Tailoring the Surface Properties of Bi2O2NCN by in Situ Activation for Augmented Photoelectrochemical Water Oxidation on WO3 and CuWO4 Heterojunction Photoanodes

Bismuth(III) oxide-carbodiimide (Bi2O2NCN) has been recently discovered as a novel mixed-anion semiconductor, which is structurally related to bismuth oxides and oxysulfides. Given the structural versatility of these layered structures, we investigated the unexplored photochemical properties of the target compound for photoelectrochemical (PEC) water oxidation. Although Bi2O2NCN does not generate a noticeable photocurrent as a single photoabsorber, the fabrication of heterojunctions with the WO3 thin film electrode shows an upsurge of current density from 0.9 to 1.1 mA cm–2 at 1.23 V vs reversible hydrogen electrode (RHE) under 1 sun (AM 1.5G) illumination in phosphate electrolyte (pH 7.0). Mechanistic analysis and structural analysis using powder X-ray diffraction (XRD), scanning electron microscopy (SEM), X-ray photoelectron spectroscopy (XPS), and scanning transmission electron microscopy energy-dispersive X-ray spectroscopy (STEM EDX) indicate that Bi2O2NCN transforms during operating conditions in situ to a core–shell structure Bi2O2NCN/BiPO4. When compared to WO3/BiPO4, the in situ electrolyte-activated WO3/Bi2O2NCN photoanode shows a higher photocurrent density due to superior charge separation across the oxide/oxide-carbodiimide interface layer. Changing the electrolyte from phosphate to sulfate results in a lower photocurrent and shows that the electrolyte determines the surface chemistry and mediates the PEC activity of the metal oxide-carbodiimide. A similar trend could be observed for CuWO4 thin film photoanodes. These results show the potential of metal oxide-carbodiimides as relatively novel representatives of mixed-anion compounds and shed light on the importance of the control over the surface chemistry to enable the in situ activation.


■ INTRODUCTION
The development of clean, renewable, and long-term sustainable energy sources to help prevent impending climate change while sustaining the global population and economic growth is a colossal challenge. 1,2 To this end, harnessing solar energy through energy conversion technologies represents a promising piece of the puzzle. 2−7 One such pathway uses PEC cells to obtain hydrogen from water upon solar illumination. 7−10 Photochemical water-splitting includes the hydrogen evolution reaction (HER) and the oxygen evolution reaction (OER), which have to be accomplished simultaneously. 11 Kinetics and stability are considered to be critical issues in the water-splitting process, which limits the PEC water-splitting efficiency. 12 Currently, significant efforts are focused on developing efficient photoanodes to accelerate the sluggish four-electron transfer oxidation reaction and reduce energy consumption. 13−16 A plethora of various semiconductors have been extensively investigated, including binary and ternary oxides. 17−22 Because the semiconductor surface suffers from low catalytic activity, it is crucial to optimize the charge transport ability and recombination rate of the photogenerated carriers. 23 Modifying the photoanode of a light-absorption semiconductor with photoelectrocatalysts or electrocatalysts is a promising strategy to tackle the kinetics demand. 24−27 Different light absorbers and electrocatalysts has been used to construct advantageous heterojunction photoanodes. 7,28−31 Bismuth(III) based semiconductors, such as Bi 2 O 3 , BiPO 4 , BiVO 4 , and the mixed-anion compounds Bi 2 O 2 S and BiOCl, have received much attention as photoanodes for PEC watersplitting. 32−42 The merits of such bismuth oxides include nontoxicity, low cost, chemical stability, and good photo-chemical transport properties which are proposed to originate from the dispersed nature of the Bi states in the vicinity of the valence band edge (VBE) and conduction band edge (CBE), thereby providing efficient electron−hole separation. 38 Meanwhile, metal carbodiimides have recently drawn considerable interest as novel materials in photochemical energy conversion systems. 43−48 Indeed, composite heterojunction photoanodes modified with metal carbodiimides have been shown to display a high charge separation efficiency. 46 Up to now, there has only been one example of the application of mixed-anion compounds based on a metal oxide-carbodiimide, i.e., Sn 2 ONCN, for water-splitting. 48 We recently discovered the novel compound Bi 2 O 2 NCN, which is a semiconductor with an electronic band gap of 1.8 eV and displays a layered structure. 49 Driven by curiosity regarding if this compound may be also suited for water oxidation, we have discovered in our present study that the title compound undergoes an electrolyte-mediated in situ activation when coupled to WO 3 and CuWO 4 photoanodes.

■ EXPERIMENTAL SECTION
Synthesis of Bi 2 O 2 NCN. Bi 2 O 2 NCN was prepared in an argonfilled glovebox by a solid-state metathesis reaction. 49 BiOCl and Na 2 NCN were mixed in a 2:1 molar ratio and ground in an agate mortar under argon. The reaction mixture of 500 mg was sealed in an open dry glass capillary (8 mm). The obtained sample was then placed in a glass ampule and was heated in a tube furnace under flowing argon gas to 350°C for 2 h, with heating and cooling rates of 2°C min −1 . The resultant powder was subsequently opened in air. After being washed with deionized water and dried in an oven at 80°C for 4 h, the product Bi 2 O 2 NCN was obtained. Synthesis of BiPO 4 . A 0.485 g portion of Bi(NO 3 ) 3 ·5H 2 O was dissolved in 90 mL of aqueous solution containing 10% glycerol by ultrasound. When completely dissolved, 0.136 g of KH 2 PO 4 was added into the above mixture under vigorous stirring which was maintained under stirring for 2 h. The resultant white suspension was centrifuged and washed alternately with deionized water and ethanol three times before being oven-dried at 120°C for 8 h to obtain a powder sample of hexagonal BiPO 4 . 50 Synthesis of WO 3 Thin Films. WO 3 thin films were produced on fluorine doped tin oxide (FTO) glass (4 cm × 1.8 cm, 2.2 mm thick, Sigma-Aldrich) by a hydrothermal synthesis method. FTO substrates were ultrasonically cleaned in diluted nitric acid, acetone, and ethanol for 15 min each in sequence and then dried in an ambient atmosphere. A 0.165 g portion of sodium tungstate dihydrate (Na 2 WO 4 ·2H 2 O, 99.9%, Acros Organics) and a 0.126 g portion of H 2 C 2 O 4 ·2H 2 O were dissolved in 5 and 10 mL of deionized water by stirring, respectively. The two solutions were then mixed with stirring, and 10 mL of 1 M HCl was added and stirred for 10 min. A 6 mL portion of the mixed solution was transferred to a 20 mL Teflon-lined stainless steel autoclave, where a FTO substrate was placed inside with the conducting side facing down and leaning against the inner wall. The autoclave was tightly sealed and heated at 180°C for 2 h, and then it was cooled to room temperature. After that, the FTO glass was cautiously washed with water and dried in the air. The monoclinic WO 3 thin film grown on an FTO substrate could be achieved after annealing at 550°C for 1 h and then cooling to room temperature under an ambient atmosphere.
Synthesis of CuWO 4 Thin Films. CuWO 4 electrodes were prepared as in our previous work. 48 A 1.26 g portion of sodium tungstate dihydrate (Na 2 WO 4 ·2H 2 O, 99.9%, Acros Organics) was dissolved in 15 mL of deionized water, and 1 mL hydrogen peroxide (30%) was added to the tungstate solution. The latter was stirred for 20 min at room temperature. A 25 mL portion of deionized water and 25 mL of isopropanol (>99.7%,) were added to the solution. A 0.73 g portion of copper(II) nitrate trihydrate (Cu(NO 3 ) 2 ·3H 2 O, >99%, Sigma) in 10 mL of deionized water was added to the tungsten precursor solution. The pH value was adjusted to 1.2 by nitric acid, and the solution was used for electrochemical deposition on FTO glass. The deposition was performed in a three-electrode setup with platinum wire and 1 M Ag/AgCl as a counter electrode and a reference electrode, respectively. The potential was swept in the range from −0.9 to +0.2 V vs 1 M Ag/AgCl for 12 cycles at a scan rate of 50 mV s −1 . After that, the working electrode was washed with deionized water, dried at room temperature, and heated at 450°C for 2 h under ambient atmosphere. The excess copper oxide was etched by immersing the electrode into 0.5 M HCl for acidic treatment. The CuWO 4 thin film grown on FTO substrate could be achieved after annealing one more time at 450°C for 30 min under ambient atmosphere.
Preparation Structural Characterization. Powder XRD patterns were recorded in transmission mode on a STOE STADI-P diffractometer (Cu Kα 1 radiation) operating with a DECTRIS Mythen 1K detector at a scan rate of 2°min −1 in the 2θ range from 10°to 90°.
SEM images were recorded on a Leo Supra 35VP SMT (Zeiss) thermal field emission scanning electron microscope operating at an accelerating voltage of 10.0 kV. TEM images were recorded on a Themis Z TEM (Thermo Fisher), and a SuperX energy-dispersive X-ray (EDX) detector operating at 300 kV in the scanning TEM mode was used for elemental mapping. XPS spectra were recorded by a hemispherical VG SCIENTA R3000 analyzer using a monochromatized aluminum source Al Kα (E = 1486.6 eV) at constant pass energy of 100 eV. The binding energies were referenced to the Au 4f core level (E b = 84.0 eV). The composition and chemical surrounding of the sample surface were determined on the basis of the areas and binding energies of Na 1s, K 2p, P 2p, O 1s, N 1s, C 1s, and Bi 4f photoelectron peaks. The fitting of the high-resolution spectra was obtained through the Casa XPS software.
Ultraviolet−visible (UV−vis) spectroscopy was performed on a Shimadzu UV-2600 spectrophotometer. Measurements were recorded in absorbance mode. The Tauc plots were calculated by the Kubelka− Munk function F(R) = (1 − R) 2 /2R for determination of the electronic band gap.
PEC Measurements. The PEC experiments were measured with a potentiostat (Gamry instruments) using an electrochemical cell operating in a three-electrode setup system. In this system, the photoanode, platinum wire, and a 1 M Ag/AgCl electrode function as a working electrode, counter electrode, and reference electrode, respectively. All current values of the electrodes were recorded vs 1 M Ag/AgCl reference electrode and converted to vs RHE according to A solar light simulator (class-AAA 94023A, Newport) with an ozonefree 450 W xenon short-arc lamp was used to illuminate the photoanode with AM 1.5G simulated visible light, which was calibrated with a Si reference cell (LOT-Quantum Design, Germany). Milli-Q water (18.3 Ω cm) was used to prepare the 0.  Ch oxide chalcogenides (Ch = S, Se, and Te; n.b., the S analogue adopts an orthorhombically distorted low-symmetry modification), thereby highlighting the divalent nitride or pseudochalcogenide nature of NCN 2− . 49 The higher degree of electronegativity of NCN 2− relative to sulfide yields a band gap that is intermediate between that of Bi 2 O 2 S (1.12 eV) and β-Bi 2 O 3 (2.48 eV) ( Figure S1). 51,52 For structural characterization, a portion of the Bi 2 O 2 NCN photoanode was physically removed after PEC OER in KP i at pH 7.0. SEM images show that the Bi 2 O 2 NCN exhibits porous globular shapes, the morphology of which shows no visible change before ( Figure 1a) and after (Figure 1b) PEC water oxidation. The PXRD patterns of Bi 2 O 2 NCN show that the compound is structurally stable in the bulk (Figure 1c). Due to the low amount of catalyst loading to WO 3 thin films, the corresponding PXRD patterns of the modified photoanodes contain only the diffraction peaks of WO 3 , which remain unchanged after PEC OER (Figure 1d).
The band gaps of Bi 2 O 2 NCN and WO 3 were determined from the UV−vis absorption spectra (Figure 2a,b). After conversion to Kubelka−Munk-transformed reflectance spectra, the band gaps of Bi 2 O 2 NCN and WO 3 were determined to be    Figure S3). In contrast, the bare Bi 2 O 2 NCN and BiPO 4 photoanodes with the same amount of material as for the composite photoanode developed only a negligible photocurrent density under the same operation conditions. Significantly, the produced photocurrent density of the composite photoanodes is higher than the sum of its individual components, indicating that a synergistic effect occurs between the WO 3 and Bi 2 O 2 NCN or BiPO 4 catalysts. This trend is more visible in CA at 1.23 V vs RHE under chopped backlight AM 1.5G illumination ( Figure  3b). The improved PEC activity was consistent with the increase of IPCE after functionalization with Bi 2 O 2 NCN ( Figure S4). The prolonged CA of the composite electrode is shown in Figure S5.
To understand the origin of the increased photocurrent density upon modification, the hole collection efficiency (η hc measured) was studied by introducing Na 2 SO 3 as a hole scavenger. The oxidation reaction of sulfite to sulfate is faster than the oxidation of water. Measurements were performed in 0.1 M KP i (pH 7.0) with or without 0.05 M Na 2 SO 3 under backlight AM 1.5G illumination (Figure 3c). In this way, the number of holes reaching the semiconductor−electrolyte interface in the reaction can be estimated. The η hc can be calculated by the ratio of photocurrent density for oxidation of sulfite (J Na2SO3 ) and water (J H2O ): η hc = J H2O /J Na2SO3 . In comparison with the WO 3 photoanode, the η hc of the composite photoanode is increased (Figure 3d), indicating that the reactivity of the surface is augmented after modification with Bi 2 O 2 NCN or BiPO 4 . It should be noted that functionalization with BiPO 4 gave slightly higher η hc values between 0.8 and 1.23 V in comparison to the Bi 2 O 2 NCNderived phosphate catalyst. Nevertheless, the advantage of the latter is the semiconducting core for improved charge separation.
To investigate the kinetics of charge transfer in the composite electrode system, we carried out EIS measurements. Figure 3e shows the EIS Nyquist plots which were measured in KP i electrolyte (pH 7) at a bias of 1 V RHE under AM 1.5G illumination. The Nyquist plot could be interpreted by the equivalent circuit as displayed in the inset. In the equivalent circuit, R s simulates the series resistance, Q 1 simulates the constant phase element (CPE) for the electrolyte/electrode  MS analysis was conducted in 0.1 M KP i electrolyte with an applied frequency of 100 Hz. The positive slopes of the plots are in agreement with the expected n-type behavior, while the reduced slope after modification of WO 3 hints toward improved charge-carrier transport (Figure 3f). Furthermore, the charge-carrier density of WO 3 @Bi 2 O 2 NCN was higher than that of WO 3 @bare BiPO 4 . The determined flat band potentials are within the range of WO 3 , i.e., the relatively bulk thin film before modification.
Structural Analysis after PEC Water Oxidation. To further identify the origin for improved PEC activity after modification, we analyzed the surface properties of a The dominant doublet, corresponding to the Bi 3+ species, is observed at 159.7 eV (Bi 4f 7/2 ) and 165.0 eV (Bi 4f 5/2 ). Furthermore, traces of metallic Bi appear on the surface, which are manifested by the photoemission at 157.3 eV (Bi 4f 7/2 ) and 162.6 eV (Bi 4f 5/2 ). It can be assumed that a significant amount of bismuth forms the phosphate phase, i.e., a core−shell structure Bi 2 O 2 NCN@ BiPO x . The formation of this shell is confirmed by the XPS P 2p spectrum (Figure 4), in which two peaks at 133.1 eV (P 2p 3/2 ) and 134.0 eV (P 2p 1/2 ) typical of phosphate species are identified. Phosphorus was also structurally determined by TEM EDX analysis (Figure 4).
Complementary structural analysis by means of high-angle annular dark field (HAADF) images shows the elemental mapping obtained with EDX. The presence of bismuth, oxygen, nitrogen, carbon, and phosphorus was also confirmed, which indicates that the carbodiimide is not further oxidized upon formation of the protective phosphate shell. The situation is therefore comparable to the stabilization of metastable metal oxynitride photoanodes. Attempts to characterize the Bi 2 O 2 NCN electrode after PEC water oxidation by means of HRTEM were not successful due to beam damage. Figure S6a shows the TEM image of the Bi 2 O 2 NCN particles, which decomposed rapidly into Bi 2 O 3 nanosheets ( Figure S6b−d). The beam damage is highly likely due to the carbodiimide anion of the Bi 2 O 2 NCN.
On the basis of a combination of XPS, TEM, SEM, and XRD analyses, the improved WO 3 photoanode performance after modification with Bi 2 O 2 NCN is attributed to the formation of an oxide/oxide-carbodiimide heterojunction. A phosphate-type shell on the Bi 2 O 2 NCN surface was formed after the PEC experiment in a phosphate electrolyte. The catalytic activation of the heterojunction was formed between the BiPO x shell with Bi 2 O 2 NCN core and WO 3 . The observed PEC behavior of Bi 2 O 2 NCN is different from CoNCN, which retains the same chemical composition on the surface and the bulk but is similar to MnNCN. 44,47 The latter is known to form an amorphous manganese phosphate shell.
A control experiment was conducted to investigate the effect of the electrolyte-mediation on the PEC behavior. LSV and CA results show that the WO 3 photoanode has almost identical photocurrent density in 1 M Na 2 SO 4 electrolyte and 0.1 M KP i electrolyte at 1.23 V vs RHE under illumination ( Figure 5). It is worth mentioning that the WO 3 electrode starts to produce higher photocurrents in Na 2 SO 4 electrolyte than in KP i electrolyte above approximately 1.20 V vs RHE. Since most CA data are compared in the literature at 1.23 V vs RHE, we have chosen the thermodynamic potential for the CA. After modification with Bi 2 O 2 NCN, a higher photocurrent density is generated in KP i electrolyte than in Na 2 SO 4 at 1.23 V vs RHE (Figure 5b). This result indicates bismuth phosphate to have the dominant role to augment the charge-carrier transport.
The interface formation of semiconductors with different VBE and CBE positions can result in improved charge-carrier separation. 35 The energy band diagram for the given semiconductors sheds light on the origin ( Figure 6). Compared with the CBE position of WO 3 , the higher CBE position of Bi 2 O 2 NCN facilitates the injection of photogenerated electrons into WO 3 with a concomitant diffusion of the photogenerated holes from WO 3 to Bi 2 O 2 NCN. WO 3 / BiPO 4 and WO 3 /Bi 2 O 3 heterojunction catalysts have been previously evaluated with respect to their PEC degradation of rhodamine B. 56−58 These results demonstrate an increased region of the absorption spectrum under visible light illumination and an efficient transfer and separation of charge carriers by synergistic effect between its components. Moreover, a surface oxygen vacancy may also be induced for WO 3 / Bi 2 O 2 NCN@BiPO x , which is a complex structure system. 59,60 As a result, a synergistic effect of the novel WO 3 /Bi 2 O 2 NCN@ BiPO x heterojunctions led to a boosted photocatalytic performance of the reaction system.
We investigated whether Bi 2 O 2 NCN can be coupled with other semiconductor materials besides the binary WO 3 to form a heterojunction photoanode with improved performance. We chose the ternary oxide CuWO 4 which exhibits a narrower band gap than the current best-performing oxidic semiconductor BiVO 4 used for photoanodes. 48 Similar to the measurements on the WO 3 photoanodes, the fabricated thin films of CuWO 4 were evaluated for PEC water oxidation with respect to the following parameters: (i) sulfate vs phosphate electrolyte and (ii) Bi 2 O 2 NCN vs BiPO 4 . The summarized results of the LSV and CA presented in Figure 7 show that the in-situ-activated Bi 2 O 2 NCN outperforms BiPO 4 . The photocurrent density of the composite photoanode is increased by 85% at 1.23 V vs RHE upon modification with Bi 2 O 2 NCN in comparison to the pristine CuWO 4 photoanode. These observations are consistent with the results obtained for the WO 3 photoanodes and demonstrate that the Bi 2 O 2 NCN@ BiPO 4 core−shell structure, which is only formed in the phosphate electrolyte, can be successfully applied to other oxide semiconductors if the band gaps are matched. Moreover, the PEC OER could be increased after modification with a layer of CoP i as cocatalyst on the surface of WO 3 /Bi 2 O 2 NCN and CuWO 4 /Bi 2 O 2 NCN photoanodes ( Figure 8).

■ CONCLUSION
The photochemical behavior of Bi 2 O 2 NCN and its application as a functional modification material to WO 3 and CuWO 4 electrodes for PEC OER have been investigated. The modified photoanode shows an augmented photocurrent effect during PEC water oxidation as a consequence of electrolyte-mediated in situ activation to a Bi 2 O 2 NCN@BiPO 4 core−shell structure, which has been confirmed by complementary XPS, XRD, and STEM EDX analysis. The Bi 2 O 2 NCN@BiPO 4 core−shell structure outperforms a bare BiPO 4 catalyst while the semiconducting oxide-carbodiimide core facilitates chargecarrier separation across the formed type-II heterojunction. Changing the electrolyte from phosphate to sulfate results in a lower photocurrent and shows that the electrolyte determines the surface chemistry and mediates the PEC activity of the metal oxide-carbodiimide. The results illustrate the potential of metal oxide-carbodiimides as relatively novel representatives of mixed-anion compounds. The study demonstrates that the incorporation of the less ionic carbodiimide anion into an oxidic structure increases the theoretical light absorption, but that at the same time the labile NCN 2− anion opens opportunities to tailor the surface chemistry to alter charge transfer kinetics.