Multivalent Cu sites synergistically adjust carbonaceous intermediates adsorption for electrocatalytic ethanol production

The electrocatalytic carbon dioxide reduction (CO₂RR) to produce high-value chemicals and fuels using renewable electricity provides a promising carbon-neutral route^1,2,3. Multi-carbon (C₂) products, such as ethylene and ethanol, have the prospect of higher energy density and market value than C₁ products⁴. Notably, significant progress has been made in producing gaseous ethylene via the CO₂RR^5,6. In contrast, there has been limited progress to date in the electrochemical conversion of CO₂ into liquid ethanol with high volumetric energy density⁷. Due to the dual constraints of thermodynamics and kinetics, CO₂ electrocatalysts do not provide the majority of ethanol production, preferring ethylene instead^8,9. Therefore, increasing ethanol selectivity relative to hydrocarbon is highly desired, but remains a challenge.

Previous studies have suggested that ethanol production can be optimized by modulating and stabilizing the key intermediate (*HCCOH) to undergo the hydrogenation pathway of *HCCOH to *HCCHOH (ethanol pathway) instead of dehydroxylation way to *HCC (ethylene pathway) over the copper (Cu)-based catalysts^10,11,12,13. This optimization is achieved mainly through strategies such as tandem catalysis¹⁴, metal-support interactions¹⁵, multi-element doping^{12,13,16,17,18}, interfacial interaction¹¹ and various surface binding sites^19,20. However, such Cu-based catalysts usually undergo electrochemical reduction under negative potential, which hinders their application in ethanol production. We hypothesize that ethanol production can be enhanced by synergistic effect of multivalent Cu sites with different adsorption capacities for carbonaceous intermediates, and improving their electrochemical stability under CO₂RR conditions via suitable coordination interaction.

In this study, based on the strategy proposed above, we constructed a multivalent Cu-based coordination polymer catalyst Cu-2,3,7,8-tetraaminophenazine-1,4,6,9-tetraone (Cu-TAPT) with stable Cu²⁺ and Cu⁺ sites, and applied it as the catalyst in the ethanol synthesis via CO₂RR. With the cooperative promotion of Cu²⁺ and Cu⁺ sites, the Faradaic efficiency (FE) ratio of ethanol to ethylene over the Cu-TAPT catalyst was 3.14:1, significantly outperforming the reference catalyst Cu-2,3,5,6-tetraaminobenzoquinone (Cu-TABQ). In situ X-ray absorption and Raman spectroscopic analysis suggested that the multivalent Cu sites are stabilized by the double confinement of N and O. Density functional theory (DFT) calculations further revealed that the synergy of multivalent Cu sites played a crucial role in hindering the breakage of the C–O bond in *HCCOH and facilitating the hydrogenation of C atoms. This process effectively retains the OH group, thereby promoting the selectivity of ethanol in CO₂RR.

Material synthesis and characterization

Cu-TAPT was obtained through the coordination reaction involving 2,3,5,6-Tetraaminobenzoquinone (TABQ) ligands and equimolar Cu²⁺ ions in deionized water at 120 °C. Cu-TABQ was prepared by substituting deionized water with dimethyl sulfoxide, following the similar procedure (Fig. 1a and Supplementary Fig. 1). Due to the solvent effect, the dimerization of the ligands in deionized water caused the partial reduction of Cu²⁺ into Cu⁺, resulting in structural differences. Powder X-ray diffraction (XRD) showed a broad diffraction peak of Cu-TAPT, suggesting an amorphous nature (Supplementary Fig. 2), which facilitates exposure of more active sites, as confirmed by the Brunauer–Emmett–Teller (BET) results (Supplementary Fig. 3)²¹. Scanning electron microscopy (SEM) images revealed distinct granule morphology of Cu-TAPT, showing the difference of Cu-TABQ and TABQ (Supplementary Fig. 4). Energy-dispersive X-ray (EDX) element mappings (Supplementary Fig. 5 and Supplementary Table 1) confirmed that Cu, N and O atoms were homogeneous distribution throughout Cu-TAPT. The contents of Cu in Cu-TAPT and Cu-TABQ obtained from inductively coupled plasma optical emission spectrometer (ICP-OES) were 27.6 and 28.2 wt%, respectively (Supplementary Table 2). Transmission electron microscopy (TEM) image showed the amorphous morphology and the absence of nanoparticles or clusters in Cu-TAPT (Supplementary Fig. 6), which is consistent with the XRD results. According to the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images of Cu-TAPT and Cu-TABQ, ultrahigh-density bright dots were observed, suggesting atomically dispersion of Cu sites due to coordination (Fig. 1b and Supplementary Fig. 7).

a Schematic illustration of the Cu-TAPT and Cu-TABQ synthetic process. b HAADF-STEM image of Cu-TAPT catalyst. c, d N 1 s spectra c and O 1 s spectra d of the TABQ, Cu-TABQ and Cu-TAPT. e Cu 2p spectra of Cu-TAPT catalyst. f The first derivatives of the XANES region of Cu foil, Cu-TAPT, Cu-TABQ, Cu₂O, CuO, CuPc and the inset is the content of Cu⁺ and Cu²⁺ in Cu-TAPT from first-order derivative of XANES. g Wavelet transforms (WT) for the k³-weighted Cu K-edge EXAFS signals of Cu-TAPT.

Full size image

According to high-resolution X-ray photoelectron spectroscopy (XPS), the shifted binding energy of N and O in Cu-TAPT suggested that the electrons of N and O atoms were delocalized and were coordinated with Cu (Fig. 1c, d)²², which was also confirmed by the Fourier transform infrared spectroscopy (FT-IR) results (Supplementary Fig. 8). Different from the typical satellite peaks of Cu²⁺ in Cu-TABQ (Supplementary Fig. 9)²³, the ratio of Cu²⁺ and Cu⁺ in Cu-TAPT was 1:1.84 (Fig. 1e, Supplementary Fig. 10 and Supplementary Table 3), agreed well with the theoretically-predicted value 1:2. The absorption threshold position in X-ray absorption near edge structure (XANES) of Cu-TABQ was basically the same as that of CuO and copper phthalocyanine (CuPc), indicating the similar oxidation state of +2²⁴. Interestingly, Cu-TAPT exhibited an absorption edge position between CuO and Cu₂O, suggesting the coexistence of bivalent and monovalent of Cu and the ratio was 1:1.85 (Fig. 1f and Supplementary Fig. 11a), in accordance with the XPS results. The Fourier transforms extended X-ray absorption fine structure (EXAFS) spectra of Cu-TAPT displayed a prominent peak that is different from Cu–O and Cu–N and has a higher coordination number. Moreover, no peaks attributed to metal Cu–Cu bonds were observed around 2.2 Å, indicating that Cu sites are atomically distributed in Cu-TAPT (Supplementary Fig. 11b)²⁵. The fitting results of EXAFS spectra (Supplementary Fig. 12 and Supplementary Table 4) showed that the average coordination number (CN) of Cu–O and Cu–N in Cu-TAPT was 4.10 and 4.22, respectively, which was different from that of Cu-TABQ (CN_Cu–N = 4.04) (Supplementary Fig. 13 and Supplementary Table 4). The differences in these structures can also be clearly observed from wavelet transform (WT) spectra (Fig. 1g and Supplementary Fig. 14). These characterizations agree with the building unit of the Cu-TAPT catalyst, which comprises Cu in different valence states and TAPT monomers extending to both ends through Cu–N and Cu–O coordination bonds. Such adjacent Cu sites with different valence states are expected to have synergistic effect for CO₂RR, as substantiated in the subsequent section.

Electrocatalytic CO₂RR performance

The electrocatalytic CO₂RR activity and selectivity of the as-synthesized catalysts were tested in a flow cell reactor with 1 M KOH electrolyte, and the product was detected by gas chromatography (GC) and ¹H nuclear magnetic resonance (NMR) spectroscopy. Cu-TAPT exhibited a lower onset potential than that of Cu-TABQ in linear sweep voltammetry (LSV) curves, suggesting that the multivalent catalyst could effectively reduce the overpotential of CO₂RR (Supplementary Fig. 15). FE_ethanol increased and then decreased with applied potentials over Cu-TAPT in the range from −0.70 V to −0.85 V vs. RHE (Fig. 2a and Supplementary Figs. 16,17). At −0.80 V vs. RHE, FE_ethanol and FEC₂₊ reached its maxima of 54.3 ± 3% and 84.9 ± 2%, respectively (Supplementary Figs. 17). Cu-TAPT showed 2.31 times increase in FE_ethanol and 1.43 times increase in FEC₂₊ compared to the Cu-TABQ, which only reached the highest FE_ethanol and FEC₂₊ of 23.5% and 60%, respectively, at a more negative potential of −0.98 V vs. RHE (Supplementary Figs. 17-19). In particular, the FE ratio of ethanol to ethylene was almost more than 2 under the applied potential range, and even exceeds 3.1 times at −0.8 V vs. RHE (Fig. 2b), suggesting a preference for ethanol on Cu-TAPT. Compared with Cu-TABQ, Cu-TAPT exhibited greater current density, and the total current density and ethanol partial current density could reach 429 mA cm⁻² and 233 mA cm⁻² at −0.8 V vs. RHE, respectively (Supplementary Figs. 20,21). The C₂₊ and ethanol formation rates over Cu-TAPT increased with applied potentials and reached the maximum of 1,205 and 724 μmol h⁻¹ cm⁻² at −0.8 V vs. RHE, which were about 2.53 times and 4.16 times higher than that on Cu-TABQ at −0.98 V vs. RHE, respectively (Fig. 2c). The Cu-TAPT catalyst outperformed most of start-of-the-art Cu-based catalysts in FE_ethanol or current density or formation rate (Supplementary Table 5). Moreover, the mass activity, specific activity and turnover frequency (TOF) of Cu-TAPT were significantly better than those of Cu-TABQ (Supplementary Figs. 22,23). Based on the results above, we propose a mechanism involving concerted catalysis by different valence Cu sites to elucidate the variations in catalytic performance between Cu-TAPT and Cu-TABQ (Fig. 2d). Cu sites in different valence states exhibit distinct adsorption capacities for intermediates, thereby facilitating coupling process, promoting ethanol production, and inhibiting the formation of hydrocarbon products. Further specific details are discussed in the following sections.

a FEs of all products under different applied potentials over Cu-TAPT in 1 M KOH electrolyte. b FE ratio of ethanol/ethylene at different potentials over Cu-TAPT. c, The formation rates of C₂₊ products and ethanol under different applied potentials over Cu-TAPT and Cu-TABQ in 1 M KOH electrolyte. d, Schematic illustration of C − C coupling formation process on Cu-TAPT based on the synergistic effect of Cu²⁺ and Cu⁺. The ball of dark blue, red, gray, azure, pink and white denote Cu²⁺, Cu⁺, C, N, O and H, respectively. e Time-dependent total current density, FE_ethanol and FE_C2+ on Cu-TAPT were recorded at −0.8 V vs. RHE for at least 30 h. Error bars in accordance with the standard deviation of at least three independent measurements. All the potentials involved in this figure have been iR corrected (R = 1.89 Ω). The potential without iR is shown in Supplementary Fig. 17.

Full size image

The properties of the electrode and electrolyte interfaces were further investigated via electrochemical double-layer capacitance and electrochemical impedance spectroscopy (EIS). The electrochemical active surface area (ECSA) of Cu-TAPT was 2.52 times of that of Cu-TABQ (Supplementary Fig. 24 and Table 6)¹⁸. The Nyquist plots showed that Cu-TAPT had a smaller charge transfer resistance (R_ct) than Cu-TABQ (Supplementary Fig. 25 and Supplementary Table 7), which further reflected the more favorable charge transfer kinetics over Cu-TAPT for CO₂RR²⁶. Similarly, Cu-TAPT exhibited greater ionic conductivity, which is favorable to electrocatalytic reaction (Supplementary Table 7)²⁷. In addition, the long-term durability of Cu-TAPT was also evaluated by a continuous CO₂RR at −0.8 V vs. RHE for 30 h, and the FE_ethanol and FE_C2+ products did not decrease obviously (Fig. 2e). As a comparison, the performance results for Cu-TABQ revealed a rapid decline in the FE for C₂₊ products within 7 h (Supplementary Fig. 26).

Various characterizations were performed to ulteriorly unveil the structural stability after CO₂RR. The XRD pattern of Cu-TAPT indicated no metallic Cu appeared after CO₂RR (Supplementary Fig. 27). Meanwhile, the XPS results suggested that the ratio of Cu²⁺ and Cu⁺ and N 1 s and O 1 s spectra in Cu-TAPT remains basically unchanged before and after the CO₂RR process, whereas the molecular structure of Cu-TABQ could not be stabilized, and almost decomposes (Supplementary Fig. 28). Cu-TAPT could maintain the coexistence of Cu²⁺ and Cu⁺ with a ratio of 1:1.78 even after a long electrolysis process for 30 h (Supplementary Fig. 29). Additionally, the morphology of Cu-TAPT generally remained unchanged after CO₂RR (Supplementary Fig. 30), which further indicated the excellent stability. Taken together, the Cu-TAPT catalyst has promising potential of application for the conversion of CO₂ to ethanol products.

In situ characterization

To investigate the oxidation state and structural evolutions of catalysts during the CO₂RR process, in situ XAS studies were carried out via a customized flow cell (Supplementary Fig. 31). The in situ XANES spectra of Cu-TAPT revealed an unchanged Cu oxidation state (Cu²⁺ and Cu⁺) for 2 h under −0.98 V vs. RHE (Fig. 3a, b), indicating that the active Cu sites in Cu-TAPT were stable during CO₂RR process. Consistent with XANES results, no Cu–Cu bond at around 2.24 Å was discerned in the EXAFS (Supplementary Fig. 32), demonstrating that Cu remained as coordinated and atomically dispersed state of Cu-TAPT under CO₂RR condition²⁸. These results were also confirmed by the potential-dependence results (Fig. 3c and Supplementary Fig. 33). On the contrary, the oxidation state of Cu decreases with decreasing potential from OCP to −0.98 V vs. RHE of Cu-TABQ (Supplementary Fig. 34). The major peak ascribed to Cu–Cu bonds at approximately 2.2 Å emerged, implying that the catalysts underwent severe electrochemical reconstruction and transformed into metallic Cu particles. Linear combination fitting (LCF) of the XANES spectra was further conducted to calculate the chemical valence of Cu-TAPT, finding no metallic Cu (Fig. 3d and Supplementary Fig. 35). Likewise, in situ Raman spectroscopy of Cu-TAPT was conducted at the same constant potential of −0.98 V vs. RHE under CO₂ atmosphere for 2 h (Fig. 3e). No considerable signal change was discovered for Cu-TPAT throughout extended operation, implying that the coordination structure of Cu-TAPT had high stability. We attribute these results to mutual constraints between Cu, N, and O atoms arising from asymmetric coordination^29,30,31,32.

a, b Time-dependent in situ Cu K-edge XANES a and first-order derivatives of the XANES region b of Cu-TAPT catalyst recorded at −0.98 V vs. RHE under CO₂ atmosphere. c Potential-dependence of in situ first-order derivatives of the XANES spectra of the Cu-TAPT under CO₂RR test. d Variation in the valence state of Cu species in Cu-TAPT catalyst with time. e In situ Raman spectra of Cu-TAPT catalyst operating at −0.98 V vs. RHE under CO₂ atmosphere with a 532 nm laser. f Total energy as a function of simulation time during the AIMD simulations under 300 K and the electric field of −2.3 V/Angstorm. All the potentials involved in this figure are not iR corrected.

Full size image

In addition, DFT and ab initio molecular dynamics (AIMD) calculations were used to further interpret the enhanced stability of Cu-TAPT (Model information in Supplementary Data 1). The crystal orbital Hamilton populations (COHP) were investigated for Cu and N or O atom bond strength of Cu-TAPT, Cu-TABQ and CuO (Supplementary Figs. 36,37 and Supplementary Table 8). According to the COHP theory, the negative and positive values of integral crystal orbital Hamilton populations (ICOHP) represent bonding and anti-bonding interactions respectively, and the zero value represents non-bonding interactions³³. The more negative the ICOHP value, the stronger the bond strength. The energy-weighted ICOHP sums up to the fermi level of Cu–N and Cu–O bonds of Cu-TAPT were −1.18 and −1.09 eV, respectively, better than Cu-TABQ (Cu–N) and CuO (Cu–O). Therefore, this asymmetric coordination can effectively improve the stability of Cu–N or Cu–O, thereby improving the electrochemical stability during the CO₂RR process, which is consistent with the results of in-situ XAS experiments. Subsequently, AIMD simulations were performed in the canonical ensemble (NVT) (Fig. 3f and Supplementary Fig. 38). The system energy remains basically unchanged under 300 K and electric field, indicating that the structure has excellent stability. Combined with the aforementioned in situ XAS and Raman results, as well as DFT calculations, we can conclude that the strong asymmetric coordination has ultra-high electrochemical stability and resistance to reconstruction over a broad potential window and long-term electrolysis during the CO₂RR process, allowing Cu-TAPT to maintain excellent catalytic performance during the electrolysis.

Furthermore, the difference in ion transport capabilities based on Cu-TAPT and Cu-TABQ was echoed in Bode plots, obtained by in situ EIS measurements during electrically driven transport of single-component solutions in the flow cell (Supplementary Fig. 39). The characteristic frequency (f) which corresponds to the local maxima in the Bode phase plot has an inverse correlation to the overall impedance of the electrode^34,35. Under the applied potential of OCP, the f_Cu-TAPT was higher than f_Cu-TABQ, indicating that lower impedance and better ion transport capability of Cu-TAPT (Supplementary Fig. 40). Likewise, as the potential increased, the f_Cu-TAPT gradually increased and a pronounced decrease in the low-frequency peak in the corresponding phase angle, indicating that the ion transport impedance of Cu-TAPT further decreased under the action of electric field and was better than Cu-TABQ (Supplementary Fig. 39)³⁶. As expected, similar behaviors were also reflected in the corresponding Nyquist plots (Supplementary Fig. 41).

Reaction mechanism investigation

As is well-known that the formation of multi-carbon products undergoes the C–C coupling process and then further reduction in CO₂RR³⁷. To gain insight into the C–C coupling of multivalent Cu sites on the Cu-TAPT electrode, the Raman measurements under the potential of OCP and −0.78 V vs. RHE (optimum potential) were conducted to explore the adsorption behavior of the chemical intermediates (Supplementary Fig. 42). The observed peaks in the range 1800–2200 cm⁻¹ can be ascribed to the *CO, including *CO_atop (2094 cm⁻¹) and *CO_bridge (1978 cm⁻¹) (Supplementary Fig. 43)¹⁴. Furthermore, the hydrocarbonate (HCO₃⁻) peak at 1345 cm⁻¹ on Cu-TAPT was almost absent in Raman spectra under −0.78 V vs. RHE (Supplementary Fig. 44)³⁸, which indicates that the local pH value on the electrode during CO₂RR is high. The high pH value of the electrode surface is believed to be beneficial to C–C coupling, lower the energy barrier for CO₂ activation and inhibit the evolution of H₂, and help improve the activity of CO₂RR³⁹.

To complement the above findings, we further performed in situ attenuated total reflection Fourier-transform infrared spectroscopy (ATR − FTIR) to monitor the co-adsorption behavior on the Cu-TAPT and Cu-TABQ catalyst (Supplementary Figs. 45,46). Previous studies have indicated that *CO_bridge protonation is energetically more favorable than *CO_atop⁴⁰. Therefore, C–C coupling can be promoted under the asymmetry between *CO and *CHO (or *COH) after the *CO_bridge protonation step, and the result is also confirmed by Gibbs free energy calculations (Supplementary Fig. 47, Supplementary Table 9 and Supplementary Data 1). Notably, the adsorbed *CHO (or *COH) intermediate on Cu-TAPT was also observed from the ATR − FTIR spectrum and increased with the applied potential, strongly supporting this process (Supplementary Fig. 46). Therefore, due to the adsorption capacity difference of different valence Cu sites for CO, the energy barrier for the formation of asymmetric C–C coupling is low based on the synergistic effect of Cu²⁺ and Cu⁺, which helps to the generation of C₂ products⁴¹. The peaks at 1339 and 1542 cm⁻¹ are indexed to the absorbed *OC₂H₅ and *OCCOH on the catalysts, respectively (Fig. 4a, b). A ratio value of *OC₂H₅/*OCCOH on Cu-TAPT was larger than that on Cu-TABQ (Fig. 4c)¹². This is ascribed to that the multivalent Cu sites induced unbalanced coupling environment, which accelerates the breakage of the Cu–COH bond, thus enhancing the production of ethanol intermediates, which agrees with the report that diversity in *CO binding sites enhances the formation of ethanol product¹⁹.

a, b In situ ATR–FTIR spectra obtained under different potentials over Cu-TAPT a and Cu-TABQ b catalyst under CO₂RR. c Potential dependence of *OC₂H₅/*OCCOH ratio obtained for Cu-TAPT and Cu-TABQ. d Gibbs free energy diagram for the hydrogenation (*HCCOH  to  *HCCHOH) and dehydroxylation (*HCCOH to *HCC) steps on Cu-TAPT and Cu-TABQ. e The COHP of C–OH and Cu–COH in *HCCOH over Cu-TAPT. f Charge transfer in C–OH and Cu–COH in *HCCOH adsorbed on Cu-TAPT and Cu-TABQ. All the potentials involved in this figure are not iR corrected.

Full size image

Subsequently, DFT was carried out to study the intrinsic role of synergistic catalysis in promoting the selectivity of ethanol (Model information in Supplementary Data 1). Most previous mechanistic studies have suggested that formations of both ethylene and ethanol proceeds via further transformation of the *HCCOH intermediate. The *HCCOH evolves into a hydrocarbon intermediate (*HCC) or a hydroxyl-containing intermediate (*HCCHOH), serving as the formation of ethylene and ethanol. When the *HCCOH intermediate is hydrogenated and dehydrated to generate *HCC, hydrocarbons (ethylene) are generated, while when the hydroxyl group is retained to generate the *HCCHOH intermediate, ethanol is finally generated. Therefore, the distribution of ethylene and ethanol products is mainly regulated by regulating the strength of the Cu–COH and C–OH bonds on the intermediate^11,15,42. We further calculated the Gibbs free energy of *HCCOH converted into *HCC and *HCCHOH to examine the impact of cooperative catalysis on C₂ product distribution (Fig. 4d and Supplementary Table 9). Compared with Cu-TABQ, the synergistic effect of Cu²⁺ and Cu⁺ over Cu-TAPT can promote the hydrogenation of *HCCOH into hydroxyl-containing intermediates, reducing the reaction energy by 1.15 eV. The proton-electron coupling step of *HCCHOH is far superior to *HCC, providing an explanation for enhanced selectivity to ethanol on Cu-TAPT (Supplementary Fig. 48).

It is believed that the hydrogenation step involves the cleavage of the Cu–COH bond to form a C–H bond, while the dehydroxylation step requires the cleavage of the C–OH bond. COHP and charge density analysis were therefore used to quantify the relative strength of the Cu–COH and C–OH bonds in *HCCOH (Fig. 4e,f and Supplementary Table 10). Compared with Cu–COH, the energy-weighted ICOHP summed up to the fermi level for C–OH is more negative, and more charge transfer from C to OH group in Cu-TAPT on account of the synergistic effect of Cu²⁺ and Cu⁺, indicating that the C–OH bond is stronger than Cu–COH bond, thus promoting ethanol production.

In summary, we have achieved highly efficient synthesis of ethanol in CO₂RR by the synergy of Cu²⁺ and Cu⁺ sites on polymer catalyst Cu-TAPT, which suppresses the production of hydrocarbons and promotes the generation of ethanol. Cu-TAPT achieved a FE of 54.3 ± 3% towards ethanol at the current density of 429 mA cm⁻², and the ratio of ethanol to ethylene reached 3.14:1. The coordination between Cu²⁺ and Cu⁺ with the functional groups in Cu-TAPT makes the catalyst very stable in CO₂RR. Additionally, cooperative catalysis of the multivalent Cu sites can stabilize the hydroxyl-containing C₂ intermediate (*HCCOH) and promotes the hydrogenation of C atoms (*HCCHOH), enhancing the selectivity to ethanol. We believe that of design of catalysts with multivalent active sites with synergistic effect can be used to prepared other efficient and selective catalysts for CO₂RR.

Chemicals

Copper nitrate trihydrate (Cu(NO₃)₂·3H₂O, 99%) was obtained from Beijing Innochem Science & Technology Co., Ltd. 2,3,5,6-Tetraaminobenzoquinone (TABQ, 97%) was purchased from Energy Chemical. Deuterated water (D₂O, AR) was provided by Shanghai Meryer Biochemical Technology Co., LTD. Potassium hydroxide (KOH, AR), isopropanol (i-PrOH, AR), dimethylsulfoxide (DMSO, AR) and phenol (C₆H₅OH, AR) were all purchased from Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China). CO₂ and Ar (Shanghai Chemistry Industrial Zone Pujiang Special Type Gas Co., Ltd.) had a purity of 99.999% and were used as received. Anion exchange membrane FUMA–FAA–3–PK-130 (0.11–0.13 mm thick, 1.1–1.4 meg g⁻¹ exchange capacity) was obtained from FuMA-Tech. Perfluorinated resin solution containing Nafion^TM 1100 W ( ~ 5 wt.% in lower aliphatic alcohols and water) was purchased from Sigma-Aldrich. Carbon paper-based gas diffusion layers (GDLs, Sigracet 29 BC) and nickel foam were provided by the Fuel Cell Store. Deionized water (resistivity 18.2 MΩ cm) was used for all experiments.

Synthesis of Cu-TAPT and Cu-TABQ

0.15 mmol of Cu(NO₃)₂·3H₂O and 0.05 mmol TABQ monomer were added into 20.0 mL H₂O and heated for 72 h at 120 °C. After cooling to room temperature, the precipitate was filtered and alternately washed with deionized water and ethanol. Subsequently, the Cu-TAPT was obtained by drying at 60 °C. Cu-TABQ was prepared with a similar procedure, and the only difference was that 20 mL DMSO was used as the solvent instead of water. The contents of Cu in Cu-TAPT and Cu-TABQ obtained from ICP-OES were 27.6 and 28.2 wt%, respectively.

Characterizations

XRD patterns were recorded with an X-ray diffractometer (Rigaku Ultima VI X-ray) with Cu-Kα radiation, and the scan speed was 5°/min. The morphology of catalysts was characterized by FESEM (JSM IT-800) equipped with EDS operated at 5 kV. High resolution transmission electron microscopy (HRTEM) images were obtained on Tecnai G2 F30 operated at 300 kV. HAADF-STEM images were achieved on an aberration-corrected electron projection microscope (FEI Titan G2 60-300). XPS spectra were obtained on an AXIS Supra surface analysis instrument using a monochromatic Al Kα X-ray beam (1,486.6 eV). FTIR spectra was determined utilizing an IRTracer-100 infrared spectrometer with a broad range of 400–4,000 cm⁻¹. The Cu content was quantitatively determined by the ICP-OES (Agilent-5110). Nitrogen adsorption-desorption isotherms were measured using an automated gas sorption analyzer (AutoChem Discovery). X-ray absorption spectroscopy (XAS) measurements were carried out at the 4B9A beamline at Beijing Synchrotron Radiation Facility (BSRF) and the BL13SSW beamline at Shanghai Synchrotron Radiation Facility (SSRF), China. The Cu K-edge X-ray absorption spectra of Cu-TAPT and Cu-TABQ electrodes were obtained, and Cu foil, CuO and Cu₂O and CuPc were used as references. The EXAFS oscillations χ(k) as a function of photoelectron wave number k was extracted by following standard procedures. The theoretical paths were generated using FEFF and the models were completed in a conventional way using the fitting program ARTEMIS. Part of the computational work was provided by “Ceshigo Research Service (www.ceshigo.com)” for technical support.

In situ Raman measurements

Stability measurements: In-situ Raman measurements were performed on a Renishaw inVia Reflex Raman microscope in a modified flow cell under CO₂ atmosphere by using a water immersion objective lens (63x) with a 532 nm laser, with 5 s integration and averaging 20 scans per region. The modified flow-cell setup consisted of a three-electrode setup, in which platinum wire and Ag/AgCl electrodes were used as the counter and reference electrodes, respectively.

Reaction mechanism study: In situ Raman study was conducted on a Renishaw inVia Reflex Raman spectrometer with a 785 nm solid laser as an excitation source. The measurements were performed in a custom-built flow cell (Gaossunion) using graphite rods and Ag/AgCl electrodes as the anode electrode and reference electrode, respectively. A working electrode was prepared by spray brushing the catalyst onto the GDL. The electrolyte (0.5 M KOH was used as electrolytes to avoid scattering of the Raman signal by bubbles at high currents) was pumped into the GDE using a syringe pump. CO₂ was passed into the back of the GDE.

In situ Attenuated Total Reflection Fourier-transform Infrared Spectroscopy (ATR–FTIR) measurements

In situ electrochemical ATR–FTIR measurements were carried out by utilizing an IRTracer-100 infrared spectrometer equipped with a liquid nitrogen-cooled MCT detector with a broad range of 1000–4000 cm⁻¹. A hemisphere silicon prism was polished with diamond abrasion paste and sonicated with deionized water. The prim then was immersed in a piranha solution (98% H₂SO₄: 30% H₂O₂, v/v = 3/1) for 2 h. Electrocatalyst ink was consistent with electrochemical tests. 50 μL of the catalyst ink was dropped onto the prism and left to dry slowly. The above prism was mounted in a homemade electrochemical three electrode cell with an Ag/AgCl reference electrode and a platinum-wire counter electrode. The catalyst electrode was tightly pressed against the hemisphere silicon prism crystal window with a micrometre-scale gap to reduce the loss of infrared light. The electrolyte was 0.5 M KHCO₃, which was constantly purged with CO₂ during the experiment. The spectra under OCP were recorded for comparison.

In situ X-ray absorption spectroscopy (XAS) measurements

In situ XAS spectra were measured at the 4B9A station of the Beijing Synchrotron Radiation Facility based on CO₂RR measurement conditions. A custom-built in situ electrochemical cell was used with Pt mesh counter electrode and Ag/AgCl reference electrode. The acquisition of each spectrum took around 18 minutes. Data analysis was performed using Athena and Artemis software.

XAFS analysis

XAS data were analyzed using ATHENA and ARTEMIS from the Demeter package software. The normalized EXAFS function was transformed, χ(E), in energy space to χ(k), where k is the photoelectron wave vector. χ(k) was multiplied by k³ to amplify the EXAFS oscillations in the mid-k region for assessing the interatomic interaction of Cu atoms. To differentiate the Cu K-edge EXAFS oscillation from different coordination shells, we applied Fourier transformation (from k space to R space) of the k³-weighted χ(k) with a k range of 2.5–12.4 Å⁻¹. The Cu K edge EXAFS data were extracted from 2.5–12.4 Å⁻¹ in k-space, and the Fourier transformed data were analyzed in the range of 1.0–2.8 Å for the prepared materials and with a k-space of 2.5–13.5 Å⁻¹ and R-range of 1.0–3.0 Å for the reference materials. The experimental data were fitted with theoretical backscattering paths calculated using FEFF6, based upon CIF files from the Materials Project. The fits were performed based on k³ weighting. The amplitude reduction factor, S₀², was determined by fitting a reference copper foil, using the first single-scattering paths with tabulated coordination numbers of bulk copper (fcc–Cu metal: mp–30, symmetrized). The value of S₀² was determined to be 0.82. For Wavelet Transform analysis, the χ(k) exported from Athena was performed to provide the overall distribution.

Electrocatalytic CO₂ reduction reaction (CO₂RR) measurements

All the electrocatalytic experiments were performed in a flow cell (101017, Gaossunion) on the electrochemical workstation (CHI 660E, Shanghai CH Instruments Co., China) at ambient pressure and temperature. Typically, 6 mg of catalyst and 60 μL of 5 wt% Nafion 1100 W were dispersed in 1 mL i-PrOH and sonicated for 20 min. Then, the resulting ink was dropped onto a piece of 1.5 cm × 2.5 cm GDL to prepare gas diffusion electrodes (GDEs). The prepared electrodes were dried under an infrared lamp around ambient environment for 10 min. In the flow cell, the prepared GDEs, Ag/AgCl electrode (saturated KCl) and nickel foam electrode were employed as the working, reference and counter electrodes, respectively. The electrolyte was 1 M KOH aqueous solution. Anode chamber and cathode chamber were separated by an anionic membrane. 40 mL 1 M KOH was circulated in cathode chamber and anode chamber at a constant rate of 20 mL min⁻¹ by peristaltic pump. The gaseous CO₂ was flowed by the backside of the GDE at a rate of 20 SCCM, controlled by a mass flow meter. The geometric surface area of the catalyst was 1 cm². All the potentials were reported with respect to the RHE using the formula^14,25:

The ohmic losses was used to measure between the working and reference electrodes via EIS at the beginning of the electrolysis, with 70% IR compensation (i, current; R, uncompensated resistance) for manual correction of the potential.

During the stability test, the violent fluctuation of current was ascribed to the device itself, which would cause electrolyte flooding and carbonate precipitation in strong alkali environment to block the gas channel. Therefore, the flow cell needed to be disassembled and reassembled. Also, the electrolyte was also replaced in total circulation volume at each orange arrow, leading to the recovery of the catalytic performance.

Electrochemical measurements

All the electrocatalytic tests were conducted at ambient pressure and temperature in a flow cell. LSV curves were conducted with scan rate of 100 mV s⁻¹. The C_dl value was obtained by recording cyclic voltammetry (CV) at various scan rates in the non-Faradaic region by flow cell.

where C_dl and C_s are the double layer capacitance and specific capacitance, respectively.

EIS measurement was carried out on the electrochemical analyzer in a frequency range from 10⁵Hz to 10⁻² Hz with an amplitude of 5 mV at an open circuit potential and equilibrium potential. The frequency and overall impedance as related by the following Equation³⁵:

The data obtained from the EIS measurements were fitted by the Zview software (Version 3.1, Scribner Associates, USA). From EIS data, it is straightforward to evaluate the electrical parameters. The intercept corresponding to the real impedance axis gives the R_s (also electrochemical cell resistance) and then the ionic conductivity can be obtained using the following Equation³⁶:

where L and S are the sample thickness and contact area, respectively.

CO₂RR product analysis

The gas products were analyzed by the GC (Agilent-8890), which was equipped with the flame ionization detectors (FID) and thermal conductivity detectors (TCD). The liquid products were quantified by NMR spectrometer (Bruker; Ascend 500 MHz) in D₂O with C₆H₅OH (200 mM) and DMSO (5 mM) as internal standards (Supplementary Figs. 16,19). Solvent pre-saturation technique was implemented to suppress the water peak. The moles of a liquid product were calculated from integral areas and calibration curves. After the quantification, the FE of each product was calculated as follows:

Where Q is the amount of charge passed through the working electrode; F is the Faraday constant (96,485 C mol⁻¹); C is the concentration of product; V is the volume of the electrolyte; n is the number of the electrons transfer for 1 mole product in electrocatalysis.

The calculation of TOF:

Where I_product is the partial current for certain product (A); n is the number of electrons transferred for product formation (which is 12 for ethanol production.); m_cat is the catalyst mass in the electrode (g); ω is in loading in the catalyst. M_Cu is the atomic mass of Cu, 63.55 g mol⁻¹.

Computational details

DFT calculations were performed on the Dmol³ module by the Materials Studio 2020^43,44. The structures of intermediates in electrocatalysis were optimized, followed by the calculation of the frequency. The generalized gradient approximation (GGA) with the Perdew-Burke-Ernzerhof (PBE) function and TS for DFT-D correction were employed in the calculations⁴⁵. The convergence tolerance of energy, force and displacement convergence were set as 1 × 10⁻⁵ Ha, 2 × 10⁻³ Ha/Å and 5 × 10⁻³ Å, respectively. The bond configuration analysis is based on the COHP theory⁴⁶. The energy contribution of electrons filling the electron orbitals of a crystal is divided into the energy contribution of the overlapping atomic/fragment (molecule) orbital pairs. The time step for the AIMD simulation is 2 fs, and the canonical ensemble condition (NVT) and the electric field (Z-axis) was imposed with a target temperature of 300 K⁴⁷. The smearing function was employed to improve the convergence of states near the Fermi level with a smearing width of 0.005 eV.

The free energy of a species was calculated according to the standard formula:

where ΔE is the DFT total energy, ΔZPE stands for the correction of zero-point energy, T denotes the thermodynamic temperature (298.15 K), and ΔS is the variation of entropy. ΔZPE and ΔS are calculated with the vibrational frequencies of adsorbates and molecules as calculated within DFT.