English

• Solar-driven CO₂ reduction now is a promising technology to solve the environmental problem and energy crisis [1,2]. The carbon-contained products were various, and among them, carbon monoxide (CO) is a momentous constituent for bulk chemical manufacturing. Therefore, the photocatalytic CO₂-to-CO conversion with high activity and selectivity is highly sought after. However, the dissociation energy of the C═O bond in CO₂ molecules is as high as 750 kJ/mol, which makes the CO₂ photoreduction an arduous reaction [3,4]. Up to now, the reported CO yields and selectivity in CO₂ photoreduction reaction were relatively low [5,6], which greatly restricted the application of solar-driven CO₂ reduction technology. Seeking a high-performance photocatalyst for CO₂-to-CO directional conversion is the hotspot and difficulty in the photocatalytic field.

  Recently, bismuth-based materials, such as Bi₂O₃ [7,8], BiOBr [9,10], BiOI₃ [11,12], and BiOCl [13,14], have been proved to be promising photocatalysts for their unique electronic structures and layered crystal structures. Among them, BiOCl has received widespread attention [15–17]. However, its solar-to-fuel efficiency is restricted greatly by its wide band gap. Various strategies have been tried to improve the photocatalytic activity of BiOCl, such as exposed facets regulation [18–20], heterojunction configuration [21,22], (non)metal doping [23,24], and so on. Recently, studies have suggested that introducing oxygen vacancies in heterojunction structure can efficiently promote the photocatalytic performance [25–27]. Nevertheless, despite considerable efforts devoted to improving the performances of BiOCl-based heterojunction photocatalysts, the intrinsic mechanisms currently remain ambiguous.

  In CO₂ photoreduction reaction, the ultimate photocatalytic performance is a result of multiple factors working together, such as the photocarrier-separation efficiency, the adsorption type of CO₂ molecules, the activation energy of CO₂ molecules, and the ^*CO intermediates formation energy. Studies have shown that affected by the reaction environment, the surface structure and interface structure of photocatalysts were dynamically changing [28,29], which caused the changes in surface atomic configuration, electronic structure, and spatial distribution of active sites. For instance, by the selective photo-deposition of PtO_x species on BiOIO₃ surface, under light irradiation, the surface adsorbed CO₂ molecules transformed into ^*CO intermediates, leading to the emergence of Bi³⁺ species and the expansion of crystalline structure of BiOIO₃ nanocrystals [12]. Furthermore, the Pt₁-ZnIn₂S₄@CoO_x system underwent a hole-induced oxidative self-reconstruction of CoO_x, which continuously scavenged photogenerated holes and effectively suppressed photocorrosion. Consequently, it achieved hydrogen evolution rates of 23.88 mmol g⁻¹ h⁻¹ in natural seawater and 48.99 mmol g⁻¹ h⁻¹ in deionized water [30]. Therefore, investigating the dynamic evolutions of crystal structure of BiOCl-based materials in CO₂ photoreduction process and its fluence on the photocatalytic activity is a reasonable way to explore the essence of CO₂ photoreduction reaction, which is also beneficial for the rational designs of efficient CO₂ reduction photocatalysts.

  Herein, by adopting quasi in-situ characterization techniques, we explored the dynamic structure and chemical-state variations on BiO_1-xCl/C₃N₅ in CO₂ photoreduction process with BiO_1-xCl and BiOCl/C₃N₅ counterparts for comparison. The related studies clearly revealed that in BiO_1-XCl/C₃N₅, electron offset could efficiently be promoted from interfacial N atoms to Bi atoms, inducing intimate interface contact and strengthening the adsorption of CO₂ molecules. It resulted in the low oxidation state Bi atoms and the local contraction of BiO_1-xCl structures. Under light irradiation, the interface interaction was further enhanced, which promoted the activation of CO₂ molecule and the interfacial electron transport. Owing to the photoinduced electrons enrichment on interface Bi sites, the adsorbed CO₂ molecules were transformed into ^*CO intermediates with the expansion of crystalline structures. As expected, BiO_1-xCl/C₃N₅ exhibited the highest CO₂-to-CO conversion performance in comparison to BiO_1-xCl and BiOCl/C₃N₅. These findings might provide a feasible strategy to elucidate the dynamic evolution process of catalytically active sites for the significantly enhanced CO₂-to-CO directional conversion, which offers new insights for the fabrication of highly efficient CO₂ photo-reduction catalysts.

  BiO_1-xCl was prepared via a simple hydrothermal process [31]. Bi(NO₃)₃·5H₂O (1 g) was first dissolved in 25 mL of ethylene glycol and stirred for 30 min to ensure the complete dissolution. After that, 5 mL of saturated NaCl solution was added. After stirring for another 60 min, the mixture was transferred into a 100 mL hydrothermal autoclave and kept at 160 ℃ for 5 h. The BiO_1-xCl product was collected after washing, centrifugation, and freeze-dried treatment.

  5-Amino-1,2,4-triazole (5 g) and NH₄Cl (5 g) were milled to form an inhomogeneous solid mixture and then was treated at 550 ℃ for 3 h with a heating rate of 10 ℃/min. The product was naturally cooled and the obtained sample was bulk C₃N₅.

  Bulk C₃N₅ (1 g) was processed into 40 mL of concentrated H₂SO₄ with continuous stirring for 12 h. The yellowish product was washed to neutral by deionized water and then dried under vacuum at 60 ℃ for 12 h.

  The synthesis procedure of BiO_1-xCl/C₃N₅ was similar to that of BiO_1-xCl, except a certain amount of prepared C₃N₅ nanosheets was injected into the mixture solution of Bi(NO₃)₃·5H₂O and ethylene glycol.

  Sample morphology was measured by transmission electron microscopy (TEM, FEI Tecnai TF20). XRD pattern was collected by X'Pert PRO, PANalytical with a scanning rate of 0.06°/s. The chemical valence and elemental composition of the catalyst surface were determined by ESCALAB 250 Xi X-ray photoelectron spectroscopy. UV–vis diffuse reflectance spectra were characterized on a UV-2550 (Shimadzu) spectrometer. Both steady and time-resolved Photoluminescence (PL) spectra were recorded on Fluorescence Spectrophotometer F-7000 and HORIBA FL-3, respectively. EPR measurements were recorded using Bruker EMX nano.

  The synchronous irradiation X-ray diffraction (SI-XRD) was performed using a 300 W xenon lamp as the light source to record XRD data before and after CO₂ adsorption as well as light irradiation. The CO₂-preadsorption process was presented in Fig. S1 (Supporting information). Specially, after being made into thin sheets with suitable size, the photocatalyst was placed in a sealed adsorption chamber and purged with Ar (50 mL/min) for 30 min to remove the surface adsorbed impurities. Subsequently, a mixture gas of CO₂ and water vapor was introduced into adsorption chamber for 60 min. Finally, the import valve and export valve of sealed chamber were closed and kept it for 60 min at room temperature to reach the adsorption equilibrium.

  In-situ diffuse reflectance Fourier transform infrared spectroscopy (SI-DRIFTS) was carried out on a Bruker VERTEX 70 infrared spectrometer with a 300 W Xe lamp (PLS-SXE300, Perfectlight Technology Co, Ltd., Beijing) as light source. A mixed atmosphere (CO₂ + H₂O vapor) was injected into the reactor for 30 min at room temperature. After reaching the sorption equilibrium, SI-DRIFTS data were recorded under alternating dark and light conditions.

  Synchronous irradiation X-ray photoelectron spectroscopy (SI-XPS): The synchronous irradiation X-ray photoelectron spectroscopy (SI-XPS) instrument was equipped with a 300 W Xe lamp as light source. Before SI-XPS measurements, after being made into thin sheets with suitable size, the photocatalyst was placed in a sealed adsorption chamber and purged with Ar (50 mL/min) at 120 ℃ for 30 min to remove the surface adsorbed impurities. Subsequently, a CO₂/H₂O (vapor) was purged into the chamber at room temperature for 20 min to reach the adsorption equilibrium. Finally, the as-prepared samples were quickly transferred to the analysis chamber of XPS instrument (ESCALAB 250Xi) for the SI-XPS measurements. The variations of XPS spectra were recorded by controlling light irradiation (on or off). The CO₂-preadsorption process was presented in Fig. S1.

  The photocatalytic CO₂ reduction activity was carried out under the irradiation of 300 W Xe lamp (PLS-SXE300, Perfectlight Technology Co, Ltd., Beijing) equipped with a 420 nm cutoff filter (Perfect Light, Microsolar 300, 420–780 nm). In a typical reaction, 100 mg samples were dispersed uniformly in pure water. The hybrid suspension was then transferred into the tailor-made vessel. Before irradiation, ultrahigh purity CO₂ (99.999%) gas was bubbled into the suspension at 300 mL/min for 30 min to remove air and saturate the solution. The reactor temperature was kept at 25 ℃ with a cooling water circulator. After 5 h magnetic stirring under light irradiation, the suspension was centrifuged to collect the catalyst. The products were analyzed using Agilent GC 6890 N gas chromatograph with thermal conductivity and flame ionization detectors.

  For comparison, the CO₂ photoreduction performance with sacrificial reagent MeCN addition was also carried out. The testing procedure was similar with MeCN solution replacing pure water.

  In isotopic labeling experiment, ¹³CO₂ gas was employed instead of ¹²CO₂ gas as the carbon source, and H₂¹⁸O was utilized instead of H₂¹⁶O. The gas products were measured by GC–MS spectrometry.

  The photoelectrochemical data were all measured by a three-electrode photoelectrochemical cell with sample photoelectrode as working electrode, Pt sheet as counter electrode, and Ag/AgCl as reference electrode. Additionally, the electrolyte solution and light source were 0.2 mol/L Na₂SO₄ (pH 6.8) solution and 300 W xenon lamp, respectively. The working electrode was prepared as following: Catalyst powder (5 mg) was dispersed into 250 μL of isopropanol, 750 μL of H₂O and 20 μL 5% Nafion, and then sonicated for 30 min. 50 μL of the sample mixture was drop-coated onto FTO glass and dried at 60 ℃ to obtain working electrode.

  All DFT calculations data was obtained by using the Quantum Espresso code [32]. The exchange-correlation energy was approximated by the generalized gradient approximation with the Perdew-Burke-Ernzerhof (PBE) functional. A 500 eV plane-wave kinetic energy cutoff was chosen. Geometry optimization was conducted with the Brillouin zone. The atomic positions were fully relaxed until a maximum energy difference and residual force on atoms, respectively. The vacuum slab of 15 Å was added to prevent the periodic interference of the unit cell.

  Fig. 1a illustrated the basic synthetic procedure for in-situ growth of oxygen vacancies BiOCl on C₃N₅ nanosheets (BiO_1-xCl/C₃N₅). As a new type of N-rich carbon nitride, C₃N₅ was chosen for its azo-linked structure, which has been proven to be ideal support to metal atoms. Additionally, it exhibited a enhanced visible-light response to provide photo-generated charge carriers [33]. Briefly, C₃N₅ nanosheets have been obtained by previously reported method [34], which were subsequently used as the in-situ growth substrate for BiO_1-xCl nanoparticles. The XRD pattern of C₃N₅ presented a main characteristic peak at 27° (Fig. 1b) [33]. The absence of a (100) peak, a specific feature of in-plane packing, suggests the distortion of C₃N₅ framework and the broadening of nanochannel distance between heptazine units due to azo (–N═N–) bridging linkage [33,35,36]. Also, both BiOCl and BiO_1-xCl were the tetragonal structure (JCPDS No 06-0249) (Fig. 1b and Fig. S2 in Supporting information). Notably, the oxygen vacancies treatment did not damage the crystalline structure of BiOCl. In BiO_1-xCl/C₃N₅ heterojunction, the XRD peaks of BiO_1-xCl and C₃N₅ were clearly observed. As shown in Fig. 1, Fig. 1, BiO_1-XCl nanosheets were junctionalized with C₃N₅, confirming the formation of heterojunction structure. The spacing value of 0.275 nm was assigned to the BiO_1-xCl (110) plane (Fig. 1e), which showed an obvious boundary line with C₃N₅ (Fig. 1e). The uniformly distributed C, O, Bi, Cl, and N elements were the directional evidence for the tight heterojunction interface between BiO_1-xCl and C₃N₅ (Fig. 1f).

  Figure 1

  

  Figure 1.  (a) The synthetic procedure of BiO_1-xCl/C₃N₅. (b) XRD, (c-e) TEM, and HRTEM of BiO_1-xCl/C₃N₅. (f) Elemental mapping images of BiO_1-xCl/C₃N₅. (g) Electron paramagnetic resonance spectra (EPR) of BiOCl, BiO_1-xCl, and BiO_1-xCl/C₃N₅. (h) Raman spectra of BiOCl, BiO_1-xCl, and BiO_1-xCl/C₃N₅.

  DownLoad: Full-Size Img PowerPoint

  The defective nature of BiO_1-xCl/C₃N₅ was verified via EPR spectra (Fig. 1g). Usually, the typical peak of vacancies at g = 2.004 was the signal for oxygen vacancies [37,38]. In comparison of BiOCl, BiO_1-xCl showed a stronger signal, indicating the successful introduction of oxygen vacancies [13]. Interestingly, the intensity of BiO_1-xCl/C₃N₅ showed no obvious different to that of BiO_1-xCl, probably due to the absence of the authentic bonding effect between BiO_1-xCl and C₃N₅ components. The result of Raman spectra further proved the existence of oxygen vacancies in BiO_1-xCl/C₃N₅. As shown in Fig. 1h, BiOCl showed the characteristic peaks at 61, 143, 198, and 397 cm⁻¹, which were assigned to active modes of A_1g, A_1g, B_1g, and 2Eg, respectively [39]. These peaks of BiO_1-xCl were much weaker and slightly shifted to lower wavenumbers direction, suggesting the existence of abundant oxygen vacancies and metallic Bi [40]. The combination of C₃N₅ component did not change the oxygen vacancy concentration for the no obvious change peak intensities in BiO_1-xCl/C₃N₅.

  The XPS survey spectra of BiO_1-xCl/C₃N₅ consisted of Bi, O, Cl, C, and N elements (Fig. S3 and Table S1 in Supporting information), suggesting the heterojunction structure of BiO_1-xCl/C₃N₅. Fig. 2a presented the O 1s spectra. For BiO_1-XCl, the peak at 529.8 eV was the crystal lattice O of [Bi₂O₂]²⁺ crystal structure [41,42], and the peak at 531.9 eV originated from the oxygen vacancies [43,44]. The ~0.14 eV binding energy shifting between BiO_1-xCl and BiO_1-xCl/C₃N₅ was from the interface interaction in BiO_1-xCl/C₃N₅. Similar conditions were also be detected on C 1s and Cl 2p spectra (Figs. S4 and S5 in Supporting information). Referring to Fig. 2b, two peaks at 158.9 and 164.4 eV in BiO_1-xCl corresponded to Bi³⁺ 4f_5/2 and 4f_7/2, respectively. While in BiO_1-xCl/C₃N₅, both peaks shifted down by ~0.16 eV for the increase in electron density of Bi atoms [45]. Correspondingly, the N 1s exhibited an opposite change. In BiO_1-xCl, the peaks of two-coordinated (N2C) and three-coordinated (N3C) were detected at 398.5 and 399.8 eV, respectively (Fig. 2c) [33]. After constructing C₃N₅ component, these peaks shifted up by ~0.19 eV, indicating that the decrease in electron density in N nucleus. Thus, in BiO_1-xCl/C₃N₅, the electrons shifted from the N atoms to Bi atoms. The similar situation was not detected in BiOCl/C₃N₅ counterpoint (Figs. S6 and S7 in Supporting information). Nevertheless, no peaks belong to Bi-N bond was detected in both Bi 4f (163.4 eV) and N 1s (398.18 eV) XPS spectra [46], which suggested in BiO_1-xCl/C₃N₅ heterojunction, BiO_1-xCl and C₃N₅ was connected by the strong electrostatic effect [47,48]. In addition, with the introduction of high concentration oxygen vacancies, the band gap (E_g) of BiO_1-xCl was narrowed to 2.81 eV from 3.44 eV of BiOCl, a distinctive feature of defect-containing semiconductors (Fig. 2d and Fig. S8 in Supporting information) [49]. Similarly, the E_g of C₃N₅ was 2.19 eV. Moreover, the valence band potentials (E_VB) of C₃N₅ and BiO_1-xCl were determined by VB-XPS with light illumination (SI-VB XPS). The E_VB of C₃N₅ and BiO_1-xCl were 1.31 and 2.16 eV, respectively (Fig. 2e). Meanwhile, the corresponding the E_CB of C₃N₅ and BiO_1-xCl were −0.88 and −0.65 eV, respectively. Based on the above analysis, the band structures of BiO_1-xCl and C₃N₅ were depicted in Fig. 2f. It can be observed that BiO_1-xCl/C₃N₅ met simultaneously the thermodynamic conditions for CO₂ reduction and H₂O oxidization.

  Figure 2

  

  Figure 2.  O 1s (a), Bi 4f (b), and N 1s (c) XPS spectra of BiO_1-xCl/C₃N₅. (d) UV–vis DRS spectra of C₃N₅, BiOCl, BiO_1-XCl, and BiO_1-xCl/C₃N₅. (e) Synergism illumination VB XPS (SI-VB XPS) spectra of C₃N₅ and BiO_1-xCl. (f) Band structure alignments of BiO_1-xCl/C₃N₅.

  DownLoad: Full-Size Img PowerPoint

  Meanwhile, the weight ratios of BiO_1-xCl and C₃N₅ in BiO_1-xCl/C₃N₅ heterojunction had great influence on the photocatalytic CO₂ conversion efficiency and CO selectivity. BiO_1-xCl/C₃N₅ with 20 wt% BiO_1-xCl loading showed the highest CO production rate (Fig. S9 in Supporting information) and selectivity (Fig. 3a). The CO production rate of BiO_1-xCl/C₃N₅ was 207.3 μmol g⁻¹ h⁻¹ with a 96.1% selectivity (Fig. 3a), which was a 5.4-fold enhancement compared to that of BiOCl/C₃N₅. However, there was still a small amount of CO produced under the Ar atmosphere (Fig. S10 in Supporting information), which may be caused by the self-decomposition of C₃N₅ or added surfactants [50,51]. With sacrificial reagent addition, the CO evolution rate reached 1096.32 μmol g⁻¹ h⁻¹ (Fig. S11 in Supporting information). Furthermore, the obtained photocatalytic performance of BiO_1-xCl/C₃N₅ was among the highest of reported works of Bi-based photocatalysis on CO₂ photoreduction to date (Fig. 3b and Table S2 in Supporting information). Meanwhile, BiO_1-xCl/C₃N₅ also exhibited excellent photocatalytic stability and structure stability (Fig. 3c, Figs. S12 and S13 in Supporting information)

  Figure 3

  

  Figure 3.  (a) The photocatalytic CO₂ reduction performance of BiO_1-xCl/C₃N₅. (b) Comparison of photocatalytic performance in CO₂ reduction of BiOCl-based photocatalysts. (c) The photostability of BiO_1-xCl/C₃N₅. (d) ¹²C isotope-labelled experiment result of BiO_1-xCl/C₃N₅. I-t curve (e), EIS plot (f), PL (g), and time-resolved PL (h) of BiO_1-xCl/C₃N₅.

  DownLoad: Full-Size Img PowerPoint

  Furthermore, the isotopic labeling experiment was conducted by using ¹³CO₂ as a carbon source. As shown in Fig. 3d, the ¹³CO peak at m/z = 29 and ¹³COOH peak at m/z = 46 could be evidently detected in BiO_1-xCl/C₃N₅ reaction system, confirming that CO should be produced from photocatalytic CO₂ reduction instead of impurities or pollutants [52–54]. Additionally, the oxidation products have also been explored through the H₂¹⁸O isotopic labeling experiment. As illustrated in Fig. S14 (Supporting information), ¹⁸O (m/z = 18) and ¹⁸O₂ (m/z = 36) were simultaneously detected, indicating that the O₂ molecular product should be derived from water oxidation, which simultaneously provided electron/proton for CO₂ reduction into CO.

  To investigate the origins of the photocatalytic performance, the electron dynamic behaviors were studied first. As shown in Fig. 3e, BiO_1-xCl/C₃N₅ displayed the highest photocurrent density compared to BiO_1-xCl and BiOCl/C₃N₅, a result of the quicker charge transfer and separation rate after constructing the heterojunction. Meanwhile, similar conclusions were obtained from the electrochemical impedance spectroscopy (EIS). BiO_1-xCl/C₃N₅ performed with the minimum radius (Fig. 3f), presenting that the fastest charge transfer rate and the lowest recombination ability of BiO_1-xCl/C₃N₅ [55–57]. This was corroborated by the reduced photoluminescence (PL) intensity (Fig. 3g), where BiO_1-xCl/C₃N₅ displayed significantly quenched signals compared to BiO_1-xCl and BiOCl/C₃N₅, highlighting suppressed electron-hole recombination [58–60]. To probe the specific charge transfer dynamics, the TRPL spectra were collected (Fig. 3h and Table S3 in Supporting information). The lifetime decay curve was well fitted by a biexponential kinetic function (R² > 0.98). The average carrier lifetime (τ_avg) of BiO_1-xCl/C₃N₅ was 6.55 ns, which was much longer than these of BiO_1-xCl (2.92 ns) and BiOCl/C₃N₅ (5.01 ns), indicating the lower recombination ratios during the charge carrier transfer process under the excitation state [61,62]. Note that the BiO_1-xCl/C₃N₅ showed a longer lifetime than BiOCl/C₃N₅, which maybe from the strong interface interaction, a main reason for the enhanced CO₂ photoreduction performance [63,64].

  The effects of CO₂ absorption under darkness and light excitation on the structural evolution of BiO_1-xCl/C₃N₅ heterojunction were explored by synchronous irradiation X-ray diffraction (SI-XRD) technique [65]. As depicted in Fig. 4a, the CO₂ adsorption on BiO_1-xCl/C₃N₅ surface induced the evident shift of diffraction peaks toward the low D-spacing values, indicating that the shrinkage of crystalline structure owing to the electron deviated from BiO_1-xCl/C₃N₅ to CO₂ molecules. Similar phenomena could also be observed in BiOCl/C₃N₅ (Fig. 4b) and BiO_1-xCl (Fig. 4c), respectively. Moreover, under light irradiation, the diffraction peaks of BiO_1-xCl/C₃N₅ shifted to the high D-spacing direction, indicating that the formed intermediates induced the changes of crystalline structure in reverse. Similar phenomenon could be observed in the BiOCl/C₃N₅ with a smaller shift amplitude. However, no evident change could be observed on BiO_1-XCl under light irradiation, revealing its relatively poor capability for the formation of reaction intermediates.

  Figure 4

  

  Figure 4.  The effect of CO₂ adsorption and light illimitation on the D-spacing evolution of BiO_1-xCl/C₃N₅ (a), BiOCl/C₃N₅ (b), and BiO_1-xCl (c). Synchronous irradiation diffuse reflectance Fourier transform infrared spectroscopy (SI-DRIFTS) BiO_1-xCl/C₃N₅ (d), BiOCl/C₃N₅ (e), and BiO_1-xCl (f).

  DownLoad: Full-Size Img PowerPoint

  The reaction intermediates during CO₂ photoreduction were determined by synchronous irradiation diffuse reflectance Fourier transform infrared spectroscopy (SI-DRIFTS). Under the darkness, compared to BiO_1-xCl and BiOCl/C₃N₅, BiO_1-xCl/C₃N₅ presented a much higher CO₂ adsorption peak (~2300–2400 cm⁻¹) and H₂O adsorption peak (~1600–1700 cm⁻¹) (Fig. S15 in Supporting information), which revealed that the surface of BiO_1-xCl/C₃N₅ was more favorable for the adsorption of CO₂ molecules [66]. Upon light irradiation on BiO_1-xCl/C₃N₅, the peaks at 1504, 1531, and 1267 cm⁻¹ assigned to ^*CO₂ intermediates could be evidently detected in Fig. 4d. Interestingly, the emergence of ^*COOH at 1698 cm⁻¹ indicated the protonation of ^*CO₂ [67,68], which was the key for CO₂-to-CO conversion [69,70]. With increasing irradiation time, the peaks of ^*CO intermediates could be detected evidently. By comparison, the ^*CO peaks on BiO_1-xCl and BiOCl/C₃N₅ were much smaller, which confirmed the low CO production efficiencies of BiO_1-xCl and BiOCl/C₃N₅ (Fig. 4e and f).

  Moreover, the dynamic evolutions of chemical bonds and electronic structures of BiO_1-xCl/C₃N₅ during CO₂ reduction process were studied via synchronous irradiation X-ray photoelectron spectroscopy (SI-XPS) (Fig. 5 and Table S4 in Supporting information) [38,71,72]. After CO₂ absorption (Fig. 5a), the peaks at 289.3 and 286.2 eV in C 1s XPS spectrum can be indexed to ^*CO₂ and ^*CO intermediates originating from the adsorption and dissociation of CO₂ molecules [12]. Additionally, the slight peak at 290.8 eV indexed to carbonate species (CO₃²⁻), which was consistent with the DRIFTS results. Simultaneously, two evident peaks at 166.7 and 166.0 eV in Bi 4f spectrum should be attributed to ^*CO and ^*CO₂ species on exposed Bi active sites (Fig. 5b). Under light irradiation, the intensities of CO₃²⁻ and ^*CO₂ peaks decreased evidently in C 1s region, while ^*CO peak increased obviously, indicating the dissociation of CO₂ molecules. Simultaneously, an evident decrease in Bi-CO₂^* peak accompanying with increase in Bi-CO^* peak was detected in Bi 4f spectrum. As illustrated in Fig. 5c, the O 1s peaks can be fitted into six peaks at 534.7, 533.3, 532.2, 531.9, 531.5, and 530.4 eV, which were assigned to H₂O, O═C, OH, O_v, O–C, and lattice O, respectively. Upon light irradiation, the intensities of H₂O and O═C peaks decreased evidently, while OH and O–C peaks increased obviously, revealing the dissociation of CO₂ and H₂O molecules over the BiO_1-xCl/C₃N₅ surface. More importantly, all these chemical states and bond evolutions were dynamically reversible by controlling the light or dark conditions (Figs. 5d–f). Furthermore, Ar ion etching has been performed to remove the adsorbed CO₂ molecules, and all the related changes have been disappeared (Figs. 5g–i, Fig. S16 and Table S5 in Supporting information), further confirming the effects of adsorption and dissociation of CO₂ on the above chemical states and bond evolutions. For comparison, the related experiments of CO₂ adsorption have also been performed on BiO_1-xCl and BiOCl/C₃N₅, respectively. As illustrated in Figs. S17 and S18 (Supporting information), both BiO_1-xCl and BiOCl/C₃N₅ exhibited the relatively low CO₂ adsorption peaks compared with BiO_1-xCl/C₃N₅, while no significant bond evolution could be detected, demonstrating the relatively poor CO₂ reduction capability. Therefore, the combination of C₃N₅ and BiO_1-xCl could effectively promote the adsorption of CO₂ molecules owing to the electron enrichment on Bi active sites. Under light illumination, the adsorbed ^*CO₂ molecules were transformed into ^*CO intermediates with the transformation of high oxidation state Bi atoms to Bi³⁺.

  Figure 5

  

  Figure 5.  High-resolution SI-XPS spectra of C 1s (a), Bi 4f (b), and O 1s (c) on BiO_1-xCl/C₃N₅ with CO₂ adsorption under darkness and light irradiation. Effect of light illumination on the high-resolution SI-XPS spectra of C 1s (d), Bi 4f (e), and O 1s (f) of BiO_1-xCl/C₃N₅ with CO₂ adsorption. C 1s (g), Bi 4f (h), and O 1s (i) of XPS spectra BiO_1-xCl/C₃N₅ after Ar ion etching. SPV mapping of the surface potential distribution of BiO_1-xCl/C₃N₅ (j), BiOCl/C₃N₅ (k), and BiO_1-xCl (l).

  DownLoad: Full-Size Img PowerPoint

  Furthermore, the surface charge distribution has been investigated by surface photovoltage (SPV) measurements. Fig. 5j confirmed that the light excitation induced the significant variations of surface potential, and the surface potential of BiO_1-xCl/C₃N₅ was >138 mV, which was much higher than these of BiOCl/C₃N₅ (~74 mV) and BiO_1-xCl (~32.5 mV) (Fig. 5k and l). Therefore, BiO_1-xCl/C₃N₅ effectively promoted the charge separation, photoinduced electron migration, and CO₂ adsorption.

  The radicals during CO₂ photoreduction are also the keys to unveil the reaction mechanism, which were investigated via the in-situ EPR spectra. As shown in Figs. 6a and b, the drastically reduced TEMPO signal and the increased DMPO–OH signal demonstrated that BiO_1-xCl/C₃N₅ produced more photogenerated electrons and promoted more H₂O molecules dissociation under light illumination. When CO₂ was continually injected into BiO_1-xCl/C₃N₅ system, the TEMPO signal showed no obvious changes during irradiation (Fig. 6c), suggesting that the photoinduced electrons were largely captured by CO₂ during photoreduction [73]. Also, even though CO₂ was injected, no any DMPO–OH signal was detected in BiO_1-xCl/C₃N₅ system under dark, while switching to light illumination, DMPO–OH signal appeared and was higher with the illumination time prolonging (Fig. 6d), revealing the indispensable role of light in promoting water hydrolysis. Consequently, the CO₂ reduction process was promoted by the protons and photoinduced electrons (CO₂ + 2e⁻ + 2H⁺ → CO + H₂O). As a result, the ample photoinduced electrons and protons endowed BiO_1-xCl/C₃N₅ favorable CO₂ photoreduction performance.

  Figure 6

  

  Figure 6.  In-situ EPR patterns of TEMPO (a) and DMPO—OH (b) before and after light illumination on BiO_1-xCl/C₃N₅. In-situ EPR patterns of TEMPO (c) and DMPO—OH (d) before and after light illumination on BiO_1-xCl/C₃N₅ with CO₂ injection. (e) Structure models of BiO_1-xCl/C₃N₅ and BiOCl/C₃N₅ after optimization. (f) The optimized structure models of BiO_1-xCl/C₃N₅ and BiOCl/C₃N₅ after CO₂ adsorption. The adsorption energies of ^*CO (g) and ^*CO₂ (h) intermediates on BiO_1-xCl/C₃N₅ surface. Standard Gibbs free energy profile of H₂O molecular dissociation (i) and CO₂ molecular reduction (j) on BiO_1-xCl/C₃N₅ surface.

  DownLoad: Full-Size Img PowerPoint

  Temperature-programmed desorption (TPD) experiments were performed to study the interactions between BiO_1-xCl/C₃N₅ and the reaction gases CO₂ and the product CO. The maximum CO₂ desorption temperature of BiO_1-xCl/C₃N₅ was 400.8 ℃, which was higher than these of BiO_1-xCl (346.1 ℃) and BiOCl/C₃N₅ (323.7 ℃), demonstrating the stronger interaction between CO₂ molecules and BiO_1-xCl/C₃N₅ (Fig. S19 in Supporting information). In contrast, BiO_1-xCl/C₃N₅ exhibited lower maximum CO desorption temperature (316.2 ℃) compared to BiO_1-xCl (350.2 ℃) and BiOCl/C₃N₅ (328.7 ℃) (Fig. S20 in Supporting information), which illustrated that the CO molecules were more easier to release from BiO_1-xCl/C₃N₅ [16,74]. Consequently, the CO₂-to-CO conversion was enhanced in BiO_1-xCl/C₃N₅ system.

  The interface interaction of BiO_1-xCl/C₃N₅ was studied theoretically with an electron introduction. As depicted in Fig. 6e, compared to BiOCl/C₃N₅, the C₃N₅ component on BiO_1-xCl/C₃N₅ deformed obviously with a deformation energy of −2.26 eV, almost twice that of BiOCl/C₃N₅ (−1.19 eV). The obvious deformation energy difference may originate from the amplified interfacial interaction. Additionally, as a result of the stronger interfacial interaction, the distance between BiO_1-xCl and C₃N₅ (2.31 Å) was shorter in comparison to that in BiOCl/C₃N₅ (2.79 Å), and the binding energy E_b in BiO_1-xCl/C₃N₅ (−4.03 eV) was much smaller than that in BiOCl/C₃N₅ (−2.91 eV), showing the stronger interfacial interaction between C₃N₅ and BiO_1-xCl (Fig. S21 in Supporting information).

  Herein, the dynamic CO₂ conversion on BiO_1-xCl/C₃N₅ was simulated. To model CO₂ adsorption on BiO_1-xCl/C₃N₅, a CO₂ molecule was placed above BiO_1-xCls side. Five adsorption sites were designed (Fig. S22 in Supporting information), including Bi atom, Bi atom adjected O_v (Bi-O_v), O atom, O_v site, and Cl atom. Also, adsorption energy (E_ads) was calculated as E_ads = E_{CO2@BiO1-xCl/C3N5} − E_{BiO1-xCl/C3N5} − E_CO2, where E_{CO2@BiO1-xCl/C3N5} was the CO₂ adsorption energy on BiO_1-xCl/C₃N₅. As a result, E_{BiO1-xCl/C3N5} and E_CO2 referred to the energies of BiO_1-xCl/C₃N₅ and a CO₂ molecule, respectively. The E_ads values for Bi, Bi-O_v, O, O_v, and Cl were −0.24, −0.32, −0.02, −0.14, and 0.15 eV, respectively. The Bi-O_v site was the most suitable site towards CO₂ adsorption.

  In order to identify the dynamic evolution of CO₂ molecule on Bi-O_v site, one electron was introduced into the calculation models. As illustrated in Fig. 6f, after geometry optimization, the CO₂ molecule adsorbed on Bi-O_v site induced the deformation of BiO_1-xCl and C₃N₅ components in the interface, while there was no so obvious deformation in BiOCl/C₃N₅, implying the bigger interaction strength of CO₂ on BiO_1-xCl/C₃N₅. The CO₂ molecule adsorbed on BiOCl/C₃N₅ showed no obvious deformation with a distance of 2.81 Å on BiOCl layer. While on BiO_1-xCl/C₃N₅, the distance between CO₂ molecule and BiO_1-xCl layer was decreased to 2.53 Å. The CO₂ molecule distorted with O═C═O angle of 135.12°, and C and O atoms formed Bi–C and Bi–O bonds with Bi atoms. By introducing an electron, the C–O bond in Bi–O–C–Bi group was stretched from 1.18 Å to 1.33 Å and the O═C═O angle was reduced to 128.06°, which indicated the approaching dissociation of CO₂ molecule [16]. In summary, BiO_1-xCl/C₃N₅ enhanced the interaction between exposed Bi atoms and CO₂ molecules and then destructed the linear structure of CO₂ molecules, thereby achieving the efficient activation of CO₂ molecules.

  From the energy aspect, the adsorption energy of CO₂ molecules on BiO_1-xCl/C₃N₅ surface was bigger compared to these on BiOCl/C₃N₅ and BiO_1-xCl (Fig. 6g), revealing that oxygen vacancies in BiO_1-xCl/C₃N₅ enhanced the interaction between electrons and CO₂. While the adsorption energies (E_ad^*CO) of BiO_1-xCl/C₃N₅ was smaller than that of BiOCl/C₃N₅, which demonstrated that ^*CO intermediates on BiO_1-xCl/C₃N₅ was easier to release (Fig. 6h). Moreover, the free energy of H₂O dissociation on BiO_1-xCl/C₃N₅ was calculated to be ca. −0.52 eV, which was smaller than these of BiOCl/C₃N₅ (1.55 eV) and BiO_1-xCl (−0.23 eV) (Fig. 6i), confirming that the BiO_1-xCl/C₃N₅ accelerated H₂O dissociation and facilitated proton-feeding [75]. Besides, the ^*H acquisition ability of BiO_1-xCl/C₃N₅ provided the enough ^*H for the formation of ^*COOH intermediates, thereby enhancing the photocatalytic reduction process [76].

  Gibbs free energies of the CO₂ reduction elementary steps were performed (Fig. 6j). Generally, in CO₂-to-CO conversion process, the ^*COOH and ^*CO intermediates were the keys [77]. The ^*COOH intermediates was first formed on Bi sites via the process of ^*CO₂ + H⁺ +e⁻ → ^*COOH, which was considered as the rate-determining step for the CO₂-to-CO conversion [78]. The smallest ^*COOH formation energy on BiO_1-xCl/C₃N₅ was from the compact heterojunction structure and optimized electron states. In the second hydrogenation step, ^*COOH was converted into ^*CO specie and remained absorbed on the catalyst surface. Owing to the negative energy for ^*CO intermediates desorption, the adsorbed CO molecules were easily to form product CO rather than the products involved with more than two-electron processes (such as CH₃OH and CH₄). From the above analysis, the CO₂ reduction produced CO as the final product on BiO_1-xCl/C₃N₅.

  Based on the above demonstrations, a potential mechanism has been proposed to clarify the charge separation, electron transport, structure and bond evolutions over BiO_1-xCl/C₃N₅ during the photocatalytic CO₂ reduction. Firstly, the strong interfacial interaction enriched the electrons on Bi sites, facilitating the CO₂ molecules adsorption. Subsequently, the strong electron-attracting ability of absorbed CO₂ molecules pushed up the oxidation state of Bi atoms and the shrinkage of BiO_1-xCl/C₃N₅ crystalline structures. Upon light illumination, the photoinduced electron further strengthened the interaction between CO₂ molecule and Bi active site, which efficiently destroyed the linear structure of CO₂ molecule, facilitating ^*CO₂ to form ^*COOH intermediates. Then, ^*COOH was further reduced to ^*CO via reacting with proton and photoinduced electron. Finally, the ^*CO desorbed and formed CO gas, accompanying with the restoration of the crystalline structures of BiO_1-xCl/C₃N₅. Thus, the dynamic structure and chemical state evolutions of BiO_1-xCl/C₃N₅ during photoreaction were the main reasons for the high CO₂-to-CO conversion.

  In summary, using BiO_1-xCl/C₃N₅ as a model photocatalyst in CO₂ reduction, we revealed the strong relationship between dynamic structure evolution and photocatalytic performance. More specifically, owing to the combination of C₃N₅ and BiO_1-xCl, the interface electron enriched on Bi sites, which effectively promoted the adsorption of CO₂ molecules. Under light illumination, at the joint action of photoinduced electrons and protons, the adsorbed structure-twisted CO₂ molecules were transformed into ^*CO intermediates, leading to restoration of crystalline structure and valence state of Bi active sites. As expected, BiO_1-xCl/C₃N₅ photocatalyst exhibited excellent CO₂-to-CO conversion activity with an CO evolution rate of 207.3 μmol g⁻¹ h⁻¹. These findings might help to reveal the origin of photocatalytic performance of catalysts during CO₂ conversion and provide a new insight to design efficient and stable photocatalysts for CO₂ reduction.

  CRediT authorship contribution statement

  Zhilei Chen: Writing – original draft, Visualization, Validation, Software, Resources, Methodology, Investigation, Formal analysis, Data curation. Zhaoxia Li: Writing – original draft, Validation, Software, Resources, Methodology, Investigation, Formal analysis, Data curation. Yunjiang Yu: Writing – original draft, Validation, Resources, Methodology, Investigation, Formal analysis, Data curation. Mingdeng Xiang: Writing – original draft, Resources, Methodology, Investigation, Formal analysis. Cheng Ding: Writing – review & editing, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation. Entian Cui: Writing – original draft, Validation, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation. Jizhou Jiang: Writing – review & editing, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

  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 is supported by the National Natural Science Foundation of China (No. 62004143), the Key Project of Scientific Research Plan of Hubei Provincial Department of Education (No. D20241501), the Major Project of Natural Science Foundation of Jiangsu Universities, China (No. 23KJA150010), the Innovation Project of Engineering Research Center of Phosphorus Resources Development and Utilization of Ministry of Education (No. LCX202404), and the Major Project of Natural Science Foundation of Jiangsu Universities, China (No. 23KJA150010).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.112313.

  
  
• 1. [1]

     D. Wen, N. Wang, J. Peng, et al., J. Mater. Sci. Technol. 226 (2025) 93–108. doi: 10.1016/j.jmst.2024.12.010

  2. [2]

     H. Zhou, W. Ye, J. Jiang, et al., Carbon Lett. 34 (2024) 1569–1591. doi: 10.1007/s42823-024-00748-8

  3. [3]

     P. Huang, J. Huang, S.A. Pantovich, et al., J. Am. Chem. Soc. 140 (2018) 16042–16047. doi: 10.1021/jacs.8b10380

  4. [4]

     H. Yue, Z. Jinfeng, R. Olim, et al., Comp. Funct. Mater. 1 (2025) 20250103. doi: 10.63823/20250103

  5. [5]

     J. Tang, C. Guo, T. Wang, et al., Carbon Neutralization 3 (2024) 557–583. doi: 10.1002/cnl2.121

  6. [6]

     Z. Sun, N. Talreja, H. Tao, et al., Angew. Chem. Int. Ed. 57 (2018) 7610–7627. doi: 10.1002/anie.201710509

  7. [7]

     Y. Wang, J. Sheng, X. Zhao, et al., Chin. Chem. Lett. 34 (2023) 107967. doi: 10.1016/j.cclet.2022.107967

  8. [8]

     H. Lu, Q. Hao, T. Chen, et al., Appl. Catal. B: Environ. 237 (2018) 59–67. doi: 10.1016/j.apcatb.2018.05.069

  9. [9]

     J. Deng, D. Xu, J. Zhang, et al., J. Mater. Sci. Technol. 180 (2024) 150–159. doi: 10.1016/j.jmst.2023.04.05309

  10. [10]

      A. Helal, F.A. Harraz, A.A. Ismail, et al., Appl. Catal. B: Environ. 213 (2017) 18–27. doi: 10.1016/j.apcatb.2017.05.009

  11. [11]

      F. Huang, Y. Liu, F. Wang, et al., Sci. China Mater. 68 (2025) 1561–1569. doi: 10.1007/s40843-024-3290-9

  12. [12]

      Y. Lan, Y. Zhang, X. Huang, et al., Angew. Chem. Int. Ed. 63 (2024) e202407736. doi: 10.1002/anie.202407736

  13. [13]

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

  14. [14]

      M. Guan, C. Xiao, J. Zhang, et al., J. Am. Chem. Soc. 135 (2013) 10411–10417. doi: 10.1021/ja402956f

  15. [15]

      S. Wang, Q. Chen, T. Gao, et al., J. Mater. Sci. Technol. 215 (2025) 1–14.

  16. [16]

      Y. Shi, J. Li, C. Mao, et al., Nat. Commun. 12 (2021) 5923. doi: 10.1038/s41467-021-26219-6

  17. [17]

      T. Peng, Y. Wang, C.L. Dong, et al., Nano-Micro Lett. 17 (2025) 223. doi: 10.1007/s40820-025-01723-2

  18. [18]

      K. Xu, L. Wang, H. Feng, et al., J. Mater. Sci. Technol. 77 (2021) 217–222. doi: 10.1016/j.jmst.2020.10.008

  19. [19]

      Z.Q. Li, X.H. Chen, T. Li, et al., Appl. Surf. Sci. 615 (2023) 156283. doi: 10.1016/j.apsusc.2022.156283

  20. [20]

      C. Tan, L. Ai, L. Wang, et al., ACS Appl. Nano Mater. 6 (2023) 21216–21225. doi: 10.1021/acsanm.3c04327

  21. [21]

      Y. Wu, B. Yuan, M. Li, et al., Chem. Sci. 6 (2015) 1873–1878. doi: 10.1039/C4SC03229B

  22. [22]

      M. Guo, Z. Zhou, S. Yan, et al., Sci. Rep. 10 (2020) 18401. doi: 10.1038/s41598-020-75003-x

  23. [23]

      C. Wang, N. Liu, X. Zhao, et al., J. Mater. Sci. Technol. 164 (2023) 188–197. doi: 10.1016/j.jmst.2023.03.066

  24. [24]

      J. Liu, H. Wang, M.J. Chang, et al., Sep. Purif. Technol. 301 (2022) 121953. doi: 10.1016/j.seppur.2022.121953

  25. [25]

      K. Su, L. Zheng, M. Liu, et al., Small 20 (2024) e2405551. doi: 10.1002/smll.202405551

  26. [26]

      Y. Xin, J. Tian, X. Xiong, et al., Adv. Mater. 37 (2025) e2417589. doi: 10.1002/adma.202417589

  27. [27]

      D. Zu, Y. Ying, Q. Wei, et al., Angew. Chem. Int. Ed. 63 (2024) e202405756. doi: 10.1002/anie.202405756

  28. [28]

      D. Wen, N. Wang, J. Peng, et al., Chin. J. Catal. 69 (2025) 58–74. doi: 10.1016/S1872-2067(24)60183-X

  29. [29]

      E. Cui, Y. Lu, J. Jiang, et al., Chin. J. Catal. 59 (2024) 126–136. doi: 10.1016/S1872-2067(23)64630-3

  30. [30]

      C. Yuan, H. Yin, J. Li, et al., Nat. Commun. 16 (2025) 6607. doi: 10.1038/s41467-025-62033-0

  31. [31]

      S. Cheng, Z. Sun, K.H. Lim, et al., ACS Appl. Nano Mater. 6 (2023) 3608–3617. doi: 10.1021/acsanm.2c05364

  32. [32]

      P. Giannozzi, S. Baroni, N. Bonini, et al., J. Phys. Condens. Matter. 21 (2009) 395502. doi: 10.1088/0953-8984/21/39/395502

  33. [33]

      P. Kumar, E. Vahidzadeh, U.K. Thakur, et al., J. Am. Chem. Soc. 141 (2019) 5415–5436. doi: 10.1021/jacs.9b00144

  34. [34]

      J. Zhang, Z. Li, J. He, et al., ACS Catal. 13 (2022) 785–795. doi: 10.3390/e24060785

  35. [35]

      R.C. Pawar, S. Kang, J.H. Park, et al., Sci. Rep. 6 (2016) 31147. doi: 10.1038/srep31147$10.1016/j.apcatb.2021.120679

  36. [36]

      D. Liu, S. Chen, Y. Zhang, et al., Appl. Catal. B: Environ. 333 (2023) 122805. doi: 10.1016/j.apcatb.2023.122805

  37. [37]

      X. Zhao, K. Gao, S. Xue, et al., Chin. Chem. Lett. 36 (2025) 110309. doi: 10.1016/j.cclet.2024.110309

  38. [38]

      Y. Zhang, Z. Xu, Q. Wang, et al., Appl. Catal. B: Environ. 299 (2021) 120679. doi: 10.1016/j.apcatb.2021.120679

  39. [39]

      S. Weng, B. Chen, L. Xie, et al., J. Mater. Chem. A 1 (2013) 3068. doi: 10.1039/c2ta01004f

  40. [40]

      H. Li, J. Shi, K. Zhao, et al., Nanoscale 6 (2014) 14168–14173. doi: 10.1039/C4NR04810E

  41. [41]

      Z. Wu, M.-C. Chong, S. Zhang, et al., Sci. China Chem. 67 (2024) 1839–1864. doi: 10.1007/s11426-023-1943-5

  42. [42]

      S. Cheng, Z. Sun, K.H. Lim, et al., ACS Catal. 13 (2023) 7221–7229. doi: 10.1021/acscatal.3c00219

  43. [43]

      L. Wang, R. Wang, T. Qiu, et al., Nano Lett. 21 (2021) 10260–10266. doi: 10.1021/acs.nanolett.1c03249

  44. [44]

      X. Zhao, Y. Xia, H. Li, et al., Appl. Catal. B: Environ. 297 (2021) 120426. doi: 10.1016/j.apcatb.2021.120426

  45. [45]

      C. Liu, Y. Ren, Z. Wang, et al., J. Colloid Interface Sci. 607 (2022) 423–430. doi: 10.3390/chemosensors10100423

  46. [46]

      X. Zhang, L. Ai, C. Tan, et al., Sep. Purif. Technol. 355 (2025) 129542. doi: 10.1016/j.seppur.2024.129542

  47. [47]

      C. Zhou, Z. Jie, S. Xue, et al., Sci. Sin. Chim. 53 (2023) 1588–1599. doi: 10.1109/jiot.2022.3210378

  48. [48]

      J. Wu, Y. Chen, L. Zhang, et al., J. Ind. Eng. Chem. 129 (2024) 424–434. doi: 10.1016/j.jiec.2023.09.002

  49. [49]

      P. Makuła, M. Pacia, W. Macyk, J. Phys. Chem. Lett. 9 (2018) 6814–6817. doi: 10.1021/acs.jpclett.8b02892

  50. [50]

      Y. Zhang, D. Yao, B. Xia, et al., ACS Energy Lett. 7 (2022) 1611–1617. doi: 10.1021/acsenergylett.2c00427

  51. [51]

      J. Zhang, T. Yuan, H. Wan, et al., Sci. China Chem. 60 (2017) 1546–1553. doi: 10.1007/s11426-017-9125-y

  52. [52]

      S. Wang, J. Henzie, B. Jiang, et al., Nat. Commun. 14 (2023) 2534. doi: 10.3390/buildings13102534

  53. [53]

      J. Di, X. Zhao, C. Lian, et al., Nano Energy 61 (2019) 54–59. doi: 10.1016/j.nanoen.2019.04.029

  54. [54]

      S. Li, F. Chen, S. Chu, et al., Small 19 (2022) 2203559.

  55. [55]

      Z. Li, Y. Zhou, Y. Zhou, et al., Nat. Commun. 14 (2023) 5742. doi: 10.1038/s41467-023-41522-0

  56. [56]

      X. Liu, G. Dawson, K. Papadikis, et al., J. Ind. Eng. Chem. 145 (2025) 561–576. doi: 10.1016/j.jiec.2024.10.051

  57. [57]

      S. Li, M. Cai, C. Wang, et al., J. Mater. Sci. Technol. 123 (2022) 177–190. doi: 10.1016/j.jmst.2022.02.012

  58. [58]

      W. Cai, Y. Tanaka, X. Zhu, et al., Nano Res. 17 (2024) 7027–7038. doi: 10.1007/s12274-024-6736-9

  59. [59]

      A. Amari, H.S.S. Aljibori, Z. Algarni, et al., J. Ind. Eng. Chem. 140 (2024) 599–616. doi: 10.1016/j.jiec.2024.08.002

  60. [60]

      B. Wang, H. Chen, F. Huang, et al., Appl. Catal. B: Environ. 374 (2025) 125394. doi: 10.1016/j.apcatb.2025.125394

  61. [61]

      B. Wang, W. Zhang, G. Liu, et al., Adv. Funct. Mater. 32 (2022) 2202885. doi: 10.1002/adfm.202202885

  62. [62]

      J.I. Khan, F.H. Isikgor, E. Ugur, et al., ACS Energy Lett. 6 (2021) 4155–4164. doi: 10.1021/acsenergylett.1c01931

  63. [63]

      Z. Zhou, J. Wang, M. Reheimujiang, et al., J. Mater. Sci. Technol. 213 (2025) 241–251. doi: 10.1016/j.jmst.2024.05.080

  64. [64]

      F. Xu, K. Meng, B. Cheng, et al., Nat. Commun. 11 (2020) 4613. doi: 10.1038/s41467-020-18350-7

  65. [65]

      Y. Wang, C. Liu, Y. Ren, et al., J. Am. Chem. Soc. 144 (2022) 5335–5341. doi: 10.1021/jacs.1c11747

  66. [66]

      B. Wang, H. Chen, W. Zhang, et al., Adv. Mater. 36 (2024) e2312676. doi: 10.1002/adma.202312676

  67. [67]

      X. Sun, T. Xian, C. Sun, et al., J. Mater. Sci. Technol. 228 (2025) 256–268. doi: 10.1016/j.jmst.2024.12.039

  68. [68]

      H. Li, C. Gao, G. Yang, et al., Chin. Chem. Lett. 36 (2025) 110547. doi: 10.1016/j.cclet.2024.110547

  69. [69]

      J. Zhang, J. Fu, K. Dai, J. Mater. Sci. Technol. 116 (2022) 192–198. doi: 10.3390/drones6080192

  70. [70]

      Z. Yao, H. Cheng, Y. Xu, et al., Nat. Commun. 15 (2024) 9881. doi: 10.1038/s41467-024-53529-2

  71. [71]

      W. Yan, Y. Zhang, Y. Bi, Angew. Chem. Int. Ed. 63 (2024) e202316459. doi: 10.1002/anie.202316459

  72. [72]

      W. Weng, Z. Lin, H. Zhang, et al., JACS Au 3 (2023) 3391–3399. doi: 10.1021/jacsau.3c00554

  73. [73]

      Y. He, S. Dai, J. Sheng, et al., Proc. Natl. Acad. Sci. U. S. A. 121 (2024) e2322107121. doi: 10.1073/pnas.2322107121

  74. [74]

      Y. Shi, G. Zhan, H. Li, et al., Adv. Mater. 33 (2021) 2100143. doi: 10.1002/adma.202100143

  75. [75]

      S. Chen, Z. Zhang, W. Jiang, et al., J. Am. Chem. Soc. 144 (2022) 12807–12815. doi: 10.1021/jacs.2c03875

  76. [76]

      L. Wang, T. Yang, B. Feng, et al., Chin. J. Catal. 54 (2023) 265–277. doi: 10.1016/S1872-2067(23)64546-2

  77. [77]

      S. Hu, P. Qiao, X. Yi, et al., Angew. Chem. Int. Ed. 62 (2023) e202304585. doi: 10.1002/anie.202304585

  78. [78]

      S. Wang, J. Wang, Y. Wang, et al., ACS Catal. 14 (2024) 10760–10788. doi: 10.1021/acscatal.4c01712

• 

  Figure 1  (a) The synthetic procedure of BiO_1-xCl/C₃N₅. (b) XRD, (c-e) TEM, and HRTEM of BiO_1-xCl/C₃N₅. (f) Elemental mapping images of BiO_1-xCl/C₃N₅. (g) Electron paramagnetic resonance spectra (EPR) of BiOCl, BiO_1-xCl, and BiO_1-xCl/C₃N₅. (h) Raman spectra of BiOCl, BiO_1-xCl, and BiO_1-xCl/C₃N₅.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  O 1s (a), Bi 4f (b), and N 1s (c) XPS spectra of BiO_1-xCl/C₃N₅. (d) UV–vis DRS spectra of C₃N₅, BiOCl, BiO_1-XCl, and BiO_1-xCl/C₃N₅. (e) Synergism illumination VB XPS (SI-VB XPS) spectra of C₃N₅ and BiO_1-xCl. (f) Band structure alignments of BiO_1-xCl/C₃N₅.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) The photocatalytic CO₂ reduction performance of BiO_1-xCl/C₃N₅. (b) Comparison of photocatalytic performance in CO₂ reduction of BiOCl-based photocatalysts. (c) The photostability of BiO_1-xCl/C₃N₅. (d) ¹²C isotope-labelled experiment result of BiO_1-xCl/C₃N₅. I-t curve (e), EIS plot (f), PL (g), and time-resolved PL (h) of BiO_1-xCl/C₃N₅.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  The effect of CO₂ adsorption and light illimitation on the D-spacing evolution of BiO_1-xCl/C₃N₅ (a), BiOCl/C₃N₅ (b), and BiO_1-xCl (c). Synchronous irradiation diffuse reflectance Fourier transform infrared spectroscopy (SI-DRIFTS) BiO_1-xCl/C₃N₅ (d), BiOCl/C₃N₅ (e), and BiO_1-xCl (f).

  下载: 全尺寸图片 幻灯片
  

  Figure 5  High-resolution SI-XPS spectra of C 1s (a), Bi 4f (b), and O 1s (c) on BiO_1-xCl/C₃N₅ with CO₂ adsorption under darkness and light irradiation. Effect of light illumination on the high-resolution SI-XPS spectra of C 1s (d), Bi 4f (e), and O 1s (f) of BiO_1-xCl/C₃N₅ with CO₂ adsorption. C 1s (g), Bi 4f (h), and O 1s (i) of XPS spectra BiO_1-xCl/C₃N₅ after Ar ion etching. SPV mapping of the surface potential distribution of BiO_1-xCl/C₃N₅ (j), BiOCl/C₃N₅ (k), and BiO_1-xCl (l).

  下载: 全尺寸图片 幻灯片
  

  Figure 6  In-situ EPR patterns of TEMPO (a) and DMPO—OH (b) before and after light illumination on BiO_1-xCl/C₃N₅. In-situ EPR patterns of TEMPO (c) and DMPO—OH (d) before and after light illumination on BiO_1-xCl/C₃N₅ with CO₂ injection. (e) Structure models of BiO_1-xCl/C₃N₅ and BiOCl/C₃N₅ after optimization. (f) The optimized structure models of BiO_1-xCl/C₃N₅ and BiOCl/C₃N₅ after CO₂ adsorption. The adsorption energies of ^*CO (g) and ^*CO₂ (h) intermediates on BiO_1-xCl/C₃N₅ surface. Standard Gibbs free energy profile of H₂O molecular dissociation (i) and CO₂ molecular reduction (j) on BiO_1-xCl/C₃N₅ surface.

  下载: 全尺寸图片 幻灯片
•