Synergistic Interaction Between N‐Heterocyclic Carbene (NHC)‐Anchored Cu(I) Atom and In Situ‐Generated Cu Nanoparticles in a Post‐Modified MOF‐808 Framework Promoting C─C Coupling in CO₂ Reduction

ABSTRACT

The electrochemical reduction of CO₂ to high‐value chemicals like C₂H₄ represents a promising route for sustainable energy and CO₂ mitigation. However, its efficiency remains constrained by the high energy barrier for C─C bond formation and the competing hydrogen evolution reaction (HER). Herein, we report a synergistic catalyst for electrochemical CO₂ reduction reaction (eCO₂RR), which comprises an N‐heterocyclic carbene (NHC)‐anchored Cu(I) atom and in situ electrogenerated Cu nanoparticles in a MOF‐808 framework. The catalyst exhibits outstanding performance in a neutral electrolyte, achieving a high Faradaic efficiency (FE) of 61.0% for C₂H₄ and 79.5% for total C₂₊ products. Experimental and theoretical studies reveal a dual role of the NHC ligand in eCO₂RR: (i) promoting proton transfer and *CO hydrogenation via a robust hydrogen‐bonding network with interfacial water and (ii) stabilizing key intermediates such as *CHO and *COCHO through its strong electron‐donating ability. The synergy between a molecular NHC‐Cu(I) atom and adjacent Cu nanoparticles in a tailored microenvironment suppresses HER and lowers the energy barrier for asymmetric C─C coupling. This work offers a strategic design concept for constructing molecular‐nanostructured synergistic active sites in metal‐organic frameworks (MOFs) to advance electrocatalytic CO₂ conversion.

The catalyst undergoes in situ reconstruction within MOF‐808 pores, yielding coexisting Cu nanoparticles and NHC‐Cu(I) species. The NHC ligand facilitates water dissociation via a strong hydrogen‐bonding network and transfers *H to the NHC‐Cu(I) site for *CO hydrogenation to *CHO. This *CHO intermediate subsequently couples asymmetrically with *CO adsorbed on adjacent Cu nanoparticles, boosting C₂H₄ production.

1. Introduction

The conversion of CO₂ into valuable chemicals and fuels offers a promising path toward carbon neutrality [1, 2]. The electrochemical CO₂ reduction reaction (eCO₂RR), driven by renewable electricity, provides a sustainable means to reduce CO₂ emissions. Among the products, C₂H₄ is particularly valuable as an industrial feedstock [3]. However, the selective production of multi‐carbon (C₂₊) products such as C₂H₄ remains challenging. This reaction involves complex multi‐step proton‐coupled electron transfer and is hindered by two main factors: the high energy barrier for C─C coupling and the competing hydrogen evolution reaction (HER) [4]. Cu is considered the primary metal capable of catalyzing *CO dimerization to produce C₂₊ products, owing to its moderate *CO adsorption energy [5]. However, the ill‐defined structures of many conventional Cu catalysts (e.g., oxide‐derived and nanostructured Cu) often lead to ambiguous reaction pathways and working mechanisms, hindering rational catalyst design. Therefore, developing efficient Cu‐based catalysts that promote C─C coupling and elucidating their underlying mechanisms are essential for advancing the selective production of C₂₊ via eCO₂RR.

Several strategies have been developed to enhance CO₂ electroreduction and promote C–C coupling, including bimetallic coupling [6, 7, 8], metal doping [9, 10], and surface microenvironment regulation [11]. Among these, tailoring the surface microenvironment has proven highly effective in stabilizing key reaction intermediates and suppressing HER, thereby facilitating selective C₂₊ formation [12, 13, 14]. Metal‐organic frameworks (MOFs), with their well‐defined crystalline and periodic structures, provide an ideal platform for constructing uniform, atomic‐level catalytic sites [15, 16]. Such structural regularity enables precise control over reaction pathways and offers valuable mechanistic insights. More importantly, organic ligand functionalization exploits this structural tunability to modulate the local microenvironment and introduce functional groups that facilitate proton transfer and intermediate hydrogenation while suppressing HER, thereby enhancing C₂₊ selectivity.

N‐Heterocyclic carbenes (NHCs), known for their σ‐donation character, can anchor metal atoms through covalent metal‐C bonds. The robust electronic interaction between NHC ligands and Cu centers effectively modulates the electronic structure of Cu, fine‐tunes the binding strength of surface intermediates, and promotes the adsorption of key intermediates [17] while maintaining remarkable structural stability under electrocatalytic conditions [18]. Although conventional NHC‐metal complexes typically feature alkyl or aryl substituents, Hupp and colleagues developed an NHC‐based MOF incorporating metal‐containing side groups, synthesized directly from metal salts and imidazole ligands [19]. Weak organic bases such as dimethylformamide (DMF) and triethylamine (Et₃N) facilitate deprotonation at the C2 position of imidazolium salts, enabling the in situ formation of NHC‐metal motifs in the presence of Cu salts [20, 21]. This strategy circumvents the risk of MOF structural degradation associated with the use of inorganic strong bases such as potassium tert‐butoxide (KOtBu). Yaghi and coworkers introduced benzimidazole ligands into MOF‐808 via post‐synthetic modification (PSM), followed by metallization to form bis(μ‐oxo) dicopper sites [22]. Considering the potential of adjacent dicopper sites as active centers for eCO₂RR, we hypothesized that incorporating NHC‐Cu as a secondary active site into the MOF framework would introduce synergistic centers for C‐C coupling and expand the catalytic versatility of NHC‐metal motifs in CO₂ reduction.

Herein, we report a Cu‐N‐NHC‐M808 catalyst synthesized by a modified method, incorporating both bis(μ‐oxo) dicopper sites and NHC‐Cu(I) atoms into the pores of MOF‐808. This design exploits the different stabilities of Cu motifs: Cu‐N and Cu‐O are easily reduced to metallic Cu at negative potentials [23, 24, 25, 26], whereas Cu‐C bonds endow superior structural stability [18, 27]. Under operando conditions, the metastable bis(μ‐oxo) dicopper sites reconstruct into Cu nanoparticles. These nanoparticles cooperate with the adjacent stable NHC‐Cu(I) atom, creating a synergistic effect that promotes C‐C coupling and enhances C₂₊ production efficiency. Additionally, the ‐NH groups on the NHC ligand alter the aqueous microenvironment by forming a strong hydrogen‐bonding network with interfacial water molecules, which facilitates proton transfer for *CO hydrogenation while suppressing HER. By contrast, functionalization with indazole alone forms Cu─O/N units without Cu─C bonds. During electrolysis, such Cu─O/N systems undergo rapid Cu reconstruction, leading to aggregated Cu nanoparticles and poor catalytic stability. Cu–N–NHC MOF, however, exhibits enhanced CO₂‐to‐C₂H₄ performance and better stability. This study demonstrates the potential of integrating NHC‐Cu atoms with in situ‐generated nanoparticles within MOF pores to advance electrocatalysis through a synergy strategy, thereby optimizing eCO₂RR for the selective production of value‐added chemicals and fuels.

2. Results and Discussion

2.1. Synthesis and Structural Characterization of Pre‐Catalysts

The Cu–N–NHC–M808 and Cu–N–M808 catalysts were prepared via a post‐synthetic modification (PSM) of the Zr‐based MOF‐808 parent platform. (See the Supporting Information, Experimental Section, for details.) Specifically, 1H‐benzo[d]imidazole‐6‐carboxylic acid (Bzz) and 2H‐indazole‐5‐carboxylic acid (Iza) were employed to substitute the terminal formate anions, coordinated water molecules, and hydroxide groups decorating the MOF skeletons. Subsequent metalation of these grafted organic ligands introduced Cu─N, Cu–O, and Cu─C sites into the MOF frameworks, schematically illustrated in Figure 1a.

FIGURE 1.

(a) Design and synthesis of Cu–N–NHC‐M808 and Cu–N–M808. Atom labeling scheme: C, gray; O, red; N, green; Cu, orange; Zr, cyan polyhedral; H, light pink; I, violet. (b) XRD patterns of simulated MOF‐808 (black), as‐synthesized MOF‐808 (red), Bzz‐M808 (blue), and Cu–N–NHC–M808 (orange). (c) Solid‐state ¹³C NMR of Bzz‐M808 and Cu–N–NHC–M808. (d) FT‐IR spectra of Bzz (black), as‐synthesized MOF‐808 (red), Bzz‐M808 (blue), and Cu‐N‐NHC‐M808 (orange). (e) TEM image and elemental distribution mapping image of Cu‐N‐NHC‐M808. (f, g) AC HAADF‐STEM images of Cu‐N‐NHC‐M808.

MOF‐808 was first synthesized via a room‐temperature aqueous‐phase reaction, yielding particles with an average size of 35 nm (Figure S1). Materials with smaller particle sizes generally offer more uniform dispersion and superior electrocatalytic performance due to improved active site accessibility [28]. The as‐prepared MOF‐808 was then impregnated in a concentrated Bzz solution under heating to promote ligand exchange and coordination, resulting in the modified material denoted as Bzz‐M808. Subsequent metalation with Cu yielded Cu‐Bzz‐M808. A final base treatment and second metallization afforded the target catalyst Cu‐N‐NHC‐M808. Iza was grafted onto MOF‐808 under identical conditions to produce Iza‐M808, which after metalation gave Cu‐N‐M808. Powder x‐ray diffraction (PXRD) confirms the high crystallinity and phase purity of the as‐synthesized MOFs (Figure 1b and Figure S2), confirming framework integrity throughout ligand incorporation and Cu coordination. During metalation of Cu‐N‐NHC‐M808, Cu(I) ions initially coordinate with N and O atoms from the grafted ligands and atmospheric oxygen to form bis(μ‐oxo) dicopper sites [22]. Subsequent treatment with Et₃N deprotonates the imidazolium C2 position, generating neutral NHC ligands [21]. The resultant carbene lone pairs then coordinate to Cu ions, forming a stable atomic NHC‐Cu(I) site. This robust site features tunable electronic properties and exceptional stability under reducing conditions [18] and is proposed to act synergistically with in situ‐generated Cu nanoparticles to facilitate the critical C‐C coupling step. In contrast, the pyrazole‐type units in Cu‐N‐M808 coordinate with Cu ions to form planar trinuclear Cu clusters within the MOF pores. Such trinuclear Cu structures have been reported for eCO₂RR toward C₂H₄ [29, 30]. However, the discrete trinuclear Cu clusters, which act as the sole active sites, reconstruct into larger Cu nanoparticles during eCO₂RR, leading to inferior catalytic performance.

The successful grafting of organic ligands (Bzz and Iza) onto the MOF‐808 framework is initially confirmed by FT‐IR spectroscopy. For Bzz‐M808, two new characteristic bands emerge at 1290 and 1020 cm⁻¹ (Figure 1d), assigned to the aromatic C─N stretching vibration [31, 32] and the C2─H stretching vibration of the imidazole ring [33], respectively. In contrast, Iza‐M808 shows a characteristic C–N stretching vibration of the indazole ring at 1209 cm⁻¹, along with typical N–H stretching vibrations above 3209 cm⁻¹ (Figure S3). Further evidence for the successful ligand substitution is provided by ¹H nuclear magnetic resonance (NMR) analysis of digested samples (Figures S4 and S5). Upon metalation, the intensity of the N‐H stretching vibrations in Iza‐M808 markedly decreases, indicating Cu coordination to indazole nitrogen. For the Bzz‐derived framework, the transformation into a metal‐carbene framework (Cu–N–NHC–M808) is monitored by tracking the imidazolium C–H vibrations. Notably, the C2–H stretching vibration at 1020 cm⁻¹ attenuates after metalation (Figure 1d), providing direct spectroscopic support for C2 deprotonation and formation of neutral NHC‐Cu species [33]. Solid‐state ¹³C NMR (SS NMR) spectroscopy further corroborates this: Bzz‐M808 exhibits a distinct resonance at 170 ppm for the imidazolium C2 carbon (Figure 1c), which shifts downfield to 174–178 ppm after metalation. This range is a hallmark of carbene carbon in NHC–Cu complexes [18]. This downfield shift confirms the formation of a covalent Cu–C bond and the construction of NHC–Cu active sites in the MOF matrix.

The morphology of the as‐synthesized MOFs is clearly observed using scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Both MOF‐derived catalysts exhibit uniform nanoparticles with a consistent size distribution centered around 40 nm (Figure 1e–g, Figures S6–S8). Energy‐dispersive x‐ray spectroscopy (EDX) elemental mapping reveals homogeneous distribution of Cu, C, N, O, and Zr throughout the MOF, confirming successful incorporation of Cu species and Bzz/Iza ligands. High‐angle annular dark‐field scanning TEM (HAADF‐STEM) images of Cu‐N‐NHC‐M808 (Figure 1f,g) reveal numerous discrete bright spots attributable to Zr and Cu atoms. Although distinguishing between Zr (Z = 40) and Cu (Z = 29) by contrast alone is challenging due to their similar atomic numbers [34], the absence of observable nanoparticles suggests that the Cu species exist predominantly as atomically dispersed sites rather than aggregated metallic Cu nanoparticles. X‐ray photoelectron spectroscopy (XPS) is used to probe the surface electronic states and elucidate the coordination environment of the metal centers. As depicted in Figure 2b and Figure S9, the N 1s spectra of both metalated samples are highly similar and exhibit a new peak at approximately 400.0 eV, which is assigned to the Cu‐coordinated N signal [35, 36]. For Cu–N–NHC–M808, the contributions at 399.1 eV and 400.9 eV correspond to the C═N─C and C─N─C moieties of the NHC ligand, respectively [37]. Compared to the un‐metalated framework, the N 1s peaks shift positively by ca. 0.4 eV (from 398.7 to 399.1 eV and from 400.5 to 400.9 eV), confirming effective Cu–N coordination.

FIGURE 2.

(a) Cu 2p spectra of Cu–N–NHC–M808 and Cu–N–M808. (b) N 1s spectra of Cu‐N‐NHC‐M808 and Bzz‐M808. (c) Cu K‐edge XANES spectra and (d) corresponding EXAFS spectra of Cu‐N‐NHC‐M808, Cu‐N‐M808, and Cu‐N‐NHC‐N901 with CuO, Cu₂O, CuPc, and Cu foil as references. (e) Cu K‐edge XANES spectra and (f) corresponding EXAFS spectra of Cu–N–NHC–M808 as a function of reaction time at −1.2 V versus RHE with CuO, Cu₂O, and Cu foil as references. (g) Wavelet transforms for the EXAFS spectra of Cu‐N‐NHC‐M808 as a function of reaction time at −1.2 V versus RHE with Cu foil as reference.

The local atomic coordination and oxidation states of Cu in Cu–N–NHC–M808 and Cu–N–M808 are determined by x‐ray absorption spectroscopy (XAS). The Cu K‐edge x‐ray absorption near‐edge structure (XANES) spectra of Cu–N–NHC–M808 (Figure 2c) exhibit a weak absorption feature near 8978 eV, corresponding to the 1s → 3d electronic transition characteristic of Cu²⁺ [38]. A second distinct feature appears in the range of 8982–8984 eV, attributed to the 1s → 4p transition of Cu⁺ [39]. Additionally, a shoulder peak at 8985.5 eV is associated with the 1s → 4p transition of Cu²⁺ [40, 41]. These spectral characteristics indicate that Cu–N–NHC–M808 contains a mixture of Cu⁺ and Cu²⁺ species. In contrast, the absorption edge of Cu–N–M808 is located near that of CuO, with no discernible Cu⁺ 1s→4p transition, suggesting an average oxidation state close to +2. This conclusion is consistent with the Cu 2p XPS (Figure 2a). The local coordination environments are further probed by Fourier‐transformed extended x‐ray absorption fine structure (EXAFS) spectroscopy. As shown in Figure 2d,g and Figure S10, a prominent peak is observed at 1.5 Å in R‐space, which is ascribed to Cu–C/N/O scattering paths. A weaker peak observed at approximately 2.5 Å in Cu–N–NHC–M808 and at 3.0 Å in Cu–N–M808 corresponds to Cu–O–Cu and Cu–Cu bonds, respectively, indicating weak Cu‐Cu interactions in both catalysts. EXAFS fitting (Tables  S2 and  S3) gives coordination numbers (CNs) of Cu–N, Cu–C, and Cu–O–Cu that are approximately 1 in Cu–N–NHC–M808, confirming dinuclear Cu sites, in agreement with a previous report [22]. For Cu‐N‐M808, the CN of Cu–Nx is calculated to be approximately 2, demonstrating each Cu atom is coordinated by two N atoms. These resolved local structures align with the results derived from XPS, FT‐IR, and ¹³C SS NMR spectroscopy, validating the proposed Cu coordination configurations.

2.2. Electrochemical CO₂RR Performance

The eCO₂RR performance was evaluated in a three‐electrode gas diffusion H‐cell with CO₂‐saturated 0.1 M KHCO₃ as the electrolyte. All potentials were measured against an Ag/AgCl reference electrode and converted to the reversible hydrogen electrode (RHE) scale. Gaseous products were quantified and identified by gas chromatography (GC) (Figures S11 and S12), and liquid products were analyzed by ¹H NMR (Figure S13). To preliminarily evaluate the CO₂RR activity of the catalysts, linear sweep voltammetry (LSV) tests were performed under both CO₂‐ and Ar‐saturated electrolytes. As depicted in Figure 3a, both Cu‐N‐NHC‐M808 and Cu‐N‐M808 exhibit higher current densities as well as lower onset potential responses under CO₂ atmosphere compared to an Ar atmosphere. This distinct response confirms efficient CO₂ reduction catalyzed by these two MOF‐based catalysts. Control experiments with Bzz‐M808 and parent MOF‐808 in CO₂‐saturated 0.1 M KHCO₃ (Figure S14) show significantly lower current densities, underscoring the essential role of Cu active sites in promoting the CO₂RR.

FIGURE 3.

Electrochemical CO₂RR performance of different catalysts. (a) LSV curves of Cu‐N‐NHC‐M808, Cu‐N‐M808, and Cu‐N‐NHC‐N901 in CO₂‐ or Ar‐saturated 0.1 M KHCO₃ electrolyte. The Faradaic efficiencies of (b) Cu‐N‐NHC‐M808, (c) Cu‐N‐M808, and (d) Cu‐N‐NHC‐N901 for various products. The FE of (e) C₂H₄ and (f) total C₂₊ products at different applied potentials. (g) Catalytic stability test of Cu‐N‐NHC‐M808 at −1.2 V versus RHE. (h) Comparison of the C₂H₄ FE for Cu‐N‐NHC MOF with other reported Cu‐based MOF catalysts.

The product distribution of eCO₂RR was further evaluated by controlled‐potential electrolysis within a potential range of −1.0 to −1.5 V versus RHE. As shown in Figure 3b, C₂H₄ is the dominant product on Cu‐N‐NHC‐M808, reaching a maximum Faradaic efficiency (FE) of 61.0% at −1.2 V versus RHE, with a total C₂₊ FE of 79.5%. HER is effectively suppressed (FE < 10%). Notably, Cu‐N‐NHC‐M808 maintains high C₂H₄ selectivity (∼52.0%) over a broad potential window from −1.1 to −1.4 V. In contrast, Cu‐N‐M808 exhibits inferior performance, affording FEs of 50.5% for C₂H₄ and 70.0% for total C₂₊ products under the same conditions, highlighting the influence of the different Cu coordination environments. Pristine Cu nanoparticles (Figure S15) show moderate C₂H₄ selectivity but substantially higher CO and H₂ FEs than Cu‐N‐NHC‐M808. Mechanism investigation reveals that the superior performance of Cu‐N‐NHC‐M808 arises from synergistic interplay between NHC‐Cu(I) moiety and coexisting Cu nanoparticles, which suppresses HER and promotes C‐C coupling (vide infra). Control samples Bzz‐M808 (Figure S16a) and MOF‐808 (Figure S16b) produce mainly H₂. These results highlight the essential role of Cu active sites in eCO₂RR and explain the lower current densities observed for these controls in LSV.

To identify the carbon source of the products under operating conditions, ¹³C isotope‐labeling experiments were performed using gas chromatography‐mass spectrometry (GC‐MS). As shown in Figure S17, when ¹³CO₂ is employed as the feed gas, distinct signals at m/z = 30 (¹³C₂H₄) and m/z = 29 (fragments from ¹³CO/¹³CO₂) are detected, whereas only a signal at m/z = 28 appears with unlabeled ¹²CO₂. These results unambiguously confirm that the carbon in the reduction products originates from CO₂. Furthermore, in situ differential electrochemical mass spectrometry (DEMS) measurements were carried out in both CO₂‐saturated and Ar‐saturated electrolytes (Figure S18). Under CO₂, signals for H₂ (m/z = 2), CH₄ (m/z = 15), C₂H₄ (m/z = 26), and CO (m/z = 28) are detected, while no carbon‐containing products are detected under Ar. These results provide direct evidence that the carbon atoms derive exclusively from CO₂, ruling out contributions from electrolyte decomposition or other sources.

The stability of Cu‐N‐NHC‐M808 was evaluated at −1.2 V versus RHE. The catalyst exhibits exceptional stability, maintaining a C₂H₄ selectivity of > 52% over 9 h of electrolysis (Figure 3g). In contrast, Cu‐N‐M808, which lacks the NHC‐Cu(I) motif, shows lower initial FEs for C₂H₄ and total C_{2 +} products, and its C₂H₄ FE drops rapidly to 33% within 4 h, accompanied by a marked increase in HER (Figure S19). These results confirm that the NHC‐Cu(I) active site in Cu‐N‐NHC‐M808 not only enhances the selectivity toward C₂₊ products (especially C₂H₄) but also helps to stabilize key intermediates. The underlying mechanism is further analyzed by operando studies and DFT calculations in the following sections.

Considering the suitable pore size of NU‐901 (Figure S20), we extended our catalyst design to this framework to further verify the critical role of NHC‐Cu(I) in promoting C─C coupling. Following the same synthetic procedure used for Cu‐N‐NHC‐M808, we obtained Cu‐N‐NHC‐N901. The resulting material exhibits a rod‐like morphology with clear lattice fringes in HR‐TEM (Figure S21), indicating good crystallinity. Elemental mapping shows homogeneous distribution of Zr, Cu, C, N, and O, confirming successful incorporation of ligands and Cu species, and PXRD verifies phase purity (Figure S22a). Critically, the formation of NHC‐Cu motif is confirmed by multiple techniques: IR spectra show attenuation of the N‐H and imidazole C2‐H stretches (Figure S22b); XPS reveals a N 1s peak at 400.2 eV characteristic of Cu‐N coordination (Figure S23); XANES spectra (Figure 2c) display a 1s → 3d transition typical of Cu²⁺ species, along with 1s→4p transitions corresponding to both Cu⁺ and Cu²⁺, similar to those observed for Cu‐N‐NHC‐M808. These results indicate that Cu‐N‐NHC‐N901 also contains a mixture of Cu⁺ and Cu²⁺ species, corresponding to XPS results (Figure S39). Furthermore, EXAFS fitting gives CNs of ∼1.0 for both Cu‐N and Cu‐C and identifies a Cu‐O‐Cu scattering path at 2.5 Å with a CN of 1.3 (Figures S24 and S25, Table S4), supporting dinuclear Cu units. Electrocatalytic performance was evaluated under identical conditions. As depicted in Figure 3d, Cu‐N‐NHC‐N901 achieves FEs of 59.9% for C₂H₄ and 84.1% for total C₂₊ products at −1.3 V versus RHE. Stability tests show a C₂H₄ FE above 47.0% after 6 h, outperforming Cu‐N‐M808 (Figure S26). The Cu‐N‐NHC MOF catalysts developed in this work outperform most reported Cu‐based catalysts in terms of FEs for C₂H₄ and C₂₊ products, as highlighted in Figure 3h and summarized in Table S5. This comparison underscores the superior performance of the Cu‐N‐NHC MOF system for CO₂‐to‐C₂₊ conversion.

Electrochemical double‐layer capacitance (C _dl) measurements were performed to estimate the electrochemical active surface area (ECSA) (Figure S27a). Cu‐N‐NHC MOF exhibits a larger ECSA than Cu‐N MOF, indicating more accessible active sites and accounting for its enhanced CO₂RR performance. Furthermore, electrochemical impedance spectroscopy (EIS) was employed to probe the charge transfer characteristics. As displayed in Figure S27b, the Nyquist plots for all three catalysts exhibit small semicircular radii, indicating low charge transfer resistance (R _ct), which facilitates efficient electron transfer and promotes the overall reaction kinetics.

2.3. Structural Dynamics of Catalysts in eCO₂RR

The structural and morphological evolution of the catalysts after eCO₂RR was characterized by TEM, XRD, IR, XPS, and XAS. For Cu‐N‐NHC‐M808, HR‐TEM (Figure S28) reveals that the overall morphology remains largely intact after 1 h of electrolysis, with in situ‐generated Cu nanoparticles (average diameter: 1.3 nm) confined within the MOF framework. The corresponding PXRD patterns show no discernible diffraction peaks attributable to these nanoparticles, likely due to their small size and high dispersion within the framework (Figure S29). Upon prolonged electrolysis, diffraction features associated with the Cu(111) plane gradually emerge, while the characteristic MOF peaks remain visible, suggesting partial reconstruction and aggregation of Cu nanoparticles under operating conditions. This is further confirmed by HR‐TEM images (Figure S30), which reveal lattice fringes with a spacing of 0.208 nm, corresponding to the Cu(111) plane, validating the formation of Cu nanoparticles that serve as active sites alongside adjacent NHC‐Cu(I) during eCO₂RR. For Cu‐N‐M808, Cu nanoparticles are also observed after 1 h electrolysis with an average size of 3.8 nm (Figures S31 and S32). In contrast, post‐electrolytic Cu‐N‐NHC‐N901 largely retains its rod‐like morphology, with uniformly dispersed Cu(111) nanoparticles of ∼1.9 nm (Figures S33 and S34), which are smaller than those formed on Cu‐N‐M808. We attribute the restrained nanoparticle growth in the Cu‐N‐NHC MOFs to the presence of NHC‐Cu(I) site, which acts as a secondary active site and helps distribute the current, mitigating localized overgrowth. In contrast, Cu‐N‐M808 possesses only one type of active site, leading to electron conduction concentrated on the reconstructed Cu nanoparticles and thus accelerated growth. Additionally, the free NHC within MOF structure may further restrict the over‐aggregation of in situ‐generated Cu nanoparticles via lone‐pair donation to vacant Cu orbitals.

The IR spectra of the three catalysts after electrolysis exhibit largely unchanged characteristic peaks (Figure S35), indicating that their structural frameworks remain essentially intact under operating conditions. In situ Raman spectroscopy further confirms this, as all major spectral features persist when the potential is shifted from OCP to −1.5 V versus RHE (Figures  S36 and S37). Moreover, the XPS N1s spectra exhibit a consistent shift toward lower binding energy (Figures S38b–S40b), suggesting electron transfer from the in situ‐generated Cu nanoparticles to the organic ligands via Cu‐N interactions. The persistence of the characteristic peak at approximately 400 eV confirms that the newly formed Cu nanoparticles are stabilized by nitrogen atoms from the ligands, while also implying partial dissociation of the original Cu‐N coordination during electrolysis.

The oxidation states of Cu species after electrolysis were probed by XPS. After 1 h, Cu⁰ species are detected in all three materials, confirming the partial reduction and formation of Cu nanoparticles (Figures S38–S40). After prolonged electrolysis, Cu⁺ species are retained in Cu‐N‐NHC‐M808 due to the NHC‐Cu(I) moiety, semi‐quantitative peak fitting of the XPS Auger spectra reveals a Cu⁺:Cu⁰ ratio of 1:4.58, whereas Cu‐N‐M808 is almost completely reduced to Cu⁰ (Figure S41). XAS was used to monitor the evolution of the local coordination environment and electronic structure during eCO₂RR. For Cu‐N‐NHC‐M808 (Figure 2e), the 1s → 3d pre‐edge peak at ∼8978 eV vanishes upon negative potential application, confirming the reduction of Cu²⁺ species [42]. The Cu K‐edge XANES spectra exhibit a progressive negative shift with electrolysis time, signifying a gradual decrease in the average Cu oxidation state, which eventually stabilizes between Cu⁰ and Cu⁺. From the first derivative of the XANES spectra (Figure S42a) and linear fitting, the average Cu valence state (Figure S42b) is determined to be approximately 0.15, corresponding to a ratio of intact NHC‐Cu(I) site to aggregated Cu nanoparticles of 1:5.67. This value is consistent with our optimized structural model (Figure S43) and, despite a slight difference from the XPS surface result (1:4.58), reflects the expected bulk composition. The configuration evolution of Cu is further characterized by EXAFS analysis (Figure 2f, Figure S45, and Table S2). The relative intensity of the Cu‐Cu coordination shell increases markedly with electrolysis time, while the contributions from Cu‐O/N coordination gradually diminish. In the early reaction stage (0‐2 h), the CN of Cu‐Cu increases from 0 to 1.6, accompanied by a decrease in Cu‐O and Cu‐N CNs, suggesting the formation of Cu clusters. Upon prolonged electrolysis, the Cu‐Cu peak further intensifies, with the CN rising significantly to 4.8 after 5 h and 5.9 after 9 h, indicating progressive aggregation of Cu nanoparticles. Concurrently, Cu–O and Cu–N CNs decrease to 0.7. Notably, the Cu–C coordination remains intact even after prolonged electrolysis, underscoring the robustness of the NHC–Cu(I) unit. This evolution gives rise to a hybrid structure of metallic Cu nanoparticles coexisting with stable NHC–Cu(I). Wavelet‐transformed (WT) analysis (Figure 2g) further differentiates contributions from Cu nanoclusters and Cu–C species. The WT contour plots show that the k‐space maximum of the Cu–Cu scattering path progressively shifts to higher values with increasing electrolysis time, reflecting dynamic evolution. Concurrently, the intensity of the Cu–O/N/C path gradually diminishes but remains visible after long‐term electrolysis, attributable to the persistent Cu–C coordination [43].

Time‐dependent operando XAS was also employed to monitor the structural evolution of Cu‐N‐NHC‐M808 (Figure S46). Due to the high resistance of the in situ electrolytic cell, the current density achieved at the target potential is far below practical levels, substantially slowing catalyst reconstruction [23]. Consequently, prolonged electrolysis was required to capture the structural changes. Despite this, XANES reveals a shift of the absorption edge toward lower energies relative to open‐circuit potential (OCP), indicating a decrease in the average Cu oxidation. Concurrently, EXAFS exhibits a gradual attenuation of the Cu–N/O/C scattering contributions, accompanied by an increase in the Cu–Cu peak intensity over the extended measurement period. These trends are consistent with the ex‐situ XAS observations and confirm the formation of Cu nanoparticles during the electrocatalytic process. The stability of the NHC‐Cu(I) unit after long‐term electrolysis is further corroborated by ¹³C SS NMR spectroscopy (Figure S47). The C2 carbon of the NHC‐Cu(I) moiety exhibits a chemical shift at 172.8 ppm; the slight deviation from its initial value is ascribed to the subtle changes in the local chemical environment resulting from partial structural reorganization, which weakens the π‐backdonation and consequently reduces the paramagnetic deshielding effect.

For Cu‐N‐M808, time‐dependent XANES spectra reveal a progressive shift in the Cu oxidation state from +2 to 0 with increasing electrolysis time (Figure S48a). Correspondingly, EXAFS analysis shows a gradual increase in the Cu‐Cu CN, accompanied by a decrease in the Cu‐O and Cu‐N CNs (Figure S45b and S48b, Table S3), indicating progressive reconstruction of the Cu‐O/N coordination environment and the gradual formation of Cu nanoparticles. This trend mirrors that of Cu‐N‐NHC‐M808, but with a more pronounced evolution of the Cu‐Cu shell and a more drastic decrease in Cu‐O/N coordination after prolonged electrolysis, as evidenced by EXAFS and WT analysis (Figure S49). This observation is also consistent with the optimized structural model (Figure S44).

2.4. Understanding the CO₂RR Mechanism

To elucidate the reaction mechanism of the eCO₂RR, operando attenuated total reflectance Fourier transform infrared (ATR‐FTIR) spectroscopy measurements were performed. As shown in Figure 4, the peak observed at 2051 cm⁻¹ corresponds to a top‐adsorbed *CO [44], and the band at 1620 cm⁻¹ to the bending vibration of adsorbed water. The peak at 1370 cm⁻¹ is assigned to the *COOH intermediate, involved in *CO formation. Additionally, peaks located at 1477 and 1060 cm⁻¹ are assigned to *CHO [45] and *COH [46] respectively, suggesting a potential hydrogenation pathway from *CO to these species. Consequently, the C–C coupling could be triggered between *CO and *CHO/*COH. Other spectral features associated with C–C coupling are also detected. Specifically, a characteristic peak near 1560 cm⁻¹ can be classified as *COCHO (or *COCOH) intermediate, which plays a critical role in C₂H₄ production [47, 48]. These spectral features are observed across all three catalysts and intensify with increasingly negative applied potentials, consistent with their high C₂H₄ selectivity. Critically, at more negative applied potentials, the spectroscopic features on the Cu–N–NHC MOF are markedly more intense than those on Cu‐N MOF (Figure 4a–c). The integrated intensity of the *COCHO band is normalized relative to its value at −0.2 V versus RHE (Figure 4i). The normalized area for Cu–N–NHC MOF is substantially larger than that for Cu–N MOF, revealing that the Cu–N–NHC MOF offers a more favorable local reaction environment, which facilitates the formation and stabilization of key intermediates involved in the pathway toward C₂H₄ production [49].

FIGURE 4.

In situ ATR‐SEIRAS characterization during CO₂RR. (a–c) Spectra collected at different applied potentials on Cu‐N‐NHC‐N901 (a), Cu–N–M808 (b), and Cu–N–NHC‐M808 (c). (d–f) Corresponding spectra in the O–H stretching region for Cu‐N‐NHC‐N901 (d), Cu‐N‐M808 (e), and Cu–N–NHC–M808 (f). (g) Relative area of the deconvoluted peaks for strongly H‐bonded water, weakly H‐bonded water, and free water at −1.2 V versus RHE over the three catalysts. (h) Deconvolution of the O–H stretching band at −1.2 V versus RHE for the three catalysts. (i) Normalized peak area of *COCHO as a function of the applied potential for the three catalysts.

The aqueous microenvironment at the electrode‐electrolyte interface of Cu–N–NHC MOF and Cu–N MOF was further probed by in situ ATR‐SEIRAS under applied potentials (Figure 4d–f). The O–H stretching region (3000–3800 cm⁻¹) reveals three distinct hydrogen‐bonding environments: free water (∼3620 cm⁻¹), weakly H‐bonded water (∼3480 cm⁻¹), and strongly H‐bonded water (∼3300 cm⁻¹) [14, 50, 51, 52]. All catalysts exhibit an increase in the intensity of all three water bands as the potential becomes more negative, with a particularly pronounced increase in the proportion of weakly and strongly H‐bonded water. At more negative potentials, Cu–N–NHC MOF exhibits a greater abundance of strongly H‐bonded water and an overall redshift in band maxima compared to Cu‐N MOF. Deconvolution at −1.2 V versus RHE (Figure 4g,h) quantifies the contributions: strongly H‐bonded water dominates (> 50%) in all samples. Cu‐N‐NHC MOF shows a lower contribution from free water (< 5%) than Cu–N MOF (∼10%), and a higher contribution from strongly H‐bonded water. This indicates that water molecules are more effectively confined within a robust hydrogen‐bonding network at the Cu–N–NHC MOF interface. Combined with the CO₂RR performance (Figure 3), we conclude that the NHC moiety facilitates the formation of a structured hydrogen‐bonding environment at the electrode‐electrolyte interface that suppresses HER and thereby steers the product distribution of CO₂RR [51, 53, 54, 55].

DFT calculations were performed to evaluate the free energy changes of reaction intermediates for Cu–N–NHC MOF and Cu–N MOF. In accordance with XAS, TEM, XRD, and other characterization results, both catalysts undergo in situ reconstruction during electrolysis, transforming CuO_xN_y species into Cu(111) crystals. Accordingly, we constructed a catalytic model comprising Bzz or Iza ligands partially coordinated on a Cu(111) substrate. As shown in Figure 5a, the rate‐determining step (RDS) for both catalysts is the conversion of *CO₂ to *COOH, with a free energy change (ΔG) of 1.13 eV for Cu‐N‐NHC MOF and 1.68 eV for Cu‐N MOF. This 0.55 eV lower barrier on Cu‐N‐NHC MOF aligns with the stronger *COOH signal at 1370 cm^{− 1} observed by in situ FTIR (Figure 4a,c), confirming that the NHC unit enhances CO2 activation on the Cu(111) surface.

FIGURE 5.

(a) Free energy profiles of C₂H₄ formation on Cu–N–NHC MOF and Cu‐N MOF. (b) Free energy changes for the key reaction steps on Cu–N–NHC MOF. (c) Free energy diagram of H₂O dissociation on Cu–N–NHC MOF and on Cu–N MOF. (d) Formation rates (H₂ and D₂) and KIE of H/D (H₂/D₂) of Cu–N–NHC–M808, Cu–N–M808, and Bzz‐M808. (e) Charge density difference plots for Cu–N–NHC MOF (up) and Cu–N MOF (bottom); blue area for charge depletion and yellow area for charge accumulation.

Based on operando ATR‐FTIR observations, *CO intermediates preferentially undergo asymmetric coupling with *CHO or *COH rather than symmetric CO‐CO coupling on all three catalysts. DFT calculations for Cu–N–NHC MOF (Figure 5b) give a ΔG of 1.03 eV for symmetric coupling, and ΔG values of −0.11 and 0.00 eV for the asymmetric CO─CHO and CO─COH pathways, respectively, indicating that the asymmetric *CO–*CHO pathway dominates. The *CO hydrogenation steps to *CHO and *COH (Figure  S50) have barriers of 0.71 and 0.95 eV, respectively, confirming a thermodynamic preference for *CHO formation. The asymmetric C─C coupling barrier along the *CHO‐to‐C₂H₄ pathway is substantially lower on Cu‐N‐NHC MOF (−0.11 eV) than on Cu‐N MOF (0.25 eV) (Figure 5a), highlighting the role of the NHC moiety in promoting asymmetric C─C coupling.

Furthermore, the NHC‐Cu site lowers the energy barrier for *CO hydrogenation to *CHO (0.71 eV) relative to slab Cu(111) (1.00 eV) (Figure S51b). This enhancement is attributed to the adjacent imidazole –NH groups, which serve as proton donors to promote *CO hydrogenation at the NHC‐Cu(I) site. In situ FTIR spectroscopy reveals that NHC facilitates the formation of a strong hydrogen‐bonding network with water. We propose that this network elongates the O‐H bond in water, thereby facilitating proton transfer and promoting *CO hydrogenation [56].

To further verify the role of NHC in regulating water dissociation, we calculated the dissociation energy of H₂O. As shown in Figure 5c, Cu–N–NHC MOF exhibits a more negative ΔG for H₂O dissociation, indicating that the NHC site promotes interfacial water dissociation to generate *H species, which enhance *CO hydrogenation to *CHO and suppress the competing HER. Figure S52 proposes a detailed proton‐transfer mechanism: the ‐NH group on the NHC motif forms a strong hydrogen bond with an adjacent H₂O molecule; proton transfer from –NH to the oxygen atom of H₂O generates a hydronium ion (H₃O⁺), which then donates a proton to *CO to form *CHO. The deprotonated nitrogen then acts as a proton acceptor, reforming hydrogen bonds with incoming H₂O molecules. The N site further elongates the O–H bond of coordinated H₂O molecules, sustaining continuous water dissociation and proton shuttling.

To experimentally validate the proton‐transfer mechanism proposed in Figure S52, we investigated the kinetic isotope effect (KIE) of Bzz‐M808, Cu–N–NHC–M808, and Cu–N–M808 by replacing H₂O with D₂O (Figure 5d). Bzz‐M808 exhibits a KIE of 4.34 for H₂ evolution, indicating that water dissociation is the rate‐determining step for H₂ production and ‐NH group facilitates proton hopping via a Grotthuss mechanism [31]. The KIE value for H₂ evolution of Cu‐N‐NHC‐M808 (1.99) is higher than that of Cu–N–M808 (1.77), suggesting that the NHC ligand promotes water dissociation more effectively, generating a larger flux of *H species through the Grotthuss pathway. For C₂H₄ formation, the KIE values are close to 1 (1.18 for Cu–N–NHC–M808 vs. 1.30 for Cu–N–M808) (Figure S53), demonstrating that water dissociation is not the rate‐determining step for C₂H₄ production [57]. The lower KIE on Cu–N–NHC–M808 implies faster proton transfer to *CO on the NHC‐containing catalyst, rendering the hydrogenation step less sensitive to isotopic substitution. This finding is consistent with our theoretical calculations showing that the NHC ligand facilitates water dissociation [58].

The *CHO intermediate may also undergo hydrogenation with *H to form *CHOH, a key intermediate leading to CH₄. However, the ΔG for this pathway is obviously higher than that for C─C coupling (Figure S54), indicating that the C₂H₄ formation route is thermodynamically favored over methane, consistent with experimental product selectivity. Bader charge analysis was performed to evaluate the charge transfer between the organic ligands (Bzz or Iza) and the Cu(111) support. As shown in Figure 5e, the calculated charge‐density distributions reveal that Cu‐N‐NHC MOF donates fewer electrons to the ligands than Cu‐N MOF, resulting in a higher electron density on the Cu(111) surface, which favors the adsorption and activation of key intermediates, thereby lowering the activation energy barrier for RDS. To further elucidate the electronic structure of the catalyst upon adsorption of key intermediates and to understand the synergistic effect between NHC‐Cu(I) site and Cu nanoparticles in promoting C–C coupling, we quantified the electron transfer in the three catalytic models. As summarized in Figure S55, the electrons transferred from Cu‐N‐NHC MOF to adsorbed *CO (0.35 e), *CHO (0.42 e), and *COCHO (0.76 e) are significantly greater than those from Cu─N MOF and Cu(111). This demonstrates that the NHC‐Cu site not only facilitates the formation and adsorption of *CO and *CHO but also cooperates with Cu nanoparticles to promote asymmetric C─C coupling. Moreover, the carbon atom of these intermediates bonded to the NHC–Cu site exhibits a pronounced electron environment (highlighted by the red dashed circle), further underscoring the role of enhanced electronic interactions in stabilizing reaction intermediates. We also tracked the Bader charge evolution at the NHC‐Cu site during *CO hydrogenation and C─C coupling (Figure S56 and Table S6). During *CO hydrogenation, the Bader charge of Cu increases from 10.99 to 11.00 e, whereas that of the adjacent carbon atom decreases from 3.67 to 3.64 e. Similarly, in the C─C coupling process, the Cu charge increases from 10.57 to 10.67 e, accompanied by a decrease in the carbon charge from 3.69 to 3.64 e. This consistent electron transfer from carbon to Cu enhances the adsorption and stabilization of key intermediates on the NHC‐Cu site and consequently lowers the overall reaction energy barrier. Partial density of states (PDOS) analysis reveals distinct electronic structures for the two catalysts (Figure S57). The Cu d‐band centers of Cu‐N‐NHC MOF and Cu‐N MOF are located at −2.35 and −2.39 eV, respectively. Upon adsorption of the *COCHO intermediate, these values shift to −2.33 and −2.38 eV. Notably, the d‐band center of Cu‐N‐NHC MOF lies closer to the Fermi level than that of Cu‐N MOF, indicating stronger interaction with reaction intermediates, which contributes to the enhanced eCO₂RR performance of Cu‐N‐NHC MOF.

3. Conclusion

In summary, we have successfully constructed a heterogeneous catalyst via PSM of MOF‐808, incorporating both bis(μ‐oxo) dicopper sites and NHC‐anchored Cu(I) sites. Under electrolysis conditions, the bis(μ‐oxo) dicopper sites undergo in situ‐reconstruction into metallic Cu nanoparticles, which function as synergistic active sites together with the stable NHC‐anchored Cu(I) site. The resulting catalyst, Cu‐N‐NHC‐M808, exhibits outstanding performance in the eCO₂RR to C₂H₄ in a neutral electrolyte, achieving a maximum C₂H₄ FE of 61.0% and a total C₂₊ product FE of 79.5% at −1.2 V versus RHE. Operando ATR‐FTIR spectroscopy, KIE measurement, and DFT calculations collectively demonstrate that the NHC site facilitates proton transfer by forming a robust hydrogen‐bonding network with interfacial water molecules. This interaction effectively lowers the energy barrier for *CO hydrogenation to *CHO at the NHC‐Cu site, thereby promoting subsequent asymmetric C–C coupling with *CO adsorbed on adjacent Cu nanoparticles to form the *COCHO intermediate. Moreover, the σ‐donating nature of the NHC ligand enhances the local electron density of the neighboring Cu(I) atom, stabilizing key intermediates such as *CHO and *COCHO through stronger electronic interactions, and concurrently improving the overall electrocatalytic stability of the framework. In contrast, the control catalyst Cu‐N‐M808, prepared using Iza ligands instead of Bzz and lacking NHC‐Cu(I) site, shows significantly lower C₂H₄ selectivity and stability. This work establishes a versatile strategy for constructing synergistic active sites inside MOFs, combining a molecular NHC‐Cu(I) unit with in situ‐derived Cu nanoparticles to steer the eCO₂RR pathway selectively toward C₂H₄ while boosting catalytic durability. This approach opens new avenues for rational design of advanced synergistic electrocatalysts, paving the way for energy‐efficient CO₂ conversion into value‐added hydrocarbons.

Author Contributions

Li‐Hong Jia: conceptualization, data curation, investigation, writing – original draft. Xian‐Ming Zhang: conceptualization, validation, supervision, funding acquisition, project administration, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: anie73158‐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.