Thank you for visiting Nature.com. The version of browser you are using has limited CSS support. For best results, we recommend that you use a newer version of your browser (or disable Compatibility Mode in Internet Explorer). In the meantime, to ensure ongoing support, we are displaying the site without styling or JavaScript.
       Electrochemical reduction of carbon dioxide to formic acid is a promising way to improve carbon dioxide utilization and has potential applications as a hydrogen storage medium. In this work, a zero-gap membrane electrode assembly architecture is developed for the direct electrochemical synthesis of formic acid from carbon dioxide. A key technological advancement is the perforated cation exchange membrane, which, when used in a forward biased bipolar membrane configuration, allows formic acid formed at the membrane interface to be displaced through the anodic flow field in concentrations as low as 0.25 M. Without additional sandwich components between the anode and cathode the concept aims to leverage existing battery materials and designs common in fuel cells and hydrogen electrolysis, allowing for a faster transition to scale-up and commercialization. In a 25 cm2 cell, the perforated cation exchange membrane configuration provides >75% Faraday efficiency for formic acid at <2 V and 300 mA/cm2. More importantly, a 55-hour stability test at 200 mA/cm2 showed stable Faraday efficiency and cell voltage. A techno-economic analysis is used to illustrate ways to achieve cost parity with current formic acid production methods.
       Electrochemical reduction of carbon dioxide to formic acid using renewable electricity has been shown to reduce production costs by up to 75%1 compared to traditional fossil fuel-based methods. As indicated in the literature2,3, formic acid has a wide range of applications, from an efficient and economical means of storing and transporting hydrogen to a feedstock for the chemical industry4,5 or the biomass industry6. Formic acid has even been identified as a feedstock for subsequent conversion into sustainable jet fuel intermediates using metabolic engineering7,8. With the development of formic acid economics1,9, several research works have focused on optimizing catalyst selectivity10,11,12,13,14,15,16. However, many efforts continue to focus on small H-cells or liquid flow cells operating at low current densities (<50 mA/cm2). To reduce costs, achieve commercialization and increase subsequent market penetration, electrochemical carbon dioxide reduction (CO2R) must be performed at high current densities (≥200 mA/cm2) and Faraday efficiency (FE)17 while maximizing material utilization and using battery components from the Technology fuel cells and water electrolysis allow CO2R devices to take advantage of economies of scale18. In addition, to increase the utility of production and avoid additional downstream processing, formic acid should be used as the final product rather than formate salts19.
       In this direction, recent efforts have been made to develop industrially relevant CO2R formate/formic acid based gas diffusion electrode (GDE) devices. A comprehensive review by Fernandez-Caso et al.20 summarizes all electrochemical cell configurations for the continuous reduction of CO2 to formic acid/formate. In general, all existing configurations can be divided into three main categories: 1. Flow-through catholytes19,21,22,23,24,25,26,27, 2. Single membrane (cation exchange membrane (CEM)28 or anion exchange membrane (AEM)29 and 3. Sandwich configuration15,30,31,32. Simplified cross-sections of these configurations are shown in Figure 1a. For the flow configuration of the catholyte, an electrolyte chamber is created between the membrane and the cathode of the GDE. Flow-through catholyte is used to create ion channels in the cathode layer of the catalyst33, although its need for controlling formate selectivity is debated34. However, this configuration was used by Chen et al. Using a SnO2 cathode on a carbon substrate with a 1.27 mm thick catholyte layer, up to 90% FE 35 at 500 mA/cm2 was achieved. The combination of a thick catholyte layer and a reverse-biased bipolar membrane (BPM) that limits ion transfer provides an operating voltage of 6 V and an energy efficiency of 15%. To improve energy efficiency, Li et al., using a single CEM configuration, achieved an FE 29 of 93.3% at a fractional current density of 51.7 mA/cm2. Diaz-Sainz et al.28 used a filter press with a single CEM membrane at a current density of 45 mA/cm2. However, all methods produced formate rather than the preferred product, formic acid. In addition to additional processing requirements, in CEM configurations, formats such as KCOOH can quickly accumulate in the GDE and flow field, causing transport restrictions and eventual cell failure.
       Comparison of the three most prominent CO2R to formate/formic acid conversion device configurations and the architecture proposed in this study. b Comparison of total current and formate/formic acid yield for catholyte configurations, sandwich configurations, single CEM configurations in the literature (shown in Supplementary Table S1) and our work. Open marks indicate the production of formate solution, and solid marks indicate the production of formic acid. *Configuration shown using hydrogen at the anode. c Zero-gap MEA configuration using a composite bipolar membrane with a perforated cation exchange layer operating in forward bias mode.
       To prevent formate formation, Proietto et al. 32 used a splitless filter press configuration in which deionized water flows through the interlayer. The system can achieve >70% CE in the current density range of 50–80 mA/cm2. Similarly, Yang et al. 14 proposed the use of a solid electrolyte interlayer between the CEM and AEM to promote the formation of formic acid. Yang et al.31,36 achieved 91.3% FE in a 5 cm2 cell at 200 mA/cm2, producing a 6.35 wt% formic acid solution. Xia et al. Using a similar configuration, 83% conversion of carbon dioxide (CO2) to formic acid FE was achieved at 200 mA/cm2, and the system durability was tested for 100 hours 30 minutes. Although small-scale results are promising, the increased cost and complexity of porous ion exchange resins make it difficult to scale interlayer configurations to larger systems (e.g., 1000 cm2).
       To visualize the net effect of the different designs, we tabulated the formate/formic acid production per kWh for all the systems mentioned earlier and plotted them in Figure 1b. It is clear here that any system containing a catholyte or interlayer will peak its performance at low current densities and degrade at higher current densities, where the ohmic limit may determine the cell voltage. Moreover, although the energy-efficient CEM configuration provides the highest molar formic acid production per kWh, salt buildup can lead to rapid performance degradation at high current densities.
       To mitigate the previously discussed failure modes, we developed a membrane electrode assembly (MEA) containing a composite forward biased BPM with a perforated cation exchange membrane (PCEM). The architecture is shown in Figure 1c. Hydrogen (H2) is introduced into the anode to generate protons through a hydrogen oxidation reaction (HOR). A PCEM layer is introduced into the BPM system to allow formate ions generated at the cathode to pass through the AEM, combine with protons to form formic acid at the BPM interface and interstitial pores of the CEM, and then exit through the GDE anode and flow field. . Using this configuration, we achieved >75% FE of formic acid at <2 V and 300 mA/cm2 for a 25 cm2 cell area. Most importantly, the design utilizes commercially available components and hardware architectures for fuel cell and water electrolysis plants, allowing for a faster time to scale. Catholyte configurations contain catholyte flow chambers which can cause a pressure imbalance between the gas and liquid phases, especially in larger cell configurations. For sandwich structures with porous layers of fluid flow, significant efforts are required to optimize the porous intermediate layer to reduce pressure drop and carbon dioxide accumulation within the intermediate layer. Both of these can lead to disruption of cellular communications. It is also difficult to produce free-standing thin porous layers on a large scale. In contrast, the proposed new configuration is a zero-gap MEA configuration that does not contain a flow chamber or intermediate layer. Compared to other existing electrochemical cells, the proposed configuration is unique in that it allows direct synthesis of formic acid in a scalable, energy-efficient, zero-gap configuration.
       To suppress hydrogen evolution, large-scale CO2 reduction efforts have used MEA and AEM membrane configurations in combination with high molar concentration electrolytes (e.g., 1-10 M KOH) to create alkaline conditions at the cathode (as shown in Figure 2a). In these configurations, formate ions formed at the cathode pass through the membrane as negatively charged species, then KCOOH is formed and exits the system through the anodic KOH stream. Although formate FE and cell voltage were initially favorable as shown in Figure 2b, stability testing resulted in a reduction in FE of approximately 30% in just 10 h (Figure S1a–c). It should be noted that the use of 1 M KOH anolyte is critical to minimize anodic overvoltage in alkaline oxygen evolution reaction (OER) systems37 and achieve ion accessibility within the cathode catalyst bed33. When the anolyte concentration is reduced to 0.1 M KOH, both cell voltage and formic acid oxidation (loss of formic acid) increase (Figure S1d), illustrating a zero-sum trade-off. The degree of formate oxidation was assessed using the overall mass balance; for more details, see the “Methods” section. The performance using MEA and single CEM membrane configurations was also studied, and the results are shown in Figure S1f,g. FE formate collected from the cathode was >60% at 200 mA/cm2 at the start of the test, but rapidly degraded within two hours due to cathode salt accumulation discussed previously (Figure S11).
       Schematic of a zero-gap MEA with CO2R at the cathode, hydrogen oxidation reaction (HOR) or OER at the anode, and one AEM membrane in between. b FE and cell voltage for this configuration with 1 M KOH and OER flowing at the anode. Error bars represent the standard deviation of three different measurements. in FE and system cell voltage with H2 and HOR at the anode. Different colors are used to distinguish formate and formic acid production. d Schematic diagram of MEA with BPM shifted forward in the middle. FE and battery voltage versus time at 200 mA/cm2 using this configuration. f Cross-sectional image of a forward-biased BPM MEA after a short test.
       To produce formic acid, hydrogen is supplied to a Pt-on-carbon (Pt/C) catalyst at the anode. As shown in Figure 2d, a forward-biased BPM generating protons at the anode has been previously investigated to achieve formic acid production. The BPM tuning unit failed after 40 minutes of operation at a current of 200 mA/cm2, accompanied by a voltage surge of more than 5 V (Fig. 2e). After testing, obvious delamination was observed at the CEM/AEM interface. Besides formate, anions such as carbonate, bicarbonate and hydroxide can also pass through the AEM membrane and react with protons at the CEM/AEM interface to produce CO2 gas and liquid water, leading to BPM delamination (Fig. 2f) and , ultimately leading to cell failure.
       Based on the performance and failure mechanisms of the above configuration, a new MEA architecture is proposed as shown in Figure 1c and detailed in Figure 3a38. Here, the PCEM layer provides a path for the migration of formic acid and anions from the CEM/AEM interface, thereby reducing the accumulation of the substance. At the same time, the PCEM interstitial pathway directs formic acid into the diffusion medium and flow field, reducing the possibility of formic acid oxidation. The polarization results using AEMs with thicknesses of 80, 40 and 25 mm are shown in Figure 3b. As expected, although the overall cell voltage increases with increasing AEM thickness, using a thicker AEM prevents back diffusion of formic acid, thereby increasing cathode pH and decreasing H2 production (Fig. 3c–e).
       a Illustration of MEA structure with AEM and perforated CEM and different formic acid transport pathways. b Cell voltage at different current densities and different AEM thicknesses. in EE at various current densities with an AEM thickness of 80 μm (d) 40 μm, e) 25 μm. Error bars represent standard deviation measured from three separate samples. f Simulation results of formic acid concentration and pH value at the CEM/AEM interface at different AEM thicknesses. f PC and pH in the cathode layer of the catalyst with different AEM film thicknesses. g Two-dimensional distribution of formic acid concentration with CEM/AEM interface and perforation.
       Figure S2 shows the distribution of formic acid concentration and pH across the MEA thickness using Poisson-Nernst-Planck finite element modeling. It is not surprising that the highest concentration of formic acid, 0.23 mol/L, is observed at the CEM/AEM interface, since formic acid is formed at this interface. The concentration of formic acid through the AEM decreases more rapidly as the thickness of the AEM increases, indicating greater resistance to mass transfer and less formic acid flux due to back diffusion. Figures 3 f and g show the pH and formic acid values ​​in the cathode catalyst bed caused by back diffusion and the two-dimensional distribution of formic acid concentration, respectively. The thinner the AEM membrane, the higher the concentration of formic acid near the cathode, and the pH of the cathode becomes acidic. Therefore, although thicker AEM membranes result in higher ohmic losses, they are critical to preventing back diffusion of formic acid to the cathode and maximizing the high purity of the FE formic acid system. Finally, increasing the AEM thickness to 80 μm resulted in FE >75% for formic acid at <2 V and 300 mA/cm2 for a 25 cm2 cell area.
       To test the stability of this PECM-based architecture, the battery current was maintained at 200 mA/cm2 for 55 hours. The overall results are shown in Figure 4, with results from the first 3 hours highlighted in Figure S3. When using the Pt/C anodic catalyst, the cell voltage increased sharply within the first 30 min (Figure S3a). Over a longer period of time, the cell voltage remained almost constant, providing a degradation rate of 0.6 mV/h (Fig. 4a). At the start of the test, the PV of formic acid collected at the anode was 76.5% and the PV of hydrogen collected at the cathode was 19.2%. After the first hour of testing, the hydrogen FE dropped to 13.8%, indicating improved formate selectivity. However, the oxidation rate of formic acid in the system dropped to 62.7% in 1 hour, and the oxidation rate of anodic formic acid increased from almost zero at the beginning of the test to 17.0%. Subsequently, the FE of H2, CO, formic acid and the rate of anodic oxidation of formic acid remained stable during the experiment. The increase in formic acid oxidation during the first hour may be due to the accumulation of formic acid at the PCEM/AEM interface. As the concentration of formic acid increases, it not only exits through the perforation of the membrane, but also diffuses through the FEM itself and enters the Pt/C anode layer. Since formic acid is a liquid at 60°C, its accumulation can cause mass transfer problems and result in preferential oxidation over hydrogen.
       a Cell voltage versus time (200 mA/cm2, 60 °C). The inset shows an optical microscope image of a cross-section of an MEA with a perforated EM. Scale bar: 300 µm. b Purity of PE and formic acid as a function of time at 200 mA/cm2 using a Pt/C anode.
       The morphology of the samples at the beginning of testing (BOT) during preparation and at the end of testing (EOT) after 55 h of stability testing was characterized using nano-X-ray computed tomography (nano-CT), as shown in Figure 5 a. The EOT sample has a larger catalyst particle size with a diameter of 1207 nm compared to 930 nm for BOT. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images and energy-dispersive X-ray spectroscopy (EDS) results are shown in Figure 5b. While the BOT catalyst layer contains most of the smaller catalyst particles as well as some larger agglomerates, in the EOT stage the catalyst layer can be divided into two distinct regions: one with significantly larger solid particles and the other with more porous regions. Number of smaller particles. The EDS image shows that the large solid particles are rich in Bi, possibly metallic Bi, and the porous regions are rich in oxygen. When the cell is operated at 200 mA/cm2, the negative potential of the cathode will cause reduction of Bi2O3, as evidenced by the in situ X-ray absorption spectroscopy results discussed below. HAADF-STEM and EDS mapping results show that Bi2O3 undergoes a reduction process, causing them to lose oxygen and agglomerate into larger metal particles. X-ray diffraction patterns of the BOT and EOT cathodes confirm the interpretation of the EDS data (Fig. 5c): only crystalline Bi2O3 was detected in the BOT cathode, and crystalline bimetal was found in the EOT cathode. To understand the effect of cathode potential on the oxidation state of the Bi2O3 cathode catalyst, the temperature was varied from open circuit potential (+0.3 V vs RHE) to -1.5 V (vs RHE). It is observed that the Bi2O3 phase begins to be reduced at -0.85 V relative to RHE, and a decrease in the intensity of the white line in the edge region of the spectrum indicates that metallic Bi is reduced to 90% of RHE at -1.1. V against RHE (Fig. 5d). Regardless of the mechanism, the overall selectivity of formate at the cathode is essentially unchanged, as inferred from H2 and CO FE and formic acid formation, despite significant changes in cathode morphology, catalyst oxidation state, and microcrystalline structure.
       a Three-dimensional structure of the catalyst layer and distribution of catalyst particles obtained using nano-X-ray CT. Scale bar: 10 µm. b Top 2: HAADF-STEM images of cathode layers of BOT and EOT catalysts. Scale bar: 1 µm. Bottom 2: Enlarged HADF-STEM and EDX images of the cathode layer of the EOT catalyst. Scale bar: 100 nm. c X-ray diffraction patterns of BOT and EOT cathode samples. d In situ X-ray absorption spectra of Bi2O3 electrode in 0.1 M KOH as a function of potential (0.8 V to -1.5 V vs. RHE).
       To determine exactly what opportunities exist for improving energy efficiency by inhibiting formic acid oxidation, an H2 reference electrode was used to identify the contribution of voltage loss39. At current densities less than 500 mA/cm2, the cathode potential remains below -1.25 V. The anodic potential is divided into two main parts: the exchange current density HOR and the theoretical overvoltage HOR 40 predicted by the previously measured Bulter-Volmer equation, and the remaining part is due to oxidation formic acid. Due to the much slower reaction kinetics compared to HOR41, the small rate of formic acid oxidation reaction at the anode can result in a significant increase in the anodic potential. The results show that complete inhibition of formic acid anodic oxidation can eliminate nearly 500 mV overvoltage.
       To test this estimate, the flow rate of deionized water (DI) at the anode inlet was varied to reduce the concentration of effluent formic acid. Figures 6b and c show FE, formic acid concentration, and cell voltage as a function of DI flux at the anode at 200 mA/cm2. As the deionized water flow rate increased from 3.3 mL/min to 25 mL/min, the formic acid concentration at the anode decreased from 0.27 mol/L to 0.08 mol/L. In comparison, using the sandwich structure proposed by Xia et al. 30 a formic acid concentration of 1.8 mol/L was obtained at 200 mA/cm2. Decreasing the concentration improves the overall FE of formic acid and reduces the FE of H2 as the cathode pH becomes more alkaline due to decreased back diffusion of formic acid. The reduced formic acid concentration at maximum DI flow also virtually eliminated formic acid oxidation, resulting in a total cell voltage of just under 1.7 V at 200 mA/cm2. Battery temperature also affects overall performance, and the results are shown in Figure S10. However, PCEM-based architectures can significantly improve energy efficiency in inhibiting formic acid oxidation, whether through the use of anodic catalysts with improved hydrogen selectivity toward formic acid or through device operation.
       a Cell voltage breakdown using cell reference H2 electrode operating at 60 °C, Pt/C anode and 80 µm AEM. b FE and formic acid concentrations collected at 200 mA/cm2 using different flow rates of anodic deionized water. c When the anode collects formic acid in different concentrations, the cell voltage is 200 mA/cm2. Error bars represent the standard deviation of three different measurements. d Minimum selling price broken down by performance at various deionized water flow rates using national industrial average electricity prices of US$0.068/kWh and US$4.5/kg hydrogen. (*: The minimum oxidation state of formic acid at the anode is assumed to be 10 M FA, the national average industrial electricity price is $0.068/kWh, and hydrogen is $4.5/kg. **: The minimum oxidation state is assumed formic acid. The concentration of FA at the anode is 1.3 M anode, the expected future electricity price is $0.03/kWh, and the dotted line represents the market price of 85 wt% FA.
       A techno-economic analysis (TEA) was conducted to obtain the minimum selling price of the fuel assemblies under a range of operating conditions, as shown in Figure 5d. Methods and background data for TEA can be found in the SI. When the LC concentration in the anode exhaust is higher, despite the higher cell voltage, the overall cost of the fuel assembly is reduced due to the reduction in separation cost. If anodic oxidation of formic acid can be minimized through catalyst development or electrode technology, the combination of lower cell voltage (1.66 V) and higher FA concentration in the effluent (10 M) would reduce the cost of electrochemical FA production to 0.74 US dollars/kg (based on electricity). price) $0.068/kWh and $4.5/kg hydrogen42. Moreover, when combined with the projected future cost of renewable electricity of $0.03/kWh and hydrogen of $2.3/kg, the FA wastewater target is reduced to 1.3 million, resulting in a final projected production cost is US$0.66/kg43. This is comparable to current market prices. Thus, future efforts focused on electrode materials and structures could further reduce anodization while allowing operation at lower cell voltages to produce higher LC concentrations.
       In summary, we have studied several zero-gap MEA structures for CO2 reduction to formic acid and proposed a structure containing a composite forward-biased bipolar membrane including a perforated cation exchange membrane (PECM) to facilitate the membrane mass transfer interface for the resulting formic acid. . This configuration generates >96% formic acid at concentrations up to 0.25 M (at an anode DI flow rate of 3.3 mL/min). At higher DI flow rates (25 mL/min), this configuration provided a current density of >80% FE of 200 mA/cm2 at 1.7 V using a 25 cm2 cell area. At moderate anodic DI rates (10 mL/min), the PECM configuration maintained stable voltage and high formic acid FE levels for 55 h of testing at 200 mA/cm2. The high stability and selectivity achieved by commercially available catalysts and polymeric membrane materials can be further enhanced by combining them with optimized electrocatalysts. Subsequent work will focus on adjusting operating conditions, anode catalyst selectivity, and MEA structure to reduce formic acid oxidation, resulting in a more concentrated effluent at lower cell voltages. The simple approach to using carbon dioxide for formic acid presented here eliminates the need for anolyte and catholyte chambers, sandwich components, and specialty materials, thereby increasing cell energy efficiency and reducing system complexity, making it easier to scale up. The proposed configuration provides a platform for the future development of technically and economically viable CO2 conversion plants.
       Unless otherwise stated, all chemical grade materials and solvents were used as received. Bismuth oxide catalyst (Bi2O3, 80 nm) was purchased from US Research Nanomaterials, Inc. Polymer powder (AP1-CNN8-00-X) was provided by IONOMR. Omnisolv® brand N-propanol (nPA) and ultrapure water (18.2 Ω, Milli–Q® Advantage A10 water purification system) were purchased from Millipore Sigma. ACS certified methanol and acetone are purchased from VWR Chemicals BDH® and Fisher Chemical, respectively. The polymer powder was mixed with a mixture of acetone and methanol in a ratio of 1:1 by weight to obtain a polymer dispersion with a concentration of 6.5 wt.%. Prepare catalytic ink by mixing 20g Bi2O3, ultrapure water, nPA and ionomer dispersion in a 30ml jar. The composition contained 30 wt.% catalyst, a mass ratio of ionomer to catalyst of 0.02 and a mass ratio of alcohol to water of 2:3 (40 wt.% nPA). Before mixing, 70g of Glen Mills 5mm zirconia grinding material was added to the mixture. Samples were placed on a Fisherbrand™ digital bottle roller at 80 rpm for 26 hours. Allow the ink to sit for 20 minutes before applying. Bi2O3 ink was applied to a Qualtech automatic applicator (QPI-AFA6800) using a 1/2″ x 16″ laboratory wirewound refill (RD Specialties – 60 mil diameter) at 22°C. 5 mL of catalytic ink was applied to a 7.5 x 8 inch Sigraacet 39 BB carbon gas diffusion carrier (fuel cell storage) by rod deposition at a fixed average speed of 55 mm/sec. Transfer these coated electrodes to an oven and dry at 80 °C. The rod coating process and images of the GDE coating are shown in Figures S4a and b. An X-ray fluorescence (XRF) instrument (Fischerscope® XDV-SDD, Fischer-Technolgy Inc. USA) confirmed that the coated GDE loading was 3.0 mg Bi2O3/cm2.
       For composite membrane configurations containing anion exchange membrane (AEM) and perforated CEM. Nafion NC700 (Chemours, USA) with a nominal thickness of 15 µm was used as the CEM layer. The anodic catalyst was sprayed directly onto the FEM with an ionomer to carbon ratio of 0.83 and a coverage area of ​​25 cm2. Supported platinum with a large surface area (50 wt.% Pt/C, TEC 10E50E, TANAKA precious metal) with a loading of 0.25 mg Pt/cm2 was used as the anode catalyst. Nafion D2020 (Ion Power, USA) was used as an ionomer for the anode layer of the catalyst. CEM perforation is performed by cutting parallel lines on the CEM film at 3mm intervals. Details of the perforation process are shown in Figures S12b and c. Using X-ray computed tomography, it was confirmed that the perforation gap was 32.6 μm, as shown in Figure S12d and e. During cell assembly, a catalyst-coated perforated CEM membrane was placed on a 25 cm2 Toray paper (5 wt% PTFE treated, Fuel Cell Store, USA). An AEM membrane (PiperION, Versogen, USA) with a thickness of 25, 40 or 80 μm was placed on top of the CEM and then on the GDE cathode. The AEM membrane was cut into 7.5 × 7.5 cm pieces to cover the entire flow field and soaked overnight in 1 M potassium hydroxide solution before assembly. Both the anode and cathode use PTFE spacers that are thick enough to achieve an optimal GDE compression of 18%. Details of the battery assembly process are shown in Figure S12a.
       During testing, the assembled cell was maintained at 60 °C (30, 60, and 80 °C for temperature dependence studies) with 0.8 L/min of hydrogen gas supplied to the anode and 2 L/min of carbon dioxide supplied to the cathode. Both the anodic and cathodic air streams were humidified at 100% relative humidity and 259 kPa absolute cathodic pressure. During operation, the cathode gas stream was mixed with 1 M KOH solution at a rate of 2 mL/min to promote utilization of the cathode catalyst bed and ionic conduction. Mix a stream of anode gas with deionized water at a rate of 10 ml/min to remove formic acid at the anode. Details of the device inputs and outputs are shown in Figure S5. The cathode exhaust gas contains CO2 and generates CO and H2. The water vapor is removed through a condenser (low temperature heat exchanger at 2°C). The remaining gas will be collected for gas timing analysis. The anode flow will also pass through a condenser to separate the liquid from the gas. The wastewater will be collected in clean vials and analyzed using liquid chronometry to quantify the formic acid produced. Electrochemical tests were performed using a Garmy potentiostat (reference number 30K, Gamry, USA). Before measuring the polarization curve, the cell was conditioned 4 times in the range from 0 to 250 mA/cm2 using linear voltammetry with a scan rate of 2.5 mA/cm2. Polarization curves were obtained in galvanostatic mode with the cell held at a certain current density for 4 minutes before sampling the cathode gas and anolyte liquid.
       We use a hydrogen reference electrode in the MEA to separate the cathode and anodic potentials. The structure of the reference electrode is shown in Figure S6a. A Nafion membrane (Nafion 211, IonPower, USA) was used as an ionic bridge to connect the MEA membrane and the reference electrode. One end of the Nafion strip was connected to a 1 cm2 gas diffusion electrode (GDE) loaded with 0.25 mg Pt/cm2 (50 wt% Pt/C, TEC10E50E, TANAKA Precious Metals) sputtered onto 29BC carbon paper (Fuel Cell Store, USA). ). Special polyetheretherketone (PEEK) hardware is used to gas seal and ensure good contact between the GDE and Nafion strips, and to connect the reference electrode to the fuel cell hardware. The other end of the Nafion strip is connected to the protruding edge of the CEM battery. Figure S6b shows the cross section of the reference electrode integrated with the MEA.
       After the exhaust gas passes through the condenser and gas-liquid separator, gas samples are taken from the cathode. The collected gas was analyzed at least three times using a 4900 Micro GC (10 μm molecular sieve, Agilent). Samples were collected in inert multi-layer aluminum foil gas sample bags Supel™ (Sigma-Aldrich) for a specified period of time (30 seconds) and manually inserted into the microgas chromatograph within two hours of collection. The injection temperature was set at 110°C. Carbon monoxide (CO) and hydrogen (H2) were separated on a heated (105 °C) pressurized (28 psi) 10 m MS5A column using argon (Matheson Gas-Matheson Purity) as the carrier gas. These connections are detected using the built-in Thermal Conductivity Detector (TCD). GC chromatograms and CO and H2 calibration curves are shown in Figure S7. Liquid formic acid samples were collected from the anode for a specified time (120 seconds) and filtered using a 0.22 μm PTFE syringe filter into 2 mL vials. Liquid products in the vials were analyzed using an Agilent 1260 Infinity II bioinert high-performance liquid chromatography (HPLC) system, into which 20 μl of sample was injected through an autosampler (G5668A) with a mobile phase of 4 mM sulfuric acid (H2SO4). ) at a flow rate of 0.6 ml/min (quaternary pump G5654A). Products were separated on a heated (35°C, column oven G7116A) Aminex HPX-87H 300 × 7.8 mm (Bio-Rad) preceded by a Micro-Guard Cation H guard column. Formic acid was detected using a diode array detector (DAD). at a wavelength of 210 nm and a bandwidth of 4 nm. The HPL chromatogram and formic acid standard calibration curve are shown in Figure S7.
       The gas products (CO and H2) FE are calculated using the following equation, and the total moles of gas are calculated using the ideal gas equation:
       Among them: \({n}_{i}\): the number of electrons in an electrochemical reaction. \(F\): Faraday’s constant. \({C}_{i}\): HPLC liquid product concentration. \(V\): volume of liquid sample collected over a fixed time t. \(j\): current density. \(A\): Geometric area of ​​the electrode (25 cm2). \(t\): sampling time period. \(P\): absolute pressure. \({x}_{i}\): Mole percent of gas determined by GC. \(R\): gas constant. \(T\): temperature.
       The concentration of anodic cations was quantified using inductively coupled plasma atomic emission spectroscopy (ICP-OES). Cations that may leach or diffuse into the anode include Ti, Pt, Bi and K. With the exception of K, all other cations were below the detection limit. Form ions in the solution leaving the anode to pair with protons or other cations. Therefore, the purity of formic acid can be calculated as
       Formate/FA production represents the amount of FA produced per kWh of electricity consumed using a particular MEA configuration, in mol/kWh. It is calculated based on current density, cell voltage and Faraday efficiency under specific operating conditions.
       Calculate the amount of formic acid oxidized at the anode based on the overall mass balance. Three competing reactions occur at the cathode: hydrogen evolution, reduction of CO2 to CO, and reduction of CO2 to formic acid. Because we have formic acid oxidation process in Anton, formic acid FE can be divided into two parts: formic acid collection and formic acid oxidation. The overall mass balance can be written as:
       We used GC to quantify the amounts of formic acid, hydrogen, and CO collected by HPLC. It should be noted that most of the formic acid was collected from the anode using the setup shown in Supplementary Figure S5. The amount of formate collected from the cathode chamber is insignificant, approximately two orders of magnitude less, and amounts to less than 0.5% of the total amount of SC.
       The continuous transport model used here is based on previous work on similar systems34. A coupled system of Poisson-Nerst-Planck (PNP) equations is used to determine water concentration and electrostatic potential in electronically and ionicly conducting phases. A detailed overview of the underlying equations and model geometry is given in the SI.
       This system determines the concentration of eight aqueous substances (\({{{{{{\rm{C}}}}}}}{{{{{{\rm{O}}}}}}}}_{2 \left ({{{{{{\rm{aq}}}}}}\right)}\), \({{{{{{\rm{H}}}}}}}}^{+ }\ ), \ ({{{{{\rm{O}}}}}}{{{{{{\rm{H}}}}}}^{-}\), \({{{ {{{ \rm{ HCO}}}}}}}_{3}^{-}\), \({{{{{{\rm{CO}}}}}}_{3}^{ 2-} \ ),\ ({{{{{\rm{HCOOH}}}}}}\), \({{{{{{\rm{HCOO}}}}}}}}^{- }\) and \({{{ {{{\rm{K}}}}}}^{+}\)), electrostatic potential in the ionic conducting phase (\({\phi }_{I}\ )) and anodic and cathodic electron conductivity. Electrostatic potentials in phases (\({\phi }_{A}\) and \({\phi }_{C}\) respectively). Instead, neither local electrical neutrality nor charge distribution functions are realized, the space charge region is solved directly using Poisson’s equation; This approach allows us to directly model Donnan repulsion effects at the CEM|AEM, CEM|Pore, and AEM|Pore interfaces. In addition, porous electrode theory (PET) is used to describe the charge transport in the anodic and cathodic layers of the catalyst. To the best of the authors’ knowledge, this work represents the first application of PET in systems with multiple space charge regions.
       GDE BOT and EOT cathode samples were tested using a Zeiss Xradia 800 Ultra with an 8.0 keV X-ray source, absorption and wide field modes, and image fusion1. 901 images were collected from -90° to 90° with an exposure time of 50 seconds. Reconstruction was performed using a back projection filter with a voxel size of 64 nm. Analysis of segmentation and particle size distribution was carried out using specially written code.
       Electron microscopic characterization involves embedding the test MEAs in epoxy resin in preparation for ultrathin sectioning with a diamond knife. The cross section of each MEA was cut to a thickness of 50 to 75 nm. A Talos F200X transmission electron microscope (Thermo Fisher Scientific) was used for scanning transmission electron microscopy (STEM) and energy-dispersive X-ray spectroscopy (EDS) measurements. The microscope is equipped with an EDS Super-X system with 4 windowless SDD detectors and operates at 200 kV.
       Powder X-ray diffraction patterns (PXRD) were obtained on a Bruker Advance D8 powder X-ray diffractometer with Ni-filtered Cu Kα radiation operating at 40 kV and 40 mA. The scanning range is from 10° to 60°, the step size is 0.005°, and the data acquisition speed is 1 second per step.
       The RAS spectrum at the edge of the Bi2O3 Bi L3 catalyst was measured as a function of potential using a homemade cell. Bi2O3 catalytic ionomer ink was prepared using 26.1 mg Bi2O3 mixed with 156.3 μL ionomer solution (6.68%) and neutralized with 1 M KOH, water (157 μL) and isopropyl alcohol (104 μL) to obtain ionomer ink. The catalyst coefficient is 0.4. The ink was applied to graphene sheets in rectangular spots (10×4 mm) until the Bi2O3 catalyst loading reached 0.5 mg/cm2. The rest of the graphene sheet is coated with Kapton to isolate these areas from the electrolyte. The catalyst-coated graphene sheet was inserted between two PTFEs and secured to the cell body (PEEK) with screws, Figure S8. Hg/HgO (1 M NaOH) served as the reference electrode, and carbon paper served as the counter electrode. The Hg/HgO reference electrode was calibrated using a platinum wire immersed in hydrogen-saturated 0.1 M KOH to convert all measured potentials to a reversible hydrogen electrode (RHE) scale. XRD spectra were obtained by monitoring the potential of a Bi2O3/graphene sheet working electrode immersed in 0.1 M KOH, heated to 30 °C. The electrolyte circulates in the battery, with the electrolyte inlet at the bottom of the cell and the outlet at the top to ensure that the electrolyte contacts the catalyst layer when bubbles form. A CH Instruments 760e potentiostat was used to control the working electrode potential. The potential sequence was an open circuit potential: -100, -200, -300, -400, -500, -800, -850, -900, -1000, -1100, -1500 and +700 mV depending on RHE. All iR potentials have been adjusted.
       Bi L3 edge (~13424 eV for Bi metal) X-ray absorption fine structure (XAFS) spectroscopy was performed on channel 10-ID, Advanced Photon Source (APS), Argonne National Fluorescence Laboratory. National Model Measurement Laboratory. A two-crystal Si(111) monochromator cooled with liquid nitrogen was used to tune the X-ray energy, and a rhodium-coated mirror was used to attenuate the harmonic content. Scan energies were varied from 13200 to 14400 eV, and fluorescence was measured using a 5 × 5 silicon PIN diode array without filters or Soller slits. The zero crossing energy of the second derivative is calibrated at 13271.90 eV through the L2 edge of the Pt foil. Due to the thickness of the electrochemical cell, it was not possible to simultaneously measure the spectrum of the reference standard. Thus, the calculated scan-to-scan change in incident X-ray energy is ±0.015 eV based on repeated measurements throughout the experiment. The thickness of the Bi2O3 layer leads to a certain degree of self-absorption of fluorescence; the electrodes maintain a fixed orientation relative to the incident beam and detector, making all scans virtually identical. Near-field XAFS spectrum was used to determine the oxidation state and chemical form of bismuth by comparison with the XANES region of Bi and Bi2O3 standards using the linear combination fitting algorithm of Athena software (version 0.9.26). by code IFEFFIT 44.
       The data supporting the figures in this article and other conclusions of this study are available from the corresponding author on reasonable request.
       Crandall B.S., Brix T., Weber R.S. and Jiao F. Techno-economic assessment of green media supply chains H2. Energy Fuels 37, 1441–1450 (2023).
       Younas M, Rezakazemi M, Arbab MS, Shah J and Rehman V. Green hydrogen storage and delivery: dehydrogenation of formic acid using highly active homogeneous and heterogeneous catalysts. internationality. J. Gidrog. Energy 47, 11694–11724 (2022).
       Nie, R. et al. Recent progress in catalytic transfer hydrogenation of formic acid over heterogeneous transition metal catalysts. AKS catalogue. 11, 1071–1095 (2021).
       Rahimi, A., Ulbrich, A., Kuhn, J. J., and Stahl, S. S. Formic acid-induced depolymerization of oxidized lignin to aromatic compounds. Nature 515, 249–252 (2014).
       Schuler E. et al. Formic acid serves as a key intermediate for CO 2 utilization. green. Chemical. 24, 8227–8258 (2022).
       Zhou, H. et al. Rapid non-destructive fractionation (≤15 min) of biomass using flow-through formic acid for overall enhancement of carbohydrate and lignin content. Chemistry and Chemistry 12, 1213–1221 (2019).
       Calvi, C. H. et al. Enhanced growth of Cupriavidus necator H16 on formate using adaptive laboratory evolutionary information engineering. Metabolites. engineer. 75, 78–90 (2023).
       Ishai, O. and Lindner, S.N. Gonzalez de la Cruz, J., Tenenboim, H. and Bar-Even, A. Bioeconomics of formates. current. Opinion. Chemical. biology. 35, 1–9 (2016).