Artificial photosynthesis can directly utilize inexhaustible solar energy to reduce CO₂ into high value-added fuels or chemicals, which can not only provide partial sustainable energy supply, but also effectively alleviate global climate change induced by the excessive emission of greenhouse gas [1-6]. In an ideal artificial photosynthesis system, water is a desirable electron source for CO₂ photoreduction, which furnishes a "green chemistry" route to generate value-added chemicals without employing any expensive sacrificial reagents [7]. That is, the redox capacities of photocatalysts should straddles the requirements of both the CO₂ reduction and H₂O oxidation potentials [8]. Based on this prerequisite, a variety of semiconductor materials with wide bandgap have been widely exploited as candidates for photocatalytic CO₂ reduction coupled with H₂O oxidation, such as metal oxides and sulfides [9-11], metal-organic frameworks [12], layered double hydroxides [13], conjugated polymers [14] and so on, but the photocatalytic performance is limited by the weak visible light response and severe charge recombination [15]. Up to now, most of the researches are focused on the amelioration of light-harvesting capacity and photogenerated carrier separation efficiency [16-18].

Among various semiconductor materials, halide perovskite (HP) nanocrystals (NCs) have been widely employed as promising candidate catalysts in the field of photocatalysis [19-22], owing to their fascinating photophysical properties such as robust light-harvesting ability and high separation efficiency of photogenerated carrier [23]. The CO₂ photoreduction activity of HP NC based catalysts have been notably meliorated through multifarious strategies such as morphology modification [24, 25], decorating reduction [26-29] or oxidation cocatalysts [30-36]. However, the activity of current halide perovskite based catalysts for artificial is still low owing the deficient active site and/or insufficient separation of photogenerated carriers. Therefore, it is highly imperative to establish viable strategies to improve the performance of HP materials to meet the prerepuisite of practical applications. Recenlty, we have demonstrated that ultra-thin two-dimensional (2D) HP nanosheet (NS) exhibits higher catalytic activity for CO₂ photoreduction than conventional HP NCs, due to the increased low coordination metal atom ratio and short carrier diffusion distance of ultra-thin 2D HP NS [24]. In addition, it is widely recognized that the 2D materials are conducive to effective contact and charge separation between components in heterostructure [37, 38]. Based on these analyses, herein we reported a simple lattice-matched in-situ growth strategy for the construction of 2D CsPbBr₃ based heterostructure with perovskite oxide (SrTiO₃) as substrate to improve the photocatalytic CO₂ reduction performance of HP materirals. SrTiO₃ (space group Pm3m) possess the same crystal structure with cubic CsPbBr₃, and its lattice constant is exactly 2/3 of CsPbBr₃ (a = 3.905 vs. 5.830 Å). Therefore, the two cells of CsPbBr₃ happen to match the three cells of SrTiO₃ in a disproportionate manner, and the lattice mismatch factor is only 0.47% [39]. This low mismatch provides a reasonable condition for CsPbBr₃ to in-situ grown tightly on the SrTiO₃ surface. Furthermore, considering that SrTiO₃ NSs surface were peculiarly prone to precipitate strontium oxides [40, 41], SrTiO₃ NSs were treated by NH₄F etching to form Ti−O terminated surface (coded as T-SrTiO₃), which is beneficial to the interfacial charge separation. As expected, the resultant 2D/2D T-SrTiO₃/CsPbBr₃ heterostructure display significantly enhanced activity for CO₂ photoreduction.

The T-SrTiO₃/CsPbBr₃ composites were prepared by a two-step synthesis method as shown in Fig. S1 (Supporting information), and specific processes were described in Supporting information. Briefly, the SrTiO₃ NSs obtained by ethylene glycol assisted solvothermal method [42] were first treated with hot water and NH₄F solution to form Ti−O terminated T-SrTiO₃. Then the as-prepared T-SrTiO₃ NSs as a substrate were added into the hot-injection reaction system for in-situ growth of CsPbBr₃ NSs to generate T-SrTiO₃/CsPbBr₃ composite. For comparison, the SrTiO₃/CsPbBr₃ composites with untreated SrTiO₃ NSs were also synthesized under the same conditions. Transmission electron microscopy (TEM) measurements demonstrated that the as-prepared SrTiO₃ samples are layered shape structures with transverse size of about 20−50 nm (Fig. 1a), and the morphology does not change obviously after NH₄F treatment (Fig. 1b). Both the high-resolution TEM (HRTEM) images of SrTiO₃ and T-SrTiO₃ display 0.19 nm lattice spacing (insets in Figs. 1a and b), corresponding to the (200) crystal plane of SrTiO₃. The TEM images of composites (Fig. S2a in Supporting information and Fig. 1c) reveal that the surfaces of SrTiO₃ and T-SrTiO₃ NSs are successfully decorated with dispersed CsPbBr₃ NSs, the sizes of which (~20 nm) are similar to that obtained by identical condition without adding perovskite oxides substrate (Fig. S3 in Supporting information). The corresponding HRTEM images (Fig. S2b in Supporting information and Fig. 1d) exhibit distinct lattice spacing of 0.19 and 0.29 nm, which can be assigned to the (200) lattice planes of cubic phase SrTiO₃ and CsPbBr₃, respectively. High-angle annular dark-field scanning TEM (HAADF-STEM) and energy-dispersive X-ray spectroscopy (EDS) mapping measurements (Figs. S4 and S5 in Supporting information) further confirmed the successful generation of CsPbBr₃ on the perovskite oxide substrates, showing local concentrated element distributions of Cs, Pb and Br.

We further resorted to Fourier transform infrared (FTIR) spectroscopy to explore the influence of NH₄F etching on the surface structure of SrTiO₃. As shown in Fig. 2a, there are distinct characteristic peaks at 2848, 1514, 1371 and 1081 cm⁻¹ for SrTiO₃, which can be attributed to the vibrations of C−H, COO⁻, Sr−O and C−C−O, respectively [43]. These characteristic peaks obviously decrease or even disappear in T-SrTiO₃, indicating that most of the oxides and incomplete reactants were removed on its surface during the process of treatment. In addition, the stretching vibrations peak of NH₃⁺ at 1630 cm⁻¹ [44] in T-SrTiO₃ is noticeably increased in comparison with that in SrTiO₃, which may be related to the introduction of NH₄F. Meanwhile, the broad absorption peaks at the range of 736 and 559 cm⁻¹ assigned to the Ti−O−Ti stretching vibrations and the TiO₆ octahedron crystal lattice vibrations [45] have obviously enhanced intensity after treatment, indicating the exposure of Ti and O atoms on the surface of the T-SrTiO₃ NSs, which provides favorable condition for the tight growth of CsPbBr₃. Additionally, the contents of Sr, Ti and O elements on the surface of SrTiO₃ before and after treatment have been analyzed by XPS measurements, and the results showed that the content ratio of Sr, Ti and O elements on the surface of initial SrTiO₃ is close to 1:1:3, while the corresponding content is 0.55:1:3 after surface treatment (Table S1 in Supporting information). This result provides further strong evidence that Sr was etched during surface treatment.

Furthermore, the powder X-ray diffraction (PXRD) and ultraviolet-visible diffuse reflectance spectroscopy (UV−vis DRS) measurements (Fig. 2b and Fig. S6 in Supporting information) confirmed that the etching of trace NH₄F solution only treated the surface of SrTiO₃, while did not change its internal crystal structure and optical absorption. PXRD patterns (Fig. 2b) of the composites have the characteristic diffraction signals of SrTiO₃ (JCPDS No. 35−0734) [46] and CsPbBr₃ (JCPDS No. 54−0752) [47]. The absorption edges of T-SrTiO₃ and CsPbBr₃ NSs are 378 and 537 nm, respectively, which can also be clearly observed in T-SrTiO₃/CsPbBr₃ and SrTiO₃/CsPbBr₃ composites (Fig. S6). In addition, the X-ray photoelectron spectroscopy (XPS) measurements showed that the binding energy of T-SrTiO₃/CsPbBr₃ for Sr 3d, Ti 2p and O 1s have perceptible shift to high binding energy direction compared to those of pristine SrTiO₃, while the signals for Cs 3d, Pb 4f and Br 3d move in the opposite direction with respect to individual CsPbBr₃ as presented in Figs. 2c, d and Fig. S7 (Supporting information). The corresponding binding energies in SrTiO₃/CsPbBr₃ also have a similar change trend, but the amplitude is obviously weaker, indicating that there are stronger electron coupling between T-SrTiO₃ and CsPbBr₃ due to the tight contact between T-SrTiO₃ and CsPbBr₃.

The band gaps (E_g) of T-SrTiO₃ and CsPbBr₃ were determined by the Tauc plots derived from UV−vis DRS spectra (Fig. S8 in Supporting information), be 3.24 and 2.35 eV, respectively. The valence band edge potentials (E_VB) of T-SrTiO₃ and CsPbBr₃ can be calculated to be 2.34 and 1.23 V relative to normal hydrogen electrode (NHE) by the ultraviolet photoelectron spectroscopy (UPS) measurements (Fig. S9 in Supporting information). Combined with E_g and E_VB values, we can derive the conduction band edge potentials (E_CB) of T-SrTiO₃ and CsPbBr₃ to be −0.90 and −1.12 V (vs. NHE), respectively. Apparently, the T-SrTiO₃/CsPbBr₃ heterostructure should show staggered band arrangement as depicted in Fig. 3a. In addition, both T-SrTiO₃ and CsPbBr₃ feature n-type semiconductor characteristics because their Mott−Schottky plots have positive slopes at different frequencies (Fig. S10 in Supporting information). Therefore, their Fermi level (E_F) positions should be biased towards the edges of the corresponding conduction bands, which can be confirmed by the UPS spectra measurements (Fig. S9), locating at −0.52 and −0.41 eV (vs. NHE) for T-SrTiO₃ and CsPbBr₃ (Fig. 3a). When T-SrTiO₃ and CsPbBr₃ are in close contact, this difference in E_F will drive the transfer of interface free electrons from T-SrTiO₃ to CsPbBr₃ to achieve the E_F balance of the system. As a result, a built-in electric field pointing from T-SrTiO₃ to CsPbBr₃ will form at the heterostructure interface.

The resultant built-in electric field and band bending (Fig. S11 in Supporting information) will facilitate the transfer of photogenerated electrons from CsPbBr₃ to T-SrTiO₃, and vice versa for photogenerated holes. The in-situ irradiated XPS measurements were further performed to analyze the interfacial charge transfer orientation in T-SrTiO₃/CsPbBr₃ heterostructure. As shown in Fig. 3b, upon light irradiation, the binding energy of Br 3d in T-SrTiO₃/CsPbBr₃ shifted positively, which implies that hole accumulation occurred in CsPbBr₃. Meanwhile, the binding energies of Ti 2p (Fig. 3c) and O 1s (Fig. 3d) in T-SrTiO₃/CsPbBr₃ shifted negatively after light irradiation, indicating that the electron density increased around Ti and O cores. These light-induced binding energy shifts confirmed that the photogenerated electrons in CsPbBr₃ can transfer to SrTiO₃, and the holes in CsPbBr₃ can transfer to SrTiO₃, which accords with the traditional Ⅱ type double charge transfer mechanism.

The dynamics of photogenerated carriers in T-SrTiO₃/CsPbBr₃ heterostructure was further scrutinized by time-resolved photoluminescence (TRPL) measurements. As shown in Fig. 4a, the TRPL decay traces can be well fitted with a three-exponential function to determine the average lifetimes of PL decays (Table S2 in Supporting information). The TRPL decay curve of pure CsPbBr₃ NSs denote the radiative and nonradiative processes of photogenerated excitons in CsPbBr₃, showing an average lifetime of 43.88 ns. Decorating CsPbBr₃ on the surface of SrTiO₃ leads to a distinctly accelerated PL decay, and the average PL lifetime is shortened to 23.02 ns, indicating the occurrence of charge transfer between SrTiO₃ and CsPbBr₃ owing to their favorable energy-offset and the formation built-in electric field (Fig. S11). It is noted that replacing SrTiO₃ with T-SrTiO₃ results in a further significant acceleration of PL decay with an average PL lifetime of 7.64 ns, which indicates faster charge transfer at the T-SrTiO₃/CsPbBr₃ heterostructure interface in comparison with that at SrTiO₃/CsPbBr₃ counterpart, owing to the stronger electron coupling between T-SrTiO₃ and CsPbBr₃ (Figs. 2c and d). The efficient charge separation in T-SrTiO₃/CsPbBr₃ can be further proved by photocurrent response (I−t) and electrochemical impedance spectroscopy (EIS) experiments. As shown in Fig. 4b, the photocurrent density of T-SrTiO₃/CsPbBr₃ is significantly higher than those of SrTiO₃, CsPbBr₃ and SrTiO₃/CsPbBr₃. T-SrTiO₃/CsPbBr₃ also shows a smaller diameter of semicircular arc (Fig. 4c) than those of SrTiO₃, CsPbBr₃ and SrTiO₃/CsPbBr₃.

A gas-solid reactor (Fig. S12 in Supporting information) was employed to evaluate the photocatalytic CO₂ reduction activities of the as-prepared catalysts with water vapor as the electron donor. A 300 W xenon lamp was used as the illuminant with a light intensity of 100 mW/cm². The main reduction product was confirmed by gas chromatography analysis to be CO, and there are no other reduction products that could not be ignored (Fig. S13 in Supporting information). The CO formation rates of all photocatalysts were compiled in Fig. 4d. Both the SrTiO₃ and CsPbBr₃ NSs exhibit very poor photocatalytic activity for CO₂ reduction, with the CO formation rates of 12.1 ± 2.3 and 18.3 ± 3.1 µmol g⁻¹ h⁻¹, respectively, owing to the weak light absorption capacity and/or insufficient charge separation efficiency. SrTiO₃/CsPbBr₃ heterostructure exhibits a significant enhanced activity for CO₂ photoreduction, owing to the improved charge separation. Increasing the content of SrTiO₃ results in a volcanic tendency for CO₂ photoreduction activity (Fig. S14 in Supporting information). The evolution of CO reaches an optimal value of 65.5 ± 3.4 µmol g⁻¹ h⁻¹ (Fig. 4d), which is about 5.4 and 3.6 times higher than those of individual SrTiO₃ and CsPbBr₃ nanosheets, respectively. Moreover, the T-SrTiO₃/CsPbBr₃ heterostructure exhibits a further manifest enhancement of photocatalytic activity for CO₂ conversion due to the accelerated interfacial charge transfer. The CO generation rate of T-SrTiO₃/CsPbBr₃ reaches up to 120.2 ± 4.9 µmol g⁻¹ h⁻¹, and the corresponding electron consumption (R_electron = 2n_CO, n_CO denote the yield of CO) yield is 240.4 ± 9.8 µmol g⁻¹ h⁻¹, which surpasses the reported halide perovskite-based photocatalysts under the same conditions (Table S3 in Supporting information).

In addition, the photocatalytic stability of T-SrTiO₃/CsPbBr₃ was further estimated by cyclic test. As shown in Fig. S15 (Supporting information), the percentage decrease of CO formation rate after 5 cycles is less than 10%, indicating that T-SrTiO₃/CsPbBr₃ has good stability in the gas−solid reaction system. Further XRD and XPS measurements confirmed that the structure of T-SrTiO₃/CsPbBr₃ heterostructure is well maintained after the photocatalytic reaction (Figs. S16 and S17 in Supporting information). Control experiments were further carried out to determine the source of CO using T-SrTiO₃/CsPbBr₃ as photocatalyst. As shown in Fig. S18 (Supporting information), the product cannot be detected in the absence of light irradiation or photocatalyst, indicating that the CO₂ reduction is initiated by light on the photocatalyst. The blank experiment without CO₂ showed trace CO, which should be the result of the decomposition of residual surface ligands, indicating that the carbon source of CO mainly comes from CO₂. In addition, the control experiment in the absence of water vapor also showed negligible CO production, demonstrating that the electron source of CO₂ photoreduction came from water oxidation. These inferences can be further confirmed by ¹³CO₂ and H₂¹⁸O isotope labeling experiments (Fig. S19 in Supporting information). Mass spectrometry (MS) measurements can clearly detect signals at m/z = 29 corresponding to ¹³CO and at m/z = 36 assigning to ¹⁸O₂.

To summarize, we have successfully constructed 2D/2D T-SrTiO₃/CsPbBr₃ heterostructure by lattice-matched in-situ growth of CsPbBr₃ NSs on surface treated SrTiO₃ NSs. Thermodynamic characterization has demonstrated formation of type Ⅱ heterostructure between T-SrTiO₃ and CsPbBr₃. The favorable energy-offset and strong electron coupling at heterostructure interface result in an efficient charge separation as confirmed by joint XPS, TRPL and electrochemical investigation, which endows T-SrTiO₃/CsPbBr₃ heterostructure exhibiting significantly enhanced photocatalytic performance for CO₂ reduction, reaching up to an exciting CO yield of 120.2 ± 4.9 µmol g⁻¹ h⁻¹ in the absence of any organic sacrificial agents, which is about 10 and 7 times higher than that of pure SrTiO₃ and CsPbBr₃ NSs, respectively. This work provides a simple and versatile strategy for the construction of HP based heterostructures with speedy charge separation to achieve efficient photocatalytic CO₂ reduction.

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.

This work was financially supported by the Natural Science Foundation of Tianjin City (No. 17JCJQJC43800), the National Key R&D Program of China (No. 2017YFA0700104), NSFC (Nos. 21931007, U21A20286), and the 111 Project of China (No. D17003).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.07.025.