Unraveling of cocatalysts photodeposited selectively on facets of BiVO4 to boost solar water splitting

Bismuth vanadate (BiVO4) has been widely investigated as a photocatalyst or photoanode for solar water splitting, but its activity is hindered by inefficient cocatalysts and limited understanding of the underlying mechanism. Here we demonstrate significantly enhanced water oxidation on the particulate BiVO4 photocatalyst via in situ facet-selective photodeposition of dual-cocatalysts that exist separately as metallic Ir nanoparticles and nanocomposite of FeOOH and CoOOH (denoted as FeCoOx), as revealed by advanced techniques. The mechanism of water oxidation promoted by the dual-cocatalysts is experimentally and theoretically unraveled, and mainly ascribed to the synergistic effect of the spatially separated dual-cocatalysts (Ir, FeCoOx) on both interface charge separation and surface catalysis. Combined with the H2-evolving photocatalysts, we finally construct a Z-scheme overall water splitting system using [Fe(CN)6]3−/4− as the redox mediator, whose apparent quantum efficiency at 420 nm and solar-to-hydrogen conversion efficiency are optimized to be 12.3% and 0.6%, respectively. Artificial photosynthesis offers an integrated means to convert light to fuel, but efficiencies are often low. Here, authors report a Z-scheme system utilizing Ir and FeCoOx co-catalysts to enhance charge separation on BiVO4 facets that achieves high quantum efficiencies for overall water splitting.

P articulate photocatalytic overall water splitting (OWS) based on inorganic semiconductor materials with relative good photostability has been demonstrated as one of the most promising ways of realizing scalable and economically viable solar hydrogen production to address energy-and environmentrelated issues [1][2][3][4][5][6][7][8][9][10][11][12] . To achieve high solar-to-hydrogen (STH) energy conversion efficiency, it is necessary to increase the quantum efficiency of photocatalytic OWS over a wide range of wavelengths, particularly the use of visible light 11,12 . However, extended visible light utilization is generally accompanied by a decreased driving force of the photogenerated carriers to make charge separation extremely difficult. Furthermore, the construction of OWS systems faces serious challenges originating from the sluggish kinetics of water oxidation involving uphill energy barrier and multiple electron transfer 13 . Consequently, visible-light-driven photocatalytic OWS systems are not only limited in number, but also show lower efficiency than those driven by ultraviolet light [14][15][16][17][18][19] . Accordingly, it is highly desirable to precisely design and modify photocatalysts with efficient visible light utilization for promotion of water oxidation.
N-type monoclinic bismuth vanadate (BiVO 4 ) has emerged as one of the most promising visible-light-responsive photocatalysts and photoanodes for water oxidation since Kudo's report in 1998 [20][21][22][23][24][25][26][27][28] . Owing to its advantages, such as efficient light absorption in the visible light region, good carrier mobility, controllable exposed facets, and non-toxic properties, BiVO 4 semiconductor has been widely and successfully employed as the water oxidation photocatalyst for the assembly of Z-scheme OWS systems using solid conductor (i.e., Au, reduced graphene oxide) or redox couple (i.e., Fe 3+/2+ , [Fe(CN) 6 ] 3−/4− ) as electron mediator 15,16,[29][30][31] . Specifically, our previous work revealed that the spatial separation of photogenerated electrons and holes can be achieved on the anisotropic facets of BiVO 4 32 , based on which reduction and oxidation cocatalysts are selectively deposited on different facets to remarkably promote its water oxidation and the efficiency of OWS under visible light 16 . Although the BiVO 4 photocatalyst has been widely investigated for the assembly of artificial Z-scheme OWS systems, the apparent quantum efficiency (AQE) and STH conversion efficiency achieved so far are still considerably below what is expected. This is mainly due to the shortage of effective cocatalyst regulation and the lack of in-depth understanding of the microscopic mechanisms behind it 33,34 . Notably, for the assembly of the redox couple-mediated Z-scheme OWS system shown in Fig. 1, the loading of effective cocatalysts is extremely important not only for the acceleration of interfacial electron transfer between the H 2 -evolving photocatalyst (HEP) and the O 2 -evolving photocatalyst (OEP), but also for the promotion of surface reaction kinetics of water splitting [35][36][37][38] . Therefore, it is a long-term task to develop innovative cocatalysts and unravel their structures as well as influence mechanism on water splitting.
In this study, we address the sluggish water oxidation of BiVO 4 via in situ photodeposition of dual innovative cocatalysts, with emphasis on elucidating the local structures of the cocatalysts and the mechanism of promotion of water oxidation. We demonstrate that the nanocomposite of FeOOH and CoOOH (denoted as FeCoO x ) in situ formed on the {110} facet of BiVO 4 not only lowers Gibbs free energy barrier of water oxidation, but also makes a better promotion on the electron transfer as well as charge separation compared with the commonly used CoO x cocatalyst. Furthermore, the Ir cocatalyst in situ deposited on the {010} facet of BiVO 4 was found to exhibit superior reduction ability of [Fe(CN) 6 ] 3− ions to our previously reported Au. Based on the facet-selective loading of the innovative dual-cocatalysts, the evolution rate of O 2 on the BiVO 4 was significantly enhanced, and a particulate Z-scheme OWS system with an AQE of 12.3% at 420 nm and a STH of 0.6%, was finally fabricated using [Fe(CN) 6 ] 3−/4− as a redox mediator and ZrO 2 /TaON or MgTa 2 O 6−x N y /TaON as the HEP. Our results demonstrate the importance and effectiveness of developing suitable cocatalysts for enhancing interfacial charge separation and surface water oxidation kinetics in promoting solar energy conversion.

Results
Structural characterizations of cocatalysts photodeposited. The anisotropic BiVO 4 with exposed {010} and {110} facets was prepared according to our previous study 39 . The Ir nanoparticles and FeCoO x nanocomposite were in situ photodeposited on the surface of BiVO 4 from an aqueous solution containing the precursors K 2 IrCl 6 , CoSO 4 , and redox [Fe(CN) 6 ] 3− ions. The asobtained sample is hereafter denoted as Ir-FeCoO x /BiVO 4 . As expected from our previous findings on the spatial separation of photogenerated electrons and holes on the anisotropic BiVO 4 32 , the Ir nanoparticles, and FeCoO x nanocomposite were clearly observed to be selectively deposited on the {010} and {110} facets of BiVO 4 , respectively (Fig. 2a, b). For comparison, the in situ photodeposition of single Ir or CoO x particles on BiVO 4 was similarly obtained (denoted as Ir/BiVO 4 and CoO x /BiVO 4 ), and the sample was characterized by field-emission scanning electron microscopy (FESEM) to further confirm the facet-selective deposition ( Supplementary Fig. 1 Fig. 1b), demonstrating the possible interaction between Fe and Co-based compounds. And the change in the long wavelength range of UV-Vis diffuse reflectance spectra (DRS) can confirm the successful deposition of the dual-cocatalysts ( Supplementary Fig. 2). The deposited Ir species were verified to exist as metallic Ir nanoparticles by means of X-ray absorption near edge structure (XANES) spectroscopy ( Supplementary Fig. 3) and high-resolution transmission electron microscopy image (Supplementary Fig. 4).
To unravel the formation of the FeCoO x nanocomposite on the {110} facets of BiVO 4 , the existing state and dispersion of both Fe and Co elements on FeCoO x /BiVO 4 (free of Ir nanoparticles to rule out its possible interference during characterization) were first analyzed. According to the elemental mapping results shown in Supplementary Fig. 5, both Fe and Co species are similarly located and dispersed, accompanied by the existence of O, which further demonstrates that Co and Fe combine together in the form of oxidation state during the photo-oxidation process. The deposition of Fe should result from redox ions in the reaction solution. The coexistence of both Fe and Co can be further revealed by electron energy loss spectroscopy analysis ( Supplementary Fig. 6). And their oxidation states can be confirmed to be Fe 3+ and Co 3+ by the Fe and Co K-edge XANES measurements through comparing with the reference materials ( Fig. 2c, d, respectively).
Second, the radial distance space spectra χ(R) of Fe and Co in FeCoO x /BiVO 4 and their corresponding references were analyzed, which provides more convincing support for the formation of nanocomposite. As shown in Fig. 2e and Supplementary Fig. 7 Supplementary Fig. 8a, b) and the first derivative of the normalized XANES μ(E) spectra ( Supplementary Fig. 8c, d). In particular, as shown in Supplementary Fig. 8c Third, the formation of the nanocomposite can be further verified by the results of quantitative χ(R) space spectra fitting and wavelet transform of χ(k). As seen in Supplementary Table 1 Table 1) provide a quantitative illustration of the existence of Fe-O-Co bond originating from the nanocomposite structure. As similarly revealed in Fig. 2h Fig. 10).
Effect of reduction and oxidation cocatalysts. As shown in Fig. 1, the water oxidation process of OEP is strongly dependent on both the reduction and oxidation cocatalysts. Therefore, understanding the effect of deposited Ir and FeCoO x cocatalysts is highly desirable. As depicted in Fig. 3a, the ability of the deposited metallic Ir to reduce [Fe(CN) 6 ] 3− ions was evaluated and found to exhibit a much higher cathode current than that of our previously reported Au nanoparticles on BiVO 4 16 , indicating its superior performance in activating and reducing the [Fe(CN) 6 ] 3− ions. In addition, the deposition of Ir or Au cocatalyst on the surface of BiVO 4 can significantly decrease the charge-transfer resistance (R ct ) across the semiconductor/electrolyte interface (Fig. 3b), further revealing the effectiveness of the deposited cocatalysts in accelerating the electron transfer from BiVO 4 to the [Fe(CN) 6 ] 3− ions (values of R s and R ct listed in Supplementary  Table 2). Meanwhile, the promotion effect of Ir is better than that of Au.
To determine the effect of the FeCoO x nanocomposite, the efficiencies of charge separation and injection (denoted as η sep and η inj , respectively) on the FeCoO x /BiVO 4 photoanode (CoO x / BiVO 4 and BiVO 4 as references) were evaluated by referring to a previous photoelectrochemical analysis 22 . Figure 3c and Supplementary Fig. 11a show that the current of the BiVO 4 photoanode can be remarkably promoted after the deposition of FeCoO x and CoO x in both cases, with and without the use of a hole scavenger, among which FeCoO x exhibits a much better promotion effect than CoO x . On this basis, both η sep and η inj on the FeCoO x /BiVO 4 photoanode were calculated to be higher than that of the CoO x /BiVO 4 photoanode ( Supplementary Fig. 11c, d nanocomposite on both the separation of photogenerated carriers and the injection of holes into the reaction solution (i.e., surface reaction) with respect to CoO x . The excellent promotion of FeCoO x on the surface reaction can be further supported by the electrochemical impedance spectroscopy (EIS) results given in Fig. 3d and Supplementary Table 3, based on which the R ct resistance on the FeCoO x /BiVO 4 electrode is the smallest among the three electrodes investigated. On the other hand, the superior promotion effect of FeCoO x on the charge separation can also be evidenced by its larger open-circuit potential (OCP) on the FeCoO x /BiVO 4 compared with CoO x /BiVO 4 (Fig. 3e). Based on the previous result that a larger difference of OCPs under dark and illumination conditions corresponds to more intense band bending 40 , therefore the FeCoO x /BiVO 4 sample can be deduced to own a more intense band bending than the CoO x /BiVO 4 sample, leading to a significantly improved η sep and the more intense band bending should result from the p-n heterojunction between FeCoO x and BiVO 4 41 .
Encouraged by the understanding of the functionalities of both reduction and oxidation cocatalysts (i.e., Ir and FeCoO x ), the synergistic effect of dual-cocatalysts on the charge separation was examined using the surface photovoltage (SPV) spectrum. As shown in Fig. 3f, the sample with both Ir and FeCoO x deposited exhibits a greater SPV amplitude with respect to the sample with single Ir or FeCoO x loaded. It should be mentioned that a much better promotion effect is also observed for the sample with facetselective deposition of Ir and FeCoO x compared to that with facet-selective deposition of Au and CoO x . These results reveal the importance of both facet-selective deposition of dualcocatalysts and the development of innovative cocatalysts for maximizing the promotion effect.
Density functional theory calculations on the O 2 -evolving reaction. Density functional theory (DFT) calculations were performed to further elucidate the microscopic mechanism of the promotion effect of the FeCoO x cocatalyst on the O 2 -evolving reaction (OER) from the viewpoint of both surface catalysis and interfacial charge transfer. As shown in Fig. 4a-c, the CoO x -FeO x -CoO x -FeO x and CoO x -CoO x -CoO x -CoO x clusters were simply extracted and placed on the {110} facets of BiVO 4 to simulate the FeCoO x /BiVO 4 and CoO x /BiVO 4 interfaces, respectively, which origin from the structure of EXAFS measurement. And the schematic of the whole OER mechanism on the FeCoO x /BiVO 4 and CoO x /BiVO 4 is given in Supplementary  Fig. 12 and illustrated in detail in supporting information. Fig. 4d and e presents the Gibbs free energy change diagram of the four elementary steps of OER on the surface of FeCoO x /BiVO 4 and CoO x /BiVO 4 , during which the Co and Fe sites on FeCoO x / BiVO 4 were mainly considered as the active sites, respectively. It was demonstrated that the rate-determining step of FeCoO x / BiVO 4 (Co or Fe site) and CoO x /BiVO 4 is the adsorption of one OH − to form OOH * from O * . The largest decrease of the Gibbs free energy barrier was observed for FeCoO x /BiVO 4 (Co site), whereas the OER performance of FeCoO x /BiVO 4 (Fe site) is much weaker than that of the corresponding Co site, suggesting that the Co site acts as the main OER site.
Next, we plotted the calculated densities of states (DOS) of the BiVO 4 {110} surface, CoO x /BiVO 4 {110} interface, and FeCoO x / BiVO 4 {110} interface (Fig. 4f, g, h and Supplementary Fig. 13). Both CoO x and FeCoO x are set to be located at the {110} facets of BiVO 4 . For the bare BiVO 4 {110} surface, there is a direct bandgap about 2.1 eV between the valence band and conduction band (Fig. 4f). However, when the CoO x -FeO x -CoO x -FeO x cluster is settled on the {110} surface of BiVO 4 , a mixed band mainly composed of Co 3d, Fe 3d, and O 2p states emerges between the valence band and conduction band (Fig. 4g). It has been demonstrated that the localization of photoexcited holes, as well as subsequent charge separation can be promoted through the formation of mixed bands 42 . In addition, the DOS of the CoO x /BiVO 4 {110} interface (Fig. 4h) has no similar result as that of the FeCoO x /BiVO 4 {110} interface (Fig. 4g, bandgap = 2.0 eV), implying that the loading of FeCoO x on BiVO 4 should have better charge separation. In order to microscopically understand the better electron transfer on the FeCoO x with respect to the CoO x , their bader charges were calculated and compared. As given in Supplementary Table 4, the changing trend of bader charge on the Co active site after introduction of Fe (increase from 1.2 a.u. in CoO x /BiVO 4 to 1.3 a.u. in FeCoO x /BiVO 4 ) is in line with the changing one of experimental valence state ( Supplementary  Fig. 14). Compared to the CoO x /BiVO 4 , the higher bader charge on the Co active site in FeCoO x /BiVO 4 indicates its stronger oxidation capacity as well as more beneficial electron transfer 43 . In addition, as shown in Supplementary Fig. 15, the d-band center (Ed) value of Co active sites in FeCoO x /BiVO 4 was calculated as −1.63 eV, which is sharply increased with respect to the CoO x /BiVO 4 (−2.56 eV). This demonstrates that the electronic structure of Co active sites can be well modulated and optimized in the FeCoO x /BiVO 4 due to the introduction of Fe atoms to get much stronger adsorption properties to the OER intermediates according to the d-band center theory 44,45 . According to previous experimental and theoretical demonstration, the Fe site is relatively inactive during the OER process 46,47 . So we deduce that the role of Fe is to assist in modifying the geometric and electronic structure of Co in the OER together with our results that very limited contributions of Fe 3d states are observed for the mixed band (Fig. 4g). These conclusions from DFT calculation well match with the aforementioned experimental results.
Photocatalytic performances of Z-scheme OWS. The modified BiVO 4 was employed as an OEP for the assembly of efficient Z-scheme OWS systems together with ZrO 2 /TaON or MgTa 2 O 6−x N y /TaON as a HEP under visible light irradiation. The HEPs were prepared and modified with cocatalysts according to previously reported procedures 16,48,49 , and the diffraction structure and morphology features were coarsely revealed by their powder X-ray diffraction patterns and FESEM images (Supplementary Figs. 16,17). The contents of deposited Ir and Co on BiVO 4 were optimized to be 0.8 and 0.2 wt%, respectively, via the photocatalytic O 2 evolution reaction (Supplementary Figs. 18,19). As seen in Fig. 5a, stable evolution curves of H 2 and O 2 with the stoichiometric molar ratio of 2:1 can be observed at the experimental region using the optimized photocatalysts, indicating the successful achievement of the OWS process. Moreover, regardless of using ZrO 2 /TaON or MgTa 2 O 6−x N y /TaON as the HEP, similar OWS activities with the initial rates of H 2 and O 2 evolution (ca. 160 and 80 μmol/h, respectively) were separately observed, implying that the O 2 evolution on BiVO 4 is the ratedetermining step, as similarly observed in our previous study 16 . It should be pointed out that when the Ir-CoO x (Imp.)/BiVO 4 with Ir and Co randomly impregnated is employed as the OEP, the OWS will not be achieved owing to the significantly decreased O 2 -evolving activity (Supplementary Fig. 20). This indicates the importance of facet-selective deposition of dual-cocatalysts in promoting the O 2 -evolving activity and fabricating a successful OWS system. The multiple cycles of time-course curves shown in Supplementary Figs. 21 and 22 demonstrate the good photostability of the system constructed in this study. The AQE value of OWS as a function of absorption wavelength was found to be in good accordance with the UV-Vis DRS of the OEP and HEP, indicating that the Z-scheme OWS system is driven by visible light excitation (Fig. 5b and Supplementary Fig. 23). The optimal AQE value of OWS at 420 ± 10 nm is 12.3%, and the AQE value at the 500 ± 10 nm is about 3%, demonstrating the wide visible light utilization. According to the activity measurements under the irradiation of AM 1.5 G (Fig. 5c), the STH energy conversion efficiency was calculated to be 0.6%. To the best of our knowledge, both the AQE and STH values should be the highest among the suspending particulate photocatalytic OWS systems using inorganic semiconductor materials with visible light utilization, regardless of one-step or two-step (i.e., Z-scheme) systems.

Discussion
Here, we show a highly efficient Z-scheme OWS system with benchmarked AQE and STH value over particulate inorganic semiconductor photocatalysts with visible light utilization. The success is mainly ascribed to the in situ facet-selective photodeposition of innovative dual-cocatalysts (Ir nanoparticles and FeCoO x nanocomposite), based on which the sluggish water oxidation on BiVO 4 can be largely overcome. Besides the finding and structural unraveling of efficient cocatalysts, the microscopic work mechanism of both reduction and oxidation cocatalysts on the interfacial charge transfer and surface catalysis has been well elucidated respectively. These results should be encouraging and enlightening to the design and assembly of OWS systems for more efficient solar-to-chemical energy conversion.  48,49 . BiVO 4 was chosen as the OEP, which was similarly synthesized according to our previous hydrothermal process 39 . Typically, 10 mmol NH 4 VO 3 and 10 mmol Bi(NO 3 ) 3 ·5H 2 O were dissolved in 2.0 M nitric acid solution, whose pH value was then adjusted to be about 0.5 with  Measurements of AQE and STH conversion efficiency. The AQE was measured using a Pyrex top-irradiation-type reaction vessel and a 300 W xenon lamp fitted with band-pass filters (ZBPA420, Asahi Spectra Co., FWHM: 10 nm). The number of photons reaching the solution was measured using a calibrated Si photodiode (LS-100, EKO Instruments Co., LTD.), and the AQE (ϕ) was calculated using the following Eq. (1):  6 ] aqueous solution and phosphate buffer solution (pH = 6, 0.1 M) were used as the electrolyte. The potential of the working electrode was controlled by a potentiostat (CHI 660E) for the LSV test and potentiostat (Solartron analytic AMETEK) for the EIS test. Before the measurement, the solution was purged with argon gas. The Nyquist plots calculated from EIS were performed from 100,000 to 0.1 Hz. Data were fitted using Zview software. Current-voltage (J-V) curves under irradiation and darkness were recorded on an electrochemical workstation (CHI 660E). The OCP of photoanode were recorded under illumination and darkness using electrochemical workstation (Solartron analytic AMETEK). A 300 W xenon lamp was used as the light source and the irradiation intensity was high enough to produce a flat band condition of the photoanodes. The electrolyte for J-V curves and OCP was 1 M KBi (pH = 9). 0.2 M Na 2 SO 3 was added to the electrolyte as a hole scavenger if necessary.

Data availability
The data that support the findings of this study are available from the source data. Source data are provided with this paper.