【新华网直播】“3·15”国际消费者权益日年主题宣传活动新闻发布会

Introduction

In natural photosystem I, chlorophyll P700 is excited followed by the transport of electrons to the catalytic center Fe₂S₂ cluster to produce reduced nicotinamide adenine dinucleotide phosphate,^1,2 which provides a stable charge/mass input for efficient CO₂ fixing and chemical manufacturing (Calvin-Benson cycle).³ Basically, the unique Fe₄S₄ redox protein clusters offer a robust structural foundation for the electron transfer,⁴ diverse amino acid residues (e.g., Cys/Glu/Ser, etc.) promote the efficient proton transport,⁵ while the ribulose bisphosphate provides an alkaline site for CO₂ binding (Figure?1a)^6,7 and functional regionalization, which synergistically contributes to the highly efficient CO₂ photoconversion. Fascinated by the delicate structures and efficient processes of natural photosynthesis, substantial endeavors have been devoted to exploiting artificial structures to recapitulate the biological charge/mass transfer processes.^8–13 Covalent organic frameworks (COFs), characterized by their covalently linked skeletons, high crystallinity, and precise chemical structures, have held great promise in artificial photosynthesis.^14–17 For example, state-of-the-art COFs have achieved (1) highly efficient electron transfer with a build-in field,¹⁸ D-π-A configuration,^19,20 redox junction,²¹ metal anchoring,^22,23 and delocalized π-configuration;^24–26 (2) remarkable proton conductivities and proton-coupled electron transfer with sulfonic,²⁷ imidazole,²⁸ pyrrole,²⁹ etc.; and (3) excellent CO₂ adsorption with an amino environment,^30,31 various metal sites,^32–34 and unique porous structure.³⁵ Despite this progress, a significant challenge remains in uniting the rapid transport of electrons, protons, and CO₂ in one COF while realizing the collaborative functional regionalization of the three processes.

• Download figure
• Download PowerPoint

Metal porphyrins and their derivatives, with delocalized π-electron tetrapyrrolic structures,^36,37 provide a molecularly precise platform for CO₂ binding^38,39 and charge transport (electron relay).⁴⁰ Nevertheless, the highly symmetric structure of the conventional porphyrins limits the directionality of electron and proton transport, thereby constraining the enhancement of artificial photosynthesis activity (Figure?1b top). Recent reports in the field of small molecules have demonstrated that porphyrin desymmetrization (e.g., O doping,⁴¹N-confused configuration⁴²) can regulate the kinetics and/or thermodynamics of CO₂ reduction. Drawing inspiration from these findings, it is promising to leverage the inherent merits of COFs in charge and mass transport by employing a desymmetrization strategy to modulate the directional charge and mass transport in COF toward higher catalytic performance. Taking the substitution of one pyrrole with thiophene in porphyrin as an example, three advantages can be achieved: (1) the more aromatic thiophene ring, with its available d-orbitals,^43,44 can serve as a bridge for electron transport; (2) S atom in thiophene acts as an electron donor (ED),⁴⁵ which increases the charge density of metal site to promote CO₂ binding; (3) the difference in electronegativity between thiophene S and pyrrole N creates a variation in the gradient of the charge density, which can be exploited to achieve the spatial separation of the electron and proton transport channels (Figure?1b bottom). However, the construction of crystalline COFs incorporating desymmetrized porphyrin moieties remains an enormous synthetic challenge in this field, which is highly desirable, yet rarely reported.

Herein, as a proof-of-concept, we construct a crystalline COF with a well-defined metal-thiophene-tripyrrole (NiSN₃Por) moiety via the Knoevenagel condensation reaction. As expected, this unique symmetry-breaking NiSN₃Por configuration exhibits excellent performance in photocatalytic CO₂ conversion. When compared with the D_4h symmetric NiN₄Por-PDAN-COF, the photocatalytic CO generation rate of NiSN₃Por-PDAN-COF is increased by ～65 times under visible light conditions (xenon lamp, λ?>?420?nm). In-depth mechanistic investigations reveal that the NiSN₃Por moiety in COF is endowed with a clear division of labor including thiophene as an electron transport region, pyrrole as a proton conduction channel, and metal site as a CO₂ binding site, which cooperatively facilitates the performance of CO₂ photoconversion.

Experimental Methods

Synthesis of NiSN₃Por-PDAN-COF

A pyrex tube was charged with Ni–SN₃(Cl)Por-CHO (16.74?mg, 0.02?mmol), p-phenylenediacetonitrile (PDAN) (9.37?mg, 0.06?mmol), o-dichlorobenzene (o-DCB) (1?mL), acetonitrile (ACN) (2?μL), and aqueous 1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU) (100?μL, 5?M). After being degassed through three freeze-pump-thaw cycles and then sealed under vacuum, the tube was heated at 80?°C for 5?days. The resulting precipitate was collected by filtration, exhaustively washed by Soxhlet extraction with tetrahydrofuran (THF) and dichloromethane (DCM) for 24?h, and dried under vacuum at 80?°C. The product was isolated as a black powder (18.34?mg, 85.1% yield).

Synthesis of NiN₄Por-PDAN-COF

A pyrex tube was charged with Ni–N₄Por-CHO (15.66?mg, 0.02?mmol), PDAN (9.37?mg, 0.06?mmol), o-DCB (2?mL), ACN (10?μL), and aqueous DBU (100?μL, 5?M). After being degassed through three freeze-pump-thaw cycles and then sealed under vacuum, the tube was heated at 120?°C for 7?days. The resulting precipitate was collected by filtration, exhaustively washed by Soxhlet extraction with THF and DCM for 24?h, and dried under vacuum at 80?°C. The product was isolated as a red powder (17.01?mg, 83.2% yield).

Photocatalytic experiments

Photocatalytic CO₂ reduction experiments were carried out in ～15?mL glass tubes (Agilent Technologies Co. Ltd, China) with COF (1?mg), triethylamine (TEA) (1?mL), Ru(bpy)₃Cl₂ (6.5?mg), and 2,2′-bipyridine (15?mg) in 5?mL pure water. The solution was bubbled with CO₂ (1?atm) for 30?min. Four hundred microliters of CH₄ (20.01% in N₂) were injected into the system and functioned as the internal standard for quantitative analysis. Then the system was subjected to irradiation under 465?nm light-emitting diodes (LEDs) (～20?mW/cm²). The gaseous products in the headspace of the tubes were probed by gas chromatography (GC), and the liquid products for CO₂ reduction were examined by nuclear magnetic resonance (NMR). Further optimized CO₂ photoreduction experiments were carried out in ～15?mL glass tubes (Agilent Company) with COF (0.5?mg), Ru(bpy)₃Cl₂ (10?mg), 2,2′-bipyridine (5?mg), and 50?mg dimethylphenyl benzimidazoline (BIH) in a mixture of 5?mL ACN and 1?mL water. The system was subjected to irradiation under an Xe lamp with a 420?nm filter.

Density functional theory calculations

The initial crystal structure of the COFs with overlapping (AA) stacking was constructed through the Materials Visualizer BIOVIA Materials Studio 2019 (19.2), and then these structures together with the cells were optimized with the generalized gradient approximation Perdew–Burke–Ernzerhof exchange-correlation functional and double numerical-polarized basis set (4.4 version) adopting Effective Core Potentials with DMol³ module of BIOVIA Material Studio 2019 software.

The lattice constants of the COFs were optimized to 26.043?×?26.043?×?4.065?? (α?=?β?=?γ?=?90°) for the NiN₄Por-PDAN-COF, and 25.691?×?25.691?×?4.715?? (α?=?β?=?γ?=?90°) for the NiSN₃Por-PDAN-COF. The convergence criteria for structure optimization and energy calculation were set as: self-consistent field (SCF) tolerance (1?×?10^?5?Ha), energy tolerance (2?×?10^?5?Ha), a maximum force tolerance (4?×?10^?5?Ha/?), and maximum displacement tolerance (5?×?10^?5??).

Additional materials, synthetic procedures, and calculation details can be found in the Supporting Information.

Results and Discussion

The target COF (NiSN₃Por-PDAN-COF) and its symmetric counterpart (NiN₄Por-PDAN-COF) were synthesized from aldehyde monomers using the Knoevenagel condensation reaction (Figure?2a, see detailed synthetic procedures in Supporting Information Figures S1–S4). Herein, PDAN was employed as a bridging unit to introduce C=C bonds in the COF to guarantee a higher charge transfer efficiency.⁴⁶ Notably, axial Cl^? as an anion, binds with Ni in an ionic bond state,⁴¹ which separates from the Ni center before contacting with CO₂ (vide infra). The chemical structure of obtained COFs was first assessed by a combination of Fourier transform infrared (FT-IR) spectroscopy and solid-state NMR (ssNMR) spectroscopy. From the FT-IR spectra, compared with the two monomers (NiN₄Por-CHO and NiSN₃Por-CHO), the peaks of two COFs at ～1670?cm^?1 were significantly decreased, while the signals at ～3025 and ～2216?cm^?1 appeared, indicating that –CHO is converted into a –CH=C(CN)– group ( Supporting Information Figure S5).^47,48 Additionally, in ssNMR spectra ( Supporting Information Figure S6), the disappearance of the –CHO signal and the emergence of the –CN signal confirm again the formation of the –CH=C(CN)– unit. From thermogravimetric analysis, both COFs are thermally stable above 260?°C ( Supporting Information Figure S7).

• Download figure
• Download PowerPoint

Then, structural simulation combined with powder X-ray diffraction (PXRD) analysis was employed to determine the stacking patterns of both COFs. Considering the variability of the thiophene and pyrrole structures, we established three possible patterns of stacking, that is, direct S–S interaction between the two layers, S interaction with ortho-site, and S interaction with para-site (Figure?2b). Geometry optimization results show that the stacking mode corresponding to S with a para-site exhibits the lowest energy, that is, a staggered-like pattern for the thiophene structure in the porphyrin COF stacking is the most stable (Figure?2c). Experimentally, the PXRD pattern of NiSN₃Por-PDAN-COF exhibits main peaks at ～3.5°, 7.0°, and 10.5°, corresponding to the (100), (200), and (300) crystal facets, respectively. After Rietveld refinement, the corresponding para-site NiSN₃Por-PDAN-COF pattern well matches with the experimental data, with final R_wp and R_p values converging to 3.91% and 2.80%, respectively (Figure?2d and Supporting Information Figure S8). Similarly, the symmetric counterpart (NiN₄Por-PDAN-COF) exhibits an AA stacking model, with R_wp and R_p values of 2.23% and 1.77%, respectively ( Supporting Information Figures S9 and S10). We also conducted a nitrogen sorption isotherm measurement for both COFs at 77?K to investigate their porous properties. Pore size evaluation from the isotherms of both COFs shows a main peak centered at ～2.7?nm ( Supporting Information Figure S11), which is consistent with the calculated pore diameter (diagonal measurement in Figure?2e,f).

To gain a deep understanding of the microstructure of the two COFs, scanning electron microscopy (SEM) was conducted. SEM images reveal that these two COFs are endowed with the same morphology of stacked sheets ( Supporting Information Figure S12). High-resolution transmission electron microscopy (HR-TEM) analysis provides more accurate structures of the two COFs. As described in Figure?3a,b, an interplanar spacing of 2.6?nm is identified, indicating the (100) crystal facet of NiSN₃Por-PDAN-COFs. In addition, the porous character of NiSN₃Por-PDAN-COF can be clearly observed, which is in good agreement with the structural simulation results (Figure?3b). Similar TEM results for NiN₄Por-PDAN-COF are presented in Supporting Information Figure S13. Furthermore, high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) was employed to analyze the main elemental distribution of NiSN₃Por-PDAN-COF. As shown in Figure?3c, the S, N, and Ni elements are evenly distributed on the 2D plane. High-resolution image shows that the high-bright spots are homogeneously distributed in the COFs (Figure?3d). Combined with electron energy loss spectroscopy (EELS) analysis (Figure?3e), we can confirm that the isolated Ni atoms are uniformly anchored in the COF structure, without the formation of Ni clusters during the COF growth. In addition, the total Ni content in NiSN₃Por-PDAN-COF and NiN₄Por-PDAN-COF is determined to be 5.13 and 5.51?wt %, respectively, by inductively coupled plasma optical emission spectrometry, which matches well with the theoretical values ( Supporting Information Table S1).

• Download figure
• Download PowerPoint

To deeply investigate the electronic and spatial configurations of catalytic sites in both COFs, synchrotron radiation X-ray absorption spectroscopy (XAS) was performed. As portrayed in Figure?3f, the intensity of the pre-edge peak for NiSN₃Por-PDAN-COF in X-ray absorption near edge structure spectra (XANES), originating from electronic transitions from the 1s to 3d orbitals, obviously increased, while the shoulder peak at ～8339?eV (1s to 4p_z transition) decreased compared with the symmetric structures (NiN₄Por-PDAN-COF), indicating that the D_4h symmetry of the coordination environment around Ni atom is broken.⁴¹ Meanwhile, compared with NiN₄Por-PDAN-COF, the white line intensity of NiSN₃Por-PDAN-COF decreases, illustrating that the charge density around Ni is increased. In addition, by means of quadratic differential analysis for the absorption edge ( Supporting Information Figure S14), the average oxidation number of NiSN₃ is lower than that of the NiN₄ structure, also verifying that the charge density increased around Ni atoms for NiSN₃ configuration. Combining the changes in the intensity of the pre-edge peak, absorption edge, and white line peak, we can conclude that the introduction of S atoms not only destroys the D_4h symmetry of Ni coordination environment, but also increases the charge density of the catalytic sites.

In the extended X-ray absorption fine structure (EXAFS) region, compared with the NiN₄ structure, the amplitude for NiSN₃ in k-space spectra decreases, illustrating that the coordination number for the Ni atom is reduced ( Supporting Information Figure S15). Fourier transform R-space analysis shows that the main peak of NiSN₃ at ～1.5??, which is attributed to Ni–N single scattering, declines compared with the NiN₄ counterpart, also illustrating that coordination number in the first coordination layer (Ni–N) decreases. Meanwhile, the peak at ～2.0?? appears, which can be ascribed to Ni–S and Ni–Cl bonds. Additionally, no characteristic peak of Ni–Ni (at ～2.2??) is observed for both NiN₄ and NiSN₃ structures, suggesting that the isolated Ni atoms are atomically dispersed in the porphyrinic skeletons (Figure?3g). The detailed quantitative fitting results ( Supporting Information Figure S16 and Table S2) further confirm the configurations of both COFs. Compared with NiN₄, the distance of the Ni–N bond for NiSN₃ decreases from 1.93 to 1.82??. This phenomenon is due to the compression of the Ni–N bond by the bulky S atoms. Continuous Cauchy wavelet transform (CCWT) analysis by combination of both k and R space provides more information on the coordination environments of Ni sites.⁴⁹ Taking Ni foil at a maximum of ～7.6??^?1 as a Ni–Ni bond standard, no signal can be observed at this position for both NiN₄ and NiSN₃ structures, further confirming atomically dispersed Ni sites. Obviously, the maximum corresponding vector k position in CCWT of NiSN₃ shifts slightly to a lower wavenumber compared with the NiN₄ structure (Figure?3h–k). This result is observed because of a decreasing bond length between Ni and N, which is well consistent with the above EXAFS fitting result.

The performance of photocatalytic CO₂ conversion of both COFs was then investigated with [Ru(bpy)₃Cl₂] as a photosensitizer (PS) and TEA as an electron donor (ED) in pure water with 1?atm CO₂ under 465?±?5?nm LED light irradiation (～20?mW cm^?2). Excitingly, compared with the symmetric structure (NiN₄Por-PDAN-COF), the CO generation rate of NiSN₃Por-PDAN-COF increases more than 27.6 times, and the corresponding CO generation rate is calculated as 713.8?μmol g^?1 h^?1 with a selectivity of 90.4%, albeit in a pure water environment (Figure?4a). The control experiments of the photocatalytic system are discussed in Supporting Information Figure S17. It illustrates that the light source, PS, COF catalyst, and ED are all necessary for the light reaction system. Noticeably, a negligible CO signal can be detected in the absence of Ru(bpy)₃Cl₂, indicating the cocatalyst nature of two COFs in the photocatalytic CO₂-to-CO conversion.^38,50 Compared with NiSN₃Por-CHO and PDAN monomers, the activity and selectivity of NiSN₃Por-PDAN-COF for CO generation are improved, which may be attributed to the inherent merits of COF with favored charge and mass transport (Figure?4b). This conclusion can also be inferred from the NiN₄Por-PDAN-COF catalytic system ( Supporting Information Figure S18). With the aid of ¹H-NMR and GC-flame ionization detector (FID) analysis, there is no signal of CH₃OH or HCOOH, and only a trace amount of CH₄ is observed ( Supporting Information Figure S19), verifying that the system scarcely produces HCOOH, CH₃OH, or CH₄ from CO₂ photoconversion. In addition, we examined the catalytic stability of the system for NiSN₃Por-PDAN-COF. After three catalytic cycles (10?h?×?3), the activity for CO production is well maintained (Figure?4c). Moreover, the FT-IR spectra ( Supporting Information Figure S20) and PXRD patterns ( Supporting Information Figure S21) of the recovered NiN₄Por-PDAN-COF and NiSN₃Por-PDAN-COF after 30?h reveal that the chemical structure and the crystalline structure are maintained during the photocatalysis, indicating its excellent stability. Furthermore, we performed a ¹³C labeling test by using ¹³CO₂ as the C-source. The products are almost completely converted into ¹³CO (m/z?=?29) for NiSN₃Por-PDAN-COF (Figure?4d), confirming that the CO product is indeed generated from reactant CO₂, instead of PS, COF catalyst, or ED.

• Download figure
• Download PowerPoint

To further elevate the photocatalytic performance, ACN with good CO₂ solubility and electron-donating reagent dimethylphenyl BIH were introduced into the system. Under a xenon-lamp condition (λ?>?420?nm), the CO production rate of NiSN₃Por-PDAN-COF reaches 22.6?mmol g^?1 h^?1 with a selectivity of 92.2%, which is comparable with the reported COF cocatalysts ( Supporting Information Table S3). In contrast to the NiN₄Por counterpart, the CO generation rate was elevated by ～65 times for NiSN₃Por-PDAN-COF (Figure?4e). Inspired by the satisfactory result, the realistic sunlight condition was adopted. On a cloudy day (May 17th, ～24?°C) with 4?h of illumination, CO production reached ～19.7?mmol g^?1, indicating natural sunlight can be employed for CO₂ photoconversion with good effect (Figure?4f).

Considering the cocatalyst nature of both COFs (vide ante), the electrochemical techniques and theoretical calculations were then performed to study the electron transfer processes of the catalytic systems. In the cyclic voltammetry (CV) curves, quasi-reversible waves were observed, and the corresponding half-wave potentials (E_1/2) for NiSN₃Por-PDAN-COF and NiN₄Por-PDAN-COF were calculated as ?0.65 and ?0.63?V (vs normal hydrogen electrode) respectively, both higher than the potential of Ru*(II/III) (?0.84?V) in water. These results suggest that the electron transfer from the excited state of Ru(bpy)₃²⁺ to the two COFs is thermodynamically feasible ( Supporting Information Figure S22). To investigate the number of electrons transferred per lattice unit of COFs, a combination of normal pulse voltammetry (NPV) and diffusion-ordered ¹H-NMR spectroscopy was employed ( Supporting Information Figure S23).^51,52 For a NiN₄Por monomer, the number of electrons transferred is calculated as ～1.80 at ?1.23?V (vs Ag/AgCl), demonstrating that each reduction event in the NiN₄Por-COF unit is associated with a two-electron reduction process.⁵³ Interestingly, for a NiSN₃Por monomer, CV and NPV show two obvious waves, which can be calculated as ～0.98 and ～1.90 electrons transferred at ?0.20 and ?0.76?V (vs Ag/AgCl), attributing to the Cl^? shedding and two-electron extraction behaviors, respectively. The difference between the electrochemical processes of the monomer and COF may be assigned to the highly conjugated structure in COF, which allows the two electrochemical processes in COF to overlap. Guided by electron transfer investigation, the detailed catalytic cycles for NiN₄Por-PDAN-COF and NiSN₃Por-PDAN-COF are shown in Supporting Information Figure S24. Upon visible-light irradiation, Ru(II) is excited followed by the transfer of electrons to the catalytic centers. For NiN₄Por-PDAN-COF, each structural unit harvests two electrons, subsequently combining with H⁺ and CO₂ to generate Ni-COOH species. For NiSN₃-PDAN-COF, each structural cell can capture three electrons, the first electron is inputted accompanied by the removal of Cl^? to generate [NiSN₃Por-PDAN]⁰, followed by the two-electron injection to form [NiSN₃Por-PDAN]^2? intermediate species, further reacting with H⁺ and CO₂ to produce Ni-COOH species. Finally, *COOH undergoes another H⁺ transfer process to generate *CO. Noticeably, Cl^?-removed [NiSN₃Por-PDAN]⁰ as an initial cocatalyst is involved in the catalytic cycle. Both COFs experience EEC (E?=?electron transfer, C?=?chemical reaction)⁵⁴ rather than ECE processes.

Femtosecond transient absorption spectroscopy (fs-TAS) was employed to evaluate the ligand-to-metal site electron transfer rates. As depicted in Figure?5a,b, the contour map of fs-TAS shows that the kinetic behavior lasts ～1000?ps for the NiN₄Por structure, while the NiSN₃Por decays within ～50?ps. For the detailed analysis of fs-TAS for NiN₄Por, the peaks at 430, 520, and 550?nm correspond to the S₂ state, Q-band ground-state bleaching (GSB), and S₁ state kinetic events, respectively (Figure?5c).⁵⁵ For NiSN₃Por counterpart, the peaks at 460, 510, 560, and 580?nm represent the B-band GSB, S₂ state, Q-band GSB, and S₁ state kinetic behaviors, respectively (Figure?5d). As the electron transport goes through the ligand-unoccupied orbital to reach the metal catalytic site, the S₁ state kinetic decay is analyzed in detail (Figure?5e) and the results of the two-exponential fit are summarized in Supporting Information Table S4. Referring to the reported metal porphyrin description,⁵⁶ the short-lived parameter corresponds to the ligand-to-metal charge transfer (LMCT) transition and the long-lived result represents the metal-to-ligand charge transfer (MLCT) process. Therefore, compared to the NiN₄Por structure, both the LMCT and MLCT events are substantially accelerated for NiSN₃Por (Figure?5f). Correspondingly, the steady-state and transient photoluminescence intensities of Ru(bpy)₃Cl₂ are more effectively quenched by introducing NiSN₃Por-PDAN-COF as a cocatalyst, further indicating that the Ni center in NiSN₃Por moiety displays a higher photoelectron extraction efficiency than that of the NiN₄Por counterpart ( Supporting Information Figure S25). Additionally, from the photocurrent response analysis, a higher photocurrent density is observed for NiSN₃Por-PDAN-COF under identical conditions ( Supporting Information Figure S26a), indicating that the transfer efficiency of electrons is higher than that of NiN₄Por-PDAN-COF. Moreover, the NiSN₃Por-PDAN-COF electrode exhibits a smaller resistance in electrochemical impedance spectroscopy (EIS) measurements compared with the NiN₄Por-PDAN-COF ( Supporting Information Figure S26b and Table S5), thereby contributing to the more feasible electron transport behavior.

• Download figure
• Download PowerPoint

To gain a deeper understanding of the reasons behind the accelerated LMCT process with NiSN₃Por configuration, the orbital-related projected density of state (PDOS) calculations were dissected in detail (Figure?5g). For NiN₄Por, the unoccupied d_x2-y2 orbitals dominate above the Fermi energy level and the splitting energy value is calculated as 1.257?eV between the d_x2-y2 and d_z2 orbitals. On the contrary, when thiophene replaces a pyrrole ring, the d_x2-y2 orbitals are broadened and part of the region appears on the Fermi energy level. At the same time, the splitting energy value between the d_x2-y2 and d_z2 orbital decreases to 0.287?eV, leading the NiSN₃ structure to exhibit a conductor-like property (i.e., higher electronic conductivity). Inspired by the fs-TAS and PDOS analysis, a clear electron transport pathway is illustrated (Figure?5h). Compared to NiN₄Por, thiophene in NiSN₃Por possesses a unique S d-orbital^41,42 involved in the porphyrin conjugation and acts as a high-speed tunnel, thereby accelerating the LMCT process. Meanwhile, the decrease in splitting energy also contributes to the d-electron transition process, which further facilitates electron injection into CO₂ molecules.

Furthermore, proton transport serves as another key factor for artificial photosynthesis. Mulliken charge analysis shows that the S atom in thiophene moiety exhibits a positive value while the N atom in pyrrole presents a negative result, illustrating that the S atom and N atom behave electrophilicity and nucleophilicity (i.e., H⁺ adsorption), respectively (Figure?6a,b). Electrostatic potential (ESP) distributions exhibit an electrostatic barrier forming around the central metal for NiN₄Por, which hinders the electron transfer from the macrocyclic ligand to the metal site. In contrast, for the NiSN₃Por structure, the thiophene moiety breaks through the electrostatic barrier and generates a high-speed channel for electron transport. Meanwhile, the remaining three pyrrole structures in NiSN₃ moiety form an electron-rich domain that offers targets for proton migration (Figure?6c). To unveil the reason behind this phenomenon, Bader charge analysis was employed to evaluate the electronic gains and losses. It is found that the introduction of S can effectively increase the charge density of central Ni and the remaining N atoms ( Supporting Information Figure S27). This result illustrates that electron-donating S in the thiophene creates a charge density gradient difference in the (δ⁺)S–Ni–N(δ^?) configuration as well as maintains a high charge density around Ni, which well corresponds to the above XANES analysis (vide supra).

• Download figure
• Download PowerPoint

Guided by the above results, H atom and CO₂ adsorption events on different sites were carefully simulated. Taking NiN₄Por structure as a reference, the adsorption energy of H at the S site is the highest, which means that the S site has a strong repulsion with H. On the contrary, H interacts ortho- and para-N with lower adsorption energies. Specifically, H on the para-N endows the most stable structure, that is, H prefers to adsorb on the para-N site ( Supporting Information Figure S28). Moreover, CO₂ favors the S–N orientation on the NiSN₃ configuration from the energy analysis ( Supporting Information Figure S29). Inspired by the adsorption behaviors of H and CO₂, we calculated the adsorption energy of the key intermediate COOH with different orientations (Figure?6d). As a result, the *COOH energies match well with the above H and CO₂ adsorption behaviors. That is, H prefers to interact with CO₂ from the para-pyrrole side, thus CO₂ can be smoothly converted to COOH via this pathway (Figure?6e). Moreover, the differential charge density of COOH with the catalytic site was analyzed to verify this event (Figure?6f). Compared with the NiN₄Por-COOH assembly, COOH on the NiSN₃Por presents more electron accumulation and depletion processes, indicating again that COOH formation is favored on the NiSN₃Por moiety. Additionally, compared to the electron-rich region of pyrrole N in NiN₄Por, the electron-poor region of thiophene S acts as an electron channel to accelerate electron transfer, while the electron-rich domain in para-pyrrole N contributes to the H⁺ migration. As a consequence, these three processes, that is, electron transfer, proton transport, and CO₂ binding, are spatially separated and functionally regionalized to synergistically promote CO₂ to CO conversion. In contrast, COOH on the NiN₄Por with a uniform structure provides a lack of target points that hinders the directional migration of protons and electrons, resulting in sluggish COOH formation.

Experimentally, FT-IR spectroscopy was employed to verify the aforementioned density functional theory (DFT) calculations. Here, to reduce the interference of H₂O on the infrared window, we used D₂O as a reactant to study the change in infrared spectra.⁵⁷ As shown in Figure?6g, the signal at ～1450 and 1660?cm^?1 is assigned to *COOD, and the signal at ～2010?cm^?1 is attributed to *CO. Remarkably, it can be found that the increased rate of *COOD signals on the NiSN₃Por-PDAN-COF surface is much faster than that of the NiN₄ counterpart with the same extension of illumination time. This phenomenon indicates that the NiSN₃ asymmetric structure is more conducive to forming Ni-COOH intermediate, thereby significantly accelerating the CO₂-to-CO reaction process, which matches well with the above theoretical models of collaboratively functional regionalization.

Conclusions

In summary, mimicking natural photosynthesis, we reported a desymmetrization strategy in porphyrinic COF that segregates electron transfer, proton transport, and CO₂ binding domains to realize their respective functions. Compared with the conventional D_4h symmetric NiN₄-typed COF, this unique symmetry-breaking NiSN₃ unit in COF photoconverts CO₂ to CO with higher activity and selectivity. Further advanced spectroscopy and DFT thermodynamics calculations unveiled that, in NiSN₃Por-typed COF, the thiophene region bears the responsibility of the fast electron channel, the pyrrole moiety plays the role of the proton hopping target, and the metal acts as the CO₂ binding site. These three regions are spatially separated but yet work in tandem to promote CO₂ photoconversion. We anticipate that the desymmetrization principle demonstrated in our system can offer a paradigm to develop other COFs for more accurate recapitulation of natural biological processes and applications in challenging artificial fields (e.g., H₂O oxidation, O₂ reduction, N₂ reduction, etc.).

Supporting Information

Supporting Information is available and includes materials, synthetic procedures, construction of COFs, additional characterizations, CO₂ reduction control experiments, detailed DFT calculations, and NMR data of relevant monomers including Figures S1–S38 and Tables S1–S5 where Figures S30–S38 are the NMR of the monomer.

Conflict of Interest

There is no conflict of interest to report.

Funding Information

This research was made possible as a result of a generous grant from Research Grants Council (RGC) Senior Research Fellowship Scheme (grant no. SRFS2021-5S01), the Hong Kong Research Grants Council (PolyU 15307321), Research Institute for Smart Energy (CDAQ), Research Centre for Nanoscience and Nanotechnology (CE2H), Miss Clarea Au for the Endowed Professorship in Energy (847?S), National Key R&D Program of China (grant nos. 2022YFA1502903 and 2022YFA0911900), the National Natural Science Foundation of China (grant nos. 22309156, 22088102 and 21971251), and the New Cornerstone Science Foundation.