Asymmetric Copper‐Sulphur Sites Promote C–C Coupling for Selective CO2 Electroreduction to C2 Products

Sustainable multicarbon e‐chemicals are of particular interest due to their potential future, high market values, and demand. In the direct electrocatalytic formation of multicarbon e‐chemicals from CO2, the elementary C–C coupling by CO dimerization is considered the rate‐limiting step. Here, a generalized surface structural design principle of asymmetric metal pair sites is proposed, explored, and experimentally tested in order to promote CO dimerization on surfaces. First a computational model of N‐doped Cu2S layers featuring adjacent, electronically asymmetric Cuδ1+‐Cuδ2+ (0 < δ1+ < δ2+ < 1) metal atomic pairs evidenced by their non‐uniform charge distribution is considered. The electronic asymmetry resulted in distinct CO adsorption energies and the associated self‐adjusting structures, which lowered C–C coupling energy barriers significantly. The computational hypotheses are experimentally tested using X‐ray photoelectron spectroscopy of Cu‐N moieties within N‐doped Cu2S layers. In‐situ Fourier‐transform infrared spectroscopy confirms linear *CO and *CO‐CO adsorption configuration by the peaks of ≈2080 and 1920 cm−1, respectively. After N‐doping, the catalytically C2 faradaic efficiency can significantly be elevated to 14.72% due to the promotion of C–C coupling.


Introduction
Our energy and climate crisis calls for a future net zero carbon emission scenario, largely relying on today's emerging renewable electricity-based energy technologies. [1]Apart from decarbonization of the energy system, strategies to achieve net zero carbon emissions must include ways to close the anthropogenic carbon cycle in the chemical sector through electricity-based, hence electrochemical, valorization of CO 2 into value-added compounds.Among these value-added chemicals, ethylene (C 2 H 4 ), an DOI: 10.1002/aenm.202304224intermediate in the production of ≈50-60% of all plastics [2] with high energy densities and market prices, has shown promising application prospects. [3]To date, catalyst materials capable of catalytically forming C 2 H 4 are still restricted to copper (Cu)-based materials, as it is the only metal that has shown a propensity to activate C-C coupling through balanced *CO adsorption.Still, Cu often requires high overpotentials to achieve a reasonable reaction rate and selectivity towards the high-valued C 2 products, limiting its widespread practical applications.
There is growing consensus that surface oxygen species along with electrophilic Cu + species with low coordination in oxide-derived materials play a critical role in the activity and selectivity toward C 2+ products. [4]Prior studies have suggested that Cu + /Cu 0 surface species synergistically promote CO 2 reduction to C 2+ products due to an enhanced ability for CO 2 activation and C-C coupling. [5]Yet, more recent work put focuses on controlling the localized surface electronic structure by combining Cu surface atomic sites with other metal atoms in atomic proximity, aiming to obtain C-C coupling on asymmetric atom pairs, thereby accelerating C 2 production by CO dimerization. [6]However, Cu bimetallic systems with Ag, Pd, or Au turned out to have a preference for C 1 rather than C 2 products.To move beyond bimetallic Cu-M alloy catalyst surfaces, designing surfaces with asymmetric diatomic Cu-Cu motifs with distinct electronic structures can be an alternative: CO 2 molecules could be efficiently adsorbed and activated to CO on each of the two Cu sites, while C-C coupling would occur more likely due to the asymmetric nature of the atomic pair.Therefore, designing model surfaces with adjacent Cu sites with asymmetric electronic structures could hold the key for enhanced reactive C-C coupling.
Bearing the above-mentioned concept in mind, we explore the design of adjacent, yet distinct ("asymmetric") Cu + sites with asymmetric electronic localization on thin Cu chalcogenide catalyst layers.We investigate and test the hypothesis whether asymmetric site motifs on Cu 2 S thin films can steer catalytic selectivity of surface reactions.More specifically, we explore whether adjacent asymmetric Cu + sites can enable more effective C-C coupling during CO 2 conversion to C 2+ products such as ethylene.Figure 1 illustrates the concept of a 2D Cu-based chalcogenide thin-layer catalyst where 1) the thin-layer morphology can help expose abundant low-coordinated Cu sites, which serve as catalytic active sites for CO dimerization; 2) dopant metal atom species modify and control the localized electronic structure of adjacent metal surface sites, thereby improving the activation of CO 2 and the C-C coupling process. [7]

Results and Discussion
We take the chalcocite-type Cu 2 S material as a model catalyst and catalyst surface.In the bulk of Cu 2 S materials, each Cu(I) atom exhibits periodic trigonal coordination with sulfur atoms. [8]n the surface of thin layers, the exposed surface Cu + site exhibits lower coordination and lower oxidation state in the range of 0 <  < 1.This can be inferred from the Bader charge analysis, in which the Cu atoms in the bulk of the Cu 2 S material exhibited a charge of 0.30.Correspondingly, as to the adjacent Cu sites of Cu 2 S thin layers, the values decreased to 0.23 and 0.10, respectively, suggesting a lowering of Cu oxidation state and the non-uniform electronic configuration within the thin layers.This distinct charge distribution could be further enlarged with the introduction of nitrogen dopants.As illustrated in Figure 2A, the charge of the corresponding Cu pair in N-doped Cu 2 S layer shifted to 0.29 and 0.06, wherein the Cu atom with the N coordination possesses a higher oxidation state.Such dissimilar charge allocation induced a non-uniform electronic polarization among the adjacent Cu atom, implying the formation of the asymmetric Cu 1+ -Cu 2+ site pair (0 <  1 + <  2 + < 1, Figure 2B).Clearly, with the enhanced disparity of charge redistribution, the asymmetric degree of the site pair in N-doped Cu 2 S was higher than that in pristine layers.To clarify how these asymmetric Cu 1+ -Cu 2+ pair sites could affect the C-C coupling process, the CO adsorption on both surfaces was investigated.As shown in Figure 2C, the CO adsorption energies of asymmetric Cu 1+ -Cu 2+ pair sites in Cu 2 S layers were −1.12 and −1.39 eV, while for the N-doped case, the values were decreased to −0.75 and −1.14 eV, respectively.That is, the CO binding strength was slightly diminished with nitrogen doping, which can facilitate a subsequent CO-CO dimerization. [1,9]Furthermore, the difference in CO adsorption between two adjacent disparate Cu sites scaled with the degree of asymmetry (disparity).Based on asymmetric CO adsorption on N-doped Cu 2 S, the reaction barrier of C-C coupling (0.79 eV) was lowered as compared to that of non-doped layers (1.00 eV, Figure 2D,E ; Figure S1, Supporting Information).Specially, one can notice that the nitrogen doping of Cu 2 S layer also triggered the self-adjustment of water layer configuration in the initial 2*CO state, which approached closely to the final *OCCO state C-C coupling, thus promoting the C-C coupling process (Figure S2, Supporting Information).Overall, our theoretical results suggested that the electronic and geometric asymmetry of the active sites induced by the nitrogen doping led to the localized electronic polarization, and then manipulated the corresponding adsorption and coupling behavior of the reactive *CO intermediates.As a result, the reaction energy barrier was lowered and thus the selectivity of C 2 products could be effectively regulated during CO 2 electroreduction.
To test our theoretical hypotheses experimentally, asymmetric pairs of Cu-S surface sites were synthesized using nitrogen (N) doping inside the chalcocite-type Cu 2 S thin layers (Figure 3A inset). [8]We successfully synthesized N-doped copper sulfide thin layers by nitriding the precursor of copper sulfide thin film layers (Figure S3, Supporting Information).X-ray diffraction (XRD) pattern of the products in Figure 3A could be readily indexed to chalcocite Cu 2 S, corresponding to JCPDS No. 72-1071.Moreover, transmission electron microscopy (TEM) image in Figure 3B exhibited their freestanding and sheet-like morphology with a lateral size of ca.200 nm.High-resolution TEM image in Figure 3C revealed that the interplanar distances of 3.77 and 3.26 Å corresponded to the (101) and (012) lattice plane dis-tances of the chalcocite Cu 2 S.Meanwhile, their dihedral angle of 78°was also consistent with the calculated angle between (101) and (012) planes, implying the [120] orientation of the synthesized Cu 2 S thin layers.Furthermore, as revealed by the X-ray photoelectron spectra (XPS) in Figure 3D, the N 1s spectra showed a peak with binding energy of 399 eV, which could be attributed to the interaction of Cu species with nitrogen, [10] indicating successful doping in the Cu 2 S thin layers and the formation of S-Cu-N sites, also denoted as N-doped Cu 2 S thin layers.By comparison, the undoped and N doped-CuS + Cu 7 S 4 mixed phase thin layers were fabricated by tuning the annealing atmosphere and temperature (denoted as CuS + Cu 7 S 4 thin layers and Ndoped CuS + Cu 7 S 4 thin layers, Figures S4 and S5, Supporting Information).
To explore the effect of N doping on the ability to form C 2 products by CO dimerization on Cu 1+ -Cu 2+ site pairs, electrocatalytic activity and faradaic efficiency tests during CO 2 electroreduction were conducted for N-doped Cu 2 S thin layers, Ndoped (CuS+Cu 7 S 4 ), and undoped reference materials.The electrochemical measurements were carried using stationary glassy carbon (GC) backing electrode on which the catalytic active Cu-S catalyst was loaded.Experiments were carried out using an H-cell at a catalyst loading of 0.10 mg cm −2 .Table S1 (Supporting Information), Figures 4A-C, and S6 (Supporting Information) correlate structural catalyst characteristics with experimental electrocatalytic product efficiencies.Structurally, XPS quantitative analysis in Table S1 (Supporting Information) showed the amount of N dopants and the degree of low-coordinated trigonal Cu increased monotonically in the order "undoped CuS", "Ndoped Cu mixed phase" and "N-doped Cu 2 S thin layers".The product selectivity (Figure 4A,B; Figures S6, S7 and Table S1, Supporting Information) evidenced that the undoped CuS thin layers (Figure 4B) exhibited a faradic efficiency for formic acid, FE(HCOOH), of 20.64% at −0.98 V RHE with trace amount of ethylene; by contrast, at the same electrode potential, the N-doped Cu 2 S thin layers (Figure 4A-C) was able to produce a relatively high FE(C 2 H 4 ) for ethylene of 14.72% at a lower FE(HCOOH) of 13.68% as compared to the commercial Cu 2 S and also the reported data for Cu 2 S. [11] We attribute this ability to form C 2 products, such as ethylene, to the presence of reactive Cu 1+ -Cu 2+ site pairs after N doping, where adsorbed CO dimerization can occur.A similar comparison was done for the undoped and Ndoped (CuS + Cu 7 S 4 ) mixed-phase thin-layer catalysts.While the undoped mixed-phase catalyst showed a FE(HCOOH) of 22.01% and no C 2 products, the N-doped (CuS + Cu 7 S 4 ) mixed-phase evidenced the formation of ethylene at lower FE(HCOOH).These results suggest the emergence of reactive adjacent Cu 1+ -Cu 2+ site pairs for promotion of C-C coupling process by CO dimer-ization.Besides, the formation of HCOOH likely followed the C-bound intermediate route for the N-doped Cu 2 S thin layers, as evidenced by the observed decrease in HCOOH FE and the concurrent increase in C 2 H 4 FE.In other words, the Cu 1+ -Cu 2+ sites impeded HCOOH desorption, inducing the generation of CO intermediates and subsequent CO dimerization.To verify the chemical stability of the Cu sulfide crystal phases over the 15 minute-measurements, X-ray diffraction pattern and N photoemission spectroscopy were carried out after the electrochemical tests: Figure S8 (Supporting Information) confirms the structural, chemical, and morphological stability of the N doping in the crystalline Cu 2 S catalysts during the electrochemical test protocols.While these techniques do not provide chemical information on the topmost surface layer, they exclude bulk structural transformations into metallic Cu or the leaching or N dopants from the bulk of the catalyst.
To investigate the CO 2 activation process on Cu sulfides at the catalyst surface, in-situ Fourier-transform infrared spectroscopy (FT-IR) measurements were performed to probe surface reaction intermediates on N-doped Cu sulfide thin layers.These were compared to those of undoped and N-doped (CuS + Cu 7 S 4 ) thin layers, as well as undoped CuS thin layers.As shown in Figure 4D,E   potential, a distinct band of an absorbed *CO intermediate at ≈2080 cm −1 emerged.Similarly, for N-doped Cu 2 S thin layers, two IR bands aroused with increasing current density (Figure 4D), where the peak at ≈2080 cm -1 indicated the stretching vibration mode of absorbed CO* at a top site.According to Masuda and Hwang's previous work, [12] the additional peak at lower wavenumber (≈1920 cm -1 ) stemmed from interactions between absorbed CO* intermediates (inset in Figure 3D), which benefits C 2+ production.By contrast, the N-doped (CuS + Cu 7 S 4 ) mixed-phase thin layers only revealed the band at ≈2080 cm -1 , while no interaction was observed at ≈1920 cm −1 , pointing to a much less favorable C-C coupling on this electrode, which is consistent with the experimental CO 2 RR performance.These findings demonstrated the formation of key intermediates of *CO, and account for the dominant C-C coupling process during the electrolysis.

Conclusion
In conclusion, this contribution has combined theory and experiments to explore catalytic reactivity effects on N-doped Cu sulfides, on which the formation of electronically asymmetric Cu site pairs was predicted to induce lower barriers for C-C coupling and C 2 product formation.Computational Cu-S thin layer material designs featured Cu 1+ -Cu 2+ site pairs (0 <  1 + ≠  2 + ) with asymmetric charge distribution, able to kinetically accelerate C-C coupling to produce C 2 hydrocarbons during CO 2 electroreduction.Experimentally, N-doped crystalline Cu 2 S thin layers and their undoped references were successfully synthesized and electrochemically tested.The distinct ability of the N-doped thin film catalysts to generate C 2 compounds, such as ethylene, was confirmed suggesting catalytically active Cu-S site pairs on the doped surfaces.Stability tests excluded structural and compositional degradation of the thin films into metallic Cu.Calculations were combined with in-situ FT-IR spectroscopic investigations of the surface of the Cu sulfide catalysts.The in-situ results unveiled clear difference in surface binding of CO be-tween doped and undoped catalysts that appear to support the hypothesis that distinct N-doped (asymmetric) Cu 1+ -Cu 2+ motifs could be responsible for the reactive ability to dimerize CO.This work demonstrates yet another example of a simple general strategy to effectively steer the selectivity toward C-C products during CO 2 electroreduction by the rational design of surface site pairs.

Experimental Section
Sample Preparation: Synthesis of CuS Thin Layers: A modified procedure based on the literature [13] was proposed to increase the output of CuS thin layers.In a 50 mL flask, 300 mg CuCl was introduced into a mixture of 10 mL octylamine and 10 mL oleylamine.The system was heated to 100 °C for 30 min under a flowing N 2 atmosphere.Then the solution was further heated to 130 °C for 3 h.Afterward, 288 mg of sulfur (S) was rapidly added into the above system and kept at 95 °C for 5 h.The obtained product underwent multiple washes with absolute ethanol and cyclohexane before being dried under vacuum.
Sample Preparation: Synthesis of N-doped Cu 2 S Thin Layers: The Ndoped Cu 2 S thin layers were obtained by annealing the obtained precursors of CuS thin layers in ammonia.The annealing process was conducted at a heating rate of 10 °C min −1 , and the samples were maintained at 400 °C for 1 h.The obtained powders were collected for further characterization.
Sample Preparation: Synthesis of N-doped Cu 7 S 4 +CuS Thin Layers: The N-doped Cu 7 S 4 +CuS thin layers were obtained by annealing the obtained precursors of CuS thin layers in ammonia.The annealing process was carried out at a heating rate of 10 °C min −1 , and the samples were maintained at 300 °C for 1 h.Subsequently, the obtained powders were collected for further characterization.
Sample Preparation: Synthesis of Cu 7 S 4 +CuS Thin Layers: The Cu 7 S 4 +CuS thin layers were synthesized by annealing the obtained precursors of CuS thin layers in a nitrogen (N 2 ) atmosphere.The annealing process was conducted at a heating rate of 10 °C min −1 , and the samples were maintained at 300 °C for 1 h.The obtained powders were then collected for further characterization.
Characterizations: XRD patterns were acquired using a D8 Advance Diffractometer (Bruker) equipped with a Lynx Eye Detector and KFL Cu 2K X-ray tube.The measurement was carried out at a step size of 0.04°within a 2 range spanning from 10°to 70°.TEM analysis was conducted using a FEI Tecnai G 2 Microscope 20 S-Twin with a LaB6-cathode at 200 kV accelerating voltage (ZELMI Centrum, Technical University Berlin).The samples were dispersed in ethanol, ultrasonicated, and drop-dried onto Cu-grids with a holey carbon film.The analysis of the images was performed using software from ImageJ.
HR-TEM was conducted using a FEI TITAN 80-300n equipped with a high brightness field emission gun (FEG) at ZELMI Centrum, Technical University Berlin.
SEM images were obtained using a JEOL 7401F instrument (Tokyo, Japan) with an accelerating voltage of 10 kV.
XPS measurements were acquired using a K-Alpha + X-ray Photoelectron Spectrometer System (Thermo Scientific).The instrument features a Hemispheric 180°dual-focus analyzer with a 128-channel detector.
Physisorption measurements were conducted by obtaining Kr physisorption isotherms at 77 K using an Autosorb-1 (QUANTACHROME).In order to minimize dead volume, the glass tubes utilized were filled with a glass rod and glass wool.
Prior to measurements, the samples were degassed under vacuum at 90 C for a minimum of 24 h to remove any adsorbates.The Brunauer-Emmett-Teller (BET) method was applied in a pressure range of 0.04 ≤ p/p0 ≥ 0.2 to calculate the overall surface area, employing a multipoint fit.
Operando FT-IR spectroscopy was carried out in a modified attenuated total reflection (ATR)-FT-IR setup.The catalysts were dispersed in water, and the catalyst solution was directly deposited onto the ATR prism to avoid intensity loss associated with the absence of a sputtered layer of a conducting material like Au. Electrochemical connection was established through a carbon cloth material, providing an electrolyte reservoir over the catalyst to minimize mass transport limitations and polarization effects.Spectra were collected in a custom-made glass cell using a Bruker Vertex 70v FT-IR spectrometer equipped with a Mercury-Cadmium-Telluride (MCT) detector cooled with liquid nitrogen.A platinum mesh served as the counter electrode, and a reversible hydrogen electrode (RHE) was employed as a reference, fabricated by purging H 2 over a Pt-mesh.In situ, electrochemical measurements were controlled using a Metrohm Autolab PGSTAT204 potentiostat.An unpolarized beam was focused with a Pike Veemax II onto the sample spot of the cell, with a spectral resolution set to 4 cm −1 .A total of 100 to 107 interferograms were collected and averaged for each presented spectrum.The reference spectrum RE1 was collected in the same electrolyte (0.1 M CO 2 saturated KHCO 3 ) immediately before the investigated potential scan at the respective start potential.A ZnSe hemisphere served as the IR window, and the electrocatalyst sample was deposited on the prism at the IR beam ATR focus spot, contacted with Toray Paper 030 carbon cloth and a Pine glassy carbon rod to fixate the carbon cloth.The entire beam pathway was under vacuum for more than 24 h prior to each measurement.The positions of the bands were determined by taking the band maximum, or, for bipolar bands, by taking the mean value of the maximum and the minimum.The potential was scanned from -100 mV versus RHE to the required potential and held for 10 min, with the background spectrum taken before applying any potential.
Electrochemical Measurements: The CO 2 electrochemical reduction tests were carried out in an H-cell, separated by an anion exchange membrane and operating in a 0.1 M CO 2 -saturated KHCO 3 solution.Polished glassy carbon, platinum mesh, and Ag/AgCl electrode was used as working electrodes, counter electrode (CE), and reference electrode, respectively, measured with a Biologic SP 300 potentiostat.Before and during the electrochemical reaction the working compartment was purged continuously with CO 2 at a flow rate of 30 sccm (standard cubic centimeters per minute) from the bottom of the cell, and the gas atmosphere was controlled with mass-flow controller.All reported potentials were corrected for ohmic drop, determined by electrochemical impedance spectroscopy.EC-Lab software was used to automatically correct 50% of the ohmic drop, while the remaining 50% was corrected manually.For each measurement, fresh electrolyte was used to ensure that adsorbates from previous experiments did not influence the measurements.The volatile products were qualitatively analyzed by a Shimadzu gas chromatograph, while the nonvolatile products were quantified by a Shimadzu gas chromatograph with a special vaporizer head and HPLC-MS (Agilent G6120AA).
Computational Details: All the theoretical calculations were carried out by using the non-empirical Perdew-Burke-Ernzerhof (PBE) exchangecorrelation functional implemented in the CP2K code. [14]The Qucikstep method was employed with the wave functions expanded to the localized double -valence-polarized basis set. [15]To prevent any surface interactions, a vacuum layer of 15 Å was set in this work.The kinetic energy cutoff was adopted to be 500 Ry.The Goedecker-Teter-Hutter pseudopotential was selected to consider the core-electron interactions. [15]The minimum energy path and energy barriers for C-C coupling were simulated by using the climbing-image nudged elastic band (CI-NEB) method. [16]

Figure 1 .
Figure 1.Illustration of the key hypothesis: The rate of C-C coupling by CO dimerization can be manipulated using structurally asymmetric Metal 1+ -Metal 2+ pair sites (0 <  1 + ≠  2 +).Shown is a well-ordered CuS x surface facet with surface Cu atoms adsorbing CO 2 molecules.Dopant atoms introduce electronic asymmetries in Cu site pairs by adjusting electron-withdrawing/donating properties.Asymmetric CO chemisorption energies favor CO dimerization and hence conversion of CO 2 to C 2 products.By contrast, sites with even electronic distributions feature strong dipole-dipole repulsion forces during CO 2 activation, which seriously hinders CO dimerization.

Figure 2 .
Figure 2. A) Bader charge analysis for the Cu 1+ -Cu 2+ pair sites of the Cu 2 S and the N-doped Cu 2 S thin layers.B) The charge density difference for the Cu 1+ -Cu 2+ pair sites of the N-doped Cu 2 S thin layers with the top (left) and side (right) view.The isosurface is set to be 0.006 e Å −3 in which the yellow bubbles stand for electrons while blue bubbles represent holes.C) The CO adsorption energies for the Cu 2 S and the N-doped Cu 2 S thin layers on different Cu active sites.D) The reaction coordinate diagram of C-C coupling on the Cu 2 S and the N-doped Cu 2 S thin surfaces.E) The reaction intermediate configurations of C-C coupling on N-doped Cu 2 S thin layer.The brown, blue, plum red, gray, and red balls illustrate the Cu, S, N, C, and O atoms.

Figure 3 .
Figure 3. Characterization of the N-doped Cu 2 S thin layers.A) XRD pattern (inset: the crystal structure of Cu 2 S), B) TEM image; C) HRTEM image, and D) N 1s XPS spectra.
and Figure S9 (Supporting Information), no obvious IR bands were detected at anodic electrode potentials for the undoped CuS thin layer near −0.1 V versus V RHE , i.e. prior to the onset of CO 2 electrolysis.With more cathodic reactive electrode

Figure 4 .
Figure 4. CO 2 electroreduction faradaic efficiencies of A) the N-doped Cu 2 S thin layers and B) commercial Cu 2 S, respectively.C) Faradaic efficiency of ethylene for the N-doped Cu 2 S thin layers and commercial Cu 2 S at different potentials.In-situ FT-IR spectra of D) the N-doped Cu 2 S thin layers and E) commercial Cu 2 S, respectively recorded at different potential intervals.