential, V
Distance, mm
0 5 10 15 20
-10
-5
0
5
10
10 nm
100 nm
1 mm
Field, MV/m
Distance, mm
0 5 10 15 20
-0.2
-0.1
0.0
0.1
0.2
10 nm
100 nm
1 mm
Potential, V
Distance, mm
0 5 10 15 20
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6 10 nm
100 nm
1 mm
Field, MV/m
Distance, mm
(a)
(b)
(d)
(e)
(c)
(f)

55
separations of 10-100 nm, which are much smaller than typical domain sizes (~1-10mm). Simple
arguments predict that the surface potential above the unscreened surfaces and electric field
above the completely screened surfaces scale linearly and reciprocally with domain size, while
electric field over the unscreened surfaces and potential over the screened surfaces are virtually
domain size-independent.
For anisotropic ferroelectrics with dielectric constants
e
xx =
e
yy =
e
x,
e
zz =
e
z the domain
structure is defined as P(x) = P z , (0<x<L/2), -P z , (L/2<x<L), where L is characteristic domain
size and is uniform in y direction. The adsorbate charge density is s(x) = -s (0<x<L/2) and s
(L/2<x<L). Potential in the air, in the adsorbate layer and in the ferroelectrics are denoted as
j
1,
j
2 and
j
3 respectively. These potentials can be found as solutions of Laplace equations
0
1
2
=Ñ
j
, for z > h,
0
2
2
=Ñ
j
, in the screening layer h>z >0, and
0
23
2
23
2
=
¶
¶
+
¶
¶zx
zx
j
e
j
e
, z < 0 (III.4)
In ferroelectrics, the corresponding boundary conditions are the continuity of potential at the
interfaces
( ) ()
hh 21
j
j
=
,
( ) ( )
00 32
jj
=
and the normal component of the displacement vectors
are as
( )
x
zz
s
j
e
j
e
=
¶
¶
-
¶
¶2
2
1
0
for z = h,
( )
xP
z
z
z
=
¶
¶
-
¶
¶
32
2
j
e
j
e
for z = 0 (III.5)
Of interest are potentials in the completely screened case, s = -P, in which case
( )
( ) ( )
÷
ø
ö
ç
è
æ+
-
÷
ø
ö
ç
è
æ+
+
+
-=
å
=Lzn
xp
Lxn
h
n
zx
zx
s122
e
122
sin
12n1
4
0
02
1
pp
eeepe see
j
(III.6)
( )
( ) ( )
÷
÷
ø
ö
ç
ç
è
æ+
÷
ø
ö
ç
è
æ+
+
+
=å
=z
x
n
zx
sLzn
xp
Lx
nh
e
epp
eeepe se
j
122
e
12
2
sin
12n14
0
02
0
3
(III.7)
The potential difference between the domains of opposite polarity is therefore:
( )
2
02
2
2
Δ
e
s
eeee see
j
h
h
zx
zx
s
»
+
=
(III.8)
Depolarization energy density for the screened case is
( )
Pdx
L
E
Lsss
el
ò
+-=
031
1
jj
or
2
2
e
hP
E
s
el
=
.
In the unscreened case (h = 0,
s
= 0), so

56
( )
( ) ( ) ( )
÷
ø
ö
ç
è
æ+
-
÷
ø
ö
ç
è
æ+
+
+
=
å
=Lzn
xp
LxnLP
n
zx
u122
e
122
sin
12n12
02
0
2
1
pp
eeep
j
(III.9)
( )
()( ) (
)
÷
÷
ø
ö
ç
ç
è
æ+
÷
ø
ö
ç
è
æ+
+
+
=å
=z
x
n
zx
uLzn
xp
Lx
nLP
e
epp
e
e
ep
j
122
e
1
22
sin
1
2n1
2
02
0
2
3
(III.10)
Depolarization energy density in the unscreened case is
( )
3
0
2
837
p
z
ee
e
zx
u
el LP
E+
=
(III.11)
where z(3) » 1.202 is zeta function.
Potential in the partially screened case is represented as the sum of the potential in the
completely screened case with a surface charge density s and in the unscreened case with a
polarization, P-
s
. Therefore, depolarization energy is
( ) ( ){ }
dxP
L
E
Lsusu
el
ò+++=
01133
1
jjsjj
(III.12)
where
u
1
j
and
u
3
j
are given by equations (III.9 and III.10) with P-
s
rather P.
Introducing the degree of screening, a, such that s = -a P, depolarization energy density
is calculated as
( ) ( )
ï
þ
ï
ý
ü
ï
î
ï
í
ì++-
+
=2
2
2
0
3
2
0
2
237
1
e
ee
a
e
e
a
p
z
a
eee
zx
zx
el hhL
P
E
(III.13)
The experimentally observed non-uniform image contrast within the domain can be
attributed either to the potential variation above the surface and corresponding change of the
capacitive interaction, or to the variation in the surface charge density and normal electric field
that results in additional Coulombic interaction between the tip and the surface. However, the
expected (almost) complete screening of polarization suggests that the dipole layer model is far
more likely. Experimentally observed potential difference of 0.140 V can be ascribed to a 0.20
nm double layer of a dielectric constant
e
1 = 80 (H2O) on a ferroelectric substrate (external
screening) or a 9.5 nm depletion layer in a ferroelectrics with a dielectric constant
e
2 = 3000
(intrinsic screening). While the former estimate is reasonable for a molecular adsorbate layer or
occupation/depletion of surface states, the latter is unreasonably small for a depletion layer width
in a semiconductor with a low charge carrier concentration (~1 mm). Moreover, potential

57
differences between c+-a and c--a domains are almost equal, suggesting that the screening is
symmetric. This is not the case if the screening is due to the free carriers in materials with a
predominant electron or hole conduction, in which the width of accumulation layer for the
polarization charge opposite to the majority carrier charge and width of depletion layer for the
polarization charge similar to the majority carrier charge are vastly different. Thus, surface
adsorption can be expected to be the dominant mechanism for polarization screening on a
ferroelectric surface in ambient conditions, though a minor contribution from intrinsic screening
cannot be excluded. The ionic nature of screening charges is consistent with multiple other
observations on non-ferroelectric surfaces, including charge retention on electronically
conductive doped SrTiO3 surfaces and potential inversion on the grain boundaries in
electroceramic materials.[60]
Following initial studies, the surface potential on pristine ferroelectric surfaces was
explored by several groups, most notably Rosenman and Rosenwaks and Kitamura. Exemplarily,
Shvebelman et al.[275] explored potential contrast on KTiOPO4 surface and observed potential
difference between the domains of the order of 40 mV. This contrast was attributed purely by
internal screening by mobile ions, with corresponding screening length of the order of 10 nm. In
comparison, Kitamura group has explored surface potential of LiNbO3 under different
conditions, and observed strong dependence of KPFM contrast on environment, as could be
expected for screening by ionic species.[281]
Furthermore, the surface potential dependence on the microstructure of the films was
explored by several groups. Choi et al. reported strong dependence of surface potential on grain
size in the polycrystalline materials.[282] Kim et al. further have explored the origin of its
influence on the surface potential in polycrystalline thin films and found that grain boundary can
be responsible for the local surface potential distribution.[283, 284] While the epitaxial films
show rather uniform surface potential distribution, the polycrystalline films show local surface
potential distribution over the whole film surface. Since grain boundaries can act as conducting
paths of current flow, electric charge can easily move through the grain boundaries during and/or
after switching procedure.
Finally, surface potential of ferroelectric polymers was explored in a series of work by
Matsushige et al.[285, 286] The authors ascribe the experimentally observed negative work
function of the in-situ deposited PVDF to the molecule-substrate interactions that results in the

58
preferential dipole orientations. The comparison between PFM and KPFM data illustrated the
presence of frozen dipoles in the material, and also illustrate that polarization switching results in
injection of additional charges on material surfaces, i.e. formation of over-screened regions. This
behavior was later found to be universal for tip-induced polarization switching, as discussed in
detail in Section II.2.4.
Further, Iwata el al.[287] explored surface potential of domain wall in Pb(Zn1/3Nb2/3)O3
(PZN)- PbTiO3 (PT). The surface potential lowering in the a-domain was larger than that in the
c-domain, the behavior attributed to the fact that the dielectric constant perpendicular to the
ferroelectric polarization is larger than the parallel one.
III.3. Domain dynamics under constant conditions
Significant insight in the mechanisms of screening phenomena can be inferred from the
observations of surface potential dynamics. Indeed, potential over static domain structures
correspond to the thermodynamic minimum, whereas details and kinetics of screening
phenomena are not observable. At the same time, perturbation of the domain state though
domain wall motion, temperature change, or external field displaces the system form equilibrium
state and allows direct observation of the kinetics and spatial localization of screening processes.
These observations are summarized in this section.
Insight into screening mechanism can be obtained from observation of potential evolution
during domain wall motion driven by wall tension (note that application of external electric
fields via tip or lateral electrodes will result in the charge injection and surface electric fields and
field-driven propagation of screening charges, as was observed for non-ferroelectric surfaces).
Figure III.4 illustrates KPFM images of c+ - c- domain structures in BTO (100) obtained at a 12 h
interval. The shrinking of the negative domain results in a dark rim in the direction of domain
wall motion. The formation of the rim can be understood assuming that the screening charge
relaxation is a slow process as compared to polarization dynamics. Interestingly, simple
considerations imply that a negative rim in the direction of wall motion is possible only if
domain related potential features are determined by the screening charges, rather than
polarization charges. In other words, for c+ domain with positive polarization charge surface
potential is actually negative, whereas for c- domain with negative polarization charge the
surface potential is positive.

59
Figure III.4. (a,d) Surface potential images of c+-c- domain region BaTiO3 (100) acquired at 12
h interval, (b,e) average profiles along the boxes and (c,f) the scheme of surface charge
distribution. Reprinted with permission from [Kalinin S V and Bonnell D A 2001 Phys, Rev. B
63 125411]. Copyright (2001) by The American Physical Society [278].
While at the first glance surprising, this conclusion is fully consistent with the charge
distribution of fully screened surface. Indeed, the screening charge is equal in magnitude to
polarization charge and is located outside of the ferroelectric material, leading to the opposite
potential of dipole layer compared to sign of polarization charge. Interestingly, this behavior is
ubiquitous for oxide surfaces in ambient and this charge inversion was reported to the grain
boundaries in electroceramic materials, etc. It is also important to note that this behavior can be
observed only for well-equilibrated surface in the absence of charge injection (e.g. from in-plane
electrodes or SPM tip), since in the latter case the strong Coulombic forces form injected charges
dominate the signal.
III.4. Lateral fields
Many applications of ferroelectric materials are based on the capacitor geometry, referring to the
ferroelectric slab between the parallel electrodes. However, these systems are poorly amenable to
(a)
(b)
(d)
(e)
(c)
(f)

60
experimental observations. The need for understanding of these system as well as emergence of
ferroelectric race track memory like devices nucleated research effort in studying ferroelectric
devices between interdigitated electrodes.
III.4.1. Lateral switching
Laterally applied fields through in-plane capacitor structures could allow exploring a direct
domain wall motion and growth behavior. Balke et al.[288] directly observed nucleation sites as
well as forward and lateral growth stages of domain formation by combination of vertical (out-of
plane) and lateral (in-plane) PFMs in the planar electrode structures of BiFeO3 (BFO) as
presented in figure III.5. The authors found that the location of the nucleation sites is correlated
with a switching direction and the nucleation can be controlled by the bias history of the sample.
Based on the results, the authors also demonstrated the manipulation of the nucleation through
the domain structure engineering. In a similar planar electrode structure, Xi Zou et al[289]
explored polarization fatigue mechanism in BFO by combination of PFM and KPFM. The
authors found that the charged domain walls were formed due to the interaction between local
space charges and polarization charges during the cyclic switching and, further found that the
fatigue occurred due to the negative charge accumulation at the electrode/film interfaces.
Figure III.5. IP PFM images showing the switching sequence for positive switching voltages.
Reprinted with permission from [Balke N, Gajek M, Tagantsev A K, Martin L W, Chu Y,

61
Ramesh R and Kalinin S V 2010 Adv. Funct. Mater. 20 3466-75] Copyright 2010 Publisher
Wiley-VCH Verlag GmbH & Co. KGaA [288].
Whereas figure III.5 present rather usual domain wall motion, unusual domain wall
motion has been reported in the planar electrode structures as well.[290, 291] McQuaid et
al.[291] reported spontaneously formed flux-closure patterns after removal of a uniform applied
electric field. Further, Sharma et al.[290] observed a unique domain wall motion of
superdomains, which are distinct bundles of a-c domains, in thin single-crystalline lamellae of
BaTiO3.
Figure III.6. (a) VPFM amplitude image of the initial domain structure of the BaTiO3 lamella
indicating the presence of a single superdomain. (b) LPFM amplitude (left panels) and phase
(right panels) imaging of the area marked by a dotted line in (a) showing nucleation and growth
of a new superdomain. The black arrows to the right of the amplitude panels indicate the slow
scan direction during the PFM imaging (which could be used to deduce the time fl ow direction).
In (i) and (ii), a small nucleated superdomain, not yet resolved in the phase signal, is revealed by
the local reduction in LPFM amplitude signal. In (iii), the growth of a needle superdomain is
apparent and the lateral movement of two distinct boundaries is observed in (iv) and (v). (c)
VPFM amplitude image of the domain structure after backswitching where one of the

62
superdomain boundaries is faintly visible. (d,e) A schematic illustration of the domain switching
process visualized in (b). Reprinted with permission from [Sharma P, McQuaid R F P, McGilly
L J, Gregg J M and Gruverman A 2013 Adv. Mater. 25 1323-30]. Copyright 2013 Publisher
Wiley-VCH Verlag GmbH & Co. KGaA [290].
III.4.2. Lateral charge injection
As briefly mentioned in section III.4.1, the application of electric field can induce charge
injection at lateral electrodes, that can clearly be visualized as injection and propagation of
charged regions.[261] As shown in figure III.3, charge injection from the lateral electrode and
propagation of the charges regions were observed in the surface potential images. As increasing
the biasing time from 10 min to 1 h shown in figures III.7(a-c), the surface potential contrast at
the nanowire becomes more faintly due to the surface charge mobility.[261] Further, after the
negative bias was switched off, the dark contrast in the surface potential image [see figure
III.7(e)] gradually becomes wider due to the surface charge mobility as well.
Figure III.7. Surface potential on the biased nanowire after (a) 10 min, (b) 20 min, and (c) 1 h
scanning illustrating the smearing of potential contrast due to the mobile charge effect. (d)
Surface potential at -10 V and (e) immediately after the bias is off illustrates the formation of
charged negative halo. (f) After a week time, the halo disappears. The scale is 6 V [(a)-(c)], 10 V
(d), and 1 V [(e) and (f)]. Reprinted with permission from [Kalinin S V, Shin J, Jesse S,

63
Geohegan D, Baddorf A P, Lilach Y, Moskovits M and Kolmakov A 2005 J. Appl. Phys. 98
044503]. Copyright 2005, AIP Publishing LLC [261].
As a historical parallel, Shockley and Prim[292] reported space-charge limited emission
for a planar electrode structures consisting of semiconductor layers. Strelcov et al.[293] reported
that the charge injection can induce an metal-insulation transition because of removal of the
Coulomb gap when charge injection increases the carrier concentration beyond a certain critical
value. These charge injection were recently visualized by time-resolved KPFM (Tr-KPFM).[294-
296] Figure III.8 presents voltage and time dependent surface potential behavior of which
information can delivery charge dynamics in the planar electrode structures. The authors found
that, while surface potential relaxation is correlated with surface charge motion at low biases and
temperatures, ionic charge dynamics can be a dominant role in conductivity at high biases and
temperatures.

64
Figure III.8. (a-c) Line-averaged surface potential vs time and position, and interelectrode
current vs time as measured at room temperature with 5Vappliedbetweenthe electrodes. (d-f)
Same with 90Vappliedto electrodes; gray rectangles show positions of the lateral electrodes.
Arrows and color bars in (b,e) indicate time sequence of the curves. Averaging was performed
over spatial locations equidistant from the grounded electrode, which created a set of “lines”,
whose positions are indicated with the color bars in (a,d). Reprinted with permission from
[Strelcov E, Jesse S, Huang Y, Teng Y, Kravchenko I I, Chu Y and Kalinin S V 2013 ACS Nano
7 6806-15]. Copyright (2013) ACS Publications [295].
III.4.3. Charge dynamics on ferroelectric surfaces
An interesting set of studies of surface potential behavior on ferroelectric surfaces as affected by
lateral electric field and as a function of humidity was recently reported by Volinsky group.[297-
299] The authors reported complete screening (disappearance of surface potential contrast) for
strong humidity[297] and emergence of unscreened component for lower humidity. The authors
also reported that application of strong electric bias across ferroelectric surface leads to
reversible inversion of surface potential, as illustrated in figure III.9.[297, 298] This behavior can
be attributed to partial removal of screening charged by sufficiently strong lateral electric field,
leading to preponderance of Coulombic forces due to unscreened polarization charge.
Figure III.9. (a) Topography image of the c domain. Surface potential map of the c domain at
(b) 18.1 V and (c) 18.2 V showing surface potential inversion. (d) Complete recovery upon

65
switching the electric field off. Reprinted with permission from [He D Y, Qiao L J and Volinsky
A A 2011 J. Appl. Phys. 110 074104]. Copyright 2011, AIP Publishing LLC [297].
Strelcov et al.[294] reported charge injection through the lateral electrodes on the LiNbO3
surface. They performed time-resolved KPFM at different polarizing biases. As shown in figure
III.10(a), surface potential distribution for polarizing biases lower than 30 V is linear and there is
only a very slight potential fluctuation on the line profile. However, as increasing the polarizing
biases, surface potential curve grows larger and laterally extends from the biased electrode (see
50 and 70 V in figure III.10 (b,c). This difference between the highest surface potential and the
polarizing bias at the electrode becomes larger. This behavior is attributed to the charge injection,
which can be either electronic or ionic nature, from the biased electrode. Further, they directly
showed the rearrangement of the screening charges by an application of the bias through the time
dependent transient behavior.
Figure III.10. Time evolution of the line-averaged surface potential and potential-distance profi
le as measured at the polarizing biases of (a, d) +30 V, (b, f) + 50V, (c, g) +70 V; electrodes
positions are shown with golden rectangles; color bars in (a–c) indicate distance from the biased
electrode; color bars in (d–g) indicates time; measurements performed with a flexible cantilever.
Reprinted with permission from [Strelcov E, Ievlev A V, Jesse S, Kravchenko I I, Shur V Y and

66
Kalinin S V 2014 Adv. Mater. 26 958-63]. Copyright 2014 Publisher Wiley-VCH Verlag GmbH
& Co. KGaA [294].
The authors further explored electrochemical polarization and relaxation. To quantify the
electrochemical polarization/relaxation processes, they used a phenomenological model by
fitting the surface potential evolution with an exponential decay function as follow:
t
f
t
eBA
-
×+=
(3)
where φ is the surface potential, A is the offset, B is the pre-exponential factor, and τ is the mean
lifetime. As shown in figure III.11(a), transient surface potential was observed because the bias
application to the electrodes rearranges the screen charges through the charge injection. Further,
as seen in figure III.11(b), the central domain has lower surface potential due to incomplete
switching.
Figure III.11. Electrochemical-polarization by a 90 V bias pulse: (a) time-averaged surface
potential distribution; (b) flattened time-averaged surface potential distribution showing
ferroelectric domains. Fitting coefficients maps for: (c) electrochemical-polarization; (d)

67
electrochemical-relaxation. The lower part of the images was recorded on the grounded
electrode; the top row was adjacent to the biased electrode. The scale bar is 10 μm. Reprinted
with permission from [Strelcov E, Ievlev A V, Jesse S, Kravchenko I I, Shur V Y and Kalinin S
V 2014 Adv. Mater. 26 958-63]. Copyright 2014 Publisher Wiley-VCH Verlag GmbH & Co.
KGaA [294].
III.5. Variable temperature experiments
The second aspect of domain and screening charge dynamics is their evolution with temperature.
The temperature-dependent crystallographic structure and polarization in ferroelectrics are
typically when known from scattering studies and can be with high degree of prediction be
estimated from Ginzburg-Landau type theories. Hence, exploring kinetics and thermodynamics
of surface properties (potential, piezoresponse, and topography) during the temperature changes
provides direct insight into screening charge behavior.
The first studies of temperature-dependent ferroelectric properties were reported by
Abplanalp by PFM[300] and Hamazaki et al.[301] by topographic imaging. In their original
publication,[301] they explored temperature dependence of surface topography of materials such
as BTO and NaKC4H4O6∙4H2O, establishing excellent agrement between expected and
experimentally measured corrugation values at the ferroelastic domain walls. Extensive imaging
studies of temperature induced domain dynamics by PFM were reported by Allegrini group.[302,
303] They have reported domain formation and relaxation kinetics subjected to thermal cycles,
behavior relevant to the local defect states. This early work was predominantly focused on the
evolution of domain structures inside the material, however, quantitative studies of domain
specific response as related to image formation mechanisms in SPM and screening mechanisms
were of less interest.
The systematic study of domain specific properties of ferroelectric surfaces by KPFM
and PFM were reported by Kalinin and Bonnell274, 275, 306 as discussed below. In particular, here
we summarize the observations of domain-specific potential evolution on heating and cooling
through ferroelectric phase transition that led to observation of spurious potential increases above
TC, heating and cooling below phase transition that lead to observation of temperature induced
potential inversion (TIPI), discuss these observation in the context of qualitative screening
model, and discuss analysis of thermodynamics of screening process from this data.

68
III.5.1. Surface potential evolution on heating
Polarization and charge dynamics on ferroelectric BTO (100) surfaces during on heating through
ferroelectric phase transition were explored in Refs.[274, 304, 305] The temperature dependence
of topographic structure and surface potential is illustrated in figure III.12. The surface
topography illustrates the presence of corrugations due to ferroelastic domain walls between a
(in plane) and c (out of plane) domains. While the relative number and orientation of domains
does not change below TC= 130°C, the surface corrugation angle, which is directly related to c/a
ratio in the tetragonal unit cell, changes with temperature as shown in figure III.13.[306] Note
the agreement between experimentally measured corrugation angle and the value calculated from
the temperature dependence of a/c ratio in BTO, suggesting that domain structure is unclamped.
The corrugation angle is established immediately after temperature change and remains constant
under isothermal conditions.
The initial observations of domain potential contrast exhibit more complex dynamics, as
illustrated in figure III.13(b).[304] In this case, a stepwise increase in temperature results in an
increase of domain potential contrast, whereas isothermal annealing for ~30 min results in
decrease of potential contrast. This behavior is somewhat unexpected, since polarization and
hence surface polarization charge decreases with temperature, in agreement with both bulk data
and measured corrugation angles.

69
Figure III.12. (Top) Surface topography and (bottom) potential distribution at BTO (100)
surface before ferroelectric phase transition at (a,b) 125°C, (c,d) 4 min after transition and (e,f)
after 2.5 h annealing at 140 °C. Scale is (b) 0.1 V, (d) 0.5 V and (f) 0.05 V. Reprinted with
permission from [Kalinin S V and Bonnell D A 2001 Phys. Rev. B 63 125411]. Copyright (2001)
by The American Physical Society [278].
Figure III.13. (a) Temperature dependence of average corrugation angle on heating (▲) and
cooling (▼) as compared to the calculated value (dashed line) and (b) domain potential contrast
(a)
(c)
(b)
(e)
(d)
(f)
10 mm

70
in KPFM measurements below Curie temperature TC. Panel (a) is reprinted with permission from
[Kalinin S V and Bonnell D A 2000 J. Appl. Phys. 87 3950]. Copyright 2000, AIP Publishing
LLC [305]. Panel (b) is reprinted with permission from [Kalinin S V and Bonnell D A 2001 Appl.
Phys. Lett. 78 1116]. Copyright 2001, AIP Publishing LLC [304].
Even more intriguing phenomena are observed on ferroelectric transition during heating
above TC. In this case, ferroelectric polarization disappears as indicated by the absence of
characteristic surface corrugations. However, the domain related potential features persist and
potential amplitudes increase by almost 2 orders of magnitude. These potentials are metastable
and rapidly decay with time, as illustrated in figure III.12 (image was acquired from bottom to
top 4 min after the transition, total acquisition time - 11 min). After annealing for ~2 h, domain
related potential contrast disappears.
Figure III.14. (a,c,e) Surface topography and (b,d,f) surface potential distribution on BTO (100)
surface above (a,b) TC, (c,d) during the transition and (e,f) 1 h after the transition. Images are
acquired from bottom to top. Scale is (a,c,e) 30 nm, (b) 0.05 V, (d,f) 0.1 V. Reprinted with
(a)
(c)
(b)
(e)
(d)
(f)
10 mm

71
permission from [Kalinin S V and Bonnell D A 2000 J. Appl. Phys. 87 3950]. Copyright 2000,
AIP Publishing LLC [305].
The sequence of events on the reverse transition is similar to those in the forward
transition. Figure III.14(a-f) shows surface topography and surface potential distribution above
the transition temperature, during the transition, and below the transition. It can be seen that
during the transition the apparent topography is very volatile for ~30 s, then the new domain
structure forms. New domains are oriented in the same direction as before the first transition;
however, the size of the domains differs. At the transition surface potential exhibits large
unstable potential amplitudes that may be attributed to depolarization currents associated with
the formation of domain structure. After a relaxation period, the surface potential stabilizes and
is again closely related to the new domain structure. Relaxation of newly formed potential
features occurs much slower than on transition to the paraelectric phase.
The observation of slow potential dynamics on heating and spurious potentials above TC
can be explained only assuming that the measured potential values are due to the interplay
between (fast) polarization dynamics and (slow) screening charge dynamics. The sign of the
surface potentials in this case is that of the screening charge. On increasing the temperature,
polarization charge decreases, whereas screening charge relaxes only slowly with time. Hence,
the increased contribution of screening charge to measured contrast results in increased surface
potential. However, the system is far away of thermodynamics equilibrium and excess screening
charge dissipates with time, leading to reduction of potential. Similar behavior is observed on
increasing the temperature above TC; however, the amount of polarization charge change in this
case is considerably larger. Hence, the spurious surface potential features represent the effect of
(now uncompensated) screening charges that can be observed directly by KPFM.
It is interesting to speculate what may be the role of these screening charges on other
properties of ferroelectric surfaces. In particular, many groups reported the observation of weak
surface ferroelectricity above bulk ferroelectric phase transitions. Given high polarizability of
ferroelectric materials and potential for field-induced ferroelectric phase transition above TC, the
presence of the surface charges can result in surface ferroelectric phase.

72
III.5.2. Temperature induced domain potential inversion
Further insight in polarization and screening charge dynamics can be obtained from variable
temperature experiments below TC, as reported in Ref.[306]. The evolution of surface potential
and surface topography for a-c domain region on BTO (100) surface containing both c+ and c-
domains is illustrated in figure III.15(a-d). On increasing the temperature, the domain potential
contrast increases, in agreement with earlier observations. Here, the relaxation of potential under
isothermal conditions was explored, and the potential was found to decay with time to a stable
and lower value. Typically, relaxation times of the order of 15 m – 1 h were observed, in
agreement with other studies of ionic charge dynamics on oxide surfaces.
An unusual behavior is observed on decreasing the temperature, as illustrated in figure
III.15(e-h). After a temperature decrease from 70°C to 50°C, the potential contrast between
domains inverts, i.e. a positive c domain becomes negative. The potential difference between the
domains decreases with time, passes through an isopotential point corresponding to zero domain
potential contrast, and finally establishes an equilibrium value. This temperature induced domain
potential inversion can be readily explained in the context of surface screening model. In this
case, decrease of temperature results in the increase of polarization bound charge. For a certain
amount of time, the Coulombic contribution of uncompensated polarization charge dominates the
signal. Accumulation of screening charges inverts the contrast back to the case when polarity is
dominated by the screening charges.
(e)
(g)
(h)
(f)
(a)
(b)
(c)
(d)

73
Figure III.15. (a) Surface topography and (b) surface potential of the ferroelectric domain
structure on a BTO (100) surface at T = 50 °C. Surface potential (c) after heating from 50 °C to
70°C and (d) after annealing at 70 °C for 50 min. (e) Surface potential at T = 90 °C. Surface
potential (f) during cooling from 90°C to 70°C, (g) at 70°C and (h) after annealing at 70°C for 50
min. Reprinted with permission from [Kalinin S V, Johnson C Y and Bonnell D A 2002 J. Appl.
Phys. 91 3816]. Copyright 2002, AIP Publishing LLC [306].
Figure III.16. Time dependence of domain potential contrast on (a) heating and (b) cooling.
Solid lines are exponential fits. (c) Time constant for relaxation process on heating in Arrhenius
coordinates and (d) temperature dependence of equilibrium domain potential contrast on heating
(▲) and cooling (▼) and fit by equation (III.14) (solid line). Reprinted with permission from
[Kalinin S V, Johnson C Y and Bonnell D A 2002 J. Appl. Phys. 91 3816]. Copyright 2002, AIP
Publishing LLC [306].
Experimentally measured kinetics of surface potential relaxation suggests that typical
relaxation times are of the order of 10 min – 1 h, allowing the isothermal kinetics studies. The

74
time dependence of domain potential contrast on heating and cooling is shown in figure
III.16(a,b).[306] The kinetics of domain potential contrast, D
j
, was found to roughly follow the
exponential law
()
tj
j
/
t
A-
+
=exp
ΔΔ 0
, where
t
is relaxation time and A is a prefactor. The
temperature dependence of the potential redistribution time is shown in figure III.16(c). The
redistribution time is almost temperature independent, with associated relaxation energy of ~4
kJ/mole, suggesting that the kinetics of relaxation process is limited by the transport of chemical
species to the surface. The characteristic redistribution time is ~ 20 min and is close to the
relaxation time for domain potential contrast above Tc (15 min). Notably, the SPM based studies
have extremely limited bandwidth (~3-10 min for imaging experiments, 1-3 s for line imaging
even excluding time for surface stabilization, typically of the order of 1-3 min). Hence, the
presence of faster relaxation processes cannot be established from SPM data and measured
dependence represents only the tails of relaxation process.
III.5.3. Thermodynamics of screening
Fortuitously, the potential relaxation on heating and cooling is sufficiently fast so that
thermodynamics equilibrium can be achieved in the experimentally accessible time scales. In
particular, the relaxation both on heating and cooling to the same final temperature result in the
same equilibrium value of domain potential contrast,
0
Δ
j
. The temperature dependence of
domain potential contrast in the temperature interval 30 °C < T < 100 °C is linear
T.
.Vdc 4
103
50590
Δ
-
×-
=
. (III.14)
Interestingly, the zero potential difference corresponding to temperature ~110 °C, well below the
Curie temperature of BTO (Tc = 130°C).
Experimentally observed dependence of surface potential on temperature suggests that
equilibrium degree of screening depends on temperature. While at low temperatures screening is
close to unity and measured potential is dominated by that of the dipole layer, on increasing the
temperature the degree of screening decreases. At 110 °C, the contributions of the dipole layer
and Coulombic forces from unscreened polarization charged are mutually compensated,
corresponding to the absence of potential contrast. At higher temperatures, the sign of measured
potential is that of polarization charges. Note that the temperature corresponding to the
isopotential point depends on humidity. While no systematic KPFM vs. T, p(H2O) studies were

75
reported, isothermal variable humidity data by Volinsky group[297, 299] suggest that at room
temperature isopotential point corresponds to 45 % relative humidity.
The observed dependence of potential contrast and degree of screening on temperature
naturally leads to the question on whether thermodynamic parameters of the process can be
obtained from these data.[306] This requires solution of two related problems, namely, (a)
establishing the thermodynamics of screening process in terms of enthalpy and entropy of
screening process and temperature-dependent ferroelectric properties of material, and (b)
establishing the relationship between the effective potential measured by KPFM and degree of
screening. Below we discuss both problems.
To describe the thermodynamics of screening surface, we assume that the surface of a
ferroelectric material is characterized by a polarization charge density
nP×=
s
, where P is the
polarization vector and n is the unit normal to the surface. The free energy for screening process
by external ionic charges can be written as:
( ) ( )
dw
aa
el EST
qN
P
H
qN
P
T,ET
,E +
-+= adsads ΔΔ
aaaa
(III.15)
where q = 1.602×10-19 C is electron charge, P is spontaneous polarization, Na = 6.022×1023 mol-1
is Avogadro number,
a
is the degree of screening and T is the temperature. Experimentally, the
degree of screening is very close to unity,
1
»
a
. Therefore, it is convenient to introduce as the
small parameter the fraction of the unscreened charge,
ag
-
=1
. The enthalpy and entropy of
adsorption are denoted DHads and DSads, respectively. Experimentally, the domain wall area is
observed to be constant during the measurement and hence the corresponding free energy, Edw, is
assumed independent on the degree of screening. The electrostatic contribution to the free
energy,
( )
TEel ,
g
, can be derived as:
( ) ( ) ( ) ( )
ï
þ
ï
ý
ü
ï
î
ï
í
ì-+-+
+
=2
2
2
0
3
2
0
211
237
,
eee
g
e
e
g
p
z
g
eee
g
zx
zx
el hhL
P
TE
(III.16)
where L is the domain size, h is the screening layer width,
2
e
is the dielectric constant of the
screening layer,
x
e
and
z
e
are the dielectric constants of the ferroelectric and
e
0 = 8.854×10-12
F/m is the dielectric constant of vacuum.

76
The temperature dependence of the equilibrium screening can be obtained from the
condition of the minimum of free energy
( )
0
,=
¶
¶
g
g
TE
. Since
( )
TEel ,
g
is a quadratic function of
g
, this condition can be written as
( ) ( )
TbTb
Eel 21 +-=
¶
¶
g
g
, (III.17)
where b1 and b2 are material-dependent constants that can be estimated from the Ginzburg-
Landau theory and parameters of domain structure. Calculated temperature dependencies of b1
and b2 for domain size L = 10 mm,
e
2 = 80 (water) and h = 0.1 nm are shown on figure III.17. It is
clearly seen that b1 and b2 are only weakly temperature dependent. The physical origin of this
behavior is that the product
e
x
e
y is only weakly temperature dependent. Consequently, b1 and b2
can be approximated by their room temperature values. At T = 25°C for h = 0.1 nm b1= 26.67
J/m2 and b2 = 0.02034 J/m2. Hence, equation (III.18) suggests that the degree of screening is a
linear function of temperature.
Figure III.17. (a) Charge distribution on the partially screened anisotropic ferroelectric surface
and (b) temperature dependence of material constants. Reprinted with permission from [Kalinin
S V, Johnson C Y and Bonnell D A 2002 J. Appl. Phys. 91 3816]. Copyright 2002, AIP
Publishing LLC [306].
From equation (III.15) the equilibrium degree of screening is defined by
( ) ( ) ( )
( ) ( )
TbS
qN
P
T
Tb T
b
TbH
qN
P
Tads
a
ads
a1
1
2
1
ΔΔ +
+-=
g
, (III.18)
where b1(T) and b2(T) are temperature dependent coefficients defined by domain structure and
material properties.

77
The second problem in the analysis of the KPFM data is the establishing the relationship
between the measured effective potential and degree of screening. This analysis can be
performed for a defined geometry of the SPM probe using image charge methods developed by
Belaidi and others.[307] For a typical metal coated tip used in the KPFM measurements with
q
= 17°, H » 10 mm and tip-surface separation z = 50-100 nm equation (III.18) can be
approximated as
( ) ( )
2121
181ΔEEH.VVV
dc
---=
, (III.19)
since the logarithmic term is only weakly dependent on the tip length. Under the experimental
conditions (lift height 100 nm), the deviation between the true and measured domain potential
difference does not exceed ~30 % and the uncertainties in the other parameters (tip shape model
and materials properties) are expected to be comparable.
Using the representation of a partially screened ferroelectric surface as a superposition of
completely unscreened and completely screened regions, the potential difference between
domains of opposite polarity is
( )
( )
zx
zx
s
Ph
eee
eee
gj
+
-=
0
2
2
1Δ
, (III.20)
while the difference in the normal component of the electric field is
zx
u
P
E
eee
g
+
=
0
Δ
. (III.21)
Hence, the measured potential difference between the domains is
( )
( )
1
0
02
4
ln
2
1Δ
-
÷
ø
ö
ç
è
æ
+
-
+
-= d
HP
H
Ph
V
zxzx
zx
dc
e
ee
g
b
ee
e
eee
g
, (III.22)
i.e. domain potential contrast is a linear function of degree of screening.
When the force acting on the biased tip above a partially screened surface is written as
ztipzsurftip
z
EVCVV
dz
dC
F+-=
2
)(
(III.23)
,where EZ is the normal component of the electric field due to unscreened polarization charge,
the combination of equations (III.22 and III.23) for a tip length of 10 mm yields the temperature
dependence of equilibrium degree of screening
T
65
1023.110627.1
--
×+×=
g
. (III.24)

78
A comparison of equations (III.18 and III.24) allows the enthalpy, DHads, and entropy, DSads, of
adsorption to be determined as DHads = 164.6 kJ/mole, DSads = -126.6 J/mole K. The enthalpy
and entropy of adsorption thus obtained are within expected values in spite of the approximations
inherent in this approach. Moreover, from equations (III.22 and III.24) the Coulombic
contribution to the effective potential can be estimated as <10-20 % thus validating our previous
conclusion that the surface is completely screened at room temperature. The nature of the
screening charges cannot be determined from these experiments; however, these results are
consistent with the well-known fact that water and hydroxyl groups, -OH, adsorb on oxide
surfaces in air. The adsorbed water can provide the charge required to screen the polarization
bound charge, since corresponding polarization charge densities are of order of 0.25 C/m2
corresponding to 2.6×10-6 mole/m2. For a typical metal oxide surface with characteristic unit cell
size of ~ 4 Å, this corresponds to the coverage of order of 0.25 ml. Dissociative adsorption of
water as a dominant screening mechanism on BTO surface in air was verified using temperature
programmed desorption experiments on poled BTO crystals. At the same perovskite (100)
surfaces are free from the midgap surface states; therefore, the screening on BTO (100) surface
cannot be attributed to the surface states filling.
III.6. Environmental effects
The screening on ferroelectric surfaces can be expected to be strongly affected by the
environment. First and foremost, the thermodynamics of screening process is directly controlled
by the chemical potential of screening species. Furthermore, the properties of the screening layer
will be strongly affected by the presence of e.g. wetting water layer, since associated high
dielectric constant will strongly enhance dissociation of neutral species in charged ions. Here, we
discuss the SPM based studies of environmental effects on screening as explored through the
measured surface potentials.
As an example of such study, Kitamura group has explored surface potential of LiNbO3
under different conditions, and observed strong dependence of KPFM contrast on environment,
as could be expected for screening by ionic species.[281] The KPFM studies of BTO surface in
ultrahigh vacuum are reported by Watanabe.[308, 309]. Interestingly, the authors report that the
potential difference between c+ and c- domains is ~100 mV, close to the value observed in
ambient environments. The authors ascribe this behavior to the intrinsic shielding mechanisms

79
and link this behavior to the formation of surface electron layers, in agreement with metallic-like
conductivity of BTO surface reported by the same group.[139, 310] Notably, metallic
conductance was recently discovered on the ferroelectric domain walls,[17] confirming this
original explanation.
However, it should be noted that potential difference between domains of opposite
polarity for intrinsic screening is expected to be of the order of band gap and hence of the order
of 3 V for BTO.[311] The experimentally observed ~100 mV potential difference between the
domains can be ascribed either to the finite resolution of KPFM, or the screening by surface
adsorbates that persist on transfer the sample from ambient to UHV environment.
Figure III.18. The time dependence of the maximum potential V normalized to the maximum
potential V0 of the first potential distribution after charging. Reprinted with permission from
[Debska M 2005 J. Electrostatics 63 1017]. Copyright (2005) with permission from Elsevier
[312].
It is also possible that charge dynamics can be affected by local defects during and/or
after switching and relative humidity.[284, 312] Since grain boundaries in polycrystalline films
are relatively conductive, as mentioned before, the grain boundaries can affect screen charge
migration on the film surfaces.[284] The charge dynamics explored macroscopically by
induction probe shows faster relaxation under the high relative humidity in triglycine sulfate
crystal (TGS) in figure III.18. Since the TGS crystal surface is dissolved by the water layer, the

80
surface properties can be modified under the high humidity which can affect surface charge
relaxation. This indicates the predominance of surface conduction under high relative humidity
in TGS.[312]
Further systematic studies of environmental and humidity effects on domain-related
surface potentials in ferroelectrics are reported by Volinsky group. The authors report complete
disappearance of surface contrast for strong humidity. [313] Finally, it is well recognized that the
thermodynamics of domain structure and hence screening can be affected not only by status of
the top surface, but also behavior on the interfaces, especially for this films. An attempt to
explore these phenomena in beveled PZT/Pt structures were reported by Lu et al.[314] The
authors report on the presence of 240 nm transition layer visible both as a gradual decrease of the
piezoresponse signal and a variation of the surface potential. The effect of this layer on retention
behavior was further explored.
Several limitations of the KPFM and related force based SPM techniques for exploring
screening phenomena on oxide surfaces should be mentioned. By definition, these techniques
measure all electrostatic forces acting on the probe, and separation of dipole layer and
uncompensated charge contributions requires numerical analysis of the tip-surface
interactions.[244, 307, 315-317]. Such de-convolution has relatively low veracity, and encounter
additional problems on topographically-non-uniform surfaces. The future progress can be
achieved by probes that can separate electric field and surface potential effects. One such
approach can be based on field effect probes. [318]
Another interesting direction can be based on using probes with defined and controlled
charge state (as opposed to bias). This can be achieved using chemically functionalized
probes,[319] similar to approach used in chemical force microscopy.[320, 321]
The theoretically expected role of screening charges on thermodynamics of domain
formation suggests that tip-induced domain switching will be strongly dependent on
environmental factors, most notably humidity. In recent years, humidity effects on tip-induced
domain switching were explored by several groups.[322, 323] Shur et al.[323] has demonstrated
the significant role of humidity on domain growth kinetics, behavior ascribed to the conductivity
and charge mobility of screening depolarization charges. They have further analyzed the kinetics
of the switching process is controlled by the interplay of the field of surface charges and

81
depolarization and screening field. It should also be noted that similar phenomena were explored
in the context of charge writing and tip induced electrochemistry.
III.7. Interplay between screening and size effects in ferroelectric nanostructures
The interesting application of the screening charge dynamics to explore ferroelectric size effects
was reported by Spanier et al.[324] Using single-crystalline nanowires[325] with controlled size,
they used EFM to create the charged regions. The charge stability is directly related to the
ferroelectric state of material, with charge stable in ferroelectric state when it is coupled to
polarization, and unstable in ferroelectric state. Also note that while charge provides only
indirect evidence on ferroelectric state, the non-contact measurements are much less destructive.
For example, equivalent measurements by PFM would imply application of high electric fields to
the material, and may affect the polarization distributions inside the material (especially close to
TC).
Figure III.19. (a) ferroelectric phase transition temperature, TC, as a function of dNW. The solid
circles are the experimentally determined TC. The magenta solid line is the result of the fit to the
data using the 1/dNW scaling relation. The inset plots TC as a function of dNW and illustrates the
inverse-diameter dependence. (b) (left) A topographic image and (right) a spatial map of the
electric polarization of an 11 nm diameter nanowire. (c) Time series of EFM images for a 25 nm
diameter nanowire following the polarization writing below and above the phase transition
temperature. The writing and reading conditions were identical to those in (b). Reprinted with
permission from [Spanier J E, Kolpak A M, Urban J J, Grinberg I, Ouyang L, Yun W S, Rappe A
M and Park H 2006 Nano Lett. 6 735-9]. Copyright (2006) ACS Publications [324].

82
As presented in figure III.19, the authors also reported that the Curie temperature is
depressed as the nanowire diameter (dNW) decreases, following a 1/dNW scaling.[324] The
diameter at which TC falls below room temperature is determined to be similar to 3 nm, and
extrapolation of the data indicates that nanowires with dNW as small as 0.8 nm can support
ferroelectricity at lower temperatures. The authors further analyzed the possible screening
mechanisms, and provided evidence towards polarization stabilization by adsorption of charge
ions. In particular, they report that ionic screening to be more efficient than that by conductive
electrodes, in agreement with later studies by Gruverman and others.[326]
Groten et al.[227] reported thickness dependent pyroelectric effect and its correlation
with the spontaneous polarization. The authors found that the induced temperature change
decreases with increasing thickness of the substrate in the ferroelectric semicrystalline
copolymer thin films. The authors further observed the temperature-induced change in the
spontaneous polarization through the observation of the screening behavior. Later, Johann and
Soergel[327] explored surface potential response to the induced temperature change in the
lithium niobate. The authors found the linear relationship between the surface potential and the
induced temperature change and, in particular, also directly probed relaxation of the surface
potential under the 1 °C of temperature change.
IV. Tip induced switching
In general, when sufficient voltage is applied through the AFM tip in the ferroelectric materials,
switching events can occur accompanied with charge injection. The injected charge often
induces unexpected back switching due to the electric field induced by the injected charges.
These injected charges are thermodynamically unstable because excess amounts of charges can
be injected onto the ferroelectric surfaces during the bias application. As a corollary, the
collection of the injected charges is feasible when a grounded tip is contacted with the charged
ferroelectric surfaces. However, in fact, the charge injection can occur in not only ferroelectric
and but also non-ferroelectric surfaces. Furthermore, it has been discovered that the switching
events are possible when high pressure is applied to the ferroelectric surfaces, the behavior is
originally attributed to flexoelectricity[328, 329] but that can also be explained by pressure

83
induced charge dynamics. In this section, tip induced switching in both ferroelectric and non-
ferroelectric surfaces are reviewed.
IV.1. Charge injection in non-ferroelectric surfaces
When a voltage is applied to the sample surface through the AFM tip, charge injection can occur
on the sample surface. In such a case, the injected charges can be trapped on the sample surface
or in the bulk of the sample.[330] On the SiO2 surfaces, the injected charges can be further
detrapped by migration along the sample surface by Ohm’s law and/or tunneling toward the
underlying conducting layer if the film thickness is sufficiently thin. If the relaxation time for the
detrapping of the injected charges is not significantly fast compared to the AFM imaging time,
the injected changed can be imaged by either EFM or KPFM.
Figure IV.1. Tip induced surface potential as a function of poling bias on pristine (as-grown)
surface (in TiO2 samples A and B fabricated under the same conditions) and surface uniformly
prepoled by ±8 V. Surface potential for each bias was averaged from each rectangular pattern.
Inset shows the surface potential image of the area scanned with the applied biases from -13 to
+13 V to the conductive probe for pristine TiO2 thin film. The scale bar is 1.5 μm. Reprinted
with permission from [Kim Y, Morozovska A N, Kumar A, Jesse S, Eliseev E A, Alibart F,
Strukov D and Kalinin S V 2012 ACS Nano 6 7026-33]. Copyright (2012) ACS Publications
[331].
Shown in figure IV.1 is the example of the charge injection on the TiO2 surface. The tip-
induced surface potential was measured by KPFM after sequential biasing of the sample surface

84
as a function of bias in a similar manner with figure IV.1.[331] As shown in figure IV.1, the
surface potential is strongly dependent on the sign of the pre-biasing, i.e. pre-poling, and it is
very analogous to the KPFM hysteresis loop. These changes in the surface potential are an
expected effect of bulk charge injection as mentioned above.[332] As applying a negative
(positive) bias to the AFM tip, negative (positive) charges are injected onto the sample surface.
As a result, the negatively (positively) biased regions show relatively high (low) surface potential.
However, since an initial surface potential depends on the pre-biasing state, hysteric behavior
was observed in figure IV.1. The surface potential in the biased non-ferroelectric surfaces is
mostly likely related to the charge injection.
IV.2. Potential evolution during switching
The charge dynamics and its corresponding surface potential under an application of a bias onto
the ferroelectric surfaces can be somewhat different compared to the non-ferroelectric surfaces
due to the existence of spontaneous polarization. Hence, of a paramount interest is the role of
screening charges on the kinetics of tip (or, more generally, top electrode) induced switching
processes. The polarization switching induced by micro-patterned top electrodes is of extreme
interest in the context of electro-optical device fabrication.[333, 334] However, the wave of
interest to these applications emerged in the context of PFM based studies of domain growth, and
especially ferroelectric based data storage.[318, 335, 336] In fact, multiple authors reported
ferroelectric bit sizes as small as several nanometers. J.-M. Triscone group reported that the
minimum bit size can be determined by the radius of the AFM tip using PFM.[337] Using
SNDM, Y. Cho group have demonstrated domains as small as 2.5 nm, corresponding to 10.1
Tb/in2 storage density directly approaching molecular limits.[338] Further, N. Tayebi et al.
demonstrate 2 nm (radius) of bit size, corresponding 40 Tbit/in2, using single-walled carbon
nanotube tip.[339]
The strong interest to tip-induced domain writing has spurred extensive theoretical effort
for describing thermodynamics and kinetics of this process. Thermodynamics of domain
formation in rigid dielectric approximation (i.e. polarization magnitude adopt bulk values within
and outside the domain and changes jumpwise at the boundary) was explored in a series of works
by Molotskii[340] and Morozovska,[341] with several special cases also reported by
Durkan,[342] Emelyanov,[343] and Kalinin.[344] The dynamics in rigid approximation and

85
assuming the viscous damping of domain walls was explored by Molotskii et al.[345] Finally,
Chen et al. and Morozovska et al. explored the thermodynamics of switching assuming
Ginzburg-Landau dynamics for polarization field, predicting intrinsic character of switching
process in agreement with experimental data.[346, 347]
The notable aspect of this theoretical effort was the observation of the non-trivial role of
electrostatic depolarization energy, and hence screening phenomena, on domain energy. In fact,
Morozovska et al., demonstrated that the nucleation bias and activation energy for domain
switching diverge with degree of screening, and, for unscreened surface, tip-induced polarization
switching is thermodynamically impossible.[62] This observation suggests that while screening
cannot be observed directly by PFM-based studies of domain dynamics, the indirect effects of
screening on domain dynamics will be very significant. Furthermore, some information on
charge dynamics during switching can be obtained by combination of PFM and KPFM studies,
as described in the rest of this section.
The screen charge behavior during switching is complicated due to the charge
compensation between polarization, screen, and injected charges.[277, 286, 348] Early work
done by Chen et al. show that the surface charge trap is a dominant effect over the ferroelectric
polarization when an external electric field is applied to the tip on PZT thin films.[286] It has
also been demonstrated that the surface potential depends on pulse voltage and duration applied
to the ferroelectric films.[286, 349] This work shows the injected charges during switching on
the ferroelectric surfaces dominantly contributed to the resulted surface potential measured by
KPFM. Indeed, in most of reports, the sign of the surface charge on the switched region was
observed as an opposite to polarization charge due to the injected charges during switching.[350].
However, the dominant charge source can be dependent on the applying an electric field.[350]
While polarization change can contribute dominantly to the surface potential under a low electric
field, surface charge trap is a dominant effect under applying a high electric field. The
contribution from surface charge trap is dependent on the materials. For instance, the SrBi2Ta2O9
(SBT) thin films can easily trap surface charges than that of the PZT thin films. Also, the
enhanced contribution from polarization charge can be observed after a grounded tip scan.

86
Figure IV.2. (a) KPFM surface potential distribution and (b) PFM phase image of the area
scanned with the applied voltage biases from −8 to 8 V with a 2 V step (from top to bottom) to
the bottom electrode. The black scale bar presents 2 µm. Schematic depiction of (c) polarization
states and charge distributions on different applied biases and (d) the number of charges that
remained after charge compensation. The rectangular and circular charges show, respectively,
polarization and screen charges. The green circular charge represents the screen charge of the as-
deposited state, and the blue charge represents the charge that remained after charge
compensation. Reproduced with permission from [Kim Y, Bae C, Ryu K, Ko H, Kim Y K, Hong
S and Shin H 2009 Appl. Phys. Lett. 94 032907]. Copyright 2009, AIP Publishing LLC [277].
Finally, Kim et al. reported the origin of surface potential during switching by
combination of PFM and KPFM.[277] Since PFM and KPFM allow exploring information on
the polarization and the surface potential, respectively, and an amount of injected charges is
dependent on the magnitude of applying voltage, the field dependent studies by combination of
PFM and KPFM can provide full information on the charge sources during switching on the
ferroelectric surfaces. The resultant absolute value and sign of surface potentials are defined by
the interplay of polarization, injected charges, and as-deposited screening charges (see figure
IV.2.). Notably, injected charges on the film surfaces can be removed by subsequent grounded
tip scans which are correlated to the amounts of residual charges.[277, 351]
After the switching, screening charges are relaxed as a function of elapsed time. If
polarization is stable after switching, the charge dynamics can be governed by Coulombic
repulsion between screen charges.[284, 352, 353] Hence, the surface charges after applying a

87
high electric field show faster relaxation due to larger Coulombic interaction between screen
charges.[284, 353] In particular, in the polycrystalline films, the grain boundaries can affect
relaxation of screen charge migration on the film surfaces since grain boundaries in
polycrystalline films are relatively conductive.[283, 284]
IV.3. Backswitching
An interesting aspect of the screening charge dynamics and its role of polarization switching in
PFM is so called anomalous domain switching and formation of bubble domains.
Experimentally, this phenomenon manifests as back-switching of polarization under the tip, i.e.
formation of a small domain with polarization orientation against the field. This effect was first
reported by Abplanalp et al.[354] and attributed to the higher-order polarization switching driven
by Maxwell stress as shown in figure IV.3.
Figure IV.3. (a) Domain pattern created by a single voltage pulse applied under high mechanical
force, imaged by PFM. The dark center at the place the tip was positioned during poling
corresponds to an antiparallel alignment of the spontaneous polarization and poling field. (b)
Piezoresponse measured along the horizontal line indicated in (a). (c) Local hysteresis loops
measured under an applied force on a 4 μm thick BTO thin film. The solid lines correspond to
increasing voltage and the dashed lines correspond to decreasing voltage. Under high mechanical
load the response switches sign again (arrow), resulting in the domain pattern shown in (a). (d)

88
Dependence of the coercive voltage (points) and the critical voltage UC at which the piezoelectric
response deff is maximal (squares) is shown as a function of the third root of the total force
present during domain formation. These experimental points separate areas of different switching
mechanisms. The applied electric field results in the indicated Maxwell stress. The dashed area
to the left was not accessible during these experiments because the Maxwell stress could not be
complensated. Reprinted with permission from [Abplanalp M, Fousek J and Günter P 2001 Phys.
Rev. Lett. 86 5799]. Copyright (2001) by The American Physical Society [354].
This interpretation was challenged by Bühlmann et al,[355] who attributed this behavior
to the effect of injected charges under the tip. This effect can be illustrated as follows (see figure
IV.4). Application of the positive bias to the tip creates downward electric field in the tip-surface
junction that results in the formation of downward oriented ferroelectric domain. However, this
process is also associated with injection of positive charges on the sample surface that are
generated in the tip surface junction and spread away from it under collective action of diffusion
and migration transport. This charge spot can be easily detected by KPFM, as reported by several
groups,[280, 351, 356] and the fact that measured potential is opposite to polarization charge
suggests that the surface is over-screened.
On turning off the tip bias while still in contact with the surface, the tip is now effectively
grounded. However, the presence of sluggish injected surface charges results in the onset of
electric field oriented opposite to initial field. The action of this field can result in polarization
back-switching, giving rise to the bubble domains.
Figure IV.4. Schematic presentation of the anti-poling effect: (left) poling at high voltage with
charge injection and (right) collection of surface charges by grounded tip and formation of

89
opposite field in the upper part of the film, causing polarization switching. The zigzag shaped
domain wall between the upper and lower film regions helps to reduce electrostatic energy of the
head-to-head configuration. In the center, schematic representation of the contributing potentials,
according the results of the spherical model. Reprinted with permission from [Bühlmann S, Colla
E and Muralt P 2005 Phys. Rev. B 72 214120]. Copyright (2005) by The American Physical
Society [355].
Following this initial observation, the anomalous switching was explored by a number of
authors, including Kim,[351, 356] Brugere,[357] Li,[358] Kan,[359] Kholkin,[280]
Soergel,[360] and others. In particular, the injection charge assisted polarization reversal was
not observed on epitaxial films with a thickness below 130 nm explored by Bühlmann et al.[355]
However, Kim et al show the effect can occur even in 50 nm polycrystalline thin films due to the
higher defect density.[356]
Figure IV.5. PFM images before (a) and after (b) illumination of an inverse domain with a band-
gap UV light. Reprinted with permission from [Kholkin A L, Bdikin I K, Shvartsman V V and
Pertsev N A 2007 Nanotechnology 18 095502]. © IOP Publishing. Reproduced by permission of
IOP Publishing. All right reserved (2007) [280].
Later, Kholkin et al. [280] reported that its origin is indeed correlated with trapped
charges. They performed the similar experiments and observed the anomalous switching as well.
To explore the origin of the observed anomalous switching, the anomalous switched region was
monitored before and after illumination with a band-gap UV light. As presented in figure IV.5,
the anomalous switched region was disappeared after the illumination with UV light. The

90
illumination with a band-gap UV light to the ferroelectric can create free charge carriers of both
signs, which neutralize the trapped charges and, accordingly, eliminate the internal field.
The role of injected charge and screening charge dynamics on polarization switching
suggest that significant care must be made in the interpretation of the PFM switching
experiments, including parameters such as minimal switchable domain size, and kinetics and
thermodynamics of switching process. For example, universally observed nearly-logarithmic
time dependence and almost linear voltage dependence of domain size reported by multiple
groups[337, 361, 362] can be indicative of the kinetics of screening charge spreading, rather than
intrinsic aspect of polarization dynamics and domain wall pinning.
IV.4 Charge collection phenomena
Contacting the surface with a grounded tip just after an application of a voltage bias could induce
anomalous switching as discussed above. In addition, contacting or scanning the surface with a
grounded tip can collect surface charges. In some cases, contacting the surface with a grounded
tip can both induce anomalous switching and collect surface charges.
Kim et al reported the path dependent polarization reversal of the effect.[351] The
grounded tip induces polarization reversal as well as surface charge transfer as shown in figure
IV.6. Lilienblum and Soergel further have explored the effect through controlling of retracting of
the tip, applying biases, and contact force.[360] They reported that this polarization reversal can
happen by charge injection solely, but by ferroelastic switching. Kholkin group shows the
inversed domain after illumination with UV light can be erased, which suggests that the effect is
originated from the trapped charges (see Figure IV.5).[363]

91
Figure IV.6. (a) Schematic of a trace of writing process for dot patterned domain. When the
back-poling process finishes, the grounded tip is moved to the center of the box (gray line). Then,
the grounded tip is moved to the left corner for writing process (violet line). The tip is moved to
follow the trace of the blue line to write dots. Finally, the tip is moved to the center of the box
(violet line). (b) Surface potential image acquired by KPFM. (c) PFM amplitude and phase
images in the position A of (a) after the dot pattern. (d) Schematic of charge injection and
transfer mechanisms: (left-hand) charge injection, (middle) polarization reversal, and (right-
hand) charge transfer. The charge with dotted circle presents the transferred screen charge by a
grounded tip. The bound charge near the ferroelectric surface is drawn as a rectangular shape.
Reprinted with permission from [Kim Y, Kim J, Bühlmann S, Hong S, Kim Y K, Kim S-H and
No K 2008 Phys. Stat. sol. (RRL) 2 74]. Copyright 2008 Publisher Wiley-VCH Verlag GmbH &
Co. KGaA [351]
As shown in figure IV.6(a), the black dots represent the locations of applied voltage for
switching (see figure IV.6(c)) and solid lines represent trajectory of the grounded tip. As shown
in figure IV.6(b), another mechanism produced when a grounded tip is in contact with the
ferroelectric surface was identified and was referred to this phenomenon as screen charge
transfer due to the electrostatic interaction. The dark lines, which are identical with the travel
path of the grounded tip during pattern writing, can clearly be seen. Polarization can be excluded
as a reason for the dark lines in KPFM based on the PFM results of figure IV.6(b). Rather,

92
during pattern writing, the grounded tip removed the screen charges present on the ferroelectric
surface.
Figure IV.7. KPFM surface potential distributions after (a) the poling process and the (b) first
and (c) third grounded tip scans. (d) Surface potential line profiles obtained from (a)-(c).
Reproduced with permission from [Kim Y, Bae C, Ryu K, Ko H, Kim Y K, Hong S and Shin H
2009 Appl. Phys. Lett. 94 032907]. Copyright 2009, AIP Publishing LLC [277].
Further, it was found that the amount of charge collection is dependent on the magnitude
of applied voltage. The surface potential was resulted from competition between different
sources of charges, e.g. polarization and injected charges, on the ferroelectric surfaces. Hence,
the complex surface potential can be observed as shown in figure IV.7. Figures IV.7(a-c) show
the surface potential evolution of the ferroelectric surfaces after contact mode scans by a
grounded tip. When the electrically grounded tip was scanned by contact mode, the mobile
screen charges on the ferroelectric surfaces will be swept to the grounded tip because of the
electrical potential difference between the tip and the surface screen charge. Figure IV.7(a)
shows the surface potential image acquired immediately after the poling process. Figure IV.7(b)
and IV.7(c) present the surface potential images after the first and third scans by a grounded tip.
As shown in figure IV.7, the surface potential contrast of biased regions changed, however the

93
amount of the changed surface potential is strongly dependent on the initial surface potential
value. Even though the surface potential behavior is rather complicated, this clearly presents that
the larger surface potential, i.e. larger amount of surface charges, can be more readily removed
compared to the smaller surface potential. That is, the charge collection is dependent on the
amount of surface charges.
Figure IV.8. Influence of screening charges on the CGM signal patterns. (a−c) Three
consecutive CGM SFMB scan images; (d−f) continuous EFM phase images at (d) 0−2 min, (e)
4−6 min, and (f) 12−14 min after the three CGM scans; (g−i) three consecutive CGM SFMB
scan images after 20 min rescreening. Reprinted with permission from [Tong S, Jung I W, Choi
Y Y, Hong S and Roelofs A 2016 ACS Nano 10 2568-74]. Copyright (2016) ACS Publications
[364].

94
Typically, the charge collection can be monitored by performing KPFM measurements
before and after the grounded tip scans. Hong et al. reported that imaging mechanism of charge
gradient microscopy (CGM) is also related to the charge collection.[364-367] They collected the
current from CGM probe while scanning a ferroelectric surface. It is expected that the measured
current dominantly originate from displacement current and charge flow from removal of screen
charges. As shown in figure IV.8(f), EFM image show relatively small contrast, indicating fully
screen state. In such a situation, the CGM imaging was performed for three times. As shown in
figure IV.8(g-i), the domain signal was predominantly observed in the first CGM image,
however the domain wall signal was observed in the third CGM image. The domain signal
originates from the removal of surface screen charges.
Even though charge collection on the ferroelectric surfaces was discussed above, the
similar charge collection is also possible even in non-ferroelectric surfaces.[368] In the recent
report by Lee et al., charge injection was performed on the poly(methyl methacrylate) (PMMA)
film surface by a similar manner. Even in the non-ferroelectric surfaces, unstable excess charges
on the PMMA surface can readily move to the AFM tip, which is grounded during the scanning
process. Obviously of interest is the interplay between charge collection and ferroelectric
polarization, to be explored in the future.
IV.5. Pressure effects
Figure IV.3 shows that the domain switching can be also related to the mechanically stimulus.
Even though in this case it primarily originated from the internal electric field generating
between a grounded tip and trapped charges, the multiple coupling mechanisms between
polarization and pressure suggest potentially broader class of pressure induced phenomena on
ferroelectric surfaces.

95
Figure IV.9. Mechanically induced reversal of ferroelectric polarization. (A and B) PFM phase
(A) and amplitude (B) images of the bidomain pattern electrically written in the BaTiO3 film. (C)
Single-point PFM hysteresis loops of the BaTiO3 film. (D and E) PFM phase (D) and amplitude
(E) images of the same area after the 1 by 1 μm2 area in the center (denoted by a dashed-line
frame) has been scanned with the tip under an incrementally increasing loading force. The
loading force was increasing in the bottom-up direction [denoted by a black arrow in (E)] from
150 to 1500 nN. (F) PFM amplitude as a function of the loading force obtained by cross-section
analysis along the white vertical line in (E). Reprinted with permission from [Lu H, Bark C W.
Esque de los Ojos D, Alcala J, Eom C B, Catalan G and Gruverman A 2012 Science 336 59-61].
Reprinted with permission from AAAS (2012) [328].
Indeed, Lu et al. demonstrate that the strain gradient induced by the tip can mechanically
switch the ferroelectric polarization.[328, 329] As presented in figure IV.9(d-f), the phase was
reversed in the left half of the image and the amplitude decreases up to around 750 nN then
increases again as increasing the loading force. On the basis of the phase and amplitude images,
it can be concluded that the switching occurs at around 750 nN. The authors argued that this
behavior originated from a flexoelectric effect. Although flexoelectricity is generally very weak
compared to the piezoelectricity, strain gradient grow in inverse proportion to the relation length,
indicating that flexoelectric effect can be very large at the nanoscale. In this context however it

96
should be noted that pressure induced effects on screening charge can give rise to similar
behavior, and the future will undoubtedly see further studies in this direction.
V. Non-local interactions driven by screening charges
As discussed throughout the review, ferroelectric polarization is unstable in the absence of
screening by internal or external charge carriers. Correspondingly, screening processes will be
inherently involved in any polarization reversal processes, and can significantly affect their
thermodynamics and kinetics. While for system with metallic electrodes the screening charges
are readily available and hence only the thermodynamics of the system is affected by the detailed
structure of space charge-layers in ferroelectric, the situation can be completely different if the
screening charges are not available or possess slow dynamics. In general case, the studies of such
cases are hindered by the availability of the hierarchy of screening mechanisms. For example,
rapid changes of polarization due to temperature change or switching in ambient environment
will first cause fast, but energetically inefficient internal screening, and subsequent transition to
slow but energetically favored ionic screening. However, in some cases the role of screening
processes on domain dynamics can be clearly visualized via backswitching phenomena, or via
non trivial non-local effects on domain dynamics stemming from the mass conservation laws of
ionic species
V.1. Spatiotemporal chaos during ferroelectric domain switching
The chaotic phenomena during tip-induced domain switching were recently explored by
Ievlev et al.[63, 369] Here, switching of the ferroelectric LiNbO3 was studies as a function of
temperature and humidity as shown in figure V.1. Application of bias pulse to the tip at
predetermined locations results in polarization switching that was visualized via PFM imaging in
a usual fashion. In the large regions of the V-T-d parameter space (where V is tip bias, T is
temperature, and d is distance between domain centers) the domain size is uniform, indicative of
the high quality of material. However, unusual effects were observed for small domain
separations, including non-equal domain sizes and pronounced transient behavior, intermittency,
and period tripling or the formation of non-periodic or long-periodic structures, as illustrated in
figure V.1. Variation of point spacing and tip bias (for a fixed bias pulse duration of 250 ms)
were used to construct the diagram of switching behaviors as shown in figure V.2(a). Here,

97
application of biases below the nucleation threshold did not lead to domain formation (region I).
Large biases resulted in continuous domain switching and the formation of “stripe” domain
(region II). Application of moderate biases for large separations resulted in well-defined domain
chains formed by uniform domains at each point of bias application, long explored for
information storage.[197, 235, 362, 370] For very large separations, this corresponds to non-
interacting domains in region IIIa. The uniform chains can also form for weak domain
interaction, as shown in region IIIb. The boundary between the two was established based on the
transient behavior in the chain, namely whether the first domain is larger than subsequent ones.
Finally, intermittency and quasiperiodic switching was observed in region IV.
Figure V.1. (a-e) Evolution of the domain patterns for variable-temperature measurements. Note
the rapid change of the switching character between 100 and 150 °C from period doubling (a) to
uniform chains (b-d) to long-range periodicity (e). (f-l) Evolution of switching pattern as a
function of relative humidity at room temperature. The data illustrates a transition between
uniform switching (f), period doubling (g,h), long range periodicity (i) and back to uniform
switching (j,k,l). Switching disappears at high humidity (decrease of domain sizes from j to l; no
domain formed at 90%). Measurements in a-e are taken at 40% humidity (defined at RT),

98
measurements (f-l) are taken at RT. Reprinted with permission from [Ievlev A V, Jesse S,
Morozovska A N, Strelcov E, Eliseev E A, Pershin Y V, Kumar A, Shur V Ya and Kalinin S V
2014 Nat. Phys. 10 59] [63].
Figure V.2. (a) Phase diagram of switching behavior as a function of bias and domain spacing.
Shown are regions of (I) no switching, (II) continuous switching, (IIIa) isolated uniform
domains, (IIIb) uniform domains with transients, and (IV) quasiperiodic and chaotic behavior.
(b) Phase diagram of switching behavior as a function of dimensionless voltage
( )
0
REUu cr
=
and domain spacing
0
Rd=d
calculated for h=1. Note the qualitative similarity between
experimental diagram in figure (a) and theoretical figure (b). Also shown are axes obtained by
rescaling to experimentally measurable parameters for
30
0
=R
nm,
20
0=REcr
V. Using bulk

99
materials parameters PS=0.75 C/m2, e33=30, e11 = 84, the critical field for wall motion can be
estimated as
=
cr
E
0.67 V/nm and effective width of water layer is H=60 nm. Note the
agreement between calculated patterns and those observed experimentally in Figure (a). (c)
Dependence
( )
nn x
x1+
calculated from equation (V.4) using equation (V.2) for and parameters
4=d
,
10=
l
, and humidity
10
=
h
and different values of
32
u=m
indicated near the curves. (d)
Experimentally determined recursive relation
( )
nn
xqx =
+1
and smoothened dependence.
Suppression of domain nucleation in the proximity of pre-existent domains is clearly seen (xn >
0.5), as well as a small upward trend for xn < 0.4. Reprinted with permission from [Ievlev A V,
Jesse S, Morozovska A N, Strelcov E, Eliseev E A, Pershin Y V, Kumar A, Shur V Ya and
Kalinin S V 2014 Nat. Phys. 10 59] [63].
To explain the observed behaviors, the authors analyzed the ionic screening model
assuming the conservation of the surface ionic species.[53, 60] In this case, polarization
switching in the selected region results in the formation of the excess uncompensated charge, and
can be represented as:
([+P] – OH-) + H2O + 2e- = ([-P] – H+) + 2OH- (V.1)
Here, ([+P] – OH-) is the positive polarization charge bound with the screening hydroxyl group
(see Refs.[244, 306] for discussion of equilibrium degree of screening), and ([-P] – H+) is the
negative polarization charge bound with a screening proton. The formed hydroxyls diffuse
laterally across the sample surface, creating a “halo” surrounding the switched area. This “halo”
will in turn suppress the polarization switching in the adjacent regions via the classical Le
Chatelier effect.
This analysis was confirmed by subsequent studies of switching behaviors as a function
of temperature and humidity. On increasing the temperature for parameters corresponding to
intermittent switching at room temperature, transition to the regular writing occurred. A series of
remarkable transitions was observed on increasing the humidity. For (nominally) zero humidity,
the writing yields a series of the equal sized domains. For higher humidity, the onset of period
doubling is observed at 25% RH. At higher humidity, quasiperiodic structures are observed for
35% and long-range periodicity (up to 8-10 domains) is observed for 50 % RH. For 60%
humidity the domains again become uniform. For further increase of humidity, the domain size

100
starts to reduce, transient behavior becomes less pronounced, and for (nominally) 90% humidity
and above domains no longer form at this bias. This behavior is also confirmed by locally
measured hysteresis loops. These observations can be immediately rationalized by considering
the T and humidity effect on the mobility of screening charges. Namely, for low humidity the
thickness of the adsorbed water layer on the surface is minimal, and hence the charge halo
around the domain is highly localized and does not affect the formation of adjacent domains. For
intermediate humidity values, the charges are localized in the vicinity of the formed domain, and
intermittent dynamics are observed. Finally, for high humidity the charge spreads rapidly across
the sample surface, obviating this effect. At the same time, the water layer effectively screens the
surface, precluding polarization switching.
The authors have further proceed to quantify this behavior by treating the domain
sequences as a classical time series for which transition to complex dynamic behaviors occur are
well-known.[371, 372] By analyzing the structure of the depolarizing field created by the charge
halo under some non-restrictive assumptions on its density profile, the authors have
demonstrated the non-monotonic effect of existing domain on the next domain in the chain, and
were further able to reduce this dynamics to a classical logistic map ,
( )
nnn zzz -a=
+1
1
,
properties of which as a function of a master parameter a are extremely well studied.[372] This
analysis allowed to reconstruct the theoretical domain behavior diagram in dimensionless units
which was very similar to experimental observations. Furthermore, matching the corresponding
material-related parameters yielded the realistic estimates for the Peierls field and domain wall
energies. Finally, the direct observation of recursive relationship was demonstrated.

101
Figure V.3. (a)-(d) Shapes of the domains explored after switching by sequence of bipolar
triangular pulses at 75% of relative humidity. Usw = 80 V; (a) 1, (b) 2, (c) 3 and (d) 5 cycles. (e)
Evolution of the domain structure as a result of tip-induced switching by sequence of two bipolar
triangular pulses with amplitude Usw = 80 V. Shown in the signals of amplitude (a)-(d) and
combined piezoresponse signal (e). Reprinted with permission from [Ievlev A V, Morozovska A
N, Eliseev E A, Shur V Ya and Kalinin S V 2014 Nat. Commun. 5 4545] [369].
The spectroscopic measurements demonstrated that normal polarization reversal was
observed after application of the electric field with direction opposite to Ps immediately after
application of bias. Application of the nominally non-switching rectangular pulses also leads to
polarization reversal, i.e. switching against applied electric field. However, in this case switching
is observed at the falling edge of the switching pulse, rather than during application of the field

102
as is the case for normal switching. PFM visualization after abnormal switching showed
formation of the ring-shaped domains with radius above 120 nm, significantly exceeding the size
of the domains formed as a result of normal switching. The authors postulated that the abnormal
switching is induced by the redistribution of the screening charges on the surface and top layer of
ferroelectric. On application of tip bias, applied electric filed along Ps is screened by surface and
bulk charges of the opposite sign. After switching the bias off, these accumulated sluggish
charges produce electric field that induces polarization reversal nominally against the field. The
ring shape of the domain can then be explained by back switching of polarization, as is directly
confirmed by observed time dependencies of the piezoresponse signals.
V.2. Domain shape instabilities during switching
Similar effects were observed in the shape of individual domains formed during bipolar
switching, as shown in figure V.3. Here, switching process gave rise to highly non-trivial domain
shapes at the different stages of triangular wave. Interestingly, application of the electric field
with the same direction as spontaneous polarization (Usw > 0 V) led to formation of the ring-
shaped domains. Application of the electric field with opposite direction led to normal switching
with formation of circular domains. The subsequent cycling leads to the loss of stability of the
ring domains, with the formation of well-defined satellites

103
Figure V.4. Switching by sequences of positive and negative triangular pulses. (a) Scheme of
switching procedure. (b) Used switching combinations and corresponding codes. (c) Examples of
the domains formed after switching by different sequences (3 switching cycles). (d) Domain
shapes as a function of sequence and number of switching cycles. Scale bar is 500 nm. Reprinted
with permission from [Ievlev A V, Morozovska A N, Eliseev E A, Shur V Ya and Kalinin S V
2014 Nat. Commun. 5 4545] [369].
These observations further enabled information coding in a single domain shape, as
shown in figure V.4. Each negative pulse is labeled as “0”, positive as “1”. Combinations of 3
consequent pulses have been used. This enables 8 possible unique switching combinations from
000 to 111 (figure V.4(b)). Complex investigation of the tip-induced switching as a function of
used sequence and number of switching cycles demonstrated unexpectedly wide variety of the
domain morphologies (figure V.4(d)). In particular, one switching cycle showed formation of the

104
domains of two types: circular and ring-like. Domain shape was defined by last pulse in the
sequence only. Sequences ending by negative pulse (000, 010, 100, 110) led to formation small
circular domains while sequences ending by positive pulse (001, 011, 101, 111) – ring-like
domains. However, increasing of the number of switching cycles revealed dependence of the
resulted domain shape on the structure of whole sequence. Switching by 3-5 cycles allowed
differentiation of 6-7 different domain morphologies. Subsequently, the specially trained neural
network was able to recognize the 3 element sequence with 80% probability, illustrating the
information encoding in the domain shape.
V.3. Modeling of screening effects impact on the tip-induced polarization switching
For most SPM experiments in humid atmosphere surface screening charges dynamics appear
under the tip-induced polarization reversal can be attributed to existing of the water meniscus
between tip and the sample surface. Results of the numerical modeling of spatial distributions of
the tip-induced electric potential and field Etip under the sample surface are shown in figure V.5
for different height hm of the meniscus.

105
Figure V.5 Results of numerical simulations of z-component the electric field produced by
biased SPM tip in presence of water meniscus. (a) Scheme of the model; (b), (c) 2D maps of the
spatial distribution of EtipZ for (b) hm = 25 nm and (c) hm = 100 nm; (d), (e) distributions of EtipZ
and potential along polar direction at 10 nm under the surface for different hm. Reprinted with
permission from [Ievlev A V, Morozovska A N, Shur V Ya and Kalinin S V 2014 Appl. Phys.
Lett. 104 092908]. Copyright 2014, AIP Publishing LLC [64].
As can be seen from figures V.5(d-e), the appearance of the meniscus leads to decreasing
of the electric field in the area under the tip and it spatial delocalization. Using an effective point
charge model[362, 373-375] for the description of SPM tip field, simulated distributions of
)(rEz
tip
can be fitted by the following expression:
( )
()
2
3
2
2*
*
0
*
)
(
12
)( rd
dQ
rE
c
a
a
c
z
tip
+
e
e
+
e
epe
=
(V.2)
At that the values of the effective tip radius d* and effective charge Q* are fitting parameters
extracted from either experiment or numerical modeling. Typical dependences of the Q* and d*
on the meniscus height hm are shown in figure V.6. Particularly, effective tip radius d* changes
for the small heights hm£ 100 nm and saturates for larger heights. Q* has short part of the growth
at hm £ 100 nm and then asymptotically decreases with hm increase. Extracted values of the
effective tip radius and charge allows to describe the growth of the isolated domain with
presence of the water meniscus using the model[62].
Figure V.6. Values of the effective charge Q* and effective tip radius d* vs. meniscus height in
the point charge model. Reprinted with permission from [Ievlev A V, Morozovska A N, Shur V

106
Ya and Kalinin S V 2014 Appl. Phys. Lett. 104 092908]. Copyright 2014, AIP Publishing LLC
[64].
Namely, to fit the point data shown in figure V.6 the following trial functions have been
selected:
÷
÷
ø
ö
ç
ç
è
æ-
-= -
¥
¥
¥
q
h
h
e
ddd
dhd
m
0
m
*1)(
,
÷
÷
ø
ö
ç
ç
è
æ÷
÷
ø
ö
ç
ç
è
æc-
+
c= -
¥¥ q
h
h
e
d
d
d
d
Qh
Qm
0
0
0
m
*1
)
(
, (V.3)
Here d*(0) = d0 and d*(∞) = d
∞
, at that hm = 0 corresponds to the dry atmosphere and in this
case tip has effective radius d0. Effective tip charge
00 dUQ sw
@
is proportional to switching bias
Usw applied to the tip and its radius. The limit hm =
∞
corresponds to the tip under the water
(liquid PFM). In this case the tip has effective radius d
∞
which is higher than the tip radius in dry
air due to delocalization of the field. Effective charge is still proportional to the tip radius d
∞
and
bias Usw, but reduced by factor due to higher permittivity of the water εw = 80. hq and hd are
characteristic distances for the effective tip radius and effective charge vs. meniscus size
dependences correspondingly.
To relate the meniscus height hm with the value of relative humidity H as
÷
ø
ö
ç
è
æ-= H
H
hHh
cr
m
exp)(
0
, where the parameters h0 and Hcr are defined from the fitting of
experimental data. Finally, domain radius as a function of relative humidity H and bias Usw was
be obtained:[64]
1
)( ),(
)
(),
(
*
*
*
0
-
÷
÷
ø
ö
ç
ç
è
æ
b+@
a
GH
dU HUQ
H
drH
Ur
cr
sw
sw
, (V.4)
where is a power factor 2/3 < α < 2, dimensionless parameter β reflects the tip form-factor and
has the order of unity, Ucr has sense of a critical voltage for a "dry" ferroelectric surface without
water meniscus (H = 0). The effective tip parameters d* and Q* are given by equation (V.3), G~1
is a factor reflecting the tip geometry.
Note that equations (V.2-4) consider the influence of the top water layer on polarization
reversal due to redistribution of the electric field produced by the tip only. But the influence

107
doesn’t limited by this phenomenon. In addition, the presence of the top water layer changes all
screening conditions by redistribution of the charge carriers across the layer. This phenomenon
can be used for explanation of the inconsistence between these results and those in refs.[322,
323] In the both papers growth of the micron-sized domains was observed. At such distances
redistribution of the electric filed caused by water meniscus is not so pronounced and cannot lead
to essential change of the domain kinetics. However external screening caused by charge carriers
in the adsorbed surface layer supports the switching far from the tip and leads to formation of the
large domains.
Finally, a self-consistent theoretical approach capable to describe the features of the
anisotropic nanodomain formation induced by a strongly inhomogeneous electric field of
charged SPM tip on non-polar cuts of ferroelectrics has been proposed.[376] Here it was shown
that a threshold field is an anisotropic function that is specified from the polar properties and
lattice pinning anisotropy of a given ferroelectric in a self-consistent way. The proposed method
for the calculation of the anisotropic threshold field is not material-specific, thus the field should
be anisotropic in all ferroelectrics with the spontaneous polarization anisotropy along the main
crystallographic directions. The most evident examples are uniaxial ferroelectrics, layered
ferroelectric perovskites and low symmetry incommensurate ferroelectrics. Obtained results
quantitatively describe difference in several times in nanodomain length experimentally observed
on X- and Y-cuts of LiNbO3[377] and can give insight into the anisotropic dynamics of nanoscale
polarization reversal in strongly inhomogeneous electric fields. The threshold fields along
different crystallographic directions can be significantly different due to crystal anisotropy of the
inter-atomic relief and energy barriers. 3D-atomic structure of LiNbO3 crystallographic cuts are
shown in figures V.7(a-c) using the coordinates from Boysen and Altorfer.[378]

108
+X
p
Y
=2.575 Å
+Z
Nb
Li
O
+Y
+Y
(b) LN-structure at X-cut
(a) LN-structure at Z-cut
(c) LN-structure at Y-cut
+Y
+X
p
z
=2.310 Å
+Z
+Y
p
x
=4.460 Å
step
Figure V.7. Atomic structure of LiNbO3 Z-cut (XY plane) (a) X-cut (ZY plane) (b) and Y-cut
(ZX-plane) (c). Big blue balls are Nb atoms, smaller green balls are Li atoms and the smallest
red balls are O atoms. Reprinted with permission from [Morozovska A N, Ievlev A V,
Obukhovskii V V, Fomichov Y, Varenyk O V, Shur V Ya, Kalinin S V and Eliseev E A 2016
Phys. Rev. B 93 165439]. Copyright (2016) by The American Physical Society [376].
A suggested step-like path of the domain wall motion in the polar direction Z on the non-
polar X- and Y-cuts is shown by an elementary step in figures. V.7(b-c). The step-like path is
defined as the elementary Li-Li distances in the directions perpendicular to the polar Z-axes.
Then using rhombohedral lattice parameters of LiNbO3, а = 5.15 Å, c = 13.86 Å, angle
ac = 55o53'[379, 380], the minimal distances
][abc
p
between the equilibrium positions of
uncharged domain wall planes at different crystallographic cuts, are
460.423
]100
[
»= ap
Å for
X-cut,
575.
22
]010[
»= ap
Å for Y-cut and
310.26
]001[
»= c
p
Å for Z-cut.
The anisotropic threshold field accounts for the anisotropy of the minimal distance
between the equilibrium positions of the uncharged domain wall in the Ishibashi formula[381] in
the following way:
( )
( )
( )
][
2
3
][
2/7
4][
exp2
abcabcS
abc
th
pw
pwPeE p-a
p-=
, (V.5)
Here the half-width of the domain wall
w
is normalized on the minimal distance
][abc
p
between
the equilibrium positions of the uncharged domain wall plane propagating in the crystallographic

109
direction [abc]. Hereinafter we associate [001] with a Z-cut, [010] with Y-cut and [001] with X-
cut.
Figure V.8 illustrates the anisotropic threshold field calculated using equation (V.4) for
LiNbO3 parameters a, PS and different domain wall half-width w, since the latter can be strongly
affected by depolarization field and depends on the wall bound charge (e.g. incline angle with
respect to the polar direction). As one can see, the value of Eth differs on the one or even several
orders of magnitude for different direction of the domain wall motion. In addition it strongly
decreases with
][abc
p
increase and vary in the range (10-3 – 10+2) kV/mm. Eth monotonically and
rapidly decreases with w increase more than 1 Å for any period
][abc
p
. Note, that smaller w
values are unlikely physical. At fixed w> 1 Å the highest fields correspond to the smallest period
]
[abc
p
, i.e.
X
th
Y
th
Z
th
EEE <<
since
XYZ ppp <<
. This exactly means that the threshold field is the
smallest for Z-cut, intermediate for Y-cut and the highest for the X-cut of the crystal.
Equation (V.5) allows us to estimate the ratio of the threshold and activation fields in
different directions for known distances
][abc
p
[figure V.8(b)]. The ratio of the threshold fields
Y
th
X
th EE
changes from 1.5 to 5, the ratios
Z
th
Y
th
EE
and
Z
th
X
th
EE
are within the range of 1.5 to
100 for realistic values of the domain wall width. The ratio of the activation fields
Y
a
X
a
EE
changes from 1.5 to 10, the ratios
Z
a
Y
a
E
E
and
Z
a
X
a
EE
are within the range of 0 to 3.
1
2
3
4
5
0.01
0.1
1
10
100
X-cut
Y-cut
Z-cut
DW half-width w (Å)
E
th
(kV/mm)
(a)
1
2
3
4
5
2
5
10
20
50
100
Ratio of E
th
(b)
E
thX
/E
thZ
DW half-width w (Å)
E
thX
/E
thY
E
thY
/E
thZ
Figure V.8. (a) Threshold fields dependence on the domain wall half-width w calculated within
Suzuki-Ishibashi model for LiNbO3 parameters a = -1.95´109 m/F, PS = 0.735 C/m2. (b) The

110
threshold fields' ratio vs. the domain wall half-width w. Reprinted with permission from
[Morozovska A N, Ievlev A V, Obukhovskii V V, Fomichov Y, Varenyk O V, Shur V Ya,
Kalinin S V and Eliseev E A 2016 Phys. Rev. B 93 165439]. Copyright (2016) by The American
Physical Society [376].
Note, that DFT calculations of the Eth values for three crystallographic directions can be
very helpful for both the anisotropic model verification and comparison with experiment.[58]
V.4. Screening effects during in-plane domain switching
Application of the bis to the tip in contact with the surface with p[urely in-plane
polarization can give rise to in-plane domain switching due to radial part of tip field. Such
switching necessarily leads to formation of charged domain walls, screened by bulk and surface
charges. Hence, observations of such domains and their time evolution provides insight into
screening mehcnisms. Alikin et al[377] measured experimentally the domain shape and sizes on
the non-polar X- and Y-cuts of CLN. Corresponding domain length and width at the non-polar
surfaces of the CLN are shown by symbols with error bars in figures. V.9(a) and (b). Solid
curves are the best fitting within the afore-presented anisotropic model[376]. Using the fitting
parameters we calculated that the threshold field
th
E
for X-cut is about 420 kV/mm and about
250 kV/mm for Y-cut. Note that due to the scattering of the width for X- and especially Y-cut
the experimental data for domain width is much less reliable than the data for length. So that it
has a little sense to conclude about the method validity from the domain width fitting. However a
reasonable agreement with experiment by changing the fitting parameters for both domain length
and width is present.

111
(a)
Z
X
r
l
Y=0
q
(b)
(c)
500 nm
Y-cut
X-cut
10
100
1000
10000
500
1000
1500
2000
2500
Length l (nm)
Pulse duration t (ms)
(d)
X-cut
Y-cut
10
100
1000
10000
100
200
300
400
X-cut
Y-cut
Width r (nm)
Pulse duration t (ms)
(e)
Figure V.9. (a) Sketch of the domain shape in the XZ plane (Y=0) induced by AFM probe on
congruent LiNbO3 (CLN) non-polar cuts. (b-c) Experimentally observed domain shape from
Alikin et al on (b) Y- and (c) X-cuts of 20-mm-thick CLN. Dependencies of (a) domain length
and (b) domain width vs. the switching pulse duration on X- and Y-cuts of LiNbO3. Symbols
with error bars are experimental data from Alikin et al for X- and Y- cuts of 20-mm-thick CLN
placed in dry nitrogen, solid curves are the fitting. Reprinted with permission from [Morozovska
A N, Ievlev A V, Obukhovskii V V, Fomichov Y, Varenyk O V, Shur V Ya, Kalinin S V and
Eliseev E A 2016 Phys. Rev. B 93 165439]. Copyright (2016) by The American Physical Society
[376].
Finally, we discuss the question about the difference in domain depth for the cases when
the writing electric field acts on different LiNbO3 cuts. As it was reported earlier by Molotskii et
al[382] for the case of nanodomain formation on polar Z-cut their depth (called length because of
the radial symmetry of domain cross-section) can reach micron distances due to the breakdown
effect. Alikin et al concluded from a selective etching that the domain depth on the Y-cut is
rather small in comparison with the one on the X-cut. Moreover, "Y-cut domains" most likely

112
remained nanosized in Y-direction, while "X-cut domain" can be much deeper, but not needle-
like as "Z-cut domains". Anisotropic approach can explain these facts, because it account for the
anisotropy of lattice barriers and depolarization effects at the charged domain walls. In
particular, the longest needle-like shape of Z-cut domains is conditioned by the smallest
threshold field
( )
Zth pE
and domain breakdown in Z-direction takes place. The smallest depth Y-
cut domain in X-direction originated from inequality
( ) ( ) ( )
ZthYthXth
p
EpEpE >>>
, since the
smaller is the threshold field the bigger is the domain size. These speculations can be quantified
using the
()
][abcth pE
ratios for different crystallographic cuts presented in figure V.9 and
corresponding minimal distance
»
Z
p
2.310 Å,
»
Y
p
2.575 Å,
»
X
p
4.469 Å.
The authors have argued that observed domain-domain interactions and complex shapes
may potentially be used for a variety of information technology applications. However, the most
significant aspect of these observations is the direct visualization of the role that slow screening
charge dynamics can play in ferroelectric switching. While classical observations of domain wall
motion cannot distinguish the possible mechanisms – e.g. pinning in the material vs. screening
charge dynamics, the observation of long range interactions, back switching, and complex
domain shapes clearly illustrate the role of screening processes in domain switching.
VI. Chemical and light effects
As discussed above, the surface state of ferroelectrics is very sensitive to external environment.
Correspondingly, it can be affected by chemistry and instant photo carriers via photovoltaic
effects. Here, we discuss the chemical switching, light effects, and internal screening
photodeposition on the ferroelectric surfaces.
VI.1. Chemical switching
Reversible chemical switching was demonstrated in PbTiO3 thin film by using x-ray scattering
measurements.[24] In their work, they reported reversal chemical switching assisted by amount
of oxygen content. This chemically induced changes in ferroelectric domain structures was
observed using SPM.

113
Figure VI.1. PFM phase images of different states: [(a) and (d)] as-deposited, [(b) and (e)
oxygen plasma treated for 10 min, and (c) and (f) after subsequent annealing at 350 °C for 30
min, for BFO thin films of [(a)-(c)] 30 nm and [(d)-(f)] 60 nm thickness. The scale bars of (c)
and (f) correspond to 400 nm and 300 nm, respectively. Reprinted with permission from [Kim Y,
Vrejoiu I, Hesse D and Alexe M 2010 Appl. Phys. Lett. 96 202902]. Copyright 2010, AIP
Publishing LLC [383].
Reversible chemical switching was also reported in BiFeO3 (BFO) thin films using PFM.[383]
The oxygen plasma and vacuum thermal annealing were used to chemically control the
polarization switching as similar to atmospheres with high and low oxygen partial pressures,
respectively, in ref. [134] As shown in figure VI.1, reversible chemical switching was well
achieved by surface chemical reactions by the oxygen plasma and subsequent vacuum thermal
annealing. They further directly observed that the later domain growth during the plasma
treatment depends on the duration of the treatment.
The chemical switching of the ferroelectric materials is strongly correlated with
chemistry of ferroelectric surfaces.[384] Bonnell group reported that chemical species such as
methanol and CO2 adsorb on BaTiO3 (BTO) surfaces at specific defect sties, here oxygen
vacancy sites.[157] Further, they presented adsorption and reaction of chemical species on the
ferroelectric surfaces are strongly dependent on the ferroelectric polarization.[156, 385, 386]

114
Figure VI.2. Influence of CO2 adsorption on the surface potential of BTO (001) and a PZT thin
film. (a) Surface potential maps of c+ and c− domains on BTO (001), which were poled by an
external electrical field applied by a conductive AFM tip scanning over the surface. Ferroelectric
domains were poled positively (negatively) in the dark (bright) area and showed negative
(positive) surface potential due to the opposite compensating charges. The areas surrounding c+
and c− domains represent in-plane a domains. (b) Surface potential maps after exposure to 30 L
of CO2. (c) Average surface potential versus CO2 dose on BTO (001). (d) θ(L)/θmax versus CO2
dose. S is proportional to the slope of the line. (e–h) Corresponding data for CO2 adsorption on
the polycrystalline PZT thin film. Reprinted with permission from [Li D, Zhao M H, Garra J,
Kolpak A M, Rappe A M, Bonnell D A and Vohs J M 2008 Nat. Mater. 7 473-7] [186].

115
Later, Bonnell group directly observed effect of ferroelectric polarization on the
physisorption energies as shown in figure VI.2. Figure VI.2 shows surface potential as a function
of CO2 dose and corresponding coverage of CO2 on BTO and PZT thin films. They found that
the gas-phase molecule such CO2 first physisorbs to the ferroelectric oxide surfaces then it
diffuse until it chemisorbes at the oxygen vacancy sites. The amount of chemisorption is
dependent on the polarization direction, CO2 dose, as well as materials.
VI.2. Light effects
A wealth of information on the role of screening phenomena on domain dynamics can be
obtained using studies under strong light irradiation. The original exploration of this area dates
back to 60ies to early work of Fridkin, Sturman and others, area generally referred as
photoferroelectrics or ferroelectric semiconductors.[311, 387-390] The central premise of these
studies was that irradiation of ferroelectric surface with sub-band gap light results in generation
of carriers in the near-surface layers. The interaction of photogenerated carriers and polarization
fields give rise to a broad set of phenomena ranging from formation of unusual domain
structures,[388] conductivity,[389] significant changes in coercive fields for polarization
switching,[391] and elastic[390] and electromechanical responses.[392] The theory of these
phenomena including formation of spatially-inhomogeneous polarization patterns and dynamic
solutions (moving solitons, etc) was considered by a number of authors.[393]
However, these early observations were significantly hindered by the lack of high-
resolution probes of domain structures. More recently, Gruverman et al.[394] reported that UV
illumination of the ferroelectric thin films results in a voltage shift by an interaction between
photo induced and polarization charges. In particular, they directly observed domain pinning
induced by the illumination at the nanoscale using PFM.[394] In a similar study, Shao et al. has
explored to directly observe photo induced carrier screening on the ferroelectric surfaces.[395]
As presented in figure VI.3, surface potential is rapidly reduced by UV illumination and recovers
after the switch-off of the UV illumination. These relaxation phenomena are due to the charging
and discharging of photo induced carriers.

116
Figure VI.3. (a) AFM topography of a BTO (001) surface that shows three corrugations due to
a-c domain walls. z scale is 200 nm. (b) The surface potential image of the same area as shown
in (a) shows (c) domains with curved domain walls. z scale is 0.25 V (c) The surface potential
image when the UV light is on. z scale is 0.25 V. Time dependence of surface potential contrast
between c+ and c− domains immediately after UV light is switched (a) on and (b) off. Reprinted
with permission from [Shao R, Nikiforov M P and Bonnell D A 2006 Appl. Phys. Lett. 89
112904]. Copyright 2006, AIP Publishing LLC [395]._
Finally, recent years have seen a dramatic growth of interest to photovoltaic phenomena
in ferroelectric materials following the works by Yang,[396] Choi,[397] and Alexe.[398, 399]
Alexe et al.[398] reported unusual photovoltaic effect in BiFeO3 single crystals. As shown in
Figure VI.4, the local and macroscopic open circuit voltages from the photocurrent-voltage
characteristics are well matched to each other. However, if one can calculate the current density,
one can find that the current density of the local measurements is significantly higher than that of
the macroscopic measurements. This is due to the fact that the AFM tip is very efficient in
collecting the photo induced carriers.

117
Figure VI.4. Photocurrent–voltage characteristics. (a) Measured macroscopically across the
entire crystal by illumination in the region marked in the lower right inset, and (b) measured by
probing with the AFM tip in the middle of the illuminated area. The upper left inset in (a)
schematically shows the measurement setup and the connection to the macroscopic
measurement; the lower right inset in (a) shows the crystal with silver electrodes marked by
arrows and approximately the illuminated area by the laser spot. The inset in (b) shows the
measurement setup for the local measurements by the AFM tip. Reprinted with permission from
[Alexe M and Hesse D 2011 Nat. Commun. 2 256] [398].
Yang et al.[396] studied ferroelectric domain wall dependent photovoltaic effects in the
BFO thin films. They found that current-voltage characteristics in two different electrode
geometries (one is perpendicular and the other is parallel to the domain walls) reveal
significantly different photovoltaic behaviors (see figure VI.5). Because the illuminated domain
wall enables a more efficient separation of the excitons, a net voltage observed across the sample
is resulted from the combination of the domain walls and excess charge carries by illumination.

118
Figure VI.5. Light and dark I–V measurements. (a,b) Schematics of the perpendicular (DW⊥)
(a) and the parallel (DW∥) (b) device geometries. (c,d) Corresponding I–V measurements of the
DW⊥ (c) and DW∥ (d) devices, respectively. Reprinted with permission from [Yang S Y,
Seidel J, Byrnes S J, Shafer P, Yang C-H, Rossell M D, Yu P, Chu Y-H, Scott J F, Ager J W,
Martin L W and Ramesh R 2010 Nat. Nanotech. 5 143-7] [396].
VI.3. Domain dependent photodeposition
The SPM studies of the ferroelectric surfaces are directly sensitive to the electrostatic fields
above the surface (KPFM and EFM) or bulk material properties in the near-surface layer
(KPFM). As such, they are more sensitive to the external screening behavior, whereas the
structure of potential and field distributions inside material is generally inaccessible by SPM and
can be explored by depth-resolved electron microscopy and scattering methods. However,
certain insight into the internal screening on ferroelectric surfaces can be obtained from the
observation of photovoltaic, chemical and photoelectrochemical properties of ferroelectrics. In
these cases, light irradiation of ferroelectric results in generation of electron-hole pairs, and their
subsequent dynamics can be explored though SPM of photochemical reactivity.
A notable example of such behavior is domain specific photochemical deposition of
metal and oxides on ferroelectrics as originally reported and extensively explored by Rohrer and
Giocondi.[400, 401] As presented in figure VI.8, the authors reported that illumination of BTO
surface in AgNO3 solutions lead to preferential silver deposition on positive domains, whereas

119
negative domains remained uncovered by silver. This behavior can be readily explained
assuming that the positive polarization charge results in downward band bending. Under
illumination, the photogenerated electron-hole pair is split by the near-surface field, and electron
moves towards surface. In the absence of electrochemically active species, electron accumulation
will result in the onset of flat band conditions, similarly to open solar cell. However, in the
presence of reducible cations chemical reaction and deposition of metallic silver is possible.
Notably, positive carriers are extracted in negative domains and likely result in water oxidation,
completing mass and electron balance.
Figure VI.6. Topographic AFM images of the surface of a BaTiO3 single crystal. (a) Before the
reactions. (b) The same area of the surface after illumination in an aqueous AgNO3 solution. The
white contrast corresponds to silver. (c) The same area of the surface after it was cleaned and
illuminated in an aqueous lead acetate solution. The white contrast corresponds to lead
containing deposits. The ranges of the vertical black-to-white contrast in topography (a−c) are 80,
100, and 110 nm, respectively. Reprinted with permission from [Giocondi J L and Rohrer G S
2001 J. Phys. Chem. B 105 8275-7]. Copyright (2001) ACS Publications [400].

120
The evidence for this mechanism can be obtained from observations of photochemical
reactivity in poorly conducting materials such as LiNbO3 (LNO). As shown in Figure VI.7, metal
deposition on the periodically poled lithium niobate (PPLN) surface was observed only in the
vicinity of ferroelectric domain walls, i.e. compensation is possible only on the length scale of
carrier diffusion in the materials.[402]
Figure VI.7. (a) Topographic image of the PPLN sample before deposition; ((b), (c))
corresponding PFM phase and amplitude images, respectively, of the PPLN sample; (d)
topographic image of the same sample after deposition (inset: higher-resolution topographic
image illustrating the structure of the lines as formed by silver particles). Reprinted with
permission from [Hanson J N, Rodriguez B J, Nemanich R J and Gruverman A 2006
Nanotechnology 17 4946-9]. © IOP Publishing. Reproduced by permission of IOP Publishing.
All right reserved (2006) [402].

121
In addition to the photochemical deposition, there are many tries for understanding and
probing of screening phenomena on ferroelectric surface as discussed in Section VI.2.[280, 281,
395, 403, 404] Beyond fundamental probing of screening phenomena on the ferroelectric
surfaces, these studies open the pathway for polarization controlled creation of metallic and other
nanostructures. In this ferroelectric lithography as pioneered by Bonnell group,[405] the charged
SPM tip is used to create domain pattern and subsequent metal photodeposition creates
nanostructure. The latter can then be studied by e.g. contacting to outside worlds, chemical
modification, etc., or potentially transferred to different substrate. Notably, the metal
photodeposition is not associated with significant changes in ferroelectric surfaces and hence can
be performed multiple times, effectively providing a master. This approach was recently
extensively explored by Gruverman,[394] Eng.[406] Kitamura[407] and others.[405] As an
example, Eng et al., prepared Pt nanostructures through creation of domain patterns and
subsequent photochemical reduction.[406] Further, as shown in figure VI.8, Kalinin et al.,
controlled local ferroelectric domains by application of an electric field through the conductive
AFM tip then the subsequent photo reduction enabled deposition of Ag nanoparticles only on c+
domains in the pattern.
Figure VI.8. Surface topography (a) and piezoresponse image (b) of PZT thin film. The inset
shows that the PFM contrast is not random but is due to the small (~50-100 nm) ferroelectric

122
domains associated with grains. PFM image (c) of lines patterned with alternating +10 and -10
Vdc. Surface topography (d) after deposition of Ag nanoparticles. Note one-to-one
correspondence between tip-induced polarization distribution and metal deposition pattern. The
features consist of closely packed metal nanoparticles of 3-10 nm. Piezoresponse image of
checkerboard domain structure fabricated using in-house lithographic system (e) and SEM image
of corresponding silver photodeposition pattern (f). Reprinted with permission from [Kalinin S V,
Bonnell D A, Alvarez T, Lei X, Hu Z, Ferris J H, Zhang Q and Dunn S 2002 Nano Lett. 2 589-
93]. Copyright (2002) ACS Publications [405].
VII. Conclusion and outlook
Ferroelectric surfaces are the inherent part of ferroelectric materials, and via modern scanning
probe imaging techniques provide the window into the bulk physics and domain dynamics.
However, the presence of the normal polarization component at surfaces necessitates polarization
screening that can be realized via surface band bending or dissociative chemisorption of reactive
species. Consequently, screening charge dynamics becomes non-trivial component of the static
and dynamic properties of ferroelectric surfaces and near surface regions. Polarization cannot be
switched unless screening charge redistribute – hence transport of screening species can become
rate-limiting step of switching phenomena. The conservation of screening charge species can
give rise to nontrivial static and dynamic phenomena, as illustrated by bistability of domain wall
motion and formation of chaotic structures. However, it also controls switching that leads to
classical domain geometries, and hence should be considered along with bulk wall motion and
pinning mechanisms. Surface screening significantly affects the basic structure of the
ferroelectric domain wall-surface junctions, and hence can be highly relevant for domain wall
conduction mechanisms as observed by AFM.[13, 15, 18, 21, 22, 408-410] Finally, in thin films
the coupling between the ferroelectricity and surface electrochemistry can give rise to the
continuum of coupled ferroelectric-electrochemical states, reminiscent of relaxors.
Scanning probe microscopy techniques have provided valuable insight into structure and
properties of ferroelectric surfaces via dual observables of surface potential and
electromechanical response. Combined with variable temperature and time dependent studies,
these observations provide the definitive evidence towards surface ionic screening. However,
observations such as metal photodeposition suggest that there are also changes in the surface

123
electronic structures. These considerations necessitate local structure- and functionality sensitive
probing via electron microscopy and scattering techniques, as e.g. demonstrated recently by
Evans group for switching effect on local crystallography.[411]. In the future, direct correlation
between probe and X-ray imaging tools in the form of multimodal imaging will allow to enable
direct data mining [412] of structure property relationships of ferroelectric surfaces.
Finally, understanding of ferroelectric surfaces necessitates the development and
mesoscopic theoretical models that capture and combine ferroelectric and (electro)chemical
degrees of freedom, and can be applied for other polar oxide surfaces.
Acknowledgement
Research was supported (S.V.K. and D.F.) by the U.S. Department of Energy, Office of Science,
Basic Energy Sciences, Materials Sciences and Engineering Division. This research was
conducted at the Center for Nanophase Materials Sciences, which is a DOE Office of Science
User Facility. This work was partially supported (Y.K.) by Basic Science Research program
through the National Research Foundation of Korea funded by the Ministry of Science, ICT &
Future Planning (NRF-2014R1A4A1008474).

124
References
[1] Tagantsev A K, Cross L E and Fousek J 2010 Domains in Ferroic Crystals and Thin Films (New
York: Springer)
[2] Scott J F 2007 Science 315 954-9
[3] Setter N, Damjanovic D, Eng L, Fox G, Gevorgian S, Hong S, Kingon A, Kohlstedt H, Park N Y,
Stephenson G B, Stolitchnov I, Tagantsev A K, Taylor D V, Yamada T and Streiffer S 2006 J.
Appl. Phys. 100 051606
[4] Vugmeister B E 2006 Phys. Rev. B 73 174117.
[5] Glinchuk M D and Stephanovich V A 1999 J. Appl. Phys. 85 1722-6
[6] Glinchuk M D and Stephanovich V A 1998 J. Phys.-Condens. Matter 10 11081-94
[7] Westphal V, Kleemann W and Glinchuk M D 1992 Phys. Rev. Lett. 68 847-50
[8] Spaldin N A and Fiebig M 2005 Science 309 391-2
[9] Fiebig M 2005 J. Phys. D-Appl. Phys. 38 R123-R52
[10] Fiebig M, Lottermoser T, Frohlich D, Goltsev A V and Pisarev R V 2002 Nature 419 818-20
[11] Wang J, Neaton J B, Zheng H, Nagarajan V, Ogale S B, Liu B, Viehland D, Vaithyanathan V,
Schlom D G, Waghmare U V, Spaldin N A, Rabe K M, Wuttig M and Ramesh R 2003 Science
299 1719-22
[12] Cheong S W and Mostovoy M 2007 Nat. Mater. 6 13-20
[13] Seidel J, Martin L W, He Q, Zhan Q, Chu Y H, Rother A, Hawkridge M E, Maksymovych P, Yu
P, Gajek M, Balke N, Kalinin S V, Gemming S, Wang F, Catalan G, Scott J F, Spaldin N A,
Orenstein J and Ramesh R 2009 Nat. Mater. 8 229-34
[14] Daraktchiev M, Catalan G and Scott J F 2010 Phys. Rev. B 81 224118
[15] Vasudevan R K, Wu W D, Guest J R, Baddorf A P, Morozovska A N, Eliseev E A, Balke N,
Nagarajan V, Maksymovych P and Kalinin S V 2013 Adv. Funct. Mater. 23 2592-616
[16] Balke N, Winchester B, Ren W, Chu Y H, Morozovska A N, Eliseev E A, Huijben M, Vasudevan
R K, Maksymovych P, Britson J, Jesse S, Kornev I, Ramesh R, Bellaiche L, Chen L Q and
Kalinin S V 2012 Nat. Phys. 8 81-8
[17] Maksymovych P, Morozovska A N, Yu P, Eliseev E A, Chu Y H, Ramesh R, Baddorf A P and
Kalinin S V 2012 Nano Lett. 12 209-13
[18] Eliseev E A, Morozovska A N, Svechnikov G S, Maksymovych P and Kalinin S V 2012 Phys.
Rev. B 85 045312
[19] Morozovska A N, Vasudevan R K, Maksymovych P, Kalinin S V and Eliseev E A 2012 Phys.
Rev. B 86 085315
[20] Maksymovych P, Seidel J, Chu Y H, Wu P P, Baddorf A P, Chen L Q, Kalinin S V and Ramesh
R 2011 Nano Lett. 11 1906-12
[21] Farokhipoor S and Noheda B 2011 Phys. Rev. Lett. 107 127601
[22] Guyonnet J, Gaponenko I, Gariglio S and Paruch P 2011 Adv. Mater. 23 5377-82
[23] Stachiotti M G and Sepliarsky M 2011 Phys. Rev. Lett. 106 137601
[24] Fong D D, Kolpak A M, Eastman J A, Streiffer S K, Fuoss P H, Stephenson G B, Thompson C,
Kim D M, Choi K J, Eom C B, Grinberg I and Rappe A M 2006 Phys. Rev. Lett. 96 127601
[25] Fister T T, Fong D D, Eastman J A, Iddir H, Zapol P, Fuoss P H, Balasubramanian M, Gordon R
A, Balasubramaniam K R and Salvador P A 2011 Phys. Rev. Lett. 106 037401
[26] Highland M J, Fister T T, Fong D D, Fuoss P H, Thompson C, Eastman J A, Streiffer S K and
Stephenson G B 2011 Phys. Rev. Lett. 107 187602
[27] Streiffer S K, Eastman J A, Fong D D, Thompson C, Munkholm A, Murty M V R, Auciello O,
Bai G R and Stephenson G B 2002 Phys. Rev. Lett. 89 067601
[28] Jia C L, Urban K W, Alexe M, Hesse D and Vrejoiu I 2011 Science 331 1420-3

125
[29] Jia C L, Nagarajan V, He J Q, Houben L, Zhao T, Ramesh R, Urban K and Waser R 2007 Nat.
Mater. 6 64-9
[30] Jia C L, Mi S B, Urban K, Vrejoiu I, Alexe M and Hesse D 2008 Nat. Mater. 7 57-61
[31] Borisevich A Y, Chang H J, Huijben M, Oxley M P, Okamoto S, Niranjan M K, Burton J D,
Tsymbal E Y, Chu Y H, Yu P, Ramesh R, Kalinin S V and Pennycook S J 2010 Phys. Rev. Lett.
105 087204
[32] Borisevich A Y, Morozovska A N, Kim Y M, Leonard D, Oxley M P, Biegalski M D, Eliseev E
A and Kalinin S V 2012 Phys. Rev. Lett. 109 065702
[33] Borisevich A Y, Lupini A R, He J, Eliseev E A, Morozovska A N, Svechnikov G S, Yu P, Chu Y
H, Ramesh R, Pantelides S T, Kalinin S V and Pennycook S J 2012 Phys. Rev. B 86 140102
[34] Chang H J, Kalinin S V, Morozovska A N, Huijben M, Chu Y H, Yu P, Ramesh R, Eliseev E A,
Svechnikov G S, Pennycook S J and Borisevich A Y 2011 Adv. Mater. 23 2474-9
[35] Kingsmith R D and Vanderbilt D 1993 Phys. Rev. B 47 1651-4
[36] Vanderbilt D and Kingsmith R D 1993 Phys. Rev. B 48 4442-55
[37] Resta R, Posternak M and Baldereschi A 1993 Phys. Rev. Lett. 70 1010-3
[38] Dawber M, Rabe K M and Scott J F 2005 Rev. Mod. Phys. 77 1083-130
[39] Rosenman G, Shur D, Krasik Y E and Dunaevsky A 2000 J. Appl. Phys. 88 6109-61
[40] Puchkarev V F and Mesyats G A 1995 J. Appl. Phys. 78 5633-7
[41] Riege H 1994 Nucl. Instrum. Methods Phys. Res. Sect. A-Accel. Spectrom. Dect. Assoc. Equip.
340 80-9
[42] Rosenman G and Rez I 1993 J. Appl. Phys. 73 1904-8
[43] Klopfer M, Wolowiec T, Satchouk V, Alivov Y and Molloi S 2012 Nucl. Instrum. Methods Phys.
Res. Sect. A-Accel. Spectrom. Dect. Assoc. Equip. 689 47-51
[44] Naranjo B, Gimzewski J K and Putterman S 2005 Nature 434 1115-7
[45] Thompson C, Fong D D, Wang R V, Jiang F, Streiffer S K, Latifi K, Eastman J A, Fuoss P H and
Stephenson G B 2008 Appl. Phys. Lett. 93 182901
[46] Ohtomo A, Muller D A, Grazul J L and Hwang H Y 2002 Nature 419 378-80
[47] Mannhart J and Schlom D G 2010 Science 327 1607-11
[48] Zhong Z C, Xu P X and Kelly P J 2010 Phys. Rev. B 82 165127
[49] Kalabukhov A, Gunnarsson R, Borjesson J, Olsson E, Claeson T and Winkler D 2007 Phys. Rev.
B 75 121404
[50] Cheng G L, Siles P F, Bi F, Cen C, Bogorin D F, Bark C W, Folkman C M, Park J W, Eom C B,
Medeiros-Ribeiro G and Levy J 2011 Nat. Nanotech. 6 343-7
[51] Bi F, Bogorin D F, Cen C, Bark C W, Park J W, Eom C B and Levy J 2010 Appl. Phys. Lett. 97
173110
[52] Bark C W, Sharma P, Wang Y, Baek S H, Lee S, Ryu S, Folkman C M, Paudel T R, Kumar A,
Kalinin S V, Sokolov A, Tsymbal E Y, Rzchowski M S, Gruverman A and Eom C B 2012 Nano
Lett. 12 1765-71
[53] Kumar A, Arruda T M, Kim Y, Ivanov I N, Jesse S, Bark C W, Bristowe N C, Artacho E,
Littlewood P B, Eom C B and Kalinin S V 2012 ACS Nano 6 3841-52
[54] Kittel C 1995 Introduction to solid state physics (Wiley)
[55] Catalan G, Scott J F, Schilling A and Gregg J M 2007 J. Phys.-Condens. Matter 19 132201
[56] Dawber M, Chandra P, Littlewood P B and Scott J F 2003 J. Phys.-Condens. Matter 15 L393-L8
[57] Chisholm M F, Luo W D, Oxley M P, Pantelides S T and Lee H N 2010 Phys. Rev. Lett. 105
197602
[58] Aravind V R, Morozovska A N, Bhattacharyya S, Lee D, Jesse S, Grinberg I, Li Y L, Choudhury
S, Wu P, Seal K, Rappe A M, Svechnikov S V, Eliseev E A, Phillpot S R, Chen L Q, Gopalan V
and Kalinin S V 2010 Phys. Rev. B 82 024111
[59] Bocher L, Gloter A, Crassous A, Garcia V, March K, Zobelli A, Valencia S, Enouz-Vedrenne S,
Moya X, Marthur N D, Deranlot C, Fusil S, Bouzehouane K, Bibes M, Barthelemy A, Colliex C
and Stephan O 2012 Nano Lett. 12 376-82

126
[60] Kalinin S V and Bonnell D A 2004 Nano Lett. 4 555-60
[61] Jesse S, Baddorf A P and Kalinin S V 2006 Appl. Phys. Lett. 88 062908
[62] Morozovska A N, Svechnikov S V, Eliseev E A, Jesse S, Rodriguez B J and Kalinin S V 2007 J.
Appl. Phys. 102 114108
[63] Ievlev A V, Jesse S, Morozovska A N, Strelcov E, Eliseev E A, Pershin Y V, Kumar A, Shur V Y
and Kalinin S V 2014 Nat. Phys. 10 59-66
[64] Ievlev A V, Morozovska A N, Shur V Y and Kalinin S V 2014 Appl. Phys. Lett. 104 092908
[65] Bardeen J 1947 Phys. Rev. 71 717-27
[66] Kreher K 1980 Kristall und Technik 15 1392
[67] Pintilie L, Stancu V, Trupina L and Pintilie I 2010 Phys. Rev. B 82 085319
[68] Morozovska A N, Eliseev E A, Svechnikov S V, Krutov A D, Shur V Y, Borisevich A Y,
Maksymovych P and Kalinin S V 2010 Phys. Rev. B 81 205308
[69] Itskovsky M A 1974 Fiz. Tverd. Tela (Leningrad) 16 2065
[70] Tagantsev A K, Cross L E and Fousek J 2010 Domains in ferroic crystals and thin film (New
York: Springer)
[71] Eliseev E A, Kalinin S V and Morozovska A N 2015 J. Appl. Phys. 117 034102
[72] Bratkovsky A M and Levanyuk A P 2009 J. Comput. Theor. Nanos. 6 465-89
[73] Bratkovsky A M and Levanyuk A P 2011 Phys. Rev. B. 84 045401
[74] Chenskii E and Tarasenko V 1982 Zh. Eksp. Teor. Fiz. 83 1089-99
[75] Jia C L, Urban K W, Alexe M, Hesse D and Vrejoiu I 2011 Science 331 1420-3
[76] Tang Y L, Zhu Y L, Ma X L, Borisevich A Y, Morozovska A N, Eliseev E A, Wang W Y, Wang
Y J, Xu Y B, Zhang Z D and Pennycook S J 2015 Science 348 547-51
[77] Darinskii B M, Lazarev A P and Sidorkin A S 1989 Fizika tverdogo tela 31 287-9.
[78] Eliseev E A, Morozovska A N, Kalinin S V, Li Y L, Shen J, Glinchuk M D, Chen L Q and
Gopalan V 2009 J. Appl. Phys. 106 084102
[79] Shur V Y, Gruverman A, Kuminov V and Tonkachyova N 1990 Ferroelectrics 111 197-206
[80] Eliseev E A, Morozovska A N, Svechnikov G S, Rumyantsev E L, Shishkin E I, Shur V Y and
Kalinin S V 2008 Phys. Rev. B 78 245409
[81] Schmickler W 1996 Interfacial Electrochemistry (Oxford University Press, Oxford) ch 4
[82] Timoshenko S P and Goodier J N 1970 Theory of Elasticity (New York: McGraw-Hill)
[83] Landau L D and Lifshitz E M 1986 Course of Theoretical Physics: Vol. 7 Theory of Elasticity
(Butterworth-Heinemann, Oxford) 3 109
[84] Glinchuk M D, Eliseev E A, Stephanovich V A and Farhi R 2003 J. Appl. Phys. 93 1150-9
[85] Glinchuk M D, Eliseev E A, Deineka A, Jastrabik L, Suchaneck G, Sandner T, Gerlach G and
Hrabovsky M 2001 Integr. Ferroelectr. 38 745-54
[86] Woo C H and Zheng Y 2008 Appl. Phys. A-Mater. 91 59-63
[87] Damon Y and Cho Y 2006 Jpn. J. Appl. Phys. 2 45 L1304-L6
[88] Daimon Y and Cho Y 2007 Appl. Phys. Lett. 90 192906
[89] Kalinin S V and Bonnell D A 2002 Electrically Based Microstructural Characterization III, ed.
Gerhardt R A et al. pp 101-6
[90] Tagantsev A K and Gerra G 2006 J. Appl. Phys. 100 051607
[91] Catalan G, Bea H, Fusil S, Bibes M, Paruch P, Barthelemy A and Scott J F 2008 Phys. Rev. Lett.
100 027602
[92] This behavior can be illustrated by considering a simple system consisting of the finite strip with
width b with constant surface charge placed at the distance H from the electrode. In the case the
full field distribution can be obtained in the evident form. The field behaves as at the film surface
near the edge of the strip and vanishes with b increase.
[93] Sawa A 2008 Mater. Today 11 28-36
[94] Magyari-Kope B, Tendulkar M, Park S G, Lee H D and Nishi Y 2011 Nanotechnology 22 254029
[95] Shang D S, Shi L, Sun J R and Shen B G 2011 Nanotechnology 22 254008

127
[96] Kim Y, Kelly S J, Morozovska A, Rahani E K, Strelcov E, Eliseev E, Jesse S, Biegalski M D,
Balke N, Benedek N, Strukov D, Aarts J, Hwang I, Oh S, Choi J S, Choi T, Park B H, Shenoy V
B, Maksymovych P and Kalinin S V 2013 Nano Lett. 13 4068-74
[97] Waser R and Aono M 2007 Nat. Mater. 6 833-40
[98] Szot K, Rogala M, Speier W, Klusek Z, Besmehn A and Waser R 2011 Nanotechnology 22
254001
[99] Morozovska A N, Eliseev E A and Kalinin S V 2010 Appl. Phys. Lett. 96 222906
[100] Morozovska A N, Eliseev E A, Balke N and Kalinin S V 2010 J. Appl. Phys. 108 053712
[101] Balke N, Jesse S, Morozovska A N, Eliseev E, Chung D W, Kim Y, Adamczyk L, Garcia R E,
Dudney N and Kalinin S V 2010 Nat. Nanotech. 5 749-54
[102] Morozovska A N, Eliseev E A, Varenyk O V, Kim Y, Strelcov E, Tselev A, Morozovsky N V
and Kalinin S V 2014 J. Appl. Phys. 116 066808
[103] McLachlan M A, McComb D W, Ryan M P, Morozovska A N, Eliseev E A, Payzant E A, Jesse
S, Seal K, Baddorf A P and Kalinin S V 2011 Adv. Funct. Mater. 21 941-7
[104] Morozovska A N, Eliseev E A, Svechnikov G S and Kalinin S V 2011 Phys. Rev. B 84 045402
[105] Morozovska A N, Eliseev E A and Kalinin S V 2012 J. Appl. Phys. 111 014114
[106] Morozovska A N, Eliseev E A, Bravina S L, Ciucci F, Svechnikov G S, Chen L Q and Kalinin S
V 2012 J. Appl. Phys. 111 014107
[107] Chua L 1971 IEEE Trans. Circuit Theory 18 507-19
[108] Chua L O and Sung Mo K 1976 P. IEEE 64 209-23
[109] Mathur N D 2008 Nature 455 E13
[110] Gil Y, Umurhan O M and Riess I 2007 Solid State Ion. 178 1-12
[111] Gil Y, Umurhan O M and Riess I 2008 J. Appl. Phys. 104 084504
[112] Allen S M and Cahn J W 1979 Acta Metall. 27 1085-95
[113] Zhang X C, Shyy W and Sastry A M 2007 J. Electrochem. Soc. 154 A910-A6
[114] Tang M, Huang H Y, Meethong N, Kao Y H, Carter W C and Chiang Y M 2009 Chem. Mater. 21
1557-71
[115] Tang M, Carter W C, Belak J F and Chiang Y-M 2010 Electrochim. Acta 56 969-76
[116] Morozovska A N, Glinchuk M D, 2016 J. Appl. Phys. 119, 094109
[117] Morozovska A N, Eliseev E A and Kalinin S V (unpublished)
[118] Highland M J, Fister T T, Fong D D, Fuoss P H, Thompson C, Eastman J A, Streiffer S K and
Stephenson G B 2011 Phys. Rev. Lett. 107 187602
[119] Stephenson G B and Highland M J 2011 Phys. Rev. B 84 064107
[120] Morozovska A N, Eliseev E A, Genenko Y A, Vorotiahin I S, Silibin M V, Cao Y, Kim Y,
Glinchuk M D and Kalinin S V 2016 Phys. Rev. B 94 174101
[121] Waser R, Böttger U and Tiedke S 2004 Polar Oxides: Properties, Characterization, and Imaging
(Wiley) ch 8
[122] Fong D D and Thompson C 2006 Annu. Rev. Mater. Res. 36 431-65
[123] Shin J, Nascimento V B, Borisevich A Y, Plummer E W, Kalinin S V and Baddorf A P 2008
Phys. Rev. B 77 245437
[124] Shin J, Nascimento V B, Geneste G, Rundgren J, Plummer E W, Dkhil B, Kalinin S V and
Baddorf A P 2009 Nano Lett. 9 3720-5
[125] Wang J L, Gaillard F, Pancotti A, Gautier B, Niu G, Vilquin B, Pillard V, Rodrigues G L M P
and Barrett N 2012 J. Phys. Chem. C 116 21802-9
[126] Stephenson G B, Fong D D, Murty M V R, Streiffer S K, Eastman J A, Auciello O, Fuoss P H,
Munkholm A, Aanerud M E M and Thompson C 2003 Physica B 336 81-9
[127] Rusanov A I 2005 Surf. Sci. Rep. 58 111-239
[128] Cammarata R C 2009 Solid State Phys. 61 1-75
[129] Munkholm A, Streiffer S K, Murty M V R, Eastman J A, Thompson C, Auciello O, Thompson L,
Moore J F and Stephenson G B 2002 Phys. Rev. Lett. 88 016101

128
[130] Bousquet E, Dawber M, Stucki N, Lichtensteiger C, Hermet P, Gariglio S, Triscone J M and
Ghosez P 2008 Nature 452 732-6
[131] Fong D D, Stephenson G B, Streiffer S K, Eastman J A, Auciello O, Fuoss P H and Thompson C
2004 Science 304 1650-3
[132] Stephenson G B and Elder K R 2006 J. Appl. Phys. 100 051601
[133] Fong D D, Cionca C, Yacoby Y, Stephenson G B, Eastman J A, Fuoss P H, Streiffer S K,
Thompson C, Clarke R, Pindak R and Stern E A 2005 Phys. Rev. B. 71 144112
[134] Wang R V, Fong D D, Jiang F, Highland M J, Fuoss P H, Thompson C, Kolpak A M, Eastman J
A, Streiffer S K, Rappe A M and Stephenson G B 2009 Phys. Rev. Lett. 102 047601
[135] Highland M J, Fister T T, Richard M I, Fong D D, Fuoss P H, Thompson C, Eastman J A,
Streiffer S K and Stephenson G B 2010 Phys. Rev. Lett. 105 167601
[136] Alahmed Z and Fu H X 2007 Phys. Rev. B 76 224101
[137] Apostol N G, Stoflea L E, Lungu G A, Tache C A, Popescu D G, Pintilie L and Teodorescu C M
2013 Mater. Sci. Eng. B-Adv. 178 1317-22
[138] Stoflea L E, Apostol N G, Trupina L and Teodorescu C M 2014 J. Mater. Chem. A 2 14386-92
[139] Watanabe Y, Okano M and Masuda A 2001 Phys. Rev. Lett. 86 332-5
[140] He J, Stephenson G B and Nakhmanson S M 2012 J. Appl. Phys. 112 054112
[141] Hong S, Nakhmanson S M and Fong D D 2016 Rep. Prog. Phys. 79 076501
[142] Liviu Cristian Tănase N G A, Laura Elena Abramiuc, Cristian Alexandru Tache, Luminița Hrib,
Lucian Trupină, Lucian Pintilie and Cristian Mihail Teodorescu 2016 Sci. Rep. 6 35301
[143] Nassreddine S, Morfin F, Niu G, Vilquin B, Gaillard F and Piccolo L 2014 Surf. Interface Anal.
46 721-5
[144] Kakekhani A and Ismail-Beigi S 2016 J. Mater. Chem. A 4 5235-46
[145] Kakekhani A and Ismail-Beigi S 2016 Phys. Chem. Chem. Phys. 18 19676-95
[146] Kolpak A M, Grinberg I and Rappe A M 2007 Phys. Rev. Lett. 98 166101
[147] Kim S, Schoenberg M R and Rappe A M 2011 Phys. Rev. Lett. 107 076102
[148] Yasunobu Inoue I Y, and Kazunori Sat 1984 J. Phys. Chem. 88 1148-51
[149] Yun Y, Pilet N, Schwarz U D and Altman E I 2009 Surf. Sci. 603 3145-54
[150] Garrity K, Kakekhani A, Kolpak A and Ismail-Beigi S 2013 Phys. Rev. B 88 045401
[151] Kakekhani A and Ismail-Beigi S 2015 ACS Catal. 5 4537-45
[152] Khan M A, Nadeem M A and Idrissn H 2016 Surf. Sci. Rep. 71 1-31
[153] Yun Y, Kampschulte L, Li M, Liao D and Altman E I 2007 J. Phys. Chem. C 111 13951-6
[154] Yun Y and Altman E I 2007 J. Am. Chem. Soc. 129 15684-9
[155] Kakekhani A, Ismail-Beigi S and Altman E I 2016 Surf. Sci. 650 302-16
[156] Garra J, Vohs J M and Bonnell D A 2009 Surf. Sci. 603 1106-14
[157] Garra J, Vohs J M and Bonnell D A 2009 J. Vac. Sci. Technol. A 27 L13-L7
[158] Popescu D G, Husanu M A, Trupina L, Hrib L, Pintilie L, Barinov A, Lizzit S, Lacovig P and
Teodorescu C M 2015 Phys. Chem. Chem. Phys. 17 509-20
[159] Husanu M A, Popescu D G, Tache C A, Apostol N G, Barinov A, Lizzit S, Lacovig P and
Teodorescu C M 2015 Appl. Surf. Sci. 352 73-81
[160] Rault J E, Mentes T O, Locatelli A and Barrett N 2014 Sci. Rep. 4 6792
[161] Thomas A G, Muryn C A, Hardman P J, Dürr H A, Owen I W, Thornton G, Quinn F M,
Rosenberg R A, Love P J and Rehn V 1994 Surf. Sci. 307-309 355-9
[162] Li D B and Bonnell D A 2008 Annu. Rev. Mater. Res. 38 351-68
[163] Nelson C T, Gao P, Jokisaari J R, Heikes C, Adamo C, Melville A, Baek S H, Folkman C M,
Winchester B, Gu Y J, Liu Y M, Zhang K, Wang E G, Li J Y, Chen L Q, Eom C B, Schlom D G
and Pan X Q 2011 Science 334 968-71
[164] Cherifi S, Hertel R, Fusil S, Bea H, Bouzehouane K, Allibe J, Bibes M and Barthelemy A 2010
Phys. Status Solidi-R 4 22-4
[165] Wang Y G, Dec J and Kleemann W 1998 J. Appl. Phys. 84 6795-9

129
[166] Gruverman A L, Hatano J and Tokumoto H 1997 Jpn. J. Appl. Phys. Part 1 - Regul. Pap. Short
Notes Rev. Pap. 36 2207-11
[167] Takashige M, Hamazaki S, Fukurai N and Shimizu F 1997 J. Phys. Soc. Jpn. 66 1848-9
[168] Takashige M, Hamazaki S I, Fukurai N, Shimizu F and Kojima S 1997 Ferroelectrics 203 221-5
[169] Takashige M, Hamazaki S I, Fukurai N, Shimizu F and Kojima S 1996 Jpn. J. Appl. Phys. Part 1
- Regul. Pap. Short Notes Rev. Pap. 35 5181-4
[170] Balakumar S, Xu J B, Ma J X, Ganesamoorthy S and Wilson I H 1997 Jpn. J. Appl. Phys. Part 1 -
Regul. Pap. Short Notes Rev. Pap. 36 5566-9
[171] Pang G K H and Baba-Kishi K Z 1998 J. Phys. D-Appl. Phys. 31 2846-53
[172] Eng L M, Friedrich M, Fousek J and Gunter P 1996 Ferroelectrics 186 49-52
[173] Eng L M, Friedrich M, Fousek J and Gunter P 1996 J. Vac. Sci. Technol. B 14 1191-6
[174] Bluhm H, Wiesendanger R and Meyer K P 1996 J. Vac. Sci. Technol. B B 14 1180-3
[175] Shilo D, Ravichandran G and Bhattacharya K 2004 Nat. Mater. 3 453-7
[176] Bluhm H, Schwarz U D and Wiesendanger R 1998 Phys. Rev. B 57 161-9
[177] Correia A, Massanell J, Garcia N, Levanyuk A P, Zlatkin A and Przeslawski J 1996 Appl. Phys.
Lett. 68 2796-8
[178] Tolstikhina A L, Belugina N V and Shikin S A 2000 Ultramicroscopy 82 149-52
[179] Bluhm H, Schwarz U D and Meyer K P 1995 Appl. Phys. A-Mater. Sci. Process. 61 525-33
[180] Labardi M, Allegrini M, Fuso F, Leccabue F, Watts B, Ascoli C and Frediani C 1995 Integr.
Ferroelectr. 8 143-50
[181] Qi Y B, Park J Y, Hendriksen B L M, Ogletree D F and Salmeron M 2008 Phys. Rev. B 77
184105
[182] Park J Y, Qi Y B, Ogletree D F, Thiel P A and Salmeron M 2007 Phys. Rev. B 76 064108
[183] Park J Y, Ogletree D F, Thiel P A and Salmeron M 2006 Science 313 186
[184] Rabe U, Kopycinska M, Hirsekorn S, Saldana J M, Schneider G A and Arnold W 2002 J. Phys.
D-Appl. Phys. 35 2621-35
[185] Liu X X, Heiderhoff R, Abicht H P and Balk L J 2002 J. Phys. D-Appl. Phys. 35 74-87
[186] Li D B, Zhao M H, Garra J, Kolpak A M, Rappe A M, Bonnell D A and Vohs J M 2008 Nat.
Mater. 7 473-7
[187] Yin Q R, Zeng H R, Yu H F and Li G R 2006 J. Mater. Sci. 41 259-70
[188] Yu H F, Zeng H R, Ding A L, Li G R, Luo H S and Yin Q R 2006 Integr. Ferroelectr. 78 309-17
[189] Yu H F, Zeng H R, Ding A L, Li G R, Luo H S and Yin Q R 2005 Integr. Ferroelectr. 73 165-71
[190] Yu H F, Zeng H R, Ma X D, Chu R Q, Li G R, Luo H S and Yin Q R 2005 Phys. Status Solidi A-
Appl. Res. 202 R10-R2
[191] Jesse S and Kalinin S V 2011 J. Phys. D-Appl. Phys. 44 464006
[192] Kalinin S V, Karapetian E and Kachanov M 2004 Phys. Rev. B 70 184101
[193] Karapetian E, Kachanov M and Kalinin S V 2005 Philos. Mag. 85 1017-51
[194] Hong J W, Park S I and Khim Z G 1999 Rev. Sci. Instrum. 70 1735-9
[195] Takata K, Kushida K and Torii K 1995 Jpn. J. Appl. Phys. Part 1 - Regul. Pap. Short Notes Rev.
Pap. 34 2890-4
[196] Franke K and Weihnacht M 1995 Ferroelectr. Lett. Sect. 19 25-33
[197] Franke K, Besold J, Haessler W and Seegebarth C 1994 Surf. Sci. 302 L283-L8
[198] Nonnenmacher M, Oboyle M P and Wickramasinghe H K 1991 Appl. Phys. Lett. 58 2921-3
[199] S. Sadewasser and T. Glatzel E 2011 Kelvin Probe Force Microscopy (New York: Springer
Science)
[200] Kitamura S and Iwatsuki M 1998 Appl. Phys. Lett. 72 3154-6
[201] Zerweck U, Loppacher C, Otto T, Grafstrom S and Eng L M 2005 Phys. Rev. B 71 125424
[202] Loppacher C, Zerweck U, Teich S, Beyreuther E, Otto T, Grafstrom S and Eng L M 2005
Nanotechnology 16 S1-S6
[203] Sommerhalter C, Matthes T W, Glatzel T, Jager-Waldau A and Lux-Steiner M C 1999 Appl.
Phys. Lett. 75 286-8

130
[204] Ziegler D and Stemmer A 2011 Nanotechnology 22 075501
[205] Liscio A, Palermo V and Samori P 2010 Accounts Chem. Res. 43 541-50
[206] Kalinin S V and Bonnell D A 2002 Phys. Rev. B 65 125408
[207] Huey B D, Ramanujan C, Bobji M, Blendell J, White G, Szoszkiewicz R and Kulik A 2004 J.
Electroceram. 13 287-91
[208] Harnagea C, Alexe M, Hesse D and Pignolet A 2003 Appl. Phys. Lett. 83 338-40
[209] Seal K, Jesse S, Rodriguez B J, Baddorf A P and Kalinin S V 2007 Appl. Phys. Lett. 91 232904
[210] Jesse S, Baddorf A P and Kalinin S V 2006 Nanotechnology 17 1615-28
[211] Hong S, Shin H, Woo J and No K 2002 Appl. Phys. Lett. 80 1453-5
[212] Hong S, Woo J, Shin H, Jeon J U, Pak Y E, Colla E L, Setter N, Kim E and No K 2001 J. Appl.
Phys. 89 1377-86
[213] Gruverman A, Auciello O and Tokumoto H 1998 Annu. Rev. Mater. Sci. 28 101-23
[214] Gruverman A, Auciello O, Ramesh R and Tokumoto H 1997 Nanotechnology 8 A38-A43
[215] Eng L M, Bammerlin M, Loppacher C, Guggisberg M, Bennewitz R, Luthi R, Meyer E, Huser T,
Heinzelmann H and Guntherodt H J 1999 Ferroelectrics 222 153-62
[216] Bonnell D A, Kalinin S V, Kholkin A L and Gruverman A 2009 MRS Bulletin 34 648-57
[217] Gruverman A and Kholkin A 2006 Rep. Prog. Phys. 69 2443-74
[218] Gruverman A and Kalinin S V 2006 J. Mater. Sci. 41 107-16
[219] Kalinin S V, Rar A and Jesse S 2006 IEEE T. Ultrason. Ferr. 53 2226-52
[220] Balke N, Bdikin I, Kalinin S V and Kholkin A L 2009 J. Am. Ceram. Soc. 92 1629-47
[221] Kalinin S V, Morozovska A N, Chen L Q and Rodriguez B J 2010 Rep. Prog. Phys. 73 056502
[222] Kalinin S V, Setter N and Kholkin A L 2009 MRS Bulletin 34 634-42
[223] Soergel E 2011 J. Phys. D-Appl. Phys. 44 464003
[224] Kalinin S V, Jesse S, Rodriguez B J, Seal K, Baddorf A P, Zhao T, Chu Y H, Ramesh R, Eliseev
E A, Morozovska A N, Mirman B and Karapetian E 2007 Jpn. J. Appl. Phys. Part 1 - Regul. Pap.
Short Notes Rev. Pap. 46 5674-85
[225] Cunningham S, Larkin I A and Davis J H 1998 Appl. Phys. Lett. 73 123-5
[226] Terris B D, Stern J E, Rugar D and Mamin H J 1989 Phys. Rev. Lett. 63 2669-72
[227] Groten J, Zirkl M, Jakopic G, Leitner A and Stadlober B 2010 Phys. Rev. B 82 054112
[228] Canet-Ferrer J, Martin-Carron L, Martinez-Pastor J, Valdes J L, Pena A, Carvajal J J and Diaz F
2007 J. Microsc.-Oxford 226 133-9
[229] Levy J, Hubert C and Trivelli A 2000 J. Chem. Phys. 112 7848-55
[230] Eng L M and Guntherodt H J 2000 Ferroelectrics 236 35
[231] Sakai A, Sasaki N, Tamate T and Ninomiya T 2003 Ferroelectrics 284 189-93
[232] Odagawa H and Cho Y 2000 Surf. Sci. 463 L621-L5
[233] Odagawa H and Cho Y 2000 Jpn. J. Appl. Phys. Part 1 - Regul. Pap. Short Notes Rev. Pap. 39
5719-22
[234] Cho Y S 2011 J. Mater. Res. 26 2007-16
[235] Tanaka K, Kurihashi Y, Uda T, Daimon Y, Odagawa N, Hirose R, Hiranaga Y and Cho Y 2008
Jpn. J. Appl. Phys. 47 3311-25
[236] Sugimura H, Ishida Y, Hayashi K, Takai O and Nakagiri N 2002 Appl. Phys. Lett. 80 1459-61
[237] Fujihira M 1999 Annu. Rev. Mater. Sci. 29 353-80
[238] Melitz W, Shen J, Kummel A C and Lee S 2011 Surf. Sci. Rep. 66 1-27
[239] Baumgart C, Helm M and Schmidt H 2009 Phys. Rev. B 80 085305
[240] Jacobs H O, Knapp H F and Stemmer A 1999 Rev. Sci. Instrum. 70 1756-60
[241] Charrier D S H, Kemerink M, Smalbrugge B E, de Vries T and Janssen R A J 2008 ACS Nano 2
622-6
[242] Sadewasser S, Glatzel T, Shikler R, Rosenwaks Y and Lux-Steiner M C 2003 Appl. Surf. Sci. 210
32-6
[243] Blumel A, Plank H, Klug A, Fisslthaler E, Sezen M, Grogger W and List E J W 2010 Rev. Sci.
Instrum. 81 056107

131
[244] Kalinin S V and Bonnell D A 2001 Phys. Rev. B 63 125411
[245] Kalinin S V and Bonnell D A 2000 Phys. Rev. B 62 10419-30
[246] Wu Y and Shannon M A 2006 Rev. Sci. Instrum. 77 043711
[247] Efimov A and Cohen S R 2000 J. Vac. Sci. Technol. A-Vac. Surf. Films 18 1051-5
[248] Strassburg E, Boag A and Rosenwaks Y 2005 Rev. Sci. Instrum. 76 083705
[249] Jesse S, Guo S, Kumar A, Rodriguez B J, Proksch R and Kalinin S V 2010 Nanotechnology 21
405703
[250] Sadewasser S, Jelinek P, Fang C K, Custance O, Yamada Y, Sugimoto Y, Abe M and Morita S
2009 Phys. Rev. Lett. 103 266103
[251] Bocquet F, Nony L, Loppacher C and Glatzel T 2008 Phys. Rev. B 78 035410
[252] Barth C, Foster A S, Henry C R and Shluger A L 2011 Adv. Mater. 23 477-501
[253] Okamoto K, Sugawara Y and Morita S 2003 Jpn. J. Appl. Phys. Part 1 - Regul. Pap. Short Notes
Rev. Pap. 42 7163-8
[254] Lee W, Lee M, Kim Y B and Prinz F B 2009 Nanotechnology 20 445706
[255] Glatzel T, Marron D F, Schedel-Niedrig T, Sadewasser S and Lux-Steiner M C 2002 Appl. Phys.
Lett. 81 2017-9
[256] Visoly-Fisher I, Cohen S R, Gartsman K, Ruzin A and Cahen D 2006 Adv. Funct. Mater. 16 649-
60
[257] Coffey D C and Ginger D S 2006 Nat. Mater. 5 735-40
[258] Palermo V, Palma M and Samori P 2006 Adv. Mater. 18 145-64
[259] Ellison D J, Lee B, Podzorov V and Frisbie C D 2011 Adv. Mater. 23 502-7
[260] Hoppe H, Glatzel T, Niggemann M, Hinsch A, Lux-Steiner M C and Sariciftci N S 2005 Nano
Lett. 5 269-74
[261] Kalinin S V, Shin J, Jesse S, Geohegan D, Baddorf A P, Lilach Y, Moskovits M and Kolmakov A
2005 J. Appl. Phys. 98 044503
[262] Schujman S B, Vajtai R, Biswas S, Dewhirst B, Schowalter L J and Ajayan P 2002 Appl. Phys.
Lett. 81 541-3
[263] Woodside M T and McEuen P L 2002 Science 296 1098-101
[264] McCormick K L, Woodside M T, Huang M, Wu M S, McEuen P L, Duruoz C and Harris J S
1999 Phys. Rev. B 59 4654-7
[265] Miyato Y, Kobayashi K, Matsushige K and Yamada H 2005 Jpn. J. Appl. Phys. Part 1 - Regul.
Pap. Short Notes Rev. Pap. 44 1633-6
[266] Huey B D and Bonnell D A 2000 Appl. Phys. Lett. 76 1012-4
[267] Bonnell D A, Huey B and Carroll D 1995 Solid State Ion. 75 35-42
[268] Kalinin S V, Suchomel M R, Davies P K and Bonnell D A 2002 J. Am. Ceram. Soc. 85 3011-7
[269] Tanimoto M and Vatel O 1996 J. Vac. Sci. Technol. B 14 1547-51
[270] Vatel O and Tanimoto M 1995 J. Appl. Phys. 77 2358-62
[271] Chung S Y, Kim I D and Kang S J L 2004 Nat. Mater. 3 774-8
[272] Schneider G A, Felten F and McMeeking R M 2003 Acta Mater. 51 2235-41
[273] Kalinin S V and Bonnell D A 1999 Z. Metallkde. 90 983-9
[274] Kalinin S V and Bonnell D A 2000 Ferroelectric Thin Films VIII, ed R W Schwartz et al. pp 327-
32
[275] Shvebelman M M, Agronin A G, Urenski R P, Rosenwaks Y and Rosenman G I 2002 Nano Lett.
2 455-8
[276] Kramer B 2001 Advances in Solid State Physics 41 (Berlin: Springer-Verlag Berlin) pp 287-98
[277] Kim Y, Bae C, Ryu K, Ko H, Kim Y K, Hong S and Shin H 2009 Appl. Phys. Lett. 94 032907
[278] Kalinin S V and Bonnell D A 2001 Phys. Rev. B 63 125411
[279] Kalinin S V, Bonnell D A, Freitag M and Johnson A T 2002 Appl. Phys. Lett. 81 754-6
[280] Kholkin A L, Bdikin I K, Shvartsman V V and Pertsev N A 2007 Nanotechnology 18 095502
[281] Liu X Y, Kitamura K and Terabe K 2006 Appl. Phys. Lett. 89 132905
[282] Choi S, Heo J, Kim D and Chung I S 2004 Thin Solid Films 464 277-81

132
[283] Kim Y, Buhlmann S, Kim J, Park M, No K, Kim Y K and Hong S 2007 Appl. Phys. Lett. 91
052906
[284] Kim Y, Park M, Buhlmann S, Hong S, Kim Y K, Ko H, Kim J and No K 2010 J. Appl. Phys. 107
054103
[285] Chen X Q, Yamada H, Horiuchi T and Matsushige K 1999 Jpn. J. Appl. Phys. Part 1 - Regul.
Pap. Short Notes Rev. Pap. 38 3932-5
[286] Chen X Q, Yamada H, Horiuchi T, Matsushige K, Watanabe S, Kawai M and Weiss P S 1999 J.
Vac. Sci. Technol. B 17 1930-4
[287] Iwata M, Katsuraya K, Suzuki I, Maeda M, Yasuda N and Ishibashi Y 2005 Mater. Sci. Eng. B-
Solid State Mater. Adv. Technol. 120 88-90
[288] Balke N, Gajek M, Tagantsev A K, Martin L W, Chu Y H, Ramesh R and Kalinin S V 2010 Adv.
Funct. Mater. 20 3466-75
[289] Zou X, You L, Chen W G, Ding H, Wu D, Wu T, Chen L and Wang J 2012 ACS Nano 6 8997-
9004
[290] Sharma P, McQuaid R G P, McGilly L J, Gregg J M and Gruverman A 2013 Adv. Mater. 25
1323-30
[291] McQuaid R G P, McGilly L J, Sharma P, Gruverman A and Gregg J M 2011 Nat. Commun. 2
404
[292] Prim W S a R C 1953 Phys. Rev. 90 753-8
[293] Strelcov E, Tselev A, Ivanov I, Budai J D, Zhang J, Tischler J Z, Kravchenko I, Kalinin S V and
Kolmakov A 2012 Nano Lett. 12 6198-205
[294] Strelcov E, Ievlev A V, Jesse S, Kravchenko I I, Shur V Y and Kalinin S V 2014 Adv. Mater. 26
958-63
[295] Strelcov E, Jesse S, Huang Y L, Teng Y C, Kravchenko I I, Chu Y H and Kalinin S V 2013 ACS
Nano 7 6806-15
[296] Ding J L, Strelcov E, Kalinin S V and Bassiri-Gharb N 2015 Nano Lett. 15 3669-76
[297] He D Y, Qiao L J and Volinsky A A 2011 J. Appl. Phys. 110 074104
[298] He D Y, Qiao L J, Volinsky A A, Bai Y and Guo L Q 2011 Phys. Rev. B 84 024101
[299] He D Y, Qiao L J, Volinsky A A, Bai Y, Wu M and Chu W Y 2011 Appl. Phys. Lett. 98 062905
[300] Abplanalp M 2001 PhD thesis Swiss Federal Institute of Technology, Switzerland
[301] Hamazaki S I, Takahashi Y, Shimizu F and Takashige M 2001 Ferroelectrics 251 101-8
[302] Likodimos V, Labardi M, Orlik X K, Pardi L, Allegrini M, Emonin S and Marti O 2001 Phys.
Rev. B 63 064104
[303] Likodimos V, Labardi M and Allegrini M 2002 Phys. Rev. B 66 024104
[304] Kalinin S V and Bonnell D A 2001 Appl. Phys. Lett. 78 1116-8
[305] Kalinin S V and Bonnell D A 2000 J. Appl. Phys. 87 3950-7
[306] Kalinin S V, Johnson C Y and Bonnell D A 2002 J. Appl. Phys. 91 3816-23
[307] Belaidi S, Girard P and Leveque G 1997 J. Appl. Phys. 81 1023-30
[308] Kaku S, Eto H, Nakamura K and Watanabe Y 2009 ISAF: 2009 18th IEEE International
Symposium on the Applications of Ferroelectrics (New York: IEEE) pp 29-32
[309] Watanabe Y 2011 Ferroelectrics 419 28-32
[310] Watanabe Y, Kaku S, Matsumoto D, Urakami Y and Cheong S W 2009 Ferroelectrics 379 381-
91
[311] Fridkin V M 1980 Ferroelectric semiconductors (New York: Consultants Bureau)
[312] Debska M 2005 J Electrostat 63 1017-23
[313] He D Y, Qiao L J, Volinsky A A, Bai Y, Wu M and Chu W Y 2011 Appl. Phys. Lett. 98 062905
[314] Lu X M, Schlaphof F, Grafstrom S, Loppacher C, Eng L M, Suchaneck G and Gerlach G 2002
Appl. Phys. Lett. 81 3215-7
[315] Fumagalli L, Gramse G, Esteban-Ferrer D, Edwards M A and Gomila G 2010 Appl. Phys. Lett.
96 183107
[316] Gramse G, Casuso I, Toset J, Fumagalli L and Gomila G 2009 Nanotechnology 20 395702

133
[317] Fumagalli L, Ferrari G, Sampietro M and Gomila G 2007 Appl. Phys. Lett. 91 243110
[318] Ko H, Ryu K, Park H, Park C, Jeon D, Kim Y K, Jung J, Min D K, Kim Y, Lee H N, Park Y,
Shin H and Hong S 2011 Nano Lett. 11 1428-33
[319] Nicklaus M, Pignolet A, Harnagea C and Ruediger A 2011 Appl. Phys. Lett. 98 162901
[320] Noy A, Vezenov D V and Lieber C M 1997 Annu. Rev. Mater. Sci. 27 381-421
[321] Noy A, Frisbie C D, Rozsnyai L F, Wrighton M S and Lieber C M 1995 J. Am. Chem. Soc. 117
7943-51
[322] Dahan D, Molotskii M, Rosenman G and Rosenwaks Y 2006 Appl. Phys. Lett. 89 152902
[323] Shur V Y, Ievlev A V, Nikolaeva E V, Shishkin E I and Neradovskiy M M 2011 J. Appl. Phys.
110 052017
[324] Spanier J E, Kolpak A M, Urban J J, Grinberg I, Lian O Y, Yun W S, Rappe A M and Park H
2006 Nano Lett. 6 735-9
[325] Urban J J, Spanier J E, Lian O Y, Yun W S and Park H 2003 Adv. Mater. 15 423-6
[326] Gruverman A and Tanaka M 2001 J. Appl. Phys. 89 1836-43
[327] Johann F and Soergel E 2009 Appl. Phys. Lett. 95 232906
[328] Lu H, Bark C W, de los Ojos D E, Alcala J, Eom C B, Catalan G and Gruverman A 2012 Science
336 59-61
[329] Lu H, Kim D J, Bark C W, Ryu S, Eom C B, Tsymbal E Y and Gruverman A 2012 Nano Lett. 12
6289-92
[330] Buh G H, Chung H J and Kuk Y 2001 Appl. Phys. Lett. 79 2010-2
[331] Kim Y, Morozovska A N, Kumar A, Jesse S, Eliseev E A, Alibart F, Strukov D and Kalinin S V
2012 ACS Nano 6 7026-33
[332] Schirmeisen A, Taskiran A, Fuchs H, Bracht H, Murugavel S and Roling B 2007 Phys. Rev. Lett.
98 225901
[333] Nakamura M, Takekawa S, Liu Y W and Kitamura K 2009 Jpn. J. Appl. Phys. 48 020214
[334] Korkishko Y N, Fedorov V A, Baranov E A, Alkaev A N, Morozova T V, Kostritskii S M and
Laurell F 2001 Ferroelectrics 264 1983-8
[335] Kim Y, Cho Y, Hong S, Buhlmann S, Park H, Min D K, Kim S H and No K 2006 Appl. Phys.
Lett. 89 172909
[336] Kim Y, Cho Y, Hong S, Buhlmann S, Park H, Min D K, Kim S H and No K 2006 Appl. Phys.
Lett. 89 162907
[337] Tybell T, Paruch P, Giamarchi T and Triscone J M 2002 Phys. Rev. Lett. 89 097601
[338] Cho Y S, Hashimoto S, Odagawa N, Tanaka K and Hiranaga Y 2005 Appl. Phys. Lett. 87 232907
[339] Tayebi N, Narui Y, Franklin N, Collier C P, Giapis K P, Nishi Y and Zhang Y G 2010 Appl. Phys.
Lett. 96 023103
[340] Molotskii M and Winebrand E 2005 Phys. Rev. B 71 132103
[341] Morozovska A N and Eliseev E A 2006 Physica B 373 54-63
[342] Durkan C, Welland M E, Chu D P and Migliorato P 1999 Phys. Rev. B 60 16198-204
[343] Emelyanov A Y 2005 Phys. Rev. B 71 132102
[344] Kalinin S V, Gruverman A, Rodriguez B J, Shin J, Baddorf A P, Karapetian E and Kachanov M
2005 J. Appl. Phys. 97 074305
[345] Molotskii M I and Shvebelman M M 2005 Philos. Mag. 85 1637-55
[346] Kalinin S V, Rodriguez B J, Jesse S, Chu Y H, Zhao T, Ramesh R, Choudhury S, Chen L Q,
Eliseev E A and Morozovska A N 2007 Proc. Natl. Acad. Sci. USA. 104 20204-9
[347] Chen L Q and Shen J 1998 Comput. Phys. Commun. 108 147-58
[348] Son J Y, Bang S H and Cho J H 2003 Appl. Phys. Lett. 82 3505-7
[349] Kim Y, Hong S, Kim S H and No K 2006 J. Electroceram. 17 185-8
[350] Son J Y, Kyhm K and Cho J H 2006 Appl. Phys. Lett. 89 092907
[351] Kim Y, Kim J, Buhlmann S, Hong S, Kim Y K, Kim S H and No K 2008 Phys. Status Solidi-R 2
74-6

134
[352] Kim J, Kim Y, No K, Buhlmann S, Hong S, Nam Y W and Kim S H 2006 Integ. Ferroelectr. 85
25-30
[353] Jang D M, Heo J, Yi I S and Chung I S 2002 Jpn. J. Appl. Phys. Part 1 - Regul. Pap. Short Notes
Rev. Pap. 41 6739-42
[354] Abplanalp M, Fousek J and Gunter P 2001 Phys. Rev. Lett. 86 5799-802
[355] Buhlmann S, Colla E and Muralt P 2005 Phys. Rev. B 72 214120
[356] Kim Y, Buhlmann S, Hong S, Kim S H and No K 2007 Appl. Phys. Lett. 90 072910
[357] Brugere A, Gidon S and Gautier B 2011 J. Appl. Phys. 110 024102
[358] Li Q A, Liu Y, Schiemer J, Smith P, Li Z R, Withers R L and Xu Z 2011 Appl. Phys. Lett. 98
092908
[359] Kan Y, Bo H F, Lu X M, Cai W, Liu Y F and Zhu J S 2008 Appl. Phys. Lett. 92 172910
[360] Lilienblum M and Soergel E 2011 J. Appl. Phys. 110 052012
[361] Paruch P, Giamarchi T, Tybell T and Triscone J M 2006 J. Appl. Phys. 100 051608
[362] Rodriguez B J, Nemanich R J, Kingon A, Gruverman A, Kalinin S V, Terabe K, Liu X Y and
Kitamura K 2005 Appl. Phys. Lett. 86 012906
[363] Kholkin A L, Bdikin I K, Shvartsman V V and Pertsev N A 2007 Nanotechnology 18 095502
[364] Tong S, Jung I W, Choi Y Y, Hong S and Roelofs A 2016 ACS Nano 10 2568-74
[365] Choi Y Y, Tong S, Ducharme S, Roelofs A and Hong S 2016 Sci. Rep. 6 25087
[366] Hong S, Tong S, Park W I, Hiranaga Y, Cho Y S and Roelofs A 2014 Natl. Acad. Sci. USA. 111
6566-9
[367] Tong S, Park W I, Choi Y Y, Stan L, Hong S and Roelofs A 2015 Phys. Rev. Appl. 3 014003
[368] Lee K Y, Kim S K, Lee J H, Seol D, Gupta M K, Kim Y and Kim S W 2016 Adv. Funct. Mater.
26 3067-73
[369] Ievlev A V, Morozovska A N, Eliseev E A, Shur V Y and Kalinin S V 2014 Nat. Commun. 5
4545
[370] Shen J, Zeng H Z, Wang Z H, Lu S B, Huang H D and Liu J S 2006 Appl. Surf. Sci. 252 8018-21
[371] Gleick J 2008 Chaos: Making a New Science (Penguin Books)
[372] Strogatz S H 2001 Nonlinear Dynamics And Chaos: With Applications To Physics, Biology,
Chemistry, And Engineering (Westview Press)
[373] Agronin A, Molotskii M, Rosenwaks Y, Rosenman G, Rodriguez B J, Kingon A I and
Gruverman A 2006 J. Appl. Phys. 99 104102
[374] Morozovska A N, Svechnikov S V, Eliseev E A and Kalinin S V 2007 Phys. Rev. B 76 054123
[375] Morozovska A N, Eliseev E A and Kalinin S V 2007 J. Appl. Phys. 102 014109
[376] Morozovska A N, Ievlev A V, Obukhovskii V V, Fomichov Y, Varenyk O V, Shur V Y, Kalinin
S V and Eliseev E A 2016 Phys. Rev. B 93 165439
[377] Alikin D O, Ievlev A V, Turygin A P, Lobov A I, Kalinin S V and Shur V Y 2015 Appl. Phys.
Lett. 106 182902
[378] Altorfer H B a F 1994 Acta Cryst. B50 405-14
[379] S.C. Abrahams J M R, J.L. Bernstein 1966 J. Phys. Chem. Solids 27 997-1012
[380] Yōichi Shiozaki T M 1963 J. Phys. Chem. Solids 24 1057-61
[381] Ishibashi Y 1979 J. Phys. Soc. Jpn. 46 1254-7
[382] Molotskii M, Agronin A, Urenski P, Shvebelman M, Rosenman G and Rosenwaks Y 2003 Phys.
Rev. Lett. 90 107601
[383] Kim Y, Vrejoiu I, Hesse D and Alexe M 2010 Appl. Phys. Lett. 96 202902
[384] Garrity K, Kolpak A M, Ismail-Beigi S and Altman E I 2010 Adv. Mater. 22 2969-73
[385] Zhao M H, Bonnell D A and Vohs J M 2008 Surf. Sci. 602 2849-55
[386] Zhao M H, Bonnell D A and Vohs J M 2009 Surf. Sci. 603 284-90
[387] Sturman V I B a B I 1980 Sov. Phys. Usp. 23 199-223
[388] ForsberghJr P W 1956 Encyclopedia of Physics 4 264-392
[389] Bolk T R, Grekov A A, Kosonogo.Na, Rodin A I and Fridkin V M 1971 Sov. Phys, Cryst. 16 198

135
[390] Fridkin V M, Nitsche R, Korchagina N, Kosonogov N A, Magomadov R, Rodin A I and
Verkhovskaya K A 1979 Phys. Status Solidi A-Appl. Res. 54 231-7
[391] Warren W L and Dimos D 1994 Appl. Phys. Lett. 64 866-8
[392] Kholkin A L, Iakovlev S O and Baptista J L 2001 Appl. Phys. Lett. 79 2055-7
[393] Mamin R F 1997 J. Exp. Theor. Phys. 84 808-13
[394] Gruverman A, Rodriguez B J, Nemanich R J and Kingon A I 2002 J. Appl. Phys. 92 2734-9
[395] Shao R, Nikiforov M P and Bonnell D A 2006 Appl. Phys. Lett. 89 112904
[396] Yang S Y, Seidel J, Byrnes S J, Shafer P, Yang C H, Rossell M D, Yu P, Chu Y H, Scott J F,
Ager J W, Martin L W and Ramesh R 2010 Nat. Nanotech. 5 143-7
[397] Choi T, Lee S, Choi Y J, Kiryukhin V and Cheong S W 2009 Science 324 63-6
[398] Alexe M and Hesse D 2011 Nat. Commun. 2 256
[399] Alexe M 2012 Nano Lett. 12 2193-8
[400] Giocondi J L and Rohrer G S 2001 J. Phys. Chem. B 105 8275-7
[401] Giocondi J L and Rohrer G S 2001 Chem. Mater. 13 241-2
[402] Hanson J N, Rodriguez B J, Nemanich R J and Gruverman A 2006 Nanotechnology 17 4946-9
[403] Liu X Y, Ohuchi F and Kitamura K 2008 Funct. Mater. Lett. 1 177-82
[404] Dunn S, Cullen D, Abad-Garcia E, Bertoni C, Carter R, Howorth D and Whatmore R W 2004
Appl. Phys. Lett. 85 3537-9
[405] Kalinin S V, Bonnell D A, Alvarez T, Lei X, Hu Z, Ferris J H, Zhang Q and Dunn S 2002 Nano
Lett. 2 589-93
[406] Haussmann A, Milde P, Erler C and Eng L M 2009 Nano Lett. 9 763-8
[407] Liu X Y, Ohuchi F and Kitamura K 2008 Funct. Mater. Lett. 1 177-82
[408] Seidel J, Maksymovych P, Batra Y, Katan A, Yang S Y, He Q, Baddorf A P, Kalinin S V, Yang
C H, Yang J C, Chu Y H, Salje E K H, Wormeester H, Salmeron M and Ramesh R 2010 Phys.
Rev. Lett. 105 197603
[409] Wu W D, Guest J R, Horibe Y, Park S, Choi T, Cheong S W and Bode M 2010 Phys. Rev. Lett.
104 217601
[410] Eliseev E A, Morozovska A N, Gu Y J, Borisevich A Y, Chen L Q, Gopalan V and Kalinin S V
2012 Phys. Rev. B 86 085416
[411] Jo J Y, Chen P, Sichel R J, Baek S H, Smith R T, Balke N, Kalinin S V, Holt M V, Maser J,
Evans-Lutterodt K, Eom C B and Evans P G 2011 Nano Lett. 11 3080-4
[412] Kalinin S.V. S B G, Archibald R.K. 2015 Nat. Mater. 14 973