2.1.

${\mathsf{BiVO}}_{\mathsf{4}}$ Film Preparation and Characterization

Thin films of ${\mathrm{BiVO}}_4$ were fabricated by ultrasonic spray deposition of an aqueous precursor solution containing Bi, V, and W. The bismuth source was a solution of 0.04 M $\mathrm{Bi}{({\mathrm{NO}}_3)}_3·5{\mathrm{H}}_2\mathrm{O}$ (99.999%, Alfa Aesar, Ward Hill, Massachusetts) in 0.5 M ${\mathrm{HNO}}_3$ (Fisher Scientific, Waltham, Massachusetts). A solution of 0.04 M ${\mathrm{VOSO}}_4·3.5{\mathrm{H}}_2\mathrm{O}$ (99.9%, Sigma-Aldrich, St. Louis, Missouri) in 0.5 M ${\mathrm{HNO}}_3$ was prepared as the vanadium source. Two different tungsten-containing solutions were used in this study. A 0.04 M in W solution was prepared by dissolving silicotungstic acid (STA), ${\mathrm{H}}_4{\mathrm{O}}_{40}{\mathrm{SiW}}_{12}·24{\mathrm{H}}_2\mathrm{O}$ (certified, Fisher Scientific, Waltham, Massachusetts), in 0.5 M ${\mathrm{HNO}}_3$ . A second 0.04 M in W precursor solution was prepared from ${({\mathrm{NH}}_4)}_6{\mathrm{H}}_2{\mathrm{W}}_{12}{\mathrm{O}}_{40}·4{\mathrm{H}}_2\mathrm{O}$ (99.9%, Inframat Advanced Materials, Manchester, Connecticut) in 0.5 M ${\mathrm{HNO}}_3$ . These solutions were then combined in different volumetric ratios to obtain the desired Bi, V, and W contents of the deposited films. Undoped ${\mathrm{BiVO}}_4$ was prepared by mixing equal volumes of the bismuth- and vanadium-containing solutions. Tungsten doping was achieved by adding the STA or ammonium metatungstate (AMT) solution to the mixture. Samples with 5% tungsten doping (calculated as a percentage of the total metal content) were produced by spray depositing precursor solutions formed from three different source mixing ratios— $9\mathrm{Bi}:10\mathrm{V}:1\mathrm{W}$ , $10\mathrm{Bi}:9\mathrm{V}:1\mathrm{W}$ , and $9.5\mathrm{Bi}:9.5\mathrm{V}:1\mathrm{W}$ . Films containing 1% tungsten doping were created from a precursor containing a source mixing ratio of $10\mathrm{Bi}:9.8\mathrm{V}:0.2\mathrm{W}$ .

The aqueous precursor solution was deposited on glass or indium tin oxide (ITO)-coated glass (Delta Technologies). For each deposition, three substrates of approximately $25\times 13\mathrm{mm}$ were placed on a hot plate maintained at 420°C. The precursor solution was nebulized using a 1.7-MHz ultrasonic transducer, and the generated mist (or fog) was transported to a linear scanning spray nozzle using air as the carrier gas. The spray pyrolysis system is similar to that described by Li et al. ¹⁵ except that in our system the nozzle motion is a simple back-and-forth translation. A deposition time of approximately 10 min was required to produce films of the desired ${\mathrm{BiVO}}_4$ film thickness. Once the desired film thickness was obtained, the air flow was stopped and the coated substrates were removed from the deposition apparatus and allowed to cool in air to room temperature. Following deposition, samples were annealed in air (calcined) at 500°C for 3 h to promote formation of the desired monoclinic scheelite phase. Some samples received a second annealing treatment at 375°C in a 3% ${\mathrm{H}}_2$ atmosphere.

Once annealed, samples for PEC testing were prepared by affixing a wire to the exposed ITO substrate using a two-part silver epoxy (Epoxy Technology EE129-4, Billerica, Massachusetts). Samples were then baked for 75 min at 97°C to cure the silver epoxy. Resistance measurements were performed to ensure conductivity between the conducting substrate and the wire. Finally, the wires and exposed conducting substrate were coated with clear insulating epoxy (2 Ton insulating epoxy, Devcon, Danvers, Massachusetts) resulting in ${\mathrm{BiVO}}_4$ photoanode samples with exposed surface areas ranging from 0.3 to $1.2{\mathrm{cm}}^2$ .

An Alpha-Step 500 surface profiler was used to measure the thickness of the resulting films. A LEO 1430VP scanning electron microscope (SEM) with an Oxford energy dispersive x-ray spectrometer system (EDS) was used to analyze the film morphology and chemical composition. Power x-ray diffraction (PXRD) data were collected using Cu $\mathrm{K}\alpha$ radiation on a PANalytical X’Pert PRO powder diffractometer equipped with an X’Celerator detector. Data were collected between 5 and 70 deg $2\theta$ in scanning mode with a step size of 0.03 deg and a count time of 250 s.

2.2.

PEC Measurements

PEC measurements were performed in a three-electrode cell containing a 0.5 M ${\mathrm{Na}}_2{\mathrm{SO}}_4$ (99.5%, Fisher Scientific) electrolyte with a phosphate buffer (pH $7.00\pm 0.02$ at 25°C, Micro Essentials Laboratory, Brooklyn, New York). The ${\mathrm{BiVO}}_4$ photoanodes served as the working electrodes. A 4-M KCl $\mathrm{Ag}/\mathrm{AgCl}$ electrode (Accumet, Fisher Scientific) was used as the reference electrode and a platinum mesh served as the counter electrode. Potential measurements are reported with respect to the $\mathrm{Ag}/\mathrm{AgCl}$ reference electrode.

A Keithley 2400 SourceMeter configured as a potentiostat was used to perform linear sweep voltammetry at a sweep rate of $10\mathrm{mV}{\mathrm{s}}^{-1}$ . Voltammograms were obtained for samples illuminated by simulated solar radiation with an intensity of $0.1\mathrm{W}{\mathrm{cm}}^{-2}$ at the sample surface. The simulated solar illumination was produced with an Oriel 96000, 150-W solar simulator (Newport) with an AM1.5 filter. Photocurrent data with both frontside illumination and through-substrate illumination conditions were collected and compared to dark current measurements. To compare the PEC performance of the different size ${\mathrm{BiVO}}_4$ photoanode samples, the measured current was divided by the area exposed to the electrolyte solution to obtain the photocurrent density, $J(\mathrm{mA}{\mathrm{cm}}^{-2})$ .

3.1.

Film Morphology and Structure

As deposited, undoped samples created from the bismuth and vanadium precursors exhibited the characteristic yellow color of ${\mathrm{BiVO}}_4$ . Samples prepared with precursors containing 1% and 5% (atomic percentage) tungsten were yellow-orange. Following calcining at 500°C, all of the deposited films, including the W-containing films, had an opaque yellow color. Samples that received the subsequent annealing in 3% ${\mathrm{H}}_2$ did not exhibit a significant difference in appearance. Visible and optical microscope examination of the tungsten-doped films revealed irregularities in the film roughness across the sample area. More specifically, some areas of the films appeared rough and nonreflective while others were smoother and more reflective. Such irregularities were far less evident for undoped samples. Streaking, indicative of thickness variations due to small irregularities in the spray nozzle flow pattern, was also observed on all of the samples.

The resulting films exhibited a unique microstructure, as seen in the representative SEM images in Fig. 1. Figure 1(a) shows an undoped, calcined, and ${\mathrm{H}}_2$ -annealed sample. A high-surface-area coral-like structure comprising connected hollow spheres and spherical fragments was evident. This most likely resulted from the rapid drying and decomposition of the ultrasonically nebulized precursor droplets and is similar to particle architectures observed by Dunkle et al. ²⁰ Surface profile measurements indicated average film thicknesses between 1.7 and 2.2 μm with significant variations due to the complex morphology shown in the SEM images in Fig. 1.

Fig. 1

Scanning electron microscope images of the ${\mathrm{BiVO}}_4$ films created from precursor ratios of (a)  $10\mathrm{Bi}:10\mathrm{V}$ , (b)  $10\mathrm{Bi}:9\mathrm{V}:1\mathrm{W}$ , (c)  $9.5\mathrm{Bi}:9.5\mathrm{V}:1\mathrm{W}$ , and (d)  $9\mathrm{Bi}:10\mathrm{V}:1\mathrm{W}$ . The tungsten source for all of the films shown was silicotungstic acid (STA). Arrows in (d) indicate rod-like structures in the film. The scale bar shown represents a 5.0-μm length.

Figures 1(b)–1(d) show representative SEM images of calcined and ${\mathrm{H}}_2$ treated films doped with 5% (atomic percentage) W from precursor ratios of (b) $10\mathrm{Bi}:9\mathrm{V}:1\mathrm{W}$ , (c) $9.5\mathrm{Bi}:9.5\mathrm{V}:1\mathrm{W}$ , and (d) $9\mathrm{Bi}:10\mathrm{V}:1\mathrm{W}$ using STA as the tungsten source. No morphology changes attributable to calcining or ${\mathrm{H}}_2$ annealing were detected. Although all of the deposited films exhibited similar morphologies, small variations in the W-doped films were observed. In general, tungsten-doped films exhibited fewer and smaller hollow partial sphere structures compared to undoped samples. The most significant morphology difference was observed in Fig. 1(d), where small rod-like structures, indicated with arrows, were present throughout the film. SEM studies of all of the films indicated that such rod-like structures were present near the edges of W-doped ${\mathrm{BiVO}}_4$ films where spray coverage was minimal and the film was sparse. However, these structures were not found in the more complete coverage areas of films prepared from the $10\mathrm{Bi}:9\mathrm{V}:1\mathrm{W}$ and $9.5\mathrm{Bi}:9.5\mathrm{V}:1\mathrm{W}$ precursor solutions, suggesting that this morphology is unique to the vanadium-rich $9\mathrm{Bi}:10\mathrm{V}:1\mathrm{W}$ composition and forms when vanadium is in excess.

EDS measurements of the film compositions indicated a close relationship between the precursor composition and the atomic percentage composition of the films produced. Spectral analysis of the rod-like structures observed in Fig. 1(d) confirmed that they were vanadium and tungsten rich.

PXRD measurements of undoped samples deposited on glass, shown in Fig. 2, indicated that after calcining the films likely comprised monoclinic scheelite ${\mathrm{BiVO}}_4$ (PDF#14-0688). PXRD pattern (a) of an as-deposited film prior to calcining suggested a scheelite tetragonal structure (PDF#75-2481) and showed the presence of unidentified phases. This indicates that while the solution delivered the components to the film, they do not react completely during the deposition process. Following annealing in air at 500°C, the PXRD patterns of undoped films, such as the one shown in pattern (b), matched the monoclinic scheelite phase, showing that annealing is required to facilitate the reaction and produce ${\mathrm{BiVO}}_4$ of the desired phase. Pattern (c) was obtained from an undoped film following the ${\mathrm{H}}_2$ reducing treatment. No significant changes were observed in the PXRD patterns of undoped films that received the 3% ${\mathrm{H}}_2$ annealing treatment, suggesting that the treatment did not significantly alter the film structure or composition. Samples with low levels of doping, such as shown in pattern (d), maintain the monoclinic scheelite structure.

Fig. 2

PXRD patterns of ${\mathrm{BiVO}}_4$ thin films samples from a $10\mathrm{Bi}:10\mathrm{V}$ (equimolar) precursor (a) as deposited, (b) calcined at 500°C, and (c) following ${\mathrm{H}}_2$ annealing. Patterns (a)–(c) were collected on the same sample. PXRD patterns of W-doped films that were calcined and ${\mathrm{H}}_2$ annealed following spray deposition with precursor mole ratios of (d)  $10\mathrm{W}:9.8\mathrm{V}:0.2\mathrm{W}$ (AMT), (e)  $10\mathrm{Bi}:9\mathrm{V}:1\mathrm{W}$ (STA), (f)  $9.5\mathrm{Bi}:9.5\mathrm{V}:1\mathrm{W}$ (STA), and (g)  $9\mathrm{Bi}:10\mathrm{V}:1\mathrm{W}$ (STA).

Increasing the amount of tungsten doping on the vanadium site causes the material to crystallize in the tetragonal scheelite structure (PDF#75-2481), ^{21 , 22} which is closely related to the structure of monoclinic scheelite. The scheelite phase ( ${\mathrm{AMO}}_4$ ) contains an eight-coordinate cubic ${\mathrm{Bi}}^{3+}$ site (A site) and a four-coordinate tetrahedral ${\mathrm{V}}^{5+}$ site (M site). Previous work shows that hexavalent ions such as ${\mathrm{Mo}}^{6+}$ and ${\mathrm{W}}^{6+}$ dope exclusively on the M site, ^{21 , 22} and in the process create a vacancy on the A site to maintain charge balance. This is consistent with ionic radii: four-coordinate ${\mathrm{V}}^{5+}$ (49.5 pm) and ${\mathrm{W}}^{6+}$ (56 pm) have much smaller radii than eight-coordinate ${\mathrm{Bi}}^{3+}$ (131 pm). ²³ Increasing the amount of substitution on the M site causes the material to undergo the monoclinic to tetragonal distortion. This distortion is not seen for low amounts of substitution, but is observed for larger amounts of substitution. Sleight et al. observe this transformation at $x=0.10$ for ${\mathrm{Bi}}_{1-x}{\varphi }_x{\mathrm{V}}_{1-3x}{\mathrm{M}}_{3x}{\mathrm{O}}_4$ ( $\varphi =\text{vacancy}$ ), a 30% substitution. In the same work, they show that doping on the A-site does not produce this distortion. ²² The shape of the peaks near 35 and 47 deg $2\theta$ can be used to distinguish between the monoclinic and tetragonal phases of ${\mathrm{BiVO}}_4$ . Tetragonal ${\mathrm{BiVO}}_4$ has a single (020) reflection near 35 deg; a change to the lower-symmetry monoclinic crystal system causes this peak to separate into two distinct reflections: (200) and (020). Similarly, at 47 deg $2\theta$ , the (024) reflection for the tetragonal cell splits into three distinct reflections: (060), (240), and (042).

The film in pattern (e) was obtained from a calcined and ${\mathrm{H}}_2$ annealed film deposited with a precursor composition ratio of $10\mathrm{Bi}:9\mathrm{V}:1\mathrm{W}$ using STA as the tungsten source. In pattern (e), two impurity peaks were observed at 28 and 32.5 deg $2\theta$ . More importantly, the single peaks observed near 35 and 47 deg $2\theta$ (marked with asterisks), indicate the presence of tetragonal scheelite phase. The 10% doping on the V site ( $x=0.033$ ) is a lower doping level than thought necessary to produce this distortion. Since processing conditions are known to affect the amount of doping in these materials, ^{24 , 25} it would not be surprising if they also affected the underlying structure. Excess Bi should be observed in this film. However, it is common for impurity phases that are $<5\%$ of the film not to be observed in PXRD data, or the excess Bi could be in an amorphous phase or in the unidentified impurity peaks. The peak shapes from pattern (f), obtained from a calcined and ${\mathrm{H}}_2$ -annealed film deposited from a precursor composition ratio of $9.5\mathrm{Bi}:9.5\mathrm{V}:1\mathrm{W}$ using STA as the tungsten source, were consistent with a tetragonal cell. No additional peaks were observed, suggesting that the tungsten was incorporated in the host material or is present as an amorphous phase.

Figures 3(a) and 3(b) show a detailed comparison of the reference and sample patterns (c), (d), and (f) near 35 and 47 deg $2\theta$ , respectively, that were used to determine the crystalline phase compositions of the samples. The decrease in the number and the sharpening of the peaks as W is added indicates that the phase transition from the monoclinic to the tetragonal phase is easily seen in this figure. Allowing for broadening caused by the size of the crystallites, (c) matches the monoclinic pattern. Although (d) clearly shows the monoclinic structure, the difference in the peak shapes most likely results from the change in lattice parameters upon W doping. The presence of small amounts of the tetragonal phase in (d) or the monoclinic phase in (f) cannot be ruled out, and it is possible that these samples contained mixed phases. Even so, it is clear that the spray pyrolysis film changes from monoclinic to tetragonal as W is add to the precursor solution.

Fig. 3

PXRD patterns of samples (c), (d), and (f) from Fig. 2 near (a) 35 and (b) 47 deg $2\theta$ compared to the reference patterns of the scheelite tetragonal (PDF#75-2481) and scheelite monoclinic (PDF#14-0688) phases.

Finally, pattern (g) of Fig. 2 was obtained for a calcined and ${\mathrm{H}}_2$ -annealed film deposited from a precursor composition ratio of $9\mathrm{Bi}:10\mathrm{V}:1\mathrm{W}$ using STA as the tungsten source. The peaks near 35 and 47 deg $2\theta$ were determined to be consistent with the tetragonal phase. Additional peaks at 13, 20, 27, and 28 deg $2\theta$ are observed. As with patterns (a) and (e), the additional peaks indicate the presence of an impurity phase that could not be fully explained by patterns in the database. These peaks may result from the rod-like structures observed in the SEM images. Based on the EDS analysis of the rod-like morphologies observed in Fig. 1(d) and the precursor composition, it is believed that these peaks may result from a phase or phases rich in vanadium and tungsten. This is consistent with the crystal chemistry of this system. Since tungsten cannot substitute on the bismuth site, and the combined V plus W content is well in excess of the amount of bismuth, they could crystallize as the impurity phase observed by PXRD.

3.2.

PEC Study Results

Figure 4 shows representative linear sweep voltammograms obtained for the ${\mathrm{BiVO}}_4$ photoanodes. Black lines represent the current density obtained when the sample was illuminated from the electrolyte interface (front) side. Gray lines indicate the current density obtained when the photoanode was illuminated through the substrate (back) side. Frontside and backside illumination studies were performed to provide insight into charge transport properties. ¹² Voltammograms (a) and (b) were obtained from undoped, calcined, and ${\mathrm{H}}_2$ -annealed photoanode samples created using the same precursor preparation and deposition procedures. Trace (c) shows the dark current density of the samples, illustrating that the current produced by samples (a) and (b) was from simulated solar light exposure. Sample (a) was one of the best-performing samples created using the described fabrication methods. It produced the largest frontside illumination photocurrent density of all samples tested in this study, particularly at low-bias voltage conditions. Sample (b) was more typical of the PEC response of undoped and ${\mathrm{H}}_2$ treated samples, exhibiting a larger photocurrent when illuminated from the substrate side of the film at high-bias voltages.

Fig. 4

Linear sweep voltammograms of ${\mathrm{BiVO}}_4$ photoanodes with frontside (black) and backside (gray) illumination. Scans (a) and (b) were obtained from undoped, calcined, and ${\mathrm{H}}_2$ annealed samples and (c) is the dark current scan. Scans (d)–(f) were obtained from W-doped, calcined, and ${\mathrm{H}}_2$ annealed samples deposited from solutions of (d) $10\mathrm{Bi}:9\mathrm{V}:1\mathrm{W}$ , (e) $9.5\mathrm{Bi}:9.5\mathrm{V}:1\mathrm{W}$ , and (f) $9\mathrm{Bi}:10\mathrm{V}:1\mathrm{W}$ . AMT was the W source for these samples. Scan (g) shows the dark current measurements for samples (d)–(f).

The remaining traces show example voltammograms obtained for 5% W-doped ${\mathrm{BiVO}}_4$ photoanodes that were calcined and ${\mathrm{H}}_2$ annealed. AMT was used as the W source for these samples. Note the difference in the photocurrent density axis scale relative to that for plots (a)–(c). Curves (d), (e), and (f) were obtained from samples deposited from $10\mathrm{Bi}:9\mathrm{V}:1\mathrm{W}$ , $9.5\mathrm{Bi}:9.5\mathrm{V}:1\mathrm{W}$ , and $9\mathrm{Bi}:10\mathrm{V}:1\mathrm{W}$ precursor solutions, respectively. Plots (d) and (e) exhibited similar frontside and backside illumination responses, while sample (f) showed a significant photocurrent enhancement with backside illumination.

Over 40 photoanode samples were fabricated during the course of this study to evaluate the repeatability of the ultrasonic spray deposition process to form ${\mathrm{BiVO}}_4$ photoanodes with consistent PEC performance. Figure 5 presents a comparison of the measured PEC performance for 27 of these samples. In Fig. 5, each bar represents the photocurrent density measured at three different bias voltages: 0.5 V (black bars), 1.0 V (dark gray bars), and 1.5 V (light gray bars) with respect to the $\mathrm{Ag}/\mathrm{AgCl}$ reference electrode. Because the photocurrent density was measured for both frontside and backside illuminations, each sample is represented by a pair of bars. The bar on the left of each pair corresponds to the observed frontside illumination photocurrent while the bar on the right represents the measurements for substrate side illumination. Samples are grouped according to their composition, as indicated by the text above and below the bar groupings. The group of three samples at the right side of the graph, denoted 5% W, represent samples doped with 5% W from different precursor ratios that did not receive the ${\mathrm{H}}_2$ treatment. The group of two undoped samples near the right side of the graph also did not receive the ${\mathrm{H}}_2$ treatment. The parenthetical lower-case symbols denote the samples with voltammograms presented in Fig. 4. Finally, the upper case A and S above the sample bars denote AMT or STA, respectively, as the W source in the precursor solution.

Fig. 5

A comparison of the measured PEC performance of samples produced throughout the course of this study. Each pair of bars represents the photocurrent density measured under frontside (left) and substrate side (right) illumination conditions with 0.5-V (black bars), 1.0-V (dark gray bars), and 1.5-V (light gray bars) bias with respect to the $\mathrm{Ag}/\mathrm{AgCl}$ reference electrode. Samples are grouped according to their composition and postdeposition treatment conditions. Parenthetical lowercase letters correspond to the voltammograms presented in Fig. 4. The uppercase A and S letters designate AMT and STA as the W doping source, respectively.

Note that Fig. 5 provides a summary of representative results obtained from the linear sweep voltammetry investigation of the samples. Multiple voltammograms were acquired for each photoanode sample nonsequentially over the course of several months to verify the repeatability of the measurements. Results from repeated voltammetry scans of the same sample showed repeatable PEC performance with little to no degradation of the test samples.

The results presented in Fig. 5 clearly demonstrate that the hydrogen annealing treatment resulted in higher photocurrent production. Undoped samples that received the ${\mathrm{H}}_2$ treatment produced, on average, the greatest photocurrent densities, as shown by the first group of bars in the chart. However, while two of the undoped and ${\mathrm{H}}_2$ treated ${\mathrm{BiVO}}_4$ samples produced significant photocurrents at both low- and high-bias conditions, variability in the PEC performance of these samples is evident, even though all six of the undoped and ${\mathrm{H}}_2$ treated samples were produced using the same precursor preparation and deposition techniques.

The results presented in Fig. 5 also show that 5% tungsten doping did not produce the expected enhancement in PEC performance. This result is especially evident when comparing the PEC results from the doped and undoped air annealed samples. Hydrogen annealing improved the performance of the W-doped samples. However, significant performance variability existed between the different precursor combinations and within the duplicate samples. Of the 5% W-doped and ${\mathrm{H}}_2$ -treated samples, those produced from a precursor composition containing $9.5\mathrm{Bi}:9.5\mathrm{V}:1\mathrm{W}$ provided the greatest photocurrent densities. Even in this case, tungsten doping produced no improvement over the undoped, ${\mathrm{H}}_2$ -annealed photoanodes. Significant performance variability between samples produced with this composition is evident, particularly for films where STA was used as the W source.

Also of note is the difference in photocurrent production between frontside and backside illuminations. With the exception of photoanodes produced from the $9.5\mathrm{Bi}:9.5\mathrm{V}:1\mathrm{W}$ and $9\mathrm{Bi}:10\mathrm{V}:1\mathrm{W}$ , the frontside illumination photocurrent density was larger than the backside illumination photocurrent density at low-bias conditions. This suggests that at low bias the PEC performance was limited by low hole mobility. However, at high bias, many of these samples exhibited improved backside illumination photocurrent densities, suggesting that the high electric field improved hole transport and that frontside high-bias performance was limited by poor electron transport through the film.

The photocurrent onset potentials for the undoped ${\mathrm{BiVO}}_4$ photoanode samples under frontside illumination ranged from $-0.332$ to $-0.293\mathrm{V}$ for both samples that received the ${\mathrm{H}}_2$ treatment and those that did not. Similarly, all 5% W-doped samples, irrespective of ${\mathrm{H}}_2$ treatment, exhibited photocurrent onset potentials under frontside illumination ranging from $-0.5$ to $-0.494\mathrm{V}$ . The backside illumination photocurrent onset potentials exhibited similar behavior but with a slight (5 to 10 mV) increase in the potential. This increase is presumed to be due to reflection and absorption losses in the glass and ITO substrate.

In an attempt to better quantify the flat-band potentials of the photoanodes, a Mott-Schottky analysis was explored. Figure 6 shows the Mott-Schottky plots obtained by measuring the capacitance of the film at different bias voltages with respect to the $\mathrm{Ag}/\mathrm{AgCl}$ reference electrode when immersed in a 0.5-M ${\mathrm{Na}}_2{\mathrm{SO}}_4$ electrolyte with no illumination. A 1000 Hz, 10-mV sinusoidal perturbation of the bias voltage was used to measure the film capacitance measurements. Figure 6(a) shows the resulting Mott-Schottky plot for a sample deposited from a $10\mathrm{Bi}:9\mathrm{V}:1\mathrm{W}$ precursor ratio that was only calcined. Figure 6(b) shows the same plot for an undoped sample and three W-doped samples that were calcined and annealed in ${\mathrm{H}}_2$ . Note the difference in the ${\mathrm{C}}^{-2}$ scale for figures (a) and (b). The resulting data exhibited nearly continuous curvature with no well-defined linear region. Although not shown here, the Mott-Schottky measurements also exhibited frequency dependence, resulting in different functionalities when tested with a 500- and 200-Hz perturbation. This is believed to result from the nondense, highly microporous morphology of the films. ¹¹ Because of this, conclusive flat-band potential estimates could not be ascertained. Visual inspection of the resulting plots suggests that the W-doped samples, irrespective of ${\mathrm{H}}_2$ treatment, may have a more negative flat-band potential, consistent with the photocurrent onset potential measurements. Qualitative information about the carrier densities in the films could be evaluated. The Mott-Schottky measurements for the ${\mathrm{H}}_2$ annealed samples in Fig. 6(b) exhibited a significantly reduced slope when compared to that of the non- ${\mathrm{H}}_2$ annealed sample in Fig. 6(a), suggesting greater carrier concentrations in the ${\mathrm{H}}_2$ annealed samples.

Fig. 6

Mott-Schottky plots of (a) a 5% W-doped samples that did not receive the ${\mathrm{H}}_2$ treatment and (b) one undoped and three 5% W-doped samples that received the ${\mathrm{H}}_2$ -annealing treatment.