Investigating the impact of hydrophobicity on CO2R performanceTo investigate the impact of hydrophobicity on CO2R performances, we selected four hydrophobic polymers to construct the Model II-type catalyst layer. The chemical structures of these four polymers are illustrated in Fig. 2a. PT95 and PCR are both fully fluorinated polymers with high hydrophobicity. Polyvinylidene fluoride (PVDF) was also selected due to its similar structure with PTFE, however, it possesses the advantage of being soluble in organic solvents such as dimethylformamide and N-methyl-2-pyrrolidone40, which is critical for preparing thin film coatings on Cu particles. Additionally, we employed Nafion as the reference sample since it is widely used for preparing the catalyst layer for CO2R41,42. We employed commercial Cu nanoparticles (Cu, ~25 nm) as the model catalyst. Although this Cu is less active/selective than specifically designed Cu catalysts28,43 it has been extensively used as a benchmark catalyst for CO2R, making our conclusions more instructive. Conventional GDL preparation procedures were employed to prepare the polymer/Cu catalyst. Specifically, PCR, PT95, PVDF and Nafion were mixed with Cu in solvents capable of dissolving the polymers, followed by thorough sonication to afford homogenous catalyst ink solutions. This solution was then spray-coated to form the Cu catalyst layer on GDL. While these polymers were used as catalyst binders, our design objective also aimed for thin and uniform coatings on the catalyst surface.First, we assessed the relative hydrophobicity of the four polymer/Cu GDEs using contact angle measurements. As depicted in Supplementary Fig. 3, all four GDEs displayed excellent hydrophobicity, with contact angles greater than 140°. Note that we were not able to capture the conventional contact-angle images for the as-prepared PT95/Cu and PCR/Cu GDEs owing to their super-hydrophobicity (Supplementary Movies 1 and 2). We also examined the water uptake properties of the four polymers to further differentiate their hydrophobic levels. The results indicated that PT95 is likely more hydrophobic than PCR (Supplementary Fig. 4). Nevertheless, we believe the relative initial hydrophobicity of the four polymers is in the sequence of PT95 > PCR > PVDF>Nafion (Fig. 2b).Then, these polymer/Cu GDEs were evaluated for CO2R using a flow-cell reactor (Supplementary Fig. 5). Using the obtained data, we plotted the highest C2+ products selectivity and partial current density achieved by the four polymer/Cu GDEs against their hydrophobicity. As shown in Fig. 2b, we found that both the C2+ selectivity and current density initially scale with the increased hydrophobicity, however, they decline after reaching a certain region. This trend suggests that, while the hydrophobicity of the polymer coating plays a role in enhancing CO2R as suggested elsewhere37,38,44, it is not the sole contributor. Consequently, other variables such as polymer porosity and layer thickness, water uptake ability, CO2 diffusivity, and other related chemical/physical properties must be considered. Through studying and adjusting these parameters, we believe the microenvironment can be further optimized towards efficient CO2 to C2+ conversion.Modeling the mass balance in Model IIWe developed a direct pore-level multi-physics model (Fig. 3a, Supplementary Figs. 6 and 7, details provided in Supplementary Information) to simulate the distribution of species on the catalyst surface during CO2R, and to evaluate the CO2R performance while changing the relevant properties of the polymer coating, i.e., thickness, porosity, and ability in managing the water/gas balance at the catalyst surface. In this model, the simulation region was limited to a single pore within the catalyst layer, and a thin/porous polymer layer was employed between the catalyst surface and the electrolyte domain. The thickness of this polymer layer is defined as \({{Thk}}_{{PL}}\),. The pore width of the catalyst layer is assumed as 200 nm35. The span of the simulation space equates to the thickness of the catalyst layer (∼30 μm). To conserve computational resources, the liquid electrolyte domain is conceived to be half-sized at 100 nm, complemented with symmetrical boundary conditions. Besides, both the interface mass transfer of CO2 and the mass transfer of the bulk electrolyte are both characterized by mass flux boundary conditions, with a constant rate for the interface mass transfer of CO2 and a consistent composition of the bulk electrolyte throughout the simulation, as illustrated in Fig. 3a45,46. Transport and electrochemical kinetics are resolved in this model to investigate how the CO2R is affected by different properties of the polymer coating. As illustrated in Fig. 1, we hypothesize that the transition from Model I to II would immediately alter the balance of H2O and CO2 at the reaction layer. Thus, we chose the volume ratio of H2O to CO2 (H2O/CO2) near the catalyst surface as one key variable of our modeling. Additionally, both porosity and thickness of the polymer layer were taken into consideration in our model to assess how they affect the local microenvironment (local species concentration) and further the CO2R performance47,48. We initially assessed the impact of polymer thickness and found that it had relatively small effects on the local microenvironments (i.e., CO2 concentration, pH) and the CO2R performance, as shown in Supplementary Fig. 8. This limited effect observed in our simulations, where the polymer thickness ranges from 0 to 30 nm, can be attributed to the fact that this thickness range does not significantly influence the effective diffusion coefficients of species within the domain. Consequently, CO2 reactants and other species within the electrolyte diffuse quickly through the polymer layer, at least at rates faster than the reaction kinetics. Besides, given the significantly higher concentration of gaseous CO2 compared to dissolved CO2 in the electrolyte, and its diffusion coefficient being approximately 10,000 times greater49, the impact of the polymer layer thickness on CO2 mass transfer becomes relatively small. Separately, we did not consider a much thicker polymer layer in our model because increasing the porous layer thickness would compromise electrode conductivity. Notably, we did not consider the impact of conductivity changes (which are difficult to reflect in the simulation) when considering variations in thickness in the simulation. By contrast, the thin layer porosity influences the microenvironments and CO2R performance obviously, as it directly affects the effective diffusion coefficient of all species within the layer. Based on our results, we suggest that the desired porosity range for polymer materials should be ranging from 0.2 to 0.9 (Fig. 3b-e)50,51. If the polymer porosity is too low, the diffusion of CO2 and other species will be significantly limited, thereby impeding the CO2R.Fig. 3: Multi-physics Modeling of CO2R in Model II.a Scheme of CO2 and H2O transport in the polymer modified catalyst (left) and the graphical illustration of the modeling domains (right). The gas and liquid transport are decoupled, leading to a three-phase unsaturated layer for enhanced electron participation in desired electrochemical reaction and reduced the CO2 transport length to the catalyst surface. The source of protons for CO2R is the water taken up by the polymer coating. b Modeled CO2 availability for the desired CO2R. c Cathodic potential. d C2+ products current density and e C2+ selectivity with the variation of the polymer porosity and local H2O/CO2 ratio at the same applied cathodic potential (−1.426 V vs. SHE). The inserts in panels b and d are the enlarged areas indicated by the red dashed line. CO2 flux boundary conditions are set at the upper boundaries of both the liquid electrolyte domain (\({{Flux}}_{{{CO}}_{2}({EL})}\)) and the polymer domain (\({{Flux}}_{{{CO}}_{2}({PL})}\)). A symmetrical condition is imposed at the right boundary to model a confined pore geometry in the catalyst layer (CL). The electrolyte potential \({\psi }_{l}\) at this boundary is set to 0 V to provide a reference for solving the electric field. Due to the continuous flow of fresh 2 M KOH electrolyte into the cathode chamber, an open boundary condition is imposed at the bottom boundary, and the equilibrium concentration values of K+ (\({c}_{{K}^{+}}^{{eq}}\)) and OH− (\({c}_{{{OH}}^{-}}^{{eq}}\)) are set at this boundary (Supplementary Table 1-3). A fixed cathodic potential \({\psi }_{s}{=V}_{{app}}\) and electrolyte current density \({J}_{l}=0\) are imposed at the left boundary. Relevant source data are provided as a Source Data file.As shown in Fig. 3b, it is apparent that small H2O/CO2 ratio and high polymer porosity enable high local CO2 mass transfer efficiency at the catalyst surface (Supplementary Fig. 9), resulting in enhanced CO2R partial current density at consistent applied cathodic potentials (vs. SHE, Supplementary Fig. 10a). Besides, the local pH increases monotonically with the decrease of the H2O/CO2 ratio (Supplementary Fig. 10b). This is expected because the total amount of OH− generated from CO2R increases with higher current densities at given cathodic potentials. Notably, the OH− concentration change is more pronounced in regions with low H2O/CO2 ratio (<0.3). Previous work has suggested that high OH− concentration could lower the activation barriers of CO-CO coupling28. Thus, it may also contribute to the observed reduced cathodic potential in the low H2O/CO2 ratio region (Fig. 3c). The overall high local CO2 concentrations also favor the production of C2+ products (Fig. 3d). However, the C2+ selectivity is less sensitive to these two variables (Fig. 3e). Indeed, on one hand, the decreased cathodic potentials at a low H2O/CO2 ratio corresponding to a decrease in C2+ products selectivity, similar potential dependent trend in CO2R selectivity observed previously52,53. On the other hand, the C2+ selectivity might also be influenced by the local CO2 concentration, as higher population of unreacted *CO2 may compete with CO dimerization for the Cu-sites. The combined effects of lower cathodic potential and higher local CO2 concentration retain the C2+ selectivity despite various in the H2O/CO2 ratio and porosity. As the pore size of the catalyst layer can affect the effective diffusion of CO2 gas47, we explored the changes in the CO2R performance when the catalyst layer pore size ranged from 10 nm to 100 nm35 (Supplementary Figs. 11 and 12). First, we simulated the local CO2 concentration as the pore size increased, however, with a relatively modest increase of approximately 1.5 mM when the pore size increased tenfold, from 10 nm to 100 nm. Consequently, the cathodic current density shows a minor decrease as the pore size decreases from 100 nm to 40 nm. Notably, we observed a sharp decline in the current density when we further reduced the pore size to 10 nm, indicating that significant CO2 mass transportation limitation occurred at this scale. Furthermore, the changes in both the local pH (Supplementary Fig. 11c) and cathodic potential (Supplementary Fig. 11d) as a function of pore size show a similar trend to what was observed for the cathodic current density, indicating that the effect of the pore sizes on the CO2R is minimal unless a small pore size of <10 nm predominates in the catalyst. Overall, these simulations suggest that regulating the polymer porosity and the local H2O/CO2 ratio are effective approaches for optimizing the microenvironment and enhancing the CO2R performance.Tuning the H2O/CO2 ratio at the surface of Cu catalystMotivated by the above simulation results, we sought to find a suitable polymer that could enable us to construct the Model II-type catalyst layer and afford an optimized microenvironment for efficient CO2R. The ideal polymer candidate should be hydrophobic and possess suitable porosity to manage an optimized local H2O/CO2 ratio (determined by both the water uptake ability and CO2 adsorption capacity). Additionally, good chemical stability, especially under practical CO2R conditions, is also critical.As shown in Fig. 4a, a fully perfluorinated polymer resin (PT) was selected due to its high hydrophobicity, CO2 adsorption ability, high porosity (Supplementary Table 3), and excellent chemical stability. First, we measured the CO2 adsorption capacities of PT and the above-mentioned polymers (PT95, PCR, PVDF, Nafion) using CO2 adsorption isotherms. As shown in Fig. 4b and Supplementary Fig. 13, PT exhibits significantly higher CO2 adsorption compared to the other polymers. Note that the CO2 adsorption measurements were conducted under dry conditions, thus, it is expected that the CO2 adsorption will be reduced if severe water uptake occurs, i.e., during flooding. Then, we measured the water uptake ability of these polymers (Methods and Supplementary Fig. 14). As depicted in Fig. 4b, under identical conditions, PT absorbed only 0.58 wt% of water, which is approximately four and twenty-four times lesser than that of the PVDF and Nafion, respectively. This low water uptake ability of PT is comparable to those of PCR and PT95 (Supplementary Tables 4 and 5). Taken together, we tentatively estimated the H2O/CO2 ratio based on the polymer loadings (Supplementary Table 6) and their corresponding water and CO2 uptakes. It was observed that the PT polymer led to the lowest H2O/CO2 ratio at the catalyst surface, which is 10 and 658 times lower than that of PVDF and Nafion, respectively (Fig.  4c, Supplementary Table 3). Notably, PT also exhibits a lower local H2O/CO2 ratio compared to both PCR and PT95, even though they are more hydrophobic than PT. We believe that the low H2O/CO2 ratio of PT will lead to further improved CO2R performance. To demonstrate this assumption, we prepared the Model II-type PT/Cu GDE following the above procedure and conducted comprehensive CO2R measurements.Fig. 4: Characterizations of Cu modified by different polymers.a Chemical structure of PT polymer resin. b CO2 adsorption isotherms and water uptake of PT, PT95, PCR, PVDF and Nafion polymer membrane. c calculated local H2O/CO2 ratio based on the polymer loadings and their water uptake and CO2 adsorption ability. d F 1 s XPS spectra of PT, PCR, PT95, PVDF and Nafion. e–j, FETEM images of (e, f) PT/Cu, (g, h) PVDF/Cu and (i, j) Nafion/Cu. k–p cross-sections of (k, l) PT/Cu, (m, n) PVDF/Cu and (o, p) Nafion/Cu based GDLs. In panels k–o, false colors applied to the images for clarity, pink: catalyst layer, blue: microporous layer, orange: gas diffusion layer. The white color on the top of the GDE is the spray coated PTFE. The average polymer thicknesses have been measured at least ten times at different samples/locations, more FETEM images can be found in Supplementary Information. Relevant source data are provided as a Source Data file.Characterizations of the Model II-type Cu catalyst layerPrior to the GDL preparation, X-ray photoelectron spectroscopy (XPS) was performed to study the composition of the polymer modified Cu. As shown in Fig. 4d, all polymer-modified Cu possess an F 1 s peak, confirming the surface coating of polymers. The chemical state of Cu was then investigated using X-ray absorption spectroscopy (XAS). The acquired Cu K-edge XAS spectra for the five samples were identical (Supplementary Fig. 15), indicating that the chemical state of the Cu catalyst was not altered by the polymer coating. The Cu LMM Auger spectra (Supplementary Fig. 16) further confirm this. Additionally, the X-ray diffraction (XRD) patterns confirm that the crystalline structure of the Cu NPs was retained during the polymer coating (Supplementary Fig. 17). Note that a small fraction of diffraction peaks for Cu2O were observed in the XRD patterns, which was probably caused by air exposure. The presence of Cu2O is further supported by the XAS results, which indicate mixed valence states in the precursor Cu NP (Supplementary Fig. 15)36.The morphology of the polymer-coated Cu was examined using Transmission Electron Microscopy (TEM). As shown in Fig. 4e–i and Supplementary Fig. 18, comparable morphologies were found for all five coated Cu NPs, which are consistent with the Scanning Electron Microscopy (SEM) images (Supplementary Fig. 19). Additionally, the uniform thin layers (marked by red arrows) on the Cu surfaces indicate that the polymer coatings are evenly distributed across the catalyst surface. The thicknesses of these polymer layers were determined by the high revolution TEM (HRTEM) (Fig. 4f–j, Supplementary 17b–f and Supplementary Fig. 20). With the typical polymer loadings, average coating thickness of 3.65 nm, 4.19 nm, 4.37 nm, 4.75 nm and 4.19 nm were determined for PT, PVDF, Nafion, PCR and PT95, respectively. The catalyst-loading in each electrode was approximately 1 mg cm−2, and the polymer loading could then be determined (Supplementary Table 6).The cross-sections of the polymer/Cu GDEs were characterized by SEM. As shown in Fig. 4k–p and Supplementary Fig. 18g–l, no significant differences were observed in the catalyst-layer thickness and the structures of the microporous layer across the three samples, indicating the similar electrode geometry for the polymer-modified GDEs. However, a slightly denser catalyst-layer was observed for PT/Cu, PCR/Cu and PT95/Cu (i.e., in absence of obvious cracking), as illustrated in Fig. 4p compared to Supplementary Fig. 18h–l. This difference could be attributed to a more uniform dispersion of the PT/Cu, PCR/Cu and PT95/Cu catalyst-inks, resulting in fewer agglomerates during spray coating. As a result, one could expect that the reactants (i.e., CO2) have less restricted access to the catalyst surface54. Overall, we have successfully prepared GDEs with Mode II-type catalyst-layers. While the chemical/physical properties of the Cu catalysts are nearly identical among these GDEs, we anticipate that the different polymer-coatings will lead to distinct CO2R performances due to the different local environment created, particularly the different local H2O/CO2 ratios.Electrochemical CO2R on polymer/Cu GDEs in flow cellTo assess the validity of our model and design, CO2R catalyzed by the above GDEs was evaluated using a flow-cell reactor under identical conditions (Supplementary Fig. 5). Hereinafter, all potentials are referenced to the reversible hydrogen electrode (RHE), unless otherwise noted.As shown in the polarization curves (Fig. 5a, Supplementary Fig. 21a), PT/Cu exhibited higher apparent activity than other polymer/Cu GDEs under the same conditions, in line with simulated results (Supplementary Fig. 21b). Supplementary Fig. 22 depicts the FEs for all detectable products from the CO2R with different GDEs. In general, all five GDEs exhibit similar products distributions, suggesting similar reaction pathways involved (Supplementary Tables 7–11). Encouragingly, the PT/Cu GDEs clearly outperforms other GDEs, as evidenced by its significantly reduced HER ( < 8% from −0.1 to −2 A cm−2) and improved C2+ selectivity at high current densities (Fig. 5b, Supplementary Fig. 23). Besides, its FEC2+ increased monotonically with the current density and reached 87.4% at −2 A cm−2, exceeding those of the state-of-the-art systems under similar conditions4,35,55,56. In contrast, other GDEs reached their peak C2+ selectivity at relatively lower current densities. Specifically, PCR exhibited an FEC2+ of 86% at −1 A cm−2, while PT95/Cu, PVDF/Cu and Nafion/Cu achieved peak FEC2+ at −0.5 A cm−2, followed by rapid deactivation in CO2R (Fig. 5b). Meanwhile, the corresponding HER activity for these polymer/Cu GEDs increased sharply after reaching their peak FE2+. This increase can be attributed, on one hand, to the rapidly increasing of local H2O/CO2 ratio as the consumption of CO2 and accumulation of H2O during CO2R process, on the other hand, to the occurrence of severe flooding, especially in the case of PVDF/Cu and Nafion/Cu. The easy flooding phenomenon of PVDF/Cu and Nafion/Cu will be thoroughly discussed later.Fig. 5: Electrochemical CO2R on different polymer/Cu GDEs.a Polarization curves of the different polymer/Cu GDEs (with 85% iR compensation. The solution resistance for PT/Cu, PCR/Cu, PT95/Cu, PVDF/Cu and Nafion/Cu are 0.31 Ω, 0.35 Ω, 0.33 Ω, 0.31 Ω and 0.38 Ω, respectively). b C2+ FE and c overpotential on these polymer/Cu GDEs under the same current density. d C2+ partial current density. e H2 partial current density and f C2+ FE on these polymer/Cu GDEs under identical cathodic potentials. g stability of PT/Cu, Nafion/Cu and PVDF/Cu at −0.2 A cm−2 in 1 M KOH. h C2+ selectivity of PT/Cu and Nafion/Cu before and after stability test at −0.2 A cm−2 and −1 A cm−2 with 1 M KOH as the electrolyte. The stability tests were stopped when H2 FE higher than 15%. For panels a–f, the test conditions include 2 M KOH as the electrolyte, a CO2 flow rate of 24 sccm, and a catholyte flow rate of 2 mL/min. For all the tests in a flow cell, the anolyte flowrate is 5 mL/min. Regarding the cathodic potential in panel a, 85% iR correction was applied. For panels c–g, 100% iR correction was applied. Details regarding the iR correction can be found in the Method section. The solution resistances for all measurements in panels b–f were recorded and plotted in Supplementary Fig. 26a. The error bars in panel b, c and f represent standard deviations from at least three independent measurements. Relevant source data are provided as a Source Data file.Furthermore, PT/Cu required significantly reduced overpotentials compared to the other four GDEs, particularly at high current densities (Fig. 5c). Consequently, PT/Cu afforded significantly higher partial current densities for C2+ production at the same cathodic potential (Fig. 5d). As a result, high cathodic energy efficiency (EE) of over 50% was obtained for PT/Cu catalyzed C2+ production at high current density of −2 A cm−2 (Supplementary Fig. 24), demonstrating the promise of practical implemetations4,28. Contrastingly, other polymer/Cu GDEs, namely PCR/Cu, PT95/Cu, PVDF/Cu and Nafion/Cu achieved their highest EEC2+ at much lower current densities: −1 A cm−2 for PCR/Cu and −0.5 A cm−2 for the other three polymer/Cu GEDs. We correlated the highest FEC2+ and jC2+ achieved by these five polymer/Cu GDEs with the corresponding estimated H2O/CO2 ratio (Supplementary Fig. 25). A clear scaling trend was observed, highlighting the significance of attaining and sustaining a low local H2O/CO2 for achieving selective and stable CO2R at high rate with enhanced energy efficiency. Also, the observed reduction in cathodic potentials aligns with the simulation results (Fig. 3c, d). To exclude the possibility of over-compensation of iR, particularly at high current densities, non-compensate overpotentials were compared among these samples (Supplementary Fig. 26), and the same trends were observed. Worth noting that, although PT95 and PCR exhibit similar local H2O/CO2 ratios (PCR: 0.06, PT95: 0.07, Supplementary Table 3), their CO2R performances differ significantly. This discrepancy may be attributed to the low porosity of both polymers (<0.2 with PT95 at 0.07, PCR at 0.14). Such low porosity hinders the diffusion of local species (e. g. CO2, H2O, OH−), and substantially reducing local CO2 concentration, and negatively impacting the CO2R performances (Fig. 2b–d). Overall, polymers capable of regulating lower local H2O/CO2 ratios can facilitate efficient CO2R at higher current densities and increase energy efficiency.On the other hand, PT/Cu exhibits relatively low C2+ selectivity but higher CO selectivity compared to other polymer/Cu GDEs at low current densities, i.e., \(\le\)−0.5 A cm−2 (Fig. 5b, Supplementary Fig. 27a, Supplementary Table 12). This can be attributed to the lower cathodic potentials applied (Fig. 5d, Supplementary Fig. 27b), in agreement with our modeling results and the previously reported potential-dependent selectivity trends for CO2R52,53. To further analyze the kinetics, we examined the partial current densities of C2+, CO and H2 for these GDEs (Fig. 5d, e and Supplementary Fig. 28). As anticipated, the enhanced C2+ and CO partial currents were associated with the decrease in local H2O/CO2 ratio regulated by the polymers. However, the H2 partial current densities and their trends were close for all five GDEs, as shown in Fig. 5e, suggesting that the HER activity is not directly associated with the local H2O/CO2 ratio. We then assessed the electrochemical active surface area (ECSA) of the five GDEs. While we acknowledge that the simplified ECSA measurements may not fully reveal the exact picture of the reaction interface during CO2R, we believe that the minor differences in their ECSA are likely not the primary factor causing the difference in their CO2R performance (Supplementary Fig. 29). We further plotted the selectivity of C2+ products and CO for each polymer/Cu electrode against their potentials, as illustrated in Fig. 5f and Supplementary Fig. 27b. Interestingly, in the non/minor-flooded region (<−0.65 V), the selectivity towards C2+ and CO are similar across all these polymer/Cu electrodes. However, further increase the cathodic potential tends to induce flooding in all polymer/Cu GDEs29. PT demonstrated the best flooding resistance, as evidenced by the continued increase in FEC2+. Altogether, we believe that a polymer coating capable of regulating a low local H2O/CO2 ratio will lead to an optimized CO2R microenvironment, minimizing concentration overpotential and allowing for improved CO2R current density. Also, the polymer coating will not significantly affect product selectivity at the same applied cathodic potential within the non/minor-flooded region.Next, we investigated the relative stability of these polymers by measuring the water and gas contact angles of the post-electrolysis electrodes. As shown in Fig. 6a, all five GDEs displayed excellent hydrophobicity prior to CO2R, with water contact angles great than 140°. Note that we were unable to capture the conventional contact-angle images for the as-prepared PT/Cu GDE owing to its super-hydrophobicity (Supplementary Movie 3). Nevertheless, PT/Cu maintained consistently high-water contact angles of over 150° throughout the one hour CO2R at −0.5 A cm−2. PCR/Cu and PT95 also exhibited commendable hydrophobic stability during the test, though they were slightly less stable than PT/Cu. In contrast, both PVDF/Cu and Nafion/Cu showed a rapid decrease in water contact angles under the same conditions (Supplementary Fig. 30). Particularly, the hydrophobicity of PVDF was significantly compromised within the first 10 minutes of electrolysis (Supplementary Movie 4). This decrease in hydrophobicity is closely associated with the electrode-flooding during CO2R. When flooding occurs, the GDL draws electrolyte from the reaction interface, resulting in continuous salt-precipitation. This lead to the blockage of CO2 transport-channels57 and consequently substantially increased concentration overpotential for CO2R. The change in hydrophobicity of these electrodes can be explained by electrowetting. To mimic the electrowetting effect under in-situ test conditions, we evaluated the variation in contact angle for PT/Cu and Nafion/Cu by employing a custom-designed in situ water contact angle assessing platform. As shown in Supplementary Fig. 31, the droplet contact angle on PT/Cu exhibited a smaller change compared to that on the Nafion/Cu electrode. This observation further confirms the enhanced hydrophobic stability provided by PT modification.Fig. 6: Post CO2R, in-situ measurements and reaction kinetics analysis.a water contact angle for the five polymer/Cu GDEs after CO2R at −0.5 A cm−2. b in-situ XAS measurements. c CO2R performance of C2+ selectivity and cathodic potentials on the function of CO2 partial pressure at −0.5 A cm−2, and d the influence of the local H2O/CO2 ratio (regulated by changing the CO2 partial pressure) on CO2R performances under identical cathodic potentials. The CO2 partial pressure was adjusted by mixing CO2 with Ar. A 50% CO2 partial pressure correlates to a doubled H2O/CO2 ratio than that of 100% CO2 concentration. e Nyquist plots of electrochemical impedance spectroscopy at −0.38 V vs. RHE in flow cell, the inset shows the equivalent circuit used for the curve fitting. The symbols in this figure are original data, the solid lines are fitting data. f Tafel plots for the CO2R to C2+ products for PT/Cu, PCR/Cu, PT95/Cu, PVDF/Cu and Nafion/Cu GDEs. The average equilibrium potential from CO2R to C2+ products was deemed as 0.09 V vs. RHE. All the CO2R tests here were conducted in a flow cell, and the electrolyte is 2 M KOH. For panels c, d and f, 100% iR correction was applied. Details regarding the iR correction can be found in the Method section. For the overpotential calculation in panel f, the average equilibrium potential for C2+ products were set as 0.09 V. The error bars in panel c and d represent standard deviations from at least three independent measurements. Relevant source data are provided as a Source Data file.We also designed and conducted gas contact angles measurements for these GDEs to evaluate their catalyst layer gas affinity, mimicking the reaction conditions, i.e., in presence of aqueous electrolyte. After CO2R electrolysis at −0.5 A cm−2 for 30 min, both Nafion and PVDF displayed significantly large gas contact angles (Supplementary Fig. 32), signifying strong gas repellence and their diminished hydrophobicity (Fig. 6a). PCR, however, show a contact angle of just 23° after CO2R, suggesting a much better gas affinity. Both PT and PT95 remain good gas affinity, as the gas bubbles were quickly absorbed by the surface (Supplementary Movies 5 and 6). Furthermore, even when the testing durations and currents were extended for PT/Cu, it still maintained superior gas affinity (Supplementary Fig. 33). Comparing the stability of water hydrophobicity and gas affinity among these GDEs, PT clearly outperformed the other polymers in terms of stability across both time and current densities scales. This characteristic ensured its robustness towards flooding and enhanced CO2R reaction rate. While for PVDF, although it initially showed a low H2O/CO2 ratio, its rapidly decreased water hydrophobicity and gas affinity likely led to substantially increase in local H2O/CO2 ratio, negatively impacting CO2R performance.PT’s stable hydrophobicity was further substantiated by the elemental mapping on the cross-section of the GDEs post CO2R. We selected PT as a representative for the three relatively stable polymers and compared it with Nafion and PVDF. As shown in Supplementary Fig. 34, the potassium (K) concentration exhibits the same trend as the contact angle measurements. Specifically, K was observed throughout the entire GDEs of Nafion/Cu and PVDF/Cu. However, only limited K was found in PT/Cu, indicating negligible electrolyte flooding occurred. As a result, >150 h continuous CO2R at −0.2 A cm−2 was demonstrated on PT/Cu, with negligible changes in FEs for both CO2R products and HER. In contrast, Nafion/Cu and PVDF/Cu exhibit poor CO2R stability, for only 11 h and 4 h, respectively (Fig. 5g). Furthermore, PT/Cu maintained stability at even high current density, i.e. −1 A cm−2, for ~10 h, indicating its impressive stability at large current densities (Supplementary Fig. 35). In contrast, Nafion/Cu displayed subpar CO2R performance towards C2+ products initially at this high current density, achieving a C2+ FE of only 65%, which then decline rapidly. Overall, PT/Cu exhibits exceptional stability in maintaining the local H2O/CO2 for CO2R, owing to its high chemical stability.Mechanistic investigationsThe chemical state of the Cu catalyst during CO2R was studied through operando XAFS spectroscopy with the same flow cell. As shown in Fig. 6b, the Cu K-edge XAFS spectra acquired for all five polymer/Cu GDEs suggest that Cu NPs were reduced to metallic state under real CO2R conditions58, indicating that the enhanced CO2R activity of PT/Cu was not resulted from differences in chemical state. Besides, the morphological structures of the Cu NPs were measured before and post CO2R for each GDEs, and no obvious changes were observed (Supplementary Fig. 36). We also do not expect the formation of any special active Cu-sites during the polymer coating. Therefore, the enhanced reaction CO2R kinetics of PT/Cu is primarily attributed to the optimized microenvironment, especially the low local H2O/CO2 ratio.The influence of local H2O/CO2 ratio was further investigated by tunning the CO2 partial pressure (PCO2) during CO2R on PT/Cu. As shown in Fig. 6c, d, we observed an initially increased FEC2+ when reducing the PCO2 (Supplementary Fig. 37, Supplementary Tables 13 and 14), until a very low PCO2 of 25% was applied, likely due to the insufficient CO2 supply. We believe that this initially increased FEC2+ was caused by the increased cathodic potentials (Fig. 6c)52,53. Similar CO2 diffusion induced concentration overpotential changes were observed elewhere59. In contrast, the FEC2+ of Nafion/Cu dropped rapidly and monotonically with decreasing PCO2 (Fig. 6c), indicating a severe CO2 transfer limitation under low PCO2, likely resulting from the unsatisfactory CO2 uptake and the GDL flooding. On the other hand, we anticipate an increase of local H2O/CO2 ratio when reducing the PCO2. To further understand the interplay between CO2R performance and the local H2O/CO2 ratio, we compared the FEC2+ from CO2R under both dilute and 100% CO2 conditions, at identical applied cathodic potentials. As shown in Fig. 6d, at low PCO2 we observed that the jC2+ notably decreased (Fig. 6d, right figure) while the FEC2+ remain relatively constant (Fig. 6d, left figure), likely due to the increased H2O/CO2 ratio under the same cathodic potentials. These findings align with the simulation results in Fig. 3c–e, indicating that the local H2O/CO2 ratio substantially impacts CO2R performance. It is crucial to note that our simulation results did not account for gas diffusion limitations caused by catalyst layer flooding, explaining why the CO2R performance at 25% PCO2 diverges from these simulations. The local H2O/CO2 ratio can be further regulated by adjusting the humidification of inlet CO2. As presented in Supplementary Fig. 38a, an increase in the inlet CO2 humidification results in a gradual rise in cell voltage with increasing temperature despite the changes are minor. As a result, this higher cell voltage led to an increase in C2H4 selectivity while reducing CO selectivity, as shown in Supplementary Fig. 38b. We anticipate that higher temperature would lead to a significant increase in CO2 feed humidity, which on the other hand will result in increased H2O/CO2 ratio and consequently declined CO2R activity as we observed in this study. Taken together, this phenomenon is in line with our conclusion and further confirms the critical role of the local H2O/CO2 ratio in regulating CO2R performance.Electrochemical impedance spectroscopy (EIS) was employed to probe the CO2R kinetics. As shown in Fig. 6e, the Nyquist plots for all GDEs exhibit a single apparent semicircle. Based on the proposed equivalent circuit60, we defined Rct and Rs as the charge transfer resistance and reactor solution resistance, respectively, and conducted the curve fitting. As a result, we observed a significantly smaller Rct for PT/Cu compared to Nafion/Cu at −0.38 V. We attribute this to the enhanced conversion rate of CO2 to CO on PT/Cu, resulting higher CO coverage and further promote the C2+ formation (Fig. 6e)61,62. This is evident by the higher CO partial current density observed for PT/Cu compared to the other four GDEs (Supplementary Fig. 28). Besides, a potential dependent EIS measurement was conducted for PT/Cu (Supplementary Fig. 39, Supplementary Table 15). The characteristic frequency of the semicircle shifts to higher values with increase in the cathodic potential, corresponding to an exponential decrease in Rct63,64. On the other hand, the measured Rs for each GDE were nearly identical since the same reactor configuration was employed (Supplementary Fig. 40). Note, the similar Rs for these polymer/Cu GDEs also suggests that the ionic conductivities at the reaction interface are not the limiting factor for CO2R on the GDEs, owing to the thin nature of the polymer coating, i.e., 2-5 nm (Fig. 4e–j). To further validate that the ions transfer across the thin polymer film is not the limiting factor in our system, we simulated the limiting current density under minimal water content in the PT polymer (extreme conditions). As shown in Supplementary Fig. 41, when considering a thin polymer layer (i.e. 3 nm) and only the diffusion of K+ (as K+ is the most abundant cation in electrolyte), it was found that even with a very low water content and a polymer porosity decreased to 0.1, the corresponding limiting current density could still reach 29.2 A cm−2. This value is substantially higher than the maximum current density achieved in our experiments (i.e. −2 A cm−2). This result indicates that the thin polymer coating is unlikely to hinder the ionic transfer across the reaction interface, at least within the current density range we measured. Next, Tafel plot-type analysis was conducted to investigate the kinetics of CO2R. As shown in Fig. 6f, we observed a decrease in Tafel-slope value (from 143.4 mV dec−1 to 90.1 mV dec−1) in the order of PVDF>Nafion>PT95 > PCR > PT. The above observations imply that the polymers possess stable and low local H2O/CO2 ratio can faciliate the CO2R kinetics, which increase the reaction rates towards C2+ products. Overall, these mechanistic investigations further confirmed the significance of an optimized local H2O/CO2 ratio, which we believe can serve as an effective descriptor for preparing active and stable GDEs for CO2R.Effects of polymer coating thickness and electrolyte pHWe studied the effect of PT coating thickness on CO2R performance. By varying the PT loading, Cu GDEs with different thicknesses of PT coatings were obtained (Supplementary Fig. 42 and Supplementary Table 16). As shown in Supplementary Fig. 43, the threshold for achieving the above promotion effect is the addition of 20 µL of PT solution during the catalyst ink preparation, corresponding to a coating thickness of ~2 nm (Supplementary Fig. 44). Beyond this threshold, the FEC2+ remains relatively stable until a high PT loading (~80 µL) (Supplementary Table 17), likely due to the inefficient water uptake and/or electronic conductivity caused by the thick polymer-coating (Supplementary Fig. 45). The influence of thickness also aligns well with the simulation results (Supplementary Fig. 8). Our observations indicate that, with an optimal polymer coating thickness where conductivity was less significant, the thickness does not really influence the effective diffusion coefficients of species within the domain. As a result, CO2 reactants and other species within the electrolyte can quickly diffuse through the polymer layer, at least significantly faster than the reaction kinetics. Consequently, a moderate PT loading of 40 µL was used for the above and subsequent measurements.CO2R in acidic electrolyte has shown promise in improving the CO2 utilization efficiency65,66. Thus, we also evaluated the promotional effect of PT for acidic CO2R. As shown in Supplementary Fig. 46c, compared to Nafion/Cu, PT/Cu exhibited significantly lower overpotential (by at least >0.3 V) to achieve high current densities with high FEC2+, confirming the enhanced acidic CO2R activity of PT/Cu. Note that the issue of flooding is less severe in acidic CO2R with Nafion/Cu catalysts (Supplementary Fig. 46b, Supplementary Table 18), likely due to the reduced formation of carbonate salts. Significantly, the maximum single-pass carbon efficiency of 46.3% (based on the CO2R products) was achieved on PT/Cu at −1 A cm−2 under acidic conditions (Supplementary Fig. 47).PT coatings in MEA reactor and other CO2R catalystsWe proceeded to assess the CO2R performance of PT/Cu in a membrane-electrode-assembly (MEA) reactor (Supplementary Fig. 48). As shown in Fig. 7a, the cell voltages of PT/Cu were significantly reduced compared to those of Nafion/Cu at broad current density ranges. Additionally, the selectivity of PT/Cu towards CO2R products remained steady at \(\ge\)80%, even at high current density of −1.2 A cm−2 (Fig. 7b, Supplementary Fig. 49, Supplementary Table 19). In contrast, Nafion/Cu exhibited much lower CO2R selectivity, likely due to flooding occurring at the catalyst layer (i.e., up to 60% H2 at −1 A cm−2, Fig. 7b, right part). Compared to the state-of-the-art Cu-based MEA, PT/Cu shows notably improved CO2R current (excluding HER) at a given cell voltage, indicating enhanced energy efficiency (Supplementary Fig. 50, Supplementary Table 20)15,67,68,69,70,71,72,73,74. Specifically, an encouraging EE of 21% was achieved at high current density of −1.2 A cm−2, which could be further improved to >30% by increasing the anolyte alkalinity (Supplementary Figs. 51 and 52). When incorporating electrocatalysts with improved intrinsic activity compared to pristine Cu, and optimized anodic catalysts, we anticipate achieving further enhanced EE that meets the threshold for practical applications69,72,75. Moreover, similar enhancements in CO2R selectivity and EE were observed for PT/Cu-based MEA using acid electrolyte (pH=1.5) (Supplementary Figs. 53 and 54).Fig. 7: PT coating for MEA reactor and other CO2R catalysts.a Full cell voltage on the function of the applied currents. b CO2R products distribution of PT/Cu and Nafion/Cu. c-d CO2R performance on (c) Sn NP and (d) Ag NP catalysts in a flow cell. For the MEA test, 1 M KOH was used as the anolyte with a flowrate of 3 sccm. The CO2 flowrate in the cathode side is 24 sccm. Before introducing CO2 into the MEA cell, it is passed through a sealed water container for preliminary humidification. For panel a, no iR correction was applied, for panels c and d, 100% iR correction was applied. Details regarding the iR correction can be found in the Method section. Relevant source data are provided as a Source Data file.Furthermore, we employed PT polymer for CO2R on Sn nanoparticles (Sn NP) and Ag nanoparticles (Ag NP) for producing formate and CO, respectively. As shown in Fig. 7c, the FEformate of PT/Sn reaches >93% at high current density of −1 A cm−2, at −0.7 V, which is ~220 mV lower than that of Nafion/Sn at the same current density (Supplementary Table 21). Similarly, PT/Ag achieved high FE of ~95% for CO formation at −1 A cm−2 at −1 V. In contrast, Nafion/Ag experienced a rapid flooding under the same condition, resulting in low FECO of 35% and high overpotential of > −2.0 V (Fig. 7d, Supplementary Table 22). Overall, these results indicate that our strategy has broad applicability for CO2R across various reactor configurations and electrocatalysts.
