Introduction

Photoelectrochemical (PEC) water splitting has been considered as a promising strategy for converting solar light into hydrogen energy1,2,3. To achieve its practical applications, the design and fabrication of semiconductor photoanodes with sufficient light absorption, effective charge separation, and high surface reactivity are essentially required4,5. Among various candidates, bismuth vanadate (BiVO4) has been attracted particular attentions owing to its appropriate bandgap (2.4 eV) and suitable band-edge positions6,7,8,9,10. However, suffering from the high charge-recombination and sluggish oxygen evolution reaction (OER) kinetics, most of reported photocurrent densities of BiVO4 photoanodes are far below the theoretical expectation (7.5 mA cm−2, AM 1.5 G illumination, 100 mW cm−2)11,12,13. During past decades, diverse strategies have been developed to improve the PEC activities of BiVO4 photoanodes, including elemental doping14,15,16, facet tailoring17,18,19, and hetero-junction20,21,22,23,24, etc. Although the PEC performances have been increased to a certain extent owing to the improved carrier mobility as well as electrical conductivity, the intrinsically poor surface reactivity still seriously restricts the PEC conversion efficiency.

Recently, BiVO4 photoanodes decorated with various transition-metal catalysts have been extensively reported for remarkably promoting the OER activities25,26,27,28,29. Specifically, they could efficiently extract photo-generated holes, minimize over potential, and provide active sites, which are all beneficial to accelerate the PEC water oxidation kinetics. Among various OER catalysts, the VIII metal (Fe, Co, Ni) oxides or (oxy)hydroxides, especially for NiFe-based materials, have attracted particular interests in recent years30,31,32,33,34. For example, Domen et al35. deposited NiFe bimetallic catalyst on BiVO4 photoanodes for improving the PEC activities up to 4.2 mA cm−2 at 1.23 VRHE. Zhang and co-workers reported that BiVO4 photoanodes modified with NiFe complexes exhibited an excellent photocurrent of 5.10 mA cm−2 31. Pihosh et al.36 fabricated a WO3/BiVO4/CoPi core-shell nanostructured photoanode that achieves near 90% of the theoretical water splitting photocurrent. On this basis, Kosar et al.37 acquired a highly efficient solar-to-hydrogen conversion efficiency of 7.7% by photovoltaic cell and WO3/BiVO4/CoPi core-shell nanorods PEC cell tandem. Despite the crucial roles of OER catalysts for enhancing PEC behaviors have been well established, much less attentions focused on optimizing their electronic structures to further boost the PEC conversion efficiency, especially for bimetallic catalysts.

Herein, we reported the incorporation of non-metallic nitrogen-atom into NiFeOx catalysts to rationally tailor the electronic structure, which remarkably promoted the photocurrent density of BiVO4 photoanodes up to 6.4 mA cm−2 at 1.23 VRHE under AM 1.5 G (100 mW cm−2) with an excellent durability. The outstanding PEC performances should be attributed to the electronic reconstruction in both NiFeOx and BiVO4, resulting from the partial substitution of O sites by low electronegativity N atoms. Specifically, the weak electron-attracting capacity of N atoms led to the electron enrichments on both Fe and Ni sites. Subsequently, the electron injection from Ni atoms to lattice V sites of BiVO4 was favorable for improving the oxygen-evolution stability, while the Fe sites could effectively attract holes for promoting the PEC activity. This work firstly demonstrates the rational regulation of electronic structures in OER catalysts as well as fundamental understanding of their intrinsic roles in PEC oxygen evolution reaction.

Results

Morphology and structure characterizations

The nanoporous BiVO4 photoanodes supported on F-doped SnO2 (FTO) glass substrates were fabricated by an electrochemical deposition associated with calcination treatment10. Figure 1a shows the scanning electron microscopy (SEM) images of the obtained BiVO4 photoanodes, clearly revealing their unique worm-like porous structure with an average diameter of 200–300 nm. Additionally, the high-resolution transmission electron microscopy (HR-TEM) image (Supplementary Fig. 1) clearly indicates that these nanocrystals possess a relatively smooth surface and a lattice spacing of 0.311 nm corresponded to (−130) plane of monoclinic BiVO4 phase. Interestingly, after the decoration of N:NiFeOx catalysts, the smooth surfaces of pristine BiVO4 photoanodes transformed into a rough flocculent-structure (Fig. 1b). The HR-TEM images (Fig. 1c and Supplementary Fig. 2) clearly indicate that an amorphous layer of N:NiFeOx catalysts was uniformly covered on BiVO4 surfaces with a thickness of ~4 nm. Moreover, Fig. 1d and Supplementary Fig. 3 show the energy dispersive spectroscopy (EDS) elemental line and mapping images, revealing the uniform distributions of N, Ni and Fe elements on BiVO4 crystal surfaces. Besides, the X-ray photoelectron spectroscopy (XPS) result also confirms the successful incorporation of the nitrogen element into the NiFeOx layer (Supplementary Fig. 4). However, compared with pristine BiVO4 photoanodes, no evident peak change could be observed in the X-ray diffraction (XRD) patterns after the decoration of N:NiFeOx catalysts (Supplementary Fig. 5), which should be due to their amorphous structure and ultrathin thickness.

Fig. 1: Morphology and structure characterization of the synthetic photoanodes.
figure 1

SEM images of pristine BiVO4 (a) and BiVO4/N:NiFeOx (b) photoanodes; HR-TEM image (c) and TEM-EDS element mapping analysis (d) of BiVO4/N:NiFeOx photoanodes.

Full size image

Photoelectrochemical properties

The PEC water splitting performances of N:NiFeOx catalyst decorated BiVO4 photoanodes (marked as BiVO4/N:NiFeOx) were measured in 0.5 M K3BO3 (pH = 9.5) electrolyte under AM 1.5 G illumination (100 mW cm−2). For comparison, the PEC activities of pristine BiVO4 as well as NiFeOx decorated BiVO4 photoanodes (marked as BiVO4/NiFeOx) have also been studied. As shown in Fig. 2a and Supplementary Fig. 6, the pristine BiVO4 photoanodes exhibit a relatively low photocurrent density (2.1 mA cm−2 at 1.23 VRHE), suffering from the sluggish oxygen evolution kinetics at anode/electrolyte interfaces. Obviously, the decoration of NiFeOx catalysts on BiVO4 photoanodes could effectively enhance the PEC water oxidation activity, and the photocurrent density has been increased up to 4.4 mA cm−2 at 1.23 VRHE. Amazingly, an outstanding photocurrent density of 6.4 mA cm−2 at 1.23 VRHE has been achieved on BiVO4/N:NiFeOx photoanodes accompanied by a lower onset potential for OER (Supplementary Fig. 7), clearly indicating that the incorporation of N-atom in NiFeOx catalysts could significantly promote the oxygen evolution activity (Supplementary Fig. 8 and Figs. 9 and 10). Furthermore, their maximum half-cell applied bias photon to current efficiencies (HC-ABPE) have been calculated and shown in Fig. 2b. The HC-ABPE value of BiVO4/N:NiFeOx photoanode could be achieved up to 1.9% at 0.73 VRHE, which is much higher than that of BiVO4/NiFeOx (1.1% at 0.8 VRHE) and pristine BiVO4 (0.29% at 0.96 VRHE), respectively. Except for the high conversion efficiency, the high stability and durability of photoelectrodes are also required for future practical applications. Figure 2c shows the current-time (i-t) curves of these photoanodes operated at 1.23 VRHE. Note that due to the serious photo-corrosion and V5+ dissolution from crystal lattices, the pristine BiVO4 exhibited the relatively poor PEC stability and the photocurrent density rapidly decreased38,39,40. Although the loading of NiFeOx catalysts on BiVO4 surfaces could improve the PEC stability to a certain extent, the photocurrent density also decreased down to 2.8 mA cm−2 after 5 h test. Interestingly, BiVO4/N:NiFeOx photoanodes possess the excellent photocurrent stability during the whole test process, indicating the positive effects of N:NiFeOx on restraining V5+ dissolution from BiVO4 lattices and the obtained photoanodes with excellent structural stability (Supplementary Fig. 11, Figs. 12 and 13, and Table 1). The above results clearly reveal that the incorporation of N atoms in NiFeOx catalysts not only significantly promotes the oxygen evolution activity but also effectively enhances the PEC stability of BiVO4 photoanodes.

Fig. 2: Photoelectrochemical properties.
figure 2

a Linear-sweep voltammograms (LSV, with a scan rate of 10 mV s−1), b half-cell ABPE (HC-ABPE) results, c I–t stability tests measured at 1.23 V vs. RHE, d IPCE results at 0.6 V vs. RHE and e EIS results at 0.75 V vs. RHE under illumination for BiVO4, BiVO4/NiFeOx, and BiVO4/N:NiFeOx photoanodes. f H2 and O2 evolution of BiVO4/N:NiFeOx measured at 1.23 V vs. RHE. All the measurements were carried at 0.5 M K3BO3 (pH = 9.5) electrolyte.

Full size image

Furthermore, their incident photon to current conversion efficiencies (IPCEs) were conducted and shown in Fig. 2d (Supplementary Fig. 14). At the wavelength of 360 nm, the IPCE values of BiVO4/N:NiFeOx photoanodes could be achieved to 93%, which is much higher than BiVO4 (8%) and BiVO4/NiFeOx (54%). Figure 2e shows the electrochemical impedance spectroscopy (EIS) for further elucidating the interfacial charge transfer and oxygen evolution kinetic. According to the Nyquist plots and the fitting results (Supplementary Table 2), the calculated resistance values of BiVO4/N:NiFeOx, BiVO4/NiFeOx, and BiVO4 photoanodes were 139.5, 149.8, and 458.9 Ω, respectively, revealing the preferable capability of N:NiFeOx catalyst for facilitating interface charge transfer. Moreover, the hydrogen and oxygen amounts generated from PEC water splitting over BiVO4/N:NiFeOx photoanodes were measured by an online gas chromatography (GC). After 3 h irradiation, the amounts of H2 and O2 increased linearly up to 365.3 and 165.8 μmol, respectively (Fig. 2f). Additionally, an average Faradaic efficiency of nearly 95% has been obtained on BiVO4/N:NiFeOx photoanodes, further confirming its excellent oxygen evolution capability.

Spectrum and electrochemical analysis

The photoluminescence spectroscopy (PL) has been measured by a fluorescence spectrophotometer under laser excitation of 355 nm. As shown in Fig. 3a, two PL peaks could be clearly identified. More specifically, the peak at 470 nm is associated with FTO (SnO2) substrate (Supplementary Fig. 15). The peak at 493 nm near the absorption band edge of BiVO4 (Supplementary Fig. 16) is attributed to radiative recombination of hole in O 2p band and electron in V 3d band, which represents the charge recombination ability41,42. Specifically, the pristine BiVO4 photoanodes exhibited a very strong PL peak, demonstrating the relatively high electron-hole recombination ratios. However, after the decoration of OER catalysts, the PL peak intensities have been evidently reduced. More specifically, N:NiFeOx catalysts exhibit more efficient capability than NiFeOx for promoting the charge separation of BiVO4 photoanodes (Supplementary Fig. 17). Moreover, the time-resolved transient absorption spectra (TR-TAS) have been performed to explore the energy relaxation process and the charge carrier concentrations of the related samples under the excited state43,44,45. In addition, the decay curves were probed at 490 nm, which was attributed to hole dynamics43. As shown in Fig. 3b, c (Supplementary Table 3), the BiVO4/N:NiFeOx photoelectrodes possess higher absorption peak and longer carrier lifetime (2.69 μs) compared with BiVO4/NiFeOx (1.77 μs) and BiVO4 (1.51 μs) photoanodes. Based on the above steady/transient spectra analysis, it can be concluded that N:NiFeOx catalysts exhibited the preferable capability for promoting charge separation and extending the carriers lifetimes. Figure 3d shows their interfacial charge transfer (ηtrans) efficiencies for water oxidation reaction (Supplementary Fig. 18). The pristine BiVO4 exhibits a very low efficiency of 28% at 1.23 VRHE, while the surface deposition of NiFeOx and N:NiFeOx catalysts could effectively increase the ηtrans efficiencies up to 66.3 and 88%, respectively. Furthermore, their electrochemical OER properties under dark conditions have also been studied and shown in Fig. 3e. Obviously, BiVO4/N:NiFeOx possesses a lower overpotential and higher water oxidation current compared with BiVO4 and BiVO4/NiFeOx, further revealing its excellent OER activity. The large-scale fabrication of photoanodes should be necessarily required for future practical applications. Accordingly, the dual BiVO4/N:NiFeOx photoanodes with a relatively large area (2 × 3.5 cm2) have been fabricated, and the photocurrent could achieve up to 37 mA at 1.23 VRHE accompanied with an excellent stability of 10 h (Fig. 3f). Thus, the above results clearly demonstrate that the BiVO4/N:NiFeOx photoanodes possess the tremendous potential for practical PEC water splitting applications.

Fig. 3: Spectrum and electrochemical OER properties.
figure 3

a PL spectra; b transient absorption (TA) spectra; c the time-resolved TA curves probed at 490 nm; d charge transfer efficiencies (ηtrans); and e LSV curves in the dark of BiVO4, BiVO4/NiFeOx, and BiVO4/N:NiFeOx photoanodes. f I–t curves of the scale-up fabricated BiVO4/N:NiFeOx photoanodes (2 × 3.5 cm2) with parallel at 1.23 VRHE.

Full size image

Effects of N-incorporation into BiVO4/NiFeOx films

Furthermore, the effects of the N-incorporation on the surface chemical states and electronic structures of both NiFeOx and BiVO4 have been explored by XPS. As shown in Supplementary Figs. 19 and 20, no evident change could be detected in Bi 4f peaks of BiVO4/N:NiFeOx compared with BiVO4/NiFeOx and BiVO4, revealing the negligible influence of N-substitution on the Bi sites of BiVO4 photoanodes. Interestingly, compared with BiVO4/NiFeOx samples, a shoulder peak at lower binding energy positions could be observed in V 2p spectra of BiVO4/N:NiFeOx photoanodes (Fig. 4a), which should be attributed to the formation of V(5−x)+ species. Moreover, the relative ratio of Ni3+ species in BiVO4/N:NiFeOx has been evidently increased (Fig. 4b)46,47,48,49,50, while the Fe3+ ratio has been decreased compared with BiVO4/NiFeOx samples (Fig. 4c)51,52,53. On the basis of above results, it may be proposed that the partial substitution of O sites in NiFeOx catalysts by N atoms should enrich the electron densities in Fe and Ni sites. Furthermore, the electron-enriched Ni sites conversely donated electrons to V sites of BiVO4 for restraining V5+ dissolution and improving the PEC stability. Additionally, the electron-enriched Fe sites could efficiently attract photo-generated holes from BiVO4 surfaces, which significantly promoted the oxygen-evolution activity. Thereby, the N-incorporation in NiFeOx catalysts could effectively promote the oxygen evolution activity and stability of BiVO4 photoanodes. To further confirm the above speculations, the N atoms in N:NiFeOx catalysts were replaced by O atoms again via an oxygen plasma treatment. As shown in Supplementary Figs. 20 and 21, all the XPS peaks of Fe, Ni, Bi, and V elements were nearly consistent with BiVO4/NiFeOx photoanodes. Additionally, after the replacement of N with O atoms, the photocurrent density has been obviously reduced from 6.4 to 5 mA cm−2 at 1.23VRHE accompanied by the poor PEC water oxidation stability (Supplementary Fig. 22), confirming the crucial roles of N-incorporation in promoting the OER activity and PEC stability of BiVO4 photoanodes.

Fig. 4: Effects of N-incorporation on the surface chemical states and electronic structures of BiVO4/NiFeOx.
figure 4

XPS high-resolution a V 2p, b Ni 2p, and c Fe 2p spectra for BiVO4/N:NiFeOx and BiVO4/NiFeOx photoanodes, respectively. Schematic of charge density difference (yellow and cyan represent charge accumulation and depletion, respectively; the cut-off of the density-difference isosurface is 0.01 Å−3) of BiVO4/NiFeOx (d) and BiVO4/N:NiFeOx (e).

Full size image

DFT calculation and analysis

Furthermore, the density functional theory (DFT) calculation has been performed to reveal the change of electron densities in Fe and Ni sites after incorporation of N atoms. As shown in Fig. 4d, e, the charge density difference results clearly reveal that the electron densities at Fe and Ni sites increased significantly (yellow regions) after partial substitution of O sites with N atoms in NiFeOx catalysts. Additionally, the Bader charge analysis (Supplementary Table 4) also verified the enriched electron densities with the N-incorporation, which is highly consistent with the XPS results. Thereby, these calculation results could further provide supports on the crucial roles of N-incorporation in regulating the electronic structures of NiFeOx.

Discussion

In summary, we reported a facile N-incorporation method to rationally regulate the electronic structures of NiFeOx catalyst decorated on BiVO4 photoanodes. More detailed experiments and XPS analysis reveal that owing to the relatively low electronegativity of N atoms, their incorporation in NiFeOx catalysts facilitates the electron enrichments in Fe and Ni sites. Furthermore, the Ni sites would donate electrons to V sites on BiVO4 surface, which could efficiently restrain V5+ dissolution and improve the PEC water oxidation stability. Moreover, the enhanced hole-attracting ability of Fe sites significantly promotes the oxygen-evolution activity. As expected, the BiVO4/N:NiFeOx photoanodes exhibited an outstanding photocurrent density of 6.4 mA cm−2 at 1.23 VRHE (AM 1.5 G, 100 mW cm−2) accompanying with the enhanced PEC stability. This work provides a new insight to construct highly efficient and stable OER catalysts for fabricating high-efficiency PEC devices.

Methods

Materials

All chemicals were of analytical grade purity, obtained from Sinopharm Chemical Reagent Co., Ltd., and used as received without further purification. Deionized water (Molecular Corp., 18.25 MΩ cm) used in the synthesis was from local sources.

Synthesis of nanoporous BiVO4 photoanodes

The nanoporous BiVO4 photoanode was obtained based on the previous report10. 2 mM Bi(NO3)3·5H2O was dissolved in 0.4 M KI solution (50 mL). Then, the pH value of this solution was adjusted to 1.7 by HNO3. Subsequently, 0.23 M quinhydrone was dispersed into ethanol solution (20 mL). Finally, mixing the two solution and stirring vigorously for a few minutes to acquire the electrodeposited solution. The cathodic deposition was performed at a constant potential of −0.1 V vs. Ag/AgCl for 3 min at room temperature to obtain the BiOI electrodes, among which FTO, Ag/AgCl (4 M KCl), and platinum pair were used as working electrode (WE), reference electrode (RE) and counter electrode (CE), respectively. Then, VO(acac)2 (0.2 M, 0.2 mL) dissolved in DMSO (10 mL) solution was coated on the BiOI electrodes and heated in the air in a muffle furnace at 450 °C for 2 h (ramping rate = 2 °C/min) to convert to BiVO4. After calcination, the excess V2O5 on the electrode surface was soaked into NaOH (1 M) solution for 15 min to remove. Finally, the electrodes were rinsed with deionized water and dried in air to obtain pure BiVO4 photoanodes.

Synthesis of BiVO4/NiFeOx and BiVO4/N:NiFeOx photoanodes

The as-prepared BiVO4 electrodes were immersed into the freshly mixed metal salt solution (pH~2.8) for 15 min (2.5 mL of 10 mM FeCl3•6H2O and 7.5 mL of 10 mM NiCl2•6H2O). Then, 2 M NaOH solution was added to adjust its pH to ~8 (The electrodes were still kept in this mixed solution during pH adjustment process). Subsequently, the solution was stood for 50 min and maintained at 25 °C throughout the co-catalyst loading process. Finally, the electrodes were washed by deionized water and calcined at 300 °C for 1 h in a muffle furnace in air atmosphere to obtain the BiVO4/NiFeOx samples. The synthesis of BiVO4/N:NiFeOx and BiVO4/O2-NiFeOx photoanodes was the same as the above steps for the preparation of BiVO4/NiFeOx, except that the final calcination process is changed to a N2 or O2 plasma treatment for 5 min (a medium power of 10.5 W and a pressure of 300 Pa, Supplementary Fig. 23). The BiVO4/N:NiFeOx-O2 photoanodes were prepared via an oxygen plasma treatment BiVO4/N:NiFeOx for 5 min.

Synthesis of BiVO4/NiFeP and BiVO4/NiFeB photoanodes

The BiVO4 photoanodes were immersed into the fresh metal salt solution (2.5 mL of 10 mM FeCl3•6H2O and 7.5 mL of 10 mM NiCl2•6H2O), and then a NaBH4 aqueous solution was added dropwise. The solution was stood for 50 min. Finally, the electrodes were washed by deionized water to obtain the BiVO4/NiFeB photoanodes. Firstly, the BiVO4 films were dipped into a water solution (20 mL) A containing SnCl2 (0.8 g) and HCl (40 wt%, 0.8 mL) for 2 min. Secondly, the films were further immersed into a water solution (20 mL) B of PdCl2 (10 mM, 3.4 mL), HF (40–50 wt%, 0.16 mL) and HCl (40 wt%, 0.2 mL) for 2 min. Finally, the films were immersed into the solution C at 60 °C for 40 s and then rinsed with deionized water to obtain the BiVO4/NiFeP photoanodes. The water solution (20 mL) C contains NiSO4•6H2O (0.15 g), FeSO4•4H2O (0.15 g), NH4F (0.2 g), NaH2PO2•H2O (0.8 g), and Na3C6H5O7·2H2O (0.4 g), and the value of pH was further adjusted to 9.0 by adding ammonia54.

Measurement and characterization

Scanning electron microscopy measurements were carried out on a field-emission scanning electron microscope (SEM, SU8020). Transmission electron microscopy (TEM) measurements were performed by using a FEI Tecnai TF20 microscope operated at 200 kV. The elemental composition and chemical valence states were explored by X-ray photoelectron spectroscopy (XPS, Al-Kα, 1486.6 eV, ESCALAB 250Xi). The crystalline structures were identified by X-ray diffraction analysis (XRD, Smartlab-SE). UV-visible diffuse reflectance spectra were performed on a UV-2550 (Shimadzu) spectrometer by using BaSO4 as the reference.

Spectrum measurements

The photoluminescence (PL) spectra were tested on F-7000 fluorescence spectrophotometer (Hitachi, Tokyo Japan) under laser excitation of 355 nm. The time-resolved transient absorption (TA) spectra were performed on LP980 spectrometer (Edinburgh Instruments Ltd., model LP980), combined with a compact Q-switched Nd:YAG laser (Continuum, the USA). The probe source was a 150 W pulsed Xenon lamp for kinetic and spectral studies. The measurements were achieved with single-flash laser excitation at 355 nm (10 Hz, FWHM~7 ns) as the pump source. The kinetic traces and transient absorption spectra were collected with a Hamamatsu R928 photomultiplier tube detector (PMT) and an iStar ICCD camera (Andor Technology), respectively. The samples were placed in a film holder, which is suitable for semi-transparent materials. The obtained data were analyzed with the Edinburgh software (LP900). In addition, the decay curves were probed at 490 nm (Fig. 3c), and their fitting was based on a biexponential decay model according to the following equation and the fitting parameters have been listed in Supplementary Table 3.

$$\left({{{{{\rm{R}}}}}}({{{{{\rm{t}}}}}})={B}_{1}{e}^{(-\frac{t}{{\tau }_{1}})}+{B}_{2}{e}^{(-\frac{t}{{\tau }_{2}})}\right)$$
(1)

Photoelectrochemical measurements

The photoelectrochemical measurement was carried out on an electrochemical workstation (CHI760E) in a standard three-electrode system and a 0.5 M K3BO3 electrolyte (pH = 9.5) under AM 1.5 G simulated sunlight (100 mW cm−2). A dual-channel power and energy meters (PM320E, THORLABS) equipped with high-sensitivity S310C probe (THORLABS) was used to calibrate the AM 1.5 light intensity to 100 mW/cm2. Moreover, the solar simulator used in our experiments has been equipped with a total-reflection mirror and AM 1.5 G fitter for PEC measurements and the corresponding spectrum has been measured by a spectrometer (BLUE-Wave, StellarNet) and shown in Supplementary Fig. 24. The photocurrent vs. voltage (J–V) characteristics were determined by scanning potential from −0.6 to 1.0 V (vs. Ag/AgCl) with a scan rate of 10 mV s −1 and the applied potentials could be converted into reversible hydrogen electrode (RHE) using the Nernst equation:

$${E}_{{{{{{\mathrm{RHE}}}}}}}={E}_{{{{{{\mathrm{Ag}}}}}}/{{{{{\mathrm{AgCl}}}}}}}+0.059{{{{{\mathrm{pH}}}}}}+0.197(25\,^\circ {{{{{\rm{C}}}}}})$$
(2)

The incident photon to current efficiency (IPCE) was determined using a full solar simulator (Newport, Model 9600, 300 W Xe arc lamp) and a motorized monochromator (Oriel Cornerstone 130 1/8 m) at 0.6 VRHE in a 0.5 M K3BO3 electrolyte. The IPCE result was calculated using the equation55:

$${{{{{\mathrm{IPCE}}}}}}( \% )=\frac{1240\times I({{{{{\mathrm{mA}}}}}}/{{{{{\mathrm{c}}}}}}{{{{{{\mathrm{m}}}}}}}^{2})}{{P}_{{{{{{\mathrm{light}}}}}}}({{{{{\mathrm{mW}}}}}}/{{{{{\mathrm{c}}}}}}{{{{{{\mathrm{m}}}}}}}^{2})\times \lambda ({{{{{\mathrm{nm}}}}}})}\times 100$$
(3)

where I is the measured photocurrent density at specific wavelength, λ is the wavelength of incident light, and Plight is the measured light power density at that wavelength.

Supposing 100% Faradaic efficiency, the half-cell applied bias photon-to-current efficiency (HC-ABPE) was calculated by following equation55:

$$({{{{{\mathrm{HC}}}}}}-{{{{{\mathrm{ABPE}}}}}})( \% )=\frac{I({{{{{\mathrm{mA}}}}}}/{{{{{\mathrm{c}}}}}}{{{{{{\mathrm{m}}}}}}}^{2})\times (1.23-{V}_{{{{{{\mathrm{bias}}}}}}})(V)}{{P}_{{{{{{\mathrm{light}}}}}}}({{{{{\mathrm{mW}}}}}}/{{{{{\mathrm{c}}}}}}{{{{{{\mathrm{m}}}}}}}^{2})}\times 100$$
(4)

where I is the photocurrent density, Vbias is the applied potential, and Plight is the incident illumination power density (100 mW cm−2).

The electrochemical impedance spectroscopy (EIS) Nyquist plots were obtained at 0.75 V (vs. RHE) with a small AC amplitude of 10 mV in the frequency range of 10−2 to 105 Hz and the measured spectra were fitted with Zview software.

Surface charge transfer efficiencies (ηtrans) of BiVO4, BiVO4/NiFeOx, and BiVO4/N:NiFeOx photoanodes can be calculated using the following equation56:

$${\eta }_{{{{{{\mathrm{trans}}}}}}}=\frac{{J}^{{{{{{{\mathrm{H}}}}}}}_{2}{{{{{\mathrm{O}}}}}}}}{{J}^{{{{{{{\mathrm{H}}}}}}}_{2}{{{{{{\mathrm{O}}}}}}}_{2}}}$$
(5)

\({J}^{{{{{{{\mathrm{H}}}}}}}_{2}{{{{{\mathrm{O}}}}}}}\) and \({J}^{{{{{{{\mathrm{H}}}}}}}_{2}{{{{{{\mathrm{O}}}}}}}_{2}}\) are the photocurrent densities obtained in 0.5 M potassium borate electrolytes (pH 9.5) without and with H2O2, respectively. Additionally, a summary of recent significant progress of BiVO4-based photoanodes has been reviewed (Supplementary Table 5).

The evolution of H2 and O2 was performed in a 0.5 M K3BO3 electrolyte at 1.23 VRHE under AM 1.5 G illumination (100 mW cm−2) by an online gas analysis system (Labsolar 6 A, Beijing Perfect light Technology Co. Ltd.) and a gas chromatograph (GC 7890 A, Agilent Technologies).

The PEC performances of two parallel BiVO4/N:NiFeOx photoanodes (single area: 2 × 3.5 cm2, distance: ~1 cm) were performed at 1.23 VRHE in 0.5 M K3BO3. Specifically, the simulated solar light illuminates vertically these two photoanodes, which were connected with copper wires.

Computational method

The Vienna Ab Initio Simulation Package (VASP) code described by the projector augmented wave (PAW) method for ion-electron interaction was applied to the simulation calculations57,58. The generalized gradient approximation (GGA) expressed in the form of the Perdew-Burke-Ernzerhof (PBE) function was used to deal with exchange-correlation interactions59. The cutoff energy of 500 eV was taken into account by all calculations, and the Monkhorst-Pack k-point grid was set to 3 × 3 × 3 for bulk structure optimization, 5 × 5 × 1 for BiVO4(001)/NiFeOx and BiVO4(001)/N:NiFeOx heterostructures. The empirical correction scheme of Grimme (DFT+D2) was adopted for considering van der Waals (vdW) interaction60. The convergence criterion for Hellmann-Feynman forces and total energy were set to 0.01 eV/Å and 10−5 eV, and the vacuum space in the z-direction was greater than 20 Å to avoid the interaction between adjacent units during structural relaxation. A twelve atomic layers BiVO4 (001) slab model was used, and the bottom six atomic layers were fixed to simulate the bulk structure.