In-situ constructed Cu/CuNC interfaces for low-overpotential reduction of CO2 to ethanol

Abstract Electrochemical CO2 reduction (ECR) to high-value multi-carbon (C2+) products is critical to sustainable energy conversion, yet the high energy barrier of C-C coupling causes catalysts to suffer high overpotential and low selectivity toward specific liquid C2+ products. Here, the electronically asymmetric Cu-Cu/Cu-N-C (Cu/CuNC) interface site is found, by theoretical calculations, to enhance the adsorption of *CO intermediates and decrease the reaction barrier of C-C coupling in ECR, enabling efficient C-C coupling at low overpotential. The catalyst consisting of high-density Cu/CuNC interface sites (noted as ER-Cu/CuNC) is then accordingly designed and constructed in situ on the high-loading Cu-N-C single atomic catalysts. Systematical experiments corroborate the theoretical prediction that the ER-Cu/CuNC boosts electrocatalytic CO2-to-ethanol conversion with a Faradaic efficiency toward C2+ of 60.3% (FEethanol of 55%) at a low overpotential of −0.35 V. These findings provide new insights and an attractive approach to creating electronically asymmetric dual sites for efficient conversion of CO2 to C2+ products.


INTRODUCTION
Renewable-energy-driven electrochemical CO 2 reduction (ECR) to fuels and value-added chemicals offers a sustainable opportunity to mitigate the environmental crisis and achieve carbon neutrality [1][2][3][4]. With the advantages of high energy density and economic value, liquid-phase multi-carbon products (C 2+ products, e.g. ethanol and acetate) are considered ideal products in the chemical industry [2]. To date, Cu-based materials are known to be the most promising catalysts for harvesting C 2+ products with considerable performance [5]. However, the complicated mechanisms of the multistep pathways and high energy barriers for ECR to C 2+ products, especially the C-C coupling step, brings significant challenges to developing highperformance catalysts with well-defined active sites for selectively reducing CO 2 into desired C 2+ products [6].
Various Cu-based catalysts have been recently reported to achieve the high-efficiency production of C 2+ products like ethylene at high potentials, such as over −1.0 V (all potentials hereafter are versus reversible hydrogen electrode (RHE) [5][6][7]). The corresponding theoretical studies demonstrated that the C-C coupling step of * CO and CO-derived intermediates ( * CHO and * COH) in most pathways for overall ECR to C 2+ products possessed the highest energy barrier [1,8,9]. Although applying more negative potentials could overcome the high energy barrier of C-C coupling and enhance the selectivity toward C 2+ products, the activity of the competitive hydrogen evolution would also be enhanced simultaneously, and the total energy efficiency of the whole electrolyzer would be decreased [10]. Besides, the over-input energy would directly reduce CO 2 into more thermally stable products (methane or ethylene) instead of the desired liquid C 2+ products [5][6][7]11,12].
The key to boosting C-C coupling at low overpotentials is to decrease the kinetic energy barrier of the C-C coupling process. One efficient way is to construct the interfaces of Cu δ+ and Cu 0 on the C The Author(s) 2022. Published by Oxford University Press on behalf of China Science Publishing & Media Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. catalytic surface, which can decrease the energy of C-C coupling by promoting the adsorption and dimerization of CO intermediates, thus enhancing the activity of CO 2 -to-C 2+ conversion [13]. Heteroatoms with high electronegativity like oxygen and halides, or electron-deficiency structures like boron, were applied to construct interfaces of Cu δ+ and Cu 0 for promoting C-C coupling [11,[14][15][16]. For example, Cuenya et al. conducted a plasma-activated Cu to promote the conversion of CO 2 to C 2+ products at −0.9 V [11]; Sargent et al. used positive-valence boron to regulate the ratio of Cu δ+ to Cu 0 on the surface to enhance the selectivity toward C 2+ products at −1.1 V [14]. However, current Cu catalysts with mix-valence interfaces usually comprise a mixture of Cu, Cu + and even Cu 2+ species and introduce diverse types of Cu sites on the catalytic surfaces, significantly lowering the selectivity toward the specific product [17]. Considering that the kinetic energy levels needed for producing each C 2+ product are similar [18,19], advanced strategies for delicately constructing Cu-based catalysts with specific active sites are urgently needed to selectively convert CO 2 into specific C 2+ products.
Here, in order to develop high-performance ECR catalysts for producing desired C 2+ products, we seek an approach to the catalyst design for decreasing the kinetic energy barrier of the C-C coupling. Based on a former reported C-C coupling pathway towards C 2+ products [20], density functional theory (DFT) was applied to investigate the kinetics and thermodynamics of reactions at different interface sites and it was found that the electronically asymmetric structured interface sites that formed on the interface of Cu-Cu and Cu-N-C sites (Cu/CuNC) exhibit the low energy barrier of only 0.30 eV for the C-C coupling promoted by the delocalized electrons. More importantly, we experimentally verify this concept by constructing well-defined Cu/CuNC interface sites on a Cu nanoparticle catalyst (ER-Cu/CuNC), electrochemically reduced, in situ, from a high-loading Cu-N-C single atomic catalyst (SAC). Both in-situ and ex-situ characterizations were applied to unveil the formation process and detailed structures of the Cu/CuNC interface sites. The as-described ER-Cu/CuNC catalyst exhibited an outstanding ECR performance with 60.3% total Faradaic efficiency of C 2+ products (FE C2+ ) and an ethanol selectivity (FE ethanol ) of 55% at only −0.35 V. In contrast, the control catalysts, including similar Cu nanoparticles without Cu/CuNC interface sites, Cu-N-C SACs, and the physical combination of Cu nanoparticles and Cu-N-C SACs, all show a negligible ethanol yield. Together with the comparison with systematically designed control catalysts, these results suggest the critical role of Cu/CuNC interface sites in enhancing the conversion efficiency of ECR to C 2+ products, providing new insights and a strategy for exploring highly efficient ECR electrocatalysts by creating interfaces with electronically asymmetric dual-site centers.

RESULTS AND DISCUSSION
DFT calculations were firstly applied to investigate the influence of the Cu/CuNC interface sites on the C-C coupling process. We constructed three model systems: Cu (100) single crystal, and Cu 20 clusters supported on N-doped porous carbon and graphene, denoted as Cu (100), Cu/CuNC and Cu/C, respectively. As shown in the charge-densitydifference maps of Fig. 1a and b, the Cu/CuNC structure exhibits more significant electron transfer at the interface sites than the Cu/C structure on the charge density difference maps. Furthermore, the asymmetric electron distribution at the Cu/CuNC interface sites demonstrates that the Cu δ+ /Cu 0 structure was established with Cu-N sites as positively charged (yellow area) Cu δ+ , and the negatively charged (cyan area) top Cu atoms as Cu 0 sites. Then, we calculated the adsorption energies of key intermediate * CO. Cu/CuNC and Cu/C models show a preferred * CO adsorption on the interface sites and the top Cu sites, respectively ( Fig. 1c and Supplementary Fig. 1). It is thus reasonable to consider the Cu-Cu/Cu-N-C interface sites in Cu/CuNC and the top Cu sites in Cu-C as reaction centers. In addition, the reaction free energy calculations indicate that * CO is easily formed at the Cu/CuNC interface ( Supplementary Fig. 2), which is highly favorable for the next C-C coupling step. The C-C coupling barriers are further calculated as shown in Fig. 1d. The formation of C-C bonds is the most critical step in the process of generating C 2+ products, while it is also the most elusive [21][22][23]. Given the coordination saturation degree of carbon atoms in intermediates and the concentration of various intermediates on the catalyst surface, * CO- * CO, * CO- * CHO and * CO- * COH coupling are proposed as the three most dominant paths to produce multi-carbon products [23]. The calculated reaction energies show that the * CO- * CHO coupling is thermodynamically most favored among the three coupling paths ( Supplementary Fig. 3), thus the barrier of * CO- * CHO coupling is used to represent the formation trends of C 2+ products (Fig. 1e). It can be clearly seen that the Cu-Cu/Cu-N-C interface sites significantly decreased the reaction barrier of * CO- * CHO coupling (0.30 eV) compared to those on Cu/C (0.72 eV) and Cu (100) (0.77 eV) surface sites. Apart from the low kinetic barrier, the C-C coupling process on the Cu/CuNC interface sites is also thermodynamically more favorable. This suggested that C-C coupling is likely no longer the main obstacle to generating multi-carbon products, and the proton-coupled electron transfer (PCET) step, such as * CO to * CHO, became the rate-limiting step. The strong dependence of the PCET step on the electrode potentials could lead to the distribution of reduction products being very sensitive to the applied potential [19].
To construct the Cu/CuNC structure experimentally, we developed an electroreduction method to form the Cu/CuNC interface sites in situ from the high-density N-anchored single-site Cu sample (Cu-N-C). As illustrated in Fig. 2, the as-described ER-Cu/CuNC catalyst with abundant Cu/CuNC interface sites was constructed by electrochemically reducing a Cu-N-C SAC supported on the threedimensional (3D) honeycomb-like porous carbon ( Supplementary Fig. 4a), which was prepared according to our previous report [24]. The structure of the Cu-N-C precursor was first characterized by transmission electron microscopy (TEM) and scanning transmission electron microscopy (STEM) (Fig. 3a and b and Supplementary Fig. 4b) with energy-dispersive X-ray spectroscopy (EDS) analysis ( Supplementary Fig. 5). No particles were observed on the 3D carbon support with the uniformly elemental distribution of Cu, suggesting the atomic dispersion of Cu. No Cu peaks appeared in the powder X-ray diffraction (XRD) patterns ( Supplementary Fig. 4c), and atomic-dispersed Cu observed by the atomic-resolution high-angle annular dark-field aberration-corrected scanning transmission electron microscopy (HAADF-AC-STEM, Fig. 3c) corroborated the successful synthesis of the Cu-N-C SAC. For the ER-Cu/CuNC catalyst, Cu nanoparticles with a diameter of 3.60 ± 0.58 nm that were uniformly dispersed on the carbon support were obtained after the in-situ electroreduction process (Fig. 3d). A lattice space of 0.21 nm was observed on the high-resolution TEM image with fast Fourier transform (FFT) analysis ( Fig. 3e and f), corresponding to the (111) facet of Cu. The ER-Cu/CuNC catalyst ( Supplementary  Fig. 6) presented a typical pattern of porous carbon, representing the low content and the ultrafine crystal size of the in-situ-formed Cu nanoparticles [3]. More importantly, the combined HAADF-AC-STEM and high-resolution electron energy loss spectroscopic (EELS) images in Fig. 3g exhibit the in-situ-grown Cu nanoparticle surrounded by abundant N, implying the formation of the Cu-Cu/Cu-N interface sites. For comparison, the samples with only Cu-Cu sites (Cu nanoparticles with a diameter of 5.18 ± 0.83 nm, Cu/C, Supplementary Fig.  7), Cu-N sites (copper phthalocyanine with welldefined CuN 4 sites on carbon support, CuPc/C, Supplementary Figs 8-10), a physical combination of Cu-N and Cu-Cu sites (CuPc molecules supported on the above Cu/C, CuPc-Cu/C, Supplementary Fig. 11) and Cu nanoparticles with a diameter of 4.72 ± 0.93 nm supported on N-doped porous carbon (Cu/NC, Supplementary Figs 12 and 13) were also prepared on the same porous carbon support. To avoid the interference of metal loading on activity, the feeding of inorganic copper sources was kept the same (Supplementary Table 1). The X-ray photoelectron spectra (XPS) revealed the surface states of as-described samples. The survey spectrum verified the presence of C, N, O and Cu elements on the Cu-N-C SAC precursor (Supplementary Fig. 14a). The high-resolution N 1s spectrum ( Supplementary Fig. 14b) showed the major N species on the Cu-N-C surface was pyridinic N. Further, the combined Cu 2p spectra with Auger spectrum showed that the Cu atoms were positively charged on the Cu-N-C surface ( Supplementary Fig. 14c and d), indicating a typical state of the single-site Cu structure. For ER-Cu/CuNC, the fresh-made catalyst was directly dried under an N 2 atmosphere and immediately transferred for Ar cluster sputtering to avoid the oxidation of copper. As shown in Fig. 4a and b, the primary Cu state on the surface of the catalyst remained in Cu 0 /Cu + at 932.5 eV during the formation of the ER-Cu/CuNC. The Auger spectra of Cu LMM further revealed the coexistence of zero-valent Cu (918.5 eV) and positively charged Cu on the surface of the ER-Cu/CuNC sample [25,26]. The existence of positively charged Cu species could be attributed to the strong interaction from the surface Cu-N structures shown in Fig. 4c [27].
The operando X-ray absorption near-edge structure (XANES) measurement was applied at Cu Kedge to further investigate the evolution of the Cu chemical state and local structure during the formation of the ER-Cu/CuNC catalyst. As shown in Fig. 4d and e, compared with CuPc, the timedependent negative shift of the adsorption edge represented the decreasing average valence of Cu, demonstrating the formation of Cu 0 species from the reduction of positively charged Cu [27,28]. Moreover, consistent with in-situ XANES spectra, the extended X-ray absorption fine structure (EX-AFS) analysis shown in Fig. 4f revealed the formation of Cu-Cu bonds located at ∼2.2Å on the horizontal axis during the electrochemical treatment. By contrast, the simultaneous decreasing of Cu-N bonds located at ∼1.5Å on the horizontal axis suggested the in-situ growth of Cu nanoparticles from the N coordinated single-atomic Cu. Combined with the above EELS mapping and XPS results, these results suggest the formation of the Cu-Cu/Cu-N interface sites from Cu-N-C SACs in the ER-Cu/CuNC catalyst.
The ECR performances of as-prepared catalysts were evaluated in an H-type cell using 0.1 M KHCO 3 aqueous solution as electrolytes (please see details in Methods and the Supplementary Data). As shown in the linear sweep voltammetric curves (Fig. 5a), the larger current density before −0.5 V in CO 2 -saturated electrolytes than in Ar suggests higher ECR activity than hydrogen evolution at low overpotentials on the ER-Cu/CuNC catalyst. Then, the potential-dependent selectivity for ECR products of the ER-Cu/CuNC catalysts was analyzed in the range of −0.30 to −0.70 V. As shown in Fig. 5b and Supplementary Fig. 15, the products mostly contained H 2 and CO in the gaseous phase, and ethanol, acetate and formate in the liquid phase, respectively. The ER-Cu/CuNC catalyst exhibited excellent selectivity toward C 2+ products at low overpotentials and achieved a maximal FE ethanol of 55% and FE C2+ products of 60.3% with a total current density of 0.98 mA m −2 at −0.35 V. In addition, the maximal FE ratio of C 2+ /C 1 could reach 17.2 at −0.35 V, indicating a high C-C coupling efficiency ( Supplementary Fig. 16). No appreciable decays of current density and FE C2+ (Fig. 5c), as well as no structural changes ( Supplementary Fig.  17) after a 6 h continuous test at −0.35 V were observed, indicating the excellent stability of the ER-Cu/CuNC catalyst. The comparison of applied potential and FE ethanol between this work and other previously reported Cu-based catalysts with FE C2+ above 50% in Fig. 5d [29][30][31][32][33][34][35][36][37][38] (please see detailed information in Supplementary Table 2), suggests that the construction of Cu-Cu/Cu-N interface sites with the asymmetric electronic structure center is an efficient strategy to develop C 2+ -selective electrocatalysts at low overpotential.
To get further insights into the improved C-C coupling on the ER-Cu/CuNC catalyst, a series of control catalysts were systematically designed and evaluated for ECR under the same conditions, including similar Cu nanoparticles without Cu-N-C sites (Cu/C, Fig. 6a and Supplementary  Fig. 18), well-defined molecular model Cu-N-C sites without Cu nanoparticles (CuPc/C, Fig. 6b and Supplementary Fig. 19), the physical combination of Cu nanoparticles and molecular CuPc without the formation of the Cu/CuNC interface sites (CuPc-Cu/C, Fig. 6c and Supplementary  Fig. 20) and Cu nanoparticles on N-doped carbon support without the interface sites of Cu/CuNC (Cu/NC, Fig. 6d and Supplementary Fig. 21). Both Cu/C and CuPc/C samples delivered low-efficiency C 2+ products, indicating the sole Cu-Cu or Cu-N 4 sites could hardly perform the C-C coupling at low potential. Furthermore, the low FE C2+ obtained on the CuPc-Cu/C and Cu/NC samples suggests the necessity of the tightly combined high-density Cu-Cu/Cu-N interface sites for forming C 2+ products. The real-time ECR activity during the ER-Cu/CuNC formation is provided in Supplementary  Fig. 22. The increase in FE C2+ during the formation of Cu-Cu/Cu-N interface sites (dependent on the electrochemical reduction time) suggests that the enhanced C-C coupling activity was from the insitu-formed Cu-Cu/Cu-N interface sites. The FE of <100% was due to the ongoing in-situ reduction of Cu-N-C to Cu nanoparticles. The above results sup-port that the excellent FE C2+ at low overpotential on the ER-Cu/CuNC catalyst should be attributed to the integrated Cu-Cu and Cu-N interface sites as predicted by the DFT calculations in Fig. 1. Besides, it is noted that the ER-Cu/CuNC also delivered much higher FE C2+ at higher potentials than all controls without such interface sites, which is to be expected with boosted C-C coupling. The decrease of FE C2+ with the increase of potentials could be attributed to the limited mass transfer of CO 2 in the H-type cell, which could be further improved by the smart design of advanced electrolysis devices and the inhibition of H 2 evolution.

CONCLUSION
In conclusion, with the assistance of DFT calculations, we successfully constructed Cu/CuNC interface sites for high-efficiency CO 2 conversion to C 2+ products at low overpotentials by forming ER-Cu/CuNC electrocatalysts via in-situ electroreduction of the Cu-N-C SAC. The EELS, XPS and in-situ X-ray absorption spectra (XAS) investigations confirmed the formation of Cu-Cu/Cu-N interface sites. The resulting ER-Cu/CuNC electrocatalyst exhibited an excellent C 2+ selectivity up to 60.3% with an FE ethanol of 55% at a low potential of −0.35 V, standing out from the Cu-based ECR electrocatalysts for C 2+ products. Systematically designed control experiments revealed that such a superior ECR performance at low potential should be ascribed to significantly enhanced electrocatalytic C-C coupling on Cu-Cu/Cu-N interface sites. DFT calculations showed that the Cu/CuNC interface site with an asymmetric electronic structure center was favorable for promoting the adsorption strength of the * CO intermediate and decreasing the reaction barrier of C-C coupling compared to those on Cu-Cu sites. These findings suggest an attractive strategy to promote CO 2 -to-C 2+ conversion at low potential via designing tightly compounded interface sites and decreasing the energy barrier of the C-C coupling   process, providing new insights into advanced catalyst design for producing specific products in ECR and other electrocatalytic reactions.

DFT calculations
All DFT calculations were performed using the Vienna ab initio simulation package (VASP) with the Perdew-Burke-Ernzerhof (PBE) exchangecorrelation functional [39][40][41][42]. In structural relaxation, the total energy and the force on each relaxed atom were converged to 10 −4 eV and 0.02 eVÅ −1 , respectively. The plane-wave cutoff energy was 450 eV, and a k-mesh of 3 × 3 × 1 was adopted to sample the Brillouin zone. The van der Waals interactions were described by the density functional dispersion correction (DFT-D3). The climbing-image nudged elastic band (CI-NEB) method was used to locate the minimum-energy path [43,44]. Four images are uniformly distributed along the diffusion path connecting the initial and final states during the NEB calculation. The configurations of initial states and final states were fully relaxed.

Preparation of porous carbon
Porous carbon was prepared according to our previous report [24].

Preparation of Cu-N-C
Typically, 1.0 mmol of Cu(NO 3 ) 2 ·3H 2 O was first dissolved in 5 mL of deionized water. Then, 1.2 g of α-D-glucose and 60 mg of porous carbon were added to the above solution and sonicated for 30 min, followed by a 12 h store. The precipitates were collected by a centrifuge, dried overnight and then grounded by melamine with a mass ratio of 1:5. The obtained powder went through a pyrolyzation at 800 • C for 2 h under an Ar atmosphere.

Preparation of ER-Cu/CuNC
The ER-Cu/CuNC was prepared via the in-situ electrochemical reduction process of the as-prepared Cu-N-C. The Cu-N-C sample was firstly loaded on the working electrode following the process of working electrode preparation, then electrochemically reduced at -0.30 V vs. RHE for 2 h in 0.1 M KHCO 3 electrolytes under CO 2 reduction conditions to form the ER-Cu/CuNC catalysts. The asprepared ER-Cu/CuNC catalysts were directly applied to the ECR performance test without further processing.

Electrochemical measurements
All electrochemical measurements were conducted on a CHI660E electrochemical workstation in a standard three-electrode system with an H-cell configuration. The as-prepared electrode (please see details in the Supplementary Data) served as the working electrode, the Ag/AgCl electrode with saturated KCl solution served as the reference electrode, and the graphite rod served as the counter electrode. The anode and cathode compartments contained 15 mL 0.1 M KHCO 3 aqueous electrolyte with a headspace of 10 mL, separated by a Nafion-117 proton exchange membrane.