Electrochemical versus Photoelectrochemical Water Oxidation Kinetics on Bismuth Vanadate (Photo)anodes

This study reports a comparison of the kinetics of electrochemical (EC) versus photoelectrochemical (PEC) water oxidation on bismuth vanadate (BiVO4) photoanodes. Plots of current density versus surface hole density, determined from operando optical absorption analyses under EC and PEC conditions, are found to be indistinguishable. We thus conclude that EC water oxidation is driven by the Zener effect tunneling electrons from the valence to conduction band under strong bias, with the kinetics of both EC and PEC water oxidation being determined by the density of accumulated surface valence band holes. We further demonstrate that our combined optical absorption/current density analyses enable an operando quantification of the BiVO4 photovoltage as a function of light intensity.

cooled as above with the exception of the final layer, which was calcined at 500 ℃ for five hours before being cooled to room temperature.

Photochemical (PEC) and electrochemical (EC) characterization
The films were characterized by linear sweep voltammetry and cyclic voltammetry (CV).All EC characterization was carried out in a home-made three-electrode EC cell with a 0.5 cm 2 quartz window.A Ag/AgCl electrode in saturated KCl solution and a platinum mesh were used as the reference and counter electrodes, respectively.All the applied potentials in experiments were with reference to Ag/AgCl (sat.KCl) electrode, and were converted to the reversible hydrogen electrode (VRHE) by Equation S1.
where V is the applied potential in the experiment, and 0.198 V is the standard reduction potential of Ag/AgCl electrode in saturated KCl solution.
BiVO4 films were used as the working electrode.The linear sweep voltammetry was measured in potassium phosphate aqueous buffer (KPi, 0.1 M, pH = 7), to characterize its J-V performance of BiVO4 films.For PEC characterization, the excitation light was provided by a 365 nm LED source and the sample was illuminated from the front (i.e. the side of the BiVO4/electrolyte interface).The light intensity was 12.6 mWcm -2 (equivalent to one sun (AM 1.5)) measured by a power meter.
Based on a reported method, CV was used to confirm that the BiVO4 films were pinhole-free, namely that the underlying FTO was not participating in EC water oxidation. 2CV measurements were carried out in the presence of K4Fe(CN)6 and  Under EC conditions, we did not observe an absorption peak at ~550 nm as reported under PEC conditions. 1 Instead, the optical absorption decreased from 540 nm to 800 nm.The disappearance of the 550 nm peak could be due to extra photoinduced absorption contributed from the probe light.If the wavelength of the probe light is set below 550 nm (e.g.540 nm), its photon energy is approaching the bandgap of BiVO4 (2.4 eV).Once BiVO4 is photo-excited, photo-generated holes will accumulate on the VB of BiVO4, and the optical absorption of photo-generated holes will be superimposed on that of the potential-induced holes.The optical absorption at 460 nm increased with the potential, but there was no water oxidation current observed, even when an optical absorption was up to 0.7 m∆OD.Thus, the optical signal at 460 nm is not related to the water oxidation.The electric field strength was calculated using the equation S3: where  0 is vacuum permittivity,  is relative permittivity of BiVO4 ( = 68) 3         To understand the effect of V0 on rate law analysis, the dark rate law was also measured using V0 at 1.96 VRHE.As shown in Figure S12, although two linear regions seem observed, for a given current density the surface hole density (i.e.optical absorption of holes) at V0 = 1.96VRHE is smaller compared to V0 = 0.61 VRHE.In SP-SEC measurements, the optical absorption of surface holes at 550 nm is equal to the optical absorption when V1 is applied minus the optical absorption when V0 is applied (i.e.reference) (equation S4).
It is possible for potential-induced holes to be generated when V app is anodic of V OCP .
Therefore, when V0 was set to 0.61 VRHE (initial measured VOCP), all the potential-induced holes were measured.However, when V0 was set to 1.9 VRHE, the hole signal at this potential was used as the reference, and the measured hole signal therefore did not include any holes generated between 0.61 VRHE and 1.9 VRHE.This is the likely explanation for the shifted dark Jwo vs. ps curves for the two V0 values.
7. Comparisons of photovoltage calculated using two different methods.scale.For the onset potential method, the (photo)current threshold was defined as 0.02 mAcm -2 .
Photovoltage can be calculated by taking the difference in (photo)current onset potentials in the light and the dark from the J-V curves.Figure S14 compares the light intensity-dependence of photovoltage calculated by this onset method with the photovoltage calculated by the operando OD method described in the main text.
When plotted on a linear LED (ϕ) scale (Figure S14a), the photovoltage begins to saturate at higher light intensities when measured by both methods.When plotted on a logarithmic LED (ϕ) scale (Figure S14b), photovoltage from both methods demonstrates linear dependence on the log of light intensity at higher light intensities.
The data from both methods show a photovoltage with non-ideal behavior, although this is much stronger in the data from the OD method compared to the onset method.
Notably, when measured from the onset method, the photovoltage drops off rapidly at low light intensities whereas the photovoltage from the OD method does not deviate from its linear behavior.A detailed comparison of these two photovoltage methods and their meaning is beyond the scope of this paper but is the subject of ongoing study.
K3Fe(CN)6 redox couple (1 mM , pH = 5.95).As shown in Figure S1 (left panel), there were pronounced redox peaks with bare FTO.The anodic peak and cathodic peak centered at 0.95 and 0.77 VRHE, respectively.During oxidation, electrons transfer from [Fe(CN)6] 2+ ions to FTO generating [Fe(CN)6] 3+ .During reduction, electrons transfer from FTO to [Fe(CN)6] 3+ ions, generating [Fe(CN)6] 2+ .[Fe(CN)6] 3+ /[Fe(CN)6] 2+ was chosen as the redox couple because its reduction potential (estimated with E1/2 =0.86 VRHE) is in the BiVO4 bandgap.This means no redox peaks should be observed if a dense BiVO4 layer was deposited on FTO.As shown in Figure S1 (left panel), no clear anodic and cathodic peaks were observed for the BiVO4/FTO anode (there may be some small current contributions from FTO (e.g. at the edges) in the sample of FTO_BiVO4, but it was negligible compared to the current observed in the FTO sample), indicating electron transfers between FTO and the redox species were indeed prohibited by the BiVO4 top layer.To further consolidate our conclusion, the [Ru(bpy)3] 3+/2+ redox couple (pH = 3.0) with a more positive reduction potential was also used to avoid possible influence from deep trap states below the CB of BiVO4.The reduction potential of [Ru(bpy)3] 3+/2+ can be estimated to be E1/2 = 1.63 VRHE which is more positive than E1/2 = 0.86 VRHE of [Fe(CN)6] 3+/2+ , so the reduction potential of [Ru(bpy)3] 3+/2+ is located deeper into the bandgap of BiVO4 than that of [Fe(CN)6] 3+/2+ .Like the analysis above, from Figure S1 (right panel), we can further conclude that spin-coated BiVO4 can be treated as pinhole-free.

Figure S3 .
Figure S3.The 460 nm optical absorption and current density of BiVO4 plotted as a function of applied potential.

4 .
The dependence of PIA signal and photocurrent density on LED intensity.

Figure S4 .
Figure S4.Steady state PIA signals at 550 nm (blue) and photocurrent density (red) as a function of LED intensity.

Figure S5 .
Figure S5.Optical absorption at 550 nm and the width of the space charge layer plotted as a function of applied potential.

3 Figures
Figures S5 and S6show how the hole signal in the dark correlates with the width of the SCL and the electric field strength respectively.The width of the SCL and the electric field strength both increase with applied potential, following a similar trend to the optical absorption of surface holes.

Figure S7 .
Figure S7.The overlay of BiVO4 current density and optical absorption plots under light and dark conditions, plotted on a linear-linear scale.

Figure S8 .
Figure S8.The overlay of BiVO4 current density and optical absorption plots under light and dark conditions, plotted on a log-linear scale.

Figure S9 .
Figure S9.The rate law plots of water oxidation on different BiVO4 electrodes under PEC and EC conditions.The PIA and SP-SEC measurements were further repeated to show the reproducibility of the overlay kinetics of PEC and EC water oxidation on BiVO4 (the datasets from Dark1 and Light1 were collected from an identical BiVO4 film; the datasets from Dark2 and Light2 were collected from different BiVO4 films).

Figure
Figure S10.A log-log plot of the turnover frequency (TOF) of BiVO4 holes for water oxidation with respect to the surface hole density under PEC and EC conditions.

Figure S11 .
Figure S11.The overlays of PEC and EC water oxidation optical decay transients when the LED was switched off (light, purple) and the system was switched to open circuit (dark, orange).The only pathway to consume the surface holes generated is via water oxidation.The decay starts at decay time = 0s.

Figure
Figure S12.A comparison of rate law using different V0 values under SP-SEC characterization.

Figure S13 .
Figure S13.The repeatability comparison of photovoltage versus LED intensity (ϕ)with data collected from multiple samples, calculated from the operando photovoltage method used in Figure4(a).For Vphoto _1, the data from Figure4(a) were added for comparison.For Vphoto _2, the PIA data (optical absorption versus light intensity) were collected using a new BiVO4 sample (i.e. the same PIA dataset used in Light2 in FigureS9), and the SP-SEC data (optical absorption versus applied potential) were collected from a further BiVO4 sample (i.e. the same SP-SEC dataset used in Dark2 in FigureS9).For Vphoto _Ma, the PIA data were directly adapted from a reported work from Ma et.al,1 and the SP-SEC data were from the dataset used in Dark2.

Figure S14 .
Figure S14.Light intensity-dependence of photovoltage in BiVO4 photoanodes, derived from the onset potentials method (orange, triangle) and the operando OD method (blue, circle).LED intensity (ϕ) plotted on (a) a linear scale and (b) a log10

Figure
Figure S15.A top-view SEM image of BiVO4 films synthesized by the same method used in this paper, showing a dense structure packing by ~100 nm particles (the scale bar is 200 nm).Reproduced from Ref. 4 with permission from the Royal Society of Chemistry.

Table S1 .
A summary of orders of water oxidation in surface holes of different BiVO4 films under PEC and EC conditions.