Role of Gd in Enhancing the Charge Carrier Mobility of Spray–Deposited BiVO4 Photoanodes

The emergence of BiVO4 as one of the most promising photoanodes for solar water splitting is largely driven by the success in dopant introduction and optimization to improve its photoelectrochemical performance. To this end, although less commonly used, several trivalent ions (e.g., In3+, Gd3+) that substitute Bi3+ have also been demonstrated to be effective dopants, which can increase the photocurrent of BiVO4 photoanodes. However, the main factor behind such improvement is still unclear as various explanations were proposed in the literature. Herein, Gd3+ is introduced to substitute Bi3+ in spray‐deposited BiVO4 films, which enables up to a ∼2‐fold photocurrent increase. Further analysis suggests that Gd‐doping enhances the carrier separation in BiVO4 and does not affect the catalytic and optical properties. Indeed, time‐resolved microwave conductivity measurements reveal that the carrier mobility of BiVO4 is increased by 50% with the introduction of Gd, while the carrier lifetime is unaffected. This mobility increase is rationalized to be a result of a higher degree of monoclinic lattice distortion in Gd‐doped BiVO4, as evident from the X‐ray diffraction and Raman spectroscopy data. Overall, these findings provide important insights into the nature and the underlying role of Gd in improving the performance of BiVO4.

The emergence of BiVO 4 as one of the most promising photoanodes for solar water splitting is largely driven by the success in dopant introduction and optimization to improve its photoelectrochemical performance. To this end, although less commonly used, several trivalent ions (e.g., In 3þ , Gd 3þ ) that substitute Bi 3þ have also been demonstrated to be effective dopants, which can increase the photocurrent of BiVO 4 photoanodes. However, the main factor behind such improvement is still unclear as various explanations were proposed in the literature. Herein, Gd 3þ is introduced to substitute Bi 3þ in spray-deposited BiVO 4 films, which enables up to a $2-fold photocurrent increase. Further analysis suggests that Gd-doping enhances the carrier separation in BiVO 4 and does not affect the catalytic and optical properties. Indeed, time-resolved microwave conductivity measurements reveal that the carrier mobility of BiVO 4 is increased by 50% with the introduction of Gd, while the carrier lifetime is unaffected. This mobility increase is rationalized to be a result of a higher degree of monoclinic lattice distortion in Gd-doped BiVO 4 , as evident from the X-ray diffraction and Raman spectroscopy data. Overall, these findings provide important insights into the nature and the underlying role of Gd in improving the performance of BiVO 4 .
The substitution of Bi 3þ in BiVO 4 with trivalent ions has also been reported. Interestingly, in contrast to the Mo or W doping described earlier, the isovalent nature of the dopants, in this case, is not expected to directly modify the carrier concentration of BiVO 4 . Other roles have therefore been proposed. For example, the incorporation of In 3þ in BiVO 4 was reported to inhibit surface charge recombination due to surface state passivation. [23] Govindaraju et al. evaluated several lanthanide ions (La, Ce, Sm, and Yb) as dopants in BiVO 4 and reported multiple roles. [24] Doping with La and Ce was found to decrease the PEC performance due to an increase of the effective mass of carriers (as shown from density functional theory, DFT) and/or an introduction of a midgap state that acts as recombination center. Meanwhile, Sm and Yb doping, despite the main 3þ oxidation state, indirectly increased the carrier concentration in BiVO 4 because of the coexistence of oxygen vacancies that introduced shallow donor states and enhanced the PEC performance. Gd has also been reported as an effective dopant in BiVO 4 photocatalyst powders, and higher photocatalytic activity for the degradation of methylene blue and rhodamine B was reported. [25,26] It was suggested that the half-filled 4f electronic configuration of Gd 3þ ions on the surface of BiVO 4 is beneficial for trapping charges and enhancing charge transfer for the catalytic reactions. An optimal Gd-doping concentration also simultaneously enhances the activity, selectivity, and stability of BiVO 4 for the water oxidation to H 2 O 2 production. [27] Newhouse et al. showed that Gd doping modified the monoclinic distortion in the lattice of BiVO 4 , which was expected to decrease the hole effective mass as shown by their DFT results. [28] Further improvement was demonstrated by combining Gd doping with an additional transition metal (W or Mo) as electron donor and H 2 annealing to improve the carrier lifetime. [29] Overall, various assignments have been proposed, which demonstrates that the main role of trivalent dopants in BiVO 4 , especially Gd, is still not clear. To unravel this, further investigation on the influence of trivalent ion doping on the photoelectrochemical properties (e.g., carrier transport and catalytic activity) of BiVO 4 is still needed.
In this study, we fabricated Gd-doped and (Gd,W)-co-doped BiVO 4 thin film photoanodes using spray pyrolysis, which demonstrate up to twice as high photocurrents as compared to that of pristine BiVO 4 . This photocurrent enhancement is found to be a result of a higher charge separation efficiency, not catalytic efficiency. In contrast to W doping, [8] Gd doping enhances the carrier mobility while maintaining the relatively long carrier lifetime in BiVO 4 as measured by time-resolved microwave conductivity (TRMC). To the best of our knowledge, this is the first direct experimental study that correlates Gd doping in BiVO 4 to the modulation of its carrier transport properties. Further effects of Gd doping on the optical, surface chemical composition, morphological, and structural properties of BiVO 4 will also be discussed.

Results and Discussion
2.1. Photoelectrochemical and Optoelectronic Properties of Gd-Doped BiVO 4 As slow surface reaction kinetics (i.e., slow hole transfer across the semiconductor/electrolyte interface) have been shown to limit the photocurrent of BiVO 4 , [7] sodium sulfite (Na 2 SO 3 ) was added into the electrolyte in this study as hole scavenger to eliminate any surface catalytic limitations. [30] Deposition of various cocatalysts (e.g., NiFeOOH and CoP i ) has been demonstrated to overcome this limitation in a hole-scavenger-free electrolyte, [7,31,32] but this is not the focus of our work. The AM 1.5 photocurrent-voltage curves for BiVO 4 and 1% Gd-BiVO 4 films in 2 M KP i with and without 0.5 M Na 2 SO 3 are shown in Figure 1. In both electrolytes, 1% Gd doping results in a significantly higher photocurrent and a cathodic shift of the onset potential of BiVO 4 ; the effect is more pronounced in the presence of hole scavengers. The Gd concentration was also varied up to 3% (Bi 1Àx Gd x VO 4 ; x ¼ 0, 0.005, 0.01, 0.02, and 0.03). The crystal structure of all these samples was found to be monoclinic scheelite BiVO 4 , which will be shown and discussed in a later section of the manuscript. The AM 1.5 photocurrents at 1.23 V RHE are plotted as a function of the Gd concentration, as shown in Figure 2 and Figure S1a, Supporting Information. All Gd-doped BiVO 4 films show higher photocurrents at 1.23 V RHE than the undoped film, and 1% is found to be the optimum Gd concentration. The photocurrent of BiVO 4 and 1% Gd-BiVO 4 at 1.23 V RHE is %1.8 and 3.4 mA cm À2 , respectively. This approximately twofold photocurrent improvement underlines the effectiveness of Gd doping in BiVO 4 . We also note that the photocurrent of the 1% Gd-BiVO 4 film is of the same order as those reported for spray-pyrolyzed gradient-doped W-BiVO 4 films, [20,21] which is among the highest for nonnanostructured BiVO 4 .
We then combined Gd doping with W doping (i.e., codoping) because W has been widely used as a donor dopant to improve the electrical conductivity of BiVO 4 . [8,9,15,16] The AM 1.5 photocurrents of 1% W-doped BiVO 4 films with varying concentrations of Gd are shown in Figure S1b,c, Supporting Information (the synthesis of W-doped BiVO 4 and (Gd,W)codoped BiVO 4 films is described in Supporting Note S1, Supporting Information). Gd doping is also effective in improving the photocurrent of W-doped BiVO 4 and the same optimum concentration of 1% Gd was found. Surprisingly, the photocurrents of the (Gd,W)-codoped BiVO 4 films are lower than those of Gd-doped films. This observation is in contrast to the case of (Gd, Mo)-codoped BiVO 4 films, as reported by Newhouse et al. [28] The www.advancedsciencenews.com www.solar-rrl.com photocurrents of (Gd,Mo)-codoped BiVO 4 was enhanced when compared to the single doping with Gd or Mo, which was attributed to the lower hole effective mass. More recently, they also reported that an enhancement can also be obtained by codoping with Gd and W, although not as large as codoping with Gd and Mo. [29] We speculate that the reason for this difference is related to the difference in the preparation method. First, inkjet printing was used in their study versus spray pyrolysis in our study. The range of the dopant concentration investigated is also different. Finally, Newhouse et al. maintained the stoichiometry of Bi and V (i.e., Bi:V ¼ 1:1), [29] but in this study we adjust the Bi and V precursor concentration to accommodate for the W and Gd dopants, such that (Bi þ Gd):(V þ W) ¼ 1:1. Although undoubtedly interesting, it is clear that codoping introduces more complexity toward understanding the main role of Gd, which is beyond the scope of the current work. Throughout the remainder of this study, we therefore focus only on Gd-doped BiVO 4 films, since monodoping BiVO 4 with Gd has been shown unambiguously to be effective in various reports.
To investigate the cause behind the photocurrent improvement, we calculated the catalytic efficiency (also often called "charge injection efficiency") from the photocurrent-voltage curves in the absence and presence of Na 2 SO 3 as hole scavengers in the electrolyte. The method has been described in previous reports, [16,33,34] and the determination of catalytic efficiency is explained in Supporting Note S2, Supporting Information. The calculated catalytic efficiencies for BiVO 4 and 1% Gd-BiVO 4 are shown in Figure S2, Supporting Information. At potentials higher than %1.1 V RHE , a slight increase of the catalytic efficiency was observed with 1% Gd doping. However, this enhancement is minor compared to the photocurrent increase, suggesting that bulk modifications are the main reason instead.
To investigate the optical absorption of the films, UV-vis spectroscopy measurements were performed on the undoped BiVO 4 and 1% Gd-BiVO 4 films deposited on quartz substrates. The UV-vis absorption spectra and the photographs of the measured samples are shown in Figure 3a. No significant difference in the absorptance can be observed; the 1% Gd-BiVO 4 shows slightly lower absorptance at wavelengths lower than 475 nm, but such a difference is within the sample-to-sample variation. The optical bandgap energy (E g ) can be obtained by the Tauc equation [35] where α is the absorption coefficient, h is Planck's constant, ν is the light frequency, A is the proportional constant, and n is 1/2 or 2 for direct or indirect allowed transition, respectively. For Tauc analysis in this study, the α in Equation (1) was approximated using the Kubelka-Munk function (F) of the film's transflectance (TR): [36] F ¼ ð1 À TRÞ 2 2TR (2) The Tauc plots for direct and indirect bandgaps are shown in Figure 3b,c, respectively. Again, no changes can be observed between the direct or indirect bandgaps of BiVO 4 and 1% Gd-BiVO 4 , indicating that Gd incorporation does not affect the optical properties of our spray-deposited BiVO 4 .  The fact that no significant catalytic or absorption changes could be observed upon Gd doping suggests that the improved photoelectrochemical properties are related to the enhancement of the charge separation efficiency. A possible factor that may increase the charge separation efficiency is an increase of conductivity due to higher carrier concentration. This is, however, unlikely to be the case as Gd is an isovalent dopant and we will discuss later that the process of introducing Gd into our BiVO 4 should not be compensated for with any electronic or ionic defects. We therefore evaluate the charge transport properties of both undoped and 1% Gd-BiVO 4 by performing TRMC measurements. Briefly, both films are excited with a nanosecond laser pulse (photon energy hν > bandgap), resulting in the formation of free electrons and holes, which are then detected as a change in the microwave power reflected from the samples. This change corresponds to the change of photoconductance (ΔG), which is then monitored as a function of time (i.e., transient). The relative comparison of the peak transient can be used to evaluate the charge carrier mobility, and the charge carrier lifetime can be obtained from the transient decay. [35,37] Figure 4a shows the transient photoconductance responses, normalized to the absorption (F A ) and photon flux (I 0 ), for the undoped and 1% Gd-BiVO 4 films. The measurements were performed under backside 350 nm laser pulse illumination (i.e., hν % 3.5 eV > E g of BiVO 4 ), and the I 0 was 4.3 Â 10 13 pulse À1 cm À2 . Upon Gd doping, the peak ΔG/(F A I 0 ) transient increases by a factor of 1.5, which indicates an improvement of the charge carrier mobility. In contrast, the transient decay does not seem to be affected. This can be seen clearly from Figure 4b, in which the transient photoconductance of each sample is normalized against the peak value (i.e., ΔG/ΔG max ). Both transients overlap with each other, which indicates that the charge carrier lifetime of BiVO 4 is unaffected by Gd doping. This also suggests that the charge transport mechanism is likely to be the same in both films. Similar effects to the transient peak and decay can also be observed for other excitation wavelengths (see Figure S3, Supporting Information). Overall, the higher mobility and constant lifetime with Gd doping result in an extended charge carrier diffusion length, which can explain the enhanced charge separation and observed photocurrent improvement.

Influence of Gd Incorporation on the Chemical, Morphological, and Structural Nature of Spray-Deposited BiVO 4
We now investigate the underlying chemical, morphological, and structural changes associated with the introduction of Gd into our BiVO 4 , which may be responsible for the improvement of the carrier mobility. The presence of Gd in the Gd-doped BiVO 4 films is first confirmed by performing X-ray photoelectron spectroscopy (XPS). Bi 4f, V 2p, O 1s, and Gd 4 d core-level spectra of undoped BiVO 4 , 1% Gd-BiVO 4 , and 3% Gd-BiVO 4 films are shown in Figure 4,5, Supporting Information. Although the Gd 4 d spectra are rather noisy due to the low concentration, Gd can be detected in the Gddoped films, and the introduction of Gd does not introduce any changes to the Bi 4f, V 2p, or O 1s spectra. This is not surprising, since the isovalent substitution of Bi with Gd (i.e., Gd Â Bi ) does not require any ionic or electronic compensations. Based on the XPS spectra, we found that the Gd atomic concentration in the films matches relatively well with the amount introduced into the spray precursor solution. A Gd atomic concentration of 1.3% and 2.9% was calculated for the 1% Gd-BiVO 4 and 3% Gd-BiVO 4 films, respectively.
The Gd incorporation was then evaluated for any influences in the morphology of the BiVO 4 films. Figure 5 shows the top-view scanning electron micrographs of undoped BiVO 4 , 1% Gd-BiVO 4 , and 3% Gd-BiVO 4 films. Similar interconnected irregular grain features can be observed from all films, and the grain size is slightly increased with increasing Gd doping concentrations: 183 AE 58 nm for undoped BiVO 4 , 218 AE 57 nm for 1% Gd-BiVO 4 , and 289 AE 96 nm for 3% Gd-BiVO 4 . An increase in grain size would reduce the amount of grain boundary scattering, and this has been demonstrated to be the cause for an increase in charge carrier mobility and photocurrent. [38,39] However, the continuous increase of grain size with increasing Gd concentration does not agree with the presence of the photocurrent maximum for the 1% Gd concentration. This suggests that the cause behind the enhanced charge carrier mobility and photoelectrochemical performance in our films is unlikely to be related to the morphology change. Finally, the influence of Gd doping on the crystal structure of BiVO 4 films is investigated by performing X-ray diffraction and Raman spectroscopy measurements. Figure 6a shows the grazing incidence X-ray diffraction patterns for undoped BiVO 4 , 1% Gd-BiVO 4 , and 3% Gd-BiVO 4 films. Reference diffraction patterns of monoclinic and tetragonal scheelite BiVO 4 are also shown. The distinction between the tetragonal and monoclinic scheelite is that the local environments of V and Bi ions are more notably distorted in the monoclinic versus tetragonal structure. As a result, two different V─O bond lengths are present in monoclinic scheelite, whereas there is only one V─O bond length in tetragonal scheelite. [40] All our BiVO 4 films show diffraction peaks that match well with those of monoclinic scheelite BiVO 4 without any impurity phases. In contrast to other reports, [26,27] we do not observe a transition from monoclinic to tetragonal scheelite in our BiVO 4 films with Gd doping, which will be discussed later. Figure S6, Supporting Information, shows that the (121) peak position slightly shifts to a higher 2θ angle upon Gd doping. This shift is consistent with the notion of substituting Bi 3þ ions (ionic radius ¼ 1.17 Å) with Gd 3þ ions (1.05 Å), [41] which has also been shown in a report of Gd-doped Bi 2 O 3 powder. [42] A closer look at the {110} and {240} X-ray diffraction peaks (Figure 6b,c) reveals that the monoclinic peak splits further apart with higher Gd concentration, suggesting increased distortion of the crystal lattice with Gd incorporation in our films. To confirm this observation, Raman spectroscopy was performed to shed light on the short-range order and the V─O bond lengths in our films. The Raman spectra are indicated in Figure 7 and the positions of the assigned Raman bands are given in Table S1, Supporting Information. All the Raman peaks agree with those described in previous reports on monoclinic scheelite BiVO 4 . [28,29,43,44] Consistent with the XRD results, the structural change from monoclinic scheelite to tetragonal scheelite, which can be identified by the merging of peaks belonging to the symmetric bending mode δ s ðVO 3À 4 Þ and asymmetric bending mode δ as ðVO 3À 4 Þ into one, [28,29,44] is not observed with Gd incorporation into our films (see Figure 7b). Instead, as the discontinuity in the dashed lines shows, δ s ðVO 3À 4 Þ and δ as ðVO 3À 4 Þ slightly shift to higher wavenumber with Gd incorporation. The same shift to higher wavenumber is also observed for the symmetric and asymmetric V─O stretching mode (ν s (V─O) and ν as (V─O); see Figure 7c). Based on the empirical relationship reported by Hardcastle et al., [45,46] the V─O bond lengths can be calculated using the following equation where is the Raman stretching frequency in cm À1 (symmetric and asymmetric) and R is the V─O bond length in angstrom.  www.advancedsciencenews.com www.solar-rrl.com The calculated V─O bond lengths are tabulated in Table S1, Supporting Information. The values obtained (%1.69 and %1.77 Å) are in very good agreement with those observed previously for BiVO 4 , [40,47] and the incorporation of Gd into our BiVO 4 films induces a slight decrease in the V─O bond length. Indeed, decreased V─O bond length has been shown to correlate with the increased photocatalytic performance of BiVO 4 . [48,49] It was reported that a shorter V─O bond length in the VO 3À 4 tetrahedron provides a stronger packed structure and a greater degree of distortion of the VO 3À 4 tetrahedron, which correlates with the distortion of the Bi─O polyhedron by the lone-pair electron of Bi 3þ in the local structure of BiVO 4 . A higher degree of distortion of the local structure therefore increases the overlap between the Bi 6s and O 2p orbitals, which enhances the photogenerated hole mobility and the overall photocatalytic activity for O 2 evolution. [48,49] We therefore attribute the same effect to our Gd-doped BiVO 4 ; the shorter V─O bond length in the VO 3À 4 tetrahedron provides a higher degree of local structural distortion in monoclinic scheelite BiVO 4 . The optimal degree of distortion is obtained for the 1% Gd-BiVO 4 film, which enhances the charge carrier mobility as observed in our TRMC measurements as well as the resulting photoelectrochemical performance. Finally, we discuss the possible reason for the difference in observation between our Gd-doped BiVO 4 films and those reported in the literature. [26,28,29] Whereas reports in the literature suggest that Gd incorporation induces a transition from monoclinic to tetragonal, we instead show that this is not the case in our spray-deposited BiVO 4 films. One possible reason is that we maintain a relatively low Gd dopant concentration in our study. Another possible reason, which we believe is more likely, is related to the different synthesis procedures, specifically in the way that stoichiometry is maintained in the resulting Gd-doped BiVO 4 . In studies reported in the literature, a Bi:V ratio of 1:1 was maintained in the Gd-doped films. This can be represented by a dissolution reaction of Gd 2 O 3 into the lattice of BiVO 4 , as shown in the Kröger-Vink notation [50] The introduction of Gd is compensated for by the formation of vanadium and oxygen vacancies (V ===== V and V •• O ). These vacancies may affect the crystal structure as well as the photoelectrochemical performance of the resulting BiVO 4 . An alternative defect chemical reaction can be written in the presence of excess oxygen during dissolution, but the formation of vanadium vacancies remains In our study, however, we did not maintain the Bi:V ratio to be 1:1. Instead, we kept the (BiþGd):V ratio as 1:1. This is typically written as a codoping process in the defect chemical equation; i.e., excess V was introduced along with the Gd-doping As shown previously, using this method, the introduction of Gd as dopant is not accompanied by the formation of any vacancies. We therefore speculate that the formation of these vacancies is the driving force for the transition from the monoclinic to tetragonal phase shown in the literature. Overall, this comparison underlines the importance of controlling the stoichiometry in the doping process of complex metal oxides.

Conclusion
In summary, we successfully introduced Gd as dopants in spray-pyrolyzed BiVO 4 films. Gd doping enhances the AM 1.5 photocurrent density, and 1 at% Gd is found to be the optimal concentration, resulting in a twofold improvement of the photocurrent density. Time-resolved microwave conductivity data suggest that the photocurrent improvement is caused by the increase of charge carrier mobility, which in turn results in better charge separation. Detailed structural analysis of the films via X-ray diffraction and Raman spectroscopy measurements reveal that Gd doping introduces further distortion of the monoclinic lattice by the reduction of the V─O bond lengths. This distortion has been correlated to the increase of photogenerated hole mobility, [48,49] which is consistent with our TRMC results. Finally, by

Experimental Section
Spray Pyrolysis of BiVO 4 and Gd-Doped BiVO 4 Thin Films: BiVO 4 and Gd-doped BiVO 4 film photoanodes were prepared by spray pyrolysis, which is a low-cost, facile, and scalable process for depositing BiVO 4 thin films. [51,52] The BiVO 4 precursor solution was prepared by dissolving 0.8 mmol of Bi(NO 3 ) 3 ·5H 2 O (98%, Sigma-Aldrich) in acetic acid and dissolving 0.8 mmol VO(AcAc) 2 (99%, Alfa Aesar) in absolute ethanol. To prevent premature oxidation by trace amounts of water, triethyl orthoformate (TEOF, 98%, Fluka Analytical) was added to the Bi and V solution (TEOF reacts with water to form ethyl formate and ethanol). [53] The V solution was then added to the Bi solution, and the mixture was diluted with excess absolute ethanol to end up with a final precursor concentration of 4 mM. The ratio of acetic acid to ethanol in the final solution was 1:9. For the deposition of 0.5-3 at% Gd-doped BiVO 4 (Bi 1Àx Gd x VO 4 ; x ¼ 0.005, 0.01, 0.02, and 0.03), Gd(NO 3 ) 3 ·6H 2 O (99%, Sigma-Aldrich) was used as the Gd precursors. Varying amounts of Gd(NO 3 ) 3 ·6H 2 O were added to the Bi precursor solution. The V solution was then added to the (BiþGd) solution, resulting in a mixture containing (BiþGd):V in a 1:1 ratio. TEC 7 FTO-coated glass or quartz was used as the substrate, which was cleaned by three successive 15 min ultrasonic cleaning treatments in acetone, absolute ethanol, and deionized water. The substrate temperature during spray was maintained at 450 C. The spray deposition was carried out using an automated spray setup with a Quickmist air atomizing spray nozzle (1/4QMJAU-NC þ SUQR-200) driven by an overpressure of nitrogen gas. The nozzle-to-substrate distance was kept at 20 cm. Each spray cycle consisted of 5 s of spray time and 55 s of delay time to allow solvent evaporation. The precursor solution (180 mL) was sprayed onto the heated substrate with a spray rate of 1-1.5 mL per spray cycle. For films deposited on FTO-coated glass substrates, a thin SnO 2 layer was spray deposited onto the FTO substrate prior to the BiVO 4 deposition using 5 mL of 0.1 M SnCl 4 solution in ethyl acetate, to prevent the recombination of electrons and holes at the FTO/BiVO 4 interface according to the previously reported procedure. [9] As a final step, all of the samples were annealed at 450 C for 2 h with a heating rate of 5 C min À1 in air to further improve the crystallinity.
Characterization: UV-vis spectroscopy was performed using a Lambda 950 spectrophotometer (Perkin Elmer) equipped with an integrating sphere. TRMC measurements were performed using a home-built setup described in more detail previously. [8,37,54] In short, the BiVO 4 % and 1% Gd-BiVO 4 films deposited on quartz were placed in a microwave cavity cell. During the measurements, a change in the microwave power was reflected by the cavity upon nanosecond excitation pulses of a diodepumped Q-switched Nd:YAG laser (coupled with a wavelength tunable optical parametric oscillator, OPO) at a wavelength of 350, 410, and 450 nm. X-band microwaves were generated by a voltage-controlled oscillator (SiversIMA VO3262X). XPS was conducted with a monochromatic Al Kα source (1486.74 eV, SPECS FOCUS 500) and a hemispherical analyzer (SPECS PHOIBOS 100) to study the composition and the oxidation state of the elements. The data were calibrated using the carbon peak at 284.5 eV. Field-emission scanning electron microscopy (FESEM) was performed using a LEO GEMINI 1530 instrument from ZEISS operated with an acceleration voltage of 5 kV. The crystal structure of all samples was studied by X-ray diffraction using a Bruker D8 diffractometer with a Cu Kα radiation (λ ¼ 1.54184 Å) at 40 kV and 40 mA. Grazing incidence configuration was used with an incident angle of 1.0 . The diffraction patterns were recorded in the range of 2θ ¼ 10 -60 at a step size of 0.0225 . Raman spectroscopy was performed using a Horiba HR800 spectrometer equipped with a HeNe laser (λ ¼ 632.8 nm) as the excitation source.
Photoelectrochemical Measurements: Photoelectrochemical measurements were performed in a three-electrode configuration at room temperature, with a platinum wire as the counter electrode, an Ag/ AgCl electrode (XR300, saturated KCl and AgCl solution, Radiometer Analytical) as the reference electrode, and the BiVO 4 as the working electrode (surface area 0.24 cm 2 ). Electrical contact to the BiVO 4 films was provided by a copper wire as well as a conducting electrical tape connected to the exposed FTO substrate. The electrolyte was 2 M potassium phosphate buffer (KP i , pH%7), with and without 0.5 M Na 2 SO 3 as a hole scavenger. Potentials with respect to the reference electrode (V Ag/AgCl ) were applied by a potentiostat (EG&G PAR 273 A) and converted to the reversible hydrogen electrode scale (V RHE ) using the following equation where V 0 Ag=AgCl is the standard potential of the Ag/AgCl reference electrode (0.199 V at 25 C). AM 1.5 photocurrent-voltage curves were measured under backside illumination using a solar simulator (WACOM WXS-50S-5 H, class AAA) with a scan rate of 25 mV s À1 .

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.