Evaluation of the Efficiency of Photoelectrochemical Activity Enhancement for the Nanostructured LaFeO3 Photocathode by Surface Passivation and Co-Catalyst Deposition

Perovskite-type lanthanum iron oxide, LaFeO3, is a promising photocathode material that can achieve water splitting under visible light. However, the performance of this photoelectrode material is limited by significant electron-hole recombination. In this work, we explore different strategies to optimize the activity of a nanostructured porous LaFeO3 film, which demonstrates enhanced photoelectrocatalytic activity due to the reduced diffusion length of the charge carriers. We found that surface passivation is not an efficient approach for enhancing the photoelectrochemical performance of LaFeO3, as it is sufficiently stable under photoelectrocatalytic conditions. Instead, the deposition of a Pt co-catalyst was shown to be essential for maximizing the photoelectrochemical activity both in hydrogen evolution and oxygen reduction reactions. Illumination-induced band edge unpinning was found to be a major challenge for the further development of LaFeO3 photocathodes for water-splitting applications.


Introduction
Visible light-absorbing transition metal oxides (TMO) have attracted attention as potential active materials for photoelectrochemical (PEC) water splitting and low-carbon fuel production [1,2]. Yet, the practical application of TMO materials in photoelectrocatalysis faces numerous challenges, such as low carrier mobility, high defect densities, short charge carrier lifetimes, charge trapping in polaronic states and modest electrocatalytic activity in multistep reactions of water electrolysis, carbon dioxide reduction or pollutant degradation [3][4][5]. On the other hand, oxide-based materials show promising stability in aqueous solutions, primarily those that are alkaline and neutral. This advantage favorably distinguishes them from III-V semiconductor photoelectrodes, which show high photoelectrocatalytic activity but are hardly stable under reaction conditions [6]. Among TMO materials, perovskites are particularly interesting due to the flexibility of their electronic and crystal structure and their chemical versatility [7,8]. Most of the TMO-based materials, which were extensively studied in photoelectrochemical processes, are photoanodes [1,[9][10][11][12][13][14][15][16] (TiO 2 , Fe 2 O 3 , BiVO 4 , SrTiO 3 ), while much less attention has been paid to metal oxide photocathodes [17][18][19], except for copper-based materials. Copper-based photocathodes [20], such as Cu 2 O and CuBi 2 O 4 , show attractive activity in photoelectrochemical processes, yet their chemical and electrochemical stability under operating conditions remains a significant challenge [21] since successful application requires surface modification with protective coatings [22].
Fe-based materials (LaFeO 3 [23][24][25][26], BiFeO 3 [27], and café 2 O 4 [28]) were reported to possess rather high stability, which ensures practical interest in this class of photoelectrode materials. However, these materials show quite low external quantum efficiency (EQE) for hydrogen evolution reaction (HER), and that motivates further research aimed at enhancing their PEC performance. p-type LaFeO 3 (LFO) is an attractive perovskite-structured photocathode material, which was reported to be sufficiently stable under the photoelectrolysis conditions in alkaline media, yet shows very modest EQE [17,[23][24][25][26][29][30][31][32][33]. Various methods of improving EQE have been proposed, such as doping to tune the electronic structure and affect the charge carrier mobilities [31,33,34], nanostructuring to reduce the diffusion length [23,25], and adding a buffer gold layer to improve charge collection efficiency and charge separation [32]. Still, no consensus currently exists on the optimal strategies of LFO surface modification to enhance the photoelectrochemical activity toward the electrolysis of water.
Moderately high photocurrents (several hundreds of µA cm −2 ) on LFO films were observed only in experiments with electron scavengers [23,30,34], while the currents for hydrogen evolution are typically much lower. These experimental results imply that the main problem might be rooted in the necessary extension of the lifetime of short-lived photogenerated charges to drive the sluggish water reduction reaction. In electrocatalysis, this problem is solved by selecting highly active electrocatalytic materials, which interact with adsorbates optimally and decrease the activation energy of the limiting step in a multistep reaction. Co-catalyst deposition is a standard approach to enhance the performance of photoelectrodes [35,36], yet most of the data were reported for n-type semiconducting photoanodes [37]. Recent studies of the co-catalyst-decorated photoanodes revealed that the role of the co-catalyst in enhancing the PEC performance of a photoelectrode may differ from the expected increase in the charge transfer rate across the photoelectrode/solution interface [37]. For instance, the deposition of a CoPO 4 co-catalyst on the surface of hematite (α-Fe 2 O 3 ) did not lead to the enhancement of charge transfer kinetics, while the improvement in activity was mainly due to the reduced recombination losses [38,39]. A similar study confirmed that catalytic properties of the deposited layer might not be important for the PEC activity of photoanodes, as surface passivation and hence reduced recombination are the predominant sources of PEC activity improvement [38][39][40]. It has also been suggested that a co-catalyst can play both a catalytic role and a non-catalytic one, enhancing not only charge transfer but also the overall stability [40,41], while the predominant effect should be dependent on the nature of the photoelectrode material, the nature of the co-catalyst, as well as on the thickness and uniformity of the co-catalyst layer. It is always problematic to deduce the actual role of the co-catalyst, as one needs to address the competition between charge transfer and electron-hole recombination and understand which of these two factors is altered to a greater extent by the co-catalyst deposition [40].
There are far fewer detailed studies [42][43][44] on the co-catalyst effect for the metal oxide photocathodes, which also require co-catalysts to be efficient in multistep protoncoupled electron transfer reactions. Yet, it is of primary importance to formulate optimal strategies for the PEC performance enhancement of promising p-type transition metal oxide materials. In this study, we aim to compare the effects of surface passivation and co-catalyst deposition for a nanostructured LFO photocathode to deduce efficient strategies to control and enhance HER kinetics under photoelectrochemical conditions.

LaFeO 3 Films
Films of LFO on fluorine-doped tin oxide (FTO) substrates were fabricated using a modified procedure from ref. [21]. Briefly, 0.2 g of La(NO 3 ) 3 *6H 2 O, 0.19 g Fe(NO 3 ) 3 *9H 2 O and 0.38 g of citric acid monohydrate were dissolved in 0.5 mL of deionized water under magnetic stirring at room temperature. After the complete dissolution of the salts, 1 mL of Triton X-100 polymer and 1 mL of acetylacetone was added. The resulting mixture was stirred overnight. The solution was spin coated at 3000 rpm for 30 s onto FTO substrates, which were cleaned in acetylacetone to improve wetting with the sol components. The film was annealed at 500 • C for 20 min to remove the organic components, and the spin-coating procedure was repeated 3 times to obtain a film with an optimal thickness (ca. 300 nm). Finally, the film was annealed at 600 • C for 2 h to obtain the well-crystallized perovskite phase.

Pt Nanoparticles
Citrate-capped Pt nanoparticles were synthesized via a conventional borohydrate reduction route [45]. A total of 1 mL of 16 mM solution of H 2 PtCl 6 and 1 mL of 40 mM citric acid solution were mixed with 38 mL of deionized water and stirred for 30 min at room temperature. Then, 0.2 mL of a 50 mM NaBH 4 solution was added dropwise to the mixture, and the color of the solution changed to brownish yellow. The mixture was stirred at ambient temperature for 1 h. The resulting solution was spin coated onto FTO and LFO films for 30 s at 1000 rpm.

TiO 2 Layers
TiO 2 layers were deposited onto LFO films via spin coating. Titanium (IV) isopropoxide was mixed with acetylacetone in molar ratios of 1:8 and 1:100 and stirred overnight at room temperature. The resulting sol was spin coated on FTO or FTO/LFO films at 3000 rpm for 30 s. The films were air dried and then annealed at 250 • C to decompose the residual titanium (IV) isopropoxide.
Scanning electron microscopy (SEM) images were registered using an FEI Scios dualbeam scanning electron microscope (field emission gun, landing energy 2 kV, in-lens secondary electrons detector). The specimen cross-section was carried out using a focused ion beam (Ga + , 30 kV). To prevent unwanted specimen etching, a Pt protection layer was deposited before cross-sectioning. Transmission electron images were collected in bright-field mode using an FEI Tecnai Osiris transmission electron microscope operated at 200 kV.
Attenuated total reflectance Fourier-transform infrared (ATR-FTIR) spectra (resolution 4 cm −1 , signal averaging by 20 scans) of the films were measured with a Bruker Alpha II spectrometer equipped with a diamond ATR crystal and a KBr beamsplitter.
UV/Vis/NIR absorption spectra were measured in transmission geometry using a custom setup built with Avantes instruments: an AvaLight-DHc light source (both deuterium and halogen lamps), an AvaSpec-HS2048 spectrometer and fiber-optic light guides. The spectrum of the FTO glass substrate was used as a reference.
X-ray photoelectron spectroscopy (XPS) experiments were performed using a PHI 500 VersaProbe II spectrometer with a spherical mirror analyzer. An Al Kα monochromatic x-ray source with 1486.6 eV X-ray energy was utilized. Survey and high-resolution spectra were recorded with 1.0 eV and 0.1 eV step sizes, respectively. High-resolution XPS spectra were processed to obtain atomic concentrations following a typical procedure for XPS spectra quantifications. Photoelectron backgrounds were subtracted from the high-resolution spectra using Shirley function approximation. The binding energy (BE) of all the spectra were calibrated using the C1s peak from adventitious carbon fixed at 284.8 eV.

Photoelectrochemical Measurements
The photoelectrochemical properties of pristine LFO, LFO/Pt, and LFO/TiO 2 electrodes were characterized by means of voltammetric and chronoamperometric measurements. The FTO substrates were masked with a non-transparent epoxy resin to expose the area of ca. 0.5-1 cm 2 . All the measurements were performed in a 0.1 M NaOH solu-tion in a PTFE cell with a quartz window and separated graphite counter electrode and FTO/LaFeO 3 working electrode compartments. HgO/Hg (1 M NaOH) was used as a reference electrode. For the chopped voltammetry and chronoamperometry measurements, a high-power blue LED (ARPL-STAR-3W, λ max = 460 nm) was used as a light source. The sample was illuminated from the front at a light intensity of 45 mW/cm 2 at 460 nm (as measured by a calibrated silicone photodiode (S121C, Thorlabs) connected to a power meter (PM100A, Thorlabs)). Potentials are reported with respect to the reversible hydrogen electrode (RHE). Prior to the measurements in 0.1 M NaOH, the solution was deaerated with argon for a minimum of 40 min. For the measurements in O 2 -saturated 0.1 M NaOH solution, the solution was purged with O 2 for 40 min. Electrochemical impedance spectra were recorded under illumination in the frequency range of 100 kHz-10 mHz with a 5 mV alternating potential amplitude.

Characterization of the LFO Films
Transparent LFO films were synthesized on FTO supports via spin coating of metal oxide precursors combined with Triton X-100 surfactant, which creates pores in the film upon decomposing [25]. The XRD pattern of the LFO film is given in Figure 1a. Except for the peaks from the substrate (FTO, marked with asterisks), the only phase present is the orthorhombic modification of LaFeO 3 [46]. Characteristic peaks at 600 and 440 cm −1 , corresponding to Fe-O stretching and O-Fe-O bending vibrations, were observed in the ATR-FTIR spectra (Figure 1b) of the prepared LFO films, which additionally proves the formation of the perovskite structure [47]. The synthesized LFO films demonstrate a porous morphology due to the polymer template ( Figure 1c). The film is composed of grains with dimensions within the range of 50-150 nm, while the film thickness amounts to ca. 600 nm (as determined from an SEM image of the specimen cross-section, Figure S1). For LFO and similar materials, nanostructuring is considered essential to increase the active surface area and to reduce the diffusion length of charge carriers, which minimizes bulk recombination losses [23,48]. Figure S2 shows the XPS survey spectrum taken from the LFO film. The elemental composition given in Table S1 agrees well with the expected stoichiometry for LFO. No Sn3d line from tin in FTO supports was detected in the XPS spectra, which confirms the integrity of the film. The spectra in the O1s region reveal two well-separated peaks ( Figure S3a). The peak (a) centered at 529.2 eV can be attributed to the lattice oxygen in the perovskite crystal structure [49,50]. The wide peak (b) at higher binding energy (531.3 eV) is a well-known feature of various perovskite XPS spectra, indicating the presence of a weakly bound oxygen, although its exact assignment to a certain oxygen-containing species is controversial [50,51].
The optical properties of the film were characterized by UV-vis spectroscopic measurements ( Figure 1d shows the absorption spectrum of LFO). Optical absorption spectra of the prepared LaFeO 3 photoelectrodes show an intense broad band with a maximum near 380 nm and a long low-absorbance tail at λ > 500 nm. It is worth noting that measured spectra are sufficiently affected by the fringes due to complex interference patterns in two thin films (FTO and LFO), subsequently covering the glass substrate. In order to avoid uncertainties (which may arise due to the measured absorption spectrum being distorted by interference fringes) in the optical band gap determination, the smooth fringe-devoid transmission curve T α was calculated and then used in the Tauc plotting. T 0 in Figure 1d is the experimentally measured transmission curve. T M and T m are the envelopes, and splines are constructed as follows: in the region of weak or medium absorption (λ > 480 nm)following the maxima (T M ) or minima (T m ) of the observed interference fringes; in the region of strong absorption (λ < 480 nm)-following the T 0 curve. T α is the fringe-devoid transmission curve, calculated as T α = (T M ·T m ) 1/2 . The Tauc plot in the insert of Figure 1d is based on the T α curve, converted into absorbance units and then baseline-corrected (linear baseline). The band gap energy E g was calculated to be 2.7 eV, which is close to the earlier estimates [26,31,34]. The optical properties of the film were characterized by UV-vis spectroscopic measurements ( Figure 1d shows the absorption spectrum of LFO). Optical absorption spectra of the prepared LaFeO3 photoelectrodes show an intense broad band with a maximum near 380 nm and a long low-absorbance tail at λ > 500 nm. It is worth noting that measured spectra are sufficiently affected by the fringes due to complex interference patterns in two thin films (FTO and LFO), subsequently covering the glass substrate. In order to avoid uncertainties (which may arise due to the measured absorption spectrum being distorted by interference fringes) in the optical band gap determination, the smooth fringe-devoid transmission curve T⍺ was calculated and then used in the Tauc plotting. T0 in Figure 1d is the experimentally measured transmission curve. TM and Tm are the envelopes, and splines are constructed as follows: in the region of weak or medium absorption (λ > 480 nm)-following the maxima (TM) or minima (Tm) of the observed interference fringes; in the region of strong absorption (λ < 480 nm)-following the T0 curve. T⍺ is the fringe- Mott-Schottky plots of the LFO film are displayed in Figure 2a. The negative slope confirms the p-type nature of the LaFeO 3 material. The slope shows frequency dependence, which is typical for nanostructured films. From the linear segment of the plots, a flat band potential of 1.36 V was determined, which agrees well with the previous results for the non-doped LFO [23,31,34]. A cyclic voltammogram of a LaFeO 3 film registered under dark conditions in 0.1 M NaOH is shown in Figure 2b, and the voltammogram of bare FTO support is given for comparison. LFO voltammograms show an increase in the capacitive current in the potential range 0.5-1.2 V, and the onset of oxygen evolution is observed at potentials higher than 1.3 V. A broad peak is observed at 1.0-1.05 V, which can be attributed to sub-bandgap surface states, which are linked to intrinsic defects such as cation vacancies [31]. One may also notice cathodic currents appearing at potentials more negative than 0.6 V vs. RHE for the FTO electrode, which refer to the irreversible reduction of tin oxide in alkaline media [52]. For the LFO film, no pronounced asymmetry in the cathodic and anodic charges is observed until ca. 0.4 V vs. RHE, although the reduction of the perovskite phase in alkaline media should occur at lower potentials. The limited cathodic stability of perovskite structures under alkaline conditions was previously revealed for Mn-based perovskites at potentials lower than 0.4 V vs. RHE [53]. served at potentials higher than 1.3 V. A broad peak is observed at 1.0-1.05 V, which can be attributed to sub-bandgap surface states, which are linked to intrinsic defects such as cation vacancies [31]. One may also notice cathodic currents appearing at potentials more negative than 0.6 V vs. RHE for the FTO electrode, which refer to the irreversible reduction of tin oxide in alkaline media [52]. For the LFO film, no pronounced asymmetry in the cathodic and anodic charges is observed until ca. 0.4 V vs. RHE, although the reduction of the perovskite phase in alkaline media should occur at lower potentials. The limited cathodic stability of perovskite structures under alkaline conditions was previously revealed for Mn-based perovskites at potentials lower than 0.4 V vs. RHE [53].  RHE does not exceed ca. 10 µA cm −2 , which reflects the low PEC activity of the pristine LFO film. Typical current decays after instantaneous rises in the illumination periods are indicative of rate-controlling recombination processes [54] (the inserts in Figure 3 show the enlarged portions of the voltammograms during a 2 s illumination period to illustrate the current decays). Apparently, these losses become more pronounced with decreasing potential, which shows that charge separation due to the increase in the overpotential does not suppress recombination, as would be expected for classical semiconducting electrodes.

Photoelectrochemical Properties
To check the effect of surface passivation and suppression of surface recombination on the PEC activity enhancement for LFO, TiO2 layers were deposited on the surface of LFO from solutions with 1:8 (sample LFO/TiO2(1:8)) and 1:100 (sample LFO/TiO2(1:100)) titanium isopropoxide/acetylacetone molar ratios. An SEM image of the LFO/TiO2(1:100) electrode shows practically unaltered morphology of the film, without the presence of large TiO2 particles, which points to the uniformity of the passivating film ( Figure S4). Figure S3b shows the high-resolution XPS spectrum of the LFO/TiO2(1:100) film, which confirms the coverage of the film surface with titanium oxide. The Ti2p is a doublet, with Ti2p1/2 at 464.0 eV and the Ti2p3/2 peak at 458.2 eV. These peak positions agree well with the literature data on TiO2 [55,56].
The effect of TiO2 layers on blocking the electron transfer (ET) across the electrode/solution interface under dark conditions was examined by recording cyclic voltammograms of an FTO electrode with TiO2(1:8) and TiO2(1:100) coatings in a solution containing 0.5 M Na2SO4 and 10 mM Fe(CN)6 3as a reversible redox probe (Figure 4a). One may notice that the TiO2(1:8) layer provides complete blocking of the ET rate to FTO, while TiO2(1:100)  . The photocurrent at 0.5 V vs. RHE does not exceed ca. 10 µA cm −2 , which reflects the low PEC activity of the pristine LFO film. Typical current decays after instantaneous rises in the illumination periods are indicative of rate-controlling recombination processes [54] (the inserts in Figure 3 show the enlarged portions of the voltammograms during a 2 s illumination period to illustrate the current decays). Apparently, these losses become more pronounced with decreasing potential, which shows that charge separation due to the increase in the overpotential does not suppress recombination, as would be expected for classical semiconducting electrodes.

Photoelectrochemical Properties
To check the effect of surface passivation and suppression of surface recombination on the PEC activity enhancement for LFO, TiO 2 layers were deposited on the surface of LFO from solutions with 1:8 (sample LFO/TiO 2 (1:8)) and 1:100 (sample LFO/TiO 2 (1:100)) titanium isopropoxide/acetylacetone molar ratios. An SEM image of the LFO/TiO 2 (1:100) electrode shows practically unaltered morphology of the film, without the presence of large TiO 2 particles, which points to the uniformity of the passivating film ( Figure S4). Figure  S3b shows the high-resolution XPS spectrum of the LFO/TiO 2 (1:100) film, which confirms the coverage of the film surface with titanium oxide. The Ti2p is a doublet, with Ti2p 1/2 at 464.0 eV and the Ti2p 3/2 peak at 458.2 eV. These peak positions agree well with the literature data on TiO 2 [55,56].
The effect of TiO 2 layers on blocking the electron transfer (ET) across the electrode/solution interface under dark conditions was examined by recording cyclic voltammograms of an FTO electrode with TiO 2 (1:8) and TiO 2 (1:100) coatings in a solution containing 0.5 M Na 2 SO 4 and 10 mM Fe(CN) 6 3− as a reversible redox probe (Figure 4a). One may notice that the TiO 2 (1:8) layer provides complete blocking of the ET rate to FTO, while TiO 2 (1:100) coating strongly reduces the ET rate but does not stop the tunneling completely.          Notably, the current at 0.5 V vs. RHE is only moderately diminished, which proves that the TiO 2 (1:100) coating does not suppress the HER. To prove that the TiO 2 layer does not decrease the transmittance and the number of photogenerated charge carriers, we compared the voltammograms under front-side and back-side illumination, and the difference in current responses was found to be minor ( Figure S5). The onset of photocathodic current is close for the pristine and TiO 2 -modified films (ca. 1.2 V), yet for the LFO electrode with the TiO 2 layer, the photocurrent transients reveal spikes upon both illumination and light interruption until ca. 0.8 V. This is a signature of trapping of the photogenerated electrons at the surface, which are then consumed by fast recombination [57]. Figure S6 shows that the blocking TiO 2 (1:8) coating results in an even more pronounced surface trapping in the 0.8-1.2 V potential range and a 5-fold decrease in the photocurrent. Based on these experiments, we conclude that protective TiO 2 layers do not enhance the PEC activity of LFO, and passivation of the recombination centers at the LFO surface should be regarded as an inefficient photocathode modification strategy. Since the TiO 2 coating also suppresses the ET to the exposed FTO support, where the backward reaction may occur, we can conclude that the PEC activity of pristine LFO is not significantly affected by the possible backward hydrogen oxidation reaction (as was found to be the case for the BiVO 4 photoanode [58]).
Next, we explored the effect of a Pt co-catalyst on the PEC activity of LFO electrodes. The LFO surface was modified with citrate-stabilized 5 nm-sized Pt nanoparticles, which were synthesized via a typical borohydride reduction route. XPS spectrum confirmed the presence of Pt nanoparticles on the surface of the LFO/Pt electrode ( Figure S3c). The Pt4f region features a doublet with Pt4f 5/2 at 74.5 eV and Pt4f 7/2 at 71.2 eV, indicating that platinum particles are in the metallic state [59,60]. In addition, a prominent tail towards higher binding energies in the spectrum is observed. The shifting of the peaks to higher binding energies was observed previously for platinum nanoparticles of various sizes and is connected with the partial oxidation of the catalyst surface [61]. The deconvolution of the Pt4f spectrum has been performed to estimate shares of Pt 2+ and Pt 4+ . The result of the Pt4f spectrum fitting with Pt4f 7/2 /Pt4f 5/2 doublets is presented in Figure S7. The Pt 4f 7/2 -4f 5/2 spin orbit splitting is 3.33 eV. The doublet with Pt4f 7/2 located at 71.2eV is attributed to metallic platinum nanoparticles Pt(0). Two additional doublets with Pt4f 7/2 binding energy located at 72.8 eV and 74.6 eV correspond to Pt 2+ and Pt 4+ states, respectively [62,63].
When Pt nanoparticles were deposited onto bare FTO supports, the catalytic effect in the dark manifested itself in a shift of the HER onset potential by ca. 0.5 V (Figure 4b).
Notably, for the Pt-modified FTO electrode, the FTO reduction wave also shifts from ca. 0.2 V to 0.7 V, which points to the catalytic effect of Pt on the FTO reduction. The modification of the LFO surface with Pt nanoparticles results in a very noticeable increase in the photocurrent (ca. 30 µA cm −2 ), which is a threefold improvement over the unmodified LFO ( Figure 3c). As one can see, the photocurrent decays are also more prominent, showing that the HER kinetics facilitated by Pt are still too slow to suppress surface recombination completely, and the photocurrent decreases rapidly from its instantaneous values. EIS data in Figure S8a confirm the observed trend-the highest resistance is observed for a TiO 2 -modified electrode, while the LFO/Pt sample shows the lowest resistance, which correlates with the currents in the voltammograms in Figure 3a-c.
For the most active LFO/Pt sample, a stability test was carried out ( Figure 5). Despite the initial current decay, no further diminution of the photocurrent density or growth of the dark currents were observed in the course of 2 h measurements under chopped illumination conditions. This can be regarded as the confirmation of the stability of the LFO phase under the reaction conditions and the assignment of the observed photocurrent to the hydrogen evolution [26,64] (and possibly reduction of traces of oxygen) rather than to the LFO photocorrosion current.

Recombination and Charge Transfer Kinetics
Potentiostatic photocurrent transients were recorded to quantify the losses. Figure 6a shows characteristic decaying transients registered at diffe for a pristine LFO electrode. The decay of photocurrent with time reflects t photogenerated electrons at or near the interface, which causes a counter and provokes recombination. A steady state is achieved when the rate at w trons reach the interface is balanced by the rate at which they are consum transfer and recombination. For LFO/Pt electrodes, the steady state pho crease, yet the current decays also become higher, which may suggest an in the rate of charge transfer and recombination (Figure 6b). For the LFO/T trode (Figure 6c), the photocurrent transients are much closer to simple re off" responses, which can be regarded as a consequence of low charge sepa PEC activity. The current decays appear only at potentials more negative t Analysis of the photocurrent transients can be used to extract the valu nation rate constant (krec) and charge transfer rate constant (kct) [54]: where ( ) is the time-dependent photocurrent density, t is the time, neous photocurrent density, and is the steady-state photocurrent densi the decrease in the photocurrent decay and the increase in the steady-stat would signify reduced recombination. Notably, the procedure for recombin stant determination from photocurrent transients is only valid at short longer times, the deviations related to the build-up of minority charge carr face appear, which might alter band bending. Figure S9 shows examples o tion of the ln ( ) vs. time plots, which were used to derive informatio

Recombination and Charge Transfer Kinetics
Potentiostatic photocurrent transients were recorded to quantify the recombination losses. Figure 6a shows characteristic decaying transients registered at different potentials for a pristine LFO electrode. The decay of photocurrent with time reflects the build-up of photogenerated electrons at or near the interface, which causes a counter-flow of holes and provokes recombination. A steady state is achieved when the rate at which the electrons reach the interface is balanced by the rate at which they are consumed by charge transfer and recombination. For LFO/Pt electrodes, the steady state photocurrents increase, yet the current decays also become higher, which may suggest an increase in both the rate of charge transfer and recombination (Figure 6b). For the LFO/TiO 2 (1:100) electrode (Figure 6c), the photocurrent transients are much closer to simple rectangular "on-off" responses, which can be regarded as a consequence of low charge separation and low PEC activity. The current decays appear only at potentials more negative than 0.5 V.  Figure 7 shows the recombination and charge transfer rate constants (Table S2 collects the values of rate constants). On average, the charge transfer rate constant for LFO/Pt is ca. 5 times higher than that for LFO without the Pt nanoparticles, which shows that the co-catalyst is essential to enhance the charge transfer rate. The recombination rate constants are expected to decrease with the increase in the overpotential due to better charge Analysis of the photocurrent transients can be used to extract the values of recombination rate constant (k rec ) and charge transfer rate constant (k ct ) [54]: where i(t) is the time-dependent photocurrent density, t is the time, i 0 is the instantaneous photocurrent density, and i ss is the steady-state photocurrent density. In this case, the decrease in the photocurrent decay and the increase in the steady-state photocurrent would signify reduced recombination. Notably, the procedure for recombination rate constant determination from photocurrent transients is only valid at short times since, at longer times, the deviations related to the build-up of minority charge carriers at the surface appear, which might alter band bending. Figure S9 shows examples of the linearization of the ln vs. time plots, which were used to derive information on the values of k tr and k rec . Figure 7 shows the recombination and charge transfer rate constants (Table S2 collects the values of rate constants). On average, the charge transfer rate constant for LFO/Pt is ca. 5 times higher than that for LFO without the Pt nanoparticles, which shows that the co-catalyst is essential to enhance the charge transfer rate. The recombination rate constants are expected to decrease with the increase in the overpotential due to better charge separation associated with the changes in band bending [40]. However, for the LFO and LFO/Pt electrodes, the opposite trend is observed-the recombination rate constants show exponential growth with a change in potential from 1.0 V to 0.3 V vs. RHE. Effectively, this means that the charge transfer efficiency decreases with the increase in the overpotential. Recombination rate constants are ca. 7 times higher for LFO/Pt, which shows that the co-catalyst does not reduce recombination, although it enhances the charge transfer. For LFO/TiO 2 , the charge transfer rate constant decreases by a factor of 2 compared to pristine LFO, while the recombination rate constants drop by an order of magnitude, which is the expected behavior for a passivating coating. Notably, at potentials less positive than 0.5 V, both k rec and k tr for pristine LFO and LFO/Pt increase with the decrease in potential, which might be related to the changes in the potential drop across the Helmholtz layer (compact part of the double layer [65]).  Figure 7 shows the recombination and charge transfer rate constants (Table S2 co lects the values of rate constants). On average, the charge transfer rate constant for LFO/P is ca. 5 times higher than that for LFO without the Pt nanoparticles, which shows that th co-catalyst is essential to enhance the charge transfer rate. The recombination rate con stants are expected to decrease with the increase in the overpotential due to better charg separation associated with the changes in band bending [40]. However, for the LFO an LFO/Pt electrodes, the opposite trend is observed-the recombination rate constants show exponential growth with a change in potential from 1.0 V to 0.3 V vs. RHE. Effectively this means that the charge transfer efficiency decreases with the increase in the overpo tential. Recombination rate constants are ca. 7 times higher for LFO/Pt, which shows tha the co-catalyst does not reduce recombination, although it enhances the charge transfe For LFO/TiO2, the charge transfer rate constant decreases by a factor of 2 compared t pristine LFO, while the recombination rate constants drop by an order of magnitude which is the expected behavior for a passivating coating. Notably, at potentials less pos tive than 0.5 V, both krec and ktr for pristine LFO and LFO/Pt increase with the decrease i potential, which might be related to the changes in the potential drop across the Helm holtz layer (compact part of the double layer [65]). In the next step, we explored the kinetics of photoelectrochemical processes on mod ified LFO electrodes for a reactant with more facile charge transfer kinetics. For LaFeO based photocathodes, dissolved oxygen is often used as an electron scavenger, with th ET to O2 proceeding at a higher rate compared to ET to water molecules in alkaline medi [23,31,34]. For pristine LFO, the current for O2 reduction is ca. 3.5 times higher than th In the next step, we explored the kinetics of photoelectrochemical processes on modified LFO electrodes for a reactant with more facile charge transfer kinetics. For LaFeO 3based photocathodes, dissolved oxygen is often used as an electron scavenger, with the ET to O 2 proceeding at a higher rate compared to ET to water molecules in alkaline media [23,31,34]. For pristine LFO, the current for O 2 reduction is ca. 3.5 times higher than the current for water reduction (Figure 3d), while for TiO 2 -covered LFO, the increase only amounts to 1.5 times (Figure 3e), which implies a reduction of the ET rate due to passivation. For the LFO/Pt electrodes, the currents are enhanced by a factor of four (Figure 3f). No typical signs of surface recombination can be observed in the voltammograms for O 2 reduction; the currents are steadily growing during the illumination periods. These results show that compared to the case of water reduction, the surface recombination is minor. Figure S10 shows characteristic photocurrent transients, registered in O 2 -saturated 0.1 M NaOH solution for LFO/Pt electrodes, with stable photocurrents without the current spikes and subsequent decays (small current decays appear only at potentials less positive than 0.4 V vs. RHE, which is a significant improvement over the trend observed for water reduction). EIS data in Figure S8b reveal characteristic semicircles in the Nyquist plots for LFO, LFO/TiO 2 and LFO/Pt, with the smallest diameter being observed for the LFO/Pt sample, which confirms much faster charge transfer kinetics for the LFO photoelectrode with a Pt co-catalyst.
To explore the effect of surface coating on the hydrogen evolution kinetics further, we analyzed the dark responses of the LFO films registered after switching off the illumination ( Figure 8). Notably, for the LFO and LFO/Pt electrodes, the differences in current responses at 0.35 and 0.55 V are minor (Figure 8a It is expected that the decrease in cathodic photocurrent due to recombination up to the point where the light is switched off should be equal in magnitude to the instantaneous anodic spike, which appears due to the flow of holes to recombine with the electrons accumulated in the surface region [66]. For the LFO and LFO/Pt electrodes, the cathodic photocurrent decays after the initial spike, but there is only a small overshoot. For LFO/TiO 2 , the overshoot is practically invisible. The discrepancy between the magnitudes of cathodic current decay and anodic current overshoot can be explained by the light-induced band edge unpinning due to the build-up of electrons at the interface, which modifies the potential distribution across the LFO/electrolyte interface [54,57,66]. The potential drop across the Helmholtz layer (∆φ H , Figure 8d) increases at the expense of decreasing the potential drop across the space charge region, and the decreased band bending causes decay of the photocurrent. For LaFeO 3, the observed effect could also correspond to the reduction of surface Fe(III) states to Fe(II) states, and recombination would then be attributed to the reoxidation of Fe(II) by holes. However, since the long-term tests imply the stability of the steady-state photocurrent, we assume that such a reduction only involves the surface atoms.

Discussion
The observed trends allow us to speculate on the strategies to overcome the limitations in the use of LFO electrodes for PEC applications that hinder the advancement of this material toward commercial PEC devices. We found that the activity of a nanoporous

Discussion
The observed trends allow us to speculate on the strategies to overcome the limitations in the use of LFO electrodes for PEC applications that hinder the advancement of this material toward commercial PEC devices. We found that the activity of a nanoporous LFO film with a Pt co-catalyst exceeds that reported for the doped LaFeO 3 films, both in HER and ORR (Table S3 compares our results with the literature data). Ca. 50-75 µA cm −2 at 0.5 V vs. RHE were achieved in O 2 -saturated alkaline solutions for Zn-, Mg-, Ca-, Sror Ba-doped LFO films [31,34], while in our study, the photocurrents for the non-doped LFO with a Pt co-catalyst at 0.5 V exceed 100 µA cm −2 . In deaerated alkaline solutions, the stationary photocurrents for the doped LFO films still do not exceed several µA cm −2 , while for the LFO/Pt film, tens of µA cm −2 were obtained in this study. Much higher photocurrent densities with oxygen as an electron scavenger (up to 200 µA cm −2 ) were reported only for the LaFeO 3 film prepared by electrodeposition, yet these were thick and non-transparent deposits with a surface area much larger than that characteristic for transparent LFO film synthesized in this work [23,33]. Moreover, the photocurrents in deoxygenated alkaline solutions for such electrodeposited films did not exceed 10 µA cm −2 , which is comparable with the performance of the obtained LFO film. It can be expected that the PEC activity of the electrodeposited LFO will also improve significantly with the addition of a co-catalyst.
The results of previous studies [34] indicate that doping of the LFO structure does not lead to a noticeable increase in the charge transfer rate. The effect of doping is primarily associated with an increase in majority carrier concentration resulting from a dopantinduced increase in the Fe 4+ density caused by charge compensation upon substitution of Fe 3+ by divalent ions and, possibly, with the increase in the hole mobility [34]. We suggest that further development of LaFeO 3 -based photocathodes should combine cocatalyst deposition with the doping strategy to reduce bulk recombination by facilitating electron-hole separation.

Conclusions
The photoelectrochemical properties of a nanostructured LaFeO 3 film were demonstrated to depend greatly on the presence of a passivating coating or a co-catalyst. We found that although LFO films show quite poor activity in HER, the photocurrents can be increased by a factor of three, reaching 30 µA cm −2 if Pt nanoparticles are deposited onto the film surface. A protective TiO 2 layer does not provide any increase in the PEC activity due to surface passivation. As the LFO electrodes exhibit high stability under illumination in aqueous alkaline media, the additional surface passivation seems to be unnecessary, while the addition of a co-catalyst was found to be essential to enhance both the HER and ORR kinetics. ORR currents increased by a factor of 4 when a Pt co-catalyst was deposited onto the film, which demonstrated the importance of increasing the rate of interfacial charge transfer for LFO photocathodes. For slower HER, the build-up of electrons at the interfaces translates into band edge unpinning, which reduces the photocurrent. Our findings suggest that the nanostructured LFO photoelectrode does not require surface protection for the development of this material for PEC water-splitting applications, while the utilization of a co-catalyst is essential for enhancing the PEC activity.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.