Interfacial Differentiated Recoordination Trigger Asymmetric Bi (δ⁺)—Bi (δ⁻) Regions for Robust CO₂ Photosynthesis to CH₃COOH

ABSTRACT

Direct solar‐driven conversion of CO₂ and H₂O into high‐value‐added C₂ products, such as acetic acid, represents a critical frontier challenge in artificial photosynthesis. However, its efficiency is primarily constrained by sluggish photogenerated charge migration and the dynamics limitations for C‐C coupling. This paper reports a “disorder‐induced reconstruction” strategy that utilizes the inherent local disorder of covalent organic frameworks (COFs) to induce the construction of asymmetric active centers on the surface of Bi₂₄O₃₁Br₁₀ (BOB), inducing atomic‐scale reconstruction and spontaneously forming extended regions with asymmetric Bi(δ+)‐Bi(δ‐) bimetallic sites. Without the use of sacrificial agents or noble metal co‐catalysts, the resulting catalyst exhibits excellent performance in the conversion of CO₂ to acetate, with a yield as high as 1.03 mmol g⁻¹ h⁻¹ and a selectivity of 97.63%. The apparent quantum efficiencies of catalyst can reach 13.75% and 8.33% at 380 and 400 nm, respectively. This work reveals a previously unknown mechanism for reconstructing inorganic surfaces into asymmetric active structural units using local disorder in organic modifiers, providing a paradigm for the design of next‐generation photocatalysts.

This paper reports a “disorder‐induced reconstruction” strategy that utilizes the inherent local disorder of covalent organic frameworks (COFs) to induce the construction of asymmetric active centers on the surface of Bi₂₄O₃₁Br₁₀ (BOB). The aperiodic structural units in the COFs disrupt the BOB surface lattice, inducing atomic‐scale reconstruction and spontaneously forming extended regions with asymmetric Bi(δ+)‐Bi(δ‐) bimetallic sites, enabling outstanding photocatalytic CO₂ to acetate conversion.

1. Introduction

Solar‐driven photocatalytic reduction of carbon dioxide (CO₂) to produce high‐value‐added chemicals offers a potential solution to the global energy crisis and greenhouse effect [1, 2, 3]. However, CO₂ molecules are highly chemically inert, and their activation and conversion involve complex multi‐proton coupled electron transfer processes [4, 5, 6]. Among the many products, acetic acid (CH₃COOH), an important C₂ industrial chemical, has a formation pathway involving key C‐C coupling steps, which are both thermodynamically and kinetically challenging [7, 8]. Therefore, achieving highly selective CO₂‐to‐CH₃COOH conversion under mild conditions is currently a cutting‐edge focus of photocatalysis research [9, 10].

From the perspective of reaction mechanism, the efficient conversion of CO₂ to C₂ products places stringent dual functional requirements on catalyst design: first, it must possess strong activation ability to efficiently cleave C═O bonds to generate CO intermediates; second, it must possess suitable local geometry and electronic structure to stabilize CO and lower the energy barrier of C‐C coupling [11, 12, 13, 14]. A single active site often cannot simultaneously satisfy these two key steps. In recent years, constructing synergistic two‐site catalytic systems has become a core strategy for overcoming the selectivity bottleneck of C₂₊ products [15, 16, 17, 18]. However, precisely constructing such asymmetric active centers at the atomic scale through rational material design remains a significant challenge. The coupling of low‐dimensional semiconductor materials with covalent organic frameworks (COFs) offers new possibilities for solving this problem. COFs [19, 20, 21], with their highly ordered pore structure and tunable molecular environment, exhibit unique advantages in the CO₂ enrichment field; while bismuth‐based semiconductors (such as Bi₂₄O₃₁Br₁₀, BOB) possess good photoresponsiveness and charge separation capabilities [22, 23]. Traditional heterojunction design often pursues perfect lattice matching to construct an ideal interface, but may overlook the complex structural relaxation phenomena at the interface. In fact, when a flexible COF containing locally disordered regions comes into contact with a rigid crystalline semiconductor, subtle structural evolution may occur at the interface. This non‐periodic perturbation originating from the organic component may induce local lattice distortion or atomic reconstruction on the surface of the inorganic component, thereby spontaneously forming asymmetric coordination configurations that are difficult to achieve in a perfect crystal. This interface engineering strategy of “disorder‐induced reconstruction” opens up a completely new path for creating novel asymmetric active centers.

In this work, we modulated the interface formation of BOB by introducing a COF containing an intrinsically locally disordered structure into the BOB synthesis. This study shows that the localized aperiodic structure of the COF induces atomic‐scale reconstruction of the BOB crystal surface at the interface with BOB, spontaneously forming a large‐scale non‐periodic interface rich in Bi(δ+)‐Bi(δ‐) asymmetric sites. Moreover, the COF at this interface further facilitates the transport of photo‐generated electrons from the BOB to this location. As a result, the modified catalyst, without sacrificial agents or noble metal co‐catalysts, synergistically achieves highly efficient CO₂ activation and C‐C coupling, ultimately exhibiting CH₃COOH formation rate as high as 1.03 mmol g⁻¹ h⁻¹ with a selectivity of 97.63%. At wavelengths of 380 and 400 nm, its apparent quantum efficiency (AQE) reaches 13.75% and 8.33%, respectively. This study not only provides a highly efficient photocatalytic CO₂‐to‐acetic acid conversion system, but also reveals an atomic‐level mechanism for inducing inorganic crystal interface reconstruction by utilizing the local disorder of organic components, thereby creating asymmetric active sites, providing a new perspective for the design of photocatalysts.

2. Results

2.1. Formation of Localized Aperiodic Interfaces Rich in Asymmetric Sites

In this study, a covalent organic framework (COF) was introduced during the synthesis of Bi₂₄O₃₁Br₁₀ (BOB) to disrupt its original periodic arrangement structure. As shown in Figure 1A, during BOB formation, the localized organic aperiodic structure in the COF induced interface reconstruction at the interface with BOB. Under this influence, the positions of Bi atoms on the surface shifted and formed large‐scale distortions, resulting in an active interface rich in Bi(δ+)‐Bi(δ‐) asymmetric sites. By adjusting the amount of COF added, a series of samples, including 0.2COF/BOB, 0.5COF/BOB, 1COF/BOB, and 2COF/BOB, were prepared. X‐ray diffraction (XRD) patterns (Figure S1) showed that the main diffraction peaks of all composite materials were in good agreement with the standard card (JCPDS No. 75–0888) of monoclinic Bi₂₄O₃₁Br₁₀, indicating that the low amount of COF introduced did not significantly alter the crystal structure characteristics of BOB.

FIGURE 1.

(A) Illustration of the formation process of COF/BOB, (B) TEM image of COF, (C‐E) HAADF‐STEM images of 0.5COF/BOB. Synchrotron radiation XAFS spectra of BOB and 0.5COF/BOB at (F) Bi L₁‐edge and (G) Bi L₃‐edge. (H) EXAFS spectra of BOB and 0.5COF/BOB at Bi L₁‐edge. (I) Wavelet transform of 0.5COF/BOB.

Transmission electron microscopy (TEM) revealed that COF exhibited typical amorphous characteristics (Figure 1B), while BOB and 0.5COF/BOB existed as ultrathin 2D nanosheets (Figures S2 and S3). Aberration‐corrected high‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM) revealed the fine surface structure and atomic arrangement (Figure 1C–E) of 0.5COF/BOB. In the atomic resolution image, the lattice fringe spacing of BOB was 0.281 nm, corresponding to its (117) or (1‐17) crystal plane (Figure 1D), indicating that the (304̅) crystal plane was the predominantly exposed crystal plane. Meanwhile, the localized surface reconstruction distortion of BOB induced by COF led to partial Bi─O bond mismatch, and the resulting defects were shown as dark regions in the STEM. The O 1s spectrum of X‐ray photoelectron spectroscopy (XPS) (Figure S4C) shows peaks with binding energies of 529.3 eV and 530.9 eV, which are attributed to lattice oxygen and defect oxygen species near oxygen vacancies, respectively. This reflects the presence of numerous Bi‐O vacancy pairs in 0.5COF/BOB [24], consistent with STEM results. The dashed box in the magnified view (Figure 1E) reveals disordered shading of COF, consistent with its amorphous structure. The red bright spots in the pseudocolor image reflect the positions of Bi atoms. Unlike the ordered periodic arrangement in pristine BOB (Figure S5), the Bi atoms (marked by yellow circles) in the COF‐affected region exhibit significant anisotropic shifts, indicating the effective formation of a large‐scale asymmetric site region.

Then, Raman and Fourier transform infrared (FTIR) spectra reveals the interfacial coupling behavior at the vibrational mode level. In Figure S6, the signals at 60 and 108 cm⁻¹ are related to the Bi‐Br A_1g vibration in BOB. These signals at 0.5COF/BOB are shifted relative to BOB, suggesting the presence of local lattice distortion. The FTIR spectra (Figure S7) show a new set of characteristic peaks at 461.9 cm⁻¹ for 0.5COF/BOB, attributable to Bi─O bonds [25], whereas the physical mixture (COF‐BOB) shows no signal at this position. Furthermore, the original Bi–O‐related vibrations in BOB (538.8 cm⁻¹) undergo a shift upon the introduction of COF, further indicating that the formation of new interfacial bonds alters the local coordination environment around Bi atoms. These results collectively demonstrate that chemical interfacial coupling is achieved between the COF and BOB, rather than simple physical loading. To elucidate the local structural reconstruction and electronic state evolution induced at the atomic scale by interface defect engineering, synchrotron X‐ray absorption fine structure (XAFS) spectroscopy was used to analyze BOB and 0.5COF/BOB in detail. X‐ray absorption near‐edge structure (XANES) spectra of Bi L₁ and L₃ (Figure 1F,G) showed a similar trend of blue shift in the absorption edges of 0.5COF/BOB, revealing the electronic structure changes in Bi atoms induced by COF. Fourier transform extended X‐ray absorption fine structure (FT‐EXAFS) spectroscopy (Figure 1H) provided direct coordination information. In R‐space, the peak signal at approximately 1.58 Å is the characteristic Bi‐O peak of BOB. In 0.5COF/BOB, this peak signal reflects the lattice Bi─O of BOB and the bond lengths of Bi‐N/O at the BOB‐COF interface, but the Bi‐O signal still dominates. Interestingly, the peak exhibited two distinct directional shifts, indicating a slight contraction of the Bi─O bonds within the [Bi₂₄O₃₁] layer and an increase in the Bi‐O distance due to the deviated Bi atoms, consistent with local lattice distortion at the BOB‐COF interface. Furthermore, the Bi‐Bi scattering peak at approximately 2.7 Å shifted to a higher R value after recombination, indicating the sustained influence of the asymmetric structure of the locally distorted interface on multiple shells. To quantify these structural changes, Bi L₁ edge EXAFS data were quantitatively fitted (Figures S8 and S9, Table S1). The results show that the first coordination layer of BOB contains only Bi─O bonds with a coordination number of approximately 3.5 and a bond length of 2.27 Å. In contrast, for 0.5COF/BOB, the Bi─O bond length in the first coordination layer shortens to 2.24 Å (coordination number approximately 3.4), and a coordination layer with a Bi─O bond length of 2.43 Å (coordination number approximately 2.3) was also discovered. Combining the spatial resolution capability of wavelet transform (WT) contour maps (Figure 1I, Figure S10) with the above HAADF‐STEM results, it is implied that COF induces the reconstruction of the local coordination environment of Bi, forming a local aperiodic interface rich in asymmetric sites.

2.2. Investigation of Photocatalytic Reduction Performance

To comprehensively evaluate the intrinsic photocatalytic activity of the constructed heterojunctions, CO₂ reduction tests are conducted in a pure water system under simulated solar light irradiation (300 W Xe lamp) without the use of photosensitizers, sacrificial agents, or organic co‐solvents. As shown in Figure 2A, the series of COF/BOB photocatalysts exhibit activities significantly superior to those of the single components. Among them, 0.5COF/BOB demonstrates optimal catalytic performance, achieving an acetic acid production rate of 1.03 mmol g⁻¹ h⁻¹, which represents 3.39‐ and 20.40‐fold enhancements over pure BOB and pure COF, respectively. Furthermore, the physical mixture (COF‐BOB) also displayed improved catalytic activity (371.92 µmol g⁻¹ h⁻¹) compared to the individual components. Regarding product selectivity (Figure 2B, Figure S11), although trace byproducts such as C₂H₆, C₂H₄, CH₄, CO, and HCOOH are detected, the system demonstrates a remarkable selectivity of 97.63% for CH₃COOH, with a corresponding electron selectivity as high as 99.37%. Then, the origin of the carbon source was verified through ¹³CO₂ isotopic labeling experiments. In the ¹³CO₂ labeling experiment, a mass spectrometry signal at m/z = 61.02, corresponding to ¹³CH₃ ¹³COO⁻, is observed (Figure 2C). To further validate the carbon source, ¹H nuclear magnetic resonance (¹H NMR) analysis is performed on the liquid products. The ¹H NMR spectrum clearly displays the characteristic doublet splitting of methyl protons in the acetic acid product, arising from strong ¹³C‐¹H coupling [26], while this splitting is absent in the ¹²CO₂ control test (Figure S12). Combined with control experiments under an inert atmosphere, in the dark, and without a catalyst (Figure S13), these results indisputably prove that the carbon atoms in the product originate exclusively from the gaseous CO₂ feedstock. Furthermore, to further verify the C‐C coupling process, CO partial pressure‐dependent kinetic experiments were conducted (Figure S14A). The CH₃COOH production rate increased with increasing CO partial pressure in the range of 0.1‐0.6 atm, giving an apparent reaction order of approximately 0.5 with respect to CO. This positive kinetic dependence indicates that CO‐derived surface intermediates participate in the C‐C coupling process, providing additional kinetic evidence for the proposed *CO coupling pathway toward CH₃COOH. The activity decrease at excessive CO partial pressure may be associated with over‐coverage of active sites and competitive adsorption (Figure S14B). In addition, we further evaluated the solar‐to‐chemical energy conversion efficiency (STC) of 0.5COF/BOB under standard AM 1.5G illumination [27]. The measured STC efficiency of 0.5COF/BOB was 0.14% (Figure S15), indicating that this system possesses certain potential for solar chemical conversion. To elucidate the complete reaction mechanism, the oxidation half‐reaction products are investigated. The generation of H₂O₂ (635.5 µmol g⁻¹ h⁻¹) and O₂ (1695.5 µmol g⁻¹ h⁻¹) is detected in the reaction system (Figure S16), indicating that photogenerated holes effectively driven water oxidation to provide the necessary protons (H⁺) for CO₂ reduction, while also achieving value‐added oxidation products. Moreover, under monochromatic light irradiation at 380 and 400 nm, the apparent quantum efficiency (AQE) of 0.5COF/BOB can reach 13.75% and 8.33%, respectively (Figure 2D). Additionally, the 0.5COF/BOB maintains acetic acid yields of 1574.59 and 497.17 µmol g⁻¹ h⁻¹ under full‐spectrum irradiation (190–1100 nm) and even under visible light irradiation (cutoff filter 400–780 nm), respectively (Figures S17 and S18), demonstrating its potential for broad‐spectrum solar energy utilization. Besides, COF, BOB, and 0.5COF/BOB exhibits negligible activity decay after 30 h of continuous photocatalytic reaction (Figure 2E), and post‐cycling XRD patterns, XPS spectra, and FTIR spectra (Figures S19–S22) confirm the preservation of their structural integrity. Finally, this material was compared and evaluated with state‐of‐the‐art photocatalysts reported recently, and it ranks among the top performers in terms of both yield and selectivity (Figure 2F, Table S2).

FIGURE 2.

(A) Photoreduction of CO₂ into CH₃COOH over COF, BOB, COF/BOB and COF‐BOB. (The error bar represents the standard deviation of at least three independent experiments.) (B) Selectivity of CO₂ reduction to CH₃COOH by COF, BOB and COF/BOB. (C) Mass spectrum of the products after ¹³CO₂ reduction over 0.5COF/BOB. (D) Apparent quantum efficiency (AQE) of 0.5COF/BOB. (E) Cycling test over COF, BOB and 0.5COF/BOB for CO₂ reduction. (F) Comparison of CH₃COOH generation performance and selectivity.

2.3. Thermodynamic and Kinetic Analysis of Photogenerated Charges

To elucidate the band alignment at the heterojunction interface and its influence on photocatalytic performance, this study systematically characterized the electronic structures of BOB and COF by combining UV‐Vis diffuse reflectance spectroscopy (UV‐Vis DRS) and Mott–Schottky (MS) tests. The UV‐Vis DRS spectra (Figure S23) shows that the intrinsic absorption edges of pure BOB and COF are located at approximately 480 and 590 nm, respectively. Notably, the 0.5COF/BOB not only exhibits a red‐shift in the absorption edge relative to BOB but also displays significantly enhanced absorption intensity in the visible light region, indicating that the introduction of the COF effectively broadens the light response range and improve photon harvesting efficiency. Based on Tauc plots derived from the Kubelka‐Munk function (Figure S24), the optical band gaps (E_g) of BOB and COF are calculated to be 2.88 and 2.51 eV, respectively. The flat band potentials (E_fb) are determined via Mott‐Schottky tests (Figure S25). The results show that all samples exhibit positive slopes, characteristic of typical n‐type semiconductors. The E_fb of BOB and COF are −0.28 and −0.32 V vs. the normal hydrogen electrode (NHE), respectively. Considering that the conduction band minimum (E_CB) of an n‐type semiconductor is typically located approximately 0.1 V more negative than the flat band potential, the E_CB values for BOB and COF are calculated to be −0.38 and −0.42 V (vs. NHE), respectively. Combining this with the band gap relationship (E_VB = E_g + E_CB), their E_VB are further estimated to be 2.50 and 2.09 V (vs. NHE), respectively. The band structure schematic constructed based on the above band parameters (Figure S26) clearly indicates that a staggered band alignment is formed between the COF and BOB. This energy level matching not only thermodynamically satisfies the potential requirements for oxidizing H₂O to H₂O₂/O₂ and reducing CO₂ to CO, CH₄, HCOOH, and the target product CH₃COOH, but also provides ample thermodynamic overpotential. Specifically, the composite possesses a sufficiently negative conduction band position to drive the high‐energy‐barrier C‐C coupling reaction, while its relatively positive valence band position ensures that the holes possess strong oxidizing capability, thereby guaranteeing the thermodynamic feasibility for the overall photocatalytic reaction.

To directly elucidate the local charge separation and transfer behavior at the COF/BOB heterojunction interface on the nanoscale, in‐situ Kelvin probe force microscopy (in‐situ KPFM) under illumination is employed to characterize the surface contact potential difference (CPD) and its photoresponse (Figure 3A,B, Figures S27 and S28). Under dark conditions, the surface potential of 0.5COF/BOB is approximately −0.23 µV, significantly lower than that of pure BOB (−0.18 µV). This indicates the occurrence of static electron transfer from the COF to BOB at the interface, preliminarily confirming the formation of a built‐in electric field within the heterojunction. Upon illumination, the charge distribution of the system undergoes dynamic evolution. The CPD of pure BOB exhibits only a slight negative shift to approximately −0.20 µV, whereas 0.5COF/BOB displays a strong photovoltage response, with its CPD shifting further negatively to approximately −0.34 µV. This significant light‐induced potential drop indicates that photoexcitation did not merely bring the heterojunction interface to a simple steady state but rather further activated the charge transfer process. The COF acts not only as an electron donor but also as an efficient light‐harvesting unit, continuously injecting excited‐state electrons into the conduction band or interfacial defect states of BOB, resulting in a highly electron‐rich state on the BOB surface. This electron transport mechanism, facilitated by the built‐in electric field, achieved efficient spatial separation of photogenerated electron‐hole pairs. Consequently, the accumulation of electrons on the BOB surface provides a high‐density source of active electrons for the CO₂ reduction reaction, thereby explaining the superior photocatalytic performance of the 0.5COF/BOB.

FIGURE 3.

KPFM images in dark and light conditions of (A) BOB and (B) 0.5COF/BOB. Ultrafast transient absorption spectroscopy of (C) BOB and (D) 0.5COF/BOB. TA decay kinetics of (E) BOB and (F) 0.5COF/BOB.

To deeply elucidate the regulatory mechanism of interfacial asymmetric sites on the spatiotemporal evolution of photogenerated charge carriers, femtosecond transient absorption (fs‐TA) spectroscopy is employed to reveal the migration, trapping, and recombination dynamics of charges in 0.5COF/BOB (Figure 3C–F). Compared with pure BOB, the decay of the transient absorption signal for 0.5COF/BOB is significantly retarded, indicating an overall prolongation of carrier lifetimes. Fitting with a tri‐exponential decay model reveals a fundamental change in the kinetic processes of the composite material. The time constant τ₁ primarily corresponds to ultrafast processes such as hot carrier cooling after excitation, in‐band relaxation, and rapid capture by shallow traps. The increase in τ₁ from 2.91 ps for BOB to 4.65 ps for 0.5COF/BOB indicates that the introduction of COF slows down the rapid deactivation of carriers at the initial stage of excitation. The time constant τ₂ is typically attributed to the migration or trapping process of conduction band electrons to surface defects or active sites [28]. This value increased significantly from 44.99 ps for BOB to 111.95 ps for 0.5COF/BOB. This variation suggests that the asymmetric Bi sites induced by interfacial Bi‐N/O coordination can serve as deep‐level electron traps or metastable trapping centers, effectively capturing and temporarily storing photogenerated electrons to inhibit their rapid recombination. The time constant τ₃ represents the recombination lifetime of trapped electrons with valence band holes (or the lifetime of the charge‐separated state). This value surged from approximately 517 ps for BOB to 1034 ps for 0.5COF/BOB. This nanosecond‐scale lifetime extension implies that non‐radiative recombination pathways are substantially suppressed, thereby overcoming the kinetic mismatch between photophysical processes and surface chemical reactions, and securing a valuable time window for the sluggish CO₂ reduction steps. In addition, the average carrier lifetime (τ_avg) calculated based on the fitting results was also significantly extended from 459.33 ps for BOB to 782.56 ps for 0.5 COF/BOB, further quantitatively demonstrating that the COF/BOB interfacial coupling effect can effectively promote charge separation and stabilize long‐lived separated states. Wavelength‐resolved fs‐TA spectra (Figure S29) show negative ΔA signals for both BOB and 0.5COF/BOB, mainly arising from ground‐state bleaching after photoexcitation. Compared with BOB, 0.5COF/BOB exhibits a broader bleaching response extending to the longer‐wavelength region, suggesting that COF coupling promotes interfacial charge separation. To verify the universality of the aforementioned ultrafast spectroscopic results at nanosecond and macroscopic scales, we systematically validate them by combining steady‐state photoluminescence (PL), time‐resolved photoluminescence (TRPL), and photoelectrochemical characterizations. The fluorescence quenching observed in the PL spectra (Figure S30) confirms the suppression of radiative recombination, while the prolonged average fluorescence lifetime (from 2.3 to 3.0 ns) shown in TRPL spectra (Figure S31) is highly consistent with the fs‐TA results, further confirming that the formed asymmetric interface can effectively extract and stabilize photogenerated electrons [29]. From a macroscopic electrochemical perspective, 0.5COF/BOB exhibits the increased transient photocurrent response (Figure S32) and the lowest charge transfer resistance (EIS, Figure S33), intuitively reflecting the enhancement of charge separation efficiency [30, 31, 32]. Furthermore, the significantly enhanced positive signal in the surface photovoltage (SPV) spectra (Figure S34) reveal spatial separation characteristics wherein photogenerated holes migrate to the surface while photogenerated electrons accumulate at the interface. This is consistent with the aforementioned mechanism of electron trapping by interfacial Bi sites.

2.4. Exploration of the Underlying Reaction Mechanism

To elucidate the microscopic mechanism of CO₂ reduction at the COF/BOB interface, we thoroughly investigate the interfacial physicochemical properties and their role in modulating the reaction microenvironment. Water contact angle measurements (Figure 4A) reveal that 0.5COF/BOB exhibits a contact angle of 27°, lower than that of BOB (35°). This indicates enhanced hydrophilicity following heterojunction formation, which facilitates the construction of a proton‐rich microenvironment at the interface to continuously supply H⁺ for CO₂ reduction [33]. Zeta potential tests (Figure S35) demonstrate that the surfaces of all three materials are negatively charged. Notably, 0.5COF/BOB exhibits a more negative surface potential relative to BOB due to the introduction of the COF. The resulting higher negative charge density favors the enhanced electrostatic adsorption of CO₂ molecules and their partially charged intermediates, thereby promoting their enrichment and activation at the catalytic interface. Nitrogen adsorption‐desorption (BET) results (Figures S36 and S37) indicate the COF possesses the highest specific surface area (253.92 m²·g⁻¹). Upon introducing the COF, the specific surface area of 0.5COF/BOB increased significantly compared to BOB, suggesting that the porous skeleton of the COF provides the 0.5COF/BOB with more abundant accessible surface area and adsorption sites. This facilitates the local interfacial enrichment of CO₂ and reaction intermediates, thereby promoting reduction steps such as C‐C coupling. CO₂ adsorption isotherms confirm that 0.5COF/BOB possesses a higher CO₂ physical adsorption capacity (Figure 4C). The CO₂ temperature‐programmed desorption (CO₂‐TPD, Figure 4B) exhibits desorption peaks whose intensity and temperature distribution reveal the presence of more highly active chemisorption sites on the surface. Such strong interactions help overcome the inertness of CO₂ and lower its activation energy barrier. To in‐situ track the fate of photogenerated electrons, we compared the transient photocurrent responses under different atmospheres (Figure S38). Upon the introduction of CO₂, the photocurrent of 0.5COF/BOB is significantly decreased [34]. This indicates in a CO₂ atmosphere, a substantial amount of photogenerated electrons are rapidly captured by interfacially adsorbed CO₂ molecules and consumed in the reduction reaction. In summary, through the synergistic optimization of surface hydrophilicity, charge distribution, pore structure, and chemisorption properties, 0.5COF/BOB successfully establishes an ideal reaction microenvironment featuring both high‐concentration CO₂ enrichment and efficient proton/electron transport. This lays a robust foundation in terms of mass transport and kinetics for the highly selective conversion of CO₂ to CH₃COOH.

FIGURE 4.

(A) Static water contact‐angle measurement of BOB and 0.5COF/BOB. (B) CO₂‐TPD curves of BOB and 0.5COF/BOB. (C) CO₂ adsorption isotherms of BOB and 0.5COF/BOB. (D) In situ FTIR spectra showing the coadsorption of CO₂ on 0.5COF/BOB and BOB. DFT calculated diagram of crystal structure, electrostatic potential of (E) BOB and (F)COF/BOB. (G) The charge density difference of COF/BOB interface. An increase in charge density is denoted in yellow and decrease in blue. (H) Free energy diagrams of CO₂ reduction to acetate over the BOB and COF/BOB.

To deeply unravel the microscopic reaction pathways of CO₂ photoreduction, in situ Fourier transform infrared (FTIR) spectroscopy is employed to monitor the dynamic evolution of reaction intermediates on the catalyst surface in real‐time under simulated illumination (Figure 4D). With prolonged irradiation time, the intensities of the infrared absorption peaks corresponding to various characteristic intermediates gradually increased, indicating that photogenerated carriers can sustainably drive the stepwise reduction of CO₂. Taking the 0.5COF/BOB heterojunction as a representative example, the absorption peaks located at 1067 cm⁻¹ and 1163 cm⁻¹ in its in situ FTIR spectrum can be assigned to CHO* and CH₃O* species, respectively, both are key intermediates in the methylation pathway. The distinct absorption peak at 1705 cm⁻¹ corresponds to COOH* species, reflecting the initial activation of CO₂. Additionally, the characteristic peaks at 1366 and 1485 cm⁻¹ are associated with OCCO* species [35]. This moiety can be regarded as a typical product resulting from the coupling of adsorbed CO* and serves as a crucial precursor for C‐C coupling during acetic acid formation. The absorption peak at 1310 cm⁻¹ can be attributed to the C‐H bending vibration of hydrogenated intermediates following C‐C coupling, while the characteristic peak at 1431 cm⁻¹ corresponds to the COO⁻ stretching vibration of the *CH₃COO⁻ species, directly confirming the generation of acetic acid on the catalyst surface. Notably, BOB and 0.5COF/BOB exhibit similar characteristic peak positions for the aforementioned key intermediates, indicating that both follow an identical pathway for CH₃COOH production. However, compared with BOB, 0.5COF/BOB displays significantly stronger absorption intensities at the peak positions corresponding to key intermediates such as COOH*, CHO*, CH₃O*, OCCO*, and *CH₃COO⁻. This suggests that the Bi asymmetric dual active sites induced by Bi‐O/N coordination reconstruction more effectively facilitates the generation and accumulation of these reaction intermediates, thereby accelerating the conversion of CO₂ to CH₃COOH.

To further elucidate the reaction mechanism, density functional theory (DFT) calculations are conducted to investigate the surface charge distribution and reaction pathways for the photoreduction of CO₂ to CH₃COOH on BOB and COF/BOB. Electrostatic potential analysis reveals that the work function of the BOB surface is 5.008 eV (Figure 4E), whereas upon the introduction of COF molecules, the work function of COF/BOB significantly increases to 5.391 eV (Figure 4F), representing an increment of approximately 0.383 eV. This elevation indicates that the introduction of COF molecules induces the formation of an interfacial dipole moment pointing toward the substrate, thereby establishing a higher surface potential barrier. This makes electron escape from the BOB surface more difficult and enhances surface polarization. Density of states (DOS) analysis (Figure S39) shows the emergence of DOS signals near the Fermi level (E_f) for interfacial Bi atoms, alongside an enhancement in the total DOS of COF/BOB, signifying an alteration in the electronic structure and triggering interfacial charge redistribution. Bader charge analysis results (Figure S40) further corroborate that the 0.5COF/BOB heterojunction undergoes a significant variation in local charge distribution due to interface formation. Consequently, atomic charge states are readjusted, exhibiting a pronounced interfacial polarization effect. Charge density difference calculations (Figure 4G) reveal the presence of highly localized charge accumulation regions at the interface. Specifically, within the region marked by the dashed line, two adjacent Bi atoms in BOB exhibit increased (yellow; coordinated with N atoms) and decreased (blue; coordinated with O atoms) charge densities, respectively. Combined with characterization results such as XAFS and KPFM, it is inferred that the COF/BOB heterojunction triggers interfacial charge rearrangement through the formation of stable Bi─O coordination bonds. The bonded Bi atoms achieve an electron‐deficient state (Bi (δ⁺)) due to electron polarization toward O, while adjacent Bi atoms increase electron density via coordination with N atoms in the electron‐rich π‐conjugated COF skeleton, forming an electron‐rich state (Bi (δ⁻)), thereby constructing Bi (δ⁺)—Bi (δ⁻) asymmetric dual active sites featuring differential charge densities. This strong internal electric field induced by interfacial recoordination, combined with the asymmetric charge distribution, synergistically promotes photogenerated charge separation. Acting as an electron donor, the Bi (δ⁻) site selectively injects its electron density into the lowest unoccupied molecular orbital (2π antibonding orbital) of the adsorbed *CO intermediate, thereby weakening the C─O bond and activating the carbon atom. Simultaneously, the adjacent electron‐deficient Bi (δ⁺) site engages in strong d‐π electron feedback (back‐donation) with the π orbitals of the *CO intermediate via its d‐orbitals [36]. This synergistic orbital interaction not only facilitates electron transfer from the metal center to the adsorbate but, crucially, effectively screens the electrostatic repulsion between adjacent *CO intermediates due to the asymmetric charge distribution. This optimizes the adsorption configuration of the key OCCO* intermediate and stabilizes the transition state for C‐C coupling. Furthermore, Gibbs free energy calculations (Figure 4H) indicate that the energy barrier for the formation of the key intermediate *COOH for CO₂ activation decreases from 1.97 eV on BOB to 1.45 eV on COF/BOB. Additionally, the activation energy barrier for this C‐C coupling step decreases to 1.22 eV at the COF/BOB heterojunction interface, which is significantly lower than the 1.37 eV observed on pure BOB. Therefore, the asymmetric Bi (δ⁺)—Bi (δ⁻) dual active sites, induced by the reconstruction of interfacial Bi‐O/N coordination, effectively lower the energy barrier of the rate‐determining step through synergistic orbital interactions and charge modulation, thereby globally enhancing the reaction efficiency for the photoreduction of CO₂ to CH₃COOH.

3. Discussion

Here, effective control over the interfacial structure of Bi₂₄O₃₁Br₁₀ (BOB) was achieved by introducing covalent organic framework (COF) with intrinsic local disorder during BOB synthesis. STEM and XAFS confirm that the localized non‐periodic structure within the COF induces atomic‐scale rearrangement of the BOB surface layer, spontaneously forming a large‐area non‐periodic active interface rich in Bi(δ+)‐Bi(δ‐) asymmetric bimetallic sites. KPFM and fs‐TAS revealed at the macroscopic and microscopic scales, respectively, that the COF‐modified layer at this interface further promotes the directed transport of photoexcited electrons from BOB to the active sites. Without sacrificial agents or noble metal co‐catalysts, this catalyst synergistically achieves efficient CO₂ photoreduction to CH₃COOH rate of 1.03 mmol g⁻¹ h⁻¹ with 97.63% selectivity. At wavelengths of 380 and 400 nm, its apparent quantum efficiency reaches 13.75% and 8.33%, respectively. This strategy provides a paradigm for precisely constructing asymmetric active centers on inorganic surfaces by leveraging the intrinsic disorder of organic components, with potential for extension to diverse energy catalytic systems to advance artificial photosynthesis and carbon‐neutral technologies.

Author Contributions

Jun Xiong: investigation, resources. Yi Zhang: writing – original draft, investigation. Zhiwei Shao: data curation, investigation, writing – original draft. Wei Jiang: resources. Haipeng Zuo: data curation. Caichao Ye: investigation, software. Jun Di: conceptualization, methodology, writing – review and editing, funding acquisition, supervision.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: adma73924‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.