Thousand-hour salt precipitation-free CO₂-to-ethylene electrosynthesis at high current densities

Abstract

Salt precipitation remains a bottleneck in scaling carbon dioxide (CO₂) electroreduction to ethylene (C₂H₄), as it blocks gas transport, induces electrode flooding, and causes rapid performance degradation. Identifying electric field-driven cation accumulation and the consequent hydrophobicity loss as key factors, here, we show a wettability-engineered Cu-based electrode that decouples hydrophobicity from electronic conductivity. The electrode incorporates in situ embedded Cu active sites within an insulating, hydrophobic polymer matrix, forming localized “hydrophobic trap”. This architecture preserves interfacial hydrophobicity under operation, suppresses cation accumulation, and confines locally generated OH⁻ ions to sustain an alkaline microenvironment for C-C coupling. The optimized electrode achieves a faradaic efficiency of 75.9% for C₂H₄ at 1.2 A cm⁻² in a flow cell, and operates for over 1000 h in a membrane electrode assembly electrolyzer. The electrode maintains high productivity under low CO₂ concentrations and in the presence of flue gas impurities. Techno-economic analysis confirms the feasibility of this strategy.

The development of robust catalysts that could work under high current densities brings promise but is a challenge in CO₂ electroreduction. Here, the authors report a wettability-engineered electrode design for ethylene electrosynthesis that operates over 1000 h without salt precipitation.

Introduction

The electroreduction of carbon dioxide (CO₂RR) to ethylene (C₂H₄) represents a promising pathway toward closing carbon cycle and enabling sustainable chemical manufacturing^1–3. Among various catalysts, copper (Cu)-based materials remain uniquely capable of facilitating C-C coupling and generating multi-carbon (C₂₊) products due to their optimal binding strength for key reaction intermediates^4–7. Recent advances have pushed a Faradaic efficiency (FE) for C₂H₄ exceeding 60% at high current densities (>200 mA cm⁻²) in alkaline or bicarbonate electrolytes (e.g., KOH and KHCO₃). Strategies such as modulating the Cu oxidation state have been shown to enhance selectivity and durability^6,7, while CO₂RR in acidic media provides additional opportunities for high-efficiency conversion under different operating conditions⁸. However, maintaining high performance over extended operational durations, especially at the thousand-hour scale, remains a major challenge⁸. A primary limitation arises from salt precipitation on the cathode, which blocks active sites, hampers CO₂ diffusion, and leads to rapid electrode deactivation^9–11. This phenomenon is driven by the electric field-induced cation accumulation (e.g., K⁺) near the electrode surface, which lowers the local pH and facilitates the formation of carbonate and bicarbonate precipitates. Meanwhile, the applied electric field repels OH⁻ ions from the cathode interface, weakening interfacial alkalinity that is essential for efficient C-C coupling. As such, mitigating salt precipitation is crucial for sustaining both the selectivity and durability of CO₂-to-C₂H₄ conversion under industrial conditions.

Efforts to regulate ion transport have included the use of cation-repelling coating with positively charged groups^12–14. While these layers offer temporary protection, their hydrophilic nature often leads to water accumulation at the electrode surface, causing electrode flooding, impeding CO₂ access, and intensifying the parasitic hydrogen evolution reaction (HER)^15–17. Alternatively, hydrophobic coatings, most commonly polytetrafluoroethylene (PTFE), have shown improved water management and extended operational stability to a few hundred hours^18,19. However, such coatings remain vulnerable to cation penetration under sustained electric fields and prone to mechanical degradation or delamination, typically requiring replacement within 200 h. Moreover, the insulating nature of PTFE introduces a long-standing trade-off between maintaining hydrophobicity and ensuring efficient charge transport. For example, increasing PTFE content beyond ~50 wt% significantly reduces current density and elevates HER activity, yielding only ~100 mA cm⁻² at 70 wt% PTFE and about 40% H₂ FE at 60 wt% PTFE²⁰. These limitations underscore the need for more robust electrode architectures that can maintain long-term CO₂RR performance without relying on fragile surface modification or composite blending.

In this work, we report a wettability-guided electrode architecture that structurally decouples hydrophobicity from electronic conductivity, enabling long-term CO₂-to-C₂H₄ electrocatalysis without salt precipitation. The electrode comprises an insulating hydrophobic polytetrafluoroethylene (PTFE) framework overlaid with a sparse three-dimensional carbon nanotube (CNT) network for charge transport. Active Cu nanoparticles are in situ embedded within the PTFE/CNT matrix, forming localized hydrophobic microenvironments, akin to a “hydrophobic trap”, that isolate the active sites from cation-rich electrolyte while maintaining access to the gas-liquid-solid interface. The hydrophobic domains also trap locally generated OH⁻ ions, sustaining an alkaline interfacial microenvironment that promotes C-C coupling and enhances C₂H₄ selectivity. The optimized electrode delivers C₂H₄ FE of 75.9% and a C₂₊ FE of 90.8% at 1.2 A cm⁻² in a flow cell, and achieves over 1,000 h of stable operation at 200 mA cm⁻² in MEA without salt accumulation or performance degradation loss. A single pass carbon efficiency exceeding 95% is maintained at a low gas flow rate of 2 sccm, and a 5-cell MEA stack produces C₂H₄ at a rate of 1.2 L h⁻¹ with FE above 70%. The electrode remains highly active even under dilute CO₂ concentrations and in the presence of typical flue gas impurities such as NO, SO₂ and O₂, thus reducing the need for gas purification and lowering operational costs. Techno-economic analysis confirms the viability of this approach. Furthermore, the design concept proves universally applicable, as demonstrated by its extension to Ag- and Sn-based catalysts for the selective production of CO and HCOOH, respectively.

Results

Catalyst design and characterization

The Cu-based catalyst developed in this study addresses a longstanding challenge in CO₂-to-C₂H₄ electrosynthesis, that is, salt precipitation, which results from excessive cation aggregation on the cathode. While traditional surface-modification strategies have shown limited success^18,19,21, they often fail to provide long-term stability due to electrode flooding or coating degradation. In contrast, our approach decouples hydrophobicity from electronic conductivity within the electrode, enabling ion transport regulation in the cathode microenvironment. The key innovation lies in the incorporation of polytetrafluoroethylene (PTFE) particles, a hydrophobic and insulating material, as the primary structural component of the gas diffusion electrode, integrated with a sparse three-dimensional network of carbon nanotubes (CNTs) (PTFE/CNT matrix). This design effectively balances efficient electron transport with the creation of a stable hydrophobic microenvironment, avoiding damage to hydrophobicity by the electrochemical process.

A Cu catalyst embedded in a PTFE/CNT matrix (Cu-PC) was fabricated via in situ electrodeposition Cu nanoparticles onto the CNTs at a cathodic current density of 100 mA cm⁻² (see Supplementary Information). The hydrophobic PTFE/CNT framework stabilizes the Cu catalyst and establishes a durable microenvironment that prevents electrode flooding, enabling continuous CO₂ electroreduction without maintenance. By tuning the CNT content within the matrix, an optimal balance between hydrophobicity and electrical conductivity was achieved, ensuring stable and efficient operation. The Cu-PC catalyst containing 10 wt% CNTs displays a favorable water contact angle of 140°∼145°, effectively repelling the electrolyte to prevent flooding, while maintaining an electrical conductivity of 0.043 Ω (Fig. 1a and Supplementary Fig. 1). In contrast, a higher CNTs content (e.g., 20 wt%) reduced hydrophobicity, with a contact angle below 85°, whereas a lower CNT content reduced conductivity (Supplementary Fig. 1). The optimized hydrophobicity of the Cu-PC catalyst, with a static contact angle of about 140°, is therefore essential for sustaining long-term CO₂ electroreduction performance (Fig. 1b and Supplementary Fig. 4). To evaluate the electrochemical active surface area (ECSA), cyclic voltammetry measurements were performed (Supplementary Fig. 5). The Cu-PC catalyst (C_dl: 18.55 mF cm⁻²) exhibited a higher ECSA than that of commercial Cu reference (Cu Ref, C_dl: 10.99 mF cm⁻², Supplementary Fig. 6). The enlarged ECSA indicates improved accessibility of active sites, which underpins both the high CO₂ reduction activity and the operational stability of the Cu-PC electrode.

Fig. 1. Catalyst design and characterization.

a Schematic of Cu-PC catalyst. b Scanning electron microscopy (SEM) image of the Cu-PC catalyst, along with its air/water contact angle photo. c In-situ X-ray absorption near-edge spectroscopy (XANES) spectra at the Cu K-edge of the Cu-PC catalyst during CO₂RR at various potentials (V vs RHE, 100% iR correction, electrolyte: 1 M KOH, pH = 14). OCV: open-circuit voltage. d The heat map of the corresponding first derivatives of the Cu K-edge XANES spectra of Cu-PC catalyst during CO₂RR. e Coordination number of the first shell Cu-Cu scattering during CO₂RR. Data are presented as mean values  ±  s.d. Relevant source data are provided in the Source Data file.

Scanning electron microscopy (SEM) images confirmed the uniformly distributed Cu nanoparticles with an average size of ∼80 nm, embedded within the PTFE/CNT matrix (Fig. 1b). Elemental mapping verified the homogeneous distribution of Cu, C from CNTs, and F from PTFE across the surface (Supplementary Fig. 3). X-ray diffraction (XRD) analysis identified the metallic nature of Cu nanoparticles, with distinct diffraction peaks corresponding to Cu(111), Cu(100), and Cu(110) crystal planes (Supplementary Fig. 2). X-ray photoelectron spectroscopy (XPS) analysis of Cu-PC catalyst revealed mixed-valence states of Cu. The Cu 2p XPS spectrum showed both Cu⁰ and Cu²⁺ species, along with a prominent Cu^δ+ signal (0<δ<2) (Supplementary Fig. 7a). Cu L₃M₄₅M₄₅ Auger spectrum further supported this observation, showing a ¹G peak at 916.5 eV, intermediate between the characteristic peaks for Cu⁺ at 915.8 eV and Cu⁰ at 918.0 eV, consistent with the presence of Cu^δ+ species. A distinct peak, attributable to ³F transition, confirmed the coexistence of metallic Cu^{0 22,23} (Supplementary Fig. 7b).

To probe the dynamic electronic structure of Cu during CO₂RR, in-situ X-ray absorption near-edge spectroscopy (XANES) was performed under applied potentials (Supplementary Fig. 8). The measurement position was periodically adjusted during catalysis to minimize potential beam-induced damage, following established protocols^24,25 (details in Supplementary information). At open-circuit voltage (OCV), the Cu K-edge XANES spectrum of Cu-PC exhibited an absorption edge energy intermediate between Cu foil and CuO references, indicating a mixed oxidation state (Fig. 1c). Upon applying −0.9 V, the Cu K-edge shifted toward the metallic Cu Ref., suggesting partial reduction of Cu. However, at a potential of −1.5 V, the Cu K-edge stabilized, with no further reduction observed, consistence with the retention of Cu^δ+ state during CO₂RR (Supplementary Fig. 9). Heat maps of the first derivative of the XANES spectra further validated the stability of Cu^δ+ under operating conditions (Fig. 1d). Quantitative analysis of Cu K-edge energy shifts revealed an average Cu oxidation state of Cu of in the Cu-PC remained at approximately +0.3, even at a potential as low as −1.5 V, underscoring resistance to over-reduction and electrochemical stability (Supplementary Fig. 10). In-situ extended X-ray absorption fine structure (EXAFS) analysis provided atomic level insights into the Cu coordination environment Cu in Cu-PC. Fourier-transformed EXAFS spectra and wavelet transformation of the EXAFS data at the Cu K-edge revealed a dominant Cu-Cu bond at ~2.24 Å, with no detectable Cu-O bond (Cu-O scattering at ~1.56 Å), indicating that metallic Cu predominates during CO₂RR (Supplementary Figs. 11 and 12)²⁶. EXAFS fitting revealed a stable Cu-Cu coordination number of 10.5 ± 1.1 at OCV, increasing marginally to 11.0 ± 0.8 at −1.5 V (Supplementary Fig. 13 and Table S2), consistent with minor structural relaxation of Cu-PC during CO₂RR (Fig. 1e)^27,28. This structural stability is crucial for maintaining active sites that favor C-C coupling, thereby enhancing selectivity toward multicarbon products like C₂H₄.

Ion and gas management facilitated by the Cu-PC catalyst

To elucidate the ion management mechanisms of the Cu-PC design during CO₂-to-C₂H₄ electrosynthesis, we investigated the retention of K⁺ ions near the catalyst surface. Inductively coupled plasma optical emission spectroscopy (ICP-OES) measurements revealed a markedly reduced accumulation of K⁺ ions on the Cu-PC catalyst compared to the Cu Ref catalyst, with a decrease of 87.7% (Supplementary Figs. 16 and 17). The suppression of K⁺ enrichment is attributed to the hydrophobic PTFE/CNT matrix, which creates a localized hydrophobic microenvironment that effectively isolates the catalytic surfaces from bulk electrolytes, thereby inhibiting cation aggregation and salt precipitation during CO₂RR.

Gas management is equally crucial for maintaining high CO₂ utilization, particularly under industrially relevant low CO₂ concentrations. To evaluate this, we performed operando gas distribution studies in a flow-type cell, quantitatively tracking CO₂ and product fluxes during electrolysis. Visualization tools, including images and Sankey diagrams, directly compare CO₂ utilization and crossover behavior of Cu-PC with Cu Ref catalysts (Fig. 2a and Supplementary Figs. 14 and 15). The Cu-PC catalyst exhibited a remarkable reduction in CO₂ loss, with only 8.6% unreacted CO₂ loss, compared to 20.0% for the Cu Ref catalyst (Fig. 2b). Concurrently, gaseous products crossover into the catholyte was drastically reduced to 4.8% for Cu-PC, versus 32.3% for the Cu Ref (Fig. 2c). These results suggest that the hydrophobic PTFE/CNT architecture of Cu-PC catalyst minimizes CO₂ diffusion into the catholyte and confines reactive intermediates near active sites, thereby ensuring efficient CO₂ utilization and stable C₂H₄ selectivity. Together, the dual functionality of Cu-PC catalyst, hydrophobicity-driven ion regulation and gas confinement, optimizes the electrochemical microenvironment for sustained CO₂RR. The synergy is crucial for achieving long-term stability and high product selectivity.

Fig. 2. Gas distribution study.

a Sankey diagram illustrating the CO₂ crossover process on the Cu-PC catalyst during CO₂RR. b Percentage of unreacted CO₂ lost to the catholyte for Cu-PC and Cu Ref catalysts. c Percentage of gaseous products loss to catholyte relative to total gas products for Cu-PC and Cu Ref catalysts. Relevant source data are provided in the Source Data file.

CO₂RR performance in different electrolyzers

The CO₂RR performances of the Cu-PC and Cu Ref catalysts were systematically evaluated across multiple electrolyzer configurations to assess their industrial feasibility and long-term operational stability. Electrochemical tests were conducted in a three-electrode flow-type cell using a 1.0 M KOH aqueous electrolyte (Supplementary Fig. 18). Gas and liquid products were quantified by gas chromatography (GC) and nuclear magnetic resonance (NMR) spectroscopy, respectively (Supplementary Fig. 19). The Cu-PC catalyst exhibited optimal CO₂RR activity when electrodeposited at a current density of 100 mA cm⁻² for 60 s (Supplementary Figs. 20–23).

As illustrated in Fig. 3a, the Cu-PC catalyst showed exceptional selectivity toward C₂₊ products, particularly ethylene (C₂H₄), ethanol (C₂H₅OH), and n-propanol (C₃H₇OH), over a wide current density range of 1.2 − 1.6 A cm⁻². At 1.2 A cm⁻², the Cu-PC catalyst achieved a C₂H₄ Faradaic efficiency (FE) of 75.9% and total C₂₊ FE of 90.8%, while suppressing H₂ evolution (H₂ FE below 5%). In contrast, the Cu Ref catalysts attained only 29.8% C₂H₄ FE under identical conditions (Fig. 3b and Supplementary Figs. 24 and 25). Furthermore, the Cu-PC catalyst delivered a partial current density for C₂H₄ $\left(j_{{\mathrm{C}}_2{\mathrm{H}}_4}\right)$ across the tested range, reaching 972.8 mA cm⁻² at 1.6 A cm⁻², more than twice that of the Cu Ref catalyst (430.4 mA cm⁻²) (Fig. 3c and Supplementary Fig. 26). These enhancements stem from the accelerated C-C coupling kinetics and a stabilized hydrophobic interface, as confirmed by structural and electrochemical analysis. To quantify C−C coupling efficiency, we calculated the ratio of C₂H₄ FE to C₁ produce FE $\left({\mathrm{Ratio}}_{{\mathrm{C}}_2{\mathrm{H}}_4/{\mathrm{C}}_1}\right)$ across varying current densities (Supplementary Fig. 27). The Cu-PC catalyst reached a peak ${\mathrm{Ratio}}_{{\mathrm{C}}_2{\mathrm{H}}_4/{\mathrm{C}}_1}$ of 7.9 at 1.2 A cm⁻², 6.6-fold higher than that of Cu Ref (1.2). The enhanced C−C coupling efficiency of Cu-PC catalyst demonstrates the critical role of the hydrophobic matrix in stabilizing *CO intermediates and promoting C-C bond formation. The half-cell energy efficiency (EE_1/2), which accounts for both the cathodic CO₂RR and the anodic water oxidation, was evaluated to assess industrial feasibility (details in Supplementary Information). The Cu-PC catalyst delivered a high C₂H₄ EE_1/2 of 38.5% at −0.98 V vs RHE, outperformed the Cu Ref, which exhibited a C₂H₄ EE_1/2 of only 11.7% at −1.01 V vs RHE (Supplementary Fig. 28). This improvement in EE reflects reduced overpotentials enabled by the coating-free hydrophobic microenvironment, which minimizes parasitic energy losses while maintaining high C₂H₄ selectivity.

Fig. 3. CO₂RR performance.

a The FEs for various products on the Cu-PC catalyst during CO₂RR at different current densities (electrolyte: 1 M KOH). b Comparison of C₂H₄ FEs on Cu-PC and Cu Ref catalysts at different current densities (electrolyte: 1 M KOH). c Partial current densities of various products as a function of the applied potential (V vs RHE, 100% iR correction, electrolyte: 1 M KOH, pH=14). d The C₂H₄ FEs under various CO₂ concentration conditions for Cu-PC and Cu Ref catalysts. e CO₂RR stability of the Cu-PC catalyst operated at 200 mA cm⁻² in the MEA electrolyzer using 0.1 M KHCO₃ electrolyte (full-cell potentials without iR correction). Gray and yellow arrows indicate the time points when the electrolyte and the anion exchange membrane were refreshed, respectively. f Performance comparison of the Cu-PC catalyst with literature benchmarks (References are listed in Supplementary Information). g The C₂H₄ FEs at different current conditions in the MEA electrolyzer stack (full-cell potentials without iR correction). Error bars represent s.d. between the three independent measurements. Data are presented as mean values  ±  s.d. Relevant source data are provided in the Source Data file.

We also investigated the CO₂RR performance of the Cu-PC catalyst under varying CO₂ concentrations (100%, 75%, and 50% CO₂ mixed with argon) in the flow-type electrolyzer. The Cu-PC catalyst maintained high C₂H₄ FEs under reduced CO₂ availability, achieving 75.9%, 49.2%, and 37.7% at 100%, 75% and 50% CO₂ concentrations, respectively (Fig. 3d and Supplementary Figs. 29 and 30). In contrast, the Cu Ref catalyst exhibited a pronounced decline in performance, with a C₂H₄ FE of only 12.1% at 50% CO₂. Furthermore, the Cu-PC achieved a high $j_{{\mathrm{C}}_2{\mathrm{H}}_4}$ of 452.4 mA cm⁻² at −1.25 V vs RHE under 50% CO₂, nearly three times higher than that of the Cu Ref catalyst (163.8 mA cm⁻² at −1.55 V vs RHE, Supplementary Figs. 31–33). These results highlight that the Cu-PC catalyst enhances CO₂ mass transport and catalytic efficiency even under limited CO₂ availability, a crucial requirement for industrial applications involving variable CO₂ feedstocks.

To validate the industrial scalability of the Cu-PC catalyst, we further evaluated its CO₂RR performance in a membrane electrode assembly (MEA) electrolyzer (Supplementary Figs. 34 and 35)²⁹. In 1.0 M KOH electrolyte, the Cu-PC catalyst delivered a peak FE for C₂H₄ of 72.9% at −2.92 V, accompanied by a high partial current density of 472.5 mA cm⁻² at −2.99 V (Supplementary Fig. 36). The full-cell energy efficiency (EE_full‑cell) reached 29.2% at $j_{{\mathrm{C}}_2{\mathrm{H}}_4}$ of 401 mA cm⁻² (Supplementary Fig. 37), emphasizing the industrial relevance of the Cu-PC catalyst. Furthermore, the Cu-PC catalyst exhibited a single-pass carbon efficiency (SPCE) for C₂H₄ of 10.2%, along with a high C₂H₄ production rate (${\mathrm{R}}_{{\mathrm{C}}_2{\mathrm{H}}_4}$) of 1246.6 μmol h⁻¹ cm⁻² at −2.92 V, demonstrating both the scalability and productivity of the Cu-PC for large-scale CO₂-to-C₂H₄ electrosynthesis. To minimize the energy penalty associated with separating unreacted CO₂ and gas products, we sought to maximize SPCE in the MEA electrolyzer. Carbonate crossover through an anion exchange membrane (AEM) was identified as a major source of carbon loss, especially under low CO₂ flow rates. To mitigate this issue, an asymmetric membrane configuration was adopted, comprising an AEM at the cathode and a proton-exchange membrane (PEM) at the anode, following our previous design (Supplementary Fig. 37). Under a low CO₂ flow rate of 2 sccm, the Cu-PC catalyst achieved a peak SPCE for CO₂RR (SPCE_CO2RR) of 95.4%.

Long-term operational stability is essential for the commercial deployment of CO₂RR catalysts. In this regard, the Cu-PC catalyst was evaluated in an MEA electrolyzer for 1020 h of continuous operation at 200 mA cm⁻², maintaining a stable full-cell voltage of approximately −3.5 V and a steady C₂H₄ FE around 68% (attenuation rate of only ~0.016% hour⁻¹) in 0.1 M KHCO₃ electrolyte (Fig. 3e), performance that is competitive with state-of-the-art CO₂RR systems. (Fig. 3f). During the stability test, the AEM was replaced periodically, while the cathode was neither cleaned nor re-treated with hydrophobic components. Notably, no salt precipitation, electrode flooding, or degradation of hydrophobic layer was observed (Supplementary Fig. 38a). Post-test analysis confirmed the electrode retained its hydrophobicity, with a contact angle of ∼130°. The absence of noticeable carbonate or bicarbonate accumulation, a common failure mode in conventional systems⁹, further validated the Cu-PC electrode’s robust management of gas transport and ion distribution. SEM and XAFS analysis of the Cu-PC after CO₂ electrolysis revealed that the morphology, particle size, and chemical state of the Cu nanoparticles remained unchanged, with no evidence of aggregation (Supplementary Fig. 39). We hypothesized that the deposition of Cu onto CNTs forms a heterojunction that inhibits the Cu migration and reconstruction, thus maintaining stable performance.

To assess the scalability of the electrode design, a 5-cell MEA stack with total active area of 5 × 10 cm² was assembled (Supplementary Fig. 38b). The stack achieved a peak C₂H₄ FE of 71.1% at 500 mA cm⁻², corresponding to a high C₂H₄ production rate exceeding 1.2 L h^–1 (Fig. 3g). These results confirm the feasibility of the Cu-PC catalyst for high-throughput CO₂-to-C₂H₄ conversion, but the current MEA stack number remains far from full-scale industrialization. Recognizing that MEA conditions vary within the stack, we further investigate the salt deposition along channels at different stack positions (Supplementary Fig. 38c). The results indicate that, owing to the anti-salt precipitation capability of the Cu-PC electrode, no significant accumulation occurred across the stack, whereas the Cu Ref electrode showed noticeable precipitation near the gas inlet, likely due to higher local CO₂ concentration. To further demonstrate the generality of our electrode framework, we extended the design to alternative catalyst systems. Specifically, Ag-based system (Ag-PC) and Sn-based system (Sn-PC) variants were fabricated, achieving nearly 100% FE for CO production and over 97% FE for HCOOH production, respectively (Supplementary Fig. 38d and e). These findings highlight the broad application of our hydrophobic electrode architecture across diverse CO₂ reduction pathways and product targets.

Mechanistic insights into hydrophobic microenvironment for CO₂RR-to-C₂H₄

To uncover the role of the hydrophobic microenvironment in CO₂-to-C₂H₄, we conducted in-situ Raman spectroscopy under controlled potentials, as detailed in the Supplementary information (Supplementary Fig. 40). Cu-PC displayed distinct bands at around 282 and 370 cm⁻¹, corresponding to the frustrated rotation of *CO on Cu and Cu−CO stretching modes, respectively (Fig. 4a). A notable blue shift in the Cu−CO stretching band for the Cu-PC catalyst compared the Cu Ref indicates a stronger Cu−CO binding with enhanced *CO stabilization and facilitated C−C coupling, in agreement with prior studies^30–32.

Fig. 4. In situ characterization and DFT calculation.

a In-situ Raman spectra of the Cu-PC and Cu Ref catalysts in the range 200­−450 cm^-1 at various applied potentials (V vs RHE, 100% iR correction, electrolyte: 0.5 M KHCO₃, pH = 8.3). b Calculated surface pH values on different electrodes (V vs RHE, 100% iR correction, electrolyte: 0.5 M KHCO₃, pH = 8.3). c Ratio of the low-frequency and high-frequency bands in the range 2000−2200 cm⁻¹ for Cu-PC and Cu Ref catalysts at various applied potentials (V vs RHE, 100% iR correction, electrolyte: 0.5 M KHCO₃, pH = 8.3). d The free energy of CO adsorption (ΔG_*CO) as a function of the surface OH (θ_OH) and H₂O (θ_H2O) coverage. e *HCCOH dihydroxylation to form a hydrocarbon intermediate (*HCC) or hydrogenation to form a hydroxy-containing intermediate (*HCCHOH). Red, gray, white, and orange spheres represent O, C, H, and Cu atoms, respectively. f The corresponding atomic configurations of key intermediates. Water molecules and associated hydrogen bonding are explicitly shown. Red, gray, white, orange, and purple spheres represent O, C, H, Cu, and K atoms, respectively. g Calculated energy barriers for the *HCCOH → *HCC and *HCCOH → *HCCHOH pathway under an applied potential of −1 V (vs standard hydrogen electrode (SHE)). IS, TS, and FS denote the initial, transient, and final states, respectively. Relevant source data are provided in the Source Data file.

Notably, bands at approximately 1065 and 1013 cm⁻¹ are attributable to adsorbed carbonate (CO₃^2-) and bicarbonate (HCO₃⁻), respectively³³, enabling estimation of the local surface pH via a calibrated intensity ratio method (Supplementary Fig. 41)³⁴. As the applied potential shifted from −0.44 to −0.84 V, we observed an intensification of the CO₃²⁻ peak and a corresponding weakening of the HCO₃⁻ band, suggesting an increase in local surface pH for both the Cu-PC and Cu Ref catalysts (Supplementary Fig. 42). Quantitative analysis (Fig. 4b) showed that, at −0.44 V, the surface pH values for Cu-PC and Cu Ref were 11.1 and 10.2, respectively, both lower than the bulk electrolyte pH due to the interfacial CO₂ consumption during CO₂RR. At −0.84 V, the surface pH increased to 11.6 for the Cu-PC and 10.7 for the Cu Ref. The significant pH value enhancement for Cu-PC arises from the hydrophobic PTFE/CNT matrix in Cu-PC, which restricts electrolyte flooding to stabilize the gas-liquid-sold interface and limit K⁺ migration to the electrode surface. Reduced K⁺ accumulations minimize charge screening effects, thereby preserving OH⁻ concentration at the catalytic interface (see Fig. 1a). In contrast, the Cu Ref catalyst lacks hydrophobicity, enabling unrestricted K⁺ migration under the negative potential, which depletes interfacial OH⁻ and suppresses local pH. The elevated pH on Cu-PC favors *CO dimerization by reducing the energetic barrier for C-C coupling, consistent with the theoretical and experimental studies linking alkaline microenvironment to improved C₂₊ selectivity.

In-situ Raman analysis further revealed a prominent C≡O stretching band above 2000 cm⁻¹ for the Cu-PC catalyst, indicative of *CO intermediates primarily adsorbed at atop sites. Atop-bound *CO is recognized as a key descriptor for C−C coupling, due to its favorable geometry and electronic configuration for dimerization^31,35. Notably, a red shift in the CO_atop peak with increasing negative potentials (Supplementary Figs. 43 and 44) suggests enhanced *CO interaction with adsorbed OH⁻¹⁵, consistent with OH⁻ enrichment at the hydrophobic Cu-PC surface. This OH⁻ confinement, corroborated by surface pH measurements, lowers the kinetic barrier for *CO dimerization by stabilizing polarizable intermediates within the alkaline microenvironment. The Stark tuning slopes provided additional mechanistic insight into the interfacial electric field strength. The Cu-PC displayed a higher Stark slope (87 cm⁻¹ V⁻¹) than Cu Ref (45 cm⁻¹ V⁻¹), indicating a stronger interfacial electric field within the hydrophobic matrix (Supplementary Fig. 45)¹⁵. This enhanced local electric field facilitates CO₂ activation by stabilizing *CO intermediates and accelerating electron transfer during the reduction process. Meanwhile, the hydrophobic interface likely amplifies the local electric field intensity by lowering the local dielectric constant.

To quantify catalytic performance, we analyzed the relative intensities of the low-and high-frequency Raman bands to derive the Ratio_low/high metric. The Cu-PC catalyst exhibited a higher Ratio_low/high value than the Cu Ref catalyst (Fig. 4c), indicating enhanced *CO surface coverage and C−C coupling efficiency^32,35. The improvement originates from the hydrophobic matrix, which (1) enriches local OH⁻ concentration, stabilizing *CO intermediates through electrostatic interactions, (2) limits water ingress, thereby suppressing the competitive HER. Collectively, the hydrophobic structure establishes a synergistic microenvironment that stabilizes reaction intermediates, minimizes K⁺ accumulation, and enables over 1000 h of stable CO₂-to-C₂H₄ conversion at high current densities.

To further elucidate the mechanistic role of the hydrophobic microenvironment in CO₂RR, we performed density functional theory (DFT) calculations to investigate the free energy of *CO adsorption (ΔG_*CO) as influenced by surface OH⁻ coverage (θ_OH) and H₂O coverage (θ_H2O). These studies illuminate the interplay between hydrophobicity and local alkalinity in stabilizing *CO intermediates, critical for enhancing C-C coupling and formation of C₂₊ products (Supplementary Figs. 46–48 and Supplementary Data 1). Using a Cu(111) slab model, we systematically varied θ_OH and θ_H2O to evaluate their effects on ΔG_*CO. As shown in Fig. 4d, decreasing θ_H2O (mimicking a hydrophobic environment) leads to more negative ΔG_*CO, revealing a stronger *CO stabilization. Meanwhile, increasing θ_OH (reflecting local alkalinity) further strengthens *CO adsorption, confirming that the local OH⁻ enrichment synergizes with hydrophobicity to tune intermediate binding. These findings are consistent with our experimental data, supporting the hypothesis that the hydrophobic matrix in the Cu-PC catalyst suppresses competitive H₂O adsorption while fostering localized OH⁻ accumulation, thereby stabilizing *CO for efficient CO₂RR.

Experimental results demonstrate that the Cu-PC catalyst exhibits significantly higher C₂H₄ selectivity compared to Cu Ref catalyst. Previous mechanistic studies suggest that both C₂H₄ and C₂H₅OH pathways originate from *CO dimerization, with bifurcation occurring at the *HCCOH intermediate (Fig. 4e)³⁰. Beyond this point, *HCCOH can either undergo dehydroxylation to form the hydrocarbon intermediate *HCC (a precursor to C₂H₄) or hydrogenation to yield the hydroxy-containing intermediate *HCCHOH (leading to C₂H₅OH) (Fig. 4f). Our analysis reveals a pronounced preference for the dehydroxylation pathway on the Cu-PC catalyst. The reaction *HCCOH + e⁻ → *HCC  +  OH⁻, involving the C–OH bond cleavage, is thermodynamically favored under hydrophobic conditions owing to suppressed H₂O activity and enhanced local OH⁻ concentration. In contrast, the hydrogenation pathway, *HCCOH  +  H₂O  +  e⁻ → *HCCHOH +  OH⁻, which requires water-mediated proton transfer, is less favorable on the Cu-PC surface due to its hydrophobic, water-deficient microenvironment. To quantify these trends, we computed the reaction energies for *HCCOH conversion to *HCC (C₂H₄ pathway) and *HCCHOH (C₂H₅OH pathway) under conditions mimicking the strong local alkalinity and hydrophobicity of Cu-PC catalyst (Supplementary Figs. 49–51). As shown in Fig. 4g and Supplementary Fig. 52, within a water-deficient alkaline environment, the energy barrier for *HCCOH → *HCC (0.479 eV) is markedly lower than that for *HCCOH → *HCCHOH (1.044 eV), indicating both thermodynamic and kinetic preference for C₂H₄ production. These results demonstrate that reduced water availability at the hydrophobic interface steers selectivity toward pathways less dependent on solvent-mediated steps (e.g., *HCC formation), while the enriched local OH⁻ concentration stabilizes dehydroxylation intermediates. The hydrophobic matrix of Cu-PC serves a dual role: stabilizing *CO intermediates to promote C-C coupling and directing the reaction flux toward C₂H₄ by favoring dehydroxylation over hydrogenation. This synergy between intermediate stabilization and pathway selectivity mirrors previous findings that hydrophobic interfacial layers enhance C₂H₄ productivity by modulating local water and ion dynamics. By decoupling *CO stabilization from water-dependent hydrogenation steps, the Cu-PC achieves an optimal balance between intermediate reactivity and product selectivity, rationalizing its superior C₂H₄ Faradaic efficiency.

Industrial application feasibility and techno-economic analysis of economic viability

The direct utilization of industrial CO₂ emission, particularly from dilute point sources, such as flue gas (typical 10 ~ 15% CO₂), presents significant challenges due to the low concentrations of CO₂ and the presence of impurities, like sulfur dioxide (SO₂) and nitric oxide (NO) at levels of approximately 100 ppm, which can poison catalysts and degrade electrolyzer components^1,36,37. Despite these hurdles, integrating CO₂ electrolysis with industrial waste streams remains a promising route for sustainable chemical synthesis, if catalysts and systems are optimized to function under real-world conditions. In this study, we evaluated the Cu-PC catalyst under a simulated flue gas condition (15% CO₂ concentration) in a membrane electrode assembly (MEA) cell. The Cu-PC catalyst exhibited robust performance, maintaining a high C₂H₄ FE of 30.2% at 500 mA cm⁻², demonstrating its effectiveness under high current densities and dilute CO₂ feeds.

However, the direct utilization of flue gas in CO₂ electrolysis is often constrained by impurities such as SO₂ and NO, which can poison the catalyst and diminish CO₂RR efficiency^36,38. To evaluate the impurity tolerance of the Cu-PC, we implemented a cyclic exposure protocol, subjecting the catalyst to intermittent contact with SO₂ and NO under representative coal-fired flue gas conditions. During two-hour cycles alternating between impurity-containing and clean gas feeds, the Cu-PC exhibited minimal fluctuations in C₂H₄ selectivity (30% ± 1%) and maintained a stable cell voltage, with full recovery of performance upon impurity removal (Fig. 5a). Impurity gases are known to undergo hydrogenation on the catalyst surface, leading to electrode deactivation³⁹. The hydrophobic electrode design of Cu-PC suppressed hydrogen evolution, thereby limiting hydrogenation side reactions and enhancing resistance to impurity poisoning. Moreover, considering the presence of oxygen (O₂) in practical flue gas streams, we evaluated catalyst performance under a mixed feed of 15% CO₂ + 5% O₂. The C₂H₄ selectivity showed no significant deviation compared to the oxygen-free conditions (Supplementary Fig. 53). Together, these results demonstrate that Cu-PC possesses robust tolerance to SO₂ and NO exposure, along with rapid recovery after impurity removal. Its stability in the presence of O₂ further underscores its practical potential for industrial CO₂RR applications where flue gas impurities are unavoidable.

Fig. 5. Application feasibility and techno-economic analysis.

a Stability test of Cu-PC catalyst under 15% CO₂ concentration with intermittent impurity gas disturbance (100 ppm SO₂ and 100 ppm NO). Experiments were performed at a constant current density of 500 mA cm⁻² (full-cell potentials without iR correction, electrolyte: 1 M KOH). Green arrows represent the introduction of impurity gas. b Techno-economic analysis of CO₂-to-C₂H₄ route on Cu-PC catalyst under 15% CO₂ concentration. c Schematic diagram of the application of CO₂RR to the exhaust gas treatment of power plant. d Techno-economic analysis of C₂H₄ product. Data are presented as mean values  ±  s.d. Relevant source data are provided in the Source Data file.

A preliminary techno-economic analysis was conducted to assess the feasibility of scaling up the Cu-PC for CO₂-to-C₂H₄ conversion, based on experimental data from a single MEA cell. Assuming a catalyst lifetime of one year and an electrode surface area of 100 m², consistent with previous studies⁴, the estimated total production cost was US$ 11,925 per ton of C₂H₄ (Fig. 5b). Capital expenditures, including installation, balance-of-plant components and operational costs were identified as the main cost drivers, with the anode catalyst and anion exchange membranes contributing 38.5% and 23.3% of total costs, respectively. Notably, the low-concentration CO₂ feedstock accounted for only 1.7% of total costs, suggesting a potential economic advantage for direct utilization of fuel gas. However, the analysis does not consider CO₂ losses (e.g., carbonate formation, membrane crossover) or downstream separation costs, which must be evaluated in future large-scale assessments. Despite the low sensitivity to CO₂ feedstock price, electricity consumption remains a major challenge, driven by the system’s limited current density and relatively high operating voltage³⁶. The reduction in C₂H₄ FE under dilute CO₂ conditions impacts the techno-economic performance in two key ways: (1) increasing the energy required for product separation, as lower selectivity generates more unconverted reactants and byproducts, and (2) reducing SPCE, necessitating larger reactor volumes or higher current densities to maintain absolute C₂H₄ output, thereby elevating both capital and operating costs. To enhance economic viability, further optimization should focus on (1) improving catalyst selectivity and durability under industrial-scale current densities, (2) reducing operating voltage through tailored electrode architectures and ionomer engineering, and (3) minimizing CO₂ crossover losses via advanced membrane design.

To explore potential industrial applications, we modeled the integration of CO₂RR system with coal-fired power plants, utilizing flue gas (10~15% CO₂, typical of coal combustion exhaust) as feedstock (Fig. 5c). This integration offers two main benefits: (1) converting waste into C₂H₄, a high-value chemical feedstock, and (2) partially substituting fossil-derived C₂H₄ in industrial supply chains. A one-year techno-economic assessment (electrode area: 100 m²) was conducted to estimate the projected potential advantages Based on an annual C₂H₄ production of 114.61 tons, the projected carbon emission tax saving is ~US$ 52,365 per year (equivalent to ~US$ 456.9 per ton of C₂H₄), while revenue from C₂H₄ sales as bulk chemical could generate approximately ~US$ 99,252 per year (Fig. 5d). We noted, however, that these economic benefits still are modest relative to the production cost of C₂H₄ via CO₂RR, underscoring the need for continued optimization. Overall, the integration of our hydrophobic matrix-enhanced CO₂RR system demonstrates a promising approach for CO₂-to-C₂H₄ conversion, offering a pathway to significantly reduce carbon emissions. By combining efficient C₂H₄ production with impurity resilience and energy optimization, the Cu-PC catalyst represents a compelling and economically viable solution. Furthermore, the hydrophobic matrix could be extended to improve the selectivity, efficiency, and stability of other catalysts, highlighting a versatile approach for addressing practical challenges in CO₂ utilization.

Discussion

In this work, we present a wettability-guided Cu-based electrode design that structurally decouples hydrophobicity from electronic conductivity to mitigate salt precipitation during CO₂ electroreduction. The electrode features in situ embedded Cu nanoparticles within an insulating hydrophobic polymer matrix that acts as a “hydrophobic trap”, overlaid with a sparse conductive network enabling efficient charge transport while preserving interfacial hydrophobicity. This decoupled architecture prevents hydrophobic degradation, maintains a stable gas-liquid-solid interface, suppresses electric field-induced cation accumulation, and confines locally generated OH⁻ ions, thereby sustaining an alkaline microenvironment favorable for C-C coupling and C₂H₄ formation. In a flow cell, the optimized electrode achieves a C₂H₄ FE of 75.9% and a total C₂₊ FE of 90.8% at 1.2 A cm⁻². Furthermore, it demonstrates exceptional durability, operating stably for over 1,000 h at 200 mA cm⁻² in an MEA without detectable degradation or salt precipitation. A SPCE exceeding 95% is realized, and a 5-cell MEA stack operating at 500 mA cm⁻² produces C₂H₄ at 1.2 L h⁻¹ with an FE above 70%. The electrode also maintains high performance under low CO₂ concentration and in the presence of typical flue gas impurities (NO, SO₂, and O₂), reducing gas pretreatment requirements and overall operational costs. This wettability-engineering design is further extended to Ag- and Sn-based catalysts for selective CO and HCOOH production, respectively, demonstrating its versatility and scalability. Overall, this work established a scalable and durable platform for long-term, carbon-efficient CO₂ electroreduction.

Methods

General materials and synthesis

All chemicals purchased from commercial sources: Potassium hydroxide (KOH, 95%) and potassium bicarbonate (KHCO₃, 99.5%) from Beijing Chemical Factory, Cupric bromide (CuBr₂, 99.99%) and sodium tartrate dibasic dihydrate (C₄H₄Na₂O₆·2H₂O, 99%) from Sigma-Aldrich, and Polytetrafluoroethylene nanoparticles from Dupont were used as received. DI water was obtained from a Millipore Gradient Milli-Q water purification system. Carbon nanotubes (CNTs) from XFNANO Materials Tech Co., Ltd. were used after purification. As-received CNTs were first calcined at 500 °C in air for 5 h. After cooling to room temperature (298 K), the CNTs were transferred into a HCl aqueous solution (5 wt%) and sonicated for 60 min (power: 450 W). The purified CNTs were collected by filtration and washed extensively with DI water. Electrolyte solutions were prepared from stock solutions of higher concentration in DI water, which were then diluted to the target molarity.

The Cu-PC catalyst was prepared via an electrodeposition approach: a hydrophobic matrix of PTFE and CNTs was first sprayed onto a gas diffusion layer (Toray) with a loading of 1 mg cm⁻². An excess of PTFE/CNT matrix was prepared to account for material loss during the spray deposition process. To ensure a consistent mass loading, the electrodes were weighed both before and after spraying. Electrodeposition was proceeded with continuous CO₂ gas flow (20 standard cubic centimeters per minute, sccm) at a constant current density of −0.1 A cm⁻² for a predetermined time. The solution consisted of 0.1 M cupric bromide, 0.2 M sodium tartrate dibasic dihydrate, and 1 M KHCO₃. Constant agitation was required during electrodeposition (∼60 rpm). For comparison, the Cu Ref electrode was prepared by dispersing Cu nanoparticles (25 nm particle size (TEM); 99.5% from Sigma Aldrich) and Nafion in a mixture of isopropyl alcohol and water (with a volumetric ratio of 3:1). This ink underwent ultrasonication for 30 min and was then spray-coated on the gas diffusion layer at a loading of 1 mg cm⁻². To ensure an accurate mass loading, the electrodes were weighed (XSR105, METTLER TOLEDO) both before and after preparation.

Electrochemical measurements

The electrochemical measurements were conducted at room temperature (298 K) throughout this study. In a flow-type cell, the anode catalyst was titanium mesh-supported iridium oxide (IrO_x/Ti mesh) prepared by a dip coating and thermal decomposition method: the anode was manufactured through the following series of steps: commercial Ti mesh was etched in HCl solution (6 M) for 60 min at 80 °C. The etched Ti mesh was rinsed with DI water to eliminate any the impurities. The Ti mesh was dipped into a solution containing isopropanol, iridium (IV) oxide dehydrate, and hydrochloric acid (HCl), was dried at 100 °C for 15 min, and then underwent sintering at 500 °C for 10 min. Steps were repeated until achieving the desired loading of 2 mg cm⁻². An anion-exchange membrane (Fumasep FAA-3-PK-130, FuMA-Tech) was used to separate the cathode chamber and anode chamber. The anion-exchange membrane was soaked in 1 M KOH for 24 h before CO₂RR test. CO₂ gas was continuously supplied to the cathode chamber at gas rate of 20 sccm. 1 M KOH was used as the cathode and anode electrolytes with a flow rate of 20 mL min⁻¹. Using an electrochemical station (PARSTAT4000A), we evaluated the performance of the cathode electrode in a three-electrode system at different current densities. Hg/HgO from Gaoss Union was used as the reference electrode. The potentials of flow-type cell were recorded versus RHE scale using the following conversion:

$$E\left(\mathrm{vs}\mathrm{RHE}\right)=E\left(\mathrm{vs}\mathrm{Hg}/\mathrm{HgO}\right)+0.098\mathrm{V}+0.0591\times \mathrm{pH}+100\%\mathrm{iR}$$

The ohmic loss between the working and reference electrodes was measured using the electrochemical impedance spectroscopy technique before the test.

In a MEA electrolyzer, the cathode, an IrO_x on Ti felt (2 mg cm⁻², IrO_x-ALK from Gaoss Union), and an anion-exchange membrane (X37-50, Dioxide Materials) were compressed to form the MEA. The anion-exchange membrane was positioned between the anode and cathode chambers. CO₂ gas was continuously supplied to the cathode chamber through a humidifier with DI water at a gas flow rate of 10 sccm cm⁻². The anode electrolyte (20 mL min⁻¹) was introduced into the anode chamber via precision peristaltic pump. The performance of the cathode electrode in a two-electrode system was evaluated using an electrochemical station equipped with a customized current booster (40 A) under different current conditions. A cold trap was utilized to separate liquid products from gas products on the cathode side. The FE of liquid products was calculated based on the total amount of the products collected from both anode and cathode chambers, considering the liquid products that might have crossed over the membrane. The full-cell voltages of MEA were recorded without iR correction.

FE calculation

The FE was calculated as follows:

$$\mathrm{FE}=\frac{nF{v}_{i}G{p}_0}{RT{j}_{total}}$$

where n represents the number of electrons transferred during the CO₂ conversion process to yield the target product, v_i is the volume concentration of product in the exhaust gas from the electrochemical cell (GC data), p₀ = 1.01 bar, G is the gas flow rate (mL min⁻¹), j_total is the steady-state current, F = 96485 C mol⁻¹, R = 8.314 J mol⁻¹ K⁻¹, T is the temperature at Kelvin.

To enhance the reliability of our results, we installed high-precision mass flow controllers (MFCs) at both the inlet and outlet of the electrochemical cell. The gas flow rate has been converted by the mixed gas coefficient. Additionally, we incorporated temperature sensors within the MFC to accurately measure the true flow rate and temperature of the gas.

Full-cell energy efficiency calculation

The full-cell energy efficiency for A product is calculated as follows:

$${EE}_{full-cell}=\frac{\left(1.23+\left({-E}_A^o\right)\right)*{FE}_A}{{-E}_{full-cell}}$$

where E_A^o is the thermodynamic potential of CO₂ to product A, calculated based on the standard molar Gibbs energy of formation at 298.15 K, FE_A is the measured FE (%) of the product A, and E_full-cell is the full-cell voltage measured in the MEA system without ohmic loss correction.

ECSA evaluation

The electrochemical surface area (ECSA) of the catalysts was determined using the electrochemical double-layer capacitance (C_dl) method. Prior to measurement, all catalysts were reduced at −0.7 V versus RHE for 60 s. Cyclic voltammetry (CV) scans were performed in the non-Faradaic region in Ar-saturated 1 M KOH for ten cycles at varying scan rates. The differences between anodic and cathodic current densities (Δj) were recorded during the final scan cycle. Δj values at each scan rate were plotted against the corresponding scan rate, and the slopes of the resulting linear fit were used to calculate the double-layer capacitances for each catalyst. ECSA = R_f × S, where S stands for the geometric area of the electrode. R_f was estimated from the ratio of double-layer capacitance for the working electrode and the corresponding smooth polycrystalline Cu electrode. To account for the contribution of the PTFE/CNT matrix to the overall interfacial capacitance, the ECSA of the substrate alone was measured and subtracted from the total capacitance of the composite electrodes. This approach isolates the capacitance contribution from the Cu active sites, providing a more accurate estimation of their ECSA.

The formation rate (R) calculation

The R for each species (i) was calculated as follows:

$$R_i=\left(Q_{total}\times {\mathrm{FE}}_i\right)/\left(96485\times z_i\times t\times S\right)$$

where Q_total is the total charge passed during electrolysis, FE_i is the Faradaic Efficiency for species i, indicating the fraction of the total charge used to produce the target product, z_i represents the number of electrons required to produce one molecule of product, t represents the electrolysis time, S represents the geometric area of the electrode.

Structural and compositional analysis

¹H NMR spectra were attained by using a Bruker Ultrashield Plus 500 MHz NMR spectrometer at room temperature (298 K). X-ray diffraction (XRD) patterns were obtained on Rigaku Smartlab 9 kW-Advance using Cu-Kα radiation (λ = 0.154 nm) at 40 kV and 30 mA at RT. Contact angle measurements were performed using the sessile drop method on a video-based contact angle measuring system. A single water droplet was placed on the sample, and ~15 seconds was given before the contact angles were measured by the computer software. We are grateful to the Analysis & Testing Center of Beihang University for the facilities and the scientific and technical assistance.

Computational details

All density functional theory (DFT) calculations were employed using the Vienna Ab initio Simulation Package (VASP) with the generalized gradient approximation (GGA) and the revised Perdew–Burke–Ernzerhof (rPBE) function. Projected augmented wave (PAW) and a plane wave basis were set with a kinetic energy cutoff of 450 eV. Geometry optimizations were performed with a force convergence criterion of less than 0.05 eV Å⁻¹. To evaluate the energy of CO adsorption, the implicit solvation model implemented in VASP_sol was used to explicitly calculate the E_sol. The DFT-D3 empirical correction method was employed to account for van der Waals interactions. To fully consider the solvation effect in the C–C coupling process, explicit water molecules were optimized to form a local minimum through a hydrogen bond network. The transition state (TS) structures and reaction pathways were determined using the climbing image nudged elastic band (CI-NEB) method. The minimum energy pathway was optimized with the force-based conjugate-gradient method until the maximum force was less than 0.03 eV Å⁻¹. The harmonic vibrational frequency calculations were performed to characterize all the stationary points and obtain zero-point energy (ZPE) corrections. Cu(111) surface models were constructed with four layers, where the bottom two layers were held fixed, while the other atoms were allowed to relax. A 15 Å vacuum layer along the c direction was employed to avoid periodic interactions. Monkhorst-Pack k-points of (2 × 2 × 1) were applied for the calculations. The constant electrode potential (CEP) model was utilized to consider the effects of applied potential on the reaction mechanisms.

In situ Raman spectroscopy

In situ Raman spectroscopy was conducted on a HORIBA LabRam HR Evolution Raman spectrometer with a 633 nm solid laser as an excitation source. The measurements were carried out in a homemade flow cell with a quartz window to detect signals from the cathode GDE. Each scan accumulated a Raman spectrum from 3 acquisitions. Electrolyte (0.5 M KHCO₃) was continuously pumped over the GDE at a rate of 10 mL min⁻¹ using a syringe pump. CO₂ gas, introduced at ~20 sccm, regulated by a mass flow controller, was directed towards the back of the GDE. Working electrodes were fabricated by airbrushing catalysts onto GDE, while Pt foil was used as the counter electrode.

In situ XAS

The in situ X-ray absorption fine structure measurements were conducted on the sample at the 21 A X-ray nano diffraction beamline of Taiwan Photon Source (TPS), National Synchrotron Radiation Research Center (NSRRC). The technical support was provided by Ceshigo Research Service. This beamline features a 4-bounce channel-cut Si (111) monochromator for mono-beam X-ray nano diffraction and X-ray absorption spectroscopy. The end-station is equipped with three ionization chambers and a Lytle/SDD detector positioned after the focusing position of the KB mirror for transmission and fluorescence mode X-ray absorption spectroscopy. The photon flux on the sample ranges from 1 × 10¹¹ ~ 3 × 10⁹ photon/sec for X-ray energy spanning 6-27 keV.

CO₂RR stability test

For the stability test, the electrolyte volume was increased to 4000 mL and regularly refreshed to maintain consistent performance and prevent fluctuations in electrolyte concentration. The anion exchange membrane was replaced approximately several days. Importantly, we emphasize that the cathode electrode was neither cleaned nor were hydrophobic components added during the stability test. It is crucial to evenly distribute torque during electrolyzer retightening to prevent irreversible damage or fracture of the cathode electrode. To facilitate lower resistance and minimize mass transfer impedance, the thickness of the gasket employed is marginally less than that of the electrode, ensuring that the electrode exerts adequate compression.

Author contributions

Z.Y., F.M., and D.L. conceived and designed this project. F.M., H.Z., and W.M. synthesized the catalysts and conducted the electrochemical experiments. F.M. and H.Z. carried out the in-situ/ex-situ characterizations. F.M., W.Z., F.X,. and M.J. conducted the computational calculation and relative data analysis. Everyone contributed to the data analysis, writing, and editing of the manuscript. J.L. supervised the entire project.

Peer review

Peer review information

Nature Communications thanks Rui LIN, Masakazu Sugiyama, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

Source data are provided with this paper. Data generated and analyzed in this study are included in the manuscript, Supplementary Information. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-67719-z.