Download as pdf or txt
Download as pdf or txt
You are on page 1of 31

Available online at www.sciencedirect.

com

surface science
reports
Surface Science Reports 71 (2016) 131
www.elsevier.com/locate/surfrep

Ferroelectric polarization effect on surface chemistry and photo-catalytic


activity: A review
M.A. Khan, M.A. Nadeem, H. Idrissn
SABIC-Corporate Research and Innovation (CRI), KAUST, Thuwal 23955, Saudi Arabia
Received 12 August 2015; received in revised form 1 November 2015; accepted 11 November 2015

Abstract
The current efciency of various photocatalytic processes is limited by the recombination of photogenerated electronhole pairs in the
photocatalyst as well as the back-reaction of intermediate species. This review concentrates on the use of ferroelectric polarization to mitigate
electronhole recombination and back-reactions and therefore improve photochemical reactivity. Ferroelectric materials are considered as wide
band gap polarizable semiconductors. Depending on the surface polarization, different regions of the surface experience different extents of band
bending and promote different carriers to move to spatially different locations. This can lead to some interesting interactions at the surface such as
spatially selective adsorption and surface redox reactions. This introductory review covers the fundamental properties of ferroelectric materials,
effect of an internal electric eld/polarization on charge carrier separation, effect of the polarization on the surface photochemistry and reviews
the work done on the use of these ferroelectric materials for photocatalytic applications such as dye degradation and water splitting. The
manipulation of photogenerated charge carriers through an internal electric eld/surface polarization is a promising strategy for the design of
improved photocatalysts.
& 2016 Published by Elsevier B.V.

Keywords: Photo-catalysis; Hydrogen production; Ferroelectric; Water splitting; Adsorption; Recombination

Contents
1.
2.
3.
4.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Ferroelectric materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Ferroelectric anomalous photovoltaic effect and use in photovoltaic devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Ferroelectric materials as catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4.1. Surface chemistry of ferroelectric materials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4.1.1. Effect of ferroelectric polarization on adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4.1.2. Spatially selective oxidation and reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.2. Photocatalytic activity using ferroelectric materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2.1.Dye degradation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2.2.Water splitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5. Conclusions and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
n

Corresponding author.
E-mail address: IdrissH@SABIC.com (H. Idriss).

http://dx.doi.org/10.1016/j.surfrep.2016.01.001
0167-5729/& 2016 Published by Elsevier B.V.

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

1. Introduction
Various semiconductors such as TiO2, CdS, ZnO and WO3
have been explored for photocatalytic applications such as
hydrogen production from water splitting and dye degradation
for environmental remediation. The current efciency of these
photocatalytic processes is below what is needed for commercialization. Factors limiting the efciency include: (i) incomplete absorption of available sunlight, (ii) poor catalyst
stability, (iii) fast electronhole recombination and (iv) slow
surface redox reactions [1,2]. In general, charge carrier
separation remains the most complex and critical issue.
Understanding and addressing the electron hole recombination
issue is critical to the success of photocatalysis. Surface redox
reactions are the rate limiting step, therefore the charge

separation, diffusion and redox reactions must proceed within


the lifetimes of photo-excited carriers [35].
One concept that can be used to mitigate losses from
recombination and back reactions is to use electric elds that
may help to separate photogenerated carriers. The best-known
examples of electric elds in photo-catalysis are associated
with interfaces, such as solution/photo-catalyst interfaces,
metalphotocatalyst junctions, and pn junctions in heterojunction photocatalysts (Fig. 1) [6]. Electric elds associated
with metalphotocatalyst Schottky junctions have been widely
used to improve their activity. Photocatalysts modied with
noble metal nanoparticles such as Pt, Au, Ag, Pd etc. facilitate
the charge separation due to the electric eld at the junction/
interface formed between the co-catalyst and the semiconductor [13,7]. Similarly electric elds at pn junctions with the

Fig. 1. The internal eld enhanced photogenerated charge carrier separation: (a) ferroelectric polarization; (b) pn junctions; (c) polar surfaces; and (d) polymorph
junctions. (PC: photocatalyst; SC: semiconductor) [6].

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

use of appropriate co-catalysts such as RuOx, IrOx, CoOx,


CuOx etc. can be used for effective charge carrier separation
[3,6]. One particularly interesting result has been obtained for
non-ferroelectric SrTiO3/Si(100) where the interface to the Si
creates a polar but non-switchable distortion in the SrTiO3 that
promotes electron transfer from the Si to the SrTiO3 surface
where it can catalyze reduction reactions [8]. Polymorph
junctions such as those associated with anatase/rutile TiO2
act in a similar way. Electric elds can also arise internally in
polar semiconductors. Polar semiconductors such as pure ZnO
is known to be polar along the c axis and the surface chemistry
and resulting photocatalytic properties have been widely
studied [9,10]. Nonetheless, this electronic polarizability is
not permanent and the polarization switching is not observed
until its melting point (1975 1C) because of large activation
energy accompanied by dipole switching process [11]. Permanent and controllable internal electric elds arise only in
ferroelectric materials because of internal polar domains.
The use of ferroelectric semiconductors provide a feasible
strategy to improve photocatalytic activity by imposing a
permanent internal polarization, which can effectively separate
photoexcited carriers. Ferroelectric materials demonstrate some
unique photochemical properties related to the internal dipole
of the material arising from the non-centro-symmetric nature of
the crystal structure. There is a growing body of work which
demonstrates that materials with an internal electric eld and
spontaneous polarization behave in anomalous ways when
photoexcited [6,1216]. The polarization of ferroelectrics
affects the electrical behavior of photo-induced charge carriers.
For example, in a BiFeO3 lm, large photocurrent ow arises
when light is applied to illuminate the positively charged ( )
surface. This photocurrent is caused by polarization-induced
surface charge, migration of oxygen vacancies to the ( )
surface, inducing a heavily doped n layer [17]. This reduces
the Schottky barrier at AuBiFeO3 interface leading to the
large photocurrents observed [17].
This review will discuss internal electric elds arising from
ferroelectric semiconductors and their inuence on photochemical reactions on surfaces. We will focus on the use of
ferroelectric materials as new candidates for improved photocatalysis. This review will start with a brief introduction to the
fundamentals of ferroelectric materials and their anomalous
photovoltaic effects followed by discussion about the recent
studies on the effect of ferroelectric polarization on surface
photochemistry specically on adsorption and spatially selective oxidation and reduction reactions. Finally, we will review
recent studies on the photocatalytic activity of ferroelectric
materials specically for dye degradation and water splitting.
2. Ferroelectric materials
Ferroelectricity has attracted much attention in the last 100
years and is one of the important topics in modern physics. In
1880, Pierre and Paul-Jacques Curie rst demonstrated the
existence of piezoelectricity in various crystals like quartz,
tourmaline and Rochelle salt (KNaC4H4O6  4H2O) [1820].
Ferroelectricity was discovered by Joseph Valasek during his

systematic study of the dielectric properties of Rochelle salt


and he presented this work at the annual meeting of the
American Physical Society in Washington on 23rd April 1920
[19,21]. The major breakthrough in ferroelectric research
happened in the early 1940s, during World War II, with the
discovery of ferroelectricity in barium titanate (BaTiO3) and
other perovskite-based materials. Since then, ferroelectrics and
their physics have been studied systematically through experimental physics, computer modeling, material synthesis and
wet-chemical synthesis.
Ferroelectric materials possess a polar unit cell and have a
spontaneous electric polarization which is reversible upon the
application of an external electric eld (EF) [17,19,22,23]. This
spontaneous polarization is permanent and is caused by the
atomic re-arrangement of ions in the crystal structure. This can
either depend on the position of the ions (conventional ferroelectrics) or on the charge ordering of multiple valences (electronic ferroelectrics) [24]. This polarization is stable under a wide
range of thermal and chemical conditions and can range between
0.1 and 78 C cm  2 depending on the ferroelectric material and
its structure. This polarization induces a surface charge given by
the equation; s P  n, where s is surface charge, P is polarization vector and n is unit vector normal to the surface [14]. The
polarization in a ferroelectric material is also associated with a
depolarization eld, which acts to neutralize the polarization by
the ow of free charge and defects within the crystal or the
absorption of charged molecules/ions on the surface of the crystal
from the surrounding medium. These are known as the internal
and external screening mechanisms respectively and are shown in
Fig. 3 and have been experimentally observed for many ferroelectric materials [14,25,26].
The regions of a ferroelectric crystal with uniformly oriented
polarization are called ferroelectric domains and the plane
separating two domains is called a domain wall. Domains are
formed in ferroelectric materials in order to minimize the
depolarizing elds within the material itself. These different
domains have different polarization directions and result in a
nearly complete compensation of the polarization of the crystal.
The depolarizing elds can also be compensated by free charge at
the boundaries. Both these mechanisms were illustrated by Rappe
and coworkers who observed the formation of striped domain
patterns on PbTiO3 when grown on a non-conductive SrTiO3
substrate [28]. At the same time, the PbTiO3 lm formed single
or mono-domains on electrically conductive substrates (SrRuO3)
that can supply compensating charges. Switching between the
two ferroelectric remnant polarization states requires the growth
and shrinking of domains. As illustrated in Fig. 4, the surface at
which the polarization produces a positive potential is termed
c domain and the surface where the polarization produces a
negative potential is termed the c domain. The domains in
7x or 7y direction, i.e. lies on the surface plane are lateral a
domains. The boundaries separating domains in x/y direction with
c /c domains are called 901 boundaries, and those separating
c and c domains are 1801 boundaries.
Ferroelectric materials can be considered as wide band gap
semiconductors with a band gap between 2.9 and 3.8 eV
[6,12,16,26,29]. Currently more than 700 ferroelectrics have

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

Fig. 2. Cubic paraelectric phase of BaTiO3 (a) Tetragonal ferroelectric phase of BaTiO3 with dipole/ polarization directed up (b) or down (c) [22].

Fig. 3. A schematic diagram of external and internal screening mechanisms


through adsorbed charges and free carriers/defects in ferroelectric materials [27].

Fig. 4. c , c  and a domains in a ferroelectric crystal [14].

been found, most of them are not hydrogen bounded or oxides


[30]. Most commonly studied ferroelectrics are oxides such as
barium titanate (BaTiO3), lead titanate (PbTiO3) and lead
zirconate titanate (PbZrxTi(1 x)O3) which show robust ferroelectricity for practical applications. Within the non-oxide family,
some chalcogenides such as zinc cadmium telluride (ZnCdTe)
and chalcohalides such as antimony sulfoiodide (SbSI) based
materials show promising ferroelectric characteristics.
One of the most widely studied ferroelectric materials is
BaTiO3. It belongs to a class of perovskite oxides with the
general formula ABO3. A and B each representing a cation
element or mixture of two or more elements. In the ideal case

the high-symmetry structure has space group Pm3m with a


simple cubic lattice and a basis of 5 atoms. B is at the center of
an octahedron with 6 oxygens as nearest neighbors. These
octahedra are packed in a simple-cubic network, where the
large holes are lled by A atoms such that each A atom has 12
nearest neighbors. Structural frustration of the cubic structure
due to mist of B/A in the holes leads to a small polar
distortion. For example BaTiO3 transforms to a ferroelectric
tetragonal phase upon temperature lowering and even further
to an orthorhombic and nally rhomobohedral ferroelectric
state [31]. Each ferroelectric structure forms a subgroup of the
cubic paraelectric structure. The rst transition to the tetragonal state (P4mm) is depicted in Fig. 2 [31]. This state occurs
below 393 K (120 1C) and is stable until 278 K. It has a
polarization of 26 C/cm2 [22]. The spontaneous polarization
in BaTiO3 lies along the c-axis and each transition in BaTiO3
is accompanied by small atomic displacement dominated by
the displacement of the Ti ion relative to the oxygen
octahedron network and a macroscopic strain.
Ferroelectric materials have also found widespread application in multilayer ceramic capacitors, gate dielectrics, waveguide modulators, IR detectors, and holographic memory [32
34]. The innite potential for applications originates from their
unique semiconducting properties and the strong inversion
symmetry breaking with a spontaneous electric polarization.
The spontaneous polarization of materials could provide new
possibilities to design photovoltaic devices by promoting the
separation of photo-excited carriers to a desirable extent.
Therefore, ferroelectric materials have recently been the focus
for utilization in photovoltaic devices and photocatalytic
applications.
3. Ferroelectric anomalous photovoltaic effect and use in
photovoltaic devices
In traditional photovoltaic (PV) effect the electric eld at the
pn junction is responsible for pulling the charge carriers away
and the resulting photocurrent from the device. The magnitude
of open circuit voltage (Voc) is determined by the quasi-Fermi
energy difference of photogenerated electrons and holes which

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

Fig. 5. The working principle of (a) pn junction solar cells and (b) FE-PV devices.(Efp Fermi level of p-type semiconductor and Efn Fermi level of n-type
semiconductor) [16]. (c) Schematic of four domains (three domain walls) in an order array of 711 domain walls and Corresponding band diagram showing the
valence band (VB) and conduction band (CB) across these domains and domain walls in the dark. Section (i) illustrates a photon hitting in the bulk of a domain and
section (ii) a photon hitting at a domain wall. (d) Evolution of band structure upon illumination of the domain wall array [38].

is limited by the bandgap of the semiconductors (Fig. 5(a)). In


the past, anomalous photovoltaic effects were seen in polycrystalline materials, where each microscopic grain can act as a
photovoltaic. Then the grains add in series, so that the overall
open-circuit voltage across the sample is large, potentially
much larger than the bandgap. This mechanism inevitably
suffers from the granular interface being poorly controlled. The
anomalous ferroelectric photovoltaic (FE-PV) effect was discovered about half a century ago in a variety of semiconducting ferroelectric materials [1316]. The FE-PV effect is
distinctly different from typical semiconductor pn junctions
such that the ferroelectric polarization is the driving force for
the photocurrent. This is illustrated in Fig. 5(b) [16]. The
ferroelectric domains acts like a photovoltaic and each domain
wall acts like a contact connecting the adjacent photovoltaics.
Again, domains add in series, so that the overall open-circuit
voltage is large. By switching the direction of polarization, it is
possible to change the photocurrent direction in a FE-PV
device [3537].
Yang and coworkers proposed a model for the FE-PV effect
described above using studies on ferroelectric BiFeO3 [38].
Fig. 5(c) is a schematic of the model domain structure showing
a series of 4 domains and 3 domain walls with thickness of
 150 nm and 2 nm respectively. The corresponding valence
(VB) and conduction (CB) position in dark conditions are also
shown. When light shines on BFO, photoexcited carriers in the
bulk are expected to quickly recombine (Fig. 5(c-i)), resulting
in no net photo effect. If the light hits the domain wall (Fig. 5
(c-ii)), the signicantly higher local electric eld enables a
more efcient separation of the excitons, creating a net
imbalance in charge carriers near the domain walls and
resulting in the band diagram shown in Fig. 5(d). This effect

means that, under illumination, a net voltage is observed across


the entire sample, resulting from the combined effect of the
domain walls. This is a similar picture to a classic pn
junction. The key difference is the magnitude of the electric
eld that drives charge separation. In a classic silicon-based
system (VOC  0.7 V; depletion layer thickness,  1 mm), an
effective electric eld of  7 kV/cm is obtained (compared
with the BFO system, with a eld of  50 kV/cm) for each
domain wall. This unique working mechanism on using
ferroelectric materials provides another viable route to convert
light into electric energy and to drive reactions for photocatalysis. It is important to note that the FE-PV effect
described above will not occur in materials which have 1801
domain boundaries, as the polarization in the two neighboring
domains will cancel each other, resulting in zero electric eld
in the domain walls.
It is also important to differentiate the bulk photovoltaic
effect seen in single crystals of ferroelectric materials with a
non-centrosymmetric structure and can result in a giant
photovoltage. This is specically called the bulk photovoltaic
effect because it occurs due to the non-centrosymmetry and not
due to the ferroelectric domains. The photovoltage can be as
large as 104 volts in some cases, e.g. in LiNbO3 bulk crystals
[16]. This effect has been observed and studied in many
ferroelectric materials such as lithium niobate (LiNbO3)
[39,40], barium titanate (BaTiO3) [41], lead zirconium titanate
(Pb(ZrTi)O3) [42] and bismuth ferrite (BiFeO3) [36,38].
To achieve a high power output a solar cell needs to show
high photovoltage, high photocurrent, and of course high
quantum efciency. Unfortunately, in ferroelectric photovoltaics, quantum efciencies remain at best on the order of 1%.
This is mainly due to very small output photocurrent densities

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

in the order of  nAcm  2 [36,43,44]. Thus ferroelectric


photovoltaics have remained an academic curiosity rather than
having any practical application. This view has recently
changed following reports that the low conversion efciencies
can be overcome by large above-bandgap photovoltages [38],
possibility of tip-enhanced photovoltaic effects at the nanoscale [45], better engineered ferroelectric materials [38,44], new
photocurrent extraction techniques [43], and particularly the
hybridization of ferroelectric materials with traditional pn
junction photovoltaics which have yielded comparable, or
superior, device performances to regular pn junction devices
[15,46].
Recently Huang and coworkers[15] demonstrated a hybrid
device using ferroelectric materials and a traditional bulk
heterojunction photovoltaic cell. The device consisted of a
thin interfacial layer of ferroelectric poly(vinylideneuorideco-triuoroethylene) [P(VDF-TrFE)] on an active layer of poly
(3-hexylthiophene):phenyl-C61-butyric acid methyl ester
(P3HT:PCBM). The ferroelectric layer was deposited by the
Langmuir Blodgett (LB) method followed by a thermal
annealing process at 135 1C to convert it into the ferroelectric
phase. The interfacial ferroelectric layer imposes a large
permanent internal electric eld on the photoinduced charge
carriers in the active layer signicantly improving the device
efciency from 1% to 2% without the ferroelectric layer to 4
5% after poling the ferroelectric layer. These enhanced
efciencies are 10 to 20% higher than those achieved by other
methods, such as morphology and electrode work-function
optimization. After poling the ferroelectric layer, one type of
charge carrier is partially or completely compensated by the
metal electrode while the other charge carriers in the ferroelectric layer generates an electric eld penetrating into the
semiconducting active layer.
As shown in Fig. 6(a) and (b), applying a positive voltage to
the Al electrode poles the ferroelectric layer. These dipoles get
aligned with positive ferroelectric polarization charges at the
interface with P3HT:PCBM layer and negative polarization
charges at interface with Al electrode. The negative polarization charges are neutralized by the Al electrode due to the large
density of free charges in metal. These positive polarization
charges cannot be compensated by the low-concentration of
free charges in the P3HT:PCBM layer, which leaves a large
uncompensated internal eld in the semiconductor. This
additional internal eld facilitates the dissociation of the bound
electronhole pairs and charge collection leading to increased
short circuit current (Jsc), ll factor (FF) and Voc as seen in Fig.
6(b) [15]. The power conversion efciencies (PCEs) of poled
FE-OPV devices were about twice larger compared to the
devices without ferroelectric layers for many types of active
layers tested. From the enhanced photocurrent in the FE-OPV
devices, the additional electric eld induced by the ferroelectric layers was estimated to be about 12 Vmm  1, which is
much larger than the built-in electric eld ( 4 V mm  1)
caused by the work function difference between typical
electrodes used in solar cells.
Chaudhary and coworkers [47] systematically studied
hybrid FE-OPV devices where the ferroelectric materials was

Fig. 6. (a) Schematics of FE-OPV and working principle with the ferroelectric
polymer at the interface; (b) Photocurrents of P3HT:PC70BM devices without
an FE layer (magenta line) and with an FE layer before poling (black squares),
and with an FE layer after positive poling (red squares) [15]. (For interpretation
of the references to color in this gure legend, the reader is referred to the web
version of this article.).

used at the interface and in the bulk of active layer by blending


with the p and n type photoactive semiconductors. They
demonstrated that by incorporating ferroelectric dipoles as
additives in OPV active layers leads to localized enhancements
of electric eld which in turn leads to more effective exciton
dissociation. By mixing a small amount of P(VDF-TrFE)
polymer into the bulk P3HT:PCBM lms, power conversion
efciencies increase by nearly 50%, and internal quantum
efciencies approach 100% indicating complete exciton dissociation at certain photon energies. According to a classic
dipole-electric eld model: E 4sf/0FE, where s is the
surface charge density and f is the volume fraction of the
dipoles, the electric eld generated by the P(VDF-TrFE) was
estimated to be as large as 240 Vmm  1 when the volume
fraction of P(VDF-TrFE) was 3%. Improved dissociation of
excitons lead to signicant improvement in device performance. Upon using 10% P(VDF-TrFE) concentration Jsc
improved from 9.6 to 11.3 mA cm  2, Voc improved from
0.55 to 0.57 V, FF improved from 48 to 60 and overall PCE
improved from 2.5% to 3.9% [47]. The higher exciton
dissociation rate in the device with the blended ferroelectric
polymer was supported by a shorter photoluminescence lifetime in the P(VDF-TrFE) mixed P3HT:PCBM lm (73 ps)
than that in the control lm without the ferroelectric polymer

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

(100 ps) implying an increased exciton dissociation rate caused


by the local electrical eld [47]. A similar effect has also been
observed for ferroelectric eld effect transistors (FeFETs)
where the polarization of ferroelectric gate layer can lead to
effective exciton separation in the semiconductor layer leading
to high mobility and on/off ratios [32,48]. The ferroelectric
polarization results in very high semiconductor surface charge
density of up to 26 mCm  2, which is of a magnitude that is
very difcult to obtain with conventional FETs because they
are limited by dielectric breakdown of the gate insulator [49].
4. Ferroelectric materials as catalysts
Since the 80s and 90s Ferroelectrics, were also known to be
wide band gap semiconductors that produce photogenerated
charge carriers under band gap illumination. This enables them
to drive photocatalytic reactions. In 1976, the photoassisted
electrolysis of water using ferroelectric BaTiO3 was reported in
a paper by Nasby and Quinn, who showed that BaTiO3 has
excellent photostability and good photocurrents [50]. The
ferroelectric nature of BaTiO3 lead to some anomalous
behavior of the photoinduced charge carriers and the resulting
surface photochemistry. Since then various ferroelectrics such
as LiNbO3, BiFeO3 and PbTiO3 have been investigated for
photocatalytic applications [6,13,14]. The semiconducting
properties of ferroelectric materials and the presence of an
internal electric eld may make them an ideal candidate as
photocatalysts for various applications. Despite their potential,
there are limited studies to understand the surface photochemistry and photocatalytic performance of ferroelectrics materials.
In the next few sections we will review the work done with
ferroelectric photocatalysts to understand the effect of polarization/internal electric eld on adsorption, charge carrier
separation, redox reactions and on their photocatalytic activity.
4.1. Surface chemistry of ferroelectric materials
4.1.1. Effect of ferroelectric polarization on adsorption
There is growing interest to study the effect of ferroelectric
polarization on the adsorption (physisorption and chemisorption) properties [6,1214,51,52]. It has been recognized for
some time that the orientation of the dipoles may inuence
adsorption and reaction on ferroelectric surfaces and electronexchange reactions at liquidsolid interfaces are known to be
domain orientation (c /c  ) specic. The ferroelectric polarization and screening affects the adsorption on the surfaces of
ferroelectric semiconductors. To date, polarization dependent
adsorption studies have primarily been carried out on BaTiO3
and LiNbO3 surfaces.
Vohs and coworkers have conducted some detailed studies of
the adsorption of low molecular weight alcohols such as ethanol
and methanol on the surface of the ferroelectric BaTiO3. In one of
their work in 2008 [53], they observed domain dependent sticking
coefcients using temperature programmed desorption (TPD) and
scanning surface potential microscopy (SSPM). The authors
performed TPD studies to probe surfaceadsorbate interactions
for the BaTiO3 and CH3OH [53]. Ferroelectric domains were

Fig. 7. TPD spectra obtained following exposure of BaTiO3 (001) thin lm to


a 20 L dose of CH3OH as a function of ferroelectric polarization [53].

oriented in situ using planar electrodes or a conductive atomic


force microscope (AFM) tip under ultrahigh vacuum (UHV)
conditions. During TPD experiment methanol desorbed in two
temperature domains; low temperature peak at 365 K and a hightemperature peak at 580 K as shown in Fig. 7. The saturation
coverage of chemisorbed species was low and estimated to be
o0.20 mL. This desorption behavior is similar to that observed
for CH3OH on TiO2 and other oxide surfaces [54,55].
In earlier works, Kim et. al.[54] and Henderson et. al.[55]
showed that methanol is adsorbed on TiO2 both molecularly and
dissociatively at 200 K, and only dissociatively at 300 K. During
dissociative adsorption methoxide is adsorbed on unsaturated Ti
cations while H atom on adjacent oxygen anions. The molecularly adsorbed methanol desorbs readily below 300 K. The
surface methoxide species resulting from dissociative adsorption
are removed via two different channels: recombinative desorption
of methanol at 365 K and decomposition at higher temperatures
with some amount of re-combinative desorption of methanol. The
TPD spectra in Fig. 7 has the same shape and desorption
temperatures as a function of poling but a difference in the
intensity of the desorbing species was seen. The same trend was
observed over several experiments to show reproducibility. This
indicates that while the energy of CH3OH chemisorption is
independent of the polarization the amount of CH3OH adsorbed
changes. The amount of CH3OH adsorbed calculated from the
TPD integrated area increases in the following order:
c ounpoledoc . This indicated that the reactive sticking
coefcient S (the fraction of the impinging molecules that
ultimately react to form a chemisorbed species) and the physisorption energy depend on the substrate polarization.
This was further investigated by using AFM and SSPM
studies by studying adsorption of CO2 on the BaTiO3 surface.
Fig. 8(c) shows the surface potential changes as a function of
dosing of CO2. The average surface potentials of both the c
and c domains initially decreases linearly with dose and then
slowly as saturation coverage is approached indicating saturation coverage of CO2 on both c and c domains is nearly
equal. By plotting the coverage as a function of dosing (L), it

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

Fig. 8. Inuence of CO2 adsorption on the surface potential of BaTiO3 (001). (a) Surface potential maps of c and c domains (700 nm  700 nm) on BaTiO3
(001), which were poled by an external electrical eld 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 BaTiO3 (001).
(d) (L)/max, versus CO2 dose. S is proportional to the slope of the line [53].

is possible to quantify the effect of polarization on the reactive


sticking coefcient S using Eq. (1) below:[53]
L
SL
p

max
max N 0 2mk b T

where is coverage (fraction of sites per area N0 that are


occupied) and L is the dosing in Langmuir units. The coverage
is the fraction of sites per area, N0, that are occupied,
N N0. Fig. 8(d) shows relative coverage as function of
dosing with S proportional to the slope of the line. Slopes for
c and c differ by a factor of 3.7 indicating that the reactive
sticking coefcient is polarization dependent with c  surfaces
showing improved sticking coefcients.
In many cases for oxide surfaces, chemisorption occurs at
oxygen vacancies, step edges, or other defect sites [56]. As
illustrated in Fig. 9 the gas phase molecule rst physisorbs and
with subsequent diffusion it reaches and chemisorbes at an O
vacancy, step edges, or other defect sites.
The adsorption studies show that S is strongly affected by
ferroelectric polarization, but the chemisorption energy is not.
This means that the increased S is due to the Van der Waals
interactions between the adsorbate and ferroelectric surface which
reduces the activation barrier for physisorption. Impinging gasphase molecules become trapped into the shallow physisorption
well, dominated by van der Waals interactions and dependent on

the surface polarization. Many trapped molecules spend a short


time on the surface and then desorb; however, if one encounters
an active site (for example, an oxygen vacancy), the precursor
may react to form a chemisorbed species, in which chemical
bonds are formed.
In a follow up work Vohs and coworkers also studied the
adsorption and reaction of ethanol on BaTiO3 using temperatureprogrammed desorption (TPD) [51]. A BaTiO3 thin lm was
supported on the surface of a rutile TiO2 (110) single crystal. The
ferroelectric dipoles in the BaTiO3 thin lm were oriented in
UHV by bringing the front surface of the grounded BaTiO3/
TiO2(110) sample at 300 K into contact with a polished Cu
electrode that was biased at 732 V relative to the sample. A
SSPM image of a 20 mm  20 mm region of the as-grown,
vacuum-annealed BaTiO3 lm is displayed in Fig. 10(a) where
the average surface potential varies by less than 0.2 V throughout
the image. This is consistent with a random orientation of
ferroelectric domains in the near surface region of the lm. After
poling (Fig. 10(c)), regions with positive and net negative polarity
are clearly evident and the average surface potential between the
positively and negatively poled regions is  2 V which is an
order of magnitude greater than the surface potential variation
observed for the as-grown, unpoled sample.
Fig. 11 shows the TPD spectra for ethanol adsorbed on the
BaTiO3 surface [51]. Ethanol desorbs in two features: a low-

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

temperature peak centered at 350 K with a broad tail that


extends to higher temperatures (Fig. 11(a)). Superimposed on
this tail is a much smaller, second ethanol peak centered at
 500 K. During TPD, ethylene and acetaldehyde products
were produced appearing as overlapping peaks between 325
and 550 K, with the largest peak for both centered at 490 K.
For TiO2, the primary product between 300 and 400 K is
C2H5OH which has been attributed to reaction between
adsorbed ethoxides and hydroxyl groups. At temperatures
between 400 and 700 K the primary products observed are
ethylene and acetaldehyde with some amount of ethanol. In
many earlier studies such as Kim and Barteau [57,58] and
Gamble et al. [59] attributed the ethylene product to hydrogen elimination from adsorbed ethoxy groups followed

Fig. 9. Precursor-mediated molecular adsorption on defective ferroelectric


oxide surfaces. The gas-phase molecule physisorbs to the oxide surface, where
it diffuses until reaching and chemisorbing at an O vacancy, or eventually
desorbing. Schematic diagram of the potential energy as a function of the
distance from the surface [53].

by CO bond scission with the eliminated hydrogen reacting


with another surface ethoxy species to form ethanol. The
aldehyde product is attributed to an -hydrogen elimination
reaction [5759]. In light of the similarity between the TPD
results for BaTiO3 and TiO2, these are the likely pathways for
reaction of ethanol on the BaTiO3 thin lm surface. TPD peak
areas at a dosing of 2.5 L ethanol were calculated as a function
of poling as shown in Fig. 11(d). The total ethanol adsorbed
was found to be higher on the c poled surface. Upon a
second TPD run, the unpoled surfaces gave very similar results
to the rst run. Based on the precursor-mediated adsorption
model discussed above it was suggested that ferroelectric
polarization increases the Vander Waal interaction between
the polar ethanol molecule and BaTiO3 surface [51].
Based on the model above, the molecular structure of the
adsorbate should also affect the sticking coefcient since a more
polar adsorbate will interact more strongly with the surface
polarization of a ferroelectric. To conrm this the authors used a
more polar adsorbate i.e. 2-ouroethanol to investigate the
inuence of ferroelectric polarization [60]. 2-Fluoroethanol was
found to absorb dissociatively on the BaTiO3 surface to form an
alkoxide that reacted upon heating to produce acetaldehyde,
ethylene, and adsorbed uorine atoms which remain on the
surface. The TPD peaks of desorbed acetaldehyde shift by 12 K
on a positively poled (c ) surface compared to an unpoled
surface and the shift was consistent upon repeated experiments.
This difference in activation energy between the c surface and
unpoled surface was  4 kJmol  1 and much higher than that
observed for ethanol or methanol giving an indication that a more
polar adsorbate reacts more strongly with the surface polarization
of a ferroelectric [51,53].
Dunn and co-workers investigated the adsorption and photodecolorization of Rhodamine B dye molecule using BaTiO3
(BTO) to probe the inuence of ferroelectricity on the photocatalytic properties [61]. BaTiO3 powder was annealed at
1200 1C for 10 h to convert it from cubic (paraelectric) to
tetragonal (ferroelectric) phase. Adsorption of a dye molecule
on the surface of a catalyst is an important step in the
decolorization process. The amount of dye adsorbed by the
catalyst in the dark was determined using the UV/vis absorption
of the dye solution. The adsorption results, shown in Fig. 12,
highlight that there was an increase in dye adsorption from 0.97%

Fig. 10. (a) SSPM image of unpoled BaTiO3 thin lm. (b) Poling pattern applied to surface using bias voltage of 710 V. (c) SSPM image of BaTiO3 thin lm after
applying the poling pattern in b [51].

10

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

Fig. 11. (a, b and c) TPD spectra from BaTiO3 thin lm dosed with 2.5 L of ethanol as a function of the polarization state: Red: unpoled, Blue: c and Orange: c
(d) TPD peak areas for BaTiO3 lm dosed with 2.5 L of ethanol as a function of the polarization state [51]. (For interpretation of the references to color in this gure
legend, the reader is referred to the web version of this article.)

to 4.81% per unit surface area after annealing that is after


converting to ferroelectric phase. The authors argued that the
polar RhB cation produced a tightly bound dye layer on the polar
surfaces that resulted in higher adsorption. However, it is not
clear whether the improved adsorption and activity of the
annealed BTO samples is due to the improved crystallinity and
cleanliness or the ferroelectric polarization of the tetragonal phase.
Altman and coworkers performed some very interesting
studies on surface chemistry of ferroelectric LiNbO3(0001)
surfaces [52,62]. The adsorption of 2-propanol on the positively and negatively poled LiNbO3(0001) surfaces was
studied using temperature programmed desorption (TPD). As
seen from the TPD spectra in Fig. 13, the desorption shows a
dominant peak at 480 K on the positive surface which did not
shift as function of coverage implying a rst-order desorption
of molecularly adsorbed 2-propanol with only weak interactions between adsorbed molecules [62]. In contrast, the TPD
results for the negative poled surface show a desorption peak
centered at 370 K which also did not change as function of the
coverage. Thus 100K difference in the desorption peaks of 2propanol suggested that the polar molecule adsorbs more
strongly on the positively poled surface.
Additionally the ferroelectric polarization of the LiNbO3
leads to some unusual adsorption desorption characteristics.
Fig. 14 is a comparison of the TPD spectra as a function of

Fig. 12. Adsorption of RhB by BaTiO3 under dark conditions for 30 min. Dye
removal is scaled for surface area. The adsorption increases dramatically after
annealing to ferroelectric phase [61].

heating rates (0.25-4 K/s) for the positive and negative


surfaces. The data shows that as the heating rate was increased
from 0.25 to 4 K/s, the desorption peaks shifted to higher
temperatures by 170 K for the positive surface and 219 K for
the negative surface which were unusually large shifts. This
unusual behavior was also reected in the kinetic parameters
obtained. A plot of log of /T2 vs 1/Tp is shown in Fig. 15;

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

11

Fig. 13. TPD traces for 2-propanol on positively (a) and negatively (b) poled LiNbO3 (0001) surfaces. The samples were polished on one side. Data were collected
using a heating rate of 1 K/s while monitoring the signal at m/q45 (CH3HCOH ) for 2-propanol [62].

Fig. 14. Series of 2-propanol TPD curves measured at varying heating rates on positively (a) and negatively (b) poled LiNbO3 (0001) surfaces at exposures of 9000
and 8000 L, respectively. Since the peak temperatures were independent of coverage, the different exposures do not inuence the results [62].

where is heating rate and Tp is desorption peak temperature


[62]. The slope corresponds to  E/R and the intercept is R/E;
where R is molar gas constant, is desorption pre-exponential
and E is desorption activation energy. The data shown in Fig.
15 indicates pre-exponential factors of 10 s  1 for positively
poled surface and 10  1 s  1 for negative surface which is

vastly different than typical values which are 13 orders of


magnitude greater.
This unusual behavior of the LiNbO3 surface was explained
based on the pyroelectric effect in ferroelectrics. The surface
polarization charges in ferroelectrics are usually passivated/
screened during sample preparation by, for example, formation

12

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

Fig. 15. Log-scale plot of /Tp2 vs Tp for desorption of 2-propanol on


positively (a) and negatively (b) poled LiNbO3 (0001) surfaces. Data were
obtained from the heating rates and corresponding desorption peak
temperatures (Tp) in Fig. 14. The steeper slope for the positive surface is
indicative of a higher desorption activation energy [62].

of oxygen anion vacancies on the negative side of the crystal


and oxygen ad-anions on the positive side to stabilize the
surface [63]. As the temperature is ramped up, the bulk
polarization decreases with increasing temperature. Therefore,
the compensating/passivating surface charge densities would
be in excess of those required to screen the bulk polarization as
the temperature is increased. This results in large surface
dipoles (electric eld) in the opposite direction of the poling
direction which are build up as the temperature is ramped
during the TPD runs. Therefore, the molecules might become
more strongly bound to the surface as the temperature is
ramped up quickly creating a temperature-dependent heat of
adsorption as seen in Fig. 14.
If the ferroelectric properties of LiNbO3 play a role in the
unusually strong peak temperature dependence for a polar
molecule on heating rate, then nonpolar molecules should
display a normal temperature dependence. To test this hypothesis, Altman and co-workers also directly compared the
adsorption of a polar versus nonpolar adsorbate on LiNbO3
[52]. Fig. 16 shows the effect of ferroelectric poling on the
adsorption/desorption of a nonpolar dodecane molecule [52].
The results show that dodecane adsorption/desorption from the
two surfaces was in fact very similar. In both cases, two peaks
can be seen: a peak at 253 K that saturates at  40 L and a
peak at 230 K that does not saturate. It is clear that the peak
positions for nonpolar dodecane were unaffected by ferroelectric poling. This is contrast with a polar molecule such as
2-porpanol where there was a  100 K difference in the
desorption peaks from the positive and negative surface. The

Fig. 16. Comparison of dodecane desorption from (a) positively and (b) negatively poled LiNbO3(0001) surfaces. The data were collected using a heating rate of
1 K/s while monitoring the mass spectrometer signal at m/q 43 (C3H7 ) for dodecane. A monolayer (1 ML) was dened as the dodecane coverage when the TPD
peak at 253 K saturated [52].

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

13

Fig. 17. Dodecane TPD curves for different heating rates for (a) positively and (b) negatively poled LiNbO3(0001) surfaces at 1 ML dodecane coverage [52].

effect of heating rate on dodecane desorption was also


measured as illustrated in Fig. 17. Upon changing the heating
rate the peak temperature changed only by 17 K. In contrast to
dodecane, the data points for 2-propanol differ widely on the
positive and negative surface as seen in Fig. 14. This reinforces
the conclusion indicating that experimental artifacts were not
responsible for the large changes seen for 2-propanol. This
nding links the unusually strong dependence on heating rate
for 2-propanol to electrostatic interactions between the polar
molecules and the ferroelectric surfaces. These results further
support the hypothesis that the strong inuence of the pyroelectric effect on desorption of the polar molecule is largely
driven by electrostatic interactions. This implies that the
stronger adsorption of 2-propanol on the positively poled
surface can be related to a more favorable electrostatic
interaction. Such a favorable geometry could be easily envisioned as one that places the negatively charged O atoms
closest to the positive surface.
Apart from the study of alcohols Zhang et. al. studied the
polarization effect on adsorption of D-cysteine molecules on
ferroelectric LiNbO3 surfaces [64]. The study was done by a
combination of infrared spectra-microscopy and spatially
resolved x-ray absorption near edge spectroscopy. The positive
and negative domains on the LiNbO3 surface were mapped by
Piezo force microscopy imaging (PFM). PFM images showed
that even in the absence of any electronic excitation (optical or
thermal), the adsorption is still polarization-specic with
preferential D-cysteine deposition occurring on positive
domains. The reported effect is attributed to dipole-dipole
interaction between polar D-cysteine molecules and ferroelectric polarization dipoles. These experimental results are
backed up by simulation studies. Sanna and co-workers
studied polarization dependent adsorption of single hydrogen

atoms and hydroxyl radicals (OH) on LiNbO3(0001) surface


by means of rst-principles total energy calculations [65].
Total energy density functional calculations was performed
within the PW91 formulation of the generalized gradient
approximation (GGA), implemented in the VASP simulation
package. Details of the calculations can be found in ref [65].
Fig. 18 shows the potential energy surface (PES) which
gives an approximate idea of the stable adsorption site. The
potential energy surface (PES) for the adsorption of H atom at
the negative side (Fig. 18(b)) is much more corrugated than in
the case of the positive side (Fig. 18(a)), indicating a lower
surface mobility of the adsorbate. On the positive surface, H
avoids a position right on top of the topmost Li and on the
negative surface H avoids the Li and O atoms. It prefers an
adsorption site between cations above the lower laying oxygen
atoms. OH radical is also found to adsorb at a similar site on
both the positive and at the negative (0001) surfaces as seen in
Fig. 18(c) and (d). Optimal geometry of the OH radical was
also obtained from PES calculation by allowing the molecule
and the surface to relax without any constraints. As seen in
Fig. 19(a), on the positive surface, the polar radical OH lies
relatively at with the adsorbate oxygen forming a covalent
bond with the surface oxygen at a distance of 1.49 . At the
negative face, OH adsorbs roughly perpendicular and is
located between a Li d(OLi) 1.84 and a Nb atom d(O
Li) 2.08 as plotted in Fig. 19(b).
The adsorption energy for both adsorbates is strongly
polarization dependent with differences as high as 2 eV. In
particular, the calculated adsorption energy for H atom and OH
radical is 6.0 and 2.9 eV at the positive and 3.8 and 5.7 eV at
the negative (0001) side, respectively. The adsorption energy
for H is higher at the positive surface than at the negative
surface, while for OH the adsorption at the negative side is

14

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

Fig. 18. Potential energy surface (PES) for the adsorption of a single H atom on the (a) positive LiNbO3(0001) surface (b) negative LiNbO3(0001) surface. Potential
energy surface for the adsorption of a single OH radical on the (c) positive LiNbO3(0001) surface (d) negative LiNbO3(0001) surface. Gray, white and red balls
represent Li, Nb and oxygen atoms respectively. Adsorption energies are in eV [65]. (For interpretation of the references to color in this gure legend, the reader is
referred to the web version of this article.)

favored. This striking contrast is related to electrostatic effects,


leading to different bonding scenarios. Both hydrogen atom
and OH radicals possess one unpaired electron. While OH
reaches a stable conguration by capturing one electron (OH  ,
hydroxide), hydrogen acts as a donor. Capturing an electron
from the positive surface further increases the surface charge
while subtracting an electron from the negative side stabilizes
it. This explains why the OH adsorption at the negative side is
favored with respect to the adsorption at the positive side. Vice
versa, the adsorption of hydrogen atoms has a stabilizing effect
at the positive surface.
Other interesting studies such as the one by Rappe and coworkers who investigated metal deposition on ferroelectric
surfaces [66]. Specically the effect of ferroelectric polarization direction on the geometric properties of Pd deposited on
the positive and negative surfaces of LiNbO3 (0001) was
studied [66]. The preferred geometries and diffusion properties
of small Pd clusters were studied using density functional
theory, and these calculations were used as the basis for
Kinetic Monte Carlo simulations (KMC) of Pd deposition on a
larger scale. First, the adsorption geometries of Pd clusters on

the c and c surfaces was calculated. Then using the


nudged elastic band method a range of possible diffusion and
agglomeration processes were modeled. Finally, the activation
barriers of these processes were used as inputs for a KMC
simulation of the deposition of Pd on LiNbO3 on a larger scale.
The surface charges were passivated with one Li atom per
primitive supercell on the c surface, and one O and one Li
atom on c in accordance with earlier studies showing that
this is the most stable surface conguration and referred to as
Li0 and O0 [67]. The DFT results showed that on the positive
surface, Pd atoms favor a clustered conguration, while on the
negative surface, Pd atoms are adsorbed in a more dispersed
pattern due to suppression of diffusion and agglomeration. In
contrast to its behavior on the c surface, Pd adsorption on
the c surface substantially alters the geometry of the surface
itself. This is largely due to interactions with O0 atoms, which
along with Li0 atoms terminate the c  surface for charge
passivation. The formation of PdO0 bonds makes adsorption
onto the c  surface much more favorable than on c surface.
On the c  surface, strong Pd bonding to O0 which is the extra
surface oxygen present for charge passivation, leads to larger

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

diffusion and agglomeration barriers. This leads in turn to less


clustering and a more planar geometry overall on the c
surface than on c as illustrated in Fig. 20. The difference in
adsorption geometry predicted here might be the reason for the
difference in catalytic activity that has been observed for Pd
deposited on oppositely poled LiNbO3 surfaces [68].
4.1.2. Spatially selective oxidation and reduction
Throughout the 80s and 90s, there was substantial amount of
work on studying the semiconducting properties of ferroelectrics however, there was very limited focus on surface
photochemistry. In the year 2000, Rohrer and coworkers
published a work on the surface photochemistry for BaTiO3

Fig. 19. Charge redistribution upon adsorption of the OH radical at the


positive (a) and negative (b) LiNbO3(0001) surface [65].

15

[69]. The work is related to variations in the band bending


associated with surface polarization of ferroelectrics and the
corresponding selective surface chemistry seen in ferroelectric
materials.
As discussed earlier, Fig. 21(a), shows the internal and
external screening mechanisms in a ferroelectric material. This
screening of ferroelectric polarization results in band bending
near surface region with the formation of depletion and
accumulation layers. In c domains, the screening causes
h depletion and downward band bending. On the other hand,
in c- domains, there is h accumulation at the surface and the
bands are bent upwards. This altered electronic structure in the
vicinity of the surface dictates which type of carriers are
available for surface reactions making chemical and photochemical reactivity on ferroelectrics domain specic. Upon
photoexcitation of the semiconducting ferroelectric material,

Fig. 21. Schematic of a ferroelectric material showing (a) internal polarization


and screening mechanisms and (b) the effect of free carrier reorganization on
band structure and photoexcited carriers [61].

Fig. 20. Kinetic Monte Carlo simulations (KMC) snapshots of 1.0 ML Pd deposition on (a) c and (b) c  LiNbO3 surface at 300 K with a 0.025 ML/s deposition
rate [66].

16

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

Fig. 22. (a) Topographic AFM images of BaTiO3 surface before reaction and (b) a surface potential image recorded at the same time with dark contrast areas
corresponding to c domains and lighter areas are c domains. (c) Topographic image after reaction in a silver nitrate solution for 3 s. (d) Topographic image after
reaction in a lead acetate solution for 3 min. The black-to-white contrast in the images shown in panels (a), (b), (c), and (d) are 100 nm, 175 mV, 60 nm, and 60 nm,
respectively [71].

the excited electronhole pairs are driven toward opposite


surfaces of the material by the internal electric eld arising
from the polarization leading to selective surface chemistry
[61,69,70].
This effect of the ferroelectric polarization on reactivity can
be treated mathematically using a space charge model. It is
assumed that it is principally the charge carriers generated in
space charge regions that participates in the surface photochemical reactions. Carriers generated deeper within the bulk
of the material have further to diffuse to reach the surface
where they are more likely to recombine. Therefore, anything
that increases the spacecharge width will make more photogenerated carriers available for the reaction and, therefore, has
the potential to increase the reactivity. The width of the space
charge region in depletion (Ld) or accumulation (La) can be
written as a function of the Debye length (LD) and the surface
potential (Vs);

s

2eV s
2
Ld
LD
kT
La


p
eV s
2 1  e kT LD

where, e is the electron charge, k the Boltzmann constant (in


units of eV/K), and T the absolute temperature. Thus, the width
of the space charge region depends on the Debye length or the
surface potential. The Debye length is a characteristic of the
sample and is dependent on the dielectric constant and donor
concentration according to Eq. (4):

s


0 r kT
LD
e2 N D

where 0 is the permittivity of free space, r the dielectric


constant, and ND is the donor density.
The surface potential on the other hand depends on several
factors, including the domain polarization, the solution composition, the position of the conduction band edge, and
adsorption on the solid surface. According to the Poisson
Eq. (1), the surface potential (Vs) , is linked to surface charge
() and the remnant polarization (Pr) [70]:
2 V x

1
 Pr
0 r

A change in polarization changes the surface potential and


leads to change of the width of the space charge region leading
to changes in reactivity from location to location.
Using AFM microscopy and PFM (Piezo force microscopy),
Rohrer and workers observed the spatially selective deposition
of metal nanoparticles from their respective salt solutions on
the surface of ferroelectric BaTiO3 [6,69,7175]. The authors
observed that when BaTiO3 is illuminated by UV light in the
presence of aqueous Ag and Pb2 ions, Ag is preferentially reduced on the surface of c domains and Pb2 is
preferentially oxidized on c  domains which is illustrated in
Fig. 22. The reduction of Ag (to Ag metal) by photogenerated electrons and the oxidation of Pb2 (to Pb4 ) by
photogenerated holes is described by the reactions (6) and (7):
Ag e  Ag

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

Pb2 2H2 O 2h PbO2 4H

The tendency of electrons to move toward the c domains


drives the reduction of Ag and similarly the holes, which
move towards the c domains, drive the oxidation of Pb2
ions, both are illustrated in energy level diagrams in Fig. 23. An
important point to note here is that the c domains are screened
by negatively charged counter ions that would be forming a
Stern or double layer [70,76]. Therefore, there has to be a
mechanism that is allowing the movement of photoexcited
electrons from the surface of the ferroelectric into the solution.
This is still not well understood and the mechanism is thought to
involve tunneling to allow electrons to pass through the bound
ions and reduce the metal cations [14,70,76]. However, no
deposition is observed when photo-chemically inert substrates
such as alumina and silica are substituted for the BaTiO3.
Furthermore, when photochemically active, but paraelectric,
materials such as TiO2 and SrTiO3 are used in place of BaTiO3,
deposition of both Ag and Pb are observed. However, the
deposition is not spatially selective as seen for BaTiO3 [69].
Rohrer and co-workers also studied the photochemical
reduction of Ag by solid solutions of Ba1  xSrxTiO3 where x
was varied from x 0 to 1 [71]. When x is 1 the catalyst is a
non-ferroelectric centrosymmetric perovskite oxide (SrTiO3)
[77]. As strontium is added to BaTiO3, there is a continuous
change from spatially localized to uniform reactivity that is
complete at x 4 0.27. On SrTiO3, silver reduction is spatially
uniform. The relative heights of the silver deposits as measured
by atomic force microscopy were used to quantify the relative
reactivity.
A selection of AFM images recorded after the reactions on
Ba1  xSrxTiO3 samples are shown in Fig. 24 [71]. The white
contrast in the images corresponds to deposited silver. For the
sample with x 0.2 (Fig. 24(a)), the silver is concentrated in
stripes similar to some of the features on pure BaTiO3. With
increasing amount of Sr the reactivity changes from being
selective on the positive domains to being uniform for all
locations. For example with x 0.4, the reactivity is spatially
uniform at all locations (Fig. 24(d)). This is correlated to a loss

17

in polarization of the solid solutions with increasing Sr content.


Room-temperature data reported by Ianculescu and co-workers
indicate that the polarization decreases by  1/3, from 27 to
17.5 C/cm2 as the Sr content increases from x 0 to x 0.3.
Decreasing polarization will decrease surface potential and this
will decrease the width of space charge region as discussed in
Eqs. (2)-(5) above. This leads to the spatial uniformity of the
reduction reactions for samples with x4 0.27 [71].
Following the fundamental work on BaTiO3 surfaces, the
spatial selectivity and reactivity on PZT surfaces was investigated. Kalinin and co-workers investigated the domain specic
deposition of various metals such as Au, Ag and Pd via
photoreduction on PZT and BaTiO3 surfaces [78]. Using an
AFM tip and voltage of 7 10 V, the authors demonstrated
domain poling from 100 nm to few microns in dimension. This
lead to metal deposition in various nanostructures as illustrated
in Fig. 25. Though poling by an AFM tip is useful for
understanding the fundamental surface photochemistry of
ferroelectric materials, it is not practical for large scale
applications. Kalinin and co-workers also demonstrated that
it is also possible to locally reorient polarization with an
electron beam. Primary electrons from the beam injected into
the surface interact with the solid and emit secondary electrons
(core electrons, valence electrons Auger electrons). If the ratio
of emitted electrons to incident electrons is o 1, the surface
will be negatively charged and if 41, the surface will be
positively charged. The depth of the charged region is on the
order of the escape depth of the secondary electrons,  1 to
3 nm. This localized surface charge results in a local electric
eld that causes realignment of atomic polarization. Fig. 26
shows the poling of PZT surface using an electron beam at
20 kV for 560 s. This was the rst demonstration that atomic
polarization can be oriented with an electron beam and can, in
principle, be repeated to develop structures in which multiple
types of metal particles and several electronically or optically
active molecules can be assembled in predetermined
congurations.
Recently the surface photochemistry of some new ferroelectric materials such as BiFeO3 and LiNbO3 have also been

Fig. 23. Energy-level diagrams for BaTiO3 in aqueous solution, without illumination. The energies on the vertical axes are on the standard hydrogen electrode
scale. EV is the valence band edge, EF the Fermi level, EC the conduction band edge, Vs the surface potential, Ld the width of the space charge layer in depletion, and
La the width of the space charge layer in accumulation. The Ag reduction and Pb2 oxidation reactions are shown at their standard potentials. Images show band
bending (a) for a positive domain, (b) for a neutral domain and (c) for a negative domain [71].

18

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

Fig. 24. Representative topographic AFM images of four different Ba1  xSrxTiO3 compositions after reaction with silver: (a) x 0.2, the black-to-white contrast is
95 nm; (b) x 0.26, the black-to-white contrast is 60 nm; (c) x 0.27, the black-to-white contrast is 60 nm; and (d) x 0.4, the black-to-white contrast is 110 nm.
All of the images have a 20-m eld of view [71].

studied. Rohrer and his group published investigated the


surface redox reactions on BiFeO3 for the rst time [73].
They observed that when excited by light with energy between
2.53 and 2.70 eV, BiFeO3 also photochemically reduces silver
cations from solution in patterns corresponding to the underlying ferroelectric domain structure. Silver is preferentially
reduced on domains with a positive polarization directed
toward the surface. The amount of reduced silver depends on
whether the component of the domain polarization normal to
the surface is positive or negative. Using PFM microscopy the
authors observed areas of silver deposition correspond to
domains that appear dark in the PFM image. These areas
correspond to a phase lag of 1801 consistent with positive
domains. Bright areas in the PFM image, consistent with
negative domains, correspond to areas without signicant
silver reduction. This is consistent with previous studies on
BaTiO3.
The spatial selectivity and reactivity of LiNbO3 has also
been reported [79]. Nemanich and coworkers reported the
LiNbO3 investigated the wavelength and screening dependences of a liquid-based photoinduced deposition process for
the synthesis of Ag nanostructures on a periodically poled

lithium niobate template [79]. LiNbO3 shows a behavior


different to other ferroelectric materials such as PZT, BaTiO3
and BiFeO3 where the Ag cations are reduced mainly on the
domain boundaries rather than the c domains. LiNbO3
unlike PZT or BaTiO3 has a low density of surface defects
( 1012 cm  2). Therefore, the internal screening in LiNbO3 is
weaker and the polarization charges are mainly screened
externally via the surface absorption of charged molecules.
Thus the band bending and electric eld at c /c  domain
surfaces is weaker compared to other ferroelectric surfaces.
The authors report a strong domain boundary electric eld in
the case of LiNbO3. If the light hits the domain wall, the
signicantly higher local electric eld enables a more efcient
separation of the excitons and the photo-excited electrons near
the boundaries will migrate under the inuence of the strong
boundary electric eld. Consequently, enhanced Ag deposition
is observed along the domain boundary. Nonetheless, the
origin of this strong domain boundary electric eld is unclear
as is the reason why only photo-excited electrons migrate to
the surface at the domain boundaries.
Rohrer and his group further investigated the spatial
selectivity on oxide heterostructures with ferroelectric and

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

19

Fig. 25. Surface topography (a) and piezo-response image (b) of PZT thin lm. The inset shows that the PFM contrast is not random but is due to the small ( 50 to
100 nm) ferroelectric 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 310 nm. Piezoresponse image of checkerboard domain structure fabricated using in-house lithographic system (e) and SEM image of
corresponding silver photodeposition pattern (f) [78].

Fig. 26. Piezo-response image of PZT surface exposed to an e-beam (a). 600 s exposure at 20 kV results in formation of positive polarization orientation (light
contrast in the center), which is the consequence of negative surface charge. Short time (o 5 s) exposure results in negative poling (dark contrast at top and bottom)
from a positive surface charge. Polarization switching can be conrmed by reversing the polarization orientation by applying an electric eld (b). Note that the
polarization reversal in the central square is partial, and complete polarization reversal requires higher bias (30 V) for completion. Comparing (a) and (b) conrms
the sign of the charge deposited by the e-beam [78].

non-ferroelectric materials. TiO2 lms, 15100 nm thick were


grown on ferroelectric BaTiO3 substrates and used to photochemically reduce Ag to Ag0 and oxidize Pb2 to Pb4

under ultraviolet illumination [72]. Atomic force microscopy


was used to show that the reactions are spatially selective and
that the pattern of products on the lm surface reproduces the

20

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

pattern of products on the bare substrate. AFM microscopy


showed that Pb and Ag containing deposits form in different
complementary locations: where Ag reduces on thin c
domains with wide gaps between them while Pb is oxidized
on the wide c  domains with thin gaps in between. The
results indicate that dipolar elds from the ferroelectric
domains cause carriers generated in the substrate to travel
through the lm to react on the lm surface just as they would
on the substrate surface as illustrated in Fig. 27. The inuence
of the ferroelectric substrate on the pattern of reactants is
diminished as the TiO2 lm thickness increases and a greater
fraction of the light is absorbed in the TiO2 lm. The results
indicate that for thin (15 nm) lms, dipolar elds from the
ferroelectric domains cause carriers generated in the substrate
to travel through the lm to react on the surface. Similar to the
study of TiO2BaTiO3 heterostructures, Rohrer and coworkers also studied TiO2-BiFeO3 heterostructures consisting
of thin titania lms (10 nm) on BiFeO3 substrates [74]. The
heterostructures, when excited by visible light with energies
between 2.53 and 2.70 eV, photochemically reduce aqueous
silver cations from solution in patterns that mimic the structure
of the ferroelectric domains in the substrate. Under the same
conditions, TiO2 by itself reduces insignicant amounts of
silver due to its large band gap (3.2 eV) making it ineffective
in visible light. The observations indicate that electrons
generated in the low band gap BiFeO3 substrate are inuenced
by dipolar elds in the ferroelectric domains and transported
through the TiO2 lm to reduce silver on the surface.
4.2. Photocatalytic activity using ferroelectric materials
4.2.1.Dye degradation
There have been limited studies in literature discussing the
photocatalytic performance of ferroelectric materials in view of
the internal electric eld. The only few studies available are for
dye degradation and water splitting. In 2013 Dunn and co-

Fig. 27. Schematic illustrating that domains in the substrate promote the same
half reactions on both the bare BaTiO3 substrate (a) and with TiO2 thin lm on
top ( 15 nm) (b) [72].

workers used BaTiO3 to probe the inuence of ferroelectricity


on the decolorization of a dye molecule Rhodamine-B [61].
BaTiO3 (BTO) powder was annealed at 1200 1C for 10 h to
convert it from a cubic (paraelectric) to tetragonal (ferroelectric) phase. The XRD spectra discussed earlier in Fig. 12
(a) indicated the formation of the tetragonal phase in annealed
samples with the peak splitting at 2 451. The annealed
catalysts have larger particle size (  622 nm) and lower
surface area (0.862 m2/g) than the control samples
( 386 nm, 2 m2/g). A 3-fold increase in the decolorization
rate using annealed BaTiO3 as shown in Fig. 28 was noted.
The photo-decolorization rate increased in the following order:
BTO (slowest) o BTO annealedo Ag BTO o Ag BTO
annealed. XPS and TEM studies gave no indication that there
were any signicant changes to the composition and surface of
the BaTiO3 before and after annealing. The authors related the
enhanced ferroelectric nature of the annealed samples to
effective adsorption of the dye, enhanced charge carrier
separation which lead to the enhanced catalytic performance.
However, as mentioned earlier, it is not easy to decouple
whether the improved adsorption and activity of the annealed
BTO samples is due to the improved crystallinity/cleanliness
or the ferroelectric polarization.
In the above study, the relationship between ferroelectric
polarization and photocatalytic activity was not clear. Yang
and coworkers came up with another method to investigate the
effect of polarization on photocatalytic dye degradation [80].
They prepared nanosized ferroelectric BaTiO3 via a hydrothermal method to investigate the effect as seen in Fig. 29.
Using piezo-response force microscopy PFM and Raman
microscopy the authors found that the polarization drastically
decreases at higher temperatures for these nano-sized ferroelectric catalysts. At such small sizes, the ferroelectric polarization becomes very small (close to zero) at 80 1C. Fig. 30
The photocatalytic activities of 7.5 nm BaTiO3 were
assessed at 30 1C and 80 1C through decolorization of
Rhodamine-B (Rh B). The decolorization rate of 7.5 nm
BaTiO3 at 30 1C is  11.9% higher than for 7.5 nm BaTiO3
at 80 1C over 40 min [80]. The effect of separation of the

Fig. 28. Photo-decolorization proles of RhB by BaTiO3 under solar


simulator. The annealed catalysts of BaTiO3 show higher activity, especially
when modied by Ag nanoparticles [61].

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

21

Fig. 29. (a) TEM image, (b) HRTEM image of the as-synthesized BaTiO3 nanoparticles [80].

Fig. 30. Photocatalytic activity of BaTiO3 nanoparticles at 30 1C and 80 1C


(reaction for 40 min). The inset schematic shows the internal spontaneous
polarization of ferroelectric materials affects photocatalytic reaction [80].

Fig. 31. Photo-decolorization of RhB over BiFeO3 at pH 2 and 6.7 under AM


1.5 illumination. Decolorization at pH 2 shows greater than 95% decolorization
after 10 min, the inset shows the decolorization of RhB over nanostructured
(Degussa P25) TiO2 for comparison [81].

photoexcited carriers due to the internal electric eld could


weaken due to lower polarization at 80 1C. To eliminate the
other effects in this photocatalytic process, large BaTiO3
(4 500 nm), was prepared and its photo-decolorization was
investigated as a control sample. It is known that in bulk
BaTiO3 there is no loss in ferroelectric polarization at 80 1C.
The control results show that the photo-decolorization rate
does not change greatly at different temperatures for larger
sized particles. The results indicated that ferroelectricity can
directly affect the photocatalytic activity.
Dunn and coworkers observed a high rate of visible light
photochemical decolorization of Rhodamine-B using BiFeO3
nanoparticles [81]. A representative example of dye decolorization over the BiFeO3 nanopowder at pH 2 and 6.7 is shown in
Fig. 31. The results indicate that decolorization of RhB is more
than 95% after 10 min of illumination under simulated solar
irradiation with BFO, at pH 2. This is much better compared to
TiO2 which took 30 min for 90% decolorization of RhB at
similar pH. The high catalytic performance of BiFeO3 under solar
light is attributed to its low band gap ( 2.1 eV) and the internal
electric eld which prevents the recombination of charge carriers.
At higher pH around 6.7, the rate reduced to  90%

decolorization after 20 min on BiFeO3 nanoparticles but still


closely matched the performance of TiO2. Metal oxides surfaces
are assumed to be in an ionization state and have surfaces of
hydroxyls that can participate in acidbase reactions as well as
ion exchange. Any change in solution pH affects the surface
charge of the particles and reposition the potentials of the
photocatalytic reactions, leading to a change in the reaction rate.
The isoelectric point or point of zero charge (PZC) represents
both the pH at which the oxide surface would have zero net
charge and the pH resulting in electrically equivalent concentrations of positive and negative complexes. At normal pH values of
DI water (pH 6.7), the zeta potential value of BFO nanoparticles
is  16 mV and changing the pH to strongly acidic (pH 2.2)
reduces the zeta potential to  1.62 mV, and the isoelectric point
was extrapolated to be at pH 2 [81]. Hence at pH of 6.7, the
surface of BFO particles have a negative charge and are overcrowded due to the development of the negative charge co-ions
screening on BFO by OH  ions present in the suspension as the
dye was dissolved in DI water. This results in fewer dye
molecules to come in direct contact with the catalyst surface
and leads to the lower dye degradation rates observed. As the pH
value brought down to 2, the BFO particle surface reaches iso-

22

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

electric point allowing RhB dye molecules adsorbed at BFO


surfaces, and this is evidenced by high amount of RhB dye
photodecolourisation.
In spite of its high dye decolorization rate, the authors report
a decrease during the reaction, attributed to the photocorrosion of the BiFeO3 catalyst. The valence band of the
BiFeO3 lies above the lower energy levels of both the dye and
water. Thus, it is not possible for BiFeO3 to inject holes into
the RhB dye due to the alignment of bands while RhB dye
molecules are able to inject holes into the BiFeO3. This leads
to oxidation of the BiFeO3 and initiates photocorrosion. This
was conrmed by XPS studies before and after photoreactions
as shown in Fig. 32 and in agreement with other reports [82].
The position of Fe 2p is expected to be at 711 eV for Fe3 and
709.5 eV for Fe2 . From the XPS spectra in Fig. 32(c), the 3/2
and 1/2 spin orbit doublet components of the Fe 2p photoemission located at 711.5 and 725.8 eV respectively were
identied and assigned to Fe3 . No Fe2 and Fe peaks were
found indicating that BiFeO3 nanoparticle surface has a single
phase with Fe present in the 3 valence state. A comparison
of the spectra obtained after photoreaction indicates that there
is signicant change in the chemical environment of the Fe3
cation (Fig. 32(c)) and no visible changes in the environment
for O and Bi (Fig. 32(a) and (b)). The changes to the XPS
spectra for Fe are consistent with dissolution of the Fe within
the BiFeO3 lattice in the surrounding solution. The FeO
bonds are the least stable and so most able to corrode. It seems
likely that the pathway for photocorrosion is the dissolution
and breaking of the FeO bonds to effectively etch iron from
the surface of the BiFeO3 lattice.
Dunn and co-workers further studied the performance of
ferroelectric LiNbO3 for photocatalytic decolorization of Acid
black 1 (amido black 10B) and Rhodamine-B [83]. Lithium
niobate possess a strong polarization (6578 Ccm  2), a Curie
point of 1210 1C and band gap of 3.7 eV making it a good
candidate as a photocatalyst to decolorize dye molecules under
UV light. Powders of lithium niobate were doped with iron or
magnesium to form n-type or p-type material respectively [83].
The performance of LiNbO3 powders was compared to
standard TiO2 powder. Adsorption of the dyes was studied
after 30 min in dark conditions using UV analysis as shown in
Table 1. The results have not been normalized to surface area.
The results indicate high adsorption of dye molecules on the
LiNbO3 powders close to TiO2 powders despite its lower
surface area. Polar nature of lithium niobate powder arising
from its ferroelectricity enables a higher loading of dye
molecules on the catalyst per unit surface area. The doped
LiNbO3 powders show much higher adsorption than both
undoped LiNbO3 and TiO2 catalysts. The authors attributed
this to the effect of doping on the absorbed layer. However,
this mechanism is not clear as doping will increase charge
carriers in the bulk of LiNbO3 leading to increased internal
screening of the surface polarization resulting in a decrease in
external screening. The reduction in external screening should
cause fewer dye molecules to be adsorbed. Fig. 33. shows the
decolorization proles of the dye molecules using the different
LiNbO3 powders. The rate of decolorization of the dye

Fig. 32. XPS spectra for BiFeO3 before and after exposure to irradiation in
RhB pH 6.7 solution. (a) indicates the O1s peak with some minor changes to
the peak shape and position, (b) the Bi 1f peak with a slight shift in peak
position, and (c) the Fe 2p peak showing a signicant change in the relative
ratios of Fe3 peaks at at 711.5 and 725.8 eV after illumination [81].

Table 1
Adsorption of Acid Black-1 and Rhodamine-B by lithium niobate and TiO2
powders under dark Conditions. Dye removal % is not scaled for surface area.
(b) % dye removed during equilibrium phase of 30 minutes dark absorption
[83].
Powder

Spontaneous
polarization
(C cm  2)

Surface Area
(m2 g  1)

Acid
black 1b

Rhodamine bb

LiNbO3
Fe:LiNbO3
Mg:LiNbO3
TiO2

78
72
65

1.4
0.8
1.3
10

10.2
14.1
15.2
11.5

10.9
16.5
16.3
12.7

solutions was found to be fastest over p-type material (Mgdoped) than with the undoped powder, with the n-type proving
(Fe-doped) least effective.
The difference in decolorization rates is attributed to the
majority carrier type: electrons in n-type and holes in the ptype material. Photoexcited holes oxidize water or the hydroxyl ion to form the hydroxyl radical while electrons reduce
oxygen to form the superoxide anion radical. The data suggest
that reactions driven by hole carriers produce a greater rate of

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

decolorization than those driven by electrons because the amount


of oxygen adsorbed in the aqueous solutions will be low due to
the low solubility of oxygen at the temperatures of solution found
during the experiment, typically in the range of 5060 1C,
meaning availability of oxygen molecules would be limited while
water molecules and hydroxyl species are abundant [83]. Another
factor inuencing the reaction rate over the p-type and n-type
material in comparison to lithium niobate may be the effect of

Fig. 33. Decolorization curves of (a) Rhodamine-B and (b) Acid Black-1
under simulated solar light using lithium niobate, iron-doped lithium niobate,
or magensium-doped lithium niobate powder as the catalyst [83].

23

dopants upon impurity scattering and rate of recombination of


excited carriers. Iron with a larger atomic radius ( 156 pm)
compared to Magnesium ( 145 pm) can cause more scattering
and charge carrier recombination.
Rohrer and coworkers have done some interesting studies on
the photocatalytic dye degradation from heterostructures comprising of ferroelectric core and non-ferroelectric shell of TiO2
[8486]. In the rst study of its kind, they prepared powders
composed of microcrystalline (mc) PbTiO3 cores coated with
nanostructured (ns) TiO2 shells using a solgel method as
shown in the TEM images in Fig. 34 [86]. The core shell
structure consisting of a micron scale core is to ensure that the
volume is sufcient to support full polarization and band
bending, which will enhance charge separation. To benet
from any possible advantages imparted by dipolar elds from
ferroelectric domains, it is necessary for the core to be in the
range of about 200 nm.
Fig. 35 shows the dye degradation of methylene blue upon
irradiation to visible light ( 4 420 nm). The heterostructured
powder degrades methylene blue (MB) at a rate 4.8 times
greater than PbTiO3, TiO2, or mechanical mixtures of both. A
schematic of energy level diagrams for both negative and
positive polarizations is shown in Fig. 36(a) and (b), respectively [86]. The band gap energy of PbTiO3 and TiO2 (anatase)
is assumed to be 2.85 eV and 3.2 eV, respectively. As
discussed earlier in Fig. 23, ferroelectric polarization and
screening effect leads to band bending at the interface.
Negative surface polarization is screened by the free holes in
the PbTiO3 leading to upward band bending (increasing energy
levels) in PbTiO3. This negative surface polarization is also
screened externally by the holes in the TiO2 shell also leading
to upward band bending in TiO2. This is illustrated in Fig. 36
(a) where positive surface polarization leads to an opposite
effect with downward band bending (decreasing energy levels)
in the PbTiO3 and TiO2.
Under visible light irradiation, photoexcited charge carriers
will mainly be generated in the PbTiO3 core. In the c
domains, electrons in PbTiO3 will be drawn to the interface

Fig. 34. TEM images of heterostructured particles composed of mc-PbTiO3 and ns-TiO2 annealed at 500 1C. (a) Low magnication bright eld image of entire
particle, (b) high magnication bright eld image of detail of the PbTiO3TiO2 interface [86].

24

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

while in c domains holes are drawn to the interface. Once


these charge carriers reach the PbTiO3TiO2 interface there are
energy barriers for transmission [86]. For example, electrons
transported to the interface in c domains (Fig. 36(b))
encounter increasing energy levels in the TiO2 shell layer.
The charge carriers need to reach the TiO2 surface for the
oxidation/reduction reactions to proceed. As the TiO2 shell is
only on the order of 1015 nm, tunneling may be the dominant
process for the charge carriers to overcome the energy barriers
discussed above.
The TiO2 is nanostructured and mesoporous with a large
surface area/active sites for the dye to absorb. Therefore, when
the charge carriers reach the TiO2 surface, the photogenerated
electrons combine with the adsorbed oxygen molecules to
form superoxide radical anions, O2 . The photogenerated
holes oxidize excited dye molecules to form radicals or they
oxidize surface adsorbed H2O to form hydroxyl radicals, OH.
Thus the enhanced photochemical reactivity is attributed to
various factors such as: (i) the absorption of visible light by the
PbTiO3 core (ii) the separation of photogenerated carriers by

Fig. 35. Photochemical dye degradation with PbTiO3TiO2 annealed at


500 1C and its component phases during irradiation by visible light ( 4420
nm). Blank refers to data from a control experiment conducted without the
addition of a catalyst [86].

internal elds in the ferroelectric core and (iii) reaction at the


surface of a nanostructured TiO2 shell.
Rohrer and coworkers made a variety of coreshell photocatalysts with TiO2 shell and different cores of ferrite and
titanate based ABO3 materials [85]. To authors wanted to
investigate the use of ferrite based core materials such as
BiFeO3, LaFeO3 and YFeO3 to take advantage of their
excellent light absorption properties. The ability of these
catalysts to degrade methylene blue in visible light
(4 420 nm) was compared and is shown in Fig. 37(b) [85].
The heterostructures with titanate core materials (FeTiO3 and
PbTiO3) showed enhanced performance compared to TiO2
alone. This was consistent with earlier studies and as discussed
above [86]. The heterostructures with ferrite cores such as
BiFeO3, LaFeO3 and YFeO3 showed comparable or even
worse performance to TiO2 itself. This interesting phenomenon
suggests that the B site in ABO3/TiO2 is directly related to the
photocatalytic activity of the heterostructures. Fig. 38 compares the estimated band positions of the materials relative to
standard hydrogen electrode (SHE) level, which is dened as
0 eV. There is no obvious trend between the band edge
positions and the reactivitys of the heterostructures. The only
notable correlation is that the compounds containing Fe on the
B-site all have valence band edge positions that are above the
positions of the titanates. This suggests that a larger hole
transfer overpotential at the ferrite/TiO2 might limit the
reactivity of the ferrite based materials.
Another possible reason for the enhanced activity of the
titanates versus ferrite cores, is the continuity of TiOTi
bonds across the ATiO3/TiO2 interface. This might reduce the
barrier for transmission across the interface from the core to
the TiO2 shell. Both the Ti and Fe are octahedrally coordinated
in the core (Fe and Ti) and shell (Ti). The bonding between
Fe3 in the FeO6 octahedra to Ti4 in TiO6 octahedra will
form FeOTi bonds, which changes the polarity compared to
TiOTi interfacial bonds. It is reasonable to assume that
electrons occupying states in bands formed by the overlap of

Fig. 36. A schematic energy level diagram of PbTiO3TiO2 with (a) negative polarization and (b) positive polarization normal to the heterostructure interface. Evac,
EC, EF, Ev, and Es are the energies of the vacuum level, conduction band, Fermi level, valance band, and surface potential, respectively [86].

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

25

Fig. 37. (a) A schematic of coreshell structures with TiO2 shell and a core of ferroelectric titanate or ferrite material. Interface between the core shell is shown as
well. (b) Photo-degradation of methylene blue with (Pb,Fe)TiO3/TiO2 and (Bi, La, Y)FeO3 annealed at 500 1C and TiO2 under visible light illumination [85].

Fig. 38. Schematic of band gap and band energy levels for materials
considered here [85].

Ti and O orbitals will experience less of a potential change at


the interface than electrons in states formed from the overlap of
a Fe and O orbital, and thus experience less scattering. Thus
the better charge transfer properties at the ATiO3/TiO2 interface versus AFeO3/TiO2 interface might be the reason for the
improvement in photocatalytic activity.

4.2.2.Water splitting
Developing clean, low-cost and renewable fuel sources is
essential to meet the energy demands of a growing population.
The use of hydrogen as a fuel is one method of producing
clean energy with minimal impact to our environment.
Hydrogen fuel is currently produced by steam reforming of
natural gas. Photocatalytic water splitting with semiconducting
materials represents a greener method of hydrogen generation.
Though the process has been studied for more than two
decades and considerable progress has been made, the overall
efciency is still low efciency because of various factors as
discussed earlier [8791]. Ferroelectric materials may offer
additional useful concepts for photocatalytic water splitting
because of the available permanent electric eld which can
help with charge carrier separation and diffusion. Despite their
potential there have been very limited studies on the use of
ferroelectric materials for photocatalytic water splitting.
Some of the pioneering work on photocatalytic water splitting
using ferroelectric photocatalysts was done by Inoue and coworkers in the 80s and early 90s. They fabricated a device (Fig.
39(a)) consisting of a thin-lm (20200 nm thick) of TiO2

deposited on ferroelectric LiNbO3 substrates [92]. The ferroelectric substrates were poled prior to the TiO2 deposition. After a
thin Pt deposition (6 nm) on the TiO2 surface, photocatalytic
water splitting was demonstrated under irradiation from Xe lamp
[92]. The following four kinds of substrates were employed: a
poled single crystal of ferroelectric lithium niobate (LiNbO3) with
(1) c domain ( ), and (2) c domain ( ), (3) a poled single
crystal of non-ferroelectric lithium tantalate (LiTaO3) ( ), and
(4) Al2O3 single crystal (n).
Fig. 39(b) shows the hydrogen production rates as a function
of the substrate with 100 nm thick TiO2 and 6 nm Pt. The
activity was in the order: LiNbO3 c substrate 4LiNbO3 c
substrate 4 LiTaO3 substrate 4Al2O3 substrate. These results
were the rst indication of the role of ferroelectric polarization
in charge carrier separation and enhancement of photocatalytic
activity. Inoue and coworkers explained the enhancement
based on a band bending mechanism due to ferroelectric
polarization. This was 16 years before Rohrer and coworkers
used the same concept to report spatially selective reduction of
metal ions [69,75].
Similar to the effect reported in heterostructures of TiO2 and
PbTiO3 in Fig. 36 [86], the mechanism of charge carrier
separation depends on the ferroelectric polarization. On
negatively poled LiNbO3 crystals there is an upward band
bending on the LiNbO3TiO2 interface due to negative
polarization charge. The photoexcited holes drift towards the
interface with LiNbO3 to compensate for the negative polarization while electrons towards the interface with Pt where
reduction takes place to produce H2. This enhanced the water
splitting reaction and the movement of charge carriers is
illustrated in Fig. 39(a). For positively poled substrates the
photoexcited carries move in opposite directions leading to
lower catalytic activity seen. For nonferroelectric LiTaO3 there
is no internal electric eld to assist with charge carrier
separation, thus leading to lower activity. Lastly the Al2O3
substrates show little/zero activity. This indicates that it is not
only photoexcited carriers in the TiO2 lm which take part in
the water splitting reaction but also excited charge carriers in
the semiconducting substrates (LiNbO3 and LiTaO3) which
participate in the redox reactions.
The authors also studied the activity as a function of
the TiO2 thickness. As seen in Fig. 40, the activity peaked at

26

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

Fig. 39. (a) Schematic representation of Pt/TiO2/ferroelectric cell structure (b) H2 production in the liquid-phase reaction (Pt 6 nm; TiO2 100 nm) [92].

Fig. 40. Photocatalytic activity variations as a function of TiO2 lm thickness


(Pt 6 nm) [92].

 100 to 120 nm thick TiO2 lms. The H2 evolution is seen to


be independent of the substrate if the TiO2 lms are thinner
than  30 nm or thicker than  150 nm. This dependence is
also associated with this ferroelectric polarization effect: for
the thin TiO2 layer, the carrier densities in TiO2 are insufcient
to be affected by the polarization eld, whereas for the thicker
TiO2 layers the polarization eld effect is attenuated to a
negligibly small level.
Inoue and coworkers also studied the anomalous photovoltaic
effects in these ferroelectric crystals [93,94]. Fig. 41 shows the
photoresponse of poled potassium doped lead niobate (PKN)
oxide ferroelectric crystal, measured in air under illumination by
Xe light. The measurements were made after the deposition of
transparent Au electrodes by Ar sputtering in the direction normal
to the polarization eld. The Ag wire was connected to the

electrodes with conducting resin [94]. Photocurrent owing


through bulk by illuminating either face of the poled disks was
measured. There were no appreciable current ows before poling
as observed by dotted lines in Fig. 41. However, after poling, the
oxide shows not only a larger transient pyroelectric current but
also a signicant steady photocurrent. With the light off, the
steady-state current attenuates down to almost zero level. When
the direction of spontaneous polarization axis is reversed (bottom
half of Fig. 41), the resulting current ows in the opposite
direction. These results clearly show that the behavior of photoexcited carriers can be controlled by the polarization axis.
After these initial efforts by Inoue and coworkers, the
research activity in developing ferroelectric materials for
photocatalytic water splitting declined. In 2011 a paper by
Misra and co-workers reported the use of single-phase BiFeO3
(BFO) nanoparticles for water decomposition under simulated
solar light [95]. The BFO nanoparticles were 5060 nm in size
with a band gap of 2.1 eV as calculated from UVvis
absorption measurements. Fig. 42(a) shows the hydrogen
production rates from a 1(M) KOH solution. It showed three
times better photocatalytic activity than P25 Degussa TiO2
nanoparticles under visible light. Though no detailed analysis
was done on the effect of ferroelectric polarization on
photocatalytic activity, the authors proposed that the internal
electric eld in BFO nanoparticles helps suppress electron
hole recombination. The combination of the internal electric
eld and low band gap (visible light absorption) in BFO lead
to the enhancement in photocatalytic activity compared to P25
nanoparticles. In addition the authors also performed photoelectrochemical (PEC) experiments on the different catalysts.
The BFO nanoparticles were coated on Ti foil (BFOTi) as
catalyst (exposed area 1.0 cm2) and used as photoanode in
the PEC cell to evaluate its activity for water photoelectrolysis
in 1(M) KOH electrolyte under simulated solar light illumination (87 mW/cm2). The PEC tests were carried out using Pt as
cathode and Ag/AgCl as reference electrode. The PEC

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

activities are presented in Fig. 42(b) where the photocurrent is


recorded as a function of the applied potential. The maximum
photocurrent density obtained from BFOTi photoanode is
0.36 mA/cm2 at 0.5 VAg/AgCl. Under the same illumination
conditions, oxygen annealed Ti photoanode showed  0.3 mA/
cm2 photocurrent density.
Similarly a study by Maggard and coworkers in 2011 briey
discussed the effect of ferroelectricity on photocatalytic
performance of PbTiO3 using Pt metal as cocatalyst [96].
The effect of ferroelectric polarization was indicated by a size
effect of the catalyst on the photocatalytic activity. In general,
particle size has been shown to have a signicant impact on the
photocatalytic activity with the smaller particle sizes generally
exhibiting higher photocatalytic rates due to higher surface
areas. The authors observed an inverse effect with PbTiO3.
This was correlated to a loss of ferroelectricity in nano-sized
PbTiO3 particles. Given that the ferroelectric polarization is
diminished in nanosized particles, this effect may explain the
decreased activity observed in the smallest PbTiO3 particles.

Fig. 41. Photoresponse of PKN oxide (dotted line) before poling and (Solid
line) after poling [94].

27

What the authors failed to address was the effect of crystallinity in these different sized particles on the photocatalytic
activity.
In 2012, Rohrer and coworkers published a study on the
effect of ferroelectric polarization on photocatalytic water
splitting activity using heterostructures comprising of ferroelectric core of BaTiO3 and nanoshell of TiO2 [84]. This was
similar to the groups earlier work on dye degradation using
these heterostructured catalysts [85,86]. The catalysts had 50
nm thick nanocrystalline nc-TiO2 surrounding micron scaled
ms-BaTiO3 cores such that the heterostructured particles had
surface areas of 40100 m2/g. Fig. 43(a) shows the H2
production rates from a water/methanol (92:8%) mixture under
UV light using different catalysts annealed at 600 1C. The
metal loading on all catalysts was kept consistent at 1 wt%
Platinum.
The ms-BaTiO3/nc-TiO2 coreshell photocatalysts had the
highest rates of photocatalytic hydrogen production from
water/methanol solutions much greater than those for ncTiO2 or ms-BaTiO3 alone. When normalized to surface area
the hydrogen production rates from BaTiO3/TiO2 was  28
mmol/m2, TiO2  16 mmol/m2 and BaTiO3  2 mmol/m2, as
seen in Fig. 43(b). This is consistent with the authors earlier
work as discussed in Fig. 36. This result suggests that the
internal electric eld in the ferroelectric core enhances charge
carrier separation in the nanoshell of TiO2 and leads to the
improved activities. The movement of photoexcited electrons
and holes is domain dependant and depends on the band
bending at the interface of BaTiO3 and TiO2.
Very recently, Park and co-workers published the rst
detailed study on developing a powdered ferroelectric photocatalyst for photocatalytic water splitting. They demonstrated
the effect of poling on ferroelectric powdered catalyst and
discussed the effect on photocatalytic performance [97]. The
authors used KxNa1  xNbO3 (NKN) powder as a ferroelectric
photocatalyst, prepared by solid-state method. The hysteresis
loops were measured by making pellets and depositing
electrodes on either side. Fig. 44(a) shows the polarization

Fig. 42. (a) Volume versus time graph showing the volume of H2 evolved during the photocatalytic splitting of water. It is seen that BFO generates 3 times more H2
than commercial P25 TiO2 nanoparticles under visible light. (b) Potentiodynamic plot of BFO at 0.4 V where Ag/AgCl electrode is used as the reference electrode
[95].

28

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

Fig. 43. (a) Hydrogen production rates from mc-BTO/ns-TiO2, mc-BTO, and ns-TiO2 annealed at 600 1C. The measurements were taken using platinized (1 wt%
Pt) powders suspended in 100 mL 8% methanolwater mixture solution under a UV light. (b) Total hydrogen production per unit surface area measured after 6 h of
a platinized photocatalyst (Reproduced from Ref. [76]) [84].

Fig. 44. (a) Hysteresis curves of Na1  xKxNbO3 solid-solution ceramics (b) Illustration of the corona-poling system [97].

versus electric eld curves for the KxNa1  xNbO3 catalyst


displaying a remnant polarization (Pr) and coercive eld (Ec)
of 14.7 mC cm  2 and 14.0 kVcm  1, respectively. To study the
effect of polarization the powdered ferroelectric photocatalyst
was poled using a corona-poling method as depicted in Fig. 44
(b). This system primarily consists of two electrodes: one is
designed with a sharp-cornered shape (corona needle), and the
other is designed with a smooth, large-diameter rounded shape
(Cu disk). When the electric eld is applied on the needle
electrode, the electric-eld intensity around the sharp point
becomes much higher than it is elsewhere and strong enough
to ionize a neutral uid (air); accordingly, a plasma composed
of positive ions and free electrons is formed. These charged
particles are separated from each other by the electric eld and
separated electrons collide with other atoms/molecules producing further electron/positive-ion pairs which continue to
repeat this process. If the needle electrode is more negative
than the disk electrode, the ions are attracted to the needle and
the electrons are repelled to the disk (negative corona)
inducing a negative charge on the upper surface of the
material. The photocatalytic activity was tested under UV
light. The H2 evolution rate of the polarized ferroelectric

powder was  0.47 mmol g  1 h  1, which was 7.4 times


higher than that of the non-polarized NKN powder ( 
63 mmol g  1 h  1) as seen in Fig. 45(a) [97]. It was noted
that polarized NKN powder also showed the higher photocatalytic activity (40 mmol g  1 h  1) than non-polarized one
under Xe lamp irradiation. These results demonstrate that the
enhancement of photocatalytic activity in ferroelectric powder
caused by polarization can be mainly attributed to the internal
dipole eld. After the poling process, a permanent internal
electric eld is generated in the powder by the remnant
polarization. When the powder is irradiated with light and
electrons and holes are photogenerated, the internal eld
promotes the separation of the electrons and holes. These
results were reproducible several times. The polarization
dependence of photocatalytic activity was further conrmed
by a composition study of NKN powders. As seen in Fig. 45
(a), the highest polarization is for x 0.5. The photocatalytic
activity also showed a similar trend showing maximum
enhancement at x 0.5 [97].
In order to demonstrate this advantage of polarization more
directly, time-resolved photoluminescence (PL) lifetime measurements were performed [97]. As shown in Fig. 45(b), the

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

29

Fig. 45. (a) H2 evolution rates for polarized and non-polarized catalysts under UV light. (Reproduced from Ref. [89].) (b)Decay prole of photoluminescence from
K0.5Na0.5NbO3 powders before and after corona poling [97].

electrons that were excited in polarized NKN have the longer


life time than non-polarized one. The calculated lifetime of
charge carriers in polarized NKN powder was 3.31 ns, which
is signicantly longer than that of non-polarized one (1.64 ns).
This result clearly demonstrates that the internal eld can
inhibit the charge recombination, promoting the separation of
the electrons and holes.
5. Conclusions and outlook
In this review, we focus on the potential of ferroelectric
materials for use in photocatalytic applications. The review
starts with fundamental properties of ferroelectric materials
such as the spontaneous polarization, screening effect and the
rise of the anomalous photovoltaic effect. We then discuss the
use of these materials for solar cells and the effect of the
internal electric eld in charge carrier separation. In the
following section, we discussed the effect of the polarization
on the surface photochemistry and adsorption properties.
Lastly, we reviewed the work done on the use of these
materials for photocatalytic applications such as dye degradation and water splitting. These studies showed that depending
on the surface polarization, different regions of the surface
experience different extents of band bending and promote
different carriers to move to spatially different locations. This
leads to some unique adsorption properties and spatially
selective surface redox reactions. The recombination of
photo-generated carriers and the back reaction of H2 and O2
to reform water are frequently cited as important factors
limiting the efciency of photolysis. A considerable work is
been carried out to develop catalyst structures that spatially
separate the charge carriers and the H2 and O2 production sites.
The engineering of such spatially selective ferroelectric photocatalysts could prove critical to reducing the recombination of
electronhole pairs, back reactions of H2 and O2 and separation of H2 and O2 leading to highly efcient photocatalytic
systems for water splitting. Ferroelectric materials alone may
not provide the needed photo-catalysts for water splitting to
hydrogen. Yet, future photo-catalysts will be composed of
functional materials where both spatial and temporal congurations are determining factors. The considerable amount of

fundamental knowledge gained in the last two decades on


ferroelectric materials, once understood, is poised to contribute
into the making of new photocatalytic materials for hydrogen
production from water.
References
[1] T. Hisatomi, J. Kubota, K. Domen, Recent advances in semiconductors
for photocatalytic and photoelectrochemical water splitting, Chem. Soc.
Rev. 43 (2014) 75207535.
[2] A. Kudo, Y. Miseki, Heterogeneous photocatalyst materials for water
splitting, Chem. Soc. Rev. 38 (2009) 253278.
[3] J. Tang, D. Wang, H. Han, C. Li, Roles of cocatalysts in photocatalysis
and photoelectrocatalysis, Acc. Chem. Res. 46 (2013) 19001909.
[4] A.J. Cowan, C.J. Barnett, S.R. Pendlebury, M. Barroso, K. Sivula,
M. Grtzel, J.R. Durrant, D.R. Klug, Activation energies for the ratelimiting step in water photooxidation by nanostructured -Fe2O3 and
TiO2, J. Am. Chem. Soc. 133 (2011) 1013410140.
[5] J. Tang, J.R. Durrant, D.R. Klug, Mechanism of photocatalytic water
splitting in TiO2: reaction of water with photoholes, importance of charge
carrier dynamics, and evidence for four-hole chemistry, J. Am. Chem.
Soc. 130 (2008) 1388513891.
[6] L. Li, P.A. Salvador, G.S. Rohrer, Photocatalysts with internal electric
elds, Nanoscale. 6 (2014) 2442.
[7] K.A. Connelly, H. Idriss, The photoreaction of TiO2 and Au/TiO2 single
crystal and powder surfaces with organic adsorbates. Emphasis on
hydrogen production from renewables, Green Chem. 14 (2012) 260.
[8] L. Ji, M.D. McDaniel, S. Wang, A.B. Posadas, X. Li, H. Huang, J.C. Lee,
A.A. Demkov, A.J. Bard, J.G. Ekerdt, E.T. Yu, A silicon-based
photocathode for water reduction with an epitaxial SrTiO3 protection
layer and a nanostructured catalyst, Nat. Nano 10 (2015) 8490.
[9] Z. Zhan, Y. Wang, Z. Lin, J. Zhang, F. Huang, Study of interface electric
eld affecting the photocatalysis of ZnO, Chem. Commun. 47 (2011)
45174519.
[10] C. Tang, M.J.S. Spencer, A.S. Barnard, Activity of ZnO polar surfaces:
an insight from surface energies, Phys. Chem. Chem. Phys. 16 (2014)
2213922144.
[11] A. Onodera, Masaki Takesada, Electronic Ferroelectricity in IIVI
Semiconductor ZnO, Adv. Ferroelectr. (2012) http://dx.doi.org/10.5772/
5230.
[12] I. Grinberg, D.V. West, M. Torres, G. Gou, D.M. Stein, L. Wu, G. Chen,
E.M. Gallo, A.R. Akbashev, P.K. Davies, J.E. Spanier, A.M. Rappe,
Perovskite oxides for visible-light-absorbing ferroelectric and photovoltaic materials, Nature 503 (2013) 509512.
[13] N. Nuraje, K. Su, Perovskite ferroelectric nanomaterials, Nanoscale. 5
(2013) 87528780.
[14] D. Tiwari, S. Dunn, Photochemistry on a polarisable semi-conductor:
what do we understand today?, J. Mater. Sci. 44 (2009) 50635079.

30

M.A. Khan et al. / Surface Science Reports 71 (2016) 131

[15] Y. Yuan, T.J. Reece, P. Sharma, S. Poddar, S. Ducharme, A. Gruverman,


Y. Yang, J. Huang, Efciency enhancement in organic solar cells with
ferroelectric polymers, Nat. Mater. 10 (2011) 296302.
[16] Y. Yuan, Z. Xiao, B. Yang, J. Huang, Arising applications of ferroelectric
materials in photovoltaic devices, J. Mater. Chem. A 2 (2014) 6027.
[17] H.T. Yi, T. Choi, S.G. Choi, Y.S. Oh, S.W. Cheong, Mechanism of the
switchable photovoltaic effect in ferroelectric BiFeO3, Adv. Mater. 23
(2011) 34033407.
[18] J. Curie, P. Curie, Dveloppement, par pression, de llectricit polaire
dans les cristaux hmidres faces inclines, Comptes Rendus De. l'Acad
emie Des. Sci. 91 (1880) 294295.
[19] J. Varghese, R.W. Whatmore, J.D. Holmes, Ferroelectric nanoparticles,
wires and tubes: synthesis, characterisation and applications, J. Mater.
Chem. C. 1 (2013) 2618.
[20] J. Curie, P. Curie, Sur llectricit polaire dans les cristaux hmidres
faces inclines, C. R. Acad. Sci. Gen. 91 (1880) 383386.
[21] J. Valasek, Piezo-Electric and allied phenomena in rochelle salt, Phys.
Rev. 17 (1921) 475481.
[22] C.H. Ahn, K.M. Rabe, J.-M. Triscone, Ferroelectricity at the nanoscale:
local polarization in oxide thin lms and heterostructures, Science 303
(2004) 488491.
[23] M.A. Khan, U.S. Bhansali, H.N. Alshareef, Fabrication and characterization of all-polymer, transparent ferroelectric capacitors on exible
substrates, Org. Electron. 12 (2011) 22252229.
[24] P. Lunkenheimer, J. Mller, S. Krohns, F. Schrettle, A. Loidl,
B. Hartmann, R. Rommel, Md Souza, C. Hotta, J.A. Schlueter,
M. Lang, Multiferroicity in an organic charge-transfer salt that is
suggestive of electric-dipoledriven magnetism, Nat. Mater. 11 (2012)
755758.
[25] M.E. Lines, A.M. Glass, Principles and Applications of Ferroelectrics and
Related Materials, Clarendon Press, Oxford, 2001.
[26] T. Mitsui, I. Tatsuzaki, E. Nakamura, An Introduction to the Physics of
Ferroelectrics, Gordon and Breach Science Publishers, 1976.
[27] W.C. Yang, B.J. Rodriguez, A. Gruverman, R.J. Nemanich, Photo
electron emission microscopy of polarity-patterned materials, J. Phys.
Condens. Mater. 17 (2005) S1415.
[28] D.D. Fong, A.M. Kolpak, J.A. Eastman, S.K. Streiffer, P.H. Fuoss, G.
B. Stephenson, C. Thompson, D.M. Kim, K.J. Choi, C.B. Eom,
I. Grinberg, A.M. Rappe, Stabilization of Monodomain polarization in
ultrathin PbTiO3 lms, Phys. Rev. Lett. 96 (2006) 127601.
[29] V.M. Fridkin, Bulk photovoltaic effect in noncentrosymmetric crystals,
Crystallogr. Rep. 46 (2001) 654658.
[30] J.F. Scott, Applications of modern ferroelectrics, Science 315 (2007)
954959.
[31] G.H. Kwei, A.C. Lawson, S.J.L. Billinge, Structures of the ferroelectric
phases of barium titanate, J. Phys. Chem. 97 (1993) 23682377.
[32] J.A. Caraveo-Frescas, M.A. Khan, H.N. Alshareef, Polymer ferroelectric
eld-effect memory device with SnO channel layer exhibits record hole
mobility, Sci. Rep. 4 (2014).
[33] R. Watton, Ferroelectric materials and devices in infrared detection and
imaging, Ferroelectrics 91 (1989) 87108.
[34] Y. Xu, Ferroelectric Materials and Their Applications, Elsevier, 2013.
[35] V.M. Fridkin, Popov, Anomalous photovoltaic effect in ferroelectrics,
Sov. Phys. Usp. 21 (1978) 981991.
[36] T. Choi, S. Lee, Y.J. Choi, V. Kiryukhin, S.-W. Cheong, Switchable
ferroelectric diode and photovoltaic effect in BiFeO3, Science 324 (2009)
6366.
[37] R.V. Baltz, Theory of the Anomalous Bulk Photovoltaic Effect in
Ferroelectrics, Phys. Stat. Sol. B 89 (1978) 419429.
[38] S.Y. Yang, J. Seidel, S.J. Byrnes, P. Shafer, C.H. Yang, M.D. Rossell,
P. Yu, Y.H. Chu, J.F. Scott, J.W. Ager, L.W. Martin, R. Ramesh, Abovebandgap voltages from ferroelectric photovoltaic devices, Nat. Nanotechnol. 5 (2010) 143147.
[39] L. Arizmendi, Photonic applications of lithium niobate crystals, Phys.
Status Solidi A 201 (2004) 253283.
[40] J. Carnicero, O. Caballero, M. Carrascosa, J.M. Cabrera, Superlinear
photovoltaic currents in LiNbO3: analyses under the two-center model,
Appl. Phys. B. 79 (2004) 351358.

[41] V.S. Dharmadhikari, W.W. Grannemann, Photovoltaic properties of


ferroelectric BaTiO3 thin lms rf sputter deposited on silicon, J. Appl.
Phys. 53 (1982) 89888992.
[42] I. Masaaki, M. Yasushi, N. Takeshi, Electrical Properties of Ferroelectric
Lead Lanthanum Zirconate Titanate as an Energy Transducer for
Application to Electrostatic-Optical Motor, Jpn. J. Appl. Phys. 41 (2002)
6993.
[43] M. Qin, K. Yao, Y.C. Liang, High efcient photovoltaics in nanoscaled
ferroelectric thin lms, Appl. Phys. Lett. 93 (2008) 122904.
[44] S.Y. Yang, L.W. Martin, S.J. Byrnes, T.E. Conry, S.R. Basu, D. Paran,
L. Reichertz, J. Ihlefeld, C. Adamo, A. Melville, Y.H. Chu, C.H. Yang, J.
L. Musfeldt, D.G. Schlom, J.W. Ager, R. Ramesh, Photovoltaic effects in
BiFeO3, Appl. Phys. Lett. 95 (2009) 062909.
[45] A. Bhatnagar, A. Roy Chaudhuri, Y. Heon Kim, D. Hesse, M. Alexe,
Role of domain walls in the abnormal photovoltaic effect in BiFeO3, Nat.
Commun. 4 (2013).
[46] B. Yang, Y. Yuan, P. Sharma, S. Poddar, R. Korlacki, S. Ducharme,
A. Gruverman, R. Saraf, J. Huang, Tuning the energy level offset
between donor and acceptor with ferroelectric dipole layers for increased
efciency in bilayer organic photovoltaic cells, Adv. Mater. 24 (2012)
14551460.
[47] K.S. Nalwa, J.A. Carr, R.C. Mahadevapuram, H.K. Kodali, S. Bose,
Y. Chen, J.W. Petrich, B. Ganapathysubramanian, S. Chaudhary,
Enhanced charge separation in organic photovoltaic lms doped with
ferroelectric dipoles, Energy Environ. Sci. 5 (2012) 7042.
[48] M.A. Khan, J.A. Caraveo-Frescas, H.N. Alshareef, Hybrid dual gate
ferroelectric memory for multilevel information storage, Org. Electron. 16
(2015) 917.
[49] R.C.G. Naber, C. Tanase, P.W.M. Blom, G.H. Gelinck, A.W. Marsman,
F.J. Touwslager, S. Setayesh, D.M. de Leeuw, High-performance
solution-processed polymer ferroelectric eld-effect transistors, Nat.
Mater. 4 (2005) 243248.
[50] R.D. Nasby, R.K. Quinn, Photoassisted electrolysis of water using a
BaTiO3 electrode, Mater. Res. Bull. 11 (1976) 985992.
[51] M.H. Zhao, D.A. Bonnell, J.M. Vohs, Effect of ferroelectric polarization
on the adsorption and reaction of ethanol on BaTiO3, Surf. Sci. 602
(2008) 28492855.
[52] Y. Yun, E.I. Altman, Using ferroelectric poling to change adsorption on
oxide surfaces, J. Am. Chem. Soc. 129 (2007) 1568415689.
[53] D. Li, M.H. Zhao, J. Garra, A.M. Kolpak, A.M. Rappe, D.A. Bonnell, J.
M. Vohs, Direct in situ determination of the polarization dependence of
physisorption on ferroelectric surfaces, Nat. Mater. 7 (2008) 473477.
[54] K.S. Kim, M.A. Barteau, Reactions of Methanol on TiO2 (001) single
crystal surfaces, Surf. Sci. 223 (1989) 1332.
[55] M.A. Henderson, S. Otero-Tapia, M.E. Castro, The chemistry of
methanol on the TiO2 (110) surface : the inuence of vacancies and
coadsorbed species, Faraday Discuss. 114 (1999) 313329.
[56] U. Diebold, The surface science of titanium dioxide, Surf. Sci. Rep. 48
(2003) 53229.
[57] K.S. Kim, M.A. Barteau, Reactions of aliphatic alcohols on the {001}facetted TiO2 (001) surface, J. Mol. Catal. 63 (1990) 103117.
[58] K.S. Kim, M.A. Barteau, Adsorption and decomposition of aliphatic
alcohols on TiO2, Langmuir. 4 (1988) 533543.
[59] L. Gamble, L.S. Jung, C.T. Campbell, Decomposition and protonation of
surface ethoxys on TiO2 (110), Surf. Sci. 348 (1996) 116.
[60] M.H. Zhao, D.A. Bonnell, J.M. Vohs, Inuence of ferroelectric polarization on the energetics of the reaction of 2-uoroethanol on BaTiO3, Surf.
Sci. 603 (2009) 284290.
[61] Y. Cui, J. Briscoe, S. Dunn, Effect of ferroelectricity on solar-light-driven
photocatalytic activity of BaTiO3inuence on the carrier separation and
stern layer formation, Chem. Mater. 25 (2013) 42154223.
[62] Y. Yun, L. Kampschulte, M. Li, D. Liao, E.I. Altman, Effect of
ferroelectric poling on the adsorption of 2-propanol on LiNbO3 (0001),
J. Phys. Chem. C 111 (2007) 1395113956.
[63] Y. Yun, M. Li, D. Liao, L. Kampschulte, E.I. Altman, Geometric and
electronic structure of positively and negatively poled LiNbO3 (0001)
surfaces, Surf. Sci. 601 (2007) 46364647.

M.A. Khan et al. / Surface Science Reports 71 (2016) 131


[64] Z. Zhang, P. Sharma, C.N. Borca, P.A. Dowben, A. Gruverman,
Polarization-specic adsorption of organic molecules on ferroelectric
LiNbO3 surfaces, Appl. Phys. Lett. 97 (2010) 243702.
[65] R. Holscher, S. Sanna, W.G. Schmidt, Adsorption of OH and H at the
LiNbO3 (0001) surface, Phys. Status Solidi C 9 (2012) 13611365.
[66] S. Kim, M.R. Schoenberg, A.M. Rappe, Polarization-dependence of
palladium deposition on ferroelectric lithium niobate(0001) surfaces,
Phys. Rev. Lett. (2011).
[67] S.V. Levchenko, A.M. Rappe, Inuence of ferroelectric polarization on
the equilibrium stoichiometry of lithium niobate (0001) surfaces, Phys.
Rev. Lett. 100 (2008) 256101.
[68] Y. Inoue, I. Yoshioka, K. Sato, Polarization effects upon adsorptive and
catalytic properties. 1. Carbon monoxide oxidation over palladium
deposited on lithium niobate (LiNbO3) ferroelectrics, J. Phys. Chem. 88
(1984) 11481151.
[69] J.L. Giocondi, G.S. Rohrer, Spatial separation of photochemical oxidation
and reduction reactions on the surface of ferroelectric BaTiO3, J. Phys.
Chem. B 105 (2001) 82758277.
[70] P.M. Jones, S. Dunn, Interaction of stern layer and domain structure on
photochemistry of lead-zirconate-titanate, J. Phys. D. Appl. Phys. 42
(2009) 065408.
[71] A. Bhardwaj, N.V. Burbure, A. Gamalski, G.S. Rohrer, Composition
dependence of the photochemical reduction of Ag by Ba1  xSrxTiO3,
Chem. Mater. 22 (2010) 35273534.
[72] N.V. Burbure, P.A. Salvador, G.S. Rohrer, Photochemical reactivity of
titania lms on BaTiO3 substrates: origin of spatial selectivity, Chem.
Mater. 22 (2010) 58235830.
[73] A.M. Schultz, Y. Zhang, P.A. Salvador, G.S. Rohrer, Effect of crystal and
domain orientation on the visible-light photochemical reduction of Ag on
BiFeO3, ACS. Appl. Mater. Int. 3 (2011) 15621567.
[74] Y. Zhang, A.M. Schultz, P.A. Salvador, G.S. Rohrer, Spatially selective
visible light photocatalytic activity of TiO2/BiFeO3 heterostructures, J.
Mater. Chem. 21 (2011) 4168.
[75] J.L. Giocondi, G.S. Rohrer, Spatially selective photochemical reduction
of silver on the surface of ferroelectric barium titanate, Chem. Mater. 13
(2001) 241242.
[76] S. Dunn, P.M. Jones, D.E. Gallardo, Photochemical growth of silver
nanoparticles on c  and c domains on lead zirconate titanate thin
lms, J. Am. Chem. Soc. 129 (2007) 87248728.
[77] K.A. Mller, H. Burkard, SrTiO3: an intrinsic quantum paraelectric below
4 K, Phys. Rev. B 19 (1979) 35933602.
[78] S.V. Kalinin, D.A. Bonnell, T. Alvarez, X. Lei, Z. Hu, J.H. Ferris,
Q. Zhang, S. Dunn, Atomic polarization and local reactivity on ferroelectric surfaces: a new route toward complex nanostructures, Nano. Lett.
2 (2002) 589593.
[79] Y. Sun, R.J. Nemanich, Photoinduced Ag deposition on periodically
poled lithium niobate: wavelength and polarization screening dependence, J. Appl. Phys. 109 (2011) 104302.
[80] R. Su, Y. Shen, L. Li, D. Zhang, G. Yang, C. Gao, Y. Yang, Silvermodied nanosized ferroelectrics as a novel photocatalyst, Small 11
(2015) 202207.
[81] C. Hengky, X. Moya, N.D. Mathur, S. Dunn, Evidence of high rate
visible light photochemical decolourisation of Rhodamine B with BiFeO3
nanoparticles associated with BiFeO3 photocorrosion, RSC Adv. 2 (2012)
11843.
[82] M. Popa, S. Preda, V. Fruth, K. Sedlckov, C. Balzsi, D. Crespo, J.
M. Caldern-Moreno, BiFeO3 lms on steel substrate by the citrate
method, Thin Solid Films 517 (2009) 25812585.

31

[83] M. Stock, S. Dunn, Inuence of the ferroelectric nature of lithium niobate


to drive photocatalytic dye decolorization under articial solar light, J.
Phys. Chem. C 116 (2012) 2085420859.
[84] L. Li, G.S. Rohrer, P.A. Salvador, E. Dickey, Heterostructured ceramic
powders for photocatalytic hydrogen production: nanostructured TiO2
shells surrounding microcrystalline (Ba,Sr)TiO3 cores, J. Am. Ceram.
Soc. 95 (2012) 14141420.
[85] L. Li, X. Liu, Y. Zhang, N.T. Nuhfer, K. Barmak, P.A. Salvador, G.
S. Rohrer, Visible-light photochemical activity of heterostructured coreshell materials composed of selected ternary titanates and ferrites coated
by TiO2, ACS Appl. Mater. Int., 5, , 2013, p. 50645071.
[86] L. Li, Y. Zhang, A.M. Schultz, X. Liu, P.A. Salvador, G.S. Rohrer,
Visible light photochemical activity of heterostructured PbTiO3TiO2
coreshell particles, Catal. Sci. Technol. 2 (2012) 1945.
[87] V. Jovic, K.E. Smith, H. Idriss, G.I.N. Waterhouse, Heterojunction
synergies in titania-supported gold photocatalysts: implications for solar
hydrogen production, ChemSusChem. (2015) http://dx.doi.org/10.1002/
cssc.201500126 n/a-n/a.
[88] Z.H.N. Al-Azri, W.-T. Chen, A. Chan, V. Jovic, T. Ina, H. Idriss, G.I.
N. Waterhouse, The roles of metal co-catalysts and reaction media in
photocatalytic hydrogen production: performance evaluation of M/TiO2
photocatalysts (M Pd, Pt, Au) in different alcoholwater mixtures, J.
Catal. 329 (2015) 355367.
[89] W.-T. Chen, A. Chan, D. Sun-Waterhouse, T. Moriga, H. Idriss, G.I.
N. Waterhouse, Ni/TiO2: a promising low-cost photocatalytic system for
solar H2 production from ethanolwater mixtures, J. Catal. 326 (2015)
4353.
[90] L. Sinatra, A.P. LaGrow, W. Peng, A.R. Kirmani, A. Amassian, H. Idriss,
O.M. Bakr, A Au/Cu2OTiO2 system for photo-catalytic hydrogen
production. A pn-junction effect or a simple case of in situ reduction?,
J. Catal. 322 (2015) 109117.
[91] A.K. Wahab, S. Bashir, Y. Al-Salik, H. Idriss, Ethanol photoreactions
over AuPd/TiO2, Appl. Petrochem. Res. 4 (2014) 5562.
[92] Y. Inoue, M. Okamura, K. Sato, A. Thin-Film Semiconducting, TiO2
Combined with ferroelectrics for photoassisted water decomposition, J.
Phys. Chem. 89 (1985) 51845187.
[93] Y. Inoue, K. Sate, K. Sato, H. Miyamat, Photoassisted water decomposition by ferroelectric lead zirconate titanate ceramics with anomalous
photovoltatic effects, J. Phys. Chem. 90 (1986) 28092810.
[94] Y. Inoue, O. Hayashi, K. Sato, Photocatalytic activities of potassiumdoped lead niobates and the effect of poling, J. Chem. Soc. Faraday
Trans. 86 (1990) 22772282.
[95] J. Deng, S. Banerjee, S.K. Mohapatra, Y.R. Smith, M. Misra, Bismuth
Iron Oxide, Nanoparticles as photocatalyst for solar hydrogen generation
from water, J. Fundam. Renew. Energy. Appl. 1 (2011) 110.
[96] D. Arney, T. Watkins, P.A. Maggard, Effects of particle surface areas and
microstructures on photocatalytic H2 and O2 production over PbTiO3, J.
Am. Ceram. Soc. 94 (2011) 14831489.
[97] S. Park, C.W. Lee, M.G. Kang, S. Kim, H.J. Kim, J.E. Kwon, S.Y. Park,
C.Y. Kang, K.S. Hong, K.T. Nam, A ferroelectric photocatalyst for
enhancing hydrogen evolution: polarized particulate suspension, Phys.
Chem. Chem. Phys. 16 (2014) 1040810413.

You might also like