English

• Cu-based metal-organic frameworks (MOFs) have been recognized as promising catalysts in the CO₂ reduction reaction (CO₂RR) catalysts for generating of C₂₊ products [1–7]. It has been widely accepted that Cu-based MOFs usually undergo in-situ reconstruction, yielding Cu-base derivatives, such as Cu, Cu₂O, CuO species and their heterostructures [8–11]. These generated derivatives are identified as paramount active sites for facilitating C—C coupling in CO₂RR [11,12]. However, in many cases, distinct MOF precursors can reconstruct the same Cu-based derivatives, but leading to significantly different catalytic performances [13–15]. For example, Cu^II/adeninato/carboxylato MOF (Cu^II/ade-MOF) nanosheets were converted to Cu nanoparticles, which obtained C₂H₄ with a maximum faradaic efficiency (FE) of 45% [16]. In another case, similar Cu nanoparticles were derived from semi-conductive MOF-Cu₃(HITP)₂, where a notable production C₂H₄ was observed attaining its maximum FE of 63% [17]. The phenomenon indicates that, before the C—C coupling process, there are other activation sites which critically influence catalytic efficiency.

  Acutually, the structures of pristine MOF precursors still remain throughout the catalytic process, particularly in prolonged catalytic cases. In essence, pristine MOFs also harbor the capability to impact the CO₂RR process, even though they are commonly overlooked in discussions of the catalytic mechanism. We posit that the varied catalytic outcomes stemming from identical Cu-based derivatives are intricately linked to the presence of pristine MOFs. Unveiling the role of pristine MOFs can not only provide a new direction to design high-proformance MOF-based CO₂RR catalysts with minimize trial and error.

  Constructing a rational catalytic system to explore the role of prinstine MOF precursors is still a significant challenge. On the one hand, to avoid variations in the C—C coupling efficiency in the second step, the derivatives produced by different Cu-based MOFs during the catalytic reaction should be completely identical [18,19]. On the other hand, to mitigate the influence of MOF's residual ligands, the prinstine MOFs chosen for comparsion should prossess the same chemical component but differet structures [20]. Last but not lesat, the MOFs catalysts should maintain the srtucture during the catalytic process, forming MOF-derivatives complexes rather than being entirely destroyed [21]. Therefore, to invesitigate a relationship between precursors and catalytic performance, it was essential to ensure structural similarity and stability as well as ligand consistency.

  Here, two Cu(I) imidazole framework isomers with the identical formula [Cu(imidazole)]_n were employed as the precursors, namely CuN₂ and Cu₂N₄ based on the coordination status. The only difference between the two precursors was the Cu coordination environments (Scheme 1). In the CuN₂ precursor, one Cu atom was linearly coordinated to N atoms of two imidazoles to form zig-zag one-dimensional chain. In the Cu₂N₄ precursor, two one-dimensional chains was perpendicularly stacked with a Cu-Cu bond formed bewteen the atoms of adjacent layers. The two Cu-based MOFs were both in-situ reconstructed to nanosized Cu(111) with a large number of MOFs retained under cathodic potential, named CuN₂/Cu(111) and Cu₂N₄/Cu(111), respectively. The CO₂RR performance of CuN₂, Cu₂N₄, CuN₂/Cu(111) and Cu₂N₄/Cu(111) was evaluated, respectively. The results showed that the both CuN₂ and Cu₂N₄ precursors produced CO while both CuN₂/Cu(111) and Cu₂N₄/Cu(111) produced C₂₊ products. The CuN₂ and Cu₂N₄ precursors showed a maximum FE_CO of 43.1% and FE_CO of 26.2% for the production of CO, respectively. Then, the CuN₂/Cu(111) and Cu₂N₄/Cu(111) exhibited a final FE_C of 64.8% and 43.9%, coorespondingly. The FE_C was closely related to the FE_CO of the pristine MOF precursors. We demonstrated a auto-tandem catalysis mechanism to reveal the performance of the Cu(imidazole) system. The pristine Cu(imidazole) played an activiation function to convert CO₂ into CO, and Cu(111) catalyzed CO to C₂₊ products. This auto-tandem catalysis mechanism was confirmed by TPD-CO, in-situ ATR-SEIRAS, and DFT computation. These results conclusively illustrated that pristine CuN₂ with more exposed Cu-N sites showed lower CO adsorption energy and generated higher CO substrate concentration, accelerating the formation of C₂₊ products. This rational mechanism provides a promising way and solution for the rational design of Cu-MOF catalysts.

  Scheme 1

  

  Scheme 1.  The design of auto-tandem catalysts CuN₂/Cu(111) and Cu₂N₄/Cu(111) and the proposed CO₂RR catalytic mechanism.

  DownLoad: Full-Size Img PowerPoint

  The CuN₂ and Cu₂N₄ were prepared with the solvothermal method under different synthetic conditions according to the reported literature method [22,23]. The obtained yellowish and brownish powder was characterized with powder X-ray diffraction (PXRD), X-ray photoelectron spectroscopy (XPS), scanning electron microscopy (SEM), transmission electron microscopy (TEM), and high-resolution transmission electron microscopy (HRTEM). PXRD patterns showed that the synthetic materials were consistent with the simulated CuN₂ and Cu₂N₄, respectively. XPS results revealed that the oxidation state of CuN₂ was monovalent, and Cu₂N₄ exhibited primary monovalent and a little bit of bivalence due to additional Cu-Cu bonding (Fig. S1 in Supporting information). SEM images exhibited that both of them were block material piled up layer by layer, and the size reached the micrometer scale. The morphologies were also confirmed by HRTEM images. Besides, there had no crystal diffraction fringes from HRTEM images. The above characterizations revealed the pristine CuN₂ and Cu₂N₄ were successfully prepared and contained no Cu-based nanoparticles (Fig. S2 in Supporting information).

  According to the previously reported method [24], under an electric field, the Cu-MOF underwent these processes of CO₂ adsorption, Cu reduction, Cu-N coordinated bond breaking, Cu atom restructuring, Cu nanoparticle growth, and nanoparticle aggregation. Therefore, we reconstructed CuN₂ and Cu₂N₄ by performing 1-h electrolysis from −0.78 V to −1.28 V vs. RHE. The reconstructed electrocatalysts were named as CuN₂/Cu(111) and Cu₂N₄/Cu(111), respectively. During the reconstruction process, we collected the electrochemical data and analyzed the products. Therein, gaseous products were quantified by an online gas chromatograph (GC), and liquid products were quantified by ¹H nuclear magnetic resonance spectroscopy (¹H NMR). Meanwhile, formate was detected with a high-performance liquid chromatograph (HPLC) (see details in Supporting information).

  To evaluate the CO₂RR performance of CuN₂/Cu(111) and Cu₂N₄/Cu(111), we tested them in both H-cell and flow-cell. On the one hand, in H-cell, we measured linear sweep voltammetry (LSV) after 1-h electrolysis at −1.18 V vs. RHE in both CO₂ and N₂ saturated 0.1 mol/L KCl electrolyte (Fig. 1a). The CuN₂/Cu(111) exhibited lower onset potential and higher total current density (j_total) than Cu₂N₄/Cu(111) from −0.4 V to −1.2 V vs. RHE, illustrating better CO₂RR catalytic performance. Then, to choose applicable electrolyte, the two catalysts were tested in 0.1 mol/L KHCO₃, 0.5 mol/L KHCO₃ and 0.1 mol/L KCl electrolyte, respectively (Figs. S3 and S4 in Supporting information). The FE of various products for CuN₂/Cu(111) and Cu₂N₄/Cu(111) were calculated and compared. Obviously, the C₂₊ product selectivity in 0.1 mol/L KCl was far higher than 0.1 mol/L KHCO₃ and 0.5 mol/L KHCO_3. Therefore, we chose 0.1 mol/L KCl electrolyte as the test condition in this work. For CuN₂/Cu(111), the C₂₊ products (C₂H₄, C₂H₅OH, and n-C₃H₇OH) were detected from −0.88 V vs. RHE (Figs. S5 and S6 in Supporting information). The maximum FE_C reached 64.8%, including FE_C of 41.5%, FE_C of 18.2%, and FE_n-C of 5.1% (Fig. 1b), which was middle level among reported Cu-based materials (Table S1 in Supporting information) [25–31]. For Cu₂N₄/Cu(111), the C₂₊ products were detected from −0.98 V vs. RHE. The maximum FE_C just reached 43.9%, including FE_C of 27.1%, FE_C of 12.6%, and FE_n-C of 3.4% (Fig. 1c). During the operated potential window from −0.98 V to −1.28 V vs. RHE, the FE_C2+ of CuN₂/Cu(111) was higher than Cu₂N₄/Cu(111) (Fig. 1d). Besides, to evaluate the number of active sites of CuN₂/Cu(111) and Cu₂N₄/Cu(111), the electrical double-layer capacitor (C_dl) was measured in non-Faraday potential intervals with a different scan rate of cyclic voltammetry (CV) (Fig. S7 in Supporting information) [32–35]. The C_dl of CuN₂/Cu(111) was calculated as 1.02 mF/cm², higher than Cu₂N₄/Cu(111) of 0.32 mF/cm², indicating more active sites for CO₂RR (Fig. 1e). In addition, their stability was tested with online quantification of C₂H₄ (Fig. 1f). The FE_C and j_total were recorded during 6-h electrolysis. Both FE_C and j_total sharply increased from 5-min to 30-min electrolysis, illustrating the reconstruction of pristine CuN₂ and Cu₂N₄. Then, the FE_C2H4 almost maintained, about 40% of CuN₂/Cu(111) and 30% of Cu₂N₄/Cu(111), showing good stability. Cu(imidazole) catalysts exhibit better stability than Cu-O based MOFs during CO₂RR process due to the more stable C—N coordination bond based on the hard and soft acid and base (HSAB) theory. According to the above results, it was obvious to conclude that the CO₂RR performance of CuN₂/Cu(111) was better than Cu₂N₄/Cu(111).

  Figure 1

  

  Figure 1.  (a) LSV curves of CuN₂/Cu(111) and Cu₂N₄/Cu(111) in both CO₂ and N₂ saturated 0.1 mol/L KCl electrolyte. FEs of various products at different operated potentialsl for (b) Cu₂N₄/Cu(111) and (c) CuN₂/Cu(111), respectively. (d) The comparison of FE_C2+ for CuN₂/Cu(111) and Cu₂N₄/Cu(111) in H-cell. (e) C_dl measurements with CV method in non-Faraday potential interval under different scan rates. (f) Stability test at −1.18 V vs. RHE for 6-h electrolysis in H-cell. FEs of various products at different operated currents density in flow-cell for (g) Cu₂N₄/Cu(111), (h) CuN₂/Cu(111), respectively. (i) The comparison of FE_C2+ for CuN₂/Cu(111) and Cu₂N₄/Cu(111) in flow-cell.

  DownLoad: Full-Size Img PowerPoint

  On the other hand, the CO₂RR performance of CuN₂/Cu(111) and Cu₂N₄/Cu(111) was also tested in flow-cell (Figs. 1g-i). The FE_C of CuN₂/Cu(111) and Cu₂N₄/Cu(111) were better than H-cell, where maximum was 72.1% and 71.0% at −700 mA/cm², respectively. Within the operated current range, the performance of CuN₂/Cu(111) was better than Cu₂N₄/Cu(111). The observed phenomenon from flow-cell was similar as in H-cell. Nevertheless, the difference in C₂₊ performance between the two catalysts less obvious than in H-cell, especially at high current density, this was owing to the destruction of pristine materials from the PXRD patterns (Fig. S12 in Supporting information). Therefore, we subsequently conducted material characterization and mechanism investigation under H-cell conditions with more stable three-phase reaction interface.

  To reveal the actual catalytic sites during the above processes, the structure of CuN₂/Cu(111) and Cu₂N₄/Cu(111) was characterized by PXRD, XPS, Cu LMM Auger spectra, in-situ X-ray absorption near edge structure (XANES), in-situ X-ray emission spectroscopy (XES), HRTEM, and selected area electron diffraction (SAED), respectively (Fig. 2, Figs. S7, S8, S19 and S20 in Supporting information). PXRD patterns showed main diffraction peaks of pristine CuN₂ and Cu₂N₄ were retained (Fig. 2a), which was also observed from SEM images (Fig. S8). Meanwhile, the diffraction peaks of Cu(111) were detected at 43.2°. These results were further confirmed by HRTEM and SAED images (Figs. 2d and e). For CuN₂/Cu(111), HRTEM images showed the block of pristine CuN₂ piled up layer by layer, and some nanoparticles dispersed in it. The crystal spacing of these nanoparticles was measured as 0.209 nm of Cu(111). In addition, the bright diffraction spots of SAED images also confirmed that Cu(111) was a predominant crystal plane. The selective crystal plane derivatization may be attributed to the uniform Cu(I) distribution in the precursors. Furthermore, the Cu(111) plane was beneficial to promote the formation of C₂₊ products [36–38]. Similarly, the characterization and analysis for Cu₂N₄/Cu(111) were conducted (Fig. 2e). It showed pristine Cu₂N₄ was maintained with the formation of Cu(111) nanoparticles. It was worth noting that other very minor Cu species also existed in the two catalysts, such as 0.243 nm of Cu₂O(111). Therefore, the derivatives of CuN₂ and Cu₂N₄ were proved to be consistent, at the same time, the pristine materials were retained. Besides, the oxidation states of CuN₂/Cu(111) and Cu₂N₄/Cu(111) were investigated with Cu LMM, in-situ XANES and XES spectra. As shown in Fig. S9 (Supporting information), the main Cu LMM peak of both CuN₂/Cu(111) and Cu₂N₄/Cu(111) was recorded at the kinetic energy of 916.4 eV belonging to Cu(I), demonstrating the reservation of pristine CuN₂ and Cu₂N₄. Meanwhile, the Cu(0) peak with a kinetic energy of 918.8 eV was observed, indicating the reconstruction of Cu(imidazole). More convincingly, compared to pristine CuN₂ and Cu₂N₄, the in-situ XANES spectra of CuN₂/Cu(111) and Cu₂N₄/Cu(111) shifted to a low-energy region during the electrolysis, showing that the oxidation state of Cu(I) was reduced to Cu(0~I) (Fig. 2b). Besides, in-situ XES also exhibited the Cu oxidation state of pristine CuN₂ and Cu₂N₄ were reduced from OCP to −1.3 V vs. RHE (Fig. 2c). In especial, the reconstruction of CuN₂ was faster than Cu₂N_4, and the mixed Cu(0~I) was more than Cu₂N_4, these were what made the better catalytic performance of derived CuN₂/Cu(111). Besides, both in-situ XANES and XES results demonstrated that pristine CuN₂ and Cu₂N₄ did not fully converted to metallic Cu. Therefore, the pristine materials played an important role in these auto-tandem catalysis. Furthermore, to demonstrate the stability of the catalysts, the structure of CuN₂/Cu(111) and Cu₂N₄/Cu(111) after the 6-h stability test was analyzed. PXRD patterns showed the main diffraction peaks of pristine CuN₂ and Cu₂N₄ were still retained (Fig. S10 in Supporting information). The diffraction peak of the Cu(111) derivative was observed, and the intensity was stronger than the materials after 1-h electrolysis, indicating a continuous reconstruction process during the stability test. The Cu(111) crystal plane was also identified from HRTEM and SAED images of CuN₂/Cu(111) and Cu₂N₄/Cu(111) (Fig. S11 in Supporting information).

  Figure 2

  

  Figure 2.  The characterization of CuN₂/Cu(111) and Cu₂N₄/Cu(111) compared with pristine CuN₂ and Cu₂N₄. (a) PXRD patterns. (b) In-situ XANES spectra. The inset is enlarged part from 8975 eV to 8980 eV. (c) In-situ Cu Kβ_{1, 3} XES spectra from OCP to −1.3 V vs. RHE. OCP was circuit potential. The microscopic analysis of (d) CuN₂/Cu(111) and (e) Cu₂N₄/Cu(111). The inset is HRTEM and SAED images.

  DownLoad: Full-Size Img PowerPoint

  According to the above structural characterizations, both the pristine Cu MOFs and the same derived Cu(111) existed during the catalytic process. Based on their different catalytic results, it was reasonable to speculate that both pristine Cu MOFs and derived Cu(111) acted as catalysts during the CO₂RR process. As Cu(111) usually reported for the C—C coupling to forming C₂₊ products, we supposed that the pristine Cu MOFs played an activation function to active CO₂, forming active substance for the next step of C—C coupling. This catalysis system with two steps was named as the auto-tandem catalysis system. The different catalytic performance of CuN₂/Cu(111) and Cu₂N₄/Cu(111) was possible came from the variation of coordination environment. Particularly, the Cu atom of pristine CuN₂ was just coordinated with two N atoms, while for Cu₂N₄, Cu was not only coordinated with two N atoms but also bonded with the other Cu atom of the adjacent layer. Therefore, we legitimately attributed the different catalytic performances to the different coordination environments of the pristine materials.

  To reveal the role of coordination environment in MOF precursor, we first investigated the performance of sole CuN₂ and Cu₂N₄ by LSV and FE measurements, respectively. The LSV curve of pristine CuN₂ also showed lower onset potential and higher j_total than Cu₂N₄, indicating better catalytic performance (Fig. 3a). Then, to avoid catalyst reconstruction, short CO₂RR experiments with 50-s electrolysis were performed from −0.88 V to −1.18 V vs. RHE for CuN₂ and Cu₂N_4, and respective CO₂RR products were quantified. There were no C₂₊ products, and just CO and H₂ was detected for both CuN₂ and Cu₂N₄ (Fig. S13 in Supporting information). For pristine CuN₂, CO product was detected at −0.88 V vs. RHE, the FE_CO increased with potentials, and the maximum FE_CO reached up to 43.1% (Fig. 3b), which lies a medium level in the Cu-based materials with single metal site (Table S2 in Supporting information). Meanwhile, for pristine Cu₂N₄, the maximum FE_CO was only 24.4% (Fig. 3b). Obviously, the CO selectivity of pristine CuN₂ was much better than Cu₂N₄ during the whole operated potential window. These results illustrated the CO₂RR performance of pristine CuN₂, including onset potential and FE_CO, was better than Cu₂N₄.

  Figure 3

  

  Figure 3.  (a) LSV curves of pristine CuN₂ and Cu₂N₄ in both CO₂- and N₂-saturated 0.1 mol/L KCl electrolyte. (b) FE_CO at different potentials with 50-s electrolysis for pristine CuN₂ and Cu₂N₄, respectively. (c) FEco for the pristine materials and FE_C2+ for the reconstructed materials at various potentials. (d) After 8-h electrolysis tests for CuN₂ and Cu₂N₄ at −1.18 V vs. RHE, the pristine MOF was reduced to Cu(111), which was immediately tested to obtain FE_CO. (e) Relative rates of CO₂ reduction to CO vs. CO reduction to C₂₊ products for CuN₂ and Cu₂N₄ after 8-h electrolysis at −1.18 V vs. RHE. (f) CORR of CuN₂/Cu(111) and Cu₂N₄/Cu(111) at various CO volume concentrations with 1-h electrolysis under fixed −1.18 V vs. RHE.

  DownLoad: Full-Size Img PowerPoint

  To confirm the catalyst structure after 50-s electrolysis, the HRTEM, SEM, and PXRD characterizations were executed (Figs. S14-S16 in Supporting information). Both CuN₂ and Cu₂N₄ still exhibited pristine structures, and no Cu-based nanoparticles were observed. Therefore, the above catalytic performance factually reflected the ability of the pristine CuN₂ and Cu₂N₄ to convert CO₂. To establish a connection between MOF and Cu(111) in the auto-tandem catalysis system, we made comparison plots of FE_CO for the pristine materials and FE_C for the reconstructed materials at various potentials. As shown in Fig. 3c, both FE_C of CuN₂/Cu(111) and FE_co of CuN₂ were higher than Cu₂N₄/Cu(111) and Cu₂N₄, respectively. Therefore, the FE_C for reconstructed materials exhibited a positive correlation with FE_CO for the pristine materials.

  According to some literatures, derived Cu metal was also known to convert CO₂ into C₂₊ products through CO as an intermediate. Therefore, in the auto-tandem catalysis system, it is very important to determine which of pristine MOF and the derived Cu(111) is the main active site for converting CO₂ to CO. On one hand, pristine CuN₂ and Cu₂N₄ were destructed under 8-h electrolysis tests at −1.18 V vs. RHE, which was fully reduced to Cu(111) from PXRD and SEM analysis (Fig. S17 in Supporting information). Subsequently, the derived Cu(111) was immediately tested to obtain FE_CO (Fig. 3d). After destruction, the derived Cu(111) exhibited lower FE_CO than pristine MOF from −0.98 V to −1.18 V vs. RHE. Especially at the potential of C₂₊ products, pristine CuN₂ and Cu₂N₄ played dominant active sites of converting CO₂ to CO. On the other hand, to explore the relationship between the two steps in the auto-tandem catalysis system, we compared the relative rates of CO₂ reduction to CO vs. CO reduction to C₂₊ products for the derived Cu(111) at −1.18 V vs. RHE. (Fig. 3e). All of the rates were around 2 × 10⁻⁷ mol min⁻¹ cm⁻². The former is not much faster than the latter. Therefore, improving the rate of CO₂ reduction to CO may promote C₂₊ products selectivity due to the rate-determining step of C—C coupling.

  To further investigate the effect of the substrate CO on the C₂₊ product, control experiments of CO reduction reaction (CORR) with various CO volume concentrations were performed on CuN₂/Cu(111) and Cu₂N₄/Cu(111), respectively. The CO/N₂ mixture with CO volume concentrations of 0.2, 0.4, and 1 was employed instead of pure CO₂ for electrolysis. At these conditions, the CORR performance of CuN₂/Cu(111) and Cu₂N₄/Cu(111) was evaluated at a potential of −1.18 V vs. RHE. As shown in Fig. 3f, for the CO volume concentration of 0.2, 0.4, and 1, FE_C of CuN₂/Cu(111) was obtained as 27.3%, 42.9%, and 77.3%, respectively, while FE_C of Cu₂N₄/Cu(111) was 24.3%, 45.6%, and 75.9%, respectively. Therefore, the FE_C of CuN₂/Cu(111) and Cu₂N₄/Cu(111) significantly increased with CO volume concentration, illustrating that the CO concentration directly determines the C₂₊ product formation. Meanwhile, the obtained FE_C of CuN₂/Cu(111) and Cu₂N₄/Cu(111) was almost identical under the same CO volume concentration, revealing that CORR bypassed the first CO₂ activation step and only depended on the same active site, in this case, the Cu(111). Therefore, these results further demonstrated the rationality of the proposed auto-tandem catalysis mechanism.

  To comprehensively reveal the role of pristine CuN₂ and Cu₂N₄ in the CO₂RR process, in-situ attenuated total reflection-surface enhanced infrared reflection absorption spectroscopy (in-situ ATR-SEIRAS) was conducted to monitor the intensity difference of *CO intermediate (Figs. 4a and b, Fig. S20) [39–44]. The IR spectrum was collected from open circuit potential (OCP) to −1.28 V vs. RHE. Notably, the *COOH intermediate was monitored at 1652 cm⁻¹ and 1657 cm⁻¹, revealing the activation of CO₂. According to the literature [40,45], the band located in the region from 2000 cm⁻¹ to 2100 cm⁻¹ was attributed to the CO linear adsorption on the top sites of Cu surface (*CO), which was the key intermediate for subsequent C—C coupling step of C₂₊ products formation. On the one hand, for pristine CuN₂, the *CO intermediate with a wavenumber of 2057 cm⁻¹ was observed at −0.68 V vs. RHE. Meanwhile, the intensity of *CO intermediate increased with the potentials, indicating faster conversion of CO₂ to *CO. The *CO stretching frequency shifted to a lower wavenumber as the potentials decreased from −0.68 V to −1.28 V vs. RHE owing to the Stark effect [45–47]. On the other hand, for pristine Cu₂N₄, the *CO intermediate was observed at a more negative potential of −0.98 V vs. RHE and a slightly lower wavenumber of 2037 cm⁻¹ than CuN₂, indicating a higher energy barrier to obtain *CO intermediate. Meanwhile, to quantitatively compare the abilities of Cu sites in CuN₂ and Cu₂N₄ for the conversion of CO₂ into *CO, the *CO intermediate band intensities were plotted from −0.08 V to −1.28 V vs. RHE (Fig. 4c). Obviously, the *CO intermediate band intensity of CuN₂ was stronger than that of Cu₂N₄, and it significantly increased with an increase of potentials, which further supported that CuN₂ easily produced more *CO intermediates than Cu₂N₄.

  Figure 4

  

  Figure 4.  In-situ ATR-SEIRAS results of (a) CuN₂ and (b) Cu₂N₄. On the right were the enlarged spectra from 2090 cm⁻¹ to 2000 cm⁻¹. OCP was circuit potential. (c) *CO band intensity of CuN₂ and Cu₂N₄ from (a) and (b) at different potentials. (d) TPD-CO curve for CuN₂ and Cu₂N₄. (e) DOS calculations of CuN₂ and Cu₂N₄. (f) Gibbs free energy profiles for CO production on CuN₂ and Cu₂N₄. The inset images were the DFT calculation structure model of *COOH adsorbed on Cu site for CuN₂ and Cu₂N₄.

  DownLoad: Full-Size Img PowerPoint

  Besides, the temperature-programmed desorption of CO (TPD-CO) technique was employed to investigate the chemisorption intensity of CO for pristine CuN₂ and Cu₂N₄ (Fig. 4d). Before testing TPD-CO, the thermogravimetric analysis (TGA) was conducted to determine the pyrolysis temperature of the catalysts (Fig. S21 in Supporting information). CuN₂ and Cu₂N₄ showed almost identical TGA curves, indicating that they were both thermally stable before about 250 ℃. Therefore, we collected the TPD-CO data from 50 ℃ to 250 ℃. According to the TPD-CO results, pristine CuN₂ exhibited a much lower CO desorption temperature of 148 ℃ than 207 ℃ of Cu₂N₄, indicating a lower desorption activation energy, which facilitated the *CO desorption and subsequent C—C coupling step on Cu(111) site.

  Furthermore, to understand the different adsorption strength for intermediates during the CO₂RR process, we analyzed the density of state (DOS) for CuN₂ and Cu₂N₄. Especially, Cu 3d orbital partial density of state (PDOS) was conducted to calculate the d-band center which was regarded as the applicable descriptor to explain catalytic reactivity [48,49]. As shown in Fig. 4e, d-center value of CuN₂ and Cu₂N₄ were −1.4 eV and −2.9 eV, respectively. The position of d-band center for CuN₂ was closer to the Femi level (E_f) than Cu₂N₄, indicating stronger adsorption of intermediates [50]. Besides, the reaction pathway and energy barrier of *CO intermediate formation were explored with the DFT calculation [51–53]. The Cu sites of pristine CuN₂ and Cu₂N₄ were modeled as the active sites for CO production. The whole reaction pathway from CO₂ to CO was proposed as follows (Fig. 4f) [44]: CO₂ → *COOH → *CO → CO, where the first step CO₂ → *COOH was the rate-determining step (RDS). Meanwhile, the intermediates such as *COOH and *CO were also detected with in-situ ATR-SEIRAS spectroscopy. The RDS reaction energy of pristine CuN₂ was calculated as 0.46 eV, which was lower than pristine Cu₂N₄ with 0.57 eV. These results further demonstrated that the pristine coordination environment of Cu(imidazole) catalysts played a decisive role in CO generation, thus affecting further C₂₊ production.

  In summary, we uncovered that the pristine Cu(imidazole) played an critical activation function to convert CO₂ into CO, and subsequent the in-situ derived Cu(111) acted as the active sites for C—C coupling. This auto-tandem catalysis activity was closely related to the coordination environment of the pristine MOFs. CuN₂ with more exposed Cu-N sites showed higher ability to form CO, thus the CuN₂/Cu(111) exhibited higher C₂₊ selectivity. The catalytic active sites were explored with HRTEM, SAED, XPS, and XANES techniques. This mechanism was demonstrated with CORR control experiment, TPD-CO, in-situ ATR-SEIRAS, and DFT calculation. We anticipate that the auto-tandem catalysis mechanism will be utilized in the design and development of Cu-based MOF precursors for promoting the C₂₊ products selectivity of reconstructed catalysts. The key of applying the auto-tandem catalytic mechanism is to improve the intrinsic catalytic active and stability of pristine MOF.

  CRediT authorship contribution statement

  Xiang-Da Zhang: Conceptualization, Data curation, Formal analysis, Writing – original draft. Jian-Mei Huang: Data curation. Xiaorong Zhu: Software. Chang Liu: Investigation. Yue Yin: Investigation. Jia-Yi Huang: Investigation. Yafei Li: Supervision, Validation. Zhi-Yuan Gu: Funding acquisition, Supervision, Visualization, Writing – review & editing.

  Acknowledgments

  This work is supported by the National Natural Science Foundation of China (Nos. 22174067 and 22204078), the Natural Science Foundation of Jiangsu Province of China (No. BK20220370), Jiangsu Provincial Department of Education (No. 22KJB150009), State Key Laboratory of Analytical Chemistry for Life Science (No. SKLACLS2218), and the Priority Academic Program Development of Jiangsu Higher Education Institutions. The authors would like to thank BL11B, BL14W1, and BL17B in Shanghai Synchrotron Radiation Facility (SSRF) and BL01B in National Synchrotron Radiation Laboratory (NSRL).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2024.109937.

  
  
• 1. [1]

     J. Liu, D. Yang, Y. Zhou, et al., Angew. Chem. Int. Ed. 60 (2021) 14473–14479. doi: 10.1002/anie.202103398

  2. [2]

     D.H. Nam, O. Shekhah, G. Lee, et al., J. Am. Chem. Soc. 142 (2020) 21513–21521. doi: 10.1021/jacs.0c10774

  3. [3]

     X.F. Qiu, H.L. Zhu, J.R. Huang, et al., J. Am. Chem. Soc. 143 (2021) 7242–7246. doi: 10.1021/jacs.1c01466

  4. [4]

     J.X. Wu, S.Z. Hou, X.D. Zhang, et al., Chem. Sci. 10 (2019) 2199–2205. doi: 10.1039/c8sc04344b

  5. [5]

     A.M. Appel, J.E. Bercaw, A.B. Bocarsly, et al., Chem. Rev. 113 (2013) 6621–6658. doi: 10.1021/cr300463y

  6. [6]

     S. Nitopi, E. Bertheussen, S.B. Scott, et al., Chem. Rev. 119 (2019) 7610–7672. doi: 10.1021/acs.chemrev.8b00705

  7. [7]

     L. Majidi, A. Ahmadiparidari, N. Shan, et al., Adv. Mater. 33 (2021) 2004393.

  8. [8]

     T. Ahmad, S. Liu, M. Sajid, et al., Nano Res. Energy 1 (2022) e9120021. doi: 10.26599/nre.2022.9120021

  9. [9]

     X. Han, Z. Liu, M. Cao, et al., Chem. Mater. 34 (2022) 6713–6722. doi: 10.1021/acs.chemmater.2c00444

  10. [10]

      C. Kong, G. Jiang, Y. Sheng, et al., Chem. Eng. J. 460 (2023) 141803.

  11. [11]

      J. Wang, Y. Zhang, Y. Ma, et al., ACS Mater. Lett. 4 (2022) 2058–2079. doi: 10.1021/acsmaterialslett.2c00751

  12. [12]

      X. Tan, C. Yu, C. Zhao, et al., ACS Appl. Mater. Interfaces 11 (2019) 9904–9910. doi: 10.1021/acsami.8b19111

  13. [13]

      M.K. Kim, H.J. Kim, H. Lim, et al., Electrochim. Acta 306 (2019) 28–34. doi: 10.13000/jfmse.2019.2.31.1.28

  14. [14]

      M.R. Smith, A. Gilman, C.W. Hullfish, et al., J. Phys. Chem. C 126 (2022) 13649–13659. doi: 10.1021/acs.jpcc.2c01287

  15. [15]

      Z. Weng, Y. Wu, M. Wang, et al., Nat. Commun. 9 (2018) 415.

  16. [16]

      F. Yang, A. Chen, P.L. Deng, et al., Chem. Sci. 10 (2019) 7975–7981. doi: 10.1039/c9sc02605c

  17. [17]

      H. Sun, L. Chen, L. Xiong, et al., Nat. Commun. 12 (2021) 6823.

  18. [18]

      Y. Zhao, L. Zheng, D. Jiang, et al., Small 17 (2021) 2006590.

  19. [19]

      Q. Zhu, D. Yang, H. Liu, et al., Angew. Chem. Int. Ed. 59 (2020) 8896–8901. doi: 10.1002/anie.202001216

  20. [20]

      N. Li, P. Yan, Y. Tang, et al., Appl. Catal. B 297 (2021) 120481.

  21. [21]

      D. Yao, T. Cheng, A. Vasileff, et al., Angew. Chem. Int. Ed. 60 (2021) 18178–18184. doi: 10.1002/anie.202104747

  22. [22]

      Y.Q. Tian, H.J. Xu, L.H. Weng, et al., Eur. J. Inorg. Chem. 2004 (2004) 1813–1816.

  23. [23]

      X.C. Huang, J.P. Zhang, Y.Y. Lin, et al., Chem. Commun. (2004) 1100–1101.

  24. [24]

      T. Bao, W. Zhai, C. Yu, et al., Small Struct. 5 (2024) 2300293.

  25. [25]

      Y. Li, Z. Pei, D. Luan, et al., Angew. Chem. Int. Ed. 62 (2023) e202302128.

  26. [26]

      Y. Baek, H. Song, D. Hong, et al., J. Mater. Chem. A 10 (2022) 9393–9401. doi: 10.1039/d1ta10345h

  27. [27]

      Z. Li, Y. Fang, J. Zhang, et al., J. Mater. Chem. A 9 (2021) 19932–19939. doi: 10.1039/d1ta02565a

  28. [28]

      L. Fan, C. Xia, F. Yang, et al., Sci. Adv. 6 (2020) eaay3111.

  29. [29]

      A. Handoko, C. Ong, Y. Huang, et al., J. Phys. Chem. C 120 (2016) 20058–20067. doi: 10.1021/acs.jpcc.6b07128

  30. [30]

      A. Loiudice, P. Lobaccaro, E.A. Kamali, et al., Angew. Chem. Int. Ed. 55 (2016) 5789–5792. doi: 10.1002/anie.201601582

  31. [31]

      D. Kim, C.S. Kley, Y. Li, et al., Proc. Natl. Acad. Sci. U. S. A. 114 (2017) 10560–10565. doi: 10.1073/pnas.1711493114

  32. [32]

      Y. Sha, J. Zhang, X. Cheng, et al., Angew. Chem. Int. Ed. 61 (2022) e202200039.

  33. [33]

      J. Yuan, S. Chen, Y. Zhang, et al., Adv. Mater. 34 (2022) 2203139.

  34. [34]

      Y. Li, S.L. Zhang, W. Cheng, et al., Adv. Mater. 34 (2022) 2105204.

  35. [35]

      Y. Zou, C. Liu, C. Zhang, et al., Nat. Commun. 14 (2023) 5780.

  36. [36]

      T. Yan, P. Wang, W.Y. Sun, Small 19 (2023) 2206070.

  37. [37]

      H. Xiao, T. Cheng, W.A. Goddard Ⅲ, J. Am. Chem. Soc. 139 (2017) 130–136. doi: 10.1021/jacs.6b06846

  38. [38]

      Z. -H. Zhao, K. Zheng, N. -Y. Huang, et al., Chem. Commun. 57 (2021) 12764–12767. doi: 10.1039/d1cc05376k

  39. [39]

      B. r. Ratschmeier, B. r. Braunschweig, J. Phys. Chem. C 125 (2021) 16498–16507. doi: 10.1021/acs.jpcc.1c02898

  40. [40]

      J. Heyes, M. Dunwell, B. Xu, J. Phys. Chem. C 120 (2016) 17334–17341. doi: 10.1021/acs.jpcc.6b03065

  41. [41]

      H. Wang, Y.W. Zhou, W.B. Cai, Curr. Opin. Electrochem. 1 (2017) 73–79.

  42. [42]

      H. Guo, D.H. Si, H.J. Zhu, et al., Angew. Chem. Int. Ed. 63 (2024) e202319472.

  43. [43]

      Q.J. Wu, D.H. Si, Q. Wu, et al., Angew. Chem. Int. Ed. 62 (2023) e202215687.

  44. [44]

      H.J. Zhu, D.H. Si, H. Guo, et al., Nat. Commun. 15 (2024) 1479.

  45. [45]

      H. Miyake, M. Osawa, Chem. Lett. 33 (2004) 278–279.

  46. [46]

      D.K. Lambert, Electrochim. Acta 41 (1996) 623–630.

  47. [47]

      A. Wuttig, C. Liu, Q. Peng, et al., ACS Central. Sci. 2 (2016) 522–528. doi: 10.1021/acscentsci.6b00155

  48. [48]

      S. Jiao, X. Fu, H. Huang, Adv. Funct. Mater. 32 (2022) 2107651.

  49. [49]

      E. Santos, W. Schmickler, ChemPhysChem 7 (2006) 2282–2285. doi: 10.1002/cphc.200600441

  50. [50]

      Q. Zhang, L. Guo, J. Clust. Sci. 29 (2018) 867–877. doi: 10.1007/s10876-018-1346-x

  51. [51]

      G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) 558–561.

  52. [52]

      K. Kim, P. Wagner, K. Wagner, et al., Energy Fuel 36 (2022) 4653–4676. doi: 10.1021/acs.energyfuels.2c00400

  53. [53]

      J. Wang, L. Gan, Q. Zhang, et al., Adv. Energy Mater. 9 (2019) 1803151.

• 

  Scheme 1  The design of auto-tandem catalysts CuN₂/Cu(111) and Cu₂N₄/Cu(111) and the proposed CO₂RR catalytic mechanism.

  下载: 全尺寸图片 幻灯片
  

  Figure 1  (a) LSV curves of CuN₂/Cu(111) and Cu₂N₄/Cu(111) in both CO₂ and N₂ saturated 0.1 mol/L KCl electrolyte. FEs of various products at different operated potentialsl for (b) Cu₂N₄/Cu(111) and (c) CuN₂/Cu(111), respectively. (d) The comparison of FE_C2+ for CuN₂/Cu(111) and Cu₂N₄/Cu(111) in H-cell. (e) C_dl measurements with CV method in non-Faraday potential interval under different scan rates. (f) Stability test at −1.18 V vs. RHE for 6-h electrolysis in H-cell. FEs of various products at different operated currents density in flow-cell for (g) Cu₂N₄/Cu(111), (h) CuN₂/Cu(111), respectively. (i) The comparison of FE_C2+ for CuN₂/Cu(111) and Cu₂N₄/Cu(111) in flow-cell.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  The characterization of CuN₂/Cu(111) and Cu₂N₄/Cu(111) compared with pristine CuN₂ and Cu₂N₄. (a) PXRD patterns. (b) In-situ XANES spectra. The inset is enlarged part from 8975 eV to 8980 eV. (c) In-situ Cu Kβ_{1, 3} XES spectra from OCP to −1.3 V vs. RHE. OCP was circuit potential. The microscopic analysis of (d) CuN₂/Cu(111) and (e) Cu₂N₄/Cu(111). The inset is HRTEM and SAED images.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) LSV curves of pristine CuN₂ and Cu₂N₄ in both CO₂- and N₂-saturated 0.1 mol/L KCl electrolyte. (b) FE_CO at different potentials with 50-s electrolysis for pristine CuN₂ and Cu₂N₄, respectively. (c) FEco for the pristine materials and FE_C2+ for the reconstructed materials at various potentials. (d) After 8-h electrolysis tests for CuN₂ and Cu₂N₄ at −1.18 V vs. RHE, the pristine MOF was reduced to Cu(111), which was immediately tested to obtain FE_CO. (e) Relative rates of CO₂ reduction to CO vs. CO reduction to C₂₊ products for CuN₂ and Cu₂N₄ after 8-h electrolysis at −1.18 V vs. RHE. (f) CORR of CuN₂/Cu(111) and Cu₂N₄/Cu(111) at various CO volume concentrations with 1-h electrolysis under fixed −1.18 V vs. RHE.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  In-situ ATR-SEIRAS results of (a) CuN₂ and (b) Cu₂N₄. On the right were the enlarged spectra from 2090 cm⁻¹ to 2000 cm⁻¹. OCP was circuit potential. (c) *CO band intensity of CuN₂ and Cu₂N₄ from (a) and (b) at different potentials. (d) TPD-CO curve for CuN₂ and Cu₂N₄. (e) DOS calculations of CuN₂ and Cu₂N₄. (f) Gibbs free energy profiles for CO production on CuN₂ and Cu₂N₄. The inset images were the DFT calculation structure model of *COOH adsorbed on Cu site for CuN₂ and Cu₂N₄.

  下载: 全尺寸图片 幻灯片
•