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).

1. Introduction

Perovskite oxynitrides, which may have the advantages of both oxides and nitrides [1], emerge as an important class of materials due to their promising applications in pigments [2], dielectrics [3, 4], ferroelectrics [5], colossal magneto-resistance [6], and photocatalysis [7, 8]. They usually have smaller bandgap than those of prototypical oxides and better chemical stability than those of prototypical nitrides. Perovskite oxynitrides are currently predicted to be promising candidates for photoelectrochemical (PEC) water splitting [9], which is a potential way to harness solar energy [10-19].

To date, only some particle-assembled oxynitride films (such as LaTiO2N, BaTaO2N, SrTaO2N, and LaTaON2) have been investigated for PEC water splitting [20-24]. 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, 25–29]. Subsequently, the particle-assembled oxynitride films are prepared by electrophoretic deposition or particle transfer methods [20-24]. 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 H2 annealing to prompt the charge transport in film particles, and using necking treatment to ameliorate the charge transport among film particles [20-24]. 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 particle-assembled 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, SrTaO2N. The as-prepared SrTaO2N 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.

2. Results

In this study, we have prepared SrTaO2N photoanode films by reactive inorganic vapor deposition, as illustrated in Figure 1(a). SrCl2 (74%)/SrCO3 (26%) eutectic salt is used to generate SrCl2 vapor during the nitridation process and provide an Sr source for the formation of SrTaO2N. The CO2 released from SrCO3 upon the nitridation process is employed as an oxidant to oxidize Ta to Ta5+ [32]. Ta foil is used both as a Ta source and as a conductive substrate. The SrCl2/SrCO3 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 [33]. The vaporized flux and released CO2 react with the Ta foils under the ammonia atmosphere, resulting in the formation of SrTaO2N film on Ta foil. The morphology of the SrTaO2N films was analyzed using field-emission scanning electron microscopy (SEM). The top-down image () and cross-sectional image (Figure 1(b)) show that the SrTaO2N film is a crack-free film consisting of compact and octahedron-like grains; the close contact between the SrTaO2N film and Ta substrate is propitious to charge carrier transport.

Figure 1

Schematic illustration. (a) Schematic illustration of the reactive inorganic vapor deposition method. (b) Cross-sectional SEM image of the as-prepared SrTaO2N film.

The X-ray diffraction (XRD) pattern (shown in Figure 2(a)) confirms the synthesis of the SrTaO2N 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 SrTaO2N (1 1 0) facet. The HRTEM image and selected area electron diffraction (SAED) pattern () suggest that the SrTaO2N samples exhibit high crystallinity, which favors electron transport.

Figure 2

XRD pattern and HRTEM image. XRD pattern (a) and HRTEM image (b) of the as-prepared SrTaO2N film. The inset shows the crystal structure of SrTaO2N.

The cross section of the SrTaO2N film was further studied by TEM. As shown in Figure 3(a), the formation of a TaN subnitride layer can be observed between the SrTaO2N 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 SrTaO2N film to Ta substrate. The electrical conductivity of as-prepared SrTaO2N film photoanode is estimated to be 1.82 × 10−5 S cm−1 (derived from ), which is approximately 23.6 times of 0.77 × 10−6 S cm−1 for the particle-assembled SrTaO2N photoanode [34]. The SrTaO2N 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.

Figure 3

Cross-sectional TEM and SAED images of the SrTaO2N film. Cross-sectional TEM image (a) and SAED images (b, c) at different depths of the SrTaO2N film.

The combination of TG-DTA (thermogravimetric and differential thermal analysis) and mass spectral measurements for the SrCO3/SrCl2 eutectic salt sample and SrCO3/SrCl2 eutectic salt+Ta sample was carried out to study the formation mechanism of SrTaO2N, 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 SrCl2/SrCO3 eutectic salt, which is confirmed by the H2O signal in mass spectra. SrCO3 started to decompose in the temperature range of approximately 550°C to 800°C, according to the weight loss (Figure 4(a)) and CO2 signal in mass spectra. Compared with the SrCl2/SrCO3 eutectic salt sample, the SrCO3/SrCl2 eutectic salt+Ta sample exhibited a weight increase and a peak of H2O signal in mass spectra around 750°C, which may be an evidence of nitridation reaction.

Figure 4

The combination of TG-DTA and mass spectroscopy. Thermogravimetric analysis (TG), differential thermal gravity (DTG), and mass spectra of the SrCl2/SrCO3 eutectic salt (26 mol% SrCO3, 74 mol% SrCl2·6H2O) (a) and the SrCO3/SrCl2 eutectic salt+Ta (the mass ratio of the eutectic salt and the Ta powder is 2 : 1) (b).

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 SrCl2 flux reagent, only small amount of TaN0.04 and Ta2N was obtained, as shown in the XRD pattern of . Without the presence of carbonate, there is no CO2 released during the nitridation process, metallic Ta cannot be nitrided to high-oxidation-state nitride directly by ammonia upon this condition, which is consistent with literatures [32, 35]. Using only SrCO3 as the flux reagent, Ta3N5 instead of SrTaO2N was obtained (). These results indicate that both the SrCl2 flux vapor and CO2 released from SrCO3 are crucial for the formation of SrTaO2N.

There are no chlorine impurities left on the SrTaO2N film, according to XPS spectra (). The carbon in the films was analyzed by XPS, as shown in . The C1s spectra of the SrTaO2N film show a peak at 284.6 eV, corresponding to the surface adventitious carbon. And the TEM and HRTEM images () suggested that there is no amorphous carbon on the surface of SrTaO2N particles. Therefore, the reaction equation for the formation of SrTaO2N films may be written as follows:

In order to find the optimal composition of the SrCl2/SrCO3 flux reagent for the synthesis of SrTaO2N film photoanodes, control experiments were also carried out by nitriding Ta foils in the same condition with different compositions of the SrCl2/SrCO3 flux reagent (50%/50%, 74%/26%, and 80%/20%). The SrTaO2N film photoanode prepared with SrCl2 (74%)/SrCO3 (26%) eutectic salt exhibits the best PEC performance, as shown in . The photocurrent of SrTaO2N photoanode prepared with the SrCl2 (50%)/SrCO3 (50%) flux reagent at 1.23 V vs. RHE (2.0 mA cm−2) is only 66.7% of that of SrTaO2N photoanode prepared with SrCl2 (74%)/SrCO3 (26%) eutectic salt (3.0 mA cm−2). The melting temperatures of the SrCl2 (50%)/SrCO3 (50%) flux reagent, SrCl2 (74%)/SrCO3 (26%) eutectic salt and SrCl2 (80%)/SrCO3 (20%) flux reagent are 870°C, 670°C and 715°C, respectively, according to the phase diagram in . When using the SrCl2 (50%)/SrCO3 (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 SrCl2 (74%)/SrCO3 (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 SrTaO2N film photoanode.

Co/CoOOH catalyst layers were deposited onto the SrTaO2N photoanode films by a two-step electrodeposition method [36]. A photoassisted electrodeposition method [37] was adopted to produce a conformal CoOOH layer on the SrTaO2N film by deposition on only where photoinduced holes were generated. The SEM image indicates the conformal distribution of the CoOOH layer on the SrTaO2N photoanodes (). As illustrated in Figure 5(a), with the surface decoration Co/CoOOH catalyst layers, the SrTaO2N film photoanode synthesized by the reactive inorganic vapor deposition method (RVD SrTaO2N) 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 SrTaO2N photoanodes (particle-assembled SrTaO2N) 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 SrTaO2N photoanode, the directly prepared SrTaO2N 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 SrTaO2N 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 SrTaO2N (). 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 (). 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.

Figure 5

PEC performance and faradaic efficiency. (a) Photocurrents of the directly prepared SrTaO2N film and particle-assembled SrTaO2N photoanodes [34] in 1 M NaOH (pH = 13.6) electrolyte under AM 1.5G (100 mW cm−2) simulated sunlight and a scan rate of 30 mV s−1. (b) Wavelength dependence on the incident photon-to-current conversion efficiency (IPCE) of the SrTaO2N film photoanode with the Co/CoOOH catalyst layer measured at 1.23 V vs. RHE in 1 M NaOH (pH = 13.6) aqueous solution. (c) Gas chromatography for O2 and H2 generated from the SrTaO2N film photoanode modified with the Co/CoOOH layer and Pt counter electrode during PEC water splitting at an applied potential of 1.23 V vs. RHE for 180 min. (d) I-t curve for the SrTaO2N film photoanode and particle-assembled SrTaO2N photoanodes during PEC water splitting at 1.23 V vs. RHE (for 240 min) under AM 1.5G (100 mW cm−2) illumination.

The applied bias photon-to-current efficiency (ABPE) of the SrTaO2N films calculated by using the J-V curve (Figure 5(a)) is shown in , assuming 100% faradaic efficiency. The maximum ABPE is 0.53% at 0.85 V vs. RHE for SrTaO2N film photoanode, which is ca. 3.5 times higher than that of 0.15% at 0.9 V vs. RHE for particle-assembled SrTaO2N photoanode. Gas chromatography was used to demonstrate that H2 and O2 gases evolved during PEC water splitting in a stoichiometric ratio of 2 : 1, and the faradaic efficiency was 93% for H2 and 94% for O2, respectively (Figure 5(c)). The band gap and flat band potentials (Efb) of the SrTaO2N film photoanode were investigated by the Tauc plot () and Mott-Schottky plots (), respectively. These results show that the band gap of SrTaO2N is approximately 2.1 eV, and the conduction and valence band edges for SrTaO2N 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 SrTaO2N 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 SrTaO2N photoanodes, as shown in Figure 5(d). The improvement in the stability may be ascribed to not only the close contact between SrTaO2N grains but also the highly conductive TaN phase benefiting the electron transfer from the SrTaO2N 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 (). 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 SrTaO2N photoanodes. The Nyquist plots of the SrTaO2N 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 (Rbulk, Cbulk), and the second represents the surface processes (Rct, Csc). Rs is the electrolyte resistance. The fitted results () show that the bulk charge transfer resistance Rbulk of the SrTaO2N film photoanode is only 2% of that of the particle-assembled SrTaO2N photoanode, thus resulting in a better PEC performance.

Figure 6

Electrochemical impedance spectroscopy measurements. Nyquist plots of the SrTaO2N film photoanode (a) and particle-assembled SrTaO2N photoanode (b) recorded at 0.6 V vs. RHE; the inset shows the equivalent circuit model.

The simplicity and rapidity of this process are applicable to a wide range of oxynitride films. For example, CaTaO2N and BaTaO2N films were synthesized at the same condition by using the CaCl2/CaCO3 flux and BaCl2/BaCO3 flux, respectively, which are proven by their XRD patterns in . As-prepared CaTaO2N and BaTaO2N films are similar crack-free and large-grained films (). The HRTEM images of CaTaO2N and BaTaO2N films display a lattice spacing of 0.280 nm and 0.292 nm (), corresponding to CaTaO2N (2 0 0) and BaTaO2N (1 1 0), respectively. Moreover, SrNbO2N and BaNbO2N films were also synthesized at the same condition by using Nb foil and corresponding flux (SrCl2/SrCO3 flux for the SrNbO2N film, BaCl2/BaCO3 flux for the BaNbO2N film) as precursors. The XRD patterns confirm the synthesis of SrNbO2N and BaNbO2N films; the morphology of SrNbO2N and BaNbO2N films exhibit similar large-grained films (). The PEC performance of these films was also measured as shown in . 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.

3. Discussion

In this study, a series of perovskite oxynitride films such as ATaO2N (A = Ca, Sr, and Ba) and ANbO2N (A = Sr, Ba) have been prepared by the reactive inorganic vapor deposition method. As-prepared SrTaO2N film photoanode exhibits much better PEC performances than the traditional particle-assembled SrTaO2N photoanode. Solar water-splitting photocurrent of 3.0 mA cm−2 at 1.23 V vs. RHE over the SrTaO2N film photoanode is increased by 270%, compared with the highest photocurrent of the particle-assembled SrTaO2N 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 SrTaO2N film photoanode than the particle-assembled SrTaO2N photoanode. The improvement in the PEC performances may be ascribed to less bulk resistance of the directly prepared SrTaO2N 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.

4. Materials and Methods

4.1. Preparation of SrTaO2N, BaTaO2N, and CaTaO2N Films

Ta foils (10 mm × 15 mm × 0.2 mm, ZhongNuo Advanced Material Co., 99.95%) were cleaned in ethanol and acetone. SrTaO2N films were prepared on a Ta foil substrate by a reactive inorganic vapor deposition method under a flow of NH3. Briefly, 0.2 g of SrCl2/SrCO3 eutectic salt, which consisted of 26 mol% SrCO3 (Sinopharm Chemical Reagent Co., 99.5%) and 74 mol% SrCl2·6H2O (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. After the ammonolysis process, the obtained SrTaO2N photoanodes were washed with sufficient deionized water and dried overnight in air before testing. BaTaO2N and CaTaO2N films were synthesized by using the corresponding flux agents of BaCl2 (74%)/BaCO3 (26%) and CaCl2 (74%)/CaCO3 (26%), respectively. SrNbO2N and BaNbO2N films were synthesized at the same condition by using Nb foil as a precursor and corresponding flux of SrCl2 (74%)/SrCO3 (26%) and BaCl2 (74%)/BaCO3 (26%), respectively.

4.2. 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 10 mM CoSO4 and 40 mM potassium acetate. First, 4 mC cm−2 of cobalt was electrodeposited at -0.8 V vs. SCE. Then, the 4 mC cm−2 of the CoOOH layer was electrodeposited in chronopotentiometry mode at 0.05 mA for 80 s under illumination of a xenon lamp (Ushio, Optical Module X500).

4.3. Structural Characterization

The crystal structures of all the samples were measured by X-ray diffraction (XRD, Rigaku Ultima III) with Cu Kα radiation (λ = 1.54056 Å) at 40 kV and 40 mA. The optical absorption spectra of the SrTaO2N photoanodes were measured with a UV-visible (UV-vis) spectrophotometer (Shimadzu, UV-vis 2550). The optical band edge of the samples was determined by Tauc plots. The morphology of the films was observed by field-emission scanning electron microscopy (FE-SEM, Nova NanoSEM 230, FEI). TEM and HRTEM analysis was performed on a high-resolution transmission electron microscope (JEM-200CX). TG-DTA coupling mass spectroscopy was carried out on TG-MS equipment (NETZSCH, STA 449 F3-QMS 403 C Aëolos, Germany) in the flow of mixed N2/NH3 (4 mol%).

4.4. PEC Characterization

The PEC activity of SrTaO2N photoanodes was measured in a three-electrode configuration cell with a potentiostat (CHI633C) using 1 M NaOH (pH = 13.6) as the electrolyte, the SrTaO2N photoanodes as a working electrode, Hg/Hg2Cl2 in saturated KCl as the reference electrode, and Pt foil as the counter electrode. The CoOOH catalyst was activated by scanning from the open-circuit potential to 2.5 V vs. RHE in 1 M NaOH before the PEC measurements [36]. Measured potentials vs. Hg/Hg2Cl2 reference electrode were calibrated to reversible hydrogen electrode (RHE). AM 1.5G simulated sunlight (100 mW cm−2) 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 SrTaO2N 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.

4.5. 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.