English

• Ever since the industrial revolution, the extensive emissions of CO₂ resulting from rapid fossil fuel consumption have disrupted the natural carbon cycle, bringing the environmental crisis of global warming and frequent extreme weather events [1,2]. Consequently, utilizing green energy to convert excess carbon dioxide to restore the carbon cycle has become increasingly important [3,4]. Given that solar energy represents the largest renewable energy source available to humanity, harnessing sunlight to drive the catalytic reaction between CO₂ and H₂O to produce solar fuels, which is also called as artificial photosynthesis, is regarded as a promising yet challenging approach [5,6].

  The paramount objective in the research of artificial photosynthesis for CO₂ reduction is the development of highly efficient photocatalysts [7-9]. The photocatalytic process fundamentally consists of three core steps: light absorption, charge separation, and surface catalytic conversion, all of which collectively determine the efficiency of solar energy conversion [10]. Metal halide perovskites are considered promising candidates due to their broad light absorption range and exceptional optoelectronic properties, where these merits have been well demonstrated in the photovoltaic and LED applications [11-13]. However, during the photocatalytic reduction of CO₂ using perovskites, the photogenerated charges encounter severe recombination, primarily due to the insufficient charge separation, leading to low efficiency for solar CO₂ reduction [14,15]. In addition, the sluggish surface catalysis, especially the slow rate of water oxidation, results in reduced consumption rate of the generated charges, leading to the accumulation of excess photocharges and further exacerbating charge recombination [16]. Therefore, enabling efficient charge separation and fast catalytic kinetics become critical priorities for constructing highly efficient perovskite photocatalysts.

  Building heterojunctions has emerged as an effective strategy to develop highly efficient perovskite photocatalysts [17,18]. Heterojunctions not only establish built-in electric fields that facilitate the effective separation of photogenerated charges but also enable the integration of additional photocatalysts with excellent water oxidation capabilities to accelerate the photocatalytic conversion process [19-21]. Previous studies have successfully developed heterojunction systems based on perovskites, including W₁₈O₄₉/CsPbBr₃ [22], TiO₂/CsPbBr₃ [23], and Cs₃Bi₂I₉/BiVO₄ [24], which demonstrate enhanced photocatalytic efficiency. However, those heterojunctions are often created through simple physical mixing or bulk-to-bulk growth methods, resulting in inadequate interface contact or limited contact area [25-28]. Such inadequacies significantly hinder the effective range of the built-in electric field and the fluent charge separation at the interface, preventing the heterojunction from fully exploiting its advantages. In this regard, exploring heterojunctions with intimate while vast interfacial contact is thus intriguing.

  As an inorganic perovskite material, CsPbBr₃ (denoted as CPB) exhibits a visible-light response and demonstrates a good ability to catalyze CO₂ reduction [29,30]. Moreover, the intrinsic crystal structure of perovskites endows CsPbBr₃ with high lattice distortion tolerance, making it an ideal host for constructing heterojunctions with favorable interfacial lattice contact [31,32]. In this study, leveraging the soft lattice characteristic of halide perovskite, indium oxide (In₂O₃, denoted as IO) nanoparticles, which serve as effective water oxidation photocatalysts, were deliberately introduced into the precursors as seeds to grow IO/CPB composites. Morphological characterizations indicated that the CPB microparticles were internally and externally decorated with IO seeds, forming heterojunctions with intimate and dense contact. The charge transfer mechanism and carrier dynamics in the heterojunction were explored by in-situ X-ray photoelectron spectroscopy, surface photovoltage measurement, and transient absorption technique. The surface catalytic kinetics, including the chemisorption of reactants and the reaction intermediates, were further investigated by CO/CO₂-temperature programmed desorption and in-situ infrared spectroscopy detections. Thanks to the benefits of the heterojunctions, the IO/CPB photocatalysts demonstrated a significantly enhanced performance to reduce CO₂ to CO and CH₄, surpassing the activities of the bare CPB and IO photocatalysts.

  Fig. 1a depicts the preparation process to build the In₂O₃/CsPbBr₃ (denoted as IO/CPB) heterojunction. In₂O₃ nanoparticles were synthesized in advance using the typical precipitation method with In(NO₃)₃ and NH₃·H₂O. Subsequently, the In₂O₃/CsPbBr₃ catalysts were prepared via a wet-mechanochemical method using the ball-milling process, where CsBr, PbBr₂ precursors, and In₂O₃ nanoparticles were thoroughly mixed. X-ray diffraction (XRD) analysis confirms that the In₂O₃ (denoted as IO) powders exhibited a typical cubic phase (JCPDS No. 06–0416), while the pure CsPbBr₃ (denoted as CPB) was identified as having an orthorhombic phase (JCPDS No. 18–0364) (Fig. 1b). The incorporation of IO during ball-milling process of CsPbBr₃ yielded IO/CPB composite that displayed a mixed XRD pattern, where characteristic peaks corresponding to CPB and IO are retained (Fig. 1b). A positive relationship is also observed between the intensities of the diffraction peaks of IO and its weight ratios in the IO/CPB heterojunctions. Notably, the diffraction peak of IO at 30.58° (assigned to (222) plane) is closely aligned with the (200) plane of CsPbBr₃ at 30.70°, suggesting comparable lattice constants for these two materials at specific crystal planes. This alignment indicates some possibilities of lattice matching at the IO/CPB heterojunction interface, which is further evidenced by the nested grain structure of the IO/CPB composites.

  Figure 1

  

  Figure 1.  (a) Schematic of the ball-milling process for the preparation of IO/CPB heterojunction. (b) XRD patterns of IO, CPB and IO/CPB with different IO-to-CPB ratios. SEM images of (c) CPB, (d) IO, and (e) IO/CPB. (f) HRTEM and (g) enlarged view of IO/CPB heterojunction. (h) High-angle annular dark-field (HAADF) image and elemental mappings of Pb, Br, Cs, In, and O elements for IO/CPB heterojunction.

  DownLoad: Full-Size Img PowerPoint

  The morphologies of different catalysts were then examined using electron microscopies. Pure CsPbBr₃ powders comprised large particles with dense grain boundaries, measuring several micrometers in diameter, as shown in the scanning electron microscopy (SEM) (Fig. 1c and Fig. S1 in Supporting information) and transmission electron microscopy (TEM) images (Fig. S2 in Supporting information). The In₂O₃ powders, in contrast, consisted of small crystalline particles with diameters of <20 nanometers, as illustrated in the SEM image (Fig. 1d) and TEM images (Fig. S3 in Supporting information). For the IO/CPB composites, the SEM image in Fig. 1e and the corresponding elemental mapping (Fig. S4 in Supporting information) reveal that the large CPB granules were uniformly decorated with IO nanoparticles, forming the IO/CPB heterojunctions. Further TEM characterizations (Figs. 1f and g, and Fig. S5 in Supporting information) displayed that, compared to the regular shape of pure CPB in Fig. S2, the structure of CPB became a granule host to wrap the IO seeds, forming the IO/CPB heterojunction. Specifically, the IO crystals with a characteristic spacing of 0.29 nm were located in the surrounding and inner regions of the IO/CPB composites. This suggests that IO nanoparticles are embedded as seeds in the CPB granules, both internally and externally, indicating the formation of intimate and extensive interfacial contacts within the IO/CPB heterojunctions. Additionally, the high-angle annular dark-field (HAADF) image and corresponding elemental mappings (Fig. 1h) show that the In elements were decorated around and within the region of Pb elements, further confirming the formation of In₂O₃/CsPbBr₃ heterojunction with vast heterointerfaces.

  The energy band structures of the IO/CPB heterojunction were then investigated. UV–vis absorption spectra (Fig. S6 in Supporting information) of bare CPB, IO and IO/CPB composites show that CPB exhibited a much wider light response range and a higher light harvesting ability than IO. As the IO and CPB are indirect and direct bandgap semiconductors, respectively, the bandgaps (E_g) of CPB and IO were then determined by the Tauc plots derived from the UV–vis spectra, which are 2.28 eV for CPB and 2.70 eV for IO, respectively (Fig. 2a). Ultraviolet photoelectron spectroscopy technique was further used to determine the band position of CPB and IO. The cut-off edges (E_cutoff) for CPB and IO were determined as 17.15 and 16.71 eV, respectively (Fig. 2b). By subtracting the cut-off edge from the excitation energy (hv = 21.2 eV), the Fermi-level energy (E_f) of CPB and IO were calculated as 4.05 and 4.49 eV (hv − E_cutoff), respectively. The Fermi edges (E_edge) from the UPS spectra for CPB and IO were around 2.15 and 2.57 eV, respectively (Fig. 2c), where the E_edge represents the difference between the E_f and the valence band maximum (E_VBM). The conduction band minimum (E_CBM) then could be determined by E_VBM minus E_g. Further calibrating to the vacuum level to reversible hydrogen level (4.45 eV for 0 V_RHE), the E_f, E_VBM, and E_CBM were then determined as −0.40, 1.75, and −0.53 V_RHE for CPB, and 0.04, 2.61, and −0.09 V_RHE for IO, respectively. To further affirm the difference of the Fermi level of IO and CPB, the work functions of the two semiconductors were also evaluated by density functional theory calculations, which are determined as 4.15 and 4.68 eV for CPB and IO, respectively (Fig. S7 in Supporting information). By subtracting these values from the vacuum level, the E_f of CPB (−0.30 eV) and IO (0.13 eV) were derived, where the former remained more negative than the latter, aligning with UPS results. A schematic diagram of CPB and IO band structure was illustrated in Fig. 2f, where two semiconductors are in the condition without contact.

  Figure 2

  

  Figure 2.  (a) Tauc plots, (b) cut-off edge of UPS spectra, and (c) Fermi-edge of UPS spectra for CPB and IO catalysts. (d) Pb 4f XPS spectra for CPB, IO/CPB in dark and IO/CPB under light. (e) In 3d XPS spectra for IO, IO/CPB in dark and IO/CPB under light. Band diagrams for IO/CPB heterojunction (f) before contact, (g) after contact in dark, and (h) after contact under light.

  DownLoad: Full-Size Img PowerPoint

  X-ray photoelectron spectroscopy (XPS) was further conducted to examine the surface chemical states of different catalysts, as well as to elucidate the electron transfer mechanisms. The shifts in the binding energies of elements stem from the changes in electron densities around the constituent elements. Therefore, comparing the shifts in XPS spectra of specific elements in the heterojunctions can provide insights into the charge transfer direction at the interface. Fig. 2d depicts the XPS spectra of Pb 4f for pure CPB, IO/CPB under dark and light conditions. After the integration of IO, the peak of Pb 4f for IO/CPB measured in dark exhibited a positive shift to higher binding energies relative to the pure CPB, indicating a loss of electrons around the Pb atoms. This shift corresponds to the electron flow from CPB to IO after heterojunction formation, which is driven by the more negative Fermi level of CPB compared to IO, as well as the built-in electric field at the heterojunction interface (Fig. 2g).

  When the heterojunction was set to light exposure, the Pb 4f peaks shifted to lower binding energies, with respect to those in dark, suggesting a reverse electron flow from IO to CPB under light. We also observed similar binding energy shifts in the Cs 3d and Br 3d XPS spectra for IO/CPB, confirming the reverse electron flow direction at the heterojunction interface under dark and light (Fig. S8 in Supporting information). In contrast, the In 3d XPS spectra for In₂O₃ showed a negative shift and a positive shift under dark and light, respectively (Fig. 2e), which is in line with the electron flow directions inferred from the analysis of Pb 4f XPS spectra. Consequently, we propose a direct Z-scheme charge transfer mechanism at the IO/CPB heterojunction (Fig. 2h), analogous to mechanisms established in other heterojunction photocatalysts [33,34]. Specifically, in the direct Z-scheme IO/CPB heterojunction, the photogenerated electrons in the conduction band of IO recombine with holes in the valence band of CPB, while the more energetic photoelectrons on CPB and photoholes on IO were utilized to drive the CO₂ reduction and water oxidation reaction, respectively.

  Solar CO₂ reduction over the catalysts was then performed in a designed gas-solid phase photoreactor with continuous CO₂ and H₂O vapor flow. The reduction products were analyzed by an online gas chromatography, where CO and CH₄ were detected as the carbonaceous product and H₂ as the byproduct. Fig. 3a depicts the production rates of the detected products across different photocatalysts. The pristine CPB exhibited production rates of H₂, CO, and CH₄ at 0.34 ± 0.10, 2.01 ± 0.11, and 3.09 ± 0.18 µmol g^-1 h^-1, respectively, demonstrating a preference for CO₂ photoreduction over hydrogen evolving on CPB surface. On the contrary, the IO catalysts produced much more H₂ with a production rate of 3.92 ± 0.24 µmol g^-1 h^-1, higher than those of CO (2.03 ± 0.05 µmol g^-1 h^-1) and CH₄ (0.42 ± 0.06 µmol g^-1 h^-1) (Fig. 3a). Here, CO was identified as the primary CO₂ reduction product on the IO catalysts, while CPB reduced CO₂ to yield more CH₄, suggesting distinct preferences of CO₂ reduction on the two semiconductors. Notably, the IO catalysts produced a greater quantity of products compared to pure CPB, which might be attributed to the nanosized granular structure and, hence, the much larger surface area of IO nanoparticles (Fig. S9 in Supporting information).

  Figure 3

  

  Figure 3.  (a) Production rates, (b) total electron consumption rates and (c) product selectivities for photocatalytic CO₂ reduction using water gas on CPB, IO, and IO/CPB with different IO-to-CPB weight ratios. (d) Production rates (right) and selectivities (left) of CH₄, CO and H₂ for solar CO₂ reduction on IO/CPB and IO—CPB. (e) Production rates for control experiments: no light, no catalyst, no H₂O, and no CO₂. (f) Mass spectra of ¹³CO and ¹³CH₄ for isotopic tracing CO₂ photoreduction reaction.

  DownLoad: Full-Size Img PowerPoint

  Once the CPB was decorated with IO, the overall performances for solar CO₂ reduction over the IO/CPB composites were improved as compared to either CPB or IO. The optimized IO/CPB heterojunction with an IO-to-CPB weight ratio of 1:1 reached the highest production rates of 4.71 ± 0.40 µmol g^-1 h^-1 for CO and 16.08 ± 0.64 µmol g^-1 h^-1 for CH₄. As we counted on the total electron consumption rate during photocatalytic CO₂ reduction, the activity of IO/CPB (1:1) was measured as 140.22 ± 6.03 µmol g^-1 h^-1, which represents 4.8 times and 9.2 times that of pure CPB (29.42 ± 1.86 µmol g^-1 h^-1) and pure IO (15.26 ± 1.07 µmol g^-1 h^-1), respectively (Fig. 3b). This clearly demonstrates the superiority of the heterojunctions.

  The selectivities for different reduction products were further assessed (Fig. 3c). Pure CPB predominantly produced CH₄ with a selectivity of 84%, while IO primarily generated H₂ (51%). The integration of CPB and IO at varying ratios exhibited significantly higher selectivity for CO₂ reduction compared to hydrogen evolution, where the highest H₂ selectivity was observed as only 12.9% for the IO/CPB with a high IO weight ratio of 66.6%. This suggests that the photocatalytic CO₂ reduction predominantly occurs on the CPB surface within the IO/CPB heterojunction, further corroborating the direct Z-scheme band structure of this heterojunction. For the optimized IO/CPB (1:1), in addition to the high rate for solar CO₂ reduction, a notable CH₄ selectivity of 91.7% was achieved, which is comparable to the performances of most other reported high-efficiency perovskite-based photocatalytic systems (Table S1 in Supporting information).

  To demonstrate the advantages of the abundant and intimate contact present in the IO/CPB heterojunction, we synthesized IO—CPB composites via ball-milling IO nanoparticles together with as-prepared CPB powders. As illustrated in Fig. 3d, the physically connected IO—CPB composite exhibited H₂, CO, and CH₄ production rates of 1.61 ± 0.20, 2.81 ± 0.21, and 3.21 ± 0.21 µmol g⁻¹ h⁻¹, respectively, with the selectivity for CH₄ at only 74.4%. This performance is significantly inferior to that of the IO/CPB heterojunctions, highlighting the enhanced photocatalytic activity afforded by the close and rich interfacial interactions in heterojunction catalysts.

  The stability of the IO/CPB heterojunction for solar CO₂ reduction was further evaluated (Fig. S10 in Supporting information). The IO/CPB kept producing CH₄, CO, and H₂ during the recycling tests, where the catalyst performance (based on the e^- consumption rate) was retained at about 86% of the initial activity after 6 h, with a CH₄ selectivity being higher than 87%. Postmortem analysis by SEM and XRD characterizations further showed that the IO/CPB maintained its morphology and crystallinity (Fig. S11 in Supporting information), demonstrating the durability of the IO/CPB heterojunction for photocatalysis. The oxidation products for solar CO₂ reduction over IO/CPB (1:1) during the stability test were also measured by the online gas chromatograph, where O₂ was detected. The amounts of O₂ evolved were nonstoichiometric to the electron consumption rates, nearly 76% of the theoretical O₂ evolving value (Fig. S12 in Supporting information). We reasoned that the missing O₂ over the IO/CPB heterojunction could be possibly attributed to the yield of other undetected oxidation species.

  We further conducted control experiments and ¹³C isotopic tracking to verify the reliability of photocatalytic activity for CO₂ reduction. Fig. 3e compares the production rates under various control conditions, including no light, no catalyst, no H₂O, and no CO₂ supply. No products were detected under the condition without light irradiation, while negligible amounts of H₂ and CO were observed where no photocatalysts or water vapor was present. After accounting for the potential contributions of photochemical reactions on trace organic or water residuals, the measured activities of H₂, CO, and CH₄ evolution as discussed above were confirmed to originate from the photocatalytic reactions involving CO₂ and water vapor substrates. For the control experiment without CO₂ supply (with water vapor available), H₂ was detected as a reduction product (Fig. 3e). This implies the catalytic ability of IO/CPB composite for water splitting. Fig. 3f depicts the mass spectra for isotopic tracing of ¹³CO₂ photoreduction, showing that ¹³CO and ¹³CH₄ were detected in the output gas of the photoreactor. This observation strongly confirms that the carbonaceous products were derived from the reduction of ¹³CO₂ input.

  From the above investigations, we have demonstrated the effectiveness of heterojunction in enhancing the photocatalytic performance for solar CO₂ reduction. We analyzed the difference in surface area among the CPB, IO, and IO/CPB heterojunctions (Fig. S9), where the heterojunction exhibited a surface area of 6.8 ± 0.3 m²/g, representing a 0.8-fold increase compared to the bare CPB, which had a surface area of 3.6 ± 0.5 m²/g. This indicates that the observed 3.8-fold increase in the activity of IO/CPB cannot be solely attributed to the difference in surface area but rather to the unique functionality of the heterojunction.

  We first assessed the impact of the direct Z-scheme heterojunction on charge carrier dynamics. As indicated in the UV–vis spectra (Fig. S6), the integration of IO with CPB did not significantly enhance the light-harvesting capability of the photocatalyst. Therefore, the observed improvement in photocatalytic performance cannot be attributed to light absorption. Next, we evaluated the effect of the direct Z-scheme heterojunction on charge separation. Fig. 4a illustrates the surface photovoltage (SPV) of IO/CPB (1:1), CPB, and IO photocatalysts. The SPV is a well-established measure of surface charge density, with its value directly correlated to the charge separation efficiency of the photocatalysts. Notably, the IO/CPB direct Z-scheme heterojunction exhibited a significantly larger SPV value compared to both CPB and IO, clearly indicating enhanced charge separation facilitated by the direct Z-scheme heterojunction.

  Figure 4

  

  Figure 4.  (a) Surface photovoltage and (b) transient absorption decay kinetics of CPB, IO, and IO/CPB (1:1) catalysts.

  DownLoad: Full-Size Img PowerPoint

  The photogenerated charge carrier lifetimes of various photocatalysts were further assessed using the transient absorption (TA) technique. Both CPB and IO exhibited photoinduced transient absorption bands centered at 515 nm and 478 nm, respectively, corresponding to the signals of photogenerated holes in these two n-type semiconductors (Fig. S13 in Supporting information). The decay of the transient absorptions was probed at the band centers, while the IO/CPB heterojunction was also probed at the absorption peak of 508 nm. Fig. 4b presents the TA kinetics for the three catalysts, along with the fitting results provided in Table S2 (Supporting information). Notably, the transient absorption decay of the IO/CPB heterojunction was significantly slower than that of the individual photocatalysts. The fitted lifetime of the photogenerated holes in the IO/CPB heterojunction was determined to be 1232.06 ps, which is longer than the lifetimes of CPB (951.28 ps) and IO (899.68 ps). It is clear that the photogenerated holes survive a longer lifetime in the IO/CPB heterojunction.

  To further elucidate the positive effect of intimate heterojunction on charge separation and transfer, we have conducted photoelectrochemical characterizations, including chopped light photocurrent measurement and electrochemical impedance spectra. The IO/CPB heterojunction exhibits a higher photocurrent density and a smaller arc radius of the Nyquist plot as compared to the pure CPB and IO—CPB with fewer contacts (Fig. S14 in Supporting information), clearly supporting the promoted separation and transfer of photogenerated charges of the heterojunction.

  Hence, we can summarize that the direct Z-scheme heterojunction enhances the charge separation and prolongs the photocarrier lifetime, thereby contributing to the enhanced catalytic activity for solar CO₂ reduction.

  We then turn to investigate the surface catalytic reactions for solar CO₂ reduction over different photocatalysts. The chemisorption and activation of CO₂ and CO molecules on the catalysts were first analyzed using the temperature-programmed desorption (TPD) technique. The interaction between the surface-active sites (mainly base-site metal atoms) and the CO₂ or CO molecules could be indicated as the desorption temperature of the adsorbates, while the adsorption amounts over the catalysts could be inferred as the integrated areas of the TPD profiles, where Table S3 (Supporting information) detailed the mathematically-integrated area of the following TPD curves [35].

  As shown in the CO₂-TPD curves in Fig. 5a, the IO/CPB exhibits a slightly higher desorption temperature and a greater adsorption capacity of CO₂ as compared to the sole CPB (Fig. S15 and Table S3 in Supporting information). The elevated desorption temperature indicates a stronger bonding strength between the adsorbed CO₂ and the surface metal atoms, suggesting enhanced CO₂ activation over the IO/CPB surface. This enhanced interaction is likely due to the stronger affinity between the acidic CO₂ with the more basic Pb atoms in IO/CPB, which exhibit a lower surface electron density as evidenced by the positive shift in the Pb 4f XPS spectra (Fig. 2d). The difference in the CO₂ adsorption amount of IO/CPB with respect to CPB could be further quantified by the integrated area, where a 1.3-fold increase is determined. As we counted the 0.8-fold increase in the BET surface area, we could conclude that the formation of heterojunction did contribute to adsorbing more CO₂ molecules on the IO/CPB photocatalysts. Hence, a strengthened chemisorption and activation of CO₂ reactants is established on the IO/CPB heterojunction, also contributing to the improved photocatalytic activity.

  Figure 5

  

  Figure 5.  (a) CO₂-TPD and (b) CO-TPD profiles of CPB, IO, and IO/CPB (1:1) catalysts. (c) In-situ DRIFT spectra of IO/CPB (1:1) catalyst with the CO₂/H₂O supply under dark and light. The spectra were divided into three ranges of 3800–2600, 2550–2050, and 1800–900 cm^-1 for a better view.

  DownLoad: Full-Size Img PowerPoint

  For bare IO, which possesses a significantly larger surface area, the lower desorption temperature and the reduced adsorption amount, when normalized to the surface area, indicate a weaker adsorption of CO₂ on the IO surface. This is consistent with the poor activity of IO for solar CO₂ reduction.

  Fig. 5b presents the CO-TPD profiles of the three photocatalysts, given that the CH₄ products usually stem from the further reduction of CO intermediates. We observed comparable CO-desorption peak temperatures for CPB and IO/CPB, while a lower temperature for IO (Fig. S16 in Supporting information). In addition, the adsorption capacities of CO normalized to the surface area for the CPB and IO/CPB photocatalysts were also found to be similar and much higher than that of IO (Table S3). These findings correspond well with the higher CH₄ selectivity observed for the former two catalysts compared to the latter.

  In-situ diffuse-reflectance infrared Fourier transform spectroscopy (DRIFTS) was carried out to further identify the surface reaction intermediates to reveal the photocatalytic CO₂ reduction mechanism. As depicted in Fig. 5c, the absorption bands around 1020 and 1410 cm^-1 are assigned to the species of monodentate carbonate (m-CO₃^2–) and bidentate carbonate (b-CO₃^2–), respectively, corresponding to the chemisorption of CO₂ on the heterojunction surface [36-38]. Other adsorbed species of *OH (O—H vibration around 1578 cm^-1), CO₂ (C=O stretch at 2361 and 2340 cm^-1) and H₂O (O—H stretch in the range of 3300-3600 cm^-1) were also identified in the spectra [39]. Upon light irradiation, new absorption bands around 2971, 2877, and 2828 cm^-1 were observed, which are characteristic peaks of C—H stretch bands from either methyl species (CH₃*), methoxyl species (CH₃O*) or the absorbed CH₄ products [40]. Besides the three peaks, a new peak at 1219 cm^-1 also arose after light irradiation, which could be assigned to the C—O stretch band [40]. Together, it is then inferred that a CH₃O* intermediate is involved in the photocatalytic CO₂ reduction over IO/CPB heterojunction. Given that the CH₃O* usually evolves by the further reduction of *CO intermediates [40,41], a possible reaction path that involves CH₃O* intermediates for solar CO₂ methanation was proposed: (ⅰ) CO₂ molecules are first adsorbed on the catalyst surface to form m-CO₃^2-, b-CO₃^2- or *CO₂ adsorbates; (ⅱ) the *CO₂ adsorbates are reduced to *CO species, where some are desorbed to yield CO product; (ⅲ) the *CO adsorbates are further hydrogenated at the carbon atom to yield CH₃O*; (ⅳ) the key CH₃O* intermediates are finally reduced to evolve CH₄.

  In summary, we exploit the soft lattice characteristics of CsPbBr₃ by intentionally incorporating indium oxide nanoparticles as seeds in situ during the crystallization process of CPB perovskite. We successfully decorated the CPB crystals with IO nanoparticles both internally and externally, constructing IO/CPB heterojunction with rich and dense interface contact. Further applying in situ X-ray photoelectron spectroscopy and band structure analysis, we found that the IO/CPB heterojunction exhibited a direct Z-type charge transfer mechanism, endowing the potent photogenerated electrons and holes on CPB and IO for their participation in photocatalytic CO₂ reduction and water oxidation, respectively. Investigation on the charge carrier dynamics by surface photovoltage and transient absorption measurements clearly suggested an enhanced charge separation and extended carrier lifetime in the IO/CPB heterojunction. Surface catalytic analysis of reactant chemisorption and intermediate detection further indicates promoted CO₂ adsorption and activation. Accordingly, the IO/CPB heterojunction with optimal IO to CPB weight ratio of 1:1 shows a remarkable enhancement in CO₂ photoreduction performance, achieving peak production rates of 4.71 ± 0.40 µmol g⁻¹ h⁻¹ for CO and 16.08 ± 0.64 µmol g⁻¹ h⁻¹ for CH₄. This work proposes a simple approach to build direct Z-scheme heterojunctions, and highlights the importance of intimate and vast interfacial contacts for enhanced charge separation and hence promoted photocatalytic activity.

  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

  Juhong Lian: Writing – original draft, Methodology, Investigation, Funding acquisition, Data curation, Conceptualization. Deng Li: Writing – review & editing, Writing – original draft, Project administration, Methodology, Investigation, Formal analysis, Conceptualization. Yongmei Ma: Visualization, Validation, Methodology, Investigation. Hui Bian: Validation, Methodology, Data curation. Yifan Shao: Validation, Resources, Methodology. Zitong Wang: Methodology, Data curation. Junqing Yan: Supervision, Funding acquisition, Conceptualization. Ruibin Jiang: Visualization, Project administration, Conceptualization. Shengzhong (Frank) Liu: Supervision, Methodology, Funding acquisition. Fuxiang Zhang: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization.

  Acknowledgments

  This work was financially supported by the National Natural Science Foundation of China (Nos. 22302148, 22261160369), Natural Science Foundation of Shaanxi Province (No. 2024JC-YBQN-0468), Science and Technology Program of Shaanxi Province (No. 2024ZC-KJXX-059), Key Research and Development Project of Weinan City (No. 2024ZDYFJH-636), Scientific Research Program Funded by Education Department of Shaanxi Provincial Government (No. 22JK0379) and Fundamental Research Funds for the Central Universities (No. GK202406030).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.111394.

  
  
• 1. [1]

     D. Tong, Q. Zhang, Y. Zheng, et al., Nature 572 (2019) 373. doi: 10.1038/s41586-019-1364-3

  2. [2]

     D.I. Armstrong McKay, A. Staal, J.F. Abrams, et al., Science 377 (2022) eabn7950. doi: 10.1126/science.abn7950

  3. [3]

     S. Perathoner, G. Centi, ChemSusChem 7 (2014) 1274. doi: 10.1002/cssc.201300926

  4. [4]

     Z. Wang, H. Song, H. Liu, J. Ye, Angew. Chem. Int. Ed. 59 (2020) 8016. doi: 10.1002/anie.201907443

  5. [5]

     W. Gu, J. Hu, L. Li, et al., Environ. Sci. Technol. 58 (2024) 20434. doi: 10.1021/acs.est.4c05605

  6. [6]

     J.K. Stolarczyk, S. Bhattacharyya, L. Polavarapu, J. Feldmann, ACS Catal. 8 (2018) 3602. doi: 10.1021/acscatal.8b00791

  7. [7]

     T.H. Wang, Y.S. Lai, C.K. Tsai, et al., Environ. Sci. Technol. 58 (2024) 13717. doi: 10.1021/acs.est.3c10274

  8. [8]

     S. Chen, C. Li, K. Domen, F. Zhang, Joule 7 (2023) 2445. doi: 10.1016/j.joule.2023.10.004

  9. [9]

     A.A. Saraev, A.Y. Kurenkova, D.D. Mishchenko, et al., Trans. Tianjin Univ. 30 (2024) 140. doi: 10.1007/s12209-024-00384-3

  10. [10]

      Z. Zhou, W. Guo, T. Yang, et al., Chin. J. Struct. Chem. 43 (2024) 100245.

  11. [11]

      H. Huang, B. Pradhan, J. Hofkens, M.B.J. Roeffaers, J.A. Steele, ACS Energy Lett. 5 (2020) 1107. doi: 10.1021/acsenergylett.0c00058

  12. [12]

      J. Wang, Y. Shi, Y. Wang, Z. Li, ACS Energy Lett. 7 (2022) 2043. doi: 10.1021/acsenergylett.2c00752

  13. [13]

      N. Guo, Z. Liu, Y. Mu, et al., Chin. Chem. Lett. 33 (2022) 3039. doi: 10.1016/j.cclet.2021.09.033

  14. [14]

      L. Zhang, H. Qiu, R. Shi, et al., J. Phys. Chem. Lett. 14 (2023) 1823. doi: 10.1021/acs.jpclett.3c00198

  15. [15]

      R. Das, A. Patra, S.K. Dutta, S. Shyamal, N. Pradhan, J. Am. Chem. Soc. 144 (2022) 18629. doi: 10.1021/jacs.2c08639

  16. [16]

      Z. Chen, Y. Hu, J. Wang, et al., Chem. Mater. 32 (2020) 1517. doi: 10.1021/acs.chemmater.9b04582

  17. [17]

      F. Li, Z. Fang, Z. Xu, Q. Xiang, Energy Environ. Sci. 17 (2024) 497. doi: 10.1039/d3ee03282e

  18. [18]

      H. Huang, D. Verhaeghe, B. Weng, et al., Angew. Chem. Int. Ed. 61 (2022) e202203261. doi: 10.1002/anie.202203261

  19. [19]

      L. Chen, J.T. Ren, Z.Y. Yuan, Adv. Energy Mater. 13 (2023) 2203720. doi: 10.1002/aenm.202203720

  20. [20]

      J. Wei, X. Mu, Y. Hu, et al., Angew. Chem. Int. Ed. 62 (2023) e202302986. doi: 10.1002/anie.202302986

  21. [21]

      X. Deng, Y. Ke, J. Ding, et al., Chin. Chem. Lett. 35 (2024) 109064. doi: 10.1016/j.cclet.2023.109064

  22. [22]

      X. Jiang, Z. Chen, Y. Shu, et al., Appl. Catal. B: Environ. Energy 348 (2024) 123840. doi: 10.1016/j.apcatb.2024.123840

  23. [23]

      W. Song, K.C. Chong, G. Qi, et al., J. Am. Chem. Soc. 146 (2024) 3303. doi: 10.1021/jacs.3c12073

  24. [24]

      L. Jiang, H. Du, L. Li, et al., Trans. Tianjin Univ. 29 (2023) 462. doi: 10.1007/s12209-023-00376-9

  25. [25]

      Z. Zhang, B. Wang, H. Zhao, et al., Appl. Catal. B: Environ. 312 (2022) 121358. doi: 10.1016/j.apcatb.2022.121358

  26. [26]

      W. Hou, H. Guo, M. Wu, L. Wang, ACS Nano 17 (2023) 20560. doi: 10.1021/acsnano.3c07411

  27. [27]

      J. Wang, J. Wang, N. Li, et al., ACS Appl. Mater. Interfaces 12 (2020) 31477. doi: 10.1021/acsami.0c08152

  28. [28]

      Z. Tang, C. Wang, W. He, et al., Chin. Chem. Lett. 33 (2022) 939. doi: 10.1016/j.cclet.2021.07.020

  29. [29]

      S. Chen, H. Yin, P. Liu, Y. Wang, H. Zhao, Adv. Mater. 35 (2023) 2203836. doi: 10.1002/adma.202203836

  30. [30]

      J. Sheng, Y. He, M. Huang, et al., ACS Catal. 12 (2022) 2915. doi: 10.1021/acscatal.2c00037

  31. [31]

      T. Lanigan-Atkins, X. He, M.J. Krogstad, et al., Nat. Mater. 20 (2021) 977. doi: 10.1038/s41563-021-00947-y

  32. [32]

      Q. Qian, Z. Wan, H. Takenaka, et al., Nat. Nanotechnol. 18 (2023) 357. doi: 10.1038/s41565-022-01306-x

  33. [33]

      X. Deng, J. Zhang, K. Qi, et al., Nat. Commun. 15 (2024) 4807. doi: 10.1038/s41467-024-49004-7

  34. [34]

      S. Wan, W. Wang, B. Cheng, et al., Nat. Commun. 15 (2024) 9612. doi: 10.1038/s41467-024-53951-6

  35. [35]

      J. Fu, L. Zhu, K. Jiang, et al., Chem. Eng. J. 415 (2021) 128982. doi: 10.1016/j.cej.2021.128982

  36. [36]

      F. Liu, J. Deng, B. Su, et al., ACS Catal. 15 (2025) 1018. doi: 10.1021/acscatal.4c06685

  37. [37]

      M. Zhou, H. Wang, R. Liu, et al., Angew. Chem. Int. Ed. 63 (2024) e202407468. doi: 10.1002/anie.202407468

  38. [38]

      B. Su, M. Zheng, W. Lin, et al., Adv. Energy Mater. 13 (2023) 2203290. doi: 10.1002/aenm.202203290

  39. [39]

      B. Su, Y. Kong, S. Wang, et al., J. Am. Chem. Soc. 145 (2023) 27415. doi: 10.1021/jacs.3c08311

  40. [40]

      H. Bian, T. Liu, D. Li, et al., Chem. Eng. J. 435 (2022) 135071. doi: 10.1016/j.cej.2022.135071

  41. [41]

      W. Zhang, C. Deng, W. Wang, H. Sheng, J. Zhao, Adv. Mater. 36 (2024) 2405825. doi: 10.1002/adma.202405825

• 

  Figure 1  (a) Schematic of the ball-milling process for the preparation of IO/CPB heterojunction. (b) XRD patterns of IO, CPB and IO/CPB with different IO-to-CPB ratios. SEM images of (c) CPB, (d) IO, and (e) IO/CPB. (f) HRTEM and (g) enlarged view of IO/CPB heterojunction. (h) High-angle annular dark-field (HAADF) image and elemental mappings of Pb, Br, Cs, In, and O elements for IO/CPB heterojunction.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (a) Tauc plots, (b) cut-off edge of UPS spectra, and (c) Fermi-edge of UPS spectra for CPB and IO catalysts. (d) Pb 4f XPS spectra for CPB, IO/CPB in dark and IO/CPB under light. (e) In 3d XPS spectra for IO, IO/CPB in dark and IO/CPB under light. Band diagrams for IO/CPB heterojunction (f) before contact, (g) after contact in dark, and (h) after contact under light.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) Production rates, (b) total electron consumption rates and (c) product selectivities for photocatalytic CO₂ reduction using water gas on CPB, IO, and IO/CPB with different IO-to-CPB weight ratios. (d) Production rates (right) and selectivities (left) of CH₄, CO and H₂ for solar CO₂ reduction on IO/CPB and IO—CPB. (e) Production rates for control experiments: no light, no catalyst, no H₂O, and no CO₂. (f) Mass spectra of ¹³CO and ¹³CH₄ for isotopic tracing CO₂ photoreduction reaction.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  (a) Surface photovoltage and (b) transient absorption decay kinetics of CPB, IO, and IO/CPB (1:1) catalysts.

  下载: 全尺寸图片 幻灯片
  

  Figure 5  (a) CO₂-TPD and (b) CO-TPD profiles of CPB, IO, and IO/CPB (1:1) catalysts. (c) In-situ DRIFT spectra of IO/CPB (1:1) catalyst with the CO₂/H₂O supply under dark and light. The spectra were divided into three ranges of 3800–2600, 2550–2050, and 1800–900 cm^-1 for a better view.

  下载: 全尺寸图片 幻灯片
•