Synergistic Spatial Confining Effect and O Vacancy in WO3 Hollow Sphere for Enhanced N2 Reduction

Visible-light-driven N2 reduction into NH3 in pure H2O provides an energy-saving alternative to the Haber–Bosch process for ammonia synthesizing. However, the thermodynamic stability of N≡N and low water solubility of N2 remain the key bottlenecks. Here, we propose a solution by developing a WO3−x hollow sphere with oxygen vacancies. Experimental analysis reveals that the hollow sphere structure greatly promotes the enrichment of N2 molecules in the inner cavity and facilitates the chemisorption of N2 onto WO3−x-HS. The outer layer’s thin shell facilitates the photogenerated charge transfer and the full exposure of O vacancies as active sites. O vacancies exposed on the surface accelerate the activation of N≡N triple bonds. As such, the optimized catalyst shows a NH3 generation rate of 140.08 μmol g−1 h−1, which is 7.94 times higher than the counterpart WO3-bulk.


Introduction
Ammonia synthesis becomes one of the most important industrial reactions due to the widespread use of NH 3 as fertilizers, pharmaceuticals, chemical feedstocks, and a promising clean energy carrier [1][2][3][4].Up until now, NH 3 production has greatly relied on the Haber-Bosch nitrogen fixation process, which was awarded as one of the most significant inventions of the 20th century.However, the reductant H 2 used in this process is mainly derived from the steam reforming of biomass, and the reaction occurs under stringent reaction conditions (350-550 • C, 200-350 atm), which consumes up to ~2% of the total energy each year [5][6][7][8].The excessive consumption of fossil energy and massive carbon dioxide emissions make this process not conducive to the sustainable development of mankind.Therefore, environmentally friendly NH 3 synthesis via a renewable route is highly desirable [9,10].
Photocatalytic N 2 reduction with H 2 O assembles the advantages of low cost, safety, and environmental friendliness to produce ammonia [11,12].The mechanism of photocatalytic N 2 reduction is as follows: N 2 is first absorbed onto the surface of the photocatalyst.Due to the strong interaction between N 2 and active sites, the sturdy triple bond is weakened, and the N 2 molecule is activated.Under irradiation, the photogenerated electrons on the conduction band of semiconductors transfer to the activated N 2 molecules, reducing N 2 into NH 3 with the participation of water.The evolved NH 3 is desorbed from the catalyst surface.Since the pioneering work reported by Schrauzer on using Fe-doped TiO 2 , extensive efforts have been made to broaden the family of photocatalysts for N 2 reduction [13].So far, diverse catalysts have been developed for photocatalytic N 2 reduction, including BiOBr, InVO 4 , Bi 2 WO 6 , Bi 2 Sn 2 O 7, and so on [14][15][16][17][18].Despite much progress, the overall efficiency is still much less than satisfactory due to the bottlenecks of the extremely low solubility of N 2 in H 2 O and the large dissociation energy of N≡N (941 kJ mol −1 ) [19,20].To tackle these, morphology and structure optimization to regulate the physical and chemical properties of photocatalysts may be effective strategies [21][22][23][24][25].
Hollow materials illustrate distinctive physicochemical properties due to the unique structure of the confined thin shell layer and internal cavity [26][27][28], presenting as ideal candidates for photocatalytic N 2 reduction.The spatial confining effect of the internal cavity is beneficial for restricting N 2 , enriching N 2 molecules on the surface of the photocatalyst, thereby accelerating reaction kinetics.Moreover, the thin shell ensures the sufficient exposure of active sites.On the other hand, surface defects on photocatalysts with highly localized electronic structural changes are efficient active sites for N 2 adsorption and activation [29][30][31][32].Especially for the O defect, the electron-rich environment is beneficial for the cleavage of the N≡N triple bond via the π-back-donation into N 2 antibonding orbitals, achieving a smoother hydrogenation process [33].For instance, Mi and coworkers have reported the preparation of TiO 2 with O defects via a solid phase reduction with NaBH 4 .The normalized N 2 photofixation rate is 324.86 µmol h −1 g −1 , which is 3.85 times that of the original TiO 2 .The experimental and theoretical calculation results suggest that the introduced O defects play a dual role in accelerating the photocatalytic N 2 reduction efficiency.An optimized concentration of O defects as electron acceptors can increase the charge separation efficiency.On the other hand, N 2 adsorbed on OVs can be dissociated and activated through the transfer of electrons into the antibonding orbital of N 2 , thus weakening the strong triple bond [34].Thus, it is desirable to optimize N 2 reduction performance by introducing O defect active sites on hollow-structure photocatalysts.
WO 3 has been widely adopted as a promising photocatalyst due to its appropriate band gap (~2.4 eV), low conduction band edge potential, and fleet electron transport rate [35][36][37].Moreover, the adjustable crystal structure makes the O defect-rich WO 3 the hotspot in N 2 reduction.For instance, Wang and coworkers have reported that the light-induced O defects at the grain boundaries of porous WO 3 can greatly enhance photocatalytic N 2 reduction performance [38].Mechanistic studies reveal that the O defect regulates the band structure of WO 3 , providing sufficient driving force to trigger N 2 reduction.Moreover, O defects can serve as active sites to chemisorb N 2 molecules.Under light, the chemisorbed N 2 molecules accepted photoexcited electrons to generate N 2 H* intermediates with coupled protons, further hydrogenating into NH 3 .Chen and coworkers proposed the modification of Fe single atoms to adjust the electronic structure of O-defective WO 3 and facilitate the adsorption and conversion of N 2 [39].These works have revealed that the introduction of the O defect is an efficient approach for promoting photocatalytic N 2 reduction in thermodynamics, but the synergistic effects of the O defect and hollow structure to further promote N 2 reduction in dynamics have not been reported.
Herein, we design a series of WO 3 hollow spheres with O defects (WO 3−x -HS) to accomplish efficient photocatalytic N 2 reduction into NH 3 in H 2 O.The as-synthesized WO 3−x -HS demonstrates a superior NH 3 evolution rate of 140.1 µmol g −1 h −1 , which is 7.96 times higher than that of pristine bulk WO 3 .The experimental analysis reveals that the uniform hollow sphere structure provides WO 3−x -HS with a relatively larger specific surface area and a confined cavity for the adsorption and enrichment of N 2 molecules on the photocatalyst, while the O defect promotes the activation of the inert N≡N triple bond.

Results and Discussion
The schematic synthetization of WO 3−x -HS is illustrated in Figure 1.Specifically, the carbon sphere has been first prepared with glucose as a precursor via a hydrothermal process.Then, W 6+ is adsorbed onto the surface of the carbon sphere.The mixture is calcined at 450 • C to remove the template carbon sphere, and the hollow sphere structure WO 3−x -HS is obtained.The crystal structures of WO 3−x -HS and bulk WO 3 are investigated via X-ray diffraction (XRD).The diffraction peaks of both samples can be matched well with monoclinic tungsten trioxide (PDF#20-1324) (Figure S1).The characteristic peaks at 23.1 • , 23.7 • , 24.1 • , 28.8 • , 33.3 • , 33.6 • , and 34.0 • correspond to the (001), (020), ( 200), (111), (021), (201), and (220) planes of monoclinic WO 3 , respectively.The morphologies of the as-synthesized samples are carefully investigated via scanning electron microscopy (SEM) and transmission electron microscopy (TEM).As displayed in Figure 2a, the obtained carbon sphere exhibits a typical and uniform spherical morphology.Figure 2b-e presents the WO 3−x -HS samples synthesized with different amounts of template.When a 0.1 g carbon sphere is used, the obtained sample shows an amorphous structure, which may be due to the insufficient template.With the increasing amount of carbon sphere, the hollow sphere structure formed gradually.WO 3−x -HS synthesized with the addition of a 0.25 g template presents the most uniform and complete hollow sphere morphology.The surface is clear and smooth.As a counterpart, the WO 3 -bulk exhibits an irregularly shaped block structure consisting of small particles (Figure 2f).The TEM image (Figure 2g) further reveals the sphere structure of WO 3−x -HS.Moreover, the brightness of the outer edge of the WO 3−x -HS is lower than center position, indicating that the prepared WO 3−x -HS is a hollow sphere with a cavity structure (Figure 2h).The high-resolution TEM (HRTEM) of WO 3−x -HS is shown in Figure 2i.Distinct 0.375 nm and 0.369 nm lattice fringes are observed, which correspond to the (020) and (200) crystal planes of tungsten oxide, respectively.Moreover, the elemental mapping results reveal the co-existence of W and O with uniform dispersion in WO 3−x -HS (Figure S2).
It is well known that the photocatalytic performance is closely related to the specific surface area size and porosity of the catalyst [40,41].The N 2 adsorption/desorption isotherms and the pore size distribution of the obtained WO 3−x -HS and WO 3 -bulk products are measured and presented in Figure 3a.Both catalysts have strong interactions with N 2 in the low-pressure region and show H3-type hysteresis loops in the high-pressure region, indicating that both samples have type IV adsorption-desorption isotherms [42].The Brunauer-Emmett-Teller (BET) surface areas of WO 3−x -HS and WO 3 -bulk are determined to be 23.71m 2 g −1 and 8.53 m 2 g −1 , respectively.The larger surface areas of WO 3−x -HS are expected to expose more catalytically active sites and enhance the contact and adsorption of nitrogen molecules.The pore size distribution curves confirm the presence of mesopores (2-50 nm) in both samples, and the porosities of WO 3−x -HS and WO 3 -bulk are 0.077 cm 3 g −1 and 0.043 cm 3 g −1 , respectively.The larger porosity of WO 3−x -HS can accelerate the transport of reactants and products during N 2 photocatalytic nitrogen fixation, providing more opportunities for proton capture and the deep hydrogenation of intermediates in photocatalytic nitrogen fixation.The presence of oxygen vacancy is verified using electron paramagnetic resonance (EPR).As can be seen in Figure 3b, no signal is observed for WO 3 -bulk in the whole range.While a significant resonance peak at g-factor value of 2.003 ascribed to the unpaired electrons is observed, suggesting the existence of abundant O vacancies in WO 3−x -HS [23].The possible O vacancy formation mechanism may be that the removal of the reductive carbon template induces the escape of lattice oxygen.The optical properties of the obtained photocatalysts have been evaluated using diffuse reflectance spectroscopy (DRS).As shown in Figure 3c, the absorption edge of WO 3−x -HS shows a blue-shift when compared with WO 3 -bulk.Moreover, WO 3−x -HS exhibits stronger light absorption properties in the visible light region, which may be due to the existence of a "sub-band" induced by O vacancies.The band energies (E g value) of the catalysts are calculated using Tauc fitting curves (Figure 3d) [43].The E g values of WO 3−x -HS and WO 3 -bulk are determined to be 2.30 and 2.43 eV, respectively.The surface chemical state of elements of as-synthesized samples has been evaluated using an X-ray photoelectron spectroscope (XPS) (Figure 3e,f).The W 4f spectra of both samples are deconvoluted into two doublet peaks, associated with two different states of the W element.The peaks in WO 3 -bulk at the binding energies at 35.39 and 37.55 eV correspond to the W 4f 7/2 and W 4f 5/2 signals of W 6+ .The second doublet peaks at 34.12 and 36.67 eV are assigned to the W 4f 7/2 and W 4f 5/2 of W 5+ .Note that the W 4f binding energies of WO 3−x -HS are lower than those of WO 3 -bulk, indicating a decrease in electron density after the generation of O vacancies [44].The O 1s spectrum of WO 3 -bulk can be fitted with three peaks at the binding energies of 530.22,531.78, and 533.15 eV, which are attributed to the lattice oxygen, O in the surface hydroxyl group, and chemisorbed oxygen, respectively [45].There is a decrease in the binding energy of lattice oxygen in WO 3−x -HS when compared to that of WO 3 -bulk, which is attributable to the increased electron cloud density caused by the formation of O vacancies.
The photocatalytic nitrogen reduction reaction (NRR) activities of the as-synthesized samples have been evaluated with H 2 O and N 2 as feedstocks.Figure 4a shows the performance of WO 3−x -HS prepared with different amounts of carbon sphere as a template.It is interesting to see that the NH 3 generation rates exhibit a trend of increasing initially and decreasing afterwards with the increasing amount of carbon sphere.The WO 3−x -HS synthesized with the addition of a 0.25 g template presents the optimal NRR activity, with an ammonia production rate of 140.08 µmol g −1 h −1 .It can also be seen that the prepared WO 3−x -HS exhibits better performance than most of the reported works on photocatalytic N 2 reduction (Table S1).It seems to match with the SEM results that a higher NRR performance is achieved with a more complete hollow sphere structure.The high performance of NRR over WO 3−x is also confirmed by ion chromatography (Figure 4b).To further reveal the role of hollow structure and O vacancies in effecting the photocatalytic performance, the NRR activity of WO 3 -bulk is tested.As displayed in Figure 4c, the WO 3 -bulk exhibits a relatively low NH 3 generation rate of 17.64 µmol g −1 h −1 .The possible reasons may be that the hollow sphere structure of WO 3−x -HS with a larger specific surface area exposes more active sites of O vacancies to activate the N≡N.The hollow structure with a nanoconfined cavity promotes the reaction kinetics.The photocatalytic performance of WO 3−x -HS under different reaction conditions has been investigated.As can be seen from Figure 4d, when the gas is changed from N 2 to Ar, almost no ammonia is produced.At the same time, ammonia could not be detected without a catalyst or in dark conditions.All these results elucidate that the evolved NH 3 is indeed originated from N 2 through the photocatalytic reduction process.The WO 3−x -HS synthesized with the addition of 0.25 g of template showed the best NRR activity with an ammonia generation rate of 140.08 µmol g −1 h −1 .The prepared WO 3−x -HS outperforms most of the reported research results in photocatalytic reduction of N 2 (Table S1).This seems to be in agreement with the SEM results that a more complete hollow sphere structure can achieve higher NRR performance.
Linear scanning voltammetry (LSV) tests have been performed on WO 3−x -HS and WO 3 -bulk in Ar-and N 2 -saturated electrolyte environments to evaluate the NRR performance.A distinct current density enhancement under N 2 over Ar suggests a potential NRR process driven by both samples (Figure 5a).The photocurrent densities of WO 3−x -HS in a N 2 environment are much higher than those of WO 3 -bulk, indicating better NRR activity [46].Electrochemical characterizations have been performed to reveal the charge transfer and separation behaviors.Figure 5b shows the transient photocurrent response of the sample.The photocurrent density of WO 3−x -HS is significantly higher compared to WO 3 -bulk, indicating that the WO 3−x -HS material is more capable of photo-induced electron-hole pair separation [47].Electrochemical impedance spectroscopy (EIS) is used to monitor charge transfer behavior.As shown in Figure 5c, WO 3−x -HS presents a decreased semicircle compared to the WO 3 -bulk, indicating lower electron transfer resistance [48].These results demonstrate that the O vacancies on WO 3−x -HS contribute to the improved separation and transfer of photoinduced carriers.The evaluation of carrier separation efficiency is conducted via steady-state PL emission spectroscopy.As shown in Figure 5d, under excitation light at a wavelength of 453 nm, there is an obvious PL quenching of WO 3−x -HS compared to WO 3 -bulk, indicating the photogenerated electron-hole recombination is suppressed and the photogenerated carrier utilization efficiency is higher [49].The reason for this result may be that the hollow structure with a relatively thin shell of WO 3−x -HS is beneficial for the migration rate and separation efficiency of photogenerated carriers.The N 2 adsorption performance of the catalysts was characterized using N 2 temperatureprogrammed desorption spectrometry (N 2 -TPD).As shown in Figure 6a, both WO 3−x -HS and WO 3 -bulk show adsorption peaks at low (230-360 • C) and high temperatures (470-650 • C), which are attributed to the physical adsorption and chemical adsorption peaks of N 2 , respectively [50].The peak intensities of WO 3−x -HS are stronger than those of WO 3 -bulk, and the peak position shows a positive shift to the relatively high temperature, which certifies the much-improved N 2 affinity.These results reveal that the designed hollow sphere structure with confined inner micro-space and O vacancies are beneficial for the adsorption of N 2 molecules and for facilitating the NRR process.In situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS) has been tested to reveal the possible N 2 reduction process.The WO 3−x -HS catalyst is exposed to a vapor of N 2 and water to simulate the reaction conditions.The in situ DRIFTS is recorded with different light irradiation times.As shown in Figure 6b, the peak at 1506 cm −1 can be attributed to the OH bending mode of adsorbed water [51].When light is introduced, the newly emerged peak at 3290 cm −1 is attributed to the ν(N-H) stretching mode in the reacting intermediate [52].The signals at about 1212, 1270, and 1392 cm −1 are assigned to bending vibrations of ammonia species and intermediates σ(N-H), while the peaks at about 2883 and 2980 cm −1 are absorption peaks of NH 4 + [51,53].Moreover, these peaks intensified with prolonged irradiation time, revealing that N 2 is gradually hydrogenated to ammonia over the catalyst consistently.
Mott-Schottky tests were performed to evaluate the energy band positions of WO 3−x -HS (Figure 7a).The slope of the linear potential curves of WO 3−x -HS is positive, indicating that the as-synthesized sample is an n-type semiconductor material [54].The flat band potentials of WO 3−x -HS are determined to be −0.33V (vs.Ag/AgCl, pH = 7), corresponding to −0.133 V vs. NHE based on the equation (E NHE = E Ag/AgCl + 0.197) [55].The conduction band of n-type semiconductors is 0-0.1 V higher than the flat-band potential.Here, the voltage difference between the conduction band (CB) and the flat potential is set to 0 V; thus, the minimum conduction band (CB) for WO 3−x -HS is about −0.13 V (vs.NHE, pH = 7).Combined with the DRS result, the valence bands (VB) of WO 3−x -HS are calculated to be 2.17 V (vs.NHE, pH = 7) [56].Thus, the band structure of WO 3−x -HS is shown in Figure 7b.Apparently, it is thermodynamically favorable for the CB electrons to reduce N 2 into NH

The Preparation of Carbon Nanosphere
Carbon nanospheres were prepared by glucose under hydrothermal conditions.An amount of 6 g of glucose was dissolved in 6 mL of deionized water.The solution was transferred to a 100 mL of PTFE reactor and reacted at 180 • C for 12 h in an oven.After the hydrothermal reaction was completed, the supernatant was centrifugally washed with deionized water and ethanol until the ionic solubility of the supernatant was less than 10 ppm.The final product was dried in a vacuum oven at 40 • C for 12 h.

The Preparation of WO 3−x -HS
WO 3−x -HS was prepared by controlling the gradual hydrolysis of WCl 6 on the template carbon sphere particles.WCl 6 was used as the metal precursor and dissolved into N, N-dimethylformamide (DMF) with a concentration of 0.1 g mL −1 denoted as solution A. At the same time, various amounts of carbon nanospheres (0.1, 0.25, 0.4, and 0.5 g) and 0.1 g of cetyltrimethylammonium bromide (CTAB) were dispersed into 50 mL of DMF solution and denoted as solution B. Afterwards, solution A was dropped into solution B under continuous ultrasonic treatment for 30 min.After that, 1 mL of deionized water was added dropwise to make the WCl 6 hydrolyze.The suspension was stirred at room temperature for 24 h, then washed with anhydrous ethanol and water, and finally freeze-dried.The obtained product was calcined in a muffle furnace at 450 • C for 2 h and marked as WO 3−x -HS.

The Preparation of Bulk WO 3
Firstly, 4 g of Na 2 WO 4 •2H 2 O was dissolved in 20 mL water to obtain a homogeneous solution, then 50 mL HNO 3 solution (1 mol L −1 ) was dropped into the above solution and stirred for another hour.The precipitate was washed with deionized water and ethanol and dried in a vacuum oven at 60 • C. Finally, the dried sample was grounded and calcined at 450 • C for 2 h, denoted as WO 3 -bulk.

Photocatalytic N 2 Fixation
The photocatalytic nitrogen fixation reaction was carried out at room temperature by first adding the photocatalyst (0.015 g) and 50 mL of deionized water to a quartz reactor with a quartz top and sonicating for 3 min.The reactor was then evacuated and subsequently bubbled with ultrapure N 2 for 30 min in the darkness.Finally, the reaction was carried out under the illumination of a 300 W xenon lamp (CEL-HXF300-T3 with filter (420 nm < λ < 800 nm)) for 1 h with continuously bubbling N 2 in the whole reaction process.A quantity of the reaction solution was collected, filtered, and then assayed for ammonia concentration by the indophenol blue method.The average ammonia concentration was calculated by three parallel experiments.The calibration curve of NH 4+ measured by UV-vis spectra is displayed in Figure S3.

Characterization
The crystal structure of the photocatalysts was analyzed by X-ray electron diffraction on a Bruker D8 Advance X-ray diffractometer.The morphological and lattice structure information was characterized using scanning electron microscopy (Hitachi SU8000, Tokyo, Japan) and transmission electron microscopy (FEI Talos F200s, Hillsboro, OR, USA).Ultraviolet-visible diffuse reflectance spectroscopy was measured on a Varian Cary 500 UVvis spectrophotometer.The chemical states of the prepared samples were characterized via X-ray photoelectron spectroscopy on a VG Scientific ESCA Lab Mark II spectrometer.The binding energies of all tested elements were calibrated by C 1s at 284.6 eV.Electron paramagnetic resonance measurements were performed on a Bruker A300 EPR spectrometer.Electrochemical tests were performed on a ZENNIUM IM6 electrochemical workstation (Zahner, Kronach, Bayern, Germany).The chemisorption properties of the catalyst for N 2 were characterized using N 2 temperature-programmed desorption spectrometry on an AutoChem II 2920 (Micromeritics) instrument with a thermal conductivity detector.PL was measured by a fluorophotometer (Edinburgh FLS1000) with an excitation wave length of 453 nm.

Electrochemistry Measurement
Fluorine-doped tin oxide (FTO) glass was used for the preparation of the working electrode.The FTO glass was first washed by sonication in acetone and ethanol for 30 min.Next, 5 mg of the as-synthesized samples were added to 0.5 mL N, N-dimethylformamide.The mixture was continuously sonicated for 2 h to obtain a uniform slurry.Then, 10 µL of the obtained homogeneous slurry was dropped onto the FTO side with exposed areas of 0.25 cm 2 .The remaining parts that were not coated were sealed with epoxy resin.Finally, the electrode was dried in nature.A conventional three-electrode cell was used for the electrochemical tests.A Pt plat and a Ag/AgCl electrode were adopted as counter electrodes and reference electrodes, respectively.Na 2 SO 4 aqueous was used as an electrolyte.

Conclusions
In summary, we have developed a WO 3−x hollow sphere with oxygen defects for efficient N 2 photoreduction into NH 3 .It is revealed that the hollow sphere structure greatly promotes the enrichment of N 2 molecules in the inner cavity and facilitates the chemisorption of N 2 onto WO 3−x -HS.O vacancies exposed on the surface accelerate the activation of the N≡N triple bond.Moreover, the thin shell of the hollow structure facilitates the photogenerated charge transfer.As such, the WO 3−x -HS exhibits better photocatalytic NRR activity with an optimal NH 3 generation rate of 140.08 µmol g −1 h −1 , which is 7.94 times higher than the counterpart WO 3 -bulk.This work validates an avenue for designing active photocatalysts toward N 2 reduction via morphology and structure regulation.

Figure 1 .
Figure 1.Schematic illustration of the synthesis of WO 3−x -HS.

Figure 4 .
Figure 4. Photocatalytic N 2 reduction performance of WO 3−x -HS-w prepared with different amounts of carbon sphere as template (a); comparison of two methods for determining ammonia concentration (b); comparison of photocatalytic N 2 reduction performance of WO 3−x -HS and WO 3 -bulk (c); control experiments of WO 3−x -HS for photocatalytic N 2 reduction performance under different conditions (d).
3 and VB holes to oxidize H 2 O.A possible NRR mechanism for WO 3−x -HS has been proposed.In the reaction system, N 2 molecules are confined and enriched in the inner cavity of WO 3−x -HS.Beneficiating from the thin shell of the hollow structure, O vacancies in WO 3−x -HS are sufficiently exposed as active sites.Thus, the activation of the N≡N triple bond is achieved by the electron-rich environment of O vacancies via the π-back-donation into N 2 antibonding orbitals.Upon irradiation, photogenerated holes on the VB of WO 3−x -HS oxidize H 2 O into O 2 and release active protons.The electrons accumulated on the CB reduce the activate *N 2 into NH 3 .

Figure 7 .
Figure 7. Mott-Schottky plots of WO 3−x -HS (a) and a schematic illustration of the band structure of WO 3−x -HS (b).