Silver nanopowder (APS 20–40 nm, 99.9%, Thermo Scientific Chemicals) was employed as the cathodic catalyst, while iridium oxide nanopowder (Fuel Cell Store) served as the anodic catalyst. For the gas diffusion media, Sigracet 39 BB carbon paper and platinum-coated titanium fiber felt (Fuel Cell Store) were used as the cathode gas diffusion layers (GDL) and anode GDL, respectively. For consistency, the active area for GDL was standardized to 12 cm². To prepare the cathode catalyst ink, 30 mg of silver nanopowder was dispersed in 2 mL of isopropanol along with 40 µL of 5 wt% Nafion solution. The mixture was ultrasonicated for 20 min to ensure homogeneity. The resulting ink was then manually sprayed onto the microporous layer (MPL) side of the cathode GDL, forming the cathode catalyst layer (cCL) with a target loading of approximately 1.5 mg/cm². Then spraying process was repeated two or three times sequentially to obtain 2- and 3-layer silver electrodes. Twenty independently prepared electrodes were produced for each nominal thickness (1-, 2-, and 3-layer cCLs) under identical spraying conditions. Gravimetric analysis showed silver loadings with relative standard deviations below 5% across batches, and representative electrodes were selected for subsequent testing. The anode catalyst layer was prepared similarly by spraying IrO₂ nanopowder uniformly onto the GDL for a homogeneous coating. The Nafion 212 membrane was thoroughly hydrated in deionized water and then placed between the cathode and anode to complete the MEA arrangement, which is compressed by approximately 27% of its thickness in the customized zero-gap electrolyzer. Scanning electron microscope (SEM) photos of the electrodes were obtained through a JEOL JXA 8530F microscope.

A custom modified electrolyzer is adopted in this work, specifically titanium flow field plates shaped with spiral channels of 0.75 mm width, 0.65 mm depth and 1.3 mm pitch. The gap thickness reserved for the MEA is fixed at 550 µm. Detailed descriptions and photographs of the electrolyzer design and experimental setup have been provided in our previous publications (Zhong et al., 2025a; Zhong et al., 2025b). The electrochemical reduction of CO₂ was evaluated with a Biologic VSP-300 potentiostat coupled with a 10 A current booster. In the zero-gap membrane electrode assembly configuration, the cell voltage was recorded as the potential difference between the cathode and anode without iR correction. Electrolyte solutions were prepared by dissolving solid KHCO₃ in deionized water to concentrations ranging from 0.1 to 2 M, and circulated as the anolyte at a constant flow rate of 30 mL/min by a high-performance liquid chromatography (HPLC) pump. A mass flow controller (Brooks SLA5800 series) was used to adjust the CO₂ flow rate (in sccm), after which the gas passed through a water-filled humidification vessel. A float-type flow meter was installed at the electrolyzer outlet to monitor the products gas flow rate. Chronopotentiometry experiments were performed at a fixed current of 1.2 A, with each test lasting at least 45 min and using fresh electrodes and membranes. Linear sweep voltammetry (LSV) experiments were performed in the range of −0.5 V to −4 V relative to the counter electrode at a scan rate of 20 mV/s.

The composition of gaseous product stream was examined using a gas chromatograph (GC, SHIMADZU 2010 plus) fitted with an HP-PLOT Q column, employing high-purity argon (99.9999%) as carrier gas. Prior to analysis, both the flame ionization detector (FID) and thermal conductivity detector (TCD) were properly calibrated. The products samples were injected and analyzed at least three times every 12 min. For the initial measurement, the cell was preconditioned by maintaining polarization for an extended period of 15 min to allow for electrochemical stabilization and purging of residual gases. Faradaic efficiency for the gaseous products was calculated according to the following formula: F E i = n × m × v × p × F I × R × T × 100 % where n is the number of transferred electrons for product (i), m is the flow of outlet mixture gas, v is the volume fraction of the product (i) measured by GC, p is the pressure, F is the Faradaic constant, while I is the given current, R is the ideal gas constant and T is the ambient temperature. The energy efficiency is given by the equation: E E i = E ′ E × F E i where E ^’ is thermoneutral potential, E is the measured cell voltage of the electrolyzer.

Catalyst thickness is not only related to the available catalytic active sites but also to CO₂ transport, thus affecting the local reaction microenvironment and ion conduction pathways. Insufficient or excessive thickness will lead to limited exposure of active sites or unsatisfactory catalyst utilization. Therefore, cathodes with three different catalyst layer thicknesses are prepared for this CEM-MEA system. The commercially available silver nanoparticles used in this study have a size of 50–100 nm according to the manufacturer. The SEM cross-sections in Figure 1a reveal catalyst layer thicknesses of approximately 3.2, 6.3, and 9.5 µm for the 1-, 2-, and 3-layer silver cathodes, respectively, with deviations below 10%. These results confirm that the thickness increases proportionally with the number of spray cycles. It demonstrates that the spraying method yields catalyst layers with consistent thickness and electrochemical behavior, ensuring that comparisons among different cCL thicknesses are not affected by fabrication variability.

At a fixed current density of 100 mA/cm², the measured cell voltages are 4.10 V for the 1-layer sample, 4.06 V for 2-layer, and 4.14 V for 3-layer (Figure 1b). This negligible voltage variation suggests that the increase in catalyst layer thickness within this range does not significantly affect charge transport resistance or mass transport. The CO production rate is fairly stable, averaging between 8% and 9% from three consecutive measurements within 50 min (Figure 1c). Figure 1d presents the Faradaic efficiencies for each cathode sample over three consecutive 15-min intervals. The hydrogen FE exhibits a slight increase over time, indicating enhanced cathode wettability. In contrast, FE_CO remains stable over the same period, and increasing from 90.5% for the 1-layer sample to 95.4% for the 3-layer sample, suggesting improved accessibility of active sites at higher catalyst loadings. Similar operating voltage and high FE_CO keep the energy efficiency between 32.35% and 33.79% (Figure 1e). These observations align with prior findings in zero-gap configurations where catalyst layers of a few micrometers achieve efficient utilization without inducing significant diffusion barriers (Choi et al., 2022a; Blake et al., 2024). Weber et al. suggests that cCL thicknesses in the range of 2.5–25 µm have minimal impact on overall reaction efficiency (Weng et al., 2020), while experimental results indicate that FE_CO exceeds 90% only when the catalyst loading is significantly reduced to 0.01 mg/cm² (Romiluyi et al., 2022). Nevertheless, a clear quantitative relationship between catalyst loading and CL thickness remains unestablished. Dinh et al. proved that a reasonable catalyst thickness can boost FE_CO, which is attributed to the high availability of active sites and improved CO₂ adsorption (Dinh et al., 2018). Thicker catalyst layers may eventually introduce larger ohmic resistance or CO₂ diffusion gradient (Qi et al., 2020; Choi et al., 2022a) that leads to a slightly higher voltage observed in 3-layer case here.

Stable high FE_CO (≥90%) and EE_CO (≥32%) across thickness variations indicate that CO₂ transport and ion migration are sufficiently effective in this modified CEM-MEA system. From a techno-economic perspective, thinner catalyst layers, such as the 1-layer sample (3.2 µm thickness), can offer superior CO output per unit silver mass, improving cost efficiency and metal utilization. These results underscore the importance of optimizing catalyst thickness not only for electrochemical performance but also for cost-effectiveness and scalability in practical MEA-based CO₂ electrolyzers.

In dynamic electrochemical systems, particularly those operating under industrially relevant current densities, the ability to maintain stable performance across a wide range of CO₂ flow rates is desirable. Several studies have evaluated the performance of CO₂ electrolyzers within a relatively narrow range between 1 and 50 sccm (Ma et al., 2020; Corral et al., 2021; Sun et al., 2024). For example, Li et al. demonstrated that with a flow rate as low as 1 sccm, single-pass conversion could reach 90% in an acid-fed MEA system while Sinton et al. reported over 85% conversion at same low flow rate in alkaline conditions (O’Brien et al., 2021; Pan et al., 2022). However, while low flow rates are beneficial for CO₂ utilization, this often limits the absolute productivity and is therefore not practical for scaled-up devices with large electrode areas (Cunha and Resasco, 2023). Therefore, extending the operating window to higher flow rates is crucial to bridge laboratory performance and industrial applications.

As the CO₂ supply increases from 50 to 90 sccm, the LSV curve shifts to higher potentials, as shown in Figure 2a. Some works acknowledged that the increase in flow rate leads to microscopic changes in the local reaction environment and mass transfer path, which requires a higher electrode overpotential to maintain a fixed reaction rate, thereby increasing the voltage (Lu et al., 2017; Verma et al., 2017; Fan et al., 2024). Despite the rise in cell voltage, product selectivity and efficiency remain robust across the tested range. As shown in Figure 2b, FE_CO remains high at all tested flow rates despite a decrease in the CO fraction in the outlet stream, whereas the hydrogen Faradaic efficiency follows a non-monotonic trend, first declining and then recovering. To probe the interplay between convective transport, diffusion, and surface kinetics, key dimensionless parameters are evaluated and shown in Figure 2c. As flow increases from 50 to 90 sccm, Reynolds number rises but remains <2000, confirming laminar flow inside microchannels. Concurrently, the Sherwood number and mass transfer coefficient increase, enhancing the CO₂ flux available to the GDL. Assuming CO₂ flux through the porous GDL scales linearly with channel velocity, the Péclet number rises accordingly, reflecting increased convective transport and elevated CO₂ concentration at the GDL–catalyst interface. At 50 sccm, mass transport limits surface CO₂ availability, yielding a FE_CO of 88.5%. At 70 sccm, transport is no longer rate-limiting, producing a maximum FE_CO of 95.3%. Further increasing flow to 90 sccm slightly reduces FE_CO to 92.3%, as secondary factors such as local pH gradients, ion migration, and interfacial coverage variations increasingly modulate product distribution. These results reveal an optimal flow regime in which convective transport, diffusive flux, and surface reaction kinetics are synergistically balanced, maximizing CO selectivity and underscoring the critical role of mass transport–reaction coupling in electrochemical CO₂ reduction. Eventually, energy efficiency remains relatively stable across the relatively high flow rate range (60–90 sccm), fluctuating only between 33.3% and 34.7%, as shown in Figure 2d. This demonstrates that while the electrical energy input increases slightly with higher flow rates, the overall system performance remains consistent, supporting the feasibility of broader operational flexibility in practical electrolysis setups. The optimal CO₂ flow rate could be determined in accordance with system-specific parameters such as reactor geometry and electrode surface area (Ye et al., 2022; Subramanian et al., 2023). Proper adjustment ensures sufficient CO₂ supply while avoiding excessive gas dilution. Ideally, achieving the target production rate with a lower fraction of unreacted CO₂ in the outlet stream can reduce the need for gas recirculation and ease the burden of downstream gas purification.

Under otherwise identical conditions, the composition of anolyte in the anode chamber directly influences electrolyte resistance and can also modulate the intrinsic kinetics of the oxygen evolution reaction (OER) at the anode (Larrea et al., 2022; Li et al., 2023; Xu et al., 2025). It also modifies key interfacial phenomena through ion gradients and transport kinetics (Wheeler et al., 2020). The LSV curves (Figure 3a) reveal that as the KHCO₃ concentration increases from 0.1 M to 2.0 M, the current density at the equivalent voltage increases significantly. Sun et al. reported that higher anolyte concentrations enhance ionic conductivity, thereby reducing ohmic losses across the membrane and at the electrode-membrane interface (Xiao et al., 2022). As shown in Figure 3b, the cell voltage drops from over 4.45 V at 0.1 M to approximately 3.7 V at 2.0 M, indicating a substantial decrease in overall cell resistance. Beyond conductivity, increasing anolyte concentration also affects species transport across the membrane. More K⁺ ions migrate through the membrane toward the cathode, where they accumulate at the catalyst-electrolyte interface. Koper and co-workers have shown that such cation migration modifies local electric fields and enhances CO₂ reduction by stabilizing key reaction intermediates, while concurrently suppressing HER through pH buffering and electrostatic effects (Monteiro et al., 2021). Here, this manifests as a steep improvement in FE_CO, rising from 34.4% at 0.1 M to 93.0% at 1.0 M (Figure 3c), accompanied by a sharp drop in hydrogen FE from 64.1% to below 5%. In addition, Zhu et al. observed improved selectivity with increased anolyte K⁺ content, even under acidic conditions, underscoring the role of alkali cations in tailoring the near-surface chemical environment to favor CO₂ reduction (Li et al., 2023; Zhu et al., 2024).

Additionally, the rise in CO energy efficiency from 11.3% at 0.1 M to 33.2% at 1.0 M (Figure 3d) further confirms the synergistic effect of enhanced conductivity and local reaction environment optimization. However, further increasing the KHCO₃ concentration beyond 1.0 M yields marginal or no additional gains in FE_CO or energy efficiency. This saturation effect suggests that at sufficiently high ionic strength, ion transport is no longer rate-limiting. Moreover, elevated K⁺ concentrations can induce bicarbonate or carbonate salt precipitation within the porous gas diffusion layer, obstructing CO₂ transport and degrading long-term performance (Garg et al., 2023). Therefore, optimizing anolyte concentration is essential: it must be high enough to ensure efficient ion transport and favorable interfacial conditions, but not so high as to induce salt accumulation or flooding, which compromise CO₂ access and overall cell stability.

In practice, electrolyzer stacks may generate significant ohmic heat and therefore are commonly operated at 40 °C–70 °C (Pelzer et al., 2025). Temperature changes can directly affect the thermodynamic feasibility of electrochemical reactions and ionic conductivity while facilitating CO₂ diffusion through the porous electrode (Choi et al., 2025). Recent studies have shown that elevated temperatures can significantly enhance reaction rates and reduce overpotentials of CO₂RR as well as the resistance of Nafion membrane (Jiang et al., 2023; Rodriguez et al., 2024; Sakita and Ticianelli, 2025). For instance, Zhuang et al. developed an alkaline polymer electrolyte (APE)-based CO₂RR system that employed pure water as the anolyte and achieved higher current density and Faradaic efficiency at operating temperatures above 60 °C (Yin et al., 2019). However, reduced CO₂ solubility and inevitable water electrolysis at high temperature tends to shift products selectivity unless the electrode structure or membrane assembly suppresses those effects.

LSV results (Figure 4a) show that the current density increased significantly with temperature at all applied voltages, indicating enhanced reaction kinetics and system conductivity. The cell voltage required to sustain the current density of 0.1 A/cm² drops from 4.1 V at 25 °C to 3.6 V at 50 °C and further to 3.4 V at 75 °C (Figure 4b), which is consistent with reduced ohmic resistance and improved ion conduction at high temperatures (Endrődi et al., 2020; Shafaque et al., 2021). Through modeling studies, Weber et al. demonstrated that the reduced overpotential resulted from the exponential enhancement of charge transfer kinetics with increasing temperature, in accordance with the Arrhenius relationship (Weng et al., 2019; Weng et al., 2020). The product distribution shows exceptional stability over the examined temperature range, as FE_CO rises slightly from 93.0% at 25 °C to 94.0% at 75 °C, while the competing HER exhibits only a modest increase from 4.7% to 6.8% within the same range (Figure 4c). Although also benefiting from thermodynamics, HER here is still effectively suppressed even at high temperatures. Segar et al. also observed unexpected inactive HER in an anion exchange membrane MEA system at high temperatures (>50 °C) and believed that fully humidified CO₂ gas and increased carbonate solubility protected the cathode microenvironment for high selectivity of CO₂RR (Rodriguez et al., 2024). Likewise, the zero-gap structure and excellent mass transport guarantee overwhelming CO₂ adsorption and rapid CO desorption on the catalyst layer, and remove water transported from the more hydrated membrane at elevated temperature (Weng et al., 2019). Energy efficiency toward CO production increases from 33.2% at 25 °C to 40.3% at 75 °C (Figure 4d), which is mainly attributed to the significant decrease in operating voltage at higher temperatures due to the reduction in overall resistance, while maintaining high CO selectivity (She et al., 2024).

Moderately elevated temperatures in the range of 50 °C–75 °C can significantly enhance CO₂ reduction performance in MEA systems by promoting reaction kinetics, ion diffusion and conductivity, and water transport. Nevertheless, operation at even higher temperatures and high current introduces challenges such as membrane dehydration and catalyst degradation, underscoring the importance of integrated thermal management strategies (Pelzer et al., 2025). Moving forward, the design of CO₂ electrolysis systems should aim to optimize both thermal and electrochemical performance, with the potential to incorporate waste heat recovery from upstream or downstream industrial processes to further improve system efficiency.

Applying pressure will enhance CO₂ solubility in aqueous electrolytes, substantially increasing reactant availability in liquid-fed cathode systems and thereby improving reaction rates and selectivity (Lamaison et al., 2020; Huang et al., 2023; Sun et al., 2025). In MEA systems, pressure acts differently by directly elevating the partial pressure of CO₂ at the catalyst layer surface; despite this potential advantage, pressure effect on MEA-based CO₂ reduction is understudied (Chen B. et al., 2024). Our study employs a zero-gap electrolyzer to assess how modest CO₂ feed pressures (1–5 bar) influence the CO₂RR at controlled conditions.

The LSV curves (Figure 5a) show that increasing gas feed pressure generally increases the reaction rate relative to ambient pressure operation, but no other significant differences are observed between 1 and 5 bar. This suggests that while slight pressurization increases CO₂ availability at the catalyst interface, the overall reaction rate is still limited by other factors, such as intrinsic catalyst activity or membrane status under pressure. Despite this benefit, Figure 5b indicates that the production of CO in the outlet gas stream remained ∼8.5 mL/min across the pressure range studied. Figure 5c demonstrates that FE_CO rises to 94.7%–95.4% at pressurized conditions while the FE for H₂ concurrently restrains to 3.1%–5.4%. The improved selectivity under pressurized conditions supports the hypothesis that enhanced CO₂ transport through the GDE to the catalyst surface and increased surface coverage favors the CO₂ reduction pathway (Hussain et al., 2025). It is noteworthy that although equal nominal pressures are applied to both the gas-phase cathode and liquid-phase anode compartments in this zero-gap system via a back-pressure regulator, the measured pressure on the anode side is approximately 0.2 bar higher than that on the cathode side. This pressure differential may induce liquid electrolyte migration across the membrane, potentially contributing to the occurrence of HER (Chen Y. et al., 2024). The energy efficiency of the system also benefits from pressurization, as shown in Figure 5d, where energy efficiency increased to approximately 36.5%. This enhancement results from the combined effects of increased FE and modest reductions in cell voltage losses due to improved mass transport and gas management under higher pressure conditions.

This section demonstrates that slightly increased pressure can enhance CO₂ mass transfer and product selectivity, thereby improving the overall efficiency and feasibility of zero-gap CEM-based electrolyzers. Higher pressure operation presents engineering challenges, such as increased mechanical stress on cell components, stringent sealing requirements, precise control and monitoring equipment, while pressure-dependent gas-liquid transfer dynamics may lead to changes in product distribution. Despite this, there is still much room for development in high-pressure electrocatalysis, especially when considering its integration with other industrial chemical production and processes.

In summary, the experimental parameters, corresponding variables, and their electrochemical performance are compiled in Table 1 to facilitate a direct and quantitative comparison. High energy efficiency can only be achieved by simultaneously maximizing product Faradaic efficiency and minimizing cell voltage; therefore, studies aiming to improve energy efficiency should primarily focus on optimizing both factors.

High CO selectivity is maintained over a wide range of operating conditions; however, an unexpected pressure surge occurs after 60 min at ambient temperature and pressure due to white precipitate formation on the GDE backside (Figure 6), a phenomenon also observed in other MEA systems (Hernandez-Aldave and Andreoli, 2020; Disch et al., 2023; Park et al., 2023). This indicates that, despite employing a proton exchange membrane in a catholyte-free configuration, the cathode microenvironment alkalinizes as the reaction proceeds. High current densities enhance water uptake at the cathode GDE–CEM interface (Shafaque et al., 2020), promoting reactions between CO₂ and water (CO₂ + H₂O+ 2e⁻→ CO + 2OH⁻, 2H₂O+ 2e⁻→ H₂ + 2OH⁻) transported from the membrane (Wheeler et al., 2020; Choi et al., 2022b). Strong CO₂RR activity accelerates hydroxide generation, elevating local alkalinity and driving reactions with K⁺ migrating from the anode, leading to potassium (bi)carbonate (CO₂ + OH⁻ → HCO₃ ⁻; HCO₃ ⁻ + OH⁻ → CO₃ ²⁻ + H₂O) precipitation at the cathode (Garg et al., 2023; Bai et al., 2024). In a parallel study with the same zero-gap configuration and conditions but employing an anion exchange membrane, stable operation for over 100 h at 50 °C was achieved (Zhong et al., 2025a), underscoring the critical role of the cathode microenvironment in determining long-term stability. Although CEM primarily conducts protons, CO₂RR-driven hydroxide buildup still renders the local environment alkaline, posing a fundamental constraint on extended operation. Despite this limitation, the early-stage performance provides a robust basis for elucidating the interplay between key operational parameters and catalytic behavior, which is the central focus of this study.

This observation also highlights a dual nature of zero-gap electrolyzers. On one hand high selectivity is accessible due to efficient CO₂ reduction under compact architecture. On the other hand, long-term stability is compromised by salt precipitation, which blocks gas diffusion pathways and degrades catalytic interfaces. Some studies addressed and investigated solutions to the salt precipitation problem. Burdyny et al. concluded several effective methods for mitigating salt precipitation, such as flushing, cation species and concentration design, and voltage pulses (Endrodi et al., 2021; Sassenburg et al., 2023). Recently, Wang et al. innovatively employed superhydrophobic cathode gas flow channels and acid-humidified CO₂ gas, effectively extending operating time (Hao et al., 2025a; Hao et al., 2025b). These methods provide valuable guidance for future efforts to improve the durability of zero-gap CO₂ electrolysis systems.