INTRODUCTION
Bismuth vanadate (BiVO₄) is an n-type semiconductor with a bandgap of 2.4 eV that absorbs visible light (approximately 520 nm). It has attracted attention as a promising photoelectrode material in various applications such as photoelectrochemical (PEC) water splitting, photocatalysis, and solar energy conversion devices due to its low cost, chemical stability, and non-toxicity [1–4]. In particular, monoclinic scheelite phase BiVO₄ has been the most actively researched in experimental and theoretical studies due to its electronic structural stability and excellent light absorption properties compared to other crystalline phases [2,4]. However, the photocurrent efficiency of BiVO₄ is significantly limited due to the short electron-hole diffusion distance (<70 nm), low charge mobility (μ ≈ 10⁻2 cm2 V⁻1 s⁻1), and serious charge recombination problems at the surface and bulk [3–5].
Various approaches have been proposed to overcome this performance degradation. These include shortening charge transport pathways through nanostructuring or thin film fabrication, enhancing charge transfer rates by introducing surface catalysts, and forming heterojunctions with other semiconductors [2-6] to induce photogenerated charge separation. Among these methods, one of the simplest and most effective is doping with transition metals [7–14]. Transition metal doping involves substituting transition metal ions with similar ionic radii and electronic structures into the V5⁺ or Bi3⁺ sites within the BiVO4 crystal lattice. This can have the following effects: (i) increasing charge concentration, (ii) modulating the electronic structure of the conduction/valence bands, and (iii) suppressing charge recombination [5,7–10,14,15]. For example, when Mo6⁺ or W6⁺ ions are subs tuted into the V5⁺ site, a donor state is formed, increasing electron concentration and improving charge mobility.
However, the doping effect is highly sensitive to concentration; excessive doping can introduce defects such as lattice distortion and oxygen vacancies, which in turn act as charge recombination centers and degrade performance [16]. Therefore, to maximize performance, it is crucial to precisely control not only the type of dopant but also its optimal concentration. The traditional research approach to finding the optimal doping concentration, however, has inherent limitations. Preparing and characterizing individual samples for each discrete concentration requires significant time, labor, and material resources. Moreover, this one-at-a-time method makes it difficult to systematically understand the complex interplay between doping levels and physical properties like defect formation, lattice strain, and charge recombination dynamics. The inefficiency of this approach acts as a major bottleneck, slowing the pace of new material discovery and necessitating a more innovative strategy.
To overcome these hurdles, this study introduces a combinatorial materials science approach coupled with a high-throughput screening (HTS) system [17,18]. This strategy enables the efficient exploration of a vast compositional space by fabricating a thin-film library with a continuous compositional gradient on a single substrate and rapidly mapping its physical properties with an automated system.
In this study, we employed the HTS system based on scanning photoelectrochemical microscopy (SPECM) to elucidate the composition-structure-property relationships in Mo-doped BiVO4 photoanodes. SPECM is ideally suited for evaluating compositional libraries, as it allows for the localized investigation of the photocurrent response by scanning a laser in the visible-light range (405 nm). This study ultimately aimed to fabricate the BiVO4 compositional library with varying Mo doping concentrations via combinatorial sputtering and to identify the optimal Mo doping concentration rapidly and systematically using the SPECM-based screening methodology. Through this approach, we sought to resolve inconsistencies in previously reported Mo-doping effects and, more broadly, to provide a generalized research framework that can accelerate the development of high-performance photoanodes for solar fuel applications.
EXPERIMENTAL
Fabrication of the Mo-doped BiVO4 Combinatorial Thin-Film Library
BiVO4 thin-film libraries with a Mo concentration gradient were fabricated using a 4-target RF/DC magnetron sputtering system (ULVAC). Fluorine-doped tin oxide (FTO) coated glass substrates (7 Ω/sq, Pilkington, 50 mm × 180 mm × 1 mm) were employed. Two sputter targets were used simultaneously: a BiVO4 ceramic target (4-inch diameter, 4 mm thickness, ≥99.9% purity, Cu backing plate with Cu bonding) and a Mo-doped BiVO4 target containing 10 at.% Mo (4-inch diameter, 4 mm thickness, ≥99.9% purity, Cu backing plate with Cu bonding). High-purity Ar (99.999%) and O₂ (99.999%) gases were supplied as process gases.
Prior to deposition, substrates were ultrasonically cleaned sequentially in acetone (5 min), isopropanol (5 min), and deionized water (three rinses), followed by N2 blow-drying. A base pressure of 1.5 × 10⁻6 Torr was achieved before deposition. During sputtering, the total working pressure was maintained with Ar (30 sccm) and O2 (6 sccm) flow rates. The substrate was heated to 450°C, and the target-to-substrate distance was fixed at 150 mm. All BiVO₄ thin films were deposited at elevated substrate temperatures without any subsequent annealing step. The temperatures of 300°C and 450°C mentioned in this study correspond to the substrate temperatures maintained during the deposition process.
RF powers of 200 W and 100 W were applied to the pure BiVO4 and Mo-doped BiVO4 targets, respectively. To create a Mo doping gradient, the two targets were mounted at opposite ends of the substrate with an offset configuration, and the substrate was positioned such that the x-axis corresponded to increasing Mo flux. This offset arrangement allowed the Mo flux to vary continuously across the substrate, forming a compositional gradient. Fig. 1(a) illustrates the combinatorial sputtering process used to fabricate a Mo concentration gradient in BiVO4 thin films. A pure BiVO4 target and a Mo-doped BiVO4 target were co-sputtered under Ar plasma onto an FTO-coated glass substrate (SLG/FTO) through a shadow mask. Due to the offset geometry of the two targets, a continuous compositional gradient was formed along the substrate.
As shown in Fig. 1(b), the deposited film library was divided into 11 regions (Samples 1–11) along the gradient direction. The left side corresponds to Mo-rich compositions, whereas the right side corresponds to Mo-poor compositions. This gradient library enabled simultaneous evaluation of photoelectrochemical properties as a function of Mo doping concentration [19,20].
The substrate was fixed at a rotation angle of 135° during deposition. The rotation angle of 135° refers to the angular configuration between targets No. 2 and No. 4 in the four-target combinatorial sputtering system, which was optimized to achieve uniform co-deposition of BiVO4 films. Prior to deposition, each target was pre-sputtered for 5 min with the shutter closed, followed by plasma stabilization for an additional 5 min. The total deposition time was 90 min.
A 1 M NaOH aqueous solution was prepared by dissolving 40.0 g of sodium hydroxide (NaOH, Sigma-Aldrich, reagent grade, ≥98%, pellets) in deionized water to a final volume of 1 L. The solution was stirred until complete dissolution and stored in a polyethylene container to minimize CO2 absorption. For high-throughput screening (HTS) photoelectrochemical (PEC) measurements, 100 mL of the 1 M NaOH solution was used as the electrolyte. The solution was freshly prepared prior to each measurement to ensure reproducibility and stability.
High-Throughput Screening (HTS) Measurements
The resulting thin-film library was characterized using a custom-built high-throughput photoelectrochemical (PEC) mapping setup, as shown in Fig. 2(a-b). A 405 nm laser (140 mW) was delivered to the sample through an optical fiber and focused onto the thin-film surface to a spot size of approximately 0.2 mm. The laser beam irradiated the sample from above, and the photocurrent response was recorded using a three-electrode configuration consisting of a working electrode, a counter electrode, and a reference electrode containing 3.0 M KCl reference electrode +0.624 V vs RHE at 25°C. The thin-film library (sample size: 180 mm × 50 mm) was placed in a quartz cell filled with 100 mL of 1 M NaOH electrolyte. The XY stage was controlled by an XYZ scanning unit, enabling automated movement of the laser probe over a scan area of 172 mm × 40 mm with a step size of 2 mm in both X and Y directions. During all measurements, the working electrode potential was fixed at 0.7 V vs. the reference electrode, and photocurrent mapping was performed to generate spatially resolved PEC performance data. During the photoelectrochemical (PEC) mapping measurements, the FTO substrate was immersed in the electrolyte to ensure stable electrical contact with the working electrode.
Additional Characterizations
The optical transmittance was obtained using a UV-Vis spectrophotometer with a scan rate of 600 nm/min, a data interval of 1 nm, an averaging time of 0.1 s, and a spectral bandwidth of 2 nm. The crystal structure was characterized by X-ray diffraction (XRD) at 40 kV with a scan rate of 10°/min and a current of 50 mA over the 2θ range of 20–80°. The surface and cross-sectional morphologies were observed by scanning electron microscopy (SEM) at an accelerating voltage of 15 kV. The elemental composition and Mo distribution were determined by energy-dispersive X-ray spectroscopy (EDX) using triple normalization treatment, and all elemental analyses were performed using standard reference materials for O, V, Mo, and Bi.
RESULTS AND DISCUSSION
Surface SEM images (Fig. 3) revealed notable morphological evolution as a function of Mo content. The Mo-rich sample (Sample 1) exhibited densely packed fine grains with reduced grain boundaries, while the intermediate Mo-doped samples (Samples 3–9) exhibited irregular grains with visible voids and boundary disorder. Conversely, the Mo-poor sample (Sample 11) presented large, well-faceted grains with smooth surfaces and minimal boundary defects. These results suggest that excessive Mo incorporation suppresses grain growth and introduces structural disorder, whereas lower Mo levels promote grain coarsening and the development of high-quality crystalline domains.
Elemental analysis using EDX (Table 1) confirmed the formation of a controlled compositional gradient across the library. The Mo atomic percentage systematically decreased from 14.41 at.% in the Mo-rich region (Sample 1) to 0.96 at.% in the Mo-poor region (Sample 11) (Fig. 4(a-b)). Bi and V contents remained relatively stable across all samples, while oxygen was consistently retained as a major component. This indicates that Mo incorporation did not induce significant cation deficiency or elemental loss. Importantly, the correlation between Mo concentration and XRD results suggests that the intermediate Mo compositions, where FWHM broadening was most pronounced, contained sufficient Mo to induce lattice distortion and defect formation. In contrast, the Mo-poor sample (Sample 11) combined low Mo content with improved crystallinity, suggesting a defect-minimized lattice favorable for charge transport.
The structural characteristics of the Mo-doped BiVO4 (BVO) thin films were examined by XRD (Fig. 5a–c). At 300°C annealing (Fig. 5a), the films exhibited only weak and broad features without distinct diffraction peaks, indicating that crystallization was incomplete. In contrast, films annealed at 450°C (Fig. 5b) displayed sharp and well-defined peaks corresponding to the monoclinic scheelite-type BiVO4 phase, confirming that high-temperature annealing is essential for phase formation and improved crystallinity.
The XRD patterns of the Mo-doped BiVO4 films (Fig. 5b) reveal a systematic evolution of the crystal structure with increasing Mo content. The undoped BiVO4 exhibits characteristic diffraction peaks corresponding to the monoclinic scheelite (ms) phase, while the Mo-doped samples display gradual peak shifts and intensity variations indicative of a transition toward the tetragonal scheelite (tz) structure. This ms–tz structural evolution suggests that Mo incorporation influences the lattice symmetry, likely through substitutional doping at V sites, which modifies local bonding environments and lattice strain.
Furthermore, quantitative analysis of the full width at half maximum (FWHM) and grain size (Fig. 5c) revealed a strong correlation between Mo content and crystallinity. Intermediate Mo-doped samples (Samples 5–7) exhibited significant peak broadening and reduced grain size, consistent with lattice distortion and increased microstrain. In contrast, the Mo-rich (Sample 1) and Mo-poor (Sample 11) samples showed narrower peaks and larger grain sizes, suggesting improved crystallinity.
Films annealed at 450°C exhibited significantly improved crystallinity compared to those annealed at 300°C, as evidenced by sharp diffraction peaks and larger grain sizes. In particular, the Mo-poor region (Sample 11, 0.96 at.% Mo) showed the narrowest FWHM values and the largest grain size, indicating a defect-minimized lattice structure.
This structural advantage translates directly into superior PEC activity. The high photocurrent observed in the Mo-poor samples can be attributed to enhanced crystallinity and enlarged grains, which reduce the density of grain boundaries and suppress charge recombination. In contrast, intermediate Mo doping levels (Samples 5–7) exhibited peak broadening and reduced grain size, consistent with lattice distortion and microstrain, which act as recombination centers. As a result, these samples displayed significantly lower photocurrents in PEC mapping.
The improvement in PEC performance observed in the Mo-poor region can be largely attributed to reduced inter-grain resistivity, resulting from enhanced crystallinity and improved grain connectivity. At the same time, Mo incorporation within the BiVO₄ lattice contributes to the electronic (n-type) doping effect by increasing the electron concentration and shifting the Fermi level upward. Therefore, both effects—enhanced charge transport across grain boundaries and increased carrier density within grains—synergistically contribute to the overall PEC enhancement in the optimally doped BiVO4 films.
UV–Vis transmittance spectra (Fig. 6a) revealed a clear absorption onset near 500 nm for all Mo-doped BiVO4 films, consistent with the expected visible-light absorption of BiVO4. Compared with bare FTO, the films exhibited significantly enhanced absorption in the visible region, confirming successful light-harvesting capability.
Tauc plot analysis (Fig. 6b) showed that the optical bandgap values varied slightly depending on Mo concentration, ranging from approximately 2.55 eV to 2.65 eV across the compositional library. The intermediate Mo-doped samples (Samples 5–7) exhibited slightly larger bandgaps (2.62–2.65 eV), while the Mo-poor (Sample 11) and Mo-rich (Samples 1–3) samples showed narrower bandgaps (2.55–2.58 eV).
Importantly, in the low Mo-doping range (1–2 at.%), the bandgap remained close to that of pristine BiVO₄, preserving effective visible-light absorption while minimizing the risk of excessive defect formation. These results imply that low-level Mo doping maintains favorable optical properties, whereas both very high and intermediate doping levels can perturb the band structure and influence charge carrier dynamics.
The PEC performance of the thin-film library was assessed by high-throughput photocurrent mapping (Fig. 7). The 2D photocurrent map (Fig. 7a) and their line profile (Fig. 7b) revealed a gradual increase in photocurrent intensity along the Mo gradient, from the Mo-rich (Sample 1) to the Mo-poor (Sample 11) side. The maximum photocurrent (0.14–0.15 mA) was observed in Sample 11 (0.96 at.% Mo), located at the Modeficient end of the substrate (Fig. 7a). These findings clearly demonstrate that the highest PEC activity was achieved not at high Mo concentrations but at low doping levels (1–2 at.%).
The observed PEC trend can be rationalized by integrating compositional, structural, and morphological analyses. First, the EDX results (Table 1) confirmed that photocurrent enhancement correlated directly with reduced Mo content.
The enhanced PEC performance of Mo-doped BiVO4 originates from a synergistic interplay between donor-induced electronic improvement and crystallinity enhancement. Upon substitution of Mo6⁺ for V5⁺, shallow donor levels are introduced near the conduction band, increasing carrier density and elevating the Fermi level, which facilitates charge separation and electrical conductivity. Concurrently, moderate Mo incorporation promotes recrystallization and grain growth, reducing grain boundary density and suppressing charge recombination. However, excessive Mo doping leads to lattice distortion and defect formation, which offset these beneficial effects [21,22].
In contrast, low-level Mo incorporation (1–2 at.%) provides sufficient donor density to improve conductivity while maintaining structural integrity, thereby optimizing charge separation and transport. Importantly, UV–Vis analysis confirmed that the bandgap remained stable at these low doping levels, ensuring sustained light absorption.
In summary, the high-throughput PEC mapping identified Sample 11 (0.96 at.% Mo) as the best-performing composition, highlighting that optimal PEC activity in Mo-doped BiVO₄ thin films is achieved at low Mo doping levels (1–2 at.%). This conclusion is strongly supported by complementary evidence from the following: (i) EDX compositional analysis, (ii) XRD peak profile analysis, (iii) SEM morphological observations, and (iv) optical bandgap evaluation.
The systematic analysis of Mo-doped BiVO4 thin films provides a comprehensive understanding of how compositional, structural, and morphological factors collectively determine PEC activity. EDX confirmed the presence of a well-controlled Mo gradient, ranging from 14.41 at.% (Sample 1) to 0.96 at.% (Sample 11). While Mo acts as a donor and can enhance carrier density, excessive incorporation was found to deteriorate structural quality, as evidenced by broadened XRD peaks and irregular grain morphology. In particular, the intermediate Mo-doped samples (Samples 5–7) exhibited the highest degree of lattice distortion and microstrain, consistent with suppressed crystallinity and increased defect formation.
In contrast, the Mo-poor sample (Sample 11, 0.96 at.%) demonstrated superior crystallinity, large grains with smooth surfaces, and minimal boundary defects, all of which are favorable for efficient charge separation and transport. Despite its low doping level, the donor density provided by Mo was sufficient to improve electrical conductivity without introducing significant recombination centers. Consequently, the best PEC performance was obtained at 1–2 at.% Mo, confirming that a delicate balance between donor introduction and defect minimization is critical for optimizing BiVO₄ photoanodes.
These results highlight a key design principle: controlled, low-level Mo doping maximizes PEC efficiency by simultaneously enhancing conductivity and preserving crystallinity, whereas excessive doping compromises performance through defect-mediated recombination.
Correlation between Structural Properties and PEC Performance. The photoelectrochemical (PEC) performance trends can be directly correlated with the structural characteristics revealed by XRD analysis.
As shown in Fig. 5b, the XRD patterns reveal a clear structural evolution from the monoclinic scheelite (ms) to the tetragonal scheelite (tz) phase with increasing Mo concentration. The pristine BiVO4 film exhibits characteristic peaks of the ms phase, whereas the Mo-doped samples show systematic shifts of the diffraction peaks toward higher 2θ angles and variations in their relative intensity ratios. In particular, the ratio between the (121) and (040) reflections gradually decreases with increasing Mo content, suggesting a progressive transition toward the more symmetric tz structure.
To further quantify this structural evolution, the relative lattice variation (Δd/d) was calculated from the peak positions at 33°, 35°, and 39°, as presented in Fig. 5c. Negative Δd/d values observed for several samples— especially at the 35° and 39° peaks—indicate peak shifts to higher angles corresponding to lattice contraction. This lattice contraction confirms that Mo incorporation into the BiVO4 lattice relaxes the inherent distortion of the ms structure and enhances crystallographic symmetry. Such structural modification reduces grain-boundary scattering and improves charge transport, contributing to the enhanced photoelectrochemical (PEC) performance discussed in Fig. 7.
Fig. 8 presents the correlation between photocurrent response, Mo content, and optical bandgap across the Mo-doped BiVO4 thin-film library annealed at 450°C. The photocurrent profile (black line) shows a gradual enhancement from the Mo-rich to the Mo-poor region, reaching a maximum photocurrent (0.14–0.15 mA) at the lowest Mo concentration (1 at.%). Photocurrent measurements were conducted using a 3.0 M KCl reference electrode +0.624 V vs RHE at 25°C. This trend indicates that low Mo incorporation is optimal for achieving efficient charge separation and transport.
The Mo content decreases continuously along the scanning direction, confirming the successful formation of a compositional gradient by combinatorial sputtering.
As shown in Fig. 6(b), the optical bandgap of Mo-doped BiVO4 films varies slightly in the range of 2.55–2.70 eV without a clear monotonic trend with increasing Mo concentration. This small fluctuation is within the experimental uncertainty and likely reflects minor structural or compositional differences rather than a systematic bandgap narrowing [23].
CONCLUSIONS
This study systematically elucidated the impact of Mo doping on the structural, optical, and photoelectrochemical properties of BiVO4 thin films prepared at 450°C. By employing a compositional-gradient library combined with high-throughput mapping, we directly correlated Mo concentration with crystallinity, grain morphology, bandgap, and PEC activity. The results revealed that excessive Mo incorporation leads to lattice distortion, grain refinement, and increased defect density, which promote carrier recombination and degrade PEC performance. In contrast, low Mo doping (1–2 at.%) effectively enhances electrical conductivity while maintaining favorable crystallinity and visible-light absorption, resulting in superior photocurrent.
The highest PEC activity (0.14–0.15 mA) was obtained from the Mo-poor region (Sample 11, 0.96 at.% Mo), establishing that optimal PEC performance is achieved at low Mo concentrations rather than in Mo-rich regimes. These findings highlight the critical role of precise dopant control in BiVO4 photoanodes and demonstrate the effectiveness of high-throughput approaches in accelerating the discovery of optimal compositions for solar fuel applications.
PDF Links
PubReader
ePub Link
Full text via DOI
Download Citation
