Alternative strategies in improving the photocatalytic and photoelectrochemical activities of visible light-driven BiVO4: a review

Hui Ling Tan , Rose Amal * and Yun Hau Ng *
Particles and Catalysis Research Group, School of Chemical Engineering, The University of New South Wales, Sydney, NSW 2052, Australia. E-mail:;

Received 22nd May 2017 , Accepted 30th June 2017

First published on 3rd July 2017

The research interest on bismuth vanadate (BiVO4) has heightened over the past decade due to its proven high activity for water oxidation and organic degradations under visible light. Although metal doping and water-oxidation cocatalyst loading have been widely demonstrated to be useful to overcome the poor electron transport and slow water oxidation kinetics of BiVO4, the efficiency of this material is still greatly limited by poor charge separation. Various efforts directed at modifying the surface and bulk properties to improve the performance of BiVO4-based materials have therefore been developed, including crystal facet engineering, coupling with graphitic carbon material, annealing treatment, and nanoscaling. This review aims to provide insights into the most recent progress in these strategies in regard to their influences on the charge separation, transport, and transfer aspects of BiVO4, all of which are crucial to govern photochemical conversion efficiency. Understanding of these charge kinetics in relation to the properties of BiVO4 is of fundamental importance for rational design of BiVO4 with optimum structures, which may serve as a general guideline for the fabrication of metal oxide photocatalysts.

image file: c7ta04441k-p1.tif

Hui Ling Tan

Hui Ling Tan received her bachelor and PhD degrees in Chemical Engineering from the University of New South Wales (UNSW) in 2012 and 2017, respectively. Currently, she is a research associate in the Particles and Catalysis Research Group (PartCat) at the UNSW. Her major research interests include fabrication of metal oxide and carbon-based materials for photocatalytic and photoelectrochemical water splitting.

image file: c7ta04441k-p2.tif

Rose Amal

Rose Amal is Australian Laureate Fellow and Scientia Professor in the School of Chemical Engineering at the University of New South Wales. She has worked in the area of particle technology for over 20 years with an emphasis on fine particle aggregation, photocatalysis, and nanoparticle synthesis. More recently, her research focus has been on the design of photocatalysts and engineering systems for solar-induced processes. Her research has produced over 300 refereed publications. More information on her research can be found at

image file: c7ta04441k-p3.tif

Yun Hau Ng

Yun Hau Ng received his PhD from Osaka University in 2009. After a brief research visit to the Radiation Laboratory at the University of Notre Dame (Prashant Kamat's group), he joined the ARC Centre of Excellence for Functional Nanomaterials at the UNSW with the Australian Postdoctoral Fellowship (APD) in 2011. He is currently a senior lecturer in the School of Chemical Engineering at the UNSW. His research is focused on the development of novel photoactive semiconductors (particles and thin films) for sunlight energy conversion. He received the Honda–Fujishima Prize in 2013 in recognition of his work in the area of photodriven water splitting. He has been named an Emerging Investigator in the Journal of Materials Chemistry A in 2016.

1. Introduction

Photocatalysis, which uses semiconductors to directly harvest and convert the renewable solar energy into chemical energy, has been regarded as a promising approach to solve our current environmental and energy challenges. In this regard, organic degradation and water splitting are the two most significant applications of photocatalysis, which are respectively directed towards environmental pollutant abatement and H2 production. Despite the different utilization of the two processes, they follow the same principal in which absorption of light energy creates electrons and holes to initiate redox reactions on the surface of the semiconductor to degrade or generate specific compounds. Therefore, the photochemical conversion efficiency of a semiconductor photocatalyst is determined by its light absorption ability and charge kinetics, which includes separation of photogenerated electrons and holes, charge transport to reaction sites, and interfacial charge transfer for chemical reactions.

To date, metal oxides are the type of semiconductor most extensively studied for the two photocatalytic applications due to their high chemical and photostability, low cost, and ease of fabrication. However, most binary metal oxides have large band gap energies attributed to their intrinsic highly positive valence bands that consist of O 2p orbitals located at ca. +3.0 V vs. NHE.1 This restricts their light absorption ability and thus leads to low solar-to-hydrogen (STH) efficiencies of the materials. For instance, with band gap energies of 3.0–3.2 eV for TiO2, 3.2–3.3 eV for ZnO, and 2.7 eV for WO3, their maximum theoretical STH efficiencies are only ca. 2% for TiO2 and ZnO,2,3 and ca. 4.5% for WO3.4 Design of metal oxides with smaller band gaps is therefore necessary for effective utilization of the solar energy.

Given that the energy levels of the conduction bands of metal oxides are generally composed of empty orbitals (LUMOs) of metal cations with d0 and d10 configurations that are not significantly negative, lowering of these bands is thermodynamically undesirable. Instead, a new valence band formed by orbitals of other elements not associated with O 2p is imperative to reduce the band gaps of metal oxides by upshifting of the valence band edge.5 To this end, ternary metal oxides with their valence bands formed by atomic orbitals of more than one element have come to the fore. Orbitals of Bi 6s in Bi3+, Ag 4d in Ag+, Pb 6s in Pb2+, and Sn 5s in Sn2+ can contribute to valence band formation above the valence bands formed by O 2p orbitals in metal oxides.5 As a result, ternary metal oxides containing Bi, Ag, Pb, or Sn such as BiVO4,6 Bi2WO6,7 AgNbO3,8 PbCrO4,9 and SnNb2O6[thin space (1/6-em)]10 have relative narrower band gaps with promising visible light absorption ability.

BiVO4, in particular with a monoclinic scheelite structure, has attracted tremendous research attention following the landmark paper by Kudo et al. on visible light-triggered water oxidation on this ternary metal oxide in the presence of Ag+ ions as electron scavengers.11 Water oxidation, which involves four-electron transfer, is the thermodynamically more demanding half reaction of water splitting in comparison to the two-electron transfer water reduction. While hybridization of Bi 6s–O 2p orbitals upshifts the valence band of monoclinic BiVO4 to a lower potential at ca. +2.5 eV vs. NHE,4 it still provides sufficient potential for this material to oxidize not only water, but also various organic compounds with a smaller band gap energy of 2.4–2.5 eV. Reduction of the band gap can potentially improve the STH efficiency of this material to about 9%,2,12,13 which is significantly higher than those of the common binary metal oxides mentioned above.

Recognition of the great potential of BiVO4 as a visible light-active photocatalyst has prompted extensive research on this material, encompassing it as the powder-type and electrode-type photocatalysts, whereby BiVO4 in the latter form is utilized as a photoanode in the photoelectrochemical system. Nevertheless, the efficiency of BiVO4 is generally low due to poor electron transport,12 slow water oxidation kinetics,14 and low carrier mobility.15 Although recent reviews have demonstrated Mo or W doping and surface modification with a water-oxidation cocatalyst as the most common approaches useful to improve the electron conductivity and water oxidation kinetics of BiVO4, respectively,16,17 coupling of doped-BiVO4 with water-oxidation cocatalyst revealed that the treated BiVO4 was predominantly limited by poor charge separation.18 Development of alternative strategies to improve the charge kinetics of BiVO4, in particular charge separation, is indispensable to achieve high-performing BiVO4.

Given that the charge kinetics of a photocatalyst is strongly affected by the surface and bulk properties of the material, strategies such as (1) crystal facet engineering, (2) coupling with graphitic carbon material, (3) annealing treatment, and (4) nanoscaling show promise to enhance the efficiency of BiVO4. We have recently investigated the impacts of each of these strategies on the charge separation, transport, and/or transfer aspects of BiVO4.19–22 In this review, a general discussion on the crystal and electronic structures of BiVO4 will first be given to facilitate understanding of BiVO4 photoactivity dependence on the crystal structure. Next, the photocatalytic applications of BiVO4 under visible light and the related mechanisms will be discussed, in which emphasis is placed on water splitting. Subsequently, the major limitations on the photoactivity of BiVO4 from the aspects of charge kinetics will be reviewed, followed by the most recent progress in strategies 1–4 to circumvent these limitations. This review includes the most up-to-date experimental and computational breakthroughs, including those from our laboratory, in order to provide insights into the fundamental dependence of metal oxides' charge kinetics on the morphological and electronic properties of the materials for rational design of optimal photocatalytic structures.

2. Fundamental properties of BiVO4

Originally BiVO4-based compounds were extensively studied as the potential substitutes for the lead-, cadmium-, and chromate-based pigments that are widely used in the coating and plastics industry because of their non-toxic nature.23 Over the past four decades, the interest in BiVO4 was mainly focused on its various technological properties such as ferroelasticity,24–29 acousto-optical,26 and ionic conductivity.30 These properties are strongly dependent on the crystal structures of BiVO4. More recent research interest on BiVO4, however, is driven by its photocatalytic activity which also depends on the crystal structures with different electronic structures.6,31 A concise overview of the different crystal structures available for BiVO4 and their corresponding optical properties and electronic structures are discussed in this section to facilitate understanding of the structure-dependent photocatalytic activity of BiVO4.

2.1 Crystal structures and phase transitions

Naturally occurring BiVO4 exists as the mineral pucherite with an orthorhombic structure. Nevertheless, normal laboratory synthesis routes do not produce BiVO4 of the similar form. Three other main crystal structures are more commonly obtained: a scheelite structure with tetragonal and monoclinic phases and a zircon structure with tetragonal phase.24,32 In all three structures, each V ion is coordinated by four O atoms in a tetrahedral site and each Bi ion is coordinated by eight O atoms. The distinction between the scheelite and zircon structures is that each Bi ion is bounded to eight VO4 tetrahedral units in the former structure, while the latter only has six VO4 around one Bi ion.16 For the scheelite structure, BiVO4 with tetragonal and monoclinic phases are dissimilar as the local environments of V and Bi ions are notably distorted in the latter. Hence, the V–O bonds in tetragonal scheelite BiVO4 are all of equal length (1.72 Å), while two different V–O bond lengths (1.77 Å and 1.69 Å) are present in monoclinic scheelite BiVO4.33

In general, low temperature synthesis such as precipitation at room temperature produces zircon-structured BiVO4, while high temperature syntheses (e.g., solid-state and melting reactions) result in the formation of monoclinic scheelite BiVO4.6 However, the crystal structure of the obtained BiVO4 is highly determined by the preparation method. Monoclinic scheelite BiVO4 is also attainable via aqueous processes at room temperature, resulting from the aging of the crystallization process of tetragonal scheelite BiVO4.31 This suggests that the formation of tetragonal scheelite BiVO4 is kinetically favorable, while monoclinic scheelite is a more thermodynamically stable form of BiVO4. The phase transitions are also possible via heat treatment: transformation of scheelite BiVO4 with monoclinic and tetragonal phases occurs reversibly at 528 K;24 tetragonal zircon BiVO4 irreversibly transforms into monoclinic scheelite at 670–770 K.6 Irreversible transition from tetragonal zircon to monoclinic scheelite has also been demonstrated to be viable through mechanical crushing.29

2.2 Optical properties and electronic structures

On the basis of the comparable band gap energies between tetragonal scheelite (2.34 eV) and monoclinic scheelite (2.41 eV) BiVO4, Tokunaga et al. suggested that the electronic structures of these scheelite-structured BiVO4's are similar.31 However, they found that monoclinic scheelite BiVO4 showed significantly higher activity for O2 evolution from aqueous AgNO3 solution (i.e., water oxidation reaction with Ag+ as the sacrificial reagent) compared to tetragonal scheelite BiVO4, under both UV and visible light irradiations.31 The distinct difference in photocatalytic activity between the two BiVO4 was attributed to the difference in the distortion of the local structure. In particular, distortion of the Bi–O polyhedron by the 6s2 lone pair of Bi3+ primarily in the monoclinic BiVO4 improves charge separation and photocatalytic activity.

On the other hand, Kudo et al. demonstrated that the water oxidation activity of monoclinic scheelite BiVO4 was higher than that of zircon-type BiVO4 under visible light irradiation, in spite of their comparable activity under UV irradiation.6 The different activities under visible light irradiation was revealed to be mainly due to the enhanced optical absorption of monoclinic scheelite BiVO4 compared to tetragonal zircon BiVO4, as portrayed in Fig. 1. Compared to zircon-type BiVO4, an additional valence band formed by 6s states of Bi3+ (i.e., Bi 6s orbitals) or a hybrid Bi 6s–O 2p orbitals was postulated to present above the O 2p orbitals in scheelite BiVO4. The electron transition from the Bi 6s or hybrid Bi 6s–O 2p valence band (that is of lower energy than the O 2p valence band) to the V 3d conduction band hence leads to smaller band gap energy of scheelite BiVO4 (2.4 eV) in comparison to that of zircon BiVO4 (2.9 eV) with V–O transition.6

image file: c7ta04441k-f1.tif
Fig. 1 Band structures of tetragonal zircon and monoclinic scheelite BiVO4 proposed by Kudo et al. Reproduced from ref. 16 with permission from The Royal Society of Chemistry.

The findings by Tokunaga et al. and Kudo et al., as described above, collectively suggest that among the three structures, monoclinic scheelite is the most active phase of BiVO4 for photocatalytic applications under visible light irradiation. Therefore, the electronic structure of monoclinic scheelite BiVO4 based on computational and experimental findings reported to date is further reviewed hereafter. Density functional theory (DFT) calculations by Walsh et al. indicated that the BiVO4 is a direct band gap semiconductor with the conduction band mainly composed of V 3d states with contributions from O 2p and Bi 6p, whereas the top of the valence band consists of Bi 6s–O 2p antibonding state.34 These structures were affirmed by Payne et al. using X-ray photoemission spectroscopy (XPS), X-ray emission spectroscopy (XES), and X-ray absorption spectroscopy (XAS),35 corroborating hybridization of Bi 6s states and O 2p states at the valence band maximum of monoclinic BiVO4 as suggested by Kudo et al.6

However, a different structure was proposed by Zhao et al. according to their DFT calculations, in which the BiVO4 was predicted as an indirect band gap semiconductor.37 The top of the valence band was estimated to be predominated by the nonbonding states of O 2pπ and Bi 6s, whereas the bottom of the conduction band is primarily the nonbonding V 3dx2y2 and 3dz2 states. The relatively smaller band gap of this BiVO4 was ascribed to the lone-pair distortion impact of Bi 6s, contributing to the upshift of the O 2p states to lower energy. A more recent study by Cooper et al. provides a comprehensive band structure of the BiVO4 (Fig. 2), which was derived from DFT calculations in conjunction with experimental evidence obtained via a combination of X-ray spectroscopies, including XAS, XES, resonant inelastic X-ray scattering (RIXS), and XPS.36 The valence band was found to primarily be composed of O 2p states, whereby hybridized Bi 6p–O sp2 and V 3d–O sp2 orbitals contribute to the respective lowest and middle regions, whereas O 2pπ and Bi 6s contribute to the top region, which is consistent with that reported by Zhao et al.37 Likewise, the conduction band consists of V 3d states with dominant dx2y2 and dz2 characters at the lowest energy edge, but contains Bi 6p character in the upper region. The band gap energy was determined to be 2.5 eV with the Fermi energy at the surface measured to be 2.0 eV above the valence band maximum, indicating that the BiVO4 is an n-type semiconductor.

image file: c7ta04441k-f2.tif
Fig. 2 Energy level diagram representing the electronic structure of monoclinic scheelite BiVO4 proposed by Cooper et al. based on DFT calculations and experimental spectra of XAS, XES, XPS, and RIXS. All energy levels are shown with respect to vacuum. Reprinted with permission from ref. 36. Copyright 2014 American Chemical Society.

Consistently, the DFT calculations by Walsh et al.34 and Zhao et al.37 suggest that the effective masses of both holes and electrons in monoclinic scheelite BiVO4 are smaller than other oxide materials. While Walsh et al. predicted a minimum effective mass of 0.3m0 for both electrons and holes, Zhao and coworkers found the minimum effective masses of 0.9m0 for electrons and 0.7m0 for holes, where m0 represents electron rest mass. Since the drift velocity of electrons and holes is inversely proportional to the effective mass,38 the smaller effective masses suggest easier charge separation and migration to reach the surface reaction sites, indicating monoclinic scheelite BiVO4 as a promising photocatalyst candidate.

3. Photocatalytic application of BiVO4 under visible light

Depending on the change in Gibbs free energy, photocatalytic reactions can be classified into two types: downhill and uphill (Fig. 3).39 Photooxidation of organic compounds (i.e., organic degradations) that occurs irreversibly using oxygen molecules is generally a downhill reaction. This type of reaction is regarded as a photoinduced reaction because it is instigated by the reactive oxygen species (ROS) formed by the redox reactions between photogenerated charge carriers and water molecules or dissolved oxygen. Meanwhile, water splitting into H2 and O2 involves the conversion of photon energy into chemical energy, accompanied by a large positive change in Gibbs free energy (ΔG° = 273 kJ mol−1). This reaction is similar to photosynthesis by green plants, which is also an uphill reaction. Therefore, water splitting is otherwise known as artificial photosynthesis.
image file: c7ta04441k-f3.tif
Fig. 3 Two distinct classifications of photocatalytic reactions: downhill and uphill. Reproduced with permission from ref. 39. Copyright 2003 Plenum Publishing Corporation.

This section will focus on the fundamental principles of the photocatalytic reactions of organic degradation and water splitting. The comparison of BiVO4 as a potent visible light-active photocatalyst alongside other oxide materials on their respective functionalities in these applications is also covered. Given that monoclinic scheelite BiVO4 is the only structure that shows high photoactivity under visible light, the BiVO4 mentioned henceforth will refer to monoclinic scheelite BiVO4, unless otherwise stated.

3.1 Organic degradations

Environmental pollution (e.g., water and air) resulting from various anthropogenic and industrial activities is largely composed of organic compounds such as pesticides, dyes, chloro-organics, and surfactants. Great research effort has been focused on photocatalysis as a viable option for environmental remediation since the functional groups of organic compounds may be oxidized, allowing transformation or degradation of the organic contaminants into less harmful substances.40 Oxide semiconductors with highly positive valence band, in particular TiO2, have been extensively studied and shown to be effective in photooxidative degradation of numerous organic compounds.40,41

Organic degradation under aqueous phase or humid conditions can generally proceed via two pathways: direct oxidation by holes and indirect oxidation by ROS such as superoxide (O2˙) and hydroxyl (˙OH) radicals.42 ˙OH, which is one of the most active and nonselective ROS, is highly oxidizing and formed by the reaction of photogenerated holes with the adsorbed H2O or hydroxide groups present on the surface of oxide photocatalysts. Simultaneously, electrons in the conduction band are transferred to oxygen to generate O2˙ that can trigger the subsequent formations of hydroperoxyl radicals (HO2˙) and hydrogen peroxide (H2O2), depending on the reaction conditions. These ˙OH, O2˙, HO2˙, and H2O2 are the important oxidant sources involved in the photocatalytic degradation of organics by TiO2.40,41

Despite having a considerably more negative valence band potential compared to that of TiO2, the thermodynamically weaker photooxidation power of BiVO4 has also been reported to be highly efficient in the photodegradation of organic materials under visible light, including water-soluble and volatile organic compounds. The former includes organic dyes (e.g., rhodamine B,43–45 methylene blue,46,47 and methyl orange48,49) and non-ionic surfactants (e.g., 4-n-nonylphenol,50 nonylphenol,51 and 4-n-alkyl-phenol52), while an example of the latter is ethylene.53 In fact, the feasibility of photocatalytic organic degradation on BiVO4 under visible light was shown to outperform that on TiO2, whose activity is strictly restricted to UV light.

Mechanistically, ˙OH radicals were proposed by several research groups as one of the active species to be involved in the photocatalytic oxidation of organic compounds by BiVO4,43,54 similar to that reported for TiO2. For instance, on the basis of increased rhodamine B photodegradation rate with increasing pH value, one possible pathway for the photocatalytic decomposition of rhodamine B was attributed by Martinez-de La Cruz to the generation of ˙OH radicals on BiVO4 (Fig. 4).43 At higher pH, a higher concentration of hydroxide (OH) ions is present in the aqueous solution. Adsorption of the OH ions on the surface of BiVO4 and direct interaction with the photogenerated holes thus result in the formation of ˙OH radicals (OH + BiVO4(h+) → ˙OH + BiVO4), which subsequently function as the oxidant to degrade the rhodamine B molecules available on the photocatalyst surface.

image file: c7ta04441k-f4.tif
Fig. 4 Schematic of the photocatalytic degradation pathway of rhodamine B (RhB) on BiVO4via ˙OH radical formation. Reprinted with permission from ref. 43. Copyright 2009 Elsevier Ltd.

This generally accepted ˙OH-radical driven mechanism, however, was demonstrated by Kohtani et al. in their more recent investigation to be extraneous in the case of BiVO4.55 Generation of ˙OH radicals was detected through its reaction with terephthalic acid to form the highly fluorescent 2-hydroxyterephthalic acid (TAOH). The amount of TAOH formed in the presence of BiVO4 was found to be negligible in comparison to that of TiO2, suggesting the inefficiency of BiVO4 to generate ˙OH radicals by hole oxidation of OH or water molecules. This phenomenon was ascribed to the negatively positioned valence band edge of BiVO4 relative to the redox potential of ˙OH/OH (Fig. 5), indicating that ˙OH generation is thermodynamically unfavorable on BiVO4.

image file: c7ta04441k-f5.tif
Fig. 5 Energy diagram of BiVO4 in comparison to TiO2 at pH 0. Adapted from ref. 55 with permission from the PCCP Owner Societies.

3.2 Water splitting

Large-scale production of H2 as a clean (i.e., carbon-free) and renewable energy carrier is desirable to address the global energy and environmental issues. Combustion of H2 generates heat and work with water as the only byproduct. However, present industrial production of H2 mainly relies on steam reforming, in which fossil fuels (e.g., natural gas) are consumed to result in CO2 emission.
CH4 + H2O → CO + 3H2(1)
CO + H2O → CO2 + H2(2)

One ideal way to solve the energy and environmental concerns is to utilize the renewable and freely available solar energy for H2 generation from water, that is, solar water splitting. The potential of solar energy conversion to produce H2via photocatalytic decomposition of water was widely recognized after the discovery of the Honda–Fujishima effect using TiO2 electrodes in 1972.56 Since then, tremendous efforts have been devoted to investigate water splitting using semiconductor photocatalysts. Water splitting has generally been studied using two approaches: powdered photocatalyst and photoelectrode (i.e., photocatalyst immobilized on a conducting substrate), which are hereafter denoted as powder suspension (PS) and photoelectrochemical (PEC) systems, respectively.

3.2.1 Powder suspension system. Solar water splitting using powdered photocatalyst shows promise for practical H2 production on a large scale because of its simplicity and low cost. Typically, the photocatalyst powders only need to be dispersed in a water pool and then H2 will be readily obtained by exposing the suspension to sunlight. The photochemical reactions involved in solar water splitting include two half reactions: (1) oxidation of water molecules by holes for O2 generation (eqn (3)) and (2) reduction of water molecules by electrons to form H2 (eqn (4)).
2H2O → 4H+ + O2 + 4e(3)
2H+ + 2e → H2(4)

One strategy to drive an overall water splitting (i.e., simultaneous generation of H2 and O2 in stoichiometric amounts) is to use a single photocatalyst (single-step photoexcitation), as shown in Fig. 6a.57 Thermodynamically, the bottom of the conduction band of a semiconductor photocatalyst has to be more negative than the redox potential of H+/H2 (0 V vs. NHE) and the top of the valence band more positive than the redox potential of O2/H2O (1.23 V vs. NHE), in order to achieve overall water splitting.

image file: c7ta04441k-f6.tif
Fig. 6 Schematic of (a) single- and (b) two-step photoexcitation system for photocatalytic water splitting. Reprinted with permission from ref. 57. Copyright 2011 Elsevier B.V.

Fig. 7 shows the band potentials of various semiconductors relative to the redox potentials of water splitting. It is noteworthy that a number of oxide semiconductors such as TiO2, SrTiO3, Ta2O5, KTaO3, and ZrO2 have suitable band structures for splitting water into H2 and O2. However, their application is primarily limited to UV light due to their intrinsically large band gaps, which is also the major demerit of tantalate photocatalysts that were shown to be active for overall water splitting.39 On the other hand, despite having suitable band potentials and a narrow band gap responsive to visible light, CdS is unstable and not active for overall water splitting because it is susceptible to photocorrosion. In the absence of a sacrificial electron donor, S2− anions in CdS rather than H2O molecules are preferentially oxidized by the photogenerated holes, resulting in the decomposition of CdS into Cd2+ and S according to the equation below:58,59

CdS + 2h+ → Cd2+ + S(5)

image file: c7ta04441k-f7.tif
Fig. 7 Comparison of the band structures of various photocatalysts with respect to the redox potentials of water splitting. Reproduced from ref. 60 with permission from the PCCP Owner Societies.

In fact, the number of semiconductors that are visible light-active, photochemically stable, and have appropriate conduction and valence band potentials for overall water splitting, are very limited. Band gap engineering of oxide photocatalysts is therefore essential for the development of new photocatalysts that are applicable for overall water splitting under visible light irradiation. For example, formation of a midgap electron donor level via doping, introduction of a new valence band by element other than O 2p (e.g., oxynitrides and oxysulfides), and the production of solid solutions have been demonstrated to be efficient for band gap narrowing (Fig. 8).5,57 To date, (Ga1−xZnx)(N1−xOx) solid solution is one of the few visible light-driven photocatalysts reported to be active for overall water splitting.61

image file: c7ta04441k-f8.tif
Fig. 8 Band gap engineering approaches to design visible light-active oxide photocatalysts. CB, VB, and DL represent the conduction band, valence band, and electron donor level, respectively. Reproduced from ref. 5 with permission from The Royal Society of Chemistry.

An alternative strategy for splitting water into H2 and O2 is to employ a two-step photoexcitation mechanism (Fig. 6b), similar to that in natural photosynthesis by green plants and is known as the Z-scheme.5,57,62 This system was first introduced by Bard in 1979,63 which involves two different photocatalysts: one that is responsible for H2 evolution and the other for O2 evolution. A suitable electron mediator (e.g., redox couple Red/Ox such as IO3/I and Fe3+/Fe2+) is typically required to shuttle the photogenerated electrons from the O2-evolving photocatalyst to the H2-evolving photocatalyst. However, the presence of a reversible ionic redox couple may promote undesirable backward reactions on the photocatalysts (Fig. 9). For example, on the O2-evolving photocatalyst, the thermodynamically more favorable oxidation of reductant (Red) to oxidant (Ox) may compete with oxidation of water to O2. Likewise, reduction of Ox to Red may proceed preferentially over water reduction to H2 on the H2-evolving photocatalyst.57 Such backward reactions suppress forward reactions (H2 and O2 generations), thus strongly hindering the efficiency of the Z-scheme system for overall water splitting. In regard to this, studies related to Z-scheme water splitting have also been directed towards using graphene as the solid-state electron mediator64,65 to promote direct electron transfer between photocatalyst particles via electrostatic interaction without an ionic redox couple as the electron mediator.62

image file: c7ta04441k-f9.tif
Fig. 9 Schematic of the undesirable backward reactions along with the forward reactions in the two-step photoexcitation system. Reprinted with permission from ref. 57. Copyright 2011 Elsevier B.V.

The Z-scheme system offers a greater potential over the single-step system for water splitting under visible light. Visible light-driven semiconductor photocatalysts that are active only for either water reduction or oxidation can potentially be employed for the construction of the Z-scheme system. Consequently, the ability of a great variety of semiconductor photocatalysts for H2 or O2 evolution from water (i.e., half reactions of water splitting) has been widely conducted and evaluated in the presence of sacrificial reagents. Addition of a reducing reagent (i.e., hole scavenger) such as alcohol and sulfide ions promotes consumption of the photogenerated holes via oxidation of the reducing agent. This then enriches the electrons in the photocatalyst for an enhanced H2 evolution reaction (Fig. 10a). On the other hand, in the presence of an oxidizing agent (i.e., electron scavenger) such as Ag+ and Fe3+, the photogenerated electrons are readily used for reducing the oxidizing agent and results in an enhanced O2 evolution reaction (Fig. 10b). However, one should be aware that the photocatalysts tested to be active for the half reactions are not definite to be active for overall water splitting in the absence of sacrificial reagents.5

image file: c7ta04441k-f10.tif
Fig. 10 Photocatalytic H2 or O2 evolution from the aqueous solutions containing sacrificial reagents. Reproduced from ref. 5 with permission from The Royal Society of Chemistry.

Although H2 production is primarily the main focus of water splitting, the O2 evolution half reaction plays an equally significant role in dictating the efficiency of the water splitting reaction since it supplies the electrons needed for water reduction to form H2. In fact, water oxidation is the more challenging half reaction for water splitting because it involves the abstraction of four electrons (refer to eqn (3)), whereas only two electrons are needed for water reduction (refer to eqn (4)). Among the visible light-responsive oxide photocatalysts, WO3[thin space (1/6-em)]66 and BiVO4[thin space (1/6-em)]6,11 have been revealed to be active for O2 generation from water in the presence of a sacrificial reagent, attributed to their deep valence bands despite the insufficient conduction band potential for water reduction, as presented in Fig. 7. However, greater attention is devoted to BiVO4 because of its ability to absorb a larger portion of the solar spectrum, judging from its relatively smaller band gap (2.4–2.5 eV in comparison to 2.7 eV for WO3).67 While the ability of BiVO4 to oxidize water has been proven using Fe3+ as the oxidizing reagent,68 the studies related to water oxidation on BiVO4 are extensively executed using the electron-scavenging Ag+ ions.6,31,33,47

Interestingly, the potential of BiVO4 to function as the O2-evolving photocatalyst in visible light-driven Z-scheme water splitting has also been demonstrated. The construction of the Z-scheme system constituting of BiVO4 as the O2-evolving photocatalyst, Pt/SrTiO3:Rh (Rh-doped SrTiO3) as the H2-evolving photocatalyst and Fe3+/Fe2+ as the redox couple was first reported by Kato et al. to be active for stoichiometric production of H2 and O2 (2[thin space (1/6-em)]:[thin space (1/6-em)]1) under visible light,69 whereby the mechanism of overall water splitting is schematically displayed in Fig. 11. A subsequent study by Sasaki et al. showed that replacement of the Pt cocatalyst on the SrTiO3:Rh photocatalyst with a Ru cocatalyst helped to suppress the undesirable backward reactions (e.g., water formation from H2 and O2, oxidation of Fe2+ by O2, and reduction of Fe3+ by H2) particularly observed using Pt/SrTiO3:Rh.70 This study indicates that Ru/SrTiO3:Rh is a better H2-evolving photocatalyst to construct the (Ru/SrTiO3:Rh)–(BiVO4)–(Fe3+/Fe2+) Z-scheme system for overall water splitting under visible light. Such a Z-scheme system composed of Ru/SrTiO3:Rh and BiVO4 was also shown to split water into H2 and O2 by using a [Co(bpy)3]3+/2+ redox couple,71 a [Co(phen)3]3+/2+ redox couple,71 or solid reduced graphene oxide (RGO)64 as the electron mediator. On the other hand, overall water splitting was also observed to proceed in an electron mediator-free (Ru/SrTiO3:Rh)–(BiVO4) Z-scheme system, in which the interparticle electron transfer was driven by the electrostatic interaction between the H2- and O2-evolving photocatalysts achieved via pH adjustment.62

image file: c7ta04441k-f11.tif
Fig. 11 Mechanism of overall water splitting using the Z-scheme system constructed by BiVO4 and Pt/SrTiO3:Rh. Reprinted with permission from ref. 70. Copyright 2008 Elsevier Inc.

Apart from Pt/SrTiO3:Rh and Ru/SrTiO3:Rh, a very recent study by Iwase et al. also demonstrated photocorrosive metal sulfides as potent H2-evolving photocatalysts in combination with CoOx-loaded BiVO4 as the O2-evolving photocatalyst for Z-scheme water splitting in the presence of RGO as the electron mediator.65 Pt/CuGaS2 and Pt/ZnGeS4 were the only two metal sulfides found to be active in such a system combination. As opposed to the Z-scheme systems containing Rh-doped SrTiO3 as the H2-evolving photocatalyst where stoichiometric water splitting was observed to proceed using bare BiVO4 as the O2-evolving photocatalyst, loading of CoOx as the water-oxidation cocatalyst on BiVO4 was revealed to be necessary to promote electron injection from BiVO4 into the metal sulfide via RGO for water splitting reaction to occur.

3.2.2 Photoelectrochemical system. The report of the Honda–Fujishima effect in 1972 has formed the basis of PEC water splitting. As opposed to the PS system, water reduction and oxidation reactions proceed on two physically separated electrodes in a PEC system. This allows separate production of H2 and O2 to prevent the backward reaction of the two gases to form water, which is the major setback in the PS system. Another advantage of the PEC system is the immobilization of the photocatalyst particles on a conducting substrate, enabling easy removal and separation of the photocatalyst from water.

A basic PEC water splitting configuration includes a single light-absorption component (i.e., a p-type or n-type semiconductor) and a counter electrode. Principally, the photoinduced minority carriers in the semiconductor are driven to the semiconductor–electrolyte interface due to the electric field at the interface, whereas the majority carriers are transported to the counter electrode (e.g., Pt) through an external circuit.72 In the case of n-type semiconductor, electrons as the majority carriers are transferred to the counter electrode to conduct water reduction for H2 formation, while holes (the minority carriers) oxidize water to O2 on the semiconductor surface, as illustrated in Fig. 12a. Therefore, n-type semiconductor functions as the photoanode of the PEC cell. On the contrary, a p-type semiconductor with holes as the majority carriers works as the photocathode, in which water is being reduced on the semiconductor surface but oxidized on the counter electrode (Fig. 12b). However, given that most of the photochemically stable and visible light-active semiconductors do not possess suitable thermodynamic potentials for water splitting into H2 and O2, an external bias is generally needed to drive the overall water splitting. The provision of external bias by a power supply is undesirable from the perspective of the energy consumption.

image file: c7ta04441k-f12.tif
Fig. 12 Configurations of PEC water splitting using (a) n-type semiconductor, (b) p-type semiconductor, and (c) both n-type and p-type semiconductors for a tandem system. Reprinted with permission from ref. 57. Copyright 2011 Elsevier B.V.

One ideal way to achieve visible light-driven PEC water splitting without the need of electrical energy from an external source is to construct a tandem system composed of two light-absorption components as the photoanode and photocathode, as schematically shown in Fig. 12c. Upon light illumination, both the photoanode and photocathode can be simultaneously excited to prompt spontaneous water redox reactions. Such a tandem system eliminates the stringent requirement of band potentials; the p-type semiconductor only needs to have a sufficiently negative conduction band to reduce water and the n-type semiconductor mainly requires a highly positive valence band to oxidize water. This suggests that a great variety of semiconductors with small band gaps can potentially be employed in such a PEC tandem system. However, Zhang et al. stated that the conduction band edge of the n-type semiconductor has to be more positive or of similar potential than the valence band edge of the p-type semiconductor,73 indicating that careful selection of the semiconductor pair is crucial for the construction of the tandem system.

BiVO4, which is an n-type semiconductor, has been recognized as one of the most promising photoanode materials.12,67,74 This is not only because of its small band gap of 2.4–2.5 eV and deep valence band, but also due to its conduction band edge that is only slightly below the redox potential for H+/H2, whose potential is relatively very negative compared to other visible light-active n-type semiconductors. The potential of the photogenerated electrons that are used for water reduction at the cathode is determined by the conduction band position of the photoanode. A higher (i.e., more negative) conduction band position results in a more negative photocurrent onset potential at the anode half reaction, suggesting that a lower external bias is needed and thus a higher cell efficiency can be achieved.16

3.2.3 Difference in charge separation mechanisms in powder suspension and photoelectrochemical systems. When a semiconductor is brought into contact with a solution containing a redox couple, there exists a disparate electrochemical potential across the interface. In the case of a n-type semiconductor that is immersed in a solution that contains strong oxidizing species (i.e., lower redox level (EF,redox) than the Fermi energy (EF) in the semiconductor), as illustrated in Fig. 13, electrons will flow from the semiconductor into the solution until the equilibrium is established (EF = EF,redox). The transfer of electrons away from the semiconductor causes band bending in the outer region of the semiconductor, in which the conduction and valence bands bend upward. Such a band bending develops an internal electric field (Vsc) within the semiconductor, whereby lesser electrons are present in the region of the semiconductor near the interface in comparison to the bulk. The depletion region is known as the space charge layer (SCL).
image file: c7ta04441k-f13.tif
Fig. 13 Schematic representation of the energy diagram of the semiconductor–solution interface (a) before and (b) after thermodynamic equilibration for an n-type semiconductor. ECB, EVB, and Eg represent the conduction band, valence band, and band gap energy of the semiconductor, respectively.

In the PS system, charge separation within the suspended semiconductor photocatalyst primarily relies on the SCL formation at the semiconductor–solution interface. For BiVO4 that is n-type semiconductor, the built-in electric field in the SCL drives the photogenerated electrons toward the bulk and holes are directed to the BiVO4-solution interface, promoting the separation of electrons and holes. The extent of electron–hole pair recombination within the bulk of the semiconductor (i.e., volume recombination) can therefore be alleviated by the formation of the SCL. However, the SCL thickness (dsc) is restricted by the particle size since it cannot exceed the radius of the particle (dsc < d/2).75 As such, a well-developed SCL is only present in large crystals but absent in nanosized particles, as portrayed in Fig. 14.

image file: c7ta04441k-f14.tif
Fig. 14 Formation of space charge layers in (a) large and (b) small semiconductor particles. Reproduced with permission from ref. 76. Copyright 1994 Springer-Verlag.

Additionally, since both the photogenerated electrons and holes have to be present on the surface of the same particle to conduct redox reactions, they can also be combined at the semiconductor surface (i.e., surface recombination). Therefore, the charge separation efficiency of the suspended semiconductor photocatalyst in the PS system is determined by both the volume and surface recombination,77 as schematically represented in Fig. 15a.

image file: c7ta04441k-f15.tif
Fig. 15 Schematic of the different charge separation mechanisms in the (a) PS and (b) PEC systems for water oxidation. Reprinted with permission from ref. 77. Copyright 2009 International Association for Hydrogen Energy.

On the other hand, charge separation in the PEC system is facilitated by the externally applied bias.78 Given that the photogenerated electrons are transported to the cathode and physically separated from the holes at the anode, surface recombination of these charge carriers is not an issue in the PEC system. Instead, volume recombination arising from interparticle charge transport plays a critical role in governing the charge separation (Fig. 15b).77 Since the semiconductor particles are immobilized on a conducting substrate, the photogenerated electrons have to diffuse through the thickness of the particle film to be collected at the back substrate, while holes have to migrate to the semiconductor–electrolyte interface to initiate oxidation reactions. Such migration of electrons and holes may result in their recombination at the grain boundaries.

3.3 Limitations of BiVO4

On the basis of the 2.4 eV band gap, in which up to 11% of the standard AM1.5 solar spectrum can be absorbed, BiVO4 can theoretically produce a maximum photocurrent of 7.6 mA cm−2 (assuming all incident photons with an energy greater than 2.4 eV are absorbed), corresponding to a STH conversion efficiency of 9.3%.12 However, the actual energy conversion efficiency obtained by BiVO4, reported to date, is far below its theoretical value. This is attributed to the poor electron transport, slow water oxidation kinetics, and low carrier mobility, contributing to the poor charge separation of BiVO4.

Despite relatively small effective carrier masses having been predicted,34,37 the performance of the BiVO4 film was demonstrated by Liang et al. to be limited by poor electron transport.12 This was proven by the incident photon-to-electron conversion efficiency (IPCE) obtained from backside illumination of the BiVO4 photoelectrode (illumination through the substrate) being greater than that from the frontside illumination (illumination through the electrolyte), as presented in Fig. 16a. A similar phenomenon was also observed by Zhong et al., where higher photocurrent densities were achieved with backside illumination of their BiVO4 photoanodes, indicating that electron transport is slower than hole transport in the BiVO4 films.18 The poor electron transport properties of BiVO4 may be ascribed to the disconnected VO4 tetrahedral units in the crystal structure of the material, suggesting that the photoexcited electrons in the V 3d conduction band have to hop between the VO4 tetrahedra.12 Indeed, the charge transport in BiVO4 single crystals from 250 to 400 K has been proposed to be dominated by a small polaron hopping mechanism that is associated with low electron mobility.79 Poor wave function overlap between V 3d and Bi 6p orbitals in the conduction band of BiVO4 to cause electron localization was also observed using DFT calculations.36 In contrast, a theoretical study on hole transport in BiVO4 revealed a relatively weak hole localization in this material,80 again signifying that the charge transport in BiVO4 is primarily limited by electron mobility.

image file: c7ta04441k-f16.tif
Fig. 16 (a) IPCE spectra of BiVO4 for frontside and backside illuminations. (b) Comparison of the IPCE spectra of BiVO4 and 1% W-doped BiVO4, both measured with backside illumination. Adapted with permission from ref. 12. Copyright 2011 American Chemical Society. (c) Photocurrent–voltage curves of BiVO4 (with and without the addition of H2O2 in the electrolyte), along with that of Co–Pi modified BiVO4. (d) Efficiencies of surface water oxidation catalysis (ϕox) and charge separation (ϕsep) of BiVO4 and Co–Pi modified BiVO4 as a function of applied bias. Reproduced from ref. 13 with permission (2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim). (e) ϕox and ϕsep of W-doped BiVO4 and Co–Pi modified W-doped BiVO4 as a function of applied bias. Adapted with permission from ref. 18. Copyright 2011 American Chemical Society.

One approach that has been widely proven to effectively improve the electron transport in BiVO4 involves introduction of impurity elements into the material, which is known as doping. Earlier efforts focused on BiVO4 singly doped with metal. For example, Ye et al. demonstrated that among the various metal dopants (W, Fe, B, Cu, Zn, Ti, Nb, Sn, Co, Pd, Rb, Ru, Ag, Ga, Sr, and Ir), only W-doped BiVO4 resulted in enhanced photocurrent and IPCE (obtained via frontside illumination) in comparison to undoped-BiVO4.81 Using backside illumination, significant enhancement of the IPCE spectrum of the BiVO4 photoelectrode was also observed upon incorporation of 1% W (Fig. 16b), manifesting improved electron mobility ensuing from W-doping.12 Subsequently, Luo et al. revealed that in addition to W, Mo is another metal dopant that can enhance the PEC performance of BiVO4, while Ta, Zr, Si, Ti, La, Fe, Sr, Zn, and Ag are ineffective.82 The beneficial role of W and Mo dopants to improve the electron transport in BiVO4 was found to be due to increased carrier density,81,82 which is in accordance with DFT calculations that predict the two dopants as shallow electron donors.83 Following these findings, BiVO4 codoped with Mo and W has also been prepared and shown to outperform that of BiVO4 singly doped with Mo or W.84,85

Another performance-limiting factor for BiVO4 is slow hole transfer across the semiconductor–electrolyte interface (i.e., slow water oxidation kinetics). This is evidenced by about threefold enhancement of the photocurrent density of BiVO4 in the presence of H2O2 (Fig. 16c),13,14 which is a highly efficient hole scavenger that has oxidation kinetics at least 10 times faster than those of H2O attributed to its lower reduction potential of +0.68 V with respect to +1.23 V vs. RHE for H2O.18 The slow water oxidation kinetics of BiVO4 can be overcome by modifying the surface with water-oxidation cocatalysts. While various transition metal catalysts such as IrO2, RuO2, and RhO2 are known as good oxygen evolution catalysts,82,86,87 materials based on earth-abundant elements are favorable. With regard to this, the potential of Co–Pi,13,14,18,88–91 Co3O4,92,93 CoO,94 FeOOH,95–97 and NiOOH97 as the earth-abundant water-oxidation cocatalysts for BiVO4 has recently been demonstrated. Of these cocatalysts, Co–Pi is the most commonly studied cocatalyst to improve the water oxidation kinetics of BiVO4 in both the PEC and PS systems. Remarkably, the photocurrent densities of Co–Pi deposited BiVO4 photoelectrodes were found to be comparable to those of BiVO4 performing H2O2 oxidation (Fig. 16c),13,14 indicating that Co–Pi is highly effective in capturing holes to promote water oxidation of BiVO4 photoanode, whereby near-complete suppression of surface electron–hole recombination can be achieved.18 In spite of that, a significant fraction (60–80%) of the photogenerated charge carriers still succumbs to recombination, as shown in Fig. 16d.

In light of the different roles of doping and water-oxidation cocatalyst in improving BiVO4 performance, as described above, Zhong et al. studied the performance of W-doped BiVO4 interfaced with Co–Pi catalyst.18 In the tested potential range of 0.4–1.5 V vs. RHE (Fig. 16e), although catalytic efficiencies of close to 100% for surface water oxidation (ϕox) on W-doped BiVO4 can be achieved after Co–Pi modification, the resulting charge separation yields (ϕsep) still remain significantly low (approximately 0.1–0.3 (i.e., 70–90% of the electron–hole pairs are recombined)). The results thus suggest bulk carrier recombination as the primary bottleneck limiting the efficiency of BiVO4. Further development of BiVO4 materials should be focused on improving charge separation efficiency.

The poor charge separation of BiVO4 was revealed by Abdi et al. to be due to its intrinsic exceptionally low carrier mobility of ∼4 × 10−2 cm2 V−1 s−1 (under ∼1 sun illumination conditions).15 However, the poor carrier mobility is compensated by the unexpectedly long carrier lifetime (40 ns) and diffusion length (70 nm), which are responsible for the high quantum efficiencies reported for BiVO4. Table 1 compares the dynamic properties of BiVO4 to those of typical metal oxides, such as Fe2O3, Cu2O, and WO3. Clearly, the carrier mobility of BiVO4 is at least 1–2 orders of magnitude lower than that of the other metal oxides, while its carrier lifetime is 1–3 orders of magnitude longer.15

Table 1 Comparison of the carrier mobility, lifetime, and diffusion length of BiVO4 with those of several typical metal oxides. Reprinted with permission from ref. 15. Copyright 2013 American Chemical Society
Material Carrier mobility, μ [cm2 V−1 s−1] Carrier lifetime, τ Diffusion length, L [nm]
Fe2O3 0.5 3 ps 2–4
Cu2O 6 40 ps 25
WO3 10 1–9 ps 150–500
BiVO4 0.044 40 ns 70

4. Alternative strategies for improving the charge kinetics of BiVO4

As described in Sections 2 and 3, BiVO4 possesses many features (e.g., visible band gap, deep valence band, and long carrier lifetime) that make it a highly promising oxide semiconductor photocatalyst for solar water oxidation and organic degradations. However, full exploitation of the valuable features of BiVO4 is hampered by its intrinsic poor charge kinetics such as charge separation, transport, and transfer. While state-of-the-art BiVO4 often incorporate n-type dopant (i.e., Mo and W) and water-oxidation cocatalyst to overcome the respective electron transport and water oxidation kinetics limitations of BiVO4, the performance of BiVO4 is still greatly hindered by poor charge separation, particularly in the bulk. Exploration of other strategies, aside from doping and surface modification with water-oxidation cocatalyst, has shown the potentials of (1) crystal facet engineering, (2) composite structure with graphitic carbon material, (3) annealing treatment, and (4) nanoscaling to overcome the shortcomings of this material. This section aims to provide an overview of strategies 1–4, whereby relevant findings from our most recent studies will also be incorporated and discussed herein.

4.1 Crystal facet engineering

Given that photochemical redox reactions essentially proceed on the surface, the physicochemical and electronic properties of the exposed surfaces of a photocatalyst, which are dependent on the surface atomic arrangement and coordination, play a vital part in determining the reactivity and activity of the material. As such, design and morphological control of crystal facets is a strategy commonly employed to optimize the performance of various semiconductor photocatalysts.

On the one hand, the photoreactivity of different crystal facets exposed on a photocatalyst was demonstrated to vary because of the different surface atomic structures, surface electronic structures, and selective reactant adsorption.98–103 These have been extensively studied for the benchmark oxide photocatalyst, TiO2. Although anatase TiO2 crystals obtained without addition of a surfactant or directing agent in the synthesis processes are dominated by the thermodynamically stable {101} facets with minority {001} surfaces,104 both theoretical and experimental studies indicated that the latter is relatively more reactive than the former.105–108 The higher reactivity of the {001} surface was conventionally ascribed to it having a higher percentage of surface undercoordinated Ti5c atoms (100%) in comparison to that of {101} with 50% Ti5c and 50% Ti6c atoms.99,109 As opposed to the conventional understanding, Pan et al. revealed that the photoreactivity order of the different facets on anatase TiO2 is {001} < {101} < {010} for ˙OH radical generation and H2 evolution, accredited to the cooperative attributes of surface unsaturated Ti5c atom density and facet-specific conduction band potential.99

Likewise, crystal facet-engineered bismuth-based ternary metal oxides such as Bi2WO6,110 Bi2MoO6,111,112 and BiVO4[thin space (1/6-em)]113–115 were found to exhibit enhanced photocatalytic activities, all of which were related to the {010} facets. In the case of BiVO4 in particular, Zhang et al. reported that BiVO4 nanosheets with a preferred (010) surface orientation, synthesized in the presence of sodium dodecyl benzene sulfonate as the morphology-directing agent, showed faster degradation of N,N,N′,N′-tetraethylated rhodamine than that of bulk BiVO4.113 Xi et al. demonstrated that BiVO4 nanoplates with exposed {010} facets displayed enhanced degradation of rhodamine B and photocatalytic O2 generation in comparison to BiVO4 nanorods and microcrystals.114 By controlling the exposure extent of BiVO4 (040) surface (corresponding to {010} facets) using different amounts of TiCl3 as the directing agent, Wang et al. revealed a correlation between the photocatalytic water oxidation activity of BiVO4 with the {010} facets, as exhibited in Fig. 17.115

image file: c7ta04441k-f17.tif
Fig. 17 Correlation between the O2 evolution rate with the exposure extent of the {010} facet on BiVO4. Adapted from ref. 115 with permission (2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim).

On the other hand, the presence of different crystal facets on a photocatalyst polyhedron was demonstrated to drive spatial separation of photogenerated electrons and holes on the surface, which was proposed to be due to the potential difference in the conduction and valence band energy levels of different facets arising from the varied surface electronic structures associated with them.116,117 Such a phenomenon was first observed by Ohno et al. on TiO2, in which formation of Pt deposits via photoreduction of Pt4+ primarily proceeded on the {110} facets of rutile and {101} facets of anatase, whereas PbO2 particles resulting from photooxidation of Pb2+ were seen mainly on the {011} facets of rutile and {001} facets of anatase.116 This study thus infers that rutile {110} and anatase {101} facets are the active reduction sites, while rutile {011} and anatase {001} facets function as the active oxidation sites. Using a single-molecule imaging and kinetic analyses of the fluorescence of redox-responsive fluorogenic dyes, Tachikawa et al. also demonstrated that the {101} facets of anatase TiO2 are a more efficient reduction site than the {001} facets due to the preferential electron trapping on the {101} facets.118

The facet-selective metal and metal oxide depositions, indicative of spatial separation of the photogenerated electrons and holes, have also been demonstrated on other photocatalysts.117,119 Particularly, Li et al. showed that the photoreduction of Au3+, Pt4+, and Ag+ ions to metal deposits and the photooxidation of Pb2+ and Mn2+ to metal oxide deposits took place selectively on the {010} and {110} facets of BiVO4 crystal, respectively,117 as depicted in Fig. 18. The preferential occurrences of photoreduction reactions on the {010} facets and photooxidation on the {110} facets were attributed to the relatively lower conduction and valence band energy levels of the {010} facets (Fig. 19a), which was supported by DFT calculations.117 The difference in surface energy levels suggests that electron transfer from {110} to {010} facets is thermodynamically feasible, leading to electron accumulation on the {010} facets and hole accumulation on the {110} facets (Fig. 19b) for them to function as the redox functional facets.

image file: c7ta04441k-f18.tif
Fig. 18 SEM images of the simultaneous depositions of metal (Au, Pt, or Ag) and metal oxide ((a) MnOx or (b) PbO2) on the surface of dual-faceted BiVO4. Reprinted with permission from ref. 117. Copyright 2013 Nature Publishing Group.

image file: c7ta04441k-f19.tif
Fig. 19 Schematic diagrams of (a) different conduction and valence band energy levels of the {010} and {110} facets of BiVO4 (adapted with permission from ref. 117. Copyright 2013 Nature Publishing Group.) and (b) spatial separation of photogenerated electrons and holes on the two facets (reproduced from ref. 93 with permission from The Royal Society of Chemistry).

On the basis of the different energy structures of the {010} and {110} facets of BiVO4, Li et al. designed two types of BiVO4–TiO2 heterojunctions by interfacing TiO2 with either the {010} or the {110} facets of BiVO4.120 They revealed that the two heterojunctions (TiO2–BiVO4 {010} and TiO2–BiVO4 {110}) have different activities for PEC photocurrent generation and photocatalytic degradation, owing to the different energy band alignments between TiO2 and the two facets of BiVO4. Meanwhile, the roles of the {010} and {110} facets as the respective reduction and oxidation functional facets of BiVO4 are further substantiated in numerous studies.93,121 For instance, Li et al. presented that the reduction cocatalysts (Pt, Au, or Ag) and oxidation cocatalysts (MnOx or Co3O4) could be selectively deposited onto the {010} and {110} facets of BiVO4 by sequential photoreduction and photooxidation deposition methods.93 The deposition of cocatalysts on the right facets of the BiVO4 (i.e., reduction cocatalyst on the electron-rich {010} facets and oxidation cocatalyst on the hole-rich {110} facets) was shown to be highly efficient to consume the photogenerated electrons and holes, resulting in enhanced photocatalytic activities with respect to the BiVO4 with randomly distributed reduction and oxidation cocatalysts. For a Z-scheme comprised of the hybridization of platinized photosystem I (PSI) protein and BiVO4, Kim et al. also demonstrated that photoreduction of Au3+ and Ag+ ions to the respective metal deposits occurred preferentially on the reduction functional {010} facets of BiVO4 to allow selective extraction of electrons from BiVO4 to be transferred to the PSI for H2 generation reaction.121

Although the phenomenon of charge separation on the different facets of a semiconductor can be deduced from the results of selective photodepositions of metal and metal oxide and the DFT calculations, direct evidence for the anisotropic charge transfer/separation on the surface of semiconductor is lacking. For this purpose, Zhu et al. employed spatially resolved surface photovoltage spectroscopy (SRSPS) on single dual-faceted BiVO4 crystal and observed a highly anisotropic photogenerated hole transfer to the {110} facets. This was indicated by the 70 times stronger SPS signal on the {110} facets compared to that of the {010} facets.122 Tachikawa et al. studied the charge carrier dynamics of dual-faceted BiVO4 using single-particle photoluminescence (PL) spectroscopy and discovered that trapped holes are favorably located on the {110} facets of the BiVO4 crystal, while electrons are uniformly distributed over the crystal.123

The findings from the abovementioned experimental results of metal/metal oxide selective photodeposition, theoretical DFT calculations and in situ microscopic techniques consistently infer that photogenerated holes are preferentially transferred to and accumulated on the {110} facets of BiVO4 relative to the {010} facets, allowing {110} to function as the active oxidation site. This may seem to be contradicting with the observations of the previously mentioned studies that suggest a correlation between the photooxidation performances of BiVO4 (including degradation of organic molecules and water oxidation for O2 evolution) with the {010} facets,113–115 which are supposedly the reduction functional facet. However, it has to be noted that the BiVO4 samples used for comparison in the said studies were of multiple morphological differences (e.g., particle size, types of exposed crystal facets, and surface area). Therefore, the observed enhanced activity of BiVO4 with preferentially exposed {010} facets should not be solely attributed to the {010} facets.

A theoretical study conducted by Yang et al. has indicated that water oxidation on the {010} facets of BiVO4 is more favorable than that on the {110} facets because {010} has higher charge carrier mobility, easier water adsorption, and lower energy barrier.38 Despite these valuable computational calculations, the crystal facet effects on the charge transfer efficiency of BiVO4 to the reactants (electron donor/acceptor), which may be one of the underlying reasons for the crystal facet-dependent photoactivity of BiVO4, remain poorly understood. This motivated us to investigate the influence of {010}/{110} relative exposure extent on the photoactivity of BiVO4 with respect to the charge transfer ability of the material.

Recently, we demonstrated that the population of photogenerated electrons and holes on BiVO4 are proportional to the respective surface areas of the {010} and {110} facets and the relative exposure extent of these two facets controls the charge transfer ability and photoactivity of BiVO4.19 Two dual-faceted BiVO4's with comparable size and surface area but distinctly different dominant exposed facets, one which is {010}-dominant and the other {110}-dominant (Fig. 20a), were prepared. The BiVO4 with dominating {010} facets unambiguously showed better photooxidation activities than the {110}-dominant BiVO4, as evidenced in photocatalytic water oxidation and 2,4-dichlorophenoxyacetic acid photodegradation shown in Fig. 20b. Steady-state PL measurements in the presence of charge scavengers revealed opposing behaviors of the two BiVO4 (Fig. 20c): PL intensity of {010}-dominant BiVO4 was more significantly quenched in response to hole-scavenging CH3OH when compared to that of electron-scavenging AgNO3, while the reverse scenario was observed for {110}-dominant BiVO4. This suggested that {010}-dominant BiVO4 has relatively more electrons than holes available on the surface, while the {110}-dominant BiVO4 surface has more holes than electrons. A greater extent of electron trapping was also found in {110}-dominant BiVO4 based on time-resolved PL analysis and flat band potential estimation from the Mott–Schottky relationship, attributed to the fact that longer distance electrons have to diffuse from the bulk of the thick BiVO4 particle to the {010} surface. As illustrated in Fig. 20d, the restricted numbers of electrons on the small {010} surface and greater degree of electron trapping led to inefficient electron extraction and increased charge recombination, resulting in the poor photooxidation activities of {110}-dominant BiVO4 despite the presence of a large oxidation functional surface. In contrast, enlargement of the {010} reduction functional facet relative to that of the {110} oxidation functional facet (represented by the {010}-dominant BiVO4) is essential to facilitate electron transfer on the BiVO4 surface for effective photooxidation performance.

image file: c7ta04441k-f20.tif
Fig. 20 (a) SEM images of dual-faceted BiVO4 samples, one is {010}-dominant and the other {110}-dominant. (b) Photocatalytic activities of the two BiVO4 samples in (left) water oxidation and (right) 2,4-dichlorophenoxyacetic acid under visible light illumination. (c) Different quenching degrees of the PL band edge emission band of the two BiVO4 samples following the addition of varying concentrations of AgNO3 and CH3OH as the respective electron and hole scavengers. Insets show the representative steady-state PL spectra of each BiVO4 in the presence of charge scavengers. (d) Schematic illustration of the different charge transfer abilities of the BiVO4 samples with different relative exposure extents of {010} and {110} facets. Reprinted with permission from ref. 19. Copyright 2016 American Chemical Society.

4.2 Compositing with graphitic carbon material

Coupling of semiconductor with a conductor (i.e., carbon material) is a promising approach shown to be effective in enhancing the charge separation and transport properties of the semiconductor. Owing to the abundance of delocalized electrons from the conjugative π-system, graphitic carbon materials such as fullerenes,124,125 carbon nanotubes,126,127 and graphene128 are good in accepting and shuttling the photogenerated electrons from semiconductor photocatalysts and hence effectively separating the electron–hole pairs. However, graphene has attracted more immense research interest than the other two carbon allotropes because of its unique two-dimensionality, contributing to its exceptionally high specific surface area (∼2600 m2 g−1)129 that is readily accessible for surface adsorption of chemical molecules (i.e., organic reactants). The single-atom-thick nature of the graphene sheet also allows it to be optically transparent, which is a highly desirable characteristic to minimize the light-shielding-effect when it is being coupled with a semiconductor that works as the main photon absorber.

Among the numerous preparation methods of graphene reported to date (e.g., mechanical cleavage,130 epitaxial growth,131 and chemical vapor deposition132), chemical exfoliation of graphite to graphene oxide (GO) followed by reduction via a chemical, thermal, electrochemical, photocatalytic, sonochemical, photothermal, or microwave-reduction method is the most widely employed approach due to the simplicity, reliability, low cost, and large scalability of the process.128 In fact, the wide-ranging synthesis methods of graphene-based semiconductor photocatalysts have shown GO as the chief precursor of graphene.128 While the presence of various oxygen functional groups (i.e., hydroxyl, epoxy, carboxyl, and carbonyl) on the surface of GO offers potential for functionalization and provides reactive sites for the nucleation and growth of nanoparticles, partial restoration of the conjugated sp2 network via reduction to obtain reduced graphene oxide (RGO) is essential to resurrect the conductivity of GO.

The pioneering work on the synthesis of graphene-based semiconductor photocatalyst was reported by Williams et al. in 2008.133 Apart from showing the practicability of in situ photocatalytic reduction of GO by TiO2 (under UV light) to obtain RGO–TiO2 nanocomposite, the work also elucidated the mechanism of electron transfer pathway from photoexcited TiO2 to GO, showcasing the electron-accepting attribute of graphene or graphene derivatives. A subsequent study by Lightcap et al. demonstrated the stepwise transfer of photogenerated electrons from TiO2 to GO for RGO formation and then from RGO to Ag+ ions, resulting in the production of a RGO–Ag–TiO2 composite. The observation of the distinct Ag nanoparticle anchored site from that of the TiO2 proved the electron-shuttling capability of RGO.134 Such excellent electron-accepting and shuttling properties of RGO are generally associated with the longer carrier lifetime, lower charge recombination, and improved photoactivities of various RGO-based oxide semiconductors, including TiO2,135,136 WO3,137 ZnO,138 and SnO2[thin space (1/6-em)]139 to name a few.

The study on RGO–BiVO4 composite was first conducted by Ng et al. via photocatalytic reduction of GO by BiVO4 under visible light.140 The resulting RGO–BiVO4 exhibited tenfold enhancement in PEC photocurrent compared to BiVO4, as displayed in Fig. 21a. Interestingly, the photocurrent generated by RGO–BiVO4 under visible light was higher than that produced by UV-excited TiO2. IPCE action spectra of BiVO4 and RGO–BiVO4 (Fig. 21b) were examined to confirm that the photocurrent generation only occurs upon photoexcitation of BiVO4 and therefore eliminates the possibilities of improved photocurrent of RGO–BiVO4 attributed to the enhanced light absorption ensuing from RGO addition and contribution of extra electrons from the excited RGO. On the contrary, the considerably increased photoconversion efficiency of RGO–BiVO4 was proven due to slower charge recombination rate, as evidenced by the longer transient time constant (the time at which ln[thin space (1/6-em)]D = −1, where D = (I(t)I(st))/(I(in)I(st)), I(t) is the photocurrent at a time t, I(in) the photocurrent at time t = 0, and I(st) the steady-state photocurrent) of RGO–BiVO4 in comparison to that of BiVO4 (Fig. 21c). This study thus suggests that the incorporation of RGO primarily provides a low-resistant electron pathway to facilitate electron transport from the photoexcited BiVO4 to the conductive electrode (schematically illustrated in Fig. 21d), enhancing charge separation and PEC photocurrent of RGO–BiVO4. Following this report, the superior performance of RGO–BiVO4 over bare BiVO4 in different photocatalytic applications such as organic degradations,141–145 water oxidation,146 and reduction of CO2 and Cr(VI)147 were also reported. Given the profound interest on graphene, coupling of BiVO4 with the other graphitic carbon allotropes is relatively less investigated, whereby only limited studies on carbon nanotube–BiVO4 composites were reported to date.148–150 Akin to RGO–BiVO4, the carbon nanotube–BiVO4 composites were also found to outperform their bare BiVO4 counterparts.

image file: c7ta04441k-f21.tif
Fig. 21 (a) Current–potential curves of BiVO4, RGO–BiVO4, and TiO2. Both BiVO4 and RGO–BiVO4 were illuminated by visible light, whereas those of TiO2 were by UV light. (b) IPCE and UV-vis diffuse reflectance spectra of BiVO4 and RGO–BiVO4. (c) Plots of normalized transient photocurrent of BiVO4 and RGO–BiVO4 as a function of time. (d) Schematic illustration of the role of RGO in facilitating electron transport in RGO–BiVO4 material. Adapted from Ng et al. Reproduced with permission from ref. 140. Copyright 2010 American Chemical Society.

Notably, research available with respect to graphene-based semiconductors predominantly focuses on highlighting graphene-mediated enhanced photoactivity and streamlining their preparation methods. Considering that the improved charge separation of graphene–semiconductor composite fundamentally relies on the charge transfer at the interface, a few studies have been devoted to maximizing the interfacial contact between graphene and the semiconductor via the construction of a graphene-wrapped semiconductor structure.151,152 For instance, Wang et al. reported the production of RGO–BiVO4 composites with the BiVO4 polyhedrons fully covered with RGO sheets, resulting in enhanced charge separation and high activity of the material for methylene blue photodegradation,152 as schematically depicted in Fig. 22a. Despite this, the influences of the semiconductor morphological changes (i.e., size, shape, and exposed crystal facets) on its charge interactions with graphene are scarcely studied. The photocatalytic activities of various semiconductors have been extensively shown to be affected by their morphological properties. Understanding the relationships between graphene–semiconductor charge interactions and the semiconductor's morphological properties is indispensable to rationalize the design of highly functionalized graphene–semiconductor materials.

image file: c7ta04441k-f22.tif
Fig. 22 Schematic illustrations of (a) RGO-wrapped BiVO4 for enhanced charge separation efficiency and improved activity for photocatalytic degradation of methylene blue (MB) (Reprinted with permission from ref. 152. Copyright 2014 American Chemical Society.) and (b) efficient electron and hole separation via electron transfer through RGO and hole migration through the internal structure of BiVO4 for improved photodegradation of rhodamine B (RhB) (reproduced from ref. 153 with permission from The Royal Society of Chemistry).

As previously described in Section 4.1, BiVO4 with well-defined {010} and {110} facets have been extensively suggested to have efficient charge separation since photogenerated electrons from the {110} facets with relatively higher (i.e., more negative) conduction band potential are preferentially transferred to the {010} facets of lower conduction band potential, while holes move in the opposite direction. However, recombination between electrons and holes is inevitable during their transition between the two facets via a single channel in the BiVO4 particle. To further optimize the charge separation efficiency of the dual-faceted BiVO4, Feng et al. proposed to incorporate RGO into the faceted BiVO4 as an additional channel that only allows transport of electrons, realizing dual-channel charge separation with high selectivity.153 As portrayed in Fig. 22b, the RGO sheet was suggested to serve as an external channel for electron migration from the {110} surface to the electron-accumulating {010} surface, whereas hole transfer in the opposite direction happens in the internal structure of BiVO4. The existence of such thermodynamically feasible distinctive migration channels of electrons and holes was supported by the emergence of a new photoluminescence emission peak in the RGO–BiVO4 sample, whose energy corresponds to the difference of the energy levels between the conduction band of {110} facets and the valence band of the {010} facets.153 This study triggered our interest to probe the facet-dependent charge interactions between graphene and BiVO4, which may be helpful to extend the application of crystal facet engineering in the fabrication of graphene–semiconductor materials.

Recently, we reported that the degree of photocurrent enhancement for the dual-faceted BiVO4 in the presence of RGO is correlated to the exposure extent of {010} facets, attributed to the different electronic properties between graphene/BiVO4 {010} and graphene/BiVO4 {110} interfaces.20 Three BiVO4 samples with different {010}/{110} relative exposure extents were synthesized and separately coupled with RGO. Note that the exposure extent of {010} relative to {110} on the BiVO4 samples follows an increasing trend of BiVO4_1.00 M < BiVO4_0.75 M < BiVO4_0.50 M, as supported by the increasing B/A degree of truncation calculated for the samples (Fig. 23a). While the PEC photocurrents of all BiVO4 samples were observed to be improved after incorporation of RGO, the degree of enhancement was found to consistently follow a similar trend of RGO–BiVO4_1.00 M < RGO–BiVO4_0.75 M < RGO–BiVO4_0.50 M under different applied potentials, as shown in Fig. 23b. Electrochemical impedance spectroscopy (EIS) analysis (Fig. 23c) suggest that the greater extent of enhancement in RGO–BiVO4 was aroused from the decrease in charge transfer resistance, indicating the beneficial role of the {010} facets to expedite electron transfer from BiVO4 to RGO. DFT calculations on the electronic properties of the interfaces between graphene and the two facets of BiVO4 confirmed that charge transport is easier through the graphene/BiVO4 {010} interface because it has a relatively smaller Schottky barrier of 0.09 eV than graphene/BiVO4 {110} interface (0.71 eV). The graphene/BiVO4 {010} interface was also found to have a higher binding energy than that of the graphene/BiVO4 {110} interface, as reflected by the shorter surface distance between graphene and BiVO4 {010} in comparison to graphene and BiVO4 {110} (Fig. 23d). Density of states (DOS) analysis revealed that while a gap of about 0.3 eV is present between the band edge of BiVO4 {110} facets and the Fermi level of graphene to render a semiconducting nature of the graphene/BiVO4 {110} interface, the graphene/BiVO4 {010} interface with negligible gap is metallic, affirming that the latter is the more favorable pathway to conduct charges.

image file: c7ta04441k-f23.tif
Fig. 23 (a) SEM images of dual-faceted BiVO4 samples with increasing {010}/{110} relative exposure extent (BiVO4_1.00 M < BiVO4_0.75 M < BiVO4_0.50 M), as evidenced by the increase of the average B/A degree of truncation. (b) Degree of photocurrent enhancement for each BiVO4 sample after incorporation of RGO, measured under different applied potentials. (c) EIS Nyquist spectra of the BiVO4 samples and their corresponding RGO–BiVO4 samples, obtained at 0 V vs. Ag/AgCl under light illumination. (d) Optimized separation at graphene/BiVO4 {010} and graphene/BiVO4 {110} interfaces and their corresponding DOS. Reproduced from ref. 20 with permission (2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim).

4.3 Annealing treatment

Annealing has been generally employed as a post-synthesis treatment, which is useful to promote crystal grain growth and to tailor the phase composition of oxide photocatalysts. The increase of grain size results in the improve of crystallinity,154–157 whereas the temperature dependence phase transformation allows such heat treatment to expedite formation of the more thermodynamically stable structure of the materials.29,158,159 Apart from that, annealing-mediated morphological changes such as particle size and shape are also commonly observed due to the minimization of surface free energy in response to the provision of thermal energy.155,157,160 These changes in the physicochemical properties affect the charge generation, separation, and transport behaviors of the oxide photocatalysts, indicating that the intrinsic poor charge properties of the materials can be addressed in part by annealing treatment.

Although earlier studies on thermal annealing were predominantly performed in an oxygen-rich atmosphere (e.g., air and O2 gases), more recent studies have extended the heat treatment to an oxygen-deficient environment such as nitrogen (N2), argon (Ar), and H2 gases.3,161–163 Annealing under both oxygen-rich and deficient conditions was demonstrated capable to tailor the physicochemical properties of the treated metal oxides as mentioned before. However, heat treatment in an oxygen-deficient atmosphere was revealed to induce the formation of oxygen defects known as oxygen vacancies, which on the other hand are inhibited in an oxygen-rich atmosphere. For instance, thermal treatment of N-doped TiO2 particles in N2 and air atmospheres was demonstrated by Yamada et al. to have opposite effects on oxygen vacancy formation in the material: increased after N2 treatment but decreased after air treatment.161 Likewise, Mtangi et al. reported that a new defect possibly related to oxygen vacancy was exhibited by ZnO after H2 and Ar annealing, but absent after O2 annealing.163 The formation of oxygen vacancies contributes to the self-doping effect, which can tune the optical and electronic properties of metal oxides.164

Among the oxygen-deficient gases, tremendous efforts have been focused on H2 annealing (i.e., hydrogenation) because of its strong reducing ability shown to be highly efficient to instigate oxygen vacancy formation in various metal oxides. Black TiO2 nanocrystals with a highly disordered surface covering the TiO2 core obtained via hydrogenation of TiO2, were first reported by Chen et al.165 Midgap electronic states were revealed to be created in the black TiO2, as supported by the significant reduced band gap from 3.30 eV of the untreated, white TiO2 to ∼1.54 eV for the treated TiO2. On the other hand, subjection of TiO2,166 ZnO,167 and WO3[thin space (1/6-em)]168 to H2 treatment (based on the separate studies conducted by Li and coworkers) consistently showed an increase of oxygen vacancies in the respective treated metal oxides.3 The introduction of oxygen vacancies, which are known as the shallow donors of metal oxides, was demonstrated to result in improved donor density, charge transport properties, and photoactivity of all the hydrogenated samples.

A more recent study of Li and coworkers has also validated the beneficial role of hydrogenation on the PEC performance of BiVO4,169 similar to that reported for other metal oxides. The visible light absorption ability of BiVO4 was enhanced upon hydrogenation, suggesting the presence of midgap states in the hydrogenated BiVO4 (H-BiVO4) samples. The creation of oxygen vacancies in H-BiVO4 was confirmed by the increase of surface adsorbed oxygen species Oads (represented by the peak located at a higher binding energy relative to the O 1s peak corresponding to the surface lattice oxygen Olatt of BiVO4), evolution of V4+ species, and negative shift of the Bi 4f peaks (indicative of the reduction of Bi3+ in BiVO4 to lower oxidation states), as depicted in Fig. 24a–c, respectively. Surface oxygen vacancies are positively charged and therefore inclined to adsorb negatively charged species such as OH, O, O2−, and O22−.170 Increased adsorption of these negative species would lead to the formation of V4+ species due to electron transfer to the cation V5+ to maintain the electroneutrality of the oxide material.171 As such, the Oads and V4+ species are associated with oxygen vacancies in BiVO4.172,173 In addition, partial reduction of Bi3+ has also been correlated with oxygen vacancies.174 The oxygen vacancy formation was proved to increase the donor density of the H-BiVO4 by one order of magnitude, as evidenced by the considerably smaller slope of its Mott–Schottky plot in comparison with that of BiVO4 (Fig. 24d). Such increase of donor density was consistent with the DFT calculations, which predicted oxygen vacancy as a shallow donor for BiVO4.169,175 EIS measurements manifested that the Nyquist plot arc radius of H-BiVO4 (Fig. 24e) was greatly reduced, signifying a smaller charge transfer resistance at the semiconductor–electrolyte interface. The enhanced charge transport ability attributable to the increased donor density of H-BiVO4 thus explained its improved PEC photocurrent with respect to BiVO4.

image file: c7ta04441k-f24.tif
Fig. 24 (a) O 1s, (b) V 2p, and (c) Bi 4f XPS spectra of BiVO4 (black) and H–BiVO4 samples heated at 300 (red) and 400 °C (blue). (d) Mott–Schottky and (e) Nyquist plots of BiVO4 and H-BiVO4 (300 °C) obtained under dark conditions. Adapted with permission from ref. 169. Copyright 2013 American Chemical Society.

A later study on the effects of H2 treatment on the structural, optical, electronic, and PEC properties of BiVO4 conducted by Singh et al. also presented findings similar to those reported by Li and coworkers (as detailed above).171 In light of H2 treatment as a straightforward approach to improve the charge transport properties of metal oxides, it has also been combined with the deposition of different water-oxidation cocatalysts as a synergistic approach to develop highly efficient BiVO4 photoanodes.176,177

H2, which is a flammable gas, has to be handled with extra care during the experimental setup for hydrogenation treatment. Furthermore, the strong reducing ability of H2 limits the temperature range applicable for hydrogenation treatment to avoid undesirable reduction effects. For instance, unfavorable reduction of the SnO2 layer on the fluorine-doped tin oxide (FTO) substrate to metallic Sn was observed for the TiO2/FTO photoelectrode hydrogenated at temperatures higher than 450 °C.166 This then restrains the extent of improvement achievable for the physicochemical properties (e.g., crystallinity) for the treated material, which are generally in proportion to the annealing temperature. Hence, N2 or Ar gases may be a safer substitute for H2 to provide an oxygen-deficient atmosphere for the annealing process.

However, studies on annealing treatment of metal oxides in N2 and Ar atmospheres with clear elucidation of the effects on the structural, optical, and electronic properties of the treated materials have been scarce. Recently, annealing of nanoporous BiVO4 photoanodes in a N2 flow was demonstrated to result in enhanced photon absorption, carrier density, and mobility because of the combined effects of N-doping and oxygen vacancy formation.178 Following this study, we examined the practicability of Ar annealing to induce changes on the properties of BiVO4,22 similar to those obtained via H2 and N2 annealing. Exploration of the impacts of Ar annealing on the properties of oxide materials, particularly oxygen vacancy formation, is imperative because Ar is an oxygen-deficient gas prevalently used in various experimental setups operated at elevated temperature due to its inertness.

In our study, dual-faceted BiVO4 was subjected to Ar annealing as a post-synthesis treatment in the temperature range of 300–700 °C.22 Evidently, photoactivities of the BiVO4 for photocatalytic O2 evolution (Fig. 25a) and PEC photocurrent generation (Fig. 25b) were shown to be affected by the annealing temperature. X-ray diffraction (XRD), Raman, and UV-vis analyses indicated that while treatment at 300 °C has little effect, higher temperatures of 500 and 700 °C enhance the crystallinity, induce greater local structure distortion, and reduce the band gap energy, all of which are advantageous for charge separation in the treated BiVO4. Annealing at 500 °C or higher was also found to introduce new midgap states in the treated BiVO4, as suggested by the negative shift of the surface photovoltage spectroscopy (SPS) photoonsets of the respective samples with reference to that of untreated BiVO4 (Fig. 25d). Detection of enhanced electron paramagnetic resonance (EPR) signal related to V4+ species at g = 1.977 proved that these new midgap states are indeed oxygen vacancies (Fig. 25e). However, the charge separation and transport properties of the dual-faceted BiVO4 were demonstrated to be optimized at 500 °C treatment, as reflected by its best photoactivity. Further increase of the annealing temperature to 700 °C was found to be undesirable due to creation of surface bismuth vacancies, as verified by the emergence of higher-binding-energy shoulders at XPS Bi 4f peaks associated with BiVO4 (Fig. 25f). As shallow electron acceptors, bismuth vacancies cause electron trapping in the 700 °C-treated BiVO4, as corroborated by the inversion of the SPS signal displayed in Fig. 25c. Additionally, diminution of the well-developed {010} and {110} facets also occurred upon 700 °C annealing, deteriorating the charge separation efficiency of the material (Fig. 25g).

image file: c7ta04441k-f25.tif
Fig. 25 (a) Photocatalytic O2 evolution and (b) photocurrent generation from dual-faceted BiVO4 and the samples annealed at 300, 500, and 700 °C under visible light. The treated samples are denoted as ArX, where X represents the annealing temperature. (c) SPS spectra and (d) variation in the SPS photoonsets with respect to annealing temperature. (e) Room temperature EPR and (f) XPS Bi 4f spectra of all the samples. (g) SEM images of Pt and MnOx-loaded on (left) dual-faceted BiVO4 (representative of the untreated, 300, and 500 °C-treated BiVO4) and (right) 700 °C-treated BiVO4. Presence of well-developed facets is demonstrated to be essential for effective charge separation as reflected by the selective photodepositions of Pt on the {010} and MnOx on the {110} surfaces as mainly seen on the dual-faceted BiVO4. Reproduced from ref. 22 with permission (2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim).

4.4 Nanoscaling

Upon photoexcitation, the photoinduced electrons and holes have to migrate from the bulk of the semiconductor photocatalyst to the surface to take part in redox reactions. In the absence of an external field, the charge carriers move by diffusion. The average distance a carrier moves from the point of generation and before recombination is defined as the diffusion length (L) which relies on the lifetime τ and mobility μ of the carrier.15
image file: c7ta04441k-t1.tif(6)
where D is the carrier diffusion coefficient that is related to μ by the Einstein relation:179
image file: c7ta04441k-t2.tif(7)

In order to optimize the collection of both electrons and holes at the semiconductor surface, a semiconductor film thickness and the dimensions of the suspended semiconductor have to be in the same range as the diffusion lengths of the charge carriers,75 as illustrated in Fig. 26. Clearly, this can be achieved by nanostructuring the surface of the film (Fig. 26b) and downsizing the suspended particle to nanoscale (Fig. 26c).

image file: c7ta04441k-f26.tif
Fig. 26 Charge collection in (a) flat films, (b) nanostructured films, and (c) suspended particles. Le and Lh denote the respective diffusion lengths of electrons and holes, whereas d represents the film or particle thickness. Adapted with permission from ref. 75. Copyright 2015 Springer International Publishing Switzerland.

The nanostructuring/nanoscaling approach is useful to overcome the limitations of oxide photocatalysts such as short carrier lifetimes and low mobility.75 In view of this, nanostructured morphologies are extensively adopted in studies related to BiVO4 photoanodes,74,82,96,97,180 which have been shown to suffer from poor carrier mobility. For instance, Kim and Choi fabricated a nanostructured BiVO4 electrode composed of particles (mean size 76 ± 5 nm) smaller than its hole diffusion length (∼100 nm).97 The resulting electrode was demonstrated to have a remarkably high charge separation efficiency of 90% at 1.23 V vs. RHE in comparison to the typical value of <30% reported for the BiVO4 photoanode, attributed to the short charge transport pathways. Additionally, the porous structure ensuing from the nanoparticle network also allows the electrolyte to easily penetrate into the semiconductor film, increasing the contact area between the semiconductor and the electrolyte and promoting interfacial charge transfer to suppress charge recombination.74,82,96,97,180

On the other hand, nanowire/nanorod arrays are another type of nanostructure potent for enhanced charge separation and transport properties of semiconductor electrodes. While both nanoparticle and nanowire/nanorod provide high surface area per electrode volume, the one-dimensional nanowire/nanorod enables vectorial charge transport (i.e., the majority carriers move along the wire/rod to the electron collector back substrate, while the minority carriers travel radially outward to the semiconductor–electrolyte interface) which is a more effective mechanism of charge separation compared to the non-directional carrier movement in a nanoparticle-based electrode.181 To this end, Su et al. prepared WO3–BiVO4 heterojunction films with conformal deposition of nanoporous BiVO4 particles on the WO3 nanorod arrays, which were shown to have better IPCE than the planar WO3–BiVO4 heterojunction films accrediting to the higher surface area and faster charge separation at the WO3–BiVO4 interface of the nanorod structure.182 Such WO3–BiVO4 nanostructure was further refined by Rao et al. into a core–shell nanowire structure, which was demonstrated to be one of the best-performing BiVO4-based photoanodes, generating an AM1.5 photocurrent of 3.1 mA cm−2 at 1.23 V vs. RHE.183

It is obvious that contemporary research on nanoscale photocatalyst is stimulated by the prospects of efficient charge carrier extraction and enlarged active surface area. However, the large specific surface area of nanomaterials commonly results in the increase of grain boundaries, surface states, and defects, which are reflected by the decrease of crystallinity.16 The presence of these defect sites promotes charge recombination on the surface of photocatalyst. Moreover, separation of electrons and holes in small particles is harder to achieve because of the smaller or the absence of SCL, which is essential to generate a built-in electric field as the internal mechanism of charge separation (previously explained in Section 3.2.3).75 As a result, the effects of particle size on the activity of semiconductor photocatalyst are a complicated issue, particularly in different configurations for water splitting (i.e., PS and particulate PEC systems).

In fact, there has been contradiction in the literature regarding the size effects of WO3 particles on its water oxidation performance in the PS and PEC systems. For example, using three WO3 samples with different morphologies and particle sizes (namely, nanodots, nanoplates, and microcrystals with average diameters of 32 ± 16 nm, 475 ± 98 nm by 58 ± 16 nm, and 2 μm, respectively), Newton and Osterloh observed that both the photocatalytic O2 evolution activity (Fig. 27a) and the PEC photocurrent (Fig. 27b) of the WO3 concordantly increased with the decrease of particle size.184 The plots of O2 evolution rate and photocurrent magnitude against the inverse particle diameter (displayed in Fig. 27c and d, respectively) indicate a linear relationship in both cases, suggesting that the two systems are governed by minority hole transport which is facilitated by the decrease of particle size. Conversely, Hong et al. reported the opposite behaviors of PS and PEC systems with respect to the particle size of WO3.77 Four WO3 samples with different particle sizes in the range of 30–500 nm were obtained via calcination at different temperatures. As shown in Fig. 28, larger particle size generates higher amount of O2 but lower photocurrent. Such contrary effects of particle size on the water oxidation performance of WO3 in the PS and PEC systems were ascribed to the different charge separation mechanisms (refer to Section 3.2.3), whereby efficient charge separation in the PS system is provided by large particle with high crystallinity and well-developed SCL, while the PEC system relies on the reduced hole diffusion length by small particle.

image file: c7ta04441k-f27.tif
Fig. 27 (a) Time profile of O2 evolution and (b) current–potential curves of WO3 with different particle sizes. The corresponding photocatalytic rates of process (a) and (b) as a function of inverse particle diameter are displayed in (c) and (d), respectively. Reproduced with permission from ref. 184. Copyright 2016 Springer Science + Business Media New York.

image file: c7ta04441k-f28.tif
Fig. 28 Photoactivities of WO3 in PS and PEC systems as a function of particle size. Reprinted with permission from ref. 77. Copyright 2009 International Association for Hydrogen Energy.

While it is relatively recent that BiVO4 is recognized as an auspicious visible light-active photocatalyst highly active for water oxidation, which is viable through both the PS and PEC systems, the nanoscaling effect on its activity in the PS system is scarcely studied relative to the PEC system. A very few studies associated with the preparation of nanosized BiVO4 particles suggested that the suspended nanoparticles have poor activity for photocatalytic O2 evolution from aqueous AgNO3 solution.78,185 These results may suggest that nanoscaling is detrimental to the performance of BiVO4 in PS system, which is in contrast with that observed in the PEC system as aforementioned. However, direct comparison of the particle size dependence of BiVO4 water oxidation photoactivities in the PS and PEC systems based on the separate studies is inexpedient due to the variant properties of the photocatalyst materials being investigated. Instead, a systematic comparison should be executed using materials of the same properties in the two systems to understand the fundamental differences of the two in order to propel further advancement of BiVO4-based materials in the respective systems.

For this purpose, we directly compared the water oxidation photoactivity of three BiVO4 samples with distinctly different particle sizes (Fig. 29a) in the PEC and PS systems.21 The average particle size decreases from 833 ± 249 nm for sample A, to 374 ± 171 nm for sample B, and 123 ± 24 nm for sample C. XRD and XPS analyses indicated that the particle crystallinity deteriorates with the decrease of BiVO4 particle size, arising from the increase of surface defect states. In the PEC system, each BiVO4 sample was first made into a particulate photoanode by dropcasting the powder onto a conducting FTO substrate and evaluated for its photocurrent density that was generated during oxidation of water on the electrode surface. As shown in Fig. 29b, photoanode made of smaller BiVO4 particles was found to generate higher photocurrent density due to improved charge transfer ability, as suggested by the decreasing Nyquist arc radius (Fig. 29c). However, suspension of the powder BiVO4 samples in aqueous AgNO3 solution for water oxidation (known as the PS system) exhibited an opposite trend to that in the PEC system. Greater amount of O2 was evolved by the BiVO4 sample with larger particle size (Fig. 29d). These contrary effects of particle size on the water oxidation performance of BiVO4 in PEC and PS systems suggest different governing factors of the two systems, that is, charge transport for the former and charge separation for the latter (Fig. 29e). Presence of an external applied bias facilitates charge separation in the PEC system. Despite having more surface defect sites for charge recombination, smaller BiVO4 particles have enhanced PEC performance due to improved charge transport properties arising from better interparticle and particle/substrate contacts. On the other hand, the better O2 evolution ability of large BiVO4 particles demonstrated that the PS system efficiency is predominantly determined by charge separation efficiency of the photocatalyst instead of charge transport. In addition to improved crystallinity to suppress surface charge recombination, presence of SCL mainly in the significantly large particle is needed to generate a built-in electric field essential for efficient charge separation in the suspended photocatalyst.

image file: c7ta04441k-f29.tif
Fig. 29 (a) SEM images of BiVO4 powders with decreasing particle size from sample A to C. (b) Current–voltage curves, (c) Nyquist plots, and (d) photocatalytic O2 evolution activity of the BiVO4 samples. The dashed and full lines in panel b represent scans under dark and visible light conditions, respectively. (e) Schematic diagram illustrating charge transport and charge separation as the respective governing factors of BiVO4 photoactivity in the PEC and PS systems. Reprinted with permission from ref. 21. Copyright 2016 American Chemical Society.

5. Summary and outlook

The overview of the physical, optical, and electronic properties of BiVO4, in particular that of the monoclinic scheelite structure has shown the promise of BiVO4 as a visible light-active photocatalyst with a high oxidizing potential. In fact, the abilities of BiVO4 for photocatalytic organic degradations, photocatalytic water oxidation, Z-scheme overall water splitting, and PEC water splitting under visible light have been widely demonstrated. Despite the state-of-the-art BiVO4 often involving metal doping and coupling with water-oxidation cocatalyst to respectively increase the electron conductivity and surface water oxidation kinetics of the material, the charge separation efficiency of the modified BiVO4 remains below 30%.18 Development of alternative strategies involving modifications on morphology, composition, and electronic properties to improve the charge kinetics of BiVO4, encompassing charge separation, transport, and transfer, is therefore essential. Recent advances in crystal facet engineering, composite structure, annealing treatment, and nanoscaling show their great potential in controlling and enhancing the charge kinetics of BiVO4. The insights into these strategies in this review provide understanding on BiVO4's charge kinetics in relation to its surface and bulk properties, which is of fundamental importance for rational design and advancing progress of BiVO4 and other oxide materials as photocatalysts.

One major goal of photocatalyst development is to achieve effective separation, transport, and transfer of photogenerated electrons and holes for optimum performance of the photocatalyst. However, the strategies developed thus far are often found to have limited capabilities in targeting simultaneous enhancement toward all three aspects of charge kinetics. For instance, while crystal facet engineering of BiVO4 with exposed {010} and {110} facets promotes spatial separation of photogenerated electrons and holes on the two surfaces and enlargement of {010} increases the charge transfer efficiency, the electron transport properties of the material can only be improved via Ar annealing as a post-synthesis treatment. Thus, further direction of research shall delve into combination of multiple strategies to optimize the three primary steps in charge kinetics.

It is undeniable that the charge kinetics of photocatalysts, which involve multiple steps and mechanisms, are indeed a complicated field of study. While the existing mechanisms are proposed mainly based on the theories and the experimental data derived from various characterization tools, a deeper understanding of charge kinetics is often restricted by the limitations of the characterization techniques and real-life experimental conditions. Characterization techniques with high spatial, temporal, and spectral resolutions are indispensable for in situ and direct observations of charge kinetics with respect to the morphological, compositional, and electronic changes of a photocatalyst. Additionally, given the multidisciplinary nature of the field, integrating materials science, physical chemistry, and surface science, collaboration between researchers across different disciplines is also necessary to go beyond their knowledge boundaries for the development of new and effective photocatalytic systems.


We thank the Australian Research Council for financial support (Discovery Project DP170102895).


  1. D. Scaife, Sol. Energy, 1980, 25, 41–54 CrossRef CAS.
  2. M. S. Prévot and K. Sivula, J. Phys. Chem. C, 2013, 117, 17879–17893 Search PubMed.
  3. G. Wang, Y. Ling and Y. Li, Nanoscale, 2012, 4, 6682–6691 RSC.
  4. S. J. Hong, S. Lee, J. S. Jang and J. S. Lee, Energy Environ. Sci., 2011, 4, 1781–1787 CAS.
  5. A. Kudo and Y. Miseki, Chem. Soc. Rev., 2009, 38, 253–278 RSC.
  6. A. Kudo, K. Omori and H. Kato, J. Am. Chem. Soc., 1999, 121, 11459–11467 CrossRef CAS.
  7. H. Fu, C. Pan, W. Yao and Y. Zhu, J. Phys. Chem. B, 2005, 109, 22432–22439 CrossRef CAS PubMed.
  8. G. Li, N. Yang, W. Wang and W. Zhang, Electrochim. Acta, 2010, 55, 7235–7239 CrossRef CAS.
  9. D. Errandonea, A. Muñoz, P. Rodriguez-Hernandez, J. E. Proctor, F. Sapiña and M. Bettinelli, Inorg. Chem., 2015, 54, 7524–7535 CrossRef CAS PubMed.
  10. Y. Hosogi, Y. Shimodaira, H. Kato, H. Kobayashi and A. Kudo, Chem. Mater., 2008, 20, 1299–1307 CrossRef CAS.
  11. A. Kudo, K. Ueda, H. Kato and I. Mikami, Catal. Lett., 1998, 53, 229–230 CrossRef CAS.
  12. Y. Liang, T. Tsubota, L. P. Mooij and R. van de Krol, J. Phys. Chem. C, 2011, 115, 17594–17598 CAS.
  13. F. F. Abdi, N. Firet and R. van de Krol, ChemCatChem, 2013, 5, 490–496 CrossRef CAS.
  14. F. F. Abdi and R. van de Krol, J. Phys. Chem. C, 2012, 116, 9398–9404 CAS.
  15. F. F. Abdi, T. J. Savenije, M. M. May, B. Dam and R. van de Krol, J. Phys. Chem. Lett., 2013, 4, 2752–2757 CrossRef CAS.
  16. Y. Park, K. J. McDonald and K.-S. Choi, Chem. Soc. Rev., 2013, 42, 2321–2337 RSC.
  17. Z.-F. Huang, L. Pan, J.-J. Zou, X. Zhang and L. Wang, Nanoscale, 2014, 6, 14044–14063 RSC.
  18. D. K. Zhong, S. Choi and D. R. Gamelin, J. Am. Chem. Soc., 2011, 133, 18370–18377 CrossRef CAS PubMed.
  19. H. L. Tan, X. Wen, R. Amal and Y. H. Ng, J. Phys. Chem. Lett., 2016, 7, 1400–1405 CrossRef CAS PubMed.
  20. H. L. Tan, H. A. Tahini, X. Wen, R. J. Wong, X. Tan, A. Iwase, A. Kudo, R. Amal, S. C. Smith and Y. H. Ng, Small, 2016, 12, 5295–5302 CrossRef CAS PubMed.
  21. H. L. Tan, R. Amal and Y. H. Ng, ACS Appl. Mater. Interfaces, 2016, 8, 28607–28614 CAS.
  22. H. L. Tan, A. Suyanto, A. T. D. Denko, W. H. Saputera, R. Amal, F. E. Osterloh and Y. H. Ng, Part. Part. Syst. Charact., 2017, 34, 1600290 CrossRef.
  23. H. Endriss, in High Performance Pigments, Wiley-VCH Verlag GmbH & Co. KGaA, 2003, pp. 7–12,  DOI:10.1002/3527600493.ch2.
  24. J. D. Bierlein and A. W. Sleight, Solid State Commun., 1975, 16, 69–70 CrossRef CAS.
  25. A. Pinczuk, B. Welber and F. Dacol, Solid State Commun., 1979, 29, 515–518 CrossRef CAS.
  26. C. Manolikas and S. Amelinckx, Phys. Status Solidi A, 1980, 60, 167–172 CrossRef CAS.
  27. W. David and I. Wood, J. Phys. C: Solid State Phys., 1983, 16, 5149 CrossRef CAS.
  28. A. Lim, S. Choh and M. Jang, Ferroelectrics, 1989, 94, 389–394 CrossRef CAS.
  29. A. K. Bhattacharya, K. K. Mallick and A. Hartridge, Mater. Lett., 1997, 30, 7–13 CrossRef CAS.
  30. K. Hirota, G. Komatsu, M. Yamashita, H. Takemura and O. Yamaguchi, Mater. Res. Bull., 1992, 27, 823–830 CrossRef CAS.
  31. S. Tokunaga, H. Kato and A. Kudo, Chem. Mater., 2001, 13, 4624–4628 CrossRef CAS.
  32. A. R. Lim, S. H. Choh and M. S. Jang, J. Phys.: Condens. Matter, 1995, 7, 7309 CrossRef CAS.
  33. J. Yu and A. Kudo, Adv. Funct. Mater., 2006, 16, 2163–2169 CrossRef CAS.
  34. A. Walsh, Y. Yan, M. N. Huda, M. M. Al-Jassim and S.-H. Wei, Chem. Mater., 2009, 21, 547–551 CrossRef CAS.
  35. D. Payne, M. Robinson, R. Egdell, A. Walsh, J. McNulty, K. Smith and L. Piper, Appl. Phys. Lett., 2011, 98, 212110 CrossRef.
  36. J. K. Cooper, S. Gul, F. M. Toma, L. Chen, P.-A. Glans, J. Guo, J. W. Ager, J. Yano and I. D. Sharp, Chem. Mater., 2014, 26, 5365–5373 CrossRef CAS.
  37. Z. Zhao, Z. Li and Z. Zou, Phys. Chem. Chem. Phys., 2011, 13, 4746–4753 RSC.
  38. J. Yang, D. Wang, X. Zhou and C. Li, Chem.–Eur. J., 2013, 19, 1320–1326 CrossRef CAS PubMed.
  39. A. Kudo, Catal. Surv. Asia, 2003, 7, 31–38 CrossRef CAS.
  40. O. Carp, C. L. Huisman and A. Reller, Prog. Solid State Chem., 2004, 32, 33–177 CrossRef CAS.
  41. A. Fujishima, T. N. Rao and D. A. Tryk, J. Photochem. Photobiol., C, 2000, 1, 1–21 CrossRef CAS.
  42. W. Y. Teoh, J. A. Scott and R. Amal, J. Phys. Chem. Lett., 2012, 3, 629–639 CrossRef CAS PubMed.
  43. A. Martinez-de La Cruz and U. G. Pérez, Mater. Res. Bull., 2010, 45, 135–141 CrossRef CAS.
  44. L. Zhou, W. Wang, L. Zhang, H. Xu and W. Zhu, J. Phys. Chem. C, 2007, 111, 13659–13664 CAS.
  45. Y. Zheng, J. Wu, F. Duan and Y. Xie, Chem. Lett., 2007, 36, 520–521 CrossRef CAS.
  46. X. Zhang, Z. Ai, F. Jia, L. Zhang, X. Fan and Z. Zou, Mater. Chem. Phys., 2007, 103, 162–167 CrossRef CAS.
  47. A. Iwase, H. Kato and A. Kudo, J. Sol. Energy Eng., 2010, 132, 021106 CrossRef.
  48. L. Zhou, W. Wang, S. Liu, L. Zhang, H. Xu and W. Zhu, J. Mol. Catal. A: Chem., 2006, 252, 120–124 CrossRef CAS.
  49. L. Ge, Mater. Chem. Phys., 2008, 107, 465–470 CrossRef CAS.
  50. S. Kohtani, S. Makino, A. Kudo, K. Tokumura, Y. Ishigaki, T. Matsunaga, O. Nikaido, K. Hayakawa and R. Nakagaki, Chem. Lett., 2002, 660–661 CrossRef CAS.
  51. S. Kohtani, M. Koshiko, A. Kudo, K. Tokumura, Y. Ishigaki, A. Toriba, K. Hayakawa and R. Nakagaki, Appl. Catal., B, 2003, 46, 573–586 CrossRef CAS.
  52. S. Kohtani, J. Hiro, N. Yamamoto, A. Kudo, K. Tokumura and R. Nakagaki, Catal. Commun., 2005, 6, 185–189 CrossRef CAS.
  53. S. M. Thalluri, M. Hussain, G. Saracco, J. Barber and N. Russo, Ind. Eng. Chem. Res., 2014, 53, 2640–2646 CrossRef CAS.
  54. S. Kohtani, M. Tomohiro, K. Tokumura and R. Nakagaki, Appl. Catal., B, 2005, 58, 265–272 CrossRef CAS.
  55. S. Kohtani, K. Yoshida, T. Maekawa, A. Iwase, A. Kudo, H. Miyabe and R. Nakagaki, Phys. Chem. Chem. Phys., 2008, 10, 2986–2992 RSC.
  56. A. Fujishima, Nature, 1972, 238, 37–38 CrossRef CAS PubMed.
  57. R. Abe, J. Photochem. Photobiol., C, 2010, 11, 179–209 CrossRef CAS.
  58. A. J. Bard and M. S. Wrighton, J. Electrochem. Soc., 1977, 124, 1706–1710 CrossRef CAS.
  59. H. Gerischer, J. Electroanal. Chem. Interfacial Electrochem., 1977, 82, 133–143 CrossRef CAS.
  60. L. Yang, H. Zhou, T. Fan and D. Zhang, Phys. Chem. Chem. Phys., 2014, 16, 6810–6826 RSC.
  61. K. Maeda, T. Takata, M. Hara, N. Saito, Y. Inoue, H. Kobayashi and K. Domen, J. Am. Chem. Soc., 2005, 127, 8286–8287 CrossRef CAS PubMed.
  62. Y. Sasaki, H. Nemoto, K. Saito and A. Kudo, J. Phys. Chem. C, 2009, 113, 17536–17542 CAS.
  63. A. J. Bard, J. Photochem., 1979, 10, 59–75 CrossRef CAS.
  64. A. Iwase, Y. H. Ng, Y. Ishiguro, A. Kudo and R. Amal, J. Am. Chem. Soc., 2011, 133, 11054–11057 CrossRef CAS PubMed.
  65. A. Iwase, S. Yoshino, T. Takayama, Y. H. Ng, R. Amal and A. Kudo, J. Am. Chem. Soc., 2016, 138, 10260–10264 CrossRef CAS PubMed.
  66. J. R. Darwent and A. Mills, J. Chem. Soc., Faraday Trans. 2, 1982, 78, 359–367 RSC.
  67. K. Sayama, A. Nomura, T. Arai, T. Sugita, R. Abe, M. Yanagida, T. Oi, Y. Iwasaki, Y. Abe and H. Sugihara, J. Phys. Chem. B, 2006, 110, 11352–11360 CrossRef CAS PubMed.
  68. H. Liu, R. Nakamura and Y. Nakato, J. Electrochem. Soc., 2005, 152, G856–G861 CrossRef.
  69. H. Kato, M. Hori, R. Konta, Y. Shimodaira and A. Kudo, Chem. Lett., 2004, 33, 1348–1349 CrossRef CAS.
  70. Y. Sasaki, A. Iwase, H. Kato and A. Kudo, J. Catal., 2008, 259, 133–137 CrossRef CAS.
  71. Y. Sasaki, H. Kato and A. Kudo, J. Am. Chem. Soc., 2013, 135, 5441–5449 CrossRef CAS PubMed.
  72. M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi, E. A. Santori and N. S. Lewis, Chem. Rev., 2010, 110, 6446–6473 CrossRef CAS PubMed.
  73. K. Zhang, M. Ma, P. Li, D. H. Wang and J. H. Park, Adv. Energy Mater., 2016, 6, 1600602 CrossRef.
  74. K. Sayama, A. Nomura, Z. Zou, R. Abe, Y. Abe and H. Arakawa, Chem. Commun., 2003, 2908–2909,  10.1039/b310428a.
  75. F. E. Osterloh, in Solar Energy for Fuels, Springer International Publishing, 2015, pp. 105–142 Search PubMed.
  76. R. Memming, in Electron Transfer I, Springer, 1994, pp. 105–181 Search PubMed.
  77. S. J. Hong, H. Jun, P. H. Borse and J. S. Lee, Int. J. Hydrogen Energy, 2009, 34, 3234–3242 CrossRef CAS.
  78. A. Iwase and A. Kudo, J. Mater. Chem., 2010, 20, 7536–7542 RSC.
  79. A. J. Rettie, H. C. Lee, L. G. Marshall, J.-F. Lin, C. Capan, J. Lindemuth, J. S. McCloy, J. Zhou, A. J. Bard and C. B. Mullins, J. Am. Chem. Soc., 2013, 135, 11389–11396 CrossRef CAS PubMed.
  80. K. E. Kweon and G. S. Hwang, Phys. Rev. B: Condens. Matter Mater. Phys., 2013, 87, 205202 CrossRef.
  81. H. Ye, J. Lee, J. S. Jang and A. J. Bard, J. Phys. Chem. C, 2010, 114, 13322–13328 CAS.
  82. W. Luo, Z. Yang, Z. Li, J. Zhang, J. Liu, Z. Zhao, Z. Wang, S. Yan, T. Yu and Z. Zou, Energy Environ. Sci., 2011, 4, 4046–4051 CAS.
  83. K. P. S. Parmar, H. J. Kang, A. Bist, P. Dua, J. S. Jang and J. S. Lee, ChemSusChem, 2012, 5, 1926–1934 CrossRef CAS PubMed.
  84. H. S. Park, K. E. Kweon, H. Ye, E. Paek, G. S. Hwang and A. J. Bard, J. Phys. Chem. C, 2011, 115, 17870–17879 CAS.
  85. S. P. Berglund, A. J. Rettie, S. Hoang and C. B. Mullins, Phys. Chem. Chem. Phys., 2012, 14, 7065–7075 RSC.
  86. G. Beni, L. Schiavone, J. Shay, W. Dautremont-Smith and B. Schneider, Nature, 1979, 282, 281–283 CrossRef CAS.
  87. A. Harriman, I. J. Pickering, J. M. Thomas and P. A. Christensen, J. Chem. Soc., Faraday Trans. 1, 1988, 84, 2795–2806 RSC.
  88. D. Wang, R. Li, J. Zhu, J. Shi, J. Han, X. Zong and C. Li, J. Phys. Chem. C, 2012, 116, 5082–5089 CAS.
  89. T. H. Jeon, W. Choi and H. Park, Phys. Chem. Chem. Phys., 2011, 13, 21392–21401 RSC.
  90. M. Zhou, J. Bao, W. Bi, Y. Zeng, R. Zhu, M. Tao and Y. Xie, ChemSusChem, 2012, 5, 1420–1425 CrossRef CAS PubMed.
  91. S. K. Pilli, T. G. Deutsch, T. E. Furtak, J. A. Turner, L. D. Brown and A. M. Herring, Phys. Chem. Chem. Phys., 2012, 14, 7032–7039 RSC.
  92. H. Ye, H. S. Park and A. J. Bard, J. Phys. Chem. C, 2011, 115, 12464–12470 CAS.
  93. R. Li, H. Han, F. Zhang, D. Wang and C. Li, Energy Environ. Sci., 2014, 7, 1369–1376 CAS.
  94. Q. Jia, K. Iwashina and A. Kudo, Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 11564–11569 CrossRef CAS PubMed.
  95. J. A. Seabold and K.-S. Choi, J. Am. Chem. Soc., 2012, 134, 2186–2192 CrossRef CAS PubMed.
  96. K. J. McDonald and K.-S. Choi, Energy Environ. Sci., 2012, 5, 8553–8557 CAS.
  97. T. W. Kim and K.-S. Choi, Science, 2014, 343, 990–994 CrossRef CAS PubMed.
  98. P. Morris Hotsenpiller, J. Bolt, W. Farneth, J. Lowekamp and G. Rohrer, J. Phys. Chem. B, 1998, 102, 3216–3226 CrossRef.
  99. J. Pan, G. Liu, G. Q. M. Lu and H. M. Cheng, Angew. Chem., Int. Ed., 2011, 50, 2133–2137 CrossRef CAS PubMed.
  100. C. G. Read, E. M. Steinmiller and K.-S. Choi, J. Am. Chem. Soc., 2009, 131, 12040–12041 CrossRef CAS PubMed.
  101. D. J. Martin, N. Umezawa, X. Chen, J. Ye and J. Tang, Energy Environ. Sci., 2013, 6, 3380–3386 CAS.
  102. X. Zhou, J. Lan, G. Liu, K. Deng, Y. Yang, G. Nie, J. Yu and L. Zhi, Angew. Chem., 2012, 124, 182–186 CrossRef.
  103. Y. P. Xie, G. Liu, L. Yin and H.-M. Cheng, J. Mater. Chem., 2012, 22, 6746–6751 RSC.
  104. H. G. Yang, C. H. Sun, S. Z. Qiao, J. Zou, G. Liu, S. C. Smith, H. M. Cheng and G. Q. Lu, Nature, 2008, 453, 638–641 CrossRef CAS PubMed.
  105. X.-Q. Gong and A. Selloni, J. Phys. Chem. B, 2005, 109, 19560–19562 CrossRef CAS PubMed.
  106. D. Zhang, G. Li, X. Yang and C. Y. Jimmy, Chem. Commun., 2009, 4381–4383 RSC.
  107. G. Liu, C. Sun, H. G. Yang, S. C. Smith, L. Wang, G. Q. M. Lu and H.-M. Cheng, Chem. Commun., 2010, 46, 755–757 RSC.
  108. X. Han, Q. Kuang, M. Jin, Z. Xie and L. Zheng, J. Am. Chem. Soc., 2009, 131, 3152–3153 CrossRef CAS PubMed.
  109. A. Selloni, Nat. Mater., 2008, 7, 613–615 CrossRef CAS PubMed.
  110. C. Zhang and Y. Zhu, Chem. Mater., 2005, 17, 3537–3545 CrossRef CAS.
  111. Y. Zheng, F. Duan, J. Wu, L. Liu, M. Chen and Y. Xie, J. Mol. Catal. A: Chem., 2009, 303, 9–14 CrossRef CAS.
  112. T. Zhou, J. Hu and J. Li, Appl. Catal., B, 2011, 110, 221–230 CrossRef CAS.
  113. L. Zhang, D. Chen and X. Jiao, J. Phys. Chem. B, 2006, 110, 2668–2673 CrossRef CAS PubMed.
  114. G. Xi and J. Ye, Chem. Commun., 2010, 46, 1893–1895 RSC.
  115. D. Wang, H. Jiang, X. Zong, Q. Xu, Y. Ma, G. Li and C. Li, Chem.–Eur. J., 2011, 17, 1275–1282 CrossRef CAS PubMed.
  116. T. Ohno, K. Sarukawa and M. Matsumura, New J. Chem., 2002, 26, 1167–1170 RSC.
  117. R. Li, F. Zhang, D. Wang, J. Yang, M. Li, J. Zhu, X. Zhou, H. Han and C. Li, Nat. Commun., 2013, 4, 1432 CrossRef PubMed.
  118. T. Tachikawa, S. Yamashita and T. Majima, J. Am. Chem. Soc., 2011, 133, 7197–7204 CrossRef CAS PubMed.
  119. L. Zhang, W. Wang, S. Sun, D. Jiang and E. Gao, Appl. Catal., B, 2015, 162, 470–474 CrossRef CAS.
  120. H. Li, H. Yu, X. Quan, S. Chen and H. Zhao, Adv. Funct. Mater., 2015, 25, 3074–3080 CrossRef CAS.
  121. Y. Kim, D. Shin, W. J. Chang, H. L. Jang, C. W. Lee, H. E. Lee and K. T. Nam, Adv. Funct. Mater., 2015, 25, 2369–2377 CrossRef CAS.
  122. J. Zhu, F. Fan, R. Chen, H. An, Z. Feng and C. Li, Angew. Chem., 2015, 127, 9239–9242 CrossRef.
  123. T. Tachikawa, T. Ochi and Y. Kobori, ACS Catal., 2016, 6, 2250–2256 CrossRef CAS.
  124. S. Zhu, T. Xu, H. Fu, J. Zhao and Y. Zhu, Environ. Sci. Technol., 2007, 41, 6234–6239 CrossRef CAS PubMed.
  125. H. Fu, T. Xu, S. Zhu and Y. Zhu, Environ. Sci. Technol., 2008, 42, 8064–8069 CrossRef CAS PubMed.
  126. L. Jiang and L. Gao, Mater. Chem. Phys., 2005, 91, 313–316 CrossRef CAS.
  127. A. Kongkanand, R. Martínez Domínguez and P. V. Kamat, Nano Lett., 2007, 7, 676–680 CrossRef CAS PubMed.
  128. Q. Xiang, J. Yu and M. Jaroniec, Chem. Soc. Rev., 2012, 41, 782–796 RSC.
  129. S. Stankovich, D. A. Dikin, G. H. Dommett, K. M. Kohlhaas, E. J. Zimney, E. A. Stach, R. D. Piner, S. T. Nguyen and R. S. Ruoff, Nature, 2006, 442, 282–286 CrossRef CAS PubMed.
  130. K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva and A. A. Firsov, Science, 2004, 306, 666–669 CrossRef CAS PubMed.
  131. X. Li, Epitaxial Graphene Films on Silicon Carbide: Growth, Characterization, and Devices, ProQuest, 2008 Search PubMed.
  132. Z. Sun, Z. Yan, J. Yao, E. Beitler, Y. Zhu and J. M. Tour, Nature, 2010, 468, 549–552 CrossRef CAS PubMed.
  133. G. Williams, B. Seger and P. V. Kamat, ACS Nano, 2008, 2, 1487–1491 CrossRef CAS PubMed.
  134. I. V. Lightcap, T. H. Kosel and P. V. Kamat, Nano Lett., 2010, 10, 577–583 CrossRef CAS PubMed.
  135. H. Zhang, X. Lv, Y. Li, Y. Wang and J. Li, ACS Nano, 2009, 4, 380–386 CrossRef PubMed.
  136. N. J. Bell, Y. H. Ng, A. Du, H. Coster, S. C. Smith and R. Amal, J. Phys. Chem. C, 2011, 115, 6004–6009 CAS.
  137. X. An, C. Y. Jimmy, Y. Wang, Y. Hu, X. Yu and G. Zhang, J. Mater. Chem., 2012, 22, 8525–8531 RSC.
  138. B. Li and H. Cao, J. Mater. Chem., 2011, 21, 3346–3349 RSC.
  139. J. Zhang, Z. Xiong and X. Zhao, J. Mater. Chem., 2011, 21, 3634–3640 RSC.
  140. Y. H. Ng, A. Iwase, A. Kudo and R. Amal, J. Phys. Chem. Lett., 2010, 1, 2607–2612 CrossRef CAS.
  141. S. B. Gawande and S. R. Thakare, Int. Nano Lett., 2012, 2, 1–7 CrossRef.
  142. Y. Yan, S. Sun, Y. Song, X. Yan, W. Guan, X. Liu and W. Shi, J. Hazard. Mater., 2013, 250–251, 106–114 CrossRef CAS PubMed.
  143. Y. Fu, X. Sun and X. Wang, Mater. Chem. Phys., 2011, 131, 325–330 CrossRef CAS.
  144. S. Dong, Y. Cui, Y. Wang, Y. Li, L. Hu, J. Sun and J. Sun, Chem. Eng. J., 2014, 249, 102–110 CrossRef CAS.
  145. Y. Wang, W. Wang, H. Mao, Y. Lu, J. Lu, J. Huang, Z. Ye and B. Lu, ACS Appl. Mater. Interfaces, 2014, 6, 12698–12706 CAS.
  146. Y. Wang, F. Li, H. Li, L. Bai and L. Sun, Chem. Commun., 2016, 52, 3050–3053 RSC.
  147. A. Wang, S. Shen, Y. Zhao and W. Wu, J. Colloid Interface Sci., 2015, 445, 330–336 CrossRef CAS PubMed.
  148. X. Zhou, J. Yu, Y. Zhang, D. Yu and W. Lu, Rare Met., 2011, 30, 199–202 CrossRef CAS.
  149. Y. Zhang, J. Yu, D. Yu, X. Zhou and W. Lu, Rare Met., 2011, 30, 192–198 CrossRef CAS.
  150. B. Liu, Z. Li, S. Xu, X. Ren, D. Han and D. Lu, J. Phys. Chem. Solids, 2014, 75, 977–983 CrossRef CAS.
  151. C. Wang, D. Meng, J. Sun, J. Memon, Y. Huang and J. Geng, Adv. Mater. Interfaces, 2014, 1, 1300150 CrossRef.
  152. T. Wang, C. Li, J. Ji, Y. Wei, P. Zhang, S. Wang, X. Fan and J. Gong, ACS Sustainable Chem. Eng., 2014, 2, 2253–2258 CrossRef CAS.
  153. W. Feng, S. Weng, Z. Zheng, Z. Fang and P. Liu, Chem. Commun., 2015, 51, 11186–11189 RSC.
  154. B. M. Reddy, A. Khan, Y. Yamada, T. Kobayashi, S. Loridant and J.-C. Volta, J. Phys. Chem. B, 2003, 107, 11475–11484 CrossRef CAS.
  155. H. S. Kang, J. S. Kang, J. W. Kim and S. Y. Lee, J. Appl. Phys., 2004, 95, 1246–1250 CrossRef CAS.
  156. S. W. Oh, H. J. Bang, Y. C. Bae and Y.-K. Sun, J. Power Sources, 2007, 173, 502–509 CrossRef CAS.
  157. J. Yu and B. Wang, Appl. Catal., B, 2010, 94, 295–302 CrossRef CAS.
  158. D. Hanaor and C. Sorrell, J. Mater. Sci., 2011, 46, 855–874 CrossRef CAS.
  159. J.-G. Yu, H.-G. Yu, B. Cheng, X.-J. Zhao, J. C. Yu and W.-K. Ho, J. Phys. Chem. B, 2003, 107, 13871–13879 CrossRef CAS.
  160. C. Ng, Y. H. Ng, A. Iwase and R. Amal, ACS Appl. Mater. Interfaces, 2013, 5, 5269–5275 CAS.
  161. K. Yamada, H. Yamane, S. Matsushima, H. Nakamura, K. Ohira, M. Kouya and K. Kumada, Thin Solid Films, 2008, 516, 7482–7487 CrossRef CAS.
  162. Y. Chae, S. Mori and M. Suzuki, Thin Solid Films, 2009, 517, 4260–4263 CrossRef CAS.
  163. W. Mtangi, F. D. Auret, W. Meyer, M. Legodi, P. J. van Rensburg, S. Coelho, M. Diale and J. Nel, J. Appl. Phys., 2012, 111, 094504 CrossRef.
  164. F. Zuo, L. Wang, T. Wu, Z. Zhang, D. Borchardt and P. Feng, J. Am. Chem. Soc., 2010, 132, 11856–11857 CrossRef CAS PubMed.
  165. X. Chen, L. Liu, Y. Y. Peter and S. S. Mao, Science, 2011, 331, 746–750 CrossRef CAS PubMed.
  166. G. Wang, H. Wang, Y. Ling, Y. Tang, X. Yang, R. C. Fitzmorris, C. Wang, J. Z. Zhang and Y. Li, Nano Lett., 2011, 11, 3026–3033 CrossRef CAS PubMed.
  167. X. Lu, G. Wang, S. Xie, J. Shi, W. Li, Y. Tong and Y. Li, Chem. Commun., 2012, 48, 7717–7719 RSC.
  168. G. Wang, Y. Ling, H. Wang, X. Yang, C. Wang, J. Z. Zhang and Y. Li, Energy Environ. Sci., 2012, 5, 6180–6187 CAS.
  169. G. Wang, Y. Ling, X. Lu, F. Qian, Y. Tong, J. Z. Zhang, V. Lordi, C. Rocha Leao and Y. Li, J. Phys. Chem. C, 2013, 117, 10957–10964 CAS.
  170. Y. Zhang, Y. Guo, H. Duan, H. Li, C. Sun and H. Liu, Phys. Chem. Chem. Phys., 2014, 16, 24519–24526 RSC.
  171. A. P. Singh, N. Kodan, A. Dey, S. Krishnamurthy and B. R. Mehta, Int. J. Hydrogen Energy, 2015, 40, 4311–4319 CrossRef CAS.
  172. H. Jiang, H. Dai, X. Meng, K. Ji, L. Zhang and J. Deng, Appl. Catal., B, 2011, 105, 326–334 CrossRef CAS.
  173. Y. Liu, H. Dai, J. Deng, L. Zhang and C. T. Au, Nanoscale, 2012, 4, 2317–2325 RSC.
  174. Y. Huang, H. Li, M.-S. Balogun, W. Liu, Y. Tong, X. Lu and H. Ji, ACS Appl. Mater. Interfaces, 2014, 6, 22920–22927 CAS.
  175. W.-J. Yin, S.-H. Wei, M. M. Al-Jassim, J. Turner and Y. Yan, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 83, 155102 CrossRef.
  176. J. Gan, X. Lu, B. B. Rajeeva, R. Menz, Y. Tong and Y. Zheng, ChemElectroChem, 2015, 2, 1385–1395 CrossRef CAS.
  177. Y. Zhang, D. Wang, X. Zhang, Y. Chen, L. Kong, P. Chen, Y. Wang, C. Wang, L. Wang and Y. Liu, Electrochim. Acta, 2016, 195, 51–58 CrossRef CAS.
  178. T. W. Kim, Y. Ping, G. A. Galli and K.-S. Choi, Nat. Commun., 2015, 6, 8769 CrossRef CAS PubMed.
  179. G. Hodes and P. V. Kamat, J. Phys. Chem. Lett., 2015, 6, 4090–4092 CrossRef CAS PubMed.
  180. J. A. Seabold, K. Zhu and N. R. Neale, Phys. Chem. Chem. Phys., 2014, 16, 1121–1131 RSC.
  181. Z. Jiang, Y. Tang, Q. Tay, Y. Zhang, O. I. Malyi, D. Wang, J. Deng, Y. Lai, H. Zhou and X. Chen, Adv. Energy Mater., 2013, 3, 1368–1380 CrossRef CAS.
  182. J. Su, L. Guo, N. Bao and C. A. Grimes, Nano Lett., 2011, 11, 1928–1933 CrossRef CAS PubMed.
  183. P. M. Rao, L. Cai, C. Liu, I. S. Cho, C. H. Lee, J. M. Weisse, P. Yang and X. Zheng, Nano Lett., 2014, 14, 1099–1105 CrossRef CAS PubMed.
  184. K. A. Newton and F. E. Osterloh, Top. Catal., 2016, 1–7 Search PubMed.
  185. Y. K. Kho, W. Y. Teoh, A. Iwase, L. Mädler, A. Kudo and R. Amal, ACS Appl. Mater. Interfaces, 2011, 3, 1997–2004 CAS.

This journal is © The Royal Society of Chemistry 2017