Field of the invention

The invention relates to methods for producing formic acid.

Background of the invention

In order to mitigate the effects of the rising concentration of CO2 on our climate, there is a need to find ways to capture CO2 and use it as a feedstock in the production of chemical building blocks, synthetic fuels and building materials. This way, the carbon cycle may be closed and chemical production can be decoupled from fossil resources, if the captured CO2 is of non-fossil origin [Katelhon et at. (2019) Proc. Natl. Acad. Sci. USA 166(23), 11187-11194]. This is generally known as Carbon Capture and Utilization (CCU).

Meanwhile, renewable energy technologies are being developed to cater to the world's increasing energy demand without emitting large amounts of additional CO2 into the atmosphere. With an increasing share of variable renewable energy sources, the need for grid balancing services by storage of energy grows [Victoria et a/. (2019) Energy Co nvers. Manag. 201, 111977]. Batteries are being used for short-term storage of electricity. However, when large capacity storage or import of energy are desired, energy storage in gaseous or liquid molecules is more attractive. A German study by Dena revealed that relying on a technology mix with CCU fuels, as opposed to full electrification scenarios, can help reach climate targets for Germany at a lowered cost of around €600 billion. This is mainly due to the fact that we can rely on an existing infrastructure (pipelines, trucks, etc.) and end-use appliances, lowering the capital expenses.

The electrochemical carbon dioxide reduction reaction (CO2RR) provides a way to combine CO2 utilization and renewable energy storage into chemical products. Formic acid, the CO2 conversion product that is aimed for in this work, can be used as a fuel in fuel cells (Direct Formic Acid Fuel Cells or indirectly in Hydrogen Fuel Cells), as a chemical product, for example an antibacterial agent, or as a syngas equivalent for the production of a range of chemicals, referred to as the 'formate bio-economy' [Morrison et at. (2019) J. Electrochem. Soc. 166(4), E77-E86]. A techno-economic analysis by Spurgeon et at. projected that electrochemical reduction of CO2 to formic acid could be close to economic competitiveness on the chemical market [Spurgeon (2018) Energy Environ. Sci. 11, 1536-1551]. Extensive work has therefore been devoted to the development of heterogeneous electrocatalysts with high selectivity and low overpotential for formic acid synthesis. Most research in this area focuses on liquid phase CO2 reduction, where CO2 is sparged into an electrolyte, often a bicarbonate solution, and is converted on a metal-based cathode. Metals such as Sn, Pb, Bi, In, etc. have been observed to reach high faradaic efficiencies towards formate [Lu et at. (2014) ChemElectroChem 1(5), 836-849; Qiao et at. (2014) Chem. Soc. Rev. 43(2), 631-675; Endrodi et a/. (2017) Prog. Energy Combust. Sci. 62, 133-154; Du et al. (2017) Appl. Electrochem. 47(6), 661-678; Jhong et al. (2013) Curr. Opin. Chem. Eng. 2(2), 191-199.

Especially Sn(-oxide) based electrodes have been researched extensively, as it is an earth abundant, low cost and safe metal with electronic properties that sit near the Sabatier optimum for the *OCHO intermediate [Ross et a/. (2019) Nat. Catal. 2 648 -658; Feaster et al. (2017) ACS Catal. 7(7), 4822-4827], The faradaic efficiency increases even further for certain bimetallic catalysts, which is often prescribed to synergistic electronic effects, where the positive oxidation state of the active metal catalyst is maintained, thereby stabilizing the reaction intermediate towards formic acid and suppressing hydrogen evolution. Examples of bimetallic catalysts are PdSn, PbSn and CuSn, reaching faradaic efficiencies towards formate of over 95% in a liquid medium [Bai et al. (2017) Angew Chem Int Ed Engl. 56(40), 12219-12223; Moore (2017) ChemSusChem 10, 3512- 3519; Zheng et al. (2018) Nat. Catal. 218, 55-61].

A major disadvantage of liquid electrolyte systems, is that the current density is strongly limited by CO2 diffusion. Solvation dynamics of CO2 in aqueous solutions are slow, such that the cathodic overpotential shows an optimum value. When the overpotential is increased further, the hydrogen evolution reaction takes over, as this is not diffusion limited to the same extent as the CO2 reduction reaction. Current densities, although strongly dependent on the electrolyte concentration [Lv et al. (2014) J. Power Sources 253, 276-281], are observed to be limited to a few tens of mA/cm ² , which is too small to be commercially viable [Ross et al. cited above].

Additionally, at neutral electrolyte pH values, formate is produced instead of formic acid. The product would therefore need to be acidified and distilled from a highly concentrated salt solution, which greatly increases the cost of the process. A life cycle assessment performed by Thonemann et al. showed that the high thermal energy demand needed for the distillation of a low-concentration formic acid product leads to a positive net global warming impact for electrochemically produced formic acid, thereby negating the large carbon savings of the product synthesis itself [Thonemann and Pizzol (2019) Energy Environ. Sci. 12, 2253- 2263].

Centi et al. therefore stated that the feasibility of CO2 conversion to fuels depends on the possibility to form liquid fuels under solventless conditions, that can be easily collected without the need for distillation [Centi and Perathoner (2010) ChemSusChem 3(2), 195-208]. Therefore, in recent years, attention has been pointed towards semi- or full gas phase systems, where CO2 is made available to the catalyst in a high concentration and/or the product is generated in an almost water-free environment [Xia et at. (2019) Nat. Energy 4 776 —785; Lee et at. (2013) ECS Trans. 53(29), 41-47; De Mot et al. (2019) Chem. Eng. J. 378, 122224; Yang et al. (2017) J. C02 Util. 20 208 -217].

For example, Lee et al. have reported on a completely gas-phase system producing formic acid. Here, H2 is used as the source of protons, that are transported to the cathode through a Nation® membrane, where they are combined with CO2 and electrons to form formic acid. Faradaic efficiencies towards formic acid are rather low, with values ranging from 5 to 12.5%. The competitive hydrogen evolution reaction (HER) accounts for most of the remaining production [Lee et al. (2013) cited above; Lee et al. (2015) Mater. Chem. A 3(6), 3029- 3034]. Park et al. used a similar reactor cell and reported increased faradaic efficiencies at increased cell temperatures and water vapour contents of CO2 [Lee et al. (2018) Angew. Chemie - Int. Ed. 57(23), 6883-6887. To our knowledge, no examples exist of gas-phase systems with an exclusive formation of formic acid. Products such as CO, H2, etc. are always formed as by-products. Additionally, the CO2 conversion of these systems is very low (< 5%). Spurgeon et al. have discussed the importance of recycling unreacted CO2, as this lowers the cost of the carbon capture process relative to the rate of liquid product formation, and proposed to separate unreacted CO2 from the other gaseous products and recycle the non-converted CO2 [Spurgeon (2018) cited above]. WO2014202854, WO2014202855 and US2017321334 disclose methods of electrochemically reducing CO2 into formic acid into a gas phase recirculation system. Summary of the invention

A recirculated, gas phase electrochemical reactor system for the production of formic acid from CO2 is proposed. The reactor system consists of an electrochemical cell, producing formic acid, which is evacuated by the cathodic gas flow. Then, formic acid is separated by condensation while the non-converted CO2 and gaseous by-products are recycled to the reactor inlet.

A mathematical simulation of this reactor system was made, in which parameters such as system pressure and reduction potential are analysed.

From this, it is shown that a highly concentrated formic acid product can be obtained, along with a high CO2 conversion. Moreover, it is shown that by recirculating the undesired gaseous by-products, such as H2, the selectivity of the electrochemical reaction towards formic acid can be increased significantly, even at relatively low pressures. For the electrochemical parameters used in this simulation, it is shown that the specific energy consumption (kWh/kg) for formic acid production can be reduced by approximately 25% at a system pressure of 1 barg, and even further when the system pressure is increased.

This work urges to broaden the scope of research to integrated system design in order to fill the gap between laboratory reactors and industrially relevant systems. We herein present that recycling gases in the cathodic outflow of the reactor back to the reactor inflow will have a beneficial effect on the selectivity of the catalyst towards the desired formic acid vapour product. This follows from the Nernst equation, where higher partial pressures of these by-products will increase the potential required to form these products, thereby shifting the equilibrium in favour of formic acid production.

The invention is further summarised in the following statements:

1. A method of electrochemically reducing CO2 into formic acid in a gas phase recirculation system, wherein a cycle of the method comprises the steps of: a) introducing a gas comprising a source of H into the anode compartment of a reactor cell, and introducing a gas comprising CO2 in the cathode compartment of said reactor cell, wherein the cathode compartment of the reactor cell comprises an electrocatalyst, thereby reducing CO2 to HCOOH, with generating H2 and CO as by-products, b) transferring the gas comprising the products obtained in step a) to a separator removing HCOOH from the system, c) removing part of the gas obtained after removal of HCOOH in step b) from the system via an adjustable regulator valve, and d) supplementing the remaining part of the gas in the system obtained after step c) with carbon dioxide, and introducing the carbon dioxide supplemented gas to the reactor cell.

2. The method according to statement 1, wherein the source of H is selected from the group consisting of H2 gas, a gas comprising H2, NH3 gas, a gas comprising NH3, liquid water or water vapour. Herein the water can be pure water or waste water.

3. The method according to statement 1 or 2, wherein the separator is selected from the group consisting of a condenser converting gaseous formic acid into a liquid, a formic acid selective membrane and a solvent absorbing formic acid.

4. The method according to statement 3, wherein the flow and temperature of the gas are adjusted such that at least 75 %, at least 80 %, at least 85%, at least 90%, at least 95 % or at least 98% of the formic acid content in the gas is condensed.

5. The method according to any one of statements 1 to 4, wherein at least 25 cycles are performed.

6. The method according to any one of statements 1 to 5, wherein the cathode catalyst shows activity towards CO2 reduction and selectivity towards formic acid production, and is selected from the group consisting of a single transition metal, a bimetallic electrocatalyst or a trimetallic electrocatalyst.

7. The method according to statement 6, wherein one of the transition metals is p- and d-group metal catalysts such as Sn, Cu, Pd, Pb, Bi or Ag.

8. The method according to any one of statements 1 to 7, wherein the CO2 reduction in step a) is performed at a pressure of between 1, 5, 10 to 50, 75, 80 or 100 bar (g), or between 100, 150, or 200 to 400, 450 or 500 bar.

9. The method according to any one of statements 1 to 8, wherein the C02 reduction in step a) is performed at a temperature of between 0, 10, 20, or 30 °C to 80, 90, 95 or 100 °C.

10. The method according to any one of statements 1 to 9, wherein the C02 inlet pressure in step a) is between 5, 10 to 50, 75, 80 or 100 bar (g), or between 100, 150, or 200 to 400, 450 or 500 bar.

11. The method according to any one of statements 1 to 10, wherein less than 25 %, less than 15 %, less than 10 %, less than 5 % of the gas is released via the regulator valve. 12. The method according to any one of statements 1 to 11, wherein the gas released from the regulatory valve is reintroduced into the reactor cell as a source of hydrogen.

13. A continuous closed loop gas phase recirculation system for electrochemically reducing CO2 into formic acid, the system comprising, arranged along the direction of the movement of the gas:

- a reaction chamber comprising an electrocatalyst, connected via the anode compartment to an inlet for a gas comprising a source of H and connected via the cathode compartment to an inlet for a gas comprising CO2,

- a separator for collecting formic acid, an outlet for condensed formic acid, and

- a regulator valve connected to an outlet.

The system further can comprise a means for pressurising the gas, typically placed in front of the reactor chamber.

14. The system according to statement 13, wherein the separator is a condenser.

15. The system according to statement 14, further comprising a connection from the regulator valve to the reactor chamber.

16. A method of electrochemically reducing CO2 into formic acid in a gas phase recirculation system, wherein a cycle of the method comprises the steps of: a) introducing a gas or liquid comprising a source of H into the anode compartment of a reactor cell, and introducing a gas comprising CO2 in the cathode compartment of said reactor cell, wherein the cathode compartment of the reactor cell comprises an electrocatalyst, thereby reducing CO2 to HCOOH, with generating H2 and CO as byproducts, b) removing HCOOH from the system, c) removing part of the gas obtained after removal of HCOOH in step b) from the system via an adjustable regulator valve set at a pressure which is lower than the reactor pressure, and d) supplementing the remaining part of the gas in the system obtained after step c) with carbon dioxide, and introducing the carbon dioxide supplemented gas to the reactor cell.

17. The method according to statement 16, comprising in step b) transferring the gas comprising the products obtained in step a) to a separator removing HCOOH from the system.

18. The method according to statement 16 or 17, wherein the source of H is selected from the group consisting of H2 gas, a gas comprising H2, NHsgas, a gas comprising NH3, liquid water or water vapor. 19. The method according to any of statements 16 to 18, wherein the separator is selected from the group consisting of a condenser converting gaseous formic acid into a liquid, a formic acid selective membrane and a solvent absorbing formic acid.

20. The method according to statement 19, wherein the flow and temperature of the gas are adjusted such that at least 75 % of the formic acid content in the gas is condensed.

21. The method according to any one of statements 16 to 20, wherein at least 25 cycles are performed.

22. The method according to any one of statements 16 to 23, wherein the cathode catalyst shows activity towards CO2 reduction and selectivity towards formic acid production, and is selected from the group consisting of a single metal, a bimetallic electrocatalyst or a trimetallic electrocatalyst.

23. The method according to statement 22, wherein one of the metals is p- and d-block metal catalysts such as Sn, Cu, Pd, Pb, Bi or Ag.

24. The method according to any one of statements 16 to 23, wherein the CO2 reduction in step a) is performed at a pressure of between 1 and 100 bar (g), or between 100 and 500 bar.

25. The method according to any one of statements 16 to 24, wherein the CO2 reduction in step a) is performed at a temperature of between 0°C and 100 °C.

26. The method according to any one of statements 16 to 25, wherein the CO2 inlet pressure in step a) is between 1 and 500 bar (g).

27. The method according to any one of statements 16 to 16, wherein less than 25 % of the gas is released via the regulator valve.

28. The method according to any one of statements 16 to 27, wherein the gas released from the regulatory valve is reintroduced into the reactor cell as a source of H .

29. A continuous closed loop gas phase recirculation system for electrochemically reducing CO2 into formic acid, by the method of statement 16, the system comprising, arranged along the direction of the movement of the gas:

- a reaction cell comprising an electrocatalyst in the cathode compartment of the reaction cell, connected via the anode compartment to an inlet for a gas comprising a source of H and connected via the cathode compartment to an inlet for a gas comprising CO2,

- a separator for collecting formic acid, an outlet for condensed formic acid, and - an adjustable regulator valve connected to an outlet set at a pressure which is lower than the reactor pressure.

30. The system according to statement 29, wherein the separator is a condenser.

31. The system according to statement 30, further comprising a connection from the regulator valve to the anode compartment of the reactor cell.

Detailed description

Figure legends:

Figure 1: schematic representation of a recirculated electrochemical CO2- reduction system for the production of formic acid (HCOOH).

Figure 2: experimental (dashed) and simulated (full line) IV-characteristics for the reduction of CO2 to formic acid (CO2 reduction reaction 'CO2RR')· H2 (hydrogen evolution reaction 'HER') is formed as the main side product. Experimental data from Lee et al. (2015) cited above.

Figure 3: Partial current density of the hydrogen evolution reaction ('HER') over time.

Figure 4: evolution of faradaic efficiency towards formic acid over an arbitrary time scale at different system pressures. The cathodic potential was kept at a constant value of -0.7V vs RHE.

Figure 5: faradaic efficiency towards formic acid as a function of the CO2 supply, at a system pressure of 10 barg.

Figure 6: A : influence of cathodic reduction potential on the faradaic efficiency towards formic acid in the base case scenario. B: visualisation of trade-off between selectivity and productivity at different cathodic potentials (V vs RHE).

Figures 7 and 8 show embodiments of systems for the generation of formic acid.

A system is disclosed where either water (vapour) or H2 is oxidised at the anode, protons are transported through a membrane and CO2 is reduced to formic acid (HCOOH) at the cathode.

Anode half reactions:

H2O -> Vi 0 ₂ (g) + 2 H ⁺ + 2 e E° = 1.229 V vs RHE H ₂ (g) -> 2 H ⁺ + 2 e E° = 0 V vs RHE Cathode half reaction:

C0 ₂ (g)+ 2 H ⁺ + 2 e - HCOOH (I) E° = -0.251 V vs RHE

Net overall reaction: C0 ₂ (g) + H2O (I) -> ¹ /2 0 ₂ (g) + HCOOH (I) E _ceii = - 1.48 V

When water is oxidised, at 25°C and partial pressures for all gases of 1 atmosphere, a theoretical cell potential of 1.48 V is needed to drive this reaction. The cell potential for this vapour phase formic acid production is 52 mV higher than what is needed for formic acid production in an acidic electrolyte in a liquid phase cell (-1.428 V), as the Gibbs free energy of pure HCOOH (I) is slightly higher (-346 kJ/mol) than that of HCOOH (aq) in solution (-356 kJ/mol). The theoretical specific energy consumption for this reaction is 1.72 kWh/kg HCOOH. Formic acid is then vaporized and carried out of the cell with the gas stream. The gas stream then goes to a condenser, where formic acid vapour is condensed along with water. The permanent gases, together with any formic acid and water that were not liquefied, are recycled to the cathode inflow. A schematic representation of the looping system is given in Figure 1. As the reaction within the looping system proceeds over time, the composition of the recirculated gas phase within the system changes. In order to investigate the effect of the changing gas composition on the reaction selectivity, a mathematical simulation was made. A mathematical simulation was made where the composition of a unit volume of gas was determined at several points in the reactor system, as shown in Figure 1: before (A) and after (B) the electrochemical reactor cell, after the condenser (C) and after a regulator valve (D).

Here, a summary of the model development will be given. A complete overview of the applied physical and mathematical equations can be found in the examples . In order to construct a model relevant to the proposed system, experimental data was extracted from a publication on full-gas phase experiments performed by Lee et al. (2015) cited above. In that work, faradaic efficiencies towards formic acid between 5 and 12.5% were reached in a fully gas-phase, solid electrolyte system. The remainder of the current was attributed to H2 formation [Lee et al. (2015) cited above]. Other products, mostly CO, have frequently been observed during electrochemical CO2 reduction reactions on Sn(oxide)-based catalysts, but, as they were not observed in this data set, they are not accounted for in our simulation. In general, it is observed that on catalysts that are selective towards formic acid, no liquid products are formed other than formic acid (Xia et al. cited above] Water is carried out of the reactor cell with the gas flow due to the presence of a water loaded membrane. The gas composition at point B in the recirculation system therefore consists of H2, HCOOH, CO2 and H2O.

In order to determine the exact composition of the gas coming out of the reactor cell, the partial current densities at the applied cell potential need to be calculated. Therefore, the current-voltage (IV) characteristics were simulated.

Figure 2 shows the experimental and simulated IV-characteristics. At the optimal cathodic potential (-0.7 V vs RHE), the faradaic efficiency towards formic acid derived from the simulated IV-curves is 12.5%, which matches the experimental value of 12.5% reported by Lee et al. (2015) cited above.

The amount of formic acid in the gas stream at point B is determined by the product yield and the vapour pressure at the reactor temperature. When the amount of formic acid exceeds the maximum vapour pressure, not all formic acid that was produced can be carried out of the reactor cell due to saturation of the gaseous flow. Then, condensation occurs. The equations and Antoine coefficients that were used to calculate the vapour pressures of water and formic acid can be found in the examples.

In the condenser, water and formic acid vapour are liquefied due to the lower vapour pressure at the condenser temperature. Depending on the temperature difference that is required, either air or liquid cooling can be used. In this simulation, the operation of the condenser is simulated by assuming that the vapour is in thermodynamic equilibrium with the formic acid-water condensate within the condenser.

The permanent gases (H2 and CO2) and the non-condensed water and formic acid vapour are then recirculated to the cathodic inlet.

As H2 is produced out of non-gaseous intermediates, the hydrogen evolution reaction (HER) would cause the system pressure to endlessly increase. The integration of an adjustable regulator valve behind the condenser gives the opportunity to release some of the gases and maintain the system pressure at a constant value. The valve is set at a pressure which is lower than the reactor pressure. After the regulator valve (point D), additional CO2 is supplied through a valve which is set at the reactor pressure. This way, the gas composition can be tuned and sufficient reagent can be supplied for the reaction to proceed at a desired steady-state gas composition and pressure.

The gas composition values at point D serve as the initial conditions for another iteration of the equations of the reactor cell, to simulate the looping of the gas flow.

As can be derived from the Nernst equation, the IV characteristics of an electrochemical system are dependent on the gas composition. As the reaction within the looping system proceeds, the increasing partial pressure of H2 in the gas mixture causes a shift in the Nernst equilibrium (Vo) of the hydrogen evolution reaction. This effect is shown in Figure 3, where it can be seen that over time, there is a shift in the IV-curve of the HER towards more negative potentials. In Figure 3, it is assumed that the system operates at a pressure of 10 barg. These conditions will be referred to as the 'base case scenario'. The parameters that were employed for this base case scenario are given in the examples.

This shift due to the changing composition of the gas mixture leads to a significant increase in the faradaic efficiency towards formic acid, which is shown in Figure 4.

As can be seen in Figure 4, at a pressure of 1 barg, an increase in the faradaic efficiency towards formic acid from 12.5% to 16.5% is obtained. This leads to a decrease in the specific energy consumption, going from 20.5 kWh/kg formic acid in a system without byproduct recirculation to 15.4 kWh/kg in a system with recirculation (Table 1). It can also be seen that when the pressure within the system is increased, the product selectivity can be increased even further. At an overpressure of 100 barg, a faradaic efficiency of 25.1% was obtained for this specific set of parameters, and the energy efficiency of the process is doubled. Table 1 : overview of product selectivity towards formic acid and specific energy consumption of the electrochemical reaction as a function of the applied system pressure. For the specific energy consumption, H2O oxidation at the anode was assumed, with an anodic overpotential of 0.28 V. Additional overpotential losses were not accounted for.

By reducing the energy costs to drive the electrochemical reactor, the operational cost of the process can be lowered significantly. For a more complete analysis, other operational costs need to be taken into account, such as the downstream product processing cost, the costs involved in the condensation and the compression costs of the make-up CO2 gas, if this is not supplied from a high- pressure waste stream. When working at high pressures, an additional cost to consider is the added capital cost of high-pressure resistant equipment. For the downstream product processing, two products can be considered. The first product is the formic acid/water mixture from the condenser. As mentioned previously, the purity of the formic acid product is of high importance, as the distillation step was identified as a main contributor to the global warming impact of the production process of formic acid by CO2 reduction [Thonemann and Pizzol (2019) Energy Environ. Sci. 12, 2253-2263; Thonemann and Schulte (2019)

Environ. Sci. Technol. 53, 21, 12320-12329. Of course, the demand for the purity of the formic acid product depends on the application that is envisioned. Formic acid is commercially sold in concentrations > 85%. Some smaller scale applications exist where lower concentrations are employed (e.g. 49-51%). Over time, the use of formic acid as a fuel in DFAFC's or as a liquid organic hydrogen carrier (LOHC) will most likely gain interest. In DFAFC's, formic acid concentrations of 2-12 M have been observed to lead to optimal cell performance, considering a trade-off between current and fuel crossover [Aslam etai. (2012) APCBEE Procedia 3, 33-39]. For the base case scenario, it was determined by simulation that the concentration of the formic acid solution in the condenser is approximately 7.8 M (~ 34 wt%), which implies that, depending on the desired application, this product could either be valorised without further processing, or with a limited need for distillation. Optimisation of this product purity by tuning parameters such as the flowrate, reactor dimensions, condenser temperature etc. is possible and will be an important aspect to ensure a cost-effective process.

The second product is the gaseous mixture that is released from the regulator valve. For the base case scenario, the product distribution is given in Table 2. Here, it can be seen that per kg formic acid produced inside the condenser, 0.16 kg H2 and 1.27 kg CO2 are removed from the recirculation system through the regulator valve. At the base case scenario, 99.9% of the produced formic acid can be found in the condenser, while 0.1% disappears through the regulator valve. Table 2 : product distribution during steady-state operation in the base case scenario.

The amount and composition of this by-product is dependent on the regulator valve setting and can be tuned. When less CO2 is supplied, product selectivity increases compared to a pure CO2 supply. Figure 5 shows how the steady state faradaic efficiency increases as CO2 supply is varied, for a system operating at the base case conditions (10 barg). For example, '2 bar CO2 supply' implies that gas is released at the regulator valve until the pressure is reduced to 8 barg, and CO2 is supplied until the pressure increases back to 10 barg. Evidently, when this much product gases are exchanged for pure CO2, the effect of by-product recirculation on the product selectivity is small, more exhaust gases are produced and the CO2 conversion is lowered.

The CO2 supply can, of course, not be decreased endlessly. When the regulator valve pressure is chosen very close or equal to the desired system pressure, instead of converging to a steady-state, the looping system would fill up with H2 and/or other by-product gases, causing the formic acid formation to stop due to a lack of reagent.

One could also argue that a lower partial pressure of CO2 lowers the CO2 availability in the system, which could have a negative effect on the CO2RR. In liquid phase systems, a high CO2 pressure has a beneficial effect on the product selectivity. This is caused by the partial current density for CO2 reduction being limited by CO2 diffusion through the electrolyte. Increasing the system pressure increases the CO2 solubility in the electrolyte and therefore also the limiting partial current density for CO2 reduction [Morrison et at. (2019) J. Electrochem. Soc. 166(4), E77-E86]. This is exemplified by Vlugt et a/., who used a high pressure experimental setup and observed that when a catholyte is pressurized with a high CO2 pressure of 40-50 bar, an increased formate selectivity and production rate can be reached [Ramdin et a/. (2019) Ind. Eng. Chem. Res. 58(5), 1834-1847]. For fully gas phase systems, the CO2 availability can be assumed to be less of a limiting factor than in liquid electrolyte systems. This is reasoned by the fact that the distance the CO2 has to diffuse through the electrolyte is much smaller. On the electrode, an ionomer and/or liquid film is present of which the thickness is typically in the order of magnitude of only a few tens of nm's. The diffusion distance in a zero gap system is therefore much smaller than in a liquid system. Endrodi et al. have researched the effect of having a pressurized, pure CO2 gas feed in a zero gap CO2 electrolyzer, and observed an initial increase in the current density for the CO2RR when increasing the pressure from 1 to 2 bar, but no significant improvement with further increase of the CO2 pressure [Endrodi et al. (2019) Prog. Energy Combust. Sci. 62, 133-154]. This is an indication that diffusion of CO2 is less of a limiting factor in gas phase systems than in liquid systems.

In Table 3, an overview is given of how the regulator valve pressure setting influences the CO2 partial pressure, the simulated faradaic efficiency towards formic acid and the composition of the regulator valve exhaust. Decreasing the regulator valve pressure implies that more gas is released per kg of liquid formic acid product, reducing the CO2 conversion, and that the CO2 partial pressure in the system is higher.

Experimental research will further explore the presence of a trade-off, where increasing the CO2 partial pressure might benefit the CO2RR, but, in this recirculation system, it will also increase the amount of by-products released from the regulator valve, and simultaneously decrease the partial pressure of H2, thereby negatively affecting the product selectivity and energy efficiency towards formic acid.

A possible valorisation opportunity for the by-products released from the regulator valve is to re-use them as reducing agents. This can be done by recirculating these products to the anodic inlet of the electrochemical cell. The anodic outlet, then containing mostly CO2, could then be used to supplement the CO2 make-up gas stream. This adds to the circular approach of the proposed system.

Table 3 : Selectivity, product distribution and CO2 partial pressure during steady- state operation for different regulator valve pressure settings, at a system pressure of 10 barg.

An important aspect to consider while optimising the operation of the recirculated system, is the cell potential that is applied. Figure 6 (left) shows the influence of the cathodic potential on the product selectivity for the base case scenario parameters. At a reduction potential of -0.7 V vs RHE, the increase in selectivity is strongly visible, while at higher reduction potentials, the selectivity increase is smaller. A higher H2 pressure will be needed to significantly enhance the product selectivity at these higher reduction potentials.

Meanwhile, at higher overpotentials, the partial current density is higher, causing the formic acid yield to be higher. Here, the benefit of a higher productivity has to be weighed against the decreased product selectivity and energy efficiency. This effect is visualised on Figure 6 (right), where the trade-off is shown between the productivity and the selectivity towards formic acid at steady-state operation of the recirculation system. By increasing the pressure (100 barg), an increase in both productivity and selectivity can be realised. Increasing the pressure, however, is expected to also lead to more crossover of the product towards the anode [Endrodi et ai (2019) Prog. Energy Combust. Sci. 62, 133-154. The present invention describes a recirculated reactor system with a fully gas- phase electrochemical cell for the production of formic acid from CO2. In this system, unconverted CO2 is recycled to the cathode inflow, together with by product hydrogen gas.

Based on a mathematical simulation, it was shown that by recirculation, with careful tuning of the CO2 supply pressure and electrical overpotential, a significant enhancement of the selectivity towards formic acid can be realised. Moreover, a concentrated formic acid product can be obtained, avoiding large costs of downstream processing.

For a commonly used Sn catalyst in a zero gap cell, as employed by Lee et al. (2015), it was simulated that a selectivity increase from 12.5% to 21.5% can be realised by recirculating the by-products of the reaction at a relatively low pressure of 10 barg. This results in a significant decrease in the specific energy consumption of the CO2 reduction reaction.

These findings are especially relevant for scale-up and industrial application, where there is an interest to work with cheap, abundant, stable catalysts that might intrinsically not be very selective.

Depending on the valorisation opportunities of the gaseous and liquid products, the system can be optimised by tuning the parameters of the system components. To this purpose, this model is a useful tool to determine the settings of the different recirculation loop components in order to reach the desired steady-state operation. Experimental validation of the proposed system is needed. Focus should be given to analysis based on steady-state gas compositions, where the product selectivity is analysed in gaseous mixtures of CO2, H2 and potentially other by-product gases (e.g. CO) with varying compositions and pressure ranges.

Combining these efforts with further research into catalyst and reactor cell design, where increasingly high current densities in electrochemical CO2RR systems are being reported, will help to underline the promising nature of CCU technologies.

Example 1:

Model description

The simulation of the looping system was performed using Excel. The governing parameters were: flow rate, CO2 pressure valve set point, regulator valve set point, electrode surface area, reactor cell temperature, cathodic potential, condenser temperature and the volume fraction (relative humidity) of H2O coming out of the reactor cell. The gas composition was calculated at each point of the looped system and varied in time. Based on the changing partial pressures in the gaseous inflow, the current- voltage (IV) curves were continuously recalculated to determine the faradaic efficiency based on the partial current densities. In the following equations, the concentration of the different compounds in the gas phase was used (C [mol/cm ³ ] = n/V = P/RT) as the functional unit. For clarity, some constant for the conversion of units were left out of the equations.

A -> B

. H ₂ : mol

CH2,B = CH2,A + production _H2 cm ³

With PCD (partial current density) calculated numerically from IV-curves, that were simulated through following equations :

V VQ + Po _h m ^" I ^" VAct ^" I ^" Vconc

V ₀ = E ₀ - * T rea ,ctor * In Q n * F j * membrane thickness Vohm membrane conductivity

Herein, Q represents the reaction coefficient of partial pressures of the involved species. j _{(0 c)} Represents the (exchange) current density of the cathodic reactions, n the number of moles of electrons transferred in the balanced equation, a _c the charge transfer coefficient and a (V) and b (m ² /A) phenomenological coefficients. The parameters a, jo, a and b were adjusted to fit the experimental data within physically plausible ranges. The chosen parameters thus do not represent real (measured) physical properties of the catalyst material. The parameter values can be found in table 4. Table 4: exchange current densities, charge transfer coefficients and phenomenological coefficients (a,b) that were used to fit the simulated IV curves with the experimental data from Lee et al (2015) cited above.

For the calculation of qo _hm, properties of a Nafion 117 membrane are assumed, with a membrane thickness of 0.0183 cm and a conductivity of 0.01 S/cm. HCOOH: analogous to H2, but the possibility of condensation of formic acid inside the reactor was taken into account :

CHCOOH.B ⁼ C _HC OOH,A + production _HCOOH — HCOOH _liq

With the amount of condensed HCOOH determined by the liquid-vapour equilibrium at the reactor temperature. Here, the Antoine equation was used, for which the constants can be found in Table 5.

C0 ₂ :

Assuming that HCOOH is the only C-containing compound that was formed :

CCO2 _, B ⁼ _C O2 _, A — production _HCOOH

H2O :

It was assumed that the gas coming out of the reactor has a relative humidity of 50% due to the presence of a water-containing membrane. The accuracy of this assumption is dependent on several parameters, such as the type of membrane, the flow rate and the reactor dimensions. Due to the limited data on water evaporation from membranes during electrochemical operation, an accurate value is rather complex to determine. The assumption of a RH of 50% is expected to more accurately describe the situation in which a proton exchange membrane (PEM) is used than an anion exchange membrane (AEM), as with a PEM, the cathode side of the membrane is kept water loaded due to the vehicular proton transport mechanism, while for an AEM, water is transported in the opposite direction, leading to a more dried out membrane and therefore presumably lower humidities of the gaseous reactor outflow.

P _H ° ₂₀ [Pa] * 0.5

^L H20,B - n _T n ¹ reactor

Table 5: Antoine constants used to calculate the vapour pressures of formic acid and water

B ^ C

C0 ₂ , H ₂ : It is assumed that the permanent gases are not removed from the gas stream by the condenser, therefore : O2,C ⁼ Cc02,B HCOOH, H2O:

Water and formic acid vapour are liquefied inside the condenser if their concentration in the gas phase exceeds the maximal vapour pressure at the temperature of the condenser.

The maximal vapour pressure is determined assuming that within the condenser, the gas is in equilibrium with the produced formic acid-water condensate, which is variable over time.

Hereby, Raoult's Law of ideal mixtures is followed:

Pi = xi * P°

Where P, is the partial vapour pressure of a component (either water or formic acid) and x, the molar fraction in the liquid mixture. The gas leaving the condenser carries the amount of water and formic acid according to their vapour pressure at the condenser temperature: Thus, the amount of water and formic acid that is recovered inside the condenser is determined as follows : condenser yields = C _{l B} C _{l c}

As formic acid in water is, in practice, not an ideal mixture, this assumption will lead to a discrepancy. Formic acid forms an azeotrope in water, with the azeotropic composition also strongly depending on the temperature [Garcia-Payo et al. (2002) J. Memb. Sci. 198(2), 197-210]. Due to limited availability of data, this effect was not taken into account. Based on the Antoine parameters for pure solutions, formic acid is calculated to be more volatile than water. However, for dilute formic acid-water mixtures, it was observed that formic acid is slightly less volatile than water. Therefore, an error is made in positive direction: more formic acid will stay in the condensate than is obtained through the model, leading to a more concentrated product than what is now discussed. C ^ D

In between point C and D in the recirculation system, there is an adjustable valve. A valve setting (P _D ) at a lower pressure than the system pressure at point C is chosen in order be able to supply sufficient CO2 reagent for the reaction to proceed efficiently, and tune the steady-state gas composition to the desired value. The valve is simulated through following equation:

It is assumed that the gas is well-mixed and all gas- and vapour components are removed through the valve according to their relative abundance in the gas mixture at point C. D -> A

The gas composition values at point D are continuously re-inputted into the equations of the reactor cell to simulate the looping of the gas flows.

As CO2 is supplied as a make-up gas, the CO2 concentration going back into the reactor is determined through following equation :

(PA - Pp)

P-C02.AI — Cc02,D + R * T

Model assumptions Starting conditions

For fitting of the initial IV-curve (Figure 2), experimental data was extracted from Lee et a/. [12, figures 4 and 5]. Partial pressures of 0.1 bar H2O, 0.1 bar H2, 0.1 bar formic acid and 0.7 bar CO2 were assumed, adding up to a total of 1 bar (0 barg). An initial concentration inside the condenser of 50 vol % HCOOH was assumed.

Electrochemical reaction

As this model is used as a proof of concept and is therefore rather simplified, the following assumptions are made :

Firstly, it is assumed that the catalyst is stable over the course of the operation. This means that the faradaic efficiencies and current density do not shift due to inherent changes to the catalyst. Secondly, it is assumed that the recirculated gas phase products do not re-adsorb or cause poisoning or other detrimental effects on the Sn-catalyst. The local pH where the cathodic reaction occurs is assumed to be constant, with a proton activity of 1. Pressure increase is assumed to not results in an increase of the activity coefficient of the gases, for which an activity coefficient of 1 is assumed. The effect of temperature on the free energy of the molecules (entropy factor) is not taken into account in this model. It is also assumed that the anode reaction (H2O oxidation) does not work as a rate-limiting factor. Finally, it is assumed that there is no leaching of formic acid through the membrane to the anode side

Base case scenario

In the base case scenario, the following assumptions were made :

The system pressure, which corresponds to the CO2 make-up valve pressure setting, was assumed to be 11 bar (10 barg), while the regulator valve setting (P _D ) was assumed to be 10.99 bar. The cathodic potential was chosen to be -0.7 V vs RHE. The total gas volume within the loop was assumed to be 20 ml, while the flowrate was set to 20 ml/min. The geometric electrode surface area was assumed to be 9 cm ² , corresponding to the active area used by Lee et al cited above]. The reactor residence time was assumed to be 10 seconds. Finally, the reactor and condenser temperature were assumed to be 40°C and 0°C respectively.