Introduction

Electrochemically converting waste CO2 into multicarbon (C2+) product ethylene, presents a sustainable route to carbon-neutral chemical production1,2,3,4,5. While alkaline or near-neutral electrolytes enhance CO2 electro-reduction reaction (CO2RR) in Faradaic efficiency (FE) of reduction products and current density6,7, suppress the hydrogen evolution reaction (HER), they suffer from substantial CO2 loss due to carbonate formation, necessitating energy intensive separation and reducing overall energy efficiency8,9,10. Moreover, the carbonate salt precipitation greatly limits operational stability by breaking the gas-liquid-solid three-phase interface11,12,13,14.

Operating CO2RR in strongly acidic media offers an efficient route to addressing above issues15,16,17. The H+ ions can regenerate the carbonate into CO2, thus preventing salt precipitation and minimizing CO2 loss. However, in acidic media, HER prevails as proton reduction is kinetically favored over carbon–carbon (C−C) coupling18. Extensive experimental and theoretical studies have shown that cationic copper (Cu) promotes the formation of C2 products (such as ethylene and ethanol), whereas metallic Cu primarily generate methane and hydrogen19,20,21. However, theoretical calculations indicated that cationic Cu tends to reduce to metallic form, particularly under acidic condition22. Cu/ZrO2 catalysts with the Cu−O−M interface are widely employed in thermocatalysis23,24, where ZrO2 with high work function, acts as an electron donor, making the reduction of interfacial Cu species more challenging. Nevertheless, the Cu−O−M interface is generally unstable in the acidic media. Therefore, constructing a more stable Cu−OM interface is crucial for enhancing ethylene selectivity and partial current density during CO2 reduction process, especially under acidic conditions8,25,26.

In this study, we design a heterostructure of 5% yttrium-stabilized ZrO2 supported CuO (YSZ/CuO) catalyst, characterized by abundant oxygen vacancies on the support. This YSZ/CuO catalyst exhibits a more stable interfacial cationic Cu compared to a ZrO2 supported CuO (ZrO2/CuO) and unsupported CuO catalysts under same reaction conditions. Remarkably, the YSZ/CuO catalyst achieves competitive ethylene selectivity of 68.7% and partial current density of 545.0 mA·cm–2 under acidic media (pH = 2), surpassing the performance of most reported acidic CO2RR catalysts. In-situ characterization studies demonstrate that cationic Cu in YSZ/CuO is more resistant to reduction during electrolysis. Density functional theory (DFT) calculations further confirm that Y-doping enhances the charge density of interfacial Cu atoms and shortens the Cu-O bond, thereby protecting cationic Cu from reduction. Additionally, the presence of interfacial cationic Cu lowers the energy barrier for the rate-determining step of *CO–*CO coupling, subsequently promoting ethylene formation.

Results

Ex-situ characterizations

Yttrium-doped ZrO2 (YSZ) was prepared using a co-precipitation method, followed by the dispersion of CuO on YSZ through an additional precipitation process (see “Methods” for details). The incorporation of Y into ZrO2 lattice was confirmed by a downshift in the characteristic peak of ZrO2 in X-ray diffraction spectra due to the larger ion radius of Y3+ (Supplementary Fig. 1a)27,28, while Y2O3 phase appeared only when the Y loading reached 10 wt% (Supplementary Fig. 1b). CuO crystallite phase is visible in both ZrO2/CuO and YSZ/CuO. Scanning electron microscopy (SEM) images of ZrO2/CuO and YSZ/CuO (Supplementary Figs. 2 and 3) reveal that Y-doping results in a more uniform distribution of catalyst nanoparticles. Transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM) (Fig. 1a–d and Supplementary Fig. 4) show larger CuO particles (50–200 nm) surrounded by approximately 5 nm YSZ in YSZ/CuO, forming abundant CuO-YSZ interfacial boundaries.

Fig. 1: Morphological and fine structural characterization.
Fig. 1: Morphological and fine structural characterization.The alternative text for this image may have been generated using AI.
Full size image

a–c TEM (a, b) and HRTEM (c) images of the YSZ/CuO catalyst. The right-hand image in c shows a magnified view of the orange square, revealing the detailed interfacial structure. d HAADF-STEM image (left) and the corresponding EDX mapping images (Cu, Zr and Y) of the YSZ/CuO catalyst. e, f XANES spectra of the Zr K-edge (e) of YSZ/CuO, ZrO2/CuO catalysts, and Cu K-edge (f) of YSZ/CuO, ZrO2/CuO and CuO catalysts. g WT k3-weighted EXAFS contour plots of YSZ/CuO. h The fitting EXAFS spectra of YSZ/CuO, ZrO2/CuO and CuO.

X-ray absorption spectroscopy (XAS) and X-ray photoelectron spectroscopy (XPS) were used to explore electron transfer at the interfacial boundaries. Comparative analysis of Zr K-edge absorption between YSZ/CuO and ZrO2/CuO (Fig. 1e) indicates that Y-doping lowers the average oxidation state of Zr, further confirmed by the Zr 3d XPS spectrum (Supplementary Fig. 5). The decreased oxidation state of Zr in YSZ is reasonable as the incorporation of Y into ZrO2 lattice lowers the stoichiometric ratio of O/metal due to the nature of Y2O3, similar to aluminum doping in SiO2 in zeolites. Therefore, more oxygen vacancies are expected in YSZ compared to pristine ZrO2. Additionally, XAS of the Y K-edge (Supplementary Figs. 6–8) reveals an oxidized Y state in YSZ/CuO (along with Y 3d XPS spectrum, Supplementary Fig. 9), with additional longer first-shell Y−O coordination environment compared to that in Y2O3 (Supplementary Fig. 8). Since the unit cell of Y2O3 is larger than ZrO2 (Supplementary Fig. 1b), it is unlikely that the Y−O bond length is elongated after Y incorporated in ZrO2 lattice. Thus, the elongated Y-O bond in YSZ/CuO likely either arises from the O shift towards Zr atoms in Y−O−Zr or new Y-O-Cu coordination at the interface. In addition to formation of oxygen vacancies, the O shifts toward Zr atoms in Y−O−Zr could also lead to the decreased valence of Zr. The average valence state of Cu (Fig. 1f) decreases in YSZ/CuO compared to ZrO2/CuO and CuO samples, indicating electron transfer from YSZ to Cu, possibly through the Zr/Y−O−Cu coordination. In the WT extended X-ray absorption fine structure (EXAFS) profiles (Fig. 1g and Supplementary Fig. 10), there are two main signals, R = 1.2-1.5 Å (k = 5 Å–1) assigning to the first shell Cu-O coordination29, and R = 3.5–4.2 Å (k = 12.5 Å–1) assigning to Cu−(O)−Zr/Y/Cu coordination. Cu−Cu coordination could not be easily separated by Cu−O−Zr/Y in the WT EXAFS profiles, while they could be separated in EXAFS fitting results (Fig. 1h). EXAFS fitting results reveal a coordination number (C.N.) of 2.3 for Cu−O−Zr/Y in YSZ/CuO, which is higher than 1.5 in ZrO2/CuO (Fig. 1h, Supplementary Fig. 11, Table 1). Overall, the role of Y in YSZ/CuO could be summarized in two aspects: (i) contributing to smaller CuO and ZrO2 particles size and the formation of electron-deficient Zr4+ in the YSZ; (ii) enhancing the interfacial Cu−O−Zr/Y density, providing richer interfacial Cu sites for CO2RR.

CO2 electroreduction performance

The CO2RR performance was evaluated in a flow-cell at constant current densities from 400 to 1000 mA·cm–2. We optimized the concentration of K+ ions in the catholyte and selected 3 M KCl as the default electrolyte for the following research (Supplementary Figs. 12 and 13). Systematic screening of the doping amount of Y and the loading of CuO are displayed (Supplementary Figs. 14–17 and Fig. 2a), which illustrate the selectivity of all gas products and the total potential of YSZ/CuO at different current densities. Encouragingly, YSZ/CuO exhibits a faradaic efficiency (FE) for ethylene of 68.7% (FEC2 = 82.2%) at 800 mA·cm–2, and maintaining 66.3% of FEethylene at 1000 mA·cm–2. Under more acidic conditions (pH = 1), the FEethylene is 53.7% at 1000 mA·cm–2 (Supplementary Fig. 18). Control samples (ZrO2/CuO, Y2O3/CuO, and pure CuO catalysts), synthesized using the same method as YSZ/CuO, show lower ethylene selectivity, with maxima of FEethylene 51.1%, 46.0%, and 25.6% under similar conditions, respectively (Supplementary Fig. 19a–c and Fig. 2b). The YSZ/CuO shows lower HER selectivity (FE < 10%) than other counterparts, indicating the importance of YSZ with appropriate Y/Cu ratio in suppressing H2 production (Supplementary Fig. 20). The YSZ/CuO catalyst also demonstrates significantly higher current densities for both C2H4 and C2 products (Fig. 2c). To assess the role of acidic electrolytes in minimizing carbonate formation and improving CO2 utilization, we evaluated the CO2 utilization efficiency of YSZ/CuO at different CO2 flow rates (Supplementary Fig. 21a) and a constant current density of 800 mA·cm–2. At a CO2 flow rate of 5 sccm, a high CO2 utilization efficiency of 56.8 % was achieved (Supplementary Fig. 21b). At 800 mA·cm–2, the partial current densities for C2H4 and C2 products on YSZ/CuO are 545.0 and 653.1 mA·cm–2, respectively, which are 1.3 and 1.2 times higher than those of ZrO2/CuO, and 3.3 and 2.6 times higher than those of pure CuO. At 1000 mA·cm–2, the partial current densities of C2H4 and C2 products could even reach 663.2 and 808.0 mA·cm–2, respectively, which is competitive in acidic CO2RR conditions (Supplementary Table 2).

Fig. 2: CO2RR performance in acidic (pH = 2) electrolyte of YSZ/CuO catalyst.
Fig. 2: CO2RR performance in acidic (pH = 2) electrolyte of YSZ/CuO catalyst.The alternative text for this image may have been generated using AI.
Full size image

a Product distribution at constant current densities in 3 M K+. b Comparison of partial current density of ethylene at different current densities for YSZ/CuO, ZrO2/CuO, Y2O3/CuO and CuO catalysts. The error bars in (a) and (b) correspond to the s.d. of three independent measurements. Data are given as average ± s.d. c Comparison of partial current density of ethylene and C2 at different current densities for YSZ/CuO, ZrO2/CuO, and CuO catalysts. d Stability of YSZ/CuO at 400 mA·cm–2. All potentials without iR correction.

In addition, no significant changes in FEethylene and potential are observed after long-term testing over 14 h (Fig. 2d). The nanoparticle size in YSZ/CuO remains stable after stability test (Supplementary Fig. 22). Electrochemical impedance spectroscopy (EIS) (Supplementary Fig. 23) and electrochemical double-layer capacitance (Cdl) measurements (Supplementary Figs. 24 and 25) indicate the lowest charge resistance and highest electrochemical double-layer capacitance for YSZ/CuO, contributing to its competitive performance at high current densities. A comparison of Cdl with ethylene selectivity (Supplementary Fig. 26) excludes the effect of ECSA on FEethylene due to Y doping. Additionally, contact angles (CA) measurements (Supplementary Fig. 27) show minimal variation after Y doping compared to ZrO2/CuO, ruling out the effect of electrode wettability. In summary, the YSZ/CuO catalyst demonstrates efficient acidic CO2RR with a high C2H4 FE of 68.7% and a high current density of 545.0 mA·cm–2, marking a significant advancement in acidic CO2RR to ethylene.

Stabilization of cationic Cu revealed by in-situ XAS

In-situ X-ray absorption fine structure spectroscopy provides valuable insights into the changes in valence and coordination environments of Cu sites during CO2RR. The Cu K-edge XANES spectra reveals that under a potential of –1.3 V vs. RHE, the adsorption edge position of YSZ/CuO (Fig. 3a and Supplementary Fig. 28a) aligns closely with that of Cu2O, while for ZrO2/CuO its shifts between Cu foil and Cu2O, with CuO reducing to metallic Cu (Fig. 3b and Supplementary Fig. 28b, c). This suggests that cationic Cu species in ZrO2/CuO and YSZ/CuO are maintained under harsh conditions, with cationic Cu in YSZ/CuO remaining in a higher valence state. XRD spectra of spent catalysts show a predominant metallic Cu phase in two catalysts, but a broad peak for Cu2O is present only in spent YSZ/CuO (Supplementary Fig. 29). This indicates that YSZ helps protect interfacial cationic Cu from reduction and affects the reduction potential of subsurface Cu ions. Ex-situ XPS spectra (Fig. 3c) illustrates that the proportion of Cu+ species in spent YSZ/CuO, ZrO2/CuO, and CuO catalysts is 63%, 55%, and 42%, respectively30. Considering that CuO is almost fully reduced under reaction condition (Supplementary Fig. 28c), the presence of partial Cu+ species likely result from reoxidation when the spent samples are exposed to air before ex-situ XPS testing. Nonetheless, the higher proportion of Cu+ species in YSZ/CuO in XPS, alongside in-situ XAS results, suggesting that cationic Cu in YSZ/CuO is more resistant to reduction compared to ZrO2/CuO. The evolution of Cu species at various applied potential (Fig. 3d, e, Supplementary Fig. 30) clearly demonstrate that Cu+ in YSZ/CuO is more resistant to reduction to Cu0 compared to that in ZrO2/CuO under the same conditions31. The average oxide states of Cu species, calculated from the percentage of each species, quantitatively show that the decline in Cu valence is slower in YSZ/CuO as increasing applied potential. For example, at –1.3 V vs. RHE, where the Cu valence states in YSZ/CuO, ZrO2/CuO, and CuO are 1.15, 0.76, and 0.19, respectively, indicating that Y-doping plays a key role in preventing the reduction of cationic Cu (Supplementary Table 3).

Fig. 3: Dynamics of Cu species under CO2RR conditions.
Fig. 3: Dynamics of Cu species under CO2RR conditions.The alternative text for this image may have been generated using AI.
Full size image

a, b In-situ Cu K-edge XANES spectra of YSZ/CuO (a) and ZrO2/CuO (b) at applied potentials. c XPS spectra of Cu LMM, showing comparison of Cu+/Cu0 composition ratios on spent YSZ/CuO, ZrO2/CuO and CuO. d, e Evolution of various Cu species fractions as a function of the applied potential for YSZ/CuO (d) and ZrO2/CuO (e). f, g In-situ WT k3-weighted EXAFS contour plots of YSZ/CuO (f) and ZrO2/CuO (g) at applied potentials.

WT-EXAFS profiles (Fig. 3f, g and Supplementary Fig. 31) reveal two main signals corresponding to first shell Cu−O and Cu−(O)−Zr/Y/Cu coordination. During CO2RR, the peak intensity of Cu−O coordination significantly decreased, while Cu−Cu coordination related to metallic Cu gradually emerged, indicating the reduction of CuO into metallic Cu in both ZrO2/CuO and YSZ/CuO. However, the Cu−(O)−Zr/Y/Cu coordination in YSZ/CuO persists at –1.3 V vs. RHE, whereas it significantly decreases to a low level at –1.0 V vs. RHE for ZrO2/CuO (Fig. 3f, g). Although the peaks related to first shell Cu-O coordination diminishes, particularly for ZrO2/CuO, the peak for Cu−O−Zr/Y remains relatively stable in both ZrO2/CuO and YSZ/CuO (Supplementary Figs. 32–34). This suggests that the reduction of Cu at high applied potentials mainly originate from CuO particles rather than interfacial Cu−(O)−Zr/Y. The decreased first-shell Cu−O coordination number is accompanied by increased Cu-Cu coordination numbers (Supplementary Tables 4–6), which are 9.5, 6.9, 5.9 for CuO, ZrO2/CuO and YSZ/CuO, respectively, at −1.3 V vs. RHE. The Cu−(O)−Zr/Y coordination number in YSZ/CuO remains unchanged when the applied potentials increase to –1.3 V vs. RHE, whereas it decreases from 1.5 to 0.7 in ZrO2/CuO. These variations in the coordination environment suggest that Y-doping stabilizes interfacial cationic Cu by inhibiting its reduction during CO2RR. The relatively higher Cu valence in YSZ/CuO corresponds to its lower C.N. of Cu-Cu, and their influence on FEethylene is summarized (Supplementary Figs. 35 and 36). Both YSZ/CuO and ZrO2/CuO catalysts undergo further reduction after treatment at –3 V vs. RHE (Supplementary Fig. 37). Importantly, the YSZ/CuO catalyst maintains a noticeably higher average Cu valence than ZrO2/CuO. Moreover, neither catalyst is fully reduced to the metallic Cu0 state (as represented by Cu foil), even under these extreme conditions. Average oxidation states of Cu species during the electrolysis show positive correlation with ethylene selectivity (Supplementary Fig. 36c).

Theoretical calculations

DFT was employed to elucidate the enhanced stability of interfacial Cu in YSZ/CuO and its superior CO2RR performance. The CuO-ZrO2 heterostructures with various facet orientations were first investigated based on TEM and XRD results (Supplementary Figs. 1 and 38). The calculated formation energies indicate that the CuO(111)-ZrO2(111) heterojunction is thermodynamically the most stable (Supplementary Figs. 39 and 40). Additionally, the stability of ZrO2 upon doping a small amount of Y (5% in this work) has been previously reported32,33, and was further confirmed by the AIMD simulations (Supplementary Fig. 41), which shows no significant structural changes in ZrO2 at this doping level. Based on these findings, we constructed a CuO(111) and ZrO2 cluster model, along with four different Y-doped ZrO2 models (Supplementary Figs. 42 and 43). These cluster models were built by stacking the smallest repeating unit on a bilayer ZrO2(111) surface. Therefore, we constructed ZrO2/CuO and YSZ/CuO heterojunction models with (111) facet interfaces, without considering substantial structural change in ZrO2 upon Y doping34,35. The ZrO2/CuO heterojunction (Supplementary Fig. 44) reveals two distinct Zr−O−Cu interfaces depending on the origin of the oxygen atom, resulting in two configurations: CuO−Zr (O sourced from CuO, denoted as the α interface) and Cu−OZr (O sourced from ZrO2, denoted as the β interface). Similarly, four YSZ/CuO heterojunctions could be formed based on the Y doping position (Supplementary Fig. 45), all exhibiting similar α and β interfaces. Differential charge density distributions of ZrO2/CuO (Supplementary Fig. 46) indicate electron transfer as Zr→O→Cu in the α interface, whereas electron transfer form of Cu→O in the β interface. Bader charge analysis (Supplementary Tables 7 and 8) suggests the richest electrons accumulation on the Cu atom at α interface (0.06 eV) when Y directly bonds with interfacial O (Fig. 4a), higher than the undoped ZrO2/CuO (0.03 eV) and other Y doped cases. Consequently, we selected this Y-doped YSZ structure for further study. Differential charge density distributions confirm that Y doping (Fig. 4b, Supplementary Fig. 47) enhances electron accumulation around Cu atoms at the α interface. This suggests that the Y atom at the α interface enhances the Cu−O bond by transferring more electrons to Cu atom, and the formation of more α interface helps stabilize the Cu−O bond. XPS analysis (Supplementary Fig. 48) indicates an increase in Zrq+ (q <4) species after the CO2RR reaction, particularly for YSZ/CuO36. Additionally, electron paramagnetic resonance (EPR) signals observed at g = 2.006 after CO2RR indicate an increase in O vacancy in both ZrO2/CuO and YSZ/CuO (Supplementary Fig. 49). This is supported by Raman spectra of fresh and spent catalysts, showing higher possibility of O vacancy both fresh and spent YSZ/CuO. Interestingly, the Eg and B1g vibration modes of YSZ/CuO (Supplementary Fig. 50a) shift to lower wavenumbers after electroreduction, while those of ZrO2/CuO (Supplementary Fig. 50b) show no significant shift. This indicates that Y-doping might weaken the O−Zr−O bond during reduction, thereby increasing the density of O vacancy37,38. The rich O vacancy in YSZ facilitates the formation of the CuO−Zr α interface with O from CuO39. In addition, the integrated Crystal Orbital Hamilton Population (ICOHP) analysis (Fig. 4c, d) unveils a substantial improvement in the overlap of the Cu-O bond at the α interface after Y doping, transitioning from −0.53 eV to –0.59 eV. Further analysis indicates an increase in bonding components within the 3dyz−2px and 3dxy−2pz orbital interactions between Cu and O atoms at the α interface (Supplementary Fig. 51). All above indicate the enhanced stabilization of the Cu-O bond at the α interface due to Y doping.

Fig. 4: Theoretical simulation.
Fig. 4: Theoretical simulation.The alternative text for this image may have been generated using AI.
Full size image

a Schematic diagram of YSZ/CuO heterojunction. b The differential charge density distributions and Bader charge analysis of YSZ/CuO heterojunction. c, d The projected Crystal Orbital Hamilton Population (pCOHP) between the Cu and O atoms on α-interface of YSZ/CuO (c) and ZrO2/CuO (d). e In-situ ATR-SEIRAS spectra of CO2RR on YSZ/CuO catalyst. f, g Reaction free energy calculation models of ZrO2/Cu (f) and YSZ/O−Cu (g). h Reaction free energy calculation for the CO2 reduction to *OCCOH on YSZ/O−Cu and ZrO2/Cu catalysts.

In-situ attenuated total reflection-surface-enhanced infrared absorption spectroscopy (ATR-SEIRAS) conducted at different applied potentials (Fig. 4e, Supplementary Fig. 52) reveals a peak around 2030 cm−1 corresponding to the *CO atop intermediate40, which appears after −0.6 V vs. RHE on YSZ/CuO and ZrO2/CuO and redshifts with increasing voltage (2037 cm−1 → 2000 cm−1 and 2060 cm−1 → 2029 cm−1, respectively). This peak is absent in CuO, indicating that interfacial cationic Cu, rather than metallic Cu, is the active site for stabilizing *CO. The redshift of the *CO peak in YSZ/CuO relative to ZrO2/CuO indicates stronger *CO adsorption on YSZ/CuO. DFT calculations reveal a volcano-type relationship between *CO adsorption strength and the Cu valence state: *CO adsorption strengthens from Cu0 to Cu+ but weakens from Cu+ to Cu2+ (Supplementary Figs. 53 and 54). Consistent with in-situ XAS results (Fig. 3a–e), Cu+ in YSZ/CuO exhibits greater resistance to further reduction to Cu0, which accounts for the observed redshift of *CO peak in YSZ/CuO compared to ZrO2/CuO. Moreover, a stronger adsorption strength and larger adsorption capacity around 1590 cm−1, ascribed to the formation of *COCOH41, are observed on YSZ/CuO, suggesting higher C–C coupling activity compared to ZrO2/CuO. In-situ XAS experiments reveal rapid reduction of CuO to Cu within ZrO2/CuO, while interfacial cationic Cu is partially retained in YSZ/CuO. To investigate the CO2RR mechanism, we constructed ZrO2/Cu and YSZ/O-retained Cu (YSZ/O-Cu) heterojunctions models, where Cu is chosen instead of CuO, as it will be reduced under reaction conditions, and interfacial oxygen is retained in YSZ/O−Cu to reflect its higher Cu valence (Fig. 4f, g, Supplementary Figs. 55, 56, 58, and 59). The selection of these heterojunctions models is further supported by benchmarking them against support-free oxygen-modified Cu models and CuOx crystallite structures with varying Cu/O ratios (Supplementary Figs. 60–64). Compared with ZrO2/Cu and YSZ/O−Cu, the polarized Cu model does not accurately describe the observed catalytic activity trends, confirming that interfacial cationic Cu species—stabilized by the ZrO2 and YSZ support—are the key active sites. ATR-SEIRAS measurements revealed the presence of *CO and *COCOH intermediates, while no isolated *COH species were detected, leaving the sequence of *CO protonation and C–C coupling unresolved. Accordingly, we explored two plausible reaction pathways: (i) two *CO species first undergo C–C coupling, followed by protonation to form *COCOH21; and (ii) *CO is initially hydrogenated to *COH, which subsequently couples with another *CO8. In pathway (i), the rate-determining step is the C–C coupling, with reaction energy of 1.15 eV on ZrO2/Cu and 0.77 eV on YSZ/O−Cu (Fig. 4h, Supplementary Table 10). In pathway (ii), the rate-determining step is *CO protonation, with higher reaction energy of 1.33 eV on ZrO2/Cu and 0.99 eV on YSZ/O−Cu (Supplementary Fig. 57 and Table 9). These results indicate that pathway (i) is more energetically favorable, with the reaction energy for C–C coupling substantially lowered on YSZ/O−Cu (from 1.15 eV to 0.77 eV). Moreover, the free energy of *CO adsorption is more negative on YSZ/O−Cu (–0.11 eV) than on ZrO2/Cu (0.42 eV), suggesting stronger *CO binding on YSZ/O−Cu, consistent with in-situ ATR-SEIRAS result (Fig. 4e). By comparing the CO2RR performance of various catalysts at the same applied potentials used in the in-situ ATR-SEIRAS and XAS experiments, we conclude that cationic Cu species stabilized by Y-doping enhance *CO adsorption and facilitates C–C coupling reaction via the pathway (i), thereby contributing to superior ethylene selectivity of YSZ/CuO.

Discussion

In summary, we developed Y-doped ZrO2 (YSZ) supported Cu catalysts for acidic CO2RR. Y doping induced increased oxygen vacancies on the YSZ surface, facilitating the formation of CuO−Zr/Y with oxygen sourced from CuO. This led to a shortened Cu-O bond length compared to CuO and ZrO2/CuO, enhancing the stability of cationic Cu at the interface in YSZ/CuO under reducing conditions. YSZ/CuO achieved a high FEethylene of 68.7% for ethylene production, corresponding to a partial current density of 545.0 mA·cm−2 in pH = 2 electrolyte. Moreover, durability over 14 h with negligible changes in FEethylene and potential was observed for YSZ/CuO. DFT calculations indicated that the cationic Cu formed by the YSZ cluster helped facilitate the formation of *OCCOH, thus exhibiting competitive catalytic performance. This study demonstrated a strategy for stabilizing cationic copper during acidic CO2RR through interface engineering.

Methods

Synthesis of ZrO2 and YSZ

The synthesis of ZrO2 powder involved a hydrothermal method combined with an alkaline hydrolysis precipitation (AHP) process. Zirconium carbonate basic (ZCB, CAS: 57219-64-4, ZrO2 content > 40%), sodium hydroxide (NaOH, 99.99%), hydrogen peroxide (H2O2, 30%), and nitric acid were procured from Aladdin Co., Ltd. (Shanghai, China). Initially, 10 g NaOH and 10 g ZCB were dissolved in 20 g deionized water and heated in a hydrothermal kettle at 110 °C for 6 h. Upon cooling, the resulting mixture was transferred to a 100 mL round-bottomed flask and diluted with 42.5 mL of H2O. Subsequently, 5.1 mL of H2O2 was added at 50 °C for 5 h with continuous stirring. The excess solution was then removed via centrifugation, and the wet solids were dried overnight. For YSZ (yttria-stabilized zirconia) with varying Y content, yttrium nitrate (1%YSZ, 0.257 g; 3%YSZ, 0.770 g; 5%YSZ, 1.283 g; 10%YSZ, 2.566 g) was introduced during the hydrothermal step (110 °C 6 h), following the same procedure as the synthesis of ZrO2.

Synthesis of ZrO2/CuO and YSZ/CuO catalysts

0.1 g ZrO2 or YSZ was dispersed in 40 mL deionized water within a 100 mL round-bottom flask. Subsequently, x grams of Cu(NO3)2·3H2O (99.0–102.0%, SCR) dissolved in 12.5 mL of deionized water were added (YSZ/CuO-0.3, x = 0.09375; YSZ/CuO-0.6, x = 0.3775; YSZ/CuO-0.8, x = 0.8808, 0.2, 0.6, 0.8 represented molar percentage of CuO). The mixture was vigorously stirred, followed by the gradual addition of 0.5 M Na2CO3 (>99.9%, Macklin) solution until the pH reached 9–10. After stirring for 1 h, the mixture was allowed to stand for another hour prior to filtration. The resulting precipitates were thoroughly washed with deionized water, followed by drying at 60 °C. Finally, the ZrO2/CuO or YSZ/CuO were obtained after annealing the dried power at 600 °C for 4 h in an air with a heating rate of 2 °C·min−1.

Synthesis of CuO catalyst

0.2 g of Cu(NO3)2·3H2O was annealed at 600 °C for 2 h, with a heating rate of 2 °C·min−1.

Electrode preparation

A catalyst ink was prepared by mixing 15 mg of the as-synthesized catalyst with 0.75 mL of isopropanol (>99.7%, SCR), 0.2 mL of deionized water, and 50 μL of Nafion solution (Nafion5%-D520). The ink was then sprayed onto a commercial carbon paper sheet measuring 2.5 cm × 2.5 cm, resulting in an area loading of approximately 2.5 mg·cm−2 (the quality difference of carbon paper sheet before and after spraying).

Electrolyte preparation

As for catholyte, 1 M, 2 M and 3 M potassium chloride solution (>99.5%, SCR) were prepared, respectively, and the pH value was adjusted to 2.0 ± 0.1 with sulfuric acid solution (95.0–98.0%, SCR). And 0.25 M potassium sulfate solution (adjusted the pH to 2.0 ± 0.1 with sulfuric acid solution) served as the anolyte. All electrolytes were prepared 10 minutes before testing and stored at room temperature.

Characterizations

Powder X-ray diffraction patterns were obtained using a PIXcel 1 d diffractometer with Cu Kα radiation (λ = 1.5406 nm, 40 kV, 40 mA) at a scan rate of 2 min−1 in the 2θ range from 5 to 80°. Morphological characterization of the materials was conducted using a scanning electron microscope (SU-8010) with a beam voltage of 15 kV. High-resolution transmission electron microscopy (HRTEM) and elemental mapping images were acquired using a 2100 F microscope operating at an acceleration voltage of 200 kV and equipped with energy dispersive X-ray spectroscopy (EDX). X-ray photoelectron spectroscopy (XPS) analysis was performed using an Escalab 250Xi system. Electron paramagnetic resonance (EPR) spectroscopy (EMXplus-9.5/12, Bruker) was employed to analyze chemical information on the catalyst surface. The structure of the catalysts was investigated using Raman spectroscopy with a Raman microscope. The yttrium (Y) content in the catalysts was quantified using inductively coupled plasma optical emission spectrometry (ICP-OES, Varian T30-ES). The surface area was determined using the Brunauer-Emmett-Teller (BET) method.

X-ray absorption spectroscopy measurements

X-ray absorption fine structure (XAFS) spectroscopy was carried out using the RapidXAFS 2 M (Anhui Absorption Spectroscopy Analysis Instrument Co., Ltd.) by transmission (or fluorescence mode at 20 kV and 20 mA, and the Ge (440) spherically bent crystal analyzer with a radius of curvature of 500 mm was used for Zn Si (733). The in-situ X-ray absorption electrochemical cell was single-tank electrolytic cell (Supplementary Fig. 66a, b), consist of Ag/AgCl electrode and Pt as the reference electrode counter electrode, respectively. 0.5 M KHCO3 was used as the electrolyte, and circulated through peristaltic pumps at a rate of 20 mL·min−1. The signals were recorded after a 5-min electrolysis at different applied potential (–0.1 V, –0.3 V, –0.7 V, –1.0 V, –1.3 V vs. RHE, since the overall resistance of the in-situ XAS cell was larger than that of the home-made flow cell, the voltage selected for the in-situ experiment cannot be fully matched with the electrochemical experiment), to gain the stable data before the collection of XAFS spectra with constantly flowed gaseous CO2 of 30 mL·min−1.

Surface-enhanced infrared absorption spectroscopy

The modified electrochemical cell, supplied by Shanghai Yuanfang Tech, was used for conducting in-situ attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy experiments. The Thermo Nicolet spectrometer (IS50) was coupled with the cell, and a mercury-cadmium-telluride (MCT) detector cooled by liquid nitrogen was utilized for this purpose.

Electrochemical measurements

Electrochemical studies were performed under constant current conditions using a home-made electrochemical flow cell comprising a gas chamber, a cathodic chamber, and an anodic chamber (Supplementary Fig. 67a, b). The catalyst layer faced the cathodic chamber, with a geometric active surface area of 2 cm2. The cathode and anode chambers were separated by a CEM (Nafion 117, 2.5 cm × 2.5 cm, 0.183 mm). Catholyte and anolyte were applied through separated silicone tubes that each connected to a peristaltic pump, offering a constant flow rate of approximately 30 mL·min−1. All electrochemical tests were conducted on a CHI 660E instrument combined with a CHI 680 C instrument, utilizing an Ag/AgCl electrode (Gaoss Union, saturated potassium chloride solution) and IrOx/Ti-mesh (Dacheng) as the reference and counter electrodes, respectively.

For performance evaluation, the flow rate of CO2 through the gas chamber was maintained at a constant 30 mL·min–1 using a mass flow controller (MFC). All performance tests were performed using potentiostatic electrolysis. Gas products collected from the outlet of the gas chamber were analyzed online after 5-min electrolysis using a gas chromatograph (Agilent 8860 GC System) equipped with both a flame ionization detector and a thermal conductivity detector. And the flow rate of the outlet gas is detected by a soap film flowmeter (JCL-2010). Liquid products were collected after 20-min electrolysis and analyzed using 1H NMR spectroscopy (600 MHz, Agilent DD2 NMR spectrometer) with water suppression. D2O and dimethyl sulfoxide were utilized as the lock solvent and internal reference, respectively.

For CO2 utilization efficiency tests, the CO2 flow rate through the gas chamber was precisely controlled at 30, 20, 10, and 5 mL·min–1 using a mass flow controller (MFC), while all other experimental conditions were kept identical to those described above.

Double-layer capacitance (Cdl) measurements were conducted to assess the electrochemically active surface area of the electrode. Cyclic voltammetry (CV) was performed over a potential range from 0 to −0.1 V vs. Ag/AgCl, employing scan rates of 20, 30, 40, 50, 60, 80, 100, and 120 mV·s–1.

Additionally, electrochemical impedance spectroscopy (EIS) measurements were carried out by applying an alternating current voltage with a 5 mV amplitude over a frequency range from 105 Hz to 0.01 Hz.

All potentials were converted to RHE scale via the equation:

$${E}_{{RHE}}={E}_{{Ag}/{AgCl}}+0.197+0.059\times {pH}$$

The product quantifications of CO2 reduction

The faradaic efficiencies (FE) of all products can be calculated as follows:

$${FE}=\frac{{N}_{{product}}}{{N}_{{total}}}\times 100\%$$

For gaseous products, the amount of gas in each vial (the volume of the sample loop (V0) in our gas chromatograph is 1 cm3 is:

$$n=\frac{P\times {V}_{0}}{R\times T}=\frac{1.013\times {10}^{5}\times 1\times {10}^{-6}}{8.314\times 298.15}=4.09\times {10}^{-5}\,{mol}$$

The number of electrons required to form 1 molecule of CH4, C2H4, CO or H2 are, respectively, 8, 12, 2 or 2. For example, the number of electrons (Nethylene) needed to get x0 ppm of ethylene is:

$${N}_{{ethylene}}={x}_{0}\times n\times {N}_{A}\times 12e\,({N}_{A}=96485\,{{{\rm{C}}}}\cdot {{mol}}^{-1})$$

Total number of electrons (Ntotal) measured during this sampling period:

$${N}_{{total}}=\frac{I\times t}{e}=\frac{I\times \frac{{V}_{0}}{v}}{e}$$

where the flow rate of the outlet gas (v) is detected by a soap film flowmeter.

For liquid products, the volume of catholyte is V = 10 cm3. The number of electrons required to form 1 molecule of HCOOH, C2H5OH, CH3COOH and C3H7OH are, respectively, 2, 12, 8, 18. For example, the number of electrons (Nethanol) needed is:

$${N}_{{ethanol}}={C}_{{ethanol}}\times V\times {N}_{A}\times 12e$$

Total number of electrons (Ntotal) measured during this sampling period:

$${N}_{{total}}=\frac{I\times t}{e}\,(t=20\,min )$$

The partial current density of target product can be calculated as follows:

$${j}_{{product}}={j}_{{total}}\times {FE}$$

The CO2 utilization efficiency (UE) analysis

The CO2 utilization efficiency (UE) can be calculated using the following method42,43:

$${UE}=\frac{{V}_{{{CO}}_{2}\,{to}\,{gas}\,{products}}+{V}_{{{CO}}_{2}\,{to}\,{liquid}\,{products}}}{{V}_{{{CO}}_{2}\,{consumed}}}$$

The velocity of gas products converted from CO2 \({V}_{{{CO}}_{2}\,{to}\,{gas}\,{products}}\) can be calculated as:

$${V}_{{{CO}}_{2}\,{to}\,{gas}\,{products}}=({x}_{{CO}}+{x}_{{{CH}}_{4}}+2{x}_{{{{{\rm{C}}}}}_{2}{{{H}}}_{4}})\times {V}_{{outlet}}$$

The velocity of gas products converted from CO2 \({V}_{{{CO}}_{2}\,{to}\,{gas}\,{products}}\) can be calculated as:

$$ {V}_{{{CO}}_{2}\,{to}\,{liquid}\,{products}}=\\ \left({x}_{{HCOOH}}+{x}_{{{CH}}_{3}{OH}}+2{x}_{{{{{\rm{C}}}}}_{2}{{{H}}}_{5}{OH}}+{2x}_{{{CH}}_{3}{COOH}}+{3x}_{{{{{\rm{C}}}}}_{3}{{{H}}}_{7}{OH}}\right)\times \frac{{RT}}{P}$$

The gas velocity of total consumption of CO2 \({V}_{{{CO}}_{2}\,{consumed}}\) can be calculated as:

$${V}_{{{CO}}_{2}\,{consumed}}={V}_{{{CO}}_{2}\,{to}\,{gas}\,{prosucts}}+{V}_{{{CO}}_{2}\,{to}\,{liquid}\,{products}}+{V}_{{{CO}}_{2}\,{loss}}$$

The gas velocity of CO2 loss in solution \({V}_{{{CO}}_{2}\,{loss}}\) can be calculated as:

$${V}_{{loss}}= {V}_{{inlet}}-\left({V}_{{outlet}}-{V}_{{{{H}}}_{2}}\right)-{x}_{{{{{\rm{C}}}}}_{2}{{{H}}}_{4}}\times {V}_{{ouelet}}-({x}_{{HCOOH}}+{x}_{{{CH}}_{3}{OH}} \\ +2{x}_{{{{{\rm{C}}}}}_{2}{{{H}}}_{5}{OH}}+2{x}_{{{CH}}_{3}{COOH}}+3{x}_{{{{{\rm{C}}}}}_{3}{{{H}}}_{7}{OH}})\times \frac{{RT}}{P}$$

The gas velocity of H2 generation \({V}_{{{H}}_{2}}\) can be calculated as:

$${V}_{{{{H}}}_{2}}={x}_{{{{H}}}_{2}}\times {V}_{{outlet}}$$

DFT calculation

Density functional theory (DFT) calculations were conducted using the Vienna Ab initio Software Package (VASP)44,45, utilizing the generalized gradient approximation (GGA)46 with the Perdew-Burke-Ernzerhof (PBE)47 exchange-correlation functional. Projector augmented wave (PAW) potentials were employed to describe the core electrons, and Grimme D348 correction was included in the calculations to consider dispersion interactions between the adsorbates and catalysts. A comparison of extended ZrO2/YSZ and ZrO2/YSZ cluster-based heterojunction model reveals minimal discrepancies in calculated CO adsorption energies (Supplementary Fig. 65), validating the use of cluster models for computational efficiency. To approximate the experimental atomic ratios while maintaining reasonable computational efficiency, we constructed truncated models including 9-metal-atom clusters for YSZ and ZrO2, each with 18 O atoms (either 9 Zr atoms or 8 Zr+1 Y atoms). As the size of Cu species are larger than the YSZ and ZrO2, larger truncated slabs of CuO (111) and Cu (111) were used to model the initial and reduced Cu states, respectively. The CuO (111) slabs in ZrO2/CuO and YSZ/CuO were modeled using a 6×4 unit cell with three atomic layers (72 Cu and 72 O atoms). A vacuum of 25 Å was applied to minimize the effect of the slab on the calculations. Cu (111) slabs for free energy calculations were modeled using an 8×5 unit cell with three atomic layers (120 Cu atoms) and the same 25 Å vacuum spacing.

A plane-wave basis set with an energy cutoff of 400 eV was employed, as for electronic structure calculations, the Brillouin zone was sampled at a 3 × 2 × 1 k-point mesh. Convergence tolerances for force and energy were set to 0.03 eV·Å–1 and 10–5 eV. For structure relaxations, the convergence tolerances for force and energy were adjusted to 0.05 eV·Å–1 and 10–5 eV, respectively. Only the gamma point in reciprocal space was utilized for these thermodynamic. Spin polarization was included in the calculations. As shown in Supplementary Tables 11 and 12, the magnetic moments of all Cu, Zr, and Y atoms at the ZrO2/CuO and YSZ/CuO interfaces were extremely weak, confirming that their influence on catalysis is negligible. DFT + U corrections were found to have negligible impact on both CO adsorption energy (ΔE*CO) and C–C coupling energy barriers (ΔGC–C) (Supplementary Table 13), thus, to optimize computational resources, these corrections were not employed in the final calculations.

AIMD simulations

Ab initio molecular dynamics (AIMD) simulations were conducted using the Vienna Ab initio Simulation Package (VASP), version 5.4.4. The simulation employed a time step of 1 femtosecond (fs), and canonical ensemble conditions (NVT) were maintained through a Nose-Hoover thermostat set to a target temperature of 298 Kelvin (K) to simulate reaction temperature of CO2RR. Initial equilibrium was achieved through a 10 picosecond (ps) relaxation period, followed by a 20 ps production phase for data collection and analysis. Notably, spin polarization and dipole corrections were omitted from the calculations. The plane-wave basis cutoff energy was set to 400 electron volts (eV), and the Perdew-Burke-Ernzerhof with Grimme’s D3 dispersion correction (PBE-D3) method was utilized to account for van der Waals interactions. Notably, calculations were performed solely at the gamma point in reciprocal space.