Interface engineering for light-driven water oxidation: unravelling the passivating and catalytic mechanism in BiVO₄ overlayers

Abstract

Artificial photosynthetic approaches require the combination of light absorbers interfaced with overlayers that enhance charge transport and collection to perform catalytic reactions. Despite numerous efforts that have coupled various catalysts to light absorbing semiconductors, the optimization of semiconductor/catalyst as well as catalyst/electrolyte interfaces and the identification of the role of the catalyst still remain a key challenge. Herein, we assemble (NiFeCoCe)O_x multi-component overlayers, interfaced with bismuth vanadate photoanodes, and determine the roles of different elements on promoting interfacial charge transfer and catalytic reaction over competitive photocarrier recombination loss processes. Through this understanding, and aided by complementary macroscopic photoelectrochemical measurements and nanoscale atomic force microscopy techniques, a bifunctional (CoFeCe/NiFe)O_x overlayer was rationally engineered. The resulting multi-functional coating yields BiVO₄ photoanodes with almost 100% efficient surface collection of holes under oxygen evolving reaction conditions. The (CoFeCe)O_x component excels at efficient capture and transport of photogenerated holes in BiVO₄ through the availability of redox active states, whereas (NiFe)O_x plays a vital role in reducing charge recombination at the BiVO₄/electrolyte interface. In addition, this study supports the hypothesis that catalytic sites act as electronically active trap states on uncoated BiVO₄ photoanodes.

---

1. Introduction

The adverse effect of intensive fossil fuel usage on our environment has motivated the search for clean and renewable energy technologies. However, for many of these technologies, energy storage still remains a limiting factor.¹ By converting the large but intermittent solar energy flux into chemical fuels, artificial photosynthesis holds promise to overcome this challenge and provide a sustained energy supply for our societal needs.^2,3 Photoelectrochemical (PEC) water splitting and CO₂ reduction represent intriguing routes to produce clean fuels using sunlight.^4–7 As the electrons and protons needed for solar fuel production hinge on the oxidation of water, an efficient photoanode is a prerequisite for a viable solar fuel technology.

Among promising semiconductor photoanode materials, bismuth vanadate (BiVO₄) has emerged as one of the most investigated systems in unassisted water-splitting devices, as it captures a substantial portion of the visible spectrum and possesses band energetics that are well-suited for the oxygen evolution reaction (OER).⁸ However, the bare surface of this material suffers from poor catalytic activity for the OER and possesses high concentrations of electronically active surface states, leading to significant charge recombination losses^9,10 and corrosion.¹¹ Numerous studies have demonstrated that BiVO₄ needs to be paired with catalysts to achieve suitable photocurrents for the OER.^12–19 However, different working mechanisms have been reported for these catalysts, depending on the physicochemical nature of the BiVO₄/catalyst interface. For example, Durrant, et al. concluded that the origin of the enhanced photocurrent resulting from a cobalt phosphate (CoPi) catalyst treatment is primarily due to improved charge transfer across the BiVO₄/electrolyte interface rather than the catalytic function of CoPi.²⁰ This argument was strengthened by a recent report of van de Krol, et al., which highlights that CoPi passivates the surface states of BiVO₄ without significantly influencing the charge transfer kinetics.²¹ On the other hand, Gamelin and co-workers came to a different conclusion that CoPi promoted charge transfer for the OER, comparing with H₂O₂ as a hole scavenger.¹³ In addition, Li, et al., argued that a cobalt borate (CoBi) catalyst accelerates interfacial charge transfer at the BiVO₄/electrolyte interface.¹⁴ Choi, et al., demonstrated that a FeOOH overlayer creates a favorable interface with BiVO₄ as well as acts as an OER catalyst.¹⁵

The different interpretations of the improvement mechanisms for the proposed overlayers highlight the variety of demands placed on the catalyst coating, as well as the challenge in discerning the prominent function(s) due to the confluence of these phenomena during photoelectrochemical reactions. Specifically, a good catalyst layer deposited on a light absorber needs to out-compete the surface states in capturing holes in order to be efficient for OER catalysis. The rate of these competing kinetic pathways determines whether the passivation mechanism will dominate over the catalytic one or vice versa. Disentangling the respective passivation and catalytic roles of light absorber overlayers is currently needed to provide the necessary insight for further development of highly efficient and stable integrated photoanodes. In order to address this issue, multi-component overlayers with complimentary hole transport and catalytic functions are practically required. While a few examples of multi-component catalytic layers have been reported to improve the efficiency of BiVO₄,^16,17 the charge transfer and collection mechanism in these overlayers still remains unclear.

Here, via an individual component analysis, we rationally design (NiFeCoCe)O_x multi-component coatings on BiVO₄ photoanodes to investigate how the specific properties of the light absorber/catalyst interface define the interfacial charge transfer and collection mechanism, as well as to simultaneously optimize the passivating and catalytic functions. Based on our high-throughput direct discovery of BiVO₄/multi-component functional interfaces,^18,19 we deposit (NiFeCoCe)O_x multi-component overlayers by photoelectrochemical deposition (PED). PED is a straightforward method that is well suited for producing diverse compositions of coatings and creating interfacial structures with BiVO₄.²² In addition, PED inherently places catalysts at specific locations on the BiVO₄/electrolyte interface where the photogenerated holes are most readily available without coating the underlying FTO.¹⁵ However, to the best of our knowledge, attempts to deposit controlled multi-element compositions by this versatile technique are rare.²²

To clarify the role of the different components and enable the rational design of a targeted multi-component overlayer, we studied the individual (Ni, Fe, Co, Ce) oxides in multi-component (NiFe, CoFeCe, and CoFeCe/NiFe) oxides. Thus, we demonstrate that an optimized light absorber/overlayer interface necessitates a dual hole transport/catalyst coating. We combined a suite of photoelectrochemical (PEC) and microscopy techniques to characterize the nanoscale distribution and functionality of the catalyst on the photoanode surface. Thus, we correlate this information with macroscopic PEC measurements to provide a deeper understanding of charge transfer at the light absorber/overlayer interface. Analysis of the role of the individual overlayer components shows that while CeO_x is excellent at passivating active surface states, it reduces the BiVO₄ catalytic activity under OER conditions. Specifically, the CeO_x overlayer produces discontinuous films that together with the photoelectrochemical findings support that surface trap states in BiVO₄ may be responsible for OER catalysis. In addition, using PEC characterization we identify (NiFe)O_x as the most active surface catalyst, and (CoFeCe)O_x as an interfacial charge transfer and collection layer in (NiFeCoCe)O_x multi-component coatings. Furthermore, we investigate the interface energetics and dynamics of multi-component BiVO₄/(CoFeCe)O_x using open circuit potential (OCP) and transient photocurrent (TPC) measurements. These measurements demonstrate that (CoFeCe)O_x is able to extract and collect holes from BiVO₄ by introducing surface redox states for the OER. Based on the understanding of (NiFeCoCe)O_x, we optimize the BiVO₄/overlayer interface by sequentially integrating (CoFeCe)O_x as a hole transport and collection layer and (NiFe)O_x as a surface catalyst with BiVO₄. The resulting photoanode achieved near-complete suppression of interface losses between 0.8 and 1.23 V vs. reverse hydrogen electrode (RHE) for solar water oxidation. In addition, we utilize atomic force microscopy (AFM) to discern and map the overlayer morphology on top of the BiVO₄ photoanode and provide subsequent insights with photoconductive AFM (pc-AFM) measurements to reveal that the holes collected by (CoFeCe)O_x can be efficiently used by (NiFe)O_x for water oxidation. These findings shed light on interface engineering and optimization with BiVO₄/catalyst assemblies, thereby offering guidelines to accelerate the discovery of functional components for PEC water splitting devices.

2. Results and discussion

In the present work, spin-coated, undoped BiVO₄ thin films on fluorine doped tin oxide (FTO)/glass substrates were prepared as previously reported.¹¹ As a starting point for understanding the properties of BiVO₄/(NiFeCoCe)O_x multi-component overlayers, we investigate the specific role of Ni, Fe, Co and Ce by individually depositing each overlayer component (NiO_x, FeO_x, CoO_x and CeO_x) on BiVO₄ films with PED under constant current density. J–E measurements in 0.1 M NaOH solution with and without H₂O₂ as a hole scavenger were conducted (Fig. 1a and b). Measurements with H₂O₂ as a hole scavenger are performed to minimize recombination losses associated with slow catalytic processes and assess the quality of the semiconductor/overlayer assembly for capturing light and separating/transporting charge. In the absence of H₂O₂, the measured photocurrent is lower due to the significant competition between the sluggish kinetics of the OER and surface recombination.

Fig. 1 PEC and AFM characteristics of individual catalyst components. Current–potential curves of BiVO₄, BiVO₄/NiO_x, BiVO₄/FeO_x, and BiVO₄/CeO_x photoanodes in the dark (dashed lines) and under AM 1.5G simulated sunlight at 100 mW cm⁻² (solid lines) in (a) 0.1 M NaOH + H₂O₂ aqueous solution (pH = 13) and (b) 0.1 M NaOH aqueous solution (pH = 13). Overlay images of the sample topography and the corresponding adhesion maps for (c) BiVO₄, (d) BiVO₄/FeO_x, (e) BiVO₄/NiO_x and (f) BiVO₄/CeO_x.

The J–E curves in the presence of H₂O₂ (Fig. 1a) show onset potentials at 0.4 V vs. RHE for BiVO₄/NiO_x, BiVO₄/FeO_x, BiVO₄/CeO_x and pristine BiVO₄ photoanodes. FeO_x shows slightly more favourable onset characteristics at 0.4–0.8 V vs. RHE. As the applied potential increases, the corresponding photocurrent densities of all the samples increase rapidly, as expected since charge extraction is a function of the applied potential. The pristine BiVO₄ photoanode exhibits a photocurrent density of 1.5 mA cm⁻² at 1.23 V vs. RHE, whereas BiVO₄/NiO_x, BiVO₄/FeO_x, and BiVO₄/CeO_x show about 1.9, 1.9, and 2.2 mA cm⁻², respectively, at the same potential. Unexpectedly, while CeO_x is not catalytically active for the OER,²³ BiVO₄/CeO_x exhibits the highest photocurrent density in the presence of H₂O₂ at 0.8–1.4 V vs. RHE. The enhanced performance of BiVO₄ on different overlayers may be attributed to improved light absorption or charge separation/transport. According to UV-Vis measurements, the addition of overlayers with PED barely alters light absorption of BiVO₄ (Fig. S1†). Moreover, all the samples exhibit similar onset potentials (∼0.4 V vs. RHE) as shown in Fig. 1a, thus implying similar interfacial energetics for driving charge separation.

Therefore, the enhanced performances of BiVO₄/NiO_x, BiVO₄/FeO_x, and BiVO₄/CeO_x with H₂O₂ may be ascribed to facilitated charge transport by passivating intrinsic surface states of BiVO₄. Specifically, CeO_x is the most effective at passivating surface states at the high potential region, and FeO_x is excellent at passivating surface states at the onset region. These observations agree well with previous reports on the excellent interface properties of BiVO₄/FeO_x (ref. 16) and BiVO₄/CeO_x.²⁴

By comparing the photocurrents for water and H₂O₂ oxidation, we can examine the surface hole injection efficiencies in BiVO₄/catalyst assemblies as P_{hole injection} = J^H₂O/J^H₂O₂. The corresponding J–E curves of NiO_x, FeO_x, and CeO_x loaded BiVO₄ photoanodes without H₂O₂ are shown in Fig. 1b. BiVO₄/NiO_x and BiVO₄/FeO_x exhibit an enhanced photocurrent density at 1.23 V vs. RHE (1.3 and 1.5 mA cm⁻²) with respect to bare BiVO₄ (0.7 mA cm⁻²). However, relative to the pristine BiVO_4, the BiVO₄/CeO_x assembly shows a significantly reduced photocurrent density for water oxidation over the entire operating potential range. This result contrasts sharply with the superior performance of BiVO₄/CeO_x in the presence of H₂O₂ (Fig. 1a), thus confirming that the catalytic activity of CeO_x for the OER is very poor. Thus, a large portion of the holes that reach the surface tend to undergo recombination instead of interfacial charge transfer, despite the favourable properties of CeO_x for passivating electronically active surface defects.

In order to estimate the probability that the holes reaching the BiVO₄ surface can effectively contribute to the OER, the hole injection efficiency at 1.23 V vs. RHE was calculated for these BiVO₄/catalyst assemblies (Table S1†). For pristine BiVO_4, the hole injection efficiency is only 48% at 1.23 V vs. RHE. This finding reveals significant kinetic competition between surface recombination and water oxidation, which is in line with previous reports.^14,15 Once coupled with NiO_x or FeO_x, this value goes up to 79% or 67%, respectively. This result is in good agreement with the enhanced photocurrent with respect to pristine BiVO₄ in Fig. 1b, thereby implying that both NiO_x and FeO_x retain BiVO₄ catalytic activity and introduce catalytic sites for the OER onto BiVO₄ surfaces. However, CeO_x results in a hole injection efficiency of 27% at 1.23 V vs. RHE. Thus, CeO_x appears to dramatically reduce the overall catalytic activity of the photoelectrode, presumably due to coating the moderately active semiconductor surface with a more chemically inert material, CeO_x.

In order to better understand the role of each individual component, we analysed the morphology and nanomechanical adhesion properties of the overlayer with atomic force microscopy (AFM). This technique allows for direct visualization of the distribution of different oxide overlayer nanoparticles on the BiVO₄ semiconductor. Specifically, we use the topography images overlayed by the corresponding adhesion maps to provide nanoscale information on the surface structure of the BiVO₄/overlayer assembly. Fig. 1c–f show the overlay images for bare BiVO₄, BiVO₄/FeO_x, BiVO₄/NiO_x, and BiVO₄/CeO_x, respectively. The pristine BiVO₄ film exhibits individual BiVO₄ grains ranging between 50 and 200 nm in size with an overall smooth surface. After depositing FeO_x, NiO_x and CeO_x, it is possible to identify 5–10 nm nanoparticles dispersed on the surfaces of the BiVO₄ grains with different degrees of coverage. Specifically, it is interesting to notice that CeO_x particles are homogenously yet discontinuously dispersed on top of the BiVO₄ light absorber (Fig. 1f). This morphology together with the PEC measurements support that while CeO_x passivates electronically active trap states in BiVO₄, it also reduces the catalytically active states. Consequently, this finding leads us to conclude that the surface trap states may be the catalytic states of bare BiVO₄, thus explaining why the BiVO₄/CeO_x shows reduced catalytic activity towards the OER.

In contrast to NiO_x, FeO_x, and CeO_x, CoO_x did not show a significant photoresponse under either front or back-side illumination but greatly enhanced the dark current for the OER (Fig. S2a†). It should also be noted that CoO_x was only deposited for 5 min, as the potential required for deposition increased sharply (Fig. S2b†), whereas the standard deposition time for all other samples was 60 min. This modified deposition approach avoided massive growth of CoO_x on BiVO₄, and prevented CoO_x from affecting light absorption of BiVO₄ (Fig. S2c†). CoO_x deposition eliminates the built-in potential and, thus, the driving force for photocarrier separation. A similar phenomenon was observed for the case of IrO_x on WO₃, which yields an ohmic solid/solid contact.²⁵ This conclusion explains the observation that the BiVO₄/CoO_x anode behaves like a conducting electrocatalyst for the OER. Thus, the BiVO₄/CoO_x junction energetics are poorly suited for creating efficient photoelectrodes.

Based on the above PEC characterization and the charge injection efficiency values of the individual overlayer components, we summarized the individual roles of the metal oxide overlayers in Fig. S3.† It is clear that NiO_x is the main catalytic component in the (NiFeCoCe)O_x multi-component coating. On the other hand, CeO_x and CoO_x modify the interfacial properties of BiVO₄, by passivating electronically active trap states and decreasing the built in potential, respectively. In addition, FeO_x has a moderate impact on both catalytic and passivating functions. Based on these insights, we rationally designed, synthesized, and tested a multi-layer coating with the aim of optimizing both the semiconductor/overlayer interface and the catalytic activity. To this end, we modified NiO_x with FeO_x to form a Ni_0.8Fe_0.2O_x catalytic layer at the outer surface by PED using mixed 0.08 M nickel(II) nitrate hexahydrate, and 0.02 M iron nitrate(III) nonahydrate as the plating solution, which provides a superior catalytic activity relative to the pure NiO_x or FeO_x (Fig. S4†). Given the beneficial interface of BiVO₄/FeO_x and BiVO₄/CeO_x, we attempted to combine FeO_x and CeO_x into an interface layer for efficiently collecting holes over the entire operating potential range. However, the combination of the Fe and Ce species results in poor coating of BiVO₄ with PED, which greatly limits the performance of BiVO₄/(FeCe)O_x (Fig S5a and c†). The different morphologies of CeO_x and (FeCe)O_x indicate that the interaction between the Fe and Ce species leads to a different growth process of the ternary oxide, as supported by the enlarged potential for depositing (FeCe)O_x (Fig. S5b†). Interestingly, we found that incorporating 80% of Co into the (FeCe) precursor solution not only leads to the uniform coating of (CoFeCe)O_x on BiVO₄ with PED but also offers additional redox active sites that are available for hole collection and transport, as demonstrated below. The X-ray photoelectron spectroscopy (XPS) measurements revealed that the deposited (CoFeCe)O_x consists of Co^(II/III), Fe^(III) and Ce^(III/IV) in the form of oxyhydroxides (Fig. S6†). Since XPS is a surface sensitive technique, inductively coupled plasma mass spectrometry (ICP-MS) was used for complementary analysis of the overall composition of (CoFeCe)O_x. The ICP-MS results reveal that (CoFeCe)O_x is composed of Co: 40%, Fe: 10% and Ce: 50% (Co_0.4Fe_0.1Ce_0.5O_x), thus indicating a preferential deposition of CeO_x on BiVO₄ during PED. In addition, energy-dispersive X-ray spectroscopy (EDX) maps, obtained on the scanning transmission electron microscope (STEM), highlight the presence of the overlayer atop the BiVO₄ (Fig. S7†), which shows uniform coating on BiVO₄ (Fig. 2a). As a result, BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x shows superior onset characteristics as well as an optimal photocurrent density of 2.4 mA cm⁻² at 1.23 V vs. RHE with H₂O₂, which is higher than BiVO₄/CeO_x (2.2 mA cm⁻²), BiVO₄/FeO_x (1.9 mA cm⁻²) or BiVO₄/(FeCe)O_x(1.2 mA cm⁻²) (Fig. S5a†) demonstrating that the BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x interface is excellent for transporting holes.

Fig. 2 Characterization of (CoFeCe)O_x overlayer. (a) Overlay images of the sample topography and the corresponding adhesion maps for BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x, (b) Current–potential curves of BiVO₄/FeO_x, BiVO₄/CeO_x and BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x photoanodes in the dark (dashed lines) and under AM 1.5G simulated sunlight at 100 mW cm⁻² in 0.1 M NaOH + H₂O₂ aqueous solution (pH = 13). (c) OCP values of pristine BiVO₄, BiVO₄/FeO_x, BiVO₄/CeO_x and BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x photoanodes in 0.1 M NaOH solution at pH 13 under chopped illumination. (d) Accumulated charge versus potential curves obtained from transient photocurrent data of BiVO₄, BiVO₄/FeO_x, BiVO₄/CeO_x and BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x photoanodes collected in 0.1 M NaOH solution at pH 13 under background white light bias at 1 sun intensity.

To investigate the interface energetics of the Co_0.4Fe_0.1Ce_0.5O_x layer, we monitored the open circuit potential (OCP) profiles and the charge carrier dynamics using transient photocurrent (TCP) measurements (Fig. 2c and d). For a photoelectrode, the changes of OCP reflect charge migration driven by the internal energetics at the electrode/electrolyte interface.¹⁷ As shown in Fig. 2c, the OCP of the pristine BiVO₄ photoanode in darkness is at 0.6 V vs. RHE. Under illumination, the OCP of BiVO₄ immediately reached an initial value (black dashed line), which further increased over time. This effect indicates that holes are accumulating on the surface under illumination, which can be ascribed to the saturation of the surface states by photo-induced processes, analogous to the recently reported photocharging effect on BiVO₄.²⁶ Upon addition of CeO_x, the OCP of BiVO₄ shifts positively, which indicates the physical passivation effect of CeO_x on partially removing surface states of BiVO₄, in line with previous findings.²⁴ In contrast, the addition of FeO_x, or Co_0.4Fe_0.1Ce_0.5O_x marginally alters the resting potential of BiVO₄ under dark conditions, suggesting that these overlayers provide marginal physical passivation of intrinsic surface states on BiVO₄. However, the photocharging processes of BiVO₄ disappear upon overlayer loading, which indicates those overlayers may offer alternative paths for hole transfer other than through the intrinsic surface states, therefore functionally passivating surface states of BiVO₄. This finding can be further supported by the different transient OCPs upon switching off the light. When the light is switched off, the OCP of BiVO₄ slowly decays to the same position as in the dark. This slow process reveals that the photogenerated holes originally accumulated on the surface of BiVO₄ under illumination and charge recombination with the majority of electrons is required to re-establish equilibrium when illumination is stopped. In contrast, after depositing CeO_x on the top, the OCP of BiVO₄/CeO_x goes back to the same position as in the dark without an apparent decay process within the time scales detectable with our measurements. The absence of a decay process further supports the passivation effect of CeO_x, as mentioned earlier. When the light is turned off, we observe a different behaviour with FeO_x and Co_0.4Fe_0.1Ce_0.5O_x. Specifically, the OCP of BiVO₄/FeO_x immediately jumps to a new value, which is 10 mV higher than the initial dark equilibrium value and shows an “opposite” charge recombination decay, compared to the pristine BiVO₄. This phenomenon is even more pronounced with Co_0.4Fe_0.1Ce_0.5O_x. The OCP of BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x jumps to a position which is 300 mV higher than the initial dark equilibrium value and the “opposite” charge recombination decay for BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x is much more significant than that for BiVO₄/FeO_x. The presence of an offset with respect to the initial dark equilibrium suggests that there might be a change in the oxidation state of the FeO_x and Co_0.4Fe_0.1Ce_0.5O_x layers, compatible with the presence of photogenerated holes of BiVO₄. In other words, this finding proves that FeO_x and Co_0.4Fe_0.1Ce_0.5O_x can extract and collect photogenerated holes from BiVO₄ under illumination, building up positively charged species, thereby establishing a new interfacial energetic alignment. The behavior of the Co_0.4Fe_0.1Ce_0.5O_x overlayer implies that the hole extraction ability mainly results from the Co species. Indeed, in the absence of Co, the OCP profile of (FeCe)O_x shows no apparent charge recombination decay upon switching off the light (Fig. S8†), which further demonstrates the important role of Co redox active species in extracting holes from BiVO₄. Notably, BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x is exposed to light for only 3 min and requires about 60 min to recover the original resting level of OCP in the dark (Fig. S9†). These results indicate that hole extraction and collection from BiVO₄ into Co_0.4Fe_0.1Ce_0.5O_x is energetically favourable.

To gain more insight into the charge carrier dynamics between (CoFeCe)O_x and BiVO₄ under operating conditions, TPC measurements were employed using a pulsed 340 nm LED and a continuous white light source. The photocurrent transients upon light perturbation are indicative of capacitive charge accumulation, which offers the possibility for electrons to recombine. The number of charges accumulated at the electrode/electrolyte interface can be qualitatively determined by integrating the photocurrent transients during the monitored surface recombination process. The potential peaks of the accumulation maximum most probably represent the energy levels of the surface redox states, which shed light on the interfacial charge capture/collection on BiVO₄ surfaces.^27–29 To depict this clearly, a plot of the accumulated charge versus potential is shown in Fig. 2d, which is obtained by integrating the photocurrent transients measured in Fig. S10.† Pristine BiVO₄ shows the most pronounced charge accumulation at the electrode/electrolyte interface. When covering the BiVO₄ photoanode surface with overlayers, charge accumulation is attenuated and occurs at more cathodic potentials. The cathodic shift provides strong evidence that the overlayers out-compete the intrinsic surface states for capturing holes, thus providing alternative paths for hole transfer. In particular, the addition of FeO_x cathodically shifts the maximum charge accumulation, indicating that the Fe redox states have more driving force to capture and accumulate holes over the intrinsic surface states on BiVO₄. In contrast, almost no charge accumulation can be calculated from the TPC results of BiVO₄/CeO_x (Fig. S10c†), once again, confirming the passivation effect that prevents hole accumulation via surface states. Interestingly, adding Ce into FeO_x leads to a great reduction in the magnitude of charge accumulation, further indicating that Ce contributes to removing surface states (Fig. S11†). These measurements, together with the OCP results, revealed the role of CeO_x in passivating surface states of BiVO₄ and creating a favourable interface for hole transport, consistent with our previous study on the BiVO₄/CeO_x interface.²⁴ Moreover, a further shift in the energy levels of redox states to 0.7 V vs. RHE is observed due to the presence of Co_0.4Fe_0.1Ce_0.5O_x. This finding suggests that Co offers additional redox states for accepting surface holes. Interestingly, CV measurements of BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x in the dark also show redox features at 0.7 V vs. RHE, produced by Co and Fe species (Fig. S12†). Given the findings above, it is very likely that Co_0.4Fe_0.1Ce_0.5O_x acts as an interface component that extracts/captures photogenerated holes from BiVO₄, out-competing intrinsic surface states by providing surface redox states (Co and Fe species), which lie at more favorable energy levels for hole capture from BiVO₄, while CeO_x mainly provides excellent interface quality with BiVO₄ for hole transport by passivating the intrinsic surface states of the semiconductor.

Although the Co_0.4Fe_0.1Ce_0.5O_x coatings excel at efficient capture and collection of photogenerated holes in BiVO₄, the significant differences of the BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x photoanode for H₂O₂ and H₂O oxidation suggests that the holes collected by Co_0.4Fe_0.1Ce_0.5O_x cannot be efficiently used for the OER, leading to significant surface recombination losses, as shown in Fig. 3. Accordingly, we further assembled a Ni_0.8Fe_0.2O_x catalytic layer on top of the BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x photoanode, as presented in Fig. S13.† As a result, BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x photoanode shows superior performance to BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x and BiVO₄/Ni_0.8Fe_0.2O_x photoanode for OER. In addition, the J–E curve of the BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x photoanode for the OER approaches that one for H₂O₂ oxidation. Strikingly, 85–100% of the surface collected holes are used by Ni_0.8Fe_0.2O_x for the OER at 0.8–1.23 V vs. RHE, as determined by comparing the J–E curves of BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x with (dashed line) and without H₂O₂ hole scavenger (solid line). This finding demonstrates the almost complete elimination of interfacial charge recombination losses with interface optimization. Moreover, the PEC performances of the BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x and BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x photoanodes show similar results with H₂O₂. These results demonstrate that the addition of Ni_0.8Fe_0.2O_x greatly improves the interfacial hole transfer of the BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x photoanode for the OER without altering its charge separation/transport abilities, whereas Co_0.4Fe_0.1Ce_0.5O_x remedies losses at the BiVO₄/Ni_0.8Fe_0.2O_x interface. To demonstrate the generality of our findings, we further performed the PEC characterization of BiVO₄, BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x and BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x at pH 7 (0.1 M KPi), and at pH 9 (0.1 M NaBi).³⁰ As a result, the addition of Co_0.4Fe_0.1Ce_0.5O_x and Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x overlayers exhibit similar effect on BiVO₄ as in the pH 13 electrolyte (Fig. S14†)

Fig. 3 The PEC characterization of the (NiFeCoCe)O_x integrated BiVO₄ photoanodes. Current–potential curves of the BiVO₄/Ni_0.8Fe_0.2O_x, BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x, and BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x photoanodes in the dark (dotted lines) under AM 1.5G simulated sunlight at 100 mW cm⁻² photoanodes in 0.1 M NaOH aqueous solution (pH = 13), compared to current–potential curves of BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x and BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x photoanodes (dashed lines) in 0.1 M NaOH + H₂O₂ aqueous solution (pH = 13).

In order to provide further insights into the charge-transport mechanism through the BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x interfaces, we investigated the current distributions in BiVO₄/Ni_0.8Fe_0.2O_x, BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x, and BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x by pc-AFM in the dark and under illumination. As shown in Fig. S15† and 4, pc-AFM measurements in the dark at an applied sample bias of 1.75 V reveal bright spots in the current map. We previously observed similar bright spots on pristine BiVO₄ and assigned these to shunts between the metal-coated probe and the underling FTO substrate.³¹ The photoelectrodeposited catalysts are sparsely distributed on the BiVO₄ surface, and preferentially deposited on top of the grains, not on the grain boundaries or pinholes, as demonstrated with the overlay maps of BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x in Fig. 2a and BiVO₄/Ni_0.8Fe_0.2O_x, BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x in Fig. S16.† Therefore, only a minor role is assigned to a shunt pathway compared to charge transport across the FTO/BiVO₄/overlayer.³²

Fig. 4 The pc-AFM measurements. Topographic and current maps collected for BiVO₄/Ni_0.8Fe_0.2O_x (a and b), BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x (c and d), and BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x (e and f).

Furthermore, pc-AFM measurements were also conducted using above band gap illumination (E_ph = 3.06 eV, 100 mW cm⁻²). Maps of the current under illumination (Fig. 4b, d and f), for the BiVO₄/Ni_0.8Fe_0.2O_x and the BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x anodes exhibit spots of increased conductivity, whereas an almost homogenous current distribution is observed for the BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x anode. In all three samples, under illumination the current mainly increases on top of the grain and not at the grain boundaries. The corresponding median photocurrents are 0.08 pA for BiVO₄/Ni_0.8Fe_0.2O_x, 0.28 pA for BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x, and 0.38 pA for BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x. These findings are in agreement with the PEC measurements presented above, which showed improved performance for BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x. The higher photocurrent value of BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x compared to BiVO₄/Ni_0.8Fe_0.2O_x might indicate that hole extraction/collection from BiVO₄ is more feasible via the Co_0.4Fe_0.1Ce_0.5O_x overlayer. Accordingly, the photo-conductivity of BiVO₄/Ni_0.8Fe_0.2O_x is greatly enhanced once Co_0.4Fe_0.1Ce_0.5O_x is inserted between Ni_0.8Fe_0.2O_x and BiVO₄. These results not only confirm the role of Co_0.4Fe_0.1Ce_0.5O_x as a hole mediator/collector, but also verify that the holes collected by Co_0.4Fe_0.1Ce_0.5O_x can be efficiently used by Ni_0.8Fe_0.2O_x to drive the OER.

3. Conclusions

In summary, we have investigated the critical interfacial roles of individual component metal oxides in the (NiFeCoCe)O_x multi-component overlayers by interfacing each of them with BiVO₄. We have identified (NiFe)O_x as the surface catalytic component and (CoFeCe)O_x as the charge capture/collection component via available redox active states, by using PEC characterization with and without a H₂O₂ hole scavenger. The charge capture/collection mechanism of (CoFeCe)O_x from BiVO₄ was also studied in detail with OCP and TPC measurements. Accordingly, we further optimized BiVO₄/(NiFeCoCe)O_x by serially interfacing the (CoFeCe)O_x hole collector and (NiFe)O_x catalyst with BiVO₄. In addition, our findings corroborate that BiVO₄ surface trap states may be catalytically active for OER catalysis. The resulting integrated photoanode almost eliminates interfacial losses across the BiVO₄/electrolyte interface over the range of 0.8–1.23 V vs. RHE for the OER. The effective charge transfer through the (CoFeCe)O_x hole collector and (NiFe)O_x catalyst interfaces was further verified by pc-AFM measurements. This study highlights the roles of (NiFeCoCe)O_x multi-component overlayers in preferentially capturing holes over the surface states of BiVO₄ as well as catalyzing the OER. This work demonstrates the significance of interface optimization in the multi-component overlayer-integrated BiVO₄ photoanodes, establishing foundational knowledge to design the next generation of integrated photo-active materials for solar fuel production.

4. Experimental

4.1 Fabrication of BiVO₄ films

Spin-coated bismuth vanadate (BiVO₄) thin films were deposited on the fluorine-doped tin oxide (FTO)-coated side of a 10 cm × 10 cm glass plate (TEC-15 Sigma Aldrich), following a procedure from our previous report.¹¹ The FTO/glass substrates were thoroughly washed with isopropanol (Sigma Aldrich, ≥98%), detergent (Alconex) in deionized water, and pure deionized water, dried with a nitrogen gun and treated with an ozone cleaner (Jelight Model 42) for 10 min.

For spin coating, typically, 15 mL of a 0.2 M solution of bismuth(III) nitrate pentahydrate (Sigma Aldrich, ≥98%) in acetylacetone (Sigma Aldrich, ≥99%) and 100 mL of a 0.03 M solution of vanadium(IV)-oxyacetylacetonate in acetylacetone were prepared separately and sonicated for 10 min. Then, the two solutions were mixed together and further sonicated for 5 min. The resulting solution was filtered with 0.45 μm nylon filters (Thermo Fisher Scientific), and dispensed onto the 10 cm × 10 cm FTO/glass plate. The substrate was then spun two times at 1000 rpm for 6 s on a spin coater (Laurell Technologies) with an acceleration rate of 150 rpm s⁻¹ and annealed in air for 10 min at 500 °C in a muffle furnace (Cole-Parmer). This spin coating-annealing procedure was repeated nine times. After the last spin coating cycle, the substrate was annealed for 2 h at 500 °C to yield a final thickness of ∼50 nm.

4.2 Photoelectrochemical deposition (PED) of catalysts

Catalysts were produced via PED on the BiVO₄ electrodes. A 0.1 M solution of various metal nitrates (nickel(II) nitrate hexahydrate, Sigma Aldrich, ≥97%; iron nitrate(III) nonahydrate Sigma Aldrich, ≥98%; cobalt(II) nitrate hexahydrate, Strem chemicals, ≥99%; and cerium nitrate(III) hexahydrate, Sigma Aldrich, ≥99%) was used as an electrolyte. PED was conducted in an undivided cell at a constant current density of 50 μA cm⁻² over a period of 60 min under AM 1.5G-simulated sunlight. The BiVO₄ electrode was used as the working electrode, Ag/AgCl as the reference electrode and a Pt wire as the counter electrode. The light irradiation came from the back side of electrodes. The 0.1 M metal nitrate precursor solution of (CoFeCe)O_x was prepared by mixing 0.08 M cobalt(II) nitrate hexahydrate, 0.01 M iron nitrate(III) nonahydrate and 0.01 M cerium nitrate(III) hexahydrate. Likewise, the 0.1 M metal nitrate precursor solution of (NiFe)O_x was prepared by mixing 0.08 M nickel(II) nitrate hexahydrate, and 0.02 M iron nitrate(III) nonahydrate. The reported ratios in the chemical formulae were provided by ICP-MS results. The BiVO₄/Co_0.4Fe_0.1Ce_0.5O_x/Ni_0.8Fe_0.2O_x sample was assembled by depositing Co_0.4Fe_0.1Ce_0.5O_x for 15 min and then Ni_0.8Fe_0.2O_x for the rest 45 min.

4.3 UV-Vis transmission and reflectance measurements

Transmission and spectral reflectance measurements were performed on a Shimadzu SolidSpec-3700 UV/Vis/NIR spectrometer using an integrating sphere.

4.4 Photoelectrochemical (PEC) and open circuit potential (OCP) measurements

The PEC test was conducted in a single-compartment, three-electrode electrochemical cell with a potentiostat (Bio-Logic, SP-300) under simulated AM 1.5G solar light irradiation (100 mW cm⁻², 16S-300-002, Solar Light). The fabricated electrode, a platinum wire, and Ag/AgCl electrode were used as the working, counter, and reference electrodes, respectively. Unless expressly mentioned, 0.1 M NaOH aqueous solution (pH = 13) was used as an electrolyte after saturation with N₂ gas for 20 min. The measured potentials were converted to V vs. RHE (E_RHE = E_Ag/AgCl + 0.193 V + 0.059 V × pH). The photocurrent was measured by linear sweep voltammetery with a scan rate of 10 mV s⁻¹. The light irradiation came from the front side of the electrodes for all cases. OCP measurements were performed under open circuit condition in the same electrolyte. The OCP reading in the dark or under illumination was obtained after a stabilization process in the dark (10 min).

4.5 Transient photocurrent (TPC) measurements

TPC measurements were conducted in a similar setup to PEC measurements. The main difference is the working electrode was exposed to a continuous ∼100 mW cm⁻² light bias from a home-built white light-emitting diode (LED) array. A 340 nm LED (M340L4, ThorLabs) with ∼10 nm spectral full width at half maximum (FWHM) and ∼10 mW cm⁻² power density was used as the modulating light source, alternating between 150 ms on-time and 203 ms off-time. The applied potential was increased from 0.56 to 1.36 V vs. RHE at 0.05 V intervals. After passing through a current pre-amplifier (SR570, Stanford Research Systems), the photocurrent transients induced by the 340 nm LED were collected at a frequency of 10⁵ Hz and averaged over 100 light pulses.

4.6 X-ray photoelectron spectroscopy (XPS) and inductively coupled plasma mass spectrometry (ICP-MS)

The near-surface chemical composition was analyzed by XPS using a monochromatized Al Kα source (hν = 1486.6 eV) on a Kratos Axis Ultra DLD system with a pass energy of 20 eV. Spectral fitting was conducted using the Casa XPS analysis software. Spectral energy positions were corrected by shifting the C 1s core level position to 284.8 eV and curves were fit with quasi-Voigt lines following Shirley background subtraction. ICP-MS was performed using an Agilent 7900 system run with the He mode. The internal standard was Ge for Ni, Fe, Co and Tb for Ce, selected based on their first ionization potentials and M/Z. The samples were digested with 1 mL of 70% HNO₃ (>99.999% trace metals basis, 225711, Aldrich) and diluted to 5 mL in total with 1% HNO₃ solution as the original sample solutions.

4.7 Transmission electron microscopy (TEM) measurements

TEM measurements were conducted on a FEI TitanX 60-300 microscope at NCEM using a double tilt holder and an accelerating voltage of 200 kV. The images were acquired with a HAADF detector for STEM. EDS was collected using a Bruker windowless EDS detector.

4.8 Atomic force microscopy (AFM) measurements

All the AFM measurements were performed with a commercial AFM system (Bruker Dimension Icon). PeakForce Quantitative Nanoscale Mechanical (PF-QNM) mode was used for recording the topography and adhesion maps simultaneously. For all the PF-QNM measurements, uncoated Si-probes (RFESP-75) with spring constants of 3.0 N m⁻¹ were used.

For pc-AFM mapping, PeakForce TUNA mode was used, whereby a positive bias of 1.75 V was applied to the FTO back contact. For all the pc-AFM measurements, the sample was illuminated from the FTO-side using a specially designed illumination setup.³¹ As the light source, a CW laser diode with a wavelength of 405 nm and an illumination intensity of about 100 mW cm⁻¹ was used. The pc-AFM measurements were performed with conductive PtIr-coated probes (Bruker SCM-PIT) exhibiting a nominal spring constant of 2.8 N m⁻¹.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This study is based on work performed at the Joint Center for Artificial Photosynthesis, a DOE Energy Innovation Hub, supported through the Office of Science of the U.S. Department of Energy under Award Number DE-SC0004993. Work at the Molecular Foundry was supported by the Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.

Notes and references

1. J. Su and L. Vayssieres, ACS Energy Lett., 2016, 1, 121–135.
2. N. S. Lewis, Science, 2007, 315, 798–801.
3. S. Dahl and I. Chorkendorff, Nat. Mater., 2012, 11, 100–101.
4. M. Gratzel, Nature, 2001, 414, 338–344.
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. X. Chen, C. Li, M. Gratzel, R. Kostecki and S. S. Mao, Chem. Soc. Rev., 2012, 41, 7909–7937.
7. B. A. Pinaud, J. D. Benck, L. C. Seitz, A. J. Forman, Z. Chen, T. G. Deutsch, B. D. James, K. N. Baum, G. N. Baum, S. Ardo, H. Wang, E. Miller and T. F. Jaramillo, Energy Environ. Sci., 2013, 6, 1983–2002.
8. I. D. Sharp, J. K. Cooper, F. M. Toma and R. Buonsanti, ACS Energy Lett., 2017, 2, 139–150.
9. F. F. Abdi, T. J. Savenije, M. M. May, B. Dam and R. van de Krol, J. Phys. Chem. Lett., 2013, 4, 2752–2757.
10. B. Pattengale and J. Huang, Phys. Chem. Chem. Phys., 2017, 19, 6831–6837.
11. F. M. Toma, J. K. Cooper, V. Kunzelmann, M. T. McDowell, J. Yu, D. M. Larson, N. J. Borys, C. Abelyan, J. W. Beeman and K. M. Yu, Nat. Commun., 2016, 7, 12012.
12. H. Ye, H. S. Park and A. J. Bard, J. Phys. Chem. C, 2011, 115, 12464–12470.
13. D. K. Zhong, S. Choi and D. R. Gamelin, J. Am. Chem. Soc., 2011, 133, 18370–18377.
14. C. Ding, J. Shi, D. Wang, Z. Wang, N. Wang, G. Liu, F. Xiong and C. Li, Phys. Chem. Chem. Phys., 2013, 15, 4589–4595.
15. J. A. Seabold and K.-S. Choi, J. Am. Chem. Soc., 2012, 134, 2186–2192.
16. T. W. Kim and K. S. Choi, Science, 2014, 343, 990–994.
17. M. Zhong, T. Hisatomi, Y. Kuang, J. Zhao, M. Liu, A. Iwase, Q. Jia, H. Nishiyama, T. Minegishi, M. Nakabayashi, N. Shibata, R. Niishiro, C. Katayama, H. Shibano, M. Katayama, A. Kudo, T. Yamada and K. Domen, J. Am. Chem. Soc., 2015, 137, 5053–5060.
18. D. Guevarra, A. Shinde, S. K. Suram, I. D. Sharp, F. M. Toma, J. A. Haber and J. M. Gregoire, Energy Environ. Sci., 2016, 9, 565–580.
19. A. Shinde, D. Guevarra, G. Liu, I. D. Sharp, F. M. Toma, J. M. Gregoire and J. A. Haber, ACS Appl. Mater. Interfaces, 2016, 8, 23696–23705.
20. Y. Ma, A. Kafizas, S. R. Pendlebury, F. L. Formal and J. R. Durrant, Adv. Funct. Mater., 2016, 26, 4951–4960.
21. C. Zachaus, F. F. Abdi, L. M. Peter and R. van de Krol, Chem. Sci., 2017, 8, 3712–3719.
22. G. V. Govindaraju, G. P. Wheeler, D. Lee and K.-S. Choi, Chem. Mater., 2017, 29, 355–370.
23. J. A. Haber, E. Anzenburg, J. Yano, C. Kisielowski and J. M. Gregoire, Adv. Energy Mater., 2015, 5, 1402307.
24. A. Shinde, G. Li, L. Zhou, D. Guevarra, S. K. Suram, F. M. Toma, Q. Yan, J. A. Haber, J. B. Neaton and J. M. Gregoire, J. Mater. Chem. A, 2016, 4, 14356–14363.
25. J. M. Spurgeon, J. M. Velazquez and M. T. McDowell, Phys. Chem. Chem. Phys., 2014, 16, 3623–3631.
26. B. J. Trzesniewski and W. A. Smith, J. Mater. Chem. A, 2016, 4, 2919–2926.
27. F. Le Formal, K. Sivula and M. Grätzel, J. Phys. Chem. C, 2012, 116, 26707–26720.
28. F. A. L. Laskowski, J. Qiu, M. R. Nellist, S. Z. Oener, A. M. Gordon and S. W. Boettcher, Sustainable Energy Fuels, 2018.
29. G. Liu, J. Shi, F. Zhang, Z. Chen, J. Han, C. Ding, S. Chen, Z. Wang, H. Han and C. Li, Angew. Chem., Int. Ed., 2014, 53, 7295–7299.
30. L. Zhou, A. Shinde, D. Guevarra, F. M. Toma, H. S. Stein, J. M. Gregoire and J. A. Haber, ACS Appl. Energy Mater., 2018, 1, 5766–5771.
31. J. Eichhorn, C. Kastl, J. K. Cooper, D. Ziegler, A. M. Schwartzberg, I. D. Sharp and F. M. Toma, Nat. Commun., 2018, 9, 2597.
32. J. Qiu, H. Hajibabaei, M. R. Nellist, F. A. L. Laskowski, S. Z. Oener, T. W. Hamann and S. W. Boettcher, ACS Energy Lett., 2018, 3, 961–969.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8se00473k

---