The Self-Passivation Mechanism in Degradation of BiVO4 Photoanode.

BiVO4 is a promising photoanode material for solar-assisted water splitting in a photoelectrochemical cell but has a propensity to degrade. Investigations carried out here in 0.1 M Na2SO4 electrolyte showed that degradation is by dissolution of V in the electrolyte while Bi is retained on the anode probably in the form of solid Bi oxide (Bi2O3, Bi4O7). Accumulation of Bi oxide on the anode surface leads to passivation from further degradation. Thermodynamic modeling of possible degradation reactions has provided theoretical support to this mechanism. This self-passivation is accompanied by a decrease in photocurrent density, but it protects the anode against extensive photocorrosion and contributes to long-term stability. This is a more definitive understanding of degradation of BiVO4 during water splitting in a photoelectrochemical cell. This understanding is imperative for both fundamental and applied research.

Introduction

Artificial photosynthesis, which captures and stores solar energy in the chemical bonds of a fuel, is a potential solution to the energy storage problem (Barber, 2009). Solar-assisted water splitting to produce hydrogen fuel has received significant research interest in this regard (Gust et al., 2009, Lewis and Nocera, 2006, Maeda and Domen, 2010, Sivula and van de Krol, 2016). The photoelectrochemical cell (PEC) is a versatile tool for photolysis of water where a semiconductor is used to harvest solar energy and an external bias is applied to facilitate the water splitting reactions (Blankenship et al., 2011, Grätzel, 2001, Li et al., 2013, Walter et al., 2010). Such electrochemical water photolysis was first reported by Fujima and Honda using a TiO2 (band gap = 3.0–3.2 eV) photoanode (Fujishima and Honda, 1972). For more efficient photolysis, a narrower bandgap and a valence band edge above 2.0 eV versus reversible hydrogen electrode (RHE) is mandatory to provide sufficient overpotential for holes to oxidize water at the anode. Simultaneously, a negative conduction band edge is required at the cathode for electrons to reduce water. The efficiency of light absorption is principally determined by the bandgap of the semiconductor, which is the basis of theoretical calculations of solar to hydrogen conversion efficiency. The bandgap is also a measure of the stability of a compound, and thus semiconductors with narrow bandgaps are usually vulnerable to degradation in photoelectrodes (Grätzel, 2001).

BiVO4 photoanode has a narrower bandgap of 2.4 eV versus RHE, which contributes to a high, theoretical solar-to-hydrogen conversion efficiency of 9.2% (Gan et al., 2014, Park et al., 2013). This remarkable efficiency has drawn tremendous research interest, yet its vulnerability to photocorrosion has been its weakness. BiVO4 suffers from chemical instability in both acidic and alkaline conditions. Besides, its degradation is notably promoted by applied bias and light illumination when used as a photoanode in PEC (Lichterman et al., 2013, McDowell et al., 2014). Researchers have examined mainly three strategies to protect BiVO4 electrodes against photocorrosion:

Adopt an external passivation layer on the photoanode to avoid direct contact with the electrolyte (Fan et al., 1983, Hu et al., 2014, McDowell et al., 2014),

Use co-catalysts to alter the thermodynamic reduction/oxidation potential of photo-generated holes/electrons, to facilitate oxygen/hydrogen evolution reactions instead of detrimental side reactions (Kim and Choi, 2014, Seabold and Choi, 2012, Zhong and Gamelin, 2009),

Manipulate the electrolyte composition to retard dissolution. For example, when the electrolyte is saturated with V ions, dissolution of BiVO4 photoanode is suppressed (Lee and Choi, 2017).

Considerable experimental effort has been devoted to the enhancement of stability via the three strategies, but the effort expended in understanding the degradation mechanism is limited. Several studies have fragmentally indicated the anodic photo-degradation propensity of the V ion in BiVO4 (Ding et al., 2013, Lee and Choi, 2017, McDowell et al., 2014), which probably prompted Choi et al. to saturate the electrolyte with V ions to retard its degradation (Lee and Choi, 2017). Although, this approach reduced photodegradation, it did not throw any light on the degradation mechanism. A recent study by Toma et al. did confirm the dissolution of V from BiVO4, but it concluded that the rates of dissolution of both Bi and V would eventually become stoichiometric upon extensive corrosion (Toma et al., 2016). In this paper we propose a variation to this mechanism of degradation by in-depth mechanistic and theoretical studies that will contribute to the fundamental understanding. BiVO4 has emerged as a versatile platform for understanding the photoelectrochemical behavior of transition metal oxide semiconductors (Sharp et al., 2017), where the observed material behavior could potentially be applied to similar types of materials.

Results

BiVO4 is usually used after doping with a higher valence metal ion (than V), such as Mo or W, to improve its electron conduction and charge separation (Zhao et al., 2018a, Zhao et al., 2018b). In this work, we adopted n-type doping by 3 at% Mo, which enhances the photocurrent density (Luo et al., 2011). A 0.1 M Na2SO4 aqueous solution at pH = 5.7 was used as the electrolyte unless stated otherwise. The absorbance, reflectance, and transmittance data of the anode are in Figure S1. Performance of the photoanode was evaluated using linear scanning voltammetry (LSV) and chronoamperometry. Typical results obtained in a trial (Figure 1A) show an onset potential of ∼0.6 V versus RHE and a photocurrent density of ∼1.0 mA cm−2 at 1.23 V versus RHE (corresponding dark current is shown in Figure S2). Figure 1B shows the decline of the photocurrent (at 1.23 V versus RHE) normalized to the initial value with time, for 50 h of continuous PEC testing. The photocurrent decreased sharply to about 25% of the initial value in 4 h followed by a gradual decrease till the end of the test. This transition from short-term instability, in the time range 0–4 h, to long-term stability will be discussed later.

Figure 1

Performance and Stability of Mo-BiVO4 Photoanode

(A) LSV curve of the Mo-BiVO4 photoanode test under AM 1.5 G solar illumination and 1.23 V versus RHE bias, in 0.1 M Na2SO4 electrolyte (pH = 5.7). Dark current has already been netted from the photocurrent.

(B) Chronoamperometry of the test in (A) for continuous 50-h run.

(C) Chronoamperometry of Mo-BiVO4 photoanode under AM 1.5 G solar illumination and 1.23 V versus RHE bias in 0.1 M KPi solution (pH = 10), with and without dissolved V2O5 species.

Trials were done in 0.1 M KPi electrolyte at pH = 10 saturated with V5+ ions, and without these ions, to assess the effect of pre-existing V ions on the degradation of the Mo-BiVO4 photoanode. Saturation with V5+ ions was achieved by dissolving 0.1 M V2O5 powder in the electrolyte. A different electrolyte and pH became necessary to dissolve adequate V2O5 to saturate the solution with V5+. Figure 1C shows that photocurrent in the V5+ saturated electrolyte became stable in 10 min, whereas that in the V-free electrolyte decreased continuously and eventually reached zero. Therefore, saturating the electrolyte with V5+ ions enhances the stability of Mo-BiVO4 photoanode.

After running the photoanode samples for different times under dark and illuminated conditions, concentrations of Bi, V, and Mo ions in the electrolyte were measured using an Inductively Coupled Plasma–Mass Spectrometer (ICP-MS). The applied bias, illumination, and temperature were controlled well so that they were not variable parameters. From the measured concentrations (see Table S1), average anode dissolution rates were calculated assuming spatially uniform dissolution over the entire photoelectrode (see Supplemental Information for calculation details). This gives different values when calculated for the concentrations of each of the elements Bi, V, and Mo. The initial thickness of the Mo-BiVO4 film was 100 nm, which was reported previously (Yao et al., 2018). Tracking the dissolution rate through dissolved Mo could be inaccurate because some Mo could have deposited on the surface of BiVO4 anode as MoO3 during the anode synthesis (Luo et al., 2012). Yet, Mo could be used to calibrate V since it substitutes for V in the BiVO4 crystal. The dissolution rates of Mo and V were consistent throughout, indicating the validity of these ICP-MS measurements.

Table 1 shows the uniform dissolution rates calculated from elemental concentrations, and the prevailing Bi:V atomic ratio in the electrolyte after some tests. It is striking that the Bi:V atomic ratio deviates significantly from the stoichiometric ratio 1:1 present in the BiVO4 anode. There is excessive V (and Mo) compared with Bi in the electrolyte. The purely chemical dissolution rates are given by the dark and open circuit (E) conditions, whereas the “PEC” condition indicates the photo-corrosion rates under both applied bias (1.23 V. versus RHE) and illumination. The chemical dissolution rates of V and Mo are higher than that of Bi by several orders of magnitude. The corrosion rates of all elements increase when either illumination or bias is applied, yet the Bi dissolution is always lower by a significant margin. Hence, V and Mo are more susceptible to dissolution under all conditions tested here. Table 1 shows the average thickness affected in the photoanode calculated using the rates of V dissolution. Interestingly, the affected thickness for three (first, fourth, and fifth) samples are very similar, even though their testing conditions are very different implying that the degradation has met strong resistance after an initial transient affecting a certain thickness. Such a resistance must have developed internally within the photoanode itself. This is supported by the chronoamperometry curve in Figure 1B, which shows that the photocurrent subsides to a steady value after a sharp reduction in the first 4 h.

Table 1

Dissolution Parameters Computed from ICP-MS Results of the Electrolyte after Photolysis under Different Conditions in 0.1 M Na2SO4 at pH = 5.7

Light Illumination	Applied Bias (V versus RHE)	Duration (h)	Dissolution Rate on Bi Basis (nm h−1)	Dissolution Rate on V Basis (nm h−1)	Dissolution Rate on Mo Basis (nm h−1)	Bi:V	Affected Thickness (nm)
Dark	EOC	99	2.4 × 10−6	0.19	0.17	1:78,750	18.7
Dark	1.23	4	4.0 × 10−3	2.78	0.91	1:695	11.1
Light	EOC	4	6.5 × 10−4	1.01	4.48	1:1,554	4.0
Light	1.23	4	0.19	4.56	4.64	1:24	18.2
Light	1.23	50	3.1 × 10−5	0.45	1.20	1:14,407	22.5

See also Tables S1–S3. The affected thickness was calculated based on V dissolution rate. The calculation of dissolution rates and affected thicknesses are explained in Supplemental Information. The open circuit voltage EOC is shown in Figure S3 for both dark and light conditions.

For comparison, another set of photolysis trials was done using 0.1 M Na2SO4 electrolyte at pH = 7.0. The results are shown in Table S2. Decrease in the degradation rates is evident compared with pH = 5.7 probably due to relieved corrosion under the neutral condition. The calculated affected thicknesses for 4 h PEC and 90 h soaking conditions are similar indicating the development of self-resistance as for the case with pH = 5.7. The corresponding data for corrosion experiments conducted in pH = 8.3 electrolyte is shown in Tables S1 and S3. It must be noted that Toma et al. also examined the degradation of BiVO4 in varied pH solutions. They too found the dissolution of V to be much faster than Bi in some pH electrolytes (Toma et al., 2016).

These results show that the degradation of Mo-BiVO4 in 0.1 M Na2SO4 at pH = 5.7 and 7.0 is due to the preferential dissolution of V (and Mo) compared with Bi. Then, one needs to determine the destiny of the Bi atoms released from the BiVO4 crystals in the anode film as it degrades. Precipitation of solids was not detected in the electrolyte. This implies that Bi probably remains on the anode. Therefore, we scrutinized the anodes after the test runs by X-Ray Photoelectron Spectroscopy (XPS) for possible changes in its chemical composition. A summary of the calculated XPS surface chemical compositions is given in Table 2. The stoichiometric ratio of Bi:V = 1:1 was confirmed in pristine Mo-BiVO4 to validate the XPS technique. After 4 h of PEC testing in pH = 5.7 electrolyte, the Bi:V ratio increased to approximately 1.5:1.0; after 50 h it increased to about 2:1. These results compliment the ICP-MS results, which showed negligible levels of Bi in the electrolyte, and confirm our hypothesis that Bi remains on the anode as it degrades. We attribute this Bi enrichment on the anode surface to the observed resistance to degradation, which amounts to self-passivation. Equation 1 shows a possible self-passivation reaction, where the solid Bi2O3 product could shield the surface of the BiVO4 and protect it from further corrosion.

Table 2

XPS Quantitative Analysis of the Surface Composition of Different Photoanode Samples after Test Runs in 0.1 M Na2SO4 Electrolyte at pH = 5.7

Sample Condition	Bi (at.%)	V (at.%)	Mo (at.%)
Pristine	48.5	48.5	3.1
After 4 h degradation	58.7	38.2	3.1
After 50 h degradation	63.9	31.6	4.5

This reaction, noted as R1, could be enabled by specific conditions, which will be clarified later in a theoretical discussion. Figure 2 shows the relevant high-resolution XPS spectra of Bi, V, and Mo. Shifts in the Bi 4f peaks evident in Figure 2A after degradation indicate that the product of degradation has a slightly lower binding energy (BE), signifying the formation of different Bi compounds on the surface, which probably constitutes the solid passivation layer. The BE of the Bi 4f7/2 and Bi 4f5/2 peaks in pristine Mo-BiVO4 was measured to be 159.1 and 164.4 eV, respectively, consistent with a previous report (Liu et al., 2017). After long-term photodegradation, these peaks shifted to 158.9 and 164.2 eV, respectively. Figure 2B shows a shoulder around 531–534 eV in O 1s peak in the pristine Mo-BiVO4, which is diminished after degradation. This was observed by previous workers also and was attributed to BE = 531.5 eV of the adsorbed O-H species on the surface of the as-fabricated anode (Qin et al., 2014, Zhang et al., 2014). The changes evident in the Mo spectrum of Figure 2C could be attributed to the dissolution of some MoO3 that might have formed on the surface of Mo-BiVO4 anode during the anode fabrication (solubility of MoO3 is 0.490 g/100 mL at 28°C) (Liu et al., 2017).

Figure 2

Characterization of Pristine Mo-BiVO4 Photoanode and after Photolysis for 50 h in 0.1 M Na2SO4 Electrolyte at pH = 5.7

(A–C) High-resolution XPS spectra of Bi, V and O, and Mo.

(D) Nyquist plots.

Results of Electrochemical Impedance Spectroscopy (EIS) on the photoanode before and after photolysis are shown in Figure 2D. The diameter of the semicircle increased significantly after degradation, signifying an increase in the interfacial charge transfer resistance (R) in the photoelectrode. Since all testing conditions remained unchanged, this increase of R is likely due to a passivation layer, such as B2O3, which could increase the barrier for hole transport to the electrolyte at interface due to its energetically deeper valence band (Myung et al., 2011).

Figure 1B shows that 75% of the initial photocurrent was lost in the first 4 h, but the thickness affected, calculated using V concentration, is only 20% of that affected in 50 h. Therefore, we could attribute the loss of photocurrent in the first 4 h to charge transport retardation caused by the formation of the passivation layer rather than to rapid dissolution of BiVO4.

Figure 3 shows the scanning electron micrographs of the photoanode before and after photolysis. Its porous nanostructure is evident. The surface before photolysis is relatively smooth, whereas after photolysis for 50 h (Figure 3B), it is degraded and rougher. However, its characteristic nanoporous structure is preserved. The roughness increase is probably caused by non-uniform degradation because of protection at specific locations from passivation. Therefore, microscopically, the photodegradation is not uniform across the illuminated area.

Figure 3

Scanning electron microscopy Images of the Mo-BiVO4 Photoanode

(A) Pristine sample.

(B) Sample after photolysis for 50 h in 0.1 M Na2SO4 electrolyte at pH = 5.7.

See also Figure S4.

With these determinative proofs, we could confirm the advent of a self-passivation mechanism in the photodegradation of Mo-BiVO4 photoanode. Next, we employ computational modeling to validate the mechanism as well as to predict degradation products by theoretical means. First, we model the dark condition. The thermodynamic stability of pure Mo-BiVO4 was simulated via a Pourbaix diagram established by Toma et al., which allows prediction of thermodynamic equilibrium and decomposition products (Hu et al., 2018, Persson et al., 2012, Pourbaix, 1966, Toma et al., 2016).

Figure 4A shows that Mo-BiVO4 is thermodynamically stable only in relatively neutral electrolyte and at low external voltages. It would degrade in acidic or basic conditions, or when the voltage approaches the oxygen evolution potential. In comparison with the modeling results on pure BiVO4, Mo-doping did not cause any material differences in its decomposition manner.

Figure 4

Thermodynamic Modeling Results

(A) The Pourbaix diagram of 50%–48%–2% Bi–V–Mo system in aqueous solution, assuming Bi, V, and Mo ion concentration at 10−8 mol kg−1. Regions are labeled for stable phase(s) of: A – MoO3(s) + Bi3+ + VO4−; B – Bi4O7(s) + VO4− + MoO42−; C – MoO3(s) + BiO+ + VO4−; D – BiO+ + VO4− + MoO42−; E – MoO3(s) + Bi3+ + VO2+; F – MoO3(s) + BiO+ + VO2+; G – MoO3(s) + Bi3+ + VO2+; H – MoO3(s) + BiO+ +VO2+; I – BiO+ + VO2+ + MoO42−; J – BiO+ + VO2+ + MoO42−; K – MoO2(s) + Bi3+ + VO2+; L – Bi3+ + VO2+ + Mo3+; M – BiVO4(s) + BiO+ +MoO42−; N – BiO+ + HVO42− + MoO42−; O – Bi4O7(s) + HVO42− + MoO42−; P – Bi4O7(s) + VO43− + MoO42−; Q – BiO+ + VO2+ + Mo3+; R – V3Mo(s) + Bi(s) + VO2+; S – Mo(s) + Bi(s) + VO2+; T – Bi(s) + VO2+ + Mo3+; U – Bi(s) + V2O3(s) + Mo3+; V – BiO+ + V2O3(s) + Mo3+; W – MoO2(s) + BiO+ + VO2+; X – MoO2(s) + VO2(s) + BiO+; Y – MoO2(s) + BiVO4(s) + BiO+; Z – MoO2(s) + V3O5(s) + BiO+; A1 – MoO2(s) + V2O3(s) + BiO+; B1 – MoO42− + V3O5(s) + BiO+; C1 – MoO42− + V2O3(s) + BiO+; D1 – Bi2O3(s) + MoO42− + HVO42−; E1 – Bi2O3(s) + MoO42− + VO43−; F1 – V3Mo(s) + Bi(s) + V2O3(s); G1 – Mo(s) + Bi(s) + V2O3(s); H1 – MoO2(s) + Bi(s) + V2O3(s); I1 – MoO2(s) + Bi2O3 (s) + V2O3(s); J1 – Bi2O3 (s) + V2O3(s) + MoO42−; K1 – Bi(s) + V2O3(s) + MoO42−; L1 – Bi(s) + HVO42− + MoO42−; M1 – Bi(s) + VO43− + MoO42−; N1 – Bi(s) + MoO2 + VO43−; O1 – Bi(s) + Mo(s)+ VO43−; P1 – V(s) + V3Mo(s) + Bi(s).

(B) The calculated ϕox and ϕre for the possible degradation processes of BiVO4 with photo-generated electrons and holes.

Simulation reveals two thermodynamically possible scenarios for degradation.

A decomposition-based process: The solids Bi4O7 (B, O, P), Bi2O3 (D1, E1, I1, J1), and soluble ionic species VO43−, HVO42−, VO3− could form during degradation.

An ion-exchange process: Another major region surrounding the stability region of Mo-BiVO4 in the diagram involves the cation BiO+, which would form solids with anions such as Cl−, SO42− (Bassett, 1965).

The Pourbaix diagram is based on thermodynamics only, whereas the actual mechanism will be influenced by kinetics and environmental factors also.

Then, we take illumination into account where photo-generated holes/electrons could participate in the self-oxidation/-reduction of BiVO4 in addition to water splitting. Below we list all plausible oxidation/reduction reactions (R2–R17) to the best of our knowledge:

R2–R10 are oxidation reactions and R11–R17 are reduction reactions. Their propensity to occur could be assessed by their Gibbs free energy change.

Theoretically, thermodynamic oxidation and reduction potentials (ϕ or ϕ) are indicative of the tendency for photo-corrosion of a semiconductor (Chen and Wang, 2012, Hu et al., 2018). Its resistance to photo-corrosion largely depends on the alignment of its ϕ relative to ϕ(O2/H2O) for the photoanode and ϕ relative to ϕ(H+/H2) for the photocathode. Corrosion is thermodynamically favorable when ϕ is smaller than ϕ(O2/H2O) or when ϕ is higher than ϕ(H+/H2) where the values are defined by the listed reactions R1 to R17 and calculated based on changes in Gibbs free energy given by Equations 18 and 19,where, n is the number of photo-generated holes/electrons, G(products) and G(reactants) stand for the Gibbs free energy of products and reactants, respectively, which are available in the literature (Jain et al., 2013, Lide, 2003–2004).

As shown in Figure 4B, the calculated ϕ values for R9 and R10 are lower than ϕ(O2/H2O), and ϕ for R12, R13 are higher than ϕ(H+/H2), which makes these four degradation reactions thermodynamically favorable. Therefore, BiVO4 would be susceptible to both oxidation of holes and reduction of electrons in aqueous solution under illumination. Among them, R10 appears to be the most probable reaction and is also consistent with our experimental observation as V2O5 could get dissolved in the electrolyte, whereas Bi oxide solid could be retained on the photoanode. Since we applied a bias of 1.23 V versus RHE to mimic the condition of electrochemical solar water splitting, the environment around Mo-BiVO4 photoanode should be oxidative. The R10 reaction itself occurs in strongly oxidative conditions, which could emanate from a high voltage bias, accumulation of photo-generated holes at the surface, or abundance of reactive oxygen species (OH− and O−) during the water splitting process. The stability of BiVO4 is determined by the competitive reactions between the self-oxidation/reduction and water splitting. By introducing co-catalysts, the water-splitting potential could be enhanced in comparison with self-oxidation/reduction reactions, thereby improving the long-term stability of the semiconductor.

Discussion

Accumulation of Bi on the photoanode, accompanied by development of self-resistance to further corrosion was proven by ICP-MS, XPS, chronoamperometry, and EIS investigations in this study. Hence, a passivation layer appears to form by degradation of the BiVO4 on the surface of the photoanode under our photolysis conditions. This self-passivation mechanism is supported by thermodynamic computations, which predict two categories of degradation products, namely, bismuth oxides (Bi2O3, Bi4O7) and (BiO)2SO4. Next, we scrutinize our samples for these predicted degradation products.

XPS was used to examine the possibility of (BiO)2SO4 formation. Normally, the BE of S 2p peak in sulfates is in the region 168–172 eV (Wagner and Mullenberg, 1979, Wahlqvist and Shchukarev, 2007). Since no peaks were detected in this BE range in the degraded anode samples, as evident in Figures 2A and S4, the possibility of (BiO)2SO4 formation is not substantiated. This makes bismuth oxides as the possible degradation products.

Subsequent characterizations by XPS, UV-vis, and Fourier-transform infrared spectroscopy (FTIR) provided less definitive results to conclusively identify the degradation products but gave some important evidence that are included in Supplemental Information. To be specific, the Bi 4f peaks in commercial Bi2O3 are shifted to lower BE (158.5 and 163.8 eV from Figure S7) compared with those in BiVO4 (159.1 and 164.4 eV), which concurs with Bi peak shifts observed in the photoanode samples after degradation (Figure 2A). From the UV-vis spectra (Figure S8), after degradation the photoanode showed increased absorbance in the range of 310–460 nm, which coincides with the main absorbance range of Bi2O3. Besides, we detected VO43− by FTIR spectra (Figure S9) in the electrolyte after photolysis, which indicates the formation of bismuth oxides according to those degradation reactions discussed earlier. Combining these experimental results, we conclude that bismuth oxides are the most probable degradation products because their formation is not negated by investigations on the photoanode and the electrolyte. We would suggest Bi2O3 or Bi4O7 to be the major components of degradation products on the photoanode. Bi4O7 is more thermodynamically favorable in the strongly oxidative condition used for PEC photolysis.

In summary, we used Mo-doped BiVO4 as a platform to elucidate the mechanism of degradation of BiVO4 photoanode during photoelectrochemical water splitting in 0.1 M Na2SO4 electrolyte at pH = 5.7. The results gave strong evidence for the preferential dissolution of V (probably as VO43−, HVO42−, VO3−) during photolysis, in comparison with Bi. Besides, strong evidence obtained indicates that Bi remains on the anode surface as a solid remnant product, likely as Bi2O3 or Bi4O7. Complementary thermodynamic modeling and analyses of possible reactions under the prevailing conditions gave this degradation mechanism a sound theoretical backup. BiVO4 undergoes even a purely chemical dissolution in the said electrolyte, but it gets accelerated by illumination and external bias application during photolysis. Enrichment of the surface by the Bi-containing remnant product, a consequence of degradation, appears to passivate the photoanode and protect it from extensive degradation over long periods of photolysis. This unforeseen benefit, however, comes at the expense of charge-transport kinetics, which results in a decrease in photocurrent density. Evidence to conclusively identify the passivation compound was not found, but experimental and theoretical verifications support the possibility of the oxides Bi2O3 and/or Bi4O7 as the major constituents of the passivation layer.

Limitations of the Study

Although this study advances the current understanding of the degradation mechanism of BiVO4 in PEC photolysis, the investigations were in a specific electrolyte only. The degradation of BiVO4 in other common electrolytes remains to be elucidated. The degradation product could not be conclusively identified experimentally, although thermodynamic modeling pointed to either Bi2O3 or Bi4O7. A kinetic model needs to be developed to describe the degradation process accurately once the degradation product is confirmed.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.