Surface hydroxide promotes CO2 electrolysis to ethylene in acidic conditions

Performing CO2 reduction in acidic conditions enables high single-pass CO2 conversion efficiency. However, a faster kinetics of the hydrogen evolution reaction compared to CO2 reduction limits the selectivity toward multicarbon products. Prior studies have shown that adsorbed hydroxide on the Cu surface promotes CO2 reduction in neutral and alkaline conditions. We posited that limited adsorbed hydroxide species in acidic CO2 reduction could contribute to a low selectivity to multicarbon products. Here we report an electrodeposited Cu catalyst that suppresses hydrogen formation and promotes selective CO2 reduction in acidic conditions. Using in situ time-resolved Raman spectroscopy, we show that a high concentration of CO and OH on the catalyst surface promotes C-C coupling, a finding that we correlate with evidence of increased CO residence time. The optimized electrodeposited Cu catalyst achieves a 60% faradaic efficiency for ethylene and 90% for multicarbon products. When deployed in a slim flow cell, the catalyst attains a 20% energy efficiency to ethylene, and 30% to multicarbon products.

Performing CO 2 reduction in acidic conditions enables high single-pass CO 2 conversion efficiency. However, a faster kinetics of the hydrogen evolution reaction compared to CO 2 reduction limits the selectivity toward multicarbon products. Prior studies have shown that adsorbed hydroxide on the Cu surface promotes CO 2 reduction in neutral and alkaline conditions. We posited that limited adsorbed hydroxide species in acidic CO 2 reduction could contribute to a low selectivity to multicarbon products. Here we report an electrodeposited Cu catalyst that suppresses hydrogen formation and promotes selective CO 2 reduction in acidic conditions. Using in situ time-resolved Raman spectroscopy, we show that a high concentration of CO and OH on the catalyst surface promotes C-C coupling, a finding that we correlate with evidence of increased CO residence time. The optimized electrodeposited Cu catalyst achieves a 60% faradaic efficiency for ethylene and 90% for multicarbon products. When deployed in a slim flow cell, the catalyst attains a 20% energy efficiency to ethylene, and 30% to multicarbon products.
The electrocatalysis community has demonstrated high selectivity toward multicarbon (C 2+ ) products in the CO 2 reduction (CO 2 R) reaction [1][2][3][4][5] . In particular, the faradaic efficiency (FE) to ethylene (C 2 H 4 ) in CO 2 R has increased at an impressive rate in systems operating in neutral and alkaline conditions 2,6-11 . Unfortunately, the single-pass carbon efficiency (SPCE) (i.e. the utilization of CO 2 ) has till now remained low 12,13 . A primary cause is the rapid conversion of CO 2 to (bi)carbonate in these mid-to high-pH conditions. This imposes a substantial energy cost penalty for carbonate regeneration and CO 2 (re)capture 12,14 .
Performing CO 2 R in acidic conditions provides a potential route to high SPCE of CO 2 12,14,15 . However, to date, the selectivity of CO 2 R in acidic conditions has been limited by the competing hydrogen evolution reaction (HER) 12,14 , which benefits from a faster kinetics and a lower reaction overpotential 16 . Consequently, the energy efficiency (EE) for producing C 2+ chemicals in acidic CO 2 R has remained well below that in neutral and alkaline conditions. The activity and selectivity of CO 2 R can be tuned by tailoring the interaction between reaction intermediates and the Cu surface 12,[17][18][19][20][21][22] . Achieving a high coverage and long residence time of CO on the surface of the catalyst can help acidic CO 2 R compete successfully against HER 22 . Previous CO 2 R studies in neutral/alkaline conditions showed that adsorbed hydroxyl groups (OH * , * indicating surface site) form on the Cu surface during CO 2 R due to an increase in local alkalinity [23][24][25] . Such species have been shown to affect the adsorption energy of CO and the reaction rate of C-C coupling 26 . In acidic CO 2 R, the local pH at the electrode surface is near-neutral under moderate current densities 14 , a fact that will limit the concentration of surface OH * and lead to low C 2+ selectivity in acidic CO 2 R.
Here we focus therefore on a strategy to increase the concentration of OH * on the catalyst surface during acidic CO 2 R. We report that it suppresses HER and enhances C-C coupling. Using density functional theory (DFT) calculations, we found an increase in the CO adsorption energy and a decrease in the barrier for C-C coupling when a submonolayer coverage of OH * exists on the Cu surface. By tailoring synthesis conditions, we synthesized catalysts that maintain a high concentration of co-adsorbed CO and OH during acidic CO 2 R. Using time-resolved in situ Raman spectroscopy, we find that the residence time of CO on the catalyst surface increases in the presence of OH, a finding we correlate with higher C 2+ selectivity and suppressed HER activity. The method of catalyst synthesis involves electrodepositing a Cu catalyst in situ during CO 2 reduction (EC-Cu). This leads to achieving >50% and 90% FE for C 2 H 4 and C 2+ , respectively, with 10 h of operating stability. When we add poly(amino acid) promotors, the catalyst achieves a C 2 H 4 FE of 60% and a C 2 H 4 EE of 20%.

DFT calculations
We first applied DFT calculations to investigate the free energy of CO adsorption (ΔG CO* ) on Cu(100) as a function of the surface coverage of CO (θ CO ) and OH (θ OH ). We chose Cu(100) as the model surface due to its high activity for the C-C coupling reactions 10,[27][28][29] and investigated the dependence of ΔG CO* on the θ CO and θ OH up to 3/9 of a monolayer (ML) ( Supplementary Fig. S1). We found that ΔG CO* is similar at different θ CO without co-adsorbed OH (Fig. 1a). However, an increase in CO binding strength is observed for θ CO between 1/9 and 2/9 ML when the θ OH is at 2/9 ML, indicating that the ΔG CO* can be fine-tuned with co-adsorbed OH. Comparing the barrier for C-C coupling with and without OH * (Fig. 1b, Supplementary Figs. S2-5), we found a lower activation energy for CO dimerization and a lower free energy for OCCO (ΔG OCCO* ) on Cu(100) with nearby adsorbed OH. A similar trend on the influence of OH * in CO dimerization is also observed for Cu(111) (Supplementary Figs. S6−10). DFT calculation results suggest that introducing OH * species in acidic CO 2 R could be a promising strategy to suppress HER and promote C 2+ formation.
In situ electrodeposition of catalysts during acidic CO 2 R Previous work reported that Cu catalysts with a surface OH * are obtained at high pH by electrochemically reducing a pre-deposited Cu(OH) 2 film 26 . We electrodeposited Cu catalysts (EC-Cu) in acidic conditions by reducing Cu 2+ species during CO 2 R, which generates a locally alkaline condition (H 3 O + mass transport limitation) necessary for the formation of Cu(OH) x pre-catalyst and provides a reductive potential needed to reduce the as-formed Cu(OH) x into Cu (Fig. 2a). The local formation of Cu(OH) x pre-catalyst was found to be necessary to obtain high performance of EC-Cu catalysts-an observation arrived at by comparing with control samples prepared under various deposition conditions ( Supplementary Fig. S11). The EC-Cu catalyst produced by Cu(OH) x -mediated in situ deposition differs in morphology and structure from the electrodeposited Cu catalysts obtained at locally acidic conditions. It also demonstrates a higher C 2+ selectivity (vide infra).
We performed electrodeposition of Cu on a sputtered Cu (sCu) covered polytetrafluoroethylene (PTFE) gas diffusion layer (sCu/ PTFE) in a flow cell during acidic CO 2 R (Fig. 2a). The sCu catalyst, a control, showed a C 2 H 4 FE of 25% and a H 2 FE of 26% (Fig. 1b) 9,11 . During electrodeposition, the selectivity of EC-Cu towards C 2 H 4 improved, reaching~50% C 2 H 4 FE, while the HER selectivity was reduced substantially (Fig. 2c). The FE to C 1 products (CO and CH 4 ) was <1%, lower than that of sCu. We also observed that deposition duration of >14 min has a negative effect on the FE of C 2 H 4 with a concomitant increase in HER.
We relate the time-dependent CO 2 R activity to the morphology change of the deposited catalysts during synthesis. Scanning electron microscopy (SEM) images of EC-Cu show that dendritic structures gradually form on the sCu/PTFE surface with a longer deposition duration (Fig. 2d, e). A complete dendritic structure was formed at 14 min, corresponding to the best performance.
In a Cu 2+ free electrolyte, the optimally deposited EC-Cu catalyst (at 14 min) maintained high CO 2 R selectivities, excluding CO 2 R catalysis by solvated Cu 2+ ions ( Supplementary Fig. S12). Transmission electron microscopy (TEM) shows that different Cu facets exist (Fig. 2f). This is consistent with OH − adsorption experiments showing a mixture of (100), (110), and (111) surface crystal orientations on EC-Cu, and the ratios of these facets differ greatly from that of sCu (Supplementary Fig. S13). Operando X-ray absorption spectroscopy (XAS) was used to track the oxidation state and local bonding configuration of Cu as a function of deposition duration (Fig. 2g, Supplementary Fig. S14). The EC-Cu catalysts were metallic and showed Cu coordination similar to that of sCu.

Experimental mechanistic studies
We employed time-resolved in situ Raman spectroscopy to probe the surface-adsorbed CO-an important reaction intermediate in CO 2 R. When CO chemisorbs on the Cu catalysts, its characteristic stretching frequency (ν(CO atop )) ranges from 1850−2100 cm −1 . The values of ν(CO atop ) depend on multiple factors, including the adsorption site (atop vs. bridge sites) and surface CO coverage (θ CO ). Based on the peak positions, we assign the peaks at 2082 and 2042 cm −1 to CO molecules adsorbed at the atop sites of the Cu surface on sCu and EC-Cu, respectively (Fig. 3a). The high-frequency band (HFB) at 2082 cm −1 is assigned to CO atop on isolated step sites, and the lower frequency band (LFB) at 2042 cm −1 is attributed to CO atop on terrace sites 17 30 . A consequence of stronger CO atop binding on EC-Cu is a longer residence time on the catalyst surface, which could benefit CO-CO coupling for C 2+ product formation. We evaluate CO residence time on the Cu surface by monitoring the decrease in ν(CO atop ) peak intensity after changing the applied potential from high (−1.5 V) to low (−1.2 V) rate conditions. For sCu and EC-Cu, the ν(CO atop ) peak intensity decays exponentially with time constants (τ) of 0.33 and 0.61 s, respectively (Fig. 3d, e). Similar τ values are also observed for the corresponding Cu-CO peak at 360 and 378 cm −1 ( Supplementary Fig. S16, S17). The larger τ values of the 2042 and 378 cm −1 peaks indicate a longer residence time of CO atop on the surface of EC-Cu, which can promote C-C coupling in addition to having a high θ CO . This effect is also in line with the higher C 2+ FE for the EC-Cu (>90%) compared with sCu (<50%). Furthermore, the degraded EC-Cu samples (<65% C 2+ FE) show a high-frequency ν(CO atop ) band with short τ, reaffirming the importance of longer residence time for adsorbed CO atop (Fig. 3c, Supplementary Figs. S18, 19). The larger τ value for EC-Cu is closely associated with the surface OH * group. The greater intensity of the Cu-OH peak at 531 cm −1 for EC-Cu suggests a greater concentration of OH * at the surface compared to sCu under the same applied potential (Fig. 3a). DFT calculations show that the ΔG CO* can increase with the number of nearby adsorbed OH. Experimentally, we observed the interactions between co-adsorbed CO and OH based on a red shift in the ν(CO atop ) peak position to 2042 cm −1 and a lower frequency shoulder at 1972 cm −1 (Fig. 3a, Supplementary Note 1) 26 . Combining in situ characterization of surface OH and CO atop , we propose that EC-Cu enables high surface coverage of co-adsorbed OH and CO, which increases the residence time of CO on the catalyst surface and promotes C-C coupling. Evidence from Raman spectroscopy indicates the presence of OH * on the surface of Cu catalysts. However, since we cannot observe and quantify the concentration of locally generated hydroxyl ions by Raman spectroscopy, we do not rule out the effect of this species on the CO 2 R selectivity (Supplementary Figs. S20, 21).

Acidic CO 2 R performance
We prepared EC-Cu on different conductive substrates and found that sCu/PTFE resulted in the highest FE of C 2 H 4 and C 2+ . The interaction between the 200 nm Cu film (sCu) and electrodeposited Cu species (EC-Cu) has little influence on the performance of the catalysts (Supplementary Fig. S22). The thickness of Cu in sCu also has a minimal impact on the FE of C 2 H 4 , but it significantly affects the stability of EC-Cu. Additionally, the substrates' structure and composition can influence EC-Cu performance ( Supplementary Fig. S23) [31][32][33] . A 200 nm Cu film deposited on a PTFE-coated gas diffusion layer was used as the substrate in all experiments unless otherwise specified. We found that the optimal deposition time of EC-Cu decreased with greater CuSO 4 concentrations (Fig. 4a). We identified two deposition conditions that produce highly selective catalysts for C 2 H 4 as EC-Cu-1 and EC-Cu-2 (optimal electrodeposition duration at 5 mM and 7 mM CuSO 4 ). The EC-Cu catalysts synthesized using the second deposition condition achieved a maximum C 2 H 4 FE of 56%. Over a long-time operation, the average FE for C 2 H 4 , C 2+, and H 2 was maintained at 53%, 90%, and 10%, respectively (Fig. 4b).
In previous acidic CO 2 R works, high current densities were necessary to increase the local pH and promote C-C coupling on Cu catalysts, and a high K + concentration was employed to suppress HER ( Supplementary Fig. S24) 12,14,34 . This was observed for sCu, which showed a lower FE to C 2 H 4 at lower current densities (Fig. 4c, <10% at 75 mA/cm 2 ). In contrast, EC-Cu demonstrated >45% FE to C 2 H 4 from 75 to 250 mA/cm 2 and was able to achieve high C 2 H 4 FE at lower K + concentrations (Fig. 4c, d and Supplementary Figs. S25, 26). We attribute this high selectivity partially to a high surface OH * concentration during CO 2 R, which stabilizes CO and reduces the C-C coupling barrier.
The stability of EC-Cu was also improved compared with sCu, from less than 0.5 h to 10 h (Supplementary Figs. S27-29). We note that the degradation of dendritic structures in EC-Cu accompanies a performance decrease, which correlates with a decrease in surface OH, and a shorter CO residence time (Supplementary Figs. S18, 19). To further increase the FE of C 2 H 4 , we introduced amide-bearing polymers during the catalyst deposition process. We evaluated five poly(amino acids) containing abundant amide groups and different functional side chains and found that poly(Lys, Phe) showed a stable C 2 H 4 FE of 60% (Fig. 4e, Supplementary Figs. S30, 31). A similar promoting effect of amine/amide functional groups toward the C 2 H 4 selectivity in CO 2 R has been reported previously 6,35 . Deploying the poly(Lys, Phe)-modified EC-Cu catalyst, we obtained a SPCE of 70% towards C 2+ products (Fig. 4f) and stable operation for 10 h with 55% FE of C 2 H 4 in acidic conditions (Fig. 4g). Based on a full cell voltage of 3.4 V ( Supplementary  Fig. S32), the EE towards C 2 H 4 is 20%, which is 1.6-fold greater than the previous acidic CO 2 -to-C 2 H 4 record. Figure 3h summarizes the result in comparison with the previous best acidic CO 2 R performance (Supplementary Table 1-3). Based on the performance summary in Fig. 3h, we achieved CO 2 -to-C 2 H 4 conversion with an energy requirement of~266 GJ ton −1 , lower than previous neutral and alkaline CO 2 R systems 13 . Despite the promising performance, practical implementation requires further improvements in stability and selectivity.
In summary, we report an EC-Cu catalyst that selects for C 2 H 4 and C 2+ products in acidic CO 2 R. Preparation via in situ electrodeposition led to a higher surface coverage of OH, and we found that this species interacts with the CO intermediate and increases the CO binding energy. This observation is consistent with DFT calculations showing that sub-monolayer coverage of OH can increase the free energy of CO adsorption and reduce the barrier for C-C coupling. Implementing the EC-Cu catalyst in acidic CO 2 R achieves a high FE toward multicarbon products (60% for C 2 H 4 , 90% for C 2+ ), a high SPCE of CO 2 to C 2+ products (70%), and a high EE (20% for C 2 H 4 , 30% for C 2+ ) simultaneously.   water-cooled Si (111) and InSn (111) double-crystal monochromator covering a photon energy range from 1.7 to 10.0 keV. The XAS data was processed with Demeter (v.0.9.26) 36 . In situ Raman spectroscopy (Renishaw Invia Raman) was performed to investigate CO speciation on sCu and EC-Cu during CO 2 R.

CO 2 reduction in a flow cell
Three chambers constitute the flow cell setup: anolyte chamber, catholyte chamber, and gas flow chamber. Each compartment has an inlet and outlet. The electrolytes were circulated through the flow cell using peristaltic pumps. CO 2 (Linde Gas) flowing through the gas flow chamber was controlled using a mass flow controller (Brooks Instrument, GF100). The CO 2 flow rate was fixed at 40 standard cubic centimeters per minute (sccm) unless stated otherwise. During the CO 2 R reaction, the catalyst-coated working electrodes, Nafion 117 membrane, and a platinum mesh (Fisher Scientific, AA41814FF) counter electrode were sandwiched between the three chambers. The working electrode potential was measured against Ag/AgCl reference electrode (CH Instruments, CHI111P) that was inserted into the catholyte chamber.

CO 2 reduction in a full cell
A slim flow cell was used to measure the full cell performance. The slim flow cell also contains anolyte, catholyte, and gas flow compartments. The catholyte chamber is made of PEEK and has a total thickness of 1/16". The reduced dimension is intended to lower ohmic losses between the cathode and the anode. In the full cell measurement, oxygen evolution reaction takes place on the anode and it was catalyzed by iridium oxide catalyst supported on titanium felt (IrO x -Ti) 37

Electrochemical measurement
All electrochemical measurements were performed using a potentiostat (Autolab PGSTAT302N). The CO 2 R performance was evaluated in a flow cell (three-electrode setup) or a slim flow cell (full cell, twoelectrode setup) under galvanostatic modes. Gas-phase products were quantified using gas chromatography (GC, Shimadzu, GC-2014) equipped with a thermal conductivity detector (TCD) for the detection of H 2 , O 2 , N 2 , and CO, and a flame ionization detector (FID). Helium (Linde, 99.999%) was used as the carrier gas. Liquid products were analyzed using 1 H NMR spectroscopy (600 MHz, Agilent DD2 NMR Spectrometer) with water suppression. Dimethyl sulfoxide (DMSO) was used as the internal reference and deuterium oxide (D 2 O) as the lock solvent. The Faradaic efficiency (FE) of the gas product was calculated based on the following equation: where n is the number of electrons transferred, F is the Faraday constant, v is the CO 2 flow rate, r is the concentration of the gas product in parts-per-million (ppm), i is the total current and V m is the unit molar volume of gas. The energy efficiency (EE) for the formation of ethylene is calculated as follows: where FE C2H4 denotes the FE of ethylene (C 2 H 4 ), E o a and E o c are the standard reduction potentials for the anode and cathode (CO 2 -toethylene) reactions, respectively.
The single-pass carbon efficiency (SPCE) of CO 2 for a particular product is determined as follows: where j is the partial current density of a specific product, n is the number of electrons transferred for every molecule of the product

In situ Raman spectroscopy
A custom-made cell was used to carry out in situ Raman spectroscopy.
In an epi-illumination configuration, a 785 nm laser was used as the excitation source. The laser power was kept lower than 0.20 mW in all experiments to minimize sample damage. The scattered Raman light was collected by a water immersion objective (Leica, 63×, NA 0.9). Raman spectrometer calibration was done with a Si standard.

Electrochemical OH − adsorption
Electrochemical OH − adsorption was performed in an N 2 -saturated 1 M KOH electrolyte. Linear sweep voltammetry was carried out between −0.2 and +0.6 V versus RHE with a sweep rate of 100 mV s −1 . Before the experiment, all copper catalysts were reduced at −0.6 V versus RHE for 3 min.

DFT Calculations
All density functional theory (DFT) calculations were carried out using the Vienna ab initio simulation package (VASP) [38][39][40][41] . The generalized gradient approximation was used with the Perdew-Burke-Ernzerhof exchange-correlation functional 42 . The projector-augmented wave method 43,44 was used to describe the electron-ion interactions and the cut-of energy for the plane wave basis set was 450 eV. To illustrate the long-range dispersion interactions between the adsorbates and catalysts, the D3 correction method was employed 45,46 . Brillouin zone integration was accomplished using a 3 × 3 × 1 and 2 × 3 × 1 Monkhorst-Pack k-point mesh for CO adsorption energy calculations and C-C coupling calculations. A vacuum region of more than 15 Å thickness was included along the perpendicular direction to avoid artificial interactions.
A periodic six-layer model and p(3 × 3) or p(6 × 3) super cell were chosen. A monolayer of charged water molecules was included in C-C coupling calculations. Adsorption geometries of the different states were optimized by a force-based conjugate gradient algorithm, whereas the transition states were located using the climbing imagenudged elastic band method 47 . During the calculations, the three lower layers were fixed and the three upper layers together with water molecules and adsorbates were allowed to relax. The Gibbs free energy (ΔG) was calculated by converting the electronic energy using the equation: ΔG = ΔE + ΔZPE + ∫ ΔC p dT − TΔS, where ΔE, ΔZPE, ΔC p , and ΔS are the differences in electronic energy, zero-point energy, heat capacity and entropy, respectively, and T was set to room temperature (298.15 K).
Energy assessment for the acidic CO 2 R electrolyzer Energy assessment of the acidic CO 2 R electrolyzer was performed using an energy evaluation model similar to that reported in refs. 13,48. This section provides an overview of the model used to obtain the energy intensity of ethylene production in an acidic CO 2 R electrolyzer. To calculate the energy intensity, we utilized the experimentally achieved performance metrics under various operating conditions (for example, under various CO 2 input flow rates). The metrics inputted to the energy model include Faradaic efficiency, single-pass conversion efficiency, full-cell voltage, and current density-all towards ethylene (which is the major product of the acidic CO 2 R system). The model considers the presence of hydrogen (the product of competing HER) in addition to ethylene and CO 2 at the cathode gas stream. The model also considers oxygen (the product of oxygen evolution) as the only product in the anode gas stream. The model considers a pressure swing adsorption (PSA) gas separation unit at the cathodic downstream to recover ethylene from the hydrogen side product and unreacted CO 2 . The CO 2 recovered from the downstream of the cathode is considered to be returned to the cathode inlet for utilization in CO 2 R. An electrolyte requirement of 100 L per m 2 of the active electrode geometric area is considered. This consideration is based on an estimation of the electrolyte requirement for a lab-scale electrolyzer with a geometric flow field area of 5 cm 2 . The electrolyte is considered to be circulating through a closed loop and used for 1 year without replacement. An example calculation for the energy intensity associated with the electrolyzer electricity and cathode separation is provided in Supplementary Note 2.

Data availability
All the data supporting the findings of this study are available within the article and its Supplementary Information. All other relevant source data are available from the corresponding authors upon reasonable request.