Maintaining local alkalinity of CO-electroreduction full cell by silica-confined electrocatalysts in membrane electrode assembly

Abstract

Carbon monoxide electroreduction in alkaline membrane electrode assembly represents an effective approach to achieve carbon neutrality. However, its performance is currently limited by the insufficient modulation of local alkalinity at a full cell level. In this work, we reveal that confining the in situ generated hydroxide at cathode and enriching the bulk hydroxide to anode are the key factors for an efficient carbon monoxide–electroreduction full cell. We thereby propose a silica-confined strategy for electrocatalyst design to maintain high local alkalinity at both cathode and anode by the strong Lewis acid–base interaction between highly electrophilic silicon atom and hydroxide. The developed copper/silica cathode and cobalt/silica anode successfully promote cathodic multicarbon formation and anodic oxygen evolution, thereby improving the full cell energy efficiency. Even under high-rate electrolysis at 900 milliamperes per square centimeter, the selectivity and energy efficiency of multicarbon products remain above 80 and 30%. This achievement highlights the significance of modulating the dynamic hydroxide transport at full cell in enhancing carbon monoxide electroreduction performance.

Customized dynamic OH⁻ transport in membrane electrode assembly promotes energy-efficient CO electroreduction.

INTRODUCTION

Membrane electrode assembly (MEA) as a state-of-the-art electrochemical device holds promise on energy-efficient CO₂ conversion and high-value multicarbon (C₂₊) products (1–4). Substantial efforts have been devoted to modulating Cu-based catalysts for enhanced C₂₊ selectivity (5–8). High local alkalinity facilitates the efficient formation of C₂₊ products (9–12). However, direct CO₂ electrolysis to C₂₊ products on Cu-based catalysts in alkaline MEA is limited by poor device performance caused by carbonate precipitation (13–16). As a necessary step in CO₂ electrolysis, CO reduction reaction (CORR) has the prominent advantage of high operation stability in alkaline environment (17–21). A cascade CO₂ conversion route hence emerges, in which CO is efficiently generated via a two-electron transfer and then supplied to alkaline MEA for C₂₊ synthesis (22–25). Although the CO electrolysis operates under alkaline environment, there is currently a lack of understanding regarding how to maintain high local alkalinity at both cathode and anode within a full cell. This gap poses challenges for achieving efficient CO-to-C₂₊ conversion under high current densities.

In a typical flow cell configuration, the catholyte OH⁻ can facilely reach the cathode through diffusion (Fig. 1A). Increasing catholyte alkalinity thereby creates a strong alkaline microenvironment and facilitates C₂₊ formation in a flow cell. However, in the catholyte-free MEA configuration, the transport of anolyte OH⁻ across anion exchange membrane (AEM) to cathode is limited (Fig. 1B). As revealed by the OH⁻ crossover measurement (Fig. 1C), the slight change in the pH of K₂SO₄ catholyte confirms the restricted crossover of KOH anolyte through the quaternary ammonia poly(N-methyl-piperidine-co-p-terphenyl) (QAPPT) AEM. Using Cu nanoparticles as the model catalyst (fig. S1), increasing anolyte alkalinity in MEA has a relatively limited effect on improving the selectivity for C₂₊ products (fig. S2). The evident OH⁻ adsorption peak on Cu crystal after CORR in KCl catholyte (Fig. 1D) demonstrates that the cathodic local OH⁻ in catholyte-free MEA originates from the in situ generated OH⁻ during CORR rather than from bulk OH⁻. The crux of maintaining high local alkalinity at the cathode to improve C₂₊ selectivity hence lies in the confinement of in situ generated OH⁻.

Fig. 1. Origin of local OH⁻ in catholyte-free MEA.

Schematic illustrations on bulk OH⁻ transport in (A) flow cell and (B) MEA configurations. (C) Measurement on anolyte OH⁻ transport across AEM under open circuit voltage (OCV). (D) In situ OH⁻ adsorption curves recorded on Cu electrode before and after CORR at −100 mA cm⁻² in 1 M KCl. j represents current density. (E) CORR performances on Cu cathodes modified with cation exchange ionomer (CEI) (i.e., Nafion) or anion exchange ionomer (AEI) (i.e., QAPPT). (F) Schematic illustration on in situ OH⁻ transport under electric field. (G) In situ OH⁻ adsorption curves recorded on Cu electrode after CORR or hydrogen evolution reaction (HER) at different current densities in 1 M KCl.

Ionomers are widely used to modulate ion transport in the microenvironment (26–28). Anion exchange ionomer (AEI; e.g. QAPPT) conducts in situ OH⁻ to the anode, leading to decreased local alkalinity (fig. S3A). Alternatively, cation exchange ionomer (CEI; e.g. Nafion) retains in situ OH⁻ through electrostatic repulsion, thereby maintaining the high local alkalinity (fig. S3B). This substantial effect of CEI is verified by the increased C₂₊ Faradaic efficiency (FE) during CORR in alkaline MEA using Cu cathode, QAPPT AEM, IrO₂ anode, and 1 M KOH anolyte (Fig. 1E). However, under high current densities, the C₂₊ selectivity on the CEI-modified Cu cathode still decreases markedly. This performance decline is attributed to the migration of in situ OH⁻ to the anode under electric field (Fig. 1F), as verified by the diminishing OH⁻ adsorption peak on Cu crystal along with increased current density (Fig. 1G). At the anode side, oxygen evolution reaction (OER) occurs via the discharge of OH⁻. AEI conducts bulk OH⁻ to the anode, leading to improved OER performance (fig. S4). However, the migration of bulk OH⁻ through AEI is insufficient to support the OER under high current densities (29). The lack of OH⁻ in the anodic microenvironment causes sluggish OER kinetics and low energy efficiency (EE). Therefore, simultaneously confining the in situ generated OH⁻ at the cathode and enriching the bulk OH⁻ to the anode within the full cell are the key factors for achieving energy-efficient CO-to-C₂₊ conversion under high current densities.

We here propose a silica-confined strategy for electrocatalyst design to maintain high local alkalinity at both cathode and anode within the full cell. The vacant 3d orbital of electrophilic Si atom has strong coordination ability toward the lone pair electrons of OH⁻. This strong Lewis acid–base interaction enables the successful modulation of the dynamic OH⁻ transport. The strategy is first validated on the Cu/SiO₂ cathode, which restricts the migration of in situ generated OH⁻ from cathode to anode (Fig. 2A). Extending the strategy to the Co/SiO₂ anode enhances the migration of bulk OH⁻ from anolyte to anode. On the basis of the maintained local alkalinity within the CO-electroreduction full cell, cathodic C₂₊ formation and anodic OER are simultaneously promoted. High C₂₊ FE above 80% and EE above 30% are achieved at an industrial current density of 900 mA cm⁻², outperforming most of reported alkaline MEA. This study demonstrates that the catalyst modification strategy based on Lewis acid–base interaction is an effective approach for optimizing membrane electrolysis devices. This strategy offers universal significance for electrolysis devices involving proton-coupled electron transfer.

Fig. 2. Characterization of Cu/SiO₂.

(A) Schematic illustration on the confinement of in situ OH⁻ by SiO₂ under electric field. (B) SEM image, (C) TEM image, and element mappings of Cu/SiO₂. (D) Cu K-edge x-ray absorption near edge structure (XANES) spectra, (E) Fourier-transformed Cu K-edge extended x-ray absorption fine structure (EXAFS) spectra, and (F) wavelet transformation plots of Cu K-edge EXAFS spectra of Cu, Cu/SiO₂, and the reference materials. FT, Fourier-transformed.

RESULTS

Synthesis and characterization of Cu/SiO₂

Cu silicate precursor is first prepared by a hydrothermal method using SiO₂ spheres (fig. S5) as the template. The urchin-like hollow structure with uniform distribution of Cu, Si, and O elements throughout amorphous Cu silicate spheres is revealed by scanning electron microscopy (SEM), transmission electron microscopy (TEM), selected-area electron diffraction (SAED), and element mapping (fig. S6, A to F). Thermal reduction of Cu silicate in H₂ atmosphere at 800°C leads to Cu/SiO₂ (Fig. 2B). The fine Cu nanoparticles are confirmed by TEM image (fig. S6G). Cu (111) crystal face on the nanoparticles is revealed by high-resolution TEM and SAED (fig. S6H). Element mappings (Fig. 2C) demonstrate the hollow structure of SiO₂, and Cu/SiO₂ is further verified by x-ray diffraction (XRD) and x-ray photoelectron spectroscopy (XPS) for Cu 2p, Cu LMM, and Si 2p (fig. S7). The Cu cubic crystal corresponding to joint committee on powder diffraction standards (JCPDS) card no.70-3038 is demonstrated in fig. S7A. The metallic Cu state is also supported by fig. S7 (B and C). The absence of Si─O─Cu bond in Cu/SiO₂ is revealed by fig. S7D. The Cu content in Cu/SiO₂ is 40.6 wt % according to inductively coupled plasma atomic emission spectroscopy (ICP-AES).

As revealed by the Cu K-edge x-ray absorption near edge structure (XANES) spectra (Fig. 2D), the valence state of Cu in Cu/SiO₂ and Cu catalysts is close to metallic Cu foil. As shown in the Fourier-transformed Cu K-edge extended x-ray absorption fine structure (EXAFS) spectra (Fig. 2E), the evident Cu─Cu coordination at ~2.2 Å and the absence of Cu─O coordination at ~1.6 Å in Cu/SiO₂ and Cu catalysts indicate their native metallic Cu coordination architecture, which is further confirmed in the wavelet transformation plots (Fig. 2F). Pb underpotential deposition (UPD)–stripping test reveals that, with the same Cu metal loading, the electrochemical active surface areas (ECSAs) of Cu/SiO₂ and Cu are comparable (fig. S8). Therefore, the difference in the microenvironment induced by SiO₂ becomes the primary factor in regulating C₂₊ selectivity.

Confining in situ OH⁻ for enhanced cathodic CORR

Density functional theory (DFT) calculations (30–34) demonstrate the strong Lewis acid–base interaction between SiO₂ and OH⁻. Cu (111) supported with Si₆O₁₃ cluster is used as the theoretical model for Cu/SiO_2. The stable structure is verified by on-the-fly machine learning force field (MLFF) ab initio molecular dynamics (AIMD) simulation at 1000 K for 20 ps with a stepsize of 1 fs (fig. S9). As can be seen in Fig. 3A, Si serves as the active site for *OH adsorption (* denoting adsorption site), which exhibits substantially enhanced adsorption compared to Cu site (fig. S10). This strong Lewis acid–base interaction is also validated by NH₃ temperature-programmed desorption equipped with mass spectrometry (TPD-MS). As shown in fig. S11, after compositing SiO₂, the NH₃ desorption peak is substantially enhanced. In the OH⁻ adsorption curves recorded in 1 M KOH (Fig. 3B), the characteristic OH⁻ adsorption peaks occur at more negative potentials on Cu/SiO₂. The facile OH⁻ adsorption corresponds to high OH⁻ availability, as further confirmed by the enhanced K⁺ adsorption (Fig. 3C). In the catholyte-free MEA configuration, as OH⁻ accumulates in the cathodic microenvironment, anolyte K⁺ migrates to the cathode under electric field to maintain the electric neutrality. The adsorption of K⁺ on cathode contributes to stabilizing *CO and thereby facilitating C₂₊ formation via cation-enhanced C-C coupling (35–38). In situ Raman spectroscopy demonstrates the strong *CO adsorption on Cu/SiO₂ (Fig. 3D). The increased peak intensity and red-shifted peak position of Cu-CO along with cathodic polarization signify the enhanced adsorption and increased coverage of *CO on Cu/SiO₂.

Fig. 3. Enhanced CORR via maintained cathodic local alkalinity.

(A) Adsorption energy (E_ad) of *OH on different Si and Cu sites of Cu/SiO₂; the lines represent the average value. Avg, average. (B) OH⁻ adsorption curves recorded on Cu/SiO₂ and Cu electrodes in 1 M KOH. (C) ECSA-normalized K⁺ adsorption on Cu/SiO₂ and Cu electrodes. (D) In situ Raman spectra during CORR recorded on Cu/SiO₂ and Cu electrodes. (E) FE of all products during CORR on Cu/SiO₂. a.u., arbitrary units. (F) FE of C₂₊ products on Cu/SiO₂ and Cu. (G) Current densities of C₂₊ products and cell voltages using Cu/SiO₂ and Cu. (H) Free energy diagrams for 2*CO coupling to *OCCO on Cu and Cu/SiO₂. TS, transition state. Error bars represent the mean ± SD from three independent measurements.

The CORR performance on Cu/SiO₂ is assessed in alkaline MEA with CEI-modified cathode, QAPPT AEM, AEI-modified IrO₂ anode, and 1 M KOH anolyte. The C₂₊ products and hydrogen compose all the products from CORR, and C₁ product is absent (Fig. 3E), with the C₂₊ products such as ethylene, ethanol, acetate, and n-propanol being predominant. Even at current density up to 900 mA cm⁻², the FE of C₂₊ products on Cu/SiO₂ appears above 80%, in contrast to the sharply decreased C₂₊ FE to ~50% at 500 mA cm⁻² on Cu (Fig. 3F). In a similar cell voltage, the partial current density of C₂₊ products increases by three times from 246 mA cm⁻² at 2.89 V on Cu to 727 mA cm⁻² at 2.88 V on Cu/SiO₂ (Fig. 3G). Electrochemical impedance spectroscopy (EIS; fig. S12A) reveals the smaller charge transfer resistance (R_ct) of Cu/SiO₂, which also has high turnover frequency (TOF; fig. S12B) and high single-pass conversion efficiency (SPCE) of 73% at 900 mA cm⁻² (fig. S12C). The excellent C₂₊ activity on Cu/SiO₂ is supported by the reduced energy barrier for 2*CO coupling to *OCCO (Fig. 3H). The enhanced adsorption of *CO and *OCCO on Cu/SiO₂ (fig. S13A) contributes to increasing *CO coverage and stabilizing *OCCO. Cu/SiO₂ with adsorbed *OCCO has a shifted d-band center (ε_d) toward the Fermi level compared to Cu (fig. S13B). Charge difference density and Bader charge demonstrate the increased charge transfer to *OCCO (fig. S13, C and D). The main product shifts from acetate on Cu to ethylene on Cu/SiO₂ (fig. S14A). DFT results verify the selectivity difference between Cu and Cu/SiO₂ (fig. S14, B to D), in which the hydrogenation of *OCCOH to *HOCCOH is favored on Cu/SiO₂, leading to the formation of ethylene as the main product (3, 39). Varying the CEI fraction in the catalyst layer of Cu cathode has a limited influence on CORR performance (fig. S15). Cu/SiO₂ exhibits much high C₂₊ selectivity than Cu with varied CEI fraction, demonstrating that the Lewis acid–base interaction induced by SiO₂ governs the synergistic effect on improving C₂₊ selectivity. The effectiveness of the SiO₂-confined strategy in maintaining local alkalinity to promote C₂₊ formation is thus validated on Cu/SiO₂ cathode by the excellent CORR performance and DFT results.

Enriching bulk OH⁻ for enhanced anodic OER

High local alkalinity is also crucial for the anodic OER, as increasing the anolyte alkalinity substantially reduces the cell voltage (Fig. 4A). However, because of the rapid consumption of local OH⁻, this strategy tends to fail under high current densities. To enhance OH⁻ transfer from bulk to microenvironment, we extended the SiO₂-confined strategy to anodic Co/SiO₂ catalyst. As shown in Fig. 4B, the bulk OH⁻ is conducted through AEI and then enriched in the microenvironment by SiO₂, therefore maintaining high local alkalinity under high current densities. Co/SiO₂ is obtained by thermal reduction of Co silicate in H₂ atmosphere (figs. S16 to S18). The Co cubic crystal and hexagonal crystal corresponding to JCPDS cards nos. 89-4307 and 89-4308 are demonstrated in fig. S18A. The metallic Co state with slight surface oxidation of Co/SiO₂ is revealed by fig. S18B. The absence of Si─O─Co bond in Co/SiO₂ is confirmed by fig. S18C. The morphology of hollow SiO₂ spheres loaded with Co nanoparticles is revealed by TEM image and element mappings (Fig. 4C). The Co content in Co/SiO₂ is 44.9 wt % according to the ICP-AES result. As shown in fig. S19, using Si₆O₁₃ cluster–supported Co (111) as the theoretical model for Co/SiO₂, the DFT calculations demonstrate that Si serves as the active site for *OH adsorption, which exhibits substantially enhanced adsorption compared to Co site (fig. S20). Charge difference density and Bader charge confirm the increased charge transfer from Co/SiO₂ to *OH (fig. S21). Because SiO₂ is inert to the OER, Co serves as the active site. DFT results verify the migration of *OH from Si site to Co site during OER (fig. S22). The enhanced *OH adsorption increases the *OH coverage on Co surface, thereby facilitating the OER kinetics, as revealed by the decreased energy barrier for the rate-determining step (*O to *OOH) at high *OH coverage (fig. S23). NH₃ TPD-MS test (fig. S24) verifies the enhanced Lewis acid property of Co/SiO₂, compared with Co (fig. S25). Co/SiO₂ exhibits accelerated catalyst reconstruction, as revealed by the negatively shifted anodic oxidation peak (P_a) in the cyclic voltammetry curve (fig. S26). In situ Raman spectroscopy demonstrates the higher OER activity on Co/SiO₂ by the obvious *OOH adsorption peak (Fig. 4D), which is further confirmed by the linear scan voltammetry (LSV) curves recorded in 1 M KOH (Fig. 4E). As the bulk OH⁻ is enriched by SiO₂ via Lewis acid–base interaction, the high local alkalinity facilitates the evolution of Co to CoOOH. The Lewis acid property of SiO₂ and the valence-state evolution of Co to CoOOH jointly drive the improvement in OER performance. An industrial-grade current density of 200 mA cm⁻² can be driven at an overpotential of only 438 mV on Co/SiO₂, much lower than Co (481 mV). The higher OER activity of Co/SiO₂ is also verified by the smaller R_ct in the EIS results (Fig. 4F). When comparing with the widely used noble IrO₂, Co/SiO₂ delivers higher current density and smaller R_ct (Fig. 4G) under high overpotentials. The outstanding OER activity makes Co/SiO₂ an industrially viable anode for high-rate electrolysis.

Fig. 4. Enhanced OER via maintained anodic local alkalinity.

(A) Cell voltages recorded in 0.5 M K₂SO₄ or 1 M KOH at 100 mA cm⁻² with Cu cathode and IrO₂ anode. (B) Schematic illustration on the enrichment of bulk OH⁻ by SiO₂. (C) TEM image and element mappings of Co/SiO₂. (D) In situ Raman spectra during OER recorded on Co/SiO₂ and Co electrodes. (E) LSV curves for OER recorded on Co/SiO₂, Co, and IrO₂ electrodes at 5 mV s⁻¹ in 1 M KOH. (F) EIS results recorded at 1.53 V versus RHE on Co/SiO₂, Co, and IrO₂ electrodes from 100 kHz to 0.1 Hz with 5-mV amplitude. CPE, constant phase element. (G) Comparation on the R_ct of Co/SiO₂ and IrO₂ electrodes at different applied potentials. (H) LSV curves for methanol oxidation reaction (MOR) recorded on Co/SiO₂ and IrO₂ electrodes at 5 mV s⁻¹ in 1 M KOH with 1 M methanol. (I) Potential change from OER to MOR on Co/SiO₂ and IrO₂ electrodes at 100 mA cm⁻². (J) OER durability test on Co/SiO₂ electrode in 1 M KOH at 100 mA cm⁻².

Methanol oxidation reaction (MOR) in alkaline solution involves the nucleophilic attack of OH⁻ to methanol, which is more active on the OH⁻-rich surfaces (40). The SiO₂-confined strategy of enriching bulk OH⁻ via Lewis acid–base interaction is validated using methanol as the local alkalinity probe. The higher MOR activity on Co/SiO₂ is revealed by the LSV curves (Fig. 4H), which is also supported by the larger potential change (Fig. 4I) and bigger current difference (fig. S27) between OER and MOR. These results confirm the high local alkalinity on Co/SiO₂. A poisoning test using thiourea (CS(NH₂)₂) as Lewis base confirms the critical role of SiO₂ in enriching bulk OH⁻ (fig. S28). The marked decreased OER current density after adding thiourea reveals the poor OH⁻ availability around the blocked Lewis acid site. Durability test at 100 mA cm⁻² demonstrates the high stability of Co/SiO₂ during a 40-hour operation (Fig. 4J). The rapid consumption of local OH⁻ during OER at high current density leads to reduced OH⁻ concentration in the anodic microenvironment compared with the bulk anolyte, which keeps SiO₂ stable during OER in alkaline solution. The versatility of the SiO₂-confined strategy is thereby validated on the Co/SiO₂ anode through the enhanced OER activity.

Full cell CO electrolysis

The integration of Cu/SiO₂ cathode and Co/SiO₂ anode (Cu/SiO₂ || Co/SiO₂) results in negligible change in the distribution of C₂₊ products (Fig. 5A). Benefiting from the accelerated OER kinetics, the cell voltage during high-rate electrolysis at 900 mA cm⁻² is substantially decreased from 2.88 V using IrO₂ anode to 2.64 V using Co/SiO₂ anode (Fig. 5B). The EE of C₂₊ products correspondingly increases to 30.6% (Fig. 5C). More obviously, C₂₊ current density of 735 mA cm⁻² can be obtained at 2.64 V using Co/SiO₂ anode, in contrast to only 416 mA cm⁻² using IrO₂ anode even at higher cell voltage of 2.72 V (Fig. 5D). The improvement becomes more pronounced when compared to the original Cu || IrO₂ configuration. Durability test at 500 mA cm⁻² demonstrates the high stability of the Cu/SiO₂ || Co/SiO₂ full cell during a 22-hour operation (fig. S29). The performance achieved in this work outperforms most of reported alkaline MEA configurations (table S1). An energy-efficient full cell CO electrolysis is thereby achieved through the proposed SiO₂-confined strategy (Fig. 5E). The strong Lewis acid–base interaction between SiO₂ and OH⁻ enables the successful maintenance of high local alkalinity at both cathode and anode under high current densities. Specifically, Cu/SiO₂ cathode confines the in situ generated OH⁻ from migration to anode, enhancing the CO conversion to C₂₊ products. Co/SiO₂ anode enriches the bulk OH⁻ to support the high-rate OER, improving the full cell EE.

Fig. 5. Performance of full cell CO electrolysis.

(A) FE of all products during CORR with Cu/SiO₂ cathode, Co/SiO₂ anode, and 1 M KOH anolyte. Comparation on (B) cell voltage, (C) EE of C₂₊ products, and (D) current density of C₂₊ products with different cathode and anode configurations. (E) Schematic illustration on the SiO₂-confined strategy to maintain local alkalinity at both cathode and anode within the full cell. Error bars represent the mean ± SD from three independent measurements.

DISCUSSION

Alkaline MEA has the prominent advantages of compact configuration and low internal resistance, representing an advanced direction of electrochemical technologies. Unlike the classic electrode/solution interface, in the electrode/polyelectrolyte interface involved in alkaline MEA, water is no longer a prevalent solvent, and the charge carriers in the alkaline polyelectrolyte are solely the OH⁻ ions. Gaining an in-depth understanding of the dynamic evolution of OH⁻ in alkaline MEA serves as the scientific foundation for designing high-efficiency electrocatalysts to enhance device performance. The present study takes CO electrolysis in alkaline MEA as the research focus, revealing that confining in situ generated OH⁻ at the cathode and enhancing the transport of bulk OH⁻ to the anode are the key factors in improving device performance. On the basis of the aforementioned transport mechanism of OH⁻ within the MEA, leveraging the strong Lewis acid–base interaction between highly electrophilic Si atom and OH⁻, the present study proposes a SiO₂-confined electrocatalyst strategy. We developed cathodic Cu/SiO₂ and anodic Co/SiO₂ electrocatalysts that effectively maintain high local alkalinity at both electrodes under high current densities, thereby simultaneously enhancing the performance of cathodic CO reduction and anodic OER. This achievement enables the CO electrolysis in alkaline MEA to attain 80% selectivity and 30% EE for C₂₊ products at an industrial current density of 900 mA cm⁻². The strategy of using Lewis acid–confined electrocatalysts to effectively maintain high local alkalinity at both electrodes not only represents an effective material strategy for enhancing the performance of CO electrolysis but also provides universal scientific guidance for proton-coupled electron transfer processes such as water electrolysis, electrochemical ammonia synthesis, carbon-based small molecule electrolysis, and fuel cells.

MATERIALS AND METHODS

Chemicals

All chemicals were used without further purification. Milli-Q ultrapure water (>18 megohm cm) was used in all experiments. Tetraethyl orthosilicate (TEOS; 28.4 wt %), isopropanol (99.7 wt %), NH₃·H₂O (25 to 28 wt %), NH₄Cl (99.5 wt %), Cu(CH₃COO)₂·H₂O (99.5 wt %), and Co(CH₃COO)₂·4H₂O (98 wt %) were purchased from Sinopharm Chemical Reagent. Cu nanoparticles (10 to 30 nm and 99.9 wt %) and Co nanoparticles (30 nm and 99.9 wt %) were purchased from Macklin. IrO₂ (85 wt % Ir content) was purchased from Shaanxi Kaida Chemical Engineering. Anhydrous ethanol and dimethyl sulfoxide (DMSO; 99.9%) were obtained from Shanghai Hushi Experimental Equipment. Deuteroxide (D₂O; 99.99%) was supplied by Damas-beta Reagent.

Material synthesis

SiO₂ spheres were prepared by the following method. A total of 0.6 ml of TEOS, 64 ml of isopropanol, 23.5 ml of deionized (DI) water, and 13 ml NH₃·H₂O was injected into a beaker with pipette and then magnetically stirred at 35°C for 1 hour. A total of 5 ml of TEOS was added dropwise into the above solution. After 2 hours of stirring, the product was collected by centrifugation and washed with DI water until the pH of solution is 7. After drying at 60°C for 24 hours, SiO₂ spheres were obtained.

Cu silicate and Co silicate were prepared by a hydrothermal method using the resulted SiO₂ spheres as the template. For Cu silicate, 120 mg of SiO₂ was first dispersed in 60 ml of DI water through ultrasonication, and then 12 mmol of NH₄Cl, 1.2 mmol of Cu(CH₃COO)₂, and 1.2 ml of NH₃·H₂O were added to the above mixture under stirring. The mixture was transferred into a Teflon-lined autoclave and maintained at 140°C for 10 hours. After the autoclave was cooled down to room temperature naturally, the product was collected by centrifugation and washed with DI water and ethanol. After drying at 60°C for 24 hours, Cu silicate was obtained. For Co silicate, 60 mg of SiO₂, 24 mmol of NH₄Cl, 0.6 mmol of Co(CH₃COO)₂, and 0.4 ml of NH₃·H₂O were used. The hydrothermal process was maintained at 100°C for 20 hours.

Cu/SiO₂ and Co/SiO₂ were prepared by thermal reduction of the resulted Cu silicate and Co silicate in 5% H₂ and 95% Ar atmosphere at 800°C for 6 hours.

Material characterizations

The XRD was carried out using a Rigaku MiniFlex 600 x-ray diffractometer with a Cu Kα source at a wavelength of 0.154 nm. The SEM images were obtained by field emission SEM (Zeiss, GeminiSEM 500). The TEM images and elemental mappings were obtained by TEM (JEOL, JEM-F200) equipped with an energy-dispersive x-ray spectroscopy (JED-2300T). XPS measurements were performed using a Thermo Fisher Scientific ESCALAB Xi+ spectrometer system with monochromatic Al Kα radiation. All of the binding energies were calibrated by the C 1s peak at 284.8 eV. The x-ray absorption spectroscopy spectra of Cu K-edge were measured at the TLS-07A beamline of National Synchrotron Radiation Research Center (NSRRC). Cu foil, Cu₂O, and CuO were used as reference materials. NH₃ TPD was conducted by an iChem 700 instrument equipped with an online iMS 770 mass spectrometers (NH₃ TPD-MS) using 1% NH₃ and 99% He. The metal content of the samples was determined by ICP-AES (Agilent, 5110).

Electrode preparation

To prepare the electrode for cathodic CORR, 5 mg of Cu/SiO₂ and 2 mg of carbon black (VXC-72R) were dispersed in 1 ml of mixed solution containing ethanol and ionomer solution. The ink was drop coated on the gas diffusion layer (YLS-30T) to fabricate the gas diffusion electrode (GDE) with catalyst loading of 2.5 mg cm⁻². Nafion 117 solution (~5 wt % in a mixture of lower aliphatic alcohols and water; Sigma-Aldrich) and the QAPPT solution (5 wt %; Alkymer) were used for CEI and AEI, respectively. The ionomer content in the catalyst layer was 10 wt %. For Cu GDE, to get the same Cu metal loading, 2 mg of Cu was used to prepare the ink, and the catalyst loading was 1.0 mg cm⁻². The ionomer content in the catalyst layer was 5, 10, or 20 wt %.

To prepare the electrode for anodic OER, 4 mg of Co/SiO₂ or 4 mg of IrO₂ was dispersed in 1 ml of mixed solution containing ethanol and ionomer solution. The ink was drop coated on the Ni foam (for alkaline solution) or Ti felt (for neutral solution) to get the catalyst loading of 2.0 mg cm⁻². Nafion 117 solution and QAPPT solution were used for CEI and AEI, respectively. The ionomer content in the catalyst layer was 10 wt %. For Co electrode, to get the same Co metal loading, 1.8 mg of Co was used to prepare the ink, and the catalyst loading was 0.9 mg cm⁻².

Pb UPD-stripping test

Pb UPD-stripping test was conducted in batch cell with Ag/AgCl reference electrode and graphite rod counter electrode at 25°C. Pb UPD was performed at −0.4 V versus reversible hydrogen electrode (RHE) for 10 min in a mixed solution of 0.15 M HClO₄ and 0.002 M Pb(ClO₄)₂, and then Pb stripping curve was recorded from −0.4 to 0.2 V versus RHE at 10 mV s⁻¹. Background curve was recorded without adding Pb(ClO₄)₂ to the solution. The ECSA was determined on the basis of the monolayer of Pb coverage on Cu and 2e⁻ Pb oxidation with a conversion factor of 310 μC cm⁻² (4).

OH⁻ adsorption test

OH⁻ adsorption test was conducted in flow cell with Hg/HgO reference electrode and IrO₂/Ni foam counter electrode at 25°C. LSV curve was recorded in 1 M KOH catholyte from open circuit potential (OCP) to 0.55 V versus RHE at 5 mV s⁻¹. For in situ OH⁻ adsorption test, LSV curve was recorded in 1 M KCl catholyte from OCP to 0.85 V versus RHE at 5 mV s⁻¹, using Ag/AgCl as reference electrode.

K⁺ adsorption test

K⁺ adsorption test was conducted in batch cell with Hg/HgO reference electrode and graphite rod counter electrode at 25°C. K⁺ adsorption was performed at −1.2 V versus RHE for 60 s in 1 M KOH, and then the electrode was transferred into pure water with potential applied. Adsorbed K⁺ was released into pure water after removing the applied potential. This process was repeated three times before the K⁺ content was determined by ICP-AES.

CO adsorption–stripping test

CO adsorption–stripping test was conducted in batch cell with Ag/AgCl reference electrode and graphite rod counter electrode at 25°C. Background curve for Cu oxidation was recorded in 0.1 M HClO₄ under N₂ atmosphere from 0 to 1.5 V versus RHE at 50 mV s⁻¹. CO adsorption was performed at 0.1 V versus RHE for 10 min under CO atmosphere. The excess CO gas was removed by purging N₂ gas for 20 min. CO stripping curve was then recorded from 0 to 1.5 V versus RHE at 50 mV s⁻¹.

In situ Raman test

In situ Raman spectroscopy was performed using a custom-made electrochemical operando cell with a Renishaw Qontor spectrometer at room temperature. A 633-nm laser beam and a ×50 objective lens were used for spectral acquisition. A gold electrode loaded with catalyst was used as working electrode. Hg/HgO and Pt wire were used as reference and counter electrodes, respectively. CO- or O₂-saturated 1 M KOH solution was used as electrolyte. Raman spectra were collected at different applied potentials, with an exposure time of 30 s per measurement.

CORR in flow cell

CORR in flow cell was conducted in a three-electrode configuration using the fabricated GDE as working electrode, Hg/HgO (for alkaline KOH solution) or Hg/Hg₂SO₄ (for neutral K₂SO₄ solution) as reference electrode, and the fabricated catalyst-loaded Ni foam or Ti felt as counter electrode at 25°C. The active area of the working electrode was 2 cm². The cathode and anode compartments were separated by AEM (FAB-PK-130). A total of 95% CO and 5% N₂ (as internal standard) gas was flowed into the cathode compartment at 10 ml min⁻¹ by a gas flow meter. The catholyte and anolyte consisted of 1 M KOH or 0.5 M K₂SO₄ and were circulated at a flow rate of 5 ml min⁻¹ using a peristaltic pump.

CORR in MEA

CORR in MEA was conducted using the fabricated GDE as cathode and catalyst-loaded Ni foam or Ti felt as anode at 25°C. The active area of both electrodes was 1 cm². Cathode and anode were separated by AEM (QAPPT, 25-μm thickness; Alkymer). The thickness ratio between the cathode/anode electrodes and the gasket was set to about 1:0.8. A total of 95% CO and 5% N₂ (as internal standard) gas was supplied to cathode at a flow rate of 5 ml min⁻¹. A total of 1 M KOH or 0.5 M K₂SO₄ was fed to anode at a flow rate of 5 ml min⁻¹. EIS was performed using a Solartron analytical 1470E system over a frequency range of 100 kHz to 0.1 Hz with 5-mV amplitude.

Determination of products

The gaseous products were analyzed by on-line gas chromatography (Shimadzu, GC-2014) equipped with thermal conductivity detector and flame ionization detector. A total of 5% N₂ was used as the internal standard to quantify the gaseous products. The liquid products were quantified by ¹H nuclear magnetic resonance on Bruker AVANCE III HD (400 MHz). DMSO was used as the internal standard. To prepare the sample solution, 10 μl of 140 mM DMSO solution and 57 μl of D₂O were added to 500 μl of the collected anolyte solution. The mixed solution was then analyzed using a water suppression method.

The FE (%) of the product was calculated as

$$\text{FE}=Q_{\text{product}}/Q_{\text{total}}\times 100\%=\left(\mathit{nNF}\right)/\left(\mathit{It}\right)\times 100\%$$

where $Q_{\text{product}}$ (in coulombs) and $Q_{\text{total}}$ (in coulombs) represented the charge consumption for the target product and the total reaction, n was the number of electrons transferred for product formation, N (in moles) was the mole of the target product, F was the Faradaic constant (96485 C mol⁻¹), I (in amperes) was the reaction current, and t (in seconds) was the reaction time.

The EE (%) of the product was calculated as

$$\text{EE}=\text{FE}\times V_{\text{theoretical}}/V_{\text{applied}}\times 100\%$$

where $V_{\text{theoretical}}$ was the theoretical cell voltage for product formation according to thermodynamic data, and $V_{\text{applied}}$ was the applied cell voltage.

The TOF (per hour) of the product was calculated as

$$\text{TOF}=(N/t)/(m_{\text{Cu}}/M_{\text{Cu}})\times 3600$$

where N (in moles) was the mole of the target product, t (in seconds) was the reaction time, m_Cu (in grams) was the Cu loading in the electrode, and M_Cu was the molar mass of Cu (63.55 g mol⁻¹).

The SPCE (%) of the product was calculated as

$$\text{SPCE}={\text{CO}}_{\text{converted}}/{\text{CO}}_{\text{input}}\times 100\%$$

where ${\text{CO}}_{\text{converted}}$ (in moles) was the mole of the converted CO according to the mole of the target product and ${\text{CO}}_{\text{input}}$ (in moles) was the mole of the input CO according to the gas flow rate (5 ml/min and 95% CO).

OER activity measurements

OER activity measurements were conducted in a batch cell with O₂-saturated 1 M KOH at 25°C. The fabricated catalyst-loaded Ni foam was used as working electrode. A homemade RHE and a graphite rod were used as reference and counter electrodes, respectively. LSV curve was recorded from OCP to 1.8 V versus RHE at 5 mV s⁻¹. EIS was performed over a frequency range of 100 kHz to 0.1 Hz with 5-mV amplitude. The presented LSV curves were corrected by 85% current × resistance (IR) compensation according to the EIS results. MOR activity measurements were conducted in a batch cell at 25°C. A mixed solution containing 1 M KOH and 1 M methanol was used as the electrolyte, which was saturated with N₂ before measurements. The fabricated catalyst-loaded Ni foam was used as working electrode. A homemade RHE and a graphite rod were used as reference and counter electrodes, respectively. LSV curve was recorded from OCP to 1.8 V versus RHE at 5 mV s⁻¹. EIS was performed over a frequency range of 100 kHz to 0.1 Hz with 5-mV amplitude. The presented LSV curves were corrected by 85% IR compensation according to the EIS results.

DFT calculations

DFT calculations were performed using generalized gradient approximation Perdew-Burke-Ernzerhof functional (30) and projector-augmented wave method (31) as implemented in Vienna ab initio simulation package (32). The first-order Methfessel-Paxton electron smearing scheme (sigma = 0.2 eV) was used for modeling the face-centered cubic metallic copper (Cu) and cobalt (Co). Grimme’s dispersion (D3) correction (33) was included for all calculations. Plane-wave kinetic energy cutoff of 400 eV was applied for the spin polarization calculations with the energy and force convergence criteria of 10⁻⁴ eV and 0.03 eV/Å, respectively. The typical (111) surface was modeled by a four-layered p(4 × 4) slab containing 64 atoms, in which the top two layers were allowed to relax and the bottom two layers were kept fixed. A thick vacuum of ~15 Å was added to the vertical direction for screening the lateral interaction. A 3 × 3 × 1 Monkhorst-Pack k-point was applied for the slab modeling. The atomic cluster of Si₆O₁₃ was built on the Cu (111) surface referring to the medium coverage of SiO_x/Cu. The initial configuration was equilibrated for 20 ps (stepsize = 1 fs) by on-the-fly MLFF AIMD simulation at 1000 K (fig. S9), which was fully relaxed by standard atomic relaxation. The *OH adsorption, C-C coupling of 2*CO, and OER were modeling based on the clean and SiO_x-supported Cu and Co surfaces, respectively. The transition states were located by combing climbing-image nudge elastic band and quasi-Newton minimization methods. Frequency calculations were based on the second-order derivatives of high-accuracy (10⁻⁶ eV) total energy with respect to ionic positions (POTIM = 0.02 Å). Gibbs free energy corrections were obtained by VASPKIT program (34).

Supplementary Materials

This PDF file includes:

Figs. S1 to S29

Table S1