English

• Converting greenhouse gasses CO₂ in atmosphere into value-added chemical products holds promise to address the current and future demand of energy supply and environmental disasters [1, 2]. Although photo-induced CO₂ reduction on a semiconductor is appealing and attractive [3-5], the difficulties in thermodynamics and kinetics make its process complicated and challenging. In particular, the adsorption and activation of high chemical inertness of linear CO₂ on the surface of semiconductor is very difficult [6]. Oxygen vacancies are the most prevalent and widely studied anion defects with a relatively lower formation energy on heterogeneous oxide surfaces [7-11]. Normally, oxygen vacancies on the catalyst surfaces can effectively modulate both the coordination structures and electronic states of reactive molecules, thus influencing the reaction kinetics and further altering the reaction mechanisms [12-14]. However, an in-depth scenario depicting the interactions between oxygen vacancies and reactive molecular in complicated reactions is still unclear, particularly for the challenging CO₂ reduction reactions [15]. Oxygen vacancies can favor the CO₂ binding and subsequent CO₂ activation, as well as the final dissociation to promote the catalytic process [16-18]. On the other hand, oxygen vacancies have extensively recognized to serve as the activation sites for the thermodynamically stable molecules such as CO₂ and N₂ [19-22]. However, in a two-step thermochemical CO₂ reduction process, oxygen vacancies created by endothermic reduction of metal oxides were found to induce the effective dissociation of CO₂ molecules acting as Lewis basic sites [23]. Although oxygen vacancies can facilitate the CO₂ reduction in numerous reactions, intrinsic mechanism on how the high chemical inertness of linear CO₂ molecules are activated and converted is inconclusive and has not fully elucidated.

  It is accepted that the CO₂ adsorption and activation on a heterogeneous photocatalyst involves the formation of a partially charged species CO₂^δ− to lower the activation barrier, in which process electron transfer from surface atoms to CO₂ molecules is dominated [24]. Subsequent reduction of CO₂^•‒ into carbon-containing products involves the electron and proton transfer and breaking C=O bonds [25, 26]. The complex chemical environment of the photocatalyst surfaces leads to the formation of diverse intermediates, further causing complicated reaction pathways [27]. Several possible reaction pathways were proposed such as the formaldehyde pathway [28], the carbene pathway [29], but the mechanism from CO₂ adsorption to catalytic reaction is still unclear. Furthermore, utilizing water as the electron donor is the ultimate solution of CO₂ photoreduction, but the oxygen evolution half-reaction involves complicated four-electron transfer process and O=O bond formation, which makes the CO₂ reduction process coupling with water oxidation more difficult.

  In this work, taking hexagonal tungsten oxide (WO₃), a semiconductor photocatalyst with unfavorable band structure as a prototype, we experimentally demonstrate the continuous CO₂ reduction to selectively produce CO under light irradiation can be achieved on the hexagonal WO₃. Investigations indicate that light irradiation induces the in-situ formation of oxygen vacancies on the hexagonal WO₃, and the oxygen vacancies promote the absorption and activation of CO₂ molecules, thus making the photo-induced CO₂ reduction takes place via an oxygen vacancies-involved process.

  Hexagonal tungsten trioxide (WO₃) has been intensively studied as a candidate photocatalyst thanks to its advantages in light harvesting properties and chemical stabilities over a wide range of acidity [7-10, 12, 30]. It is widely-held that the conduction band edge of hexagonal WO₃ locates at ~0.2 eV vs. RHE, which is more positive than CO₂ reduction to carbon-containing products energy levels, namely, its conduction band is thermodynamically unfavorable to drive the direct reduction of CO₂ molecules. Analyzing from the microstructure, hexagonal WO₃ is composed of WO₆ octahedral unit cells which share its corner oxygen atom to construct a defect perovskite-like ReO₃ structure (Fig. 1a). The intrinsic structure of hexagonal WO₃ leads to incidental oxygen atom leakage, creating oxygen vacancies. Correspondingly, the decrease ratio of structural W=O double bonds in the unit cell to single bonds is proposed to correlate the oxygen vacancies generation process [31]. Chromic properties result from the color centers is common in the WO₃ material, generally being attributed to electron excess localized on W ions (W⁵⁺), which is closely related to the oxygen vacancies [32].

  Figure 1

  

  Figure 1.  Schematic structure and characterization of as-prepared hexagonal WO₃. (a) Scheme illustration of hexagonal WO₃ crystal structure. (b) XRD pattern of as-prepared WO₃. (c) SEM image of hexagonal WO₃. (d, e) High-resolution TEM images of hexagonal WO₃.

  DownLoad: Full-Size Img PowerPoint

  The hexagonal WO₃ was employed in this work and prepared via a hydrothermal process as reported [33]. Detailed synthesis, characterization and CO₂ photoreduction process was shown in Supporting information. All the XRD diffraction peaks of as-synthesized sample can be well indexed to hexagonal phase WO₃ with no characteristic peak of sub stoichiometric WO₃ or hydrates was observed (Fig. 1b). The morphology of the hexagonal WO₃ exhibits a rod-like shape with diameter of ~50 nm and rod length of 1–2 µm (Figs. 1c and d, Figs. S1 and S2 in Supporting information). A regular lattice fringe spacing about 0.39 nm was shown in high-resolution TEM image, which can be attributed to the representative (001) plane of the hexagonal WO₃, demonstrating that the prepared WO₃ nanorod grew along the [001] crystal direction (Fig. 1e). Mott-Schottky analysis was applied to determine the conduction band edge of as-prepared WO₃ (Fig. S3 in Supporting information). As an n-type semiconductor, the conduction band edge of hexagonal WO₃ was estimated to be 0.07 eV, similar to the results previously reported [34]. The band gap (E_g) was calculated to be 2.68 eV (Figs. S4 and S5 in Supporting information).

  Photo-induced CO₂ reduction on the hexagonal WO₃ was examined under CO₂ atmosphere with the presence of water vapor using Xe lamp as light source. As shown in Fig. 2a, surprisingly, we observed that the continuous producing of CO gas can be obviously detected under light irradiation, which means that CO₂ can be reduced to CO under this condition. As contrast, no CO product was detected in the same condition without light irradiation, illustrating that the CO₂ reduction should be responsive to the light-induced photoreduction process (Fig. 2b). Furthermore, we verified that either no CO was detected in the case when the hexagonal WO₃ was absence (Fig. S6 in Supporting information). Structural evolution under the illumination condition was further confirmed by Raman spectroscopy (Fig. S7 in Supporting information). Moreover, CO₂ reduction performance was also examined under different wavelength ranges. As depicted in Fig. 2c, CO₂ photoreduction towards CO is significantly correlated with the light absorption ranges of hexagonal WO₃, confirming that the CO₂ reduction is associated with the photo excitation of hexagonal WO₃ catalyst and the subsequent charge-involved photoreduction process. Additionally, comparative experiments with or without introducing H₂O vapor were conducted. As can be seen from Fig. 2d, a remarkable difference of the performance was clearly observed under the cases with or without H₂O. When the H₂O vapor was introduced into the reactor, the photoreduction activity was more than an order of magnitude of that without H₂O, indicating that H₂O molecules exactly participate in the CO₂ photoreduction process.

  Figure 2

  

  Figure 2.  (a) Photocatalytic activities of CO₂ reduction on hexagonal WO₃. (b) Control experiments of CO₂ reduction under light and dark conditions. (c) CO₂ photoreduction to CO on hexagonal WO₃ under different wavelengths. (d) The comparison of CO₂ photoreduction with or without introducing H₂O vapor. (e, f) Mass spectra extracted from GC–MS analysis of product with isotope labeling using ¹³CO₂ and H₂¹⁸O.

  DownLoad: Full-Size Img PowerPoint

  Besides, the photocatalytic activity of CO₂ reduction in gas and liquid phase were also estimated. Under the same irradiation condition, the photocatalytic activity in gas phase is slightly higher than that in liquid phase, which may due to the weak solubility of CO₂ (Fig. S8 in Supporting information).

  To certify that the reliable product comes from CO₂ photoreduction rather than other contaminants, isotope labeling experiment was performed using ¹³CO₂ as the reactant, while the product was analyzed by GC–MS. As shown in Fig. 2e, for non-labeled CO₂, molecular ion peak at m/z = 28 indicates that the product is CO, while replacing the reactant by ¹³CO₂, molecular ion peak at m/z = 29 evidently showed up without any other products, indicating the selective formation of ¹³CO, giving clear evidence that the produced CO is originated from the CO₂ input with a high selectivity. To verify the attendance of water, control experiment with ¹⁸O labeling H₂O was also examined. In the mass spectrum shown in Fig. 2f, two peaks appeared at m/z values of 34 and 36, which can be attributed to the ionic ¹⁶O¹⁸O and ¹⁸O₂, respectively. The detected O₂ undoubtedly confirms that the water oxidation takes place during CO₂ photoreduction. Therefore, we can come to a conclusion that the CO₂ photoreduction can be realized on the hexagonal WO₃. It should be noted that although H₂O molecules were confirmed to participate in the reaction process, no detectable H₂ was observed in the experiment, which may be due to fact that the hydrogen species formed in the reaction adsorb on the surface of the catalyst. To investigate the phenomenon, examination of H₂ adsorption was conducted for the hexagonal WO₃ with or without light treatment (Fig. S9 in Supporting information). As expected, we found that the hexagonal WO₃ after photocatalytic reaction under light irradiation shows a drastically enhanced H₂ adsorption capability compared to the fresh WO₃, which is proposed to be the reason why no H₂ but only CO was detected in the experiment.

  The results above evidently demonstrate the occurrence of photo-induced CO₂ reduction to CO on the hexagonal WO₃, however, how the reaction takes place is still unclear. Considering the structure of hexagonal WO₃, each oxygen atom bounds to two W atoms in the lattice, whereas there may be missing oxygen atoms at the boundary or corner sites on the surface. In addition, the photogenerated holes with high oxidation potential may oxidize the lattice oxygen of the hexagonal WO₃ and create oxygen vacancies on the surface, which attends the CO₂ reduction and activation subsequently. As depicted in Fig. 3a, when the hexagonal WO₃ was irradiated under light, the apparent color gradually changes from light gray to dark blue with irradiation time prolonging. UV–vis diffusion reflection spectroscopy was then employed to investigate the absorbance during this color-changing process (Fig. 3b). Obviously, the absorption in the region of 500–800 nm dramatically increases, while the absorption at the band edge keeps nearly constant. In general, the light absorption increase in visible light region for semiconductors can be attributed to the formation of low-valued metal species [35]. Here, the increase in light absorption can be attributed to the generation of pentavalent tungsten (W⁵⁺), which also correlates the generation of oxygen vacancies during the CO₂ photoreduction reaction on the hexagonal WO₃. Electron spin resonance (ESR) spectra were therefore employed to further verify the formation of oxygen vacancies. As depicted in Fig. 3c, no significant signal was observed in dark condition, whereas an obvious signal at g = 2.002 appears once light irradiated, which clearly evidenced the formation of single-electron-trapped oxygen vacancies. Furthermore, the hexagonal WO₃ samples before and after irradiation exhibit comparable XRD diffraction peaks and intensities (Fig. 3d), indicating the maintained crystallinity of the hexagonal WO₃ with light-induced formation of oxygen vacancies, which was also supported by the unchanged lattice spacing of the hexagonal WO₃ compared with the fresh ones (Fig. S10 in Supporting information). Fig. 3e depicts the Raman spectra of the hexagonal WO₃ before and after light irradiation. The bands with Raman shift at 760 cm⁻¹ and 817 cm⁻¹ can be ascribed to O-W-O stretching mode, whilst peaks at 250 and 324 cm⁻¹ belong to the stretching mode of W-O-W [36]. The Raman peak at 940 cm⁻¹ can be assigned to the W=O double bond stretching mode located at the boundaries. After light irradiation to generate oxygen vacancies, the intensity of all characteristic peaks dramatically decreases. Besides, the defects of nanocrystals can be reflected on the width of Raman shift [37]. The peak width at half-height at around 806 cm⁻¹ decreased obviously after irradiation, which supports the formation of oxygen vacancies laterally.

  Figure 3

  

  Figure 3.  (a) Photographs of the hexagonal WO₃ under light irradiation. (b) UV–visible spectra of the hexagonal WO₃ before and after light irradiation. (c) In-situ EPR spectra of the hexagonal WO₃ before and after light irradiation. (d) XRD pattern of the hexagonal WO₃ before and after the photocatalytic reaction. (e) In-situ Raman spectra of the hexagonal WO₃ before and after light irradiation. In-situ FTIR spectra of (f) CO₂ and (g) CO₂ and water vapor interaction with fresh WO₃ in the dark and under photoirradiation subsequently. (h) CO₂ and water vapor interaction with pre-irradiated WO₃ in the dark and under photoirradiation subsequently.

  DownLoad: Full-Size Img PowerPoint

  To further build a connection between the oxygen vacancies formation and CO₂ photoreduction and provide an unambiguous interpretation of the water function of in the reaction process, in-situ transient Fourier transform infrared spectroscopy (FTIR) experiment were conducted (Figs. 3f-h). The spectra were obtained by subtracting WO₃ background in Ar atmosphere. When sole CO₂ was introduced to the reaction system, no obvious weak was shown in Fig. 3f, which can be attributed to the weak absorption of CO₂. When WO₃ was under irradiation, two negative peaks corresponding to COOH⁻ and HOCOO⁻ were shown, which may due to the reduction of chemical absorbed CO₂ [38]. As for the situation that CO₂ purged through the water (Fig. 3g), a broaden peak at around 1650 cm⁻¹ arose, which belongs to the –OH, besides, a weak peak at 1425 cm⁻¹ assigned to HCO₃⁻ occurred, which indicate that attendance of water promotes the formation of HCO₃⁻ intermediate. Besides, under the same irradiation condition, the peaks corresponding to COOH⁻ and HOCOO⁻ decrease with a larger magnitude, which support the inference above. It is noticeable that after irradiation the peak around 1425 cm⁻¹ reoccurred, suggesting that HCO₃⁻ can be formed with the help of water. In addition, the formation and decomposition of the HCO₃⁻ keeps dynamic balance during the irradiation.

  To figure out the function of oxygen vacancies, the catalyst was pre-irradiated in Ar to induce oxygen vacancies formation, followed by the same experiment steps as above (Fig. 3h). The spectra were normalized according to the transmittance. It is noticeable that when the catalyst was kept in the dark state under humid CO₂ atmosphere, no peak corresponding to CO₂ reaction intermediate was found, which imply that photo-generated electrons and holes are indispensable. Different from the fresh catalyst, the spectrum of the catalyst containing oxygen vacancies shows a small peak corresponding to HOCOO⁻ species after CO₂ adsorption. A plausible explanation for is that one oxygen atom of CO₂ fills the oxygen vacancy and the other oxygen atom bind with hydroxy group on the surface to form a bending structure [39]. Compared with the spectrum of intrinsic WO₃ under same condition, a larger hydroxyl peak was shown. During the irradiation, the peak belongs to COOH⁻ decreases but is different from that HOCOO⁻ species. With the help of H₂O molecules, the oxygen vacancies facilitate the formation of HOCOO⁻, namely HCO₃⁻ intermediate and consequently promote CO₂ reduction.

  The aforementioned results proved the creation of oxygen vacancies via the light-induced process during photocatalytic reactions on the hexagonal WO₃ while the photogenerated electrons will reduce W⁶⁺ to W⁵⁺. It should be emphasized that such oxygen vacancies generation process is reversible when the catalyst was exposed to O₂-containing atmosphere. The created oxygen vacancies on the surface of hexagonal WO₃ can be eliminated by O₂ (Figs. S11 and S12 in Supporting information), and the oxygen vacancies can be recovered again after irradiated by light, completing the oxygen vacancies-containing chemical redox loop on the surface. The recyclable CO₂ photoreduction during this reversible oxygen vacancies-containing process were also confirmed, further confirming the indispensable role of oxygen vacancies in this process (Fig. S13 in Supporting information). It can be therefore concluded that irradiation of WO₃ created the oxygen vacancies on the surface, which serves as the activation site of CO₂ reduction. The existence of light-induced oxygen vacancies makes it possible for hexagonal WO₃ to drive CO₂ reduction, whereas an appropriate amount of oxygen vacancies is needed. To further verify the essential role oxygen vacancies on the photo-induced CO₂ reduction on the hexagonal WO₃, we synthesized a series of oxygen vacancies-containing samples via treating the hexagonal WO₃ under the atmosphere of H₂ gas (Fig. S14 in Supporting information). The reduction of WO₃ to produce oxygen vacancies were confirmed by the UV–vis spectra of the samples treated under different conditions. The photocatalytic activities increase obviously with increasing the content of oxygen vacancies to reach the maximum. Excess amount of the oxygen vacancies may destroy the structure of hexagonal WO₃ and result in a decline in photocatalytic activities.

  DFT calculation was also utilized to verify the function of oxygen vacancies. We first simulated the most possible oxygen vacancy sites on the hexagonal WO₃ surface. As shown in Fig. S15a (Supporting information), the comparison of three modes of oxygen vacancy were presented on the representative (100) surface. The relative defect formation energy was calculated during removing an oxygen atom from the corresponding location, the structure of Type Ⅲ with lowest defect formation energy was found to be the most stable oxygen vacancy site. The adsorption mode of CO₂ molecules on the perfect (100) and defective (100) surfaces was then calculated (Figs. S15b-e in Supporting information). When CO₂ molecule adsorbs on a perfect (100) surface (Figs. S15b and d), the adsorption energy is −0.37 eV. Compared with the C=O in CO₂ molecule, the C=O in adsorbed CO₂ is not significantly elongated compared with the non-absorbed case (1.25 Å vs. 1.17 Å). In addition, the C atom in CO₂ molecule forms a weak link with the surface oxygen atom, causing a bending configuration with a 136° carbon-oxygen double bond angle. In contrast, the adsorption energy of CO₂ molecule adsorbed on the oxygen vacancy-containing (100) surface is −2.09 eV, which is much larger in magnitude than that on the perfect (100) surface (Figs. S15c and e). The results illustrate that CO₂ molecules can be easily adsorbed on the oxygen vacancy sites. Moreover, the C=O bond in absorbed CO₂ is obviously elongated from 1.17 Å to 1.457 Å and the bond angle of C=O was reduced to 124.8°. It is acceptable that the O atom in CO₂ adsorbed onto the oxygen vacancy promotes the electron transfer from low-value W⁵⁺ to the O sites, leading to the elongation of C=O bond, which further intensifies the electron cloud mismatching between C and O atoms. The elongated C=O bond at this scale is much easier to break, signifying that C=O is weakened by the presence of oxygen vacancies, which effectively promotes the activation of CO₂ molecules.

  Based on the results above, oxygen vacancies can be induced on the surface of hexagonal WO₃ under irradiation, which serves as the activation sites of CO₂. When CO₂ and water were introduced with light irradiation on WO₃, a plausible reaction mechanism involved the formation of oxygen vacancies and intermediate on the hexagonal WO₃ photocatalyst was proposed (Scheme 1). Initially, oxygen vacancies were induced to generate on the surface of the hexagonal WO₃ by the irradiation of light. Then, oxygen atom in CO₂ molecule adsorbed on the light-induced oxygen vacancies of the hexagonal WO₃ surface to form bended *CO₂, following by protonated by –OH adsorbed on the adjacent W atom to form adsorbed HCO₃⁻ and HCOO⁻ intermediate. Differ from the mechanism previously reported on hydrated WO₃·0.33H₂O [29], since the attendance of water will lead to the formation of adsorbed H, the H species will form a hydrogen bond with the O atom in CO₂. Combining with the electron transfer from the adjacent W⁵⁺, the breaking of C-O band induces filling an oxygen vacancy and formation of CO and H₂O. The following irradiation will regenerate the oxygen vacancies and drive the successive reaction. The existence of photo-induced oxygen vacancies is a double-edged sword, since the presence of a small amount of oxygen vacancies will promote CO₂ activation, but a large amount of that will destroy the surface structure.

  Scheme 1

  

  Scheme 1.  Schematic illustration of the proposed mechanism for photo-induced CO₂ reduction on the hexagonal WO₃.

  DownLoad: Full-Size Img PowerPoint

  In summary, we experimentally realize the photo-induced CO₂ reduction to CO on the hexagonal WO₃ via an oxygen vacancies-involved process. Isotope labeling experiments verified the formation of oxygen with the attendance of water and lattice oxygen. The photo-induced formation of oxygen vacancies was demonstrated to closely correlate the activities of CO₂ photoreduction. Combining with the in-situ FTIR, a mechanism involved oxygen vacancy was proposed that with the assistance of adsorbed water, CO₂ adsorbed on oxygen vacancy was protonated to HCO₃⁻ intermediate and finally reduce to CO. Our work provides insight on the mechanism for carbon dioxide reduction on oxygen vacancy sites.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  Acknowledgments

  This work was performed by the Fundamental Research Center of Artificial Photosynthesis (FReCAP), and supported by National Natural Science Foundation of China (No. 22088102). This work was also supported by National Natural Science Foundation of China (No. 22090033). R. Li would like to thank the support from Youth Innovation Promotion Association of Chinese Academy of Sciences and Dalian Institute of Chemical Physics, CAS.

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.02.006.

  
  
• 1. [1]

     S. Navarro-Jaén, M. Virginie, M. Robert, et al., Nat. Rev. Chem. 5 (2021) 564–579. doi: 10.1038/s41570-021-00289-y

  2. [2]

     Y.X. Li, D.P. Hui, J.C. Zhao, et al., Nat. Commun. 12 (2021) 123. doi: 10.1038/s41467-020-20444-1

  3. [3]

     X.X. Chang, T. Wang, J.L. Gong, Energy Environ. Sci. 9 (2016) 2177–2196. doi: 10.1039/C6EE00383D

  4. [4]

     H.L. Huo, D. Liu, A. Li, et al., Nanoscale 12 (2020) 13912–13917. doi: 10.1039/D0NR02944K

  5. [5]

     G.G. Zhang, G.S. Li, X.C. Wang, et al., Angew. Chem. Int. Ed. 58 (2019) 3433–3437. doi: 10.1002/anie.201811938

  6. [6]

     W.G. Tu, Y. Zhou, Z.Z. Zou, Adv. Mater. 26 (2014) 4607–4626. doi: 10.1002/adma.201400087

  7. [7]

     H. Li, J. Li, Z.H. Ai, et al., Angew. Chem. Int. Ed. 57 (2018) 122–138. doi: 10.1002/anie.201705628

  8. [8]

     G.C. Xi, S.X. Ouyang, J.H. Ye, et al., Angew. Chem. Int. Ed. 51 (2012) 2395–2399. doi: 10.1002/anie.201107681

  9. [9]

     L.C. Wang, Y. Wang, Z.F. Liu, et al., J. Mater. Chem. A 4 (2016) 5314–5322. doi: 10.1039/C5TA10180H

  10. [10]

      D.L. Liu, C.H. Wang, B. Zhang, et al., Chem 5 (2019) 376–389. doi: 10.1016/j.chempr.2018.11.001

  11. [11]

      J.Z. Liao, K.L. Li, Y.J. Sun, et al., Chin. Chem. Lett. 31 (2020) 2737–2741. doi: 10.1016/j.cclet.2020.03.081

  12. [12]

      L. Liang, X.D. Li, Y. Xie, et al., Joule 2 (2018) 1004–1016. doi: 10.1016/j.joule.2018.02.019

  13. [13]

      Y.F. Sun, S. Gao, F.C. Lei, et al., Chem. Soc. Rev. 44 (2015) 623–636. doi: 10.1039/C4CS00236A

  14. [14]

      Y. Liu, X.L. Dong, B. Dong, et al., Colloids Surf. A 621 (2021) 126582. doi: 10.1016/j.colsurfa.2021.126582

  15. [15]

      J. Yang, S.Y. Hu, Y.R. Fang, et al., ACS Catal. 9 (2019) 9751–9763. doi: 10.1021/acscatal.9b02408

  16. [16]

      W. Pipornpong, R. Wanbayor, V. Ruangpornvisuti, Appl. Surf. Sci. 257 (2011) 10322–10328. doi: 10.1016/j.apsusc.2011.06.013

  17. [17]

      L.J. Liu, C.Y. Zhao, Y. Li, et al., J. Phys. Chem. C 116 (2012) 7904–7912. doi: 10.1021/jp300932b

  18. [18]

      T. Li, X. Dong, Y.H. Sun, et al., Appl. Surf. Sci. 526 (2020) 146578. doi: 10.1016/j.apsusc.2020.146578

  19. [19]

      X. Chen, X. Zhang, Y.H. Li, et al., Appl. Catal. B 281 (2021) 119516. doi: 10.1016/j.apcatb.2020.119516

  20. [20]

      Z.L. Wang, X. Mao, P. Chen, et al., Angew. Chem. Int. Ed. 131 (2019) 1042–1046. doi: 10.1002/ange.201810583

  21. [21]

      W.H. Kong, R. Zhang, X.P. Sun, et al., Nanoscale 11 (2019) 19274–19277. doi: 10.1039/C9NR03678D

  22. [22]

      J.L. Wang, S.H. Kang, H.M. Zhang, et al., Chin. Chem. Lett. 32 (2021) 2833–2836. doi: 10.1016/j.cclet.2021.01.020

  23. [23]

      Z.P. Chen, Q.Q. Jiang, F. Cheng, et al., J. Mater. Chem. A 7 (2019) 6099–6112. doi: 10.1039/C8TA11957K

  24. [24]

      A. Wagner, C.D. Sahm, E. Reisner, et al., Nat. Catal. 3 (2020) 775–786. doi: 10.1038/s41929-020-00512-x

  25. [25]

      S.N. Habisreutinger, L. Schmidt-mende, J.K. Stolarczyk, Angew. Chem. Int. Ed. 52 (2013) 7372–7408. doi: 10.1002/anie.201207199

  26. [26]

      A. Li, J.L. Gong, M. Antonietti, et al., Angew. Chem. Int. Ed. 58 (2019) 14549–14555. doi: 10.1002/anie.201908058

  27. [27]

      G.G. Zhang, Z.A. Lan, X.C. Wang, ChemCatChem 9 (2015) 1422–1423.

  28. [28]

      M. Gattrell, N. Gupta, A. Co, J. Electrochem. Soc. 594 (2006) 1–19.

  29. [29]

      S.S. Tan, L. Zou, E. Hu, Catal. Today 131 (2008) 125–129. doi: 10.1016/j.cattod.2007.10.011

  30. [30]

      T.Y. Peng, D.N. Keng, J.R. Xiao, et al., J. Solid State Chem. 194 (2012) 250–256. doi: 10.1016/j.jssc.2012.05.016

  31. [31]

      E. Geber, R.J. Ackermann, Inorg. Chem. 5 (1966) 136–142. doi: 10.1021/ic50035a033

  32. [32]

      R. Chatten, A.V. Chadwick, A.D. Rougier, et al., J. Phys. Chem. B 109 (2005) 3146–3156.

  33. [33]

      J.M. Wang, E. Khoo, P.S. Lee, et al., J. Phys. Chem. C 112 (2008) 14306–14312. doi: 10.1021/jp804035r

  34. [34]

      M.L. Kang, J.M. Liang, F.F. Wang, et al., Mater. Res. Bull. 121 (2020) 110614. doi: 10.1016/j.materresbull.2019.110614

  35. [35]

      W. Zhang, H.L. Hei, Y. Tian, et al., Nano Energy 66 (2019) 104113. doi: 10.1016/j.nanoen.2019.104113

  36. [36]

      M.F. Daniel, B. Desbat, J.C. Lassegues, J. Solid State Chem. 67 (1987) 235–247. doi: 10.1016/0022-4596(87)90359-8

  37. [37]

      Y.T. Wang, J.M. Cai, M.Q. Wu, et al., Appl. Catal. B 239 (2018) 398–407. doi: 10.1016/j.apcatb.2018.08.029

  38. [38]

      S.M. Sun, W. Motonori, J. Wu, et al., J. Am. Chem. Soc. 140 (2018) 6474–6482. doi: 10.1021/jacs.8b03316

  39. [39]

      Y.F. Ji, F. Luo, J. Am. Chem. Soc. 138 (2016) 15896–15902. doi: 10.1021/jacs.6b05695

• 

  Figure 1  Schematic structure and characterization of as-prepared hexagonal WO₃. (a) Scheme illustration of hexagonal WO₃ crystal structure. (b) XRD pattern of as-prepared WO₃. (c) SEM image of hexagonal WO₃. (d, e) High-resolution TEM images of hexagonal WO₃.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (a) Photocatalytic activities of CO₂ reduction on hexagonal WO₃. (b) Control experiments of CO₂ reduction under light and dark conditions. (c) CO₂ photoreduction to CO on hexagonal WO₃ under different wavelengths. (d) The comparison of CO₂ photoreduction with or without introducing H₂O vapor. (e, f) Mass spectra extracted from GC–MS analysis of product with isotope labeling using ¹³CO₂ and H₂¹⁸O.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) Photographs of the hexagonal WO₃ under light irradiation. (b) UV–visible spectra of the hexagonal WO₃ before and after light irradiation. (c) In-situ EPR spectra of the hexagonal WO₃ before and after light irradiation. (d) XRD pattern of the hexagonal WO₃ before and after the photocatalytic reaction. (e) In-situ Raman spectra of the hexagonal WO₃ before and after light irradiation. In-situ FTIR spectra of (f) CO₂ and (g) CO₂ and water vapor interaction with fresh WO₃ in the dark and under photoirradiation subsequently. (h) CO₂ and water vapor interaction with pre-irradiated WO₃ in the dark and under photoirradiation subsequently.

  下载: 全尺寸图片 幻灯片
  

  Scheme 1  Schematic illustration of the proposed mechanism for photo-induced CO₂ reduction on the hexagonal WO₃.

  下载: 全尺寸图片 幻灯片
•