Surface-Modiﬁed Ta 3 N 5 Photoanodes for Sunlight-Driven Overall Water Splitting by Photoelectrochemical Cells

: The development of visible-light-responsive semiconductor-based photoelectrodes is a prerequisite for the construction of efﬁcient photoelectrochemical (PEC) cells for solar water splitting. Surface modiﬁcation with an electrocatalyst on the photoelectrode is effective for maximizing the water splitting efﬁciency of the PEC cell. Herein, we investigate the effects of surface modiﬁcation of Ta 3 N 5 photoanodes with electrocatalysts consisting of Ni, Fe, and Co oxides, and their mixture, on the PEC oxygen evolution reaction (OER) performance. Among the investigated samples, NiFeO x -modiﬁed Ta 3 N 5 (NiFeO x /Ta 3 N 5 ) photoanodes showed the lowest onset potential for OER. A PEC cell with a parallel conﬁguration consisting of a NiFeO x /Ta 3 N 5 photoanode and an Al-doped La 5 Ti 2 Cu 0.9 Ag 0.1 S 5 O 7 (LTCA:Al) photocathode exhibited stoichiometric hydrogen and oxygen generation from water splitting, without any external bias voltage. The solar-to-hydrogen energy conversion efﬁciency (STH) of this cell for water splitting was found to be 0.2% at 1 min after the start of the reaction. In addition, water splitting by a PEC cell with a tandem conﬁguration incorporating a NiFeO x /Ta 3 N 5 transparent photoanode prepared on a quartz insulating substrate as a front-side electrode and a LTCA:Al photocathode as a back side electrode was demonstrated, and the STH was found to be 0.04% at the initial stage of the reaction.


Introduction
The conversion of solar energy into hydrogen through the photoelectrochemical (PEC) water splitting reaction using semiconductor photoelectrodes is seen as one of the most promising ways to produce chemical fuels that can be stored and are transportable [1][2][3][4][5][6]. One approach to PEC water splitting involves the use of a PEC cell that combines a photocathode for the hydrogen evolution reaction (HER) and a photoanode for the oxygen evolution reaction (OER) [1,[7][8][9]. This PEC cell can relax the requirements for water splitting, such as the positions of the conduction band minimum and valence band maximum of the individual photoelectrodes [8,9], through a two-step light excitation process. This process makes it possible to drive a spontaneous overall water splitting reaction under simulated sunlight illumination without the application of an external bias voltage.
Two different configurations of PEC cells, consisting of a photoanode-photocathode combination, have been proposed: One is a parallel configuration (parallel PEC cell) where the photoelectrodes are placed side-by-side [8][9][10][11]. The other is a tandem configuration (tandem PEC cell) that has a stacked structure, composed of a transparent photoelectrode on the front side, and a second photoelectrode capable of responding to the light transmitted through the front-side photoelectrode [7,[12][13][14]. Schematic illustrations of the PEC cells are given in Figure S1 in the supporting information (SI). Tandem PEC cells have the potential to achieve high solar-to-hydrogen energy conversion efficiency (STH) values, because the unit area of light illumination is the area of the front-side photoelectrode, leading to potentially increased STH values per unit area. The STH and the working potential (E work ) for the PEC cell are determined by photocurrent matching, from the intersection of the current-potential (i-E) curves for the photoanode and photocathode [1,7]. Although overall water splitting by PEC cells has been examined, parallel PEC cells have shown insufficient STH levels-typically less than 2% [8][9][10][11]15]. In the case of tandem PEC cells, higher STH values have been reported, but still typically less than 4% [7,[12][13][14]16]. Huang et al. demonstrated an STH of 3.17% using tandem PEC cells consisting of a BiVO 4 -based photoanode and a Cu 2 ZnSnS 4 -based photocathode [14]. In addition, Kobayashi et al. reported an STH of 3.7% using a tandem PEC cell composed of a BiVO 4 -based photoanode and a CuIn 0.5 Ga 0.5 Se 2 (CIGS)-based photocathode [13]. Ta 3 N 5 is considered among the most promising semiconductor photoanodes for PEC oxygen evolution because its band structure is well suited to water splitting [17][18][19]. A Ta 3 N 5 -based photoanode is expected to generate a photocurrent density of 12.9 mA cm −2 based on its bandgap of~2.1 eV under simulated sunlight, assuming a quantum efficiency of 100% [20]. The theoretical limiting photocurrent density corresponds to an STH value of 15.9% under 1 sun illumination. Despite the promising properties of Ta 3 N 5 , instability and a positive onset potential in PEC oxygen evolution are problems to be solved. Surface modification with an electrocatalyst for the OER and insertion of a protective layer have been shown to be effective strategies for enhancing the PEC oxygen evolution efficiency of Ta 3 N 5 . Liu et al. demonstrated a photocurrent density of 12.1 mA cm −2 at 1.23 V vs. the reversible hydrogen electrode (RHE) under simulated sunlight [20]. This is close to the theoretical maximum photocurrent value for Ta 3 N 5 . The authors employed Ta 3 N 5based photoanodes modified with a hierarchical structure consisting of a TiO x electronblocking layer, FeOOH/Ni(OH) 2 hole-storage layers, and an Ir-Co molecular catalyst for the OER [20]. Nevertheless, the onset potential for PEC oxygen evolution on high-quality Ta 3 N 5 was found to be approximately 0.65 V vs. the RHE. This means that application of an external bias voltage is essential to drive PEC water splitting on Ta 3 N 5 -based photoanodes.
The intrinsic semiconducting properties of Ta 3 N 5 associated with light absorption, extinction via electron-hole pair separation, carrier diffusion, and carrier transport are crucial parameters that affect water splitting efficiency [21]. The crystallinity of Ta 3 N 5 is reduced by both bulk and surface defects, which cause the formation of carrier recombination sites and surface states [22][23][24][25][26][27]. In addition, at the Ta 3 N 5 /electrolyte solution interface, the catalytic efficiency of the electrocatalyst and the mass transfer of reactants should also be determined in order to clarify the overall efficiency of water splitting. To date, several electrocatalysts for the OER-such as Co(OH) x [28][29][30], Co 3 O 4 [31,32], CoPi [33][34][35], Ni(OH) x [36], and a mixed metal oxide (Ni-Fe-Co) [37,38]-have been used to decorate the surface of Ta 3 N 5 , leading to enhanced PEC oxygen evolution efficiency. In these studies, the magnitude of the photocurrent generated at 1.23 V vs. the RHE has mainly been discussed; however, onset potentials have been found around 0.6-0.8 V vs. the RHE [39][40][41]. Nurlaela et al. demonstrated that the modification with the better OER electrocatalyst improves photocatalytic activity more drastically, in a comparative study of both electrocatalytic and photocatalytic testing using a Ta 3 N 5 photocatalyst system [42]. This implies that the onset potential for PEC oxygen evolution on Ta 3 N 5 strictly reflects the electrochemical properties of the electrocatalyst that is decorated on the semiconductor surface. Investigations to clarify the impact of the electrocatalyst on PEC oxygen evolution are expected to provide us with strategies to improve the water splitting efficiency, leading to the development of a more efficient PEC cell.
To accomplish overall water splitting with a PEC cell consisting of a photoanodephotocathode combination, photocurrent values at E work of the PEC cell are quite important, especially in the vicinity of the onset potential of the photoelectrodes. In the present study, we focused on the surface modification of Ta 3 N 5 photoanodes as a means of enhancing the efficiency of overall water splitting in the PEC cell. Electrocatalysts such as mono-metal oxides and mixed metal oxides consisting of Ni, Fe, and Co species were loaded onto the surface of Ta 3 N 5 photoanodes in order to observe their effect on PEC properties, such as the photocurrent density and onset potential. First, Ta 3 N 5 photoanodes prepared on Ta-metal substrates (Ta 3 N 5 /Ta) were utilized to investigate the effect of different electrocatalysts on PEC oxygen evolution. The electrocatalysts were loaded onto the surface of the Ta 3 N 5 through the drop-casting of a non-aqueous solution containing metal-organic precursors, followed by heating in air [43][44][45]. This surface modification method is relatively simple compared to other modification methods, such as (photo-assisted) electrodeposition. Based on the resulting data, the present study considers overall water splitting using a parallel PEC cell composed of a Ta 3 N 5 /Ta photoanode and an Al-doped La 5 Ti 2 Cu 0.9 Ag 0.1 S 5 O 7 (LTCA:Al) particulate-based photocathode, which has an absorption edge wavelength of 710 nm [9,11]. Next, a transparent Ta 3 N 5 photoanode prepared on a double-sided polished quartz insulating substrate (Ta 3 N 5 /SiO 2 ) was employed in order to construct the tandem PEC cell. Ta 3 N 5 /SiO 2 can generate a PEC oxygen evolution via the intrinsic conductivity of the Ta 3 N 5 itself, even without using a transparent conductive substrate [43]. The present study considers overall water splitting using a tandem PEC cell composed of a Ta 3 N 5 /SiO 2 transparent photoanode as the front-side electrode and a LTCA:Al photocathode as the back-side electrode. Figure 1 summarizes the photocurrent-potential (i-E) curves for Ta 3 N 5 /Ta photoanodes modified with various electrocatalysts (NiFeCoO x , NiFeO x , CoO x , NiO x , and FeO x ) for the OER. Under simulated sunlight illumination, Ta 3 N 5 /Ta photoanodes produced an anodic photocurrent regardless of the composition of the electrocatalyst. NiFeCoO xand NiFeO x -modified Ta 3 N 5 /Ta photoanodes (NiFeCoO x /Ta 3 N 5 /Ta in Figure 1a and NiFeO x /Ta 3 N 5 /Ta in Figure 1b) clearly exhibited an anodic photocurrent at electrode potentials above 0.6 V vs. the RHE. The photocurrent generated by the NiFeCoO x /Ta 3 N 5 /Ta and NiFeO x /Ta 3 N 5 /Ta photoanodes increased with increasing electrode potential. At electrode potentials below 0.6 V vs. the RHE, these photoanodes did not exhibit any appreciable photocurrent. The photocurrent densities generated by NiFeCoO x /Ta 3 N 5 /Ta and NiFeO x /Ta 3 N 5 /Ta at 1.23 V vs. the RHE were 4.5 mA cm −2 and 4.0 mA cm −2 , respectively. On the other hand, CoO x -and NiO x -modified Ta 3 N 5 /Ta photoanodes (CoO x /Ta 3 N 5 /Ta in Figure 1c and NiO x /Ta 3 N 5 /Ta in Figure 1d) also exhibited an anodic photocurrent at electrode potentials above 0.7 V and 0.8 V vs. the RHE, respectively. For comparison, the previously reported CoP i -modified Ta 3 N 5 /Ta (CoP i /Ta 3 N 5 /Ta) [34] photoanodes also exhibited an anodic photocurrent at electrode potentials above 0.7 V vs. the RHE, in good accordance with the onset potential for the CoO x /Ta 3 N 5 /Ta photoanode. The photocurrent density generated by the CoO x /Ta 3 N 5 /Ta photoanode was found to be 4.5 mA cm −2 at 1.23 V vs. the RHE, which is comparable to that for the NiFeCoO x /Ta 3 N 5 /Ta photoanode. Surface modification with NiFeCoOx and NiFeOx electrocatalysts significantly im proved the photocurrent at around 0.7 V vs. the RHE, which is the expected Ework valu for the PEC cell. The cathodic shift of the onset potential is consistent with the lower over potential for the mixed metal oxide catalysts (NiFeOx and NiFeCoOx) [42,46]. In contrast the photocurrent generated by the NiOx/Ta3N5/Ta photoanode increased linearly as function of photoelectrode potential, indicating that the hole-transfer efficiency for NiO at the solid/liquid interface is poor compared to that for the mixed metal oxide electrocat alysts (NiFeOx and NiFeCoOx). As shown in Figure 1e, the FeOx-modified Ta3N5/T (FeOx/Ta3N5/Ta) photoanode shows a low photocurrent of less than 0.1 mA cm −2 at 1.23 V vs. the RHE. Note that the i-E curve obtained for FeOx/Ta3N5/Ta photoanodes shows a low anodic photocurrent compared to that for a bare Ta3N5/Ta photoanode (see Figure S2 in the SI), indicating that the FeOx catalyst did not act effectively as an electrocatalyst for th OER under the present conditions. It should be noted that the visible-light absorption o FeOx in the present loading amount is negligible (see Figure S3 in the SI), indicating tha light blocking by the FeOx is not a cause of the poor PEC performance of the FeOx/Ta3N5/T photoanode. Figure 2 shows typical current-time (i-t) curves for NiFeCoOx/Ta3N5/Ta The potential sweep rate (v) and intermittent irradiation period were 10 mV s −1 and 6 s, respectively.

PEC Characteristics of Surface-Modified Ta 3 N 5 Photoanodes
Surface modification with NiFeCoO x and NiFeO x electrocatalysts significantly improved the photocurrent at around 0.7 V vs. the RHE, which is the expected E work value for the PEC cell. The cathodic shift of the onset potential is consistent with the lower overpotential for the mixed metal oxide catalysts (NiFeO x and NiFeCoO x ) [42,46]. In contrast, the photocurrent generated by the NiO x /Ta 3 N 5 /Ta photoanode increased linearly as a function of photoelectrode potential, indicating that the hole-transfer efficiency for NiO x at the solid/liquid interface is poor compared to that for the mixed metal oxide electrocatalysts (NiFeO x and NiFeCoO x ). As shown in Figure 1e, the FeO x -modified Ta 3 N 5 /Ta (FeO x /Ta 3 N 5 /Ta) photoanode shows a low photocurrent of less than 0.1 mA cm −2 at 1.23 V vs. the RHE. Note that the i-E curve obtained for FeO x /Ta 3 N 5 /Ta photoanodes shows a low anodic photocurrent compared to that for a bare Ta 3 N 5 /Ta photoanode (see Figure S2 in the SI), indicating that the FeO x catalyst did not act effectively as an electrocatalyst for the OER under the present conditions. It should be noted that the visible-light absorption of FeO x in the present loading amount is negligible (see Figure S3 in the SI), indicating that light blocking by the FeO x is not a cause of the poor PEC performance of the FeO x /Ta 3 N 5 /Ta photoanode. Figure 2 shows typical current-time (i-t) curves for NiFeCoO x /Ta 3 N 5 /Ta, NiFeO x /Ta 3 N 5 /Ta and CoO x /Ta 3 N 5 /Ta photoanodes recorded at 0.7 V vs. the RHE under simulated sunlight. Note that because NiO x /Ta 3 N 5 /Ta and FeO x /Ta 3 N 5 /Ta did not generate a meaningful anodic photocurrent at 0.7 V vs. the RHE, evaluation results for these photoanodes are not shown here. After the start of light illumination, a spike-like decrease in photocurrent was observed over a period of several seconds. This was at least partially attributable to double-layer charging by a non-Faradaic process. To accurately evaluate the PEC oxygen evolution, the photocurrent observed 1 min after the start of light illumination will be discussed. The NiFeCoO x /Ta 3 N 5 /Ta photoanode produced a photocurrent density of approximately 0.37 mA cm −2 at 1 min after the start of light illumination; however, the photocurrent decreased with time, and was 0.28 mA cm −2 after 5 min. On the other hand, the NiFeO x /Ta 3 N 5 /Ta photoanode showed an initial photocurrent density of 0.40 mA cm −2 , and maintained a value of 0.38 mA cm −2 after 5 min had passed. Haleem et al. reported that Ni-Fe-Co mixed metal oxide-modified Ta 3 N 5 photoanodes showed a stable photocurrent of 4 mA cm −2 at 1.23 V vs. the RHE for over 2 h [37]. The difference in current stability is most likely caused by the different preparation protocol used in that study, which adopted photo-electrodeposition for the loading of Ni-Fe-Co oxide on Ta 3 N 5 . CoO x /Ta 3 N 5 /Ta photoanodes showed an anodic photocurrent of 0.03 mA cm −2 , which is clearly lower than the values for NiFeCoO x /Ta 3 N 5 /Ta and NiFeO x /Ta 3 N 5 /Ta photoanodes. Judging from the i-E and i-t characteristics of the surface-modified Ta 3 N 5 /Ta photoanodes, the NiFeO x /Ta 3 N 5 /Ta photoanode appears promising for the construction of PEC cells for overall water splitting. Therefore, the NiFeO x /Ta 3 N 5 /Ta photoanodes were selected for further study. a meaningful anodic photocurrent at 0.7 V vs. the RHE, evaluation results for these photoanodes are not shown here. After the start of light illumination, a spike-like decrease in photocurrent was observed over a period of several seconds. This was at least partially attributable to double-layer charging by a non-Faradaic process. To accurately evaluate the PEC oxygen evolution, the photocurrent observed 1 min after the start of light illumination will be discussed. The NiFeCoOx/Ta3N5/Ta photoanode produced a photocurrent density of approximately 0.37 mA cm −2 at 1 min after the start of light illumination; however, the photocurrent decreased with time, and was 0.28 mA cm −2 after 5 min. On the other hand, the NiFeOx/Ta3N5/Ta photoanode showed an initial photocurrent density of 0.40 mA cm −2 , and maintained a value of 0.38 mA cm −2 after 5 min had passed. Haleem et al. reported that Ni-Fe-Co mixed metal oxide-modified Ta3N5 photoanodes showed a stable photocurrent of 4 mA cm −2 at 1.23 V vs. the RHE for over 2 h [37]. The difference in current stability is most likely caused by the different preparation protocol used in that study, which adopted photo-electrodeposition for the loading of Ni-Fe-Co oxide on Ta3N5. CoOx/Ta3N5/Ta photoanodes showed an anodic photocurrent of 0.03 mA cm −2 , which is clearly lower than the values for NiFeCoOx/Ta3N5/Ta and NiFeOx/Ta3N5/Ta photoanodes. Judging from the i-E and i-t characteristics of the surface-modified Ta3N5/Ta photoanodes, the NiFeOx/Ta3N5/Ta photoanode appears promising for the construction of PEC cells for overall water splitting. Therefore, the NiFeOx/Ta3N5/Ta photoanodes were selected for further study.  Figure 3 shows the wavelength dependence of the incident photon-to-current conversion efficiency (IPCE) for NiFeOx/Ta3N5/Ta photoanodes recorded at 0.7, 0.9, and 1.1 V vs. the RHE. The IPCE spectra demonstrate that NiFeOx/Ta3N5/Ta photoanodes can generate a photocurrent under light illumination up to a wavelength of 600 nm, which corresponds to the absorption edge wavelength for Ta3N5. Generally, the IPCE increased as the wavelength decreased and the electrode potential became more positive, because of the shorter light penetration depth and the wider depletion layer at the solid/liquid interface. For illumination with 420 nm monochromatic light, the IPCEs at 0.7, 0.9, and 1.1 V vs. the RHE were found to be 8%, 32%, and 52%, respectively. Intriguingly, the IPCE values at 0.7 V and 1.1 V were comparable at the wavelength of 580 nm, indicating that most of the photocurrent originated from energy conversion utilizing visible light in the short wave-  Figure 3 shows the wavelength dependence of the incident photon-to-current conversion efficiency (IPCE) for NiFeO x /Ta 3 N 5 /Ta photoanodes recorded at 0.7, 0.9, and 1.1 V vs. the RHE. The IPCE spectra demonstrate that NiFeO x /Ta 3 N 5 /Ta photoanodes can generate a photocurrent under light illumination up to a wavelength of 600 nm, which corresponds to the absorption edge wavelength for Ta 3 N 5 . Generally, the IPCE increased as the wavelength decreased and the electrode potential became more positive, because of the shorter light penetration depth and the wider depletion layer at the solid/liquid interface. For illumination with 420 nm monochromatic light, the IPCEs at 0.7, 0.9, and 1.1 V vs. the RHE were found to be 8%, 32%, and 52%, respectively. Intriguingly, the IPCE values at 0.7 V and 1.1 V were comparable at the wavelength of 580 nm, indicating that most of the photocurrent originated from energy conversion utilizing visible light in the short wavelength region. As shown in Figure S4 in the SI, SEM images revealed NiFeOx deposited on the Ta3N5 surface, while part of the exposed surface of Ta3N5 was clearly observed, indicating that a boundary was formed between the NiFeOx-covered part and the exposed part of the Ta3N5. SEM energy-dispersive X-ray spectroscopic analysis showed that the elements of Ni, Fe, and O attributable to NiFeOx were distributed on the Ta3N5, but the exposed Ta3N5 not covered by NiFeOx was also observed (see Figure S5 in the SI). These results suggest that the exposed Ta3N5 causes the decrease in photocurrent with time, partially due to the self-oxidation of Ta3N5 under light illumination [31,43,44]. XPS spectra revealed that the Ta3N5 surface was covered with NiFeOx, as Ni and Fe peaks were observed, unlike in the case of the XPS spectra of bare Ta3N5/Ta (see Figure S6 in the SI). The XPS data also indicated that NiFeOx covers the Ta3N5 surface, since the Ta and N peak intensities weakened after NiFeOx modification. Nevertheless, the XPS peaks due to N and Ta were clearly observed even after NiFeOx modification. This indicates that the Ta3N5 surface is partially exposed, in accordance with the SEM results.

Overall Water Splitting by Parallel PEC Cell
Since the NiFeOx/Ta3N5/Ta photoanode can generate an appreciable anodic photocurrent at 0.7 V vs. the RHE, it is possible to achieve spontaneous overall water splitting without the application of an external bias voltage, using only an appropriate photocathode with an onset potential more positive than 0.7 V vs. the RHE. In the present study, NiFeOx/Ta3N5/Ta and Al-doped La5Ti2Cu0.9Ag0.1S5O7 (LTCA:Al) were employed as the photoanode and photocathode for a parallel PEC cell (see Figure S7 in the SI). The LTCA:Al photocathode surface was modified with a Pt catalyst in order to promote the HER, and then sequentially modified with a TiO2 protection layer and a CdS layer in order to promote charge separation through the formation of p-n junctions. This was done because the LTCA:Al modified with Pt, TiO2, and CdS (Pt/TiO2/CdS/LTCA:Al) photocathode is capable of driving PEC hydrogen evolution with an onset potential of 0.9 V vs. the RHE in a highly alkaline aqueous solution in a stable manner. The details of the preparation and the PEC properties of the surface-modified LTCA:Al photocathode have been reported elsewhere [9,11]. Figure 4a shows the i-E curves for a NiFeOx/Ta3N5/Ta photoanode with a geometrical As shown in Figure S4 in the SI, SEM images revealed NiFeO x deposited on the Ta 3 N 5 surface, while part of the exposed surface of Ta 3 N 5 was clearly observed, indicating that a boundary was formed between the NiFeO x -covered part and the exposed part of the Ta 3 N 5 . SEM energy-dispersive X-ray spectroscopic analysis showed that the elements of Ni, Fe, and O attributable to NiFeO x were distributed on the Ta 3 N 5 , but the exposed Ta 3 N 5 not covered by NiFeO x was also observed (see Figure S5 in the SI). These results suggest that the exposed Ta 3 N 5 causes the decrease in photocurrent with time, partially due to the self-oxidation of Ta 3 N 5 under light illumination [31,43,44]. XPS spectra revealed that the Ta 3 N 5 surface was covered with NiFeO x , as Ni and Fe peaks were observed, unlike in the case of the XPS spectra of bare Ta 3 N 5 /Ta (see Figure S6 in the SI). The XPS data also indicated that NiFeO x covers the Ta 3 N 5 surface, since the Ta and N peak intensities weakened after NiFeO x modification. Nevertheless, the XPS peaks due to N and Ta were clearly observed even after NiFeO x modification. This indicates that the Ta 3 N 5 surface is partially exposed, in accordance with the SEM results.

Overall Water Splitting by Parallel PEC Cell
Since the NiFeO x /Ta 3 N 5 /Ta photoanode can generate an appreciable anodic photocurrent at 0.7 V vs. the RHE, it is possible to achieve spontaneous overall water splitting without the application of an external bias voltage, using only an appropriate photocathode with an onset potential more positive than 0.7 V vs. the RHE. In the present study, NiFeO x /Ta 3 N 5 /Ta and Al-doped La 5 Ti 2 Cu 0.9 Ag 0.1 S 5 O 7 (LTCA:Al) were employed as the photoanode and photocathode for a parallel PEC cell (see Figure S7 in the SI). The LTCA:Al photocathode surface was modified with a Pt catalyst in order to promote the HER, and then sequentially modified with a TiO 2 protection layer and a CdS layer in order to promote charge separation through the formation of p-n junctions. This was done because the LTCA:Al modified with Pt, TiO 2, and CdS (Pt/TiO 2 /CdS/LTCA:Al) photocathode is capable of driving PEC hydrogen evolution with an onset potential of 0.9 V vs. the RHE in a highly alkaline aqueous solution in a stable manner. The details of the preparation and the PEC properties of the surface-modified LTCA:Al photocathode have been reported elsewhere [9,11]. Figure 4a shows the i-E curves for a NiFeO x /Ta 3 N 5 /Ta photoanode with a geometrical electrode area of 0.7 cm 2 and an LTCA:Al photocathode with a geometrical electrode area of 0.6 cm 2 under simulated sunlight. In principle, the E work and STH for the PEC cell should be determined by current matching; thus, the operational photocurrent and the E work for this cell were expected to be 0.22 mA and 0.68 V vs. the RHE, respectively. It should be noted that the total geometrical electrode area (1.30 cm 2 ) for both photoelectrodes must be considered when calculating the photocurrent density and STH values.   Figure 4b shows the time course of the photocurrent density for the non-biased parallel PEC cell. The photocurrent density at 1 min after starting the reaction was found to be 0.17 mA cm −2 , which corresponds to an STH of 0.2%. Although the photocurrent density for the PEC cell gradually decreased with time, the photocurrent and STH remained 0.08 mA cm −2 and 0.1%, respectively, even after 6 h of illumination. The Ework value for the PEC cell 1 min after initiating reaction was found to be 0.67 V vs. the RHE, as shown in Figure S8 in the SI. The Ework value gradually shifted to more positive potentials over time, reaching 0.7 V vs. the RHE after 6 h. This shift in Ewok during the water splitting reaction suggests that the decrease in photocurrent for the PEC cell is attributable to degradation of the NiFeOx/Ta3N5/Ta photoanode rather than degradation of the Pt/TiO2/CdS/LTCA:Al photocathode. The detachment of the NiFeOx electrocatalyst on the Ta3N5 surface is one possible reason for the gradual decrease in the photocurrent of the Ta3N5/Ta photoanode. The photocurrent density at around 0.7 V vs. the RHE decreased significantly after the  Figure 4b shows the time course of the photocurrent density for the non-biased parallel PEC cell. The photocurrent density at 1 min after starting the reaction was found to be 0.17 mA cm −2 , which corresponds to an STH of 0.2%. Although the photocurrent density for the PEC cell gradually decreased with time, the photocurrent and STH remained 0.08 mA cm −2 and 0.1%, respectively, even after 6 h of illumination. The E work value for the PEC cell 1 min after initiating reaction was found to be 0.67 V vs. the RHE, as shown in Figure S8 in the SI. The E work value gradually shifted to more positive potentials over time, reaching 0.7 V vs. the RHE after 6 h. This shift in E wok during the water splitting reaction suggests that the decrease in photocurrent for the PEC cell is attributable to degradation of the NiFeO x /Ta 3 N 5 /Ta photoanode rather than degradation of the Pt/TiO 2 /CdS/LTCA:Al photocathode. The detachment of the NiFeO x electrocatalyst on the Ta 3 N 5 surface is one possible reason for the gradual decrease in the photocurrent of the Ta 3 N 5 /Ta photoanode. The photocurrent density at around 0.7 V vs. the RHE decreased significantly after the water splitting reaction (see Figure S9 in the SI). A partial removal of NiFeO x from the photoanode surface, giving rise to a decrease in the surface coverage of the NiFeO x , was revealed by the results of the SEM observation after the water splitting reaction (see Figure  S10 in the SI).
Hydrogen and oxygen were generated by the parallel PEC cell at a stoichiometric ratio of 2:1 over 6 h, as shown in Figure 4c. The amounts of evolved hydrogen and oxygen were consistent with the values calculated from the total charge calculated from the recorded photocurrent, indicating that the Faradaic efficiency for overall water splitting for the cell is~97%. We note that the apparent delay in gas detection for the initial 1 h is due to the time required for homogenization of the gas phase in the measurement system. The average gas generation rates over the 6 h period were approximately 1.8 µmol cm −2 h −1 for hydrogen and 0.9 µmol cm −2 h −1 for oxygen. The hydrogen and oxygen generation rates were approximately 1.6 times higher than those previously reported for a parallel PEC cell composed of an Ir-Co catalyst-loaded BaTaO 2 N photoanode and a Pt/TiO 2 /CdS/LTCA:Al photocathode with average gas generation rates of 1.1 µmol cm −2 h −1 for hydrogen and 0.55 µmol cm −2 h −1 for oxygen [9]. The slightly higher STH for the parallel PEC cell incorporating the Ta 3 N 5 -LTCA:Al combination originates from the improved photocurrent for the PEC cells at E work .

Overall Water Splitting by Tandem PEC Cell
Next, overall water splitting by a tandem PEC cell using surface-modified Ta 3 N 5 photoanodes was assessed. In this trial, a transparent Ta 3 N 5 photoanode fabricated on an insulating quartz substrate (Ta 3 N 5 /SiO 2 ) was employed for the front-side photoanode in the cell. The optical characteristics of the Ta 3 N 5 /SiO 2 transparent photoanodes required tuning, because the water splitting efficiency of the cell is determined by the intensity and the wavelength of the light transmitted through the front-side photoanode. As shown in Figure S11 in the SI, as the thickness of the Ta 3 N 5 films increased, the color changed from pale orange to dark red. Figure 5a shows the UV-vis transmission spectra of the Ta 3 N 5 /SiO 2 photoanodes with different thicknesses of Ta 3 N 5 . In the film thickness range of 100-800 nm, these photoanodes showed an optical transmittance of over 60% at wavelengths longer than 600 nm. The fringes observed in the spectra are ascribable to optical interference. The 800 nm thick Ta 3 N 5 film absorbed most of the visible light at wavelengths up to 600 nm, which is the absorption edge wavelength for Ta 3 N 5 . A plot of the absorption coefficient for the Ta 3 N 5 /SiO 2 photoanode as a function of wavelength is shown in Figure S12a in the SI. The absorption coefficient for Ta 3 N 5 /SiO 2 appears to monotonically decrease with increasing wavelength. The values are in the order of 10 5~1 0 4 cm −1 in the range 350-600 nm. Ziani et al. reported absorption coefficients for Ta 3 N 5 between 8 × 10 5 and 1 × 10 4 cm −1 for wavelengths from 300 to 600 nm [27], which are in good agreement with the results of the present study. Figure S12b in the SI displays a plot of light intensity as a function of the optical path length in Ta 3 N 5 /SiO 2 . When Ta 3 N 5 /SiO 2 is irradiated with light with a wavelength shorter than 450 nm, the light can be completely absorbed by Ta 3 N 5 with a thickness of 300 nm. On the other hand, light with wavelengths of 500 and 550 nm is fully absorbed when the Ta 3 N 5 /SiO 2 is 800 nm thick. This shows that using Ta 3 N 5 with a thickness of 800 nm is preferable in order to absorb light at wavelengths relevant to the PEC oxygen evolution on Ta 3 N 5 /SiO 2 (350-600 nm). Figure S13 in the SI shows the UV-vis transmission spectra of the NiFeO x -modified Ta 3 N 5 /SiO 2 photoanodes with 800 nm-thick Ta 3 N 5 . Even after NiFeO x modifications, the optical transmittance at wavelengths longer than 600 nm remained over 60%, which can apply to the tandem PEC cell. Our previous study has demonstrated that an increase in the loading amount of NiFeO x decreases the transmittance of Ta 3 N 5 /SiO 2 photoanodes, while enhancing the OER performance of NiFeO x /Ta 3 N 5 /SiO 2 [47]. However, when the loading amount of NiFeO x is larger than 0.1 µmol cm −2 , the PEC performance decreases because of the decrease in the light intensity reaching the Ta 3 N 5 [47]. Therefore, Ta 3 N 5 /SiO 2 photoanodes with the loading amount of 0.03 µmol cm −2 , which is same amount of NiFeO x in Ta 3 N 5 /Ta, was employed in order to investigate the PEC performance on Ta 3 N 5 /SiO 2 photoanodes. ts 2021, 11, 584 9 of 15 0.03 µ mol cm −2 , which is same amount of NiFeOx in Ta3N5/Ta, was employed in order to investigate the PEC performance on Ta3N5/SiO2 photoanodes. Figure 5b presents typical i-E curves acquired for NiFeOx/Ta3N5/SiO2 photoanodes with 800 nm thick Ta3N5 under simulated sunlight illumination from the front side (Ta3N5) and from the back side (SiO2). Under light illumination from the front side, the photoanode shows a photocurrent density of 4.4 mA cm −2 at 1.23 V vs. the RHE. This value is consistent with that for previously reported NiFeOx-modified Ta3N5/SiO2 photoanodes [43,47]. The i-t curves for the NiFeOx/Ta3N5/SiO2 transparent photoanode recorded at 1.23 V vs. the RHE (see Figure S14 in the SI) demonstrate that the initial photocurrent density of 4.4 mA cm −2 generated by the photoanode was well consistent with photocurrent value on the i-E curves (Figure 5b). The photocurrent for the NiFeOx/Ta3N5/SiO2 at 1.23 V vs. the RHE gradually decreased to 70% over 30 min. This gradual decrease of the photocurrent is most likely due to the self-photo-oxidation of the Ta3N5 surface [47]. On the other hand, when the light irradiated the electrode from the back side, the photocurrent significantly decreased to 0.4 mA cm −2 at 1.23 V vs. the RHE. As shown in Figure S15 in the SI, the photocurrent at 1.23 V vs. the RHE tended to increase as the Ta3N5 film's thickness increased when irradiation was from the front side. The photocurrent at 0.7 V vs. the RHE which is the expected Ework of the PEC cell also increased with the increase in the Ta3N5 film's thickness. In contrast, when irradiation was from the back side, the photocurrent tended to decrease as the thickness of Ta3N5 increased. Light with short wavelengths, such as 350 nm and 400 nm, is completely absorbed within about 100 nm of the SiO2/Ta3N5 interface (see Figure S12b in the SI). This means that the generated carriers need to travel at least 700 nm to reach the Ta3N5/electrolyte interface. In addition, the holes generated by shorter wavelength light may not be transferred effectively to the Ta3N5/electrolyte interface due to carrier recombination, leading to a decrease in photocurrent. The onset potential, which is the emergence of the anodic photocurrent, clearly shifted from 0.6 V vs. the RHE to 1.0 V vs. the RHE when the illumination direction changed from the front side to the back side. This result indicates that the light needs to reach the depletion region of Ta3N5 in order to cause holes to accumulate at the electrified Ta3N5/electrolyte interface effectively.  and from the back side (SiO 2 ). Under light illumination from the front side, the photoanode shows a photocurrent density of 4.4 mA cm −2 at 1.23 V vs. the RHE. This value is consistent with that for previously reported NiFeO x -modified Ta 3 N 5 /SiO 2 photoanodes [43,47]. The i-t curves for the NiFeO x /Ta 3 N 5 /SiO 2 transparent photoanode recorded at 1.23 V vs. the RHE (see Figure S14 in the SI) demonstrate that the initial photocurrent density of 4.4 mA cm −2 generated by the photoanode was well consistent with photocurrent value on the i-E curves (Figure 5b). The photocurrent for the NiFeO x /Ta 3 N 5 /SiO 2 at 1.23 V vs. the RHE gradually decreased to 70% over 30 min. This gradual decrease of the photocurrent is most likely due to the self-photo-oxidation of the Ta 3 N 5 surface [47]. On the other hand, when the light irradiated the electrode from the back side, the photocurrent significantly decreased to 0.4 mA cm −2 at 1.23 V vs. the RHE. As shown in Figure S15 in the SI, the photocurrent at 1.23 V vs. the RHE tended to increase as the Ta 3 N 5 film's thickness increased when irradiation was from the front side. The photocurrent at 0.7 V vs. the RHE which is the expected E work of the PEC cell also increased with the increase in the Ta 3 N 5 film's thickness. In contrast, when irradiation was from the back side, the photocurrent tended to decrease as the thickness of Ta 3 N 5 increased. Light with short wavelengths, such as 350 nm and 400 nm, is completely absorbed within about 100 nm of the SiO 2 /Ta 3 N 5 interface (see Figure S12b in the SI). This means that the generated carriers need to travel at least 700 nm to reach the Ta 3 N 5 /electrolyte interface. In addition, the holes generated by shorter wavelength light may not be transferred effectively to the Ta 3 N 5 /electrolyte interface due to carrier recombination, leading to a decrease in photocurrent. The onset potential, which is the emergence of the anodic photocurrent, clearly shifted from 0.6 V vs. the RHE to 1.0 V vs. the RHE when the illumination direction changed from the front side to the back side. This result indicates that the light needs to reach the depletion region of Ta 3 N 5 in order to cause holes to accumulate at the electrified Ta 3 N 5 /electrolyte interface effectively.
Finally, the overall water splitting reaction using a tandem PEC cell composed of a combination of a NiFeO x /Ta 3 N 5 /SiO 2 photoanode on top of a Pt/TiO 2 /CdS/LTCA:Al photocathode is discussed. Based on the aforementioned discussion, a NiFeO x /Ta 3 N 5 /SiO 2 photoanode with a Ta 3 N 5 thickness of 800 nm was employed as the front-side transparent photoanode in the tandem PEC cell. The Pt/TiO 2 /CdS/LTCA:Al photocathode has an absorption edge wavelength of 710 nm; thus, this photocathode can exhibit the PEC hydrogen evolution reaction by utilizing the light transmitted through the transparent NiFeO x /Ta 3 N 5 /SiO 2 (see Figure S16 in the SI). Figure 6a shows i-E curves for the NiFeO x /Ta 3 N 5 /SiO 2 transparent photoanode and the Pt/TiO 2 /CdS/Al:LTCA photocathode. Note that the photocurrent density of Pt/TiO 2 /CdS/Al:LTCA photocathode was calculated by dividing the photocurrent value by the geometrical electrode area of the LTCA:Al photocathode (0.26 cm 2 ). On the other hand, the photocurrent density of the Pt/TiO 2 /CdS/LTCA:Al photocathode placed behind the NiFeO x /Ta 3 N 5 /SiO 2 transparent photoanode was calculated using the geometrical electrode area of Ta 3 N 5 (0.67 cm 2 ) of the front-side electrode. The cathodic photocurrent density at 0.7 V vs. the RHE generated by the Pt/TiO 2 /CdS/LTCA:Al was reduced to about 25% when it was placed behind the NiFeO x /Ta 3 N 5 /SiO 2 photoanode. The intersection of the photocurrent density and the expected E work for the tandem PEC cell were found to be 0.04 mA cm −2 and 0.64 V vs. the RHE, respectively. Finally, the overall water splitting reaction using a tandem PEC cell composed of a combination of a NiFeOx/Ta3N5/SiO2 photoanode on top of a Pt/TiO2/CdS/LTCA:Al photocathode is discussed. Based on the aforementioned discussion, a NiFeOx/Ta3N5/SiO2 photoanode with a Ta3N5 thickness of 800 nm was employed as the front-side transparent photoanode in the tandem PEC cell. The Pt/TiO2/CdS/LTCA:Al photocathode has an absorption edge wavelength of 710 nm; thus, this photocathode can exhibit the PEC hydrogen evolution reaction by utilizing the light transmitted through the transparent NiFeOx/Ta3N5/SiO2 (see Figure S16 in the SI). Figure 6a shows i-E curves for the NiFeOx/Ta3N5/SiO2 transparent photoanode and the Pt/TiO2/CdS/Al:LTCA photocathode. Note that the photocurrent density of Pt/TiO2/CdS/Al:LTCA photocathode was calculated by dividing the photocurrent value by the geometrical electrode area of the LTCA:Al photocathode (0.26 cm 2 ). On the other hand, the photocurrent density of the Pt/TiO2/CdS/LTCA:Al photocathode placed behind the NiFeOx/Ta3N5/SiO2 transparent photoanode was calculated using the geometrical electrode area of Ta3N5 (0.67 cm 2 ) of the front-side electrode. The cathodic photocurrent density at 0.7 V vs. the RHE generated by the Pt/TiO2/CdS/LTCA:Al was reduced to about 25% when it was placed behind the NiFeOx/Ta3N5/SiO2 photoanode. The intersection of the photocurrent density and the expected Ework for the tandem PEC cell were found to be 0.04 mA cm −2 and 0.64 V vs. the RHE, respectively.  Figure 6b shows the time course of the photocurrent generated by the non-biased tandem PEC cell. The tandem cell generates a photocurrent of 0.036 mA cm −2 under continuous light illumination, which is in relatively good agreement with the expected photocurrent estimated from the intersection of the i-E curves (see Figure 6a). The photocurrent density of 0.036 mA cm −2 corresponds to an STH of 0.04%, assuming that the Faradaic efficiency for water splitting is 100%. The photocurrent generated by the tandem PEC cell decreased over time, and the photocurrent after 10 min of reaction was found to be 0.02 mA cm −2 . Akagi et al. reported an STH of approximately 0.06% for overall water splitting by a tandem PEC cell composed of a Mg-doped Ta3N5 transparent photoanode fabricated on a conductive carbon nanotube-coated quartz substrate and a solid solution ZnSe-CIGS-based photocathode with an absorption edge wavelength of 900 nm [40]. Although the present results are comparable with those for the previously reported tandem PEC cell  Figure 6b shows the time course of the photocurrent generated by the non-biased tandem PEC cell. The tandem cell generates a photocurrent of 0.036 mA cm −2 under continuous light illumination, which is in relatively good agreement with the expected photocurrent estimated from the intersection of the i-E curves (see Figure 6a). The photocurrent density of 0.036 mA cm −2 corresponds to an STH of 0.04%, assuming that the Faradaic efficiency for water splitting is 100%. The photocurrent generated by the tandem PEC cell decreased over time, and the photocurrent after 10 min of reaction was found to be 0.02 mA cm −2 . Akagi et al. reported an STH of approximately 0.06% for overall water splitting by a tandem PEC cell composed of a Mg-doped Ta 3 N 5 transparent photoanode fabricated on a conductive carbon nanotube-coated quartz substrate and a solid solution ZnSe-CIGS-based photocathode with an absorption edge wavelength of 900 nm [40]. Al-though the present results are comparable with those for the previously reported tandem PEC cell using a transparent Ta 3 N 5 -based photoanode, the use of Ta 3 N 5 prepared directly on an insulating substrate made it possible to construct the current tandem PEC cell. Our results represent progress in the development of tandem PEC cells that combine a metal oxysulfide-based particulate photocathode and a nitride-based transparent photoanode prepared on an insulator. This will further expand the list of potential narrow bandgap materials applicable to the construction of tandem-type PEC cells.

Preparation of Ta 3 N 5 -Based Photoanodes
A Ta 3 N 5 thin film was prepared following previously reported methods [34]. An approximately 500 nm-thick TaO δ film was deposited on mirror-polished Ta substrates (10 × 10 × 0.3 mm) via radio frequency (RF) magnetron sputtering (ES-205L, Eiko, Tokyo, Japan). The sputtering target was a Ta metal disk (3N pure (99.9%), Kojundo Chem. Lab. Co., Saitama, Japan) 60 mm in diameter and 3 mm thick. The Ta target was pre-sputtered in an Ar gas atmosphere for 10 min in order to clean its surface. A gas purifier (VICI Metronics Inc., Poulsbo, Washington, USA) was connected to the incoming Ar line in order to eliminate impurities. The base pressure in the sputtering chamber was 3 × 10 −7 Pa. The precursor films were prepared at an RF power of 70 W for 2.5 h in Ar at a 4.0 Pa working pressure. The resulting films were subsequently heated at 1073 K for 1 h in a flow of NH 3 (200 sccm) with a temperature ramp of 10 K min −1 . In addition, Ta 3 N 5 thin films were prepared on double-side polished quartz (SiO 2 ) insulating substrates (10 × 10 × 0.5 mm) following a modified version of previously reported methods [43]. Ta-metal films with thicknesses of 75, 150, 300, 450, and 600 nm were sputtered at an RF power of 70 W in Ar at a working pressure of 3.0 Pa. The sputtering rate for the films under these conditions was 4.0 nm min −1 . These precursor films were heated in a flow consisting of a mixture of NH 3 and N 2 (NH 3 = 100 sccm and N 2 = 100 sccm) at 1073 K with a temperature ramp of 10 K min −1 . Precursor films with thicknesses of 75, 150, and 300 nm were heated at 1073 K for 0.5 h in order to produce Ta 3 N 5 films with thicknesses of 100, 200, and 400 nm, while films with thicknesses of 450 and 600 nm were heated at 1073 K for 1 h in order to produce the 600 nm and 800 nm thick Ta 3 N 5 .

Surface Modification of Ta 3 N 5 -Based Photoanodes
Metal oxide catalysts to promote the OER were deposited on the surface of Ta 3 N 5 . Deposition was accomplished by combining precursor solutions of iron(III) 2-ethylhexanoate (Wako Chemicals, Richmond, VA, USA), nickel(II) 2-ethlyhexanoate (Wako), and Co(III) 2-ethyl hexanoate (Wako) dissolved in hexane (HPLC grade, Wako). The precursor solutions of FeO x , NiO x , CoO x , NiFeO x , and NiFeCoO x were prepared as follow: For the mono-metal oxide catalysts (FeO x , NiO x , and CoO x ), 10 µL of metal complex was added to 10 mL of hexane, and then 40 µL of this solution was drop-cast on the surface of the Ta 3 N 5 surface, with a coverage area of approximately 0.7 cm 2 . For the mixed metal oxide catalysts (NiFeO x and NiFeCoO x ), 10 µL of each metal complex was added to 10 mL of hexane, and then 20 µL of the solutions were drop-cast on the Ta 3 N 5 surface, with a coverage area of approximately 0.7 cm 2 . The loading amount of these catalysts was 0.03 µmol cm −2 . This was followed by heating at 413 K in air for 45 min. Subsequently, the samples were rinsed with a copious amount of hexane in order to remove residual precursors, followed by drying in air. Indium metals were soldered onto the back side of Ta substrate of Ta 3 N 5 /Ta and onto the edge of the Ta 3 N 5 surface of the Ta 3 N 5 /SiO 2 . Following this, additional indium was soldered to connect the lead wire, and then the unused region was covered with epoxy resin as insulation. The schematic illustration of the Ta 3 N 5 /Ta and Ta 3 N 5 /SiO 2 is given in Figure S17 in the SI.

PEC Measurements
PEC measurements for the Ta 3 N 5 -based photoanodes were performed with a threeelectrode configuration using a Ag/AgCl/sat'd KCl reference electrode and a coiled Pt wire counter electrode. The electrode potential was controlled using a potentiostat (HSV-110, Hokuto Denko Corp., Tokyo, Japan). A surface-modified Ta 3 N 5 photoanode with an exposed area of 0.7 to 0.5 cm 2 was positioned in a glass cell containing 0.1 M sodium phosphate aqueous solution with the pH adjusted to 13 by the addition of NaOH as the electrolyte. The electrode potentials were converted to values versus the reversible hydrogen electrode (vs. the RHE) using the Nernst equation. A solar simulator equipped with an AM 1.5G filter (XES-40S2, SAN-EI Electric Co. Ltd., Osaka, Japan) was employed as the light source. The intensity of the light was adjusted to AM 1.5G with 100 mW cm −2 . Incident photon-to-current conversion efficiency (IPCE) values were obtained using a MAX-302 Xe light source (Asahi Spectra Co. Ltd., Tokyo, Japan) equipped with bandpass filters to provide monochromatic light. IPCEs were calculated using the equation: where λ is the wavelength of the monochromatic incident light (nm), J is the photocurrent density (mA cm −2 ), and P is the intensity of the incident light (mW cm −2 ). Overall water splitting was performed in two differently configured PEC cells consisting of a Ta 3 N 5 photoanode and an LTCA:Al photocathode. For the parallel PEC cell, the surface-modified Ta 3 N 5 /Ta photoanode and the LTCA:Al photocathode were placed side-by-side. For the tandem PEC cell, the surface-modified Ta 3 N 5 /SiO 2 transparent photoanode was placed in front of the LTCA:Al photocathode. The Ta 3 N 5 -based photoanode and the LTCA:Al-based photocathode were connected to one another via lead wires. The LTCA:Al photocathode was prepared using the particle transfer method [9,11,48]. Prior to the construction of the cells, the LTCA:Al photocathodes were sequentially modified with a CdS layer via chemical bath deposition, a TiO 2 surface protection layer was deposited via drop-casting of a tetraisopropyl orthotitanate aqueous solution, and a Pt catalyst was deposited via photoelectrochemical deposition [9]. The resulting photocurrent and quantities of evolved gases were simultaneously determined using a multimeter and a micro gas chromatograph (GC3000A, Agilent Technologies Inc. Santa Clata, CA, USA).
The STH values for the cells were calculated from the i-t curves for the PEC cells using the following equation [44]: where J is the photocurrent density (mA cm −2 ) and P is the intensity of simulated sunlight (AM 1.5G, 100 mW cm −2 ). Note that the geometrical electrode areas for the parallel PEC cell and the tandem PEC cell were the total effective area of both photoelectrodes (Ta 3 N 5 -LTCA:Al) and the effective area of the front-side Ta 3 N 5 photoanode, respectively.

Conclusions
The surface of Ta 3 N 5 thin film-based photoanodes was modified with electrocatalysts consisting of metal oxides in order to clarify their effect on the PEC oxygen evolution efficiency. Surface modification with mixed metal oxides such as NiFeO x and NiFeCoO x improved the onset potential to 0.6 V vs. the RHE, while the photoanodes with surface modification by the mono-metal oxides-namely, NiO x , FeO x , and CoO x -exhibited poor onset potentials as high as 0.7 V vs. the RHE. This trend for the onset potential is attributable to the overpotential for the electrocatalytic OER performance of the applied materials. The NiFeO x /Ta 3 N 5 photoanode exhibited a significantly enhanced photocurrent, achieving a value of 0.40 mA cm −2 at 0.7 V vs. the RHE. A parallel PEC cell composed of a NiFeO x /Ta 3 N 5 photoanode and a Pt/TiO 2 /CdS/LTCA:Al photocathode exhibited non-biased, spontaneous overall water splitting, with stoichiometric hydrogen and oxygen evolution, and a Faradaic efficiency of~97%. The STH for the parallel PEC cell was found to be 0.2% after 1 min of simulated sunlight illumination. A tandem PEC cell composed of a NiFeO x /Ta 3 N 5 /SiO 2 transparent photoanode as the front-side electrode and a Pt/TiO 2 /CdS/LTCA:Al photocathode behind it generated a self-standing photocurrent density of 0.036 mA cm −2 without the application of an external bias voltage. The feasibility of water splitting by tandem PEC cells consisting of a nitride-based transparent photoanode and an oxysulfide-based particulate photocathode was established in the present study. The photocurrent density and the STH value obtained by the tandem PEC cells are far from sufficient for practical applications, and are lower than those reported for PEC cells. There are still many problems to be solved, such as the low photocurrent at electrode potentials below 0.7 V vs. the RHE, and the low stability of the Ta 3 N 5 photoanodes. The photocurrent can be enhanced by further surface modification and improvement of the semiconductor properties of Ta 3 N 5 . Introducing an oxide-based protective layer on the NiFeO x electrocatalyst might be helpful in stabilizing the Ta 3 N 5 photoanode. Further improvement of the performance of the photocathode and the tandem PEC cell, enabling us to accurately evaluate the gas generation rates and the Faradaic efficiency, are also issues for future work.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/catal11050584/s1, Figure S1: schematic illustration of PEC cells consisting of photoanodephotocathode combinations; Figure S2: i-E curve for a bare Ta 3 N 5 photoanode; Figure S3: UV-vis transmission spectrum of an FeO x -deposited quartz substrate; Figure S4: SEM images; Figure S5: SEM-EDX analysis; Figure S6: XPS spectra; Figure S7: photograph of a parallel PEC cell; Figure S8: time course of the working potential of a PEC cell; Figure S9: i-E curves for a NiFeO x /Ta 3 N 5 photoanode after water splitting reaction; Figure S10: SEM images after water splitting reaction; Figure S11: photograph of a Ta 3 N 5 /SiO 2 photoanode; Figure S12: absorption coefficient spectrum and plot of light intensity as function of optical path length in Ta 3 N 5 ; Figure S13: UV-vis transmission spectrum for a NiFeO x /Ta 3 N 5 /SiO 2 transparent photoanode; Figure S14: i-t curve for a NiFeO x /Ta 3 N 5 /SiO 2 transparent photoanode at 1.23 V vs. the RHE; Figure S15: i-E curves for Ta 3 N 5 /SiO 2 transparent photoanodes with different film thicknesses; Figure S16: photograph of a tandem PEC cell; Figure S17: schematic illustration of the cross-sectional structures of Ta 3 N 5 /Ta and Ta 3 N 5 /SiO 2 photoanodes.

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