Reactive Inorganic Vapor Deposition of Perovskite Oxynitride Films for Solar Energy Conversion

The synthesis of perovskite oxynitrides, which are promising photoanode candidates for solar energy conversion, is normally accomplished by high-temperature ammonolysis of oxides and carbonate precursors, thus making the deposition of their planar films onto conductive substrates challenging. Here, we proposed a facile strategy to prepare a series of perovskite oxynitride films. Taking SrTaO2N as a prototype, we prepared SrTaO2N films on Ta foils under NH3 flow by utilizing the vaporized SrCl2/SrCO3 eutectic salt. The SrTaO2N films exhibit solar water-splitting photocurrents of 3.0 mA cm−2 at 1.23 V vs. RHE (reversible hydrogen electrode), which increases by 270% compared to the highest photocurrent (1.1 mA cm−2 at 1.23 V vs. RHE) of SrTaO2N reported in the literature. This strategy may also be applied to directly prepare a series of perovskite oxynitride films on conductive substrates such as ATaO2N (A = Ca, Ba) and ANbO2N (A = Sr, Ba).

To date, only some particle-assembled oxynitride films (such as LaTiO 2 N, BaTaO 2 N, SrTaO 2 N, and LaTaON 2 ) have been investigated for PEC water splitting [20][21][22][23][24]. In order to prepare the particle-assembled oxynitride films, the oxynitride powders are firstly synthesized by nitridation of corresponding oxide precursors at high temperature under ammonia flow usually at 800~1300°C [5,[25][26][27][28][29]. Subsequently, the particle-assembled oxynitride films are prepared by electrophoretic deposition or particle transfer methods [20][21][22][23][24]. However, these particle-assembled films often suffer from poor charge carrier transport not only among film particles but also between the film particles and the conductive substrate. Also, the efficiency loss may even occur at the exposed region of the underlying conductive substrate [30,31]. Various efforts have been made to improve their PEC performances, such as increasing the particle crystallinity or H 2 annealing to prompt the charge transport in film particles, and using necking treatment to ameliorate the charge transport among film particles [20][21][22][23][24]. Unfortunately, the electron-hole recombination at the interface between the film particles and conductive substrate is still serious in these particle-assembled films, thus hampering their solar-to-hydrogen efficiency.
It is indispensable and challenging to directly synthesize oxynitride photoanode films on a conductive substrate for overcoming the main shortcomings of the particleassembled films, including the poor charge carrier transport and the efficiency loss at the exposed conductive substrate. Here, we have presented a strategy of reactive inorganic vapor deposition and prepared a series of oxynitride films, for example, SrTaO 2 N. The as-prepared SrTaO 2 N photoanode films exhibit a significant increase in the photocurrent as well as photochemical stability and a cathodic shift in onset potential for water oxidation, in comparison with the particle-assembled photoanode films.

Results
In this study, we have prepared SrTaO 2 N photoanode films by reactive inorganic vapor deposition, as illustrated in Figure 1(a). SrCl 2 (74%)/SrCO 3 (26%) eutectic salt is used to generate SrCl 2 vapor during the nitridation process and provide an Sr source for the formation of SrTaO 2 N. The CO 2 released from SrCO 3 upon the nitridation process is employed as an oxidant to oxidize Ta to Ta 5+ [32]. Ta foil is used both as a Ta source and as a conductive substrate. The SrCl 2 /SrCO 3 eutectic salt may be evaporated during the ammonolysis process at 950°C for 2 hours, which is approximately 280°C higher than their melting temperature, according to the phase diagram in Figure S1 [33]. The vaporized flux and released CO 2 react with the Ta foils under the ammonia atmosphere, resulting in the formation of SrTaO 2 N film on Ta foil. The morphology of the SrTaO 2 N films was analyzed using field-emission scanning electron microscopy (SEM). The top-down image ( Figure S2) and cross-sectional image (Figure 1(b)) show that the SrTaO 2 N film is a crack-free film consisting of compact and octahedron-like grains; the close contact between the SrTaO 2 N film and Ta substrate is propitious to charge carrier transport.
The X-ray diffraction (XRD) pattern (shown in Figure 2(a)) confirms the synthesis of the SrTaO 2 N film, based on the JCPDS file (PDF#40-0662). The high-resolution transmission electron microscopy (HRTEM) image of Figure 2(b) displays a lattice spacing of 0.286 nm, corresponding to the SrTaO 2 N (1 1 0) facet. The HRTEM image and selected area electron diffraction (SAED) pattern ( Figure S3) suggest that the SrTaO 2 N samples exhibit high crystallinity, which favors electron transport.
The cross section of the SrTaO 2 N film was further studied by TEM. As shown in Figure 3(a), the formation of a TaN subnitride layer can be observed between the SrTaO 2 N film and Ta substrate, which is suggested by selected area electron diffraction (SAED) images (Figures 3(b) and 3(c)). The presence of a highly conductive TaN phase provides an efficient electron transfer pathway that benefits the electron transfer from the SrTaO 2 N film to Ta substrate. The electrical conductivity of as-prepared SrTaO 2 N film photoanode is estimated to be 1:82 × 10 −5 S cm −1 (derived from Figure S4), which is approximately 23.6 times of 0:77 × 10 −6 S cm −1 for the particle-assembled SrTaO 2 N photoanode [34]. The SrTaO 2 N films prepared by our method exhibit much higher conductivity than the particle-assembled films, which may be ascribed to the reduction in the interfacial resistance.
The combination of TG-DTA (thermogravimetric and differential thermal analysis) and mass spectral measurements for the SrCO 3 /SrCl 2 eutectic salt sample and SrCO 3 /SrCl 2 eutectic salt+Ta sample was carried out to study the formation mechanism of SrTaO 2 N, as shown in Figure 4.
The weight loss in the temperature range of approximately 70°C to 170°C is caused by the loss of crystal water in SrCl 2 /SrCO 3 eutectic salt, which is confirmed by the H 2 O signal in mass spectra. SrCO 3 started to decompose in the temperature range of approximately 550°C to 800°C, according to the weight loss (Figure 4(a)) and CO 2 signal in mass spectra. Compared with the SrCl 2 /SrCO 3 eutectic salt sample, the SrCO 3 /SrCl 2 eutectic salt+Ta sample exhibited a weight increase and a peak of H 2 O signal in mass spectra around 750°C, which may be an evidence of nitridation reaction.
Control experiments were also carried out by nitriding Ta foils in the same condition with different components of the flux reagent. In the case of the SrCl 2 flux reagent, only small amount of TaN 0.04 and Ta 2 N was obtained, as shown in the XRD pattern of Figure S5a. Without the presence of carbonate, there is no CO 2 released during the nitridation process, metallic Ta cannot be nitrided to high-oxidationstate nitride directly by ammonia upon this condition, which is consistent with literatures [32,35]. Using only SrCO 3 as the flux reagent, Ta 3 N 5 instead of SrTaO 2 N was obtained ( Figure S5b). These results indicate that both the SrCl 2 flux vapor and CO 2 released from SrCO 3 are crucial for the formation of SrTaO 2 N.
There are no chlorine impurities left on the SrTaO 2 N film, according to XPS spectra ( Figure S6). The carbon in the films was analyzed by XPS, as shown in Figure S7. The C1s spectra of the SrTaO 2 N film show a peak at 284.6 eV, corresponding to the surface adventitious carbon. And the TEM and HRTEM images ( Figure S8) suggested that there is no amorphous carbon on the surface of SrTaO 2 N particles. Therefore, the reaction equation for the formation of SrTaO 2 N films may be written as follows: In order to find the optimal composition of the SrCl 2 /SrCO 3 flux reagent for the synthesis of SrTaO 2 N film photoanodes, control experiments were also carried out by nitriding Ta foils in the same condition with different compositions of the SrCl 2 /SrCO 3 flux reagent (50%/50%, 74%/26%, and 80%/20%). The SrTaO 2 N film photoanode prepared with SrCl 2 (74%)/SrCO 3 (26%) eutectic salt exhibits the best PEC performance, as shown in Figure S9. The photocurrent of SrTaO 2 N photoanode prepared with the SrCl 2 (50%)/SrCO 3 (50%) flux reagent at 1.23 V vs. RHE (2.0 mA cm -2 ) is only 66.7% of that of SrTaO 2 N photoanode prepared with SrCl 2 (74%)/SrCO 3 (26%) eutectic salt (3.0 mA cm -2 ). The melting temperatures of the SrCl 2 (50%)/SrCO 3 (50%) flux reagent, SrCl 2 (74%)/SrCO 3 (26%) eutectic salt and SrCl 2 (80%)/SrCO 3 (20%) flux reagent are 870°C, 670°C and 715°C, respectively, according to the phase diagram in Figure S1. When using the SrCl 2 (50%)/SrCO 3 (50%) flux reagent as a precursor, approximately half of the flux reagent remained in the alumina combustion boat after the 2 hours of the nitridation process. These results suggested that the SrCl 2 (74%)/SrCO 3 (26%) eutectic salt with the lowest melting temperature may introduce the highest Sr partial pressure during the nitridation process, thus resulting in the best PEC performance of SrTaO 2 N film photoanode.
Co/CoOOH catalyst layers were deposited onto the SrTaO 2 N photoanode films by a two-step electrodeposition method [36]. A photoassisted electrodeposition method [37] was adopted to produce a conformal CoOOH layer on the SrTaO 2 N film by deposition on only where photoinduced holes were generated. The SEM image indicates the conformal distribution of the CoOOH layer on the SrTaO 2 N photoanodes ( Figure S10a). As illustrated in Figure 5(a), with the surface decoration Co/CoOOH catalyst layers, the SrTaO 2 N film photoanode synthesized by the reactive inorganic vapor deposition method (RVD SrTaO 2 N) exhibits a solar photocurrent of 3.0 mA cm -2 at 1.23 V vs. RHE and a low onset potential for water oxidation at 0.55 V vs. RHE. The highest PEC performance of particle-assembled SrTaO 2 N photoanodes (particle-assembled SrTaO 2 N) in the literature exhibits a solar photocurrent of 1.1 mA cm -2 at 1.23 V vs. RHE and an onset potential of 0.6 V vs. RHE for water oxidation ( Figure 5(a), blue curves) [34]. Compared with the particle-assembled SrTaO 2 N photoanode, the directly prepared SrTaO 2 N film photoanode shows a ca. 270% enhancement in the photocurrent at 1.23 V vs. RHE and a ca. 50 mV cathodic shift in onset potential for water oxidation. The wavelength dependence of the incident photon-to-current conversion efficiency (IPCE) of the SrTaO 2 N film photoanode is shown in Figure 5(b). The onset wavelength for the photocurrents is about 600 nm, which agrees well with the optical absorption edge of SrTaO 2 N ( Figure S11a). The integrated photocurrent is 2.9 mA cm -2 at 1.23 V vs. RHE, which is calculated by multiplication of the measured IPCE spectra with the AM 1.5G standard solar spectral distribution ( Figure S12). It agrees well with the experimental photocurrent of 3.0 mA cm -2 at 1.23 V vs. RHE in Figure 5(a). The small difference between the integrated and experimental photocurrents suggests that the IPCE spectra and the photocurrent under AM 1.5G illumination are credible.
The applied bias photon-to-current efficiency (ABPE) of the SrTaO 2 N films calculated by using the J-V curve ( Figure 5(a)) is shown in Figure S13, assuming 100%  faradaic efficiency. The maximum ABPE is 0.53% at 0.85 V vs. RHE for SrTaO 2 N film photoanode, which is ca. 3.5 times higher than that of 0.15% at 0.9 V vs. RHE for particleassembled SrTaO 2 N photoanode. Gas chromatography was used to demonstrate that H 2 and O 2 gases evolved during PEC water splitting in a stoichiometric ratio of 2 : 1, and the faradaic efficiency was 93% for H 2 and 94% for O 2 , respectively ( Figure 5(c)). The band gap and flat band potentials (E fb ) of the SrTaO 2 N film photoanode were investigated by the Tauc plot ( Figure S11b) and Mott-Schottky plots ( Figure S14), respectively. These results show that the band gap of SrTaO 2 N is approximately 2.1 eV, and the conduction and valence band edges for SrTaO 2 N are estimated to be -0.5 and 1.6 V vs. RHE, respectively, which agrees well with previous report [38]. Photochemical stability is a vital parameter for the application of solar water splitting. The SrTaO 2 N film photoanode exhibits an initial photocurrent of approximately 2.8 mA cm -2 and retains approximately 86% of its initial activity after AM 1.5G (100 mW cm -2 ) irradiation for 5 hours, which shows significant improvement compared with the particle-assembled SrTaO 2 N photoanodes, as shown in Figure 5(d). The improvement in the stability may be ascribed to not only the close contact between SrTaO 2 N grains but also the highly conductive TaN phase benefiting the electron transfer from the SrTaO 2 N film to Ta substrate. The gradual drop of the photocurrent is partly caused by the peeling off and the dissolution of the CoOOH layer in strong base (1 M NaOH, pH = 13:6) during the stability testing ( Figure S10b). Similar phenomenon has also been observed that the CoOOH electrocatalyst shows a steady decrease in the current density for oxygen evolution reaction in 1 M KOH [36].
Electrochemical impedance spectroscopy measurements were performed in a three-electrode configuration with frequencies ranging from 0.1 Hz to 100 kHz, in order to further comprehend the charge transport of these two SrTaO 2 N photoanodes. The Nyquist plots of the SrTaO 2 N photoanodes (shown in Figure 6) reveal two semicircles. The impedance spectra have been modelled to an equivalent circuit model consisting of two parallel RC elements connected in series, which is shown in the inset of Figure 6(b). The first RC element represents the bulk processes (R bulk , C bulk ), and the second represents the surface processes (R ct , C sc ). R s is the electrolyte resistance. The fitted results (Table S1) show that the bulk charge transfer resistance R bulk of the SrTaO 2 N film photoanode is only 2% of that of the particle-assembled SrTaO 2 N photoanode, thus resulting in a better PEC performance.
The simplicity and rapidity of this process are applicable to a wide range of oxynitride films. For example, CaTaO 2 N and BaTaO 2 N films were synthesized at the same condition by using the CaCl 2 /CaCO 3 flux and BaCl 2 /BaCO 3 flux, respectively, which are proven by their XRD patterns in Figure S15. As-prepared CaTaO 2 N and BaTaO 2 N films are similar crack-free and large-grained films ( Figure S16). The HRTEM images of CaTaO 2 N and BaTaO 2 N films display a lattice spacing of 0.280 nm and 0.292 nm ( Figure S17), corresponding to CaTaO 2 N (2 0 0) and BaTaO 2 N (1 1 0), respectively. Moreover, SrNbO 2 N and BaNbO 2 N films were also synthesized at the same condition by using Nb foil and corresponding flux (SrCl 2 /SrCO 3 flux for the SrNbO 2 N film, BaCl 2 /BaCO 3 flux for the BaNbO 2 N film) as precursors. The XRD patterns confirm the synthesis of SrNbO 2 N and BaNbO 2 N films; the morphology of SrNbO 2 N and BaNbO 2 N films exhibit similar large-grained films ( Figure S18). The PEC performance of these films was also measured as shown in Figure S19. The PEC performance of these oxynitride films may be further optimized by choosing the optimal composition of the corresponding flux reagent to generate higher Ca or Ba partial pressure during the nitridation process.

Discussion
In this study, a series of perovskite oxynitride films such as ATaO 2 N (A = Ca, Sr, and Ba) and ANbO 2 N (A = Sr, Ba) have been prepared by the reactive inorganic vapor deposition method. As-prepared SrTaO 2 N film photoanode exhibits much better PEC performances than the traditional particleassembled SrTaO 2 N photoanode. Solar water-splitting photocurrent of 3.0 mA cm -2 at 1.23 V vs. RHE over the SrTaO 2 N film photoanode is increased by 270%, compared with the highest photocurrent of the particle-assembled SrTaO 2 N photoanode in the literature. Also, the photocurrent remains 86% of its initial activity after AM 1.5G illumination for 5 h, suggesting much better photochemical stability of the SrTaO 2 N film photoanode than the particle-assembled SrTaO 2 N photoanode. The improvement in the PEC performances may be ascribed to less bulk resistance of the directly prepared SrTaO 2 N films. The concept of reactive inorganic vapor deposition growth may cast light on the fabrication of oxide, oxyhalide, and oxysulfide films, as well as other oxynitride films. and 74 mol% SrCl 2 ·6H 2 O (Sinopharm Chemical Reagent Co., 99.5%), was placed in an alumina combustion boat. Ta foils were put on the top of the alumina combustion boat in a sealed quartz tube furnace. The Ta foils were then annealed at 950°C for 2 hours under a flow of ammonia gas (flow rate: 250 sccm) at a rate of 10°C/min. The samples were then cooled to room temperature with the flow of ammonia gas.

Photoassisted Electrodeposition of Co/CoOOH Catalysts.
A two-step electrodeposition method [36] was introduced to deposit a Co/CoOOH catalyst layer from a solution of  ) was obtained from a Newport Sol3A Class AAA simulator, and the light intensity was calibrated at 100 mW cm -2 by a standard reference Newport 91150 silicon cell before testing. The backsides of the SrTaO 2 N films were covered with AB glue and dried overnight before the PEC testing. The wavelength dependence of the incident photon-to-current efficiency (IPCE) was measured under monochromatic light irradiation from a xenon lamp (Ushio, Optical Module X500) equipped with 330-600 nm band pass filters. The light intensity of each wavelength was obtained with a photometer (Newport, 840-C). The applied bias photon-to-current efficiency (ABPE) was calculated from the J-V curve of the photoanode under AM 1.5G simulated sunlight.

Electrochemical Impedance Spectroscopy Measurements.
Electrochemical impedance spectroscopy measurements were performed using an electrochemical analyzer (Solartron 1260+1287) with a frequency range of 0.1 Hz-100 kHz. All the measurements were obtained at room temperature at an applied potential of 0.6 V vs. RHE under dark conditions. The EIS spectral data were fitted to the electrical analogue using ZView software.  Figure 6: Electrochemical impedance spectroscopy measurements. Nyquist plots of the SrTaO 2 N film photoanode (a) and particle-assembled SrTaO 2 N photoanode (b) recorded at 0.6 V vs. RHE; the inset shows the equivalent circuit model. Figure S1: phase diagram of the system SrCl 2 -SrCO 3 . Figure  S2: top-down SEM image of the SrTaO 2 N film. Figure S3: SAED pattern of the SrTaO 2 N film. Figure S4: I-V characteristics of the SrTaO 2 N film. Figure S5: XRD patterns of control experiments. Figure S6: XPS spectra from 220 eV to 180 eV. Figure S7: XPS spectra of C1s. Figure S8: TEM and HRTEM images of the peeled SrTaO2N particles. Figure S9: photocurrents of the SrTaO 2 N film prepared with different compositions of the SrCl 2 /SrCO 3 flux reagent. Figure S10: SEM image of the CoOOH catalyst layer. Figure S11: UV-vis diffuse reflectance spectra and Tauc plot of the SrTaO 2 N film. Figure S12: derived photocurrent of the SrTaO 2 N film. Figure S13: ABPE of SrTaO 2 N film photoanode and particle-assembled SrTaO 2 N photoanode. Figure S14: Mott-Schottky plots of the SrTaO 2 N film. Figure S15: XRD patterns of CaTaO 2 N and BaTaO 2 N films. Figure S16: SEM images of CaTaO 2 N and BaTaO 2 N films. Figure S17: HRTEM images of CaTaO 2 N and BaTaO 2 N films. Figure  S18: XRD patterns and SEM images of SrNbO 2 N and BaN-bO 2 N films. Figure S19: J-V curves of the CaTaO 2 N film, BaTaO 2 N film, SrNbO 2 N film, and BaNbO 2 N film. Table  S1: the fitted values of R bulk .