top of page

Scale-up of electrochemical flow cell towards industrial CO₂ reduction to potassium formate

Dr. Arthur G. Fink, Dr. Fabiola Navarro-Pardo, Prof. Jason R. Tavares, Dr. Ulrich Legrand

Graphical Abstract

CO₂ electroreduction enables the production of valuable chemicals using captured CO₂ and clean electricity. We report the scale-up of a CO₂ reduction reaction electrochemical cell to industrial size. While having commercially relevant performance, further challenges are discussed in the present study.


The CO₂ electroreduction reaction (CO₂RR) presents a pathway to decarbonize the manufacturing industry by using clean electricity and CO₂ as feedstocks instead of relying on fossil fuels. Although known for over 100 years, this technology has yet only been developed at bench-top scale (1–100 cm²). In this manuscript, we report CO₂ electroreduction to potassium formate (HCOOK) in a stack of two 1526-cm² three-compartments electrochemical cells with gas diffusion electrode (GDE) cathodes. In this stack, we achieved over 60 % current selectivity towards HCOOK at current densities of 200 mA cm−2 and under cell voltages of ~4.0 V. We reached these performance metrics by tuning electrolyte composition and cell architecture. We also show that a minimum of +10-14 kPa of pressure difference must be applied between gaseous and catholyte compartments to enable the CO₂RR to take place. We emphasize the challenges associated with scaling-up a CO₂ electrochemical cell, specifically by demonstrating that optimal operation parameters are strongly correlated to cell architecture. This study demonstrates the feasibility of developing CO₂RR electrochemical cells to an industrial scale.


Global carbon dioxide (CO₂) emissions reached 36.8 Gt in 2022 and it is likely that the +1.5 °C objective decided upon during the Paris Agreement of 2015 will be exceeded over the 2023–2027 period. While common effort is rising to fight climate change and many available technologies are already deployed to considerably reduce greenhouse gas (GHG) emissions, many chemicals are still manufactured with heavy usage of fossil fuels serving as feedstocks, energy sources, and for transportation. One such chemical is potassium formate (HCOOK), a compound produced for its use as a green non-corrosive de-icing agent and drilling fluid. HCOOK had an estimated market of 716 M US$ in 2022 (expected CAGR of 4.2 % from 2022 to 2029). This chemical is synthesized mainly in the Middle-East and Asia indirectly through the reaction of carbon monoxide (CO) and methanol (MeOH), both sourced from fossil fuels, at high pressures and moderate temperatures (4500 kPa and 80 °C), to produce formic acid (HCOOH). HCOOH reacts spontaneously with potassium hydroxide (KOH) to yield HCOOK. The overall carbon footprint of HCOOK production alone is estimated to be around 2.67 CO₂ ton per ton of HCOOK (ton CO₂eHCOOK) when we consider only the production of formic acid from fossil sources (1.41 ton CO₂eHCOOK), and the production of KOH (1.26 ton CO₂eHCOOK). This situation offers an opportunity to decarbonize the potassium formate industry.

Potassium formate can alternatively be produced directly through the CO₂ electroreduction reaction (CO₂RR, Eq. 1 powered by renewable electricity.

This approach enables the use of renewable clean energy (e. g., hydro, solar) to drive the reaction, local production to prevent long-distance transportation, and a circular economy of CO₂ used here as a feedstock rather than a product. At an industrial scale, the implementation of CO2RR would therefore ultimately reduce GHG associated with HCOOK manufacture. Over the last decade, advances in the field of CO₂ reduction have been tremendous: selective reactions typically performed in batch cells (H-cells) at 1–10 mA cm−2 since the mid 1980 s were successfully translated into flow cells to reach hundreds of mA cm−2 (Figure 1). Selectivity towards a specific product is achieved by tuning the heterogeneous metal catalyst employed: (i) formate and formic acid are generated with lead, tin, bismuth, or indium-based catalysts; (ii) CO is produced over silver or gold catalysts; and (iii) copper-based catalysts can produce a variety of single and multi-carbon products (e. g., MeOH, methane, ethanol, ethylene, etc.). Recent techno-economic assessment studies forecast that to be economically viable at the industrial scale, CO2RR to formate should operate ≥100 mA cm−2, with >80 % of current selectivity (Faradaic efficiency, FE), with a cell voltage <3 V, while retaining these performance metrics for thousands of hours. Although most of these performance metrics have been fulfilled at the bench-top scale, the development of an electrolyzer at the industrial scale remains unreported.

Figure 1

Pathway towards industrialization of CO2RR to formate in three-compartments cells containing GDE cathodes. As of 2023, most of the literature available experimented using batch cells (H-cells) where CO₂ is bubbled in the catholyte. These cells typically have electrodes of ~1 cm² and are operated at around 10 mA cm−2 due to the solubility limit of CO₂ in aqueous solvents and the distance between the electrodes. Over the last decade, literature also demonstrated the use of a single flow cell capable of reaching ~100 cm² in size and operate at ~200 mA cm−2. Such a benchtop cell, however, can only produce ~25 g of potassium formate per hour with a FE of 80 %. Industrialization of CO2RR towards formate will likely require larger cells (~2000 cm²) in stacks. These larger cells would each be capable of producing ~0.5 kg of HCOOK per hour.

The largest electrolyzers to date are roughly sized at ~100 cm², which would only generate ~220 kg of HCOOK per year (0.025 kgHCOOK h−1) when considering a 80 % FE at 200 mA cm−2(Figure 1). The work of Li et al. in 2007 makes an exception with a 320 cm² two-compartment cell without a gas-diffusion electrode. While this study reports good selectivity (~60 % Faradaic efficiency at 300 mA cm−2), their cell is fed with pressurized gaseous CO₂ (~420 kPa(abs)) in the catholyte. The development of larger reactors (~2000 cm²) is unexplored territory probably due to its elevated cost of construction and operation. Moreover, the current stream of research mainly focuses on developing high-performance catalysts, while neglecting the cost and scalability of these catalysts. Researchers also widely still test these catalysts in the controlled conditions of an H-cell (98 % of papers on CO2RR published between 2007 and 2017) that are not representative of an industrial flow reactor. Scaling-up requires specific size-dependant considerations such as: mass transfer of reactants and products; current distribution; material availability; and production capabilities. To the best of our knowledge, this work has not been tackled yet for cells containing gas diffusion electrode (GDE) cathodes.

In this paper, we report the first pre-industrial 1526 cm² electrochemical cell, a 60-fold increase from the largest three-compartment electrolyzer reported to date. We achieved operation at 200 mA cm−2 at cell voltages as low as 3.86 V by tuning the electrolyte composition (initial and during electrolysis), the pressure difference between catholyte and CO₂, and the electrochemical cell architecture. We show that these parameters under-reported in the literature play a crucial role in the electrochemical production of HCOOK at scale. Based on our findings, we demonstrate the stable production of HCOOK in a stack of two 1526 cm² cells with a FE>60 % stable over 1-hour batches corresponding to the carbonation of the electrolyte. These performance metrics represents the state-of-the-art in terms of HCOOK production from CO₂ electrolysis with a GDE cathode. We estimate that under continuous operation, this stack of electrochemical cells would be capable of producing >5 tons of HCOOK per year, laying a clear path towards industrial-scale production of valuable chemicals through the CO2RR.

Results and Discussion

All experiments reported in this study were performed in a custom-built 3-compartment electrochemical cell (Figures 2a,b, S1). The anolyte compartment is fed with 18.9 L of a 4 M KOH solution (diluted for stock 45 w/w% solution, Quadra Chemicals) at 5.7 liter per minute (LPM) and the catholyte compartment is similarly fed at 5.7 LPM with 18.9 L of a 1 to 4 M KOH solution (Figure 2c). Both anolyte and catholyte are continuously recirculated. A cathode current collector and a gas diffusion electrode (GDE) composed of a carbon cloth, a microporous layer, a Sn-based catalyst (proprietary and supplied by Electro Carbon Inc.), and a hydrophobic layer, are placed between the catholyte and the CO₂ compartments. The catalyst is spray-coated on the microporous layer to a loading of ~0.9 mg cm−2. Dry gaseous CO₂ (99.99 %, <20 ppm H2O, Air Liquide) is delivered to the CO₂ compartment at 14.2-33 standard liter per minute (SLPM). Each layer is gas and liquid sealed with silicon gaskets. The multilayered 1526 cm² GDE (corresponding to the geometric exposed area of catalyst) serves to support the Sn-based electrocatalyst and manage the transport of water, electrons, and CO₂ reactants to the catalytic sites. For all experiments, a constant current of 305 A is applied between the cathode current collector and the anode (where the oxygen evolution reaction, OER, evolves; Eq. 2 by means of a 0–500 A and 240 V DC power supply (Volteq). Anolyte and catholyte were sampled every 15 min and analyzed by permanganate titrations and high-performance liquid chromatography (HPLC) to determine HCOOK concentrations, while acid-base titrations were performed to determine hydroxides, carbonates, and bicarbonates concentrations.

Figure 2

(a) Exploded view of the cell architecture used in this study. The anolyte compartment is fed with a 4 M KOH solution and is separated from the cathode compartment by a microporous separator. The catholyte compartment is fed with a 2 M KOH solution and is separated from the CO₂ compartment with a gas diffusion electrode (GDE) cathode coated with the CO₂ electrocatalyst. A current (305 A) is applied between the anode and the cathode GDE to produce HCOOK that exits the cell in the catholyte. (b) A picture of the 1526 cm² cell used in this study. (c) Flow and circuits diagram of the electrochemical cell used in this study.

Lowering cell voltage in a scaled-up architecture

CO₂ electro-reduction is an energy intensive process due to the high stability of the CO₂ molecule. Therefore, the reduction of the energy consumption related to this process is one of the key factors for commercialization. We identified that a fixed current of 305 A was required to achieve the HCOOK production rates we target. This chosen current (200 mA cm−2, defined by geometric surface area) correlates well with many recent techno-economic studies that suggest that 100–200 mA cm−2 is an industrial current density target.13, 39 Since applied current is hence fixed, the energy consumed will be reduced only through the modulation of the cell voltage required to deliver the 305 A, while maintaining the same FE. In this study, we evaluated and followed three parameters that are susceptible to affect the cell voltage: (i) KOH concentration in the catholyte; (ii) catholyte compartment thickness; and (iii) chemical composition of the catholyte throughout electrolysis.

We first modulated the KOH catholyte concentrations to lower the cell voltage (i. e., the voltage applied between the anode and cathode). We tested 3 different KOH initial concentrations of the catholyte ([KOH]catholyte; 1, 2, and 4 M) and reported the cell voltages read over the first 30 min of electrolysis at 305 A (Figure 3a). We observed that the cell voltage at time=0 min dropped by 2.07 V when we increased the [KOH]catholyte from 1 M (5.85 V) to 4 M (3.78 V). The voltage for [KOH]catholyte of 2 M (4.74 V) was only ~0.98 V higher than with 4 M. Yet, calculations based on the Ohmic drop caused by the loss of electrolyte conductivity only predicts a potential difference of ~0.10 V (see Methods for calculation; Table S1). We note that we measure the total cell voltage, hence parameters other than catholyte Ohmic drop will influence the cell voltage and explain this disparity (e. g., voltage drop through membrane, cathode and anode overpotentials, diffusion overpotential, electrolyte conductivity).50 In particular, the voltage drop through the membrane is probably causing the ~1 V voltage difference between each concentration tested. Indeed, the FAAM-15 microporous separator used in this study is advised for KOH concentrations of 6 to 12 M, a range far away the 1 M KOH concentration tested here.51, 52 Moreover, these microporous separators are incompatible with carbonates and bicarbonates. The 1 M KOH was almost fully carbonated in 15 min (Figure S2), which explains the increase in voltage observed in Fig 3a. While 4 M KOH is the best candidate in terms of cell voltage, and thus input energy required, this KOH concentration also means twice the amount of K2CO3 that needs to be separated from the HCOOK produced in an industrial setup. The saved price of electricity between 2 M and 4 M KOH is minimal (0.02 $CAD reduction per hour of electrolysis at 305 A with Hydro Quebec 2023 rate of 0.11 $CAD/kWh). Therefore, we concluded that 2 M KOH was the best tested condition for decreasing energy consumption. All experiments presented below are performed at this [KOH]catholyte.

Figure 3

(a) Cell voltage as a function of time in the catholyte during electrolysis at 305 A total current for the three [KOH] tested (1 M in orange, 2 M in green, and 4 M in blue). (b) Cell voltage as a function of distance between the microporous separator and the cathode GDE surface. Values are averaged over the first 30 min of electrolysis at 305 A total current. (c) Cell voltage as a function of time during electrolysis at 305 A with a 2 M KOH initial catholyte and with a membrane-GDE distance of 2.6 mm. (d) Concentrations of KOH (grey), K₂CO (blue), KHCO (orange), and HCOOK (green) during electrolysis at 305 A with a 2 M KOH initial catholyte and with a membrane-GDE distance of 2.6 mm.

We then modulated the distance between the microporous separator and the GDE (membrane-GDE distance) from 7.7 to 2.6 mm and averaged the cell voltage values over the first 30 min of electrolysis at 305 A (Figure 3b). We observed that shortening the membrane-GDE distance three-fold enabled a reduction of 0.27 V (4.41±0.11 V and 4.14±0.11 V at 7.7 and 2.6 mm, respectively). We found that the predicted Ohmic drop change between the two distances is 0.20 V (Table S1). This value falls within the error range of our experimental observations. Indeed, the nature of the electrolytes remained similar in this set of experiments compared to the experiments with different [KOH]catholyte. We concluded that modulating the distance between the microporous separator and the GDE does not have a significant effect on cell voltage in our electrochemical cell architecture. Yet, the thinner configuration offers a practical advantage such as providing a more compact system, hence minimizing the system footprint.

Lastly, we followed the evolution of the voltage during electrolysis (2 M KOH initial catholyte with a membrane-GDE distance of 2.6-mm) at 305 A (Figure 3c). We observed that the voltage slowly increased over the first 90 min from 3.70 V to 4.44 V, then changed to a fast increase regime to reach 7.58 V after 165 min of electrolysis. We performed acid-base analyses on the aliquots taken during electrolysis and found that KOH had completely been carbonated into K₂CO after 90 min of electrolysis (Eq. 3), and that 0.9 M of KHCO were in the electrolyte after 165 min of electrolysis (Figure 3d). The conversion of hydroxides species into carbonates and bicarbonates is caused by reaction with electrochemically unreacted CO₂ that is present in the alkaline electrolyte from diffusion through the GDE (Eqs. 4–5).53 The increase of K+ in the catholyte during electrolysis comes from diffusion of the anolyte through the microporous separator. We note that the sharper rise in cell voltage comes with the emergence of KHCO3 (around 100 min), while the consumption of KOH into K2CO leads to a less drastic change. The change in electrolyte conductivity (40.5, 16.6, and 12.5 S/m for 2 M KOH, 1 M K₂CO, and 2 M KHCO, respectively) cannot solely explain the voltage change recorded for solutions containing KHCO compared to the ones containing mostly K₂CO (Table S1). The microporous separator used in this study is not advised to use in the presence of bicarbonates. We envision that the sharp rise in voltage during emergence of bicarbonate species is due to incompatibility between the microporous separator (FAAM-15) with the lower pH associated with the formation of bicarbonates.52 Indeed, these microporous separators do not have charged functional groups that promote surface-hopping of hydroxides, but are porous membrane separators that relies on wicking for hydroxide transfer.54 This diffusion is promoted in high alkaline environments.55 We conclude that in this reactor architecture, the cell voltage remains low in the presence of KOH and K₂CO species (<4.5 V), but high cell voltages (>6 V) arise in the presence of bicarbonate species. This situation makes KHCO3-containing electrolytes incompatible with HCOOK production with FAAM membranes.

Define pressures to maximize Faradaic efficiency for HCOOK

As described in the previous paragraph, the production of carbonated species in the catholyte is caused by the diffusion of electrochemically unreacted CO₂ to the catholyte through the porous GDE. The diffusion of CO₂ to the catalyst layer (facing the catholyte) is crucial because it enables the CO2RR reaction to take place through the availability of CO₂ and proton donors (e. g., water) at the catalyst surface. Due to the porous nature of the GDE, the catholyte and CO₂ pressures (Pcatholyte and PCO2, respectively) must be carefully controlled.56-58 In the case that Pcatholyte >PCO2 (Figure 4a Case 1), the catholyte permeates through the GDE to the CO₂ compartement (flow-through configuration).57 Besides not delivering the CO₂ reactant to the catalyst, two issues are common to this situation: (i) catholyte droplets can hang on the gas side of the GDE; and (ii) the catholyte can dry which precipitates carbonates onto the GDE. Both phenomena actively block transport of CO₂ to the electrocatalyst, hence prevent electroreduction of CO₂. Moreover, the second phenomenon (precipitate formation) must be avoided due to its non-reversible nature as no liquid is passed through the CO₂ compartment to dissolve the solids. In the case that Pcatholyte = PCO2 (Figure 4a Case 2), the CO2RR is limited by the slow diffusion of CO₂ to the catalyst layer, which is incompatible with the high current densities applied in our electrochemical cell and promotes the competitive hydrogen evolution reaction (HER; Eq. 6). Lastly, in the case that Pcatholyte < PCO2 (Figure 4a Case 3), the situation is ideal for electroreduction of CO₂. Nonetheless, we have seen that bicarbonates are to be avoided to maintain a low cell voltage. The formation of carbonated species (i. e., HCO3− and CO₃²−) arises when excess CO₂ goes through the catalyst layer electrochemically unreacted. Therefore, the range of pressure difference between Pcatholyte and PCO2 giving the highest Faradaic efficiency for HCOOK (FEHCOOK) must be determined for a given electrochemical cell architecture and operation conditions. This pressure difference must be neither too large to prevent formation of bicarbonates in the catholyte, nor too low to still ensure CO₂ permeation to the catalyst, and as homogenous as possible over the entire surface area.

Figure 4

(a) Schematic showing the different pressure regimes for the pressures in the CO₂ compartement and the catholyte compartment (i. e., PCO2 < Pcatholyte, PCO2=Pcatholyte, and PCO2 > Pcatholyte). (b) Faradaic efficiencies for HCOOK (FEHCOOK) as a function of pressure difference between the CO₂ compartment and catholyte compartment during electrolysis at 305 A with a 2 M KOH initial catholyte and with a membrane-GDE distance of 7.7 mm.

To address this required precise pressure control, many researchers avoided a three-compartment cell and rather fed CO₂ dissolved in the catholyte in a two-compartment cell, hence bypassing the gaseous CO₂ compartment. However, the saturated CO₂ concentration in aqueous electrolytes at ambient pressure and temperature is only 0.033 M, a value too low to enable the high yield production of HCOOK. Moreover, there is no three-compartment CO₂ electrochemical cell producing formate as tall (40 cm) as the one we employ in this study. Indeed, in 2023, the average CO₂ electrolyzer targeting formate is 1–100 cm² (about 1–9 cm tall). That is, real-world considerations such as the pressure difference within the electrolyte compartment due to hydrostatic pressure for such benchtop reactors, were not considered relevant. Yet, this parameter could drastically influence the gradient of pressures within our cell. If we consider only a hydrostatic pressure difference between the inlet (at the bottom) and outlet (at the top) of our cell (i. e., a distance of 40 cm), we obtain a pressure difference of 4 kPa. This pressure difference would need to be taken in account to satisfy Pcatholyte<PCO2 throughout the entire electrochemical cell.

We set out to investigate the impact of pressure difference between the CO₂ and catholyte compartments (∆PCO2–Catholyte = 3 – 14 kPa ; measured at the outlets of the cell) on the Faradaic efficiencies for HCOOK during electrolysis at 305 A (Figure 4b). No formate was produced for ∆PCO2–Catholyte ≤ 10 kPa. The FEHCOOK reached ~40 % for a ∆PCO2–Catholyte of 11 kPa, a difference of pressure associated with the onset of CO₂ electroreduction. Finally, the FEHCOOK values plateaued around 55 % for ∆PCO2–Catholyte ≥ 12 kPa. We did not observe flooding for all ∆PCO2–Catholyt tested. This data teaches us that the minimal ∆PCO2–Catholyt to produce formate is ~7 kPa greater than the hydrostatic pressure. Indeed, pressure drops are also caused by the changes in pipe/compartment architecture and by the flow pattern in the catholyte compartment. It is also worth mentioning that the ∆PCO2–Catholyt values reported herein are from pressures that were measured on the outlet pipes exiting the electrochemical cell. Therefore, the actual local pressure in the electrolyzer compartments might differ from our readings. For these tests, we also measured the inlet pressures of the catholyte and the CO₂ piping just before the cell. We found a pressure drop between outlet (146 kPa(abs)) and inlet (162 kPa(abs)) of catholyte of 16 kPa. On the CO₂ side, we found no pressure drop between outlet and inlet (i. e., 160 kPa(abs) for both). We conducted the same set of experiment with a membrane-GDE distance of 2.6 mm and observed that the onset of CO₂ electroreduction was for ∆PCO2–Catholyt values ~14 kPa and FEHCOOK values plateaued above 17 kPa (Figure S3). For this lower membrane-GDE distance, we found a pressure drop between outlet (149 kPa(abs)) and inlet (168 kPa(abs)) of catholyte of 19 kPa, and again no pressure drop between outlet and inlet (i. e., 167 kPa(abs) for both) for the CO₂. The increase in ∆PCO2–Catholyt required to onset CO2RR with the thinner catholyte compartment can be rationalized by the increase in pressure drop caused by the higher velocities of catholyte. Indeed, the flow rate is kept constant for both cell architectures (5.7 LPM), while the volume of the catholyte compartment decreased from 1463 to 494 cm3 for the membrane-GDE distances of 7.7 and 2.6 mm, respectively. We conclude that ∆PCO2–Catholyt plays a key role in CO₂ electroreduction efficiency and must be adjusted to the cell architecture. The optimal conditions for high FEHCOOK values reported here are only valid for this cell architecture and the set operation parameters (e. g., flow rate, GDE composition).

Operation of the electrochemical cell

In the previous sections, we saw that: (i) a catholyte composed of 2 M KOH provided acceptable cell voltages; (ii) thickness of the cell had only a moderate impact on cell voltage; (iii) cell voltage rapidly increased due to the incompatibility between our microporous separator and the bicarbonates formed after 90 min of electrolysis; and (iv) that a minimum ∆PCO2–Catholyt needs to be applied to produce HCOOK. We therefore performed electrolysis at 305 A (0.2 A cm−2) for 90 min with a membrane-GDE distance of 7.7 mm, at a ∆PCO2–Catholyt of 14 kPa, and with 2 M KOH catholyte and recorded the FEHCOOK values (Figure 5a). We observed that the FEHCOOK values were stable at (63.1±3.0)% over the 90 min with a slight decrease (about 4 %) after 45 min. We also observed that some HCOOK was found in the anolyte after 75 min. This observation is consistent with the non-selective microporous separator employed here. The use of ion-exchange membranes that are stable in alkaline media yet preventing formate cross-over from catholyte to anolyte will be required in further development. We also tested the electrolysis with a membrane-GDE distance of 2.6 mm at a ∆PCO2–Catholyt of 17 kPa and observed a stable FEHCOOK of (51.5±3.2)% (Figure S4). No formate was found in the anolyte for this latter test. We thus conclude that high efficiencies can be retained for only about 90 min. In an industrial setup, batches of electrolytes could be circulated before extracting HCOOK. Therefore, the 90-min limitation we present here would not be an issue in industrialization as long as the catalyst does not degrade.

Figure 5

(a) Faradaic efficiencies for HCOOK (FEHCOOK; in green for HCOOK quantified in the catholyte, in orange for HCOOK quantified in the anolyte, and in black for the total amount of HCOOK quantified) as a function of time during electrolysis at 305 A with a 2 M KOH initial catholyte, ΔPCO2–Catholyte of 14 kPa, and with a membrane-GDE distance of 7.7 mm. (b) FEHCOOK as a function of time during electrolysis in a stack of two 1526–cm² cells each operated at 305 A, with a 2 M KOH initial catholyte, ΔPCO2–Catholyte of 14 kPa, and with membrane-GDE distance of 7.7 mm.

In an industrial HCOOK production plant, a stack of electrochemical cells, processing batches of electrolytes or running continuously before separating HCOOK, will likely be the system of choice. We therefore set out to build a stack of two 1526 cm² cells (Figs. S5-S6) to test if our system retains its performance when operated in stacks. We flowed the electrolytes in parallel between each respective compartment and supplied the electrical current in series (305 A applied). We kept the flow rates and electrolyte volumes constant and employed a KOH catholyte concentration of 4 M to compensate for the formation of carbonates and bicarbonates. We observed a FEHCOOK of (61.8±4.2)% stable over the hour of electrolysis (Figure 5b), while the applied voltage was 8.12±0.19 V over the first 30 min of electrolysis (4.29±0.08 V and 3.83±0.12 V for each respective cell). Over the hour of experiment, we reached 0.28 M of HCOOK in the catholyte. In a commercial unit, higher HCOOK concentrations could be achieved to facilitate purification of HCOOK by, for instance, lowering the ratio of electrolyte volume to electrode area. Some studies report HCOOK concentrations over 1 M for longer electrolysis duration. The experiment in the stack of 2 cells gave a total energy spent of 0.45 kWh mol−1 formate. These results demonstrate that the electrolytic performance of a single cell can be retained in a stack of two cells. Moreover, these results are aligned with the state-of-the-art of CO2RR to formate with GDE cathodes (Table 1). We note that out of the five studies reporting >10 cm² of geometric cathode area, only the work of Chen et al. (25 cm²) operated at industrially-relevant current densities (≥200 mA cm−2) at near-ambient pressures and temperatures.

Table 1 State-of-the-art reports of CO₂ electroreduction to formate or formic acid in flow cells.

[a] NRS stands for nanorod@sheet. [b] NPs stands for nanoparticles. [c] Unknown flow cell architecture. [d] This study operated at high pressures of 5000 kPa. [e] This study operated at high pressures of 420 kPa(abs).


In conclusion, we reported herein the largest electrochemical cell for CO₂ reduction towards formate known to date in the available scientific literature. We demonstrated that at this scale, electrolyte composition, pressure difference between catholyte and CO₂ compartments, and cell architecture were key parameters to modulate both energy required (through voltage) and electrical yield (Faradaic efficiency) at values in agreement with industrial targets. Based on our findings, we were able to achieve FEHCOOK values >60 % with cell voltage <3.9 V during electrolysis at 200 mA cm−2. Although these values are shy of the 80 % Faradaic efficiency that techno-economic assessment report for industrialization, we believe that these values are attainable with further research (e. g., catalyst development, modelling of electrolyte flow). In this direction, we also demonstrated a stack of two 1526 cm² cells that retained the voltage and efficiencies of a single cell. This situation will hopefully promote work towards further understanding of the long-term operation and stability of these large GDE cells, but also towards a stack of more cells and, potentially, industrialization of CO₂ reduction towards formate.

Experimental Section


Starch (from potato) and potassium iodide (KI; 99 % ReagentPlus) were purchased from Sigma Aldrich. Potassium permanganate (KMnO3, 99.3 %, Reagent Grade) was supplied by Blackbook supply. Sulfuric acid (H2SO4, 95.0-98.0 %, ACS reagent) was supplied by J. T. Baker. Concentrated KOH stock solution (45 % w/w in H2O) was supplied by Quadra Chemicals. CO₂ gas (99.99 %, <20 ppm H2O) was supplied by Air Liquide. All chemicals are employed as-received, without further purification. The gas diffusion layer (GDL, W1S1011) was supplied by Fuel Cell Store. The FAAM-15 microporous separator was supplied by Fumatech.

Cell architecture

The electrochemical cell is a 3-compartment custom-made cell (Figure 2a) where the cathode catalyst is placed on a porous gas diffusion layer (GDL) at the interface between the gaseous CO₂ and the liquid catholyte. The compartment pieces are made of high-density polyethylene (HDPE). Two stainless steel endplates (12.7-mm thick) sandwich the anolyte, the catholyte, and the CO₂ compartments. A Ni-based anode (proprietary and supplied by Electro Carbon Inc.), where the oxygen evolution reaction (OER; Eq. 2) takes place, and a microporous separator is placed between the anolyte and catholyte compartments. The two steel endplates are placed on either side of the cell to close it leak proof and ensure the cell mechanic integrity during operation. A machined tin-covered copper piece ensures the electric current delivery to the GDL. This cathode current collector has 16 openings of 9.0 by 10.6 cm with bands 1.5 cm wide (Figure S1). The openings expose 1526 cm² of GDE. To ensure that the electrodes and microporous separator remain in place during the cell operation, supports HDPE plastic mesh layers with wide opening (polyethylene plastic mesh 4×4, McMaster Carr) are placed in the three compartments. These meshes also serve as turbulence promoters and constitute the flow-fields of each compartment. Silicone gaskets are placed between the different components to seal the electrochemical cell. The geometries of current collectors and compartments are described in Figure S1.

Electrolyzer setup

The electrochemical cell setup employed in this study is presented in Figure 2b. CO₂ and electrolytes are respectively contained in a gas cylinder and plastic drums. Electrolytes are fed by means of 93 W circulation pumps. We read the pressures of each of the three compartments by placing a pressure gauge on the tubing of each CO₂, catholyte, and anolyte outlet streams. The pressure in each compartment is only controlled by a back-pressure valve placed right after the pressure gauge. Absolute pressures at the outlet were between 136 and 149 kPa(abs) for the anolyte, between 136 and 149 kPa(abs) for the catholyte, and between 149 and 167 kPa(abs) for the CO₂. The CO₂ exiting the cell is sent to an exhaust while the anolyte and catholyte are injected back to their respective storage container. Both electrolyte containers are ventilated to avoid gas (CO₂, O2, H2) accumulation. Catholyte and anolyte solutions are continuously cooled with a chiller with a maximal capacity of 3.3 kW. The temperature of each electrolyte was stable in the range of 20–25 °C during all electrolysis experiments.

Product quantification

HCOOK quantification

HCOOK concentration was determined by high-performance liquid chromatography (HPLC) using a 1260 Agilent Infinity II HPLC equipped with a DAD G7115A. Samples were acidified using HCl and calibrations were performed with known amounts of formic acid (HCOOH). These analyses were performed at the Centre National en Électrochimie et en Technologies Environnementales (CNETE) in Shawinigan, QC.

HCOOK concentration was also determined through titration following the ACS Reagent Chemicals procedure for HCOONa.67 In short, an appropriate volume (0.250–1.000 mL) of the sampled aliquot electrolyte (catholyte or anolyte; V sample) is mixed with ~20 mL of deionized water and 5.000 mL (VKMnO) of 0.020 M (KMnO) KMnO aqueous solution. This mixture is boiled for 10 min, under stirring and with a watch glass, then quickly cooled down to room-temperature in an ice bath. In this order: ~2 mL of 5 M HSO; ~200 mg of KI; and 3–4 drops of starch indicators are added. The resulting solution is titrated with 0.100 M ( Na₂S₂O₃ ) NaSO until the solution is transparent (Vtitrant ). The concentration of HCOOK (HCOOK ) is given by the Equation 7 below:

Acid-base titrations

Hydroxides (OH−; Eq. 8), carbonates (CO²−; Eq. 9), and bicarbonates (HCO−; Eq. 10) respective concentrations were determined by acid titration (HCl 0.1 M) using an automatic titrator InMotion Flex and T5 from Mettler Toledo. Calibrations were performed using known quantities of Na₂CO in triplicates. These analyses were performed at CNETE in Shawinigan, QC.

Faradaic efficiency (FE) calculation

Faradaic efficiencies values for HCOOK (FEHCOOK ) was calculated based on the Equation 11 below:

Where HCOOK is the concentration of HCOOK in either the anolyte or catholyte in mol L⁻¹, V electrolyte is the volume of the anolyte or catholyte in L, is the number of electrons per mole of HCOOK product involved in the reduction reaction (=2), F is Faraday's constant (F =96,485 C mol⁻¹), I is the total current in A, and t is the elapsed time of electrolysis in s. If not otherwise stated, FEHCOOK values reported in this manuscript are for the formate quantified in the catholyte only.

Calculation of the Ohmic drop in electrolyte

The theoretical Ohmic drop due to electrolyte resistance (∆Ø electrolyte) can be calculated based on the experimental conductivity values (urn:x-wiley:18673880:media:cctc202300977:cctc202300977-math-0041 ) found for aqueous KOH solutions of 2 M (40.5 S m−1) and 4 M (61.5 S m−1), as well as 1 M K2CO (16.6 S m−1) and 2 M KHCO3 (12.5 S m−1) in Equation 12 and Ohm's law (Eq. 13. Ohmic drops are reported in Table S1.

Where Relectrolyte is the electrical resistance of the electrolyte in Ohms between the GDE and the microporous separator, is the distance in m between the GDE and the microporous separator, A is the geometric area of the electrode (0.1526 m2), and I is the total applied current (305 A).

Author Contributions

U. Legrand and A. Fink conceptualized the project. F. Navarro-Pardo helped with performing titrations. A. Fink performed all experiments and wrote the original draft. All authors contributed to manuscript review and editing.


We acknowledge Dr. Antonio Ramirez and Christian Désilets (CNETE) for their help on analyzing the electrolyte samples, as well as the catalyst synthesis. This research was fully funded by Electro Carbon Inc.

Conflict of interest

The author U. Legrand is the co-founder of Electro Carbon Inc., a start-up aiming to commercialize CO₂ electroreduction technology and thus acknowledge personal financial interest in this research. A. Fink and F. Navarro-Pardo are employees from Electro Carbon Inc. This conflict of interest has not led any of the co-authors to bias or otherwise modify any of the methods and/or results reported here.

Link to original post



For more information about Electro Carbon, our world’s cleanest potassium formate, or to submit a speculative application.

Thank you for your interest

bottom of page