Sponge-inspired catalyst design for durable acidic CO₂ reduction at low K⁺ concentration

Abstract

Acidic electrochemical CO₂ reduction (CO₂RR) typically requires K⁺ ions to create a local H⁺-depleted microenvironment, suppressing competing hydrogen evolution reaction (HER). Excessive localized K⁺ causes salt precipitation, compromising electrolysis stability. Achieving stable operation with high Faradaic efficiency (FE) at low K⁺ concentrations remains a crucial challenge for conventional nanomaterials. Inspired by water-trapping function of sponges, we design a three-dimensional interconnected porous cubic SnO₂ electrocatalyst (SnO₂ sponge) that confines OH^– within porous channels to consume proton influx from the bulk, enabling durable acidic CO₂RR towards formic acid (HCOOH). Theoretical and experimental studies reveal the SnO₂ sponge sustains substantially higher OH^– concentration than dispersed SnO₂ nanoparticles. At pH 1.82, the SnO₂ sponge achieves 94.5% FE_HCOOH at 800 mA cm^–2. With only 0.075 M K⁺, it retains 95.2% FE_HCOOH at 400 mA cm^–2. Notably, it enables continuous HCOOH production at 400 mA cm^–2 with 97.7% FE_HCOOH for over 390 h without cleaning. This work provides a promising strategy for durable and efficient CO₂RR in acidic media with low K⁺ concentrations.

Acidic CO₂ reduction typically requires high K⁺ to suppress hydrogen evolution, which can cause salt precipitation and system instability. Here, the authors report a porous SnO₂ sponge that traps OH⁻ ions, enabling stable and highly efficient formic acid production at low K⁺ concentrations.

Introduction

The electrochemical CO₂ reduction reaction (CO₂RR) offers a promising route for converting CO₂ into valuable chemicals and thus contributing to carbon neutrality.^1–4 In recent years, acidic CO₂RR has attracted considerable attention as a strategy to overcome the low CO₂ utilization typically encountered under CO₂RR in alkaline or neutral conditions⁵. Nevertheless, a major challenge in acidic CO₂RR is suppressing the competing hydrogen evolution reaction (HER), which proceeds much faster than CO₂RR in strongly acidic media. Effective suppression of HER requires precise regulation of proton concentration within the catalyst’s local microenvironment.² A widely adopted strategy involves introducing high concentrations of alkali metal ions such as Na⁺, K⁺, and Cs⁺ to form a hydrated cation layer at the outer Helmholtz plane (OHP).⁶ This confined cation layer establishes a local electric field that repels excess H⁺ and elevates the local pH at the catalyst surface, thereby enabling acidic CO₂RR with high Faradaic efficiency (FE).^4,6,7 Whereas, the accompanying high local pH and excessive K⁺ can induce salt accumulation, as observed in neutral and alkaline systems, leading to gas diffusion layer (GDL) blockage^3,8,9 and limiting operation duration to only tens of hours^3,6,10,11. Therefore, reducing the reliance on K⁺ while maintaining a high local pH remains a central challenge to achieve durable acidic CO₂RR.

Various strategies have been explored to minimize reliance on alkali metal ions. The most widely adopted approach involves introducing positively charged functional groups onto the catalyst to mimic the hydrated K⁺ layer^9,12–14. Another strategy is the design of a heterogeneous adlayer, which effectively blocks proton transport from the bulk electrolyte.¹⁵ Tip effect has also been employed to enrich K⁺ at the OHP, thereby reducing the required bulk K⁺ concentration.⁸ These approaches improve CO₂ reduction activity and operational stability under low-K⁺ or K⁺-free conditions, but they inevitably increase electrode resistance, which in turn requires high applied potentials and results in low cathodic energy conversion efficiency (ECE) at industrial current densities.

Through constructing reservoir structures such as nanosheets and nanocavities^16–19, it is possible to design a structure with seepage behavior that slows the exchange of electrolyte between the catalyst and the surrounding bulk environment.²⁰ This can restrict proton flux and create locally alkaline environments through OH^– confinement in the pores during reaction without the need for additives, significantly reducing the required bulk K⁺ concentrations. This behavior is similar to a sponge after absorbing water. Inspired by the water-trapping capability of sponges, catalysts with a three-dimensional (3D) interconnected porous network (sponge-like) structure not only provide a large specific surface area and abundant active sites, but also restrict electrolyte flow, facilitating the formation of a high local pH environment, as proton influx from the bulk is much slower than local proton consumption (Fig. 1a).²¹ In this work, we carried out theoretical and experimental practices on this catalyst design concept. Simulations indicate that the peak OH^– concentration surrounding the sponge model system is orders of magnitude higher than that surrounding dispersed nanoparticles and solid cube model systems. Accordingly, a 3D interconnected porous cubic SnO₂ (denoted as SnO₂ sponge) was synthesized by hydrothermal-assisted acid-etching of CaSnO₃. The SnO₂ sponge features a high specific surface area of 116.6 m² g^–1 and rich active sites, maintaining a robust alkaline microenvironment during CO₂RR. In pH 1.82 electrolytes, it achieves a FE_HCOOH of 94.5% at 800 mA cm^–2 and requires only 0.075 M K⁺ to obtain high FE_HCOOH ( > 90%) at 400 mA cm^–2, with a peak single-pass CO₂ conversion efficiency (SPCE) of 80.2%. Owing to the stable alkaline microenvironment enabled by the SnO₂ sponge, continuous electrolysis at 400 mA cm^–2 was extended from 20 h with SnO₂ nanoparticles (FE_HCOOH declining from 76.7% to 33.9%) to over 390 h without the need for cleaning operations, while maintaining a high FE_HCOOH of 97.7%.

Fig. 1. Simulations.

a Schematic diagram of the SnO₂ sponge applied in CO₂RR in an acidic electrolyte. b The simulated local pH distribution diagrams for the porous cube model. c Simulated local pH variation on the porous cube and dispersed nanoparticles catalysts in the diffusion layer with 1.0 M bulk K⁺. d Simulated peak pH for different catalysts under various K⁺ concentrations. All simulations were conducted at –1.27 V_RHE in an acidic electrolyte (pH = 1.82). Source data are provided in a Source Data file.

Results

Proof of concept with simulations

We constructed three ideal model systems to mimic a sponge-like catalyst, a dispersed nanoparticles catalyst, and a solid microparticles catalyst. Then, simulations with COMSOL were conducted to evaluate the differences in OH^– concentration among these three models, thereby enabling a comparison of their capacities to generate and sustain localized reaction microenvironments. More specifically, the porous cube was used to model a sponge-structured system, the dispersed particle array represented a non-sponge, loosely packed structure, and the solid cube served as a non-porous structure. As shown in Figs. 1b, c, S1, the porous cube shows multiple high-pH zones, with a two-dimensional (2D) slice exhibiting a pH > 10.6 under 1.0 M bulk K⁺ at –1.27 V_RHE, particularly in the pore regions. This indicates the capability of the porous cube to retain OH^– and buffer the surrounding environment. In contrast, dispersed nanoparticles and a solid cube in a similarly sized region result in only moderate pH elevations, with a peak pH of approximately 6.8 and 7.0, respectively.

The pH variation from the catalyst to the bulk electrolyte within the diffusion layer further underscores this disparity among the sponge, dispersed nanoparticles, and solid microparticles structures. As shown in Figs. 1c, d, S1 and S2, under both 1.0 M K⁺ and 0 M K⁺, noticeable pH differences and pH-maintaining regions are observed for the porous cube, dispersed nanoparticles and solid cube. Figure 1d and S3 reveal that the peak pH is a function of bulk K⁺ concentration. In acidic electrolytes with low added K⁺, the porous cube model achieves a higher peak pH compared with the nanoparticles and solid cube models. The porous cube model exhibits a steep increase in pH with increasing K⁺ concentration, reaching a plateau at 0.3 M K⁺ with a pH of 10.2. Conversely, the nanoparticles and solid cube models exhibit minimal response, achieving only a pH of 6.4 and 6.9, respectively, at 0.3 M K⁺. As shown in Fig. S4, the peak OH^– concentration in the porous cube is orders of magnitude higher than those observed in the nanoparticles and solid cube models under 1.0 M K⁺. Collectively, the geometric confinement and spatial ion management of the porous cube demonstrate a promising capacity to generate and maintain an alkaline microenvironment, suggesting that K⁺ concentration can be reduced when using the sponge-structured porous cube as a catalyst in practical applications.

Preparation and structural characterization

A 3D interconnected porous cubic SnO₂ (SnO₂ sponge) catalyst was synthesized via a hydrothermal-assisted acid-etching approach (Fig. 2a). Bulk CaSnO₃, serving as the precursor, exhibits a typical perovskite structure with a microcube morphology (Fig. S5) and was used to form porous SnO₂ by dissolving Ca²⁺ with hydrochloric acid. To further evaluate the advantages of the sponge-type SnO₂, SnO₂ nanoparticles (denoted as SnO₂ NPs) and SnO₂ solid microparticles (denoted as SnO₂ solid MPs) were prepared as reference samples. The SnO₂ NPs were synthesized using the same protocol as the SnO₂ sponge, but in the absence of Ca²⁺ (Fig. S6). The powder X-ray diffraction (PXRD) pattern (Fig. S7) of the SnO₂ sponge matches well with that of SnO₂, confirming the successful removal of Ca²⁺ and phase purity of the final product.

Fig. 2. Synthesis and structural characterization.

a Schematic illustration of the synthesis process of the SnO₂ sponge. b–d Low-magnification HAADF-STEM image (b), high magnification HAADF-STEM image (c), and HR-TEM image (d) of the SnO₂ sponge. e EDS elemental mapping of Sn, O and Ca in a sponge cube. f Nitrogen adsorption-desorption isotherms of SnO₂ sponge, CaSn(OH)₆ cube, and CaSnO₃ cube. g Pore size distribution, cumulative pore volume, and corresponding porosity of the SnO₂ sponge. Source data are provided in a Source Data file.

Rigorous characterization of SnO₂ sponge using high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM), high-resolution transmission electron microscopy (HR-TEM), and scanning electron microscopy (SEM) images, along with HAADF-STEM tomographic analysis, reveal an average lateral dimension of approximately 0.6 μm (Fig. S5), which is comparable to the macroscopic size of the SnO₂ solid MPs (Fig. S8). Notably, the SnO₂ sponge exhibits a well-defined 3D interconnected architecture composed of numerous nanoparticles (Fig. 2b, c, Video S1), with particle sizes comparable to those of the SnO₂ NPs control sample (Fig. S9). This structural feature indicates that the SnO₂ sponge can be regarded as an interconnected framework assembled from nanoscale building blocks, whereas SnO₂ NPs represent a loosely packed and poorly interconnected assembly of similar primary particles, making them an appropriate comparison system for probing the effect of confining channels. The HR-TEM image (Fig. 2d) reveals lattice spacings of 0.33 nm and 0.18 nm, corresponding to the (110) and (211) facets of SnO₂, respectively, which are consistent with PXRD results (Fig. S7). Scanning transmission electron microscopy energy-dispersive X-ray spectroscopy (STEM-EDS) elemental mapping shows that only Sn and O elements exist in the porous cubic SnO₂ (Figs. 2e, S10, Table S1), indicating complete removal of Ca²⁺. Further evidence of the porous character of the SnO₂ sponge is provided by its significantly high Brunauer–Emmett–Teller (BET) surface area of 116.6 m² g^–1, which is nearly 10 times higher than that of the CaSnO₃ precursor and SnO₂ solid MPs, and also exceeds that of SnO₂ NPs (Figs. 2f, S11). As shown in Fig. 2g, the pore size distribution indicates abundant mesopores with an average pore size of 3.7 nm and an overall porosity of 53.5%.

Electrochemical CO₂ reduction performance

The CO₂RR performance of the SnO₂ sponge was evaluated in a three-electrode flow cell using H₂SO₄ and K₂SO₄ solutions (electrolyte pH 1.82, Fig. S12, Video S2). Liquid and gaseous products were quantified using nuclear magnetic resonance (NMR) spectroscopy and gas chromatography (GC), respectively. As shown in Fig. 3a, the linear sweep voltammetry (LSV) curves provide insight into the electrochemical behavior of the SnO₂ sponge. To evaluate the performance of catalysts, we conducted the tests at 1.0 M K⁺ as usual. As shown in Figs. 3b, S13 and S14, the SnO₂ sponge catalyst exhibited over 90% FE for HCOOH (FE_HCOOH) across the entire range of 0–800 mA cm^–2, reaching a peak FE_HCOOH of 96.4% and a maximum HCOOH production rate of 39.2 mmol m⁻² s⁻¹. As illustrated in Fig. S15, only trace amounts of CO and H₂ were detected. Compared with SnO₂ NPs and SnO₂ solid MPs, the SnO₂ sponge delivered significantly higher FE_HCOOH and j_HCOOH across all tested current densities. Notably, at 800 mA cm^–2, the SnO₂ sponge achieved a FE_HCOOH of 94.5%, substantially surpassing the 70.9% and 68.2% observed for SnO₂ NPs and SnO₂ solid MPs, respectively. The SnO₂ NPs and SnO₂ solid MPs show comparable FE_HCOOH over the range of 0–800 mA cm^–2, indicating that the presence of internal porous channels plays an important role in achieving enhanced CO₂RR performance. CO₂RR using SnO₂ sponge in a strong buffering electrolyte (H₃PO₄ + KH₂PO₄, pH 1.82, 1.0 M K⁺) resulted in a low FE_HCOOH (Fig. S16), further confirming that the moderate alkaline microenvironment is the key factor for performance enhancement. This experiment indicates that the observed performance enhancement arises from the generation of alkaline conditions by porous structure, rather than from any inherent efficiency of CO₂ electroreduction under acidic conditions. To exhibit the OH^‒ confinement effect of SnO₂ sponge, the performance evaluation was also conducted at 0.25 M K⁺. As shown in Fig. S17, the SnO₂ sponge maintained 95.1% FE_HCOOH at 600 mA cm⁻², while SnO₂ NPs exhibited only 49.7%. This enhanced performance highlights the advantage of the sponge structure in promoting CO₂-to-HCOOH conversion in acidic electrolytes, particularly under industrial current densities. In addition, the SnO₂ sponge demonstrated strong pH tolerance (Fig. 3c), maintaining FE_HCOOH above 90% across a wide pH range (1.3–13.6), demonstrating its durability under diverse electrochemical conditions. The decrease in FE at pH 1 (Fig. S18) may be associated with enhanced proton diffusion and possible instability of SnO₂ structure under strongly acidic reductive conditions.

Fig. 3. Electrochemical CO₂ reduction measurements.

a LSV curves of SnO₂ sponge and SnO₂ NPs under CO₂ and Ar atmospheres in a flow cell using an acidic electrolyte. b Current-dependent FE_HCOOH at 1.0 M K⁺ electrolyte for SnO₂ sponge and SnO₂ NPs. c pH-dependent ${\mathrm{FE}}_{{\mathrm{HCOOH}/\mathrm{HCOO}}^{-}}$ and corresponding cathodic ECE at 400 mA cm^–2. d FE_HCOOH of SnO₂ sponge and SnO₂ NPs in electrolytes with various K⁺ concentrations at 400 mA cm^–2. e CO₂ SPCE of SnO₂ sponge under varying CO₂ flow rates at 400 mA cm^–2 at 0.075 M K⁺. f Photographs of the backside of carbon paper after electrolysis for 10 h at 1.0 M K⁺ and 390 h at 0.25 M K⁺. g, h Electrochemical stability of SnO₂ sponge (g) over 390 h and SnO₂ NPs (h) over 20 h at 400 mA cm^–2 in 0.25 M K⁺ electrolyte. Error bars represent the standard deviation of three independent measurements. Source data are provided in a Source Data file.

To investigate the K⁺ tolerance threshold of SnO₂ sponge, CO₂RR experiments were conducted on both SnO₂ sponge and dispersed SnO₂ NPs under various K⁺ concentrations. The pH value was fixed to exclude variations in proton diffusion rates on the catalyst surface. As shown in Fig. 3d, both SnO₂ sponge and SnO₂ NPs maintained over 90% FE_HCOOH at a K⁺ concentration of 0.25 M, indicating that this concentration is sufficient to form a positive charge layer at the OHP, effectively repelling H⁺ transfer. At 0.075 M K⁺, the SnO₂ sponge retained 95.2% FE_HCOOH, whereas the FE_HCOOH of SnO₂ NPs decreased to 69.2%. This lower K⁺ operating threshold suggests that the sponge structure can efficiently store OH^–, while repelling proton transport at lower K⁺ concentrations, thereby inhibiting the kinetically favorable HER. At the optimized K⁺ concentration (0.075 M), a CO₂ SPCE of 80.2% with a FE_HCOOH of 92.1% was achieved by optimizing the CO₂ flow rate (Fig. 3e), emphasizing the high CO₂ utilization capability of the SnO₂ sponge in acidic media.

Due to the strong restriction of OH^– by the SnO₂ sponge, the stable microenvironment enables the catalyst to maintain high efficiency and stability at industrial-level current densities, particularly without requiring cleaning operations. An initial stability test was conducted at 400 mA cm⁻² at a 1.0 M K⁺ electrolyte. The results revealed that even in acidic electrolytes, severe salt precipitation occurred on the backside of the GDL, hindering the contact pathway between CO₂ and the catalyst, thereby limiting stable operation to only 100 h (Figs. 3f, S19). This finding aligns with previous literature reports.^9,12 Periodic salt removal is inevitable and can be achieved through methods such as flushing with water or solvents²², or self-cleaning by switching between operational and regenerative voltages²³. To further enhance the system’s capacity to mitigate salt precipitation, a 0.25 M K⁺ electrolyte was selected for long-term stability testing. Under the low K⁺ concentration, no significant salt precipitation was observed on the backside of the carbon paper, emphasizing the importance of using low K⁺ concentration electrolytes in CO₂RR. As a result, the SnO₂ sponge continuously operated at 400 mA cm⁻² for over 390 h with an average FE_HCOOH of 97.7%, without the need for cleanup operations (Fig. 3g). To maintain the electrolyte pH (Fig. S20) and ensure efficient CO₂ utilization, the electrolyte was refreshed every 12 h, allowing the SPCE to sustain at 29.7%. As shown in Table S2, the results were compared with previously reported works, and the high activity and FE_HCOOH of the SnO₂ sponge render it competitive among the catalyst performances in CO₂RR-to-HCOOH. In contrast, dispersed SnO₂ NPs experienced rapid performance degradation within 20 h (FE_HCOOH decreased from 76.7% to 33.9%) due to fast proton transport from the bulk electrolyte to the catalyst surface, making HER the dominant reaction (Fig. 3h). The SnO₂ solid MPs retain a FE_HCOOH of 62.4% for 100 h (Fig. S21). The behavior is likely associated with a larger particle size and improved resistance to self-reduction. This highlights the importance of the SnO₂ sponge in creating a microenvironment resistant to proton interference. The comprehensive CO₂RR performance of the SnO₂ sponge positions it as a promising candidate catalyst for industrial-level CO₂-to-HCOOH conversion.

The SnO₂ sponge remains stable after continuous CO₂RR operation. To explore the state of the SnO₂ sponge after CO₂RR, post-electrolysis samples were characterized using HR-TEM, HAADF-STEM, STEM-EDS elemental mapping, X-ray photoelectron spectroscopy (XPS), and X-ray absorption spectroscopy (XAS) analyses. After long-term stability tests, the SnO₂ sponge retains its 3D porous cubic structure with preserved lattice features (Figs. 4a, b, S22). As shown in Fig. S23, XPS analysis of SnO₂ sponge after 100 h of electrolysis and SnO₂ NPs after 20 h revealed that the Sn 3d_5/2 peak could be deconvoluted into two peaks at approximately 486.98 and 485.19 eV, while the Sn 3d_3/2 peak was split into two peaks at approximately 495.33 and 493.60 eV, corresponding to Sn⁴⁺ and Sn⁰, respectively.²⁴ The presence of Sn⁰ demonstrates slight self-reduction of the SnO₂ sponge after 100 h of electrolysis, whereas the bulk phase remains stable, which is consistent with XAS observations probing Sn valence variation and atomic coordination. Fig. 4c, d show that the X-ray absorption near-edge structure (XANES) spectra of the Sn K-edge for SnO₂ sponge after electrolysis exhibit no shift compared to pristine SnO₂ sponge. The R-space spectra from extended X-ray absorption fine structure (EXAFS) reveal nearly identical Sn–O distances for both samples. In contrast, the absorption edge of SnO₂ NPs after electrolysis is negatively shifted toward the Sn foil reference. The Sn–O distance signal significantly diminishes after electrolysis, while the characteristic Sn–Sn distance from typical metallic Sn emerges.

Fig. 4. Morphology and structural characterization of SnO₂ sponge and SnO₂ NPs after electrolysis at 400 cm^–2.

a, b TEM (a) and HR-TEM (b) images of SnO₂ sponge after 60 h of electrolysis. c XANES spectra of the Sn K-edge. d k²-weighted EXAFS spectra. The spectra of Sn foil and SnO₂ are shown as references. Source data are provided in a Source Data file.

In CO₂RR, the structural integrity of catalysts is crucial for long-term stability, especially under high current densities^25,26. These results demonstrate good stability of the SnO₂ sponge even after prolonged continuous operation. The SnO₂ sponge, with its 3D interconnected porous framework, stabilizes the local alkaline environment, elevates CO₂ transfer, and enhances the interactions between intermediates and the Sn^δ+ sites. These provide enhanced resistance to structural evolution. In contrast to SnO₂ NPs, dispersed catalysts are more prone to reduction under similar conditions due to their lack of structural confinement and more exposed surfaces.

Observation of local pH variations

To thoroughly assess the high-efficiency OH⁻ confinement capability of the SnO₂ sponge, in situ fluorescence analysis was performed to observe local pH variations on the catalyst surface (Figs. 5a, S24). Dansyl acid, a fluorescent probe that exhibits increased photoluminescence (PL) intensity at around 510 nm as the pH rises from 2.77 to 6.70 (Fig. S25), was utilized to reflect local pH changes.^26,27 We performed the CO₂RR with SnO₂ sponge, SnO₂ NPs, and carbon black in an electrolyte containing Dansyl acid (pH 1.70, 1.0 M K⁺, 10⁻⁴ M Dansyl acid). As shown in Figs. 5b, d, S25, the local pH on the SnO₂ sponge reached 3.83 at −1.1 V_RHE and rapidly increased to 6.0 at −1.5 V_RHE. This indicates that the sponge structure can efficiently accumulate OH⁻ at low potential and effectively confine OH⁻ at higher applied potentials. In contrast, on the surface of SnO₂ NPs (Figs. 5c, d, S25), the pH also increases with potential increases, reaching a peak of 4.86 at −1.5 V_RHE. However, unlike the SnO₂ sponge, the pH rapidly decreases with further potential increases. This behavior is primarily attributed to faster proton migration from the bulk electrolyte to the cathode, driven by the electric field and concentration gradient. To rule out the possibility of physical adsorption of Dansyl acid under the influence of the electric field, we employed carbon black as a blank control, which showed no significant pH changes at various potentials (Figs. S25, S26). Hence, the sponge demonstrates a great capacity for OH⁻ storage, thereby protecting it from proton etching.

Fig. 5. Revealing local pH variations and impact on CO₂RR performance.

a Schematic illustration of local pH variations revealed by in situ fluorescence imaging. b, c Potential-dependent in situ fluorescence mapping (intensity at 510 nm) visualizing local pH over SnO₂ sponge (b) and SnO₂ NPs (c). Scale bar, 1 μm. d Local pH values of SnO₂ sponge and SnO₂ NPs under various applied potentials extracted from fluorescence signals. e Schematic illustrating the distinct effects of K⁺ concentration under identical CO₂ supply conditions during CO₂RR. f FE_HCOOH on SnO₂ sponge at different K⁺ concentrations under 3.5 sccm CO₂ flow at 400 mA cm⁻². g LSV curves of SnO₂ sponge and SnO₂ NPs in N₂-saturated acidic electrolytes (1.0 M K⁺) with varying pH values. h Time-dependent evolution of FE_HCOOH on SnO₂ sponge. Each bar denotes the FE_HCOOH within a defined time interval. i Comparison of FE_HCOOH between pristine catalysts and catalysts after electrolysis at 1.0 M K⁺ in pure H₂SO₄ electrolyte. Red column: SnO₂ sponge; blue column: SnO₂ NPs. Inset: schematic of the operational configuration. Source data are provided in a Source Data file.

To provide additional compelling evidence for the OH⁻ confinement characteristic in SnO₂ sponge, multiple characterization methods were employed. First, the SPCE of SnO₂ sponge exhibits a strong sensitivity to the K⁺ concentration. Theoretically, carbon efficiency in an acidic medium can reach 100%, as supported by previous studies.^2,3 However, the SPCE of SnO₂ sponge decreases significantly with increasing K⁺ concentration. As shown in Fig. 5e, f, the SPCE reaches 80.2% at 0.075 M K⁺ at 400 mA cm⁻², but drops to 49.6% at 1.0 M K⁺. This decline is attributed to the pronounced OH⁻ confinement effect in the SnO₂ sponge, leading to the formation of locally concentrated KOH. The localized KOH then reacts with a portion of the flowing CO₂ to form carbonates, thereby reducing the supply of available CO₂—an effect analogous to what is observed in alkaline electrolytes. As the bulk K⁺ concentration decreases, CO₂ loss is mitigated. In addition, as shown in Fig. S27, under high K⁺ concentration, increasing the CO₂ flow rate leads to an improvement in FE_HCOOH, further supporting the notion that CO₂ is consumed by the locally formed KOH.

LSV curves were recorded to evaluate proton diffusion behavior in N₂-saturated electrolytes with varying pH values and a constant K⁺ concentration of 1.0 M, where HER dominates. At low applied potentials, the HER behaviors over SnO₂ sponge and SnO₂ NPs reflect the rate of proton diffusion from the bulk electrolyte toward the catalyst interior. As shown in Fig. 5g, both SnO₂ sponge and SnO₂ NPs exhibited increasing current density with rising applied potential and decreasing pH from 1.82 to 1.28. Between −1.0 and −1.5 V_RHE, mass transfer on SnO₂ NPs was hindered, as evidenced by the current density plateau, which was more pronounced at higher bulk pH (1.82 and 1.47). When the bulk pH was set to 1.28, no significant current plateau was observed, indicating that protons can diffuse rapidly on the SnO₂ NPs surface. In contrast, SnO₂ sponge did not exhibit a significant current plateau, suggesting that protons on the catalyst surface are rapidly depleted at low applied potential, inhibiting HER. This demonstrates that OH⁻ confinement within the sponge structure efficiently neutralizes diffusing protons, stabilizing the microenvironment at the catalyst surface.

We also investigated the temporal evolution of FE_HCOOH to evaluate OH⁻ coverage on the catalyst surface. As shown in Figs. 5h, S28, FE_HCOOH was measured at multiple time points within an 8-min period at 400 mA cm⁻². FE_HCOOH exhibited a gradual increase over time. By calculating the FE_HCOOH values at corresponding time intervals, we determined that the FE for CO₂-to-HCOOH conversion was 69.1% from 0 to 30 s, 81.3% from 30 to 60 s, and 96.4% from 60 to 120 s. This indicates that the dense sponge structure efficiently minimizes intermixing between bulk and local electrolytes, prompting rapid proton depletion on the catalyst surface within 1 min without replenishment, thereby suppressing HER. Simultaneously, the high concentration of confined OH⁻ prevents proton attack on the catalyst surface and interior, ensuring sustained high FE_HCOOH.

We also compared the performance of pristine SnO₂ sponge and SnO₂ sponge electrolyzed at 1.0 M K⁺ in pure H₂SO₄ electrolyte. As shown in Fig. 5i, the pristine SnO₂ sponge exhibited only 10.4% FE_HCOOH at 400 mA cm⁻² in pure H₂SO₄ due to competing proton migration. However, after electrolysis at 1.0 M K⁺ electrolyte followed by rinsing and transferring to pure H₂SO₄, the treated SnO₂ sponge exhibited a FE_HCOOH of 48.2%, indicating efficient OH⁻ accumulation within the sponge structure. In contrast, the FE_HCOOH of SnO₂ NPs in pure H₂SO₄ showed negligible changes after treatment. To confirm that OH⁻ confinement in SnO₂ sponge is the primary factor responsible for achieving high FE, while K⁺ plays a secondary, less significant role, we compared the K⁺ concentration within SnO₂ sponge and SnO₂ NPs over various conducting times at 400 mA cm^-2. As shown in Fig. S29, K⁺ retention within the catalyst framework remains similar across both testing times and catalyst structures, indicating that K⁺ does not significantly affect the FE. This further supports the conclusion that OH⁻ confinement in SnO₂ sponge is the dominant factor driving high FE.

These results validate the design concept of the SnO₂ sponge: its efficient confinement of OH⁻ within the structure allows only minimal K⁺ required for maintaining a moderate alkaline environment, providing sufficient resistance against proton attack to achieve high FE and mitigate salt accumulation, ultimately ensuring longer operation.

Discussion

In summary, we demonstrate the effectiveness of a sponge-structured SnO₂ catalyst that establishes a stable alkaline microenvironment, enabling industrial CO₂-to-HCOOH conversion in an acidic electrolyte with low K⁺ concentration. The 3D interconnected network of SnO₂ sponge confines and concentrates OH⁻ within the porous channels, reducing the reliance on K⁺ while simultaneously enhancing both catalytic activity and long-term stability. Theoretical and experimental results demonstrate the OH^– confinement capability of SnO₂ sponge compared to dispersed SnO₂ NPs and SnO₂ solid MPs. The SnO₂ sponge catalyst achieves efficient CO₂ conversion over a broad potential range (0 − 800 mA cm⁻²), reaching a FE_HCOOH of 94.5% at 800 mA cm^–2. Moreover, it sustains continuous HCOOH production at a j_HCOOH of 390 mA cm^–2 for over 390 h at 0.25 M K⁺ without cleanup operations. This work validates the feasibility of sponge-like catalysts in acidic CO₂RR at low K⁺. The strategy can be extended to other CO₂RR catalytic systems, providing a solution for efficient acidic CO₂RR under low-K⁺ conditions.

Methods

SnO₂ sponge synthesis

In a typical procedure, the CaSn(OH)₆ and CaSnO₃ cubes were synthesized according to previous reports.^28,29 CaCl₂ (Aladdin Reagent Co., Ltd., 99%, 2.2 mmol), Na₂SnO₃ (Shanghai Macklin Biochemical Co., Ltd., 98%, 2.0 mmol), and PVP (Meryer (Shanghai) Chemical Technology Co., Ltd., 50 mg) were dissolved in 40 ml of deionized water with vigorous stirring. The solution was then transferred into a 50 mL Teflon-lined stainless steel autoclave to react at 140 °C for 10 h. After cooling to 25 °C, the products were filtered and dried for 6 h at 60 °C. CaSn(OH)₆ cubes were calcined in air at 500 °C for 2 h to obtain CaSnO₃ cubes. To etch off Ca ions, CaSnO₃ cubes (100 mg) were dissolved in 0.08 M HCl (Sinopharm Chemical Reagent Co., Ltd., 40 ml). The solution was then irradiated via microwave oven (800 W) at 160 °C for 90 min. After cooling to 25 °C, SnO₂ sponges were washed with deionized water and dried for 6 h at 60 °C.

SnO₂ nanoparticles and SnO₂ solid microparticles synthesis

SnO₂ NPs were synthesized using a method similar to that used for SnO₂ sponge, except that CaCl₂ was not added. SnO₂ solid MPs were purchased from J&K Scientific.

Preparation of working electrode

The catalyst ink was prepared by mixing 30 mg of catalyst, 15 mg of carbon black (XC-72), 4.5 ml of ethanol, 30 μl of 1 wt% PTFE emulsion, and 150 μl of Nafion solution, followed by sonication for 30 min. The ink was then air-brushed onto a gas diffusion layer (YLS-30T GDL) as the cathode electrode with a mass loading of 1.5 mg cm⁻².

COMSOL simulation

A three-dimensional reaction-diffusion model was built to model the microenvironment of solution species inside a porous cube, a solid cube, and nanoparticles through COMSOL Multiphysics 6.2. The structures of the catalyst model were constructed based on the experimental SEM images to perform finite element method (FEM) simulations. The porous cube, the solid cube, and nanoparticles have a similar size (0.6 × 0.6 × 0.6 m³). Eleven of the species (CO₂ (aq), CO(g), H₂(g), OH⁻, H⁺, HCO₃⁻, CO₃²⁻, HSO₄⁻, SO₄²⁻, HCOO⁻ and K⁺) were considered in the electrolyte within the modeled domain. All the simulations were carried out at −1.27 V_RHE, and the concentrations of all species will satisfy the steady-state mass conservation equation,

$$\nabla \cdot N_i=R_i$$ 1

where N_i and R_i are the flux and the net reaction rate in the solution of species i, respectively.

The diffusion and migration of species i is expressed as,

$$N_i=-D_i\frac{d{c}_i}{dx}-\frac{z_i{D}_{i}F{c}_i}{\text{RT}}\frac{d\varphi }{dx}$$ 2

where D_i, c_i and z_i represent the diffusion coefficient, concentration, and charge of species i, respectively, while Φ is the electrolyte ionic potential and x is the position. F is the Faraday constant, R is the ideal gas constant, and T is the temperature.

The heterogeneous reactions occur at the cathode surface in the electrocatalytic CO₂ reduction process.

$$\text{C}{\text{O}}_2\left(aq\right)+{\text{H}}_2\text{O}+2{\text{e}}^{-}\rightleftharpoons \text{CO}\left(g\right)+2\text{O}{\text{H}}^{-}$$ 3

$$\text{C}{\text{O}}_2\left(aq\right)+2{\text{H}}_2\text{O}+2{\text{e}}^{-}\rightleftharpoons \text{HCOOH}\left(aq\right)+2\text{O}{\text{H}}^{-}$$ 4

$$2{\text{H}}_2\text{O}+2{\text{e}}^{-}\rightleftharpoons H_2\left(g\right)+2\text{O}{\text{H}}^{-}$$ 5

The fluxes of each species at cathode surface are expressed as,

$$R_i=0\left(i=\text{HC}{\text{O}}_3^{-},\text{C}{\text{O}}_3^{2-},\text{HS}{\text{O}}_4^{-},\text{S}{\text{O}}_4^{2-},{\text{K}}^{+},{\text{H}}^{+}\right)$$ 6

$$R_{\text{C}{\text{O}}_2}=\frac{j_{tot}F{E}_{\text{HCOOH}}}{n_{\text{HCOOH}}F}+\frac{j_{tot}F{E}_{\text{CO}}}{n_{\text{CO}}F}$$ 7

$$R_{\text{CO}}=-\frac{j_{tot}F{E}_{\text{CO}}}{n_{\text{CO}}F}$$ 8

$$R_{\text{HCOOH}}=-\frac{j_{tot}F{E}_{\text{HCOOH}}}{n_{\text{HCOOH}}F}$$ 9

$$R_{\text{O}{\text{H}}^{-}}=-\frac{j_{tot}}{n_{\text{O}{\text{H}}^{-}}F}$$ 10

for all catalysts, the FE_HCOOH = 94.5%, FE_CO = 6.7%, $F{E}_{H_2}=4.4\%$ species are obtained from the experimental results, a total current density of j_tot = 800 mA cm⁻², where n_HCOOH = 2, n_CO = 2 are the numbers of electrons consumed to reduce CO₂ to HCOOH and CO, and $n_{{\mathrm{OH}}^{-}}=1$ is the number of electrons consumed per OH⁻ produced.

In addition, the homogeneous reactions occur in the electrolyte are expressed as,

$$\text{C}{\text{O}}_2\left(aq\right)+\text{O}{\text{H}}^{-}\begin{array}{c}\hfill k_1^f\hfill \\ \hfill \rightleftharpoons \hfill \\ \hfill k_1^r\hfill \end{array}\text{HC}{\text{O}}_3^{-}$$ 11

$$\text{HC}{\text{O}}_3^{-}+\text{O}{\text{H}}^{-}\begin{array}{c}\hfill k_2^f\hfill \\ \hfill \rightleftharpoons \hfill \\ \hfill k_2^r\hfill \end{array}\text{C}{\text{O}}_3^{2-}+{\text{H}}_{2}O$$ 12

$$H_2\text{O}\begin{array}{c}\hfill k_w^f\hfill \\ \hfill \rightleftharpoons \hfill \\ \hfill k_w^r\hfill \end{array}{H}^{+}+\text{O}{\text{H}}^{-}$$ 13

the values of the rate constants will be listed in the Tables S3, S4, and then R_i of species i due to the homogeneous reactions can be expressed as:

$$R_{H^{+}}=-k_w^r{c}_{H^{+}}{c}_{\text{O}{\text{H}}^{-}}+k_w^f$$ 14

$$R_{\text{O}{\text{H}}^{-}}=-k_1^f{c}_{\text{C}{\text{O}}_2}{c}_{\text{O}{\text{H}}^{-}}-k_2^f{c}_{\text{HC}{\text{O}}_3^{-}}{c}_{\text{O}{\text{H}}^{-}}-k_w^f{c}_{H^{+}}{c}_{\text{O}{\text{H}}^{-}}+k_1^r{c}_{\text{HC}{\text{O}}_3^{-}}+k_2^r{c}_{\text{C}{\text{O}}_3^{2-}}+k_w^f$$ 15

$$R_{\text{HC}{\text{O}}_3^{-}}=k_1^f{c}_{\text{C}{\text{O}}_2}{c}_{\text{O}{\text{H}}^{-}}-k_2^f{c}_{\text{HC}{\text{O}}_3^{-}}{c}_{\text{O}{\text{H}}^{-}}-k_1^r{c}_{\text{HC}{\text{O}}_3^{-}}+k_2^r{c}_{\text{C}{\text{O}}_3^{2-}}$$ 16

$$R_{\text{C}{\text{O}}_3^{2-}}=k_2^f{c}_{\text{HC}{\text{O}}_3^{-}}{c}_{\text{O}{\text{H}}^{-}}-k_2^r{c}_{\text{C}{\text{O}}_3^{2-}}$$ 17

$$R_{\text{C}{\text{O}}_2}=-k_1^f{c}_{\text{C}{\text{O}}_2}{c}_{\text{O}{\text{H}}^{-}}+k_1^r{c}_{\text{HC}{\text{O}}_3^{-}}$$ 18

the boundary conditions of the bulk solution are set to K₂SO₄ solutions saturated with CO₂.

The solubility of CO₂ in electrolyte solutions obeys Henry’s law,

$$c_{\text{C}{\text{O}}_2,\text{saturated}}=H_{\text{C}{\text{O}}_2}{P}_{\text{C}{\text{O}}_2}$$ 19

where $P_{{\mathrm{CO}}_2}$ is the partial pressure of CO₂, which is equal to 1 bar, $H_{{\mathrm{CO}}_2}$ is Henry’s constant and is given as a function of temperature T (298 K),

$$\ln H_{\text{C}{\text{O}}_2}=93.4517\times \frac{100}{T}+23.3585\times \ln \left(\frac{T}{100}\right)-60.2409$$ 20

The current i_l is governed by the charge conservation and current conservation,

$$\sum z_i{c}_i=0$$ 21

$$\nabla i_l=F\sum z_i{R}_i+Q_l$$ 22

where i_l is current density in the liquid phase, and Q_l is electrolyte current source.

Electrochemical measurements

All electrochemical measurements were conducted on a CHI electrochemical workstation at 25 °C. A typical three-electrode setup was used in a flow cell, with the experimental configuration illustrated in Fig. S12. The catalyst on the GDL was used as the working electrode (the exposed area of the electrode was 0.5 × 2 cm²). An iridium-loaded titanium mesh was employed as the anode for water oxidation. An Ag/AgCl electrode in saturated KCl was adopted as the reference electrode. The anolyte and catholyte chambers were separated by a proton exchange membrane (Nafion 115, 125 μm). The pH of the anodic and cathodic electrolytes was adjusted using H₂SO₄ and KOH, as monitored by a pH meter. The K⁺ concentrations were adjusted to various levels (0, 0.05, 0.075, 0.1, 0.25, 0.5, 1.0 M), with pH adjusted using H₂SO₄. The prepared electrolyte was stored in sealed containers. The catholyte flow rate was 5 mL min^–1, controlled by a peristaltic pump. The CO₂ flow rate, controlled by a gas flowmeter, was set to 15 sccm during CO₂RR under 0–800 mA cm^–2 and 10 sccm during the stability test. The potential was calculated according to the expression:

$$E_{\text{RHE}}=E_{\text{Ag}/\text{AgCl}}+0.059\times \text{pH}+0.198-0.85\times i{R}_{\text{Correct.}}$$ 23

All potential values reported are uncorrected unless otherwise noted. During the stability test, the 3.5 L electrolyte was refreshed periodically every 12 h.

CO₂RR product quantification

The gaseous products (CO and H₂) were quantified using a gas chromatograph (GC, FULI INSTRUMENTS GC9790Plus) equipped with a thermal conductivity detector (TCD) for quantifying H₂ and a flame ionization detector for quantifying CO. The FEs of CO and H₂ were calculated using the following equation:

$$FE=\frac{z\times \frac{P_{0}V}{RT}\times F}{Q_{total}}\times 100\%$$ 24

where z is the number of electrons required for producing a product, P₀ is the atmospheric pressure (1.013 × 10⁵ Pa), V is the volume of the gaseous product, R is the ideal gas constant (8.314 J mol⁻¹ K⁻¹), T is the reaction temperature (298 K), F is the Faradaic constant (96,485 C mol⁻¹), and Q_total is the charge passed through the catalyst.

The liquid products were analyzed using a ¹H-nuclear magnetic resonance (NMR, Bruker Ascend 500 MHz) spectrometer, employing the internal standard method. The FE of HCOOH was determined using the following equation:

$$FE=\frac{zcVF}{Q_{total}}\times 100\%$$ 25

where c and V are the concentration of HCOOH products and the electrolyte volume, respectively.

The production rate for HCOOH was calculated using the following equation:

$$\mathrm{Production}\mathrm{rate}=\frac{Q_{total}\times F{E}_{\mathrm{HCOOH}}}{zFtS}$$ 26

where t is the electrolysis time, and S is the exposed area of the electrode (1 cm²).

The cathodic ECE of the HCOOH product was calculated using the following formula:

$$ECE=\frac{1.23-E_{\text{HCOOH}}}{1.23-E_{\text{Cathode}}}\times F{E}_{\text{HCOOH}}\times 100\%$$ 27

where E_HCOOH is the theoretical equilibrium potential of HCOOH formation, which is −0.2 V_RHE.

The SPCE of CO₂ was calculated using

$$SPCE=\frac{\frac{\left(j\times F{E}_{\mathrm{HCOOH}}\times 60s\right)}{2\times F}}{\frac{V\times 1min}{V_m}}$$ 28

where j is the total current, V is the CO₂ flow rate, V_m is the unit molar volume of CO₂ at 25 °C (24.5 L mol⁻¹).

Characterization

Transmission electron microscopy (TEM), high-resolution TEM (HRTEM), high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images, and HAADF-STEM tomographic analysis were conducted on a Talos F200X G2 (FEI) transmission electron microscope under an acceleration voltage of 200 kV. Scanning electron microscopy (SEM) was performed using a Zeiss Gemini 450 microscope system. X-ray powder diffraction (XRD) was carried out with a Bruker D8 Advance X-ray diffractometer using Cu Kα radiation at a scanning rate of 9°/min in the 2θ range from 10° to 80°. N₂ adsorption isotherms and BET surface areas were obtained using a Micromeritics 3FLEX system. X-ray photoelectron spectroscopy (XPS) studies were performed using an ESCALAB Xi+ (Thermo Fisher) spectrometer with Al Kα radiation. Sn K-edge XAFS analyses were performed with Si (111) crystal monochromators at the BL14W Beam line at the Shanghai Synchrotron Radiation Facility (SSRF), Shanghai, China. Fluorescence spectra were collected using a fluorescence spectrophotometer (Horiba LabRAM Odyssey).

In situ fluorescence experiments

The in situ fluorescence system was performed in a flow cell (Fig. S24). The Pt plate and Ag/AgCl electrode served as the counter electrode and reference electrode, respectively. The electrolyte (1.0 M K⁺, pH 1.82) was prepared by mixing H₂SO₄ (Sinopharm Chemical Reagent Co., Ltd.) and K₂SO₄ (Shanghai Macklin Biochemical Co., Ltd., 99%) with 1.0 × 10⁻⁴ M Dansyl acid (J&K Scientific, 98%). The CO₂ and electrolyte flow rates were 5 sccm and 5 mL min^–1 regulated by a mass flow controller and a flow battery pump, respectively. The SnO₂ sponge and SnO₂ NPs on carbon paper were used as the working electrode, with a loading mass similar to that used in CO₂RR. Calibration of the pH-fluorescence intensity curves (Figs. 4d, S25a) was carried out in a mixture of 0.5 M K₂SO₄ and 1.0 × 10⁻⁴ M Dansyl acid with different pH values ranging from 1.70 to 6.70, adjusted with H₂SO₄. Potential-dependent in situ fluorescence spectra were recorded from −0.9 to −1.8 V_RHE. Time-dependent in situ fluorescence mapping spectra were recorded at various applied potentials.

Author contributions

Conceptualization: K.Z., B.J., L.S., and B.Z. Investigation: K.Z., B.J., and Y.S. Visualization: K.Z., W.S., and B.J. COMSOL simulation: W.S. and T.W. Writing: K.Z., W.S., B.J., T.W., L.S., and B.Z.

Peer review

Peer review information

Nature Communications thanks Junwei Fu and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data generated in this study are provided in the paper, Supplementary Information and Source Data file. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-72463-z.