An integrated n-Si/BiVO4 photoelectrode with an interfacial bi-layer for unassisted solar water splitting

Integrated n-Si/BiVO4 is one of the most promising candidates for unbiased photoelectrochemical water splitting. However, a direct connection between n-Si and BiVO4 will not attain overall water splitting due to the small band offset as well as the interfacial defects at the n-Si/BiVO4 interface that severely impede carrier separation and transport, limiting the photovoltage generation. This paper describes the design and fabrication of an integrated n-Si/BiVO4 device with enhanced photovoltage extracted from the interfacial bi-layer for unassisted water splitting. An Al2O3/indium tin oxide (ITO) interfacial bi-layer was inserted at the n-Si/BiVO4 interface, which promotes the interfacial carrier transport by enlarging the band offset while healing interfacial defects. When coupled to a separate cathode for hydrogen evolution, spontaneous water splitting could be realized with this n-Si/Al2O3/ITO/BiVO4 tandem anode, with an average solar-to-hydrogen (STH) efficiency of 0.62% for over 1000 hours.

S3 temperature to obtain 1 M precursor solution. Then the precursor solution was spin coated on the Si/Al2O3/ITO, ITO and FTO at a substrate temperature of 60 °C. The spin coating speed was optimized to 1000 rpm for 20 s followed by 3000 rpm for 40 s to obtain approximately BiVO4 thickness. In order to obtain compatible annealing conditions for each layer, the annealing condition was optimized to 400 °C for 0.5 h in the air. The obtained electrodes are soaked in 1 M KOH for 10 min to strip the excess VOx species 3 . A facile photoetching approach was applied to enhance the charge separation by immersing the obtained film in 1 M potassium borate buffer solution (KBi, pH 9.0) containing 0.2 M Na2SO3 (98%) under simulated air mass (AM) 1.5 G illumination for 10 min.

The deposition of NiFe(OH) x catalyst.
For the catalyst, a 20 μL precursor solution containing 30 mM sodium citrate, 5 mM Fe(NO3)3, and 5 mM Ni(NO3)2 was dropped cast onto the BiVO4 film and then dried at 60 °C for 1 h. The obtained electrodes are soaked in 1 M KOH for 5 min to covert the complex into hydroxide, followed by rinsing with deionized water and drying with N2 to remove the excess hydroxide 4 .

Photoelectrochemical measurements.
The PEC measurements of the photoanode were performed in a three-electrode configuration using a Hg/HgO reference, and a Pt foil counter electrode. Ga-In alloy (75.5:24.5 wt%, ≥99.99%) was rubbed on the back of Si to form an ohmic contact. The exposed edges were sealed with an epoxy adhesive. Sulfite oxidation reaction (SOR) was performed in 1.0 M potassium borate buffer solution (KBi, pH 9.0) containing 0.2 M Na2SO3 under simulated AM 1.5 G sunlight illumination.
An electrochemical workstation (CompactStat.e20250, IVIUM) was used to measure the J-V curves and chronoamperometry under the irradiation provided by a xenon lamp (PLS-FX300HU, Beijing Perfectlight), equipped with an AM 1.5 G filter. The light intensity was adjusted to 100 mW cm -2 against a calibrated Si photodiode (Thorlabs, Inc.). The J-V curves were measured at 50 mV s -1 . The active areas of the working electrode were determined by the software Image J.
The solar-to-hydrogen (STH) efficiency of the electrodes above was calculated from the J-V curves with an assumption of 100% Faradaic efficiency (FE), according to the equation STH = I × V × FE / P 6 . Where I (mA cm -2 ) is the photocurrent density, V (1.23 V) is the thermodynamic water splitting potential (based on ∆G 0 ), and P (mW cm -2 ) is the incident illumination intensity (100 mW cm -2 in this work).

S4
The OCP measurement was first performed in the dark. After the potential had stabilized, the light was turned on 7 . The OCP value was the average of more than three different devices.
The sustainable water oxidation performance of the integrated photoanode were conducted without bias. The water oxidation process under illumination tends to oxidize the Ni and Fe species to NiFe(OH)x with high catalytic activity, as well as enrich the surface oxygen vacancies of BiVO4 that promote carrier separation, which may be responsible for the increase of photocurrent. After first 100 h stability, the test was stopped due to a power outage, the integrated photoanode was left in air without light illumination and bias for 2 days. Part of oxygen vacancies were filled by the oxygen from air during the 2 days, leading to the initial decrease in photocurrent.
The second and third light-off duration were short, which shown less effect on the photocurrent.
The H2 production of the PEC tandem cells in Table S1 were calculated from the stability curves with an assumption of 100% Faradaic efficiency and 1 cm 2 electrode area, according to the ideal gas equation PV=nRT. Where P (Pa) is the pressure of the ideal gas, V (m 3 ) is the volume of the ideal gas, n (mole) is the amount of ideal gas measured in terms of moles, R is the gas constant (8.314 J K -1 mol -1 ), T (K) is the temperature. FE=αnF/It, where F is Faraday's constant, α is the number of electrons exchanged, I (mA cm -2 ) is the total current and t (h) the stability.
The Jabs (the rate of photon absorption expressed as a current density, which is calculated assuming the absorbed photon-to-current conversion efficiency (APCE) to be 100%) of BiVO4 photoanode was calculated by measuring the light absorbance (A) and integrating it with respect to the AM 1.5G solar spectrum 8 , where A= 100%-T-R (T and R is the light transmittance and reflectance investigated by UV-vis spectroscopy, respectively).
The charge separation efficiency (Փsep) and charge injection efficiency (Փox) were calculated from the equations: JPEC = Jabs × Փsep × Փox, where JPEC is the measured photocurrent density 8 . For SOR with extremely fast oxidation kinetics, surface recombination is negligible, Փox becomes 1, thus Փsep= JSOR/Jabs. Upon the loading of NiFe(OH)x for OER, the Փox can be further calculated.

Characterization.
The top-down and cross-sectional morphologies of the devices were imaged using a field emission scanning electron microscope (FE-SEM, Hitachi S-4800, 5 kV). The thicknesses of the ITO and