Black phosphorene as a hole extraction layer boosting solar water splitting of oxygen evolution catalysts

As the development of oxygen evolution co-catalysts (OECs) is being actively undertaken, the tailored integration of those OECs with photoanodes is expected to be a plausible avenue for achieving highly efficient solar-assisted water splitting. Here, we demonstrate that a black phosphorene (BP) layer, inserted between the OEC and BiVO4 can improve the photoelectrochemical performance of pre-optimized OEC/BiVO4 (OEC: NiOOH, MnOx, and CoOOH) systems by 1.2∼1.6-fold, while the OEC overlayer, in turn, can suppress BP self-oxidation to achieve a high durability. A photocurrent density of 4.48 mA·cm−2 at 1.23 V vs reversible hydrogen electrode (RHE) is achieved by the NiOOH/BP/BiVO4 photoanode. It is found that the intrinsic p-type BP can boost hole extraction from BiVO4 and prolong holes trapping lifetime on BiVO4 surface. This work sheds light on the design of BP-based devices for application in solar to fuel conversion, and also suggests a promising nexus between semiconductor and electrocatalyst.

P hotoelectrochemical (PEC) water splitting on polycrystalline BiVO 4 photoanodes has attracted considerable attention in recent years due to the narrow bandgap (2.4-2.5 eV) and deep valence band edge of BiVO 4 , which enable visible light harvesting and water oxidation 1,2 . However, the occurrence of surface/ bulk charge recombination due to the poor charge transport characteristics and short hole-diffusion length (<70 nm) of BiVO 4 leaves room to improve the PEC performance of BiVO 4 photoanodes 3,4 . Heteroatom doping [5][6][7] , component or structural tuning [8][9][10] , and loading of oxygen evolution co-catalysts (OECs) [11][12][13][14] are identified as the most promising approaches for overcoming these drawbacks and improving the PEC performance of BiVO 4 photoanodes. Among these methods, OEC loading can strongly suppress surface recombination in BiVO 4 photoanodes and also shift the photocurrent onset potential close to its flat-band potential for water oxidation, which is the most significant feature for achieving unbiased solar water splitting 15,16 .
Recently, van de Krol et al. re-stated the roles of some OECs, such as Co-Pi and RuO x , from the perspectives of the surface reaction kinetics and surface recombination. The researchers mainly pointed out that the water oxidation capability of OECs is strongly limited by their small thermodynamic driving force caused by insufficient hole extraction from the photoanodes 17 . This limitation means that for one of the best existing photoanode materials with an OEC, BiVO 4 /OEC, the band bending of BiVO 4 at the electrode/electrolyte interface must be optimized [18][19][20] . For instance, Kim and Choi demonstrated that the incorporation of a FeOOH compound can accelerate hole transport from BiVO 4 to the NiOOH OEC because the hole transport resistance of FeOOH is lower than that of NiOOH 13 . Zhong et al. suggested that the deposition of p-NiO on the CoO x OEC/BiVO 4 surface to form a p-n junction interface can be beneficial for rapid hole extraction to reduce bulk charge recombination in BiVO 4 21 . Gong's group directly employed a p-Co 3 O 4 OEC instead of CoO x and proved that a p-type semiconductor having OEC functions can result in simultaneous enhancements in the hole extraction and water oxidation capabilities of the BiVO 4 photoanode 22 . Therefore, promoting hole extraction from BiVO 4 to OECs by improving their interface resistance still holds broad interest and significance for enhancing the PEC performance.
As a novel 2D family of materials, exfoliated black phosphorene (BP) layers that are 2-20 nm thick can show p-type semiconductor properties with high hole mobility (1000 cm 2 V −3 s), which are caused by the unavoidable presence of oxygen species 23,24 . On the other hand, its bandgap properties, which are dependent on the number of layers, result in a tunable bandgap between the bulk value of 0.3 eV to the monolayer value of 2.1 eV; therefore, BP is considered a photoabsorber of visible and nearinfrared solar light for solar light harvesting 25 . Apart from several compelling succeeds in the application of exfoliated BP as a photocatalyst for H 2 generation and water splitting [26][27][28][29][30] , employing exfoliated BP and its tailored integration with photoanodes to enhance the PEC performance for highly efficient water splitting has not been given much attention 31 .
In this study, we first demonstrate that the insertion of exfoliated BP nanosheets with ∼4 layers between BiVO 4 photoanodes and conventional OEC layers can lead to ultra-rapid hole extraction. The electrochemical analysis reveals a built-in p/n electric field formed by the BP/BiVO 4 heterostructure, in which the space-charge region results in an upward shift in the energy level of the BP nanosheets. After coating of the photoanode with an additional thin OEC layer (NiOOH), the interfacial band-edge energetics strongly drive holes from BiVO 4 to the NiOOH surface for efficient water oxidation. As a result, NiOOH/BP/BiVO 4 achieves a photocurrent density of 4.48 mA·cm −2 at a bias of 1.23 V vs. RHE, which is 4.2 times higher than that of pure BiVO 4 and 1.5 times higher than that of NiOOH/BiVO 4 . Moreover, the hole extraction role of the BP nanosheets is successfully evidenced by two other OECs (MnO x and CoOOH), demonstrating the potential of BP as an auxiliary to enhance water oxidation.

Results
Characterization of the BP/BiVO 4 photoanode. BP nanosheets were synthesized by liquid exfoliation of bulk BP particles and dispersed in isopropanol (IPA) under an N 2 atmosphere (Supplementary Fig. 1). The redshifted Raman signals of the BP nanosheets confirm the successful exfoliation of bulk BP (Supplementary Fig. 2). The atomic force microscopy (AFM) image of the exfoliated BP nanosheet layers shows a distinct 2D morphology with an average thickness of ∼2.2 nm, corresponding to 4 layers ( Supplementary Fig. 3) 32 . High-resolution transmission electron microscopy (HR-TEM) images of the exfoliated BP nanosheets display clear lattice fringes with a d-spacing of 0.34 nm, corresponding to the (040) plane ( Supplementary Fig. 4). A nanoporous BiVO 4 photoanode was fabricated by using an electro-deposited BiOI film as a precursor based on the previous method 13 . The thickness of the as-prepared BiVO 4 photoanode was ca. 1 µm (Fig. 1a). Considering that the lateral size of BP is larger than the pore size of BiVO 4 film, the depositing BP on BiVO 4 photoanode was assisted by centrifuge-coated method (See experimental section for detail). Compared to the morphology of the pure BiVO 4 photoanode ( Fig. 1b and Supplementary Fig. 5a), the SEM image of BP/BiVO 4 does not reveal the presence of BP nanosheets on surface of BiVO 4 photoanode ( Fig. 1c and Supplementary Fig. 5b), which is in stark contrast to that the BiVO 4 photoanode is immersed into the BP dispersion by natural adsorption or deposition ( Supplementary Fig. 6). X-ray diffraction (XRD) analysis demonstrates monoclinic BiVO 4 crystal, which remains unchanged after the deposition of BP, but a small diffraction peak of BP can be detected ( Supplementary Fig. 7). However, although the observation on BP/BiVO 4 by TEM image could not distinguish the presence of BP nanosheets (Fig. 1d), the electron diffraction spot confirms the co-existence of polycrystalline BiVO4 and BP components (insert in Fig. 1d). Highangle annular dark-field scanning TEM-energy-dispersive spectroscopy (HAADF-STEM-EDX) reveals an obvious sheet-like distribution pattern of P element clinging to BiVO 4 particles (Fig. 1e). Further enlarged HAADF-STEM image exhibits a bright area on the BiVO 4 particle, which can be identified as a BP sheet ( Supplementary Fig. 8). However, the structure incompatibility between the 2D BP sheets and 3D BiVO 4 nanopores may lead to uncovered part of BiVO 4 by BP. Nevertheless, its high-resolution TEM (HR-TEM) image shows the distinct interface of BP/BiVO 4 , in which the lattice spacing of 0.212 nm corresponds to the (051) planes of monoclinic BiVO 4 14 , while the other lattice spacing of 0.54 nm is consistent with the interlayer distance of BP along the c-axis 33 . Compared to the BP/BiVO 4 prepared by the centrifugecoated method, the naturally deposited BP/BiVO 4 shows the poor connection between the BP sheet and BiVO 4 particles (Supplementary Fig. 9). The conductivity of the BiVO 4 photoanode and BP nanosheets was investigated with Mott-Schottky plots. As shown in Fig. 1g, BiVO 4 is a typical n-type semiconductor with a Fermi energy of 0.324 V vs NHE; the n-type behaviour is usually caused by the presence of oxygen defects 1 . In contrast, the straight line of exfoliated BP displays a negative slope with a Fermi level of 0.588 V vs NHE, indicating p-type conductivity. The intimate contact between BiVO 4 and the BP nanosheets makes it easy to construct an electric field from the p/n junctions. High-resolution X-ray photoelectron spectroscopy (XPS) of Bi, O, and V can determine the built-in potential and band offsets at the interface 34 . As shown in the Bi 4f (Fig. 2a) and O 1s/V 2p (Fig. 2b) XPS spectra, the Bi 4f 7/2 core-level and O 2p core-level spectra of BP/BiVO 4 shift by 0.3 and 1.58 eV, respectively, to higher binding energies in comparison to those of pure BiVO 4 , whereas the V 2p 3/2 core-level spectra of BP/BiVO 4 and BiVO 4 are almost the same. Accordingly, the binding energy of the P 2p core-level spectrum of BP shifts by ∼0.41 eV to lower energy after integration with BiVO 4 (Fig. 2c). The differential charge density diagram of the BP/BiVO 4 heterointerface was compared with that of the clean BiVO 4 surface by Bader charge analysis, which can further reveal the charge transfer direction (Fig. 2d). In detail, BP with a 1 × 3 supercell donates 0.12 e to BiVO 4 with a 1 × 2 supercell, and the transferred charges are mainly distributed on the interfacial O atoms with negligible effect on the V atoms. Density of states (DOS) calculations were then conducted to determine the electronic structure of BiVO 4 . The calculated valance band (VB) maximum and conduction band (CB) minimum of monoclinic BiVO 4 with a standard space group of C2/c is mainly comprised of O 2p and V 3d orbitals ( Supplementary  Fig. 10), which is in good agreement with a previous report 35 . Remarkably, BP profoundly influences the VB electronic structure of BiVO 4 through the overlap of the P 2p and O 2p orbitals independent of the CB minimum of BiVO 4 (Fig. 2e). As the O 2p and V 3d orbitals contribute to the VB of BiVO 4 , the charge transfer that occurs on the O and V atoms at the p/n junction might be implicated in upward bending of the VB. As shown in Fig. 2f, The VB position is shifted upward from 2.48 eV for BiVO 4 to 2.23 eV for BP/BiVO 4 , implying that the positrons (holes) are the dominant carriers across the p/n junction under reverse bias 36 . According to the above experimental and theoretical results, Fig. 2g summarizes the possible band offsets and built-in potential of the BP/BiVO 4 heterojunction. Since the BP/BiVO 4 photoanode performs under external bias to facilitate electron transport from BP/BiVO 4 to the counter electrode in PEC water splitting, the external bias is therefore regarded as reverse bias. Positrons as the dominant carriers across the BP/BiVO 4 heterointerface would promote hole extraction from BiVO 4 to BP under external bias. To determine the hole extraction as a function of applied bias, the in-situ ultrafast transient absorption (TA) spectroscopy is performed to evaluate the hole trapping behaviours. As reported previously, the TA signal of surface-trapped holes for BP/BiVO 4 hybrid is mainly located in the wavelength ranging from 400 to 700 nm 27 . Therefore, the TA signal at 500 nm for BiVO 4 and BP/BiVO 4 anodes, respectively under open circuit potential and 0.8 V vs. Ag/AgCl in 0.5 M phosphate buffer (KPi, pH 7.1) are monitored. Compared with BiVO 4 anode, the intensity of the absorption signal for BP/BiVO 4 anode is significantly increased as anodic shifting the applied bias (Supplementary Fig. 11). In Fig. 3a, b, the decay signal was further fitted to a biexponential decay model with a fast component (τ 1 ) and a slow component (τ 2 ) 37,38 , and the lifetimes are summarized in Fig. 3c. Since the fast component, τ 1 , is associated with the hole being trapped at near band edge, the decreased τ 1 values for both BiVO 4 and BP/BiVO 4 photoanodes under applied bias can be considered the external bias that boosts hole transport 37  trapped at photoanode/electrolyte interface, which is expected to yield long-lived holes for water oxidation 37 . For BP/BiVO 4 anode, the two-fold increase in slow component suggests that the increased number of holes are extracted from BiVO 4 to photoanode/electrolyte interface under applied bias, which is consistent with our assumption from their band alignment.
PEC Performance Evaluation of BP/BiVO 4 anode. The improvement in the PEC performance of BiVO 4 imparted by the p-n junction formed with BP was investigated by measuring the J-V curves in 0.5 M phosphate buffer (KPi, pH 7.1) from rear side illumination (AM 1.5G, 100 mW cm −2 ). As shown in Fig. 3d, an obvious improvement in the photocurrent density of BiVO 4 is observed in the presence of BP, whereas the dark current density demonstrates that the water oxidation kinetics are slower for BP/BiVO 4 than for BiVO 4 ( Supplementary Fig. 12). These results imply that the enhanced photocurrent density of BP/BiVO 4 does not originate from surface water oxidation. As a consequence, a NiOOH layer that is a well-defined OEC for water oxidation was electro-deposited on the BP/BiVO 4 electrode to enhance the water oxidation kinetics ( Supplementary Fig. 13 Fig. 15). These results clearly illustrate that the superior PEC performance of the NiOOH/BP/BiVO 4 photoanode relative to that of NiOOH/ FeOOH/BiVO 4 results from the interfacial behaviour associated with hole transfer. To understand how hole transfer was improved by the presence of BP nanosheets, the charge separation efficiency (η sep ) was calculated by the following equation 40 : where J HS is the photocurrent density measured in a hole scavenger-containing electrolyte and J abs is associated with the maximum photocurrent density (J max ) and light harvesting  Fig. 16). However, the BP nanosheets might not act as an efficient photosensitizer to inject electrons, as the BP/BiVO 4 photoanode does not show a detectable photocurrent response at 520 nm in the presence of Na 2 SO 3 as a hole scavenger (Supplementary Fig. 17). As a result, J max and LHE can be established based on only the absorption of BiVO 4 ranging from 300 to 515 nm. The LHE and calculated J max value are shown in Supplementary Fig. 18 and the J HS is measured in the presence of Na 2 SO 3 as a hole scavenger ( Supplementary  Fig. 19). The corresponding η sep value is shown in Fig. 3f. The BP nanosheets clearly significantly improve the η sep of BiVO 4 in the entire voltage region. Excluding the possibility of electron injection by the excited BP layer, the enhanced η sep directly points to efficient hole extraction from the excited BiVO 4 photoanode to the BP layer. Nevertheless, due to its poor water oxidation ability ( Supplementary Fig. 12), the charge transfer efficiency (η tran ) of the BP/BiVO 4 photoanode, which is calculated as J Ph /J HS (J Ph is the photocurrent density measured in KPi electrolyte) 41 , is lower than that of the BiVO 4 photoanode (Supplementary Fig. 20). To understand the charge separation and transfer limitation, the J-V curves of BiVO 4 and BP/BiVO 4 photoanodes were measured from front illumination. As shown in Supplementary Fig. 21a, the photocurrent densities of BP/BiVO 4 photoanode are still better than that of BiVO 4 in the KPi electrolyte with and without hole scavenger. The calculated η sep is similar to the result obtained by rear illumination (Supplementary Fig. 21b), whereas the calculated η tran (Supplementary Fig. 21c) display different tendency from rear illumination. The enhanced η tran of BP/BiVO 4 photoanode from front illumination is unexpected, which might be ascribed to the effect of surface passivation by BP layers on the reduction of surface recombination that compensates poor water oxidation ability 42 . Therefore, the subsequent introduction of the NiOOH overlayer on BP/BiVO 4 is crucial for improving the hole capability towards efficient water oxidation. η tran of NiOOH/BP/ BiVO 4 (90%) is ∼2.7 times higher than that of BP/BiVO 4 , which has a value of 1.23 V vs RHE (Supplementary Fig. 22 and Fig. 3g), indicating the occurrence of strong synergistic effects between NiOOH and BP.
Hole Extraction Behaviour of BP Nanosheets. The capability of BP to perform hole extraction is further proven by electrochemical impedance spectroscopy (EIS) conducted at 0.6 V vs RHE in KPi electrolyte (Fig. 4a). The Nyquist plots were fitted by an equivalent circuit, as shown in the inset of Fig. 4a 41 . Based on the smaller charge transport resistance of BP/BiVO 4 relative to that of BiVO 4 , the results clearly show that the hole extraction behaviour of the BP nanosheets is superior to that of FeOOH. Furthermore, the bulk capacitances (C bulk ) for all photoanodes are almost same, which can be ascribed to the redox process of V 4+ /V 5+ 19,41 . The capacitances at the electrode/electrolyte interface for BP/BiVO 4 , NiOOH/FeOOH/BiVO 4 and NiOOH/BP/BiVO 4 are significantly increased, which can be related to the surface layer that modifies surface state of BiVO 4 43 . Figure 4b demonstrates the occurrence of long-lived hole storage by the BP nanosheet layer based on analysis of the transient cathodic current. t 1 , corresponding to photocurrent quenching under instantaneous light-off conditions, gradually decays to a steady state (t 2 ), thus showing a cathodic current. The delay in the steady-state cathodic current indicates that the separated holes that reach the electrode/electrolyte interface are not involved in water oxidation but are instead stored at the electrode surface. Therefore, the large value of t 2 -t 1 for BP/BiVO 4 indicates the presence of long-lived holes at the surface of BiVO 4 [44][45][46] . The charge storage behaviour of the NiOOH/BP/BiVO 4 photoanode against an applied bias can be calculated from the transient-state photocurrent based on the chronoamperometry curve measured under chopped illumination and linear sweep voltammetry (LSV) curves, respectively (Supplementary Fig. 23). The photocurrent drop from the transient state to the steady state can be ascribed to the number of holes stored 43 . Compared to the NiOOH/BiVO 4 photoanode, the number of holes stored by the NiOOH/BP/BiVO 4 photoanode is obviously higher across the entire potential region, especially at low bias (Fig. 4c). The fate of the holes extracted is to reach the surface, then participates in the water oxidation reaction at high potentials, where the injection barrier no longer impedes the charge transfer from the electrode to the electrolyte. These results further demonstrate the strong capability of the BP nanosheet layer to perform hole extraction towards water oxidation occurring at surface of OECs.
To further illustrate the impressive role of the BP nanosheet layer in hole extraction, two other well-defined OECs, CoOOH, and MnO x , were respectively spin-coated and photo-deposited on the BP/BiVO 4 photoanode surface (Supplementary Fig. 24). The enhancement in the PEC performance is evidenced by comparison of the catalysts deposited on BiVO 4 photoanodes, as demonstrated by the cyclic voltammetry (CV) curves in Fig. 4d. The enhancement factors induced by the BP layer are summarized in Supplementary Table 2, in which an average 1.5-fold enhancement is observed, and the NiOOH OEC overlayer demonstrates the highest PEC performance, which arises from its superior water oxidation capability (Supplementary Fig. 25).
PEC Water Splitting of NiOOH/BP/BiVO 4 anode. The BP nanosheet layer buried underneath the NiOOH layer exhibits a current density of 4.46 mA cm −2 at 1.23 V vs NHE for at least 200 min, as shown in Fig. 5a. Without the outmost NiOOH layer or with pure BiVO 4 (BP/BiVO 4 or BiVO 4 photoanode), the steady-state current density at 1.23 V vs NHE gradually drops. The fading of the photocurrent density can be ascribed to anodic photocorrosion of BiVO 4 by the surface-accumulated holes that arise from an insufficient water oxidation capability 47,48 . In addition, the oxygen gas surrounding BP may cause its self-oxidation, as determined by P 2p XPS analysis ( Supplementary  Fig. 26). Moreover, BiVO 4 combined with BP and NiOOH layers demonstrates superior PEC performance relative to other OEC/ BiVO 4 photoanodes and competitive values with those improved OEC/BiVO 4 photoanodes (Supplementary Table 3). The gas evolved from the NiOOH/BP/BiVO 4 photoanode was measured from the photocurrent density at 1.23 V vs RHE. As shown in Fig. 5b, the linear fitting plots of both H 2 and O 2 nearly overlap with the theoretical number of electrons. The excellent Faradic efficiency for O 2 evolution indicates that the hole-storing behaviour of BP does not impede the oxygen evolution reaction taking place at the NiOOH surface.
Furthermore, the long-term durability of the NiOOH/BP/ BiVO 4 photoanode was investigated at 1.23 V vs RHE. This test was performed for 60 h and rested for 2 h with an interval of 20 h. As shown in Fig. 5c, the photocurrent density of the NiOOH/BP/BiVO 4 photoanode is stable with a slight fluctuation between 4.31 and 4.56 mA/cm 2 . The morphology after testing is well-maintained (Insert in Fig. 5c). The result is a sharp contrast to the CoOOH/BP/BiVO 4 photoanode which shows a rapid decrease in photocurrent density after continuous testing of 2 h (Supplementary Fig. 27). The XPS results of NiOOH/BP/BiVO 4 photoanodes before and after long-term testing are shown in Supplementary Fig. 28, which indicates that the BP in the NiOOH/BP/BiVO 4 photoanode is oxidized to a lesser extent. However, the BP in the CoOOH/BP/BiVO 4 is almost completely oxidized ( Supplementary Fig. 29). It is a fact that the BP is able to be slowly oxidized during PEC testing, whereas the electrodeposited NiOOH is believed to have conformal coverage which impedes the oxidization of BP.

Discussion
In this study, we have demonstrated that a layer of BP nanosheets can serve as an excellent hole extraction layer in a BiVO 4 /OEC photoanode for solar water splitting. The BP nanosheets, which were exfoliated from layered bulk BP, had the unique merit of ptype conductivity, hence enabling the formation of a p/n heterojunction with BiVO 4 , which facilitated hole transfer from BiVO 4 to the OEC surface. As a result, the NiOOH/BP/BiVO 4 photoanode exhibited a photocurrent density of 4.48 mA·cm −2 at 1.23 V vs RHE under AM 1.5 illumination, which was 4.2 times higher than that of pure BiVO 4 and 1.5 times higher than that of NiOOH/BiVO 4 . The BP layer was found to store separated holes and then transfer them to the OEC surface, and this impressive function was universal for other OECs, such as CoOOH and MnO x . Moreover, the burying of the BP nanosheets by the OEC layer alleviated self-oxidation, thereby prolonging the stability of photoelectrochemical water splitting by BiVO 4 . Our work shows the potential for application of BP in solar energy conversion devices, nevertheless, a uniform coating of BP on photoanodes with strongly coupled interface is still desired for further optimization.

Methods
Materials. All chemical reagents were purchased from Aldrich without further purification. FTO was purchased from TEC-8, Pilkilton with a resistance of 14 Ω. BP crystal was purchased from Mukenano Co. LTD.
Synthesis of BP/BiVO 4 Photoanodes. BiVO 4 electrodes were prepared based on Lee and Chio's method 13 . Bulk BP crystals were exfoliated by ultrasonication in a mixture of γ-butyrolactone (GBL) and IPA. Briefly, 20 mg of BP crystals was dispersed into 20 mL of the mixture and sonicated for 10 h at 300 W. The resultant dispersion was centrifuged at 2000 rpm for 60 min The exfoliated BP sheets were dispersed again in IPA at a concentration of 0.02 mg/mL and stored under flowing N 2 . The as-prepared BiVO 4 photoanode with a size of 1 × 2 cm 2 was placed against the wall of a 50 mL centrifuge tube with the sample side facing up. Then, 50 mL BP/ IPA dispersion was added to the centrifuge tube and centrifuged by 1000 rpm for 1 min For self-absorption or deposition of BP on BiVO 4 photoanode, the as-prepared BiVO 4 photoanode with a size of 1 × 2 cm 2 was immersed in 10 mL of the BP sheet dispersion for 2 h in a glovebox. All BP/BiVO 4 photoanode was further dried at 50 ℃ in a vacuum oven. Material Characterization. SEM images of the products were recorded on a fieldemission scanning electron microscope (JSM-7000F, Japan). The XRD patterns were obtained with a D500/5000 diffractometer operated in Bragg-Brentano geometry and equipped with a Cu-Kα radiation source. The HR-TEM observations were performed on a JEOL JEM-AFM 200F (Japan) electron microscope with (Cscorrected/energy-dispersive X-ray spectroscopy (EDS)/EELS). The VB-XPS and XPS measurements were performed on an auger electron spectroscopy (AES) XPS instrument (ESCA2000 from VG Microtech in England) equipped with an aluminum anode (Al Kα, λ = 1486.6 eV). The UV-Vis DRS spectra were recorded using a UV-Vis spectrophotometer (Shimadzu UV-2550). Raman spectra were measured by using LabRam Aramis equipment (Horriba Jovin Yvon Inc., US). The AFM was measured by using Bruker Multimodel-8 equipment.
In-situ time-resolved transient absorption spectroscopy. The Laser flash photoelectrochemical water splitting measurements with transmission detection were performed with the third harmonic of the Nd;YAG laser (10Hz, NT342A, EKSPLA, 355 nm (3.5 mJ/pulse)) as excitation source and the Xe lamp (continuous wave, 300W, Newport) as the probe light source, in a three electrode system with working electrode (3 × 3 cm 2 ), Pt counter electrode and Ag@AgCl reference electrode, and N 2 saturated 0.5 M KPi buffer electrolyte. The transmitted probe light was focused on a monochromator (Princeton Instruments, Acton SpectraPro SP-2300). The output of the monochromator was monitored using a photomultiplier Tube (PDS-1, Dongwoo Optron). The transient signals were passed through an amplifier (SR445A, Stanford Research Systems) and then recorded by a digital oscilloscope (350MHz, MDO4034C, Tektronix). Photoanodes were placed in a sealed reactor with Argon purged phosphate buffer electrolyte. The applied bias was controlled with a PGSTAT204 potentiostat (Metro Autolab).
Computational method. Density functional theory calculations are performed using the plane-wave basis sets in VASP code 51 . The ion-electron interaction is treated by the projected-augmented wave (PAW) approximation 52 . The exchangecorrelation functional is expressed with generalized gradient approximation Perdew−Burke−Ernzerhof (PBE-GGA) 52 . The energy cutoff for the plane-wave basis is set to 400 eV and the convergence threshold is set as 10 −4 eV in energy and 0.01 eVÅ −1 in force. DFT+D2 method is adopted to describe the Grimme vdw correction during structure simulation and electronic calculation 53 Prior to the PEC measurements, the electrolyte was purged with N 2 to remove dissolved oxygen. In a typical J-V measurement, linear sweep voltammetry was conducted at a scan rate of 20 mV s −1 . The potentiostatic mode was used to measure the electrochemical impedance spectra (EIS) with an AC voltage amplitude of 5 mV and a frequency range of 0.01-100 kHz under AM 1.5G illumination. When doing the record, a silver paste was painted on the top to increase the conductivity and an aperture was used to determine the contact area between the samples and the electrolyte.
The gas evolution was carried out in a quartz reactor, which was sealed with rubber plugs and Parafilm. The electrode (1.5 cm 2 ) was immersed in the electrolyte in a three-electrode configuration with a 1.23 V vs RHE. Prior to the reaction and the sealing process, the electrolyte was purged with N 2 gas. 1 mL of gas was analyzed by gas chromatography (Agilent Technologies 7890A GC system, USA) using a 5 Å molecular sieve column and Ar as the carrier gas. The experimental error for the evolution of H 2 and O 2 was considered to be ≈3%.
The theoretical electron number as a function of the J-t curve was calculated on the basis of an area of 1.5 cm 2 .
Theoretical electron number ¼ Z t¼200min t¼0min current density 1:5 6:24146 10 18 ð4Þ The photocurrent-to-H 2 conversion efficiency and photocurrent-to-O 2 conversion efficiency were determined on the basis of their linear slopes (i.e., theoretical electron number 2 for the photocurrent-to-H 2 conversion efficiency and theoretical electron number 4 for the photocurrent-to-O 2 conversion efficiency).
Calculation of the theoretical photocurrent in BiVO 4 photoanodes. The single photon energy is calculated from Eq. (5) where E(λ) is the photon energy (J), h is Planck's constant (6.626 × 10 −34 Js), C is the speed of light (3 × 10 8 m s −1 ) and λ is the photon wavelength (nm). The solar photon flux is then calculated according to Eq. (6) Flux where Flux(λ) is the solar photon flux (m −2 s −1 nm −1 ), and P(λ) is the solar power flux (W m −2 nm −1 ).1 The theoretical maximum photocurrent density under solar illumination (AM1.5), J max (A m −2 ), is then calculated by integrating the solar photon flux between 300 to 515 nm, shown in Eq. (7): where e is the elementary charge (1.602 × 10 −19 C). The theoretical photocurrent of such BiVO 4 photoanodes is accordingly calculated to be 6.87 mA cm −2 based our solar spectra.

Data availability
The data supporting the findings of this study are available within the article and its supplementary information files and from the corresponding author upon reasonable request.