Pure CuBi2O4 Photoelectrodes with Increased Stability by Rapid Thermal Processing of Bi2O3/CuO Grown by Pulsed Laser Deposition

A new method for enhancing the charge separation and photo‐electrochemical stability of CuBi2O4 photoelectrodes by sequentially depositing Bi2O3 and CuO layers on fluorine‐doped tin oxide substrates with pulsed laser deposition (PLD), followed by rapid thermal processing (RTP), resulting in phase‐pure, highly crystalline films after 10 min at 650 °C, is reported. Conventional furnace annealing of similar films for 72 h at 500 °C do not result in phase‐pure CuBi2O4. The combined PLD and RTP approach allow excellent control of the Bi:Cu stoichiometry and results in photoelectrodes with superior electronic properties compared to photoelectrodes fabricated through spray pyrolysis. The low photocurrents of the CuBi2O4 photocathodes fabricated through PLD/RTP in this study are primarily attributed to their low specific surface area, lack of CuO impurities, and limited, slow charge transport in the undoped films. Bare (without protection layers) CuBi2O4 photoelectrodes made with PLD/RTP shows a photocurrent decrease of only 26% after 5 h, which represents the highest stability reported to date for this material. The PLD/RTP fabrication approach offers new possibilities of fabricating complex metal oxides photoelectrodes with a high degree of crystallinity and good electronic properties at higher temperatures than the thermal stability of glass‐based transparent conductive substrates would allow.


Introduction
In recent years, the ternary oxide CuBi 2 O 4 has been gaining significant interest as a photocathode (p-type semiconductor) material for photo-electrochemical (PEC) water splitting owing to its appealing electronic properties. [1,2] Its reported bandgap energy necessary to fabricate them with high quality and purity, minimizing impurities and defects.
The phase diagram of Bi 2 O 3 and CuO ( Figure S1, Supporting Information), displays two significant challenges in the fabrication of CuBi 2 O 4 thin films photoelectrodes. [16][17][18] First, the solid-state reaction of Bi 2 O 3 + CuO → CuBi 2 O 4 begins at temperatures above ≈600 °C until CuBi 2 O 4 melts incongruently to CuO + liquid at 820 ± 20 °C. These temperatures are too high for most commonly used transparent conductive substrates on which PEC photoelectrodes are typically deposited. The glass transition temperature, T g , of typical soda-lime glass used for transparent conductive oxides (TCO) ranges between ≈540 and 570 °C. [19][20][21] A precise T g will depend on the chemical composition of the glass used by each TCO substrate manufacturer. The most widely used TCO films are tin-doped In 2 O 3 (ITO) and fluorine-doped SnO 2 (FTO). The conductivity of ITO films already starts to decrease above ≈350 °C, whereas FTO films are more stable, but also start to lose conductivity above ≈600 °C. [22,23] However, reports show that FTO remains stable at 800-1000 °C for short heating times (≈10 1 -10 3 s). [20,24,25] Second, there is no Cu x Bi 2−x O 4 solid solution (CuBi 2 O 4 is a line compound), resulting in the segregation of either CuO or Bi 2 O 3 when Bi:Cu stoichiometry ≠ 2. Note that both CuO and Bi 2 O 3 are photoelectrochemically active materials, Figure S2 (Supporting Information). To mitigate the gap between the desired reaction temperatures of CuBi 2 O 4 and the practical fabrication conditions of photoelectrodes thin films, many approaches utilize alternative syntheses methods, a compromise that researchers also employ in the fabrication of other photo active materials, [26,27] especially for complex metal oxide (CMO) photo electrodes. Deposition techniques such as spray-pyrolysis, spin coating, drop-casting, inkjet printing, and electrochemical deposition are widely used to fabricate CMO photoelectrodes, and CuBi 2 O 4 in partic ular. [1,3,5,9,11,[28][29][30] These chemical techniques rely on the use of a solvent, molecular precursors, and stabilizers, which are known to affect the formation mechanism, the films' morphology, and therefore its electronic properties. However, these films typically consist of particles with ill-defined shapes, different roughness (porosity), and crystallinity, as well as fluctuations in stoichiometry. Different photo-electrochemically active phases intermixed in a photoelectrode film can influence the charge carrier dynamics in the film's bulk (photogeneration, separation, and transport), [11] as well as the film's surface (recombination vs catalysis), [31] and the photo-electrochemical stability of the photoelectrode. [32] Therefore, a fabrication approach that enables the synthesis of high-purity CuBi 2 O 4 thin films and allows for uniform thermal processing without adversely affecting the properties of the transparent conductive oxide substrate is necessary.
Here, we present a new fabrication approach for CuBi 2 O 4 by combining pulsed laser deposition (PLD) to sequentially deposit the binary oxides Bi 2 O 3 and CuO on FTO substrates with rapid thermal processing (RTP) to carry out the Bi 2 O 3 + CuO → CuBi 2 O 4 reaction. The process is illustrated in Scheme 1. A comparative study with conventional furnace heating (FH) as the post-deposition thermal treatment reveals the importance of rapid radiative heating in the processing of pure-phase complex metal oxides photoelectrodes. We show that for both heating methods, a preferred and faster formation of CuBi 2 O 4 occurs in FTO/Bi 2 O 3 /CuO in comparison to FTO/CuO/Bi 2 O 3 films. Remarkably, we discovered that FTO/Bi 2 O 3 /CuO films heated by RTP resulted in phase-pure CuBi 2 O 4 films, whereas similar films heated for 72 h by FH did not. The phase-pure CuBi 2 O 4 films comprised of micron-sized particles, and their optical, electronic, and photo-electrochemical properties were investigated and compared to porous CuBi 2 O 4 photoelectrodes made by the spray-pyrolysis technique, which is currently the benchmark technique to fabricate CuBi 2 O 4 photoelectrodes. [5,9,28] The phase-pure CuBi 2 O 4 photoelectrodes made by combined PLD and RTP demonstrate superior electronic properties and unprecedented photo-electrochemical stability without protection layers and catalysts.

Structural Characterization
To achieve a pure CuBi 2 O 4 film with a reported optimal thickness of ≈250 nm, [5] and based on the bulk density values of all three oxides (see the Experimental Section), the thickness of deposited Bi 2 O 3 throughout the study was kept at 240 nm. The CuO thickness was varied between 150-360 nm to find the correct thickness ratio that would correspond to a Bi:Cu stoichiometry of 2:1. X-ray diffractograms and Raman micro scopy, complemented with scanning electron microscopy (SEM) images confirm that under our deposition conditions, a CuO film thickness of 240 nm on top of a 240 nm thick Bi 2 O 3 was the correct thicknesses ratio to achieve a single CuBi 2 O 4 phase. A CuO thickness of 240 nm is the only thickness for which we found no secondary Bi 2 O 3 or CuO phases with X-ray diffraction (XRD) and Raman analysis ( Figure S3, Supporting Information) and no CuO particles on the surface of the film by SEM analysis ( Figure S4, Supporting Information). Figure 1a shows X-ray diffractograms of Bi 2 O 3 /CuO on FTO glass substrates, as-deposited, after FH for 72 h at 500 °C in air, and after RTP for 10 min under a controlled temperature of 650 °C in one bar of oxygen. We note that inductively coupled plasma mass spectrometry (ICP-MS) measurements of dissolved films confirm an identical Bi:Cu ratio in all three samples, eliminating any concern of losing Bi 2 O 3 during heating due to its high volatility. [33] The as-deposited two-layer sample shows Bi 2 O 3 and CuO phases, as well as the formation of CuBi 2 O 4 as evidence by the peaks at 28 The strong tendency of the RTP-grown films to form the CuBi 2 O 4 phase is also observed for films in which the order of the layers was reversed, i.e., when the CuO layer was deposited first (CuO/Bi 2 O 3 films, Figure S5, Supporting Information). These films do not show any CuBi 2 O 4 in their as-deposited form. After FH-treatment, these films show similar amounts of Bi 2 O 3 and only a small amount of CuBi 2 O 4 , whereas RTP-grown films show a clear reduction of the amount of Bi 2 O 3 and a large increase in the amount of CuBi 2 O 4 . Nevertheless, a complete conversion to CuBi 2 O 4 did not occur after similar processing times. The reason for this is not yet fully understood. However, the importance of the order of the deposition and the fact that as-deposited CuO/Bi 2 O 3 films show different crystallographic orientations than the Bi 2 O 3 /CuO films ( Figure S5, Supporting Information) provide important clues on the formation mechanism of CuBi 2 O 4 , which will be discussed in more detail below. Figure 1b shows the Raman microscopy analysis of the FH and RTP films with a CuBi 2 O 4 single crystal and single-layer Bi 2 O 3 and CuO films deposited on FTO substrates as references. As CuBi 2 O 4 shares similar bonds with Bi 2 O 3 and CuO, many of its Raman peaks are located at comparable wavenumbers as the binary oxides. [34][35][36] When comparing the Bi 2 O 3 /CuO films with a single crystal, Bi 2 O 3, and CuO references, it is clear that the FH film still contains significant amounts of Bi 2 O 3 and CuO. This is evidenced by the peaks at 184, 209, and 444 for Bi 2 O 3 , and 294 and 342 cm −1 for CuO. These peaks are absent in the single crystal and RTP film, which is consistent with the XRD results and confirms that RTP-grown films strongly favor the formation of the CuBi 2 O 4 phase. Furthermore, in the RTP film, all CuBi 2 O 4 Raman peaks showed a shift to higher wavenumbers. This typically indicates a decrease in the atomic bond lengths, which one would expect if the structure transforms into a more ordered and crystalline state. This is indeed what we expect based on the detected decrease in XRD peaks widths (vide supra) and change in grain size that we will discuss in more detail below. In addition to a more ordered and crystalline phase, the effect of strain in the PLD/RTP film is also a possibility, as the Raman peaks are blue-shifted relative to those of the CuBi 2 O 4 single crystal.
The film morphologies and elemental compositions were characterized using SEM along with energy-dispersive X-ray (EDX), seen in Figure 2. The Bi 2 O 3 /CuO + RTP film, Figure 2a, consists of tightly packed, large elongated particles, which closely resembles the morphology of Bi 2 O 3 particles, Figure S6 (Supporting Information). In contrast, the Bi 2 O 3 /CuO + FH film, Figure 2b, was more porous and comprised smaller nanostructured particles that form larger aggregates, closely resembling the morphology of a CuO film, Figure S6 (Supporting Information). Cross-sectional imaging and elemental mapping of both films support the XRD and Raman data, by showing that the Bi 2 O 3 /CuO + RTP film consists of a single layer with Cu and Bi distributed homogeneously throughout its thickness, as shown in Figure 2c-  To examine the effect of RTP heating times on the formation of CuBi 2 O 4 , additional Bi 2 O 3 /CuO films were heated for 1, 5, 10, 15, and 20 min. Careful analysis of the X-ray diffractograms and Raman spectra show that heating durations of 1 and 5 min are not sufficient to complete the solid-state reaction, Figure S7 (Supporting Information). Heating the films for longer than 10 min results in an increased crystallinity of the CuBi 2 O 4 phase. Since these gaps can lead to electrochemical shorts between the electrolyte and the FTO back-contact, films treated for 15 and 20 min. will not be considered any further.
Next, the reason for attaining a pure phase of CuBi 2 O 4 after heating for 10 min by RTP, which did not occur after heating for 72 h in a furnace, was investigated. Spinolo et al. studied the chemical diffusion and the reactivity between two pressed pellets of Bi 2 O 3 and CuO. [37,38] These authors confirmed experimentally that Cu 2+ and O 2− are the mobile species that concurrently diffuse into the Bi 2 O 3 layer and measured an ambipolar diffusion coefficient of 2 × 10 −12 cm 2 s −1 at 750 °C (in air). We Adv. Funct. Mater. 2020, 30,1910832  have used the following equation to approximate the time needed for the diffusion of the relevant species when heating our films in a furnace [39] Here, D is the diffusion coefficient, x the mean distance traveled by the mobile species, and t is the elapsed time since diffusion began. Based on Equation (1), at 750 °C, for D = 2 × 10 −12 cm 2 s −1 , the CuO would fully diffuse into a 240 nm Bi 2 O 3 film in about 2.4 min. However, even after heating in a furnace Bi 2 O 3 /CuO and Cu/Bi films on quartz substrates for much longer times (1 h at 750 °C in air), the transformation is still not complete ( Figure S8, Supporting Information). This indicates that the transformation to CuBi 2 O 4 is not only governed by diffusion. Other processes, such as nucleation, are likely to play a role during the formation of the CuBi 2 O 4 phase.
Our interpretation of the diffusion and grain growth processes during RTP and FH is illustrated in Scheme 2. At t 0 , the initial conditions of both RTP and FH systems are similar, with CuBi 2 O 4 grains at the Bi 2 O 3 /CuO interface in the as-deposited films, acting as nucleation sites. In films heated by RTP, the high heating rate (10 K s −1 ) brings the system into a steady state at 650 °C in ≈1 min. In this state, fast diffusion of Cu ions into the Bi 2 O 3 layer occurs, and the nucleation of CuBi 2 O 4 begins throughout the film. After ≈5 min, the homogeneous distribution of nucleation sites leads to rapid grain growth, which culminates in a single-phase CuBi 2 O 4 after 10 min. However, in films heated in a conventional furnace, a slower heating rate (10 K min −1 ) brings the system into a steady-state only after ≈1 h, during which slow diffusion and some grain growth occur. From this point onward, larger CuBi 2 O 4 grains impede Cu ions diffusing further into the Bi 2 O 3 layer (represented by illustrating the system after ≈3 h). As a result, an incomplete phase transformation in the Cu-Bi-O layer is observed, even after heating for 72 h.

Scheme 2.
Illustrative interpretation of the diffusion and grain growth processes during RTP and FH. Temperature versus time profiles of the RTP and Furnace (FH), and the effect of heating rates on diffusion and nucleation process in the CuO-Bi 2 O 3 system. In the RTP system, high heating rates promote fast diffusion of Cu ions into the Bi 2 O 3 layer, which lead to homogeneous nucleation and rapid grain growth of CuBi 2 O 4 throughout the Bi 2 O 3 thickness. In the Furnace (FH) system, low heating rates lead to slow diffusion into the Bi 2 O 3 layer and grain growth, which impedes homogeneous nucleation, resulting in incomplete phase transformation. and have a similar thickness of ≈250 nm. The transmittance of the CuBi 2 O 4 produced by combined PLD and RTP differs by two features. First, it is more transparent throughout the entire visible range. Second, two peaks are apparent in the spectrum, centered at 530 and 650 nm, whereas the transmittance of the CuBi 2 O 4 made by spray pyrolysis increases nearly monotonically with increasing wavelength.
Typically, differences in the absorption intensity and shape of films with similar thicknesses are attributed to variations in morphology and compositional changes. The higher transparency of PLD/RTP-grown CuBi 2 O 4 films can be explained by less scattering from the large elongated grains than from the porous CuBi 2 O 4 film made by spray pyrolysis, Figure S9 (Supporting Information). Figure 4b compares the absorption of both films to PLD/RTP-grown CuBi 2 O 4 films with varying thicknesses of pristine CuO on top of a 240 nm Bi 2 O 3 (deposited before heating by RTP), along with the absorption of a bare FTO substrate. The peaks at ≈490 and 600 nm in the PLD/RTP-grown CuBi 2 O 4 can result from either constructive optical interference, morphology or they reflect the density of states in the intrinsic electronic structure of CuBi 2 O 4 . A possible constructive optical interference can be excluded, as the peaks are invariant in films with thicknesses ranging from 250-360 nm. Reports of porous CuBi 2 O 4 have shown evidence of similar peaks, ruling out differences in morphology as the reason as well.
The reported peaks were, however, correlated with a Bi:Cu stoichiometry ≈2. [5,11,40] When the reported stoichiometry was < 2 (i.e., segregation of CuO), a nearly monotonically change in absorption with increasing wavelength was evident. [5,11] Analysis of the XRD patterns and Raman spectra of a CuBi 2 O 4 film made by spray pyrolysis displays evidence of a secondary crystalline CuO phase, seen in Figure S10 (Supporting Information). An examination of the absorption of a 250 nm CuO film deposited on FTO shows a monotonic decrease with increasing wavelength, similar to the CuBi 2 O 4 made by spray-pyrolysis, Figure S11 (Supporting Information). Therefore, it can be concluded that the absorption peaks at ≈490 and 600 nm originate from the intrinsic electronic structure of CuBi 2 O 4 . Moreover, it appears that these intrinsic absorption peaks are only clearly visible in CuBi 2 O 4 films made with PLD + RTP; films made by, e.g., spray pyrolysis exhibit optical scattering, which hides these features. Different fabrication methods can affect not only the films' morphology, but their electronic properties as well, and consequently their stability. Therefore, it is important to distinguish between the way CuO is formed in both methods and its possible effects on the photocathode properties. In the Bi 2 O 3 + CuO reaction in the RTP, CuO can segregate out at the surface on the compact CuBi 2 O 4 layer by merely controlling the pristine CuO thickness (vide supra). However, in the sprayed films, CuO is uncontrollably formed and intermixed with CuBi 2 O 4 in the porous layer, formed by pyrolyzing a solution containing bismuth and copper precursors. We discuss below the properties of the CuBi 2 O 4 films, which are differently influenced by the fabrication methods.
To assess the optoelectronic quality and the bandgap energies of the films, light-modulated surface photovoltage (SPV) measurements were conducted and compared with analyses of a single crystal of CuBi 2 O 4 as a reference. The SPV method is an established technique for the contactless characterization of semiconductors, relying on the analysis of light-induced changes of the surface potential. The strength of using modulated over steady-state SPV measurements is that modulation of the light excitation results in lock-in amplification of the SPV signals and enable high sensitivity to charge separation processes even in extremely thin photoactive layers or single nanocrystals, even though the absolute signal amplitudes are lower than for conventional steady-state Kelvin probe measurements. [41] Modulated SPV signals depend on photogeneration, charge separation, charge transport, and charge recombination and back transfer processes. This means that only those absorption events will result in SPV signals for which photogenerated charge carriers can follow the modulation frequency. Furthermore, SPV spectra can provide information on the bandgap energy as well as on photoactive electronic defect states located within the bandgap. A detailed account of the bandgap energy  (E g ) extraction from the light-modulated SPV measurements is presented in Figure S12 (Supporting Information). Figure 5a shows the PV amplitude spectra of a CuBi 2 O 4 crystal before and after cleaving (the corresponding Raman spectra are seen in Figure S13, Supporting Information). The negative in-phase signals (not shown) are characteristic for a p-type semiconductor with a space-charge region at the surface. The E g for both single crystals is 1.9 ± 0.02 eV, Figure S12 (Supporting Information). The exponential tail widths, E t , are extracted from the inverse of the slope in the bandgap region, and amount to 90 and 130 meV for the cleaved and uncleaved crystals, respectively. Smaller E t values indicate higher electrically ordered photoabsorbers with low sub-bandgap trap densities near the band edge and, therefore, reduced recombination losses, which is in general beneficial for photoelectrode performance. [42] Thus, the cleaved crystals show less surface disorder, as one would indeed expect. This is also consistent with the higher PV amplitude signal for the cleaved sample, which reflects a higher efficiency of charge separation.
The PV amplitude spectra of the CuBi 2 O 4 films made by combined PLD and RTP and by spray-pyrolysis (illuminated from both sides, back and front) are compared and presented in Figure 5b. For CuBi 2 O 4 made by combined PLD and RTP, E t and E g values were 90 meV and 1.98 ± 0.05 eV when measured under front illumination. Under back illumination, the E t and E g values were 80 meV and 1.87 ± 0.05 eV. On the other hand, upon front illumination of the CuBi 2 O 4 made by spray-pyrolysis, the SPV signals show E t and E g values of 150 meV and 1.65 ± 0.05 eV, respectively. However, no SPV signals were detected under back illumination. This can be due to the following reasons: i) as a result of high absorption coefficient, supra-bandgap photons do not penetrate throughout the entire film, and ii) effective charge separation only takes place at the front side. We wish to highlight that in photo-electrochemical conditions, the electrolyte acts as a selective contact, which explains why the more porous CuBi 2 O 4 photoelectrode made by spray-pyrolysis is PEC-active (but not SPV-active) upon back illumination (vide infra). At 2.5 eV, the PV amplitudes were 80 and 5 µV for the CuBi 2 O 4 made by combined PLD and RTP and spray-pyrolysis. Therefore, the modulated charge separation was much more efficient for the PLD/RTP samples. Reported density functional theory calculations of the electronic band structures of CuBi 2 O 4 denote a bandgap energy of ≈1.9 eV for a pristine CuBi 2 O 4 lattice and a lower one of ≈1.75 eV for a CuBi 2 O 4 lattice with Cu vacancies. [43] Thus, the lower bandgap energy obtained for the CuBi 2 O 4 photoelectrode made by spray pyrolysis (1.65 ± 0.05 eV) may be due to a higher defect concentration in the form of Cu vacancies. This is consistent with the larger tail energy of the sprayed samples (150 meV vs 90 meV for the PLD/RTP samples). In addition, the presence of crystalline CuO (E g ≈ 1.4 eV) [44][45][46] in the sprayed samples may also lead to a lower effective bandgap of the sample.
When comparing the bandgap and tail energy values (E g and E t ) obtained for the different samples, the strong similarities between the CuBi 2 O 4 made by PLD/RTP and the single crystal demonstrates the superior electronic properties of the PLD/ RTP films compared to the sprayed CuBi 2 O 4 samples. Finally, we note a strong feature at ≈3.1 eV for the PLD/RTP samples that is present for both front and back illumination, but not for the CuBi 2 O 4 made by spray pyrolysis. This transition at higher energies can represent an additional allowed electronic transition related to the intrinsic band structure of CuBi 2 O 4 , as suggested by Sharma et al., [43] or can hint at traces of Bi 2 O 3 in the film. Bi 2 O 3 is a wide-bandgap semiconductor with E g that can range from 2.4 to 3.96 eV, depending on its phase and the preparation method. [12,47,48] The origin of the feature at 3.1 eV is beyond the scope of this study and will be addressed in a separate publication.

Photo-Electrochemical Performance
Photo-electrochemical studies were conducted on CuBi 2 O 4 photo electrodes made by PLD/RTP with different Bi:Cu ratios.  The latter was controlled by the thickness of the CuO layer that was deposited on top of a 240 nm Bi 2 O 3 layer. Figure 6 depicts the current density values (at 0.6 V vs RHE) as a function of the CuO thickness. Typically, films of single phase CuBi 2 O 4 showed photocurrents of ≈0.1 mA cm −2 (at 0.6 V vs RHE) in the presence of H 2 O 2 as an electron scavenger. This is substantially lower than previously reported record photocurrent values for CuBi 2 O 4 , [5,9,11,28] the reason for this will be discussed below. However, the addition of a ≈30 nm layer of CuO particles on top of the CuBi 2 O 4 increased the photocurrents to ≈−0.3 to −0.5 mA cm −2 and kept increasing for thicker films of pristine CuO (denser CuO top-layers), Figure S14 (Supporting Information). Despite their lower photocurrents, from this point onward in the PEC and stability studies, we focus our effort on pure CuBi 2 O 4 films to eliminate any effects of CuO on their investigated properties. Figure 7 displays chopped linear sweep voltammetry (LSV) scans of CuBi 2 O 4 photoelectrodes made by PLD/RTP and spray-pyrolysis under simulated AM1.5 illumination. Scans for back-and front-illumination are presented, as well as LSVs with H 2 O 2 . The insets show a magnified area of the LSVs without H 2 O 2 , revealing photocurrent onset potentials of ≈1.1 and 0.95 V versus RHE for the PLD/RTP and the spray-pyrolysis photoelectrodes, respectively. Furthermore, in contrast to previous reports, the PLD/RTP photoelectrodes show higher photo currents when illuminated from the front side. This implies that under our synthesis conditions, the transport of the electrons, not of the holes, limits the photoconversion efficiency of pristine CuBi 2 O 4 . [1] A similar observation is evident in LSVs of PLD/RTP-grown CuBi 2 O 4 with a CuO top-layer, Figure S15a (Supporting Information). In contrast, the Bi 2 O 3 /CuO photoelectrodes treated with FH for 72 h showed higher photocurrents when illuminated from the back side, see Figure S15b (Supporting Information). A full understanding of the carrier dynamics of these electrodes, which can be described as FTO/Bi 2 O 3 /CuBi 2 O 4 /CuO (vide supra) and consisted of no less than three photoactive phases, is beyond the scope of this study.
The photocurrents of the sprayed samples are considerably higher than those of the PLD/RTP samples (−1.89/−1.68 mA cm −2 at 0.6 V versus RHE for back/front illumination of sprayed samples compared to −0.03/−0.09 mA cm −2 at 0.6 V versus RHE, for back/front illumination of PLD/RTP samples, all in the presence of H 2 O 2 ). The lower photocurrents of the PLD/RTP films can be attributed to one or more of the following factors: i) low specific surface area, ii) lack of photocurrent contributions from impurity phases, such as CuO, iii) un-optimized film thickness or iv) the higher bandgap energy (vide supra) which decreases the available light-harvesting range, Figure S16   SPV measurements, they are undoped. It is known that the performance of undoped metal oxide photoelectrodes is often limited by poor conductivity. This may be due to a low concentration of majority carriers, [5,49] but more often, it is caused by slow charge transport due to the formation of small polarons. While doping with shallow donors (for n-type semiconductors) or shallow acceptors (for p-type semiconductors) can increase the carrier concentration, like in conventional (non-oxide) semiconductors, it may also enhance carrier transport by the formation of small polaron bands or by lowering the polaron hopping energy. [32,50,51] As an example, studies of the charge carrier transport and PEC performance of doped and undoped BiVO 4 single crystals and films showed remarkable improvements in conductivity and photocurrents with Mo or W doping concentrations as low as ≈0.2-2% (both Mo and W are shallow donors in BiVO 4 ). [52][53][54] We believe that the primary reasons for the lower photocurrents in our CuBi 2 O 4 films are low specific surface area, [55] lack of CuO impurities and limited, slow charge transport in the undoped films. However, we cannot completely exclude the influence of a higher bandgap and un-optimized thickness of the films. The maximum current density decrease due to a higher bandgap would be 40% (1.9 and 1.65 eV corresponds to theoretical photocurrent densities of 17.04 and 24.16 mA cm −2 respectively), see Figure S16 (Supporting Information). Furthermore, reports of CuBi 2 O 4 with thicknesses ranging from 100 to 600 nm, showed a maximum change in photocurrent by a factor of 3. [1,5] The absorbed photon-to-current efficiency (APCE) measured from both sides of two CuBi 2 O 4 photoelectrodes made by both methods is seen in Figure 8. The corresponding IPCE and absorption values are shown in Figure S17 (Supporting Information). The APCE onsets of both photoelectrodes were at 640 ± 10 nm (1.93 ± 0.03 eV), seen in the inset. By examining the bandgap energies and absorption of both photoelectrodes, it is shown that the CuBi 2 O 4 made by combined PLD and RTP is considerably better in converting photons of energies close to its absorption edge. At these wavelengths, the light is absorbed uniformly in the entire film, which indicates less bulk recombination in the PLD/RTP films than in the sprayed films. This bulk recombination is typically due to defects that are energetically located deep within the bandgap, where they can act as traps for photogenerated charge carriers. The sprayed CuBi 2 O 4 films show no photocurrent at wavelengths above 650 nm, despite having a 1.65 eV (750 nm) bandgap and substantial (47%) optical absorption at 650 nm. This indicates extensive charge recombination at wavelengths close to the absorption edge, i.e., throughout the entire film thickness. The predicted AM1.5 photocurrent densities (J AM1.5 ) at 0.6 V (vs RHE) were calculated from the IPCE data (see the Experimental Section), assuming the photocurrent depends linearly on the light intensity. [56] The predicted J AM1.5 values of the CuBi 2 O 4 made by combined PLD and RTP were ≈−0.12/−0.1 mA cm −2 (back/front illumination), compared to measured photocurrents obtained by LSVs of −0.03/−0.09 mA cm −2 (back/front illumination). The predicted J AM1.5 values for the sprayed CuBi 2 O 4 films were ≈−2.71/−1.68 mA cm −2 (back/front illumination), compared to measured values obtained by LSVs of −1.89/−1.68 mA cm −2 (back/front illumination).
For both photoelectrodes, the APCE shapes are similar and the J AM1. 5 and LSVs values measured from the front side were practically identical, which implies similar CuBi 2 O 4 /electrolyte interfaces. However, the LSV values for the backside illumination show a higher J AM1.5 , which is presumably due to higher recombination at the FTO/CuBi 2 O 4 interface, [9,11,12] which causes a deviation from a linear relationship. Additionally, the APCE shapes of the backside illumination are different, with a decline in APCE below 400 nm for the combined PLD and RTP photoelectrode. Different APCE behavior between the back and front illumination is known in metal oxide photoelectrodes in general, [57] and also specifically for CuBi 2 O 4 , [1,11] to be due to poor transport of one of the carriers, or differences in absorption, dependent on the illumination side. Determining the reason for the decrease in APCE below 400 nm for backside illumination of the PLD/RTP films is beyond the scope of this work. However, we suggest a few possible explanations: i) increased recombination at the CuBi 2 O 4 /FTO interface, dominant mainly below 400 nm (also supported by lower SPV amplitudes for back-illuminated samples above 2.5 eV (490 nm)), and ii) possible traces of Bi 2 O 3 in the film that influence the overall charge transfer processes in the photoelectrode, at energies above ≈2.8 eV.

Photo-Electrochemical Stability
The photo-electrochemical stability of bare CuBi 2 O 4 (without protection layers and catalyst) made by both methods was studied under an applied potential of 0.6 V (vs RHE). Figure 9 presents stability measurements of both photocathodes, performed for 5 h (without an electron scavenger). The CuBi 2 O 4 made by spray pyrolysis showed higher initial photocurrents, consistent with the data in Figure 7. However, after ≈5 min its performance decreases below that of the pure PLD/RTP-grown illumination decreased by 26%, versus 98% for the sprayed samples. After the measurements, both photoelectrodes showed very different degrees of degradation (inset Figure 9). For the PLD/RTP photoelectrodes, the exposed area has darkened, whereas the exposed area of the sprayed samples has completely dissolved. The back-illuminated PLD/RTP sample showed lower stability than the front illuminated sample. This is probably due to higher recombination at the CuBi 2 O 4 /FTO interface, supported by the large difference between the predicted J AM1.5 and measured photocurrent for back illumination, and lower SPV amplitude of back-illuminated films implicating less efficient charge separation. As degradation of photoelectrodes is mainly driven by interface phenomena, [32] A careful examination of the dark currents in the LSVs of both photoelectrodes can shed light on the reasons for the substantial differences in PEC stability. Figure S18 (Supporting Information) shows a magnified area in both LSVs, focused on the currents and transient behavior in the dark in the 0.5-0.7 V versus RHE potential range. The CuBi 2 O 4 by PLD/RTP shows low dark currents and a lack of transients. This behavior is typically attributed to an unexposed contact between the FTO and the electrolyte, and low concentration of surface electrons, which manifest good catalytic activity. The dense CuBi 2 O 4 film made of tightly packed particles with minimal pinholes in the film, and enhanced charge separation efficiency, can support this behavior. In contrast, the sprayed CuBi 2 O 4 showed an anodic overshoot when the light was turned off, which typically suggests low catalytic activity (consistent with our SPV and APCE observations, vide supra) and an exposed contact between the FTO and the electrolyte. Furthermore, the sprayed CuBi 2 O 4 films contain a CuO impurity (a photo-electrochemically unstable oxide), [10] which can further accelerate the corrosion of the photocathode. However, the exact nature of the degradation mechanism would probably be a combination of different factors and is beyond the scope of this work.

Conclusions
We have presented a new fabrication approach for phase-pure CuBi 2 O 4 photoelectrodes by combining PLD to sequentially deposit Bi 2 O 3 and CuO on FTO substrates and RTP to carry out the Bi 2 O 3 + CuO → CuBi 2 O 4 reaction. Compact, singlephase CuBi 2 O 4 films were obtained after 10 min at 650 °C, demonstrating a viable approach to fabricate complex metal oxides thin film photoelectrodes at higher temperatures than the thermal stability of the FTO substrates would typically allow. A comparative study with conventional furnace heating revealed the significance of radiative heating in the processing of pure complex metal oxide photoelectrodes. The PLD/RTPmade photoelectrodes exhibited superior electronic properties and enhanced charge separation over CuBi 2 O 4 photoelectrodes made by spray-pyrolysis. Light-modulated surface photovoltage measurements of the PLD/RTP films determined bandgap energy of 1.9 ± 0.05 eV, compared with a CuBi 2 O 4 single crystal (1.9 eV ± 0.02 eV). In contrast, the bandgap energy of the CuBi 2 O 4 films by spray-pyrolysis was 1.65 ± 0.05 eV. The sprayed CuBi 2 O 4 films show a higher defect concentration, consistent with larger tail energies (150 meV vs 90 meV for the PLD/RTP films) and a (photoactive) CuO impurity found in the porous films, which also explains their poor stability. The photocurrent densities of the PLD/RTP films were lower than previously reported record photocurrent values of CuBi 2 O 4 . We have suggested low specific surface area, the lack of the (photoactive) CuO impurity phase, and slow charge transport of the undoped films as the primary reasons. However, the influence of other parameters such as higher bandgap of the pure phase, and unoptimized thickness of the films, cannot be excluded entirely. Furthermore, bare CuBi 2 O 4 by PLD/RTP photoelectrodes exhibited improved photo-electrochemical stability. Although CuBi 2 O 4 suffers from an intrinsic photo-electrochemical instability, the PLD/RTP films showed a stability decrease of only 26% after 5 h, which represents the highest stability reported to date. The noteworthy improvement of stability can lead to a better understanding of the corrosion mechanism of CuBi 2 O 4 , potentially shedding light on improved designs to protect it and increase its durability. To further improve the current densities and performances of CuBi 2 O 4 photoelectrodes made by PLD/ RTP, employing effective strategies such as nanostructuring, doping, and the use of co-catalysts is required. Moreover, a better understanding of the formation mechanism of CuBi 2 O 4 , primarily determining the role of Bi 2 O 3 in the reaction. If the suggested strategies are successfully implemented, combining CuBi 2 O 4 with a suitable bottom absorber can potentially result in a solar water-splitting device with high efficiencies.
Finally, complex metal oxides with a high degree of crystallinity and good electronic properties require a high thermal budget (the product of process temperature and processing time at an elevated temperature), [58] which is difficult to apply due to the low thermal stability of common TCO substrates. Therefore, extending the development and design of fabrication  methods by PLD and RTP to other complex metal-oxides photoelectrodes (e.g., α-SnWO 4 , CuFeO 2 ) is imperative to further progressing these materials as efficient photoabsorbers for solar water splitting.

Experimental Section
Films Deposition: Bi 2 O 3 and CuO films were deposited using a PLD system (PREVAC, Poland) by ablating from commercial targets of Bi 2 O 3 and CuO (99.99%, MaTeck) with a KrF-Excimer laser (248 nm, LPXPro 210, COHERENT). All films were deposited at a substrate temperature of 500 °C. The laser fluence was 1.5 J cm −2 and the repetition rate was 10 Hz. An oxygen background pressure (PO 2 ) of 1 × 10 −2 mbar was used and the target-to-substrate distance was 60 mm. Throughout the study, the thickness of the Bi 2 O 3 films was kept constant. The CuO thicknesses were adjusted experimentally to control the desired stoichiometry, which was calculated from the film thicknesses and the bulk densities of the materials (8.65, 8.9, and 6.31 g cm −3 for CuBi 2 O 4 , Bi 2 O 3 , and CuO, respectively). The substrates were FTO-coated glass (TEC 7, Pilkington) or quartz (Spectrosil 2000, Baumbauch & Co LTD), cleaned in a 1 vol% Triton solution (Triton X-100, Sigma-Aldrich), deionized water, and ethanol (≥99.99%, Sigma-Aldrich) for 10 min in each solution. Photoelectrodes made by spray-pyrolysis were fabricated using a previously reported procedure. [9] A DEKTAK profilometer was used to determine the film thicknesses.
Crystal Growth: CuBi 2 O 4 crystals were grown by the floating zone technique in a four mirror optical image furnace (Crystal Systems Corp., Japan), in ambient air at a rate of 5 mm h −1 . The high-density feed rod (D = ≈6 mm, L = ≈7 cm) was prepared from a stoichiometric mixture of high purity powders of Bi 2 O 3 (99.99%) and CuO (99.995%) by a solidstate reaction.
Furnace Heating and Rapid Thermal Processing: Conventional furnace heating of the films was performed at 500 °C in air at a heating rate of 10 K min −1 . Rapid thermal processing was conducted using a Rapid Thermal Processor (model: AS-One 100, ANNEALSYS). All RTP treatments were performed under atmospheric pressure of pure oxygen, with a heating rate of 10 K s −1 . In a typical RTP procedure, a sample is placed on a SiC wafer, used as a susceptor. An optical pyrometer was used to monitor the temperature at the back of the SiC susceptor (i.e., the nonilluminated side), which was typically at 550 °C. The surface temperatures of the bare substrates and the thin films were monitored at 650 °C by a thermocouple, attached to the sample surface with an indium contact.
Characterization: X-ray diffraction measurements were performed using a Bruker D8 diffractometer with Cu Kα radiation at 40 mA and 40 kV. Measurements were carried out in a grazing incidence geometry (angle of incidence was 2°) with a step size of 0.04° and a step duration of 6 s. The data were normalized after removing the background, without additional data averaging or noise reduction. The short-range structure and vibration modes of the films were analyzed by Raman microscopy (Dilor micro LabRam, Horiba) with a laser excitation wavelength of 635 nm and a power of 0.6 mW at the objective (spot size: ≈1 µm in diameter). UV-vis measurements were performed using a PerkinElmer Lambda 950 spectrophotometer with an integrating sphere. Two types of measurements were used to determine the optical absorption of the films: i) by independently measuring the transmittance and reflectance, or ii) by placing the films inside the integrating sphere with an offset of ≈7.5° from the incident light, and measuring the transflectance (transmittance T + reflectance R). Scanning electron microscopy imaging was carried out at a Zeiss UltraPlus or at a LEO GEMINI 1530 scanning electron microscope operated at 7 or 5 kV acceleration voltage. For energy-dispersive X-ray analysis, an UltimExtreme EDX detector with AZtec acquisition and evaluation software by Oxford Instruments was used.
Modulated surface photovoltage spectra were measured in the configuration of a parallel plate capacitor (quartz cylinder partially coated with the SnO 2 :F electrode, mica sheet as an insulator), in an ambient atmosphere. [59] The SPV signal is defined as the change in the surface potential as a result of the illumination. Front illumination (light impinging the surface) was provided by a halogen lamp, coupled to a quartz prism monochromator (SPM2) and modulated at a frequency of 8 Hz by using an optical chopper. In-phase and 90° phase-shifted SPV signals were detected with a high-impedance buffer and a dual phase lock-in amplifier (EG&G 5210). The amplitude of the modulated SPV signal is defined as the square root of the sum of the squared in-phase and 90° phase-shifted SPV signals. [60] Photo-electrochemical measurements were performed in the three-electrode configuration under the control of a potentiostat (EG&G Princeton Applied Research 273A). The studied films were connected as the working electrode in a custom-designed PEC cell with a calibrated Ag/AgCl reference electrode (XR300, Radiometer Analytical, E Ag/AgCl = 0.199 V versus normal hydrogen electrode, NHE), and a platinum wire as the counter electrode. All the measured potentials were converted to the reversible hydrogen electrode scale using the Nernst equation. The illuminated area of the sample was 0.28 cm 2 , which is identical to the area exposed to the electrolyte. All measurements were performed in an aqueous 0.3 m K 2 SO 4 and 0.2 m phosphate buffer (pH 7). Measurements with H 2 O 2 as an electron scavenger were performed with 30% hydrogen peroxide (Merck Schuchardt oHG) that was mixed with the electrolyte in a 4:1 (electrolyte:H 2 O 2 ) volume ratio. A WACOM super solar simulator (Model WXS-50S-5H, class AAA), was used as the illumination source and calibrated to closely resemble the AM1.5 global spectrum at 100 mW cm −2 . Stability measurements were performed under chopped light at 0.6 V versus RHE, without H 2 O 2 .
For incident photon-to-current efficiency (IPCE) measurements, a LOT (LSH302) lamp and Acton Research monochromator (SP2150) were used. The incident-photon-to-current conversion efficiency (IPCE) values were calculated as follows where J p is the averaged photocurrent density, P is the power density of the incident light, and λ is the wavelength. The power density was calibrated through the electrolyte contained between two quartz windows for front side illumination, and through FTO glass for backside illumination. This means that the reported IPCE values are for the CuBi 2 O 4 film itself, and not for the entire CuBi 2 O 4 photocathode/PEC cell assembly. Absorbed photon-to-current efficiency values were calculated by dividing the IPCE through the absorptance of the measured photoelectrode (corrected for the absorptance of the FTO glass), according to Equation (3) APCE % IPCE % Absorptance The predicted AM1.5 photocurrent density J AM1.5 (mA cm −2 ) of the photoelectrodes was estimated by multiplying the IPCE values with the solar photon flux Φ AM1.5 (mA cm −2 nm −1 ) and integrating it over the measured wavelengths, according to Equation (4

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.