English

• Photocatalytic reduction of carbon dioxide (CO₂) into valuable fuel is a promising approach to solve current energy and environmental problems [1]. Among them, the reduction of CO₂ to methane (CH₄) is significantly challenging due to the complex eight-electron-proton transfer process [2]. Metal halide perovskites (MHPs) possess intriguing merits of high light absorption coefficient, long charge diffusion length and adjustable band gaps, offering great potential for photocatalytic CO₂ reduction [3-7].

  For MHPs, the inherent soft crystal lattice property of perovskite contributes to high structural tolerance factor, which allows great tolerance to lattice mismatch or atomic deletion, providing a great opportunity for defect adjustment of perovskites materials [8]. Generally, defects in MHPs can trigger the reduced open-circuit voltage in solar cells or severe carrier recombination in LED, thus there is a negative correlation between the defect density and its efficiency [9,10]. Nevertheless, the contribution of defects should be discussed differently for field of photogenerated charges of surface catalysis and the improved charge photocatalysis. Some defects are believed to capture photogenerated charges for surface catalysis and improve charge separation efficiency, while some defects as deep energy level defects also trap photogenerated charges but inversely act as recombination center, leading to the carrier nonradiative recombination and reducing the charge separation efficiency [11-13]. Therefore, understanding defects formation mechanism and elucidating how the defects could boost a photocatalytic reaction is essential yet complex and challenging.

  Aiming to address the aforementioned challenges, extensive research efforts are focused on the defect engineering of halide perovskite with enhanced capabilities [14-16]. In particularly, the introduction pathway and mechanism of halogen vacancy have been extensively investigated. He et al. reported a controllable approach to synthesize the Br vacancy-rich Cs₂AgBiBr₆ via the visible light irradiation on the Cs₂AgBiBr₆ double perovskite, which indicates Br vacancies in Cs₂AgBiBr₆ can optimize the local atomic arrangement and electronic structure, contributing to a high charge separation efficiency [17]. Similarly, Pi et al. discovered that Cl-deficient 3D hierarchical Cs₂NaBiCl₆ porous microspheres were prepared via a simple grinding method by utilizing the moisture adsorption by the chlorides, revealing Cl vacancies can suppress photogenerated electron-holes recombination and reduce the free energy barrier for the generation of key intermediate [18]. These above findings disclose halogen vacancies of MHPs usually play a positive effect in photocatalysis. Nevertheless, conventional investigations pay more attention to the effect of halogen vacancies on behavior of photogenerated carriers for MHPs photocatalysis, their effects on adsorption activation of surface reaction molecules on their surfaces (such as CO₂ and CO) are not well established yet. In this regard, exploring an effective and controllable preparation method of halogen defect-rich MHPs nanocrystals is of great significance but still unfulfilled.

  According to the growth mechanism of halide perovskite, the density of surface vacancy defects could be elaborately designed and well controlled by manipulating its crystal growth. In this work, we added solvent to assist CsBr and PbBr₂ dissolution and recrystallization during the ball milling process, and changed the polarity of the solvent to modulate the dissolution speed of CsBr and PbBr₂, and then changed the crystallization growth process of CsPbBr₃ to achieve the regulation of the Br vacancy density. We found the improvement of the solvent polarity benefits the increase in the crystallization of CsPbBr₃, thereby declining the vacancy density. The results showed surface Br vacancies show a small effect on the energy band structure and a moderate effect on improving the photogenerated charge separation and carrier lifetime. By comparing the photocatalytic CO₂ reduction activity and selectivity of CsPbBr₃ with different vacancy densities, the adsorption ability of CO₂ and CO on the catalysts surface is significantly activated by Br vacancies, which is responsible for the boosted catalytic activity and selectivity.

  CsPbBr₃ catalysts were prepared via the wet-mechanochemical ball-milling process (Fig. 1a). During the process, the solvent additives with different polarities were added and the dissolution of CsBr and PbBr₂ precursors varied with the solvent polarities. According to the crystal growth theory, the crystallinity of CsPbBr₃ should be tuned by adjusting the polarity of solvent. In this work, four solvents including acetonitrile (ACN), ethanol (EA), carbon tetrachloride (CT) and hexane (HX) were used, where these solvents have gradient polarities of 6.2, 4.3, 1.6, 0.06, respectively (Table S1 in Supporting information). Hence, the corresponding catalysts were denoted as CsPbBr₃-ACN, -EA, -CT, and -HX, respectively. Note that the sample names are further simplified in the figures as the abbreviations of corresponding solvents.

  Figure 1

  

  Figure 1.  (a) Schematic of the ball-milling process to prepare CsPbBr₃ using different solvents. (b) XRD patterns. (c-f) SEM images. (g) Br 3d XPS spectra. (h) Fitted results of the Br/Pb ratio. (i) EPR spectra and (j) Br vacancy density (represented by peak intensity of the EPR spectra) for different CsPbBr₃ samples.

  DownLoad: Full-Size Img PowerPoint

  Fig. 1b shows the XRD patterns of different CsPbBr₃ catalysts prepared by ball-milling method using different solvents. All CsPbBr₃ samples could be indexed as the orthorhombic phase (JCPDS No. 18–0364), where the representative peaks corresponding to (100), (110), (002) and (200) planes were all observed [19]. The difference in the crystallinities of CsPbBr₃ samples could be further seen from the change of the crystal morphologies (Figs. 1c-f). The CsPbBr₃-ACN showed a regular grain morphology and a large grain size, indicating a good crystallinity. As the decrease of solvent polarity, the shape and size of the crystalline grains of CsPbBr₃ became more irregular and smaller, corresponding to the decrease in crystallinity. Taking the CsPbBr₃-ACN as the basis and setting its crystallinity as 1, the relative crystallinities of different CsPbBr₃ samples were obtained by comparing the peak intensities of (110), (200) and (202) planes, which verifying the effectiveness of controlling the crystallization of CsPbBr₃ by adjusting the solvent polarity (Fig. S1 in Supporting information).

  It is noted that the organic solvents tend to adsorb on the CsPbBr₃ surface and these organic residues could produce CO or CH₄ during the photocatalytic process via photochemical reactions [20]. To eliminate the interference of organic solvents, the infrared spectroscopy technique was adopted to examine the organic residues over the CsPbBr₃ catalysts (Fig. S2 in Supporting information), indicating the organics on CsPbBr₃ after ball-milling were removed during the overnight heat treatment. The surface areas were further measured by the N₂ adsorption/desorption isotherm, where the BET surface areas of CsPbBr₃—HX, -CT, -EA, and -ACN were determined as 3.41 ± 0.29, 2.98 ± 0.18, 2.21 ± 0.04, and 1.40 ± 0.02 m²/g, respectively (Fig. S3 in Supporting information). It is concluded that the surface area of CsPbBr₃ decreases with the increase of crystallinity, corresponding to the larger grain size in SEM images. X-ray photoelectron spectroscopy (XPS) was performed to study the surface element valence states of different CsPbBr₃ catalysts. As shown in the C 1s spectra (Fig. S4a in Supporting information), the four CsPbBr₃ samples had an identical binding energy for the C 1s peak, indicating that the carbon species are the same on the four catalysts. It further confirms that there are no solvent residues on the CsPbBr₃ surface. As for the Cs 3d and Pb 4f spectra, the CsPbBr₃—HX showed a slight left-shift toward a higher binding energy as compared to CsPbBr₃-ACN (Figs. S4b and c in Supporting information). This shift to higher binding energy was more obvious in the Br 3d peaks for CsPbBr₃—HX, comparing to the spectrum of CsPbBr₃-ACN (Fig. 1g). It is also observed that the shift degree of the binding energy of Br 3d increased in the order of CsPbBr₃-EA, CsPbBr₃—CT, CsPbBr₃—HX. Typically, a shift to higher binding energy in XPS spectra for halogen elements indicates an electron-deficient environment, which corresponds to the formation of halogen-vacancies on the surface of perovskites [17,21-23]. Thus, using CsPbBr₃-ACN as a baseline for comparison, we can infer that Br vacancies are present on the surfaces of the other three CsPbBr₃ samples, with vacancy density increasing in correlation with the degree of shift.

  Fig. S5 (Supporting information) depicts the steady-state photoluminescence (PL) spectra of four CsPbBr₃ catalysts, where all showed one emission band at ~552 nm. It is observed that the PL intensity actually decreased with the solvent polarity. In general, the PL intensity of perovskite is negatively related to the Br vacancy density [24,25]. Therefore, we can infer that the defect density of CsPbBr₃—HX is the highest. Besides, the surface Br/Pb ratio of CsPbBr₃ was calculated to monitor the relative Br content indirectly based on the Br 3d and Pb 4f XPS spectra, where the peak areas and atomic sensitivity factors were counted (Table S2 in Supporting information). As shown in Fig. 1h, a reduced Br/Pb ratio (2.68) was observed in the CsPbBr₃—HX sample compared with the CsPbBr₃-ACN (3.00). It is demonstrated that, with the polarity of solvent decreasing, the surface of CsPbBr₃ changed from Br-deficient (low Br/Pb ratio) to Br-rich (high Br/Pb ratio). The Br vacancy defects of our CsPbBr₃ catalysts were further detected by the low-temperature solid-state EPR spectra. As shown in Fig. 1i, all the four samples showed only one single Lorentzian line centered at a g value of about 2.001, which represents the existence of Br vacancies [26]. It is also noted that the EPR signal of CsPbBr₃—HX was the highest, corresponding to the richest Br vacancies. The relationship of Br vacancy density and the Br/Pb ratio of the four CsPbBr₃ samples was drawn in Fig. 1j, the calculated Br vacancy amount showed a quasi-linear relationship with the Br/Pb ratio, clearly supporting the rationality of our Br vacancy quantifications using the EPR and XPS results. To sum up, by adjusting the polarity of solvent used in the ball-milling process, the crystallinity of CsPbBr₃ catalyst was changed and hence the surface Br vacancy density was controlled.

  The catalytic performance of CsPbBr₃ samples for solar CO₂ reduction was evaluated in a flow microreactor with continuous CO₂ gas and H₂O vapor passing through the floating catalyst bed (Fig. S6 in Supporting information). Only CH₄ and CO were detected as the carbonaceous product, while H₂ was yielded as the byproduct (Fig. S7 in Supporting information) [27]. As shown in Fig. 2a, the CsPbBr₃-ACN, which has the lowest Br vacancy density, only yielded low production rates of 0.95 ± 0.08 and 0.56 ± 0.08 µmol g^-1 h^-1 for CH₄ and CO evolving, respectively. On the contrary, the CsPbBr₃—HX, which has the richest Br vacancy defects, showed the highest activity for solar CO₂ reduction with CH₄ and CO production rates being 17.94 ± 0.81 and 1.06 ± 0.12 µmol g^-1 h^-1, respectively. When we counted the number of electrons participating in CO₂ reduction and took the difference of BET surface areas into consideration, it was found that the surface area-normalized electron consumption rate of CsPbBr₃ for solar CO₂ reduction is positively related to the Br vacancy density (Fig. 2b), demonstrating that the Br vacancy would facilitate the photocatalytic CO₂ reduction over CsPbBr₃ catalysts.

  Figure 2

  

  Figure 2.  (a) Production rates of CH₄ and CO. (b) Relationship between Br vacancy quasi-density and the surface area-normalized electron consumption rate. (c) The ratio of CH₄/CO production rate. (d) Relationship between Br vacancy quasi-density and the surface area-normalized production rate of CH₄ and CO. (e) ¹³CH₄ and ¹³CO mass spectra for ¹³CO₂ photocatalytic reduction on CsPbBr₃—HX. (f) Stability test of CsPbBr₃—HX at 150 ℃ under 3 sun irradiation.

  DownLoad: Full-Size Img PowerPoint

  The product selectivity of different CsPbBr₃ catalysts was further investigated. As we compared the ratio of production rates for CH₄ and CO, it is shown that the CH₄/CO yield ratio of CsPbBr₃—CT was only 0.87, while that of CsPbBr₃—HX was 16.99, which is much higher than the others (1.63 for CsPbBr₃-ACN and 2.16 for CsPbBr₃-EA) (Fig. 2c). The CH₄ selectivity for the four samples was further calculated, where CsPbBr₃—HX demonstrated the highest value of 95.8% (75.6%, 79.4% and 73.2% for CsPbBr₃-ACN, -EA, -CT, respectively) (Fig. S8 in Supporting information). The photocatalytic performance of CsPbBr₃—HX in this work exceeds most of reported CsPbBr₃-based photocatalysts (Table S5 in Supporting information). As shown in Fig. 2d, the normalized yield rate of CH₄ is positively related to the Br vacancy density, while that of CO shows a volcanic relationship with the Br vacancy density. It is suggested that, besides determining the photocatalytic activity of CsPbBr₃, the Br vacancy also affects the product selectivity.

  To consolidate the above findings, the carbon source of CH₄ products was investigated via the ¹³C-labeled isotopic test, where ¹³CO₂ was used as the reactants. The mass spectra of collected gaseous products matched well with the ¹³CH₄ and ¹³CO chemicals (Fig. 2e), verifying that these products are reduced from the CO₂ input. In addition, the stability of the champion photocatalyst, CsPbBr₃—HX, for solar CO₂ conversion was studied. During the test, CH₄, CO and H₂ were continuously produced with the CH₄ selectivity keeping at around 95% (Fig. 2f). After 4-h test, the CsPbBr₃—HX catalyst maintained more than 92% of its initial activity for the production of CH₄ (from 18.02 µmol g^-1 h^-1 to 16.70 µmol g^-1 h^-1), demonstrating a reasonable durability for photocatalytic CO₂ reduction. To unravel the origins of CH₄ selectivity, we investigated the effects of Br vacancy on the photocatalytic processes, including light absorption, charge separation, and surface reaction.

  Fig. S9 (Supporting information) displays the UV–vis spectra of different CsPbBr₃ samples. The four catalysts showed small difference in the light absorbance and absorption edges. It is indicated that the surface Br vacancies have not formed the intra-band defect states and hence there is no defect-induced light absorption for our CsPbBr₃ catalysts [28]. The energy band gaps (E_g) of the four catalysts were further determined by the Tauc plots (Fig. S10 in Supporting information), where the E_g for CsPbBr₃-ACN, -EA, -CT, and -HX were 2.28, 2.27, 2.27 and 2.27 eV, respectively. The band structures were further obtained by determining the Fermi energy level (E_f), the conduction band minimum (E_CBM), and the valence band maximum (E_VBM) via the E_g and ultraviolet photoelectron spectroscopy (Table S3 in Supporting information). As shown in the schematic of band diagrams in Fig. S11 (Supporting information), the Br vacancy would gradually uplift the E_CBM and E_VBM of CsPbBr₃ by less than 0.1 eV following the sequence of CsPbBr₃-ACN, -EA, -CT, and -HX. Therefore, the Br vacancies have little influence on the band structure of CsPbBr₃, which would not contribute to the rather big difference in the photocatalytic activities.

  The effect of Br vacancy defects on the charge separation of CsPbBr₃ was then investigated by the surface photovoltage (SPV) and femtosecond transient absorption (TA) spectra. Fig. 3a shows the measured SPV signals of different CsPbBr₃ samples. CsPbBr₃—HX with abundant Br vacancy defects demonstrated the largest SPV signal, reaching 88.4 ± 5.0 mV, while CsPbBr₃-ACN with the least Br vacancy density had the smallest SPV value of 42.3 ± 3.9 mV. As the signal of SPV is directly related to the accumulation of surface charges upon irradiation, the value of SPV then measures the charge separation ability of the photocatalyst [29]. Therefore, it is suggested that the surface Br vacancy defects could facilitate the charge separation process by acting as charge-trapping states.

  Figure 3

  

  Figure 3.  (a) Surface photovoltage plots. (b) Transient absorption spectroscopy curves. In-situ DRIFT spectra of CsPbBr₃ prepared with different solvents: Wavelength range of 1810–3600 cm^-1 (c), and 1000–1550 cm^-1 (d). Test process: Before test, the reaction cell was purged with Ar at a flow rate of 30 sccm and heated at 150 ℃ for 20 min. Then, the gas input was switched to CO₂ and H₂O vapor at a flow rate of 5 sccm. A background spectrum was collected after a 15 min equilibrium. The light was then turned on and the in-situ spectra after 20 min equilibrium were recorded.

  DownLoad: Full-Size Img PowerPoint

  The charge carrier dynamics was further analyzed by the TA spectra. As shown in Fig. S12 and Table S4 (Supporting information), the TA kinetics probed at ~517 nm showed a typical absorption decay, from which the lifetime of photogenerated holes can be determined [15]. As shown in Fig. 3b, the hole signal of CsPbBr₃—HX demonstrated the slowest decay kinetics as compared to the others, corresponding to the longest hole-lifetime of 2012.1 ps. As a contrast, the absorption decay of CsPbBr₃-ACN is the fastest among them and the lifetime is only 378.6 ps. It is clear that the Br vacancy defects would prolong the lifetime of the photogenerated holes. Hence, for the charge separation part, the Br vacancy defects is supposed to promote the charge separation, allowing more photogenerated charges to survive for the surface catalytic CO₂ reaction.

  To further confirm the nature of defect-induced photocatalytic process, the effect of Br vacancy defects on the surface catalytic process was investigated. We first study the reaction mechanism by probing the possible reaction intermediates with in-situ DRIFT spectra. Figs. 3c and d display the acquired in-situ DRIFT spectra for photocatalytic CO₂ reduction on different CsPbBr₃ catalysts at 150 ℃ after 20 min illumination. For all the four samples, the characteristic bands of H₂O at 3497 cm^-1 and CO₂ at 2360 and 2336 cm^-1 were all observed [30,31]. Notably, the peak at 2145 cm^-1 corresponding to CO was only observed in spectrum of CT-CsPbBr₃, which is consistent with the high production rate of CO over CT-CsPbBr₃. In addition, the carboxylate species of *CO₂ were also identified from the bands at 1610 and 1264 cm^-1, implying that the adsorbed CO₂ molecules were activated in the same way on the four samples. The three peaks in range of 2975–2840 cm^-1, which are characteristics of C—H stretch bands (σ_C) for CH₃ group, correspond to the signals from absorbed methane (CH₄), methyl intermediate (CH₃*) and/or methoxyl intermediate (CH₃O*). The corresponding C—H vibration bands (δ_C) of CH₃ at 1462 and 1380 cm^-1 were also observed for all the four samples. Furthermore, an obvious peak at ~1110 cm^-1 was detected in all spectra of different samples, which band is the characteristic of C—O stretch bands (σ_C) [32]. Then, it is supposed that the CH₃O* intermediate is involved in the photocatalytic CO₂ reduction over the four CsPbBr₃ catalysts. Therefore, it is concluded that the Br vacancy defects tend not to change the reaction pathway for photocatalytic CO₂ reduction over CsPbBr₃.

  We then put our focus on the adsorption and activation of surface reactants of the role of Br vacancies, where the two molecules of CO₂ and CO were examined. Temperature-programmed desorption (TPD) technique was adopted to evaluate the interaction between the molecules and the catalytic sites [33,34]. As shown in the CO₂-TPD curves in Fig. 4a, the former CO₂ desorption peaks of four samples located at P1 had identical peak temperature (233 ± 1 ℃), corresponding to same CO₂ adsorption sites. While, the latter peaks of CsPbBr₃-ACN, -EA, -CT, and -HX located at P2 showed peak temperatures of 251.7, 254.5, 256.6, and 257.5 ℃ (Fig. 4b), respectively, where the desorption temperature increased with the Br-vacancy density, revealing that the vacancy-relevant sites deliver an enhanced interaction with the adsorbed CO₂ molecules. Furthermore, the amounts of CO₂ adsorbed could also be derived from the integrated area of the desorption peak. By further normalizing the adsorption amount to the BET surface area and relating this amount to the defect density, it is found that the total normalized adsorption amount of CO₂ on CsPbBr₃ increased with the vacancy density together with a gradually decrease rate (Fig. S13a in Supporting information). This phenomenon could be explained by the vacancy-induced chemisorption of CO₂ on CsPbBr₃: initially, the adsorbing capacity of CO₂ increased with the density of Br vacancy defects; then, as the chemisorbed CO₂ on the surface gradually became saturated, the adsorbing capacity tended to reach a plateau with further increasing Br vacancy density. Therefore, the increased CO₂ adsorption induced by the Br vacancy defects is thought to accelerate surface reactions and hence improve the activity.

  Figure 4

  

  Figure 4.  (a) CO₂-TPD and (b) the change of adsorption temperature at P2 desorption peaks for CO₂-TPD. (c) CO-TPD and (d) the change of adsorption temperature at P2′ desorption peaks for CO-TPD. (e) CO₂ and (f) CO adsorption energy of CsPbBr₃ with different number of defects.

  DownLoad: Full-Size Img PowerPoint

  The CO adsorption on the surface sites of CsPbBr₃ was also investigated to reveal the cause for the change of CH₄ productivity, given that *CO is one of the key intermediates for CH₄ production. As shown in Fig. 4c, the CO-TPD curves of different samples all showed desorption peaks located at P1′ and P2′, corresponding to the chemisorption of CO at the reactive sites. Similarly, the CO desorption peaks of four samples located at P1′ had fixed peak temperature corresponding to same CO adsorption sites. It is also observed that the desorption temperature of CO became higher ranging from 302 ℃ to 330 ℃ for P2′ desorption peak when the polarity of the solvents increased (Fig. 4d). It is concluded that the interaction between the active sites and the CO molecular is enhanced along with the enrichment of surface Br vacancies. In addition, it is found that the total normalized adsorption amount of CO located at P2′ on CsPbBr₃ displays continuous upsurge with the increase of defect density (Fig. S13b in Supporting information). Nevertheless, the strong interaction, along with sufficient amount of CO adsorption, contribute to the high selectivity of CH₄. As a result, due to weak CO interaction and insufficient CO adsorption amount, CsPbBr₃—CT exhibited the lowest value of CH₄ selectivity, while the opposite case is observed for CsPbBr₃—HX. These findings suggested that Br vacancy defects favor to improve CO adsorption capacity, thus leading to high CH₄ selectivity. This was mainly attributed to the fact that vacancies tended to change the local charge density on the surface, which helped to break the chemical bonds between molecules and increased the interaction between adsorbed molecules and the catalyst surface during the catalytic process, thereby improving the adsorption of CO₂ and CO [26,35].

  Further, the high activity and selectivity mechanism induced by Br vacancy defects through DFT calculations was investigated. Theoretically, different kinds of vacancies on catalyst are closely related to the CO₂ and CO adsorption energy, thereby achieving the tailored regulation of activity and selectivity [36,37]. To this end, upon keeping the crystal structural stability of perovskite, we then constructed series of Br vacancy defects with different numbers and sites using the Vienna ab initio Simulation Package (VASP). The CO₂ and CO adsorption energy of CsPbBr₃ catalyst depending on Br vacancy defects with different numbers and sites were presented in Figs. 4e and f. The value of adsorption energy of CO₂ is positive with the absence of any defects. On the contrary, the adsorption energy of CO₂ and CO change from positive to negative in the presence of defects, and the adsorption energy of two defects was significantly lower than that of a single defect. This firmly implied the increase in Br vacancy defects significantly contributed to CO₂ and CO adsorption in photocatalyst CO₂ reduction. More importantly, no matter where the two defects are distributed in the crystal, its adsorption energy of CO₂ is significantly lower than that of one defect. This indicates that the number of defects, rather than their locations, has a significant influence on the adsorption energy. Such distinctive variations in adsorption energy definitely demonstrate that the numbers of Br vacancy defects are beneficial to promote the enhancement of catalytic activity and selectivity.

  In this work, we developed a series of CsPbBr₃ photocatalysts with adjustable surface Br vacancies defects via the ball milling method using solvents with different polarities for highly efficient and selective photocatalytic CO₂ reduction in a solid-gas phase reaction. Benefiting from Br vacancy defects, the carrier separation capacity and adsorption of CO₂ and CO molecules have been boosted, thereby accelerating surface reaction kinetics in catalytic process. In addition, the interaction between the active sites and the CO molecular is enhanced as surface Br vacancies upsurge, resulting in a decrease in the production of the two-electron CO and concurrently improving the eight-electron reaction. Consequently, the richest Br vacancy defects endow CsPbBr₃—HX with high rate of CO₂ photoreduction to CH₄ up to 17.94 ± 0.81 µmol g^-1 h^-1 along with superior selectivity of 95.8%, which is 18.9-fold compared that of CsPbBr₃—CAN with the lowest Br vacancy defects. This work proposes a simple method to synthesize a Br vacancy rich CsPbBr₃ and opens an entrance to substantially enhance the photocatalytic activity and selectivity.

  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.

  CRediT authorship contribution statement

  Hui Bian: Writing – original draft. Xinyi Yuan: Writing – original draft. Nan Zhang: Data curation. Zhuo Xu: Resources. Juhong Lian: Writing – original draft. Ruibin Jiang: Writing – review & editing. Junqing Yan: Writing – review & editing. Deng Li: Writing – review & editing. Shengzhong (Frank) Liu: Writing – review & editing.

  Acknowledgments

  This work was financially supported by the National Natural Science Foundation of China (Nos. 22072081, 22302148), the Nature Science Foundation of Shaanxi in China (No. 2024JC-YBQN-0468), Scientific Research Program Funded by Education Department of Shaanxi Provincal Government (Nos. 22JK0379, 24JK0665) and Youth Talent Scientific Research Program Funded by Weinan Normal University (No. 2022RC17).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.111034.

  
  
• 1. [1]

     Z. Jiang, X. Xu, Y. Ma, et al., Nature 586 (2020) 549–554. doi: 10.1038/s41586-020-2738-2

  2. [2]

     X. Jiang, J. Huang, Z. Bi, et al., Adv. Mater. 34 (2022) 2109330.

  3. [3]

     Y. Li, Y. Lei, H. Wang, et al., Nano-Micro Lett. 15 (2023) 128.

  4. [4]

     Z. Dai, N.P. Padture, Nat. Energy 8 (2023) 1319–1327. doi: 10.1038/s41560-023-01378-6

  5. [5]

     Y. Zou, S. Li, D. Zheng, et al., Sci. China Chem. 67 (2024) 2215–2223. doi: 10.1007/s11426-024-2076-3

  6. [6]

     Y. Jiang, J.F. Liao, H.Y. Chen, et al., Chem 6 (2020) 766–780.

  7. [7]

     Y. Wang, H. Huang, Z. Zhang, et al., Appl. Catal. B 282 (2021) 119570.

  8. [8]

     H. Arandiyan, S.S. Mofarah, C.C. Sorrell, et al., Chem. Soc. Rev. 50 (2021) 10116–10211. doi: 10.1039/d0cs00639d

  9. [9]

     G. Li, Z. Su, L. Canil, et al., Science 379 (2023) 399–403.

  10. [10]

      J. Liu, Z. Liu, J. Ren, et al., Adv. Funct. Mater. 35 (2025) 2420239. doi: 10.1002/adfm.202420239

  11. [11]

      C. Wang, Z. Xie, Y. Wang, et al., Adv. Sci. 11 (2024) 2402471.

  12. [12]

      Y. Wang, H. Fan, X. Liu, et al., J. Alloys Compd. 945 (2023) 169197.

  13. [13]

      J. Wu, X. Li, W. Shi, et al., Angew. Chem. Int. Ed. 57 (2018) 8719–8723. doi: 10.1002/anie.201803514

  14. [14]

      J. Liu, S. Zhang, D. Qu, et al., Nano-Micro Lett. 17 (2024) 24.

  15. [15]

      X. Wang, K. Li, J. He, et al., Nano Energy 78 (2020)105388.

  16. [16]

      Y. Dong, K. Li, W. Luo, et al., Angew. Chem. Int. Ed. 59 (2020) 12931–12937. doi: 10.1002/anie.202002939

  17. [17]

      Z. He, Q. Tang, X. Liu, et al., Energy Fuels 35 (2021) 15005–15009. doi: 10.1021/acs.energyfuels.1c01482

  18. [18]

      J. Pi, X. Jia, Z. Long, et al., Adv. Energy Mater. 12 (2022) 2202074.

  19. [19]

      Y. Jiang, J.F. Liao, H.Y. Chen, et al., Chem 6 (2020) 766–780.

  20. [20]

      J. San Martin, N. Dang, E. Raulerson, et al., Angew. Chem. Int. Ed. 61 (2022) e202205572.

  21. [21]

      Q. Wang, G. Zhang, W. Xing, et al., Angew. Chem. Int. Ed. 62 (2023) e202307930.

  22. [22]

      J. Liu, H. Liang, B. Wei, et al., Small Struct. 4 (2023) 2200379.

  23. [23]

      J.N. Yang, Y. Song, J.S. Yao, et al., J. Am. Chem. Soc. 142 (2020) 2956–2967. doi: 10.1021/jacs.9b11719

  24. [24]

      M. Li, Y. Zhao, J. Guo, et al., Nano-Micro Lett. 15 (2023) 119. doi: 10.32014/2023.2518-170x.304

  25. [25]

      L. Song, Q. Zhang, S. Ullah, et al., J. Mater. Chem. C 11 (2023) 4526–4535. doi: 10.1039/d3tc00406f

  26. [26]

      Y. Wang, Q. Zhou, Y. Zhu, et al., Appl. Catal. B 294 (2021) 120236.

  27. [27]

      Y. Li, C. Wang, M. Song, et al., Appl. Catal. B 243 (2019) 760–770.

  28. [28]

      L. Wang, B. Zhu, J. Zhang, et al., Matter 5 (2022) 4187–4211.

  29. [29]

      B.N. Choi, J.Y. Seo, Z. An, et al., Chem. Eng. J. 430 (2022) 132807.

  30. [30]

      B. He, P. Xiao, S. Wan, et al., Angew. Chem. Int. Ed. 62 (2023) e202313172.

  31. [31]

      H. Bian, T. Liu, D. Li, et al., Chem. Eng. J. 435 (2022) 135071.

  32. [32]

      X. Yue, L. Cheng, J. Fan, et al., Appl. Catal. B 304 (2022) 120979.

  33. [33]

      W. Yan, Y. Zhang, Y. Bi, Angew. Chem. Int. Ed. 63 (2024) e202316459.

  34. [34]

      Q. Ren, Y. He, H. Wang, et al., Research 6 (2023) 0244.

  35. [35]

      X. Yan, C. Duan, S. Yu, et al., J. CO₂ Util. 79 (2024) 102648.

  36. [36]

      S. Cheng, Z. Sun, K.H. Lim, et al., Appl. Catal. B 343 (2024) 123583.

  37. [37]

      N.H. Kwon, J. Park, X. Jin, et al., ACS Nano 17 (2023) 23732–23745. doi: 10.1021/acsnano.3c07566

• 

  Figure 1  (a) Schematic of the ball-milling process to prepare CsPbBr₃ using different solvents. (b) XRD patterns. (c-f) SEM images. (g) Br 3d XPS spectra. (h) Fitted results of the Br/Pb ratio. (i) EPR spectra and (j) Br vacancy density (represented by peak intensity of the EPR spectra) for different CsPbBr₃ samples.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (a) Production rates of CH₄ and CO. (b) Relationship between Br vacancy quasi-density and the surface area-normalized electron consumption rate. (c) The ratio of CH₄/CO production rate. (d) Relationship between Br vacancy quasi-density and the surface area-normalized production rate of CH₄ and CO. (e) ¹³CH₄ and ¹³CO mass spectra for ¹³CO₂ photocatalytic reduction on CsPbBr₃—HX. (f) Stability test of CsPbBr₃—HX at 150 ℃ under 3 sun irradiation.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) Surface photovoltage plots. (b) Transient absorption spectroscopy curves. In-situ DRIFT spectra of CsPbBr₃ prepared with different solvents: Wavelength range of 1810–3600 cm^-1 (c), and 1000–1550 cm^-1 (d). Test process: Before test, the reaction cell was purged with Ar at a flow rate of 30 sccm and heated at 150 ℃ for 20 min. Then, the gas input was switched to CO₂ and H₂O vapor at a flow rate of 5 sccm. A background spectrum was collected after a 15 min equilibrium. The light was then turned on and the in-situ spectra after 20 min equilibrium were recorded.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  (a) CO₂-TPD and (b) the change of adsorption temperature at P2 desorption peaks for CO₂-TPD. (c) CO-TPD and (d) the change of adsorption temperature at P2′ desorption peaks for CO-TPD. (e) CO₂ and (f) CO adsorption energy of CsPbBr₃ with different number of defects.

  下载: 全尺寸图片 幻灯片
•