Photocatalytic generation of hydrogen by core-shell WO3/BiVO4 nanorods with ultimate water splitting efficiency

Efficient photocatalytic water splitting requires effective generation, separation and transfer of photo-induced charge carriers that can hardly be achieved simultaneously in a single material. Here we show that the effectiveness of each process can be separately maximized in a nanostructured heterojunction with extremely thin absorber layer. We demonstrate this concept on WO3/BiVO4+CoPi core-shell nanostructured photoanode that achieves near theoretical water splitting efficiency. BiVO4 is characterized by a high recombination rate of photogenerated carriers that have much shorter diffusion length than the thickness required for sufficient light absorption. This issue can be resolved by the combination of BiVO4 with more conductive WO3 nanorods in a form of core-shell heterojunction, where the BiVO4 absorber layer is thinner than the carrier diffusion length while it’s optical thickness is reestablished by light trapping in high aspect ratio nanostructures. Our photoanode demonstrates ultimate water splitting photocurrent of 6.72 mA cm−2 under 1 sun illumination at 1.23 VRHE that corresponds to ~90% of the theoretically possible value for BiVO4. We also demonstrate a self-biased operation of the photoanode in tandem with a double-junction GaAs/InGaAsP photovoltaic cell with stable water splitting photocurrent of 6.56 mA cm−2 that corresponds to the solar to hydrogen generation efficiency of 8.1%.

Fabrication of WO 3 -NRs/BiVO 4 +CoPi heterojunction photoanodes. At first, we deposited a compact ITO film with the thickness of about 150 nm on fused silica substrates (5x5cm) at normal incidence angle (α=0º) from the ITO target (99.999%, Furuchi Chem. Co.) in the Ar:O 2 (15:0.3 SCCM) mixture and working pressure of 0.6 Pa. Then, the stage was turned toward the Pt target (99.999%, Furuchi Chem. Co.) to deposit a thin Pt film (~50 nm) in Ar atmosphere at the working pressure of 2 Pa. In the next step, the Pt film was encapsulated by the deposition of the second ITO layer (150 nm), as described above. The ITO/Pt/ITO stack has a low sheet resistance of ~3-4 Ω/☐ due to the encapsulated Pt layer, which simultaneously acts as a back reflector. Since the Pt layer was encapsulated and had no contact with the electrolyte, it did not participate in the electrochemical reaction and thus in future could be substituted by a less expensive metal, such as Ag or Al. After the deposition of the ITO/Pt/ITO stack we set the sample over the third magnetron with W target (99.99%, Advantec Co.) and changed the stage position to the GLAD regime with α=85º to the substrate normal. The WO 3 -NRs were deposited in the GLAD regime by reactive sputtering in the O 2 :Ar (9.6 : 11 SCCM) mixture and low working pressure of 0.3 Pa with the constant speed of substrate rotation of 45 rpm. The fabrication of WO 3 -NRs was finalized by annealing in air at 575 °C for 4.5 h.
The precursor solution for the electrodepositon of BiVO 4 was prepared by dissolving 10 mM of Bi(NO 3 ) 3 in a solution of 35 mM VOSO 4 adjusted to pH = 0.5 with HNO 3 . The Bi(III) is soluble at pH < 2, however no film can be formed in such acidic solution. Therefore, at first the pH of the electrolyte was raised to 5.1 by 2 M sodium acetate solution and then stabilized at pH = 4.7 by adding a few drops of concentrated HNO 3 , since V(IV) starts to precipitate at pH > 5.
The electrodeposition of BiVO 4 was conducted at potentiostatic conditions in the two electrode configuration with the bias of 0.21 V applied between ITO/Pt/ITO/WO 3 -NRs as a working electrode and a Pt mesh as a counter electrode. The deposition of amorphous BiVO 4 was carried out at 55 °C by varying the deposition time from 35 to 270 s. All freshly prepared samples were rinsed with distilled water, dried in the N 2 steam and then annealed in air at 500 °C for 2 hours to convert the amorphous layer into a crystalline monoclinic The sample was biased vs a counter Pt mesh electrode at galvanostatic conditions to keep the photocurrent at ~10 μA cm -2 . The optimized deposition time was found to be 500 s. The resulting photoanodes were rinsed with distilled water and dried under a gentle N 2 flow. The photoanodes based on flat film WO 3 /BiVO 4 heterojunction were prepared by using the same fabrication procedure, but without the GLAD regime. All chemicals were purchased from Wako.
The photoelectrochemical (PEC) characterizations of the photoanodes were conducted according to the standard PEC characterization protocol in a potassium phosphate buffer solution (pH=7) by using ALS/CHI (608D) potentiostat and a standard three-electrode method with a Pt counter electrode and an Ag/AgCl reference electrode. Two electrode measurements were conducted by following the same protocol. The I-V characteristics and the photocurrent-time (J p -t) profiles were recorded under simulated solar light provided by a solar simulator (PEC-L01, Peccel Co.). The light intensity was adjusted by using an NREL calibrated photodetector.
The V RHE potential was calculated by using the Nernst equation: is the converted potential vs RHE, / is the experimental potential measured against the Ag/AgCl reference electrode, and / 0 is the standard potential of Ag/AgCl at 25° C (0.198 V).
The incident photon to current conversion efficiency (IPCE) was measured in the two-electrode configuration at the constant bias of 1V vs Pt electrode from 300 to 650 nm by using a tunable light source provided by a stabilized 500 W Xenon lamp combined with a computer-controlled double grating monochromator. The whole system was purchased from JASCO Co.
The oxygen and hydrogen evolution were directly measured in an airtight 2-electrodes PEC cell connected to a gas micro-chromatograph (Inficon 3000, EZ IQ). Prior to measurements, the PEC cell was evacuated and filled with Ar to atmospheric pressure repeatedly to eliminate air in the cell. The photoelectrode was biased at 1V vs the Pt counter electrode in pH 7 potassium phosphate buffer solution and illuminated by a simulated AM1.5 solar light. The gas probes were taken every 10 minutes by the gas micro-chromatograph.