Heating-rate controlled defect engineering in BiVO₄ thin films for enhanced photoelectrochemical water splitting

Abstract

We investigate the effect of nucleation kinetics, governed by ramping rate of annealing temperature, on the structural, chemical, and optical properties of BiVO₄ thin films and their subsequent impact on photoelectrochemical (PEC) performance, with the goal of enhancing the photocatalytic activity of BiVO₄-based photoanodes. Here, we demonstrate that controlled thermal nucleation, achieved by tuning the annealing heating rate, is an effective and scalable strategy to engineer oxygen vacancies (O_v) and modulate defect chemistry in BiVO₄ thin films. Structural and spectroscopic analyses reveal that an optimized heating rate promotes the partial reduction of V⁵⁺ to V⁴⁺, generating O_v donor states that narrow the optical band gap from 2.55 to 2.33 eV and increase carrier density. First-principles DFT calculations demonstrate that these oxygen vacancies introduce V⁴⁺ derived shallow donor states, providing theoretical validation for the experimentally observed bandgap reduction and enhanced charge transport. Atomic force microscopy (AFM) and scanning electron morphology (SEM) reveals an evolution in surface morphology with increased surface roughness and porosity, which promote electron transport and bulk conductivity. These synergistic effects result in a remarkable photocurrent density of 3.17 mA cm⁻² at 1.23 V vs. the reversible hydrogen electrode (RHE) and excellent stability for ∼4.5 hours in aqueous electrolyte. Overall, this study highlights controlled thermal nucleation as a robust strategy to tune microstructure, defect chemistry, and charge transport, paving the way for next-generation high-efficiency PEC materials.

---

1 Introduction

Photoelectrochemical (PEC) cells, which utilize photoactive semiconductors and sunlight to produce electricity and hydrogen through water splitting, represent a promising alternative to natural photosystems.^1–4 This process stores solar energy in the H–H bond, the simplest chemical bond,^5,6 and was first demonstrated using nanosized titania (TiO₂).⁷ Various semiconductors have been developed for water electrolysis,^8–10 including III-V semiconductors (e.g., GaInP₂,¹¹ AlGaAs,¹² and InP¹³) and II-VI semiconductors (e.g., ZnO,¹⁴ CdS,¹⁵ and ZnS¹⁶), which exhibit good photoactivity with solar-to-hydrogen (STH) conversion efficiencies of up to 20%. However, these materials face limitations due to their instability in aqueous electrolytes and high costs.^17–19 In contrast, metal oxides are more durable and cost-effective, but their reported STH conversion efficiencies are typically below 8%,^20–22 limited by rapid charge carrier recombination and poor charge transport.^10,23

Monoclinic scheelite BiVO₄ has emerged as a promising metal oxide photoelectrode due to its favorable band structure (E_g ∼2.3–2.6 eV), suitable band-edge position (conduction band at ∼0 V vs. RHE), and intrinsic electrical properties.^24–26 It is one of the three polymorphs of BiVO₄, i.e. pucherite (orthorhombic), dreyerite (tetragonal zircon), and clinobisvanite (monoclinic scheelite).²⁷ Its crystallographic and electronic properties show good photocatalytic activity.^28,29 However, BiVO₄ suffers from significant charge carrier recombination under illumination due to its low charge carrier mobility (∼0.043 cm² V⁻¹ s⁻¹).³⁰ The highest reported photocurrent density for pristine BiVO₄ is around 3.6 mA cm⁻²,³¹ achieved using laser ablation methods that are not scalable due to decreasing ablation efficiency over time, high energy requirements, and expensive production systems. Therefore, developing facile protocols to enhance the charge carrier transport in BiVO₄ photoanodes remains a significant challenge. Various approaches have been explored, including doping,^32,33 heterojunction formation,^34,35 plasmonics,^36,37 and nanostructuring.³⁸ Additionally, optimizing experimental parameters such as precursor concentration,^39,40 ultrasonication treatment,^41,42 thermal treatment,^43,44 and annealing conditions^45,46 has been shown to significantly impact the morphology and performance of these materials for photo-driven applications.

This work demonstrates that the heating rate during annealing is an effective parameter for controlling nucleation that governs defect formation, and the resulting microstructure in BiVO₄. By tuning oxygen vacancy concentration through this simple thermal adjustment, we provide a scalable and industry-compatible approach for improving charge transport and PEC performance. Oxygen vacancies introduce distinct electronic states by redistributing charges in BiVO₄, reducing the work function and bandgap, thereby enabling the thin films to absorb lower-energy light.^47–49 Optimizing the thermal treatment during the deposition of BiVO₄ thin films is a crucial approach to tuning the nucleation process, which defines the crystallographic properties of the final product.⁵⁰ This thermal treatment is decisive in tailoring the crystal structure and generating oxygen vacancies that enable effective charge migration from the bulk material to the surface, thereby enhancing light absorption efficiency.^47,50 Practically, the effective activation energy for nucleation and crystallization can be tuned by varying the heating rate during the annealing process.^51,52 The impact of the heating rate on the crystallization of BiVO₄ thin films was investigated using X-ray diffraction (XRD) and atomic force microscopy (AFM). The results show that the crystallinity and grain size of the films are strongly influenced by the heating rate.^53,54 Optimizing the annealing protocol through a series of experiments resulted in a remarkable five-fold increase in photocurrent density, reaching 3.17 mA cm⁻² at 1.23 V vs. RHE. X-ray photoelectron spectroscopy (XPS), UV-visible spectroscopy, and AFM analyses confirmed that tuning the annealing parameters altered the interatomic bonding, lowered the work function, and refined the film morphology to enhance the photoactivity of BiVO₄, providing a facile and effective alternative to address the limitations of charge generation and transport in BiVO₄ photoelectrodes.

2 Experimental details

2.1. Materials

All chemicals were purchased from Sigma Aldrich and used in their purest form without further treatment. The BiVO₄ ink was prepared using Bi(NO₃)₃·5H₂O (≥99.99%), ethylene glycol (≥99.0%), glacial acetic acid (≥99.7%)) deionized water, VO(acac)₂ (≥98%), and polyethylene glycol (≥99.0%). The electrolyte solution consisted of deionized water, Na₂SO₄ (≥99.0%), and Na₂SO₃ (≥98%).

2.2. Preparation of BiVO₄ ink

The synthesis of BiVO₄ thin films was performed in air. 0.6 g of Bi(NO₃)₃·5H₂O was added to 4.0 g (3.604 mL) of ethylene glycol (EG) and sonicated for 10 minutes to achieve transparency. Subsequently, 2.0 mL of glacial acetic acid and 1.0 mL of deionized water were added and sonicated for an additional 10 minutes. Adding 0.6 g of VO(acac)₂ and sonicating for 30 minutes transformed the transparent solution into a dark turquoise color. Finally, 0.472 g of polyethylene glycol (PEG) was added to control the morphology of the deposited BiVO₄ thin films. The resulting viscous solution, stabilized by the triblock copolymer Pluronic F-108, was ready for spin coating.

2.3. Fabrication of BiVO₄ film

FTO-coated glass substrates were cleaned ultrasonically in acetone, ethanol, and deionized water for 20 minutes each, followed by drying at 60 °C. The BiVO₄ films were deposited by spin coating 0.5 mL of the BiVO₄ ink onto the FTO-coated glass substrates at 1200 rpm for 10 seconds and then 1500 rpm for 35 seconds. The films were immediately transferred to a furnace for annealing at 475 °C for 1.5 hours with different heating rates (10 °C, 20 °C, 30 °C, 40 °C, 50 °C, 60 °C, and 70 °C per minute), labeled as BV-10 °C, BV-20 °C, BV-30 °C, BV-40 °C, BV-50 °C, BV-60 °C, and BV-70 °C, respectively. The samples were allowed to cool naturally to room temperature. The entire process took approximately 12 hours, producing films with a thickness of 0.2 microns. A single spin-coating step was sufficient to prepare all seven samples with different heating rates. The experimental protocol is illustrated in Fig. 1, and digital micrographs of the prepared samples are provided in the SI (Fig. S1).

Fig. 1 The schematic illustration of BiVO₄ thin films deposition protocol by spin coating method.

2.4. Characterization techniques

X-ray powder diffraction (XRD; Bruker D8 ADVANCE Plus) was performed to obtain the crystal structures of the prepared samples using Cu-Kα radiation (wavelength 1.5 Å) over a range of 10°–80° at a scan rate of 5° min⁻¹. X-ray photoelectron spectroscopy (XPS; ThermoFisher ESCALAB™ 250XIb) equipped with Al Kα radiation was conducted to explore the binding energy of all elements. UV-vis spectroscopy (Cary 50 conc UV-visible spectrophotometer) was performed in the range of 300–700 nm. AFM (Bruker diMultimode V®AFM) data were obtained using Nanoscope software and an AppNano ACTA tip (ROC = 6 nm) in tapping mode.

2.5. PEC cell assembly and measurements

The PEC performance of the BiVO₄ thin films was tested using a standard three-electrode cell system in a deionized water-based electrolyte (0.1 M Na₂SO₄ + 0.1 M Na₂SO₃; pH ∼9.26). The prepared BiVO₄ thin films were cut into smaller pieces with an effective area of 0.7 cm², and the edges were sealed with epoxy glue to prevent current leakage. Before use, the electrolyte was purged with nitrogen gas for 30 minutes to remove any residual dissolved oxygen. The BiVO₄ photoanode served as the working electrode, while a platinum wire and an Ag/AgCl electrode were used as the counter and reference electrodes, respectively. The applied potential was corrected using the Nernst equation:

E_RHE = E_SCE + 0.0591 pH + 0.209 (1)

where E_RHE is the reversible hydrogen electrode potential, and 0.209 is the conversion factor from Ag/AgCl to RHE. All PEC data were recorded using a CH Instrument electrochemical workstation (model CHI6011E). A 300W Xenon lamp (NewBet HSX-F300) provided simulated 1 sun (AM 1.5 G) illumination with a calibrated power density of 100 mW cm⁻². Transient photocurrent density was measured under chopped illumination at 1.23 V vs. RHE.⁵³ Linear sweep voltammetry (LSV) was conducted at a scan rate of 10 mV s⁻¹, and electrochemical impedance spectroscopy (EIS) was performed in the frequency range of 1 Hz to 10 kHz. The flat-band potential was estimated from Mott–Schottky (MS) plots using the capacitance data of the BiVO₄ photoanodes recorded at different biases. A sinusoidal modulation of 5 mV at 5 Hz was applied to obtain the MS plots:where C is the capacitance of photoanode, e is the electronic charge (C), ε is the dielectric constant of BiVO₄, ε^o is the permittivity of free space, V is the applied bias (versus RHE), V_f is the flat band potential (versus RHE), k is the Boltzmann constant, N_d is the donor density for n-type semiconductor (cm⁻³), A is the surface area of photoanode, and T is the temperature (K).

2.6. Computational methods

All density functional theory (DFT) calculations were performed using the Vienna Ab initio Simulation Package (VASP).^55,56 The electron–ion interactions were described using the Projector Augmented-Wave (PAW) method with the Perdew–Burke–Ernzerhof (PBE) exchange-correlation functional.^57,58 To accurately model the electronic structure of BiVO₄, the DFT + U method was employed with a Hubbard U parameter of 3.0 eV applied to the Bi 3d states. The pristine monoclinic structure was first fully relaxed to its minimum energy configuration. A 2 × 1 × 2 supercell containing 96 atoms was then constructed from this optimized cell to model bulk properties. A neutral oxygen vacancy was created by removing a single oxygen atom, after which the defective supercell was also fully relaxed. The Brillouin zone for all supercell calculations was sampled with a Γ-centered 2 × 3 × 2 k-point mesh.⁵⁹

3 Results and discussion

3.1. Crystallographic analysis

Fig. 2 shows the XRD patterns of BiVO₄ thin films prepared under different heating rates. The obtained diffraction peaks are in good correspondence with the monoclinic scheelite phase of BiVO₄ (JCPDS card number 14-0688), which is optimal for photoelectrochemical water splitting applications.⁵³ Some additional peaks were attributed to the FTO substrates, and no secondary oxide phases were detected in any of the films.

Fig. 2 XRD patterns of BiVO₄ thin films annealed at different heating rates (BV-10 °C to BV-70 °C). All samples match the monoclinic scheelite phase (JCPDS 14-0688), with increasing peak sharpness and intensity at higher heating rates, indicative of improved crystallinity. The peaks from the FTO substrate are marked with (*).

All films subjected to various thermal treatments exhibited comparable patterns, indicating no phase transformation. Further analysis revealed that BV-60 °C exhibited the highest crystallinity, with sharper diffraction peaks compared to other samples. The sharpness of the peaks is directly proportional to the crystallite size.⁶⁰ The controlled annealing process played an important role in tuning the crystallinity and consequently the electronic properties of the BiVO₄ thin films. Higher crystallinity reflects improved charge transport, which is essential for enhancing the performance of photoactive materials.^61,62 The magnified plotting of XRD peaks for better comparison is provided in the SI (Fig. S2), and the planes associated with each peak are listed in SI Table S3.

3.2. X-ray photoelectron spectroscopy (XPS)

The surface chemical status of the samples was studied using XPS. Fig. 3a shows the survey spectra of BiVO₄, confirming the presence of C, Bi, V, and O elements in all samples. The Bi 4f_7/2 and Bi 4f_5/2 peaks (Fig. 3b) for BV-20 °C, BV-40 °C, BV-50 °C, and BV-70 °C were observed at 159.4 eV and 164.7 eV, respectively, indicating the Bi³⁺ oxidation state. The peak shift is negligible among these samples, except for BV-10 °C and BV-30 °C, which exhibited a slight negative shift towards lower binding energies. The sample BV-60 °C showed the lowest binding energies (158.9 eV and 164.2 eV), suggesting a change in the local coordination environment due to the presence of oxygen vacancies and V⁴⁺ species.

Fig. 3 (a) XPS high resolution elemental survey plots confirming the presence of Bi 4f, V 2p, and O 1s bands, (b) the splitting signals of Bi 4f to Bi 4f_7/2 and Bi 4f_5/2, (c) the O 1s peaks observed for as deposited BiVO₄ samples with an obvious peak observed at 531.5 eV for BV-60 °C referring to the presence of oxygen vacancies in this sample, and (d) the splitting signals of V 2p to V 2p_3/2 and V 2p_1/2.

The V 2p_3/2 and V 2p_1/2 peaks observed at 516.8 eV and 524.6 eV (Fig. 3d) are characteristic of V⁵⁺ species.^63–67 BV-60 °C exhibited a prominent low-energy shift in both the Bi 4f and V 2p binding energies compared to other samples. This shift can be attributed to changes in the local coordination environments of Bi and V ions, specifically a reduction in coordination number due to oxygen vacancies, which likely leads to the partial reduction of Bi³⁺ and V⁵⁺ ions.⁶⁷

XPS analysis of metal oxides revealed two types of oxygen oxidation states: the peaks formed in the range of 529.5–530.5 eV are the splitting peaks of the O 1s band, characteristic of the O²⁻ oxidation state. The peaks formed in the region between 531–532 eV suggest the existence of another O 1s species with lower electron density than O²⁻, assignable to O⁻ ions.^68,69 The major peaks for BV-10 °C, BV-30 °C, BV-40 °C, and BV-70 °C were observed at 530.07 eV and 530.90 eV in the O 1s spectrum (Fig. 3c), assigned to O bonded within a regular oxide crystal (O²⁻) in BiVO₄ and adsorbed hydroxyl groups on the surface, respectively. BV-20 °C and BV-50 °C exhibited lower binding energy values without changes in oxidation states. The presence of oxygen vacancies in BV-60 °C is evident from the shoulder peak at 531.5 eV (Fig. 3c, inset), indicating the incorporation of V⁴⁺ species in the BiVO₄ lattice during the reaction between Bi(NO₃)₃ and VO(acac)₂. The partial reduction of V⁵⁺ to V⁴⁺ ions leads to the generation of abundant oxygen vacancies as shallow donors of electrons.^68,70,71 The higher concentration of V⁴⁺ species and oxygen vacancies in BV-60 °C can be attributed to its porous network structure, which provides a larger surface area for defect formation, thereby enhancing the photocatalytic performance of BiVO₄.⁷² The lower binding energy shifts in other samples for Bi and V splitting signals indicate lower crystallinity compared to BV-60 °C, consistent with the XRD results.

Collectively, the systematic shifts in the Bi 4f, V 2p, and O 1s spectra across the heating-rate series establish a clear comparative trend in defect density. The defect-related O 1s shoulder at ∼531.0 ± 0.2 eV is observed in all samples. By normalizing each spectrum to the lattice oxygen maximum (∼529.6–530.0 eV) and inspecting a fixed binding-energy window (530.8–531.6 eV), we find that the vacancy-related intensity is most pronounced for BV-60 °C, followed by BV-50 °C ≳ BV-40 °C > BV-30 °C ≈ BV-70 °C > BV-20 °C ≈ BV-10 °C. A similar trend is evident in the V 2p_3/2 spectra, where the relative contribution of the V⁴⁺ shoulder (∼516.0–516.5 eV) compared to the V⁵⁺ main peak (∼516.8 eV) is again maximized for BV-60 °C. This concurrent increase in vacancy-related oxygen and V⁴⁺ species provides consistent evidence that BV-60 °C possesses the highest oxygen-vacancy concentration. Importantly, this sample also exhibits the strongest photoelectrochemical response, including the highest photocurrent density and charge-separation efficiency, confirming the functional role of these vacancies. We emphasize that this comparison is semi-quantitative; however, by applying identical normalization and fixed energy windows across the series, the observed trend is robust and reliable.

3.3. UV-visible spectroscopy

Investigating the optical properties is crucial for understanding the PEC performance. Fig. 4(a and b) shows the UV-vis absorption spectra of the BiVO₄ thin films annealed at different heating rates, along with bandgap evaluations using the Tauc relation (eqn (3)):

αhv = C (hv − E_g)^n/2 (3)

where α is the absorption coefficient, h is the Plank's constant, v is the frequency of incident light, C is a constant which is the function of band telling parameter, and E_g is the bandgap energy (eV). For BiVO₄, n = 1 since it is an indirect bandgap semiconductor.⁷³ The visible light absorption in monoclinic scheelite BiVO₄ occurs through transitions from the valence band (composed of Bi 6s or Bi 6s + O 2p hybrid orbitals) to the V 3d conduction band.⁷⁴ All samples exhibited strong absorption in the visible light region, with the most intense absorption observed in BV-60 °C. The bandgap reduction from 2.55 eV to 2.33 eV in BV-60 °C is attributed to the introduction of shallow bands within the bandgap of BiVO₄ by oxygen vacancies.⁷⁵ These electronic defects significantly reduce the bandgap energy, promoting the photoactivity of the material, as illustrated in Fig. 4c.

Fig. 4 (a) UV-vis absorption spectra of BiVO₄ thin films prepared at different heating rates, showing improved visible-light absorption for BV-60 °C. (b) Tauc plots for bandgap estimation, with the lowest E_g (∼2.33 eV) observed for BV-60 °C. (c) Schematic representation of shallow donor levels introduced by oxygen vacancies, which facilitate charge transport. (d) AFM images revealing morphology evolution from compact grains (BV-10-40 °C) to well-defined nanocones at BV-60 °C, before collapsing into larger irregular features at BV-70 °C. The rougher, porous morphology of BV-60 °C provides enhanced surface area and charge-separation pathways, consistent with its improved PEC activity.

The optical response further clarifies the role of defect states in charge dynamics. A gradual narrowing of the band gap from ∼2.55 eV (BV-20 °C) to ∼2.33 eV (BV-60 °C), together with the emergence of sub-bandgap absorption tails, can be attributed to oxygen-vacancy/V⁴⁺ defect states just below the conduction band. These defect states effectively extend light absorption and facilitate electron transfer. Complementary PL measurements (S4) reveal that samples with the highest PL intensity, such as BV-40 °C, suffer from pronounced radiative recombination, which limits charge carrier utilization and results in lower PEC activity. In contrast, BV-60 °C exhibits strongly quenched PL, indicating suppressed recombination and better separation of photogenerated carriers. This behaviour arises from favourable band-edge modulation by optimally introduced oxygen vacancies as the defect states below the conduction band promote electron delocalization while maintaining sufficient band bending at the electrode/electrolyte interface to drive water oxidation. The same sample (BV-60 °C) delivers the highest photocurrent density, confirming that reduced PL intensity directly correlates with improved charge separation and PEC performance. Excessive vacancy formation at higher annealing rates (BV-70 °C) partially restores PL intensity, reflecting recombination via deep defect centres, and explains the corresponding decline in PEC activity.

3.4. Atomic force microscopy (AFM)

AFM images (5 × 5 µm) of the BiVO₄ thin films reveal a morphology transition from densely stacked short nanorods to long, sharp nanocones with a globular particle matrix for BV-60 °C. The average arithmetic roughness (R_a), root mean square roughness (R_q), and average maximum height (R_mp) were estimated from AFM analysis and are summarized in Table 1. The three-dimensional topographic images of the as-deposited BiVO₄ thin films are shown in Fig. 4d. BV-60 °C exhibited the highest roughness values and the most pronounced nanocones. Compared to other samples (Table 1), the vertically oriented nanocone array provides a greater contact area between the electrolyte and the photoanode, as well as a larger particle size. Since conductivity is directly proportional to the cross-sectional area of the particles, the larger grain size ensures improved bulk conductivity. Additionally, the vertically oriented porous morphology shortens the average travel distance for charge carriers before they participate in generating photocurrent. The optimal film thickness is limited by the charge carrier diffusion length.

Table 1 The average arithmetic roughness of the surface (R_a), root mean square vales of the roughness (R_q), and average maximum height (R_mp) of as prepared BiVO₄ thin films obtained from atomic force microscopy

Roughness parameter	Samples names
	BV-10 °C	BV-20 °C	BV-30 °C	BV-40 °C	BV-50 °C	BV-60 °C	BV-70 °C
Ra (nm)	15.8	15.7	17.6	21.6	17.7	29.0	20.2
Rq (nm)	19.9	19.6	22.1	27.7	22.7	37.3	25.7
Avg max height (nm)	9.21	7.45	11.4	14.3	13.4	24.0	9.2

---

3.5. Scanning electron microscopy (SEM)

SEM analysis (Fig. 5) reveals that the surface morphology of BiVO₄ strongly depends on the heating rate, evolving from compact grains at low rates (BV-10 °C) to a porous, cone-like network at BV-60 °C, before collapsing into overgrown coarse features at higher rates (BV-70 °C).

Fig. 5 SEM images of BiVO₄ thin films annealed at different heating rates, showing a transition from compact grains (BV-10 °C) to porous, cone-like features at BV-60 °C, followed by overgrown structures at BV-70 °C. The porous nanocone morphology of BV-60 °C provides higher surface area and correlates with AFM roughness and XPS evidence of oxygen vacancies, supporting its improved PEC performance.

This hierarchical and porous microstructure at BV-60 °C provides an enlarged surface area and facilitates electrolyte infiltration, consistent with AFM measurements that showed the highest surface roughness and nanocone formation for the same sample. XPS analysis further supports this correlation, as BV-60 °C exhibited the most pronounced oxygen-vacancy-related O 1s shoulder and V⁴⁺ contribution, indicating that the optimized morphology not only promotes surface contact but also provides a favourable environment for defect generation. Quantitative analysis of SEM images revealed average grain sizes of ∼295 nm, 295 nm, 275 nm, 245 nm, 282 nm, 339 nm, and 280 nm for BV-10 °C to BV-70 °C, respectively, with BV-60 °C exhibiting the largest features. This trend supports the morphological transition observed, where optimized nanocone growth at BV-60 °C provides enhanced porosity and surface area, consistent with AFM roughness and XPS vacancy signatures. Together, these results demonstrate that controlled thermal nucleation simultaneously tunes surface architecture and defect chemistry, leading to improved PEC activity.

3.6. PEC performance

3.6.1. I–t curves. The impact of the heating rate on the PEC water splitting performance of BiVO₄ films was evaluated using I–t curves (Fig. 6a) and J–V curves (Fig. 6b) with a conventional three-electrode cell system at a scan rate of 10 mV s⁻¹ under AM 1.5 G illumination (100 mW cm⁻²). Fig. 6a shows the chopped transient current density under SE (substrate to electrolyte) illumination at 1.23 V vs. RHE for water splitting. The EE illumination output current density is provided in the SI (Figure S4) and is comparatively lower for all samples due to the increased charge transfer distance for photoelectrons in this configuration. Upon irradiation, the electrodes exhibit a large photocurrent spike due to the presence of numerous photogenerated electron–hole pairs in the bulk. This spike decreases as fast charge carrier recombination balances the charge generation rate. The minimal dark current for all BiVO₄ films compared to their photocurrents indicates efficient photoresponse. BV-60 °C demonstrated a remarkable five-fold increase in photocurrent density, reaching 3.17 mA cm⁻² at 1.23 V vs. RHE compared to the initial sample (BV-10 °C). The improved PEC performance is attributed to the morphology defects of the deposited photoanodes, as confirmed by AFM and XPS analyses.

Fig. 6 (a) Chopped transient photocurrent responses of BiVO₄ thin films (BV-10 °C to BV-70 °C) measured under SE (substrate-to-electrolyte) illumination at 1.23 V vs. RHE, showing variation in current density with heating rate (highest for BV-60 °C at 3.17 mA cm⁻²). (b) Linear sweep voltammetry (J–V) curves of the samples, where differences in onset potential and photocurrent trends reflect the influence of heating rate on charge separation and PEC activity.

Furthermore, photoluminescence (PL) spectroscopy (Fig. S5, SI note 1) confirmed increased charge carrier mobility and reduced recombination due to the optimized nucleation process through controlled heating rate protocols. The 4.5-hours stability test (Fig. 7) revealed outstanding durability for BV-60 °C in aqueous electrolyte, with only a modest 2% drop in photocurrent density. While BV-10 °C lost 50% of its initial photocurrent density within one hour stability testing. The AFM micrographs justified this expectation, and a detailed discussion on this phenomenon is furnished in the next portion. These results tell that the PEC performance of BiVO₄ can be better tuned by optimizing the heating rate during the annealing process of thin films. This strategy predominantly accelerated electron transport and boosted film conductivity. The larger slope of the linear sweep voltammetry (LSV) curves of BV-60 °C substantiates this increase in performance so well as shown in Fig. 6b. To confirm that the enhanced photocurrents were not an artifact of sulfite oxidation, additional PEC measurements were performed in neutral phosphate buffer (KPi, pH 7.0) without a sacrificial agent. As shown in Fig. S8, the same activity trend was maintained, validating that the improvements arise from intrinsic film properties.

Fig. 7 Stability test of BV-60 °C at 1.23 V vs. RHE under continuous illumination for 4.5 h, showing ∼2% photocurrent loss. The inset compares BV-10 °C, which exhibits ∼50% photocurrent loss within 1 h.

3.6.2. Mott–Schottky plots. Mott–Schottky (MS) measurements were performed to determine the charge carrier density at the BiVO₄ electrode–electrolyte interface. The MS curves (Fig. 8a) were obtained from the space charge capacitance values recorded under dark conditions at 1000 Hz. The positive slopes of all curves confirm that BiVO₄ is an n-type semiconductor.⁷⁶ The slopes indicate that BV-60 °C has a higher charge carrier density compared to other samples. A positive shift in the flat band potential (E_fb) for BV-60 °C suggests increased band edge bending, which improves charge separation and aligns with other studies. The Mott–Schottky analysis assumes an ideal depletion region and perfectly ohmic contact between the semiconductor and the electrolyte. While metal oxide semiconductors like BiVO₄ may deviate from these assumptions due to interface states or contact resistances, the extracted donor densities here serve as relative indicators because all films share the same thickness, substrate, and electrochemical configuration.⁷⁷ Trends from the Mott–Schottky slopes are corroborated by complementary measurements: the systematic increase of V⁴⁺ ions and oxygen vacancies in XPS, the narrowing of the optical bandgap in UV-vis spectra, suppressed radiative recombination in PL, and reduced charge-transfer resistance in impedance spectra. Collectively, these results confirm that the BV-60 °C sample exhibits the highest effective carrier density, validating the reliability of the comparative analysis.

Fig. 8 (a) Mott–Schottky plots of BiVO₄ thin films prepared at different heating rates, measured in 0.1 M Na₂SO₄ + 0.1 M Na₂SO₃ electrolyte (pH ∼9.26) with 5 mV AC amplitude at 1 kHz. The extracted flat-band potentials (V_fb) and slopes (m) are indicated for each sample, showing clear variation with heating rate, and (b) Nyquist plots of BiVO₄ thin films under substrate-to-electrolyte illumination, showing differences in charge-transfer resistance (R_ct). BV-60 °C displays the smallest semicircle, consistent with faster interfacial charge transfer.

3.6.3. Electrochemical impedance spectroscopy (EIS). Electrochemical impedance spectroscopy was conducted to characterize the charge mobility at the BiVO₄ electrode electrolyte interface. In the Nyquist plots (Fig. 8b), the smaller the diameter of the semicircle, the better the charge carrier transport.⁷⁸ The sequence of decreasing semicircle diameters under dark conditions and illumination is BV-20 °C > BV-10 °C > BV-70 °C > BV-40 °C > BV-30 °C > BV-50 °C > BV-60 °C, indicating a significant decrease in charge carrier transfer resistance. The BV-60 °C photoanode effectively suppressed electron–hole pair recombination and exhibited efficient charge mobility, while the BV-70 °C photoanode showed a larger R_ct value than BV-30 °C, BV-40 °C, and BV-50 °C, resulting in higher charge transfer resistance. Overly rapid heating rates during annealing led to poor charge transfer at the electrode–electrolyte interface.

The onset potential of a photoanode is a critical parameter as it reflects the ease of initiating photoelectrochemical water oxidation and can be correlated with the flat band potential obtained from Mott–Schottky analysis. From the LSV curves, the onset potential of BV-60 °C and BV-70 °C is shifted towards more negative values compared to the other samples, indicating that these films require lower external bias to drive charge separation. This trend agrees well with the Mott–Schottky slopes, where BV-60 °C and BV-70 °C exhibit the lowest slopes and therefore the highest donor densities. The higher carrier density facilitates band bending and accelerates charge separation at the semiconductor/electrolyte interface, thereby lowering the onset potential. In contrast, BV-10 °C and BV-20 °C display the highest slopes, corresponding to the lowest donor densities and consequently a delayed onset potential. Although BV-70 °C shows the highest donor density and earliest onset potential, its photocurrent remains lower than BV-60 due to excessive defect states and higher interfacial resistance that promote charge recombination. Hence, BV-60 °C achieves the optimal balance between an advantageous onset potential and suppressed recombination, pronouncing its superior PEC performance. The proposed band alignment and defect-state positions (Fig. 9) are deduced from experimental observations (XPS, UV-vis, PL, and Mott–Schottky analyses).

Fig. 9 Rapid Nucleation-induced oxygen vacancy formation in BiVO₄: bandgap narrowing and suppressed recombination via shallow donor states; an illustration deduced from XPS, UV-vis, PL, and Mott–Schottky analyses.

3.6.4. Incident photon-to-current efficiency (IPCE). The IPCE spectra of the BiVO₄ thin films (Fig. 10) were measured at 1.23 V vs. RHE under AM 1.5 G illumination in neutral phosphate buffer. BV-60 °C displays the highest IPCE, reaching ∼35% at ∼420 nm, nearly twice that of BV-20 °C, which shows the weakest response. The relative IPCE intensities mirror the photocurrent trends obtained from LSV, indicating that the enhanced performance of BV-60 °C originates from improved charge separation and reduced recombination associated with oxygen-vacancy-mediated defect states. As the electrolyte does not contain sacrificial agents, the IPCE values directly reflect the intrinsic water oxidation efficiency of the BiVO₄ electrodes, confirming that defect engineering via controlled thermal nucleation is an effective strategy for boosting photoanode performance.

Fig. 10 IPCE spectra of BiVO₄ thin films (BV-10 °C to BV-70 °C) showing the highest efficiency for BV-60 °C, consistent with its superior charge separation and PEC performance.

3.7. Theoretical insights into oxygen vacancy-mediated electronic modulation in BiVO₄

To investigate the electronic origins of the enhanced photoactivity, DFT calculations were performed on pristine and oxygen deficient BiVO₄ supercells. Fig. 11a and b show the calculated band structures. The pristine system exhibits a direct bandgap of 2.6 eV, consistent with our experimental optical gap of 2.55 eV. Upon introduction of an oxygen vacancy, a defect band appears that reduces the calculated gap to 2.2 eV, corresponding to the redshift observed experimentally for BV-60 °C. The projected density of states (PDOS) for the pristine and oxygen deficient systems are shown in Fig. 11c and d, respectively. In the oxygen deficient system, a new peak composed of V 3d orbitals emerges approximately 0.4 eV below the conduction band minimum, representing V⁴⁺ derived shallow donor states.⁷⁹ This feature is consistent with the V⁴⁺ species identified in our XPS analysis.

Fig. 11 (a and b) Electronic band structures of pristine and oxygen-deficient BiVO₄, showing bandgap narrowing from 2.6 to 2.2 eV due to vacancy-induced defect states. (c and d) Total and projected density of states (PDOS) illustrating atomic contributions and the emergence of V-derived donor states in oxygen-deficient BiVO₄. (e and f) Planar-averaged electrostatic potential profiles used for work-function calculations, yielding Φ = 5.64 eV (primitive) and 5.32 eV (defective).

These vacancy-induced states enable lower energy electronic transitions,⁷⁹ accounting for the enhanced visible light absorption in BV-60 °C. Previous computational work has shown that concurrent optimization of band structure and oxygen vacancy concentration is critical for maximizing photoelectrochemical performance.⁸⁰ The calculated band structures (Fig. 11a and b) also reveal modified conduction band dispersion in the defective system, indicating improved electron transport. This finding aligns with the increased carrier density extracted from our Mott Schottky analysis and the reduced charge transfer resistance observed in EIS for BV-60 °C.

Work functions were calculated from planar averaged electrostatic potential profiles (Fig. 11e and f). The pristine slab yields a work function of 5.64 eV, while the oxygen deficient slab exhibits a reduced value of 5.32 eV (ΔΦ = 0.32 eV). This reduction results from the upward shift of the Fermi energy into oxygen vacancy donor states, consistent with the enhanced n type character observed in our Mott Schottky analysis.

Beyond band structure modification, the asymmetric charge distribution induced by oxygen vacancies disrupts the local symmetry of the BiVO₄ lattice, generating a local dipole moment.⁸¹ This dipole creates an internal electric field that promotes electron hole separation, consistent with the quenched photoluminescence observed in BV-60 °C (Fig. S5). Collectively, these DFT calculations confirm that oxygen vacancies introduce shallow donor states that narrow the bandgap, increase carrier density, lower the work function, and enhance charge separation, contributing to the superior PEC performance of BV 60 °C.

3.8. Mechanism of defect-mediated charge transport mechanism in BiVO₄ photoanodes induced by controlled thermal nucleation

During annealing procedure, the formation of O_v takes place in BiVO₄ lattice through a controlled heating rate protocol that encourage slight lattice oxygen loss and the concomitant partial reduction of V⁵⁺ to V⁴⁺. These oxygen vacancies act as shallow donor sites, introducing electronic states just below the conduction band (as depicted in the Fig. 9). These O_v-related defect levels lie right under the conduction band edge, which effectively dopes the BiVO₄ lattice with n-type defects. Density functional theory calculations corroborate this mechanism by revealing V⁴⁺-derived shallow donor states located near the conduction band minimum, which induce bandgap narrowing and an upward Fermi-level shift. The presence of these donor levels narrows the effective band gap from ∼2.55 eV in pristine BiVO₄ to ∼2.33 eV in O_v-rich BiVO₄ by enabling lower-energy electronic transitions. As a result of this reduction in band gap, visible light is absorbed into longer wavelengths, resulting in better photon utilization. Moreover, oxygen vacancies increase the electron carrier density (since each O_v donates two electrons), which improves electrical conductivity and facilitates charge separation. The shallow donor states trap photogenerated electrons away from the valence band, lowering the likelihood of electron–hole recombination and thus suppressing radiative recombination pathways.

Experimentally, X-ray photoelectron spectroscopy (XPS) provides evidence of this mechanism: O_v-rich BiVO₄ exhibits a shoulder in the O 1s XPS peak (around ∼531.5 eV) and an increased V⁴⁺ signal, directly indicating oxygen vacancies and the reduction of V⁵⁺ to V⁴⁺. In O_v-modified BiVO₄, the absorption edge marks a significant redshift, consistent with a narrower band gap. Likewise, photoluminescence (PL) measurements reveal a quenched emission intensity for O_v-engineered BiVO₄, signifying reduced charge-carrier recombination. These experimental observations are in agreement with the theoretical predictions of defect-induced donor levels and enhanced n-type character, establishing a direct correlation between electronic-structure modulation and photoelectrochemical performance. The synergic effects of reduced band gap increased visible-light absorption, and improved charge separation with lower recombination align with the depiction of pristine versus O_v-modified BiVO₄ in the Fig. 8. The O_v-rich BiVO₄ features defect-derived donor levels below the conduction band that boost its photoactivity, whereas the pristine BiVO₄ retains a wider band gap and suffers from more recombination, underscoring the beneficial role of oxygen vacancy engineering in BiVO₄.

4 Conclusion

BiVO₄ films were deposited by spin coating and annealed at 475 °C with varying heating rates to tune nucleation process. XRD analysis revealed that all samples exhibited a monoclinic scheelite structure. The porosity in the morphology of BV-60 °C increased the effective surface area and improved electrolyte contact, facilitating interfacial charge transfer. This sample also possessed a high concentration of oxygen vacancies and V⁴⁺ ions in its lattice, which enhanced photon absorption, charge generation, and charge mobility. These morphological and electronic defects reduced the bandgap energy from 2.55 eV to 2.33 eV and resulted in a five-fold increase in photocurrent density to 3.17 mA cm⁻² for pristine BiVO₄.

DFT calculations further demonstrated that these oxygen vacancies introduce V⁴⁺-derived shallow donor states, which contribute to bandgap narrowing and modify conduction band dispersion to facilitate charge transport. At an annealing rate of 60 °C min⁻¹, the concentration of oxygen vacancies appears sufficient to balance improved carrier mobility with reduced recombination losses without introducing excessive structural defects that may act as recombination centres, as observed at higher ramping rates (e.g., 70 °C min⁻¹). Thus, the optimized defect chemistry induced by thermal treatment tailors the band structure to improve light harvesting, charge separation, and interfacial charge transfer, collectively enhancing PEC water splitting performance.

The 4.5-hours stability test demonstrated that tuning the thermal conditions during BiVO₄ thin film deposition produces an efficiently photoactive material capable of withstanding aqueous electrolyte for extended periods, preventing short circuiting. This synthesis approach may also be extended to explore the defect-mediated intrinsic properties of other functional metal oxides, advancing their potential for photocatalytic applications.

Author contributions

Rafiqat Ul Rasool: conceptualization, methodology, investigation, data curation, formal analysis, visualization, writing-original draft, review & editing. Syed Bilal Ahmed: formal analysis, theoretical calculations, writing-review & editing. Ruiqin Zhang: supervision, resources, project administration, funding acquisition, writing-review & editing.

Conflicts of interest

There are no conflicts to declare.

Data availability

All data supporting the findings of this study are available within the article and its supplementary information (SI) file. The data will be available from the corresponding author upon reasonable request. Supplementary information: comparative analysis of reported BiVO₄ systems, additional experimental details, structural and spectroscopic characterization, and supporting photoelectrochemical performance data of BiVO₄ thin films. See DOI: https://doi.org/10.1039/d5ta03464g.

Acknowledgements

This work was supported by the Research Grants Council of the Hong Kong SAR (No. 11317122) and the City University of Hong Kong (No. 9231264, 7020044, 9229019, 7020096).

References

1. H. Wu, L. Zhang, A. Du, R. Irani, R. van de Krol, F. F. Abdi and Y. H. Ng, Nat. Commun., 2022, 13, 6231.
2. A. Vilanova, P. Dias, T. Lopes and A. Mendes, Chem. Soc. Rev., 2024, 53, 2388–2434.
3. D. J. Martin, P. J. T. Reardon, S. J. A. Moniz and J. Tang, J. Am. Chem. Soc., 2014, 136, 12568–12571.
4. S. Wei, X. Xia, S. Bi, S. Hu, X. Wu, H.-Y. Hsu, X. Zou, K. Huang, D. W. Zhang, Q. Sun, A. J. Bard, E. T. Yu and L. Ji, Chem. Soc. Rev., 2024, 53, 6860–6916.
5. M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi, E. A. Santori and N. S. Lewis, Chem. Rev., 2010, 110, 6446–6473.
6. A. Currao, Chimia, 2007, 61, 815–819.
7. A. Fujishima and K. Honda, Nature, 1972, 238, 37–38.
8. S.-C. Wang, F.-Q. Tang and L.-Z. Wang, J. Inorg. Mater., 2018, 33, 173–197,  DOI:10.15541/jim20170352.
9. N. Serpone and A. Emeline, J. Phys. Chem. Lett., 2012, 3, 673–677.
10. J. Wang, M. Zhou, H. Yu, Z. Zhang and Y. Sun, J. Colloid Interface Sci., 2025, 646, 35–46.
11. S. Hwang, M. Yang, C. Kim, J. Lee and J. Kim, Sustain. Energy Fuels, 2020, 4, 1437–1442.
12. J. D. Butson, J. Tournet, B. Gupta, A. Sharma, M. Lysevych, T. Haggren, C. Jagadish, H. H. Tan and S. Karuturi, ACS Appl. Mater. Interfaces, 2024, 16, 45180–45188.
13. Y. Ramprakash, D. Bose and S. Basu, Surf. Sci., 1984, 145, 175–184.
14. K.-S. Ahn, Y. Yan, S. H. Wei and J. A. Turner, Appl. Phys. Lett., 2008, 93, 163117.
15. H. S. Mansur, M. A. Orefice and A. C. Mansur, J. Chem. Soc., Faraday Trans., 1995, 91, 665–672,  10.1039/FT9959100665.
16. J. Zhang, X. Wu, Y. Liu, H. Chen and S. Li, J. Mater. Chem. A, 2015, 3, 535–541.
17. S. Chen, T. Takata and K. Domen, Nat. Rev. Mater., 2017, 2, 17050.
18. J. Jia, M. Seitz, J. Lichterman, N. Liu and H. Wang, Nat. Commun., 2016, 7, 1–6.
19. P. Guo, F. Hao, L. Chen, W. Hu and Y. Fang, J. Mater. Chem. A, 2019, 7, 2497–2506.
20. J. H. Kim, D. H. Youn, J. Y. Kim, S. K. Lim and J. S. Lee, Nat. Commun., 2016, 7, 1–9.
21. X. Shi, H. Wang, C. Chen, J. Liu and T. Zhang, Nat. Commun., 2016, 7, 1–6.
22. B. Y. Liu, J. Wang, X. Lin, Y. Zhao and H. Zhang, Angew. Chem., Int. Ed., 2023, 62, e202217346.
23. F. Wang, M. Li, Y. Xu, S. Tang and X. Wang, J. Am. Chem. Soc., 2017, 139, 15094–15103.
24. A. Walsh, J. L. F. Da Silva and S. H. Wei, Chem. Mater., 2009, 21, 547–551.
25. K. Sivula, F. Le Formal and M. Grätzel, ChemSusChem, 2011, 4, 432–449.
26. F. F. Abdi, L. Han, A. H. M. Smets, M. Zeman, B. Dam and R. van de Krol, J. Phys. Chem. Lett., 2013, 4, 2752–2757.
27. A. Bhattacharya, K. Mallick and A. Hartridge, Mater. Lett., 1997, 30, 7–13.
28. Y. Zhao, C. Zhang, H. Xu, L. Wang and G. Chen, Chem.–Eur. J., 2008, 14, 1601–1606.
29. J. K. Cooper, S. Gul, F. M. Toma, I. D. Sharp, L. Y. Lin and J. W. Ager, J. Phys. Chem. C, 2015, 119, 2969–2974.
30. F. Ye, M. Chen, J. Zhang, L. Wei and S. Huang, Appl. Catal., B, 2018, 227, 258–265.
31. H. S. Han, S. Shin, D. Jeong, Y. J. Jang and J. S. Lee, Energy Environ. Sci., 2018, 11, 1299–1306.
32. F. F. Abdi, L. Han, A. H. M. Smets, M. Zeman, B. Dam and R. van de Krol, Nat. Commun., 2013, 4, 1–7.
33. K. P. S. Parmar, H. J. Kang, J. S. Lee and J. H. Kim, ChemSusChem, 2012, 5, 1926–1934.
34. J.-S. Yang and J.-J. Wu, Nano Energy, 2017, 32, 232–240.
35. L. Zhou, M. Xu, C. Wang, W. Li and X. Zhang, Nano Lett., 2016, 16, 3463–3474.
36. L. Zhang, L. O. Herrmann and J. J. Baumberg, Sci. Rep., 2015, 5, 1–12.
37. W. Yang, Q. Li, Z. Li, T. Wang and X. Wang, Nanoscale Res. Lett., 2016, 11, 1–8.
38. Y. Wei, Y. Zhang, X. Liu and H. Li, Nano Res., 2016, 9, 1561–1569.
39. J. S. Nyarige, T. P. Krüger and M. Diale, Mater. Today Commun., 2020, 25, 101459.
40. H. Baqiah, M. Ali, N. Z. Noor and A. Salleh, Optik, 2020, 206, 164303.
41. S. Zhang, J. Wang, M. Li, Y. Chen and X. Liu, Appl. Catal., B, 2020, 262, 118279.
42. Z. Masoumi, M. Tayebi and B.-K. Lee, Ultrason. Sonochem., 2021, 72, 105403.
43. B. Wickman, A. Bastos da Silva Fanta, A. Burrows, A. Hellman, J. B. Wagner and B. Iandolo, Sci. Rep., 2017, 7, 40500.
44. L. Shi, S. Zhuo, M. Abulikemu, G. Mettela, T. Palaniselvam, S. Rasul, B. Tang, B. Yan, N. B. Saleh and P. Wang, RSC Adv., 2018, 8, 29179–29188,  10.1039/C8RA04887H.
45. V. D. Das, J. Sathyanarayanan and L. Damodare, Surf. Coat. Technol., 1997, 94, 669–671.
46. F. F. Abdi and R. van de Krol, J. Phys. Chem. C, 2012, 116, 9398–9404.
47. H. Fu, Q. Qi, Y. Li, J. Pan and C. Zhong, Nanomaterials, 2024, 14, 1270.
48. B. D. Khadka, S. Sagadevan, S. Kato and T. Soga, Phys. B: Condens. Matter, 2024, 682, 415878.
49. R. Chen, M. Li, S. Hu and J. Wang, ChemElectroChem, 2024, 11, e202400600.
50. G. C. Tu, J. Y. Chen, Y. Li, C. W. Chang, W. J. Chang, Z. X. Zhen, H. M. Chen and C. M. Jiang, ACS Appl. Electron. Mater., 2024, 6, 1872–1885.
51. A. Choi, M. Jung and Y. K. Lee, Electrochem. Solid-State Lett., 2009, 12, 7–10.
52. J. R. Frade, J. Am. Ceram. Soc., 2005, 81, 2654–2660.
53. M. Lamers, S. Fiechter, D. Friedrich, F. F. Abdi and R. van de Krol, J. Mater. Chem. A, 2018, 6, 18694–18700,  10.1039/C8TA06269B.
54. Q. Zhang, S. Jiang and Y. Li, Surf. Rev. Lett., 2007, 14, 141–145.
55. G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1993, 47, 558.
56. G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 49, 14251.
57. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865.
58. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953.
59. H. J. Monkhorst and J. D. Pack, Phys. Rev. B: Condens. Matter Mater. Phys., 1976, 13, 5188.
60. S. A. A. Terohid, S. Heidari, A. Jafari and S. Asgary, Appl. Phys. A, 2018, 124, 1–9.
61. D. Jena and A. Konar, Phys. Rev. Lett., 2007, 98, 136805.
62. Y. S. Son, Y. Kim, J. Lee, G. Y. Lee, W. T. Park, Y. Y. Noh, C. E. Park and T. Park, J. Am. Chem. Soc., 2016, 138, 8096–8103.
63. K. Zhang, L. Wang, T. Liu and X. Li, Nat. Commun., 2019, 10, 1–10.
64. W. Luo, M. Li, Q. Chen and R. Zhang, Energy Environ. Sci., 2011, 4, 4046–4051.
65. L. Zhang, D. Chen and X. Jiao, J. Phys. Chem. B, 2006, 110, 2668–2673.
66. M. F. R. Samsudin, R. Bashiri, N. M. Mohamed, Y. H. Ng and S. Sufian, Appl. Surf. Sci., 2020, 504, 144417.
67. F. S. Hegner, D. Forrer, J. R. Galán-Mascarós, N. López and A. Selloni, J. Phys. Chem. Lett., 2019, 10, 6672–6678,  DOI:10.1021/acs.jpclett.9b02552.
68. M. Huang, J. Bian, W. Xiong, C. Huang and R. Zhang, J. Mater. Chem. A, 2018, 6, 3602–3609,  10.1039/C7TA11132K.
69. F. Li, H. Wu, Z. Liu and Y. Yang, Phys. Chem. Chem. Phys., 2017, 19, 21862–21868.
70. G. Tan, Y. Zhang, L. Zhang and Z. Zhang, Ceram. Int., 2014, 40, 9541–9547.
71. M. Wang, Q. Li, J. Zhou and L. Zhang, J. Alloys Compd., 2015, 648, 1109–1115.
72. C. Q. Sun, Prog. Solid State Chem., 2006, 35, 1–159.
73. Z. Zhao, X. Li, H. Liu and Y. Wang, Solid State Sci., 2013, 18, 98–104.
74. J. Yu and A. Kudo, Chem. Lett., 2005, 34, 850–851.
75. J.-M. Wu, Y.-H. Lin and H.-S. Tsai, Appl. Catal., B, 2018, 221, 187–195.
76. A. Hankin, M. Bedoya-Lora, J. C. Alexander, A. Regoutz and G. H. Kelsall, J. Mater. Chem. A, 2019, 7, 26162–26176.
77. A. Venugopal, R. Kas, K. Hau and W. A. Smith, J. Am. Chem. Soc., 2021, 143, 18581–18591,  DOI:10.1021/jacs.1c08245.
78. Y. Bicer and I. Dincer, Chem. Eng. Process. Process Intensif, 2017, 117, 141–153,  DOI:10.1016/j.cep.2017.04.001.
79. O. El Ouardi, H. Ladib, B. Vigolo, J. Alami and M. Makha, Int. J. Hydrogen Energy, 2025, 157, 150384,  DOI:10.1016/j.ijhydene.2025.150384.
80. P. An, Y. Zhou, H. Xiu, Y. Wang and Y. Wang, J. Colloid Interface Sci., 2026, 140231.
81. J. Zhang, Q. Jia and M. Li, Dalton Trans., 2026, 55, 3,  10.1039/D5DT02584B.

---