Bipolar membrane electrolyzers enable high single-pass CO2 electroreduction to multicarbon products

Electrochemical CO2 reduction in membrane-electrode-assembly (MEA) reactors often suffers major carbon losses because CO2 reacts with alkaline or neutral electrolytes to form (bi)carbonates that cross over to the anode, necessitating energy-intensive separation of CO2 from O2. Bipolar membranes (BPMs) can block CO2 crossover and regenerate dissolved CO2 from (bi)carbonates, but the acidic cation-exchange layer (CEL) adjacent to the cathode degrades CO2 reduction (CO2RR) selectivity. Buffering catholytes placed between BPM and cathode maintain high local pH at the catalyst but regenerate CO2 near the CEL surface, creating long diffusion paths and poor reactant availability, limiting single-pass utilization (SPU) to ~15% for C2+ and ~6% with flowing catholyte. Prior BPM-based systems therefore did not surpass the ~25% SPU limit for C2+ in neutral media. This work proposes a stationary, non-buffering catholyte of controlled thickness between the BPM and the cathode to extend proton penetration, regenerate CO2 closer to the cathode, maintain high local pH, and thereby overcome mass-transfer limitations. A model-guided design targets a regenerated CO2 diffusion path of ~10–12 µm to balance regeneration and diffusion, enabling high SPU while sustaining C2+ selectivity.

Previous CO2RR electrolyzer configurations include alkaline/neutral AEM-based MEAs that exhibit significant CO2 crossover as carbonate with CO2/O2 at the anode ~2 across 100–300 mA cm−2, imposing large anodic separation energy. BPMs have been explored to block CO2 crossover and reconvert (bi)carbonates to CO2, but commercial BPMs acidify the cathode and deteriorate CO2RR selectivity. Strategies using buffering catholytes (e.g., KHCO3) between BPM and cathode maintained cathode pH >12 but regenerated CO2 at the CEL surface, creating thick diffusion layers and limiting SPU (~15% for C2+; ~6% with flowing catholyte). Table 1 compares designs: neutral/alkaline AEM systems reached max SPU ~24–30% on Cu; BPMs with solid polymer or thick bicarbonate layers favor C1 products and did not exceed the 25% SPU limit for C2+. Acidic systems have recently achieved high SPU (>75%) and eliminated anodic separation energy, but at higher cell voltages and/or lower ethylene selectivity, increasing overall energy intensity. These gaps motivate a BPM-based approach that maintains high local pH, minimizes diffusion distance of regenerated CO2, and suppresses crossover.

The study combined finite-element simulations and experiments. A 1D multiphysics COMSOL model simulated species transport, homogeneous carbonate equilibria, water dissociation, and heterogeneous CO2RR/HER kinetics across a gas diffusion layer (50 µm), Cu catalyst layer (~0.1 µm), stationary catholyte (SC) layer (16–250 µm), and a CEL boundary. Constant-concentration boundaries were used (e.g., 37.8 mM CO2 in the GDL, constant proton generation at the CEL at set current density). Two catholytes were examined: non-buffering 0.5 M K2SO4 and buffering KHCO3. Simulations computed local pH and dissolved CO2 profiles, locating the effective diffusion boundary (1% drop from bulk CO2) and quantifying diffusion layer thicknesses. Experimentally, a stationary-catholyte BPMEA (SC-BPMEA) was assembled: a Cu nanoparticle catalyst (1.5 mg cm−2) was spray-deposited on a hydrophobic carbon GDL; the anode was IrO2 on Ti felt (1.5 mg cm−2). A custom BPM (Nafion 212 CEL | TiO2 nanoparticle water-dissociation layer | Piperion AEL) under reverse bias contacted a porous support saturated with 0.5 M K2SO4 forming the SC-layer (thicknesses 16, 65, 125, 250 µm). The anode side was fed with 0.1 M KHCO3 unless otherwise noted. Cell voltages were measured without iR correction using a potentiostat. Gas products, CO2 and O2 were quantified by GC; outlet flows were measured by bubble column; liquid products were quantified by 1H NMR. CO2 SPU was assessed under restricted CO2 feed (varied flow rates normalized per cm2). Electrochemical impedance spectroscopy probed ohmic changes. A cation-exchange membrane variant (SC-CEMEA) with acidic anolyte (K2SO4 + H3PO4, pH 1.84–2.37) tested feasibility of replacing BPM. An energy assessment quantified electricity, cathodic separation, and anodic separation energies for ethylene production, comparing SC-BPMEA to state-of-the-art neutral-MEA and acidic systems, using literature-based separations models and assuming <0.5% CO2 crossover for BPM/acidic cases.

• Modeling showed that in non-buffering catholytes (0.5 M K2SO4) the local cathode pH remains >11 for SC-layer thicknesses of 65–250 µm, favoring CO2RR over HER; at 16 µm, pH drops to ~8.7, increasing HER.
• Simulated diffusion layer thicknesses of regenerated dissolved CO2 decrease with thinner SC-layers: ~75 µm (250 µm layer), ~35 µm (125 µm), ~12 µm (65 µm), ~5 µm (16 µm). Keeping the diffusion path length ~10–12 µm balances regeneration and diffusion.
• The SC-BPMEA suppressed CO2 crossover: anode CO2/O2 ratio ~0.06 at 200 mA cm−2 (one order of magnitude lower than AEMEA, ~2), with crossover <0.5% of total CO2 input.
• Cell voltage depended on SC-layer thickness at 200 mA cm−2: 250 µm, 5.1 V; 125 µm, higher than 65 µm; 65 µm, 3.8 V; 16 µm, 4.4 V (affected by low-porosity support). For 65 µm at 200 mA cm−2, voltage breakdown totaled 3.82 V with small BPM water-dissociation overpotential contribution.
• Gas product Faradaic efficiencies (10 sccm cm−2 CO2 at 35 °C): with 65/125/250 µm SC-layers, H2 FE ~20% at 200 mA cm−2 and ethylene FE ~35–43%, indicating maintained high local pH. At 16 µm, H2 FE rose to 88% at 200 mA cm−2.
• Under restricted CO2 (200 mA cm−2, 35 °C), total CO2 SPU increased with decreasing SC-layer thickness: up to 21% (250 µm), 61% (125 µm), and 78% (65 µm). For 65 µm, as CO2 feed decreased from 1.17 to 0.58 and 0.29 sccm cm−2, C2+ FE decreased from 49% to 48% and 34%, while H2 FE increased from 23% to 31% and 64%, consistent with CO2 mass-transport limitation.
• The optimized 65 µm SC-BPMEA achieved single-pass CO2 utilization of 78%, surpassing the previous 25% SPU limit for C2+ in neutral systems and reducing downstream CO2 separation energy by ~10× compared to past systems.
• Stability: >50 h at 200 mA cm−2 with limited CO2 availability (1.42 sccm cm−2) with minimal crossover.
• Energy assessment (per ton ethylene): SC-BPMEA overall energy ~470 GJ/t at 10 sccm cm−2 and ~410 GJ/t at 1.17 sccm cm−2, versus neutral-MEA 499 GJ/t, acidic flow cell 637 GJ/t, acidic MEA 465 GJ/t. Optimal tradeoff at ~35% SPU minimized cathodic separation energy to ~15 GJ/t without large FE penalties.
• Attempted SC-CEMEA showed lower initial voltage and reasonable selectivity but suffered instability: CO2RR selectivity declined to ~100% H2 after ~3 h, electrolyte ejection due to water imbalance, and need for continuous acid/salt addition, indicating BPM is required for steady operation.

By introducing a stationary, non-buffering catholyte of optimized thickness between the BPM and the catalyst, the electrolyzer regenerates CO2 from (bi)carbonate close to the cathode while maintaining a highly alkaline local microenvironment. This shortens the diffusion path of regenerated CO2, improves mass transport, and eliminates CO2 crossover, thereby enabling high SPU without sacrificing cell voltage or selectivity. Simulations predicted and experiments confirmed that a ~65 µm SC-layer (diffusion length ~12 µm) balances regeneration rate and CO2 flux, delivering SPU up to 78% at 200 mA cm−2 with ethylene FE around 40% and cell voltage ~3.8 V—comparable to neutral AEMEA systems but with negligible anodic CO2. Thicker SC-layers extend diffusion distances, lowering CO2 availability under limited feed and reducing SPU; thinner layers reduce local pH and increase HER. The energy analysis shows that high SPU in SC-BPMEA substantially lowers cathodic and eliminates anodic separation energy while maintaining competitive electricity consumption, yielding lower overall energy intensity than neutral and acidic benchmarks. Attempts to replace BPM with a CEM indicate that BPM’s junction and ion selectivity are crucial for carbon balance and water management in sustained operation.

This work presents a BPM-based MEA architecture with a stationary, non-buffering catholyte layer that overcomes carbonate crossover losses and mass-transfer limits, enabling high single-pass CO2 utilization (up to 78%) for multicarbon (C2+) products at industrially relevant current densities and cell voltages. Modeling-guided control of catholyte thickness establishes a short diffusion path for regenerated CO2 while maintaining high local pH, yielding high selectivity and low energy intensity by minimizing downstream separations. The SC-BPMEA platform offers a robust route to evaluate CO2RR catalysts under high-utilization conditions and can inform other electrochemical processes requiring decoupled microenvironments. Future improvements could include alternative catholytes (e.g., ionic liquids or organic salts), optimized porous support structure and hydrophobicity, enhanced cathode catalysts and processing, and BPMs with lower water-dissociation losses.

Selectivity to C2+ products decreases as CO2 feed is restricted to maximize SPU, increasing HER due to mass-transport limitations. Extremely thin SC-layers (e.g., 16 µm) reduce local pH and favor HER. Thicker SC-layers (>125 µm) cause longer diffusion distances, CO2 bubble formation near the CEL, higher ohmic losses, and diminished SPU. The CEM-based variant showed instability from water imbalance and co-ion transport, requiring continuous acid/salt addition and leading to electrolyte ejection and selectivity loss. Although >50 h stability was demonstrated under limited CO2, longer-term durability and operation at higher current densities were not reported. Achieving simultaneously very high SPU and maximum ethylene FE remains a tradeoff under restricted CO2 supply.